diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3/config.json b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3/config.json new file mode 100644 index 0000000000000000000000000000000000000000..c31576a9e46a0f10b3f568fd1ab5cd51d7e1adec --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3/config.json @@ -0,0 +1,75 @@ +{ + "action_dim": 32, + "action_head_cfg": { + "action_dim": 32, + "action_horizon": 16, + "add_pos_embed": true, + "backbone_embedding_dim": 2048, + "diffusion_model_cfg": { + "attention_head_dim": 48, + "cross_attention_dim": 2048, + "dropout": 0.2, + "final_dropout": true, + "interleave_self_attention": true, + "norm_type": "ada_norm", + "num_attention_heads": 32, + "num_layers": 16, + "output_dim": 1024, + "positional_embeddings": null + }, + "hidden_size": 1024, + "input_embedding_dim": 1536, + "max_action_dim": 32, + "max_state_dim": 64, + "model_dtype": "float32", + "noise_beta_alpha": 1.5, + "noise_beta_beta": 1.0, + "noise_s": 0.999, + "num_inference_timesteps": 4, + "num_target_vision_tokens": 32, + "num_timestep_buckets": 1000, + "tune_diffusion_model": true, + "tune_projector": true, + "use_vlln": true, + "vl_self_attention_cfg": { + "attention_head_dim": 64, + "dropout": 0.2, + "final_dropout": true, + "num_attention_heads": 32, + "num_layers": 4, + "positional_embeddings": null + } + }, + "action_horizon": 16, + "architectures": [ + "GR00T_N1_5_MGD" + ], + "attn_implementation": null, + "backbone_cfg": { + "eagle_path": "NVEagle/eagle_er-qwen3_1_7B-Siglip2_400M_stage1_5_128gpu_er_v7_1mlp_nops", + "load_bf16": false, + "project_to_dim": null, + "reproject_vision": false, + "select_layer": 12, + "tune_llm": false, + "tune_visual": true, + "use_flash_attention": true + }, + "compute_dtype": "bfloat16", + "hidden_size": 2048, + "mgd_enabled": false, + "mgd_fm_loss_weight": 1.0, + "mgd_loss_weight": 0.0, + "mgd_pretrained_projector_path": null, + "mgd_sequence_hidden_dim": 512, + "mgd_target_dim": 512, + "mgd_target_pooling": "flatten", + "mgd_target_projection": "frozen_random", + "mgd_token_mask_ratio": 0.3, + "mgd_use_cosine_loss": true, + "mgd_use_mse_loss": false, + "model_dtype": "float32", + "model_type": "gr00t_n1_5", + "torch_dtype": "bfloat16", + "transformers_version": "4.51.3" +} diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3/model-00001-of-00002.safetensors b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3/model-00001-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b20e8c93aa6ecc7578225ae89a0947f4015b02f1 --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3/model-00001-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:84b77399325519ae31615104eb691ea1873a33771c7e81d5f21a7e5773b978e5 +size 4999367032 diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3/model-00002-of-00002.safetensors b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3/model-00002-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..cd5333358bb6b0f8e26923c27712a62ab51591f0 --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3/model-00002-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6632ed8555e6976f7883f956088758204eb8b5d0e923d6598c08fbb8c1989688 +size 2643339748 diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3/model.safetensors.index.json b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..579ccf56b744c24c73f571cce697c12343122f56 --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3/model.safetensors.index.json @@ -0,0 +1,916 @@ +{ + "metadata": { + "total_size": 7642572676 + }, + "weight_map": { + "action_head.action_decoder.layer1.W": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer1.b": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer2.W": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer2.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W1.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W1.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W2.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W2.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W3.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W3.b": "model-00002-of-00002.safetensors", + "action_head.future_tokens.weight": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_1.bias": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_1.weight": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_2.bias": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_2.weight": "model-00002-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_1.bias": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_1.weight": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_2.bias": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.position_embedding.weight": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer1.W": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer1.b": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer2.W": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer2.b": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vlln.bias": "model-00002-of-00002.safetensors", + "action_head.vlln.weight": "model-00002-of-00002.safetensors", + "action_target_projector.projector.bias": "model-00002-of-00002.safetensors", + "action_target_projector.projector.weight": "model-00002-of-00002.safetensors", + "backbone.eagle_model.language_model.lm_head.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.embed_tokens.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.mlp1.0.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.mlp1.0.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.patch_embedding.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.patch_embedding.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.position_embedding.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.in_proj_bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.in_proj_weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.layernorm.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.probe": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.post_layernorm.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.post_layernorm.weight": "model-00001-of-00002.safetensors", + "sequence_mgd_head.attn_score.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.attn_score.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.out_proj.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.out_proj.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.0.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.0.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.2.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.2.weight": "model-00002-of-00002.safetensors" + } +} diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3/trainer_state.json b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..1e5f9160facba5271c0989cbd88f45b4463e97d0 --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3/trainer_state.json @@ -0,0 +1,30043 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 8.728542333430317, + "eval_steps": 500, + "global_step": 30000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.002909514111143439, + "flow_matching_loss": 0.12651809230446814, + "grad_norm": 0.7402766346931458, + "learning_rate": 1.8e-07, + "loss": 0.1265, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10 + }, + { + "epoch": 0.005819028222286878, + "flow_matching_loss": 0.1152798742055893, + "grad_norm": 0.6575862765312195, + "learning_rate": 3.8e-07, + "loss": 0.1153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20 + }, + { + "epoch": 0.008728542333430317, + "flow_matching_loss": 0.11522509008646012, + "grad_norm": 0.7150079011917114, + "learning_rate": 5.800000000000001e-07, + "loss": 0.1152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 30 + }, + { + "epoch": 0.011638056444573755, + "flow_matching_loss": 0.09427751079201699, + "grad_norm": 0.46293339133262634, + "learning_rate": 7.799999999999999e-07, + "loss": 0.0943, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 40 + }, + { + "epoch": 0.014547570555717195, + "flow_matching_loss": 0.0906400889158249, + "grad_norm": 0.40832841396331787, + "learning_rate": 9.8e-07, + "loss": 0.0906, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 50 + }, + { + "epoch": 0.017457084666860634, + "flow_matching_loss": 0.08181314021348954, + "grad_norm": 0.43843352794647217, + "learning_rate": 1.18e-06, + "loss": 0.0818, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 60 + }, + { + "epoch": 0.020366598778004074, + "flow_matching_loss": 0.07855656705796718, + "grad_norm": 0.4632027745246887, + "learning_rate": 1.38e-06, + "loss": 0.0786, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 70 + }, + { + "epoch": 0.02327611288914751, + "flow_matching_loss": 0.06030846759676933, + "grad_norm": 0.3045955300331116, + "learning_rate": 1.5800000000000001e-06, + "loss": 0.0603, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 80 + }, + { + "epoch": 0.02618562700029095, + "flow_matching_loss": 0.062237000465393065, + "grad_norm": 0.2801671624183655, + "learning_rate": 1.7800000000000001e-06, + "loss": 0.0622, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 90 + }, + { + "epoch": 0.02909514111143439, + "flow_matching_loss": 0.05256380774080753, + "grad_norm": 0.20143912732601166, + "learning_rate": 1.98e-06, + "loss": 0.0526, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 100 + }, + { + "epoch": 0.03200465522257783, + "flow_matching_loss": 0.056555157899856566, + "grad_norm": 0.2408137023448944, + "learning_rate": 2.1800000000000003e-06, + "loss": 0.0566, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 110 + }, + { + "epoch": 0.03491416933372127, + "flow_matching_loss": 0.048005517944693564, + "grad_norm": 0.1884280890226364, + "learning_rate": 2.38e-06, + "loss": 0.048, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 120 + }, + { + "epoch": 0.037823683444864704, + "flow_matching_loss": 0.04668699577450752, + "grad_norm": 0.19780811667442322, + "learning_rate": 2.58e-06, + "loss": 0.0467, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 130 + }, + { + "epoch": 0.04073319755600815, + "flow_matching_loss": 0.04790678024291992, + "grad_norm": 0.1967543363571167, + "learning_rate": 2.78e-06, + "loss": 0.0479, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 140 + }, + { + "epoch": 0.043642711667151585, + "flow_matching_loss": 0.04863724298775196, + "grad_norm": 0.272660493850708, + "learning_rate": 2.98e-06, + "loss": 0.0486, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 150 + }, + { + "epoch": 0.04655222577829502, + "flow_matching_loss": 0.04353667423129082, + "grad_norm": 0.222812220454216, + "learning_rate": 3.18e-06, + "loss": 0.0435, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 160 + }, + { + "epoch": 0.049461739889438465, + "flow_matching_loss": 0.03733670599758625, + "grad_norm": 0.21927696466445923, + "learning_rate": 3.38e-06, + "loss": 0.0373, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 170 + }, + { + "epoch": 0.0523712540005819, + "flow_matching_loss": 0.0360936326906085, + "grad_norm": 0.24436259269714355, + "learning_rate": 3.58e-06, + "loss": 0.0361, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 180 + }, + { + "epoch": 0.055280768111725345, + "flow_matching_loss": 0.037410495616495607, + "grad_norm": 0.19850803911685944, + "learning_rate": 3.7800000000000002e-06, + "loss": 0.0374, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 190 + }, + { + "epoch": 0.05819028222286878, + "flow_matching_loss": 0.03579694926738739, + "grad_norm": 0.1686488837003708, + "learning_rate": 3.98e-06, + "loss": 0.0358, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 200 + }, + { + "epoch": 0.06109979633401222, + "flow_matching_loss": 0.037802538648247716, + "grad_norm": 0.17236050963401794, + "learning_rate": 4.18e-06, + "loss": 0.0378, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 210 + }, + { + "epoch": 0.06400931044515566, + "flow_matching_loss": 0.03397677205502987, + "grad_norm": 0.17126667499542236, + "learning_rate": 4.3799999999999996e-06, + "loss": 0.034, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 220 + }, + { + "epoch": 0.06691882455629909, + "flow_matching_loss": 0.038775010406970976, + "grad_norm": 0.16270321607589722, + "learning_rate": 4.58e-06, + "loss": 0.0388, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 230 + }, + { + "epoch": 0.06982833866744254, + "flow_matching_loss": 0.03883931208401918, + "grad_norm": 0.1858592927455902, + "learning_rate": 4.78e-06, + "loss": 0.0388, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 240 + }, + { + "epoch": 0.07273785277858598, + "flow_matching_loss": 0.03434933796525001, + "grad_norm": 0.19044414162635803, + "learning_rate": 4.980000000000001e-06, + "loss": 0.0343, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 250 + }, + { + "epoch": 0.07564736688972941, + "flow_matching_loss": 0.03208901342004537, + "grad_norm": 0.17155373096466064, + "learning_rate": 5.18e-06, + "loss": 0.0321, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 260 + }, + { + "epoch": 0.07855688100087285, + "flow_matching_loss": 0.03438600935041904, + "grad_norm": 0.1744360476732254, + "learning_rate": 5.38e-06, + "loss": 0.0344, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 270 + }, + { + "epoch": 0.0814663951120163, + "flow_matching_loss": 0.030176587775349616, + "grad_norm": 0.16271138191223145, + "learning_rate": 5.58e-06, + "loss": 0.0302, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 280 + }, + { + "epoch": 0.08437590922315973, + "flow_matching_loss": 0.03063834421336651, + "grad_norm": 0.13861629366874695, + "learning_rate": 5.7800000000000006e-06, + "loss": 0.0306, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 290 + }, + { + "epoch": 0.08728542333430317, + "flow_matching_loss": 0.035166236013174056, + "grad_norm": 0.20718400180339813, + "learning_rate": 5.98e-06, + "loss": 0.0352, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 300 + }, + { + "epoch": 0.09019493744544661, + "flow_matching_loss": 0.03251210562884808, + "grad_norm": 0.1849493533372879, + "learning_rate": 6.18e-06, + "loss": 0.0325, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 310 + }, + { + "epoch": 0.09310445155659004, + "flow_matching_loss": 0.025926103442907335, + "grad_norm": 0.1793784201145172, + "learning_rate": 6.38e-06, + "loss": 0.0259, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 320 + }, + { + "epoch": 0.09601396566773349, + "flow_matching_loss": 0.032622674107551576, + "grad_norm": 0.1985716074705124, + "learning_rate": 6.58e-06, + "loss": 0.0326, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 330 + }, + { + "epoch": 0.09892347977887693, + "flow_matching_loss": 0.03596934173256159, + "grad_norm": 0.26673901081085205, + "learning_rate": 6.78e-06, + "loss": 0.036, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 340 + }, + { + "epoch": 0.10183299389002037, + "flow_matching_loss": 0.028649488650262355, + "grad_norm": 0.16154420375823975, + "learning_rate": 6.98e-06, + "loss": 0.0286, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 350 + }, + { + "epoch": 0.1047425080011638, + "flow_matching_loss": 0.03157542757689953, + "grad_norm": 0.14434140920639038, + "learning_rate": 7.180000000000001e-06, + "loss": 0.0316, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 360 + }, + { + "epoch": 0.10765202211230725, + "flow_matching_loss": 0.02920713685452938, + "grad_norm": 0.1379987746477127, + "learning_rate": 7.3800000000000005e-06, + "loss": 0.0292, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 370 + }, + { + "epoch": 0.11056153622345069, + "flow_matching_loss": 0.027275916002690793, + "grad_norm": 0.1460859328508377, + "learning_rate": 7.5799999999999994e-06, + "loss": 0.0273, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 380 + }, + { + "epoch": 0.11347105033459412, + "flow_matching_loss": 0.026951910369098185, + "grad_norm": 0.2659739553928375, + "learning_rate": 7.780000000000002e-06, + "loss": 0.027, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 390 + }, + { + "epoch": 0.11638056444573756, + "flow_matching_loss": 0.027327189594507216, + "grad_norm": 0.14026117324829102, + "learning_rate": 7.98e-06, + "loss": 0.0273, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 400 + }, + { + "epoch": 0.11929007855688101, + "flow_matching_loss": 0.030604574270546435, + "grad_norm": 0.2799255847930908, + "learning_rate": 8.18e-06, + "loss": 0.0306, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 410 + }, + { + "epoch": 0.12219959266802444, + "flow_matching_loss": 0.026950815878808498, + "grad_norm": 0.1688050776720047, + "learning_rate": 8.38e-06, + "loss": 0.027, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 420 + }, + { + "epoch": 0.12510910677916787, + "flow_matching_loss": 0.03261478859931231, + "grad_norm": 0.21647225320339203, + "learning_rate": 8.58e-06, + "loss": 0.0326, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 430 + }, + { + "epoch": 0.12801862089031132, + "flow_matching_loss": 0.027793361619114876, + "grad_norm": 0.16190461814403534, + "learning_rate": 8.78e-06, + "loss": 0.0278, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 440 + }, + { + "epoch": 0.13092813500145475, + "flow_matching_loss": 0.026850407384335993, + "grad_norm": 0.1569625586271286, + "learning_rate": 8.98e-06, + "loss": 0.0269, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 450 + }, + { + "epoch": 0.13383764911259818, + "flow_matching_loss": 0.024850488640367985, + "grad_norm": 0.1453407108783722, + "learning_rate": 9.18e-06, + "loss": 0.0249, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 460 + }, + { + "epoch": 0.13674716322374164, + "flow_matching_loss": 0.026034651882946493, + "grad_norm": 0.15146100521087646, + "learning_rate": 9.38e-06, + "loss": 0.026, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 470 + }, + { + "epoch": 0.13965667733488507, + "flow_matching_loss": 0.027936949953436853, + "grad_norm": 0.14753958582878113, + "learning_rate": 9.580000000000001e-06, + "loss": 0.0279, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 480 + }, + { + "epoch": 0.1425661914460285, + "flow_matching_loss": 0.02663248684257269, + "grad_norm": 0.16540265083312988, + "learning_rate": 9.780000000000001e-06, + "loss": 0.0266, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 490 + }, + { + "epoch": 0.14547570555717196, + "flow_matching_loss": 0.025918115489184858, + "grad_norm": 0.13860824704170227, + "learning_rate": 9.980000000000001e-06, + "loss": 0.0259, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 500 + }, + { + "epoch": 0.1483852196683154, + "flow_matching_loss": 0.023474160581827164, + "grad_norm": 0.2924189865589142, + "learning_rate": 1.0179999999999999e-05, + "loss": 0.0235, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 510 + }, + { + "epoch": 0.15129473377945882, + "flow_matching_loss": 0.026986394077539444, + "grad_norm": 0.14106540381908417, + "learning_rate": 1.0379999999999999e-05, + "loss": 0.027, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 520 + }, + { + "epoch": 0.15420424789060228, + "flow_matching_loss": 0.02684167120605707, + "grad_norm": 0.1683470904827118, + "learning_rate": 1.058e-05, + "loss": 0.0268, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 530 + }, + { + "epoch": 0.1571137620017457, + "flow_matching_loss": 0.028073015995323657, + "grad_norm": 0.19711211323738098, + "learning_rate": 1.078e-05, + "loss": 0.0281, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 540 + }, + { + "epoch": 0.16002327611288913, + "flow_matching_loss": 0.022154679894447325, + "grad_norm": 0.16642902791500092, + "learning_rate": 1.098e-05, + "loss": 0.0222, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 550 + }, + { + "epoch": 0.1629327902240326, + "flow_matching_loss": 0.026503290981054306, + "grad_norm": 0.13388291001319885, + "learning_rate": 1.118e-05, + "loss": 0.0265, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 560 + }, + { + "epoch": 0.16584230433517602, + "flow_matching_loss": 0.02676616171374917, + "grad_norm": 0.14508862793445587, + "learning_rate": 1.1380000000000001e-05, + "loss": 0.0268, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 570 + }, + { + "epoch": 0.16875181844631945, + "flow_matching_loss": 0.024827801063656806, + "grad_norm": 0.17462259531021118, + "learning_rate": 1.1580000000000001e-05, + "loss": 0.0248, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 580 + }, + { + "epoch": 0.1716613325574629, + "flow_matching_loss": 0.025570261664688586, + "grad_norm": 0.16522862017154694, + "learning_rate": 1.178e-05, + "loss": 0.0256, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 590 + }, + { + "epoch": 0.17457084666860634, + "flow_matching_loss": 0.025263234786689282, + "grad_norm": 0.16251014173030853, + "learning_rate": 1.198e-05, + "loss": 0.0253, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 600 + }, + { + "epoch": 0.17748036077974977, + "flow_matching_loss": 0.022936346009373665, + "grad_norm": 0.14700005948543549, + "learning_rate": 1.2180000000000002e-05, + "loss": 0.0229, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 610 + }, + { + "epoch": 0.18038987489089323, + "flow_matching_loss": 0.02522765640169382, + "grad_norm": 0.12662570178508759, + "learning_rate": 1.238e-05, + "loss": 0.0252, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 620 + }, + { + "epoch": 0.18329938900203666, + "flow_matching_loss": 0.02583275381475687, + "grad_norm": 0.14435622096061707, + "learning_rate": 1.258e-05, + "loss": 0.0258, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 630 + }, + { + "epoch": 0.18620890311318009, + "flow_matching_loss": 0.025874245166778564, + "grad_norm": 0.16966219246387482, + "learning_rate": 1.278e-05, + "loss": 0.0259, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 640 + }, + { + "epoch": 0.18911841722432354, + "flow_matching_loss": 0.02542715221643448, + "grad_norm": 0.127089723944664, + "learning_rate": 1.298e-05, + "loss": 0.0254, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 650 + }, + { + "epoch": 0.19202793133546697, + "flow_matching_loss": 0.025847407430410384, + "grad_norm": 0.20329344272613525, + "learning_rate": 1.3180000000000001e-05, + "loss": 0.0258, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 660 + }, + { + "epoch": 0.1949374454466104, + "flow_matching_loss": 0.0242184916511178, + "grad_norm": 0.17702051997184753, + "learning_rate": 1.338e-05, + "loss": 0.0242, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 670 + }, + { + "epoch": 0.19784695955775386, + "flow_matching_loss": 0.02345710527151823, + "grad_norm": 0.16560539603233337, + "learning_rate": 1.358e-05, + "loss": 0.0235, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 680 + }, + { + "epoch": 0.2007564736688973, + "flow_matching_loss": 0.020882812794297934, + "grad_norm": 0.10396488010883331, + "learning_rate": 1.378e-05, + "loss": 0.0209, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 690 + }, + { + "epoch": 0.20366598778004075, + "flow_matching_loss": 0.0227653332054615, + "grad_norm": 0.17184723913669586, + "learning_rate": 1.3980000000000002e-05, + "loss": 0.0228, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 700 + }, + { + "epoch": 0.20657550189118418, + "flow_matching_loss": 0.024216826446354388, + "grad_norm": 0.14497768878936768, + "learning_rate": 1.4180000000000001e-05, + "loss": 0.0242, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 710 + }, + { + "epoch": 0.2094850160023276, + "flow_matching_loss": 0.022863735258579255, + "grad_norm": 0.13112197816371918, + "learning_rate": 1.438e-05, + "loss": 0.0229, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 720 + }, + { + "epoch": 0.21239453011347106, + "flow_matching_loss": 0.023854022845625877, + "grad_norm": 0.14173072576522827, + "learning_rate": 1.458e-05, + "loss": 0.0239, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 730 + }, + { + "epoch": 0.2153040442246145, + "flow_matching_loss": 0.025482414104044438, + "grad_norm": 0.12373295426368713, + "learning_rate": 1.4779999999999999e-05, + "loss": 0.0255, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 740 + }, + { + "epoch": 0.21821355833575792, + "flow_matching_loss": 0.025454910658299922, + "grad_norm": 0.2192395031452179, + "learning_rate": 1.498e-05, + "loss": 0.0255, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 750 + }, + { + "epoch": 0.22112307244690138, + "flow_matching_loss": 0.02401951365172863, + "grad_norm": 0.20380628108978271, + "learning_rate": 1.518e-05, + "loss": 0.024, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 760 + }, + { + "epoch": 0.2240325865580448, + "flow_matching_loss": 0.027595036663115025, + "grad_norm": 0.20759063959121704, + "learning_rate": 1.5380000000000002e-05, + "loss": 0.0276, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 770 + }, + { + "epoch": 0.22694210066918824, + "flow_matching_loss": 0.023473415058106184, + "grad_norm": 0.15538272261619568, + "learning_rate": 1.558e-05, + "loss": 0.0235, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 780 + }, + { + "epoch": 0.2298516147803317, + "flow_matching_loss": 0.02240349128842354, + "grad_norm": 0.1468908190727234, + "learning_rate": 1.578e-05, + "loss": 0.0224, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 790 + }, + { + "epoch": 0.23276112889147513, + "flow_matching_loss": 0.02415908668190241, + "grad_norm": 0.17530480027198792, + "learning_rate": 1.598e-05, + "loss": 0.0242, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 800 + }, + { + "epoch": 0.23567064300261856, + "flow_matching_loss": 0.020658717211335897, + "grad_norm": 0.12499396502971649, + "learning_rate": 1.618e-05, + "loss": 0.0207, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 810 + }, + { + "epoch": 0.23858015711376201, + "flow_matching_loss": 0.02262230236083269, + "grad_norm": 0.16670989990234375, + "learning_rate": 1.6380000000000002e-05, + "loss": 0.0226, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 820 + }, + { + "epoch": 0.24148967122490544, + "flow_matching_loss": 0.022605573665350674, + "grad_norm": 0.11775291711091995, + "learning_rate": 1.658e-05, + "loss": 0.0226, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 830 + }, + { + "epoch": 0.24439918533604887, + "flow_matching_loss": 0.02424959894269705, + "grad_norm": 0.12897488474845886, + "learning_rate": 1.6780000000000002e-05, + "loss": 0.0242, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 840 + }, + { + "epoch": 0.24730869944719233, + "flow_matching_loss": 0.0247517803683877, + "grad_norm": 0.1541481763124466, + "learning_rate": 1.698e-05, + "loss": 0.0248, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 850 + }, + { + "epoch": 0.25021821355833573, + "flow_matching_loss": 0.024185372423380612, + "grad_norm": 0.14106076955795288, + "learning_rate": 1.718e-05, + "loss": 0.0242, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 860 + }, + { + "epoch": 0.2531277276694792, + "flow_matching_loss": 0.02409915644675493, + "grad_norm": 0.13129174709320068, + "learning_rate": 1.7380000000000003e-05, + "loss": 0.0241, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 870 + }, + { + "epoch": 0.25603724178062265, + "flow_matching_loss": 0.025851870886981487, + "grad_norm": 0.18664920330047607, + "learning_rate": 1.758e-05, + "loss": 0.0259, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 880 + }, + { + "epoch": 0.2589467558917661, + "flow_matching_loss": 0.02311691138893366, + "grad_norm": 0.13099530339241028, + "learning_rate": 1.778e-05, + "loss": 0.0231, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 890 + }, + { + "epoch": 0.2618562700029095, + "flow_matching_loss": 0.026105239987373352, + "grad_norm": 0.12370136380195618, + "learning_rate": 1.798e-05, + "loss": 0.0261, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 900 + }, + { + "epoch": 0.26476578411405294, + "flow_matching_loss": 0.023170758225023746, + "grad_norm": 0.14878924190998077, + "learning_rate": 1.818e-05, + "loss": 0.0232, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 910 + }, + { + "epoch": 0.26767529822519637, + "flow_matching_loss": 0.021206838730722665, + "grad_norm": 0.15685400366783142, + "learning_rate": 1.838e-05, + "loss": 0.0212, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 920 + }, + { + "epoch": 0.27058481233633985, + "flow_matching_loss": 0.02472804393619299, + "grad_norm": 0.1351301372051239, + "learning_rate": 1.8579999999999998e-05, + "loss": 0.0247, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 930 + }, + { + "epoch": 0.2734943264474833, + "flow_matching_loss": 0.023353499360382557, + "grad_norm": 0.14254583418369293, + "learning_rate": 1.878e-05, + "loss": 0.0234, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 940 + }, + { + "epoch": 0.2764038405586267, + "flow_matching_loss": 0.022570143081247808, + "grad_norm": 0.1343407928943634, + "learning_rate": 1.898e-05, + "loss": 0.0226, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 950 + }, + { + "epoch": 0.27931335466977014, + "flow_matching_loss": 0.023178067244589328, + "grad_norm": 0.1260804831981659, + "learning_rate": 1.918e-05, + "loss": 0.0232, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 960 + }, + { + "epoch": 0.28222286878091357, + "flow_matching_loss": 0.02370437849313021, + "grad_norm": 0.17035196721553802, + "learning_rate": 1.938e-05, + "loss": 0.0237, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 970 + }, + { + "epoch": 0.285132382892057, + "flow_matching_loss": 0.02057368094101548, + "grad_norm": 0.13011594116687775, + "learning_rate": 1.958e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 980 + }, + { + "epoch": 0.2880418970032005, + "flow_matching_loss": 0.02727743536233902, + "grad_norm": 0.15859198570251465, + "learning_rate": 1.978e-05, + "loss": 0.0273, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 990 + }, + { + "epoch": 0.2909514111143439, + "flow_matching_loss": 0.022164654452353716, + "grad_norm": 0.11995124816894531, + "learning_rate": 1.9980000000000002e-05, + "loss": 0.0222, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1000 + }, + { + "epoch": 0.29386092522548735, + "flow_matching_loss": 0.021997209824621677, + "grad_norm": 0.1803901046514511, + "learning_rate": 2.018e-05, + "loss": 0.022, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1010 + }, + { + "epoch": 0.2967704393366308, + "flow_matching_loss": 0.027047692239284514, + "grad_norm": 0.13500265777111053, + "learning_rate": 2.038e-05, + "loss": 0.027, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1020 + }, + { + "epoch": 0.2996799534477742, + "flow_matching_loss": 0.020784050039947034, + "grad_norm": 0.1290229856967926, + "learning_rate": 2.0580000000000003e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1030 + }, + { + "epoch": 0.30258946755891764, + "flow_matching_loss": 0.021048144903033972, + "grad_norm": 0.10664508491754532, + "learning_rate": 2.078e-05, + "loss": 0.021, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1040 + }, + { + "epoch": 0.3054989816700611, + "flow_matching_loss": 0.020910587813705205, + "grad_norm": 0.13554227352142334, + "learning_rate": 2.0980000000000002e-05, + "loss": 0.0209, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1050 + }, + { + "epoch": 0.30840849578120455, + "flow_matching_loss": 0.018838377203792334, + "grad_norm": 0.13401055335998535, + "learning_rate": 2.118e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1060 + }, + { + "epoch": 0.311318009892348, + "flow_matching_loss": 0.022762462217360734, + "grad_norm": 0.15523949265480042, + "learning_rate": 2.1380000000000002e-05, + "loss": 0.0228, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1070 + }, + { + "epoch": 0.3142275240034914, + "flow_matching_loss": 0.02361538903787732, + "grad_norm": 0.18030945956707, + "learning_rate": 2.1580000000000003e-05, + "loss": 0.0236, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1080 + }, + { + "epoch": 0.31713703811463484, + "flow_matching_loss": 0.021665265876799823, + "grad_norm": 0.17576953768730164, + "learning_rate": 2.178e-05, + "loss": 0.0217, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1090 + }, + { + "epoch": 0.32004655222577827, + "flow_matching_loss": 0.022684145998209716, + "grad_norm": 0.10969354957342148, + "learning_rate": 2.198e-05, + "loss": 0.0227, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1100 + }, + { + "epoch": 0.32295606633692175, + "flow_matching_loss": 0.020700398832559586, + "grad_norm": 0.12754011154174805, + "learning_rate": 2.2179999999999998e-05, + "loss": 0.0207, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1110 + }, + { + "epoch": 0.3258655804480652, + "flow_matching_loss": 0.021664462611079217, + "grad_norm": 0.1397317796945572, + "learning_rate": 2.238e-05, + "loss": 0.0217, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1120 + }, + { + "epoch": 0.3287750945592086, + "flow_matching_loss": 0.02394861588254571, + "grad_norm": 0.15131065249443054, + "learning_rate": 2.258e-05, + "loss": 0.0239, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1130 + }, + { + "epoch": 0.33168460867035204, + "flow_matching_loss": 0.023953361436724663, + "grad_norm": 0.14684942364692688, + "learning_rate": 2.278e-05, + "loss": 0.024, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1140 + }, + { + "epoch": 0.3345941227814955, + "flow_matching_loss": 0.02011026395484805, + "grad_norm": 0.15347738564014435, + "learning_rate": 2.298e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1150 + }, + { + "epoch": 0.3375036368926389, + "flow_matching_loss": 0.022831694688647985, + "grad_norm": 0.1693112850189209, + "learning_rate": 2.3179999999999998e-05, + "loss": 0.0228, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1160 + }, + { + "epoch": 0.3404131510037824, + "flow_matching_loss": 0.02607177309691906, + "grad_norm": 0.18736708164215088, + "learning_rate": 2.338e-05, + "loss": 0.0261, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1170 + }, + { + "epoch": 0.3433226651149258, + "flow_matching_loss": 0.02353391982614994, + "grad_norm": 0.14527125656604767, + "learning_rate": 2.358e-05, + "loss": 0.0235, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1180 + }, + { + "epoch": 0.34623217922606925, + "flow_matching_loss": 0.020372791215777397, + "grad_norm": 0.15006881952285767, + "learning_rate": 2.378e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1190 + }, + { + "epoch": 0.3491416933372127, + "flow_matching_loss": 0.02451947666704655, + "grad_norm": 0.24306942522525787, + "learning_rate": 2.398e-05, + "loss": 0.0245, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1200 + }, + { + "epoch": 0.3520512074483561, + "flow_matching_loss": 0.021212649811059237, + "grad_norm": 0.14053194224834442, + "learning_rate": 2.4180000000000002e-05, + "loss": 0.0212, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1210 + }, + { + "epoch": 0.35496072155949954, + "flow_matching_loss": 0.02220342755317688, + "grad_norm": 0.17332050204277039, + "learning_rate": 2.438e-05, + "loss": 0.0222, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1220 + }, + { + "epoch": 0.357870235670643, + "flow_matching_loss": 0.022382692992687227, + "grad_norm": 0.17098546028137207, + "learning_rate": 2.4580000000000002e-05, + "loss": 0.0224, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1230 + }, + { + "epoch": 0.36077974978178645, + "flow_matching_loss": 0.02123220395296812, + "grad_norm": 0.1380387842655182, + "learning_rate": 2.478e-05, + "loss": 0.0212, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1240 + }, + { + "epoch": 0.3636892638929299, + "flow_matching_loss": 0.020996518526226284, + "grad_norm": 0.18676133453845978, + "learning_rate": 2.498e-05, + "loss": 0.021, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1250 + }, + { + "epoch": 0.3665987780040733, + "flow_matching_loss": 0.021872975677251816, + "grad_norm": 0.1378878951072693, + "learning_rate": 2.5180000000000003e-05, + "loss": 0.0219, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1260 + }, + { + "epoch": 0.36950829211521674, + "flow_matching_loss": 0.020465939212590455, + "grad_norm": 0.15487609803676605, + "learning_rate": 2.538e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1270 + }, + { + "epoch": 0.37241780622636017, + "flow_matching_loss": 0.022392774187028407, + "grad_norm": 0.1305442601442337, + "learning_rate": 2.5580000000000002e-05, + "loss": 0.0224, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1280 + }, + { + "epoch": 0.37532732033750366, + "flow_matching_loss": 0.022139374911785126, + "grad_norm": 0.1365901231765747, + "learning_rate": 2.578e-05, + "loss": 0.0221, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1290 + }, + { + "epoch": 0.3782368344486471, + "flow_matching_loss": 0.020555764716118575, + "grad_norm": 0.1513245701789856, + "learning_rate": 2.5980000000000002e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1300 + }, + { + "epoch": 0.3811463485597905, + "flow_matching_loss": 0.01913353120908141, + "grad_norm": 0.12608614563941956, + "learning_rate": 2.618e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1310 + }, + { + "epoch": 0.38405586267093395, + "flow_matching_loss": 0.017857550736516715, + "grad_norm": 0.1288081407546997, + "learning_rate": 2.6379999999999998e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1320 + }, + { + "epoch": 0.3869653767820774, + "flow_matching_loss": 0.025874372571706772, + "grad_norm": 0.1576732099056244, + "learning_rate": 2.658e-05, + "loss": 0.0259, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1330 + }, + { + "epoch": 0.3898748908932208, + "flow_matching_loss": 0.023091320879757404, + "grad_norm": 0.15448147058486938, + "learning_rate": 2.678e-05, + "loss": 0.0231, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1340 + }, + { + "epoch": 0.3927844050043643, + "flow_matching_loss": 0.022764113359153272, + "grad_norm": 0.21157103776931763, + "learning_rate": 2.698e-05, + "loss": 0.0228, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1350 + }, + { + "epoch": 0.3956939191155077, + "flow_matching_loss": 0.022034947481006383, + "grad_norm": 0.1277129203081131, + "learning_rate": 2.718e-05, + "loss": 0.022, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1360 + }, + { + "epoch": 0.39860343322665115, + "flow_matching_loss": 0.022598444670438766, + "grad_norm": 0.15181100368499756, + "learning_rate": 2.738e-05, + "loss": 0.0226, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1370 + }, + { + "epoch": 0.4015129473377946, + "flow_matching_loss": 0.01962289456278086, + "grad_norm": 0.14532554149627686, + "learning_rate": 2.758e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1380 + }, + { + "epoch": 0.404422461448938, + "flow_matching_loss": 0.020831809006631376, + "grad_norm": 0.14978381991386414, + "learning_rate": 2.778e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1390 + }, + { + "epoch": 0.4073319755600815, + "flow_matching_loss": 0.021022525057196616, + "grad_norm": 0.1824600249528885, + "learning_rate": 2.798e-05, + "loss": 0.021, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1400 + }, + { + "epoch": 0.4102414896712249, + "flow_matching_loss": 0.020296704210340976, + "grad_norm": 0.2004566490650177, + "learning_rate": 2.818e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1410 + }, + { + "epoch": 0.41315100378236835, + "flow_matching_loss": 0.02267464492470026, + "grad_norm": 0.20057177543640137, + "learning_rate": 2.838e-05, + "loss": 0.0227, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1420 + }, + { + "epoch": 0.4160605178935118, + "flow_matching_loss": 0.020953496918082237, + "grad_norm": 0.1387690156698227, + "learning_rate": 2.858e-05, + "loss": 0.021, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1430 + }, + { + "epoch": 0.4189700320046552, + "flow_matching_loss": 0.01959231598302722, + "grad_norm": 0.1446612924337387, + "learning_rate": 2.8780000000000002e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1440 + }, + { + "epoch": 0.42187954611579864, + "flow_matching_loss": 0.020412160828709603, + "grad_norm": 0.12393714487552643, + "learning_rate": 2.898e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1450 + }, + { + "epoch": 0.42478906022694213, + "flow_matching_loss": 0.025717341899871828, + "grad_norm": 0.1529090851545334, + "learning_rate": 2.9180000000000002e-05, + "loss": 0.0257, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1460 + }, + { + "epoch": 0.42769857433808556, + "flow_matching_loss": 0.020548769179731606, + "grad_norm": 0.15063682198524475, + "learning_rate": 2.938e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1470 + }, + { + "epoch": 0.430608088449229, + "flow_matching_loss": 0.021907895989716052, + "grad_norm": 0.15320034325122833, + "learning_rate": 2.958e-05, + "loss": 0.0219, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1480 + }, + { + "epoch": 0.4335176025603724, + "flow_matching_loss": 0.02173078302294016, + "grad_norm": 0.19347907602787018, + "learning_rate": 2.9780000000000003e-05, + "loss": 0.0217, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1490 + }, + { + "epoch": 0.43642711667151585, + "flow_matching_loss": 0.020798430033028125, + "grad_norm": 0.18330763280391693, + "learning_rate": 2.998e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1500 + }, + { + "epoch": 0.4393366307826593, + "flow_matching_loss": 0.02285180166363716, + "grad_norm": 0.16106203198432922, + "learning_rate": 2.9999992618302022e-05, + "loss": 0.0229, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1510 + }, + { + "epoch": 0.44224614489380276, + "flow_matching_loss": 0.01902921888977289, + "grad_norm": 0.13432370126247406, + "learning_rate": 2.999996710133069e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1520 + }, + { + "epoch": 0.4451556590049462, + "flow_matching_loss": 0.021931240707635878, + "grad_norm": 0.14586353302001953, + "learning_rate": 2.9999923357984923e-05, + "loss": 0.0219, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1530 + }, + { + "epoch": 0.4480651731160896, + "flow_matching_loss": 0.02247059755027294, + "grad_norm": 0.1815539300441742, + "learning_rate": 2.9999861388317874e-05, + "loss": 0.0225, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1540 + }, + { + "epoch": 0.45097468722723305, + "flow_matching_loss": 0.02291085822507739, + "grad_norm": 0.15305081009864807, + "learning_rate": 2.9999781192404847e-05, + "loss": 0.0229, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1550 + }, + { + "epoch": 0.4538842013383765, + "flow_matching_loss": 0.023928939551115035, + "grad_norm": 0.13521987199783325, + "learning_rate": 2.9999682770343283e-05, + "loss": 0.0239, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1560 + }, + { + "epoch": 0.4567937154495199, + "flow_matching_loss": 0.02371554709970951, + "grad_norm": 0.14803048968315125, + "learning_rate": 2.9999566122252777e-05, + "loss": 0.0237, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1570 + }, + { + "epoch": 0.4597032295606634, + "flow_matching_loss": 0.02157682301476598, + "grad_norm": 0.2209833562374115, + "learning_rate": 2.9999431248275065e-05, + "loss": 0.0216, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1580 + }, + { + "epoch": 0.4626127436718068, + "flow_matching_loss": 0.022071784175932408, + "grad_norm": 0.12957961857318878, + "learning_rate": 2.9999278148574027e-05, + "loss": 0.0221, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1590 + }, + { + "epoch": 0.46552225778295026, + "flow_matching_loss": 0.023543705232441425, + "grad_norm": 0.158281147480011, + "learning_rate": 2.9999106823335705e-05, + "loss": 0.0235, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1600 + }, + { + "epoch": 0.4684317718940937, + "flow_matching_loss": 0.022064955066889526, + "grad_norm": 0.1297406107187271, + "learning_rate": 2.999891727276827e-05, + "loss": 0.0221, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1610 + }, + { + "epoch": 0.4713412860052371, + "flow_matching_loss": 0.020030682533979417, + "grad_norm": 0.12434624135494232, + "learning_rate": 2.9998709497102037e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1620 + }, + { + "epoch": 0.47425080011638054, + "flow_matching_loss": 0.023519768193364144, + "grad_norm": 0.16366934776306152, + "learning_rate": 2.9998483496589485e-05, + "loss": 0.0235, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1630 + }, + { + "epoch": 0.47716031422752403, + "flow_matching_loss": 0.01937630381435156, + "grad_norm": 0.14078906178474426, + "learning_rate": 2.9998239271505217e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1640 + }, + { + "epoch": 0.48006982833866746, + "flow_matching_loss": 0.020832094084471463, + "grad_norm": 0.11857575923204422, + "learning_rate": 2.9997976822145995e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1650 + }, + { + "epoch": 0.4829793424498109, + "flow_matching_loss": 0.01899214219301939, + "grad_norm": 0.13087424635887146, + "learning_rate": 2.9997696148830714e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1660 + }, + { + "epoch": 0.4858888565609543, + "flow_matching_loss": 0.02199583500623703, + "grad_norm": 0.13007161021232605, + "learning_rate": 2.9997397251900423e-05, + "loss": 0.022, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1670 + }, + { + "epoch": 0.48879837067209775, + "flow_matching_loss": 0.02129652528092265, + "grad_norm": 0.13224643468856812, + "learning_rate": 2.999708013171831e-05, + "loss": 0.0213, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1680 + }, + { + "epoch": 0.4917078847832412, + "flow_matching_loss": 0.02493185978382826, + "grad_norm": 0.14975720643997192, + "learning_rate": 2.9996744788669705e-05, + "loss": 0.0249, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1690 + }, + { + "epoch": 0.49461739889438466, + "flow_matching_loss": 0.02526523657143116, + "grad_norm": 0.13280656933784485, + "learning_rate": 2.999639122316208e-05, + "loss": 0.0253, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1700 + }, + { + "epoch": 0.4975269130055281, + "flow_matching_loss": 0.02108570607379079, + "grad_norm": 0.11641683429479599, + "learning_rate": 2.999601943562505e-05, + "loss": 0.0211, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1710 + }, + { + "epoch": 0.5004364271166715, + "flow_matching_loss": 0.02090233061462641, + "grad_norm": 0.20237520337104797, + "learning_rate": 2.9995629426510377e-05, + "loss": 0.0209, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1720 + }, + { + "epoch": 0.503345941227815, + "flow_matching_loss": 0.021247651614248752, + "grad_norm": 0.14620548486709595, + "learning_rate": 2.9995221196291952e-05, + "loss": 0.0212, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1730 + }, + { + "epoch": 0.5062554553389584, + "flow_matching_loss": 0.020271640736609697, + "grad_norm": 0.13686472177505493, + "learning_rate": 2.999479474546582e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1740 + }, + { + "epoch": 0.5091649694501018, + "flow_matching_loss": 0.023515575379133225, + "grad_norm": 0.12365816533565521, + "learning_rate": 2.9994350074550153e-05, + "loss": 0.0235, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1750 + }, + { + "epoch": 0.5120744835612453, + "flow_matching_loss": 0.0204609427601099, + "grad_norm": 0.13997909426689148, + "learning_rate": 2.9993887184085273e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1760 + }, + { + "epoch": 0.5149839976723887, + "flow_matching_loss": 0.023571821860969067, + "grad_norm": 0.17511078715324402, + "learning_rate": 2.9993406074633628e-05, + "loss": 0.0236, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1770 + }, + { + "epoch": 0.5178935117835322, + "flow_matching_loss": 0.02037672223523259, + "grad_norm": 0.14515075087547302, + "learning_rate": 2.9992906746779824e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1780 + }, + { + "epoch": 0.5208030258946756, + "flow_matching_loss": 0.022009742725640536, + "grad_norm": 0.2095533013343811, + "learning_rate": 2.9992389201130576e-05, + "loss": 0.022, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1790 + }, + { + "epoch": 0.523712540005819, + "flow_matching_loss": 0.022732518240809442, + "grad_norm": 0.13079993426799774, + "learning_rate": 2.999185343831476e-05, + "loss": 0.0227, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1800 + }, + { + "epoch": 0.5266220541169625, + "flow_matching_loss": 0.020039835851639508, + "grad_norm": 0.17014791071414948, + "learning_rate": 2.999129945898338e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1810 + }, + { + "epoch": 0.5295315682281059, + "flow_matching_loss": 0.02188774850219488, + "grad_norm": 0.14110754430294037, + "learning_rate": 2.999072726380957e-05, + "loss": 0.0219, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1820 + }, + { + "epoch": 0.5324410823392494, + "flow_matching_loss": 0.019706594571471215, + "grad_norm": 0.17599229514598846, + "learning_rate": 2.9990136853488603e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1830 + }, + { + "epoch": 0.5353505964503927, + "flow_matching_loss": 0.02451040465384722, + "grad_norm": 0.1738787442445755, + "learning_rate": 2.9989528228737878e-05, + "loss": 0.0245, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1840 + }, + { + "epoch": 0.5382601105615362, + "flow_matching_loss": 0.022099800314754246, + "grad_norm": 0.16295179724693298, + "learning_rate": 2.9988901390296935e-05, + "loss": 0.0221, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1850 + }, + { + "epoch": 0.5411696246726797, + "flow_matching_loss": 0.023093668278306723, + "grad_norm": 0.1691749393939972, + "learning_rate": 2.9988256338927442e-05, + "loss": 0.0231, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1860 + }, + { + "epoch": 0.5440791387838231, + "flow_matching_loss": 0.021478073298931123, + "grad_norm": 0.1209549754858017, + "learning_rate": 2.9987593075413198e-05, + "loss": 0.0215, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1870 + }, + { + "epoch": 0.5469886528949666, + "flow_matching_loss": 0.020987271144986153, + "grad_norm": 0.14189954102039337, + "learning_rate": 2.9986911600560128e-05, + "loss": 0.021, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1880 + }, + { + "epoch": 0.5498981670061099, + "flow_matching_loss": 0.023604264203459024, + "grad_norm": 0.11578726768493652, + "learning_rate": 2.998621191519629e-05, + "loss": 0.0236, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1890 + }, + { + "epoch": 0.5528076811172534, + "flow_matching_loss": 0.020255362894386054, + "grad_norm": 0.14984484016895294, + "learning_rate": 2.9985494020171873e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1900 + }, + { + "epoch": 0.5557171952283969, + "flow_matching_loss": 0.02165202545002103, + "grad_norm": 0.10238403081893921, + "learning_rate": 2.9984757916359176e-05, + "loss": 0.0217, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1910 + }, + { + "epoch": 0.5586267093395403, + "flow_matching_loss": 0.021266025211662053, + "grad_norm": 0.13811178505420685, + "learning_rate": 2.998400360465264e-05, + "loss": 0.0213, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1920 + }, + { + "epoch": 0.5615362234506838, + "flow_matching_loss": 0.022728396765887737, + "grad_norm": 0.16254161298274994, + "learning_rate": 2.998323108596883e-05, + "loss": 0.0227, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1930 + }, + { + "epoch": 0.5644457375618271, + "flow_matching_loss": 0.021735641919076443, + "grad_norm": 0.12847159802913666, + "learning_rate": 2.9982440361246422e-05, + "loss": 0.0217, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1940 + }, + { + "epoch": 0.5673552516729706, + "flow_matching_loss": 0.02034592879936099, + "grad_norm": 0.12550145387649536, + "learning_rate": 2.998163143144622e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1950 + }, + { + "epoch": 0.570264765784114, + "flow_matching_loss": 0.021219035144895315, + "grad_norm": 0.20467646420001984, + "learning_rate": 2.9980804297551155e-05, + "loss": 0.0212, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1960 + }, + { + "epoch": 0.5731742798952575, + "flow_matching_loss": 0.022345809638500212, + "grad_norm": 0.1302085965871811, + "learning_rate": 2.997995896056627e-05, + "loss": 0.0223, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1970 + }, + { + "epoch": 0.576083794006401, + "flow_matching_loss": 0.021793764363974334, + "grad_norm": 0.13161998987197876, + "learning_rate": 2.997909542151873e-05, + "loss": 0.0218, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1980 + }, + { + "epoch": 0.5789933081175443, + "flow_matching_loss": 0.0229428518563509, + "grad_norm": 0.15670669078826904, + "learning_rate": 2.9978213681457823e-05, + "loss": 0.0229, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1990 + }, + { + "epoch": 0.5819028222286878, + "flow_matching_loss": 0.02006754232570529, + "grad_norm": 0.14403729140758514, + "learning_rate": 2.9977313741454933e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2000 + }, + { + "epoch": 0.5848123363398312, + "flow_matching_loss": 0.020235104486346245, + "grad_norm": 0.11565768718719482, + "learning_rate": 2.997639560260358e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2010 + }, + { + "epoch": 0.5877218504509747, + "flow_matching_loss": 0.02247671950608492, + "grad_norm": 0.11874642223119736, + "learning_rate": 2.9975459266019387e-05, + "loss": 0.0225, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2020 + }, + { + "epoch": 0.5906313645621182, + "flow_matching_loss": 0.021873071137815715, + "grad_norm": 0.15350204706192017, + "learning_rate": 2.9974504732840092e-05, + "loss": 0.0219, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2030 + }, + { + "epoch": 0.5935408786732616, + "flow_matching_loss": 0.020492740906774998, + "grad_norm": 0.11642514169216156, + "learning_rate": 2.997353200422555e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2040 + }, + { + "epoch": 0.596450392784405, + "flow_matching_loss": 0.02311016507446766, + "grad_norm": 0.16441276669502258, + "learning_rate": 2.99725410813577e-05, + "loss": 0.0231, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2050 + }, + { + "epoch": 0.5993599068955484, + "flow_matching_loss": 0.020867803134024143, + "grad_norm": 0.1289816051721573, + "learning_rate": 2.997153196544063e-05, + "loss": 0.0209, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2060 + }, + { + "epoch": 0.6022694210066919, + "flow_matching_loss": 0.0236697094514966, + "grad_norm": 0.15350182354450226, + "learning_rate": 2.997050465770049e-05, + "loss": 0.0237, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2070 + }, + { + "epoch": 0.6051789351178353, + "flow_matching_loss": 0.018655717093497514, + "grad_norm": 0.16157972812652588, + "learning_rate": 2.9969459159385562e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2080 + }, + { + "epoch": 0.6080884492289788, + "flow_matching_loss": 0.022290799021720886, + "grad_norm": 0.13758738338947296, + "learning_rate": 2.9968395471766237e-05, + "loss": 0.0223, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2090 + }, + { + "epoch": 0.6109979633401222, + "flow_matching_loss": 0.021410304214805366, + "grad_norm": 0.13813166320323944, + "learning_rate": 2.996731359613498e-05, + "loss": 0.0214, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2100 + }, + { + "epoch": 0.6139074774512656, + "flow_matching_loss": 0.018860566429793835, + "grad_norm": 0.09354589879512787, + "learning_rate": 2.996621353380638e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2110 + }, + { + "epoch": 0.6168169915624091, + "flow_matching_loss": 0.02026448491960764, + "grad_norm": 0.1668606698513031, + "learning_rate": 2.9965095286117113e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2120 + }, + { + "epoch": 0.6197265056735525, + "flow_matching_loss": 0.02399362251162529, + "grad_norm": 0.15909512341022491, + "learning_rate": 2.996395885442596e-05, + "loss": 0.024, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2130 + }, + { + "epoch": 0.622636019784696, + "flow_matching_loss": 0.0201635486446321, + "grad_norm": 0.11935274302959442, + "learning_rate": 2.996280424011379e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2140 + }, + { + "epoch": 0.6255455338958394, + "flow_matching_loss": 0.02124578580260277, + "grad_norm": 0.15929892659187317, + "learning_rate": 2.9961631444583565e-05, + "loss": 0.0212, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2150 + }, + { + "epoch": 0.6284550480069828, + "flow_matching_loss": 0.01915343189612031, + "grad_norm": 0.14989565312862396, + "learning_rate": 2.9960440469260347e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2160 + }, + { + "epoch": 0.6313645621181263, + "flow_matching_loss": 0.018989718053489923, + "grad_norm": 0.20345935225486755, + "learning_rate": 2.995923131559128e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2170 + }, + { + "epoch": 0.6342740762292697, + "flow_matching_loss": 0.0203954896889627, + "grad_norm": 0.22839701175689697, + "learning_rate": 2.99580039850456e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2180 + }, + { + "epoch": 0.6371835903404132, + "flow_matching_loss": 0.01964216437190771, + "grad_norm": 0.15754613280296326, + "learning_rate": 2.9956758479114635e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2190 + }, + { + "epoch": 0.6400931044515565, + "flow_matching_loss": 0.02545248279348016, + "grad_norm": 0.20317167043685913, + "learning_rate": 2.9955494799311787e-05, + "loss": 0.0255, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2200 + }, + { + "epoch": 0.6430026185627, + "flow_matching_loss": 0.02222902700304985, + "grad_norm": 0.17232325673103333, + "learning_rate": 2.995421294717254e-05, + "loss": 0.0222, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2210 + }, + { + "epoch": 0.6459121326738435, + "flow_matching_loss": 0.02296419097110629, + "grad_norm": 0.15567393600940704, + "learning_rate": 2.995291292425448e-05, + "loss": 0.023, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2220 + }, + { + "epoch": 0.6488216467849869, + "flow_matching_loss": 0.02026752345263958, + "grad_norm": 0.13663436472415924, + "learning_rate": 2.9951594732137248e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2230 + }, + { + "epoch": 0.6517311608961304, + "flow_matching_loss": 0.02085584159940481, + "grad_norm": 0.10754940658807755, + "learning_rate": 2.995025837242257e-05, + "loss": 0.0209, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2240 + }, + { + "epoch": 0.6546406750072737, + "flow_matching_loss": 0.021514910366386177, + "grad_norm": 0.15662507712841034, + "learning_rate": 2.994890384673426e-05, + "loss": 0.0215, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2250 + }, + { + "epoch": 0.6575501891184172, + "flow_matching_loss": 0.022080511040985584, + "grad_norm": 0.12468431144952774, + "learning_rate": 2.9947531156718174e-05, + "loss": 0.0221, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2260 + }, + { + "epoch": 0.6604597032295607, + "flow_matching_loss": 0.019860007148236037, + "grad_norm": 0.1430625021457672, + "learning_rate": 2.994614030404228e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2270 + }, + { + "epoch": 0.6633692173407041, + "flow_matching_loss": 0.018129666149616242, + "grad_norm": 0.13728761672973633, + "learning_rate": 2.994473129039659e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2280 + }, + { + "epoch": 0.6662787314518476, + "flow_matching_loss": 0.020138048753142356, + "grad_norm": 0.126656174659729, + "learning_rate": 2.9943304117493183e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2290 + }, + { + "epoch": 0.669188245562991, + "flow_matching_loss": 0.020590228587388994, + "grad_norm": 0.17060840129852295, + "learning_rate": 2.9941858787066207e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2300 + }, + { + "epoch": 0.6720977596741344, + "flow_matching_loss": 0.021117928437888622, + "grad_norm": 0.13715219497680664, + "learning_rate": 2.9940395300871888e-05, + "loss": 0.0211, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2310 + }, + { + "epoch": 0.6750072737852778, + "flow_matching_loss": 0.021206411346793175, + "grad_norm": 0.2067960649728775, + "learning_rate": 2.993891366068849e-05, + "loss": 0.0212, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2320 + }, + { + "epoch": 0.6779167878964213, + "flow_matching_loss": 0.02503648540005088, + "grad_norm": 0.14835835993289948, + "learning_rate": 2.993741386831634e-05, + "loss": 0.025, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2330 + }, + { + "epoch": 0.6808263020075648, + "flow_matching_loss": 0.019045776780694725, + "grad_norm": 0.1208658292889595, + "learning_rate": 2.9935895925577837e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2340 + }, + { + "epoch": 0.6837358161187082, + "flow_matching_loss": 0.024493254255503416, + "grad_norm": 0.12565602362155914, + "learning_rate": 2.9934359834317425e-05, + "loss": 0.0245, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2350 + }, + { + "epoch": 0.6866453302298516, + "flow_matching_loss": 0.021724103949964046, + "grad_norm": 0.16723789274692535, + "learning_rate": 2.993280559640159e-05, + "loss": 0.0217, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2360 + }, + { + "epoch": 0.689554844340995, + "flow_matching_loss": 0.020020887814462186, + "grad_norm": 0.19413404166698456, + "learning_rate": 2.9931233213718887e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2370 + }, + { + "epoch": 0.6924643584521385, + "flow_matching_loss": 0.020381979830563068, + "grad_norm": 0.16622212529182434, + "learning_rate": 2.992964268817991e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2380 + }, + { + "epoch": 0.695373872563282, + "flow_matching_loss": 0.020980837102979422, + "grad_norm": 0.09416759759187698, + "learning_rate": 2.9928034021717286e-05, + "loss": 0.021, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2390 + }, + { + "epoch": 0.6982833866744254, + "flow_matching_loss": 0.021069782599806786, + "grad_norm": 0.11358066648244858, + "learning_rate": 2.9926407216285707e-05, + "loss": 0.0211, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2400 + }, + { + "epoch": 0.7011929007855688, + "flow_matching_loss": 0.01833305452018976, + "grad_norm": 0.15401336550712585, + "learning_rate": 2.9924762273861894e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2410 + }, + { + "epoch": 0.7041024148967122, + "flow_matching_loss": 0.025914757139980794, + "grad_norm": 0.08981619030237198, + "learning_rate": 2.992309919644461e-05, + "loss": 0.0259, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2420 + }, + { + "epoch": 0.7070119290078557, + "flow_matching_loss": 0.022039265278726815, + "grad_norm": 0.1530817747116089, + "learning_rate": 2.9921417986054633e-05, + "loss": 0.022, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2430 + }, + { + "epoch": 0.7099214431189991, + "flow_matching_loss": 0.020719315856695175, + "grad_norm": 0.12303377687931061, + "learning_rate": 2.9919718644734815e-05, + "loss": 0.0207, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2440 + }, + { + "epoch": 0.7128309572301426, + "flow_matching_loss": 0.021972264070063828, + "grad_norm": 0.19529691338539124, + "learning_rate": 2.991800117455e-05, + "loss": 0.022, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2450 + }, + { + "epoch": 0.715740471341286, + "flow_matching_loss": 0.020045795291662217, + "grad_norm": 0.17893297970294952, + "learning_rate": 2.9916265577587085e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2460 + }, + { + "epoch": 0.7186499854524294, + "flow_matching_loss": 0.02092741709202528, + "grad_norm": 0.16204465925693512, + "learning_rate": 2.9914511855954986e-05, + "loss": 0.0209, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2470 + }, + { + "epoch": 0.7215594995635729, + "flow_matching_loss": 0.018655744008719922, + "grad_norm": 0.11348774284124374, + "learning_rate": 2.9912740011784626e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2480 + }, + { + "epoch": 0.7244690136747163, + "flow_matching_loss": 0.024263223446905614, + "grad_norm": 0.1605375111103058, + "learning_rate": 2.991095004722898e-05, + "loss": 0.0243, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2490 + }, + { + "epoch": 0.7273785277858598, + "flow_matching_loss": 0.021514955908060074, + "grad_norm": 0.15240730345249176, + "learning_rate": 2.9909141964463017e-05, + "loss": 0.0215, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2500 + }, + { + "epoch": 0.7302880418970032, + "flow_matching_loss": 0.02233523866161704, + "grad_norm": 0.12396831065416336, + "learning_rate": 2.990731576568372e-05, + "loss": 0.0223, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2510 + }, + { + "epoch": 0.7331975560081466, + "flow_matching_loss": 0.02118651382625103, + "grad_norm": 0.14014463126659393, + "learning_rate": 2.99054714531101e-05, + "loss": 0.0212, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2520 + }, + { + "epoch": 0.7361070701192901, + "flow_matching_loss": 0.021331666875630616, + "grad_norm": 0.1123218983411789, + "learning_rate": 2.9903609028983178e-05, + "loss": 0.0213, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2530 + }, + { + "epoch": 0.7390165842304335, + "flow_matching_loss": 0.018901433050632476, + "grad_norm": 0.14786991477012634, + "learning_rate": 2.990172849556596e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2540 + }, + { + "epoch": 0.741926098341577, + "flow_matching_loss": 0.022340271808207034, + "grad_norm": 0.14928804337978363, + "learning_rate": 2.989982985514348e-05, + "loss": 0.0223, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2550 + }, + { + "epoch": 0.7448356124527203, + "flow_matching_loss": 0.023524999152868986, + "grad_norm": 0.17094506323337555, + "learning_rate": 2.9897913110022766e-05, + "loss": 0.0235, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2560 + }, + { + "epoch": 0.7477451265638638, + "flow_matching_loss": 0.018899698089808226, + "grad_norm": 0.14348086714744568, + "learning_rate": 2.989597826253284e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2570 + }, + { + "epoch": 0.7506546406750073, + "flow_matching_loss": 0.021146958880126476, + "grad_norm": 0.1792125105857849, + "learning_rate": 2.989402531502473e-05, + "loss": 0.0211, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2580 + }, + { + "epoch": 0.7535641547861507, + "flow_matching_loss": 0.018091379757970573, + "grad_norm": 0.14535175263881683, + "learning_rate": 2.989205426987144e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2590 + }, + { + "epoch": 0.7564736688972942, + "flow_matching_loss": 0.023184876702725888, + "grad_norm": 0.1466989517211914, + "learning_rate": 2.9890065129467988e-05, + "loss": 0.0232, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2600 + }, + { + "epoch": 0.7593831830084375, + "flow_matching_loss": 0.020928794890642165, + "grad_norm": 0.14920556545257568, + "learning_rate": 2.9888057896231364e-05, + "loss": 0.0209, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2610 + }, + { + "epoch": 0.762292697119581, + "flow_matching_loss": 0.021490638703107835, + "grad_norm": 0.14517757296562195, + "learning_rate": 2.988603257260054e-05, + "loss": 0.0215, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2620 + }, + { + "epoch": 0.7652022112307245, + "flow_matching_loss": 0.02009591059759259, + "grad_norm": 0.11595700681209564, + "learning_rate": 2.988398916103648e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2630 + }, + { + "epoch": 0.7681117253418679, + "flow_matching_loss": 0.021099250949919223, + "grad_norm": 0.1229499876499176, + "learning_rate": 2.988192766402212e-05, + "loss": 0.0211, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2640 + }, + { + "epoch": 0.7710212394530114, + "flow_matching_loss": 0.02253248393535614, + "grad_norm": 0.23053398728370667, + "learning_rate": 2.9879848084062378e-05, + "loss": 0.0225, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2650 + }, + { + "epoch": 0.7739307535641547, + "flow_matching_loss": 0.020950458012521267, + "grad_norm": 0.10988275706768036, + "learning_rate": 2.9877750423684132e-05, + "loss": 0.021, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2660 + }, + { + "epoch": 0.7768402676752982, + "flow_matching_loss": 0.021244869288057088, + "grad_norm": 0.14633390307426453, + "learning_rate": 2.987563468543624e-05, + "loss": 0.0212, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2670 + }, + { + "epoch": 0.7797497817864416, + "flow_matching_loss": 0.02237392859533429, + "grad_norm": 0.15412537753582, + "learning_rate": 2.9873500871889533e-05, + "loss": 0.0224, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2680 + }, + { + "epoch": 0.7826592958975851, + "flow_matching_loss": 0.022243255004286765, + "grad_norm": 0.12816695868968964, + "learning_rate": 2.9871348985636778e-05, + "loss": 0.0222, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2690 + }, + { + "epoch": 0.7855688100087286, + "flow_matching_loss": 0.021288671158254148, + "grad_norm": 0.1895957738161087, + "learning_rate": 2.986917902929273e-05, + "loss": 0.0213, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2700 + }, + { + "epoch": 0.788478324119872, + "flow_matching_loss": 0.02046503983438015, + "grad_norm": 0.11689028143882751, + "learning_rate": 2.9866991005494095e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2710 + }, + { + "epoch": 0.7913878382310154, + "flow_matching_loss": 0.02033822741359472, + "grad_norm": 0.17702433466911316, + "learning_rate": 2.986478491689952e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2720 + }, + { + "epoch": 0.7942973523421588, + "flow_matching_loss": 0.0205597722902894, + "grad_norm": 0.11141970753669739, + "learning_rate": 2.986256076618961e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2730 + }, + { + "epoch": 0.7972068664533023, + "flow_matching_loss": 0.019387800153344868, + "grad_norm": 0.13355182111263275, + "learning_rate": 2.986031855606692e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2740 + }, + { + "epoch": 0.8001163805644458, + "flow_matching_loss": 0.01762328380718827, + "grad_norm": 0.11869662255048752, + "learning_rate": 2.9858058289255954e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2750 + }, + { + "epoch": 0.8030258946755892, + "flow_matching_loss": 0.019221416488289834, + "grad_norm": 0.1435195952653885, + "learning_rate": 2.985577996850314e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2760 + }, + { + "epoch": 0.8059354087867326, + "flow_matching_loss": 0.021838541235774757, + "grad_norm": 0.11328020691871643, + "learning_rate": 2.9853483596576856e-05, + "loss": 0.0218, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2770 + }, + { + "epoch": 0.808844922897876, + "flow_matching_loss": 0.0188876261934638, + "grad_norm": 0.15067404508590698, + "learning_rate": 2.9851169176267412e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2780 + }, + { + "epoch": 0.8117544370090195, + "flow_matching_loss": 0.020593919418752195, + "grad_norm": 0.14279307425022125, + "learning_rate": 2.9848836710387044e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2790 + }, + { + "epoch": 0.814663951120163, + "flow_matching_loss": 0.020433297753334044, + "grad_norm": 0.1693989783525467, + "learning_rate": 2.9846486201769917e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2800 + }, + { + "epoch": 0.8175734652313064, + "flow_matching_loss": 0.023273638822138308, + "grad_norm": 0.1416156142950058, + "learning_rate": 2.9844117653272122e-05, + "loss": 0.0233, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2810 + }, + { + "epoch": 0.8204829793424498, + "flow_matching_loss": 0.02148872260004282, + "grad_norm": 0.16504934430122375, + "learning_rate": 2.9841731067771676e-05, + "loss": 0.0215, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2820 + }, + { + "epoch": 0.8233924934535932, + "flow_matching_loss": 0.018802146147936582, + "grad_norm": 0.10215882211923599, + "learning_rate": 2.9839326448168492e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2830 + }, + { + "epoch": 0.8263020075647367, + "flow_matching_loss": 0.01945648221299052, + "grad_norm": 0.0979677215218544, + "learning_rate": 2.983690379738442e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2840 + }, + { + "epoch": 0.8292115216758801, + "flow_matching_loss": 0.019712160527706146, + "grad_norm": 0.09688778221607208, + "learning_rate": 2.9834463118363204e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2850 + }, + { + "epoch": 0.8321210357870236, + "flow_matching_loss": 0.023611902631819247, + "grad_norm": 0.1286144107580185, + "learning_rate": 2.98320044140705e-05, + "loss": 0.0236, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2860 + }, + { + "epoch": 0.835030549898167, + "flow_matching_loss": 0.020526218600571154, + "grad_norm": 0.1090836450457573, + "learning_rate": 2.9829527687493868e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2870 + }, + { + "epoch": 0.8379400640093104, + "flow_matching_loss": 0.019920181669294833, + "grad_norm": 0.09602289646863937, + "learning_rate": 2.9827032941642765e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2880 + }, + { + "epoch": 0.8408495781204539, + "flow_matching_loss": 0.021748722158372404, + "grad_norm": 0.1438419222831726, + "learning_rate": 2.9824520179548535e-05, + "loss": 0.0217, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2890 + }, + { + "epoch": 0.8437590922315973, + "flow_matching_loss": 0.020381800085306167, + "grad_norm": 0.10709547251462936, + "learning_rate": 2.9821989404264428e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2900 + }, + { + "epoch": 0.8466686063427408, + "flow_matching_loss": 0.01932899858802557, + "grad_norm": 0.16418059170246124, + "learning_rate": 2.9819440618865572e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2910 + }, + { + "epoch": 0.8495781204538843, + "flow_matching_loss": 0.022586369421333075, + "grad_norm": 0.13189060986042023, + "learning_rate": 2.981687382644898e-05, + "loss": 0.0226, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2920 + }, + { + "epoch": 0.8524876345650276, + "flow_matching_loss": 0.021724587865173818, + "grad_norm": 0.10292370617389679, + "learning_rate": 2.9814289030133556e-05, + "loss": 0.0217, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2930 + }, + { + "epoch": 0.8553971486761711, + "flow_matching_loss": 0.023768472857773303, + "grad_norm": 0.17424839735031128, + "learning_rate": 2.9811686233060058e-05, + "loss": 0.0238, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2940 + }, + { + "epoch": 0.8583066627873145, + "flow_matching_loss": 0.019892978109419347, + "grad_norm": 0.1525665819644928, + "learning_rate": 2.9809065438391138e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2950 + }, + { + "epoch": 0.861216176898458, + "flow_matching_loss": 0.020516314078122377, + "grad_norm": 0.13506649434566498, + "learning_rate": 2.9806426649311304e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2960 + }, + { + "epoch": 0.8641256910096013, + "flow_matching_loss": 0.020059478655457496, + "grad_norm": 0.13228435814380646, + "learning_rate": 2.980376986902694e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2970 + }, + { + "epoch": 0.8670352051207448, + "flow_matching_loss": 0.019994856789708138, + "grad_norm": 0.13561752438545227, + "learning_rate": 2.9801095100766276e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2980 + }, + { + "epoch": 0.8699447192318883, + "flow_matching_loss": 0.018456903751939535, + "grad_norm": 0.187338724732399, + "learning_rate": 2.979840234777942e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2990 + }, + { + "epoch": 0.8728542333430317, + "flow_matching_loss": 0.018692684452980758, + "grad_norm": 0.1214597299695015, + "learning_rate": 2.979569161333831e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3000 + }, + { + "epoch": 0.8757637474541752, + "flow_matching_loss": 0.01953063430264592, + "grad_norm": 0.16285152733325958, + "learning_rate": 2.9792962900736744e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3010 + }, + { + "epoch": 0.8786732615653186, + "flow_matching_loss": 0.020422083884477617, + "grad_norm": 0.12688519060611725, + "learning_rate": 2.9790216213290372e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3020 + }, + { + "epoch": 0.881582775676462, + "flow_matching_loss": 0.017397477757185698, + "grad_norm": 0.09373781830072403, + "learning_rate": 2.9787451554336672e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3030 + }, + { + "epoch": 0.8844922897876055, + "flow_matching_loss": 0.021014456823468208, + "grad_norm": 0.13910874724388123, + "learning_rate": 2.9784668927234972e-05, + "loss": 0.021, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3040 + }, + { + "epoch": 0.8874018038987489, + "flow_matching_loss": 0.018149933964014053, + "grad_norm": 0.15231499075889587, + "learning_rate": 2.9781868335366418e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3050 + }, + { + "epoch": 0.8903113180098924, + "flow_matching_loss": 0.019832478370517492, + "grad_norm": 0.15695752203464508, + "learning_rate": 2.9779049782134e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3060 + }, + { + "epoch": 0.8932208321210358, + "flow_matching_loss": 0.019459139835089446, + "grad_norm": 0.15979814529418945, + "learning_rate": 2.9776213270962523e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3070 + }, + { + "epoch": 0.8961303462321792, + "flow_matching_loss": 0.02011569542810321, + "grad_norm": 0.1659284383058548, + "learning_rate": 2.977335880529861e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3080 + }, + { + "epoch": 0.8990398603433226, + "flow_matching_loss": 0.020429057721048593, + "grad_norm": 0.1249007135629654, + "learning_rate": 2.977048638861072e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3090 + }, + { + "epoch": 0.9019493744544661, + "flow_matching_loss": 0.02125987159088254, + "grad_norm": 0.08941806852817535, + "learning_rate": 2.97675960243891e-05, + "loss": 0.0213, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3100 + }, + { + "epoch": 0.9048588885656096, + "flow_matching_loss": 0.019485225435346365, + "grad_norm": 0.10928992927074432, + "learning_rate": 2.976468771614582e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3110 + }, + { + "epoch": 0.907768402676753, + "flow_matching_loss": 0.020080822985619307, + "grad_norm": 0.1311078518629074, + "learning_rate": 2.9761761467414736e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3120 + }, + { + "epoch": 0.9106779167878964, + "flow_matching_loss": 0.021977593004703522, + "grad_norm": 0.1041993647813797, + "learning_rate": 2.9758817281751538e-05, + "loss": 0.022, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3130 + }, + { + "epoch": 0.9135874308990398, + "flow_matching_loss": 0.020318437833338976, + "grad_norm": 0.1284831166267395, + "learning_rate": 2.975585516273367e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3140 + }, + { + "epoch": 0.9164969450101833, + "flow_matching_loss": 0.01916202763095498, + "grad_norm": 0.11871717125177383, + "learning_rate": 2.975287511396039e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3150 + }, + { + "epoch": 0.9194064591213268, + "flow_matching_loss": 0.01866158451884985, + "grad_norm": 0.10943996161222458, + "learning_rate": 2.9749877139052745e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3160 + }, + { + "epoch": 0.9223159732324702, + "flow_matching_loss": 0.017674814630299806, + "grad_norm": 0.08927900344133377, + "learning_rate": 2.9746861241653555e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3170 + }, + { + "epoch": 0.9252254873436137, + "flow_matching_loss": 0.01906758304685354, + "grad_norm": 0.11900405585765839, + "learning_rate": 2.9743827425427416e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3180 + }, + { + "epoch": 0.928135001454757, + "flow_matching_loss": 0.02028451403602958, + "grad_norm": 0.14512042701244354, + "learning_rate": 2.9740775694060704e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3190 + }, + { + "epoch": 0.9310445155659005, + "flow_matching_loss": 0.019958041980862616, + "grad_norm": 0.11058254539966583, + "learning_rate": 2.973770605126156e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3200 + }, + { + "epoch": 0.9339540296770439, + "flow_matching_loss": 0.020238740276545285, + "grad_norm": 0.11448169499635696, + "learning_rate": 2.9734618500759886e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3210 + }, + { + "epoch": 0.9368635437881874, + "flow_matching_loss": 0.018700968008488417, + "grad_norm": 0.16157980263233185, + "learning_rate": 2.9731513046307356e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3220 + }, + { + "epoch": 0.9397730578993309, + "flow_matching_loss": 0.016944552958011627, + "grad_norm": 0.11694961041212082, + "learning_rate": 2.972838969167738e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3230 + }, + { + "epoch": 0.9426825720104742, + "flow_matching_loss": 0.020207651983946563, + "grad_norm": 0.10977567732334137, + "learning_rate": 2.972524844066513e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3240 + }, + { + "epoch": 0.9455920861216177, + "flow_matching_loss": 0.019841692596673965, + "grad_norm": 0.13688714802265167, + "learning_rate": 2.9722089297087524e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3250 + }, + { + "epoch": 0.9485016002327611, + "flow_matching_loss": 0.021290661115199328, + "grad_norm": 0.2145812064409256, + "learning_rate": 2.971891226478322e-05, + "loss": 0.0213, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3260 + }, + { + "epoch": 0.9514111143439046, + "flow_matching_loss": 0.0191860256716609, + "grad_norm": 0.09808003902435303, + "learning_rate": 2.9715717347612615e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3270 + }, + { + "epoch": 0.9543206284550481, + "flow_matching_loss": 0.02020351653918624, + "grad_norm": 0.10663800686597824, + "learning_rate": 2.971250454945783e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3280 + }, + { + "epoch": 0.9572301425661914, + "flow_matching_loss": 0.02531038299202919, + "grad_norm": 0.11563810706138611, + "learning_rate": 2.970927387422272e-05, + "loss": 0.0253, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3290 + }, + { + "epoch": 0.9601396566773349, + "flow_matching_loss": 0.018790072202682494, + "grad_norm": 0.1452322155237198, + "learning_rate": 2.9706025325832858e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3300 + }, + { + "epoch": 0.9630491707884783, + "flow_matching_loss": 0.019478535279631615, + "grad_norm": 0.12236349284648895, + "learning_rate": 2.9702758908235534e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3310 + }, + { + "epoch": 0.9659586848996218, + "flow_matching_loss": 0.021755362674593926, + "grad_norm": 0.14682868123054504, + "learning_rate": 2.9699474625399765e-05, + "loss": 0.0218, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3320 + }, + { + "epoch": 0.9688681990107652, + "flow_matching_loss": 0.01903050011023879, + "grad_norm": 0.11524717509746552, + "learning_rate": 2.9696172481316246e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3330 + }, + { + "epoch": 0.9717777131219086, + "flow_matching_loss": 0.01952268844470382, + "grad_norm": 0.1307864785194397, + "learning_rate": 2.9692852479997407e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3340 + }, + { + "epoch": 0.9746872272330521, + "flow_matching_loss": 0.019781590532511474, + "grad_norm": 0.11583775281906128, + "learning_rate": 2.9689514625477364e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3350 + }, + { + "epoch": 0.9775967413441955, + "flow_matching_loss": 0.019400075636804105, + "grad_norm": 0.11487513035535812, + "learning_rate": 2.968615892181192e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3360 + }, + { + "epoch": 0.980506255455339, + "flow_matching_loss": 0.02086751237511635, + "grad_norm": 0.14031338691711426, + "learning_rate": 2.9682785373078567e-05, + "loss": 0.0209, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3370 + }, + { + "epoch": 0.9834157695664824, + "flow_matching_loss": 0.018048079777508975, + "grad_norm": 0.15604496002197266, + "learning_rate": 2.9679393983376493e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3380 + }, + { + "epoch": 0.9863252836776258, + "flow_matching_loss": 0.017073636781424283, + "grad_norm": 0.10981956869363785, + "learning_rate": 2.9675984756826545e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3390 + }, + { + "epoch": 0.9892347977887693, + "flow_matching_loss": 0.022145511209964754, + "grad_norm": 0.12312600016593933, + "learning_rate": 2.9672557697571273e-05, + "loss": 0.0221, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3400 + }, + { + "epoch": 0.9921443118999127, + "flow_matching_loss": 0.01717163771390915, + "grad_norm": 0.12287737429141998, + "learning_rate": 2.966911280977486e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3410 + }, + { + "epoch": 0.9950538260110562, + "flow_matching_loss": 0.0161151017062366, + "grad_norm": 0.1094001978635788, + "learning_rate": 2.966565009762318e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3420 + }, + { + "epoch": 0.9979633401221996, + "flow_matching_loss": 0.020848034881055355, + "grad_norm": 0.151205375790596, + "learning_rate": 2.966216956532375e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3430 + }, + { + "epoch": 1.000872854233343, + "flow_matching_loss": 0.0232934458181262, + "grad_norm": 0.14234405755996704, + "learning_rate": 2.9658671217105744e-05, + "loss": 0.0233, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3440 + }, + { + "epoch": 1.0037823683444864, + "flow_matching_loss": 0.020832740142941474, + "grad_norm": 0.1396530717611313, + "learning_rate": 2.965515505721999e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3450 + }, + { + "epoch": 1.00669188245563, + "flow_matching_loss": 0.022671780455857514, + "grad_norm": 0.14545868337154388, + "learning_rate": 2.9651621089938948e-05, + "loss": 0.0227, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3460 + }, + { + "epoch": 1.0096013965667734, + "flow_matching_loss": 0.018209088966250418, + "grad_norm": 0.12708811461925507, + "learning_rate": 2.9648069319556725e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3470 + }, + { + "epoch": 1.0125109106779169, + "flow_matching_loss": 0.02079339399933815, + "grad_norm": 0.12868502736091614, + "learning_rate": 2.9644499750389057e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3480 + }, + { + "epoch": 1.0154204247890601, + "flow_matching_loss": 0.018977133091539145, + "grad_norm": 0.14295896887779236, + "learning_rate": 2.9640912386773305e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3490 + }, + { + "epoch": 1.0183299389002036, + "flow_matching_loss": 0.018898451793938877, + "grad_norm": 0.13518980145454407, + "learning_rate": 2.9637307233068455e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3500 + }, + { + "epoch": 1.021239453011347, + "flow_matching_loss": 0.01905636815354228, + "grad_norm": 0.10575452446937561, + "learning_rate": 2.9633684293655108e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3510 + }, + { + "epoch": 1.0241489671224906, + "flow_matching_loss": 0.021478781290352345, + "grad_norm": 0.1210121139883995, + "learning_rate": 2.963004357293548e-05, + "loss": 0.0215, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3520 + }, + { + "epoch": 1.027058481233634, + "flow_matching_loss": 0.02171181347221136, + "grad_norm": 0.10813646018505096, + "learning_rate": 2.962638507533339e-05, + "loss": 0.0217, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3530 + }, + { + "epoch": 1.0299679953447773, + "flow_matching_loss": 0.01943965209648013, + "grad_norm": 0.09064453095197678, + "learning_rate": 2.962270880529425e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3540 + }, + { + "epoch": 1.0328775094559208, + "flow_matching_loss": 0.020593015011399984, + "grad_norm": 0.10025422275066376, + "learning_rate": 2.9619014767285084e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3550 + }, + { + "epoch": 1.0357870235670643, + "flow_matching_loss": 0.01893206937238574, + "grad_norm": 0.1261240839958191, + "learning_rate": 2.9615302965794488e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3560 + }, + { + "epoch": 1.0386965376782078, + "flow_matching_loss": 0.01960571752861142, + "grad_norm": 0.14627711474895477, + "learning_rate": 2.961157340533266e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3570 + }, + { + "epoch": 1.0416060517893513, + "flow_matching_loss": 0.022046187613159417, + "grad_norm": 0.11756478250026703, + "learning_rate": 2.9607826090431364e-05, + "loss": 0.022, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3580 + }, + { + "epoch": 1.0445155659004945, + "flow_matching_loss": 0.02065265765413642, + "grad_norm": 0.13326922059059143, + "learning_rate": 2.9604061025643936e-05, + "loss": 0.0207, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3590 + }, + { + "epoch": 1.047425080011638, + "flow_matching_loss": 0.020617244113236666, + "grad_norm": 0.10782354325056076, + "learning_rate": 2.9600278215545292e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3600 + }, + { + "epoch": 1.0503345941227815, + "flow_matching_loss": 0.01872306801378727, + "grad_norm": 0.14750133454799652, + "learning_rate": 2.95964776647319e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3610 + }, + { + "epoch": 1.053244108233925, + "flow_matching_loss": 0.018100361805409193, + "grad_norm": 0.11138572543859482, + "learning_rate": 2.9592659377821787e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3620 + }, + { + "epoch": 1.0561536223450685, + "flow_matching_loss": 0.018203713465481997, + "grad_norm": 0.13820771872997284, + "learning_rate": 2.9588823359454528e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3630 + }, + { + "epoch": 1.0590631364562118, + "flow_matching_loss": 0.019488974846899508, + "grad_norm": 0.1093984916806221, + "learning_rate": 2.958496961429126e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3640 + }, + { + "epoch": 1.0619726505673552, + "flow_matching_loss": 0.02180482614785433, + "grad_norm": 0.15499867498874664, + "learning_rate": 2.9581098147014628e-05, + "loss": 0.0218, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3650 + }, + { + "epoch": 1.0648821646784987, + "flow_matching_loss": 0.020217375084757804, + "grad_norm": 0.10867831110954285, + "learning_rate": 2.9577208962328843e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3660 + }, + { + "epoch": 1.0677916787896422, + "flow_matching_loss": 0.019563346449285746, + "grad_norm": 0.09984540939331055, + "learning_rate": 2.9573302064959633e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3670 + }, + { + "epoch": 1.0707011929007855, + "flow_matching_loss": 0.018501034658402204, + "grad_norm": 0.1080268993973732, + "learning_rate": 2.9569377459654235e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3680 + }, + { + "epoch": 1.073610707011929, + "flow_matching_loss": 0.020160446316003798, + "grad_norm": 0.13288435339927673, + "learning_rate": 2.9565435151181424e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3690 + }, + { + "epoch": 1.0765202211230724, + "flow_matching_loss": 0.019586923718452453, + "grad_norm": 0.12863220274448395, + "learning_rate": 2.9561475144331473e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3700 + }, + { + "epoch": 1.079429735234216, + "flow_matching_loss": 0.020003524143248796, + "grad_norm": 0.11212433874607086, + "learning_rate": 2.9557497443916162e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3710 + }, + { + "epoch": 1.0823392493453594, + "flow_matching_loss": 0.021355896722525358, + "grad_norm": 0.11464273184537888, + "learning_rate": 2.9553502054768774e-05, + "loss": 0.0214, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3720 + }, + { + "epoch": 1.0852487634565027, + "flow_matching_loss": 0.01978323683142662, + "grad_norm": 0.09873249381780624, + "learning_rate": 2.954948898174409e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3730 + }, + { + "epoch": 1.0881582775676462, + "flow_matching_loss": 0.019663280248641966, + "grad_norm": 0.11640917509794235, + "learning_rate": 2.954545822971836e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3740 + }, + { + "epoch": 1.0910677916787896, + "flow_matching_loss": 0.021951702050864698, + "grad_norm": 0.09276483952999115, + "learning_rate": 2.9541409803589335e-05, + "loss": 0.022, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3750 + }, + { + "epoch": 1.0939773057899331, + "flow_matching_loss": 0.021807590313255788, + "grad_norm": 0.15656784176826477, + "learning_rate": 2.9537343708276235e-05, + "loss": 0.0218, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3760 + }, + { + "epoch": 1.0968868199010766, + "flow_matching_loss": 0.019357286300510167, + "grad_norm": 0.10243184864521027, + "learning_rate": 2.953325994871975e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3770 + }, + { + "epoch": 1.0997963340122199, + "flow_matching_loss": 0.018047819565981626, + "grad_norm": 0.11579737067222595, + "learning_rate": 2.9529158529882023e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3780 + }, + { + "epoch": 1.1027058481233634, + "flow_matching_loss": 0.020540405996143818, + "grad_norm": 0.12591861188411713, + "learning_rate": 2.9525039456746677e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3790 + }, + { + "epoch": 1.1056153622345068, + "flow_matching_loss": 0.020707310922443867, + "grad_norm": 0.12741418182849884, + "learning_rate": 2.952090273431877e-05, + "loss": 0.0207, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3800 + }, + { + "epoch": 1.1085248763456503, + "flow_matching_loss": 0.0181847607716918, + "grad_norm": 0.1309894323348999, + "learning_rate": 2.9516748367624808e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3810 + }, + { + "epoch": 1.1114343904567938, + "flow_matching_loss": 0.019265200197696685, + "grad_norm": 0.1425621658563614, + "learning_rate": 2.951257636171274e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3820 + }, + { + "epoch": 1.114343904567937, + "flow_matching_loss": 0.02000597510486841, + "grad_norm": 0.13353976607322693, + "learning_rate": 2.9508386721651944e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3830 + }, + { + "epoch": 1.1172534186790806, + "flow_matching_loss": 0.019608167093247177, + "grad_norm": 0.13634268939495087, + "learning_rate": 2.9504179452533228e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3840 + }, + { + "epoch": 1.120162932790224, + "flow_matching_loss": 0.018607161939144135, + "grad_norm": 0.14058677852153778, + "learning_rate": 2.9499954559468826e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3850 + }, + { + "epoch": 1.1230724469013675, + "flow_matching_loss": 0.020137833990156652, + "grad_norm": 0.11448176950216293, + "learning_rate": 2.9495712047592378e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3860 + }, + { + "epoch": 1.1259819610125108, + "flow_matching_loss": 0.021158502064645292, + "grad_norm": 0.10674801468849182, + "learning_rate": 2.949145192205893e-05, + "loss": 0.0212, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3870 + }, + { + "epoch": 1.1288914751236543, + "flow_matching_loss": 0.018845943082123996, + "grad_norm": 0.16949953138828278, + "learning_rate": 2.9487174188044953e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3880 + }, + { + "epoch": 1.1318009892347978, + "flow_matching_loss": 0.018683666829019785, + "grad_norm": 0.13859093189239502, + "learning_rate": 2.9482878850748276e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3890 + }, + { + "epoch": 1.1347105033459413, + "flow_matching_loss": 0.01973863448947668, + "grad_norm": 0.13878273963928223, + "learning_rate": 2.9478565915388157e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3900 + }, + { + "epoch": 1.1376200174570847, + "flow_matching_loss": 0.02186323022469878, + "grad_norm": 0.1285182386636734, + "learning_rate": 2.9474235387205207e-05, + "loss": 0.0219, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3910 + }, + { + "epoch": 1.140529531568228, + "flow_matching_loss": 0.022561856545507908, + "grad_norm": 0.18367153406143188, + "learning_rate": 2.946988727146143e-05, + "loss": 0.0226, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3920 + }, + { + "epoch": 1.1434390456793715, + "flow_matching_loss": 0.01756394309923053, + "grad_norm": 0.1205090656876564, + "learning_rate": 2.9465521573440196e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3930 + }, + { + "epoch": 1.146348559790515, + "flow_matching_loss": 0.019704478234052657, + "grad_norm": 0.13710808753967285, + "learning_rate": 2.946113829844624e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3940 + }, + { + "epoch": 1.1492580739016585, + "flow_matching_loss": 0.017835623864084483, + "grad_norm": 0.14823487401008606, + "learning_rate": 2.945673745180566e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3950 + }, + { + "epoch": 1.152167588012802, + "flow_matching_loss": 0.019996625278145073, + "grad_norm": 0.12896005809307098, + "learning_rate": 2.945231903886589e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3960 + }, + { + "epoch": 1.1550771021239452, + "flow_matching_loss": 0.019948563631623985, + "grad_norm": 0.13482925295829773, + "learning_rate": 2.9447883064995728e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3970 + }, + { + "epoch": 1.1579866162350887, + "flow_matching_loss": 0.02486538589000702, + "grad_norm": 0.11081795394420624, + "learning_rate": 2.9443429535585288e-05, + "loss": 0.0249, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3980 + }, + { + "epoch": 1.1608961303462322, + "flow_matching_loss": 0.02374758366495371, + "grad_norm": 0.15054279565811157, + "learning_rate": 2.9438958456046044e-05, + "loss": 0.0237, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3990 + }, + { + "epoch": 1.1638056444573757, + "flow_matching_loss": 0.01966148177161813, + "grad_norm": 0.12416628003120422, + "learning_rate": 2.9434469831810765e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4000 + }, + { + "epoch": 1.1667151585685192, + "flow_matching_loss": 0.021495402604341508, + "grad_norm": 0.09240670502185822, + "learning_rate": 2.9429963668333566e-05, + "loss": 0.0215, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4010 + }, + { + "epoch": 1.1696246726796624, + "flow_matching_loss": 0.018995161447674037, + "grad_norm": 0.13091585040092468, + "learning_rate": 2.9425439971089845e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4020 + }, + { + "epoch": 1.172534186790806, + "flow_matching_loss": 0.018500068876892328, + "grad_norm": 0.15096819400787354, + "learning_rate": 2.9420898745576336e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4030 + }, + { + "epoch": 1.1754437009019494, + "flow_matching_loss": 0.01895195757970214, + "grad_norm": 0.09951987862586975, + "learning_rate": 2.9416339997311044e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4040 + }, + { + "epoch": 1.1783532150130929, + "flow_matching_loss": 0.017339306883513927, + "grad_norm": 0.16477146744728088, + "learning_rate": 2.941176373183329e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4050 + }, + { + "epoch": 1.1812627291242364, + "flow_matching_loss": 0.02029628148302436, + "grad_norm": 0.14931151270866394, + "learning_rate": 2.9407169954703653e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4060 + }, + { + "epoch": 1.1841722432353796, + "flow_matching_loss": 0.019167711678892373, + "grad_norm": 0.14891639351844788, + "learning_rate": 2.9402558671504018e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4070 + }, + { + "epoch": 1.187081757346523, + "flow_matching_loss": 0.0202009123750031, + "grad_norm": 0.08993399143218994, + "learning_rate": 2.9397929887837524e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4080 + }, + { + "epoch": 1.1899912714576666, + "flow_matching_loss": 0.01912785479798913, + "grad_norm": 0.12530530989170074, + "learning_rate": 2.939328360932858e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4090 + }, + { + "epoch": 1.19290078556881, + "flow_matching_loss": 0.019485727418214083, + "grad_norm": 0.1257530003786087, + "learning_rate": 2.938861984162285e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4100 + }, + { + "epoch": 1.1958102996799536, + "flow_matching_loss": 0.01991983186453581, + "grad_norm": 0.10403003543615341, + "learning_rate": 2.938393859038726e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4110 + }, + { + "epoch": 1.1987198137910968, + "flow_matching_loss": 0.019259783159941434, + "grad_norm": 0.11392809450626373, + "learning_rate": 2.9379239861309964e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4120 + }, + { + "epoch": 1.2016293279022403, + "flow_matching_loss": 0.016559949330985545, + "grad_norm": 0.09835024923086166, + "learning_rate": 2.937452366010036e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4130 + }, + { + "epoch": 1.2045388420133838, + "flow_matching_loss": 0.022062919940799473, + "grad_norm": 0.08878886699676514, + "learning_rate": 2.936978999248908e-05, + "loss": 0.0221, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4140 + }, + { + "epoch": 1.2074483561245273, + "flow_matching_loss": 0.019182106480002403, + "grad_norm": 0.13247188925743103, + "learning_rate": 2.936503886422798e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4150 + }, + { + "epoch": 1.2103578702356708, + "flow_matching_loss": 0.020828545093536377, + "grad_norm": 0.12607493996620178, + "learning_rate": 2.9360270281090122e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4160 + }, + { + "epoch": 1.213267384346814, + "flow_matching_loss": 0.01960375178605318, + "grad_norm": 0.10769419372081757, + "learning_rate": 2.935548424886979e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4170 + }, + { + "epoch": 1.2161768984579575, + "flow_matching_loss": 0.019240689184516668, + "grad_norm": 0.14196056127548218, + "learning_rate": 2.935068077338246e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4180 + }, + { + "epoch": 1.219086412569101, + "flow_matching_loss": 0.018125166557729243, + "grad_norm": 0.11816045641899109, + "learning_rate": 2.934585986046481e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4190 + }, + { + "epoch": 1.2219959266802445, + "flow_matching_loss": 0.017781476862728594, + "grad_norm": 0.11440300196409225, + "learning_rate": 2.93410215159747e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4200 + }, + { + "epoch": 1.2249054407913877, + "flow_matching_loss": 0.018480805121362208, + "grad_norm": 0.11997628211975098, + "learning_rate": 2.933616574579118e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4210 + }, + { + "epoch": 1.2278149549025312, + "flow_matching_loss": 0.01751372255384922, + "grad_norm": 0.13667099177837372, + "learning_rate": 2.933129255581447e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4220 + }, + { + "epoch": 1.2307244690136747, + "flow_matching_loss": 0.017776093259453774, + "grad_norm": 0.1129881888628006, + "learning_rate": 2.9326401951965954e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4230 + }, + { + "epoch": 1.2336339831248182, + "flow_matching_loss": 0.020316970907151698, + "grad_norm": 0.19248142838478088, + "learning_rate": 2.9321493940188165e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4240 + }, + { + "epoch": 1.2365434972359617, + "flow_matching_loss": 0.01876587523147464, + "grad_norm": 0.14058232307434082, + "learning_rate": 2.931656852644482e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4250 + }, + { + "epoch": 1.239453011347105, + "flow_matching_loss": 0.01890244139358401, + "grad_norm": 0.1320091187953949, + "learning_rate": 2.9311625716720748e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4260 + }, + { + "epoch": 1.2423625254582484, + "flow_matching_loss": 0.016232216637581585, + "grad_norm": 0.11126454174518585, + "learning_rate": 2.9306665517021927e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4270 + }, + { + "epoch": 1.245272039569392, + "flow_matching_loss": 0.019796877354383468, + "grad_norm": 0.10708227008581161, + "learning_rate": 2.9301687933375483e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4280 + }, + { + "epoch": 1.2481815536805354, + "flow_matching_loss": 0.019929695595055818, + "grad_norm": 0.10741990059614182, + "learning_rate": 2.929669297182963e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4290 + }, + { + "epoch": 1.2510910677916787, + "flow_matching_loss": 0.019358795322477817, + "grad_norm": 0.12188580632209778, + "learning_rate": 2.929168063845373e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4300 + }, + { + "epoch": 1.2540005819028222, + "flow_matching_loss": 0.02070425422862172, + "grad_norm": 0.11918968707323074, + "learning_rate": 2.9286650939338236e-05, + "loss": 0.0207, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4310 + }, + { + "epoch": 1.2569100960139656, + "flow_matching_loss": 0.017745612561702727, + "grad_norm": 0.10381732136011124, + "learning_rate": 2.9281603880594706e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4320 + }, + { + "epoch": 1.2598196101251091, + "flow_matching_loss": 0.020806472189724444, + "grad_norm": 0.12861894071102142, + "learning_rate": 2.9276539468355796e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4330 + }, + { + "epoch": 1.2627291242362526, + "flow_matching_loss": 0.020982307195663453, + "grad_norm": 0.1616789698600769, + "learning_rate": 2.9271457708775244e-05, + "loss": 0.021, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4340 + }, + { + "epoch": 1.2656386383473959, + "flow_matching_loss": 0.01866462789475918, + "grad_norm": 0.12223922461271286, + "learning_rate": 2.9266358608027863e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4350 + }, + { + "epoch": 1.2685481524585394, + "flow_matching_loss": 0.018269345816224813, + "grad_norm": 0.14737433195114136, + "learning_rate": 2.9261242172309543e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4360 + }, + { + "epoch": 1.2714576665696828, + "flow_matching_loss": 0.019210759457200764, + "grad_norm": 0.12061114609241486, + "learning_rate": 2.925610840783723e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4370 + }, + { + "epoch": 1.2743671806808263, + "flow_matching_loss": 0.02246705526486039, + "grad_norm": 0.17330051958560944, + "learning_rate": 2.9250957320848946e-05, + "loss": 0.0225, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4380 + }, + { + "epoch": 1.2772766947919698, + "flow_matching_loss": 0.020086084492504597, + "grad_norm": 0.1284935027360916, + "learning_rate": 2.924578891760374e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4390 + }, + { + "epoch": 1.280186208903113, + "flow_matching_loss": 0.020921478513628244, + "grad_norm": 0.13293075561523438, + "learning_rate": 2.924060320438171e-05, + "loss": 0.0209, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4400 + }, + { + "epoch": 1.2830957230142566, + "flow_matching_loss": 0.020981478318572045, + "grad_norm": 0.17335684597492218, + "learning_rate": 2.9235400187483978e-05, + "loss": 0.021, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4410 + }, + { + "epoch": 1.2860052371254, + "flow_matching_loss": 0.02017575381323695, + "grad_norm": 0.1661846935749054, + "learning_rate": 2.9230179873232716e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4420 + }, + { + "epoch": 1.2889147512365435, + "flow_matching_loss": 0.021438610926270484, + "grad_norm": 0.1257018893957138, + "learning_rate": 2.922494226797108e-05, + "loss": 0.0214, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4430 + }, + { + "epoch": 1.291824265347687, + "flow_matching_loss": 0.018998367618769406, + "grad_norm": 0.13150636851787567, + "learning_rate": 2.9219687378063275e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4440 + }, + { + "epoch": 1.2947337794588303, + "flow_matching_loss": 0.019688482210040093, + "grad_norm": 0.14847339689731598, + "learning_rate": 2.921441520989447e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4450 + }, + { + "epoch": 1.2976432935699738, + "flow_matching_loss": 0.017121675331145524, + "grad_norm": 0.09760269522666931, + "learning_rate": 2.9209125769870854e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4460 + }, + { + "epoch": 1.3005528076811173, + "flow_matching_loss": 0.0173819312825799, + "grad_norm": 0.10750922560691833, + "learning_rate": 2.92038190644196e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4470 + }, + { + "epoch": 1.3034623217922607, + "flow_matching_loss": 0.020000479370355605, + "grad_norm": 0.14462816715240479, + "learning_rate": 2.9198495099988843e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4480 + }, + { + "epoch": 1.3063718359034042, + "flow_matching_loss": 0.019060518126934767, + "grad_norm": 0.10630142688751221, + "learning_rate": 2.9193153883047712e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4490 + }, + { + "epoch": 1.3092813500145475, + "flow_matching_loss": 0.02040344001725316, + "grad_norm": 0.14214695990085602, + "learning_rate": 2.9187795420086287e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4500 + }, + { + "epoch": 1.312190864125691, + "flow_matching_loss": 0.016186579782515764, + "grad_norm": 0.10136432200670242, + "learning_rate": 2.9182419717615602e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4510 + }, + { + "epoch": 1.3151003782368345, + "flow_matching_loss": 0.01786073427647352, + "grad_norm": 0.1113603413105011, + "learning_rate": 2.917702678216765e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4520 + }, + { + "epoch": 1.318009892347978, + "flow_matching_loss": 0.018625627737492323, + "grad_norm": 0.12369287014007568, + "learning_rate": 2.9171616620295347e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4530 + }, + { + "epoch": 1.3209194064591214, + "flow_matching_loss": 0.01832020692527294, + "grad_norm": 0.15089717507362366, + "learning_rate": 2.9166189238572554e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4540 + }, + { + "epoch": 1.3238289205702647, + "flow_matching_loss": 0.019018799345940352, + "grad_norm": 0.11599308997392654, + "learning_rate": 2.9160744643594052e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4550 + }, + { + "epoch": 1.3267384346814082, + "flow_matching_loss": 0.0192041864618659, + "grad_norm": 0.1317102164030075, + "learning_rate": 2.915528284197554e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4560 + }, + { + "epoch": 1.3296479487925517, + "flow_matching_loss": 0.01736687235534191, + "grad_norm": 0.1145939752459526, + "learning_rate": 2.914980384035362e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4570 + }, + { + "epoch": 1.3325574629036951, + "flow_matching_loss": 0.021554567199200392, + "grad_norm": 0.12938949465751648, + "learning_rate": 2.9144307645385787e-05, + "loss": 0.0216, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4580 + }, + { + "epoch": 1.3354669770148386, + "flow_matching_loss": 0.021568352542817592, + "grad_norm": 0.11842968314886093, + "learning_rate": 2.9138794263750454e-05, + "loss": 0.0216, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4590 + }, + { + "epoch": 1.338376491125982, + "flow_matching_loss": 0.017527136765420438, + "grad_norm": 0.1170065626502037, + "learning_rate": 2.913326370214688e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4600 + }, + { + "epoch": 1.3412860052371254, + "flow_matching_loss": 0.022438178956508636, + "grad_norm": 0.1408499926328659, + "learning_rate": 2.9127715967295234e-05, + "loss": 0.0224, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4610 + }, + { + "epoch": 1.3441955193482689, + "flow_matching_loss": 0.0188970192335546, + "grad_norm": 0.13196496665477753, + "learning_rate": 2.912215106593653e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4620 + }, + { + "epoch": 1.3471050334594123, + "flow_matching_loss": 0.019031760562211276, + "grad_norm": 0.10530880093574524, + "learning_rate": 2.911656900483265e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4630 + }, + { + "epoch": 1.3500145475705558, + "flow_matching_loss": 0.017637229431420563, + "grad_norm": 0.14364442229270935, + "learning_rate": 2.9110969790766322e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4640 + }, + { + "epoch": 1.352924061681699, + "flow_matching_loss": 0.019008156936615707, + "grad_norm": 0.15376168489456177, + "learning_rate": 2.910535343054112e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4650 + }, + { + "epoch": 1.3558335757928426, + "flow_matching_loss": 0.018715053051710128, + "grad_norm": 0.1134209856390953, + "learning_rate": 2.909971993098145e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4660 + }, + { + "epoch": 1.358743089903986, + "flow_matching_loss": 0.019854200072586536, + "grad_norm": 0.11885932087898254, + "learning_rate": 2.9094069298932552e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4670 + }, + { + "epoch": 1.3616526040151296, + "flow_matching_loss": 0.018717249482870103, + "grad_norm": 0.173568457365036, + "learning_rate": 2.908840154126047e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4680 + }, + { + "epoch": 1.364562118126273, + "flow_matching_loss": 0.02138984641060233, + "grad_norm": 0.11760532110929489, + "learning_rate": 2.9082716664852063e-05, + "loss": 0.0214, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4690 + }, + { + "epoch": 1.3674716322374163, + "flow_matching_loss": 0.020969637855887412, + "grad_norm": 0.12221749871969223, + "learning_rate": 2.9077014676615e-05, + "loss": 0.021, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4700 + }, + { + "epoch": 1.3703811463485598, + "flow_matching_loss": 0.019146033376455308, + "grad_norm": 0.12373429536819458, + "learning_rate": 2.9071295583477724e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4710 + }, + { + "epoch": 1.3732906604597033, + "flow_matching_loss": 0.019553594663739203, + "grad_norm": 0.18360206484794617, + "learning_rate": 2.906555939238948e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4720 + }, + { + "epoch": 1.3762001745708465, + "flow_matching_loss": 0.01856996091082692, + "grad_norm": 0.1547669768333435, + "learning_rate": 2.905980611032028e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4730 + }, + { + "epoch": 1.3791096886819902, + "flow_matching_loss": 0.018859682139009237, + "grad_norm": 0.11328162252902985, + "learning_rate": 2.9054035744260904e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4740 + }, + { + "epoch": 1.3820192027931335, + "flow_matching_loss": 0.021247651614248752, + "grad_norm": 0.14880675077438354, + "learning_rate": 2.9048248301222885e-05, + "loss": 0.0212, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4750 + }, + { + "epoch": 1.384928716904277, + "flow_matching_loss": 0.0204268254339695, + "grad_norm": 0.14653463661670685, + "learning_rate": 2.904244378823852e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4760 + }, + { + "epoch": 1.3878382310154205, + "flow_matching_loss": 0.019560932461172343, + "grad_norm": 0.1130187138915062, + "learning_rate": 2.9036622212360836e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4770 + }, + { + "epoch": 1.3907477451265637, + "flow_matching_loss": 0.021731292456388475, + "grad_norm": 0.10508779436349869, + "learning_rate": 2.9030783580663596e-05, + "loss": 0.0217, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4780 + }, + { + "epoch": 1.3936572592377072, + "flow_matching_loss": 0.020548449549824, + "grad_norm": 0.20009881258010864, + "learning_rate": 2.902492790024129e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4790 + }, + { + "epoch": 1.3965667733488507, + "flow_matching_loss": 0.0172319483011961, + "grad_norm": 0.10355977714061737, + "learning_rate": 2.901905517820912e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4800 + }, + { + "epoch": 1.3994762874599942, + "flow_matching_loss": 0.022524194046854972, + "grad_norm": 0.15807634592056274, + "learning_rate": 2.9013165421703e-05, + "loss": 0.0225, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4810 + }, + { + "epoch": 1.4023858015711377, + "flow_matching_loss": 0.020549187436699867, + "grad_norm": 0.1320662498474121, + "learning_rate": 2.9007258637879534e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4820 + }, + { + "epoch": 1.405295315682281, + "flow_matching_loss": 0.01981898080557585, + "grad_norm": 0.10965550690889359, + "learning_rate": 2.9001334833916025e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4830 + }, + { + "epoch": 1.4082048297934244, + "flow_matching_loss": 0.02017290974035859, + "grad_norm": 0.1510964035987854, + "learning_rate": 2.899539401701046e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4840 + }, + { + "epoch": 1.411114343904568, + "flow_matching_loss": 0.01761751538142562, + "grad_norm": 0.1066637933254242, + "learning_rate": 2.8989436194381474e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4850 + }, + { + "epoch": 1.4140238580157114, + "flow_matching_loss": 0.02111377930268645, + "grad_norm": 0.12935081124305725, + "learning_rate": 2.8983461373268404e-05, + "loss": 0.0211, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4860 + }, + { + "epoch": 1.4169333721268549, + "flow_matching_loss": 0.02015951508656144, + "grad_norm": 0.12256127595901489, + "learning_rate": 2.8977469560931205e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4870 + }, + { + "epoch": 1.4198428862379981, + "flow_matching_loss": 0.019500618521124124, + "grad_norm": 0.11455879360437393, + "learning_rate": 2.8971460764650502e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4880 + }, + { + "epoch": 1.4227524003491416, + "flow_matching_loss": 0.017201552167534828, + "grad_norm": 0.10316871851682663, + "learning_rate": 2.8965434991727545e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4890 + }, + { + "epoch": 1.4256619144602851, + "flow_matching_loss": 0.018755893502384425, + "grad_norm": 0.09314754605293274, + "learning_rate": 2.895939224948422e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4900 + }, + { + "epoch": 1.4285714285714286, + "flow_matching_loss": 0.019229982607066633, + "grad_norm": 0.13427837193012238, + "learning_rate": 2.895333254526303e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4910 + }, + { + "epoch": 1.431480942682572, + "flow_matching_loss": 0.022341964673250912, + "grad_norm": 0.12000983953475952, + "learning_rate": 2.8947255886427084e-05, + "loss": 0.0223, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4920 + }, + { + "epoch": 1.4343904567937154, + "flow_matching_loss": 0.020814536046236752, + "grad_norm": 0.12465659528970718, + "learning_rate": 2.8941162280360095e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4930 + }, + { + "epoch": 1.4372999709048588, + "flow_matching_loss": 0.0183053825981915, + "grad_norm": 0.13812005519866943, + "learning_rate": 2.8935051734466368e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4940 + }, + { + "epoch": 1.4402094850160023, + "flow_matching_loss": 0.019715505931526423, + "grad_norm": 0.10634562373161316, + "learning_rate": 2.8928924256170792e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4950 + }, + { + "epoch": 1.4431189991271458, + "flow_matching_loss": 0.02020000768825412, + "grad_norm": 0.11548534035682678, + "learning_rate": 2.8922779852918836e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4960 + }, + { + "epoch": 1.4460285132382893, + "flow_matching_loss": 0.01955066928640008, + "grad_norm": 0.10274205356836319, + "learning_rate": 2.8916618532176526e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4970 + }, + { + "epoch": 1.4489380273494326, + "flow_matching_loss": 0.019864471908658744, + "grad_norm": 0.14873306453227997, + "learning_rate": 2.891044030143045e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4980 + }, + { + "epoch": 1.451847541460576, + "flow_matching_loss": 0.019037802703678608, + "grad_norm": 0.12509863078594208, + "learning_rate": 2.890424516818774e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4990 + }, + { + "epoch": 1.4547570555717195, + "flow_matching_loss": 0.02002840079367161, + "grad_norm": 0.10798213630914688, + "learning_rate": 2.8898033139976057e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5000 + }, + { + "epoch": 1.457666569682863, + "flow_matching_loss": 0.019962809421122073, + "grad_norm": 0.12182355672121048, + "learning_rate": 2.889180422434362e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5010 + }, + { + "epoch": 1.4605760837940065, + "flow_matching_loss": 0.020827846322208642, + "grad_norm": 0.15948912501335144, + "learning_rate": 2.8885558428859136e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5020 + }, + { + "epoch": 1.4634855979051498, + "flow_matching_loss": 0.0204831057228148, + "grad_norm": 0.14672935009002686, + "learning_rate": 2.8879295761111835e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5030 + }, + { + "epoch": 1.4663951120162932, + "flow_matching_loss": 0.019081580732017756, + "grad_norm": 0.07762162387371063, + "learning_rate": 2.8873016228711452e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5040 + }, + { + "epoch": 1.4693046261274367, + "flow_matching_loss": 0.017685319669544695, + "grad_norm": 0.12403806298971176, + "learning_rate": 2.8866719839288215e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5050 + }, + { + "epoch": 1.4722141402385802, + "flow_matching_loss": 0.01899409117177129, + "grad_norm": 0.16185012459754944, + "learning_rate": 2.886040660049283e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5060 + }, + { + "epoch": 1.4751236543497237, + "flow_matching_loss": 0.01830744557082653, + "grad_norm": 0.11183091998100281, + "learning_rate": 2.8854076519996467e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5070 + }, + { + "epoch": 1.478033168460867, + "flow_matching_loss": 0.018303022906184196, + "grad_norm": 0.09673533588647842, + "learning_rate": 2.884772960549078e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5080 + }, + { + "epoch": 1.4809426825720104, + "flow_matching_loss": 0.019492703676223754, + "grad_norm": 0.1263529509305954, + "learning_rate": 2.8841365864687867e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5090 + }, + { + "epoch": 1.483852196683154, + "flow_matching_loss": 0.019129863753914834, + "grad_norm": 0.09041758626699448, + "learning_rate": 2.8834985305320276e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5100 + }, + { + "epoch": 1.4867617107942974, + "flow_matching_loss": 0.02022917876020074, + "grad_norm": 0.11968643963336945, + "learning_rate": 2.882858793514098e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5110 + }, + { + "epoch": 1.489671224905441, + "flow_matching_loss": 0.01907501993700862, + "grad_norm": 0.11285536736249924, + "learning_rate": 2.8822173761923398e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5120 + }, + { + "epoch": 1.4925807390165842, + "flow_matching_loss": 0.019703442137688397, + "grad_norm": 0.16226078569889069, + "learning_rate": 2.8815742793461343e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5130 + }, + { + "epoch": 1.4954902531277277, + "flow_matching_loss": 0.022600498888641594, + "grad_norm": 0.16061252355575562, + "learning_rate": 2.880929503756906e-05, + "loss": 0.0226, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5140 + }, + { + "epoch": 1.4983997672388711, + "flow_matching_loss": 0.020170036889612674, + "grad_norm": 0.110908143222332, + "learning_rate": 2.880283050208117e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5150 + }, + { + "epoch": 1.5013092813500144, + "flow_matching_loss": 0.020044369157403708, + "grad_norm": 0.1428176462650299, + "learning_rate": 2.8796349194852704e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5160 + }, + { + "epoch": 1.504218795461158, + "flow_matching_loss": 0.01762789497151971, + "grad_norm": 0.12255964428186417, + "learning_rate": 2.8789851123759052e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5170 + }, + { + "epoch": 1.5071283095723014, + "flow_matching_loss": 0.016656992398202418, + "grad_norm": 0.08746975660324097, + "learning_rate": 2.8783336296695985e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5180 + }, + { + "epoch": 1.5100378236834449, + "flow_matching_loss": 0.02086135158315301, + "grad_norm": 0.15272243320941925, + "learning_rate": 2.877680472157964e-05, + "loss": 0.0209, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5190 + }, + { + "epoch": 1.5129473377945883, + "flow_matching_loss": 0.018120726756751537, + "grad_norm": 0.1029312014579773, + "learning_rate": 2.8770256406346483e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5200 + }, + { + "epoch": 1.5158568519057316, + "flow_matching_loss": 0.018238282576203347, + "grad_norm": 0.10517055541276932, + "learning_rate": 2.8763691358953348e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5210 + }, + { + "epoch": 1.5187663660168753, + "flow_matching_loss": 0.01812648056074977, + "grad_norm": 0.11472605913877487, + "learning_rate": 2.8757109587377377e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5220 + }, + { + "epoch": 1.5216758801280186, + "flow_matching_loss": 0.018506610207259654, + "grad_norm": 0.10459251701831818, + "learning_rate": 2.8750511099616047e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5230 + }, + { + "epoch": 1.524585394239162, + "flow_matching_loss": 0.018230716232210398, + "grad_norm": 0.13472667336463928, + "learning_rate": 2.8743895903687143e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5240 + }, + { + "epoch": 1.5274949083503055, + "flow_matching_loss": 0.018906720355153082, + "grad_norm": 0.14485566318035126, + "learning_rate": 2.8737264007628753e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5250 + }, + { + "epoch": 1.5304044224614488, + "flow_matching_loss": 0.021160504966974258, + "grad_norm": 0.17228569090366364, + "learning_rate": 2.8730615419499247e-05, + "loss": 0.0212, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5260 + }, + { + "epoch": 1.5333139365725925, + "flow_matching_loss": 0.018793236743658782, + "grad_norm": 0.11260773241519928, + "learning_rate": 2.8723950147377303e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5270 + }, + { + "epoch": 1.5362234506837358, + "flow_matching_loss": 0.018410925380885602, + "grad_norm": 0.14049914479255676, + "learning_rate": 2.871726819936184e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5280 + }, + { + "epoch": 1.5391329647948793, + "flow_matching_loss": 0.017452908959239722, + "grad_norm": 0.15281878411769867, + "learning_rate": 2.8710569583572068e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5290 + }, + { + "epoch": 1.5420424789060228, + "flow_matching_loss": 0.018716686870902777, + "grad_norm": 0.12281883507966995, + "learning_rate": 2.870385430814742e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5300 + }, + { + "epoch": 1.544951993017166, + "flow_matching_loss": 0.0176226912997663, + "grad_norm": 0.11591055244207382, + "learning_rate": 2.8697122381247605e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5310 + }, + { + "epoch": 1.5478615071283097, + "flow_matching_loss": 0.019679184071719646, + "grad_norm": 0.13119739294052124, + "learning_rate": 2.8690373811052545e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5320 + }, + { + "epoch": 1.550771021239453, + "flow_matching_loss": 0.01820288533344865, + "grad_norm": 0.13280276954174042, + "learning_rate": 2.8683608605762378e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5330 + }, + { + "epoch": 1.5536805353505965, + "flow_matching_loss": 0.018775815796107053, + "grad_norm": 0.091239333152771, + "learning_rate": 2.8676826773597485e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5340 + }, + { + "epoch": 1.55659004946174, + "flow_matching_loss": 0.017419944610446693, + "grad_norm": 0.15205912292003632, + "learning_rate": 2.8670028322798413e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5350 + }, + { + "epoch": 1.5594995635728832, + "flow_matching_loss": 0.019118846114724874, + "grad_norm": 0.15730036795139313, + "learning_rate": 2.8663213261625932e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5360 + }, + { + "epoch": 1.562409077684027, + "flow_matching_loss": 0.018265718687325717, + "grad_norm": 0.15268570184707642, + "learning_rate": 2.865638159836098e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5370 + }, + { + "epoch": 1.5653185917951702, + "flow_matching_loss": 0.018881552014499904, + "grad_norm": 0.09835398197174072, + "learning_rate": 2.8649533341304677e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5380 + }, + { + "epoch": 1.5682281059063137, + "flow_matching_loss": 0.020548896864056587, + "grad_norm": 0.07549668848514557, + "learning_rate": 2.8642668498778297e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5390 + }, + { + "epoch": 1.5711376200174572, + "flow_matching_loss": 0.01912550302222371, + "grad_norm": 0.1386328935623169, + "learning_rate": 2.8635787079123268e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5400 + }, + { + "epoch": 1.5740471341286004, + "flow_matching_loss": 0.019477459229528905, + "grad_norm": 0.11879459768533707, + "learning_rate": 2.8628889090701175e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5410 + }, + { + "epoch": 1.576956648239744, + "flow_matching_loss": 0.01951815662905574, + "grad_norm": 0.12619857490062714, + "learning_rate": 2.862197454189371e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5420 + }, + { + "epoch": 1.5798661623508874, + "flow_matching_loss": 0.018207450676709414, + "grad_norm": 0.1473788619041443, + "learning_rate": 2.8615043441102708e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5430 + }, + { + "epoch": 1.5827756764620309, + "flow_matching_loss": 0.017285762913525106, + "grad_norm": 0.12563709914684296, + "learning_rate": 2.860809579675011e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5440 + }, + { + "epoch": 1.5856851905731744, + "flow_matching_loss": 0.01779331313446164, + "grad_norm": 0.13562650978565216, + "learning_rate": 2.860113161727796e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5450 + }, + { + "epoch": 1.5885947046843176, + "flow_matching_loss": 0.01668977504596114, + "grad_norm": 0.11321182548999786, + "learning_rate": 2.8594150911148393e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5460 + }, + { + "epoch": 1.5915042187954611, + "flow_matching_loss": 0.019253403320908545, + "grad_norm": 0.11993864923715591, + "learning_rate": 2.8587153686843626e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5470 + }, + { + "epoch": 1.5944137329066046, + "flow_matching_loss": 0.01942837266251445, + "grad_norm": 0.12306574732065201, + "learning_rate": 2.858013995286594e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5480 + }, + { + "epoch": 1.597323247017748, + "flow_matching_loss": 0.01880584079772234, + "grad_norm": 0.14904756844043732, + "learning_rate": 2.857310971773769e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5490 + }, + { + "epoch": 1.6002327611288916, + "flow_matching_loss": 0.022374465689063074, + "grad_norm": 0.09933476150035858, + "learning_rate": 2.8566062990001274e-05, + "loss": 0.0224, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5500 + }, + { + "epoch": 1.6031422752400348, + "flow_matching_loss": 0.017133787740021944, + "grad_norm": 0.10517794638872147, + "learning_rate": 2.8558999778219126e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5510 + }, + { + "epoch": 1.6060517893511783, + "flow_matching_loss": 0.018309002835303546, + "grad_norm": 0.09561958909034729, + "learning_rate": 2.8551920090973722e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5520 + }, + { + "epoch": 1.6089613034623218, + "flow_matching_loss": 0.016501423716545106, + "grad_norm": 0.1084713414311409, + "learning_rate": 2.854482393686754e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5530 + }, + { + "epoch": 1.6118708175734653, + "flow_matching_loss": 0.020688424352556467, + "grad_norm": 0.13586421310901642, + "learning_rate": 2.8537711324523086e-05, + "loss": 0.0207, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5540 + }, + { + "epoch": 1.6147803316846088, + "flow_matching_loss": 0.019431835040450096, + "grad_norm": 0.13148713111877441, + "learning_rate": 2.8530582262582845e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5550 + }, + { + "epoch": 1.617689845795752, + "flow_matching_loss": 0.02191240880638361, + "grad_norm": 0.13474299013614655, + "learning_rate": 2.8523436759709306e-05, + "loss": 0.0219, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5560 + }, + { + "epoch": 1.6205993599068955, + "flow_matching_loss": 0.01959659168496728, + "grad_norm": 0.13190722465515137, + "learning_rate": 2.8516274824584926e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5570 + }, + { + "epoch": 1.623508874018039, + "flow_matching_loss": 0.02251738514751196, + "grad_norm": 0.1609591692686081, + "learning_rate": 2.8509096465912136e-05, + "loss": 0.0225, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5580 + }, + { + "epoch": 1.6264183881291823, + "flow_matching_loss": 0.018510091863572598, + "grad_norm": 0.09757319837808609, + "learning_rate": 2.8501901692413315e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5590 + }, + { + "epoch": 1.629327902240326, + "flow_matching_loss": 0.021406957414001226, + "grad_norm": 0.09985985606908798, + "learning_rate": 2.8494690512830788e-05, + "loss": 0.0214, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5600 + }, + { + "epoch": 1.6322374163514692, + "flow_matching_loss": 0.02082693288102746, + "grad_norm": 0.09843172132968903, + "learning_rate": 2.8487462935926823e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5610 + }, + { + "epoch": 1.6351469304626127, + "flow_matching_loss": 0.017858053930103777, + "grad_norm": 0.0760171189904213, + "learning_rate": 2.8480218970483606e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5620 + }, + { + "epoch": 1.6380564445737562, + "flow_matching_loss": 0.020932516269385814, + "grad_norm": 0.14808383584022522, + "learning_rate": 2.847295862530324e-05, + "loss": 0.0209, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5630 + }, + { + "epoch": 1.6409659586848995, + "flow_matching_loss": 0.01759991515427828, + "grad_norm": 0.1468019187450409, + "learning_rate": 2.846568190920773e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5640 + }, + { + "epoch": 1.6438754727960432, + "flow_matching_loss": 0.020527474023401736, + "grad_norm": 0.10361718386411667, + "learning_rate": 2.8458388831038965e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5650 + }, + { + "epoch": 1.6467849869071864, + "flow_matching_loss": 0.019008374121040107, + "grad_norm": 0.11944040656089783, + "learning_rate": 2.8451079399658725e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5660 + }, + { + "epoch": 1.64969450101833, + "flow_matching_loss": 0.0213087591342628, + "grad_norm": 0.15964806079864502, + "learning_rate": 2.8443753623948665e-05, + "loss": 0.0213, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5670 + }, + { + "epoch": 1.6526040151294734, + "flow_matching_loss": 0.01653672391548753, + "grad_norm": 0.10433870553970337, + "learning_rate": 2.8436411512810288e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5680 + }, + { + "epoch": 1.6555135292406167, + "flow_matching_loss": 0.020477885846048594, + "grad_norm": 0.1059308722615242, + "learning_rate": 2.842905307516495e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5690 + }, + { + "epoch": 1.6584230433517604, + "flow_matching_loss": 0.016016767639666796, + "grad_norm": 0.12804830074310303, + "learning_rate": 2.8421678319953854e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5700 + }, + { + "epoch": 1.6613325574629036, + "flow_matching_loss": 0.020519766397774218, + "grad_norm": 0.13237659633159637, + "learning_rate": 2.841428725613801e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5710 + }, + { + "epoch": 1.6642420715740471, + "flow_matching_loss": 0.01919695008546114, + "grad_norm": 0.10482087731361389, + "learning_rate": 2.8406879892698272e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5720 + }, + { + "epoch": 1.6671515856851906, + "flow_matching_loss": 0.022824862599372865, + "grad_norm": 0.11630039662122726, + "learning_rate": 2.8399456238635274e-05, + "loss": 0.0228, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5730 + }, + { + "epoch": 1.6700610997963339, + "flow_matching_loss": 0.020564477425068617, + "grad_norm": 0.12934935092926025, + "learning_rate": 2.8392016302969457e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5740 + }, + { + "epoch": 1.6729706139074776, + "flow_matching_loss": 0.01713141370564699, + "grad_norm": 0.09672720730304718, + "learning_rate": 2.8384560094741055e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5750 + }, + { + "epoch": 1.6758801280186209, + "flow_matching_loss": 0.017388729937374593, + "grad_norm": 0.10527375340461731, + "learning_rate": 2.8377087623010054e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5760 + }, + { + "epoch": 1.6787896421297643, + "flow_matching_loss": 0.0180427435785532, + "grad_norm": 0.10753776133060455, + "learning_rate": 2.8369598896856213e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5770 + }, + { + "epoch": 1.6816991562409078, + "flow_matching_loss": 0.020248808432370424, + "grad_norm": 0.13654041290283203, + "learning_rate": 2.836209392537904e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5780 + }, + { + "epoch": 1.684608670352051, + "flow_matching_loss": 0.016587851103395224, + "grad_norm": 0.11213497072458267, + "learning_rate": 2.8354572717697787e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5790 + }, + { + "epoch": 1.6875181844631948, + "flow_matching_loss": 0.017910706996917724, + "grad_norm": 0.08665506541728973, + "learning_rate": 2.8347035282951428e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5800 + }, + { + "epoch": 1.690427698574338, + "flow_matching_loss": 0.01791443843394518, + "grad_norm": 0.10648215562105179, + "learning_rate": 2.8339481630298653e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5810 + }, + { + "epoch": 1.6933372126854815, + "flow_matching_loss": 0.017467097379267216, + "grad_norm": 0.07802577316761017, + "learning_rate": 2.833191176891787e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5820 + }, + { + "epoch": 1.696246726796625, + "flow_matching_loss": 0.018080213200300933, + "grad_norm": 0.08774275332689285, + "learning_rate": 2.832432570800717e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5830 + }, + { + "epoch": 1.6991562409077683, + "flow_matching_loss": 0.016392524633556605, + "grad_norm": 0.09495893120765686, + "learning_rate": 2.8316723456784332e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5840 + }, + { + "epoch": 1.702065755018912, + "flow_matching_loss": 0.019120931904762984, + "grad_norm": 0.1025930643081665, + "learning_rate": 2.8309105024486808e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5850 + }, + { + "epoch": 1.7049752691300553, + "flow_matching_loss": 0.019098191987723112, + "grad_norm": 0.11956751346588135, + "learning_rate": 2.830147042037172e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5860 + }, + { + "epoch": 1.7078847832411987, + "flow_matching_loss": 0.017716273479163647, + "grad_norm": 0.09644756466150284, + "learning_rate": 2.8293819653715816e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5870 + }, + { + "epoch": 1.7107942973523422, + "flow_matching_loss": 0.01870635589584708, + "grad_norm": 0.12819872796535492, + "learning_rate": 2.828615273381551e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5880 + }, + { + "epoch": 1.7137038114634855, + "flow_matching_loss": 0.015151969622820616, + "grad_norm": 0.09492373466491699, + "learning_rate": 2.827846966998683e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5890 + }, + { + "epoch": 1.7166133255746292, + "flow_matching_loss": 0.01797046633437276, + "grad_norm": 0.11459405720233917, + "learning_rate": 2.8270770471565422e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5900 + }, + { + "epoch": 1.7195228396857725, + "flow_matching_loss": 0.01813667267560959, + "grad_norm": 0.15836870670318604, + "learning_rate": 2.826305514790654e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5910 + }, + { + "epoch": 1.722432353796916, + "flow_matching_loss": 0.01977492021396756, + "grad_norm": 0.12300648540258408, + "learning_rate": 2.825532370838503e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5920 + }, + { + "epoch": 1.7253418679080594, + "flow_matching_loss": 0.021208139322698116, + "grad_norm": 0.15771807730197906, + "learning_rate": 2.824757616239532e-05, + "loss": 0.0212, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5930 + }, + { + "epoch": 1.7282513820192027, + "flow_matching_loss": 0.019439497869461773, + "grad_norm": 0.11723114550113678, + "learning_rate": 2.8239812519351404e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5940 + }, + { + "epoch": 1.7311608961303462, + "flow_matching_loss": 0.02009304566308856, + "grad_norm": 0.16762574017047882, + "learning_rate": 2.823203278868685e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5950 + }, + { + "epoch": 1.7340704102414897, + "flow_matching_loss": 0.019415396451950073, + "grad_norm": 0.12293820828199387, + "learning_rate": 2.8224236979854752e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5960 + }, + { + "epoch": 1.7369799243526332, + "flow_matching_loss": 0.017473978642374276, + "grad_norm": 0.08871529996395111, + "learning_rate": 2.821642510232776e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5970 + }, + { + "epoch": 1.7398894384637766, + "flow_matching_loss": 0.01665355879813433, + "grad_norm": 0.09165602177381516, + "learning_rate": 2.8208597165598046e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5980 + }, + { + "epoch": 1.74279895257492, + "flow_matching_loss": 0.019570414163172244, + "grad_norm": 0.14009873569011688, + "learning_rate": 2.8200753179177285e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5990 + }, + { + "epoch": 1.7457084666860634, + "flow_matching_loss": 0.017459666635841132, + "grad_norm": 0.1078985184431076, + "learning_rate": 2.819289315259666e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6000 + }, + { + "epoch": 1.7486179807972069, + "flow_matching_loss": 0.020017506461590527, + "grad_norm": 0.14468605816364288, + "learning_rate": 2.8185017095406845e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6010 + }, + { + "epoch": 1.7515274949083504, + "flow_matching_loss": 0.018135637603700162, + "grad_norm": 0.14033393561840057, + "learning_rate": 2.8177125017177998e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6020 + }, + { + "epoch": 1.7544370090194938, + "flow_matching_loss": 0.021071703359484672, + "grad_norm": 0.16608628630638123, + "learning_rate": 2.8169216927499727e-05, + "loss": 0.0211, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6030 + }, + { + "epoch": 1.757346523130637, + "flow_matching_loss": 0.017256941832602023, + "grad_norm": 0.121796615421772, + "learning_rate": 2.816129283598112e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6040 + }, + { + "epoch": 1.7602560372417806, + "flow_matching_loss": 0.018350136838853358, + "grad_norm": 0.10709434002637863, + "learning_rate": 2.815335275225068e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6050 + }, + { + "epoch": 1.763165551352924, + "flow_matching_loss": 0.017989817541092634, + "grad_norm": 0.12471825629472733, + "learning_rate": 2.8145396685956374e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6060 + }, + { + "epoch": 1.7660750654640673, + "flow_matching_loss": 0.017530438676476478, + "grad_norm": 0.10398363322019577, + "learning_rate": 2.813742464676556e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6070 + }, + { + "epoch": 1.768984579575211, + "flow_matching_loss": 0.02223329562693834, + "grad_norm": 0.12240432947874069, + "learning_rate": 2.8129436644365025e-05, + "loss": 0.0222, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6080 + }, + { + "epoch": 1.7718940936863543, + "flow_matching_loss": 0.019121537636965513, + "grad_norm": 0.09736226499080658, + "learning_rate": 2.8121432688460938e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6090 + }, + { + "epoch": 1.7748036077974978, + "flow_matching_loss": 0.0183946643024683, + "grad_norm": 0.09501925855875015, + "learning_rate": 2.8113412788778867e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6100 + }, + { + "epoch": 1.7777131219086413, + "flow_matching_loss": 0.01878406051546335, + "grad_norm": 0.12148749828338623, + "learning_rate": 2.8105376955063744e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6110 + }, + { + "epoch": 1.7806226360197845, + "flow_matching_loss": 0.020606681145727633, + "grad_norm": 0.11325923353433609, + "learning_rate": 2.8097325197079862e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6120 + }, + { + "epoch": 1.7835321501309283, + "flow_matching_loss": 0.02008127812296152, + "grad_norm": 0.11031731963157654, + "learning_rate": 2.808925752461087e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6130 + }, + { + "epoch": 1.7864416642420715, + "flow_matching_loss": 0.018576754629611968, + "grad_norm": 0.12853418290615082, + "learning_rate": 2.808117394745975e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6140 + }, + { + "epoch": 1.789351178353215, + "flow_matching_loss": 0.016922535840421915, + "grad_norm": 0.11872372031211853, + "learning_rate": 2.807307447544881e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6150 + }, + { + "epoch": 1.7922606924643585, + "flow_matching_loss": 0.018211338948458434, + "grad_norm": 0.08386844396591187, + "learning_rate": 2.806495911841967e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6160 + }, + { + "epoch": 1.7951702065755017, + "flow_matching_loss": 0.016328243538737297, + "grad_norm": 0.10095055401325226, + "learning_rate": 2.8056827886233255e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6170 + }, + { + "epoch": 1.7980797206866455, + "flow_matching_loss": 0.017897058092057706, + "grad_norm": 0.13274455070495605, + "learning_rate": 2.8048680788769783e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6180 + }, + { + "epoch": 1.8009892347977887, + "flow_matching_loss": 0.018188320100307465, + "grad_norm": 0.15242157876491547, + "learning_rate": 2.8040517835928742e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6190 + }, + { + "epoch": 1.8038987489089322, + "flow_matching_loss": 0.01832429226487875, + "grad_norm": 0.1345454901456833, + "learning_rate": 2.803233903762889e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6200 + }, + { + "epoch": 1.8068082630200757, + "flow_matching_loss": 0.019449283368885517, + "grad_norm": 0.11482206732034683, + "learning_rate": 2.8024144403808235e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6210 + }, + { + "epoch": 1.809717777131219, + "flow_matching_loss": 0.018642341159284113, + "grad_norm": 0.1261654943227768, + "learning_rate": 2.8015933944424037e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6220 + }, + { + "epoch": 1.8126272912423627, + "flow_matching_loss": 0.017872029915452003, + "grad_norm": 0.1406889408826828, + "learning_rate": 2.800770766945277e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6230 + }, + { + "epoch": 1.815536805353506, + "flow_matching_loss": 0.01845394754782319, + "grad_norm": 0.07824287563562393, + "learning_rate": 2.7999465588890135e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6240 + }, + { + "epoch": 1.8184463194646494, + "flow_matching_loss": 0.017713198065757753, + "grad_norm": 0.1495400071144104, + "learning_rate": 2.799120771275105e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6250 + }, + { + "epoch": 1.821355833575793, + "flow_matching_loss": 0.017567190621048213, + "grad_norm": 0.12494136393070221, + "learning_rate": 2.7982934051069597e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6260 + }, + { + "epoch": 1.8242653476869362, + "flow_matching_loss": 0.017152978293597698, + "grad_norm": 0.09832169115543365, + "learning_rate": 2.7974644613899063e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6270 + }, + { + "epoch": 1.8271748617980799, + "flow_matching_loss": 0.01631726399064064, + "grad_norm": 0.09938570111989975, + "learning_rate": 2.796633941131189e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6280 + }, + { + "epoch": 1.8300843759092231, + "flow_matching_loss": 0.020150656159967185, + "grad_norm": 0.1652933955192566, + "learning_rate": 2.795801845339969e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6290 + }, + { + "epoch": 1.8329938900203666, + "flow_matching_loss": 0.01760119628161192, + "grad_norm": 0.09582813084125519, + "learning_rate": 2.7949681750273208e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6300 + }, + { + "epoch": 1.83590340413151, + "flow_matching_loss": 0.016486515197902916, + "grad_norm": 0.11537952721118927, + "learning_rate": 2.7941329312062322e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6310 + }, + { + "epoch": 1.8388129182426534, + "flow_matching_loss": 0.017091987747699024, + "grad_norm": 0.09537938237190247, + "learning_rate": 2.7932961148916037e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6320 + }, + { + "epoch": 1.841722432353797, + "flow_matching_loss": 0.01981869200244546, + "grad_norm": 0.10912882536649704, + "learning_rate": 2.7924577271002456e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6330 + }, + { + "epoch": 1.8446319464649403, + "flow_matching_loss": 0.017597956117242575, + "grad_norm": 0.12476957589387894, + "learning_rate": 2.7916177688508782e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6340 + }, + { + "epoch": 1.8475414605760838, + "flow_matching_loss": 0.021393010579049587, + "grad_norm": 0.10056480020284653, + "learning_rate": 2.7907762411641302e-05, + "loss": 0.0214, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6350 + }, + { + "epoch": 1.8504509746872273, + "flow_matching_loss": 0.01655219476670027, + "grad_norm": 0.09373020380735397, + "learning_rate": 2.7899331450625367e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6360 + }, + { + "epoch": 1.8533604887983706, + "flow_matching_loss": 0.016919801477342844, + "grad_norm": 0.11811617761850357, + "learning_rate": 2.7890884815705397e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6370 + }, + { + "epoch": 1.8562700029095143, + "flow_matching_loss": 0.018225272372364997, + "grad_norm": 0.10383623093366623, + "learning_rate": 2.788242251714484e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6380 + }, + { + "epoch": 1.8591795170206575, + "flow_matching_loss": 0.01976971998810768, + "grad_norm": 0.08779042214155197, + "learning_rate": 2.787394456522619e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6390 + }, + { + "epoch": 1.862089031131801, + "flow_matching_loss": 0.02126378733664751, + "grad_norm": 0.1354113072156906, + "learning_rate": 2.7865450970250965e-05, + "loss": 0.0213, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6400 + }, + { + "epoch": 1.8649985452429445, + "flow_matching_loss": 0.020412834361195564, + "grad_norm": 0.14462250471115112, + "learning_rate": 2.7856941742539676e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6410 + }, + { + "epoch": 1.8679080593540878, + "flow_matching_loss": 0.01904315808787942, + "grad_norm": 0.10963939875364304, + "learning_rate": 2.784841689243184e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6420 + }, + { + "epoch": 1.8708175734652313, + "flow_matching_loss": 0.018189170770347117, + "grad_norm": 0.12888513505458832, + "learning_rate": 2.783987643028595e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6430 + }, + { + "epoch": 1.8737270875763747, + "flow_matching_loss": 0.019831146113574505, + "grad_norm": 0.1690405309200287, + "learning_rate": 2.783132036647948e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6440 + }, + { + "epoch": 1.8766366016875182, + "flow_matching_loss": 0.018027864303439854, + "grad_norm": 0.10561926662921906, + "learning_rate": 2.7822748711408852e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6450 + }, + { + "epoch": 1.8795461157986617, + "flow_matching_loss": 0.01946938652545214, + "grad_norm": 0.15325064957141876, + "learning_rate": 2.7814161475489428e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6460 + }, + { + "epoch": 1.882455629909805, + "flow_matching_loss": 0.016154635325074197, + "grad_norm": 0.1087401956319809, + "learning_rate": 2.780555866915552e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6470 + }, + { + "epoch": 1.8853651440209485, + "flow_matching_loss": 0.017828434985131026, + "grad_norm": 0.12222432345151901, + "learning_rate": 2.7796940302860338e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6480 + }, + { + "epoch": 1.888274658132092, + "flow_matching_loss": 0.0197286743670702, + "grad_norm": 0.10177840292453766, + "learning_rate": 2.7788306387076018e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6490 + }, + { + "epoch": 1.8911841722432354, + "flow_matching_loss": 0.02003141222521663, + "grad_norm": 0.1077129915356636, + "learning_rate": 2.7779656932293584e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6500 + }, + { + "epoch": 1.894093686354379, + "flow_matching_loss": 0.021216689515858887, + "grad_norm": 0.0892796441912651, + "learning_rate": 2.777099194902293e-05, + "loss": 0.0212, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6510 + }, + { + "epoch": 1.8970032004655222, + "flow_matching_loss": 0.01879407772794366, + "grad_norm": 0.08499927073717117, + "learning_rate": 2.776231144779283e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6520 + }, + { + "epoch": 1.8999127145766657, + "flow_matching_loss": 0.01796602588146925, + "grad_norm": 0.1007700264453888, + "learning_rate": 2.775361543915091e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6530 + }, + { + "epoch": 1.9028222286878091, + "flow_matching_loss": 0.016223870404064655, + "grad_norm": 0.15638108551502228, + "learning_rate": 2.774490393366365e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6540 + }, + { + "epoch": 1.9057317427989524, + "flow_matching_loss": 0.018480258528143167, + "grad_norm": 0.10980281978845596, + "learning_rate": 2.7736176941916346e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6550 + }, + { + "epoch": 1.9086412569100961, + "flow_matching_loss": 0.018437505606561898, + "grad_norm": 0.17012175917625427, + "learning_rate": 2.772743447451311e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6560 + }, + { + "epoch": 1.9115507710212394, + "flow_matching_loss": 0.01966326581314206, + "grad_norm": 0.13892583549022675, + "learning_rate": 2.7718676542076872e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6570 + }, + { + "epoch": 1.9144602851323829, + "flow_matching_loss": 0.016835830360651016, + "grad_norm": 0.13839183747768402, + "learning_rate": 2.770990315524934e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6580 + }, + { + "epoch": 1.9173697992435264, + "flow_matching_loss": 0.019104960560798644, + "grad_norm": 0.08989879488945007, + "learning_rate": 2.7701114324691015e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6590 + }, + { + "epoch": 1.9202793133546696, + "flow_matching_loss": 0.01974287647753954, + "grad_norm": 0.11082018166780472, + "learning_rate": 2.7692310061081145e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6600 + }, + { + "epoch": 1.9231888274658133, + "flow_matching_loss": 0.018229635246098042, + "grad_norm": 0.09218133985996246, + "learning_rate": 2.7683490375117742e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6610 + }, + { + "epoch": 1.9260983415769566, + "flow_matching_loss": 0.020802381727844477, + "grad_norm": 0.11059550195932388, + "learning_rate": 2.7674655277517556e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6620 + }, + { + "epoch": 1.9290078556881, + "flow_matching_loss": 0.02003654707223177, + "grad_norm": 0.14356720447540283, + "learning_rate": 2.7665804779016072e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6630 + }, + { + "epoch": 1.9319173697992436, + "flow_matching_loss": 0.02060510702431202, + "grad_norm": 0.10042020678520203, + "learning_rate": 2.765693889036747e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6640 + }, + { + "epoch": 1.9348268839103868, + "flow_matching_loss": 0.022569483239203693, + "grad_norm": 0.1281530112028122, + "learning_rate": 2.764805762234465e-05, + "loss": 0.0226, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6650 + }, + { + "epoch": 1.9377363980215305, + "flow_matching_loss": 0.01670844629406929, + "grad_norm": 0.11978866159915924, + "learning_rate": 2.7639160985739182e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6660 + }, + { + "epoch": 1.9406459121326738, + "flow_matching_loss": 0.019993682764470576, + "grad_norm": 0.11956313997507095, + "learning_rate": 2.7630248991361317e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6670 + }, + { + "epoch": 1.9435554262438173, + "flow_matching_loss": 0.016700246185064316, + "grad_norm": 0.11147811263799667, + "learning_rate": 2.762132165003998e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6680 + }, + { + "epoch": 1.9464649403549608, + "flow_matching_loss": 0.01704650726169348, + "grad_norm": 0.10005667060613632, + "learning_rate": 2.7612378972622722e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6690 + }, + { + "epoch": 1.949374454466104, + "flow_matching_loss": 0.01750051025301218, + "grad_norm": 0.11528822779655457, + "learning_rate": 2.7603420969975743e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6700 + }, + { + "epoch": 1.9522839685772477, + "flow_matching_loss": 0.02058243751525879, + "grad_norm": 0.14216676354408264, + "learning_rate": 2.7594447652983862e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6710 + }, + { + "epoch": 1.955193482688391, + "flow_matching_loss": 0.023338209744542836, + "grad_norm": 0.15836761891841888, + "learning_rate": 2.7585459032550497e-05, + "loss": 0.0233, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6720 + }, + { + "epoch": 1.9581029967995345, + "flow_matching_loss": 0.019279392156749963, + "grad_norm": 0.14713028073310852, + "learning_rate": 2.7576455119597684e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6730 + }, + { + "epoch": 1.961012510910678, + "flow_matching_loss": 0.017403384391218423, + "grad_norm": 0.10096863657236099, + "learning_rate": 2.7567435925066012e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6740 + }, + { + "epoch": 1.9639220250218212, + "flow_matching_loss": 0.02195101575925946, + "grad_norm": 0.12051745504140854, + "learning_rate": 2.7558401459914665e-05, + "loss": 0.022, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6750 + }, + { + "epoch": 1.966831539132965, + "flow_matching_loss": 0.01902120839804411, + "grad_norm": 0.08476331830024719, + "learning_rate": 2.7549351735121362e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6760 + }, + { + "epoch": 1.9697410532441082, + "flow_matching_loss": 0.018765654414892197, + "grad_norm": 0.10845906287431717, + "learning_rate": 2.7540286761682377e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6770 + }, + { + "epoch": 1.9726505673552517, + "flow_matching_loss": 0.019052738882601263, + "grad_norm": 0.12779732048511505, + "learning_rate": 2.75312065506125e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6780 + }, + { + "epoch": 1.9755600814663952, + "flow_matching_loss": 0.01819583456963301, + "grad_norm": 0.09755697101354599, + "learning_rate": 2.7522111112945052e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6790 + }, + { + "epoch": 1.9784695955775384, + "flow_matching_loss": 0.019504572823643686, + "grad_norm": 0.09685169905424118, + "learning_rate": 2.751300045973184e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6800 + }, + { + "epoch": 1.9813791096886821, + "flow_matching_loss": 0.0199792823754251, + "grad_norm": 0.09749870002269745, + "learning_rate": 2.750387460204317e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6810 + }, + { + "epoch": 1.9842886237998254, + "flow_matching_loss": 0.021008232980966567, + "grad_norm": 0.1061653196811676, + "learning_rate": 2.7494733550967827e-05, + "loss": 0.021, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6820 + }, + { + "epoch": 1.9871981379109689, + "flow_matching_loss": 0.01700979396700859, + "grad_norm": 0.10215368866920471, + "learning_rate": 2.748557731761304e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6830 + }, + { + "epoch": 1.9901076520221124, + "flow_matching_loss": 0.019343967083841562, + "grad_norm": 0.1380034238100052, + "learning_rate": 2.74764059131045e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6840 + }, + { + "epoch": 1.9930171661332556, + "flow_matching_loss": 0.01757895275950432, + "grad_norm": 0.09404817968606949, + "learning_rate": 2.7467219348586322e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6850 + }, + { + "epoch": 1.9959266802443993, + "flow_matching_loss": 0.018919487483799458, + "grad_norm": 0.10472332686185837, + "learning_rate": 2.7458017635221056e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6860 + }, + { + "epoch": 1.9988361943555426, + "flow_matching_loss": 0.01745895128697157, + "grad_norm": 0.11759072542190552, + "learning_rate": 2.7448800784189652e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6870 + }, + { + "epoch": 2.001745708466686, + "flow_matching_loss": 0.019107850547879935, + "grad_norm": 0.09371015429496765, + "learning_rate": 2.7439568806691448e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6880 + }, + { + "epoch": 2.0046552225778296, + "flow_matching_loss": 0.018006999790668488, + "grad_norm": 0.07147903740406036, + "learning_rate": 2.7430321713944167e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6890 + }, + { + "epoch": 2.007564736688973, + "flow_matching_loss": 0.01997232558205724, + "grad_norm": 0.1254059374332428, + "learning_rate": 2.7421059517183906e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6900 + }, + { + "epoch": 2.0104742508001165, + "flow_matching_loss": 0.02010758752003312, + "grad_norm": 0.16575628519058228, + "learning_rate": 2.7411782227665097e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6910 + }, + { + "epoch": 2.01338376491126, + "flow_matching_loss": 0.018700932338833808, + "grad_norm": 0.1305495649576187, + "learning_rate": 2.7402489856660528e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6920 + }, + { + "epoch": 2.016293279022403, + "flow_matching_loss": 0.01908630384132266, + "grad_norm": 0.13278566300868988, + "learning_rate": 2.7393182415461307e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6930 + }, + { + "epoch": 2.019202793133547, + "flow_matching_loss": 0.017158983927220105, + "grad_norm": 0.09328961372375488, + "learning_rate": 2.7383859915376853e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6940 + }, + { + "epoch": 2.02211230724469, + "flow_matching_loss": 0.01924540037289262, + "grad_norm": 0.1477542221546173, + "learning_rate": 2.7374522367734875e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6950 + }, + { + "epoch": 2.0250218213558338, + "flow_matching_loss": 0.019124919176101686, + "grad_norm": 0.15845683217048645, + "learning_rate": 2.7365169783881378e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6960 + }, + { + "epoch": 2.027931335466977, + "flow_matching_loss": 0.01857250416651368, + "grad_norm": 0.11981357634067535, + "learning_rate": 2.735580217518064e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6970 + }, + { + "epoch": 2.0308408495781203, + "flow_matching_loss": 0.019608942698687314, + "grad_norm": 0.09181609749794006, + "learning_rate": 2.7346419553015175e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6980 + }, + { + "epoch": 2.033750363689264, + "flow_matching_loss": 0.017771827150136234, + "grad_norm": 0.10161527991294861, + "learning_rate": 2.733702192878577e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6990 + }, + { + "epoch": 2.0366598778004072, + "flow_matching_loss": 0.017237311881035566, + "grad_norm": 0.08972230553627014, + "learning_rate": 2.732760931391141e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7000 + }, + { + "epoch": 2.039569391911551, + "flow_matching_loss": 0.01937702801078558, + "grad_norm": 0.09767349064350128, + "learning_rate": 2.731818171982932e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7010 + }, + { + "epoch": 2.042478906022694, + "flow_matching_loss": 0.01661846935749054, + "grad_norm": 0.0894271582365036, + "learning_rate": 2.730873915799491e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7020 + }, + { + "epoch": 2.0453884201338375, + "flow_matching_loss": 0.01607099426910281, + "grad_norm": 0.10574827343225479, + "learning_rate": 2.7299281639881783e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7030 + }, + { + "epoch": 2.048297934244981, + "flow_matching_loss": 0.018687658477574587, + "grad_norm": 0.1452738344669342, + "learning_rate": 2.728980917698172e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7040 + }, + { + "epoch": 2.0512074483561245, + "flow_matching_loss": 0.017808857839554548, + "grad_norm": 0.12131457030773163, + "learning_rate": 2.7280321780804655e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7050 + }, + { + "epoch": 2.054116962467268, + "flow_matching_loss": 0.015875842049717905, + "grad_norm": 0.09585986286401749, + "learning_rate": 2.7270819462878666e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7060 + }, + { + "epoch": 2.0570264765784114, + "flow_matching_loss": 0.02055332288146019, + "grad_norm": 0.09519877284765244, + "learning_rate": 2.7261302234749965e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7070 + }, + { + "epoch": 2.0599359906895547, + "flow_matching_loss": 0.01629904406145215, + "grad_norm": 0.12086743861436844, + "learning_rate": 2.7251770107982888e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7080 + }, + { + "epoch": 2.0628455048006984, + "flow_matching_loss": 0.017739154677838088, + "grad_norm": 0.14438055455684662, + "learning_rate": 2.7242223094159856e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7090 + }, + { + "epoch": 2.0657550189118417, + "flow_matching_loss": 0.01703035328537226, + "grad_norm": 0.09205318987369537, + "learning_rate": 2.7232661204881408e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7100 + }, + { + "epoch": 2.0686645330229854, + "flow_matching_loss": 0.0181176302023232, + "grad_norm": 0.0749276801943779, + "learning_rate": 2.7223084451766123e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7110 + }, + { + "epoch": 2.0715740471341286, + "flow_matching_loss": 0.018373394012451173, + "grad_norm": 0.1115708202123642, + "learning_rate": 2.721349284645067e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7120 + }, + { + "epoch": 2.074483561245272, + "flow_matching_loss": 0.01914216037839651, + "grad_norm": 0.11786515265703201, + "learning_rate": 2.7203886400589752e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7130 + }, + { + "epoch": 2.0773930753564156, + "flow_matching_loss": 0.017512196861207485, + "grad_norm": 0.08773602545261383, + "learning_rate": 2.7194265125856105e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7140 + }, + { + "epoch": 2.080302589467559, + "flow_matching_loss": 0.016131309792399406, + "grad_norm": 0.10608745366334915, + "learning_rate": 2.718462903394049e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7150 + }, + { + "epoch": 2.0832121035787026, + "flow_matching_loss": 0.019255019817501308, + "grad_norm": 0.12565390765666962, + "learning_rate": 2.7174978136551662e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7160 + }, + { + "epoch": 2.086121617689846, + "flow_matching_loss": 0.01975020905956626, + "grad_norm": 0.16263680160045624, + "learning_rate": 2.716531244541637e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7170 + }, + { + "epoch": 2.089031131800989, + "flow_matching_loss": 0.019192702881991864, + "grad_norm": 0.1377885937690735, + "learning_rate": 2.715563197227934e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7180 + }, + { + "epoch": 2.091940645912133, + "flow_matching_loss": 0.019697252195328474, + "grad_norm": 0.11945761740207672, + "learning_rate": 2.714593672890327e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7190 + }, + { + "epoch": 2.094850160023276, + "flow_matching_loss": 0.017656457237899305, + "grad_norm": 0.12310130149126053, + "learning_rate": 2.7136226727068783e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7200 + }, + { + "epoch": 2.0977596741344193, + "flow_matching_loss": 0.018130894657224415, + "grad_norm": 0.10869112610816956, + "learning_rate": 2.712650197857445e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7210 + }, + { + "epoch": 2.100669188245563, + "flow_matching_loss": 0.015464584156870843, + "grad_norm": 0.09250783175230026, + "learning_rate": 2.7116762495236763e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7220 + }, + { + "epoch": 2.1035787023567063, + "flow_matching_loss": 0.01928229983896017, + "grad_norm": 0.12655623257160187, + "learning_rate": 2.7107008288890106e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7230 + }, + { + "epoch": 2.10648821646785, + "flow_matching_loss": 0.020645223557949066, + "grad_norm": 0.1267855316400528, + "learning_rate": 2.709723937138676e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7240 + }, + { + "epoch": 2.1093977305789933, + "flow_matching_loss": 0.016452715918421745, + "grad_norm": 0.11875329166650772, + "learning_rate": 2.7087455754596883e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7250 + }, + { + "epoch": 2.112307244690137, + "flow_matching_loss": 0.016613583639264108, + "grad_norm": 0.12252247333526611, + "learning_rate": 2.7077657450408498e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7260 + }, + { + "epoch": 2.1152167588012802, + "flow_matching_loss": 0.01875978959724307, + "grad_norm": 0.1177014410495758, + "learning_rate": 2.7067844470727456e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7270 + }, + { + "epoch": 2.1181262729124235, + "flow_matching_loss": 0.019320190139114856, + "grad_norm": 0.15774577856063843, + "learning_rate": 2.7058016827477455e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7280 + }, + { + "epoch": 2.121035787023567, + "flow_matching_loss": 0.020403694454580547, + "grad_norm": 0.12605337798595428, + "learning_rate": 2.7048174532600018e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7290 + }, + { + "epoch": 2.1239453011347105, + "flow_matching_loss": 0.0171829123981297, + "grad_norm": 0.10918740183115005, + "learning_rate": 2.7038317598054455e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7300 + }, + { + "epoch": 2.1268548152458537, + "flow_matching_loss": 0.01947987247258425, + "grad_norm": 0.1331186592578888, + "learning_rate": 2.702844603581787e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7310 + }, + { + "epoch": 2.1297643293569974, + "flow_matching_loss": 0.02042197557166219, + "grad_norm": 0.11829070001840591, + "learning_rate": 2.701855985788514e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7320 + }, + { + "epoch": 2.1326738434681407, + "flow_matching_loss": 0.017492032144218683, + "grad_norm": 0.11200670897960663, + "learning_rate": 2.7008659076268912e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7330 + }, + { + "epoch": 2.1355833575792844, + "flow_matching_loss": 0.018832398857921363, + "grad_norm": 0.10687265545129776, + "learning_rate": 2.6998743702999562e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7340 + }, + { + "epoch": 2.1384928716904277, + "flow_matching_loss": 0.017905874457210303, + "grad_norm": 0.10314186662435532, + "learning_rate": 2.6988813750125202e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7350 + }, + { + "epoch": 2.141402385801571, + "flow_matching_loss": 0.023601799644529818, + "grad_norm": 0.1182151734828949, + "learning_rate": 2.6978869229711668e-05, + "loss": 0.0236, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7360 + }, + { + "epoch": 2.1443118999127146, + "flow_matching_loss": 0.02245865873992443, + "grad_norm": 0.13742539286613464, + "learning_rate": 2.6968910153842484e-05, + "loss": 0.0225, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7370 + }, + { + "epoch": 2.147221414023858, + "flow_matching_loss": 0.018705550488084553, + "grad_norm": 0.13163797557353973, + "learning_rate": 2.6958936534618874e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7380 + }, + { + "epoch": 2.1501309281350016, + "flow_matching_loss": 0.020284469239413738, + "grad_norm": 0.13644683361053467, + "learning_rate": 2.694894838415971e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7390 + }, + { + "epoch": 2.153040442246145, + "flow_matching_loss": 0.01644694060087204, + "grad_norm": 0.09499238431453705, + "learning_rate": 2.6938945714601555e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7400 + }, + { + "epoch": 2.155949956357288, + "flow_matching_loss": 0.017744527477771044, + "grad_norm": 0.10439356416463852, + "learning_rate": 2.6928928538098587e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7410 + }, + { + "epoch": 2.158859470468432, + "flow_matching_loss": 0.018028839025646448, + "grad_norm": 0.12389776110649109, + "learning_rate": 2.6918896866822615e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7420 + }, + { + "epoch": 2.161768984579575, + "flow_matching_loss": 0.017645984143018722, + "grad_norm": 0.10141152888536453, + "learning_rate": 2.690885071296308e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7430 + }, + { + "epoch": 2.164678498690719, + "flow_matching_loss": 0.018127983435988426, + "grad_norm": 0.188055157661438, + "learning_rate": 2.6898790088726993e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7440 + }, + { + "epoch": 2.167588012801862, + "flow_matching_loss": 0.018003125116229056, + "grad_norm": 0.11409605294466019, + "learning_rate": 2.6888715006338968e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7450 + }, + { + "epoch": 2.1704975269130053, + "flow_matching_loss": 0.02029329026117921, + "grad_norm": 0.13632600009441376, + "learning_rate": 2.6878625478041172e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7460 + }, + { + "epoch": 2.173407041024149, + "flow_matching_loss": 0.01840378250926733, + "grad_norm": 0.15372392535209656, + "learning_rate": 2.6868521516093342e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7470 + }, + { + "epoch": 2.1763165551352923, + "flow_matching_loss": 0.017146605625748636, + "grad_norm": 0.13934482634067535, + "learning_rate": 2.6858403132772745e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7480 + }, + { + "epoch": 2.179226069246436, + "flow_matching_loss": 0.015697774291038514, + "grad_norm": 0.09667491167783737, + "learning_rate": 2.6848270340374162e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7490 + }, + { + "epoch": 2.1821355833575793, + "flow_matching_loss": 0.019864925276488067, + "grad_norm": 0.12849313020706177, + "learning_rate": 2.68381231512099e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7500 + }, + { + "epoch": 2.1850450974687226, + "flow_matching_loss": 0.017223574686795473, + "grad_norm": 0.1567932367324829, + "learning_rate": 2.6827961577609745e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7510 + }, + { + "epoch": 2.1879546115798663, + "flow_matching_loss": 0.020675304532051086, + "grad_norm": 0.10655330121517181, + "learning_rate": 2.681778563192097e-05, + "loss": 0.0207, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7520 + }, + { + "epoch": 2.1908641256910095, + "flow_matching_loss": 0.01732751177623868, + "grad_norm": 0.12565752863883972, + "learning_rate": 2.6807595326508308e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7530 + }, + { + "epoch": 2.1937736398021532, + "flow_matching_loss": 0.016748631466180088, + "grad_norm": 0.09178011119365692, + "learning_rate": 2.679739067375394e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7540 + }, + { + "epoch": 2.1966831539132965, + "flow_matching_loss": 0.01769283264875412, + "grad_norm": 0.10233199596405029, + "learning_rate": 2.6787171686057487e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7550 + }, + { + "epoch": 2.1995926680244398, + "flow_matching_loss": 0.02076631663367152, + "grad_norm": 0.0899515300989151, + "learning_rate": 2.677693837583597e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7560 + }, + { + "epoch": 2.2025021821355835, + "flow_matching_loss": 0.01593199549242854, + "grad_norm": 0.10503878444433212, + "learning_rate": 2.6766690755523836e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7570 + }, + { + "epoch": 2.2054116962467267, + "flow_matching_loss": 0.016337306704372168, + "grad_norm": 0.0933985561132431, + "learning_rate": 2.675642883757291e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7580 + }, + { + "epoch": 2.2083212103578704, + "flow_matching_loss": 0.017446743324398994, + "grad_norm": 0.11551081389188766, + "learning_rate": 2.674615263445239e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7590 + }, + { + "epoch": 2.2112307244690137, + "flow_matching_loss": 0.017489498667418956, + "grad_norm": 0.09626580029726028, + "learning_rate": 2.673586215864882e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7600 + }, + { + "epoch": 2.214140238580157, + "flow_matching_loss": 0.016690688207745552, + "grad_norm": 0.10648716241121292, + "learning_rate": 2.6725557422666113e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7610 + }, + { + "epoch": 2.2170497526913007, + "flow_matching_loss": 0.017160642333328723, + "grad_norm": 0.13215602934360504, + "learning_rate": 2.6715238439025493e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7620 + }, + { + "epoch": 2.219959266802444, + "flow_matching_loss": 0.018863613717257977, + "grad_norm": 0.09900104999542236, + "learning_rate": 2.6704905220265493e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7630 + }, + { + "epoch": 2.2228687809135876, + "flow_matching_loss": 0.01829357910901308, + "grad_norm": 0.1289888173341751, + "learning_rate": 2.6694557778941947e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7640 + }, + { + "epoch": 2.225778295024731, + "flow_matching_loss": 0.018411530833691358, + "grad_norm": 0.09086525440216064, + "learning_rate": 2.6684196127627977e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7650 + }, + { + "epoch": 2.228687809135874, + "flow_matching_loss": 0.02259277496486902, + "grad_norm": 0.10775649547576904, + "learning_rate": 2.6673820278913963e-05, + "loss": 0.0226, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7660 + }, + { + "epoch": 2.231597323247018, + "flow_matching_loss": 0.0184589559212327, + "grad_norm": 0.12871241569519043, + "learning_rate": 2.6663430245407544e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7670 + }, + { + "epoch": 2.234506837358161, + "flow_matching_loss": 0.016867988649755715, + "grad_norm": 0.11540054529905319, + "learning_rate": 2.6653026039733585e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7680 + }, + { + "epoch": 2.237416351469305, + "flow_matching_loss": 0.01605006419122219, + "grad_norm": 0.11315573751926422, + "learning_rate": 2.6642607674534176e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7690 + }, + { + "epoch": 2.240325865580448, + "flow_matching_loss": 0.02202965011820197, + "grad_norm": 0.0985175296664238, + "learning_rate": 2.663217516246862e-05, + "loss": 0.022, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7700 + }, + { + "epoch": 2.2432353796915914, + "flow_matching_loss": 0.018754064664244653, + "grad_norm": 0.15313781797885895, + "learning_rate": 2.66217285162134e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7710 + }, + { + "epoch": 2.246144893802735, + "flow_matching_loss": 0.016832628659904002, + "grad_norm": 0.11416077613830566, + "learning_rate": 2.6611267748462175e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7720 + }, + { + "epoch": 2.2490544079138783, + "flow_matching_loss": 0.017864371929317713, + "grad_norm": 0.14785654842853546, + "learning_rate": 2.6600792871925764e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7730 + }, + { + "epoch": 2.2519639220250216, + "flow_matching_loss": 0.01711286697536707, + "grad_norm": 0.11531225591897964, + "learning_rate": 2.659030389933213e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7740 + }, + { + "epoch": 2.2548734361361653, + "flow_matching_loss": 0.017955686897039413, + "grad_norm": 0.10034838318824768, + "learning_rate": 2.6579800843426358e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7750 + }, + { + "epoch": 2.2577829502473086, + "flow_matching_loss": 0.01792715918272734, + "grad_norm": 0.10758007317781448, + "learning_rate": 2.656928371697067e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7760 + }, + { + "epoch": 2.2606924643584523, + "flow_matching_loss": 0.02318886574357748, + "grad_norm": 0.13957729935646057, + "learning_rate": 2.6558752532744346e-05, + "loss": 0.0232, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7770 + }, + { + "epoch": 2.2636019784695955, + "flow_matching_loss": 0.016971156001091003, + "grad_norm": 0.0829559937119484, + "learning_rate": 2.654820730354377e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7780 + }, + { + "epoch": 2.2665114925807393, + "flow_matching_loss": 0.016257997322827577, + "grad_norm": 0.13538013398647308, + "learning_rate": 2.6537648042182406e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7790 + }, + { + "epoch": 2.2694210066918825, + "flow_matching_loss": 0.01719806743785739, + "grad_norm": 0.09995758533477783, + "learning_rate": 2.6527074761490735e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7800 + }, + { + "epoch": 2.2723305208030258, + "flow_matching_loss": 0.01840660870075226, + "grad_norm": 0.1083717867732048, + "learning_rate": 2.6516487474316303e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7810 + }, + { + "epoch": 2.2752400349141695, + "flow_matching_loss": 0.015520579647272825, + "grad_norm": 0.1478813886642456, + "learning_rate": 2.650588619352365e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7820 + }, + { + "epoch": 2.2781495490253127, + "flow_matching_loss": 0.01949509233236313, + "grad_norm": 0.1400599628686905, + "learning_rate": 2.649527093199434e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7830 + }, + { + "epoch": 2.281059063136456, + "flow_matching_loss": 0.01699040047824383, + "grad_norm": 0.11697930097579956, + "learning_rate": 2.6484641702626914e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7840 + }, + { + "epoch": 2.2839685772475997, + "flow_matching_loss": 0.01729865912348032, + "grad_norm": 0.1082041785120964, + "learning_rate": 2.647399851833689e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7850 + }, + { + "epoch": 2.286878091358743, + "flow_matching_loss": 0.018088943511247634, + "grad_norm": 0.09038826078176498, + "learning_rate": 2.6463341392056735e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7860 + }, + { + "epoch": 2.2897876054698867, + "flow_matching_loss": 0.016865708865225315, + "grad_norm": 0.10212551802396774, + "learning_rate": 2.6452670336735867e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7870 + }, + { + "epoch": 2.29269711958103, + "flow_matching_loss": 0.01776736481115222, + "grad_norm": 0.14641106128692627, + "learning_rate": 2.644198536534063e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7880 + }, + { + "epoch": 2.295606633692173, + "flow_matching_loss": 0.019673588685691357, + "grad_norm": 0.13974891602993011, + "learning_rate": 2.6431286490854255e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7890 + }, + { + "epoch": 2.298516147803317, + "flow_matching_loss": 0.0186963408254087, + "grad_norm": 0.1766127347946167, + "learning_rate": 2.6420573726276903e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7900 + }, + { + "epoch": 2.30142566191446, + "flow_matching_loss": 0.017929500434547663, + "grad_norm": 0.129961758852005, + "learning_rate": 2.6409847084625575e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7910 + }, + { + "epoch": 2.304335176025604, + "flow_matching_loss": 0.017995546292513608, + "grad_norm": 0.10265321284532547, + "learning_rate": 2.6399106578934167e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7920 + }, + { + "epoch": 2.307244690136747, + "flow_matching_loss": 0.018350721336901188, + "grad_norm": 0.10721010714769363, + "learning_rate": 2.63883522222534e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7930 + }, + { + "epoch": 2.3101542042478904, + "flow_matching_loss": 0.019481508154422046, + "grad_norm": 0.1303674578666687, + "learning_rate": 2.6377584027650827e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7940 + }, + { + "epoch": 2.313063718359034, + "flow_matching_loss": 0.01860106997191906, + "grad_norm": 0.10469726473093033, + "learning_rate": 2.636680200821083e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7950 + }, + { + "epoch": 2.3159732324701774, + "flow_matching_loss": 0.019746093172580005, + "grad_norm": 0.1155790239572525, + "learning_rate": 2.635600617703457e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7960 + }, + { + "epoch": 2.318882746581321, + "flow_matching_loss": 0.020961021166294815, + "grad_norm": 0.12804356217384338, + "learning_rate": 2.6345196547240007e-05, + "loss": 0.021, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7970 + }, + { + "epoch": 2.3217922606924644, + "flow_matching_loss": 0.017027804628014565, + "grad_norm": 0.09620936214923859, + "learning_rate": 2.6334373131961854e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7980 + }, + { + "epoch": 2.3247017748036076, + "flow_matching_loss": 0.018877256009727717, + "grad_norm": 0.12749458849430084, + "learning_rate": 2.6323535944351583e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7990 + }, + { + "epoch": 2.3276112889147513, + "flow_matching_loss": 0.018214420508593322, + "grad_norm": 0.11715508997440338, + "learning_rate": 2.63126849975774e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8000 + }, + { + "epoch": 2.3305208030258946, + "flow_matching_loss": 0.01641616625711322, + "grad_norm": 0.10143211483955383, + "learning_rate": 2.6301820304824234e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8010 + }, + { + "epoch": 2.3334303171370383, + "flow_matching_loss": 0.017267664894461632, + "grad_norm": 0.11031204462051392, + "learning_rate": 2.6290941879293703e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8020 + }, + { + "epoch": 2.3363398312481816, + "flow_matching_loss": 0.01848909417167306, + "grad_norm": 0.09595046192407608, + "learning_rate": 2.628004973420412e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8030 + }, + { + "epoch": 2.339249345359325, + "flow_matching_loss": 0.018946431949734686, + "grad_norm": 0.1284899115562439, + "learning_rate": 2.626914388279047e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8040 + }, + { + "epoch": 2.3421588594704685, + "flow_matching_loss": 0.018054174259305, + "grad_norm": 0.11697356402873993, + "learning_rate": 2.6258224338304394e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8050 + }, + { + "epoch": 2.345068373581612, + "flow_matching_loss": 0.018473714496940373, + "grad_norm": 0.11610624939203262, + "learning_rate": 2.6247291114014166e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8060 + }, + { + "epoch": 2.347977887692755, + "flow_matching_loss": 0.017379160784184933, + "grad_norm": 0.11207600682973862, + "learning_rate": 2.6236344223204683e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8070 + }, + { + "epoch": 2.3508874018038988, + "flow_matching_loss": 0.018568482156842946, + "grad_norm": 0.10574880987405777, + "learning_rate": 2.6225383679177453e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8080 + }, + { + "epoch": 2.353796915915042, + "flow_matching_loss": 0.017806814145296813, + "grad_norm": 0.11534494161605835, + "learning_rate": 2.6214409495250568e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8090 + }, + { + "epoch": 2.3567064300261857, + "flow_matching_loss": 0.019752371218055488, + "grad_norm": 0.08168254792690277, + "learning_rate": 2.620342168475869e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8100 + }, + { + "epoch": 2.359615944137329, + "flow_matching_loss": 0.016818151529878377, + "grad_norm": 0.08644966781139374, + "learning_rate": 2.619242026105306e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8110 + }, + { + "epoch": 2.3625254582484727, + "flow_matching_loss": 0.01936910366639495, + "grad_norm": 0.1375429928302765, + "learning_rate": 2.618140523750143e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8120 + }, + { + "epoch": 2.365434972359616, + "flow_matching_loss": 0.01882415795698762, + "grad_norm": 0.12533721327781677, + "learning_rate": 2.6170376627488103e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8130 + }, + { + "epoch": 2.3683444864707592, + "flow_matching_loss": 0.01747430134564638, + "grad_norm": 0.11289052665233612, + "learning_rate": 2.6159334444413873e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8140 + }, + { + "epoch": 2.371254000581903, + "flow_matching_loss": 0.019434971921145915, + "grad_norm": 0.10017641633749008, + "learning_rate": 2.614827870169604e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8150 + }, + { + "epoch": 2.374163514693046, + "flow_matching_loss": 0.018909587431699036, + "grad_norm": 0.07706894725561142, + "learning_rate": 2.6137209412768367e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8160 + }, + { + "epoch": 2.3770730288041895, + "flow_matching_loss": 0.01869430923834443, + "grad_norm": 0.1335970014333725, + "learning_rate": 2.612612659108109e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8170 + }, + { + "epoch": 2.379982542915333, + "flow_matching_loss": 0.017046215385198592, + "grad_norm": 0.12985338270664215, + "learning_rate": 2.611503025010088e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8180 + }, + { + "epoch": 2.3828920570264764, + "flow_matching_loss": 0.020755794923752548, + "grad_norm": 0.12446025013923645, + "learning_rate": 2.6103920403310832e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8190 + }, + { + "epoch": 2.38580157113762, + "flow_matching_loss": 0.01814684635028243, + "grad_norm": 0.09741411358118057, + "learning_rate": 2.6092797064210465e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8200 + }, + { + "epoch": 2.3887110852487634, + "flow_matching_loss": 0.018157309014350177, + "grad_norm": 0.13824057579040527, + "learning_rate": 2.6081660246315685e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8210 + }, + { + "epoch": 2.391620599359907, + "flow_matching_loss": 0.01631672354415059, + "grad_norm": 0.10004331171512604, + "learning_rate": 2.6070509963158774e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8220 + }, + { + "epoch": 2.3945301134710504, + "flow_matching_loss": 0.017104656342417, + "grad_norm": 0.1138915941119194, + "learning_rate": 2.6059346228288375e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8230 + }, + { + "epoch": 2.3974396275821936, + "flow_matching_loss": 0.018677235767245292, + "grad_norm": 0.0963015928864479, + "learning_rate": 2.604816905526948e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8240 + }, + { + "epoch": 2.4003491416933374, + "flow_matching_loss": 0.019066191371530296, + "grad_norm": 0.11263193190097809, + "learning_rate": 2.603697845768341e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8250 + }, + { + "epoch": 2.4032586558044806, + "flow_matching_loss": 0.018209832720458508, + "grad_norm": 0.14251479506492615, + "learning_rate": 2.6025774449127792e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8260 + }, + { + "epoch": 2.406168169915624, + "flow_matching_loss": 0.018481634464114903, + "grad_norm": 0.1271229088306427, + "learning_rate": 2.6014557043216562e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8270 + }, + { + "epoch": 2.4090776840267676, + "flow_matching_loss": 0.017520945332944394, + "grad_norm": 0.09441065788269043, + "learning_rate": 2.6003326253579913e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8280 + }, + { + "epoch": 2.411987198137911, + "flow_matching_loss": 0.019297345262020826, + "grad_norm": 0.11211907118558884, + "learning_rate": 2.599208209386432e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8290 + }, + { + "epoch": 2.4148967122490546, + "flow_matching_loss": 0.017434975132346153, + "grad_norm": 0.11075133830308914, + "learning_rate": 2.5980824577732502e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8300 + }, + { + "epoch": 2.417806226360198, + "flow_matching_loss": 0.01923596803098917, + "grad_norm": 0.1307036280632019, + "learning_rate": 2.5969553718863395e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8310 + }, + { + "epoch": 2.4207157404713415, + "flow_matching_loss": 0.017740354500710966, + "grad_norm": 0.11574627459049225, + "learning_rate": 2.595826953095216e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8320 + }, + { + "epoch": 2.423625254582485, + "flow_matching_loss": 0.018470351118594408, + "grad_norm": 0.11021478474140167, + "learning_rate": 2.594697202771014e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8330 + }, + { + "epoch": 2.426534768693628, + "flow_matching_loss": 0.018054230883717538, + "grad_norm": 0.12198447436094284, + "learning_rate": 2.5935661222864888e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8340 + }, + { + "epoch": 2.4294442828047718, + "flow_matching_loss": 0.019958736840635538, + "grad_norm": 0.2172105312347412, + "learning_rate": 2.592433713016008e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8350 + }, + { + "epoch": 2.432353796915915, + "flow_matching_loss": 0.017673453968018293, + "grad_norm": 0.10383239388465881, + "learning_rate": 2.5912999763355553e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8360 + }, + { + "epoch": 2.4352633110270583, + "flow_matching_loss": 0.014516102150082588, + "grad_norm": 0.07924241572618484, + "learning_rate": 2.590164913622729e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8370 + }, + { + "epoch": 2.438172825138202, + "flow_matching_loss": 0.01799259763211012, + "grad_norm": 0.10421398282051086, + "learning_rate": 2.589028526256737e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8380 + }, + { + "epoch": 2.4410823392493453, + "flow_matching_loss": 0.016724864859133957, + "grad_norm": 0.09783376008272171, + "learning_rate": 2.587890815618397e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8390 + }, + { + "epoch": 2.443991853360489, + "flow_matching_loss": 0.016822417732328176, + "grad_norm": 0.07887395471334457, + "learning_rate": 2.5867517830901343e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8400 + }, + { + "epoch": 2.4469013674716322, + "flow_matching_loss": 0.018963457271456718, + "grad_norm": 0.11532431095838547, + "learning_rate": 2.5856114300559812e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8410 + }, + { + "epoch": 2.4498108815827755, + "flow_matching_loss": 0.01839853757992387, + "grad_norm": 0.15835551917552948, + "learning_rate": 2.584469757901574e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8420 + }, + { + "epoch": 2.452720395693919, + "flow_matching_loss": 0.016642373241484164, + "grad_norm": 0.10590657591819763, + "learning_rate": 2.5833267680141526e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8430 + }, + { + "epoch": 2.4556299098050625, + "flow_matching_loss": 0.019187586847692728, + "grad_norm": 0.1292397528886795, + "learning_rate": 2.5821824617825564e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8440 + }, + { + "epoch": 2.458539423916206, + "flow_matching_loss": 0.022141394950449467, + "grad_norm": 0.1447700709104538, + "learning_rate": 2.5810368405972265e-05, + "loss": 0.0221, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8450 + }, + { + "epoch": 2.4614489380273494, + "flow_matching_loss": 0.01814094465225935, + "grad_norm": 0.09536813944578171, + "learning_rate": 2.5798899058501996e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8460 + }, + { + "epoch": 2.4643584521384927, + "flow_matching_loss": 0.018272387515753507, + "grad_norm": 0.1252065896987915, + "learning_rate": 2.578741658935111e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8470 + }, + { + "epoch": 2.4672679662496364, + "flow_matching_loss": 0.017819214053452014, + "grad_norm": 0.09329923242330551, + "learning_rate": 2.5775921012471875e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8480 + }, + { + "epoch": 2.4701774803607797, + "flow_matching_loss": 0.017887706495821477, + "grad_norm": 0.09350134432315826, + "learning_rate": 2.5764412341832505e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8490 + }, + { + "epoch": 2.4730869944719234, + "flow_matching_loss": 0.02019619569182396, + "grad_norm": 0.14066758751869202, + "learning_rate": 2.5752890591417125e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8500 + }, + { + "epoch": 2.4759965085830666, + "flow_matching_loss": 0.017168483603745698, + "grad_norm": 0.08700524270534515, + "learning_rate": 2.5741355775225744e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8510 + }, + { + "epoch": 2.47890602269421, + "flow_matching_loss": 0.017552052810788154, + "grad_norm": 0.10267877578735352, + "learning_rate": 2.5729807907274254e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8520 + }, + { + "epoch": 2.4818155368053536, + "flow_matching_loss": 0.017653680127114057, + "grad_norm": 0.17677582800388336, + "learning_rate": 2.5718247001594403e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8530 + }, + { + "epoch": 2.484725050916497, + "flow_matching_loss": 0.01996102323755622, + "grad_norm": 0.12516221404075623, + "learning_rate": 2.5706673072233776e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8540 + }, + { + "epoch": 2.4876345650276406, + "flow_matching_loss": 0.016668419633060694, + "grad_norm": 0.12142006307840347, + "learning_rate": 2.5695086133255795e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8550 + }, + { + "epoch": 2.490544079138784, + "flow_matching_loss": 0.01914818277582526, + "grad_norm": 0.11591973155736923, + "learning_rate": 2.5683486198739687e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8560 + }, + { + "epoch": 2.493453593249927, + "flow_matching_loss": 0.021771664172410964, + "grad_norm": 0.1027606874704361, + "learning_rate": 2.5671873282780452e-05, + "loss": 0.0218, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8570 + }, + { + "epoch": 2.496363107361071, + "flow_matching_loss": 0.018516417872160674, + "grad_norm": 0.1943928748369217, + "learning_rate": 2.5660247399488887e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8580 + }, + { + "epoch": 2.499272621472214, + "flow_matching_loss": 0.017364402115345002, + "grad_norm": 0.13641197979450226, + "learning_rate": 2.564860856299154e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8590 + }, + { + "epoch": 2.5021821355833573, + "flow_matching_loss": 0.01722828410565853, + "grad_norm": 0.11722321063280106, + "learning_rate": 2.5636956787430693e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8600 + }, + { + "epoch": 2.505091649694501, + "flow_matching_loss": 0.019486129842698575, + "grad_norm": 0.11279989778995514, + "learning_rate": 2.5625292086964346e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8610 + }, + { + "epoch": 2.5080011638056443, + "flow_matching_loss": 0.018044565990567207, + "grad_norm": 0.09527190029621124, + "learning_rate": 2.5613614475766212e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8620 + }, + { + "epoch": 2.510910677916788, + "flow_matching_loss": 0.019480511732399463, + "grad_norm": 0.0912218913435936, + "learning_rate": 2.5601923968025694e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8630 + }, + { + "epoch": 2.5138201920279313, + "flow_matching_loss": 0.019488959200680257, + "grad_norm": 0.14378765225410461, + "learning_rate": 2.5590220577947862e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8640 + }, + { + "epoch": 2.516729706139075, + "flow_matching_loss": 0.018067649099975824, + "grad_norm": 0.1359298676252365, + "learning_rate": 2.5578504319753427e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8650 + }, + { + "epoch": 2.5196392202502182, + "flow_matching_loss": 0.017446452099829913, + "grad_norm": 0.07841566950082779, + "learning_rate": 2.556677520767876e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8660 + }, + { + "epoch": 2.5225487343613615, + "flow_matching_loss": 0.018837802577763796, + "grad_norm": 0.10287534445524216, + "learning_rate": 2.5555033255975835e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8670 + }, + { + "epoch": 2.525458248472505, + "flow_matching_loss": 0.019351077917963266, + "grad_norm": 0.12134762853384018, + "learning_rate": 2.5543278478912228e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8680 + }, + { + "epoch": 2.5283677625836485, + "flow_matching_loss": 0.01629063729196787, + "grad_norm": 0.12428268045186996, + "learning_rate": 2.55315108907711e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8690 + }, + { + "epoch": 2.5312772766947917, + "flow_matching_loss": 0.015176665410399437, + "grad_norm": 0.09856949001550674, + "learning_rate": 2.551973050585119e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8700 + }, + { + "epoch": 2.5341867908059355, + "flow_matching_loss": 0.020080906338989735, + "grad_norm": 0.13869337737560272, + "learning_rate": 2.5507937338466768e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8710 + }, + { + "epoch": 2.5370963049170787, + "flow_matching_loss": 0.018747863359749317, + "grad_norm": 0.11289133876562119, + "learning_rate": 2.549613140294764e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8720 + }, + { + "epoch": 2.5400058190282224, + "flow_matching_loss": 0.016904752887785434, + "grad_norm": 0.14586141705513, + "learning_rate": 2.548431271363914e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8730 + }, + { + "epoch": 2.5429153331393657, + "flow_matching_loss": 0.01755953850224614, + "grad_norm": 0.10746952891349792, + "learning_rate": 2.5472481284902092e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8740 + }, + { + "epoch": 2.5458248472505094, + "flow_matching_loss": 0.01592195304110646, + "grad_norm": 0.10186891257762909, + "learning_rate": 2.5460637131112788e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8750 + }, + { + "epoch": 2.5487343613616527, + "flow_matching_loss": 0.015822473634034395, + "grad_norm": 0.14161339402198792, + "learning_rate": 2.5448780266662995e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8760 + }, + { + "epoch": 2.551643875472796, + "flow_matching_loss": 0.020721744652837513, + "grad_norm": 0.14276030659675598, + "learning_rate": 2.5436910705959926e-05, + "loss": 0.0207, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8770 + }, + { + "epoch": 2.5545533895839396, + "flow_matching_loss": 0.019006393756717442, + "grad_norm": 0.09929308295249939, + "learning_rate": 2.542502846342621e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8780 + }, + { + "epoch": 2.557462903695083, + "flow_matching_loss": 0.01804532827809453, + "grad_norm": 0.15963177382946014, + "learning_rate": 2.5413133553499892e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8790 + }, + { + "epoch": 2.560372417806226, + "flow_matching_loss": 0.014831932168453932, + "grad_norm": 0.11650434881448746, + "learning_rate": 2.540122599063442e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8800 + }, + { + "epoch": 2.56328193191737, + "flow_matching_loss": 0.01597219193354249, + "grad_norm": 0.10672661662101746, + "learning_rate": 2.538930578929859e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8810 + }, + { + "epoch": 2.566191446028513, + "flow_matching_loss": 0.018311738222837447, + "grad_norm": 0.11531367897987366, + "learning_rate": 2.537737296397658e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8820 + }, + { + "epoch": 2.569100960139657, + "flow_matching_loss": 0.018171353731304407, + "grad_norm": 0.13357745110988617, + "learning_rate": 2.5365427529167903e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8830 + }, + { + "epoch": 2.5720104742508, + "flow_matching_loss": 0.020644170697778463, + "grad_norm": 0.12549816071987152, + "learning_rate": 2.5353469499387375e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8840 + }, + { + "epoch": 2.574919988361944, + "flow_matching_loss": 0.019695629738271235, + "grad_norm": 0.11018286645412445, + "learning_rate": 2.534149888916514e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8850 + }, + { + "epoch": 2.577829502473087, + "flow_matching_loss": 0.017654611356556417, + "grad_norm": 0.12477482110261917, + "learning_rate": 2.5329515713046616e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8860 + }, + { + "epoch": 2.5807390165842303, + "flow_matching_loss": 0.019994399417191742, + "grad_norm": 0.12826046347618103, + "learning_rate": 2.531751998559249e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8870 + }, + { + "epoch": 2.583648530695374, + "flow_matching_loss": 0.016907496843487024, + "grad_norm": 0.12897151708602905, + "learning_rate": 2.5305511721378703e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8880 + }, + { + "epoch": 2.5865580448065173, + "flow_matching_loss": 0.018340784776955842, + "grad_norm": 0.1113208681344986, + "learning_rate": 2.529349093499642e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8890 + }, + { + "epoch": 2.5894675589176606, + "flow_matching_loss": 0.01846808511763811, + "grad_norm": 0.09089729934930801, + "learning_rate": 2.528145764105205e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8900 + }, + { + "epoch": 2.5923770730288043, + "flow_matching_loss": 0.016681012604385613, + "grad_norm": 0.1333075612783432, + "learning_rate": 2.5269411854167166e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8910 + }, + { + "epoch": 2.5952865871399475, + "flow_matching_loss": 0.018110657390207053, + "grad_norm": 0.12166335433721542, + "learning_rate": 2.5257353588978534e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8920 + }, + { + "epoch": 2.598196101251091, + "flow_matching_loss": 0.021474546287208796, + "grad_norm": 0.09476490318775177, + "learning_rate": 2.5245282860138083e-05, + "loss": 0.0215, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8930 + }, + { + "epoch": 2.6011056153622345, + "flow_matching_loss": 0.01893823118880391, + "grad_norm": 0.13775447010993958, + "learning_rate": 2.5233199682312894e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8940 + }, + { + "epoch": 2.604015129473378, + "flow_matching_loss": 0.017000349890440702, + "grad_norm": 0.09897399693727493, + "learning_rate": 2.5221104070185163e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8950 + }, + { + "epoch": 2.6069246435845215, + "flow_matching_loss": 0.020492124557495116, + "grad_norm": 0.11504624038934708, + "learning_rate": 2.5208996038452204e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8960 + }, + { + "epoch": 2.6098341576956647, + "flow_matching_loss": 0.019105795491486787, + "grad_norm": 0.12346111983060837, + "learning_rate": 2.5196875601826403e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8970 + }, + { + "epoch": 2.6127436718068084, + "flow_matching_loss": 0.018135680258274077, + "grad_norm": 0.14503632485866547, + "learning_rate": 2.518474277503525e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8980 + }, + { + "epoch": 2.6156531859179517, + "flow_matching_loss": 0.019711473491042852, + "grad_norm": 0.11491666734218597, + "learning_rate": 2.5172597572821266e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8990 + }, + { + "epoch": 2.618562700029095, + "flow_matching_loss": 0.018357941415160894, + "grad_norm": 0.11367522925138474, + "learning_rate": 2.5160440009942015e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9000 + }, + { + "epoch": 2.6214722141402387, + "flow_matching_loss": 0.019697507563978433, + "grad_norm": 0.13070686161518097, + "learning_rate": 2.514827010117009e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9010 + }, + { + "epoch": 2.624381728251382, + "flow_matching_loss": 0.01901723342016339, + "grad_norm": 0.10969605296850204, + "learning_rate": 2.5136087861293068e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9020 + }, + { + "epoch": 2.627291242362525, + "flow_matching_loss": 0.01987721547484398, + "grad_norm": 0.11607061326503754, + "learning_rate": 2.5123893305113525e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9030 + }, + { + "epoch": 2.630200756473669, + "flow_matching_loss": 0.017318934388458728, + "grad_norm": 0.10792668163776398, + "learning_rate": 2.5111686447448997e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9040 + }, + { + "epoch": 2.633110270584812, + "flow_matching_loss": 0.017892606742680072, + "grad_norm": 0.10012281686067581, + "learning_rate": 2.5099467303131966e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9050 + }, + { + "epoch": 2.636019784695956, + "flow_matching_loss": 0.019041915144771337, + "grad_norm": 0.09521325677633286, + "learning_rate": 2.508723588700984e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9060 + }, + { + "epoch": 2.638929298807099, + "flow_matching_loss": 0.018794159963727, + "grad_norm": 0.14929801225662231, + "learning_rate": 2.5074992213944952e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9070 + }, + { + "epoch": 2.641838812918243, + "flow_matching_loss": 0.017727776430547237, + "grad_norm": 0.08827722072601318, + "learning_rate": 2.506273629881451e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9080 + }, + { + "epoch": 2.644748327029386, + "flow_matching_loss": 0.017265769094228743, + "grad_norm": 0.10038930177688599, + "learning_rate": 2.5050468156510617e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9090 + }, + { + "epoch": 2.6476578411405294, + "flow_matching_loss": 0.018754428438842296, + "grad_norm": 0.10447995364665985, + "learning_rate": 2.5038187801940215e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9100 + }, + { + "epoch": 2.650567355251673, + "flow_matching_loss": 0.016702421009540558, + "grad_norm": 0.1396123617887497, + "learning_rate": 2.5025895250025088e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9110 + }, + { + "epoch": 2.6534768693628163, + "flow_matching_loss": 0.019304819405078888, + "grad_norm": 0.15071332454681396, + "learning_rate": 2.501359051570186e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9120 + }, + { + "epoch": 2.6563863834739596, + "flow_matching_loss": 0.016930746287107466, + "grad_norm": 0.08059664070606232, + "learning_rate": 2.500127361392193e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9130 + }, + { + "epoch": 2.6592958975851033, + "flow_matching_loss": 0.01640694970265031, + "grad_norm": 0.0869600847363472, + "learning_rate": 2.49889445596515e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9140 + }, + { + "epoch": 2.6622054116962466, + "flow_matching_loss": 0.021432064566761257, + "grad_norm": 0.1333371102809906, + "learning_rate": 2.497660336787154e-05, + "loss": 0.0214, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9150 + }, + { + "epoch": 2.6651149258073903, + "flow_matching_loss": 0.015290356427431106, + "grad_norm": 0.10086368769407272, + "learning_rate": 2.496425005357775e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9160 + }, + { + "epoch": 2.6680244399185336, + "flow_matching_loss": 0.018643063493072987, + "grad_norm": 0.0996362715959549, + "learning_rate": 2.4951884631780578e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9170 + }, + { + "epoch": 2.6709339540296773, + "flow_matching_loss": 0.01684821117669344, + "grad_norm": 0.10013402998447418, + "learning_rate": 2.493950711750518e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9180 + }, + { + "epoch": 2.6738434681408205, + "flow_matching_loss": 0.020808419212698936, + "grad_norm": 0.11447049677371979, + "learning_rate": 2.49271175257914e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9190 + }, + { + "epoch": 2.676752982251964, + "flow_matching_loss": 0.018195441365242003, + "grad_norm": 0.09246514737606049, + "learning_rate": 2.4914715871693757e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9200 + }, + { + "epoch": 2.6796624963631075, + "flow_matching_loss": 0.018711960408836603, + "grad_norm": 0.12238156050443649, + "learning_rate": 2.4902302170281438e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9210 + }, + { + "epoch": 2.6825720104742508, + "flow_matching_loss": 0.01853279732167721, + "grad_norm": 0.1322554349899292, + "learning_rate": 2.4889876436638252e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9220 + }, + { + "epoch": 2.685481524585394, + "flow_matching_loss": 0.01914047123864293, + "grad_norm": 0.10900459438562393, + "learning_rate": 2.4877438685862643e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9230 + }, + { + "epoch": 2.6883910386965377, + "flow_matching_loss": 0.016694821789860724, + "grad_norm": 0.1145913153886795, + "learning_rate": 2.486498893306765e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9240 + }, + { + "epoch": 2.691300552807681, + "flow_matching_loss": 0.01820406150072813, + "grad_norm": 0.12925522029399872, + "learning_rate": 2.485252719338089e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9250 + }, + { + "epoch": 2.6942100669188247, + "flow_matching_loss": 0.018606740329414607, + "grad_norm": 0.10851932317018509, + "learning_rate": 2.4840053481944563e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9260 + }, + { + "epoch": 2.697119581029968, + "flow_matching_loss": 0.021229518018662928, + "grad_norm": 0.1254948377609253, + "learning_rate": 2.4827567813915398e-05, + "loss": 0.0212, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9270 + }, + { + "epoch": 2.7000290951411117, + "flow_matching_loss": 0.018862832430750133, + "grad_norm": 0.09968658536672592, + "learning_rate": 2.481507020446466e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9280 + }, + { + "epoch": 2.702938609252255, + "flow_matching_loss": 0.01657056463882327, + "grad_norm": 0.13691283762454987, + "learning_rate": 2.4802560668778122e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9290 + }, + { + "epoch": 2.705848123363398, + "flow_matching_loss": 0.01873939875513315, + "grad_norm": 0.13037773966789246, + "learning_rate": 2.479003922205605e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9300 + }, + { + "epoch": 2.708757637474542, + "flow_matching_loss": 0.01775156017392874, + "grad_norm": 0.12276570498943329, + "learning_rate": 2.4777505879513183e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9310 + }, + { + "epoch": 2.711667151585685, + "flow_matching_loss": 0.015943690203130245, + "grad_norm": 0.09815452992916107, + "learning_rate": 2.4764960656378714e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9320 + }, + { + "epoch": 2.7145766656968284, + "flow_matching_loss": 0.017172303795814515, + "grad_norm": 0.14564093947410583, + "learning_rate": 2.4752403567896274e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9330 + }, + { + "epoch": 2.717486179807972, + "flow_matching_loss": 0.020977266319096088, + "grad_norm": 0.09277895092964172, + "learning_rate": 2.47398346293239e-05, + "loss": 0.021, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9340 + }, + { + "epoch": 2.7203956939191154, + "flow_matching_loss": 0.01857603481039405, + "grad_norm": 0.1031019315123558, + "learning_rate": 2.4727253855934055e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9350 + }, + { + "epoch": 2.723305208030259, + "flow_matching_loss": 0.017864160146564244, + "grad_norm": 0.10502457618713379, + "learning_rate": 2.4714661263013548e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9360 + }, + { + "epoch": 2.7262147221414024, + "flow_matching_loss": 0.01679808972403407, + "grad_norm": 0.09576518833637238, + "learning_rate": 2.4702056865863575e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9370 + }, + { + "epoch": 2.729124236252546, + "flow_matching_loss": 0.015140633191913366, + "grad_norm": 0.10912797600030899, + "learning_rate": 2.4689440679799672e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9380 + }, + { + "epoch": 2.7320337503636893, + "flow_matching_loss": 0.018017912097275258, + "grad_norm": 0.11989324539899826, + "learning_rate": 2.4676812720151678e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9390 + }, + { + "epoch": 2.7349432644748326, + "flow_matching_loss": 0.018241909611970187, + "grad_norm": 0.12084964662790298, + "learning_rate": 2.466417300226377e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9400 + }, + { + "epoch": 2.7378527785859763, + "flow_matching_loss": 0.019421902671456336, + "grad_norm": 0.10170818865299225, + "learning_rate": 2.4651521541494394e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9410 + }, + { + "epoch": 2.7407622926971196, + "flow_matching_loss": 0.015686621889472008, + "grad_norm": 0.11352090537548065, + "learning_rate": 2.4638858353216267e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9420 + }, + { + "epoch": 2.743671806808263, + "flow_matching_loss": 0.017482594400644303, + "grad_norm": 0.09335094690322876, + "learning_rate": 2.4626183452816355e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9430 + }, + { + "epoch": 2.7465813209194065, + "flow_matching_loss": 0.017492544278502463, + "grad_norm": 0.12270035594701767, + "learning_rate": 2.4613496855695855e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9440 + }, + { + "epoch": 2.74949083503055, + "flow_matching_loss": 0.017922431509941817, + "grad_norm": 0.15701769292354584, + "learning_rate": 2.4600798577270183e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9450 + }, + { + "epoch": 2.752400349141693, + "flow_matching_loss": 0.019336151517927648, + "grad_norm": 0.1461864411830902, + "learning_rate": 2.4588088632968944e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9460 + }, + { + "epoch": 2.7553098632528368, + "flow_matching_loss": 0.018631018698215485, + "grad_norm": 0.08929374814033508, + "learning_rate": 2.4575367038235916e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9470 + }, + { + "epoch": 2.7582193773639805, + "flow_matching_loss": 0.017927236575633287, + "grad_norm": 0.10746867954730988, + "learning_rate": 2.4562633808529043e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9480 + }, + { + "epoch": 2.7611288914751237, + "flow_matching_loss": 0.017648316361010073, + "grad_norm": 0.08776035159826279, + "learning_rate": 2.4549888959320396e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9490 + }, + { + "epoch": 2.764038405586267, + "flow_matching_loss": 0.015023109037429094, + "grad_norm": 0.10450907796621323, + "learning_rate": 2.4537132506096162e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9500 + }, + { + "epoch": 2.7669479196974107, + "flow_matching_loss": 0.018658894579857588, + "grad_norm": 0.11059699952602386, + "learning_rate": 2.452436446435664e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9510 + }, + { + "epoch": 2.769857433808554, + "flow_matching_loss": 0.019413833506405354, + "grad_norm": 0.11791469901800156, + "learning_rate": 2.451158484961621e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9520 + }, + { + "epoch": 2.7727669479196972, + "flow_matching_loss": 0.014963779971003533, + "grad_norm": 0.10038059204816818, + "learning_rate": 2.4498793677403304e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9530 + }, + { + "epoch": 2.775676462030841, + "flow_matching_loss": 0.01780060539022088, + "grad_norm": 0.08903767913579941, + "learning_rate": 2.4485990963260398e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9540 + }, + { + "epoch": 2.778585976141984, + "flow_matching_loss": 0.017459281906485557, + "grad_norm": 0.13311856985092163, + "learning_rate": 2.4473176722744e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9550 + }, + { + "epoch": 2.7814954902531275, + "flow_matching_loss": 0.015717105753719807, + "grad_norm": 0.10795249789953232, + "learning_rate": 2.446035097142463e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9560 + }, + { + "epoch": 2.784405004364271, + "flow_matching_loss": 0.018468617647886276, + "grad_norm": 0.1450192630290985, + "learning_rate": 2.444751372488677e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9570 + }, + { + "epoch": 2.7873145184754144, + "flow_matching_loss": 0.01978381248190999, + "grad_norm": 0.15731918811798096, + "learning_rate": 2.443466499872889e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9580 + }, + { + "epoch": 2.790224032586558, + "flow_matching_loss": 0.018316347897052766, + "grad_norm": 0.1473993957042694, + "learning_rate": 2.4421804808563413e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9590 + }, + { + "epoch": 2.7931335466977014, + "flow_matching_loss": 0.017746127117425202, + "grad_norm": 0.09787359088659286, + "learning_rate": 2.440893317001667e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9600 + }, + { + "epoch": 2.796043060808845, + "flow_matching_loss": 0.0165681267157197, + "grad_norm": 0.08473402261734009, + "learning_rate": 2.4396050098728925e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9610 + }, + { + "epoch": 2.7989525749199884, + "flow_matching_loss": 0.017780386842787266, + "grad_norm": 0.09737222641706467, + "learning_rate": 2.4383155610354314e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9620 + }, + { + "epoch": 2.8018620890311317, + "flow_matching_loss": 0.016885223053395747, + "grad_norm": 0.09918737411499023, + "learning_rate": 2.437024972056087e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9630 + }, + { + "epoch": 2.8047716031422754, + "flow_matching_loss": 0.018101406283676624, + "grad_norm": 0.11662585288286209, + "learning_rate": 2.4357332445030452e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9640 + }, + { + "epoch": 2.8076811172534186, + "flow_matching_loss": 0.017353063076734544, + "grad_norm": 0.08747350424528122, + "learning_rate": 2.434440379945878e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9650 + }, + { + "epoch": 2.810590631364562, + "flow_matching_loss": 0.015075351111590862, + "grad_norm": 0.12871745228767395, + "learning_rate": 2.4331463799555366e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9660 + }, + { + "epoch": 2.8135001454757056, + "flow_matching_loss": 0.01685200147330761, + "grad_norm": 0.10702962428331375, + "learning_rate": 2.4318512461043544e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9670 + }, + { + "epoch": 2.816409659586849, + "flow_matching_loss": 0.015541659109294415, + "grad_norm": 0.09791617095470428, + "learning_rate": 2.4305549799660397e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9680 + }, + { + "epoch": 2.8193191736979926, + "flow_matching_loss": 0.016661832761019468, + "grad_norm": 0.09886577725410461, + "learning_rate": 2.429257583115679e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9690 + }, + { + "epoch": 2.822228687809136, + "flow_matching_loss": 0.017360482923686506, + "grad_norm": 0.06914464384317398, + "learning_rate": 2.4279590571297316e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9700 + }, + { + "epoch": 2.8251382019202795, + "flow_matching_loss": 0.01740957023575902, + "grad_norm": 0.13738781213760376, + "learning_rate": 2.4266594035860288e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9710 + }, + { + "epoch": 2.828047716031423, + "flow_matching_loss": 0.01752264518290758, + "grad_norm": 0.11445640027523041, + "learning_rate": 2.425358624063773e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9720 + }, + { + "epoch": 2.830957230142566, + "flow_matching_loss": 0.01471738638356328, + "grad_norm": 0.10679417848587036, + "learning_rate": 2.424056720143532e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9730 + }, + { + "epoch": 2.8338667442537098, + "flow_matching_loss": 0.019716431759297846, + "grad_norm": 0.10168182104825974, + "learning_rate": 2.422753693407244e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9740 + }, + { + "epoch": 2.836776258364853, + "flow_matching_loss": 0.018578156363219023, + "grad_norm": 0.11125647276639938, + "learning_rate": 2.421449545438209e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9750 + }, + { + "epoch": 2.8396857724759963, + "flow_matching_loss": 0.017307613603770734, + "grad_norm": 0.0808723047375679, + "learning_rate": 2.4201442778210887e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9760 + }, + { + "epoch": 2.84259528658714, + "flow_matching_loss": 0.01769512863829732, + "grad_norm": 0.12272462248802185, + "learning_rate": 2.418837892141907e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9770 + }, + { + "epoch": 2.8455048006982833, + "flow_matching_loss": 0.02004918046295643, + "grad_norm": 0.12469298392534256, + "learning_rate": 2.4175303899880458e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9780 + }, + { + "epoch": 2.848414314809427, + "flow_matching_loss": 0.019331635162234306, + "grad_norm": 0.10463939607143402, + "learning_rate": 2.4162217729482427e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9790 + }, + { + "epoch": 2.8513238289205702, + "flow_matching_loss": 0.015476855635643005, + "grad_norm": 0.09390590339899063, + "learning_rate": 2.4149120426125918e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9800 + }, + { + "epoch": 2.854233343031714, + "flow_matching_loss": 0.015193192567676306, + "grad_norm": 0.11378953605890274, + "learning_rate": 2.413601200572538e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9810 + }, + { + "epoch": 2.857142857142857, + "flow_matching_loss": 0.01657526018097997, + "grad_norm": 0.0737103521823883, + "learning_rate": 2.4122892484208794e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9820 + }, + { + "epoch": 2.8600523712540005, + "flow_matching_loss": 0.016937067918479442, + "grad_norm": 0.08735332638025284, + "learning_rate": 2.4109761877517603e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9830 + }, + { + "epoch": 2.862961885365144, + "flow_matching_loss": 0.01758886268362403, + "grad_norm": 0.09700547903776169, + "learning_rate": 2.4096620201606737e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9840 + }, + { + "epoch": 2.8658713994762874, + "flow_matching_loss": 0.02008426235988736, + "grad_norm": 0.1742132157087326, + "learning_rate": 2.4083467472444573e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9850 + }, + { + "epoch": 2.8687809135874307, + "flow_matching_loss": 0.017028515040874482, + "grad_norm": 0.136953204870224, + "learning_rate": 2.4070303706012912e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9860 + }, + { + "epoch": 2.8716904276985744, + "flow_matching_loss": 0.01861600819975138, + "grad_norm": 0.11618804186582565, + "learning_rate": 2.4057128918306975e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9870 + }, + { + "epoch": 2.8745999418097177, + "flow_matching_loss": 0.017757633421570062, + "grad_norm": 0.11619840562343597, + "learning_rate": 2.4043943125335373e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9880 + }, + { + "epoch": 2.8775094559208614, + "flow_matching_loss": 0.01839942131191492, + "grad_norm": 0.09752203524112701, + "learning_rate": 2.403074634312008e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9890 + }, + { + "epoch": 2.8804189700320046, + "flow_matching_loss": 0.01724341232329607, + "grad_norm": 0.09007328003644943, + "learning_rate": 2.4017538587696442e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9900 + }, + { + "epoch": 2.8833284841431484, + "flow_matching_loss": 0.016943175811320543, + "grad_norm": 0.1364452987909317, + "learning_rate": 2.4004319875113118e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9910 + }, + { + "epoch": 2.8862379982542916, + "flow_matching_loss": 0.020891787391155957, + "grad_norm": 0.07505422830581665, + "learning_rate": 2.3991090221432096e-05, + "loss": 0.0209, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9920 + }, + { + "epoch": 2.889147512365435, + "flow_matching_loss": 0.021578687895089387, + "grad_norm": 0.1287870556116104, + "learning_rate": 2.397784964272865e-05, + "loss": 0.0216, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9930 + }, + { + "epoch": 2.8920570264765786, + "flow_matching_loss": 0.0175129315815866, + "grad_norm": 0.11321669816970825, + "learning_rate": 2.3964598155091335e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9940 + }, + { + "epoch": 2.894966540587722, + "flow_matching_loss": 0.018398033268749715, + "grad_norm": 0.10984573513269424, + "learning_rate": 2.3951335774621952e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9950 + }, + { + "epoch": 2.897876054698865, + "flow_matching_loss": 0.01679343041032553, + "grad_norm": 0.11517667025327682, + "learning_rate": 2.3938062517435548e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9960 + }, + { + "epoch": 2.900785568810009, + "flow_matching_loss": 0.016608820017427205, + "grad_norm": 0.10601761192083359, + "learning_rate": 2.392477839966038e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9970 + }, + { + "epoch": 2.903695082921152, + "flow_matching_loss": 0.017462366074323655, + "grad_norm": 0.1268858164548874, + "learning_rate": 2.3911483437437904e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9980 + }, + { + "epoch": 2.9066045970322953, + "flow_matching_loss": 0.016092582140117884, + "grad_norm": 0.10744545608758926, + "learning_rate": 2.3898177646922757e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9990 + }, + { + "epoch": 2.909514111143439, + "flow_matching_loss": 0.01754728490486741, + "grad_norm": 0.10470502078533173, + "learning_rate": 2.3884861044282722e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10000 + }, + { + "epoch": 2.9124236252545828, + "flow_matching_loss": 0.016317529138177633, + "grad_norm": 0.10439707338809967, + "learning_rate": 2.3871533645698732e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10010 + }, + { + "epoch": 2.915333139365726, + "flow_matching_loss": 0.018675052281469106, + "grad_norm": 0.1266223043203354, + "learning_rate": 2.3858195467364833e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10020 + }, + { + "epoch": 2.9182426534768693, + "flow_matching_loss": 0.018477196618914605, + "grad_norm": 0.09685253351926804, + "learning_rate": 2.3844846525488166e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10030 + }, + { + "epoch": 2.921152167588013, + "flow_matching_loss": 0.016728088445961475, + "grad_norm": 0.09574057161808014, + "learning_rate": 2.3831486836288957e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10040 + }, + { + "epoch": 2.9240616816991563, + "flow_matching_loss": 0.016437359154224396, + "grad_norm": 0.10411965847015381, + "learning_rate": 2.381811641600048e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10050 + }, + { + "epoch": 2.9269711958102995, + "flow_matching_loss": 0.015908531937748195, + "grad_norm": 0.09857218712568283, + "learning_rate": 2.3804735280869073e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10060 + }, + { + "epoch": 2.9298807099214432, + "flow_matching_loss": 0.016936283186078073, + "grad_norm": 0.09966739267110825, + "learning_rate": 2.3791343447154064e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10070 + }, + { + "epoch": 2.9327902240325865, + "flow_matching_loss": 0.018945996277034284, + "grad_norm": 0.12095125019550323, + "learning_rate": 2.37779409311278e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10080 + }, + { + "epoch": 2.9356997381437298, + "flow_matching_loss": 0.016743403859436512, + "grad_norm": 0.10312169045209885, + "learning_rate": 2.3764527749075596e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10090 + }, + { + "epoch": 2.9386092522548735, + "flow_matching_loss": 0.018618556391447782, + "grad_norm": 0.12963880598545074, + "learning_rate": 2.3751103917295735e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10100 + }, + { + "epoch": 2.9415187663660167, + "flow_matching_loss": 0.018892863392829896, + "grad_norm": 0.11460694670677185, + "learning_rate": 2.373766945209944e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10110 + }, + { + "epoch": 2.9444282804771604, + "flow_matching_loss": 0.01756260162219405, + "grad_norm": 0.1071109026670456, + "learning_rate": 2.3724224369810856e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10120 + }, + { + "epoch": 2.9473377945883037, + "flow_matching_loss": 0.01608498189598322, + "grad_norm": 0.08531112223863602, + "learning_rate": 2.3710768686767015e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10130 + }, + { + "epoch": 2.9502473086994474, + "flow_matching_loss": 0.016982991620898245, + "grad_norm": 0.10533571988344193, + "learning_rate": 2.3697302419317856e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10140 + }, + { + "epoch": 2.9531568228105907, + "flow_matching_loss": 0.014702930580824614, + "grad_norm": 0.0773758813738823, + "learning_rate": 2.3683825583826146e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10150 + }, + { + "epoch": 2.956066336921734, + "flow_matching_loss": 0.016111402958631515, + "grad_norm": 0.1195482462644577, + "learning_rate": 2.3670338196667528e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10160 + }, + { + "epoch": 2.9589758510328776, + "flow_matching_loss": 0.020318750757724047, + "grad_norm": 0.13365133106708527, + "learning_rate": 2.3656840274230443e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10170 + }, + { + "epoch": 2.961885365144021, + "flow_matching_loss": 0.017067603580653668, + "grad_norm": 0.11440785974264145, + "learning_rate": 2.3643331832916133e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10180 + }, + { + "epoch": 2.964794879255164, + "flow_matching_loss": 0.014671525824815035, + "grad_norm": 0.10639658570289612, + "learning_rate": 2.362981288913863e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10190 + }, + { + "epoch": 2.967704393366308, + "flow_matching_loss": 0.016085063852369786, + "grad_norm": 0.17825418710708618, + "learning_rate": 2.3616283459324737e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10200 + }, + { + "epoch": 2.970613907477451, + "flow_matching_loss": 0.016763005591928958, + "grad_norm": 0.10118815302848816, + "learning_rate": 2.3602743559913973e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10210 + }, + { + "epoch": 2.973523421588595, + "flow_matching_loss": 0.019717623945325613, + "grad_norm": 0.13510224223136902, + "learning_rate": 2.35891932073586e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10220 + }, + { + "epoch": 2.976432935699738, + "flow_matching_loss": 0.01604130519554019, + "grad_norm": 0.09963656216859818, + "learning_rate": 2.357563241812357e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10230 + }, + { + "epoch": 2.979342449810882, + "flow_matching_loss": 0.0159688962623477, + "grad_norm": 0.13893656432628632, + "learning_rate": 2.3562061208686522e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10240 + }, + { + "epoch": 2.982251963922025, + "flow_matching_loss": 0.015565995220094919, + "grad_norm": 0.13211432099342346, + "learning_rate": 2.354847959553776e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10250 + }, + { + "epoch": 2.9851614780331683, + "flow_matching_loss": 0.018195226229727267, + "grad_norm": 0.11872857064008713, + "learning_rate": 2.353488759518022e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10260 + }, + { + "epoch": 2.988070992144312, + "flow_matching_loss": 0.018017878010869025, + "grad_norm": 0.09985178709030151, + "learning_rate": 2.352128522412946e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10270 + }, + { + "epoch": 2.9909805062554553, + "flow_matching_loss": 0.01866810694336891, + "grad_norm": 0.11552257090806961, + "learning_rate": 2.3507672498913653e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10280 + }, + { + "epoch": 2.9938900203665986, + "flow_matching_loss": 0.0167852483689785, + "grad_norm": 0.11633085459470749, + "learning_rate": 2.3494049436073537e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10290 + }, + { + "epoch": 2.9967995344777423, + "flow_matching_loss": 0.016972436103969812, + "grad_norm": 0.14454902708530426, + "learning_rate": 2.348041605216242e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10300 + }, + { + "epoch": 2.9997090485888855, + "flow_matching_loss": 0.018019679933786392, + "grad_norm": 0.08419682830572128, + "learning_rate": 2.3466772363746146e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10310 + }, + { + "epoch": 3.0026185627000292, + "flow_matching_loss": 0.015126134175807238, + "grad_norm": 0.10072670131921768, + "learning_rate": 2.345311838740309e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10320 + }, + { + "epoch": 3.0055280768111725, + "flow_matching_loss": 0.018482847046107052, + "grad_norm": 0.11606498062610626, + "learning_rate": 2.343945413972411e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10330 + }, + { + "epoch": 3.0084375909223158, + "flow_matching_loss": 0.017323090136051177, + "grad_norm": 0.13375163078308105, + "learning_rate": 2.3425779637312568e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10340 + }, + { + "epoch": 3.0113471050334595, + "flow_matching_loss": 0.0179236164316535, + "grad_norm": 0.1048947274684906, + "learning_rate": 2.3412094896784264e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10350 + }, + { + "epoch": 3.0142566191446027, + "flow_matching_loss": 0.017225937452167273, + "grad_norm": 0.11143108457326889, + "learning_rate": 2.339839993476745e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10360 + }, + { + "epoch": 3.0171661332557465, + "flow_matching_loss": 0.016434815619140862, + "grad_norm": 0.09285429120063782, + "learning_rate": 2.3384694767902804e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10370 + }, + { + "epoch": 3.0200756473668897, + "flow_matching_loss": 0.016194479912519454, + "grad_norm": 0.08686324208974838, + "learning_rate": 2.3370979412843384e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10380 + }, + { + "epoch": 3.022985161478033, + "flow_matching_loss": 0.01770535483956337, + "grad_norm": 0.08524306863546371, + "learning_rate": 2.3357253886254648e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10390 + }, + { + "epoch": 3.0258946755891767, + "flow_matching_loss": 0.018003031518310307, + "grad_norm": 0.11782679706811905, + "learning_rate": 2.3343518204814407e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10400 + }, + { + "epoch": 3.02880418970032, + "flow_matching_loss": 0.016942134778946637, + "grad_norm": 0.10616131126880646, + "learning_rate": 2.3329772385212798e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10410 + }, + { + "epoch": 3.0317137038114637, + "flow_matching_loss": 0.016461779829114676, + "grad_norm": 0.13641123473644257, + "learning_rate": 2.3316016444152296e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10420 + }, + { + "epoch": 3.034623217922607, + "flow_matching_loss": 0.014909493271261454, + "grad_norm": 0.06871248781681061, + "learning_rate": 2.3302250398347668e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10430 + }, + { + "epoch": 3.03753273203375, + "flow_matching_loss": 0.016172694880515336, + "grad_norm": 0.108500637114048, + "learning_rate": 2.328847426452595e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10440 + }, + { + "epoch": 3.040442246144894, + "flow_matching_loss": 0.01703843893483281, + "grad_norm": 0.10586956143379211, + "learning_rate": 2.3274688059426456e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10450 + }, + { + "epoch": 3.043351760256037, + "flow_matching_loss": 0.018480130285024644, + "grad_norm": 0.12220495194196701, + "learning_rate": 2.3260891799800715e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10460 + }, + { + "epoch": 3.046261274367181, + "flow_matching_loss": 0.015348528139293193, + "grad_norm": 0.0967290848493576, + "learning_rate": 2.3247085502412482e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10470 + }, + { + "epoch": 3.049170788478324, + "flow_matching_loss": 0.017317990958690643, + "grad_norm": 0.11687956005334854, + "learning_rate": 2.3233269184037717e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10480 + }, + { + "epoch": 3.0520803025894674, + "flow_matching_loss": 0.018929157964885236, + "grad_norm": 0.09205896407365799, + "learning_rate": 2.3219442861464547e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10490 + }, + { + "epoch": 3.054989816700611, + "flow_matching_loss": 0.02002443019300699, + "grad_norm": 0.10793694853782654, + "learning_rate": 2.3205606551493255e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10500 + }, + { + "epoch": 3.0578993308117544, + "flow_matching_loss": 0.014948683977127075, + "grad_norm": 0.11142867803573608, + "learning_rate": 2.319176027093627e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10510 + }, + { + "epoch": 3.060808844922898, + "flow_matching_loss": 0.016024147998541592, + "grad_norm": 0.14907735586166382, + "learning_rate": 2.3177904036618117e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10520 + }, + { + "epoch": 3.0637183590340413, + "flow_matching_loss": 0.019831259548664094, + "grad_norm": 0.11247463524341583, + "learning_rate": 2.316403786537544e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10530 + }, + { + "epoch": 3.0666278731451846, + "flow_matching_loss": 0.01797578949481249, + "grad_norm": 0.10682319104671478, + "learning_rate": 2.3150161774056933e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10540 + }, + { + "epoch": 3.0695373872563283, + "flow_matching_loss": 0.018484918121248483, + "grad_norm": 0.14719845354557037, + "learning_rate": 2.313627577952336e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10550 + }, + { + "epoch": 3.0724469013674716, + "flow_matching_loss": 0.015501275286078453, + "grad_norm": 0.15278083086013794, + "learning_rate": 2.312237989864752e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10560 + }, + { + "epoch": 3.0753564154786153, + "flow_matching_loss": 0.017269372660666703, + "grad_norm": 0.0841541662812233, + "learning_rate": 2.3108474148314205e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10570 + }, + { + "epoch": 3.0782659295897585, + "flow_matching_loss": 0.015326974913477898, + "grad_norm": 0.06376605480909348, + "learning_rate": 2.309455854542023e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10580 + }, + { + "epoch": 3.081175443700902, + "flow_matching_loss": 0.01658603884279728, + "grad_norm": 0.09831462800502777, + "learning_rate": 2.308063310687435e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10590 + }, + { + "epoch": 3.0840849578120455, + "flow_matching_loss": 0.016292280238121747, + "grad_norm": 0.09819848835468292, + "learning_rate": 2.306669784959729e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10600 + }, + { + "epoch": 3.0869944719231888, + "flow_matching_loss": 0.015205122530460358, + "grad_norm": 0.10081220418214798, + "learning_rate": 2.3052752790521703e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10610 + }, + { + "epoch": 3.0899039860343325, + "flow_matching_loss": 0.01849516239017248, + "grad_norm": 0.09997093677520752, + "learning_rate": 2.3038797946592152e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10620 + }, + { + "epoch": 3.0928135001454757, + "flow_matching_loss": 0.01744763944298029, + "grad_norm": 0.0980069562792778, + "learning_rate": 2.3024833334765087e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10630 + }, + { + "epoch": 3.095723014256619, + "flow_matching_loss": 0.02000811751931906, + "grad_norm": 0.09620320796966553, + "learning_rate": 2.301085897200883e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10640 + }, + { + "epoch": 3.0986325283677627, + "flow_matching_loss": 0.018227444868534802, + "grad_norm": 0.09213646501302719, + "learning_rate": 2.299687487530354e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10650 + }, + { + "epoch": 3.101542042478906, + "flow_matching_loss": 0.013959370367228986, + "grad_norm": 0.09715485572814941, + "learning_rate": 2.2982881061641228e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10660 + }, + { + "epoch": 3.1044515565900497, + "flow_matching_loss": 0.016296377312391995, + "grad_norm": 0.09127617627382278, + "learning_rate": 2.296887754802569e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10670 + }, + { + "epoch": 3.107361070701193, + "flow_matching_loss": 0.018549071904271842, + "grad_norm": 0.11791244149208069, + "learning_rate": 2.295486435147251e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10680 + }, + { + "epoch": 3.110270584812336, + "flow_matching_loss": 0.017758280411362647, + "grad_norm": 0.11405063420534134, + "learning_rate": 2.294084148900905e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10690 + }, + { + "epoch": 3.11318009892348, + "flow_matching_loss": 0.01741497255861759, + "grad_norm": 0.09575730562210083, + "learning_rate": 2.292680897767441e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10700 + }, + { + "epoch": 3.116089613034623, + "flow_matching_loss": 0.01821881579235196, + "grad_norm": 0.08646342903375626, + "learning_rate": 2.291276683451941e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10710 + }, + { + "epoch": 3.1189991271457664, + "flow_matching_loss": 0.01838805992156267, + "grad_norm": 0.09808541089296341, + "learning_rate": 2.2898715076606584e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10720 + }, + { + "epoch": 3.12190864125691, + "flow_matching_loss": 0.017404169123619796, + "grad_norm": 0.11980163305997849, + "learning_rate": 2.2884653721010135e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10730 + }, + { + "epoch": 3.1248181553680534, + "flow_matching_loss": 0.017665161285549402, + "grad_norm": 0.08293423801660538, + "learning_rate": 2.2870582784815946e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10740 + }, + { + "epoch": 3.127727669479197, + "flow_matching_loss": 0.01661165514960885, + "grad_norm": 0.10824603587388992, + "learning_rate": 2.2856502285121525e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10750 + }, + { + "epoch": 3.1306371835903404, + "flow_matching_loss": 0.01735125118866563, + "grad_norm": 0.08644290268421173, + "learning_rate": 2.2842412239036004e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10760 + }, + { + "epoch": 3.133546697701484, + "flow_matching_loss": 0.016071054339408874, + "grad_norm": 0.11482983827590942, + "learning_rate": 2.2828312663680125e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10770 + }, + { + "epoch": 3.1364562118126273, + "flow_matching_loss": 0.014091669302433728, + "grad_norm": 0.06087778881192207, + "learning_rate": 2.2814203576186194e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10780 + }, + { + "epoch": 3.1393657259237706, + "flow_matching_loss": 0.015398448705673218, + "grad_norm": 0.08957405388355255, + "learning_rate": 2.2800084993698093e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10790 + }, + { + "epoch": 3.1422752400349143, + "flow_matching_loss": 0.01842161752283573, + "grad_norm": 0.11071638017892838, + "learning_rate": 2.2785956933371224e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10800 + }, + { + "epoch": 3.1451847541460576, + "flow_matching_loss": 0.018017819337546825, + "grad_norm": 0.13078415393829346, + "learning_rate": 2.2771819412372505e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10810 + }, + { + "epoch": 3.148094268257201, + "flow_matching_loss": 0.01753325602039695, + "grad_norm": 0.11918125301599503, + "learning_rate": 2.2757672447880374e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10820 + }, + { + "epoch": 3.1510037823683446, + "flow_matching_loss": 0.015870712138712405, + "grad_norm": 0.10457432270050049, + "learning_rate": 2.2743516057084712e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10830 + }, + { + "epoch": 3.153913296479488, + "flow_matching_loss": 0.016512397117912768, + "grad_norm": 0.11132578551769257, + "learning_rate": 2.2729350257186877e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10840 + }, + { + "epoch": 3.1568228105906315, + "flow_matching_loss": 0.017727589420974254, + "grad_norm": 0.11217792332172394, + "learning_rate": 2.2715175065399655e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10850 + }, + { + "epoch": 3.159732324701775, + "flow_matching_loss": 0.016912432666867972, + "grad_norm": 0.12988872826099396, + "learning_rate": 2.270099049894723e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10860 + }, + { + "epoch": 3.162641838812918, + "flow_matching_loss": 0.017145905923098326, + "grad_norm": 0.07921086996793747, + "learning_rate": 2.268679657506519e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10870 + }, + { + "epoch": 3.1655513529240618, + "flow_matching_loss": 0.0170176618732512, + "grad_norm": 0.11568465828895569, + "learning_rate": 2.2672593311000498e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10880 + }, + { + "epoch": 3.168460867035205, + "flow_matching_loss": 0.016752590984106065, + "grad_norm": 0.18105168640613556, + "learning_rate": 2.2658380724011452e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10890 + }, + { + "epoch": 3.1713703811463487, + "flow_matching_loss": 0.01617611153051257, + "grad_norm": 0.12367796152830124, + "learning_rate": 2.264415883136769e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10900 + }, + { + "epoch": 3.174279895257492, + "flow_matching_loss": 0.017189905606210233, + "grad_norm": 0.09360115975141525, + "learning_rate": 2.262992765035015e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10910 + }, + { + "epoch": 3.1771894093686353, + "flow_matching_loss": 0.01694434005767107, + "grad_norm": 0.09885549545288086, + "learning_rate": 2.2615687198251062e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10920 + }, + { + "epoch": 3.180098923479779, + "flow_matching_loss": 0.017331210430711508, + "grad_norm": 0.1371932327747345, + "learning_rate": 2.2601437492373914e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10930 + }, + { + "epoch": 3.1830084375909222, + "flow_matching_loss": 0.01680336371064186, + "grad_norm": 0.12286802381277084, + "learning_rate": 2.2587178550033444e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10940 + }, + { + "epoch": 3.185917951702066, + "flow_matching_loss": 0.01528843753039837, + "grad_norm": 0.13752678036689758, + "learning_rate": 2.2572910388555615e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10950 + }, + { + "epoch": 3.188827465813209, + "flow_matching_loss": 0.015500336792320013, + "grad_norm": 0.11631854623556137, + "learning_rate": 2.255863302527758e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10960 + }, + { + "epoch": 3.1917369799243525, + "flow_matching_loss": 0.01956510078161955, + "grad_norm": 0.09008416533470154, + "learning_rate": 2.2544346477547685e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10970 + }, + { + "epoch": 3.194646494035496, + "flow_matching_loss": 0.016137733217328788, + "grad_norm": 0.07498758286237717, + "learning_rate": 2.253005076272544e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10980 + }, + { + "epoch": 3.1975560081466394, + "flow_matching_loss": 0.019458981230854988, + "grad_norm": 0.11913537979125977, + "learning_rate": 2.2515745898181477e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10990 + }, + { + "epoch": 3.200465522257783, + "flow_matching_loss": 0.016269123833626508, + "grad_norm": 0.1177680641412735, + "learning_rate": 2.250143190129756e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11000 + }, + { + "epoch": 3.2033750363689264, + "flow_matching_loss": 0.01806772630661726, + "grad_norm": 0.07835470885038376, + "learning_rate": 2.2487108789466547e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11010 + }, + { + "epoch": 3.2062845504800697, + "flow_matching_loss": 0.01762646622955799, + "grad_norm": 0.14429393410682678, + "learning_rate": 2.247277658009236e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11020 + }, + { + "epoch": 3.2091940645912134, + "flow_matching_loss": 0.017010372877120972, + "grad_norm": 0.11572926491498947, + "learning_rate": 2.245843529059e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11030 + }, + { + "epoch": 3.2121035787023566, + "flow_matching_loss": 0.016172098740935326, + "grad_norm": 0.090363509953022, + "learning_rate": 2.244408493838547e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11040 + }, + { + "epoch": 3.2150130928135003, + "flow_matching_loss": 0.018178763892501594, + "grad_norm": 0.10014654695987701, + "learning_rate": 2.242972554091581e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11050 + }, + { + "epoch": 3.2179226069246436, + "flow_matching_loss": 0.017245830222964286, + "grad_norm": 0.0881604254245758, + "learning_rate": 2.2415357115629045e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11060 + }, + { + "epoch": 3.220832121035787, + "flow_matching_loss": 0.015890311822295188, + "grad_norm": 0.10409723222255707, + "learning_rate": 2.2400979679984162e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11070 + }, + { + "epoch": 3.2237416351469306, + "flow_matching_loss": 0.01699820552021265, + "grad_norm": 0.12658771872520447, + "learning_rate": 2.2386593251451096e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11080 + }, + { + "epoch": 3.226651149258074, + "flow_matching_loss": 0.016235327813774347, + "grad_norm": 0.1314588040113449, + "learning_rate": 2.2372197847510722e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11090 + }, + { + "epoch": 3.2295606633692175, + "flow_matching_loss": 0.017558663804084062, + "grad_norm": 0.0919298455119133, + "learning_rate": 2.2357793485654813e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11100 + }, + { + "epoch": 3.232470177480361, + "flow_matching_loss": 0.016546535398811103, + "grad_norm": 0.1198749765753746, + "learning_rate": 2.2343380183386017e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11110 + }, + { + "epoch": 3.235379691591504, + "flow_matching_loss": 0.017924441117793323, + "grad_norm": 0.12414274364709854, + "learning_rate": 2.232895795821787e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11120 + }, + { + "epoch": 3.2382892057026478, + "flow_matching_loss": 0.01876690620556474, + "grad_norm": 0.1049628034234047, + "learning_rate": 2.2314526827674723e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11130 + }, + { + "epoch": 3.241198719813791, + "flow_matching_loss": 0.016548378113657235, + "grad_norm": 0.08105495572090149, + "learning_rate": 2.230008680929177e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11140 + }, + { + "epoch": 3.2441082339249343, + "flow_matching_loss": 0.014510602690279484, + "grad_norm": 0.08756295591592789, + "learning_rate": 2.2285637920614987e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11150 + }, + { + "epoch": 3.247017748036078, + "flow_matching_loss": 0.01853625774383545, + "grad_norm": 0.11693926155567169, + "learning_rate": 2.2271180179201143e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11160 + }, + { + "epoch": 3.2499272621472213, + "flow_matching_loss": 0.01787342820316553, + "grad_norm": 0.11376772820949554, + "learning_rate": 2.225671360261775e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11170 + }, + { + "epoch": 3.252836776258365, + "flow_matching_loss": 0.015038911160081625, + "grad_norm": 0.0990724116563797, + "learning_rate": 2.2242238208443068e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11180 + }, + { + "epoch": 3.2557462903695082, + "flow_matching_loss": 0.019704859983175994, + "grad_norm": 0.0719190314412117, + "learning_rate": 2.2227754014266063e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11190 + }, + { + "epoch": 3.258655804480652, + "flow_matching_loss": 0.017659429740160704, + "grad_norm": 0.12255430966615677, + "learning_rate": 2.22132610376864e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11200 + }, + { + "epoch": 3.261565318591795, + "flow_matching_loss": 0.018480885215103625, + "grad_norm": 0.10277733206748962, + "learning_rate": 2.219875929631441e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11210 + }, + { + "epoch": 3.2644748327029385, + "flow_matching_loss": 0.015381670836359263, + "grad_norm": 0.07806986570358276, + "learning_rate": 2.2184248807771074e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11220 + }, + { + "epoch": 3.267384346814082, + "flow_matching_loss": 0.021141335275024174, + "grad_norm": 0.1219012662768364, + "learning_rate": 2.216972958968801e-05, + "loss": 0.0211, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11230 + }, + { + "epoch": 3.2702938609252254, + "flow_matching_loss": 0.016800382919609547, + "grad_norm": 0.1181412860751152, + "learning_rate": 2.215520165970743e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11240 + }, + { + "epoch": 3.2732033750363687, + "flow_matching_loss": 0.017858489695936442, + "grad_norm": 0.11031405627727509, + "learning_rate": 2.2140665035482143e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11250 + }, + { + "epoch": 3.2761128891475124, + "flow_matching_loss": 0.01683491449803114, + "grad_norm": 0.10728054493665695, + "learning_rate": 2.2126119734675518e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11260 + }, + { + "epoch": 3.2790224032586557, + "flow_matching_loss": 0.01657544309273362, + "grad_norm": 0.10907420516014099, + "learning_rate": 2.2111565774961463e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11270 + }, + { + "epoch": 3.2819319173697994, + "flow_matching_loss": 0.014962619543075562, + "grad_norm": 0.14752265810966492, + "learning_rate": 2.2097003174024417e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11280 + }, + { + "epoch": 3.2848414314809427, + "flow_matching_loss": 0.01515916856005788, + "grad_norm": 0.10144834220409393, + "learning_rate": 2.2082431949559315e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11290 + }, + { + "epoch": 3.2877509455920864, + "flow_matching_loss": 0.016396564245224, + "grad_norm": 0.09552954137325287, + "learning_rate": 2.2067852119271558e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11300 + }, + { + "epoch": 3.2906604597032296, + "flow_matching_loss": 0.01629395941272378, + "grad_norm": 0.08745764195919037, + "learning_rate": 2.2053263700877026e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11310 + }, + { + "epoch": 3.293569973814373, + "flow_matching_loss": 0.01624776190146804, + "grad_norm": 0.09962816536426544, + "learning_rate": 2.2038666712102015e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11320 + }, + { + "epoch": 3.2964794879255166, + "flow_matching_loss": 0.016029024496674538, + "grad_norm": 0.10830114781856537, + "learning_rate": 2.2024061170683246e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11330 + }, + { + "epoch": 3.29938900203666, + "flow_matching_loss": 0.018747856095433234, + "grad_norm": 0.111957848072052, + "learning_rate": 2.200944709436783e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11340 + }, + { + "epoch": 3.302298516147803, + "flow_matching_loss": 0.017974776960909367, + "grad_norm": 0.09599035233259201, + "learning_rate": 2.1994824500913246e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11350 + }, + { + "epoch": 3.305208030258947, + "flow_matching_loss": 0.01856102151796222, + "grad_norm": 0.10287068039178848, + "learning_rate": 2.1980193408087322e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11360 + }, + { + "epoch": 3.30811754437009, + "flow_matching_loss": 0.016954929940402506, + "grad_norm": 0.07657893002033234, + "learning_rate": 2.1965553833668224e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11370 + }, + { + "epoch": 3.311027058481234, + "flow_matching_loss": 0.017982215993106364, + "grad_norm": 0.11776580661535263, + "learning_rate": 2.195090579544441e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11380 + }, + { + "epoch": 3.313936572592377, + "flow_matching_loss": 0.01847560964524746, + "grad_norm": 0.1188497543334961, + "learning_rate": 2.193624931121462e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11390 + }, + { + "epoch": 3.3168460867035203, + "flow_matching_loss": 0.01797729916870594, + "grad_norm": 0.12526482343673706, + "learning_rate": 2.1921584398787872e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11400 + }, + { + "epoch": 3.319755600814664, + "flow_matching_loss": 0.01792053123936057, + "grad_norm": 0.08912550657987595, + "learning_rate": 2.1906911075983418e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11410 + }, + { + "epoch": 3.3226651149258073, + "flow_matching_loss": 0.015748875681310892, + "grad_norm": 0.12167323380708694, + "learning_rate": 2.1892229360630728e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11420 + }, + { + "epoch": 3.325574629036951, + "flow_matching_loss": 0.015499299019575119, + "grad_norm": 0.10049828886985779, + "learning_rate": 2.187753927056947e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11430 + }, + { + "epoch": 3.3284841431480943, + "flow_matching_loss": 0.015336861182004214, + "grad_norm": 0.09212835133075714, + "learning_rate": 2.186284082364949e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11440 + }, + { + "epoch": 3.3313936572592375, + "flow_matching_loss": 0.016964916419237853, + "grad_norm": 0.08828458189964294, + "learning_rate": 2.1848134037730786e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11450 + }, + { + "epoch": 3.3343031713703812, + "flow_matching_loss": 0.016696213465183973, + "grad_norm": 0.07656654715538025, + "learning_rate": 2.183341893068349e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11460 + }, + { + "epoch": 3.3372126854815245, + "flow_matching_loss": 0.015292736329138279, + "grad_norm": 0.09438396990299225, + "learning_rate": 2.1818695520387842e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11470 + }, + { + "epoch": 3.340122199592668, + "flow_matching_loss": 0.01570647247135639, + "grad_norm": 0.08726302534341812, + "learning_rate": 2.1803963824734173e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11480 + }, + { + "epoch": 3.3430317137038115, + "flow_matching_loss": 0.01728242002427578, + "grad_norm": 0.11854422837495804, + "learning_rate": 2.1789223861622887e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11490 + }, + { + "epoch": 3.3459412278149547, + "flow_matching_loss": 0.016598236747086048, + "grad_norm": 0.12029577046632767, + "learning_rate": 2.1774475648964428e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11500 + }, + { + "epoch": 3.3488507419260984, + "flow_matching_loss": 0.017056545894593, + "grad_norm": 0.12479795515537262, + "learning_rate": 2.1759719204679267e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11510 + }, + { + "epoch": 3.3517602560372417, + "flow_matching_loss": 0.01869436539709568, + "grad_norm": 0.09090341627597809, + "learning_rate": 2.174495454669787e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11520 + }, + { + "epoch": 3.3546697701483854, + "flow_matching_loss": 0.0197309167124331, + "grad_norm": 0.09225929528474808, + "learning_rate": 2.1730181692960692e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11530 + }, + { + "epoch": 3.3575792842595287, + "flow_matching_loss": 0.019821944274008275, + "grad_norm": 0.09403005987405777, + "learning_rate": 2.171540066141814e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11540 + }, + { + "epoch": 3.360488798370672, + "flow_matching_loss": 0.018573809880763292, + "grad_norm": 0.13481643795967102, + "learning_rate": 2.170061147003056e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11550 + }, + { + "epoch": 3.3633983124818156, + "flow_matching_loss": 0.018604302499443292, + "grad_norm": 0.1271612048149109, + "learning_rate": 2.1685814136768225e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11560 + }, + { + "epoch": 3.366307826592959, + "flow_matching_loss": 0.017109748162329196, + "grad_norm": 0.1411367654800415, + "learning_rate": 2.167100867961128e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11570 + }, + { + "epoch": 3.369217340704102, + "flow_matching_loss": 0.018394350353628398, + "grad_norm": 0.09613040834665298, + "learning_rate": 2.1656195116549756e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11580 + }, + { + "epoch": 3.372126854815246, + "flow_matching_loss": 0.01822853982448578, + "grad_norm": 0.09703870117664337, + "learning_rate": 2.1641373465583527e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11590 + }, + { + "epoch": 3.375036368926389, + "flow_matching_loss": 0.0158661350607872, + "grad_norm": 0.11327365040779114, + "learning_rate": 2.1626543744722292e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11600 + }, + { + "epoch": 3.377945883037533, + "flow_matching_loss": 0.019152140337973834, + "grad_norm": 0.11236138641834259, + "learning_rate": 2.161170597198557e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11610 + }, + { + "epoch": 3.380855397148676, + "flow_matching_loss": 0.02053636135533452, + "grad_norm": 0.11315835267305374, + "learning_rate": 2.1596860165402645e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11620 + }, + { + "epoch": 3.38376491125982, + "flow_matching_loss": 0.021152041014283897, + "grad_norm": 0.1852436065673828, + "learning_rate": 2.1582006343012576e-05, + "loss": 0.0212, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11630 + }, + { + "epoch": 3.386674425370963, + "flow_matching_loss": 0.01667035399004817, + "grad_norm": 0.10755116492509842, + "learning_rate": 2.156714452286416e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11640 + }, + { + "epoch": 3.3895839394821063, + "flow_matching_loss": 0.017705397587269543, + "grad_norm": 0.11165253818035126, + "learning_rate": 2.155227472301591e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11650 + }, + { + "epoch": 3.39249345359325, + "flow_matching_loss": 0.015161607321351767, + "grad_norm": 0.09655166417360306, + "learning_rate": 2.1537396961536027e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11660 + }, + { + "epoch": 3.3954029677043933, + "flow_matching_loss": 0.016920079290866853, + "grad_norm": 0.13598240911960602, + "learning_rate": 2.1522511256502403e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11670 + }, + { + "epoch": 3.3983124818155366, + "flow_matching_loss": 0.015691134519875048, + "grad_norm": 0.10236183553934097, + "learning_rate": 2.1507617626002566e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11680 + }, + { + "epoch": 3.4012219959266803, + "flow_matching_loss": 0.016802864242345096, + "grad_norm": 0.1524566262960434, + "learning_rate": 2.1492716088133685e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11690 + }, + { + "epoch": 3.4041315100378235, + "flow_matching_loss": 0.018287409096956253, + "grad_norm": 0.0904138833284378, + "learning_rate": 2.1477806661002536e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11700 + }, + { + "epoch": 3.4070410241489673, + "flow_matching_loss": 0.016678264550864696, + "grad_norm": 0.10474843531847, + "learning_rate": 2.146288936272548e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11710 + }, + { + "epoch": 3.4099505382601105, + "flow_matching_loss": 0.015692057088017464, + "grad_norm": 0.11613913625478745, + "learning_rate": 2.1447964211428433e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11720 + }, + { + "epoch": 3.4128600523712542, + "flow_matching_loss": 0.018244205228984355, + "grad_norm": 0.10587284713983536, + "learning_rate": 2.1433031225246865e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11730 + }, + { + "epoch": 3.4157695664823975, + "flow_matching_loss": 0.017409776616841555, + "grad_norm": 0.11218385398387909, + "learning_rate": 2.1418090422325766e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11740 + }, + { + "epoch": 3.4186790805935408, + "flow_matching_loss": 0.019279838912189005, + "grad_norm": 0.10095519572496414, + "learning_rate": 2.1403141820819623e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11750 + }, + { + "epoch": 3.4215885947046845, + "flow_matching_loss": 0.018170748278498648, + "grad_norm": 0.10987858474254608, + "learning_rate": 2.1388185438892377e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11760 + }, + { + "epoch": 3.4244981088158277, + "flow_matching_loss": 0.01757259462028742, + "grad_norm": 0.11579645425081253, + "learning_rate": 2.1373221294717465e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11770 + }, + { + "epoch": 3.427407622926971, + "flow_matching_loss": 0.01417988771572709, + "grad_norm": 0.10730372369289398, + "learning_rate": 2.1358249406477724e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11780 + }, + { + "epoch": 3.4303171370381147, + "flow_matching_loss": 0.017681704461574556, + "grad_norm": 0.10584768652915955, + "learning_rate": 2.1343269792365408e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11790 + }, + { + "epoch": 3.433226651149258, + "flow_matching_loss": 0.017728570196777583, + "grad_norm": 0.08703809976577759, + "learning_rate": 2.132828247058217e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11800 + }, + { + "epoch": 3.4361361652604017, + "flow_matching_loss": 0.017994286119937898, + "grad_norm": 0.10803775489330292, + "learning_rate": 2.1313287459339004e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11810 + }, + { + "epoch": 3.439045679371545, + "flow_matching_loss": 0.018324954621493816, + "grad_norm": 0.1347346305847168, + "learning_rate": 2.1298284776856276e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11820 + }, + { + "epoch": 3.4419551934826886, + "flow_matching_loss": 0.018565405160188675, + "grad_norm": 0.06970669329166412, + "learning_rate": 2.128327444136365e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11830 + }, + { + "epoch": 3.444864707593832, + "flow_matching_loss": 0.018317176122218372, + "grad_norm": 0.11515316367149353, + "learning_rate": 2.1268256471100105e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11840 + }, + { + "epoch": 3.447774221704975, + "flow_matching_loss": 0.018038914911448954, + "grad_norm": 0.12508738040924072, + "learning_rate": 2.1253230884313893e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11850 + }, + { + "epoch": 3.450683735816119, + "flow_matching_loss": 0.015852873492985965, + "grad_norm": 0.12313637882471085, + "learning_rate": 2.123819769926251e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11860 + }, + { + "epoch": 3.453593249927262, + "flow_matching_loss": 0.01872999928891659, + "grad_norm": 0.10618789494037628, + "learning_rate": 2.1223156934212702e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11870 + }, + { + "epoch": 3.4565027640384054, + "flow_matching_loss": 0.017988124489784242, + "grad_norm": 0.12369155883789062, + "learning_rate": 2.1208108607440418e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11880 + }, + { + "epoch": 3.459412278149549, + "flow_matching_loss": 0.01659356150776148, + "grad_norm": 0.09880479425191879, + "learning_rate": 2.119305273723078e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11890 + }, + { + "epoch": 3.4623217922606924, + "flow_matching_loss": 0.01717121135443449, + "grad_norm": 0.08065689355134964, + "learning_rate": 2.1177989341878106e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11900 + }, + { + "epoch": 3.465231306371836, + "flow_matching_loss": 0.018069065548479556, + "grad_norm": 0.12991857528686523, + "learning_rate": 2.116291843968583e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11910 + }, + { + "epoch": 3.4681408204829793, + "flow_matching_loss": 0.017142867390066385, + "grad_norm": 0.12521864473819733, + "learning_rate": 2.114784004896653e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11920 + }, + { + "epoch": 3.4710503345941226, + "flow_matching_loss": 0.02138552479445934, + "grad_norm": 0.12859982252120972, + "learning_rate": 2.113275418804187e-05, + "loss": 0.0214, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11930 + }, + { + "epoch": 3.4739598487052663, + "flow_matching_loss": 0.0222785959020257, + "grad_norm": 0.16727638244628906, + "learning_rate": 2.1117660875242583e-05, + "loss": 0.0223, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11940 + }, + { + "epoch": 3.4768693628164096, + "flow_matching_loss": 0.01618236294016242, + "grad_norm": 0.0837680995464325, + "learning_rate": 2.1102560128908478e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11950 + }, + { + "epoch": 3.4797788769275533, + "flow_matching_loss": 0.020458326302468778, + "grad_norm": 0.15740150213241577, + "learning_rate": 2.1087451967388384e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11960 + }, + { + "epoch": 3.4826883910386965, + "flow_matching_loss": 0.01741088181734085, + "grad_norm": 0.11970353126525879, + "learning_rate": 2.1072336409040135e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11970 + }, + { + "epoch": 3.48559790514984, + "flow_matching_loss": 0.01673882994800806, + "grad_norm": 0.1275339126586914, + "learning_rate": 2.1057213472230573e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11980 + }, + { + "epoch": 3.4885074192609835, + "flow_matching_loss": 0.015940543077886105, + "grad_norm": 0.08817119151353836, + "learning_rate": 2.1042083175335478e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11990 + }, + { + "epoch": 3.4914169333721268, + "flow_matching_loss": 0.018669878784567117, + "grad_norm": 0.12141905725002289, + "learning_rate": 2.1026945536739595e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12000 + }, + { + "epoch": 3.4943264474832705, + "flow_matching_loss": 0.019294316321611403, + "grad_norm": 0.12255833297967911, + "learning_rate": 2.1011800574836586e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12010 + }, + { + "epoch": 3.4972359615944137, + "flow_matching_loss": 0.015684737451374532, + "grad_norm": 0.10143955796957016, + "learning_rate": 2.0996648308028992e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12020 + }, + { + "epoch": 3.500145475705557, + "flow_matching_loss": 0.01617130832746625, + "grad_norm": 0.10058043152093887, + "learning_rate": 2.0981488754728266e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12030 + }, + { + "epoch": 3.5030549898167007, + "flow_matching_loss": 0.016253302060067653, + "grad_norm": 0.08706089109182358, + "learning_rate": 2.0966321933354685e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12040 + }, + { + "epoch": 3.505964503927844, + "flow_matching_loss": 0.01740820500999689, + "grad_norm": 0.09055259823799133, + "learning_rate": 2.095114786233736e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12050 + }, + { + "epoch": 3.5088740180389877, + "flow_matching_loss": 0.018042267579585314, + "grad_norm": 0.1165284514427185, + "learning_rate": 2.0935966560114236e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12060 + }, + { + "epoch": 3.511783532150131, + "flow_matching_loss": 0.01578328385949135, + "grad_norm": 0.08231764286756516, + "learning_rate": 2.092077804513201e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12070 + }, + { + "epoch": 3.514693046261274, + "flow_matching_loss": 0.01676584631204605, + "grad_norm": 0.07438531517982483, + "learning_rate": 2.0905582335846167e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12080 + }, + { + "epoch": 3.517602560372418, + "flow_matching_loss": 0.016876632813364267, + "grad_norm": 0.09866189956665039, + "learning_rate": 2.0890379450720927e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12090 + }, + { + "epoch": 3.520512074483561, + "flow_matching_loss": 0.018746224325150253, + "grad_norm": 0.12102433294057846, + "learning_rate": 2.087516940822923e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12100 + }, + { + "epoch": 3.5234215885947044, + "flow_matching_loss": 0.016773126274347305, + "grad_norm": 0.12379666417837143, + "learning_rate": 2.085995222685271e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12110 + }, + { + "epoch": 3.526331102705848, + "flow_matching_loss": 0.016812533792108297, + "grad_norm": 0.12517516314983368, + "learning_rate": 2.084472792508167e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12120 + }, + { + "epoch": 3.5292406168169914, + "flow_matching_loss": 0.017638929188251495, + "grad_norm": 0.15685361623764038, + "learning_rate": 2.082949652141508e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12130 + }, + { + "epoch": 3.532150130928135, + "flow_matching_loss": 0.01628329660743475, + "grad_norm": 0.12842494249343872, + "learning_rate": 2.081425803436054e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12140 + }, + { + "epoch": 3.5350596450392784, + "flow_matching_loss": 0.01590052107349038, + "grad_norm": 0.1048445776104927, + "learning_rate": 2.079901248243423e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12150 + }, + { + "epoch": 3.537969159150422, + "flow_matching_loss": 0.017485375329852105, + "grad_norm": 0.07970736920833588, + "learning_rate": 2.078375988416095e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12160 + }, + { + "epoch": 3.5408786732615654, + "flow_matching_loss": 0.016384985484182834, + "grad_norm": 0.08735865354537964, + "learning_rate": 2.076850025807404e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12170 + }, + { + "epoch": 3.5437881873727086, + "flow_matching_loss": 0.018678644485771657, + "grad_norm": 0.12117670476436615, + "learning_rate": 2.0753233622715374e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12180 + }, + { + "epoch": 3.5466977014838523, + "flow_matching_loss": 0.015944008715450764, + "grad_norm": 0.1258944571018219, + "learning_rate": 2.073795999663536e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12190 + }, + { + "epoch": 3.5496072155949956, + "flow_matching_loss": 0.018125721160322426, + "grad_norm": 0.10161641240119934, + "learning_rate": 2.07226793983929e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12200 + }, + { + "epoch": 3.552516729706139, + "flow_matching_loss": 0.02057289909571409, + "grad_norm": 0.1699497401714325, + "learning_rate": 2.070739184655536e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12210 + }, + { + "epoch": 3.5554262438172826, + "flow_matching_loss": 0.015062516368925572, + "grad_norm": 0.09202244877815247, + "learning_rate": 2.069209735969855e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12220 + }, + { + "epoch": 3.558335757928426, + "flow_matching_loss": 0.01681271577253938, + "grad_norm": 0.07867659628391266, + "learning_rate": 2.067679595640672e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12230 + }, + { + "epoch": 3.5612452720395695, + "flow_matching_loss": 0.017111766338348388, + "grad_norm": 0.10032163560390472, + "learning_rate": 2.066148765527252e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12240 + }, + { + "epoch": 3.564154786150713, + "flow_matching_loss": 0.014879765454679728, + "grad_norm": 0.09489116072654724, + "learning_rate": 2.0646172474896976e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12250 + }, + { + "epoch": 3.5670643002618565, + "flow_matching_loss": 0.01701114745810628, + "grad_norm": 0.12384049594402313, + "learning_rate": 2.063085043388947e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12260 + }, + { + "epoch": 3.5699738143729998, + "flow_matching_loss": 0.019035689905285834, + "grad_norm": 0.07971587032079697, + "learning_rate": 2.0615521550867732e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12270 + }, + { + "epoch": 3.572883328484143, + "flow_matching_loss": 0.014741199649870395, + "grad_norm": 0.09967884421348572, + "learning_rate": 2.0600185844457803e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12280 + }, + { + "epoch": 3.5757928425952867, + "flow_matching_loss": 0.014029958657920361, + "grad_norm": 0.09597716480493546, + "learning_rate": 2.058484333329401e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12290 + }, + { + "epoch": 3.57870235670643, + "flow_matching_loss": 0.019820270501077174, + "grad_norm": 0.1120486706495285, + "learning_rate": 2.0569494036018955e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12300 + }, + { + "epoch": 3.5816118708175733, + "flow_matching_loss": 0.018747544195502997, + "grad_norm": 0.11949411779642105, + "learning_rate": 2.055413797128347e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12310 + }, + { + "epoch": 3.584521384928717, + "flow_matching_loss": 0.01659454395994544, + "grad_norm": 0.11973525583744049, + "learning_rate": 2.0538775157746634e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12320 + }, + { + "epoch": 3.5874308990398602, + "flow_matching_loss": 0.015145210456103087, + "grad_norm": 0.08775419741868973, + "learning_rate": 2.0523405614075706e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12330 + }, + { + "epoch": 3.590340413151004, + "flow_matching_loss": 0.017117340955883263, + "grad_norm": 0.08138681948184967, + "learning_rate": 2.0508029358946132e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12340 + }, + { + "epoch": 3.593249927262147, + "flow_matching_loss": 0.01764583317562938, + "grad_norm": 0.0809263065457344, + "learning_rate": 2.0492646411041515e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12350 + }, + { + "epoch": 3.596159441373291, + "flow_matching_loss": 0.01654229750856757, + "grad_norm": 0.07279617339372635, + "learning_rate": 2.0477256789053587e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12360 + }, + { + "epoch": 3.599068955484434, + "flow_matching_loss": 0.015723094902932645, + "grad_norm": 0.08566262573003769, + "learning_rate": 2.0461860511682194e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12370 + }, + { + "epoch": 3.6019784695955774, + "flow_matching_loss": 0.01521032229065895, + "grad_norm": 0.09379515051841736, + "learning_rate": 2.0446457597635254e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12380 + }, + { + "epoch": 3.604887983706721, + "flow_matching_loss": 0.016054441779851915, + "grad_norm": 0.08445299416780472, + "learning_rate": 2.0431048065628763e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12390 + }, + { + "epoch": 3.6077974978178644, + "flow_matching_loss": 0.0166753813624382, + "grad_norm": 0.10817159712314606, + "learning_rate": 2.0415631934386764e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12400 + }, + { + "epoch": 3.6107070119290077, + "flow_matching_loss": 0.014427537750452756, + "grad_norm": 0.09671463072299957, + "learning_rate": 2.04002092226413e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12410 + }, + { + "epoch": 3.6136165260401514, + "flow_matching_loss": 0.016047271061688663, + "grad_norm": 0.13417033851146698, + "learning_rate": 2.038477994913242e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12420 + }, + { + "epoch": 3.6165260401512946, + "flow_matching_loss": 0.01891305362805724, + "grad_norm": 0.07872102409601212, + "learning_rate": 2.036934413260816e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12430 + }, + { + "epoch": 3.619435554262438, + "flow_matching_loss": 0.01413246439769864, + "grad_norm": 0.09566221386194229, + "learning_rate": 2.0353901791824473e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12440 + }, + { + "epoch": 3.6223450683735816, + "flow_matching_loss": 0.01661413749679923, + "grad_norm": 0.10580331832170486, + "learning_rate": 2.0338452945545275e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12450 + }, + { + "epoch": 3.6252545824847253, + "flow_matching_loss": 0.01428342228755355, + "grad_norm": 0.0752774178981781, + "learning_rate": 2.0322997612542366e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12460 + }, + { + "epoch": 3.6281640965958686, + "flow_matching_loss": 0.01816737586632371, + "grad_norm": 0.12033716589212418, + "learning_rate": 2.0307535811595427e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12470 + }, + { + "epoch": 3.631073610707012, + "flow_matching_loss": 0.015212892834097148, + "grad_norm": 0.11295986920595169, + "learning_rate": 2.0292067561492013e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12480 + }, + { + "epoch": 3.6339831248181556, + "flow_matching_loss": 0.017590536922216415, + "grad_norm": 0.07237235456705093, + "learning_rate": 2.02765928810275e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12490 + }, + { + "epoch": 3.636892638929299, + "flow_matching_loss": 0.0152421735227108, + "grad_norm": 0.15359525382518768, + "learning_rate": 2.0261111789005087e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12500 + }, + { + "epoch": 3.639802153040442, + "flow_matching_loss": 0.01863300073891878, + "grad_norm": 0.1249886080622673, + "learning_rate": 2.024562430423576e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12510 + }, + { + "epoch": 3.642711667151586, + "flow_matching_loss": 0.014845115598291159, + "grad_norm": 0.10090803354978561, + "learning_rate": 2.0230130445538273e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12520 + }, + { + "epoch": 3.645621181262729, + "flow_matching_loss": 0.01823920113965869, + "grad_norm": 0.11375551670789719, + "learning_rate": 2.0214630231739127e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12530 + }, + { + "epoch": 3.6485306953738723, + "flow_matching_loss": 0.015360770933330058, + "grad_norm": 0.10929226130247116, + "learning_rate": 2.0199123681672543e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12540 + }, + { + "epoch": 3.651440209485016, + "flow_matching_loss": 0.016584882978349924, + "grad_norm": 0.11754951626062393, + "learning_rate": 2.0183610814180437e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12550 + }, + { + "epoch": 3.6543497235961593, + "flow_matching_loss": 0.018782926257699728, + "grad_norm": 0.10129260271787643, + "learning_rate": 2.016809164811241e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12560 + }, + { + "epoch": 3.657259237707303, + "flow_matching_loss": 0.016712196171283722, + "grad_norm": 0.1570688784122467, + "learning_rate": 2.015256620232571e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12570 + }, + { + "epoch": 3.6601687518184463, + "flow_matching_loss": 0.018084443360567092, + "grad_norm": 0.09489896893501282, + "learning_rate": 2.0137034495685222e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12580 + }, + { + "epoch": 3.66307826592959, + "flow_matching_loss": 0.016229477850720285, + "grad_norm": 0.06646712124347687, + "learning_rate": 2.012149654706343e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12590 + }, + { + "epoch": 3.6659877800407332, + "flow_matching_loss": 0.015570734720677137, + "grad_norm": 0.09550966322422028, + "learning_rate": 2.010595237534041e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12600 + }, + { + "epoch": 3.6688972941518765, + "flow_matching_loss": 0.014061974547803401, + "grad_norm": 0.09348545223474503, + "learning_rate": 2.0090401999403793e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12610 + }, + { + "epoch": 3.67180680826302, + "flow_matching_loss": 0.016194103751331568, + "grad_norm": 0.13463793694972992, + "learning_rate": 2.007484543814875e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12620 + }, + { + "epoch": 3.6747163223741635, + "flow_matching_loss": 0.016982166934758423, + "grad_norm": 0.13265246152877808, + "learning_rate": 2.0059282710477972e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12630 + }, + { + "epoch": 3.6776258364853067, + "flow_matching_loss": 0.018310912884771824, + "grad_norm": 0.12172620743513107, + "learning_rate": 2.0043713835301647e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12640 + }, + { + "epoch": 3.6805353505964504, + "flow_matching_loss": 0.015956800244748593, + "grad_norm": 0.10520347207784653, + "learning_rate": 2.0028138831537417e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12650 + }, + { + "epoch": 3.6834448647075937, + "flow_matching_loss": 0.016105532180517912, + "grad_norm": 0.07500375062227249, + "learning_rate": 2.0012557718110387e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12660 + }, + { + "epoch": 3.6863543788187374, + "flow_matching_loss": 0.01723590064793825, + "grad_norm": 0.09930115193128586, + "learning_rate": 1.9996970513953077e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12670 + }, + { + "epoch": 3.6892638929298807, + "flow_matching_loss": 0.016428377851843833, + "grad_norm": 0.10746174305677414, + "learning_rate": 1.9981377238005405e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12680 + }, + { + "epoch": 3.6921734070410244, + "flow_matching_loss": 0.01768616451881826, + "grad_norm": 0.134182408452034, + "learning_rate": 1.9965777909214674e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12690 + }, + { + "epoch": 3.6950829211521676, + "flow_matching_loss": 0.015212783310562372, + "grad_norm": 0.08404561877250671, + "learning_rate": 1.9950172546535545e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12700 + }, + { + "epoch": 3.697992435263311, + "flow_matching_loss": 0.016414856258779763, + "grad_norm": 0.09199991822242737, + "learning_rate": 1.9934561168930002e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12710 + }, + { + "epoch": 3.7009019493744546, + "flow_matching_loss": 0.018307274021208287, + "grad_norm": 0.07482583820819855, + "learning_rate": 1.991894379536734e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12720 + }, + { + "epoch": 3.703811463485598, + "flow_matching_loss": 0.018137746304273606, + "grad_norm": 0.0816798061132431, + "learning_rate": 1.9903320444824137e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12730 + }, + { + "epoch": 3.706720977596741, + "flow_matching_loss": 0.014850107952952385, + "grad_norm": 0.08787763863801956, + "learning_rate": 1.9887691136284245e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12740 + }, + { + "epoch": 3.709630491707885, + "flow_matching_loss": 0.01622796105220914, + "grad_norm": 0.08518777042627335, + "learning_rate": 1.9872055888738745e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12750 + }, + { + "epoch": 3.712540005819028, + "flow_matching_loss": 0.018287668004631996, + "grad_norm": 0.09095745533704758, + "learning_rate": 1.985641472118593e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12760 + }, + { + "epoch": 3.715449519930172, + "flow_matching_loss": 0.019061079900711774, + "grad_norm": 0.1228397861123085, + "learning_rate": 1.9840767652631294e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12770 + }, + { + "epoch": 3.718359034041315, + "flow_matching_loss": 0.016998581402003766, + "grad_norm": 0.09928615391254425, + "learning_rate": 1.9825114702087513e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12780 + }, + { + "epoch": 3.7212685481524588, + "flow_matching_loss": 0.017400463484227656, + "grad_norm": 0.09567832946777344, + "learning_rate": 1.9809455888574388e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12790 + }, + { + "epoch": 3.724178062263602, + "flow_matching_loss": 0.017205290123820306, + "grad_norm": 0.10800719261169434, + "learning_rate": 1.9793791231118864e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12800 + }, + { + "epoch": 3.7270875763747453, + "flow_matching_loss": 0.015379883348941803, + "grad_norm": 0.09229813516139984, + "learning_rate": 1.9778120748754958e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12810 + }, + { + "epoch": 3.729997090485889, + "flow_matching_loss": 0.016021823696792125, + "grad_norm": 0.06971389055252075, + "learning_rate": 1.97624444605238e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12820 + }, + { + "epoch": 3.7329066045970323, + "flow_matching_loss": 0.01656747506931424, + "grad_norm": 0.08995677530765533, + "learning_rate": 1.9746762385473552e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12830 + }, + { + "epoch": 3.7358161187081755, + "flow_matching_loss": 0.01552039198577404, + "grad_norm": 0.0883893072605133, + "learning_rate": 1.9731074542659414e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12840 + }, + { + "epoch": 3.7387256328193192, + "flow_matching_loss": 0.018693336565047503, + "grad_norm": 0.12100815773010254, + "learning_rate": 1.9715380951143603e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12850 + }, + { + "epoch": 3.7416351469304625, + "flow_matching_loss": 0.016572430543601514, + "grad_norm": 0.13556957244873047, + "learning_rate": 1.9699681629995295e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12860 + }, + { + "epoch": 3.744544661041606, + "flow_matching_loss": 0.01742819668725133, + "grad_norm": 0.10914882272481918, + "learning_rate": 1.968397659829066e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12870 + }, + { + "epoch": 3.7474541751527495, + "flow_matching_loss": 0.015750265307724477, + "grad_norm": 0.08067692071199417, + "learning_rate": 1.966826587511279e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12880 + }, + { + "epoch": 3.750363689263893, + "flow_matching_loss": 0.017666925396770238, + "grad_norm": 0.1602693349123001, + "learning_rate": 1.965254947955169e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12890 + }, + { + "epoch": 3.7532732033750364, + "flow_matching_loss": 0.01509360009804368, + "grad_norm": 0.08864778280258179, + "learning_rate": 1.963682743070427e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12900 + }, + { + "epoch": 3.7561827174861797, + "flow_matching_loss": 0.015593391004949808, + "grad_norm": 0.09607668220996857, + "learning_rate": 1.962109974767429e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12910 + }, + { + "epoch": 3.7590922315973234, + "flow_matching_loss": 0.01651570899412036, + "grad_norm": 0.10573388636112213, + "learning_rate": 1.960536644957238e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12920 + }, + { + "epoch": 3.7620017457084667, + "flow_matching_loss": 0.018648765329271556, + "grad_norm": 0.11771620064973831, + "learning_rate": 1.9589627555515974e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12930 + }, + { + "epoch": 3.76491125981961, + "flow_matching_loss": 0.016282585076987745, + "grad_norm": 0.10193521529436111, + "learning_rate": 1.9573883084629317e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12940 + }, + { + "epoch": 3.7678207739307537, + "flow_matching_loss": 0.01810954287648201, + "grad_norm": 0.13011033833026886, + "learning_rate": 1.9558133056043422e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12950 + }, + { + "epoch": 3.770730288041897, + "flow_matching_loss": 0.016909771133214235, + "grad_norm": 0.09285027533769608, + "learning_rate": 1.9542377488896062e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12960 + }, + { + "epoch": 3.77363980215304, + "flow_matching_loss": 0.015523257572203875, + "grad_norm": 0.09501209855079651, + "learning_rate": 1.9526616402331733e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12970 + }, + { + "epoch": 3.776549316264184, + "flow_matching_loss": 0.015719993319362403, + "grad_norm": 0.11441008001565933, + "learning_rate": 1.9510849815501638e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12980 + }, + { + "epoch": 3.7794588303753276, + "flow_matching_loss": 0.016090051271021366, + "grad_norm": 0.1377420872449875, + "learning_rate": 1.949507774756368e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12990 + }, + { + "epoch": 3.782368344486471, + "flow_matching_loss": 0.018523840140551327, + "grad_norm": 0.11636623740196228, + "learning_rate": 1.9479300217682394e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13000 + }, + { + "epoch": 3.785277858597614, + "flow_matching_loss": 0.017384209390729665, + "grad_norm": 0.09801468253135681, + "learning_rate": 1.946351724502898e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13010 + }, + { + "epoch": 3.788187372708758, + "flow_matching_loss": 0.017453391570597887, + "grad_norm": 0.09506893903017044, + "learning_rate": 1.9447728848781227e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13020 + }, + { + "epoch": 3.791096886819901, + "flow_matching_loss": 0.016339645069092513, + "grad_norm": 0.16902174055576324, + "learning_rate": 1.943193504812353e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13030 + }, + { + "epoch": 3.7940064009310444, + "flow_matching_loss": 0.015980910044163465, + "grad_norm": 0.08795704692602158, + "learning_rate": 1.9416135862246846e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13040 + }, + { + "epoch": 3.796915915042188, + "flow_matching_loss": 0.018725888058543205, + "grad_norm": 0.15500669181346893, + "learning_rate": 1.9400331310348673e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13050 + }, + { + "epoch": 3.7998254291533313, + "flow_matching_loss": 0.01676732785999775, + "grad_norm": 0.12860733270645142, + "learning_rate": 1.9384521411633033e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13060 + }, + { + "epoch": 3.8027349432644746, + "flow_matching_loss": 0.014008757378906011, + "grad_norm": 0.08703599125146866, + "learning_rate": 1.936870618531045e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13070 + }, + { + "epoch": 3.8056444573756183, + "flow_matching_loss": 0.01812824448570609, + "grad_norm": 0.10077287256717682, + "learning_rate": 1.9352885650597904e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13080 + }, + { + "epoch": 3.8085539714867616, + "flow_matching_loss": 0.017955259047448636, + "grad_norm": 0.1057751476764679, + "learning_rate": 1.9337059826718847e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13090 + }, + { + "epoch": 3.8114634855979053, + "flow_matching_loss": 0.01590264206752181, + "grad_norm": 0.13822762668132782, + "learning_rate": 1.932122873290314e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13100 + }, + { + "epoch": 3.8143729997090485, + "flow_matching_loss": 0.01823499519377947, + "grad_norm": 0.10861235111951828, + "learning_rate": 1.9305392388387054e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13110 + }, + { + "epoch": 3.8172825138201922, + "flow_matching_loss": 0.01595664266496897, + "grad_norm": 0.09811568260192871, + "learning_rate": 1.928955081241325e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13120 + }, + { + "epoch": 3.8201920279313355, + "flow_matching_loss": 0.017158600874245165, + "grad_norm": 0.08317399024963379, + "learning_rate": 1.9273704024230725e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13130 + }, + { + "epoch": 3.8231015420424788, + "flow_matching_loss": 0.015312578808516265, + "grad_norm": 0.09728337824344635, + "learning_rate": 1.925785204309483e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13140 + }, + { + "epoch": 3.8260110561536225, + "flow_matching_loss": 0.0174286269582808, + "grad_norm": 0.10158412158489227, + "learning_rate": 1.9241994888267212e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13150 + }, + { + "epoch": 3.8289205702647657, + "flow_matching_loss": 0.015408447664231061, + "grad_norm": 0.0612046979367733, + "learning_rate": 1.9226132579015815e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13160 + }, + { + "epoch": 3.831830084375909, + "flow_matching_loss": 0.016741381026804447, + "grad_norm": 0.08908824622631073, + "learning_rate": 1.9210265134614832e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13170 + }, + { + "epoch": 3.8347395984870527, + "flow_matching_loss": 0.015872610453516244, + "grad_norm": 0.07075361162424088, + "learning_rate": 1.9194392574344712e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13180 + }, + { + "epoch": 3.837649112598196, + "flow_matching_loss": 0.01904375208541751, + "grad_norm": 0.08760780096054077, + "learning_rate": 1.9178514917492103e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13190 + }, + { + "epoch": 3.8405586267093397, + "flow_matching_loss": 0.014829815458506345, + "grad_norm": 0.0913081169128418, + "learning_rate": 1.9162632183349866e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13200 + }, + { + "epoch": 3.843468140820483, + "flow_matching_loss": 0.01641531931236386, + "grad_norm": 0.08150303363800049, + "learning_rate": 1.914674439121701e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13210 + }, + { + "epoch": 3.8463776549316266, + "flow_matching_loss": 0.016703038569539787, + "grad_norm": 0.08840328454971313, + "learning_rate": 1.9130851560398715e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13220 + }, + { + "epoch": 3.84928716904277, + "flow_matching_loss": 0.016910781618207693, + "grad_norm": 0.08222091197967529, + "learning_rate": 1.911495371020625e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13230 + }, + { + "epoch": 3.852196683153913, + "flow_matching_loss": 0.01453474760055542, + "grad_norm": 0.11402494460344315, + "learning_rate": 1.909905085995702e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13240 + }, + { + "epoch": 3.855106197265057, + "flow_matching_loss": 0.01657362263649702, + "grad_norm": 0.0779091939330101, + "learning_rate": 1.908314302897448e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13250 + }, + { + "epoch": 3.8580157113762, + "flow_matching_loss": 0.017768303211778403, + "grad_norm": 0.093662790954113, + "learning_rate": 1.9067230236588142e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13260 + }, + { + "epoch": 3.8609252254873434, + "flow_matching_loss": 0.01624351511709392, + "grad_norm": 0.08935526013374329, + "learning_rate": 1.905131250213355e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13270 + }, + { + "epoch": 3.863834739598487, + "flow_matching_loss": 0.01657905811443925, + "grad_norm": 0.09218782186508179, + "learning_rate": 1.9035389844952256e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13280 + }, + { + "epoch": 3.8667442537096304, + "flow_matching_loss": 0.0173229293897748, + "grad_norm": 0.10502084344625473, + "learning_rate": 1.901946228439179e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13290 + }, + { + "epoch": 3.869653767820774, + "flow_matching_loss": 0.0151486043818295, + "grad_norm": 0.1248641088604927, + "learning_rate": 1.9003529839805637e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13300 + }, + { + "epoch": 3.8725632819319173, + "flow_matching_loss": 0.01655611805617809, + "grad_norm": 0.10235421359539032, + "learning_rate": 1.8987592530553217e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13310 + }, + { + "epoch": 3.875472796043061, + "flow_matching_loss": 0.020584689825773238, + "grad_norm": 0.1556289792060852, + "learning_rate": 1.897165037599987e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13320 + }, + { + "epoch": 3.8783823101542043, + "flow_matching_loss": 0.01806502928957343, + "grad_norm": 0.07555297017097473, + "learning_rate": 1.8955703395516814e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13330 + }, + { + "epoch": 3.8812918242653476, + "flow_matching_loss": 0.01561225214973092, + "grad_norm": 0.11786383390426636, + "learning_rate": 1.8939751608481134e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13340 + }, + { + "epoch": 3.8842013383764913, + "flow_matching_loss": 0.01606062548235059, + "grad_norm": 0.13601112365722656, + "learning_rate": 1.892379503427575e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13350 + }, + { + "epoch": 3.8871108524876345, + "flow_matching_loss": 0.019636988081037997, + "grad_norm": 0.10336559265851974, + "learning_rate": 1.8907833692289412e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13360 + }, + { + "epoch": 3.890020366598778, + "flow_matching_loss": 0.013888938911259174, + "grad_norm": 0.08544784039258957, + "learning_rate": 1.889186760191666e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13370 + }, + { + "epoch": 3.8929298807099215, + "flow_matching_loss": 0.01624129302799702, + "grad_norm": 0.09615042060613632, + "learning_rate": 1.887589678255779e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13380 + }, + { + "epoch": 3.895839394821065, + "flow_matching_loss": 0.016002936661243437, + "grad_norm": 0.1391596496105194, + "learning_rate": 1.8859921253618852e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13390 + }, + { + "epoch": 3.898748908932208, + "flow_matching_loss": 0.017012601252645254, + "grad_norm": 0.11406459659337997, + "learning_rate": 1.884394103451163e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13400 + }, + { + "epoch": 3.9016584230433518, + "flow_matching_loss": 0.01714625870808959, + "grad_norm": 0.110779769718647, + "learning_rate": 1.8827956144653583e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13410 + }, + { + "epoch": 3.9045679371544955, + "flow_matching_loss": 0.019186852872371672, + "grad_norm": 0.0898725688457489, + "learning_rate": 1.8811966603467874e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13420 + }, + { + "epoch": 3.9074774512656387, + "flow_matching_loss": 0.01856397623196244, + "grad_norm": 0.17817018926143646, + "learning_rate": 1.8795972430383293e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13430 + }, + { + "epoch": 3.910386965376782, + "flow_matching_loss": 0.017138399370014667, + "grad_norm": 0.08084523677825928, + "learning_rate": 1.8779973644834273e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13440 + }, + { + "epoch": 3.9132964794879257, + "flow_matching_loss": 0.014715242758393287, + "grad_norm": 0.10876669734716415, + "learning_rate": 1.876397026626085e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13450 + }, + { + "epoch": 3.916205993599069, + "flow_matching_loss": 0.01642067525535822, + "grad_norm": 0.0690922811627388, + "learning_rate": 1.8747962314108633e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13460 + }, + { + "epoch": 3.919115507710212, + "flow_matching_loss": 0.017024180572479963, + "grad_norm": 0.1068263128399849, + "learning_rate": 1.873194980782879e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13470 + }, + { + "epoch": 3.922025021821356, + "flow_matching_loss": 0.017240972630679606, + "grad_norm": 0.07683034241199493, + "learning_rate": 1.871593276687804e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13480 + }, + { + "epoch": 3.924934535932499, + "flow_matching_loss": 0.015127575397491455, + "grad_norm": 0.07133612036705017, + "learning_rate": 1.869991121071858e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13490 + }, + { + "epoch": 3.9278440500436425, + "flow_matching_loss": 0.017091592773795128, + "grad_norm": 0.09297561645507812, + "learning_rate": 1.8683885158818117e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13500 + }, + { + "epoch": 3.930753564154786, + "flow_matching_loss": 0.0189496791921556, + "grad_norm": 0.08095396310091019, + "learning_rate": 1.8667854630649825e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13510 + }, + { + "epoch": 3.9336630782659294, + "flow_matching_loss": 0.014315936155617237, + "grad_norm": 0.13264933228492737, + "learning_rate": 1.86518196456923e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13520 + }, + { + "epoch": 3.936572592377073, + "flow_matching_loss": 0.015383055713027716, + "grad_norm": 0.0802878737449646, + "learning_rate": 1.8635780223429565e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13530 + }, + { + "epoch": 3.9394821064882164, + "flow_matching_loss": 0.01657079830765724, + "grad_norm": 0.08909593522548676, + "learning_rate": 1.8619736383351017e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13540 + }, + { + "epoch": 3.94239162059936, + "flow_matching_loss": 0.01753278626129031, + "grad_norm": 0.09342725574970245, + "learning_rate": 1.8603688144951454e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13550 + }, + { + "epoch": 3.9453011347105034, + "flow_matching_loss": 0.01704461667686701, + "grad_norm": 0.09295453131198883, + "learning_rate": 1.8587635527730983e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13560 + }, + { + "epoch": 3.9482106488216466, + "flow_matching_loss": 0.01829156596213579, + "grad_norm": 0.08780011534690857, + "learning_rate": 1.8571578551195056e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13570 + }, + { + "epoch": 3.9511201629327903, + "flow_matching_loss": 0.017546477634459733, + "grad_norm": 0.09026364237070084, + "learning_rate": 1.855551723485441e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13580 + }, + { + "epoch": 3.9540296770439336, + "flow_matching_loss": 0.019383129384368657, + "grad_norm": 0.13824330270290375, + "learning_rate": 1.8539451598225066e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13590 + }, + { + "epoch": 3.956939191155077, + "flow_matching_loss": 0.0164991470053792, + "grad_norm": 0.10461151599884033, + "learning_rate": 1.8523381660828276e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13600 + }, + { + "epoch": 3.9598487052662206, + "flow_matching_loss": 0.014032720401883126, + "grad_norm": 0.1088300496339798, + "learning_rate": 1.850730744219054e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13610 + }, + { + "epoch": 3.962758219377364, + "flow_matching_loss": 0.015306173823773861, + "grad_norm": 0.08262323588132858, + "learning_rate": 1.849122896184354e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13620 + }, + { + "epoch": 3.9656677334885075, + "flow_matching_loss": 0.014927633758634328, + "grad_norm": 0.10251221060752869, + "learning_rate": 1.8475146239324146e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13630 + }, + { + "epoch": 3.968577247599651, + "flow_matching_loss": 0.015456491895020009, + "grad_norm": 0.09771323204040527, + "learning_rate": 1.8459059294174396e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13640 + }, + { + "epoch": 3.9714867617107945, + "flow_matching_loss": 0.017321035824716093, + "grad_norm": 0.14955934882164001, + "learning_rate": 1.844296814594143e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13650 + }, + { + "epoch": 3.9743962758219378, + "flow_matching_loss": 0.014413911569863557, + "grad_norm": 0.08577822893857956, + "learning_rate": 1.8426872814177516e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13660 + }, + { + "epoch": 3.977305789933081, + "flow_matching_loss": 0.016526913084089757, + "grad_norm": 0.1011245995759964, + "learning_rate": 1.8410773318440003e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13670 + }, + { + "epoch": 3.9802153040442247, + "flow_matching_loss": 0.01951951263472438, + "grad_norm": 0.14346736669540405, + "learning_rate": 1.839466967829129e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13680 + }, + { + "epoch": 3.983124818155368, + "flow_matching_loss": 0.017316418047994376, + "grad_norm": 0.09068089723587036, + "learning_rate": 1.8378561913298822e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13690 + }, + { + "epoch": 3.9860343322665113, + "flow_matching_loss": 0.01913882251828909, + "grad_norm": 0.08934729546308517, + "learning_rate": 1.8362450043035053e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13700 + }, + { + "epoch": 3.988943846377655, + "flow_matching_loss": 0.017345335427671672, + "grad_norm": 0.11823779344558716, + "learning_rate": 1.834633408707742e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13710 + }, + { + "epoch": 3.9918533604887982, + "flow_matching_loss": 0.014759994577616453, + "grad_norm": 0.09085006266832352, + "learning_rate": 1.8330214065008336e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13720 + }, + { + "epoch": 3.994762874599942, + "flow_matching_loss": 0.01924119433388114, + "grad_norm": 0.09332837909460068, + "learning_rate": 1.831408999641514e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13730 + }, + { + "epoch": 3.997672388711085, + "flow_matching_loss": 0.015985803306102754, + "grad_norm": 0.09846356511116028, + "learning_rate": 1.8297961900890107e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13740 + }, + { + "epoch": 4.000581902822229, + "flow_matching_loss": 0.019217167887836693, + "grad_norm": 0.1231658011674881, + "learning_rate": 1.8281829798030385e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13750 + }, + { + "epoch": 4.003491416933372, + "flow_matching_loss": 0.018439241871237754, + "grad_norm": 0.13085968792438507, + "learning_rate": 1.8265693707438002e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13760 + }, + { + "epoch": 4.0064009310445154, + "flow_matching_loss": 0.013935388764366508, + "grad_norm": 0.1002269983291626, + "learning_rate": 1.824955364871982e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13770 + }, + { + "epoch": 4.009310445155659, + "flow_matching_loss": 0.015538275986909867, + "grad_norm": 0.11298969388008118, + "learning_rate": 1.8233409641487553e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13780 + }, + { + "epoch": 4.012219959266803, + "flow_matching_loss": 0.016157521028071643, + "grad_norm": 0.10717200487852097, + "learning_rate": 1.8217261705357677e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13790 + }, + { + "epoch": 4.015129473377946, + "flow_matching_loss": 0.015492352191358805, + "grad_norm": 0.0930967628955841, + "learning_rate": 1.8201109859951462e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13800 + }, + { + "epoch": 4.018038987489089, + "flow_matching_loss": 0.015741464961320163, + "grad_norm": 0.12380781769752502, + "learning_rate": 1.818495412489492e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13810 + }, + { + "epoch": 4.020948501600233, + "flow_matching_loss": 0.017037717159837484, + "grad_norm": 0.11595480889081955, + "learning_rate": 1.8168794519818802e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13820 + }, + { + "epoch": 4.023858015711376, + "flow_matching_loss": 0.018775954749435185, + "grad_norm": 0.12244854867458344, + "learning_rate": 1.8152631064358537e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13830 + }, + { + "epoch": 4.02676752982252, + "flow_matching_loss": 0.01862484896555543, + "grad_norm": 0.10676400363445282, + "learning_rate": 1.8136463778154258e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13840 + }, + { + "epoch": 4.029677043933663, + "flow_matching_loss": 0.015818431600928308, + "grad_norm": 0.10603279620409012, + "learning_rate": 1.812029268085074e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13850 + }, + { + "epoch": 4.032586558044806, + "flow_matching_loss": 0.014308629184961319, + "grad_norm": 0.08487431704998016, + "learning_rate": 1.8104117792097387e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13860 + }, + { + "epoch": 4.03549607215595, + "flow_matching_loss": 0.01913582533597946, + "grad_norm": 0.09508536010980606, + "learning_rate": 1.808793913154822e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13870 + }, + { + "epoch": 4.038405586267094, + "flow_matching_loss": 0.018032890651375054, + "grad_norm": 0.11633582413196564, + "learning_rate": 1.8071756718861838e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13880 + }, + { + "epoch": 4.041315100378237, + "flow_matching_loss": 0.017773021711036563, + "grad_norm": 0.0897398442029953, + "learning_rate": 1.805557057370139e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13890 + }, + { + "epoch": 4.04422461448938, + "flow_matching_loss": 0.017445030994713305, + "grad_norm": 0.0869385153055191, + "learning_rate": 1.8039380715734574e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13900 + }, + { + "epoch": 4.047134128600524, + "flow_matching_loss": 0.017188047152012588, + "grad_norm": 0.10933095961809158, + "learning_rate": 1.802318716463359e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13910 + }, + { + "epoch": 4.0500436427116675, + "flow_matching_loss": 0.016068013664335013, + "grad_norm": 0.10384562611579895, + "learning_rate": 1.8006989940075138e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13920 + }, + { + "epoch": 4.05295315682281, + "flow_matching_loss": 0.014259934891015291, + "grad_norm": 0.12237749248743057, + "learning_rate": 1.7990789061740368e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13930 + }, + { + "epoch": 4.055862670933954, + "flow_matching_loss": 0.014692302141338588, + "grad_norm": 0.10231764614582062, + "learning_rate": 1.7974584549314875e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13940 + }, + { + "epoch": 4.058772185045098, + "flow_matching_loss": 0.017078329157084227, + "grad_norm": 0.09250355511903763, + "learning_rate": 1.7958376422488672e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13950 + }, + { + "epoch": 4.0616816991562406, + "flow_matching_loss": 0.016436827182769776, + "grad_norm": 0.09822327643632889, + "learning_rate": 1.7942164700956156e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13960 + }, + { + "epoch": 4.064591213267384, + "flow_matching_loss": 0.018575597647577526, + "grad_norm": 0.10899528861045837, + "learning_rate": 1.79259494044161e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13970 + }, + { + "epoch": 4.067500727378528, + "flow_matching_loss": 0.015052048582583666, + "grad_norm": 0.07966610044240952, + "learning_rate": 1.7909730552571617e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13980 + }, + { + "epoch": 4.070410241489671, + "flow_matching_loss": 0.015752366837114097, + "grad_norm": 0.1086639016866684, + "learning_rate": 1.7893508165130143e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13990 + }, + { + "epoch": 4.0733197556008145, + "flow_matching_loss": 0.01411684425547719, + "grad_norm": 0.08196080476045609, + "learning_rate": 1.787728226180341e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14000 + }, + { + "epoch": 4.076229269711958, + "flow_matching_loss": 0.016998186986893416, + "grad_norm": 0.14692853391170502, + "learning_rate": 1.786105286230742e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14010 + }, + { + "epoch": 4.079138783823102, + "flow_matching_loss": 0.01643937798216939, + "grad_norm": 0.07970968633890152, + "learning_rate": 1.784481998636242e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14020 + }, + { + "epoch": 4.082048297934245, + "flow_matching_loss": 0.014707861840724945, + "grad_norm": 0.090290367603302, + "learning_rate": 1.782858365369289e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14030 + }, + { + "epoch": 4.084957812045388, + "flow_matching_loss": 0.014193193707615136, + "grad_norm": 0.10465805977582932, + "learning_rate": 1.78123438840275e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14040 + }, + { + "epoch": 4.087867326156532, + "flow_matching_loss": 0.019041665829718113, + "grad_norm": 0.12046783417463303, + "learning_rate": 1.779610069709911e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14050 + }, + { + "epoch": 4.090776840267675, + "flow_matching_loss": 0.016730032209306956, + "grad_norm": 0.08127912878990173, + "learning_rate": 1.7779854112644713e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14060 + }, + { + "epoch": 4.093686354378819, + "flow_matching_loss": 0.01306318761780858, + "grad_norm": 0.07202616333961487, + "learning_rate": 1.7763604150405446e-05, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14070 + }, + { + "epoch": 4.096595868489962, + "flow_matching_loss": 0.015167648065835237, + "grad_norm": 0.08973774313926697, + "learning_rate": 1.7747350830126548e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14080 + }, + { + "epoch": 4.099505382601105, + "flow_matching_loss": 0.015405882429331542, + "grad_norm": 0.08842409402132034, + "learning_rate": 1.7731094171557332e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14090 + }, + { + "epoch": 4.102414896712249, + "flow_matching_loss": 0.015053561143577098, + "grad_norm": 0.08970632404088974, + "learning_rate": 1.771483419445117e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14100 + }, + { + "epoch": 4.105324410823393, + "flow_matching_loss": 0.015787344425916672, + "grad_norm": 0.13172154128551483, + "learning_rate": 1.7698570918565474e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14110 + }, + { + "epoch": 4.108233924934536, + "flow_matching_loss": 0.0175326113589108, + "grad_norm": 0.08798792213201523, + "learning_rate": 1.768230436366165e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14120 + }, + { + "epoch": 4.111143439045679, + "flow_matching_loss": 0.016544082667678595, + "grad_norm": 0.08165581524372101, + "learning_rate": 1.7666034549505102e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14130 + }, + { + "epoch": 4.114052953156823, + "flow_matching_loss": 0.014953741524368524, + "grad_norm": 0.09946196526288986, + "learning_rate": 1.7649761495865185e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14140 + }, + { + "epoch": 4.1169624672679666, + "flow_matching_loss": 0.014449213538318873, + "grad_norm": 0.10985726863145828, + "learning_rate": 1.7633485222515188e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14150 + }, + { + "epoch": 4.119871981379109, + "flow_matching_loss": 0.0158359351567924, + "grad_norm": 0.10079121589660645, + "learning_rate": 1.761720574923233e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14160 + }, + { + "epoch": 4.122781495490253, + "flow_matching_loss": 0.01686147563159466, + "grad_norm": 0.08332313597202301, + "learning_rate": 1.7600923095797704e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14170 + }, + { + "epoch": 4.125691009601397, + "flow_matching_loss": 0.016212908551096916, + "grad_norm": 0.11558276414871216, + "learning_rate": 1.758463728199626e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14180 + }, + { + "epoch": 4.12860052371254, + "flow_matching_loss": 0.017184327729046343, + "grad_norm": 0.11055291444063187, + "learning_rate": 1.7568348327616803e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14190 + }, + { + "epoch": 4.131510037823683, + "flow_matching_loss": 0.0159779473207891, + "grad_norm": 0.1032731831073761, + "learning_rate": 1.755205625245195e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14200 + }, + { + "epoch": 4.134419551934827, + "flow_matching_loss": 0.017496406007558106, + "grad_norm": 0.1304956078529358, + "learning_rate": 1.7535761076298104e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14210 + }, + { + "epoch": 4.137329066045971, + "flow_matching_loss": 0.017571727372705938, + "grad_norm": 0.07045839726924896, + "learning_rate": 1.7519462818955448e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14220 + }, + { + "epoch": 4.1402385801571135, + "flow_matching_loss": 0.017406838294118644, + "grad_norm": 0.19756193459033966, + "learning_rate": 1.7503161500227896e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14230 + }, + { + "epoch": 4.143148094268257, + "flow_matching_loss": 0.01562606794759631, + "grad_norm": 0.10599926114082336, + "learning_rate": 1.7486857139923093e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14240 + }, + { + "epoch": 4.146057608379401, + "flow_matching_loss": 0.017404246609658003, + "grad_norm": 0.1170332208275795, + "learning_rate": 1.747054975785236e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14250 + }, + { + "epoch": 4.148967122490544, + "flow_matching_loss": 0.01572293508797884, + "grad_norm": 0.09225881099700928, + "learning_rate": 1.745423937383072e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14260 + }, + { + "epoch": 4.1518766366016875, + "flow_matching_loss": 0.017494051437824964, + "grad_norm": 0.1169407069683075, + "learning_rate": 1.7437926007676825e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14270 + }, + { + "epoch": 4.154786150712831, + "flow_matching_loss": 0.015322569105774165, + "grad_norm": 0.08328565955162048, + "learning_rate": 1.742160967921294e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14280 + }, + { + "epoch": 4.157695664823974, + "flow_matching_loss": 0.015415052277967333, + "grad_norm": 0.12098155170679092, + "learning_rate": 1.7405290408264954e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14290 + }, + { + "epoch": 4.160605178935118, + "flow_matching_loss": 0.016877013631165026, + "grad_norm": 0.09920068830251694, + "learning_rate": 1.738896821466232e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14300 + }, + { + "epoch": 4.163514693046261, + "flow_matching_loss": 0.01928565325215459, + "grad_norm": 0.10219766944646835, + "learning_rate": 1.737264311823804e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14310 + }, + { + "epoch": 4.166424207157405, + "flow_matching_loss": 0.01646290151402354, + "grad_norm": 0.12037648260593414, + "learning_rate": 1.7356315138828645e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14320 + }, + { + "epoch": 4.169333721268548, + "flow_matching_loss": 0.014531752560287714, + "grad_norm": 0.10641574114561081, + "learning_rate": 1.7339984296274174e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14330 + }, + { + "epoch": 4.172243235379692, + "flow_matching_loss": 0.01852541109547019, + "grad_norm": 0.09486116468906403, + "learning_rate": 1.7323650610418134e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14340 + }, + { + "epoch": 4.175152749490835, + "flow_matching_loss": 0.014818787109106779, + "grad_norm": 0.0813441053032875, + "learning_rate": 1.730731410110749e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14350 + }, + { + "epoch": 4.178062263601978, + "flow_matching_loss": 0.01567042078822851, + "grad_norm": 0.08676321804523468, + "learning_rate": 1.7290974788192655e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14360 + }, + { + "epoch": 4.180971777713122, + "flow_matching_loss": 0.015122364088892937, + "grad_norm": 0.07350590825080872, + "learning_rate": 1.727463269152743e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14370 + }, + { + "epoch": 4.183881291824266, + "flow_matching_loss": 0.016098495572805405, + "grad_norm": 0.11782407760620117, + "learning_rate": 1.7258287830968996e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14380 + }, + { + "epoch": 4.186790805935408, + "flow_matching_loss": 0.018132514972239732, + "grad_norm": 0.12841656804084778, + "learning_rate": 1.7241940226377908e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14390 + }, + { + "epoch": 4.189700320046552, + "flow_matching_loss": 0.01573459869250655, + "grad_norm": 0.10482030361890793, + "learning_rate": 1.7225589897618043e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14400 + }, + { + "epoch": 4.192609834157696, + "flow_matching_loss": 0.014083197340369225, + "grad_norm": 0.08213619887828827, + "learning_rate": 1.7209236864556588e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14410 + }, + { + "epoch": 4.195519348268839, + "flow_matching_loss": 0.013584069535136222, + "grad_norm": 0.0584077350795269, + "learning_rate": 1.719288114706403e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14420 + }, + { + "epoch": 4.198428862379982, + "flow_matching_loss": 0.016869401186704637, + "grad_norm": 0.09442660212516785, + "learning_rate": 1.7176522765014104e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14430 + }, + { + "epoch": 4.201338376491126, + "flow_matching_loss": 0.017672218661755324, + "grad_norm": 0.09274273365736008, + "learning_rate": 1.7160161738283782e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14440 + }, + { + "epoch": 4.20424789060227, + "flow_matching_loss": 0.014600632712244987, + "grad_norm": 0.07790569216012955, + "learning_rate": 1.714379808675326e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14450 + }, + { + "epoch": 4.207157404713413, + "flow_matching_loss": 0.014218398090451955, + "grad_norm": 0.09977280348539352, + "learning_rate": 1.712743183030592e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14460 + }, + { + "epoch": 4.210066918824556, + "flow_matching_loss": 0.015235679876059293, + "grad_norm": 0.09977210313081741, + "learning_rate": 1.7111062988828294e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14470 + }, + { + "epoch": 4.2129764329357, + "flow_matching_loss": 0.013707359042018651, + "grad_norm": 0.1697809249162674, + "learning_rate": 1.709469158221008e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14480 + }, + { + "epoch": 4.215885947046843, + "flow_matching_loss": 0.01685498310253024, + "grad_norm": 0.11442065984010696, + "learning_rate": 1.7078317630344083e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14490 + }, + { + "epoch": 4.2187954611579865, + "flow_matching_loss": 0.015165155474096536, + "grad_norm": 0.07933083176612854, + "learning_rate": 1.706194115312618e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14500 + }, + { + "epoch": 4.22170497526913, + "flow_matching_loss": 0.01747297877445817, + "grad_norm": 0.09733022749423981, + "learning_rate": 1.704556217045536e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14510 + }, + { + "epoch": 4.224614489380274, + "flow_matching_loss": 0.015648717805743218, + "grad_norm": 0.11268153041601181, + "learning_rate": 1.7029180702233612e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14520 + }, + { + "epoch": 4.227524003491417, + "flow_matching_loss": 0.01616940489038825, + "grad_norm": 0.08891375362873077, + "learning_rate": 1.7012796768365975e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14530 + }, + { + "epoch": 4.2304335176025605, + "flow_matching_loss": 0.014578462857753039, + "grad_norm": 0.11706782877445221, + "learning_rate": 1.6996410388760465e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14540 + }, + { + "epoch": 4.233343031713704, + "flow_matching_loss": 0.018178456649184226, + "grad_norm": 0.10057071596384048, + "learning_rate": 1.698002158332809e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14550 + }, + { + "epoch": 4.236252545824847, + "flow_matching_loss": 0.016716297063976527, + "grad_norm": 0.09005194902420044, + "learning_rate": 1.6963630371982783e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14560 + }, + { + "epoch": 4.239162059935991, + "flow_matching_loss": 0.015047491155564785, + "grad_norm": 0.0877184197306633, + "learning_rate": 1.6947236774641416e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14570 + }, + { + "epoch": 4.242071574047134, + "flow_matching_loss": 0.01444267649203539, + "grad_norm": 0.09514611214399338, + "learning_rate": 1.6930840811223755e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14580 + }, + { + "epoch": 4.244981088158277, + "flow_matching_loss": 0.01539029013365507, + "grad_norm": 0.1086859405040741, + "learning_rate": 1.6914442501652448e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14590 + }, + { + "epoch": 4.247890602269421, + "flow_matching_loss": 0.014503851672634483, + "grad_norm": 0.06945018470287323, + "learning_rate": 1.6898041865852974e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14600 + }, + { + "epoch": 4.250800116380565, + "flow_matching_loss": 0.01662211799994111, + "grad_norm": 0.09680130332708359, + "learning_rate": 1.6881638923753663e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14610 + }, + { + "epoch": 4.2537096304917075, + "flow_matching_loss": 0.0150479088537395, + "grad_norm": 0.0797659382224083, + "learning_rate": 1.6865233695285632e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14620 + }, + { + "epoch": 4.256619144602851, + "flow_matching_loss": 0.015315295569598675, + "grad_norm": 0.09130281955003738, + "learning_rate": 1.6848826200382782e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14630 + }, + { + "epoch": 4.259528658713995, + "flow_matching_loss": 0.015640498045831917, + "grad_norm": 0.06954661756753922, + "learning_rate": 1.6832416458981762e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14640 + }, + { + "epoch": 4.262438172825139, + "flow_matching_loss": 0.014442320633679628, + "grad_norm": 0.07033764570951462, + "learning_rate": 1.681600449102196e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14650 + }, + { + "epoch": 4.265347686936281, + "flow_matching_loss": 0.016288541723042727, + "grad_norm": 0.0768030509352684, + "learning_rate": 1.6799590316445468e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14660 + }, + { + "epoch": 4.268257201047425, + "flow_matching_loss": 0.018902813363820315, + "grad_norm": 0.09257551282644272, + "learning_rate": 1.678317395519705e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14670 + }, + { + "epoch": 4.271166715158569, + "flow_matching_loss": 0.0162323041819036, + "grad_norm": 0.09417937695980072, + "learning_rate": 1.676675542722413e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14680 + }, + { + "epoch": 4.274076229269712, + "flow_matching_loss": 0.014544641319662332, + "grad_norm": 0.09866730868816376, + "learning_rate": 1.6750334752476776e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14690 + }, + { + "epoch": 4.276985743380855, + "flow_matching_loss": 0.01785870613530278, + "grad_norm": 0.11645770072937012, + "learning_rate": 1.673391195090765e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14700 + }, + { + "epoch": 4.279895257491999, + "flow_matching_loss": 0.015270494762808085, + "grad_norm": 0.08106483519077301, + "learning_rate": 1.6717487042472006e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14710 + }, + { + "epoch": 4.282804771603142, + "flow_matching_loss": 0.0160136335529387, + "grad_norm": 0.085543192923069, + "learning_rate": 1.6701060047127655e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14720 + }, + { + "epoch": 4.285714285714286, + "flow_matching_loss": 0.017300919536501168, + "grad_norm": 0.12364611029624939, + "learning_rate": 1.6684630984834947e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14730 + }, + { + "epoch": 4.288623799825429, + "flow_matching_loss": 0.015039634983986616, + "grad_norm": 0.06580080091953278, + "learning_rate": 1.666819987555675e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14740 + }, + { + "epoch": 4.291533313936573, + "flow_matching_loss": 0.013938609138131141, + "grad_norm": 0.09449799358844757, + "learning_rate": 1.6651766739258392e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14750 + }, + { + "epoch": 4.294442828047716, + "flow_matching_loss": 0.017270491272211076, + "grad_norm": 0.11116847395896912, + "learning_rate": 1.6635331595907704e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14760 + }, + { + "epoch": 4.2973523421588595, + "flow_matching_loss": 0.01560730254277587, + "grad_norm": 0.0735214501619339, + "learning_rate": 1.6618894465474923e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14770 + }, + { + "epoch": 4.300261856270003, + "flow_matching_loss": 0.01706037884578109, + "grad_norm": 0.08395195752382278, + "learning_rate": 1.660245536793271e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14780 + }, + { + "epoch": 4.303171370381146, + "flow_matching_loss": 0.014473474584519863, + "grad_norm": 0.1251356601715088, + "learning_rate": 1.6586014323256126e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14790 + }, + { + "epoch": 4.30608088449229, + "flow_matching_loss": 0.01667649280279875, + "grad_norm": 0.07805212587118149, + "learning_rate": 1.656957135142259e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14800 + }, + { + "epoch": 4.3089903986034335, + "flow_matching_loss": 0.016665185429155828, + "grad_norm": 0.0882173702120781, + "learning_rate": 1.655312647241186e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14810 + }, + { + "epoch": 4.311899912714576, + "flow_matching_loss": 0.013680392317473889, + "grad_norm": 0.09620925039052963, + "learning_rate": 1.6536679706206015e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14820 + }, + { + "epoch": 4.31480942682572, + "flow_matching_loss": 0.018011457938700914, + "grad_norm": 0.07885660231113434, + "learning_rate": 1.6520231072789427e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14830 + }, + { + "epoch": 4.317718940936864, + "flow_matching_loss": 0.0171472136862576, + "grad_norm": 0.10724294930696487, + "learning_rate": 1.6503780592148737e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14840 + }, + { + "epoch": 4.320628455048007, + "flow_matching_loss": 0.013781437370926141, + "grad_norm": 0.07718419283628464, + "learning_rate": 1.648732828427283e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14850 + }, + { + "epoch": 4.32353796915915, + "flow_matching_loss": 0.016224464122205972, + "grad_norm": 0.09346777945756912, + "learning_rate": 1.647087416915281e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14860 + }, + { + "epoch": 4.326447483270294, + "flow_matching_loss": 0.016495107766240837, + "grad_norm": 0.09114187210798264, + "learning_rate": 1.6454418266781987e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14870 + }, + { + "epoch": 4.329356997381438, + "flow_matching_loss": 0.017638651840388774, + "grad_norm": 0.12305302917957306, + "learning_rate": 1.6437960597155824e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14880 + }, + { + "epoch": 4.3322665114925805, + "flow_matching_loss": 0.014799921587109566, + "grad_norm": 0.11526305973529816, + "learning_rate": 1.642150118027194e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14890 + }, + { + "epoch": 4.335176025603724, + "flow_matching_loss": 0.018105693720281126, + "grad_norm": 0.12174665182828903, + "learning_rate": 1.640504003613009e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14900 + }, + { + "epoch": 4.338085539714868, + "flow_matching_loss": 0.017812772002071143, + "grad_norm": 0.09828537702560425, + "learning_rate": 1.6388577184732107e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14910 + }, + { + "epoch": 4.340995053826011, + "flow_matching_loss": 0.01652163304388523, + "grad_norm": 0.0879681408405304, + "learning_rate": 1.6372112646081907e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14920 + }, + { + "epoch": 4.343904567937154, + "flow_matching_loss": 0.01848255880177021, + "grad_norm": 0.16025499999523163, + "learning_rate": 1.6355646440185464e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14930 + }, + { + "epoch": 4.346814082048298, + "flow_matching_loss": 0.01673642946407199, + "grad_norm": 0.0814657211303711, + "learning_rate": 1.6339178587050764e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14940 + }, + { + "epoch": 4.349723596159441, + "flow_matching_loss": 0.01528903776779771, + "grad_norm": 0.10057863593101501, + "learning_rate": 1.632270910668781e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14950 + }, + { + "epoch": 4.352633110270585, + "flow_matching_loss": 0.014203892927616835, + "grad_norm": 0.10205753147602081, + "learning_rate": 1.6306238019108562e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14960 + }, + { + "epoch": 4.355542624381728, + "flow_matching_loss": 0.016157666314393283, + "grad_norm": 0.10272819548845291, + "learning_rate": 1.6289765344326952e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14970 + }, + { + "epoch": 4.358452138492872, + "flow_matching_loss": 0.017005117144435644, + "grad_norm": 0.08041521906852722, + "learning_rate": 1.6273291102358833e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14980 + }, + { + "epoch": 4.361361652604015, + "flow_matching_loss": 0.01551317647099495, + "grad_norm": 0.0937652513384819, + "learning_rate": 1.6256815313221963e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14990 + }, + { + "epoch": 4.364271166715159, + "flow_matching_loss": 0.015615325700491667, + "grad_norm": 0.07066230475902557, + "learning_rate": 1.6240337996935972e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15000 + }, + { + "epoch": 4.367180680826302, + "flow_matching_loss": 0.01727598300203681, + "grad_norm": 0.11289616674184799, + "learning_rate": 1.6223859173522367e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15010 + }, + { + "epoch": 4.370090194937445, + "flow_matching_loss": 0.01786585748195648, + "grad_norm": 0.09734094142913818, + "learning_rate": 1.6207378863004455e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15020 + }, + { + "epoch": 4.372999709048589, + "flow_matching_loss": 0.014421155489981175, + "grad_norm": 0.09651602059602737, + "learning_rate": 1.6190897085407382e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15030 + }, + { + "epoch": 4.3759092231597325, + "flow_matching_loss": 0.015193953458219767, + "grad_norm": 0.10185321420431137, + "learning_rate": 1.617441386075805e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15040 + }, + { + "epoch": 4.378818737270876, + "flow_matching_loss": 0.01604761229828, + "grad_norm": 0.11549616605043411, + "learning_rate": 1.6157929209085142e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15050 + }, + { + "epoch": 4.381728251382019, + "flow_matching_loss": 0.01599924871698022, + "grad_norm": 0.08693745732307434, + "learning_rate": 1.614144315041906e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15060 + }, + { + "epoch": 4.384637765493163, + "flow_matching_loss": 0.01714960914105177, + "grad_norm": 0.12913542985916138, + "learning_rate": 1.612495570479192e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15070 + }, + { + "epoch": 4.3875472796043065, + "flow_matching_loss": 0.01953795300796628, + "grad_norm": 0.09705273061990738, + "learning_rate": 1.6108466892237517e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15080 + }, + { + "epoch": 4.390456793715449, + "flow_matching_loss": 0.01845652228221297, + "grad_norm": 0.11559299379587173, + "learning_rate": 1.609197673279132e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15090 + }, + { + "epoch": 4.393366307826593, + "flow_matching_loss": 0.017266194615513087, + "grad_norm": 0.08085296303033829, + "learning_rate": 1.6075485246490428e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15100 + }, + { + "epoch": 4.396275821937737, + "flow_matching_loss": 0.017480923887342216, + "grad_norm": 0.07242101430892944, + "learning_rate": 1.605899245337355e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15110 + }, + { + "epoch": 4.3991853360488795, + "flow_matching_loss": 0.01665394501760602, + "grad_norm": 0.10832604765892029, + "learning_rate": 1.6042498373480983e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15120 + }, + { + "epoch": 4.402094850160023, + "flow_matching_loss": 0.014900316204875707, + "grad_norm": 0.06682789325714111, + "learning_rate": 1.602600302685459e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15130 + }, + { + "epoch": 4.405004364271167, + "flow_matching_loss": 0.01736470116302371, + "grad_norm": 0.10420133918523788, + "learning_rate": 1.6009506433537774e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15140 + }, + { + "epoch": 4.40791387838231, + "flow_matching_loss": 0.013668314088135958, + "grad_norm": 0.06599612534046173, + "learning_rate": 1.599300861357545e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15150 + }, + { + "epoch": 4.4108233924934535, + "flow_matching_loss": 0.014864539168775082, + "grad_norm": 0.11788338422775269, + "learning_rate": 1.5976509587014032e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15160 + }, + { + "epoch": 4.413732906604597, + "flow_matching_loss": 0.016544056497514247, + "grad_norm": 0.0697181448340416, + "learning_rate": 1.5960009373901384e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15170 + }, + { + "epoch": 4.416642420715741, + "flow_matching_loss": 0.01614631898701191, + "grad_norm": 0.11715347319841385, + "learning_rate": 1.5943507994286823e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15180 + }, + { + "epoch": 4.419551934826884, + "flow_matching_loss": 0.018591429572552443, + "grad_norm": 0.10873400419950485, + "learning_rate": 1.5927005468221082e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15190 + }, + { + "epoch": 4.422461448938027, + "flow_matching_loss": 0.01606637900695205, + "grad_norm": 0.07983242720365524, + "learning_rate": 1.5910501815756285e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15200 + }, + { + "epoch": 4.425370963049171, + "flow_matching_loss": 0.015022598206996918, + "grad_norm": 0.08836698532104492, + "learning_rate": 1.589399705694593e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15210 + }, + { + "epoch": 4.428280477160314, + "flow_matching_loss": 0.015822744462639093, + "grad_norm": 0.08869028091430664, + "learning_rate": 1.587749121184485e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15220 + }, + { + "epoch": 4.431189991271458, + "flow_matching_loss": 0.016069013718515636, + "grad_norm": 0.11042685806751251, + "learning_rate": 1.5860984300509206e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15230 + }, + { + "epoch": 4.434099505382601, + "flow_matching_loss": 0.017266084905713797, + "grad_norm": 0.0901445746421814, + "learning_rate": 1.5844476342996453e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15240 + }, + { + "epoch": 4.437009019493744, + "flow_matching_loss": 0.01576445922255516, + "grad_norm": 0.11595486104488373, + "learning_rate": 1.5827967359365306e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15250 + }, + { + "epoch": 4.439918533604888, + "flow_matching_loss": 0.015422432404011487, + "grad_norm": 0.0917731299996376, + "learning_rate": 1.5811457369675758e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15260 + }, + { + "epoch": 4.442828047716032, + "flow_matching_loss": 0.020163973793387413, + "grad_norm": 0.1155005693435669, + "learning_rate": 1.5794946393988982e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15270 + }, + { + "epoch": 4.445737561827175, + "flow_matching_loss": 0.01649732617661357, + "grad_norm": 0.10163987427949905, + "learning_rate": 1.5778434452367382e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15280 + }, + { + "epoch": 4.448647075938318, + "flow_matching_loss": 0.015724303014576436, + "grad_norm": 0.0977906659245491, + "learning_rate": 1.576192156487452e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15290 + }, + { + "epoch": 4.451556590049462, + "flow_matching_loss": 0.017930114269256593, + "grad_norm": 0.14176473021507263, + "learning_rate": 1.574540775157511e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15300 + }, + { + "epoch": 4.4544661041606055, + "flow_matching_loss": 0.014501907862722874, + "grad_norm": 0.11660414934158325, + "learning_rate": 1.5728893032534998e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15310 + }, + { + "epoch": 4.457375618271748, + "flow_matching_loss": 0.013243012689054012, + "grad_norm": 0.07668542116880417, + "learning_rate": 1.571237742782112e-05, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15320 + }, + { + "epoch": 4.460285132382892, + "flow_matching_loss": 0.01598684210330248, + "grad_norm": 0.0718831717967987, + "learning_rate": 1.569586095750149e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15330 + }, + { + "epoch": 4.463194646494036, + "flow_matching_loss": 0.014765835367143154, + "grad_norm": 0.10625175386667252, + "learning_rate": 1.5679343641645193e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15340 + }, + { + "epoch": 4.466104160605179, + "flow_matching_loss": 0.016342782881110907, + "grad_norm": 0.07233001291751862, + "learning_rate": 1.5662825500322304e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15350 + }, + { + "epoch": 4.469013674716322, + "flow_matching_loss": 0.016837434470653535, + "grad_norm": 0.09161531180143356, + "learning_rate": 1.5646306553603932e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15360 + }, + { + "epoch": 4.471923188827466, + "flow_matching_loss": 0.01837548762559891, + "grad_norm": 0.10622671991586685, + "learning_rate": 1.5629786821562154e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15370 + }, + { + "epoch": 4.47483270293861, + "flow_matching_loss": 0.014784779213368893, + "grad_norm": 0.08131495863199234, + "learning_rate": 1.561326632427001e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15380 + }, + { + "epoch": 4.4777422170497525, + "flow_matching_loss": 0.017050490807741882, + "grad_norm": 0.11842180788516998, + "learning_rate": 1.5596745081801447e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15390 + }, + { + "epoch": 4.480651731160896, + "flow_matching_loss": 0.015684643667191266, + "grad_norm": 0.09263888746500015, + "learning_rate": 1.5580223114231348e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15400 + }, + { + "epoch": 4.48356124527204, + "flow_matching_loss": 0.014076963160187006, + "grad_norm": 0.10360424965620041, + "learning_rate": 1.5563700441635445e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15410 + }, + { + "epoch": 4.486470759383183, + "flow_matching_loss": 0.01809463743120432, + "grad_norm": 0.07650944590568542, + "learning_rate": 1.5547177084090358e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15420 + }, + { + "epoch": 4.4893802734943264, + "flow_matching_loss": 0.015365965757519006, + "grad_norm": 0.11915356665849686, + "learning_rate": 1.5530653061673514e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15430 + }, + { + "epoch": 4.49228978760547, + "flow_matching_loss": 0.017584039084613323, + "grad_norm": 0.09902108460664749, + "learning_rate": 1.5514128394463167e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15440 + }, + { + "epoch": 4.495199301716613, + "flow_matching_loss": 0.01571573754772544, + "grad_norm": 0.08676744252443314, + "learning_rate": 1.549760310253834e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15450 + }, + { + "epoch": 4.498108815827757, + "flow_matching_loss": 0.015400867909193039, + "grad_norm": 0.07576404511928558, + "learning_rate": 1.5481077205978816e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15460 + }, + { + "epoch": 4.5010183299389, + "flow_matching_loss": 0.013449622225016356, + "grad_norm": 0.0705718845129013, + "learning_rate": 1.5464550724865135e-05, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15470 + }, + { + "epoch": 4.503927844050043, + "flow_matching_loss": 0.017731491569429637, + "grad_norm": 0.10060626268386841, + "learning_rate": 1.5448023679278517e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15480 + }, + { + "epoch": 4.506837358161187, + "flow_matching_loss": 0.01677383203059435, + "grad_norm": 0.09220478683710098, + "learning_rate": 1.543149608930088e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15490 + }, + { + "epoch": 4.509746872272331, + "flow_matching_loss": 0.01860901080071926, + "grad_norm": 0.08575331419706345, + "learning_rate": 1.5414967975014805e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15500 + }, + { + "epoch": 4.512656386383474, + "flow_matching_loss": 0.015396214369684458, + "grad_norm": 0.08871474117040634, + "learning_rate": 1.539843935650351e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15510 + }, + { + "epoch": 4.515565900494617, + "flow_matching_loss": 0.016409564390778543, + "grad_norm": 0.10582870990037918, + "learning_rate": 1.5381910253850834e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15520 + }, + { + "epoch": 4.518475414605761, + "flow_matching_loss": 0.014621280133724213, + "grad_norm": 0.13695944845676422, + "learning_rate": 1.5365380687141183e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15530 + }, + { + "epoch": 4.521384928716905, + "flow_matching_loss": 0.016633154824376106, + "grad_norm": 0.1086900532245636, + "learning_rate": 1.5348850676459544e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15540 + }, + { + "epoch": 4.524294442828047, + "flow_matching_loss": 0.016752814687788487, + "grad_norm": 0.07390608638525009, + "learning_rate": 1.5332320241891438e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15550 + }, + { + "epoch": 4.527203956939191, + "flow_matching_loss": 0.017726988811045886, + "grad_norm": 0.09585577994585037, + "learning_rate": 1.5315789403522908e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15560 + }, + { + "epoch": 4.530113471050335, + "flow_matching_loss": 0.018756733275949955, + "grad_norm": 0.0875697135925293, + "learning_rate": 1.529925818144047e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15570 + }, + { + "epoch": 4.5330229851614785, + "flow_matching_loss": 0.017603451292961834, + "grad_norm": 0.06943368166685104, + "learning_rate": 1.5282726595731126e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15580 + }, + { + "epoch": 4.535932499272621, + "flow_matching_loss": 0.014357708487659692, + "grad_norm": 0.07813708484172821, + "learning_rate": 1.5266194666482304e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15590 + }, + { + "epoch": 4.538842013383765, + "flow_matching_loss": 0.017316557839512824, + "grad_norm": 0.11420274525880814, + "learning_rate": 1.5249662413781866e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15600 + }, + { + "epoch": 4.541751527494909, + "flow_matching_loss": 0.015961054619401692, + "grad_norm": 0.13688023388385773, + "learning_rate": 1.5233129857718054e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15610 + }, + { + "epoch": 4.5446610416060516, + "flow_matching_loss": 0.01527927340939641, + "grad_norm": 0.10849982500076294, + "learning_rate": 1.521659701837948e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15620 + }, + { + "epoch": 4.547570555717195, + "flow_matching_loss": 0.01488520372658968, + "grad_norm": 0.09187113493680954, + "learning_rate": 1.5200063915855107e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15630 + }, + { + "epoch": 4.550480069828339, + "flow_matching_loss": 0.016544935200363396, + "grad_norm": 0.1198318749666214, + "learning_rate": 1.5183530570234212e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15640 + }, + { + "epoch": 4.553389583939482, + "flow_matching_loss": 0.015307972207665443, + "grad_norm": 0.13333529233932495, + "learning_rate": 1.5166997001606361e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15650 + }, + { + "epoch": 4.5562990980506255, + "flow_matching_loss": 0.016619101259857418, + "grad_norm": 0.137334406375885, + "learning_rate": 1.5150463230061417e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15660 + }, + { + "epoch": 4.559208612161769, + "flow_matching_loss": 0.014383634552359581, + "grad_norm": 0.11802871525287628, + "learning_rate": 1.5133929275689455e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15670 + }, + { + "epoch": 4.562118126272912, + "flow_matching_loss": 0.013825292978435755, + "grad_norm": 0.09984635561704636, + "learning_rate": 1.5117395158580795e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15680 + }, + { + "epoch": 4.565027640384056, + "flow_matching_loss": 0.014200825989246369, + "grad_norm": 0.11393607407808304, + "learning_rate": 1.510086089882595e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15690 + }, + { + "epoch": 4.567937154495199, + "flow_matching_loss": 0.01653757756575942, + "grad_norm": 0.08561106026172638, + "learning_rate": 1.5084326516515599e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15700 + }, + { + "epoch": 4.570846668606343, + "flow_matching_loss": 0.01602890104986727, + "grad_norm": 0.08250420540571213, + "learning_rate": 1.5067792031740582e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15710 + }, + { + "epoch": 4.573756182717486, + "flow_matching_loss": 0.01401349538937211, + "grad_norm": 0.11003727465867996, + "learning_rate": 1.505125746459185e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15720 + }, + { + "epoch": 4.57666569682863, + "flow_matching_loss": 0.015454203262925148, + "grad_norm": 0.08349286764860153, + "learning_rate": 1.5034722835160466e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15730 + }, + { + "epoch": 4.579575210939773, + "flow_matching_loss": 0.015734692104160786, + "grad_norm": 0.12764522433280945, + "learning_rate": 1.5018188163537566e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15740 + }, + { + "epoch": 4.582484725050916, + "flow_matching_loss": 0.017423460353165866, + "grad_norm": 0.08878914266824722, + "learning_rate": 1.5001653469814331e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15750 + }, + { + "epoch": 4.58539423916206, + "flow_matching_loss": 0.016137593239545823, + "grad_norm": 0.1537633240222931, + "learning_rate": 1.4985118774081978e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15760 + }, + { + "epoch": 4.588303753273204, + "flow_matching_loss": 0.017437493056058885, + "grad_norm": 0.07317952811717987, + "learning_rate": 1.4968584096431708e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15770 + }, + { + "epoch": 4.591213267384346, + "flow_matching_loss": 0.016291189147159456, + "grad_norm": 0.09388724714517593, + "learning_rate": 1.4952049456954731e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15780 + }, + { + "epoch": 4.59412278149549, + "flow_matching_loss": 0.01673245495185256, + "grad_norm": 0.0870712473988533, + "learning_rate": 1.4935514875742182e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15790 + }, + { + "epoch": 4.597032295606634, + "flow_matching_loss": 0.016887236945331098, + "grad_norm": 0.12846586108207703, + "learning_rate": 1.4918980372885143e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15800 + }, + { + "epoch": 4.599941809717777, + "flow_matching_loss": 0.016820732969790697, + "grad_norm": 0.11485950648784637, + "learning_rate": 1.4902445968474591e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15810 + }, + { + "epoch": 4.60285132382892, + "flow_matching_loss": 0.015478447452187538, + "grad_norm": 0.13132734596729279, + "learning_rate": 1.4885911682601378e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15820 + }, + { + "epoch": 4.605760837940064, + "flow_matching_loss": 0.01464445684105158, + "grad_norm": 0.164974644780159, + "learning_rate": 1.4869377535356235e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15830 + }, + { + "epoch": 4.608670352051208, + "flow_matching_loss": 0.015067234355956315, + "grad_norm": 0.0945645123720169, + "learning_rate": 1.4852843546829704e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15840 + }, + { + "epoch": 4.611579866162351, + "flow_matching_loss": 0.017439099587500097, + "grad_norm": 0.09182863682508469, + "learning_rate": 1.483630973711214e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15850 + }, + { + "epoch": 4.614489380273494, + "flow_matching_loss": 0.018054576124995947, + "grad_norm": 0.09114368259906769, + "learning_rate": 1.4819776126293677e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15860 + }, + { + "epoch": 4.617398894384638, + "flow_matching_loss": 0.01638011299073696, + "grad_norm": 0.08503157645463943, + "learning_rate": 1.4803242734464214e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15870 + }, + { + "epoch": 4.620308408495781, + "flow_matching_loss": 0.01583732310682535, + "grad_norm": 0.09942633658647537, + "learning_rate": 1.4786709581713383e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15880 + }, + { + "epoch": 4.6232179226069245, + "flow_matching_loss": 0.018788494169712067, + "grad_norm": 0.10680214315652847, + "learning_rate": 1.4770176688130524e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15890 + }, + { + "epoch": 4.626127436718068, + "flow_matching_loss": 0.015565091650933028, + "grad_norm": 0.12351426482200623, + "learning_rate": 1.475364407380466e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15900 + }, + { + "epoch": 4.629036950829212, + "flow_matching_loss": 0.01567549994215369, + "grad_norm": 0.10863669961690903, + "learning_rate": 1.4737111758824476e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15910 + }, + { + "epoch": 4.631946464940355, + "flow_matching_loss": 0.015478389523923396, + "grad_norm": 0.0930393636226654, + "learning_rate": 1.4720579763278289e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15920 + }, + { + "epoch": 4.6348559790514985, + "flow_matching_loss": 0.01629438167437911, + "grad_norm": 0.10764170438051224, + "learning_rate": 1.4704048107254042e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15930 + }, + { + "epoch": 4.637765493162642, + "flow_matching_loss": 0.015313482051715255, + "grad_norm": 0.09405120462179184, + "learning_rate": 1.4687516810839246e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15940 + }, + { + "epoch": 4.640675007273785, + "flow_matching_loss": 0.017218116857111453, + "grad_norm": 0.11763253808021545, + "learning_rate": 1.4670985894120997e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15950 + }, + { + "epoch": 4.643584521384929, + "flow_matching_loss": 0.015868105739355088, + "grad_norm": 0.0771610289812088, + "learning_rate": 1.465445537718591e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15960 + }, + { + "epoch": 4.646494035496072, + "flow_matching_loss": 0.016373687610030174, + "grad_norm": 0.1077762022614479, + "learning_rate": 1.463792528012012e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15970 + }, + { + "epoch": 4.649403549607215, + "flow_matching_loss": 0.01729726856574416, + "grad_norm": 0.16635751724243164, + "learning_rate": 1.4621395623009252e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15980 + }, + { + "epoch": 4.652313063718359, + "flow_matching_loss": 0.01682655420154333, + "grad_norm": 0.10431315004825592, + "learning_rate": 1.4604866425938407e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15990 + }, + { + "epoch": 4.655222577829503, + "flow_matching_loss": 0.01496501974761486, + "grad_norm": 0.0764966607093811, + "learning_rate": 1.4588337708992115e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16000 + }, + { + "epoch": 4.6581320919406455, + "flow_matching_loss": 0.013920735381543637, + "grad_norm": 0.11357522010803223, + "learning_rate": 1.4571809492254318e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16010 + }, + { + "epoch": 4.661041606051789, + "flow_matching_loss": 0.016891164518892766, + "grad_norm": 0.12507833540439606, + "learning_rate": 1.4555281795808371e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16020 + }, + { + "epoch": 4.663951120162933, + "flow_matching_loss": 0.014697388838976621, + "grad_norm": 0.10588080435991287, + "learning_rate": 1.4538754639736966e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16030 + }, + { + "epoch": 4.666860634274077, + "flow_matching_loss": 0.016216529253870248, + "grad_norm": 0.06744769960641861, + "learning_rate": 1.4522228044122174e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16040 + }, + { + "epoch": 4.669770148385219, + "flow_matching_loss": 0.016799982637166977, + "grad_norm": 0.09289780259132385, + "learning_rate": 1.4505702029045359e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16050 + }, + { + "epoch": 4.672679662496363, + "flow_matching_loss": 0.017579679656773806, + "grad_norm": 0.09772516041994095, + "learning_rate": 1.4489176614587189e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16060 + }, + { + "epoch": 4.675589176607507, + "flow_matching_loss": 0.014983798563480377, + "grad_norm": 0.08012356609106064, + "learning_rate": 1.4472651820827595e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16070 + }, + { + "epoch": 4.67849869071865, + "flow_matching_loss": 0.016721233166754244, + "grad_norm": 0.08177757263183594, + "learning_rate": 1.4456127667845766e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16080 + }, + { + "epoch": 4.681408204829793, + "flow_matching_loss": 0.01912945881485939, + "grad_norm": 0.10400960594415665, + "learning_rate": 1.4439604175720104e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16090 + }, + { + "epoch": 4.684317718940937, + "flow_matching_loss": 0.012834546994417906, + "grad_norm": 0.10350847989320755, + "learning_rate": 1.4423081364528212e-05, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16100 + }, + { + "epoch": 4.687227233052081, + "flow_matching_loss": 0.01583029543980956, + "grad_norm": 0.09240668267011642, + "learning_rate": 1.4406559254346863e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16110 + }, + { + "epoch": 4.690136747163224, + "flow_matching_loss": 0.014887699950486421, + "grad_norm": 0.0815926268696785, + "learning_rate": 1.4390037865251978e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16120 + }, + { + "epoch": 4.693046261274367, + "flow_matching_loss": 0.01585311833769083, + "grad_norm": 0.12424477189779282, + "learning_rate": 1.4373517217318597e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16130 + }, + { + "epoch": 4.69595577538551, + "flow_matching_loss": 0.016680989786982535, + "grad_norm": 0.11314666271209717, + "learning_rate": 1.4356997330620875e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16140 + }, + { + "epoch": 4.698865289496654, + "flow_matching_loss": 0.016453519556671382, + "grad_norm": 0.14098045229911804, + "learning_rate": 1.434047822523203e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16150 + }, + { + "epoch": 4.7017748036077975, + "flow_matching_loss": 0.014999171998351813, + "grad_norm": 0.10631545633077621, + "learning_rate": 1.4323959921224322e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16160 + }, + { + "epoch": 4.704684317718941, + "flow_matching_loss": 0.01477600708603859, + "grad_norm": 0.12730175256729126, + "learning_rate": 1.4307442438669066e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16170 + }, + { + "epoch": 4.707593831830084, + "flow_matching_loss": 0.01677406169474125, + "grad_norm": 0.12828218936920166, + "learning_rate": 1.4290925797636546e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16180 + }, + { + "epoch": 4.710503345941228, + "flow_matching_loss": 0.016940113343298435, + "grad_norm": 0.13664846122264862, + "learning_rate": 1.4274410018196039e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16190 + }, + { + "epoch": 4.7134128600523715, + "flow_matching_loss": 0.01806974336504936, + "grad_norm": 0.09339256584644318, + "learning_rate": 1.4257895120415783e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16200 + }, + { + "epoch": 4.716322374163514, + "flow_matching_loss": 0.017555908486247062, + "grad_norm": 0.09442771971225739, + "learning_rate": 1.4241381124362933e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16210 + }, + { + "epoch": 4.719231888274658, + "flow_matching_loss": 0.017330969963222743, + "grad_norm": 0.09120869636535645, + "learning_rate": 1.4224868050103546e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16220 + }, + { + "epoch": 4.722141402385802, + "flow_matching_loss": 0.014031673688441515, + "grad_norm": 0.08312079310417175, + "learning_rate": 1.4208355917702573e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16230 + }, + { + "epoch": 4.725050916496945, + "flow_matching_loss": 0.01307161501608789, + "grad_norm": 0.09512653946876526, + "learning_rate": 1.41918447472238e-05, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16240 + }, + { + "epoch": 4.727960430608088, + "flow_matching_loss": 0.017109466902911662, + "grad_norm": 0.10294341295957565, + "learning_rate": 1.4175334558729869e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16250 + }, + { + "epoch": 4.730869944719232, + "flow_matching_loss": 0.016447204817086458, + "grad_norm": 0.09401630610227585, + "learning_rate": 1.4158825372282211e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16260 + }, + { + "epoch": 4.733779458830376, + "flow_matching_loss": 0.017095954902470113, + "grad_norm": 0.11541885882616043, + "learning_rate": 1.4142317207941042e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16270 + }, + { + "epoch": 4.7366889729415185, + "flow_matching_loss": 0.017041157837957142, + "grad_norm": 0.07604272663593292, + "learning_rate": 1.4125810085765339e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16280 + }, + { + "epoch": 4.739598487052662, + "flow_matching_loss": 0.02057704348117113, + "grad_norm": 0.1292538344860077, + "learning_rate": 1.410930402581281e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16290 + }, + { + "epoch": 4.742508001163806, + "flow_matching_loss": 0.01745749395340681, + "grad_norm": 0.07660280168056488, + "learning_rate": 1.4092799048139876e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16300 + }, + { + "epoch": 4.745417515274949, + "flow_matching_loss": 0.015458076912909745, + "grad_norm": 0.09201618283987045, + "learning_rate": 1.4076295172801647e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16310 + }, + { + "epoch": 4.748327029386092, + "flow_matching_loss": 0.01575237577781081, + "grad_norm": 0.12828324735164642, + "learning_rate": 1.4059792419851884e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16320 + }, + { + "epoch": 4.751236543497236, + "flow_matching_loss": 0.014425818156450988, + "grad_norm": 0.0933956578373909, + "learning_rate": 1.4043290809342984e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16330 + }, + { + "epoch": 4.754146057608379, + "flow_matching_loss": 0.01707879714667797, + "grad_norm": 0.09927718341350555, + "learning_rate": 1.4026790361325958e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16340 + }, + { + "epoch": 4.757055571719523, + "flow_matching_loss": 0.014000436011701822, + "grad_norm": 0.11219330877065659, + "learning_rate": 1.4010291095850415e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16350 + }, + { + "epoch": 4.759965085830666, + "flow_matching_loss": 0.016825653985142706, + "grad_norm": 0.14069361984729767, + "learning_rate": 1.3993793032964515e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16360 + }, + { + "epoch": 4.76287459994181, + "flow_matching_loss": 0.01364974696189165, + "grad_norm": 0.09010961651802063, + "learning_rate": 1.3977296192714956e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16370 + }, + { + "epoch": 4.765784114052953, + "flow_matching_loss": 0.015066609624773264, + "grad_norm": 0.0876690000295639, + "learning_rate": 1.396080059514696e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16380 + }, + { + "epoch": 4.768693628164097, + "flow_matching_loss": 0.017465018667280674, + "grad_norm": 0.10394837707281113, + "learning_rate": 1.3944306260304232e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16390 + }, + { + "epoch": 4.77160314227524, + "flow_matching_loss": 0.01386024821549654, + "grad_norm": 0.11151094734668732, + "learning_rate": 1.3927813208228936e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16400 + }, + { + "epoch": 4.774512656386383, + "flow_matching_loss": 0.015331445075571538, + "grad_norm": 0.11912214756011963, + "learning_rate": 1.39113214589617e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16410 + }, + { + "epoch": 4.777422170497527, + "flow_matching_loss": 0.0158418751321733, + "grad_norm": 0.07806462049484253, + "learning_rate": 1.3894831032541545e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16420 + }, + { + "epoch": 4.7803316846086705, + "flow_matching_loss": 0.015097732655704021, + "grad_norm": 0.0944133922457695, + "learning_rate": 1.3878341949005895e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16430 + }, + { + "epoch": 4.783241198719814, + "flow_matching_loss": 0.01471991604194045, + "grad_norm": 0.07223670184612274, + "learning_rate": 1.3861854228390546e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16440 + }, + { + "epoch": 4.786150712830957, + "flow_matching_loss": 0.01612548651173711, + "grad_norm": 0.07202550768852234, + "learning_rate": 1.3845367890729621e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16450 + }, + { + "epoch": 4.789060226942101, + "flow_matching_loss": 0.013630046509206295, + "grad_norm": 0.07807764410972595, + "learning_rate": 1.3828882956055591e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16460 + }, + { + "epoch": 4.7919697410532445, + "flow_matching_loss": 0.01485900729894638, + "grad_norm": 0.08449715375900269, + "learning_rate": 1.38123994443992e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16470 + }, + { + "epoch": 4.794879255164387, + "flow_matching_loss": 0.018542506638914345, + "grad_norm": 0.11931698024272919, + "learning_rate": 1.3795917375789465e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16480 + }, + { + "epoch": 4.797788769275531, + "flow_matching_loss": 0.016703213192522527, + "grad_norm": 0.13453280925750732, + "learning_rate": 1.3779436770253658e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16490 + }, + { + "epoch": 4.800698283386675, + "flow_matching_loss": 0.012933958508074283, + "grad_norm": 0.11741986125707626, + "learning_rate": 1.3762957647817258e-05, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16500 + }, + { + "epoch": 4.8036077974978175, + "flow_matching_loss": 0.016039397288113832, + "grad_norm": 0.12320774048566818, + "learning_rate": 1.374648002850397e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16510 + }, + { + "epoch": 4.806517311608961, + "flow_matching_loss": 0.014504279103130102, + "grad_norm": 0.10664926469326019, + "learning_rate": 1.3730003932335642e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16520 + }, + { + "epoch": 4.809426825720105, + "flow_matching_loss": 0.015281917434185744, + "grad_norm": 0.11018478125333786, + "learning_rate": 1.3713529379332292e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16530 + }, + { + "epoch": 4.812336339831248, + "flow_matching_loss": 0.013989289896562696, + "grad_norm": 0.07959562540054321, + "learning_rate": 1.3697056389512054e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16540 + }, + { + "epoch": 4.8152458539423915, + "flow_matching_loss": 0.015128422994166613, + "grad_norm": 0.08648963272571564, + "learning_rate": 1.3680584982891162e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16550 + }, + { + "epoch": 4.818155368053535, + "flow_matching_loss": 0.01583993323147297, + "grad_norm": 0.05856424942612648, + "learning_rate": 1.3664115179483921e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16560 + }, + { + "epoch": 4.821064882164679, + "flow_matching_loss": 0.015518709179013967, + "grad_norm": 0.10231895744800568, + "learning_rate": 1.364764699930271e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16570 + }, + { + "epoch": 4.823974396275822, + "flow_matching_loss": 0.017138865869492294, + "grad_norm": 0.09670226275920868, + "learning_rate": 1.363118046235791e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16580 + }, + { + "epoch": 4.826883910386965, + "flow_matching_loss": 0.01698900405317545, + "grad_norm": 0.08938953280448914, + "learning_rate": 1.3614715588657921e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16590 + }, + { + "epoch": 4.829793424498109, + "flow_matching_loss": 0.019100060500204564, + "grad_norm": 0.08692535758018494, + "learning_rate": 1.3598252398209118e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16600 + }, + { + "epoch": 4.832702938609252, + "flow_matching_loss": 0.015301388688385487, + "grad_norm": 0.0932508334517479, + "learning_rate": 1.358179091101582e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16610 + }, + { + "epoch": 4.835612452720396, + "flow_matching_loss": 0.01619074512273073, + "grad_norm": 0.08034513145685196, + "learning_rate": 1.3565331147080299e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16620 + }, + { + "epoch": 4.838521966831539, + "flow_matching_loss": 0.01498173289000988, + "grad_norm": 0.0770285576581955, + "learning_rate": 1.3548873126402716e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16630 + }, + { + "epoch": 4.841431480942683, + "flow_matching_loss": 0.018046111892908813, + "grad_norm": 0.112299345433712, + "learning_rate": 1.3532416868981117e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16640 + }, + { + "epoch": 4.844340995053826, + "flow_matching_loss": 0.015609795693308115, + "grad_norm": 0.11869283765554428, + "learning_rate": 1.3515962394811401e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16650 + }, + { + "epoch": 4.84725050916497, + "flow_matching_loss": 0.01612294539809227, + "grad_norm": 0.08659136295318604, + "learning_rate": 1.3499509723887313e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16660 + }, + { + "epoch": 4.850160023276112, + "flow_matching_loss": 0.01821644539013505, + "grad_norm": 0.09286724030971527, + "learning_rate": 1.3483058876200396e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16670 + }, + { + "epoch": 4.853069537387256, + "flow_matching_loss": 0.016458834894001483, + "grad_norm": 0.12307055294513702, + "learning_rate": 1.3466609871739987e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16680 + }, + { + "epoch": 4.8559790514984, + "flow_matching_loss": 0.018688556086272002, + "grad_norm": 0.13667599856853485, + "learning_rate": 1.3450162730493172e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16690 + }, + { + "epoch": 4.8588885656095435, + "flow_matching_loss": 0.015130303055047988, + "grad_norm": 0.0834074541926384, + "learning_rate": 1.3433717472444781e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16700 + }, + { + "epoch": 4.861798079720686, + "flow_matching_loss": 0.013449243828654289, + "grad_norm": 0.09138760715723038, + "learning_rate": 1.3417274117577343e-05, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16710 + }, + { + "epoch": 4.86470759383183, + "flow_matching_loss": 0.01423934604972601, + "grad_norm": 0.08911828696727753, + "learning_rate": 1.3400832685871102e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16720 + }, + { + "epoch": 4.867617107942974, + "flow_matching_loss": 0.014873762615025043, + "grad_norm": 0.09398024529218674, + "learning_rate": 1.3384393197303936e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16730 + }, + { + "epoch": 4.870526622054117, + "flow_matching_loss": 0.014392315130680799, + "grad_norm": 0.08927252143621445, + "learning_rate": 1.336795567185138e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16740 + }, + { + "epoch": 4.87343613616526, + "flow_matching_loss": 0.017125553637742996, + "grad_norm": 0.09010444581508636, + "learning_rate": 1.3351520129486573e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16750 + }, + { + "epoch": 4.876345650276404, + "flow_matching_loss": 0.0151263945735991, + "grad_norm": 0.08611121773719788, + "learning_rate": 1.3335086590180252e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16760 + }, + { + "epoch": 4.879255164387548, + "flow_matching_loss": 0.015405936259776353, + "grad_norm": 0.08573740720748901, + "learning_rate": 1.3318655073900708e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16770 + }, + { + "epoch": 4.8821646784986905, + "flow_matching_loss": 0.015253206714987755, + "grad_norm": 0.08772850781679153, + "learning_rate": 1.3302225600613796e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16780 + }, + { + "epoch": 4.885074192609834, + "flow_matching_loss": 0.0146696493960917, + "grad_norm": 0.06823518127202988, + "learning_rate": 1.3285798190282869e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16790 + }, + { + "epoch": 4.887983706720978, + "flow_matching_loss": 0.016272342950105666, + "grad_norm": 0.09771410375833511, + "learning_rate": 1.3269372862868776e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16800 + }, + { + "epoch": 4.890893220832121, + "flow_matching_loss": 0.016038994677364825, + "grad_norm": 0.2075277864933014, + "learning_rate": 1.3252949638329845e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16810 + }, + { + "epoch": 4.8938027349432645, + "flow_matching_loss": 0.015781829692423345, + "grad_norm": 0.13833604753017426, + "learning_rate": 1.3236528536621834e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16820 + }, + { + "epoch": 4.896712249054408, + "flow_matching_loss": 0.017318316083401442, + "grad_norm": 0.09890660643577576, + "learning_rate": 1.3220109577697943e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16830 + }, + { + "epoch": 4.899621763165551, + "flow_matching_loss": 0.014804049022495746, + "grad_norm": 0.11172100156545639, + "learning_rate": 1.3203692781508745e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16840 + }, + { + "epoch": 4.902531277276695, + "flow_matching_loss": 0.016105281375348567, + "grad_norm": 0.097085140645504, + "learning_rate": 1.3187278168002198e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16850 + }, + { + "epoch": 4.905440791387838, + "flow_matching_loss": 0.015211483696475625, + "grad_norm": 0.10681768506765366, + "learning_rate": 1.3170865757123606e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16860 + }, + { + "epoch": 4.908350305498981, + "flow_matching_loss": 0.015113549027591943, + "grad_norm": 0.19113628566265106, + "learning_rate": 1.3154455568815592e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16870 + }, + { + "epoch": 4.911259819610125, + "flow_matching_loss": 0.014365700911730528, + "grad_norm": 0.08625302463769913, + "learning_rate": 1.3138047623018087e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16880 + }, + { + "epoch": 4.914169333721269, + "flow_matching_loss": 0.01406736634671688, + "grad_norm": 0.06524065136909485, + "learning_rate": 1.3121641939668293e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16890 + }, + { + "epoch": 4.917078847832412, + "flow_matching_loss": 0.017736222688108684, + "grad_norm": 0.12185309082269669, + "learning_rate": 1.3105238538700659e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16900 + }, + { + "epoch": 4.919988361943555, + "flow_matching_loss": 0.014388202410191297, + "grad_norm": 0.08607664704322815, + "learning_rate": 1.3088837440046866e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16910 + }, + { + "epoch": 4.922897876054699, + "flow_matching_loss": 0.014414375089108944, + "grad_norm": 0.0902622640132904, + "learning_rate": 1.3072438663635787e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16920 + }, + { + "epoch": 4.925807390165843, + "flow_matching_loss": 0.015497835632413626, + "grad_norm": 0.13240748643875122, + "learning_rate": 1.3056042229393497e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16930 + }, + { + "epoch": 4.928716904276985, + "flow_matching_loss": 0.014833873510360718, + "grad_norm": 0.13472877442836761, + "learning_rate": 1.30396481572432e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16940 + }, + { + "epoch": 4.931626418388129, + "flow_matching_loss": 0.01463161539286375, + "grad_norm": 0.1181359738111496, + "learning_rate": 1.3023256467105237e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16950 + }, + { + "epoch": 4.934535932499273, + "flow_matching_loss": 0.013484005536884069, + "grad_norm": 0.08763053268194199, + "learning_rate": 1.3006867178897067e-05, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16960 + }, + { + "epoch": 4.9374454466104165, + "flow_matching_loss": 0.014024478010833263, + "grad_norm": 0.07843764126300812, + "learning_rate": 1.2990480312533213e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16970 + }, + { + "epoch": 4.940354960721559, + "flow_matching_loss": 0.016519707441329957, + "grad_norm": 0.10667947679758072, + "learning_rate": 1.2974095887925258e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16980 + }, + { + "epoch": 4.943264474832703, + "flow_matching_loss": 0.01681738840416074, + "grad_norm": 0.07688947767019272, + "learning_rate": 1.2957713924981837e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16990 + }, + { + "epoch": 4.946173988943847, + "flow_matching_loss": 0.0169492956250906, + "grad_norm": 0.1174900010228157, + "learning_rate": 1.2941334443608572e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17000 + }, + { + "epoch": 4.94908350305499, + "flow_matching_loss": 0.01638036370277405, + "grad_norm": 0.07732567936182022, + "learning_rate": 1.2924957463708073e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17010 + }, + { + "epoch": 4.951993017166133, + "flow_matching_loss": 0.015642932057380675, + "grad_norm": 0.12911543250083923, + "learning_rate": 1.2908583005179924e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17020 + }, + { + "epoch": 4.954902531277277, + "flow_matching_loss": 0.014157989155501127, + "grad_norm": 0.09044159948825836, + "learning_rate": 1.2892211087920624e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17030 + }, + { + "epoch": 4.95781204538842, + "flow_matching_loss": 0.01611937377601862, + "grad_norm": 0.10393234342336655, + "learning_rate": 1.287584173182361e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17040 + }, + { + "epoch": 4.9607215594995635, + "flow_matching_loss": 0.015084230713546276, + "grad_norm": 0.07779600471258163, + "learning_rate": 1.2859474956779188e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17050 + }, + { + "epoch": 4.963631073610707, + "flow_matching_loss": 0.015954060200601815, + "grad_norm": 0.0646548792719841, + "learning_rate": 1.2843110782674532e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17060 + }, + { + "epoch": 4.96654058772185, + "flow_matching_loss": 0.013670385908335447, + "grad_norm": 0.08797180652618408, + "learning_rate": 1.2826749229393655e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17070 + }, + { + "epoch": 4.969450101832994, + "flow_matching_loss": 0.0166532751172781, + "grad_norm": 0.11257839947938919, + "learning_rate": 1.2810390316817381e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17080 + }, + { + "epoch": 4.9723596159441374, + "flow_matching_loss": 0.01615113439038396, + "grad_norm": 0.0954844057559967, + "learning_rate": 1.2794034064823346e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17090 + }, + { + "epoch": 4.975269130055281, + "flow_matching_loss": 0.015235940739512444, + "grad_norm": 0.10236464440822601, + "learning_rate": 1.277768049328593e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17100 + }, + { + "epoch": 4.978178644166424, + "flow_matching_loss": 0.01802357444539666, + "grad_norm": 0.08032819628715515, + "learning_rate": 1.2761329622076269e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17110 + }, + { + "epoch": 4.981088158277568, + "flow_matching_loss": 0.014065607078373432, + "grad_norm": 0.09858310222625732, + "learning_rate": 1.274498147106221e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17120 + }, + { + "epoch": 4.983997672388711, + "flow_matching_loss": 0.01649904455989599, + "grad_norm": 0.09618568420410156, + "learning_rate": 1.272863606010829e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17130 + }, + { + "epoch": 4.986907186499854, + "flow_matching_loss": 0.01712132040411234, + "grad_norm": 0.10167849063873291, + "learning_rate": 1.2712293409075739e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17140 + }, + { + "epoch": 4.989816700610998, + "flow_matching_loss": 0.014961433596909046, + "grad_norm": 0.0812094658613205, + "learning_rate": 1.2695953537822413e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17150 + }, + { + "epoch": 4.992726214722142, + "flow_matching_loss": 0.015660582482814787, + "grad_norm": 0.0735451951622963, + "learning_rate": 1.2679616466202792e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17160 + }, + { + "epoch": 4.995635728833285, + "flow_matching_loss": 0.015377873741090297, + "grad_norm": 0.1480407565832138, + "learning_rate": 1.2663282214067967e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17170 + }, + { + "epoch": 4.998545242944428, + "flow_matching_loss": 0.017426365055143833, + "grad_norm": 0.10199553519487381, + "learning_rate": 1.264695080126559e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17180 + }, + { + "epoch": 5.001454757055572, + "flow_matching_loss": 0.013831419963389635, + "grad_norm": 0.10483542829751968, + "learning_rate": 1.263062224763986e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17190 + }, + { + "epoch": 5.004364271166716, + "flow_matching_loss": 0.015288940351456403, + "grad_norm": 0.08037164807319641, + "learning_rate": 1.2614296573031522e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17200 + }, + { + "epoch": 5.007273785277858, + "flow_matching_loss": 0.015316598769277334, + "grad_norm": 0.09553515166044235, + "learning_rate": 1.259797379727781e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17210 + }, + { + "epoch": 5.010183299389002, + "flow_matching_loss": 0.01827065572142601, + "grad_norm": 0.13313153386116028, + "learning_rate": 1.2581653940212423e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17220 + }, + { + "epoch": 5.013092813500146, + "flow_matching_loss": 0.01744036441668868, + "grad_norm": 0.08782559633255005, + "learning_rate": 1.2565337021665537e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17230 + }, + { + "epoch": 5.016002327611289, + "flow_matching_loss": 0.014413054566830397, + "grad_norm": 0.08721254765987396, + "learning_rate": 1.2549023061463739e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17240 + }, + { + "epoch": 5.018911841722432, + "flow_matching_loss": 0.01613846719264984, + "grad_norm": 0.09684908390045166, + "learning_rate": 1.2532712079430038e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17250 + }, + { + "epoch": 5.021821355833576, + "flow_matching_loss": 0.0167623252607882, + "grad_norm": 0.09035143256187439, + "learning_rate": 1.2516404095383813e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17260 + }, + { + "epoch": 5.024730869944719, + "flow_matching_loss": 0.01505950205028057, + "grad_norm": 0.11153571307659149, + "learning_rate": 1.2500099129140796e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17270 + }, + { + "epoch": 5.0276403840558626, + "flow_matching_loss": 0.015750865824520587, + "grad_norm": 0.09106987714767456, + "learning_rate": 1.2483797200513065e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17280 + }, + { + "epoch": 5.030549898167006, + "flow_matching_loss": 0.014228528551757335, + "grad_norm": 0.08125599473714828, + "learning_rate": 1.2467498329308992e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17290 + }, + { + "epoch": 5.03345941227815, + "flow_matching_loss": 0.01494707865640521, + "grad_norm": 0.10148292034864426, + "learning_rate": 1.2451202535333253e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17300 + }, + { + "epoch": 5.036368926389293, + "flow_matching_loss": 0.01603147778660059, + "grad_norm": 0.09836816042661667, + "learning_rate": 1.2434909838386768e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17310 + }, + { + "epoch": 5.0392784405004365, + "flow_matching_loss": 0.017079062387347223, + "grad_norm": 0.1032223328948021, + "learning_rate": 1.2418620258266703e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17320 + }, + { + "epoch": 5.04218795461158, + "flow_matching_loss": 0.017065037880092858, + "grad_norm": 0.10315623134374619, + "learning_rate": 1.2402333814766432e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17330 + }, + { + "epoch": 5.045097468722723, + "flow_matching_loss": 0.01322889355942607, + "grad_norm": 0.0844450369477272, + "learning_rate": 1.2386050527675515e-05, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17340 + }, + { + "epoch": 5.048006982833867, + "flow_matching_loss": 0.015328942891210317, + "grad_norm": 0.09483646601438522, + "learning_rate": 1.2369770416779693e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17350 + }, + { + "epoch": 5.05091649694501, + "flow_matching_loss": 0.014826100785285234, + "grad_norm": 0.13956689834594727, + "learning_rate": 1.2353493501860828e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17360 + }, + { + "epoch": 5.053826011056153, + "flow_matching_loss": 0.016885695047676563, + "grad_norm": 0.10460329055786133, + "learning_rate": 1.2337219802696905e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17370 + }, + { + "epoch": 5.056735525167297, + "flow_matching_loss": 0.01508739059790969, + "grad_norm": 0.09932221472263336, + "learning_rate": 1.2320949339062008e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17380 + }, + { + "epoch": 5.059645039278441, + "flow_matching_loss": 0.016095047909766437, + "grad_norm": 0.12081830203533173, + "learning_rate": 1.2304682130726281e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17390 + }, + { + "epoch": 5.0625545533895835, + "flow_matching_loss": 0.018000547122210265, + "grad_norm": 0.11717186868190765, + "learning_rate": 1.2288418197455913e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17400 + }, + { + "epoch": 5.065464067500727, + "flow_matching_loss": 0.01487006302922964, + "grad_norm": 0.09577067196369171, + "learning_rate": 1.2272157559013126e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17410 + }, + { + "epoch": 5.068373581611871, + "flow_matching_loss": 0.014045093022286893, + "grad_norm": 0.08171246945858002, + "learning_rate": 1.2255900235156123e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17420 + }, + { + "epoch": 5.071283095723015, + "flow_matching_loss": 0.015001492947340012, + "grad_norm": 0.09241514652967453, + "learning_rate": 1.2239646245639087e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17430 + }, + { + "epoch": 5.074192609834157, + "flow_matching_loss": 0.013240514881908893, + "grad_norm": 0.09344059973955154, + "learning_rate": 1.2223395610212142e-05, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17440 + }, + { + "epoch": 5.077102123945301, + "flow_matching_loss": 0.012916357908397912, + "grad_norm": 0.08109825104475021, + "learning_rate": 1.2207148348621348e-05, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17450 + }, + { + "epoch": 5.080011638056445, + "flow_matching_loss": 0.017113077640533447, + "grad_norm": 0.12527799606323242, + "learning_rate": 1.219090448060866e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17460 + }, + { + "epoch": 5.082921152167588, + "flow_matching_loss": 0.017015609331429003, + "grad_norm": 0.11158061772584915, + "learning_rate": 1.217466402591191e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17470 + }, + { + "epoch": 5.085830666278731, + "flow_matching_loss": 0.015692807827144863, + "grad_norm": 0.108421690762043, + "learning_rate": 1.2158427004264785e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17480 + }, + { + "epoch": 5.088740180389875, + "flow_matching_loss": 0.017578754853457213, + "grad_norm": 0.07610312104225159, + "learning_rate": 1.2142193435396788e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17490 + }, + { + "epoch": 5.091649694501018, + "flow_matching_loss": 0.0139323765411973, + "grad_norm": 0.07445422559976578, + "learning_rate": 1.2125963339033234e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17500 + }, + { + "epoch": 5.094559208612162, + "flow_matching_loss": 0.013924924097955227, + "grad_norm": 0.09771090000867844, + "learning_rate": 1.2109736734895237e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17510 + }, + { + "epoch": 5.097468722723305, + "flow_matching_loss": 0.01732043279334903, + "grad_norm": 0.09871049970388412, + "learning_rate": 1.2093513642699634e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17520 + }, + { + "epoch": 5.100378236834449, + "flow_matching_loss": 0.018016692716628313, + "grad_norm": 0.08884324878454208, + "learning_rate": 1.2077294082159024e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17530 + }, + { + "epoch": 5.103287750945592, + "flow_matching_loss": 0.017186560574918984, + "grad_norm": 0.10663319379091263, + "learning_rate": 1.2061078072981697e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17540 + }, + { + "epoch": 5.1061972650567355, + "flow_matching_loss": 0.01525884997099638, + "grad_norm": 0.08080203831195831, + "learning_rate": 1.2044865634871626e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17550 + }, + { + "epoch": 5.109106779167879, + "flow_matching_loss": 0.015759120415896177, + "grad_norm": 0.07992065697908401, + "learning_rate": 1.2028656787528468e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17560 + }, + { + "epoch": 5.112016293279022, + "flow_matching_loss": 0.01610362809151411, + "grad_norm": 0.06954941153526306, + "learning_rate": 1.2012451550647491e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17570 + }, + { + "epoch": 5.114925807390166, + "flow_matching_loss": 0.01680936189368367, + "grad_norm": 0.09724979102611542, + "learning_rate": 1.1996249943919589e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17580 + }, + { + "epoch": 5.1178353215013095, + "flow_matching_loss": 0.016928659286350012, + "grad_norm": 0.13867530226707458, + "learning_rate": 1.1980051987031237e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17590 + }, + { + "epoch": 5.120744835612452, + "flow_matching_loss": 0.016838896740227937, + "grad_norm": 0.09064623713493347, + "learning_rate": 1.1963857699664483e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17600 + }, + { + "epoch": 5.123654349723596, + "flow_matching_loss": 0.014893598482012748, + "grad_norm": 0.09808480739593506, + "learning_rate": 1.1947667101496911e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17610 + }, + { + "epoch": 5.12656386383474, + "flow_matching_loss": 0.013510762620717287, + "grad_norm": 0.08183302730321884, + "learning_rate": 1.1931480212201633e-05, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17620 + }, + { + "epoch": 5.129473377945883, + "flow_matching_loss": 0.01892154198139906, + "grad_norm": 0.093246228992939, + "learning_rate": 1.1915297051447237e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17630 + }, + { + "epoch": 5.132382892057026, + "flow_matching_loss": 0.016878945473581552, + "grad_norm": 0.11501695215702057, + "learning_rate": 1.1899117638897793e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17640 + }, + { + "epoch": 5.13529240616817, + "flow_matching_loss": 0.01653000619262457, + "grad_norm": 0.07555361837148666, + "learning_rate": 1.1882941994212811e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17650 + }, + { + "epoch": 5.138201920279314, + "flow_matching_loss": 0.017064678203314542, + "grad_norm": 0.10985544323921204, + "learning_rate": 1.1866770137047221e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17660 + }, + { + "epoch": 5.1411114343904565, + "flow_matching_loss": 0.013424926344305277, + "grad_norm": 0.09500008076429367, + "learning_rate": 1.1850602087051363e-05, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17670 + }, + { + "epoch": 5.1440209485016, + "flow_matching_loss": 0.015106313116848468, + "grad_norm": 0.11393161118030548, + "learning_rate": 1.183443786387094e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17680 + }, + { + "epoch": 5.146930462612744, + "flow_matching_loss": 0.01702026091516018, + "grad_norm": 0.12441559135913849, + "learning_rate": 1.1818277487147005e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17690 + }, + { + "epoch": 5.149839976723887, + "flow_matching_loss": 0.014682843443006277, + "grad_norm": 0.1027454063296318, + "learning_rate": 1.180212097651594e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17700 + }, + { + "epoch": 5.15274949083503, + "flow_matching_loss": 0.013969277869910002, + "grad_norm": 0.06825157254934311, + "learning_rate": 1.1785968351609422e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17710 + }, + { + "epoch": 5.155659004946174, + "flow_matching_loss": 0.01457156501710415, + "grad_norm": 0.0829966738820076, + "learning_rate": 1.1769819632054427e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17720 + }, + { + "epoch": 5.158568519057318, + "flow_matching_loss": 0.015252304635941983, + "grad_norm": 0.08775914460420609, + "learning_rate": 1.1753674837473165e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17730 + }, + { + "epoch": 5.161478033168461, + "flow_matching_loss": 0.013341128919273615, + "grad_norm": 0.08870364725589752, + "learning_rate": 1.1737533987483081e-05, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17740 + }, + { + "epoch": 5.164387547279604, + "flow_matching_loss": 0.015322318207472563, + "grad_norm": 0.11159223318099976, + "learning_rate": 1.1721397101696835e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17750 + }, + { + "epoch": 5.167297061390748, + "flow_matching_loss": 0.012571405526250601, + "grad_norm": 0.09277341514825821, + "learning_rate": 1.1705264199722257e-05, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17760 + }, + { + "epoch": 5.170206575501891, + "flow_matching_loss": 0.014967778511345387, + "grad_norm": 0.10813047736883163, + "learning_rate": 1.1689135301162356e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17770 + }, + { + "epoch": 5.173116089613035, + "flow_matching_loss": 0.01548331305384636, + "grad_norm": 0.08932217210531235, + "learning_rate": 1.1673010425615258e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17780 + }, + { + "epoch": 5.176025603724178, + "flow_matching_loss": 0.01565873371437192, + "grad_norm": 0.10245919227600098, + "learning_rate": 1.1656889592674208e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17790 + }, + { + "epoch": 5.178935117835321, + "flow_matching_loss": 0.014707857882604003, + "grad_norm": 0.09105143696069717, + "learning_rate": 1.1640772821927533e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17800 + }, + { + "epoch": 5.181844631946465, + "flow_matching_loss": 0.01601873692125082, + "grad_norm": 0.08785451203584671, + "learning_rate": 1.162466013295864e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17810 + }, + { + "epoch": 5.1847541460576085, + "flow_matching_loss": 0.017720568645745515, + "grad_norm": 0.09593483060598373, + "learning_rate": 1.160855154534595e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17820 + }, + { + "epoch": 5.187663660168752, + "flow_matching_loss": 0.015069333650171757, + "grad_norm": 0.08920294791460037, + "learning_rate": 1.1592447078662931e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17830 + }, + { + "epoch": 5.190573174279895, + "flow_matching_loss": 0.016806321032345296, + "grad_norm": 0.09380742907524109, + "learning_rate": 1.1576346752478025e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17840 + }, + { + "epoch": 5.193482688391039, + "flow_matching_loss": 0.0161131476983428, + "grad_norm": 0.08387945592403412, + "learning_rate": 1.1560250586354643e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17850 + }, + { + "epoch": 5.1963922025021825, + "flow_matching_loss": 0.01445962069556117, + "grad_norm": 0.07746019959449768, + "learning_rate": 1.1544158599851146e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17860 + }, + { + "epoch": 5.199301716613325, + "flow_matching_loss": 0.015599421132355929, + "grad_norm": 0.08869531750679016, + "learning_rate": 1.1528070812520809e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17870 + }, + { + "epoch": 5.202211230724469, + "flow_matching_loss": 0.01455440353602171, + "grad_norm": 0.10149404406547546, + "learning_rate": 1.1511987243911821e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17880 + }, + { + "epoch": 5.205120744835613, + "flow_matching_loss": 0.015160079300403594, + "grad_norm": 0.0846322625875473, + "learning_rate": 1.1495907913567228e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17890 + }, + { + "epoch": 5.2080302589467555, + "flow_matching_loss": 0.01354620922356844, + "grad_norm": 0.10278790444135666, + "learning_rate": 1.1479832841024936e-05, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17900 + }, + { + "epoch": 5.210939773057899, + "flow_matching_loss": 0.013893352448940277, + "grad_norm": 0.11013484001159668, + "learning_rate": 1.1463762045817675e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17910 + }, + { + "epoch": 5.213849287169043, + "flow_matching_loss": 0.017055634502321483, + "grad_norm": 0.10997111350297928, + "learning_rate": 1.1447695547472964e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17920 + }, + { + "epoch": 5.216758801280186, + "flow_matching_loss": 0.01437161723151803, + "grad_norm": 0.08549386262893677, + "learning_rate": 1.1431633365513132e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17930 + }, + { + "epoch": 5.2196683153913295, + "flow_matching_loss": 0.01563930595293641, + "grad_norm": 0.08036988228559494, + "learning_rate": 1.1415575519455235e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17940 + }, + { + "epoch": 5.222577829502473, + "flow_matching_loss": 0.015575766656547784, + "grad_norm": 0.09741239994764328, + "learning_rate": 1.1399522028811069e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17950 + }, + { + "epoch": 5.225487343613617, + "flow_matching_loss": 0.01539246141910553, + "grad_norm": 0.09641503542661667, + "learning_rate": 1.1383472913087143e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17960 + }, + { + "epoch": 5.22839685772476, + "flow_matching_loss": 0.017427484272047877, + "grad_norm": 0.0925009697675705, + "learning_rate": 1.1367428191784643e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17970 + }, + { + "epoch": 5.231306371835903, + "flow_matching_loss": 0.013023786153644324, + "grad_norm": 0.08477144688367844, + "learning_rate": 1.1351387884399413e-05, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17980 + }, + { + "epoch": 5.234215885947047, + "flow_matching_loss": 0.014051786437630654, + "grad_norm": 0.09657323360443115, + "learning_rate": 1.1335352010421953e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17990 + }, + { + "epoch": 5.23712540005819, + "flow_matching_loss": 0.014566648844629526, + "grad_norm": 0.10304946452379227, + "learning_rate": 1.1319320589337352e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18000 + }, + { + "epoch": 5.240034914169334, + "flow_matching_loss": 0.014092113636434079, + "grad_norm": 0.10444691777229309, + "learning_rate": 1.1303293640625299e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18010 + }, + { + "epoch": 5.242944428280477, + "flow_matching_loss": 0.015533959865570069, + "grad_norm": 0.10191742330789566, + "learning_rate": 1.1287271183760052e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18020 + }, + { + "epoch": 5.24585394239162, + "flow_matching_loss": 0.015870952885597943, + "grad_norm": 0.10765163600444794, + "learning_rate": 1.1271253238210398e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18030 + }, + { + "epoch": 5.248763456502764, + "flow_matching_loss": 0.01475082766264677, + "grad_norm": 0.0836259126663208, + "learning_rate": 1.1255239823439663e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18040 + }, + { + "epoch": 5.251672970613908, + "flow_matching_loss": 0.01463257083669305, + "grad_norm": 0.06737785786390305, + "learning_rate": 1.1239230958905652e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18050 + }, + { + "epoch": 5.254582484725051, + "flow_matching_loss": 0.014649744611233473, + "grad_norm": 0.08940593898296356, + "learning_rate": 1.1223226664060641e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18060 + }, + { + "epoch": 5.257491998836194, + "flow_matching_loss": 0.01488332860171795, + "grad_norm": 0.09199492633342743, + "learning_rate": 1.1207226958351362e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18070 + }, + { + "epoch": 5.260401512947338, + "flow_matching_loss": 0.016103813704103232, + "grad_norm": 0.14211200177669525, + "learning_rate": 1.1191231861218955e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18080 + }, + { + "epoch": 5.2633110270584815, + "flow_matching_loss": 0.016207800433039665, + "grad_norm": 0.09535577893257141, + "learning_rate": 1.1175241392098988e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18090 + }, + { + "epoch": 5.266220541169624, + "flow_matching_loss": 0.014710820466279983, + "grad_norm": 0.07229983806610107, + "learning_rate": 1.1159255570421376e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18100 + }, + { + "epoch": 5.269130055280768, + "flow_matching_loss": 0.015076295845210553, + "grad_norm": 0.094437375664711, + "learning_rate": 1.1143274415610408e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18110 + }, + { + "epoch": 5.272039569391912, + "flow_matching_loss": 0.014375308016315102, + "grad_norm": 0.07598117738962173, + "learning_rate": 1.1127297947084692e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18120 + }, + { + "epoch": 5.274949083503055, + "flow_matching_loss": 0.01450026948004961, + "grad_norm": 0.0808749571442604, + "learning_rate": 1.1111326184257135e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18130 + }, + { + "epoch": 5.277858597614198, + "flow_matching_loss": 0.01866740854457021, + "grad_norm": 0.11327411234378815, + "learning_rate": 1.109535914653495e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18140 + }, + { + "epoch": 5.280768111725342, + "flow_matching_loss": 0.013697562459856271, + "grad_norm": 0.09326351433992386, + "learning_rate": 1.1079396853319582e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18150 + }, + { + "epoch": 5.283677625836486, + "flow_matching_loss": 0.014859538152813911, + "grad_norm": 0.10304900258779526, + "learning_rate": 1.1063439324006724e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18160 + }, + { + "epoch": 5.2865871399476285, + "flow_matching_loss": 0.017370480950921773, + "grad_norm": 0.10586794465780258, + "learning_rate": 1.1047486577986285e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18170 + }, + { + "epoch": 5.289496654058772, + "flow_matching_loss": 0.015619388502091169, + "grad_norm": 0.07839105278253555, + "learning_rate": 1.1031538634642353e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18180 + }, + { + "epoch": 5.292406168169916, + "flow_matching_loss": 0.01543574221432209, + "grad_norm": 0.11321055889129639, + "learning_rate": 1.1015595513353172e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18190 + }, + { + "epoch": 5.295315682281059, + "flow_matching_loss": 0.014115274511277675, + "grad_norm": 0.07511354982852936, + "learning_rate": 1.0999657233491156e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18200 + }, + { + "epoch": 5.2982251963922025, + "flow_matching_loss": 0.014746448770165444, + "grad_norm": 0.081536203622818, + "learning_rate": 1.0983723814422808e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18210 + }, + { + "epoch": 5.301134710503346, + "flow_matching_loss": 0.017672530841082335, + "grad_norm": 0.09940407425165176, + "learning_rate": 1.0967795275508735e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18220 + }, + { + "epoch": 5.304044224614489, + "flow_matching_loss": 0.015052763745188713, + "grad_norm": 0.06013740971684456, + "learning_rate": 1.095187163610361e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18230 + }, + { + "epoch": 5.306953738725633, + "flow_matching_loss": 0.014621639158576727, + "grad_norm": 0.08241251856088638, + "learning_rate": 1.0935952915556163e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18240 + }, + { + "epoch": 5.309863252836776, + "flow_matching_loss": 0.015127378981560469, + "grad_norm": 0.08762352913618088, + "learning_rate": 1.0920039133209137e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18250 + }, + { + "epoch": 5.31277276694792, + "flow_matching_loss": 0.01636416558176279, + "grad_norm": 0.10146971046924591, + "learning_rate": 1.0904130308399283e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18260 + }, + { + "epoch": 5.315682281059063, + "flow_matching_loss": 0.01612417371943593, + "grad_norm": 0.09929634630680084, + "learning_rate": 1.0888226460457318e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18270 + }, + { + "epoch": 5.318591795170207, + "flow_matching_loss": 0.015011216280981898, + "grad_norm": 0.0925179049372673, + "learning_rate": 1.087232760870792e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18280 + }, + { + "epoch": 5.32150130928135, + "flow_matching_loss": 0.016490601096302272, + "grad_norm": 0.09322243183851242, + "learning_rate": 1.0856433772469686e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18290 + }, + { + "epoch": 5.324410823392493, + "flow_matching_loss": 0.013861291296780109, + "grad_norm": 0.09384167939424515, + "learning_rate": 1.0840544971055135e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18300 + }, + { + "epoch": 5.327320337503637, + "flow_matching_loss": 0.016193929966539143, + "grad_norm": 0.08073306828737259, + "learning_rate": 1.0824661223770655e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18310 + }, + { + "epoch": 5.330229851614781, + "flow_matching_loss": 0.015003992803394794, + "grad_norm": 0.07304829359054565, + "learning_rate": 1.0808782549916498e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18320 + }, + { + "epoch": 5.333139365725923, + "flow_matching_loss": 0.018336824607104063, + "grad_norm": 0.13413552939891815, + "learning_rate": 1.079290896878675e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18330 + }, + { + "epoch": 5.336048879837067, + "flow_matching_loss": 0.015611177496612071, + "grad_norm": 0.09483392536640167, + "learning_rate": 1.07770404996693e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18340 + }, + { + "epoch": 5.338958393948211, + "flow_matching_loss": 0.013845154270529746, + "grad_norm": 0.0775378867983818, + "learning_rate": 1.0761177161845848e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18350 + }, + { + "epoch": 5.3418679080593545, + "flow_matching_loss": 0.015189172327518463, + "grad_norm": 0.08496595919132233, + "learning_rate": 1.0745318974591841e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18360 + }, + { + "epoch": 5.344777422170497, + "flow_matching_loss": 0.01573526384308934, + "grad_norm": 0.0922253429889679, + "learning_rate": 1.0729465957176469e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18370 + }, + { + "epoch": 5.347686936281641, + "flow_matching_loss": 0.01769481198862195, + "grad_norm": 0.07992938160896301, + "learning_rate": 1.0713618128862642e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18380 + }, + { + "epoch": 5.350596450392785, + "flow_matching_loss": 0.014826903957873583, + "grad_norm": 0.09232024848461151, + "learning_rate": 1.0697775508906969e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18390 + }, + { + "epoch": 5.353505964503928, + "flow_matching_loss": 0.015872006118297578, + "grad_norm": 0.09404067695140839, + "learning_rate": 1.068193811655972e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18400 + }, + { + "epoch": 5.356415478615071, + "flow_matching_loss": 0.016646180860698225, + "grad_norm": 0.08865050971508026, + "learning_rate": 1.0666105971064835e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18410 + }, + { + "epoch": 5.359324992726215, + "flow_matching_loss": 0.013963623158633708, + "grad_norm": 0.11355552077293396, + "learning_rate": 1.0650279091659852e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18420 + }, + { + "epoch": 5.362234506837358, + "flow_matching_loss": 0.01697285184636712, + "grad_norm": 0.1253308355808258, + "learning_rate": 1.0634457497575926e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18430 + }, + { + "epoch": 5.3651440209485015, + "flow_matching_loss": 0.014187132474035025, + "grad_norm": 0.08767858147621155, + "learning_rate": 1.061864120803778e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18440 + }, + { + "epoch": 5.368053535059645, + "flow_matching_loss": 0.014922519773244857, + "grad_norm": 0.11501943320035934, + "learning_rate": 1.0602830242263701e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18450 + }, + { + "epoch": 5.370963049170788, + "flow_matching_loss": 0.016335638845339416, + "grad_norm": 0.08682071417570114, + "learning_rate": 1.0587024619465506e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18460 + }, + { + "epoch": 5.373872563281932, + "flow_matching_loss": 0.01377326943911612, + "grad_norm": 0.07096885144710541, + "learning_rate": 1.0571224358848517e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18470 + }, + { + "epoch": 5.3767820773930755, + "flow_matching_loss": 0.014787849225103854, + "grad_norm": 0.10005169361829758, + "learning_rate": 1.0555429479611537e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18480 + }, + { + "epoch": 5.379691591504219, + "flow_matching_loss": 0.015072980988770723, + "grad_norm": 0.07161211967468262, + "learning_rate": 1.0539640000946835e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18490 + }, + { + "epoch": 5.382601105615362, + "flow_matching_loss": 0.01442957492545247, + "grad_norm": 0.09415078163146973, + "learning_rate": 1.0523855942040111e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18500 + }, + { + "epoch": 5.385510619726506, + "flow_matching_loss": 0.01702263243496418, + "grad_norm": 0.08266570419073105, + "learning_rate": 1.0508077322070496e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18510 + }, + { + "epoch": 5.388420133837649, + "flow_matching_loss": 0.016348663810640575, + "grad_norm": 0.0988672599196434, + "learning_rate": 1.0492304160210495e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18520 + }, + { + "epoch": 5.391329647948792, + "flow_matching_loss": 0.013677653018385172, + "grad_norm": 0.09534667432308197, + "learning_rate": 1.0476536475625985e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18530 + }, + { + "epoch": 5.394239162059936, + "flow_matching_loss": 0.016046375408768652, + "grad_norm": 0.12787432968616486, + "learning_rate": 1.0460774287476195e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18540 + }, + { + "epoch": 5.39714867617108, + "flow_matching_loss": 0.01528844740241766, + "grad_norm": 0.1124769002199173, + "learning_rate": 1.044501761491366e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18550 + }, + { + "epoch": 5.400058190282223, + "flow_matching_loss": 0.01481919726356864, + "grad_norm": 0.10649170726537704, + "learning_rate": 1.0429266477084235e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18560 + }, + { + "epoch": 5.402967704393366, + "flow_matching_loss": 0.01646175803616643, + "grad_norm": 0.09168415516614914, + "learning_rate": 1.0413520893127034e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18570 + }, + { + "epoch": 5.40587721850451, + "flow_matching_loss": 0.013533691503107547, + "grad_norm": 0.09073606133460999, + "learning_rate": 1.0397780882174424e-05, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18580 + }, + { + "epoch": 5.408786732615654, + "flow_matching_loss": 0.01645332742482424, + "grad_norm": 0.0739060640335083, + "learning_rate": 1.0382046463351998e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18590 + }, + { + "epoch": 5.411696246726796, + "flow_matching_loss": 0.015620932262390853, + "grad_norm": 0.1512128710746765, + "learning_rate": 1.0366317655778568e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18600 + }, + { + "epoch": 5.41460576083794, + "flow_matching_loss": 0.01702398182824254, + "grad_norm": 0.07886990159749985, + "learning_rate": 1.035059447856611e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18610 + }, + { + "epoch": 5.417515274949084, + "flow_matching_loss": 0.014694052189588547, + "grad_norm": 0.08293836563825607, + "learning_rate": 1.0334876950819777e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18620 + }, + { + "epoch": 5.420424789060227, + "flow_matching_loss": 0.014120027143508195, + "grad_norm": 0.07572274655103683, + "learning_rate": 1.0319165091637837e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18630 + }, + { + "epoch": 5.42333430317137, + "flow_matching_loss": 0.014811498485505581, + "grad_norm": 0.0691199079155922, + "learning_rate": 1.0303458920111686e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18640 + }, + { + "epoch": 5.426243817282514, + "flow_matching_loss": 0.01349575649946928, + "grad_norm": 0.09769117087125778, + "learning_rate": 1.0287758455325803e-05, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18650 + }, + { + "epoch": 5.429153331393657, + "flow_matching_loss": 0.01470928369089961, + "grad_norm": 0.08284135907888412, + "learning_rate": 1.0272063716357729e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18660 + }, + { + "epoch": 5.432062845504801, + "flow_matching_loss": 0.012673334777355194, + "grad_norm": 0.10697383433580399, + "learning_rate": 1.0256374722278057e-05, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18670 + }, + { + "epoch": 5.434972359615944, + "flow_matching_loss": 0.016147681698203086, + "grad_norm": 0.1116928681731224, + "learning_rate": 1.0240691492150388e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18680 + }, + { + "epoch": 5.437881873727088, + "flow_matching_loss": 0.015470081195235252, + "grad_norm": 0.08340135961771011, + "learning_rate": 1.0225014045031334e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18690 + }, + { + "epoch": 5.440791387838231, + "flow_matching_loss": 0.01587310144677758, + "grad_norm": 0.1074332520365715, + "learning_rate": 1.0209342399970468e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18700 + }, + { + "epoch": 5.4437009019493745, + "flow_matching_loss": 0.01438881354406476, + "grad_norm": 0.09992524236440659, + "learning_rate": 1.019367657601031e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18710 + }, + { + "epoch": 5.446610416060518, + "flow_matching_loss": 0.016198138240724802, + "grad_norm": 0.10447093844413757, + "learning_rate": 1.0178016592186323e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18720 + }, + { + "epoch": 5.449519930171661, + "flow_matching_loss": 0.014047264400869609, + "grad_norm": 0.09940777719020844, + "learning_rate": 1.0162362467526861e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18730 + }, + { + "epoch": 5.452429444282805, + "flow_matching_loss": 0.017739979550242424, + "grad_norm": 0.08233474940061569, + "learning_rate": 1.0146714221053159e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18740 + }, + { + "epoch": 5.4553389583939484, + "flow_matching_loss": 0.016112689301371573, + "grad_norm": 0.10418575257062912, + "learning_rate": 1.0131071871779314e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18750 + }, + { + "epoch": 5.458248472505091, + "flow_matching_loss": 0.015143813844770194, + "grad_norm": 0.08765964955091476, + "learning_rate": 1.0115435438712249e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18760 + }, + { + "epoch": 5.461157986616235, + "flow_matching_loss": 0.015532385837286711, + "grad_norm": 0.09142987430095673, + "learning_rate": 1.0099804940851716e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18770 + }, + { + "epoch": 5.464067500727379, + "flow_matching_loss": 0.015146846137940883, + "grad_norm": 0.0737798735499382, + "learning_rate": 1.0084180397190238e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18780 + }, + { + "epoch": 5.4669770148385215, + "flow_matching_loss": 0.015078038070350885, + "grad_norm": 0.11539913713932037, + "learning_rate": 1.006856182671311e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18790 + }, + { + "epoch": 5.469886528949665, + "flow_matching_loss": 0.017047253809869288, + "grad_norm": 0.12038485705852509, + "learning_rate": 1.0052949248398367e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18800 + }, + { + "epoch": 5.472796043060809, + "flow_matching_loss": 0.013391355890780688, + "grad_norm": 0.0955917164683342, + "learning_rate": 1.003734268121676e-05, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18810 + }, + { + "epoch": 5.475705557171953, + "flow_matching_loss": 0.016188598610460757, + "grad_norm": 0.1029442548751831, + "learning_rate": 1.0021742144131744e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18820 + }, + { + "epoch": 5.478615071283095, + "flow_matching_loss": 0.014522118028253317, + "grad_norm": 0.1315307468175888, + "learning_rate": 1.000614765609944e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18830 + }, + { + "epoch": 5.481524585394239, + "flow_matching_loss": 0.015164781920611858, + "grad_norm": 0.09535947442054749, + "learning_rate": 9.990559236068628e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18840 + }, + { + "epoch": 5.484434099505383, + "flow_matching_loss": 0.0167053135111928, + "grad_norm": 0.12874488532543182, + "learning_rate": 9.974976902980703e-06, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18850 + }, + { + "epoch": 5.487343613616526, + "flow_matching_loss": 0.013924761768430471, + "grad_norm": 0.08283086121082306, + "learning_rate": 9.959400675769668e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18860 + }, + { + "epoch": 5.490253127727669, + "flow_matching_loss": 0.014659265894442797, + "grad_norm": 0.08974171429872513, + "learning_rate": 9.943830573362106e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18870 + }, + { + "epoch": 5.493162641838813, + "flow_matching_loss": 0.014038912113755941, + "grad_norm": 0.07040951400995255, + "learning_rate": 9.928266614677168e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18880 + }, + { + "epoch": 5.496072155949957, + "flow_matching_loss": 0.014894255064427852, + "grad_norm": 0.07736583799123764, + "learning_rate": 9.912708818626524e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18890 + }, + { + "epoch": 5.4989816700611, + "flow_matching_loss": 0.014519116375595331, + "grad_norm": 0.09539593756198883, + "learning_rate": 9.897157204114369e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18900 + }, + { + "epoch": 5.501891184172243, + "flow_matching_loss": 0.013973113941028714, + "grad_norm": 0.06844270974397659, + "learning_rate": 9.881611790037375e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18910 + }, + { + "epoch": 5.504800698283387, + "flow_matching_loss": 0.016346335411071777, + "grad_norm": 0.12365614622831345, + "learning_rate": 9.866072595284684e-06, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18920 + }, + { + "epoch": 5.50771021239453, + "flow_matching_loss": 0.014966575894504786, + "grad_norm": 0.08798620849847794, + "learning_rate": 9.850539638737894e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18930 + }, + { + "epoch": 5.5106197265056736, + "flow_matching_loss": 0.013043543975800275, + "grad_norm": 0.0708589255809784, + "learning_rate": 9.835012939271006e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18940 + }, + { + "epoch": 5.513529240616817, + "flow_matching_loss": 0.015452075470238923, + "grad_norm": 0.11233743280172348, + "learning_rate": 9.819492515750418e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18950 + }, + { + "epoch": 5.51643875472796, + "flow_matching_loss": 0.014804276265203952, + "grad_norm": 0.10568483918905258, + "learning_rate": 9.803978387034918e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18960 + }, + { + "epoch": 5.519348268839104, + "flow_matching_loss": 0.016668171714991332, + "grad_norm": 0.07416079193353653, + "learning_rate": 9.788470571975624e-06, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18970 + }, + { + "epoch": 5.5222577829502475, + "flow_matching_loss": 0.01654977323487401, + "grad_norm": 0.09609606117010117, + "learning_rate": 9.772969089416008e-06, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18980 + }, + { + "epoch": 5.52516729706139, + "flow_matching_loss": 0.01523995092138648, + "grad_norm": 0.15692247450351715, + "learning_rate": 9.757473958191822e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18990 + }, + { + "epoch": 5.528076811172534, + "flow_matching_loss": 0.015775987040251495, + "grad_norm": 0.08109759539365768, + "learning_rate": 9.741985197131118e-06, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19000 + }, + { + "epoch": 5.530986325283678, + "flow_matching_loss": 0.014428446348756552, + "grad_norm": 0.07566675543785095, + "learning_rate": 9.726502825054195e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19010 + }, + { + "epoch": 5.533895839394821, + "flow_matching_loss": 0.014066411182284355, + "grad_norm": 0.0918038859963417, + "learning_rate": 9.711026860773595e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19020 + }, + { + "epoch": 5.536805353505964, + "flow_matching_loss": 0.016933999489992857, + "grad_norm": 0.07168366760015488, + "learning_rate": 9.69555732309408e-06, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19030 + }, + { + "epoch": 5.539714867617108, + "flow_matching_loss": 0.017278622649610043, + "grad_norm": 0.12344551831483841, + "learning_rate": 9.68009423081259e-06, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19040 + }, + { + "epoch": 5.542624381728252, + "flow_matching_loss": 0.015545894112437963, + "grad_norm": 0.08579353243112564, + "learning_rate": 9.664637602718249e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19050 + }, + { + "epoch": 5.5455338958393945, + "flow_matching_loss": 0.01789169302210212, + "grad_norm": 0.10678859800100327, + "learning_rate": 9.649187457592313e-06, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19060 + }, + { + "epoch": 5.548443409950538, + "flow_matching_loss": 0.01630345806479454, + "grad_norm": 0.1156182587146759, + "learning_rate": 9.63374381420816e-06, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19070 + }, + { + "epoch": 5.551352924061682, + "flow_matching_loss": 0.016397836431860924, + "grad_norm": 0.0800369456410408, + "learning_rate": 9.618306691331279e-06, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19080 + }, + { + "epoch": 5.554262438172826, + "flow_matching_loss": 0.014292547572404146, + "grad_norm": 0.14236299693584442, + "learning_rate": 9.602876107719228e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19090 + }, + { + "epoch": 5.557171952283968, + "flow_matching_loss": 0.016286798845976592, + "grad_norm": 0.09245256334543228, + "learning_rate": 9.587452082121621e-06, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19100 + }, + { + "epoch": 5.560081466395112, + "flow_matching_loss": 0.01727361176162958, + "grad_norm": 0.10632941126823425, + "learning_rate": 9.57203463328011e-06, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19110 + }, + { + "epoch": 5.562990980506255, + "flow_matching_loss": 0.01588900377973914, + "grad_norm": 0.09513022005558014, + "learning_rate": 9.556623779928341e-06, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19120 + }, + { + "epoch": 5.565900494617399, + "flow_matching_loss": 0.016548850946128367, + "grad_norm": 0.09619451314210892, + "learning_rate": 9.541219540791949e-06, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19130 + }, + { + "epoch": 5.568810008728542, + "flow_matching_loss": 0.015874387044459582, + "grad_norm": 0.1273002028465271, + "learning_rate": 9.525821934588551e-06, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19140 + }, + { + "epoch": 5.571719522839686, + "flow_matching_loss": 0.016004559956490992, + "grad_norm": 0.09699638932943344, + "learning_rate": 9.510430980027684e-06, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19150 + }, + { + "epoch": 5.574629036950829, + "flow_matching_loss": 0.017177335266023875, + "grad_norm": 0.09560845792293549, + "learning_rate": 9.495046695810809e-06, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19160 + }, + { + "epoch": 5.577538551061973, + "flow_matching_loss": 0.012820242159068584, + "grad_norm": 0.08884304016828537, + "learning_rate": 9.47966910063128e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19170 + }, + { + "epoch": 5.580448065173116, + "flow_matching_loss": 0.016035723406821488, + "grad_norm": 0.07282299548387527, + "learning_rate": 9.464298213174325e-06, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19180 + }, + { + "epoch": 5.583357579284259, + "flow_matching_loss": 0.016168579272925854, + "grad_norm": 0.09352463483810425, + "learning_rate": 9.448934052117024e-06, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19190 + }, + { + "epoch": 5.586267093395403, + "flow_matching_loss": 0.01663033328950405, + "grad_norm": 0.08862287551164627, + "learning_rate": 9.433576636128284e-06, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19200 + }, + { + "epoch": 5.5891766075065465, + "flow_matching_loss": 0.013932885322719812, + "grad_norm": 0.08939788490533829, + "learning_rate": 9.41822598386881e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19210 + }, + { + "epoch": 5.59208612161769, + "flow_matching_loss": 0.016612326074391604, + "grad_norm": 0.11943159252405167, + "learning_rate": 9.402882113991094e-06, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19220 + }, + { + "epoch": 5.594995635728833, + "flow_matching_loss": 0.016370157618075608, + "grad_norm": 0.11206604540348053, + "learning_rate": 9.387545045139381e-06, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19230 + }, + { + "epoch": 5.597905149839977, + "flow_matching_loss": 0.015516960620880127, + "grad_norm": 0.08948666602373123, + "learning_rate": 9.372214795949655e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19240 + }, + { + "epoch": 5.6008146639511205, + "flow_matching_loss": 0.017340993136167528, + "grad_norm": 0.13310565054416656, + "learning_rate": 9.356891385049618e-06, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19250 + }, + { + "epoch": 5.603724178062263, + "flow_matching_loss": 0.014936304278671741, + "grad_norm": 0.10669916123151779, + "learning_rate": 9.341574831058664e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19260 + }, + { + "epoch": 5.606633692173407, + "flow_matching_loss": 0.013082978408783674, + "grad_norm": 0.10970792174339294, + "learning_rate": 9.326265152587844e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19270 + }, + { + "epoch": 5.609543206284551, + "flow_matching_loss": 0.018059417605400085, + "grad_norm": 0.16421793401241302, + "learning_rate": 9.310962368239861e-06, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19280 + }, + { + "epoch": 5.6124527203956935, + "flow_matching_loss": 0.014959539100527764, + "grad_norm": 0.09610509872436523, + "learning_rate": 9.295666496609036e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19290 + }, + { + "epoch": 5.615362234506837, + "flow_matching_loss": 0.01796846864745021, + "grad_norm": 0.08771530538797379, + "learning_rate": 9.280377556281306e-06, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19300 + }, + { + "epoch": 5.618271748617981, + "flow_matching_loss": 0.015293535776436329, + "grad_norm": 0.07302651554346085, + "learning_rate": 9.265095565834172e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19310 + }, + { + "epoch": 5.621181262729124, + "flow_matching_loss": 0.017110577318817377, + "grad_norm": 0.13181094825267792, + "learning_rate": 9.249820543836688e-06, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19320 + }, + { + "epoch": 5.6240907768402675, + "flow_matching_loss": 0.014460364542901515, + "grad_norm": 0.0871080681681633, + "learning_rate": 9.234552508849454e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19330 + }, + { + "epoch": 5.627000290951411, + "flow_matching_loss": 0.015226195380091668, + "grad_norm": 0.10906129330396652, + "learning_rate": 9.21929147942456e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19340 + }, + { + "epoch": 5.629909805062555, + "flow_matching_loss": 0.013012525252997875, + "grad_norm": 0.06704418361186981, + "learning_rate": 9.204037474105611e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19350 + }, + { + "epoch": 5.632819319173698, + "flow_matching_loss": 0.013972265738993883, + "grad_norm": 0.06453999131917953, + "learning_rate": 9.188790511427653e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19360 + }, + { + "epoch": 5.635728833284841, + "flow_matching_loss": 0.017831349465996026, + "grad_norm": 0.1309843212366104, + "learning_rate": 9.173550609917186e-06, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19370 + }, + { + "epoch": 5.638638347395985, + "flow_matching_loss": 0.014169430639594794, + "grad_norm": 0.09645996242761612, + "learning_rate": 9.158317788092124e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19380 + }, + { + "epoch": 5.641547861507128, + "flow_matching_loss": 0.01633834382519126, + "grad_norm": 0.20101642608642578, + "learning_rate": 9.143092064461783e-06, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19390 + }, + { + "epoch": 5.644457375618272, + "flow_matching_loss": 0.01655197562649846, + "grad_norm": 0.07899781316518784, + "learning_rate": 9.127873457526855e-06, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19400 + }, + { + "epoch": 5.647366889729415, + "flow_matching_loss": 0.014670562371611596, + "grad_norm": 0.09988106042146683, + "learning_rate": 9.112661985779382e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19410 + }, + { + "epoch": 5.650276403840559, + "flow_matching_loss": 0.015583471767604351, + "grad_norm": 0.09457389265298843, + "learning_rate": 9.09745766770274e-06, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19420 + }, + { + "epoch": 5.653185917951702, + "flow_matching_loss": 0.014050995279103518, + "grad_norm": 0.08110184967517853, + "learning_rate": 9.082260521771601e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19430 + }, + { + "epoch": 5.656095432062846, + "flow_matching_loss": 0.018048284109681845, + "grad_norm": 0.09030544012784958, + "learning_rate": 9.067070566451936e-06, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19440 + }, + { + "epoch": 5.659004946173989, + "flow_matching_loss": 0.011745583731681108, + "grad_norm": 0.17938661575317383, + "learning_rate": 9.051887820200966e-06, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19450 + }, + { + "epoch": 5.661914460285132, + "flow_matching_loss": 0.014530333131551743, + "grad_norm": 0.07666829228401184, + "learning_rate": 9.03671230146717e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19460 + }, + { + "epoch": 5.664823974396276, + "flow_matching_loss": 0.014555516373366117, + "grad_norm": 0.10931665450334549, + "learning_rate": 9.021544028690228e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19470 + }, + { + "epoch": 5.6677334885074195, + "flow_matching_loss": 0.0138834229670465, + "grad_norm": 0.07684733718633652, + "learning_rate": 9.006383020301024e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19480 + }, + { + "epoch": 5.670643002618562, + "flow_matching_loss": 0.0142855285666883, + "grad_norm": 0.07090671360492706, + "learning_rate": 8.991229294721613e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19490 + }, + { + "epoch": 5.673552516729706, + "flow_matching_loss": 0.014164840057492255, + "grad_norm": 0.07812577486038208, + "learning_rate": 8.976082870365191e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19500 + }, + { + "epoch": 5.67646203084085, + "flow_matching_loss": 0.018126088567078112, + "grad_norm": 0.08453090488910675, + "learning_rate": 8.960943765636105e-06, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19510 + }, + { + "epoch": 5.679371544951993, + "flow_matching_loss": 0.015097272768616676, + "grad_norm": 0.10491485148668289, + "learning_rate": 8.945811998929792e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19520 + }, + { + "epoch": 5.682281059063136, + "flow_matching_loss": 0.017781312484294175, + "grad_norm": 0.09539864957332611, + "learning_rate": 8.93068758863277e-06, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19530 + }, + { + "epoch": 5.68519057317428, + "flow_matching_loss": 0.014432825613766909, + "grad_norm": 0.09082391858100891, + "learning_rate": 8.915570553122628e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19540 + }, + { + "epoch": 5.688100087285424, + "flow_matching_loss": 0.015715800132602452, + "grad_norm": 0.09584973007440567, + "learning_rate": 8.900460910767982e-06, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19550 + }, + { + "epoch": 5.6910096013965665, + "flow_matching_loss": 0.014880405180156231, + "grad_norm": 0.07994678616523743, + "learning_rate": 8.885358679928486e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19560 + }, + { + "epoch": 5.69391911550771, + "flow_matching_loss": 0.01607328038662672, + "grad_norm": 0.09963509440422058, + "learning_rate": 8.870263878954767e-06, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19570 + }, + { + "epoch": 5.696828629618854, + "flow_matching_loss": 0.01510404348373413, + "grad_norm": 0.11040499806404114, + "learning_rate": 8.855176526188428e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19580 + }, + { + "epoch": 5.699738143729997, + "flow_matching_loss": 0.01296973805874586, + "grad_norm": 0.08333750814199448, + "learning_rate": 8.840096639962033e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19590 + }, + { + "epoch": 5.7026476578411405, + "flow_matching_loss": 0.014538337290287019, + "grad_norm": 0.1020808294415474, + "learning_rate": 8.825024238599056e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19600 + }, + { + "epoch": 5.705557171952284, + "flow_matching_loss": 0.014532447047531604, + "grad_norm": 0.08364713191986084, + "learning_rate": 8.809959340413891e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19610 + }, + { + "epoch": 5.708466686063428, + "flow_matching_loss": 0.014676829986274242, + "grad_norm": 0.07449062168598175, + "learning_rate": 8.79490196371181e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19620 + }, + { + "epoch": 5.711376200174571, + "flow_matching_loss": 0.015442149713635445, + "grad_norm": 0.0826849415898323, + "learning_rate": 8.779852126788949e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19630 + }, + { + "epoch": 5.714285714285714, + "flow_matching_loss": 0.013652070797979832, + "grad_norm": 0.0702415332198143, + "learning_rate": 8.764809847932277e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19640 + }, + { + "epoch": 5.717195228396857, + "flow_matching_loss": 0.015134587977081537, + "grad_norm": 0.10152850300073624, + "learning_rate": 8.749775145419577e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19650 + }, + { + "epoch": 5.720104742508001, + "flow_matching_loss": 0.017437273543328045, + "grad_norm": 0.0928562730550766, + "learning_rate": 8.734748037519427e-06, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19660 + }, + { + "epoch": 5.723014256619145, + "flow_matching_loss": 0.017643696162849665, + "grad_norm": 0.09676487743854523, + "learning_rate": 8.719728542491195e-06, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19670 + }, + { + "epoch": 5.725923770730288, + "flow_matching_loss": 0.01406277883797884, + "grad_norm": 0.0852268785238266, + "learning_rate": 8.704716678584976e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19680 + }, + { + "epoch": 5.728833284841431, + "flow_matching_loss": 0.014812090899795294, + "grad_norm": 0.10161467641592026, + "learning_rate": 8.689712464041596e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19690 + }, + { + "epoch": 5.731742798952575, + "flow_matching_loss": 0.014756974577903748, + "grad_norm": 0.09579680860042572, + "learning_rate": 8.674715917092592e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19700 + }, + { + "epoch": 5.734652313063719, + "flow_matching_loss": 0.013033432699739934, + "grad_norm": 0.07381702214479446, + "learning_rate": 8.65972705596018e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19710 + }, + { + "epoch": 5.737561827174861, + "flow_matching_loss": 0.014792820159345865, + "grad_norm": 0.07879967987537384, + "learning_rate": 8.64474589885725e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19720 + }, + { + "epoch": 5.740471341286005, + "flow_matching_loss": 0.014963204693049192, + "grad_norm": 0.08320292085409164, + "learning_rate": 8.629772463987315e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19730 + }, + { + "epoch": 5.743380855397149, + "flow_matching_loss": 0.014947652257978917, + "grad_norm": 0.11602301150560379, + "learning_rate": 8.614806769544507e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19740 + }, + { + "epoch": 5.7462903695082925, + "flow_matching_loss": 0.014623335376381875, + "grad_norm": 0.07268206030130386, + "learning_rate": 8.59984883371356e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19750 + }, + { + "epoch": 5.749199883619435, + "flow_matching_loss": 0.017156309448182583, + "grad_norm": 0.08862058073282242, + "learning_rate": 8.584898674669772e-06, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19760 + }, + { + "epoch": 5.752109397730579, + "flow_matching_loss": 0.01418432341888547, + "grad_norm": 0.09330777078866959, + "learning_rate": 8.569956310579001e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19770 + }, + { + "epoch": 5.755018911841723, + "flow_matching_loss": 0.012988925166428088, + "grad_norm": 0.0849771648645401, + "learning_rate": 8.555021759597627e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19780 + }, + { + "epoch": 5.757928425952866, + "flow_matching_loss": 0.01610731240361929, + "grad_norm": 0.07338185608386993, + "learning_rate": 8.54009503987253e-06, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19790 + }, + { + "epoch": 5.760837940064009, + "flow_matching_loss": 0.016141128074377776, + "grad_norm": 0.062320828437805176, + "learning_rate": 8.525176169541095e-06, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19800 + }, + { + "epoch": 5.763747454175153, + "flow_matching_loss": 0.014695723261684179, + "grad_norm": 0.07618138194084167, + "learning_rate": 8.510265166731151e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19810 + }, + { + "epoch": 5.766656968286296, + "flow_matching_loss": 0.01345332795754075, + "grad_norm": 0.0837467685341835, + "learning_rate": 8.495362049560964e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19820 + }, + { + "epoch": 5.7695664823974395, + "flow_matching_loss": 0.01623114524409175, + "grad_norm": 0.08757782727479935, + "learning_rate": 8.480466836139241e-06, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19830 + }, + { + "epoch": 5.772475996508583, + "flow_matching_loss": 0.015307913348078728, + "grad_norm": 0.10643594712018967, + "learning_rate": 8.465579544565066e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19840 + }, + { + "epoch": 5.775385510619726, + "flow_matching_loss": 0.014848119951784611, + "grad_norm": 0.07635879516601562, + "learning_rate": 8.450700192927897e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19850 + }, + { + "epoch": 5.77829502473087, + "flow_matching_loss": 0.016095220856368542, + "grad_norm": 0.1064440906047821, + "learning_rate": 8.435828799307551e-06, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19860 + }, + { + "epoch": 5.7812045388420135, + "flow_matching_loss": 0.015013992320746183, + "grad_norm": 0.08645132184028625, + "learning_rate": 8.420965381774172e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19870 + }, + { + "epoch": 5.784114052953157, + "flow_matching_loss": 0.015638354886323213, + "grad_norm": 0.1081518605351448, + "learning_rate": 8.406109958388222e-06, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19880 + }, + { + "epoch": 5.7870235670643, + "flow_matching_loss": 0.014562873169779778, + "grad_norm": 0.09967754036188126, + "learning_rate": 8.391262547200434e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19890 + }, + { + "epoch": 5.789933081175444, + "flow_matching_loss": 0.013047370687127113, + "grad_norm": 0.08170332759618759, + "learning_rate": 8.376423166251816e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19900 + }, + { + "epoch": 5.792842595286587, + "flow_matching_loss": 0.017600651644170285, + "grad_norm": 0.09942895919084549, + "learning_rate": 8.361591833573612e-06, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19910 + }, + { + "epoch": 5.79575210939773, + "flow_matching_loss": 0.014997601602226495, + "grad_norm": 0.09377032518386841, + "learning_rate": 8.346768567187287e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19920 + }, + { + "epoch": 5.798661623508874, + "flow_matching_loss": 0.01946730250492692, + "grad_norm": 0.11969394236803055, + "learning_rate": 8.331953385104516e-06, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19930 + }, + { + "epoch": 5.801571137620018, + "flow_matching_loss": 0.014407995715737342, + "grad_norm": 0.09347108006477356, + "learning_rate": 8.317146305327134e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19940 + }, + { + "epoch": 5.804480651731161, + "flow_matching_loss": 0.01545794690027833, + "grad_norm": 0.08432816714048386, + "learning_rate": 8.302347345847146e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19950 + }, + { + "epoch": 5.807390165842304, + "flow_matching_loss": 0.018422001134604217, + "grad_norm": 0.12776117026805878, + "learning_rate": 8.287556524646683e-06, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19960 + }, + { + "epoch": 5.810299679953448, + "flow_matching_loss": 0.013104374101385475, + "grad_norm": 0.08408716320991516, + "learning_rate": 8.272773859697978e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19970 + }, + { + "epoch": 5.813209194064592, + "flow_matching_loss": 0.014666048903018236, + "grad_norm": 0.08823202550411224, + "learning_rate": 8.257999368963374e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19980 + }, + { + "epoch": 5.816118708175734, + "flow_matching_loss": 0.013088092301040887, + "grad_norm": 0.06563981622457504, + "learning_rate": 8.243233070395266e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19990 + }, + { + "epoch": 5.819028222286878, + "flow_matching_loss": 0.015100381802767515, + "grad_norm": 0.09071839600801468, + "learning_rate": 8.2284749819361e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20000 + }, + { + "epoch": 5.821937736398022, + "flow_matching_loss": 0.012694002781063318, + "grad_norm": 0.07158033549785614, + "learning_rate": 8.213725121518346e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20010 + }, + { + "epoch": 5.824847250509165, + "flow_matching_loss": 0.014742857497185468, + "grad_norm": 0.09939808398485184, + "learning_rate": 8.198983507064469e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20020 + }, + { + "epoch": 5.827756764620308, + "flow_matching_loss": 0.01608961643651128, + "grad_norm": 0.07647663354873657, + "learning_rate": 8.18425015648692e-06, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20030 + }, + { + "epoch": 5.830666278731452, + "flow_matching_loss": 0.015369795076549054, + "grad_norm": 0.07909273356199265, + "learning_rate": 8.169525087688122e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20040 + }, + { + "epoch": 5.833575792842595, + "flow_matching_loss": 0.012291476503014565, + "grad_norm": 0.07584989070892334, + "learning_rate": 8.154808318560413e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20050 + }, + { + "epoch": 5.836485306953739, + "flow_matching_loss": 0.013238877290859818, + "grad_norm": 0.07822920382022858, + "learning_rate": 8.140099866986058e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20060 + }, + { + "epoch": 5.839394821064882, + "flow_matching_loss": 0.015093533601611853, + "grad_norm": 0.12121448665857315, + "learning_rate": 8.125399750837202e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20070 + }, + { + "epoch": 5.842304335176026, + "flow_matching_loss": 0.015465649031102657, + "grad_norm": 0.10718119144439697, + "learning_rate": 8.110707987975888e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20080 + }, + { + "epoch": 5.845213849287169, + "flow_matching_loss": 0.013204418122768402, + "grad_norm": 0.08539771288633347, + "learning_rate": 8.096024596253982e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20090 + }, + { + "epoch": 5.8481233633983125, + "flow_matching_loss": 0.013805221300572157, + "grad_norm": 0.08248013257980347, + "learning_rate": 8.081349593513197e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20100 + }, + { + "epoch": 5.851032877509456, + "flow_matching_loss": 0.015829539205878974, + "grad_norm": 0.12917262315750122, + "learning_rate": 8.066682997585045e-06, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20110 + }, + { + "epoch": 5.853942391620599, + "flow_matching_loss": 0.014957948494702578, + "grad_norm": 0.09622350335121155, + "learning_rate": 8.052024826290819e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20120 + }, + { + "epoch": 5.856851905731743, + "flow_matching_loss": 0.014861858077347278, + "grad_norm": 0.09007716178894043, + "learning_rate": 8.037375097441574e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20130 + }, + { + "epoch": 5.8597614198428865, + "flow_matching_loss": 0.014314686879515648, + "grad_norm": 0.10503590106964111, + "learning_rate": 8.022733828838124e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20140 + }, + { + "epoch": 5.86267093395403, + "flow_matching_loss": 0.014519699569791555, + "grad_norm": 0.06827547401189804, + "learning_rate": 8.008101038270988e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20150 + }, + { + "epoch": 5.865580448065173, + "flow_matching_loss": 0.016358167584985494, + "grad_norm": 0.11323074996471405, + "learning_rate": 7.99347674352038e-06, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20160 + }, + { + "epoch": 5.868489962176317, + "flow_matching_loss": 0.01620288034901023, + "grad_norm": 0.08518040180206299, + "learning_rate": 7.978860962356202e-06, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20170 + }, + { + "epoch": 5.8713994762874595, + "flow_matching_loss": 0.013887952547520399, + "grad_norm": 0.11200303584337234, + "learning_rate": 7.964253712537997e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20180 + }, + { + "epoch": 5.874308990398603, + "flow_matching_loss": 0.014788771606981754, + "grad_norm": 0.08147960901260376, + "learning_rate": 7.949655011814968e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20190 + }, + { + "epoch": 5.877218504509747, + "flow_matching_loss": 0.01364389406517148, + "grad_norm": 0.08265066891908646, + "learning_rate": 7.9350648779259e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20200 + }, + { + "epoch": 5.880128018620891, + "flow_matching_loss": 0.01796096684411168, + "grad_norm": 0.09025844931602478, + "learning_rate": 7.92048332859919e-06, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20210 + }, + { + "epoch": 5.8830375327320334, + "flow_matching_loss": 0.016896114312112333, + "grad_norm": 0.12025701254606247, + "learning_rate": 7.90591038155278e-06, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20220 + }, + { + "epoch": 5.885947046843177, + "flow_matching_loss": 0.016350913699716328, + "grad_norm": 0.0979309007525444, + "learning_rate": 7.891346054494194e-06, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20230 + }, + { + "epoch": 5.888856560954321, + "flow_matching_loss": 0.01571773560717702, + "grad_norm": 0.10279200226068497, + "learning_rate": 7.876790365120447e-06, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20240 + }, + { + "epoch": 5.891766075065464, + "flow_matching_loss": 0.01431666761636734, + "grad_norm": 0.09049319475889206, + "learning_rate": 7.862243331118088e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20250 + }, + { + "epoch": 5.894675589176607, + "flow_matching_loss": 0.015154402144253254, + "grad_norm": 0.09839759767055511, + "learning_rate": 7.847704970163127e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20260 + }, + { + "epoch": 5.897585103287751, + "flow_matching_loss": 0.014907071739435196, + "grad_norm": 0.07311265170574188, + "learning_rate": 7.833175299921046e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20270 + }, + { + "epoch": 5.900494617398895, + "flow_matching_loss": 0.014763335511088372, + "grad_norm": 0.08247099071741104, + "learning_rate": 7.81865433804676e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20280 + }, + { + "epoch": 5.903404131510038, + "flow_matching_loss": 0.01497803684324026, + "grad_norm": 0.08772189915180206, + "learning_rate": 7.80414210218461e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20290 + }, + { + "epoch": 5.906313645621181, + "flow_matching_loss": 0.016593714896589517, + "grad_norm": 0.07641516625881195, + "learning_rate": 7.789638609968332e-06, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20300 + }, + { + "epoch": 5.909223159732325, + "flow_matching_loss": 0.015432843286544085, + "grad_norm": 0.09379000216722488, + "learning_rate": 7.775143879021037e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20310 + }, + { + "epoch": 5.912132673843468, + "flow_matching_loss": 0.013020847178995609, + "grad_norm": 0.08472144603729248, + "learning_rate": 7.760657926955187e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20320 + }, + { + "epoch": 5.915042187954612, + "flow_matching_loss": 0.013715212047100068, + "grad_norm": 0.09002477675676346, + "learning_rate": 7.746180771372584e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20330 + }, + { + "epoch": 5.917951702065755, + "flow_matching_loss": 0.01515366518869996, + "grad_norm": 0.07133544236421585, + "learning_rate": 7.731712429864325e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20340 + }, + { + "epoch": 5.920861216176898, + "flow_matching_loss": 0.01549188168719411, + "grad_norm": 0.09162233024835587, + "learning_rate": 7.717252920010828e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20350 + }, + { + "epoch": 5.923770730288042, + "flow_matching_loss": 0.014827012456953526, + "grad_norm": 0.08685954660177231, + "learning_rate": 7.702802259381742e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20360 + }, + { + "epoch": 5.9266802443991855, + "flow_matching_loss": 0.015481170359998941, + "grad_norm": 0.0795360878109932, + "learning_rate": 7.688360465535999e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20370 + }, + { + "epoch": 5.929589758510328, + "flow_matching_loss": 0.018322527687996627, + "grad_norm": 0.09816224873065948, + "learning_rate": 7.673927556021734e-06, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20380 + }, + { + "epoch": 5.932499272621472, + "flow_matching_loss": 0.013261767290532588, + "grad_norm": 0.08828863501548767, + "learning_rate": 7.659503548376284e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20390 + }, + { + "epoch": 5.935408786732616, + "flow_matching_loss": 0.015770493447780608, + "grad_norm": 0.09939364343881607, + "learning_rate": 7.645088460126193e-06, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20400 + }, + { + "epoch": 5.9383183008437594, + "flow_matching_loss": 0.012876563798636198, + "grad_norm": 0.06197980046272278, + "learning_rate": 7.630682308787145e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20410 + }, + { + "epoch": 5.941227814954902, + "flow_matching_loss": 0.013671321142464877, + "grad_norm": 0.0972200259566307, + "learning_rate": 7.6162851118639754e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20420 + }, + { + "epoch": 5.944137329066046, + "flow_matching_loss": 0.012689969036728144, + "grad_norm": 0.10450563579797745, + "learning_rate": 7.601896886850632e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20430 + }, + { + "epoch": 5.94704684317719, + "flow_matching_loss": 0.015169254411011935, + "grad_norm": 0.08710049092769623, + "learning_rate": 7.587517651230168e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20440 + }, + { + "epoch": 5.9499563572883325, + "flow_matching_loss": 0.014753962401300669, + "grad_norm": 0.08427312970161438, + "learning_rate": 7.573147422474702e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20450 + }, + { + "epoch": 5.952865871399476, + "flow_matching_loss": 0.013172070309519768, + "grad_norm": 0.07502274215221405, + "learning_rate": 7.55878621804543e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20460 + }, + { + "epoch": 5.95577538551062, + "flow_matching_loss": 0.013973633758723736, + "grad_norm": 0.0839131772518158, + "learning_rate": 7.544434055392564e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20470 + }, + { + "epoch": 5.958684899621764, + "flow_matching_loss": 0.018288365565240383, + "grad_norm": 0.11403992772102356, + "learning_rate": 7.530090951955334e-06, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20480 + }, + { + "epoch": 5.961594413732906, + "flow_matching_loss": 0.014549379888921975, + "grad_norm": 0.10721032321453094, + "learning_rate": 7.515756925161959e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20490 + }, + { + "epoch": 5.96450392784405, + "flow_matching_loss": 0.012494714651256799, + "grad_norm": 0.07749555259943008, + "learning_rate": 7.501431992429643e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20500 + }, + { + "epoch": 5.967413441955194, + "flow_matching_loss": 0.01509668594226241, + "grad_norm": 0.11724540591239929, + "learning_rate": 7.487116171164519e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20510 + }, + { + "epoch": 5.970322956066337, + "flow_matching_loss": 0.012724138703197241, + "grad_norm": 0.07744082063436508, + "learning_rate": 7.472809478761672e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20520 + }, + { + "epoch": 5.97323247017748, + "flow_matching_loss": 0.014648662321269513, + "grad_norm": 0.08199375867843628, + "learning_rate": 7.4585119326050775e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20530 + }, + { + "epoch": 5.976141984288624, + "flow_matching_loss": 0.016558182984590532, + "grad_norm": 0.10800734907388687, + "learning_rate": 7.444223550067599e-06, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20540 + }, + { + "epoch": 5.979051498399767, + "flow_matching_loss": 0.014470775658264757, + "grad_norm": 0.08014080673456192, + "learning_rate": 7.429944348510964e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20550 + }, + { + "epoch": 5.981961012510911, + "flow_matching_loss": 0.013216481264680623, + "grad_norm": 0.08594901114702225, + "learning_rate": 7.415674345285763e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20560 + }, + { + "epoch": 5.984870526622054, + "flow_matching_loss": 0.016285757534205914, + "grad_norm": 0.156072735786438, + "learning_rate": 7.401413557731387e-06, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20570 + }, + { + "epoch": 5.987780040733197, + "flow_matching_loss": 0.015712118800729515, + "grad_norm": 0.09884929656982422, + "learning_rate": 7.387162003176037e-06, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20580 + }, + { + "epoch": 5.990689554844341, + "flow_matching_loss": 0.014028676133602858, + "grad_norm": 0.08929216116666794, + "learning_rate": 7.372919698936697e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20590 + }, + { + "epoch": 5.9935990689554846, + "flow_matching_loss": 0.015344497375190258, + "grad_norm": 0.07552594691514969, + "learning_rate": 7.358686662319104e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20600 + }, + { + "epoch": 5.996508583066628, + "flow_matching_loss": 0.015149612352252007, + "grad_norm": 0.09330376237630844, + "learning_rate": 7.344462910617752e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20610 + }, + { + "epoch": 5.999418097177771, + "flow_matching_loss": 0.012959240190684795, + "grad_norm": 0.09843109548091888, + "learning_rate": 7.330248461115832e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20620 + }, + { + "epoch": 6.002327611288915, + "flow_matching_loss": 0.01608596360310912, + "grad_norm": 0.09333177655935287, + "learning_rate": 7.316043331085242e-06, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20630 + }, + { + "epoch": 6.0052371254000585, + "flow_matching_loss": 0.014555901009589433, + "grad_norm": 0.07338928431272507, + "learning_rate": 7.30184753778655e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20640 + }, + { + "epoch": 6.008146639511201, + "flow_matching_loss": 0.013672070670872926, + "grad_norm": 0.09724044799804688, + "learning_rate": 7.2876610984689935e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20650 + }, + { + "epoch": 6.011056153622345, + "flow_matching_loss": 0.015052701532840728, + "grad_norm": 0.12020207941532135, + "learning_rate": 7.273484030370423e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20660 + }, + { + "epoch": 6.013965667733489, + "flow_matching_loss": 0.012365911062806845, + "grad_norm": 0.07322954386472702, + "learning_rate": 7.259316350717324e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20670 + }, + { + "epoch": 6.0168751818446315, + "flow_matching_loss": 0.016082635335624217, + "grad_norm": 0.08942261338233948, + "learning_rate": 7.245158076724757e-06, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20680 + }, + { + "epoch": 6.019784695955775, + "flow_matching_loss": 0.015228016208857298, + "grad_norm": 0.07150641828775406, + "learning_rate": 7.23100922559636e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20690 + }, + { + "epoch": 6.022694210066919, + "flow_matching_loss": 0.016735372692346574, + "grad_norm": 0.0959414541721344, + "learning_rate": 7.216869814524323e-06, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20700 + }, + { + "epoch": 6.025603724178063, + "flow_matching_loss": 0.014235061034560204, + "grad_norm": 0.10229992866516113, + "learning_rate": 7.202739860689357e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20710 + }, + { + "epoch": 6.0285132382892055, + "flow_matching_loss": 0.016678054910153152, + "grad_norm": 0.1032709851861, + "learning_rate": 7.188619381260698e-06, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20720 + }, + { + "epoch": 6.031422752400349, + "flow_matching_loss": 0.015377639699727297, + "grad_norm": 0.07680055499076843, + "learning_rate": 7.174508393396057e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20730 + }, + { + "epoch": 6.034332266511493, + "flow_matching_loss": 0.012772881565615535, + "grad_norm": 0.08102019131183624, + "learning_rate": 7.160406914241615e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20740 + }, + { + "epoch": 6.037241780622636, + "flow_matching_loss": 0.014425070956349373, + "grad_norm": 0.09131346642971039, + "learning_rate": 7.146314960931996e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20750 + }, + { + "epoch": 6.040151294733779, + "flow_matching_loss": 0.012584412097930908, + "grad_norm": 0.09542366862297058, + "learning_rate": 7.132232550590247e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20760 + }, + { + "epoch": 6.043060808844923, + "flow_matching_loss": 0.013930397341027856, + "grad_norm": 0.10717722773551941, + "learning_rate": 7.118159700327839e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20770 + }, + { + "epoch": 6.045970322956066, + "flow_matching_loss": 0.01458190158009529, + "grad_norm": 0.09973037987947464, + "learning_rate": 7.104096427244603e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20780 + }, + { + "epoch": 6.04887983706721, + "flow_matching_loss": 0.013323500193655491, + "grad_norm": 0.0662616416811943, + "learning_rate": 7.090042748428737e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20790 + }, + { + "epoch": 6.051789351178353, + "flow_matching_loss": 0.012565383268520237, + "grad_norm": 0.09182285517454147, + "learning_rate": 7.075998680956797e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20800 + }, + { + "epoch": 6.054698865289497, + "flow_matching_loss": 0.01584730139002204, + "grad_norm": 0.11162320524454117, + "learning_rate": 7.061964241893639e-06, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20810 + }, + { + "epoch": 6.05760837940064, + "flow_matching_loss": 0.013903214503079653, + "grad_norm": 0.07896742969751358, + "learning_rate": 7.047939448292442e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20820 + }, + { + "epoch": 6.060517893511784, + "flow_matching_loss": 0.01569086229428649, + "grad_norm": 0.1082025021314621, + "learning_rate": 7.033924317194644e-06, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20830 + }, + { + "epoch": 6.063427407622927, + "flow_matching_loss": 0.012992203701287508, + "grad_norm": 0.09758712351322174, + "learning_rate": 7.0199188656299524e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20840 + }, + { + "epoch": 6.06633692173407, + "flow_matching_loss": 0.014252115366980433, + "grad_norm": 0.07667660713195801, + "learning_rate": 7.005923110616313e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20850 + }, + { + "epoch": 6.069246435845214, + "flow_matching_loss": 0.015129730757325887, + "grad_norm": 0.06999492645263672, + "learning_rate": 6.9919370691598855e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20860 + }, + { + "epoch": 6.0721559499563575, + "flow_matching_loss": 0.015000399248674513, + "grad_norm": 0.07569798827171326, + "learning_rate": 6.977960758255024e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20870 + }, + { + "epoch": 6.0750654640675, + "flow_matching_loss": 0.01262803040444851, + "grad_norm": 0.08031195402145386, + "learning_rate": 6.963994194884276e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20880 + }, + { + "epoch": 6.077974978178644, + "flow_matching_loss": 0.015534536819905043, + "grad_norm": 0.08397657424211502, + "learning_rate": 6.950037396018328e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20890 + }, + { + "epoch": 6.080884492289788, + "flow_matching_loss": 0.015505019668489694, + "grad_norm": 0.08622857928276062, + "learning_rate": 6.9360903786160056e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20900 + }, + { + "epoch": 6.083794006400931, + "flow_matching_loss": 0.01589081920683384, + "grad_norm": 0.07250192016363144, + "learning_rate": 6.9221531596242485e-06, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20910 + }, + { + "epoch": 6.086703520512074, + "flow_matching_loss": 0.016091607883572577, + "grad_norm": 0.06725704669952393, + "learning_rate": 6.908225755978089e-06, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20920 + }, + { + "epoch": 6.089613034623218, + "flow_matching_loss": 0.01408340148627758, + "grad_norm": 0.0956074446439743, + "learning_rate": 6.8943081846006466e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20930 + }, + { + "epoch": 6.092522548734362, + "flow_matching_loss": 0.015640151128172873, + "grad_norm": 0.11014663428068161, + "learning_rate": 6.88040046240307e-06, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20940 + }, + { + "epoch": 6.0954320628455045, + "flow_matching_loss": 0.015504722576588392, + "grad_norm": 0.0751420184969902, + "learning_rate": 6.866502606284566e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20950 + }, + { + "epoch": 6.098341576956648, + "flow_matching_loss": 0.015865883603692055, + "grad_norm": 0.08682925254106522, + "learning_rate": 6.852614633132335e-06, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20960 + }, + { + "epoch": 6.101251091067792, + "flow_matching_loss": 0.015482798218727112, + "grad_norm": 0.08605664223432541, + "learning_rate": 6.8387365598215665e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20970 + }, + { + "epoch": 6.104160605178935, + "flow_matching_loss": 0.01567097371444106, + "grad_norm": 0.09475165605545044, + "learning_rate": 6.82486840321544e-06, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20980 + }, + { + "epoch": 6.1070701192900785, + "flow_matching_loss": 0.01452607922255993, + "grad_norm": 0.114878810942173, + "learning_rate": 6.811010180165069e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20990 + }, + { + "epoch": 6.109979633401222, + "flow_matching_loss": 0.013593205064535142, + "grad_norm": 0.0687895119190216, + "learning_rate": 6.797161907509499e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21000 + }, + { + "epoch": 6.112889147512365, + "flow_matching_loss": 0.017575876135379077, + "grad_norm": 0.09155207872390747, + "learning_rate": 6.783323602075688e-06, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21010 + }, + { + "epoch": 6.115798661623509, + "flow_matching_loss": 0.01633753962814808, + "grad_norm": 0.09353403002023697, + "learning_rate": 6.7694952806784784e-06, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21020 + }, + { + "epoch": 6.118708175734652, + "flow_matching_loss": 0.014141821209341288, + "grad_norm": 0.09187426418066025, + "learning_rate": 6.7556769601205944e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21030 + }, + { + "epoch": 6.121617689845796, + "flow_matching_loss": 0.01649776231497526, + "grad_norm": 0.10660210996866226, + "learning_rate": 6.741868657192594e-06, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21040 + }, + { + "epoch": 6.124527203956939, + "flow_matching_loss": 0.01537354914471507, + "grad_norm": 0.11110921204090118, + "learning_rate": 6.728070388672864e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21050 + }, + { + "epoch": 6.127436718068083, + "flow_matching_loss": 0.013735521258786321, + "grad_norm": 0.11367841809988022, + "learning_rate": 6.714282171327604e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21060 + }, + { + "epoch": 6.130346232179226, + "flow_matching_loss": 0.013082112837582826, + "grad_norm": 0.07216137647628784, + "learning_rate": 6.700504021910791e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21070 + }, + { + "epoch": 6.133255746290369, + "flow_matching_loss": 0.016537775564938784, + "grad_norm": 0.07912547141313553, + "learning_rate": 6.68673595716419e-06, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21080 + }, + { + "epoch": 6.136165260401513, + "flow_matching_loss": 0.014564961474388839, + "grad_norm": 0.1084476187825203, + "learning_rate": 6.6729779938172856e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21090 + }, + { + "epoch": 6.139074774512657, + "flow_matching_loss": 0.01524391332641244, + "grad_norm": 0.12367519736289978, + "learning_rate": 6.659230148587311e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21100 + }, + { + "epoch": 6.141984288623799, + "flow_matching_loss": 0.013810659945011138, + "grad_norm": 0.08542274683713913, + "learning_rate": 6.645492438179188e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21110 + }, + { + "epoch": 6.144893802734943, + "flow_matching_loss": 0.014550638617947698, + "grad_norm": 0.09230313450098038, + "learning_rate": 6.631764879285531e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21120 + }, + { + "epoch": 6.147803316846087, + "flow_matching_loss": 0.01577924070879817, + "grad_norm": 0.09892401844263077, + "learning_rate": 6.618047488586614e-06, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21130 + }, + { + "epoch": 6.1507128309572305, + "flow_matching_loss": 0.014371089730411768, + "grad_norm": 0.11355171352624893, + "learning_rate": 6.604340282750371e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21140 + }, + { + "epoch": 6.153622345068373, + "flow_matching_loss": 0.014261819329112769, + "grad_norm": 0.07833594083786011, + "learning_rate": 6.590643278432342e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21150 + }, + { + "epoch": 6.156531859179517, + "flow_matching_loss": 0.01251658508554101, + "grad_norm": 0.07248411327600479, + "learning_rate": 6.576956492275682e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21160 + }, + { + "epoch": 6.159441373290661, + "flow_matching_loss": 0.0152420312166214, + "grad_norm": 0.09716183692216873, + "learning_rate": 6.563279940911124e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21170 + }, + { + "epoch": 6.162350887401804, + "flow_matching_loss": 0.014127713069319725, + "grad_norm": 0.14172044396400452, + "learning_rate": 6.549613640956963e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21180 + }, + { + "epoch": 6.165260401512947, + "flow_matching_loss": 0.016210054513067007, + "grad_norm": 0.10248475521802902, + "learning_rate": 6.535957609019053e-06, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21190 + }, + { + "epoch": 6.168169915624091, + "flow_matching_loss": 0.013248946610838174, + "grad_norm": 0.10477298498153687, + "learning_rate": 6.5223118616907545e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21200 + }, + { + "epoch": 6.171079429735234, + "flow_matching_loss": 0.0157174919731915, + "grad_norm": 0.0667007565498352, + "learning_rate": 6.508676415552939e-06, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21210 + }, + { + "epoch": 6.1739889438463775, + "flow_matching_loss": 0.013064334820955991, + "grad_norm": 0.11479857563972473, + "learning_rate": 6.495051287173952e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21220 + }, + { + "epoch": 6.176898457957521, + "flow_matching_loss": 0.015511668287217617, + "grad_norm": 0.08994153141975403, + "learning_rate": 6.481436493109621e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21230 + }, + { + "epoch": 6.179807972068665, + "flow_matching_loss": 0.015633545722812413, + "grad_norm": 0.10834354162216187, + "learning_rate": 6.467832049903194e-06, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21240 + }, + { + "epoch": 6.182717486179808, + "flow_matching_loss": 0.012771842442452908, + "grad_norm": 0.08341916650533676, + "learning_rate": 6.454237974085361e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21250 + }, + { + "epoch": 6.1856270002909515, + "flow_matching_loss": 0.012797381915152072, + "grad_norm": 0.09186943620443344, + "learning_rate": 6.440654282174203e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21260 + }, + { + "epoch": 6.188536514402095, + "flow_matching_loss": 0.014382623881101609, + "grad_norm": 0.09477914124727249, + "learning_rate": 6.427080990675187e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21270 + }, + { + "epoch": 6.191446028513238, + "flow_matching_loss": 0.013488629646599293, + "grad_norm": 0.09829434752464294, + "learning_rate": 6.41351811608114e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21280 + }, + { + "epoch": 6.194355542624382, + "flow_matching_loss": 0.014386853575706482, + "grad_norm": 0.07434675097465515, + "learning_rate": 6.399965674872228e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21290 + }, + { + "epoch": 6.197265056735525, + "flow_matching_loss": 0.014695226168259978, + "grad_norm": 0.07444892078638077, + "learning_rate": 6.386423683515957e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21300 + }, + { + "epoch": 6.200174570846668, + "flow_matching_loss": 0.016634052991867064, + "grad_norm": 0.0731363371014595, + "learning_rate": 6.372892158467119e-06, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21310 + }, + { + "epoch": 6.203084084957812, + "flow_matching_loss": 0.012946562189608813, + "grad_norm": 0.07544246315956116, + "learning_rate": 6.359371116167791e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21320 + }, + { + "epoch": 6.205993599068956, + "flow_matching_loss": 0.013499987591058017, + "grad_norm": 0.08449385315179825, + "learning_rate": 6.345860573047314e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21330 + }, + { + "epoch": 6.208903113180099, + "flow_matching_loss": 0.014751081727445126, + "grad_norm": 0.09153780341148376, + "learning_rate": 6.332360545522268e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21340 + }, + { + "epoch": 6.211812627291242, + "flow_matching_loss": 0.017367703281342982, + "grad_norm": 0.1092320904135704, + "learning_rate": 6.31887104999647e-06, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21350 + }, + { + "epoch": 6.214722141402386, + "flow_matching_loss": 0.015254906564950942, + "grad_norm": 0.0889820083975792, + "learning_rate": 6.305392102860924e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21360 + }, + { + "epoch": 6.21763165551353, + "flow_matching_loss": 0.013685392122715712, + "grad_norm": 0.09787064790725708, + "learning_rate": 6.291923720493814e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21370 + }, + { + "epoch": 6.220541169624672, + "flow_matching_loss": 0.013627137802541257, + "grad_norm": 0.0879652202129364, + "learning_rate": 6.278465919260507e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21380 + }, + { + "epoch": 6.223450683735816, + "flow_matching_loss": 0.01308881603181362, + "grad_norm": 0.09709520637989044, + "learning_rate": 6.26501871551349e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21390 + }, + { + "epoch": 6.22636019784696, + "flow_matching_loss": 0.01529724495485425, + "grad_norm": 0.08594167232513428, + "learning_rate": 6.251582125592394e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21400 + }, + { + "epoch": 6.229269711958103, + "flow_matching_loss": 0.01437316443771124, + "grad_norm": 0.08828581124544144, + "learning_rate": 6.2381561658239386e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21410 + }, + { + "epoch": 6.232179226069246, + "flow_matching_loss": 0.014651543088257312, + "grad_norm": 0.08017995953559875, + "learning_rate": 6.2247408525219276e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21420 + }, + { + "epoch": 6.23508874018039, + "flow_matching_loss": 0.014934381283819676, + "grad_norm": 0.11915018409490585, + "learning_rate": 6.2113362019872335e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21430 + }, + { + "epoch": 6.237998254291533, + "flow_matching_loss": 0.014622548967599869, + "grad_norm": 0.0965372622013092, + "learning_rate": 6.197942230507772e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21440 + }, + { + "epoch": 6.240907768402677, + "flow_matching_loss": 0.014003767911344767, + "grad_norm": 0.08255574107170105, + "learning_rate": 6.184558954358475e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21450 + }, + { + "epoch": 6.24381728251382, + "flow_matching_loss": 0.014733434747904539, + "grad_norm": 0.10898052901029587, + "learning_rate": 6.1711863898012965e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21460 + }, + { + "epoch": 6.246726796624964, + "flow_matching_loss": 0.016282930690795182, + "grad_norm": 0.07867442816495895, + "learning_rate": 6.157824553085155e-06, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21470 + }, + { + "epoch": 6.249636310736107, + "flow_matching_loss": 0.01486074784770608, + "grad_norm": 0.08837274461984634, + "learning_rate": 6.144473460445945e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21480 + }, + { + "epoch": 6.2525458248472505, + "flow_matching_loss": 0.01539599196985364, + "grad_norm": 0.09146788716316223, + "learning_rate": 6.131133128106503e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21490 + }, + { + "epoch": 6.255455338958394, + "flow_matching_loss": 0.013307681120932103, + "grad_norm": 0.07570185512304306, + "learning_rate": 6.117803572276584e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21500 + }, + { + "epoch": 6.258364853069537, + "flow_matching_loss": 0.016201992612332107, + "grad_norm": 0.12349525839090347, + "learning_rate": 6.104484809152866e-06, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21510 + }, + { + "epoch": 6.261274367180681, + "flow_matching_loss": 0.015515747293829919, + "grad_norm": 0.13466811180114746, + "learning_rate": 6.091176854918892e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21520 + }, + { + "epoch": 6.2641838812918245, + "flow_matching_loss": 0.013700751587748528, + "grad_norm": 0.1418861746788025, + "learning_rate": 6.077879725745092e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21530 + }, + { + "epoch": 6.267093395402968, + "flow_matching_loss": 0.014083158597350121, + "grad_norm": 0.0844516009092331, + "learning_rate": 6.064593437788727e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21540 + }, + { + "epoch": 6.270002909514111, + "flow_matching_loss": 0.015890226140618323, + "grad_norm": 0.07763144373893738, + "learning_rate": 6.051318007193881e-06, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21550 + }, + { + "epoch": 6.272912423625255, + "flow_matching_loss": 0.01400866275653243, + "grad_norm": 0.08481937646865845, + "learning_rate": 6.038053450091469e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21560 + }, + { + "epoch": 6.275821937736398, + "flow_matching_loss": 0.01435112627223134, + "grad_norm": 0.0991041287779808, + "learning_rate": 6.0247997825991705e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21570 + }, + { + "epoch": 6.278731451847541, + "flow_matching_loss": 0.014716031029820442, + "grad_norm": 0.08979982137680054, + "learning_rate": 6.011557020821444e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21580 + }, + { + "epoch": 6.281640965958685, + "flow_matching_loss": 0.013747359067201615, + "grad_norm": 0.07798193395137787, + "learning_rate": 5.99832518084949e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21590 + }, + { + "epoch": 6.284550480069829, + "flow_matching_loss": 0.013441804144531489, + "grad_norm": 0.08103671669960022, + "learning_rate": 5.98510427876124e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21600 + }, + { + "epoch": 6.2874599941809715, + "flow_matching_loss": 0.015865613892674445, + "grad_norm": 0.10015171766281128, + "learning_rate": 5.971894330621344e-06, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21610 + }, + { + "epoch": 6.290369508292115, + "flow_matching_loss": 0.013971061073243618, + "grad_norm": 0.08132471144199371, + "learning_rate": 5.958695352481134e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21620 + }, + { + "epoch": 6.293279022403259, + "flow_matching_loss": 0.01373159484937787, + "grad_norm": 0.09168661385774612, + "learning_rate": 5.945507360378608e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21630 + }, + { + "epoch": 6.296188536514402, + "flow_matching_loss": 0.01396366637200117, + "grad_norm": 0.08040449023246765, + "learning_rate": 5.9323303703384225e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21640 + }, + { + "epoch": 6.299098050625545, + "flow_matching_loss": 0.014924115128815174, + "grad_norm": 0.07798832654953003, + "learning_rate": 5.919164398371857e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21650 + }, + { + "epoch": 6.302007564736689, + "flow_matching_loss": 0.012947468459606171, + "grad_norm": 0.09433355182409286, + "learning_rate": 5.906009460476818e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21660 + }, + { + "epoch": 6.304917078847833, + "flow_matching_loss": 0.015347700845450164, + "grad_norm": 0.09348020702600479, + "learning_rate": 5.892865572637789e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21670 + }, + { + "epoch": 6.307826592958976, + "flow_matching_loss": 0.015381815563887357, + "grad_norm": 0.14881053566932678, + "learning_rate": 5.8797327508258396e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21680 + }, + { + "epoch": 6.310736107070119, + "flow_matching_loss": 0.012381136417388916, + "grad_norm": 0.09428149461746216, + "learning_rate": 5.866611010998584e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21690 + }, + { + "epoch": 6.313645621181263, + "flow_matching_loss": 0.01584968464449048, + "grad_norm": 0.07433199137449265, + "learning_rate": 5.853500369100172e-06, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21700 + }, + { + "epoch": 6.316555135292406, + "flow_matching_loss": 0.014808907732367515, + "grad_norm": 0.11005672067403793, + "learning_rate": 5.840400841061265e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21710 + }, + { + "epoch": 6.31946464940355, + "flow_matching_loss": 0.015020939148962497, + "grad_norm": 0.09106137603521347, + "learning_rate": 5.827312442799034e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21720 + }, + { + "epoch": 6.322374163514693, + "flow_matching_loss": 0.014704703632742167, + "grad_norm": 0.1287667155265808, + "learning_rate": 5.814235190217114e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21730 + }, + { + "epoch": 6.325283677625836, + "flow_matching_loss": 0.01314592882990837, + "grad_norm": 0.07293686270713806, + "learning_rate": 5.801169099205599e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21740 + }, + { + "epoch": 6.32819319173698, + "flow_matching_loss": 0.015304486360400915, + "grad_norm": 0.11144223809242249, + "learning_rate": 5.788114185641022e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21750 + }, + { + "epoch": 6.3311027058481235, + "flow_matching_loss": 0.014351527672261, + "grad_norm": 0.08712425082921982, + "learning_rate": 5.775070465386328e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21760 + }, + { + "epoch": 6.334012219959266, + "flow_matching_loss": 0.013979410659521819, + "grad_norm": 0.09488712251186371, + "learning_rate": 5.7620379542908765e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21770 + }, + { + "epoch": 6.33692173407041, + "flow_matching_loss": 0.014947292394936085, + "grad_norm": 0.08862662315368652, + "learning_rate": 5.749016668190395e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21780 + }, + { + "epoch": 6.339831248181554, + "flow_matching_loss": 0.01262530218809843, + "grad_norm": 0.06896983087062836, + "learning_rate": 5.736006622906973e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21790 + }, + { + "epoch": 6.3427407622926975, + "flow_matching_loss": 0.014934257697314024, + "grad_norm": 0.07584088295698166, + "learning_rate": 5.723007834249034e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21800 + }, + { + "epoch": 6.34565027640384, + "flow_matching_loss": 0.014640997909009457, + "grad_norm": 0.11755198985338211, + "learning_rate": 5.710020318011346e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21810 + }, + { + "epoch": 6.348559790514984, + "flow_matching_loss": 0.01474936492741108, + "grad_norm": 0.08969475328922272, + "learning_rate": 5.697044089974955e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21820 + }, + { + "epoch": 6.351469304626128, + "flow_matching_loss": 0.016009271517395972, + "grad_norm": 0.08218991756439209, + "learning_rate": 5.68407916590721e-06, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21830 + }, + { + "epoch": 6.3543788187372705, + "flow_matching_loss": 0.01422044513747096, + "grad_norm": 0.10357143729925156, + "learning_rate": 5.671125561561713e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21840 + }, + { + "epoch": 6.357288332848414, + "flow_matching_loss": 0.015723252110183238, + "grad_norm": 0.0849858745932579, + "learning_rate": 5.658183292678315e-06, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21850 + }, + { + "epoch": 6.360197846959558, + "flow_matching_loss": 0.015195124130696059, + "grad_norm": 0.08636932820081711, + "learning_rate": 5.645252374983091e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21860 + }, + { + "epoch": 6.363107361070702, + "flow_matching_loss": 0.015353558026254177, + "grad_norm": 0.09853178262710571, + "learning_rate": 5.6323328241883224e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21870 + }, + { + "epoch": 6.3660168751818444, + "flow_matching_loss": 0.015335786063224078, + "grad_norm": 0.0857633501291275, + "learning_rate": 5.61942465599249e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21880 + }, + { + "epoch": 6.368926389292988, + "flow_matching_loss": 0.0151323264464736, + "grad_norm": 0.11707190424203873, + "learning_rate": 5.60652788608023e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21890 + }, + { + "epoch": 6.371835903404132, + "flow_matching_loss": 0.013489203806966544, + "grad_norm": 0.06911199539899826, + "learning_rate": 5.593642530122336e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21900 + }, + { + "epoch": 6.374745417515275, + "flow_matching_loss": 0.014316807221621274, + "grad_norm": 0.13739722967147827, + "learning_rate": 5.58076860377573e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21910 + }, + { + "epoch": 6.377654931626418, + "flow_matching_loss": 0.014114025700837373, + "grad_norm": 0.10763666778802872, + "learning_rate": 5.567906122683439e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21920 + }, + { + "epoch": 6.380564445737562, + "flow_matching_loss": 0.016906302981078626, + "grad_norm": 0.1071816012263298, + "learning_rate": 5.555055102474604e-06, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21930 + }, + { + "epoch": 6.383473959848705, + "flow_matching_loss": 0.015219568461179733, + "grad_norm": 0.09414507448673248, + "learning_rate": 5.542215558764421e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21940 + }, + { + "epoch": 6.386383473959849, + "flow_matching_loss": 0.014404393173754216, + "grad_norm": 0.1029241606593132, + "learning_rate": 5.529387507154138e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21950 + }, + { + "epoch": 6.389292988070992, + "flow_matching_loss": 0.013327156752347946, + "grad_norm": 0.11015768349170685, + "learning_rate": 5.516570963231058e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21960 + }, + { + "epoch": 6.392202502182135, + "flow_matching_loss": 0.01454530395567417, + "grad_norm": 0.07902165502309799, + "learning_rate": 5.503765942568484e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21970 + }, + { + "epoch": 6.395112016293279, + "flow_matching_loss": 0.015206039510667324, + "grad_norm": 0.07797252386808395, + "learning_rate": 5.49097246072573e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21980 + }, + { + "epoch": 6.398021530404423, + "flow_matching_loss": 0.013977123145014047, + "grad_norm": 0.09059818089008331, + "learning_rate": 5.4781905332480805e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21990 + }, + { + "epoch": 6.400931044515566, + "flow_matching_loss": 0.014765760581940412, + "grad_norm": 0.10718679428100586, + "learning_rate": 5.4654201756667795e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22000 + }, + { + "epoch": 6.403840558626709, + "flow_matching_loss": 0.014425118640065192, + "grad_norm": 0.10121788829565048, + "learning_rate": 5.452661403499017e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22010 + }, + { + "epoch": 6.406750072737853, + "flow_matching_loss": 0.014177028462290764, + "grad_norm": 0.1216454803943634, + "learning_rate": 5.4399142322479005e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22020 + }, + { + "epoch": 6.4096595868489965, + "flow_matching_loss": 0.014428845886141062, + "grad_norm": 0.07582981139421463, + "learning_rate": 5.427178677402452e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22030 + }, + { + "epoch": 6.412569100960139, + "flow_matching_loss": 0.013865289371460676, + "grad_norm": 0.09993346780538559, + "learning_rate": 5.41445475443757e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22040 + }, + { + "epoch": 6.415478615071283, + "flow_matching_loss": 0.015131273958832025, + "grad_norm": 0.08500932157039642, + "learning_rate": 5.4017424788140194e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22050 + }, + { + "epoch": 6.418388129182427, + "flow_matching_loss": 0.011009389301761985, + "grad_norm": 0.07140389084815979, + "learning_rate": 5.389041865978414e-06, + "loss": 0.011, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22060 + }, + { + "epoch": 6.4212976432935704, + "flow_matching_loss": 0.014288040110841393, + "grad_norm": 0.08130437135696411, + "learning_rate": 5.37635293136319e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22070 + }, + { + "epoch": 6.424207157404713, + "flow_matching_loss": 0.014234142005443573, + "grad_norm": 0.0731760710477829, + "learning_rate": 5.363675690386609e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22080 + }, + { + "epoch": 6.427116671515857, + "flow_matching_loss": 0.01336234281770885, + "grad_norm": 0.11267790198326111, + "learning_rate": 5.351010158452708e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22090 + }, + { + "epoch": 6.430026185627001, + "flow_matching_loss": 0.013409920036792755, + "grad_norm": 0.06556492298841476, + "learning_rate": 5.338356350951308e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22100 + }, + { + "epoch": 6.4329356997381435, + "flow_matching_loss": 0.015179566200822591, + "grad_norm": 0.10534709692001343, + "learning_rate": 5.325714283257974e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22110 + }, + { + "epoch": 6.435845213849287, + "flow_matching_loss": 0.012335841218009591, + "grad_norm": 0.08375781774520874, + "learning_rate": 5.313083970734013e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22120 + }, + { + "epoch": 6.438754727960431, + "flow_matching_loss": 0.015870262496173383, + "grad_norm": 0.06890735030174255, + "learning_rate": 5.300465428726439e-06, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22130 + }, + { + "epoch": 6.441664242071574, + "flow_matching_loss": 0.01368562076240778, + "grad_norm": 0.1132785826921463, + "learning_rate": 5.287858672567979e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22140 + }, + { + "epoch": 6.444573756182717, + "flow_matching_loss": 0.01520066000521183, + "grad_norm": 0.09270496666431427, + "learning_rate": 5.275263717577028e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22150 + }, + { + "epoch": 6.447483270293861, + "flow_matching_loss": 0.017382794246077537, + "grad_norm": 0.09375622123479843, + "learning_rate": 5.2626805790576435e-06, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22160 + }, + { + "epoch": 6.450392784405004, + "flow_matching_loss": 0.014089121017605066, + "grad_norm": 0.12049616873264313, + "learning_rate": 5.250109272299523e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22170 + }, + { + "epoch": 6.453302298516148, + "flow_matching_loss": 0.014667649753391743, + "grad_norm": 0.08771178871393204, + "learning_rate": 5.237549812577987e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22180 + }, + { + "epoch": 6.456211812627291, + "flow_matching_loss": 0.013788379728794098, + "grad_norm": 0.07048629224300385, + "learning_rate": 5.225002215153973e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22190 + }, + { + "epoch": 6.459121326738435, + "flow_matching_loss": 0.013927537761628628, + "grad_norm": 0.09473507851362228, + "learning_rate": 5.21246649527399e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22200 + }, + { + "epoch": 6.462030840849578, + "flow_matching_loss": 0.01698842840269208, + "grad_norm": 0.08875071257352829, + "learning_rate": 5.199942668170118e-06, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22210 + }, + { + "epoch": 6.464940354960722, + "flow_matching_loss": 0.012876168824732304, + "grad_norm": 0.08389011025428772, + "learning_rate": 5.187430749059986e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22220 + }, + { + "epoch": 6.467849869071865, + "flow_matching_loss": 0.013478673994541168, + "grad_norm": 0.08263185620307922, + "learning_rate": 5.174930753146763e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22230 + }, + { + "epoch": 6.470759383183008, + "flow_matching_loss": 0.01280059153214097, + "grad_norm": 0.074292853474617, + "learning_rate": 5.162442695619114e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22240 + }, + { + "epoch": 6.473668897294152, + "flow_matching_loss": 0.014083294477313758, + "grad_norm": 0.09589681029319763, + "learning_rate": 5.149966591651215e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22250 + }, + { + "epoch": 6.4765784114052956, + "flow_matching_loss": 0.013614076003432275, + "grad_norm": 0.09338157624006271, + "learning_rate": 5.137502456402705e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22260 + }, + { + "epoch": 6.479487925516438, + "flow_matching_loss": 0.015143313817679882, + "grad_norm": 0.08858052641153336, + "learning_rate": 5.125050305018683e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22270 + }, + { + "epoch": 6.482397439627582, + "flow_matching_loss": 0.01643245220184326, + "grad_norm": 0.08281692117452621, + "learning_rate": 5.112610152629687e-06, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22280 + }, + { + "epoch": 6.485306953738726, + "flow_matching_loss": 0.013461715076118708, + "grad_norm": 0.09917773306369781, + "learning_rate": 5.10018201435167e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22290 + }, + { + "epoch": 6.488216467849869, + "flow_matching_loss": 0.016281958762556316, + "grad_norm": 0.1006360575556755, + "learning_rate": 5.087765905286004e-06, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22300 + }, + { + "epoch": 6.491125981961012, + "flow_matching_loss": 0.014684343989938498, + "grad_norm": 0.07552444934844971, + "learning_rate": 5.0753618405194245e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22310 + }, + { + "epoch": 6.494035496072156, + "flow_matching_loss": 0.014127829391509295, + "grad_norm": 0.10098833590745926, + "learning_rate": 5.062969835124042e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22320 + }, + { + "epoch": 6.4969450101833, + "flow_matching_loss": 0.015225479938089847, + "grad_norm": 0.06166338920593262, + "learning_rate": 5.05058990415731e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22330 + }, + { + "epoch": 6.4998545242944425, + "flow_matching_loss": 0.015307170897722244, + "grad_norm": 0.08563750237226486, + "learning_rate": 5.038222062662007e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22340 + }, + { + "epoch": 6.502764038405586, + "flow_matching_loss": 0.015494873281568289, + "grad_norm": 0.1221560537815094, + "learning_rate": 5.025866325666237e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22350 + }, + { + "epoch": 6.50567355251673, + "flow_matching_loss": 0.01236799405887723, + "grad_norm": 0.0888020321726799, + "learning_rate": 5.013522708183383e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22360 + }, + { + "epoch": 6.508583066627873, + "flow_matching_loss": 0.016216397937387227, + "grad_norm": 0.11905719339847565, + "learning_rate": 5.001191225212096e-06, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22370 + }, + { + "epoch": 6.5114925807390165, + "flow_matching_loss": 0.014225606340914965, + "grad_norm": 0.08967415243387222, + "learning_rate": 4.9888718917363e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22380 + }, + { + "epoch": 6.51440209485016, + "flow_matching_loss": 0.015078908391296863, + "grad_norm": 0.08093687146902084, + "learning_rate": 4.9765647227251395e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22390 + }, + { + "epoch": 6.517311608961304, + "flow_matching_loss": 0.014520673826336861, + "grad_norm": 0.09818815439939499, + "learning_rate": 4.964269733132992e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22400 + }, + { + "epoch": 6.520221123072447, + "flow_matching_loss": 0.015419813338667154, + "grad_norm": 0.10345466434955597, + "learning_rate": 4.951986937899428e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22410 + }, + { + "epoch": 6.52313063718359, + "flow_matching_loss": 0.015916680078953506, + "grad_norm": 0.1051754504442215, + "learning_rate": 4.9397163519492e-06, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22420 + }, + { + "epoch": 6.526040151294734, + "flow_matching_loss": 0.014178875368088484, + "grad_norm": 0.11941234767436981, + "learning_rate": 4.9274579901922275e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22430 + }, + { + "epoch": 6.528949665405877, + "flow_matching_loss": 0.015387834515422583, + "grad_norm": 0.06912901252508163, + "learning_rate": 4.9152118675235705e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22440 + }, + { + "epoch": 6.531859179517021, + "flow_matching_loss": 0.0156737495213747, + "grad_norm": 0.07152557373046875, + "learning_rate": 4.902977998823432e-06, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22450 + }, + { + "epoch": 6.534768693628164, + "flow_matching_loss": 0.01565415789373219, + "grad_norm": 0.0733279287815094, + "learning_rate": 4.890756398957114e-06, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22460 + }, + { + "epoch": 6.537678207739307, + "flow_matching_loss": 0.012859377078711987, + "grad_norm": 0.0904780775308609, + "learning_rate": 4.8785470827750075e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22470 + }, + { + "epoch": 6.540587721850451, + "flow_matching_loss": 0.013477918598800898, + "grad_norm": 0.0879134088754654, + "learning_rate": 4.866350065112588e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22480 + }, + { + "epoch": 6.543497235961595, + "flow_matching_loss": 0.012195089552551507, + "grad_norm": 0.09046684950590134, + "learning_rate": 4.854165360790381e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22490 + }, + { + "epoch": 6.546406750072737, + "flow_matching_loss": 0.013685567956417798, + "grad_norm": 0.09293665736913681, + "learning_rate": 4.8419929846139425e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22500 + }, + { + "epoch": 6.549316264183881, + "flow_matching_loss": 0.01573388483375311, + "grad_norm": 0.13131694495677948, + "learning_rate": 4.829832951373872e-06, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22510 + }, + { + "epoch": 6.552225778295025, + "flow_matching_loss": 0.015571770630776883, + "grad_norm": 0.09955277293920517, + "learning_rate": 4.817685275845746e-06, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22520 + }, + { + "epoch": 6.5551352924061685, + "flow_matching_loss": 0.012419113796204329, + "grad_norm": 0.07453470677137375, + "learning_rate": 4.805549972790143e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22530 + }, + { + "epoch": 6.558044806517311, + "flow_matching_loss": 0.015879227221012114, + "grad_norm": 0.11883842945098877, + "learning_rate": 4.7934270569525966e-06, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22540 + }, + { + "epoch": 6.560954320628455, + "flow_matching_loss": 0.014902953524142504, + "grad_norm": 0.07515626400709152, + "learning_rate": 4.7813165430635874e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22550 + }, + { + "epoch": 6.563863834739599, + "flow_matching_loss": 0.012829450983554124, + "grad_norm": 0.07510501146316528, + "learning_rate": 4.769218445838544e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22560 + }, + { + "epoch": 6.566773348850742, + "flow_matching_loss": 0.013986039068549871, + "grad_norm": 0.09028057008981705, + "learning_rate": 4.757132779977787e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22570 + }, + { + "epoch": 6.569682862961885, + "flow_matching_loss": 0.014982333406805992, + "grad_norm": 0.07966925203800201, + "learning_rate": 4.745059560166542e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22580 + }, + { + "epoch": 6.572592377073029, + "flow_matching_loss": 0.013937093503773212, + "grad_norm": 0.06684848666191101, + "learning_rate": 4.732998801074911e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22590 + }, + { + "epoch": 6.575501891184173, + "flow_matching_loss": 0.013745475187897682, + "grad_norm": 0.08520538359880447, + "learning_rate": 4.7209505173578445e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22600 + }, + { + "epoch": 6.5784114052953155, + "flow_matching_loss": 0.015956676937639712, + "grad_norm": 0.07257088273763657, + "learning_rate": 4.7089147236551555e-06, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22610 + }, + { + "epoch": 6.581320919406459, + "flow_matching_loss": 0.0164893988519907, + "grad_norm": 0.09605865925550461, + "learning_rate": 4.696891434591464e-06, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22620 + }, + { + "epoch": 6.584230433517602, + "flow_matching_loss": 0.015560597646981478, + "grad_norm": 0.0967395007610321, + "learning_rate": 4.684880664776199e-06, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22630 + }, + { + "epoch": 6.587139947628746, + "flow_matching_loss": 0.013300700765103101, + "grad_norm": 0.08985591679811478, + "learning_rate": 4.6728824288035784e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22640 + }, + { + "epoch": 6.5900494617398895, + "flow_matching_loss": 0.014948249980807304, + "grad_norm": 0.08191078901290894, + "learning_rate": 4.660896741252586e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22650 + }, + { + "epoch": 6.592958975851033, + "flow_matching_loss": 0.015381798474118113, + "grad_norm": 0.08795366436243057, + "learning_rate": 4.6489236166869714e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22660 + }, + { + "epoch": 6.595868489962176, + "flow_matching_loss": 0.016078782081604005, + "grad_norm": 0.0863547995686531, + "learning_rate": 4.6369630696552e-06, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22670 + }, + { + "epoch": 6.59877800407332, + "flow_matching_loss": 0.014249049965292216, + "grad_norm": 0.08548824489116669, + "learning_rate": 4.625015114690474e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22680 + }, + { + "epoch": 6.601687518184463, + "flow_matching_loss": 0.016250970773398876, + "grad_norm": 0.08084236085414886, + "learning_rate": 4.613079766310682e-06, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22690 + }, + { + "epoch": 6.604597032295606, + "flow_matching_loss": 0.0145519167650491, + "grad_norm": 0.08253823220729828, + "learning_rate": 4.601157039018396e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22700 + }, + { + "epoch": 6.60750654640675, + "flow_matching_loss": 0.016251626517623664, + "grad_norm": 0.09551717340946198, + "learning_rate": 4.589246947300849e-06, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22710 + }, + { + "epoch": 6.610416060517894, + "flow_matching_loss": 0.016011226829141377, + "grad_norm": 0.10668443143367767, + "learning_rate": 4.577349505629936e-06, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22720 + }, + { + "epoch": 6.613325574629037, + "flow_matching_loss": 0.013369128759950399, + "grad_norm": 0.08106508105993271, + "learning_rate": 4.565464728462165e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22730 + }, + { + "epoch": 6.61623508874018, + "flow_matching_loss": 0.015407718811184168, + "grad_norm": 0.06540941447019577, + "learning_rate": 4.553592630238663e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22740 + }, + { + "epoch": 6.619144602851324, + "flow_matching_loss": 0.014067528676241637, + "grad_norm": 0.1197061687707901, + "learning_rate": 4.541733225385146e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22750 + }, + { + "epoch": 6.622054116962468, + "flow_matching_loss": 0.014332662895321846, + "grad_norm": 0.08744443207979202, + "learning_rate": 4.529886528311905e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22760 + }, + { + "epoch": 6.62496363107361, + "flow_matching_loss": 0.014180146809667348, + "grad_norm": 0.08625650405883789, + "learning_rate": 4.5180525534138045e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22770 + }, + { + "epoch": 6.627873145184754, + "flow_matching_loss": 0.014734474010765553, + "grad_norm": 0.10375946760177612, + "learning_rate": 4.506231315070235e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22780 + }, + { + "epoch": 6.630782659295898, + "flow_matching_loss": 0.01507950210943818, + "grad_norm": 0.09765426814556122, + "learning_rate": 4.494422827645115e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22790 + }, + { + "epoch": 6.633692173407041, + "flow_matching_loss": 0.011922177206724883, + "grad_norm": 0.07131493836641312, + "learning_rate": 4.482627105486865e-06, + "loss": 0.0119, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22800 + }, + { + "epoch": 6.636601687518184, + "flow_matching_loss": 0.014463500399142504, + "grad_norm": 0.08295010030269623, + "learning_rate": 4.470844162928412e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22810 + }, + { + "epoch": 6.639511201629328, + "flow_matching_loss": 0.012564603658393025, + "grad_norm": 0.06271038949489594, + "learning_rate": 4.459074014287129e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22820 + }, + { + "epoch": 6.642420715740471, + "flow_matching_loss": 0.011439814185723662, + "grad_norm": 0.07281585782766342, + "learning_rate": 4.447316673864869e-06, + "loss": 0.0114, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22830 + }, + { + "epoch": 6.645330229851615, + "flow_matching_loss": 0.014419043343514204, + "grad_norm": 0.09395404160022736, + "learning_rate": 4.435572155947903e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22840 + }, + { + "epoch": 6.648239743962758, + "flow_matching_loss": 0.014042683597654104, + "grad_norm": 0.12108415365219116, + "learning_rate": 4.423840474806929e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22850 + }, + { + "epoch": 6.651149258073902, + "flow_matching_loss": 0.013793862238526345, + "grad_norm": 0.06737612187862396, + "learning_rate": 4.412121644697043e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22860 + }, + { + "epoch": 6.654058772185045, + "flow_matching_loss": 0.015471157990396023, + "grad_norm": 0.08385740220546722, + "learning_rate": 4.400415679857727e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22870 + }, + { + "epoch": 6.6569682862961885, + "flow_matching_loss": 0.015467752143740654, + "grad_norm": 0.06278073042631149, + "learning_rate": 4.388722594512838e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22880 + }, + { + "epoch": 6.659877800407332, + "flow_matching_loss": 0.01500282222405076, + "grad_norm": 0.09328419715166092, + "learning_rate": 4.377042402870577e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22890 + }, + { + "epoch": 6.662787314518475, + "flow_matching_loss": 0.013927439227700233, + "grad_norm": 0.05935673788189888, + "learning_rate": 4.365375119123476e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22900 + }, + { + "epoch": 6.665696828629619, + "flow_matching_loss": 0.013175918906927108, + "grad_norm": 0.056028813123703, + "learning_rate": 4.353720757448383e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22910 + }, + { + "epoch": 6.6686063427407625, + "flow_matching_loss": 0.015832109469920398, + "grad_norm": 0.10077142715454102, + "learning_rate": 4.342079332006445e-06, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22920 + }, + { + "epoch": 6.671515856851906, + "flow_matching_loss": 0.01516365958377719, + "grad_norm": 0.09578988701105118, + "learning_rate": 4.330450856943101e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22930 + }, + { + "epoch": 6.674425370963049, + "flow_matching_loss": 0.014323683083057403, + "grad_norm": 0.10486209392547607, + "learning_rate": 4.318835346388036e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22940 + }, + { + "epoch": 6.677334885074193, + "flow_matching_loss": 0.012820971431210637, + "grad_norm": 0.09260717034339905, + "learning_rate": 4.307232814455189e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22950 + }, + { + "epoch": 6.680244399185336, + "flow_matching_loss": 0.01313297445885837, + "grad_norm": 0.07006571441888809, + "learning_rate": 4.29564327524274e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22960 + }, + { + "epoch": 6.683153913296479, + "flow_matching_loss": 0.014760083146393299, + "grad_norm": 0.1131540834903717, + "learning_rate": 4.284066742833061e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22970 + }, + { + "epoch": 6.686063427407623, + "flow_matching_loss": 0.01549757132306695, + "grad_norm": 0.08046171069145203, + "learning_rate": 4.272503231292741e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22980 + }, + { + "epoch": 6.688972941518767, + "flow_matching_loss": 0.014695211872458458, + "grad_norm": 0.11836641281843185, + "learning_rate": 4.260952754672532e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22990 + }, + { + "epoch": 6.6918824556299095, + "flow_matching_loss": 0.013484826125204564, + "grad_norm": 0.08300895988941193, + "learning_rate": 4.249415327007353e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23000 + }, + { + "epoch": 6.694791969741053, + "flow_matching_loss": 0.013900620676577092, + "grad_norm": 0.09442862868309021, + "learning_rate": 4.237890962316264e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23010 + }, + { + "epoch": 6.697701483852197, + "flow_matching_loss": 0.013140367716550827, + "grad_norm": 0.07078865170478821, + "learning_rate": 4.226379674602452e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23020 + }, + { + "epoch": 6.70061099796334, + "flow_matching_loss": 0.015845660213381053, + "grad_norm": 0.10436124354600906, + "learning_rate": 4.214881477853227e-06, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23030 + }, + { + "epoch": 6.703520512074483, + "flow_matching_loss": 0.015077829733490944, + "grad_norm": 0.0742732435464859, + "learning_rate": 4.2033963860399745e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23040 + }, + { + "epoch": 6.706430026185627, + "flow_matching_loss": 0.016868235915899275, + "grad_norm": 0.10739132016897202, + "learning_rate": 4.1919244131181696e-06, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23050 + }, + { + "epoch": 6.709339540296771, + "flow_matching_loss": 0.013716255873441696, + "grad_norm": 0.09317190200090408, + "learning_rate": 4.180465573027335e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23060 + }, + { + "epoch": 6.712249054407914, + "flow_matching_loss": 0.014102874137461186, + "grad_norm": 0.08880341798067093, + "learning_rate": 4.169019879691046e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23070 + }, + { + "epoch": 6.715158568519057, + "flow_matching_loss": 0.014954393543303014, + "grad_norm": 0.07924759387969971, + "learning_rate": 4.157587347016891e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23080 + }, + { + "epoch": 6.718068082630201, + "flow_matching_loss": 0.013298056460916996, + "grad_norm": 0.07671986520290375, + "learning_rate": 4.14616798889649e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23090 + }, + { + "epoch": 6.720977596741344, + "flow_matching_loss": 0.015559778409078717, + "grad_norm": 0.10841615498065948, + "learning_rate": 4.134761819205425e-06, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23100 + }, + { + "epoch": 6.723887110852488, + "flow_matching_loss": 0.015305323339998723, + "grad_norm": 0.06480435281991959, + "learning_rate": 4.1233688518032794e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23110 + }, + { + "epoch": 6.726796624963631, + "flow_matching_loss": 0.014734790055081248, + "grad_norm": 0.08516449481248856, + "learning_rate": 4.1119891005335795e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23120 + }, + { + "epoch": 6.729706139074775, + "flow_matching_loss": 0.013907054718583823, + "grad_norm": 0.08598562330007553, + "learning_rate": 4.1006225792237895e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23130 + }, + { + "epoch": 6.732615653185918, + "flow_matching_loss": 0.012438493128865958, + "grad_norm": 0.07528328895568848, + "learning_rate": 4.089269301685314e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23140 + }, + { + "epoch": 6.7355251672970615, + "flow_matching_loss": 0.015916272439062594, + "grad_norm": 0.11116472631692886, + "learning_rate": 4.077929281713453e-06, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23150 + }, + { + "epoch": 6.738434681408204, + "flow_matching_loss": 0.015251775179058314, + "grad_norm": 0.11898446828126907, + "learning_rate": 4.066602533087398e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23160 + }, + { + "epoch": 6.741344195519348, + "flow_matching_loss": 0.012976460903882981, + "grad_norm": 0.09304897487163544, + "learning_rate": 4.0552890695702175e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23170 + }, + { + "epoch": 6.744253709630492, + "flow_matching_loss": 0.014536272175610066, + "grad_norm": 0.06986381858587265, + "learning_rate": 4.04398890490883e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23180 + }, + { + "epoch": 6.7471632237416355, + "flow_matching_loss": 0.015323119051754475, + "grad_norm": 0.10118892043828964, + "learning_rate": 4.032702052834013e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23190 + }, + { + "epoch": 6.750072737852778, + "flow_matching_loss": 0.013649956090375781, + "grad_norm": 0.07955934852361679, + "learning_rate": 4.021428527060352e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23200 + }, + { + "epoch": 6.752982251963922, + "flow_matching_loss": 0.012966891285032034, + "grad_norm": 0.08659923821687698, + "learning_rate": 4.010168341286241e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23210 + }, + { + "epoch": 6.755891766075066, + "flow_matching_loss": 0.014283372741192579, + "grad_norm": 0.07874425500631332, + "learning_rate": 3.998921509193868e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23220 + }, + { + "epoch": 6.7588012801862085, + "flow_matching_loss": 0.013423302117735148, + "grad_norm": 0.0925198644399643, + "learning_rate": 3.987688044449193e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23230 + }, + { + "epoch": 6.761710794297352, + "flow_matching_loss": 0.01579404026269913, + "grad_norm": 0.12715397775173187, + "learning_rate": 3.9764679607019385e-06, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23240 + }, + { + "epoch": 6.764620308408496, + "flow_matching_loss": 0.013759645400568843, + "grad_norm": 0.10955600440502167, + "learning_rate": 3.965261271585561e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23250 + }, + { + "epoch": 6.76752982251964, + "flow_matching_loss": 0.01669608009979129, + "grad_norm": 0.0740373283624649, + "learning_rate": 3.954067990717251e-06, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23260 + }, + { + "epoch": 6.7704393366307825, + "flow_matching_loss": 0.01269497787579894, + "grad_norm": 0.08233688026666641, + "learning_rate": 3.9428881316978935e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23270 + }, + { + "epoch": 6.773348850741926, + "flow_matching_loss": 0.014798053726553916, + "grad_norm": 0.10099361091852188, + "learning_rate": 3.931721708112075e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23280 + }, + { + "epoch": 6.77625836485307, + "flow_matching_loss": 0.013870008382946252, + "grad_norm": 0.08883556723594666, + "learning_rate": 3.920568733528047e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23290 + }, + { + "epoch": 6.779167878964213, + "flow_matching_loss": 0.013798294588923454, + "grad_norm": 0.06761809438467026, + "learning_rate": 3.909429221497734e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23300 + }, + { + "epoch": 6.782077393075356, + "flow_matching_loss": 0.01510107098147273, + "grad_norm": 0.08067288249731064, + "learning_rate": 3.898303185556689e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23310 + }, + { + "epoch": 6.7849869071865, + "flow_matching_loss": 0.012252869317308068, + "grad_norm": 0.07427465170621872, + "learning_rate": 3.887190639224097e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23320 + }, + { + "epoch": 6.787896421297643, + "flow_matching_loss": 0.013050753436982632, + "grad_norm": 0.06290048360824585, + "learning_rate": 3.876091596002746e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23330 + }, + { + "epoch": 6.790805935408787, + "flow_matching_loss": 0.016220013750717042, + "grad_norm": 0.12369818985462189, + "learning_rate": 3.865006069379019e-06, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23340 + }, + { + "epoch": 6.79371544951993, + "flow_matching_loss": 0.01293764212168753, + "grad_norm": 0.0815839171409607, + "learning_rate": 3.853934072822881e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23350 + }, + { + "epoch": 6.796624963631073, + "flow_matching_loss": 0.013989817257970572, + "grad_norm": 0.07069406658411026, + "learning_rate": 3.8428756197878514e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23360 + }, + { + "epoch": 6.799534477742217, + "flow_matching_loss": 0.015924100670963524, + "grad_norm": 0.10681464523077011, + "learning_rate": 3.831830723710992e-06, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23370 + }, + { + "epoch": 6.802443991853361, + "flow_matching_loss": 0.01461092997342348, + "grad_norm": 0.06824325770139694, + "learning_rate": 3.820799398012889e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23380 + }, + { + "epoch": 6.805353505964504, + "flow_matching_loss": 0.014728863444179296, + "grad_norm": 0.08603712916374207, + "learning_rate": 3.80978165609765e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23390 + }, + { + "epoch": 6.808263020075647, + "flow_matching_loss": 0.013009082712233066, + "grad_norm": 0.09025676548480988, + "learning_rate": 3.798777511352864e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23400 + }, + { + "epoch": 6.811172534186791, + "flow_matching_loss": 0.014665239118039608, + "grad_norm": 0.06268615275621414, + "learning_rate": 3.787786977149613e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23410 + }, + { + "epoch": 6.8140820482979345, + "flow_matching_loss": 0.015484282467514276, + "grad_norm": 0.06645560264587402, + "learning_rate": 3.776810066842427e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23420 + }, + { + "epoch": 6.816991562409077, + "flow_matching_loss": 0.015485009271651506, + "grad_norm": 0.061919983476400375, + "learning_rate": 3.7658467937692866e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23430 + }, + { + "epoch": 6.819901076520221, + "flow_matching_loss": 0.01363168740645051, + "grad_norm": 0.06912839412689209, + "learning_rate": 3.7548971712516e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23440 + }, + { + "epoch": 6.822810590631365, + "flow_matching_loss": 0.013079597894102336, + "grad_norm": 0.1410837173461914, + "learning_rate": 3.743961212594197e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23450 + }, + { + "epoch": 6.8257201047425085, + "flow_matching_loss": 0.01551303118467331, + "grad_norm": 0.07896018028259277, + "learning_rate": 3.7330389310852975e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23460 + }, + { + "epoch": 6.828629618853651, + "flow_matching_loss": 0.01644034143537283, + "grad_norm": 0.09911727905273438, + "learning_rate": 3.7221303399965008e-06, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23470 + }, + { + "epoch": 6.831539132964795, + "flow_matching_loss": 0.01558038080111146, + "grad_norm": 0.07836750894784927, + "learning_rate": 3.711235452582775e-06, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23480 + }, + { + "epoch": 6.834448647075939, + "flow_matching_loss": 0.013096069311723113, + "grad_norm": 0.0815233513712883, + "learning_rate": 3.7003542820824347e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23490 + }, + { + "epoch": 6.8373581611870815, + "flow_matching_loss": 0.015426663123071194, + "grad_norm": 0.08288561552762985, + "learning_rate": 3.6894868417171252e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23500 + }, + { + "epoch": 6.840267675298225, + "flow_matching_loss": 0.013220719713717698, + "grad_norm": 0.07118070870637894, + "learning_rate": 3.678633144691818e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23510 + }, + { + "epoch": 6.843177189409369, + "flow_matching_loss": 0.015953687578439714, + "grad_norm": 0.10277502238750458, + "learning_rate": 3.667793204194776e-06, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23520 + }, + { + "epoch": 6.846086703520512, + "flow_matching_loss": 0.015406224969774484, + "grad_norm": 0.06642258912324905, + "learning_rate": 3.656967033397544e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23530 + }, + { + "epoch": 6.8489962176316554, + "flow_matching_loss": 0.015998510271310808, + "grad_norm": 0.10257942974567413, + "learning_rate": 3.646154645454948e-06, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23540 + }, + { + "epoch": 6.851905731742799, + "flow_matching_loss": 0.015213276911526918, + "grad_norm": 0.052887432277202606, + "learning_rate": 3.6353560535050524e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23550 + }, + { + "epoch": 6.854815245853942, + "flow_matching_loss": 0.015122593566775321, + "grad_norm": 0.09402100741863251, + "learning_rate": 3.624571270669172e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23560 + }, + { + "epoch": 6.857724759965086, + "flow_matching_loss": 0.015614959225058556, + "grad_norm": 0.08969062566757202, + "learning_rate": 3.6138003100518296e-06, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23570 + }, + { + "epoch": 6.860634274076229, + "flow_matching_loss": 0.013843958638608456, + "grad_norm": 0.12225569039583206, + "learning_rate": 3.6030431847407595e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23580 + }, + { + "epoch": 6.863543788187373, + "flow_matching_loss": 0.014846247993409634, + "grad_norm": 0.0943889394402504, + "learning_rate": 3.592299907806881e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23590 + }, + { + "epoch": 6.866453302298516, + "flow_matching_loss": 0.014818381611257791, + "grad_norm": 0.06585124135017395, + "learning_rate": 3.5815704923042892e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23600 + }, + { + "epoch": 6.86936281640966, + "flow_matching_loss": 0.014092313870787621, + "grad_norm": 0.099324069917202, + "learning_rate": 3.5708549512702392e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23610 + }, + { + "epoch": 6.872272330520803, + "flow_matching_loss": 0.01487248670309782, + "grad_norm": 0.11814488470554352, + "learning_rate": 3.5601532977251217e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23620 + }, + { + "epoch": 6.875181844631946, + "flow_matching_loss": 0.013458778103813528, + "grad_norm": 0.11356566101312637, + "learning_rate": 3.5494655446724544e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23630 + }, + { + "epoch": 6.87809135874309, + "flow_matching_loss": 0.013205097243189811, + "grad_norm": 0.09578078240156174, + "learning_rate": 3.538791705098866e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23640 + }, + { + "epoch": 6.881000872854234, + "flow_matching_loss": 0.014721087645739317, + "grad_norm": 0.13309697806835175, + "learning_rate": 3.5281317919740718e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23650 + }, + { + "epoch": 6.883910386965377, + "flow_matching_loss": 0.017367717064917087, + "grad_norm": 0.07497905194759369, + "learning_rate": 3.5174858182508825e-06, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23660 + }, + { + "epoch": 6.88681990107652, + "flow_matching_loss": 0.014582452364265919, + "grad_norm": 0.08556178957223892, + "learning_rate": 3.506853796865153e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23670 + }, + { + "epoch": 6.889729415187664, + "flow_matching_loss": 0.012884492613375187, + "grad_norm": 0.09921980649232864, + "learning_rate": 3.496235740735789e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23680 + }, + { + "epoch": 6.892638929298807, + "flow_matching_loss": 0.0135859876871109, + "grad_norm": 0.07906624674797058, + "learning_rate": 3.485631662764738e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23690 + }, + { + "epoch": 6.89554844340995, + "flow_matching_loss": 0.01301106046885252, + "grad_norm": 0.09686422348022461, + "learning_rate": 3.475041575836952e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23700 + }, + { + "epoch": 6.898457957521094, + "flow_matching_loss": 0.013715807814151049, + "grad_norm": 0.08254915475845337, + "learning_rate": 3.4644654928203796e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23710 + }, + { + "epoch": 6.901367471632238, + "flow_matching_loss": 0.01534593738615513, + "grad_norm": 0.0834808200597763, + "learning_rate": 3.4539034265659663e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23720 + }, + { + "epoch": 6.904276985743381, + "flow_matching_loss": 0.014327230956405401, + "grad_norm": 0.08321879059076309, + "learning_rate": 3.4433553899076163e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23730 + }, + { + "epoch": 6.907186499854524, + "flow_matching_loss": 0.014451778307557107, + "grad_norm": 0.061378322541713715, + "learning_rate": 3.432821395662189e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23740 + }, + { + "epoch": 6.910096013965668, + "flow_matching_loss": 0.016007500514388085, + "grad_norm": 0.07999450713396072, + "learning_rate": 3.4223014566294785e-06, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23750 + }, + { + "epoch": 6.913005528076811, + "flow_matching_loss": 0.015633990056812763, + "grad_norm": 0.08298293501138687, + "learning_rate": 3.4117955855922e-06, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23760 + }, + { + "epoch": 6.9159150421879545, + "flow_matching_loss": 0.014301945827901364, + "grad_norm": 0.08933848887681961, + "learning_rate": 3.4013037953159855e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23770 + }, + { + "epoch": 6.918824556299098, + "flow_matching_loss": 0.014055878482758999, + "grad_norm": 0.06922994554042816, + "learning_rate": 3.390826098549345e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23780 + }, + { + "epoch": 6.921734070410242, + "flow_matching_loss": 0.013960854429751635, + "grad_norm": 0.08900177478790283, + "learning_rate": 3.3803625080236684e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23790 + }, + { + "epoch": 6.924643584521385, + "flow_matching_loss": 0.014415589440613986, + "grad_norm": 0.06717699021100998, + "learning_rate": 3.369913036453205e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23800 + }, + { + "epoch": 6.927553098632528, + "flow_matching_loss": 0.01534127672202885, + "grad_norm": 0.1446162611246109, + "learning_rate": 3.3594776965350434e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23810 + }, + { + "epoch": 6.930462612743672, + "flow_matching_loss": 0.015040928218513726, + "grad_norm": 0.08802912384271622, + "learning_rate": 3.34905650094911e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23820 + }, + { + "epoch": 6.933372126854815, + "flow_matching_loss": 0.017083355877548455, + "grad_norm": 0.10448340326547623, + "learning_rate": 3.338649462358145e-06, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23830 + }, + { + "epoch": 6.936281640965959, + "flow_matching_loss": 0.014049726724624633, + "grad_norm": 0.08511403948068619, + "learning_rate": 3.3282565934076765e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23840 + }, + { + "epoch": 6.939191155077102, + "flow_matching_loss": 0.015915956068783998, + "grad_norm": 0.10706716030836105, + "learning_rate": 3.31787790672602e-06, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23850 + }, + { + "epoch": 6.942100669188245, + "flow_matching_loss": 0.014723649993538857, + "grad_norm": 0.06836281716823578, + "learning_rate": 3.307513414924255e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23860 + }, + { + "epoch": 6.945010183299389, + "flow_matching_loss": 0.013444055151194335, + "grad_norm": 0.08041879534721375, + "learning_rate": 3.2971631305962246e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23870 + }, + { + "epoch": 6.947919697410533, + "flow_matching_loss": 0.01240438623353839, + "grad_norm": 0.08601761609315872, + "learning_rate": 3.2868270663184964e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23880 + }, + { + "epoch": 6.950829211521675, + "flow_matching_loss": 0.013248140830546618, + "grad_norm": 0.07014002650976181, + "learning_rate": 3.276505234650362e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23890 + }, + { + "epoch": 6.953738725632819, + "flow_matching_loss": 0.014944649394601583, + "grad_norm": 0.07840918749570847, + "learning_rate": 3.266197648133821e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23900 + }, + { + "epoch": 6.956648239743963, + "flow_matching_loss": 0.013518750574439764, + "grad_norm": 0.07076043635606766, + "learning_rate": 3.255904319293562e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23910 + }, + { + "epoch": 6.9595577538551066, + "flow_matching_loss": 0.012040359061211347, + "grad_norm": 0.09234936535358429, + "learning_rate": 3.2456252606369485e-06, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23920 + }, + { + "epoch": 6.962467267966249, + "flow_matching_loss": 0.014245411474257707, + "grad_norm": 0.0867217481136322, + "learning_rate": 3.235360484654011e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23930 + }, + { + "epoch": 6.965376782077393, + "flow_matching_loss": 0.013906595949083566, + "grad_norm": 0.10886604338884354, + "learning_rate": 3.225110003817418e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23940 + }, + { + "epoch": 6.968286296188537, + "flow_matching_loss": 0.015962945576757193, + "grad_norm": 0.12642517685890198, + "learning_rate": 3.214873830582467e-06, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23950 + }, + { + "epoch": 6.97119581029968, + "flow_matching_loss": 0.016628112085163592, + "grad_norm": 0.08680090308189392, + "learning_rate": 3.2046519773870803e-06, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23960 + }, + { + "epoch": 6.974105324410823, + "flow_matching_loss": 0.01591313788667321, + "grad_norm": 0.09093563258647919, + "learning_rate": 3.1944444566517654e-06, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23970 + }, + { + "epoch": 6.977014838521967, + "flow_matching_loss": 0.013329457305371762, + "grad_norm": 0.07049701362848282, + "learning_rate": 3.18425128077963e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23980 + }, + { + "epoch": 6.979924352633111, + "flow_matching_loss": 0.01388767482712865, + "grad_norm": 0.08359670639038086, + "learning_rate": 3.174072462156342e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23990 + }, + { + "epoch": 6.9828338667442535, + "flow_matching_loss": 0.013165712356567383, + "grad_norm": 0.08165507018566132, + "learning_rate": 3.163908013150124e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24000 + }, + { + "epoch": 6.985743380855397, + "flow_matching_loss": 0.01583755649626255, + "grad_norm": 0.06622534990310669, + "learning_rate": 3.153757946111741e-06, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24010 + }, + { + "epoch": 6.988652894966541, + "flow_matching_loss": 0.014304531551897526, + "grad_norm": 0.08171389997005463, + "learning_rate": 3.1436222733744776e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24020 + }, + { + "epoch": 6.991562409077684, + "flow_matching_loss": 0.013287617964670062, + "grad_norm": 0.06681279838085175, + "learning_rate": 3.133501007254138e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24030 + }, + { + "epoch": 6.9944719231888275, + "flow_matching_loss": 0.011539644002914429, + "grad_norm": 0.1058812364935875, + "learning_rate": 3.1233941600490123e-06, + "loss": 0.0115, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24040 + }, + { + "epoch": 6.997381437299971, + "flow_matching_loss": 0.014381324220448733, + "grad_norm": 0.10661996155977249, + "learning_rate": 3.113301744039875e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24050 + }, + { + "epoch": 7.000290951411114, + "flow_matching_loss": 0.014738664776086808, + "grad_norm": 0.10041991621255875, + "learning_rate": 3.103223771489959e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24060 + }, + { + "epoch": 7.003200465522258, + "flow_matching_loss": 0.0148294388782233, + "grad_norm": 0.10860630869865417, + "learning_rate": 3.09316025464495e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24070 + }, + { + "epoch": 7.006109979633401, + "flow_matching_loss": 0.014451954513788223, + "grad_norm": 0.07793910056352615, + "learning_rate": 3.0831112057329775e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24080 + }, + { + "epoch": 7.009019493744544, + "flow_matching_loss": 0.013766560330986976, + "grad_norm": 0.0981958732008934, + "learning_rate": 3.073076636964579e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24090 + }, + { + "epoch": 7.011929007855688, + "flow_matching_loss": 0.015134428534656763, + "grad_norm": 0.17631356418132782, + "learning_rate": 3.063056560532698e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24100 + }, + { + "epoch": 7.014838521966832, + "flow_matching_loss": 0.012943075690418481, + "grad_norm": 0.06869272142648697, + "learning_rate": 3.0530509886126807e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24110 + }, + { + "epoch": 7.017748036077975, + "flow_matching_loss": 0.013626334443688392, + "grad_norm": 0.0909290611743927, + "learning_rate": 3.043059933362236e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24120 + }, + { + "epoch": 7.020657550189118, + "flow_matching_loss": 0.014845193270593882, + "grad_norm": 0.09007915109395981, + "learning_rate": 3.0330834069214337e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24130 + }, + { + "epoch": 7.023567064300262, + "flow_matching_loss": 0.013301121722906827, + "grad_norm": 0.08213095366954803, + "learning_rate": 3.023121421412703e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24140 + }, + { + "epoch": 7.026476578411406, + "flow_matching_loss": 0.013067401479929686, + "grad_norm": 0.0700477808713913, + "learning_rate": 3.0131739889407896e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24150 + }, + { + "epoch": 7.029386092522548, + "flow_matching_loss": 0.015728855226188897, + "grad_norm": 0.06115733087062836, + "learning_rate": 3.0032411215927645e-06, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24160 + }, + { + "epoch": 7.032295606633692, + "flow_matching_loss": 0.01463871793821454, + "grad_norm": 0.09296828508377075, + "learning_rate": 2.9933228314379975e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24170 + }, + { + "epoch": 7.035205120744836, + "flow_matching_loss": 0.015334279276430607, + "grad_norm": 0.0894642248749733, + "learning_rate": 2.9834191305281405e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24180 + }, + { + "epoch": 7.038114634855979, + "flow_matching_loss": 0.012319618090987205, + "grad_norm": 0.08764513581991196, + "learning_rate": 2.973530030897135e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24190 + }, + { + "epoch": 7.041024148967122, + "flow_matching_loss": 0.01615680046379566, + "grad_norm": 0.0720524862408638, + "learning_rate": 2.9636555445611646e-06, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24200 + }, + { + "epoch": 7.043933663078266, + "flow_matching_loss": 0.01505158757790923, + "grad_norm": 0.12520870566368103, + "learning_rate": 2.953795683518659e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24210 + }, + { + "epoch": 7.04684317718941, + "flow_matching_loss": 0.013619162980467081, + "grad_norm": 0.08142644912004471, + "learning_rate": 2.9439504597502815e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24220 + }, + { + "epoch": 7.049752691300553, + "flow_matching_loss": 0.013034352706745267, + "grad_norm": 0.0805656835436821, + "learning_rate": 2.934119885218902e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24230 + }, + { + "epoch": 7.052662205411696, + "flow_matching_loss": 0.012600676715373993, + "grad_norm": 0.08817769587039948, + "learning_rate": 2.9243039718696045e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24240 + }, + { + "epoch": 7.05557171952284, + "flow_matching_loss": 0.013948963955044746, + "grad_norm": 0.09132271260023117, + "learning_rate": 2.914502731629641e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24250 + }, + { + "epoch": 7.058481233633983, + "flow_matching_loss": 0.013141225185245275, + "grad_norm": 0.06754092872142792, + "learning_rate": 2.904716176408451e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24260 + }, + { + "epoch": 7.0613907477451265, + "flow_matching_loss": 0.01394431982189417, + "grad_norm": 0.06645453721284866, + "learning_rate": 2.8949443180976172e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24270 + }, + { + "epoch": 7.06430026185627, + "flow_matching_loss": 0.013113988563418388, + "grad_norm": 0.07038965076208115, + "learning_rate": 2.885187168570864e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24280 + }, + { + "epoch": 7.067209775967413, + "flow_matching_loss": 0.01568224569782615, + "grad_norm": 0.08050306886434555, + "learning_rate": 2.875444739684058e-06, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24290 + }, + { + "epoch": 7.070119290078557, + "flow_matching_loss": 0.014869879744946957, + "grad_norm": 0.09009259194135666, + "learning_rate": 2.865717043275165e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24300 + }, + { + "epoch": 7.0730288041897005, + "flow_matching_loss": 0.015265162754803895, + "grad_norm": 0.1003505289554596, + "learning_rate": 2.8560040911642528e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24310 + }, + { + "epoch": 7.075938318300844, + "flow_matching_loss": 0.01397160543128848, + "grad_norm": 0.08778946846723557, + "learning_rate": 2.846305895153472e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24320 + }, + { + "epoch": 7.078847832411987, + "flow_matching_loss": 0.013099048659205437, + "grad_norm": 0.08546307682991028, + "learning_rate": 2.8366224670270487e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24330 + }, + { + "epoch": 7.081757346523131, + "flow_matching_loss": 0.01460324777290225, + "grad_norm": 0.08919332176446915, + "learning_rate": 2.8269538185512535e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24340 + }, + { + "epoch": 7.084666860634274, + "flow_matching_loss": 0.014840057119727135, + "grad_norm": 0.08139149099588394, + "learning_rate": 2.817299961474417e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24350 + }, + { + "epoch": 7.087576374745417, + "flow_matching_loss": 0.014719145465642213, + "grad_norm": 0.08747754245996475, + "learning_rate": 2.8076609075268816e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24360 + }, + { + "epoch": 7.090485888856561, + "flow_matching_loss": 0.014131759479641914, + "grad_norm": 0.0829755887389183, + "learning_rate": 2.798036668421006e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24370 + }, + { + "epoch": 7.093395402967705, + "flow_matching_loss": 0.01355854570865631, + "grad_norm": 0.08808032423257828, + "learning_rate": 2.7884272558511433e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24380 + }, + { + "epoch": 7.0963049170788475, + "flow_matching_loss": 0.014214759040623904, + "grad_norm": 0.1176396980881691, + "learning_rate": 2.7788326814936456e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24390 + }, + { + "epoch": 7.099214431189991, + "flow_matching_loss": 0.015254991035908461, + "grad_norm": 0.18960964679718018, + "learning_rate": 2.7692529570068176e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24400 + }, + { + "epoch": 7.102123945301135, + "flow_matching_loss": 0.01680911621078849, + "grad_norm": 0.09635687619447708, + "learning_rate": 2.759688094030933e-06, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24410 + }, + { + "epoch": 7.105033459412278, + "flow_matching_loss": 0.0141983050853014, + "grad_norm": 0.09011179208755493, + "learning_rate": 2.750138104188201e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24420 + }, + { + "epoch": 7.107942973523421, + "flow_matching_loss": 0.015427737589925527, + "grad_norm": 0.116549551486969, + "learning_rate": 2.740602999082759e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24430 + }, + { + "epoch": 7.110852487634565, + "flow_matching_loss": 0.015776320919394494, + "grad_norm": 0.12605592608451843, + "learning_rate": 2.731082790300654e-06, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24440 + }, + { + "epoch": 7.113762001745709, + "flow_matching_loss": 0.01865340182557702, + "grad_norm": 0.10537190735340118, + "learning_rate": 2.7215774894098474e-06, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24450 + }, + { + "epoch": 7.116671515856852, + "flow_matching_loss": 0.011798585159704089, + "grad_norm": 0.10976215451955795, + "learning_rate": 2.7120871079601696e-06, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24460 + }, + { + "epoch": 7.119581029967995, + "flow_matching_loss": 0.012117305118590594, + "grad_norm": 0.08976855129003525, + "learning_rate": 2.7026116574833314e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24470 + }, + { + "epoch": 7.122490544079139, + "flow_matching_loss": 0.015622575301676989, + "grad_norm": 0.08397896587848663, + "learning_rate": 2.6931511494928953e-06, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24480 + }, + { + "epoch": 7.125400058190282, + "flow_matching_loss": 0.011889915307983756, + "grad_norm": 0.06583145260810852, + "learning_rate": 2.683705595484275e-06, + "loss": 0.0119, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24490 + }, + { + "epoch": 7.128309572301426, + "flow_matching_loss": 0.014216344896703958, + "grad_norm": 0.06377292424440384, + "learning_rate": 2.6742750069347014e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24500 + }, + { + "epoch": 7.131219086412569, + "flow_matching_loss": 0.014751075860112905, + "grad_norm": 0.0943383201956749, + "learning_rate": 2.664859395303239e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24510 + }, + { + "epoch": 7.134128600523712, + "flow_matching_loss": 0.013865306694060563, + "grad_norm": 0.0669148713350296, + "learning_rate": 2.655458772030739e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24520 + }, + { + "epoch": 7.137038114634856, + "flow_matching_loss": 0.012914431001991033, + "grad_norm": 0.07629422843456268, + "learning_rate": 2.6460731485398403e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24530 + }, + { + "epoch": 7.1399476287459995, + "flow_matching_loss": 0.013906699046492577, + "grad_norm": 0.08695345371961594, + "learning_rate": 2.6367025362349695e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24540 + }, + { + "epoch": 7.142857142857143, + "flow_matching_loss": 0.013718688022345304, + "grad_norm": 0.0809130147099495, + "learning_rate": 2.6273469465022965e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24550 + }, + { + "epoch": 7.145766656968286, + "flow_matching_loss": 0.01398463686928153, + "grad_norm": 0.11323913931846619, + "learning_rate": 2.618006390709753e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24560 + }, + { + "epoch": 7.14867617107943, + "flow_matching_loss": 0.013623181823641061, + "grad_norm": 0.0716513991355896, + "learning_rate": 2.60868088020699e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24570 + }, + { + "epoch": 7.1515856851905735, + "flow_matching_loss": 0.013491208478808404, + "grad_norm": 0.0744086503982544, + "learning_rate": 2.599370426325382e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24580 + }, + { + "epoch": 7.154495199301716, + "flow_matching_loss": 0.014759252499788999, + "grad_norm": 0.09412649273872375, + "learning_rate": 2.5900750403780093e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24590 + }, + { + "epoch": 7.15740471341286, + "flow_matching_loss": 0.014976971689611674, + "grad_norm": 0.07438778877258301, + "learning_rate": 2.580794733659639e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24600 + }, + { + "epoch": 7.160314227524004, + "flow_matching_loss": 0.013435850758105517, + "grad_norm": 0.0813860148191452, + "learning_rate": 2.5715295174467236e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24610 + }, + { + "epoch": 7.1632237416351465, + "flow_matching_loss": 0.013847048301249743, + "grad_norm": 0.07312171906232834, + "learning_rate": 2.5622794029973746e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24620 + }, + { + "epoch": 7.16613325574629, + "flow_matching_loss": 0.013677220698446036, + "grad_norm": 0.08021225035190582, + "learning_rate": 2.553044401551351e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24630 + }, + { + "epoch": 7.169042769857434, + "flow_matching_loss": 0.012541955895721912, + "grad_norm": 0.0745791420340538, + "learning_rate": 2.54382452433005e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24640 + }, + { + "epoch": 7.171952283968578, + "flow_matching_loss": 0.011492510000243783, + "grad_norm": 0.06309379637241364, + "learning_rate": 2.53461978253649e-06, + "loss": 0.0115, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24650 + }, + { + "epoch": 7.1748617980797205, + "flow_matching_loss": 0.014324071258306504, + "grad_norm": 0.07071365416049957, + "learning_rate": 2.5254301873553065e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24660 + }, + { + "epoch": 7.177771312190864, + "flow_matching_loss": 0.01284454008564353, + "grad_norm": 0.10694614052772522, + "learning_rate": 2.5162557499527184e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24670 + }, + { + "epoch": 7.180680826302008, + "flow_matching_loss": 0.01276822416111827, + "grad_norm": 0.08707603067159653, + "learning_rate": 2.5070964814765335e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24680 + }, + { + "epoch": 7.183590340413151, + "flow_matching_loss": 0.014605947677046061, + "grad_norm": 0.09120558202266693, + "learning_rate": 2.4979523930561295e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24690 + }, + { + "epoch": 7.186499854524294, + "flow_matching_loss": 0.014178356062620879, + "grad_norm": 0.1126769557595253, + "learning_rate": 2.4888234958024333e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24700 + }, + { + "epoch": 7.189409368635438, + "flow_matching_loss": 0.015926412492990493, + "grad_norm": 0.10227059572935104, + "learning_rate": 2.479709800807915e-06, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24710 + }, + { + "epoch": 7.192318882746581, + "flow_matching_loss": 0.012333379499614239, + "grad_norm": 0.07101229578256607, + "learning_rate": 2.4706113191465767e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24720 + }, + { + "epoch": 7.195228396857725, + "flow_matching_loss": 0.012839423585683107, + "grad_norm": 0.11575376242399216, + "learning_rate": 2.46152806187393e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24730 + }, + { + "epoch": 7.198137910968868, + "flow_matching_loss": 0.014925766550004483, + "grad_norm": 0.08130719512701035, + "learning_rate": 2.452460040026989e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24740 + }, + { + "epoch": 7.201047425080012, + "flow_matching_loss": 0.014196109585464, + "grad_norm": 0.0756433755159378, + "learning_rate": 2.4434072646242555e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24750 + }, + { + "epoch": 7.203956939191155, + "flow_matching_loss": 0.013366247527301311, + "grad_norm": 0.06709844619035721, + "learning_rate": 2.4343697466657015e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24760 + }, + { + "epoch": 7.206866453302299, + "flow_matching_loss": 0.015955084562301637, + "grad_norm": 0.09726931899785995, + "learning_rate": 2.42534749713277e-06, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24770 + }, + { + "epoch": 7.209775967413442, + "flow_matching_loss": 0.013478817977011203, + "grad_norm": 0.05198851600289345, + "learning_rate": 2.4163405269883425e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24780 + }, + { + "epoch": 7.212685481524585, + "flow_matching_loss": 0.01490011252462864, + "grad_norm": 0.07226741313934326, + "learning_rate": 2.407348847176735e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24790 + }, + { + "epoch": 7.215594995635729, + "flow_matching_loss": 0.014689784590154886, + "grad_norm": 0.08749768882989883, + "learning_rate": 2.398372468623688e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24800 + }, + { + "epoch": 7.2185045097468725, + "flow_matching_loss": 0.014946876931935549, + "grad_norm": 0.1149258092045784, + "learning_rate": 2.3894114022363446e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24810 + }, + { + "epoch": 7.221414023858015, + "flow_matching_loss": 0.015387535654008388, + "grad_norm": 0.11329058557748795, + "learning_rate": 2.3804656589032514e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24820 + }, + { + "epoch": 7.224323537969159, + "flow_matching_loss": 0.015077253151685, + "grad_norm": 0.08026096969842911, + "learning_rate": 2.3715352494943237e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24830 + }, + { + "epoch": 7.227233052080303, + "flow_matching_loss": 0.013415753841400146, + "grad_norm": 0.07196564227342606, + "learning_rate": 2.3626201848608582e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24840 + }, + { + "epoch": 7.2301425661914465, + "flow_matching_loss": 0.013940725289285184, + "grad_norm": 0.09749576449394226, + "learning_rate": 2.353720475835494e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24850 + }, + { + "epoch": 7.233052080302589, + "flow_matching_loss": 0.01850587734952569, + "grad_norm": 0.10049186646938324, + "learning_rate": 2.344836133232215e-06, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24860 + }, + { + "epoch": 7.235961594413733, + "flow_matching_loss": 0.013526275847107172, + "grad_norm": 0.11583643406629562, + "learning_rate": 2.335967167846342e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24870 + }, + { + "epoch": 7.238871108524877, + "flow_matching_loss": 0.01520079029724002, + "grad_norm": 0.11711856722831726, + "learning_rate": 2.3271135904544993e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24880 + }, + { + "epoch": 7.2417806226360195, + "flow_matching_loss": 0.014551408030092717, + "grad_norm": 0.08107373118400574, + "learning_rate": 2.3182754118146206e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24890 + }, + { + "epoch": 7.244690136747163, + "flow_matching_loss": 0.012840803433209657, + "grad_norm": 0.07280663400888443, + "learning_rate": 2.309452642665923e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24900 + }, + { + "epoch": 7.247599650858307, + "flow_matching_loss": 0.015533262584358453, + "grad_norm": 0.08611547201871872, + "learning_rate": 2.3006452937289065e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24910 + }, + { + "epoch": 7.25050916496945, + "flow_matching_loss": 0.015391664672642946, + "grad_norm": 0.07071756571531296, + "learning_rate": 2.291853375705327e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24920 + }, + { + "epoch": 7.2534186790805935, + "flow_matching_loss": 0.015476942341774703, + "grad_norm": 0.10620583593845367, + "learning_rate": 2.283076899278197e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24930 + }, + { + "epoch": 7.256328193191737, + "flow_matching_loss": 0.015478803124278784, + "grad_norm": 0.158305823802948, + "learning_rate": 2.274315875111763e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24940 + }, + { + "epoch": 7.25923770730288, + "flow_matching_loss": 0.012328649405390024, + "grad_norm": 0.08065329492092133, + "learning_rate": 2.265570313851493e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24950 + }, + { + "epoch": 7.262147221414024, + "flow_matching_loss": 0.01550623644143343, + "grad_norm": 0.08298835903406143, + "learning_rate": 2.2568402261240685e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24960 + }, + { + "epoch": 7.265056735525167, + "flow_matching_loss": 0.01516477670520544, + "grad_norm": 0.09317749738693237, + "learning_rate": 2.248125622537373e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24970 + }, + { + "epoch": 7.267966249636311, + "flow_matching_loss": 0.013006569352000952, + "grad_norm": 0.07825423777103424, + "learning_rate": 2.239426513680467e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24980 + }, + { + "epoch": 7.270875763747454, + "flow_matching_loss": 0.013126688497141003, + "grad_norm": 0.07252056151628494, + "learning_rate": 2.2307429101235915e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24990 + }, + { + "epoch": 7.273785277858598, + "flow_matching_loss": 0.016631345078349114, + "grad_norm": 0.07057835161685944, + "learning_rate": 2.222074822418145e-06, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25000 + }, + { + "epoch": 7.276694791969741, + "flow_matching_loss": 0.013238988770172, + "grad_norm": 0.08830960839986801, + "learning_rate": 2.213422261096668e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25010 + }, + { + "epoch": 7.279604306080884, + "flow_matching_loss": 0.013071057526394725, + "grad_norm": 0.09530115127563477, + "learning_rate": 2.204785236672835e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25020 + }, + { + "epoch": 7.282513820192028, + "flow_matching_loss": 0.014754211530089378, + "grad_norm": 0.06121613085269928, + "learning_rate": 2.1961637596414536e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25030 + }, + { + "epoch": 7.285423334303172, + "flow_matching_loss": 0.015613781381398439, + "grad_norm": 0.06988568603992462, + "learning_rate": 2.1875578404784275e-06, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25040 + }, + { + "epoch": 7.288332848414315, + "flow_matching_loss": 0.012992769479751587, + "grad_norm": 0.10253952443599701, + "learning_rate": 2.1789674896407583e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25050 + }, + { + "epoch": 7.291242362525458, + "flow_matching_loss": 0.016131370048969983, + "grad_norm": 0.07271061092615128, + "learning_rate": 2.170392717566534e-06, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25060 + }, + { + "epoch": 7.294151876636602, + "flow_matching_loss": 0.013385040126740933, + "grad_norm": 0.13305294513702393, + "learning_rate": 2.161833534674909e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25070 + }, + { + "epoch": 7.2970613907477455, + "flow_matching_loss": 0.016781340446323158, + "grad_norm": 0.12453014403581619, + "learning_rate": 2.1532899513661015e-06, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25080 + }, + { + "epoch": 7.299970904858888, + "flow_matching_loss": 0.013119752053171396, + "grad_norm": 0.09980730712413788, + "learning_rate": 2.1447619780213682e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25090 + }, + { + "epoch": 7.302880418970032, + "flow_matching_loss": 0.014230965543538332, + "grad_norm": 0.09392115473747253, + "learning_rate": 2.136249625003001e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25100 + }, + { + "epoch": 7.305789933081176, + "flow_matching_loss": 0.015129288472235203, + "grad_norm": 0.087183877825737, + "learning_rate": 2.1277529026543107e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25110 + }, + { + "epoch": 7.308699447192319, + "flow_matching_loss": 0.013069019559770823, + "grad_norm": 0.08635169267654419, + "learning_rate": 2.1192718212996188e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25120 + }, + { + "epoch": 7.311608961303462, + "flow_matching_loss": 0.015527315437793732, + "grad_norm": 0.08820019662380219, + "learning_rate": 2.110806391244235e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25130 + }, + { + "epoch": 7.314518475414606, + "flow_matching_loss": 0.015679367538541555, + "grad_norm": 0.08156926184892654, + "learning_rate": 2.1023566227744596e-06, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25140 + }, + { + "epoch": 7.317427989525749, + "flow_matching_loss": 0.0138743843883276, + "grad_norm": 0.0640530064702034, + "learning_rate": 2.0939225261575563e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25150 + }, + { + "epoch": 7.3203375036368925, + "flow_matching_loss": 0.014235263876616954, + "grad_norm": 0.0961003229022026, + "learning_rate": 2.0855041116417473e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25160 + }, + { + "epoch": 7.323247017748036, + "flow_matching_loss": 0.014080692082643509, + "grad_norm": 0.08768102526664734, + "learning_rate": 2.0771013894562003e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25170 + }, + { + "epoch": 7.32615653185918, + "flow_matching_loss": 0.014371582493185996, + "grad_norm": 0.12016171962022781, + "learning_rate": 2.0687143698110113e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25180 + }, + { + "epoch": 7.329066045970323, + "flow_matching_loss": 0.014258750155568123, + "grad_norm": 0.0708664208650589, + "learning_rate": 2.060343062897208e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25190 + }, + { + "epoch": 7.3319755600814664, + "flow_matching_loss": 0.014215134549885988, + "grad_norm": 0.06035872548818588, + "learning_rate": 2.0519874788867127e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25200 + }, + { + "epoch": 7.33488507419261, + "flow_matching_loss": 0.014812026266008615, + "grad_norm": 0.08752720803022385, + "learning_rate": 2.043647627932347e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25210 + }, + { + "epoch": 7.337794588303753, + "flow_matching_loss": 0.014338266989216208, + "grad_norm": 0.09384816139936447, + "learning_rate": 2.0353235201678193e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25220 + }, + { + "epoch": 7.340704102414897, + "flow_matching_loss": 0.01444038413465023, + "grad_norm": 0.06799960136413574, + "learning_rate": 2.0270151657076995e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25230 + }, + { + "epoch": 7.34361361652604, + "flow_matching_loss": 0.01287532178685069, + "grad_norm": 0.06484532356262207, + "learning_rate": 2.0187225746474286e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25240 + }, + { + "epoch": 7.346523130637183, + "flow_matching_loss": 0.014621039573103189, + "grad_norm": 0.09197019040584564, + "learning_rate": 2.0104457570632845e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25250 + }, + { + "epoch": 7.349432644748327, + "flow_matching_loss": 0.014295019768178463, + "grad_norm": 0.058059025555849075, + "learning_rate": 2.0021847230123775e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25260 + }, + { + "epoch": 7.352342158859471, + "flow_matching_loss": 0.015139515977352858, + "grad_norm": 0.07314454764127731, + "learning_rate": 1.9939394825326494e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25270 + }, + { + "epoch": 7.355251672970613, + "flow_matching_loss": 0.01318295681849122, + "grad_norm": 0.08144327998161316, + "learning_rate": 1.9857100456428385e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25280 + }, + { + "epoch": 7.358161187081757, + "flow_matching_loss": 0.014203503774479032, + "grad_norm": 0.054150622338056564, + "learning_rate": 1.977496422342492e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25290 + }, + { + "epoch": 7.361070701192901, + "flow_matching_loss": 0.012850214447826146, + "grad_norm": 0.07243940234184265, + "learning_rate": 1.9692986226119345e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25300 + }, + { + "epoch": 7.363980215304045, + "flow_matching_loss": 0.014566637389361859, + "grad_norm": 0.10153475403785706, + "learning_rate": 1.9611166564122658e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25310 + }, + { + "epoch": 7.366889729415187, + "flow_matching_loss": 0.012606734596192836, + "grad_norm": 0.07959485054016113, + "learning_rate": 1.9529505336853456e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25320 + }, + { + "epoch": 7.369799243526331, + "flow_matching_loss": 0.014409357868134975, + "grad_norm": 0.11986146122217178, + "learning_rate": 1.9448002643537833e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25330 + }, + { + "epoch": 7.372708757637475, + "flow_matching_loss": 0.014391877595335245, + "grad_norm": 0.08762197941541672, + "learning_rate": 1.9366658583209205e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25340 + }, + { + "epoch": 7.375618271748618, + "flow_matching_loss": 0.014386576134711504, + "grad_norm": 0.07910101860761642, + "learning_rate": 1.9285473254708363e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25350 + }, + { + "epoch": 7.378527785859761, + "flow_matching_loss": 0.01367629151791334, + "grad_norm": 0.08921607583761215, + "learning_rate": 1.9204446756683092e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25360 + }, + { + "epoch": 7.381437299970905, + "flow_matching_loss": 0.013233462255448103, + "grad_norm": 0.13044871389865875, + "learning_rate": 1.912357918758822e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25370 + }, + { + "epoch": 7.384346814082049, + "flow_matching_loss": 0.013972825091332196, + "grad_norm": 0.07918258756399155, + "learning_rate": 1.9042870645685478e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25380 + }, + { + "epoch": 7.387256328193192, + "flow_matching_loss": 0.014089339319616557, + "grad_norm": 0.08464474231004715, + "learning_rate": 1.8962321229043322e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25390 + }, + { + "epoch": 7.390165842304335, + "flow_matching_loss": 0.012981378706172109, + "grad_norm": 0.08367118239402771, + "learning_rate": 1.8881931035536943e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25400 + }, + { + "epoch": 7.393075356415479, + "flow_matching_loss": 0.015563595201820135, + "grad_norm": 0.1107674390077591, + "learning_rate": 1.8801700162847952e-06, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25410 + }, + { + "epoch": 7.395984870526622, + "flow_matching_loss": 0.01585817662999034, + "grad_norm": 0.0889921560883522, + "learning_rate": 1.87216287084645e-06, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25420 + }, + { + "epoch": 7.3988943846377655, + "flow_matching_loss": 0.013393113436177374, + "grad_norm": 0.08265678584575653, + "learning_rate": 1.8641716769680895e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25430 + }, + { + "epoch": 7.401803898748909, + "flow_matching_loss": 0.01340421037748456, + "grad_norm": 0.08415795117616653, + "learning_rate": 1.8561964443597662e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25440 + }, + { + "epoch": 7.404713412860052, + "flow_matching_loss": 0.015283449925482273, + "grad_norm": 0.1022995263338089, + "learning_rate": 1.848237182712147e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25450 + }, + { + "epoch": 7.407622926971196, + "flow_matching_loss": 0.012013914436101914, + "grad_norm": 0.08412887901067734, + "learning_rate": 1.8402939016964788e-06, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25460 + }, + { + "epoch": 7.410532441082339, + "flow_matching_loss": 0.015550176985561848, + "grad_norm": 0.08224915713071823, + "learning_rate": 1.832366610964601e-06, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25470 + }, + { + "epoch": 7.413441955193482, + "flow_matching_loss": 0.015298072248697281, + "grad_norm": 0.08247583359479904, + "learning_rate": 1.8244553201489166e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25480 + }, + { + "epoch": 7.416351469304626, + "flow_matching_loss": 0.014674827456474304, + "grad_norm": 0.06594474613666534, + "learning_rate": 1.8165600388623882e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25490 + }, + { + "epoch": 7.41926098341577, + "flow_matching_loss": 0.01601538322865963, + "grad_norm": 0.11508312076330185, + "learning_rate": 1.8086807766985331e-06, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25500 + }, + { + "epoch": 7.422170497526913, + "flow_matching_loss": 0.014366422034800052, + "grad_norm": 0.12571381032466888, + "learning_rate": 1.8008175432313946e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25510 + }, + { + "epoch": 7.425080011638056, + "flow_matching_loss": 0.014491050876677037, + "grad_norm": 0.08747198432683945, + "learning_rate": 1.7929703480155429e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25520 + }, + { + "epoch": 7.4279895257492, + "flow_matching_loss": 0.014611645136028528, + "grad_norm": 0.0880393385887146, + "learning_rate": 1.7851392005860595e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25530 + }, + { + "epoch": 7.430899039860344, + "flow_matching_loss": 0.014907400123775005, + "grad_norm": 0.09487324953079224, + "learning_rate": 1.7773241104585243e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25540 + }, + { + "epoch": 7.433808553971486, + "flow_matching_loss": 0.015122691821306943, + "grad_norm": 0.09806613624095917, + "learning_rate": 1.7695250871290108e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25550 + }, + { + "epoch": 7.43671806808263, + "flow_matching_loss": 0.015159394592046738, + "grad_norm": 0.15396586060523987, + "learning_rate": 1.761742140074074e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25560 + }, + { + "epoch": 7.439627582193774, + "flow_matching_loss": 0.015621849708259106, + "grad_norm": 0.09534913301467896, + "learning_rate": 1.7539752787507208e-06, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25570 + }, + { + "epoch": 7.442537096304917, + "flow_matching_loss": 0.014497640077024698, + "grad_norm": 0.11840719729661942, + "learning_rate": 1.7462245125964233e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25580 + }, + { + "epoch": 7.44544661041606, + "flow_matching_loss": 0.016483363322913648, + "grad_norm": 0.06121818348765373, + "learning_rate": 1.7384898510290931e-06, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25590 + }, + { + "epoch": 7.448356124527204, + "flow_matching_loss": 0.015397515147924423, + "grad_norm": 0.07807585597038269, + "learning_rate": 1.730771303447069e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25600 + }, + { + "epoch": 7.451265638638348, + "flow_matching_loss": 0.014273709710687399, + "grad_norm": 0.10457358509302139, + "learning_rate": 1.7230688792291204e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25610 + }, + { + "epoch": 7.454175152749491, + "flow_matching_loss": 0.013734094984829425, + "grad_norm": 0.07309585064649582, + "learning_rate": 1.7153825877344175e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25620 + }, + { + "epoch": 7.457084666860634, + "flow_matching_loss": 0.014791169296950102, + "grad_norm": 0.13388752937316895, + "learning_rate": 1.7077124383025278e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25630 + }, + { + "epoch": 7.459994180971778, + "flow_matching_loss": 0.012770591117441655, + "grad_norm": 0.0687701478600502, + "learning_rate": 1.7000584402534058e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25640 + }, + { + "epoch": 7.462903695082921, + "flow_matching_loss": 0.015157066378742456, + "grad_norm": 0.0802060216665268, + "learning_rate": 1.6924206028873774e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25650 + }, + { + "epoch": 7.4658132091940645, + "flow_matching_loss": 0.013230806216597557, + "grad_norm": 0.08851282298564911, + "learning_rate": 1.6847989354851434e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25660 + }, + { + "epoch": 7.468722723305208, + "flow_matching_loss": 0.014478008542209864, + "grad_norm": 0.10913188010454178, + "learning_rate": 1.6771934473077443e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25670 + }, + { + "epoch": 7.471632237416351, + "flow_matching_loss": 0.015343158598989248, + "grad_norm": 0.09487875550985336, + "learning_rate": 1.669604147596563e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25680 + }, + { + "epoch": 7.474541751527495, + "flow_matching_loss": 0.014167239703238011, + "grad_norm": 0.09464231133460999, + "learning_rate": 1.662031045573319e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25690 + }, + { + "epoch": 7.4774512656386385, + "flow_matching_loss": 0.014590151887387038, + "grad_norm": 0.07055854797363281, + "learning_rate": 1.6544741504400402e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25700 + }, + { + "epoch": 7.480360779749782, + "flow_matching_loss": 0.015095755364745856, + "grad_norm": 0.1715504676103592, + "learning_rate": 1.6469334713790712e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25710 + }, + { + "epoch": 7.483270293860925, + "flow_matching_loss": 0.014109355956315994, + "grad_norm": 0.07374084740877151, + "learning_rate": 1.639409017553048e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25720 + }, + { + "epoch": 7.486179807972069, + "flow_matching_loss": 0.012406782805919647, + "grad_norm": 0.07889340072870255, + "learning_rate": 1.6319007981048884e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25730 + }, + { + "epoch": 7.489089322083212, + "flow_matching_loss": 0.015388664416968822, + "grad_norm": 0.09356334805488586, + "learning_rate": 1.6244088221577896e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25740 + }, + { + "epoch": 7.491998836194355, + "flow_matching_loss": 0.013121352437883616, + "grad_norm": 0.07786783576011658, + "learning_rate": 1.6169330988152081e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25750 + }, + { + "epoch": 7.494908350305499, + "flow_matching_loss": 0.013016394479200244, + "grad_norm": 0.0962240993976593, + "learning_rate": 1.6094736371608475e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25760 + }, + { + "epoch": 7.497817864416643, + "flow_matching_loss": 0.014045420940965414, + "grad_norm": 0.10364485532045364, + "learning_rate": 1.602030446258666e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25770 + }, + { + "epoch": 7.5007273785277855, + "flow_matching_loss": 0.01426596799865365, + "grad_norm": 0.09808256477117538, + "learning_rate": 1.5946035351528388e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25780 + }, + { + "epoch": 7.503636892638929, + "flow_matching_loss": 0.016736858151853086, + "grad_norm": 0.09613136202096939, + "learning_rate": 1.5871929128677621e-06, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25790 + }, + { + "epoch": 7.506546406750073, + "flow_matching_loss": 0.013927958067506551, + "grad_norm": 0.08262886106967926, + "learning_rate": 1.5797985884080412e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25800 + }, + { + "epoch": 7.509455920861216, + "flow_matching_loss": 0.013803528295829893, + "grad_norm": 0.08583671599626541, + "learning_rate": 1.572420570758476e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25810 + }, + { + "epoch": 7.512365434972359, + "flow_matching_loss": 0.014288275316357613, + "grad_norm": 0.08124036341905594, + "learning_rate": 1.5650588688840567e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25820 + }, + { + "epoch": 7.515274949083503, + "flow_matching_loss": 0.013209140580147504, + "grad_norm": 0.08170871436595917, + "learning_rate": 1.5577134917299413e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25830 + }, + { + "epoch": 7.518184463194647, + "flow_matching_loss": 0.015455606579780578, + "grad_norm": 0.10286328941583633, + "learning_rate": 1.5503844482214618e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25840 + }, + { + "epoch": 7.52109397730579, + "flow_matching_loss": 0.013236380321905017, + "grad_norm": 0.08329058438539505, + "learning_rate": 1.5430717472640932e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25850 + }, + { + "epoch": 7.524003491416933, + "flow_matching_loss": 0.01496637249365449, + "grad_norm": 0.07304197549819946, + "learning_rate": 1.5357753977434541e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25860 + }, + { + "epoch": 7.526913005528077, + "flow_matching_loss": 0.015527189429849386, + "grad_norm": 0.1027088314294815, + "learning_rate": 1.5284954085253043e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25870 + }, + { + "epoch": 7.52982251963922, + "flow_matching_loss": 0.01336254170164466, + "grad_norm": 0.12249969691038132, + "learning_rate": 1.5212317884555115e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25880 + }, + { + "epoch": 7.532732033750364, + "flow_matching_loss": 0.015465813130140305, + "grad_norm": 0.11580519378185272, + "learning_rate": 1.5139845463600605e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25890 + }, + { + "epoch": 7.535641547861507, + "flow_matching_loss": 0.015508448611944914, + "grad_norm": 0.09402089565992355, + "learning_rate": 1.5067536910450352e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25900 + }, + { + "epoch": 7.538551061972651, + "flow_matching_loss": 0.013945201132446527, + "grad_norm": 0.10661599785089493, + "learning_rate": 1.4995392312966044e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25910 + }, + { + "epoch": 7.541460576083794, + "flow_matching_loss": 0.013377075828611851, + "grad_norm": 0.11245696991682053, + "learning_rate": 1.492341175881014e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25920 + }, + { + "epoch": 7.5443700901949375, + "flow_matching_loss": 0.012829165905714035, + "grad_norm": 0.07892875373363495, + "learning_rate": 1.4851595335445895e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25930 + }, + { + "epoch": 7.547279604306081, + "flow_matching_loss": 0.01431407118216157, + "grad_norm": 0.09936215728521347, + "learning_rate": 1.4779943130136953e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25940 + }, + { + "epoch": 7.550189118417224, + "flow_matching_loss": 0.014183853939175605, + "grad_norm": 0.08584640175104141, + "learning_rate": 1.4708455229947537e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25950 + }, + { + "epoch": 7.553098632528368, + "flow_matching_loss": 0.013170688413083553, + "grad_norm": 0.11614738404750824, + "learning_rate": 1.4637131721742148e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25960 + }, + { + "epoch": 7.5560081466395115, + "flow_matching_loss": 0.014658765308558941, + "grad_norm": 0.08599349111318588, + "learning_rate": 1.4565972692185626e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25970 + }, + { + "epoch": 7.558917660750654, + "flow_matching_loss": 0.0164495300501585, + "grad_norm": 0.09029784053564072, + "learning_rate": 1.4494978227742855e-06, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25980 + }, + { + "epoch": 7.561827174861798, + "flow_matching_loss": 0.016184354666620493, + "grad_norm": 0.07734526693820953, + "learning_rate": 1.442414841467885e-06, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25990 + }, + { + "epoch": 7.564736688972942, + "flow_matching_loss": 0.013729310315102338, + "grad_norm": 0.10368485748767853, + "learning_rate": 1.4353483339058498e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26000 + }, + { + "epoch": 7.5676462030840845, + "flow_matching_loss": 0.01597363334149122, + "grad_norm": 0.07272447645664215, + "learning_rate": 1.4282983086746498e-06, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26010 + }, + { + "epoch": 7.570555717195228, + "flow_matching_loss": 0.014344613719731569, + "grad_norm": 0.11480733007192612, + "learning_rate": 1.4212647743407325e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26020 + }, + { + "epoch": 7.573465231306372, + "flow_matching_loss": 0.014082200545817613, + "grad_norm": 0.0768648087978363, + "learning_rate": 1.4142477394505065e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26030 + }, + { + "epoch": 7.576374745417516, + "flow_matching_loss": 0.015683251526206733, + "grad_norm": 0.10372936725616455, + "learning_rate": 1.4072472125303299e-06, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26040 + }, + { + "epoch": 7.5792842595286585, + "flow_matching_loss": 0.013438956625759602, + "grad_norm": 0.07471299916505814, + "learning_rate": 1.4002632020865031e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26050 + }, + { + "epoch": 7.582193773639802, + "flow_matching_loss": 0.016687551513314247, + "grad_norm": 0.09393119812011719, + "learning_rate": 1.393295716605258e-06, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26060 + }, + { + "epoch": 7.585103287750946, + "flow_matching_loss": 0.013121997192502021, + "grad_norm": 0.09687773138284683, + "learning_rate": 1.386344764552742e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26070 + }, + { + "epoch": 7.588012801862089, + "flow_matching_loss": 0.01440229844301939, + "grad_norm": 0.08882716298103333, + "learning_rate": 1.3794103543750246e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26080 + }, + { + "epoch": 7.590922315973232, + "flow_matching_loss": 0.01256960555911064, + "grad_norm": 0.10186247527599335, + "learning_rate": 1.3724924944980637e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26090 + }, + { + "epoch": 7.593831830084376, + "flow_matching_loss": 0.013002964667975903, + "grad_norm": 0.09939970821142197, + "learning_rate": 1.365591193327712e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26100 + }, + { + "epoch": 7.59674134419552, + "flow_matching_loss": 0.016229519341140985, + "grad_norm": 0.08316872268915176, + "learning_rate": 1.3587064592496984e-06, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26110 + }, + { + "epoch": 7.599650858306663, + "flow_matching_loss": 0.01667974768206477, + "grad_norm": 0.08944620192050934, + "learning_rate": 1.3518383006296292e-06, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26120 + }, + { + "epoch": 7.602560372417806, + "flow_matching_loss": 0.012816027086228133, + "grad_norm": 0.08499332517385483, + "learning_rate": 1.344986725812959e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26130 + }, + { + "epoch": 7.605469886528949, + "flow_matching_loss": 0.013902797643095256, + "grad_norm": 0.12604153156280518, + "learning_rate": 1.3381517431250046e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26140 + }, + { + "epoch": 7.608379400640093, + "flow_matching_loss": 0.014410585863515735, + "grad_norm": 0.09451030194759369, + "learning_rate": 1.3313333608709095e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26150 + }, + { + "epoch": 7.611288914751237, + "flow_matching_loss": 0.013486937526613474, + "grad_norm": 0.09920443594455719, + "learning_rate": 1.3245315873356528e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26160 + }, + { + "epoch": 7.61419842886238, + "flow_matching_loss": 0.01414848817512393, + "grad_norm": 0.08340596407651901, + "learning_rate": 1.3177464307840304e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26170 + }, + { + "epoch": 7.617107942973523, + "flow_matching_loss": 0.012670376151800156, + "grad_norm": 0.08979196846485138, + "learning_rate": 1.310977899460642e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26180 + }, + { + "epoch": 7.620017457084667, + "flow_matching_loss": 0.014435863262042404, + "grad_norm": 0.1018541231751442, + "learning_rate": 1.3042260015899006e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26190 + }, + { + "epoch": 7.6229269711958105, + "flow_matching_loss": 0.01488469410687685, + "grad_norm": 0.17944790422916412, + "learning_rate": 1.2974907453759948e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26200 + }, + { + "epoch": 7.625836485306953, + "flow_matching_loss": 0.013551262952387334, + "grad_norm": 0.0849783644080162, + "learning_rate": 1.2907721390028937e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26210 + }, + { + "epoch": 7.628745999418097, + "flow_matching_loss": 0.014603732619434595, + "grad_norm": 0.08946771919727325, + "learning_rate": 1.2840701906343417e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26220 + }, + { + "epoch": 7.631655513529241, + "flow_matching_loss": 0.014756117854267359, + "grad_norm": 0.07071249186992645, + "learning_rate": 1.2773849084138312e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26230 + }, + { + "epoch": 7.6345650276403845, + "flow_matching_loss": 0.013744156528264283, + "grad_norm": 0.09330040961503983, + "learning_rate": 1.2707163004646177e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26240 + }, + { + "epoch": 7.637474541751527, + "flow_matching_loss": 0.012940215039998293, + "grad_norm": 0.07672855257987976, + "learning_rate": 1.2640643748896879e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26250 + }, + { + "epoch": 7.640384055862671, + "flow_matching_loss": 0.01688549844548106, + "grad_norm": 0.0716031864285469, + "learning_rate": 1.257429139771753e-06, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26260 + }, + { + "epoch": 7.643293569973815, + "flow_matching_loss": 0.012862515728920698, + "grad_norm": 0.07436873018741608, + "learning_rate": 1.2508106031732553e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26270 + }, + { + "epoch": 7.6462030840849575, + "flow_matching_loss": 0.014593726862221957, + "grad_norm": 0.09999596327543259, + "learning_rate": 1.2442087731363366e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26280 + }, + { + "epoch": 7.649112598196101, + "flow_matching_loss": 0.01365247517824173, + "grad_norm": 0.09416908025741577, + "learning_rate": 1.237623657682847e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26290 + }, + { + "epoch": 7.652022112307245, + "flow_matching_loss": 0.012104412261396646, + "grad_norm": 0.06528446078300476, + "learning_rate": 1.2310552648143208e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26300 + }, + { + "epoch": 7.654931626418388, + "flow_matching_loss": 0.016590338479727507, + "grad_norm": 0.06850750744342804, + "learning_rate": 1.2245036025119738e-06, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26310 + }, + { + "epoch": 7.6578411405295315, + "flow_matching_loss": 0.013681519869714975, + "grad_norm": 0.07243537902832031, + "learning_rate": 1.2179686787366962e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26320 + }, + { + "epoch": 7.660750654640675, + "flow_matching_loss": 0.015494188293814659, + "grad_norm": 0.06914651393890381, + "learning_rate": 1.2114505014290329e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26330 + }, + { + "epoch": 7.663660168751818, + "flow_matching_loss": 0.017981476336717605, + "grad_norm": 0.11846418678760529, + "learning_rate": 1.2049490785091833e-06, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26340 + }, + { + "epoch": 7.666569682862962, + "flow_matching_loss": 0.0134786244481802, + "grad_norm": 0.08003823459148407, + "learning_rate": 1.1984644178769949e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26350 + }, + { + "epoch": 7.669479196974105, + "flow_matching_loss": 0.011728116124868394, + "grad_norm": 0.08736734092235565, + "learning_rate": 1.1919965274119381e-06, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26360 + }, + { + "epoch": 7.672388711085249, + "flow_matching_loss": 0.014800511673092842, + "grad_norm": 0.08602020144462585, + "learning_rate": 1.18554541497311e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26370 + }, + { + "epoch": 7.675298225196392, + "flow_matching_loss": 0.015539678186178208, + "grad_norm": 0.11496956646442413, + "learning_rate": 1.1791110883992213e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26380 + }, + { + "epoch": 7.678207739307536, + "flow_matching_loss": 0.016519681829959155, + "grad_norm": 0.13769543170928955, + "learning_rate": 1.1726935555085817e-06, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26390 + }, + { + "epoch": 7.681117253418679, + "flow_matching_loss": 0.015384826855733991, + "grad_norm": 0.1136244535446167, + "learning_rate": 1.166292824099105e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26400 + }, + { + "epoch": 7.684026767529822, + "flow_matching_loss": 0.012879026588052511, + "grad_norm": 0.08076633512973785, + "learning_rate": 1.1599089019482768e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26410 + }, + { + "epoch": 7.686936281640966, + "flow_matching_loss": 0.01271235542371869, + "grad_norm": 0.13672170042991638, + "learning_rate": 1.1535417968131718e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26420 + }, + { + "epoch": 7.68984579575211, + "flow_matching_loss": 0.015005742106586695, + "grad_norm": 0.09101788699626923, + "learning_rate": 1.147191516430418e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26430 + }, + { + "epoch": 7.692755309863253, + "flow_matching_loss": 0.014742507040500641, + "grad_norm": 0.08846329152584076, + "learning_rate": 1.140858068516205e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26440 + }, + { + "epoch": 7.695664823974396, + "flow_matching_loss": 0.01288983402773738, + "grad_norm": 0.08208402246236801, + "learning_rate": 1.1345414607662736e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26450 + }, + { + "epoch": 7.69857433808554, + "flow_matching_loss": 0.012644060607999563, + "grad_norm": 0.06444975733757019, + "learning_rate": 1.1282417008558937e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26460 + }, + { + "epoch": 7.7014838521966835, + "flow_matching_loss": 0.0148300401866436, + "grad_norm": 0.11039318144321442, + "learning_rate": 1.1219587964398674e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26470 + }, + { + "epoch": 7.704393366307826, + "flow_matching_loss": 0.014631668478250504, + "grad_norm": 0.09907735884189606, + "learning_rate": 1.1156927551525198e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26480 + }, + { + "epoch": 7.70730288041897, + "flow_matching_loss": 0.014226983953267335, + "grad_norm": 0.09774810075759888, + "learning_rate": 1.109443584607675e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26490 + }, + { + "epoch": 7.710212394530114, + "flow_matching_loss": 0.015014115162193776, + "grad_norm": 0.0890064537525177, + "learning_rate": 1.1032112923986708e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26500 + }, + { + "epoch": 7.713121908641257, + "flow_matching_loss": 0.01567679541185498, + "grad_norm": 0.11297817528247833, + "learning_rate": 1.0969958860983299e-06, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26510 + }, + { + "epoch": 7.7160314227524, + "flow_matching_loss": 0.012857725657522679, + "grad_norm": 0.10039937496185303, + "learning_rate": 1.0907973732589537e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26520 + }, + { + "epoch": 7.718940936863544, + "flow_matching_loss": 0.015613959170877934, + "grad_norm": 0.07995288819074631, + "learning_rate": 1.0846157614123215e-06, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26530 + }, + { + "epoch": 7.721850450974687, + "flow_matching_loss": 0.013830576464533806, + "grad_norm": 0.08299081772565842, + "learning_rate": 1.078451058069671e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26540 + }, + { + "epoch": 7.7247599650858305, + "flow_matching_loss": 0.014550984837114811, + "grad_norm": 0.11256161332130432, + "learning_rate": 1.0723032707217056e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26550 + }, + { + "epoch": 7.727669479196974, + "flow_matching_loss": 0.013725569797679782, + "grad_norm": 0.11044242978096008, + "learning_rate": 1.066172406838558e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26560 + }, + { + "epoch": 7.730578993308118, + "flow_matching_loss": 0.01863508513197303, + "grad_norm": 0.07841916382312775, + "learning_rate": 1.0600584738698127e-06, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26570 + }, + { + "epoch": 7.733488507419261, + "flow_matching_loss": 0.014974233508110047, + "grad_norm": 0.09103778004646301, + "learning_rate": 1.0539614792444735e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26580 + }, + { + "epoch": 7.7363980215304045, + "flow_matching_loss": 0.01288365083746612, + "grad_norm": 0.06741131097078323, + "learning_rate": 1.0478814303709627e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26590 + }, + { + "epoch": 7.739307535641548, + "flow_matching_loss": 0.016357136191800237, + "grad_norm": 0.11258289217948914, + "learning_rate": 1.041818334637109e-06, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26600 + }, + { + "epoch": 7.742217049752691, + "flow_matching_loss": 0.013823993131518364, + "grad_norm": 0.08826430141925812, + "learning_rate": 1.0357721994101526e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26610 + }, + { + "epoch": 7.745126563863835, + "flow_matching_loss": 0.012800157628953457, + "grad_norm": 0.12401054054498672, + "learning_rate": 1.0297430320367135e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26620 + }, + { + "epoch": 7.748036077974978, + "flow_matching_loss": 0.014802435692399741, + "grad_norm": 0.10554634779691696, + "learning_rate": 1.0237308398428014e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26630 + }, + { + "epoch": 7.750945592086122, + "flow_matching_loss": 0.014699221309274434, + "grad_norm": 0.06795179843902588, + "learning_rate": 1.0177356301337925e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26640 + }, + { + "epoch": 7.753855106197265, + "flow_matching_loss": 0.014088507369160652, + "grad_norm": 0.1082763597369194, + "learning_rate": 1.0117574101944348e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26650 + }, + { + "epoch": 7.756764620308409, + "flow_matching_loss": 0.014182930532842874, + "grad_norm": 0.07277444750070572, + "learning_rate": 1.0057961872888293e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26660 + }, + { + "epoch": 7.7596741344195515, + "flow_matching_loss": 0.013717705942690372, + "grad_norm": 0.08637826889753342, + "learning_rate": 9.998519686604235e-07, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26670 + }, + { + "epoch": 7.762583648530695, + "flow_matching_loss": 0.013829255662858485, + "grad_norm": 0.08695773780345917, + "learning_rate": 9.939247615320047e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26680 + }, + { + "epoch": 7.765493162641839, + "flow_matching_loss": 0.014398554060608148, + "grad_norm": 0.10284141451120377, + "learning_rate": 9.880145731056834e-07, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26690 + }, + { + "epoch": 7.768402676752983, + "flow_matching_loss": 0.015509345661848783, + "grad_norm": 0.08637217432260513, + "learning_rate": 9.821214105629034e-07, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26700 + }, + { + "epoch": 7.771312190864125, + "flow_matching_loss": 0.014553907513618469, + "grad_norm": 0.08115831017494202, + "learning_rate": 9.762452810644068e-07, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26710 + }, + { + "epoch": 7.774221704975269, + "flow_matching_loss": 0.01463586469180882, + "grad_norm": 0.0805380642414093, + "learning_rate": 9.703861917502487e-07, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26720 + }, + { + "epoch": 7.777131219086413, + "flow_matching_loss": 0.017284551355987787, + "grad_norm": 0.1376451551914215, + "learning_rate": 9.645441497397744e-07, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26730 + }, + { + "epoch": 7.780040733197556, + "flow_matching_loss": 0.013920481409877538, + "grad_norm": 0.08100453019142151, + "learning_rate": 9.587191621316154e-07, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26740 + }, + { + "epoch": 7.782950247308699, + "flow_matching_loss": 0.012997050024569034, + "grad_norm": 0.0718446895480156, + "learning_rate": 9.529112360036802e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26750 + }, + { + "epoch": 7.785859761419843, + "flow_matching_loss": 0.015607160050421953, + "grad_norm": 0.10284649580717087, + "learning_rate": 9.471203784131449e-07, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26760 + }, + { + "epoch": 7.788769275530987, + "flow_matching_loss": 0.014275641879066825, + "grad_norm": 0.05955590307712555, + "learning_rate": 9.41346596396453e-07, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26770 + }, + { + "epoch": 7.79167878964213, + "flow_matching_loss": 0.01401993166655302, + "grad_norm": 0.0697559267282486, + "learning_rate": 9.355898969692889e-07, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26780 + }, + { + "epoch": 7.794588303753273, + "flow_matching_loss": 0.016225067898631096, + "grad_norm": 0.09603418409824371, + "learning_rate": 9.29850287126589e-07, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26790 + }, + { + "epoch": 7.797497817864417, + "flow_matching_loss": 0.01511576510965824, + "grad_norm": 0.10034032166004181, + "learning_rate": 9.24127773842518e-07, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26800 + }, + { + "epoch": 7.80040733197556, + "flow_matching_loss": 0.013024144899100065, + "grad_norm": 0.08808476477861404, + "learning_rate": 9.18422364070467e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26810 + }, + { + "epoch": 7.8033168460867035, + "flow_matching_loss": 0.013718207646161319, + "grad_norm": 0.11380468308925629, + "learning_rate": 9.127340647430538e-07, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26820 + }, + { + "epoch": 7.806226360197847, + "flow_matching_loss": 0.014902166742831468, + "grad_norm": 0.0633590817451477, + "learning_rate": 9.070628827720929e-07, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26830 + }, + { + "epoch": 7.80913587430899, + "flow_matching_loss": 0.012279572756960988, + "grad_norm": 0.07184822112321854, + "learning_rate": 9.014088250486052e-07, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26840 + }, + { + "epoch": 7.812045388420134, + "flow_matching_loss": 0.01436708690598607, + "grad_norm": 0.1172545775771141, + "learning_rate": 8.95771898442807e-07, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26850 + }, + { + "epoch": 7.8149549025312774, + "flow_matching_loss": 0.015115803480148316, + "grad_norm": 0.08704890310764313, + "learning_rate": 8.90152109804091e-07, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26860 + }, + { + "epoch": 7.81786441664242, + "flow_matching_loss": 0.013269464671611785, + "grad_norm": 0.11073615401983261, + "learning_rate": 8.845494659610365e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26870 + }, + { + "epoch": 7.820773930753564, + "flow_matching_loss": 0.013734037149697542, + "grad_norm": 0.09884457290172577, + "learning_rate": 8.789639737213812e-07, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26880 + }, + { + "epoch": 7.823683444864708, + "flow_matching_loss": 0.0133093754760921, + "grad_norm": 0.07919954508543015, + "learning_rate": 8.733956398720244e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26890 + }, + { + "epoch": 7.826592958975851, + "flow_matching_loss": 0.01327903252094984, + "grad_norm": 0.06709621846675873, + "learning_rate": 8.678444711790157e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26900 + }, + { + "epoch": 7.829502473086994, + "flow_matching_loss": 0.013865892309695482, + "grad_norm": 0.08620923012495041, + "learning_rate": 8.623104743875493e-07, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26910 + }, + { + "epoch": 7.832411987198138, + "flow_matching_loss": 0.013035459723323584, + "grad_norm": 0.09651099145412445, + "learning_rate": 8.567936562219547e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26920 + }, + { + "epoch": 7.835321501309282, + "flow_matching_loss": 0.014050630573183297, + "grad_norm": 0.088788703083992, + "learning_rate": 8.512940233856847e-07, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26930 + }, + { + "epoch": 7.838231015420424, + "flow_matching_loss": 0.012157002277672291, + "grad_norm": 0.09055937081575394, + "learning_rate": 8.458115825613122e-07, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26940 + }, + { + "epoch": 7.841140529531568, + "flow_matching_loss": 0.013459870126098394, + "grad_norm": 0.08899789303541183, + "learning_rate": 8.403463404105199e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26950 + }, + { + "epoch": 7.844050043642712, + "flow_matching_loss": 0.014839519700035452, + "grad_norm": 0.10551975667476654, + "learning_rate": 8.348983035740926e-07, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26960 + }, + { + "epoch": 7.846959557753856, + "flow_matching_loss": 0.011690373811870814, + "grad_norm": 0.07346683740615845, + "learning_rate": 8.294674786719064e-07, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26970 + }, + { + "epoch": 7.849869071864998, + "flow_matching_loss": 0.015125592704862356, + "grad_norm": 0.08091080188751221, + "learning_rate": 8.240538723029295e-07, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26980 + }, + { + "epoch": 7.852778585976142, + "flow_matching_loss": 0.014446925278753043, + "grad_norm": 0.11006640642881393, + "learning_rate": 8.186574910451983e-07, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26990 + }, + { + "epoch": 7.855688100087286, + "flow_matching_loss": 0.014847372379153966, + "grad_norm": 0.07608263939619064, + "learning_rate": 8.132783414558326e-07, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27000 + }, + { + "epoch": 7.858597614198429, + "flow_matching_loss": 0.013304058089852333, + "grad_norm": 0.07424747943878174, + "learning_rate": 8.079164300710023e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27010 + }, + { + "epoch": 7.861507128309572, + "flow_matching_loss": 0.013989751599729061, + "grad_norm": 0.06410663574934006, + "learning_rate": 8.025717634059326e-07, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27020 + }, + { + "epoch": 7.864416642420716, + "flow_matching_loss": 0.015000274777412415, + "grad_norm": 0.0780554786324501, + "learning_rate": 7.972443479549035e-07, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27030 + }, + { + "epoch": 7.867326156531859, + "flow_matching_loss": 0.014619177579879761, + "grad_norm": 0.09003885835409164, + "learning_rate": 7.919341901912236e-07, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27040 + }, + { + "epoch": 7.870235670643003, + "flow_matching_loss": 0.014562879595905543, + "grad_norm": 0.12149740755558014, + "learning_rate": 7.866412965672365e-07, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27050 + }, + { + "epoch": 7.873145184754146, + "flow_matching_loss": 0.013813414983451367, + "grad_norm": 0.08700816333293915, + "learning_rate": 7.813656735143044e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27060 + }, + { + "epoch": 7.876054698865289, + "flow_matching_loss": 0.0165361643768847, + "grad_norm": 0.11035886406898499, + "learning_rate": 7.761073274428076e-07, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27070 + }, + { + "epoch": 7.878964212976433, + "flow_matching_loss": 0.01615977780893445, + "grad_norm": 0.08272876590490341, + "learning_rate": 7.708662647421355e-07, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27080 + }, + { + "epoch": 7.8818737270875765, + "flow_matching_loss": 0.013443589769303798, + "grad_norm": 0.08264555037021637, + "learning_rate": 7.65642491780672e-07, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27090 + }, + { + "epoch": 7.88478324119872, + "flow_matching_loss": 0.01403747098520398, + "grad_norm": 0.0833025872707367, + "learning_rate": 7.604360149057916e-07, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27100 + }, + { + "epoch": 7.887692755309863, + "flow_matching_loss": 0.01376321678981185, + "grad_norm": 0.12671393156051636, + "learning_rate": 7.552468404438568e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27110 + }, + { + "epoch": 7.890602269421007, + "flow_matching_loss": 0.013517692405730485, + "grad_norm": 0.07264502346515656, + "learning_rate": 7.500749747002022e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27120 + }, + { + "epoch": 7.89351178353215, + "flow_matching_loss": 0.014541818574070931, + "grad_norm": 0.07544995844364166, + "learning_rate": 7.449204239591345e-07, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27130 + }, + { + "epoch": 7.896421297643293, + "flow_matching_loss": 0.013569011399522424, + "grad_norm": 0.09822897613048553, + "learning_rate": 7.397831944839167e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27140 + }, + { + "epoch": 7.899330811754437, + "flow_matching_loss": 0.01249367455020547, + "grad_norm": 0.06591733545064926, + "learning_rate": 7.346632925167706e-07, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27150 + }, + { + "epoch": 7.902240325865581, + "flow_matching_loss": 0.012637842353433371, + "grad_norm": 0.08468794822692871, + "learning_rate": 7.295607242788566e-07, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27160 + }, + { + "epoch": 7.905149839976724, + "flow_matching_loss": 0.013640935439616442, + "grad_norm": 0.07588551193475723, + "learning_rate": 7.244754959702765e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27170 + }, + { + "epoch": 7.908059354087867, + "flow_matching_loss": 0.01558383209630847, + "grad_norm": 0.11370569467544556, + "learning_rate": 7.194076137700628e-07, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27180 + }, + { + "epoch": 7.910968868199011, + "flow_matching_loss": 0.014143766928464174, + "grad_norm": 0.09403953701257706, + "learning_rate": 7.143570838361696e-07, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27190 + }, + { + "epoch": 7.913878382310154, + "flow_matching_loss": 0.01455815164372325, + "grad_norm": 0.08013255894184113, + "learning_rate": 7.093239123054696e-07, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27200 + }, + { + "epoch": 7.916787896421297, + "flow_matching_loss": 0.015230774041265249, + "grad_norm": 0.09509396553039551, + "learning_rate": 7.043081052937389e-07, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27210 + }, + { + "epoch": 7.919697410532441, + "flow_matching_loss": 0.013414326217025518, + "grad_norm": 0.08665838092565536, + "learning_rate": 6.993096688956558e-07, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27220 + }, + { + "epoch": 7.922606924643585, + "flow_matching_loss": 0.013523074332624674, + "grad_norm": 0.09980209916830063, + "learning_rate": 6.943286091847884e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27230 + }, + { + "epoch": 7.925516438754728, + "flow_matching_loss": 0.014195089880377054, + "grad_norm": 0.08111656457185745, + "learning_rate": 6.89364932213602e-07, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27240 + }, + { + "epoch": 7.928425952865871, + "flow_matching_loss": 0.014992231596261263, + "grad_norm": 0.09460663795471191, + "learning_rate": 6.844186440134253e-07, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27250 + }, + { + "epoch": 7.931335466977015, + "flow_matching_loss": 0.015518494881689548, + "grad_norm": 0.06977856904268265, + "learning_rate": 6.794897505944692e-07, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27260 + }, + { + "epoch": 7.934244981088158, + "flow_matching_loss": 0.012946877162903547, + "grad_norm": 0.0778139978647232, + "learning_rate": 6.745782579457993e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27270 + }, + { + "epoch": 7.937154495199302, + "flow_matching_loss": 0.014711342053487897, + "grad_norm": 0.06675028055906296, + "learning_rate": 6.696841720353453e-07, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27280 + }, + { + "epoch": 7.940064009310445, + "flow_matching_loss": 0.014336035214364528, + "grad_norm": 0.06870561838150024, + "learning_rate": 6.648074988098834e-07, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27290 + }, + { + "epoch": 7.942973523421589, + "flow_matching_loss": 0.014212092850357294, + "grad_norm": 0.08791530877351761, + "learning_rate": 6.599482441950322e-07, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27300 + }, + { + "epoch": 7.945883037532732, + "flow_matching_loss": 0.014652222860604525, + "grad_norm": 0.07506697624921799, + "learning_rate": 6.551064140952417e-07, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27310 + }, + { + "epoch": 7.9487925516438755, + "flow_matching_loss": 0.014635540265589952, + "grad_norm": 0.09945543110370636, + "learning_rate": 6.502820143937943e-07, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27320 + }, + { + "epoch": 7.951702065755019, + "flow_matching_loss": 0.013941418845206499, + "grad_norm": 0.09140612930059433, + "learning_rate": 6.454750509527874e-07, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27330 + }, + { + "epoch": 7.954611579866162, + "flow_matching_loss": 0.014846179075539112, + "grad_norm": 0.055106110870838165, + "learning_rate": 6.406855296131387e-07, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27340 + }, + { + "epoch": 7.957521093977306, + "flow_matching_loss": 0.014060169644653796, + "grad_norm": 0.10889783501625061, + "learning_rate": 6.359134561945662e-07, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27350 + }, + { + "epoch": 7.9604306080884495, + "flow_matching_loss": 0.01321932701393962, + "grad_norm": 0.08292141556739807, + "learning_rate": 6.311588364955883e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27360 + }, + { + "epoch": 7.963340122199592, + "flow_matching_loss": 0.012752023059874773, + "grad_norm": 0.08623439073562622, + "learning_rate": 6.264216762935187e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27370 + }, + { + "epoch": 7.966249636310736, + "flow_matching_loss": 0.014384616166353226, + "grad_norm": 0.09475784003734589, + "learning_rate": 6.21701981344453e-07, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27380 + }, + { + "epoch": 7.96915915042188, + "flow_matching_loss": 0.011815527267754079, + "grad_norm": 0.06265874952077866, + "learning_rate": 6.16999757383262e-07, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27390 + }, + { + "epoch": 7.9720686645330225, + "flow_matching_loss": 0.014273636043071747, + "grad_norm": 0.08070427179336548, + "learning_rate": 6.123150101235952e-07, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27400 + }, + { + "epoch": 7.974978178644166, + "flow_matching_loss": 0.015497570298612118, + "grad_norm": 0.0988779217004776, + "learning_rate": 6.076477452578594e-07, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27410 + }, + { + "epoch": 7.97788769275531, + "flow_matching_loss": 0.012895004265010357, + "grad_norm": 0.0713910311460495, + "learning_rate": 6.029979684572262e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27420 + }, + { + "epoch": 7.980797206866454, + "flow_matching_loss": 0.014675142150372267, + "grad_norm": 0.09220994263887405, + "learning_rate": 5.983656853716097e-07, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27430 + }, + { + "epoch": 7.9837067209775965, + "flow_matching_loss": 0.015200679190456867, + "grad_norm": 0.09250426292419434, + "learning_rate": 5.937509016296689e-07, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27440 + }, + { + "epoch": 7.98661623508874, + "flow_matching_loss": 0.01627507312223315, + "grad_norm": 0.08132074773311615, + "learning_rate": 5.89153622838805e-07, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27450 + }, + { + "epoch": 7.989525749199884, + "flow_matching_loss": 0.015534671209752559, + "grad_norm": 0.08134594559669495, + "learning_rate": 5.845738545851431e-07, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27460 + }, + { + "epoch": 7.992435263311027, + "flow_matching_loss": 0.012810801435261965, + "grad_norm": 0.08360132575035095, + "learning_rate": 5.800116024335334e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27470 + }, + { + "epoch": 7.99534477742217, + "flow_matching_loss": 0.01399339260533452, + "grad_norm": 0.1137876883149147, + "learning_rate": 5.75466871927543e-07, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27480 + }, + { + "epoch": 7.998254291533314, + "flow_matching_loss": 0.01222856342792511, + "grad_norm": 0.053628161549568176, + "learning_rate": 5.709396685894463e-07, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27490 + }, + { + "epoch": 8.001163805644458, + "flow_matching_loss": 0.01230713101103902, + "grad_norm": 0.07380056381225586, + "learning_rate": 5.664299979202264e-07, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27500 + }, + { + "epoch": 8.004073319755602, + "flow_matching_loss": 0.014430622011423111, + "grad_norm": 0.0715230330824852, + "learning_rate": 5.61937865399555e-07, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27510 + }, + { + "epoch": 8.006982833866743, + "flow_matching_loss": 0.016282184794545174, + "grad_norm": 0.1141456589102745, + "learning_rate": 5.574632764857974e-07, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27520 + }, + { + "epoch": 8.009892347977887, + "flow_matching_loss": 0.012921040412038565, + "grad_norm": 0.0835268571972847, + "learning_rate": 5.53006236616001e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27530 + }, + { + "epoch": 8.012801862089031, + "flow_matching_loss": 0.014137991610914469, + "grad_norm": 0.08788561075925827, + "learning_rate": 5.485667512058873e-07, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27540 + }, + { + "epoch": 8.015711376200175, + "flow_matching_loss": 0.013151416089385748, + "grad_norm": 0.07605785876512527, + "learning_rate": 5.441448256498544e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27550 + }, + { + "epoch": 8.018620890311318, + "flow_matching_loss": 0.015708757936954497, + "grad_norm": 0.10121772438287735, + "learning_rate": 5.397404653209559e-07, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27560 + }, + { + "epoch": 8.021530404422462, + "flow_matching_loss": 0.014817082695662976, + "grad_norm": 0.09706418216228485, + "learning_rate": 5.35353675570906e-07, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27570 + }, + { + "epoch": 8.024439918533606, + "flow_matching_loss": 0.011736938310787082, + "grad_norm": 0.07967354357242584, + "learning_rate": 5.309844617300674e-07, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27580 + }, + { + "epoch": 8.027349432644748, + "flow_matching_loss": 0.01396767534315586, + "grad_norm": 0.07872552424669266, + "learning_rate": 5.266328291074468e-07, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27590 + }, + { + "epoch": 8.030258946755891, + "flow_matching_loss": 0.014391190093010663, + "grad_norm": 0.11363880336284637, + "learning_rate": 5.222987829906861e-07, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27600 + }, + { + "epoch": 8.033168460867035, + "flow_matching_loss": 0.01211719112470746, + "grad_norm": 0.0967118963599205, + "learning_rate": 5.179823286460644e-07, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27610 + }, + { + "epoch": 8.036077974978179, + "flow_matching_loss": 0.01505623497068882, + "grad_norm": 0.08515620231628418, + "learning_rate": 5.13683471318478e-07, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27620 + }, + { + "epoch": 8.038987489089322, + "flow_matching_loss": 0.015143895521759987, + "grad_norm": 0.10426545888185501, + "learning_rate": 5.094022162314449e-07, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27630 + }, + { + "epoch": 8.041897003200466, + "flow_matching_loss": 0.015849158726632596, + "grad_norm": 0.09241590648889542, + "learning_rate": 5.051385685870924e-07, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27640 + }, + { + "epoch": 8.044806517311608, + "flow_matching_loss": 0.01393915992230177, + "grad_norm": 0.09204084426164627, + "learning_rate": 5.008925335661546e-07, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27650 + }, + { + "epoch": 8.047716031422752, + "flow_matching_loss": 0.013787961658090353, + "grad_norm": 0.08409593254327774, + "learning_rate": 4.966641163279662e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27660 + }, + { + "epoch": 8.050625545533896, + "flow_matching_loss": 0.013712106179445982, + "grad_norm": 0.08371444046497345, + "learning_rate": 4.924533220104521e-07, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27670 + }, + { + "epoch": 8.05353505964504, + "flow_matching_loss": 0.014461463037878274, + "grad_norm": 0.07894019782543182, + "learning_rate": 4.882601557301231e-07, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27680 + }, + { + "epoch": 8.056444573756183, + "flow_matching_loss": 0.016643000300973654, + "grad_norm": 0.07991410791873932, + "learning_rate": 4.840846225820717e-07, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27690 + }, + { + "epoch": 8.059354087867327, + "flow_matching_loss": 0.012122046947479249, + "grad_norm": 0.06661070883274078, + "learning_rate": 4.79926727639966e-07, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27700 + }, + { + "epoch": 8.06226360197847, + "flow_matching_loss": 0.015022685658186674, + "grad_norm": 0.0769040510058403, + "learning_rate": 4.7578647595603817e-07, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27710 + }, + { + "epoch": 8.065173116089612, + "flow_matching_loss": 0.018842409551143646, + "grad_norm": 0.11071794480085373, + "learning_rate": 4.716638725610889e-07, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27720 + }, + { + "epoch": 8.068082630200756, + "flow_matching_loss": 0.015069541055709124, + "grad_norm": 0.08245305716991425, + "learning_rate": 4.6755892246446606e-07, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27730 + }, + { + "epoch": 8.0709921443119, + "flow_matching_loss": 0.01351510128006339, + "grad_norm": 0.09481793642044067, + "learning_rate": 4.634716306540715e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27740 + }, + { + "epoch": 8.073901658423043, + "flow_matching_loss": 0.015060655446723104, + "grad_norm": 0.09230012446641922, + "learning_rate": 4.5940200209635244e-07, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27750 + }, + { + "epoch": 8.076811172534187, + "flow_matching_loss": 0.014393373113125562, + "grad_norm": 0.11050088703632355, + "learning_rate": 4.5535004173628657e-07, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27760 + }, + { + "epoch": 8.07972068664533, + "flow_matching_loss": 0.013089052867144346, + "grad_norm": 0.09733860194683075, + "learning_rate": 4.5131575449739225e-07, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27770 + }, + { + "epoch": 8.082630200756475, + "flow_matching_loss": 0.014780598133802414, + "grad_norm": 0.057471808046102524, + "learning_rate": 4.472991452817099e-07, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27780 + }, + { + "epoch": 8.085539714867616, + "flow_matching_loss": 0.014118118584156037, + "grad_norm": 0.06639894098043442, + "learning_rate": 4.433002189697938e-07, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27790 + }, + { + "epoch": 8.08844922897876, + "flow_matching_loss": 0.015409441478550435, + "grad_norm": 0.09329140931367874, + "learning_rate": 4.393189804207187e-07, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27800 + }, + { + "epoch": 8.091358743089904, + "flow_matching_loss": 0.014563405700027943, + "grad_norm": 0.09496444463729858, + "learning_rate": 4.353554344720651e-07, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27810 + }, + { + "epoch": 8.094268257201048, + "flow_matching_loss": 0.0155177628621459, + "grad_norm": 0.07675004005432129, + "learning_rate": 4.314095859399153e-07, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27820 + }, + { + "epoch": 8.097177771312191, + "flow_matching_loss": 0.01617257446050644, + "grad_norm": 0.07051413506269455, + "learning_rate": 4.274814396188459e-07, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27830 + }, + { + "epoch": 8.100087285423335, + "flow_matching_loss": 0.012742032576352359, + "grad_norm": 0.10519028455018997, + "learning_rate": 4.235710002819271e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27840 + }, + { + "epoch": 8.102996799534477, + "flow_matching_loss": 0.013521303003653885, + "grad_norm": 0.05269400775432587, + "learning_rate": 4.196782726807097e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27850 + }, + { + "epoch": 8.10590631364562, + "flow_matching_loss": 0.013458708766847849, + "grad_norm": 0.09800061583518982, + "learning_rate": 4.158032615452251e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27860 + }, + { + "epoch": 8.108815827756764, + "flow_matching_loss": 0.015027925930917263, + "grad_norm": 0.08801549673080444, + "learning_rate": 4.1194597158397853e-07, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27870 + }, + { + "epoch": 8.111725341867908, + "flow_matching_loss": 0.015203026402741671, + "grad_norm": 0.11950559914112091, + "learning_rate": 4.081064074839408e-07, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27880 + }, + { + "epoch": 8.114634855979052, + "flow_matching_loss": 0.0156282477080822, + "grad_norm": 0.1182948648929596, + "learning_rate": 4.042845739105433e-07, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27890 + }, + { + "epoch": 8.117544370090195, + "flow_matching_loss": 0.01520431786775589, + "grad_norm": 0.12647713720798492, + "learning_rate": 4.004804755076746e-07, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27900 + }, + { + "epoch": 8.12045388420134, + "flow_matching_loss": 0.016291917022317647, + "grad_norm": 0.09244107455015182, + "learning_rate": 3.9669411689767063e-07, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27910 + }, + { + "epoch": 8.123363398312481, + "flow_matching_loss": 0.016295634023845196, + "grad_norm": 0.09047530591487885, + "learning_rate": 3.929255026813194e-07, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27920 + }, + { + "epoch": 8.126272912423625, + "flow_matching_loss": 0.012801210628822446, + "grad_norm": 0.11147058010101318, + "learning_rate": 3.8917463743783797e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27930 + }, + { + "epoch": 8.129182426534769, + "flow_matching_loss": 0.014537422638386487, + "grad_norm": 0.12046769261360168, + "learning_rate": 3.854415257248839e-07, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27940 + }, + { + "epoch": 8.132091940645912, + "flow_matching_loss": 0.013633953034877777, + "grad_norm": 0.0998714491724968, + "learning_rate": 3.8172617207853864e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27950 + }, + { + "epoch": 8.135001454757056, + "flow_matching_loss": 0.015175402071326971, + "grad_norm": 0.12115988880395889, + "learning_rate": 3.7802858101330773e-07, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27960 + }, + { + "epoch": 8.1379109688682, + "flow_matching_loss": 0.013528841640800237, + "grad_norm": 0.10901642590761185, + "learning_rate": 3.743487570221121e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27970 + }, + { + "epoch": 8.140820482979342, + "flow_matching_loss": 0.014031935576349496, + "grad_norm": 0.06236205995082855, + "learning_rate": 3.7068670457628517e-07, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27980 + }, + { + "epoch": 8.143729997090485, + "flow_matching_loss": 0.015371669735759496, + "grad_norm": 0.08368196338415146, + "learning_rate": 3.6704242812556576e-07, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27990 + }, + { + "epoch": 8.146639511201629, + "flow_matching_loss": 0.015128017961978912, + "grad_norm": 0.09764424711465836, + "learning_rate": 3.634159320980951e-07, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28000 + }, + { + "epoch": 8.149549025312773, + "flow_matching_loss": 0.016297946963459254, + "grad_norm": 0.08828964084386826, + "learning_rate": 3.5980722090040663e-07, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28010 + }, + { + "epoch": 8.152458539423916, + "flow_matching_loss": 0.01375255435705185, + "grad_norm": 0.09276210516691208, + "learning_rate": 3.5621629891742447e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28020 + }, + { + "epoch": 8.15536805353506, + "flow_matching_loss": 0.013318331120535732, + "grad_norm": 0.0975356251001358, + "learning_rate": 3.526431705124583e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28030 + }, + { + "epoch": 8.158277567646204, + "flow_matching_loss": 0.014954595174640418, + "grad_norm": 0.11649443209171295, + "learning_rate": 3.4908784002719517e-07, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28040 + }, + { + "epoch": 8.161187081757346, + "flow_matching_loss": 0.014357408974319696, + "grad_norm": 0.07320227473974228, + "learning_rate": 3.4555031178169773e-07, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28050 + }, + { + "epoch": 8.16409659586849, + "flow_matching_loss": 0.014700469374656678, + "grad_norm": 0.11027166992425919, + "learning_rate": 3.420305900743942e-07, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28060 + }, + { + "epoch": 8.167006109979633, + "flow_matching_loss": 0.012518169451504945, + "grad_norm": 0.08402913808822632, + "learning_rate": 3.385286791820802e-07, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28070 + }, + { + "epoch": 8.169915624090777, + "flow_matching_loss": 0.0164953182451427, + "grad_norm": 0.10801161825656891, + "learning_rate": 3.350445833599086e-07, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28080 + }, + { + "epoch": 8.17282513820192, + "flow_matching_loss": 0.01530859274789691, + "grad_norm": 0.13929477334022522, + "learning_rate": 3.3157830684138303e-07, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28090 + }, + { + "epoch": 8.175734652313064, + "flow_matching_loss": 0.013373258896172047, + "grad_norm": 0.07522106170654297, + "learning_rate": 3.281298538383559e-07, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28100 + }, + { + "epoch": 8.178644166424208, + "flow_matching_loss": 0.01734298998489976, + "grad_norm": 0.08473938703536987, + "learning_rate": 3.2469922854102387e-07, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28110 + }, + { + "epoch": 8.18155368053535, + "flow_matching_loss": 0.01420684801414609, + "grad_norm": 0.09398151189088821, + "learning_rate": 3.21286435117919e-07, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28120 + }, + { + "epoch": 8.184463194646494, + "flow_matching_loss": 0.015331870689988137, + "grad_norm": 0.07414539903402328, + "learning_rate": 3.1789147771590754e-07, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28130 + }, + { + "epoch": 8.187372708757637, + "flow_matching_loss": 0.014888697862625122, + "grad_norm": 0.07240217179059982, + "learning_rate": 3.14514360460183e-07, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28140 + }, + { + "epoch": 8.190282222868781, + "flow_matching_loss": 0.014316740073263645, + "grad_norm": 0.08101446181535721, + "learning_rate": 3.1115508745426293e-07, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28150 + }, + { + "epoch": 8.193191736979925, + "flow_matching_loss": 0.014792121294885874, + "grad_norm": 0.06910941004753113, + "learning_rate": 3.0781366277997716e-07, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28160 + }, + { + "epoch": 8.196101251091068, + "flow_matching_loss": 0.015220180340111255, + "grad_norm": 0.061839617788791656, + "learning_rate": 3.044900904974762e-07, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28170 + }, + { + "epoch": 8.19901076520221, + "flow_matching_loss": 0.013912210799753666, + "grad_norm": 0.08843021094799042, + "learning_rate": 3.011843746452081e-07, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28180 + }, + { + "epoch": 8.201920279313354, + "flow_matching_loss": 0.01455027861520648, + "grad_norm": 0.10615979135036469, + "learning_rate": 2.9789651923993464e-07, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28190 + }, + { + "epoch": 8.204829793424498, + "flow_matching_loss": 0.011309586744755507, + "grad_norm": 0.07719965279102325, + "learning_rate": 2.9462652827670855e-07, + "loss": 0.0113, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28200 + }, + { + "epoch": 8.207739307535642, + "flow_matching_loss": 0.013671706151217222, + "grad_norm": 0.071086585521698, + "learning_rate": 2.913744057288764e-07, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28210 + }, + { + "epoch": 8.210648821646785, + "flow_matching_loss": 0.013013689406216144, + "grad_norm": 0.07643253356218338, + "learning_rate": 2.8814015554807396e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28220 + }, + { + "epoch": 8.213558335757929, + "flow_matching_loss": 0.012544802576303481, + "grad_norm": 0.1023239940404892, + "learning_rate": 2.849237816642192e-07, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28230 + }, + { + "epoch": 8.216467849869073, + "flow_matching_loss": 0.014227741304785014, + "grad_norm": 0.07822541147470474, + "learning_rate": 2.817252879855109e-07, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28240 + }, + { + "epoch": 8.219377363980215, + "flow_matching_loss": 0.015068450570106506, + "grad_norm": 0.08152998983860016, + "learning_rate": 2.785446783984219e-07, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28250 + }, + { + "epoch": 8.222286878091358, + "flow_matching_loss": 0.015640700608491896, + "grad_norm": 0.09379686415195465, + "learning_rate": 2.7538195676769076e-07, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28260 + }, + { + "epoch": 8.225196392202502, + "flow_matching_loss": 0.015475532412528992, + "grad_norm": 0.12204372137784958, + "learning_rate": 2.722371269363216e-07, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28270 + }, + { + "epoch": 8.228105906313646, + "flow_matching_loss": 0.0142522226087749, + "grad_norm": 0.06549375504255295, + "learning_rate": 2.6911019272558125e-07, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28280 + }, + { + "epoch": 8.23101542042479, + "flow_matching_loss": 0.014438975695520639, + "grad_norm": 0.09007986634969711, + "learning_rate": 2.6600115793498857e-07, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28290 + }, + { + "epoch": 8.233924934535933, + "flow_matching_loss": 0.014029392506927252, + "grad_norm": 0.09302502870559692, + "learning_rate": 2.629100263423151e-07, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28300 + }, + { + "epoch": 8.236834448647077, + "flow_matching_loss": 0.01718112751841545, + "grad_norm": 0.07874174416065216, + "learning_rate": 2.598368017035763e-07, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28310 + }, + { + "epoch": 8.239743962758219, + "flow_matching_loss": 0.012891411129385234, + "grad_norm": 0.05943836644291878, + "learning_rate": 2.5678148775303e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28320 + }, + { + "epoch": 8.242653476869362, + "flow_matching_loss": 0.013200584333389997, + "grad_norm": 0.09453283995389938, + "learning_rate": 2.537440882031683e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28330 + }, + { + "epoch": 8.245562990980506, + "flow_matching_loss": 0.01563205486163497, + "grad_norm": 0.09480831027030945, + "learning_rate": 2.5072460674472385e-07, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28340 + }, + { + "epoch": 8.24847250509165, + "flow_matching_loss": 0.01318454770371318, + "grad_norm": 0.07186613976955414, + "learning_rate": 2.4772304704664683e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28350 + }, + { + "epoch": 8.251382019202794, + "flow_matching_loss": 0.015050746314227582, + "grad_norm": 0.11611827462911606, + "learning_rate": 2.447394127561181e-07, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28360 + }, + { + "epoch": 8.254291533313937, + "flow_matching_loss": 0.013450238853693008, + "grad_norm": 0.07939040660858154, + "learning_rate": 2.41773707498531e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28370 + }, + { + "epoch": 8.25720104742508, + "flow_matching_loss": 0.01494966596364975, + "grad_norm": 0.08989791572093964, + "learning_rate": 2.388259348774996e-07, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28380 + }, + { + "epoch": 8.260110561536223, + "flow_matching_loss": 0.013957982091233135, + "grad_norm": 0.09590578079223633, + "learning_rate": 2.358960984748454e-07, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28390 + }, + { + "epoch": 8.263020075647367, + "flow_matching_loss": 0.012688054423779248, + "grad_norm": 0.08672603964805603, + "learning_rate": 2.3298420185059565e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28400 + }, + { + "epoch": 8.26592958975851, + "flow_matching_loss": 0.013778523541986942, + "grad_norm": 0.07772721350193024, + "learning_rate": 2.3009024854297844e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28410 + }, + { + "epoch": 8.268839103869654, + "flow_matching_loss": 0.012798740016296505, + "grad_norm": 0.07847995311021805, + "learning_rate": 2.272142420684209e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28420 + }, + { + "epoch": 8.271748617980798, + "flow_matching_loss": 0.013769618794322014, + "grad_norm": 0.07164575159549713, + "learning_rate": 2.243561859215426e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28430 + }, + { + "epoch": 8.274658132091941, + "flow_matching_loss": 0.013079442735761403, + "grad_norm": 0.1008741483092308, + "learning_rate": 2.2151608357514897e-07, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28440 + }, + { + "epoch": 8.277567646203083, + "flow_matching_loss": 0.013535825721919536, + "grad_norm": 0.08328311145305634, + "learning_rate": 2.186939384802361e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28450 + }, + { + "epoch": 8.280477160314227, + "flow_matching_loss": 0.012845247797667981, + "grad_norm": 0.08149824291467667, + "learning_rate": 2.1588975406597266e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28460 + }, + { + "epoch": 8.28338667442537, + "flow_matching_loss": 0.01452464833855629, + "grad_norm": 0.10001993924379349, + "learning_rate": 2.131035337397097e-07, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28470 + }, + { + "epoch": 8.286296188536515, + "flow_matching_loss": 0.014525190647691488, + "grad_norm": 0.09195733070373535, + "learning_rate": 2.1033528088696575e-07, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28480 + }, + { + "epoch": 8.289205702647658, + "flow_matching_loss": 0.01274981740862131, + "grad_norm": 0.06892085075378418, + "learning_rate": 2.0758499887143012e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28490 + }, + { + "epoch": 8.292115216758802, + "flow_matching_loss": 0.017312448238953948, + "grad_norm": 0.13304658234119415, + "learning_rate": 2.0485269103495462e-07, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28500 + }, + { + "epoch": 8.295024730869944, + "flow_matching_loss": 0.014396510273218154, + "grad_norm": 0.06144195795059204, + "learning_rate": 2.021383606975502e-07, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28510 + }, + { + "epoch": 8.297934244981088, + "flow_matching_loss": 0.012701548542827368, + "grad_norm": 0.05568660423159599, + "learning_rate": 1.9944201115738524e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28520 + }, + { + "epoch": 8.300843759092231, + "flow_matching_loss": 0.01210336983203888, + "grad_norm": 0.050354622304439545, + "learning_rate": 1.9676364569077732e-07, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28530 + }, + { + "epoch": 8.303753273203375, + "flow_matching_loss": 0.014289540704339743, + "grad_norm": 0.08226069808006287, + "learning_rate": 1.9410326755219477e-07, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28540 + }, + { + "epoch": 8.306662787314519, + "flow_matching_loss": 0.013367579597979785, + "grad_norm": 0.07951254397630692, + "learning_rate": 1.9146087997424677e-07, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28550 + }, + { + "epoch": 8.309572301425662, + "flow_matching_loss": 0.01286365333944559, + "grad_norm": 0.09568344801664352, + "learning_rate": 1.8883648616768335e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28560 + }, + { + "epoch": 8.312481815536806, + "flow_matching_loss": 0.01516064889729023, + "grad_norm": 0.08495879173278809, + "learning_rate": 1.862300893213903e-07, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28570 + }, + { + "epoch": 8.315391329647948, + "flow_matching_loss": 0.016331378277391194, + "grad_norm": 0.10091210901737213, + "learning_rate": 1.8364169260238595e-07, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28580 + }, + { + "epoch": 8.318300843759092, + "flow_matching_loss": 0.01413948806002736, + "grad_norm": 0.10334788262844086, + "learning_rate": 1.8107129915581777e-07, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28590 + }, + { + "epoch": 8.321210357870235, + "flow_matching_loss": 0.015149091184139252, + "grad_norm": 0.08024650067090988, + "learning_rate": 1.785189121049541e-07, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28600 + }, + { + "epoch": 8.32411987198138, + "flow_matching_loss": 0.013446892891079188, + "grad_norm": 0.09790278971195221, + "learning_rate": 1.7598453455118567e-07, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28610 + }, + { + "epoch": 8.327029386092523, + "flow_matching_loss": 0.014606676576659084, + "grad_norm": 0.08446257561445236, + "learning_rate": 1.7346816957402256e-07, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28620 + }, + { + "epoch": 8.329938900203667, + "flow_matching_loss": 0.014502758905291558, + "grad_norm": 0.07711825519800186, + "learning_rate": 1.7096982023108389e-07, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28630 + }, + { + "epoch": 8.33284841431481, + "flow_matching_loss": 0.014967344235628843, + "grad_norm": 0.10981005430221558, + "learning_rate": 1.6848948955809963e-07, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28640 + }, + { + "epoch": 8.335757928425952, + "flow_matching_loss": 0.012771748844534159, + "grad_norm": 0.07632391154766083, + "learning_rate": 1.66027180568904e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28650 + }, + { + "epoch": 8.338667442537096, + "flow_matching_loss": 0.013692122697830201, + "grad_norm": 0.1041378453373909, + "learning_rate": 1.6358289625543865e-07, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28660 + }, + { + "epoch": 8.34157695664824, + "flow_matching_loss": 0.015123107004910707, + "grad_norm": 0.08704020828008652, + "learning_rate": 1.6115663958773618e-07, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28670 + }, + { + "epoch": 8.344486470759383, + "flow_matching_loss": 0.012272137682884932, + "grad_norm": 0.07857710868120193, + "learning_rate": 1.5874841351392832e-07, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28680 + }, + { + "epoch": 8.347395984870527, + "flow_matching_loss": 0.01465540351346135, + "grad_norm": 0.11494773626327515, + "learning_rate": 1.56358220960236e-07, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28690 + }, + { + "epoch": 8.35030549898167, + "flow_matching_loss": 0.014203933812677861, + "grad_norm": 0.08643573522567749, + "learning_rate": 1.5398606483096932e-07, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28700 + }, + { + "epoch": 8.353215013092813, + "flow_matching_loss": 0.013486722577363252, + "grad_norm": 0.08828597515821457, + "learning_rate": 1.5163194800852265e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28710 + }, + { + "epoch": 8.356124527203956, + "flow_matching_loss": 0.014624262787401676, + "grad_norm": 0.08831058442592621, + "learning_rate": 1.4929587335336614e-07, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28720 + }, + { + "epoch": 8.3590340413151, + "flow_matching_loss": 0.012534939963370562, + "grad_norm": 0.07287882268428802, + "learning_rate": 1.4697784370405587e-07, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28730 + }, + { + "epoch": 8.361943555426244, + "flow_matching_loss": 0.013801057823002338, + "grad_norm": 0.07031117379665375, + "learning_rate": 1.4467786187721376e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28740 + }, + { + "epoch": 8.364853069537387, + "flow_matching_loss": 0.015021517965942621, + "grad_norm": 0.092593714594841, + "learning_rate": 1.4239593066753265e-07, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28750 + }, + { + "epoch": 8.367762583648531, + "flow_matching_loss": 0.016323238797485828, + "grad_norm": 0.09415299445390701, + "learning_rate": 1.4013205284777953e-07, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28760 + }, + { + "epoch": 8.370672097759675, + "flow_matching_loss": 0.013675452116876841, + "grad_norm": 0.0977901741862297, + "learning_rate": 1.3788623116877397e-07, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28770 + }, + { + "epoch": 8.373581611870817, + "flow_matching_loss": 0.015185480751097203, + "grad_norm": 0.08400719612836838, + "learning_rate": 1.3565846835940477e-07, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28780 + }, + { + "epoch": 8.37649112598196, + "flow_matching_loss": 0.013430658634752035, + "grad_norm": 0.08755921572446823, + "learning_rate": 1.334487671266116e-07, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28790 + }, + { + "epoch": 8.379400640093104, + "flow_matching_loss": 0.014014257490634919, + "grad_norm": 0.06689438968896866, + "learning_rate": 1.3125713015538998e-07, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28800 + }, + { + "epoch": 8.382310154204248, + "flow_matching_loss": 0.015272505255416036, + "grad_norm": 0.09110616147518158, + "learning_rate": 1.290835601087864e-07, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28810 + }, + { + "epoch": 8.385219668315392, + "flow_matching_loss": 0.015364799834787845, + "grad_norm": 0.09241130948066711, + "learning_rate": 1.2692805962789323e-07, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28820 + }, + { + "epoch": 8.388129182426535, + "flow_matching_loss": 0.014922642149031162, + "grad_norm": 0.09043923020362854, + "learning_rate": 1.2479063133184698e-07, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28830 + }, + { + "epoch": 8.391038696537677, + "flow_matching_loss": 0.014261433854699135, + "grad_norm": 0.11975493282079697, + "learning_rate": 1.2267127781782516e-07, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28840 + }, + { + "epoch": 8.393948210648821, + "flow_matching_loss": 0.014064511004835367, + "grad_norm": 0.06606819480657578, + "learning_rate": 1.205700016610395e-07, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28850 + }, + { + "epoch": 8.396857724759965, + "flow_matching_loss": 0.014657105412334203, + "grad_norm": 0.08033055812120438, + "learning_rate": 1.1848680541474088e-07, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28860 + }, + { + "epoch": 8.399767238871108, + "flow_matching_loss": 0.018531225994229315, + "grad_norm": 0.1858014166355133, + "learning_rate": 1.1642169161020955e-07, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28870 + }, + { + "epoch": 8.402676752982252, + "flow_matching_loss": 0.014781397813931108, + "grad_norm": 0.07953601330518723, + "learning_rate": 1.1437466275675156e-07, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28880 + }, + { + "epoch": 8.405586267093396, + "flow_matching_loss": 0.013318697176873683, + "grad_norm": 0.10187871009111404, + "learning_rate": 1.1234572134170229e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28890 + }, + { + "epoch": 8.40849578120454, + "flow_matching_loss": 0.015234979242086411, + "grad_norm": 0.09704533964395523, + "learning_rate": 1.1033486983041463e-07, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28900 + }, + { + "epoch": 8.411405295315681, + "flow_matching_loss": 0.016399186197668315, + "grad_norm": 0.08608464896678925, + "learning_rate": 1.0834211066626409e-07, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28910 + }, + { + "epoch": 8.414314809426825, + "flow_matching_loss": 0.01332681318745017, + "grad_norm": 0.06860868632793427, + "learning_rate": 1.0636744627064044e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28920 + }, + { + "epoch": 8.417224323537969, + "flow_matching_loss": 0.015355230029672384, + "grad_norm": 0.1399124264717102, + "learning_rate": 1.0441087904294766e-07, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28930 + }, + { + "epoch": 8.420133837649113, + "flow_matching_loss": 0.013893887866288424, + "grad_norm": 0.09447777271270752, + "learning_rate": 1.0247241136059904e-07, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28940 + }, + { + "epoch": 8.423043351760256, + "flow_matching_loss": 0.01703433021903038, + "grad_norm": 0.058854229748249054, + "learning_rate": 1.0055204557901709e-07, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28950 + }, + { + "epoch": 8.4259528658714, + "flow_matching_loss": 0.01566518424078822, + "grad_norm": 0.07704672962427139, + "learning_rate": 9.864978403162361e-08, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28960 + }, + { + "epoch": 8.428862379982544, + "flow_matching_loss": 0.013877123687416315, + "grad_norm": 0.07394403219223022, + "learning_rate": 9.676562902984965e-08, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28970 + }, + { + "epoch": 8.431771894093686, + "flow_matching_loss": 0.014969744905829429, + "grad_norm": 0.1390329748392105, + "learning_rate": 9.489958286311883e-08, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28980 + }, + { + "epoch": 8.43468140820483, + "flow_matching_loss": 0.01324465312063694, + "grad_norm": 0.09416630119085312, + "learning_rate": 9.30516477988541e-08, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28990 + }, + { + "epoch": 8.437590922315973, + "flow_matching_loss": 0.01320120575837791, + "grad_norm": 0.05692411959171295, + "learning_rate": 9.122182608247098e-08, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29000 + }, + { + "epoch": 8.440500436427117, + "flow_matching_loss": 0.013874347414821386, + "grad_norm": 0.06829813867807388, + "learning_rate": 8.94101199373759e-08, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29010 + }, + { + "epoch": 8.44340995053826, + "flow_matching_loss": 0.014637418370693923, + "grad_norm": 0.0982196033000946, + "learning_rate": 8.761653156495964e-08, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29020 + }, + { + "epoch": 8.446319464649404, + "flow_matching_loss": 0.01454687276855111, + "grad_norm": 0.10084112733602524, + "learning_rate": 8.584106314460549e-08, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29030 + }, + { + "epoch": 8.449228978760548, + "flow_matching_loss": 0.014487104024738073, + "grad_norm": 0.10519708693027496, + "learning_rate": 8.408371683367111e-08, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29040 + }, + { + "epoch": 8.45213849287169, + "flow_matching_loss": 0.015199813060462475, + "grad_norm": 0.06820158660411835, + "learning_rate": 8.234449476750172e-08, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29050 + }, + { + "epoch": 8.455048006982834, + "flow_matching_loss": 0.015182297211140394, + "grad_norm": 0.08833839744329453, + "learning_rate": 8.062339905941351e-08, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29060 + }, + { + "epoch": 8.457957521093977, + "flow_matching_loss": 0.015259871818125248, + "grad_norm": 0.10787395387887955, + "learning_rate": 7.89204318007003e-08, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29070 + }, + { + "epoch": 8.460867035205121, + "flow_matching_loss": 0.014203662751242518, + "grad_norm": 0.08389528840780258, + "learning_rate": 7.723559506062849e-08, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29080 + }, + { + "epoch": 8.463776549316265, + "flow_matching_loss": 0.01254896828904748, + "grad_norm": 0.07434823364019394, + "learning_rate": 7.556889088643381e-08, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29090 + }, + { + "epoch": 8.466686063427408, + "flow_matching_loss": 0.014537858590483666, + "grad_norm": 0.07465790957212448, + "learning_rate": 7.392032130331793e-08, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29100 + }, + { + "epoch": 8.46959557753855, + "flow_matching_loss": 0.015037787146866322, + "grad_norm": 0.11713410168886185, + "learning_rate": 7.228988831445015e-08, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29110 + }, + { + "epoch": 8.472505091649694, + "flow_matching_loss": 0.01680860761553049, + "grad_norm": 0.1110420674085617, + "learning_rate": 7.06775939009574e-08, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29120 + }, + { + "epoch": 8.475414605760838, + "flow_matching_loss": 0.015649963822215795, + "grad_norm": 0.09807878732681274, + "learning_rate": 6.908344002193256e-08, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29130 + }, + { + "epoch": 8.478324119871981, + "flow_matching_loss": 0.01509512895718217, + "grad_norm": 0.0649191290140152, + "learning_rate": 6.750742861442116e-08, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29140 + }, + { + "epoch": 8.481233633983125, + "flow_matching_loss": 0.01531128603965044, + "grad_norm": 0.0837479829788208, + "learning_rate": 6.594956159342636e-08, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29150 + }, + { + "epoch": 8.484143148094269, + "flow_matching_loss": 0.014229300245642663, + "grad_norm": 0.0852886512875557, + "learning_rate": 6.440984085190394e-08, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29160 + }, + { + "epoch": 8.487052662205413, + "flow_matching_loss": 0.013946122583001852, + "grad_norm": 0.08564610779285431, + "learning_rate": 6.288826826075733e-08, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29170 + }, + { + "epoch": 8.489962176316554, + "flow_matching_loss": 0.015957748144865037, + "grad_norm": 0.07361208647489548, + "learning_rate": 6.138484566884261e-08, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29180 + }, + { + "epoch": 8.492871690427698, + "flow_matching_loss": 0.01406710147857666, + "grad_norm": 0.110297791659832, + "learning_rate": 5.989957490296183e-08, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29190 + }, + { + "epoch": 8.495781204538842, + "flow_matching_loss": 0.012309949658811093, + "grad_norm": 0.08567453175783157, + "learning_rate": 5.843245776785466e-08, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29200 + }, + { + "epoch": 8.498690718649986, + "flow_matching_loss": 0.013508777879178525, + "grad_norm": 0.1604529321193695, + "learning_rate": 5.6983496046210115e-08, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29210 + }, + { + "epoch": 8.50160023276113, + "flow_matching_loss": 0.01411912301555276, + "grad_norm": 0.0803595632314682, + "learning_rate": 5.5552691498653165e-08, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29220 + }, + { + "epoch": 8.504509746872273, + "flow_matching_loss": 0.014799521304666996, + "grad_norm": 0.10188739746809006, + "learning_rate": 5.4140045863743106e-08, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29230 + }, + { + "epoch": 8.507419260983415, + "flow_matching_loss": 0.015497564896941185, + "grad_norm": 0.07447807490825653, + "learning_rate": 5.274556085798188e-08, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29240 + }, + { + "epoch": 8.510328775094559, + "flow_matching_loss": 0.012959363590925931, + "grad_norm": 0.08522007614374161, + "learning_rate": 5.136923817579742e-08, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29250 + }, + { + "epoch": 8.513238289205702, + "flow_matching_loss": 0.01534451674669981, + "grad_norm": 0.07738329470157623, + "learning_rate": 5.001107948955197e-08, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29260 + }, + { + "epoch": 8.516147803316846, + "flow_matching_loss": 0.014370167069137097, + "grad_norm": 0.12198609858751297, + "learning_rate": 4.867108644953711e-08, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29270 + }, + { + "epoch": 8.51905731742799, + "flow_matching_loss": 0.015105118602514267, + "grad_norm": 0.15379412472248077, + "learning_rate": 4.734926068397038e-08, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29280 + }, + { + "epoch": 8.521966831539133, + "flow_matching_loss": 0.015091814380139113, + "grad_norm": 0.10260328650474548, + "learning_rate": 4.604560379899536e-08, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29290 + }, + { + "epoch": 8.524876345650277, + "flow_matching_loss": 0.013421854330226779, + "grad_norm": 0.09394656866788864, + "learning_rate": 4.476011737867658e-08, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29300 + }, + { + "epoch": 8.52778585976142, + "flow_matching_loss": 0.013804426044225692, + "grad_norm": 0.06654728204011917, + "learning_rate": 4.349280298500291e-08, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29310 + }, + { + "epoch": 8.530695373872563, + "flow_matching_loss": 0.017148305475711823, + "grad_norm": 0.10863077640533447, + "learning_rate": 4.224366215787923e-08, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29320 + }, + { + "epoch": 8.533604887983707, + "flow_matching_loss": 0.013339214120060205, + "grad_norm": 0.07158472388982773, + "learning_rate": 4.101269641513139e-08, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29330 + }, + { + "epoch": 8.53651440209485, + "flow_matching_loss": 0.014804770797491073, + "grad_norm": 0.10242751985788345, + "learning_rate": 3.979990725249627e-08, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29340 + }, + { + "epoch": 8.539423916205994, + "flow_matching_loss": 0.01570573654025793, + "grad_norm": 0.07406982779502869, + "learning_rate": 3.860529614363006e-08, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29350 + }, + { + "epoch": 8.542333430317138, + "flow_matching_loss": 0.01446047257632017, + "grad_norm": 0.07898923009634018, + "learning_rate": 3.742886454009664e-08, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29360 + }, + { + "epoch": 8.545242944428281, + "flow_matching_loss": 0.013784945104271173, + "grad_norm": 0.12123512476682663, + "learning_rate": 3.627061387137087e-08, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29370 + }, + { + "epoch": 8.548152458539423, + "flow_matching_loss": 0.013016645424067974, + "grad_norm": 0.08418362587690353, + "learning_rate": 3.5130545544835305e-08, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29380 + }, + { + "epoch": 8.551061972650567, + "flow_matching_loss": 0.014412074908614158, + "grad_norm": 0.09018685668706894, + "learning_rate": 3.400866094578348e-08, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29390 + }, + { + "epoch": 8.55397148676171, + "flow_matching_loss": 0.014499035757035017, + "grad_norm": 0.09186900407075882, + "learning_rate": 3.290496143740829e-08, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29400 + }, + { + "epoch": 8.556881000872854, + "flow_matching_loss": 0.013651481829583645, + "grad_norm": 0.07218698412179947, + "learning_rate": 3.18194483608103e-08, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29410 + }, + { + "epoch": 8.559790514983998, + "flow_matching_loss": 0.014459491847082973, + "grad_norm": 0.09881356358528137, + "learning_rate": 3.075212303498942e-08, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29420 + }, + { + "epoch": 8.562700029095142, + "flow_matching_loss": 0.01851292671635747, + "grad_norm": 0.14355333149433136, + "learning_rate": 2.9702986756846574e-08, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29430 + }, + { + "epoch": 8.565609543206284, + "flow_matching_loss": 0.010514516104012727, + "grad_norm": 0.06168727949261665, + "learning_rate": 2.867204080118202e-08, + "loss": 0.0105, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29440 + }, + { + "epoch": 8.568519057317427, + "flow_matching_loss": 0.012980909179896116, + "grad_norm": 0.11659830063581467, + "learning_rate": 2.7659286420693708e-08, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29450 + }, + { + "epoch": 8.571428571428571, + "flow_matching_loss": 0.016918274387717248, + "grad_norm": 0.22715991735458374, + "learning_rate": 2.6664724845968934e-08, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29460 + }, + { + "epoch": 8.574338085539715, + "flow_matching_loss": 0.01374113904312253, + "grad_norm": 0.07331649959087372, + "learning_rate": 2.5688357285499343e-08, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29470 + }, + { + "epoch": 8.577247599650859, + "flow_matching_loss": 0.014043162669986486, + "grad_norm": 0.08543414622545242, + "learning_rate": 2.473018492565926e-08, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29480 + }, + { + "epoch": 8.580157113762002, + "flow_matching_loss": 0.015119913220405578, + "grad_norm": 0.08399637043476105, + "learning_rate": 2.3790208930720702e-08, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29490 + }, + { + "epoch": 8.583066627873146, + "flow_matching_loss": 0.015533708408474923, + "grad_norm": 0.12467196583747864, + "learning_rate": 2.2868430442843368e-08, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29500 + }, + { + "epoch": 8.585976141984288, + "flow_matching_loss": 0.01414270419627428, + "grad_norm": 0.08635060489177704, + "learning_rate": 2.196485058207298e-08, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29510 + }, + { + "epoch": 8.588885656095432, + "flow_matching_loss": 0.014634006191045046, + "grad_norm": 0.08957154303789139, + "learning_rate": 2.1079470446346282e-08, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29520 + }, + { + "epoch": 8.591795170206575, + "flow_matching_loss": 0.013496777601540088, + "grad_norm": 0.06706652790307999, + "learning_rate": 2.0212291111482707e-08, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29530 + }, + { + "epoch": 8.594704684317719, + "flow_matching_loss": 0.014957328792661428, + "grad_norm": 0.08290869742631912, + "learning_rate": 1.9363313631187707e-08, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29540 + }, + { + "epoch": 8.597614198428863, + "flow_matching_loss": 0.012337153311818838, + "grad_norm": 0.0891013965010643, + "learning_rate": 1.853253903704777e-08, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29550 + }, + { + "epoch": 8.600523712540006, + "flow_matching_loss": 0.014811934996396303, + "grad_norm": 0.08858301490545273, + "learning_rate": 1.7719968338533733e-08, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29560 + }, + { + "epoch": 8.603433226651148, + "flow_matching_loss": 0.013082755636423827, + "grad_norm": 0.09776012599468231, + "learning_rate": 1.6925602522992467e-08, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29570 + }, + { + "epoch": 8.606342740762292, + "flow_matching_loss": 0.01717021120712161, + "grad_norm": 0.10332617163658142, + "learning_rate": 1.6149442555658533e-08, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29580 + }, + { + "epoch": 8.609252254873436, + "flow_matching_loss": 0.013716396037489175, + "grad_norm": 0.07300597429275513, + "learning_rate": 1.5391489379635858e-08, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29590 + }, + { + "epoch": 8.61216176898458, + "flow_matching_loss": 0.014508464187383652, + "grad_norm": 0.0833553895354271, + "learning_rate": 1.4651743915909399e-08, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29600 + }, + { + "epoch": 8.615071283095723, + "flow_matching_loss": 0.01322893239557743, + "grad_norm": 0.10105647146701813, + "learning_rate": 1.3930207063340139e-08, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29610 + }, + { + "epoch": 8.617980797206867, + "flow_matching_loss": 0.014401822350919247, + "grad_norm": 0.09036899358034134, + "learning_rate": 1.322687969866343e-08, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29620 + }, + { + "epoch": 8.62089031131801, + "flow_matching_loss": 0.01536096129566431, + "grad_norm": 0.12893082201480865, + "learning_rate": 1.2541762676488989e-08, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29630 + }, + { + "epoch": 8.623799825429153, + "flow_matching_loss": 0.01562698264606297, + "grad_norm": 0.09105856716632843, + "learning_rate": 1.1874856829297564e-08, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29640 + }, + { + "epoch": 8.626709339540296, + "flow_matching_loss": 0.014204379543662072, + "grad_norm": 0.11356008797883987, + "learning_rate": 1.1226162967442611e-08, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29650 + }, + { + "epoch": 8.62961885365144, + "flow_matching_loss": 0.01348894601687789, + "grad_norm": 0.0770334005355835, + "learning_rate": 1.059568187914861e-08, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29660 + }, + { + "epoch": 8.632528367762584, + "flow_matching_loss": 0.01218818798661232, + "grad_norm": 0.08303683251142502, + "learning_rate": 9.983414330511087e-09, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29670 + }, + { + "epoch": 8.635437881873727, + "flow_matching_loss": 0.014948657248169184, + "grad_norm": 0.0972728505730629, + "learning_rate": 9.389361065489932e-09, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29680 + }, + { + "epoch": 8.638347395984871, + "flow_matching_loss": 0.015118073113262654, + "grad_norm": 0.07631339132785797, + "learning_rate": 8.81352280591774e-09, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29690 + }, + { + "epoch": 8.641256910096015, + "flow_matching_loss": 0.016150859044864775, + "grad_norm": 0.09784317761659622, + "learning_rate": 8.255900251491478e-09, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29700 + }, + { + "epoch": 8.644166424207157, + "flow_matching_loss": 0.014934702683240176, + "grad_norm": 0.09401462227106094, + "learning_rate": 7.716494079775816e-09, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29710 + }, + { + "epoch": 8.6470759383183, + "flow_matching_loss": 0.013166244048625231, + "grad_norm": 0.08865852653980255, + "learning_rate": 7.195304946199799e-09, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29720 + }, + { + "epoch": 8.649985452429444, + "flow_matching_loss": 0.013800548017024994, + "grad_norm": 0.08763492107391357, + "learning_rate": 6.692333484056845e-09, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29730 + }, + { + "epoch": 8.652894966540588, + "flow_matching_loss": 0.013015625718981027, + "grad_norm": 0.09982739388942719, + "learning_rate": 6.207580304506411e-09, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29740 + }, + { + "epoch": 8.655804480651732, + "flow_matching_loss": 0.01583666019141674, + "grad_norm": 0.11751998960971832, + "learning_rate": 5.741045996567329e-09, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29750 + }, + { + "epoch": 8.658713994762875, + "flow_matching_loss": 0.014649682585150004, + "grad_norm": 0.08662352710962296, + "learning_rate": 5.2927311271244725e-09, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29760 + }, + { + "epoch": 8.661623508874017, + "flow_matching_loss": 0.014005054719746113, + "grad_norm": 0.08532943576574326, + "learning_rate": 4.8626362409204265e-09, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29770 + }, + { + "epoch": 8.664533022985161, + "flow_matching_loss": 0.015095721511170269, + "grad_norm": 0.07754822820425034, + "learning_rate": 4.4507618605638126e-09, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29780 + }, + { + "epoch": 8.667442537096305, + "flow_matching_loss": 0.014074154663830996, + "grad_norm": 0.0658322274684906, + "learning_rate": 4.057108486519301e-09, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29790 + }, + { + "epoch": 8.670352051207448, + "flow_matching_loss": 0.014719462487846613, + "grad_norm": 0.08310340344905853, + "learning_rate": 3.6816765971126043e-09, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29800 + }, + { + "epoch": 8.673261565318592, + "flow_matching_loss": 0.015255775302648544, + "grad_norm": 0.10329257696866989, + "learning_rate": 3.324466648528812e-09, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29810 + }, + { + "epoch": 8.676171079429736, + "flow_matching_loss": 0.013790425937622785, + "grad_norm": 0.11584015190601349, + "learning_rate": 2.9854790748123917e-09, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29820 + }, + { + "epoch": 8.67908059354088, + "flow_matching_loss": 0.014121217839419841, + "grad_norm": 0.08774961531162262, + "learning_rate": 2.664714287865522e-09, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29830 + }, + { + "epoch": 8.681990107652021, + "flow_matching_loss": 0.014284798875451088, + "grad_norm": 0.08351629972457886, + "learning_rate": 2.362172677446428e-09, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29840 + }, + { + "epoch": 8.684899621763165, + "flow_matching_loss": 0.015877093095332383, + "grad_norm": 0.09413779526948929, + "learning_rate": 2.0778546111727137e-09, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29850 + }, + { + "epoch": 8.687809135874309, + "flow_matching_loss": 0.015656255558133124, + "grad_norm": 0.06286724656820297, + "learning_rate": 1.811760434516363e-09, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29860 + }, + { + "epoch": 8.690718649985453, + "flow_matching_loss": 0.013904739916324616, + "grad_norm": 0.07024824619293213, + "learning_rate": 1.5638904708087377e-09, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29870 + }, + { + "epoch": 8.693628164096596, + "flow_matching_loss": 0.014962677471339702, + "grad_norm": 0.09886088967323303, + "learning_rate": 1.3342450212355805e-09, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29880 + }, + { + "epoch": 8.69653767820774, + "flow_matching_loss": 0.015051961597055196, + "grad_norm": 0.09117791801691055, + "learning_rate": 1.122824364835351e-09, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29890 + }, + { + "epoch": 8.699447192318882, + "flow_matching_loss": 0.014378390740603209, + "grad_norm": 0.06521416455507278, + "learning_rate": 9.296287585042196e-10, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29900 + }, + { + "epoch": 8.702356706430026, + "flow_matching_loss": 0.01367904506623745, + "grad_norm": 0.07205648720264435, + "learning_rate": 7.546584369960697e-10, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29910 + }, + { + "epoch": 8.70526622054117, + "flow_matching_loss": 0.014763976959511638, + "grad_norm": 0.06399117410182953, + "learning_rate": 5.97913612912504e-10, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29920 + }, + { + "epoch": 8.708175734652313, + "flow_matching_loss": 0.017035243194550277, + "grad_norm": 0.09678608924150467, + "learning_rate": 4.593944767161684e-10, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29930 + }, + { + "epoch": 8.711085248763457, + "flow_matching_loss": 0.014553564134985209, + "grad_norm": 0.06919198483228683, + "learning_rate": 3.391011967207591e-10, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29940 + }, + { + "epoch": 8.7139947628746, + "flow_matching_loss": 0.015397820062935353, + "grad_norm": 0.10789285600185394, + "learning_rate": 2.370339190910231e-10, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29950 + }, + { + "epoch": 8.716904276985744, + "flow_matching_loss": 0.013118467479944228, + "grad_norm": 0.10595181584358215, + "learning_rate": 1.531927678510847e-10, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29960 + }, + { + "epoch": 8.719813791096886, + "flow_matching_loss": 0.014010875765234233, + "grad_norm": 0.10274787992238998, + "learning_rate": 8.757784487611886e-11, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29970 + }, + { + "epoch": 8.72272330520803, + "flow_matching_loss": 0.013644915632903576, + "grad_norm": 0.0954645425081253, + "learning_rate": 4.018922989235119e-11, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29980 + }, + { + "epoch": 8.725632819319173, + "flow_matching_loss": 0.013218238297849894, + "grad_norm": 0.08894939720630646, + "learning_rate": 1.1026980482053972e-11, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29990 + }, + { + "epoch": 8.728542333430317, + "flow_matching_loss": 0.012512949667870999, + "grad_norm": 0.07938306778669357, + "learning_rate": 9.113208188082922e-14, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 30000 + }, + { + "epoch": 8.728542333430317, + "step": 30000, + "total_flos": 0.0, + "train_loss": 0.017189961644013724, + "train_runtime": 40539.3311, + "train_samples_per_second": 94.723, + "train_steps_per_second": 0.74 + } + ], + "logging_steps": 10, + "max_steps": 30000, + "num_input_tokens_seen": 0, + "num_train_epochs": 9, + "save_steps": 1, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": false, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 0.0, + "train_batch_size": 64, + "trial_name": null, + "trial_params": null +} diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3/training_args.bin b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..3fb69730436b9d669cbd13fc372a81d9baff6809 --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:72a1321eb6d04e9b5eee4e4df9ec47bfb0f5b146e55b70ab4b6c35fa3591b00a +size 5969 diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/resolved_config.yaml b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/resolved_config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..e51f908252d5bbfbcb72b643d81736ad49b64def --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/resolved_config.yaml @@ -0,0 +1,32 @@ +experiment_name: mgd_sweep_from_trained_base +policy_type: groot_mgd +base_model_path: ${HOME}/groot_robocasa/robocasa_v2/MGD_Processing_line_Only/mgd_token_mask_ratio_0.3/phase2 +output_root: ${HOME}/groot_robocasa/robocasa_v2/MGD_Processing_line_Only/dit_core_only +sweep: + mgd_token_mask_ratio: + - 0.3 +dataset: + dataset_soup: my_atomic26_human +training: + num_gpus: 2 + batch_size: 64 +model: null +phases: +- name: phase3_mgd_fm + max_steps: 30000 + save_steps: 0 + trainable: + preset: freeze_processing_line + tune_llm: false + tune_visual: false + tune_projector: true + tune_diffusion_model: true + losses: + mgd_enabled: false + mgd_fm_loss_weight: 1.0 + mgd_loss_weight: 0.0 + mgd_token_mask_ratio: 0.3 + mgd_use_cosine_loss: true + mgd_use_mse_loss: false +resolved_sweep: + mgd_token_mask_ratio: 0.3 diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/mgd_dit_core_only1.yaml b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/mgd_dit_core_only1.yaml new file mode 100644 index 0000000000000000000000000000000000000000..943b4c83d04b770b33d9c2cab0f01a26d211a03b --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/mgd_dit_core_only1.yaml @@ -0,0 +1,30 @@ +experiment_name: mgd_sweep_from_trained_base +policy_type: groot_mgd +base_model_path: ${HOME}/groot_robocasa/robocasa_v2/MGD_Processing_line_Only/mgd_token_mask_ratio_${sweep.mgd_token_mask_ratio}/phase2 +output_root: ${HOME}/groot_robocasa/robocasa_v2/MGD_Processing_line_Only/dit_core_only +sweep: + mgd_token_mask_ratio: + - 0.5 +dataset: + dataset_soup: my_atomic26_human +training: + num_gpus: 1 + batch_size: 128 +model: null +phases: +- name: phase3_mgd_fm + max_steps: 30000 + save_steps: 0 + trainable: + preset: freeze_processing_line + tune_llm: false + tune_visual: false + tune_projector: true + tune_diffusion_model: true + losses: + mgd_enabled: false + mgd_fm_loss_weight: 1.0 + mgd_loss_weight: 0.0 + mgd_token_mask_ratio: ${sweep.mgd_token_mask_ratio} + mgd_use_cosine_loss: true + mgd_use_mse_loss: false diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/phase3/config.json b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/phase3/config.json new file mode 100644 index 0000000000000000000000000000000000000000..2f760935085343e0e47fc3b4677451045980ca46 --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/phase3/config.json @@ -0,0 +1,75 @@ +{ + "action_dim": 32, + "action_head_cfg": { + "action_dim": 32, + "action_horizon": 16, + "add_pos_embed": true, + "backbone_embedding_dim": 2048, + "diffusion_model_cfg": { + "attention_head_dim": 48, + "cross_attention_dim": 2048, + "dropout": 0.2, + "final_dropout": true, + "interleave_self_attention": true, + "norm_type": "ada_norm", + "num_attention_heads": 32, + "num_layers": 16, + "output_dim": 1024, + "positional_embeddings": null + }, + "hidden_size": 1024, + "input_embedding_dim": 1536, + "max_action_dim": 32, + "max_state_dim": 64, + "model_dtype": "float32", + "noise_beta_alpha": 1.5, + "noise_beta_beta": 1.0, + "noise_s": 0.999, + "num_inference_timesteps": 4, + "num_target_vision_tokens": 32, + "num_timestep_buckets": 1000, + "tune_diffusion_model": true, + "tune_projector": true, + "use_vlln": true, + "vl_self_attention_cfg": { + "attention_head_dim": 64, + "dropout": 0.2, + "final_dropout": true, + "num_attention_heads": 32, + "num_layers": 4, + "positional_embeddings": null + } + }, + "action_horizon": 16, + "architectures": [ + "GR00T_N1_5_MGD" + ], + "attn_implementation": null, + "backbone_cfg": { + "eagle_path": "NVEagle/eagle_er-qwen3_1_7B-Siglip2_400M_stage1_5_128gpu_er_v7_1mlp_nops", + "load_bf16": false, + "project_to_dim": null, + "reproject_vision": false, + "select_layer": 12, + "tune_llm": false, + "tune_visual": true, + "use_flash_attention": true + }, + "compute_dtype": "bfloat16", + "hidden_size": 2048, + "mgd_enabled": false, + "mgd_fm_loss_weight": 1.0, + "mgd_loss_weight": 0.0, + "mgd_pretrained_projector_path": null, + "mgd_sequence_hidden_dim": 512, + "mgd_target_dim": 512, + "mgd_target_pooling": "flatten", + "mgd_target_projection": "frozen_random", + "mgd_token_mask_ratio": 0.5, + "mgd_use_cosine_loss": true, + "mgd_use_mse_loss": false, + "model_dtype": "float32", + "model_type": "gr00t_n1_5", + "torch_dtype": "bfloat16", + "transformers_version": "4.51.3" +} diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/phase3/experiment_cfg/metadata.json b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/phase3/experiment_cfg/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..300eb070963337016e16465f50ba8351ea2eb25c --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/phase3/experiment_cfg/metadata.json @@ -0,0 +1,431 @@ +{ + "new_embodiment": { + "statistics": { + "state": { + "base_position": { + "max": [ + 7.3139495849609375, + 0.4876587688922882, + 0.7196521759033203 + ], + "min": [ + -4.94293737411499, + -6.890198230743408, + 0.6996827125549316 + ], + "mean": [ + 2.749288365724937, + -2.0455216239909983, + 0.700532551020533 + ], + "std": [ + 1.6786295897046262, + 1.312897969434244, + 0.00133751758906258 + ], + "q01": [ + -1.4241907881148037, + -5.1716547935950885, + 0.6999670898768614 + ], + "q99": [ + 6.198111545831555, + -0.5873531334104489, + 0.7038833014242587 + ] + }, + "base_rotation": { + "max": [ + 0.0, + 0.0, + 1.0, + 1.0 + ], + "min": [ + 0.0, + 0.0, + -1.0, + 0.0 + ], + "mean": [ + 0.0, + 0.0, + 0.23084999587450247, + 0.6238431088581847 + ], + "std": [ + 0.0, + 0.0, + 0.6556404371644047, + 0.3572252105732042 + ], + "q01": [ + 0.0, + 0.0, + -0.9999999999999999, + 1.7482354593273349e-06 + ], + "q99": [ + 0.0, + 0.0, + 0.9999999999999999, + 0.9999999999999999 + ] + }, + "end_effector_position_relative": { + "max": [ + 0.9014528393745422, + 0.8003877401351929, + 0.9829942584037781 + ], + "min": [ + -0.37471598386764526, + -0.8472502827644348, + -0.25070279836654663 + ], + "mean": [ + 0.28667332601863704, + -0.038315473170876704, + 0.45974658906777444 + ], + "std": [ + 0.17067058828305154, + 0.23569952037784195, + 0.21950702354181487 + ], + "q01": [ + 0.015172948963784228, + -0.44450502063220615, + 0.23314444103329338 + ], + "q99": [ + 0.5609069689572412, + 0.38454179128009547, + 0.7028102736473852 + ] + }, + "end_effector_rotation_relative": { + "max": [ + 0.9999998807907104, + 0.9984455108642578, + 0.9434727430343628, + 0.9062229990959167 + ], + "min": [ + -0.9999930262565613, + -0.9988337755203247, + -0.9618149995803833, + 2.1872274658107926e-07 + ], + "mean": [ + -0.25672297401336, + 0.024425335806687216, + -0.08740977164483847, + 0.16608739644018397 + ], + "std": [ + 0.7932585208392383, + 0.3102260195580416, + 0.3772472576315148, + 0.17451959300158773 + ], + "q01": [ + -0.99379006987882, + -0.5478102050038828, + -0.6101777952971636, + 0.002274085014134748 + ], + "q99": [ + 0.8804856038896288, + 0.5910611092873037, + 0.5216058042993562, + 0.5231965974012255 + ] + }, + "gripper_qpos": { + "max": [ + 0.055869169533252716, + 0.010916369967162609 + ], + "min": [ + -0.011436971835792065, + -0.05664053186774254 + ], + "mean": [ + 0.031651551516052555, + -0.03161481387920421 + ], + "std": [ + 0.013119769222217526, + 0.01306078225192402 + ], + "q01": [ + 0.0060999246446777336, + -0.04062601653964198 + ], + "q99": [ + 0.04054891621517283, + -0.0059163567113456345 + ] + } + }, + "action": { + "base_motion": { + "max": [ + 1.0, + 1.0, + 1.0, + 0.0 + ], + "min": [ + -1.0, + -1.0, + -1.0, + 0.0 + ], + "mean": [ + 0.008144896753718373, + -0.00018893135146078263, + -0.0008739062727221845, + 0.0 + ], + "std": [ + 0.11030045424364411, + 0.10148082570313594, + 0.08975698373467506, + 0.0 + ], + "q01": [ + -0.07287373067581518, + -0.09446994199569948, + -0.07702818259249572, + 0.0 + ], + "q99": [ + 0.04485485414407898, + 0.09626941605965177, + 0.07610665906122742, + 0.0 + ] + }, + "control_mode": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.9221974313425939 + ], + "std": [ + 0.38670194342612674 + ], + "q01": [ + -1.0 + ], + "q99": [ + -0.384693946883099 + ] + }, + "end_effector_position": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.01976129923017491, + -0.020870631344490034, + -0.05993476517349181 + ], + "std": [ + 0.4347024990804053, + 0.4221662976569997, + 0.3845506843044066 + ], + "q01": [ + -0.8835792939767807, + -0.9280919251713778, + -0.783361071470956 + ], + "q99": [ + 0.7622084438449307, + 0.8625125252609077, + 0.7470974753250706 + ] + }, + "end_effector_rotation": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.008089673068544335, + -0.026498254627927348, + 0.0029083551743059005 + ], + "std": [ + 0.11216734503733773, + 0.13206002613798629, + 0.12615870412692864 + ], + "q01": [ + -0.2814627972846672, + -0.4342386516115439, + -0.34489755628340574 + ], + "q99": [ + 0.31879353266939386, + 0.28411797683220563, + 0.3597718671669894 + ] + }, + "gripper_close": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.3824228874769045 + ], + "std": [ + 0.9240015096469786 + ], + "q01": [ + -1.0 + ], + "q99": [ + 0.5597272305813592 + ] + } + } + }, + "modalities": { + "video": { + "robot0_eye_in_hand": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + }, + "robot0_agentview_left": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + }, + "robot0_agentview_right": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + } + }, + "state": { + "base_position": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "base_rotation": { + "absolute": true, + "rotation_type": "quaternion", + "shape": [ + 4 + ], + "continuous": true + }, + "end_effector_position_relative": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "end_effector_rotation_relative": { + "absolute": true, + "rotation_type": "quaternion", + "shape": [ + 4 + ], + "continuous": true + }, + "gripper_qpos": { + "absolute": true, + "rotation_type": null, + "shape": [ + 2 + ], + "continuous": true + } + }, + "action": { + "base_motion": { + "absolute": true, + "rotation_type": null, + "shape": [ + 4 + ], + "continuous": true + }, + "control_mode": { + "absolute": true, + "rotation_type": null, + "shape": [ + 1 + ], + "continuous": true + }, + "end_effector_position": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "end_effector_rotation": { + "absolute": true, + "rotation_type": "axis_angle", + "shape": [ + 3 + ], + "continuous": true + }, + "gripper_close": { + "absolute": true, + "rotation_type": null, + "shape": [ + 1 + ], + "continuous": true + } + } + }, + "embodiment_tag": "new_embodiment" + } +} \ No newline at end of file diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/phase3/model-00001-of-00002.safetensors b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/phase3/model-00001-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..a7343abe56edceb9504a5f16122179c5c885b9fa --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/phase3/model-00001-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8eef2ab5cc4e7fa26e55ac2ace3014b3ae1a00ff8cf2f7fdff33be3116e3c9f7 +size 4999367032 diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/phase3/model-00002-of-00002.safetensors b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/phase3/model-00002-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..7ae4f3adb4dae5885331015110a6973fa521a70c --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/phase3/model-00002-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3feded9f5c354cf03aaeb3673dcf36d4698109c407b0ddae473d1dff2555d030 +size 2643339748 diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/phase3/model.safetensors.index.json b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/phase3/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..579ccf56b744c24c73f571cce697c12343122f56 --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/phase3/model.safetensors.index.json @@ -0,0 +1,916 @@ +{ + "metadata": { + "total_size": 7642572676 + }, + "weight_map": { + "action_head.action_decoder.layer1.W": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer1.b": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer2.W": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer2.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W1.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W1.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W2.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W2.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W3.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W3.b": "model-00002-of-00002.safetensors", + "action_head.future_tokens.weight": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_1.bias": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_1.weight": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_2.bias": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_2.weight": "model-00002-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_1.bias": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_1.weight": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_2.bias": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.position_embedding.weight": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer1.W": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer1.b": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer2.W": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer2.b": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vlln.bias": "model-00002-of-00002.safetensors", + "action_head.vlln.weight": "model-00002-of-00002.safetensors", + "action_target_projector.projector.bias": "model-00002-of-00002.safetensors", + "action_target_projector.projector.weight": "model-00002-of-00002.safetensors", + "backbone.eagle_model.language_model.lm_head.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.embed_tokens.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.mlp1.0.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.mlp1.0.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.patch_embedding.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.patch_embedding.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.position_embedding.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.in_proj_bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.in_proj_weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.layernorm.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.probe": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.post_layernorm.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.post_layernorm.weight": "model-00001-of-00002.safetensors", + "sequence_mgd_head.attn_score.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.attn_score.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.out_proj.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.out_proj.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.0.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.0.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.2.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.2.weight": "model-00002-of-00002.safetensors" + } +} diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/phase3/trainer_state.json b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/phase3/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..7bb0c91a8a1ba1b12497b27daf1e28349d32d08d --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/phase3/trainer_state.json @@ -0,0 +1,30043 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 8.728542333430317, + "eval_steps": 500, + "global_step": 30000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.002909514111143439, + "flow_matching_loss": 0.11957852989435196, + "grad_norm": 0.6391264796257019, + "learning_rate": 1.8e-07, + "loss": 0.1196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10 + }, + { + "epoch": 0.005819028222286878, + "flow_matching_loss": 0.10896031707525253, + "grad_norm": 0.5143718719482422, + "learning_rate": 3.8e-07, + "loss": 0.109, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20 + }, + { + "epoch": 0.008728542333430317, + "flow_matching_loss": 0.10506509095430375, + "grad_norm": 0.5473178625106812, + "learning_rate": 5.800000000000001e-07, + "loss": 0.1051, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 30 + }, + { + "epoch": 0.011638056444573755, + "flow_matching_loss": 0.09418077915906906, + "grad_norm": 0.4443802237510681, + "learning_rate": 7.799999999999999e-07, + "loss": 0.0942, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 40 + }, + { + "epoch": 0.014547570555717195, + "flow_matching_loss": 0.08383444249629975, + "grad_norm": 0.334856241941452, + "learning_rate": 9.8e-07, + "loss": 0.0838, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 50 + }, + { + "epoch": 0.017457084666860634, + "flow_matching_loss": 0.07655692845582962, + "grad_norm": 0.26298630237579346, + "learning_rate": 1.18e-06, + "loss": 0.0766, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 60 + }, + { + "epoch": 0.020366598778004074, + "flow_matching_loss": 0.06602602563798428, + "grad_norm": 0.2860617935657501, + "learning_rate": 1.38e-06, + "loss": 0.066, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 70 + }, + { + "epoch": 0.02327611288914751, + "flow_matching_loss": 0.06293820664286613, + "grad_norm": 0.21859042346477509, + "learning_rate": 1.5800000000000001e-06, + "loss": 0.0629, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 80 + }, + { + "epoch": 0.02618562700029095, + "flow_matching_loss": 0.05282442457973957, + "grad_norm": 0.24048349261283875, + "learning_rate": 1.7800000000000001e-06, + "loss": 0.0528, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 90 + }, + { + "epoch": 0.02909514111143439, + "flow_matching_loss": 0.05028784424066544, + "grad_norm": 0.1738506406545639, + "learning_rate": 1.98e-06, + "loss": 0.0503, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 100 + }, + { + "epoch": 0.03200465522257783, + "flow_matching_loss": 0.048641151189804076, + "grad_norm": 0.2052169293165207, + "learning_rate": 2.1800000000000003e-06, + "loss": 0.0486, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 110 + }, + { + "epoch": 0.03491416933372127, + "flow_matching_loss": 0.04896783232688904, + "grad_norm": 0.1721595823764801, + "learning_rate": 2.38e-06, + "loss": 0.049, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 120 + }, + { + "epoch": 0.037823683444864704, + "flow_matching_loss": 0.04284493736922741, + "grad_norm": 0.1367134302854538, + "learning_rate": 2.58e-06, + "loss": 0.0428, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 130 + }, + { + "epoch": 0.04073319755600815, + "flow_matching_loss": 0.039705104380846026, + "grad_norm": 0.11662483215332031, + "learning_rate": 2.78e-06, + "loss": 0.0397, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 140 + }, + { + "epoch": 0.043642711667151585, + "flow_matching_loss": 0.04180329218506813, + "grad_norm": 0.16507379710674286, + "learning_rate": 2.98e-06, + "loss": 0.0418, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 150 + }, + { + "epoch": 0.04655222577829502, + "flow_matching_loss": 0.03667588848620653, + "grad_norm": 0.13730306923389435, + "learning_rate": 3.18e-06, + "loss": 0.0367, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 160 + }, + { + "epoch": 0.049461739889438465, + "flow_matching_loss": 0.04107833988964558, + "grad_norm": 0.17700107395648956, + "learning_rate": 3.38e-06, + "loss": 0.0411, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 170 + }, + { + "epoch": 0.0523712540005819, + "flow_matching_loss": 0.037821179069578645, + "grad_norm": 0.1482573300600052, + "learning_rate": 3.58e-06, + "loss": 0.0378, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 180 + }, + { + "epoch": 0.055280768111725345, + "flow_matching_loss": 0.033722754940390585, + "grad_norm": 0.11779645830392838, + "learning_rate": 3.7800000000000002e-06, + "loss": 0.0337, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 190 + }, + { + "epoch": 0.05819028222286878, + "flow_matching_loss": 0.034226584061980245, + "grad_norm": 0.1451074630022049, + "learning_rate": 3.98e-06, + "loss": 0.0342, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 200 + }, + { + "epoch": 0.06109979633401222, + "flow_matching_loss": 0.034252094104886055, + "grad_norm": 0.1266142725944519, + "learning_rate": 4.18e-06, + "loss": 0.0343, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 210 + }, + { + "epoch": 0.06400931044515566, + "flow_matching_loss": 0.03504473008215427, + "grad_norm": 0.12142464518547058, + "learning_rate": 4.3799999999999996e-06, + "loss": 0.035, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 220 + }, + { + "epoch": 0.06691882455629909, + "flow_matching_loss": 0.03148159850388765, + "grad_norm": 0.12539684772491455, + "learning_rate": 4.58e-06, + "loss": 0.0315, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 230 + }, + { + "epoch": 0.06982833866744254, + "flow_matching_loss": 0.030874002538621426, + "grad_norm": 0.0804213136434555, + "learning_rate": 4.78e-06, + "loss": 0.0309, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 240 + }, + { + "epoch": 0.07273785277858598, + "flow_matching_loss": 0.03129255045205355, + "grad_norm": 0.11944226920604706, + "learning_rate": 4.980000000000001e-06, + "loss": 0.0313, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 250 + }, + { + "epoch": 0.07564736688972941, + "flow_matching_loss": 0.02998856231570244, + "grad_norm": 0.10760107636451721, + "learning_rate": 5.18e-06, + "loss": 0.03, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 260 + }, + { + "epoch": 0.07855688100087285, + "flow_matching_loss": 0.031979494355618955, + "grad_norm": 0.14510394632816315, + "learning_rate": 5.38e-06, + "loss": 0.032, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 270 + }, + { + "epoch": 0.0814663951120163, + "flow_matching_loss": 0.027600772306323053, + "grad_norm": 0.0853445753455162, + "learning_rate": 5.58e-06, + "loss": 0.0276, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 280 + }, + { + "epoch": 0.08437590922315973, + "flow_matching_loss": 0.029718828946352006, + "grad_norm": 0.12685206532478333, + "learning_rate": 5.7800000000000006e-06, + "loss": 0.0297, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 290 + }, + { + "epoch": 0.08728542333430317, + "flow_matching_loss": 0.028926638327538966, + "grad_norm": 0.11411578953266144, + "learning_rate": 5.98e-06, + "loss": 0.0289, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 300 + }, + { + "epoch": 0.09019493744544661, + "flow_matching_loss": 0.0288217443972826, + "grad_norm": 0.10206004232168198, + "learning_rate": 6.18e-06, + "loss": 0.0288, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 310 + }, + { + "epoch": 0.09310445155659004, + "flow_matching_loss": 0.02908130194991827, + "grad_norm": 0.1455816924571991, + "learning_rate": 6.38e-06, + "loss": 0.0291, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 320 + }, + { + "epoch": 0.09601396566773349, + "flow_matching_loss": 0.028966113366186617, + "grad_norm": 0.10206829011440277, + "learning_rate": 6.58e-06, + "loss": 0.029, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 330 + }, + { + "epoch": 0.09892347977887693, + "flow_matching_loss": 0.026145823672413825, + "grad_norm": 0.1263730227947235, + "learning_rate": 6.78e-06, + "loss": 0.0261, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 340 + }, + { + "epoch": 0.10183299389002037, + "flow_matching_loss": 0.02507718987762928, + "grad_norm": 0.13555730879306793, + "learning_rate": 6.98e-06, + "loss": 0.0251, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 350 + }, + { + "epoch": 0.1047425080011638, + "flow_matching_loss": 0.027567699924111365, + "grad_norm": 0.08892174810171127, + "learning_rate": 7.180000000000001e-06, + "loss": 0.0276, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 360 + }, + { + "epoch": 0.10765202211230725, + "flow_matching_loss": 0.027296516858041287, + "grad_norm": 0.10770472139120102, + "learning_rate": 7.3800000000000005e-06, + "loss": 0.0273, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 370 + }, + { + "epoch": 0.11056153622345069, + "flow_matching_loss": 0.02820488251745701, + "grad_norm": 0.12191107124090195, + "learning_rate": 7.5799999999999994e-06, + "loss": 0.0282, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 380 + }, + { + "epoch": 0.11347105033459412, + "flow_matching_loss": 0.028291431628167628, + "grad_norm": 0.0956963449716568, + "learning_rate": 7.780000000000002e-06, + "loss": 0.0283, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 390 + }, + { + "epoch": 0.11638056444573756, + "flow_matching_loss": 0.02576776556670666, + "grad_norm": 0.13158927857875824, + "learning_rate": 7.98e-06, + "loss": 0.0258, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 400 + }, + { + "epoch": 0.11929007855688101, + "flow_matching_loss": 0.02399444915354252, + "grad_norm": 0.10368754714727402, + "learning_rate": 8.18e-06, + "loss": 0.024, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 410 + }, + { + "epoch": 0.12219959266802444, + "flow_matching_loss": 0.025058373250067233, + "grad_norm": 0.10552521049976349, + "learning_rate": 8.38e-06, + "loss": 0.0251, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 420 + }, + { + "epoch": 0.12510910677916787, + "flow_matching_loss": 0.027486556395888328, + "grad_norm": 0.13808134198188782, + "learning_rate": 8.58e-06, + "loss": 0.0275, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 430 + }, + { + "epoch": 0.12801862089031132, + "flow_matching_loss": 0.027584309317171575, + "grad_norm": 0.12872029840946198, + "learning_rate": 8.78e-06, + "loss": 0.0276, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 440 + }, + { + "epoch": 0.13092813500145475, + "flow_matching_loss": 0.02612177673727274, + "grad_norm": 0.12289468944072723, + "learning_rate": 8.98e-06, + "loss": 0.0261, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 450 + }, + { + "epoch": 0.13383764911259818, + "flow_matching_loss": 0.024657025188207626, + "grad_norm": 0.11251690238714218, + "learning_rate": 9.18e-06, + "loss": 0.0247, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 460 + }, + { + "epoch": 0.13674716322374164, + "flow_matching_loss": 0.025887068547308444, + "grad_norm": 0.11474744975566864, + "learning_rate": 9.38e-06, + "loss": 0.0259, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 470 + }, + { + "epoch": 0.13965667733488507, + "flow_matching_loss": 0.02497375961393118, + "grad_norm": 0.11526566743850708, + "learning_rate": 9.580000000000001e-06, + "loss": 0.025, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 480 + }, + { + "epoch": 0.1425661914460285, + "flow_matching_loss": 0.024761319160461426, + "grad_norm": 0.09640290588140488, + "learning_rate": 9.780000000000001e-06, + "loss": 0.0248, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 490 + }, + { + "epoch": 0.14547570555717196, + "flow_matching_loss": 0.025818328000605108, + "grad_norm": 0.10522473603487015, + "learning_rate": 9.980000000000001e-06, + "loss": 0.0258, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 500 + }, + { + "epoch": 0.1483852196683154, + "flow_matching_loss": 0.025872234813869, + "grad_norm": 0.1299685835838318, + "learning_rate": 1.0179999999999999e-05, + "loss": 0.0259, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 510 + }, + { + "epoch": 0.15129473377945882, + "flow_matching_loss": 0.021965932659804822, + "grad_norm": 0.10535416007041931, + "learning_rate": 1.0379999999999999e-05, + "loss": 0.022, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 520 + }, + { + "epoch": 0.15420424789060228, + "flow_matching_loss": 0.021229392476379873, + "grad_norm": 0.0916435495018959, + "learning_rate": 1.058e-05, + "loss": 0.0212, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 530 + }, + { + "epoch": 0.1571137620017457, + "flow_matching_loss": 0.024864858854562043, + "grad_norm": 0.11862961947917938, + "learning_rate": 1.078e-05, + "loss": 0.0249, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 540 + }, + { + "epoch": 0.16002327611288913, + "flow_matching_loss": 0.022710948903113603, + "grad_norm": 0.08918412029743195, + "learning_rate": 1.098e-05, + "loss": 0.0227, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 550 + }, + { + "epoch": 0.1629327902240326, + "flow_matching_loss": 0.021623315662145613, + "grad_norm": 0.06377934664487839, + "learning_rate": 1.118e-05, + "loss": 0.0216, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 560 + }, + { + "epoch": 0.16584230433517602, + "flow_matching_loss": 0.02545337565243244, + "grad_norm": 0.1168038472533226, + "learning_rate": 1.1380000000000001e-05, + "loss": 0.0255, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 570 + }, + { + "epoch": 0.16875181844631945, + "flow_matching_loss": 0.022344311326742174, + "grad_norm": 0.0990239679813385, + "learning_rate": 1.1580000000000001e-05, + "loss": 0.0223, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 580 + }, + { + "epoch": 0.1716613325574629, + "flow_matching_loss": 0.02600078023970127, + "grad_norm": 0.0860186517238617, + "learning_rate": 1.178e-05, + "loss": 0.026, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 590 + }, + { + "epoch": 0.17457084666860634, + "flow_matching_loss": 0.023707061633467676, + "grad_norm": 0.10180020332336426, + "learning_rate": 1.198e-05, + "loss": 0.0237, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 600 + }, + { + "epoch": 0.17748036077974977, + "flow_matching_loss": 0.02173575069755316, + "grad_norm": 0.10598171502351761, + "learning_rate": 1.2180000000000002e-05, + "loss": 0.0217, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 610 + }, + { + "epoch": 0.18038987489089323, + "flow_matching_loss": 0.023042372614145278, + "grad_norm": 0.10324563831090927, + "learning_rate": 1.238e-05, + "loss": 0.023, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 620 + }, + { + "epoch": 0.18329938900203666, + "flow_matching_loss": 0.022666584141552447, + "grad_norm": 0.1065848097205162, + "learning_rate": 1.258e-05, + "loss": 0.0227, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 630 + }, + { + "epoch": 0.18620890311318009, + "flow_matching_loss": 0.023166095651686192, + "grad_norm": 0.10722683370113373, + "learning_rate": 1.278e-05, + "loss": 0.0232, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 640 + }, + { + "epoch": 0.18911841722432354, + "flow_matching_loss": 0.022956989146769045, + "grad_norm": 0.11428894847631454, + "learning_rate": 1.298e-05, + "loss": 0.023, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 650 + }, + { + "epoch": 0.19202793133546697, + "flow_matching_loss": 0.019625614769756793, + "grad_norm": 0.11275286972522736, + "learning_rate": 1.3180000000000001e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 660 + }, + { + "epoch": 0.1949374454466104, + "flow_matching_loss": 0.025806786492466927, + "grad_norm": 0.11304866522550583, + "learning_rate": 1.338e-05, + "loss": 0.0258, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 670 + }, + { + "epoch": 0.19784695955775386, + "flow_matching_loss": 0.022468149475753308, + "grad_norm": 0.11920002847909927, + "learning_rate": 1.358e-05, + "loss": 0.0225, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 680 + }, + { + "epoch": 0.2007564736688973, + "flow_matching_loss": 0.02155338190495968, + "grad_norm": 0.1312607377767563, + "learning_rate": 1.378e-05, + "loss": 0.0216, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 690 + }, + { + "epoch": 0.20366598778004075, + "flow_matching_loss": 0.020712326094508172, + "grad_norm": 0.11397044360637665, + "learning_rate": 1.3980000000000002e-05, + "loss": 0.0207, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 700 + }, + { + "epoch": 0.20657550189118418, + "flow_matching_loss": 0.022502155043184757, + "grad_norm": 0.09693106263875961, + "learning_rate": 1.4180000000000001e-05, + "loss": 0.0225, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 710 + }, + { + "epoch": 0.2094850160023276, + "flow_matching_loss": 0.02017089566215873, + "grad_norm": 0.08406130969524384, + "learning_rate": 1.438e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 720 + }, + { + "epoch": 0.21239453011347106, + "flow_matching_loss": 0.022927170246839525, + "grad_norm": 0.11150752753019333, + "learning_rate": 1.458e-05, + "loss": 0.0229, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 730 + }, + { + "epoch": 0.2153040442246145, + "flow_matching_loss": 0.021252985298633575, + "grad_norm": 0.09255439043045044, + "learning_rate": 1.4779999999999999e-05, + "loss": 0.0213, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 740 + }, + { + "epoch": 0.21821355833575792, + "flow_matching_loss": 0.02149612121284008, + "grad_norm": 0.10574609786272049, + "learning_rate": 1.498e-05, + "loss": 0.0215, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 750 + }, + { + "epoch": 0.22112307244690138, + "flow_matching_loss": 0.021513335593044757, + "grad_norm": 0.09783576428890228, + "learning_rate": 1.518e-05, + "loss": 0.0215, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 760 + }, + { + "epoch": 0.2240325865580448, + "flow_matching_loss": 0.02217772863805294, + "grad_norm": 0.1033083125948906, + "learning_rate": 1.5380000000000002e-05, + "loss": 0.0222, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 770 + }, + { + "epoch": 0.22694210066918824, + "flow_matching_loss": 0.022127522714436056, + "grad_norm": 0.10406232625246048, + "learning_rate": 1.558e-05, + "loss": 0.0221, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 780 + }, + { + "epoch": 0.2298516147803317, + "flow_matching_loss": 0.022202673368155956, + "grad_norm": 0.12757836282253265, + "learning_rate": 1.578e-05, + "loss": 0.0222, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 790 + }, + { + "epoch": 0.23276112889147513, + "flow_matching_loss": 0.0221654886379838, + "grad_norm": 0.11735603958368301, + "learning_rate": 1.598e-05, + "loss": 0.0222, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 800 + }, + { + "epoch": 0.23567064300261856, + "flow_matching_loss": 0.020244254916906356, + "grad_norm": 0.11844217032194138, + "learning_rate": 1.618e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 810 + }, + { + "epoch": 0.23858015711376201, + "flow_matching_loss": 0.021805757470428944, + "grad_norm": 0.10603825747966766, + "learning_rate": 1.6380000000000002e-05, + "loss": 0.0218, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 820 + }, + { + "epoch": 0.24148967122490544, + "flow_matching_loss": 0.019202703423798085, + "grad_norm": 0.12730568647384644, + "learning_rate": 1.658e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 830 + }, + { + "epoch": 0.24439918533604887, + "flow_matching_loss": 0.021799416095018388, + "grad_norm": 0.0999128445982933, + "learning_rate": 1.6780000000000002e-05, + "loss": 0.0218, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 840 + }, + { + "epoch": 0.24730869944719233, + "flow_matching_loss": 0.024295849446207284, + "grad_norm": 0.11548066139221191, + "learning_rate": 1.698e-05, + "loss": 0.0243, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 850 + }, + { + "epoch": 0.25021821355833573, + "flow_matching_loss": 0.02110443785786629, + "grad_norm": 0.14502982795238495, + "learning_rate": 1.718e-05, + "loss": 0.0211, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 860 + }, + { + "epoch": 0.2531277276694792, + "flow_matching_loss": 0.020604066736996173, + "grad_norm": 0.0936955064535141, + "learning_rate": 1.7380000000000003e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 870 + }, + { + "epoch": 0.25603724178062265, + "flow_matching_loss": 0.022041736356914043, + "grad_norm": 0.09237299114465714, + "learning_rate": 1.758e-05, + "loss": 0.022, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 880 + }, + { + "epoch": 0.2589467558917661, + "flow_matching_loss": 0.021292009763419627, + "grad_norm": 0.07801612466573715, + "learning_rate": 1.778e-05, + "loss": 0.0213, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 890 + }, + { + "epoch": 0.2618562700029095, + "flow_matching_loss": 0.022785918414592744, + "grad_norm": 0.09062657505273819, + "learning_rate": 1.798e-05, + "loss": 0.0228, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 900 + }, + { + "epoch": 0.26476578411405294, + "flow_matching_loss": 0.02075223308056593, + "grad_norm": 0.09298733621835709, + "learning_rate": 1.818e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 910 + }, + { + "epoch": 0.26767529822519637, + "flow_matching_loss": 0.021669294498860837, + "grad_norm": 0.11384975165128708, + "learning_rate": 1.838e-05, + "loss": 0.0217, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 920 + }, + { + "epoch": 0.27058481233633985, + "flow_matching_loss": 0.021932714246213438, + "grad_norm": 0.1148248165845871, + "learning_rate": 1.8579999999999998e-05, + "loss": 0.0219, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 930 + }, + { + "epoch": 0.2734943264474833, + "flow_matching_loss": 0.021899109706282616, + "grad_norm": 0.13721060752868652, + "learning_rate": 1.878e-05, + "loss": 0.0219, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 940 + }, + { + "epoch": 0.2764038405586267, + "flow_matching_loss": 0.021844013407826424, + "grad_norm": 0.11574936658143997, + "learning_rate": 1.898e-05, + "loss": 0.0218, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 950 + }, + { + "epoch": 0.27931335466977014, + "flow_matching_loss": 0.022226464096456768, + "grad_norm": 0.08913720399141312, + "learning_rate": 1.918e-05, + "loss": 0.0222, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 960 + }, + { + "epoch": 0.28222286878091357, + "flow_matching_loss": 0.021070724166929723, + "grad_norm": 0.0918000116944313, + "learning_rate": 1.938e-05, + "loss": 0.0211, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 970 + }, + { + "epoch": 0.285132382892057, + "flow_matching_loss": 0.020725973695516587, + "grad_norm": 0.11576782912015915, + "learning_rate": 1.958e-05, + "loss": 0.0207, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 980 + }, + { + "epoch": 0.2880418970032005, + "flow_matching_loss": 0.021653497219085695, + "grad_norm": 0.08966407924890518, + "learning_rate": 1.978e-05, + "loss": 0.0217, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 990 + }, + { + "epoch": 0.2909514111143439, + "flow_matching_loss": 0.02201499305665493, + "grad_norm": 0.12441512942314148, + "learning_rate": 1.9980000000000002e-05, + "loss": 0.022, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1000 + }, + { + "epoch": 0.29386092522548735, + "flow_matching_loss": 0.022179529070854187, + "grad_norm": 0.13032536208629608, + "learning_rate": 2.018e-05, + "loss": 0.0222, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1010 + }, + { + "epoch": 0.2967704393366308, + "flow_matching_loss": 0.02164872195571661, + "grad_norm": 0.13288572430610657, + "learning_rate": 2.038e-05, + "loss": 0.0216, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1020 + }, + { + "epoch": 0.2996799534477742, + "flow_matching_loss": 0.021323128789663314, + "grad_norm": 0.14666040241718292, + "learning_rate": 2.0580000000000003e-05, + "loss": 0.0213, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1030 + }, + { + "epoch": 0.30258946755891764, + "flow_matching_loss": 0.02106987340375781, + "grad_norm": 0.11167845129966736, + "learning_rate": 2.078e-05, + "loss": 0.0211, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1040 + }, + { + "epoch": 0.3054989816700611, + "flow_matching_loss": 0.022299323789775373, + "grad_norm": 0.09753567725419998, + "learning_rate": 2.0980000000000002e-05, + "loss": 0.0223, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1050 + }, + { + "epoch": 0.30840849578120455, + "flow_matching_loss": 0.019890730641782282, + "grad_norm": 0.13520494103431702, + "learning_rate": 2.118e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1060 + }, + { + "epoch": 0.311318009892348, + "flow_matching_loss": 0.02100431462749839, + "grad_norm": 0.10593137890100479, + "learning_rate": 2.1380000000000002e-05, + "loss": 0.021, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1070 + }, + { + "epoch": 0.3142275240034914, + "flow_matching_loss": 0.02043596040457487, + "grad_norm": 0.144779771566391, + "learning_rate": 2.1580000000000003e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1080 + }, + { + "epoch": 0.31713703811463484, + "flow_matching_loss": 0.01877970974892378, + "grad_norm": 0.11020537465810776, + "learning_rate": 2.178e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1090 + }, + { + "epoch": 0.32004655222577827, + "flow_matching_loss": 0.021806272957473994, + "grad_norm": 0.15087948739528656, + "learning_rate": 2.198e-05, + "loss": 0.0218, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1100 + }, + { + "epoch": 0.32295606633692175, + "flow_matching_loss": 0.02225855067372322, + "grad_norm": 0.1395636796951294, + "learning_rate": 2.2179999999999998e-05, + "loss": 0.0223, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1110 + }, + { + "epoch": 0.3258655804480652, + "flow_matching_loss": 0.019476784858852625, + "grad_norm": 0.12086892127990723, + "learning_rate": 2.238e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1120 + }, + { + "epoch": 0.3287750945592086, + "flow_matching_loss": 0.021062810719013215, + "grad_norm": 0.11054575443267822, + "learning_rate": 2.258e-05, + "loss": 0.0211, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1130 + }, + { + "epoch": 0.33168460867035204, + "flow_matching_loss": 0.018803967535495757, + "grad_norm": 0.12821991741657257, + "learning_rate": 2.278e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1140 + }, + { + "epoch": 0.3345941227814955, + "flow_matching_loss": 0.020003264583647252, + "grad_norm": 0.13383738696575165, + "learning_rate": 2.298e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1150 + }, + { + "epoch": 0.3375036368926389, + "flow_matching_loss": 0.021575142815709113, + "grad_norm": 0.15301638841629028, + "learning_rate": 2.3179999999999998e-05, + "loss": 0.0216, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1160 + }, + { + "epoch": 0.3404131510037824, + "flow_matching_loss": 0.0210911032743752, + "grad_norm": 0.09102558344602585, + "learning_rate": 2.338e-05, + "loss": 0.0211, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1170 + }, + { + "epoch": 0.3433226651149258, + "flow_matching_loss": 0.022730602882802486, + "grad_norm": 0.08975644409656525, + "learning_rate": 2.358e-05, + "loss": 0.0227, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1180 + }, + { + "epoch": 0.34623217922606925, + "flow_matching_loss": 0.019879920966923238, + "grad_norm": 0.08419822156429291, + "learning_rate": 2.378e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1190 + }, + { + "epoch": 0.3491416933372127, + "flow_matching_loss": 0.020650241617113353, + "grad_norm": 0.1551811397075653, + "learning_rate": 2.398e-05, + "loss": 0.0207, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1200 + }, + { + "epoch": 0.3520512074483561, + "flow_matching_loss": 0.02122237253934145, + "grad_norm": 0.0947592556476593, + "learning_rate": 2.4180000000000002e-05, + "loss": 0.0212, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1210 + }, + { + "epoch": 0.35496072155949954, + "flow_matching_loss": 0.02069281041622162, + "grad_norm": 0.1302022635936737, + "learning_rate": 2.438e-05, + "loss": 0.0207, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1220 + }, + { + "epoch": 0.357870235670643, + "flow_matching_loss": 0.020118600130081175, + "grad_norm": 0.08868718892335892, + "learning_rate": 2.4580000000000002e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1230 + }, + { + "epoch": 0.36077974978178645, + "flow_matching_loss": 0.01999882245436311, + "grad_norm": 0.09383807331323624, + "learning_rate": 2.478e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1240 + }, + { + "epoch": 0.3636892638929299, + "flow_matching_loss": 0.021627848595380784, + "grad_norm": 0.10378488898277283, + "learning_rate": 2.498e-05, + "loss": 0.0216, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1250 + }, + { + "epoch": 0.3665987780040733, + "flow_matching_loss": 0.02108713462948799, + "grad_norm": 0.10800045728683472, + "learning_rate": 2.5180000000000003e-05, + "loss": 0.0211, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1260 + }, + { + "epoch": 0.36950829211521674, + "flow_matching_loss": 0.01979217482730746, + "grad_norm": 0.0819505825638771, + "learning_rate": 2.538e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1270 + }, + { + "epoch": 0.37241780622636017, + "flow_matching_loss": 0.02296731136739254, + "grad_norm": 0.10492303222417831, + "learning_rate": 2.5580000000000002e-05, + "loss": 0.023, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1280 + }, + { + "epoch": 0.37532732033750366, + "flow_matching_loss": 0.018664024583995343, + "grad_norm": 0.12010547518730164, + "learning_rate": 2.578e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1290 + }, + { + "epoch": 0.3782368344486471, + "flow_matching_loss": 0.020160029642283916, + "grad_norm": 0.12036475539207458, + "learning_rate": 2.5980000000000002e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1300 + }, + { + "epoch": 0.3811463485597905, + "flow_matching_loss": 0.020206746086478232, + "grad_norm": 0.10722903907299042, + "learning_rate": 2.618e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1310 + }, + { + "epoch": 0.38405586267093395, + "flow_matching_loss": 0.020989946834743024, + "grad_norm": 0.11839815229177475, + "learning_rate": 2.6379999999999998e-05, + "loss": 0.021, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1320 + }, + { + "epoch": 0.3869653767820774, + "flow_matching_loss": 0.021519336104393005, + "grad_norm": 0.10365170985460281, + "learning_rate": 2.658e-05, + "loss": 0.0215, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1330 + }, + { + "epoch": 0.3898748908932208, + "flow_matching_loss": 0.020084141939878463, + "grad_norm": 0.11617409437894821, + "learning_rate": 2.678e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1340 + }, + { + "epoch": 0.3927844050043643, + "flow_matching_loss": 0.021676960028707982, + "grad_norm": 0.09898514300584793, + "learning_rate": 2.698e-05, + "loss": 0.0217, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1350 + }, + { + "epoch": 0.3956939191155077, + "flow_matching_loss": 0.020147782191634177, + "grad_norm": 0.12194157391786575, + "learning_rate": 2.718e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1360 + }, + { + "epoch": 0.39860343322665115, + "flow_matching_loss": 0.01957361977547407, + "grad_norm": 0.10781208425760269, + "learning_rate": 2.738e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1370 + }, + { + "epoch": 0.4015129473377946, + "flow_matching_loss": 0.018152609560638668, + "grad_norm": 0.09992589056491852, + "learning_rate": 2.758e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1380 + }, + { + "epoch": 0.404422461448938, + "flow_matching_loss": 0.02023630989715457, + "grad_norm": 0.09094756096601486, + "learning_rate": 2.778e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1390 + }, + { + "epoch": 0.4073319755600815, + "flow_matching_loss": 0.019676712807267904, + "grad_norm": 0.10377799719572067, + "learning_rate": 2.798e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1400 + }, + { + "epoch": 0.4102414896712249, + "flow_matching_loss": 0.022427573427557947, + "grad_norm": 0.13727609813213348, + "learning_rate": 2.818e-05, + "loss": 0.0224, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1410 + }, + { + "epoch": 0.41315100378236835, + "flow_matching_loss": 0.019265454821288587, + "grad_norm": 0.09035764634609222, + "learning_rate": 2.838e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1420 + }, + { + "epoch": 0.4160605178935118, + "flow_matching_loss": 0.021049279905855655, + "grad_norm": 0.11919813603162766, + "learning_rate": 2.858e-05, + "loss": 0.021, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1430 + }, + { + "epoch": 0.4189700320046552, + "flow_matching_loss": 0.019644626323133706, + "grad_norm": 0.10342154651880264, + "learning_rate": 2.8780000000000002e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1440 + }, + { + "epoch": 0.42187954611579864, + "flow_matching_loss": 0.020715662091970444, + "grad_norm": 0.09389200806617737, + "learning_rate": 2.898e-05, + "loss": 0.0207, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1450 + }, + { + "epoch": 0.42478906022694213, + "flow_matching_loss": 0.020945943519473077, + "grad_norm": 0.12466754019260406, + "learning_rate": 2.9180000000000002e-05, + "loss": 0.0209, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1460 + }, + { + "epoch": 0.42769857433808556, + "flow_matching_loss": 0.02250794190913439, + "grad_norm": 0.17169566452503204, + "learning_rate": 2.938e-05, + "loss": 0.0225, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1470 + }, + { + "epoch": 0.430608088449229, + "flow_matching_loss": 0.019605210702866315, + "grad_norm": 0.11155330389738083, + "learning_rate": 2.958e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1480 + }, + { + "epoch": 0.4335176025603724, + "flow_matching_loss": 0.018920379504561425, + "grad_norm": 0.11541755497455597, + "learning_rate": 2.9780000000000003e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1490 + }, + { + "epoch": 0.43642711667151585, + "flow_matching_loss": 0.019303594809025527, + "grad_norm": 0.12945598363876343, + "learning_rate": 2.998e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1500 + }, + { + "epoch": 0.4393366307826593, + "flow_matching_loss": 0.01928238393738866, + "grad_norm": 0.11383146047592163, + "learning_rate": 2.9999992618302022e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1510 + }, + { + "epoch": 0.44224614489380276, + "flow_matching_loss": 0.02000527232885361, + "grad_norm": 0.15020206570625305, + "learning_rate": 2.999996710133069e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1520 + }, + { + "epoch": 0.4451556590049462, + "flow_matching_loss": 0.01922872643917799, + "grad_norm": 0.12204582989215851, + "learning_rate": 2.9999923357984923e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1530 + }, + { + "epoch": 0.4480651731160896, + "flow_matching_loss": 0.019270368199795485, + "grad_norm": 0.10549847036600113, + "learning_rate": 2.9999861388317874e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1540 + }, + { + "epoch": 0.45097468722723305, + "flow_matching_loss": 0.021559222601354122, + "grad_norm": 0.13807307183742523, + "learning_rate": 2.9999781192404847e-05, + "loss": 0.0216, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1550 + }, + { + "epoch": 0.4538842013383765, + "flow_matching_loss": 0.01943699046969414, + "grad_norm": 0.13433963060379028, + "learning_rate": 2.9999682770343283e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1560 + }, + { + "epoch": 0.4567937154495199, + "flow_matching_loss": 0.02126980908215046, + "grad_norm": 0.10999911278486252, + "learning_rate": 2.9999566122252777e-05, + "loss": 0.0213, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1570 + }, + { + "epoch": 0.4597032295606634, + "flow_matching_loss": 0.0185450810007751, + "grad_norm": 0.11278800666332245, + "learning_rate": 2.9999431248275065e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1580 + }, + { + "epoch": 0.4626127436718068, + "flow_matching_loss": 0.01823292300105095, + "grad_norm": 0.10676553100347519, + "learning_rate": 2.9999278148574027e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1590 + }, + { + "epoch": 0.46552225778295026, + "flow_matching_loss": 0.01963558318093419, + "grad_norm": 0.09251893311738968, + "learning_rate": 2.9999106823335705e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1600 + }, + { + "epoch": 0.4684317718940937, + "flow_matching_loss": 0.019634858518838883, + "grad_norm": 0.14447124302387238, + "learning_rate": 2.999891727276827e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1610 + }, + { + "epoch": 0.4713412860052371, + "flow_matching_loss": 0.01726821893826127, + "grad_norm": 0.11315406858921051, + "learning_rate": 2.9998709497102037e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1620 + }, + { + "epoch": 0.47425080011638054, + "flow_matching_loss": 0.020811807550489902, + "grad_norm": 0.12515220046043396, + "learning_rate": 2.9998483496589485e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1630 + }, + { + "epoch": 0.47716031422752403, + "flow_matching_loss": 0.020409489050507545, + "grad_norm": 0.10874652117490768, + "learning_rate": 2.9998239271505217e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1640 + }, + { + "epoch": 0.48006982833866746, + "flow_matching_loss": 0.022171822655946017, + "grad_norm": 0.09445799142122269, + "learning_rate": 2.9997976822145995e-05, + "loss": 0.0222, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1650 + }, + { + "epoch": 0.4829793424498109, + "flow_matching_loss": 0.019888615421950816, + "grad_norm": 0.11820948868989944, + "learning_rate": 2.9997696148830714e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1660 + }, + { + "epoch": 0.4858888565609543, + "flow_matching_loss": 0.019783002324402332, + "grad_norm": 0.08661482483148575, + "learning_rate": 2.9997397251900423e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1670 + }, + { + "epoch": 0.48879837067209775, + "flow_matching_loss": 0.018650043196976183, + "grad_norm": 0.1063147559762001, + "learning_rate": 2.999708013171831e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1680 + }, + { + "epoch": 0.4917078847832412, + "flow_matching_loss": 0.019747700728476047, + "grad_norm": 0.11096528172492981, + "learning_rate": 2.9996744788669705e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1690 + }, + { + "epoch": 0.49461739889438466, + "flow_matching_loss": 0.020599117409437893, + "grad_norm": 0.11614083498716354, + "learning_rate": 2.999639122316208e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1700 + }, + { + "epoch": 0.4975269130055281, + "flow_matching_loss": 0.01648150905966759, + "grad_norm": 0.11915189772844315, + "learning_rate": 2.999601943562505e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1710 + }, + { + "epoch": 0.5004364271166715, + "flow_matching_loss": 0.0214836566708982, + "grad_norm": 0.12210331857204437, + "learning_rate": 2.9995629426510377e-05, + "loss": 0.0215, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1720 + }, + { + "epoch": 0.503345941227815, + "flow_matching_loss": 0.02088084127753973, + "grad_norm": 0.13143905997276306, + "learning_rate": 2.9995221196291952e-05, + "loss": 0.0209, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1730 + }, + { + "epoch": 0.5062554553389584, + "flow_matching_loss": 0.019048825930804016, + "grad_norm": 0.12355560809373856, + "learning_rate": 2.999479474546582e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1740 + }, + { + "epoch": 0.5091649694501018, + "flow_matching_loss": 0.019093702919781207, + "grad_norm": 0.10883723944425583, + "learning_rate": 2.9994350074550153e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1750 + }, + { + "epoch": 0.5120744835612453, + "flow_matching_loss": 0.01897429833188653, + "grad_norm": 0.12529560923576355, + "learning_rate": 2.9993887184085273e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1760 + }, + { + "epoch": 0.5149839976723887, + "flow_matching_loss": 0.02036708313971758, + "grad_norm": 0.11655668914318085, + "learning_rate": 2.9993406074633628e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1770 + }, + { + "epoch": 0.5178935117835322, + "flow_matching_loss": 0.019621549919247628, + "grad_norm": 0.12911736965179443, + "learning_rate": 2.9992906746779824e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1780 + }, + { + "epoch": 0.5208030258946756, + "flow_matching_loss": 0.01990507422015071, + "grad_norm": 0.10910261422395706, + "learning_rate": 2.9992389201130576e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1790 + }, + { + "epoch": 0.523712540005819, + "flow_matching_loss": 0.02075831890106201, + "grad_norm": 0.07461588084697723, + "learning_rate": 2.999185343831476e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1800 + }, + { + "epoch": 0.5266220541169625, + "flow_matching_loss": 0.02014348916709423, + "grad_norm": 0.11058427393436432, + "learning_rate": 2.999129945898338e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1810 + }, + { + "epoch": 0.5295315682281059, + "flow_matching_loss": 0.01883417833596468, + "grad_norm": 0.09275602549314499, + "learning_rate": 2.999072726380957e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1820 + }, + { + "epoch": 0.5324410823392494, + "flow_matching_loss": 0.0203109884634614, + "grad_norm": 0.10750189423561096, + "learning_rate": 2.9990136853488603e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1830 + }, + { + "epoch": 0.5353505964503927, + "flow_matching_loss": 0.01849959949031472, + "grad_norm": 0.09932545572519302, + "learning_rate": 2.9989528228737878e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1840 + }, + { + "epoch": 0.5382601105615362, + "flow_matching_loss": 0.018869185727089642, + "grad_norm": 0.10849635303020477, + "learning_rate": 2.9988901390296935e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1850 + }, + { + "epoch": 0.5411696246726797, + "flow_matching_loss": 0.020195501390844584, + "grad_norm": 0.07934395223855972, + "learning_rate": 2.9988256338927442e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1860 + }, + { + "epoch": 0.5440791387838231, + "flow_matching_loss": 0.018912001233547925, + "grad_norm": 0.08885929733514786, + "learning_rate": 2.9987593075413198e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1870 + }, + { + "epoch": 0.5469886528949666, + "flow_matching_loss": 0.019494645390659572, + "grad_norm": 0.10698315501213074, + "learning_rate": 2.9986911600560128e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1880 + }, + { + "epoch": 0.5498981670061099, + "flow_matching_loss": 0.01851522019132972, + "grad_norm": 0.09530776739120483, + "learning_rate": 2.998621191519629e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1890 + }, + { + "epoch": 0.5528076811172534, + "flow_matching_loss": 0.020170450396835803, + "grad_norm": 0.10968459397554398, + "learning_rate": 2.9985494020171873e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1900 + }, + { + "epoch": 0.5557171952283969, + "flow_matching_loss": 0.018374338280409575, + "grad_norm": 0.09446757286787033, + "learning_rate": 2.9984757916359176e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1910 + }, + { + "epoch": 0.5586267093395403, + "flow_matching_loss": 0.018384858034551144, + "grad_norm": 0.11678633093833923, + "learning_rate": 2.998400360465264e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1920 + }, + { + "epoch": 0.5615362234506838, + "flow_matching_loss": 0.018652132991701365, + "grad_norm": 0.14418740570545197, + "learning_rate": 2.998323108596883e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1930 + }, + { + "epoch": 0.5644457375618271, + "flow_matching_loss": 0.019151432253420354, + "grad_norm": 0.10495369136333466, + "learning_rate": 2.9982440361246422e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1940 + }, + { + "epoch": 0.5673552516729706, + "flow_matching_loss": 0.01788393370807171, + "grad_norm": 0.10260157287120819, + "learning_rate": 2.998163143144622e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1950 + }, + { + "epoch": 0.570264765784114, + "flow_matching_loss": 0.022401940356940032, + "grad_norm": 0.12304387986660004, + "learning_rate": 2.9980804297551155e-05, + "loss": 0.0224, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1960 + }, + { + "epoch": 0.5731742798952575, + "flow_matching_loss": 0.018053389620035887, + "grad_norm": 0.08812795579433441, + "learning_rate": 2.997995896056627e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1970 + }, + { + "epoch": 0.576083794006401, + "flow_matching_loss": 0.01937941098585725, + "grad_norm": 0.124569371342659, + "learning_rate": 2.997909542151873e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1980 + }, + { + "epoch": 0.5789933081175443, + "flow_matching_loss": 0.022355991415679455, + "grad_norm": 0.16409650444984436, + "learning_rate": 2.9978213681457823e-05, + "loss": 0.0224, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 1990 + }, + { + "epoch": 0.5819028222286878, + "flow_matching_loss": 0.02111262958496809, + "grad_norm": 0.13631121814250946, + "learning_rate": 2.9977313741454933e-05, + "loss": 0.0211, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2000 + }, + { + "epoch": 0.5848123363398312, + "flow_matching_loss": 0.01932366006076336, + "grad_norm": 0.12480654567480087, + "learning_rate": 2.997639560260358e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2010 + }, + { + "epoch": 0.5877218504509747, + "flow_matching_loss": 0.01828473322093487, + "grad_norm": 0.10486571490764618, + "learning_rate": 2.9975459266019387e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2020 + }, + { + "epoch": 0.5906313645621182, + "flow_matching_loss": 0.017612296901643277, + "grad_norm": 0.10991387814283371, + "learning_rate": 2.9974504732840092e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2030 + }, + { + "epoch": 0.5935408786732616, + "flow_matching_loss": 0.018814872298389674, + "grad_norm": 0.095576211810112, + "learning_rate": 2.997353200422555e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2040 + }, + { + "epoch": 0.596450392784405, + "flow_matching_loss": 0.019239315204322337, + "grad_norm": 0.09007420390844345, + "learning_rate": 2.99725410813577e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2050 + }, + { + "epoch": 0.5993599068955484, + "flow_matching_loss": 0.018721137940883637, + "grad_norm": 0.12215665727853775, + "learning_rate": 2.997153196544063e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2060 + }, + { + "epoch": 0.6022694210066919, + "flow_matching_loss": 0.019014161452651022, + "grad_norm": 0.08025005459785461, + "learning_rate": 2.997050465770049e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2070 + }, + { + "epoch": 0.6051789351178353, + "flow_matching_loss": 0.020035501383244992, + "grad_norm": 0.12162505090236664, + "learning_rate": 2.9969459159385562e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2080 + }, + { + "epoch": 0.6080884492289788, + "flow_matching_loss": 0.019283664971590044, + "grad_norm": 0.09975284337997437, + "learning_rate": 2.9968395471766237e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2090 + }, + { + "epoch": 0.6109979633401222, + "flow_matching_loss": 0.017876154370605947, + "grad_norm": 0.10556112229824066, + "learning_rate": 2.996731359613498e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2100 + }, + { + "epoch": 0.6139074774512656, + "flow_matching_loss": 0.018078568018972875, + "grad_norm": 0.11613994091749191, + "learning_rate": 2.996621353380638e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2110 + }, + { + "epoch": 0.6168169915624091, + "flow_matching_loss": 0.019239773973822594, + "grad_norm": 0.14407341182231903, + "learning_rate": 2.9965095286117113e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2120 + }, + { + "epoch": 0.6197265056735525, + "flow_matching_loss": 0.018044556118547918, + "grad_norm": 0.11327681690454483, + "learning_rate": 2.996395885442596e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2130 + }, + { + "epoch": 0.622636019784696, + "flow_matching_loss": 0.018167013954371213, + "grad_norm": 0.09858638048171997, + "learning_rate": 2.996280424011379e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2140 + }, + { + "epoch": 0.6255455338958394, + "flow_matching_loss": 0.01969497688114643, + "grad_norm": 0.1150483638048172, + "learning_rate": 2.9961631444583565e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2150 + }, + { + "epoch": 0.6284550480069828, + "flow_matching_loss": 0.018967631552368403, + "grad_norm": 0.10183852910995483, + "learning_rate": 2.9960440469260347e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2160 + }, + { + "epoch": 0.6313645621181263, + "flow_matching_loss": 0.01922054812312126, + "grad_norm": 0.11118480563163757, + "learning_rate": 2.995923131559128e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2170 + }, + { + "epoch": 0.6342740762292697, + "flow_matching_loss": 0.018541223369538784, + "grad_norm": 0.09912718832492828, + "learning_rate": 2.99580039850456e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2180 + }, + { + "epoch": 0.6371835903404132, + "flow_matching_loss": 0.019695677421987058, + "grad_norm": 0.10524135082960129, + "learning_rate": 2.9956758479114635e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2190 + }, + { + "epoch": 0.6400931044515565, + "flow_matching_loss": 0.019582299143075944, + "grad_norm": 0.11900739371776581, + "learning_rate": 2.9955494799311787e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2200 + }, + { + "epoch": 0.6430026185627, + "flow_matching_loss": 0.019446412939578293, + "grad_norm": 0.09284541010856628, + "learning_rate": 2.995421294717254e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2210 + }, + { + "epoch": 0.6459121326738435, + "flow_matching_loss": 0.019145808089524506, + "grad_norm": 0.12375140190124512, + "learning_rate": 2.995291292425448e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2220 + }, + { + "epoch": 0.6488216467849869, + "flow_matching_loss": 0.017825152166187762, + "grad_norm": 0.10429172217845917, + "learning_rate": 2.9951594732137248e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2230 + }, + { + "epoch": 0.6517311608961304, + "flow_matching_loss": 0.018849579617381096, + "grad_norm": 0.13928836584091187, + "learning_rate": 2.995025837242257e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2240 + }, + { + "epoch": 0.6546406750072737, + "flow_matching_loss": 0.019289367739111184, + "grad_norm": 0.12333313375711441, + "learning_rate": 2.994890384673426e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2250 + }, + { + "epoch": 0.6575501891184172, + "flow_matching_loss": 0.016385636944323777, + "grad_norm": 0.10467875748872757, + "learning_rate": 2.9947531156718174e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2260 + }, + { + "epoch": 0.6604597032295607, + "flow_matching_loss": 0.01866155294701457, + "grad_norm": 0.1073923408985138, + "learning_rate": 2.994614030404228e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2270 + }, + { + "epoch": 0.6633692173407041, + "flow_matching_loss": 0.01914365654811263, + "grad_norm": 0.0848807618021965, + "learning_rate": 2.994473129039659e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2280 + }, + { + "epoch": 0.6662787314518476, + "flow_matching_loss": 0.018379815109074115, + "grad_norm": 0.09243578463792801, + "learning_rate": 2.9943304117493183e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2290 + }, + { + "epoch": 0.669188245562991, + "flow_matching_loss": 0.019703232031315566, + "grad_norm": 0.07869881391525269, + "learning_rate": 2.9941858787066207e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2300 + }, + { + "epoch": 0.6720977596741344, + "flow_matching_loss": 0.019653220474720002, + "grad_norm": 0.06921471655368805, + "learning_rate": 2.9940395300871888e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2310 + }, + { + "epoch": 0.6750072737852778, + "flow_matching_loss": 0.01848568581044674, + "grad_norm": 0.10677073895931244, + "learning_rate": 2.993891366068849e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2320 + }, + { + "epoch": 0.6779167878964213, + "flow_matching_loss": 0.017731867823749782, + "grad_norm": 0.08929085731506348, + "learning_rate": 2.993741386831634e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2330 + }, + { + "epoch": 0.6808263020075648, + "flow_matching_loss": 0.018697912525385618, + "grad_norm": 0.10084296017885208, + "learning_rate": 2.9935895925577837e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2340 + }, + { + "epoch": 0.6837358161187082, + "flow_matching_loss": 0.020753286965191364, + "grad_norm": 0.10279800742864609, + "learning_rate": 2.9934359834317425e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2350 + }, + { + "epoch": 0.6866453302298516, + "flow_matching_loss": 0.019533946178853512, + "grad_norm": 0.1161407083272934, + "learning_rate": 2.993280559640159e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2360 + }, + { + "epoch": 0.689554844340995, + "flow_matching_loss": 0.018656597286462785, + "grad_norm": 0.08206971734762192, + "learning_rate": 2.9931233213718887e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2370 + }, + { + "epoch": 0.6924643584521385, + "flow_matching_loss": 0.020320266485214233, + "grad_norm": 0.09677283465862274, + "learning_rate": 2.992964268817991e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2380 + }, + { + "epoch": 0.695373872563282, + "flow_matching_loss": 0.01947930259630084, + "grad_norm": 0.11219554394483566, + "learning_rate": 2.9928034021717286e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2390 + }, + { + "epoch": 0.6982833866744254, + "flow_matching_loss": 0.018299046996980907, + "grad_norm": 0.1100105494260788, + "learning_rate": 2.9926407216285707e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2400 + }, + { + "epoch": 0.7011929007855688, + "flow_matching_loss": 0.021406875737011433, + "grad_norm": 0.10984180867671967, + "learning_rate": 2.9924762273861894e-05, + "loss": 0.0214, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2410 + }, + { + "epoch": 0.7041024148967122, + "flow_matching_loss": 0.020138508174568415, + "grad_norm": 0.10705562680959702, + "learning_rate": 2.992309919644461e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2420 + }, + { + "epoch": 0.7070119290078557, + "flow_matching_loss": 0.018548611644655467, + "grad_norm": 0.08153270930051804, + "learning_rate": 2.9921417986054633e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2430 + }, + { + "epoch": 0.7099214431189991, + "flow_matching_loss": 0.01939406953752041, + "grad_norm": 0.09491351246833801, + "learning_rate": 2.9919718644734815e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2440 + }, + { + "epoch": 0.7128309572301426, + "flow_matching_loss": 0.018421135656535626, + "grad_norm": 0.08720671385526657, + "learning_rate": 2.991800117455e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2450 + }, + { + "epoch": 0.715740471341286, + "flow_matching_loss": 0.02005987484008074, + "grad_norm": 0.09193934500217438, + "learning_rate": 2.9916265577587085e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2460 + }, + { + "epoch": 0.7186499854524294, + "flow_matching_loss": 0.019460596051067112, + "grad_norm": 0.09338230639696121, + "learning_rate": 2.9914511855954986e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2470 + }, + { + "epoch": 0.7215594995635729, + "flow_matching_loss": 0.01961137177422643, + "grad_norm": 0.07846778631210327, + "learning_rate": 2.9912740011784626e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2480 + }, + { + "epoch": 0.7244690136747163, + "flow_matching_loss": 0.01948198825120926, + "grad_norm": 0.12197179347276688, + "learning_rate": 2.991095004722898e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2490 + }, + { + "epoch": 0.7273785277858598, + "flow_matching_loss": 0.01944761462509632, + "grad_norm": 0.12225307524204254, + "learning_rate": 2.9909141964463017e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2500 + }, + { + "epoch": 0.7302880418970032, + "flow_matching_loss": 0.019709926564246416, + "grad_norm": 0.1040320172905922, + "learning_rate": 2.990731576568372e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2510 + }, + { + "epoch": 0.7331975560081466, + "flow_matching_loss": 0.0187241074629128, + "grad_norm": 0.11408799141645432, + "learning_rate": 2.99054714531101e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2520 + }, + { + "epoch": 0.7361070701192901, + "flow_matching_loss": 0.018447735998779535, + "grad_norm": 0.12192201614379883, + "learning_rate": 2.9903609028983178e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2530 + }, + { + "epoch": 0.7390165842304335, + "flow_matching_loss": 0.01776484688743949, + "grad_norm": 0.11018020659685135, + "learning_rate": 2.990172849556596e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2540 + }, + { + "epoch": 0.741926098341577, + "flow_matching_loss": 0.01990019017830491, + "grad_norm": 0.11644422262907028, + "learning_rate": 2.989982985514348e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2550 + }, + { + "epoch": 0.7448356124527203, + "flow_matching_loss": 0.018186735361814498, + "grad_norm": 0.07951903343200684, + "learning_rate": 2.9897913110022766e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2560 + }, + { + "epoch": 0.7477451265638638, + "flow_matching_loss": 0.019962443970143795, + "grad_norm": 0.13351088762283325, + "learning_rate": 2.989597826253284e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2570 + }, + { + "epoch": 0.7506546406750073, + "flow_matching_loss": 0.019971180334687234, + "grad_norm": 0.10336884111166, + "learning_rate": 2.989402531502473e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2580 + }, + { + "epoch": 0.7535641547861507, + "flow_matching_loss": 0.017464050464332102, + "grad_norm": 0.1109415739774704, + "learning_rate": 2.989205426987144e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2590 + }, + { + "epoch": 0.7564736688972942, + "flow_matching_loss": 0.018906620889902116, + "grad_norm": 0.09036893397569656, + "learning_rate": 2.9890065129467988e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2600 + }, + { + "epoch": 0.7593831830084375, + "flow_matching_loss": 0.018178500607609747, + "grad_norm": 0.11626835912466049, + "learning_rate": 2.9888057896231364e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2610 + }, + { + "epoch": 0.762292697119581, + "flow_matching_loss": 0.018914533499628305, + "grad_norm": 0.10080782324075699, + "learning_rate": 2.988603257260054e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2620 + }, + { + "epoch": 0.7652022112307245, + "flow_matching_loss": 0.01995226228609681, + "grad_norm": 0.11216042190790176, + "learning_rate": 2.988398916103648e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2630 + }, + { + "epoch": 0.7681117253418679, + "flow_matching_loss": 0.018609867244958878, + "grad_norm": 0.09995784610509872, + "learning_rate": 2.988192766402212e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2640 + }, + { + "epoch": 0.7710212394530114, + "flow_matching_loss": 0.018661878257989883, + "grad_norm": 0.08435508608818054, + "learning_rate": 2.9879848084062378e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2650 + }, + { + "epoch": 0.7739307535641547, + "flow_matching_loss": 0.017954350262880326, + "grad_norm": 0.07912006974220276, + "learning_rate": 2.9877750423684132e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2660 + }, + { + "epoch": 0.7768402676752982, + "flow_matching_loss": 0.017441228777170182, + "grad_norm": 0.09589036554098129, + "learning_rate": 2.987563468543624e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2670 + }, + { + "epoch": 0.7797497817864416, + "flow_matching_loss": 0.017584876623004675, + "grad_norm": 0.11459698528051376, + "learning_rate": 2.9873500871889533e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2680 + }, + { + "epoch": 0.7826592958975851, + "flow_matching_loss": 0.017514295037835838, + "grad_norm": 0.10205350071191788, + "learning_rate": 2.9871348985636778e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2690 + }, + { + "epoch": 0.7855688100087286, + "flow_matching_loss": 0.01924346461892128, + "grad_norm": 0.10688585788011551, + "learning_rate": 2.986917902929273e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2700 + }, + { + "epoch": 0.788478324119872, + "flow_matching_loss": 0.01929849525913596, + "grad_norm": 0.10314787924289703, + "learning_rate": 2.9866991005494095e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2710 + }, + { + "epoch": 0.7913878382310154, + "flow_matching_loss": 0.016679305117577314, + "grad_norm": 0.09474411606788635, + "learning_rate": 2.986478491689952e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2720 + }, + { + "epoch": 0.7942973523421588, + "flow_matching_loss": 0.016576969437301158, + "grad_norm": 0.07706945389509201, + "learning_rate": 2.986256076618961e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2730 + }, + { + "epoch": 0.7972068664533023, + "flow_matching_loss": 0.017964319605380297, + "grad_norm": 0.11411388218402863, + "learning_rate": 2.986031855606692e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2740 + }, + { + "epoch": 0.8001163805644458, + "flow_matching_loss": 0.02012732233852148, + "grad_norm": 0.12172304838895798, + "learning_rate": 2.9858058289255954e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2750 + }, + { + "epoch": 0.8030258946755892, + "flow_matching_loss": 0.01814053561538458, + "grad_norm": 0.095218226313591, + "learning_rate": 2.985577996850314e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2760 + }, + { + "epoch": 0.8059354087867326, + "flow_matching_loss": 0.017857825849205256, + "grad_norm": 0.105841726064682, + "learning_rate": 2.9853483596576856e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2770 + }, + { + "epoch": 0.808844922897876, + "flow_matching_loss": 0.02097136788070202, + "grad_norm": 0.10726771503686905, + "learning_rate": 2.9851169176267412e-05, + "loss": 0.021, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2780 + }, + { + "epoch": 0.8117544370090195, + "flow_matching_loss": 0.020247659645974636, + "grad_norm": 0.13044854998588562, + "learning_rate": 2.9848836710387044e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2790 + }, + { + "epoch": 0.814663951120163, + "flow_matching_loss": 0.020375383924692868, + "grad_norm": 0.08266475796699524, + "learning_rate": 2.9846486201769917e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2800 + }, + { + "epoch": 0.8175734652313064, + "flow_matching_loss": 0.020383878983557223, + "grad_norm": 0.10482478141784668, + "learning_rate": 2.9844117653272122e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2810 + }, + { + "epoch": 0.8204829793424498, + "flow_matching_loss": 0.018226347956806423, + "grad_norm": 0.10711780190467834, + "learning_rate": 2.9841731067771676e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2820 + }, + { + "epoch": 0.8233924934535932, + "flow_matching_loss": 0.018655291851609947, + "grad_norm": 0.12391617894172668, + "learning_rate": 2.9839326448168492e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2830 + }, + { + "epoch": 0.8263020075647367, + "flow_matching_loss": 0.01902230428531766, + "grad_norm": 0.12120126187801361, + "learning_rate": 2.983690379738442e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2840 + }, + { + "epoch": 0.8292115216758801, + "flow_matching_loss": 0.017729255277663468, + "grad_norm": 0.07339368015527725, + "learning_rate": 2.9834463118363204e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2850 + }, + { + "epoch": 0.8321210357870236, + "flow_matching_loss": 0.019286504853516816, + "grad_norm": 0.11923196911811829, + "learning_rate": 2.98320044140705e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2860 + }, + { + "epoch": 0.835030549898167, + "flow_matching_loss": 0.020085631869733334, + "grad_norm": 0.08904014527797699, + "learning_rate": 2.9829527687493868e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2870 + }, + { + "epoch": 0.8379400640093104, + "flow_matching_loss": 0.01717304652556777, + "grad_norm": 0.10706599056720734, + "learning_rate": 2.9827032941642765e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2880 + }, + { + "epoch": 0.8408495781204539, + "flow_matching_loss": 0.01837529828771949, + "grad_norm": 0.11302343755960464, + "learning_rate": 2.9824520179548535e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2890 + }, + { + "epoch": 0.8437590922315973, + "flow_matching_loss": 0.01735926577821374, + "grad_norm": 0.09746816754341125, + "learning_rate": 2.9821989404264428e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2900 + }, + { + "epoch": 0.8466686063427408, + "flow_matching_loss": 0.018904269952327014, + "grad_norm": 0.0711943730711937, + "learning_rate": 2.9819440618865572e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2910 + }, + { + "epoch": 0.8495781204538843, + "flow_matching_loss": 0.01671912157908082, + "grad_norm": 0.0913877859711647, + "learning_rate": 2.981687382644898e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2920 + }, + { + "epoch": 0.8524876345650276, + "flow_matching_loss": 0.018588849902153017, + "grad_norm": 0.07872047275304794, + "learning_rate": 2.9814289030133556e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2930 + }, + { + "epoch": 0.8553971486761711, + "flow_matching_loss": 0.017932650446891785, + "grad_norm": 0.08371059596538544, + "learning_rate": 2.9811686233060058e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2940 + }, + { + "epoch": 0.8583066627873145, + "flow_matching_loss": 0.016837144922465085, + "grad_norm": 0.08201240003108978, + "learning_rate": 2.9809065438391138e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2950 + }, + { + "epoch": 0.861216176898458, + "flow_matching_loss": 0.018669214565306902, + "grad_norm": 0.11473485827445984, + "learning_rate": 2.9806426649311304e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2960 + }, + { + "epoch": 0.8641256910096013, + "flow_matching_loss": 0.019387588649988175, + "grad_norm": 0.10103511810302734, + "learning_rate": 2.980376986902694e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2970 + }, + { + "epoch": 0.8670352051207448, + "flow_matching_loss": 0.01833098204806447, + "grad_norm": 0.09916017204523087, + "learning_rate": 2.9801095100766276e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2980 + }, + { + "epoch": 0.8699447192318883, + "flow_matching_loss": 0.01812260625883937, + "grad_norm": 0.09585433453321457, + "learning_rate": 2.979840234777942e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 2990 + }, + { + "epoch": 0.8728542333430317, + "flow_matching_loss": 0.01848535854369402, + "grad_norm": 0.0918707624077797, + "learning_rate": 2.979569161333831e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3000 + }, + { + "epoch": 0.8757637474541752, + "flow_matching_loss": 0.019470172561705114, + "grad_norm": 0.10949733108282089, + "learning_rate": 2.9792962900736744e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3010 + }, + { + "epoch": 0.8786732615653186, + "flow_matching_loss": 0.018322648573666812, + "grad_norm": 0.08958007395267487, + "learning_rate": 2.9790216213290372e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3020 + }, + { + "epoch": 0.881582775676462, + "flow_matching_loss": 0.017954500485211612, + "grad_norm": 0.0973551943898201, + "learning_rate": 2.9787451554336672e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3030 + }, + { + "epoch": 0.8844922897876055, + "flow_matching_loss": 0.019681660644710063, + "grad_norm": 0.1024533361196518, + "learning_rate": 2.9784668927234972e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3040 + }, + { + "epoch": 0.8874018038987489, + "flow_matching_loss": 0.017294289357960225, + "grad_norm": 0.0963742583990097, + "learning_rate": 2.9781868335366418e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3050 + }, + { + "epoch": 0.8903113180098924, + "flow_matching_loss": 0.020310640707612036, + "grad_norm": 0.1114104837179184, + "learning_rate": 2.9779049782134e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3060 + }, + { + "epoch": 0.8932208321210358, + "flow_matching_loss": 0.01913429582491517, + "grad_norm": 0.10190286487340927, + "learning_rate": 2.9776213270962523e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3070 + }, + { + "epoch": 0.8961303462321792, + "flow_matching_loss": 0.0171782236546278, + "grad_norm": 0.10903666913509369, + "learning_rate": 2.977335880529861e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3080 + }, + { + "epoch": 0.8990398603433226, + "flow_matching_loss": 0.01622651591897011, + "grad_norm": 0.08125019818544388, + "learning_rate": 2.977048638861072e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3090 + }, + { + "epoch": 0.9019493744544661, + "flow_matching_loss": 0.018778041750192643, + "grad_norm": 0.08775831013917923, + "learning_rate": 2.97675960243891e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3100 + }, + { + "epoch": 0.9048588885656096, + "flow_matching_loss": 0.018334154784679414, + "grad_norm": 0.13157816231250763, + "learning_rate": 2.976468771614582e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3110 + }, + { + "epoch": 0.907768402676753, + "flow_matching_loss": 0.01778433732688427, + "grad_norm": 0.09423839300870895, + "learning_rate": 2.9761761467414736e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3120 + }, + { + "epoch": 0.9106779167878964, + "flow_matching_loss": 0.016575857251882552, + "grad_norm": 0.07193324714899063, + "learning_rate": 2.9758817281751538e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3130 + }, + { + "epoch": 0.9135874308990398, + "flow_matching_loss": 0.016401616297662257, + "grad_norm": 0.10706868022680283, + "learning_rate": 2.975585516273367e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3140 + }, + { + "epoch": 0.9164969450101833, + "flow_matching_loss": 0.01829546596854925, + "grad_norm": 0.07814232259988785, + "learning_rate": 2.975287511396039e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3150 + }, + { + "epoch": 0.9194064591213268, + "flow_matching_loss": 0.017188655305653812, + "grad_norm": 0.09994473308324814, + "learning_rate": 2.9749877139052745e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3160 + }, + { + "epoch": 0.9223159732324702, + "flow_matching_loss": 0.017632752005010844, + "grad_norm": 0.10922437906265259, + "learning_rate": 2.9746861241653555e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3170 + }, + { + "epoch": 0.9252254873436137, + "flow_matching_loss": 0.019671374931931496, + "grad_norm": 0.11382201313972473, + "learning_rate": 2.9743827425427416e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3180 + }, + { + "epoch": 0.928135001454757, + "flow_matching_loss": 0.017244889214634896, + "grad_norm": 0.11926373094320297, + "learning_rate": 2.9740775694060704e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3190 + }, + { + "epoch": 0.9310445155659005, + "flow_matching_loss": 0.01974462904036045, + "grad_norm": 0.09741265326738358, + "learning_rate": 2.973770605126156e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3200 + }, + { + "epoch": 0.9339540296770439, + "flow_matching_loss": 0.02007892346009612, + "grad_norm": 0.09367163479328156, + "learning_rate": 2.9734618500759886e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3210 + }, + { + "epoch": 0.9368635437881874, + "flow_matching_loss": 0.018431354593485594, + "grad_norm": 0.12068048119544983, + "learning_rate": 2.9731513046307356e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3220 + }, + { + "epoch": 0.9397730578993309, + "flow_matching_loss": 0.018067772407084702, + "grad_norm": 0.12101586163043976, + "learning_rate": 2.972838969167738e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3230 + }, + { + "epoch": 0.9426825720104742, + "flow_matching_loss": 0.016623569931834937, + "grad_norm": 0.08586813509464264, + "learning_rate": 2.972524844066513e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3240 + }, + { + "epoch": 0.9455920861216177, + "flow_matching_loss": 0.01916105281561613, + "grad_norm": 0.10989166796207428, + "learning_rate": 2.9722089297087524e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3250 + }, + { + "epoch": 0.9485016002327611, + "flow_matching_loss": 0.018497587833553552, + "grad_norm": 0.08806995302438736, + "learning_rate": 2.971891226478322e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3260 + }, + { + "epoch": 0.9514111143439046, + "flow_matching_loss": 0.016648671496659516, + "grad_norm": 0.11021354794502258, + "learning_rate": 2.9715717347612615e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3270 + }, + { + "epoch": 0.9543206284550481, + "flow_matching_loss": 0.018388393986970186, + "grad_norm": 0.10511526465415955, + "learning_rate": 2.971250454945783e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3280 + }, + { + "epoch": 0.9572301425661914, + "flow_matching_loss": 0.01740486789494753, + "grad_norm": 0.09247571229934692, + "learning_rate": 2.970927387422272e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3290 + }, + { + "epoch": 0.9601396566773349, + "flow_matching_loss": 0.019228547997772693, + "grad_norm": 0.10625152289867401, + "learning_rate": 2.9706025325832858e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3300 + }, + { + "epoch": 0.9630491707884783, + "flow_matching_loss": 0.019225964974611997, + "grad_norm": 0.10350123047828674, + "learning_rate": 2.9702758908235534e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3310 + }, + { + "epoch": 0.9659586848996218, + "flow_matching_loss": 0.020313945598900318, + "grad_norm": 0.0902726873755455, + "learning_rate": 2.9699474625399765e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3320 + }, + { + "epoch": 0.9688681990107652, + "flow_matching_loss": 0.019294144678860904, + "grad_norm": 0.12675310671329498, + "learning_rate": 2.9696172481316246e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3330 + }, + { + "epoch": 0.9717777131219086, + "flow_matching_loss": 0.017797700595110655, + "grad_norm": 0.11727787554264069, + "learning_rate": 2.9692852479997407e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3340 + }, + { + "epoch": 0.9746872272330521, + "flow_matching_loss": 0.01769801564514637, + "grad_norm": 0.11914455145597458, + "learning_rate": 2.9689514625477364e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3350 + }, + { + "epoch": 0.9775967413441955, + "flow_matching_loss": 0.02099292604252696, + "grad_norm": 0.11208023130893707, + "learning_rate": 2.968615892181192e-05, + "loss": 0.021, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3360 + }, + { + "epoch": 0.980506255455339, + "flow_matching_loss": 0.01800288539379835, + "grad_norm": 0.1126713976264, + "learning_rate": 2.9682785373078567e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3370 + }, + { + "epoch": 0.9834157695664824, + "flow_matching_loss": 0.02023499784991145, + "grad_norm": 0.1046132817864418, + "learning_rate": 2.9679393983376493e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3380 + }, + { + "epoch": 0.9863252836776258, + "flow_matching_loss": 0.01896311677992344, + "grad_norm": 0.14581292867660522, + "learning_rate": 2.9675984756826545e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3390 + }, + { + "epoch": 0.9892347977887693, + "flow_matching_loss": 0.01821328653022647, + "grad_norm": 0.10426869988441467, + "learning_rate": 2.9672557697571273e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3400 + }, + { + "epoch": 0.9921443118999127, + "flow_matching_loss": 0.021029244549572466, + "grad_norm": 0.11001290380954742, + "learning_rate": 2.966911280977486e-05, + "loss": 0.021, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3410 + }, + { + "epoch": 0.9950538260110562, + "flow_matching_loss": 0.018004207126796245, + "grad_norm": 0.09236831963062286, + "learning_rate": 2.966565009762318e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3420 + }, + { + "epoch": 0.9979633401221996, + "flow_matching_loss": 0.018140034284442663, + "grad_norm": 0.09382433444261551, + "learning_rate": 2.966216956532375e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3430 + }, + { + "epoch": 1.000872854233343, + "flow_matching_loss": 0.01851024329662323, + "grad_norm": 0.10031302273273468, + "learning_rate": 2.9658671217105744e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3440 + }, + { + "epoch": 1.0037823683444864, + "flow_matching_loss": 0.018028015177696942, + "grad_norm": 0.12977974116802216, + "learning_rate": 2.965515505721999e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3450 + }, + { + "epoch": 1.00669188245563, + "flow_matching_loss": 0.019814448338001965, + "grad_norm": 0.08032368868589401, + "learning_rate": 2.9651621089938948e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3460 + }, + { + "epoch": 1.0096013965667734, + "flow_matching_loss": 0.016713718045502902, + "grad_norm": 0.08694867789745331, + "learning_rate": 2.9648069319556725e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3470 + }, + { + "epoch": 1.0125109106779169, + "flow_matching_loss": 0.018012770265340806, + "grad_norm": 0.0849858820438385, + "learning_rate": 2.9644499750389057e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3480 + }, + { + "epoch": 1.0154204247890601, + "flow_matching_loss": 0.018896611500531436, + "grad_norm": 0.07064961642026901, + "learning_rate": 2.9640912386773305e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3490 + }, + { + "epoch": 1.0183299389002036, + "flow_matching_loss": 0.017089961376041175, + "grad_norm": 0.08646975457668304, + "learning_rate": 2.9637307233068455e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3500 + }, + { + "epoch": 1.021239453011347, + "flow_matching_loss": 0.017704608850181103, + "grad_norm": 0.09774674475193024, + "learning_rate": 2.9633684293655108e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3510 + }, + { + "epoch": 1.0241489671224906, + "flow_matching_loss": 0.017291402257978917, + "grad_norm": 0.07876400649547577, + "learning_rate": 2.963004357293548e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3520 + }, + { + "epoch": 1.027058481233634, + "flow_matching_loss": 0.016510333213955165, + "grad_norm": 0.07524371892213821, + "learning_rate": 2.962638507533339e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3530 + }, + { + "epoch": 1.0299679953447773, + "flow_matching_loss": 0.01775587061420083, + "grad_norm": 0.07280609756708145, + "learning_rate": 2.962270880529425e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3540 + }, + { + "epoch": 1.0328775094559208, + "flow_matching_loss": 0.017568336520344018, + "grad_norm": 0.08158526569604874, + "learning_rate": 2.9619014767285084e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3550 + }, + { + "epoch": 1.0357870235670643, + "flow_matching_loss": 0.017553476337343454, + "grad_norm": 0.07003781199455261, + "learning_rate": 2.9615302965794488e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3560 + }, + { + "epoch": 1.0386965376782078, + "flow_matching_loss": 0.018184498883783817, + "grad_norm": 0.07239402830600739, + "learning_rate": 2.961157340533266e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3570 + }, + { + "epoch": 1.0416060517893513, + "flow_matching_loss": 0.017108201421797274, + "grad_norm": 0.10114186257123947, + "learning_rate": 2.9607826090431364e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3580 + }, + { + "epoch": 1.0445155659004945, + "flow_matching_loss": 0.01818251721560955, + "grad_norm": 0.09527178108692169, + "learning_rate": 2.9604061025643936e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3590 + }, + { + "epoch": 1.047425080011638, + "flow_matching_loss": 0.018562502786517145, + "grad_norm": 0.08926309645175934, + "learning_rate": 2.9600278215545292e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3600 + }, + { + "epoch": 1.0503345941227815, + "flow_matching_loss": 0.016624342650175095, + "grad_norm": 0.08077188581228256, + "learning_rate": 2.95964776647319e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3610 + }, + { + "epoch": 1.053244108233925, + "flow_matching_loss": 0.019081162102520467, + "grad_norm": 0.07033433765172958, + "learning_rate": 2.9592659377821787e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3620 + }, + { + "epoch": 1.0561536223450685, + "flow_matching_loss": 0.016871089302003383, + "grad_norm": 0.07533829659223557, + "learning_rate": 2.9588823359454528e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3630 + }, + { + "epoch": 1.0590631364562118, + "flow_matching_loss": 0.018510181363672017, + "grad_norm": 0.08127737045288086, + "learning_rate": 2.958496961429126e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3640 + }, + { + "epoch": 1.0619726505673552, + "flow_matching_loss": 0.018526287842541934, + "grad_norm": 0.12658695876598358, + "learning_rate": 2.9581098147014628e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3650 + }, + { + "epoch": 1.0648821646784987, + "flow_matching_loss": 0.01906443266198039, + "grad_norm": 0.07055851817131042, + "learning_rate": 2.9577208962328843e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3660 + }, + { + "epoch": 1.0677916787896422, + "flow_matching_loss": 0.01712263636291027, + "grad_norm": 0.08822812885046005, + "learning_rate": 2.9573302064959633e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3670 + }, + { + "epoch": 1.0707011929007855, + "flow_matching_loss": 0.01943627716973424, + "grad_norm": 0.10270032286643982, + "learning_rate": 2.9569377459654235e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3680 + }, + { + "epoch": 1.073610707011929, + "flow_matching_loss": 0.019090043753385542, + "grad_norm": 0.11913636326789856, + "learning_rate": 2.9565435151181424e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3690 + }, + { + "epoch": 1.0765202211230724, + "flow_matching_loss": 0.018570251110941172, + "grad_norm": 0.09321325272321701, + "learning_rate": 2.9561475144331473e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3700 + }, + { + "epoch": 1.079429735234216, + "flow_matching_loss": 0.01618027873337269, + "grad_norm": 0.08255013823509216, + "learning_rate": 2.9557497443916162e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3710 + }, + { + "epoch": 1.0823392493453594, + "flow_matching_loss": 0.017737521417438985, + "grad_norm": 0.10429845005273819, + "learning_rate": 2.9553502054768774e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3720 + }, + { + "epoch": 1.0852487634565027, + "flow_matching_loss": 0.018365352880209684, + "grad_norm": 0.10847841203212738, + "learning_rate": 2.954948898174409e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3730 + }, + { + "epoch": 1.0881582775676462, + "flow_matching_loss": 0.018504612054675816, + "grad_norm": 0.08747928589582443, + "learning_rate": 2.954545822971836e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3740 + }, + { + "epoch": 1.0910677916787896, + "flow_matching_loss": 0.017504666186869144, + "grad_norm": 0.09519334882497787, + "learning_rate": 2.9541409803589335e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3750 + }, + { + "epoch": 1.0939773057899331, + "flow_matching_loss": 0.018213493190705775, + "grad_norm": 0.0996343269944191, + "learning_rate": 2.9537343708276235e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3760 + }, + { + "epoch": 1.0968868199010766, + "flow_matching_loss": 0.017857956886291503, + "grad_norm": 0.08563773334026337, + "learning_rate": 2.953325994871975e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3770 + }, + { + "epoch": 1.0997963340122199, + "flow_matching_loss": 0.018901527021080257, + "grad_norm": 0.1031501367688179, + "learning_rate": 2.9529158529882023e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3780 + }, + { + "epoch": 1.1027058481233634, + "flow_matching_loss": 0.015986056718975305, + "grad_norm": 0.1090025007724762, + "learning_rate": 2.9525039456746677e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3790 + }, + { + "epoch": 1.1056153622345068, + "flow_matching_loss": 0.01665734425187111, + "grad_norm": 0.10821790993213654, + "learning_rate": 2.952090273431877e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3800 + }, + { + "epoch": 1.1085248763456503, + "flow_matching_loss": 0.01686789495870471, + "grad_norm": 0.08292470872402191, + "learning_rate": 2.9516748367624808e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3810 + }, + { + "epoch": 1.1114343904567938, + "flow_matching_loss": 0.01849607201293111, + "grad_norm": 0.08133069425821304, + "learning_rate": 2.951257636171274e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3820 + }, + { + "epoch": 1.114343904567937, + "flow_matching_loss": 0.019985859096050263, + "grad_norm": 0.09021090716123581, + "learning_rate": 2.9508386721651944e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3830 + }, + { + "epoch": 1.1172534186790806, + "flow_matching_loss": 0.0183102079667151, + "grad_norm": 0.08560869842767715, + "learning_rate": 2.9504179452533228e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3840 + }, + { + "epoch": 1.120162932790224, + "flow_matching_loss": 0.018546116072684526, + "grad_norm": 0.13800564408302307, + "learning_rate": 2.9499954559468826e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3850 + }, + { + "epoch": 1.1230724469013675, + "flow_matching_loss": 0.01735885962843895, + "grad_norm": 0.08593789488077164, + "learning_rate": 2.9495712047592378e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3860 + }, + { + "epoch": 1.1259819610125108, + "flow_matching_loss": 0.01754489177837968, + "grad_norm": 0.08763536810874939, + "learning_rate": 2.949145192205893e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3870 + }, + { + "epoch": 1.1288914751236543, + "flow_matching_loss": 0.01871220227330923, + "grad_norm": 0.10825508832931519, + "learning_rate": 2.9487174188044953e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3880 + }, + { + "epoch": 1.1318009892347978, + "flow_matching_loss": 0.018772626109421254, + "grad_norm": 0.09162640571594238, + "learning_rate": 2.9482878850748276e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3890 + }, + { + "epoch": 1.1347105033459413, + "flow_matching_loss": 0.017574034724384548, + "grad_norm": 0.08956382423639297, + "learning_rate": 2.9478565915388157e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3900 + }, + { + "epoch": 1.1376200174570847, + "flow_matching_loss": 0.015855283103883268, + "grad_norm": 0.09416811168193817, + "learning_rate": 2.9474235387205207e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3910 + }, + { + "epoch": 1.140529531568228, + "flow_matching_loss": 0.01824382385239005, + "grad_norm": 0.08198630809783936, + "learning_rate": 2.946988727146143e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3920 + }, + { + "epoch": 1.1434390456793715, + "flow_matching_loss": 0.01752053163945675, + "grad_norm": 0.09703607857227325, + "learning_rate": 2.9465521573440196e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3930 + }, + { + "epoch": 1.146348559790515, + "flow_matching_loss": 0.017832505516707897, + "grad_norm": 0.10077141970396042, + "learning_rate": 2.946113829844624e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3940 + }, + { + "epoch": 1.1492580739016585, + "flow_matching_loss": 0.01803307430818677, + "grad_norm": 0.10195383429527283, + "learning_rate": 2.945673745180566e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3950 + }, + { + "epoch": 1.152167588012802, + "flow_matching_loss": 0.018184073269367218, + "grad_norm": 0.0788196325302124, + "learning_rate": 2.945231903886589e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3960 + }, + { + "epoch": 1.1550771021239452, + "flow_matching_loss": 0.018025304563343524, + "grad_norm": 0.08540042489767075, + "learning_rate": 2.9447883064995728e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3970 + }, + { + "epoch": 1.1579866162350887, + "flow_matching_loss": 0.01903387373313308, + "grad_norm": 0.09530279785394669, + "learning_rate": 2.9443429535585288e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3980 + }, + { + "epoch": 1.1608961303462322, + "flow_matching_loss": 0.020587613712996243, + "grad_norm": 0.11037687212228775, + "learning_rate": 2.9438958456046044e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 3990 + }, + { + "epoch": 1.1638056444573757, + "flow_matching_loss": 0.017603846732527018, + "grad_norm": 0.10139672458171844, + "learning_rate": 2.9434469831810765e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4000 + }, + { + "epoch": 1.1667151585685192, + "flow_matching_loss": 0.01641847686842084, + "grad_norm": 0.07711735367774963, + "learning_rate": 2.9429963668333566e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4010 + }, + { + "epoch": 1.1696246726796624, + "flow_matching_loss": 0.018554951902478932, + "grad_norm": 0.1207059696316719, + "learning_rate": 2.9425439971089845e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4020 + }, + { + "epoch": 1.172534186790806, + "flow_matching_loss": 0.016874324530363083, + "grad_norm": 0.073297418653965, + "learning_rate": 2.9420898745576336e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4030 + }, + { + "epoch": 1.1754437009019494, + "flow_matching_loss": 0.017677681427448987, + "grad_norm": 0.12059706449508667, + "learning_rate": 2.9416339997311044e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4040 + }, + { + "epoch": 1.1783532150130929, + "flow_matching_loss": 0.018725256156176328, + "grad_norm": 0.10837530344724655, + "learning_rate": 2.941176373183329e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4050 + }, + { + "epoch": 1.1812627291242364, + "flow_matching_loss": 0.018330416921526194, + "grad_norm": 0.1228313073515892, + "learning_rate": 2.9407169954703653e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4060 + }, + { + "epoch": 1.1841722432353796, + "flow_matching_loss": 0.017222365736961363, + "grad_norm": 0.09575636684894562, + "learning_rate": 2.9402558671504018e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4070 + }, + { + "epoch": 1.187081757346523, + "flow_matching_loss": 0.018668787088245153, + "grad_norm": 0.0976257175207138, + "learning_rate": 2.9397929887837524e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4080 + }, + { + "epoch": 1.1899912714576666, + "flow_matching_loss": 0.017625673115253447, + "grad_norm": 0.10228557884693146, + "learning_rate": 2.939328360932858e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4090 + }, + { + "epoch": 1.19290078556881, + "flow_matching_loss": 0.018832901399582623, + "grad_norm": 0.08967453241348267, + "learning_rate": 2.938861984162285e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4100 + }, + { + "epoch": 1.1958102996799536, + "flow_matching_loss": 0.016413830872625112, + "grad_norm": 0.08499201387166977, + "learning_rate": 2.938393859038726e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4110 + }, + { + "epoch": 1.1987198137910968, + "flow_matching_loss": 0.017256331630051136, + "grad_norm": 0.08433124423027039, + "learning_rate": 2.9379239861309964e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4120 + }, + { + "epoch": 1.2016293279022403, + "flow_matching_loss": 0.017418409790843727, + "grad_norm": 0.06862207502126694, + "learning_rate": 2.937452366010036e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4130 + }, + { + "epoch": 1.2045388420133838, + "flow_matching_loss": 0.017175285890698434, + "grad_norm": 0.11601414531469345, + "learning_rate": 2.936978999248908e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4140 + }, + { + "epoch": 1.2074483561245273, + "flow_matching_loss": 0.017895869445055723, + "grad_norm": 0.11593836545944214, + "learning_rate": 2.936503886422798e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4150 + }, + { + "epoch": 1.2103578702356708, + "flow_matching_loss": 0.017054905649274588, + "grad_norm": 0.09781889617443085, + "learning_rate": 2.9360270281090122e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4160 + }, + { + "epoch": 1.213267384346814, + "flow_matching_loss": 0.018594330083578824, + "grad_norm": 0.09650484472513199, + "learning_rate": 2.935548424886979e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4170 + }, + { + "epoch": 1.2161768984579575, + "flow_matching_loss": 0.017697219830006362, + "grad_norm": 0.11229865998029709, + "learning_rate": 2.935068077338246e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4180 + }, + { + "epoch": 1.219086412569101, + "flow_matching_loss": 0.016808687802404167, + "grad_norm": 0.07744954526424408, + "learning_rate": 2.934585986046481e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4190 + }, + { + "epoch": 1.2219959266802445, + "flow_matching_loss": 0.018246331345289947, + "grad_norm": 0.07977195829153061, + "learning_rate": 2.93410215159747e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4200 + }, + { + "epoch": 1.2249054407913877, + "flow_matching_loss": 0.017191207874566315, + "grad_norm": 0.06756410002708435, + "learning_rate": 2.933616574579118e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4210 + }, + { + "epoch": 1.2278149549025312, + "flow_matching_loss": 0.01754891499876976, + "grad_norm": 0.11626248806715012, + "learning_rate": 2.933129255581447e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4220 + }, + { + "epoch": 1.2307244690136747, + "flow_matching_loss": 0.019722874648869036, + "grad_norm": 0.07443170994520187, + "learning_rate": 2.9326401951965954e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4230 + }, + { + "epoch": 1.2336339831248182, + "flow_matching_loss": 0.017063596099615098, + "grad_norm": 0.0938928872346878, + "learning_rate": 2.9321493940188165e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4240 + }, + { + "epoch": 1.2365434972359617, + "flow_matching_loss": 0.017973087914288043, + "grad_norm": 0.0905243456363678, + "learning_rate": 2.931656852644482e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4250 + }, + { + "epoch": 1.239453011347105, + "flow_matching_loss": 0.01711770324036479, + "grad_norm": 0.08731282502412796, + "learning_rate": 2.9311625716720748e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4260 + }, + { + "epoch": 1.2423625254582484, + "flow_matching_loss": 0.018538788426667452, + "grad_norm": 0.08525966852903366, + "learning_rate": 2.9306665517021927e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4270 + }, + { + "epoch": 1.245272039569392, + "flow_matching_loss": 0.017017473839223383, + "grad_norm": 0.09485193341970444, + "learning_rate": 2.9301687933375483e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4280 + }, + { + "epoch": 1.2481815536805354, + "flow_matching_loss": 0.019961184356361628, + "grad_norm": 0.09641290456056595, + "learning_rate": 2.929669297182963e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4290 + }, + { + "epoch": 1.2510910677916787, + "flow_matching_loss": 0.018259262572973965, + "grad_norm": 0.09967435151338577, + "learning_rate": 2.929168063845373e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4300 + }, + { + "epoch": 1.2540005819028222, + "flow_matching_loss": 0.018136469274759294, + "grad_norm": 0.08038952201604843, + "learning_rate": 2.9286650939338236e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4310 + }, + { + "epoch": 1.2569100960139656, + "flow_matching_loss": 0.017794671840965746, + "grad_norm": 0.12716582417488098, + "learning_rate": 2.9281603880594706e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4320 + }, + { + "epoch": 1.2598196101251091, + "flow_matching_loss": 0.01874416312202811, + "grad_norm": 0.11686268448829651, + "learning_rate": 2.9276539468355796e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4330 + }, + { + "epoch": 1.2627291242362526, + "flow_matching_loss": 0.01879497552290559, + "grad_norm": 0.14414700865745544, + "learning_rate": 2.9271457708775244e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4340 + }, + { + "epoch": 1.2656386383473959, + "flow_matching_loss": 0.01588035998865962, + "grad_norm": 0.09816278517246246, + "learning_rate": 2.9266358608027863e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4350 + }, + { + "epoch": 1.2685481524585394, + "flow_matching_loss": 0.0186996141448617, + "grad_norm": 0.12927688658237457, + "learning_rate": 2.9261242172309543e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4360 + }, + { + "epoch": 1.2714576665696828, + "flow_matching_loss": 0.01746635101735592, + "grad_norm": 0.08532265573740005, + "learning_rate": 2.925610840783723e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4370 + }, + { + "epoch": 1.2743671806808263, + "flow_matching_loss": 0.01657203268259764, + "grad_norm": 0.0956910103559494, + "learning_rate": 2.9250957320848946e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4380 + }, + { + "epoch": 1.2772766947919698, + "flow_matching_loss": 0.018387085013091563, + "grad_norm": 0.09720522165298462, + "learning_rate": 2.924578891760374e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4390 + }, + { + "epoch": 1.280186208903113, + "flow_matching_loss": 0.017194008361548184, + "grad_norm": 0.09337832033634186, + "learning_rate": 2.924060320438171e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4400 + }, + { + "epoch": 1.2830957230142566, + "flow_matching_loss": 0.017932888865470887, + "grad_norm": 0.12469319999217987, + "learning_rate": 2.9235400187483978e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4410 + }, + { + "epoch": 1.2860052371254, + "flow_matching_loss": 0.01848166957497597, + "grad_norm": 0.09238400310277939, + "learning_rate": 2.9230179873232716e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4420 + }, + { + "epoch": 1.2889147512365435, + "flow_matching_loss": 0.017232669051736593, + "grad_norm": 0.142038494348526, + "learning_rate": 2.922494226797108e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4430 + }, + { + "epoch": 1.291824265347687, + "flow_matching_loss": 0.018479018844664095, + "grad_norm": 0.07484936714172363, + "learning_rate": 2.9219687378063275e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4440 + }, + { + "epoch": 1.2947337794588303, + "flow_matching_loss": 0.01841676440089941, + "grad_norm": 0.11738834530115128, + "learning_rate": 2.921441520989447e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4450 + }, + { + "epoch": 1.2976432935699738, + "flow_matching_loss": 0.018146234937012194, + "grad_norm": 0.10958316177129745, + "learning_rate": 2.9209125769870854e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4460 + }, + { + "epoch": 1.3005528076811173, + "flow_matching_loss": 0.020034205354750156, + "grad_norm": 0.11122437566518784, + "learning_rate": 2.92038190644196e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4470 + }, + { + "epoch": 1.3034623217922607, + "flow_matching_loss": 0.019206910766661166, + "grad_norm": 0.1333494484424591, + "learning_rate": 2.9198495099988843e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4480 + }, + { + "epoch": 1.3063718359034042, + "flow_matching_loss": 0.017751287575811146, + "grad_norm": 0.0669863149523735, + "learning_rate": 2.9193153883047712e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4490 + }, + { + "epoch": 1.3092813500145475, + "flow_matching_loss": 0.018790558725595475, + "grad_norm": 0.1233728751540184, + "learning_rate": 2.9187795420086287e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4500 + }, + { + "epoch": 1.312190864125691, + "flow_matching_loss": 0.019940192345529795, + "grad_norm": 0.0998883992433548, + "learning_rate": 2.9182419717615602e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4510 + }, + { + "epoch": 1.3151003782368345, + "flow_matching_loss": 0.018871088698506355, + "grad_norm": 0.11406343430280685, + "learning_rate": 2.917702678216765e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4520 + }, + { + "epoch": 1.318009892347978, + "flow_matching_loss": 0.016867879033088683, + "grad_norm": 0.08203241229057312, + "learning_rate": 2.9171616620295347e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4530 + }, + { + "epoch": 1.3209194064591214, + "flow_matching_loss": 0.018540493119508027, + "grad_norm": 0.11034803837537766, + "learning_rate": 2.9166189238572554e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4540 + }, + { + "epoch": 1.3238289205702647, + "flow_matching_loss": 0.017267818655818702, + "grad_norm": 0.06641562283039093, + "learning_rate": 2.9160744643594052e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4550 + }, + { + "epoch": 1.3267384346814082, + "flow_matching_loss": 0.01710999021306634, + "grad_norm": 0.0827474445104599, + "learning_rate": 2.915528284197554e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4560 + }, + { + "epoch": 1.3296479487925517, + "flow_matching_loss": 0.0171516721136868, + "grad_norm": 0.07506970316171646, + "learning_rate": 2.914980384035362e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4570 + }, + { + "epoch": 1.3325574629036951, + "flow_matching_loss": 0.017941261269152166, + "grad_norm": 0.08414442092180252, + "learning_rate": 2.9144307645385787e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4580 + }, + { + "epoch": 1.3354669770148386, + "flow_matching_loss": 0.01724579920992255, + "grad_norm": 0.08578333258628845, + "learning_rate": 2.9138794263750454e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4590 + }, + { + "epoch": 1.338376491125982, + "flow_matching_loss": 0.018318196292966604, + "grad_norm": 0.08672690391540527, + "learning_rate": 2.913326370214688e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4600 + }, + { + "epoch": 1.3412860052371254, + "flow_matching_loss": 0.017566915322095155, + "grad_norm": 0.0760151743888855, + "learning_rate": 2.9127715967295234e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4610 + }, + { + "epoch": 1.3441955193482689, + "flow_matching_loss": 0.01760668670758605, + "grad_norm": 0.09774770587682724, + "learning_rate": 2.912215106593653e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4620 + }, + { + "epoch": 1.3471050334594123, + "flow_matching_loss": 0.01794936563819647, + "grad_norm": 0.07294467836618423, + "learning_rate": 2.911656900483265e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4630 + }, + { + "epoch": 1.3500145475705558, + "flow_matching_loss": 0.018472930509597064, + "grad_norm": 0.07920266687870026, + "learning_rate": 2.9110969790766322e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4640 + }, + { + "epoch": 1.352924061681699, + "flow_matching_loss": 0.01784668294712901, + "grad_norm": 0.0815187469124794, + "learning_rate": 2.910535343054112e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4650 + }, + { + "epoch": 1.3558335757928426, + "flow_matching_loss": 0.016710187681019305, + "grad_norm": 0.11634989827871323, + "learning_rate": 2.909971993098145e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4660 + }, + { + "epoch": 1.358743089903986, + "flow_matching_loss": 0.019587599392980336, + "grad_norm": 0.09780547767877579, + "learning_rate": 2.9094069298932552e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4670 + }, + { + "epoch": 1.3616526040151296, + "flow_matching_loss": 0.01849818490445614, + "grad_norm": 0.10789816826581955, + "learning_rate": 2.908840154126047e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4680 + }, + { + "epoch": 1.364562118126273, + "flow_matching_loss": 0.01533505255356431, + "grad_norm": 0.062022749334573746, + "learning_rate": 2.9082716664852063e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4690 + }, + { + "epoch": 1.3674716322374163, + "flow_matching_loss": 0.018212305568158626, + "grad_norm": 0.0765312984585762, + "learning_rate": 2.9077014676615e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4700 + }, + { + "epoch": 1.3703811463485598, + "flow_matching_loss": 0.016794453840702772, + "grad_norm": 0.09144172817468643, + "learning_rate": 2.9071295583477724e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4710 + }, + { + "epoch": 1.3732906604597033, + "flow_matching_loss": 0.016384642105549574, + "grad_norm": 0.08697270601987839, + "learning_rate": 2.906555939238948e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4720 + }, + { + "epoch": 1.3762001745708465, + "flow_matching_loss": 0.017887113709002735, + "grad_norm": 0.08140653371810913, + "learning_rate": 2.905980611032028e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4730 + }, + { + "epoch": 1.3791096886819902, + "flow_matching_loss": 0.018021827936172484, + "grad_norm": 0.0906154066324234, + "learning_rate": 2.9054035744260904e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4740 + }, + { + "epoch": 1.3820192027931335, + "flow_matching_loss": 0.016529027745127677, + "grad_norm": 0.08482995629310608, + "learning_rate": 2.9048248301222885e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4750 + }, + { + "epoch": 1.384928716904277, + "flow_matching_loss": 0.017980134487152098, + "grad_norm": 0.07993477582931519, + "learning_rate": 2.904244378823852e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4760 + }, + { + "epoch": 1.3878382310154205, + "flow_matching_loss": 0.015756020601838828, + "grad_norm": 0.09419599175453186, + "learning_rate": 2.9036622212360836e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4770 + }, + { + "epoch": 1.3907477451265637, + "flow_matching_loss": 0.017025388684123753, + "grad_norm": 0.08274397999048233, + "learning_rate": 2.9030783580663596e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4780 + }, + { + "epoch": 1.3936572592377072, + "flow_matching_loss": 0.0172802759334445, + "grad_norm": 0.12208238989114761, + "learning_rate": 2.902492790024129e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4790 + }, + { + "epoch": 1.3965667733488507, + "flow_matching_loss": 0.01849865010008216, + "grad_norm": 0.09283638745546341, + "learning_rate": 2.901905517820912e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4800 + }, + { + "epoch": 1.3994762874599942, + "flow_matching_loss": 0.0165410072542727, + "grad_norm": 0.08133002370595932, + "learning_rate": 2.9013165421703e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4810 + }, + { + "epoch": 1.4023858015711377, + "flow_matching_loss": 0.01856268374249339, + "grad_norm": 0.08000727742910385, + "learning_rate": 2.9007258637879534e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4820 + }, + { + "epoch": 1.405295315682281, + "flow_matching_loss": 0.016700625605881213, + "grad_norm": 0.0753961056470871, + "learning_rate": 2.9001334833916025e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4830 + }, + { + "epoch": 1.4082048297934244, + "flow_matching_loss": 0.016661651153117418, + "grad_norm": 0.08640418946743011, + "learning_rate": 2.899539401701046e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4840 + }, + { + "epoch": 1.411114343904568, + "flow_matching_loss": 0.017137428838759662, + "grad_norm": 0.09864164888858795, + "learning_rate": 2.8989436194381474e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4850 + }, + { + "epoch": 1.4140238580157114, + "flow_matching_loss": 0.017694329656660557, + "grad_norm": 0.08597861230373383, + "learning_rate": 2.8983461373268404e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4860 + }, + { + "epoch": 1.4169333721268549, + "flow_matching_loss": 0.01748534757643938, + "grad_norm": 0.10314324498176575, + "learning_rate": 2.8977469560931205e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4870 + }, + { + "epoch": 1.4198428862379981, + "flow_matching_loss": 0.01680967016145587, + "grad_norm": 0.07918404042720795, + "learning_rate": 2.8971460764650502e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4880 + }, + { + "epoch": 1.4227524003491416, + "flow_matching_loss": 0.01772168409079313, + "grad_norm": 0.07928013056516647, + "learning_rate": 2.8965434991727545e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4890 + }, + { + "epoch": 1.4256619144602851, + "flow_matching_loss": 0.01727563962340355, + "grad_norm": 0.08083848655223846, + "learning_rate": 2.895939224948422e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4900 + }, + { + "epoch": 1.4285714285714286, + "flow_matching_loss": 0.01505086775869131, + "grad_norm": 0.06903116405010223, + "learning_rate": 2.895333254526303e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4910 + }, + { + "epoch": 1.431480942682572, + "flow_matching_loss": 0.017383875139057637, + "grad_norm": 0.09588684141635895, + "learning_rate": 2.8947255886427084e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4920 + }, + { + "epoch": 1.4343904567937154, + "flow_matching_loss": 0.017861897218972446, + "grad_norm": 0.08520248532295227, + "learning_rate": 2.8941162280360095e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4930 + }, + { + "epoch": 1.4372999709048588, + "flow_matching_loss": 0.018031459674239157, + "grad_norm": 0.058098725974559784, + "learning_rate": 2.8935051734466368e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4940 + }, + { + "epoch": 1.4402094850160023, + "flow_matching_loss": 0.01812290884554386, + "grad_norm": 0.08408401161432266, + "learning_rate": 2.8928924256170792e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4950 + }, + { + "epoch": 1.4431189991271458, + "flow_matching_loss": 0.018396809231489897, + "grad_norm": 0.09920460730791092, + "learning_rate": 2.8922779852918836e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4960 + }, + { + "epoch": 1.4460285132382893, + "flow_matching_loss": 0.019034580700099467, + "grad_norm": 0.09478297829627991, + "learning_rate": 2.8916618532176526e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4970 + }, + { + "epoch": 1.4489380273494326, + "flow_matching_loss": 0.01682168859988451, + "grad_norm": 0.10513750463724136, + "learning_rate": 2.891044030143045e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4980 + }, + { + "epoch": 1.451847541460576, + "flow_matching_loss": 0.01601092265918851, + "grad_norm": 0.1039203405380249, + "learning_rate": 2.890424516818774e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 4990 + }, + { + "epoch": 1.4547570555717195, + "flow_matching_loss": 0.017031166702508926, + "grad_norm": 0.0710081160068512, + "learning_rate": 2.8898033139976057e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5000 + }, + { + "epoch": 1.457666569682863, + "flow_matching_loss": 0.016943196207284926, + "grad_norm": 0.12970642745494843, + "learning_rate": 2.889180422434362e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5010 + }, + { + "epoch": 1.4605760837940065, + "flow_matching_loss": 0.016532422881573437, + "grad_norm": 0.08096943795681, + "learning_rate": 2.8885558428859136e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5020 + }, + { + "epoch": 1.4634855979051498, + "flow_matching_loss": 0.01627649664878845, + "grad_norm": 0.08339137583971024, + "learning_rate": 2.8879295761111835e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5030 + }, + { + "epoch": 1.4663951120162932, + "flow_matching_loss": 0.01844171164557338, + "grad_norm": 0.10386183857917786, + "learning_rate": 2.8873016228711452e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5040 + }, + { + "epoch": 1.4693046261274367, + "flow_matching_loss": 0.018290637247264385, + "grad_norm": 0.1413814276456833, + "learning_rate": 2.8866719839288215e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5050 + }, + { + "epoch": 1.4722141402385802, + "flow_matching_loss": 0.016997105814516543, + "grad_norm": 0.10948627442121506, + "learning_rate": 2.886040660049283e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5060 + }, + { + "epoch": 1.4751236543497237, + "flow_matching_loss": 0.016289071924984456, + "grad_norm": 0.09320095181465149, + "learning_rate": 2.8854076519996467e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5070 + }, + { + "epoch": 1.478033168460867, + "flow_matching_loss": 0.01503765545785427, + "grad_norm": 0.07379238307476044, + "learning_rate": 2.884772960549078e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5080 + }, + { + "epoch": 1.4809426825720104, + "flow_matching_loss": 0.018831222504377364, + "grad_norm": 0.09777366369962692, + "learning_rate": 2.8841365864687867e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5090 + }, + { + "epoch": 1.483852196683154, + "flow_matching_loss": 0.01591841457411647, + "grad_norm": 0.0771479606628418, + "learning_rate": 2.8834985305320276e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5100 + }, + { + "epoch": 1.4867617107942974, + "flow_matching_loss": 0.016431513521820307, + "grad_norm": 0.07898079603910446, + "learning_rate": 2.882858793514098e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5110 + }, + { + "epoch": 1.489671224905441, + "flow_matching_loss": 0.017069196980446576, + "grad_norm": 0.1075642928481102, + "learning_rate": 2.8822173761923398e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5120 + }, + { + "epoch": 1.4925807390165842, + "flow_matching_loss": 0.01585097620263696, + "grad_norm": 0.08737059682607651, + "learning_rate": 2.8815742793461343e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5130 + }, + { + "epoch": 1.4954902531277277, + "flow_matching_loss": 0.01661896863952279, + "grad_norm": 0.09448487311601639, + "learning_rate": 2.880929503756906e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5140 + }, + { + "epoch": 1.4983997672388711, + "flow_matching_loss": 0.017361583095043896, + "grad_norm": 0.11666061729192734, + "learning_rate": 2.880283050208117e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5150 + }, + { + "epoch": 1.5013092813500144, + "flow_matching_loss": 0.016455515287816525, + "grad_norm": 0.06508040428161621, + "learning_rate": 2.8796349194852704e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5160 + }, + { + "epoch": 1.504218795461158, + "flow_matching_loss": 0.01716578146442771, + "grad_norm": 0.09014807641506195, + "learning_rate": 2.8789851123759052e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5170 + }, + { + "epoch": 1.5071283095723014, + "flow_matching_loss": 0.017792343720793725, + "grad_norm": 0.09673328697681427, + "learning_rate": 2.8783336296695985e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5180 + }, + { + "epoch": 1.5100378236834449, + "flow_matching_loss": 0.01618854282423854, + "grad_norm": 0.08644711226224899, + "learning_rate": 2.877680472157964e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5190 + }, + { + "epoch": 1.5129473377945883, + "flow_matching_loss": 0.01862327391281724, + "grad_norm": 0.07266746461391449, + "learning_rate": 2.8770256406346483e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5200 + }, + { + "epoch": 1.5158568519057316, + "flow_matching_loss": 0.017012276593595742, + "grad_norm": 0.07603176683187485, + "learning_rate": 2.8763691358953348e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5210 + }, + { + "epoch": 1.5187663660168753, + "flow_matching_loss": 0.016088925302028656, + "grad_norm": 0.08934494853019714, + "learning_rate": 2.8757109587377377e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5220 + }, + { + "epoch": 1.5216758801280186, + "flow_matching_loss": 0.01628910694271326, + "grad_norm": 0.07086964696645737, + "learning_rate": 2.8750511099616047e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5230 + }, + { + "epoch": 1.524585394239162, + "flow_matching_loss": 0.0166813756339252, + "grad_norm": 0.09887402504682541, + "learning_rate": 2.8743895903687143e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5240 + }, + { + "epoch": 1.5274949083503055, + "flow_matching_loss": 0.019000981654971837, + "grad_norm": 0.10313404351472855, + "learning_rate": 2.8737264007628753e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5250 + }, + { + "epoch": 1.5304044224614488, + "flow_matching_loss": 0.015572401788085699, + "grad_norm": 0.0866900384426117, + "learning_rate": 2.8730615419499247e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5260 + }, + { + "epoch": 1.5333139365725925, + "flow_matching_loss": 0.019560024421662093, + "grad_norm": 0.13202694058418274, + "learning_rate": 2.8723950147377303e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5270 + }, + { + "epoch": 1.5362234506837358, + "flow_matching_loss": 0.017563256621360778, + "grad_norm": 0.09223472326993942, + "learning_rate": 2.871726819936184e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5280 + }, + { + "epoch": 1.5391329647948793, + "flow_matching_loss": 0.017023221775889396, + "grad_norm": 0.0844714343547821, + "learning_rate": 2.8710569583572068e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5290 + }, + { + "epoch": 1.5420424789060228, + "flow_matching_loss": 0.016103014629334212, + "grad_norm": 0.07358773052692413, + "learning_rate": 2.870385430814742e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5300 + }, + { + "epoch": 1.544951993017166, + "flow_matching_loss": 0.01680693160742521, + "grad_norm": 0.09081678837537766, + "learning_rate": 2.8697122381247605e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5310 + }, + { + "epoch": 1.5478615071283097, + "flow_matching_loss": 0.01836457820609212, + "grad_norm": 0.0926685482263565, + "learning_rate": 2.8690373811052545e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5320 + }, + { + "epoch": 1.550771021239453, + "flow_matching_loss": 0.016051399521529675, + "grad_norm": 0.07901865243911743, + "learning_rate": 2.8683608605762378e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5330 + }, + { + "epoch": 1.5536805353505965, + "flow_matching_loss": 0.01812496967613697, + "grad_norm": 0.09506777673959732, + "learning_rate": 2.8676826773597485e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5340 + }, + { + "epoch": 1.55659004946174, + "flow_matching_loss": 0.01791005441918969, + "grad_norm": 0.09674572199583054, + "learning_rate": 2.8670028322798413e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5350 + }, + { + "epoch": 1.5594995635728832, + "flow_matching_loss": 0.0170031463727355, + "grad_norm": 0.12036973237991333, + "learning_rate": 2.8663213261625932e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5360 + }, + { + "epoch": 1.562409077684027, + "flow_matching_loss": 0.018219267297536135, + "grad_norm": 0.0873119980096817, + "learning_rate": 2.865638159836098e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5370 + }, + { + "epoch": 1.5653185917951702, + "flow_matching_loss": 0.017394870799034835, + "grad_norm": 0.08233197778463364, + "learning_rate": 2.8649533341304677e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5380 + }, + { + "epoch": 1.5682281059063137, + "flow_matching_loss": 0.014961966685950756, + "grad_norm": 0.06513916701078415, + "learning_rate": 2.8642668498778297e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5390 + }, + { + "epoch": 1.5711376200174572, + "flow_matching_loss": 0.017580207064747812, + "grad_norm": 0.07748878747224808, + "learning_rate": 2.8635787079123268e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5400 + }, + { + "epoch": 1.5740471341286004, + "flow_matching_loss": 0.01698612906038761, + "grad_norm": 0.08867599815130234, + "learning_rate": 2.8628889090701175e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5410 + }, + { + "epoch": 1.576956648239744, + "flow_matching_loss": 0.01570911183953285, + "grad_norm": 0.09579691290855408, + "learning_rate": 2.862197454189371e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5420 + }, + { + "epoch": 1.5798661623508874, + "flow_matching_loss": 0.01685804072767496, + "grad_norm": 0.06944193691015244, + "learning_rate": 2.8615043441102708e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5430 + }, + { + "epoch": 1.5827756764620309, + "flow_matching_loss": 0.017252314370125532, + "grad_norm": 0.1117226630449295, + "learning_rate": 2.860809579675011e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5440 + }, + { + "epoch": 1.5856851905731744, + "flow_matching_loss": 0.016684918198734522, + "grad_norm": 0.10219967365264893, + "learning_rate": 2.860113161727796e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5450 + }, + { + "epoch": 1.5885947046843176, + "flow_matching_loss": 0.016875190660357475, + "grad_norm": 0.10607108473777771, + "learning_rate": 2.8594150911148393e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5460 + }, + { + "epoch": 1.5915042187954611, + "flow_matching_loss": 0.018059553019702433, + "grad_norm": 0.10895755141973495, + "learning_rate": 2.8587153686843626e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5470 + }, + { + "epoch": 1.5944137329066046, + "flow_matching_loss": 0.016144927684217693, + "grad_norm": 0.10499639809131622, + "learning_rate": 2.858013995286594e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5480 + }, + { + "epoch": 1.597323247017748, + "flow_matching_loss": 0.01809315672144294, + "grad_norm": 0.09541905671358109, + "learning_rate": 2.857310971773769e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5490 + }, + { + "epoch": 1.6002327611288916, + "flow_matching_loss": 0.01778232790529728, + "grad_norm": 0.10060036182403564, + "learning_rate": 2.8566062990001274e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5500 + }, + { + "epoch": 1.6031422752400348, + "flow_matching_loss": 0.017414864525198935, + "grad_norm": 0.09510762989521027, + "learning_rate": 2.8558999778219126e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5510 + }, + { + "epoch": 1.6060517893511783, + "flow_matching_loss": 0.01616199463605881, + "grad_norm": 0.08502375334501266, + "learning_rate": 2.8551920090973722e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5520 + }, + { + "epoch": 1.6089613034623218, + "flow_matching_loss": 0.017623922694474457, + "grad_norm": 0.07633312791585922, + "learning_rate": 2.854482393686754e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5530 + }, + { + "epoch": 1.6118708175734653, + "flow_matching_loss": 0.017661013826727866, + "grad_norm": 0.09117305278778076, + "learning_rate": 2.8537711324523086e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5540 + }, + { + "epoch": 1.6147803316846088, + "flow_matching_loss": 0.016406095679849388, + "grad_norm": 0.08659131079912186, + "learning_rate": 2.8530582262582845e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5550 + }, + { + "epoch": 1.617689845795752, + "flow_matching_loss": 0.017479505948722364, + "grad_norm": 0.09384981542825699, + "learning_rate": 2.8523436759709306e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5560 + }, + { + "epoch": 1.6205993599068955, + "flow_matching_loss": 0.018003450240939856, + "grad_norm": 0.0737551599740982, + "learning_rate": 2.8516274824584926e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5570 + }, + { + "epoch": 1.623508874018039, + "flow_matching_loss": 0.015221840143203736, + "grad_norm": 0.08111722022294998, + "learning_rate": 2.8509096465912136e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5580 + }, + { + "epoch": 1.6264183881291823, + "flow_matching_loss": 0.017960367631167172, + "grad_norm": 0.07046273350715637, + "learning_rate": 2.8501901692413315e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5590 + }, + { + "epoch": 1.629327902240326, + "flow_matching_loss": 0.01822669245302677, + "grad_norm": 0.07960425317287445, + "learning_rate": 2.8494690512830788e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5600 + }, + { + "epoch": 1.6322374163514692, + "flow_matching_loss": 0.016237550415098667, + "grad_norm": 0.06337831914424896, + "learning_rate": 2.8487462935926823e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5610 + }, + { + "epoch": 1.6351469304626127, + "flow_matching_loss": 0.018720805179327726, + "grad_norm": 0.09085789322853088, + "learning_rate": 2.8480218970483606e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5620 + }, + { + "epoch": 1.6380564445737562, + "flow_matching_loss": 0.01724295550957322, + "grad_norm": 0.08947379887104034, + "learning_rate": 2.847295862530324e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5630 + }, + { + "epoch": 1.6409659586848995, + "flow_matching_loss": 0.015557727962732314, + "grad_norm": 0.08142384141683578, + "learning_rate": 2.846568190920773e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5640 + }, + { + "epoch": 1.6438754727960432, + "flow_matching_loss": 0.014966677408665419, + "grad_norm": 0.08896787464618683, + "learning_rate": 2.8458388831038965e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5650 + }, + { + "epoch": 1.6467849869071864, + "flow_matching_loss": 0.016115553118288516, + "grad_norm": 0.09310554713010788, + "learning_rate": 2.8451079399658725e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5660 + }, + { + "epoch": 1.64969450101833, + "flow_matching_loss": 0.01682263296097517, + "grad_norm": 0.13455618917942047, + "learning_rate": 2.8443753623948665e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5670 + }, + { + "epoch": 1.6526040151294734, + "flow_matching_loss": 0.019589664414525032, + "grad_norm": 0.1290634125471115, + "learning_rate": 2.8436411512810288e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5680 + }, + { + "epoch": 1.6555135292406167, + "flow_matching_loss": 0.018188603036105633, + "grad_norm": 0.11339693516492844, + "learning_rate": 2.842905307516495e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5690 + }, + { + "epoch": 1.6584230433517604, + "flow_matching_loss": 0.018213918153196573, + "grad_norm": 0.0911894291639328, + "learning_rate": 2.8421678319953854e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5700 + }, + { + "epoch": 1.6613325574629036, + "flow_matching_loss": 0.016742206830531358, + "grad_norm": 0.08298409730195999, + "learning_rate": 2.841428725613801e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5710 + }, + { + "epoch": 1.6642420715740471, + "flow_matching_loss": 0.016323442477732895, + "grad_norm": 0.09323465079069138, + "learning_rate": 2.8406879892698272e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5720 + }, + { + "epoch": 1.6671515856851906, + "flow_matching_loss": 0.016671096812933682, + "grad_norm": 0.06632257252931595, + "learning_rate": 2.8399456238635274e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5730 + }, + { + "epoch": 1.6700610997963339, + "flow_matching_loss": 0.015876694489270447, + "grad_norm": 0.07370807230472565, + "learning_rate": 2.8392016302969457e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5740 + }, + { + "epoch": 1.6729706139074776, + "flow_matching_loss": 0.01548408530652523, + "grad_norm": 0.0818537175655365, + "learning_rate": 2.8384560094741055e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5750 + }, + { + "epoch": 1.6758801280186209, + "flow_matching_loss": 0.018204374704509974, + "grad_norm": 0.10679682344198227, + "learning_rate": 2.8377087623010054e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5760 + }, + { + "epoch": 1.6787896421297643, + "flow_matching_loss": 0.019166892766952513, + "grad_norm": 0.09296752512454987, + "learning_rate": 2.8369598896856213e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5770 + }, + { + "epoch": 1.6816991562409078, + "flow_matching_loss": 0.018760529905557634, + "grad_norm": 0.0971725657582283, + "learning_rate": 2.836209392537904e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5780 + }, + { + "epoch": 1.684608670352051, + "flow_matching_loss": 0.017140307836234568, + "grad_norm": 0.09415116161108017, + "learning_rate": 2.8354572717697787e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5790 + }, + { + "epoch": 1.6875181844631948, + "flow_matching_loss": 0.017298791743814945, + "grad_norm": 0.09866904467344284, + "learning_rate": 2.8347035282951428e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5800 + }, + { + "epoch": 1.690427698574338, + "flow_matching_loss": 0.020154422149062157, + "grad_norm": 0.10787111520767212, + "learning_rate": 2.8339481630298653e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5810 + }, + { + "epoch": 1.6933372126854815, + "flow_matching_loss": 0.01932883858680725, + "grad_norm": 0.09033793956041336, + "learning_rate": 2.833191176891787e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5820 + }, + { + "epoch": 1.696246726796625, + "flow_matching_loss": 0.017310749925673007, + "grad_norm": 0.09649154543876648, + "learning_rate": 2.832432570800717e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5830 + }, + { + "epoch": 1.6991562409077683, + "flow_matching_loss": 0.015698580630123617, + "grad_norm": 0.08289851248264313, + "learning_rate": 2.8316723456784332e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5840 + }, + { + "epoch": 1.702065755018912, + "flow_matching_loss": 0.01670493567362428, + "grad_norm": 0.10048871487379074, + "learning_rate": 2.8309105024486808e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5850 + }, + { + "epoch": 1.7049752691300553, + "flow_matching_loss": 0.017347517609596252, + "grad_norm": 0.0845789685845375, + "learning_rate": 2.830147042037172e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5860 + }, + { + "epoch": 1.7078847832411987, + "flow_matching_loss": 0.0181811167858541, + "grad_norm": 0.07265578955411911, + "learning_rate": 2.8293819653715816e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5870 + }, + { + "epoch": 1.7107942973523422, + "flow_matching_loss": 0.01749199368059635, + "grad_norm": 0.12034767866134644, + "learning_rate": 2.828615273381551e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5880 + }, + { + "epoch": 1.7137038114634855, + "flow_matching_loss": 0.01654358869418502, + "grad_norm": 0.10932344943284988, + "learning_rate": 2.827846966998683e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5890 + }, + { + "epoch": 1.7166133255746292, + "flow_matching_loss": 0.01763467360287905, + "grad_norm": 0.08873020857572556, + "learning_rate": 2.8270770471565422e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5900 + }, + { + "epoch": 1.7195228396857725, + "flow_matching_loss": 0.016993628069758415, + "grad_norm": 0.0922614261507988, + "learning_rate": 2.826305514790654e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5910 + }, + { + "epoch": 1.722432353796916, + "flow_matching_loss": 0.018817862402647732, + "grad_norm": 0.11580375581979752, + "learning_rate": 2.825532370838503e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5920 + }, + { + "epoch": 1.7253418679080594, + "flow_matching_loss": 0.017556344904005528, + "grad_norm": 0.0897599458694458, + "learning_rate": 2.824757616239532e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5930 + }, + { + "epoch": 1.7282513820192027, + "flow_matching_loss": 0.01715866168960929, + "grad_norm": 0.09729140996932983, + "learning_rate": 2.8239812519351404e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5940 + }, + { + "epoch": 1.7311608961303462, + "flow_matching_loss": 0.01777117932215333, + "grad_norm": 0.07686830312013626, + "learning_rate": 2.823203278868685e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5950 + }, + { + "epoch": 1.7340704102414897, + "flow_matching_loss": 0.017890595458447935, + "grad_norm": 0.08235519379377365, + "learning_rate": 2.8224236979854752e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5960 + }, + { + "epoch": 1.7369799243526332, + "flow_matching_loss": 0.01989372055977583, + "grad_norm": 0.0808558464050293, + "learning_rate": 2.821642510232776e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5970 + }, + { + "epoch": 1.7398894384637766, + "flow_matching_loss": 0.019226994179189206, + "grad_norm": 0.0824810117483139, + "learning_rate": 2.8208597165598046e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5980 + }, + { + "epoch": 1.74279895257492, + "flow_matching_loss": 0.018084902595728636, + "grad_norm": 0.11257638037204742, + "learning_rate": 2.8200753179177285e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 5990 + }, + { + "epoch": 1.7457084666860634, + "flow_matching_loss": 0.017362800147384404, + "grad_norm": 0.11582588404417038, + "learning_rate": 2.819289315259666e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6000 + }, + { + "epoch": 1.7486179807972069, + "flow_matching_loss": 0.01877316916361451, + "grad_norm": 0.10098263621330261, + "learning_rate": 2.8185017095406845e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6010 + }, + { + "epoch": 1.7515274949083504, + "flow_matching_loss": 0.016554318461567165, + "grad_norm": 0.07611538469791412, + "learning_rate": 2.8177125017177998e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6020 + }, + { + "epoch": 1.7544370090194938, + "flow_matching_loss": 0.01637625163421035, + "grad_norm": 0.08150531351566315, + "learning_rate": 2.8169216927499727e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6030 + }, + { + "epoch": 1.757346523130637, + "flow_matching_loss": 0.01620745100080967, + "grad_norm": 0.08863822370767593, + "learning_rate": 2.816129283598112e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6040 + }, + { + "epoch": 1.7602560372417806, + "flow_matching_loss": 0.01692084800451994, + "grad_norm": 0.07439881563186646, + "learning_rate": 2.815335275225068e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6050 + }, + { + "epoch": 1.763165551352924, + "flow_matching_loss": 0.016922676749527455, + "grad_norm": 0.0853864997625351, + "learning_rate": 2.8145396685956374e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6060 + }, + { + "epoch": 1.7660750654640673, + "flow_matching_loss": 0.017206930555403233, + "grad_norm": 0.08500128984451294, + "learning_rate": 2.813742464676556e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6070 + }, + { + "epoch": 1.768984579575211, + "flow_matching_loss": 0.01654683379456401, + "grad_norm": 0.1160147488117218, + "learning_rate": 2.8129436644365025e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6080 + }, + { + "epoch": 1.7718940936863543, + "flow_matching_loss": 0.016722906846553087, + "grad_norm": 0.06942442804574966, + "learning_rate": 2.8121432688460938e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6090 + }, + { + "epoch": 1.7748036077974978, + "flow_matching_loss": 0.017990529723465442, + "grad_norm": 0.09589885175228119, + "learning_rate": 2.8113412788778867e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6100 + }, + { + "epoch": 1.7777131219086413, + "flow_matching_loss": 0.016799407079815865, + "grad_norm": 0.0750080943107605, + "learning_rate": 2.8105376955063744e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6110 + }, + { + "epoch": 1.7806226360197845, + "flow_matching_loss": 0.017199462000280618, + "grad_norm": 0.07226303219795227, + "learning_rate": 2.8097325197079862e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6120 + }, + { + "epoch": 1.7835321501309283, + "flow_matching_loss": 0.016121744737029077, + "grad_norm": 0.10195241868495941, + "learning_rate": 2.808925752461087e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6130 + }, + { + "epoch": 1.7864416642420715, + "flow_matching_loss": 0.016745352186262608, + "grad_norm": 0.06571152061223984, + "learning_rate": 2.808117394745975e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6140 + }, + { + "epoch": 1.789351178353215, + "flow_matching_loss": 0.017415294609963893, + "grad_norm": 0.08149969577789307, + "learning_rate": 2.807307447544881e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6150 + }, + { + "epoch": 1.7922606924643585, + "flow_matching_loss": 0.017331584822386503, + "grad_norm": 0.09414371848106384, + "learning_rate": 2.806495911841967e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6160 + }, + { + "epoch": 1.7951702065755017, + "flow_matching_loss": 0.017356385476887226, + "grad_norm": 0.08183462172746658, + "learning_rate": 2.8056827886233255e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6170 + }, + { + "epoch": 1.7980797206866455, + "flow_matching_loss": 0.017430007550865413, + "grad_norm": 0.08893103152513504, + "learning_rate": 2.8048680788769783e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6180 + }, + { + "epoch": 1.8009892347977887, + "flow_matching_loss": 0.015677034854888916, + "grad_norm": 0.0707133412361145, + "learning_rate": 2.8040517835928742e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6190 + }, + { + "epoch": 1.8038987489089322, + "flow_matching_loss": 0.015951901953667403, + "grad_norm": 0.08220255374908447, + "learning_rate": 2.803233903762889e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6200 + }, + { + "epoch": 1.8068082630200757, + "flow_matching_loss": 0.014752757921814918, + "grad_norm": 0.07485468685626984, + "learning_rate": 2.8024144403808235e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6210 + }, + { + "epoch": 1.809717777131219, + "flow_matching_loss": 0.017506221681833266, + "grad_norm": 0.07357978075742722, + "learning_rate": 2.8015933944424037e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6220 + }, + { + "epoch": 1.8126272912423627, + "flow_matching_loss": 0.015915028750896454, + "grad_norm": 0.07209871709346771, + "learning_rate": 2.800770766945277e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6230 + }, + { + "epoch": 1.815536805353506, + "flow_matching_loss": 0.016900410316884516, + "grad_norm": 0.0774451270699501, + "learning_rate": 2.7999465588890135e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6240 + }, + { + "epoch": 1.8184463194646494, + "flow_matching_loss": 0.01702676070854068, + "grad_norm": 0.07061811536550522, + "learning_rate": 2.799120771275105e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6250 + }, + { + "epoch": 1.821355833575793, + "flow_matching_loss": 0.01653643324971199, + "grad_norm": 0.08194387704133987, + "learning_rate": 2.7982934051069597e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6260 + }, + { + "epoch": 1.8242653476869362, + "flow_matching_loss": 0.016485044546425343, + "grad_norm": 0.07407239079475403, + "learning_rate": 2.7974644613899063e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6270 + }, + { + "epoch": 1.8271748617980799, + "flow_matching_loss": 0.017232183087617158, + "grad_norm": 0.10221776366233826, + "learning_rate": 2.796633941131189e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6280 + }, + { + "epoch": 1.8300843759092231, + "flow_matching_loss": 0.017522669583559036, + "grad_norm": 0.0904102623462677, + "learning_rate": 2.795801845339969e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6290 + }, + { + "epoch": 1.8329938900203666, + "flow_matching_loss": 0.016194858588278294, + "grad_norm": 0.09380383044481277, + "learning_rate": 2.7949681750273208e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6300 + }, + { + "epoch": 1.83590340413151, + "flow_matching_loss": 0.015287991892546416, + "grad_norm": 0.07979407906532288, + "learning_rate": 2.7941329312062322e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6310 + }, + { + "epoch": 1.8388129182426534, + "flow_matching_loss": 0.017290185578167437, + "grad_norm": 0.09106510877609253, + "learning_rate": 2.7932961148916037e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6320 + }, + { + "epoch": 1.841722432353797, + "flow_matching_loss": 0.015430368855595589, + "grad_norm": 0.07037898153066635, + "learning_rate": 2.7924577271002456e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6330 + }, + { + "epoch": 1.8446319464649403, + "flow_matching_loss": 0.01609450001269579, + "grad_norm": 0.05681585147976875, + "learning_rate": 2.7916177688508782e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6340 + }, + { + "epoch": 1.8475414605760838, + "flow_matching_loss": 0.01707817576825619, + "grad_norm": 0.09189490973949432, + "learning_rate": 2.7907762411641302e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6350 + }, + { + "epoch": 1.8504509746872273, + "flow_matching_loss": 0.018344346061348916, + "grad_norm": 0.10156885534524918, + "learning_rate": 2.7899331450625367e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6360 + }, + { + "epoch": 1.8533604887983706, + "flow_matching_loss": 0.017775400914251803, + "grad_norm": 0.08959219604730606, + "learning_rate": 2.7890884815705397e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6370 + }, + { + "epoch": 1.8562700029095143, + "flow_matching_loss": 0.018578748032450677, + "grad_norm": 0.0971435010433197, + "learning_rate": 2.788242251714484e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6380 + }, + { + "epoch": 1.8591795170206575, + "flow_matching_loss": 0.017189202923327686, + "grad_norm": 0.11674308031797409, + "learning_rate": 2.787394456522619e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6390 + }, + { + "epoch": 1.862089031131801, + "flow_matching_loss": 0.018492572568356992, + "grad_norm": 0.0848507434129715, + "learning_rate": 2.7865450970250965e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6400 + }, + { + "epoch": 1.8649985452429445, + "flow_matching_loss": 0.01633382961153984, + "grad_norm": 0.09937480092048645, + "learning_rate": 2.7856941742539676e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6410 + }, + { + "epoch": 1.8679080593540878, + "flow_matching_loss": 0.016189302131533623, + "grad_norm": 0.08764338493347168, + "learning_rate": 2.784841689243184e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6420 + }, + { + "epoch": 1.8708175734652313, + "flow_matching_loss": 0.017714053113013505, + "grad_norm": 0.08089357614517212, + "learning_rate": 2.783987643028595e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6430 + }, + { + "epoch": 1.8737270875763747, + "flow_matching_loss": 0.016881211940199137, + "grad_norm": 0.10263040661811829, + "learning_rate": 2.783132036647948e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6440 + }, + { + "epoch": 1.8766366016875182, + "flow_matching_loss": 0.017497071344405413, + "grad_norm": 0.11991552263498306, + "learning_rate": 2.7822748711408852e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6450 + }, + { + "epoch": 1.8795461157986617, + "flow_matching_loss": 0.015395635645836592, + "grad_norm": 0.08834826201200485, + "learning_rate": 2.7814161475489428e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6460 + }, + { + "epoch": 1.882455629909805, + "flow_matching_loss": 0.017094421107321976, + "grad_norm": 0.07557949423789978, + "learning_rate": 2.780555866915552e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6470 + }, + { + "epoch": 1.8853651440209485, + "flow_matching_loss": 0.018055094685405493, + "grad_norm": 0.08473949879407883, + "learning_rate": 2.7796940302860338e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6480 + }, + { + "epoch": 1.888274658132092, + "flow_matching_loss": 0.01640403354540467, + "grad_norm": 0.07579516619443893, + "learning_rate": 2.7788306387076018e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6490 + }, + { + "epoch": 1.8911841722432354, + "flow_matching_loss": 0.017296364065259695, + "grad_norm": 0.09215551614761353, + "learning_rate": 2.7779656932293584e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6500 + }, + { + "epoch": 1.894093686354379, + "flow_matching_loss": 0.017104742862284184, + "grad_norm": 0.07609230279922485, + "learning_rate": 2.777099194902293e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6510 + }, + { + "epoch": 1.8970032004655222, + "flow_matching_loss": 0.01584559753537178, + "grad_norm": 0.09556776285171509, + "learning_rate": 2.776231144779283e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6520 + }, + { + "epoch": 1.8999127145766657, + "flow_matching_loss": 0.016443815547972918, + "grad_norm": 0.09666924923658371, + "learning_rate": 2.775361543915091e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6530 + }, + { + "epoch": 1.9028222286878091, + "flow_matching_loss": 0.01686391718685627, + "grad_norm": 0.09146234393119812, + "learning_rate": 2.774490393366365e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6540 + }, + { + "epoch": 1.9057317427989524, + "flow_matching_loss": 0.016857012733817102, + "grad_norm": 0.08843562006950378, + "learning_rate": 2.7736176941916346e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6550 + }, + { + "epoch": 1.9086412569100961, + "flow_matching_loss": 0.01760939313098788, + "grad_norm": 0.09796661883592606, + "learning_rate": 2.772743447451311e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6560 + }, + { + "epoch": 1.9115507710212394, + "flow_matching_loss": 0.015726991835981606, + "grad_norm": 0.07968240231275558, + "learning_rate": 2.7718676542076872e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6570 + }, + { + "epoch": 1.9144602851323829, + "flow_matching_loss": 0.015904546342790128, + "grad_norm": 0.06395082175731659, + "learning_rate": 2.770990315524934e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6580 + }, + { + "epoch": 1.9173697992435264, + "flow_matching_loss": 0.017076310887932777, + "grad_norm": 0.08119373023509979, + "learning_rate": 2.7701114324691015e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6590 + }, + { + "epoch": 1.9202793133546696, + "flow_matching_loss": 0.016194525733590127, + "grad_norm": 0.0991370677947998, + "learning_rate": 2.7692310061081145e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6600 + }, + { + "epoch": 1.9231888274658133, + "flow_matching_loss": 0.016300099436193705, + "grad_norm": 0.10749704390764236, + "learning_rate": 2.7683490375117742e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6610 + }, + { + "epoch": 1.9260983415769566, + "flow_matching_loss": 0.016560526192188264, + "grad_norm": 0.06251049041748047, + "learning_rate": 2.7674655277517556e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6620 + }, + { + "epoch": 1.9290078556881, + "flow_matching_loss": 0.01695288047194481, + "grad_norm": 0.12784874439239502, + "learning_rate": 2.7665804779016072e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6630 + }, + { + "epoch": 1.9319173697992436, + "flow_matching_loss": 0.015560354571789503, + "grad_norm": 0.09243197739124298, + "learning_rate": 2.765693889036747e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6640 + }, + { + "epoch": 1.9348268839103868, + "flow_matching_loss": 0.01643256563693285, + "grad_norm": 0.08910775184631348, + "learning_rate": 2.764805762234465e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6650 + }, + { + "epoch": 1.9377363980215305, + "flow_matching_loss": 0.018131051398813724, + "grad_norm": 0.0979708731174469, + "learning_rate": 2.7639160985739182e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6660 + }, + { + "epoch": 1.9406459121326738, + "flow_matching_loss": 0.017591935582458973, + "grad_norm": 0.07146275788545609, + "learning_rate": 2.7630248991361317e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6670 + }, + { + "epoch": 1.9435554262438173, + "flow_matching_loss": 0.018344883993268012, + "grad_norm": 0.09048923850059509, + "learning_rate": 2.762132165003998e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6680 + }, + { + "epoch": 1.9464649403549608, + "flow_matching_loss": 0.015901629626750947, + "grad_norm": 0.09246297180652618, + "learning_rate": 2.7612378972622722e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6690 + }, + { + "epoch": 1.949374454466104, + "flow_matching_loss": 0.016860281769186258, + "grad_norm": 0.10724200308322906, + "learning_rate": 2.7603420969975743e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6700 + }, + { + "epoch": 1.9522839685772477, + "flow_matching_loss": 0.018016140814870595, + "grad_norm": 0.09994521737098694, + "learning_rate": 2.7594447652983862e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6710 + }, + { + "epoch": 1.955193482688391, + "flow_matching_loss": 0.01829990539699793, + "grad_norm": 0.10873758792877197, + "learning_rate": 2.7585459032550497e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6720 + }, + { + "epoch": 1.9581029967995345, + "flow_matching_loss": 0.015357098821550608, + "grad_norm": 0.10149592161178589, + "learning_rate": 2.7576455119597684e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6730 + }, + { + "epoch": 1.961012510910678, + "flow_matching_loss": 0.016721124574542046, + "grad_norm": 0.0814293622970581, + "learning_rate": 2.7567435925066012e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6740 + }, + { + "epoch": 1.9639220250218212, + "flow_matching_loss": 0.01712523363530636, + "grad_norm": 0.06600968539714813, + "learning_rate": 2.7558401459914665e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6750 + }, + { + "epoch": 1.966831539132965, + "flow_matching_loss": 0.01638155207037926, + "grad_norm": 0.06897532939910889, + "learning_rate": 2.7549351735121362e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6760 + }, + { + "epoch": 1.9697410532441082, + "flow_matching_loss": 0.017254160810261966, + "grad_norm": 0.11103256046772003, + "learning_rate": 2.7540286761682377e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6770 + }, + { + "epoch": 1.9726505673552517, + "flow_matching_loss": 0.017970383726060392, + "grad_norm": 0.09817861020565033, + "learning_rate": 2.75312065506125e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6780 + }, + { + "epoch": 1.9755600814663952, + "flow_matching_loss": 0.018382926005870104, + "grad_norm": 0.09734764695167542, + "learning_rate": 2.7522111112945052e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6790 + }, + { + "epoch": 1.9784695955775384, + "flow_matching_loss": 0.01765251439064741, + "grad_norm": 0.10186561942100525, + "learning_rate": 2.751300045973184e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6800 + }, + { + "epoch": 1.9813791096886821, + "flow_matching_loss": 0.015632458962500095, + "grad_norm": 0.0832919180393219, + "learning_rate": 2.750387460204317e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6810 + }, + { + "epoch": 1.9842886237998254, + "flow_matching_loss": 0.015918474271893503, + "grad_norm": 0.08892107754945755, + "learning_rate": 2.7494733550967827e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6820 + }, + { + "epoch": 1.9871981379109689, + "flow_matching_loss": 0.01633006604388356, + "grad_norm": 0.08667997270822525, + "learning_rate": 2.748557731761304e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6830 + }, + { + "epoch": 1.9901076520221124, + "flow_matching_loss": 0.018007144890725614, + "grad_norm": 0.07802540808916092, + "learning_rate": 2.74764059131045e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6840 + }, + { + "epoch": 1.9930171661332556, + "flow_matching_loss": 0.017149584274739026, + "grad_norm": 0.08695584535598755, + "learning_rate": 2.7467219348586322e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6850 + }, + { + "epoch": 1.9959266802443993, + "flow_matching_loss": 0.016799836698919535, + "grad_norm": 0.07436618953943253, + "learning_rate": 2.7458017635221056e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6860 + }, + { + "epoch": 1.9988361943555426, + "flow_matching_loss": 0.017902564816176893, + "grad_norm": 0.07455611228942871, + "learning_rate": 2.7448800784189652e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6870 + }, + { + "epoch": 2.001745708466686, + "flow_matching_loss": 0.017821810860186815, + "grad_norm": 0.0956747904419899, + "learning_rate": 2.7439568806691448e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6880 + }, + { + "epoch": 2.0046552225778296, + "flow_matching_loss": 0.015173871815204621, + "grad_norm": 0.09084337949752808, + "learning_rate": 2.7430321713944167e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6890 + }, + { + "epoch": 2.007564736688973, + "flow_matching_loss": 0.015359226893633605, + "grad_norm": 0.08174826949834824, + "learning_rate": 2.7421059517183906e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6900 + }, + { + "epoch": 2.0104742508001165, + "flow_matching_loss": 0.016740091610699892, + "grad_norm": 0.08032640069723129, + "learning_rate": 2.7411782227665097e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6910 + }, + { + "epoch": 2.01338376491126, + "flow_matching_loss": 0.018670203164219856, + "grad_norm": 0.08388732373714447, + "learning_rate": 2.7402489856660528e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6920 + }, + { + "epoch": 2.016293279022403, + "flow_matching_loss": 0.015359473507851362, + "grad_norm": 0.08058849722146988, + "learning_rate": 2.7393182415461307e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6930 + }, + { + "epoch": 2.019202793133547, + "flow_matching_loss": 0.01730056619271636, + "grad_norm": 0.10241051018238068, + "learning_rate": 2.7383859915376853e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6940 + }, + { + "epoch": 2.02211230724469, + "flow_matching_loss": 0.016866291780024766, + "grad_norm": 0.07475873827934265, + "learning_rate": 2.7374522367734875e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6950 + }, + { + "epoch": 2.0250218213558338, + "flow_matching_loss": 0.01663920246064663, + "grad_norm": 0.08385030180215836, + "learning_rate": 2.7365169783881378e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6960 + }, + { + "epoch": 2.027931335466977, + "flow_matching_loss": 0.014750133268535137, + "grad_norm": 0.08441538363695145, + "learning_rate": 2.735580217518064e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6970 + }, + { + "epoch": 2.0308408495781203, + "flow_matching_loss": 0.01725252689793706, + "grad_norm": 0.08109751343727112, + "learning_rate": 2.7346419553015175e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6980 + }, + { + "epoch": 2.033750363689264, + "flow_matching_loss": 0.017660737317055464, + "grad_norm": 0.10495557636022568, + "learning_rate": 2.733702192878577e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 6990 + }, + { + "epoch": 2.0366598778004072, + "flow_matching_loss": 0.01593604953959584, + "grad_norm": 0.09454198181629181, + "learning_rate": 2.732760931391141e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7000 + }, + { + "epoch": 2.039569391911551, + "flow_matching_loss": 0.01654547341167927, + "grad_norm": 0.07170224189758301, + "learning_rate": 2.731818171982932e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7010 + }, + { + "epoch": 2.042478906022694, + "flow_matching_loss": 0.015454737935215235, + "grad_norm": 0.10838077962398529, + "learning_rate": 2.730873915799491e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7020 + }, + { + "epoch": 2.0453884201338375, + "flow_matching_loss": 0.016959916707128285, + "grad_norm": 0.07772323489189148, + "learning_rate": 2.7299281639881783e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7030 + }, + { + "epoch": 2.048297934244981, + "flow_matching_loss": 0.014778068009763955, + "grad_norm": 0.06683853268623352, + "learning_rate": 2.728980917698172e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7040 + }, + { + "epoch": 2.0512074483561245, + "flow_matching_loss": 0.01618355754762888, + "grad_norm": 0.08981888741254807, + "learning_rate": 2.7280321780804655e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7050 + }, + { + "epoch": 2.054116962467268, + "flow_matching_loss": 0.017350093089044095, + "grad_norm": 0.08273127675056458, + "learning_rate": 2.7270819462878666e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7060 + }, + { + "epoch": 2.0570264765784114, + "flow_matching_loss": 0.01599160796031356, + "grad_norm": 0.07325915992259979, + "learning_rate": 2.7261302234749965e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7070 + }, + { + "epoch": 2.0599359906895547, + "flow_matching_loss": 0.015203469712287188, + "grad_norm": 0.10310603678226471, + "learning_rate": 2.7251770107982888e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7080 + }, + { + "epoch": 2.0628455048006984, + "flow_matching_loss": 0.016120374388992786, + "grad_norm": 0.07476403564214706, + "learning_rate": 2.7242223094159856e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7090 + }, + { + "epoch": 2.0657550189118417, + "flow_matching_loss": 0.016491062846034764, + "grad_norm": 0.07918092608451843, + "learning_rate": 2.7232661204881408e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7100 + }, + { + "epoch": 2.0686645330229854, + "flow_matching_loss": 0.015142364241182805, + "grad_norm": 0.09591260552406311, + "learning_rate": 2.7223084451766123e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7110 + }, + { + "epoch": 2.0715740471341286, + "flow_matching_loss": 0.016970221512019636, + "grad_norm": 0.10381370037794113, + "learning_rate": 2.721349284645067e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7120 + }, + { + "epoch": 2.074483561245272, + "flow_matching_loss": 0.017051411606371403, + "grad_norm": 0.09185776859521866, + "learning_rate": 2.7203886400589752e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7130 + }, + { + "epoch": 2.0773930753564156, + "flow_matching_loss": 0.017098691128194333, + "grad_norm": 0.10021667927503586, + "learning_rate": 2.7194265125856105e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7140 + }, + { + "epoch": 2.080302589467559, + "flow_matching_loss": 0.016829704586416482, + "grad_norm": 0.11007888615131378, + "learning_rate": 2.718462903394049e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7150 + }, + { + "epoch": 2.0832121035787026, + "flow_matching_loss": 0.017500566504895686, + "grad_norm": 0.1074223592877388, + "learning_rate": 2.7174978136551662e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7160 + }, + { + "epoch": 2.086121617689846, + "flow_matching_loss": 0.018054713401943444, + "grad_norm": 0.1136183813214302, + "learning_rate": 2.716531244541637e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7170 + }, + { + "epoch": 2.089031131800989, + "flow_matching_loss": 0.01648937426507473, + "grad_norm": 0.08568757772445679, + "learning_rate": 2.715563197227934e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7180 + }, + { + "epoch": 2.091940645912133, + "flow_matching_loss": 0.015366742014884948, + "grad_norm": 0.07440062612295151, + "learning_rate": 2.714593672890327e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7190 + }, + { + "epoch": 2.094850160023276, + "flow_matching_loss": 0.01750759230926633, + "grad_norm": 0.10559970140457153, + "learning_rate": 2.7136226727068783e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7200 + }, + { + "epoch": 2.0977596741344193, + "flow_matching_loss": 0.015514460671693087, + "grad_norm": 0.07742292433977127, + "learning_rate": 2.712650197857445e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7210 + }, + { + "epoch": 2.100669188245563, + "flow_matching_loss": 0.017607845552265643, + "grad_norm": 0.07515960931777954, + "learning_rate": 2.7116762495236763e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7220 + }, + { + "epoch": 2.1035787023567063, + "flow_matching_loss": 0.015851312689483164, + "grad_norm": 0.09743686765432358, + "learning_rate": 2.7107008288890106e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7230 + }, + { + "epoch": 2.10648821646785, + "flow_matching_loss": 0.015351446624845267, + "grad_norm": 0.06548064947128296, + "learning_rate": 2.709723937138676e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7240 + }, + { + "epoch": 2.1093977305789933, + "flow_matching_loss": 0.016595125664025545, + "grad_norm": 0.07743366807699203, + "learning_rate": 2.7087455754596883e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7250 + }, + { + "epoch": 2.112307244690137, + "flow_matching_loss": 0.016051541734486818, + "grad_norm": 0.09489785134792328, + "learning_rate": 2.7077657450408498e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7260 + }, + { + "epoch": 2.1152167588012802, + "flow_matching_loss": 0.01579417735338211, + "grad_norm": 0.08436472713947296, + "learning_rate": 2.7067844470727456e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7270 + }, + { + "epoch": 2.1181262729124235, + "flow_matching_loss": 0.0188221450895071, + "grad_norm": 0.09410113841295242, + "learning_rate": 2.7058016827477455e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7280 + }, + { + "epoch": 2.121035787023567, + "flow_matching_loss": 0.01571461632847786, + "grad_norm": 0.09739621728658676, + "learning_rate": 2.7048174532600018e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7290 + }, + { + "epoch": 2.1239453011347105, + "flow_matching_loss": 0.016110573802143334, + "grad_norm": 0.09729532152414322, + "learning_rate": 2.7038317598054455e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7300 + }, + { + "epoch": 2.1268548152458537, + "flow_matching_loss": 0.016490278113633395, + "grad_norm": 0.06990286707878113, + "learning_rate": 2.702844603581787e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7310 + }, + { + "epoch": 2.1297643293569974, + "flow_matching_loss": 0.016419134754687547, + "grad_norm": 0.10305049270391464, + "learning_rate": 2.701855985788514e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7320 + }, + { + "epoch": 2.1326738434681407, + "flow_matching_loss": 0.01617075828835368, + "grad_norm": 0.10823473334312439, + "learning_rate": 2.7008659076268912e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7330 + }, + { + "epoch": 2.1355833575792844, + "flow_matching_loss": 0.017440353985875845, + "grad_norm": 0.09205766767263412, + "learning_rate": 2.6998743702999562e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7340 + }, + { + "epoch": 2.1384928716904277, + "flow_matching_loss": 0.0160367825999856, + "grad_norm": 0.1336563527584076, + "learning_rate": 2.6988813750125202e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7350 + }, + { + "epoch": 2.141402385801571, + "flow_matching_loss": 0.016865062713623046, + "grad_norm": 0.09867194294929504, + "learning_rate": 2.6978869229711668e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7360 + }, + { + "epoch": 2.1443118999127146, + "flow_matching_loss": 0.01690604668110609, + "grad_norm": 0.08355677127838135, + "learning_rate": 2.6968910153842484e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7370 + }, + { + "epoch": 2.147221414023858, + "flow_matching_loss": 0.015676092728972434, + "grad_norm": 0.08776777982711792, + "learning_rate": 2.6958936534618874e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7380 + }, + { + "epoch": 2.1501309281350016, + "flow_matching_loss": 0.01612120782956481, + "grad_norm": 0.10372947156429291, + "learning_rate": 2.694894838415971e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7390 + }, + { + "epoch": 2.153040442246145, + "flow_matching_loss": 0.01571722310036421, + "grad_norm": 0.09617681056261063, + "learning_rate": 2.6938945714601555e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7400 + }, + { + "epoch": 2.155949956357288, + "flow_matching_loss": 0.015271288715302945, + "grad_norm": 0.09277535229921341, + "learning_rate": 2.6928928538098587e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7410 + }, + { + "epoch": 2.158859470468432, + "flow_matching_loss": 0.017560992669314147, + "grad_norm": 0.08877278119325638, + "learning_rate": 2.6918896866822615e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7420 + }, + { + "epoch": 2.161768984579575, + "flow_matching_loss": 0.01533149303868413, + "grad_norm": 0.09108344465494156, + "learning_rate": 2.690885071296308e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7430 + }, + { + "epoch": 2.164678498690719, + "flow_matching_loss": 0.017757116910070182, + "grad_norm": 0.10470380634069443, + "learning_rate": 2.6898790088726993e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7440 + }, + { + "epoch": 2.167588012801862, + "flow_matching_loss": 0.017198972310870884, + "grad_norm": 0.10377341508865356, + "learning_rate": 2.6888715006338968e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7450 + }, + { + "epoch": 2.1704975269130053, + "flow_matching_loss": 0.01663223747164011, + "grad_norm": 0.09103534370660782, + "learning_rate": 2.6878625478041172e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7460 + }, + { + "epoch": 2.173407041024149, + "flow_matching_loss": 0.019528802204877137, + "grad_norm": 0.114688441157341, + "learning_rate": 2.6868521516093342e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7470 + }, + { + "epoch": 2.1763165551352923, + "flow_matching_loss": 0.014382893871515989, + "grad_norm": 0.06969216465950012, + "learning_rate": 2.6858403132772745e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7480 + }, + { + "epoch": 2.179226069246436, + "flow_matching_loss": 0.016963027883321047, + "grad_norm": 0.10866544395685196, + "learning_rate": 2.6848270340374162e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7490 + }, + { + "epoch": 2.1821355833575793, + "flow_matching_loss": 0.014895957242697478, + "grad_norm": 0.0828837901353836, + "learning_rate": 2.68381231512099e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7500 + }, + { + "epoch": 2.1850450974687226, + "flow_matching_loss": 0.017039693985134365, + "grad_norm": 0.06893998384475708, + "learning_rate": 2.6827961577609745e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7510 + }, + { + "epoch": 2.1879546115798663, + "flow_matching_loss": 0.01618116656318307, + "grad_norm": 0.07202392816543579, + "learning_rate": 2.681778563192097e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7520 + }, + { + "epoch": 2.1908641256910095, + "flow_matching_loss": 0.016526246815919875, + "grad_norm": 0.09681548178195953, + "learning_rate": 2.6807595326508308e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7530 + }, + { + "epoch": 2.1937736398021532, + "flow_matching_loss": 0.0179961153306067, + "grad_norm": 0.10063367336988449, + "learning_rate": 2.679739067375394e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7540 + }, + { + "epoch": 2.1966831539132965, + "flow_matching_loss": 0.01832523727789521, + "grad_norm": 0.08261393010616302, + "learning_rate": 2.6787171686057487e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7550 + }, + { + "epoch": 2.1995926680244398, + "flow_matching_loss": 0.01695553045719862, + "grad_norm": 0.10449963063001633, + "learning_rate": 2.677693837583597e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7560 + }, + { + "epoch": 2.2025021821355835, + "flow_matching_loss": 0.015904280357062817, + "grad_norm": 0.08271070569753647, + "learning_rate": 2.6766690755523836e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7570 + }, + { + "epoch": 2.2054116962467267, + "flow_matching_loss": 0.01575610777363181, + "grad_norm": 0.10787206143140793, + "learning_rate": 2.675642883757291e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7580 + }, + { + "epoch": 2.2083212103578704, + "flow_matching_loss": 0.016596919670701028, + "grad_norm": 0.09200070798397064, + "learning_rate": 2.674615263445239e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7590 + }, + { + "epoch": 2.2112307244690137, + "flow_matching_loss": 0.01805265499278903, + "grad_norm": 0.0885273888707161, + "learning_rate": 2.673586215864882e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7600 + }, + { + "epoch": 2.214140238580157, + "flow_matching_loss": 0.01647864207625389, + "grad_norm": 0.09317810833454132, + "learning_rate": 2.6725557422666113e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7610 + }, + { + "epoch": 2.2170497526913007, + "flow_matching_loss": 0.015808716975152494, + "grad_norm": 0.1048714816570282, + "learning_rate": 2.6715238439025493e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7620 + }, + { + "epoch": 2.219959266802444, + "flow_matching_loss": 0.016592751350253822, + "grad_norm": 0.09508407860994339, + "learning_rate": 2.6704905220265493e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7630 + }, + { + "epoch": 2.2228687809135876, + "flow_matching_loss": 0.01720146369189024, + "grad_norm": 0.08996113389730453, + "learning_rate": 2.6694557778941947e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7640 + }, + { + "epoch": 2.225778295024731, + "flow_matching_loss": 0.01671403869986534, + "grad_norm": 0.10598383098840714, + "learning_rate": 2.6684196127627977e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7650 + }, + { + "epoch": 2.228687809135874, + "flow_matching_loss": 0.015097912773489952, + "grad_norm": 0.0661490336060524, + "learning_rate": 2.6673820278913963e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7660 + }, + { + "epoch": 2.231597323247018, + "flow_matching_loss": 0.01522701745852828, + "grad_norm": 0.07215132564306259, + "learning_rate": 2.6663430245407544e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7670 + }, + { + "epoch": 2.234506837358161, + "flow_matching_loss": 0.017201013490557672, + "grad_norm": 0.08673103898763657, + "learning_rate": 2.6653026039733585e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7680 + }, + { + "epoch": 2.237416351469305, + "flow_matching_loss": 0.01728763971477747, + "grad_norm": 0.0706494078040123, + "learning_rate": 2.6642607674534176e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7690 + }, + { + "epoch": 2.240325865580448, + "flow_matching_loss": 0.017370521556586027, + "grad_norm": 0.07108838111162186, + "learning_rate": 2.663217516246862e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7700 + }, + { + "epoch": 2.2432353796915914, + "flow_matching_loss": 0.015931936725974084, + "grad_norm": 0.10340288281440735, + "learning_rate": 2.66217285162134e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7710 + }, + { + "epoch": 2.246144893802735, + "flow_matching_loss": 0.01726842364296317, + "grad_norm": 0.09005376696586609, + "learning_rate": 2.6611267748462175e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7720 + }, + { + "epoch": 2.2490544079138783, + "flow_matching_loss": 0.01579492585733533, + "grad_norm": 0.0758841335773468, + "learning_rate": 2.6600792871925764e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7730 + }, + { + "epoch": 2.2519639220250216, + "flow_matching_loss": 0.015560926869511604, + "grad_norm": 0.08643247187137604, + "learning_rate": 2.659030389933213e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7740 + }, + { + "epoch": 2.2548734361361653, + "flow_matching_loss": 0.017243430763483048, + "grad_norm": 0.09427540004253387, + "learning_rate": 2.6579800843426358e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7750 + }, + { + "epoch": 2.2577829502473086, + "flow_matching_loss": 0.01680444134399295, + "grad_norm": 0.07737462222576141, + "learning_rate": 2.656928371697067e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7760 + }, + { + "epoch": 2.2606924643584523, + "flow_matching_loss": 0.016614085622131826, + "grad_norm": 0.07463520020246506, + "learning_rate": 2.6558752532744346e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7770 + }, + { + "epoch": 2.2636019784695955, + "flow_matching_loss": 0.017885309271514417, + "grad_norm": 0.09467581659555435, + "learning_rate": 2.654820730354377e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7780 + }, + { + "epoch": 2.2665114925807393, + "flow_matching_loss": 0.017834890447556973, + "grad_norm": 0.12022972851991653, + "learning_rate": 2.6537648042182406e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7790 + }, + { + "epoch": 2.2694210066918825, + "flow_matching_loss": 0.015965427830815317, + "grad_norm": 0.09629964083433151, + "learning_rate": 2.6527074761490735e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7800 + }, + { + "epoch": 2.2723305208030258, + "flow_matching_loss": 0.01523082023486495, + "grad_norm": 0.08877071738243103, + "learning_rate": 2.6516487474316303e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7810 + }, + { + "epoch": 2.2752400349141695, + "flow_matching_loss": 0.015103575401008129, + "grad_norm": 0.06224657595157623, + "learning_rate": 2.650588619352365e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7820 + }, + { + "epoch": 2.2781495490253127, + "flow_matching_loss": 0.015976173616945744, + "grad_norm": 0.08615821599960327, + "learning_rate": 2.649527093199434e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7830 + }, + { + "epoch": 2.281059063136456, + "flow_matching_loss": 0.014502124860882759, + "grad_norm": 0.06663165241479874, + "learning_rate": 2.6484641702626914e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7840 + }, + { + "epoch": 2.2839685772475997, + "flow_matching_loss": 0.016006383392959834, + "grad_norm": 0.10475897789001465, + "learning_rate": 2.647399851833689e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7850 + }, + { + "epoch": 2.286878091358743, + "flow_matching_loss": 0.016521460376679897, + "grad_norm": 0.08100198209285736, + "learning_rate": 2.6463341392056735e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7860 + }, + { + "epoch": 2.2897876054698867, + "flow_matching_loss": 0.015193694457411765, + "grad_norm": 0.0940919816493988, + "learning_rate": 2.6452670336735867e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7870 + }, + { + "epoch": 2.29269711958103, + "flow_matching_loss": 0.017112257797271012, + "grad_norm": 0.08626744896173477, + "learning_rate": 2.644198536534063e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7880 + }, + { + "epoch": 2.295606633692173, + "flow_matching_loss": 0.01693006381392479, + "grad_norm": 0.11797182261943817, + "learning_rate": 2.6431286490854255e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7890 + }, + { + "epoch": 2.298516147803317, + "flow_matching_loss": 0.01807980826124549, + "grad_norm": 0.09922970831394196, + "learning_rate": 2.6420573726276903e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7900 + }, + { + "epoch": 2.30142566191446, + "flow_matching_loss": 0.016918319277465343, + "grad_norm": 0.10104074329137802, + "learning_rate": 2.6409847084625575e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7910 + }, + { + "epoch": 2.304335176025604, + "flow_matching_loss": 0.0159209243953228, + "grad_norm": 0.08861827105283737, + "learning_rate": 2.6399106578934167e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7920 + }, + { + "epoch": 2.307244690136747, + "flow_matching_loss": 0.018168943468481302, + "grad_norm": 0.12202727049589157, + "learning_rate": 2.63883522222534e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7930 + }, + { + "epoch": 2.3101542042478904, + "flow_matching_loss": 0.015838419552892447, + "grad_norm": 0.08724018931388855, + "learning_rate": 2.6377584027650827e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7940 + }, + { + "epoch": 2.313063718359034, + "flow_matching_loss": 0.017250462993979455, + "grad_norm": 0.11001388728618622, + "learning_rate": 2.636680200821083e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7950 + }, + { + "epoch": 2.3159732324701774, + "flow_matching_loss": 0.016607213579118252, + "grad_norm": 0.10062267631292343, + "learning_rate": 2.635600617703457e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7960 + }, + { + "epoch": 2.318882746581321, + "flow_matching_loss": 0.016539415810257196, + "grad_norm": 0.1028779000043869, + "learning_rate": 2.6345196547240007e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7970 + }, + { + "epoch": 2.3217922606924644, + "flow_matching_loss": 0.016135498136281966, + "grad_norm": 0.09629010409116745, + "learning_rate": 2.6334373131961854e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7980 + }, + { + "epoch": 2.3247017748036076, + "flow_matching_loss": 0.015930989291518927, + "grad_norm": 0.08179789781570435, + "learning_rate": 2.6323535944351583e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 7990 + }, + { + "epoch": 2.3276112889147513, + "flow_matching_loss": 0.01520235175266862, + "grad_norm": 0.06850016117095947, + "learning_rate": 2.63126849975774e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8000 + }, + { + "epoch": 2.3305208030258946, + "flow_matching_loss": 0.01624268973246217, + "grad_norm": 0.08197231590747833, + "learning_rate": 2.6301820304824234e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8010 + }, + { + "epoch": 2.3334303171370383, + "flow_matching_loss": 0.015210281778126954, + "grad_norm": 0.10971414297819138, + "learning_rate": 2.6290941879293703e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8020 + }, + { + "epoch": 2.3363398312481816, + "flow_matching_loss": 0.016861541103571652, + "grad_norm": 0.09198158979415894, + "learning_rate": 2.628004973420412e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8030 + }, + { + "epoch": 2.339249345359325, + "flow_matching_loss": 0.017163799982517956, + "grad_norm": 0.10308793932199478, + "learning_rate": 2.626914388279047e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8040 + }, + { + "epoch": 2.3421588594704685, + "flow_matching_loss": 0.015297060925513506, + "grad_norm": 0.07291965186595917, + "learning_rate": 2.6258224338304394e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8050 + }, + { + "epoch": 2.345068373581612, + "flow_matching_loss": 0.016296430490911007, + "grad_norm": 0.08919917047023773, + "learning_rate": 2.6247291114014166e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8060 + }, + { + "epoch": 2.347977887692755, + "flow_matching_loss": 0.015698900818824767, + "grad_norm": 0.09717602282762527, + "learning_rate": 2.6236344223204683e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8070 + }, + { + "epoch": 2.3508874018038988, + "flow_matching_loss": 0.016794399172067643, + "grad_norm": 0.08262206614017487, + "learning_rate": 2.6225383679177453e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8080 + }, + { + "epoch": 2.353796915915042, + "flow_matching_loss": 0.015983247291296722, + "grad_norm": 0.07528749853372574, + "learning_rate": 2.6214409495250568e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8090 + }, + { + "epoch": 2.3567064300261857, + "flow_matching_loss": 0.01953901518136263, + "grad_norm": 0.08688241988420486, + "learning_rate": 2.620342168475869e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8100 + }, + { + "epoch": 2.359615944137329, + "flow_matching_loss": 0.017705923598259687, + "grad_norm": 0.10984530299901962, + "learning_rate": 2.619242026105306e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8110 + }, + { + "epoch": 2.3625254582484727, + "flow_matching_loss": 0.015780732687562703, + "grad_norm": 0.0604446604847908, + "learning_rate": 2.618140523750143e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8120 + }, + { + "epoch": 2.365434972359616, + "flow_matching_loss": 0.0166504867374897, + "grad_norm": 0.07224846631288528, + "learning_rate": 2.6170376627488103e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8130 + }, + { + "epoch": 2.3683444864707592, + "flow_matching_loss": 0.0170687647536397, + "grad_norm": 0.09462336450815201, + "learning_rate": 2.6159334444413873e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8140 + }, + { + "epoch": 2.371254000581903, + "flow_matching_loss": 0.01878861226141453, + "grad_norm": 0.10542982071638107, + "learning_rate": 2.614827870169604e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8150 + }, + { + "epoch": 2.374163514693046, + "flow_matching_loss": 0.014758850354701281, + "grad_norm": 0.09903089702129364, + "learning_rate": 2.6137209412768367e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8160 + }, + { + "epoch": 2.3770730288041895, + "flow_matching_loss": 0.017610883247107266, + "grad_norm": 0.10778328776359558, + "learning_rate": 2.612612659108109e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8170 + }, + { + "epoch": 2.379982542915333, + "flow_matching_loss": 0.01624194998294115, + "grad_norm": 0.10709578543901443, + "learning_rate": 2.611503025010088e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8180 + }, + { + "epoch": 2.3828920570264764, + "flow_matching_loss": 0.01568342950195074, + "grad_norm": 0.07311269640922546, + "learning_rate": 2.6103920403310832e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8190 + }, + { + "epoch": 2.38580157113762, + "flow_matching_loss": 0.01595008047297597, + "grad_norm": 0.07027493417263031, + "learning_rate": 2.6092797064210465e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8200 + }, + { + "epoch": 2.3887110852487634, + "flow_matching_loss": 0.017098479252308608, + "grad_norm": 0.08756469935178757, + "learning_rate": 2.6081660246315685e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8210 + }, + { + "epoch": 2.391620599359907, + "flow_matching_loss": 0.015159698110073805, + "grad_norm": 0.07350801676511765, + "learning_rate": 2.6070509963158774e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8220 + }, + { + "epoch": 2.3945301134710504, + "flow_matching_loss": 0.015296839829534292, + "grad_norm": 0.0750538781285286, + "learning_rate": 2.6059346228288375e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8230 + }, + { + "epoch": 2.3974396275821936, + "flow_matching_loss": 0.016342896874994038, + "grad_norm": 0.07158420979976654, + "learning_rate": 2.604816905526948e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8240 + }, + { + "epoch": 2.4003491416933374, + "flow_matching_loss": 0.015469567757099867, + "grad_norm": 0.06751872599124908, + "learning_rate": 2.603697845768341e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8250 + }, + { + "epoch": 2.4032586558044806, + "flow_matching_loss": 0.01574276462197304, + "grad_norm": 0.08411609381437302, + "learning_rate": 2.6025774449127792e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8260 + }, + { + "epoch": 2.406168169915624, + "flow_matching_loss": 0.015600107423961163, + "grad_norm": 0.088570237159729, + "learning_rate": 2.6014557043216562e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8270 + }, + { + "epoch": 2.4090776840267676, + "flow_matching_loss": 0.015241039637476205, + "grad_norm": 0.09559714794158936, + "learning_rate": 2.6003326253579913e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8280 + }, + { + "epoch": 2.411987198137911, + "flow_matching_loss": 0.016027992777526378, + "grad_norm": 0.09286770224571228, + "learning_rate": 2.599208209386432e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8290 + }, + { + "epoch": 2.4148967122490546, + "flow_matching_loss": 0.01734726456925273, + "grad_norm": 0.12336403876543045, + "learning_rate": 2.5980824577732502e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8300 + }, + { + "epoch": 2.417806226360198, + "flow_matching_loss": 0.015072564221918583, + "grad_norm": 0.11323237419128418, + "learning_rate": 2.5969553718863395e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8310 + }, + { + "epoch": 2.4207157404713415, + "flow_matching_loss": 0.0173687182366848, + "grad_norm": 0.0942811444401741, + "learning_rate": 2.595826953095216e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8320 + }, + { + "epoch": 2.423625254582485, + "flow_matching_loss": 0.015977287106215952, + "grad_norm": 0.0802612379193306, + "learning_rate": 2.594697202771014e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8330 + }, + { + "epoch": 2.426534768693628, + "flow_matching_loss": 0.016625703126192094, + "grad_norm": 0.07412584125995636, + "learning_rate": 2.5935661222864888e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8340 + }, + { + "epoch": 2.4294442828047718, + "flow_matching_loss": 0.017746695782989264, + "grad_norm": 0.09889943152666092, + "learning_rate": 2.592433713016008e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8350 + }, + { + "epoch": 2.432353796915915, + "flow_matching_loss": 0.013939985074102879, + "grad_norm": 0.08268271386623383, + "learning_rate": 2.5912999763355553e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8360 + }, + { + "epoch": 2.4352633110270583, + "flow_matching_loss": 0.015081868972629309, + "grad_norm": 0.0791548639535904, + "learning_rate": 2.590164913622729e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8370 + }, + { + "epoch": 2.438172825138202, + "flow_matching_loss": 0.016892579011619092, + "grad_norm": 0.09906744956970215, + "learning_rate": 2.589028526256737e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8380 + }, + { + "epoch": 2.4410823392493453, + "flow_matching_loss": 0.01623848723247647, + "grad_norm": 0.06402256339788437, + "learning_rate": 2.587890815618397e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8390 + }, + { + "epoch": 2.443991853360489, + "flow_matching_loss": 0.016782090160995723, + "grad_norm": 0.11139404028654099, + "learning_rate": 2.5867517830901343e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8400 + }, + { + "epoch": 2.4469013674716322, + "flow_matching_loss": 0.015975214634090663, + "grad_norm": 0.07841439545154572, + "learning_rate": 2.5856114300559812e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8410 + }, + { + "epoch": 2.4498108815827755, + "flow_matching_loss": 0.01692858589813113, + "grad_norm": 0.11154390871524811, + "learning_rate": 2.584469757901574e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8420 + }, + { + "epoch": 2.452720395693919, + "flow_matching_loss": 0.014338553790003062, + "grad_norm": 0.061450738459825516, + "learning_rate": 2.5833267680141526e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8430 + }, + { + "epoch": 2.4556299098050625, + "flow_matching_loss": 0.015503286756575108, + "grad_norm": 0.09279503673315048, + "learning_rate": 2.5821824617825564e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8440 + }, + { + "epoch": 2.458539423916206, + "flow_matching_loss": 0.01585025144740939, + "grad_norm": 0.08564874529838562, + "learning_rate": 2.5810368405972265e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8450 + }, + { + "epoch": 2.4614489380273494, + "flow_matching_loss": 0.016200677119195463, + "grad_norm": 0.07401000708341599, + "learning_rate": 2.5798899058501996e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8460 + }, + { + "epoch": 2.4643584521384927, + "flow_matching_loss": 0.014444562513381243, + "grad_norm": 0.07762070000171661, + "learning_rate": 2.578741658935111e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8470 + }, + { + "epoch": 2.4672679662496364, + "flow_matching_loss": 0.015426216647028923, + "grad_norm": 0.07920286059379578, + "learning_rate": 2.5775921012471875e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8480 + }, + { + "epoch": 2.4701774803607797, + "flow_matching_loss": 0.015282637905329466, + "grad_norm": 0.0873798355460167, + "learning_rate": 2.5764412341832505e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8490 + }, + { + "epoch": 2.4730869944719234, + "flow_matching_loss": 0.016729251109063627, + "grad_norm": 0.11981558799743652, + "learning_rate": 2.5752890591417125e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8500 + }, + { + "epoch": 2.4759965085830666, + "flow_matching_loss": 0.01765403039753437, + "grad_norm": 0.07959409803152084, + "learning_rate": 2.5741355775225744e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8510 + }, + { + "epoch": 2.47890602269421, + "flow_matching_loss": 0.016284286603331567, + "grad_norm": 0.1161184161901474, + "learning_rate": 2.5729807907274254e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8520 + }, + { + "epoch": 2.4818155368053536, + "flow_matching_loss": 0.016473979409784078, + "grad_norm": 0.09296166896820068, + "learning_rate": 2.5718247001594403e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8530 + }, + { + "epoch": 2.484725050916497, + "flow_matching_loss": 0.01506030298769474, + "grad_norm": 0.07755027711391449, + "learning_rate": 2.5706673072233776e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8540 + }, + { + "epoch": 2.4876345650276406, + "flow_matching_loss": 0.015932923182845116, + "grad_norm": 0.07190908491611481, + "learning_rate": 2.5695086133255795e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8550 + }, + { + "epoch": 2.490544079138784, + "flow_matching_loss": 0.017024032678455113, + "grad_norm": 0.07529281824827194, + "learning_rate": 2.5683486198739687e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8560 + }, + { + "epoch": 2.493453593249927, + "flow_matching_loss": 0.01660259123891592, + "grad_norm": 0.0828472226858139, + "learning_rate": 2.5671873282780452e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8570 + }, + { + "epoch": 2.496363107361071, + "flow_matching_loss": 0.016667408775538205, + "grad_norm": 0.08625348657369614, + "learning_rate": 2.5660247399488887e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8580 + }, + { + "epoch": 2.499272621472214, + "flow_matching_loss": 0.01603578059002757, + "grad_norm": 0.07573306560516357, + "learning_rate": 2.564860856299154e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8590 + }, + { + "epoch": 2.5021821355833573, + "flow_matching_loss": 0.015590169373899699, + "grad_norm": 0.09372639656066895, + "learning_rate": 2.5636956787430693e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8600 + }, + { + "epoch": 2.505091649694501, + "flow_matching_loss": 0.017080802656710147, + "grad_norm": 0.07568460702896118, + "learning_rate": 2.5625292086964346e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8610 + }, + { + "epoch": 2.5080011638056443, + "flow_matching_loss": 0.017413893528282643, + "grad_norm": 0.1157645732164383, + "learning_rate": 2.5613614475766212e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8620 + }, + { + "epoch": 2.510910677916788, + "flow_matching_loss": 0.016290823742747308, + "grad_norm": 0.07187263667583466, + "learning_rate": 2.5601923968025694e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8630 + }, + { + "epoch": 2.5138201920279313, + "flow_matching_loss": 0.014991325419396162, + "grad_norm": 0.08803549408912659, + "learning_rate": 2.5590220577947862e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8640 + }, + { + "epoch": 2.516729706139075, + "flow_matching_loss": 0.01642779540270567, + "grad_norm": 0.10450978577136993, + "learning_rate": 2.5578504319753427e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8650 + }, + { + "epoch": 2.5196392202502182, + "flow_matching_loss": 0.01629939591512084, + "grad_norm": 0.07677729427814484, + "learning_rate": 2.556677520767876e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8660 + }, + { + "epoch": 2.5225487343613615, + "flow_matching_loss": 0.014951254427433013, + "grad_norm": 0.06766960769891739, + "learning_rate": 2.5555033255975835e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8670 + }, + { + "epoch": 2.525458248472505, + "flow_matching_loss": 0.015571688674390316, + "grad_norm": 0.08217111974954605, + "learning_rate": 2.5543278478912228e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8680 + }, + { + "epoch": 2.5283677625836485, + "flow_matching_loss": 0.01702773626893759, + "grad_norm": 0.09228511899709702, + "learning_rate": 2.55315108907711e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8690 + }, + { + "epoch": 2.5312772766947917, + "flow_matching_loss": 0.01727589061483741, + "grad_norm": 0.09963161498308182, + "learning_rate": 2.551973050585119e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8700 + }, + { + "epoch": 2.5341867908059355, + "flow_matching_loss": 0.015458132140338421, + "grad_norm": 0.08121296018362045, + "learning_rate": 2.5507937338466768e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8710 + }, + { + "epoch": 2.5370963049170787, + "flow_matching_loss": 0.014396661799401045, + "grad_norm": 0.08168414235115051, + "learning_rate": 2.549613140294764e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8720 + }, + { + "epoch": 2.5400058190282224, + "flow_matching_loss": 0.015005306154489518, + "grad_norm": 0.09581513702869415, + "learning_rate": 2.548431271363914e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8730 + }, + { + "epoch": 2.5429153331393657, + "flow_matching_loss": 0.017050480004400015, + "grad_norm": 0.07747106999158859, + "learning_rate": 2.5472481284902092e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8740 + }, + { + "epoch": 2.5458248472505094, + "flow_matching_loss": 0.016053544357419014, + "grad_norm": 0.10051104426383972, + "learning_rate": 2.5460637131112788e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8750 + }, + { + "epoch": 2.5487343613616527, + "flow_matching_loss": 0.016183933429419994, + "grad_norm": 0.10861380398273468, + "learning_rate": 2.5448780266662995e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8760 + }, + { + "epoch": 2.551643875472796, + "flow_matching_loss": 0.017637730110436678, + "grad_norm": 0.07638845592737198, + "learning_rate": 2.5436910705959926e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8770 + }, + { + "epoch": 2.5545533895839396, + "flow_matching_loss": 0.01613210644572973, + "grad_norm": 0.07043460756540298, + "learning_rate": 2.542502846342621e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8780 + }, + { + "epoch": 2.557462903695083, + "flow_matching_loss": 0.017090037930756807, + "grad_norm": 0.09041484445333481, + "learning_rate": 2.5413133553499892e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8790 + }, + { + "epoch": 2.560372417806226, + "flow_matching_loss": 0.017914092727005483, + "grad_norm": 0.0740489810705185, + "learning_rate": 2.540122599063442e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8800 + }, + { + "epoch": 2.56328193191737, + "flow_matching_loss": 0.015627609007060526, + "grad_norm": 0.07219716906547546, + "learning_rate": 2.538930578929859e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8810 + }, + { + "epoch": 2.566191446028513, + "flow_matching_loss": 0.018298513442277908, + "grad_norm": 0.10164990276098251, + "learning_rate": 2.537737296397658e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8820 + }, + { + "epoch": 2.569100960139657, + "flow_matching_loss": 0.018487638793885708, + "grad_norm": 0.08901666104793549, + "learning_rate": 2.5365427529167903e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8830 + }, + { + "epoch": 2.5720104742508, + "flow_matching_loss": 0.015038092061877251, + "grad_norm": 0.09261022508144379, + "learning_rate": 2.5353469499387375e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8840 + }, + { + "epoch": 2.574919988361944, + "flow_matching_loss": 0.017097650188952686, + "grad_norm": 0.07215695828199387, + "learning_rate": 2.534149888916514e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8850 + }, + { + "epoch": 2.577829502473087, + "flow_matching_loss": 0.014233542513102293, + "grad_norm": 0.09334525465965271, + "learning_rate": 2.5329515713046616e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8860 + }, + { + "epoch": 2.5807390165842303, + "flow_matching_loss": 0.015516739059239625, + "grad_norm": 0.07499765604734421, + "learning_rate": 2.531751998559249e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8870 + }, + { + "epoch": 2.583648530695374, + "flow_matching_loss": 0.015188798680901528, + "grad_norm": 0.08905529975891113, + "learning_rate": 2.5305511721378703e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8880 + }, + { + "epoch": 2.5865580448065173, + "flow_matching_loss": 0.015663477778434753, + "grad_norm": 0.10457529127597809, + "learning_rate": 2.529349093499642e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8890 + }, + { + "epoch": 2.5894675589176606, + "flow_matching_loss": 0.015633022971451284, + "grad_norm": 0.08079066872596741, + "learning_rate": 2.528145764105205e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8900 + }, + { + "epoch": 2.5923770730288043, + "flow_matching_loss": 0.015501982904970646, + "grad_norm": 0.07249439507722855, + "learning_rate": 2.5269411854167166e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8910 + }, + { + "epoch": 2.5952865871399475, + "flow_matching_loss": 0.01829608529806137, + "grad_norm": 0.07552612572908401, + "learning_rate": 2.5257353588978534e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8920 + }, + { + "epoch": 2.598196101251091, + "flow_matching_loss": 0.016371575929224492, + "grad_norm": 0.1026063933968544, + "learning_rate": 2.5245282860138083e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8930 + }, + { + "epoch": 2.6011056153622345, + "flow_matching_loss": 0.015549315325915813, + "grad_norm": 0.08165484666824341, + "learning_rate": 2.5233199682312894e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8940 + }, + { + "epoch": 2.604015129473378, + "flow_matching_loss": 0.015000383090227842, + "grad_norm": 0.055332690477371216, + "learning_rate": 2.5221104070185163e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8950 + }, + { + "epoch": 2.6069246435845215, + "flow_matching_loss": 0.016515743173658847, + "grad_norm": 0.08178418129682541, + "learning_rate": 2.5208996038452204e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8960 + }, + { + "epoch": 2.6098341576956647, + "flow_matching_loss": 0.015181224979460239, + "grad_norm": 0.0980820506811142, + "learning_rate": 2.5196875601826403e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8970 + }, + { + "epoch": 2.6127436718068084, + "flow_matching_loss": 0.015769525058567525, + "grad_norm": 0.07630413770675659, + "learning_rate": 2.518474277503525e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8980 + }, + { + "epoch": 2.6156531859179517, + "flow_matching_loss": 0.01592537760734558, + "grad_norm": 0.07979630678892136, + "learning_rate": 2.5172597572821266e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 8990 + }, + { + "epoch": 2.618562700029095, + "flow_matching_loss": 0.015098175685852765, + "grad_norm": 0.07829853147268295, + "learning_rate": 2.5160440009942015e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9000 + }, + { + "epoch": 2.6214722141402387, + "flow_matching_loss": 0.015383510012179613, + "grad_norm": 0.09768007695674896, + "learning_rate": 2.514827010117009e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9010 + }, + { + "epoch": 2.624381728251382, + "flow_matching_loss": 0.015145967528223992, + "grad_norm": 0.08862598240375519, + "learning_rate": 2.5136087861293068e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9020 + }, + { + "epoch": 2.627291242362525, + "flow_matching_loss": 0.015580752026289701, + "grad_norm": 0.09620499610900879, + "learning_rate": 2.5123893305113525e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9030 + }, + { + "epoch": 2.630200756473669, + "flow_matching_loss": 0.01559119364246726, + "grad_norm": 0.0745769813656807, + "learning_rate": 2.5111686447448997e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9040 + }, + { + "epoch": 2.633110270584812, + "flow_matching_loss": 0.016066277492791416, + "grad_norm": 0.07894081622362137, + "learning_rate": 2.5099467303131966e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9050 + }, + { + "epoch": 2.636019784695956, + "flow_matching_loss": 0.015427187085151672, + "grad_norm": 0.07842984795570374, + "learning_rate": 2.508723588700984e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9060 + }, + { + "epoch": 2.638929298807099, + "flow_matching_loss": 0.01619447683915496, + "grad_norm": 0.06613453477621078, + "learning_rate": 2.5074992213944952e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9070 + }, + { + "epoch": 2.641838812918243, + "flow_matching_loss": 0.015470034163445234, + "grad_norm": 0.07790035754442215, + "learning_rate": 2.506273629881451e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9080 + }, + { + "epoch": 2.644748327029386, + "flow_matching_loss": 0.01765516111627221, + "grad_norm": 0.08576702326536179, + "learning_rate": 2.5050468156510617e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9090 + }, + { + "epoch": 2.6476578411405294, + "flow_matching_loss": 0.01496027149260044, + "grad_norm": 0.07592269778251648, + "learning_rate": 2.5038187801940215e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9100 + }, + { + "epoch": 2.650567355251673, + "flow_matching_loss": 0.01610537599772215, + "grad_norm": 0.06679565459489822, + "learning_rate": 2.5025895250025088e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9110 + }, + { + "epoch": 2.6534768693628163, + "flow_matching_loss": 0.016818077117204667, + "grad_norm": 0.0946272611618042, + "learning_rate": 2.501359051570186e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9120 + }, + { + "epoch": 2.6563863834739596, + "flow_matching_loss": 0.016765352711081503, + "grad_norm": 0.08136779814958572, + "learning_rate": 2.500127361392193e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9130 + }, + { + "epoch": 2.6592958975851033, + "flow_matching_loss": 0.015825557336211205, + "grad_norm": 0.0812956914305687, + "learning_rate": 2.49889445596515e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9140 + }, + { + "epoch": 2.6622054116962466, + "flow_matching_loss": 0.016228922177106142, + "grad_norm": 0.07115481793880463, + "learning_rate": 2.497660336787154e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9150 + }, + { + "epoch": 2.6651149258073903, + "flow_matching_loss": 0.015736792329698802, + "grad_norm": 0.0937192440032959, + "learning_rate": 2.496425005357775e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9160 + }, + { + "epoch": 2.6680244399185336, + "flow_matching_loss": 0.017475182097405194, + "grad_norm": 0.10165403038263321, + "learning_rate": 2.4951884631780578e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9170 + }, + { + "epoch": 2.6709339540296773, + "flow_matching_loss": 0.01548903388902545, + "grad_norm": 0.06392738968133926, + "learning_rate": 2.493950711750518e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9180 + }, + { + "epoch": 2.6738434681408205, + "flow_matching_loss": 0.01827444154769182, + "grad_norm": 0.10442104190587997, + "learning_rate": 2.49271175257914e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9190 + }, + { + "epoch": 2.676752982251964, + "flow_matching_loss": 0.01667042775079608, + "grad_norm": 0.09201356023550034, + "learning_rate": 2.4914715871693757e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9200 + }, + { + "epoch": 2.6796624963631075, + "flow_matching_loss": 0.01748319622129202, + "grad_norm": 0.11368203908205032, + "learning_rate": 2.4902302170281438e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9210 + }, + { + "epoch": 2.6825720104742508, + "flow_matching_loss": 0.016765958070755003, + "grad_norm": 0.10431733727455139, + "learning_rate": 2.4889876436638252e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9220 + }, + { + "epoch": 2.685481524585394, + "flow_matching_loss": 0.016432017739862203, + "grad_norm": 0.09550170600414276, + "learning_rate": 2.4877438685862643e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9230 + }, + { + "epoch": 2.6883910386965377, + "flow_matching_loss": 0.014995853323489428, + "grad_norm": 0.07246292382478714, + "learning_rate": 2.486498893306765e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9240 + }, + { + "epoch": 2.691300552807681, + "flow_matching_loss": 0.016036652494221926, + "grad_norm": 0.0898195207118988, + "learning_rate": 2.485252719338089e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9250 + }, + { + "epoch": 2.6942100669188247, + "flow_matching_loss": 0.01691264696419239, + "grad_norm": 0.09035919606685638, + "learning_rate": 2.4840053481944563e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9260 + }, + { + "epoch": 2.697119581029968, + "flow_matching_loss": 0.015686024725437165, + "grad_norm": 0.09571180492639542, + "learning_rate": 2.4827567813915398e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9270 + }, + { + "epoch": 2.7000290951411117, + "flow_matching_loss": 0.01712434496730566, + "grad_norm": 0.07744565606117249, + "learning_rate": 2.481507020446466e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9280 + }, + { + "epoch": 2.702938609252255, + "flow_matching_loss": 0.01630834210664034, + "grad_norm": 0.0672847181558609, + "learning_rate": 2.4802560668778122e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9290 + }, + { + "epoch": 2.705848123363398, + "flow_matching_loss": 0.015662869717925788, + "grad_norm": 0.0879279226064682, + "learning_rate": 2.479003922205605e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9300 + }, + { + "epoch": 2.708757637474542, + "flow_matching_loss": 0.01594325704500079, + "grad_norm": 0.06991494446992874, + "learning_rate": 2.4777505879513183e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9310 + }, + { + "epoch": 2.711667151585685, + "flow_matching_loss": 0.015211998857557774, + "grad_norm": 0.09269137680530548, + "learning_rate": 2.4764960656378714e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9320 + }, + { + "epoch": 2.7145766656968284, + "flow_matching_loss": 0.014068638160824775, + "grad_norm": 0.08727968484163284, + "learning_rate": 2.4752403567896274e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9330 + }, + { + "epoch": 2.717486179807972, + "flow_matching_loss": 0.014367791917175054, + "grad_norm": 0.09582386165857315, + "learning_rate": 2.47398346293239e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9340 + }, + { + "epoch": 2.7203956939191154, + "flow_matching_loss": 0.016246006544679403, + "grad_norm": 0.10427840054035187, + "learning_rate": 2.4727253855934055e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9350 + }, + { + "epoch": 2.723305208030259, + "flow_matching_loss": 0.0161787249147892, + "grad_norm": 0.06124483048915863, + "learning_rate": 2.4714661263013548e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9360 + }, + { + "epoch": 2.7262147221414024, + "flow_matching_loss": 0.016100141033530236, + "grad_norm": 0.10400816798210144, + "learning_rate": 2.4702056865863575e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9370 + }, + { + "epoch": 2.729124236252546, + "flow_matching_loss": 0.016270245891064405, + "grad_norm": 0.08667192608118057, + "learning_rate": 2.4689440679799672e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9380 + }, + { + "epoch": 2.7320337503636893, + "flow_matching_loss": 0.015190940909087658, + "grad_norm": 0.09287340193986893, + "learning_rate": 2.4676812720151678e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9390 + }, + { + "epoch": 2.7349432644748326, + "flow_matching_loss": 0.015566894691437483, + "grad_norm": 0.06336111575365067, + "learning_rate": 2.466417300226377e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9400 + }, + { + "epoch": 2.7378527785859763, + "flow_matching_loss": 0.01602795282378793, + "grad_norm": 0.07335017621517181, + "learning_rate": 2.4651521541494394e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9410 + }, + { + "epoch": 2.7407622926971196, + "flow_matching_loss": 0.016490754764527084, + "grad_norm": 0.0796826034784317, + "learning_rate": 2.4638858353216267e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9420 + }, + { + "epoch": 2.743671806808263, + "flow_matching_loss": 0.016662957798689603, + "grad_norm": 0.0941772311925888, + "learning_rate": 2.4626183452816355e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9430 + }, + { + "epoch": 2.7465813209194065, + "flow_matching_loss": 0.01469989325851202, + "grad_norm": 0.07162295281887054, + "learning_rate": 2.4613496855695855e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9440 + }, + { + "epoch": 2.74949083503055, + "flow_matching_loss": 0.014436995424330235, + "grad_norm": 0.07710712403059006, + "learning_rate": 2.4600798577270183e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9450 + }, + { + "epoch": 2.752400349141693, + "flow_matching_loss": 0.016104507073760033, + "grad_norm": 0.09488355368375778, + "learning_rate": 2.4588088632968944e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9460 + }, + { + "epoch": 2.7553098632528368, + "flow_matching_loss": 0.015470258146524429, + "grad_norm": 0.08227191120386124, + "learning_rate": 2.4575367038235916e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9470 + }, + { + "epoch": 2.7582193773639805, + "flow_matching_loss": 0.015377181861549617, + "grad_norm": 0.092167928814888, + "learning_rate": 2.4562633808529043e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9480 + }, + { + "epoch": 2.7611288914751237, + "flow_matching_loss": 0.016913816425949336, + "grad_norm": 0.11207162588834763, + "learning_rate": 2.4549888959320396e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9490 + }, + { + "epoch": 2.764038405586267, + "flow_matching_loss": 0.015725202020257713, + "grad_norm": 0.05276364088058472, + "learning_rate": 2.4537132506096162e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9500 + }, + { + "epoch": 2.7669479196974107, + "flow_matching_loss": 0.014307956118136645, + "grad_norm": 0.10448216646909714, + "learning_rate": 2.452436446435664e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9510 + }, + { + "epoch": 2.769857433808554, + "flow_matching_loss": 0.017557715345174075, + "grad_norm": 0.09501653909683228, + "learning_rate": 2.451158484961621e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9520 + }, + { + "epoch": 2.7727669479196972, + "flow_matching_loss": 0.017688268143683673, + "grad_norm": 0.09256640821695328, + "learning_rate": 2.4498793677403304e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9530 + }, + { + "epoch": 2.775676462030841, + "flow_matching_loss": 0.017225933074951173, + "grad_norm": 0.0798371285200119, + "learning_rate": 2.4485990963260398e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9540 + }, + { + "epoch": 2.778585976141984, + "flow_matching_loss": 0.016357009205967186, + "grad_norm": 0.10337930917739868, + "learning_rate": 2.4473176722744e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9550 + }, + { + "epoch": 2.7814954902531275, + "flow_matching_loss": 0.018202284071594476, + "grad_norm": 0.08487426489591599, + "learning_rate": 2.446035097142463e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9560 + }, + { + "epoch": 2.784405004364271, + "flow_matching_loss": 0.016866580676287414, + "grad_norm": 0.08879834413528442, + "learning_rate": 2.444751372488677e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9570 + }, + { + "epoch": 2.7873145184754144, + "flow_matching_loss": 0.01857468169182539, + "grad_norm": 0.05794541910290718, + "learning_rate": 2.443466499872889e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9580 + }, + { + "epoch": 2.790224032586558, + "flow_matching_loss": 0.014288917556405068, + "grad_norm": 0.07364710420370102, + "learning_rate": 2.4421804808563413e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9590 + }, + { + "epoch": 2.7931335466977014, + "flow_matching_loss": 0.018690633680671455, + "grad_norm": 0.08846189826726913, + "learning_rate": 2.440893317001667e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9600 + }, + { + "epoch": 2.796043060808845, + "flow_matching_loss": 0.01478832308202982, + "grad_norm": 0.08696884661912918, + "learning_rate": 2.4396050098728925e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9610 + }, + { + "epoch": 2.7989525749199884, + "flow_matching_loss": 0.015587813965976239, + "grad_norm": 0.06989570707082748, + "learning_rate": 2.4383155610354314e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9620 + }, + { + "epoch": 2.8018620890311317, + "flow_matching_loss": 0.01854343991726637, + "grad_norm": 0.10138457268476486, + "learning_rate": 2.437024972056087e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9630 + }, + { + "epoch": 2.8047716031422754, + "flow_matching_loss": 0.015503582637757063, + "grad_norm": 0.0890192985534668, + "learning_rate": 2.4357332445030452e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9640 + }, + { + "epoch": 2.8076811172534186, + "flow_matching_loss": 0.016867614164948464, + "grad_norm": 0.09498883038759232, + "learning_rate": 2.434440379945878e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9650 + }, + { + "epoch": 2.810590631364562, + "flow_matching_loss": 0.016877350118011235, + "grad_norm": 0.12455708533525467, + "learning_rate": 2.4331463799555366e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9660 + }, + { + "epoch": 2.8135001454757056, + "flow_matching_loss": 0.015054069552570581, + "grad_norm": 0.07714275270700455, + "learning_rate": 2.4318512461043544e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9670 + }, + { + "epoch": 2.816409659586849, + "flow_matching_loss": 0.015206705499440431, + "grad_norm": 0.12289591878652573, + "learning_rate": 2.4305549799660397e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9680 + }, + { + "epoch": 2.8193191736979926, + "flow_matching_loss": 0.016265198402106763, + "grad_norm": 0.08894221484661102, + "learning_rate": 2.429257583115679e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9690 + }, + { + "epoch": 2.822228687809136, + "flow_matching_loss": 0.017174868565052747, + "grad_norm": 0.083670474588871, + "learning_rate": 2.4279590571297316e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9700 + }, + { + "epoch": 2.8251382019202795, + "flow_matching_loss": 0.016140423901379108, + "grad_norm": 0.09961961209774017, + "learning_rate": 2.4266594035860288e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9710 + }, + { + "epoch": 2.828047716031423, + "flow_matching_loss": 0.016245375387370586, + "grad_norm": 0.09243503212928772, + "learning_rate": 2.425358624063773e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9720 + }, + { + "epoch": 2.830957230142566, + "flow_matching_loss": 0.014185183309018612, + "grad_norm": 0.08667387813329697, + "learning_rate": 2.424056720143532e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9730 + }, + { + "epoch": 2.8338667442537098, + "flow_matching_loss": 0.015114926639944315, + "grad_norm": 0.08263567090034485, + "learning_rate": 2.422753693407244e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9740 + }, + { + "epoch": 2.836776258364853, + "flow_matching_loss": 0.0148558066226542, + "grad_norm": 0.09814136475324631, + "learning_rate": 2.421449545438209e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9750 + }, + { + "epoch": 2.8396857724759963, + "flow_matching_loss": 0.017969620134681463, + "grad_norm": 0.0857098400592804, + "learning_rate": 2.4201442778210887e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9760 + }, + { + "epoch": 2.84259528658714, + "flow_matching_loss": 0.016136865131556988, + "grad_norm": 0.07886181771755219, + "learning_rate": 2.418837892141907e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9770 + }, + { + "epoch": 2.8455048006982833, + "flow_matching_loss": 0.01588539471849799, + "grad_norm": 0.07093358039855957, + "learning_rate": 2.4175303899880458e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9780 + }, + { + "epoch": 2.848414314809427, + "flow_matching_loss": 0.015897923056036233, + "grad_norm": 0.06196693331003189, + "learning_rate": 2.4162217729482427e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9790 + }, + { + "epoch": 2.8513238289205702, + "flow_matching_loss": 0.014537472557276488, + "grad_norm": 0.12211130559444427, + "learning_rate": 2.4149120426125918e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9800 + }, + { + "epoch": 2.854233343031714, + "flow_matching_loss": 0.014733227156102657, + "grad_norm": 0.0765785351395607, + "learning_rate": 2.413601200572538e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9810 + }, + { + "epoch": 2.857142857142857, + "flow_matching_loss": 0.015478320326656103, + "grad_norm": 0.07804866135120392, + "learning_rate": 2.4122892484208794e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9820 + }, + { + "epoch": 2.8600523712540005, + "flow_matching_loss": 0.016539573855698108, + "grad_norm": 0.09049414098262787, + "learning_rate": 2.4109761877517603e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9830 + }, + { + "epoch": 2.862961885365144, + "flow_matching_loss": 0.017017710953950882, + "grad_norm": 0.08025554567575455, + "learning_rate": 2.4096620201606737e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9840 + }, + { + "epoch": 2.8658713994762874, + "flow_matching_loss": 0.014235160406678915, + "grad_norm": 0.06290726363658905, + "learning_rate": 2.4083467472444573e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9850 + }, + { + "epoch": 2.8687809135874307, + "flow_matching_loss": 0.014623530395328998, + "grad_norm": 0.0755898728966713, + "learning_rate": 2.4070303706012912e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9860 + }, + { + "epoch": 2.8716904276985744, + "flow_matching_loss": 0.01587908938527107, + "grad_norm": 0.0824689120054245, + "learning_rate": 2.4057128918306975e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9870 + }, + { + "epoch": 2.8745999418097177, + "flow_matching_loss": 0.015053980238735675, + "grad_norm": 0.08085981756448746, + "learning_rate": 2.4043943125335373e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9880 + }, + { + "epoch": 2.8775094559208614, + "flow_matching_loss": 0.016866143513470887, + "grad_norm": 0.09555201232433319, + "learning_rate": 2.403074634312008e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9890 + }, + { + "epoch": 2.8804189700320046, + "flow_matching_loss": 0.016470300406217574, + "grad_norm": 0.09441875666379929, + "learning_rate": 2.4017538587696442e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9900 + }, + { + "epoch": 2.8833284841431484, + "flow_matching_loss": 0.014890693034976721, + "grad_norm": 0.06815000623464584, + "learning_rate": 2.4004319875113118e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9910 + }, + { + "epoch": 2.8862379982542916, + "flow_matching_loss": 0.01593627892434597, + "grad_norm": 0.08270730078220367, + "learning_rate": 2.3991090221432096e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9920 + }, + { + "epoch": 2.889147512365435, + "flow_matching_loss": 0.01612103981897235, + "grad_norm": 0.07483769208192825, + "learning_rate": 2.397784964272865e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9930 + }, + { + "epoch": 2.8920570264765786, + "flow_matching_loss": 0.015913822036236525, + "grad_norm": 0.08344195038080215, + "learning_rate": 2.3964598155091335e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9940 + }, + { + "epoch": 2.894966540587722, + "flow_matching_loss": 0.016424527578055858, + "grad_norm": 0.06890333443880081, + "learning_rate": 2.3951335774621952e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9950 + }, + { + "epoch": 2.897876054698865, + "flow_matching_loss": 0.01693023629486561, + "grad_norm": 0.07427181303501129, + "learning_rate": 2.3938062517435548e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9960 + }, + { + "epoch": 2.900785568810009, + "flow_matching_loss": 0.018289812467992307, + "grad_norm": 0.08422599732875824, + "learning_rate": 2.392477839966038e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9970 + }, + { + "epoch": 2.903695082921152, + "flow_matching_loss": 0.015879007149487735, + "grad_norm": 0.08955325186252594, + "learning_rate": 2.3911483437437904e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9980 + }, + { + "epoch": 2.9066045970322953, + "flow_matching_loss": 0.015280448831617832, + "grad_norm": 0.090792715549469, + "learning_rate": 2.3898177646922757e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 9990 + }, + { + "epoch": 2.909514111143439, + "flow_matching_loss": 0.014453674945980311, + "grad_norm": 0.07854966074228287, + "learning_rate": 2.3884861044282722e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10000 + }, + { + "epoch": 2.9124236252545828, + "flow_matching_loss": 0.01468636654317379, + "grad_norm": 0.10502863675355911, + "learning_rate": 2.3871533645698732e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10010 + }, + { + "epoch": 2.915333139365726, + "flow_matching_loss": 0.015492429211735725, + "grad_norm": 0.07289799302816391, + "learning_rate": 2.3858195467364833e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10020 + }, + { + "epoch": 2.9182426534768693, + "flow_matching_loss": 0.013739456050097942, + "grad_norm": 0.07094506919384003, + "learning_rate": 2.3844846525488166e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10030 + }, + { + "epoch": 2.921152167588013, + "flow_matching_loss": 0.014847536757588386, + "grad_norm": 0.07667036354541779, + "learning_rate": 2.3831486836288957e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10040 + }, + { + "epoch": 2.9240616816991563, + "flow_matching_loss": 0.01514466879889369, + "grad_norm": 0.08083973079919815, + "learning_rate": 2.381811641600048e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10050 + }, + { + "epoch": 2.9269711958102995, + "flow_matching_loss": 0.015857574995607138, + "grad_norm": 0.08967342972755432, + "learning_rate": 2.3804735280869073e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10060 + }, + { + "epoch": 2.9298807099214432, + "flow_matching_loss": 0.015380233153700828, + "grad_norm": 0.07347968220710754, + "learning_rate": 2.3791343447154064e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10070 + }, + { + "epoch": 2.9327902240325865, + "flow_matching_loss": 0.015328123327344656, + "grad_norm": 0.07243114709854126, + "learning_rate": 2.37779409311278e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10080 + }, + { + "epoch": 2.9356997381437298, + "flow_matching_loss": 0.01633472153916955, + "grad_norm": 0.07957731932401657, + "learning_rate": 2.3764527749075596e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10090 + }, + { + "epoch": 2.9386092522548735, + "flow_matching_loss": 0.015757478773593903, + "grad_norm": 0.09710264950990677, + "learning_rate": 2.3751103917295735e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10100 + }, + { + "epoch": 2.9415187663660167, + "flow_matching_loss": 0.01745665296912193, + "grad_norm": 0.08406912535429001, + "learning_rate": 2.373766945209944e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10110 + }, + { + "epoch": 2.9444282804771604, + "flow_matching_loss": 0.017084371577948333, + "grad_norm": 0.0887690931558609, + "learning_rate": 2.3724224369810856e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10120 + }, + { + "epoch": 2.9473377945883037, + "flow_matching_loss": 0.016416219621896745, + "grad_norm": 0.11708499491214752, + "learning_rate": 2.3710768686767015e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10130 + }, + { + "epoch": 2.9502473086994474, + "flow_matching_loss": 0.015348739642649888, + "grad_norm": 0.0643615648150444, + "learning_rate": 2.3697302419317856e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10140 + }, + { + "epoch": 2.9531568228105907, + "flow_matching_loss": 0.016016786731779576, + "grad_norm": 0.12805472314357758, + "learning_rate": 2.3683825583826146e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10150 + }, + { + "epoch": 2.956066336921734, + "flow_matching_loss": 0.01521172933280468, + "grad_norm": 0.07232817262411118, + "learning_rate": 2.3670338196667528e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10160 + }, + { + "epoch": 2.9589758510328776, + "flow_matching_loss": 0.014402250479906797, + "grad_norm": 0.08671021461486816, + "learning_rate": 2.3656840274230443e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10170 + }, + { + "epoch": 2.961885365144021, + "flow_matching_loss": 0.015610834211111068, + "grad_norm": 0.09499255567789078, + "learning_rate": 2.3643331832916133e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10180 + }, + { + "epoch": 2.964794879255164, + "flow_matching_loss": 0.01793188862502575, + "grad_norm": 0.09580863267183304, + "learning_rate": 2.362981288913863e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10190 + }, + { + "epoch": 2.967704393366308, + "flow_matching_loss": 0.01597502799704671, + "grad_norm": 0.08713854849338531, + "learning_rate": 2.3616283459324737e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10200 + }, + { + "epoch": 2.970613907477451, + "flow_matching_loss": 0.014374320860952139, + "grad_norm": 0.08800195902585983, + "learning_rate": 2.3602743559913973e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10210 + }, + { + "epoch": 2.973523421588595, + "flow_matching_loss": 0.014742946065962315, + "grad_norm": 0.09330254048109055, + "learning_rate": 2.35891932073586e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10220 + }, + { + "epoch": 2.976432935699738, + "flow_matching_loss": 0.015739009622484444, + "grad_norm": 0.12108491361141205, + "learning_rate": 2.357563241812357e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10230 + }, + { + "epoch": 2.979342449810882, + "flow_matching_loss": 0.015564217418432235, + "grad_norm": 0.07671075314283371, + "learning_rate": 2.3562061208686522e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10240 + }, + { + "epoch": 2.982251963922025, + "flow_matching_loss": 0.01604644991457462, + "grad_norm": 0.09530340135097504, + "learning_rate": 2.354847959553776e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10250 + }, + { + "epoch": 2.9851614780331683, + "flow_matching_loss": 0.015127280447632074, + "grad_norm": 0.08848127722740173, + "learning_rate": 2.353488759518022e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10260 + }, + { + "epoch": 2.988070992144312, + "flow_matching_loss": 0.015365494973957538, + "grad_norm": 0.07028676569461823, + "learning_rate": 2.352128522412946e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10270 + }, + { + "epoch": 2.9909805062554553, + "flow_matching_loss": 0.015208727028220893, + "grad_norm": 0.07755360007286072, + "learning_rate": 2.3507672498913653e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10280 + }, + { + "epoch": 2.9938900203665986, + "flow_matching_loss": 0.015964175574481487, + "grad_norm": 0.10339933633804321, + "learning_rate": 2.3494049436073537e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10290 + }, + { + "epoch": 2.9967995344777423, + "flow_matching_loss": 0.015841569844633342, + "grad_norm": 0.07640321552753448, + "learning_rate": 2.348041605216242e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10300 + }, + { + "epoch": 2.9997090485888855, + "flow_matching_loss": 0.01562760816887021, + "grad_norm": 0.08482292294502258, + "learning_rate": 2.3466772363746146e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10310 + }, + { + "epoch": 3.0026185627000292, + "flow_matching_loss": 0.015086142625659705, + "grad_norm": 0.08816058933734894, + "learning_rate": 2.345311838740309e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10320 + }, + { + "epoch": 3.0055280768111725, + "flow_matching_loss": 0.014634321257472039, + "grad_norm": 0.09501238912343979, + "learning_rate": 2.343945413972411e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10330 + }, + { + "epoch": 3.0084375909223158, + "flow_matching_loss": 0.015055603347718716, + "grad_norm": 0.07990912348031998, + "learning_rate": 2.3425779637312568e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10340 + }, + { + "epoch": 3.0113471050334595, + "flow_matching_loss": 0.01646934598684311, + "grad_norm": 0.09662150591611862, + "learning_rate": 2.3412094896784264e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10350 + }, + { + "epoch": 3.0142566191446027, + "flow_matching_loss": 0.01717370990663767, + "grad_norm": 0.08356604725122452, + "learning_rate": 2.339839993476745e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10360 + }, + { + "epoch": 3.0171661332557465, + "flow_matching_loss": 0.015665557608008383, + "grad_norm": 0.07903715968132019, + "learning_rate": 2.3384694767902804e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10370 + }, + { + "epoch": 3.0200756473668897, + "flow_matching_loss": 0.013255877234041692, + "grad_norm": 0.07235628366470337, + "learning_rate": 2.3370979412843384e-05, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10380 + }, + { + "epoch": 3.022985161478033, + "flow_matching_loss": 0.014915210753679275, + "grad_norm": 0.0669739842414856, + "learning_rate": 2.3357253886254648e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10390 + }, + { + "epoch": 3.0258946755891767, + "flow_matching_loss": 0.01399024147540331, + "grad_norm": 0.0789620578289032, + "learning_rate": 2.3343518204814407e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10400 + }, + { + "epoch": 3.02880418970032, + "flow_matching_loss": 0.0152158847078681, + "grad_norm": 0.07901521027088165, + "learning_rate": 2.3329772385212798e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10410 + }, + { + "epoch": 3.0317137038114637, + "flow_matching_loss": 0.015124813746660947, + "grad_norm": 0.07819824665784836, + "learning_rate": 2.3316016444152296e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10420 + }, + { + "epoch": 3.034623217922607, + "flow_matching_loss": 0.015654110908508302, + "grad_norm": 0.08373551815748215, + "learning_rate": 2.3302250398347668e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10430 + }, + { + "epoch": 3.03753273203375, + "flow_matching_loss": 0.015310317277908325, + "grad_norm": 0.09592315554618835, + "learning_rate": 2.328847426452595e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10440 + }, + { + "epoch": 3.040442246144894, + "flow_matching_loss": 0.015259873028844595, + "grad_norm": 0.08175301551818848, + "learning_rate": 2.3274688059426456e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10450 + }, + { + "epoch": 3.043351760256037, + "flow_matching_loss": 0.01654182290658355, + "grad_norm": 0.07946216315031052, + "learning_rate": 2.3260891799800715e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10460 + }, + { + "epoch": 3.046261274367181, + "flow_matching_loss": 0.015810263995081186, + "grad_norm": 0.08434632420539856, + "learning_rate": 2.3247085502412482e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10470 + }, + { + "epoch": 3.049170788478324, + "flow_matching_loss": 0.015070579294115306, + "grad_norm": 0.07897178083658218, + "learning_rate": 2.3233269184037717e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10480 + }, + { + "epoch": 3.0520803025894674, + "flow_matching_loss": 0.01744923125952482, + "grad_norm": 0.08933214098215103, + "learning_rate": 2.3219442861464547e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10490 + }, + { + "epoch": 3.054989816700611, + "flow_matching_loss": 0.014710999373346568, + "grad_norm": 0.07698569446802139, + "learning_rate": 2.3205606551493255e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10500 + }, + { + "epoch": 3.0578993308117544, + "flow_matching_loss": 0.016022367868572474, + "grad_norm": 0.08459034562110901, + "learning_rate": 2.319176027093627e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10510 + }, + { + "epoch": 3.060808844922898, + "flow_matching_loss": 0.016146218962967397, + "grad_norm": 0.06623293459415436, + "learning_rate": 2.3177904036618117e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10520 + }, + { + "epoch": 3.0637183590340413, + "flow_matching_loss": 0.014846988394856453, + "grad_norm": 0.07118456810712814, + "learning_rate": 2.316403786537544e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10530 + }, + { + "epoch": 3.0666278731451846, + "flow_matching_loss": 0.016045124363154174, + "grad_norm": 0.09385216981172562, + "learning_rate": 2.3150161774056933e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10540 + }, + { + "epoch": 3.0695373872563283, + "flow_matching_loss": 0.01423181276768446, + "grad_norm": 0.08517618477344513, + "learning_rate": 2.313627577952336e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10550 + }, + { + "epoch": 3.0724469013674716, + "flow_matching_loss": 0.014636414125561714, + "grad_norm": 0.09296092391014099, + "learning_rate": 2.312237989864752e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10560 + }, + { + "epoch": 3.0753564154786153, + "flow_matching_loss": 0.01664766389876604, + "grad_norm": 0.09841103106737137, + "learning_rate": 2.3108474148314205e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10570 + }, + { + "epoch": 3.0782659295897585, + "flow_matching_loss": 0.015077568311244249, + "grad_norm": 0.08104795962572098, + "learning_rate": 2.309455854542023e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10580 + }, + { + "epoch": 3.081175443700902, + "flow_matching_loss": 0.015044816303998232, + "grad_norm": 0.08692153543233871, + "learning_rate": 2.308063310687435e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10590 + }, + { + "epoch": 3.0840849578120455, + "flow_matching_loss": 0.017141340766102076, + "grad_norm": 0.07784572243690491, + "learning_rate": 2.306669784959729e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10600 + }, + { + "epoch": 3.0869944719231888, + "flow_matching_loss": 0.015262208692729474, + "grad_norm": 0.09102693200111389, + "learning_rate": 2.3052752790521703e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10610 + }, + { + "epoch": 3.0899039860343325, + "flow_matching_loss": 0.015302827395498753, + "grad_norm": 0.0829348936676979, + "learning_rate": 2.3038797946592152e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10620 + }, + { + "epoch": 3.0928135001454757, + "flow_matching_loss": 0.015304843988269567, + "grad_norm": 0.08671735227108002, + "learning_rate": 2.3024833334765087e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10630 + }, + { + "epoch": 3.095723014256619, + "flow_matching_loss": 0.014335061516612768, + "grad_norm": 0.06356758624315262, + "learning_rate": 2.301085897200883e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10640 + }, + { + "epoch": 3.0986325283677627, + "flow_matching_loss": 0.01520595671609044, + "grad_norm": 0.07158395648002625, + "learning_rate": 2.299687487530354e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10650 + }, + { + "epoch": 3.101542042478906, + "flow_matching_loss": 0.015587270818650722, + "grad_norm": 0.07198874652385712, + "learning_rate": 2.2982881061641228e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10660 + }, + { + "epoch": 3.1044515565900497, + "flow_matching_loss": 0.01595425121486187, + "grad_norm": 0.08251528441905975, + "learning_rate": 2.296887754802569e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10670 + }, + { + "epoch": 3.107361070701193, + "flow_matching_loss": 0.01611713245511055, + "grad_norm": 0.06464014202356339, + "learning_rate": 2.295486435147251e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10680 + }, + { + "epoch": 3.110270584812336, + "flow_matching_loss": 0.015240859426558017, + "grad_norm": 0.07106533646583557, + "learning_rate": 2.294084148900905e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10690 + }, + { + "epoch": 3.11318009892348, + "flow_matching_loss": 0.014856058545410633, + "grad_norm": 0.09394925832748413, + "learning_rate": 2.292680897767441e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10700 + }, + { + "epoch": 3.116089613034623, + "flow_matching_loss": 0.016244260128587484, + "grad_norm": 0.07793572545051575, + "learning_rate": 2.291276683451941e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10710 + }, + { + "epoch": 3.1189991271457664, + "flow_matching_loss": 0.015232953336089849, + "grad_norm": 0.09921017289161682, + "learning_rate": 2.2898715076606584e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10720 + }, + { + "epoch": 3.12190864125691, + "flow_matching_loss": 0.013086782768368721, + "grad_norm": 0.06439618021249771, + "learning_rate": 2.2884653721010135e-05, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10730 + }, + { + "epoch": 3.1248181553680534, + "flow_matching_loss": 0.016818947438150645, + "grad_norm": 0.10020428895950317, + "learning_rate": 2.2870582784815946e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10740 + }, + { + "epoch": 3.127727669479197, + "flow_matching_loss": 0.014591674879193306, + "grad_norm": 0.06057731434702873, + "learning_rate": 2.2856502285121525e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10750 + }, + { + "epoch": 3.1306371835903404, + "flow_matching_loss": 0.01649843892082572, + "grad_norm": 0.10226329416036606, + "learning_rate": 2.2842412239036004e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10760 + }, + { + "epoch": 3.133546697701484, + "flow_matching_loss": 0.015220282226800918, + "grad_norm": 0.08608047664165497, + "learning_rate": 2.2828312663680125e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10770 + }, + { + "epoch": 3.1364562118126273, + "flow_matching_loss": 0.01581614101305604, + "grad_norm": 0.08744465559720993, + "learning_rate": 2.2814203576186194e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10780 + }, + { + "epoch": 3.1393657259237706, + "flow_matching_loss": 0.01531647825613618, + "grad_norm": 0.07064255326986313, + "learning_rate": 2.2800084993698093e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10790 + }, + { + "epoch": 3.1422752400349143, + "flow_matching_loss": 0.015096652973443269, + "grad_norm": 0.09361318498849869, + "learning_rate": 2.2785956933371224e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10800 + }, + { + "epoch": 3.1451847541460576, + "flow_matching_loss": 0.01618814719840884, + "grad_norm": 0.07742153853178024, + "learning_rate": 2.2771819412372505e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10810 + }, + { + "epoch": 3.148094268257201, + "flow_matching_loss": 0.01446524541825056, + "grad_norm": 0.07723217457532883, + "learning_rate": 2.2757672447880374e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10820 + }, + { + "epoch": 3.1510037823683446, + "flow_matching_loss": 0.01546453284099698, + "grad_norm": 0.09021779894828796, + "learning_rate": 2.2743516057084712e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10830 + }, + { + "epoch": 3.153913296479488, + "flow_matching_loss": 0.015567675977945328, + "grad_norm": 0.08657471090555191, + "learning_rate": 2.2729350257186877e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10840 + }, + { + "epoch": 3.1568228105906315, + "flow_matching_loss": 0.0146649525500834, + "grad_norm": 0.07971818745136261, + "learning_rate": 2.2715175065399655e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10850 + }, + { + "epoch": 3.159732324701775, + "flow_matching_loss": 0.014686002396047115, + "grad_norm": 0.10466691106557846, + "learning_rate": 2.270099049894723e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10860 + }, + { + "epoch": 3.162641838812918, + "flow_matching_loss": 0.015355703327804804, + "grad_norm": 0.07063841074705124, + "learning_rate": 2.268679657506519e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10870 + }, + { + "epoch": 3.1655513529240618, + "flow_matching_loss": 0.015317789558321238, + "grad_norm": 0.0990372970700264, + "learning_rate": 2.2672593311000498e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10880 + }, + { + "epoch": 3.168460867035205, + "flow_matching_loss": 0.01491024810820818, + "grad_norm": 0.0790305882692337, + "learning_rate": 2.2658380724011452e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10890 + }, + { + "epoch": 3.1713703811463487, + "flow_matching_loss": 0.014408760797232389, + "grad_norm": 0.07145271450281143, + "learning_rate": 2.264415883136769e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10900 + }, + { + "epoch": 3.174279895257492, + "flow_matching_loss": 0.01596514331176877, + "grad_norm": 0.08453219383955002, + "learning_rate": 2.262992765035015e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10910 + }, + { + "epoch": 3.1771894093686353, + "flow_matching_loss": 0.014927462488412858, + "grad_norm": 0.09978280961513519, + "learning_rate": 2.2615687198251062e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10920 + }, + { + "epoch": 3.180098923479779, + "flow_matching_loss": 0.016310163866728546, + "grad_norm": 0.0764085054397583, + "learning_rate": 2.2601437492373914e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10930 + }, + { + "epoch": 3.1830084375909222, + "flow_matching_loss": 0.016740838531404734, + "grad_norm": 0.06889592111110687, + "learning_rate": 2.2587178550033444e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10940 + }, + { + "epoch": 3.185917951702066, + "flow_matching_loss": 0.015051318891346454, + "grad_norm": 0.0708032101392746, + "learning_rate": 2.2572910388555615e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10950 + }, + { + "epoch": 3.188827465813209, + "flow_matching_loss": 0.01482566650956869, + "grad_norm": 0.064992256462574, + "learning_rate": 2.255863302527758e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10960 + }, + { + "epoch": 3.1917369799243525, + "flow_matching_loss": 0.017122190352529286, + "grad_norm": 0.07450073957443237, + "learning_rate": 2.2544346477547685e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10970 + }, + { + "epoch": 3.194646494035496, + "flow_matching_loss": 0.013338679913431406, + "grad_norm": 0.07609791308641434, + "learning_rate": 2.253005076272544e-05, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10980 + }, + { + "epoch": 3.1975560081466394, + "flow_matching_loss": 0.017076717037707568, + "grad_norm": 0.09863721579313278, + "learning_rate": 2.2515745898181477e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 10990 + }, + { + "epoch": 3.200465522257783, + "flow_matching_loss": 0.016562686953693628, + "grad_norm": 0.08072847872972488, + "learning_rate": 2.250143190129756e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11000 + }, + { + "epoch": 3.2033750363689264, + "flow_matching_loss": 0.015399628784507513, + "grad_norm": 0.10498252511024475, + "learning_rate": 2.2487108789466547e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11010 + }, + { + "epoch": 3.2062845504800697, + "flow_matching_loss": 0.015554165095090866, + "grad_norm": 0.0737716555595398, + "learning_rate": 2.247277658009236e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11020 + }, + { + "epoch": 3.2091940645912134, + "flow_matching_loss": 0.014117324445396661, + "grad_norm": 0.07801555842161179, + "learning_rate": 2.245843529059e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11030 + }, + { + "epoch": 3.2121035787023566, + "flow_matching_loss": 0.01441505951806903, + "grad_norm": 0.0937659963965416, + "learning_rate": 2.244408493838547e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11040 + }, + { + "epoch": 3.2150130928135003, + "flow_matching_loss": 0.014626496750861407, + "grad_norm": 0.06170578673481941, + "learning_rate": 2.242972554091581e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11050 + }, + { + "epoch": 3.2179226069246436, + "flow_matching_loss": 0.014931832719594241, + "grad_norm": 0.07572762668132782, + "learning_rate": 2.2415357115629045e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11060 + }, + { + "epoch": 3.220832121035787, + "flow_matching_loss": 0.015417482610791922, + "grad_norm": 0.07334236055612564, + "learning_rate": 2.2400979679984162e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11070 + }, + { + "epoch": 3.2237416351469306, + "flow_matching_loss": 0.01797833852469921, + "grad_norm": 0.10359115898609161, + "learning_rate": 2.2386593251451096e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11080 + }, + { + "epoch": 3.226651149258074, + "flow_matching_loss": 0.015619401074945926, + "grad_norm": 0.08786690980195999, + "learning_rate": 2.2372197847510722e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11090 + }, + { + "epoch": 3.2295606633692175, + "flow_matching_loss": 0.014547743089497089, + "grad_norm": 0.09775540977716446, + "learning_rate": 2.2357793485654813e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11100 + }, + { + "epoch": 3.232470177480361, + "flow_matching_loss": 0.015903067588806153, + "grad_norm": 0.061824627220630646, + "learning_rate": 2.2343380183386017e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11110 + }, + { + "epoch": 3.235379691591504, + "flow_matching_loss": 0.014599829260259867, + "grad_norm": 0.06892683357000351, + "learning_rate": 2.232895795821787e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11120 + }, + { + "epoch": 3.2382892057026478, + "flow_matching_loss": 0.014280756842345, + "grad_norm": 0.08416149020195007, + "learning_rate": 2.2314526827674723e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11130 + }, + { + "epoch": 3.241198719813791, + "flow_matching_loss": 0.0155974212102592, + "grad_norm": 0.0943167433142662, + "learning_rate": 2.230008680929177e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11140 + }, + { + "epoch": 3.2441082339249343, + "flow_matching_loss": 0.016756225004792215, + "grad_norm": 0.0660262480378151, + "learning_rate": 2.2285637920614987e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11150 + }, + { + "epoch": 3.247017748036078, + "flow_matching_loss": 0.014585093501955271, + "grad_norm": 0.07345934957265854, + "learning_rate": 2.2271180179201143e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11160 + }, + { + "epoch": 3.2499272621472213, + "flow_matching_loss": 0.01695232056081295, + "grad_norm": 0.08071698993444443, + "learning_rate": 2.225671360261775e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11170 + }, + { + "epoch": 3.252836776258365, + "flow_matching_loss": 0.015945157408714293, + "grad_norm": 0.06679379940032959, + "learning_rate": 2.2242238208443068e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11180 + }, + { + "epoch": 3.2557462903695082, + "flow_matching_loss": 0.015291140601038932, + "grad_norm": 0.07699058204889297, + "learning_rate": 2.2227754014266063e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11190 + }, + { + "epoch": 3.258655804480652, + "flow_matching_loss": 0.016107641626149416, + "grad_norm": 0.07417980581521988, + "learning_rate": 2.22132610376864e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11200 + }, + { + "epoch": 3.261565318591795, + "flow_matching_loss": 0.016982611920684575, + "grad_norm": 0.08416034281253815, + "learning_rate": 2.219875929631441e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11210 + }, + { + "epoch": 3.2644748327029385, + "flow_matching_loss": 0.01496037794277072, + "grad_norm": 0.08660227805376053, + "learning_rate": 2.2184248807771074e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11220 + }, + { + "epoch": 3.267384346814082, + "flow_matching_loss": 0.015199510287493468, + "grad_norm": 0.06262461096048355, + "learning_rate": 2.216972958968801e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11230 + }, + { + "epoch": 3.2702938609252254, + "flow_matching_loss": 0.014879797771573066, + "grad_norm": 0.07628375291824341, + "learning_rate": 2.215520165970743e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11240 + }, + { + "epoch": 3.2732033750363687, + "flow_matching_loss": 0.013696678634732961, + "grad_norm": 0.08823966234922409, + "learning_rate": 2.2140665035482143e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11250 + }, + { + "epoch": 3.2761128891475124, + "flow_matching_loss": 0.01529853567481041, + "grad_norm": 0.07645256817340851, + "learning_rate": 2.2126119734675518e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11260 + }, + { + "epoch": 3.2790224032586557, + "flow_matching_loss": 0.014607799891382455, + "grad_norm": 0.09255778044462204, + "learning_rate": 2.2111565774961463e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11270 + }, + { + "epoch": 3.2819319173697994, + "flow_matching_loss": 0.015212753508239984, + "grad_norm": 0.06922861188650131, + "learning_rate": 2.2097003174024417e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11280 + }, + { + "epoch": 3.2848414314809427, + "flow_matching_loss": 0.01532007846981287, + "grad_norm": 0.07110816985368729, + "learning_rate": 2.2082431949559315e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11290 + }, + { + "epoch": 3.2877509455920864, + "flow_matching_loss": 0.015359319094568492, + "grad_norm": 0.09647352993488312, + "learning_rate": 2.2067852119271558e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11300 + }, + { + "epoch": 3.2906604597032296, + "flow_matching_loss": 0.016029846854507923, + "grad_norm": 0.06821310520172119, + "learning_rate": 2.2053263700877026e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11310 + }, + { + "epoch": 3.293569973814373, + "flow_matching_loss": 0.01601950218901038, + "grad_norm": 0.0653790608048439, + "learning_rate": 2.2038666712102015e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11320 + }, + { + "epoch": 3.2964794879255166, + "flow_matching_loss": 0.01600588848814368, + "grad_norm": 0.07359831780195236, + "learning_rate": 2.2024061170683246e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11330 + }, + { + "epoch": 3.29938900203666, + "flow_matching_loss": 0.016894906386733057, + "grad_norm": 0.08427239954471588, + "learning_rate": 2.200944709436783e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11340 + }, + { + "epoch": 3.302298516147803, + "flow_matching_loss": 0.01615295372903347, + "grad_norm": 0.10443778336048126, + "learning_rate": 2.1994824500913246e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11350 + }, + { + "epoch": 3.305208030258947, + "flow_matching_loss": 0.017092229519039392, + "grad_norm": 0.07021120935678482, + "learning_rate": 2.1980193408087322e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11360 + }, + { + "epoch": 3.30811754437009, + "flow_matching_loss": 0.015939043182879687, + "grad_norm": 0.08641776442527771, + "learning_rate": 2.1965553833668224e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11370 + }, + { + "epoch": 3.311027058481234, + "flow_matching_loss": 0.015527061093598605, + "grad_norm": 0.06871704757213593, + "learning_rate": 2.195090579544441e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11380 + }, + { + "epoch": 3.313936572592377, + "flow_matching_loss": 0.015334117412567138, + "grad_norm": 0.09730838239192963, + "learning_rate": 2.193624931121462e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11390 + }, + { + "epoch": 3.3168460867035203, + "flow_matching_loss": 0.014886845368891954, + "grad_norm": 0.0696071907877922, + "learning_rate": 2.1921584398787872e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11400 + }, + { + "epoch": 3.319755600814664, + "flow_matching_loss": 0.014826918579638004, + "grad_norm": 0.06843945384025574, + "learning_rate": 2.1906911075983418e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11410 + }, + { + "epoch": 3.3226651149258073, + "flow_matching_loss": 0.013193984609097242, + "grad_norm": 0.08574014157056808, + "learning_rate": 2.1892229360630728e-05, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11420 + }, + { + "epoch": 3.325574629036951, + "flow_matching_loss": 0.01494209235534072, + "grad_norm": 0.08976937085390091, + "learning_rate": 2.187753927056947e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11430 + }, + { + "epoch": 3.3284841431480943, + "flow_matching_loss": 0.01659237714484334, + "grad_norm": 0.11180431395769119, + "learning_rate": 2.186284082364949e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11440 + }, + { + "epoch": 3.3313936572592375, + "flow_matching_loss": 0.015749915968626738, + "grad_norm": 0.0947609543800354, + "learning_rate": 2.1848134037730786e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11450 + }, + { + "epoch": 3.3343031713703812, + "flow_matching_loss": 0.015376142505556346, + "grad_norm": 0.09999451041221619, + "learning_rate": 2.183341893068349e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11460 + }, + { + "epoch": 3.3372126854815245, + "flow_matching_loss": 0.015470939502120019, + "grad_norm": 0.05818157643079758, + "learning_rate": 2.1818695520387842e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11470 + }, + { + "epoch": 3.340122199592668, + "flow_matching_loss": 0.013636673521250486, + "grad_norm": 0.09030024707317352, + "learning_rate": 2.1803963824734173e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11480 + }, + { + "epoch": 3.3430317137038115, + "flow_matching_loss": 0.015031356085091829, + "grad_norm": 0.13053344190120697, + "learning_rate": 2.1789223861622887e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11490 + }, + { + "epoch": 3.3459412278149547, + "flow_matching_loss": 0.01568855345249176, + "grad_norm": 0.11050936579704285, + "learning_rate": 2.1774475648964428e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11500 + }, + { + "epoch": 3.3488507419260984, + "flow_matching_loss": 0.015165898017585278, + "grad_norm": 0.11162535846233368, + "learning_rate": 2.1759719204679267e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11510 + }, + { + "epoch": 3.3517602560372417, + "flow_matching_loss": 0.01603771522641182, + "grad_norm": 0.06844819337129593, + "learning_rate": 2.174495454669787e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11520 + }, + { + "epoch": 3.3546697701483854, + "flow_matching_loss": 0.01752981897443533, + "grad_norm": 0.0929417759180069, + "learning_rate": 2.1730181692960692e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11530 + }, + { + "epoch": 3.3575792842595287, + "flow_matching_loss": 0.014415863528847694, + "grad_norm": 0.10741381347179413, + "learning_rate": 2.171540066141814e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11540 + }, + { + "epoch": 3.360488798370672, + "flow_matching_loss": 0.016136183124035596, + "grad_norm": 0.10457614064216614, + "learning_rate": 2.170061147003056e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11550 + }, + { + "epoch": 3.3633983124818156, + "flow_matching_loss": 0.017379524558782576, + "grad_norm": 0.08770859241485596, + "learning_rate": 2.1685814136768225e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11560 + }, + { + "epoch": 3.366307826592959, + "flow_matching_loss": 0.014251954201608897, + "grad_norm": 0.0676586702466011, + "learning_rate": 2.167100867961128e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11570 + }, + { + "epoch": 3.369217340704102, + "flow_matching_loss": 0.01751046357676387, + "grad_norm": 0.07541871070861816, + "learning_rate": 2.1656195116549756e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11580 + }, + { + "epoch": 3.372126854815246, + "flow_matching_loss": 0.015153719484806061, + "grad_norm": 0.07343531399965286, + "learning_rate": 2.1641373465583527e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11590 + }, + { + "epoch": 3.375036368926389, + "flow_matching_loss": 0.015052888542413712, + "grad_norm": 0.07062983512878418, + "learning_rate": 2.1626543744722292e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11600 + }, + { + "epoch": 3.377945883037533, + "flow_matching_loss": 0.014208865258842707, + "grad_norm": 0.07133804261684418, + "learning_rate": 2.161170597198557e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11610 + }, + { + "epoch": 3.380855397148676, + "flow_matching_loss": 0.014730602409690619, + "grad_norm": 0.0738903284072876, + "learning_rate": 2.1596860165402645e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11620 + }, + { + "epoch": 3.38376491125982, + "flow_matching_loss": 0.016088690049946307, + "grad_norm": 0.0890546441078186, + "learning_rate": 2.1582006343012576e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11630 + }, + { + "epoch": 3.386674425370963, + "flow_matching_loss": 0.014838141109794378, + "grad_norm": 0.06722988933324814, + "learning_rate": 2.156714452286416e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11640 + }, + { + "epoch": 3.3895839394821063, + "flow_matching_loss": 0.015889868140220642, + "grad_norm": 0.07635676115751266, + "learning_rate": 2.155227472301591e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11650 + }, + { + "epoch": 3.39249345359325, + "flow_matching_loss": 0.015074862632900476, + "grad_norm": 0.10639965534210205, + "learning_rate": 2.1537396961536027e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11660 + }, + { + "epoch": 3.3954029677043933, + "flow_matching_loss": 0.015277268923819065, + "grad_norm": 0.0752687081694603, + "learning_rate": 2.1522511256502403e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11670 + }, + { + "epoch": 3.3983124818155366, + "flow_matching_loss": 0.015159731544554234, + "grad_norm": 0.08551640063524246, + "learning_rate": 2.1507617626002566e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11680 + }, + { + "epoch": 3.4012219959266803, + "flow_matching_loss": 0.016628528572618963, + "grad_norm": 0.07033184170722961, + "learning_rate": 2.1492716088133685e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11690 + }, + { + "epoch": 3.4041315100378235, + "flow_matching_loss": 0.014843813702464103, + "grad_norm": 0.07217340171337128, + "learning_rate": 2.1477806661002536e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11700 + }, + { + "epoch": 3.4070410241489673, + "flow_matching_loss": 0.016312492452561854, + "grad_norm": 0.07894200086593628, + "learning_rate": 2.146288936272548e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11710 + }, + { + "epoch": 3.4099505382601105, + "flow_matching_loss": 0.01553204944357276, + "grad_norm": 0.09515032172203064, + "learning_rate": 2.1447964211428433e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11720 + }, + { + "epoch": 3.4128600523712542, + "flow_matching_loss": 0.013957070000469685, + "grad_norm": 0.08343199640512466, + "learning_rate": 2.1433031225246865e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11730 + }, + { + "epoch": 3.4157695664823975, + "flow_matching_loss": 0.018415308836847544, + "grad_norm": 0.07973859459161758, + "learning_rate": 2.1418090422325766e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11740 + }, + { + "epoch": 3.4186790805935408, + "flow_matching_loss": 0.0146258813329041, + "grad_norm": 0.0724276602268219, + "learning_rate": 2.1403141820819623e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11750 + }, + { + "epoch": 3.4215885947046845, + "flow_matching_loss": 0.015520735085010529, + "grad_norm": 0.0834338590502739, + "learning_rate": 2.1388185438892377e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11760 + }, + { + "epoch": 3.4244981088158277, + "flow_matching_loss": 0.015677267871797084, + "grad_norm": 0.07127588987350464, + "learning_rate": 2.1373221294717465e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11770 + }, + { + "epoch": 3.427407622926971, + "flow_matching_loss": 0.01666880175471306, + "grad_norm": 0.07928606122732162, + "learning_rate": 2.1358249406477724e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11780 + }, + { + "epoch": 3.4303171370381147, + "flow_matching_loss": 0.015719879325479268, + "grad_norm": 0.06755295395851135, + "learning_rate": 2.1343269792365408e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11790 + }, + { + "epoch": 3.433226651149258, + "flow_matching_loss": 0.016388945654034613, + "grad_norm": 0.09301071614027023, + "learning_rate": 2.132828247058217e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11800 + }, + { + "epoch": 3.4361361652604017, + "flow_matching_loss": 0.01569692250341177, + "grad_norm": 0.085683174431324, + "learning_rate": 2.1313287459339004e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11810 + }, + { + "epoch": 3.439045679371545, + "flow_matching_loss": 0.01459726169705391, + "grad_norm": 0.08098357915878296, + "learning_rate": 2.1298284776856276e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11820 + }, + { + "epoch": 3.4419551934826886, + "flow_matching_loss": 0.015756877791136503, + "grad_norm": 0.08477999269962311, + "learning_rate": 2.128327444136365e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11830 + }, + { + "epoch": 3.444864707593832, + "flow_matching_loss": 0.015641486924141645, + "grad_norm": 0.07628018409013748, + "learning_rate": 2.1268256471100105e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11840 + }, + { + "epoch": 3.447774221704975, + "flow_matching_loss": 0.014458597265183925, + "grad_norm": 0.061489708721637726, + "learning_rate": 2.1253230884313893e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11850 + }, + { + "epoch": 3.450683735816119, + "flow_matching_loss": 0.016034630592912436, + "grad_norm": 0.08221996575593948, + "learning_rate": 2.123819769926251e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11860 + }, + { + "epoch": 3.453593249927262, + "flow_matching_loss": 0.01497499030083418, + "grad_norm": 0.08176316320896149, + "learning_rate": 2.1223156934212702e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11870 + }, + { + "epoch": 3.4565027640384054, + "flow_matching_loss": 0.0155030257999897, + "grad_norm": 0.07133590430021286, + "learning_rate": 2.1208108607440418e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11880 + }, + { + "epoch": 3.459412278149549, + "flow_matching_loss": 0.015317054651677608, + "grad_norm": 0.06520935893058777, + "learning_rate": 2.119305273723078e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11890 + }, + { + "epoch": 3.4623217922606924, + "flow_matching_loss": 0.01432994194328785, + "grad_norm": 0.0804227665066719, + "learning_rate": 2.1177989341878106e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11900 + }, + { + "epoch": 3.465231306371836, + "flow_matching_loss": 0.01584387058392167, + "grad_norm": 0.06417582929134369, + "learning_rate": 2.116291843968583e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11910 + }, + { + "epoch": 3.4681408204829793, + "flow_matching_loss": 0.01601812206208706, + "grad_norm": 0.09675505012273788, + "learning_rate": 2.114784004896653e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11920 + }, + { + "epoch": 3.4710503345941226, + "flow_matching_loss": 0.01629976453259587, + "grad_norm": 0.06801251322031021, + "learning_rate": 2.113275418804187e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11930 + }, + { + "epoch": 3.4739598487052663, + "flow_matching_loss": 0.014379793219268322, + "grad_norm": 0.08308740705251694, + "learning_rate": 2.1117660875242583e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11940 + }, + { + "epoch": 3.4768693628164096, + "flow_matching_loss": 0.015599736105650663, + "grad_norm": 0.09500250220298767, + "learning_rate": 2.1102560128908478e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11950 + }, + { + "epoch": 3.4797788769275533, + "flow_matching_loss": 0.014109668787568808, + "grad_norm": 0.07440651953220367, + "learning_rate": 2.1087451967388384e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11960 + }, + { + "epoch": 3.4826883910386965, + "flow_matching_loss": 0.01485594203695655, + "grad_norm": 0.0827968567609787, + "learning_rate": 2.1072336409040135e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11970 + }, + { + "epoch": 3.48559790514984, + "flow_matching_loss": 0.016974958404898644, + "grad_norm": 0.07440180331468582, + "learning_rate": 2.1057213472230573e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11980 + }, + { + "epoch": 3.4885074192609835, + "flow_matching_loss": 0.017234695609658957, + "grad_norm": 0.09330865740776062, + "learning_rate": 2.1042083175335478e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 11990 + }, + { + "epoch": 3.4914169333721268, + "flow_matching_loss": 0.014584231749176979, + "grad_norm": 0.08215612173080444, + "learning_rate": 2.1026945536739595e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12000 + }, + { + "epoch": 3.4943264474832705, + "flow_matching_loss": 0.01606825226917863, + "grad_norm": 0.07790908962488174, + "learning_rate": 2.1011800574836586e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12010 + }, + { + "epoch": 3.4972359615944137, + "flow_matching_loss": 0.014840306714177132, + "grad_norm": 0.05681914463639259, + "learning_rate": 2.0996648308028992e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12020 + }, + { + "epoch": 3.500145475705557, + "flow_matching_loss": 0.014231746643781662, + "grad_norm": 0.11534086614847183, + "learning_rate": 2.0981488754728266e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12030 + }, + { + "epoch": 3.5030549898167007, + "flow_matching_loss": 0.015246598236262798, + "grad_norm": 0.07636135816574097, + "learning_rate": 2.0966321933354685e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12040 + }, + { + "epoch": 3.505964503927844, + "flow_matching_loss": 0.014846167713403701, + "grad_norm": 0.07598406821489334, + "learning_rate": 2.095114786233736e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12050 + }, + { + "epoch": 3.5088740180389877, + "flow_matching_loss": 0.015623101498931646, + "grad_norm": 0.07993654161691666, + "learning_rate": 2.0935966560114236e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12060 + }, + { + "epoch": 3.511783532150131, + "flow_matching_loss": 0.014811740815639496, + "grad_norm": 0.09625408053398132, + "learning_rate": 2.092077804513201e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12070 + }, + { + "epoch": 3.514693046261274, + "flow_matching_loss": 0.015382303670048714, + "grad_norm": 0.0879182294011116, + "learning_rate": 2.0905582335846167e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12080 + }, + { + "epoch": 3.517602560372418, + "flow_matching_loss": 0.015659973304718734, + "grad_norm": 0.08180984854698181, + "learning_rate": 2.0890379450720927e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12090 + }, + { + "epoch": 3.520512074483561, + "flow_matching_loss": 0.015476400032639503, + "grad_norm": 0.085471510887146, + "learning_rate": 2.087516940822923e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12100 + }, + { + "epoch": 3.5234215885947044, + "flow_matching_loss": 0.015680381935089825, + "grad_norm": 0.07904262095689774, + "learning_rate": 2.085995222685271e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12110 + }, + { + "epoch": 3.526331102705848, + "flow_matching_loss": 0.01445759292691946, + "grad_norm": 0.06339608877897263, + "learning_rate": 2.084472792508167e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12120 + }, + { + "epoch": 3.5292406168169914, + "flow_matching_loss": 0.014933250844478607, + "grad_norm": 0.07699710875749588, + "learning_rate": 2.082949652141508e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12130 + }, + { + "epoch": 3.532150130928135, + "flow_matching_loss": 0.014755536336451769, + "grad_norm": 0.07096560299396515, + "learning_rate": 2.081425803436054e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12140 + }, + { + "epoch": 3.5350596450392784, + "flow_matching_loss": 0.015896176639944314, + "grad_norm": 0.09391763061285019, + "learning_rate": 2.079901248243423e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12150 + }, + { + "epoch": 3.537969159150422, + "flow_matching_loss": 0.016461908631026746, + "grad_norm": 0.06897689402103424, + "learning_rate": 2.078375988416095e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12160 + }, + { + "epoch": 3.5408786732615654, + "flow_matching_loss": 0.01433317419141531, + "grad_norm": 0.07040475308895111, + "learning_rate": 2.076850025807404e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12170 + }, + { + "epoch": 3.5437881873727086, + "flow_matching_loss": 0.016563307587057353, + "grad_norm": 0.06450559198856354, + "learning_rate": 2.0753233622715374e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12180 + }, + { + "epoch": 3.5466977014838523, + "flow_matching_loss": 0.01387160662561655, + "grad_norm": 0.05456933379173279, + "learning_rate": 2.073795999663536e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12190 + }, + { + "epoch": 3.5496072155949956, + "flow_matching_loss": 0.015134937688708305, + "grad_norm": 0.062114790081977844, + "learning_rate": 2.07226793983929e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12200 + }, + { + "epoch": 3.552516729706139, + "flow_matching_loss": 0.017116803489625455, + "grad_norm": 0.07330848276615143, + "learning_rate": 2.070739184655536e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12210 + }, + { + "epoch": 3.5554262438172826, + "flow_matching_loss": 0.016279212012887, + "grad_norm": 0.08307880908250809, + "learning_rate": 2.069209735969855e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12220 + }, + { + "epoch": 3.558335757928426, + "flow_matching_loss": 0.016737996879965068, + "grad_norm": 0.10090626776218414, + "learning_rate": 2.067679595640672e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12230 + }, + { + "epoch": 3.5612452720395695, + "flow_matching_loss": 0.014578737970441579, + "grad_norm": 0.0924069732427597, + "learning_rate": 2.066148765527252e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12240 + }, + { + "epoch": 3.564154786150713, + "flow_matching_loss": 0.014851547870784998, + "grad_norm": 0.09856583178043365, + "learning_rate": 2.0646172474896976e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12250 + }, + { + "epoch": 3.5670643002618565, + "flow_matching_loss": 0.014652405120432376, + "grad_norm": 0.08253976702690125, + "learning_rate": 2.063085043388947e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12260 + }, + { + "epoch": 3.5699738143729998, + "flow_matching_loss": 0.01444775452837348, + "grad_norm": 0.08341598510742188, + "learning_rate": 2.0615521550867732e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12270 + }, + { + "epoch": 3.572883328484143, + "flow_matching_loss": 0.014345444086939097, + "grad_norm": 0.07259522378444672, + "learning_rate": 2.0600185844457803e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12280 + }, + { + "epoch": 3.5757928425952867, + "flow_matching_loss": 0.014955624379217625, + "grad_norm": 0.06408794969320297, + "learning_rate": 2.058484333329401e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12290 + }, + { + "epoch": 3.57870235670643, + "flow_matching_loss": 0.015570421051234008, + "grad_norm": 0.07086075842380524, + "learning_rate": 2.0569494036018955e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12300 + }, + { + "epoch": 3.5816118708175733, + "flow_matching_loss": 0.014438960142433644, + "grad_norm": 0.09264744818210602, + "learning_rate": 2.055413797128347e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12310 + }, + { + "epoch": 3.584521384928717, + "flow_matching_loss": 0.014104956574738026, + "grad_norm": 0.07077410817146301, + "learning_rate": 2.0538775157746634e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12320 + }, + { + "epoch": 3.5874308990398602, + "flow_matching_loss": 0.013657112792134284, + "grad_norm": 0.08152949810028076, + "learning_rate": 2.0523405614075706e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12330 + }, + { + "epoch": 3.590340413151004, + "flow_matching_loss": 0.014723737724125385, + "grad_norm": 0.05459417402744293, + "learning_rate": 2.0508029358946132e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12340 + }, + { + "epoch": 3.593249927262147, + "flow_matching_loss": 0.014982997626066207, + "grad_norm": 0.06143946945667267, + "learning_rate": 2.0492646411041515e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12350 + }, + { + "epoch": 3.596159441373291, + "flow_matching_loss": 0.01595185389742255, + "grad_norm": 0.10025094449520111, + "learning_rate": 2.0477256789053587e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12360 + }, + { + "epoch": 3.599068955484434, + "flow_matching_loss": 0.014266297779977321, + "grad_norm": 0.08835582435131073, + "learning_rate": 2.0461860511682194e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12370 + }, + { + "epoch": 3.6019784695955774, + "flow_matching_loss": 0.014878789894282818, + "grad_norm": 0.09285423159599304, + "learning_rate": 2.0446457597635254e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12380 + }, + { + "epoch": 3.604887983706721, + "flow_matching_loss": 0.015094347018748521, + "grad_norm": 0.08217331022024155, + "learning_rate": 2.0431048065628763e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12390 + }, + { + "epoch": 3.6077974978178644, + "flow_matching_loss": 0.015729272831231357, + "grad_norm": 0.0708748996257782, + "learning_rate": 2.0415631934386764e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12400 + }, + { + "epoch": 3.6107070119290077, + "flow_matching_loss": 0.01575240436941385, + "grad_norm": 0.07710058987140656, + "learning_rate": 2.04002092226413e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12410 + }, + { + "epoch": 3.6136165260401514, + "flow_matching_loss": 0.014913765713572502, + "grad_norm": 0.07109581679105759, + "learning_rate": 2.038477994913242e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12420 + }, + { + "epoch": 3.6165260401512946, + "flow_matching_loss": 0.016417440492659807, + "grad_norm": 0.07884085178375244, + "learning_rate": 2.036934413260816e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12430 + }, + { + "epoch": 3.619435554262438, + "flow_matching_loss": 0.01597119215875864, + "grad_norm": 0.09533672779798508, + "learning_rate": 2.0353901791824473e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12440 + }, + { + "epoch": 3.6223450683735816, + "flow_matching_loss": 0.014754175208508968, + "grad_norm": 0.0970696434378624, + "learning_rate": 2.0338452945545275e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12450 + }, + { + "epoch": 3.6252545824847253, + "flow_matching_loss": 0.015970749128609896, + "grad_norm": 0.07719206809997559, + "learning_rate": 2.0322997612542366e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12460 + }, + { + "epoch": 3.6281640965958686, + "flow_matching_loss": 0.01575769213959575, + "grad_norm": 0.09329578280448914, + "learning_rate": 2.0307535811595427e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12470 + }, + { + "epoch": 3.631073610707012, + "flow_matching_loss": 0.015780258271843196, + "grad_norm": 0.0748058333992958, + "learning_rate": 2.0292067561492013e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12480 + }, + { + "epoch": 3.6339831248181556, + "flow_matching_loss": 0.016214762441813947, + "grad_norm": 0.09164014458656311, + "learning_rate": 2.02765928810275e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12490 + }, + { + "epoch": 3.636892638929299, + "flow_matching_loss": 0.015178795717656613, + "grad_norm": 0.07162585109472275, + "learning_rate": 2.0261111789005087e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12500 + }, + { + "epoch": 3.639802153040442, + "flow_matching_loss": 0.01577666401863098, + "grad_norm": 0.08761367201805115, + "learning_rate": 2.024562430423576e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12510 + }, + { + "epoch": 3.642711667151586, + "flow_matching_loss": 0.015112879686057568, + "grad_norm": 0.06986001878976822, + "learning_rate": 2.0230130445538273e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12520 + }, + { + "epoch": 3.645621181262729, + "flow_matching_loss": 0.01534789465367794, + "grad_norm": 0.07255560159683228, + "learning_rate": 2.0214630231739127e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12530 + }, + { + "epoch": 3.6485306953738723, + "flow_matching_loss": 0.015469399094581605, + "grad_norm": 0.0966716930270195, + "learning_rate": 2.0199123681672543e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12540 + }, + { + "epoch": 3.651440209485016, + "flow_matching_loss": 0.014341234602034092, + "grad_norm": 0.07361998409032822, + "learning_rate": 2.0183610814180437e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12550 + }, + { + "epoch": 3.6543497235961593, + "flow_matching_loss": 0.016720857936888933, + "grad_norm": 0.07613180577754974, + "learning_rate": 2.016809164811241e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12560 + }, + { + "epoch": 3.657259237707303, + "flow_matching_loss": 0.015492786001414061, + "grad_norm": 0.07348378747701645, + "learning_rate": 2.015256620232571e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12570 + }, + { + "epoch": 3.6601687518184463, + "flow_matching_loss": 0.015598589554429054, + "grad_norm": 0.08892955631017685, + "learning_rate": 2.0137034495685222e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12580 + }, + { + "epoch": 3.66307826592959, + "flow_matching_loss": 0.01454709293320775, + "grad_norm": 0.06663551181554794, + "learning_rate": 2.012149654706343e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12590 + }, + { + "epoch": 3.6659877800407332, + "flow_matching_loss": 0.016047302167862654, + "grad_norm": 0.08209995925426483, + "learning_rate": 2.010595237534041e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12600 + }, + { + "epoch": 3.6688972941518765, + "flow_matching_loss": 0.015839790645986795, + "grad_norm": 0.08488032966852188, + "learning_rate": 2.0090401999403793e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12610 + }, + { + "epoch": 3.67180680826302, + "flow_matching_loss": 0.015500704944133758, + "grad_norm": 0.08656375110149384, + "learning_rate": 2.007484543814875e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12620 + }, + { + "epoch": 3.6747163223741635, + "flow_matching_loss": 0.01557289930060506, + "grad_norm": 0.08931925147771835, + "learning_rate": 2.0059282710477972e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12630 + }, + { + "epoch": 3.6776258364853067, + "flow_matching_loss": 0.01441049836575985, + "grad_norm": 0.06347820907831192, + "learning_rate": 2.0043713835301647e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12640 + }, + { + "epoch": 3.6805353505964504, + "flow_matching_loss": 0.01488736756145954, + "grad_norm": 0.07370192557573318, + "learning_rate": 2.0028138831537417e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12650 + }, + { + "epoch": 3.6834448647075937, + "flow_matching_loss": 0.014337416645139456, + "grad_norm": 0.08057195693254471, + "learning_rate": 2.0012557718110387e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12660 + }, + { + "epoch": 3.6863543788187374, + "flow_matching_loss": 0.015252651926130057, + "grad_norm": 0.07211877405643463, + "learning_rate": 1.9996970513953077e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12670 + }, + { + "epoch": 3.6892638929298807, + "flow_matching_loss": 0.014883488696068526, + "grad_norm": 0.07587935030460358, + "learning_rate": 1.9981377238005405e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12680 + }, + { + "epoch": 3.6921734070410244, + "flow_matching_loss": 0.014429643470793962, + "grad_norm": 0.08962063491344452, + "learning_rate": 1.9965777909214674e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12690 + }, + { + "epoch": 3.6950829211521676, + "flow_matching_loss": 0.013441132102161645, + "grad_norm": 0.07572191208600998, + "learning_rate": 1.9950172546535545e-05, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12700 + }, + { + "epoch": 3.697992435263311, + "flow_matching_loss": 0.01582878902554512, + "grad_norm": 0.07448488473892212, + "learning_rate": 1.9934561168930002e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12710 + }, + { + "epoch": 3.7009019493744546, + "flow_matching_loss": 0.014469576254487037, + "grad_norm": 0.10022494941949844, + "learning_rate": 1.991894379536734e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12720 + }, + { + "epoch": 3.703811463485598, + "flow_matching_loss": 0.014997255895286799, + "grad_norm": 0.09445875883102417, + "learning_rate": 1.9903320444824137e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12730 + }, + { + "epoch": 3.706720977596741, + "flow_matching_loss": 0.015600121766328811, + "grad_norm": 0.07282574474811554, + "learning_rate": 1.9887691136284245e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12740 + }, + { + "epoch": 3.709630491707885, + "flow_matching_loss": 0.016763430461287497, + "grad_norm": 0.09825586527585983, + "learning_rate": 1.9872055888738745e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12750 + }, + { + "epoch": 3.712540005819028, + "flow_matching_loss": 0.014882063400000334, + "grad_norm": 0.06771338731050491, + "learning_rate": 1.985641472118593e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12760 + }, + { + "epoch": 3.715449519930172, + "flow_matching_loss": 0.015473625343292951, + "grad_norm": 0.0662817433476448, + "learning_rate": 1.9840767652631294e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12770 + }, + { + "epoch": 3.718359034041315, + "flow_matching_loss": 0.015894069988280534, + "grad_norm": 0.061968859285116196, + "learning_rate": 1.9825114702087513e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12780 + }, + { + "epoch": 3.7212685481524588, + "flow_matching_loss": 0.014881713222712279, + "grad_norm": 0.07768125087022781, + "learning_rate": 1.9809455888574388e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12790 + }, + { + "epoch": 3.724178062263602, + "flow_matching_loss": 0.016949024703353643, + "grad_norm": 0.09999608993530273, + "learning_rate": 1.9793791231118864e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12800 + }, + { + "epoch": 3.7270875763747453, + "flow_matching_loss": 0.015344189573079347, + "grad_norm": 0.08304872363805771, + "learning_rate": 1.9778120748754958e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12810 + }, + { + "epoch": 3.729997090485889, + "flow_matching_loss": 0.015131905023008586, + "grad_norm": 0.10192036628723145, + "learning_rate": 1.97624444605238e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12820 + }, + { + "epoch": 3.7329066045970323, + "flow_matching_loss": 0.01452829958871007, + "grad_norm": 0.06647082418203354, + "learning_rate": 1.9746762385473552e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12830 + }, + { + "epoch": 3.7358161187081755, + "flow_matching_loss": 0.015444056503474713, + "grad_norm": 0.09458453208208084, + "learning_rate": 1.9731074542659414e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12840 + }, + { + "epoch": 3.7387256328193192, + "flow_matching_loss": 0.016327164974063633, + "grad_norm": 0.10168302804231644, + "learning_rate": 1.9715380951143603e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12850 + }, + { + "epoch": 3.7416351469304625, + "flow_matching_loss": 0.015480757225304842, + "grad_norm": 0.10346796363592148, + "learning_rate": 1.9699681629995295e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12860 + }, + { + "epoch": 3.744544661041606, + "flow_matching_loss": 0.014520786050707103, + "grad_norm": 0.08600758016109467, + "learning_rate": 1.968397659829066e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12870 + }, + { + "epoch": 3.7474541751527495, + "flow_matching_loss": 0.0139972859993577, + "grad_norm": 0.0809231773018837, + "learning_rate": 1.966826587511279e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12880 + }, + { + "epoch": 3.750363689263893, + "flow_matching_loss": 0.014974929951131343, + "grad_norm": 0.07978958636522293, + "learning_rate": 1.965254947955169e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12890 + }, + { + "epoch": 3.7532732033750364, + "flow_matching_loss": 0.016817395016551016, + "grad_norm": 0.059731822460889816, + "learning_rate": 1.963682743070427e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12900 + }, + { + "epoch": 3.7561827174861797, + "flow_matching_loss": 0.014251423068344593, + "grad_norm": 0.0708240196108818, + "learning_rate": 1.962109974767429e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12910 + }, + { + "epoch": 3.7590922315973234, + "flow_matching_loss": 0.014946888294070959, + "grad_norm": 0.09659311175346375, + "learning_rate": 1.960536644957238e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12920 + }, + { + "epoch": 3.7620017457084667, + "flow_matching_loss": 0.01531663304194808, + "grad_norm": 0.08539491891860962, + "learning_rate": 1.9589627555515974e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12930 + }, + { + "epoch": 3.76491125981961, + "flow_matching_loss": 0.015591784007847309, + "grad_norm": 0.07342422753572464, + "learning_rate": 1.9573883084629317e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12940 + }, + { + "epoch": 3.7678207739307537, + "flow_matching_loss": 0.015442391578108072, + "grad_norm": 0.06626555323600769, + "learning_rate": 1.9558133056043422e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12950 + }, + { + "epoch": 3.770730288041897, + "flow_matching_loss": 0.014002919103950262, + "grad_norm": 0.06999494880437851, + "learning_rate": 1.9542377488896062e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12960 + }, + { + "epoch": 3.77363980215304, + "flow_matching_loss": 0.014945900067687035, + "grad_norm": 0.05268583074212074, + "learning_rate": 1.9526616402331733e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12970 + }, + { + "epoch": 3.776549316264184, + "flow_matching_loss": 0.014918960630893707, + "grad_norm": 0.08314146101474762, + "learning_rate": 1.9510849815501638e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12980 + }, + { + "epoch": 3.7794588303753276, + "flow_matching_loss": 0.01502603879198432, + "grad_norm": 0.09538402408361435, + "learning_rate": 1.949507774756368e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 12990 + }, + { + "epoch": 3.782368344486471, + "flow_matching_loss": 0.015883041266351938, + "grad_norm": 0.07004507631063461, + "learning_rate": 1.9479300217682394e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13000 + }, + { + "epoch": 3.785277858597614, + "flow_matching_loss": 0.015340164862573146, + "grad_norm": 0.07908450067043304, + "learning_rate": 1.946351724502898e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13010 + }, + { + "epoch": 3.788187372708758, + "flow_matching_loss": 0.014865301176905632, + "grad_norm": 0.06623215973377228, + "learning_rate": 1.9447728848781227e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13020 + }, + { + "epoch": 3.791096886819901, + "flow_matching_loss": 0.01474988404661417, + "grad_norm": 0.08199667185544968, + "learning_rate": 1.943193504812353e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13030 + }, + { + "epoch": 3.7940064009310444, + "flow_matching_loss": 0.014598784409463406, + "grad_norm": 0.06327107548713684, + "learning_rate": 1.9416135862246846e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13040 + }, + { + "epoch": 3.796915915042188, + "flow_matching_loss": 0.013952832203358412, + "grad_norm": 0.07784723490476608, + "learning_rate": 1.9400331310348673e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13050 + }, + { + "epoch": 3.7998254291533313, + "flow_matching_loss": 0.01637130118906498, + "grad_norm": 0.11010678857564926, + "learning_rate": 1.9384521411633033e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13060 + }, + { + "epoch": 3.8027349432644746, + "flow_matching_loss": 0.015689415950328113, + "grad_norm": 0.08306363224983215, + "learning_rate": 1.936870618531045e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13070 + }, + { + "epoch": 3.8056444573756183, + "flow_matching_loss": 0.015155722852796316, + "grad_norm": 0.08190848678350449, + "learning_rate": 1.9352885650597904e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13080 + }, + { + "epoch": 3.8085539714867616, + "flow_matching_loss": 0.014823535922914743, + "grad_norm": 0.10564404726028442, + "learning_rate": 1.9337059826718847e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13090 + }, + { + "epoch": 3.8114634855979053, + "flow_matching_loss": 0.013628110382705927, + "grad_norm": 0.08159254491329193, + "learning_rate": 1.932122873290314e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13100 + }, + { + "epoch": 3.8143729997090485, + "flow_matching_loss": 0.01560172438621521, + "grad_norm": 0.06844203174114227, + "learning_rate": 1.9305392388387054e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13110 + }, + { + "epoch": 3.8172825138201922, + "flow_matching_loss": 0.015737169701606036, + "grad_norm": 0.0820152759552002, + "learning_rate": 1.928955081241325e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13120 + }, + { + "epoch": 3.8201920279313355, + "flow_matching_loss": 0.014866608940064908, + "grad_norm": 0.11021903902292252, + "learning_rate": 1.9273704024230725e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13130 + }, + { + "epoch": 3.8231015420424788, + "flow_matching_loss": 0.01565711610019207, + "grad_norm": 0.0738174244761467, + "learning_rate": 1.925785204309483e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13140 + }, + { + "epoch": 3.8260110561536225, + "flow_matching_loss": 0.014496132358908653, + "grad_norm": 0.07491621375083923, + "learning_rate": 1.9241994888267212e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13150 + }, + { + "epoch": 3.8289205702647657, + "flow_matching_loss": 0.01579592386260629, + "grad_norm": 0.0811389908194542, + "learning_rate": 1.9226132579015815e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13160 + }, + { + "epoch": 3.831830084375909, + "flow_matching_loss": 0.017404117342084646, + "grad_norm": 0.07972889393568039, + "learning_rate": 1.9210265134614832e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13170 + }, + { + "epoch": 3.8347395984870527, + "flow_matching_loss": 0.015355046186596156, + "grad_norm": 0.07584629207849503, + "learning_rate": 1.9194392574344712e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13180 + }, + { + "epoch": 3.837649112598196, + "flow_matching_loss": 0.014824478514492511, + "grad_norm": 0.06667227298021317, + "learning_rate": 1.9178514917492103e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13190 + }, + { + "epoch": 3.8405586267093397, + "flow_matching_loss": 0.016842051409184933, + "grad_norm": 0.06817618757486343, + "learning_rate": 1.9162632183349866e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13200 + }, + { + "epoch": 3.843468140820483, + "flow_matching_loss": 0.01513356128707528, + "grad_norm": 0.07528634369373322, + "learning_rate": 1.914674439121701e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13210 + }, + { + "epoch": 3.8463776549316266, + "flow_matching_loss": 0.01531507158651948, + "grad_norm": 0.08590903878211975, + "learning_rate": 1.9130851560398715e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13220 + }, + { + "epoch": 3.84928716904277, + "flow_matching_loss": 0.013833930902183055, + "grad_norm": 0.05860443413257599, + "learning_rate": 1.911495371020625e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13230 + }, + { + "epoch": 3.852196683153913, + "flow_matching_loss": 0.015372515469789506, + "grad_norm": 0.1030806228518486, + "learning_rate": 1.909905085995702e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13240 + }, + { + "epoch": 3.855106197265057, + "flow_matching_loss": 0.015737086720764636, + "grad_norm": 0.08523275703191757, + "learning_rate": 1.908314302897448e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13250 + }, + { + "epoch": 3.8580157113762, + "flow_matching_loss": 0.016124567296355963, + "grad_norm": 0.0719156339764595, + "learning_rate": 1.9067230236588142e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13260 + }, + { + "epoch": 3.8609252254873434, + "flow_matching_loss": 0.014044593367725611, + "grad_norm": 0.10529834032058716, + "learning_rate": 1.905131250213355e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13270 + }, + { + "epoch": 3.863834739598487, + "flow_matching_loss": 0.014580122940242291, + "grad_norm": 0.08640825003385544, + "learning_rate": 1.9035389844952256e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13280 + }, + { + "epoch": 3.8667442537096304, + "flow_matching_loss": 0.01789060728624463, + "grad_norm": 0.08943463861942291, + "learning_rate": 1.901946228439179e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13290 + }, + { + "epoch": 3.869653767820774, + "flow_matching_loss": 0.014929715357720853, + "grad_norm": 0.07823865860700607, + "learning_rate": 1.9003529839805637e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13300 + }, + { + "epoch": 3.8725632819319173, + "flow_matching_loss": 0.01413062671199441, + "grad_norm": 0.05171496793627739, + "learning_rate": 1.8987592530553217e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13310 + }, + { + "epoch": 3.875472796043061, + "flow_matching_loss": 0.01737849349156022, + "grad_norm": 0.09016017615795135, + "learning_rate": 1.897165037599987e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13320 + }, + { + "epoch": 3.8783823101542043, + "flow_matching_loss": 0.014546248503029347, + "grad_norm": 0.06757014989852905, + "learning_rate": 1.8955703395516814e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13330 + }, + { + "epoch": 3.8812918242653476, + "flow_matching_loss": 0.015926550049334763, + "grad_norm": 0.08513949066400528, + "learning_rate": 1.8939751608481134e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13340 + }, + { + "epoch": 3.8842013383764913, + "flow_matching_loss": 0.015233403258025646, + "grad_norm": 0.073252372443676, + "learning_rate": 1.892379503427575e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13350 + }, + { + "epoch": 3.8871108524876345, + "flow_matching_loss": 0.01518319258466363, + "grad_norm": 0.07889381051063538, + "learning_rate": 1.8907833692289412e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13360 + }, + { + "epoch": 3.890020366598778, + "flow_matching_loss": 0.01534176552668214, + "grad_norm": 0.06803280115127563, + "learning_rate": 1.889186760191666e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13370 + }, + { + "epoch": 3.8929298807099215, + "flow_matching_loss": 0.015776305738836528, + "grad_norm": 0.0681178942322731, + "learning_rate": 1.887589678255779e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13380 + }, + { + "epoch": 3.895839394821065, + "flow_matching_loss": 0.015209975559264422, + "grad_norm": 0.07514775544404984, + "learning_rate": 1.8859921253618852e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13390 + }, + { + "epoch": 3.898748908932208, + "flow_matching_loss": 0.01608736077323556, + "grad_norm": 0.10330091416835785, + "learning_rate": 1.884394103451163e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13400 + }, + { + "epoch": 3.9016584230433518, + "flow_matching_loss": 0.014346883445978165, + "grad_norm": 0.06634590029716492, + "learning_rate": 1.8827956144653583e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13410 + }, + { + "epoch": 3.9045679371544955, + "flow_matching_loss": 0.014579633716493845, + "grad_norm": 0.07627879828214645, + "learning_rate": 1.8811966603467874e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13420 + }, + { + "epoch": 3.9074774512656387, + "flow_matching_loss": 0.016319816652685403, + "grad_norm": 0.10773918777704239, + "learning_rate": 1.8795972430383293e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13430 + }, + { + "epoch": 3.910386965376782, + "flow_matching_loss": 0.014405209477990866, + "grad_norm": 0.09140139818191528, + "learning_rate": 1.8779973644834273e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13440 + }, + { + "epoch": 3.9132964794879257, + "flow_matching_loss": 0.01390032097697258, + "grad_norm": 0.08395938575267792, + "learning_rate": 1.876397026626085e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13450 + }, + { + "epoch": 3.916205993599069, + "flow_matching_loss": 0.01453215116634965, + "grad_norm": 0.0703798308968544, + "learning_rate": 1.8747962314108633e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13460 + }, + { + "epoch": 3.919115507710212, + "flow_matching_loss": 0.014888463728129863, + "grad_norm": 0.07989762723445892, + "learning_rate": 1.873194980782879e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13470 + }, + { + "epoch": 3.922025021821356, + "flow_matching_loss": 0.014136711228638887, + "grad_norm": 0.06615065783262253, + "learning_rate": 1.871593276687804e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13480 + }, + { + "epoch": 3.924934535932499, + "flow_matching_loss": 0.014244797639548779, + "grad_norm": 0.0751226618885994, + "learning_rate": 1.869991121071858e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13490 + }, + { + "epoch": 3.9278440500436425, + "flow_matching_loss": 0.01558488653972745, + "grad_norm": 0.07030288130044937, + "learning_rate": 1.8683885158818117e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13500 + }, + { + "epoch": 3.930753564154786, + "flow_matching_loss": 0.013695767801254987, + "grad_norm": 0.07937522232532501, + "learning_rate": 1.8667854630649825e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13510 + }, + { + "epoch": 3.9336630782659294, + "flow_matching_loss": 0.015679743234068157, + "grad_norm": 0.0662912130355835, + "learning_rate": 1.86518196456923e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13520 + }, + { + "epoch": 3.936572592377073, + "flow_matching_loss": 0.015433083847165107, + "grad_norm": 0.11082117259502411, + "learning_rate": 1.8635780223429565e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13530 + }, + { + "epoch": 3.9394821064882164, + "flow_matching_loss": 0.016111167799681424, + "grad_norm": 0.080501027405262, + "learning_rate": 1.8619736383351017e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13540 + }, + { + "epoch": 3.94239162059936, + "flow_matching_loss": 0.014421588554978371, + "grad_norm": 0.055343423038721085, + "learning_rate": 1.8603688144951454e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13550 + }, + { + "epoch": 3.9453011347105034, + "flow_matching_loss": 0.015340539254248142, + "grad_norm": 0.08445049077272415, + "learning_rate": 1.8587635527730983e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13560 + }, + { + "epoch": 3.9482106488216466, + "flow_matching_loss": 0.014655168913304806, + "grad_norm": 0.08315341174602509, + "learning_rate": 1.8571578551195056e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13570 + }, + { + "epoch": 3.9511201629327903, + "flow_matching_loss": 0.0132493382319808, + "grad_norm": 0.07376855611801147, + "learning_rate": 1.855551723485441e-05, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13580 + }, + { + "epoch": 3.9540296770439336, + "flow_matching_loss": 0.014801244996488095, + "grad_norm": 0.08220674842596054, + "learning_rate": 1.8539451598225066e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13590 + }, + { + "epoch": 3.956939191155077, + "flow_matching_loss": 0.015522986184805632, + "grad_norm": 0.07899833470582962, + "learning_rate": 1.8523381660828276e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13600 + }, + { + "epoch": 3.9598487052662206, + "flow_matching_loss": 0.015388775616884232, + "grad_norm": 0.09519131481647491, + "learning_rate": 1.850730744219054e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13610 + }, + { + "epoch": 3.962758219377364, + "flow_matching_loss": 0.01417188048362732, + "grad_norm": 0.07668602466583252, + "learning_rate": 1.849122896184354e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13620 + }, + { + "epoch": 3.9656677334885075, + "flow_matching_loss": 0.01639600405469537, + "grad_norm": 0.07650433480739594, + "learning_rate": 1.8475146239324146e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13630 + }, + { + "epoch": 3.968577247599651, + "flow_matching_loss": 0.013482035044580698, + "grad_norm": 0.07764936983585358, + "learning_rate": 1.8459059294174396e-05, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13640 + }, + { + "epoch": 3.9714867617107945, + "flow_matching_loss": 0.015584578178822995, + "grad_norm": 0.07656349241733551, + "learning_rate": 1.844296814594143e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13650 + }, + { + "epoch": 3.9743962758219378, + "flow_matching_loss": 0.01471948130056262, + "grad_norm": 0.10196517407894135, + "learning_rate": 1.8426872814177516e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13660 + }, + { + "epoch": 3.977305789933081, + "flow_matching_loss": 0.014948063902556897, + "grad_norm": 0.07997869700193405, + "learning_rate": 1.8410773318440003e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13670 + }, + { + "epoch": 3.9802153040442247, + "flow_matching_loss": 0.01395276077091694, + "grad_norm": 0.08206649124622345, + "learning_rate": 1.839466967829129e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13680 + }, + { + "epoch": 3.983124818155368, + "flow_matching_loss": 0.014522182196378708, + "grad_norm": 0.07001964747905731, + "learning_rate": 1.8378561913298822e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13690 + }, + { + "epoch": 3.9860343322665113, + "flow_matching_loss": 0.013379440829157829, + "grad_norm": 0.06686311215162277, + "learning_rate": 1.8362450043035053e-05, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13700 + }, + { + "epoch": 3.988943846377655, + "flow_matching_loss": 0.014914518874138593, + "grad_norm": 0.1170818954706192, + "learning_rate": 1.834633408707742e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13710 + }, + { + "epoch": 3.9918533604887982, + "flow_matching_loss": 0.015666478686034678, + "grad_norm": 0.10649386048316956, + "learning_rate": 1.8330214065008336e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13720 + }, + { + "epoch": 3.994762874599942, + "flow_matching_loss": 0.01572956470772624, + "grad_norm": 0.07949764281511307, + "learning_rate": 1.831408999641514e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13730 + }, + { + "epoch": 3.997672388711085, + "flow_matching_loss": 0.015082302689552306, + "grad_norm": 0.09257955104112625, + "learning_rate": 1.8297961900890107e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13740 + }, + { + "epoch": 4.000581902822229, + "flow_matching_loss": 0.013616311084479093, + "grad_norm": 0.09366997331380844, + "learning_rate": 1.8281829798030385e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13750 + }, + { + "epoch": 4.003491416933372, + "flow_matching_loss": 0.014855632465332747, + "grad_norm": 0.09710550308227539, + "learning_rate": 1.8265693707438002e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13760 + }, + { + "epoch": 4.0064009310445154, + "flow_matching_loss": 0.014842930156737566, + "grad_norm": 0.09384413808584213, + "learning_rate": 1.824955364871982e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13770 + }, + { + "epoch": 4.009310445155659, + "flow_matching_loss": 0.014005714654922485, + "grad_norm": 0.07173143327236176, + "learning_rate": 1.8233409641487553e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13780 + }, + { + "epoch": 4.012219959266803, + "flow_matching_loss": 0.015977469459176063, + "grad_norm": 0.07374656945466995, + "learning_rate": 1.8217261705357677e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13790 + }, + { + "epoch": 4.015129473377946, + "flow_matching_loss": 0.01483305199071765, + "grad_norm": 0.11079880595207214, + "learning_rate": 1.8201109859951462e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13800 + }, + { + "epoch": 4.018038987489089, + "flow_matching_loss": 0.013477620761841535, + "grad_norm": 0.1053965613245964, + "learning_rate": 1.818495412489492e-05, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13810 + }, + { + "epoch": 4.020948501600233, + "flow_matching_loss": 0.016141959093511106, + "grad_norm": 0.06615139544010162, + "learning_rate": 1.8168794519818802e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13820 + }, + { + "epoch": 4.023858015711376, + "flow_matching_loss": 0.015326834749430418, + "grad_norm": 0.06525973230600357, + "learning_rate": 1.8152631064358537e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13830 + }, + { + "epoch": 4.02676752982252, + "flow_matching_loss": 0.015027104504406452, + "grad_norm": 0.06568615883588791, + "learning_rate": 1.8136463778154258e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13840 + }, + { + "epoch": 4.029677043933663, + "flow_matching_loss": 0.015257480647414922, + "grad_norm": 0.07240273058414459, + "learning_rate": 1.812029268085074e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13850 + }, + { + "epoch": 4.032586558044806, + "flow_matching_loss": 0.016022813506424427, + "grad_norm": 0.08979135751724243, + "learning_rate": 1.8104117792097387e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13860 + }, + { + "epoch": 4.03549607215595, + "flow_matching_loss": 0.016239756904542445, + "grad_norm": 0.0773392915725708, + "learning_rate": 1.808793913154822e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13870 + }, + { + "epoch": 4.038405586267094, + "flow_matching_loss": 0.015460313204675913, + "grad_norm": 0.08641500771045685, + "learning_rate": 1.8071756718861838e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13880 + }, + { + "epoch": 4.041315100378237, + "flow_matching_loss": 0.015027948655188084, + "grad_norm": 0.09780822694301605, + "learning_rate": 1.805557057370139e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13890 + }, + { + "epoch": 4.04422461448938, + "flow_matching_loss": 0.015985894575715066, + "grad_norm": 0.06996931880712509, + "learning_rate": 1.8039380715734574e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13900 + }, + { + "epoch": 4.047134128600524, + "flow_matching_loss": 0.015214587468653918, + "grad_norm": 0.09220394492149353, + "learning_rate": 1.802318716463359e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13910 + }, + { + "epoch": 4.0500436427116675, + "flow_matching_loss": 0.015239682048559189, + "grad_norm": 0.08730532974004745, + "learning_rate": 1.8006989940075138e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13920 + }, + { + "epoch": 4.05295315682281, + "flow_matching_loss": 0.013948339875787497, + "grad_norm": 0.07310034334659576, + "learning_rate": 1.7990789061740368e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13930 + }, + { + "epoch": 4.055862670933954, + "flow_matching_loss": 0.01551080159842968, + "grad_norm": 0.08298029750585556, + "learning_rate": 1.7974584549314875e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13940 + }, + { + "epoch": 4.058772185045098, + "flow_matching_loss": 0.016402858030050992, + "grad_norm": 0.0760258287191391, + "learning_rate": 1.7958376422488672e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13950 + }, + { + "epoch": 4.0616816991562406, + "flow_matching_loss": 0.01656693695113063, + "grad_norm": 0.07214951515197754, + "learning_rate": 1.7942164700956156e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13960 + }, + { + "epoch": 4.064591213267384, + "flow_matching_loss": 0.01397476801648736, + "grad_norm": 0.08165285736322403, + "learning_rate": 1.79259494044161e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13970 + }, + { + "epoch": 4.067500727378528, + "flow_matching_loss": 0.014646024908870459, + "grad_norm": 0.08159542083740234, + "learning_rate": 1.7909730552571617e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13980 + }, + { + "epoch": 4.070410241489671, + "flow_matching_loss": 0.015780727937817575, + "grad_norm": 0.10157568007707596, + "learning_rate": 1.7893508165130143e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 13990 + }, + { + "epoch": 4.0733197556008145, + "flow_matching_loss": 0.015913321543484925, + "grad_norm": 0.07735645025968552, + "learning_rate": 1.787728226180341e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14000 + }, + { + "epoch": 4.076229269711958, + "flow_matching_loss": 0.015310766361653804, + "grad_norm": 0.07655656337738037, + "learning_rate": 1.786105286230742e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14010 + }, + { + "epoch": 4.079138783823102, + "flow_matching_loss": 0.01502293711528182, + "grad_norm": 0.08246251940727234, + "learning_rate": 1.784481998636242e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14020 + }, + { + "epoch": 4.082048297934245, + "flow_matching_loss": 0.016965329740196468, + "grad_norm": 0.10165663808584213, + "learning_rate": 1.782858365369289e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14030 + }, + { + "epoch": 4.084957812045388, + "flow_matching_loss": 0.01510065170004964, + "grad_norm": 0.11116847395896912, + "learning_rate": 1.78123438840275e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14040 + }, + { + "epoch": 4.087867326156532, + "flow_matching_loss": 0.016220449563115837, + "grad_norm": 0.059428513050079346, + "learning_rate": 1.779610069709911e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14050 + }, + { + "epoch": 4.090776840267675, + "flow_matching_loss": 0.01618818137794733, + "grad_norm": 0.07638953626155853, + "learning_rate": 1.7779854112644713e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14060 + }, + { + "epoch": 4.093686354378819, + "flow_matching_loss": 0.01594740841537714, + "grad_norm": 0.07663699984550476, + "learning_rate": 1.7763604150405446e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14070 + }, + { + "epoch": 4.096595868489962, + "flow_matching_loss": 0.015631217695772647, + "grad_norm": 0.07886145263910294, + "learning_rate": 1.7747350830126548e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14080 + }, + { + "epoch": 4.099505382601105, + "flow_matching_loss": 0.015890952292829753, + "grad_norm": 0.07668419927358627, + "learning_rate": 1.7731094171557332e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14090 + }, + { + "epoch": 4.102414896712249, + "flow_matching_loss": 0.014469109661877155, + "grad_norm": 0.06442062556743622, + "learning_rate": 1.771483419445117e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14100 + }, + { + "epoch": 4.105324410823393, + "flow_matching_loss": 0.012871387787163258, + "grad_norm": 0.06407693773508072, + "learning_rate": 1.7698570918565474e-05, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14110 + }, + { + "epoch": 4.108233924934536, + "flow_matching_loss": 0.013115144986659288, + "grad_norm": 0.06688642501831055, + "learning_rate": 1.768230436366165e-05, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14120 + }, + { + "epoch": 4.111143439045679, + "flow_matching_loss": 0.016165948752313853, + "grad_norm": 0.07322639226913452, + "learning_rate": 1.7666034549505102e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14130 + }, + { + "epoch": 4.114052953156823, + "flow_matching_loss": 0.016100504156202077, + "grad_norm": 0.06608857214450836, + "learning_rate": 1.7649761495865185e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14140 + }, + { + "epoch": 4.1169624672679666, + "flow_matching_loss": 0.01454313714057207, + "grad_norm": 0.08379761129617691, + "learning_rate": 1.7633485222515188e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14150 + }, + { + "epoch": 4.119871981379109, + "flow_matching_loss": 0.01610182672739029, + "grad_norm": 0.07886158674955368, + "learning_rate": 1.761720574923233e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14160 + }, + { + "epoch": 4.122781495490253, + "flow_matching_loss": 0.01586903017014265, + "grad_norm": 0.07624044269323349, + "learning_rate": 1.7600923095797704e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14170 + }, + { + "epoch": 4.125691009601397, + "flow_matching_loss": 0.014557879511266946, + "grad_norm": 0.09627790749073029, + "learning_rate": 1.758463728199626e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14180 + }, + { + "epoch": 4.12860052371254, + "flow_matching_loss": 0.013889676611870527, + "grad_norm": 0.11454810947179794, + "learning_rate": 1.7568348327616803e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14190 + }, + { + "epoch": 4.131510037823683, + "flow_matching_loss": 0.015228831861168145, + "grad_norm": 0.07685293257236481, + "learning_rate": 1.755205625245195e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14200 + }, + { + "epoch": 4.134419551934827, + "flow_matching_loss": 0.013746820762753487, + "grad_norm": 0.08030282706022263, + "learning_rate": 1.7535761076298104e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14210 + }, + { + "epoch": 4.137329066045971, + "flow_matching_loss": 0.013496614899486304, + "grad_norm": 0.08963262289762497, + "learning_rate": 1.7519462818955448e-05, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14220 + }, + { + "epoch": 4.1402385801571135, + "flow_matching_loss": 0.01493463758379221, + "grad_norm": 0.07776547968387604, + "learning_rate": 1.7503161500227896e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14230 + }, + { + "epoch": 4.143148094268257, + "flow_matching_loss": 0.014659299422055483, + "grad_norm": 0.07679387181997299, + "learning_rate": 1.7486857139923093e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14240 + }, + { + "epoch": 4.146057608379401, + "flow_matching_loss": 0.014898673817515373, + "grad_norm": 0.09313786029815674, + "learning_rate": 1.747054975785236e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14250 + }, + { + "epoch": 4.148967122490544, + "flow_matching_loss": 0.015903094317764044, + "grad_norm": 0.07560387253761292, + "learning_rate": 1.745423937383072e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14260 + }, + { + "epoch": 4.1518766366016875, + "flow_matching_loss": 0.013232088647782803, + "grad_norm": 0.07156052440404892, + "learning_rate": 1.7437926007676825e-05, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14270 + }, + { + "epoch": 4.154786150712831, + "flow_matching_loss": 0.014724692143499852, + "grad_norm": 0.07507237046957016, + "learning_rate": 1.742160967921294e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14280 + }, + { + "epoch": 4.157695664823974, + "flow_matching_loss": 0.015805251710116864, + "grad_norm": 0.07447381317615509, + "learning_rate": 1.7405290408264954e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14290 + }, + { + "epoch": 4.160605178935118, + "flow_matching_loss": 0.014357852935791015, + "grad_norm": 0.055837150663137436, + "learning_rate": 1.738896821466232e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14300 + }, + { + "epoch": 4.163514693046261, + "flow_matching_loss": 0.014213691931217909, + "grad_norm": 0.07191658765077591, + "learning_rate": 1.737264311823804e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14310 + }, + { + "epoch": 4.166424207157405, + "flow_matching_loss": 0.014819152653217316, + "grad_norm": 0.07382476329803467, + "learning_rate": 1.7356315138828645e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14320 + }, + { + "epoch": 4.169333721268548, + "flow_matching_loss": 0.01553381560370326, + "grad_norm": 0.07806475460529327, + "learning_rate": 1.7339984296274174e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14330 + }, + { + "epoch": 4.172243235379692, + "flow_matching_loss": 0.01489891717210412, + "grad_norm": 0.09105653315782547, + "learning_rate": 1.7323650610418134e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14340 + }, + { + "epoch": 4.175152749490835, + "flow_matching_loss": 0.014844807796180248, + "grad_norm": 0.09984341263771057, + "learning_rate": 1.730731410110749e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14350 + }, + { + "epoch": 4.178062263601978, + "flow_matching_loss": 0.014979499392211437, + "grad_norm": 0.07517286390066147, + "learning_rate": 1.7290974788192655e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14360 + }, + { + "epoch": 4.180971777713122, + "flow_matching_loss": 0.015357793681323529, + "grad_norm": 0.08493130654096603, + "learning_rate": 1.727463269152743e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14370 + }, + { + "epoch": 4.183881291824266, + "flow_matching_loss": 0.015724270697683097, + "grad_norm": 0.07967507839202881, + "learning_rate": 1.7258287830968996e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14380 + }, + { + "epoch": 4.186790805935408, + "flow_matching_loss": 0.015233211684972048, + "grad_norm": 0.07800158858299255, + "learning_rate": 1.7241940226377908e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14390 + }, + { + "epoch": 4.189700320046552, + "flow_matching_loss": 0.014747006166726351, + "grad_norm": 0.07308946549892426, + "learning_rate": 1.7225589897618043e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14400 + }, + { + "epoch": 4.192609834157696, + "flow_matching_loss": 0.015788565017282964, + "grad_norm": 0.09524010866880417, + "learning_rate": 1.7209236864556588e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14410 + }, + { + "epoch": 4.195519348268839, + "flow_matching_loss": 0.014461743272840976, + "grad_norm": 0.07999199628829956, + "learning_rate": 1.719288114706403e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14420 + }, + { + "epoch": 4.198428862379982, + "flow_matching_loss": 0.016062608733773232, + "grad_norm": 0.1039251759648323, + "learning_rate": 1.7176522765014104e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14430 + }, + { + "epoch": 4.201338376491126, + "flow_matching_loss": 0.01680302172899246, + "grad_norm": 0.07494623214006424, + "learning_rate": 1.7160161738283782e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14440 + }, + { + "epoch": 4.20424789060227, + "flow_matching_loss": 0.015759934671223164, + "grad_norm": 0.06992339342832565, + "learning_rate": 1.714379808675326e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14450 + }, + { + "epoch": 4.207157404713413, + "flow_matching_loss": 0.01689430568367243, + "grad_norm": 0.1049075722694397, + "learning_rate": 1.712743183030592e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14460 + }, + { + "epoch": 4.210066918824556, + "flow_matching_loss": 0.01518243681639433, + "grad_norm": 0.10233863443136215, + "learning_rate": 1.7111062988828294e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14470 + }, + { + "epoch": 4.2129764329357, + "flow_matching_loss": 0.01587499277666211, + "grad_norm": 0.08596283942461014, + "learning_rate": 1.709469158221008e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14480 + }, + { + "epoch": 4.215885947046843, + "flow_matching_loss": 0.015418623015284539, + "grad_norm": 0.0785202756524086, + "learning_rate": 1.7078317630344083e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14490 + }, + { + "epoch": 4.2187954611579865, + "flow_matching_loss": 0.014982212614268065, + "grad_norm": 0.08539605885744095, + "learning_rate": 1.706194115312618e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14500 + }, + { + "epoch": 4.22170497526913, + "flow_matching_loss": 0.013810612075030804, + "grad_norm": 0.07026781886816025, + "learning_rate": 1.704556217045536e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14510 + }, + { + "epoch": 4.224614489380274, + "flow_matching_loss": 0.015287307277321815, + "grad_norm": 0.0818885937333107, + "learning_rate": 1.7029180702233612e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14520 + }, + { + "epoch": 4.227524003491417, + "flow_matching_loss": 0.01548102181404829, + "grad_norm": 0.080250583589077, + "learning_rate": 1.7012796768365975e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14530 + }, + { + "epoch": 4.2304335176025605, + "flow_matching_loss": 0.015137349348515272, + "grad_norm": 0.07762870192527771, + "learning_rate": 1.6996410388760465e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14540 + }, + { + "epoch": 4.233343031713704, + "flow_matching_loss": 0.014925900567322969, + "grad_norm": 0.08936086297035217, + "learning_rate": 1.698002158332809e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14550 + }, + { + "epoch": 4.236252545824847, + "flow_matching_loss": 0.015466941148042678, + "grad_norm": 0.08747684955596924, + "learning_rate": 1.6963630371982783e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14560 + }, + { + "epoch": 4.239162059935991, + "flow_matching_loss": 0.015566913038492202, + "grad_norm": 0.07859471440315247, + "learning_rate": 1.6947236774641416e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14570 + }, + { + "epoch": 4.242071574047134, + "flow_matching_loss": 0.01602203520014882, + "grad_norm": 0.09522400796413422, + "learning_rate": 1.6930840811223755e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14580 + }, + { + "epoch": 4.244981088158277, + "flow_matching_loss": 0.015669668186455966, + "grad_norm": 0.09615881741046906, + "learning_rate": 1.6914442501652448e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14590 + }, + { + "epoch": 4.247890602269421, + "flow_matching_loss": 0.014179198537021875, + "grad_norm": 0.07316907495260239, + "learning_rate": 1.6898041865852974e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14600 + }, + { + "epoch": 4.250800116380565, + "flow_matching_loss": 0.014304464962333441, + "grad_norm": 0.06751125305891037, + "learning_rate": 1.6881638923753663e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14610 + }, + { + "epoch": 4.2537096304917075, + "flow_matching_loss": 0.015227616764605045, + "grad_norm": 0.07092851400375366, + "learning_rate": 1.6865233695285632e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14620 + }, + { + "epoch": 4.256619144602851, + "flow_matching_loss": 0.014966520294547082, + "grad_norm": 0.07415245473384857, + "learning_rate": 1.6848826200382782e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14630 + }, + { + "epoch": 4.259528658713995, + "flow_matching_loss": 0.01451623197644949, + "grad_norm": 0.07499248534440994, + "learning_rate": 1.6832416458981762e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14640 + }, + { + "epoch": 4.262438172825139, + "flow_matching_loss": 0.014770927652716637, + "grad_norm": 0.07175338268280029, + "learning_rate": 1.681600449102196e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14650 + }, + { + "epoch": 4.265347686936281, + "flow_matching_loss": 0.014665864594280719, + "grad_norm": 0.07027586549520493, + "learning_rate": 1.6799590316445468e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14660 + }, + { + "epoch": 4.268257201047425, + "flow_matching_loss": 0.013845999538898469, + "grad_norm": 0.07819302380084991, + "learning_rate": 1.678317395519705e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14670 + }, + { + "epoch": 4.271166715158569, + "flow_matching_loss": 0.016228861827403306, + "grad_norm": 0.076007179915905, + "learning_rate": 1.676675542722413e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14680 + }, + { + "epoch": 4.274076229269712, + "flow_matching_loss": 0.014277228061109782, + "grad_norm": 0.07354190945625305, + "learning_rate": 1.6750334752476776e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14690 + }, + { + "epoch": 4.276985743380855, + "flow_matching_loss": 0.014257513172924518, + "grad_norm": 0.09136087447404861, + "learning_rate": 1.673391195090765e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14700 + }, + { + "epoch": 4.279895257491999, + "flow_matching_loss": 0.014636557269841432, + "grad_norm": 0.09176922589540482, + "learning_rate": 1.6717487042472006e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14710 + }, + { + "epoch": 4.282804771603142, + "flow_matching_loss": 0.016269149631261824, + "grad_norm": 0.08882725238800049, + "learning_rate": 1.6701060047127655e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14720 + }, + { + "epoch": 4.285714285714286, + "flow_matching_loss": 0.014009154122322797, + "grad_norm": 0.06758830696344376, + "learning_rate": 1.6684630984834947e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14730 + }, + { + "epoch": 4.288623799825429, + "flow_matching_loss": 0.01577377216890454, + "grad_norm": 0.06822438538074493, + "learning_rate": 1.666819987555675e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14740 + }, + { + "epoch": 4.291533313936573, + "flow_matching_loss": 0.016556976083666086, + "grad_norm": 0.0625971108675003, + "learning_rate": 1.6651766739258392e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14750 + }, + { + "epoch": 4.294442828047716, + "flow_matching_loss": 0.015623989142477512, + "grad_norm": 0.08312332630157471, + "learning_rate": 1.6635331595907704e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14760 + }, + { + "epoch": 4.2973523421588595, + "flow_matching_loss": 0.014134983625262975, + "grad_norm": 0.0933290421962738, + "learning_rate": 1.6618894465474923e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14770 + }, + { + "epoch": 4.300261856270003, + "flow_matching_loss": 0.014747746102511883, + "grad_norm": 0.09101437032222748, + "learning_rate": 1.660245536793271e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14780 + }, + { + "epoch": 4.303171370381146, + "flow_matching_loss": 0.01505565932020545, + "grad_norm": 0.09943174570798874, + "learning_rate": 1.6586014323256126e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14790 + }, + { + "epoch": 4.30608088449229, + "flow_matching_loss": 0.014893659763038159, + "grad_norm": 0.07880614697933197, + "learning_rate": 1.656957135142259e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14800 + }, + { + "epoch": 4.3089903986034335, + "flow_matching_loss": 0.014372179564088584, + "grad_norm": 0.08433464169502258, + "learning_rate": 1.655312647241186e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14810 + }, + { + "epoch": 4.311899912714576, + "flow_matching_loss": 0.016091348510235547, + "grad_norm": 0.08304093778133392, + "learning_rate": 1.6536679706206015e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14820 + }, + { + "epoch": 4.31480942682572, + "flow_matching_loss": 0.01472490355372429, + "grad_norm": 0.0659429207444191, + "learning_rate": 1.6520231072789427e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14830 + }, + { + "epoch": 4.317718940936864, + "flow_matching_loss": 0.013412334769964219, + "grad_norm": 0.06995540112257004, + "learning_rate": 1.6503780592148737e-05, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14840 + }, + { + "epoch": 4.320628455048007, + "flow_matching_loss": 0.015463532041758299, + "grad_norm": 0.07858077436685562, + "learning_rate": 1.648732828427283e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14850 + }, + { + "epoch": 4.32353796915915, + "flow_matching_loss": 0.013993626181036233, + "grad_norm": 0.08520372211933136, + "learning_rate": 1.647087416915281e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14860 + }, + { + "epoch": 4.326447483270294, + "flow_matching_loss": 0.014067402575165033, + "grad_norm": 0.08398837596178055, + "learning_rate": 1.6454418266781987e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14870 + }, + { + "epoch": 4.329356997381438, + "flow_matching_loss": 0.015399997308850288, + "grad_norm": 0.09523718804121017, + "learning_rate": 1.6437960597155824e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14880 + }, + { + "epoch": 4.3322665114925805, + "flow_matching_loss": 0.015567205101251601, + "grad_norm": 0.0763525739312172, + "learning_rate": 1.642150118027194e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14890 + }, + { + "epoch": 4.335176025603724, + "flow_matching_loss": 0.015831305459141732, + "grad_norm": 0.07996321469545364, + "learning_rate": 1.640504003613009e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14900 + }, + { + "epoch": 4.338085539714868, + "flow_matching_loss": 0.015092579182237387, + "grad_norm": 0.07606387138366699, + "learning_rate": 1.6388577184732107e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14910 + }, + { + "epoch": 4.340995053826011, + "flow_matching_loss": 0.014973573014140129, + "grad_norm": 0.07518859952688217, + "learning_rate": 1.6372112646081907e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14920 + }, + { + "epoch": 4.343904567937154, + "flow_matching_loss": 0.015584727562963963, + "grad_norm": 0.06764617562294006, + "learning_rate": 1.6355646440185464e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14930 + }, + { + "epoch": 4.346814082048298, + "flow_matching_loss": 0.015131119545549155, + "grad_norm": 0.0923362523317337, + "learning_rate": 1.6339178587050764e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14940 + }, + { + "epoch": 4.349723596159441, + "flow_matching_loss": 0.014864864107221366, + "grad_norm": 0.09131257236003876, + "learning_rate": 1.632270910668781e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14950 + }, + { + "epoch": 4.352633110270585, + "flow_matching_loss": 0.015050220862030983, + "grad_norm": 0.07823561877012253, + "learning_rate": 1.6306238019108562e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14960 + }, + { + "epoch": 4.355542624381728, + "flow_matching_loss": 0.01654695477336645, + "grad_norm": 0.10205482691526413, + "learning_rate": 1.6289765344326952e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14970 + }, + { + "epoch": 4.358452138492872, + "flow_matching_loss": 0.015807953476905823, + "grad_norm": 0.06716150045394897, + "learning_rate": 1.6273291102358833e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14980 + }, + { + "epoch": 4.361361652604015, + "flow_matching_loss": 0.01418626019731164, + "grad_norm": 0.08109015226364136, + "learning_rate": 1.6256815313221963e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 14990 + }, + { + "epoch": 4.364271166715159, + "flow_matching_loss": 0.013870825339108706, + "grad_norm": 0.06968723237514496, + "learning_rate": 1.6240337996935972e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15000 + }, + { + "epoch": 4.367180680826302, + "flow_matching_loss": 0.015030915476381778, + "grad_norm": 0.08531917631626129, + "learning_rate": 1.6223859173522367e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15010 + }, + { + "epoch": 4.370090194937445, + "flow_matching_loss": 0.015157500188797713, + "grad_norm": 0.06523136794567108, + "learning_rate": 1.6207378863004455e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15020 + }, + { + "epoch": 4.372999709048589, + "flow_matching_loss": 0.013722294755280019, + "grad_norm": 0.07243933528661728, + "learning_rate": 1.6190897085407382e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15030 + }, + { + "epoch": 4.3759092231597325, + "flow_matching_loss": 0.01632539490237832, + "grad_norm": 0.0734253078699112, + "learning_rate": 1.617441386075805e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15040 + }, + { + "epoch": 4.378818737270876, + "flow_matching_loss": 0.015991386491805315, + "grad_norm": 0.10683631896972656, + "learning_rate": 1.6157929209085142e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15050 + }, + { + "epoch": 4.381728251382019, + "flow_matching_loss": 0.014014955004677176, + "grad_norm": 0.06084362417459488, + "learning_rate": 1.614144315041906e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15060 + }, + { + "epoch": 4.384637765493163, + "flow_matching_loss": 0.017047799471765757, + "grad_norm": 0.08103403449058533, + "learning_rate": 1.612495570479192e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15070 + }, + { + "epoch": 4.3875472796043065, + "flow_matching_loss": 0.016107522509992124, + "grad_norm": 0.08495491743087769, + "learning_rate": 1.6108466892237517e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15080 + }, + { + "epoch": 4.390456793715449, + "flow_matching_loss": 0.014140655938535928, + "grad_norm": 0.07165364176034927, + "learning_rate": 1.609197673279132e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15090 + }, + { + "epoch": 4.393366307826593, + "flow_matching_loss": 0.013967789895832538, + "grad_norm": 0.09411190450191498, + "learning_rate": 1.6075485246490428e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15100 + }, + { + "epoch": 4.396275821937737, + "flow_matching_loss": 0.014032442681491375, + "grad_norm": 0.08396762609481812, + "learning_rate": 1.605899245337355e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15110 + }, + { + "epoch": 4.3991853360488795, + "flow_matching_loss": 0.014782024826854468, + "grad_norm": 0.08314624428749084, + "learning_rate": 1.6042498373480983e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15120 + }, + { + "epoch": 4.402094850160023, + "flow_matching_loss": 0.014922874327749015, + "grad_norm": 0.09240645915269852, + "learning_rate": 1.602600302685459e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15130 + }, + { + "epoch": 4.405004364271167, + "flow_matching_loss": 0.015184484422206879, + "grad_norm": 0.0910346582531929, + "learning_rate": 1.6009506433537774e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15140 + }, + { + "epoch": 4.40791387838231, + "flow_matching_loss": 0.014332036394625903, + "grad_norm": 0.09902123361825943, + "learning_rate": 1.599300861357545e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15150 + }, + { + "epoch": 4.4108233924934535, + "flow_matching_loss": 0.01413049902766943, + "grad_norm": 0.07101614028215408, + "learning_rate": 1.5976509587014032e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15160 + }, + { + "epoch": 4.413732906604597, + "flow_matching_loss": 0.015009857714176178, + "grad_norm": 0.04130292311310768, + "learning_rate": 1.5960009373901384e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15170 + }, + { + "epoch": 4.416642420715741, + "flow_matching_loss": 0.014895289670675993, + "grad_norm": 0.08816580474376678, + "learning_rate": 1.5943507994286823e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15180 + }, + { + "epoch": 4.419551934826884, + "flow_matching_loss": 0.01442231060937047, + "grad_norm": 0.11421776562929153, + "learning_rate": 1.5927005468221082e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15190 + }, + { + "epoch": 4.422461448938027, + "flow_matching_loss": 0.014334757812321186, + "grad_norm": 0.08985734730958939, + "learning_rate": 1.5910501815756285e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15200 + }, + { + "epoch": 4.425370963049171, + "flow_matching_loss": 0.017768604215234518, + "grad_norm": 0.10437354445457458, + "learning_rate": 1.589399705694593e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15210 + }, + { + "epoch": 4.428280477160314, + "flow_matching_loss": 0.013526966050267219, + "grad_norm": 0.06458356976509094, + "learning_rate": 1.587749121184485e-05, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15220 + }, + { + "epoch": 4.431189991271458, + "flow_matching_loss": 0.014334492199122906, + "grad_norm": 0.07025489211082458, + "learning_rate": 1.5860984300509206e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15230 + }, + { + "epoch": 4.434099505382601, + "flow_matching_loss": 0.01299317916855216, + "grad_norm": 0.06986387073993683, + "learning_rate": 1.5844476342996453e-05, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15240 + }, + { + "epoch": 4.437009019493744, + "flow_matching_loss": 0.012946550734341145, + "grad_norm": 0.06992793083190918, + "learning_rate": 1.5827967359365306e-05, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15250 + }, + { + "epoch": 4.439918533604888, + "flow_matching_loss": 0.013781459257006645, + "grad_norm": 0.09060268849134445, + "learning_rate": 1.5811457369675758e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15260 + }, + { + "epoch": 4.442828047716032, + "flow_matching_loss": 0.014552890788763762, + "grad_norm": 0.08867060393095016, + "learning_rate": 1.5794946393988982e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15270 + }, + { + "epoch": 4.445737561827175, + "flow_matching_loss": 0.014612020272761584, + "grad_norm": 0.07358456403017044, + "learning_rate": 1.5778434452367382e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15280 + }, + { + "epoch": 4.448647075938318, + "flow_matching_loss": 0.014483146741986275, + "grad_norm": 0.07012782990932465, + "learning_rate": 1.576192156487452e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15290 + }, + { + "epoch": 4.451556590049462, + "flow_matching_loss": 0.014289630856364966, + "grad_norm": 0.07718883454799652, + "learning_rate": 1.574540775157511e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15300 + }, + { + "epoch": 4.4544661041606055, + "flow_matching_loss": 0.015394666884094476, + "grad_norm": 0.09731829166412354, + "learning_rate": 1.5728893032534998e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15310 + }, + { + "epoch": 4.457375618271748, + "flow_matching_loss": 0.012836019974201918, + "grad_norm": 0.07412710040807724, + "learning_rate": 1.571237742782112e-05, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15320 + }, + { + "epoch": 4.460285132382892, + "flow_matching_loss": 0.013701215013861656, + "grad_norm": 0.06988117098808289, + "learning_rate": 1.569586095750149e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15330 + }, + { + "epoch": 4.463194646494036, + "flow_matching_loss": 0.015245670918375254, + "grad_norm": 0.08507321774959564, + "learning_rate": 1.5679343641645193e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15340 + }, + { + "epoch": 4.466104160605179, + "flow_matching_loss": 0.014805424027144909, + "grad_norm": 0.0634031891822815, + "learning_rate": 1.5662825500322304e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15350 + }, + { + "epoch": 4.469013674716322, + "flow_matching_loss": 0.016491972841322422, + "grad_norm": 0.0807950347661972, + "learning_rate": 1.5646306553603932e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15360 + }, + { + "epoch": 4.471923188827466, + "flow_matching_loss": 0.013970221858471632, + "grad_norm": 0.08035554736852646, + "learning_rate": 1.5629786821562154e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15370 + }, + { + "epoch": 4.47483270293861, + "flow_matching_loss": 0.016221597231924532, + "grad_norm": 0.09006202220916748, + "learning_rate": 1.561326632427001e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15380 + }, + { + "epoch": 4.4777422170497525, + "flow_matching_loss": 0.013975668884813785, + "grad_norm": 0.07957533746957779, + "learning_rate": 1.5596745081801447e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15390 + }, + { + "epoch": 4.480651731160896, + "flow_matching_loss": 0.013437022175639867, + "grad_norm": 0.08115212619304657, + "learning_rate": 1.5580223114231348e-05, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15400 + }, + { + "epoch": 4.48356124527204, + "flow_matching_loss": 0.015148298721760512, + "grad_norm": 0.09862697124481201, + "learning_rate": 1.5563700441635445e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15410 + }, + { + "epoch": 4.486470759383183, + "flow_matching_loss": 0.014630807191133499, + "grad_norm": 0.07992416620254517, + "learning_rate": 1.5547177084090358e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15420 + }, + { + "epoch": 4.4893802734943264, + "flow_matching_loss": 0.013019058294594288, + "grad_norm": 0.06291518360376358, + "learning_rate": 1.5530653061673514e-05, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15430 + }, + { + "epoch": 4.49228978760547, + "flow_matching_loss": 0.013219672534614802, + "grad_norm": 0.07245729863643646, + "learning_rate": 1.5514128394463167e-05, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15440 + }, + { + "epoch": 4.495199301716613, + "flow_matching_loss": 0.016330114286392926, + "grad_norm": 0.09696550667285919, + "learning_rate": 1.549760310253834e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15450 + }, + { + "epoch": 4.498108815827757, + "flow_matching_loss": 0.013116530794650316, + "grad_norm": 0.08611389994621277, + "learning_rate": 1.5481077205978816e-05, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15460 + }, + { + "epoch": 4.5010183299389, + "flow_matching_loss": 0.014116436708718538, + "grad_norm": 0.09240510314702988, + "learning_rate": 1.5464550724865135e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15470 + }, + { + "epoch": 4.503927844050043, + "flow_matching_loss": 0.015336194448173046, + "grad_norm": 0.09991602599620819, + "learning_rate": 1.5448023679278517e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15480 + }, + { + "epoch": 4.506837358161187, + "flow_matching_loss": 0.013883461616933346, + "grad_norm": 0.09660439193248749, + "learning_rate": 1.543149608930088e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15490 + }, + { + "epoch": 4.509746872272331, + "flow_matching_loss": 0.0151103631593287, + "grad_norm": 0.07507980614900589, + "learning_rate": 1.5414967975014805e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15500 + }, + { + "epoch": 4.512656386383474, + "flow_matching_loss": 0.015098787751048804, + "grad_norm": 0.08128144592046738, + "learning_rate": 1.539843935650351e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15510 + }, + { + "epoch": 4.515565900494617, + "flow_matching_loss": 0.01432015746831894, + "grad_norm": 0.08194848150014877, + "learning_rate": 1.5381910253850834e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15520 + }, + { + "epoch": 4.518475414605761, + "flow_matching_loss": 0.015895967744290827, + "grad_norm": 0.08704596012830734, + "learning_rate": 1.5365380687141183e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15530 + }, + { + "epoch": 4.521384928716905, + "flow_matching_loss": 0.015021363086998463, + "grad_norm": 0.07523554563522339, + "learning_rate": 1.5348850676459544e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15540 + }, + { + "epoch": 4.524294442828047, + "flow_matching_loss": 0.013903049007058143, + "grad_norm": 0.06428109109401703, + "learning_rate": 1.5332320241891438e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15550 + }, + { + "epoch": 4.527203956939191, + "flow_matching_loss": 0.016111268568783997, + "grad_norm": 0.08954824507236481, + "learning_rate": 1.5315789403522908e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15560 + }, + { + "epoch": 4.530113471050335, + "flow_matching_loss": 0.014935821387916803, + "grad_norm": 0.0713387206196785, + "learning_rate": 1.529925818144047e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15570 + }, + { + "epoch": 4.5330229851614785, + "flow_matching_loss": 0.013843467459082603, + "grad_norm": 0.06443995982408524, + "learning_rate": 1.5282726595731126e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15580 + }, + { + "epoch": 4.535932499272621, + "flow_matching_loss": 0.014464623294770718, + "grad_norm": 0.10092808306217194, + "learning_rate": 1.5266194666482304e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15590 + }, + { + "epoch": 4.538842013383765, + "flow_matching_loss": 0.015624248888343573, + "grad_norm": 0.07754936814308167, + "learning_rate": 1.5249662413781866e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15600 + }, + { + "epoch": 4.541751527494909, + "flow_matching_loss": 0.014295940473675727, + "grad_norm": 0.10332988202571869, + "learning_rate": 1.5233129857718054e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15610 + }, + { + "epoch": 4.5446610416060516, + "flow_matching_loss": 0.0127662293612957, + "grad_norm": 0.06545239686965942, + "learning_rate": 1.521659701837948e-05, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15620 + }, + { + "epoch": 4.547570555717195, + "flow_matching_loss": 0.014642744418233633, + "grad_norm": 0.1132861077785492, + "learning_rate": 1.5200063915855107e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15630 + }, + { + "epoch": 4.550480069828339, + "flow_matching_loss": 0.014675755146890878, + "grad_norm": 0.0766669437289238, + "learning_rate": 1.5183530570234212e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15640 + }, + { + "epoch": 4.553389583939482, + "flow_matching_loss": 0.014484820887446404, + "grad_norm": 0.06509560346603394, + "learning_rate": 1.5166997001606361e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15650 + }, + { + "epoch": 4.5562990980506255, + "flow_matching_loss": 0.01397568229585886, + "grad_norm": 0.078965924680233, + "learning_rate": 1.5150463230061417e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15660 + }, + { + "epoch": 4.559208612161769, + "flow_matching_loss": 0.014735007099807263, + "grad_norm": 0.062429189682006836, + "learning_rate": 1.5133929275689455e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15670 + }, + { + "epoch": 4.562118126272912, + "flow_matching_loss": 0.014139562658965588, + "grad_norm": 0.08984331786632538, + "learning_rate": 1.5117395158580795e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15680 + }, + { + "epoch": 4.565027640384056, + "flow_matching_loss": 0.014524329639971256, + "grad_norm": 0.06667698919773102, + "learning_rate": 1.510086089882595e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15690 + }, + { + "epoch": 4.567937154495199, + "flow_matching_loss": 0.014795872196555138, + "grad_norm": 0.09919349104166031, + "learning_rate": 1.5084326516515599e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15700 + }, + { + "epoch": 4.570846668606343, + "flow_matching_loss": 0.01512451721355319, + "grad_norm": 0.08123046159744263, + "learning_rate": 1.5067792031740582e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15710 + }, + { + "epoch": 4.573756182717486, + "flow_matching_loss": 0.015394161827862263, + "grad_norm": 0.1065877228975296, + "learning_rate": 1.505125746459185e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15720 + }, + { + "epoch": 4.57666569682863, + "flow_matching_loss": 0.014578016474843025, + "grad_norm": 0.08976095914840698, + "learning_rate": 1.5034722835160466e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15730 + }, + { + "epoch": 4.579575210939773, + "flow_matching_loss": 0.013344048336148263, + "grad_norm": 0.07386556267738342, + "learning_rate": 1.5018188163537566e-05, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15740 + }, + { + "epoch": 4.582484725050916, + "flow_matching_loss": 0.0140821504406631, + "grad_norm": 0.06501052528619766, + "learning_rate": 1.5001653469814331e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15750 + }, + { + "epoch": 4.58539423916206, + "flow_matching_loss": 0.01398624125868082, + "grad_norm": 0.07561297714710236, + "learning_rate": 1.4985118774081978e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15760 + }, + { + "epoch": 4.588303753273204, + "flow_matching_loss": 0.014233425352722406, + "grad_norm": 0.08032281696796417, + "learning_rate": 1.4968584096431708e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15770 + }, + { + "epoch": 4.591213267384346, + "flow_matching_loss": 0.014516256004571914, + "grad_norm": 0.0841061919927597, + "learning_rate": 1.4952049456954731e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15780 + }, + { + "epoch": 4.59412278149549, + "flow_matching_loss": 0.015196537598967552, + "grad_norm": 0.08409544825553894, + "learning_rate": 1.4935514875742182e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15790 + }, + { + "epoch": 4.597032295606634, + "flow_matching_loss": 0.013386325351893902, + "grad_norm": 0.08574692159891129, + "learning_rate": 1.4918980372885143e-05, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15800 + }, + { + "epoch": 4.599941809717777, + "flow_matching_loss": 0.01491455240175128, + "grad_norm": 0.08868012577295303, + "learning_rate": 1.4902445968474591e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15810 + }, + { + "epoch": 4.60285132382892, + "flow_matching_loss": 0.015310697630047799, + "grad_norm": 0.08070006966590881, + "learning_rate": 1.4885911682601378e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15820 + }, + { + "epoch": 4.605760837940064, + "flow_matching_loss": 0.014509843662381172, + "grad_norm": 0.07494336366653442, + "learning_rate": 1.4869377535356235e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15830 + }, + { + "epoch": 4.608670352051208, + "flow_matching_loss": 0.013870447874069214, + "grad_norm": 0.07871320843696594, + "learning_rate": 1.4852843546829704e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15840 + }, + { + "epoch": 4.611579866162351, + "flow_matching_loss": 0.014847736433148384, + "grad_norm": 0.08036062121391296, + "learning_rate": 1.483630973711214e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15850 + }, + { + "epoch": 4.614489380273494, + "flow_matching_loss": 0.01454058000817895, + "grad_norm": 0.09614580124616623, + "learning_rate": 1.4819776126293677e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15860 + }, + { + "epoch": 4.617398894384638, + "flow_matching_loss": 0.014092055894434453, + "grad_norm": 0.08636300265789032, + "learning_rate": 1.4803242734464214e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15870 + }, + { + "epoch": 4.620308408495781, + "flow_matching_loss": 0.014255230128765107, + "grad_norm": 0.06117107719182968, + "learning_rate": 1.4786709581713383e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15880 + }, + { + "epoch": 4.6232179226069245, + "flow_matching_loss": 0.015391869191080331, + "grad_norm": 0.08179186284542084, + "learning_rate": 1.4770176688130524e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15890 + }, + { + "epoch": 4.626127436718068, + "flow_matching_loss": 0.016708184219896793, + "grad_norm": 0.09183014184236526, + "learning_rate": 1.475364407380466e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15900 + }, + { + "epoch": 4.629036950829212, + "flow_matching_loss": 0.014158191904425622, + "grad_norm": 0.08650922775268555, + "learning_rate": 1.4737111758824476e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15910 + }, + { + "epoch": 4.631946464940355, + "flow_matching_loss": 0.014552545547485352, + "grad_norm": 0.08347480744123459, + "learning_rate": 1.4720579763278289e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15920 + }, + { + "epoch": 4.6348559790514985, + "flow_matching_loss": 0.01603951258584857, + "grad_norm": 0.07432330399751663, + "learning_rate": 1.4704048107254042e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15930 + }, + { + "epoch": 4.637765493162642, + "flow_matching_loss": 0.015504970960319043, + "grad_norm": 0.06635594367980957, + "learning_rate": 1.4687516810839246e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15940 + }, + { + "epoch": 4.640675007273785, + "flow_matching_loss": 0.015857152361422776, + "grad_norm": 0.10431186854839325, + "learning_rate": 1.4670985894120997e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15950 + }, + { + "epoch": 4.643584521384929, + "flow_matching_loss": 0.01458518197759986, + "grad_norm": 0.08237912505865097, + "learning_rate": 1.465445537718591e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15960 + }, + { + "epoch": 4.646494035496072, + "flow_matching_loss": 0.01357204644009471, + "grad_norm": 0.0918801799416542, + "learning_rate": 1.463792528012012e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15970 + }, + { + "epoch": 4.649403549607215, + "flow_matching_loss": 0.017459400463849306, + "grad_norm": 0.06362897902727127, + "learning_rate": 1.4621395623009252e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15980 + }, + { + "epoch": 4.652313063718359, + "flow_matching_loss": 0.015372572839260102, + "grad_norm": 0.07856226712465286, + "learning_rate": 1.4604866425938407e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 15990 + }, + { + "epoch": 4.655222577829503, + "flow_matching_loss": 0.01423191586509347, + "grad_norm": 0.06872551143169403, + "learning_rate": 1.4588337708992115e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16000 + }, + { + "epoch": 4.6581320919406455, + "flow_matching_loss": 0.014382013771682977, + "grad_norm": 0.06902491301298141, + "learning_rate": 1.4571809492254318e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16010 + }, + { + "epoch": 4.661041606051789, + "flow_matching_loss": 0.016723122913390398, + "grad_norm": 0.08575188368558884, + "learning_rate": 1.4555281795808371e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16020 + }, + { + "epoch": 4.663951120162933, + "flow_matching_loss": 0.01502839382737875, + "grad_norm": 0.09838937222957611, + "learning_rate": 1.4538754639736966e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16030 + }, + { + "epoch": 4.666860634274077, + "flow_matching_loss": 0.013034147769212722, + "grad_norm": 0.07068893313407898, + "learning_rate": 1.4522228044122174e-05, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16040 + }, + { + "epoch": 4.669770148385219, + "flow_matching_loss": 0.015559915732592345, + "grad_norm": 0.07665523886680603, + "learning_rate": 1.4505702029045359e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16050 + }, + { + "epoch": 4.672679662496363, + "flow_matching_loss": 0.015184084419161081, + "grad_norm": 0.09281795471906662, + "learning_rate": 1.4489176614587189e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16060 + }, + { + "epoch": 4.675589176607507, + "flow_matching_loss": 0.014278464205563068, + "grad_norm": 0.0919865220785141, + "learning_rate": 1.4472651820827595e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16070 + }, + { + "epoch": 4.67849869071865, + "flow_matching_loss": 0.014968716073781252, + "grad_norm": 0.07334495335817337, + "learning_rate": 1.4456127667845766e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16080 + }, + { + "epoch": 4.681408204829793, + "flow_matching_loss": 0.014573677815496922, + "grad_norm": 0.08618137985467911, + "learning_rate": 1.4439604175720104e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16090 + }, + { + "epoch": 4.684317718940937, + "flow_matching_loss": 0.014227053429931403, + "grad_norm": 0.067264124751091, + "learning_rate": 1.4423081364528212e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16100 + }, + { + "epoch": 4.687227233052081, + "flow_matching_loss": 0.014692451898008585, + "grad_norm": 0.10153862088918686, + "learning_rate": 1.4406559254346863e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16110 + }, + { + "epoch": 4.690136747163224, + "flow_matching_loss": 0.014821956120431422, + "grad_norm": 0.0748070776462555, + "learning_rate": 1.4390037865251978e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16120 + }, + { + "epoch": 4.693046261274367, + "flow_matching_loss": 0.014645923487842083, + "grad_norm": 0.07195401936769485, + "learning_rate": 1.4373517217318597e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16130 + }, + { + "epoch": 4.69595577538551, + "flow_matching_loss": 0.013721899781376123, + "grad_norm": 0.10976959764957428, + "learning_rate": 1.4356997330620875e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16140 + }, + { + "epoch": 4.698865289496654, + "flow_matching_loss": 0.013732035458087922, + "grad_norm": 0.05918185040354729, + "learning_rate": 1.434047822523203e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16150 + }, + { + "epoch": 4.7017748036077975, + "flow_matching_loss": 0.016583086363971232, + "grad_norm": 0.09673801809549332, + "learning_rate": 1.4323959921224322e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16160 + }, + { + "epoch": 4.704684317718941, + "flow_matching_loss": 0.013821309991180896, + "grad_norm": 0.0769217386841774, + "learning_rate": 1.4307442438669066e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16170 + }, + { + "epoch": 4.707593831830084, + "flow_matching_loss": 0.014570187497884034, + "grad_norm": 0.06258105486631393, + "learning_rate": 1.4290925797636546e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16180 + }, + { + "epoch": 4.710503345941228, + "flow_matching_loss": 0.014515000022947788, + "grad_norm": 0.07292283326387405, + "learning_rate": 1.4274410018196039e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16190 + }, + { + "epoch": 4.7134128600523715, + "flow_matching_loss": 0.015164814982563257, + "grad_norm": 0.07036242634057999, + "learning_rate": 1.4257895120415783e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16200 + }, + { + "epoch": 4.716322374163514, + "flow_matching_loss": 0.015823520347476007, + "grad_norm": 0.1164296418428421, + "learning_rate": 1.4241381124362933e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16210 + }, + { + "epoch": 4.719231888274658, + "flow_matching_loss": 0.016368670947849752, + "grad_norm": 0.09073644876480103, + "learning_rate": 1.4224868050103546e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16220 + }, + { + "epoch": 4.722141402385802, + "flow_matching_loss": 0.014774301089346409, + "grad_norm": 0.07728391885757446, + "learning_rate": 1.4208355917702573e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16230 + }, + { + "epoch": 4.725050916496945, + "flow_matching_loss": 0.014340469986200333, + "grad_norm": 0.07799604535102844, + "learning_rate": 1.41918447472238e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16240 + }, + { + "epoch": 4.727960430608088, + "flow_matching_loss": 0.014757714699953794, + "grad_norm": 0.09126769751310349, + "learning_rate": 1.4175334558729869e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16250 + }, + { + "epoch": 4.730869944719232, + "flow_matching_loss": 0.014838451333343983, + "grad_norm": 0.08212810009717941, + "learning_rate": 1.4158825372282211e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16260 + }, + { + "epoch": 4.733779458830376, + "flow_matching_loss": 0.014319622702896596, + "grad_norm": 0.0762224942445755, + "learning_rate": 1.4142317207941042e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16270 + }, + { + "epoch": 4.7366889729415185, + "flow_matching_loss": 0.01468813568353653, + "grad_norm": 0.05543021485209465, + "learning_rate": 1.4125810085765339e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16280 + }, + { + "epoch": 4.739598487052662, + "flow_matching_loss": 0.016467265505343675, + "grad_norm": 0.08326709270477295, + "learning_rate": 1.410930402581281e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16290 + }, + { + "epoch": 4.742508001163806, + "flow_matching_loss": 0.014605036098510027, + "grad_norm": 0.10428942739963531, + "learning_rate": 1.4092799048139876e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16300 + }, + { + "epoch": 4.745417515274949, + "flow_matching_loss": 0.014631668943911791, + "grad_norm": 0.07454412430524826, + "learning_rate": 1.4076295172801647e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16310 + }, + { + "epoch": 4.748327029386092, + "flow_matching_loss": 0.013392662908881903, + "grad_norm": 0.08118512481451035, + "learning_rate": 1.4059792419851884e-05, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16320 + }, + { + "epoch": 4.751236543497236, + "flow_matching_loss": 0.015646727196872233, + "grad_norm": 0.06738359481096268, + "learning_rate": 1.4043290809342984e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16330 + }, + { + "epoch": 4.754146057608379, + "flow_matching_loss": 0.014073270931839943, + "grad_norm": 0.06856400519609451, + "learning_rate": 1.4026790361325958e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16340 + }, + { + "epoch": 4.757055571719523, + "flow_matching_loss": 0.014685491938143969, + "grad_norm": 0.08104920387268066, + "learning_rate": 1.4010291095850415e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16350 + }, + { + "epoch": 4.759965085830666, + "flow_matching_loss": 0.014405467920005322, + "grad_norm": 0.06822150200605392, + "learning_rate": 1.3993793032964515e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16360 + }, + { + "epoch": 4.76287459994181, + "flow_matching_loss": 0.014871872961521149, + "grad_norm": 0.10137245804071426, + "learning_rate": 1.3977296192714956e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16370 + }, + { + "epoch": 4.765784114052953, + "flow_matching_loss": 0.013979556784033776, + "grad_norm": 0.09212284535169601, + "learning_rate": 1.396080059514696e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16380 + }, + { + "epoch": 4.768693628164097, + "flow_matching_loss": 0.0162409876473248, + "grad_norm": 0.09794077277183533, + "learning_rate": 1.3944306260304232e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16390 + }, + { + "epoch": 4.77160314227524, + "flow_matching_loss": 0.014263976644724608, + "grad_norm": 0.0828828364610672, + "learning_rate": 1.3927813208228936e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16400 + }, + { + "epoch": 4.774512656386383, + "flow_matching_loss": 0.013555902522057295, + "grad_norm": 0.0660286694765091, + "learning_rate": 1.39113214589617e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16410 + }, + { + "epoch": 4.777422170497527, + "flow_matching_loss": 0.013314308505505324, + "grad_norm": 0.08267956227064133, + "learning_rate": 1.3894831032541545e-05, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16420 + }, + { + "epoch": 4.7803316846086705, + "flow_matching_loss": 0.013946569617837667, + "grad_norm": 0.06778787076473236, + "learning_rate": 1.3878341949005895e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16430 + }, + { + "epoch": 4.783241198719814, + "flow_matching_loss": 0.014925912115722894, + "grad_norm": 0.07828576117753983, + "learning_rate": 1.3861854228390546e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16440 + }, + { + "epoch": 4.786150712830957, + "flow_matching_loss": 0.014729516115039587, + "grad_norm": 0.08249921351671219, + "learning_rate": 1.3845367890729621e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16450 + }, + { + "epoch": 4.789060226942101, + "flow_matching_loss": 0.015345989447087049, + "grad_norm": 0.09129665791988373, + "learning_rate": 1.3828882956055591e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16460 + }, + { + "epoch": 4.7919697410532445, + "flow_matching_loss": 0.013588733691722154, + "grad_norm": 0.06301514059305191, + "learning_rate": 1.38123994443992e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16470 + }, + { + "epoch": 4.794879255164387, + "flow_matching_loss": 0.013186279777437448, + "grad_norm": 0.08729828894138336, + "learning_rate": 1.3795917375789465e-05, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16480 + }, + { + "epoch": 4.797788769275531, + "flow_matching_loss": 0.014674048591405153, + "grad_norm": 0.07717011868953705, + "learning_rate": 1.3779436770253658e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16490 + }, + { + "epoch": 4.800698283386675, + "flow_matching_loss": 0.014055029395967722, + "grad_norm": 0.07736579328775406, + "learning_rate": 1.3762957647817258e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16500 + }, + { + "epoch": 4.8036077974978175, + "flow_matching_loss": 0.013815392181277275, + "grad_norm": 0.0796428769826889, + "learning_rate": 1.374648002850397e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16510 + }, + { + "epoch": 4.806517311608961, + "flow_matching_loss": 0.01315568359568715, + "grad_norm": 0.07976143062114716, + "learning_rate": 1.3730003932335642e-05, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16520 + }, + { + "epoch": 4.809426825720105, + "flow_matching_loss": 0.013944532535970212, + "grad_norm": 0.07132702320814133, + "learning_rate": 1.3713529379332292e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16530 + }, + { + "epoch": 4.812336339831248, + "flow_matching_loss": 0.013842474948614836, + "grad_norm": 0.07411444932222366, + "learning_rate": 1.3697056389512054e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16540 + }, + { + "epoch": 4.8152458539423915, + "flow_matching_loss": 0.014342760108411312, + "grad_norm": 0.0819624662399292, + "learning_rate": 1.3680584982891162e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16550 + }, + { + "epoch": 4.818155368053535, + "flow_matching_loss": 0.012819636892527343, + "grad_norm": 0.07260286062955856, + "learning_rate": 1.3664115179483921e-05, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16560 + }, + { + "epoch": 4.821064882164679, + "flow_matching_loss": 0.013659362960606814, + "grad_norm": 0.07077622413635254, + "learning_rate": 1.364764699930271e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16570 + }, + { + "epoch": 4.823974396275822, + "flow_matching_loss": 0.012741822376847267, + "grad_norm": 0.06072103604674339, + "learning_rate": 1.363118046235791e-05, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16580 + }, + { + "epoch": 4.826883910386965, + "flow_matching_loss": 0.013578874990344048, + "grad_norm": 0.06319291144609451, + "learning_rate": 1.3614715588657921e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16590 + }, + { + "epoch": 4.829793424498109, + "flow_matching_loss": 0.014027332328259945, + "grad_norm": 0.0997951552271843, + "learning_rate": 1.3598252398209118e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16600 + }, + { + "epoch": 4.832702938609252, + "flow_matching_loss": 0.015883042383939026, + "grad_norm": 0.06984248012304306, + "learning_rate": 1.358179091101582e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16610 + }, + { + "epoch": 4.835612452720396, + "flow_matching_loss": 0.014560313243418933, + "grad_norm": 0.06538951396942139, + "learning_rate": 1.3565331147080299e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16620 + }, + { + "epoch": 4.838521966831539, + "flow_matching_loss": 0.016424162313342094, + "grad_norm": 0.07660477608442307, + "learning_rate": 1.3548873126402716e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16630 + }, + { + "epoch": 4.841431480942683, + "flow_matching_loss": 0.012269951775670052, + "grad_norm": 0.06121942773461342, + "learning_rate": 1.3532416868981117e-05, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16640 + }, + { + "epoch": 4.844340995053826, + "flow_matching_loss": 0.015092259366065263, + "grad_norm": 0.08641815185546875, + "learning_rate": 1.3515962394811401e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16650 + }, + { + "epoch": 4.84725050916497, + "flow_matching_loss": 0.01519012525677681, + "grad_norm": 0.10017720609903336, + "learning_rate": 1.3499509723887313e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16660 + }, + { + "epoch": 4.850160023276112, + "flow_matching_loss": 0.014194582030177117, + "grad_norm": 0.08840885013341904, + "learning_rate": 1.3483058876200396e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16670 + }, + { + "epoch": 4.853069537387256, + "flow_matching_loss": 0.014235251024365425, + "grad_norm": 0.06737853586673737, + "learning_rate": 1.3466609871739987e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16680 + }, + { + "epoch": 4.8559790514984, + "flow_matching_loss": 0.014257429353892802, + "grad_norm": 0.0949140414595604, + "learning_rate": 1.3450162730493172e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16690 + }, + { + "epoch": 4.8588885656095435, + "flow_matching_loss": 0.012631885055452586, + "grad_norm": 0.08374945819377899, + "learning_rate": 1.3433717472444781e-05, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16700 + }, + { + "epoch": 4.861798079720686, + "flow_matching_loss": 0.013917269837111234, + "grad_norm": 0.07398287951946259, + "learning_rate": 1.3417274117577343e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16710 + }, + { + "epoch": 4.86470759383183, + "flow_matching_loss": 0.014922299329191447, + "grad_norm": 0.08838286995887756, + "learning_rate": 1.3400832685871102e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16720 + }, + { + "epoch": 4.867617107942974, + "flow_matching_loss": 0.015351281687617301, + "grad_norm": 0.08320791274309158, + "learning_rate": 1.3384393197303936e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16730 + }, + { + "epoch": 4.870526622054117, + "flow_matching_loss": 0.014377421233803034, + "grad_norm": 0.08866966515779495, + "learning_rate": 1.336795567185138e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16740 + }, + { + "epoch": 4.87343613616526, + "flow_matching_loss": 0.014532910566776991, + "grad_norm": 0.0696142315864563, + "learning_rate": 1.3351520129486573e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16750 + }, + { + "epoch": 4.876345650276404, + "flow_matching_loss": 0.014980449620634318, + "grad_norm": 0.08698644489049911, + "learning_rate": 1.3335086590180252e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16760 + }, + { + "epoch": 4.879255164387548, + "flow_matching_loss": 0.015058276988565922, + "grad_norm": 0.0822606235742569, + "learning_rate": 1.3318655073900708e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16770 + }, + { + "epoch": 4.8821646784986905, + "flow_matching_loss": 0.013919273298233748, + "grad_norm": 0.07582014799118042, + "learning_rate": 1.3302225600613796e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16780 + }, + { + "epoch": 4.885074192609834, + "flow_matching_loss": 0.01336130676791072, + "grad_norm": 0.07408515363931656, + "learning_rate": 1.3285798190282869e-05, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16790 + }, + { + "epoch": 4.887983706720978, + "flow_matching_loss": 0.013689538650214671, + "grad_norm": 0.06974935531616211, + "learning_rate": 1.3269372862868776e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16800 + }, + { + "epoch": 4.890893220832121, + "flow_matching_loss": 0.014184916391968727, + "grad_norm": 0.0813729390501976, + "learning_rate": 1.3252949638329845e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16810 + }, + { + "epoch": 4.8938027349432645, + "flow_matching_loss": 0.015482174325734377, + "grad_norm": 0.08019409328699112, + "learning_rate": 1.3236528536621834e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16820 + }, + { + "epoch": 4.896712249054408, + "flow_matching_loss": 0.014624634850770235, + "grad_norm": 0.07172039151191711, + "learning_rate": 1.3220109577697943e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16830 + }, + { + "epoch": 4.899621763165551, + "flow_matching_loss": 0.014635954238474368, + "grad_norm": 0.08011574298143387, + "learning_rate": 1.3203692781508745e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16840 + }, + { + "epoch": 4.902531277276695, + "flow_matching_loss": 0.014664537832140922, + "grad_norm": 0.09822243452072144, + "learning_rate": 1.3187278168002198e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16850 + }, + { + "epoch": 4.905440791387838, + "flow_matching_loss": 0.014527781028300523, + "grad_norm": 0.09535945951938629, + "learning_rate": 1.3170865757123606e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16860 + }, + { + "epoch": 4.908350305498981, + "flow_matching_loss": 0.013727824855595826, + "grad_norm": 0.08278372138738632, + "learning_rate": 1.3154455568815592e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16870 + }, + { + "epoch": 4.911259819610125, + "flow_matching_loss": 0.014565099030733109, + "grad_norm": 0.08199728280305862, + "learning_rate": 1.3138047623018087e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16880 + }, + { + "epoch": 4.914169333721269, + "flow_matching_loss": 0.014319510199129581, + "grad_norm": 0.09505245089530945, + "learning_rate": 1.3121641939668293e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16890 + }, + { + "epoch": 4.917078847832412, + "flow_matching_loss": 0.01437838664278388, + "grad_norm": 0.09958186745643616, + "learning_rate": 1.3105238538700659e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16900 + }, + { + "epoch": 4.919988361943555, + "flow_matching_loss": 0.013476345408707856, + "grad_norm": 0.08008596301078796, + "learning_rate": 1.3088837440046866e-05, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16910 + }, + { + "epoch": 4.922897876054699, + "flow_matching_loss": 0.01429281458258629, + "grad_norm": 0.07631220668554306, + "learning_rate": 1.3072438663635787e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16920 + }, + { + "epoch": 4.925807390165843, + "flow_matching_loss": 0.01456923158839345, + "grad_norm": 0.08522520959377289, + "learning_rate": 1.3056042229393497e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16930 + }, + { + "epoch": 4.928716904276985, + "flow_matching_loss": 0.014662631135433913, + "grad_norm": 0.08634315431118011, + "learning_rate": 1.30396481572432e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16940 + }, + { + "epoch": 4.931626418388129, + "flow_matching_loss": 0.017257215082645418, + "grad_norm": 0.07365034520626068, + "learning_rate": 1.3023256467105237e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16950 + }, + { + "epoch": 4.934535932499273, + "flow_matching_loss": 0.013243357837200164, + "grad_norm": 0.05530531331896782, + "learning_rate": 1.3006867178897067e-05, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16960 + }, + { + "epoch": 4.9374454466104165, + "flow_matching_loss": 0.014088227227330209, + "grad_norm": 0.07021339237689972, + "learning_rate": 1.2990480312533213e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16970 + }, + { + "epoch": 4.940354960721559, + "flow_matching_loss": 0.015809244010597467, + "grad_norm": 0.06388810276985168, + "learning_rate": 1.2974095887925258e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16980 + }, + { + "epoch": 4.943264474832703, + "flow_matching_loss": 0.014120496436953544, + "grad_norm": 0.064362071454525, + "learning_rate": 1.2957713924981837e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 16990 + }, + { + "epoch": 4.946173988943847, + "flow_matching_loss": 0.0133256901986897, + "grad_norm": 0.06198849529027939, + "learning_rate": 1.2941334443608572e-05, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17000 + }, + { + "epoch": 4.94908350305499, + "flow_matching_loss": 0.014105560164898634, + "grad_norm": 0.07580011337995529, + "learning_rate": 1.2924957463708073e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17010 + }, + { + "epoch": 4.951993017166133, + "flow_matching_loss": 0.014293930120766163, + "grad_norm": 0.0650518462061882, + "learning_rate": 1.2908583005179924e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17020 + }, + { + "epoch": 4.954902531277277, + "flow_matching_loss": 0.016169224679470063, + "grad_norm": 0.06986358761787415, + "learning_rate": 1.2892211087920624e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17030 + }, + { + "epoch": 4.95781204538842, + "flow_matching_loss": 0.012169232312589883, + "grad_norm": 0.06352328509092331, + "learning_rate": 1.287584173182361e-05, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17040 + }, + { + "epoch": 4.9607215594995635, + "flow_matching_loss": 0.014230998419225215, + "grad_norm": 0.06794797629117966, + "learning_rate": 1.2859474956779188e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17050 + }, + { + "epoch": 4.963631073610707, + "flow_matching_loss": 0.014136901125311852, + "grad_norm": 0.09690054506063461, + "learning_rate": 1.2843110782674532e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17060 + }, + { + "epoch": 4.96654058772185, + "flow_matching_loss": 0.013950208015739918, + "grad_norm": 0.0981927216053009, + "learning_rate": 1.2826749229393655e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17070 + }, + { + "epoch": 4.969450101832994, + "flow_matching_loss": 0.01558370068669319, + "grad_norm": 0.07120291143655777, + "learning_rate": 1.2810390316817381e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17080 + }, + { + "epoch": 4.9723596159441374, + "flow_matching_loss": 0.013906182814389467, + "grad_norm": 0.1037883460521698, + "learning_rate": 1.2794034064823346e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17090 + }, + { + "epoch": 4.975269130055281, + "flow_matching_loss": 0.014103555306792259, + "grad_norm": 0.08093751221895218, + "learning_rate": 1.277768049328593e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17100 + }, + { + "epoch": 4.978178644166424, + "flow_matching_loss": 0.014909974485635757, + "grad_norm": 0.09241246432065964, + "learning_rate": 1.2761329622076269e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17110 + }, + { + "epoch": 4.981088158277568, + "flow_matching_loss": 0.013594804238528014, + "grad_norm": 0.07650624960660934, + "learning_rate": 1.274498147106221e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17120 + }, + { + "epoch": 4.983997672388711, + "flow_matching_loss": 0.013011347502470016, + "grad_norm": 0.069866843521595, + "learning_rate": 1.272863606010829e-05, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17130 + }, + { + "epoch": 4.986907186499854, + "flow_matching_loss": 0.015285687800496817, + "grad_norm": 0.07052939385175705, + "learning_rate": 1.2712293409075739e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17140 + }, + { + "epoch": 4.989816700610998, + "flow_matching_loss": 0.014901444409042596, + "grad_norm": 0.08401324599981308, + "learning_rate": 1.2695953537822413e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17150 + }, + { + "epoch": 4.992726214722142, + "flow_matching_loss": 0.014991042017936707, + "grad_norm": 0.09070569276809692, + "learning_rate": 1.2679616466202792e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17160 + }, + { + "epoch": 4.995635728833285, + "flow_matching_loss": 0.01724343542009592, + "grad_norm": 0.09058485925197601, + "learning_rate": 1.2663282214067967e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17170 + }, + { + "epoch": 4.998545242944428, + "flow_matching_loss": 0.016057119145989417, + "grad_norm": 0.07499388605356216, + "learning_rate": 1.264695080126559e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17180 + }, + { + "epoch": 5.001454757055572, + "flow_matching_loss": 0.013174483273178338, + "grad_norm": 0.08485870808362961, + "learning_rate": 1.263062224763986e-05, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17190 + }, + { + "epoch": 5.004364271166716, + "flow_matching_loss": 0.01475791959092021, + "grad_norm": 0.0808720663189888, + "learning_rate": 1.2614296573031522e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17200 + }, + { + "epoch": 5.007273785277858, + "flow_matching_loss": 0.01492586052045226, + "grad_norm": 0.07600932568311691, + "learning_rate": 1.259797379727781e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17210 + }, + { + "epoch": 5.010183299389002, + "flow_matching_loss": 0.013432506565004587, + "grad_norm": 0.08669573813676834, + "learning_rate": 1.2581653940212423e-05, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17220 + }, + { + "epoch": 5.013092813500146, + "flow_matching_loss": 0.014336793962866069, + "grad_norm": 0.08033598214387894, + "learning_rate": 1.2565337021665537e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17230 + }, + { + "epoch": 5.016002327611289, + "flow_matching_loss": 0.014789143856614829, + "grad_norm": 0.09820553660392761, + "learning_rate": 1.2549023061463739e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17240 + }, + { + "epoch": 5.018911841722432, + "flow_matching_loss": 0.014397968538105488, + "grad_norm": 0.06980694085359573, + "learning_rate": 1.2532712079430038e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17250 + }, + { + "epoch": 5.021821355833576, + "flow_matching_loss": 0.015867799799889327, + "grad_norm": 0.08698073774576187, + "learning_rate": 1.2516404095383813e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17260 + }, + { + "epoch": 5.024730869944719, + "flow_matching_loss": 0.015494402032345533, + "grad_norm": 0.1154361218214035, + "learning_rate": 1.2500099129140796e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17270 + }, + { + "epoch": 5.0276403840558626, + "flow_matching_loss": 0.014247441850602626, + "grad_norm": 0.08798468858003616, + "learning_rate": 1.2483797200513065e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17280 + }, + { + "epoch": 5.030549898167006, + "flow_matching_loss": 0.014925187267363072, + "grad_norm": 0.10355237871408463, + "learning_rate": 1.2467498329308992e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17290 + }, + { + "epoch": 5.03345941227815, + "flow_matching_loss": 0.015004857070744037, + "grad_norm": 0.058941714465618134, + "learning_rate": 1.2451202535333253e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17300 + }, + { + "epoch": 5.036368926389293, + "flow_matching_loss": 0.015615834947675466, + "grad_norm": 0.07094793021678925, + "learning_rate": 1.2434909838386768e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17310 + }, + { + "epoch": 5.0392784405004365, + "flow_matching_loss": 0.012983580958098173, + "grad_norm": 0.07840409129858017, + "learning_rate": 1.2418620258266703e-05, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17320 + }, + { + "epoch": 5.04218795461158, + "flow_matching_loss": 0.015236461628228426, + "grad_norm": 0.07920239865779877, + "learning_rate": 1.2402333814766432e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17330 + }, + { + "epoch": 5.045097468722723, + "flow_matching_loss": 0.013295307569205762, + "grad_norm": 0.08205486088991165, + "learning_rate": 1.2386050527675515e-05, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17340 + }, + { + "epoch": 5.048006982833867, + "flow_matching_loss": 0.012957388907670975, + "grad_norm": 0.07248560339212418, + "learning_rate": 1.2369770416779693e-05, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17350 + }, + { + "epoch": 5.05091649694501, + "flow_matching_loss": 0.013592023402452469, + "grad_norm": 0.07421214133501053, + "learning_rate": 1.2353493501860828e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17360 + }, + { + "epoch": 5.053826011056153, + "flow_matching_loss": 0.01527148773893714, + "grad_norm": 0.10148954391479492, + "learning_rate": 1.2337219802696905e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17370 + }, + { + "epoch": 5.056735525167297, + "flow_matching_loss": 0.015120538789778948, + "grad_norm": 0.06719441711902618, + "learning_rate": 1.2320949339062008e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17380 + }, + { + "epoch": 5.059645039278441, + "flow_matching_loss": 0.015150679368525743, + "grad_norm": 0.10341599583625793, + "learning_rate": 1.2304682130726281e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17390 + }, + { + "epoch": 5.0625545533895835, + "flow_matching_loss": 0.014969553798437119, + "grad_norm": 0.07313315570354462, + "learning_rate": 1.2288418197455913e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17400 + }, + { + "epoch": 5.065464067500727, + "flow_matching_loss": 0.01470245597884059, + "grad_norm": 0.09451326727867126, + "learning_rate": 1.2272157559013126e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17410 + }, + { + "epoch": 5.068373581611871, + "flow_matching_loss": 0.014056878909468652, + "grad_norm": 0.07537028193473816, + "learning_rate": 1.2255900235156123e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17420 + }, + { + "epoch": 5.071283095723015, + "flow_matching_loss": 0.013893928192555905, + "grad_norm": 0.09515228867530823, + "learning_rate": 1.2239646245639087e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17430 + }, + { + "epoch": 5.074192609834157, + "flow_matching_loss": 0.013839660491794348, + "grad_norm": 0.09155604243278503, + "learning_rate": 1.2223395610212142e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17440 + }, + { + "epoch": 5.077102123945301, + "flow_matching_loss": 0.014405519422143697, + "grad_norm": 0.060941897332668304, + "learning_rate": 1.2207148348621348e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17450 + }, + { + "epoch": 5.080011638056445, + "flow_matching_loss": 0.014006106834858655, + "grad_norm": 0.06991862505674362, + "learning_rate": 1.219090448060866e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17460 + }, + { + "epoch": 5.082921152167588, + "flow_matching_loss": 0.015146032441407442, + "grad_norm": 0.07323029637336731, + "learning_rate": 1.217466402591191e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17470 + }, + { + "epoch": 5.085830666278731, + "flow_matching_loss": 0.013844878878444433, + "grad_norm": 0.06799393147230148, + "learning_rate": 1.2158427004264785e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17480 + }, + { + "epoch": 5.088740180389875, + "flow_matching_loss": 0.01452613091096282, + "grad_norm": 0.07743974775075912, + "learning_rate": 1.2142193435396788e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17490 + }, + { + "epoch": 5.091649694501018, + "flow_matching_loss": 0.01370912017300725, + "grad_norm": 0.06681108474731445, + "learning_rate": 1.2125963339033234e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17500 + }, + { + "epoch": 5.094559208612162, + "flow_matching_loss": 0.013859811332076788, + "grad_norm": 0.06763824075460434, + "learning_rate": 1.2109736734895237e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17510 + }, + { + "epoch": 5.097468722723305, + "flow_matching_loss": 0.014519299659878016, + "grad_norm": 0.07905834168195724, + "learning_rate": 1.2093513642699634e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17520 + }, + { + "epoch": 5.100378236834449, + "flow_matching_loss": 0.014625543542206288, + "grad_norm": 0.07283028960227966, + "learning_rate": 1.2077294082159024e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17530 + }, + { + "epoch": 5.103287750945592, + "flow_matching_loss": 0.014715418498963118, + "grad_norm": 0.0842815563082695, + "learning_rate": 1.2061078072981697e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17540 + }, + { + "epoch": 5.1061972650567355, + "flow_matching_loss": 0.013832768145948648, + "grad_norm": 0.09055306762456894, + "learning_rate": 1.2044865634871626e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17550 + }, + { + "epoch": 5.109106779167879, + "flow_matching_loss": 0.014508138597011565, + "grad_norm": 0.0697324350476265, + "learning_rate": 1.2028656787528468e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17560 + }, + { + "epoch": 5.112016293279022, + "flow_matching_loss": 0.012834996730089188, + "grad_norm": 0.07877586781978607, + "learning_rate": 1.2012451550647491e-05, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17570 + }, + { + "epoch": 5.114925807390166, + "flow_matching_loss": 0.012734392285346985, + "grad_norm": 0.07297014445066452, + "learning_rate": 1.1996249943919589e-05, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17580 + }, + { + "epoch": 5.1178353215013095, + "flow_matching_loss": 0.012701442185789346, + "grad_norm": 0.0695299282670021, + "learning_rate": 1.1980051987031237e-05, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17590 + }, + { + "epoch": 5.120744835612452, + "flow_matching_loss": 0.01277652671560645, + "grad_norm": 0.08383118361234665, + "learning_rate": 1.1963857699664483e-05, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17600 + }, + { + "epoch": 5.123654349723596, + "flow_matching_loss": 0.015230201464146376, + "grad_norm": 0.07866144180297852, + "learning_rate": 1.1947667101496911e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17610 + }, + { + "epoch": 5.12656386383474, + "flow_matching_loss": 0.014541693776845933, + "grad_norm": 0.07512625306844711, + "learning_rate": 1.1931480212201633e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17620 + }, + { + "epoch": 5.129473377945883, + "flow_matching_loss": 0.012751576118171215, + "grad_norm": 0.07307911664247513, + "learning_rate": 1.1915297051447237e-05, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17630 + }, + { + "epoch": 5.132382892057026, + "flow_matching_loss": 0.01326951738446951, + "grad_norm": 0.07951916009187698, + "learning_rate": 1.1899117638897793e-05, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17640 + }, + { + "epoch": 5.13529240616817, + "flow_matching_loss": 0.0145078394562006, + "grad_norm": 0.06883057206869125, + "learning_rate": 1.1882941994212811e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17650 + }, + { + "epoch": 5.138201920279314, + "flow_matching_loss": 0.014405203424394131, + "grad_norm": 0.0774250403046608, + "learning_rate": 1.1866770137047221e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17660 + }, + { + "epoch": 5.1411114343904565, + "flow_matching_loss": 0.014974320307374, + "grad_norm": 0.0801144614815712, + "learning_rate": 1.1850602087051363e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17670 + }, + { + "epoch": 5.1440209485016, + "flow_matching_loss": 0.013733332697302102, + "grad_norm": 0.07801534235477448, + "learning_rate": 1.183443786387094e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17680 + }, + { + "epoch": 5.146930462612744, + "flow_matching_loss": 0.014694625418633223, + "grad_norm": 0.0861586406826973, + "learning_rate": 1.1818277487147005e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17690 + }, + { + "epoch": 5.149839976723887, + "flow_matching_loss": 0.014064605440944433, + "grad_norm": 0.06927123665809631, + "learning_rate": 1.180212097651594e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17700 + }, + { + "epoch": 5.15274949083503, + "flow_matching_loss": 0.015349931176751852, + "grad_norm": 0.07302751392126083, + "learning_rate": 1.1785968351609422e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17710 + }, + { + "epoch": 5.155659004946174, + "flow_matching_loss": 0.013842426054179669, + "grad_norm": 0.06895443797111511, + "learning_rate": 1.1769819632054427e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17720 + }, + { + "epoch": 5.158568519057318, + "flow_matching_loss": 0.014253380242735147, + "grad_norm": 0.07778438925743103, + "learning_rate": 1.1753674837473165e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17730 + }, + { + "epoch": 5.161478033168461, + "flow_matching_loss": 0.013100327830761671, + "grad_norm": 0.059446923434734344, + "learning_rate": 1.1737533987483081e-05, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17740 + }, + { + "epoch": 5.164387547279604, + "flow_matching_loss": 0.013806027546525, + "grad_norm": 0.0643453598022461, + "learning_rate": 1.1721397101696835e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17750 + }, + { + "epoch": 5.167297061390748, + "flow_matching_loss": 0.013737361878156662, + "grad_norm": 0.07297095656394958, + "learning_rate": 1.1705264199722257e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17760 + }, + { + "epoch": 5.170206575501891, + "flow_matching_loss": 0.015510430093854666, + "grad_norm": 0.07082313299179077, + "learning_rate": 1.1689135301162356e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17770 + }, + { + "epoch": 5.173116089613035, + "flow_matching_loss": 0.013255203794687986, + "grad_norm": 0.07747276872396469, + "learning_rate": 1.1673010425615258e-05, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17780 + }, + { + "epoch": 5.176025603724178, + "flow_matching_loss": 0.0152359863743186, + "grad_norm": 0.08447699248790741, + "learning_rate": 1.1656889592674208e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17790 + }, + { + "epoch": 5.178935117835321, + "flow_matching_loss": 0.015629077143967152, + "grad_norm": 0.1200331598520279, + "learning_rate": 1.1640772821927533e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17800 + }, + { + "epoch": 5.181844631946465, + "flow_matching_loss": 0.013760239910334348, + "grad_norm": 0.08382770419120789, + "learning_rate": 1.162466013295864e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17810 + }, + { + "epoch": 5.1847541460576085, + "flow_matching_loss": 0.013536443933844566, + "grad_norm": 0.08971860259771347, + "learning_rate": 1.160855154534595e-05, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17820 + }, + { + "epoch": 5.187663660168752, + "flow_matching_loss": 0.013499094918370248, + "grad_norm": 0.0659145936369896, + "learning_rate": 1.1592447078662931e-05, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17830 + }, + { + "epoch": 5.190573174279895, + "flow_matching_loss": 0.015971735958009957, + "grad_norm": 0.07897434383630753, + "learning_rate": 1.1576346752478025e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17840 + }, + { + "epoch": 5.193482688391039, + "flow_matching_loss": 0.013998060021549463, + "grad_norm": 0.08359299600124359, + "learning_rate": 1.1560250586354643e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17850 + }, + { + "epoch": 5.1963922025021825, + "flow_matching_loss": 0.01304289735853672, + "grad_norm": 0.0731058418750763, + "learning_rate": 1.1544158599851146e-05, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17860 + }, + { + "epoch": 5.199301716613325, + "flow_matching_loss": 0.01395136322826147, + "grad_norm": 0.0802319198846817, + "learning_rate": 1.1528070812520809e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17870 + }, + { + "epoch": 5.202211230724469, + "flow_matching_loss": 0.013076271954923868, + "grad_norm": 0.07824064046144485, + "learning_rate": 1.1511987243911821e-05, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17880 + }, + { + "epoch": 5.205120744835613, + "flow_matching_loss": 0.013538339547812939, + "grad_norm": 0.06238299980759621, + "learning_rate": 1.1495907913567228e-05, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17890 + }, + { + "epoch": 5.2080302589467555, + "flow_matching_loss": 0.012480934709310531, + "grad_norm": 0.07444217056035995, + "learning_rate": 1.1479832841024936e-05, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17900 + }, + { + "epoch": 5.210939773057899, + "flow_matching_loss": 0.014742120262235403, + "grad_norm": 0.08569744974374771, + "learning_rate": 1.1463762045817675e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17910 + }, + { + "epoch": 5.213849287169043, + "flow_matching_loss": 0.01336787510663271, + "grad_norm": 0.07056215405464172, + "learning_rate": 1.1447695547472964e-05, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17920 + }, + { + "epoch": 5.216758801280186, + "flow_matching_loss": 0.014095176011323929, + "grad_norm": 0.06577488034963608, + "learning_rate": 1.1431633365513132e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17930 + }, + { + "epoch": 5.2196683153913295, + "flow_matching_loss": 0.013916104938834906, + "grad_norm": 0.07080544531345367, + "learning_rate": 1.1415575519455235e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17940 + }, + { + "epoch": 5.222577829502473, + "flow_matching_loss": 0.012974327336996794, + "grad_norm": 0.076821468770504, + "learning_rate": 1.1399522028811069e-05, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17950 + }, + { + "epoch": 5.225487343613617, + "flow_matching_loss": 0.013941044453531504, + "grad_norm": 0.07235971093177795, + "learning_rate": 1.1383472913087143e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17960 + }, + { + "epoch": 5.22839685772476, + "flow_matching_loss": 0.012575377989560366, + "grad_norm": 0.06894263625144958, + "learning_rate": 1.1367428191784643e-05, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17970 + }, + { + "epoch": 5.231306371835903, + "flow_matching_loss": 0.014110368583351374, + "grad_norm": 0.07341308891773224, + "learning_rate": 1.1351387884399413e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17980 + }, + { + "epoch": 5.234215885947047, + "flow_matching_loss": 0.014685844257473945, + "grad_norm": 0.08293548971414566, + "learning_rate": 1.1335352010421953e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 17990 + }, + { + "epoch": 5.23712540005819, + "flow_matching_loss": 0.01524152709171176, + "grad_norm": 0.11656209826469421, + "learning_rate": 1.1319320589337352e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18000 + }, + { + "epoch": 5.240034914169334, + "flow_matching_loss": 0.014020074717700481, + "grad_norm": 0.07882246375083923, + "learning_rate": 1.1303293640625299e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18010 + }, + { + "epoch": 5.242944428280477, + "flow_matching_loss": 0.013929210789501666, + "grad_norm": 0.07227837294340134, + "learning_rate": 1.1287271183760052e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18020 + }, + { + "epoch": 5.24585394239162, + "flow_matching_loss": 0.014053185191005468, + "grad_norm": 0.07670918107032776, + "learning_rate": 1.1271253238210398e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18030 + }, + { + "epoch": 5.248763456502764, + "flow_matching_loss": 0.01381723051890731, + "grad_norm": 0.09517525136470795, + "learning_rate": 1.1255239823439663e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18040 + }, + { + "epoch": 5.251672970613908, + "flow_matching_loss": 0.01412128945812583, + "grad_norm": 0.0841747596859932, + "learning_rate": 1.1239230958905652e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18050 + }, + { + "epoch": 5.254582484725051, + "flow_matching_loss": 0.014334895368665456, + "grad_norm": 0.07077088207006454, + "learning_rate": 1.1223226664060641e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18060 + }, + { + "epoch": 5.257491998836194, + "flow_matching_loss": 0.011991481855511665, + "grad_norm": 0.07647416740655899, + "learning_rate": 1.1207226958351362e-05, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18070 + }, + { + "epoch": 5.260401512947338, + "flow_matching_loss": 0.012789621390402317, + "grad_norm": 0.05713275074958801, + "learning_rate": 1.1191231861218955e-05, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18080 + }, + { + "epoch": 5.2633110270584815, + "flow_matching_loss": 0.01304421704262495, + "grad_norm": 0.08305735886096954, + "learning_rate": 1.1175241392098988e-05, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18090 + }, + { + "epoch": 5.266220541169624, + "flow_matching_loss": 0.013603393826633692, + "grad_norm": 0.0990513414144516, + "learning_rate": 1.1159255570421376e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18100 + }, + { + "epoch": 5.269130055280768, + "flow_matching_loss": 0.01425888789817691, + "grad_norm": 0.0787111148238182, + "learning_rate": 1.1143274415610408e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18110 + }, + { + "epoch": 5.272039569391912, + "flow_matching_loss": 0.014824128057807685, + "grad_norm": 0.08488097786903381, + "learning_rate": 1.1127297947084692e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18120 + }, + { + "epoch": 5.274949083503055, + "flow_matching_loss": 0.014517893455922604, + "grad_norm": 0.08265654742717743, + "learning_rate": 1.1111326184257135e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18130 + }, + { + "epoch": 5.277858597614198, + "flow_matching_loss": 0.014811432175338268, + "grad_norm": 0.08194331079721451, + "learning_rate": 1.109535914653495e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18140 + }, + { + "epoch": 5.280768111725342, + "flow_matching_loss": 0.014137065783143044, + "grad_norm": 0.07347078621387482, + "learning_rate": 1.1079396853319582e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18150 + }, + { + "epoch": 5.283677625836486, + "flow_matching_loss": 0.01434409050270915, + "grad_norm": 0.09449905902147293, + "learning_rate": 1.1063439324006724e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18160 + }, + { + "epoch": 5.2865871399476285, + "flow_matching_loss": 0.014212740398943425, + "grad_norm": 0.07819166034460068, + "learning_rate": 1.1047486577986285e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18170 + }, + { + "epoch": 5.289496654058772, + "flow_matching_loss": 0.012921795994043351, + "grad_norm": 0.06693124771118164, + "learning_rate": 1.1031538634642353e-05, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18180 + }, + { + "epoch": 5.292406168169916, + "flow_matching_loss": 0.014065996371209621, + "grad_norm": 0.0788087248802185, + "learning_rate": 1.1015595513353172e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18190 + }, + { + "epoch": 5.295315682281059, + "flow_matching_loss": 0.015297961980104446, + "grad_norm": 0.08602231740951538, + "learning_rate": 1.0999657233491156e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18200 + }, + { + "epoch": 5.2982251963922025, + "flow_matching_loss": 0.015396787598729134, + "grad_norm": 0.11109869927167892, + "learning_rate": 1.0983723814422808e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18210 + }, + { + "epoch": 5.301134710503346, + "flow_matching_loss": 0.013839747942984103, + "grad_norm": 0.08248860388994217, + "learning_rate": 1.0967795275508735e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18220 + }, + { + "epoch": 5.304044224614489, + "flow_matching_loss": 0.014963547699153423, + "grad_norm": 0.118488609790802, + "learning_rate": 1.095187163610361e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18230 + }, + { + "epoch": 5.306953738725633, + "flow_matching_loss": 0.01330208983272314, + "grad_norm": 0.07503626495599747, + "learning_rate": 1.0935952915556163e-05, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18240 + }, + { + "epoch": 5.309863252836776, + "flow_matching_loss": 0.015089655946940183, + "grad_norm": 0.08472942560911179, + "learning_rate": 1.0920039133209137e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18250 + }, + { + "epoch": 5.31277276694792, + "flow_matching_loss": 0.014178075548261403, + "grad_norm": 0.07356730103492737, + "learning_rate": 1.0904130308399283e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18260 + }, + { + "epoch": 5.315682281059063, + "flow_matching_loss": 0.014837050437927246, + "grad_norm": 0.06548425555229187, + "learning_rate": 1.0888226460457318e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18270 + }, + { + "epoch": 5.318591795170207, + "flow_matching_loss": 0.013993861619383097, + "grad_norm": 0.07497602701187134, + "learning_rate": 1.087232760870792e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18280 + }, + { + "epoch": 5.32150130928135, + "flow_matching_loss": 0.01361953942105174, + "grad_norm": 0.08522956073284149, + "learning_rate": 1.0856433772469686e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18290 + }, + { + "epoch": 5.324410823392493, + "flow_matching_loss": 0.014280647691339254, + "grad_norm": 0.0748744085431099, + "learning_rate": 1.0840544971055135e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18300 + }, + { + "epoch": 5.327320337503637, + "flow_matching_loss": 0.012320141308009624, + "grad_norm": 0.07844077050685883, + "learning_rate": 1.0824661223770655e-05, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18310 + }, + { + "epoch": 5.330229851614781, + "flow_matching_loss": 0.0167470739223063, + "grad_norm": 0.12603890895843506, + "learning_rate": 1.0808782549916498e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18320 + }, + { + "epoch": 5.333139365725923, + "flow_matching_loss": 0.014190875925123692, + "grad_norm": 0.06117792800068855, + "learning_rate": 1.079290896878675e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18330 + }, + { + "epoch": 5.336048879837067, + "flow_matching_loss": 0.01464774226769805, + "grad_norm": 0.08312884718179703, + "learning_rate": 1.07770404996693e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18340 + }, + { + "epoch": 5.338958393948211, + "flow_matching_loss": 0.012319139763712883, + "grad_norm": 0.10944189876317978, + "learning_rate": 1.0761177161845848e-05, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18350 + }, + { + "epoch": 5.3418679080593545, + "flow_matching_loss": 0.015030022151768208, + "grad_norm": 0.06116833910346031, + "learning_rate": 1.0745318974591841e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18360 + }, + { + "epoch": 5.344777422170497, + "flow_matching_loss": 0.014492313284426928, + "grad_norm": 0.08094902336597443, + "learning_rate": 1.0729465957176469e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18370 + }, + { + "epoch": 5.347686936281641, + "flow_matching_loss": 0.01333428779616952, + "grad_norm": 0.0730765238404274, + "learning_rate": 1.0713618128862642e-05, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18380 + }, + { + "epoch": 5.350596450392785, + "flow_matching_loss": 0.0139237679541111, + "grad_norm": 0.05362412706017494, + "learning_rate": 1.0697775508906969e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18390 + }, + { + "epoch": 5.353505964503928, + "flow_matching_loss": 0.013760816212743521, + "grad_norm": 0.060967400670051575, + "learning_rate": 1.068193811655972e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18400 + }, + { + "epoch": 5.356415478615071, + "flow_matching_loss": 0.014660908747464419, + "grad_norm": 0.0787210762500763, + "learning_rate": 1.0666105971064835e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18410 + }, + { + "epoch": 5.359324992726215, + "flow_matching_loss": 0.013746879808604718, + "grad_norm": 0.09715977311134338, + "learning_rate": 1.0650279091659852e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18420 + }, + { + "epoch": 5.362234506837358, + "flow_matching_loss": 0.013131294772028923, + "grad_norm": 0.07471799105405807, + "learning_rate": 1.0634457497575926e-05, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18430 + }, + { + "epoch": 5.3651440209485015, + "flow_matching_loss": 0.014621380716562271, + "grad_norm": 0.0822833850979805, + "learning_rate": 1.061864120803778e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18440 + }, + { + "epoch": 5.368053535059645, + "flow_matching_loss": 0.013868337869644165, + "grad_norm": 0.06635168194770813, + "learning_rate": 1.0602830242263701e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18450 + }, + { + "epoch": 5.370963049170788, + "flow_matching_loss": 0.013565827300772071, + "grad_norm": 0.07220041751861572, + "learning_rate": 1.0587024619465506e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18460 + }, + { + "epoch": 5.373872563281932, + "flow_matching_loss": 0.01360450629144907, + "grad_norm": 0.05307985842227936, + "learning_rate": 1.0571224358848517e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18470 + }, + { + "epoch": 5.3767820773930755, + "flow_matching_loss": 0.014258207194507122, + "grad_norm": 0.07804977148771286, + "learning_rate": 1.0555429479611537e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18480 + }, + { + "epoch": 5.379691591504219, + "flow_matching_loss": 0.013797235395759343, + "grad_norm": 0.07200499624013901, + "learning_rate": 1.0539640000946835e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18490 + }, + { + "epoch": 5.382601105615362, + "flow_matching_loss": 0.012477932963520288, + "grad_norm": 0.06504882127046585, + "learning_rate": 1.0523855942040111e-05, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18500 + }, + { + "epoch": 5.385510619726506, + "flow_matching_loss": 0.013339301571249962, + "grad_norm": 0.05906660482287407, + "learning_rate": 1.0508077322070496e-05, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18510 + }, + { + "epoch": 5.388420133837649, + "flow_matching_loss": 0.013868662621825933, + "grad_norm": 0.08986791223287582, + "learning_rate": 1.0492304160210495e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18520 + }, + { + "epoch": 5.391329647948792, + "flow_matching_loss": 0.015176227409392595, + "grad_norm": 0.08087968826293945, + "learning_rate": 1.0476536475625985e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18530 + }, + { + "epoch": 5.394239162059936, + "flow_matching_loss": 0.01600948264822364, + "grad_norm": 0.0920604020357132, + "learning_rate": 1.0460774287476195e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18540 + }, + { + "epoch": 5.39714867617108, + "flow_matching_loss": 0.013185323029756547, + "grad_norm": 0.06203969195485115, + "learning_rate": 1.044501761491366e-05, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18550 + }, + { + "epoch": 5.400058190282223, + "flow_matching_loss": 0.012988746911287308, + "grad_norm": 0.08109024167060852, + "learning_rate": 1.0429266477084235e-05, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18560 + }, + { + "epoch": 5.402967704393366, + "flow_matching_loss": 0.014999587927013636, + "grad_norm": 0.08345464617013931, + "learning_rate": 1.0413520893127034e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18570 + }, + { + "epoch": 5.40587721850451, + "flow_matching_loss": 0.013098410982638597, + "grad_norm": 0.0789119228720665, + "learning_rate": 1.0397780882174424e-05, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18580 + }, + { + "epoch": 5.408786732615654, + "flow_matching_loss": 0.013907326105982065, + "grad_norm": 0.07319469749927521, + "learning_rate": 1.0382046463351998e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18590 + }, + { + "epoch": 5.411696246726796, + "flow_matching_loss": 0.01491400832310319, + "grad_norm": 0.0759134367108345, + "learning_rate": 1.0366317655778568e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18600 + }, + { + "epoch": 5.41460576083794, + "flow_matching_loss": 0.01303564989939332, + "grad_norm": 0.09008701890707016, + "learning_rate": 1.035059447856611e-05, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18610 + }, + { + "epoch": 5.417515274949084, + "flow_matching_loss": 0.013703648746013642, + "grad_norm": 0.06842056661844254, + "learning_rate": 1.0334876950819777e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18620 + }, + { + "epoch": 5.420424789060227, + "flow_matching_loss": 0.014206151198595762, + "grad_norm": 0.08440735936164856, + "learning_rate": 1.0319165091637837e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18630 + }, + { + "epoch": 5.42333430317137, + "flow_matching_loss": 0.014173902198672294, + "grad_norm": 0.06344017386436462, + "learning_rate": 1.0303458920111686e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18640 + }, + { + "epoch": 5.426243817282514, + "flow_matching_loss": 0.014974792581051588, + "grad_norm": 0.06240127235651016, + "learning_rate": 1.0287758455325803e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18650 + }, + { + "epoch": 5.429153331393657, + "flow_matching_loss": 0.014034784585237502, + "grad_norm": 0.06940951943397522, + "learning_rate": 1.0272063716357729e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18660 + }, + { + "epoch": 5.432062845504801, + "flow_matching_loss": 0.013105412293225528, + "grad_norm": 0.060240402817726135, + "learning_rate": 1.0256374722278057e-05, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18670 + }, + { + "epoch": 5.434972359615944, + "flow_matching_loss": 0.0144074821844697, + "grad_norm": 0.06963799148797989, + "learning_rate": 1.0240691492150388e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18680 + }, + { + "epoch": 5.437881873727088, + "flow_matching_loss": 0.012875640019774436, + "grad_norm": 0.07454155385494232, + "learning_rate": 1.0225014045031334e-05, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18690 + }, + { + "epoch": 5.440791387838231, + "flow_matching_loss": 0.013245443347841501, + "grad_norm": 0.07150642573833466, + "learning_rate": 1.0209342399970468e-05, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18700 + }, + { + "epoch": 5.4437009019493745, + "flow_matching_loss": 0.013530796952545643, + "grad_norm": 0.0615660659968853, + "learning_rate": 1.019367657601031e-05, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18710 + }, + { + "epoch": 5.446610416060518, + "flow_matching_loss": 0.013718448393046857, + "grad_norm": 0.0727241039276123, + "learning_rate": 1.0178016592186323e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18720 + }, + { + "epoch": 5.449519930171661, + "flow_matching_loss": 0.016086266189813615, + "grad_norm": 0.08528797328472137, + "learning_rate": 1.0162362467526861e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18730 + }, + { + "epoch": 5.452429444282805, + "flow_matching_loss": 0.014361578039824963, + "grad_norm": 0.05863480642437935, + "learning_rate": 1.0146714221053159e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18740 + }, + { + "epoch": 5.4553389583939484, + "flow_matching_loss": 0.014821816515177489, + "grad_norm": 0.10753107070922852, + "learning_rate": 1.0131071871779314e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18750 + }, + { + "epoch": 5.458248472505091, + "flow_matching_loss": 0.01580682899802923, + "grad_norm": 0.06376317143440247, + "learning_rate": 1.0115435438712249e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18760 + }, + { + "epoch": 5.461157986616235, + "flow_matching_loss": 0.015841268002986908, + "grad_norm": 0.10683897882699966, + "learning_rate": 1.0099804940851716e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18770 + }, + { + "epoch": 5.464067500727379, + "flow_matching_loss": 0.013363798428326846, + "grad_norm": 0.07325812429189682, + "learning_rate": 1.0084180397190238e-05, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18780 + }, + { + "epoch": 5.4669770148385215, + "flow_matching_loss": 0.01513790600001812, + "grad_norm": 0.07019786536693573, + "learning_rate": 1.006856182671311e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18790 + }, + { + "epoch": 5.469886528949665, + "flow_matching_loss": 0.014613651018589736, + "grad_norm": 0.09300339221954346, + "learning_rate": 1.0052949248398367e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18800 + }, + { + "epoch": 5.472796043060809, + "flow_matching_loss": 0.014719314686954022, + "grad_norm": 0.07203436642885208, + "learning_rate": 1.003734268121676e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18810 + }, + { + "epoch": 5.475705557171953, + "flow_matching_loss": 0.014135550428181886, + "grad_norm": 0.08211177587509155, + "learning_rate": 1.0021742144131744e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18820 + }, + { + "epoch": 5.478615071283095, + "flow_matching_loss": 0.014328669290989637, + "grad_norm": 0.08199828863143921, + "learning_rate": 1.000614765609944e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18830 + }, + { + "epoch": 5.481524585394239, + "flow_matching_loss": 0.013131102174520492, + "grad_norm": 0.08433878421783447, + "learning_rate": 9.990559236068628e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18840 + }, + { + "epoch": 5.484434099505383, + "flow_matching_loss": 0.013476364128291608, + "grad_norm": 0.06485473364591599, + "learning_rate": 9.974976902980703e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18850 + }, + { + "epoch": 5.487343613616526, + "flow_matching_loss": 0.014918736554682255, + "grad_norm": 0.07683218270540237, + "learning_rate": 9.959400675769668e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18860 + }, + { + "epoch": 5.490253127727669, + "flow_matching_loss": 0.014722555316984654, + "grad_norm": 0.075995534658432, + "learning_rate": 9.943830573362106e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18870 + }, + { + "epoch": 5.493162641838813, + "flow_matching_loss": 0.015584193542599678, + "grad_norm": 0.08984941989183426, + "learning_rate": 9.928266614677168e-06, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18880 + }, + { + "epoch": 5.496072155949957, + "flow_matching_loss": 0.014371374063193798, + "grad_norm": 0.08136141300201416, + "learning_rate": 9.912708818626524e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18890 + }, + { + "epoch": 5.4989816700611, + "flow_matching_loss": 0.01358921229839325, + "grad_norm": 0.09291591495275497, + "learning_rate": 9.897157204114369e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18900 + }, + { + "epoch": 5.501891184172243, + "flow_matching_loss": 0.01319147488102317, + "grad_norm": 0.06537451595067978, + "learning_rate": 9.881611790037375e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18910 + }, + { + "epoch": 5.504800698283387, + "flow_matching_loss": 0.014242850244045258, + "grad_norm": 0.07002439349889755, + "learning_rate": 9.866072595284684e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18920 + }, + { + "epoch": 5.50771021239453, + "flow_matching_loss": 0.014595110714435578, + "grad_norm": 0.0660470575094223, + "learning_rate": 9.850539638737894e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18930 + }, + { + "epoch": 5.5106197265056736, + "flow_matching_loss": 0.01525071607902646, + "grad_norm": 0.08294948935508728, + "learning_rate": 9.835012939271006e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18940 + }, + { + "epoch": 5.513529240616817, + "flow_matching_loss": 0.01372428284958005, + "grad_norm": 0.08121013641357422, + "learning_rate": 9.819492515750418e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18950 + }, + { + "epoch": 5.51643875472796, + "flow_matching_loss": 0.013317237049341202, + "grad_norm": 0.06462250649929047, + "learning_rate": 9.803978387034918e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18960 + }, + { + "epoch": 5.519348268839104, + "flow_matching_loss": 0.016907600592821835, + "grad_norm": 0.07488692551851273, + "learning_rate": 9.788470571975624e-06, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18970 + }, + { + "epoch": 5.5222577829502475, + "flow_matching_loss": 0.013380698207765818, + "grad_norm": 0.06947974860668182, + "learning_rate": 9.772969089416008e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18980 + }, + { + "epoch": 5.52516729706139, + "flow_matching_loss": 0.013738674949854613, + "grad_norm": 0.06764917075634003, + "learning_rate": 9.757473958191822e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 18990 + }, + { + "epoch": 5.528076811172534, + "flow_matching_loss": 0.013209135830402374, + "grad_norm": 0.06537255644798279, + "learning_rate": 9.741985197131118e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19000 + }, + { + "epoch": 5.530986325283678, + "flow_matching_loss": 0.012434760574251413, + "grad_norm": 0.06341597437858582, + "learning_rate": 9.726502825054195e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19010 + }, + { + "epoch": 5.533895839394821, + "flow_matching_loss": 0.01286294530145824, + "grad_norm": 0.06695058941841125, + "learning_rate": 9.711026860773595e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19020 + }, + { + "epoch": 5.536805353505964, + "flow_matching_loss": 0.013298728410154581, + "grad_norm": 0.07664590328931808, + "learning_rate": 9.69555732309408e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19030 + }, + { + "epoch": 5.539714867617108, + "flow_matching_loss": 0.01533393943682313, + "grad_norm": 0.09982796758413315, + "learning_rate": 9.68009423081259e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19040 + }, + { + "epoch": 5.542624381728252, + "flow_matching_loss": 0.013436759077012538, + "grad_norm": 0.08752105385065079, + "learning_rate": 9.664637602718249e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19050 + }, + { + "epoch": 5.5455338958393945, + "flow_matching_loss": 0.013410498481243848, + "grad_norm": 0.07009952515363693, + "learning_rate": 9.649187457592313e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19060 + }, + { + "epoch": 5.548443409950538, + "flow_matching_loss": 0.013525201100856065, + "grad_norm": 0.06905335187911987, + "learning_rate": 9.63374381420816e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19070 + }, + { + "epoch": 5.551352924061682, + "flow_matching_loss": 0.014533462934195995, + "grad_norm": 0.08070108294487, + "learning_rate": 9.618306691331279e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19080 + }, + { + "epoch": 5.554262438172826, + "flow_matching_loss": 0.01508248345926404, + "grad_norm": 0.08117672055959702, + "learning_rate": 9.602876107719228e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19090 + }, + { + "epoch": 5.557171952283968, + "flow_matching_loss": 0.013818306941539049, + "grad_norm": 0.09232920408248901, + "learning_rate": 9.587452082121621e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19100 + }, + { + "epoch": 5.560081466395112, + "flow_matching_loss": 0.01401618393138051, + "grad_norm": 0.06534752994775772, + "learning_rate": 9.57203463328011e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19110 + }, + { + "epoch": 5.562990980506255, + "flow_matching_loss": 0.013245942629873752, + "grad_norm": 0.0608723908662796, + "learning_rate": 9.556623779928341e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19120 + }, + { + "epoch": 5.565900494617399, + "flow_matching_loss": 0.012943431548774243, + "grad_norm": 0.08211256563663483, + "learning_rate": 9.541219540791949e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19130 + }, + { + "epoch": 5.568810008728542, + "flow_matching_loss": 0.014719156827777625, + "grad_norm": 0.06107352301478386, + "learning_rate": 9.525821934588551e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19140 + }, + { + "epoch": 5.571719522839686, + "flow_matching_loss": 0.013904716819524765, + "grad_norm": 0.0753227025270462, + "learning_rate": 9.510430980027684e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19150 + }, + { + "epoch": 5.574629036950829, + "flow_matching_loss": 0.013522779569029808, + "grad_norm": 0.056349169462919235, + "learning_rate": 9.495046695810809e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19160 + }, + { + "epoch": 5.577538551061973, + "flow_matching_loss": 0.013702734280377627, + "grad_norm": 0.09230038523674011, + "learning_rate": 9.47966910063128e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19170 + }, + { + "epoch": 5.580448065173116, + "flow_matching_loss": 0.014062075223773719, + "grad_norm": 0.07966207712888718, + "learning_rate": 9.464298213174325e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19180 + }, + { + "epoch": 5.583357579284259, + "flow_matching_loss": 0.012519476935267448, + "grad_norm": 0.0641256645321846, + "learning_rate": 9.448934052117024e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19190 + }, + { + "epoch": 5.586267093395403, + "flow_matching_loss": 0.01410133419558406, + "grad_norm": 0.0813443511724472, + "learning_rate": 9.433576636128284e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19200 + }, + { + "epoch": 5.5891766075065465, + "flow_matching_loss": 0.012764712329953909, + "grad_norm": 0.07929657399654388, + "learning_rate": 9.41822598386881e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19210 + }, + { + "epoch": 5.59208612161769, + "flow_matching_loss": 0.01461963662877679, + "grad_norm": 0.05423362925648689, + "learning_rate": 9.402882113991094e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19220 + }, + { + "epoch": 5.594995635728833, + "flow_matching_loss": 0.015321825724095106, + "grad_norm": 0.08455118536949158, + "learning_rate": 9.387545045139381e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19230 + }, + { + "epoch": 5.597905149839977, + "flow_matching_loss": 0.013162200059741735, + "grad_norm": 0.06820191442966461, + "learning_rate": 9.372214795949655e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19240 + }, + { + "epoch": 5.6008146639511205, + "flow_matching_loss": 0.014720608107745647, + "grad_norm": 0.07933379709720612, + "learning_rate": 9.356891385049618e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19250 + }, + { + "epoch": 5.603724178062263, + "flow_matching_loss": 0.014043298736214637, + "grad_norm": 0.06991490721702576, + "learning_rate": 9.341574831058664e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19260 + }, + { + "epoch": 5.606633692173407, + "flow_matching_loss": 0.01379737388342619, + "grad_norm": 0.06425200402736664, + "learning_rate": 9.326265152587844e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19270 + }, + { + "epoch": 5.609543206284551, + "flow_matching_loss": 0.013860025443136692, + "grad_norm": 0.0696304515004158, + "learning_rate": 9.310962368239861e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19280 + }, + { + "epoch": 5.6124527203956935, + "flow_matching_loss": 0.014556194189935923, + "grad_norm": 0.07177449762821198, + "learning_rate": 9.295666496609036e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19290 + }, + { + "epoch": 5.615362234506837, + "flow_matching_loss": 0.013794043008238078, + "grad_norm": 0.06869437545537949, + "learning_rate": 9.280377556281306e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19300 + }, + { + "epoch": 5.618271748617981, + "flow_matching_loss": 0.01628536283969879, + "grad_norm": 0.06978055089712143, + "learning_rate": 9.265095565834172e-06, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19310 + }, + { + "epoch": 5.621181262729124, + "flow_matching_loss": 0.012428372167050838, + "grad_norm": 0.09864863008260727, + "learning_rate": 9.249820543836688e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19320 + }, + { + "epoch": 5.6240907768402675, + "flow_matching_loss": 0.014627764001488686, + "grad_norm": 0.08128707855939865, + "learning_rate": 9.234552508849454e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19330 + }, + { + "epoch": 5.627000290951411, + "flow_matching_loss": 0.013230200204998255, + "grad_norm": 0.07800736278295517, + "learning_rate": 9.21929147942456e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19340 + }, + { + "epoch": 5.629909805062555, + "flow_matching_loss": 0.01255423789843917, + "grad_norm": 0.07449322938919067, + "learning_rate": 9.204037474105611e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19350 + }, + { + "epoch": 5.632819319173698, + "flow_matching_loss": 0.013917815499007702, + "grad_norm": 0.0706578940153122, + "learning_rate": 9.188790511427653e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19360 + }, + { + "epoch": 5.635728833284841, + "flow_matching_loss": 0.01381744146347046, + "grad_norm": 0.07333850860595703, + "learning_rate": 9.173550609917186e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19370 + }, + { + "epoch": 5.638638347395985, + "flow_matching_loss": 0.014906936138868333, + "grad_norm": 0.099968321621418, + "learning_rate": 9.158317788092124e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19380 + }, + { + "epoch": 5.641547861507128, + "flow_matching_loss": 0.013764949236065149, + "grad_norm": 0.05746352672576904, + "learning_rate": 9.143092064461783e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19390 + }, + { + "epoch": 5.644457375618272, + "flow_matching_loss": 0.012484820792451501, + "grad_norm": 0.047256603837013245, + "learning_rate": 9.127873457526855e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19400 + }, + { + "epoch": 5.647366889729415, + "flow_matching_loss": 0.013861607294529676, + "grad_norm": 0.07210560142993927, + "learning_rate": 9.112661985779382e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19410 + }, + { + "epoch": 5.650276403840559, + "flow_matching_loss": 0.014828621596097945, + "grad_norm": 0.08028285950422287, + "learning_rate": 9.09745766770274e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19420 + }, + { + "epoch": 5.653185917951702, + "flow_matching_loss": 0.014000285137444734, + "grad_norm": 0.08074948191642761, + "learning_rate": 9.082260521771601e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19430 + }, + { + "epoch": 5.656095432062846, + "flow_matching_loss": 0.013784564472734928, + "grad_norm": 0.08663112670183182, + "learning_rate": 9.067070566451936e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19440 + }, + { + "epoch": 5.659004946173989, + "flow_matching_loss": 0.014770177751779556, + "grad_norm": 0.08776795864105225, + "learning_rate": 9.051887820200966e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19450 + }, + { + "epoch": 5.661914460285132, + "flow_matching_loss": 0.013020809832960367, + "grad_norm": 0.06020805612206459, + "learning_rate": 9.03671230146717e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19460 + }, + { + "epoch": 5.664823974396276, + "flow_matching_loss": 0.013973519951105118, + "grad_norm": 0.06758236140012741, + "learning_rate": 9.021544028690228e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19470 + }, + { + "epoch": 5.6677334885074195, + "flow_matching_loss": 0.013431855663657188, + "grad_norm": 0.08342023193836212, + "learning_rate": 9.006383020301024e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19480 + }, + { + "epoch": 5.670643002618562, + "flow_matching_loss": 0.012797566130757332, + "grad_norm": 0.07798338681459427, + "learning_rate": 8.991229294721613e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19490 + }, + { + "epoch": 5.673552516729706, + "flow_matching_loss": 0.015186396986246109, + "grad_norm": 0.07702511548995972, + "learning_rate": 8.976082870365191e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19500 + }, + { + "epoch": 5.67646203084085, + "flow_matching_loss": 0.013011136092245579, + "grad_norm": 0.06734813749790192, + "learning_rate": 8.960943765636105e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19510 + }, + { + "epoch": 5.679371544951993, + "flow_matching_loss": 0.013332582451403141, + "grad_norm": 0.07860258966684341, + "learning_rate": 8.945811998929792e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19520 + }, + { + "epoch": 5.682281059063136, + "flow_matching_loss": 0.012676289584487677, + "grad_norm": 0.06208709254860878, + "learning_rate": 8.93068758863277e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19530 + }, + { + "epoch": 5.68519057317428, + "flow_matching_loss": 0.013011361099779606, + "grad_norm": 0.06711776554584503, + "learning_rate": 8.915570553122628e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19540 + }, + { + "epoch": 5.688100087285424, + "flow_matching_loss": 0.013220662064850331, + "grad_norm": 0.06706066429615021, + "learning_rate": 8.900460910767982e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19550 + }, + { + "epoch": 5.6910096013965665, + "flow_matching_loss": 0.01316726803779602, + "grad_norm": 0.07574845850467682, + "learning_rate": 8.885358679928486e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19560 + }, + { + "epoch": 5.69391911550771, + "flow_matching_loss": 0.012383767776191235, + "grad_norm": 0.06340748816728592, + "learning_rate": 8.870263878954767e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19570 + }, + { + "epoch": 5.696828629618854, + "flow_matching_loss": 0.013981906231492758, + "grad_norm": 0.07075715810060501, + "learning_rate": 8.855176526188428e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19580 + }, + { + "epoch": 5.699738143729997, + "flow_matching_loss": 0.014002531953155994, + "grad_norm": 0.07737661153078079, + "learning_rate": 8.840096639962033e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19590 + }, + { + "epoch": 5.7026476578411405, + "flow_matching_loss": 0.01345181129872799, + "grad_norm": 0.06957145780324936, + "learning_rate": 8.825024238599056e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19600 + }, + { + "epoch": 5.705557171952284, + "flow_matching_loss": 0.014417755044996738, + "grad_norm": 0.06828322261571884, + "learning_rate": 8.809959340413891e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19610 + }, + { + "epoch": 5.708466686063428, + "flow_matching_loss": 0.013552786596119405, + "grad_norm": 0.052430395036935806, + "learning_rate": 8.79490196371181e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19620 + }, + { + "epoch": 5.711376200174571, + "flow_matching_loss": 0.013859143014997244, + "grad_norm": 0.09208868443965912, + "learning_rate": 8.779852126788949e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19630 + }, + { + "epoch": 5.714285714285714, + "flow_matching_loss": 0.014707072917371989, + "grad_norm": 0.07321837544441223, + "learning_rate": 8.764809847932277e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19640 + }, + { + "epoch": 5.717195228396857, + "flow_matching_loss": 0.013189580477774143, + "grad_norm": 0.0803893432021141, + "learning_rate": 8.749775145419577e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19650 + }, + { + "epoch": 5.720104742508001, + "flow_matching_loss": 0.01605983395129442, + "grad_norm": 0.08422481268644333, + "learning_rate": 8.734748037519427e-06, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19660 + }, + { + "epoch": 5.723014256619145, + "flow_matching_loss": 0.013996883947402238, + "grad_norm": 0.07347878813743591, + "learning_rate": 8.719728542491195e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19670 + }, + { + "epoch": 5.725923770730288, + "flow_matching_loss": 0.014734817482531071, + "grad_norm": 0.07745840400457382, + "learning_rate": 8.704716678584976e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19680 + }, + { + "epoch": 5.728833284841431, + "flow_matching_loss": 0.01462929304689169, + "grad_norm": 0.08333516865968704, + "learning_rate": 8.689712464041596e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19690 + }, + { + "epoch": 5.731742798952575, + "flow_matching_loss": 0.014246050082147122, + "grad_norm": 0.07650727778673172, + "learning_rate": 8.674715917092592e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19700 + }, + { + "epoch": 5.734652313063719, + "flow_matching_loss": 0.013957061059772968, + "grad_norm": 0.09963200241327286, + "learning_rate": 8.65972705596018e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19710 + }, + { + "epoch": 5.737561827174861, + "flow_matching_loss": 0.013646051660180092, + "grad_norm": 0.06418734043836594, + "learning_rate": 8.64474589885725e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19720 + }, + { + "epoch": 5.740471341286005, + "flow_matching_loss": 0.01441462356597185, + "grad_norm": 0.07088744640350342, + "learning_rate": 8.629772463987315e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19730 + }, + { + "epoch": 5.743380855397149, + "flow_matching_loss": 0.013560963235795498, + "grad_norm": 0.07893220335245132, + "learning_rate": 8.614806769544507e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19740 + }, + { + "epoch": 5.7462903695082925, + "flow_matching_loss": 0.011633584834635257, + "grad_norm": 0.07044023275375366, + "learning_rate": 8.59984883371356e-06, + "loss": 0.0116, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19750 + }, + { + "epoch": 5.749199883619435, + "flow_matching_loss": 0.01374082574620843, + "grad_norm": 0.059705816209316254, + "learning_rate": 8.584898674669772e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19760 + }, + { + "epoch": 5.752109397730579, + "flow_matching_loss": 0.014201650209724904, + "grad_norm": 0.09106852859258652, + "learning_rate": 8.569956310579001e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19770 + }, + { + "epoch": 5.755018911841723, + "flow_matching_loss": 0.013414052408188581, + "grad_norm": 0.09440051019191742, + "learning_rate": 8.555021759597627e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19780 + }, + { + "epoch": 5.757928425952866, + "flow_matching_loss": 0.015476438403129577, + "grad_norm": 0.08630137145519257, + "learning_rate": 8.54009503987253e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19790 + }, + { + "epoch": 5.760837940064009, + "flow_matching_loss": 0.012779364548623561, + "grad_norm": 0.06602420657873154, + "learning_rate": 8.525176169541095e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19800 + }, + { + "epoch": 5.763747454175153, + "flow_matching_loss": 0.014344696793705225, + "grad_norm": 0.06605277210474014, + "learning_rate": 8.510265166731151e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19810 + }, + { + "epoch": 5.766656968286296, + "flow_matching_loss": 0.013928807619959117, + "grad_norm": 0.07499410957098007, + "learning_rate": 8.495362049560964e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19820 + }, + { + "epoch": 5.7695664823974395, + "flow_matching_loss": 0.013621231261640786, + "grad_norm": 0.08595845848321915, + "learning_rate": 8.480466836139241e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19830 + }, + { + "epoch": 5.772475996508583, + "flow_matching_loss": 0.013450589217245578, + "grad_norm": 0.061808791011571884, + "learning_rate": 8.465579544565066e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19840 + }, + { + "epoch": 5.775385510619726, + "flow_matching_loss": 0.015342164039611816, + "grad_norm": 0.10419429838657379, + "learning_rate": 8.450700192927897e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19850 + }, + { + "epoch": 5.77829502473087, + "flow_matching_loss": 0.01352948397397995, + "grad_norm": 0.061743445694446564, + "learning_rate": 8.435828799307551e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19860 + }, + { + "epoch": 5.7812045388420135, + "flow_matching_loss": 0.013265693560242653, + "grad_norm": 0.07601521909236908, + "learning_rate": 8.420965381774172e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19870 + }, + { + "epoch": 5.784114052953157, + "flow_matching_loss": 0.013604666013270617, + "grad_norm": 0.09289446473121643, + "learning_rate": 8.406109958388222e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19880 + }, + { + "epoch": 5.7870235670643, + "flow_matching_loss": 0.013449338171631098, + "grad_norm": 0.06888096034526825, + "learning_rate": 8.391262547200434e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19890 + }, + { + "epoch": 5.789933081175444, + "flow_matching_loss": 0.014733711257576942, + "grad_norm": 0.08243697136640549, + "learning_rate": 8.376423166251816e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19900 + }, + { + "epoch": 5.792842595286587, + "flow_matching_loss": 0.014213358238339423, + "grad_norm": 0.07165886461734772, + "learning_rate": 8.361591833573612e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19910 + }, + { + "epoch": 5.79575210939773, + "flow_matching_loss": 0.014538467675447465, + "grad_norm": 0.06593390554189682, + "learning_rate": 8.346768567187287e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19920 + }, + { + "epoch": 5.798661623508874, + "flow_matching_loss": 0.0134547533467412, + "grad_norm": 0.06419916450977325, + "learning_rate": 8.331953385104516e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19930 + }, + { + "epoch": 5.801571137620018, + "flow_matching_loss": 0.013423800468444824, + "grad_norm": 0.06922859698534012, + "learning_rate": 8.317146305327134e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19940 + }, + { + "epoch": 5.804480651731161, + "flow_matching_loss": 0.014095025975257159, + "grad_norm": 0.07326412200927734, + "learning_rate": 8.302347345847146e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19950 + }, + { + "epoch": 5.807390165842304, + "flow_matching_loss": 0.01449103830382228, + "grad_norm": 0.06709704548120499, + "learning_rate": 8.287556524646683e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19960 + }, + { + "epoch": 5.810299679953448, + "flow_matching_loss": 0.012989426264539361, + "grad_norm": 0.056809037923812866, + "learning_rate": 8.272773859697978e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19970 + }, + { + "epoch": 5.813209194064592, + "flow_matching_loss": 0.013232431933283806, + "grad_norm": 0.09189996868371964, + "learning_rate": 8.257999368963374e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19980 + }, + { + "epoch": 5.816118708175734, + "flow_matching_loss": 0.0134742833673954, + "grad_norm": 0.07713286578655243, + "learning_rate": 8.243233070395266e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 19990 + }, + { + "epoch": 5.819028222286878, + "flow_matching_loss": 0.013946933019906283, + "grad_norm": 0.07694512605667114, + "learning_rate": 8.2284749819361e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20000 + }, + { + "epoch": 5.821937736398022, + "flow_matching_loss": 0.014757164008915424, + "grad_norm": 0.07786765694618225, + "learning_rate": 8.213725121518346e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20010 + }, + { + "epoch": 5.824847250509165, + "flow_matching_loss": 0.014996055513620377, + "grad_norm": 0.08803462982177734, + "learning_rate": 8.198983507064469e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20020 + }, + { + "epoch": 5.827756764620308, + "flow_matching_loss": 0.014623521734029055, + "grad_norm": 0.08339009433984756, + "learning_rate": 8.18425015648692e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20030 + }, + { + "epoch": 5.830666278731452, + "flow_matching_loss": 0.0139027233235538, + "grad_norm": 0.07796301692724228, + "learning_rate": 8.169525087688122e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20040 + }, + { + "epoch": 5.833575792842595, + "flow_matching_loss": 0.014880090206861495, + "grad_norm": 0.092775858938694, + "learning_rate": 8.154808318560413e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20050 + }, + { + "epoch": 5.836485306953739, + "flow_matching_loss": 0.013261178508400917, + "grad_norm": 0.0720883309841156, + "learning_rate": 8.140099866986058e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20060 + }, + { + "epoch": 5.839394821064882, + "flow_matching_loss": 0.012893917225301265, + "grad_norm": 0.07374638319015503, + "learning_rate": 8.125399750837202e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20070 + }, + { + "epoch": 5.842304335176026, + "flow_matching_loss": 0.01385287856683135, + "grad_norm": 0.07515527307987213, + "learning_rate": 8.110707987975888e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20080 + }, + { + "epoch": 5.845213849287169, + "flow_matching_loss": 0.014749199524521828, + "grad_norm": 0.07491724193096161, + "learning_rate": 8.096024596253982e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20090 + }, + { + "epoch": 5.8481233633983125, + "flow_matching_loss": 0.013229366485029459, + "grad_norm": 0.0765443816781044, + "learning_rate": 8.081349593513197e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20100 + }, + { + "epoch": 5.851032877509456, + "flow_matching_loss": 0.013214761950075626, + "grad_norm": 0.0906120091676712, + "learning_rate": 8.066682997585045e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20110 + }, + { + "epoch": 5.853942391620599, + "flow_matching_loss": 0.013111216295510531, + "grad_norm": 0.10625690221786499, + "learning_rate": 8.052024826290819e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20120 + }, + { + "epoch": 5.856851905731743, + "flow_matching_loss": 0.015779935475438833, + "grad_norm": 0.09113506972789764, + "learning_rate": 8.037375097441574e-06, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20130 + }, + { + "epoch": 5.8597614198428865, + "flow_matching_loss": 0.013436411134898662, + "grad_norm": 0.07908552139997482, + "learning_rate": 8.022733828838124e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20140 + }, + { + "epoch": 5.86267093395403, + "flow_matching_loss": 0.014688399713486433, + "grad_norm": 0.09008011966943741, + "learning_rate": 8.008101038270988e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20150 + }, + { + "epoch": 5.865580448065173, + "flow_matching_loss": 0.014431864488869905, + "grad_norm": 0.07192651182413101, + "learning_rate": 7.99347674352038e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20160 + }, + { + "epoch": 5.868489962176317, + "flow_matching_loss": 0.013540288247168065, + "grad_norm": 0.059935007244348526, + "learning_rate": 7.978860962356202e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20170 + }, + { + "epoch": 5.8713994762874595, + "flow_matching_loss": 0.013977660052478313, + "grad_norm": 0.08810340613126755, + "learning_rate": 7.964253712537997e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20180 + }, + { + "epoch": 5.874308990398603, + "flow_matching_loss": 0.013802854204550385, + "grad_norm": 0.06040380895137787, + "learning_rate": 7.949655011814968e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20190 + }, + { + "epoch": 5.877218504509747, + "flow_matching_loss": 0.013088070787489414, + "grad_norm": 0.06618435680866241, + "learning_rate": 7.9350648779259e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20200 + }, + { + "epoch": 5.880128018620891, + "flow_matching_loss": 0.013462936785072089, + "grad_norm": 0.06056733801960945, + "learning_rate": 7.92048332859919e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20210 + }, + { + "epoch": 5.8830375327320334, + "flow_matching_loss": 0.014848919026553631, + "grad_norm": 0.07528995722532272, + "learning_rate": 7.90591038155278e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20220 + }, + { + "epoch": 5.885947046843177, + "flow_matching_loss": 0.012801005970686674, + "grad_norm": 0.07406371086835861, + "learning_rate": 7.891346054494194e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20230 + }, + { + "epoch": 5.888856560954321, + "flow_matching_loss": 0.013847187533974648, + "grad_norm": 0.0773933008313179, + "learning_rate": 7.876790365120447e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20240 + }, + { + "epoch": 5.891766075065464, + "flow_matching_loss": 0.013645104505121708, + "grad_norm": 0.06827635318040848, + "learning_rate": 7.862243331118088e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20250 + }, + { + "epoch": 5.894675589176607, + "flow_matching_loss": 0.013626991864293813, + "grad_norm": 0.06843166053295135, + "learning_rate": 7.847704970163127e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20260 + }, + { + "epoch": 5.897585103287751, + "flow_matching_loss": 0.013644114509224892, + "grad_norm": 0.07399114966392517, + "learning_rate": 7.833175299921046e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20270 + }, + { + "epoch": 5.900494617398895, + "flow_matching_loss": 0.013299732701852918, + "grad_norm": 0.07332953810691833, + "learning_rate": 7.81865433804676e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20280 + }, + { + "epoch": 5.903404131510038, + "flow_matching_loss": 0.014063518401235342, + "grad_norm": 0.08378478139638901, + "learning_rate": 7.80414210218461e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20290 + }, + { + "epoch": 5.906313645621181, + "flow_matching_loss": 0.013520581554621458, + "grad_norm": 0.07136157900094986, + "learning_rate": 7.789638609968332e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20300 + }, + { + "epoch": 5.909223159732325, + "flow_matching_loss": 0.014487337227910757, + "grad_norm": 0.07059333473443985, + "learning_rate": 7.775143879021037e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20310 + }, + { + "epoch": 5.912132673843468, + "flow_matching_loss": 0.014042341988533736, + "grad_norm": 0.08462020009756088, + "learning_rate": 7.760657926955187e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20320 + }, + { + "epoch": 5.915042187954612, + "flow_matching_loss": 0.01333738248795271, + "grad_norm": 0.0887245386838913, + "learning_rate": 7.746180771372584e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20330 + }, + { + "epoch": 5.917951702065755, + "flow_matching_loss": 0.013730249833315611, + "grad_norm": 0.0699739158153534, + "learning_rate": 7.731712429864325e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20340 + }, + { + "epoch": 5.920861216176898, + "flow_matching_loss": 0.013774011377245187, + "grad_norm": 0.08325962722301483, + "learning_rate": 7.717252920010828e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20350 + }, + { + "epoch": 5.923770730288042, + "flow_matching_loss": 0.015258386079221965, + "grad_norm": 0.08618428558111191, + "learning_rate": 7.702802259381742e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20360 + }, + { + "epoch": 5.9266802443991855, + "flow_matching_loss": 0.014751589111983775, + "grad_norm": 0.09495894610881805, + "learning_rate": 7.688360465535999e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20370 + }, + { + "epoch": 5.929589758510328, + "flow_matching_loss": 0.01393575118854642, + "grad_norm": 0.08616695553064346, + "learning_rate": 7.673927556021734e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20380 + }, + { + "epoch": 5.932499272621472, + "flow_matching_loss": 0.01504011070355773, + "grad_norm": 0.08938861638307571, + "learning_rate": 7.659503548376284e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20390 + }, + { + "epoch": 5.935408786732616, + "flow_matching_loss": 0.01316290646791458, + "grad_norm": 0.09314657002687454, + "learning_rate": 7.645088460126193e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20400 + }, + { + "epoch": 5.9383183008437594, + "flow_matching_loss": 0.014024066366255283, + "grad_norm": 0.05843929573893547, + "learning_rate": 7.630682308787145e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20410 + }, + { + "epoch": 5.941227814954902, + "flow_matching_loss": 0.014066519681364298, + "grad_norm": 0.06680796295404434, + "learning_rate": 7.6162851118639754e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20420 + }, + { + "epoch": 5.944137329066046, + "flow_matching_loss": 0.01297402698546648, + "grad_norm": 0.060835324227809906, + "learning_rate": 7.601896886850632e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20430 + }, + { + "epoch": 5.94704684317719, + "flow_matching_loss": 0.014955730922520161, + "grad_norm": 0.15657220780849457, + "learning_rate": 7.587517651230168e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20440 + }, + { + "epoch": 5.9499563572883325, + "flow_matching_loss": 0.014421947579830886, + "grad_norm": 0.08897427469491959, + "learning_rate": 7.573147422474702e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20450 + }, + { + "epoch": 5.952865871399476, + "flow_matching_loss": 0.01303082974627614, + "grad_norm": 0.059640828520059586, + "learning_rate": 7.55878621804543e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20460 + }, + { + "epoch": 5.95577538551062, + "flow_matching_loss": 0.013911883253604174, + "grad_norm": 0.06590395420789719, + "learning_rate": 7.544434055392564e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20470 + }, + { + "epoch": 5.958684899621764, + "flow_matching_loss": 0.013857920933514834, + "grad_norm": 0.06536798179149628, + "learning_rate": 7.530090951955334e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20480 + }, + { + "epoch": 5.961594413732906, + "flow_matching_loss": 0.014002930279821158, + "grad_norm": 0.06690935045480728, + "learning_rate": 7.515756925161959e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20490 + }, + { + "epoch": 5.96450392784405, + "flow_matching_loss": 0.013378847390413284, + "grad_norm": 0.10565229505300522, + "learning_rate": 7.501431992429643e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20500 + }, + { + "epoch": 5.967413441955194, + "flow_matching_loss": 0.013727994449436665, + "grad_norm": 0.06711839884519577, + "learning_rate": 7.487116171164519e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20510 + }, + { + "epoch": 5.970322956066337, + "flow_matching_loss": 0.014269156102091073, + "grad_norm": 0.07583244144916534, + "learning_rate": 7.472809478761672e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20520 + }, + { + "epoch": 5.97323247017748, + "flow_matching_loss": 0.013395992852747441, + "grad_norm": 0.0664479061961174, + "learning_rate": 7.4585119326050775e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20530 + }, + { + "epoch": 5.976141984288624, + "flow_matching_loss": 0.013059158436954021, + "grad_norm": 0.06291623413562775, + "learning_rate": 7.444223550067599e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20540 + }, + { + "epoch": 5.979051498399767, + "flow_matching_loss": 0.012821360025554895, + "grad_norm": 0.06246177479624748, + "learning_rate": 7.429944348510964e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20550 + }, + { + "epoch": 5.981961012510911, + "flow_matching_loss": 0.014922027755528689, + "grad_norm": 0.10775981843471527, + "learning_rate": 7.415674345285763e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20560 + }, + { + "epoch": 5.984870526622054, + "flow_matching_loss": 0.013516869209706784, + "grad_norm": 0.06878874450922012, + "learning_rate": 7.401413557731387e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20570 + }, + { + "epoch": 5.987780040733197, + "flow_matching_loss": 0.012658418715000152, + "grad_norm": 0.07063031941652298, + "learning_rate": 7.387162003176037e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20580 + }, + { + "epoch": 5.990689554844341, + "flow_matching_loss": 0.013461919967085123, + "grad_norm": 0.06531249731779099, + "learning_rate": 7.372919698936697e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20590 + }, + { + "epoch": 5.9935990689554846, + "flow_matching_loss": 0.015170445758849382, + "grad_norm": 0.08464805036783218, + "learning_rate": 7.358686662319104e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20600 + }, + { + "epoch": 5.996508583066628, + "flow_matching_loss": 0.012818939611315727, + "grad_norm": 0.06139141321182251, + "learning_rate": 7.344462910617752e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20610 + }, + { + "epoch": 5.999418097177771, + "flow_matching_loss": 0.013043529726564885, + "grad_norm": 0.06412608921527863, + "learning_rate": 7.330248461115832e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20620 + }, + { + "epoch": 6.002327611288915, + "flow_matching_loss": 0.01338115893304348, + "grad_norm": 0.06606021523475647, + "learning_rate": 7.316043331085242e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20630 + }, + { + "epoch": 6.0052371254000585, + "flow_matching_loss": 0.014607445616275073, + "grad_norm": 0.08576945215463638, + "learning_rate": 7.30184753778655e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20640 + }, + { + "epoch": 6.008146639511201, + "flow_matching_loss": 0.014650371577590703, + "grad_norm": 0.0753631666302681, + "learning_rate": 7.2876610984689935e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20650 + }, + { + "epoch": 6.011056153622345, + "flow_matching_loss": 0.013512080255895852, + "grad_norm": 0.06642290949821472, + "learning_rate": 7.273484030370423e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20660 + }, + { + "epoch": 6.013965667733489, + "flow_matching_loss": 0.014321845024824142, + "grad_norm": 0.08033954352140427, + "learning_rate": 7.259316350717324e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20670 + }, + { + "epoch": 6.0168751818446315, + "flow_matching_loss": 0.01292652329429984, + "grad_norm": 0.06816756725311279, + "learning_rate": 7.245158076724757e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20680 + }, + { + "epoch": 6.019784695955775, + "flow_matching_loss": 0.013356755953282118, + "grad_norm": 0.05639943107962608, + "learning_rate": 7.23100922559636e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20690 + }, + { + "epoch": 6.022694210066919, + "flow_matching_loss": 0.014520961977541446, + "grad_norm": 0.08841076493263245, + "learning_rate": 7.216869814524323e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20700 + }, + { + "epoch": 6.025603724178063, + "flow_matching_loss": 0.013761247135698795, + "grad_norm": 0.0628279596567154, + "learning_rate": 7.202739860689357e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20710 + }, + { + "epoch": 6.0285132382892055, + "flow_matching_loss": 0.012860330194234848, + "grad_norm": 0.07391525059938431, + "learning_rate": 7.188619381260698e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20720 + }, + { + "epoch": 6.031422752400349, + "flow_matching_loss": 0.014246062375605106, + "grad_norm": 0.05804761126637459, + "learning_rate": 7.174508393396057e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20730 + }, + { + "epoch": 6.034332266511493, + "flow_matching_loss": 0.014035649597644806, + "grad_norm": 0.08102737367153168, + "learning_rate": 7.160406914241615e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20740 + }, + { + "epoch": 6.037241780622636, + "flow_matching_loss": 0.013018817361444234, + "grad_norm": 0.07982810586690903, + "learning_rate": 7.146314960931996e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20750 + }, + { + "epoch": 6.040151294733779, + "flow_matching_loss": 0.013623427972197533, + "grad_norm": 0.10457555204629898, + "learning_rate": 7.132232550590247e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20760 + }, + { + "epoch": 6.043060808844923, + "flow_matching_loss": 0.013978890888392926, + "grad_norm": 0.06905245780944824, + "learning_rate": 7.118159700327839e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20770 + }, + { + "epoch": 6.045970322956066, + "flow_matching_loss": 0.013053791597485542, + "grad_norm": 0.06040935963392258, + "learning_rate": 7.104096427244603e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20780 + }, + { + "epoch": 6.04887983706721, + "flow_matching_loss": 0.013516444247215986, + "grad_norm": 0.07075101137161255, + "learning_rate": 7.090042748428737e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20790 + }, + { + "epoch": 6.051789351178353, + "flow_matching_loss": 0.01284086937084794, + "grad_norm": 0.07539436221122742, + "learning_rate": 7.075998680956797e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20800 + }, + { + "epoch": 6.054698865289497, + "flow_matching_loss": 0.01232607001438737, + "grad_norm": 0.06344272196292877, + "learning_rate": 7.061964241893639e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20810 + }, + { + "epoch": 6.05760837940064, + "flow_matching_loss": 0.014897357299923896, + "grad_norm": 0.07526150345802307, + "learning_rate": 7.047939448292442e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20820 + }, + { + "epoch": 6.060517893511784, + "flow_matching_loss": 0.012764256540685892, + "grad_norm": 0.06881296634674072, + "learning_rate": 7.033924317194644e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20830 + }, + { + "epoch": 6.063427407622927, + "flow_matching_loss": 0.013440481945872306, + "grad_norm": 0.06596582382917404, + "learning_rate": 7.0199188656299524e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20840 + }, + { + "epoch": 6.06633692173407, + "flow_matching_loss": 0.012699814140796661, + "grad_norm": 0.07813560962677002, + "learning_rate": 7.005923110616313e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20850 + }, + { + "epoch": 6.069246435845214, + "flow_matching_loss": 0.013788906764239073, + "grad_norm": 0.08119695633649826, + "learning_rate": 6.9919370691598855e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20860 + }, + { + "epoch": 6.0721559499563575, + "flow_matching_loss": 0.012904537934809923, + "grad_norm": 0.062298137694597244, + "learning_rate": 6.977960758255024e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20870 + }, + { + "epoch": 6.0750654640675, + "flow_matching_loss": 0.013342053163796663, + "grad_norm": 0.07072030752897263, + "learning_rate": 6.963994194884276e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20880 + }, + { + "epoch": 6.077974978178644, + "flow_matching_loss": 0.014421308133751153, + "grad_norm": 0.08403238654136658, + "learning_rate": 6.950037396018328e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20890 + }, + { + "epoch": 6.080884492289788, + "flow_matching_loss": 0.01406254032626748, + "grad_norm": 0.05878273397684097, + "learning_rate": 6.9360903786160056e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20900 + }, + { + "epoch": 6.083794006400931, + "flow_matching_loss": 0.01279848925769329, + "grad_norm": 0.05908052623271942, + "learning_rate": 6.9221531596242485e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20910 + }, + { + "epoch": 6.086703520512074, + "flow_matching_loss": 0.013967563072219491, + "grad_norm": 0.07026775181293488, + "learning_rate": 6.908225755978089e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20920 + }, + { + "epoch": 6.089613034623218, + "flow_matching_loss": 0.01340379510074854, + "grad_norm": 0.07203994691371918, + "learning_rate": 6.8943081846006466e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20930 + }, + { + "epoch": 6.092522548734362, + "flow_matching_loss": 0.014152648579329252, + "grad_norm": 0.09151970595121384, + "learning_rate": 6.88040046240307e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20940 + }, + { + "epoch": 6.0954320628455045, + "flow_matching_loss": 0.013432913646101951, + "grad_norm": 0.07707943767309189, + "learning_rate": 6.866502606284566e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20950 + }, + { + "epoch": 6.098341576956648, + "flow_matching_loss": 0.013512122631072997, + "grad_norm": 0.06905467808246613, + "learning_rate": 6.852614633132335e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20960 + }, + { + "epoch": 6.101251091067792, + "flow_matching_loss": 0.014145964104682208, + "grad_norm": 0.075456902384758, + "learning_rate": 6.8387365598215665e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20970 + }, + { + "epoch": 6.104160605178935, + "flow_matching_loss": 0.014378399308770894, + "grad_norm": 0.07432951778173447, + "learning_rate": 6.82486840321544e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20980 + }, + { + "epoch": 6.1070701192900785, + "flow_matching_loss": 0.015034446585923433, + "grad_norm": 0.07796572893857956, + "learning_rate": 6.811010180165069e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 20990 + }, + { + "epoch": 6.109979633401222, + "flow_matching_loss": 0.012520063109695912, + "grad_norm": 0.08578699082136154, + "learning_rate": 6.797161907509499e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21000 + }, + { + "epoch": 6.112889147512365, + "flow_matching_loss": 0.012584083434194326, + "grad_norm": 0.07925533503293991, + "learning_rate": 6.783323602075688e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21010 + }, + { + "epoch": 6.115798661623509, + "flow_matching_loss": 0.012253145501017571, + "grad_norm": 0.08869442343711853, + "learning_rate": 6.7694952806784784e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21020 + }, + { + "epoch": 6.118708175734652, + "flow_matching_loss": 0.013936540205031633, + "grad_norm": 0.0829630047082901, + "learning_rate": 6.7556769601205944e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21030 + }, + { + "epoch": 6.121617689845796, + "flow_matching_loss": 0.01557383881881833, + "grad_norm": 0.06480823457241058, + "learning_rate": 6.741868657192594e-06, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21040 + }, + { + "epoch": 6.124527203956939, + "flow_matching_loss": 0.013954813312739133, + "grad_norm": 0.08003394305706024, + "learning_rate": 6.728070388672864e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21050 + }, + { + "epoch": 6.127436718068083, + "flow_matching_loss": 0.013074188027530909, + "grad_norm": 0.06920632719993591, + "learning_rate": 6.714282171327604e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21060 + }, + { + "epoch": 6.130346232179226, + "flow_matching_loss": 0.013159275893121958, + "grad_norm": 0.07120053470134735, + "learning_rate": 6.700504021910791e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21070 + }, + { + "epoch": 6.133255746290369, + "flow_matching_loss": 0.012831538729369641, + "grad_norm": 0.04894782975316048, + "learning_rate": 6.68673595716419e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21080 + }, + { + "epoch": 6.136165260401513, + "flow_matching_loss": 0.014128577802330256, + "grad_norm": 0.06352882832288742, + "learning_rate": 6.6729779938172856e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21090 + }, + { + "epoch": 6.139074774512657, + "flow_matching_loss": 0.013059527147561312, + "grad_norm": 0.089161217212677, + "learning_rate": 6.659230148587311e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21100 + }, + { + "epoch": 6.141984288623799, + "flow_matching_loss": 0.012751314416527748, + "grad_norm": 0.08149219304323196, + "learning_rate": 6.645492438179188e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21110 + }, + { + "epoch": 6.144893802734943, + "flow_matching_loss": 0.012835640367120505, + "grad_norm": 0.055885475128889084, + "learning_rate": 6.631764879285531e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21120 + }, + { + "epoch": 6.147803316846087, + "flow_matching_loss": 0.013199290819466114, + "grad_norm": 0.09167298674583435, + "learning_rate": 6.618047488586614e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21130 + }, + { + "epoch": 6.1507128309572305, + "flow_matching_loss": 0.014526690077036618, + "grad_norm": 0.06155386567115784, + "learning_rate": 6.604340282750371e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21140 + }, + { + "epoch": 6.153622345068373, + "flow_matching_loss": 0.01342006828635931, + "grad_norm": 0.0793628916144371, + "learning_rate": 6.590643278432342e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21150 + }, + { + "epoch": 6.156531859179517, + "flow_matching_loss": 0.013493011053651572, + "grad_norm": 0.06371542811393738, + "learning_rate": 6.576956492275682e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21160 + }, + { + "epoch": 6.159441373290661, + "flow_matching_loss": 0.014578517433255911, + "grad_norm": 0.06251689046621323, + "learning_rate": 6.563279940911124e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21170 + }, + { + "epoch": 6.162350887401804, + "flow_matching_loss": 0.011952918861061335, + "grad_norm": 0.06362323462963104, + "learning_rate": 6.549613640956963e-06, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21180 + }, + { + "epoch": 6.165260401512947, + "flow_matching_loss": 0.014548214431852103, + "grad_norm": 0.08182073384523392, + "learning_rate": 6.535957609019053e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21190 + }, + { + "epoch": 6.168169915624091, + "flow_matching_loss": 0.01357681406661868, + "grad_norm": 0.05403905361890793, + "learning_rate": 6.5223118616907545e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21200 + }, + { + "epoch": 6.171079429735234, + "flow_matching_loss": 0.01395200053229928, + "grad_norm": 0.07819663733243942, + "learning_rate": 6.508676415552939e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21210 + }, + { + "epoch": 6.1739889438463775, + "flow_matching_loss": 0.014665459468960761, + "grad_norm": 0.07384762912988663, + "learning_rate": 6.495051287173952e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21220 + }, + { + "epoch": 6.176898457957521, + "flow_matching_loss": 0.013386011682450771, + "grad_norm": 0.06337407231330872, + "learning_rate": 6.481436493109621e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21230 + }, + { + "epoch": 6.179807972068665, + "flow_matching_loss": 0.014396184496581555, + "grad_norm": 0.06983184069395065, + "learning_rate": 6.467832049903194e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21240 + }, + { + "epoch": 6.182717486179808, + "flow_matching_loss": 0.01518079824745655, + "grad_norm": 0.08792704343795776, + "learning_rate": 6.454237974085361e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21250 + }, + { + "epoch": 6.1856270002909515, + "flow_matching_loss": 0.012457642331719398, + "grad_norm": 0.0681145116686821, + "learning_rate": 6.440654282174203e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21260 + }, + { + "epoch": 6.188536514402095, + "flow_matching_loss": 0.014071778766810894, + "grad_norm": 0.08124446868896484, + "learning_rate": 6.427080990675187e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21270 + }, + { + "epoch": 6.191446028513238, + "flow_matching_loss": 0.01398466075770557, + "grad_norm": 0.08365116268396378, + "learning_rate": 6.41351811608114e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21280 + }, + { + "epoch": 6.194355542624382, + "flow_matching_loss": 0.013002427946776151, + "grad_norm": 0.06582597643136978, + "learning_rate": 6.399965674872228e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21290 + }, + { + "epoch": 6.197265056735525, + "flow_matching_loss": 0.014462905190885066, + "grad_norm": 0.07184753566980362, + "learning_rate": 6.386423683515957e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21300 + }, + { + "epoch": 6.200174570846668, + "flow_matching_loss": 0.01407859562896192, + "grad_norm": 0.09338876605033875, + "learning_rate": 6.372892158467119e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21310 + }, + { + "epoch": 6.203084084957812, + "flow_matching_loss": 0.014183247555047273, + "grad_norm": 0.0934266522526741, + "learning_rate": 6.359371116167791e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21320 + }, + { + "epoch": 6.205993599068956, + "flow_matching_loss": 0.012562664225697517, + "grad_norm": 0.0759960487484932, + "learning_rate": 6.345860573047314e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21330 + }, + { + "epoch": 6.208903113180099, + "flow_matching_loss": 0.01439686557278037, + "grad_norm": 0.07659218460321426, + "learning_rate": 6.332360545522268e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21340 + }, + { + "epoch": 6.211812627291242, + "flow_matching_loss": 0.014106225594878197, + "grad_norm": 0.09161235392093658, + "learning_rate": 6.31887104999647e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21350 + }, + { + "epoch": 6.214722141402386, + "flow_matching_loss": 0.012935131508857011, + "grad_norm": 0.08210441470146179, + "learning_rate": 6.305392102860924e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21360 + }, + { + "epoch": 6.21763165551353, + "flow_matching_loss": 0.012861873675137758, + "grad_norm": 0.06348702311515808, + "learning_rate": 6.291923720493814e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21370 + }, + { + "epoch": 6.220541169624672, + "flow_matching_loss": 0.01403709566220641, + "grad_norm": 0.07163286209106445, + "learning_rate": 6.278465919260507e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21380 + }, + { + "epoch": 6.223450683735816, + "flow_matching_loss": 0.012616245355457068, + "grad_norm": 0.10804980248212814, + "learning_rate": 6.26501871551349e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21390 + }, + { + "epoch": 6.22636019784696, + "flow_matching_loss": 0.012905366905033588, + "grad_norm": 0.08127621561288834, + "learning_rate": 6.251582125592394e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21400 + }, + { + "epoch": 6.229269711958103, + "flow_matching_loss": 0.01336753936484456, + "grad_norm": 0.07373716682195663, + "learning_rate": 6.2381561658239386e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21410 + }, + { + "epoch": 6.232179226069246, + "flow_matching_loss": 0.012651353422552348, + "grad_norm": 0.10665709525346756, + "learning_rate": 6.2247408525219276e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21420 + }, + { + "epoch": 6.23508874018039, + "flow_matching_loss": 0.01331365630030632, + "grad_norm": 0.06875603646039963, + "learning_rate": 6.2113362019872335e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21430 + }, + { + "epoch": 6.237998254291533, + "flow_matching_loss": 0.012810017354786396, + "grad_norm": 0.06800201535224915, + "learning_rate": 6.197942230507772e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21440 + }, + { + "epoch": 6.240907768402677, + "flow_matching_loss": 0.013947608321905136, + "grad_norm": 0.07994898408651352, + "learning_rate": 6.184558954358475e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21450 + }, + { + "epoch": 6.24381728251382, + "flow_matching_loss": 0.01400615433230996, + "grad_norm": 0.08925843983888626, + "learning_rate": 6.1711863898012965e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21460 + }, + { + "epoch": 6.246726796624964, + "flow_matching_loss": 0.013187952153384686, + "grad_norm": 0.0774691253900528, + "learning_rate": 6.157824553085155e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21470 + }, + { + "epoch": 6.249636310736107, + "flow_matching_loss": 0.013222523499280214, + "grad_norm": 0.06112616881728172, + "learning_rate": 6.144473460445945e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21480 + }, + { + "epoch": 6.2525458248472505, + "flow_matching_loss": 0.013364616129547358, + "grad_norm": 0.09023316949605942, + "learning_rate": 6.131133128106503e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21490 + }, + { + "epoch": 6.255455338958394, + "flow_matching_loss": 0.013325887639075518, + "grad_norm": 0.07730896770954132, + "learning_rate": 6.117803572276584e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21500 + }, + { + "epoch": 6.258364853069537, + "flow_matching_loss": 0.013505143113434315, + "grad_norm": 0.08541591465473175, + "learning_rate": 6.104484809152866e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21510 + }, + { + "epoch": 6.261274367180681, + "flow_matching_loss": 0.014753915183246136, + "grad_norm": 0.08976373076438904, + "learning_rate": 6.091176854918892e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21520 + }, + { + "epoch": 6.2641838812918245, + "flow_matching_loss": 0.014273176621645688, + "grad_norm": 0.06672367453575134, + "learning_rate": 6.077879725745092e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21530 + }, + { + "epoch": 6.267093395402968, + "flow_matching_loss": 0.014067056216299533, + "grad_norm": 0.06204155087471008, + "learning_rate": 6.064593437788727e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21540 + }, + { + "epoch": 6.270002909514111, + "flow_matching_loss": 0.012958420999348163, + "grad_norm": 0.07347016781568527, + "learning_rate": 6.051318007193881e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21550 + }, + { + "epoch": 6.272912423625255, + "flow_matching_loss": 0.013255035411566496, + "grad_norm": 0.08531838655471802, + "learning_rate": 6.038053450091469e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21560 + }, + { + "epoch": 6.275821937736398, + "flow_matching_loss": 0.013811799231916666, + "grad_norm": 0.07470180839300156, + "learning_rate": 6.0247997825991705e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21570 + }, + { + "epoch": 6.278731451847541, + "flow_matching_loss": 0.013600754830986262, + "grad_norm": 0.07131385803222656, + "learning_rate": 6.011557020821444e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21580 + }, + { + "epoch": 6.281640965958685, + "flow_matching_loss": 0.01325705014169216, + "grad_norm": 0.05311321094632149, + "learning_rate": 5.99832518084949e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21590 + }, + { + "epoch": 6.284550480069829, + "flow_matching_loss": 0.014485093299299479, + "grad_norm": 0.0680558905005455, + "learning_rate": 5.98510427876124e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21600 + }, + { + "epoch": 6.2874599941809715, + "flow_matching_loss": 0.013706157077103853, + "grad_norm": 0.07896234840154648, + "learning_rate": 5.971894330621344e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21610 + }, + { + "epoch": 6.290369508292115, + "flow_matching_loss": 0.01459311619400978, + "grad_norm": 0.062067266553640366, + "learning_rate": 5.958695352481134e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21620 + }, + { + "epoch": 6.293279022403259, + "flow_matching_loss": 0.01266593225300312, + "grad_norm": 0.0762583315372467, + "learning_rate": 5.945507360378608e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21630 + }, + { + "epoch": 6.296188536514402, + "flow_matching_loss": 0.014999859035015106, + "grad_norm": 0.0864642858505249, + "learning_rate": 5.9323303703384225e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21640 + }, + { + "epoch": 6.299098050625545, + "flow_matching_loss": 0.013467316515743732, + "grad_norm": 0.10498881340026855, + "learning_rate": 5.919164398371857e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21650 + }, + { + "epoch": 6.302007564736689, + "flow_matching_loss": 0.01315827462822199, + "grad_norm": 0.08467771112918854, + "learning_rate": 5.906009460476818e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21660 + }, + { + "epoch": 6.304917078847833, + "flow_matching_loss": 0.014299181196838618, + "grad_norm": 0.07239816337823868, + "learning_rate": 5.892865572637789e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21670 + }, + { + "epoch": 6.307826592958976, + "flow_matching_loss": 0.013373817130923272, + "grad_norm": 0.06359751522541046, + "learning_rate": 5.8797327508258396e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21680 + }, + { + "epoch": 6.310736107070119, + "flow_matching_loss": 0.013608177565038205, + "grad_norm": 0.05801743268966675, + "learning_rate": 5.866611010998584e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21690 + }, + { + "epoch": 6.313645621181263, + "flow_matching_loss": 0.013848559930920601, + "grad_norm": 0.06648078560829163, + "learning_rate": 5.853500369100172e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21700 + }, + { + "epoch": 6.316555135292406, + "flow_matching_loss": 0.01289151906967163, + "grad_norm": 0.06789704412221909, + "learning_rate": 5.840400841061265e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21710 + }, + { + "epoch": 6.31946464940355, + "flow_matching_loss": 0.012534800916910172, + "grad_norm": 0.07673487067222595, + "learning_rate": 5.827312442799034e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21720 + }, + { + "epoch": 6.322374163514693, + "flow_matching_loss": 0.011932060308754443, + "grad_norm": 0.0708501935005188, + "learning_rate": 5.814235190217114e-06, + "loss": 0.0119, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21730 + }, + { + "epoch": 6.325283677625836, + "flow_matching_loss": 0.012880152091383934, + "grad_norm": 0.09044141322374344, + "learning_rate": 5.801169099205599e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21740 + }, + { + "epoch": 6.32819319173698, + "flow_matching_loss": 0.01305748987942934, + "grad_norm": 0.08360414206981659, + "learning_rate": 5.788114185641022e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21750 + }, + { + "epoch": 6.3311027058481235, + "flow_matching_loss": 0.01337619349360466, + "grad_norm": 0.06152678653597832, + "learning_rate": 5.775070465386328e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21760 + }, + { + "epoch": 6.334012219959266, + "flow_matching_loss": 0.01417469447478652, + "grad_norm": 0.10867657512426376, + "learning_rate": 5.7620379542908765e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21770 + }, + { + "epoch": 6.33692173407041, + "flow_matching_loss": 0.013875481858849525, + "grad_norm": 0.06831344962120056, + "learning_rate": 5.749016668190395e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21780 + }, + { + "epoch": 6.339831248181554, + "flow_matching_loss": 0.014637913275510072, + "grad_norm": 0.0568140484392643, + "learning_rate": 5.736006622906973e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21790 + }, + { + "epoch": 6.3427407622926975, + "flow_matching_loss": 0.014577757753431796, + "grad_norm": 0.05983157828450203, + "learning_rate": 5.723007834249034e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21800 + }, + { + "epoch": 6.34565027640384, + "flow_matching_loss": 0.013965468388050795, + "grad_norm": 0.05902672931551933, + "learning_rate": 5.710020318011346e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21810 + }, + { + "epoch": 6.348559790514984, + "flow_matching_loss": 0.01444997675716877, + "grad_norm": 0.07469973713159561, + "learning_rate": 5.697044089974955e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21820 + }, + { + "epoch": 6.351469304626128, + "flow_matching_loss": 0.014425041899085046, + "grad_norm": 0.058961521834135056, + "learning_rate": 5.68407916590721e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21830 + }, + { + "epoch": 6.3543788187372705, + "flow_matching_loss": 0.012464318796992302, + "grad_norm": 0.06402431428432465, + "learning_rate": 5.671125561561713e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21840 + }, + { + "epoch": 6.357288332848414, + "flow_matching_loss": 0.014658653642982244, + "grad_norm": 0.06572264432907104, + "learning_rate": 5.658183292678315e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21850 + }, + { + "epoch": 6.360197846959558, + "flow_matching_loss": 0.012549732904881239, + "grad_norm": 0.054180342704057693, + "learning_rate": 5.645252374983091e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21860 + }, + { + "epoch": 6.363107361070702, + "flow_matching_loss": 0.014381828717887402, + "grad_norm": 0.05498868227005005, + "learning_rate": 5.6323328241883224e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21870 + }, + { + "epoch": 6.3660168751818444, + "flow_matching_loss": 0.013947288691997527, + "grad_norm": 0.09776416420936584, + "learning_rate": 5.61942465599249e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21880 + }, + { + "epoch": 6.368926389292988, + "flow_matching_loss": 0.013974087685346604, + "grad_norm": 0.08617467433214188, + "learning_rate": 5.60652788608023e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21890 + }, + { + "epoch": 6.371835903404132, + "flow_matching_loss": 0.012676784163340926, + "grad_norm": 0.06929370760917664, + "learning_rate": 5.593642530122336e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21900 + }, + { + "epoch": 6.374745417515275, + "flow_matching_loss": 0.015684251580387353, + "grad_norm": 0.0818684920668602, + "learning_rate": 5.58076860377573e-06, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21910 + }, + { + "epoch": 6.377654931626418, + "flow_matching_loss": 0.013344424497336149, + "grad_norm": 0.08244957029819489, + "learning_rate": 5.567906122683439e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21920 + }, + { + "epoch": 6.380564445737562, + "flow_matching_loss": 0.014805641956627369, + "grad_norm": 0.08186900615692139, + "learning_rate": 5.555055102474604e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21930 + }, + { + "epoch": 6.383473959848705, + "flow_matching_loss": 0.013509374391287566, + "grad_norm": 0.06675300747156143, + "learning_rate": 5.542215558764421e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21940 + }, + { + "epoch": 6.386383473959849, + "flow_matching_loss": 0.013714950624853372, + "grad_norm": 0.05708472430706024, + "learning_rate": 5.529387507154138e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21950 + }, + { + "epoch": 6.389292988070992, + "flow_matching_loss": 0.014757471345365047, + "grad_norm": 0.0721307247877121, + "learning_rate": 5.516570963231058e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21960 + }, + { + "epoch": 6.392202502182135, + "flow_matching_loss": 0.0133464558981359, + "grad_norm": 0.07982407510280609, + "learning_rate": 5.503765942568484e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21970 + }, + { + "epoch": 6.395112016293279, + "flow_matching_loss": 0.013324925862252712, + "grad_norm": 0.07761149108409882, + "learning_rate": 5.49097246072573e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21980 + }, + { + "epoch": 6.398021530404423, + "flow_matching_loss": 0.013628732413053513, + "grad_norm": 0.08745390176773071, + "learning_rate": 5.4781905332480805e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 21990 + }, + { + "epoch": 6.400931044515566, + "flow_matching_loss": 0.013379075285047292, + "grad_norm": 0.0676802396774292, + "learning_rate": 5.4654201756667795e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22000 + }, + { + "epoch": 6.403840558626709, + "flow_matching_loss": 0.012372338585555554, + "grad_norm": 0.06918632984161377, + "learning_rate": 5.452661403499017e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22010 + }, + { + "epoch": 6.406750072737853, + "flow_matching_loss": 0.012910364009439946, + "grad_norm": 0.08009369671344757, + "learning_rate": 5.4399142322479005e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22020 + }, + { + "epoch": 6.4096595868489965, + "flow_matching_loss": 0.013682690355926752, + "grad_norm": 0.0697103887796402, + "learning_rate": 5.427178677402452e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22030 + }, + { + "epoch": 6.412569100960139, + "flow_matching_loss": 0.013462392427027226, + "grad_norm": 0.09547895938158035, + "learning_rate": 5.41445475443757e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22040 + }, + { + "epoch": 6.415478615071283, + "flow_matching_loss": 0.013490174245089293, + "grad_norm": 0.07074116915464401, + "learning_rate": 5.4017424788140194e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22050 + }, + { + "epoch": 6.418388129182427, + "flow_matching_loss": 0.014390055648982525, + "grad_norm": 0.05957217514514923, + "learning_rate": 5.389041865978414e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22060 + }, + { + "epoch": 6.4212976432935704, + "flow_matching_loss": 0.013776670210063457, + "grad_norm": 0.06613312661647797, + "learning_rate": 5.37635293136319e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22070 + }, + { + "epoch": 6.424207157404713, + "flow_matching_loss": 0.014812473487108947, + "grad_norm": 0.10950074344873428, + "learning_rate": 5.363675690386609e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22080 + }, + { + "epoch": 6.427116671515857, + "flow_matching_loss": 0.014358712639659644, + "grad_norm": 0.059068094938993454, + "learning_rate": 5.351010158452708e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22090 + }, + { + "epoch": 6.430026185627001, + "flow_matching_loss": 0.013437368627637624, + "grad_norm": 0.08689611405134201, + "learning_rate": 5.338356350951308e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22100 + }, + { + "epoch": 6.4329356997381435, + "flow_matching_loss": 0.012738493084907532, + "grad_norm": 0.08115945756435394, + "learning_rate": 5.325714283257974e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22110 + }, + { + "epoch": 6.435845213849287, + "flow_matching_loss": 0.014465564768761397, + "grad_norm": 0.06740902364253998, + "learning_rate": 5.313083970734013e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22120 + }, + { + "epoch": 6.438754727960431, + "flow_matching_loss": 0.0163935667835176, + "grad_norm": 0.08645253628492355, + "learning_rate": 5.300465428726439e-06, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22130 + }, + { + "epoch": 6.441664242071574, + "flow_matching_loss": 0.013616079930216074, + "grad_norm": 0.06534223258495331, + "learning_rate": 5.287858672567979e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22140 + }, + { + "epoch": 6.444573756182717, + "flow_matching_loss": 0.013818234950304032, + "grad_norm": 0.0572090670466423, + "learning_rate": 5.275263717577028e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22150 + }, + { + "epoch": 6.447483270293861, + "flow_matching_loss": 0.012545681186020374, + "grad_norm": 0.07186184823513031, + "learning_rate": 5.2626805790576435e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22160 + }, + { + "epoch": 6.450392784405004, + "flow_matching_loss": 0.013784188590943814, + "grad_norm": 0.06590911746025085, + "learning_rate": 5.250109272299523e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22170 + }, + { + "epoch": 6.453302298516148, + "flow_matching_loss": 0.012599368393421174, + "grad_norm": 0.09146437048912048, + "learning_rate": 5.237549812577987e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22180 + }, + { + "epoch": 6.456211812627291, + "flow_matching_loss": 0.014612434897571802, + "grad_norm": 0.08692826330661774, + "learning_rate": 5.225002215153973e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22190 + }, + { + "epoch": 6.459121326738435, + "flow_matching_loss": 0.013037483114749194, + "grad_norm": 0.06001284718513489, + "learning_rate": 5.21246649527399e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22200 + }, + { + "epoch": 6.462030840849578, + "flow_matching_loss": 0.012439053039997816, + "grad_norm": 0.06237281858921051, + "learning_rate": 5.199942668170118e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22210 + }, + { + "epoch": 6.464940354960722, + "flow_matching_loss": 0.013310950435698032, + "grad_norm": 0.06958480924367905, + "learning_rate": 5.187430749059986e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22220 + }, + { + "epoch": 6.467849869071865, + "flow_matching_loss": 0.01437306934967637, + "grad_norm": 0.07408503443002701, + "learning_rate": 5.174930753146763e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22230 + }, + { + "epoch": 6.470759383183008, + "flow_matching_loss": 0.012378826271742582, + "grad_norm": 0.06287696212530136, + "learning_rate": 5.162442695619114e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22240 + }, + { + "epoch": 6.473668897294152, + "flow_matching_loss": 0.012902287859469652, + "grad_norm": 0.08866557478904724, + "learning_rate": 5.149966591651215e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22250 + }, + { + "epoch": 6.4765784114052956, + "flow_matching_loss": 0.014118932094424963, + "grad_norm": 0.07543319463729858, + "learning_rate": 5.137502456402705e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22260 + }, + { + "epoch": 6.479487925516438, + "flow_matching_loss": 0.014378662779927253, + "grad_norm": 0.07786840200424194, + "learning_rate": 5.125050305018683e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22270 + }, + { + "epoch": 6.482397439627582, + "flow_matching_loss": 0.013949399814009666, + "grad_norm": 0.06162134185433388, + "learning_rate": 5.112610152629687e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22280 + }, + { + "epoch": 6.485306953738726, + "flow_matching_loss": 0.014580365642905236, + "grad_norm": 0.06187082827091217, + "learning_rate": 5.10018201435167e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22290 + }, + { + "epoch": 6.488216467849869, + "flow_matching_loss": 0.011973320227116346, + "grad_norm": 0.0684242695569992, + "learning_rate": 5.087765905286004e-06, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22300 + }, + { + "epoch": 6.491125981961012, + "flow_matching_loss": 0.012989166751503944, + "grad_norm": 0.08950722962617874, + "learning_rate": 5.0753618405194245e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22310 + }, + { + "epoch": 6.494035496072156, + "flow_matching_loss": 0.013042198866605759, + "grad_norm": 0.05878158658742905, + "learning_rate": 5.062969835124042e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22320 + }, + { + "epoch": 6.4969450101833, + "flow_matching_loss": 0.013409404084086418, + "grad_norm": 0.0706595927476883, + "learning_rate": 5.05058990415731e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22330 + }, + { + "epoch": 6.4998545242944425, + "flow_matching_loss": 0.013332903943955898, + "grad_norm": 0.07735515385866165, + "learning_rate": 5.038222062662007e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22340 + }, + { + "epoch": 6.502764038405586, + "flow_matching_loss": 0.015594257693737745, + "grad_norm": 0.08529850840568542, + "learning_rate": 5.025866325666237e-06, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22350 + }, + { + "epoch": 6.50567355251673, + "flow_matching_loss": 0.014552695769816637, + "grad_norm": 0.07035086303949356, + "learning_rate": 5.013522708183383e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22360 + }, + { + "epoch": 6.508583066627873, + "flow_matching_loss": 0.014857903495430947, + "grad_norm": 0.07600998133420944, + "learning_rate": 5.001191225212096e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22370 + }, + { + "epoch": 6.5114925807390165, + "flow_matching_loss": 0.012877288274466991, + "grad_norm": 0.08999856561422348, + "learning_rate": 4.9888718917363e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22380 + }, + { + "epoch": 6.51440209485016, + "flow_matching_loss": 0.012458019005134702, + "grad_norm": 0.07064348459243774, + "learning_rate": 4.9765647227251395e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22390 + }, + { + "epoch": 6.517311608961304, + "flow_matching_loss": 0.012317993957549334, + "grad_norm": 0.06244400888681412, + "learning_rate": 4.964269733132992e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22400 + }, + { + "epoch": 6.520221123072447, + "flow_matching_loss": 0.013677875138819217, + "grad_norm": 0.06266000121831894, + "learning_rate": 4.951986937899428e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22410 + }, + { + "epoch": 6.52313063718359, + "flow_matching_loss": 0.01320616789162159, + "grad_norm": 0.07671651244163513, + "learning_rate": 4.9397163519492e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22420 + }, + { + "epoch": 6.526040151294734, + "flow_matching_loss": 0.013066184241324663, + "grad_norm": 0.06471577286720276, + "learning_rate": 4.9274579901922275e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22430 + }, + { + "epoch": 6.528949665405877, + "flow_matching_loss": 0.012017757818102836, + "grad_norm": 0.07541292160749435, + "learning_rate": 4.9152118675235705e-06, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22440 + }, + { + "epoch": 6.531859179517021, + "flow_matching_loss": 0.013801049906760454, + "grad_norm": 0.08353924751281738, + "learning_rate": 4.902977998823432e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22450 + }, + { + "epoch": 6.534768693628164, + "flow_matching_loss": 0.013974407128989697, + "grad_norm": 0.060540370643138885, + "learning_rate": 4.890756398957114e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22460 + }, + { + "epoch": 6.537678207739307, + "flow_matching_loss": 0.013436923641711474, + "grad_norm": 0.07175809890031815, + "learning_rate": 4.8785470827750075e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22470 + }, + { + "epoch": 6.540587721850451, + "flow_matching_loss": 0.013483384158462286, + "grad_norm": 0.06997458636760712, + "learning_rate": 4.866350065112588e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22480 + }, + { + "epoch": 6.543497235961595, + "flow_matching_loss": 0.011747473757714033, + "grad_norm": 0.08805360645055771, + "learning_rate": 4.854165360790381e-06, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22490 + }, + { + "epoch": 6.546406750072737, + "flow_matching_loss": 0.012278868863359094, + "grad_norm": 0.07783672213554382, + "learning_rate": 4.8419929846139425e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22500 + }, + { + "epoch": 6.549316264183881, + "flow_matching_loss": 0.014619407430291175, + "grad_norm": 0.08615543693304062, + "learning_rate": 4.829832951373872e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22510 + }, + { + "epoch": 6.552225778295025, + "flow_matching_loss": 0.013019076734781265, + "grad_norm": 0.07616326212882996, + "learning_rate": 4.817685275845746e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22520 + }, + { + "epoch": 6.5551352924061685, + "flow_matching_loss": 0.014324491005390882, + "grad_norm": 0.07834882289171219, + "learning_rate": 4.805549972790143e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22530 + }, + { + "epoch": 6.558044806517311, + "flow_matching_loss": 0.013236889429390431, + "grad_norm": 0.07489871978759766, + "learning_rate": 4.7934270569525966e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22540 + }, + { + "epoch": 6.560954320628455, + "flow_matching_loss": 0.01274797460064292, + "grad_norm": 0.08917434513568878, + "learning_rate": 4.7813165430635874e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22550 + }, + { + "epoch": 6.563863834739599, + "flow_matching_loss": 0.013948941975831986, + "grad_norm": 0.07994019985198975, + "learning_rate": 4.769218445838544e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22560 + }, + { + "epoch": 6.566773348850742, + "flow_matching_loss": 0.013383117970079184, + "grad_norm": 0.07774246484041214, + "learning_rate": 4.757132779977787e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22570 + }, + { + "epoch": 6.569682862961885, + "flow_matching_loss": 0.012692431826144457, + "grad_norm": 0.06253936886787415, + "learning_rate": 4.745059560166542e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22580 + }, + { + "epoch": 6.572592377073029, + "flow_matching_loss": 0.012691887281835078, + "grad_norm": 0.0599326528608799, + "learning_rate": 4.732998801074911e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22590 + }, + { + "epoch": 6.575501891184173, + "flow_matching_loss": 0.01280516739934683, + "grad_norm": 0.06588095426559448, + "learning_rate": 4.7209505173578445e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22600 + }, + { + "epoch": 6.5784114052953155, + "flow_matching_loss": 0.014797658193856478, + "grad_norm": 0.108277827501297, + "learning_rate": 4.7089147236551555e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22610 + }, + { + "epoch": 6.581320919406459, + "flow_matching_loss": 0.012664176430553198, + "grad_norm": 0.05974648892879486, + "learning_rate": 4.696891434591464e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22620 + }, + { + "epoch": 6.584230433517602, + "flow_matching_loss": 0.014306690264493226, + "grad_norm": 0.08135362714529037, + "learning_rate": 4.684880664776199e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22630 + }, + { + "epoch": 6.587139947628746, + "flow_matching_loss": 0.014345387928187847, + "grad_norm": 0.07090885937213898, + "learning_rate": 4.6728824288035784e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22640 + }, + { + "epoch": 6.5900494617398895, + "flow_matching_loss": 0.014230594597756862, + "grad_norm": 0.07416519522666931, + "learning_rate": 4.660896741252586e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22650 + }, + { + "epoch": 6.592958975851033, + "flow_matching_loss": 0.014115734305232764, + "grad_norm": 0.07003627717494965, + "learning_rate": 4.6489236166869714e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22660 + }, + { + "epoch": 6.595868489962176, + "flow_matching_loss": 0.01167108230292797, + "grad_norm": 0.07473664730787277, + "learning_rate": 4.6369630696552e-06, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22670 + }, + { + "epoch": 6.59877800407332, + "flow_matching_loss": 0.013322941493242979, + "grad_norm": 0.06458549946546555, + "learning_rate": 4.625015114690474e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22680 + }, + { + "epoch": 6.601687518184463, + "flow_matching_loss": 0.012455717101693154, + "grad_norm": 0.06708625704050064, + "learning_rate": 4.613079766310682e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22690 + }, + { + "epoch": 6.604597032295606, + "flow_matching_loss": 0.012809778098016977, + "grad_norm": 0.06664470583200455, + "learning_rate": 4.601157039018396e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22700 + }, + { + "epoch": 6.60750654640675, + "flow_matching_loss": 0.011927387677133083, + "grad_norm": 0.079779714345932, + "learning_rate": 4.589246947300849e-06, + "loss": 0.0119, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22710 + }, + { + "epoch": 6.610416060517894, + "flow_matching_loss": 0.012552298791706561, + "grad_norm": 0.06724656373262405, + "learning_rate": 4.577349505629936e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22720 + }, + { + "epoch": 6.613325574629037, + "flow_matching_loss": 0.013801949005573988, + "grad_norm": 0.08309617638587952, + "learning_rate": 4.565464728462165e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22730 + }, + { + "epoch": 6.61623508874018, + "flow_matching_loss": 0.014249459188431502, + "grad_norm": 0.07396794855594635, + "learning_rate": 4.553592630238663e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22740 + }, + { + "epoch": 6.619144602851324, + "flow_matching_loss": 0.012820951733738184, + "grad_norm": 0.08411047607660294, + "learning_rate": 4.541733225385146e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22750 + }, + { + "epoch": 6.622054116962468, + "flow_matching_loss": 0.012647601962089538, + "grad_norm": 0.09513940662145615, + "learning_rate": 4.529886528311905e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22760 + }, + { + "epoch": 6.62496363107361, + "flow_matching_loss": 0.013887646794319152, + "grad_norm": 0.08614011108875275, + "learning_rate": 4.5180525534138045e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22770 + }, + { + "epoch": 6.627873145184754, + "flow_matching_loss": 0.0144280762411654, + "grad_norm": 0.07358040660619736, + "learning_rate": 4.506231315070235e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22780 + }, + { + "epoch": 6.630782659295898, + "flow_matching_loss": 0.013163793366402388, + "grad_norm": 0.07864868640899658, + "learning_rate": 4.494422827645115e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22790 + }, + { + "epoch": 6.633692173407041, + "flow_matching_loss": 0.012885838840156793, + "grad_norm": 0.0875745415687561, + "learning_rate": 4.482627105486865e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22800 + }, + { + "epoch": 6.636601687518184, + "flow_matching_loss": 0.012179369479417801, + "grad_norm": 0.05601033940911293, + "learning_rate": 4.470844162928412e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22810 + }, + { + "epoch": 6.639511201629328, + "flow_matching_loss": 0.013040865957736968, + "grad_norm": 0.07324472069740295, + "learning_rate": 4.459074014287129e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22820 + }, + { + "epoch": 6.642420715740471, + "flow_matching_loss": 0.01375096058472991, + "grad_norm": 0.061665039509534836, + "learning_rate": 4.447316673864869e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22830 + }, + { + "epoch": 6.645330229851615, + "flow_matching_loss": 0.012608497217297555, + "grad_norm": 0.06747863441705704, + "learning_rate": 4.435572155947903e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22840 + }, + { + "epoch": 6.648239743962758, + "flow_matching_loss": 0.013383533526211976, + "grad_norm": 0.07583832740783691, + "learning_rate": 4.423840474806929e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22850 + }, + { + "epoch": 6.651149258073902, + "flow_matching_loss": 0.013770818710327148, + "grad_norm": 0.06424327939748764, + "learning_rate": 4.412121644697043e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22860 + }, + { + "epoch": 6.654058772185045, + "flow_matching_loss": 0.012279549334198236, + "grad_norm": 0.07237479835748672, + "learning_rate": 4.400415679857727e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22870 + }, + { + "epoch": 6.6569682862961885, + "flow_matching_loss": 0.013779634702950716, + "grad_norm": 0.10594811290502548, + "learning_rate": 4.388722594512838e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22880 + }, + { + "epoch": 6.659877800407332, + "flow_matching_loss": 0.013413378689438104, + "grad_norm": 0.07461182028055191, + "learning_rate": 4.377042402870577e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22890 + }, + { + "epoch": 6.662787314518475, + "flow_matching_loss": 0.012193301040679217, + "grad_norm": 0.07149967551231384, + "learning_rate": 4.365375119123476e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22900 + }, + { + "epoch": 6.665696828629619, + "flow_matching_loss": 0.012400860339403153, + "grad_norm": 0.06420587748289108, + "learning_rate": 4.353720757448383e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22910 + }, + { + "epoch": 6.6686063427407625, + "flow_matching_loss": 0.0127859428524971, + "grad_norm": 0.08046424388885498, + "learning_rate": 4.342079332006445e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22920 + }, + { + "epoch": 6.671515856851906, + "flow_matching_loss": 0.013156667817384005, + "grad_norm": 0.07746373862028122, + "learning_rate": 4.330450856943101e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22930 + }, + { + "epoch": 6.674425370963049, + "flow_matching_loss": 0.013050335831940174, + "grad_norm": 0.07540087401866913, + "learning_rate": 4.318835346388036e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22940 + }, + { + "epoch": 6.677334885074193, + "flow_matching_loss": 0.01263108616694808, + "grad_norm": 0.0715625062584877, + "learning_rate": 4.307232814455189e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22950 + }, + { + "epoch": 6.680244399185336, + "flow_matching_loss": 0.013175473175942897, + "grad_norm": 0.06669336557388306, + "learning_rate": 4.29564327524274e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22960 + }, + { + "epoch": 6.683153913296479, + "flow_matching_loss": 0.01316832723096013, + "grad_norm": 0.0659121423959732, + "learning_rate": 4.284066742833061e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22970 + }, + { + "epoch": 6.686063427407623, + "flow_matching_loss": 0.012881690915673972, + "grad_norm": 0.08752994984388351, + "learning_rate": 4.272503231292741e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22980 + }, + { + "epoch": 6.688972941518767, + "flow_matching_loss": 0.014578353054821492, + "grad_norm": 0.09754519909620285, + "learning_rate": 4.260952754672532e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 22990 + }, + { + "epoch": 6.6918824556299095, + "flow_matching_loss": 0.014796303771436214, + "grad_norm": 0.08505729585886002, + "learning_rate": 4.249415327007353e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23000 + }, + { + "epoch": 6.694791969741053, + "flow_matching_loss": 0.013581185881048442, + "grad_norm": 0.09604617953300476, + "learning_rate": 4.237890962316264e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23010 + }, + { + "epoch": 6.697701483852197, + "flow_matching_loss": 0.013079790864139795, + "grad_norm": 0.08292097598314285, + "learning_rate": 4.226379674602452e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23020 + }, + { + "epoch": 6.70061099796334, + "flow_matching_loss": 0.013464473187923431, + "grad_norm": 0.08257924020290375, + "learning_rate": 4.214881477853227e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23030 + }, + { + "epoch": 6.703520512074483, + "flow_matching_loss": 0.013366041425615548, + "grad_norm": 0.07382367551326752, + "learning_rate": 4.2033963860399745e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23040 + }, + { + "epoch": 6.706430026185627, + "flow_matching_loss": 0.01186885442584753, + "grad_norm": 0.055135373026132584, + "learning_rate": 4.1919244131181696e-06, + "loss": 0.0119, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23050 + }, + { + "epoch": 6.709339540296771, + "flow_matching_loss": 0.013424212764948606, + "grad_norm": 0.06911183893680573, + "learning_rate": 4.180465573027335e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23060 + }, + { + "epoch": 6.712249054407914, + "flow_matching_loss": 0.011955693410709501, + "grad_norm": 0.07754072546958923, + "learning_rate": 4.169019879691046e-06, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23070 + }, + { + "epoch": 6.715158568519057, + "flow_matching_loss": 0.01323012774810195, + "grad_norm": 0.08831287175416946, + "learning_rate": 4.157587347016891e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23080 + }, + { + "epoch": 6.718068082630201, + "flow_matching_loss": 0.013057221658527851, + "grad_norm": 0.07804005593061447, + "learning_rate": 4.14616798889649e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23090 + }, + { + "epoch": 6.720977596741344, + "flow_matching_loss": 0.011813301593065262, + "grad_norm": 0.06287261843681335, + "learning_rate": 4.134761819205425e-06, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23100 + }, + { + "epoch": 6.723887110852488, + "flow_matching_loss": 0.012514743395149708, + "grad_norm": 0.08171433955430984, + "learning_rate": 4.1233688518032794e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23110 + }, + { + "epoch": 6.726796624963631, + "flow_matching_loss": 0.01213830653578043, + "grad_norm": 0.0632585734128952, + "learning_rate": 4.1119891005335795e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23120 + }, + { + "epoch": 6.729706139074775, + "flow_matching_loss": 0.012153969053179026, + "grad_norm": 0.06418593972921371, + "learning_rate": 4.1006225792237895e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23130 + }, + { + "epoch": 6.732615653185918, + "flow_matching_loss": 0.012737945653498173, + "grad_norm": 0.06217648833990097, + "learning_rate": 4.089269301685314e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23140 + }, + { + "epoch": 6.7355251672970615, + "flow_matching_loss": 0.011958282627165317, + "grad_norm": 0.08387526124715805, + "learning_rate": 4.077929281713453e-06, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23150 + }, + { + "epoch": 6.738434681408204, + "flow_matching_loss": 0.01344501655548811, + "grad_norm": 0.08157632499933243, + "learning_rate": 4.066602533087398e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23160 + }, + { + "epoch": 6.741344195519348, + "flow_matching_loss": 0.013352810218930245, + "grad_norm": 0.08237400650978088, + "learning_rate": 4.0552890695702175e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23170 + }, + { + "epoch": 6.744253709630492, + "flow_matching_loss": 0.013337617088109256, + "grad_norm": 0.081477589905262, + "learning_rate": 4.04398890490883e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23180 + }, + { + "epoch": 6.7471632237416355, + "flow_matching_loss": 0.013405650667846204, + "grad_norm": 0.07833698391914368, + "learning_rate": 4.032702052834013e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23190 + }, + { + "epoch": 6.750072737852778, + "flow_matching_loss": 0.013357904460281133, + "grad_norm": 0.06393641233444214, + "learning_rate": 4.021428527060352e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23200 + }, + { + "epoch": 6.752982251963922, + "flow_matching_loss": 0.013832518458366394, + "grad_norm": 0.05421995371580124, + "learning_rate": 4.010168341286241e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23210 + }, + { + "epoch": 6.755891766075066, + "flow_matching_loss": 0.01359674073755741, + "grad_norm": 0.07249770313501358, + "learning_rate": 3.998921509193868e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23220 + }, + { + "epoch": 6.7588012801862085, + "flow_matching_loss": 0.01328016333281994, + "grad_norm": 0.07241462171077728, + "learning_rate": 3.987688044449193e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23230 + }, + { + "epoch": 6.761710794297352, + "flow_matching_loss": 0.015134180057793855, + "grad_norm": 0.08209222555160522, + "learning_rate": 3.9764679607019385e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23240 + }, + { + "epoch": 6.764620308408496, + "flow_matching_loss": 0.012993327621370555, + "grad_norm": 0.06588281691074371, + "learning_rate": 3.965261271585561e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23250 + }, + { + "epoch": 6.76752982251964, + "flow_matching_loss": 0.01275297049432993, + "grad_norm": 0.07698290050029755, + "learning_rate": 3.954067990717251e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23260 + }, + { + "epoch": 6.7704393366307825, + "flow_matching_loss": 0.012368461955338717, + "grad_norm": 0.07327555119991302, + "learning_rate": 3.9428881316978935e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23270 + }, + { + "epoch": 6.773348850741926, + "flow_matching_loss": 0.014484945405274629, + "grad_norm": 0.06738276779651642, + "learning_rate": 3.931721708112075e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23280 + }, + { + "epoch": 6.77625836485307, + "flow_matching_loss": 0.012577784899622201, + "grad_norm": 0.05898813530802727, + "learning_rate": 3.920568733528047e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23290 + }, + { + "epoch": 6.779167878964213, + "flow_matching_loss": 0.013249537535011768, + "grad_norm": 0.06367369741201401, + "learning_rate": 3.909429221497734e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23300 + }, + { + "epoch": 6.782077393075356, + "flow_matching_loss": 0.013080786634236573, + "grad_norm": 0.08027568459510803, + "learning_rate": 3.898303185556689e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23310 + }, + { + "epoch": 6.7849869071865, + "flow_matching_loss": 0.013767514564096927, + "grad_norm": 0.09910979121923447, + "learning_rate": 3.887190639224097e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23320 + }, + { + "epoch": 6.787896421297643, + "flow_matching_loss": 0.01205474054440856, + "grad_norm": 0.060933541506528854, + "learning_rate": 3.876091596002746e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23330 + }, + { + "epoch": 6.790805935408787, + "flow_matching_loss": 0.01407859642058611, + "grad_norm": 0.07226121425628662, + "learning_rate": 3.865006069379019e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23340 + }, + { + "epoch": 6.79371544951993, + "flow_matching_loss": 0.012378246989101171, + "grad_norm": 0.06077126786112785, + "learning_rate": 3.853934072822881e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23350 + }, + { + "epoch": 6.796624963631073, + "flow_matching_loss": 0.01279826508834958, + "grad_norm": 0.11632164567708969, + "learning_rate": 3.8428756197878514e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23360 + }, + { + "epoch": 6.799534477742217, + "flow_matching_loss": 0.013852845970541238, + "grad_norm": 0.08741648495197296, + "learning_rate": 3.831830723710992e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23370 + }, + { + "epoch": 6.802443991853361, + "flow_matching_loss": 0.012309214193373918, + "grad_norm": 0.07019244134426117, + "learning_rate": 3.820799398012889e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23380 + }, + { + "epoch": 6.805353505964504, + "flow_matching_loss": 0.013724331324920059, + "grad_norm": 0.10938937216997147, + "learning_rate": 3.80978165609765e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23390 + }, + { + "epoch": 6.808263020075647, + "flow_matching_loss": 0.014336697943508624, + "grad_norm": 0.09971470385789871, + "learning_rate": 3.798777511352864e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23400 + }, + { + "epoch": 6.811172534186791, + "flow_matching_loss": 0.013575168885290622, + "grad_norm": 0.08182620257139206, + "learning_rate": 3.787786977149613e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23410 + }, + { + "epoch": 6.8140820482979345, + "flow_matching_loss": 0.01306431544944644, + "grad_norm": 0.0722547248005867, + "learning_rate": 3.776810066842427e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23420 + }, + { + "epoch": 6.816991562409077, + "flow_matching_loss": 0.012784101441502571, + "grad_norm": 0.06513359397649765, + "learning_rate": 3.7658467937692866e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23430 + }, + { + "epoch": 6.819901076520221, + "flow_matching_loss": 0.012299287132918835, + "grad_norm": 0.05964573472738266, + "learning_rate": 3.7548971712516e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23440 + }, + { + "epoch": 6.822810590631365, + "flow_matching_loss": 0.013486617524176837, + "grad_norm": 0.09415420889854431, + "learning_rate": 3.743961212594197e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23450 + }, + { + "epoch": 6.8257201047425085, + "flow_matching_loss": 0.013028914667665958, + "grad_norm": 0.05950843542814255, + "learning_rate": 3.7330389310852975e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23460 + }, + { + "epoch": 6.828629618853651, + "flow_matching_loss": 0.014827377535402775, + "grad_norm": 0.07325046509504318, + "learning_rate": 3.7221303399965008e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23470 + }, + { + "epoch": 6.831539132964795, + "flow_matching_loss": 0.01272427411749959, + "grad_norm": 0.09654416888952255, + "learning_rate": 3.711235452582775e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23480 + }, + { + "epoch": 6.834448647075939, + "flow_matching_loss": 0.012641285918653012, + "grad_norm": 0.06549131870269775, + "learning_rate": 3.7003542820824347e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23490 + }, + { + "epoch": 6.8373581611870815, + "flow_matching_loss": 0.01434050127863884, + "grad_norm": 0.07705140113830566, + "learning_rate": 3.6894868417171252e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23500 + }, + { + "epoch": 6.840267675298225, + "flow_matching_loss": 0.012918719463050366, + "grad_norm": 0.06565580517053604, + "learning_rate": 3.678633144691818e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23510 + }, + { + "epoch": 6.843177189409369, + "flow_matching_loss": 0.01315924758091569, + "grad_norm": 0.06412441283464432, + "learning_rate": 3.667793204194776e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23520 + }, + { + "epoch": 6.846086703520512, + "flow_matching_loss": 0.010768975596874952, + "grad_norm": 0.06377682834863663, + "learning_rate": 3.656967033397544e-06, + "loss": 0.0108, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23530 + }, + { + "epoch": 6.8489962176316554, + "flow_matching_loss": 0.014235550072044134, + "grad_norm": 0.11339374631643295, + "learning_rate": 3.646154645454948e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23540 + }, + { + "epoch": 6.851905731742799, + "flow_matching_loss": 0.012568653002381325, + "grad_norm": 0.06091293692588806, + "learning_rate": 3.6353560535050524e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23550 + }, + { + "epoch": 6.854815245853942, + "flow_matching_loss": 0.012308130413293839, + "grad_norm": 0.06215246394276619, + "learning_rate": 3.624571270669172e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23560 + }, + { + "epoch": 6.857724759965086, + "flow_matching_loss": 0.013055917713791132, + "grad_norm": 0.07203428447246552, + "learning_rate": 3.6138003100518296e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23570 + }, + { + "epoch": 6.860634274076229, + "flow_matching_loss": 0.013073442317545415, + "grad_norm": 0.054761312901973724, + "learning_rate": 3.6030431847407595e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23580 + }, + { + "epoch": 6.863543788187373, + "flow_matching_loss": 0.01434020409360528, + "grad_norm": 0.10924729704856873, + "learning_rate": 3.592299907806881e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23590 + }, + { + "epoch": 6.866453302298516, + "flow_matching_loss": 0.012352512683719397, + "grad_norm": 0.07767850160598755, + "learning_rate": 3.5815704923042892e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23600 + }, + { + "epoch": 6.86936281640966, + "flow_matching_loss": 0.013451638445258141, + "grad_norm": 0.07494913041591644, + "learning_rate": 3.5708549512702392e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23610 + }, + { + "epoch": 6.872272330520803, + "flow_matching_loss": 0.013869556691497564, + "grad_norm": 0.08847980201244354, + "learning_rate": 3.5601532977251217e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23620 + }, + { + "epoch": 6.875181844631946, + "flow_matching_loss": 0.012346963956952095, + "grad_norm": 0.06703509390354156, + "learning_rate": 3.5494655446724544e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23630 + }, + { + "epoch": 6.87809135874309, + "flow_matching_loss": 0.013439282868057489, + "grad_norm": 0.07025184482336044, + "learning_rate": 3.538791705098866e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23640 + }, + { + "epoch": 6.881000872854234, + "flow_matching_loss": 0.013571306783705949, + "grad_norm": 0.08834854513406754, + "learning_rate": 3.5281317919740718e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23650 + }, + { + "epoch": 6.883910386965377, + "flow_matching_loss": 0.011514554172754288, + "grad_norm": 0.07961033284664154, + "learning_rate": 3.5174858182508825e-06, + "loss": 0.0115, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23660 + }, + { + "epoch": 6.88681990107652, + "flow_matching_loss": 0.01433885945007205, + "grad_norm": 0.06444607675075531, + "learning_rate": 3.506853796865153e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23670 + }, + { + "epoch": 6.889729415187664, + "flow_matching_loss": 0.01267557665705681, + "grad_norm": 0.07557566463947296, + "learning_rate": 3.496235740735789e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23680 + }, + { + "epoch": 6.892638929298807, + "flow_matching_loss": 0.01323378849774599, + "grad_norm": 0.08679758012294769, + "learning_rate": 3.485631662764738e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23690 + }, + { + "epoch": 6.89554844340995, + "flow_matching_loss": 0.013517966307699681, + "grad_norm": 0.06883301585912704, + "learning_rate": 3.475041575836952e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23700 + }, + { + "epoch": 6.898457957521094, + "flow_matching_loss": 0.012650252226740122, + "grad_norm": 0.06750419735908508, + "learning_rate": 3.4644654928203796e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23710 + }, + { + "epoch": 6.901367471632238, + "flow_matching_loss": 0.012091747112572193, + "grad_norm": 0.06011442467570305, + "learning_rate": 3.4539034265659663e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23720 + }, + { + "epoch": 6.904276985743381, + "flow_matching_loss": 0.012967042252421379, + "grad_norm": 0.07578044384717941, + "learning_rate": 3.4433553899076163e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23730 + }, + { + "epoch": 6.907186499854524, + "flow_matching_loss": 0.013312952546402811, + "grad_norm": 0.05966169387102127, + "learning_rate": 3.432821395662189e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23740 + }, + { + "epoch": 6.910096013965668, + "flow_matching_loss": 0.01387314973399043, + "grad_norm": 0.07922252267599106, + "learning_rate": 3.4223014566294785e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23750 + }, + { + "epoch": 6.913005528076811, + "flow_matching_loss": 0.014905789587646722, + "grad_norm": 0.08457576483488083, + "learning_rate": 3.4117955855922e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23760 + }, + { + "epoch": 6.9159150421879545, + "flow_matching_loss": 0.012990378495305776, + "grad_norm": 0.05490583926439285, + "learning_rate": 3.4013037953159855e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23770 + }, + { + "epoch": 6.918824556299098, + "flow_matching_loss": 0.014326185639947652, + "grad_norm": 0.07260622084140778, + "learning_rate": 3.390826098549345e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23780 + }, + { + "epoch": 6.921734070410242, + "flow_matching_loss": 0.012019568216055631, + "grad_norm": 0.0552930124104023, + "learning_rate": 3.3803625080236684e-06, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23790 + }, + { + "epoch": 6.924643584521385, + "flow_matching_loss": 0.01547076292335987, + "grad_norm": 0.11994647234678268, + "learning_rate": 3.369913036453205e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23800 + }, + { + "epoch": 6.927553098632528, + "flow_matching_loss": 0.013233691826462746, + "grad_norm": 0.06702899187803268, + "learning_rate": 3.3594776965350434e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23810 + }, + { + "epoch": 6.930462612743672, + "flow_matching_loss": 0.013987453561276198, + "grad_norm": 0.06085817888379097, + "learning_rate": 3.34905650094911e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23820 + }, + { + "epoch": 6.933372126854815, + "flow_matching_loss": 0.01239976929500699, + "grad_norm": 0.06913836300373077, + "learning_rate": 3.338649462358145e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23830 + }, + { + "epoch": 6.936281640965959, + "flow_matching_loss": 0.012844567652791739, + "grad_norm": 0.06143622472882271, + "learning_rate": 3.3282565934076765e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23840 + }, + { + "epoch": 6.939191155077102, + "flow_matching_loss": 0.013333524577319622, + "grad_norm": 0.07414962351322174, + "learning_rate": 3.31787790672602e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23850 + }, + { + "epoch": 6.942100669188245, + "flow_matching_loss": 0.012160107120871544, + "grad_norm": 0.0696955993771553, + "learning_rate": 3.307513414924255e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23860 + }, + { + "epoch": 6.945010183299389, + "flow_matching_loss": 0.014105782844126225, + "grad_norm": 0.06021328270435333, + "learning_rate": 3.2971631305962246e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23870 + }, + { + "epoch": 6.947919697410533, + "flow_matching_loss": 0.012647306313738227, + "grad_norm": 0.05701705440878868, + "learning_rate": 3.2868270663184964e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23880 + }, + { + "epoch": 6.950829211521675, + "flow_matching_loss": 0.014173644315451383, + "grad_norm": 0.08307893574237823, + "learning_rate": 3.276505234650362e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23890 + }, + { + "epoch": 6.953738725632819, + "flow_matching_loss": 0.0133087407797575, + "grad_norm": 0.07117629796266556, + "learning_rate": 3.266197648133821e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23900 + }, + { + "epoch": 6.956648239743963, + "flow_matching_loss": 0.013530096691101789, + "grad_norm": 0.09680023044347763, + "learning_rate": 3.255904319293562e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23910 + }, + { + "epoch": 6.9595577538551066, + "flow_matching_loss": 0.012747881188988685, + "grad_norm": 0.06917626410722733, + "learning_rate": 3.2456252606369485e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23920 + }, + { + "epoch": 6.962467267966249, + "flow_matching_loss": 0.012068322021514177, + "grad_norm": 0.06282377243041992, + "learning_rate": 3.235360484654011e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23930 + }, + { + "epoch": 6.965376782077393, + "flow_matching_loss": 0.01312935994938016, + "grad_norm": 0.07004904747009277, + "learning_rate": 3.225110003817418e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23940 + }, + { + "epoch": 6.968286296188537, + "flow_matching_loss": 0.014184669684618711, + "grad_norm": 0.11056961119174957, + "learning_rate": 3.214873830582467e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23950 + }, + { + "epoch": 6.97119581029968, + "flow_matching_loss": 0.013077223859727382, + "grad_norm": 0.06396222859621048, + "learning_rate": 3.2046519773870803e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23960 + }, + { + "epoch": 6.974105324410823, + "flow_matching_loss": 0.01376980347558856, + "grad_norm": 0.06966537982225418, + "learning_rate": 3.1944444566517654e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23970 + }, + { + "epoch": 6.977014838521967, + "flow_matching_loss": 0.013524494040757418, + "grad_norm": 0.0663403794169426, + "learning_rate": 3.18425128077963e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23980 + }, + { + "epoch": 6.979924352633111, + "flow_matching_loss": 0.013540730904787779, + "grad_norm": 0.08207444101572037, + "learning_rate": 3.174072462156342e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 23990 + }, + { + "epoch": 6.9828338667442535, + "flow_matching_loss": 0.011592795746400953, + "grad_norm": 0.061608705669641495, + "learning_rate": 3.163908013150124e-06, + "loss": 0.0116, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24000 + }, + { + "epoch": 6.985743380855397, + "flow_matching_loss": 0.012418582849204541, + "grad_norm": 0.07592878490686417, + "learning_rate": 3.153757946111741e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24010 + }, + { + "epoch": 6.988652894966541, + "flow_matching_loss": 0.012359482422471046, + "grad_norm": 0.07717892527580261, + "learning_rate": 3.1436222733744776e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24020 + }, + { + "epoch": 6.991562409077684, + "flow_matching_loss": 0.012033763714134693, + "grad_norm": 0.06441289931535721, + "learning_rate": 3.133501007254138e-06, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24030 + }, + { + "epoch": 6.9944719231888275, + "flow_matching_loss": 0.013655327167361975, + "grad_norm": 0.08312865346670151, + "learning_rate": 3.1233941600490123e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24040 + }, + { + "epoch": 6.997381437299971, + "flow_matching_loss": 0.013295511901378631, + "grad_norm": 0.08324886858463287, + "learning_rate": 3.113301744039875e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24050 + }, + { + "epoch": 7.000290951411114, + "flow_matching_loss": 0.013923694007098674, + "grad_norm": 0.07051778584718704, + "learning_rate": 3.103223771489959e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24060 + }, + { + "epoch": 7.003200465522258, + "flow_matching_loss": 0.012239718856289983, + "grad_norm": 0.0754796490073204, + "learning_rate": 3.09316025464495e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24070 + }, + { + "epoch": 7.006109979633401, + "flow_matching_loss": 0.013776348996907472, + "grad_norm": 0.08114205300807953, + "learning_rate": 3.0831112057329775e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24080 + }, + { + "epoch": 7.009019493744544, + "flow_matching_loss": 0.013368572387844325, + "grad_norm": 0.06838233768939972, + "learning_rate": 3.073076636964579e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24090 + }, + { + "epoch": 7.011929007855688, + "flow_matching_loss": 0.012081048637628555, + "grad_norm": 0.05184592679142952, + "learning_rate": 3.063056560532698e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24100 + }, + { + "epoch": 7.014838521966832, + "flow_matching_loss": 0.012163841351866721, + "grad_norm": 0.08445174992084503, + "learning_rate": 3.0530509886126807e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24110 + }, + { + "epoch": 7.017748036077975, + "flow_matching_loss": 0.012078788690268993, + "grad_norm": 0.0796457901597023, + "learning_rate": 3.043059933362236e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24120 + }, + { + "epoch": 7.020657550189118, + "flow_matching_loss": 0.013791468925774097, + "grad_norm": 0.08410722017288208, + "learning_rate": 3.0330834069214337e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24130 + }, + { + "epoch": 7.023567064300262, + "flow_matching_loss": 0.013339803740382194, + "grad_norm": 0.09685815870761871, + "learning_rate": 3.023121421412703e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24140 + }, + { + "epoch": 7.026476578411406, + "flow_matching_loss": 0.013773251511156558, + "grad_norm": 0.07762318849563599, + "learning_rate": 3.0131739889407896e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24150 + }, + { + "epoch": 7.029386092522548, + "flow_matching_loss": 0.0127343506552279, + "grad_norm": 0.07465093582868576, + "learning_rate": 3.0032411215927645e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24160 + }, + { + "epoch": 7.032295606633692, + "flow_matching_loss": 0.013217219337821006, + "grad_norm": 0.07047083973884583, + "learning_rate": 2.9933228314379975e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24170 + }, + { + "epoch": 7.035205120744836, + "flow_matching_loss": 0.012204410322010518, + "grad_norm": 0.06911014765501022, + "learning_rate": 2.9834191305281405e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24180 + }, + { + "epoch": 7.038114634855979, + "flow_matching_loss": 0.01300343768671155, + "grad_norm": 0.07424776256084442, + "learning_rate": 2.973530030897135e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24190 + }, + { + "epoch": 7.041024148967122, + "flow_matching_loss": 0.012008262239396572, + "grad_norm": 0.07900994271039963, + "learning_rate": 2.9636555445611646e-06, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24200 + }, + { + "epoch": 7.043933663078266, + "flow_matching_loss": 0.01315589053556323, + "grad_norm": 0.08802741020917892, + "learning_rate": 2.953795683518659e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24210 + }, + { + "epoch": 7.04684317718941, + "flow_matching_loss": 0.012994121666997671, + "grad_norm": 0.09260044246912003, + "learning_rate": 2.9439504597502815e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24220 + }, + { + "epoch": 7.049752691300553, + "flow_matching_loss": 0.013760613091289998, + "grad_norm": 0.0671023279428482, + "learning_rate": 2.934119885218902e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24230 + }, + { + "epoch": 7.052662205411696, + "flow_matching_loss": 0.011709636403247714, + "grad_norm": 0.07491301745176315, + "learning_rate": 2.9243039718696045e-06, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24240 + }, + { + "epoch": 7.05557171952284, + "flow_matching_loss": 0.011811607237905263, + "grad_norm": 0.054729197174310684, + "learning_rate": 2.914502731629641e-06, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24250 + }, + { + "epoch": 7.058481233633983, + "flow_matching_loss": 0.014450524374842644, + "grad_norm": 0.07196380943059921, + "learning_rate": 2.904716176408451e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24260 + }, + { + "epoch": 7.0613907477451265, + "flow_matching_loss": 0.013633861392736434, + "grad_norm": 0.10089542716741562, + "learning_rate": 2.8949443180976172e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24270 + }, + { + "epoch": 7.06430026185627, + "flow_matching_loss": 0.013279086165130138, + "grad_norm": 0.0828099399805069, + "learning_rate": 2.885187168570864e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24280 + }, + { + "epoch": 7.067209775967413, + "flow_matching_loss": 0.0144026062451303, + "grad_norm": 0.07810799777507782, + "learning_rate": 2.875444739684058e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24290 + }, + { + "epoch": 7.070119290078557, + "flow_matching_loss": 0.012314677145332098, + "grad_norm": 0.07579539716243744, + "learning_rate": 2.865717043275165e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24300 + }, + { + "epoch": 7.0730288041897005, + "flow_matching_loss": 0.014299785066395999, + "grad_norm": 0.06534378230571747, + "learning_rate": 2.8560040911642528e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24310 + }, + { + "epoch": 7.075938318300844, + "flow_matching_loss": 0.013996868208050729, + "grad_norm": 0.06601939350366592, + "learning_rate": 2.846305895153472e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24320 + }, + { + "epoch": 7.078847832411987, + "flow_matching_loss": 0.014250252489000559, + "grad_norm": 0.08378820866346359, + "learning_rate": 2.8366224670270487e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24330 + }, + { + "epoch": 7.081757346523131, + "flow_matching_loss": 0.011393609549850225, + "grad_norm": 0.05324731767177582, + "learning_rate": 2.8269538185512535e-06, + "loss": 0.0114, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24340 + }, + { + "epoch": 7.084666860634274, + "flow_matching_loss": 0.012587910983711481, + "grad_norm": 0.07163026183843613, + "learning_rate": 2.817299961474417e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24350 + }, + { + "epoch": 7.087576374745417, + "flow_matching_loss": 0.012789037358015776, + "grad_norm": 0.0645531490445137, + "learning_rate": 2.8076609075268816e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24360 + }, + { + "epoch": 7.090485888856561, + "flow_matching_loss": 0.013863526284694672, + "grad_norm": 0.07644204050302505, + "learning_rate": 2.798036668421006e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24370 + }, + { + "epoch": 7.093395402967705, + "flow_matching_loss": 0.012814339715987444, + "grad_norm": 0.09393570572137833, + "learning_rate": 2.7884272558511433e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24380 + }, + { + "epoch": 7.0963049170788475, + "flow_matching_loss": 0.013429556041955948, + "grad_norm": 0.05040053278207779, + "learning_rate": 2.7788326814936456e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24390 + }, + { + "epoch": 7.099214431189991, + "flow_matching_loss": 0.014519502688199282, + "grad_norm": 0.06803322583436966, + "learning_rate": 2.7692529570068176e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24400 + }, + { + "epoch": 7.102123945301135, + "flow_matching_loss": 0.01496996358036995, + "grad_norm": 0.08708856999874115, + "learning_rate": 2.759688094030933e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24410 + }, + { + "epoch": 7.105033459412278, + "flow_matching_loss": 0.013233906030654908, + "grad_norm": 0.07699897140264511, + "learning_rate": 2.750138104188201e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24420 + }, + { + "epoch": 7.107942973523421, + "flow_matching_loss": 0.012809615302830934, + "grad_norm": 0.0747479796409607, + "learning_rate": 2.740602999082759e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24430 + }, + { + "epoch": 7.110852487634565, + "flow_matching_loss": 0.01238290909677744, + "grad_norm": 0.05632628872990608, + "learning_rate": 2.731082790300654e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24440 + }, + { + "epoch": 7.113762001745709, + "flow_matching_loss": 0.01327802212908864, + "grad_norm": 0.051959872245788574, + "learning_rate": 2.7215774894098474e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24450 + }, + { + "epoch": 7.116671515856852, + "flow_matching_loss": 0.01134128523990512, + "grad_norm": 0.07458533346652985, + "learning_rate": 2.7120871079601696e-06, + "loss": 0.0113, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24460 + }, + { + "epoch": 7.119581029967995, + "flow_matching_loss": 0.013349977415055036, + "grad_norm": 0.06368128955364227, + "learning_rate": 2.7026116574833314e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24470 + }, + { + "epoch": 7.122490544079139, + "flow_matching_loss": 0.012562236189842224, + "grad_norm": 0.0845566913485527, + "learning_rate": 2.6931511494928953e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24480 + }, + { + "epoch": 7.125400058190282, + "flow_matching_loss": 0.012606816738843918, + "grad_norm": 0.08125726133584976, + "learning_rate": 2.683705595484275e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24490 + }, + { + "epoch": 7.128309572301426, + "flow_matching_loss": 0.013678825367242098, + "grad_norm": 0.07409604638814926, + "learning_rate": 2.6742750069347014e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24500 + }, + { + "epoch": 7.131219086412569, + "flow_matching_loss": 0.013686820492148399, + "grad_norm": 0.0936979278922081, + "learning_rate": 2.664859395303239e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24510 + }, + { + "epoch": 7.134128600523712, + "flow_matching_loss": 0.015125338081270456, + "grad_norm": 0.09818891435861588, + "learning_rate": 2.655458772030739e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24520 + }, + { + "epoch": 7.137038114634856, + "flow_matching_loss": 0.01205681897699833, + "grad_norm": 0.11127855628728867, + "learning_rate": 2.6460731485398403e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24530 + }, + { + "epoch": 7.1399476287459995, + "flow_matching_loss": 0.014237461239099502, + "grad_norm": 0.0842173844575882, + "learning_rate": 2.6367025362349695e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24540 + }, + { + "epoch": 7.142857142857143, + "flow_matching_loss": 0.013627373520284891, + "grad_norm": 0.07710163295269012, + "learning_rate": 2.6273469465022965e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24550 + }, + { + "epoch": 7.145766656968286, + "flow_matching_loss": 0.013048865739256144, + "grad_norm": 0.06570237129926682, + "learning_rate": 2.618006390709753e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24560 + }, + { + "epoch": 7.14867617107943, + "flow_matching_loss": 0.012613368593156338, + "grad_norm": 0.058185964822769165, + "learning_rate": 2.60868088020699e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24570 + }, + { + "epoch": 7.1515856851905735, + "flow_matching_loss": 0.012759647518396377, + "grad_norm": 0.06023018807172775, + "learning_rate": 2.599370426325382e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24580 + }, + { + "epoch": 7.154495199301716, + "flow_matching_loss": 0.014306988939642907, + "grad_norm": 0.06926777213811874, + "learning_rate": 2.5900750403780093e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24590 + }, + { + "epoch": 7.15740471341286, + "flow_matching_loss": 0.012614351883530616, + "grad_norm": 0.0714886337518692, + "learning_rate": 2.580794733659639e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24600 + }, + { + "epoch": 7.160314227524004, + "flow_matching_loss": 0.014344602916389703, + "grad_norm": 0.06699349731206894, + "learning_rate": 2.5715295174467236e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24610 + }, + { + "epoch": 7.1632237416351465, + "flow_matching_loss": 0.012466399976983667, + "grad_norm": 0.060588542371988297, + "learning_rate": 2.5622794029973746e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24620 + }, + { + "epoch": 7.16613325574629, + "flow_matching_loss": 0.0136436621658504, + "grad_norm": 0.06389699131250381, + "learning_rate": 2.553044401551351e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24630 + }, + { + "epoch": 7.169042769857434, + "flow_matching_loss": 0.01439961064606905, + "grad_norm": 0.08247794955968857, + "learning_rate": 2.54382452433005e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24640 + }, + { + "epoch": 7.171952283968578, + "flow_matching_loss": 0.013059384655207396, + "grad_norm": 0.10571398586034775, + "learning_rate": 2.53461978253649e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24650 + }, + { + "epoch": 7.1748617980797205, + "flow_matching_loss": 0.012258846126496793, + "grad_norm": 0.07228204607963562, + "learning_rate": 2.5254301873553065e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24660 + }, + { + "epoch": 7.177771312190864, + "flow_matching_loss": 0.013164240680634976, + "grad_norm": 0.060404930263757706, + "learning_rate": 2.5162557499527184e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24670 + }, + { + "epoch": 7.180680826302008, + "flow_matching_loss": 0.012725449353456496, + "grad_norm": 0.061132289469242096, + "learning_rate": 2.5070964814765335e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24680 + }, + { + "epoch": 7.183590340413151, + "flow_matching_loss": 0.013678031507879496, + "grad_norm": 0.08273901045322418, + "learning_rate": 2.4979523930561295e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24690 + }, + { + "epoch": 7.186499854524294, + "flow_matching_loss": 0.01191875832155347, + "grad_norm": 0.07530416548252106, + "learning_rate": 2.4888234958024333e-06, + "loss": 0.0119, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24700 + }, + { + "epoch": 7.189409368635438, + "flow_matching_loss": 0.014158102590590716, + "grad_norm": 0.08369842916727066, + "learning_rate": 2.479709800807915e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24710 + }, + { + "epoch": 7.192318882746581, + "flow_matching_loss": 0.013087652344256639, + "grad_norm": 0.09264060854911804, + "learning_rate": 2.4706113191465767e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24720 + }, + { + "epoch": 7.195228396857725, + "flow_matching_loss": 0.012982225371524691, + "grad_norm": 0.0836668461561203, + "learning_rate": 2.46152806187393e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24730 + }, + { + "epoch": 7.198137910968868, + "flow_matching_loss": 0.012134405691176653, + "grad_norm": 0.07441049069166183, + "learning_rate": 2.452460040026989e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24740 + }, + { + "epoch": 7.201047425080012, + "flow_matching_loss": 0.013944985345005989, + "grad_norm": 0.07621421664953232, + "learning_rate": 2.4434072646242555e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24750 + }, + { + "epoch": 7.203956939191155, + "flow_matching_loss": 0.012367180921137333, + "grad_norm": 0.0761420950293541, + "learning_rate": 2.4343697466657015e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24760 + }, + { + "epoch": 7.206866453302299, + "flow_matching_loss": 0.01172742024064064, + "grad_norm": 0.0716809332370758, + "learning_rate": 2.42534749713277e-06, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24770 + }, + { + "epoch": 7.209775967413442, + "flow_matching_loss": 0.013584075309336186, + "grad_norm": 0.05760239437222481, + "learning_rate": 2.4163405269883425e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24780 + }, + { + "epoch": 7.212685481524585, + "flow_matching_loss": 0.013777763582766056, + "grad_norm": 0.06101386994123459, + "learning_rate": 2.407348847176735e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24790 + }, + { + "epoch": 7.215594995635729, + "flow_matching_loss": 0.013429543375968933, + "grad_norm": 0.07487001270055771, + "learning_rate": 2.398372468623688e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24800 + }, + { + "epoch": 7.2185045097468725, + "flow_matching_loss": 0.012228234857320785, + "grad_norm": 0.06392122060060501, + "learning_rate": 2.3894114022363446e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24810 + }, + { + "epoch": 7.221414023858015, + "flow_matching_loss": 0.012323313299566508, + "grad_norm": 0.07510365545749664, + "learning_rate": 2.3804656589032514e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24820 + }, + { + "epoch": 7.224323537969159, + "flow_matching_loss": 0.012571858149021864, + "grad_norm": 0.06851806491613388, + "learning_rate": 2.3715352494943237e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24830 + }, + { + "epoch": 7.227233052080303, + "flow_matching_loss": 0.01299173878505826, + "grad_norm": 0.0656377524137497, + "learning_rate": 2.3626201848608582e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24840 + }, + { + "epoch": 7.2301425661914465, + "flow_matching_loss": 0.013223565369844436, + "grad_norm": 0.08375021815299988, + "learning_rate": 2.353720475835494e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24850 + }, + { + "epoch": 7.233052080302589, + "flow_matching_loss": 0.011923977453261615, + "grad_norm": 0.06097510829567909, + "learning_rate": 2.344836133232215e-06, + "loss": 0.0119, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24860 + }, + { + "epoch": 7.235961594413733, + "flow_matching_loss": 0.013761421758681535, + "grad_norm": 0.06924326717853546, + "learning_rate": 2.335967167846342e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24870 + }, + { + "epoch": 7.238871108524877, + "flow_matching_loss": 0.012687346246093512, + "grad_norm": 0.07396054267883301, + "learning_rate": 2.3271135904544993e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24880 + }, + { + "epoch": 7.2417806226360195, + "flow_matching_loss": 0.012041572760790587, + "grad_norm": 0.07609918713569641, + "learning_rate": 2.3182754118146206e-06, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24890 + }, + { + "epoch": 7.244690136747163, + "flow_matching_loss": 0.012917874939739704, + "grad_norm": 0.066389299929142, + "learning_rate": 2.309452642665923e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24900 + }, + { + "epoch": 7.247599650858307, + "flow_matching_loss": 0.013467625342309475, + "grad_norm": 0.09029030054807663, + "learning_rate": 2.3006452937289065e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24910 + }, + { + "epoch": 7.25050916496945, + "flow_matching_loss": 0.013243778143078089, + "grad_norm": 0.06256437301635742, + "learning_rate": 2.291853375705327e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24920 + }, + { + "epoch": 7.2534186790805935, + "flow_matching_loss": 0.013294597156345844, + "grad_norm": 0.06800330430269241, + "learning_rate": 2.283076899278197e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24930 + }, + { + "epoch": 7.256328193191737, + "flow_matching_loss": 0.013315598294138909, + "grad_norm": 0.06033209338784218, + "learning_rate": 2.274315875111763e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24940 + }, + { + "epoch": 7.25923770730288, + "flow_matching_loss": 0.013028513081371785, + "grad_norm": 0.06581635028123856, + "learning_rate": 2.265570313851493e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24950 + }, + { + "epoch": 7.262147221414024, + "flow_matching_loss": 0.013104200642555953, + "grad_norm": 0.07426039129495621, + "learning_rate": 2.2568402261240685e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24960 + }, + { + "epoch": 7.265056735525167, + "flow_matching_loss": 0.013127768971025943, + "grad_norm": 0.0683746412396431, + "learning_rate": 2.248125622537373e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24970 + }, + { + "epoch": 7.267966249636311, + "flow_matching_loss": 0.015658169891685246, + "grad_norm": 0.09765830636024475, + "learning_rate": 2.239426513680467e-06, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24980 + }, + { + "epoch": 7.270875763747454, + "flow_matching_loss": 0.0113798966165632, + "grad_norm": 0.07160168886184692, + "learning_rate": 2.2307429101235915e-06, + "loss": 0.0114, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 24990 + }, + { + "epoch": 7.273785277858598, + "flow_matching_loss": 0.01377755831927061, + "grad_norm": 0.08232659846544266, + "learning_rate": 2.222074822418145e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25000 + }, + { + "epoch": 7.276694791969741, + "flow_matching_loss": 0.012920684739947319, + "grad_norm": 0.06931172311306, + "learning_rate": 2.213422261096668e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25010 + }, + { + "epoch": 7.279604306080884, + "flow_matching_loss": 0.01247064326889813, + "grad_norm": 0.07889042049646378, + "learning_rate": 2.204785236672835e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25020 + }, + { + "epoch": 7.282513820192028, + "flow_matching_loss": 0.012855493929237127, + "grad_norm": 0.07014771550893784, + "learning_rate": 2.1961637596414536e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25030 + }, + { + "epoch": 7.285423334303172, + "flow_matching_loss": 0.014467110205441714, + "grad_norm": 0.08160828799009323, + "learning_rate": 2.1875578404784275e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25040 + }, + { + "epoch": 7.288332848414315, + "flow_matching_loss": 0.012672700826078654, + "grad_norm": 0.058091193437576294, + "learning_rate": 2.1789674896407583e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25050 + }, + { + "epoch": 7.291242362525458, + "flow_matching_loss": 0.012809596210718154, + "grad_norm": 0.06610129773616791, + "learning_rate": 2.170392717566534e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25060 + }, + { + "epoch": 7.294151876636602, + "flow_matching_loss": 0.0133501716889441, + "grad_norm": 0.054400645196437836, + "learning_rate": 2.161833534674909e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25070 + }, + { + "epoch": 7.2970613907477455, + "flow_matching_loss": 0.013675120938569308, + "grad_norm": 0.053762346506118774, + "learning_rate": 2.1532899513661015e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25080 + }, + { + "epoch": 7.299970904858888, + "flow_matching_loss": 0.012978781666606664, + "grad_norm": 0.08040399849414825, + "learning_rate": 2.1447619780213682e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25090 + }, + { + "epoch": 7.302880418970032, + "flow_matching_loss": 0.012538180407136678, + "grad_norm": 0.055258795619010925, + "learning_rate": 2.136249625003001e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25100 + }, + { + "epoch": 7.305789933081176, + "flow_matching_loss": 0.013060044962912798, + "grad_norm": 0.057001613080501556, + "learning_rate": 2.1277529026543107e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25110 + }, + { + "epoch": 7.308699447192319, + "flow_matching_loss": 0.013655280694365501, + "grad_norm": 0.058778002858161926, + "learning_rate": 2.1192718212996188e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25120 + }, + { + "epoch": 7.311608961303462, + "flow_matching_loss": 0.010538684530183673, + "grad_norm": 0.05762752890586853, + "learning_rate": 2.110806391244235e-06, + "loss": 0.0105, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25130 + }, + { + "epoch": 7.314518475414606, + "flow_matching_loss": 0.013550690282136202, + "grad_norm": 0.07295507192611694, + "learning_rate": 2.1023566227744596e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25140 + }, + { + "epoch": 7.317427989525749, + "flow_matching_loss": 0.013786217477172614, + "grad_norm": 0.07096055150032043, + "learning_rate": 2.0939225261575563e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25150 + }, + { + "epoch": 7.3203375036368925, + "flow_matching_loss": 0.012375852558761835, + "grad_norm": 0.095338836312294, + "learning_rate": 2.0855041116417473e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25160 + }, + { + "epoch": 7.323247017748036, + "flow_matching_loss": 0.012291854154318571, + "grad_norm": 0.09436263144016266, + "learning_rate": 2.0771013894562003e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25170 + }, + { + "epoch": 7.32615653185918, + "flow_matching_loss": 0.0126788261346519, + "grad_norm": 0.06214461475610733, + "learning_rate": 2.0687143698110113e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25180 + }, + { + "epoch": 7.329066045970323, + "flow_matching_loss": 0.013309497106820344, + "grad_norm": 0.1134079322218895, + "learning_rate": 2.060343062897208e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25190 + }, + { + "epoch": 7.3319755600814664, + "flow_matching_loss": 0.012689704913645983, + "grad_norm": 0.06954015791416168, + "learning_rate": 2.0519874788867127e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25200 + }, + { + "epoch": 7.33488507419261, + "flow_matching_loss": 0.014092088025063277, + "grad_norm": 0.0855197012424469, + "learning_rate": 2.043647627932347e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25210 + }, + { + "epoch": 7.337794588303753, + "flow_matching_loss": 0.013058912847191096, + "grad_norm": 0.06703419238328934, + "learning_rate": 2.0353235201678193e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25220 + }, + { + "epoch": 7.340704102414897, + "flow_matching_loss": 0.012271210271865129, + "grad_norm": 0.07556241005659103, + "learning_rate": 2.0270151657076995e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25230 + }, + { + "epoch": 7.34361361652604, + "flow_matching_loss": 0.014575333055108785, + "grad_norm": 0.0758795291185379, + "learning_rate": 2.0187225746474286e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25240 + }, + { + "epoch": 7.346523130637183, + "flow_matching_loss": 0.012605851236730813, + "grad_norm": 0.06054588779807091, + "learning_rate": 2.0104457570632845e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25250 + }, + { + "epoch": 7.349432644748327, + "flow_matching_loss": 0.012461605109274387, + "grad_norm": 0.0611952468752861, + "learning_rate": 2.0021847230123775e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25260 + }, + { + "epoch": 7.352342158859471, + "flow_matching_loss": 0.011523432936519385, + "grad_norm": 0.05084116384387016, + "learning_rate": 1.9939394825326494e-06, + "loss": 0.0115, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25270 + }, + { + "epoch": 7.355251672970613, + "flow_matching_loss": 0.012106937356293201, + "grad_norm": 0.05574171245098114, + "learning_rate": 1.9857100456428385e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25280 + }, + { + "epoch": 7.358161187081757, + "flow_matching_loss": 0.01310834102332592, + "grad_norm": 0.07533165067434311, + "learning_rate": 1.977496422342492e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25290 + }, + { + "epoch": 7.361070701192901, + "flow_matching_loss": 0.01304339999333024, + "grad_norm": 0.07708322256803513, + "learning_rate": 1.9692986226119345e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25300 + }, + { + "epoch": 7.363980215304045, + "flow_matching_loss": 0.014299921691417694, + "grad_norm": 0.08604677021503448, + "learning_rate": 1.9611166564122658e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25310 + }, + { + "epoch": 7.366889729415187, + "flow_matching_loss": 0.012930291332304477, + "grad_norm": 0.07884234935045242, + "learning_rate": 1.9529505336853456e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25320 + }, + { + "epoch": 7.369799243526331, + "flow_matching_loss": 0.014421731792390346, + "grad_norm": 0.07091029733419418, + "learning_rate": 1.9448002643537833e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25330 + }, + { + "epoch": 7.372708757637475, + "flow_matching_loss": 0.011736518330872059, + "grad_norm": 0.06459653377532959, + "learning_rate": 1.9366658583209205e-06, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25340 + }, + { + "epoch": 7.375618271748618, + "flow_matching_loss": 0.012224908731877804, + "grad_norm": 0.07143402099609375, + "learning_rate": 1.9285473254708363e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25350 + }, + { + "epoch": 7.378527785859761, + "flow_matching_loss": 0.013674024678766727, + "grad_norm": 0.09013223648071289, + "learning_rate": 1.9204446756683092e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25360 + }, + { + "epoch": 7.381437299970905, + "flow_matching_loss": 0.013178820349276065, + "grad_norm": 0.07861611992120743, + "learning_rate": 1.912357918758822e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25370 + }, + { + "epoch": 7.384346814082049, + "flow_matching_loss": 0.012334919348359109, + "grad_norm": 0.07647135108709335, + "learning_rate": 1.9042870645685478e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25380 + }, + { + "epoch": 7.387256328193192, + "flow_matching_loss": 0.01341190179809928, + "grad_norm": 0.05840656906366348, + "learning_rate": 1.8962321229043322e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25390 + }, + { + "epoch": 7.390165842304335, + "flow_matching_loss": 0.014032341074198484, + "grad_norm": 0.06756342202425003, + "learning_rate": 1.8881931035536943e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25400 + }, + { + "epoch": 7.393075356415479, + "flow_matching_loss": 0.012930939253419638, + "grad_norm": 0.06421398371458054, + "learning_rate": 1.8801700162847952e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25410 + }, + { + "epoch": 7.395984870526622, + "flow_matching_loss": 0.01330068688839674, + "grad_norm": 0.07883691787719727, + "learning_rate": 1.87216287084645e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25420 + }, + { + "epoch": 7.3988943846377655, + "flow_matching_loss": 0.013317269645631314, + "grad_norm": 0.09124581515789032, + "learning_rate": 1.8641716769680895e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25430 + }, + { + "epoch": 7.401803898748909, + "flow_matching_loss": 0.012218899372965098, + "grad_norm": 0.062327321618795395, + "learning_rate": 1.8561964443597662e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25440 + }, + { + "epoch": 7.404713412860052, + "flow_matching_loss": 0.013039053231477738, + "grad_norm": 0.07481712102890015, + "learning_rate": 1.848237182712147e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25450 + }, + { + "epoch": 7.407622926971196, + "flow_matching_loss": 0.01274887048639357, + "grad_norm": 0.056531988084316254, + "learning_rate": 1.8402939016964788e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25460 + }, + { + "epoch": 7.410532441082339, + "flow_matching_loss": 0.013257852755486965, + "grad_norm": 0.047009702771902084, + "learning_rate": 1.832366610964601e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25470 + }, + { + "epoch": 7.413441955193482, + "flow_matching_loss": 0.01259139897301793, + "grad_norm": 0.05993375554680824, + "learning_rate": 1.8244553201489166e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25480 + }, + { + "epoch": 7.416351469304626, + "flow_matching_loss": 0.012274976540356874, + "grad_norm": 0.0734214186668396, + "learning_rate": 1.8165600388623882e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25490 + }, + { + "epoch": 7.41926098341577, + "flow_matching_loss": 0.013438132591545582, + "grad_norm": 0.07600793242454529, + "learning_rate": 1.8086807766985331e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25500 + }, + { + "epoch": 7.422170497526913, + "flow_matching_loss": 0.012121975794434548, + "grad_norm": 0.05948229506611824, + "learning_rate": 1.8008175432313946e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25510 + }, + { + "epoch": 7.425080011638056, + "flow_matching_loss": 0.01191346487030387, + "grad_norm": 0.08612322062253952, + "learning_rate": 1.7929703480155429e-06, + "loss": 0.0119, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25520 + }, + { + "epoch": 7.4279895257492, + "flow_matching_loss": 0.013862775871530175, + "grad_norm": 0.07376115769147873, + "learning_rate": 1.7851392005860595e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25530 + }, + { + "epoch": 7.430899039860344, + "flow_matching_loss": 0.012960921786725522, + "grad_norm": 0.05657510459423065, + "learning_rate": 1.7773241104585243e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25540 + }, + { + "epoch": 7.433808553971486, + "flow_matching_loss": 0.012099429173395038, + "grad_norm": 0.07718539983034134, + "learning_rate": 1.7695250871290108e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25550 + }, + { + "epoch": 7.43671806808263, + "flow_matching_loss": 0.013707152009010315, + "grad_norm": 0.0667220875620842, + "learning_rate": 1.761742140074074e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25560 + }, + { + "epoch": 7.439627582193774, + "flow_matching_loss": 0.012684958055615425, + "grad_norm": 0.08278442174196243, + "learning_rate": 1.7539752787507208e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25570 + }, + { + "epoch": 7.442537096304917, + "flow_matching_loss": 0.01366911232471466, + "grad_norm": 0.07153280079364777, + "learning_rate": 1.7462245125964233e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25580 + }, + { + "epoch": 7.44544661041606, + "flow_matching_loss": 0.014841235801577568, + "grad_norm": 0.0994393453001976, + "learning_rate": 1.7384898510290931e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25590 + }, + { + "epoch": 7.448356124527204, + "flow_matching_loss": 0.01342151453718543, + "grad_norm": 0.07436629384756088, + "learning_rate": 1.730771303447069e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25600 + }, + { + "epoch": 7.451265638638348, + "flow_matching_loss": 0.0129134276881814, + "grad_norm": 0.0859152302145958, + "learning_rate": 1.7230688792291204e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25610 + }, + { + "epoch": 7.454175152749491, + "flow_matching_loss": 0.014512934908270835, + "grad_norm": 0.08573093265295029, + "learning_rate": 1.7153825877344175e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25620 + }, + { + "epoch": 7.457084666860634, + "flow_matching_loss": 0.012800470925867558, + "grad_norm": 0.05831589549779892, + "learning_rate": 1.7077124383025278e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25630 + }, + { + "epoch": 7.459994180971778, + "flow_matching_loss": 0.012330307718366384, + "grad_norm": 0.06885861605405807, + "learning_rate": 1.7000584402534058e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25640 + }, + { + "epoch": 7.462903695082921, + "flow_matching_loss": 0.01293637678027153, + "grad_norm": 0.050423312932252884, + "learning_rate": 1.6924206028873774e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25650 + }, + { + "epoch": 7.4658132091940645, + "flow_matching_loss": 0.013649784866720438, + "grad_norm": 0.06327241659164429, + "learning_rate": 1.6847989354851434e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25660 + }, + { + "epoch": 7.468722723305208, + "flow_matching_loss": 0.011483726277947425, + "grad_norm": 0.06406784802675247, + "learning_rate": 1.6771934473077443e-06, + "loss": 0.0115, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25670 + }, + { + "epoch": 7.471632237416351, + "flow_matching_loss": 0.013094308972358703, + "grad_norm": 0.08338239789009094, + "learning_rate": 1.669604147596563e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25680 + }, + { + "epoch": 7.474541751527495, + "flow_matching_loss": 0.012733108550310134, + "grad_norm": 0.07019822299480438, + "learning_rate": 1.662031045573319e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25690 + }, + { + "epoch": 7.4774512656386385, + "flow_matching_loss": 0.012831385992467404, + "grad_norm": 0.06902701407670975, + "learning_rate": 1.6544741504400402e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25700 + }, + { + "epoch": 7.480360779749782, + "flow_matching_loss": 0.011672129109501838, + "grad_norm": 0.07583564519882202, + "learning_rate": 1.6469334713790712e-06, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25710 + }, + { + "epoch": 7.483270293860925, + "flow_matching_loss": 0.013492080755531788, + "grad_norm": 0.08908869326114655, + "learning_rate": 1.639409017553048e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25720 + }, + { + "epoch": 7.486179807972069, + "flow_matching_loss": 0.012594279460608959, + "grad_norm": 0.05291089415550232, + "learning_rate": 1.6319007981048884e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25730 + }, + { + "epoch": 7.489089322083212, + "flow_matching_loss": 0.012221523933112622, + "grad_norm": 0.06040407344698906, + "learning_rate": 1.6244088221577896e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25740 + }, + { + "epoch": 7.491998836194355, + "flow_matching_loss": 0.013818352203816175, + "grad_norm": 0.08165033161640167, + "learning_rate": 1.6169330988152081e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25750 + }, + { + "epoch": 7.494908350305499, + "flow_matching_loss": 0.012833732645958662, + "grad_norm": 0.07168484479188919, + "learning_rate": 1.6094736371608475e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25760 + }, + { + "epoch": 7.497817864416643, + "flow_matching_loss": 0.012817358877509833, + "grad_norm": 0.10635149478912354, + "learning_rate": 1.602030446258666e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25770 + }, + { + "epoch": 7.5007273785277855, + "flow_matching_loss": 0.012820358015596866, + "grad_norm": 0.07729887217283249, + "learning_rate": 1.5946035351528388e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25780 + }, + { + "epoch": 7.503636892638929, + "flow_matching_loss": 0.011788236815482378, + "grad_norm": 0.07085059583187103, + "learning_rate": 1.5871929128677621e-06, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25790 + }, + { + "epoch": 7.506546406750073, + "flow_matching_loss": 0.013153692241758108, + "grad_norm": 0.08461789041757584, + "learning_rate": 1.5797985884080412e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25800 + }, + { + "epoch": 7.509455920861216, + "flow_matching_loss": 0.013119586743414402, + "grad_norm": 0.08148098737001419, + "learning_rate": 1.572420570758476e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25810 + }, + { + "epoch": 7.512365434972359, + "flow_matching_loss": 0.014783224556595087, + "grad_norm": 0.07059703767299652, + "learning_rate": 1.5650588688840567e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25820 + }, + { + "epoch": 7.515274949083503, + "flow_matching_loss": 0.014253124594688416, + "grad_norm": 0.07809881120920181, + "learning_rate": 1.5577134917299413e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25830 + }, + { + "epoch": 7.518184463194647, + "flow_matching_loss": 0.012570601049810648, + "grad_norm": 0.08103885501623154, + "learning_rate": 1.5503844482214618e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25840 + }, + { + "epoch": 7.52109397730579, + "flow_matching_loss": 0.012107066623866559, + "grad_norm": 0.05658887326717377, + "learning_rate": 1.5430717472640932e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25850 + }, + { + "epoch": 7.524003491416933, + "flow_matching_loss": 0.013507523015141487, + "grad_norm": 0.07372228056192398, + "learning_rate": 1.5357753977434541e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25860 + }, + { + "epoch": 7.526913005528077, + "flow_matching_loss": 0.013177156448364258, + "grad_norm": 0.0726289302110672, + "learning_rate": 1.5284954085253043e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25870 + }, + { + "epoch": 7.52982251963922, + "flow_matching_loss": 0.012900643981993198, + "grad_norm": 0.06559240072965622, + "learning_rate": 1.5212317884555115e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25880 + }, + { + "epoch": 7.532732033750364, + "flow_matching_loss": 0.012548348773270846, + "grad_norm": 0.07608447223901749, + "learning_rate": 1.5139845463600605e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25890 + }, + { + "epoch": 7.535641547861507, + "flow_matching_loss": 0.012182949110865593, + "grad_norm": 0.07705956697463989, + "learning_rate": 1.5067536910450352e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25900 + }, + { + "epoch": 7.538551061972651, + "flow_matching_loss": 0.010484562814235687, + "grad_norm": 0.07828722149133682, + "learning_rate": 1.4995392312966044e-06, + "loss": 0.0105, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25910 + }, + { + "epoch": 7.541460576083794, + "flow_matching_loss": 0.012635072227567434, + "grad_norm": 0.07037383317947388, + "learning_rate": 1.492341175881014e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25920 + }, + { + "epoch": 7.5443700901949375, + "flow_matching_loss": 0.013254914432764053, + "grad_norm": 0.06791326403617859, + "learning_rate": 1.4851595335445895e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25930 + }, + { + "epoch": 7.547279604306081, + "flow_matching_loss": 0.013586146710440516, + "grad_norm": 0.06689749658107758, + "learning_rate": 1.4779943130136953e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25940 + }, + { + "epoch": 7.550189118417224, + "flow_matching_loss": 0.01177665013819933, + "grad_norm": 0.07148636132478714, + "learning_rate": 1.4708455229947537e-06, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25950 + }, + { + "epoch": 7.553098632528368, + "flow_matching_loss": 0.012249214015901089, + "grad_norm": 0.07021035999059677, + "learning_rate": 1.4637131721742148e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25960 + }, + { + "epoch": 7.5560081466395115, + "flow_matching_loss": 0.012855331040918827, + "grad_norm": 0.05520514398813248, + "learning_rate": 1.4565972692185626e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25970 + }, + { + "epoch": 7.558917660750654, + "flow_matching_loss": 0.012447944236919284, + "grad_norm": 0.06433721631765366, + "learning_rate": 1.4494978227742855e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25980 + }, + { + "epoch": 7.561827174861798, + "flow_matching_loss": 0.013791683688759803, + "grad_norm": 0.05606786161661148, + "learning_rate": 1.442414841467885e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 25990 + }, + { + "epoch": 7.564736688972942, + "flow_matching_loss": 0.013292414788156747, + "grad_norm": 0.07977280765771866, + "learning_rate": 1.4353483339058498e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26000 + }, + { + "epoch": 7.5676462030840845, + "flow_matching_loss": 0.012237032130360603, + "grad_norm": 0.0483664870262146, + "learning_rate": 1.4282983086746498e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26010 + }, + { + "epoch": 7.570555717195228, + "flow_matching_loss": 0.011142915533855557, + "grad_norm": 0.061599250882864, + "learning_rate": 1.4212647743407325e-06, + "loss": 0.0111, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26020 + }, + { + "epoch": 7.573465231306372, + "flow_matching_loss": 0.013078386243432761, + "grad_norm": 0.06774485856294632, + "learning_rate": 1.4142477394505065e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26030 + }, + { + "epoch": 7.576374745417516, + "flow_matching_loss": 0.01283123278990388, + "grad_norm": 0.09033896028995514, + "learning_rate": 1.4072472125303299e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26040 + }, + { + "epoch": 7.5792842595286585, + "flow_matching_loss": 0.01255258172750473, + "grad_norm": 0.06380046904087067, + "learning_rate": 1.4002632020865031e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26050 + }, + { + "epoch": 7.582193773639802, + "flow_matching_loss": 0.012843643315136433, + "grad_norm": 0.07161778956651688, + "learning_rate": 1.393295716605258e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26060 + }, + { + "epoch": 7.585103287750946, + "flow_matching_loss": 0.013148422073572873, + "grad_norm": 0.05444419011473656, + "learning_rate": 1.386344764552742e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26070 + }, + { + "epoch": 7.588012801862089, + "flow_matching_loss": 0.012031706795096398, + "grad_norm": 0.07700629532337189, + "learning_rate": 1.3794103543750246e-06, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26080 + }, + { + "epoch": 7.590922315973232, + "flow_matching_loss": 0.012030580546706915, + "grad_norm": 0.06295322626829147, + "learning_rate": 1.3724924944980637e-06, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26090 + }, + { + "epoch": 7.593831830084376, + "flow_matching_loss": 0.01177627956494689, + "grad_norm": 0.049774814397096634, + "learning_rate": 1.365591193327712e-06, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26100 + }, + { + "epoch": 7.59674134419552, + "flow_matching_loss": 0.01392419608309865, + "grad_norm": 0.06342910975217819, + "learning_rate": 1.3587064592496984e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26110 + }, + { + "epoch": 7.599650858306663, + "flow_matching_loss": 0.014274237211793661, + "grad_norm": 0.09233888983726501, + "learning_rate": 1.3518383006296292e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26120 + }, + { + "epoch": 7.602560372417806, + "flow_matching_loss": 0.011621550004929305, + "grad_norm": 0.0687977597117424, + "learning_rate": 1.344986725812959e-06, + "loss": 0.0116, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26130 + }, + { + "epoch": 7.605469886528949, + "flow_matching_loss": 0.01190667850896716, + "grad_norm": 0.14373205602169037, + "learning_rate": 1.3381517431250046e-06, + "loss": 0.0119, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26140 + }, + { + "epoch": 7.608379400640093, + "flow_matching_loss": 0.012780462950468063, + "grad_norm": 0.055491507053375244, + "learning_rate": 1.3313333608709095e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26150 + }, + { + "epoch": 7.611288914751237, + "flow_matching_loss": 0.013227985426783561, + "grad_norm": 0.08118662983179092, + "learning_rate": 1.3245315873356528e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26160 + }, + { + "epoch": 7.61419842886238, + "flow_matching_loss": 0.01240280456840992, + "grad_norm": 0.07563182711601257, + "learning_rate": 1.3177464307840304e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26170 + }, + { + "epoch": 7.617107942973523, + "flow_matching_loss": 0.012079085130244494, + "grad_norm": 0.05798719823360443, + "learning_rate": 1.310977899460642e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26180 + }, + { + "epoch": 7.620017457084667, + "flow_matching_loss": 0.01311165578663349, + "grad_norm": 0.07265697419643402, + "learning_rate": 1.3042260015899006e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26190 + }, + { + "epoch": 7.6229269711958105, + "flow_matching_loss": 0.01389027489349246, + "grad_norm": 0.08371012657880783, + "learning_rate": 1.2974907453759948e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26200 + }, + { + "epoch": 7.625836485306953, + "flow_matching_loss": 0.013912614714354277, + "grad_norm": 0.07087790220975876, + "learning_rate": 1.2907721390028937e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26210 + }, + { + "epoch": 7.628745999418097, + "flow_matching_loss": 0.013536378741264343, + "grad_norm": 0.070008784532547, + "learning_rate": 1.2840701906343417e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26220 + }, + { + "epoch": 7.631655513529241, + "flow_matching_loss": 0.013747624401003123, + "grad_norm": 0.0680910050868988, + "learning_rate": 1.2773849084138312e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26230 + }, + { + "epoch": 7.6345650276403845, + "flow_matching_loss": 0.011536851711571216, + "grad_norm": 0.07855898886919022, + "learning_rate": 1.2707163004646177e-06, + "loss": 0.0115, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26240 + }, + { + "epoch": 7.637474541751527, + "flow_matching_loss": 0.012234514066949486, + "grad_norm": 0.0681988000869751, + "learning_rate": 1.2640643748896879e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26250 + }, + { + "epoch": 7.640384055862671, + "flow_matching_loss": 0.012548262905329465, + "grad_norm": 0.05895388498902321, + "learning_rate": 1.257429139771753e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26260 + }, + { + "epoch": 7.643293569973815, + "flow_matching_loss": 0.012987198401242494, + "grad_norm": 0.07444070279598236, + "learning_rate": 1.2508106031732553e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26270 + }, + { + "epoch": 7.6462030840849575, + "flow_matching_loss": 0.012397184036672115, + "grad_norm": 0.061005447059869766, + "learning_rate": 1.2442087731363366e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26280 + }, + { + "epoch": 7.649112598196101, + "flow_matching_loss": 0.013311865646392107, + "grad_norm": 0.08798597007989883, + "learning_rate": 1.237623657682847e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26290 + }, + { + "epoch": 7.652022112307245, + "flow_matching_loss": 0.012807992938905954, + "grad_norm": 0.09124507009983063, + "learning_rate": 1.2310552648143208e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26300 + }, + { + "epoch": 7.654931626418388, + "flow_matching_loss": 0.012266082037240267, + "grad_norm": 0.0875726267695427, + "learning_rate": 1.2245036025119738e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26310 + }, + { + "epoch": 7.6578411405295315, + "flow_matching_loss": 0.012627287488430738, + "grad_norm": 0.09540621936321259, + "learning_rate": 1.2179686787366962e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26320 + }, + { + "epoch": 7.660750654640675, + "flow_matching_loss": 0.014216389041393995, + "grad_norm": 0.08137194812297821, + "learning_rate": 1.2114505014290329e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26330 + }, + { + "epoch": 7.663660168751818, + "flow_matching_loss": 0.014367375895380974, + "grad_norm": 0.06338533759117126, + "learning_rate": 1.2049490785091833e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26340 + }, + { + "epoch": 7.666569682862962, + "flow_matching_loss": 0.014211819786578418, + "grad_norm": 0.06757400929927826, + "learning_rate": 1.1984644178769949e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26350 + }, + { + "epoch": 7.669479196974105, + "flow_matching_loss": 0.012798390164971352, + "grad_norm": 0.06758058071136475, + "learning_rate": 1.1919965274119381e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26360 + }, + { + "epoch": 7.672388711085249, + "flow_matching_loss": 0.012690815702080726, + "grad_norm": 0.07073196023702621, + "learning_rate": 1.18554541497311e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26370 + }, + { + "epoch": 7.675298225196392, + "flow_matching_loss": 0.012623388040810823, + "grad_norm": 0.06943254917860031, + "learning_rate": 1.1791110883992213e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26380 + }, + { + "epoch": 7.678207739307536, + "flow_matching_loss": 0.012302027270197868, + "grad_norm": 0.06511859595775604, + "learning_rate": 1.1726935555085817e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26390 + }, + { + "epoch": 7.681117253418679, + "flow_matching_loss": 0.013017509318888187, + "grad_norm": 0.06441745162010193, + "learning_rate": 1.166292824099105e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26400 + }, + { + "epoch": 7.684026767529822, + "flow_matching_loss": 0.012132040224969387, + "grad_norm": 0.06387417018413544, + "learning_rate": 1.1599089019482768e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26410 + }, + { + "epoch": 7.686936281640966, + "flow_matching_loss": 0.013589953817427159, + "grad_norm": 0.0675065740942955, + "learning_rate": 1.1535417968131718e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26420 + }, + { + "epoch": 7.68984579575211, + "flow_matching_loss": 0.012828375585377217, + "grad_norm": 0.06698855757713318, + "learning_rate": 1.147191516430418e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26430 + }, + { + "epoch": 7.692755309863253, + "flow_matching_loss": 0.01225200816988945, + "grad_norm": 0.06951673328876495, + "learning_rate": 1.140858068516205e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26440 + }, + { + "epoch": 7.695664823974396, + "flow_matching_loss": 0.012968241423368453, + "grad_norm": 0.07945740967988968, + "learning_rate": 1.1345414607662736e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26450 + }, + { + "epoch": 7.69857433808554, + "flow_matching_loss": 0.013866188656538724, + "grad_norm": 0.06808175146579742, + "learning_rate": 1.1282417008558937e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26460 + }, + { + "epoch": 7.7014838521966835, + "flow_matching_loss": 0.01109476275742054, + "grad_norm": 0.05359599366784096, + "learning_rate": 1.1219587964398674e-06, + "loss": 0.0111, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26470 + }, + { + "epoch": 7.704393366307826, + "flow_matching_loss": 0.014301272202283144, + "grad_norm": 0.07413151860237122, + "learning_rate": 1.1156927551525198e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26480 + }, + { + "epoch": 7.70730288041897, + "flow_matching_loss": 0.012859277240931988, + "grad_norm": 0.07639714330434799, + "learning_rate": 1.109443584607675e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26490 + }, + { + "epoch": 7.710212394530114, + "flow_matching_loss": 0.012814912293106318, + "grad_norm": 0.07460569590330124, + "learning_rate": 1.1032112923986708e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26500 + }, + { + "epoch": 7.713121908641257, + "flow_matching_loss": 0.010979409143328667, + "grad_norm": 0.06333214789628983, + "learning_rate": 1.0969958860983299e-06, + "loss": 0.011, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26510 + }, + { + "epoch": 7.7160314227524, + "flow_matching_loss": 0.013121406454592944, + "grad_norm": 0.07062237709760666, + "learning_rate": 1.0907973732589537e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26520 + }, + { + "epoch": 7.718940936863544, + "flow_matching_loss": 0.014736282359808683, + "grad_norm": 0.08114725351333618, + "learning_rate": 1.0846157614123215e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26530 + }, + { + "epoch": 7.721850450974687, + "flow_matching_loss": 0.012898176675662398, + "grad_norm": 0.08635713160037994, + "learning_rate": 1.078451058069671e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26540 + }, + { + "epoch": 7.7247599650858305, + "flow_matching_loss": 0.011656746733933687, + "grad_norm": 0.076515793800354, + "learning_rate": 1.0723032707217056e-06, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26550 + }, + { + "epoch": 7.727669479196974, + "flow_matching_loss": 0.012701408099383115, + "grad_norm": 0.08632136136293411, + "learning_rate": 1.066172406838558e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26560 + }, + { + "epoch": 7.730578993308118, + "flow_matching_loss": 0.01372673511505127, + "grad_norm": 0.08342651277780533, + "learning_rate": 1.0600584738698127e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26570 + }, + { + "epoch": 7.733488507419261, + "flow_matching_loss": 0.012027610000222922, + "grad_norm": 0.060280416160821915, + "learning_rate": 1.0539614792444735e-06, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26580 + }, + { + "epoch": 7.7363980215304045, + "flow_matching_loss": 0.012871930655092002, + "grad_norm": 0.06948684900999069, + "learning_rate": 1.0478814303709627e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26590 + }, + { + "epoch": 7.739307535641548, + "flow_matching_loss": 0.012626357562839984, + "grad_norm": 0.07391028106212616, + "learning_rate": 1.041818334637109e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26600 + }, + { + "epoch": 7.742217049752691, + "flow_matching_loss": 0.013215960375964642, + "grad_norm": 0.06805191189050674, + "learning_rate": 1.0357721994101526e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26610 + }, + { + "epoch": 7.745126563863835, + "flow_matching_loss": 0.011777885910123586, + "grad_norm": 0.06617675721645355, + "learning_rate": 1.0297430320367135e-06, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26620 + }, + { + "epoch": 7.748036077974978, + "flow_matching_loss": 0.014371013455092906, + "grad_norm": 0.06166217848658562, + "learning_rate": 1.0237308398428014e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26630 + }, + { + "epoch": 7.750945592086122, + "flow_matching_loss": 0.012715864833444357, + "grad_norm": 0.08461878448724747, + "learning_rate": 1.0177356301337925e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26640 + }, + { + "epoch": 7.753855106197265, + "flow_matching_loss": 0.012742830999195575, + "grad_norm": 0.07615501433610916, + "learning_rate": 1.0117574101944348e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26650 + }, + { + "epoch": 7.756764620308409, + "flow_matching_loss": 0.0142555829603225, + "grad_norm": 0.08782289177179337, + "learning_rate": 1.0057961872888293e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26660 + }, + { + "epoch": 7.7596741344195515, + "flow_matching_loss": 0.014334477111697197, + "grad_norm": 0.07511448115110397, + "learning_rate": 9.998519686604235e-07, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26670 + }, + { + "epoch": 7.762583648530695, + "flow_matching_loss": 0.013113170489668846, + "grad_norm": 0.07824164628982544, + "learning_rate": 9.939247615320047e-07, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26680 + }, + { + "epoch": 7.765493162641839, + "flow_matching_loss": 0.013630472868680955, + "grad_norm": 0.08294370025396347, + "learning_rate": 9.880145731056834e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26690 + }, + { + "epoch": 7.768402676752983, + "flow_matching_loss": 0.011904085148125887, + "grad_norm": 0.07212171703577042, + "learning_rate": 9.821214105629034e-07, + "loss": 0.0119, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26700 + }, + { + "epoch": 7.771312190864125, + "flow_matching_loss": 0.013267873041331767, + "grad_norm": 0.06830432265996933, + "learning_rate": 9.762452810644068e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26710 + }, + { + "epoch": 7.774221704975269, + "flow_matching_loss": 0.012770872190594673, + "grad_norm": 0.07221584022045135, + "learning_rate": 9.703861917502487e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26720 + }, + { + "epoch": 7.777131219086413, + "flow_matching_loss": 0.012591711012646556, + "grad_norm": 0.05683606490492821, + "learning_rate": 9.645441497397744e-07, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26730 + }, + { + "epoch": 7.780040733197556, + "flow_matching_loss": 0.012788553535938264, + "grad_norm": 0.05289946123957634, + "learning_rate": 9.587191621316154e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26740 + }, + { + "epoch": 7.782950247308699, + "flow_matching_loss": 0.01221606694161892, + "grad_norm": 0.07604807615280151, + "learning_rate": 9.529112360036802e-07, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26750 + }, + { + "epoch": 7.785859761419843, + "flow_matching_loss": 0.01271046306937933, + "grad_norm": 0.07585832476615906, + "learning_rate": 9.471203784131449e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26760 + }, + { + "epoch": 7.788769275530987, + "flow_matching_loss": 0.012375692836940289, + "grad_norm": 0.06289848685264587, + "learning_rate": 9.41346596396453e-07, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26770 + }, + { + "epoch": 7.79167878964213, + "flow_matching_loss": 0.013712527696043253, + "grad_norm": 0.07865127921104431, + "learning_rate": 9.355898969692889e-07, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26780 + }, + { + "epoch": 7.794588303753273, + "flow_matching_loss": 0.013340232986956834, + "grad_norm": 0.05315452069044113, + "learning_rate": 9.29850287126589e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26790 + }, + { + "epoch": 7.797497817864417, + "flow_matching_loss": 0.013532031606882811, + "grad_norm": 0.08113444596529007, + "learning_rate": 9.24127773842518e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26800 + }, + { + "epoch": 7.80040733197556, + "flow_matching_loss": 0.012681691348552704, + "grad_norm": 0.06716790795326233, + "learning_rate": 9.18422364070467e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26810 + }, + { + "epoch": 7.8033168460867035, + "flow_matching_loss": 0.013341965340077876, + "grad_norm": 0.06646155565977097, + "learning_rate": 9.127340647430538e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26820 + }, + { + "epoch": 7.806226360197847, + "flow_matching_loss": 0.011977728363126517, + "grad_norm": 0.08402863889932632, + "learning_rate": 9.070628827720929e-07, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26830 + }, + { + "epoch": 7.80913587430899, + "flow_matching_loss": 0.013303213939070701, + "grad_norm": 0.06793270260095596, + "learning_rate": 9.014088250486052e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26840 + }, + { + "epoch": 7.812045388420134, + "flow_matching_loss": 0.012101596593856812, + "grad_norm": 0.06043362617492676, + "learning_rate": 8.95771898442807e-07, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26850 + }, + { + "epoch": 7.8149549025312774, + "flow_matching_loss": 0.01237841546535492, + "grad_norm": 0.07232126593589783, + "learning_rate": 8.90152109804091e-07, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26860 + }, + { + "epoch": 7.81786441664242, + "flow_matching_loss": 0.01260833302512765, + "grad_norm": 0.0723879411816597, + "learning_rate": 8.845494659610365e-07, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26870 + }, + { + "epoch": 7.820773930753564, + "flow_matching_loss": 0.012015944626182318, + "grad_norm": 0.055900465697050095, + "learning_rate": 8.789639737213812e-07, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26880 + }, + { + "epoch": 7.823683444864708, + "flow_matching_loss": 0.0141375245526433, + "grad_norm": 0.10025962442159653, + "learning_rate": 8.733956398720244e-07, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26890 + }, + { + "epoch": 7.826592958975851, + "flow_matching_loss": 0.011661839764565229, + "grad_norm": 0.05434897169470787, + "learning_rate": 8.678444711790157e-07, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26900 + }, + { + "epoch": 7.829502473086994, + "flow_matching_loss": 0.01329895220696926, + "grad_norm": 0.07360409945249557, + "learning_rate": 8.623104743875493e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26910 + }, + { + "epoch": 7.832411987198138, + "flow_matching_loss": 0.011700825206935406, + "grad_norm": 0.06889282166957855, + "learning_rate": 8.567936562219547e-07, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26920 + }, + { + "epoch": 7.835321501309282, + "flow_matching_loss": 0.012411681888625025, + "grad_norm": 0.05514807626605034, + "learning_rate": 8.512940233856847e-07, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26930 + }, + { + "epoch": 7.838231015420424, + "flow_matching_loss": 0.013622082956135272, + "grad_norm": 0.07261635363101959, + "learning_rate": 8.458115825613122e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26940 + }, + { + "epoch": 7.841140529531568, + "flow_matching_loss": 0.013382064178586006, + "grad_norm": 0.06534697115421295, + "learning_rate": 8.403463404105199e-07, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26950 + }, + { + "epoch": 7.844050043642712, + "flow_matching_loss": 0.013899722788482904, + "grad_norm": 0.0829702615737915, + "learning_rate": 8.348983035740926e-07, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26960 + }, + { + "epoch": 7.846959557753856, + "flow_matching_loss": 0.01394154028967023, + "grad_norm": 0.06289325654506683, + "learning_rate": 8.294674786719064e-07, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26970 + }, + { + "epoch": 7.849869071864998, + "flow_matching_loss": 0.012227454036474229, + "grad_norm": 0.06706035882234573, + "learning_rate": 8.240538723029295e-07, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26980 + }, + { + "epoch": 7.852778585976142, + "flow_matching_loss": 0.014013869501650334, + "grad_norm": 0.09092912077903748, + "learning_rate": 8.186574910451983e-07, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 26990 + }, + { + "epoch": 7.855688100087286, + "flow_matching_loss": 0.013561711274087429, + "grad_norm": 0.07205277681350708, + "learning_rate": 8.132783414558326e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27000 + }, + { + "epoch": 7.858597614198429, + "flow_matching_loss": 0.012738823611289262, + "grad_norm": 0.05128224194049835, + "learning_rate": 8.079164300710023e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27010 + }, + { + "epoch": 7.861507128309572, + "flow_matching_loss": 0.012938710395246744, + "grad_norm": 0.07306075096130371, + "learning_rate": 8.025717634059326e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27020 + }, + { + "epoch": 7.864416642420716, + "flow_matching_loss": 0.012652888894081116, + "grad_norm": 0.050664253532886505, + "learning_rate": 7.972443479549035e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27030 + }, + { + "epoch": 7.867326156531859, + "flow_matching_loss": 0.013250813633203507, + "grad_norm": 0.06696998327970505, + "learning_rate": 7.919341901912236e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27040 + }, + { + "epoch": 7.870235670643003, + "flow_matching_loss": 0.01301035787910223, + "grad_norm": 0.07620339840650558, + "learning_rate": 7.866412965672365e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27050 + }, + { + "epoch": 7.873145184754146, + "flow_matching_loss": 0.013091527577489614, + "grad_norm": 0.05864829197525978, + "learning_rate": 7.813656735143044e-07, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27060 + }, + { + "epoch": 7.876054698865289, + "flow_matching_loss": 0.015439269319176673, + "grad_norm": 0.07775416225194931, + "learning_rate": 7.761073274428076e-07, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27070 + }, + { + "epoch": 7.878964212976433, + "flow_matching_loss": 0.01208119154907763, + "grad_norm": 0.07371184974908829, + "learning_rate": 7.708662647421355e-07, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27080 + }, + { + "epoch": 7.8818737270875765, + "flow_matching_loss": 0.013441320974379778, + "grad_norm": 0.07182905077934265, + "learning_rate": 7.65642491780672e-07, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27090 + }, + { + "epoch": 7.88478324119872, + "flow_matching_loss": 0.011770930513739586, + "grad_norm": 0.07131637632846832, + "learning_rate": 7.604360149057916e-07, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27100 + }, + { + "epoch": 7.887692755309863, + "flow_matching_loss": 0.012733687087893486, + "grad_norm": 0.06346842646598816, + "learning_rate": 7.552468404438568e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27110 + }, + { + "epoch": 7.890602269421007, + "flow_matching_loss": 0.012267770152539015, + "grad_norm": 0.08467470109462738, + "learning_rate": 7.500749747002022e-07, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27120 + }, + { + "epoch": 7.89351178353215, + "flow_matching_loss": 0.014162540528923273, + "grad_norm": 0.05807049199938774, + "learning_rate": 7.449204239591345e-07, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27130 + }, + { + "epoch": 7.896421297643293, + "flow_matching_loss": 0.012174832914024592, + "grad_norm": 0.06340792775154114, + "learning_rate": 7.397831944839167e-07, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27140 + }, + { + "epoch": 7.899330811754437, + "flow_matching_loss": 0.012255798745900392, + "grad_norm": 0.0591316893696785, + "learning_rate": 7.346632925167706e-07, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27150 + }, + { + "epoch": 7.902240325865581, + "flow_matching_loss": 0.012497935723513365, + "grad_norm": 0.07291005551815033, + "learning_rate": 7.295607242788566e-07, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27160 + }, + { + "epoch": 7.905149839976724, + "flow_matching_loss": 0.01368441954255104, + "grad_norm": 0.07010766863822937, + "learning_rate": 7.244754959702765e-07, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27170 + }, + { + "epoch": 7.908059354087867, + "flow_matching_loss": 0.013609485048800708, + "grad_norm": 0.09366802871227264, + "learning_rate": 7.194076137700628e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27180 + }, + { + "epoch": 7.910968868199011, + "flow_matching_loss": 0.012198489531874657, + "grad_norm": 0.07857147604227066, + "learning_rate": 7.143570838361696e-07, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27190 + }, + { + "epoch": 7.913878382310154, + "flow_matching_loss": 0.011665806733071804, + "grad_norm": 0.07387354224920273, + "learning_rate": 7.093239123054696e-07, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27200 + }, + { + "epoch": 7.916787896421297, + "flow_matching_loss": 0.012706049717962742, + "grad_norm": 0.06353069096803665, + "learning_rate": 7.043081052937389e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27210 + }, + { + "epoch": 7.919697410532441, + "flow_matching_loss": 0.01239713579416275, + "grad_norm": 0.06387545168399811, + "learning_rate": 6.993096688956558e-07, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27220 + }, + { + "epoch": 7.922606924643585, + "flow_matching_loss": 0.01392208132892847, + "grad_norm": 0.06471024453639984, + "learning_rate": 6.943286091847884e-07, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27230 + }, + { + "epoch": 7.925516438754728, + "flow_matching_loss": 0.013023110665380954, + "grad_norm": 0.08426330238580704, + "learning_rate": 6.89364932213602e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27240 + }, + { + "epoch": 7.928425952865871, + "flow_matching_loss": 0.01304463716223836, + "grad_norm": 0.06377990543842316, + "learning_rate": 6.844186440134253e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27250 + }, + { + "epoch": 7.931335466977015, + "flow_matching_loss": 0.011838026624172927, + "grad_norm": 0.05850611999630928, + "learning_rate": 6.794897505944692e-07, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27260 + }, + { + "epoch": 7.934244981088158, + "flow_matching_loss": 0.011643405631184578, + "grad_norm": 0.06821798533201218, + "learning_rate": 6.745782579457993e-07, + "loss": 0.0116, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27270 + }, + { + "epoch": 7.937154495199302, + "flow_matching_loss": 0.013462896458804607, + "grad_norm": 0.09847799688577652, + "learning_rate": 6.696841720353453e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27280 + }, + { + "epoch": 7.940064009310445, + "flow_matching_loss": 0.012438499554991721, + "grad_norm": 0.0605456717312336, + "learning_rate": 6.648074988098834e-07, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27290 + }, + { + "epoch": 7.942973523421589, + "flow_matching_loss": 0.012334281299263238, + "grad_norm": 0.08166225999593735, + "learning_rate": 6.599482441950322e-07, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27300 + }, + { + "epoch": 7.945883037532732, + "flow_matching_loss": 0.013688144274055958, + "grad_norm": 0.08016535639762878, + "learning_rate": 6.551064140952417e-07, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27310 + }, + { + "epoch": 7.9487925516438755, + "flow_matching_loss": 0.012790690921247006, + "grad_norm": 0.07792257517576218, + "learning_rate": 6.502820143937943e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27320 + }, + { + "epoch": 7.951702065755019, + "flow_matching_loss": 0.01342535400763154, + "grad_norm": 0.08440513908863068, + "learning_rate": 6.454750509527874e-07, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27330 + }, + { + "epoch": 7.954611579866162, + "flow_matching_loss": 0.01235794834792614, + "grad_norm": 0.08133941888809204, + "learning_rate": 6.406855296131387e-07, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27340 + }, + { + "epoch": 7.957521093977306, + "flow_matching_loss": 0.01319186231121421, + "grad_norm": 0.06799145042896271, + "learning_rate": 6.359134561945662e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27350 + }, + { + "epoch": 7.9604306080884495, + "flow_matching_loss": 0.01200198447331786, + "grad_norm": 0.060073595494031906, + "learning_rate": 6.311588364955883e-07, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27360 + }, + { + "epoch": 7.963340122199592, + "flow_matching_loss": 0.012581582553684712, + "grad_norm": 0.10197519510984421, + "learning_rate": 6.264216762935187e-07, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27370 + }, + { + "epoch": 7.966249636310736, + "flow_matching_loss": 0.012986408732831477, + "grad_norm": 0.06462494283914566, + "learning_rate": 6.21701981344453e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27380 + }, + { + "epoch": 7.96915915042188, + "flow_matching_loss": 0.014058680273592472, + "grad_norm": 0.058957260102033615, + "learning_rate": 6.16999757383262e-07, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27390 + }, + { + "epoch": 7.9720686645330225, + "flow_matching_loss": 0.011486744321882725, + "grad_norm": 0.056843101978302, + "learning_rate": 6.123150101235952e-07, + "loss": 0.0115, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27400 + }, + { + "epoch": 7.974978178644166, + "flow_matching_loss": 0.013243414368480444, + "grad_norm": 0.09613320976495743, + "learning_rate": 6.076477452578594e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27410 + }, + { + "epoch": 7.97788769275531, + "flow_matching_loss": 0.013841983955353498, + "grad_norm": 0.11645399034023285, + "learning_rate": 6.029979684572262e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27420 + }, + { + "epoch": 7.980797206866454, + "flow_matching_loss": 0.011886368552222849, + "grad_norm": 0.06327846646308899, + "learning_rate": 5.983656853716097e-07, + "loss": 0.0119, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27430 + }, + { + "epoch": 7.9837067209775965, + "flow_matching_loss": 0.013328510150313378, + "grad_norm": 0.07263582199811935, + "learning_rate": 5.937509016296689e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27440 + }, + { + "epoch": 7.98661623508874, + "flow_matching_loss": 0.01349923647940159, + "grad_norm": 0.07780887931585312, + "learning_rate": 5.89153622838805e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27450 + }, + { + "epoch": 7.989525749199884, + "flow_matching_loss": 0.012073251558467746, + "grad_norm": 0.045072682201862335, + "learning_rate": 5.845738545851431e-07, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27460 + }, + { + "epoch": 7.992435263311027, + "flow_matching_loss": 0.013600567076355218, + "grad_norm": 0.10172402113676071, + "learning_rate": 5.800116024335334e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27470 + }, + { + "epoch": 7.99534477742217, + "flow_matching_loss": 0.013160080462694169, + "grad_norm": 0.15186269581317902, + "learning_rate": 5.75466871927543e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27480 + }, + { + "epoch": 7.998254291533314, + "flow_matching_loss": 0.012103529646992683, + "grad_norm": 0.08182662725448608, + "learning_rate": 5.709396685894463e-07, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27490 + }, + { + "epoch": 8.001163805644458, + "flow_matching_loss": 0.010976522881537675, + "grad_norm": 0.06037335470318794, + "learning_rate": 5.664299979202264e-07, + "loss": 0.011, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27500 + }, + { + "epoch": 8.004073319755602, + "flow_matching_loss": 0.01377509068697691, + "grad_norm": 0.08376156538724899, + "learning_rate": 5.61937865399555e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27510 + }, + { + "epoch": 8.006982833866743, + "flow_matching_loss": 0.0131288249976933, + "grad_norm": 0.058316901326179504, + "learning_rate": 5.574632764857974e-07, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27520 + }, + { + "epoch": 8.009892347977887, + "flow_matching_loss": 0.013312409073114395, + "grad_norm": 0.060219451785087585, + "learning_rate": 5.53006236616001e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27530 + }, + { + "epoch": 8.012801862089031, + "flow_matching_loss": 0.012958084791898727, + "grad_norm": 0.07643022388219833, + "learning_rate": 5.485667512058873e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27540 + }, + { + "epoch": 8.015711376200175, + "flow_matching_loss": 0.013588641304522753, + "grad_norm": 0.06345991790294647, + "learning_rate": 5.441448256498544e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27550 + }, + { + "epoch": 8.018620890311318, + "flow_matching_loss": 0.011964343767613173, + "grad_norm": 0.07162240892648697, + "learning_rate": 5.397404653209559e-07, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27560 + }, + { + "epoch": 8.021530404422462, + "flow_matching_loss": 0.013982978090643883, + "grad_norm": 0.0637969970703125, + "learning_rate": 5.35353675570906e-07, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27570 + }, + { + "epoch": 8.024439918533606, + "flow_matching_loss": 0.012996282801032067, + "grad_norm": 0.07577679306268692, + "learning_rate": 5.309844617300674e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27580 + }, + { + "epoch": 8.027349432644748, + "flow_matching_loss": 0.012534740380942822, + "grad_norm": 0.06356759369373322, + "learning_rate": 5.266328291074468e-07, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27590 + }, + { + "epoch": 8.030258946755891, + "flow_matching_loss": 0.014026763383299112, + "grad_norm": 0.06428677588701248, + "learning_rate": 5.222987829906861e-07, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27600 + }, + { + "epoch": 8.033168460867035, + "flow_matching_loss": 0.012560094147920609, + "grad_norm": 0.08340828865766525, + "learning_rate": 5.179823286460644e-07, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27610 + }, + { + "epoch": 8.036077974978179, + "flow_matching_loss": 0.013333381339907645, + "grad_norm": 0.056733693927526474, + "learning_rate": 5.13683471318478e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27620 + }, + { + "epoch": 8.038987489089322, + "flow_matching_loss": 0.012102200090885163, + "grad_norm": 0.06672048568725586, + "learning_rate": 5.094022162314449e-07, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27630 + }, + { + "epoch": 8.041897003200466, + "flow_matching_loss": 0.01123367426916957, + "grad_norm": 0.07147487252950668, + "learning_rate": 5.051385685870924e-07, + "loss": 0.0112, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27640 + }, + { + "epoch": 8.044806517311608, + "flow_matching_loss": 0.013323393370956182, + "grad_norm": 0.11078888922929764, + "learning_rate": 5.008925335661546e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27650 + }, + { + "epoch": 8.047716031422752, + "flow_matching_loss": 0.014161836355924606, + "grad_norm": 0.07782618701457977, + "learning_rate": 4.966641163279662e-07, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27660 + }, + { + "epoch": 8.050625545533896, + "flow_matching_loss": 0.012251785770058633, + "grad_norm": 0.06411916762590408, + "learning_rate": 4.924533220104521e-07, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27670 + }, + { + "epoch": 8.05353505964504, + "flow_matching_loss": 0.01325656920671463, + "grad_norm": 0.10096710920333862, + "learning_rate": 4.882601557301231e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27680 + }, + { + "epoch": 8.056444573756183, + "flow_matching_loss": 0.01282265093177557, + "grad_norm": 0.07032284140586853, + "learning_rate": 4.840846225820717e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27690 + }, + { + "epoch": 8.059354087867327, + "flow_matching_loss": 0.013832969404757024, + "grad_norm": 0.06450319290161133, + "learning_rate": 4.79926727639966e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27700 + }, + { + "epoch": 8.06226360197847, + "flow_matching_loss": 0.014668611530214547, + "grad_norm": 0.06648196280002594, + "learning_rate": 4.7578647595603817e-07, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27710 + }, + { + "epoch": 8.065173116089612, + "flow_matching_loss": 0.012159868609160185, + "grad_norm": 0.06535237282514572, + "learning_rate": 4.716638725610889e-07, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27720 + }, + { + "epoch": 8.068082630200756, + "flow_matching_loss": 0.011569353844970464, + "grad_norm": 0.08388612419366837, + "learning_rate": 4.6755892246446606e-07, + "loss": 0.0116, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27730 + }, + { + "epoch": 8.0709921443119, + "flow_matching_loss": 0.012968243844807149, + "grad_norm": 0.06443262100219727, + "learning_rate": 4.634716306540715e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27740 + }, + { + "epoch": 8.073901658423043, + "flow_matching_loss": 0.013094702549278737, + "grad_norm": 0.07338638603687286, + "learning_rate": 4.5940200209635244e-07, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27750 + }, + { + "epoch": 8.076811172534187, + "flow_matching_loss": 0.013768563978374004, + "grad_norm": 0.077033132314682, + "learning_rate": 4.5535004173628657e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27760 + }, + { + "epoch": 8.07972068664533, + "flow_matching_loss": 0.012963672261685133, + "grad_norm": 0.058653004467487335, + "learning_rate": 4.5131575449739225e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27770 + }, + { + "epoch": 8.082630200756475, + "flow_matching_loss": 0.012479779217392207, + "grad_norm": 0.07911240309476852, + "learning_rate": 4.472991452817099e-07, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27780 + }, + { + "epoch": 8.085539714867616, + "flow_matching_loss": 0.013357452303171157, + "grad_norm": 0.0876830667257309, + "learning_rate": 4.433002189697938e-07, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27790 + }, + { + "epoch": 8.08844922897876, + "flow_matching_loss": 0.01290185023099184, + "grad_norm": 0.08071833103895187, + "learning_rate": 4.393189804207187e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27800 + }, + { + "epoch": 8.091358743089904, + "flow_matching_loss": 0.013959024753421544, + "grad_norm": 0.071235790848732, + "learning_rate": 4.353554344720651e-07, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27810 + }, + { + "epoch": 8.094268257201048, + "flow_matching_loss": 0.014391971472650766, + "grad_norm": 0.0810093954205513, + "learning_rate": 4.314095859399153e-07, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27820 + }, + { + "epoch": 8.097177771312191, + "flow_matching_loss": 0.012666654586791993, + "grad_norm": 0.08817046880722046, + "learning_rate": 4.274814396188459e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27830 + }, + { + "epoch": 8.100087285423335, + "flow_matching_loss": 0.013068395014852286, + "grad_norm": 0.06504296511411667, + "learning_rate": 4.235710002819271e-07, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27840 + }, + { + "epoch": 8.102996799534477, + "flow_matching_loss": 0.012955585308372975, + "grad_norm": 0.06966384500265121, + "learning_rate": 4.196782726807097e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27850 + }, + { + "epoch": 8.10590631364562, + "flow_matching_loss": 0.012858011573553086, + "grad_norm": 0.08155141025781631, + "learning_rate": 4.158032615452251e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27860 + }, + { + "epoch": 8.108815827756764, + "flow_matching_loss": 0.013975854031741619, + "grad_norm": 0.0736677423119545, + "learning_rate": 4.1194597158397853e-07, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27870 + }, + { + "epoch": 8.111725341867908, + "flow_matching_loss": 0.012838307488709687, + "grad_norm": 0.05996353551745415, + "learning_rate": 4.081064074839408e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27880 + }, + { + "epoch": 8.114634855979052, + "flow_matching_loss": 0.012813475821167231, + "grad_norm": 0.08311943709850311, + "learning_rate": 4.042845739105433e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27890 + }, + { + "epoch": 8.117544370090195, + "flow_matching_loss": 0.01282686684280634, + "grad_norm": 0.10000535100698471, + "learning_rate": 4.004804755076746e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27900 + }, + { + "epoch": 8.12045388420134, + "flow_matching_loss": 0.01199403554201126, + "grad_norm": 0.060644447803497314, + "learning_rate": 3.9669411689767063e-07, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27910 + }, + { + "epoch": 8.123363398312481, + "flow_matching_loss": 0.012610387476161122, + "grad_norm": 0.0671173706650734, + "learning_rate": 3.929255026813194e-07, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27920 + }, + { + "epoch": 8.126272912423625, + "flow_matching_loss": 0.012355474382638931, + "grad_norm": 0.06751329451799393, + "learning_rate": 3.8917463743783797e-07, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27930 + }, + { + "epoch": 8.129182426534769, + "flow_matching_loss": 0.013383136224001646, + "grad_norm": 0.06770852208137512, + "learning_rate": 3.854415257248839e-07, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27940 + }, + { + "epoch": 8.132091940645912, + "flow_matching_loss": 0.01156202405691147, + "grad_norm": 0.06095369905233383, + "learning_rate": 3.8172617207853864e-07, + "loss": 0.0116, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27950 + }, + { + "epoch": 8.135001454757056, + "flow_matching_loss": 0.012757162936031819, + "grad_norm": 0.08495932817459106, + "learning_rate": 3.7802858101330773e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27960 + }, + { + "epoch": 8.1379109688682, + "flow_matching_loss": 0.013105278555303811, + "grad_norm": 0.07662546634674072, + "learning_rate": 3.743487570221121e-07, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27970 + }, + { + "epoch": 8.140820482979342, + "flow_matching_loss": 0.013023317977786063, + "grad_norm": 0.0981823280453682, + "learning_rate": 3.7068670457628517e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27980 + }, + { + "epoch": 8.143729997090485, + "flow_matching_loss": 0.013122931867837907, + "grad_norm": 0.06958580017089844, + "learning_rate": 3.6704242812556576e-07, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 27990 + }, + { + "epoch": 8.146639511201629, + "flow_matching_loss": 0.013188391644507646, + "grad_norm": 0.09407681971788406, + "learning_rate": 3.634159320980951e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28000 + }, + { + "epoch": 8.149549025312773, + "flow_matching_loss": 0.012895617354661226, + "grad_norm": 0.06586829572916031, + "learning_rate": 3.5980722090040663e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28010 + }, + { + "epoch": 8.152458539423916, + "flow_matching_loss": 0.012326311226934195, + "grad_norm": 0.07035885751247406, + "learning_rate": 3.5621629891742447e-07, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28020 + }, + { + "epoch": 8.15536805353506, + "flow_matching_loss": 0.014644635748118163, + "grad_norm": 0.06775034219026566, + "learning_rate": 3.526431705124583e-07, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28030 + }, + { + "epoch": 8.158277567646204, + "flow_matching_loss": 0.014984537940472364, + "grad_norm": 0.08339616656303406, + "learning_rate": 3.4908784002719517e-07, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28040 + }, + { + "epoch": 8.161187081757346, + "flow_matching_loss": 0.013268484454602002, + "grad_norm": 0.09428761899471283, + "learning_rate": 3.4555031178169773e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28050 + }, + { + "epoch": 8.16409659586849, + "flow_matching_loss": 0.013878103997558355, + "grad_norm": 0.0954204574227333, + "learning_rate": 3.420305900743942e-07, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28060 + }, + { + "epoch": 8.167006109979633, + "flow_matching_loss": 0.013244531862437726, + "grad_norm": 0.06296652555465698, + "learning_rate": 3.385286791820802e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28070 + }, + { + "epoch": 8.169915624090777, + "flow_matching_loss": 0.013858631812036037, + "grad_norm": 0.084579236805439, + "learning_rate": 3.350445833599086e-07, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28080 + }, + { + "epoch": 8.17282513820192, + "flow_matching_loss": 0.011442597769200803, + "grad_norm": 0.06510187685489655, + "learning_rate": 3.3157830684138303e-07, + "loss": 0.0114, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28090 + }, + { + "epoch": 8.175734652313064, + "flow_matching_loss": 0.013250317052006721, + "grad_norm": 0.06868989020586014, + "learning_rate": 3.281298538383559e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28100 + }, + { + "epoch": 8.178644166424208, + "flow_matching_loss": 0.013215628173202276, + "grad_norm": 0.06951950490474701, + "learning_rate": 3.2469922854102387e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28110 + }, + { + "epoch": 8.18155368053535, + "flow_matching_loss": 0.014493253082036972, + "grad_norm": 0.07312360405921936, + "learning_rate": 3.21286435117919e-07, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28120 + }, + { + "epoch": 8.184463194646494, + "flow_matching_loss": 0.012522770930081606, + "grad_norm": 0.07457323372364044, + "learning_rate": 3.1789147771590754e-07, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28130 + }, + { + "epoch": 8.187372708757637, + "flow_matching_loss": 0.01359171224758029, + "grad_norm": 0.04700348153710365, + "learning_rate": 3.14514360460183e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28140 + }, + { + "epoch": 8.190282222868781, + "flow_matching_loss": 0.012532502226531506, + "grad_norm": 0.06821933388710022, + "learning_rate": 3.1115508745426293e-07, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28150 + }, + { + "epoch": 8.193191736979925, + "flow_matching_loss": 0.013746024947613478, + "grad_norm": 0.08348139375448227, + "learning_rate": 3.0781366277997716e-07, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28160 + }, + { + "epoch": 8.196101251091068, + "flow_matching_loss": 0.01425830963999033, + "grad_norm": 0.05789078772068024, + "learning_rate": 3.044900904974762e-07, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28170 + }, + { + "epoch": 8.19901076520221, + "flow_matching_loss": 0.01273475270718336, + "grad_norm": 0.06216936558485031, + "learning_rate": 3.011843746452081e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28180 + }, + { + "epoch": 8.201920279313354, + "flow_matching_loss": 0.015218209475278854, + "grad_norm": 0.06656448543071747, + "learning_rate": 2.9789651923993464e-07, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28190 + }, + { + "epoch": 8.204829793424498, + "flow_matching_loss": 0.01184342335909605, + "grad_norm": 0.05962615832686424, + "learning_rate": 2.9462652827670855e-07, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28200 + }, + { + "epoch": 8.207739307535642, + "flow_matching_loss": 0.013213720172643661, + "grad_norm": 0.08668399602174759, + "learning_rate": 2.913744057288764e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28210 + }, + { + "epoch": 8.210648821646785, + "flow_matching_loss": 0.014583468157798051, + "grad_norm": 0.1074739322066307, + "learning_rate": 2.8814015554807396e-07, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28220 + }, + { + "epoch": 8.213558335757929, + "flow_matching_loss": 0.013093191105872393, + "grad_norm": 0.055379968136548996, + "learning_rate": 2.849237816642192e-07, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28230 + }, + { + "epoch": 8.216467849869073, + "flow_matching_loss": 0.012606904841959476, + "grad_norm": 0.06330244988203049, + "learning_rate": 2.817252879855109e-07, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28240 + }, + { + "epoch": 8.219377363980215, + "flow_matching_loss": 0.012128066271543503, + "grad_norm": 0.07152049988508224, + "learning_rate": 2.785446783984219e-07, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28250 + }, + { + "epoch": 8.222286878091358, + "flow_matching_loss": 0.011026451829820871, + "grad_norm": 0.06426088511943817, + "learning_rate": 2.7538195676769076e-07, + "loss": 0.011, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28260 + }, + { + "epoch": 8.225196392202502, + "flow_matching_loss": 0.013621341623365878, + "grad_norm": 0.06925376504659653, + "learning_rate": 2.722371269363216e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28270 + }, + { + "epoch": 8.228105906313646, + "flow_matching_loss": 0.013530756160616875, + "grad_norm": 0.06542079150676727, + "learning_rate": 2.6911019272558125e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28280 + }, + { + "epoch": 8.23101542042479, + "flow_matching_loss": 0.011573642492294312, + "grad_norm": 0.07104208320379257, + "learning_rate": 2.6600115793498857e-07, + "loss": 0.0116, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28290 + }, + { + "epoch": 8.233924934535933, + "flow_matching_loss": 0.013519202172756196, + "grad_norm": 0.09723030030727386, + "learning_rate": 2.629100263423151e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28300 + }, + { + "epoch": 8.236834448647077, + "flow_matching_loss": 0.011686358274891973, + "grad_norm": 0.06337190419435501, + "learning_rate": 2.598368017035763e-07, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28310 + }, + { + "epoch": 8.239743962758219, + "flow_matching_loss": 0.01355264289304614, + "grad_norm": 0.07275745272636414, + "learning_rate": 2.5678148775303e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28320 + }, + { + "epoch": 8.242653476869362, + "flow_matching_loss": 0.012549559026956559, + "grad_norm": 0.07644246518611908, + "learning_rate": 2.537440882031683e-07, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28330 + }, + { + "epoch": 8.245562990980506, + "flow_matching_loss": 0.0135580369271338, + "grad_norm": 0.06513189524412155, + "learning_rate": 2.5072460674472385e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28340 + }, + { + "epoch": 8.24847250509165, + "flow_matching_loss": 0.011599574983119965, + "grad_norm": 0.07351565361022949, + "learning_rate": 2.4772304704664683e-07, + "loss": 0.0116, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28350 + }, + { + "epoch": 8.251382019202794, + "flow_matching_loss": 0.012605345621705056, + "grad_norm": 0.08159194886684418, + "learning_rate": 2.447394127561181e-07, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28360 + }, + { + "epoch": 8.254291533313937, + "flow_matching_loss": 0.013568165712058544, + "grad_norm": 0.07246667891740799, + "learning_rate": 2.41773707498531e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28370 + }, + { + "epoch": 8.25720104742508, + "flow_matching_loss": 0.012622840888798236, + "grad_norm": 0.07399924099445343, + "learning_rate": 2.388259348774996e-07, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28380 + }, + { + "epoch": 8.260110561536223, + "flow_matching_loss": 0.01289844149723649, + "grad_norm": 0.07642663270235062, + "learning_rate": 2.358960984748454e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28390 + }, + { + "epoch": 8.263020075647367, + "flow_matching_loss": 0.011633377242833375, + "grad_norm": 0.050575993955135345, + "learning_rate": 2.3298420185059565e-07, + "loss": 0.0116, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28400 + }, + { + "epoch": 8.26592958975851, + "flow_matching_loss": 0.013370336405932904, + "grad_norm": 0.0717485323548317, + "learning_rate": 2.3009024854297844e-07, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28410 + }, + { + "epoch": 8.268839103869654, + "flow_matching_loss": 0.012039550207555294, + "grad_norm": 0.07514847815036774, + "learning_rate": 2.272142420684209e-07, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28420 + }, + { + "epoch": 8.271748617980798, + "flow_matching_loss": 0.013867509271949529, + "grad_norm": 0.08123744279146194, + "learning_rate": 2.243561859215426e-07, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28430 + }, + { + "epoch": 8.274658132091941, + "flow_matching_loss": 0.012806138303130866, + "grad_norm": 0.06769362092018127, + "learning_rate": 2.2151608357514897e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28440 + }, + { + "epoch": 8.277567646203083, + "flow_matching_loss": 0.012290109973400831, + "grad_norm": 0.06167908385396004, + "learning_rate": 2.186939384802361e-07, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28450 + }, + { + "epoch": 8.280477160314227, + "flow_matching_loss": 0.012751576770097018, + "grad_norm": 0.0783676877617836, + "learning_rate": 2.1588975406597266e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28460 + }, + { + "epoch": 8.28338667442537, + "flow_matching_loss": 0.012558369617909192, + "grad_norm": 0.06741730868816376, + "learning_rate": 2.131035337397097e-07, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28470 + }, + { + "epoch": 8.286296188536515, + "flow_matching_loss": 0.013786687701940536, + "grad_norm": 0.07224731892347336, + "learning_rate": 2.1033528088696575e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28480 + }, + { + "epoch": 8.289205702647658, + "flow_matching_loss": 0.012525017373263836, + "grad_norm": 0.08348555862903595, + "learning_rate": 2.0758499887143012e-07, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28490 + }, + { + "epoch": 8.292115216758802, + "flow_matching_loss": 0.01271587861701846, + "grad_norm": 0.06260176002979279, + "learning_rate": 2.0485269103495462e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28500 + }, + { + "epoch": 8.295024730869944, + "flow_matching_loss": 0.012587063387036324, + "grad_norm": 0.07085982710123062, + "learning_rate": 2.021383606975502e-07, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28510 + }, + { + "epoch": 8.297934244981088, + "flow_matching_loss": 0.012246784474700689, + "grad_norm": 0.06609652191400528, + "learning_rate": 1.9944201115738524e-07, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28520 + }, + { + "epoch": 8.300843759092231, + "flow_matching_loss": 0.012722001224756242, + "grad_norm": 0.06083147972822189, + "learning_rate": 1.9676364569077732e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28530 + }, + { + "epoch": 8.303753273203375, + "flow_matching_loss": 0.012918277457356454, + "grad_norm": 0.0665624588727951, + "learning_rate": 1.9410326755219477e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28540 + }, + { + "epoch": 8.306662787314519, + "flow_matching_loss": 0.01184877110645175, + "grad_norm": 0.0912712886929512, + "learning_rate": 1.9146087997424677e-07, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28550 + }, + { + "epoch": 8.309572301425662, + "flow_matching_loss": 0.013595526944845915, + "grad_norm": 0.07156643271446228, + "learning_rate": 1.8883648616768335e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28560 + }, + { + "epoch": 8.312481815536806, + "flow_matching_loss": 0.012883279100060463, + "grad_norm": 0.05490626394748688, + "learning_rate": 1.862300893213903e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28570 + }, + { + "epoch": 8.315391329647948, + "flow_matching_loss": 0.011970263067632913, + "grad_norm": 0.09323376417160034, + "learning_rate": 1.8364169260238595e-07, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28580 + }, + { + "epoch": 8.318300843759092, + "flow_matching_loss": 0.01355331065133214, + "grad_norm": 0.06173180788755417, + "learning_rate": 1.8107129915581777e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28590 + }, + { + "epoch": 8.321210357870235, + "flow_matching_loss": 0.012189502408728004, + "grad_norm": 0.07750650495290756, + "learning_rate": 1.785189121049541e-07, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28600 + }, + { + "epoch": 8.32411987198138, + "flow_matching_loss": 0.01266820365563035, + "grad_norm": 0.07237403839826584, + "learning_rate": 1.7598453455118567e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28610 + }, + { + "epoch": 8.327029386092523, + "flow_matching_loss": 0.01433944869786501, + "grad_norm": 0.06966681778430939, + "learning_rate": 1.7346816957402256e-07, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28620 + }, + { + "epoch": 8.329938900203667, + "flow_matching_loss": 0.012249280232936144, + "grad_norm": 0.07230470329523087, + "learning_rate": 1.7096982023108389e-07, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28630 + }, + { + "epoch": 8.33284841431481, + "flow_matching_loss": 0.012726344633847475, + "grad_norm": 0.057123810052871704, + "learning_rate": 1.6848948955809963e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28640 + }, + { + "epoch": 8.335757928425952, + "flow_matching_loss": 0.013213436212390661, + "grad_norm": 0.07276846468448639, + "learning_rate": 1.66027180568904e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28650 + }, + { + "epoch": 8.338667442537096, + "flow_matching_loss": 0.013235205598175526, + "grad_norm": 0.0581718347966671, + "learning_rate": 1.6358289625543865e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28660 + }, + { + "epoch": 8.34157695664824, + "flow_matching_loss": 0.01274622669443488, + "grad_norm": 0.05975045636296272, + "learning_rate": 1.6115663958773618e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28670 + }, + { + "epoch": 8.344486470759383, + "flow_matching_loss": 0.013512833695858716, + "grad_norm": 0.07526625692844391, + "learning_rate": 1.5874841351392832e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28680 + }, + { + "epoch": 8.347395984870527, + "flow_matching_loss": 0.015495777316391467, + "grad_norm": 0.09914402663707733, + "learning_rate": 1.56358220960236e-07, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28690 + }, + { + "epoch": 8.35030549898167, + "flow_matching_loss": 0.012382709234952927, + "grad_norm": 0.07004306465387344, + "learning_rate": 1.5398606483096932e-07, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28700 + }, + { + "epoch": 8.353215013092813, + "flow_matching_loss": 0.01441841321066022, + "grad_norm": 0.11182279139757156, + "learning_rate": 1.5163194800852265e-07, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28710 + }, + { + "epoch": 8.356124527203956, + "flow_matching_loss": 0.014612728357315063, + "grad_norm": 0.09309599548578262, + "learning_rate": 1.4929587335336614e-07, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28720 + }, + { + "epoch": 8.3590340413151, + "flow_matching_loss": 0.012807410582900048, + "grad_norm": 0.07489442080259323, + "learning_rate": 1.4697784370405587e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28730 + }, + { + "epoch": 8.361943555426244, + "flow_matching_loss": 0.012662817630916834, + "grad_norm": 0.07617061585187912, + "learning_rate": 1.4467786187721376e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28740 + }, + { + "epoch": 8.364853069537387, + "flow_matching_loss": 0.012615935876965523, + "grad_norm": 0.06837426871061325, + "learning_rate": 1.4239593066753265e-07, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28750 + }, + { + "epoch": 8.367762583648531, + "flow_matching_loss": 0.012145032966509461, + "grad_norm": 0.06263464689254761, + "learning_rate": 1.4013205284777953e-07, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28760 + }, + { + "epoch": 8.370672097759675, + "flow_matching_loss": 0.012858329527080058, + "grad_norm": 0.07096876949071884, + "learning_rate": 1.3788623116877397e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28770 + }, + { + "epoch": 8.373581611870817, + "flow_matching_loss": 0.012683265656232835, + "grad_norm": 0.06653342396020889, + "learning_rate": 1.3565846835940477e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28780 + }, + { + "epoch": 8.37649112598196, + "flow_matching_loss": 0.013042287994176149, + "grad_norm": 0.09980694204568863, + "learning_rate": 1.334487671266116e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28790 + }, + { + "epoch": 8.379400640093104, + "flow_matching_loss": 0.012066153204068541, + "grad_norm": 0.07713396847248077, + "learning_rate": 1.3125713015538998e-07, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28800 + }, + { + "epoch": 8.382310154204248, + "flow_matching_loss": 0.0128202011808753, + "grad_norm": 0.07240042835474014, + "learning_rate": 1.290835601087864e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28810 + }, + { + "epoch": 8.385219668315392, + "flow_matching_loss": 0.012177428510040045, + "grad_norm": 0.06319954991340637, + "learning_rate": 1.2692805962789323e-07, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28820 + }, + { + "epoch": 8.388129182426535, + "flow_matching_loss": 0.012377765867859125, + "grad_norm": 0.05570422112941742, + "learning_rate": 1.2479063133184698e-07, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28830 + }, + { + "epoch": 8.391038696537677, + "flow_matching_loss": 0.012061446346342564, + "grad_norm": 0.07644127309322357, + "learning_rate": 1.2267127781782516e-07, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28840 + }, + { + "epoch": 8.393948210648821, + "flow_matching_loss": 0.013047231175005437, + "grad_norm": 0.08634291589260101, + "learning_rate": 1.205700016610395e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28850 + }, + { + "epoch": 8.396857724759965, + "flow_matching_loss": 0.012642847094684839, + "grad_norm": 0.06040705367922783, + "learning_rate": 1.1848680541474088e-07, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28860 + }, + { + "epoch": 8.399767238871108, + "flow_matching_loss": 0.012642869073897601, + "grad_norm": 0.06766446679830551, + "learning_rate": 1.1642169161020955e-07, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28870 + }, + { + "epoch": 8.402676752982252, + "flow_matching_loss": 0.012465938832610846, + "grad_norm": 0.08120467513799667, + "learning_rate": 1.1437466275675156e-07, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28880 + }, + { + "epoch": 8.405586267093396, + "flow_matching_loss": 0.013501220662146807, + "grad_norm": 0.05903855711221695, + "learning_rate": 1.1234572134170229e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28890 + }, + { + "epoch": 8.40849578120454, + "flow_matching_loss": 0.012055949214845896, + "grad_norm": 0.06431137025356293, + "learning_rate": 1.1033486983041463e-07, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28900 + }, + { + "epoch": 8.411405295315681, + "flow_matching_loss": 0.012819670606404543, + "grad_norm": 0.06793515384197235, + "learning_rate": 1.0834211066626409e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28910 + }, + { + "epoch": 8.414314809426825, + "flow_matching_loss": 0.011801788024604321, + "grad_norm": 0.06529086083173752, + "learning_rate": 1.0636744627064044e-07, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28920 + }, + { + "epoch": 8.417224323537969, + "flow_matching_loss": 0.013998137786984444, + "grad_norm": 0.05944078043103218, + "learning_rate": 1.0441087904294766e-07, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28930 + }, + { + "epoch": 8.420133837649113, + "flow_matching_loss": 0.011432184465229511, + "grad_norm": 0.05693169683218002, + "learning_rate": 1.0247241136059904e-07, + "loss": 0.0114, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28940 + }, + { + "epoch": 8.423043351760256, + "flow_matching_loss": 0.012852784991264344, + "grad_norm": 0.07386791706085205, + "learning_rate": 1.0055204557901709e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28950 + }, + { + "epoch": 8.4259528658714, + "flow_matching_loss": 0.012738364934921264, + "grad_norm": 0.071060411632061, + "learning_rate": 9.864978403162361e-08, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28960 + }, + { + "epoch": 8.428862379982544, + "flow_matching_loss": 0.013066381588578225, + "grad_norm": 0.06994062662124634, + "learning_rate": 9.676562902984965e-08, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28970 + }, + { + "epoch": 8.431771894093686, + "flow_matching_loss": 0.01191741987131536, + "grad_norm": 0.07181645184755325, + "learning_rate": 9.489958286311883e-08, + "loss": 0.0119, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28980 + }, + { + "epoch": 8.43468140820483, + "flow_matching_loss": 0.012062098365277052, + "grad_norm": 0.05911954492330551, + "learning_rate": 9.30516477988541e-08, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 28990 + }, + { + "epoch": 8.437590922315973, + "flow_matching_loss": 0.01264209058135748, + "grad_norm": 0.07594984769821167, + "learning_rate": 9.122182608247098e-08, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29000 + }, + { + "epoch": 8.440500436427117, + "flow_matching_loss": 0.012350694090127946, + "grad_norm": 0.05738458409905434, + "learning_rate": 8.94101199373759e-08, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29010 + }, + { + "epoch": 8.44340995053826, + "flow_matching_loss": 0.011987593956291676, + "grad_norm": 0.056578561663627625, + "learning_rate": 8.761653156495964e-08, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29020 + }, + { + "epoch": 8.446319464649404, + "flow_matching_loss": 0.013049287442117929, + "grad_norm": 0.07732746005058289, + "learning_rate": 8.584106314460549e-08, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29030 + }, + { + "epoch": 8.449228978760548, + "flow_matching_loss": 0.013672502245754004, + "grad_norm": 0.07627490162849426, + "learning_rate": 8.408371683367111e-08, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29040 + }, + { + "epoch": 8.45213849287169, + "flow_matching_loss": 0.012105375621467828, + "grad_norm": 0.06678113341331482, + "learning_rate": 8.234449476750172e-08, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29050 + }, + { + "epoch": 8.455048006982834, + "flow_matching_loss": 0.012913932371884585, + "grad_norm": 0.06250522285699844, + "learning_rate": 8.062339905941351e-08, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29060 + }, + { + "epoch": 8.457957521093977, + "flow_matching_loss": 0.013064093980938196, + "grad_norm": 0.06952090561389923, + "learning_rate": 7.89204318007003e-08, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29070 + }, + { + "epoch": 8.460867035205121, + "flow_matching_loss": 0.011925913207232953, + "grad_norm": 0.06045028194785118, + "learning_rate": 7.723559506062849e-08, + "loss": 0.0119, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29080 + }, + { + "epoch": 8.463776549316265, + "flow_matching_loss": 0.01306575033813715, + "grad_norm": 0.054545897990465164, + "learning_rate": 7.556889088643381e-08, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29090 + }, + { + "epoch": 8.466686063427408, + "flow_matching_loss": 0.013650075066834688, + "grad_norm": 0.09854549914598465, + "learning_rate": 7.392032130331793e-08, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29100 + }, + { + "epoch": 8.46959557753855, + "flow_matching_loss": 0.015546680241823197, + "grad_norm": 0.08914481103420258, + "learning_rate": 7.228988831445015e-08, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29110 + }, + { + "epoch": 8.472505091649694, + "flow_matching_loss": 0.012990179564803838, + "grad_norm": 0.07046489417552948, + "learning_rate": 7.06775939009574e-08, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29120 + }, + { + "epoch": 8.475414605760838, + "flow_matching_loss": 0.013071298552677035, + "grad_norm": 0.048914603888988495, + "learning_rate": 6.908344002193256e-08, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29130 + }, + { + "epoch": 8.478324119871981, + "flow_matching_loss": 0.011695770639926195, + "grad_norm": 0.0493750236928463, + "learning_rate": 6.750742861442116e-08, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29140 + }, + { + "epoch": 8.481233633983125, + "flow_matching_loss": 0.012090576719492674, + "grad_norm": 0.05984507501125336, + "learning_rate": 6.594956159342636e-08, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29150 + }, + { + "epoch": 8.484143148094269, + "flow_matching_loss": 0.01382071441039443, + "grad_norm": 0.05642218887805939, + "learning_rate": 6.440984085190394e-08, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29160 + }, + { + "epoch": 8.487052662205413, + "flow_matching_loss": 0.012403417937457561, + "grad_norm": 0.058013807982206345, + "learning_rate": 6.288826826075733e-08, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29170 + }, + { + "epoch": 8.489962176316554, + "flow_matching_loss": 0.013050282746553421, + "grad_norm": 0.07396427541971207, + "learning_rate": 6.138484566884261e-08, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29180 + }, + { + "epoch": 8.492871690427698, + "flow_matching_loss": 0.012217097543179989, + "grad_norm": 0.06211281195282936, + "learning_rate": 5.989957490296183e-08, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29190 + }, + { + "epoch": 8.495781204538842, + "flow_matching_loss": 0.011936973966658115, + "grad_norm": 0.06047975271940231, + "learning_rate": 5.843245776785466e-08, + "loss": 0.0119, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29200 + }, + { + "epoch": 8.498690718649986, + "flow_matching_loss": 0.012397396750748158, + "grad_norm": 0.06770820170640945, + "learning_rate": 5.6983496046210115e-08, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29210 + }, + { + "epoch": 8.50160023276113, + "flow_matching_loss": 0.012652566749602556, + "grad_norm": 0.06579840928316116, + "learning_rate": 5.5552691498653165e-08, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29220 + }, + { + "epoch": 8.504509746872273, + "flow_matching_loss": 0.012185069266706705, + "grad_norm": 0.05626003444194794, + "learning_rate": 5.4140045863743106e-08, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29230 + }, + { + "epoch": 8.507419260983415, + "flow_matching_loss": 0.013780882116407156, + "grad_norm": 0.06347176432609558, + "learning_rate": 5.274556085798188e-08, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29240 + }, + { + "epoch": 8.510328775094559, + "flow_matching_loss": 0.012242414522916079, + "grad_norm": 0.055616140365600586, + "learning_rate": 5.136923817579742e-08, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29250 + }, + { + "epoch": 8.513238289205702, + "flow_matching_loss": 0.012812222260981799, + "grad_norm": 0.07311689108610153, + "learning_rate": 5.001107948955197e-08, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29260 + }, + { + "epoch": 8.516147803316846, + "flow_matching_loss": 0.012554540857672691, + "grad_norm": 0.06727994978427887, + "learning_rate": 4.867108644953711e-08, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29270 + }, + { + "epoch": 8.51905731742799, + "flow_matching_loss": 0.014720562659204007, + "grad_norm": 0.07841529697179794, + "learning_rate": 4.734926068397038e-08, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29280 + }, + { + "epoch": 8.521966831539133, + "flow_matching_loss": 0.014147276151925325, + "grad_norm": 0.06437932699918747, + "learning_rate": 4.604560379899536e-08, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29290 + }, + { + "epoch": 8.524876345650277, + "flow_matching_loss": 0.013111739978194236, + "grad_norm": 0.06689436733722687, + "learning_rate": 4.476011737867658e-08, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29300 + }, + { + "epoch": 8.52778585976142, + "flow_matching_loss": 0.013202043808996677, + "grad_norm": 0.05986233428120613, + "learning_rate": 4.349280298500291e-08, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29310 + }, + { + "epoch": 8.530695373872563, + "flow_matching_loss": 0.012906148005276919, + "grad_norm": 0.09957590699195862, + "learning_rate": 4.224366215787923e-08, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29320 + }, + { + "epoch": 8.533604887983707, + "flow_matching_loss": 0.011154662817716599, + "grad_norm": 0.07120413333177567, + "learning_rate": 4.101269641513139e-08, + "loss": 0.0112, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29330 + }, + { + "epoch": 8.53651440209485, + "flow_matching_loss": 0.014442384522408248, + "grad_norm": 0.07022010535001755, + "learning_rate": 3.979990725249627e-08, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29340 + }, + { + "epoch": 8.539423916205994, + "flow_matching_loss": 0.014020628295838833, + "grad_norm": 0.07072775810956955, + "learning_rate": 3.860529614363006e-08, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29350 + }, + { + "epoch": 8.542333430317138, + "flow_matching_loss": 0.013012716360390187, + "grad_norm": 0.06555414944887161, + "learning_rate": 3.742886454009664e-08, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29360 + }, + { + "epoch": 8.545242944428281, + "flow_matching_loss": 0.011351094581186771, + "grad_norm": 0.06440197676420212, + "learning_rate": 3.627061387137087e-08, + "loss": 0.0114, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29370 + }, + { + "epoch": 8.548152458539423, + "flow_matching_loss": 0.013685052189975977, + "grad_norm": 0.0823453813791275, + "learning_rate": 3.5130545544835305e-08, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29380 + }, + { + "epoch": 8.551061972650567, + "flow_matching_loss": 0.013926566205918789, + "grad_norm": 0.07791199535131454, + "learning_rate": 3.400866094578348e-08, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29390 + }, + { + "epoch": 8.55397148676171, + "flow_matching_loss": 0.012910888250917196, + "grad_norm": 0.05855482444167137, + "learning_rate": 3.290496143740829e-08, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29400 + }, + { + "epoch": 8.556881000872854, + "flow_matching_loss": 0.01568738454952836, + "grad_norm": 0.06846615672111511, + "learning_rate": 3.18194483608103e-08, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29410 + }, + { + "epoch": 8.559790514983998, + "flow_matching_loss": 0.012224457133561373, + "grad_norm": 0.06365924328565598, + "learning_rate": 3.075212303498942e-08, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29420 + }, + { + "epoch": 8.562700029095142, + "flow_matching_loss": 0.012991876155138016, + "grad_norm": 0.06471985578536987, + "learning_rate": 2.9702986756846574e-08, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29430 + }, + { + "epoch": 8.565609543206284, + "flow_matching_loss": 0.010391318844631315, + "grad_norm": 0.07286480069160461, + "learning_rate": 2.867204080118202e-08, + "loss": 0.0104, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29440 + }, + { + "epoch": 8.568519057317427, + "flow_matching_loss": 0.013837077841162681, + "grad_norm": 0.09886623173952103, + "learning_rate": 2.7659286420693708e-08, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29450 + }, + { + "epoch": 8.571428571428571, + "flow_matching_loss": 0.012983535882085562, + "grad_norm": 0.07786650210618973, + "learning_rate": 2.6664724845968934e-08, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29460 + }, + { + "epoch": 8.574338085539715, + "flow_matching_loss": 0.013203028403222562, + "grad_norm": 0.0669604241847992, + "learning_rate": 2.5688357285499343e-08, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29470 + }, + { + "epoch": 8.577247599650859, + "flow_matching_loss": 0.01383969197049737, + "grad_norm": 0.07313407957553864, + "learning_rate": 2.473018492565926e-08, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29480 + }, + { + "epoch": 8.580157113762002, + "flow_matching_loss": 0.011396007612347603, + "grad_norm": 0.07329302281141281, + "learning_rate": 2.3790208930720702e-08, + "loss": 0.0114, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29490 + }, + { + "epoch": 8.583066627873146, + "flow_matching_loss": 0.012226414494216442, + "grad_norm": 0.07752939313650131, + "learning_rate": 2.2868430442843368e-08, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29500 + }, + { + "epoch": 8.585976141984288, + "flow_matching_loss": 0.011755072697997093, + "grad_norm": 0.06800425797700882, + "learning_rate": 2.196485058207298e-08, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29510 + }, + { + "epoch": 8.588885656095432, + "flow_matching_loss": 0.012176356464624404, + "grad_norm": 0.07006639242172241, + "learning_rate": 2.1079470446346282e-08, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29520 + }, + { + "epoch": 8.591795170206575, + "flow_matching_loss": 0.013872453290969134, + "grad_norm": 0.058815620839595795, + "learning_rate": 2.0212291111482707e-08, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29530 + }, + { + "epoch": 8.594704684317719, + "flow_matching_loss": 0.012820127420127392, + "grad_norm": 0.0732850432395935, + "learning_rate": 1.9363313631187707e-08, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29540 + }, + { + "epoch": 8.597614198428863, + "flow_matching_loss": 0.01455988809466362, + "grad_norm": 0.07404571771621704, + "learning_rate": 1.853253903704777e-08, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29550 + }, + { + "epoch": 8.600523712540006, + "flow_matching_loss": 0.011584219802170993, + "grad_norm": 0.06814903765916824, + "learning_rate": 1.7719968338533733e-08, + "loss": 0.0116, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29560 + }, + { + "epoch": 8.603433226651148, + "flow_matching_loss": 0.012126040551811457, + "grad_norm": 0.07298994064331055, + "learning_rate": 1.6925602522992467e-08, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29570 + }, + { + "epoch": 8.606342740762292, + "flow_matching_loss": 0.01219925144687295, + "grad_norm": 0.06856200844049454, + "learning_rate": 1.6149442555658533e-08, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29580 + }, + { + "epoch": 8.609252254873436, + "flow_matching_loss": 0.013906961306929588, + "grad_norm": 0.07330061495304108, + "learning_rate": 1.5391489379635858e-08, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29590 + }, + { + "epoch": 8.61216176898458, + "flow_matching_loss": 0.012819942738860845, + "grad_norm": 0.05756188929080963, + "learning_rate": 1.4651743915909399e-08, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29600 + }, + { + "epoch": 8.615071283095723, + "flow_matching_loss": 0.015142134949564933, + "grad_norm": 0.07487346231937408, + "learning_rate": 1.3930207063340139e-08, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29610 + }, + { + "epoch": 8.617980797206867, + "flow_matching_loss": 0.012325819721445442, + "grad_norm": 0.05687498673796654, + "learning_rate": 1.322687969866343e-08, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29620 + }, + { + "epoch": 8.62089031131801, + "flow_matching_loss": 0.011502622067928315, + "grad_norm": 0.07665029913187027, + "learning_rate": 1.2541762676488989e-08, + "loss": 0.0115, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29630 + }, + { + "epoch": 8.623799825429153, + "flow_matching_loss": 0.011482204403728246, + "grad_norm": 0.07028832286596298, + "learning_rate": 1.1874856829297564e-08, + "loss": 0.0115, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29640 + }, + { + "epoch": 8.626709339540296, + "flow_matching_loss": 0.013585858047008514, + "grad_norm": 0.0728665292263031, + "learning_rate": 1.1226162967442611e-08, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29650 + }, + { + "epoch": 8.62961885365144, + "flow_matching_loss": 0.011659033037722111, + "grad_norm": 0.08169899880886078, + "learning_rate": 1.059568187914861e-08, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29660 + }, + { + "epoch": 8.632528367762584, + "flow_matching_loss": 0.012372460588812828, + "grad_norm": 0.06996914744377136, + "learning_rate": 9.983414330511087e-09, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29670 + }, + { + "epoch": 8.635437881873727, + "flow_matching_loss": 0.013657233864068984, + "grad_norm": 0.08532674610614777, + "learning_rate": 9.389361065489932e-09, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29680 + }, + { + "epoch": 8.638347395984871, + "flow_matching_loss": 0.012985961511731148, + "grad_norm": 0.08552336692810059, + "learning_rate": 8.81352280591774e-09, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29690 + }, + { + "epoch": 8.641256910096015, + "flow_matching_loss": 0.011058480618521572, + "grad_norm": 0.06328040361404419, + "learning_rate": 8.255900251491478e-09, + "loss": 0.0111, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29700 + }, + { + "epoch": 8.644166424207157, + "flow_matching_loss": 0.014783472754061222, + "grad_norm": 0.0809445008635521, + "learning_rate": 7.716494079775816e-09, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29710 + }, + { + "epoch": 8.6470759383183, + "flow_matching_loss": 0.012453899532556535, + "grad_norm": 0.06661822646856308, + "learning_rate": 7.195304946199799e-09, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29720 + }, + { + "epoch": 8.649985452429444, + "flow_matching_loss": 0.012731821648776532, + "grad_norm": 0.06765200942754745, + "learning_rate": 6.692333484056845e-09, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29730 + }, + { + "epoch": 8.652894966540588, + "flow_matching_loss": 0.013671270478516817, + "grad_norm": 0.08420855551958084, + "learning_rate": 6.207580304506411e-09, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29740 + }, + { + "epoch": 8.655804480651732, + "flow_matching_loss": 0.013048023451119661, + "grad_norm": 0.09778223931789398, + "learning_rate": 5.741045996567329e-09, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29750 + }, + { + "epoch": 8.658713994762875, + "flow_matching_loss": 0.012617102544754743, + "grad_norm": 0.08303812146186829, + "learning_rate": 5.2927311271244725e-09, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29760 + }, + { + "epoch": 8.661623508874017, + "flow_matching_loss": 0.014149058610200882, + "grad_norm": 0.0789141058921814, + "learning_rate": 4.8626362409204265e-09, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29770 + }, + { + "epoch": 8.664533022985161, + "flow_matching_loss": 0.013815926481038332, + "grad_norm": 0.08443633466959, + "learning_rate": 4.4507618605638126e-09, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29780 + }, + { + "epoch": 8.667442537096305, + "flow_matching_loss": 0.014270150940865279, + "grad_norm": 0.08866216987371445, + "learning_rate": 4.057108486519301e-09, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29790 + }, + { + "epoch": 8.670352051207448, + "flow_matching_loss": 0.01315668374300003, + "grad_norm": 0.07516106963157654, + "learning_rate": 3.6816765971126043e-09, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29800 + }, + { + "epoch": 8.673261565318592, + "flow_matching_loss": 0.012624203227460384, + "grad_norm": 0.0706789493560791, + "learning_rate": 3.324466648528812e-09, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29810 + }, + { + "epoch": 8.676171079429736, + "flow_matching_loss": 0.013683277182281017, + "grad_norm": 0.06629828363656998, + "learning_rate": 2.9854790748123917e-09, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29820 + }, + { + "epoch": 8.67908059354088, + "flow_matching_loss": 0.012416338268667459, + "grad_norm": 0.06581724435091019, + "learning_rate": 2.664714287865522e-09, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29830 + }, + { + "epoch": 8.681990107652021, + "flow_matching_loss": 0.012554429471492767, + "grad_norm": 0.05887773260474205, + "learning_rate": 2.362172677446428e-09, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29840 + }, + { + "epoch": 8.684899621763165, + "flow_matching_loss": 0.011771690659224987, + "grad_norm": 0.0729188546538353, + "learning_rate": 2.0778546111727137e-09, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29850 + }, + { + "epoch": 8.687809135874309, + "flow_matching_loss": 0.012417888920754194, + "grad_norm": 0.08576650172472, + "learning_rate": 1.811760434516363e-09, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29860 + }, + { + "epoch": 8.690718649985453, + "flow_matching_loss": 0.012143871001899242, + "grad_norm": 0.06380762159824371, + "learning_rate": 1.5638904708087377e-09, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29870 + }, + { + "epoch": 8.693628164096596, + "flow_matching_loss": 0.01224596444517374, + "grad_norm": 0.054130036383867264, + "learning_rate": 1.3342450212355805e-09, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29880 + }, + { + "epoch": 8.69653767820774, + "flow_matching_loss": 0.013569681253284217, + "grad_norm": 0.09417291730642319, + "learning_rate": 1.122824364835351e-09, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29890 + }, + { + "epoch": 8.699447192318882, + "flow_matching_loss": 0.012200099136680364, + "grad_norm": 0.06119155511260033, + "learning_rate": 9.296287585042196e-10, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29900 + }, + { + "epoch": 8.702356706430026, + "flow_matching_loss": 0.012029681727290154, + "grad_norm": 0.061947040259838104, + "learning_rate": 7.546584369960697e-10, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29910 + }, + { + "epoch": 8.70526622054117, + "flow_matching_loss": 0.013541720062494277, + "grad_norm": 0.07544105499982834, + "learning_rate": 5.97913612912504e-10, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29920 + }, + { + "epoch": 8.708175734652313, + "flow_matching_loss": 0.011795561853796243, + "grad_norm": 0.05005600303411484, + "learning_rate": 4.593944767161684e-10, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29930 + }, + { + "epoch": 8.711085248763457, + "flow_matching_loss": 0.013798499293625355, + "grad_norm": 0.08500850200653076, + "learning_rate": 3.391011967207591e-10, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29940 + }, + { + "epoch": 8.7139947628746, + "flow_matching_loss": 0.011824683379381895, + "grad_norm": 0.07519569247961044, + "learning_rate": 2.370339190910231e-10, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29950 + }, + { + "epoch": 8.716904276985744, + "flow_matching_loss": 0.012577164452522993, + "grad_norm": 0.062107235193252563, + "learning_rate": 1.531927678510847e-10, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29960 + }, + { + "epoch": 8.719813791096886, + "flow_matching_loss": 0.013511533383280039, + "grad_norm": 0.055698636919260025, + "learning_rate": 8.757784487611886e-11, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29970 + }, + { + "epoch": 8.72272330520803, + "flow_matching_loss": 0.013075003866106272, + "grad_norm": 0.0704827532172203, + "learning_rate": 4.018922989235119e-11, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29980 + }, + { + "epoch": 8.725632819319173, + "flow_matching_loss": 0.012844633590430022, + "grad_norm": 0.06343361735343933, + "learning_rate": 1.1026980482053972e-11, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 29990 + }, + { + "epoch": 8.728542333430317, + "flow_matching_loss": 0.012701811827719211, + "grad_norm": 0.0609227679669857, + "learning_rate": 9.113208188082922e-14, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.5, + "step": 30000 + }, + { + "epoch": 8.728542333430317, + "step": 30000, + "total_flos": 0.0, + "train_loss": 0.015626614911605913, + "train_runtime": 64240.4931, + "train_samples_per_second": 59.775, + "train_steps_per_second": 0.467 + } + ], + "logging_steps": 10, + "max_steps": 30000, + "num_input_tokens_seen": 0, + "num_train_epochs": 9, + "save_steps": 1, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": false, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 0.0, + "train_batch_size": 128, + "trial_name": null, + "trial_params": null +} diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/phase3/training_args.bin b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/phase3/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..eaf436152b86e62c0479be7ee1f248f79f1f57e9 --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/phase3/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1c2d0c2277b66fa3ea387a1fc73762d9e00f75ee7528c55ee6211089361ee2a2 +size 5969 diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/resolved_config.yaml b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/resolved_config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ea0e766771bef9d10f7300273d3af6b3164bdba4 --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/resolved_config.yaml @@ -0,0 +1,32 @@ +experiment_name: mgd_sweep_from_trained_base +policy_type: groot_mgd +base_model_path: ${HOME}/groot_robocasa/robocasa_v2/MGD_Processing_line_Only/mgd_token_mask_ratio_0.5/phase2 +output_root: ${HOME}/groot_robocasa/robocasa_v2/MGD_Processing_line_Only/dit_core_only +sweep: + mgd_token_mask_ratio: + - 0.5 +dataset: + dataset_soup: my_atomic26_human +training: + num_gpus: 1 + batch_size: 128 +model: null +phases: +- name: phase3_mgd_fm + max_steps: 30000 + save_steps: 0 + trainable: + preset: freeze_processing_line + tune_llm: false + tune_visual: false + tune_projector: true + tune_diffusion_model: true + losses: + mgd_enabled: false + mgd_fm_loss_weight: 1.0 + mgd_loss_weight: 0.0 + mgd_token_mask_ratio: 0.5 + mgd_use_cosine_loss: true + mgd_use_mse_loss: false +resolved_sweep: + mgd_token_mask_ratio: 0.5 diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/phase3/config.json b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/phase3/config.json new file mode 100644 index 0000000000000000000000000000000000000000..5659f250a7642da16d7e0861e44244c466ad4268 --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/phase3/config.json @@ -0,0 +1,75 @@ +{ + "action_dim": 32, + "action_head_cfg": { + "action_dim": 32, + "action_horizon": 16, + "add_pos_embed": true, + "backbone_embedding_dim": 2048, + "diffusion_model_cfg": { + "attention_head_dim": 48, + "cross_attention_dim": 2048, + "dropout": 0.2, + "final_dropout": true, + "interleave_self_attention": true, + "norm_type": "ada_norm", + "num_attention_heads": 32, + "num_layers": 16, + "output_dim": 1024, + "positional_embeddings": null + }, + "hidden_size": 1024, + "input_embedding_dim": 1536, + "max_action_dim": 32, + "max_state_dim": 64, + "model_dtype": "float32", + "noise_beta_alpha": 1.5, + "noise_beta_beta": 1.0, + "noise_s": 0.999, + "num_inference_timesteps": 4, + "num_target_vision_tokens": 32, + "num_timestep_buckets": 1000, + "tune_diffusion_model": true, + "tune_projector": true, + "use_vlln": true, + "vl_self_attention_cfg": { + "attention_head_dim": 64, + "dropout": 0.2, + "final_dropout": true, + "num_attention_heads": 32, + "num_layers": 4, + "positional_embeddings": null + } + }, + "action_horizon": 16, + "architectures": [ + "GR00T_N1_5_MGD" + ], + "attn_implementation": null, + "backbone_cfg": { + "eagle_path": "NVEagle/eagle_er-qwen3_1_7B-Siglip2_400M_stage1_5_128gpu_er_v7_1mlp_nops", + "load_bf16": false, + "project_to_dim": null, + "reproject_vision": false, + "select_layer": 12, + "tune_llm": false, + "tune_visual": true, + "use_flash_attention": true + }, + "compute_dtype": "bfloat16", + "hidden_size": 2048, + "mgd_enabled": false, + "mgd_fm_loss_weight": 1.0, + "mgd_loss_weight": 0.0, + "mgd_pretrained_projector_path": null, + "mgd_sequence_hidden_dim": 512, + "mgd_target_dim": 512, + "mgd_target_pooling": "flatten", + "mgd_target_projection": "frozen_random", + "mgd_token_mask_ratio": 0.7, + "mgd_use_cosine_loss": true, + "mgd_use_mse_loss": false, + "model_dtype": "float32", + "model_type": "gr00t_n1_5", + "torch_dtype": "bfloat16", + "transformers_version": "4.51.3" +} diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/phase3/model-00001-of-00002.safetensors b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/phase3/model-00001-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ff17ff178a2ab396e9626b7bcfd40912adeba312 --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/phase3/model-00001-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c3d91174343f68b987f67e6e2226b1adc91e5775eaa949595c700859a9d317f1 +size 4999367032 diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/phase3/model-00002-of-00002.safetensors b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/phase3/model-00002-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..112ff1370d8e4ba34a2061d4428b946376acea56 --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/phase3/model-00002-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:573640a06163e106ea9e170ef6d1c90629e58dc53fd512f72a5cd1542d4450c7 +size 2643339748 diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/phase3/trainer_state.json b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/phase3/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..b86d29984527843e488b7323d8933e7d1ea1210b --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/phase3/trainer_state.json @@ -0,0 +1,30043 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 8.728542333430317, + "eval_steps": 500, + "global_step": 30000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.002909514111143439, + "flow_matching_loss": 0.13517873287200927, + "grad_norm": 0.7548623085021973, + "learning_rate": 1.8e-07, + "loss": 0.1352, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10 + }, + { + "epoch": 0.005819028222286878, + "flow_matching_loss": 0.1243258222937584, + "grad_norm": 0.7039984464645386, + "learning_rate": 3.8e-07, + "loss": 0.1243, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20 + }, + { + "epoch": 0.008728542333430317, + "flow_matching_loss": 0.12047879993915558, + "grad_norm": 0.6892663836479187, + "learning_rate": 5.800000000000001e-07, + "loss": 0.1205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 30 + }, + { + "epoch": 0.011638056444573755, + "flow_matching_loss": 0.10773561894893646, + "grad_norm": 0.48759496212005615, + "learning_rate": 7.799999999999999e-07, + "loss": 0.1077, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 40 + }, + { + "epoch": 0.014547570555717195, + "flow_matching_loss": 0.09475416168570519, + "grad_norm": 0.42872756719589233, + "learning_rate": 9.8e-07, + "loss": 0.0948, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 50 + }, + { + "epoch": 0.017457084666860634, + "flow_matching_loss": 0.08632880225777625, + "grad_norm": 0.2774789035320282, + "learning_rate": 1.18e-06, + "loss": 0.0863, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 60 + }, + { + "epoch": 0.020366598778004074, + "flow_matching_loss": 0.07635721154510974, + "grad_norm": 0.3063410222530365, + "learning_rate": 1.38e-06, + "loss": 0.0764, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 70 + }, + { + "epoch": 0.02327611288914751, + "flow_matching_loss": 0.07213817872107028, + "grad_norm": 0.21857188642024994, + "learning_rate": 1.5800000000000001e-06, + "loss": 0.0721, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 80 + }, + { + "epoch": 0.02618562700029095, + "flow_matching_loss": 0.060456155240535735, + "grad_norm": 0.2389698624610901, + "learning_rate": 1.7800000000000001e-06, + "loss": 0.0605, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 90 + }, + { + "epoch": 0.02909514111143439, + "flow_matching_loss": 0.05783389657735825, + "grad_norm": 0.20880815386772156, + "learning_rate": 1.98e-06, + "loss": 0.0578, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 100 + }, + { + "epoch": 0.03200465522257783, + "flow_matching_loss": 0.0546309620141983, + "grad_norm": 0.20113055408000946, + "learning_rate": 2.1800000000000003e-06, + "loss": 0.0546, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 110 + }, + { + "epoch": 0.03491416933372127, + "flow_matching_loss": 0.054952782765030864, + "grad_norm": 0.18514256179332733, + "learning_rate": 2.38e-06, + "loss": 0.055, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 120 + }, + { + "epoch": 0.037823683444864704, + "flow_matching_loss": 0.04825925640761852, + "grad_norm": 0.15446971356868744, + "learning_rate": 2.58e-06, + "loss": 0.0483, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 130 + }, + { + "epoch": 0.04073319755600815, + "flow_matching_loss": 0.04477493278682232, + "grad_norm": 0.13283032178878784, + "learning_rate": 2.78e-06, + "loss": 0.0448, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 140 + }, + { + "epoch": 0.043642711667151585, + "flow_matching_loss": 0.04854634888470173, + "grad_norm": 0.1824956089258194, + "learning_rate": 2.98e-06, + "loss": 0.0485, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 150 + }, + { + "epoch": 0.04655222577829502, + "flow_matching_loss": 0.04061545822769404, + "grad_norm": 0.14136192202568054, + "learning_rate": 3.18e-06, + "loss": 0.0406, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 160 + }, + { + "epoch": 0.049461739889438465, + "flow_matching_loss": 0.045986872911453244, + "grad_norm": 0.15585896372795105, + "learning_rate": 3.38e-06, + "loss": 0.046, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 170 + }, + { + "epoch": 0.0523712540005819, + "flow_matching_loss": 0.0422821793705225, + "grad_norm": 0.14175210893154144, + "learning_rate": 3.58e-06, + "loss": 0.0423, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 180 + }, + { + "epoch": 0.055280768111725345, + "flow_matching_loss": 0.037602398172020914, + "grad_norm": 0.1564805507659912, + "learning_rate": 3.7800000000000002e-06, + "loss": 0.0376, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 190 + }, + { + "epoch": 0.05819028222286878, + "flow_matching_loss": 0.03706999123096466, + "grad_norm": 0.13885128498077393, + "learning_rate": 3.98e-06, + "loss": 0.0371, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 200 + }, + { + "epoch": 0.06109979633401222, + "flow_matching_loss": 0.03733954504132271, + "grad_norm": 0.12437935173511505, + "learning_rate": 4.18e-06, + "loss": 0.0373, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 210 + }, + { + "epoch": 0.06400931044515566, + "flow_matching_loss": 0.03875552918761969, + "grad_norm": 0.13181272149085999, + "learning_rate": 4.3799999999999996e-06, + "loss": 0.0388, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 220 + }, + { + "epoch": 0.06691882455629909, + "flow_matching_loss": 0.03419189006090164, + "grad_norm": 0.14824031293392181, + "learning_rate": 4.58e-06, + "loss": 0.0342, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 230 + }, + { + "epoch": 0.06982833866744254, + "flow_matching_loss": 0.03333702106028795, + "grad_norm": 0.09359066933393478, + "learning_rate": 4.78e-06, + "loss": 0.0333, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 240 + }, + { + "epoch": 0.07273785277858598, + "flow_matching_loss": 0.033826086483895776, + "grad_norm": 0.14490029215812683, + "learning_rate": 4.980000000000001e-06, + "loss": 0.0338, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 250 + }, + { + "epoch": 0.07564736688972941, + "flow_matching_loss": 0.032628918811678884, + "grad_norm": 0.14850202202796936, + "learning_rate": 5.18e-06, + "loss": 0.0326, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 260 + }, + { + "epoch": 0.07855688100087285, + "flow_matching_loss": 0.03430499918758869, + "grad_norm": 0.15610642731189728, + "learning_rate": 5.38e-06, + "loss": 0.0343, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 270 + }, + { + "epoch": 0.0814663951120163, + "flow_matching_loss": 0.028796425461769103, + "grad_norm": 0.09198634326457977, + "learning_rate": 5.58e-06, + "loss": 0.0288, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 280 + }, + { + "epoch": 0.08437590922315973, + "flow_matching_loss": 0.03169934377074242, + "grad_norm": 0.12405478954315186, + "learning_rate": 5.7800000000000006e-06, + "loss": 0.0317, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 290 + }, + { + "epoch": 0.08728542333430317, + "flow_matching_loss": 0.03071829564869404, + "grad_norm": 0.10975580662488937, + "learning_rate": 5.98e-06, + "loss": 0.0307, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 300 + }, + { + "epoch": 0.09019493744544661, + "flow_matching_loss": 0.03079163786023855, + "grad_norm": 0.11115937680006027, + "learning_rate": 6.18e-06, + "loss": 0.0308, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 310 + }, + { + "epoch": 0.09310445155659004, + "flow_matching_loss": 0.0309028634801507, + "grad_norm": 0.1465008705854416, + "learning_rate": 6.38e-06, + "loss": 0.0309, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 320 + }, + { + "epoch": 0.09601396566773349, + "flow_matching_loss": 0.03093602731823921, + "grad_norm": 0.10163235664367676, + "learning_rate": 6.58e-06, + "loss": 0.0309, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 330 + }, + { + "epoch": 0.09892347977887693, + "flow_matching_loss": 0.027764305099844934, + "grad_norm": 0.12403926253318787, + "learning_rate": 6.78e-06, + "loss": 0.0278, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 340 + }, + { + "epoch": 0.10183299389002037, + "flow_matching_loss": 0.0267748462036252, + "grad_norm": 0.11982909590005875, + "learning_rate": 6.98e-06, + "loss": 0.0268, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 350 + }, + { + "epoch": 0.1047425080011638, + "flow_matching_loss": 0.02850965466350317, + "grad_norm": 0.09971286356449127, + "learning_rate": 7.180000000000001e-06, + "loss": 0.0285, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 360 + }, + { + "epoch": 0.10765202211230725, + "flow_matching_loss": 0.029419366642832755, + "grad_norm": 0.11338437348604202, + "learning_rate": 7.3800000000000005e-06, + "loss": 0.0294, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 370 + }, + { + "epoch": 0.11056153622345069, + "flow_matching_loss": 0.029723187163472175, + "grad_norm": 0.11586830019950867, + "learning_rate": 7.5799999999999994e-06, + "loss": 0.0297, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 380 + }, + { + "epoch": 0.11347105033459412, + "flow_matching_loss": 0.030193422362208365, + "grad_norm": 0.0963481068611145, + "learning_rate": 7.780000000000002e-06, + "loss": 0.0302, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 390 + }, + { + "epoch": 0.11638056444573756, + "flow_matching_loss": 0.027163317985832693, + "grad_norm": 0.13135404884815216, + "learning_rate": 7.98e-06, + "loss": 0.0272, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 400 + }, + { + "epoch": 0.11929007855688101, + "flow_matching_loss": 0.025531047396361826, + "grad_norm": 0.10527969896793365, + "learning_rate": 8.18e-06, + "loss": 0.0255, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 410 + }, + { + "epoch": 0.12219959266802444, + "flow_matching_loss": 0.02641311176121235, + "grad_norm": 0.10075055807828903, + "learning_rate": 8.38e-06, + "loss": 0.0264, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 420 + }, + { + "epoch": 0.12510910677916787, + "flow_matching_loss": 0.02864238191395998, + "grad_norm": 0.13571399450302124, + "learning_rate": 8.58e-06, + "loss": 0.0286, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 430 + }, + { + "epoch": 0.12801862089031132, + "flow_matching_loss": 0.028548496961593627, + "grad_norm": 0.1290515512228012, + "learning_rate": 8.78e-06, + "loss": 0.0285, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 440 + }, + { + "epoch": 0.13092813500145475, + "flow_matching_loss": 0.02750872541218996, + "grad_norm": 0.11145783960819244, + "learning_rate": 8.98e-06, + "loss": 0.0275, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 450 + }, + { + "epoch": 0.13383764911259818, + "flow_matching_loss": 0.0259602889418602, + "grad_norm": 0.11778199672698975, + "learning_rate": 9.18e-06, + "loss": 0.026, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 460 + }, + { + "epoch": 0.13674716322374164, + "flow_matching_loss": 0.02702238503843546, + "grad_norm": 0.10918254405260086, + "learning_rate": 9.38e-06, + "loss": 0.027, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 470 + }, + { + "epoch": 0.13965667733488507, + "flow_matching_loss": 0.02621891349554062, + "grad_norm": 0.12028585374355316, + "learning_rate": 9.580000000000001e-06, + "loss": 0.0262, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 480 + }, + { + "epoch": 0.1425661914460285, + "flow_matching_loss": 0.026134964451193808, + "grad_norm": 0.10952601581811905, + "learning_rate": 9.780000000000001e-06, + "loss": 0.0261, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 490 + }, + { + "epoch": 0.14547570555717196, + "flow_matching_loss": 0.02696720100939274, + "grad_norm": 0.11399393528699875, + "learning_rate": 9.980000000000001e-06, + "loss": 0.027, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 500 + }, + { + "epoch": 0.1483852196683154, + "flow_matching_loss": 0.027005795575678348, + "grad_norm": 0.12453304976224899, + "learning_rate": 1.0179999999999999e-05, + "loss": 0.027, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 510 + }, + { + "epoch": 0.15129473377945882, + "flow_matching_loss": 0.023036616295576094, + "grad_norm": 0.10448396950960159, + "learning_rate": 1.0379999999999999e-05, + "loss": 0.023, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 520 + }, + { + "epoch": 0.15420424789060228, + "flow_matching_loss": 0.022436598688364028, + "grad_norm": 0.09651082009077072, + "learning_rate": 1.058e-05, + "loss": 0.0224, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 530 + }, + { + "epoch": 0.1571137620017457, + "flow_matching_loss": 0.026296931318938732, + "grad_norm": 0.11791712045669556, + "learning_rate": 1.078e-05, + "loss": 0.0263, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 540 + }, + { + "epoch": 0.16002327611288913, + "flow_matching_loss": 0.023659420758485795, + "grad_norm": 0.07828687876462936, + "learning_rate": 1.098e-05, + "loss": 0.0237, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 550 + }, + { + "epoch": 0.1629327902240326, + "flow_matching_loss": 0.022823915258049964, + "grad_norm": 0.07259003072977066, + "learning_rate": 1.118e-05, + "loss": 0.0228, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 560 + }, + { + "epoch": 0.16584230433517602, + "flow_matching_loss": 0.02632592711597681, + "grad_norm": 0.11944786459207535, + "learning_rate": 1.1380000000000001e-05, + "loss": 0.0263, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 570 + }, + { + "epoch": 0.16875181844631945, + "flow_matching_loss": 0.023793658055365086, + "grad_norm": 0.09833203256130219, + "learning_rate": 1.1580000000000001e-05, + "loss": 0.0238, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 580 + }, + { + "epoch": 0.1716613325574629, + "flow_matching_loss": 0.026928935945034028, + "grad_norm": 0.09137031435966492, + "learning_rate": 1.178e-05, + "loss": 0.0269, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 590 + }, + { + "epoch": 0.17457084666860634, + "flow_matching_loss": 0.02435552254319191, + "grad_norm": 0.11217307299375534, + "learning_rate": 1.198e-05, + "loss": 0.0244, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 600 + }, + { + "epoch": 0.17748036077974977, + "flow_matching_loss": 0.02295201290398836, + "grad_norm": 0.10666759312152863, + "learning_rate": 1.2180000000000002e-05, + "loss": 0.023, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 610 + }, + { + "epoch": 0.18038987489089323, + "flow_matching_loss": 0.024176337011158466, + "grad_norm": 0.11287493258714676, + "learning_rate": 1.238e-05, + "loss": 0.0242, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 620 + }, + { + "epoch": 0.18329938900203666, + "flow_matching_loss": 0.02309722136706114, + "grad_norm": 0.08762650936841965, + "learning_rate": 1.258e-05, + "loss": 0.0231, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 630 + }, + { + "epoch": 0.18620890311318009, + "flow_matching_loss": 0.023657916486263274, + "grad_norm": 0.11965446174144745, + "learning_rate": 1.278e-05, + "loss": 0.0237, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 640 + }, + { + "epoch": 0.18911841722432354, + "flow_matching_loss": 0.02374318167567253, + "grad_norm": 0.11451905220746994, + "learning_rate": 1.298e-05, + "loss": 0.0237, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 650 + }, + { + "epoch": 0.19202793133546697, + "flow_matching_loss": 0.020598486065864563, + "grad_norm": 0.10034392029047012, + "learning_rate": 1.3180000000000001e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 660 + }, + { + "epoch": 0.1949374454466104, + "flow_matching_loss": 0.026801137439906596, + "grad_norm": 0.10502073913812637, + "learning_rate": 1.338e-05, + "loss": 0.0268, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 670 + }, + { + "epoch": 0.19784695955775386, + "flow_matching_loss": 0.02331406343728304, + "grad_norm": 0.12322814762592316, + "learning_rate": 1.358e-05, + "loss": 0.0233, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 680 + }, + { + "epoch": 0.2007564736688973, + "flow_matching_loss": 0.022497034445405007, + "grad_norm": 0.1176372617483139, + "learning_rate": 1.378e-05, + "loss": 0.0225, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 690 + }, + { + "epoch": 0.20366598778004075, + "flow_matching_loss": 0.0213545604608953, + "grad_norm": 0.11412621289491653, + "learning_rate": 1.3980000000000002e-05, + "loss": 0.0214, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 700 + }, + { + "epoch": 0.20657550189118418, + "flow_matching_loss": 0.023258752934634686, + "grad_norm": 0.10565335303544998, + "learning_rate": 1.4180000000000001e-05, + "loss": 0.0233, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 710 + }, + { + "epoch": 0.2094850160023276, + "flow_matching_loss": 0.02100435020402074, + "grad_norm": 0.1109994649887085, + "learning_rate": 1.438e-05, + "loss": 0.021, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 720 + }, + { + "epoch": 0.21239453011347106, + "flow_matching_loss": 0.0238120062276721, + "grad_norm": 0.10497681051492691, + "learning_rate": 1.458e-05, + "loss": 0.0238, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 730 + }, + { + "epoch": 0.2153040442246145, + "flow_matching_loss": 0.022070358507335186, + "grad_norm": 0.10442095249891281, + "learning_rate": 1.4779999999999999e-05, + "loss": 0.0221, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 740 + }, + { + "epoch": 0.21821355833575792, + "flow_matching_loss": 0.022088124603033065, + "grad_norm": 0.11426962167024612, + "learning_rate": 1.498e-05, + "loss": 0.0221, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 750 + }, + { + "epoch": 0.22112307244690138, + "flow_matching_loss": 0.02255166620016098, + "grad_norm": 0.09108679741621017, + "learning_rate": 1.518e-05, + "loss": 0.0226, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 760 + }, + { + "epoch": 0.2240325865580448, + "flow_matching_loss": 0.02313474863767624, + "grad_norm": 0.12137779593467712, + "learning_rate": 1.5380000000000002e-05, + "loss": 0.0231, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 770 + }, + { + "epoch": 0.22694210066918824, + "flow_matching_loss": 0.022533394768834113, + "grad_norm": 0.09352660179138184, + "learning_rate": 1.558e-05, + "loss": 0.0225, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 780 + }, + { + "epoch": 0.2298516147803317, + "flow_matching_loss": 0.02317782249301672, + "grad_norm": 0.11825411766767502, + "learning_rate": 1.578e-05, + "loss": 0.0232, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 790 + }, + { + "epoch": 0.23276112889147513, + "flow_matching_loss": 0.022952629998326302, + "grad_norm": 0.12325946986675262, + "learning_rate": 1.598e-05, + "loss": 0.023, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 800 + }, + { + "epoch": 0.23567064300261856, + "flow_matching_loss": 0.020808304473757745, + "grad_norm": 0.10157302021980286, + "learning_rate": 1.618e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 810 + }, + { + "epoch": 0.23858015711376201, + "flow_matching_loss": 0.02243024669587612, + "grad_norm": 0.11749786883592606, + "learning_rate": 1.6380000000000002e-05, + "loss": 0.0224, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 820 + }, + { + "epoch": 0.24148967122490544, + "flow_matching_loss": 0.020171974878758193, + "grad_norm": 0.09594277292490005, + "learning_rate": 1.658e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 830 + }, + { + "epoch": 0.24439918533604887, + "flow_matching_loss": 0.022571573592722417, + "grad_norm": 0.10795339196920395, + "learning_rate": 1.6780000000000002e-05, + "loss": 0.0226, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 840 + }, + { + "epoch": 0.24730869944719233, + "flow_matching_loss": 0.025059936475008725, + "grad_norm": 0.13360194861888885, + "learning_rate": 1.698e-05, + "loss": 0.0251, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 850 + }, + { + "epoch": 0.25021821355833573, + "flow_matching_loss": 0.022084706090390684, + "grad_norm": 0.15949425101280212, + "learning_rate": 1.718e-05, + "loss": 0.0221, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 860 + }, + { + "epoch": 0.2531277276694792, + "flow_matching_loss": 0.02109331712126732, + "grad_norm": 0.11262931674718857, + "learning_rate": 1.7380000000000003e-05, + "loss": 0.0211, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 870 + }, + { + "epoch": 0.25603724178062265, + "flow_matching_loss": 0.02234501447528601, + "grad_norm": 0.09190438687801361, + "learning_rate": 1.758e-05, + "loss": 0.0223, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 880 + }, + { + "epoch": 0.2589467558917661, + "flow_matching_loss": 0.022109046950936316, + "grad_norm": 0.08959496766328812, + "learning_rate": 1.778e-05, + "loss": 0.0221, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 890 + }, + { + "epoch": 0.2618562700029095, + "flow_matching_loss": 0.02385210059583187, + "grad_norm": 0.09901391714811325, + "learning_rate": 1.798e-05, + "loss": 0.0239, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 900 + }, + { + "epoch": 0.26476578411405294, + "flow_matching_loss": 0.021110167913138866, + "grad_norm": 0.09209585189819336, + "learning_rate": 1.818e-05, + "loss": 0.0211, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 910 + }, + { + "epoch": 0.26767529822519637, + "flow_matching_loss": 0.022278976812958717, + "grad_norm": 0.12148077040910721, + "learning_rate": 1.838e-05, + "loss": 0.0223, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 920 + }, + { + "epoch": 0.27058481233633985, + "flow_matching_loss": 0.02242675591260195, + "grad_norm": 0.13164061307907104, + "learning_rate": 1.8579999999999998e-05, + "loss": 0.0224, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 930 + }, + { + "epoch": 0.2734943264474833, + "flow_matching_loss": 0.02240576259791851, + "grad_norm": 0.1396752893924713, + "learning_rate": 1.878e-05, + "loss": 0.0224, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 940 + }, + { + "epoch": 0.2764038405586267, + "flow_matching_loss": 0.02241366673260927, + "grad_norm": 0.12301665544509888, + "learning_rate": 1.898e-05, + "loss": 0.0224, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 950 + }, + { + "epoch": 0.27931335466977014, + "flow_matching_loss": 0.022872879542410372, + "grad_norm": 0.09548220038414001, + "learning_rate": 1.918e-05, + "loss": 0.0229, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 960 + }, + { + "epoch": 0.28222286878091357, + "flow_matching_loss": 0.022236265242099762, + "grad_norm": 0.07936926186084747, + "learning_rate": 1.938e-05, + "loss": 0.0222, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 970 + }, + { + "epoch": 0.285132382892057, + "flow_matching_loss": 0.02162749171257019, + "grad_norm": 0.12890353798866272, + "learning_rate": 1.958e-05, + "loss": 0.0216, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 980 + }, + { + "epoch": 0.2880418970032005, + "flow_matching_loss": 0.0223810825496912, + "grad_norm": 0.11116036772727966, + "learning_rate": 1.978e-05, + "loss": 0.0224, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 990 + }, + { + "epoch": 0.2909514111143439, + "flow_matching_loss": 0.022724844701588153, + "grad_norm": 0.13673432171344757, + "learning_rate": 1.9980000000000002e-05, + "loss": 0.0227, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1000 + }, + { + "epoch": 0.29386092522548735, + "flow_matching_loss": 0.02298563942313194, + "grad_norm": 0.1253710240125656, + "learning_rate": 2.018e-05, + "loss": 0.023, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1010 + }, + { + "epoch": 0.2967704393366308, + "flow_matching_loss": 0.02225575242191553, + "grad_norm": 0.14123816788196564, + "learning_rate": 2.038e-05, + "loss": 0.0223, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1020 + }, + { + "epoch": 0.2996799534477742, + "flow_matching_loss": 0.02181571526452899, + "grad_norm": 0.13402943313121796, + "learning_rate": 2.0580000000000003e-05, + "loss": 0.0218, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1030 + }, + { + "epoch": 0.30258946755891764, + "flow_matching_loss": 0.021811266243457795, + "grad_norm": 0.1251261681318283, + "learning_rate": 2.078e-05, + "loss": 0.0218, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1040 + }, + { + "epoch": 0.3054989816700611, + "flow_matching_loss": 0.02301169503480196, + "grad_norm": 0.09183633327484131, + "learning_rate": 2.0980000000000002e-05, + "loss": 0.023, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1050 + }, + { + "epoch": 0.30840849578120455, + "flow_matching_loss": 0.020500523224473, + "grad_norm": 0.12676022946834564, + "learning_rate": 2.118e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1060 + }, + { + "epoch": 0.311318009892348, + "flow_matching_loss": 0.021449005976319314, + "grad_norm": 0.09962743520736694, + "learning_rate": 2.1380000000000002e-05, + "loss": 0.0214, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1070 + }, + { + "epoch": 0.3142275240034914, + "flow_matching_loss": 0.02071178946644068, + "grad_norm": 0.14141155779361725, + "learning_rate": 2.1580000000000003e-05, + "loss": 0.0207, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1080 + }, + { + "epoch": 0.31713703811463484, + "flow_matching_loss": 0.019396833144128323, + "grad_norm": 0.11961345374584198, + "learning_rate": 2.178e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1090 + }, + { + "epoch": 0.32004655222577827, + "flow_matching_loss": 0.022519232891499995, + "grad_norm": 0.16043873131275177, + "learning_rate": 2.198e-05, + "loss": 0.0225, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1100 + }, + { + "epoch": 0.32295606633692175, + "flow_matching_loss": 0.022523126378655434, + "grad_norm": 0.15021607279777527, + "learning_rate": 2.2179999999999998e-05, + "loss": 0.0225, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1110 + }, + { + "epoch": 0.3258655804480652, + "flow_matching_loss": 0.020078240521252157, + "grad_norm": 0.13381260633468628, + "learning_rate": 2.238e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1120 + }, + { + "epoch": 0.3287750945592086, + "flow_matching_loss": 0.0220586933195591, + "grad_norm": 0.11642517894506454, + "learning_rate": 2.258e-05, + "loss": 0.0221, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1130 + }, + { + "epoch": 0.33168460867035204, + "flow_matching_loss": 0.01957835229113698, + "grad_norm": 0.15039679408073425, + "learning_rate": 2.278e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1140 + }, + { + "epoch": 0.3345941227814955, + "flow_matching_loss": 0.020641309954226018, + "grad_norm": 0.12696333229541779, + "learning_rate": 2.298e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1150 + }, + { + "epoch": 0.3375036368926389, + "flow_matching_loss": 0.022329976223409174, + "grad_norm": 0.15839174389839172, + "learning_rate": 2.3179999999999998e-05, + "loss": 0.0223, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1160 + }, + { + "epoch": 0.3404131510037824, + "flow_matching_loss": 0.02161662457510829, + "grad_norm": 0.1106647253036499, + "learning_rate": 2.338e-05, + "loss": 0.0216, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1170 + }, + { + "epoch": 0.3433226651149258, + "flow_matching_loss": 0.023337959311902523, + "grad_norm": 0.08509796857833862, + "learning_rate": 2.358e-05, + "loss": 0.0233, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1180 + }, + { + "epoch": 0.34623217922606925, + "flow_matching_loss": 0.020380011945962905, + "grad_norm": 0.10807551443576813, + "learning_rate": 2.378e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1190 + }, + { + "epoch": 0.3491416933372127, + "flow_matching_loss": 0.02128849634900689, + "grad_norm": 0.18791018426418304, + "learning_rate": 2.398e-05, + "loss": 0.0213, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1200 + }, + { + "epoch": 0.3520512074483561, + "flow_matching_loss": 0.0220167426392436, + "grad_norm": 0.0997326523065567, + "learning_rate": 2.4180000000000002e-05, + "loss": 0.022, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1210 + }, + { + "epoch": 0.35496072155949954, + "flow_matching_loss": 0.02131102532148361, + "grad_norm": 0.12576648592948914, + "learning_rate": 2.438e-05, + "loss": 0.0213, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1220 + }, + { + "epoch": 0.357870235670643, + "flow_matching_loss": 0.02047766651958227, + "grad_norm": 0.10318110138177872, + "learning_rate": 2.4580000000000002e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1230 + }, + { + "epoch": 0.36077974978178645, + "flow_matching_loss": 0.020542478747665883, + "grad_norm": 0.0817994698882103, + "learning_rate": 2.478e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1240 + }, + { + "epoch": 0.3636892638929299, + "flow_matching_loss": 0.022171679884195328, + "grad_norm": 0.09749434888362885, + "learning_rate": 2.498e-05, + "loss": 0.0222, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1250 + }, + { + "epoch": 0.3665987780040733, + "flow_matching_loss": 0.02204299531877041, + "grad_norm": 0.10710253566503525, + "learning_rate": 2.5180000000000003e-05, + "loss": 0.022, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1260 + }, + { + "epoch": 0.36950829211521674, + "flow_matching_loss": 0.0205587706528604, + "grad_norm": 0.0873231515288353, + "learning_rate": 2.538e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1270 + }, + { + "epoch": 0.37241780622636017, + "flow_matching_loss": 0.0237720413133502, + "grad_norm": 0.10638689994812012, + "learning_rate": 2.5580000000000002e-05, + "loss": 0.0238, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1280 + }, + { + "epoch": 0.37532732033750366, + "flow_matching_loss": 0.019293398037552832, + "grad_norm": 0.10804909467697144, + "learning_rate": 2.578e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1290 + }, + { + "epoch": 0.3782368344486471, + "flow_matching_loss": 0.02083021802827716, + "grad_norm": 0.13863927125930786, + "learning_rate": 2.5980000000000002e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1300 + }, + { + "epoch": 0.3811463485597905, + "flow_matching_loss": 0.020754597149789333, + "grad_norm": 0.12038977444171906, + "learning_rate": 2.618e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1310 + }, + { + "epoch": 0.38405586267093395, + "flow_matching_loss": 0.021592473424971104, + "grad_norm": 0.1173630878329277, + "learning_rate": 2.6379999999999998e-05, + "loss": 0.0216, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1320 + }, + { + "epoch": 0.3869653767820774, + "flow_matching_loss": 0.022269449755549432, + "grad_norm": 0.10789163410663605, + "learning_rate": 2.658e-05, + "loss": 0.0223, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1330 + }, + { + "epoch": 0.3898748908932208, + "flow_matching_loss": 0.020603457093238832, + "grad_norm": 0.10580725222826004, + "learning_rate": 2.678e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1340 + }, + { + "epoch": 0.3927844050043643, + "flow_matching_loss": 0.022237404994666577, + "grad_norm": 0.10034766048192978, + "learning_rate": 2.698e-05, + "loss": 0.0222, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1350 + }, + { + "epoch": 0.3956939191155077, + "flow_matching_loss": 0.020723331533372403, + "grad_norm": 0.1466626077890396, + "learning_rate": 2.718e-05, + "loss": 0.0207, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1360 + }, + { + "epoch": 0.39860343322665115, + "flow_matching_loss": 0.020558105409145357, + "grad_norm": 0.10944448411464691, + "learning_rate": 2.738e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1370 + }, + { + "epoch": 0.4015129473377946, + "flow_matching_loss": 0.01889608521014452, + "grad_norm": 0.11380863934755325, + "learning_rate": 2.758e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1380 + }, + { + "epoch": 0.404422461448938, + "flow_matching_loss": 0.021252901200205086, + "grad_norm": 0.10181940346956253, + "learning_rate": 2.778e-05, + "loss": 0.0213, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1390 + }, + { + "epoch": 0.4073319755600815, + "flow_matching_loss": 0.020224009547382592, + "grad_norm": 0.11639661341905594, + "learning_rate": 2.798e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1400 + }, + { + "epoch": 0.4102414896712249, + "flow_matching_loss": 0.022974209673702718, + "grad_norm": 0.14003221690654755, + "learning_rate": 2.818e-05, + "loss": 0.023, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1410 + }, + { + "epoch": 0.41315100378236835, + "flow_matching_loss": 0.019816460832953454, + "grad_norm": 0.11017854511737823, + "learning_rate": 2.838e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1420 + }, + { + "epoch": 0.4160605178935118, + "flow_matching_loss": 0.021631084848195314, + "grad_norm": 0.12799406051635742, + "learning_rate": 2.858e-05, + "loss": 0.0216, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1430 + }, + { + "epoch": 0.4189700320046552, + "flow_matching_loss": 0.020168902445584537, + "grad_norm": 0.09259454905986786, + "learning_rate": 2.8780000000000002e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1440 + }, + { + "epoch": 0.42187954611579864, + "flow_matching_loss": 0.02115853428840637, + "grad_norm": 0.11104249209165573, + "learning_rate": 2.898e-05, + "loss": 0.0212, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1450 + }, + { + "epoch": 0.42478906022694213, + "flow_matching_loss": 0.021364679001271723, + "grad_norm": 0.12400257587432861, + "learning_rate": 2.9180000000000002e-05, + "loss": 0.0214, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1460 + }, + { + "epoch": 0.42769857433808556, + "flow_matching_loss": 0.02315368689596653, + "grad_norm": 0.1773833930492401, + "learning_rate": 2.938e-05, + "loss": 0.0232, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1470 + }, + { + "epoch": 0.430608088449229, + "flow_matching_loss": 0.019988491758704187, + "grad_norm": 0.1312667429447174, + "learning_rate": 2.958e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1480 + }, + { + "epoch": 0.4335176025603724, + "flow_matching_loss": 0.01932177720591426, + "grad_norm": 0.11853079497814178, + "learning_rate": 2.9780000000000003e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1490 + }, + { + "epoch": 0.43642711667151585, + "flow_matching_loss": 0.02010973831638694, + "grad_norm": 0.1530364453792572, + "learning_rate": 2.998e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1500 + }, + { + "epoch": 0.4393366307826593, + "flow_matching_loss": 0.01999103892594576, + "grad_norm": 0.1405227929353714, + "learning_rate": 2.9999992618302022e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1510 + }, + { + "epoch": 0.44224614489380276, + "flow_matching_loss": 0.02071929518133402, + "grad_norm": 0.1671593338251114, + "learning_rate": 2.999996710133069e-05, + "loss": 0.0207, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1520 + }, + { + "epoch": 0.4451556590049462, + "flow_matching_loss": 0.020067112147808076, + "grad_norm": 0.1424213945865631, + "learning_rate": 2.9999923357984923e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1530 + }, + { + "epoch": 0.4480651731160896, + "flow_matching_loss": 0.020038421358913185, + "grad_norm": 0.10518350452184677, + "learning_rate": 2.9999861388317874e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1540 + }, + { + "epoch": 0.45097468722723305, + "flow_matching_loss": 0.022202354855835438, + "grad_norm": 0.1580095887184143, + "learning_rate": 2.9999781192404847e-05, + "loss": 0.0222, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1550 + }, + { + "epoch": 0.4538842013383765, + "flow_matching_loss": 0.020028229244053362, + "grad_norm": 0.1359623670578003, + "learning_rate": 2.9999682770343283e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1560 + }, + { + "epoch": 0.4567937154495199, + "flow_matching_loss": 0.021460407599806785, + "grad_norm": 0.10072334855794907, + "learning_rate": 2.9999566122252777e-05, + "loss": 0.0215, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1570 + }, + { + "epoch": 0.4597032295606634, + "flow_matching_loss": 0.01903377249836922, + "grad_norm": 0.10468201339244843, + "learning_rate": 2.9999431248275065e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1580 + }, + { + "epoch": 0.4626127436718068, + "flow_matching_loss": 0.018969965726137163, + "grad_norm": 0.10318546742200851, + "learning_rate": 2.9999278148574027e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1590 + }, + { + "epoch": 0.46552225778295026, + "flow_matching_loss": 0.020376853831112384, + "grad_norm": 0.09178975969552994, + "learning_rate": 2.9999106823335705e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1600 + }, + { + "epoch": 0.4684317718940937, + "flow_matching_loss": 0.019995273277163505, + "grad_norm": 0.13247737288475037, + "learning_rate": 2.999891727276827e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1610 + }, + { + "epoch": 0.4713412860052371, + "flow_matching_loss": 0.01762162884697318, + "grad_norm": 0.11489103734493256, + "learning_rate": 2.9998709497102037e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1620 + }, + { + "epoch": 0.47425080011638054, + "flow_matching_loss": 0.020912516117095947, + "grad_norm": 0.13112619519233704, + "learning_rate": 2.9998483496589485e-05, + "loss": 0.0209, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1630 + }, + { + "epoch": 0.47716031422752403, + "flow_matching_loss": 0.020522016286849975, + "grad_norm": 0.12172495573759079, + "learning_rate": 2.9998239271505217e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1640 + }, + { + "epoch": 0.48006982833866746, + "flow_matching_loss": 0.02268050741404295, + "grad_norm": 0.09534955769777298, + "learning_rate": 2.9997976822145995e-05, + "loss": 0.0227, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1650 + }, + { + "epoch": 0.4829793424498109, + "flow_matching_loss": 0.020290390960872172, + "grad_norm": 0.10347266495227814, + "learning_rate": 2.9997696148830714e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1660 + }, + { + "epoch": 0.4858888565609543, + "flow_matching_loss": 0.020203891675919294, + "grad_norm": 0.09111194312572479, + "learning_rate": 2.9997397251900423e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1670 + }, + { + "epoch": 0.48879837067209775, + "flow_matching_loss": 0.019269738998264074, + "grad_norm": 0.1077430248260498, + "learning_rate": 2.999708013171831e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1680 + }, + { + "epoch": 0.4917078847832412, + "flow_matching_loss": 0.020102854631841183, + "grad_norm": 0.10666920989751816, + "learning_rate": 2.9996744788669705e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1690 + }, + { + "epoch": 0.49461739889438466, + "flow_matching_loss": 0.021055846102535726, + "grad_norm": 0.11392156034708023, + "learning_rate": 2.999639122316208e-05, + "loss": 0.0211, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1700 + }, + { + "epoch": 0.4975269130055281, + "flow_matching_loss": 0.016953710466623306, + "grad_norm": 0.11685340106487274, + "learning_rate": 2.999601943562505e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1710 + }, + { + "epoch": 0.5004364271166715, + "flow_matching_loss": 0.02197800325229764, + "grad_norm": 0.1250072866678238, + "learning_rate": 2.9995629426510377e-05, + "loss": 0.022, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1720 + }, + { + "epoch": 0.503345941227815, + "flow_matching_loss": 0.02148594781756401, + "grad_norm": 0.1243630200624466, + "learning_rate": 2.9995221196291952e-05, + "loss": 0.0215, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1730 + }, + { + "epoch": 0.5062554553389584, + "flow_matching_loss": 0.019732909463346004, + "grad_norm": 0.12629297375679016, + "learning_rate": 2.999479474546582e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1740 + }, + { + "epoch": 0.5091649694501018, + "flow_matching_loss": 0.019741754233837127, + "grad_norm": 0.09974021464586258, + "learning_rate": 2.9994350074550153e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1750 + }, + { + "epoch": 0.5120744835612453, + "flow_matching_loss": 0.019712957553565502, + "grad_norm": 0.13666243851184845, + "learning_rate": 2.9993887184085273e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1760 + }, + { + "epoch": 0.5149839976723887, + "flow_matching_loss": 0.02061714939773083, + "grad_norm": 0.11800438165664673, + "learning_rate": 2.9993406074633628e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1770 + }, + { + "epoch": 0.5178935117835322, + "flow_matching_loss": 0.019873473048210143, + "grad_norm": 0.12152084708213806, + "learning_rate": 2.9992906746779824e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1780 + }, + { + "epoch": 0.5208030258946756, + "flow_matching_loss": 0.020265040826052426, + "grad_norm": 0.10752566158771515, + "learning_rate": 2.9992389201130576e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1790 + }, + { + "epoch": 0.523712540005819, + "flow_matching_loss": 0.02118642386049032, + "grad_norm": 0.07480551302433014, + "learning_rate": 2.999185343831476e-05, + "loss": 0.0212, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1800 + }, + { + "epoch": 0.5266220541169625, + "flow_matching_loss": 0.020581787824630736, + "grad_norm": 0.13145500421524048, + "learning_rate": 2.999129945898338e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1810 + }, + { + "epoch": 0.5295315682281059, + "flow_matching_loss": 0.01918804869055748, + "grad_norm": 0.1200886145234108, + "learning_rate": 2.999072726380957e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1820 + }, + { + "epoch": 0.5324410823392494, + "flow_matching_loss": 0.0212301567196846, + "grad_norm": 0.10580107569694519, + "learning_rate": 2.9990136853488603e-05, + "loss": 0.0212, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1830 + }, + { + "epoch": 0.5353505964503927, + "flow_matching_loss": 0.01925200065597892, + "grad_norm": 0.12333907932043076, + "learning_rate": 2.9989528228737878e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1840 + }, + { + "epoch": 0.5382601105615362, + "flow_matching_loss": 0.01953992424532771, + "grad_norm": 0.1042298823595047, + "learning_rate": 2.9988901390296935e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1850 + }, + { + "epoch": 0.5411696246726797, + "flow_matching_loss": 0.021093413047492505, + "grad_norm": 0.0908447727560997, + "learning_rate": 2.9988256338927442e-05, + "loss": 0.0211, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1860 + }, + { + "epoch": 0.5440791387838231, + "flow_matching_loss": 0.019173827208578586, + "grad_norm": 0.10514748096466064, + "learning_rate": 2.9987593075413198e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1870 + }, + { + "epoch": 0.5469886528949666, + "flow_matching_loss": 0.020146347768604755, + "grad_norm": 0.1138196513056755, + "learning_rate": 2.9986911600560128e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1880 + }, + { + "epoch": 0.5498981670061099, + "flow_matching_loss": 0.018930347356945276, + "grad_norm": 0.10341677069664001, + "learning_rate": 2.998621191519629e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1890 + }, + { + "epoch": 0.5528076811172534, + "flow_matching_loss": 0.02080009588971734, + "grad_norm": 0.1164831668138504, + "learning_rate": 2.9985494020171873e-05, + "loss": 0.0208, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1900 + }, + { + "epoch": 0.5557171952283969, + "flow_matching_loss": 0.018830547481775282, + "grad_norm": 0.1062106341123581, + "learning_rate": 2.9984757916359176e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1910 + }, + { + "epoch": 0.5586267093395403, + "flow_matching_loss": 0.018672756385058165, + "grad_norm": 0.11313308030366898, + "learning_rate": 2.998400360465264e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1920 + }, + { + "epoch": 0.5615362234506838, + "flow_matching_loss": 0.018890675622969866, + "grad_norm": 0.15707330405712128, + "learning_rate": 2.998323108596883e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1930 + }, + { + "epoch": 0.5644457375618271, + "flow_matching_loss": 0.01981911016628146, + "grad_norm": 0.0957782194018364, + "learning_rate": 2.9982440361246422e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1940 + }, + { + "epoch": 0.5673552516729706, + "flow_matching_loss": 0.01852161679416895, + "grad_norm": 0.11649620532989502, + "learning_rate": 2.998163143144622e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1950 + }, + { + "epoch": 0.570264765784114, + "flow_matching_loss": 0.023156235553324224, + "grad_norm": 0.12081693112850189, + "learning_rate": 2.9980804297551155e-05, + "loss": 0.0232, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1960 + }, + { + "epoch": 0.5731742798952575, + "flow_matching_loss": 0.018351116310805084, + "grad_norm": 0.08874780684709549, + "learning_rate": 2.997995896056627e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1970 + }, + { + "epoch": 0.576083794006401, + "flow_matching_loss": 0.01992413429543376, + "grad_norm": 0.10583867132663727, + "learning_rate": 2.997909542151873e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1980 + }, + { + "epoch": 0.5789933081175443, + "flow_matching_loss": 0.022667080722749235, + "grad_norm": 0.1486353576183319, + "learning_rate": 2.9978213681457823e-05, + "loss": 0.0227, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 1990 + }, + { + "epoch": 0.5819028222286878, + "flow_matching_loss": 0.021632661018520592, + "grad_norm": 0.11213609576225281, + "learning_rate": 2.9977313741454933e-05, + "loss": 0.0216, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2000 + }, + { + "epoch": 0.5848123363398312, + "flow_matching_loss": 0.019710080605000256, + "grad_norm": 0.12285921722650528, + "learning_rate": 2.997639560260358e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2010 + }, + { + "epoch": 0.5877218504509747, + "flow_matching_loss": 0.018698981683701276, + "grad_norm": 0.11094410717487335, + "learning_rate": 2.9975459266019387e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2020 + }, + { + "epoch": 0.5906313645621182, + "flow_matching_loss": 0.01849611895158887, + "grad_norm": 0.13003310561180115, + "learning_rate": 2.9974504732840092e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2030 + }, + { + "epoch": 0.5935408786732616, + "flow_matching_loss": 0.019449546188116073, + "grad_norm": 0.10292180627584457, + "learning_rate": 2.997353200422555e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2040 + }, + { + "epoch": 0.596450392784405, + "flow_matching_loss": 0.019974672980606557, + "grad_norm": 0.08814246952533722, + "learning_rate": 2.99725410813577e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2050 + }, + { + "epoch": 0.5993599068955484, + "flow_matching_loss": 0.0193723713979125, + "grad_norm": 0.1162729486823082, + "learning_rate": 2.997153196544063e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2060 + }, + { + "epoch": 0.6022694210066919, + "flow_matching_loss": 0.01953621506690979, + "grad_norm": 0.0976141020655632, + "learning_rate": 2.997050465770049e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2070 + }, + { + "epoch": 0.6051789351178353, + "flow_matching_loss": 0.020680833142250776, + "grad_norm": 0.13340094685554504, + "learning_rate": 2.9969459159385562e-05, + "loss": 0.0207, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2080 + }, + { + "epoch": 0.6080884492289788, + "flow_matching_loss": 0.019957925472408534, + "grad_norm": 0.10369983315467834, + "learning_rate": 2.9968395471766237e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2090 + }, + { + "epoch": 0.6109979633401222, + "flow_matching_loss": 0.018203141633421183, + "grad_norm": 0.10705326497554779, + "learning_rate": 2.996731359613498e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2100 + }, + { + "epoch": 0.6139074774512656, + "flow_matching_loss": 0.018429558631032704, + "grad_norm": 0.10860796272754669, + "learning_rate": 2.996621353380638e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2110 + }, + { + "epoch": 0.6168169915624091, + "flow_matching_loss": 0.019673612155020236, + "grad_norm": 0.1367141157388687, + "learning_rate": 2.9965095286117113e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2120 + }, + { + "epoch": 0.6197265056735525, + "flow_matching_loss": 0.01858607716858387, + "grad_norm": 0.1014171913266182, + "learning_rate": 2.996395885442596e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2130 + }, + { + "epoch": 0.622636019784696, + "flow_matching_loss": 0.01852706549689174, + "grad_norm": 0.09938299655914307, + "learning_rate": 2.996280424011379e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2140 + }, + { + "epoch": 0.6255455338958394, + "flow_matching_loss": 0.020228541921824217, + "grad_norm": 0.09921959042549133, + "learning_rate": 2.9961631444583565e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2150 + }, + { + "epoch": 0.6284550480069828, + "flow_matching_loss": 0.019531529862433672, + "grad_norm": 0.09321621805429459, + "learning_rate": 2.9960440469260347e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2160 + }, + { + "epoch": 0.6313645621181263, + "flow_matching_loss": 0.01977524161338806, + "grad_norm": 0.11531215906143188, + "learning_rate": 2.995923131559128e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2170 + }, + { + "epoch": 0.6342740762292697, + "flow_matching_loss": 0.018954070191830395, + "grad_norm": 0.10475543886423111, + "learning_rate": 2.99580039850456e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2180 + }, + { + "epoch": 0.6371835903404132, + "flow_matching_loss": 0.02028482425957918, + "grad_norm": 0.09762804210186005, + "learning_rate": 2.9956758479114635e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2190 + }, + { + "epoch": 0.6400931044515565, + "flow_matching_loss": 0.019801801815629006, + "grad_norm": 0.09176918119192123, + "learning_rate": 2.9955494799311787e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2200 + }, + { + "epoch": 0.6430026185627, + "flow_matching_loss": 0.019932084996253253, + "grad_norm": 0.09326978772878647, + "learning_rate": 2.995421294717254e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2210 + }, + { + "epoch": 0.6459121326738435, + "flow_matching_loss": 0.019505486264824866, + "grad_norm": 0.1010829284787178, + "learning_rate": 2.995291292425448e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2220 + }, + { + "epoch": 0.6488216467849869, + "flow_matching_loss": 0.01829868135973811, + "grad_norm": 0.11528094112873077, + "learning_rate": 2.9951594732137248e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2230 + }, + { + "epoch": 0.6517311608961304, + "flow_matching_loss": 0.019126165099442004, + "grad_norm": 0.1263420730829239, + "learning_rate": 2.995025837242257e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2240 + }, + { + "epoch": 0.6546406750072737, + "flow_matching_loss": 0.019707585498690606, + "grad_norm": 0.12261626124382019, + "learning_rate": 2.994890384673426e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2250 + }, + { + "epoch": 0.6575501891184172, + "flow_matching_loss": 0.016967683844268323, + "grad_norm": 0.08948231488466263, + "learning_rate": 2.9947531156718174e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2260 + }, + { + "epoch": 0.6604597032295607, + "flow_matching_loss": 0.019103033188730478, + "grad_norm": 0.12263049930334091, + "learning_rate": 2.994614030404228e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2270 + }, + { + "epoch": 0.6633692173407041, + "flow_matching_loss": 0.019601563178002834, + "grad_norm": 0.09547588974237442, + "learning_rate": 2.994473129039659e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2280 + }, + { + "epoch": 0.6662787314518476, + "flow_matching_loss": 0.018714132905006408, + "grad_norm": 0.09066019207239151, + "learning_rate": 2.9943304117493183e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2290 + }, + { + "epoch": 0.669188245562991, + "flow_matching_loss": 0.020278374198824167, + "grad_norm": 0.0773807168006897, + "learning_rate": 2.9941858787066207e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2300 + }, + { + "epoch": 0.6720977596741344, + "flow_matching_loss": 0.020187490433454514, + "grad_norm": 0.08610354363918304, + "learning_rate": 2.9940395300871888e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2310 + }, + { + "epoch": 0.6750072737852778, + "flow_matching_loss": 0.019189943093806505, + "grad_norm": 0.1062404066324234, + "learning_rate": 2.993891366068849e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2320 + }, + { + "epoch": 0.6779167878964213, + "flow_matching_loss": 0.018165397830307482, + "grad_norm": 0.08411315828561783, + "learning_rate": 2.993741386831634e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2330 + }, + { + "epoch": 0.6808263020075648, + "flow_matching_loss": 0.01918727345764637, + "grad_norm": 0.11183233559131622, + "learning_rate": 2.9935895925577837e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2340 + }, + { + "epoch": 0.6837358161187082, + "flow_matching_loss": 0.021217321045696735, + "grad_norm": 0.10651082545518875, + "learning_rate": 2.9934359834317425e-05, + "loss": 0.0212, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2350 + }, + { + "epoch": 0.6866453302298516, + "flow_matching_loss": 0.02012283932417631, + "grad_norm": 0.11884887516498566, + "learning_rate": 2.993280559640159e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2360 + }, + { + "epoch": 0.689554844340995, + "flow_matching_loss": 0.019339125044643878, + "grad_norm": 0.09141652286052704, + "learning_rate": 2.9931233213718887e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2370 + }, + { + "epoch": 0.6924643584521385, + "flow_matching_loss": 0.020934330020099878, + "grad_norm": 0.11127802729606628, + "learning_rate": 2.992964268817991e-05, + "loss": 0.0209, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2380 + }, + { + "epoch": 0.695373872563282, + "flow_matching_loss": 0.0200210259296, + "grad_norm": 0.10906796902418137, + "learning_rate": 2.9928034021717286e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2390 + }, + { + "epoch": 0.6982833866744254, + "flow_matching_loss": 0.01889802943915129, + "grad_norm": 0.11454474925994873, + "learning_rate": 2.9926407216285707e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2400 + }, + { + "epoch": 0.7011929007855688, + "flow_matching_loss": 0.021753561682999134, + "grad_norm": 0.13627105951309204, + "learning_rate": 2.9924762273861894e-05, + "loss": 0.0218, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2410 + }, + { + "epoch": 0.7041024148967122, + "flow_matching_loss": 0.02033474128693342, + "grad_norm": 0.11298619955778122, + "learning_rate": 2.992309919644461e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2420 + }, + { + "epoch": 0.7070119290078557, + "flow_matching_loss": 0.019044611137360334, + "grad_norm": 0.09199131280183792, + "learning_rate": 2.9921417986054633e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2430 + }, + { + "epoch": 0.7099214431189991, + "flow_matching_loss": 0.019636241346597673, + "grad_norm": 0.083233542740345, + "learning_rate": 2.9919718644734815e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2440 + }, + { + "epoch": 0.7128309572301426, + "flow_matching_loss": 0.01871565794572234, + "grad_norm": 0.09986811876296997, + "learning_rate": 2.991800117455e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2450 + }, + { + "epoch": 0.715740471341286, + "flow_matching_loss": 0.02048503402620554, + "grad_norm": 0.09106690436601639, + "learning_rate": 2.9916265577587085e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2460 + }, + { + "epoch": 0.7186499854524294, + "flow_matching_loss": 0.019900908414274453, + "grad_norm": 0.10259341448545456, + "learning_rate": 2.9914511855954986e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2470 + }, + { + "epoch": 0.7215594995635729, + "flow_matching_loss": 0.020150802936404942, + "grad_norm": 0.0728544369339943, + "learning_rate": 2.9912740011784626e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2480 + }, + { + "epoch": 0.7244690136747163, + "flow_matching_loss": 0.019740773364901543, + "grad_norm": 0.10861191898584366, + "learning_rate": 2.991095004722898e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2490 + }, + { + "epoch": 0.7273785277858598, + "flow_matching_loss": 0.01992887780070305, + "grad_norm": 0.13408491015434265, + "learning_rate": 2.9909141964463017e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2500 + }, + { + "epoch": 0.7302880418970032, + "flow_matching_loss": 0.020389516185969115, + "grad_norm": 0.13542470335960388, + "learning_rate": 2.990731576568372e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2510 + }, + { + "epoch": 0.7331975560081466, + "flow_matching_loss": 0.01940856035798788, + "grad_norm": 0.11694429069757462, + "learning_rate": 2.99054714531101e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2520 + }, + { + "epoch": 0.7361070701192901, + "flow_matching_loss": 0.01872241562232375, + "grad_norm": 0.10760446637868881, + "learning_rate": 2.9903609028983178e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2530 + }, + { + "epoch": 0.7390165842304335, + "flow_matching_loss": 0.018060475308448076, + "grad_norm": 0.11932144314050674, + "learning_rate": 2.990172849556596e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2540 + }, + { + "epoch": 0.741926098341577, + "flow_matching_loss": 0.020471792202442884, + "grad_norm": 0.12167076766490936, + "learning_rate": 2.989982985514348e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2550 + }, + { + "epoch": 0.7448356124527203, + "flow_matching_loss": 0.018810376524925232, + "grad_norm": 0.09541167318820953, + "learning_rate": 2.9897913110022766e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2560 + }, + { + "epoch": 0.7477451265638638, + "flow_matching_loss": 0.020371688809245826, + "grad_norm": 0.13169442117214203, + "learning_rate": 2.989597826253284e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2570 + }, + { + "epoch": 0.7506546406750073, + "flow_matching_loss": 0.02055797856301069, + "grad_norm": 0.1015075221657753, + "learning_rate": 2.989402531502473e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2580 + }, + { + "epoch": 0.7535641547861507, + "flow_matching_loss": 0.017610469553619622, + "grad_norm": 0.0790795162320137, + "learning_rate": 2.989205426987144e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2590 + }, + { + "epoch": 0.7564736688972942, + "flow_matching_loss": 0.019107090961188077, + "grad_norm": 0.09036987274885178, + "learning_rate": 2.9890065129467988e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2600 + }, + { + "epoch": 0.7593831830084375, + "flow_matching_loss": 0.018688016477972268, + "grad_norm": 0.12186135351657867, + "learning_rate": 2.9888057896231364e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2610 + }, + { + "epoch": 0.762292697119581, + "flow_matching_loss": 0.019101955089718105, + "grad_norm": 0.09940166026353836, + "learning_rate": 2.988603257260054e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2620 + }, + { + "epoch": 0.7652022112307245, + "flow_matching_loss": 0.020139879453927277, + "grad_norm": 0.1332274079322815, + "learning_rate": 2.988398916103648e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2630 + }, + { + "epoch": 0.7681117253418679, + "flow_matching_loss": 0.019053262192755938, + "grad_norm": 0.10380490124225616, + "learning_rate": 2.988192766402212e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2640 + }, + { + "epoch": 0.7710212394530114, + "flow_matching_loss": 0.019164971262216567, + "grad_norm": 0.08661839365959167, + "learning_rate": 2.9879848084062378e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2650 + }, + { + "epoch": 0.7739307535641547, + "flow_matching_loss": 0.01818268233910203, + "grad_norm": 0.09152336418628693, + "learning_rate": 2.9877750423684132e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2660 + }, + { + "epoch": 0.7768402676752982, + "flow_matching_loss": 0.018276084307581188, + "grad_norm": 0.10524634271860123, + "learning_rate": 2.987563468543624e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2670 + }, + { + "epoch": 0.7797497817864416, + "flow_matching_loss": 0.018180299177765848, + "grad_norm": 0.09563110768795013, + "learning_rate": 2.9873500871889533e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2680 + }, + { + "epoch": 0.7826592958975851, + "flow_matching_loss": 0.017704789340496064, + "grad_norm": 0.0965760201215744, + "learning_rate": 2.9871348985636778e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2690 + }, + { + "epoch": 0.7855688100087286, + "flow_matching_loss": 0.019652227871119976, + "grad_norm": 0.10202546417713165, + "learning_rate": 2.986917902929273e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2700 + }, + { + "epoch": 0.788478324119872, + "flow_matching_loss": 0.02003510557115078, + "grad_norm": 0.1031411811709404, + "learning_rate": 2.9866991005494095e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2710 + }, + { + "epoch": 0.7913878382310154, + "flow_matching_loss": 0.016886166110634804, + "grad_norm": 0.07532360404729843, + "learning_rate": 2.986478491689952e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2720 + }, + { + "epoch": 0.7942973523421588, + "flow_matching_loss": 0.017122420202940702, + "grad_norm": 0.09441164880990982, + "learning_rate": 2.986256076618961e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2730 + }, + { + "epoch": 0.7972068664533023, + "flow_matching_loss": 0.01825088057667017, + "grad_norm": 0.09816018491983414, + "learning_rate": 2.986031855606692e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2740 + }, + { + "epoch": 0.8001163805644458, + "flow_matching_loss": 0.020588923804461957, + "grad_norm": 0.13582023978233337, + "learning_rate": 2.9858058289255954e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2750 + }, + { + "epoch": 0.8030258946755892, + "flow_matching_loss": 0.018678577337414026, + "grad_norm": 0.09058792144060135, + "learning_rate": 2.985577996850314e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2760 + }, + { + "epoch": 0.8059354087867326, + "flow_matching_loss": 0.018190510291606187, + "grad_norm": 0.10466286540031433, + "learning_rate": 2.9853483596576856e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2770 + }, + { + "epoch": 0.808844922897876, + "flow_matching_loss": 0.021390361711382866, + "grad_norm": 0.10273987054824829, + "learning_rate": 2.9851169176267412e-05, + "loss": 0.0214, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2780 + }, + { + "epoch": 0.8117544370090195, + "flow_matching_loss": 0.020584381744265558, + "grad_norm": 0.13044871389865875, + "learning_rate": 2.9848836710387044e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2790 + }, + { + "epoch": 0.814663951120163, + "flow_matching_loss": 0.020699620340019466, + "grad_norm": 0.09167768806219101, + "learning_rate": 2.9846486201769917e-05, + "loss": 0.0207, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2800 + }, + { + "epoch": 0.8175734652313064, + "flow_matching_loss": 0.020645774342119692, + "grad_norm": 0.1127953827381134, + "learning_rate": 2.9844117653272122e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2810 + }, + { + "epoch": 0.8204829793424498, + "flow_matching_loss": 0.01860180599614978, + "grad_norm": 0.09855547547340393, + "learning_rate": 2.9841731067771676e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2820 + }, + { + "epoch": 0.8233924934535932, + "flow_matching_loss": 0.019292312301695348, + "grad_norm": 0.11446227878332138, + "learning_rate": 2.9839326448168492e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2830 + }, + { + "epoch": 0.8263020075647367, + "flow_matching_loss": 0.019281926285475492, + "grad_norm": 0.11484672874212265, + "learning_rate": 2.983690379738442e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2840 + }, + { + "epoch": 0.8292115216758801, + "flow_matching_loss": 0.018239854834973812, + "grad_norm": 0.07844091951847076, + "learning_rate": 2.9834463118363204e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2850 + }, + { + "epoch": 0.8321210357870236, + "flow_matching_loss": 0.01984337978065014, + "grad_norm": 0.12307356297969818, + "learning_rate": 2.98320044140705e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2860 + }, + { + "epoch": 0.835030549898167, + "flow_matching_loss": 0.020340707339346408, + "grad_norm": 0.09291069209575653, + "learning_rate": 2.9829527687493868e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2870 + }, + { + "epoch": 0.8379400640093104, + "flow_matching_loss": 0.017524137627333403, + "grad_norm": 0.09746887534856796, + "learning_rate": 2.9827032941642765e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2880 + }, + { + "epoch": 0.8408495781204539, + "flow_matching_loss": 0.018969420809298754, + "grad_norm": 0.11097733676433563, + "learning_rate": 2.9824520179548535e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2890 + }, + { + "epoch": 0.8437590922315973, + "flow_matching_loss": 0.017707086633890867, + "grad_norm": 0.09318418055772781, + "learning_rate": 2.9821989404264428e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2900 + }, + { + "epoch": 0.8466686063427408, + "flow_matching_loss": 0.019095855299383402, + "grad_norm": 0.06212354451417923, + "learning_rate": 2.9819440618865572e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2910 + }, + { + "epoch": 0.8495781204538843, + "flow_matching_loss": 0.017092576250433922, + "grad_norm": 0.09125030040740967, + "learning_rate": 2.981687382644898e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2920 + }, + { + "epoch": 0.8524876345650276, + "flow_matching_loss": 0.01896805511787534, + "grad_norm": 0.09255921095609665, + "learning_rate": 2.9814289030133556e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2930 + }, + { + "epoch": 0.8553971486761711, + "flow_matching_loss": 0.0185190468095243, + "grad_norm": 0.10173816978931427, + "learning_rate": 2.9811686233060058e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2940 + }, + { + "epoch": 0.8583066627873145, + "flow_matching_loss": 0.017045192420482635, + "grad_norm": 0.07469920814037323, + "learning_rate": 2.9809065438391138e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2950 + }, + { + "epoch": 0.861216176898458, + "flow_matching_loss": 0.019032335933297874, + "grad_norm": 0.12077812850475311, + "learning_rate": 2.9806426649311304e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2960 + }, + { + "epoch": 0.8641256910096013, + "flow_matching_loss": 0.020110293477773666, + "grad_norm": 0.11862783133983612, + "learning_rate": 2.980376986902694e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2970 + }, + { + "epoch": 0.8670352051207448, + "flow_matching_loss": 0.018931299820542334, + "grad_norm": 0.10683681815862656, + "learning_rate": 2.9801095100766276e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2980 + }, + { + "epoch": 0.8699447192318883, + "flow_matching_loss": 0.01842739023268223, + "grad_norm": 0.08966946601867676, + "learning_rate": 2.979840234777942e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 2990 + }, + { + "epoch": 0.8728542333430317, + "flow_matching_loss": 0.018935232795774935, + "grad_norm": 0.1229260191321373, + "learning_rate": 2.979569161333831e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3000 + }, + { + "epoch": 0.8757637474541752, + "flow_matching_loss": 0.020045480318367483, + "grad_norm": 0.11845389008522034, + "learning_rate": 2.9792962900736744e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3010 + }, + { + "epoch": 0.8786732615653186, + "flow_matching_loss": 0.018675920739769935, + "grad_norm": 0.10521237552165985, + "learning_rate": 2.9790216213290372e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3020 + }, + { + "epoch": 0.881582775676462, + "flow_matching_loss": 0.018332845624536277, + "grad_norm": 0.11342164874076843, + "learning_rate": 2.9787451554336672e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3030 + }, + { + "epoch": 0.8844922897876055, + "flow_matching_loss": 0.02057248931378126, + "grad_norm": 0.09816586226224899, + "learning_rate": 2.9784668927234972e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3040 + }, + { + "epoch": 0.8874018038987489, + "flow_matching_loss": 0.017975552752614022, + "grad_norm": 0.11504170298576355, + "learning_rate": 2.9781868335366418e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3050 + }, + { + "epoch": 0.8903113180098924, + "flow_matching_loss": 0.020855865441262722, + "grad_norm": 0.1349685788154602, + "learning_rate": 2.9779049782134e-05, + "loss": 0.0209, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3060 + }, + { + "epoch": 0.8932208321210358, + "flow_matching_loss": 0.019627457205206155, + "grad_norm": 0.10752315819263458, + "learning_rate": 2.9776213270962523e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3070 + }, + { + "epoch": 0.8961303462321792, + "flow_matching_loss": 0.0175611631013453, + "grad_norm": 0.11468195170164108, + "learning_rate": 2.977335880529861e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3080 + }, + { + "epoch": 0.8990398603433226, + "flow_matching_loss": 0.016723378840833903, + "grad_norm": 0.10321617871522903, + "learning_rate": 2.977048638861072e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3090 + }, + { + "epoch": 0.9019493744544661, + "flow_matching_loss": 0.01914457157254219, + "grad_norm": 0.08558323979377747, + "learning_rate": 2.97675960243891e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3100 + }, + { + "epoch": 0.9048588885656096, + "flow_matching_loss": 0.0188642761670053, + "grad_norm": 0.1350438892841339, + "learning_rate": 2.976468771614582e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3110 + }, + { + "epoch": 0.907768402676753, + "flow_matching_loss": 0.01791706942021847, + "grad_norm": 0.09812701493501663, + "learning_rate": 2.9761761467414736e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3120 + }, + { + "epoch": 0.9106779167878964, + "flow_matching_loss": 0.017245538905262946, + "grad_norm": 0.07750106602907181, + "learning_rate": 2.9758817281751538e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3130 + }, + { + "epoch": 0.9135874308990398, + "flow_matching_loss": 0.01673461114987731, + "grad_norm": 0.11362973600625992, + "learning_rate": 2.975585516273367e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3140 + }, + { + "epoch": 0.9164969450101833, + "flow_matching_loss": 0.01897493125870824, + "grad_norm": 0.08284789323806763, + "learning_rate": 2.975287511396039e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3150 + }, + { + "epoch": 0.9194064591213268, + "flow_matching_loss": 0.017540675587952137, + "grad_norm": 0.11142988502979279, + "learning_rate": 2.9749877139052745e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3160 + }, + { + "epoch": 0.9223159732324702, + "flow_matching_loss": 0.018251173943281174, + "grad_norm": 0.09272816777229309, + "learning_rate": 2.9746861241653555e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3170 + }, + { + "epoch": 0.9252254873436137, + "flow_matching_loss": 0.020428431034088135, + "grad_norm": 0.11305379122495651, + "learning_rate": 2.9743827425427416e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3180 + }, + { + "epoch": 0.928135001454757, + "flow_matching_loss": 0.017657786328345536, + "grad_norm": 0.10694064944982529, + "learning_rate": 2.9740775694060704e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3190 + }, + { + "epoch": 0.9310445155659005, + "flow_matching_loss": 0.01985239628702402, + "grad_norm": 0.09028897434473038, + "learning_rate": 2.973770605126156e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3200 + }, + { + "epoch": 0.9339540296770439, + "flow_matching_loss": 0.020530006662011147, + "grad_norm": 0.11404499411582947, + "learning_rate": 2.9734618500759886e-05, + "loss": 0.0205, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3210 + }, + { + "epoch": 0.9368635437881874, + "flow_matching_loss": 0.018791144527494908, + "grad_norm": 0.1309749037027359, + "learning_rate": 2.9731513046307356e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3220 + }, + { + "epoch": 0.9397730578993309, + "flow_matching_loss": 0.018499893508851527, + "grad_norm": 0.13816294074058533, + "learning_rate": 2.972838969167738e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3230 + }, + { + "epoch": 0.9426825720104742, + "flow_matching_loss": 0.017074046842753886, + "grad_norm": 0.10228951275348663, + "learning_rate": 2.972524844066513e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3240 + }, + { + "epoch": 0.9455920861216177, + "flow_matching_loss": 0.019805200211703778, + "grad_norm": 0.09924551844596863, + "learning_rate": 2.9722089297087524e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3250 + }, + { + "epoch": 0.9485016002327611, + "flow_matching_loss": 0.019117241818457843, + "grad_norm": 0.08520963788032532, + "learning_rate": 2.971891226478322e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3260 + }, + { + "epoch": 0.9514111143439046, + "flow_matching_loss": 0.017020434513688086, + "grad_norm": 0.11019929498434067, + "learning_rate": 2.9715717347612615e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3270 + }, + { + "epoch": 0.9543206284550481, + "flow_matching_loss": 0.01883316710591316, + "grad_norm": 0.1212393045425415, + "learning_rate": 2.971250454945783e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3280 + }, + { + "epoch": 0.9572301425661914, + "flow_matching_loss": 0.017708388343453407, + "grad_norm": 0.09742605686187744, + "learning_rate": 2.970927387422272e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3290 + }, + { + "epoch": 0.9601396566773349, + "flow_matching_loss": 0.019735865481197833, + "grad_norm": 0.11007943749427795, + "learning_rate": 2.9706025325832858e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3300 + }, + { + "epoch": 0.9630491707884783, + "flow_matching_loss": 0.019633201602846383, + "grad_norm": 0.09405940026044846, + "learning_rate": 2.9702758908235534e-05, + "loss": 0.0196, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3310 + }, + { + "epoch": 0.9659586848996218, + "flow_matching_loss": 0.021007805317640304, + "grad_norm": 0.1009920984506607, + "learning_rate": 2.9699474625399765e-05, + "loss": 0.021, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3320 + }, + { + "epoch": 0.9688681990107652, + "flow_matching_loss": 0.019727158918976785, + "grad_norm": 0.10755439847707748, + "learning_rate": 2.9696172481316246e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3330 + }, + { + "epoch": 0.9717777131219086, + "flow_matching_loss": 0.018424294143915176, + "grad_norm": 0.11721062660217285, + "learning_rate": 2.9692852479997407e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3340 + }, + { + "epoch": 0.9746872272330521, + "flow_matching_loss": 0.018126681353896855, + "grad_norm": 0.12135332822799683, + "learning_rate": 2.9689514625477364e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3350 + }, + { + "epoch": 0.9775967413441955, + "flow_matching_loss": 0.021455384977161884, + "grad_norm": 0.12670359015464783, + "learning_rate": 2.968615892181192e-05, + "loss": 0.0215, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3360 + }, + { + "epoch": 0.980506255455339, + "flow_matching_loss": 0.018364130333065987, + "grad_norm": 0.09889969974756241, + "learning_rate": 2.9682785373078567e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3370 + }, + { + "epoch": 0.9834157695664824, + "flow_matching_loss": 0.020633623655885457, + "grad_norm": 0.10356268286705017, + "learning_rate": 2.9679393983376493e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3380 + }, + { + "epoch": 0.9863252836776258, + "flow_matching_loss": 0.019476023223251104, + "grad_norm": 0.14022786915302277, + "learning_rate": 2.9675984756826545e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3390 + }, + { + "epoch": 0.9892347977887693, + "flow_matching_loss": 0.018836218398064375, + "grad_norm": 0.09801580011844635, + "learning_rate": 2.9672557697571273e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3400 + }, + { + "epoch": 0.9921443118999127, + "flow_matching_loss": 0.021529276482760908, + "grad_norm": 0.09024018794298172, + "learning_rate": 2.966911280977486e-05, + "loss": 0.0215, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3410 + }, + { + "epoch": 0.9950538260110562, + "flow_matching_loss": 0.01814887849614024, + "grad_norm": 0.10071894526481628, + "learning_rate": 2.966565009762318e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3420 + }, + { + "epoch": 0.9979633401221996, + "flow_matching_loss": 0.018544204998761415, + "grad_norm": 0.08996984362602234, + "learning_rate": 2.966216956532375e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3430 + }, + { + "epoch": 1.000872854233343, + "flow_matching_loss": 0.019230186380445956, + "grad_norm": 0.09342276304960251, + "learning_rate": 2.9658671217105744e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3440 + }, + { + "epoch": 1.0037823683444864, + "flow_matching_loss": 0.01844645235687494, + "grad_norm": 0.12514850497245789, + "learning_rate": 2.965515505721999e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3450 + }, + { + "epoch": 1.00669188245563, + "flow_matching_loss": 0.0201486811041832, + "grad_norm": 0.08639591932296753, + "learning_rate": 2.9651621089938948e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3460 + }, + { + "epoch": 1.0096013965667734, + "flow_matching_loss": 0.01730583868920803, + "grad_norm": 0.09231448173522949, + "learning_rate": 2.9648069319556725e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3470 + }, + { + "epoch": 1.0125109106779169, + "flow_matching_loss": 0.01858885083347559, + "grad_norm": 0.09006810188293457, + "learning_rate": 2.9644499750389057e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3480 + }, + { + "epoch": 1.0154204247890601, + "flow_matching_loss": 0.019374042563140393, + "grad_norm": 0.08764363080263138, + "learning_rate": 2.9640912386773305e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3490 + }, + { + "epoch": 1.0183299389002036, + "flow_matching_loss": 0.01749600050970912, + "grad_norm": 0.08894215524196625, + "learning_rate": 2.9637307233068455e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3500 + }, + { + "epoch": 1.021239453011347, + "flow_matching_loss": 0.017796285822987555, + "grad_norm": 0.08896324783563614, + "learning_rate": 2.9633684293655108e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3510 + }, + { + "epoch": 1.0241489671224906, + "flow_matching_loss": 0.017840042337775232, + "grad_norm": 0.08360375463962555, + "learning_rate": 2.963004357293548e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3520 + }, + { + "epoch": 1.027058481233634, + "flow_matching_loss": 0.017205945681780577, + "grad_norm": 0.09024953097105026, + "learning_rate": 2.962638507533339e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3530 + }, + { + "epoch": 1.0299679953447773, + "flow_matching_loss": 0.01806115871295333, + "grad_norm": 0.07882887870073318, + "learning_rate": 2.962270880529425e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3540 + }, + { + "epoch": 1.0328775094559208, + "flow_matching_loss": 0.018254214618355035, + "grad_norm": 0.07874168455600739, + "learning_rate": 2.9619014767285084e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3550 + }, + { + "epoch": 1.0357870235670643, + "flow_matching_loss": 0.018135436810553074, + "grad_norm": 0.06533890962600708, + "learning_rate": 2.9615302965794488e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3560 + }, + { + "epoch": 1.0386965376782078, + "flow_matching_loss": 0.018602319434285162, + "grad_norm": 0.0778808519244194, + "learning_rate": 2.961157340533266e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3570 + }, + { + "epoch": 1.0416060517893513, + "flow_matching_loss": 0.017338017653673887, + "grad_norm": 0.11539080739021301, + "learning_rate": 2.9607826090431364e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3580 + }, + { + "epoch": 1.0445155659004945, + "flow_matching_loss": 0.018564752768725157, + "grad_norm": 0.0963340625166893, + "learning_rate": 2.9604061025643936e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3590 + }, + { + "epoch": 1.047425080011638, + "flow_matching_loss": 0.018788988050073386, + "grad_norm": 0.10232310742139816, + "learning_rate": 2.9600278215545292e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3600 + }, + { + "epoch": 1.0503345941227815, + "flow_matching_loss": 0.016897898633033037, + "grad_norm": 0.07842380553483963, + "learning_rate": 2.95964776647319e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3610 + }, + { + "epoch": 1.053244108233925, + "flow_matching_loss": 0.019477718509733677, + "grad_norm": 0.08044180274009705, + "learning_rate": 2.9592659377821787e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3620 + }, + { + "epoch": 1.0561536223450685, + "flow_matching_loss": 0.017526969127357007, + "grad_norm": 0.0841904878616333, + "learning_rate": 2.9588823359454528e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3630 + }, + { + "epoch": 1.0590631364562118, + "flow_matching_loss": 0.01902558971196413, + "grad_norm": 0.07169728726148605, + "learning_rate": 2.958496961429126e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3640 + }, + { + "epoch": 1.0619726505673552, + "flow_matching_loss": 0.01901453621685505, + "grad_norm": 0.1217823326587677, + "learning_rate": 2.9581098147014628e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3650 + }, + { + "epoch": 1.0648821646784987, + "flow_matching_loss": 0.01976873641833663, + "grad_norm": 0.09083980321884155, + "learning_rate": 2.9577208962328843e-05, + "loss": 0.0198, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3660 + }, + { + "epoch": 1.0677916787896422, + "flow_matching_loss": 0.017811258602887393, + "grad_norm": 0.10659024864435196, + "learning_rate": 2.9573302064959633e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3670 + }, + { + "epoch": 1.0707011929007855, + "flow_matching_loss": 0.020120925828814508, + "grad_norm": 0.11351529508829117, + "learning_rate": 2.9569377459654235e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3680 + }, + { + "epoch": 1.073610707011929, + "flow_matching_loss": 0.019447321072220802, + "grad_norm": 0.11186040192842484, + "learning_rate": 2.9565435151181424e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3690 + }, + { + "epoch": 1.0765202211230724, + "flow_matching_loss": 0.01904837815091014, + "grad_norm": 0.08647958189249039, + "learning_rate": 2.9561475144331473e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3700 + }, + { + "epoch": 1.079429735234216, + "flow_matching_loss": 0.01641654400154948, + "grad_norm": 0.08734160661697388, + "learning_rate": 2.9557497443916162e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3710 + }, + { + "epoch": 1.0823392493453594, + "flow_matching_loss": 0.01808957802131772, + "grad_norm": 0.09798106551170349, + "learning_rate": 2.9553502054768774e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3720 + }, + { + "epoch": 1.0852487634565027, + "flow_matching_loss": 0.018561636749655008, + "grad_norm": 0.12198197841644287, + "learning_rate": 2.954948898174409e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3730 + }, + { + "epoch": 1.0881582775676462, + "flow_matching_loss": 0.01872598649933934, + "grad_norm": 0.07937467098236084, + "learning_rate": 2.954545822971836e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3740 + }, + { + "epoch": 1.0910677916787896, + "flow_matching_loss": 0.0176808575168252, + "grad_norm": 0.09905891865491867, + "learning_rate": 2.9541409803589335e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3750 + }, + { + "epoch": 1.0939773057899331, + "flow_matching_loss": 0.01882423395290971, + "grad_norm": 0.10036731511354446, + "learning_rate": 2.9537343708276235e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3760 + }, + { + "epoch": 1.0968868199010766, + "flow_matching_loss": 0.01828415635973215, + "grad_norm": 0.08646246790885925, + "learning_rate": 2.953325994871975e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3770 + }, + { + "epoch": 1.0997963340122199, + "flow_matching_loss": 0.01970659540966153, + "grad_norm": 0.11852510273456573, + "learning_rate": 2.9529158529882023e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3780 + }, + { + "epoch": 1.1027058481233634, + "flow_matching_loss": 0.016362350899726152, + "grad_norm": 0.11255858838558197, + "learning_rate": 2.9525039456746677e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3790 + }, + { + "epoch": 1.1056153622345068, + "flow_matching_loss": 0.017050331737846137, + "grad_norm": 0.10436110198497772, + "learning_rate": 2.952090273431877e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3800 + }, + { + "epoch": 1.1085248763456503, + "flow_matching_loss": 0.017412279453128576, + "grad_norm": 0.08984997123479843, + "learning_rate": 2.9516748367624808e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3810 + }, + { + "epoch": 1.1114343904567938, + "flow_matching_loss": 0.019079350214451552, + "grad_norm": 0.08538604527711868, + "learning_rate": 2.951257636171274e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3820 + }, + { + "epoch": 1.114343904567937, + "flow_matching_loss": 0.020250942930579184, + "grad_norm": 0.08933316171169281, + "learning_rate": 2.9508386721651944e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3830 + }, + { + "epoch": 1.1172534186790806, + "flow_matching_loss": 0.018861791398376227, + "grad_norm": 0.08725470304489136, + "learning_rate": 2.9504179452533228e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3840 + }, + { + "epoch": 1.120162932790224, + "flow_matching_loss": 0.018909188732504845, + "grad_norm": 0.1470394879579544, + "learning_rate": 2.9499954559468826e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3850 + }, + { + "epoch": 1.1230724469013675, + "flow_matching_loss": 0.017731254920363426, + "grad_norm": 0.10436313599348068, + "learning_rate": 2.9495712047592378e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3860 + }, + { + "epoch": 1.1259819610125108, + "flow_matching_loss": 0.017720675747841596, + "grad_norm": 0.08774556964635849, + "learning_rate": 2.949145192205893e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3870 + }, + { + "epoch": 1.1288914751236543, + "flow_matching_loss": 0.018911564722657204, + "grad_norm": 0.09040176123380661, + "learning_rate": 2.9487174188044953e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3880 + }, + { + "epoch": 1.1318009892347978, + "flow_matching_loss": 0.019208915065973997, + "grad_norm": 0.10024567693471909, + "learning_rate": 2.9482878850748276e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3890 + }, + { + "epoch": 1.1347105033459413, + "flow_matching_loss": 0.018112357053905726, + "grad_norm": 0.08633365482091904, + "learning_rate": 2.9478565915388157e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3900 + }, + { + "epoch": 1.1376200174570847, + "flow_matching_loss": 0.016116305347532035, + "grad_norm": 0.10258221626281738, + "learning_rate": 2.9474235387205207e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3910 + }, + { + "epoch": 1.140529531568228, + "flow_matching_loss": 0.018468044232577085, + "grad_norm": 0.08124295622110367, + "learning_rate": 2.946988727146143e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3920 + }, + { + "epoch": 1.1434390456793715, + "flow_matching_loss": 0.017930401768535376, + "grad_norm": 0.0944528877735138, + "learning_rate": 2.9465521573440196e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3930 + }, + { + "epoch": 1.146348559790515, + "flow_matching_loss": 0.01819923985749483, + "grad_norm": 0.10765781253576279, + "learning_rate": 2.946113829844624e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3940 + }, + { + "epoch": 1.1492580739016585, + "flow_matching_loss": 0.018257397320121528, + "grad_norm": 0.09298393130302429, + "learning_rate": 2.945673745180566e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3950 + }, + { + "epoch": 1.152167588012802, + "flow_matching_loss": 0.018548332154750824, + "grad_norm": 0.09478645026683807, + "learning_rate": 2.945231903886589e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3960 + }, + { + "epoch": 1.1550771021239452, + "flow_matching_loss": 0.018170183431357144, + "grad_norm": 0.08638930320739746, + "learning_rate": 2.9447883064995728e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3970 + }, + { + "epoch": 1.1579866162350887, + "flow_matching_loss": 0.019544500578194856, + "grad_norm": 0.09964679926633835, + "learning_rate": 2.9443429535585288e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3980 + }, + { + "epoch": 1.1608961303462322, + "flow_matching_loss": 0.020855669118463992, + "grad_norm": 0.10663013160228729, + "learning_rate": 2.9438958456046044e-05, + "loss": 0.0209, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 3990 + }, + { + "epoch": 1.1638056444573757, + "flow_matching_loss": 0.01783746536821127, + "grad_norm": 0.099392831325531, + "learning_rate": 2.9434469831810765e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4000 + }, + { + "epoch": 1.1667151585685192, + "flow_matching_loss": 0.016741718631237745, + "grad_norm": 0.08121660351753235, + "learning_rate": 2.9429963668333566e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4010 + }, + { + "epoch": 1.1696246726796624, + "flow_matching_loss": 0.018954686541110276, + "grad_norm": 0.1269601583480835, + "learning_rate": 2.9425439971089845e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4020 + }, + { + "epoch": 1.172534186790806, + "flow_matching_loss": 0.01728836726397276, + "grad_norm": 0.08181599527597427, + "learning_rate": 2.9420898745576336e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4030 + }, + { + "epoch": 1.1754437009019494, + "flow_matching_loss": 0.01811245149001479, + "grad_norm": 0.11778513342142105, + "learning_rate": 2.9416339997311044e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4040 + }, + { + "epoch": 1.1783532150130929, + "flow_matching_loss": 0.01911936802789569, + "grad_norm": 0.11700204014778137, + "learning_rate": 2.941176373183329e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4050 + }, + { + "epoch": 1.1812627291242364, + "flow_matching_loss": 0.018887192383408545, + "grad_norm": 0.10368045419454575, + "learning_rate": 2.9407169954703653e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4060 + }, + { + "epoch": 1.1841722432353796, + "flow_matching_loss": 0.017710230313241482, + "grad_norm": 0.08288384974002838, + "learning_rate": 2.9402558671504018e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4070 + }, + { + "epoch": 1.187081757346523, + "flow_matching_loss": 0.019235361274331807, + "grad_norm": 0.08389434963464737, + "learning_rate": 2.9397929887837524e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4080 + }, + { + "epoch": 1.1899912714576666, + "flow_matching_loss": 0.0179730080999434, + "grad_norm": 0.09015706926584244, + "learning_rate": 2.939328360932858e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4090 + }, + { + "epoch": 1.19290078556881, + "flow_matching_loss": 0.01913061076775193, + "grad_norm": 0.09610650688409805, + "learning_rate": 2.938861984162285e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4100 + }, + { + "epoch": 1.1958102996799536, + "flow_matching_loss": 0.016717475187033416, + "grad_norm": 0.08806820958852768, + "learning_rate": 2.938393859038726e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4110 + }, + { + "epoch": 1.1987198137910968, + "flow_matching_loss": 0.017702670209109782, + "grad_norm": 0.10046123713254929, + "learning_rate": 2.9379239861309964e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4120 + }, + { + "epoch": 1.2016293279022403, + "flow_matching_loss": 0.017854461260139943, + "grad_norm": 0.06780946999788284, + "learning_rate": 2.937452366010036e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4130 + }, + { + "epoch": 1.2045388420133838, + "flow_matching_loss": 0.01771745551377535, + "grad_norm": 0.10280607640743256, + "learning_rate": 2.936978999248908e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4140 + }, + { + "epoch": 1.2074483561245273, + "flow_matching_loss": 0.018214298877865076, + "grad_norm": 0.10675376653671265, + "learning_rate": 2.936503886422798e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4150 + }, + { + "epoch": 1.2103578702356708, + "flow_matching_loss": 0.017264979146420955, + "grad_norm": 0.09889914840459824, + "learning_rate": 2.9360270281090122e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4160 + }, + { + "epoch": 1.213267384346814, + "flow_matching_loss": 0.018937842547893526, + "grad_norm": 0.08624409884214401, + "learning_rate": 2.935548424886979e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4170 + }, + { + "epoch": 1.2161768984579575, + "flow_matching_loss": 0.018174263648688793, + "grad_norm": 0.12131497263908386, + "learning_rate": 2.935068077338246e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4180 + }, + { + "epoch": 1.219086412569101, + "flow_matching_loss": 0.017099984642118215, + "grad_norm": 0.0826491042971611, + "learning_rate": 2.934585986046481e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4190 + }, + { + "epoch": 1.2219959266802445, + "flow_matching_loss": 0.01864868113771081, + "grad_norm": 0.09836935997009277, + "learning_rate": 2.93410215159747e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4200 + }, + { + "epoch": 1.2249054407913877, + "flow_matching_loss": 0.017372097447514535, + "grad_norm": 0.08062241971492767, + "learning_rate": 2.933616574579118e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4210 + }, + { + "epoch": 1.2278149549025312, + "flow_matching_loss": 0.01775721935555339, + "grad_norm": 0.11370919644832611, + "learning_rate": 2.933129255581447e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4220 + }, + { + "epoch": 1.2307244690136747, + "flow_matching_loss": 0.020140708703547717, + "grad_norm": 0.07857515662908554, + "learning_rate": 2.9326401951965954e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4230 + }, + { + "epoch": 1.2336339831248182, + "flow_matching_loss": 0.01743037225678563, + "grad_norm": 0.10387536138296127, + "learning_rate": 2.9321493940188165e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4240 + }, + { + "epoch": 1.2365434972359617, + "flow_matching_loss": 0.018371493183076383, + "grad_norm": 0.09044521301984787, + "learning_rate": 2.931656852644482e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4250 + }, + { + "epoch": 1.239453011347105, + "flow_matching_loss": 0.017629220243543388, + "grad_norm": 0.09736171364784241, + "learning_rate": 2.9311625716720748e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4260 + }, + { + "epoch": 1.2423625254582484, + "flow_matching_loss": 0.018919091019779444, + "grad_norm": 0.09058050066232681, + "learning_rate": 2.9306665517021927e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4270 + }, + { + "epoch": 1.245272039569392, + "flow_matching_loss": 0.017430911026895046, + "grad_norm": 0.09812258183956146, + "learning_rate": 2.9301687933375483e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4280 + }, + { + "epoch": 1.2481815536805354, + "flow_matching_loss": 0.020585005823522805, + "grad_norm": 0.08601369708776474, + "learning_rate": 2.929669297182963e-05, + "loss": 0.0206, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4290 + }, + { + "epoch": 1.2510910677916787, + "flow_matching_loss": 0.018478300794959067, + "grad_norm": 0.09330283105373383, + "learning_rate": 2.929168063845373e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4300 + }, + { + "epoch": 1.2540005819028222, + "flow_matching_loss": 0.01849617874249816, + "grad_norm": 0.08385040611028671, + "learning_rate": 2.9286650939338236e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4310 + }, + { + "epoch": 1.2569100960139656, + "flow_matching_loss": 0.01822030507028103, + "grad_norm": 0.09959220886230469, + "learning_rate": 2.9281603880594706e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4320 + }, + { + "epoch": 1.2598196101251091, + "flow_matching_loss": 0.01901859128847718, + "grad_norm": 0.112197645008564, + "learning_rate": 2.9276539468355796e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4330 + }, + { + "epoch": 1.2627291242362526, + "flow_matching_loss": 0.019208294339478016, + "grad_norm": 0.1383962780237198, + "learning_rate": 2.9271457708775244e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4340 + }, + { + "epoch": 1.2656386383473959, + "flow_matching_loss": 0.016443994734436272, + "grad_norm": 0.09864553064107895, + "learning_rate": 2.9266358608027863e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4350 + }, + { + "epoch": 1.2685481524585394, + "flow_matching_loss": 0.019102632999420166, + "grad_norm": 0.09941279143095016, + "learning_rate": 2.9261242172309543e-05, + "loss": 0.0191, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4360 + }, + { + "epoch": 1.2714576665696828, + "flow_matching_loss": 0.017583046108484268, + "grad_norm": 0.08766337484121323, + "learning_rate": 2.925610840783723e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4370 + }, + { + "epoch": 1.2743671806808263, + "flow_matching_loss": 0.01670179730281234, + "grad_norm": 0.082002654671669, + "learning_rate": 2.9250957320848946e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4380 + }, + { + "epoch": 1.2772766947919698, + "flow_matching_loss": 0.01858357768505812, + "grad_norm": 0.1014467403292656, + "learning_rate": 2.924578891760374e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4390 + }, + { + "epoch": 1.280186208903113, + "flow_matching_loss": 0.017586917523294686, + "grad_norm": 0.09436162561178207, + "learning_rate": 2.924060320438171e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4400 + }, + { + "epoch": 1.2830957230142566, + "flow_matching_loss": 0.018257829174399377, + "grad_norm": 0.12024789303541183, + "learning_rate": 2.9235400187483978e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4410 + }, + { + "epoch": 1.2860052371254, + "flow_matching_loss": 0.018954668752849103, + "grad_norm": 0.08495334535837173, + "learning_rate": 2.9230179873232716e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4420 + }, + { + "epoch": 1.2889147512365435, + "flow_matching_loss": 0.017773597594350578, + "grad_norm": 0.11191771924495697, + "learning_rate": 2.922494226797108e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4430 + }, + { + "epoch": 1.291824265347687, + "flow_matching_loss": 0.018844246584922075, + "grad_norm": 0.09031171351671219, + "learning_rate": 2.9219687378063275e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4440 + }, + { + "epoch": 1.2947337794588303, + "flow_matching_loss": 0.018993313051760198, + "grad_norm": 0.11572220176458359, + "learning_rate": 2.921441520989447e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4450 + }, + { + "epoch": 1.2976432935699738, + "flow_matching_loss": 0.018619462847709656, + "grad_norm": 0.11062991619110107, + "learning_rate": 2.9209125769870854e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4460 + }, + { + "epoch": 1.3005528076811173, + "flow_matching_loss": 0.020284060575067996, + "grad_norm": 0.10694470256567001, + "learning_rate": 2.92038190644196e-05, + "loss": 0.0203, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4470 + }, + { + "epoch": 1.3034623217922607, + "flow_matching_loss": 0.019398790411651136, + "grad_norm": 0.12097244709730148, + "learning_rate": 2.9198495099988843e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4480 + }, + { + "epoch": 1.3063718359034042, + "flow_matching_loss": 0.018007945362478496, + "grad_norm": 0.09315640479326248, + "learning_rate": 2.9193153883047712e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4490 + }, + { + "epoch": 1.3092813500145475, + "flow_matching_loss": 0.019243950210511684, + "grad_norm": 0.09778282791376114, + "learning_rate": 2.9187795420086287e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4500 + }, + { + "epoch": 1.312190864125691, + "flow_matching_loss": 0.020176839735358953, + "grad_norm": 0.10870636254549026, + "learning_rate": 2.9182419717615602e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4510 + }, + { + "epoch": 1.3151003782368345, + "flow_matching_loss": 0.019200797751545907, + "grad_norm": 0.12208949029445648, + "learning_rate": 2.917702678216765e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4520 + }, + { + "epoch": 1.318009892347978, + "flow_matching_loss": 0.017416909337043762, + "grad_norm": 0.10028563439846039, + "learning_rate": 2.9171616620295347e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4530 + }, + { + "epoch": 1.3209194064591214, + "flow_matching_loss": 0.01902217846363783, + "grad_norm": 0.10554487258195877, + "learning_rate": 2.9166189238572554e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4540 + }, + { + "epoch": 1.3238289205702647, + "flow_matching_loss": 0.017868200410157443, + "grad_norm": 0.08479122817516327, + "learning_rate": 2.9160744643594052e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4550 + }, + { + "epoch": 1.3267384346814082, + "flow_matching_loss": 0.01774563081562519, + "grad_norm": 0.08744746446609497, + "learning_rate": 2.915528284197554e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4560 + }, + { + "epoch": 1.3296479487925517, + "flow_matching_loss": 0.017609895672649145, + "grad_norm": 0.07875224947929382, + "learning_rate": 2.914980384035362e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4570 + }, + { + "epoch": 1.3325574629036951, + "flow_matching_loss": 0.018285086378455162, + "grad_norm": 0.07618343830108643, + "learning_rate": 2.9144307645385787e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4580 + }, + { + "epoch": 1.3354669770148386, + "flow_matching_loss": 0.017639012169092895, + "grad_norm": 0.10082172602415085, + "learning_rate": 2.9138794263750454e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4590 + }, + { + "epoch": 1.338376491125982, + "flow_matching_loss": 0.018833895865827798, + "grad_norm": 0.10035180300474167, + "learning_rate": 2.913326370214688e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4600 + }, + { + "epoch": 1.3412860052371254, + "flow_matching_loss": 0.018262040615081788, + "grad_norm": 0.09943670779466629, + "learning_rate": 2.9127715967295234e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4610 + }, + { + "epoch": 1.3441955193482689, + "flow_matching_loss": 0.018121830467134715, + "grad_norm": 0.09781814366579056, + "learning_rate": 2.912215106593653e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4620 + }, + { + "epoch": 1.3471050334594123, + "flow_matching_loss": 0.018252192065119743, + "grad_norm": 0.0686846673488617, + "learning_rate": 2.911656900483265e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4630 + }, + { + "epoch": 1.3500145475705558, + "flow_matching_loss": 0.018967512343078852, + "grad_norm": 0.08574791252613068, + "learning_rate": 2.9110969790766322e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4640 + }, + { + "epoch": 1.352924061681699, + "flow_matching_loss": 0.01814276073127985, + "grad_norm": 0.08063971251249313, + "learning_rate": 2.910535343054112e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4650 + }, + { + "epoch": 1.3558335757928426, + "flow_matching_loss": 0.017324884422123432, + "grad_norm": 0.11089887470006943, + "learning_rate": 2.909971993098145e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4660 + }, + { + "epoch": 1.358743089903986, + "flow_matching_loss": 0.019982997607439757, + "grad_norm": 0.11462248116731644, + "learning_rate": 2.9094069298932552e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4670 + }, + { + "epoch": 1.3616526040151296, + "flow_matching_loss": 0.01900808960199356, + "grad_norm": 0.10775087773799896, + "learning_rate": 2.908840154126047e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4680 + }, + { + "epoch": 1.364562118126273, + "flow_matching_loss": 0.015799060463905334, + "grad_norm": 0.06244603917002678, + "learning_rate": 2.9082716664852063e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4690 + }, + { + "epoch": 1.3674716322374163, + "flow_matching_loss": 0.01877765329554677, + "grad_norm": 0.08301108330488205, + "learning_rate": 2.9077014676615e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4700 + }, + { + "epoch": 1.3703811463485598, + "flow_matching_loss": 0.01720654368400574, + "grad_norm": 0.07409752160310745, + "learning_rate": 2.9071295583477724e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4710 + }, + { + "epoch": 1.3732906604597033, + "flow_matching_loss": 0.016734423581510782, + "grad_norm": 0.0924043208360672, + "learning_rate": 2.906555939238948e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4720 + }, + { + "epoch": 1.3762001745708465, + "flow_matching_loss": 0.0182792947627604, + "grad_norm": 0.07868342846632004, + "learning_rate": 2.905980611032028e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4730 + }, + { + "epoch": 1.3791096886819902, + "flow_matching_loss": 0.018709181156009436, + "grad_norm": 0.09808248281478882, + "learning_rate": 2.9054035744260904e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4740 + }, + { + "epoch": 1.3820192027931335, + "flow_matching_loss": 0.016925296280533075, + "grad_norm": 0.08910417556762695, + "learning_rate": 2.9048248301222885e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4750 + }, + { + "epoch": 1.384928716904277, + "flow_matching_loss": 0.018614665232598783, + "grad_norm": 0.0961250439286232, + "learning_rate": 2.904244378823852e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4760 + }, + { + "epoch": 1.3878382310154205, + "flow_matching_loss": 0.01651146337389946, + "grad_norm": 0.11450298130512238, + "learning_rate": 2.9036622212360836e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4770 + }, + { + "epoch": 1.3907477451265637, + "flow_matching_loss": 0.01732441335916519, + "grad_norm": 0.08941829204559326, + "learning_rate": 2.9030783580663596e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4780 + }, + { + "epoch": 1.3936572592377072, + "flow_matching_loss": 0.017447572853416204, + "grad_norm": 0.12177257239818573, + "learning_rate": 2.902492790024129e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4790 + }, + { + "epoch": 1.3965667733488507, + "flow_matching_loss": 0.018938645999878646, + "grad_norm": 0.09724850207567215, + "learning_rate": 2.901905517820912e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4800 + }, + { + "epoch": 1.3994762874599942, + "flow_matching_loss": 0.017023267317563297, + "grad_norm": 0.08755195885896683, + "learning_rate": 2.9013165421703e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4810 + }, + { + "epoch": 1.4023858015711377, + "flow_matching_loss": 0.018979556020349266, + "grad_norm": 0.10531023144721985, + "learning_rate": 2.9007258637879534e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4820 + }, + { + "epoch": 1.405295315682281, + "flow_matching_loss": 0.017148247826844452, + "grad_norm": 0.06516321748495102, + "learning_rate": 2.9001334833916025e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4830 + }, + { + "epoch": 1.4082048297934244, + "flow_matching_loss": 0.01702414993196726, + "grad_norm": 0.08933217823505402, + "learning_rate": 2.899539401701046e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4840 + }, + { + "epoch": 1.411114343904568, + "flow_matching_loss": 0.017573399282991887, + "grad_norm": 0.11892300844192505, + "learning_rate": 2.8989436194381474e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4850 + }, + { + "epoch": 1.4140238580157114, + "flow_matching_loss": 0.018019800074398518, + "grad_norm": 0.09751694649457932, + "learning_rate": 2.8983461373268404e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4860 + }, + { + "epoch": 1.4169333721268549, + "flow_matching_loss": 0.018052796833217144, + "grad_norm": 0.1145012304186821, + "learning_rate": 2.8977469560931205e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4870 + }, + { + "epoch": 1.4198428862379981, + "flow_matching_loss": 0.017197118792682885, + "grad_norm": 0.1093859001994133, + "learning_rate": 2.8971460764650502e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4880 + }, + { + "epoch": 1.4227524003491416, + "flow_matching_loss": 0.018249510508030652, + "grad_norm": 0.0934465304017067, + "learning_rate": 2.8965434991727545e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4890 + }, + { + "epoch": 1.4256619144602851, + "flow_matching_loss": 0.01767230164259672, + "grad_norm": 0.10075203329324722, + "learning_rate": 2.895939224948422e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4900 + }, + { + "epoch": 1.4285714285714286, + "flow_matching_loss": 0.015482633654028178, + "grad_norm": 0.072540782392025, + "learning_rate": 2.895333254526303e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4910 + }, + { + "epoch": 1.431480942682572, + "flow_matching_loss": 0.017702196072787047, + "grad_norm": 0.09651000797748566, + "learning_rate": 2.8947255886427084e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4920 + }, + { + "epoch": 1.4343904567937154, + "flow_matching_loss": 0.01827510492876172, + "grad_norm": 0.09040627628564835, + "learning_rate": 2.8941162280360095e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4930 + }, + { + "epoch": 1.4372999709048588, + "flow_matching_loss": 0.018604743015021085, + "grad_norm": 0.07292753458023071, + "learning_rate": 2.8935051734466368e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4940 + }, + { + "epoch": 1.4402094850160023, + "flow_matching_loss": 0.018677146080881356, + "grad_norm": 0.08237838000059128, + "learning_rate": 2.8928924256170792e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4950 + }, + { + "epoch": 1.4431189991271458, + "flow_matching_loss": 0.018619131948798894, + "grad_norm": 0.08702763170003891, + "learning_rate": 2.8922779852918836e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4960 + }, + { + "epoch": 1.4460285132382893, + "flow_matching_loss": 0.019457097351551055, + "grad_norm": 0.09463224560022354, + "learning_rate": 2.8916618532176526e-05, + "loss": 0.0195, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4970 + }, + { + "epoch": 1.4489380273494326, + "flow_matching_loss": 0.017149927280843257, + "grad_norm": 0.11838459223508835, + "learning_rate": 2.891044030143045e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4980 + }, + { + "epoch": 1.451847541460576, + "flow_matching_loss": 0.01655133254826069, + "grad_norm": 0.09493520855903625, + "learning_rate": 2.890424516818774e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 4990 + }, + { + "epoch": 1.4547570555717195, + "flow_matching_loss": 0.01725783934816718, + "grad_norm": 0.073824942111969, + "learning_rate": 2.8898033139976057e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5000 + }, + { + "epoch": 1.457666569682863, + "flow_matching_loss": 0.01747996937483549, + "grad_norm": 0.11005017161369324, + "learning_rate": 2.889180422434362e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5010 + }, + { + "epoch": 1.4605760837940065, + "flow_matching_loss": 0.016852688137441875, + "grad_norm": 0.08344901353120804, + "learning_rate": 2.8885558428859136e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5020 + }, + { + "epoch": 1.4634855979051498, + "flow_matching_loss": 0.016580923087894918, + "grad_norm": 0.08651362359523773, + "learning_rate": 2.8879295761111835e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5030 + }, + { + "epoch": 1.4663951120162932, + "flow_matching_loss": 0.018782219849526882, + "grad_norm": 0.10748005658388138, + "learning_rate": 2.8873016228711452e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5040 + }, + { + "epoch": 1.4693046261274367, + "flow_matching_loss": 0.01883310228586197, + "grad_norm": 0.13379813730716705, + "learning_rate": 2.8866719839288215e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5050 + }, + { + "epoch": 1.4722141402385802, + "flow_matching_loss": 0.017361748311668635, + "grad_norm": 0.0940941795706749, + "learning_rate": 2.886040660049283e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5060 + }, + { + "epoch": 1.4751236543497237, + "flow_matching_loss": 0.016431981418281793, + "grad_norm": 0.08713187277317047, + "learning_rate": 2.8854076519996467e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5070 + }, + { + "epoch": 1.478033168460867, + "flow_matching_loss": 0.015611420106142759, + "grad_norm": 0.09247031807899475, + "learning_rate": 2.884772960549078e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5080 + }, + { + "epoch": 1.4809426825720104, + "flow_matching_loss": 0.019041858054697512, + "grad_norm": 0.09477276355028152, + "learning_rate": 2.8841365864687867e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5090 + }, + { + "epoch": 1.483852196683154, + "flow_matching_loss": 0.016526197269558906, + "grad_norm": 0.08528783917427063, + "learning_rate": 2.8834985305320276e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5100 + }, + { + "epoch": 1.4867617107942974, + "flow_matching_loss": 0.01690773954614997, + "grad_norm": 0.07359456270933151, + "learning_rate": 2.882858793514098e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5110 + }, + { + "epoch": 1.489671224905441, + "flow_matching_loss": 0.017451133392751218, + "grad_norm": 0.10380527377128601, + "learning_rate": 2.8822173761923398e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5120 + }, + { + "epoch": 1.4925807390165842, + "flow_matching_loss": 0.016315473522990943, + "grad_norm": 0.07851549983024597, + "learning_rate": 2.8815742793461343e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5130 + }, + { + "epoch": 1.4954902531277277, + "flow_matching_loss": 0.017056092247366907, + "grad_norm": 0.09480322152376175, + "learning_rate": 2.880929503756906e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5140 + }, + { + "epoch": 1.4983997672388711, + "flow_matching_loss": 0.017868447862565518, + "grad_norm": 0.11108627170324326, + "learning_rate": 2.880283050208117e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5150 + }, + { + "epoch": 1.5013092813500144, + "flow_matching_loss": 0.016785035375505686, + "grad_norm": 0.061757851392030716, + "learning_rate": 2.8796349194852704e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5160 + }, + { + "epoch": 1.504218795461158, + "flow_matching_loss": 0.017771530151367187, + "grad_norm": 0.10262252390384674, + "learning_rate": 2.8789851123759052e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5170 + }, + { + "epoch": 1.5071283095723014, + "flow_matching_loss": 0.01854334119707346, + "grad_norm": 0.10497070103883743, + "learning_rate": 2.8783336296695985e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5180 + }, + { + "epoch": 1.5100378236834449, + "flow_matching_loss": 0.016487014200538396, + "grad_norm": 0.08160553127527237, + "learning_rate": 2.877680472157964e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5190 + }, + { + "epoch": 1.5129473377945883, + "flow_matching_loss": 0.019391296431422234, + "grad_norm": 0.08666663616895676, + "learning_rate": 2.8770256406346483e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5200 + }, + { + "epoch": 1.5158568519057316, + "flow_matching_loss": 0.01724807731807232, + "grad_norm": 0.08073283731937408, + "learning_rate": 2.8763691358953348e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5210 + }, + { + "epoch": 1.5187663660168753, + "flow_matching_loss": 0.016551259718835355, + "grad_norm": 0.08751153945922852, + "learning_rate": 2.8757109587377377e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5220 + }, + { + "epoch": 1.5216758801280186, + "flow_matching_loss": 0.016597395483404397, + "grad_norm": 0.08179182559251785, + "learning_rate": 2.8750511099616047e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5230 + }, + { + "epoch": 1.524585394239162, + "flow_matching_loss": 0.01719669522717595, + "grad_norm": 0.09003862738609314, + "learning_rate": 2.8743895903687143e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5240 + }, + { + "epoch": 1.5274949083503055, + "flow_matching_loss": 0.019365734606981277, + "grad_norm": 0.10881456732749939, + "learning_rate": 2.8737264007628753e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5250 + }, + { + "epoch": 1.5304044224614488, + "flow_matching_loss": 0.01594038922339678, + "grad_norm": 0.10556256026029587, + "learning_rate": 2.8730615419499247e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5260 + }, + { + "epoch": 1.5333139365725925, + "flow_matching_loss": 0.02015107562765479, + "grad_norm": 0.1527707874774933, + "learning_rate": 2.8723950147377303e-05, + "loss": 0.0202, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5270 + }, + { + "epoch": 1.5362234506837358, + "flow_matching_loss": 0.0180386065505445, + "grad_norm": 0.09705820679664612, + "learning_rate": 2.871726819936184e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5280 + }, + { + "epoch": 1.5391329647948793, + "flow_matching_loss": 0.01734231049194932, + "grad_norm": 0.09556610137224197, + "learning_rate": 2.8710569583572068e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5290 + }, + { + "epoch": 1.5420424789060228, + "flow_matching_loss": 0.016399398911744356, + "grad_norm": 0.06832445412874222, + "learning_rate": 2.870385430814742e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5300 + }, + { + "epoch": 1.544951993017166, + "flow_matching_loss": 0.017231032717972994, + "grad_norm": 0.10136380046606064, + "learning_rate": 2.8697122381247605e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5310 + }, + { + "epoch": 1.5478615071283097, + "flow_matching_loss": 0.0188065892085433, + "grad_norm": 0.09229336678981781, + "learning_rate": 2.8690373811052545e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5320 + }, + { + "epoch": 1.550771021239453, + "flow_matching_loss": 0.016497951280325652, + "grad_norm": 0.08012223988771439, + "learning_rate": 2.8683608605762378e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5330 + }, + { + "epoch": 1.5536805353505965, + "flow_matching_loss": 0.018582467548549177, + "grad_norm": 0.09397357702255249, + "learning_rate": 2.8676826773597485e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5340 + }, + { + "epoch": 1.55659004946174, + "flow_matching_loss": 0.01852222913876176, + "grad_norm": 0.08856892585754395, + "learning_rate": 2.8670028322798413e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5350 + }, + { + "epoch": 1.5594995635728832, + "flow_matching_loss": 0.017504793591797352, + "grad_norm": 0.14570890367031097, + "learning_rate": 2.8663213261625932e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5360 + }, + { + "epoch": 1.562409077684027, + "flow_matching_loss": 0.01857088729739189, + "grad_norm": 0.09754477441310883, + "learning_rate": 2.865638159836098e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5370 + }, + { + "epoch": 1.5653185917951702, + "flow_matching_loss": 0.017908022087067366, + "grad_norm": 0.08342677354812622, + "learning_rate": 2.8649533341304677e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5380 + }, + { + "epoch": 1.5682281059063137, + "flow_matching_loss": 0.015244993846863508, + "grad_norm": 0.07602348923683167, + "learning_rate": 2.8642668498778297e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5390 + }, + { + "epoch": 1.5711376200174572, + "flow_matching_loss": 0.017927828896790742, + "grad_norm": 0.08281898498535156, + "learning_rate": 2.8635787079123268e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5400 + }, + { + "epoch": 1.5740471341286004, + "flow_matching_loss": 0.01723905373364687, + "grad_norm": 0.08341525495052338, + "learning_rate": 2.8628889090701175e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5410 + }, + { + "epoch": 1.576956648239744, + "flow_matching_loss": 0.016299184784293173, + "grad_norm": 0.08228100836277008, + "learning_rate": 2.862197454189371e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5420 + }, + { + "epoch": 1.5798661623508874, + "flow_matching_loss": 0.01724052196368575, + "grad_norm": 0.08897308260202408, + "learning_rate": 2.8615043441102708e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5430 + }, + { + "epoch": 1.5827756764620309, + "flow_matching_loss": 0.01751705687493086, + "grad_norm": 0.0952899232506752, + "learning_rate": 2.860809579675011e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5440 + }, + { + "epoch": 1.5856851905731744, + "flow_matching_loss": 0.016873243171721695, + "grad_norm": 0.0782603994011879, + "learning_rate": 2.860113161727796e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5450 + }, + { + "epoch": 1.5885947046843176, + "flow_matching_loss": 0.016978295519948006, + "grad_norm": 0.09755745530128479, + "learning_rate": 2.8594150911148393e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5460 + }, + { + "epoch": 1.5915042187954611, + "flow_matching_loss": 0.018555331230163574, + "grad_norm": 0.10776392370462418, + "learning_rate": 2.8587153686843626e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5470 + }, + { + "epoch": 1.5944137329066046, + "flow_matching_loss": 0.016409764159470798, + "grad_norm": 0.0847768560051918, + "learning_rate": 2.858013995286594e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5480 + }, + { + "epoch": 1.597323247017748, + "flow_matching_loss": 0.01864912882447243, + "grad_norm": 0.10116804391145706, + "learning_rate": 2.857310971773769e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5490 + }, + { + "epoch": 1.6002327611288916, + "flow_matching_loss": 0.018319619167596102, + "grad_norm": 0.11757565289735794, + "learning_rate": 2.8566062990001274e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5500 + }, + { + "epoch": 1.6031422752400348, + "flow_matching_loss": 0.01778229447081685, + "grad_norm": 0.08383749425411224, + "learning_rate": 2.8558999778219126e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5510 + }, + { + "epoch": 1.6060517893511783, + "flow_matching_loss": 0.016651444137096405, + "grad_norm": 0.08387260138988495, + "learning_rate": 2.8551920090973722e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5520 + }, + { + "epoch": 1.6089613034623218, + "flow_matching_loss": 0.01813643369823694, + "grad_norm": 0.09596428275108337, + "learning_rate": 2.854482393686754e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5530 + }, + { + "epoch": 1.6118708175734653, + "flow_matching_loss": 0.018199178297072648, + "grad_norm": 0.09630517661571503, + "learning_rate": 2.8537711324523086e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5540 + }, + { + "epoch": 1.6147803316846088, + "flow_matching_loss": 0.01682445714250207, + "grad_norm": 0.08891288191080093, + "learning_rate": 2.8530582262582845e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5550 + }, + { + "epoch": 1.617689845795752, + "flow_matching_loss": 0.017872560303658247, + "grad_norm": 0.10974521934986115, + "learning_rate": 2.8523436759709306e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5560 + }, + { + "epoch": 1.6205993599068955, + "flow_matching_loss": 0.0185155033133924, + "grad_norm": 0.08491750806570053, + "learning_rate": 2.8516274824584926e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5570 + }, + { + "epoch": 1.623508874018039, + "flow_matching_loss": 0.015539500396698713, + "grad_norm": 0.08791665732860565, + "learning_rate": 2.8509096465912136e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5580 + }, + { + "epoch": 1.6264183881291823, + "flow_matching_loss": 0.01817246014252305, + "grad_norm": 0.06732159852981567, + "learning_rate": 2.8501901692413315e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5590 + }, + { + "epoch": 1.629327902240326, + "flow_matching_loss": 0.018525287695229053, + "grad_norm": 0.08299463987350464, + "learning_rate": 2.8494690512830788e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5600 + }, + { + "epoch": 1.6322374163514692, + "flow_matching_loss": 0.016658343933522703, + "grad_norm": 0.06430450826883316, + "learning_rate": 2.8487462935926823e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5610 + }, + { + "epoch": 1.6351469304626127, + "flow_matching_loss": 0.01916158040985465, + "grad_norm": 0.09220156073570251, + "learning_rate": 2.8480218970483606e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5620 + }, + { + "epoch": 1.6380564445737562, + "flow_matching_loss": 0.017783908639103174, + "grad_norm": 0.08560550212860107, + "learning_rate": 2.847295862530324e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5630 + }, + { + "epoch": 1.6409659586848995, + "flow_matching_loss": 0.015903191827237605, + "grad_norm": 0.08141791820526123, + "learning_rate": 2.846568190920773e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5640 + }, + { + "epoch": 1.6438754727960432, + "flow_matching_loss": 0.015182710252702236, + "grad_norm": 0.0815165713429451, + "learning_rate": 2.8458388831038965e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5650 + }, + { + "epoch": 1.6467849869071864, + "flow_matching_loss": 0.01641483334824443, + "grad_norm": 0.08154363930225372, + "learning_rate": 2.8451079399658725e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5660 + }, + { + "epoch": 1.64969450101833, + "flow_matching_loss": 0.017261054646223784, + "grad_norm": 0.12954142689704895, + "learning_rate": 2.8443753623948665e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5670 + }, + { + "epoch": 1.6526040151294734, + "flow_matching_loss": 0.02002245970070362, + "grad_norm": 0.12919364869594574, + "learning_rate": 2.8436411512810288e-05, + "loss": 0.02, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5680 + }, + { + "epoch": 1.6555135292406167, + "flow_matching_loss": 0.01850557392463088, + "grad_norm": 0.10711091756820679, + "learning_rate": 2.842905307516495e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5690 + }, + { + "epoch": 1.6584230433517604, + "flow_matching_loss": 0.018428776878863573, + "grad_norm": 0.09152252972126007, + "learning_rate": 2.8421678319953854e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5700 + }, + { + "epoch": 1.6613325574629036, + "flow_matching_loss": 0.01702547324821353, + "grad_norm": 0.08295119553804398, + "learning_rate": 2.841428725613801e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5710 + }, + { + "epoch": 1.6642420715740471, + "flow_matching_loss": 0.016469663009047508, + "grad_norm": 0.0860595852136612, + "learning_rate": 2.8406879892698272e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5720 + }, + { + "epoch": 1.6671515856851906, + "flow_matching_loss": 0.01720836302265525, + "grad_norm": 0.08857264369726181, + "learning_rate": 2.8399456238635274e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5730 + }, + { + "epoch": 1.6700610997963339, + "flow_matching_loss": 0.01650536237284541, + "grad_norm": 0.07622256875038147, + "learning_rate": 2.8392016302969457e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5740 + }, + { + "epoch": 1.6729706139074776, + "flow_matching_loss": 0.015823934320360424, + "grad_norm": 0.08929377794265747, + "learning_rate": 2.8384560094741055e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5750 + }, + { + "epoch": 1.6758801280186209, + "flow_matching_loss": 0.018386620935052632, + "grad_norm": 0.10116942971944809, + "learning_rate": 2.8377087623010054e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5760 + }, + { + "epoch": 1.6787896421297643, + "flow_matching_loss": 0.0196587678976357, + "grad_norm": 0.10320749133825302, + "learning_rate": 2.8369598896856213e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5770 + }, + { + "epoch": 1.6816991562409078, + "flow_matching_loss": 0.019258577562868594, + "grad_norm": 0.10766518861055374, + "learning_rate": 2.836209392537904e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5780 + }, + { + "epoch": 1.684608670352051, + "flow_matching_loss": 0.01773231811821461, + "grad_norm": 0.10599896311759949, + "learning_rate": 2.8354572717697787e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5790 + }, + { + "epoch": 1.6875181844631948, + "flow_matching_loss": 0.017586765624582768, + "grad_norm": 0.09834612160921097, + "learning_rate": 2.8347035282951428e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5800 + }, + { + "epoch": 1.690427698574338, + "flow_matching_loss": 0.02070730337873101, + "grad_norm": 0.1000761091709137, + "learning_rate": 2.8339481630298653e-05, + "loss": 0.0207, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5810 + }, + { + "epoch": 1.6933372126854815, + "flow_matching_loss": 0.019719135016202927, + "grad_norm": 0.10444348305463791, + "learning_rate": 2.833191176891787e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5820 + }, + { + "epoch": 1.696246726796625, + "flow_matching_loss": 0.017742310464382172, + "grad_norm": 0.08219499886035919, + "learning_rate": 2.832432570800717e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5830 + }, + { + "epoch": 1.6991562409077683, + "flow_matching_loss": 0.01590384729206562, + "grad_norm": 0.09063272923231125, + "learning_rate": 2.8316723456784332e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5840 + }, + { + "epoch": 1.702065755018912, + "flow_matching_loss": 0.01692941039800644, + "grad_norm": 0.09858737885951996, + "learning_rate": 2.8309105024486808e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5850 + }, + { + "epoch": 1.7049752691300553, + "flow_matching_loss": 0.017827079445123673, + "grad_norm": 0.08925267308950424, + "learning_rate": 2.830147042037172e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5860 + }, + { + "epoch": 1.7078847832411987, + "flow_matching_loss": 0.018322870042175053, + "grad_norm": 0.08450651913881302, + "learning_rate": 2.8293819653715816e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5870 + }, + { + "epoch": 1.7107942973523422, + "flow_matching_loss": 0.01777884578332305, + "grad_norm": 0.10889311879873276, + "learning_rate": 2.828615273381551e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5880 + }, + { + "epoch": 1.7137038114634855, + "flow_matching_loss": 0.01694237319752574, + "grad_norm": 0.1096169501543045, + "learning_rate": 2.827846966998683e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5890 + }, + { + "epoch": 1.7166133255746292, + "flow_matching_loss": 0.018072139471769333, + "grad_norm": 0.11075051128864288, + "learning_rate": 2.8270770471565422e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5900 + }, + { + "epoch": 1.7195228396857725, + "flow_matching_loss": 0.017480346094816922, + "grad_norm": 0.10236915946006775, + "learning_rate": 2.826305514790654e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5910 + }, + { + "epoch": 1.722432353796916, + "flow_matching_loss": 0.01925774961709976, + "grad_norm": 0.11663434654474258, + "learning_rate": 2.825532370838503e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5920 + }, + { + "epoch": 1.7253418679080594, + "flow_matching_loss": 0.01799608487635851, + "grad_norm": 0.09770410507917404, + "learning_rate": 2.824757616239532e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5930 + }, + { + "epoch": 1.7282513820192027, + "flow_matching_loss": 0.017739695403724907, + "grad_norm": 0.09844247996807098, + "learning_rate": 2.8239812519351404e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5940 + }, + { + "epoch": 1.7311608961303462, + "flow_matching_loss": 0.018396802712231874, + "grad_norm": 0.09365978837013245, + "learning_rate": 2.823203278868685e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5950 + }, + { + "epoch": 1.7340704102414897, + "flow_matching_loss": 0.018304255045950412, + "grad_norm": 0.08987553417682648, + "learning_rate": 2.8224236979854752e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5960 + }, + { + "epoch": 1.7369799243526332, + "flow_matching_loss": 0.020427318289875984, + "grad_norm": 0.0771869570016861, + "learning_rate": 2.821642510232776e-05, + "loss": 0.0204, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5970 + }, + { + "epoch": 1.7398894384637766, + "flow_matching_loss": 0.01974338321015239, + "grad_norm": 0.09289658814668655, + "learning_rate": 2.8208597165598046e-05, + "loss": 0.0197, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5980 + }, + { + "epoch": 1.74279895257492, + "flow_matching_loss": 0.018540690187364817, + "grad_norm": 0.11165712773799896, + "learning_rate": 2.8200753179177285e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 5990 + }, + { + "epoch": 1.7457084666860634, + "flow_matching_loss": 0.017818270716816187, + "grad_norm": 0.11257367581129074, + "learning_rate": 2.819289315259666e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6000 + }, + { + "epoch": 1.7486179807972069, + "flow_matching_loss": 0.019261682871729136, + "grad_norm": 0.10651533305644989, + "learning_rate": 2.8185017095406845e-05, + "loss": 0.0193, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6010 + }, + { + "epoch": 1.7515274949083504, + "flow_matching_loss": 0.016773654147982597, + "grad_norm": 0.06700421124696732, + "learning_rate": 2.8177125017177998e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6020 + }, + { + "epoch": 1.7544370090194938, + "flow_matching_loss": 0.016513659060001372, + "grad_norm": 0.07833678275346756, + "learning_rate": 2.8169216927499727e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6030 + }, + { + "epoch": 1.757346523130637, + "flow_matching_loss": 0.016677586641162635, + "grad_norm": 0.10146848857402802, + "learning_rate": 2.816129283598112e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6040 + }, + { + "epoch": 1.7602560372417806, + "flow_matching_loss": 0.01751871071755886, + "grad_norm": 0.07822853326797485, + "learning_rate": 2.815335275225068e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6050 + }, + { + "epoch": 1.763165551352924, + "flow_matching_loss": 0.017435315251350402, + "grad_norm": 0.092750184237957, + "learning_rate": 2.8145396685956374e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6060 + }, + { + "epoch": 1.7660750654640673, + "flow_matching_loss": 0.017700549215078354, + "grad_norm": 0.09876800328493118, + "learning_rate": 2.813742464676556e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6070 + }, + { + "epoch": 1.768984579575211, + "flow_matching_loss": 0.016948574781417848, + "grad_norm": 0.1106359213590622, + "learning_rate": 2.8129436644365025e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6080 + }, + { + "epoch": 1.7718940936863543, + "flow_matching_loss": 0.01704831700772047, + "grad_norm": 0.07591833919286728, + "learning_rate": 2.8121432688460938e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6090 + }, + { + "epoch": 1.7748036077974978, + "flow_matching_loss": 0.018344786297529937, + "grad_norm": 0.09355084598064423, + "learning_rate": 2.8113412788778867e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6100 + }, + { + "epoch": 1.7777131219086413, + "flow_matching_loss": 0.017293549701571466, + "grad_norm": 0.07106456905603409, + "learning_rate": 2.8105376955063744e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6110 + }, + { + "epoch": 1.7806226360197845, + "flow_matching_loss": 0.017689632810652257, + "grad_norm": 0.07953126728534698, + "learning_rate": 2.8097325197079862e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6120 + }, + { + "epoch": 1.7835321501309283, + "flow_matching_loss": 0.016422489657998085, + "grad_norm": 0.085886649787426, + "learning_rate": 2.808925752461087e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6130 + }, + { + "epoch": 1.7864416642420715, + "flow_matching_loss": 0.017259439267218114, + "grad_norm": 0.07595214247703552, + "learning_rate": 2.808117394745975e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6140 + }, + { + "epoch": 1.789351178353215, + "flow_matching_loss": 0.017657240107655525, + "grad_norm": 0.075985386967659, + "learning_rate": 2.807307447544881e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6150 + }, + { + "epoch": 1.7922606924643585, + "flow_matching_loss": 0.017636528704315424, + "grad_norm": 0.1062876433134079, + "learning_rate": 2.806495911841967e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6160 + }, + { + "epoch": 1.7951702065755017, + "flow_matching_loss": 0.017750392854213714, + "grad_norm": 0.07909415662288666, + "learning_rate": 2.8056827886233255e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6170 + }, + { + "epoch": 1.7980797206866455, + "flow_matching_loss": 0.01794065209105611, + "grad_norm": 0.0911792740225792, + "learning_rate": 2.8048680788769783e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6180 + }, + { + "epoch": 1.8009892347977887, + "flow_matching_loss": 0.016185302566736937, + "grad_norm": 0.08202730864286423, + "learning_rate": 2.8040517835928742e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6190 + }, + { + "epoch": 1.8038987489089322, + "flow_matching_loss": 0.016134064085781573, + "grad_norm": 0.08715477585792542, + "learning_rate": 2.803233903762889e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6200 + }, + { + "epoch": 1.8068082630200757, + "flow_matching_loss": 0.015097574330866337, + "grad_norm": 0.09246494621038437, + "learning_rate": 2.8024144403808235e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6210 + }, + { + "epoch": 1.809717777131219, + "flow_matching_loss": 0.017987956013530492, + "grad_norm": 0.07452069222927094, + "learning_rate": 2.8015933944424037e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6220 + }, + { + "epoch": 1.8126272912423627, + "flow_matching_loss": 0.016245094127953053, + "grad_norm": 0.07510294765233994, + "learning_rate": 2.800770766945277e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6230 + }, + { + "epoch": 1.815536805353506, + "flow_matching_loss": 0.017236716393381358, + "grad_norm": 0.08328904956579208, + "learning_rate": 2.7999465588890135e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6240 + }, + { + "epoch": 1.8184463194646494, + "flow_matching_loss": 0.017453235387802125, + "grad_norm": 0.06981257349252701, + "learning_rate": 2.799120771275105e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6250 + }, + { + "epoch": 1.821355833575793, + "flow_matching_loss": 0.01704114191234112, + "grad_norm": 0.09467387199401855, + "learning_rate": 2.7982934051069597e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6260 + }, + { + "epoch": 1.8242653476869362, + "flow_matching_loss": 0.017012202739715578, + "grad_norm": 0.06669198721647263, + "learning_rate": 2.7974644613899063e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6270 + }, + { + "epoch": 1.8271748617980799, + "flow_matching_loss": 0.017525253165513278, + "grad_norm": 0.11544971168041229, + "learning_rate": 2.796633941131189e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6280 + }, + { + "epoch": 1.8300843759092231, + "flow_matching_loss": 0.017810753732919692, + "grad_norm": 0.10231969505548477, + "learning_rate": 2.795801845339969e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6290 + }, + { + "epoch": 1.8329938900203666, + "flow_matching_loss": 0.016508394572883844, + "grad_norm": 0.09667836874723434, + "learning_rate": 2.7949681750273208e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6300 + }, + { + "epoch": 1.83590340413151, + "flow_matching_loss": 0.015658417157828806, + "grad_norm": 0.09886516630649567, + "learning_rate": 2.7941329312062322e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6310 + }, + { + "epoch": 1.8388129182426534, + "flow_matching_loss": 0.017808574717491865, + "grad_norm": 0.09121502935886383, + "learning_rate": 2.7932961148916037e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6320 + }, + { + "epoch": 1.841722432353797, + "flow_matching_loss": 0.015661254711449146, + "grad_norm": 0.06457483023405075, + "learning_rate": 2.7924577271002456e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6330 + }, + { + "epoch": 1.8446319464649403, + "flow_matching_loss": 0.016410801466554402, + "grad_norm": 0.06225297227501869, + "learning_rate": 2.7916177688508782e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6340 + }, + { + "epoch": 1.8475414605760838, + "flow_matching_loss": 0.017373467423021795, + "grad_norm": 0.09945845603942871, + "learning_rate": 2.7907762411641302e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6350 + }, + { + "epoch": 1.8504509746872273, + "flow_matching_loss": 0.018526183813810347, + "grad_norm": 0.09676595777273178, + "learning_rate": 2.7899331450625367e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6360 + }, + { + "epoch": 1.8533604887983706, + "flow_matching_loss": 0.018018057476729153, + "grad_norm": 0.0964801236987114, + "learning_rate": 2.7890884815705397e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6370 + }, + { + "epoch": 1.8562700029095143, + "flow_matching_loss": 0.018945401720702647, + "grad_norm": 0.10283821821212769, + "learning_rate": 2.788242251714484e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6380 + }, + { + "epoch": 1.8591795170206575, + "flow_matching_loss": 0.017498037591576577, + "grad_norm": 0.10304134339094162, + "learning_rate": 2.787394456522619e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6390 + }, + { + "epoch": 1.862089031131801, + "flow_matching_loss": 0.01920912703499198, + "grad_norm": 0.09259947389364243, + "learning_rate": 2.7865450970250965e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6400 + }, + { + "epoch": 1.8649985452429445, + "flow_matching_loss": 0.01681995466351509, + "grad_norm": 0.10278018563985825, + "learning_rate": 2.7856941742539676e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6410 + }, + { + "epoch": 1.8679080593540878, + "flow_matching_loss": 0.016474193334579466, + "grad_norm": 0.10610830038785934, + "learning_rate": 2.784841689243184e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6420 + }, + { + "epoch": 1.8708175734652313, + "flow_matching_loss": 0.018024241365492345, + "grad_norm": 0.07568792253732681, + "learning_rate": 2.783987643028595e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6430 + }, + { + "epoch": 1.8737270875763747, + "flow_matching_loss": 0.017102495208382605, + "grad_norm": 0.09471683204174042, + "learning_rate": 2.783132036647948e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6440 + }, + { + "epoch": 1.8766366016875182, + "flow_matching_loss": 0.01784184416756034, + "grad_norm": 0.13135147094726562, + "learning_rate": 2.7822748711408852e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6450 + }, + { + "epoch": 1.8795461157986617, + "flow_matching_loss": 0.015856162924319506, + "grad_norm": 0.09402813017368317, + "learning_rate": 2.7814161475489428e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6460 + }, + { + "epoch": 1.882455629909805, + "flow_matching_loss": 0.01737728314474225, + "grad_norm": 0.07409776002168655, + "learning_rate": 2.780555866915552e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6470 + }, + { + "epoch": 1.8853651440209485, + "flow_matching_loss": 0.018084917776286603, + "grad_norm": 0.08197813481092453, + "learning_rate": 2.7796940302860338e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6480 + }, + { + "epoch": 1.888274658132092, + "flow_matching_loss": 0.016703634150326253, + "grad_norm": 0.07926367968320847, + "learning_rate": 2.7788306387076018e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6490 + }, + { + "epoch": 1.8911841722432354, + "flow_matching_loss": 0.017703006975352764, + "grad_norm": 0.09322194755077362, + "learning_rate": 2.7779656932293584e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6500 + }, + { + "epoch": 1.894093686354379, + "flow_matching_loss": 0.01739504747092724, + "grad_norm": 0.08049934357404709, + "learning_rate": 2.777099194902293e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6510 + }, + { + "epoch": 1.8970032004655222, + "flow_matching_loss": 0.016165436152368783, + "grad_norm": 0.09947182238101959, + "learning_rate": 2.776231144779283e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6520 + }, + { + "epoch": 1.8999127145766657, + "flow_matching_loss": 0.017088018357753754, + "grad_norm": 0.09252896159887314, + "learning_rate": 2.775361543915091e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6530 + }, + { + "epoch": 1.9028222286878091, + "flow_matching_loss": 0.017044087499380113, + "grad_norm": 0.09347112476825714, + "learning_rate": 2.774490393366365e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6540 + }, + { + "epoch": 1.9057317427989524, + "flow_matching_loss": 0.01690651969984174, + "grad_norm": 0.08416274189949036, + "learning_rate": 2.7736176941916346e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6550 + }, + { + "epoch": 1.9086412569100961, + "flow_matching_loss": 0.018215662520378827, + "grad_norm": 0.10287459194660187, + "learning_rate": 2.772743447451311e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6560 + }, + { + "epoch": 1.9115507710212394, + "flow_matching_loss": 0.01618689140304923, + "grad_norm": 0.07721050828695297, + "learning_rate": 2.7718676542076872e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6570 + }, + { + "epoch": 1.9144602851323829, + "flow_matching_loss": 0.016593139339238405, + "grad_norm": 0.07704292237758636, + "learning_rate": 2.770990315524934e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6580 + }, + { + "epoch": 1.9173697992435264, + "flow_matching_loss": 0.017465021181851625, + "grad_norm": 0.08089347928762436, + "learning_rate": 2.7701114324691015e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6590 + }, + { + "epoch": 1.9202793133546696, + "flow_matching_loss": 0.016676111053675412, + "grad_norm": 0.09347733110189438, + "learning_rate": 2.7692310061081145e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6600 + }, + { + "epoch": 1.9231888274658133, + "flow_matching_loss": 0.01661946829408407, + "grad_norm": 0.10289081186056137, + "learning_rate": 2.7683490375117742e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6610 + }, + { + "epoch": 1.9260983415769566, + "flow_matching_loss": 0.016842301934957504, + "grad_norm": 0.0683227926492691, + "learning_rate": 2.7674655277517556e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6620 + }, + { + "epoch": 1.9290078556881, + "flow_matching_loss": 0.01733102910220623, + "grad_norm": 0.11925417184829712, + "learning_rate": 2.7665804779016072e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6630 + }, + { + "epoch": 1.9319173697992436, + "flow_matching_loss": 0.01607190826907754, + "grad_norm": 0.09829721599817276, + "learning_rate": 2.765693889036747e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6640 + }, + { + "epoch": 1.9348268839103868, + "flow_matching_loss": 0.016802352014929058, + "grad_norm": 0.08800160139799118, + "learning_rate": 2.764805762234465e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6650 + }, + { + "epoch": 1.9377363980215305, + "flow_matching_loss": 0.018525593634694814, + "grad_norm": 0.11344432830810547, + "learning_rate": 2.7639160985739182e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6660 + }, + { + "epoch": 1.9406459121326738, + "flow_matching_loss": 0.018025882262736558, + "grad_norm": 0.07311772555112839, + "learning_rate": 2.7630248991361317e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6670 + }, + { + "epoch": 1.9435554262438173, + "flow_matching_loss": 0.01871368708088994, + "grad_norm": 0.10023481398820877, + "learning_rate": 2.762132165003998e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6680 + }, + { + "epoch": 1.9464649403549608, + "flow_matching_loss": 0.016176209878176452, + "grad_norm": 0.08521554619073868, + "learning_rate": 2.7612378972622722e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6690 + }, + { + "epoch": 1.949374454466104, + "flow_matching_loss": 0.016953679919242858, + "grad_norm": 0.09448737651109695, + "learning_rate": 2.7603420969975743e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6700 + }, + { + "epoch": 1.9522839685772477, + "flow_matching_loss": 0.018337500374764205, + "grad_norm": 0.09041756391525269, + "learning_rate": 2.7594447652983862e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6710 + }, + { + "epoch": 1.955193482688391, + "flow_matching_loss": 0.018385697435587645, + "grad_norm": 0.10844827443361282, + "learning_rate": 2.7585459032550497e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6720 + }, + { + "epoch": 1.9581029967995345, + "flow_matching_loss": 0.015564127918332815, + "grad_norm": 0.09805431216955185, + "learning_rate": 2.7576455119597684e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6730 + }, + { + "epoch": 1.961012510910678, + "flow_matching_loss": 0.016847989242523908, + "grad_norm": 0.08842213451862335, + "learning_rate": 2.7567435925066012e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6740 + }, + { + "epoch": 1.9639220250218212, + "flow_matching_loss": 0.017531768698245288, + "grad_norm": 0.07309678196907043, + "learning_rate": 2.7558401459914665e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6750 + }, + { + "epoch": 1.966831539132965, + "flow_matching_loss": 0.016719388868659736, + "grad_norm": 0.07056431472301483, + "learning_rate": 2.7549351735121362e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6760 + }, + { + "epoch": 1.9697410532441082, + "flow_matching_loss": 0.01745654083788395, + "grad_norm": 0.10567488521337509, + "learning_rate": 2.7540286761682377e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6770 + }, + { + "epoch": 1.9726505673552517, + "flow_matching_loss": 0.018537852726876736, + "grad_norm": 0.11335349082946777, + "learning_rate": 2.75312065506125e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6780 + }, + { + "epoch": 1.9755600814663952, + "flow_matching_loss": 0.018655801378190517, + "grad_norm": 0.11768301576375961, + "learning_rate": 2.7522111112945052e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6790 + }, + { + "epoch": 1.9784695955775384, + "flow_matching_loss": 0.017906893324106932, + "grad_norm": 0.1045667827129364, + "learning_rate": 2.751300045973184e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6800 + }, + { + "epoch": 1.9813791096886821, + "flow_matching_loss": 0.01591968797147274, + "grad_norm": 0.07371080666780472, + "learning_rate": 2.750387460204317e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6810 + }, + { + "epoch": 1.9842886237998254, + "flow_matching_loss": 0.016324034705758095, + "grad_norm": 0.10299493372440338, + "learning_rate": 2.7494733550967827e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6820 + }, + { + "epoch": 1.9871981379109689, + "flow_matching_loss": 0.01684019509702921, + "grad_norm": 0.08735819905996323, + "learning_rate": 2.748557731761304e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6830 + }, + { + "epoch": 1.9901076520221124, + "flow_matching_loss": 0.01826313640922308, + "grad_norm": 0.0856560543179512, + "learning_rate": 2.74764059131045e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6840 + }, + { + "epoch": 1.9930171661332556, + "flow_matching_loss": 0.017709902580827476, + "grad_norm": 0.09543632715940475, + "learning_rate": 2.7467219348586322e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6850 + }, + { + "epoch": 1.9959266802443993, + "flow_matching_loss": 0.017147672548890113, + "grad_norm": 0.074175164103508, + "learning_rate": 2.7458017635221056e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6860 + }, + { + "epoch": 1.9988361943555426, + "flow_matching_loss": 0.018140558060258626, + "grad_norm": 0.07267837971448898, + "learning_rate": 2.7448800784189652e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6870 + }, + { + "epoch": 2.001745708466686, + "flow_matching_loss": 0.018649403098970653, + "grad_norm": 0.09880408644676208, + "learning_rate": 2.7439568806691448e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6880 + }, + { + "epoch": 2.0046552225778296, + "flow_matching_loss": 0.01568167908117175, + "grad_norm": 0.09546150267124176, + "learning_rate": 2.7430321713944167e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6890 + }, + { + "epoch": 2.007564736688973, + "flow_matching_loss": 0.01583406049758196, + "grad_norm": 0.07993894070386887, + "learning_rate": 2.7421059517183906e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6900 + }, + { + "epoch": 2.0104742508001165, + "flow_matching_loss": 0.017210769560188056, + "grad_norm": 0.0679052397608757, + "learning_rate": 2.7411782227665097e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6910 + }, + { + "epoch": 2.01338376491126, + "flow_matching_loss": 0.01899720123037696, + "grad_norm": 0.08575404435396194, + "learning_rate": 2.7402489856660528e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6920 + }, + { + "epoch": 2.016293279022403, + "flow_matching_loss": 0.015685430634766816, + "grad_norm": 0.09661939740180969, + "learning_rate": 2.7393182415461307e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6930 + }, + { + "epoch": 2.019202793133547, + "flow_matching_loss": 0.017687340546399356, + "grad_norm": 0.09781120717525482, + "learning_rate": 2.7383859915376853e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6940 + }, + { + "epoch": 2.02211230724469, + "flow_matching_loss": 0.017146237194538116, + "grad_norm": 0.08979441225528717, + "learning_rate": 2.7374522367734875e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6950 + }, + { + "epoch": 2.0250218213558338, + "flow_matching_loss": 0.017145304288715125, + "grad_norm": 0.08814993500709534, + "learning_rate": 2.7365169783881378e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6960 + }, + { + "epoch": 2.027931335466977, + "flow_matching_loss": 0.015137003920972347, + "grad_norm": 0.10397441685199738, + "learning_rate": 2.735580217518064e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6970 + }, + { + "epoch": 2.0308408495781203, + "flow_matching_loss": 0.017604984156787396, + "grad_norm": 0.07871846109628677, + "learning_rate": 2.7346419553015175e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6980 + }, + { + "epoch": 2.033750363689264, + "flow_matching_loss": 0.01810803757980466, + "grad_norm": 0.09439206123352051, + "learning_rate": 2.733702192878577e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 6990 + }, + { + "epoch": 2.0366598778004072, + "flow_matching_loss": 0.016409308649599552, + "grad_norm": 0.09747392684221268, + "learning_rate": 2.732760931391141e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7000 + }, + { + "epoch": 2.039569391911551, + "flow_matching_loss": 0.017041077185422183, + "grad_norm": 0.08043479919433594, + "learning_rate": 2.731818171982932e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7010 + }, + { + "epoch": 2.042478906022694, + "flow_matching_loss": 0.015871071815490724, + "grad_norm": 0.10757005959749222, + "learning_rate": 2.730873915799491e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7020 + }, + { + "epoch": 2.0453884201338375, + "flow_matching_loss": 0.017325311060994865, + "grad_norm": 0.08819490671157837, + "learning_rate": 2.7299281639881783e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7030 + }, + { + "epoch": 2.048297934244981, + "flow_matching_loss": 0.014968813117593526, + "grad_norm": 0.05725795775651932, + "learning_rate": 2.728980917698172e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7040 + }, + { + "epoch": 2.0512074483561245, + "flow_matching_loss": 0.016627685725688936, + "grad_norm": 0.10328594595193863, + "learning_rate": 2.7280321780804655e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7050 + }, + { + "epoch": 2.054116962467268, + "flow_matching_loss": 0.017853218596428633, + "grad_norm": 0.10583736002445221, + "learning_rate": 2.7270819462878666e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7060 + }, + { + "epoch": 2.0570264765784114, + "flow_matching_loss": 0.01643746243789792, + "grad_norm": 0.07140009105205536, + "learning_rate": 2.7261302234749965e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7070 + }, + { + "epoch": 2.0599359906895547, + "flow_matching_loss": 0.015793882962316274, + "grad_norm": 0.11548943817615509, + "learning_rate": 2.7251770107982888e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7080 + }, + { + "epoch": 2.0628455048006984, + "flow_matching_loss": 0.016571137215942146, + "grad_norm": 0.07252912223339081, + "learning_rate": 2.7242223094159856e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7090 + }, + { + "epoch": 2.0657550189118417, + "flow_matching_loss": 0.01671187235042453, + "grad_norm": 0.08494837582111359, + "learning_rate": 2.7232661204881408e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7100 + }, + { + "epoch": 2.0686645330229854, + "flow_matching_loss": 0.015298077557235955, + "grad_norm": 0.08627169579267502, + "learning_rate": 2.7223084451766123e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7110 + }, + { + "epoch": 2.0715740471341286, + "flow_matching_loss": 0.01733332620933652, + "grad_norm": 0.10902882367372513, + "learning_rate": 2.721349284645067e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7120 + }, + { + "epoch": 2.074483561245272, + "flow_matching_loss": 0.01754440935328603, + "grad_norm": 0.0893215760588646, + "learning_rate": 2.7203886400589752e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7130 + }, + { + "epoch": 2.0773930753564156, + "flow_matching_loss": 0.017637354601174594, + "grad_norm": 0.10940682888031006, + "learning_rate": 2.7194265125856105e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7140 + }, + { + "epoch": 2.080302589467559, + "flow_matching_loss": 0.017195680737495424, + "grad_norm": 0.11265061050653458, + "learning_rate": 2.718462903394049e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7150 + }, + { + "epoch": 2.0832121035787026, + "flow_matching_loss": 0.018028973322361707, + "grad_norm": 0.09953073412179947, + "learning_rate": 2.7174978136551662e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7160 + }, + { + "epoch": 2.086121617689846, + "flow_matching_loss": 0.01834177169948816, + "grad_norm": 0.11197014153003693, + "learning_rate": 2.716531244541637e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7170 + }, + { + "epoch": 2.089031131800989, + "flow_matching_loss": 0.016932375449687242, + "grad_norm": 0.08841590583324432, + "learning_rate": 2.715563197227934e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7180 + }, + { + "epoch": 2.091940645912133, + "flow_matching_loss": 0.015404341835528612, + "grad_norm": 0.07792938500642776, + "learning_rate": 2.714593672890327e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7190 + }, + { + "epoch": 2.094850160023276, + "flow_matching_loss": 0.017863807640969752, + "grad_norm": 0.1147279441356659, + "learning_rate": 2.7136226727068783e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7200 + }, + { + "epoch": 2.0977596741344193, + "flow_matching_loss": 0.015937092807143928, + "grad_norm": 0.09305734932422638, + "learning_rate": 2.712650197857445e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7210 + }, + { + "epoch": 2.100669188245563, + "flow_matching_loss": 0.017726405151188375, + "grad_norm": 0.08517958223819733, + "learning_rate": 2.7116762495236763e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7220 + }, + { + "epoch": 2.1035787023567063, + "flow_matching_loss": 0.016014457400888204, + "grad_norm": 0.10112561285495758, + "learning_rate": 2.7107008288890106e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7230 + }, + { + "epoch": 2.10648821646785, + "flow_matching_loss": 0.015732124075293542, + "grad_norm": 0.06461262702941895, + "learning_rate": 2.709723937138676e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7240 + }, + { + "epoch": 2.1093977305789933, + "flow_matching_loss": 0.01685096276924014, + "grad_norm": 0.0812191590666771, + "learning_rate": 2.7087455754596883e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7250 + }, + { + "epoch": 2.112307244690137, + "flow_matching_loss": 0.016350370645523072, + "grad_norm": 0.09727034717798233, + "learning_rate": 2.7077657450408498e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7260 + }, + { + "epoch": 2.1152167588012802, + "flow_matching_loss": 0.016032659262418748, + "grad_norm": 0.06649128347635269, + "learning_rate": 2.7067844470727456e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7270 + }, + { + "epoch": 2.1181262729124235, + "flow_matching_loss": 0.01920559788122773, + "grad_norm": 0.08866606652736664, + "learning_rate": 2.7058016827477455e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7280 + }, + { + "epoch": 2.121035787023567, + "flow_matching_loss": 0.015905411448329687, + "grad_norm": 0.10585571080446243, + "learning_rate": 2.7048174532600018e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7290 + }, + { + "epoch": 2.1239453011347105, + "flow_matching_loss": 0.016320299170911313, + "grad_norm": 0.10086130350828171, + "learning_rate": 2.7038317598054455e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7300 + }, + { + "epoch": 2.1268548152458537, + "flow_matching_loss": 0.016563319601118565, + "grad_norm": 0.09234333038330078, + "learning_rate": 2.702844603581787e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7310 + }, + { + "epoch": 2.1297643293569974, + "flow_matching_loss": 0.016718290094286204, + "grad_norm": 0.09610690921545029, + "learning_rate": 2.701855985788514e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7320 + }, + { + "epoch": 2.1326738434681407, + "flow_matching_loss": 0.016672453191131354, + "grad_norm": 0.12475550174713135, + "learning_rate": 2.7008659076268912e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7330 + }, + { + "epoch": 2.1355833575792844, + "flow_matching_loss": 0.017712255660444497, + "grad_norm": 0.08569250255823135, + "learning_rate": 2.6998743702999562e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7340 + }, + { + "epoch": 2.1384928716904277, + "flow_matching_loss": 0.016600201651453973, + "grad_norm": 0.12738731503486633, + "learning_rate": 2.6988813750125202e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7350 + }, + { + "epoch": 2.141402385801571, + "flow_matching_loss": 0.01709371255710721, + "grad_norm": 0.09901578724384308, + "learning_rate": 2.6978869229711668e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7360 + }, + { + "epoch": 2.1443118999127146, + "flow_matching_loss": 0.01729667289182544, + "grad_norm": 0.06964082270860672, + "learning_rate": 2.6968910153842484e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7370 + }, + { + "epoch": 2.147221414023858, + "flow_matching_loss": 0.015839887596666814, + "grad_norm": 0.08834230154752731, + "learning_rate": 2.6958936534618874e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7380 + }, + { + "epoch": 2.1501309281350016, + "flow_matching_loss": 0.016467356868088246, + "grad_norm": 0.09291891753673553, + "learning_rate": 2.694894838415971e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7390 + }, + { + "epoch": 2.153040442246145, + "flow_matching_loss": 0.01608502585440874, + "grad_norm": 0.10061835497617722, + "learning_rate": 2.6938945714601555e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7400 + }, + { + "epoch": 2.155949956357288, + "flow_matching_loss": 0.015590994711965322, + "grad_norm": 0.08331675827503204, + "learning_rate": 2.6928928538098587e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7410 + }, + { + "epoch": 2.158859470468432, + "flow_matching_loss": 0.0180085901170969, + "grad_norm": 0.08184441924095154, + "learning_rate": 2.6918896866822615e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7420 + }, + { + "epoch": 2.161768984579575, + "flow_matching_loss": 0.01567588560283184, + "grad_norm": 0.09979327768087387, + "learning_rate": 2.690885071296308e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7430 + }, + { + "epoch": 2.164678498690719, + "flow_matching_loss": 0.01817941004410386, + "grad_norm": 0.10627494752407074, + "learning_rate": 2.6898790088726993e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7440 + }, + { + "epoch": 2.167588012801862, + "flow_matching_loss": 0.017556030116975308, + "grad_norm": 0.1143597811460495, + "learning_rate": 2.6888715006338968e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7450 + }, + { + "epoch": 2.1704975269130053, + "flow_matching_loss": 0.01713912608101964, + "grad_norm": 0.09412652254104614, + "learning_rate": 2.6878625478041172e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7460 + }, + { + "epoch": 2.173407041024149, + "flow_matching_loss": 0.020053531508892775, + "grad_norm": 0.10254962742328644, + "learning_rate": 2.6868521516093342e-05, + "loss": 0.0201, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7470 + }, + { + "epoch": 2.1763165551352923, + "flow_matching_loss": 0.014933906868100167, + "grad_norm": 0.07866193354129791, + "learning_rate": 2.6858403132772745e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7480 + }, + { + "epoch": 2.179226069246436, + "flow_matching_loss": 0.01710622850805521, + "grad_norm": 0.10630738735198975, + "learning_rate": 2.6848270340374162e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7490 + }, + { + "epoch": 2.1821355833575793, + "flow_matching_loss": 0.015119362063705922, + "grad_norm": 0.06704124063253403, + "learning_rate": 2.68381231512099e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7500 + }, + { + "epoch": 2.1850450974687226, + "flow_matching_loss": 0.017535852268338203, + "grad_norm": 0.07470168173313141, + "learning_rate": 2.6827961577609745e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7510 + }, + { + "epoch": 2.1879546115798663, + "flow_matching_loss": 0.01651258682832122, + "grad_norm": 0.06797785311937332, + "learning_rate": 2.681778563192097e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7520 + }, + { + "epoch": 2.1908641256910095, + "flow_matching_loss": 0.016675827093422413, + "grad_norm": 0.09902913123369217, + "learning_rate": 2.6807595326508308e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7530 + }, + { + "epoch": 2.1937736398021532, + "flow_matching_loss": 0.018646321073174477, + "grad_norm": 0.11385133862495422, + "learning_rate": 2.679739067375394e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7540 + }, + { + "epoch": 2.1966831539132965, + "flow_matching_loss": 0.018976420722901823, + "grad_norm": 0.09049560129642487, + "learning_rate": 2.6787171686057487e-05, + "loss": 0.019, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7550 + }, + { + "epoch": 2.1995926680244398, + "flow_matching_loss": 0.017389330454170703, + "grad_norm": 0.11519832909107208, + "learning_rate": 2.677693837583597e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7560 + }, + { + "epoch": 2.2025021821355835, + "flow_matching_loss": 0.016242699511349202, + "grad_norm": 0.08297033607959747, + "learning_rate": 2.6766690755523836e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7570 + }, + { + "epoch": 2.2054116962467267, + "flow_matching_loss": 0.016050766315311195, + "grad_norm": 0.09951205551624298, + "learning_rate": 2.675642883757291e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7580 + }, + { + "epoch": 2.2083212103578704, + "flow_matching_loss": 0.01704913517460227, + "grad_norm": 0.0927065834403038, + "learning_rate": 2.674615263445239e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7590 + }, + { + "epoch": 2.2112307244690137, + "flow_matching_loss": 0.01830225419253111, + "grad_norm": 0.08197597414255142, + "learning_rate": 2.673586215864882e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7600 + }, + { + "epoch": 2.214140238580157, + "flow_matching_loss": 0.016685486398637296, + "grad_norm": 0.08854865282773972, + "learning_rate": 2.6725557422666113e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7610 + }, + { + "epoch": 2.2170497526913007, + "flow_matching_loss": 0.01620900696143508, + "grad_norm": 0.09393265098333359, + "learning_rate": 2.6715238439025493e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7620 + }, + { + "epoch": 2.219959266802444, + "flow_matching_loss": 0.016962279193103315, + "grad_norm": 0.09990477561950684, + "learning_rate": 2.6704905220265493e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7630 + }, + { + "epoch": 2.2228687809135876, + "flow_matching_loss": 0.017767627816647293, + "grad_norm": 0.08516903221607208, + "learning_rate": 2.6694557778941947e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7640 + }, + { + "epoch": 2.225778295024731, + "flow_matching_loss": 0.0171190882101655, + "grad_norm": 0.09781587868928909, + "learning_rate": 2.6684196127627977e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7650 + }, + { + "epoch": 2.228687809135874, + "flow_matching_loss": 0.015458602365106345, + "grad_norm": 0.0773041769862175, + "learning_rate": 2.6673820278913963e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7660 + }, + { + "epoch": 2.231597323247018, + "flow_matching_loss": 0.015724196285009383, + "grad_norm": 0.06635570526123047, + "learning_rate": 2.6663430245407544e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7670 + }, + { + "epoch": 2.234506837358161, + "flow_matching_loss": 0.017557672597467898, + "grad_norm": 0.0909775123000145, + "learning_rate": 2.6653026039733585e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7680 + }, + { + "epoch": 2.237416351469305, + "flow_matching_loss": 0.01774633824825287, + "grad_norm": 0.07385165244340897, + "learning_rate": 2.6642607674534176e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7690 + }, + { + "epoch": 2.240325865580448, + "flow_matching_loss": 0.01760054137557745, + "grad_norm": 0.07193442434072495, + "learning_rate": 2.663217516246862e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7700 + }, + { + "epoch": 2.2432353796915914, + "flow_matching_loss": 0.01621604720130563, + "grad_norm": 0.10343901067972183, + "learning_rate": 2.66217285162134e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7710 + }, + { + "epoch": 2.246144893802735, + "flow_matching_loss": 0.01769486125558615, + "grad_norm": 0.09560664743185043, + "learning_rate": 2.6611267748462175e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7720 + }, + { + "epoch": 2.2490544079138783, + "flow_matching_loss": 0.016414881125092505, + "grad_norm": 0.09132366627454758, + "learning_rate": 2.6600792871925764e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7730 + }, + { + "epoch": 2.2519639220250216, + "flow_matching_loss": 0.01611247193068266, + "grad_norm": 0.09540247172117233, + "learning_rate": 2.659030389933213e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7740 + }, + { + "epoch": 2.2548734361361653, + "flow_matching_loss": 0.01768200993537903, + "grad_norm": 0.10282265394926071, + "learning_rate": 2.6579800843426358e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7750 + }, + { + "epoch": 2.2577829502473086, + "flow_matching_loss": 0.017114116996526717, + "grad_norm": 0.08815206587314606, + "learning_rate": 2.656928371697067e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7760 + }, + { + "epoch": 2.2606924643584523, + "flow_matching_loss": 0.016887757927179337, + "grad_norm": 0.07433019578456879, + "learning_rate": 2.6558752532744346e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7770 + }, + { + "epoch": 2.2636019784695955, + "flow_matching_loss": 0.01811790084466338, + "grad_norm": 0.09739366173744202, + "learning_rate": 2.654820730354377e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7780 + }, + { + "epoch": 2.2665114925807393, + "flow_matching_loss": 0.01801857678219676, + "grad_norm": 0.10852178186178207, + "learning_rate": 2.6537648042182406e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7790 + }, + { + "epoch": 2.2694210066918825, + "flow_matching_loss": 0.015986242704093455, + "grad_norm": 0.08997005969285965, + "learning_rate": 2.6527074761490735e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7800 + }, + { + "epoch": 2.2723305208030258, + "flow_matching_loss": 0.015405334066599607, + "grad_norm": 0.06968945264816284, + "learning_rate": 2.6516487474316303e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7810 + }, + { + "epoch": 2.2752400349141695, + "flow_matching_loss": 0.01569924969226122, + "grad_norm": 0.07129793614149094, + "learning_rate": 2.650588619352365e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7820 + }, + { + "epoch": 2.2781495490253127, + "flow_matching_loss": 0.016208725143224002, + "grad_norm": 0.0916820615530014, + "learning_rate": 2.649527093199434e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7830 + }, + { + "epoch": 2.281059063136456, + "flow_matching_loss": 0.0148994873277843, + "grad_norm": 0.07366766035556793, + "learning_rate": 2.6484641702626914e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7840 + }, + { + "epoch": 2.2839685772475997, + "flow_matching_loss": 0.016457920521497728, + "grad_norm": 0.09128957241773605, + "learning_rate": 2.647399851833689e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7850 + }, + { + "epoch": 2.286878091358743, + "flow_matching_loss": 0.016783568169921638, + "grad_norm": 0.08044405281543732, + "learning_rate": 2.6463341392056735e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7860 + }, + { + "epoch": 2.2897876054698867, + "flow_matching_loss": 0.015665628109127284, + "grad_norm": 0.09965658187866211, + "learning_rate": 2.6452670336735867e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7870 + }, + { + "epoch": 2.29269711958103, + "flow_matching_loss": 0.01742576202377677, + "grad_norm": 0.08097612112760544, + "learning_rate": 2.644198536534063e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7880 + }, + { + "epoch": 2.295606633692173, + "flow_matching_loss": 0.017399293836206198, + "grad_norm": 0.13303282856941223, + "learning_rate": 2.6431286490854255e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7890 + }, + { + "epoch": 2.298516147803317, + "flow_matching_loss": 0.018498538807034493, + "grad_norm": 0.09241434931755066, + "learning_rate": 2.6420573726276903e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7900 + }, + { + "epoch": 2.30142566191446, + "flow_matching_loss": 0.01708095194771886, + "grad_norm": 0.10010165721178055, + "learning_rate": 2.6409847084625575e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7910 + }, + { + "epoch": 2.304335176025604, + "flow_matching_loss": 0.0164679198525846, + "grad_norm": 0.080846406519413, + "learning_rate": 2.6399106578934167e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7920 + }, + { + "epoch": 2.307244690136747, + "flow_matching_loss": 0.01872089523822069, + "grad_norm": 0.111399345099926, + "learning_rate": 2.63883522222534e-05, + "loss": 0.0187, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7930 + }, + { + "epoch": 2.3101542042478904, + "flow_matching_loss": 0.01631526369601488, + "grad_norm": 0.09905419498682022, + "learning_rate": 2.6377584027650827e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7940 + }, + { + "epoch": 2.313063718359034, + "flow_matching_loss": 0.017783665843307973, + "grad_norm": 0.10604127496480942, + "learning_rate": 2.636680200821083e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7950 + }, + { + "epoch": 2.3159732324701774, + "flow_matching_loss": 0.016937231458723546, + "grad_norm": 0.10406281054019928, + "learning_rate": 2.635600617703457e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7960 + }, + { + "epoch": 2.318882746581321, + "flow_matching_loss": 0.016752655059099196, + "grad_norm": 0.08549328148365021, + "learning_rate": 2.6345196547240007e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7970 + }, + { + "epoch": 2.3217922606924644, + "flow_matching_loss": 0.01656844848766923, + "grad_norm": 0.10413024574518204, + "learning_rate": 2.6334373131961854e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7980 + }, + { + "epoch": 2.3247017748036076, + "flow_matching_loss": 0.016395817045122385, + "grad_norm": 0.09152618795633316, + "learning_rate": 2.6323535944351583e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 7990 + }, + { + "epoch": 2.3276112889147513, + "flow_matching_loss": 0.01562562584877014, + "grad_norm": 0.06654234975576401, + "learning_rate": 2.63126849975774e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8000 + }, + { + "epoch": 2.3305208030258946, + "flow_matching_loss": 0.016589518170803785, + "grad_norm": 0.09738603979349136, + "learning_rate": 2.6301820304824234e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8010 + }, + { + "epoch": 2.3334303171370383, + "flow_matching_loss": 0.015576600469648838, + "grad_norm": 0.09597738832235336, + "learning_rate": 2.6290941879293703e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8020 + }, + { + "epoch": 2.3363398312481816, + "flow_matching_loss": 0.017150271870195865, + "grad_norm": 0.08966708183288574, + "learning_rate": 2.628004973420412e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8030 + }, + { + "epoch": 2.339249345359325, + "flow_matching_loss": 0.017380709759891033, + "grad_norm": 0.10492178052663803, + "learning_rate": 2.626914388279047e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8040 + }, + { + "epoch": 2.3421588594704685, + "flow_matching_loss": 0.0157411927357316, + "grad_norm": 0.08523256331682205, + "learning_rate": 2.6258224338304394e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8050 + }, + { + "epoch": 2.345068373581612, + "flow_matching_loss": 0.016672222316265105, + "grad_norm": 0.0916673094034195, + "learning_rate": 2.6247291114014166e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8060 + }, + { + "epoch": 2.347977887692755, + "flow_matching_loss": 0.016005220543593168, + "grad_norm": 0.08753997087478638, + "learning_rate": 2.6236344223204683e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8070 + }, + { + "epoch": 2.3508874018038988, + "flow_matching_loss": 0.017237199656665324, + "grad_norm": 0.08104145526885986, + "learning_rate": 2.6225383679177453e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8080 + }, + { + "epoch": 2.353796915915042, + "flow_matching_loss": 0.016359394509345294, + "grad_norm": 0.0782342478632927, + "learning_rate": 2.6214409495250568e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8090 + }, + { + "epoch": 2.3567064300261857, + "flow_matching_loss": 0.019949413556605577, + "grad_norm": 0.08833711594343185, + "learning_rate": 2.620342168475869e-05, + "loss": 0.0199, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8100 + }, + { + "epoch": 2.359615944137329, + "flow_matching_loss": 0.017892360780388118, + "grad_norm": 0.11404598504304886, + "learning_rate": 2.619242026105306e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8110 + }, + { + "epoch": 2.3625254582484727, + "flow_matching_loss": 0.016043529007583858, + "grad_norm": 0.059389930218458176, + "learning_rate": 2.618140523750143e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8120 + }, + { + "epoch": 2.365434972359616, + "flow_matching_loss": 0.01699010971933603, + "grad_norm": 0.08135490864515305, + "learning_rate": 2.6170376627488103e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8130 + }, + { + "epoch": 2.3683444864707592, + "flow_matching_loss": 0.017447092942893504, + "grad_norm": 0.10265641659498215, + "learning_rate": 2.6159334444413873e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8140 + }, + { + "epoch": 2.371254000581903, + "flow_matching_loss": 0.019417719449847937, + "grad_norm": 0.1085008978843689, + "learning_rate": 2.614827870169604e-05, + "loss": 0.0194, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8150 + }, + { + "epoch": 2.374163514693046, + "flow_matching_loss": 0.01499771662056446, + "grad_norm": 0.10918427258729935, + "learning_rate": 2.6137209412768367e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8160 + }, + { + "epoch": 2.3770730288041895, + "flow_matching_loss": 0.018060886580497027, + "grad_norm": 0.08488847315311432, + "learning_rate": 2.612612659108109e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8170 + }, + { + "epoch": 2.379982542915333, + "flow_matching_loss": 0.01644488200545311, + "grad_norm": 0.10751072317361832, + "learning_rate": 2.611503025010088e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8180 + }, + { + "epoch": 2.3828920570264764, + "flow_matching_loss": 0.016033994778990746, + "grad_norm": 0.0854162946343422, + "learning_rate": 2.6103920403310832e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8190 + }, + { + "epoch": 2.38580157113762, + "flow_matching_loss": 0.016311317216604948, + "grad_norm": 0.0746743232011795, + "learning_rate": 2.6092797064210465e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8200 + }, + { + "epoch": 2.3887110852487634, + "flow_matching_loss": 0.017289732117205858, + "grad_norm": 0.0926615372300148, + "learning_rate": 2.6081660246315685e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8210 + }, + { + "epoch": 2.391620599359907, + "flow_matching_loss": 0.015510873962193727, + "grad_norm": 0.07280011475086212, + "learning_rate": 2.6070509963158774e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8220 + }, + { + "epoch": 2.3945301134710504, + "flow_matching_loss": 0.015493584703654051, + "grad_norm": 0.07811743766069412, + "learning_rate": 2.6059346228288375e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8230 + }, + { + "epoch": 2.3974396275821936, + "flow_matching_loss": 0.016811930388212205, + "grad_norm": 0.07725562900304794, + "learning_rate": 2.604816905526948e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8240 + }, + { + "epoch": 2.4003491416933374, + "flow_matching_loss": 0.015829139202833176, + "grad_norm": 0.07486899197101593, + "learning_rate": 2.603697845768341e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8250 + }, + { + "epoch": 2.4032586558044806, + "flow_matching_loss": 0.016239497065544128, + "grad_norm": 0.08499974757432938, + "learning_rate": 2.6025774449127792e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8260 + }, + { + "epoch": 2.406168169915624, + "flow_matching_loss": 0.01587969521060586, + "grad_norm": 0.0990145280957222, + "learning_rate": 2.6014557043216562e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8270 + }, + { + "epoch": 2.4090776840267676, + "flow_matching_loss": 0.015629934798926114, + "grad_norm": 0.09616070240736008, + "learning_rate": 2.6003326253579913e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8280 + }, + { + "epoch": 2.411987198137911, + "flow_matching_loss": 0.016428003273904323, + "grad_norm": 0.09419026225805283, + "learning_rate": 2.599208209386432e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8290 + }, + { + "epoch": 2.4148967122490546, + "flow_matching_loss": 0.017669115029275418, + "grad_norm": 0.1288117915391922, + "learning_rate": 2.5980824577732502e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8300 + }, + { + "epoch": 2.417806226360198, + "flow_matching_loss": 0.015752539690583944, + "grad_norm": 0.12355592101812363, + "learning_rate": 2.5969553718863395e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8310 + }, + { + "epoch": 2.4207157404713415, + "flow_matching_loss": 0.01775268539786339, + "grad_norm": 0.09511879831552505, + "learning_rate": 2.595826953095216e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8320 + }, + { + "epoch": 2.423625254582485, + "flow_matching_loss": 0.01637167762964964, + "grad_norm": 0.09923364967107773, + "learning_rate": 2.594697202771014e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8330 + }, + { + "epoch": 2.426534768693628, + "flow_matching_loss": 0.016967440955340862, + "grad_norm": 0.09250941127538681, + "learning_rate": 2.5935661222864888e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8340 + }, + { + "epoch": 2.4294442828047718, + "flow_matching_loss": 0.018339756689965726, + "grad_norm": 0.10274505615234375, + "learning_rate": 2.592433713016008e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8350 + }, + { + "epoch": 2.432353796915915, + "flow_matching_loss": 0.01422146437689662, + "grad_norm": 0.07761015743017197, + "learning_rate": 2.5912999763355553e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8360 + }, + { + "epoch": 2.4352633110270583, + "flow_matching_loss": 0.015393354929983616, + "grad_norm": 0.091717429459095, + "learning_rate": 2.590164913622729e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8370 + }, + { + "epoch": 2.438172825138202, + "flow_matching_loss": 0.017315532360225917, + "grad_norm": 0.10919360816478729, + "learning_rate": 2.589028526256737e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8380 + }, + { + "epoch": 2.4410823392493453, + "flow_matching_loss": 0.016504412237554788, + "grad_norm": 0.06290418654680252, + "learning_rate": 2.587890815618397e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8390 + }, + { + "epoch": 2.443991853360489, + "flow_matching_loss": 0.017209468688815833, + "grad_norm": 0.12301398068666458, + "learning_rate": 2.5867517830901343e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8400 + }, + { + "epoch": 2.4469013674716322, + "flow_matching_loss": 0.016197823360562323, + "grad_norm": 0.0805886760354042, + "learning_rate": 2.5856114300559812e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8410 + }, + { + "epoch": 2.4498108815827755, + "flow_matching_loss": 0.01730370447039604, + "grad_norm": 0.123382069170475, + "learning_rate": 2.584469757901574e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8420 + }, + { + "epoch": 2.452720395693919, + "flow_matching_loss": 0.014983690530061721, + "grad_norm": 0.06714499741792679, + "learning_rate": 2.5833267680141526e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8430 + }, + { + "epoch": 2.4556299098050625, + "flow_matching_loss": 0.016014325059950353, + "grad_norm": 0.09412454068660736, + "learning_rate": 2.5821824617825564e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8440 + }, + { + "epoch": 2.458539423916206, + "flow_matching_loss": 0.016019146516919137, + "grad_norm": 0.07846289128065109, + "learning_rate": 2.5810368405972265e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8450 + }, + { + "epoch": 2.4614489380273494, + "flow_matching_loss": 0.01666690530255437, + "grad_norm": 0.08016729354858398, + "learning_rate": 2.5798899058501996e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8460 + }, + { + "epoch": 2.4643584521384927, + "flow_matching_loss": 0.014872416574507951, + "grad_norm": 0.07650582492351532, + "learning_rate": 2.578741658935111e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8470 + }, + { + "epoch": 2.4672679662496364, + "flow_matching_loss": 0.015769103541970253, + "grad_norm": 0.08610288798809052, + "learning_rate": 2.5775921012471875e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8480 + }, + { + "epoch": 2.4701774803607797, + "flow_matching_loss": 0.015561928320676089, + "grad_norm": 0.07838276773691177, + "learning_rate": 2.5764412341832505e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8490 + }, + { + "epoch": 2.4730869944719234, + "flow_matching_loss": 0.01717992126941681, + "grad_norm": 0.11722113937139511, + "learning_rate": 2.5752890591417125e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8500 + }, + { + "epoch": 2.4759965085830666, + "flow_matching_loss": 0.018380965571850538, + "grad_norm": 0.07170277088880539, + "learning_rate": 2.5741355775225744e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8510 + }, + { + "epoch": 2.47890602269421, + "flow_matching_loss": 0.016638895403593777, + "grad_norm": 0.12257634103298187, + "learning_rate": 2.5729807907274254e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8520 + }, + { + "epoch": 2.4818155368053536, + "flow_matching_loss": 0.01692306622862816, + "grad_norm": 0.10204463452100754, + "learning_rate": 2.5718247001594403e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8530 + }, + { + "epoch": 2.484725050916497, + "flow_matching_loss": 0.015346733666956425, + "grad_norm": 0.07862623035907745, + "learning_rate": 2.5706673072233776e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8540 + }, + { + "epoch": 2.4876345650276406, + "flow_matching_loss": 0.016150155290961265, + "grad_norm": 0.08030939847230911, + "learning_rate": 2.5695086133255795e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8550 + }, + { + "epoch": 2.490544079138784, + "flow_matching_loss": 0.017190087027847765, + "grad_norm": 0.06853018701076508, + "learning_rate": 2.5683486198739687e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8560 + }, + { + "epoch": 2.493453593249927, + "flow_matching_loss": 0.016670877300202846, + "grad_norm": 0.08541064709424973, + "learning_rate": 2.5671873282780452e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8570 + }, + { + "epoch": 2.496363107361071, + "flow_matching_loss": 0.017014114931225777, + "grad_norm": 0.08021650463342667, + "learning_rate": 2.5660247399488887e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8580 + }, + { + "epoch": 2.499272621472214, + "flow_matching_loss": 0.016526057664304972, + "grad_norm": 0.08338150382041931, + "learning_rate": 2.564860856299154e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8590 + }, + { + "epoch": 2.5021821355833573, + "flow_matching_loss": 0.015825955662876368, + "grad_norm": 0.09757062792778015, + "learning_rate": 2.5636956787430693e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8600 + }, + { + "epoch": 2.505091649694501, + "flow_matching_loss": 0.017337079998105763, + "grad_norm": 0.08473268151283264, + "learning_rate": 2.5625292086964346e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8610 + }, + { + "epoch": 2.5080011638056443, + "flow_matching_loss": 0.017533166985958814, + "grad_norm": 0.11570984125137329, + "learning_rate": 2.5613614475766212e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8620 + }, + { + "epoch": 2.510910677916788, + "flow_matching_loss": 0.016769272927194833, + "grad_norm": 0.07681332528591156, + "learning_rate": 2.5601923968025694e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8630 + }, + { + "epoch": 2.5138201920279313, + "flow_matching_loss": 0.015227507520467042, + "grad_norm": 0.09600352495908737, + "learning_rate": 2.5590220577947862e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8640 + }, + { + "epoch": 2.516729706139075, + "flow_matching_loss": 0.01692957794293761, + "grad_norm": 0.10557198524475098, + "learning_rate": 2.5578504319753427e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8650 + }, + { + "epoch": 2.5196392202502182, + "flow_matching_loss": 0.016908188909292222, + "grad_norm": 0.07588446885347366, + "learning_rate": 2.556677520767876e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8660 + }, + { + "epoch": 2.5225487343613615, + "flow_matching_loss": 0.01524846227839589, + "grad_norm": 0.07457084953784943, + "learning_rate": 2.5555033255975835e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8670 + }, + { + "epoch": 2.525458248472505, + "flow_matching_loss": 0.015966742020100354, + "grad_norm": 0.08100665360689163, + "learning_rate": 2.5543278478912228e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8680 + }, + { + "epoch": 2.5283677625836485, + "flow_matching_loss": 0.01737617710605264, + "grad_norm": 0.09512031078338623, + "learning_rate": 2.55315108907711e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8690 + }, + { + "epoch": 2.5312772766947917, + "flow_matching_loss": 0.017802881170064212, + "grad_norm": 0.10854946821928024, + "learning_rate": 2.551973050585119e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8700 + }, + { + "epoch": 2.5341867908059355, + "flow_matching_loss": 0.015953502804040908, + "grad_norm": 0.08053968101739883, + "learning_rate": 2.5507937338466768e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8710 + }, + { + "epoch": 2.5370963049170787, + "flow_matching_loss": 0.014678433816879988, + "grad_norm": 0.08224227279424667, + "learning_rate": 2.549613140294764e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8720 + }, + { + "epoch": 2.5400058190282224, + "flow_matching_loss": 0.015496090054512024, + "grad_norm": 0.09877459704875946, + "learning_rate": 2.548431271363914e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8730 + }, + { + "epoch": 2.5429153331393657, + "flow_matching_loss": 0.01751486174762249, + "grad_norm": 0.08254294097423553, + "learning_rate": 2.5472481284902092e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8740 + }, + { + "epoch": 2.5458248472505094, + "flow_matching_loss": 0.016387844178825616, + "grad_norm": 0.09876764565706253, + "learning_rate": 2.5460637131112788e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8750 + }, + { + "epoch": 2.5487343613616527, + "flow_matching_loss": 0.016851152293384075, + "grad_norm": 0.09861771017313004, + "learning_rate": 2.5448780266662995e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8760 + }, + { + "epoch": 2.551643875472796, + "flow_matching_loss": 0.01808974929153919, + "grad_norm": 0.07365987449884415, + "learning_rate": 2.5436910705959926e-05, + "loss": 0.0181, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8770 + }, + { + "epoch": 2.5545533895839396, + "flow_matching_loss": 0.016753123793751, + "grad_norm": 0.07674849778413773, + "learning_rate": 2.542502846342621e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8780 + }, + { + "epoch": 2.557462903695083, + "flow_matching_loss": 0.01741184191778302, + "grad_norm": 0.08699735254049301, + "learning_rate": 2.5413133553499892e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8790 + }, + { + "epoch": 2.560372417806226, + "flow_matching_loss": 0.01830142503604293, + "grad_norm": 0.07171973586082458, + "learning_rate": 2.540122599063442e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8800 + }, + { + "epoch": 2.56328193191737, + "flow_matching_loss": 0.015840360336005686, + "grad_norm": 0.06969989836215973, + "learning_rate": 2.538930578929859e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8810 + }, + { + "epoch": 2.566191446028513, + "flow_matching_loss": 0.018786312825977803, + "grad_norm": 0.10847988724708557, + "learning_rate": 2.537737296397658e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8820 + }, + { + "epoch": 2.569100960139657, + "flow_matching_loss": 0.01849815044552088, + "grad_norm": 0.07948047667741776, + "learning_rate": 2.5365427529167903e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8830 + }, + { + "epoch": 2.5720104742508, + "flow_matching_loss": 0.015357869770377874, + "grad_norm": 0.08873472362756729, + "learning_rate": 2.5353469499387375e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8840 + }, + { + "epoch": 2.574919988361944, + "flow_matching_loss": 0.01741977035999298, + "grad_norm": 0.07926098257303238, + "learning_rate": 2.534149888916514e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8850 + }, + { + "epoch": 2.577829502473087, + "flow_matching_loss": 0.014289793930947781, + "grad_norm": 0.07697245478630066, + "learning_rate": 2.5329515713046616e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8860 + }, + { + "epoch": 2.5807390165842303, + "flow_matching_loss": 0.015771857276558877, + "grad_norm": 0.07355915755033493, + "learning_rate": 2.531751998559249e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8870 + }, + { + "epoch": 2.583648530695374, + "flow_matching_loss": 0.015442801732569933, + "grad_norm": 0.10635778307914734, + "learning_rate": 2.5305511721378703e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8880 + }, + { + "epoch": 2.5865580448065173, + "flow_matching_loss": 0.015771003812551497, + "grad_norm": 0.10823024064302444, + "learning_rate": 2.529349093499642e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8890 + }, + { + "epoch": 2.5894675589176606, + "flow_matching_loss": 0.016132648475468157, + "grad_norm": 0.08514950424432755, + "learning_rate": 2.528145764105205e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8900 + }, + { + "epoch": 2.5923770730288043, + "flow_matching_loss": 0.015889378543943167, + "grad_norm": 0.08292815089225769, + "learning_rate": 2.5269411854167166e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8910 + }, + { + "epoch": 2.5952865871399475, + "flow_matching_loss": 0.018755771778523923, + "grad_norm": 0.08309318870306015, + "learning_rate": 2.5257353588978534e-05, + "loss": 0.0188, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8920 + }, + { + "epoch": 2.598196101251091, + "flow_matching_loss": 0.016749916318804025, + "grad_norm": 0.09995987266302109, + "learning_rate": 2.5245282860138083e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8930 + }, + { + "epoch": 2.6011056153622345, + "flow_matching_loss": 0.015869433153420688, + "grad_norm": 0.09192490577697754, + "learning_rate": 2.5233199682312894e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8940 + }, + { + "epoch": 2.604015129473378, + "flow_matching_loss": 0.01532344613224268, + "grad_norm": 0.06348004192113876, + "learning_rate": 2.5221104070185163e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8950 + }, + { + "epoch": 2.6069246435845215, + "flow_matching_loss": 0.016924058459699153, + "grad_norm": 0.09808795154094696, + "learning_rate": 2.5208996038452204e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8960 + }, + { + "epoch": 2.6098341576956647, + "flow_matching_loss": 0.015406109113246202, + "grad_norm": 0.08630405366420746, + "learning_rate": 2.5196875601826403e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8970 + }, + { + "epoch": 2.6127436718068084, + "flow_matching_loss": 0.01622479744255543, + "grad_norm": 0.07287503778934479, + "learning_rate": 2.518474277503525e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8980 + }, + { + "epoch": 2.6156531859179517, + "flow_matching_loss": 0.016194238886237146, + "grad_norm": 0.09219347685575485, + "learning_rate": 2.5172597572821266e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 8990 + }, + { + "epoch": 2.618562700029095, + "flow_matching_loss": 0.015293483901768923, + "grad_norm": 0.09198833256959915, + "learning_rate": 2.5160440009942015e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9000 + }, + { + "epoch": 2.6214722141402387, + "flow_matching_loss": 0.0157985495403409, + "grad_norm": 0.10003931820392609, + "learning_rate": 2.514827010117009e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9010 + }, + { + "epoch": 2.624381728251382, + "flow_matching_loss": 0.01564438231289387, + "grad_norm": 0.10483098775148392, + "learning_rate": 2.5136087861293068e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9020 + }, + { + "epoch": 2.627291242362525, + "flow_matching_loss": 0.015878251660615204, + "grad_norm": 0.07080615311861038, + "learning_rate": 2.5123893305113525e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9030 + }, + { + "epoch": 2.630200756473669, + "flow_matching_loss": 0.016153452359139918, + "grad_norm": 0.0733814686536789, + "learning_rate": 2.5111686447448997e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9040 + }, + { + "epoch": 2.633110270584812, + "flow_matching_loss": 0.016378199961036444, + "grad_norm": 0.07076900452375412, + "learning_rate": 2.5099467303131966e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9050 + }, + { + "epoch": 2.636019784695956, + "flow_matching_loss": 0.015757268853485586, + "grad_norm": 0.08127396553754807, + "learning_rate": 2.508723588700984e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9060 + }, + { + "epoch": 2.638929298807099, + "flow_matching_loss": 0.016762963403016327, + "grad_norm": 0.07462532818317413, + "learning_rate": 2.5074992213944952e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9070 + }, + { + "epoch": 2.641838812918243, + "flow_matching_loss": 0.015975463017821313, + "grad_norm": 0.07611680775880814, + "learning_rate": 2.506273629881451e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9080 + }, + { + "epoch": 2.644748327029386, + "flow_matching_loss": 0.01802010489627719, + "grad_norm": 0.09511147439479828, + "learning_rate": 2.5050468156510617e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9090 + }, + { + "epoch": 2.6476578411405294, + "flow_matching_loss": 0.015333410911262035, + "grad_norm": 0.07142185419797897, + "learning_rate": 2.5038187801940215e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9100 + }, + { + "epoch": 2.650567355251673, + "flow_matching_loss": 0.01646627690643072, + "grad_norm": 0.06762517988681793, + "learning_rate": 2.5025895250025088e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9110 + }, + { + "epoch": 2.6534768693628163, + "flow_matching_loss": 0.017116518784314395, + "grad_norm": 0.09021764993667603, + "learning_rate": 2.501359051570186e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9120 + }, + { + "epoch": 2.6563863834739596, + "flow_matching_loss": 0.017094652727246286, + "grad_norm": 0.0906793549656868, + "learning_rate": 2.500127361392193e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9130 + }, + { + "epoch": 2.6592958975851033, + "flow_matching_loss": 0.01605583503842354, + "grad_norm": 0.08370396494865417, + "learning_rate": 2.49889445596515e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9140 + }, + { + "epoch": 2.6622054116962466, + "flow_matching_loss": 0.01654166504740715, + "grad_norm": 0.07795020192861557, + "learning_rate": 2.497660336787154e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9150 + }, + { + "epoch": 2.6651149258073903, + "flow_matching_loss": 0.016135876625776292, + "grad_norm": 0.09685271233320236, + "learning_rate": 2.496425005357775e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9160 + }, + { + "epoch": 2.6680244399185336, + "flow_matching_loss": 0.017911757715046404, + "grad_norm": 0.1021728366613388, + "learning_rate": 2.4951884631780578e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9170 + }, + { + "epoch": 2.6709339540296773, + "flow_matching_loss": 0.015861476492136716, + "grad_norm": 0.056547317653894424, + "learning_rate": 2.493950711750518e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9180 + }, + { + "epoch": 2.6738434681408205, + "flow_matching_loss": 0.018633324094116686, + "grad_norm": 0.0919359028339386, + "learning_rate": 2.49271175257914e-05, + "loss": 0.0186, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9190 + }, + { + "epoch": 2.676752982251964, + "flow_matching_loss": 0.017206464894115925, + "grad_norm": 0.08675181865692139, + "learning_rate": 2.4914715871693757e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9200 + }, + { + "epoch": 2.6796624963631075, + "flow_matching_loss": 0.017534404899924992, + "grad_norm": 0.09158682823181152, + "learning_rate": 2.4902302170281438e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9210 + }, + { + "epoch": 2.6825720104742508, + "flow_matching_loss": 0.017005362547934055, + "grad_norm": 0.09906627982854843, + "learning_rate": 2.4889876436638252e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9220 + }, + { + "epoch": 2.685481524585394, + "flow_matching_loss": 0.016748409159481524, + "grad_norm": 0.09424477070569992, + "learning_rate": 2.4877438685862643e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9230 + }, + { + "epoch": 2.6883910386965377, + "flow_matching_loss": 0.01537421317771077, + "grad_norm": 0.07534545660018921, + "learning_rate": 2.486498893306765e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9240 + }, + { + "epoch": 2.691300552807681, + "flow_matching_loss": 0.016430896148085593, + "grad_norm": 0.08118598908185959, + "learning_rate": 2.485252719338089e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9250 + }, + { + "epoch": 2.6942100669188247, + "flow_matching_loss": 0.01737104495987296, + "grad_norm": 0.08936554938554764, + "learning_rate": 2.4840053481944563e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9260 + }, + { + "epoch": 2.697119581029968, + "flow_matching_loss": 0.01605172045528889, + "grad_norm": 0.09989416599273682, + "learning_rate": 2.4827567813915398e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9270 + }, + { + "epoch": 2.7000290951411117, + "flow_matching_loss": 0.017455101199448108, + "grad_norm": 0.08083780854940414, + "learning_rate": 2.481507020446466e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9280 + }, + { + "epoch": 2.702938609252255, + "flow_matching_loss": 0.01657123425975442, + "grad_norm": 0.06923859566450119, + "learning_rate": 2.4802560668778122e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9290 + }, + { + "epoch": 2.705848123363398, + "flow_matching_loss": 0.015824701823294162, + "grad_norm": 0.08688946068286896, + "learning_rate": 2.479003922205605e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9300 + }, + { + "epoch": 2.708757637474542, + "flow_matching_loss": 0.01647872356697917, + "grad_norm": 0.07314660400152206, + "learning_rate": 2.4777505879513183e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9310 + }, + { + "epoch": 2.711667151585685, + "flow_matching_loss": 0.015446212142705917, + "grad_norm": 0.0827651172876358, + "learning_rate": 2.4764960656378714e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9320 + }, + { + "epoch": 2.7145766656968284, + "flow_matching_loss": 0.014446275122463704, + "grad_norm": 0.09494772553443909, + "learning_rate": 2.4752403567896274e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9330 + }, + { + "epoch": 2.717486179807972, + "flow_matching_loss": 0.014903487078845501, + "grad_norm": 0.09653677046298981, + "learning_rate": 2.47398346293239e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9340 + }, + { + "epoch": 2.7203956939191154, + "flow_matching_loss": 0.016699534375220537, + "grad_norm": 0.0915675163269043, + "learning_rate": 2.4727253855934055e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9350 + }, + { + "epoch": 2.723305208030259, + "flow_matching_loss": 0.016165184322744608, + "grad_norm": 0.07023651897907257, + "learning_rate": 2.4714661263013548e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9360 + }, + { + "epoch": 2.7262147221414024, + "flow_matching_loss": 0.01635450990870595, + "grad_norm": 0.10185030102729797, + "learning_rate": 2.4702056865863575e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9370 + }, + { + "epoch": 2.729124236252546, + "flow_matching_loss": 0.016765501443296672, + "grad_norm": 0.07673486322164536, + "learning_rate": 2.4689440679799672e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9380 + }, + { + "epoch": 2.7320337503636893, + "flow_matching_loss": 0.015447215456515551, + "grad_norm": 0.08862485736608505, + "learning_rate": 2.4676812720151678e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9390 + }, + { + "epoch": 2.7349432644748326, + "flow_matching_loss": 0.015878223162144423, + "grad_norm": 0.06391225010156631, + "learning_rate": 2.466417300226377e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9400 + }, + { + "epoch": 2.7378527785859763, + "flow_matching_loss": 0.01648207297548652, + "grad_norm": 0.08467114716768265, + "learning_rate": 2.4651521541494394e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9410 + }, + { + "epoch": 2.7407622926971196, + "flow_matching_loss": 0.017100695054978132, + "grad_norm": 0.09071175754070282, + "learning_rate": 2.4638858353216267e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9420 + }, + { + "epoch": 2.743671806808263, + "flow_matching_loss": 0.017394664511084557, + "grad_norm": 0.10810456424951553, + "learning_rate": 2.4626183452816355e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9430 + }, + { + "epoch": 2.7465813209194065, + "flow_matching_loss": 0.015240325871855021, + "grad_norm": 0.08969663828611374, + "learning_rate": 2.4613496855695855e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9440 + }, + { + "epoch": 2.74949083503055, + "flow_matching_loss": 0.014784020092338323, + "grad_norm": 0.0983208566904068, + "learning_rate": 2.4600798577270183e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9450 + }, + { + "epoch": 2.752400349141693, + "flow_matching_loss": 0.016529259085655213, + "grad_norm": 0.09840866923332214, + "learning_rate": 2.4588088632968944e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9460 + }, + { + "epoch": 2.7553098632528368, + "flow_matching_loss": 0.015878135990351438, + "grad_norm": 0.08830905705690384, + "learning_rate": 2.4575367038235916e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9470 + }, + { + "epoch": 2.7582193773639805, + "flow_matching_loss": 0.015772669203579427, + "grad_norm": 0.09171845763921738, + "learning_rate": 2.4562633808529043e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9480 + }, + { + "epoch": 2.7611288914751237, + "flow_matching_loss": 0.01696584802120924, + "grad_norm": 0.10803502053022385, + "learning_rate": 2.4549888959320396e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9490 + }, + { + "epoch": 2.764038405586267, + "flow_matching_loss": 0.01624936256557703, + "grad_norm": 0.05673908442258835, + "learning_rate": 2.4537132506096162e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9500 + }, + { + "epoch": 2.7669479196974107, + "flow_matching_loss": 0.014644589461386203, + "grad_norm": 0.1001754179596901, + "learning_rate": 2.452436446435664e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9510 + }, + { + "epoch": 2.769857433808554, + "flow_matching_loss": 0.017767826933413743, + "grad_norm": 0.08796373754739761, + "learning_rate": 2.451158484961621e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9520 + }, + { + "epoch": 2.7727669479196972, + "flow_matching_loss": 0.018013802357017994, + "grad_norm": 0.09749230742454529, + "learning_rate": 2.4498793677403304e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9530 + }, + { + "epoch": 2.775676462030841, + "flow_matching_loss": 0.017599573824554682, + "grad_norm": 0.09031965583562851, + "learning_rate": 2.4485990963260398e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9540 + }, + { + "epoch": 2.778585976141984, + "flow_matching_loss": 0.01677909353747964, + "grad_norm": 0.08946201950311661, + "learning_rate": 2.4473176722744e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9550 + }, + { + "epoch": 2.7814954902531275, + "flow_matching_loss": 0.018485461454838516, + "grad_norm": 0.08658052235841751, + "learning_rate": 2.446035097142463e-05, + "loss": 0.0185, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9560 + }, + { + "epoch": 2.784405004364271, + "flow_matching_loss": 0.017268637660890818, + "grad_norm": 0.08780259639024734, + "learning_rate": 2.444751372488677e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9570 + }, + { + "epoch": 2.7873145184754144, + "flow_matching_loss": 0.018929189164191486, + "grad_norm": 0.06488753110170364, + "learning_rate": 2.443466499872889e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9580 + }, + { + "epoch": 2.790224032586558, + "flow_matching_loss": 0.014778747409582137, + "grad_norm": 0.0704764574766159, + "learning_rate": 2.4421804808563413e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9590 + }, + { + "epoch": 2.7931335466977014, + "flow_matching_loss": 0.019173683412373067, + "grad_norm": 0.09431754797697067, + "learning_rate": 2.440893317001667e-05, + "loss": 0.0192, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9600 + }, + { + "epoch": 2.796043060808845, + "flow_matching_loss": 0.015119770728051662, + "grad_norm": 0.07841940224170685, + "learning_rate": 2.4396050098728925e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9610 + }, + { + "epoch": 2.7989525749199884, + "flow_matching_loss": 0.01604356514289975, + "grad_norm": 0.07723134011030197, + "learning_rate": 2.4383155610354314e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9620 + }, + { + "epoch": 2.8018620890311317, + "flow_matching_loss": 0.018877202458679677, + "grad_norm": 0.10199473053216934, + "learning_rate": 2.437024972056087e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9630 + }, + { + "epoch": 2.8047716031422754, + "flow_matching_loss": 0.015736716613173485, + "grad_norm": 0.09293261170387268, + "learning_rate": 2.4357332445030452e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9640 + }, + { + "epoch": 2.8076811172534186, + "flow_matching_loss": 0.01730595203116536, + "grad_norm": 0.09926071763038635, + "learning_rate": 2.434440379945878e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9650 + }, + { + "epoch": 2.810590631364562, + "flow_matching_loss": 0.017299339175224304, + "grad_norm": 0.12462867796421051, + "learning_rate": 2.4331463799555366e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9660 + }, + { + "epoch": 2.8135001454757056, + "flow_matching_loss": 0.015367075242102146, + "grad_norm": 0.07535947114229202, + "learning_rate": 2.4318512461043544e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9670 + }, + { + "epoch": 2.816409659586849, + "flow_matching_loss": 0.015377208683639764, + "grad_norm": 0.12442696839570999, + "learning_rate": 2.4305549799660397e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9680 + }, + { + "epoch": 2.8193191736979926, + "flow_matching_loss": 0.01656598299741745, + "grad_norm": 0.0962383970618248, + "learning_rate": 2.429257583115679e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9690 + }, + { + "epoch": 2.822228687809136, + "flow_matching_loss": 0.01753973327577114, + "grad_norm": 0.08245345950126648, + "learning_rate": 2.4279590571297316e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9700 + }, + { + "epoch": 2.8251382019202795, + "flow_matching_loss": 0.016294135339558126, + "grad_norm": 0.08431974053382874, + "learning_rate": 2.4266594035860288e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9710 + }, + { + "epoch": 2.828047716031423, + "flow_matching_loss": 0.01655037123709917, + "grad_norm": 0.08962531387805939, + "learning_rate": 2.425358624063773e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9720 + }, + { + "epoch": 2.830957230142566, + "flow_matching_loss": 0.014459655061364174, + "grad_norm": 0.08794696629047394, + "learning_rate": 2.424056720143532e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9730 + }, + { + "epoch": 2.8338667442537098, + "flow_matching_loss": 0.015442301612347364, + "grad_norm": 0.0685071125626564, + "learning_rate": 2.422753693407244e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9740 + }, + { + "epoch": 2.836776258364853, + "flow_matching_loss": 0.015096234437078238, + "grad_norm": 0.10656712204217911, + "learning_rate": 2.421449545438209e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9750 + }, + { + "epoch": 2.8396857724759963, + "flow_matching_loss": 0.01843647612258792, + "grad_norm": 0.07583649456501007, + "learning_rate": 2.4201442778210887e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9760 + }, + { + "epoch": 2.84259528658714, + "flow_matching_loss": 0.01676537599414587, + "grad_norm": 0.08063670247793198, + "learning_rate": 2.418837892141907e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9770 + }, + { + "epoch": 2.8455048006982833, + "flow_matching_loss": 0.016253407578915358, + "grad_norm": 0.06844998896121979, + "learning_rate": 2.4175303899880458e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9780 + }, + { + "epoch": 2.848414314809427, + "flow_matching_loss": 0.01630002735182643, + "grad_norm": 0.07681822031736374, + "learning_rate": 2.4162217729482427e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9790 + }, + { + "epoch": 2.8513238289205702, + "flow_matching_loss": 0.014875825308263303, + "grad_norm": 0.10326908528804779, + "learning_rate": 2.4149120426125918e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9800 + }, + { + "epoch": 2.854233343031714, + "flow_matching_loss": 0.01507036704570055, + "grad_norm": 0.07617391645908356, + "learning_rate": 2.413601200572538e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9810 + }, + { + "epoch": 2.857142857142857, + "flow_matching_loss": 0.015854917094111444, + "grad_norm": 0.08670378476381302, + "learning_rate": 2.4122892484208794e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9820 + }, + { + "epoch": 2.8600523712540005, + "flow_matching_loss": 0.01706076432019472, + "grad_norm": 0.09119105339050293, + "learning_rate": 2.4109761877517603e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9830 + }, + { + "epoch": 2.862961885365144, + "flow_matching_loss": 0.017240438889712096, + "grad_norm": 0.08980844914913177, + "learning_rate": 2.4096620201606737e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9840 + }, + { + "epoch": 2.8658713994762874, + "flow_matching_loss": 0.014663816709071398, + "grad_norm": 0.06786753982305527, + "learning_rate": 2.4083467472444573e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9850 + }, + { + "epoch": 2.8687809135874307, + "flow_matching_loss": 0.014959205873310566, + "grad_norm": 0.07404063642024994, + "learning_rate": 2.4070303706012912e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9860 + }, + { + "epoch": 2.8716904276985744, + "flow_matching_loss": 0.016311419662088156, + "grad_norm": 0.09461662918329239, + "learning_rate": 2.4057128918306975e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9870 + }, + { + "epoch": 2.8745999418097177, + "flow_matching_loss": 0.015203474089503288, + "grad_norm": 0.08937881141901016, + "learning_rate": 2.4043943125335373e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9880 + }, + { + "epoch": 2.8775094559208614, + "flow_matching_loss": 0.017333436757326126, + "grad_norm": 0.10069386661052704, + "learning_rate": 2.403074634312008e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9890 + }, + { + "epoch": 2.8804189700320046, + "flow_matching_loss": 0.016933331917971372, + "grad_norm": 0.08849085122346878, + "learning_rate": 2.4017538587696442e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9900 + }, + { + "epoch": 2.8833284841431484, + "flow_matching_loss": 0.015307815559208394, + "grad_norm": 0.07838859409093857, + "learning_rate": 2.4004319875113118e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9910 + }, + { + "epoch": 2.8862379982542916, + "flow_matching_loss": 0.016217452939599752, + "grad_norm": 0.0823657214641571, + "learning_rate": 2.3991090221432096e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9920 + }, + { + "epoch": 2.889147512365435, + "flow_matching_loss": 0.016659653466194867, + "grad_norm": 0.08140718191862106, + "learning_rate": 2.397784964272865e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9930 + }, + { + "epoch": 2.8920570264765786, + "flow_matching_loss": 0.0163639304228127, + "grad_norm": 0.08379080891609192, + "learning_rate": 2.3964598155091335e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9940 + }, + { + "epoch": 2.894966540587722, + "flow_matching_loss": 0.01676495922729373, + "grad_norm": 0.08751889318227768, + "learning_rate": 2.3951335774621952e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9950 + }, + { + "epoch": 2.897876054698865, + "flow_matching_loss": 0.0174841882660985, + "grad_norm": 0.07675199955701828, + "learning_rate": 2.3938062517435548e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9960 + }, + { + "epoch": 2.900785568810009, + "flow_matching_loss": 0.018901466857641935, + "grad_norm": 0.07935124635696411, + "learning_rate": 2.392477839966038e-05, + "loss": 0.0189, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9970 + }, + { + "epoch": 2.903695082921152, + "flow_matching_loss": 0.016380308009684087, + "grad_norm": 0.10204409062862396, + "learning_rate": 2.3911483437437904e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9980 + }, + { + "epoch": 2.9066045970322953, + "flow_matching_loss": 0.015777745563536883, + "grad_norm": 0.09291962534189224, + "learning_rate": 2.3898177646922757e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 9990 + }, + { + "epoch": 2.909514111143439, + "flow_matching_loss": 0.014693605341017246, + "grad_norm": 0.07019583880901337, + "learning_rate": 2.3884861044282722e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10000 + }, + { + "epoch": 2.9124236252545828, + "flow_matching_loss": 0.015012552589178085, + "grad_norm": 0.09365615248680115, + "learning_rate": 2.3871533645698732e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10010 + }, + { + "epoch": 2.915333139365726, + "flow_matching_loss": 0.01574408421292901, + "grad_norm": 0.06768190860748291, + "learning_rate": 2.3858195467364833e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10020 + }, + { + "epoch": 2.9182426534768693, + "flow_matching_loss": 0.014026197046041489, + "grad_norm": 0.06871765851974487, + "learning_rate": 2.3844846525488166e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10030 + }, + { + "epoch": 2.921152167588013, + "flow_matching_loss": 0.015159650892019271, + "grad_norm": 0.07114538550376892, + "learning_rate": 2.3831486836288957e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10040 + }, + { + "epoch": 2.9240616816991563, + "flow_matching_loss": 0.015581926424056291, + "grad_norm": 0.0827697366476059, + "learning_rate": 2.381811641600048e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10050 + }, + { + "epoch": 2.9269711958102995, + "flow_matching_loss": 0.016198268067091702, + "grad_norm": 0.08216728270053864, + "learning_rate": 2.3804735280869073e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10060 + }, + { + "epoch": 2.9298807099214432, + "flow_matching_loss": 0.015505132637917996, + "grad_norm": 0.06696473062038422, + "learning_rate": 2.3791343447154064e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10070 + }, + { + "epoch": 2.9327902240325865, + "flow_matching_loss": 0.015528784785419702, + "grad_norm": 0.06961639970541, + "learning_rate": 2.37779409311278e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10080 + }, + { + "epoch": 2.9356997381437298, + "flow_matching_loss": 0.016824864875525238, + "grad_norm": 0.09744373708963394, + "learning_rate": 2.3764527749075596e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10090 + }, + { + "epoch": 2.9386092522548735, + "flow_matching_loss": 0.016235892847180366, + "grad_norm": 0.11372315138578415, + "learning_rate": 2.3751103917295735e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10100 + }, + { + "epoch": 2.9415187663660167, + "flow_matching_loss": 0.018027981743216515, + "grad_norm": 0.09162444621324539, + "learning_rate": 2.373766945209944e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10110 + }, + { + "epoch": 2.9444282804771604, + "flow_matching_loss": 0.017611121851950885, + "grad_norm": 0.09258407354354858, + "learning_rate": 2.3724224369810856e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10120 + }, + { + "epoch": 2.9473377945883037, + "flow_matching_loss": 0.01683004153892398, + "grad_norm": 0.12609772384166718, + "learning_rate": 2.3710768686767015e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10130 + }, + { + "epoch": 2.9502473086994474, + "flow_matching_loss": 0.01585523076355457, + "grad_norm": 0.07342829555273056, + "learning_rate": 2.3697302419317856e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10140 + }, + { + "epoch": 2.9531568228105907, + "flow_matching_loss": 0.01643197564408183, + "grad_norm": 0.12254830449819565, + "learning_rate": 2.3683825583826146e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10150 + }, + { + "epoch": 2.956066336921734, + "flow_matching_loss": 0.015601888578385115, + "grad_norm": 0.07087606191635132, + "learning_rate": 2.3670338196667528e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10160 + }, + { + "epoch": 2.9589758510328776, + "flow_matching_loss": 0.014826248958706855, + "grad_norm": 0.08505092561244965, + "learning_rate": 2.3656840274230443e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10170 + }, + { + "epoch": 2.961885365144021, + "flow_matching_loss": 0.01586463861167431, + "grad_norm": 0.0915859192609787, + "learning_rate": 2.3643331832916133e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10180 + }, + { + "epoch": 2.964794879255164, + "flow_matching_loss": 0.01833362439647317, + "grad_norm": 0.09736472368240356, + "learning_rate": 2.362981288913863e-05, + "loss": 0.0183, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10190 + }, + { + "epoch": 2.967704393366308, + "flow_matching_loss": 0.01643246617168188, + "grad_norm": 0.09396586567163467, + "learning_rate": 2.3616283459324737e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10200 + }, + { + "epoch": 2.970613907477451, + "flow_matching_loss": 0.014671270921826362, + "grad_norm": 0.08434142172336578, + "learning_rate": 2.3602743559913973e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10210 + }, + { + "epoch": 2.973523421588595, + "flow_matching_loss": 0.015064767841249705, + "grad_norm": 0.09536068141460419, + "learning_rate": 2.35891932073586e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10220 + }, + { + "epoch": 2.976432935699738, + "flow_matching_loss": 0.016161991842091085, + "grad_norm": 0.13074712455272675, + "learning_rate": 2.357563241812357e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10230 + }, + { + "epoch": 2.979342449810882, + "flow_matching_loss": 0.016015428770333528, + "grad_norm": 0.08563913404941559, + "learning_rate": 2.3562061208686522e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10240 + }, + { + "epoch": 2.982251963922025, + "flow_matching_loss": 0.01646116180345416, + "grad_norm": 0.09376879036426544, + "learning_rate": 2.354847959553776e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10250 + }, + { + "epoch": 2.9851614780331683, + "flow_matching_loss": 0.01541334968060255, + "grad_norm": 0.09063588827848434, + "learning_rate": 2.353488759518022e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10260 + }, + { + "epoch": 2.988070992144312, + "flow_matching_loss": 0.015784395672380925, + "grad_norm": 0.08073409646749496, + "learning_rate": 2.352128522412946e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10270 + }, + { + "epoch": 2.9909805062554553, + "flow_matching_loss": 0.015573166776448488, + "grad_norm": 0.0770811140537262, + "learning_rate": 2.3507672498913653e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10280 + }, + { + "epoch": 2.9938900203665986, + "flow_matching_loss": 0.01614592904224992, + "grad_norm": 0.09535417705774307, + "learning_rate": 2.3494049436073537e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10290 + }, + { + "epoch": 2.9967995344777423, + "flow_matching_loss": 0.016268288623541592, + "grad_norm": 0.07732424885034561, + "learning_rate": 2.348041605216242e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10300 + }, + { + "epoch": 2.9997090485888855, + "flow_matching_loss": 0.016140823438763618, + "grad_norm": 0.09712300449609756, + "learning_rate": 2.3466772363746146e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10310 + }, + { + "epoch": 3.0026185627000292, + "flow_matching_loss": 0.015530548151582479, + "grad_norm": 0.08525227010250092, + "learning_rate": 2.345311838740309e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10320 + }, + { + "epoch": 3.0055280768111725, + "flow_matching_loss": 0.015000639762729407, + "grad_norm": 0.09672493487596512, + "learning_rate": 2.343945413972411e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10330 + }, + { + "epoch": 3.0084375909223158, + "flow_matching_loss": 0.015281609632074833, + "grad_norm": 0.07962464541196823, + "learning_rate": 2.3425779637312568e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10340 + }, + { + "epoch": 3.0113471050334595, + "flow_matching_loss": 0.01663999482989311, + "grad_norm": 0.08597230166196823, + "learning_rate": 2.3412094896784264e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10350 + }, + { + "epoch": 3.0142566191446027, + "flow_matching_loss": 0.01767744841054082, + "grad_norm": 0.08642803132534027, + "learning_rate": 2.339839993476745e-05, + "loss": 0.0177, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10360 + }, + { + "epoch": 3.0171661332557465, + "flow_matching_loss": 0.01609700610861182, + "grad_norm": 0.09332061558961868, + "learning_rate": 2.3384694767902804e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10370 + }, + { + "epoch": 3.0200756473668897, + "flow_matching_loss": 0.013702133577317, + "grad_norm": 0.07087881118059158, + "learning_rate": 2.3370979412843384e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10380 + }, + { + "epoch": 3.022985161478033, + "flow_matching_loss": 0.015010550059378147, + "grad_norm": 0.06163318455219269, + "learning_rate": 2.3357253886254648e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10390 + }, + { + "epoch": 3.0258946755891767, + "flow_matching_loss": 0.014093468617647886, + "grad_norm": 0.0791102945804596, + "learning_rate": 2.3343518204814407e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10400 + }, + { + "epoch": 3.02880418970032, + "flow_matching_loss": 0.015304413996636868, + "grad_norm": 0.07194823771715164, + "learning_rate": 2.3329772385212798e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10410 + }, + { + "epoch": 3.0317137038114637, + "flow_matching_loss": 0.015646661072969435, + "grad_norm": 0.07858846336603165, + "learning_rate": 2.3316016444152296e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10420 + }, + { + "epoch": 3.034623217922607, + "flow_matching_loss": 0.016068592108786105, + "grad_norm": 0.09775222092866898, + "learning_rate": 2.3302250398347668e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10430 + }, + { + "epoch": 3.03753273203375, + "flow_matching_loss": 0.01571628376841545, + "grad_norm": 0.0873384103178978, + "learning_rate": 2.328847426452595e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10440 + }, + { + "epoch": 3.040442246144894, + "flow_matching_loss": 0.015646505169570446, + "grad_norm": 0.08170320093631744, + "learning_rate": 2.3274688059426456e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10450 + }, + { + "epoch": 3.043351760256037, + "flow_matching_loss": 0.016881988290697335, + "grad_norm": 0.07291057705879211, + "learning_rate": 2.3260891799800715e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10460 + }, + { + "epoch": 3.046261274367181, + "flow_matching_loss": 0.016429992951452733, + "grad_norm": 0.09166331589221954, + "learning_rate": 2.3247085502412482e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10470 + }, + { + "epoch": 3.049170788478324, + "flow_matching_loss": 0.015431780461221933, + "grad_norm": 0.08466377854347229, + "learning_rate": 2.3233269184037717e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10480 + }, + { + "epoch": 3.0520803025894674, + "flow_matching_loss": 0.017797209415584804, + "grad_norm": 0.09280207008123398, + "learning_rate": 2.3219442861464547e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10490 + }, + { + "epoch": 3.054989816700611, + "flow_matching_loss": 0.015216343849897385, + "grad_norm": 0.08192425966262817, + "learning_rate": 2.3205606551493255e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10500 + }, + { + "epoch": 3.0578993308117544, + "flow_matching_loss": 0.01639921758323908, + "grad_norm": 0.07944244146347046, + "learning_rate": 2.319176027093627e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10510 + }, + { + "epoch": 3.060808844922898, + "flow_matching_loss": 0.01667764214798808, + "grad_norm": 0.07541355490684509, + "learning_rate": 2.3177904036618117e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10520 + }, + { + "epoch": 3.0637183590340413, + "flow_matching_loss": 0.01531137190759182, + "grad_norm": 0.07690460234880447, + "learning_rate": 2.316403786537544e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10530 + }, + { + "epoch": 3.0666278731451846, + "flow_matching_loss": 0.01667955182492733, + "grad_norm": 0.09209548681974411, + "learning_rate": 2.3150161774056933e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10540 + }, + { + "epoch": 3.0695373872563283, + "flow_matching_loss": 0.01457892283797264, + "grad_norm": 0.0780811533331871, + "learning_rate": 2.313627577952336e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10550 + }, + { + "epoch": 3.0724469013674716, + "flow_matching_loss": 0.014802695065736771, + "grad_norm": 0.09389405697584152, + "learning_rate": 2.312237989864752e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10560 + }, + { + "epoch": 3.0753564154786153, + "flow_matching_loss": 0.017006196081638336, + "grad_norm": 0.09417407959699631, + "learning_rate": 2.3108474148314205e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10570 + }, + { + "epoch": 3.0782659295897585, + "flow_matching_loss": 0.015392145700752735, + "grad_norm": 0.08484616875648499, + "learning_rate": 2.309455854542023e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10580 + }, + { + "epoch": 3.081175443700902, + "flow_matching_loss": 0.015228810627013445, + "grad_norm": 0.08838316798210144, + "learning_rate": 2.308063310687435e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10590 + }, + { + "epoch": 3.0840849578120455, + "flow_matching_loss": 0.017644286341965198, + "grad_norm": 0.07586672902107239, + "learning_rate": 2.306669784959729e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10600 + }, + { + "epoch": 3.0869944719231888, + "flow_matching_loss": 0.015515133552253247, + "grad_norm": 0.09400155395269394, + "learning_rate": 2.3052752790521703e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10610 + }, + { + "epoch": 3.0899039860343325, + "flow_matching_loss": 0.01581016257405281, + "grad_norm": 0.09447593241930008, + "learning_rate": 2.3038797946592152e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10620 + }, + { + "epoch": 3.0928135001454757, + "flow_matching_loss": 0.015492994152009488, + "grad_norm": 0.08281589299440384, + "learning_rate": 2.3024833334765087e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10630 + }, + { + "epoch": 3.095723014256619, + "flow_matching_loss": 0.014539913181215525, + "grad_norm": 0.06709302961826324, + "learning_rate": 2.301085897200883e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10640 + }, + { + "epoch": 3.0986325283677627, + "flow_matching_loss": 0.015493208914995194, + "grad_norm": 0.0715959295630455, + "learning_rate": 2.299687487530354e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10650 + }, + { + "epoch": 3.101542042478906, + "flow_matching_loss": 0.015957160387188196, + "grad_norm": 0.0760168731212616, + "learning_rate": 2.2982881061641228e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10660 + }, + { + "epoch": 3.1044515565900497, + "flow_matching_loss": 0.016205344721674918, + "grad_norm": 0.07975756376981735, + "learning_rate": 2.296887754802569e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10670 + }, + { + "epoch": 3.107361070701193, + "flow_matching_loss": 0.016544552613049747, + "grad_norm": 0.06024974584579468, + "learning_rate": 2.295486435147251e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10680 + }, + { + "epoch": 3.110270584812336, + "flow_matching_loss": 0.015459001902490855, + "grad_norm": 0.06071052700281143, + "learning_rate": 2.294084148900905e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10690 + }, + { + "epoch": 3.11318009892348, + "flow_matching_loss": 0.015225035324692725, + "grad_norm": 0.08121990412473679, + "learning_rate": 2.292680897767441e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10700 + }, + { + "epoch": 3.116089613034623, + "flow_matching_loss": 0.01657073274254799, + "grad_norm": 0.0827622190117836, + "learning_rate": 2.291276683451941e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10710 + }, + { + "epoch": 3.1189991271457664, + "flow_matching_loss": 0.015524777304381131, + "grad_norm": 0.0974397286772728, + "learning_rate": 2.2898715076606584e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10720 + }, + { + "epoch": 3.12190864125691, + "flow_matching_loss": 0.013268745969980955, + "grad_norm": 0.06868566572666168, + "learning_rate": 2.2884653721010135e-05, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10730 + }, + { + "epoch": 3.1248181553680534, + "flow_matching_loss": 0.01702647628262639, + "grad_norm": 0.09135475009679794, + "learning_rate": 2.2870582784815946e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10740 + }, + { + "epoch": 3.127727669479197, + "flow_matching_loss": 0.0147713971324265, + "grad_norm": 0.05529720336198807, + "learning_rate": 2.2856502285121525e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10750 + }, + { + "epoch": 3.1306371835903404, + "flow_matching_loss": 0.01686782529577613, + "grad_norm": 0.0939682200551033, + "learning_rate": 2.2842412239036004e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10760 + }, + { + "epoch": 3.133546697701484, + "flow_matching_loss": 0.015391272306442261, + "grad_norm": 0.08144635707139969, + "learning_rate": 2.2828312663680125e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10770 + }, + { + "epoch": 3.1364562118126273, + "flow_matching_loss": 0.01644293647259474, + "grad_norm": 0.08846632391214371, + "learning_rate": 2.2814203576186194e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10780 + }, + { + "epoch": 3.1393657259237706, + "flow_matching_loss": 0.015537980012595654, + "grad_norm": 0.06525404751300812, + "learning_rate": 2.2800084993698093e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10790 + }, + { + "epoch": 3.1422752400349143, + "flow_matching_loss": 0.015313633158802986, + "grad_norm": 0.08752402663230896, + "learning_rate": 2.2785956933371224e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10800 + }, + { + "epoch": 3.1451847541460576, + "flow_matching_loss": 0.01654322203248739, + "grad_norm": 0.0727604329586029, + "learning_rate": 2.2771819412372505e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10810 + }, + { + "epoch": 3.148094268257201, + "flow_matching_loss": 0.01467289524152875, + "grad_norm": 0.0652417317032814, + "learning_rate": 2.2757672447880374e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10820 + }, + { + "epoch": 3.1510037823683446, + "flow_matching_loss": 0.01585540296509862, + "grad_norm": 0.08802733570337296, + "learning_rate": 2.2743516057084712e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10830 + }, + { + "epoch": 3.153913296479488, + "flow_matching_loss": 0.015626773796975614, + "grad_norm": 0.08728941529989243, + "learning_rate": 2.2729350257186877e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10840 + }, + { + "epoch": 3.1568228105906315, + "flow_matching_loss": 0.015088558290153741, + "grad_norm": 0.09237043559551239, + "learning_rate": 2.2715175065399655e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10850 + }, + { + "epoch": 3.159732324701775, + "flow_matching_loss": 0.014962168876081705, + "grad_norm": 0.10682081431150436, + "learning_rate": 2.270099049894723e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10860 + }, + { + "epoch": 3.162641838812918, + "flow_matching_loss": 0.015765823423862457, + "grad_norm": 0.07975596934556961, + "learning_rate": 2.268679657506519e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10870 + }, + { + "epoch": 3.1655513529240618, + "flow_matching_loss": 0.01567065268754959, + "grad_norm": 0.09431754052639008, + "learning_rate": 2.2672593311000498e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10880 + }, + { + "epoch": 3.168460867035205, + "flow_matching_loss": 0.015189000125974416, + "grad_norm": 0.07538163661956787, + "learning_rate": 2.2658380724011452e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10890 + }, + { + "epoch": 3.1713703811463487, + "flow_matching_loss": 0.014898003917187453, + "grad_norm": 0.07576432824134827, + "learning_rate": 2.264415883136769e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10900 + }, + { + "epoch": 3.174279895257492, + "flow_matching_loss": 0.016239275224506855, + "grad_norm": 0.08816496282815933, + "learning_rate": 2.262992765035015e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10910 + }, + { + "epoch": 3.1771894093686353, + "flow_matching_loss": 0.015209307242184878, + "grad_norm": 0.10077665001153946, + "learning_rate": 2.2615687198251062e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10920 + }, + { + "epoch": 3.180098923479779, + "flow_matching_loss": 0.01659329645335674, + "grad_norm": 0.07828088104724884, + "learning_rate": 2.2601437492373914e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10930 + }, + { + "epoch": 3.1830084375909222, + "flow_matching_loss": 0.016919676586985587, + "grad_norm": 0.06607142835855484, + "learning_rate": 2.2587178550033444e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10940 + }, + { + "epoch": 3.185917951702066, + "flow_matching_loss": 0.01531932232901454, + "grad_norm": 0.06744671612977982, + "learning_rate": 2.2572910388555615e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10950 + }, + { + "epoch": 3.188827465813209, + "flow_matching_loss": 0.015126600116491317, + "grad_norm": 0.07025310397148132, + "learning_rate": 2.255863302527758e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10960 + }, + { + "epoch": 3.1917369799243525, + "flow_matching_loss": 0.017486049607396126, + "grad_norm": 0.08038453757762909, + "learning_rate": 2.2544346477547685e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10970 + }, + { + "epoch": 3.194646494035496, + "flow_matching_loss": 0.013477479945868254, + "grad_norm": 0.07414896786212921, + "learning_rate": 2.253005076272544e-05, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10980 + }, + { + "epoch": 3.1975560081466394, + "flow_matching_loss": 0.017282843589782715, + "grad_norm": 0.1045636385679245, + "learning_rate": 2.2515745898181477e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 10990 + }, + { + "epoch": 3.200465522257783, + "flow_matching_loss": 0.01702231392264366, + "grad_norm": 0.0807872861623764, + "learning_rate": 2.250143190129756e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11000 + }, + { + "epoch": 3.2033750363689264, + "flow_matching_loss": 0.01577789057046175, + "grad_norm": 0.09919263422489166, + "learning_rate": 2.2487108789466547e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11010 + }, + { + "epoch": 3.2062845504800697, + "flow_matching_loss": 0.01566901532933116, + "grad_norm": 0.07509168982505798, + "learning_rate": 2.247277658009236e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11020 + }, + { + "epoch": 3.2091940645912134, + "flow_matching_loss": 0.014618480019271374, + "grad_norm": 0.08487751334905624, + "learning_rate": 2.245843529059e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11030 + }, + { + "epoch": 3.2121035787023566, + "flow_matching_loss": 0.014729593880474567, + "grad_norm": 0.0931611955165863, + "learning_rate": 2.244408493838547e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11040 + }, + { + "epoch": 3.2150130928135003, + "flow_matching_loss": 0.014829451777040958, + "grad_norm": 0.05985325947403908, + "learning_rate": 2.242972554091581e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11050 + }, + { + "epoch": 3.2179226069246436, + "flow_matching_loss": 0.015509863756597042, + "grad_norm": 0.06955527514219284, + "learning_rate": 2.2415357115629045e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11060 + }, + { + "epoch": 3.220832121035787, + "flow_matching_loss": 0.015783578716218473, + "grad_norm": 0.07993081957101822, + "learning_rate": 2.2400979679984162e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11070 + }, + { + "epoch": 3.2237416351469306, + "flow_matching_loss": 0.018158215191215278, + "grad_norm": 0.09317691624164581, + "learning_rate": 2.2386593251451096e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11080 + }, + { + "epoch": 3.226651149258074, + "flow_matching_loss": 0.015982141066342593, + "grad_norm": 0.09372970461845398, + "learning_rate": 2.2372197847510722e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11090 + }, + { + "epoch": 3.2295606633692175, + "flow_matching_loss": 0.015007262583822012, + "grad_norm": 0.10607970505952835, + "learning_rate": 2.2357793485654813e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11100 + }, + { + "epoch": 3.232470177480361, + "flow_matching_loss": 0.016477022040635347, + "grad_norm": 0.06727633625268936, + "learning_rate": 2.2343380183386017e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11110 + }, + { + "epoch": 3.235379691591504, + "flow_matching_loss": 0.01497757826000452, + "grad_norm": 0.08622851222753525, + "learning_rate": 2.232895795821787e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11120 + }, + { + "epoch": 3.2382892057026478, + "flow_matching_loss": 0.014695270080119371, + "grad_norm": 0.09832178056240082, + "learning_rate": 2.2314526827674723e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11130 + }, + { + "epoch": 3.241198719813791, + "flow_matching_loss": 0.01593130109831691, + "grad_norm": 0.08771312236785889, + "learning_rate": 2.230008680929177e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11140 + }, + { + "epoch": 3.2441082339249343, + "flow_matching_loss": 0.017032976169139145, + "grad_norm": 0.07573718577623367, + "learning_rate": 2.2285637920614987e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11150 + }, + { + "epoch": 3.247017748036078, + "flow_matching_loss": 0.01483093136921525, + "grad_norm": 0.07510372996330261, + "learning_rate": 2.2271180179201143e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11160 + }, + { + "epoch": 3.2499272621472213, + "flow_matching_loss": 0.017287169490009545, + "grad_norm": 0.07174990326166153, + "learning_rate": 2.225671360261775e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11170 + }, + { + "epoch": 3.252836776258365, + "flow_matching_loss": 0.016480120364576578, + "grad_norm": 0.06329040974378586, + "learning_rate": 2.2242238208443068e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11180 + }, + { + "epoch": 3.2557462903695082, + "flow_matching_loss": 0.015432583726942539, + "grad_norm": 0.0753125250339508, + "learning_rate": 2.2227754014266063e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11190 + }, + { + "epoch": 3.258655804480652, + "flow_matching_loss": 0.01650581192225218, + "grad_norm": 0.0734996497631073, + "learning_rate": 2.22132610376864e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11200 + }, + { + "epoch": 3.261565318591795, + "flow_matching_loss": 0.017295587807893753, + "grad_norm": 0.09123699367046356, + "learning_rate": 2.219875929631441e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11210 + }, + { + "epoch": 3.2644748327029385, + "flow_matching_loss": 0.015227931365370751, + "grad_norm": 0.08374454081058502, + "learning_rate": 2.2184248807771074e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11220 + }, + { + "epoch": 3.267384346814082, + "flow_matching_loss": 0.015521557070314884, + "grad_norm": 0.06116270646452904, + "learning_rate": 2.216972958968801e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11230 + }, + { + "epoch": 3.2702938609252254, + "flow_matching_loss": 0.015325801167637109, + "grad_norm": 0.07362812757492065, + "learning_rate": 2.215520165970743e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11240 + }, + { + "epoch": 3.2732033750363687, + "flow_matching_loss": 0.014033748395740986, + "grad_norm": 0.08680620789527893, + "learning_rate": 2.2140665035482143e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11250 + }, + { + "epoch": 3.2761128891475124, + "flow_matching_loss": 0.015600281488150358, + "grad_norm": 0.07903824001550674, + "learning_rate": 2.2126119734675518e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11260 + }, + { + "epoch": 3.2790224032586557, + "flow_matching_loss": 0.015016225352883338, + "grad_norm": 0.09557031840085983, + "learning_rate": 2.2111565774961463e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11270 + }, + { + "epoch": 3.2819319173697994, + "flow_matching_loss": 0.015609236620366573, + "grad_norm": 0.07202493399381638, + "learning_rate": 2.2097003174024417e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11280 + }, + { + "epoch": 3.2848414314809427, + "flow_matching_loss": 0.015919599402695894, + "grad_norm": 0.08149658888578415, + "learning_rate": 2.2082431949559315e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11290 + }, + { + "epoch": 3.2877509455920864, + "flow_matching_loss": 0.015856925584375858, + "grad_norm": 0.09905757755041122, + "learning_rate": 2.2067852119271558e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11300 + }, + { + "epoch": 3.2906604597032296, + "flow_matching_loss": 0.016325663030147552, + "grad_norm": 0.07886487990617752, + "learning_rate": 2.2053263700877026e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11310 + }, + { + "epoch": 3.293569973814373, + "flow_matching_loss": 0.016544256079941987, + "grad_norm": 0.06999601423740387, + "learning_rate": 2.2038666712102015e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11320 + }, + { + "epoch": 3.2964794879255166, + "flow_matching_loss": 0.016369162779301404, + "grad_norm": 0.08360185474157333, + "learning_rate": 2.2024061170683246e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11330 + }, + { + "epoch": 3.29938900203666, + "flow_matching_loss": 0.017352513130754234, + "grad_norm": 0.10171817988157272, + "learning_rate": 2.200944709436783e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11340 + }, + { + "epoch": 3.302298516147803, + "flow_matching_loss": 0.016528304293751718, + "grad_norm": 0.09772343188524246, + "learning_rate": 2.1994824500913246e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11350 + }, + { + "epoch": 3.305208030258947, + "flow_matching_loss": 0.017869686894118785, + "grad_norm": 0.07630102336406708, + "learning_rate": 2.1980193408087322e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11360 + }, + { + "epoch": 3.30811754437009, + "flow_matching_loss": 0.016389637999236584, + "grad_norm": 0.08328431844711304, + "learning_rate": 2.1965553833668224e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11370 + }, + { + "epoch": 3.311027058481234, + "flow_matching_loss": 0.01576598547399044, + "grad_norm": 0.08076978474855423, + "learning_rate": 2.195090579544441e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11380 + }, + { + "epoch": 3.313936572592377, + "flow_matching_loss": 0.01562340734526515, + "grad_norm": 0.10013268142938614, + "learning_rate": 2.193624931121462e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11390 + }, + { + "epoch": 3.3168460867035203, + "flow_matching_loss": 0.015319904312491417, + "grad_norm": 0.0698503702878952, + "learning_rate": 2.1921584398787872e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11400 + }, + { + "epoch": 3.319755600814664, + "flow_matching_loss": 0.015175303351134062, + "grad_norm": 0.06953471899032593, + "learning_rate": 2.1906911075983418e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11410 + }, + { + "epoch": 3.3226651149258073, + "flow_matching_loss": 0.01349791381508112, + "grad_norm": 0.10166264325380325, + "learning_rate": 2.1892229360630728e-05, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11420 + }, + { + "epoch": 3.325574629036951, + "flow_matching_loss": 0.015225025545805692, + "grad_norm": 0.10553744435310364, + "learning_rate": 2.187753927056947e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11430 + }, + { + "epoch": 3.3284841431480943, + "flow_matching_loss": 0.016695264261215925, + "grad_norm": 0.09739246964454651, + "learning_rate": 2.186284082364949e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11440 + }, + { + "epoch": 3.3313936572592375, + "flow_matching_loss": 0.015786725282669067, + "grad_norm": 0.10237258672714233, + "learning_rate": 2.1848134037730786e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11450 + }, + { + "epoch": 3.3343031713703812, + "flow_matching_loss": 0.015703592728823424, + "grad_norm": 0.09557795524597168, + "learning_rate": 2.183341893068349e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11460 + }, + { + "epoch": 3.3372126854815245, + "flow_matching_loss": 0.015811072662472725, + "grad_norm": 0.05385303869843483, + "learning_rate": 2.1818695520387842e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11470 + }, + { + "epoch": 3.340122199592668, + "flow_matching_loss": 0.01394019676372409, + "grad_norm": 0.08661764860153198, + "learning_rate": 2.1803963824734173e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11480 + }, + { + "epoch": 3.3430317137038115, + "flow_matching_loss": 0.015274078119546175, + "grad_norm": 0.10606594383716583, + "learning_rate": 2.1789223861622887e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11490 + }, + { + "epoch": 3.3459412278149547, + "flow_matching_loss": 0.016080983635038137, + "grad_norm": 0.09117284417152405, + "learning_rate": 2.1774475648964428e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11500 + }, + { + "epoch": 3.3488507419260984, + "flow_matching_loss": 0.01548759862780571, + "grad_norm": 0.10832858830690384, + "learning_rate": 2.1759719204679267e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11510 + }, + { + "epoch": 3.3517602560372417, + "flow_matching_loss": 0.016317503154277803, + "grad_norm": 0.07406946271657944, + "learning_rate": 2.174495454669787e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11520 + }, + { + "epoch": 3.3546697701483854, + "flow_matching_loss": 0.017939995601773263, + "grad_norm": 0.09933004528284073, + "learning_rate": 2.1730181692960692e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11530 + }, + { + "epoch": 3.3575792842595287, + "flow_matching_loss": 0.014526022132486105, + "grad_norm": 0.11960159242153168, + "learning_rate": 2.171540066141814e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11540 + }, + { + "epoch": 3.360488798370672, + "flow_matching_loss": 0.01648043291643262, + "grad_norm": 0.10899977385997772, + "learning_rate": 2.170061147003056e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11550 + }, + { + "epoch": 3.3633983124818156, + "flow_matching_loss": 0.01760370349511504, + "grad_norm": 0.09907393902540207, + "learning_rate": 2.1685814136768225e-05, + "loss": 0.0176, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11560 + }, + { + "epoch": 3.366307826592959, + "flow_matching_loss": 0.0145729573443532, + "grad_norm": 0.0770871639251709, + "learning_rate": 2.167100867961128e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11570 + }, + { + "epoch": 3.369217340704102, + "flow_matching_loss": 0.017867156118154526, + "grad_norm": 0.09764286875724792, + "learning_rate": 2.1656195116549756e-05, + "loss": 0.0179, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11580 + }, + { + "epoch": 3.372126854815246, + "flow_matching_loss": 0.015663818269968034, + "grad_norm": 0.08561721444129944, + "learning_rate": 2.1641373465583527e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11590 + }, + { + "epoch": 3.375036368926389, + "flow_matching_loss": 0.015396594442427159, + "grad_norm": 0.07548549026250839, + "learning_rate": 2.1626543744722292e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11600 + }, + { + "epoch": 3.377945883037533, + "flow_matching_loss": 0.014566183742135762, + "grad_norm": 0.073089100420475, + "learning_rate": 2.161170597198557e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11610 + }, + { + "epoch": 3.380855397148676, + "flow_matching_loss": 0.01497549181804061, + "grad_norm": 0.06637361645698547, + "learning_rate": 2.1596860165402645e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11620 + }, + { + "epoch": 3.38376491125982, + "flow_matching_loss": 0.01660250285640359, + "grad_norm": 0.10184965282678604, + "learning_rate": 2.1582006343012576e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11630 + }, + { + "epoch": 3.386674425370963, + "flow_matching_loss": 0.01538337329402566, + "grad_norm": 0.07100508362054825, + "learning_rate": 2.156714452286416e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11640 + }, + { + "epoch": 3.3895839394821063, + "flow_matching_loss": 0.016463820915669204, + "grad_norm": 0.09366966038942337, + "learning_rate": 2.155227472301591e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11650 + }, + { + "epoch": 3.39249345359325, + "flow_matching_loss": 0.015530473738908767, + "grad_norm": 0.1079830601811409, + "learning_rate": 2.1537396961536027e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11660 + }, + { + "epoch": 3.3954029677043933, + "flow_matching_loss": 0.015487567894160748, + "grad_norm": 0.07285177707672119, + "learning_rate": 2.1522511256502403e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11670 + }, + { + "epoch": 3.3983124818155366, + "flow_matching_loss": 0.015399404522031545, + "grad_norm": 0.09246493130922318, + "learning_rate": 2.1507617626002566e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11680 + }, + { + "epoch": 3.4012219959266803, + "flow_matching_loss": 0.01696722535416484, + "grad_norm": 0.07571431249380112, + "learning_rate": 2.1492716088133685e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11690 + }, + { + "epoch": 3.4041315100378235, + "flow_matching_loss": 0.015158537402749061, + "grad_norm": 0.07080086320638657, + "learning_rate": 2.1477806661002536e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11700 + }, + { + "epoch": 3.4070410241489673, + "flow_matching_loss": 0.016639158315956592, + "grad_norm": 0.08529754728078842, + "learning_rate": 2.146288936272548e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11710 + }, + { + "epoch": 3.4099505382601105, + "flow_matching_loss": 0.015589340589940548, + "grad_norm": 0.0976489707827568, + "learning_rate": 2.1447964211428433e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11720 + }, + { + "epoch": 3.4128600523712542, + "flow_matching_loss": 0.014236819371581078, + "grad_norm": 0.08778992295265198, + "learning_rate": 2.1433031225246865e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11730 + }, + { + "epoch": 3.4157695664823975, + "flow_matching_loss": 0.018442128505557775, + "grad_norm": 0.0720478817820549, + "learning_rate": 2.1418090422325766e-05, + "loss": 0.0184, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11740 + }, + { + "epoch": 3.4186790805935408, + "flow_matching_loss": 0.014955327659845353, + "grad_norm": 0.08616191893815994, + "learning_rate": 2.1403141820819623e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11750 + }, + { + "epoch": 3.4215885947046845, + "flow_matching_loss": 0.015889866929501294, + "grad_norm": 0.08068066835403442, + "learning_rate": 2.1388185438892377e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11760 + }, + { + "epoch": 3.4244981088158277, + "flow_matching_loss": 0.015952305309474467, + "grad_norm": 0.07275479286909103, + "learning_rate": 2.1373221294717465e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11770 + }, + { + "epoch": 3.427407622926971, + "flow_matching_loss": 0.017077057063579558, + "grad_norm": 0.08692759275436401, + "learning_rate": 2.1358249406477724e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11780 + }, + { + "epoch": 3.4303171370381147, + "flow_matching_loss": 0.015976204723119735, + "grad_norm": 0.07428392767906189, + "learning_rate": 2.1343269792365408e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11790 + }, + { + "epoch": 3.433226651149258, + "flow_matching_loss": 0.01678070882335305, + "grad_norm": 0.09482993930578232, + "learning_rate": 2.132828247058217e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11800 + }, + { + "epoch": 3.4361361652604017, + "flow_matching_loss": 0.016101665794849396, + "grad_norm": 0.08307316899299622, + "learning_rate": 2.1313287459339004e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11810 + }, + { + "epoch": 3.439045679371545, + "flow_matching_loss": 0.015004648733884096, + "grad_norm": 0.08488436788320541, + "learning_rate": 2.1298284776856276e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11820 + }, + { + "epoch": 3.4419551934826886, + "flow_matching_loss": 0.016112047154456376, + "grad_norm": 0.0823889970779419, + "learning_rate": 2.128327444136365e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11830 + }, + { + "epoch": 3.444864707593832, + "flow_matching_loss": 0.015899555012583732, + "grad_norm": 0.07702554762363434, + "learning_rate": 2.1268256471100105e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11840 + }, + { + "epoch": 3.447774221704975, + "flow_matching_loss": 0.014794762525707484, + "grad_norm": 0.0708891823887825, + "learning_rate": 2.1253230884313893e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11850 + }, + { + "epoch": 3.450683735816119, + "flow_matching_loss": 0.016360840667039155, + "grad_norm": 0.08192126452922821, + "learning_rate": 2.123819769926251e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11860 + }, + { + "epoch": 3.453593249927262, + "flow_matching_loss": 0.01517721638083458, + "grad_norm": 0.07312550395727158, + "learning_rate": 2.1223156934212702e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11870 + }, + { + "epoch": 3.4565027640384054, + "flow_matching_loss": 0.015864239539951086, + "grad_norm": 0.07071651518344879, + "learning_rate": 2.1208108607440418e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11880 + }, + { + "epoch": 3.459412278149549, + "flow_matching_loss": 0.015618266072124243, + "grad_norm": 0.08002010732889175, + "learning_rate": 2.119305273723078e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11890 + }, + { + "epoch": 3.4623217922606924, + "flow_matching_loss": 0.014547472726553679, + "grad_norm": 0.0857280045747757, + "learning_rate": 2.1177989341878106e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11900 + }, + { + "epoch": 3.465231306371836, + "flow_matching_loss": 0.016300923004746438, + "grad_norm": 0.06980105489492416, + "learning_rate": 2.116291843968583e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11910 + }, + { + "epoch": 3.4681408204829793, + "flow_matching_loss": 0.016443762369453906, + "grad_norm": 0.10285712033510208, + "learning_rate": 2.114784004896653e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11920 + }, + { + "epoch": 3.4710503345941226, + "flow_matching_loss": 0.016655749082565306, + "grad_norm": 0.06695953011512756, + "learning_rate": 2.113275418804187e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11930 + }, + { + "epoch": 3.4739598487052663, + "flow_matching_loss": 0.014649773295968772, + "grad_norm": 0.07824932783842087, + "learning_rate": 2.1117660875242583e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11940 + }, + { + "epoch": 3.4768693628164096, + "flow_matching_loss": 0.015911860577762126, + "grad_norm": 0.09264266490936279, + "learning_rate": 2.1102560128908478e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11950 + }, + { + "epoch": 3.4797788769275533, + "flow_matching_loss": 0.014400138799101115, + "grad_norm": 0.07432914525270462, + "learning_rate": 2.1087451967388384e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11960 + }, + { + "epoch": 3.4826883910386965, + "flow_matching_loss": 0.015441007912158966, + "grad_norm": 0.08765111118555069, + "learning_rate": 2.1072336409040135e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11970 + }, + { + "epoch": 3.48559790514984, + "flow_matching_loss": 0.01724835569038987, + "grad_norm": 0.08219291269779205, + "learning_rate": 2.1057213472230573e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11980 + }, + { + "epoch": 3.4885074192609835, + "flow_matching_loss": 0.017299752216786146, + "grad_norm": 0.08180999755859375, + "learning_rate": 2.1042083175335478e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 11990 + }, + { + "epoch": 3.4914169333721268, + "flow_matching_loss": 0.015056550595909357, + "grad_norm": 0.07728607952594757, + "learning_rate": 2.1026945536739595e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12000 + }, + { + "epoch": 3.4943264474832705, + "flow_matching_loss": 0.01645749304443598, + "grad_norm": 0.07741076499223709, + "learning_rate": 2.1011800574836586e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12010 + }, + { + "epoch": 3.4972359615944137, + "flow_matching_loss": 0.015137423109263182, + "grad_norm": 0.07451406866312027, + "learning_rate": 2.0996648308028992e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12020 + }, + { + "epoch": 3.500145475705557, + "flow_matching_loss": 0.014386047888547183, + "grad_norm": 0.10213243961334229, + "learning_rate": 2.0981488754728266e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12030 + }, + { + "epoch": 3.5030549898167007, + "flow_matching_loss": 0.015523343347012997, + "grad_norm": 0.07839789986610413, + "learning_rate": 2.0966321933354685e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12040 + }, + { + "epoch": 3.505964503927844, + "flow_matching_loss": 0.014956394396722316, + "grad_norm": 0.07096131891012192, + "learning_rate": 2.095114786233736e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12050 + }, + { + "epoch": 3.5088740180389877, + "flow_matching_loss": 0.01580944927409291, + "grad_norm": 0.09348490089178085, + "learning_rate": 2.0935966560114236e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12060 + }, + { + "epoch": 3.511783532150131, + "flow_matching_loss": 0.015166023187339306, + "grad_norm": 0.10812751948833466, + "learning_rate": 2.092077804513201e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12070 + }, + { + "epoch": 3.514693046261274, + "flow_matching_loss": 0.015801558177918194, + "grad_norm": 0.07773604243993759, + "learning_rate": 2.0905582335846167e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12080 + }, + { + "epoch": 3.517602560372418, + "flow_matching_loss": 0.015770119335502385, + "grad_norm": 0.09053847193717957, + "learning_rate": 2.0890379450720927e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12090 + }, + { + "epoch": 3.520512074483561, + "flow_matching_loss": 0.015636278316378593, + "grad_norm": 0.07905237376689911, + "learning_rate": 2.087516940822923e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12100 + }, + { + "epoch": 3.5234215885947044, + "flow_matching_loss": 0.01614513248205185, + "grad_norm": 0.08440990746021271, + "learning_rate": 2.085995222685271e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12110 + }, + { + "epoch": 3.526331102705848, + "flow_matching_loss": 0.014764749817550183, + "grad_norm": 0.07110119611024857, + "learning_rate": 2.084472792508167e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12120 + }, + { + "epoch": 3.5292406168169914, + "flow_matching_loss": 0.015123569313436747, + "grad_norm": 0.07604318857192993, + "learning_rate": 2.082949652141508e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12130 + }, + { + "epoch": 3.532150130928135, + "flow_matching_loss": 0.015169519931077957, + "grad_norm": 0.07197772711515427, + "learning_rate": 2.081425803436054e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12140 + }, + { + "epoch": 3.5350596450392784, + "flow_matching_loss": 0.016316266171634196, + "grad_norm": 0.0969124287366867, + "learning_rate": 2.079901248243423e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12150 + }, + { + "epoch": 3.537969159150422, + "flow_matching_loss": 0.01687488742172718, + "grad_norm": 0.06748178601264954, + "learning_rate": 2.078375988416095e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12160 + }, + { + "epoch": 3.5408786732615654, + "flow_matching_loss": 0.014629907067865133, + "grad_norm": 0.08651870489120483, + "learning_rate": 2.076850025807404e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12170 + }, + { + "epoch": 3.5437881873727086, + "flow_matching_loss": 0.016862182319164275, + "grad_norm": 0.06576815247535706, + "learning_rate": 2.0753233622715374e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12180 + }, + { + "epoch": 3.5466977014838523, + "flow_matching_loss": 0.014397279918193817, + "grad_norm": 0.057529207319021225, + "learning_rate": 2.073795999663536e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12190 + }, + { + "epoch": 3.5496072155949956, + "flow_matching_loss": 0.015618199482560158, + "grad_norm": 0.0672595426440239, + "learning_rate": 2.07226793983929e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12200 + }, + { + "epoch": 3.552516729706139, + "flow_matching_loss": 0.017521644197404384, + "grad_norm": 0.08405184000730515, + "learning_rate": 2.070739184655536e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12210 + }, + { + "epoch": 3.5554262438172826, + "flow_matching_loss": 0.016649756859987975, + "grad_norm": 0.10807227343320847, + "learning_rate": 2.069209735969855e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12220 + }, + { + "epoch": 3.558335757928426, + "flow_matching_loss": 0.017219106573611496, + "grad_norm": 0.1086990088224411, + "learning_rate": 2.067679595640672e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12230 + }, + { + "epoch": 3.5612452720395695, + "flow_matching_loss": 0.014914821553975343, + "grad_norm": 0.09392277896404266, + "learning_rate": 2.066148765527252e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12240 + }, + { + "epoch": 3.564154786150713, + "flow_matching_loss": 0.015201032534241677, + "grad_norm": 0.11541976034641266, + "learning_rate": 2.0646172474896976e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12250 + }, + { + "epoch": 3.5670643002618565, + "flow_matching_loss": 0.01501031843945384, + "grad_norm": 0.09628836065530777, + "learning_rate": 2.063085043388947e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12260 + }, + { + "epoch": 3.5699738143729998, + "flow_matching_loss": 0.01482925321906805, + "grad_norm": 0.0906067043542862, + "learning_rate": 2.0615521550867732e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12270 + }, + { + "epoch": 3.572883328484143, + "flow_matching_loss": 0.014727403409779072, + "grad_norm": 0.07755091786384583, + "learning_rate": 2.0600185844457803e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12280 + }, + { + "epoch": 3.5757928425952867, + "flow_matching_loss": 0.015338443778455257, + "grad_norm": 0.06917943805456161, + "learning_rate": 2.058484333329401e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12290 + }, + { + "epoch": 3.57870235670643, + "flow_matching_loss": 0.015857416298240425, + "grad_norm": 0.07748106867074966, + "learning_rate": 2.0569494036018955e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12300 + }, + { + "epoch": 3.5816118708175733, + "flow_matching_loss": 0.014812857285141945, + "grad_norm": 0.0860537514090538, + "learning_rate": 2.055413797128347e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12310 + }, + { + "epoch": 3.584521384928717, + "flow_matching_loss": 0.01473366068676114, + "grad_norm": 0.07578720152378082, + "learning_rate": 2.0538775157746634e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12320 + }, + { + "epoch": 3.5874308990398602, + "flow_matching_loss": 0.013597626890987157, + "grad_norm": 0.0764647051692009, + "learning_rate": 2.0523405614075706e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12330 + }, + { + "epoch": 3.590340413151004, + "flow_matching_loss": 0.01479197470471263, + "grad_norm": 0.06410715728998184, + "learning_rate": 2.0508029358946132e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12340 + }, + { + "epoch": 3.593249927262147, + "flow_matching_loss": 0.015179800521582364, + "grad_norm": 0.07077004760503769, + "learning_rate": 2.0492646411041515e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12350 + }, + { + "epoch": 3.596159441373291, + "flow_matching_loss": 0.016236913669854403, + "grad_norm": 0.09112633019685745, + "learning_rate": 2.0477256789053587e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12360 + }, + { + "epoch": 3.599068955484434, + "flow_matching_loss": 0.014412646275013685, + "grad_norm": 0.09569782763719559, + "learning_rate": 2.0461860511682194e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12370 + }, + { + "epoch": 3.6019784695955774, + "flow_matching_loss": 0.015076385717839003, + "grad_norm": 0.09800426661968231, + "learning_rate": 2.0446457597635254e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12380 + }, + { + "epoch": 3.604887983706721, + "flow_matching_loss": 0.015450550056993962, + "grad_norm": 0.08446631580591202, + "learning_rate": 2.0431048065628763e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12390 + }, + { + "epoch": 3.6077974978178644, + "flow_matching_loss": 0.016118599195033313, + "grad_norm": 0.07184389233589172, + "learning_rate": 2.0415631934386764e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12400 + }, + { + "epoch": 3.6107070119290077, + "flow_matching_loss": 0.016177729610353707, + "grad_norm": 0.0882306694984436, + "learning_rate": 2.04002092226413e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12410 + }, + { + "epoch": 3.6136165260401514, + "flow_matching_loss": 0.015446070954203606, + "grad_norm": 0.07394906133413315, + "learning_rate": 2.038477994913242e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12420 + }, + { + "epoch": 3.6165260401512946, + "flow_matching_loss": 0.01669631330296397, + "grad_norm": 0.07907658815383911, + "learning_rate": 2.036934413260816e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12430 + }, + { + "epoch": 3.619435554262438, + "flow_matching_loss": 0.016422658693045378, + "grad_norm": 0.09465497732162476, + "learning_rate": 2.0353901791824473e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12440 + }, + { + "epoch": 3.6223450683735816, + "flow_matching_loss": 0.01496605332940817, + "grad_norm": 0.09778919070959091, + "learning_rate": 2.0338452945545275e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12450 + }, + { + "epoch": 3.6252545824847253, + "flow_matching_loss": 0.016077290009707212, + "grad_norm": 0.06875442713499069, + "learning_rate": 2.0322997612542366e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12460 + }, + { + "epoch": 3.6281640965958686, + "flow_matching_loss": 0.016417355835437776, + "grad_norm": 0.1015530526638031, + "learning_rate": 2.0307535811595427e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12470 + }, + { + "epoch": 3.631073610707012, + "flow_matching_loss": 0.01621114695444703, + "grad_norm": 0.07206564396619797, + "learning_rate": 2.0292067561492013e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12480 + }, + { + "epoch": 3.6339831248181556, + "flow_matching_loss": 0.016566062811762094, + "grad_norm": 0.09519502520561218, + "learning_rate": 2.02765928810275e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12490 + }, + { + "epoch": 3.636892638929299, + "flow_matching_loss": 0.015483225043863059, + "grad_norm": 0.08131353557109833, + "learning_rate": 2.0261111789005087e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12500 + }, + { + "epoch": 3.639802153040442, + "flow_matching_loss": 0.015879342518746852, + "grad_norm": 0.09069592505693436, + "learning_rate": 2.024562430423576e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12510 + }, + { + "epoch": 3.642711667151586, + "flow_matching_loss": 0.01518073482438922, + "grad_norm": 0.0792841836810112, + "learning_rate": 2.0230130445538273e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12520 + }, + { + "epoch": 3.645621181262729, + "flow_matching_loss": 0.015609188564121723, + "grad_norm": 0.07261446863412857, + "learning_rate": 2.0214630231739127e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12530 + }, + { + "epoch": 3.6485306953738723, + "flow_matching_loss": 0.016003567725419998, + "grad_norm": 0.10017208755016327, + "learning_rate": 2.0199123681672543e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12540 + }, + { + "epoch": 3.651440209485016, + "flow_matching_loss": 0.014717599656432868, + "grad_norm": 0.06902829557657242, + "learning_rate": 2.0183610814180437e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12550 + }, + { + "epoch": 3.6543497235961593, + "flow_matching_loss": 0.016812256444245575, + "grad_norm": 0.06748068332672119, + "learning_rate": 2.016809164811241e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12560 + }, + { + "epoch": 3.657259237707303, + "flow_matching_loss": 0.015731275826692582, + "grad_norm": 0.07679944485425949, + "learning_rate": 2.015256620232571e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12570 + }, + { + "epoch": 3.6601687518184463, + "flow_matching_loss": 0.016016950365155935, + "grad_norm": 0.09687835723161697, + "learning_rate": 2.0137034495685222e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12580 + }, + { + "epoch": 3.66307826592959, + "flow_matching_loss": 0.015028362721204757, + "grad_norm": 0.06666471064090729, + "learning_rate": 2.012149654706343e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12590 + }, + { + "epoch": 3.6659877800407332, + "flow_matching_loss": 0.01630397029221058, + "grad_norm": 0.0801204964518547, + "learning_rate": 2.010595237534041e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12600 + }, + { + "epoch": 3.6688972941518765, + "flow_matching_loss": 0.016019405517727136, + "grad_norm": 0.08529985696077347, + "learning_rate": 2.0090401999403793e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12610 + }, + { + "epoch": 3.67180680826302, + "flow_matching_loss": 0.0158347200602293, + "grad_norm": 0.07985467463731766, + "learning_rate": 2.007484543814875e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12620 + }, + { + "epoch": 3.6747163223741635, + "flow_matching_loss": 0.01611623801290989, + "grad_norm": 0.0873478502035141, + "learning_rate": 2.0059282710477972e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12630 + }, + { + "epoch": 3.6776258364853067, + "flow_matching_loss": 0.014705162309110165, + "grad_norm": 0.07241815328598022, + "learning_rate": 2.0043713835301647e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12640 + }, + { + "epoch": 3.6805353505964504, + "flow_matching_loss": 0.015239765774458647, + "grad_norm": 0.07882752269506454, + "learning_rate": 2.0028138831537417e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12650 + }, + { + "epoch": 3.6834448647075937, + "flow_matching_loss": 0.014842839818447829, + "grad_norm": 0.08252668380737305, + "learning_rate": 2.0012557718110387e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12660 + }, + { + "epoch": 3.6863543788187374, + "flow_matching_loss": 0.015467292349785566, + "grad_norm": 0.08509097248315811, + "learning_rate": 1.9996970513953077e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12670 + }, + { + "epoch": 3.6892638929298807, + "flow_matching_loss": 0.015018313191831112, + "grad_norm": 0.08223022520542145, + "learning_rate": 1.9981377238005405e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12680 + }, + { + "epoch": 3.6921734070410244, + "flow_matching_loss": 0.014997973572462797, + "grad_norm": 0.08540572971105576, + "learning_rate": 1.9965777909214674e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12690 + }, + { + "epoch": 3.6950829211521676, + "flow_matching_loss": 0.013907907623797655, + "grad_norm": 0.08100665360689163, + "learning_rate": 1.9950172546535545e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12700 + }, + { + "epoch": 3.697992435263311, + "flow_matching_loss": 0.016329466085880994, + "grad_norm": 0.08483491837978363, + "learning_rate": 1.9934561168930002e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12710 + }, + { + "epoch": 3.7009019493744546, + "flow_matching_loss": 0.014641010854393243, + "grad_norm": 0.09153957664966583, + "learning_rate": 1.991894379536734e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12720 + }, + { + "epoch": 3.703811463485598, + "flow_matching_loss": 0.015480553824454546, + "grad_norm": 0.09066426008939743, + "learning_rate": 1.9903320444824137e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12730 + }, + { + "epoch": 3.706720977596741, + "flow_matching_loss": 0.016156731359660625, + "grad_norm": 0.07582638412714005, + "learning_rate": 1.9887691136284245e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12740 + }, + { + "epoch": 3.709630491707885, + "flow_matching_loss": 0.017133331391960382, + "grad_norm": 0.11425577104091644, + "learning_rate": 1.9872055888738745e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12750 + }, + { + "epoch": 3.712540005819028, + "flow_matching_loss": 0.01545662023127079, + "grad_norm": 0.07261211425065994, + "learning_rate": 1.985641472118593e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12760 + }, + { + "epoch": 3.715449519930172, + "flow_matching_loss": 0.015672298427671195, + "grad_norm": 0.06483614444732666, + "learning_rate": 1.9840767652631294e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12770 + }, + { + "epoch": 3.718359034041315, + "flow_matching_loss": 0.01608514115214348, + "grad_norm": 0.06611791253089905, + "learning_rate": 1.9825114702087513e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12780 + }, + { + "epoch": 3.7212685481524588, + "flow_matching_loss": 0.015126891434192657, + "grad_norm": 0.07577274739742279, + "learning_rate": 1.9809455888574388e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12790 + }, + { + "epoch": 3.724178062263602, + "flow_matching_loss": 0.017325895093381404, + "grad_norm": 0.09634122252464294, + "learning_rate": 1.9793791231118864e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12800 + }, + { + "epoch": 3.7270875763747453, + "flow_matching_loss": 0.015553371980786324, + "grad_norm": 0.07159753143787384, + "learning_rate": 1.9778120748754958e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12810 + }, + { + "epoch": 3.729997090485889, + "flow_matching_loss": 0.015327459760010242, + "grad_norm": 0.09359882026910782, + "learning_rate": 1.97624444605238e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12820 + }, + { + "epoch": 3.7329066045970323, + "flow_matching_loss": 0.014849781338125468, + "grad_norm": 0.06924345344305038, + "learning_rate": 1.9746762385473552e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12830 + }, + { + "epoch": 3.7358161187081755, + "flow_matching_loss": 0.015838226396590473, + "grad_norm": 0.0921359732747078, + "learning_rate": 1.9731074542659414e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12840 + }, + { + "epoch": 3.7387256328193192, + "flow_matching_loss": 0.016693384386599064, + "grad_norm": 0.09801449626684189, + "learning_rate": 1.9715380951143603e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12850 + }, + { + "epoch": 3.7416351469304625, + "flow_matching_loss": 0.01581725450232625, + "grad_norm": 0.10012103617191315, + "learning_rate": 1.9699681629995295e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12860 + }, + { + "epoch": 3.744544661041606, + "flow_matching_loss": 0.014978116378188133, + "grad_norm": 0.08862815052270889, + "learning_rate": 1.968397659829066e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12870 + }, + { + "epoch": 3.7474541751527495, + "flow_matching_loss": 0.014335913117974997, + "grad_norm": 0.07120011001825333, + "learning_rate": 1.966826587511279e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12880 + }, + { + "epoch": 3.750363689263893, + "flow_matching_loss": 0.015277501568198204, + "grad_norm": 0.08024851232767105, + "learning_rate": 1.965254947955169e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12890 + }, + { + "epoch": 3.7532732033750364, + "flow_matching_loss": 0.017034181207418442, + "grad_norm": 0.06435678154230118, + "learning_rate": 1.963682743070427e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12900 + }, + { + "epoch": 3.7561827174861797, + "flow_matching_loss": 0.014810711517930031, + "grad_norm": 0.07694236189126968, + "learning_rate": 1.962109974767429e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12910 + }, + { + "epoch": 3.7590922315973234, + "flow_matching_loss": 0.015188704617321492, + "grad_norm": 0.09992945939302444, + "learning_rate": 1.960536644957238e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12920 + }, + { + "epoch": 3.7620017457084667, + "flow_matching_loss": 0.015435496717691422, + "grad_norm": 0.09228812903165817, + "learning_rate": 1.9589627555515974e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12930 + }, + { + "epoch": 3.76491125981961, + "flow_matching_loss": 0.015882004331797362, + "grad_norm": 0.07033485919237137, + "learning_rate": 1.9573883084629317e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12940 + }, + { + "epoch": 3.7678207739307537, + "flow_matching_loss": 0.015843075700104237, + "grad_norm": 0.06887858361005783, + "learning_rate": 1.9558133056043422e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12950 + }, + { + "epoch": 3.770730288041897, + "flow_matching_loss": 0.014563427679240704, + "grad_norm": 0.08626893162727356, + "learning_rate": 1.9542377488896062e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12960 + }, + { + "epoch": 3.77363980215304, + "flow_matching_loss": 0.015314784366637468, + "grad_norm": 0.05913437530398369, + "learning_rate": 1.9526616402331733e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12970 + }, + { + "epoch": 3.776549316264184, + "flow_matching_loss": 0.01529606683179736, + "grad_norm": 0.08114432543516159, + "learning_rate": 1.9510849815501638e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12980 + }, + { + "epoch": 3.7794588303753276, + "flow_matching_loss": 0.015286808274686337, + "grad_norm": 0.09581013023853302, + "learning_rate": 1.949507774756368e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 12990 + }, + { + "epoch": 3.782368344486471, + "flow_matching_loss": 0.016128509305417538, + "grad_norm": 0.07352163642644882, + "learning_rate": 1.9479300217682394e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13000 + }, + { + "epoch": 3.785277858597614, + "flow_matching_loss": 0.015466719213873149, + "grad_norm": 0.07060395181179047, + "learning_rate": 1.946351724502898e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13010 + }, + { + "epoch": 3.788187372708758, + "flow_matching_loss": 0.015147818066179753, + "grad_norm": 0.07406304776668549, + "learning_rate": 1.9447728848781227e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13020 + }, + { + "epoch": 3.791096886819901, + "flow_matching_loss": 0.015304163750261068, + "grad_norm": 0.09741988033056259, + "learning_rate": 1.943193504812353e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13030 + }, + { + "epoch": 3.7940064009310444, + "flow_matching_loss": 0.015030360501259565, + "grad_norm": 0.06565934419631958, + "learning_rate": 1.9416135862246846e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13040 + }, + { + "epoch": 3.796915915042188, + "flow_matching_loss": 0.014392882771790027, + "grad_norm": 0.08280763775110245, + "learning_rate": 1.9400331310348673e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13050 + }, + { + "epoch": 3.7998254291533313, + "flow_matching_loss": 0.016658001206815244, + "grad_norm": 0.11200007796287537, + "learning_rate": 1.9384521411633033e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13060 + }, + { + "epoch": 3.8027349432644746, + "flow_matching_loss": 0.01615695161744952, + "grad_norm": 0.08758462220430374, + "learning_rate": 1.936870618531045e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13070 + }, + { + "epoch": 3.8056444573756183, + "flow_matching_loss": 0.015309465024620295, + "grad_norm": 0.07756228744983673, + "learning_rate": 1.9352885650597904e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13080 + }, + { + "epoch": 3.8085539714867616, + "flow_matching_loss": 0.014990462735295296, + "grad_norm": 0.09962864220142365, + "learning_rate": 1.9337059826718847e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13090 + }, + { + "epoch": 3.8114634855979053, + "flow_matching_loss": 0.014035008288919926, + "grad_norm": 0.09485512226819992, + "learning_rate": 1.932122873290314e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13100 + }, + { + "epoch": 3.8143729997090485, + "flow_matching_loss": 0.01592204039916396, + "grad_norm": 0.07570520788431168, + "learning_rate": 1.9305392388387054e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13110 + }, + { + "epoch": 3.8172825138201922, + "flow_matching_loss": 0.016062804590910675, + "grad_norm": 0.08076439052820206, + "learning_rate": 1.928955081241325e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13120 + }, + { + "epoch": 3.8201920279313355, + "flow_matching_loss": 0.015399712882936002, + "grad_norm": 0.12258557230234146, + "learning_rate": 1.9273704024230725e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13130 + }, + { + "epoch": 3.8231015420424788, + "flow_matching_loss": 0.01596811357885599, + "grad_norm": 0.06719166785478592, + "learning_rate": 1.925785204309483e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13140 + }, + { + "epoch": 3.8260110561536225, + "flow_matching_loss": 0.014814761374145747, + "grad_norm": 0.07782695442438126, + "learning_rate": 1.9241994888267212e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13150 + }, + { + "epoch": 3.8289205702647657, + "flow_matching_loss": 0.0159940954297781, + "grad_norm": 0.06713519990444183, + "learning_rate": 1.9226132579015815e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13160 + }, + { + "epoch": 3.831830084375909, + "flow_matching_loss": 0.017756551317870618, + "grad_norm": 0.07703738659620285, + "learning_rate": 1.9210265134614832e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13170 + }, + { + "epoch": 3.8347395984870527, + "flow_matching_loss": 0.01571000972762704, + "grad_norm": 0.07831337302923203, + "learning_rate": 1.9194392574344712e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13180 + }, + { + "epoch": 3.837649112598196, + "flow_matching_loss": 0.015117396600544453, + "grad_norm": 0.06852998584508896, + "learning_rate": 1.9178514917492103e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13190 + }, + { + "epoch": 3.8405586267093397, + "flow_matching_loss": 0.01726938281208277, + "grad_norm": 0.0651521161198616, + "learning_rate": 1.9162632183349866e-05, + "loss": 0.0173, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13200 + }, + { + "epoch": 3.843468140820483, + "flow_matching_loss": 0.015599463786929845, + "grad_norm": 0.06973610818386078, + "learning_rate": 1.914674439121701e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13210 + }, + { + "epoch": 3.8463776549316266, + "flow_matching_loss": 0.015650570392608643, + "grad_norm": 0.0911756381392479, + "learning_rate": 1.9130851560398715e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13220 + }, + { + "epoch": 3.84928716904277, + "flow_matching_loss": 0.014214181620627642, + "grad_norm": 0.056858714669942856, + "learning_rate": 1.911495371020625e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13230 + }, + { + "epoch": 3.852196683153913, + "flow_matching_loss": 0.015668600518256425, + "grad_norm": 0.10028836131095886, + "learning_rate": 1.909905085995702e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13240 + }, + { + "epoch": 3.855106197265057, + "flow_matching_loss": 0.015958270244300365, + "grad_norm": 0.07811019569635391, + "learning_rate": 1.908314302897448e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13250 + }, + { + "epoch": 3.8580157113762, + "flow_matching_loss": 0.016703103110194205, + "grad_norm": 0.07165423035621643, + "learning_rate": 1.9067230236588142e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13260 + }, + { + "epoch": 3.8609252254873434, + "flow_matching_loss": 0.014458565693348645, + "grad_norm": 0.10392113029956818, + "learning_rate": 1.905131250213355e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13270 + }, + { + "epoch": 3.863834739598487, + "flow_matching_loss": 0.014913774002343417, + "grad_norm": 0.08680251985788345, + "learning_rate": 1.9035389844952256e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13280 + }, + { + "epoch": 3.8667442537096304, + "flow_matching_loss": 0.01822825213894248, + "grad_norm": 0.08185291290283203, + "learning_rate": 1.901946228439179e-05, + "loss": 0.0182, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13290 + }, + { + "epoch": 3.869653767820774, + "flow_matching_loss": 0.015454999636858702, + "grad_norm": 0.0889175534248352, + "learning_rate": 1.9003529839805637e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13300 + }, + { + "epoch": 3.8725632819319173, + "flow_matching_loss": 0.014318622462451459, + "grad_norm": 0.05035459250211716, + "learning_rate": 1.8987592530553217e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13310 + }, + { + "epoch": 3.875472796043061, + "flow_matching_loss": 0.017766905017197132, + "grad_norm": 0.08929821103811264, + "learning_rate": 1.897165037599987e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13320 + }, + { + "epoch": 3.8783823101542043, + "flow_matching_loss": 0.014778384193778037, + "grad_norm": 0.06941773742437363, + "learning_rate": 1.8955703395516814e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13330 + }, + { + "epoch": 3.8812918242653476, + "flow_matching_loss": 0.016314536333084106, + "grad_norm": 0.07863995432853699, + "learning_rate": 1.8939751608481134e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13340 + }, + { + "epoch": 3.8842013383764913, + "flow_matching_loss": 0.015667952317744493, + "grad_norm": 0.06965357810258865, + "learning_rate": 1.892379503427575e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13350 + }, + { + "epoch": 3.8871108524876345, + "flow_matching_loss": 0.015365955978631973, + "grad_norm": 0.07910340279340744, + "learning_rate": 1.8907833692289412e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13360 + }, + { + "epoch": 3.890020366598778, + "flow_matching_loss": 0.015671274717897178, + "grad_norm": 0.06519993394613266, + "learning_rate": 1.889186760191666e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13370 + }, + { + "epoch": 3.8929298807099215, + "flow_matching_loss": 0.01589350253343582, + "grad_norm": 0.0732232928276062, + "learning_rate": 1.887589678255779e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13380 + }, + { + "epoch": 3.895839394821065, + "flow_matching_loss": 0.015495924465358257, + "grad_norm": 0.07772276550531387, + "learning_rate": 1.8859921253618852e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13390 + }, + { + "epoch": 3.898748908932208, + "flow_matching_loss": 0.01668654689565301, + "grad_norm": 0.10574867576360703, + "learning_rate": 1.884394103451163e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13400 + }, + { + "epoch": 3.9016584230433518, + "flow_matching_loss": 0.014674871694296598, + "grad_norm": 0.07080578804016113, + "learning_rate": 1.8827956144653583e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13410 + }, + { + "epoch": 3.9045679371544955, + "flow_matching_loss": 0.01490477379411459, + "grad_norm": 0.07571115344762802, + "learning_rate": 1.8811966603467874e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13420 + }, + { + "epoch": 3.9074774512656387, + "flow_matching_loss": 0.01667021308094263, + "grad_norm": 0.10515370965003967, + "learning_rate": 1.8795972430383293e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13430 + }, + { + "epoch": 3.910386965376782, + "flow_matching_loss": 0.014604585058987141, + "grad_norm": 0.09119710326194763, + "learning_rate": 1.8779973644834273e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13440 + }, + { + "epoch": 3.9132964794879257, + "flow_matching_loss": 0.01420961944386363, + "grad_norm": 0.08076418936252594, + "learning_rate": 1.876397026626085e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13450 + }, + { + "epoch": 3.916205993599069, + "flow_matching_loss": 0.014849066920578479, + "grad_norm": 0.07542139291763306, + "learning_rate": 1.8747962314108633e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13460 + }, + { + "epoch": 3.919115507710212, + "flow_matching_loss": 0.015440836921334267, + "grad_norm": 0.08173234760761261, + "learning_rate": 1.873194980782879e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13470 + }, + { + "epoch": 3.922025021821356, + "flow_matching_loss": 0.014377333596348763, + "grad_norm": 0.06418035924434662, + "learning_rate": 1.871593276687804e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13480 + }, + { + "epoch": 3.924934535932499, + "flow_matching_loss": 0.014493086189031602, + "grad_norm": 0.072786346077919, + "learning_rate": 1.869991121071858e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13490 + }, + { + "epoch": 3.9278440500436425, + "flow_matching_loss": 0.015903431084007026, + "grad_norm": 0.0696580708026886, + "learning_rate": 1.8683885158818117e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13500 + }, + { + "epoch": 3.930753564154786, + "flow_matching_loss": 0.014030499663203955, + "grad_norm": 0.0814409852027893, + "learning_rate": 1.8667854630649825e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13510 + }, + { + "epoch": 3.9336630782659294, + "flow_matching_loss": 0.01608144948258996, + "grad_norm": 0.059304606169462204, + "learning_rate": 1.86518196456923e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13520 + }, + { + "epoch": 3.936572592377073, + "flow_matching_loss": 0.015775493625551463, + "grad_norm": 0.10035821050405502, + "learning_rate": 1.8635780223429565e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13530 + }, + { + "epoch": 3.9394821064882164, + "flow_matching_loss": 0.016348942276090385, + "grad_norm": 0.08098331093788147, + "learning_rate": 1.8619736383351017e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13540 + }, + { + "epoch": 3.94239162059936, + "flow_matching_loss": 0.01480183145031333, + "grad_norm": 0.05381172150373459, + "learning_rate": 1.8603688144951454e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13550 + }, + { + "epoch": 3.9453011347105034, + "flow_matching_loss": 0.015695603098720312, + "grad_norm": 0.08972971141338348, + "learning_rate": 1.8587635527730983e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13560 + }, + { + "epoch": 3.9482106488216466, + "flow_matching_loss": 0.015048960503190757, + "grad_norm": 0.0823139026761055, + "learning_rate": 1.8571578551195056e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13570 + }, + { + "epoch": 3.9511201629327903, + "flow_matching_loss": 0.013475803565233946, + "grad_norm": 0.0720219612121582, + "learning_rate": 1.855551723485441e-05, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13580 + }, + { + "epoch": 3.9540296770439336, + "flow_matching_loss": 0.015181075315922499, + "grad_norm": 0.07116485387086868, + "learning_rate": 1.8539451598225066e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13590 + }, + { + "epoch": 3.956939191155077, + "flow_matching_loss": 0.015714609995484352, + "grad_norm": 0.08450599759817123, + "learning_rate": 1.8523381660828276e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13600 + }, + { + "epoch": 3.9598487052662206, + "flow_matching_loss": 0.015774597972631456, + "grad_norm": 0.10173771530389786, + "learning_rate": 1.850730744219054e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13610 + }, + { + "epoch": 3.962758219377364, + "flow_matching_loss": 0.014625470526516438, + "grad_norm": 0.07654306292533875, + "learning_rate": 1.849122896184354e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13620 + }, + { + "epoch": 3.9656677334885075, + "flow_matching_loss": 0.016870354395359756, + "grad_norm": 0.08706289529800415, + "learning_rate": 1.8475146239324146e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13630 + }, + { + "epoch": 3.968577247599651, + "flow_matching_loss": 0.013773362524807454, + "grad_norm": 0.07280026376247406, + "learning_rate": 1.8459059294174396e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13640 + }, + { + "epoch": 3.9714867617107945, + "flow_matching_loss": 0.015972617641091348, + "grad_norm": 0.07602652907371521, + "learning_rate": 1.844296814594143e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13650 + }, + { + "epoch": 3.9743962758219378, + "flow_matching_loss": 0.015194455161690712, + "grad_norm": 0.09904441237449646, + "learning_rate": 1.8426872814177516e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13660 + }, + { + "epoch": 3.977305789933081, + "flow_matching_loss": 0.015282877162098885, + "grad_norm": 0.0880991667509079, + "learning_rate": 1.8410773318440003e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13670 + }, + { + "epoch": 3.9802153040442247, + "flow_matching_loss": 0.014428908750414849, + "grad_norm": 0.08847387135028839, + "learning_rate": 1.839466967829129e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13680 + }, + { + "epoch": 3.983124818155368, + "flow_matching_loss": 0.014933108165860177, + "grad_norm": 0.0695764422416687, + "learning_rate": 1.8378561913298822e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13690 + }, + { + "epoch": 3.9860343322665113, + "flow_matching_loss": 0.013626991678029299, + "grad_norm": 0.07035688310861588, + "learning_rate": 1.8362450043035053e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13700 + }, + { + "epoch": 3.988943846377655, + "flow_matching_loss": 0.015200061164796353, + "grad_norm": 0.0848856046795845, + "learning_rate": 1.834633408707742e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13710 + }, + { + "epoch": 3.9918533604887982, + "flow_matching_loss": 0.015958325657993554, + "grad_norm": 0.11115742474794388, + "learning_rate": 1.8330214065008336e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13720 + }, + { + "epoch": 3.994762874599942, + "flow_matching_loss": 0.016256703156977893, + "grad_norm": 0.07842010259628296, + "learning_rate": 1.831408999641514e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13730 + }, + { + "epoch": 3.997672388711085, + "flow_matching_loss": 0.01529949577525258, + "grad_norm": 0.09860032051801682, + "learning_rate": 1.8297961900890107e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13740 + }, + { + "epoch": 4.000581902822229, + "flow_matching_loss": 0.013921436574310064, + "grad_norm": 0.09491388499736786, + "learning_rate": 1.8281829798030385e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13750 + }, + { + "epoch": 4.003491416933372, + "flow_matching_loss": 0.015295592974871397, + "grad_norm": 0.10895074158906937, + "learning_rate": 1.8265693707438002e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13760 + }, + { + "epoch": 4.0064009310445154, + "flow_matching_loss": 0.015313094947487116, + "grad_norm": 0.0973781868815422, + "learning_rate": 1.824955364871982e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13770 + }, + { + "epoch": 4.009310445155659, + "flow_matching_loss": 0.014313169941306114, + "grad_norm": 0.07113306224346161, + "learning_rate": 1.8233409641487553e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13780 + }, + { + "epoch": 4.012219959266803, + "flow_matching_loss": 0.016398165095597505, + "grad_norm": 0.07743734121322632, + "learning_rate": 1.8217261705357677e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13790 + }, + { + "epoch": 4.015129473377946, + "flow_matching_loss": 0.01507912902161479, + "grad_norm": 0.10143893212080002, + "learning_rate": 1.8201109859951462e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13800 + }, + { + "epoch": 4.018038987489089, + "flow_matching_loss": 0.01370950099080801, + "grad_norm": 0.10464879125356674, + "learning_rate": 1.818495412489492e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13810 + }, + { + "epoch": 4.020948501600233, + "flow_matching_loss": 0.01678687259554863, + "grad_norm": 0.07353879511356354, + "learning_rate": 1.8168794519818802e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13820 + }, + { + "epoch": 4.023858015711376, + "flow_matching_loss": 0.015798552613705396, + "grad_norm": 0.062051720917224884, + "learning_rate": 1.8152631064358537e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13830 + }, + { + "epoch": 4.02676752982252, + "flow_matching_loss": 0.015245513059198856, + "grad_norm": 0.0666421577334404, + "learning_rate": 1.8136463778154258e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13840 + }, + { + "epoch": 4.029677043933663, + "flow_matching_loss": 0.015540968161076307, + "grad_norm": 0.07380559295415878, + "learning_rate": 1.812029268085074e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13850 + }, + { + "epoch": 4.032586558044806, + "flow_matching_loss": 0.01651590596884489, + "grad_norm": 0.08852961659431458, + "learning_rate": 1.8104117792097387e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13860 + }, + { + "epoch": 4.03549607215595, + "flow_matching_loss": 0.01656812708824873, + "grad_norm": 0.09526115655899048, + "learning_rate": 1.808793913154822e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13870 + }, + { + "epoch": 4.038405586267094, + "flow_matching_loss": 0.01594859445467591, + "grad_norm": 0.08893018215894699, + "learning_rate": 1.8071756718861838e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13880 + }, + { + "epoch": 4.041315100378237, + "flow_matching_loss": 0.015321127977222205, + "grad_norm": 0.09843612462282181, + "learning_rate": 1.805557057370139e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13890 + }, + { + "epoch": 4.04422461448938, + "flow_matching_loss": 0.016273250617086888, + "grad_norm": 0.06771542876958847, + "learning_rate": 1.8039380715734574e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13900 + }, + { + "epoch": 4.047134128600524, + "flow_matching_loss": 0.015595010202378034, + "grad_norm": 0.08702874183654785, + "learning_rate": 1.802318716463359e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13910 + }, + { + "epoch": 4.0500436427116675, + "flow_matching_loss": 0.01558934124186635, + "grad_norm": 0.08044595271348953, + "learning_rate": 1.8006989940075138e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13920 + }, + { + "epoch": 4.05295315682281, + "flow_matching_loss": 0.014209416415542363, + "grad_norm": 0.06619153916835785, + "learning_rate": 1.7990789061740368e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13930 + }, + { + "epoch": 4.055862670933954, + "flow_matching_loss": 0.01566743403673172, + "grad_norm": 0.09016405045986176, + "learning_rate": 1.7974584549314875e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13940 + }, + { + "epoch": 4.058772185045098, + "flow_matching_loss": 0.016773972660303116, + "grad_norm": 0.07450349628925323, + "learning_rate": 1.7958376422488672e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13950 + }, + { + "epoch": 4.0616816991562406, + "flow_matching_loss": 0.017084579821676014, + "grad_norm": 0.07802364230155945, + "learning_rate": 1.7942164700956156e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13960 + }, + { + "epoch": 4.064591213267384, + "flow_matching_loss": 0.01420174390077591, + "grad_norm": 0.0774056538939476, + "learning_rate": 1.79259494044161e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13970 + }, + { + "epoch": 4.067500727378528, + "flow_matching_loss": 0.015040003694593906, + "grad_norm": 0.08767130225896835, + "learning_rate": 1.7909730552571617e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13980 + }, + { + "epoch": 4.070410241489671, + "flow_matching_loss": 0.01630115704610944, + "grad_norm": 0.10135307908058167, + "learning_rate": 1.7893508165130143e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 13990 + }, + { + "epoch": 4.0733197556008145, + "flow_matching_loss": 0.016279854439198972, + "grad_norm": 0.07549937814474106, + "learning_rate": 1.787728226180341e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14000 + }, + { + "epoch": 4.076229269711958, + "flow_matching_loss": 0.015538495127111673, + "grad_norm": 0.0809144377708435, + "learning_rate": 1.786105286230742e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14010 + }, + { + "epoch": 4.079138783823102, + "flow_matching_loss": 0.015324411354959011, + "grad_norm": 0.07660394161939621, + "learning_rate": 1.784481998636242e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14020 + }, + { + "epoch": 4.082048297934245, + "flow_matching_loss": 0.0174640073440969, + "grad_norm": 0.0978534147143364, + "learning_rate": 1.782858365369289e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14030 + }, + { + "epoch": 4.084957812045388, + "flow_matching_loss": 0.015492206066846847, + "grad_norm": 0.10647391527891159, + "learning_rate": 1.78123438840275e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14040 + }, + { + "epoch": 4.087867326156532, + "flow_matching_loss": 0.016346232686191797, + "grad_norm": 0.0786377489566803, + "learning_rate": 1.779610069709911e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14050 + }, + { + "epoch": 4.090776840267675, + "flow_matching_loss": 0.01653909534215927, + "grad_norm": 0.08222456276416779, + "learning_rate": 1.7779854112644713e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14060 + }, + { + "epoch": 4.093686354378819, + "flow_matching_loss": 0.01631098398938775, + "grad_norm": 0.08705010265111923, + "learning_rate": 1.7763604150405446e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14070 + }, + { + "epoch": 4.096595868489962, + "flow_matching_loss": 0.016011272091418505, + "grad_norm": 0.08564131706953049, + "learning_rate": 1.7747350830126548e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14080 + }, + { + "epoch": 4.099505382601105, + "flow_matching_loss": 0.016094554960727692, + "grad_norm": 0.08629187196493149, + "learning_rate": 1.7731094171557332e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14090 + }, + { + "epoch": 4.102414896712249, + "flow_matching_loss": 0.014798251725733281, + "grad_norm": 0.0702090933918953, + "learning_rate": 1.771483419445117e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14100 + }, + { + "epoch": 4.105324410823393, + "flow_matching_loss": 0.013313017040491103, + "grad_norm": 0.0775623768568039, + "learning_rate": 1.7698570918565474e-05, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14110 + }, + { + "epoch": 4.108233924934536, + "flow_matching_loss": 0.013368872273713351, + "grad_norm": 0.07200277596712112, + "learning_rate": 1.768230436366165e-05, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14120 + }, + { + "epoch": 4.111143439045679, + "flow_matching_loss": 0.016602748539298773, + "grad_norm": 0.07156822830438614, + "learning_rate": 1.7666034549505102e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14130 + }, + { + "epoch": 4.114052953156823, + "flow_matching_loss": 0.016620270162820815, + "grad_norm": 0.07272769510746002, + "learning_rate": 1.7649761495865185e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14140 + }, + { + "epoch": 4.1169624672679666, + "flow_matching_loss": 0.014955591596662998, + "grad_norm": 0.08966485410928726, + "learning_rate": 1.7633485222515188e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14150 + }, + { + "epoch": 4.119871981379109, + "flow_matching_loss": 0.016474599298089743, + "grad_norm": 0.07576680928468704, + "learning_rate": 1.761720574923233e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14160 + }, + { + "epoch": 4.122781495490253, + "flow_matching_loss": 0.01619552979245782, + "grad_norm": 0.0851840078830719, + "learning_rate": 1.7600923095797704e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14170 + }, + { + "epoch": 4.125691009601397, + "flow_matching_loss": 0.014801710937172174, + "grad_norm": 0.0947892889380455, + "learning_rate": 1.758463728199626e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14180 + }, + { + "epoch": 4.12860052371254, + "flow_matching_loss": 0.014067183155566454, + "grad_norm": 0.09974844753742218, + "learning_rate": 1.7568348327616803e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14190 + }, + { + "epoch": 4.131510037823683, + "flow_matching_loss": 0.0156576294451952, + "grad_norm": 0.07054577022790909, + "learning_rate": 1.755205625245195e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14200 + }, + { + "epoch": 4.134419551934827, + "flow_matching_loss": 0.014158251788467168, + "grad_norm": 0.08261838555335999, + "learning_rate": 1.7535761076298104e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14210 + }, + { + "epoch": 4.137329066045971, + "flow_matching_loss": 0.013758059963583947, + "grad_norm": 0.09073110669851303, + "learning_rate": 1.7519462818955448e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14220 + }, + { + "epoch": 4.1402385801571135, + "flow_matching_loss": 0.015317468252032996, + "grad_norm": 0.0765247642993927, + "learning_rate": 1.7503161500227896e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14230 + }, + { + "epoch": 4.143148094268257, + "flow_matching_loss": 0.014885509200394154, + "grad_norm": 0.0904502421617508, + "learning_rate": 1.7486857139923093e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14240 + }, + { + "epoch": 4.146057608379401, + "flow_matching_loss": 0.015133786760270596, + "grad_norm": 0.08348391950130463, + "learning_rate": 1.747054975785236e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14250 + }, + { + "epoch": 4.148967122490544, + "flow_matching_loss": 0.016330687887966634, + "grad_norm": 0.08265487104654312, + "learning_rate": 1.745423937383072e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14260 + }, + { + "epoch": 4.1518766366016875, + "flow_matching_loss": 0.013599105551838875, + "grad_norm": 0.07202516496181488, + "learning_rate": 1.7437926007676825e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14270 + }, + { + "epoch": 4.154786150712831, + "flow_matching_loss": 0.015054161101579666, + "grad_norm": 0.0710628479719162, + "learning_rate": 1.742160967921294e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14280 + }, + { + "epoch": 4.157695664823974, + "flow_matching_loss": 0.016264576837420462, + "grad_norm": 0.0703674778342247, + "learning_rate": 1.7405290408264954e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14290 + }, + { + "epoch": 4.160605178935118, + "flow_matching_loss": 0.014653602987527848, + "grad_norm": 0.055504944175481796, + "learning_rate": 1.738896821466232e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14300 + }, + { + "epoch": 4.163514693046261, + "flow_matching_loss": 0.014516313932836056, + "grad_norm": 0.07429336756467819, + "learning_rate": 1.737264311823804e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14310 + }, + { + "epoch": 4.166424207157405, + "flow_matching_loss": 0.015140607953071594, + "grad_norm": 0.06966197490692139, + "learning_rate": 1.7356315138828645e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14320 + }, + { + "epoch": 4.169333721268548, + "flow_matching_loss": 0.015886259730905294, + "grad_norm": 0.0807681679725647, + "learning_rate": 1.7339984296274174e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14330 + }, + { + "epoch": 4.172243235379692, + "flow_matching_loss": 0.015203347150236367, + "grad_norm": 0.08876502513885498, + "learning_rate": 1.7323650610418134e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14340 + }, + { + "epoch": 4.175152749490835, + "flow_matching_loss": 0.01510233236476779, + "grad_norm": 0.09879422932863235, + "learning_rate": 1.730731410110749e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14350 + }, + { + "epoch": 4.178062263601978, + "flow_matching_loss": 0.015150254219770431, + "grad_norm": 0.0818834900856018, + "learning_rate": 1.7290974788192655e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14360 + }, + { + "epoch": 4.180971777713122, + "flow_matching_loss": 0.015755346044898033, + "grad_norm": 0.0790729895234108, + "learning_rate": 1.727463269152743e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14370 + }, + { + "epoch": 4.183881291824266, + "flow_matching_loss": 0.016101249400526284, + "grad_norm": 0.07995037734508514, + "learning_rate": 1.7258287830968996e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14380 + }, + { + "epoch": 4.186790805935408, + "flow_matching_loss": 0.015753670409321786, + "grad_norm": 0.07416621595621109, + "learning_rate": 1.7241940226377908e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14390 + }, + { + "epoch": 4.189700320046552, + "flow_matching_loss": 0.015176983922719956, + "grad_norm": 0.07164789736270905, + "learning_rate": 1.7225589897618043e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14400 + }, + { + "epoch": 4.192609834157696, + "flow_matching_loss": 0.01577199837192893, + "grad_norm": 0.08149939775466919, + "learning_rate": 1.7209236864556588e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14410 + }, + { + "epoch": 4.195519348268839, + "flow_matching_loss": 0.014864621870219707, + "grad_norm": 0.08579286932945251, + "learning_rate": 1.719288114706403e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14420 + }, + { + "epoch": 4.198428862379982, + "flow_matching_loss": 0.016547549236565828, + "grad_norm": 0.10186683386564255, + "learning_rate": 1.7176522765014104e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14430 + }, + { + "epoch": 4.201338376491126, + "flow_matching_loss": 0.01698241485282779, + "grad_norm": 0.06917279958724976, + "learning_rate": 1.7160161738283782e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14440 + }, + { + "epoch": 4.20424789060227, + "flow_matching_loss": 0.01597996400669217, + "grad_norm": 0.07216208428144455, + "learning_rate": 1.714379808675326e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14450 + }, + { + "epoch": 4.207157404713413, + "flow_matching_loss": 0.017160116136074065, + "grad_norm": 0.08764893561601639, + "learning_rate": 1.712743183030592e-05, + "loss": 0.0172, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14460 + }, + { + "epoch": 4.210066918824556, + "flow_matching_loss": 0.015582568570971489, + "grad_norm": 0.09027358889579773, + "learning_rate": 1.7111062988828294e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14470 + }, + { + "epoch": 4.2129764329357, + "flow_matching_loss": 0.01614631749689579, + "grad_norm": 0.08200453221797943, + "learning_rate": 1.709469158221008e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14480 + }, + { + "epoch": 4.215885947046843, + "flow_matching_loss": 0.015510893054306507, + "grad_norm": 0.08983764797449112, + "learning_rate": 1.7078317630344083e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14490 + }, + { + "epoch": 4.2187954611579865, + "flow_matching_loss": 0.015322821494191886, + "grad_norm": 0.09228964895009995, + "learning_rate": 1.706194115312618e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14500 + }, + { + "epoch": 4.22170497526913, + "flow_matching_loss": 0.014185319468379021, + "grad_norm": 0.0706309974193573, + "learning_rate": 1.704556217045536e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14510 + }, + { + "epoch": 4.224614489380274, + "flow_matching_loss": 0.015773274190723895, + "grad_norm": 0.08242277055978775, + "learning_rate": 1.7029180702233612e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14520 + }, + { + "epoch": 4.227524003491417, + "flow_matching_loss": 0.015822984743863344, + "grad_norm": 0.08430999517440796, + "learning_rate": 1.7012796768365975e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14530 + }, + { + "epoch": 4.2304335176025605, + "flow_matching_loss": 0.015575056616216897, + "grad_norm": 0.07694287598133087, + "learning_rate": 1.6996410388760465e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14540 + }, + { + "epoch": 4.233343031713704, + "flow_matching_loss": 0.015194864012300968, + "grad_norm": 0.08898156881332397, + "learning_rate": 1.698002158332809e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14550 + }, + { + "epoch": 4.236252545824847, + "flow_matching_loss": 0.016067562345415353, + "grad_norm": 0.09187214076519012, + "learning_rate": 1.6963630371982783e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14560 + }, + { + "epoch": 4.239162059935991, + "flow_matching_loss": 0.015971062891185283, + "grad_norm": 0.07520009577274323, + "learning_rate": 1.6947236774641416e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14570 + }, + { + "epoch": 4.242071574047134, + "flow_matching_loss": 0.016495497711002827, + "grad_norm": 0.0927506685256958, + "learning_rate": 1.6930840811223755e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14580 + }, + { + "epoch": 4.244981088158277, + "flow_matching_loss": 0.016172353830188512, + "grad_norm": 0.09534966200590134, + "learning_rate": 1.6914442501652448e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14590 + }, + { + "epoch": 4.247890602269421, + "flow_matching_loss": 0.01466203723102808, + "grad_norm": 0.09108094125986099, + "learning_rate": 1.6898041865852974e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14600 + }, + { + "epoch": 4.250800116380565, + "flow_matching_loss": 0.014549478702247143, + "grad_norm": 0.07445013523101807, + "learning_rate": 1.6881638923753663e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14610 + }, + { + "epoch": 4.2537096304917075, + "flow_matching_loss": 0.015781715884804724, + "grad_norm": 0.16927161812782288, + "learning_rate": 1.6865233695285632e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14620 + }, + { + "epoch": 4.256619144602851, + "flow_matching_loss": 0.015330062061548234, + "grad_norm": 0.07640359550714493, + "learning_rate": 1.6848826200382782e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14630 + }, + { + "epoch": 4.259528658713995, + "flow_matching_loss": 0.014848456252366304, + "grad_norm": 0.069781593978405, + "learning_rate": 1.6832416458981762e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14640 + }, + { + "epoch": 4.262438172825139, + "flow_matching_loss": 0.014983478747308255, + "grad_norm": 0.0657932236790657, + "learning_rate": 1.681600449102196e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14650 + }, + { + "epoch": 4.265347686936281, + "flow_matching_loss": 0.01493974458426237, + "grad_norm": 0.07052794098854065, + "learning_rate": 1.6799590316445468e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14660 + }, + { + "epoch": 4.268257201047425, + "flow_matching_loss": 0.014247909095138312, + "grad_norm": 0.08006411790847778, + "learning_rate": 1.678317395519705e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14670 + }, + { + "epoch": 4.271166715158569, + "flow_matching_loss": 0.016440741252154112, + "grad_norm": 0.06676370650529861, + "learning_rate": 1.676675542722413e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14680 + }, + { + "epoch": 4.274076229269712, + "flow_matching_loss": 0.014415490347892046, + "grad_norm": 0.07308398187160492, + "learning_rate": 1.6750334752476776e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14690 + }, + { + "epoch": 4.276985743380855, + "flow_matching_loss": 0.014666675683110952, + "grad_norm": 0.09987874329090118, + "learning_rate": 1.673391195090765e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14700 + }, + { + "epoch": 4.279895257491999, + "flow_matching_loss": 0.015028753224760294, + "grad_norm": 0.09422009438276291, + "learning_rate": 1.6717487042472006e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14710 + }, + { + "epoch": 4.282804771603142, + "flow_matching_loss": 0.016535019408911465, + "grad_norm": 0.07700705528259277, + "learning_rate": 1.6701060047127655e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14720 + }, + { + "epoch": 4.285714285714286, + "flow_matching_loss": 0.014412735030055046, + "grad_norm": 0.07577886432409286, + "learning_rate": 1.6684630984834947e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14730 + }, + { + "epoch": 4.288623799825429, + "flow_matching_loss": 0.01600094074383378, + "grad_norm": 0.07607541233301163, + "learning_rate": 1.666819987555675e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14740 + }, + { + "epoch": 4.291533313936573, + "flow_matching_loss": 0.016725027095526458, + "grad_norm": 0.07396673411130905, + "learning_rate": 1.6651766739258392e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14750 + }, + { + "epoch": 4.294442828047716, + "flow_matching_loss": 0.01592404404655099, + "grad_norm": 0.09739674627780914, + "learning_rate": 1.6635331595907704e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14760 + }, + { + "epoch": 4.2973523421588595, + "flow_matching_loss": 0.014485918916761875, + "grad_norm": 0.09431733936071396, + "learning_rate": 1.6618894465474923e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14770 + }, + { + "epoch": 4.300261856270003, + "flow_matching_loss": 0.015085088741034269, + "grad_norm": 0.09905322641134262, + "learning_rate": 1.660245536793271e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14780 + }, + { + "epoch": 4.303171370381146, + "flow_matching_loss": 0.015448873769491911, + "grad_norm": 0.09636103361845016, + "learning_rate": 1.6586014323256126e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14790 + }, + { + "epoch": 4.30608088449229, + "flow_matching_loss": 0.01539082983508706, + "grad_norm": 0.08113239705562592, + "learning_rate": 1.656957135142259e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14800 + }, + { + "epoch": 4.3089903986034335, + "flow_matching_loss": 0.0148838322609663, + "grad_norm": 0.09380308538675308, + "learning_rate": 1.655312647241186e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14810 + }, + { + "epoch": 4.311899912714576, + "flow_matching_loss": 0.01625382425263524, + "grad_norm": 0.08528212457895279, + "learning_rate": 1.6536679706206015e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14820 + }, + { + "epoch": 4.31480942682572, + "flow_matching_loss": 0.015068733505904675, + "grad_norm": 0.07940470427274704, + "learning_rate": 1.6520231072789427e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14830 + }, + { + "epoch": 4.317718940936864, + "flow_matching_loss": 0.013834102544933558, + "grad_norm": 0.07088273763656616, + "learning_rate": 1.6503780592148737e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14840 + }, + { + "epoch": 4.320628455048007, + "flow_matching_loss": 0.015843819454312326, + "grad_norm": 0.07295680046081543, + "learning_rate": 1.648732828427283e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14850 + }, + { + "epoch": 4.32353796915915, + "flow_matching_loss": 0.014252823125571013, + "grad_norm": 0.08095297962427139, + "learning_rate": 1.647087416915281e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14860 + }, + { + "epoch": 4.326447483270294, + "flow_matching_loss": 0.01428950224071741, + "grad_norm": 0.08719556033611298, + "learning_rate": 1.6454418266781987e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14870 + }, + { + "epoch": 4.329356997381438, + "flow_matching_loss": 0.015582555811852217, + "grad_norm": 0.08710618317127228, + "learning_rate": 1.6437960597155824e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14880 + }, + { + "epoch": 4.3322665114925805, + "flow_matching_loss": 0.01577791469171643, + "grad_norm": 0.08142226934432983, + "learning_rate": 1.642150118027194e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14890 + }, + { + "epoch": 4.335176025603724, + "flow_matching_loss": 0.01616771863773465, + "grad_norm": 0.09000484645366669, + "learning_rate": 1.640504003613009e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14900 + }, + { + "epoch": 4.338085539714868, + "flow_matching_loss": 0.015400654543191195, + "grad_norm": 0.06749879568815231, + "learning_rate": 1.6388577184732107e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14910 + }, + { + "epoch": 4.340995053826011, + "flow_matching_loss": 0.015366263035684825, + "grad_norm": 0.07650014758110046, + "learning_rate": 1.6372112646081907e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14920 + }, + { + "epoch": 4.343904567937154, + "flow_matching_loss": 0.015836974140256643, + "grad_norm": 0.06675205379724503, + "learning_rate": 1.6355646440185464e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14930 + }, + { + "epoch": 4.346814082048298, + "flow_matching_loss": 0.01551452288404107, + "grad_norm": 0.09733577072620392, + "learning_rate": 1.6339178587050764e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14940 + }, + { + "epoch": 4.349723596159441, + "flow_matching_loss": 0.015182586666196585, + "grad_norm": 0.08910001814365387, + "learning_rate": 1.632270910668781e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14950 + }, + { + "epoch": 4.352633110270585, + "flow_matching_loss": 0.01521865427494049, + "grad_norm": 0.08331349492073059, + "learning_rate": 1.6306238019108562e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14960 + }, + { + "epoch": 4.355542624381728, + "flow_matching_loss": 0.016933780629187822, + "grad_norm": 0.09820353984832764, + "learning_rate": 1.6289765344326952e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14970 + }, + { + "epoch": 4.358452138492872, + "flow_matching_loss": 0.01611119685694575, + "grad_norm": 0.07037333399057388, + "learning_rate": 1.6273291102358833e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14980 + }, + { + "epoch": 4.361361652604015, + "flow_matching_loss": 0.01459206622093916, + "grad_norm": 0.07711095362901688, + "learning_rate": 1.6256815313221963e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 14990 + }, + { + "epoch": 4.364271166715159, + "flow_matching_loss": 0.01418785136193037, + "grad_norm": 0.06798174232244492, + "learning_rate": 1.6240337996935972e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15000 + }, + { + "epoch": 4.367180680826302, + "flow_matching_loss": 0.015333021432161332, + "grad_norm": 0.08265598863363266, + "learning_rate": 1.6223859173522367e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15010 + }, + { + "epoch": 4.370090194937445, + "flow_matching_loss": 0.01542681697756052, + "grad_norm": 0.06597740203142166, + "learning_rate": 1.6207378863004455e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15020 + }, + { + "epoch": 4.372999709048589, + "flow_matching_loss": 0.014093794021755457, + "grad_norm": 0.06504916399717331, + "learning_rate": 1.6190897085407382e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15030 + }, + { + "epoch": 4.3759092231597325, + "flow_matching_loss": 0.016944225318729876, + "grad_norm": 0.08285428583621979, + "learning_rate": 1.617441386075805e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15040 + }, + { + "epoch": 4.378818737270876, + "flow_matching_loss": 0.01644668811932206, + "grad_norm": 0.10724488645792007, + "learning_rate": 1.6157929209085142e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15050 + }, + { + "epoch": 4.381728251382019, + "flow_matching_loss": 0.014182469993829726, + "grad_norm": 0.06410989165306091, + "learning_rate": 1.614144315041906e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15060 + }, + { + "epoch": 4.384637765493163, + "flow_matching_loss": 0.017377880215644837, + "grad_norm": 0.07490792870521545, + "learning_rate": 1.612495570479192e-05, + "loss": 0.0174, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15070 + }, + { + "epoch": 4.3875472796043065, + "flow_matching_loss": 0.016282394342124463, + "grad_norm": 0.08397471904754639, + "learning_rate": 1.6108466892237517e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15080 + }, + { + "epoch": 4.390456793715449, + "flow_matching_loss": 0.014289507921785117, + "grad_norm": 0.07151994854211807, + "learning_rate": 1.609197673279132e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15090 + }, + { + "epoch": 4.393366307826593, + "flow_matching_loss": 0.014390549622476101, + "grad_norm": 0.09439928084611893, + "learning_rate": 1.6075485246490428e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15100 + }, + { + "epoch": 4.396275821937737, + "flow_matching_loss": 0.014597580954432487, + "grad_norm": 0.08527495712041855, + "learning_rate": 1.605899245337355e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15110 + }, + { + "epoch": 4.3991853360488795, + "flow_matching_loss": 0.014864209294319152, + "grad_norm": 0.08194050937891006, + "learning_rate": 1.6042498373480983e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15120 + }, + { + "epoch": 4.402094850160023, + "flow_matching_loss": 0.015257677622139454, + "grad_norm": 0.09441344439983368, + "learning_rate": 1.602600302685459e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15130 + }, + { + "epoch": 4.405004364271167, + "flow_matching_loss": 0.015479239728301763, + "grad_norm": 0.08283742517232895, + "learning_rate": 1.6009506433537774e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15140 + }, + { + "epoch": 4.40791387838231, + "flow_matching_loss": 0.014675028994679451, + "grad_norm": 0.08990555256605148, + "learning_rate": 1.599300861357545e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15150 + }, + { + "epoch": 4.4108233924934535, + "flow_matching_loss": 0.014239749405533075, + "grad_norm": 0.07100869715213776, + "learning_rate": 1.5976509587014032e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15160 + }, + { + "epoch": 4.413732906604597, + "flow_matching_loss": 0.01515564313158393, + "grad_norm": 0.04680229723453522, + "learning_rate": 1.5960009373901384e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15170 + }, + { + "epoch": 4.416642420715741, + "flow_matching_loss": 0.015328969340771437, + "grad_norm": 0.0928611159324646, + "learning_rate": 1.5943507994286823e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15180 + }, + { + "epoch": 4.419551934826884, + "flow_matching_loss": 0.014646376390010119, + "grad_norm": 0.10583382099866867, + "learning_rate": 1.5927005468221082e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15190 + }, + { + "epoch": 4.422461448938027, + "flow_matching_loss": 0.014649094082415104, + "grad_norm": 0.0882863849401474, + "learning_rate": 1.5910501815756285e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15200 + }, + { + "epoch": 4.425370963049171, + "flow_matching_loss": 0.017967935465276242, + "grad_norm": 0.10499437898397446, + "learning_rate": 1.589399705694593e-05, + "loss": 0.018, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15210 + }, + { + "epoch": 4.428280477160314, + "flow_matching_loss": 0.013886073790490627, + "grad_norm": 0.07281343638896942, + "learning_rate": 1.587749121184485e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15220 + }, + { + "epoch": 4.431189991271458, + "flow_matching_loss": 0.014548630453646183, + "grad_norm": 0.07535326480865479, + "learning_rate": 1.5860984300509206e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15230 + }, + { + "epoch": 4.434099505382601, + "flow_matching_loss": 0.01316566588357091, + "grad_norm": 0.0699094906449318, + "learning_rate": 1.5844476342996453e-05, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15240 + }, + { + "epoch": 4.437009019493744, + "flow_matching_loss": 0.013337842281907797, + "grad_norm": 0.06649422645568848, + "learning_rate": 1.5827967359365306e-05, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15250 + }, + { + "epoch": 4.439918533604888, + "flow_matching_loss": 0.014036919921636581, + "grad_norm": 0.08271580189466476, + "learning_rate": 1.5811457369675758e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15260 + }, + { + "epoch": 4.442828047716032, + "flow_matching_loss": 0.014852597378194332, + "grad_norm": 0.08624443411827087, + "learning_rate": 1.5794946393988982e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15270 + }, + { + "epoch": 4.445737561827175, + "flow_matching_loss": 0.014842741750180721, + "grad_norm": 0.07787185907363892, + "learning_rate": 1.5778434452367382e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15280 + }, + { + "epoch": 4.448647075938318, + "flow_matching_loss": 0.014834370836615563, + "grad_norm": 0.06576157361268997, + "learning_rate": 1.576192156487452e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15290 + }, + { + "epoch": 4.451556590049462, + "flow_matching_loss": 0.014667247980833053, + "grad_norm": 0.08448494970798492, + "learning_rate": 1.574540775157511e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15300 + }, + { + "epoch": 4.4544661041606055, + "flow_matching_loss": 0.016063040308654308, + "grad_norm": 0.10453368723392487, + "learning_rate": 1.5728893032534998e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15310 + }, + { + "epoch": 4.457375618271748, + "flow_matching_loss": 0.013180234655737876, + "grad_norm": 0.07684655487537384, + "learning_rate": 1.571237742782112e-05, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15320 + }, + { + "epoch": 4.460285132382892, + "flow_matching_loss": 0.014059068448841573, + "grad_norm": 0.070323646068573, + "learning_rate": 1.569586095750149e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15330 + }, + { + "epoch": 4.463194646494036, + "flow_matching_loss": 0.015471447538584471, + "grad_norm": 0.09130179136991501, + "learning_rate": 1.5679343641645193e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15340 + }, + { + "epoch": 4.466104160605179, + "flow_matching_loss": 0.01513359798118472, + "grad_norm": 0.07127727568149567, + "learning_rate": 1.5662825500322304e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15350 + }, + { + "epoch": 4.469013674716322, + "flow_matching_loss": 0.016832927241921425, + "grad_norm": 0.07597096264362335, + "learning_rate": 1.5646306553603932e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15360 + }, + { + "epoch": 4.471923188827466, + "flow_matching_loss": 0.014380946848541498, + "grad_norm": 0.08015502244234085, + "learning_rate": 1.5629786821562154e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15370 + }, + { + "epoch": 4.47483270293861, + "flow_matching_loss": 0.016526215616613625, + "grad_norm": 0.09757106751203537, + "learning_rate": 1.561326632427001e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15380 + }, + { + "epoch": 4.4777422170497525, + "flow_matching_loss": 0.014515587501227856, + "grad_norm": 0.08200182020664215, + "learning_rate": 1.5596745081801447e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15390 + }, + { + "epoch": 4.480651731160896, + "flow_matching_loss": 0.014161148760467767, + "grad_norm": 0.09396179765462875, + "learning_rate": 1.5580223114231348e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15400 + }, + { + "epoch": 4.48356124527204, + "flow_matching_loss": 0.01577649572864175, + "grad_norm": 0.09525942802429199, + "learning_rate": 1.5563700441635445e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15410 + }, + { + "epoch": 4.486470759383183, + "flow_matching_loss": 0.015010017901659012, + "grad_norm": 0.08346281945705414, + "learning_rate": 1.5547177084090358e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15420 + }, + { + "epoch": 4.4893802734943264, + "flow_matching_loss": 0.013263867516070605, + "grad_norm": 0.05908859521150589, + "learning_rate": 1.5530653061673514e-05, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15430 + }, + { + "epoch": 4.49228978760547, + "flow_matching_loss": 0.01365661844611168, + "grad_norm": 0.067902572453022, + "learning_rate": 1.5514128394463167e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15440 + }, + { + "epoch": 4.495199301716613, + "flow_matching_loss": 0.01675456827506423, + "grad_norm": 0.1054021418094635, + "learning_rate": 1.549760310253834e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15450 + }, + { + "epoch": 4.498108815827757, + "flow_matching_loss": 0.013608244433999061, + "grad_norm": 0.07637651264667511, + "learning_rate": 1.5481077205978816e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15460 + }, + { + "epoch": 4.5010183299389, + "flow_matching_loss": 0.014567085448652506, + "grad_norm": 0.07157211750745773, + "learning_rate": 1.5464550724865135e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15470 + }, + { + "epoch": 4.503927844050043, + "flow_matching_loss": 0.015715322364121674, + "grad_norm": 0.09493041783571243, + "learning_rate": 1.5448023679278517e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15480 + }, + { + "epoch": 4.506837358161187, + "flow_matching_loss": 0.01412515128031373, + "grad_norm": 0.08700806647539139, + "learning_rate": 1.543149608930088e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15490 + }, + { + "epoch": 4.509746872272331, + "flow_matching_loss": 0.015441437344998121, + "grad_norm": 0.07135587930679321, + "learning_rate": 1.5414967975014805e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15500 + }, + { + "epoch": 4.512656386383474, + "flow_matching_loss": 0.01538073793053627, + "grad_norm": 0.07725987583398819, + "learning_rate": 1.539843935650351e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15510 + }, + { + "epoch": 4.515565900494617, + "flow_matching_loss": 0.014720884896814824, + "grad_norm": 0.08178314566612244, + "learning_rate": 1.5381910253850834e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15520 + }, + { + "epoch": 4.518475414605761, + "flow_matching_loss": 0.01633389964699745, + "grad_norm": 0.08916303515434265, + "learning_rate": 1.5365380687141183e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15530 + }, + { + "epoch": 4.521384928716905, + "flow_matching_loss": 0.015334509406238794, + "grad_norm": 0.07560893893241882, + "learning_rate": 1.5348850676459544e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15540 + }, + { + "epoch": 4.524294442828047, + "flow_matching_loss": 0.014249600749462844, + "grad_norm": 0.06754286587238312, + "learning_rate": 1.5332320241891438e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15550 + }, + { + "epoch": 4.527203956939191, + "flow_matching_loss": 0.01642288202419877, + "grad_norm": 0.0945645347237587, + "learning_rate": 1.5315789403522908e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15560 + }, + { + "epoch": 4.530113471050335, + "flow_matching_loss": 0.015222803503274918, + "grad_norm": 0.07667209208011627, + "learning_rate": 1.529925818144047e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15570 + }, + { + "epoch": 4.5330229851614785, + "flow_matching_loss": 0.014285815600305795, + "grad_norm": 0.06664964556694031, + "learning_rate": 1.5282726595731126e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15580 + }, + { + "epoch": 4.535932499272621, + "flow_matching_loss": 0.014713311195373535, + "grad_norm": 0.10595262795686722, + "learning_rate": 1.5266194666482304e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15590 + }, + { + "epoch": 4.538842013383765, + "flow_matching_loss": 0.01583753675222397, + "grad_norm": 0.07361612468957901, + "learning_rate": 1.5249662413781866e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15600 + }, + { + "epoch": 4.541751527494909, + "flow_matching_loss": 0.014366567693650722, + "grad_norm": 0.10659811645746231, + "learning_rate": 1.5233129857718054e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15610 + }, + { + "epoch": 4.5446610416060516, + "flow_matching_loss": 0.013162621669471264, + "grad_norm": 0.06413508206605911, + "learning_rate": 1.521659701837948e-05, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15620 + }, + { + "epoch": 4.547570555717195, + "flow_matching_loss": 0.015040202997624874, + "grad_norm": 0.11040350049734116, + "learning_rate": 1.5200063915855107e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15630 + }, + { + "epoch": 4.550480069828339, + "flow_matching_loss": 0.015007389895617961, + "grad_norm": 0.06700676679611206, + "learning_rate": 1.5183530570234212e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15640 + }, + { + "epoch": 4.553389583939482, + "flow_matching_loss": 0.014909560140222312, + "grad_norm": 0.06634991616010666, + "learning_rate": 1.5166997001606361e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15650 + }, + { + "epoch": 4.5562990980506255, + "flow_matching_loss": 0.014362081233412028, + "grad_norm": 0.0760212168097496, + "learning_rate": 1.5150463230061417e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15660 + }, + { + "epoch": 4.559208612161769, + "flow_matching_loss": 0.015115749556571245, + "grad_norm": 0.06701108813285828, + "learning_rate": 1.5133929275689455e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15670 + }, + { + "epoch": 4.562118126272912, + "flow_matching_loss": 0.014503232389688491, + "grad_norm": 0.08152595162391663, + "learning_rate": 1.5117395158580795e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15680 + }, + { + "epoch": 4.565027640384056, + "flow_matching_loss": 0.014756903238594531, + "grad_norm": 0.07388801127672195, + "learning_rate": 1.510086089882595e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15690 + }, + { + "epoch": 4.567937154495199, + "flow_matching_loss": 0.015100438427180052, + "grad_norm": 0.09187298268079758, + "learning_rate": 1.5084326516515599e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15700 + }, + { + "epoch": 4.570846668606343, + "flow_matching_loss": 0.015293651074171067, + "grad_norm": 0.06979867815971375, + "learning_rate": 1.5067792031740582e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15710 + }, + { + "epoch": 4.573756182717486, + "flow_matching_loss": 0.01579747796058655, + "grad_norm": 0.11245065927505493, + "learning_rate": 1.505125746459185e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15720 + }, + { + "epoch": 4.57666569682863, + "flow_matching_loss": 0.014778692740947008, + "grad_norm": 0.08328581601381302, + "learning_rate": 1.5034722835160466e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15730 + }, + { + "epoch": 4.579575210939773, + "flow_matching_loss": 0.01351548582315445, + "grad_norm": 0.07289411872625351, + "learning_rate": 1.5018188163537566e-05, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15740 + }, + { + "epoch": 4.582484725050916, + "flow_matching_loss": 0.014577453676611184, + "grad_norm": 0.07238859683275223, + "learning_rate": 1.5001653469814331e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15750 + }, + { + "epoch": 4.58539423916206, + "flow_matching_loss": 0.014226112328469754, + "grad_norm": 0.0770384669303894, + "learning_rate": 1.4985118774081978e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15760 + }, + { + "epoch": 4.588303753273204, + "flow_matching_loss": 0.014784551784396171, + "grad_norm": 0.08075158298015594, + "learning_rate": 1.4968584096431708e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15770 + }, + { + "epoch": 4.591213267384346, + "flow_matching_loss": 0.01476367199793458, + "grad_norm": 0.07672468572854996, + "learning_rate": 1.4952049456954731e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15780 + }, + { + "epoch": 4.59412278149549, + "flow_matching_loss": 0.015205531008541584, + "grad_norm": 0.08210346102714539, + "learning_rate": 1.4935514875742182e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15790 + }, + { + "epoch": 4.597032295606634, + "flow_matching_loss": 0.013733066245913505, + "grad_norm": 0.08624155819416046, + "learning_rate": 1.4918980372885143e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15800 + }, + { + "epoch": 4.599941809717777, + "flow_matching_loss": 0.01529159201309085, + "grad_norm": 0.08158335089683533, + "learning_rate": 1.4902445968474591e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15810 + }, + { + "epoch": 4.60285132382892, + "flow_matching_loss": 0.015428535081446171, + "grad_norm": 0.07578372210264206, + "learning_rate": 1.4885911682601378e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15820 + }, + { + "epoch": 4.605760837940064, + "flow_matching_loss": 0.014836748037487269, + "grad_norm": 0.06659610569477081, + "learning_rate": 1.4869377535356235e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15830 + }, + { + "epoch": 4.608670352051208, + "flow_matching_loss": 0.014191420469433069, + "grad_norm": 0.08022090047597885, + "learning_rate": 1.4852843546829704e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15840 + }, + { + "epoch": 4.611579866162351, + "flow_matching_loss": 0.015397784486413001, + "grad_norm": 0.0815025120973587, + "learning_rate": 1.483630973711214e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15850 + }, + { + "epoch": 4.614489380273494, + "flow_matching_loss": 0.014936479926109313, + "grad_norm": 0.09465974569320679, + "learning_rate": 1.4819776126293677e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15860 + }, + { + "epoch": 4.617398894384638, + "flow_matching_loss": 0.014495617244392633, + "grad_norm": 0.09433960914611816, + "learning_rate": 1.4803242734464214e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15870 + }, + { + "epoch": 4.620308408495781, + "flow_matching_loss": 0.01459124181419611, + "grad_norm": 0.06622006744146347, + "learning_rate": 1.4786709581713383e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15880 + }, + { + "epoch": 4.6232179226069245, + "flow_matching_loss": 0.015604551695287228, + "grad_norm": 0.08048195391893387, + "learning_rate": 1.4770176688130524e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15890 + }, + { + "epoch": 4.626127436718068, + "flow_matching_loss": 0.016903050895780324, + "grad_norm": 0.07583972066640854, + "learning_rate": 1.475364407380466e-05, + "loss": 0.0169, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15900 + }, + { + "epoch": 4.629036950829212, + "flow_matching_loss": 0.01428276700899005, + "grad_norm": 0.09281301498413086, + "learning_rate": 1.4737111758824476e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15910 + }, + { + "epoch": 4.631946464940355, + "flow_matching_loss": 0.014665240235626698, + "grad_norm": 0.08737551420927048, + "learning_rate": 1.4720579763278289e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15920 + }, + { + "epoch": 4.6348559790514985, + "flow_matching_loss": 0.01631218856200576, + "grad_norm": 0.07040282338857651, + "learning_rate": 1.4704048107254042e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15930 + }, + { + "epoch": 4.637765493162642, + "flow_matching_loss": 0.01603719647973776, + "grad_norm": 0.07014825940132141, + "learning_rate": 1.4687516810839246e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15940 + }, + { + "epoch": 4.640675007273785, + "flow_matching_loss": 0.01623982209712267, + "grad_norm": 0.10088696330785751, + "learning_rate": 1.4670985894120997e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15950 + }, + { + "epoch": 4.643584521384929, + "flow_matching_loss": 0.014888141863048076, + "grad_norm": 0.07753296941518784, + "learning_rate": 1.465445537718591e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15960 + }, + { + "epoch": 4.646494035496072, + "flow_matching_loss": 0.014034479949623346, + "grad_norm": 0.0899086520075798, + "learning_rate": 1.463792528012012e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15970 + }, + { + "epoch": 4.649403549607215, + "flow_matching_loss": 0.017823790106922387, + "grad_norm": 0.056695349514484406, + "learning_rate": 1.4621395623009252e-05, + "loss": 0.0178, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15980 + }, + { + "epoch": 4.652313063718359, + "flow_matching_loss": 0.01570156291127205, + "grad_norm": 0.07974360883235931, + "learning_rate": 1.4604866425938407e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 15990 + }, + { + "epoch": 4.655222577829503, + "flow_matching_loss": 0.014509990531951188, + "grad_norm": 0.07683538645505905, + "learning_rate": 1.4588337708992115e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16000 + }, + { + "epoch": 4.6581320919406455, + "flow_matching_loss": 0.014841016102582216, + "grad_norm": 0.06340067833662033, + "learning_rate": 1.4571809492254318e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16010 + }, + { + "epoch": 4.661041606051789, + "flow_matching_loss": 0.017096192669123413, + "grad_norm": 0.09022312611341476, + "learning_rate": 1.4555281795808371e-05, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16020 + }, + { + "epoch": 4.663951120162933, + "flow_matching_loss": 0.015209637023508549, + "grad_norm": 0.10046033561229706, + "learning_rate": 1.4538754639736966e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16030 + }, + { + "epoch": 4.666860634274077, + "flow_matching_loss": 0.013318375684320927, + "grad_norm": 0.07222380489110947, + "learning_rate": 1.4522228044122174e-05, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16040 + }, + { + "epoch": 4.669770148385219, + "flow_matching_loss": 0.015881820116192102, + "grad_norm": 0.08935004472732544, + "learning_rate": 1.4505702029045359e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16050 + }, + { + "epoch": 4.672679662496363, + "flow_matching_loss": 0.015626498591154815, + "grad_norm": 0.09403254091739655, + "learning_rate": 1.4489176614587189e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16060 + }, + { + "epoch": 4.675589176607507, + "flow_matching_loss": 0.014775883965194225, + "grad_norm": 0.09314518421888351, + "learning_rate": 1.4472651820827595e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16070 + }, + { + "epoch": 4.67849869071865, + "flow_matching_loss": 0.015222343523055315, + "grad_norm": 0.07281845808029175, + "learning_rate": 1.4456127667845766e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16080 + }, + { + "epoch": 4.681408204829793, + "flow_matching_loss": 0.01492083128541708, + "grad_norm": 0.0864255428314209, + "learning_rate": 1.4439604175720104e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16090 + }, + { + "epoch": 4.684317718940937, + "flow_matching_loss": 0.014423668105155229, + "grad_norm": 0.07333975285291672, + "learning_rate": 1.4423081364528212e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16100 + }, + { + "epoch": 4.687227233052081, + "flow_matching_loss": 0.014844511914998294, + "grad_norm": 0.10218643397092819, + "learning_rate": 1.4406559254346863e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16110 + }, + { + "epoch": 4.690136747163224, + "flow_matching_loss": 0.015045746695250273, + "grad_norm": 0.08316023647785187, + "learning_rate": 1.4390037865251978e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16120 + }, + { + "epoch": 4.693046261274367, + "flow_matching_loss": 0.014966365415602922, + "grad_norm": 0.0767529159784317, + "learning_rate": 1.4373517217318597e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16130 + }, + { + "epoch": 4.69595577538551, + "flow_matching_loss": 0.014191563613712788, + "grad_norm": 0.11528179794549942, + "learning_rate": 1.4356997330620875e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16140 + }, + { + "epoch": 4.698865289496654, + "flow_matching_loss": 0.014259442035108805, + "grad_norm": 0.06687159836292267, + "learning_rate": 1.434047822523203e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16150 + }, + { + "epoch": 4.7017748036077975, + "flow_matching_loss": 0.016797532420605422, + "grad_norm": 0.0910755917429924, + "learning_rate": 1.4323959921224322e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16160 + }, + { + "epoch": 4.704684317718941, + "flow_matching_loss": 0.014089447166770696, + "grad_norm": 0.07469021528959274, + "learning_rate": 1.4307442438669066e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16170 + }, + { + "epoch": 4.707593831830084, + "flow_matching_loss": 0.015136717073619365, + "grad_norm": 0.07012826949357986, + "learning_rate": 1.4290925797636546e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16180 + }, + { + "epoch": 4.710503345941228, + "flow_matching_loss": 0.014760059583932162, + "grad_norm": 0.07804161310195923, + "learning_rate": 1.4274410018196039e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16190 + }, + { + "epoch": 4.7134128600523715, + "flow_matching_loss": 0.015410854015499353, + "grad_norm": 0.06996399909257889, + "learning_rate": 1.4257895120415783e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16200 + }, + { + "epoch": 4.716322374163514, + "flow_matching_loss": 0.016130581311881543, + "grad_norm": 0.10762904584407806, + "learning_rate": 1.4241381124362933e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16210 + }, + { + "epoch": 4.719231888274658, + "flow_matching_loss": 0.016711472440510987, + "grad_norm": 0.10230723023414612, + "learning_rate": 1.4224868050103546e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16220 + }, + { + "epoch": 4.722141402385802, + "flow_matching_loss": 0.015211268421262502, + "grad_norm": 0.07675371319055557, + "learning_rate": 1.4208355917702573e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16230 + }, + { + "epoch": 4.725050916496945, + "flow_matching_loss": 0.014527850318700075, + "grad_norm": 0.07369499653577805, + "learning_rate": 1.41918447472238e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16240 + }, + { + "epoch": 4.727960430608088, + "flow_matching_loss": 0.015113794337958097, + "grad_norm": 0.0841701328754425, + "learning_rate": 1.4175334558729869e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16250 + }, + { + "epoch": 4.730869944719232, + "flow_matching_loss": 0.015314390882849693, + "grad_norm": 0.08355673402547836, + "learning_rate": 1.4158825372282211e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16260 + }, + { + "epoch": 4.733779458830376, + "flow_matching_loss": 0.014605069998651743, + "grad_norm": 0.0753682553768158, + "learning_rate": 1.4142317207941042e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16270 + }, + { + "epoch": 4.7366889729415185, + "flow_matching_loss": 0.01531632049009204, + "grad_norm": 0.06247171014547348, + "learning_rate": 1.4125810085765339e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16280 + }, + { + "epoch": 4.739598487052662, + "flow_matching_loss": 0.016668537817895412, + "grad_norm": 0.07525069266557693, + "learning_rate": 1.410930402581281e-05, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16290 + }, + { + "epoch": 4.742508001163806, + "flow_matching_loss": 0.014887715689837933, + "grad_norm": 0.10488765686750412, + "learning_rate": 1.4092799048139876e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16300 + }, + { + "epoch": 4.745417515274949, + "flow_matching_loss": 0.01480203978717327, + "grad_norm": 0.06271260976791382, + "learning_rate": 1.4076295172801647e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16310 + }, + { + "epoch": 4.748327029386092, + "flow_matching_loss": 0.013847177848219872, + "grad_norm": 0.08104492723941803, + "learning_rate": 1.4059792419851884e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16320 + }, + { + "epoch": 4.751236543497236, + "flow_matching_loss": 0.01598479747772217, + "grad_norm": 0.06800654530525208, + "learning_rate": 1.4043290809342984e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16330 + }, + { + "epoch": 4.754146057608379, + "flow_matching_loss": 0.014363607112318277, + "grad_norm": 0.07468225061893463, + "learning_rate": 1.4026790361325958e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16340 + }, + { + "epoch": 4.757055571719523, + "flow_matching_loss": 0.015031483210623264, + "grad_norm": 0.07671065628528595, + "learning_rate": 1.4010291095850415e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16350 + }, + { + "epoch": 4.759965085830666, + "flow_matching_loss": 0.014723558444529772, + "grad_norm": 0.06958729773759842, + "learning_rate": 1.3993793032964515e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16360 + }, + { + "epoch": 4.76287459994181, + "flow_matching_loss": 0.015378311835229396, + "grad_norm": 0.10647297650575638, + "learning_rate": 1.3977296192714956e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16370 + }, + { + "epoch": 4.765784114052953, + "flow_matching_loss": 0.014283940754830837, + "grad_norm": 0.08347438275814056, + "learning_rate": 1.396080059514696e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16380 + }, + { + "epoch": 4.768693628164097, + "flow_matching_loss": 0.01661686636507511, + "grad_norm": 0.10428116470575333, + "learning_rate": 1.3944306260304232e-05, + "loss": 0.0166, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16390 + }, + { + "epoch": 4.77160314227524, + "flow_matching_loss": 0.014718637522310018, + "grad_norm": 0.08969713002443314, + "learning_rate": 1.3927813208228936e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16400 + }, + { + "epoch": 4.774512656386383, + "flow_matching_loss": 0.014074893668293953, + "grad_norm": 0.07141360640525818, + "learning_rate": 1.39113214589617e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16410 + }, + { + "epoch": 4.777422170497527, + "flow_matching_loss": 0.01377620417624712, + "grad_norm": 0.08808831125497818, + "learning_rate": 1.3894831032541545e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16420 + }, + { + "epoch": 4.7803316846086705, + "flow_matching_loss": 0.014477548841387033, + "grad_norm": 0.08860398828983307, + "learning_rate": 1.3878341949005895e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16430 + }, + { + "epoch": 4.783241198719814, + "flow_matching_loss": 0.015191169362515211, + "grad_norm": 0.09091860800981522, + "learning_rate": 1.3861854228390546e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16440 + }, + { + "epoch": 4.786150712830957, + "flow_matching_loss": 0.014979330636560918, + "grad_norm": 0.09062525629997253, + "learning_rate": 1.3845367890729621e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16450 + }, + { + "epoch": 4.789060226942101, + "flow_matching_loss": 0.015573734883219004, + "grad_norm": 0.07882576435804367, + "learning_rate": 1.3828882956055591e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16460 + }, + { + "epoch": 4.7919697410532445, + "flow_matching_loss": 0.013788984064012766, + "grad_norm": 0.062931589782238, + "learning_rate": 1.38123994443992e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16470 + }, + { + "epoch": 4.794879255164387, + "flow_matching_loss": 0.013524109032005072, + "grad_norm": 0.0953712910413742, + "learning_rate": 1.3795917375789465e-05, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16480 + }, + { + "epoch": 4.797788769275531, + "flow_matching_loss": 0.014933544024825097, + "grad_norm": 0.08583877980709076, + "learning_rate": 1.3779436770253658e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16490 + }, + { + "epoch": 4.800698283386675, + "flow_matching_loss": 0.014073549304157495, + "grad_norm": 0.0683288499712944, + "learning_rate": 1.3762957647817258e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16500 + }, + { + "epoch": 4.8036077974978175, + "flow_matching_loss": 0.013906251080334187, + "grad_norm": 0.08360769599676132, + "learning_rate": 1.374648002850397e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16510 + }, + { + "epoch": 4.806517311608961, + "flow_matching_loss": 0.013594860676676035, + "grad_norm": 0.08541896194219589, + "learning_rate": 1.3730003932335642e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16520 + }, + { + "epoch": 4.809426825720105, + "flow_matching_loss": 0.014240224845707417, + "grad_norm": 0.07079464197158813, + "learning_rate": 1.3713529379332292e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16530 + }, + { + "epoch": 4.812336339831248, + "flow_matching_loss": 0.014295293670147657, + "grad_norm": 0.0695660263299942, + "learning_rate": 1.3697056389512054e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16540 + }, + { + "epoch": 4.8152458539423915, + "flow_matching_loss": 0.014810548722743988, + "grad_norm": 0.08169253915548325, + "learning_rate": 1.3680584982891162e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16550 + }, + { + "epoch": 4.818155368053535, + "flow_matching_loss": 0.013042308017611503, + "grad_norm": 0.07010281085968018, + "learning_rate": 1.3664115179483921e-05, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16560 + }, + { + "epoch": 4.821064882164679, + "flow_matching_loss": 0.014098097756505012, + "grad_norm": 0.07111611217260361, + "learning_rate": 1.364764699930271e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16570 + }, + { + "epoch": 4.823974396275822, + "flow_matching_loss": 0.013053113222122192, + "grad_norm": 0.06209981068968773, + "learning_rate": 1.363118046235791e-05, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16580 + }, + { + "epoch": 4.826883910386965, + "flow_matching_loss": 0.013987190928310156, + "grad_norm": 0.0625721663236618, + "learning_rate": 1.3614715588657921e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16590 + }, + { + "epoch": 4.829793424498109, + "flow_matching_loss": 0.014419741928577423, + "grad_norm": 0.10793200135231018, + "learning_rate": 1.3598252398209118e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16600 + }, + { + "epoch": 4.832702938609252, + "flow_matching_loss": 0.016029440611600877, + "grad_norm": 0.07609381526708603, + "learning_rate": 1.358179091101582e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16610 + }, + { + "epoch": 4.835612452720396, + "flow_matching_loss": 0.014919188246130944, + "grad_norm": 0.06430615484714508, + "learning_rate": 1.3565331147080299e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16620 + }, + { + "epoch": 4.838521966831539, + "flow_matching_loss": 0.01675456315279007, + "grad_norm": 0.07089581340551376, + "learning_rate": 1.3548873126402716e-05, + "loss": 0.0168, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16630 + }, + { + "epoch": 4.841431480942683, + "flow_matching_loss": 0.01257228832691908, + "grad_norm": 0.0631260946393013, + "learning_rate": 1.3532416868981117e-05, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16640 + }, + { + "epoch": 4.844340995053826, + "flow_matching_loss": 0.015277350880205631, + "grad_norm": 0.08085840195417404, + "learning_rate": 1.3515962394811401e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16650 + }, + { + "epoch": 4.84725050916497, + "flow_matching_loss": 0.015563887171447277, + "grad_norm": 0.10093069821596146, + "learning_rate": 1.3499509723887313e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16660 + }, + { + "epoch": 4.850160023276112, + "flow_matching_loss": 0.014402544405311347, + "grad_norm": 0.08074311167001724, + "learning_rate": 1.3483058876200396e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16670 + }, + { + "epoch": 4.853069537387256, + "flow_matching_loss": 0.014598889648914338, + "grad_norm": 0.06168543919920921, + "learning_rate": 1.3466609871739987e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16680 + }, + { + "epoch": 4.8559790514984, + "flow_matching_loss": 0.01435737805441022, + "grad_norm": 0.09279780834913254, + "learning_rate": 1.3450162730493172e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16690 + }, + { + "epoch": 4.8588885656095435, + "flow_matching_loss": 0.012929372768849135, + "grad_norm": 0.0820256695151329, + "learning_rate": 1.3433717472444781e-05, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16700 + }, + { + "epoch": 4.861798079720686, + "flow_matching_loss": 0.014277141448110342, + "grad_norm": 0.07564597576856613, + "learning_rate": 1.3417274117577343e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16710 + }, + { + "epoch": 4.86470759383183, + "flow_matching_loss": 0.015350951161235572, + "grad_norm": 0.07790574431419373, + "learning_rate": 1.3400832685871102e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16720 + }, + { + "epoch": 4.867617107942974, + "flow_matching_loss": 0.015819155611097813, + "grad_norm": 0.08499059826135635, + "learning_rate": 1.3384393197303936e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16730 + }, + { + "epoch": 4.870526622054117, + "flow_matching_loss": 0.014708140399307013, + "grad_norm": 0.09341372549533844, + "learning_rate": 1.336795567185138e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16740 + }, + { + "epoch": 4.87343613616526, + "flow_matching_loss": 0.014960625860840081, + "grad_norm": 0.07154164463281631, + "learning_rate": 1.3351520129486573e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16750 + }, + { + "epoch": 4.876345650276404, + "flow_matching_loss": 0.015358183439821005, + "grad_norm": 0.08872153609991074, + "learning_rate": 1.3335086590180252e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16760 + }, + { + "epoch": 4.879255164387548, + "flow_matching_loss": 0.015364951640367507, + "grad_norm": 0.07375823706388474, + "learning_rate": 1.3318655073900708e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16770 + }, + { + "epoch": 4.8821646784986905, + "flow_matching_loss": 0.014255099184811115, + "grad_norm": 0.07809901982545853, + "learning_rate": 1.3302225600613796e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16780 + }, + { + "epoch": 4.885074192609834, + "flow_matching_loss": 0.013852381426841021, + "grad_norm": 0.07617498934268951, + "learning_rate": 1.3285798190282869e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16790 + }, + { + "epoch": 4.887983706720978, + "flow_matching_loss": 0.014015466906130314, + "grad_norm": 0.0757310539484024, + "learning_rate": 1.3269372862868776e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16800 + }, + { + "epoch": 4.890893220832121, + "flow_matching_loss": 0.014522068481892348, + "grad_norm": 0.07520217448472977, + "learning_rate": 1.3252949638329845e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16810 + }, + { + "epoch": 4.8938027349432645, + "flow_matching_loss": 0.015661916136741637, + "grad_norm": 0.08118928968906403, + "learning_rate": 1.3236528536621834e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16820 + }, + { + "epoch": 4.896712249054408, + "flow_matching_loss": 0.015018646139651537, + "grad_norm": 0.0690036490559578, + "learning_rate": 1.3220109577697943e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16830 + }, + { + "epoch": 4.899621763165551, + "flow_matching_loss": 0.014931920357048512, + "grad_norm": 0.0732988640666008, + "learning_rate": 1.3203692781508745e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16840 + }, + { + "epoch": 4.902531277276695, + "flow_matching_loss": 0.015083711501210928, + "grad_norm": 0.0970691666007042, + "learning_rate": 1.3187278168002198e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16850 + }, + { + "epoch": 4.905440791387838, + "flow_matching_loss": 0.014582821540534497, + "grad_norm": 0.09434524178504944, + "learning_rate": 1.3170865757123606e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16860 + }, + { + "epoch": 4.908350305498981, + "flow_matching_loss": 0.014035007543861866, + "grad_norm": 0.0826631560921669, + "learning_rate": 1.3154455568815592e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16870 + }, + { + "epoch": 4.911259819610125, + "flow_matching_loss": 0.014662429131567479, + "grad_norm": 0.07778293639421463, + "learning_rate": 1.3138047623018087e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16880 + }, + { + "epoch": 4.914169333721269, + "flow_matching_loss": 0.014788937568664551, + "grad_norm": 0.09050532430410385, + "learning_rate": 1.3121641939668293e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16890 + }, + { + "epoch": 4.917078847832412, + "flow_matching_loss": 0.014965395629405975, + "grad_norm": 0.10781051963567734, + "learning_rate": 1.3105238538700659e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16900 + }, + { + "epoch": 4.919988361943555, + "flow_matching_loss": 0.013706042245030404, + "grad_norm": 0.08127835392951965, + "learning_rate": 1.3088837440046866e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16910 + }, + { + "epoch": 4.922897876054699, + "flow_matching_loss": 0.014666154328733683, + "grad_norm": 0.08154449611902237, + "learning_rate": 1.3072438663635787e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16920 + }, + { + "epoch": 4.925807390165843, + "flow_matching_loss": 0.014881400112062692, + "grad_norm": 0.08855606615543365, + "learning_rate": 1.3056042229393497e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16930 + }, + { + "epoch": 4.928716904276985, + "flow_matching_loss": 0.015088946744799613, + "grad_norm": 0.08492794632911682, + "learning_rate": 1.30396481572432e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16940 + }, + { + "epoch": 4.931626418388129, + "flow_matching_loss": 0.017518240306526423, + "grad_norm": 0.0709591656923294, + "learning_rate": 1.3023256467105237e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16950 + }, + { + "epoch": 4.934535932499273, + "flow_matching_loss": 0.013327594660222531, + "grad_norm": 0.0571567676961422, + "learning_rate": 1.3006867178897067e-05, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16960 + }, + { + "epoch": 4.9374454466104165, + "flow_matching_loss": 0.01435862695798278, + "grad_norm": 0.06819260865449905, + "learning_rate": 1.2990480312533213e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16970 + }, + { + "epoch": 4.940354960721559, + "flow_matching_loss": 0.01615955736488104, + "grad_norm": 0.06880558282136917, + "learning_rate": 1.2974095887925258e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16980 + }, + { + "epoch": 4.943264474832703, + "flow_matching_loss": 0.014487528894096612, + "grad_norm": 0.08178136497735977, + "learning_rate": 1.2957713924981837e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 16990 + }, + { + "epoch": 4.946173988943847, + "flow_matching_loss": 0.01363947121426463, + "grad_norm": 0.0730617418885231, + "learning_rate": 1.2941334443608572e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17000 + }, + { + "epoch": 4.94908350305499, + "flow_matching_loss": 0.014480241388082505, + "grad_norm": 0.08572964370250702, + "learning_rate": 1.2924957463708073e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17010 + }, + { + "epoch": 4.951993017166133, + "flow_matching_loss": 0.014388509560376406, + "grad_norm": 0.06250667572021484, + "learning_rate": 1.2908583005179924e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17020 + }, + { + "epoch": 4.954902531277277, + "flow_matching_loss": 0.016405060607939958, + "grad_norm": 0.07635243982076645, + "learning_rate": 1.2892211087920624e-05, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17030 + }, + { + "epoch": 4.95781204538842, + "flow_matching_loss": 0.012476063612848521, + "grad_norm": 0.06885851919651031, + "learning_rate": 1.287584173182361e-05, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17040 + }, + { + "epoch": 4.9607215594995635, + "flow_matching_loss": 0.014559865090996027, + "grad_norm": 0.0744883194565773, + "learning_rate": 1.2859474956779188e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17050 + }, + { + "epoch": 4.963631073610707, + "flow_matching_loss": 0.014424815122038127, + "grad_norm": 0.09917238354682922, + "learning_rate": 1.2843110782674532e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17060 + }, + { + "epoch": 4.96654058772185, + "flow_matching_loss": 0.014181149285286665, + "grad_norm": 0.11270022392272949, + "learning_rate": 1.2826749229393655e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17070 + }, + { + "epoch": 4.969450101832994, + "flow_matching_loss": 0.015917066019028425, + "grad_norm": 0.07644961029291153, + "learning_rate": 1.2810390316817381e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17080 + }, + { + "epoch": 4.9723596159441374, + "flow_matching_loss": 0.014113528933376075, + "grad_norm": 0.10615627467632294, + "learning_rate": 1.2794034064823346e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17090 + }, + { + "epoch": 4.975269130055281, + "flow_matching_loss": 0.014580399729311466, + "grad_norm": 0.07897920906543732, + "learning_rate": 1.277768049328593e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17100 + }, + { + "epoch": 4.978178644166424, + "flow_matching_loss": 0.01539627481251955, + "grad_norm": 0.08093976229429245, + "learning_rate": 1.2761329622076269e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17110 + }, + { + "epoch": 4.981088158277568, + "flow_matching_loss": 0.01402741316705942, + "grad_norm": 0.0715462937951088, + "learning_rate": 1.274498147106221e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17120 + }, + { + "epoch": 4.983997672388711, + "flow_matching_loss": 0.013447346817702054, + "grad_norm": 0.059073008596897125, + "learning_rate": 1.272863606010829e-05, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17130 + }, + { + "epoch": 4.986907186499854, + "flow_matching_loss": 0.015362286940217018, + "grad_norm": 0.06813973188400269, + "learning_rate": 1.2712293409075739e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17140 + }, + { + "epoch": 4.989816700610998, + "flow_matching_loss": 0.015282431896775962, + "grad_norm": 0.08947782963514328, + "learning_rate": 1.2695953537822413e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17150 + }, + { + "epoch": 4.992726214722142, + "flow_matching_loss": 0.015598297491669655, + "grad_norm": 0.11203235387802124, + "learning_rate": 1.2679616466202792e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17160 + }, + { + "epoch": 4.995635728833285, + "flow_matching_loss": 0.017519359569996597, + "grad_norm": 0.08747486025094986, + "learning_rate": 1.2663282214067967e-05, + "loss": 0.0175, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17170 + }, + { + "epoch": 4.998545242944428, + "flow_matching_loss": 0.01624805713072419, + "grad_norm": 0.07835464179515839, + "learning_rate": 1.264695080126559e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17180 + }, + { + "epoch": 5.001454757055572, + "flow_matching_loss": 0.013423391897231341, + "grad_norm": 0.09713799506425858, + "learning_rate": 1.263062224763986e-05, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17190 + }, + { + "epoch": 5.004364271166716, + "flow_matching_loss": 0.014986522309482098, + "grad_norm": 0.09304139763116837, + "learning_rate": 1.2614296573031522e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17200 + }, + { + "epoch": 5.007273785277858, + "flow_matching_loss": 0.015336994640529155, + "grad_norm": 0.07684627920389175, + "learning_rate": 1.259797379727781e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17210 + }, + { + "epoch": 5.010183299389002, + "flow_matching_loss": 0.013687242195010186, + "grad_norm": 0.09065864980220795, + "learning_rate": 1.2581653940212423e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17220 + }, + { + "epoch": 5.013092813500146, + "flow_matching_loss": 0.014696628041565418, + "grad_norm": 0.07852908968925476, + "learning_rate": 1.2565337021665537e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17230 + }, + { + "epoch": 5.016002327611289, + "flow_matching_loss": 0.015046160947531462, + "grad_norm": 0.08795205503702164, + "learning_rate": 1.2549023061463739e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17240 + }, + { + "epoch": 5.018911841722432, + "flow_matching_loss": 0.014818538911640644, + "grad_norm": 0.07740470767021179, + "learning_rate": 1.2532712079430038e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17250 + }, + { + "epoch": 5.021821355833576, + "flow_matching_loss": 0.016335815377533435, + "grad_norm": 0.08094668388366699, + "learning_rate": 1.2516404095383813e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17260 + }, + { + "epoch": 5.024730869944719, + "flow_matching_loss": 0.015780210215598344, + "grad_norm": 0.10990293323993683, + "learning_rate": 1.2500099129140796e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17270 + }, + { + "epoch": 5.0276403840558626, + "flow_matching_loss": 0.014551437646150588, + "grad_norm": 0.08970575779676437, + "learning_rate": 1.2483797200513065e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17280 + }, + { + "epoch": 5.030549898167006, + "flow_matching_loss": 0.015204572211951017, + "grad_norm": 0.10300686210393906, + "learning_rate": 1.2467498329308992e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17290 + }, + { + "epoch": 5.03345941227815, + "flow_matching_loss": 0.01518882978707552, + "grad_norm": 0.05672498047351837, + "learning_rate": 1.2451202535333253e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17300 + }, + { + "epoch": 5.036368926389293, + "flow_matching_loss": 0.015849341917783023, + "grad_norm": 0.07853874564170837, + "learning_rate": 1.2434909838386768e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17310 + }, + { + "epoch": 5.0392784405004365, + "flow_matching_loss": 0.013245379831641913, + "grad_norm": 0.08079039305448532, + "learning_rate": 1.2418620258266703e-05, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17320 + }, + { + "epoch": 5.04218795461158, + "flow_matching_loss": 0.015579369477927684, + "grad_norm": 0.07462076097726822, + "learning_rate": 1.2402333814766432e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17330 + }, + { + "epoch": 5.045097468722723, + "flow_matching_loss": 0.013642902020365, + "grad_norm": 0.08509200066328049, + "learning_rate": 1.2386050527675515e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17340 + }, + { + "epoch": 5.048006982833867, + "flow_matching_loss": 0.013243858236819506, + "grad_norm": 0.0675523579120636, + "learning_rate": 1.2369770416779693e-05, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17350 + }, + { + "epoch": 5.05091649694501, + "flow_matching_loss": 0.01398252546787262, + "grad_norm": 0.08487533032894135, + "learning_rate": 1.2353493501860828e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17360 + }, + { + "epoch": 5.053826011056153, + "flow_matching_loss": 0.015536256786435842, + "grad_norm": 0.10401463508605957, + "learning_rate": 1.2337219802696905e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17370 + }, + { + "epoch": 5.056735525167297, + "flow_matching_loss": 0.01538141155615449, + "grad_norm": 0.06140521541237831, + "learning_rate": 1.2320949339062008e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17380 + }, + { + "epoch": 5.059645039278441, + "flow_matching_loss": 0.015478410851210355, + "grad_norm": 0.1044408306479454, + "learning_rate": 1.2304682130726281e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17390 + }, + { + "epoch": 5.0625545533895835, + "flow_matching_loss": 0.015070775337517262, + "grad_norm": 0.07109670341014862, + "learning_rate": 1.2288418197455913e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17400 + }, + { + "epoch": 5.065464067500727, + "flow_matching_loss": 0.015024624299257994, + "grad_norm": 0.08368762582540512, + "learning_rate": 1.2272157559013126e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17410 + }, + { + "epoch": 5.068373581611871, + "flow_matching_loss": 0.014272811729460955, + "grad_norm": 0.08244291692972183, + "learning_rate": 1.2255900235156123e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17420 + }, + { + "epoch": 5.071283095723015, + "flow_matching_loss": 0.01403438588604331, + "grad_norm": 0.09499821811914444, + "learning_rate": 1.2239646245639087e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17430 + }, + { + "epoch": 5.074192609834157, + "flow_matching_loss": 0.014195795729756355, + "grad_norm": 0.0903811827301979, + "learning_rate": 1.2223395610212142e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17440 + }, + { + "epoch": 5.077102123945301, + "flow_matching_loss": 0.014830919168889522, + "grad_norm": 0.06399180740118027, + "learning_rate": 1.2207148348621348e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17450 + }, + { + "epoch": 5.080011638056445, + "flow_matching_loss": 0.014129289239645005, + "grad_norm": 0.0695425346493721, + "learning_rate": 1.219090448060866e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17460 + }, + { + "epoch": 5.082921152167588, + "flow_matching_loss": 0.015637845639139412, + "grad_norm": 0.07558012008666992, + "learning_rate": 1.217466402591191e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17470 + }, + { + "epoch": 5.085830666278731, + "flow_matching_loss": 0.014203175529837608, + "grad_norm": 0.06528320908546448, + "learning_rate": 1.2158427004264785e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17480 + }, + { + "epoch": 5.088740180389875, + "flow_matching_loss": 0.014827497955411672, + "grad_norm": 0.06931105256080627, + "learning_rate": 1.2142193435396788e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17490 + }, + { + "epoch": 5.091649694501018, + "flow_matching_loss": 0.014183285366743802, + "grad_norm": 0.06540476530790329, + "learning_rate": 1.2125963339033234e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17500 + }, + { + "epoch": 5.094559208612162, + "flow_matching_loss": 0.013953167479485273, + "grad_norm": 0.0707620307803154, + "learning_rate": 1.2109736734895237e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17510 + }, + { + "epoch": 5.097468722723305, + "flow_matching_loss": 0.014830300025641919, + "grad_norm": 0.07965666055679321, + "learning_rate": 1.2093513642699634e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17520 + }, + { + "epoch": 5.100378236834449, + "flow_matching_loss": 0.014923431538045406, + "grad_norm": 0.07956211268901825, + "learning_rate": 1.2077294082159024e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17530 + }, + { + "epoch": 5.103287750945592, + "flow_matching_loss": 0.015013416670262814, + "grad_norm": 0.08995705097913742, + "learning_rate": 1.2061078072981697e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17540 + }, + { + "epoch": 5.1061972650567355, + "flow_matching_loss": 0.014196161366999149, + "grad_norm": 0.09634124487638474, + "learning_rate": 1.2044865634871626e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17550 + }, + { + "epoch": 5.109106779167879, + "flow_matching_loss": 0.01458036880940199, + "grad_norm": 0.0687124952673912, + "learning_rate": 1.2028656787528468e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17560 + }, + { + "epoch": 5.112016293279022, + "flow_matching_loss": 0.01309160040691495, + "grad_norm": 0.07802250236272812, + "learning_rate": 1.2012451550647491e-05, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17570 + }, + { + "epoch": 5.114925807390166, + "flow_matching_loss": 0.01330098183825612, + "grad_norm": 0.07277046144008636, + "learning_rate": 1.1996249943919589e-05, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17580 + }, + { + "epoch": 5.1178353215013095, + "flow_matching_loss": 0.012888380791991949, + "grad_norm": 0.06612291187047958, + "learning_rate": 1.1980051987031237e-05, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17590 + }, + { + "epoch": 5.120744835612452, + "flow_matching_loss": 0.013074870221316814, + "grad_norm": 0.09085731208324432, + "learning_rate": 1.1963857699664483e-05, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17600 + }, + { + "epoch": 5.123654349723596, + "flow_matching_loss": 0.015562891867011786, + "grad_norm": 0.07481423020362854, + "learning_rate": 1.1947667101496911e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17610 + }, + { + "epoch": 5.12656386383474, + "flow_matching_loss": 0.014719567447900771, + "grad_norm": 0.07798118144273758, + "learning_rate": 1.1931480212201633e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17620 + }, + { + "epoch": 5.129473377945883, + "flow_matching_loss": 0.013094160053879023, + "grad_norm": 0.07637961208820343, + "learning_rate": 1.1915297051447237e-05, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17630 + }, + { + "epoch": 5.132382892057026, + "flow_matching_loss": 0.013580766133964061, + "grad_norm": 0.08318822830915451, + "learning_rate": 1.1899117638897793e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17640 + }, + { + "epoch": 5.13529240616817, + "flow_matching_loss": 0.014785446133464575, + "grad_norm": 0.07190155237913132, + "learning_rate": 1.1882941994212811e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17650 + }, + { + "epoch": 5.138201920279314, + "flow_matching_loss": 0.014735218696296215, + "grad_norm": 0.07607743889093399, + "learning_rate": 1.1866770137047221e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17660 + }, + { + "epoch": 5.1411114343904565, + "flow_matching_loss": 0.015198312420397996, + "grad_norm": 0.07823514938354492, + "learning_rate": 1.1850602087051363e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17670 + }, + { + "epoch": 5.1440209485016, + "flow_matching_loss": 0.01415485618636012, + "grad_norm": 0.07118518650531769, + "learning_rate": 1.183443786387094e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17680 + }, + { + "epoch": 5.146930462612744, + "flow_matching_loss": 0.015106230601668358, + "grad_norm": 0.08038599044084549, + "learning_rate": 1.1818277487147005e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17690 + }, + { + "epoch": 5.149839976723887, + "flow_matching_loss": 0.014461912401020526, + "grad_norm": 0.0674690455198288, + "learning_rate": 1.180212097651594e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17700 + }, + { + "epoch": 5.15274949083503, + "flow_matching_loss": 0.01579096373170614, + "grad_norm": 0.07561401277780533, + "learning_rate": 1.1785968351609422e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17710 + }, + { + "epoch": 5.155659004946174, + "flow_matching_loss": 0.014097069948911667, + "grad_norm": 0.07020734250545502, + "learning_rate": 1.1769819632054427e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17720 + }, + { + "epoch": 5.158568519057318, + "flow_matching_loss": 0.014736156444996595, + "grad_norm": 0.08190456032752991, + "learning_rate": 1.1753674837473165e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17730 + }, + { + "epoch": 5.161478033168461, + "flow_matching_loss": 0.013263844884932041, + "grad_norm": 0.05903729051351547, + "learning_rate": 1.1737533987483081e-05, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17740 + }, + { + "epoch": 5.164387547279604, + "flow_matching_loss": 0.014178628008812666, + "grad_norm": 0.0653097853064537, + "learning_rate": 1.1721397101696835e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17750 + }, + { + "epoch": 5.167297061390748, + "flow_matching_loss": 0.014202595315873623, + "grad_norm": 0.07456643879413605, + "learning_rate": 1.1705264199722257e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17760 + }, + { + "epoch": 5.170206575501891, + "flow_matching_loss": 0.015833819936960935, + "grad_norm": 0.06333242356777191, + "learning_rate": 1.1689135301162356e-05, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17770 + }, + { + "epoch": 5.173116089613035, + "flow_matching_loss": 0.013640451058745385, + "grad_norm": 0.07899776846170425, + "learning_rate": 1.1673010425615258e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17780 + }, + { + "epoch": 5.176025603724178, + "flow_matching_loss": 0.015583462081849575, + "grad_norm": 0.08531445264816284, + "learning_rate": 1.1656889592674208e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17790 + }, + { + "epoch": 5.178935117835321, + "flow_matching_loss": 0.015870503056794404, + "grad_norm": 0.11061547696590424, + "learning_rate": 1.1640772821927533e-05, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17800 + }, + { + "epoch": 5.181844631946465, + "flow_matching_loss": 0.01443715300410986, + "grad_norm": 0.0923316702246666, + "learning_rate": 1.162466013295864e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17810 + }, + { + "epoch": 5.1847541460576085, + "flow_matching_loss": 0.013770577032119036, + "grad_norm": 0.08880434930324554, + "learning_rate": 1.160855154534595e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17820 + }, + { + "epoch": 5.187663660168752, + "flow_matching_loss": 0.013682099152356386, + "grad_norm": 0.06813716888427734, + "learning_rate": 1.1592447078662931e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17830 + }, + { + "epoch": 5.190573174279895, + "flow_matching_loss": 0.0161172759719193, + "grad_norm": 0.0833578109741211, + "learning_rate": 1.1576346752478025e-05, + "loss": 0.0161, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17840 + }, + { + "epoch": 5.193482688391039, + "flow_matching_loss": 0.014250215981155634, + "grad_norm": 0.07949372380971909, + "learning_rate": 1.1560250586354643e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17850 + }, + { + "epoch": 5.1963922025021825, + "flow_matching_loss": 0.013446813449263573, + "grad_norm": 0.0812385231256485, + "learning_rate": 1.1544158599851146e-05, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17860 + }, + { + "epoch": 5.199301716613325, + "flow_matching_loss": 0.014290925487875939, + "grad_norm": 0.08268042653799057, + "learning_rate": 1.1528070812520809e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17870 + }, + { + "epoch": 5.202211230724469, + "flow_matching_loss": 0.013223837036639452, + "grad_norm": 0.08393258601427078, + "learning_rate": 1.1511987243911821e-05, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17880 + }, + { + "epoch": 5.205120744835613, + "flow_matching_loss": 0.013661114405840636, + "grad_norm": 0.06896724551916122, + "learning_rate": 1.1495907913567228e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17890 + }, + { + "epoch": 5.2080302589467555, + "flow_matching_loss": 0.012799243908375502, + "grad_norm": 0.07236266881227493, + "learning_rate": 1.1479832841024936e-05, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17900 + }, + { + "epoch": 5.210939773057899, + "flow_matching_loss": 0.015009920950978995, + "grad_norm": 0.0908389762043953, + "learning_rate": 1.1463762045817675e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17910 + }, + { + "epoch": 5.213849287169043, + "flow_matching_loss": 0.013882779143750667, + "grad_norm": 0.07920007407665253, + "learning_rate": 1.1447695547472964e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17920 + }, + { + "epoch": 5.216758801280186, + "flow_matching_loss": 0.01436904538422823, + "grad_norm": 0.06449761241674423, + "learning_rate": 1.1431633365513132e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17930 + }, + { + "epoch": 5.2196683153913295, + "flow_matching_loss": 0.014232579991221428, + "grad_norm": 0.07393350452184677, + "learning_rate": 1.1415575519455235e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17940 + }, + { + "epoch": 5.222577829502473, + "flow_matching_loss": 0.01342633031308651, + "grad_norm": 0.07196889817714691, + "learning_rate": 1.1399522028811069e-05, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17950 + }, + { + "epoch": 5.225487343613617, + "flow_matching_loss": 0.014414418954402208, + "grad_norm": 0.07985491305589676, + "learning_rate": 1.1383472913087143e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17960 + }, + { + "epoch": 5.22839685772476, + "flow_matching_loss": 0.012938365340232849, + "grad_norm": 0.0706188976764679, + "learning_rate": 1.1367428191784643e-05, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17970 + }, + { + "epoch": 5.231306371835903, + "flow_matching_loss": 0.01458335416391492, + "grad_norm": 0.07201600819826126, + "learning_rate": 1.1351387884399413e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17980 + }, + { + "epoch": 5.234215885947047, + "flow_matching_loss": 0.015005349554121495, + "grad_norm": 0.08267015218734741, + "learning_rate": 1.1335352010421953e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 17990 + }, + { + "epoch": 5.23712540005819, + "flow_matching_loss": 0.01560506345704198, + "grad_norm": 0.1175367534160614, + "learning_rate": 1.1319320589337352e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18000 + }, + { + "epoch": 5.240034914169334, + "flow_matching_loss": 0.01438123183324933, + "grad_norm": 0.078096903860569, + "learning_rate": 1.1303293640625299e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18010 + }, + { + "epoch": 5.242944428280477, + "flow_matching_loss": 0.014248014893382788, + "grad_norm": 0.08650845289230347, + "learning_rate": 1.1287271183760052e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18020 + }, + { + "epoch": 5.24585394239162, + "flow_matching_loss": 0.014431782159954309, + "grad_norm": 0.0813077986240387, + "learning_rate": 1.1271253238210398e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18030 + }, + { + "epoch": 5.248763456502764, + "flow_matching_loss": 0.014173167198896408, + "grad_norm": 0.09728267043828964, + "learning_rate": 1.1255239823439663e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18040 + }, + { + "epoch": 5.251672970613908, + "flow_matching_loss": 0.014359564613550901, + "grad_norm": 0.08393754810094833, + "learning_rate": 1.1239230958905652e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18050 + }, + { + "epoch": 5.254582484725051, + "flow_matching_loss": 0.01472691148519516, + "grad_norm": 0.08069011569023132, + "learning_rate": 1.1223226664060641e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18060 + }, + { + "epoch": 5.257491998836194, + "flow_matching_loss": 0.012260529957711697, + "grad_norm": 0.069974884390831, + "learning_rate": 1.1207226958351362e-05, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18070 + }, + { + "epoch": 5.260401512947338, + "flow_matching_loss": 0.01301980372518301, + "grad_norm": 0.053866952657699585, + "learning_rate": 1.1191231861218955e-05, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18080 + }, + { + "epoch": 5.2633110270584815, + "flow_matching_loss": 0.013286399189382792, + "grad_norm": 0.08059152960777283, + "learning_rate": 1.1175241392098988e-05, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18090 + }, + { + "epoch": 5.266220541169624, + "flow_matching_loss": 0.013815781380981208, + "grad_norm": 0.083107590675354, + "learning_rate": 1.1159255570421376e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18100 + }, + { + "epoch": 5.269130055280768, + "flow_matching_loss": 0.014473802130669355, + "grad_norm": 0.07718322426080704, + "learning_rate": 1.1143274415610408e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18110 + }, + { + "epoch": 5.272039569391912, + "flow_matching_loss": 0.015125489421188831, + "grad_norm": 0.0888260155916214, + "learning_rate": 1.1127297947084692e-05, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18120 + }, + { + "epoch": 5.274949083503055, + "flow_matching_loss": 0.014721883181482553, + "grad_norm": 0.08783269673585892, + "learning_rate": 1.1111326184257135e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18130 + }, + { + "epoch": 5.277858597614198, + "flow_matching_loss": 0.015443879924714566, + "grad_norm": 0.09099747240543365, + "learning_rate": 1.109535914653495e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18140 + }, + { + "epoch": 5.280768111725342, + "flow_matching_loss": 0.014405611716210843, + "grad_norm": 0.07240559905767441, + "learning_rate": 1.1079396853319582e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18150 + }, + { + "epoch": 5.283677625836486, + "flow_matching_loss": 0.014777471031993627, + "grad_norm": 0.09118334949016571, + "learning_rate": 1.1063439324006724e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18160 + }, + { + "epoch": 5.2865871399476285, + "flow_matching_loss": 0.014511637389659882, + "grad_norm": 0.07447124272584915, + "learning_rate": 1.1047486577986285e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18170 + }, + { + "epoch": 5.289496654058772, + "flow_matching_loss": 0.01308499276638031, + "grad_norm": 0.07157981395721436, + "learning_rate": 1.1031538634642353e-05, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18180 + }, + { + "epoch": 5.292406168169916, + "flow_matching_loss": 0.014464143570512534, + "grad_norm": 0.07140300422906876, + "learning_rate": 1.1015595513353172e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18190 + }, + { + "epoch": 5.295315682281059, + "flow_matching_loss": 0.015671083703637123, + "grad_norm": 0.08430729806423187, + "learning_rate": 1.0999657233491156e-05, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18200 + }, + { + "epoch": 5.2982251963922025, + "flow_matching_loss": 0.015552492160350085, + "grad_norm": 0.10384771227836609, + "learning_rate": 1.0983723814422808e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18210 + }, + { + "epoch": 5.301134710503346, + "flow_matching_loss": 0.014116156846284866, + "grad_norm": 0.08197542279958725, + "learning_rate": 1.0967795275508735e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18220 + }, + { + "epoch": 5.304044224614489, + "flow_matching_loss": 0.015408993791788816, + "grad_norm": 0.11041007190942764, + "learning_rate": 1.095187163610361e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18230 + }, + { + "epoch": 5.306953738725633, + "flow_matching_loss": 0.013563696667551995, + "grad_norm": 0.06684599071741104, + "learning_rate": 1.0935952915556163e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18240 + }, + { + "epoch": 5.309863252836776, + "flow_matching_loss": 0.015282709803432225, + "grad_norm": 0.07639232277870178, + "learning_rate": 1.0920039133209137e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18250 + }, + { + "epoch": 5.31277276694792, + "flow_matching_loss": 0.014508427772670984, + "grad_norm": 0.07880190759897232, + "learning_rate": 1.0904130308399283e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18260 + }, + { + "epoch": 5.315682281059063, + "flow_matching_loss": 0.015356415603309869, + "grad_norm": 0.07015882432460785, + "learning_rate": 1.0888226460457318e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18270 + }, + { + "epoch": 5.318591795170207, + "flow_matching_loss": 0.014299337938427924, + "grad_norm": 0.0790313333272934, + "learning_rate": 1.087232760870792e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18280 + }, + { + "epoch": 5.32150130928135, + "flow_matching_loss": 0.013922499679028987, + "grad_norm": 0.09185872972011566, + "learning_rate": 1.0856433772469686e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18290 + }, + { + "epoch": 5.324410823392493, + "flow_matching_loss": 0.014552885480225087, + "grad_norm": 0.0791638195514679, + "learning_rate": 1.0840544971055135e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18300 + }, + { + "epoch": 5.327320337503637, + "flow_matching_loss": 0.012597315199673175, + "grad_norm": 0.07417434453964233, + "learning_rate": 1.0824661223770655e-05, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18310 + }, + { + "epoch": 5.330229851614781, + "flow_matching_loss": 0.017045872565358878, + "grad_norm": 0.1353216916322708, + "learning_rate": 1.0808782549916498e-05, + "loss": 0.017, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18320 + }, + { + "epoch": 5.333139365725923, + "flow_matching_loss": 0.014580531045794487, + "grad_norm": 0.07357041537761688, + "learning_rate": 1.079290896878675e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18330 + }, + { + "epoch": 5.336048879837067, + "flow_matching_loss": 0.015180761460214853, + "grad_norm": 0.08923309296369553, + "learning_rate": 1.07770404996693e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18340 + }, + { + "epoch": 5.338958393948211, + "flow_matching_loss": 0.012821434997022152, + "grad_norm": 0.11213812977075577, + "learning_rate": 1.0761177161845848e-05, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18350 + }, + { + "epoch": 5.3418679080593545, + "flow_matching_loss": 0.015427530277520418, + "grad_norm": 0.06436263024806976, + "learning_rate": 1.0745318974591841e-05, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18360 + }, + { + "epoch": 5.344777422170497, + "flow_matching_loss": 0.01485648239031434, + "grad_norm": 0.08968647569417953, + "learning_rate": 1.0729465957176469e-05, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18370 + }, + { + "epoch": 5.347686936281641, + "flow_matching_loss": 0.013665487710386514, + "grad_norm": 0.07570510357618332, + "learning_rate": 1.0713618128862642e-05, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18380 + }, + { + "epoch": 5.350596450392785, + "flow_matching_loss": 0.014228413999080657, + "grad_norm": 0.061166271567344666, + "learning_rate": 1.0697775508906969e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18390 + }, + { + "epoch": 5.353505964503928, + "flow_matching_loss": 0.014105386845767499, + "grad_norm": 0.062139857560396194, + "learning_rate": 1.068193811655972e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18400 + }, + { + "epoch": 5.356415478615071, + "flow_matching_loss": 0.015023345313966274, + "grad_norm": 0.07352445274591446, + "learning_rate": 1.0666105971064835e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18410 + }, + { + "epoch": 5.359324992726215, + "flow_matching_loss": 0.014001339487731457, + "grad_norm": 0.1115109845995903, + "learning_rate": 1.0650279091659852e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18420 + }, + { + "epoch": 5.362234506837358, + "flow_matching_loss": 0.013228275813162327, + "grad_norm": 0.07127568870782852, + "learning_rate": 1.0634457497575926e-05, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18430 + }, + { + "epoch": 5.3651440209485015, + "flow_matching_loss": 0.01521924054250121, + "grad_norm": 0.08253985643386841, + "learning_rate": 1.061864120803778e-05, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18440 + }, + { + "epoch": 5.368053535059645, + "flow_matching_loss": 0.014230374246835709, + "grad_norm": 0.06952086836099625, + "learning_rate": 1.0602830242263701e-05, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18450 + }, + { + "epoch": 5.370963049170788, + "flow_matching_loss": 0.013858684152364732, + "grad_norm": 0.08153971284627914, + "learning_rate": 1.0587024619465506e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18460 + }, + { + "epoch": 5.373872563281932, + "flow_matching_loss": 0.013953071832656861, + "grad_norm": 0.061966873705387115, + "learning_rate": 1.0571224358848517e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18470 + }, + { + "epoch": 5.3767820773930755, + "flow_matching_loss": 0.014721129089593887, + "grad_norm": 0.08270831406116486, + "learning_rate": 1.0555429479611537e-05, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18480 + }, + { + "epoch": 5.379691591504219, + "flow_matching_loss": 0.01413518823683262, + "grad_norm": 0.0737101212143898, + "learning_rate": 1.0539640000946835e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18490 + }, + { + "epoch": 5.382601105615362, + "flow_matching_loss": 0.012754282169044017, + "grad_norm": 0.06758757680654526, + "learning_rate": 1.0523855942040111e-05, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18500 + }, + { + "epoch": 5.385510619726506, + "flow_matching_loss": 0.013559170626103878, + "grad_norm": 0.05786287784576416, + "learning_rate": 1.0508077322070496e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18510 + }, + { + "epoch": 5.388420133837649, + "flow_matching_loss": 0.014106608089059591, + "grad_norm": 0.09558823704719543, + "learning_rate": 1.0492304160210495e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18520 + }, + { + "epoch": 5.391329647948792, + "flow_matching_loss": 0.015606057737022639, + "grad_norm": 0.08480917662382126, + "learning_rate": 1.0476536475625985e-05, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18530 + }, + { + "epoch": 5.394239162059936, + "flow_matching_loss": 0.016465680953115226, + "grad_norm": 0.10109039396047592, + "learning_rate": 1.0460774287476195e-05, + "loss": 0.0165, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18540 + }, + { + "epoch": 5.39714867617108, + "flow_matching_loss": 0.013601236324757338, + "grad_norm": 0.058149006217718124, + "learning_rate": 1.044501761491366e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18550 + }, + { + "epoch": 5.400058190282223, + "flow_matching_loss": 0.013456345256417989, + "grad_norm": 0.07865539193153381, + "learning_rate": 1.0429266477084235e-05, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18560 + }, + { + "epoch": 5.402967704393366, + "flow_matching_loss": 0.015510769747197627, + "grad_norm": 0.0857459306716919, + "learning_rate": 1.0413520893127034e-05, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18570 + }, + { + "epoch": 5.40587721850451, + "flow_matching_loss": 0.01341849947348237, + "grad_norm": 0.0749133825302124, + "learning_rate": 1.0397780882174424e-05, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18580 + }, + { + "epoch": 5.408786732615654, + "flow_matching_loss": 0.014026033412665128, + "grad_norm": 0.07495055347681046, + "learning_rate": 1.0382046463351998e-05, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18590 + }, + { + "epoch": 5.411696246726796, + "flow_matching_loss": 0.015254741348326207, + "grad_norm": 0.0759185254573822, + "learning_rate": 1.0366317655778568e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18600 + }, + { + "epoch": 5.41460576083794, + "flow_matching_loss": 0.013468405324965715, + "grad_norm": 0.08407723158597946, + "learning_rate": 1.035059447856611e-05, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18610 + }, + { + "epoch": 5.417515274949084, + "flow_matching_loss": 0.014144762977957725, + "grad_norm": 0.06991912424564362, + "learning_rate": 1.0334876950819777e-05, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18620 + }, + { + "epoch": 5.420424789060227, + "flow_matching_loss": 0.014495187066495418, + "grad_norm": 0.08243036270141602, + "learning_rate": 1.0319165091637837e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18630 + }, + { + "epoch": 5.42333430317137, + "flow_matching_loss": 0.014482987113296986, + "grad_norm": 0.06415586918592453, + "learning_rate": 1.0303458920111686e-05, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18640 + }, + { + "epoch": 5.426243817282514, + "flow_matching_loss": 0.015259051881730556, + "grad_norm": 0.061347201466560364, + "learning_rate": 1.0287758455325803e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18650 + }, + { + "epoch": 5.429153331393657, + "flow_matching_loss": 0.014362092595547438, + "grad_norm": 0.07047771662473679, + "learning_rate": 1.0272063716357729e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18660 + }, + { + "epoch": 5.432062845504801, + "flow_matching_loss": 0.013360496424138547, + "grad_norm": 0.061755821108818054, + "learning_rate": 1.0256374722278057e-05, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18670 + }, + { + "epoch": 5.434972359615944, + "flow_matching_loss": 0.014621833804994822, + "grad_norm": 0.07258612662553787, + "learning_rate": 1.0240691492150388e-05, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18680 + }, + { + "epoch": 5.437881873727088, + "flow_matching_loss": 0.013166894763708114, + "grad_norm": 0.07106325030326843, + "learning_rate": 1.0225014045031334e-05, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18690 + }, + { + "epoch": 5.440791387838231, + "flow_matching_loss": 0.013528968673199415, + "grad_norm": 0.07468114048242569, + "learning_rate": 1.0209342399970468e-05, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18700 + }, + { + "epoch": 5.4437009019493745, + "flow_matching_loss": 0.013786572311073542, + "grad_norm": 0.06311517208814621, + "learning_rate": 1.019367657601031e-05, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18710 + }, + { + "epoch": 5.446610416060518, + "flow_matching_loss": 0.013923586159944535, + "grad_norm": 0.07382917404174805, + "learning_rate": 1.0178016592186323e-05, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18720 + }, + { + "epoch": 5.449519930171661, + "flow_matching_loss": 0.016335546411573885, + "grad_norm": 0.08474098145961761, + "learning_rate": 1.0162362467526861e-05, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18730 + }, + { + "epoch": 5.452429444282805, + "flow_matching_loss": 0.014414412807673216, + "grad_norm": 0.0573820136487484, + "learning_rate": 1.0146714221053159e-05, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18740 + }, + { + "epoch": 5.4553389583939484, + "flow_matching_loss": 0.014995524194091558, + "grad_norm": 0.09695395827293396, + "learning_rate": 1.0131071871779314e-05, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18750 + }, + { + "epoch": 5.458248472505091, + "flow_matching_loss": 0.01616803715005517, + "grad_norm": 0.06463763862848282, + "learning_rate": 1.0115435438712249e-05, + "loss": 0.0162, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18760 + }, + { + "epoch": 5.461157986616235, + "flow_matching_loss": 0.016049562580883504, + "grad_norm": 0.09228076785802841, + "learning_rate": 1.0099804940851716e-05, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18770 + }, + { + "epoch": 5.464067500727379, + "flow_matching_loss": 0.013631443306803704, + "grad_norm": 0.07269315421581268, + "learning_rate": 1.0084180397190238e-05, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18780 + }, + { + "epoch": 5.4669770148385215, + "flow_matching_loss": 0.015345499943941831, + "grad_norm": 0.06799890100955963, + "learning_rate": 1.006856182671311e-05, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18790 + }, + { + "epoch": 5.469886528949665, + "flow_matching_loss": 0.014806960895657539, + "grad_norm": 0.08976037800312042, + "learning_rate": 1.0052949248398367e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18800 + }, + { + "epoch": 5.472796043060809, + "flow_matching_loss": 0.014820606634020805, + "grad_norm": 0.07689198851585388, + "learning_rate": 1.003734268121676e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18810 + }, + { + "epoch": 5.475705557171953, + "flow_matching_loss": 0.01429534461349249, + "grad_norm": 0.07625622302293777, + "learning_rate": 1.0021742144131744e-05, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18820 + }, + { + "epoch": 5.478615071283095, + "flow_matching_loss": 0.014757600240409374, + "grad_norm": 0.08511488139629364, + "learning_rate": 1.000614765609944e-05, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18830 + }, + { + "epoch": 5.481524585394239, + "flow_matching_loss": 0.01329875085502863, + "grad_norm": 0.08029192686080933, + "learning_rate": 9.990559236068628e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18840 + }, + { + "epoch": 5.484434099505383, + "flow_matching_loss": 0.01367708882316947, + "grad_norm": 0.06573877483606339, + "learning_rate": 9.974976902980703e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18850 + }, + { + "epoch": 5.487343613616526, + "flow_matching_loss": 0.015431222878396511, + "grad_norm": 0.08339960873126984, + "learning_rate": 9.959400675769668e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18860 + }, + { + "epoch": 5.490253127727669, + "flow_matching_loss": 0.015002061612904072, + "grad_norm": 0.08041666448116302, + "learning_rate": 9.943830573362106e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18870 + }, + { + "epoch": 5.493162641838813, + "flow_matching_loss": 0.015867660660296677, + "grad_norm": 0.09842998534440994, + "learning_rate": 9.928266614677168e-06, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18880 + }, + { + "epoch": 5.496072155949957, + "flow_matching_loss": 0.014660486672073602, + "grad_norm": 0.07909104973077774, + "learning_rate": 9.912708818626524e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18890 + }, + { + "epoch": 5.4989816700611, + "flow_matching_loss": 0.013768407236784696, + "grad_norm": 0.09168185293674469, + "learning_rate": 9.897157204114369e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18900 + }, + { + "epoch": 5.501891184172243, + "flow_matching_loss": 0.01359344944357872, + "grad_norm": 0.07503817975521088, + "learning_rate": 9.881611790037375e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18910 + }, + { + "epoch": 5.504800698283387, + "flow_matching_loss": 0.014545656740665436, + "grad_norm": 0.08101405948400497, + "learning_rate": 9.866072595284684e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18920 + }, + { + "epoch": 5.50771021239453, + "flow_matching_loss": 0.014727358892560004, + "grad_norm": 0.06313575804233551, + "learning_rate": 9.850539638737894e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18930 + }, + { + "epoch": 5.5106197265056736, + "flow_matching_loss": 0.015707041788846254, + "grad_norm": 0.08458573371171951, + "learning_rate": 9.835012939271006e-06, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18940 + }, + { + "epoch": 5.513529240616817, + "flow_matching_loss": 0.014138418715447187, + "grad_norm": 0.07830741256475449, + "learning_rate": 9.819492515750418e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18950 + }, + { + "epoch": 5.51643875472796, + "flow_matching_loss": 0.013253368157893419, + "grad_norm": 0.0655209869146347, + "learning_rate": 9.803978387034918e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18960 + }, + { + "epoch": 5.519348268839104, + "flow_matching_loss": 0.017053388245403766, + "grad_norm": 0.07348814606666565, + "learning_rate": 9.788470571975624e-06, + "loss": 0.0171, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18970 + }, + { + "epoch": 5.5222577829502475, + "flow_matching_loss": 0.013441601023077964, + "grad_norm": 0.06665221601724625, + "learning_rate": 9.772969089416008e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18980 + }, + { + "epoch": 5.52516729706139, + "flow_matching_loss": 0.014102543238550425, + "grad_norm": 0.06225533038377762, + "learning_rate": 9.757473958191822e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 18990 + }, + { + "epoch": 5.528076811172534, + "flow_matching_loss": 0.013359599187970162, + "grad_norm": 0.0665641576051712, + "learning_rate": 9.741985197131118e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19000 + }, + { + "epoch": 5.530986325283678, + "flow_matching_loss": 0.012938430532813072, + "grad_norm": 0.07185734808444977, + "learning_rate": 9.726502825054195e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19010 + }, + { + "epoch": 5.533895839394821, + "flow_matching_loss": 0.013238775543868542, + "grad_norm": 0.07074296474456787, + "learning_rate": 9.711026860773595e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19020 + }, + { + "epoch": 5.536805353505964, + "flow_matching_loss": 0.013764259498566389, + "grad_norm": 0.0721181333065033, + "learning_rate": 9.69555732309408e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19030 + }, + { + "epoch": 5.539714867617108, + "flow_matching_loss": 0.015712709445506335, + "grad_norm": 0.09315396845340729, + "learning_rate": 9.68009423081259e-06, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19040 + }, + { + "epoch": 5.542624381728252, + "flow_matching_loss": 0.013718444388359785, + "grad_norm": 0.08433602750301361, + "learning_rate": 9.664637602718249e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19050 + }, + { + "epoch": 5.5455338958393945, + "flow_matching_loss": 0.013684534560889005, + "grad_norm": 0.061317309737205505, + "learning_rate": 9.649187457592313e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19060 + }, + { + "epoch": 5.548443409950538, + "flow_matching_loss": 0.013795143738389016, + "grad_norm": 0.06315181404352188, + "learning_rate": 9.63374381420816e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19070 + }, + { + "epoch": 5.551352924061682, + "flow_matching_loss": 0.014802067168056965, + "grad_norm": 0.07103651762008667, + "learning_rate": 9.618306691331279e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19080 + }, + { + "epoch": 5.554262438172826, + "flow_matching_loss": 0.015337784867733717, + "grad_norm": 0.0749947801232338, + "learning_rate": 9.602876107719228e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19090 + }, + { + "epoch": 5.557171952283968, + "flow_matching_loss": 0.014090190548449755, + "grad_norm": 0.06682690232992172, + "learning_rate": 9.587452082121621e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19100 + }, + { + "epoch": 5.560081466395112, + "flow_matching_loss": 0.014489790517836809, + "grad_norm": 0.0666239783167839, + "learning_rate": 9.57203463328011e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19110 + }, + { + "epoch": 5.562990980506255, + "flow_matching_loss": 0.01331667909398675, + "grad_norm": 0.0660175308585167, + "learning_rate": 9.556623779928341e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19120 + }, + { + "epoch": 5.565900494617399, + "flow_matching_loss": 0.013228223938494921, + "grad_norm": 0.0823737233877182, + "learning_rate": 9.541219540791949e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19130 + }, + { + "epoch": 5.568810008728542, + "flow_matching_loss": 0.01481823530048132, + "grad_norm": 0.05605817213654518, + "learning_rate": 9.525821934588551e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19140 + }, + { + "epoch": 5.571719522839686, + "flow_matching_loss": 0.014323061052709817, + "grad_norm": 0.07652457803487778, + "learning_rate": 9.510430980027684e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19150 + }, + { + "epoch": 5.574629036950829, + "flow_matching_loss": 0.013784695975482463, + "grad_norm": 0.057495638728141785, + "learning_rate": 9.495046695810809e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19160 + }, + { + "epoch": 5.577538551061973, + "flow_matching_loss": 0.014087195042520761, + "grad_norm": 0.09137474000453949, + "learning_rate": 9.47966910063128e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19170 + }, + { + "epoch": 5.580448065173116, + "flow_matching_loss": 0.01438750522211194, + "grad_norm": 0.07996245473623276, + "learning_rate": 9.464298213174325e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19180 + }, + { + "epoch": 5.583357579284259, + "flow_matching_loss": 0.01283996794372797, + "grad_norm": 0.06817197054624557, + "learning_rate": 9.448934052117024e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19190 + }, + { + "epoch": 5.586267093395403, + "flow_matching_loss": 0.014442319422960282, + "grad_norm": 0.07869201898574829, + "learning_rate": 9.433576636128284e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19200 + }, + { + "epoch": 5.5891766075065465, + "flow_matching_loss": 0.013046272285282611, + "grad_norm": 0.0733441561460495, + "learning_rate": 9.41822598386881e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19210 + }, + { + "epoch": 5.59208612161769, + "flow_matching_loss": 0.014755075238645077, + "grad_norm": 0.062093351036310196, + "learning_rate": 9.402882113991094e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19220 + }, + { + "epoch": 5.594995635728833, + "flow_matching_loss": 0.01564233349636197, + "grad_norm": 0.08595731109380722, + "learning_rate": 9.387545045139381e-06, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19230 + }, + { + "epoch": 5.597905149839977, + "flow_matching_loss": 0.013453064300119878, + "grad_norm": 0.0682637169957161, + "learning_rate": 9.372214795949655e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19240 + }, + { + "epoch": 5.6008146639511205, + "flow_matching_loss": 0.015059493947774172, + "grad_norm": 0.08128850907087326, + "learning_rate": 9.356891385049618e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19250 + }, + { + "epoch": 5.603724178062263, + "flow_matching_loss": 0.014334553480148315, + "grad_norm": 0.07265900075435638, + "learning_rate": 9.341574831058664e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19260 + }, + { + "epoch": 5.606633692173407, + "flow_matching_loss": 0.014270665310323238, + "grad_norm": 0.06629081070423126, + "learning_rate": 9.326265152587844e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19270 + }, + { + "epoch": 5.609543206284551, + "flow_matching_loss": 0.014284381829202174, + "grad_norm": 0.07268621772527695, + "learning_rate": 9.310962368239861e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19280 + }, + { + "epoch": 5.6124527203956935, + "flow_matching_loss": 0.014671500027179717, + "grad_norm": 0.07241180539131165, + "learning_rate": 9.295666496609036e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19290 + }, + { + "epoch": 5.615362234506837, + "flow_matching_loss": 0.014042776264250278, + "grad_norm": 0.07982973754405975, + "learning_rate": 9.280377556281306e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19300 + }, + { + "epoch": 5.618271748617981, + "flow_matching_loss": 0.01674653487280011, + "grad_norm": 0.07041902095079422, + "learning_rate": 9.265095565834172e-06, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19310 + }, + { + "epoch": 5.621181262729124, + "flow_matching_loss": 0.012617670092731714, + "grad_norm": 0.09313657134771347, + "learning_rate": 9.249820543836688e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19320 + }, + { + "epoch": 5.6240907768402675, + "flow_matching_loss": 0.014751604106277228, + "grad_norm": 0.08451233804225922, + "learning_rate": 9.234552508849454e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19330 + }, + { + "epoch": 5.627000290951411, + "flow_matching_loss": 0.013396016415208578, + "grad_norm": 0.07536576688289642, + "learning_rate": 9.21929147942456e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19340 + }, + { + "epoch": 5.629909805062555, + "flow_matching_loss": 0.012925148848444224, + "grad_norm": 0.07973012328147888, + "learning_rate": 9.204037474105611e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19350 + }, + { + "epoch": 5.632819319173698, + "flow_matching_loss": 0.014315337780863047, + "grad_norm": 0.07595457136631012, + "learning_rate": 9.188790511427653e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19360 + }, + { + "epoch": 5.635728833284841, + "flow_matching_loss": 0.013984286971390248, + "grad_norm": 0.07091580331325531, + "learning_rate": 9.173550609917186e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19370 + }, + { + "epoch": 5.638638347395985, + "flow_matching_loss": 0.015132176969200373, + "grad_norm": 0.104290671646595, + "learning_rate": 9.158317788092124e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19380 + }, + { + "epoch": 5.641547861507128, + "flow_matching_loss": 0.014093407336622477, + "grad_norm": 0.06379752606153488, + "learning_rate": 9.143092064461783e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19390 + }, + { + "epoch": 5.644457375618272, + "flow_matching_loss": 0.012740873033180833, + "grad_norm": 0.04558994621038437, + "learning_rate": 9.127873457526855e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19400 + }, + { + "epoch": 5.647366889729415, + "flow_matching_loss": 0.014154290035367011, + "grad_norm": 0.06654490530490875, + "learning_rate": 9.112661985779382e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19410 + }, + { + "epoch": 5.650276403840559, + "flow_matching_loss": 0.015171472821384669, + "grad_norm": 0.07357189804315567, + "learning_rate": 9.09745766770274e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19420 + }, + { + "epoch": 5.653185917951702, + "flow_matching_loss": 0.014216260518878698, + "grad_norm": 0.08858582377433777, + "learning_rate": 9.082260521771601e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19430 + }, + { + "epoch": 5.656095432062846, + "flow_matching_loss": 0.014263520203530788, + "grad_norm": 0.11146829277276993, + "learning_rate": 9.067070566451936e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19440 + }, + { + "epoch": 5.659004946173989, + "flow_matching_loss": 0.015022577531635762, + "grad_norm": 0.09028405696153641, + "learning_rate": 9.051887820200966e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19450 + }, + { + "epoch": 5.661914460285132, + "flow_matching_loss": 0.013373479805886745, + "grad_norm": 0.06368529051542282, + "learning_rate": 9.03671230146717e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19460 + }, + { + "epoch": 5.664823974396276, + "flow_matching_loss": 0.014199172891676427, + "grad_norm": 0.07499255985021591, + "learning_rate": 9.021544028690228e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19470 + }, + { + "epoch": 5.6677334885074195, + "flow_matching_loss": 0.013728784397244453, + "grad_norm": 0.08070968091487885, + "learning_rate": 9.006383020301024e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19480 + }, + { + "epoch": 5.670643002618562, + "flow_matching_loss": 0.013197594322264194, + "grad_norm": 0.08103517442941666, + "learning_rate": 8.991229294721613e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19490 + }, + { + "epoch": 5.673552516729706, + "flow_matching_loss": 0.015304145216941834, + "grad_norm": 0.07755275815725327, + "learning_rate": 8.976082870365191e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19500 + }, + { + "epoch": 5.67646203084085, + "flow_matching_loss": 0.013202411867678166, + "grad_norm": 0.0734463557600975, + "learning_rate": 8.960943765636105e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19510 + }, + { + "epoch": 5.679371544951993, + "flow_matching_loss": 0.013579751644283534, + "grad_norm": 0.0852995365858078, + "learning_rate": 8.945811998929792e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19520 + }, + { + "epoch": 5.682281059063136, + "flow_matching_loss": 0.013032276183366776, + "grad_norm": 0.05873430520296097, + "learning_rate": 8.93068758863277e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19530 + }, + { + "epoch": 5.68519057317428, + "flow_matching_loss": 0.013371981959789992, + "grad_norm": 0.07293348759412766, + "learning_rate": 8.915570553122628e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19540 + }, + { + "epoch": 5.688100087285424, + "flow_matching_loss": 0.01357826692983508, + "grad_norm": 0.07578589022159576, + "learning_rate": 8.900460910767982e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19550 + }, + { + "epoch": 5.6910096013965665, + "flow_matching_loss": 0.01335503300651908, + "grad_norm": 0.06754269450902939, + "learning_rate": 8.885358679928486e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19560 + }, + { + "epoch": 5.69391911550771, + "flow_matching_loss": 0.012880491744726896, + "grad_norm": 0.06643732637166977, + "learning_rate": 8.870263878954767e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19570 + }, + { + "epoch": 5.696828629618854, + "flow_matching_loss": 0.014076102618128061, + "grad_norm": 0.07397861778736115, + "learning_rate": 8.855176526188428e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19580 + }, + { + "epoch": 5.699738143729997, + "flow_matching_loss": 0.01436843890696764, + "grad_norm": 0.06804251670837402, + "learning_rate": 8.840096639962033e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19590 + }, + { + "epoch": 5.7026476578411405, + "flow_matching_loss": 0.013665927574038506, + "grad_norm": 0.07557784765958786, + "learning_rate": 8.825024238599056e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19600 + }, + { + "epoch": 5.705557171952284, + "flow_matching_loss": 0.01485659359022975, + "grad_norm": 0.0752531886100769, + "learning_rate": 8.809959340413891e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19610 + }, + { + "epoch": 5.708466686063428, + "flow_matching_loss": 0.013904489576816559, + "grad_norm": 0.0539797842502594, + "learning_rate": 8.79490196371181e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19620 + }, + { + "epoch": 5.711376200174571, + "flow_matching_loss": 0.01412653038278222, + "grad_norm": 0.09645655006170273, + "learning_rate": 8.779852126788949e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19630 + }, + { + "epoch": 5.714285714285714, + "flow_matching_loss": 0.015047381352633237, + "grad_norm": 0.07107476145029068, + "learning_rate": 8.764809847932277e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19640 + }, + { + "epoch": 5.717195228396857, + "flow_matching_loss": 0.013708317000418902, + "grad_norm": 0.0773264616727829, + "learning_rate": 8.749775145419577e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19650 + }, + { + "epoch": 5.720104742508001, + "flow_matching_loss": 0.016365718189626934, + "grad_norm": 0.07689926773309708, + "learning_rate": 8.734748037519427e-06, + "loss": 0.0164, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19660 + }, + { + "epoch": 5.723014256619145, + "flow_matching_loss": 0.01430869558826089, + "grad_norm": 0.08343633264303207, + "learning_rate": 8.719728542491195e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19670 + }, + { + "epoch": 5.725923770730288, + "flow_matching_loss": 0.014973381627351045, + "grad_norm": 0.0839294046163559, + "learning_rate": 8.704716678584976e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19680 + }, + { + "epoch": 5.728833284841431, + "flow_matching_loss": 0.0149082001298666, + "grad_norm": 0.08940339088439941, + "learning_rate": 8.689712464041596e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19690 + }, + { + "epoch": 5.731742798952575, + "flow_matching_loss": 0.014468807447701692, + "grad_norm": 0.08049774169921875, + "learning_rate": 8.674715917092592e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19700 + }, + { + "epoch": 5.734652313063719, + "flow_matching_loss": 0.014334093313664198, + "grad_norm": 0.09485798329114914, + "learning_rate": 8.65972705596018e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19710 + }, + { + "epoch": 5.737561827174861, + "flow_matching_loss": 0.013943265378475189, + "grad_norm": 0.07200029492378235, + "learning_rate": 8.64474589885725e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19720 + }, + { + "epoch": 5.740471341286005, + "flow_matching_loss": 0.014856323320418596, + "grad_norm": 0.07227042317390442, + "learning_rate": 8.629772463987315e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19730 + }, + { + "epoch": 5.743380855397149, + "flow_matching_loss": 0.013936101179569959, + "grad_norm": 0.08727487921714783, + "learning_rate": 8.614806769544507e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19740 + }, + { + "epoch": 5.7462903695082925, + "flow_matching_loss": 0.011969675216823817, + "grad_norm": 0.06679785251617432, + "learning_rate": 8.59984883371356e-06, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19750 + }, + { + "epoch": 5.749199883619435, + "flow_matching_loss": 0.014188953302800655, + "grad_norm": 0.06751339137554169, + "learning_rate": 8.584898674669772e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19760 + }, + { + "epoch": 5.752109397730579, + "flow_matching_loss": 0.014417252410203219, + "grad_norm": 0.09383241832256317, + "learning_rate": 8.569956310579001e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19770 + }, + { + "epoch": 5.755018911841723, + "flow_matching_loss": 0.013694533798843622, + "grad_norm": 0.09826914966106415, + "learning_rate": 8.555021759597627e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19780 + }, + { + "epoch": 5.757928425952866, + "flow_matching_loss": 0.015949592739343644, + "grad_norm": 0.07931877672672272, + "learning_rate": 8.54009503987253e-06, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19790 + }, + { + "epoch": 5.760837940064009, + "flow_matching_loss": 0.013000521529465913, + "grad_norm": 0.06406965851783752, + "learning_rate": 8.525176169541095e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19800 + }, + { + "epoch": 5.763747454175153, + "flow_matching_loss": 0.014789592660963535, + "grad_norm": 0.07257941365242004, + "learning_rate": 8.510265166731151e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19810 + }, + { + "epoch": 5.766656968286296, + "flow_matching_loss": 0.014403690025210381, + "grad_norm": 0.07238893210887909, + "learning_rate": 8.495362049560964e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19820 + }, + { + "epoch": 5.7695664823974395, + "flow_matching_loss": 0.013737293612211942, + "grad_norm": 0.0941644161939621, + "learning_rate": 8.480466836139241e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19830 + }, + { + "epoch": 5.772475996508583, + "flow_matching_loss": 0.013831727299839258, + "grad_norm": 0.06765419989824295, + "learning_rate": 8.465579544565066e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19840 + }, + { + "epoch": 5.775385510619726, + "flow_matching_loss": 0.01563181234523654, + "grad_norm": 0.1012405976653099, + "learning_rate": 8.450700192927897e-06, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19850 + }, + { + "epoch": 5.77829502473087, + "flow_matching_loss": 0.013661278132349252, + "grad_norm": 0.062113165855407715, + "learning_rate": 8.435828799307551e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19860 + }, + { + "epoch": 5.7812045388420135, + "flow_matching_loss": 0.01354493759572506, + "grad_norm": 0.0718725398182869, + "learning_rate": 8.420965381774172e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19870 + }, + { + "epoch": 5.784114052953157, + "flow_matching_loss": 0.01402540747076273, + "grad_norm": 0.09290974587202072, + "learning_rate": 8.406109958388222e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19880 + }, + { + "epoch": 5.7870235670643, + "flow_matching_loss": 0.01389267323538661, + "grad_norm": 0.07864680141210556, + "learning_rate": 8.391262547200434e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19890 + }, + { + "epoch": 5.789933081175444, + "flow_matching_loss": 0.015119002852588891, + "grad_norm": 0.08533994853496552, + "learning_rate": 8.376423166251816e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19900 + }, + { + "epoch": 5.792842595286587, + "flow_matching_loss": 0.014529847167432308, + "grad_norm": 0.06666699051856995, + "learning_rate": 8.361591833573612e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19910 + }, + { + "epoch": 5.79575210939773, + "flow_matching_loss": 0.014812623523175717, + "grad_norm": 0.06958672404289246, + "learning_rate": 8.346768567187287e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19920 + }, + { + "epoch": 5.798661623508874, + "flow_matching_loss": 0.013574155978858471, + "grad_norm": 0.05884311720728874, + "learning_rate": 8.331953385104516e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19930 + }, + { + "epoch": 5.801571137620018, + "flow_matching_loss": 0.013698898814618587, + "grad_norm": 0.07386980205774307, + "learning_rate": 8.317146305327134e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19940 + }, + { + "epoch": 5.804480651731161, + "flow_matching_loss": 0.014401399064809084, + "grad_norm": 0.07622580975294113, + "learning_rate": 8.302347345847146e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19950 + }, + { + "epoch": 5.807390165842304, + "flow_matching_loss": 0.014592710137367248, + "grad_norm": 0.06870995461940765, + "learning_rate": 8.287556524646683e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19960 + }, + { + "epoch": 5.810299679953448, + "flow_matching_loss": 0.013259923923760653, + "grad_norm": 0.054225001484155655, + "learning_rate": 8.272773859697978e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19970 + }, + { + "epoch": 5.813209194064592, + "flow_matching_loss": 0.013414818234741688, + "grad_norm": 0.08845765888690948, + "learning_rate": 8.257999368963374e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19980 + }, + { + "epoch": 5.816118708175734, + "flow_matching_loss": 0.01395297097042203, + "grad_norm": 0.07445809990167618, + "learning_rate": 8.243233070395266e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 19990 + }, + { + "epoch": 5.819028222286878, + "flow_matching_loss": 0.014504988491535187, + "grad_norm": 0.07355999201536179, + "learning_rate": 8.2284749819361e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20000 + }, + { + "epoch": 5.821937736398022, + "flow_matching_loss": 0.015099234785884619, + "grad_norm": 0.08766761422157288, + "learning_rate": 8.213725121518346e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20010 + }, + { + "epoch": 5.824847250509165, + "flow_matching_loss": 0.015444792062044143, + "grad_norm": 0.08836955577135086, + "learning_rate": 8.198983507064469e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20020 + }, + { + "epoch": 5.827756764620308, + "flow_matching_loss": 0.014990715030580758, + "grad_norm": 0.08063908666372299, + "learning_rate": 8.18425015648692e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20030 + }, + { + "epoch": 5.830666278731452, + "flow_matching_loss": 0.01430626641958952, + "grad_norm": 0.07673446089029312, + "learning_rate": 8.169525087688122e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20040 + }, + { + "epoch": 5.833575792842595, + "flow_matching_loss": 0.015210375189781189, + "grad_norm": 0.09960515052080154, + "learning_rate": 8.154808318560413e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20050 + }, + { + "epoch": 5.836485306953739, + "flow_matching_loss": 0.013684631604701281, + "grad_norm": 0.07480380684137344, + "learning_rate": 8.140099866986058e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20060 + }, + { + "epoch": 5.839394821064882, + "flow_matching_loss": 0.013105819001793861, + "grad_norm": 0.0761236622929573, + "learning_rate": 8.125399750837202e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20070 + }, + { + "epoch": 5.842304335176026, + "flow_matching_loss": 0.013975527603179216, + "grad_norm": 0.07633239775896072, + "learning_rate": 8.110707987975888e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20080 + }, + { + "epoch": 5.845213849287169, + "flow_matching_loss": 0.015163173340260983, + "grad_norm": 0.06959670782089233, + "learning_rate": 8.096024596253982e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20090 + }, + { + "epoch": 5.8481233633983125, + "flow_matching_loss": 0.013459863886237145, + "grad_norm": 0.07972502708435059, + "learning_rate": 8.081349593513197e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20100 + }, + { + "epoch": 5.851032877509456, + "flow_matching_loss": 0.01354650678113103, + "grad_norm": 0.08303926140069962, + "learning_rate": 8.066682997585045e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20110 + }, + { + "epoch": 5.853942391620599, + "flow_matching_loss": 0.013530223164707422, + "grad_norm": 0.11618082225322723, + "learning_rate": 8.052024826290819e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20120 + }, + { + "epoch": 5.856851905731743, + "flow_matching_loss": 0.016269932501018047, + "grad_norm": 0.10512907803058624, + "learning_rate": 8.037375097441574e-06, + "loss": 0.0163, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20130 + }, + { + "epoch": 5.8597614198428865, + "flow_matching_loss": 0.013669048435986042, + "grad_norm": 0.07429081201553345, + "learning_rate": 8.022733828838124e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20140 + }, + { + "epoch": 5.86267093395403, + "flow_matching_loss": 0.015007939841598272, + "grad_norm": 0.08141826838254929, + "learning_rate": 8.008101038270988e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20150 + }, + { + "epoch": 5.865580448065173, + "flow_matching_loss": 0.014831140078604222, + "grad_norm": 0.07837033271789551, + "learning_rate": 7.99347674352038e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20160 + }, + { + "epoch": 5.868489962176317, + "flow_matching_loss": 0.013828329648822546, + "grad_norm": 0.06071313098073006, + "learning_rate": 7.978860962356202e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20170 + }, + { + "epoch": 5.8713994762874595, + "flow_matching_loss": 0.014217355567961932, + "grad_norm": 0.09274650365114212, + "learning_rate": 7.964253712537997e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20180 + }, + { + "epoch": 5.874308990398603, + "flow_matching_loss": 0.014051559707149863, + "grad_norm": 0.06358139216899872, + "learning_rate": 7.949655011814968e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20190 + }, + { + "epoch": 5.877218504509747, + "flow_matching_loss": 0.013316015247255564, + "grad_norm": 0.07063622027635574, + "learning_rate": 7.9350648779259e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20200 + }, + { + "epoch": 5.880128018620891, + "flow_matching_loss": 0.013728983607143164, + "grad_norm": 0.07366438210010529, + "learning_rate": 7.92048332859919e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20210 + }, + { + "epoch": 5.8830375327320334, + "flow_matching_loss": 0.015099959261715413, + "grad_norm": 0.08330964297056198, + "learning_rate": 7.90591038155278e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20220 + }, + { + "epoch": 5.885947046843177, + "flow_matching_loss": 0.01317483289167285, + "grad_norm": 0.07613707333803177, + "learning_rate": 7.891346054494194e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20230 + }, + { + "epoch": 5.888856560954321, + "flow_matching_loss": 0.014225766528397799, + "grad_norm": 0.08330544084310532, + "learning_rate": 7.876790365120447e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20240 + }, + { + "epoch": 5.891766075065464, + "flow_matching_loss": 0.013986371271312237, + "grad_norm": 0.06707387417554855, + "learning_rate": 7.862243331118088e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20250 + }, + { + "epoch": 5.894675589176607, + "flow_matching_loss": 0.013851267565041781, + "grad_norm": 0.07314996421337128, + "learning_rate": 7.847704970163127e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20260 + }, + { + "epoch": 5.897585103287751, + "flow_matching_loss": 0.014111886359751224, + "grad_norm": 0.08555448800325394, + "learning_rate": 7.833175299921046e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20270 + }, + { + "epoch": 5.900494617398895, + "flow_matching_loss": 0.013648872030898929, + "grad_norm": 0.07605095207691193, + "learning_rate": 7.81865433804676e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20280 + }, + { + "epoch": 5.903404131510038, + "flow_matching_loss": 0.014380007702857255, + "grad_norm": 0.08584292232990265, + "learning_rate": 7.80414210218461e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20290 + }, + { + "epoch": 5.906313645621181, + "flow_matching_loss": 0.013513617962598801, + "grad_norm": 0.06928573548793793, + "learning_rate": 7.789638609968332e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20300 + }, + { + "epoch": 5.909223159732325, + "flow_matching_loss": 0.014847537316381932, + "grad_norm": 0.06868329644203186, + "learning_rate": 7.775143879021037e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20310 + }, + { + "epoch": 5.912132673843468, + "flow_matching_loss": 0.014344074949622154, + "grad_norm": 0.08376650512218475, + "learning_rate": 7.760657926955187e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20320 + }, + { + "epoch": 5.915042187954612, + "flow_matching_loss": 0.013599287252873182, + "grad_norm": 0.0847613736987114, + "learning_rate": 7.746180771372584e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20330 + }, + { + "epoch": 5.917951702065755, + "flow_matching_loss": 0.014062853064388037, + "grad_norm": 0.07099830359220505, + "learning_rate": 7.731712429864325e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20340 + }, + { + "epoch": 5.920861216176898, + "flow_matching_loss": 0.01407105652615428, + "grad_norm": 0.07714832574129105, + "learning_rate": 7.717252920010828e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20350 + }, + { + "epoch": 5.923770730288042, + "flow_matching_loss": 0.015465126931667328, + "grad_norm": 0.09067614376544952, + "learning_rate": 7.702802259381742e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20360 + }, + { + "epoch": 5.9266802443991855, + "flow_matching_loss": 0.01501097045838833, + "grad_norm": 0.09232568740844727, + "learning_rate": 7.688360465535999e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20370 + }, + { + "epoch": 5.929589758510328, + "flow_matching_loss": 0.01393065694719553, + "grad_norm": 0.0804368406534195, + "learning_rate": 7.673927556021734e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20380 + }, + { + "epoch": 5.932499272621472, + "flow_matching_loss": 0.015563889779150486, + "grad_norm": 0.08923529088497162, + "learning_rate": 7.659503548376284e-06, + "loss": 0.0156, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20390 + }, + { + "epoch": 5.935408786732616, + "flow_matching_loss": 0.01349063590168953, + "grad_norm": 0.08321642130613327, + "learning_rate": 7.645088460126193e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20400 + }, + { + "epoch": 5.9383183008437594, + "flow_matching_loss": 0.014122496731579304, + "grad_norm": 0.06501735001802444, + "learning_rate": 7.630682308787145e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20410 + }, + { + "epoch": 5.941227814954902, + "flow_matching_loss": 0.014395196828991175, + "grad_norm": 0.07032398134469986, + "learning_rate": 7.6162851118639754e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20420 + }, + { + "epoch": 5.944137329066046, + "flow_matching_loss": 0.013264912087470293, + "grad_norm": 0.06012444198131561, + "learning_rate": 7.601896886850632e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20430 + }, + { + "epoch": 5.94704684317719, + "flow_matching_loss": 0.015427252650260926, + "grad_norm": 0.1476536989212036, + "learning_rate": 7.587517651230168e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20440 + }, + { + "epoch": 5.9499563572883325, + "flow_matching_loss": 0.014675283245742321, + "grad_norm": 0.08538325130939484, + "learning_rate": 7.573147422474702e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20450 + }, + { + "epoch": 5.952865871399476, + "flow_matching_loss": 0.013292931951582432, + "grad_norm": 0.06502033025026321, + "learning_rate": 7.55878621804543e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20460 + }, + { + "epoch": 5.95577538551062, + "flow_matching_loss": 0.014537099935114383, + "grad_norm": 0.07112442702054977, + "learning_rate": 7.544434055392564e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20470 + }, + { + "epoch": 5.958684899621764, + "flow_matching_loss": 0.013996171113103629, + "grad_norm": 0.0714881643652916, + "learning_rate": 7.530090951955334e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20480 + }, + { + "epoch": 5.961594413732906, + "flow_matching_loss": 0.014244860224425792, + "grad_norm": 0.06602444499731064, + "learning_rate": 7.515756925161959e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20490 + }, + { + "epoch": 5.96450392784405, + "flow_matching_loss": 0.013637705147266388, + "grad_norm": 0.09599675983190536, + "learning_rate": 7.501431992429643e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20500 + }, + { + "epoch": 5.967413441955194, + "flow_matching_loss": 0.0139633703045547, + "grad_norm": 0.07262294739484787, + "learning_rate": 7.487116171164519e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20510 + }, + { + "epoch": 5.970322956066337, + "flow_matching_loss": 0.014682600367814302, + "grad_norm": 0.07916159182786942, + "learning_rate": 7.472809478761672e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20520 + }, + { + "epoch": 5.97323247017748, + "flow_matching_loss": 0.013564885873347522, + "grad_norm": 0.06802450865507126, + "learning_rate": 7.4585119326050775e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20530 + }, + { + "epoch": 5.976141984288624, + "flow_matching_loss": 0.013458718825131655, + "grad_norm": 0.06298468261957169, + "learning_rate": 7.444223550067599e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20540 + }, + { + "epoch": 5.979051498399767, + "flow_matching_loss": 0.013161344546824693, + "grad_norm": 0.06183139979839325, + "learning_rate": 7.429944348510964e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20550 + }, + { + "epoch": 5.981961012510911, + "flow_matching_loss": 0.015255657397210598, + "grad_norm": 0.10469596832990646, + "learning_rate": 7.415674345285763e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20560 + }, + { + "epoch": 5.984870526622054, + "flow_matching_loss": 0.013769308105111122, + "grad_norm": 0.06591331958770752, + "learning_rate": 7.401413557731387e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20570 + }, + { + "epoch": 5.987780040733197, + "flow_matching_loss": 0.01304428968578577, + "grad_norm": 0.06953418254852295, + "learning_rate": 7.387162003176037e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20580 + }, + { + "epoch": 5.990689554844341, + "flow_matching_loss": 0.013733491115272046, + "grad_norm": 0.06225340813398361, + "learning_rate": 7.372919698936697e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20590 + }, + { + "epoch": 5.9935990689554846, + "flow_matching_loss": 0.01566469995304942, + "grad_norm": 0.08662894368171692, + "learning_rate": 7.358686662319104e-06, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20600 + }, + { + "epoch": 5.996508583066628, + "flow_matching_loss": 0.01314371544867754, + "grad_norm": 0.06061619147658348, + "learning_rate": 7.344462910617752e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20610 + }, + { + "epoch": 5.999418097177771, + "flow_matching_loss": 0.013418169226497412, + "grad_norm": 0.06300526857376099, + "learning_rate": 7.330248461115832e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20620 + }, + { + "epoch": 6.002327611288915, + "flow_matching_loss": 0.013556200079619884, + "grad_norm": 0.06835675239562988, + "learning_rate": 7.316043331085242e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20630 + }, + { + "epoch": 6.0052371254000585, + "flow_matching_loss": 0.01491622133180499, + "grad_norm": 0.094411700963974, + "learning_rate": 7.30184753778655e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20640 + }, + { + "epoch": 6.008146639511201, + "flow_matching_loss": 0.014850405696779489, + "grad_norm": 0.07396261394023895, + "learning_rate": 7.2876610984689935e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20650 + }, + { + "epoch": 6.011056153622345, + "flow_matching_loss": 0.013783895596861839, + "grad_norm": 0.05937230959534645, + "learning_rate": 7.273484030370423e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20660 + }, + { + "epoch": 6.013965667733489, + "flow_matching_loss": 0.014657514076679945, + "grad_norm": 0.0807560384273529, + "learning_rate": 7.259316350717324e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20670 + }, + { + "epoch": 6.0168751818446315, + "flow_matching_loss": 0.013405325170606375, + "grad_norm": 0.07523629814386368, + "learning_rate": 7.245158076724757e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20680 + }, + { + "epoch": 6.019784695955775, + "flow_matching_loss": 0.013624921906739473, + "grad_norm": 0.059157032519578934, + "learning_rate": 7.23100922559636e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20690 + }, + { + "epoch": 6.022694210066919, + "flow_matching_loss": 0.014940113853663206, + "grad_norm": 0.08784077316522598, + "learning_rate": 7.216869814524323e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20700 + }, + { + "epoch": 6.025603724178063, + "flow_matching_loss": 0.014113074447959662, + "grad_norm": 0.06443515419960022, + "learning_rate": 7.202739860689357e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20710 + }, + { + "epoch": 6.0285132382892055, + "flow_matching_loss": 0.013223251327872276, + "grad_norm": 0.0781756266951561, + "learning_rate": 7.188619381260698e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20720 + }, + { + "epoch": 6.031422752400349, + "flow_matching_loss": 0.01453910805284977, + "grad_norm": 0.0533694289624691, + "learning_rate": 7.174508393396057e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20730 + }, + { + "epoch": 6.034332266511493, + "flow_matching_loss": 0.014316313993185758, + "grad_norm": 0.07835479080677032, + "learning_rate": 7.160406914241615e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20740 + }, + { + "epoch": 6.037241780622636, + "flow_matching_loss": 0.013392299599945545, + "grad_norm": 0.0750868171453476, + "learning_rate": 7.146314960931996e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20750 + }, + { + "epoch": 6.040151294733779, + "flow_matching_loss": 0.013830903638154268, + "grad_norm": 0.1067124530673027, + "learning_rate": 7.132232550590247e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20760 + }, + { + "epoch": 6.043060808844923, + "flow_matching_loss": 0.014013542234897614, + "grad_norm": 0.06091924384236336, + "learning_rate": 7.118159700327839e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20770 + }, + { + "epoch": 6.045970322956066, + "flow_matching_loss": 0.013451365567743779, + "grad_norm": 0.06327421963214874, + "learning_rate": 7.104096427244603e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20780 + }, + { + "epoch": 6.04887983706721, + "flow_matching_loss": 0.013756188657134771, + "grad_norm": 0.07084129750728607, + "learning_rate": 7.090042748428737e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20790 + }, + { + "epoch": 6.051789351178353, + "flow_matching_loss": 0.013082756940275431, + "grad_norm": 0.08562427014112473, + "learning_rate": 7.075998680956797e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20800 + }, + { + "epoch": 6.054698865289497, + "flow_matching_loss": 0.012663019262254238, + "grad_norm": 0.06215748190879822, + "learning_rate": 7.061964241893639e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20810 + }, + { + "epoch": 6.05760837940064, + "flow_matching_loss": 0.01539806192740798, + "grad_norm": 0.08502383530139923, + "learning_rate": 7.047939448292442e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20820 + }, + { + "epoch": 6.060517893511784, + "flow_matching_loss": 0.013169512897729874, + "grad_norm": 0.07282809168100357, + "learning_rate": 7.033924317194644e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20830 + }, + { + "epoch": 6.063427407622927, + "flow_matching_loss": 0.0137340790592134, + "grad_norm": 0.07349980622529984, + "learning_rate": 7.0199188656299524e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20840 + }, + { + "epoch": 6.06633692173407, + "flow_matching_loss": 0.012988741975277662, + "grad_norm": 0.08323325961828232, + "learning_rate": 7.005923110616313e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20850 + }, + { + "epoch": 6.069246435845214, + "flow_matching_loss": 0.014029187057167292, + "grad_norm": 0.07824993878602982, + "learning_rate": 6.9919370691598855e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20860 + }, + { + "epoch": 6.0721559499563575, + "flow_matching_loss": 0.013497325871139765, + "grad_norm": 0.06327732652425766, + "learning_rate": 6.977960758255024e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20870 + }, + { + "epoch": 6.0750654640675, + "flow_matching_loss": 0.013722582999616862, + "grad_norm": 0.06879319995641708, + "learning_rate": 6.963994194884276e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20880 + }, + { + "epoch": 6.077974978178644, + "flow_matching_loss": 0.014788540545850992, + "grad_norm": 0.09766296297311783, + "learning_rate": 6.950037396018328e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20890 + }, + { + "epoch": 6.080884492289788, + "flow_matching_loss": 0.014514585956931114, + "grad_norm": 0.06655889004468918, + "learning_rate": 6.9360903786160056e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20900 + }, + { + "epoch": 6.083794006400931, + "flow_matching_loss": 0.01320246672257781, + "grad_norm": 0.06340334564447403, + "learning_rate": 6.9221531596242485e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20910 + }, + { + "epoch": 6.086703520512074, + "flow_matching_loss": 0.014280240470543503, + "grad_norm": 0.06954594701528549, + "learning_rate": 6.908225755978089e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20920 + }, + { + "epoch": 6.089613034623218, + "flow_matching_loss": 0.013630278781056404, + "grad_norm": 0.07117926329374313, + "learning_rate": 6.8943081846006466e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20930 + }, + { + "epoch": 6.092522548734362, + "flow_matching_loss": 0.014392923377454282, + "grad_norm": 0.09033478051424026, + "learning_rate": 6.88040046240307e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20940 + }, + { + "epoch": 6.0954320628455045, + "flow_matching_loss": 0.013773900177329778, + "grad_norm": 0.07175427675247192, + "learning_rate": 6.866502606284566e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20950 + }, + { + "epoch": 6.098341576956648, + "flow_matching_loss": 0.013849829696118832, + "grad_norm": 0.06937547028064728, + "learning_rate": 6.852614633132335e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20960 + }, + { + "epoch": 6.101251091067792, + "flow_matching_loss": 0.014484397973865271, + "grad_norm": 0.09119633585214615, + "learning_rate": 6.8387365598215665e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20970 + }, + { + "epoch": 6.104160605178935, + "flow_matching_loss": 0.014576097950339317, + "grad_norm": 0.07566185295581818, + "learning_rate": 6.82486840321544e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20980 + }, + { + "epoch": 6.1070701192900785, + "flow_matching_loss": 0.015404364187270403, + "grad_norm": 0.08344325423240662, + "learning_rate": 6.811010180165069e-06, + "loss": 0.0154, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 20990 + }, + { + "epoch": 6.109979633401222, + "flow_matching_loss": 0.012731384113430977, + "grad_norm": 0.07462110370397568, + "learning_rate": 6.797161907509499e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21000 + }, + { + "epoch": 6.112889147512365, + "flow_matching_loss": 0.01294850604608655, + "grad_norm": 0.08715935051441193, + "learning_rate": 6.783323602075688e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21010 + }, + { + "epoch": 6.115798661623509, + "flow_matching_loss": 0.012590233888477086, + "grad_norm": 0.0762113481760025, + "learning_rate": 6.7694952806784784e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21020 + }, + { + "epoch": 6.118708175734652, + "flow_matching_loss": 0.014170585945248605, + "grad_norm": 0.09181688725948334, + "learning_rate": 6.7556769601205944e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21030 + }, + { + "epoch": 6.121617689845796, + "flow_matching_loss": 0.015989256836473942, + "grad_norm": 0.06950598210096359, + "learning_rate": 6.741868657192594e-06, + "loss": 0.016, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21040 + }, + { + "epoch": 6.124527203956939, + "flow_matching_loss": 0.014221334923058748, + "grad_norm": 0.07909749448299408, + "learning_rate": 6.728070388672864e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21050 + }, + { + "epoch": 6.127436718068083, + "flow_matching_loss": 0.01340945204719901, + "grad_norm": 0.06428062915802002, + "learning_rate": 6.714282171327604e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21060 + }, + { + "epoch": 6.130346232179226, + "flow_matching_loss": 0.013415731862187386, + "grad_norm": 0.06765471398830414, + "learning_rate": 6.700504021910791e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21070 + }, + { + "epoch": 6.133255746290369, + "flow_matching_loss": 0.013184472918510437, + "grad_norm": 0.05209539458155632, + "learning_rate": 6.68673595716419e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21080 + }, + { + "epoch": 6.136165260401513, + "flow_matching_loss": 0.014431306160986423, + "grad_norm": 0.06317262351512909, + "learning_rate": 6.6729779938172856e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21090 + }, + { + "epoch": 6.139074774512657, + "flow_matching_loss": 0.01336613493040204, + "grad_norm": 0.08697149157524109, + "learning_rate": 6.659230148587311e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21100 + }, + { + "epoch": 6.141984288623799, + "flow_matching_loss": 0.013025241065770387, + "grad_norm": 0.07936728745698929, + "learning_rate": 6.645492438179188e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21110 + }, + { + "epoch": 6.144893802734943, + "flow_matching_loss": 0.013131159637123347, + "grad_norm": 0.05360887572169304, + "learning_rate": 6.631764879285531e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21120 + }, + { + "epoch": 6.147803316846087, + "flow_matching_loss": 0.013403280451893807, + "grad_norm": 0.09067647904157639, + "learning_rate": 6.618047488586614e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21130 + }, + { + "epoch": 6.1507128309572305, + "flow_matching_loss": 0.014772513974457979, + "grad_norm": 0.06492524594068527, + "learning_rate": 6.604340282750371e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21140 + }, + { + "epoch": 6.153622345068373, + "flow_matching_loss": 0.013828552421182394, + "grad_norm": 0.07932925224304199, + "learning_rate": 6.590643278432342e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21150 + }, + { + "epoch": 6.156531859179517, + "flow_matching_loss": 0.013757641892880201, + "grad_norm": 0.06907790154218674, + "learning_rate": 6.576956492275682e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21160 + }, + { + "epoch": 6.159441373290661, + "flow_matching_loss": 0.015046664327383042, + "grad_norm": 0.06481286138296127, + "learning_rate": 6.563279940911124e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21170 + }, + { + "epoch": 6.162350887401804, + "flow_matching_loss": 0.012430801056325436, + "grad_norm": 0.06835377961397171, + "learning_rate": 6.549613640956963e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21180 + }, + { + "epoch": 6.165260401512947, + "flow_matching_loss": 0.014910814631730319, + "grad_norm": 0.08025582879781723, + "learning_rate": 6.535957609019053e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21190 + }, + { + "epoch": 6.168169915624091, + "flow_matching_loss": 0.013913018256425857, + "grad_norm": 0.053222622722387314, + "learning_rate": 6.5223118616907545e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21200 + }, + { + "epoch": 6.171079429735234, + "flow_matching_loss": 0.014184026513248682, + "grad_norm": 0.081902414560318, + "learning_rate": 6.508676415552939e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21210 + }, + { + "epoch": 6.1739889438463775, + "flow_matching_loss": 0.015171585604548454, + "grad_norm": 0.07757691293954849, + "learning_rate": 6.495051287173952e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21220 + }, + { + "epoch": 6.176898457957521, + "flow_matching_loss": 0.01373736709356308, + "grad_norm": 0.09276857227087021, + "learning_rate": 6.481436493109621e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21230 + }, + { + "epoch": 6.179807972068665, + "flow_matching_loss": 0.014689796976745129, + "grad_norm": 0.07690567523241043, + "learning_rate": 6.467832049903194e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21240 + }, + { + "epoch": 6.182717486179808, + "flow_matching_loss": 0.015460456162691117, + "grad_norm": 0.08984031528234482, + "learning_rate": 6.454237974085361e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21250 + }, + { + "epoch": 6.1856270002909515, + "flow_matching_loss": 0.012941820919513703, + "grad_norm": 0.08031286299228668, + "learning_rate": 6.440654282174203e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21260 + }, + { + "epoch": 6.188536514402095, + "flow_matching_loss": 0.014365424029529094, + "grad_norm": 0.08140640705823898, + "learning_rate": 6.427080990675187e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21270 + }, + { + "epoch": 6.191446028513238, + "flow_matching_loss": 0.014282458741217852, + "grad_norm": 0.085521399974823, + "learning_rate": 6.41351811608114e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21280 + }, + { + "epoch": 6.194355542624382, + "flow_matching_loss": 0.013385900761932135, + "grad_norm": 0.06316952407360077, + "learning_rate": 6.399965674872228e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21290 + }, + { + "epoch": 6.197265056735525, + "flow_matching_loss": 0.01491516325622797, + "grad_norm": 0.07573876529932022, + "learning_rate": 6.386423683515957e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21300 + }, + { + "epoch": 6.200174570846668, + "flow_matching_loss": 0.014253875147551298, + "grad_norm": 0.09048546850681305, + "learning_rate": 6.372892158467119e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21310 + }, + { + "epoch": 6.203084084957812, + "flow_matching_loss": 0.0144011159427464, + "grad_norm": 0.09952786564826965, + "learning_rate": 6.359371116167791e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21320 + }, + { + "epoch": 6.205993599068956, + "flow_matching_loss": 0.012829269841313362, + "grad_norm": 0.07530386745929718, + "learning_rate": 6.345860573047314e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21330 + }, + { + "epoch": 6.208903113180099, + "flow_matching_loss": 0.014703129138797521, + "grad_norm": 0.08028355240821838, + "learning_rate": 6.332360545522268e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21340 + }, + { + "epoch": 6.211812627291242, + "flow_matching_loss": 0.01456295931711793, + "grad_norm": 0.09765908122062683, + "learning_rate": 6.31887104999647e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21350 + }, + { + "epoch": 6.214722141402386, + "flow_matching_loss": 0.013198340870440006, + "grad_norm": 0.08704812824726105, + "learning_rate": 6.305392102860924e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21360 + }, + { + "epoch": 6.21763165551353, + "flow_matching_loss": 0.012990807089954615, + "grad_norm": 0.06474843621253967, + "learning_rate": 6.291923720493814e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21370 + }, + { + "epoch": 6.220541169624672, + "flow_matching_loss": 0.014385289791971446, + "grad_norm": 0.070151187479496, + "learning_rate": 6.278465919260507e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21380 + }, + { + "epoch": 6.223450683735816, + "flow_matching_loss": 0.013015955872833728, + "grad_norm": 0.1043245866894722, + "learning_rate": 6.26501871551349e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21390 + }, + { + "epoch": 6.22636019784696, + "flow_matching_loss": 0.01314411936327815, + "grad_norm": 0.07527510821819305, + "learning_rate": 6.251582125592394e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21400 + }, + { + "epoch": 6.229269711958103, + "flow_matching_loss": 0.013604086730629206, + "grad_norm": 0.07093317806720734, + "learning_rate": 6.2381561658239386e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21410 + }, + { + "epoch": 6.232179226069246, + "flow_matching_loss": 0.012944313418120145, + "grad_norm": 0.0957547277212143, + "learning_rate": 6.2247408525219276e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21420 + }, + { + "epoch": 6.23508874018039, + "flow_matching_loss": 0.013591976277530194, + "grad_norm": 0.0643240362405777, + "learning_rate": 6.2113362019872335e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21430 + }, + { + "epoch": 6.237998254291533, + "flow_matching_loss": 0.013068357948213816, + "grad_norm": 0.0761457085609436, + "learning_rate": 6.197942230507772e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21440 + }, + { + "epoch": 6.240907768402677, + "flow_matching_loss": 0.014157952275127172, + "grad_norm": 0.07620974630117416, + "learning_rate": 6.184558954358475e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21450 + }, + { + "epoch": 6.24381728251382, + "flow_matching_loss": 0.014501601364463568, + "grad_norm": 0.08758077770471573, + "learning_rate": 6.1711863898012965e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21460 + }, + { + "epoch": 6.246726796624964, + "flow_matching_loss": 0.013535264693200589, + "grad_norm": 0.07695110142230988, + "learning_rate": 6.157824553085155e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21470 + }, + { + "epoch": 6.249636310736107, + "flow_matching_loss": 0.013501236028969289, + "grad_norm": 0.06207139045000076, + "learning_rate": 6.144473460445945e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21480 + }, + { + "epoch": 6.2525458248472505, + "flow_matching_loss": 0.013520921021699906, + "grad_norm": 0.09831635653972626, + "learning_rate": 6.131133128106503e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21490 + }, + { + "epoch": 6.255455338958394, + "flow_matching_loss": 0.013784234505146742, + "grad_norm": 0.07329009473323822, + "learning_rate": 6.117803572276584e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21500 + }, + { + "epoch": 6.258364853069537, + "flow_matching_loss": 0.013748167827725411, + "grad_norm": 0.08190511912107468, + "learning_rate": 6.104484809152866e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21510 + }, + { + "epoch": 6.261274367180681, + "flow_matching_loss": 0.015129808615893126, + "grad_norm": 0.09078273177146912, + "learning_rate": 6.091176854918892e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21520 + }, + { + "epoch": 6.2641838812918245, + "flow_matching_loss": 0.014566926751285791, + "grad_norm": 0.07276182621717453, + "learning_rate": 6.077879725745092e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21530 + }, + { + "epoch": 6.267093395402968, + "flow_matching_loss": 0.014469590317457915, + "grad_norm": 0.058758560568094254, + "learning_rate": 6.064593437788727e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21540 + }, + { + "epoch": 6.270002909514111, + "flow_matching_loss": 0.01332934880629182, + "grad_norm": 0.07279109954833984, + "learning_rate": 6.051318007193881e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21550 + }, + { + "epoch": 6.272912423625255, + "flow_matching_loss": 0.013451437558978796, + "grad_norm": 0.08369569480419159, + "learning_rate": 6.038053450091469e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21560 + }, + { + "epoch": 6.275821937736398, + "flow_matching_loss": 0.014330238942056895, + "grad_norm": 0.07587255537509918, + "learning_rate": 6.0247997825991705e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21570 + }, + { + "epoch": 6.278731451847541, + "flow_matching_loss": 0.014002854097634554, + "grad_norm": 0.07639221847057343, + "learning_rate": 6.011557020821444e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21580 + }, + { + "epoch": 6.281640965958685, + "flow_matching_loss": 0.013703750725835562, + "grad_norm": 0.054455529898405075, + "learning_rate": 5.99832518084949e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21590 + }, + { + "epoch": 6.284550480069829, + "flow_matching_loss": 0.014731973502784967, + "grad_norm": 0.06927565485239029, + "learning_rate": 5.98510427876124e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21600 + }, + { + "epoch": 6.2874599941809715, + "flow_matching_loss": 0.014050720259547234, + "grad_norm": 0.07596131414175034, + "learning_rate": 5.971894330621344e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21610 + }, + { + "epoch": 6.290369508292115, + "flow_matching_loss": 0.014756939746439457, + "grad_norm": 0.06253991276025772, + "learning_rate": 5.958695352481134e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21620 + }, + { + "epoch": 6.293279022403259, + "flow_matching_loss": 0.012812863755971193, + "grad_norm": 0.07234927266836166, + "learning_rate": 5.945507360378608e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21630 + }, + { + "epoch": 6.296188536514402, + "flow_matching_loss": 0.015236531291157008, + "grad_norm": 0.08152704685926437, + "learning_rate": 5.9323303703384225e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21640 + }, + { + "epoch": 6.299098050625545, + "flow_matching_loss": 0.013668009825050832, + "grad_norm": 0.09131752699613571, + "learning_rate": 5.919164398371857e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21650 + }, + { + "epoch": 6.302007564736689, + "flow_matching_loss": 0.013253286853432656, + "grad_norm": 0.07726700603961945, + "learning_rate": 5.906009460476818e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21660 + }, + { + "epoch": 6.304917078847833, + "flow_matching_loss": 0.014442135114222765, + "grad_norm": 0.07749588787555695, + "learning_rate": 5.892865572637789e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21670 + }, + { + "epoch": 6.307826592958976, + "flow_matching_loss": 0.01352814668789506, + "grad_norm": 0.06412413716316223, + "learning_rate": 5.8797327508258396e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21680 + }, + { + "epoch": 6.310736107070119, + "flow_matching_loss": 0.014054200612008572, + "grad_norm": 0.060113683342933655, + "learning_rate": 5.866611010998584e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21690 + }, + { + "epoch": 6.313645621181263, + "flow_matching_loss": 0.014165872149169445, + "grad_norm": 0.05995707958936691, + "learning_rate": 5.853500369100172e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21700 + }, + { + "epoch": 6.316555135292406, + "flow_matching_loss": 0.013126329705119132, + "grad_norm": 0.06919566541910172, + "learning_rate": 5.840400841061265e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21710 + }, + { + "epoch": 6.31946464940355, + "flow_matching_loss": 0.012708070315420627, + "grad_norm": 0.08498122543096542, + "learning_rate": 5.827312442799034e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21720 + }, + { + "epoch": 6.322374163514693, + "flow_matching_loss": 0.012333287950605154, + "grad_norm": 0.06998780369758606, + "learning_rate": 5.814235190217114e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21730 + }, + { + "epoch": 6.325283677625836, + "flow_matching_loss": 0.013083042949438095, + "grad_norm": 0.09094341844320297, + "learning_rate": 5.801169099205599e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21740 + }, + { + "epoch": 6.32819319173698, + "flow_matching_loss": 0.01343791550025344, + "grad_norm": 0.0786413699388504, + "learning_rate": 5.788114185641022e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21750 + }, + { + "epoch": 6.3311027058481235, + "flow_matching_loss": 0.013511177897453309, + "grad_norm": 0.059557341039180756, + "learning_rate": 5.775070465386328e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21760 + }, + { + "epoch": 6.334012219959266, + "flow_matching_loss": 0.014486255589872599, + "grad_norm": 0.1089017242193222, + "learning_rate": 5.7620379542908765e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21770 + }, + { + "epoch": 6.33692173407041, + "flow_matching_loss": 0.01420563431456685, + "grad_norm": 0.07535963505506516, + "learning_rate": 5.749016668190395e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21780 + }, + { + "epoch": 6.339831248181554, + "flow_matching_loss": 0.014547975733876228, + "grad_norm": 0.06047477200627327, + "learning_rate": 5.736006622906973e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21790 + }, + { + "epoch": 6.3427407622926975, + "flow_matching_loss": 0.014860155060887337, + "grad_norm": 0.06856055557727814, + "learning_rate": 5.723007834249034e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21800 + }, + { + "epoch": 6.34565027640384, + "flow_matching_loss": 0.014266170468181372, + "grad_norm": 0.055201735347509384, + "learning_rate": 5.710020318011346e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21810 + }, + { + "epoch": 6.348559790514984, + "flow_matching_loss": 0.014641480334103108, + "grad_norm": 0.0675075352191925, + "learning_rate": 5.697044089974955e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21820 + }, + { + "epoch": 6.351469304626128, + "flow_matching_loss": 0.014755088929086923, + "grad_norm": 0.06041795760393143, + "learning_rate": 5.68407916590721e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21830 + }, + { + "epoch": 6.3543788187372705, + "flow_matching_loss": 0.012598422262817622, + "grad_norm": 0.06289540231227875, + "learning_rate": 5.671125561561713e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21840 + }, + { + "epoch": 6.357288332848414, + "flow_matching_loss": 0.014932323340326547, + "grad_norm": 0.06549284607172012, + "learning_rate": 5.658183292678315e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21850 + }, + { + "epoch": 6.360197846959558, + "flow_matching_loss": 0.012892310321331025, + "grad_norm": 0.05122943967580795, + "learning_rate": 5.645252374983091e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21860 + }, + { + "epoch": 6.363107361070702, + "flow_matching_loss": 0.01480625979602337, + "grad_norm": 0.060514237731695175, + "learning_rate": 5.6323328241883224e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21870 + }, + { + "epoch": 6.3660168751818444, + "flow_matching_loss": 0.014137675520032644, + "grad_norm": 0.09322214871644974, + "learning_rate": 5.61942465599249e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21880 + }, + { + "epoch": 6.368926389292988, + "flow_matching_loss": 0.014516512863337994, + "grad_norm": 0.08511187136173248, + "learning_rate": 5.60652788608023e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21890 + }, + { + "epoch": 6.371835903404132, + "flow_matching_loss": 0.01301077539101243, + "grad_norm": 0.06855368614196777, + "learning_rate": 5.593642530122336e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21900 + }, + { + "epoch": 6.374745417515275, + "flow_matching_loss": 0.015900616254657506, + "grad_norm": 0.07792884111404419, + "learning_rate": 5.58076860377573e-06, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21910 + }, + { + "epoch": 6.377654931626418, + "flow_matching_loss": 0.013457727804780006, + "grad_norm": 0.07681462168693542, + "learning_rate": 5.567906122683439e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21920 + }, + { + "epoch": 6.380564445737562, + "flow_matching_loss": 0.01522621912881732, + "grad_norm": 0.0862216055393219, + "learning_rate": 5.555055102474604e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21930 + }, + { + "epoch": 6.383473959848705, + "flow_matching_loss": 0.013562560547143221, + "grad_norm": 0.06587720662355423, + "learning_rate": 5.542215558764421e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21940 + }, + { + "epoch": 6.386383473959849, + "flow_matching_loss": 0.013954288233071566, + "grad_norm": 0.06705109030008316, + "learning_rate": 5.529387507154138e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21950 + }, + { + "epoch": 6.389292988070992, + "flow_matching_loss": 0.01495327353477478, + "grad_norm": 0.07628808170557022, + "learning_rate": 5.516570963231058e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21960 + }, + { + "epoch": 6.392202502182135, + "flow_matching_loss": 0.0136781289242208, + "grad_norm": 0.07493627071380615, + "learning_rate": 5.503765942568484e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21970 + }, + { + "epoch": 6.395112016293279, + "flow_matching_loss": 0.013529838994145393, + "grad_norm": 0.07154303789138794, + "learning_rate": 5.49097246072573e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21980 + }, + { + "epoch": 6.398021530404423, + "flow_matching_loss": 0.013687581662088633, + "grad_norm": 0.08381904661655426, + "learning_rate": 5.4781905332480805e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 21990 + }, + { + "epoch": 6.400931044515566, + "flow_matching_loss": 0.013811539392918348, + "grad_norm": 0.07027190923690796, + "learning_rate": 5.4654201756667795e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22000 + }, + { + "epoch": 6.403840558626709, + "flow_matching_loss": 0.012531904876232148, + "grad_norm": 0.07141739875078201, + "learning_rate": 5.452661403499017e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22010 + }, + { + "epoch": 6.406750072737853, + "flow_matching_loss": 0.013187241274863482, + "grad_norm": 0.08347906172275543, + "learning_rate": 5.4399142322479005e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22020 + }, + { + "epoch": 6.4096595868489965, + "flow_matching_loss": 0.013990705367177725, + "grad_norm": 0.07642899453639984, + "learning_rate": 5.427178677402452e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22030 + }, + { + "epoch": 6.412569100960139, + "flow_matching_loss": 0.013777005393058062, + "grad_norm": 0.10097537934780121, + "learning_rate": 5.41445475443757e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22040 + }, + { + "epoch": 6.415478615071283, + "flow_matching_loss": 0.013730342406779528, + "grad_norm": 0.07001534849405289, + "learning_rate": 5.4017424788140194e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22050 + }, + { + "epoch": 6.418388129182427, + "flow_matching_loss": 0.014625941682606936, + "grad_norm": 0.0626462772488594, + "learning_rate": 5.389041865978414e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22060 + }, + { + "epoch": 6.4212976432935704, + "flow_matching_loss": 0.01386255705729127, + "grad_norm": 0.0671612024307251, + "learning_rate": 5.37635293136319e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22070 + }, + { + "epoch": 6.424207157404713, + "flow_matching_loss": 0.015128411073237657, + "grad_norm": 0.11179117113351822, + "learning_rate": 5.363675690386609e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22080 + }, + { + "epoch": 6.427116671515857, + "flow_matching_loss": 0.014546551834791898, + "grad_norm": 0.06140498071908951, + "learning_rate": 5.351010158452708e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22090 + }, + { + "epoch": 6.430026185627001, + "flow_matching_loss": 0.013832816667854787, + "grad_norm": 0.08724000304937363, + "learning_rate": 5.338356350951308e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22100 + }, + { + "epoch": 6.4329356997381435, + "flow_matching_loss": 0.013059362582862378, + "grad_norm": 0.08690649271011353, + "learning_rate": 5.325714283257974e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22110 + }, + { + "epoch": 6.435845213849287, + "flow_matching_loss": 0.014837469253689051, + "grad_norm": 0.0719689205288887, + "learning_rate": 5.313083970734013e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22120 + }, + { + "epoch": 6.438754727960431, + "flow_matching_loss": 0.016650992445647717, + "grad_norm": 0.09512520581483841, + "learning_rate": 5.300465428726439e-06, + "loss": 0.0167, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22130 + }, + { + "epoch": 6.441664242071574, + "flow_matching_loss": 0.014142430014908313, + "grad_norm": 0.07049176841974258, + "learning_rate": 5.287858672567979e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22140 + }, + { + "epoch": 6.444573756182717, + "flow_matching_loss": 0.014152975939214229, + "grad_norm": 0.06172182410955429, + "learning_rate": 5.275263717577028e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22150 + }, + { + "epoch": 6.447483270293861, + "flow_matching_loss": 0.012836957722902298, + "grad_norm": 0.07724255323410034, + "learning_rate": 5.2626805790576435e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22160 + }, + { + "epoch": 6.450392784405004, + "flow_matching_loss": 0.014033023174852133, + "grad_norm": 0.0704922005534172, + "learning_rate": 5.250109272299523e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22170 + }, + { + "epoch": 6.453302298516148, + "flow_matching_loss": 0.012886614352464677, + "grad_norm": 0.08593744784593582, + "learning_rate": 5.237549812577987e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22180 + }, + { + "epoch": 6.456211812627291, + "flow_matching_loss": 0.01491421153768897, + "grad_norm": 0.08056683838367462, + "learning_rate": 5.225002215153973e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22190 + }, + { + "epoch": 6.459121326738435, + "flow_matching_loss": 0.013374029006808997, + "grad_norm": 0.06302527338266373, + "learning_rate": 5.21246649527399e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22200 + }, + { + "epoch": 6.462030840849578, + "flow_matching_loss": 0.012596352398395539, + "grad_norm": 0.06383635848760605, + "learning_rate": 5.199942668170118e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22210 + }, + { + "epoch": 6.464940354960722, + "flow_matching_loss": 0.013677599374204874, + "grad_norm": 0.06999382376670837, + "learning_rate": 5.187430749059986e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22220 + }, + { + "epoch": 6.467849869071865, + "flow_matching_loss": 0.0144259886816144, + "grad_norm": 0.07448378950357437, + "learning_rate": 5.174930753146763e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22230 + }, + { + "epoch": 6.470759383183008, + "flow_matching_loss": 0.01259779678657651, + "grad_norm": 0.05929303169250488, + "learning_rate": 5.162442695619114e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22240 + }, + { + "epoch": 6.473668897294152, + "flow_matching_loss": 0.013272299990057945, + "grad_norm": 0.09310027956962585, + "learning_rate": 5.149966591651215e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22250 + }, + { + "epoch": 6.4765784114052956, + "flow_matching_loss": 0.01456504762172699, + "grad_norm": 0.0753837525844574, + "learning_rate": 5.137502456402705e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22260 + }, + { + "epoch": 6.479487925516438, + "flow_matching_loss": 0.01469061877578497, + "grad_norm": 0.08203332871198654, + "learning_rate": 5.125050305018683e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22270 + }, + { + "epoch": 6.482397439627582, + "flow_matching_loss": 0.014532555919140577, + "grad_norm": 0.06071222946047783, + "learning_rate": 5.112610152629687e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22280 + }, + { + "epoch": 6.485306953738726, + "flow_matching_loss": 0.014859825652092696, + "grad_norm": 0.06446590274572372, + "learning_rate": 5.10018201435167e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22290 + }, + { + "epoch": 6.488216467849869, + "flow_matching_loss": 0.012366828694939613, + "grad_norm": 0.07536733895540237, + "learning_rate": 5.087765905286004e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22300 + }, + { + "epoch": 6.491125981961012, + "flow_matching_loss": 0.013367510214447976, + "grad_norm": 0.08268345892429352, + "learning_rate": 5.0753618405194245e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22310 + }, + { + "epoch": 6.494035496072156, + "flow_matching_loss": 0.013477816246449947, + "grad_norm": 0.06228843331336975, + "learning_rate": 5.062969835124042e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22320 + }, + { + "epoch": 6.4969450101833, + "flow_matching_loss": 0.013717906083911657, + "grad_norm": 0.07580100744962692, + "learning_rate": 5.05058990415731e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22330 + }, + { + "epoch": 6.4998545242944425, + "flow_matching_loss": 0.013569267932325602, + "grad_norm": 0.07993832975625992, + "learning_rate": 5.038222062662007e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22340 + }, + { + "epoch": 6.502764038405586, + "flow_matching_loss": 0.01583597157150507, + "grad_norm": 0.07751388102769852, + "learning_rate": 5.025866325666237e-06, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22350 + }, + { + "epoch": 6.50567355251673, + "flow_matching_loss": 0.01490413723513484, + "grad_norm": 0.07484705001115799, + "learning_rate": 5.013522708183383e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22360 + }, + { + "epoch": 6.508583066627873, + "flow_matching_loss": 0.0151130267418921, + "grad_norm": 0.07005965709686279, + "learning_rate": 5.001191225212096e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22370 + }, + { + "epoch": 6.5114925807390165, + "flow_matching_loss": 0.013207436725497245, + "grad_norm": 0.08515024185180664, + "learning_rate": 4.9888718917363e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22380 + }, + { + "epoch": 6.51440209485016, + "flow_matching_loss": 0.012868308648467063, + "grad_norm": 0.07462819665670395, + "learning_rate": 4.9765647227251395e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22390 + }, + { + "epoch": 6.517311608961304, + "flow_matching_loss": 0.012370591331273318, + "grad_norm": 0.058447469025850296, + "learning_rate": 4.964269733132992e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22400 + }, + { + "epoch": 6.520221123072447, + "flow_matching_loss": 0.01427555326372385, + "grad_norm": 0.06164787337183952, + "learning_rate": 4.951986937899428e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22410 + }, + { + "epoch": 6.52313063718359, + "flow_matching_loss": 0.01360704256221652, + "grad_norm": 0.08284138143062592, + "learning_rate": 4.9397163519492e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22420 + }, + { + "epoch": 6.526040151294734, + "flow_matching_loss": 0.013441438600420953, + "grad_norm": 0.06868024170398712, + "learning_rate": 4.9274579901922275e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22430 + }, + { + "epoch": 6.528949665405877, + "flow_matching_loss": 0.012208202574402093, + "grad_norm": 0.07116495072841644, + "learning_rate": 4.9152118675235705e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22440 + }, + { + "epoch": 6.531859179517021, + "flow_matching_loss": 0.01405786583200097, + "grad_norm": 0.08975136280059814, + "learning_rate": 4.902977998823432e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22450 + }, + { + "epoch": 6.534768693628164, + "flow_matching_loss": 0.014195761177688837, + "grad_norm": 0.06086152046918869, + "learning_rate": 4.890756398957114e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22460 + }, + { + "epoch": 6.537678207739307, + "flow_matching_loss": 0.013690461497753858, + "grad_norm": 0.06337469816207886, + "learning_rate": 4.8785470827750075e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22470 + }, + { + "epoch": 6.540587721850451, + "flow_matching_loss": 0.013786675222218037, + "grad_norm": 0.07446262240409851, + "learning_rate": 4.866350065112588e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22480 + }, + { + "epoch": 6.543497235961595, + "flow_matching_loss": 0.011928433179855346, + "grad_norm": 0.08531173318624496, + "learning_rate": 4.854165360790381e-06, + "loss": 0.0119, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22490 + }, + { + "epoch": 6.546406750072737, + "flow_matching_loss": 0.012541463971138, + "grad_norm": 0.08153589814901352, + "learning_rate": 4.8419929846139425e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22500 + }, + { + "epoch": 6.549316264183881, + "flow_matching_loss": 0.015013776626437903, + "grad_norm": 0.08668842911720276, + "learning_rate": 4.829832951373872e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22510 + }, + { + "epoch": 6.552225778295025, + "flow_matching_loss": 0.013323185965418816, + "grad_norm": 0.07276536524295807, + "learning_rate": 4.817685275845746e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22520 + }, + { + "epoch": 6.5551352924061685, + "flow_matching_loss": 0.014654535986483096, + "grad_norm": 0.07799229025840759, + "learning_rate": 4.805549972790143e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22530 + }, + { + "epoch": 6.558044806517311, + "flow_matching_loss": 0.013497538585215806, + "grad_norm": 0.07011746615171432, + "learning_rate": 4.7934270569525966e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22540 + }, + { + "epoch": 6.560954320628455, + "flow_matching_loss": 0.01307247458025813, + "grad_norm": 0.08940732479095459, + "learning_rate": 4.7813165430635874e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22550 + }, + { + "epoch": 6.563863834739599, + "flow_matching_loss": 0.01420173216611147, + "grad_norm": 0.0815533772110939, + "learning_rate": 4.769218445838544e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22560 + }, + { + "epoch": 6.566773348850742, + "flow_matching_loss": 0.013693376537412405, + "grad_norm": 0.07812409847974777, + "learning_rate": 4.757132779977787e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22570 + }, + { + "epoch": 6.569682862961885, + "flow_matching_loss": 0.012948618363589048, + "grad_norm": 0.05971544608473778, + "learning_rate": 4.745059560166542e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22580 + }, + { + "epoch": 6.572592377073029, + "flow_matching_loss": 0.012966985441744327, + "grad_norm": 0.06985223293304443, + "learning_rate": 4.732998801074911e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22590 + }, + { + "epoch": 6.575501891184173, + "flow_matching_loss": 0.01311846598982811, + "grad_norm": 0.0684056505560875, + "learning_rate": 4.7209505173578445e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22600 + }, + { + "epoch": 6.5784114052953155, + "flow_matching_loss": 0.015038743801414966, + "grad_norm": 0.10768185555934906, + "learning_rate": 4.7089147236551555e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22610 + }, + { + "epoch": 6.581320919406459, + "flow_matching_loss": 0.012891565170139074, + "grad_norm": 0.06173758581280708, + "learning_rate": 4.696891434591464e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22620 + }, + { + "epoch": 6.584230433517602, + "flow_matching_loss": 0.014844093658030032, + "grad_norm": 0.08286523073911667, + "learning_rate": 4.684880664776199e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22630 + }, + { + "epoch": 6.587139947628746, + "flow_matching_loss": 0.01472172662615776, + "grad_norm": 0.07527424395084381, + "learning_rate": 4.6728824288035784e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22640 + }, + { + "epoch": 6.5900494617398895, + "flow_matching_loss": 0.014453207328915595, + "grad_norm": 0.08619502931833267, + "learning_rate": 4.660896741252586e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22650 + }, + { + "epoch": 6.592958975851033, + "flow_matching_loss": 0.014401643816381694, + "grad_norm": 0.08046220988035202, + "learning_rate": 4.6489236166869714e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22660 + }, + { + "epoch": 6.595868489962176, + "flow_matching_loss": 0.011916884034872056, + "grad_norm": 0.07507841289043427, + "learning_rate": 4.6369630696552e-06, + "loss": 0.0119, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22670 + }, + { + "epoch": 6.59877800407332, + "flow_matching_loss": 0.013732102420181035, + "grad_norm": 0.08175269514322281, + "learning_rate": 4.625015114690474e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22680 + }, + { + "epoch": 6.601687518184463, + "flow_matching_loss": 0.012743378337472678, + "grad_norm": 0.0696427971124649, + "learning_rate": 4.613079766310682e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22690 + }, + { + "epoch": 6.604597032295606, + "flow_matching_loss": 0.013077616039663554, + "grad_norm": 0.0673983246088028, + "learning_rate": 4.601157039018396e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22700 + }, + { + "epoch": 6.60750654640675, + "flow_matching_loss": 0.012176554463803769, + "grad_norm": 0.07625260949134827, + "learning_rate": 4.589246947300849e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22710 + }, + { + "epoch": 6.610416060517894, + "flow_matching_loss": 0.012914742901921272, + "grad_norm": 0.06471428275108337, + "learning_rate": 4.577349505629936e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22720 + }, + { + "epoch": 6.613325574629037, + "flow_matching_loss": 0.014074949733912944, + "grad_norm": 0.08280757069587708, + "learning_rate": 4.565464728462165e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22730 + }, + { + "epoch": 6.61623508874018, + "flow_matching_loss": 0.014735533762723208, + "grad_norm": 0.06245066598057747, + "learning_rate": 4.553592630238663e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22740 + }, + { + "epoch": 6.619144602851324, + "flow_matching_loss": 0.013205146044492721, + "grad_norm": 0.08905640989542007, + "learning_rate": 4.541733225385146e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22750 + }, + { + "epoch": 6.622054116962468, + "flow_matching_loss": 0.012781511899083853, + "grad_norm": 0.09496120363473892, + "learning_rate": 4.529886528311905e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22760 + }, + { + "epoch": 6.62496363107361, + "flow_matching_loss": 0.014091704785823823, + "grad_norm": 0.08600950986146927, + "learning_rate": 4.5180525534138045e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22770 + }, + { + "epoch": 6.627873145184754, + "flow_matching_loss": 0.014796038810163736, + "grad_norm": 0.07484681904315948, + "learning_rate": 4.506231315070235e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22780 + }, + { + "epoch": 6.630782659295898, + "flow_matching_loss": 0.013411798793822528, + "grad_norm": 0.07974006235599518, + "learning_rate": 4.494422827645115e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22790 + }, + { + "epoch": 6.633692173407041, + "flow_matching_loss": 0.013176506850868464, + "grad_norm": 0.07989339530467987, + "learning_rate": 4.482627105486865e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22800 + }, + { + "epoch": 6.636601687518184, + "flow_matching_loss": 0.012595283333212137, + "grad_norm": 0.05626977980136871, + "learning_rate": 4.470844162928412e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22810 + }, + { + "epoch": 6.639511201629328, + "flow_matching_loss": 0.013309414312243462, + "grad_norm": 0.07105117291212082, + "learning_rate": 4.459074014287129e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22820 + }, + { + "epoch": 6.642420715740471, + "flow_matching_loss": 0.014154531806707383, + "grad_norm": 0.06379906088113785, + "learning_rate": 4.447316673864869e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22830 + }, + { + "epoch": 6.645330229851615, + "flow_matching_loss": 0.0131059218198061, + "grad_norm": 0.06536258012056351, + "learning_rate": 4.435572155947903e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22840 + }, + { + "epoch": 6.648239743962758, + "flow_matching_loss": 0.013507421873509885, + "grad_norm": 0.08035716414451599, + "learning_rate": 4.423840474806929e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22850 + }, + { + "epoch": 6.651149258073902, + "flow_matching_loss": 0.01399205606430769, + "grad_norm": 0.06556934118270874, + "learning_rate": 4.412121644697043e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22860 + }, + { + "epoch": 6.654058772185045, + "flow_matching_loss": 0.012543855234980583, + "grad_norm": 0.07369931042194366, + "learning_rate": 4.400415679857727e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22870 + }, + { + "epoch": 6.6569682862961885, + "flow_matching_loss": 0.014243584405630828, + "grad_norm": 0.10423941165208817, + "learning_rate": 4.388722594512838e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22880 + }, + { + "epoch": 6.659877800407332, + "flow_matching_loss": 0.013924232963472605, + "grad_norm": 0.07756797224283218, + "learning_rate": 4.377042402870577e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22890 + }, + { + "epoch": 6.662787314518475, + "flow_matching_loss": 0.01237685913220048, + "grad_norm": 0.08000265061855316, + "learning_rate": 4.365375119123476e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22900 + }, + { + "epoch": 6.665696828629619, + "flow_matching_loss": 0.012626700103282928, + "grad_norm": 0.07317151874303818, + "learning_rate": 4.353720757448383e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22910 + }, + { + "epoch": 6.6686063427407625, + "flow_matching_loss": 0.01284522544592619, + "grad_norm": 0.07941973209381104, + "learning_rate": 4.342079332006445e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22920 + }, + { + "epoch": 6.671515856851906, + "flow_matching_loss": 0.013449256215244532, + "grad_norm": 0.07649289816617966, + "learning_rate": 4.330450856943101e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22930 + }, + { + "epoch": 6.674425370963049, + "flow_matching_loss": 0.013279706891626119, + "grad_norm": 0.07581792771816254, + "learning_rate": 4.318835346388036e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22940 + }, + { + "epoch": 6.677334885074193, + "flow_matching_loss": 0.012904252484440803, + "grad_norm": 0.07037778943777084, + "learning_rate": 4.307232814455189e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22950 + }, + { + "epoch": 6.680244399185336, + "flow_matching_loss": 0.013504314236342907, + "grad_norm": 0.07645539194345474, + "learning_rate": 4.29564327524274e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22960 + }, + { + "epoch": 6.683153913296479, + "flow_matching_loss": 0.01353674503043294, + "grad_norm": 0.06743789464235306, + "learning_rate": 4.284066742833061e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22970 + }, + { + "epoch": 6.686063427407623, + "flow_matching_loss": 0.01336314743384719, + "grad_norm": 0.09549041837453842, + "learning_rate": 4.272503231292741e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22980 + }, + { + "epoch": 6.688972941518767, + "flow_matching_loss": 0.01471501039341092, + "grad_norm": 0.09776213765144348, + "learning_rate": 4.260952754672532e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 22990 + }, + { + "epoch": 6.6918824556299095, + "flow_matching_loss": 0.015146843623369932, + "grad_norm": 0.08690391480922699, + "learning_rate": 4.249415327007353e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23000 + }, + { + "epoch": 6.694791969741053, + "flow_matching_loss": 0.013952477555721998, + "grad_norm": 0.09394558519124985, + "learning_rate": 4.237890962316264e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23010 + }, + { + "epoch": 6.697701483852197, + "flow_matching_loss": 0.01340078953653574, + "grad_norm": 0.07973601669073105, + "learning_rate": 4.226379674602452e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23020 + }, + { + "epoch": 6.70061099796334, + "flow_matching_loss": 0.013898663362488151, + "grad_norm": 0.08852451294660568, + "learning_rate": 4.214881477853227e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23030 + }, + { + "epoch": 6.703520512074483, + "flow_matching_loss": 0.013894646242260934, + "grad_norm": 0.08911297470331192, + "learning_rate": 4.2033963860399745e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23040 + }, + { + "epoch": 6.706430026185627, + "flow_matching_loss": 0.012299288157373666, + "grad_norm": 0.06216337904334068, + "learning_rate": 4.1919244131181696e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23050 + }, + { + "epoch": 6.709339540296771, + "flow_matching_loss": 0.013743846490979195, + "grad_norm": 0.07950935512781143, + "learning_rate": 4.180465573027335e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23060 + }, + { + "epoch": 6.712249054407914, + "flow_matching_loss": 0.012286746874451637, + "grad_norm": 0.08001112192869186, + "learning_rate": 4.169019879691046e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23070 + }, + { + "epoch": 6.715158568519057, + "flow_matching_loss": 0.013543603010475636, + "grad_norm": 0.08397569507360458, + "learning_rate": 4.157587347016891e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23080 + }, + { + "epoch": 6.718068082630201, + "flow_matching_loss": 0.013449937105178833, + "grad_norm": 0.07333645224571228, + "learning_rate": 4.14616798889649e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23090 + }, + { + "epoch": 6.720977596741344, + "flow_matching_loss": 0.012109600380063057, + "grad_norm": 0.06658409535884857, + "learning_rate": 4.134761819205425e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23100 + }, + { + "epoch": 6.723887110852488, + "flow_matching_loss": 0.012983416765928268, + "grad_norm": 0.0855039432644844, + "learning_rate": 4.1233688518032794e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23110 + }, + { + "epoch": 6.726796624963631, + "flow_matching_loss": 0.012615562602877616, + "grad_norm": 0.06691299378871918, + "learning_rate": 4.1119891005335795e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23120 + }, + { + "epoch": 6.729706139074775, + "flow_matching_loss": 0.012416750378906726, + "grad_norm": 0.06726830452680588, + "learning_rate": 4.1006225792237895e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23130 + }, + { + "epoch": 6.732615653185918, + "flow_matching_loss": 0.012897020671516657, + "grad_norm": 0.059875376522541046, + "learning_rate": 4.089269301685314e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23140 + }, + { + "epoch": 6.7355251672970615, + "flow_matching_loss": 0.012289696279913187, + "grad_norm": 0.07970064133405685, + "learning_rate": 4.077929281713453e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23150 + }, + { + "epoch": 6.738434681408204, + "flow_matching_loss": 0.013613579235970974, + "grad_norm": 0.07670076191425323, + "learning_rate": 4.066602533087398e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23160 + }, + { + "epoch": 6.741344195519348, + "flow_matching_loss": 0.013356367778033017, + "grad_norm": 0.07442005723714828, + "learning_rate": 4.0552890695702175e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23170 + }, + { + "epoch": 6.744253709630492, + "flow_matching_loss": 0.013509290851652622, + "grad_norm": 0.08186016976833344, + "learning_rate": 4.04398890490883e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23180 + }, + { + "epoch": 6.7471632237416355, + "flow_matching_loss": 0.013760196417570114, + "grad_norm": 0.0810793787240982, + "learning_rate": 4.032702052834013e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23190 + }, + { + "epoch": 6.750072737852778, + "flow_matching_loss": 0.013471677619963884, + "grad_norm": 0.0609215646982193, + "learning_rate": 4.021428527060352e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23200 + }, + { + "epoch": 6.752982251963922, + "flow_matching_loss": 0.014117403328418732, + "grad_norm": 0.05150905251502991, + "learning_rate": 4.010168341286241e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23210 + }, + { + "epoch": 6.755891766075066, + "flow_matching_loss": 0.013881998788565397, + "grad_norm": 0.0733441561460495, + "learning_rate": 3.998921509193868e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23220 + }, + { + "epoch": 6.7588012801862085, + "flow_matching_loss": 0.013606242556124925, + "grad_norm": 0.07021297514438629, + "learning_rate": 3.987688044449193e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23230 + }, + { + "epoch": 6.761710794297352, + "flow_matching_loss": 0.015464908350259066, + "grad_norm": 0.08015312999486923, + "learning_rate": 3.9764679607019385e-06, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23240 + }, + { + "epoch": 6.764620308408496, + "flow_matching_loss": 0.013290308229625226, + "grad_norm": 0.06545493006706238, + "learning_rate": 3.965261271585561e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23250 + }, + { + "epoch": 6.76752982251964, + "flow_matching_loss": 0.01315771946683526, + "grad_norm": 0.07884573936462402, + "learning_rate": 3.954067990717251e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23260 + }, + { + "epoch": 6.7704393366307825, + "flow_matching_loss": 0.012665298208594323, + "grad_norm": 0.07523021101951599, + "learning_rate": 3.9428881316978935e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23270 + }, + { + "epoch": 6.773348850741926, + "flow_matching_loss": 0.014885528199374675, + "grad_norm": 0.06750430911779404, + "learning_rate": 3.931721708112075e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23280 + }, + { + "epoch": 6.77625836485307, + "flow_matching_loss": 0.012829640228301287, + "grad_norm": 0.06059056520462036, + "learning_rate": 3.920568733528047e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23290 + }, + { + "epoch": 6.779167878964213, + "flow_matching_loss": 0.01338359247893095, + "grad_norm": 0.06192099303007126, + "learning_rate": 3.909429221497734e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23300 + }, + { + "epoch": 6.782077393075356, + "flow_matching_loss": 0.013250590302050114, + "grad_norm": 0.08348383754491806, + "learning_rate": 3.898303185556689e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23310 + }, + { + "epoch": 6.7849869071865, + "flow_matching_loss": 0.014120639860630035, + "grad_norm": 0.09187139570713043, + "learning_rate": 3.887190639224097e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23320 + }, + { + "epoch": 6.787896421297643, + "flow_matching_loss": 0.012335626874119043, + "grad_norm": 0.06453113257884979, + "learning_rate": 3.876091596002746e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23330 + }, + { + "epoch": 6.790805935408787, + "flow_matching_loss": 0.014429711829870939, + "grad_norm": 0.08097228407859802, + "learning_rate": 3.865006069379019e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23340 + }, + { + "epoch": 6.79371544951993, + "flow_matching_loss": 0.012658544071018695, + "grad_norm": 0.0639374852180481, + "learning_rate": 3.853934072822881e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23350 + }, + { + "epoch": 6.796624963631073, + "flow_matching_loss": 0.01323060430586338, + "grad_norm": 0.1263953149318695, + "learning_rate": 3.8428756197878514e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23360 + }, + { + "epoch": 6.799534477742217, + "flow_matching_loss": 0.01429108576849103, + "grad_norm": 0.08962102979421616, + "learning_rate": 3.831830723710992e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23370 + }, + { + "epoch": 6.802443991853361, + "flow_matching_loss": 0.012685498036444188, + "grad_norm": 0.07846025377511978, + "learning_rate": 3.820799398012889e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23380 + }, + { + "epoch": 6.805353505964504, + "flow_matching_loss": 0.014020934840664268, + "grad_norm": 0.11279319226741791, + "learning_rate": 3.80978165609765e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23390 + }, + { + "epoch": 6.808263020075647, + "flow_matching_loss": 0.014618769753724337, + "grad_norm": 0.11338276416063309, + "learning_rate": 3.798777511352864e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23400 + }, + { + "epoch": 6.811172534186791, + "flow_matching_loss": 0.013870649505406618, + "grad_norm": 0.08309751749038696, + "learning_rate": 3.787786977149613e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23410 + }, + { + "epoch": 6.8140820482979345, + "flow_matching_loss": 0.0134478023275733, + "grad_norm": 0.06983308494091034, + "learning_rate": 3.776810066842427e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23420 + }, + { + "epoch": 6.816991562409077, + "flow_matching_loss": 0.013246002979576588, + "grad_norm": 0.06847703456878662, + "learning_rate": 3.7658467937692866e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23430 + }, + { + "epoch": 6.819901076520221, + "flow_matching_loss": 0.012718132603913546, + "grad_norm": 0.06628246605396271, + "learning_rate": 3.7548971712516e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23440 + }, + { + "epoch": 6.822810590631365, + "flow_matching_loss": 0.013757247850298882, + "grad_norm": 0.09513478726148605, + "learning_rate": 3.743961212594197e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23450 + }, + { + "epoch": 6.8257201047425085, + "flow_matching_loss": 0.013284018728882074, + "grad_norm": 0.05832594260573387, + "learning_rate": 3.7330389310852975e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23460 + }, + { + "epoch": 6.828629618853651, + "flow_matching_loss": 0.015290326718240977, + "grad_norm": 0.06800846755504608, + "learning_rate": 3.7221303399965008e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23470 + }, + { + "epoch": 6.831539132964795, + "flow_matching_loss": 0.012904610112309456, + "grad_norm": 0.07147835940122604, + "learning_rate": 3.711235452582775e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23480 + }, + { + "epoch": 6.834448647075939, + "flow_matching_loss": 0.012940154224634171, + "grad_norm": 0.06585895270109177, + "learning_rate": 3.7003542820824347e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23490 + }, + { + "epoch": 6.8373581611870815, + "flow_matching_loss": 0.01461790706962347, + "grad_norm": 0.07931136339902878, + "learning_rate": 3.6894868417171252e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23500 + }, + { + "epoch": 6.840267675298225, + "flow_matching_loss": 0.013374823983758688, + "grad_norm": 0.07067818194627762, + "learning_rate": 3.678633144691818e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23510 + }, + { + "epoch": 6.843177189409369, + "flow_matching_loss": 0.013430937938392163, + "grad_norm": 0.06365042924880981, + "learning_rate": 3.667793204194776e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23520 + }, + { + "epoch": 6.846086703520512, + "flow_matching_loss": 0.011083919182419777, + "grad_norm": 0.05990695580840111, + "learning_rate": 3.656967033397544e-06, + "loss": 0.0111, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23530 + }, + { + "epoch": 6.8489962176316554, + "flow_matching_loss": 0.014533934090286494, + "grad_norm": 0.11526501178741455, + "learning_rate": 3.646154645454948e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23540 + }, + { + "epoch": 6.851905731742799, + "flow_matching_loss": 0.012844994105398655, + "grad_norm": 0.057291872799396515, + "learning_rate": 3.6353560535050524e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23550 + }, + { + "epoch": 6.854815245853942, + "flow_matching_loss": 0.012489539757370948, + "grad_norm": 0.06423024088144302, + "learning_rate": 3.624571270669172e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23560 + }, + { + "epoch": 6.857724759965086, + "flow_matching_loss": 0.013338980078697205, + "grad_norm": 0.07439717650413513, + "learning_rate": 3.6138003100518296e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23570 + }, + { + "epoch": 6.860634274076229, + "flow_matching_loss": 0.013153657782822847, + "grad_norm": 0.05874113366007805, + "learning_rate": 3.6030431847407595e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23580 + }, + { + "epoch": 6.863543788187373, + "flow_matching_loss": 0.01458176476880908, + "grad_norm": 0.10942620784044266, + "learning_rate": 3.592299907806881e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23590 + }, + { + "epoch": 6.866453302298516, + "flow_matching_loss": 0.012617684807628394, + "grad_norm": 0.07176242768764496, + "learning_rate": 3.5815704923042892e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23600 + }, + { + "epoch": 6.86936281640966, + "flow_matching_loss": 0.013641841616481543, + "grad_norm": 0.07346128672361374, + "learning_rate": 3.5708549512702392e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23610 + }, + { + "epoch": 6.872272330520803, + "flow_matching_loss": 0.014166008308529853, + "grad_norm": 0.07953372597694397, + "learning_rate": 3.5601532977251217e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23620 + }, + { + "epoch": 6.875181844631946, + "flow_matching_loss": 0.012752580270171166, + "grad_norm": 0.07913528382778168, + "learning_rate": 3.5494655446724544e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23630 + }, + { + "epoch": 6.87809135874309, + "flow_matching_loss": 0.013845092710107565, + "grad_norm": 0.06812689453363419, + "learning_rate": 3.538791705098866e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23640 + }, + { + "epoch": 6.881000872854234, + "flow_matching_loss": 0.013927313033491373, + "grad_norm": 0.09185840934515, + "learning_rate": 3.5281317919740718e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23650 + }, + { + "epoch": 6.883910386965377, + "flow_matching_loss": 0.011715890932828188, + "grad_norm": 0.07415413111448288, + "learning_rate": 3.5174858182508825e-06, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23660 + }, + { + "epoch": 6.88681990107652, + "flow_matching_loss": 0.014668140932917595, + "grad_norm": 0.07439026981592178, + "learning_rate": 3.506853796865153e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23670 + }, + { + "epoch": 6.889729415187664, + "flow_matching_loss": 0.01300558764487505, + "grad_norm": 0.08233650773763657, + "learning_rate": 3.496235740735789e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23680 + }, + { + "epoch": 6.892638929298807, + "flow_matching_loss": 0.013608956709504128, + "grad_norm": 0.08220668882131577, + "learning_rate": 3.485631662764738e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23690 + }, + { + "epoch": 6.89554844340995, + "flow_matching_loss": 0.01377074709162116, + "grad_norm": 0.07619868218898773, + "learning_rate": 3.475041575836952e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23700 + }, + { + "epoch": 6.898457957521094, + "flow_matching_loss": 0.01305314158089459, + "grad_norm": 0.06711938232183456, + "learning_rate": 3.4644654928203796e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23710 + }, + { + "epoch": 6.901367471632238, + "flow_matching_loss": 0.012556754890829325, + "grad_norm": 0.05857870727777481, + "learning_rate": 3.4539034265659663e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23720 + }, + { + "epoch": 6.904276985743381, + "flow_matching_loss": 0.013217241782695056, + "grad_norm": 0.07483728229999542, + "learning_rate": 3.4433553899076163e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23730 + }, + { + "epoch": 6.907186499854524, + "flow_matching_loss": 0.013631981518119573, + "grad_norm": 0.06099158152937889, + "learning_rate": 3.432821395662189e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23740 + }, + { + "epoch": 6.910096013965668, + "flow_matching_loss": 0.013979929126799106, + "grad_norm": 0.07492247223854065, + "learning_rate": 3.4223014566294785e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23750 + }, + { + "epoch": 6.913005528076811, + "flow_matching_loss": 0.014937316719442605, + "grad_norm": 0.07960514724254608, + "learning_rate": 3.4117955855922e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23760 + }, + { + "epoch": 6.9159150421879545, + "flow_matching_loss": 0.013455716334283352, + "grad_norm": 0.05174409970641136, + "learning_rate": 3.4013037953159855e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23770 + }, + { + "epoch": 6.918824556299098, + "flow_matching_loss": 0.014701621700078249, + "grad_norm": 0.08209408074617386, + "learning_rate": 3.390826098549345e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23780 + }, + { + "epoch": 6.921734070410242, + "flow_matching_loss": 0.012206410523504019, + "grad_norm": 0.05375075340270996, + "learning_rate": 3.3803625080236684e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23790 + }, + { + "epoch": 6.924643584521385, + "flow_matching_loss": 0.015847679041326047, + "grad_norm": 0.10203468799591064, + "learning_rate": 3.369913036453205e-06, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23800 + }, + { + "epoch": 6.927553098632528, + "flow_matching_loss": 0.013374628871679306, + "grad_norm": 0.0650608241558075, + "learning_rate": 3.3594776965350434e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23810 + }, + { + "epoch": 6.930462612743672, + "flow_matching_loss": 0.014318596199154853, + "grad_norm": 0.05706459656357765, + "learning_rate": 3.34905650094911e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23820 + }, + { + "epoch": 6.933372126854815, + "flow_matching_loss": 0.012629026826471091, + "grad_norm": 0.07078385353088379, + "learning_rate": 3.338649462358145e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23830 + }, + { + "epoch": 6.936281640965959, + "flow_matching_loss": 0.01313364002853632, + "grad_norm": 0.06480158120393753, + "learning_rate": 3.3282565934076765e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23840 + }, + { + "epoch": 6.939191155077102, + "flow_matching_loss": 0.013747723400592804, + "grad_norm": 0.0762142762541771, + "learning_rate": 3.31787790672602e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23850 + }, + { + "epoch": 6.942100669188245, + "flow_matching_loss": 0.012451488059014082, + "grad_norm": 0.0718955472111702, + "learning_rate": 3.307513414924255e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23860 + }, + { + "epoch": 6.945010183299389, + "flow_matching_loss": 0.014508639182895421, + "grad_norm": 0.06115486100316048, + "learning_rate": 3.2971631305962246e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23870 + }, + { + "epoch": 6.947919697410533, + "flow_matching_loss": 0.012881204020231962, + "grad_norm": 0.057725511491298676, + "learning_rate": 3.2868270663184964e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23880 + }, + { + "epoch": 6.950829211521675, + "flow_matching_loss": 0.014414492528885603, + "grad_norm": 0.08524574339389801, + "learning_rate": 3.276505234650362e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23890 + }, + { + "epoch": 6.953738725632819, + "flow_matching_loss": 0.013726853672415018, + "grad_norm": 0.07049224525690079, + "learning_rate": 3.266197648133821e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23900 + }, + { + "epoch": 6.956648239743963, + "flow_matching_loss": 0.013927618600428104, + "grad_norm": 0.10191530734300613, + "learning_rate": 3.255904319293562e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23910 + }, + { + "epoch": 6.9595577538551066, + "flow_matching_loss": 0.012896528001874685, + "grad_norm": 0.07032246142625809, + "learning_rate": 3.2456252606369485e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23920 + }, + { + "epoch": 6.962467267966249, + "flow_matching_loss": 0.012394889444112777, + "grad_norm": 0.06066609174013138, + "learning_rate": 3.235360484654011e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23930 + }, + { + "epoch": 6.965376782077393, + "flow_matching_loss": 0.013882222305983304, + "grad_norm": 0.06754445284605026, + "learning_rate": 3.225110003817418e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23940 + }, + { + "epoch": 6.968286296188537, + "flow_matching_loss": 0.014418349042534829, + "grad_norm": 0.12997865676879883, + "learning_rate": 3.214873830582467e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23950 + }, + { + "epoch": 6.97119581029968, + "flow_matching_loss": 0.01351471794769168, + "grad_norm": 0.06625460833311081, + "learning_rate": 3.2046519773870803e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23960 + }, + { + "epoch": 6.974105324410823, + "flow_matching_loss": 0.014106313791126012, + "grad_norm": 0.0812268927693367, + "learning_rate": 3.1944444566517654e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23970 + }, + { + "epoch": 6.977014838521967, + "flow_matching_loss": 0.013983903639018535, + "grad_norm": 0.0662759467959404, + "learning_rate": 3.18425128077963e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23980 + }, + { + "epoch": 6.979924352633111, + "flow_matching_loss": 0.013791670929640532, + "grad_norm": 0.08961210399866104, + "learning_rate": 3.174072462156342e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 23990 + }, + { + "epoch": 6.9828338667442535, + "flow_matching_loss": 0.011874229181557894, + "grad_norm": 0.06505373120307922, + "learning_rate": 3.163908013150124e-06, + "loss": 0.0119, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24000 + }, + { + "epoch": 6.985743380855397, + "flow_matching_loss": 0.012723711505532264, + "grad_norm": 0.07721253484487534, + "learning_rate": 3.153757946111741e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24010 + }, + { + "epoch": 6.988652894966541, + "flow_matching_loss": 0.01268945150077343, + "grad_norm": 0.08874359726905823, + "learning_rate": 3.1436222733744776e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24020 + }, + { + "epoch": 6.991562409077684, + "flow_matching_loss": 0.012335187941789626, + "grad_norm": 0.07053457200527191, + "learning_rate": 3.133501007254138e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24030 + }, + { + "epoch": 6.9944719231888275, + "flow_matching_loss": 0.014111262746155262, + "grad_norm": 0.08076953142881393, + "learning_rate": 3.1233941600490123e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24040 + }, + { + "epoch": 6.997381437299971, + "flow_matching_loss": 0.013674123957753182, + "grad_norm": 0.08810172230005264, + "learning_rate": 3.113301744039875e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24050 + }, + { + "epoch": 7.000290951411114, + "flow_matching_loss": 0.014262439962476493, + "grad_norm": 0.06349223852157593, + "learning_rate": 3.103223771489959e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24060 + }, + { + "epoch": 7.003200465522258, + "flow_matching_loss": 0.012448655720800162, + "grad_norm": 0.0831340104341507, + "learning_rate": 3.09316025464495e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24070 + }, + { + "epoch": 7.006109979633401, + "flow_matching_loss": 0.01397148184478283, + "grad_norm": 0.08581917732954025, + "learning_rate": 3.0831112057329775e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24080 + }, + { + "epoch": 7.009019493744544, + "flow_matching_loss": 0.013586502335965634, + "grad_norm": 0.0799858421087265, + "learning_rate": 3.073076636964579e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24090 + }, + { + "epoch": 7.011929007855688, + "flow_matching_loss": 0.012459897715598345, + "grad_norm": 0.05207386240363121, + "learning_rate": 3.063056560532698e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24100 + }, + { + "epoch": 7.014838521966832, + "flow_matching_loss": 0.012314534466713666, + "grad_norm": 0.07984931021928787, + "learning_rate": 3.0530509886126807e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24110 + }, + { + "epoch": 7.017748036077975, + "flow_matching_loss": 0.012506315391510725, + "grad_norm": 0.08474420756101608, + "learning_rate": 3.043059933362236e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24120 + }, + { + "epoch": 7.020657550189118, + "flow_matching_loss": 0.01429559914395213, + "grad_norm": 0.09474306553602219, + "learning_rate": 3.0330834069214337e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24130 + }, + { + "epoch": 7.023567064300262, + "flow_matching_loss": 0.013708836026489734, + "grad_norm": 0.09282676130533218, + "learning_rate": 3.023121421412703e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24140 + }, + { + "epoch": 7.026476578411406, + "flow_matching_loss": 0.013980389665812254, + "grad_norm": 0.07729857414960861, + "learning_rate": 3.0131739889407896e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24150 + }, + { + "epoch": 7.029386092522548, + "flow_matching_loss": 0.013201931025832891, + "grad_norm": 0.06794945150613785, + "learning_rate": 3.0032411215927645e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24160 + }, + { + "epoch": 7.032295606633692, + "flow_matching_loss": 0.013416908495128155, + "grad_norm": 0.06693290919065475, + "learning_rate": 2.9933228314379975e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24170 + }, + { + "epoch": 7.035205120744836, + "flow_matching_loss": 0.012548547331243753, + "grad_norm": 0.07743176072835922, + "learning_rate": 2.9834191305281405e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24180 + }, + { + "epoch": 7.038114634855979, + "flow_matching_loss": 0.013279276713728905, + "grad_norm": 0.0734158530831337, + "learning_rate": 2.973530030897135e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24190 + }, + { + "epoch": 7.041024148967122, + "flow_matching_loss": 0.012152835633605719, + "grad_norm": 0.07103466987609863, + "learning_rate": 2.9636555445611646e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24200 + }, + { + "epoch": 7.043933663078266, + "flow_matching_loss": 0.013507293257862329, + "grad_norm": 0.08550570905208588, + "learning_rate": 2.953795683518659e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24210 + }, + { + "epoch": 7.04684317718941, + "flow_matching_loss": 0.01335873119533062, + "grad_norm": 0.08221346884965897, + "learning_rate": 2.9439504597502815e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24220 + }, + { + "epoch": 7.049752691300553, + "flow_matching_loss": 0.014059164468199015, + "grad_norm": 0.07166575640439987, + "learning_rate": 2.934119885218902e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24230 + }, + { + "epoch": 7.052662205411696, + "flow_matching_loss": 0.012004397623240948, + "grad_norm": 0.07915980368852615, + "learning_rate": 2.9243039718696045e-06, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24240 + }, + { + "epoch": 7.05557171952284, + "flow_matching_loss": 0.01213940242305398, + "grad_norm": 0.06586863100528717, + "learning_rate": 2.914502731629641e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24250 + }, + { + "epoch": 7.058481233633983, + "flow_matching_loss": 0.014714818727225066, + "grad_norm": 0.07660364359617233, + "learning_rate": 2.904716176408451e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24260 + }, + { + "epoch": 7.0613907477451265, + "flow_matching_loss": 0.013903143629431725, + "grad_norm": 0.08368014544248581, + "learning_rate": 2.8949443180976172e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24270 + }, + { + "epoch": 7.06430026185627, + "flow_matching_loss": 0.013642107602208853, + "grad_norm": 0.08161546289920807, + "learning_rate": 2.885187168570864e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24280 + }, + { + "epoch": 7.067209775967413, + "flow_matching_loss": 0.01475173719227314, + "grad_norm": 0.07863420993089676, + "learning_rate": 2.875444739684058e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24290 + }, + { + "epoch": 7.070119290078557, + "flow_matching_loss": 0.012611891701817512, + "grad_norm": 0.06599205732345581, + "learning_rate": 2.865717043275165e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24300 + }, + { + "epoch": 7.0730288041897005, + "flow_matching_loss": 0.014573428500443696, + "grad_norm": 0.06109977513551712, + "learning_rate": 2.8560040911642528e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24310 + }, + { + "epoch": 7.075938318300844, + "flow_matching_loss": 0.014236172195523977, + "grad_norm": 0.06528174132108688, + "learning_rate": 2.846305895153472e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24320 + }, + { + "epoch": 7.078847832411987, + "flow_matching_loss": 0.014609815925359726, + "grad_norm": 0.07856868952512741, + "learning_rate": 2.8366224670270487e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24330 + }, + { + "epoch": 7.081757346523131, + "flow_matching_loss": 0.011734249629080295, + "grad_norm": 0.0545320063829422, + "learning_rate": 2.8269538185512535e-06, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24340 + }, + { + "epoch": 7.084666860634274, + "flow_matching_loss": 0.013002996239811182, + "grad_norm": 0.07330109179019928, + "learning_rate": 2.817299961474417e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24350 + }, + { + "epoch": 7.087576374745417, + "flow_matching_loss": 0.012978363502770662, + "grad_norm": 0.06182662397623062, + "learning_rate": 2.8076609075268816e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24360 + }, + { + "epoch": 7.090485888856561, + "flow_matching_loss": 0.014339429419487714, + "grad_norm": 0.07838094979524612, + "learning_rate": 2.798036668421006e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24370 + }, + { + "epoch": 7.093395402967705, + "flow_matching_loss": 0.013236935622990132, + "grad_norm": 0.09148716926574707, + "learning_rate": 2.7884272558511433e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24380 + }, + { + "epoch": 7.0963049170788475, + "flow_matching_loss": 0.013799360487610102, + "grad_norm": 0.051487263292074203, + "learning_rate": 2.7788326814936456e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24390 + }, + { + "epoch": 7.099214431189991, + "flow_matching_loss": 0.01468724776059389, + "grad_norm": 0.06749580055475235, + "learning_rate": 2.7692529570068176e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24400 + }, + { + "epoch": 7.102123945301135, + "flow_matching_loss": 0.015345342736691237, + "grad_norm": 0.0819593146443367, + "learning_rate": 2.759688094030933e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24410 + }, + { + "epoch": 7.105033459412278, + "flow_matching_loss": 0.01353678060695529, + "grad_norm": 0.07594393938779831, + "learning_rate": 2.750138104188201e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24420 + }, + { + "epoch": 7.107942973523421, + "flow_matching_loss": 0.013083771336823702, + "grad_norm": 0.07168266177177429, + "learning_rate": 2.740602999082759e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24430 + }, + { + "epoch": 7.110852487634565, + "flow_matching_loss": 0.012622910365462304, + "grad_norm": 0.06300567090511322, + "learning_rate": 2.731082790300654e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24440 + }, + { + "epoch": 7.113762001745709, + "flow_matching_loss": 0.01342198308557272, + "grad_norm": 0.05351831391453743, + "learning_rate": 2.7215774894098474e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24450 + }, + { + "epoch": 7.116671515856852, + "flow_matching_loss": 0.011478703143075109, + "grad_norm": 0.06542030721902847, + "learning_rate": 2.7120871079601696e-06, + "loss": 0.0115, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24460 + }, + { + "epoch": 7.119581029967995, + "flow_matching_loss": 0.01376121724024415, + "grad_norm": 0.06712181121110916, + "learning_rate": 2.7026116574833314e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24470 + }, + { + "epoch": 7.122490544079139, + "flow_matching_loss": 0.012997860368341208, + "grad_norm": 0.08531537652015686, + "learning_rate": 2.6931511494928953e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24480 + }, + { + "epoch": 7.125400058190282, + "flow_matching_loss": 0.012936404161155223, + "grad_norm": 0.07962572574615479, + "learning_rate": 2.683705595484275e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24490 + }, + { + "epoch": 7.128309572301426, + "flow_matching_loss": 0.013959827926009893, + "grad_norm": 0.07102639228105545, + "learning_rate": 2.6742750069347014e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24500 + }, + { + "epoch": 7.131219086412569, + "flow_matching_loss": 0.01405642069876194, + "grad_norm": 0.10250754654407501, + "learning_rate": 2.664859395303239e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24510 + }, + { + "epoch": 7.134128600523712, + "flow_matching_loss": 0.015346614830195903, + "grad_norm": 0.09848271310329437, + "learning_rate": 2.655458772030739e-06, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24520 + }, + { + "epoch": 7.137038114634856, + "flow_matching_loss": 0.012335732020437718, + "grad_norm": 0.11077426373958588, + "learning_rate": 2.6460731485398403e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24530 + }, + { + "epoch": 7.1399476287459995, + "flow_matching_loss": 0.014516010601073503, + "grad_norm": 0.08722764253616333, + "learning_rate": 2.6367025362349695e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24540 + }, + { + "epoch": 7.142857142857143, + "flow_matching_loss": 0.013918992038816214, + "grad_norm": 0.06914642453193665, + "learning_rate": 2.6273469465022965e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24550 + }, + { + "epoch": 7.145766656968286, + "flow_matching_loss": 0.013358053006231784, + "grad_norm": 0.06617547571659088, + "learning_rate": 2.618006390709753e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24560 + }, + { + "epoch": 7.14867617107943, + "flow_matching_loss": 0.012914456520229578, + "grad_norm": 0.05920351669192314, + "learning_rate": 2.60868088020699e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24570 + }, + { + "epoch": 7.1515856851905735, + "flow_matching_loss": 0.013132689241319894, + "grad_norm": 0.06065352261066437, + "learning_rate": 2.599370426325382e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24580 + }, + { + "epoch": 7.154495199301716, + "flow_matching_loss": 0.014485084637999534, + "grad_norm": 0.0675547644495964, + "learning_rate": 2.5900750403780093e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24590 + }, + { + "epoch": 7.15740471341286, + "flow_matching_loss": 0.013018064945936204, + "grad_norm": 0.07292836159467697, + "learning_rate": 2.580794733659639e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24600 + }, + { + "epoch": 7.160314227524004, + "flow_matching_loss": 0.0146580308675766, + "grad_norm": 0.06828145682811737, + "learning_rate": 2.5715295174467236e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24610 + }, + { + "epoch": 7.1632237416351465, + "flow_matching_loss": 0.01275080731138587, + "grad_norm": 0.05610258877277374, + "learning_rate": 2.5622794029973746e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24620 + }, + { + "epoch": 7.16613325574629, + "flow_matching_loss": 0.013729071710258722, + "grad_norm": 0.060167279094457626, + "learning_rate": 2.553044401551351e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24630 + }, + { + "epoch": 7.169042769857434, + "flow_matching_loss": 0.014683324564248323, + "grad_norm": 0.08530737459659576, + "learning_rate": 2.54382452433005e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24640 + }, + { + "epoch": 7.171952283968578, + "flow_matching_loss": 0.01355051351711154, + "grad_norm": 0.12871025502681732, + "learning_rate": 2.53461978253649e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24650 + }, + { + "epoch": 7.1748617980797205, + "flow_matching_loss": 0.012540367711335421, + "grad_norm": 0.07559042423963547, + "learning_rate": 2.5254301873553065e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24660 + }, + { + "epoch": 7.177771312190864, + "flow_matching_loss": 0.013394430559128524, + "grad_norm": 0.0599004290997982, + "learning_rate": 2.5162557499527184e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24670 + }, + { + "epoch": 7.180680826302008, + "flow_matching_loss": 0.013001345936208964, + "grad_norm": 0.06158075109124184, + "learning_rate": 2.5070964814765335e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24680 + }, + { + "epoch": 7.183590340413151, + "flow_matching_loss": 0.013900089263916015, + "grad_norm": 0.08077487349510193, + "learning_rate": 2.4979523930561295e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24690 + }, + { + "epoch": 7.186499854524294, + "flow_matching_loss": 0.012050543073564767, + "grad_norm": 0.06680431962013245, + "learning_rate": 2.4888234958024333e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24700 + }, + { + "epoch": 7.189409368635438, + "flow_matching_loss": 0.01467258520424366, + "grad_norm": 0.08427324146032333, + "learning_rate": 2.479709800807915e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24710 + }, + { + "epoch": 7.192318882746581, + "flow_matching_loss": 0.013567049521952868, + "grad_norm": 0.09092753380537033, + "learning_rate": 2.4706113191465767e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24720 + }, + { + "epoch": 7.195228396857725, + "flow_matching_loss": 0.01344510456547141, + "grad_norm": 0.08086074888706207, + "learning_rate": 2.46152806187393e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24730 + }, + { + "epoch": 7.198137910968868, + "flow_matching_loss": 0.01235598223283887, + "grad_norm": 0.08013242483139038, + "learning_rate": 2.452460040026989e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24740 + }, + { + "epoch": 7.201047425080012, + "flow_matching_loss": 0.01429294953122735, + "grad_norm": 0.07373183220624924, + "learning_rate": 2.4434072646242555e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24750 + }, + { + "epoch": 7.203956939191155, + "flow_matching_loss": 0.012713455688208341, + "grad_norm": 0.07295826077461243, + "learning_rate": 2.4343697466657015e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24760 + }, + { + "epoch": 7.206866453302299, + "flow_matching_loss": 0.012177757546305656, + "grad_norm": 0.07526753842830658, + "learning_rate": 2.42534749713277e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24770 + }, + { + "epoch": 7.209775967413442, + "flow_matching_loss": 0.013773608487099409, + "grad_norm": 0.05993805453181267, + "learning_rate": 2.4163405269883425e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24780 + }, + { + "epoch": 7.212685481524585, + "flow_matching_loss": 0.01409003660082817, + "grad_norm": 0.06386899948120117, + "learning_rate": 2.407348847176735e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24790 + }, + { + "epoch": 7.215594995635729, + "flow_matching_loss": 0.013851764891296626, + "grad_norm": 0.07432347536087036, + "learning_rate": 2.398372468623688e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24800 + }, + { + "epoch": 7.2185045097468725, + "flow_matching_loss": 0.012573564238846303, + "grad_norm": 0.060605648905038834, + "learning_rate": 2.3894114022363446e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24810 + }, + { + "epoch": 7.221414023858015, + "flow_matching_loss": 0.012682951148599386, + "grad_norm": 0.08004429191350937, + "learning_rate": 2.3804656589032514e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24820 + }, + { + "epoch": 7.224323537969159, + "flow_matching_loss": 0.012903656903654337, + "grad_norm": 0.07317494601011276, + "learning_rate": 2.3715352494943237e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24830 + }, + { + "epoch": 7.227233052080303, + "flow_matching_loss": 0.013238308671861886, + "grad_norm": 0.06420118361711502, + "learning_rate": 2.3626201848608582e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24840 + }, + { + "epoch": 7.2301425661914465, + "flow_matching_loss": 0.013486605044454337, + "grad_norm": 0.07841554284095764, + "learning_rate": 2.353720475835494e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24850 + }, + { + "epoch": 7.233052080302589, + "flow_matching_loss": 0.012401058617979289, + "grad_norm": 0.06129325181245804, + "learning_rate": 2.344836133232215e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24860 + }, + { + "epoch": 7.235961594413733, + "flow_matching_loss": 0.014048645738512278, + "grad_norm": 0.07047711312770844, + "learning_rate": 2.335967167846342e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24870 + }, + { + "epoch": 7.238871108524877, + "flow_matching_loss": 0.013026804849505424, + "grad_norm": 0.07703019678592682, + "learning_rate": 2.3271135904544993e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24880 + }, + { + "epoch": 7.2417806226360195, + "flow_matching_loss": 0.012371654342859983, + "grad_norm": 0.09147273004055023, + "learning_rate": 2.3182754118146206e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24890 + }, + { + "epoch": 7.244690136747163, + "flow_matching_loss": 0.013368748128414154, + "grad_norm": 0.06538436561822891, + "learning_rate": 2.309452642665923e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24900 + }, + { + "epoch": 7.247599650858307, + "flow_matching_loss": 0.013684569858014584, + "grad_norm": 0.08671076595783234, + "learning_rate": 2.3006452937289065e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24910 + }, + { + "epoch": 7.25050916496945, + "flow_matching_loss": 0.01364570902660489, + "grad_norm": 0.06522215157747269, + "learning_rate": 2.291853375705327e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24920 + }, + { + "epoch": 7.2534186790805935, + "flow_matching_loss": 0.013512483984231948, + "grad_norm": 0.07479485869407654, + "learning_rate": 2.283076899278197e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24930 + }, + { + "epoch": 7.256328193191737, + "flow_matching_loss": 0.013694265577942132, + "grad_norm": 0.06213264912366867, + "learning_rate": 2.274315875111763e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24940 + }, + { + "epoch": 7.25923770730288, + "flow_matching_loss": 0.013416297547519208, + "grad_norm": 0.07697916775941849, + "learning_rate": 2.265570313851493e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24950 + }, + { + "epoch": 7.262147221414024, + "flow_matching_loss": 0.013248101249337197, + "grad_norm": 0.07165759801864624, + "learning_rate": 2.2568402261240685e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24960 + }, + { + "epoch": 7.265056735525167, + "flow_matching_loss": 0.01347733773291111, + "grad_norm": 0.06621304899454117, + "learning_rate": 2.248125622537373e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24970 + }, + { + "epoch": 7.267966249636311, + "flow_matching_loss": 0.0158517986536026, + "grad_norm": 0.09131656587123871, + "learning_rate": 2.239426513680467e-06, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24980 + }, + { + "epoch": 7.270875763747454, + "flow_matching_loss": 0.011721732653677463, + "grad_norm": 0.0716676190495491, + "learning_rate": 2.2307429101235915e-06, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 24990 + }, + { + "epoch": 7.273785277858598, + "flow_matching_loss": 0.014215456880629062, + "grad_norm": 0.07658518105745316, + "learning_rate": 2.222074822418145e-06, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25000 + }, + { + "epoch": 7.276694791969741, + "flow_matching_loss": 0.013348527625203133, + "grad_norm": 0.06635493040084839, + "learning_rate": 2.213422261096668e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25010 + }, + { + "epoch": 7.279604306080884, + "flow_matching_loss": 0.012773536890745164, + "grad_norm": 0.07704836875200272, + "learning_rate": 2.204785236672835e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25020 + }, + { + "epoch": 7.282513820192028, + "flow_matching_loss": 0.013284770864993333, + "grad_norm": 0.07035034894943237, + "learning_rate": 2.1961637596414536e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25030 + }, + { + "epoch": 7.285423334303172, + "flow_matching_loss": 0.014852165151387453, + "grad_norm": 0.08377543091773987, + "learning_rate": 2.1875578404784275e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25040 + }, + { + "epoch": 7.288332848414315, + "flow_matching_loss": 0.01274118833243847, + "grad_norm": 0.0636291354894638, + "learning_rate": 2.1789674896407583e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25050 + }, + { + "epoch": 7.291242362525458, + "flow_matching_loss": 0.013318611774593592, + "grad_norm": 0.06529203057289124, + "learning_rate": 2.170392717566534e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25060 + }, + { + "epoch": 7.294151876636602, + "flow_matching_loss": 0.013562395796179771, + "grad_norm": 0.05961434170603752, + "learning_rate": 2.161833534674909e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25070 + }, + { + "epoch": 7.2970613907477455, + "flow_matching_loss": 0.013894814904779196, + "grad_norm": 0.05506366491317749, + "learning_rate": 2.1532899513661015e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25080 + }, + { + "epoch": 7.299970904858888, + "flow_matching_loss": 0.013189655635505915, + "grad_norm": 0.08422168344259262, + "learning_rate": 2.1447619780213682e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25090 + }, + { + "epoch": 7.302880418970032, + "flow_matching_loss": 0.012749161757528781, + "grad_norm": 0.057954125106334686, + "learning_rate": 2.136249625003001e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25100 + }, + { + "epoch": 7.305789933081176, + "flow_matching_loss": 0.013357336819171905, + "grad_norm": 0.061063215136528015, + "learning_rate": 2.1277529026543107e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25110 + }, + { + "epoch": 7.308699447192319, + "flow_matching_loss": 0.013787130638957024, + "grad_norm": 0.0631231740117073, + "learning_rate": 2.1192718212996188e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25120 + }, + { + "epoch": 7.311608961303462, + "flow_matching_loss": 0.01107627279125154, + "grad_norm": 0.06361349672079086, + "learning_rate": 2.110806391244235e-06, + "loss": 0.0111, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25130 + }, + { + "epoch": 7.314518475414606, + "flow_matching_loss": 0.013862110394984483, + "grad_norm": 0.07330663502216339, + "learning_rate": 2.1023566227744596e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25140 + }, + { + "epoch": 7.317427989525749, + "flow_matching_loss": 0.013896622695028782, + "grad_norm": 0.06631974875926971, + "learning_rate": 2.0939225261575563e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25150 + }, + { + "epoch": 7.3203375036368925, + "flow_matching_loss": 0.012656226754188538, + "grad_norm": 0.0903475359082222, + "learning_rate": 2.0855041116417473e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25160 + }, + { + "epoch": 7.323247017748036, + "flow_matching_loss": 0.012560615222901106, + "grad_norm": 0.09666713327169418, + "learning_rate": 2.0771013894562003e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25170 + }, + { + "epoch": 7.32615653185918, + "flow_matching_loss": 0.013111697789281607, + "grad_norm": 0.06514808535575867, + "learning_rate": 2.0687143698110113e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25180 + }, + { + "epoch": 7.329066045970323, + "flow_matching_loss": 0.013500704802572728, + "grad_norm": 0.12469997256994247, + "learning_rate": 2.060343062897208e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25190 + }, + { + "epoch": 7.3319755600814664, + "flow_matching_loss": 0.013077361043542624, + "grad_norm": 0.0766192078590393, + "learning_rate": 2.0519874788867127e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25200 + }, + { + "epoch": 7.33488507419261, + "flow_matching_loss": 0.01428374219685793, + "grad_norm": 0.08772586286067963, + "learning_rate": 2.043647627932347e-06, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25210 + }, + { + "epoch": 7.337794588303753, + "flow_matching_loss": 0.01329247998073697, + "grad_norm": 0.07366669178009033, + "learning_rate": 2.0353235201678193e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25220 + }, + { + "epoch": 7.340704102414897, + "flow_matching_loss": 0.012601849902421237, + "grad_norm": 0.0738757848739624, + "learning_rate": 2.0270151657076995e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25230 + }, + { + "epoch": 7.34361361652604, + "flow_matching_loss": 0.014987087342888117, + "grad_norm": 0.07412946969270706, + "learning_rate": 2.0187225746474286e-06, + "loss": 0.015, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25240 + }, + { + "epoch": 7.346523130637183, + "flow_matching_loss": 0.012823039898648858, + "grad_norm": 0.06469246000051498, + "learning_rate": 2.0104457570632845e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25250 + }, + { + "epoch": 7.349432644748327, + "flow_matching_loss": 0.012757815793156624, + "grad_norm": 0.05866915360093117, + "learning_rate": 2.0021847230123775e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25260 + }, + { + "epoch": 7.352342158859471, + "flow_matching_loss": 0.011704725213348866, + "grad_norm": 0.05763552337884903, + "learning_rate": 1.9939394825326494e-06, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25270 + }, + { + "epoch": 7.355251672970613, + "flow_matching_loss": 0.012225421564653515, + "grad_norm": 0.058036766946315765, + "learning_rate": 1.9857100456428385e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25280 + }, + { + "epoch": 7.358161187081757, + "flow_matching_loss": 0.013591224234551192, + "grad_norm": 0.07278840988874435, + "learning_rate": 1.977496422342492e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25290 + }, + { + "epoch": 7.361070701192901, + "flow_matching_loss": 0.013175432104617357, + "grad_norm": 0.06886667013168335, + "learning_rate": 1.9692986226119345e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25300 + }, + { + "epoch": 7.363980215304045, + "flow_matching_loss": 0.014913164172321558, + "grad_norm": 0.07464814186096191, + "learning_rate": 1.9611166564122658e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25310 + }, + { + "epoch": 7.366889729415187, + "flow_matching_loss": 0.013248833827674389, + "grad_norm": 0.07841351628303528, + "learning_rate": 1.9529505336853456e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25320 + }, + { + "epoch": 7.369799243526331, + "flow_matching_loss": 0.014733208809047937, + "grad_norm": 0.07054108381271362, + "learning_rate": 1.9448002643537833e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25330 + }, + { + "epoch": 7.372708757637475, + "flow_matching_loss": 0.012118788715451956, + "grad_norm": 0.06721708923578262, + "learning_rate": 1.9366658583209205e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25340 + }, + { + "epoch": 7.375618271748618, + "flow_matching_loss": 0.012692227307707072, + "grad_norm": 0.06883901357650757, + "learning_rate": 1.9285473254708363e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25350 + }, + { + "epoch": 7.378527785859761, + "flow_matching_loss": 0.01361623527482152, + "grad_norm": 0.08626219630241394, + "learning_rate": 1.9204446756683092e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25360 + }, + { + "epoch": 7.381437299970905, + "flow_matching_loss": 0.013153980486094952, + "grad_norm": 0.059991322457790375, + "learning_rate": 1.912357918758822e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25370 + }, + { + "epoch": 7.384346814082049, + "flow_matching_loss": 0.01259128637611866, + "grad_norm": 0.06731656193733215, + "learning_rate": 1.9042870645685478e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25380 + }, + { + "epoch": 7.387256328193192, + "flow_matching_loss": 0.013809979800134898, + "grad_norm": 0.053773701190948486, + "learning_rate": 1.8962321229043322e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25390 + }, + { + "epoch": 7.390165842304335, + "flow_matching_loss": 0.01444683400914073, + "grad_norm": 0.07284468412399292, + "learning_rate": 1.8881931035536943e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25400 + }, + { + "epoch": 7.393075356415479, + "flow_matching_loss": 0.013573806080967189, + "grad_norm": 0.07249485701322556, + "learning_rate": 1.8801700162847952e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25410 + }, + { + "epoch": 7.395984870526622, + "flow_matching_loss": 0.013656122516840697, + "grad_norm": 0.07880706340074539, + "learning_rate": 1.87216287084645e-06, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25420 + }, + { + "epoch": 7.3988943846377655, + "flow_matching_loss": 0.013594516832381488, + "grad_norm": 0.0918034091591835, + "learning_rate": 1.8641716769680895e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25430 + }, + { + "epoch": 7.401803898748909, + "flow_matching_loss": 0.012425165437161923, + "grad_norm": 0.06750696152448654, + "learning_rate": 1.8561964443597662e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25440 + }, + { + "epoch": 7.404713412860052, + "flow_matching_loss": 0.013372705783694983, + "grad_norm": 0.07486861944198608, + "learning_rate": 1.848237182712147e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25450 + }, + { + "epoch": 7.407622926971196, + "flow_matching_loss": 0.013044159254059195, + "grad_norm": 0.0612102635204792, + "learning_rate": 1.8402939016964788e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25460 + }, + { + "epoch": 7.410532441082339, + "flow_matching_loss": 0.013573934603482486, + "grad_norm": 0.05162703990936279, + "learning_rate": 1.832366610964601e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25470 + }, + { + "epoch": 7.413441955193482, + "flow_matching_loss": 0.01289509804919362, + "grad_norm": 0.06643415987491608, + "learning_rate": 1.8244553201489166e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25480 + }, + { + "epoch": 7.416351469304626, + "flow_matching_loss": 0.012731716502457857, + "grad_norm": 0.07722578197717667, + "learning_rate": 1.8165600388623882e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25490 + }, + { + "epoch": 7.41926098341577, + "flow_matching_loss": 0.013760777097195387, + "grad_norm": 0.07765767723321915, + "learning_rate": 1.8086807766985331e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25500 + }, + { + "epoch": 7.422170497526913, + "flow_matching_loss": 0.012360454350709916, + "grad_norm": 0.0611170195043087, + "learning_rate": 1.8008175432313946e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25510 + }, + { + "epoch": 7.425080011638056, + "flow_matching_loss": 0.012308253347873688, + "grad_norm": 0.08464125543832779, + "learning_rate": 1.7929703480155429e-06, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25520 + }, + { + "epoch": 7.4279895257492, + "flow_matching_loss": 0.014114185282960535, + "grad_norm": 0.06951863318681717, + "learning_rate": 1.7851392005860595e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25530 + }, + { + "epoch": 7.430899039860344, + "flow_matching_loss": 0.013157390616834163, + "grad_norm": 0.06613007932901382, + "learning_rate": 1.7773241104585243e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25540 + }, + { + "epoch": 7.433808553971486, + "flow_matching_loss": 0.012422007601708173, + "grad_norm": 0.06958946585655212, + "learning_rate": 1.7695250871290108e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25550 + }, + { + "epoch": 7.43671806808263, + "flow_matching_loss": 0.013981592934578657, + "grad_norm": 0.06756360083818436, + "learning_rate": 1.761742140074074e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25560 + }, + { + "epoch": 7.439627582193774, + "flow_matching_loss": 0.012837771046906709, + "grad_norm": 0.07985037565231323, + "learning_rate": 1.7539752787507208e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25570 + }, + { + "epoch": 7.442537096304917, + "flow_matching_loss": 0.013995966874063015, + "grad_norm": 0.07491634786128998, + "learning_rate": 1.7462245125964233e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25580 + }, + { + "epoch": 7.44544661041606, + "flow_matching_loss": 0.01487233191728592, + "grad_norm": 0.09043553471565247, + "learning_rate": 1.7384898510290931e-06, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25590 + }, + { + "epoch": 7.448356124527204, + "flow_matching_loss": 0.01382870078086853, + "grad_norm": 0.07468879967927933, + "learning_rate": 1.730771303447069e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25600 + }, + { + "epoch": 7.451265638638348, + "flow_matching_loss": 0.013018564973026513, + "grad_norm": 0.060955777764320374, + "learning_rate": 1.7230688792291204e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25610 + }, + { + "epoch": 7.454175152749491, + "flow_matching_loss": 0.014835781510919332, + "grad_norm": 0.08731991052627563, + "learning_rate": 1.7153825877344175e-06, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25620 + }, + { + "epoch": 7.457084666860634, + "flow_matching_loss": 0.013066833931952715, + "grad_norm": 0.056844405829906464, + "learning_rate": 1.7077124383025278e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25630 + }, + { + "epoch": 7.459994180971778, + "flow_matching_loss": 0.012611515540629625, + "grad_norm": 0.07197578251361847, + "learning_rate": 1.7000584402534058e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25640 + }, + { + "epoch": 7.462903695082921, + "flow_matching_loss": 0.013212213944643735, + "grad_norm": 0.05362580716609955, + "learning_rate": 1.6924206028873774e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25650 + }, + { + "epoch": 7.4658132091940645, + "flow_matching_loss": 0.013975289277732372, + "grad_norm": 0.06716492772102356, + "learning_rate": 1.6847989354851434e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25660 + }, + { + "epoch": 7.468722723305208, + "flow_matching_loss": 0.011759980767965316, + "grad_norm": 0.06728854775428772, + "learning_rate": 1.6771934473077443e-06, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25670 + }, + { + "epoch": 7.471632237416351, + "flow_matching_loss": 0.013437493704259396, + "grad_norm": 0.08397552371025085, + "learning_rate": 1.669604147596563e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25680 + }, + { + "epoch": 7.474541751527495, + "flow_matching_loss": 0.013044299185276031, + "grad_norm": 0.0764714851975441, + "learning_rate": 1.662031045573319e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25690 + }, + { + "epoch": 7.4774512656386385, + "flow_matching_loss": 0.013367217499762774, + "grad_norm": 0.06390617787837982, + "learning_rate": 1.6544741504400402e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25700 + }, + { + "epoch": 7.480360779749782, + "flow_matching_loss": 0.01193312294781208, + "grad_norm": 0.07767977565526962, + "learning_rate": 1.6469334713790712e-06, + "loss": 0.0119, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25710 + }, + { + "epoch": 7.483270293860925, + "flow_matching_loss": 0.013833898399025202, + "grad_norm": 0.08869464695453644, + "learning_rate": 1.639409017553048e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25720 + }, + { + "epoch": 7.486179807972069, + "flow_matching_loss": 0.012996876984834671, + "grad_norm": 0.059234894812107086, + "learning_rate": 1.6319007981048884e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25730 + }, + { + "epoch": 7.489089322083212, + "flow_matching_loss": 0.012575512658804655, + "grad_norm": 0.06706538796424866, + "learning_rate": 1.6244088221577896e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25740 + }, + { + "epoch": 7.491998836194355, + "flow_matching_loss": 0.014055040944367648, + "grad_norm": 0.08129065483808517, + "learning_rate": 1.6169330988152081e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25750 + }, + { + "epoch": 7.494908350305499, + "flow_matching_loss": 0.0130197879858315, + "grad_norm": 0.07364487648010254, + "learning_rate": 1.6094736371608475e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25760 + }, + { + "epoch": 7.497817864416643, + "flow_matching_loss": 0.013128327205777168, + "grad_norm": 0.10549283772706985, + "learning_rate": 1.602030446258666e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25770 + }, + { + "epoch": 7.5007273785277855, + "flow_matching_loss": 0.013076996989548206, + "grad_norm": 0.07743141800165176, + "learning_rate": 1.5946035351528388e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25780 + }, + { + "epoch": 7.503636892638929, + "flow_matching_loss": 0.012044836953282356, + "grad_norm": 0.07060275971889496, + "learning_rate": 1.5871929128677621e-06, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25790 + }, + { + "epoch": 7.506546406750073, + "flow_matching_loss": 0.013454039767384528, + "grad_norm": 0.08691464364528656, + "learning_rate": 1.5797985884080412e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25800 + }, + { + "epoch": 7.509455920861216, + "flow_matching_loss": 0.013419202249497175, + "grad_norm": 0.08581821620464325, + "learning_rate": 1.572420570758476e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25810 + }, + { + "epoch": 7.512365434972359, + "flow_matching_loss": 0.015227650851011276, + "grad_norm": 0.07373550534248352, + "learning_rate": 1.5650588688840567e-06, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25820 + }, + { + "epoch": 7.515274949083503, + "flow_matching_loss": 0.01454822914674878, + "grad_norm": 0.06605472415685654, + "learning_rate": 1.5577134917299413e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25830 + }, + { + "epoch": 7.518184463194647, + "flow_matching_loss": 0.012643444584682584, + "grad_norm": 0.08032344281673431, + "learning_rate": 1.5503844482214618e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25840 + }, + { + "epoch": 7.52109397730579, + "flow_matching_loss": 0.012404222367331385, + "grad_norm": 0.05586589500308037, + "learning_rate": 1.5430717472640932e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25850 + }, + { + "epoch": 7.524003491416933, + "flow_matching_loss": 0.01380286468192935, + "grad_norm": 0.07586577534675598, + "learning_rate": 1.5357753977434541e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25860 + }, + { + "epoch": 7.526913005528077, + "flow_matching_loss": 0.013627897948026657, + "grad_norm": 0.0751841738820076, + "learning_rate": 1.5284954085253043e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25870 + }, + { + "epoch": 7.52982251963922, + "flow_matching_loss": 0.013027400895953178, + "grad_norm": 0.06680558621883392, + "learning_rate": 1.5212317884555115e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25880 + }, + { + "epoch": 7.532732033750364, + "flow_matching_loss": 0.012702945061028004, + "grad_norm": 0.06997482478618622, + "learning_rate": 1.5139845463600605e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25890 + }, + { + "epoch": 7.535641547861507, + "flow_matching_loss": 0.012605775147676468, + "grad_norm": 0.07440752536058426, + "learning_rate": 1.5067536910450352e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25900 + }, + { + "epoch": 7.538551061972651, + "flow_matching_loss": 0.010807568859308957, + "grad_norm": 0.06377705931663513, + "learning_rate": 1.4995392312966044e-06, + "loss": 0.0108, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25910 + }, + { + "epoch": 7.541460576083794, + "flow_matching_loss": 0.012867990788072348, + "grad_norm": 0.06984561681747437, + "learning_rate": 1.492341175881014e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25920 + }, + { + "epoch": 7.5443700901949375, + "flow_matching_loss": 0.013522308319807053, + "grad_norm": 0.06588717550039291, + "learning_rate": 1.4851595335445895e-06, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25930 + }, + { + "epoch": 7.547279604306081, + "flow_matching_loss": 0.013972195470705629, + "grad_norm": 0.07141748070716858, + "learning_rate": 1.4779943130136953e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25940 + }, + { + "epoch": 7.550189118417224, + "flow_matching_loss": 0.012063881382346153, + "grad_norm": 0.06949219107627869, + "learning_rate": 1.4708455229947537e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25950 + }, + { + "epoch": 7.553098632528368, + "flow_matching_loss": 0.012670301739126444, + "grad_norm": 0.06645029038190842, + "learning_rate": 1.4637131721742148e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25960 + }, + { + "epoch": 7.5560081466395115, + "flow_matching_loss": 0.013108430709689856, + "grad_norm": 0.06455008685588837, + "learning_rate": 1.4565972692185626e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25970 + }, + { + "epoch": 7.558917660750654, + "flow_matching_loss": 0.012778581865131856, + "grad_norm": 0.06960225850343704, + "learning_rate": 1.4494978227742855e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25980 + }, + { + "epoch": 7.561827174861798, + "flow_matching_loss": 0.014144712220877409, + "grad_norm": 0.0539361834526062, + "learning_rate": 1.442414841467885e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 25990 + }, + { + "epoch": 7.564736688972942, + "flow_matching_loss": 0.013834370113909245, + "grad_norm": 0.08327645808458328, + "learning_rate": 1.4353483339058498e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26000 + }, + { + "epoch": 7.5676462030840845, + "flow_matching_loss": 0.012613664101809263, + "grad_norm": 0.05255855992436409, + "learning_rate": 1.4282983086746498e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26010 + }, + { + "epoch": 7.570555717195228, + "flow_matching_loss": 0.011455109110102057, + "grad_norm": 0.06777452677488327, + "learning_rate": 1.4212647743407325e-06, + "loss": 0.0115, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26020 + }, + { + "epoch": 7.573465231306372, + "flow_matching_loss": 0.013400793820619584, + "grad_norm": 0.06851029396057129, + "learning_rate": 1.4142477394505065e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26030 + }, + { + "epoch": 7.576374745417516, + "flow_matching_loss": 0.013164529483765364, + "grad_norm": 0.07324803620576859, + "learning_rate": 1.4072472125303299e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26040 + }, + { + "epoch": 7.5792842595286585, + "flow_matching_loss": 0.012849375791847705, + "grad_norm": 0.06068152189254761, + "learning_rate": 1.4002632020865031e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26050 + }, + { + "epoch": 7.582193773639802, + "flow_matching_loss": 0.013119243085384369, + "grad_norm": 0.06994456797838211, + "learning_rate": 1.393295716605258e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26060 + }, + { + "epoch": 7.585103287750946, + "flow_matching_loss": 0.013393313251435756, + "grad_norm": 0.05553772300481796, + "learning_rate": 1.386344764552742e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26070 + }, + { + "epoch": 7.588012801862089, + "flow_matching_loss": 0.012170665245503188, + "grad_norm": 0.08011575788259506, + "learning_rate": 1.3794103543750246e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26080 + }, + { + "epoch": 7.590922315973232, + "flow_matching_loss": 0.012221443466842175, + "grad_norm": 0.06558685004711151, + "learning_rate": 1.3724924944980637e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26090 + }, + { + "epoch": 7.593831830084376, + "flow_matching_loss": 0.012027917802333832, + "grad_norm": 0.06178373470902443, + "learning_rate": 1.365591193327712e-06, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26100 + }, + { + "epoch": 7.59674134419552, + "flow_matching_loss": 0.014128809981048106, + "grad_norm": 0.0684836134314537, + "learning_rate": 1.3587064592496984e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26110 + }, + { + "epoch": 7.599650858306663, + "flow_matching_loss": 0.014527949038892984, + "grad_norm": 0.09612485021352768, + "learning_rate": 1.3518383006296292e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26120 + }, + { + "epoch": 7.602560372417806, + "flow_matching_loss": 0.011967008467763662, + "grad_norm": 0.07348043471574783, + "learning_rate": 1.344986725812959e-06, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26130 + }, + { + "epoch": 7.605469886528949, + "flow_matching_loss": 0.012143870070576667, + "grad_norm": 0.11264265328645706, + "learning_rate": 1.3381517431250046e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26140 + }, + { + "epoch": 7.608379400640093, + "flow_matching_loss": 0.013177009485661983, + "grad_norm": 0.05684136226773262, + "learning_rate": 1.3313333608709095e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26150 + }, + { + "epoch": 7.611288914751237, + "flow_matching_loss": 0.013223960995674133, + "grad_norm": 0.0818810984492302, + "learning_rate": 1.3245315873356528e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26160 + }, + { + "epoch": 7.61419842886238, + "flow_matching_loss": 0.01265435153618455, + "grad_norm": 0.07102756947278976, + "learning_rate": 1.3177464307840304e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26170 + }, + { + "epoch": 7.617107942973523, + "flow_matching_loss": 0.012458936031907796, + "grad_norm": 0.06177353113889694, + "learning_rate": 1.310977899460642e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26180 + }, + { + "epoch": 7.620017457084667, + "flow_matching_loss": 0.013267368264496326, + "grad_norm": 0.06846771389245987, + "learning_rate": 1.3042260015899006e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26190 + }, + { + "epoch": 7.6229269711958105, + "flow_matching_loss": 0.01437487518414855, + "grad_norm": 0.08218520134687424, + "learning_rate": 1.2974907453759948e-06, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26200 + }, + { + "epoch": 7.625836485306953, + "flow_matching_loss": 0.014479259122163057, + "grad_norm": 0.0833488404750824, + "learning_rate": 1.2907721390028937e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26210 + }, + { + "epoch": 7.628745999418097, + "flow_matching_loss": 0.013965187780559062, + "grad_norm": 0.07312171906232834, + "learning_rate": 1.2840701906343417e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26220 + }, + { + "epoch": 7.631655513529241, + "flow_matching_loss": 0.014086830615997314, + "grad_norm": 0.07482457160949707, + "learning_rate": 1.2773849084138312e-06, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26230 + }, + { + "epoch": 7.6345650276403845, + "flow_matching_loss": 0.012106719706207515, + "grad_norm": 0.08135700970888138, + "learning_rate": 1.2707163004646177e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26240 + }, + { + "epoch": 7.637474541751527, + "flow_matching_loss": 0.012582778371870518, + "grad_norm": 0.06991448253393173, + "learning_rate": 1.2640643748896879e-06, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26250 + }, + { + "epoch": 7.640384055862671, + "flow_matching_loss": 0.012748242914676666, + "grad_norm": 0.05955749750137329, + "learning_rate": 1.257429139771753e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26260 + }, + { + "epoch": 7.643293569973815, + "flow_matching_loss": 0.013253430649638177, + "grad_norm": 0.07165494561195374, + "learning_rate": 1.2508106031732553e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26270 + }, + { + "epoch": 7.6462030840849575, + "flow_matching_loss": 0.012796518113464116, + "grad_norm": 0.06198901683092117, + "learning_rate": 1.2442087731363366e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26280 + }, + { + "epoch": 7.649112598196101, + "flow_matching_loss": 0.013623378146439791, + "grad_norm": 0.0833805724978447, + "learning_rate": 1.237623657682847e-06, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26290 + }, + { + "epoch": 7.652022112307245, + "flow_matching_loss": 0.013237891625612975, + "grad_norm": 0.09005767852067947, + "learning_rate": 1.2310552648143208e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26300 + }, + { + "epoch": 7.654931626418388, + "flow_matching_loss": 0.012796226982027292, + "grad_norm": 0.08633266389369965, + "learning_rate": 1.2245036025119738e-06, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26310 + }, + { + "epoch": 7.6578411405295315, + "flow_matching_loss": 0.012880842760205269, + "grad_norm": 0.088819220662117, + "learning_rate": 1.2179686787366962e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26320 + }, + { + "epoch": 7.660750654640675, + "flow_matching_loss": 0.01447228780016303, + "grad_norm": 0.09020951390266418, + "learning_rate": 1.2114505014290329e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26330 + }, + { + "epoch": 7.663660168751818, + "flow_matching_loss": 0.014623017515987158, + "grad_norm": 0.06446749716997147, + "learning_rate": 1.2049490785091833e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26340 + }, + { + "epoch": 7.666569682862962, + "flow_matching_loss": 0.014596761204302311, + "grad_norm": 0.07044199109077454, + "learning_rate": 1.1984644178769949e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26350 + }, + { + "epoch": 7.669479196974105, + "flow_matching_loss": 0.013107201643288135, + "grad_norm": 0.06445644050836563, + "learning_rate": 1.1919965274119381e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26360 + }, + { + "epoch": 7.672388711085249, + "flow_matching_loss": 0.01301550716161728, + "grad_norm": 0.07121271640062332, + "learning_rate": 1.18554541497311e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26370 + }, + { + "epoch": 7.675298225196392, + "flow_matching_loss": 0.012729758396744728, + "grad_norm": 0.06445855647325516, + "learning_rate": 1.1791110883992213e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26380 + }, + { + "epoch": 7.678207739307536, + "flow_matching_loss": 0.012709571700543165, + "grad_norm": 0.06900112330913544, + "learning_rate": 1.1726935555085817e-06, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26390 + }, + { + "epoch": 7.681117253418679, + "flow_matching_loss": 0.013329302985221147, + "grad_norm": 0.06442254781723022, + "learning_rate": 1.166292824099105e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26400 + }, + { + "epoch": 7.684026767529822, + "flow_matching_loss": 0.012529503740370273, + "grad_norm": 0.06550145894289017, + "learning_rate": 1.1599089019482768e-06, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26410 + }, + { + "epoch": 7.686936281640966, + "flow_matching_loss": 0.013893236964941024, + "grad_norm": 0.06959931552410126, + "learning_rate": 1.1535417968131718e-06, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26420 + }, + { + "epoch": 7.68984579575211, + "flow_matching_loss": 0.01299024187028408, + "grad_norm": 0.07501140236854553, + "learning_rate": 1.147191516430418e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26430 + }, + { + "epoch": 7.692755309863253, + "flow_matching_loss": 0.012387743499130011, + "grad_norm": 0.0702316164970398, + "learning_rate": 1.140858068516205e-06, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26440 + }, + { + "epoch": 7.695664823974396, + "flow_matching_loss": 0.013311941921710969, + "grad_norm": 0.08723565191030502, + "learning_rate": 1.1345414607662736e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26450 + }, + { + "epoch": 7.69857433808554, + "flow_matching_loss": 0.014031861163675784, + "grad_norm": 0.06859581917524338, + "learning_rate": 1.1282417008558937e-06, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26460 + }, + { + "epoch": 7.7014838521966835, + "flow_matching_loss": 0.011532760970294476, + "grad_norm": 0.0635051354765892, + "learning_rate": 1.1219587964398674e-06, + "loss": 0.0115, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26470 + }, + { + "epoch": 7.704393366307826, + "flow_matching_loss": 0.014578307420015335, + "grad_norm": 0.08082127571105957, + "learning_rate": 1.1156927551525198e-06, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26480 + }, + { + "epoch": 7.70730288041897, + "flow_matching_loss": 0.013138892222195863, + "grad_norm": 0.07608868181705475, + "learning_rate": 1.109443584607675e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26490 + }, + { + "epoch": 7.710212394530114, + "flow_matching_loss": 0.013294264115393162, + "grad_norm": 0.08032077550888062, + "learning_rate": 1.1032112923986708e-06, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26500 + }, + { + "epoch": 7.713121908641257, + "flow_matching_loss": 0.011254991218447686, + "grad_norm": 0.06743570417165756, + "learning_rate": 1.0969958860983299e-06, + "loss": 0.0113, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26510 + }, + { + "epoch": 7.7160314227524, + "flow_matching_loss": 0.013446665462106466, + "grad_norm": 0.07653339207172394, + "learning_rate": 1.0907973732589537e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26520 + }, + { + "epoch": 7.718940936863544, + "flow_matching_loss": 0.015130050294101239, + "grad_norm": 0.08206222206354141, + "learning_rate": 1.0846157614123215e-06, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26530 + }, + { + "epoch": 7.721850450974687, + "flow_matching_loss": 0.013189208693802357, + "grad_norm": 0.08971752226352692, + "learning_rate": 1.078451058069671e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26540 + }, + { + "epoch": 7.7247599650858305, + "flow_matching_loss": 0.012039609532803298, + "grad_norm": 0.08913372457027435, + "learning_rate": 1.0723032707217056e-06, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26550 + }, + { + "epoch": 7.727669479196974, + "flow_matching_loss": 0.013056432735174895, + "grad_norm": 0.09037644416093826, + "learning_rate": 1.066172406838558e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26560 + }, + { + "epoch": 7.730578993308118, + "flow_matching_loss": 0.013806803524494171, + "grad_norm": 0.08915790915489197, + "learning_rate": 1.0600584738698127e-06, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26570 + }, + { + "epoch": 7.733488507419261, + "flow_matching_loss": 0.012204782385379077, + "grad_norm": 0.07305504381656647, + "learning_rate": 1.0539614792444735e-06, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26580 + }, + { + "epoch": 7.7363980215304045, + "flow_matching_loss": 0.013216180354356765, + "grad_norm": 0.06446444243192673, + "learning_rate": 1.0478814303709627e-06, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26590 + }, + { + "epoch": 7.739307535641548, + "flow_matching_loss": 0.012863661348819732, + "grad_norm": 0.07925805449485779, + "learning_rate": 1.041818334637109e-06, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26600 + }, + { + "epoch": 7.742217049752691, + "flow_matching_loss": 0.01341062057763338, + "grad_norm": 0.06890176236629486, + "learning_rate": 1.0357721994101526e-06, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26610 + }, + { + "epoch": 7.745126563863835, + "flow_matching_loss": 0.012051971163600683, + "grad_norm": 0.07198187708854675, + "learning_rate": 1.0297430320367135e-06, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26620 + }, + { + "epoch": 7.748036077974978, + "flow_matching_loss": 0.014723907876759767, + "grad_norm": 0.066949762403965, + "learning_rate": 1.0237308398428014e-06, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26630 + }, + { + "epoch": 7.750945592086122, + "flow_matching_loss": 0.013053695484995843, + "grad_norm": 0.08257106691598892, + "learning_rate": 1.0177356301337925e-06, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26640 + }, + { + "epoch": 7.753855106197265, + "flow_matching_loss": 0.013009652495384216, + "grad_norm": 0.08074992150068283, + "learning_rate": 1.0117574101944348e-06, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26650 + }, + { + "epoch": 7.756764620308409, + "flow_matching_loss": 0.014509277231991292, + "grad_norm": 0.08577080070972443, + "learning_rate": 1.0057961872888293e-06, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26660 + }, + { + "epoch": 7.7596741344195515, + "flow_matching_loss": 0.014555423986166716, + "grad_norm": 0.07558207213878632, + "learning_rate": 9.998519686604235e-07, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26670 + }, + { + "epoch": 7.762583648530695, + "flow_matching_loss": 0.013402850087732076, + "grad_norm": 0.07353921234607697, + "learning_rate": 9.939247615320047e-07, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26680 + }, + { + "epoch": 7.765493162641839, + "flow_matching_loss": 0.01379830427467823, + "grad_norm": 0.08779197186231613, + "learning_rate": 9.880145731056834e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26690 + }, + { + "epoch": 7.768402676752983, + "flow_matching_loss": 0.01229699058458209, + "grad_norm": 0.07608925551176071, + "learning_rate": 9.821214105629034e-07, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26700 + }, + { + "epoch": 7.771312190864125, + "flow_matching_loss": 0.013499239087104797, + "grad_norm": 0.06334824860095978, + "learning_rate": 9.762452810644068e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26710 + }, + { + "epoch": 7.774221704975269, + "flow_matching_loss": 0.013044402468949556, + "grad_norm": 0.06291113048791885, + "learning_rate": 9.703861917502487e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26720 + }, + { + "epoch": 7.777131219086413, + "flow_matching_loss": 0.012826426979154348, + "grad_norm": 0.06801430881023407, + "learning_rate": 9.645441497397744e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26730 + }, + { + "epoch": 7.780040733197556, + "flow_matching_loss": 0.01307748444378376, + "grad_norm": 0.05333932489156723, + "learning_rate": 9.587191621316154e-07, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26740 + }, + { + "epoch": 7.782950247308699, + "flow_matching_loss": 0.012403632514178753, + "grad_norm": 0.07605002820491791, + "learning_rate": 9.529112360036802e-07, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26750 + }, + { + "epoch": 7.785859761419843, + "flow_matching_loss": 0.013019464071840047, + "grad_norm": 0.07193410396575928, + "learning_rate": 9.471203784131449e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26760 + }, + { + "epoch": 7.788769275530987, + "flow_matching_loss": 0.012564624845981597, + "grad_norm": 0.06779354065656662, + "learning_rate": 9.41346596396453e-07, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26770 + }, + { + "epoch": 7.79167878964213, + "flow_matching_loss": 0.013998975232243537, + "grad_norm": 0.09976626187562943, + "learning_rate": 9.355898969692889e-07, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26780 + }, + { + "epoch": 7.794588303753273, + "flow_matching_loss": 0.013638800103217363, + "grad_norm": 0.056017231196165085, + "learning_rate": 9.29850287126589e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26790 + }, + { + "epoch": 7.797497817864417, + "flow_matching_loss": 0.013757371623069049, + "grad_norm": 0.07984006404876709, + "learning_rate": 9.24127773842518e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26800 + }, + { + "epoch": 7.80040733197556, + "flow_matching_loss": 0.012899720203131437, + "grad_norm": 0.06279243528842926, + "learning_rate": 9.18422364070467e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26810 + }, + { + "epoch": 7.8033168460867035, + "flow_matching_loss": 0.013541064783930778, + "grad_norm": 0.06632053852081299, + "learning_rate": 9.127340647430538e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26820 + }, + { + "epoch": 7.806226360197847, + "flow_matching_loss": 0.012200635857880116, + "grad_norm": 0.08267610520124435, + "learning_rate": 9.070628827720929e-07, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26830 + }, + { + "epoch": 7.80913587430899, + "flow_matching_loss": 0.013649710267782212, + "grad_norm": 0.07264820486307144, + "learning_rate": 9.014088250486052e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26840 + }, + { + "epoch": 7.812045388420134, + "flow_matching_loss": 0.0123151701875031, + "grad_norm": 0.05777278169989586, + "learning_rate": 8.95771898442807e-07, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26850 + }, + { + "epoch": 7.8149549025312774, + "flow_matching_loss": 0.012649506609886885, + "grad_norm": 0.07255353033542633, + "learning_rate": 8.90152109804091e-07, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26860 + }, + { + "epoch": 7.81786441664242, + "flow_matching_loss": 0.012939517945051193, + "grad_norm": 0.07000676542520523, + "learning_rate": 8.845494659610365e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26870 + }, + { + "epoch": 7.820773930753564, + "flow_matching_loss": 0.012210354488343001, + "grad_norm": 0.05871880054473877, + "learning_rate": 8.789639737213812e-07, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26880 + }, + { + "epoch": 7.823683444864708, + "flow_matching_loss": 0.014339123386889696, + "grad_norm": 0.08111872524023056, + "learning_rate": 8.733956398720244e-07, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26890 + }, + { + "epoch": 7.826592958975851, + "flow_matching_loss": 0.011976827308535576, + "grad_norm": 0.05513317510485649, + "learning_rate": 8.678444711790157e-07, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26900 + }, + { + "epoch": 7.829502473086994, + "flow_matching_loss": 0.013533831108361483, + "grad_norm": 0.07467497885227203, + "learning_rate": 8.623104743875493e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26910 + }, + { + "epoch": 7.832411987198138, + "flow_matching_loss": 0.011979317851364612, + "grad_norm": 0.06564397364854813, + "learning_rate": 8.567936562219547e-07, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26920 + }, + { + "epoch": 7.835321501309282, + "flow_matching_loss": 0.012757146079093219, + "grad_norm": 0.053038664162158966, + "learning_rate": 8.512940233856847e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26930 + }, + { + "epoch": 7.838231015420424, + "flow_matching_loss": 0.013783146627247334, + "grad_norm": 0.07722607254981995, + "learning_rate": 8.458115825613122e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26940 + }, + { + "epoch": 7.841140529531568, + "flow_matching_loss": 0.013595539052039385, + "grad_norm": 0.06695393472909927, + "learning_rate": 8.403463404105199e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26950 + }, + { + "epoch": 7.844050043642712, + "flow_matching_loss": 0.014091470744460821, + "grad_norm": 0.07290095835924149, + "learning_rate": 8.348983035740926e-07, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26960 + }, + { + "epoch": 7.846959557753856, + "flow_matching_loss": 0.014037591684609652, + "grad_norm": 0.06500615924596786, + "learning_rate": 8.294674786719064e-07, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26970 + }, + { + "epoch": 7.849869071864998, + "flow_matching_loss": 0.012481986451894045, + "grad_norm": 0.06599623709917068, + "learning_rate": 8.240538723029295e-07, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26980 + }, + { + "epoch": 7.852778585976142, + "flow_matching_loss": 0.014356099255383014, + "grad_norm": 0.09089325368404388, + "learning_rate": 8.186574910451983e-07, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 26990 + }, + { + "epoch": 7.855688100087286, + "flow_matching_loss": 0.013763805478811264, + "grad_norm": 0.075838603079319, + "learning_rate": 8.132783414558326e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27000 + }, + { + "epoch": 7.858597614198429, + "flow_matching_loss": 0.013027012906968593, + "grad_norm": 0.049189064651727676, + "learning_rate": 8.079164300710023e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27010 + }, + { + "epoch": 7.861507128309572, + "flow_matching_loss": 0.013405296113342046, + "grad_norm": 0.0826864242553711, + "learning_rate": 8.025717634059326e-07, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27020 + }, + { + "epoch": 7.864416642420716, + "flow_matching_loss": 0.012853532284498214, + "grad_norm": 0.05150274559855461, + "learning_rate": 7.972443479549035e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27030 + }, + { + "epoch": 7.867326156531859, + "flow_matching_loss": 0.013568164687603712, + "grad_norm": 0.07305805385112762, + "learning_rate": 7.919341901912236e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27040 + }, + { + "epoch": 7.870235670643003, + "flow_matching_loss": 0.013364016264677047, + "grad_norm": 0.0721023827791214, + "learning_rate": 7.866412965672365e-07, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27050 + }, + { + "epoch": 7.873145184754146, + "flow_matching_loss": 0.013341550808399915, + "grad_norm": 0.060213446617126465, + "learning_rate": 7.813656735143044e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27060 + }, + { + "epoch": 7.876054698865289, + "flow_matching_loss": 0.01575254611670971, + "grad_norm": 0.07499613612890244, + "learning_rate": 7.761073274428076e-07, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27070 + }, + { + "epoch": 7.878964212976433, + "flow_matching_loss": 0.012500405311584473, + "grad_norm": 0.08601849526166916, + "learning_rate": 7.708662647421355e-07, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27080 + }, + { + "epoch": 7.8818737270875765, + "flow_matching_loss": 0.01380010200664401, + "grad_norm": 0.07393507659435272, + "learning_rate": 7.65642491780672e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27090 + }, + { + "epoch": 7.88478324119872, + "flow_matching_loss": 0.012014451529830694, + "grad_norm": 0.05834844335913658, + "learning_rate": 7.604360149057916e-07, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27100 + }, + { + "epoch": 7.887692755309863, + "flow_matching_loss": 0.012908423319458961, + "grad_norm": 0.06702522933483124, + "learning_rate": 7.552468404438568e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27110 + }, + { + "epoch": 7.890602269421007, + "flow_matching_loss": 0.012388237752020359, + "grad_norm": 0.0785914808511734, + "learning_rate": 7.500749747002022e-07, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27120 + }, + { + "epoch": 7.89351178353215, + "flow_matching_loss": 0.01442179474979639, + "grad_norm": 0.054776839911937714, + "learning_rate": 7.449204239591345e-07, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27130 + }, + { + "epoch": 7.896421297643293, + "flow_matching_loss": 0.012719681765884162, + "grad_norm": 0.06031060591340065, + "learning_rate": 7.397831944839167e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27140 + }, + { + "epoch": 7.899330811754437, + "flow_matching_loss": 0.012541459035128355, + "grad_norm": 0.06243489682674408, + "learning_rate": 7.346632925167706e-07, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27150 + }, + { + "epoch": 7.902240325865581, + "flow_matching_loss": 0.012690915912389755, + "grad_norm": 0.07629349082708359, + "learning_rate": 7.295607242788566e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27160 + }, + { + "epoch": 7.905149839976724, + "flow_matching_loss": 0.014282960165292024, + "grad_norm": 0.07106144726276398, + "learning_rate": 7.244754959702765e-07, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27170 + }, + { + "epoch": 7.908059354087867, + "flow_matching_loss": 0.014086223766207695, + "grad_norm": 0.0932222232222557, + "learning_rate": 7.194076137700628e-07, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27180 + }, + { + "epoch": 7.910968868199011, + "flow_matching_loss": 0.012331743724644184, + "grad_norm": 0.08005638420581818, + "learning_rate": 7.143570838361696e-07, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27190 + }, + { + "epoch": 7.913878382310154, + "flow_matching_loss": 0.012016741465777159, + "grad_norm": 0.07220957428216934, + "learning_rate": 7.093239123054696e-07, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27200 + }, + { + "epoch": 7.916787896421297, + "flow_matching_loss": 0.012985778972506523, + "grad_norm": 0.05832774192094803, + "learning_rate": 7.043081052937389e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27210 + }, + { + "epoch": 7.919697410532441, + "flow_matching_loss": 0.012774398364126683, + "grad_norm": 0.060998059809207916, + "learning_rate": 6.993096688956558e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27220 + }, + { + "epoch": 7.922606924643585, + "flow_matching_loss": 0.014057501126080752, + "grad_norm": 0.07031319290399551, + "learning_rate": 6.943286091847884e-07, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27230 + }, + { + "epoch": 7.925516438754728, + "flow_matching_loss": 0.013242710009217262, + "grad_norm": 0.08141982555389404, + "learning_rate": 6.89364932213602e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27240 + }, + { + "epoch": 7.928425952865871, + "flow_matching_loss": 0.0133820915594697, + "grad_norm": 0.06384304910898209, + "learning_rate": 6.844186440134253e-07, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27250 + }, + { + "epoch": 7.931335466977015, + "flow_matching_loss": 0.012129717227071523, + "grad_norm": 0.06116066873073578, + "learning_rate": 6.794897505944692e-07, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27260 + }, + { + "epoch": 7.934244981088158, + "flow_matching_loss": 0.011830867361277341, + "grad_norm": 0.06897369772195816, + "learning_rate": 6.745782579457993e-07, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27270 + }, + { + "epoch": 7.937154495199302, + "flow_matching_loss": 0.013722779881209134, + "grad_norm": 0.10858043283224106, + "learning_rate": 6.696841720353453e-07, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27280 + }, + { + "epoch": 7.940064009310445, + "flow_matching_loss": 0.012674851343035698, + "grad_norm": 0.06284525990486145, + "learning_rate": 6.648074988098834e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27290 + }, + { + "epoch": 7.942973523421589, + "flow_matching_loss": 0.012746625114232302, + "grad_norm": 0.07707797735929489, + "learning_rate": 6.599482441950322e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27300 + }, + { + "epoch": 7.945883037532732, + "flow_matching_loss": 0.014024155493825675, + "grad_norm": 0.08320318162441254, + "learning_rate": 6.551064140952417e-07, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27310 + }, + { + "epoch": 7.9487925516438755, + "flow_matching_loss": 0.012999731954187155, + "grad_norm": 0.08398561179637909, + "learning_rate": 6.502820143937943e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27320 + }, + { + "epoch": 7.951702065755019, + "flow_matching_loss": 0.01379833435639739, + "grad_norm": 0.09337179362773895, + "learning_rate": 6.454750509527874e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27330 + }, + { + "epoch": 7.954611579866162, + "flow_matching_loss": 0.012578693963587285, + "grad_norm": 0.07512868940830231, + "learning_rate": 6.406855296131387e-07, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27340 + }, + { + "epoch": 7.957521093977306, + "flow_matching_loss": 0.013397112768143416, + "grad_norm": 0.06521967053413391, + "learning_rate": 6.359134561945662e-07, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27350 + }, + { + "epoch": 7.9604306080884495, + "flow_matching_loss": 0.012308290787041187, + "grad_norm": 0.0625908225774765, + "learning_rate": 6.311588364955883e-07, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27360 + }, + { + "epoch": 7.963340122199592, + "flow_matching_loss": 0.012749652285128832, + "grad_norm": 0.10379615426063538, + "learning_rate": 6.264216762935187e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27370 + }, + { + "epoch": 7.966249636310736, + "flow_matching_loss": 0.013245607819408178, + "grad_norm": 0.0646478459239006, + "learning_rate": 6.21701981344453e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27380 + }, + { + "epoch": 7.96915915042188, + "flow_matching_loss": 0.01437405776232481, + "grad_norm": 0.06035379692912102, + "learning_rate": 6.16999757383262e-07, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27390 + }, + { + "epoch": 7.9720686645330225, + "flow_matching_loss": 0.011757573764771222, + "grad_norm": 0.05458332970738411, + "learning_rate": 6.123150101235952e-07, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27400 + }, + { + "epoch": 7.974978178644166, + "flow_matching_loss": 0.013780077267438174, + "grad_norm": 0.09512313455343246, + "learning_rate": 6.076477452578594e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27410 + }, + { + "epoch": 7.97788769275531, + "flow_matching_loss": 0.014018448349088431, + "grad_norm": 0.12468426674604416, + "learning_rate": 6.029979684572262e-07, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27420 + }, + { + "epoch": 7.980797206866454, + "flow_matching_loss": 0.012163915066048503, + "grad_norm": 0.06791767477989197, + "learning_rate": 5.983656853716097e-07, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27430 + }, + { + "epoch": 7.9837067209775965, + "flow_matching_loss": 0.013701032474637031, + "grad_norm": 0.07600302994251251, + "learning_rate": 5.937509016296689e-07, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27440 + }, + { + "epoch": 7.98661623508874, + "flow_matching_loss": 0.013756174128502607, + "grad_norm": 0.08397292345762253, + "learning_rate": 5.89153622838805e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27450 + }, + { + "epoch": 7.989525749199884, + "flow_matching_loss": 0.012431882182136178, + "grad_norm": 0.04643155634403229, + "learning_rate": 5.845738545851431e-07, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27460 + }, + { + "epoch": 7.992435263311027, + "flow_matching_loss": 0.013994931150227785, + "grad_norm": 0.159210667014122, + "learning_rate": 5.800116024335334e-07, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27470 + }, + { + "epoch": 7.99534477742217, + "flow_matching_loss": 0.013380531407892703, + "grad_norm": 0.13695304095745087, + "learning_rate": 5.75466871927543e-07, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27480 + }, + { + "epoch": 7.998254291533314, + "flow_matching_loss": 0.012347315438091755, + "grad_norm": 0.09150289744138718, + "learning_rate": 5.709396685894463e-07, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27490 + }, + { + "epoch": 8.001163805644458, + "flow_matching_loss": 0.011247481219470501, + "grad_norm": 0.06036624684929848, + "learning_rate": 5.664299979202264e-07, + "loss": 0.0112, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27500 + }, + { + "epoch": 8.004073319755602, + "flow_matching_loss": 0.014287814404815436, + "grad_norm": 0.09041581302881241, + "learning_rate": 5.61937865399555e-07, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27510 + }, + { + "epoch": 8.006982833866743, + "flow_matching_loss": 0.013306626677513122, + "grad_norm": 0.05442555621266365, + "learning_rate": 5.574632764857974e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27520 + }, + { + "epoch": 8.009892347977887, + "flow_matching_loss": 0.013579746801406145, + "grad_norm": 0.060288604348897934, + "learning_rate": 5.53006236616001e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27530 + }, + { + "epoch": 8.012801862089031, + "flow_matching_loss": 0.013208312448114157, + "grad_norm": 0.08334948122501373, + "learning_rate": 5.485667512058873e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27540 + }, + { + "epoch": 8.015711376200175, + "flow_matching_loss": 0.013757292460650206, + "grad_norm": 0.06211887672543526, + "learning_rate": 5.441448256498544e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27550 + }, + { + "epoch": 8.018620890311318, + "flow_matching_loss": 0.01215092008933425, + "grad_norm": 0.07220452278852463, + "learning_rate": 5.397404653209559e-07, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27560 + }, + { + "epoch": 8.021530404422462, + "flow_matching_loss": 0.014286754187196493, + "grad_norm": 0.06270579248666763, + "learning_rate": 5.35353675570906e-07, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27570 + }, + { + "epoch": 8.024439918533606, + "flow_matching_loss": 0.013196620531380177, + "grad_norm": 0.08135437965393066, + "learning_rate": 5.309844617300674e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27580 + }, + { + "epoch": 8.027349432644748, + "flow_matching_loss": 0.012759868148714304, + "grad_norm": 0.06532192975282669, + "learning_rate": 5.266328291074468e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27590 + }, + { + "epoch": 8.030258946755891, + "flow_matching_loss": 0.014493829011917115, + "grad_norm": 0.06448081135749817, + "learning_rate": 5.222987829906861e-07, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27600 + }, + { + "epoch": 8.033168460867035, + "flow_matching_loss": 0.01303670760244131, + "grad_norm": 0.08315631002187729, + "learning_rate": 5.179823286460644e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27610 + }, + { + "epoch": 8.036077974978179, + "flow_matching_loss": 0.0136534390039742, + "grad_norm": 0.05826244875788689, + "learning_rate": 5.13683471318478e-07, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27620 + }, + { + "epoch": 8.038987489089322, + "flow_matching_loss": 0.012222288642078638, + "grad_norm": 0.062348540872335434, + "learning_rate": 5.094022162314449e-07, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27630 + }, + { + "epoch": 8.041897003200466, + "flow_matching_loss": 0.011667531635612249, + "grad_norm": 0.07182899862527847, + "learning_rate": 5.051385685870924e-07, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27640 + }, + { + "epoch": 8.044806517311608, + "flow_matching_loss": 0.013541365135461092, + "grad_norm": 0.11641152203083038, + "learning_rate": 5.008925335661546e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27650 + }, + { + "epoch": 8.047716031422752, + "flow_matching_loss": 0.014329439681023359, + "grad_norm": 0.07533407211303711, + "learning_rate": 4.966641163279662e-07, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27660 + }, + { + "epoch": 8.050625545533896, + "flow_matching_loss": 0.012490479741245508, + "grad_norm": 0.07190994173288345, + "learning_rate": 4.924533220104521e-07, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27670 + }, + { + "epoch": 8.05353505964504, + "flow_matching_loss": 0.013500718865543604, + "grad_norm": 0.09837368875741959, + "learning_rate": 4.882601557301231e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27680 + }, + { + "epoch": 8.056444573756183, + "flow_matching_loss": 0.013176156487315894, + "grad_norm": 0.06590796262025833, + "learning_rate": 4.840846225820717e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27690 + }, + { + "epoch": 8.059354087867327, + "flow_matching_loss": 0.014289868623018264, + "grad_norm": 0.062190551310777664, + "learning_rate": 4.79926727639966e-07, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27700 + }, + { + "epoch": 8.06226360197847, + "flow_matching_loss": 0.014840609300881625, + "grad_norm": 0.06881009042263031, + "learning_rate": 4.7578647595603817e-07, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27710 + }, + { + "epoch": 8.065173116089612, + "flow_matching_loss": 0.012188779655843973, + "grad_norm": 0.06966244429349899, + "learning_rate": 4.716638725610889e-07, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27720 + }, + { + "epoch": 8.068082630200756, + "flow_matching_loss": 0.012194887548685075, + "grad_norm": 0.07751472294330597, + "learning_rate": 4.6755892246446606e-07, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27730 + }, + { + "epoch": 8.0709921443119, + "flow_matching_loss": 0.013309936411678791, + "grad_norm": 0.0669393464922905, + "learning_rate": 4.634716306540715e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27740 + }, + { + "epoch": 8.073901658423043, + "flow_matching_loss": 0.013218438532203437, + "grad_norm": 0.07286029309034348, + "learning_rate": 4.5940200209635244e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27750 + }, + { + "epoch": 8.076811172534187, + "flow_matching_loss": 0.013950783666223288, + "grad_norm": 0.07376687973737717, + "learning_rate": 4.5535004173628657e-07, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27760 + }, + { + "epoch": 8.07972068664533, + "flow_matching_loss": 0.013260223343968392, + "grad_norm": 0.054791126400232315, + "learning_rate": 4.5131575449739225e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27770 + }, + { + "epoch": 8.082630200756475, + "flow_matching_loss": 0.012895369436591864, + "grad_norm": 0.08129460364580154, + "learning_rate": 4.472991452817099e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27780 + }, + { + "epoch": 8.085539714867616, + "flow_matching_loss": 0.013583429250866174, + "grad_norm": 0.09490743279457092, + "learning_rate": 4.433002189697938e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27790 + }, + { + "epoch": 8.08844922897876, + "flow_matching_loss": 0.013195159751921893, + "grad_norm": 0.11025794595479965, + "learning_rate": 4.393189804207187e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27800 + }, + { + "epoch": 8.091358743089904, + "flow_matching_loss": 0.01426175907254219, + "grad_norm": 0.07033360004425049, + "learning_rate": 4.353554344720651e-07, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27810 + }, + { + "epoch": 8.094268257201048, + "flow_matching_loss": 0.014666618779301643, + "grad_norm": 0.08570201694965363, + "learning_rate": 4.314095859399153e-07, + "loss": 0.0147, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27820 + }, + { + "epoch": 8.097177771312191, + "flow_matching_loss": 0.013049312587827443, + "grad_norm": 0.081409752368927, + "learning_rate": 4.274814396188459e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27830 + }, + { + "epoch": 8.100087285423335, + "flow_matching_loss": 0.013399397116154432, + "grad_norm": 0.06143257021903992, + "learning_rate": 4.235710002819271e-07, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27840 + }, + { + "epoch": 8.102996799534477, + "flow_matching_loss": 0.0132149837911129, + "grad_norm": 0.0723227709531784, + "learning_rate": 4.196782726807097e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27850 + }, + { + "epoch": 8.10590631364562, + "flow_matching_loss": 0.01321325832977891, + "grad_norm": 0.06741347908973694, + "learning_rate": 4.158032615452251e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27860 + }, + { + "epoch": 8.108815827756764, + "flow_matching_loss": 0.014370756037533284, + "grad_norm": 0.07095623016357422, + "learning_rate": 4.1194597158397853e-07, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27870 + }, + { + "epoch": 8.111725341867908, + "flow_matching_loss": 0.01309375986456871, + "grad_norm": 0.06358751654624939, + "learning_rate": 4.081064074839408e-07, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27880 + }, + { + "epoch": 8.114634855979052, + "flow_matching_loss": 0.013238781783729791, + "grad_norm": 0.0912591814994812, + "learning_rate": 4.042845739105433e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27890 + }, + { + "epoch": 8.117544370090195, + "flow_matching_loss": 0.01322575779631734, + "grad_norm": 0.10157089680433273, + "learning_rate": 4.004804755076746e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27900 + }, + { + "epoch": 8.12045388420134, + "flow_matching_loss": 0.012239594385027886, + "grad_norm": 0.06231558322906494, + "learning_rate": 3.9669411689767063e-07, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27910 + }, + { + "epoch": 8.123363398312481, + "flow_matching_loss": 0.012850325275212526, + "grad_norm": 0.0690871998667717, + "learning_rate": 3.929255026813194e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27920 + }, + { + "epoch": 8.126272912423625, + "flow_matching_loss": 0.012701636180281639, + "grad_norm": 0.06675399094820023, + "learning_rate": 3.8917463743783797e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27930 + }, + { + "epoch": 8.129182426534769, + "flow_matching_loss": 0.013564869109541178, + "grad_norm": 0.06537852436304092, + "learning_rate": 3.854415257248839e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27940 + }, + { + "epoch": 8.132091940645912, + "flow_matching_loss": 0.011868537776172161, + "grad_norm": 0.06173308193683624, + "learning_rate": 3.8172617207853864e-07, + "loss": 0.0119, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27950 + }, + { + "epoch": 8.135001454757056, + "flow_matching_loss": 0.013012092094868422, + "grad_norm": 0.07992628216743469, + "learning_rate": 3.7802858101330773e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27960 + }, + { + "epoch": 8.1379109688682, + "flow_matching_loss": 0.013255163934081793, + "grad_norm": 0.09058728814125061, + "learning_rate": 3.743487570221121e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27970 + }, + { + "epoch": 8.140820482979342, + "flow_matching_loss": 0.0133405446074903, + "grad_norm": 0.09840809553861618, + "learning_rate": 3.7068670457628517e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27980 + }, + { + "epoch": 8.143729997090485, + "flow_matching_loss": 0.013507729209959508, + "grad_norm": 0.06600094586610794, + "learning_rate": 3.6704242812556576e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 27990 + }, + { + "epoch": 8.146639511201629, + "flow_matching_loss": 0.013511857390403748, + "grad_norm": 0.09852014482021332, + "learning_rate": 3.634159320980951e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28000 + }, + { + "epoch": 8.149549025312773, + "flow_matching_loss": 0.013250138610601425, + "grad_norm": 0.0698489248752594, + "learning_rate": 3.5980722090040663e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28010 + }, + { + "epoch": 8.152458539423916, + "flow_matching_loss": 0.012766918633133173, + "grad_norm": 0.07029476016759872, + "learning_rate": 3.5621629891742447e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28020 + }, + { + "epoch": 8.15536805353506, + "flow_matching_loss": 0.014916585851460695, + "grad_norm": 0.07000543177127838, + "learning_rate": 3.526431705124583e-07, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28030 + }, + { + "epoch": 8.158277567646204, + "flow_matching_loss": 0.015137170441448689, + "grad_norm": 0.08551294356584549, + "learning_rate": 3.4908784002719517e-07, + "loss": 0.0151, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28040 + }, + { + "epoch": 8.161187081757346, + "flow_matching_loss": 0.01362446192651987, + "grad_norm": 0.09600923210382462, + "learning_rate": 3.4555031178169773e-07, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28050 + }, + { + "epoch": 8.16409659586849, + "flow_matching_loss": 0.014271253161132336, + "grad_norm": 0.08495666086673737, + "learning_rate": 3.420305900743942e-07, + "loss": 0.0143, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28060 + }, + { + "epoch": 8.167006109979633, + "flow_matching_loss": 0.013525026664137841, + "grad_norm": 0.06384831666946411, + "learning_rate": 3.385286791820802e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28070 + }, + { + "epoch": 8.169915624090777, + "flow_matching_loss": 0.014092251379042863, + "grad_norm": 0.08102622628211975, + "learning_rate": 3.350445833599086e-07, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28080 + }, + { + "epoch": 8.17282513820192, + "flow_matching_loss": 0.011740511283278465, + "grad_norm": 0.06730016320943832, + "learning_rate": 3.3157830684138303e-07, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28090 + }, + { + "epoch": 8.175734652313064, + "flow_matching_loss": 0.013650993071496486, + "grad_norm": 0.068842813372612, + "learning_rate": 3.281298538383559e-07, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28100 + }, + { + "epoch": 8.178644166424208, + "flow_matching_loss": 0.013662446476519107, + "grad_norm": 0.07014291733503342, + "learning_rate": 3.2469922854102387e-07, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28110 + }, + { + "epoch": 8.18155368053535, + "flow_matching_loss": 0.014812625106424093, + "grad_norm": 0.0725192129611969, + "learning_rate": 3.21286435117919e-07, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28120 + }, + { + "epoch": 8.184463194646494, + "flow_matching_loss": 0.012725951708853244, + "grad_norm": 0.06834663450717926, + "learning_rate": 3.1789147771590754e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28130 + }, + { + "epoch": 8.187372708757637, + "flow_matching_loss": 0.014200031571090222, + "grad_norm": 0.04834742099046707, + "learning_rate": 3.14514360460183e-07, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28140 + }, + { + "epoch": 8.190282222868781, + "flow_matching_loss": 0.012957079987972974, + "grad_norm": 0.06974779814481735, + "learning_rate": 3.1115508745426293e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28150 + }, + { + "epoch": 8.193191736979925, + "flow_matching_loss": 0.013867418933659791, + "grad_norm": 0.07879640907049179, + "learning_rate": 3.0781366277997716e-07, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28160 + }, + { + "epoch": 8.196101251091068, + "flow_matching_loss": 0.014645668771117925, + "grad_norm": 0.05560896545648575, + "learning_rate": 3.044900904974762e-07, + "loss": 0.0146, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28170 + }, + { + "epoch": 8.19901076520221, + "flow_matching_loss": 0.012946467380970716, + "grad_norm": 0.07002051919698715, + "learning_rate": 3.011843746452081e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28180 + }, + { + "epoch": 8.201920279313354, + "flow_matching_loss": 0.015496170893311501, + "grad_norm": 0.06698726117610931, + "learning_rate": 2.9789651923993464e-07, + "loss": 0.0155, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28190 + }, + { + "epoch": 8.204829793424498, + "flow_matching_loss": 0.011993461847305298, + "grad_norm": 0.060356732457876205, + "learning_rate": 2.9462652827670855e-07, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28200 + }, + { + "epoch": 8.207739307535642, + "flow_matching_loss": 0.013424586039036512, + "grad_norm": 0.08645280450582504, + "learning_rate": 2.913744057288764e-07, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28210 + }, + { + "epoch": 8.210648821646785, + "flow_matching_loss": 0.014868453424423934, + "grad_norm": 0.09956724941730499, + "learning_rate": 2.8814015554807396e-07, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28220 + }, + { + "epoch": 8.213558335757929, + "flow_matching_loss": 0.013203751854598522, + "grad_norm": 0.05462333932518959, + "learning_rate": 2.849237816642192e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28230 + }, + { + "epoch": 8.216467849869073, + "flow_matching_loss": 0.01282372111454606, + "grad_norm": 0.06800107657909393, + "learning_rate": 2.817252879855109e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28240 + }, + { + "epoch": 8.219377363980215, + "flow_matching_loss": 0.012368787359446288, + "grad_norm": 0.07386172562837601, + "learning_rate": 2.785446783984219e-07, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28250 + }, + { + "epoch": 8.222286878091358, + "flow_matching_loss": 0.011245414894074202, + "grad_norm": 0.06650472432374954, + "learning_rate": 2.7538195676769076e-07, + "loss": 0.0112, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28260 + }, + { + "epoch": 8.225196392202502, + "flow_matching_loss": 0.013842509966343642, + "grad_norm": 0.06630755960941315, + "learning_rate": 2.722371269363216e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28270 + }, + { + "epoch": 8.228105906313646, + "flow_matching_loss": 0.013804655801504851, + "grad_norm": 0.06693244725465775, + "learning_rate": 2.6911019272558125e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28280 + }, + { + "epoch": 8.23101542042479, + "flow_matching_loss": 0.011832439806312323, + "grad_norm": 0.06921738386154175, + "learning_rate": 2.6600115793498857e-07, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28290 + }, + { + "epoch": 8.233924934535933, + "flow_matching_loss": 0.013825554214417935, + "grad_norm": 0.09809104353189468, + "learning_rate": 2.629100263423151e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28300 + }, + { + "epoch": 8.236834448647077, + "flow_matching_loss": 0.011818017717450858, + "grad_norm": 0.06586253643035889, + "learning_rate": 2.598368017035763e-07, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28310 + }, + { + "epoch": 8.239743962758219, + "flow_matching_loss": 0.013971927668899297, + "grad_norm": 0.07034020125865936, + "learning_rate": 2.5678148775303e-07, + "loss": 0.014, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28320 + }, + { + "epoch": 8.242653476869362, + "flow_matching_loss": 0.012823088653385639, + "grad_norm": 0.08925905078649521, + "learning_rate": 2.537440882031683e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28330 + }, + { + "epoch": 8.245562990980506, + "flow_matching_loss": 0.013918464630842209, + "grad_norm": 0.0687389001250267, + "learning_rate": 2.5072460674472385e-07, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28340 + }, + { + "epoch": 8.24847250509165, + "flow_matching_loss": 0.011945240199565887, + "grad_norm": 0.0728069618344307, + "learning_rate": 2.4772304704664683e-07, + "loss": 0.0119, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28350 + }, + { + "epoch": 8.251382019202794, + "flow_matching_loss": 0.013036716263741255, + "grad_norm": 0.08773738890886307, + "learning_rate": 2.447394127561181e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28360 + }, + { + "epoch": 8.254291533313937, + "flow_matching_loss": 0.013783159758895635, + "grad_norm": 0.08011390268802643, + "learning_rate": 2.41773707498531e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28370 + }, + { + "epoch": 8.25720104742508, + "flow_matching_loss": 0.01283069523051381, + "grad_norm": 0.07038161903619766, + "learning_rate": 2.388259348774996e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28380 + }, + { + "epoch": 8.260110561536223, + "flow_matching_loss": 0.013156685512512922, + "grad_norm": 0.07314583659172058, + "learning_rate": 2.358960984748454e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28390 + }, + { + "epoch": 8.263020075647367, + "flow_matching_loss": 0.011981046013534068, + "grad_norm": 0.052439942955970764, + "learning_rate": 2.3298420185059565e-07, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28400 + }, + { + "epoch": 8.26592958975851, + "flow_matching_loss": 0.01374374283477664, + "grad_norm": 0.07110939174890518, + "learning_rate": 2.3009024854297844e-07, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28410 + }, + { + "epoch": 8.268839103869654, + "flow_matching_loss": 0.012233400251716375, + "grad_norm": 0.07109638303518295, + "learning_rate": 2.272142420684209e-07, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28420 + }, + { + "epoch": 8.271748617980798, + "flow_matching_loss": 0.014241647440940142, + "grad_norm": 0.08619463443756104, + "learning_rate": 2.243561859215426e-07, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28430 + }, + { + "epoch": 8.274658132091941, + "flow_matching_loss": 0.013210100494325161, + "grad_norm": 0.07044722139835358, + "learning_rate": 2.2151608357514897e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28440 + }, + { + "epoch": 8.277567646203083, + "flow_matching_loss": 0.01275319354608655, + "grad_norm": 0.06679011881351471, + "learning_rate": 2.186939384802361e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28450 + }, + { + "epoch": 8.280477160314227, + "flow_matching_loss": 0.013073664624243975, + "grad_norm": 0.07854233682155609, + "learning_rate": 2.1588975406597266e-07, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28460 + }, + { + "epoch": 8.28338667442537, + "flow_matching_loss": 0.012849995028227567, + "grad_norm": 0.0702224150300026, + "learning_rate": 2.131035337397097e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28470 + }, + { + "epoch": 8.286296188536515, + "flow_matching_loss": 0.014056440815329552, + "grad_norm": 0.07503967732191086, + "learning_rate": 2.1033528088696575e-07, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28480 + }, + { + "epoch": 8.289205702647658, + "flow_matching_loss": 0.012905588932335376, + "grad_norm": 0.08797723054885864, + "learning_rate": 2.0758499887143012e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28490 + }, + { + "epoch": 8.292115216758802, + "flow_matching_loss": 0.013185286894440651, + "grad_norm": 0.058013152331113815, + "learning_rate": 2.0485269103495462e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28500 + }, + { + "epoch": 8.295024730869944, + "flow_matching_loss": 0.012942140456289053, + "grad_norm": 0.06967217475175858, + "learning_rate": 2.021383606975502e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28510 + }, + { + "epoch": 8.297934244981088, + "flow_matching_loss": 0.012433741427958011, + "grad_norm": 0.06539319455623627, + "learning_rate": 1.9944201115738524e-07, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28520 + }, + { + "epoch": 8.300843759092231, + "flow_matching_loss": 0.012808733433485032, + "grad_norm": 0.06248525157570839, + "learning_rate": 1.9676364569077732e-07, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28530 + }, + { + "epoch": 8.303753273203375, + "flow_matching_loss": 0.013118397351354361, + "grad_norm": 0.07272321730852127, + "learning_rate": 1.9410326755219477e-07, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28540 + }, + { + "epoch": 8.306662787314519, + "flow_matching_loss": 0.012298578675836325, + "grad_norm": 0.09246067702770233, + "learning_rate": 1.9146087997424677e-07, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28550 + }, + { + "epoch": 8.309572301425662, + "flow_matching_loss": 0.013892250694334507, + "grad_norm": 0.07178626209497452, + "learning_rate": 1.8883648616768335e-07, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28560 + }, + { + "epoch": 8.312481815536806, + "flow_matching_loss": 0.013136189244687558, + "grad_norm": 0.056980378925800323, + "learning_rate": 1.862300893213903e-07, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28570 + }, + { + "epoch": 8.315391329647948, + "flow_matching_loss": 0.012258890829980374, + "grad_norm": 0.09933861345052719, + "learning_rate": 1.8364169260238595e-07, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28580 + }, + { + "epoch": 8.318300843759092, + "flow_matching_loss": 0.013935950305312871, + "grad_norm": 0.06888892501592636, + "learning_rate": 1.8107129915581777e-07, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28590 + }, + { + "epoch": 8.321210357870235, + "flow_matching_loss": 0.01257244972512126, + "grad_norm": 0.0822475254535675, + "learning_rate": 1.785189121049541e-07, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28600 + }, + { + "epoch": 8.32411987198138, + "flow_matching_loss": 0.013034752476960421, + "grad_norm": 0.07814375311136246, + "learning_rate": 1.7598453455118567e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28610 + }, + { + "epoch": 8.327029386092523, + "flow_matching_loss": 0.014508029818534851, + "grad_norm": 0.06789205223321915, + "learning_rate": 1.7346816957402256e-07, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28620 + }, + { + "epoch": 8.329938900203667, + "flow_matching_loss": 0.012659200560301542, + "grad_norm": 0.07516569644212723, + "learning_rate": 1.7096982023108389e-07, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28630 + }, + { + "epoch": 8.33284841431481, + "flow_matching_loss": 0.013029574695974589, + "grad_norm": 0.06447333842515945, + "learning_rate": 1.6848948955809963e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28640 + }, + { + "epoch": 8.335757928425952, + "flow_matching_loss": 0.013696052413433791, + "grad_norm": 0.07316471636295319, + "learning_rate": 1.66027180568904e-07, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28650 + }, + { + "epoch": 8.338667442537096, + "flow_matching_loss": 0.013545328937470912, + "grad_norm": 0.061644718050956726, + "learning_rate": 1.6358289625543865e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28660 + }, + { + "epoch": 8.34157695664824, + "flow_matching_loss": 0.012964595947414637, + "grad_norm": 0.06325668841600418, + "learning_rate": 1.6115663958773618e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28670 + }, + { + "epoch": 8.344486470759383, + "flow_matching_loss": 0.013764564879238605, + "grad_norm": 0.07233870029449463, + "learning_rate": 1.5874841351392832e-07, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28680 + }, + { + "epoch": 8.347395984870527, + "flow_matching_loss": 0.015887828823179007, + "grad_norm": 0.1032002866268158, + "learning_rate": 1.56358220960236e-07, + "loss": 0.0159, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28690 + }, + { + "epoch": 8.35030549898167, + "flow_matching_loss": 0.012551895808428526, + "grad_norm": 0.07248947769403458, + "learning_rate": 1.5398606483096932e-07, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28700 + }, + { + "epoch": 8.353215013092813, + "flow_matching_loss": 0.014529402181506158, + "grad_norm": 0.11099398881196976, + "learning_rate": 1.5163194800852265e-07, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28710 + }, + { + "epoch": 8.356124527203956, + "flow_matching_loss": 0.014822370372712612, + "grad_norm": 0.08988014608621597, + "learning_rate": 1.4929587335336614e-07, + "loss": 0.0148, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28720 + }, + { + "epoch": 8.3590340413151, + "flow_matching_loss": 0.012935156933963299, + "grad_norm": 0.07663433998823166, + "learning_rate": 1.4697784370405587e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28730 + }, + { + "epoch": 8.361943555426244, + "flow_matching_loss": 0.012955391500145196, + "grad_norm": 0.07711460441350937, + "learning_rate": 1.4467786187721376e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28740 + }, + { + "epoch": 8.364853069537387, + "flow_matching_loss": 0.012915009353309869, + "grad_norm": 0.06958425045013428, + "learning_rate": 1.4239593066753265e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28750 + }, + { + "epoch": 8.367762583648531, + "flow_matching_loss": 0.012425017822533847, + "grad_norm": 0.06907285749912262, + "learning_rate": 1.4013205284777953e-07, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28760 + }, + { + "epoch": 8.370672097759675, + "flow_matching_loss": 0.013170706480741501, + "grad_norm": 0.07378822565078735, + "learning_rate": 1.3788623116877397e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28770 + }, + { + "epoch": 8.373581611870817, + "flow_matching_loss": 0.012896844185888767, + "grad_norm": 0.06496565788984299, + "learning_rate": 1.3565846835940477e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28780 + }, + { + "epoch": 8.37649112598196, + "flow_matching_loss": 0.013529720809310674, + "grad_norm": 0.09612305462360382, + "learning_rate": 1.334487671266116e-07, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28790 + }, + { + "epoch": 8.379400640093104, + "flow_matching_loss": 0.012399725988507271, + "grad_norm": 0.07635997235774994, + "learning_rate": 1.3125713015538998e-07, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28800 + }, + { + "epoch": 8.382310154204248, + "flow_matching_loss": 0.01321950308047235, + "grad_norm": 0.07442307472229004, + "learning_rate": 1.290835601087864e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28810 + }, + { + "epoch": 8.385219668315392, + "flow_matching_loss": 0.012461153231561184, + "grad_norm": 0.06225230544805527, + "learning_rate": 1.2692805962789323e-07, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28820 + }, + { + "epoch": 8.388129182426535, + "flow_matching_loss": 0.01259643891826272, + "grad_norm": 0.057438671588897705, + "learning_rate": 1.2479063133184698e-07, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28830 + }, + { + "epoch": 8.391038696537677, + "flow_matching_loss": 0.012260766793042421, + "grad_norm": 0.07588067650794983, + "learning_rate": 1.2267127781782516e-07, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28840 + }, + { + "epoch": 8.393948210648821, + "flow_matching_loss": 0.013268992677330971, + "grad_norm": 0.09720174223184586, + "learning_rate": 1.205700016610395e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28850 + }, + { + "epoch": 8.396857724759965, + "flow_matching_loss": 0.013044176902621985, + "grad_norm": 0.06289160251617432, + "learning_rate": 1.1848680541474088e-07, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28860 + }, + { + "epoch": 8.399767238871108, + "flow_matching_loss": 0.01294081350788474, + "grad_norm": 0.06572547554969788, + "learning_rate": 1.1642169161020955e-07, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28870 + }, + { + "epoch": 8.402676752982252, + "flow_matching_loss": 0.012629870045930146, + "grad_norm": 0.08075610548257828, + "learning_rate": 1.1437466275675156e-07, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28880 + }, + { + "epoch": 8.405586267093396, + "flow_matching_loss": 0.01369920363649726, + "grad_norm": 0.06585540622472763, + "learning_rate": 1.1234572134170229e-07, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28890 + }, + { + "epoch": 8.40849578120454, + "flow_matching_loss": 0.012449014093726873, + "grad_norm": 0.07231110334396362, + "learning_rate": 1.1033486983041463e-07, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28900 + }, + { + "epoch": 8.411405295315681, + "flow_matching_loss": 0.013162690307945013, + "grad_norm": 0.06815285980701447, + "learning_rate": 1.0834211066626409e-07, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28910 + }, + { + "epoch": 8.414314809426825, + "flow_matching_loss": 0.012193144112825394, + "grad_norm": 0.06704732030630112, + "learning_rate": 1.0636744627064044e-07, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28920 + }, + { + "epoch": 8.417224323537969, + "flow_matching_loss": 0.014222809672355652, + "grad_norm": 0.06442899256944656, + "learning_rate": 1.0441087904294766e-07, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28930 + }, + { + "epoch": 8.420133837649113, + "flow_matching_loss": 0.011692884564399719, + "grad_norm": 0.06299479305744171, + "learning_rate": 1.0247241136059904e-07, + "loss": 0.0117, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28940 + }, + { + "epoch": 8.423043351760256, + "flow_matching_loss": 0.013323319144546986, + "grad_norm": 0.07683615386486053, + "learning_rate": 1.0055204557901709e-07, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28950 + }, + { + "epoch": 8.4259528658714, + "flow_matching_loss": 0.013062174897640944, + "grad_norm": 0.07336898893117905, + "learning_rate": 9.864978403162361e-08, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28960 + }, + { + "epoch": 8.428862379982544, + "flow_matching_loss": 0.013374030217528343, + "grad_norm": 0.07275928556919098, + "learning_rate": 9.676562902984965e-08, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28970 + }, + { + "epoch": 8.431771894093686, + "flow_matching_loss": 0.012268496491014958, + "grad_norm": 0.07155843079090118, + "learning_rate": 9.489958286311883e-08, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28980 + }, + { + "epoch": 8.43468140820483, + "flow_matching_loss": 0.012507267761975527, + "grad_norm": 0.05814892426133156, + "learning_rate": 9.30516477988541e-08, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 28990 + }, + { + "epoch": 8.437590922315973, + "flow_matching_loss": 0.012914602551609278, + "grad_norm": 0.07504168897867203, + "learning_rate": 9.122182608247098e-08, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29000 + }, + { + "epoch": 8.440500436427117, + "flow_matching_loss": 0.012680661957710981, + "grad_norm": 0.05745575577020645, + "learning_rate": 8.94101199373759e-08, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29010 + }, + { + "epoch": 8.44340995053826, + "flow_matching_loss": 0.01237425347790122, + "grad_norm": 0.05850376561284065, + "learning_rate": 8.761653156495964e-08, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29020 + }, + { + "epoch": 8.446319464649404, + "flow_matching_loss": 0.013304556161165238, + "grad_norm": 0.0765313059091568, + "learning_rate": 8.584106314460549e-08, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29030 + }, + { + "epoch": 8.449228978760548, + "flow_matching_loss": 0.013881264533847571, + "grad_norm": 0.08283841609954834, + "learning_rate": 8.408371683367111e-08, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29040 + }, + { + "epoch": 8.45213849287169, + "flow_matching_loss": 0.01239089211449027, + "grad_norm": 0.06698659807443619, + "learning_rate": 8.234449476750172e-08, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29050 + }, + { + "epoch": 8.455048006982834, + "flow_matching_loss": 0.013106839638203383, + "grad_norm": 0.06033868342638016, + "learning_rate": 8.062339905941351e-08, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29060 + }, + { + "epoch": 8.457957521093977, + "flow_matching_loss": 0.01334813078865409, + "grad_norm": 0.06391813606023788, + "learning_rate": 7.89204318007003e-08, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29070 + }, + { + "epoch": 8.460867035205121, + "flow_matching_loss": 0.012487060204148293, + "grad_norm": 0.062088530510663986, + "learning_rate": 7.723559506062849e-08, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29080 + }, + { + "epoch": 8.463776549316265, + "flow_matching_loss": 0.013291326072067022, + "grad_norm": 0.057236626744270325, + "learning_rate": 7.556889088643381e-08, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29090 + }, + { + "epoch": 8.466686063427408, + "flow_matching_loss": 0.013719519786536693, + "grad_norm": 0.09734895825386047, + "learning_rate": 7.392032130331793e-08, + "loss": 0.0137, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29100 + }, + { + "epoch": 8.46959557753855, + "flow_matching_loss": 0.01578335650265217, + "grad_norm": 0.08742554485797882, + "learning_rate": 7.228988831445015e-08, + "loss": 0.0158, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29110 + }, + { + "epoch": 8.472505091649694, + "flow_matching_loss": 0.01336895665153861, + "grad_norm": 0.06782568991184235, + "learning_rate": 7.06775939009574e-08, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29120 + }, + { + "epoch": 8.475414605760838, + "flow_matching_loss": 0.013114818697795272, + "grad_norm": 0.04426317289471626, + "learning_rate": 6.908344002193256e-08, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29130 + }, + { + "epoch": 8.478324119871981, + "flow_matching_loss": 0.01204625777900219, + "grad_norm": 0.051742974668741226, + "learning_rate": 6.750742861442116e-08, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29140 + }, + { + "epoch": 8.481233633983125, + "flow_matching_loss": 0.012305488437414169, + "grad_norm": 0.06015434116125107, + "learning_rate": 6.594956159342636e-08, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29150 + }, + { + "epoch": 8.484143148094269, + "flow_matching_loss": 0.014164058119058609, + "grad_norm": 0.05719580128788948, + "learning_rate": 6.440984085190394e-08, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29160 + }, + { + "epoch": 8.487052662205413, + "flow_matching_loss": 0.012747060135006904, + "grad_norm": 0.05371512472629547, + "learning_rate": 6.288826826075733e-08, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29170 + }, + { + "epoch": 8.489962176316554, + "flow_matching_loss": 0.013474225252866744, + "grad_norm": 0.07047636061906815, + "learning_rate": 6.138484566884261e-08, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29180 + }, + { + "epoch": 8.492871690427698, + "flow_matching_loss": 0.012641463428735733, + "grad_norm": 0.06298645585775375, + "learning_rate": 5.989957490296183e-08, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29190 + }, + { + "epoch": 8.495781204538842, + "flow_matching_loss": 0.012265716958791018, + "grad_norm": 0.06399785727262497, + "learning_rate": 5.843245776785466e-08, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29200 + }, + { + "epoch": 8.498690718649986, + "flow_matching_loss": 0.012626282963901758, + "grad_norm": 0.0687728002667427, + "learning_rate": 5.6983496046210115e-08, + "loss": 0.0126, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29210 + }, + { + "epoch": 8.50160023276113, + "flow_matching_loss": 0.013016362953931094, + "grad_norm": 0.0646325945854187, + "learning_rate": 5.5552691498653165e-08, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29220 + }, + { + "epoch": 8.504509746872273, + "flow_matching_loss": 0.012440611701458692, + "grad_norm": 0.0592304952442646, + "learning_rate": 5.4140045863743106e-08, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29230 + }, + { + "epoch": 8.507419260983415, + "flow_matching_loss": 0.014053878095000982, + "grad_norm": 0.06468126177787781, + "learning_rate": 5.274556085798188e-08, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29240 + }, + { + "epoch": 8.510328775094559, + "flow_matching_loss": 0.012812061235308647, + "grad_norm": 0.061748284846544266, + "learning_rate": 5.136923817579742e-08, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29250 + }, + { + "epoch": 8.513238289205702, + "flow_matching_loss": 0.013006596826016903, + "grad_norm": 0.07731618732213974, + "learning_rate": 5.001107948955197e-08, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29260 + }, + { + "epoch": 8.516147803316846, + "flow_matching_loss": 0.012975158542394638, + "grad_norm": 0.07094527035951614, + "learning_rate": 4.867108644953711e-08, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29270 + }, + { + "epoch": 8.51905731742799, + "flow_matching_loss": 0.015319705102592706, + "grad_norm": 0.08821843564510345, + "learning_rate": 4.734926068397038e-08, + "loss": 0.0153, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29280 + }, + { + "epoch": 8.521966831539133, + "flow_matching_loss": 0.01447738679125905, + "grad_norm": 0.0640617311000824, + "learning_rate": 4.604560379899536e-08, + "loss": 0.0145, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29290 + }, + { + "epoch": 8.524876345650277, + "flow_matching_loss": 0.013559564296156169, + "grad_norm": 0.07170272618532181, + "learning_rate": 4.476011737867658e-08, + "loss": 0.0136, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29300 + }, + { + "epoch": 8.52778585976142, + "flow_matching_loss": 0.013373140990734101, + "grad_norm": 0.06084325909614563, + "learning_rate": 4.349280298500291e-08, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29310 + }, + { + "epoch": 8.530695373872563, + "flow_matching_loss": 0.013142002467066049, + "grad_norm": 0.1119910255074501, + "learning_rate": 4.224366215787923e-08, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29320 + }, + { + "epoch": 8.533604887983707, + "flow_matching_loss": 0.011528885737061501, + "grad_norm": 0.08044872432947159, + "learning_rate": 4.101269641513139e-08, + "loss": 0.0115, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29330 + }, + { + "epoch": 8.53651440209485, + "flow_matching_loss": 0.014877272583544255, + "grad_norm": 0.07312719523906708, + "learning_rate": 3.979990725249627e-08, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29340 + }, + { + "epoch": 8.539423916205994, + "flow_matching_loss": 0.014207939617335796, + "grad_norm": 0.0782603919506073, + "learning_rate": 3.860529614363006e-08, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29350 + }, + { + "epoch": 8.542333430317138, + "flow_matching_loss": 0.013209477253258228, + "grad_norm": 0.07182188332080841, + "learning_rate": 3.742886454009664e-08, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29360 + }, + { + "epoch": 8.545242944428281, + "flow_matching_loss": 0.01156281353905797, + "grad_norm": 0.06423448771238327, + "learning_rate": 3.627061387137087e-08, + "loss": 0.0116, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29370 + }, + { + "epoch": 8.548152458539423, + "flow_matching_loss": 0.014099615719169378, + "grad_norm": 0.08433372527360916, + "learning_rate": 3.5130545544835305e-08, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29380 + }, + { + "epoch": 8.551061972650567, + "flow_matching_loss": 0.014189791120588779, + "grad_norm": 0.07968813180923462, + "learning_rate": 3.400866094578348e-08, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29390 + }, + { + "epoch": 8.55397148676171, + "flow_matching_loss": 0.013223117310553789, + "grad_norm": 0.05581344664096832, + "learning_rate": 3.290496143740829e-08, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29400 + }, + { + "epoch": 8.556881000872854, + "flow_matching_loss": 0.015703410468995573, + "grad_norm": 0.06435631960630417, + "learning_rate": 3.18194483608103e-08, + "loss": 0.0157, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29410 + }, + { + "epoch": 8.559790514983998, + "flow_matching_loss": 0.01250689085572958, + "grad_norm": 0.06279484182596207, + "learning_rate": 3.075212303498942e-08, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29420 + }, + { + "epoch": 8.562700029095142, + "flow_matching_loss": 0.0134184411726892, + "grad_norm": 0.06455033272504807, + "learning_rate": 2.9702986756846574e-08, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29430 + }, + { + "epoch": 8.565609543206284, + "flow_matching_loss": 0.01080277729779482, + "grad_norm": 0.07135183364152908, + "learning_rate": 2.867204080118202e-08, + "loss": 0.0108, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29440 + }, + { + "epoch": 8.568519057317427, + "flow_matching_loss": 0.014072981849312782, + "grad_norm": 0.10049489140510559, + "learning_rate": 2.7659286420693708e-08, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29450 + }, + { + "epoch": 8.571428571428571, + "flow_matching_loss": 0.013306618854403495, + "grad_norm": 0.08165105432271957, + "learning_rate": 2.6664724845968934e-08, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29460 + }, + { + "epoch": 8.574338085539715, + "flow_matching_loss": 0.013532500807195902, + "grad_norm": 0.07542365044355392, + "learning_rate": 2.5688357285499343e-08, + "loss": 0.0135, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29470 + }, + { + "epoch": 8.577247599650859, + "flow_matching_loss": 0.014202935062348843, + "grad_norm": 0.07224608957767487, + "learning_rate": 2.473018492565926e-08, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29480 + }, + { + "epoch": 8.580157113762002, + "flow_matching_loss": 0.011835665814578534, + "grad_norm": 0.07719457894563675, + "learning_rate": 2.3790208930720702e-08, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29490 + }, + { + "epoch": 8.583066627873146, + "flow_matching_loss": 0.012542288471013307, + "grad_norm": 0.0752614438533783, + "learning_rate": 2.2868430442843368e-08, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29500 + }, + { + "epoch": 8.585976141984288, + "flow_matching_loss": 0.012244785018265248, + "grad_norm": 0.06834932416677475, + "learning_rate": 2.196485058207298e-08, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29510 + }, + { + "epoch": 8.588885656095432, + "flow_matching_loss": 0.012411410547792911, + "grad_norm": 0.06936447322368622, + "learning_rate": 2.1079470446346282e-08, + "loss": 0.0124, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29520 + }, + { + "epoch": 8.591795170206575, + "flow_matching_loss": 0.014243972674012183, + "grad_norm": 0.06403341889381409, + "learning_rate": 2.0212291111482707e-08, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29530 + }, + { + "epoch": 8.594704684317719, + "flow_matching_loss": 0.012990737333893776, + "grad_norm": 0.073891282081604, + "learning_rate": 1.9363313631187707e-08, + "loss": 0.013, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29540 + }, + { + "epoch": 8.597614198428863, + "flow_matching_loss": 0.014920033048838377, + "grad_norm": 0.07430481165647507, + "learning_rate": 1.853253903704777e-08, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29550 + }, + { + "epoch": 8.600523712540006, + "flow_matching_loss": 0.01188280750066042, + "grad_norm": 0.07000619173049927, + "learning_rate": 1.7719968338533733e-08, + "loss": 0.0119, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29560 + }, + { + "epoch": 8.603433226651148, + "flow_matching_loss": 0.012207754887640476, + "grad_norm": 0.06758992373943329, + "learning_rate": 1.6925602522992467e-08, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29570 + }, + { + "epoch": 8.606342740762292, + "flow_matching_loss": 0.012458876986056567, + "grad_norm": 0.07193419337272644, + "learning_rate": 1.6149442555658533e-08, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29580 + }, + { + "epoch": 8.609252254873436, + "flow_matching_loss": 0.014115791767835617, + "grad_norm": 0.07100837677717209, + "learning_rate": 1.5391489379635858e-08, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29590 + }, + { + "epoch": 8.61216176898458, + "flow_matching_loss": 0.013086906354874372, + "grad_norm": 0.06890976428985596, + "learning_rate": 1.4651743915909399e-08, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29600 + }, + { + "epoch": 8.615071283095723, + "flow_matching_loss": 0.015219411347061396, + "grad_norm": 0.0743747130036354, + "learning_rate": 1.3930207063340139e-08, + "loss": 0.0152, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29610 + }, + { + "epoch": 8.617980797206867, + "flow_matching_loss": 0.01248831981793046, + "grad_norm": 0.05777926370501518, + "learning_rate": 1.322687969866343e-08, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29620 + }, + { + "epoch": 8.62089031131801, + "flow_matching_loss": 0.011752333212643861, + "grad_norm": 0.07623199373483658, + "learning_rate": 1.2541762676488989e-08, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29630 + }, + { + "epoch": 8.623799825429153, + "flow_matching_loss": 0.011531879007816315, + "grad_norm": 0.06842109560966492, + "learning_rate": 1.1874856829297564e-08, + "loss": 0.0115, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29640 + }, + { + "epoch": 8.626709339540296, + "flow_matching_loss": 0.013851037342101335, + "grad_norm": 0.07599944621324539, + "learning_rate": 1.1226162967442611e-08, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29650 + }, + { + "epoch": 8.62961885365144, + "flow_matching_loss": 0.011803677119314671, + "grad_norm": 0.07787354290485382, + "learning_rate": 1.059568187914861e-08, + "loss": 0.0118, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29660 + }, + { + "epoch": 8.632528367762584, + "flow_matching_loss": 0.012823010887950658, + "grad_norm": 0.07591405510902405, + "learning_rate": 9.983414330511087e-09, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29670 + }, + { + "epoch": 8.635437881873727, + "flow_matching_loss": 0.013875067606568337, + "grad_norm": 0.08333554863929749, + "learning_rate": 9.389361065489932e-09, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29680 + }, + { + "epoch": 8.638347395984871, + "flow_matching_loss": 0.013230404071509837, + "grad_norm": 0.08494426310062408, + "learning_rate": 8.81352280591774e-09, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29690 + }, + { + "epoch": 8.641256910096015, + "flow_matching_loss": 0.011368190217763186, + "grad_norm": 0.0666668638586998, + "learning_rate": 8.255900251491478e-09, + "loss": 0.0114, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29700 + }, + { + "epoch": 8.644166424207157, + "flow_matching_loss": 0.014926906116306782, + "grad_norm": 0.08105084300041199, + "learning_rate": 7.716494079775816e-09, + "loss": 0.0149, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29710 + }, + { + "epoch": 8.6470759383183, + "flow_matching_loss": 0.012734224647283554, + "grad_norm": 0.06504005193710327, + "learning_rate": 7.195304946199799e-09, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29720 + }, + { + "epoch": 8.649985452429444, + "flow_matching_loss": 0.012916474603116512, + "grad_norm": 0.0662386417388916, + "learning_rate": 6.692333484056845e-09, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29730 + }, + { + "epoch": 8.652894966540588, + "flow_matching_loss": 0.014136611949652433, + "grad_norm": 0.08607471734285355, + "learning_rate": 6.207580304506411e-09, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29740 + }, + { + "epoch": 8.655804480651732, + "flow_matching_loss": 0.01336164642125368, + "grad_norm": 0.09756533056497574, + "learning_rate": 5.741045996567329e-09, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29750 + }, + { + "epoch": 8.658713994762875, + "flow_matching_loss": 0.013054155930876731, + "grad_norm": 0.07884014397859573, + "learning_rate": 5.2927311271244725e-09, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29760 + }, + { + "epoch": 8.661623508874017, + "flow_matching_loss": 0.014235230349004268, + "grad_norm": 0.07753881067037582, + "learning_rate": 4.8626362409204265e-09, + "loss": 0.0142, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29770 + }, + { + "epoch": 8.664533022985161, + "flow_matching_loss": 0.014132514037191868, + "grad_norm": 0.08987182378768921, + "learning_rate": 4.4507618605638126e-09, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29780 + }, + { + "epoch": 8.667442537096305, + "flow_matching_loss": 0.01443347306922078, + "grad_norm": 0.09337235242128372, + "learning_rate": 4.057108486519301e-09, + "loss": 0.0144, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29790 + }, + { + "epoch": 8.670352051207448, + "flow_matching_loss": 0.013343097642064095, + "grad_norm": 0.07780446112155914, + "learning_rate": 3.6816765971126043e-09, + "loss": 0.0133, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29800 + }, + { + "epoch": 8.673261565318592, + "flow_matching_loss": 0.01294982060790062, + "grad_norm": 0.06808187067508698, + "learning_rate": 3.324466648528812e-09, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29810 + }, + { + "epoch": 8.676171079429736, + "flow_matching_loss": 0.014094784390181303, + "grad_norm": 0.06912414729595184, + "learning_rate": 2.9854790748123917e-09, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29820 + }, + { + "epoch": 8.67908059354088, + "flow_matching_loss": 0.012659616768360138, + "grad_norm": 0.0690755620598793, + "learning_rate": 2.664714287865522e-09, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29830 + }, + { + "epoch": 8.681990107652021, + "flow_matching_loss": 0.012873538583517075, + "grad_norm": 0.0632394552230835, + "learning_rate": 2.362172677446428e-09, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29840 + }, + { + "epoch": 8.684899621763165, + "flow_matching_loss": 0.012170662172138691, + "grad_norm": 0.08294575661420822, + "learning_rate": 2.0778546111727137e-09, + "loss": 0.0122, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29850 + }, + { + "epoch": 8.687809135874309, + "flow_matching_loss": 0.01278019482269883, + "grad_norm": 0.07993537932634354, + "learning_rate": 1.811760434516363e-09, + "loss": 0.0128, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29860 + }, + { + "epoch": 8.690718649985453, + "flow_matching_loss": 0.01249940125271678, + "grad_norm": 0.06431994587182999, + "learning_rate": 1.5638904708087377e-09, + "loss": 0.0125, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29870 + }, + { + "epoch": 8.693628164096596, + "flow_matching_loss": 0.012336233723908662, + "grad_norm": 0.05681069195270538, + "learning_rate": 1.3342450212355805e-09, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29880 + }, + { + "epoch": 8.69653767820774, + "flow_matching_loss": 0.013931627664715052, + "grad_norm": 0.0994863212108612, + "learning_rate": 1.122824364835351e-09, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29890 + }, + { + "epoch": 8.699447192318882, + "flow_matching_loss": 0.012655199319124222, + "grad_norm": 0.06071685999631882, + "learning_rate": 9.296287585042196e-10, + "loss": 0.0127, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29900 + }, + { + "epoch": 8.702356706430026, + "flow_matching_loss": 0.0122809954918921, + "grad_norm": 0.06998459994792938, + "learning_rate": 7.546584369960697e-10, + "loss": 0.0123, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29910 + }, + { + "epoch": 8.70526622054117, + "flow_matching_loss": 0.013850979227572679, + "grad_norm": 0.07262946665287018, + "learning_rate": 5.97913612912504e-10, + "loss": 0.0139, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29920 + }, + { + "epoch": 8.708175734652313, + "flow_matching_loss": 0.012019805982708931, + "grad_norm": 0.05133824422955513, + "learning_rate": 4.593944767161684e-10, + "loss": 0.012, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29930 + }, + { + "epoch": 8.711085248763457, + "flow_matching_loss": 0.014054147992283106, + "grad_norm": 0.0776619017124176, + "learning_rate": 3.391011967207591e-10, + "loss": 0.0141, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29940 + }, + { + "epoch": 8.7139947628746, + "flow_matching_loss": 0.012134960293769837, + "grad_norm": 0.07352397590875626, + "learning_rate": 2.370339190910231e-10, + "loss": 0.0121, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29950 + }, + { + "epoch": 8.716904276985744, + "flow_matching_loss": 0.012933912873268127, + "grad_norm": 0.06397238373756409, + "learning_rate": 1.531927678510847e-10, + "loss": 0.0129, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29960 + }, + { + "epoch": 8.719813791096886, + "flow_matching_loss": 0.013762789592146874, + "grad_norm": 0.06450065225362778, + "learning_rate": 8.757784487611886e-11, + "loss": 0.0138, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29970 + }, + { + "epoch": 8.72272330520803, + "flow_matching_loss": 0.013357210624963044, + "grad_norm": 0.07358156889677048, + "learning_rate": 4.018922989235119e-11, + "loss": 0.0134, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29980 + }, + { + "epoch": 8.725632819319173, + "flow_matching_loss": 0.013158852886408567, + "grad_norm": 0.06373626738786697, + "learning_rate": 1.1026980482053972e-11, + "loss": 0.0132, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 29990 + }, + { + "epoch": 8.728542333430317, + "flow_matching_loss": 0.013071979489177465, + "grad_norm": 0.06255932152271271, + "learning_rate": 9.113208188082922e-14, + "loss": 0.0131, + "mgd_loss": 0.0, + "mgd_token_mask_ratio_cfg": 0.699999988079071, + "step": 30000 + }, + { + "epoch": 8.728542333430317, + "step": 30000, + "total_flos": 0.0, + "train_loss": 0.016048218871404726, + "train_runtime": 61633.3056, + "train_samples_per_second": 62.304, + "train_steps_per_second": 0.487 + } + ], + "logging_steps": 10, + "max_steps": 30000, + "num_input_tokens_seen": 0, + "num_train_epochs": 9, + "save_steps": 1, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": false, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 0.0, + "train_batch_size": 128, + "trial_name": null, + "trial_params": null +} diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/phase3/training_args.bin b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/phase3/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..55f6f3526c69c2b0e7981a40852b1dc429b77d2c --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/phase3/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:58b199da585d22d1b7fb00f02701dceb547adbfb8b69cfaa1e126aba2284c0b0 +size 5969 diff --git a/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/phase3/model-00001-of-00002.safetensors b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/phase3/model-00001-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..cf711fc4f2a7984357f091d3ba166a299f4f47e1 --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/phase3/model-00001-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:003d413d1868bb05d96fb150e12b6c2924bdcfe9a2ec7f576075b2de9d6f029d +size 4999367032 diff --git a/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/phase3/model-00002-of-00002.safetensors b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/phase3/model-00002-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..71349e82b56d8b0d7de35c0ab93d64d004ad2709 --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/phase3/model-00002-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ad5f032032e20939bb00f113fb1fb6bf4c44fb830c5e774c499d092bc9c7e329 +size 2643339748 diff --git a/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/phase3/training_args.bin b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/phase3/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..ffd7a9247b249bb67338372960dd62256889172c --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/phase3/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:57e8e5b32444ef2fa141d9faad5ffaab46ab9a8993736a58a01efe0dd655a4ae +size 6033 diff --git a/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/phase3/model-00001-of-00002.safetensors b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/phase3/model-00001-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..a40c799c029c7e2af543f067facec68ac94c1297 --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/phase3/model-00001-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3b0351c4631c466cba8297e761920f24b271b74ce69021361d26673cac5defd1 +size 4999367032 diff --git a/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/phase3/model-00002-of-00002.safetensors b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/phase3/model-00002-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c409f960dfab09f6bed7ddc413b506adab01ecca --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/phase3/model-00002-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:be7c7e81a8889271fac3864bb2eb9301a45617f880127f7d01b66ea234c70291 +size 2643339748 diff --git a/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/phase3/trainer_state.json b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/phase3/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..b0507e6c5780b9ed12f41ecc10ea7dbfe21e6e78 --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/phase3/trainer_state.json @@ -0,0 +1,36043 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 8.728542333430317, + "eval_steps": 500, + "global_step": 30000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.002909514111143439, + "flow_matching_loss": 0.12287038937211037, + "grad_norm": 0.7655800580978394, + "learning_rate": 1.8e-07, + "loss": 0.1229, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10 + }, + { + "epoch": 0.005819028222286878, + "flow_matching_loss": 0.11181098520755768, + "grad_norm": 0.6593242883682251, + "learning_rate": 3.8e-07, + "loss": 0.1118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20 + }, + { + "epoch": 0.008728542333430317, + "flow_matching_loss": 0.10511413663625717, + "grad_norm": 0.6830371022224426, + "learning_rate": 5.800000000000001e-07, + "loss": 0.1051, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 30 + }, + { + "epoch": 0.011638056444573755, + "flow_matching_loss": 0.08889771699905395, + "grad_norm": 0.4798424243927002, + "learning_rate": 7.799999999999999e-07, + "loss": 0.0889, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 40 + }, + { + "epoch": 0.014547570555717195, + "flow_matching_loss": 0.0752666525542736, + "grad_norm": 0.39793306589126587, + "learning_rate": 9.8e-07, + "loss": 0.0753, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 50 + }, + { + "epoch": 0.017457084666860634, + "flow_matching_loss": 0.06627346128225327, + "grad_norm": 0.2909850478172302, + "learning_rate": 1.18e-06, + "loss": 0.0663, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 60 + }, + { + "epoch": 0.020366598778004074, + "flow_matching_loss": 0.0545802790671587, + "grad_norm": 0.28076159954071045, + "learning_rate": 1.38e-06, + "loss": 0.0546, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 70 + }, + { + "epoch": 0.02327611288914751, + "flow_matching_loss": 0.05200903229415417, + "grad_norm": 0.23693877458572388, + "learning_rate": 1.5800000000000001e-06, + "loss": 0.052, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 80 + }, + { + "epoch": 0.02618562700029095, + "flow_matching_loss": 0.041241689771413806, + "grad_norm": 0.24238331615924835, + "learning_rate": 1.7800000000000001e-06, + "loss": 0.0412, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 90 + }, + { + "epoch": 0.02909514111143439, + "flow_matching_loss": 0.03884585350751877, + "grad_norm": 0.27954593300819397, + "learning_rate": 1.98e-06, + "loss": 0.0388, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 100 + }, + { + "epoch": 0.03200465522257783, + "flow_matching_loss": 0.03645187355577946, + "grad_norm": 0.20790651440620422, + "learning_rate": 2.1800000000000003e-06, + "loss": 0.0365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 110 + }, + { + "epoch": 0.03491416933372127, + "flow_matching_loss": 0.035410002991557124, + "grad_norm": 0.19240981340408325, + "learning_rate": 2.38e-06, + "loss": 0.0354, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 120 + }, + { + "epoch": 0.037823683444864704, + "flow_matching_loss": 0.029070502519607543, + "grad_norm": 0.15042144060134888, + "learning_rate": 2.58e-06, + "loss": 0.0291, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 130 + }, + { + "epoch": 0.04073319755600815, + "flow_matching_loss": 0.025942875631153585, + "grad_norm": 0.12835712730884552, + "learning_rate": 2.78e-06, + "loss": 0.0259, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 140 + }, + { + "epoch": 0.043642711667151585, + "flow_matching_loss": 0.026672330126166342, + "grad_norm": 0.1484895795583725, + "learning_rate": 2.98e-06, + "loss": 0.0267, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 150 + }, + { + "epoch": 0.04655222577829502, + "flow_matching_loss": 0.022607591189444064, + "grad_norm": 0.12978270649909973, + "learning_rate": 3.18e-06, + "loss": 0.0226, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 160 + }, + { + "epoch": 0.049461739889438465, + "flow_matching_loss": 0.025677112117409705, + "grad_norm": 0.14417092502117157, + "learning_rate": 3.38e-06, + "loss": 0.0257, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 170 + }, + { + "epoch": 0.0523712540005819, + "flow_matching_loss": 0.02321900725364685, + "grad_norm": 0.10060111433267593, + "learning_rate": 3.58e-06, + "loss": 0.0232, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 180 + }, + { + "epoch": 0.055280768111725345, + "flow_matching_loss": 0.02032177001237869, + "grad_norm": 0.09406960755586624, + "learning_rate": 3.7800000000000002e-06, + "loss": 0.0203, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 190 + }, + { + "epoch": 0.05819028222286878, + "flow_matching_loss": 0.021089514903724193, + "grad_norm": 0.09270510077476501, + "learning_rate": 3.98e-06, + "loss": 0.0211, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 200 + }, + { + "epoch": 0.06109979633401222, + "flow_matching_loss": 0.020576616004109382, + "grad_norm": 0.09434044361114502, + "learning_rate": 4.18e-06, + "loss": 0.0206, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 210 + }, + { + "epoch": 0.06400931044515566, + "flow_matching_loss": 0.021583461202681065, + "grad_norm": 0.09462200105190277, + "learning_rate": 4.3799999999999996e-06, + "loss": 0.0216, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 220 + }, + { + "epoch": 0.06691882455629909, + "flow_matching_loss": 0.01958881700411439, + "grad_norm": 0.10815712064504623, + "learning_rate": 4.58e-06, + "loss": 0.0196, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 230 + }, + { + "epoch": 0.06982833866744254, + "flow_matching_loss": 0.018696496076881886, + "grad_norm": 0.07075441628694534, + "learning_rate": 4.78e-06, + "loss": 0.0187, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 240 + }, + { + "epoch": 0.07273785277858598, + "flow_matching_loss": 0.01898619355633855, + "grad_norm": 0.09028103202581406, + "learning_rate": 4.980000000000001e-06, + "loss": 0.019, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 250 + }, + { + "epoch": 0.07564736688972941, + "flow_matching_loss": 0.01816976359114051, + "grad_norm": 0.07521460950374603, + "learning_rate": 5.18e-06, + "loss": 0.0182, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 260 + }, + { + "epoch": 0.07855688100087285, + "flow_matching_loss": 0.01998343812301755, + "grad_norm": 0.09349961578845978, + "learning_rate": 5.38e-06, + "loss": 0.02, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 270 + }, + { + "epoch": 0.0814663951120163, + "flow_matching_loss": 0.017167134024202824, + "grad_norm": 0.0709666833281517, + "learning_rate": 5.58e-06, + "loss": 0.0172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 280 + }, + { + "epoch": 0.08437590922315973, + "flow_matching_loss": 0.01917630694806576, + "grad_norm": 0.08182869851589203, + "learning_rate": 5.7800000000000006e-06, + "loss": 0.0192, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 290 + }, + { + "epoch": 0.08728542333430317, + "flow_matching_loss": 0.01859060786664486, + "grad_norm": 0.08940938115119934, + "learning_rate": 5.98e-06, + "loss": 0.0186, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 300 + }, + { + "epoch": 0.09019493744544661, + "flow_matching_loss": 0.01831605238839984, + "grad_norm": 0.07600802928209305, + "learning_rate": 6.18e-06, + "loss": 0.0183, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 310 + }, + { + "epoch": 0.09310445155659004, + "flow_matching_loss": 0.018884282931685446, + "grad_norm": 0.09215965121984482, + "learning_rate": 6.38e-06, + "loss": 0.0189, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 320 + }, + { + "epoch": 0.09601396566773349, + "flow_matching_loss": 0.018094755429774522, + "grad_norm": 0.08244185149669647, + "learning_rate": 6.58e-06, + "loss": 0.0181, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 330 + }, + { + "epoch": 0.09892347977887693, + "flow_matching_loss": 0.0170771025121212, + "grad_norm": 0.08249742537736893, + "learning_rate": 6.78e-06, + "loss": 0.0171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 340 + }, + { + "epoch": 0.10183299389002037, + "flow_matching_loss": 0.015987787675112487, + "grad_norm": 0.07061918824911118, + "learning_rate": 6.98e-06, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 350 + }, + { + "epoch": 0.1047425080011638, + "flow_matching_loss": 0.01780217355117202, + "grad_norm": 0.07464741915464401, + "learning_rate": 7.180000000000001e-06, + "loss": 0.0178, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 360 + }, + { + "epoch": 0.10765202211230725, + "flow_matching_loss": 0.017736614029854537, + "grad_norm": 0.073550283908844, + "learning_rate": 7.3800000000000005e-06, + "loss": 0.0177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 370 + }, + { + "epoch": 0.11056153622345069, + "flow_matching_loss": 0.018482870887964964, + "grad_norm": 0.08359595388174057, + "learning_rate": 7.5799999999999994e-06, + "loss": 0.0185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 380 + }, + { + "epoch": 0.11347105033459412, + "flow_matching_loss": 0.018767212610691787, + "grad_norm": 0.06589818745851517, + "learning_rate": 7.780000000000002e-06, + "loss": 0.0188, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 390 + }, + { + "epoch": 0.11638056444573756, + "flow_matching_loss": 0.016612835135310887, + "grad_norm": 0.09178656339645386, + "learning_rate": 7.98e-06, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 400 + }, + { + "epoch": 0.11929007855688101, + "flow_matching_loss": 0.015700438432395457, + "grad_norm": 0.0743747279047966, + "learning_rate": 8.18e-06, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 410 + }, + { + "epoch": 0.12219959266802444, + "flow_matching_loss": 0.01663182508200407, + "grad_norm": 0.0703354999423027, + "learning_rate": 8.38e-06, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 420 + }, + { + "epoch": 0.12510910677916787, + "flow_matching_loss": 0.018547352775931357, + "grad_norm": 0.08598528802394867, + "learning_rate": 8.58e-06, + "loss": 0.0185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 430 + }, + { + "epoch": 0.12801862089031132, + "flow_matching_loss": 0.018368005659431218, + "grad_norm": 0.0846896842122078, + "learning_rate": 8.78e-06, + "loss": 0.0184, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 440 + }, + { + "epoch": 0.13092813500145475, + "flow_matching_loss": 0.017690911050885917, + "grad_norm": 0.07368426024913788, + "learning_rate": 8.98e-06, + "loss": 0.0177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 450 + }, + { + "epoch": 0.13383764911259818, + "flow_matching_loss": 0.016177355125546456, + "grad_norm": 0.07443322986364365, + "learning_rate": 9.18e-06, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 460 + }, + { + "epoch": 0.13674716322374164, + "flow_matching_loss": 0.017277533933520318, + "grad_norm": 0.0691395178437233, + "learning_rate": 9.38e-06, + "loss": 0.0173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 470 + }, + { + "epoch": 0.13965667733488507, + "flow_matching_loss": 0.016739646811038254, + "grad_norm": 0.07742138206958771, + "learning_rate": 9.580000000000001e-06, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 480 + }, + { + "epoch": 0.1425661914460285, + "flow_matching_loss": 0.016353982966393232, + "grad_norm": 0.06267094612121582, + "learning_rate": 9.780000000000001e-06, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 490 + }, + { + "epoch": 0.14547570555717196, + "flow_matching_loss": 0.017119845468550922, + "grad_norm": 0.07906630635261536, + "learning_rate": 9.980000000000001e-06, + "loss": 0.0171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 500 + }, + { + "epoch": 0.1483852196683154, + "flow_matching_loss": 0.017756224889308215, + "grad_norm": 0.09237691760063171, + "learning_rate": 1.0179999999999999e-05, + "loss": 0.0178, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 510 + }, + { + "epoch": 0.15129473377945882, + "flow_matching_loss": 0.014518933929502964, + "grad_norm": 0.07308480888605118, + "learning_rate": 1.0379999999999999e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 520 + }, + { + "epoch": 0.15420424789060228, + "flow_matching_loss": 0.014796417951583863, + "grad_norm": 0.07633255422115326, + "learning_rate": 1.058e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 530 + }, + { + "epoch": 0.1571137620017457, + "flow_matching_loss": 0.01719888122752309, + "grad_norm": 0.0809977725148201, + "learning_rate": 1.078e-05, + "loss": 0.0172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 540 + }, + { + "epoch": 0.16002327611288913, + "flow_matching_loss": 0.01553144007921219, + "grad_norm": 0.08279266208410263, + "learning_rate": 1.098e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 550 + }, + { + "epoch": 0.1629327902240326, + "flow_matching_loss": 0.014778489805757999, + "grad_norm": 0.082179956138134, + "learning_rate": 1.118e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 560 + }, + { + "epoch": 0.16584230433517602, + "flow_matching_loss": 0.017346266005188227, + "grad_norm": 0.0897284671664238, + "learning_rate": 1.1380000000000001e-05, + "loss": 0.0173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 570 + }, + { + "epoch": 0.16875181844631945, + "flow_matching_loss": 0.015619205124676228, + "grad_norm": 0.07371357828378677, + "learning_rate": 1.1580000000000001e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 580 + }, + { + "epoch": 0.1716613325574629, + "flow_matching_loss": 0.01821941826492548, + "grad_norm": 0.07222266495227814, + "learning_rate": 1.178e-05, + "loss": 0.0182, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 590 + }, + { + "epoch": 0.17457084666860634, + "flow_matching_loss": 0.016428532265126704, + "grad_norm": 0.08169856667518616, + "learning_rate": 1.198e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 600 + }, + { + "epoch": 0.17748036077974977, + "flow_matching_loss": 0.014954305440187454, + "grad_norm": 0.0770837813615799, + "learning_rate": 1.2180000000000002e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 610 + }, + { + "epoch": 0.18038987489089323, + "flow_matching_loss": 0.016215208265930415, + "grad_norm": 0.08322267979383469, + "learning_rate": 1.238e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 620 + }, + { + "epoch": 0.18329938900203666, + "flow_matching_loss": 0.016140831913799048, + "grad_norm": 0.08825630694627762, + "learning_rate": 1.258e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 630 + }, + { + "epoch": 0.18620890311318009, + "flow_matching_loss": 0.015933708101511002, + "grad_norm": 0.07910127937793732, + "learning_rate": 1.278e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 640 + }, + { + "epoch": 0.18911841722432354, + "flow_matching_loss": 0.016114483028650282, + "grad_norm": 0.07833167910575867, + "learning_rate": 1.298e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 650 + }, + { + "epoch": 0.19202793133546697, + "flow_matching_loss": 0.013714529853314162, + "grad_norm": 0.07532219588756561, + "learning_rate": 1.3180000000000001e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 660 + }, + { + "epoch": 0.1949374454466104, + "flow_matching_loss": 0.018379315733909607, + "grad_norm": 0.06933792680501938, + "learning_rate": 1.338e-05, + "loss": 0.0184, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 670 + }, + { + "epoch": 0.19784695955775386, + "flow_matching_loss": 0.01586674088612199, + "grad_norm": 0.07839253544807434, + "learning_rate": 1.358e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 680 + }, + { + "epoch": 0.2007564736688973, + "flow_matching_loss": 0.01520371288061142, + "grad_norm": 0.07804407179355621, + "learning_rate": 1.378e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 690 + }, + { + "epoch": 0.20366598778004075, + "flow_matching_loss": 0.014503150619566441, + "grad_norm": 0.09584196656942368, + "learning_rate": 1.3980000000000002e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 700 + }, + { + "epoch": 0.20657550189118418, + "flow_matching_loss": 0.016069578379392623, + "grad_norm": 0.07581158727407455, + "learning_rate": 1.4180000000000001e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 710 + }, + { + "epoch": 0.2094850160023276, + "flow_matching_loss": 0.014246354717761278, + "grad_norm": 0.09042142331600189, + "learning_rate": 1.438e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 720 + }, + { + "epoch": 0.21239453011347106, + "flow_matching_loss": 0.01645402628928423, + "grad_norm": 0.07873786985874176, + "learning_rate": 1.458e-05, + "loss": 0.0165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 730 + }, + { + "epoch": 0.2153040442246145, + "flow_matching_loss": 0.015115041006356477, + "grad_norm": 0.07085967808961868, + "learning_rate": 1.4779999999999999e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 740 + }, + { + "epoch": 0.21821355833575792, + "flow_matching_loss": 0.014925589226186276, + "grad_norm": 0.08381156623363495, + "learning_rate": 1.498e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 750 + }, + { + "epoch": 0.22112307244690138, + "flow_matching_loss": 0.015460956934839487, + "grad_norm": 0.10196590423583984, + "learning_rate": 1.518e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 760 + }, + { + "epoch": 0.2240325865580448, + "flow_matching_loss": 0.01630785772576928, + "grad_norm": 0.07801049947738647, + "learning_rate": 1.5380000000000002e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 770 + }, + { + "epoch": 0.22694210066918824, + "flow_matching_loss": 0.01563758449628949, + "grad_norm": 0.08088133484125137, + "learning_rate": 1.558e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 780 + }, + { + "epoch": 0.2298516147803317, + "flow_matching_loss": 0.015925957635045052, + "grad_norm": 0.08935142308473587, + "learning_rate": 1.578e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 790 + }, + { + "epoch": 0.23276112889147513, + "flow_matching_loss": 0.01611961144953966, + "grad_norm": 0.08931814879179001, + "learning_rate": 1.598e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 800 + }, + { + "epoch": 0.23567064300261856, + "flow_matching_loss": 0.0146680592559278, + "grad_norm": 0.0730084627866745, + "learning_rate": 1.618e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 810 + }, + { + "epoch": 0.23858015711376201, + "flow_matching_loss": 0.015753265377134083, + "grad_norm": 0.0696752741932869, + "learning_rate": 1.6380000000000002e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 820 + }, + { + "epoch": 0.24148967122490544, + "flow_matching_loss": 0.013805709406733513, + "grad_norm": 0.07250497490167618, + "learning_rate": 1.658e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 830 + }, + { + "epoch": 0.24439918533604887, + "flow_matching_loss": 0.015988719183951616, + "grad_norm": 0.07009713351726532, + "learning_rate": 1.6780000000000002e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 840 + }, + { + "epoch": 0.24730869944719233, + "flow_matching_loss": 0.017753837071359156, + "grad_norm": 0.0813191682100296, + "learning_rate": 1.698e-05, + "loss": 0.0178, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 850 + }, + { + "epoch": 0.25021821355833573, + "flow_matching_loss": 0.015170101914554835, + "grad_norm": 0.08393101394176483, + "learning_rate": 1.718e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 860 + }, + { + "epoch": 0.2531277276694792, + "flow_matching_loss": 0.015282752364873886, + "grad_norm": 0.07942099124193192, + "learning_rate": 1.7380000000000003e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 870 + }, + { + "epoch": 0.25603724178062265, + "flow_matching_loss": 0.01634777756407857, + "grad_norm": 0.07774577289819717, + "learning_rate": 1.758e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 880 + }, + { + "epoch": 0.2589467558917661, + "flow_matching_loss": 0.015448281168937683, + "grad_norm": 0.08429206907749176, + "learning_rate": 1.778e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 890 + }, + { + "epoch": 0.2618562700029095, + "flow_matching_loss": 0.01710456293076277, + "grad_norm": 0.08523119986057281, + "learning_rate": 1.798e-05, + "loss": 0.0171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 900 + }, + { + "epoch": 0.26476578411405294, + "flow_matching_loss": 0.01518495911732316, + "grad_norm": 0.07964563369750977, + "learning_rate": 1.818e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 910 + }, + { + "epoch": 0.26767529822519637, + "flow_matching_loss": 0.015967101883143188, + "grad_norm": 0.08049096167087555, + "learning_rate": 1.838e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 920 + }, + { + "epoch": 0.27058481233633985, + "flow_matching_loss": 0.016429709363728763, + "grad_norm": 0.08577635139226913, + "learning_rate": 1.8579999999999998e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 930 + }, + { + "epoch": 0.2734943264474833, + "flow_matching_loss": 0.016249098628759385, + "grad_norm": 0.08828327804803848, + "learning_rate": 1.878e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 940 + }, + { + "epoch": 0.2764038405586267, + "flow_matching_loss": 0.016374445892870425, + "grad_norm": 0.10110734403133392, + "learning_rate": 1.898e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 950 + }, + { + "epoch": 0.27931335466977014, + "flow_matching_loss": 0.01671481700614095, + "grad_norm": 0.09098681807518005, + "learning_rate": 1.918e-05, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 960 + }, + { + "epoch": 0.28222286878091357, + "flow_matching_loss": 0.01624807035550475, + "grad_norm": 0.06724198162555695, + "learning_rate": 1.938e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 970 + }, + { + "epoch": 0.285132382892057, + "flow_matching_loss": 0.015106600429862737, + "grad_norm": 0.0738142803311348, + "learning_rate": 1.958e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 980 + }, + { + "epoch": 0.2880418970032005, + "flow_matching_loss": 0.016135610267519952, + "grad_norm": 0.08625482767820358, + "learning_rate": 1.978e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 990 + }, + { + "epoch": 0.2909514111143439, + "flow_matching_loss": 0.016591527499258518, + "grad_norm": 0.0907081887125969, + "learning_rate": 1.9980000000000002e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1000 + }, + { + "epoch": 0.29386092522548735, + "flow_matching_loss": 0.01641002744436264, + "grad_norm": 0.09467216581106186, + "learning_rate": 2.018e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1010 + }, + { + "epoch": 0.2967704393366308, + "flow_matching_loss": 0.01609385944902897, + "grad_norm": 0.07275364547967911, + "learning_rate": 2.038e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1020 + }, + { + "epoch": 0.2996799534477742, + "flow_matching_loss": 0.01618061438202858, + "grad_norm": 0.10728253424167633, + "learning_rate": 2.0580000000000003e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1030 + }, + { + "epoch": 0.30258946755891764, + "flow_matching_loss": 0.015277042984962463, + "grad_norm": 0.08212406933307648, + "learning_rate": 2.078e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1040 + }, + { + "epoch": 0.3054989816700611, + "flow_matching_loss": 0.016740335337817668, + "grad_norm": 0.09040951728820801, + "learning_rate": 2.0980000000000002e-05, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1050 + }, + { + "epoch": 0.30840849578120455, + "flow_matching_loss": 0.014953754004091024, + "grad_norm": 0.09300144761800766, + "learning_rate": 2.118e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1060 + }, + { + "epoch": 0.311318009892348, + "flow_matching_loss": 0.01597188524901867, + "grad_norm": 0.07752863317728043, + "learning_rate": 2.1380000000000002e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1070 + }, + { + "epoch": 0.3142275240034914, + "flow_matching_loss": 0.015044747106730937, + "grad_norm": 0.10097916424274445, + "learning_rate": 2.1580000000000003e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1080 + }, + { + "epoch": 0.31713703811463484, + "flow_matching_loss": 0.013395094312727451, + "grad_norm": 0.07679927349090576, + "learning_rate": 2.178e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1090 + }, + { + "epoch": 0.32004655222577827, + "flow_matching_loss": 0.01650534477084875, + "grad_norm": 0.0793059691786766, + "learning_rate": 2.198e-05, + "loss": 0.0165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1100 + }, + { + "epoch": 0.32295606633692175, + "flow_matching_loss": 0.016863738372921943, + "grad_norm": 0.10074655711650848, + "learning_rate": 2.2179999999999998e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1110 + }, + { + "epoch": 0.3258655804480652, + "flow_matching_loss": 0.014548236411064864, + "grad_norm": 0.07688106596469879, + "learning_rate": 2.238e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1120 + }, + { + "epoch": 0.3287750945592086, + "flow_matching_loss": 0.01585365515202284, + "grad_norm": 0.08229753375053406, + "learning_rate": 2.258e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1130 + }, + { + "epoch": 0.33168460867035204, + "flow_matching_loss": 0.014446070045232772, + "grad_norm": 0.0893310010433197, + "learning_rate": 2.278e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1140 + }, + { + "epoch": 0.3345941227814955, + "flow_matching_loss": 0.014956111181527376, + "grad_norm": 0.081088125705719, + "learning_rate": 2.298e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1150 + }, + { + "epoch": 0.3375036368926389, + "flow_matching_loss": 0.016559711191803217, + "grad_norm": 0.09041618555784225, + "learning_rate": 2.3179999999999998e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1160 + }, + { + "epoch": 0.3404131510037824, + "flow_matching_loss": 0.015907222870737314, + "grad_norm": 0.06750702112913132, + "learning_rate": 2.338e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1170 + }, + { + "epoch": 0.3433226651149258, + "flow_matching_loss": 0.017233227379620075, + "grad_norm": 0.09086409211158752, + "learning_rate": 2.358e-05, + "loss": 0.0172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1180 + }, + { + "epoch": 0.34623217922606925, + "flow_matching_loss": 0.015013408474624157, + "grad_norm": 0.08044689893722534, + "learning_rate": 2.378e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1190 + }, + { + "epoch": 0.3491416933372127, + "flow_matching_loss": 0.015800530929118393, + "grad_norm": 0.0949568971991539, + "learning_rate": 2.398e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1200 + }, + { + "epoch": 0.3520512074483561, + "flow_matching_loss": 0.0163030200637877, + "grad_norm": 0.0727931335568428, + "learning_rate": 2.4180000000000002e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1210 + }, + { + "epoch": 0.35496072155949954, + "flow_matching_loss": 0.016082535218447446, + "grad_norm": 0.09800975769758224, + "learning_rate": 2.438e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1220 + }, + { + "epoch": 0.357870235670643, + "flow_matching_loss": 0.015517400484532118, + "grad_norm": 0.08067825436592102, + "learning_rate": 2.4580000000000002e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1230 + }, + { + "epoch": 0.36077974978178645, + "flow_matching_loss": 0.015083470847457647, + "grad_norm": 0.07307286560535431, + "learning_rate": 2.478e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1240 + }, + { + "epoch": 0.3636892638929299, + "flow_matching_loss": 0.01684371717274189, + "grad_norm": 0.07553724199533463, + "learning_rate": 2.498e-05, + "loss": 0.0168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1250 + }, + { + "epoch": 0.3665987780040733, + "flow_matching_loss": 0.016337019857019186, + "grad_norm": 0.0886448323726654, + "learning_rate": 2.5180000000000003e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1260 + }, + { + "epoch": 0.36950829211521674, + "flow_matching_loss": 0.015102312434464693, + "grad_norm": 0.08856600522994995, + "learning_rate": 2.538e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1270 + }, + { + "epoch": 0.37241780622636017, + "flow_matching_loss": 0.018126980494707824, + "grad_norm": 0.08168669790029526, + "learning_rate": 2.5580000000000002e-05, + "loss": 0.0181, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1280 + }, + { + "epoch": 0.37532732033750366, + "flow_matching_loss": 0.014618378411978484, + "grad_norm": 0.11214173585176468, + "learning_rate": 2.578e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1290 + }, + { + "epoch": 0.3782368344486471, + "flow_matching_loss": 0.015554662700742483, + "grad_norm": 0.11293197423219681, + "learning_rate": 2.5980000000000002e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1300 + }, + { + "epoch": 0.3811463485597905, + "flow_matching_loss": 0.015929086692631246, + "grad_norm": 0.10778398811817169, + "learning_rate": 2.618e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1310 + }, + { + "epoch": 0.38405586267093395, + "flow_matching_loss": 0.016345554497092962, + "grad_norm": 0.09025412797927856, + "learning_rate": 2.6379999999999998e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1320 + }, + { + "epoch": 0.3869653767820774, + "flow_matching_loss": 0.01676767449826002, + "grad_norm": 0.08893383294343948, + "learning_rate": 2.658e-05, + "loss": 0.0168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1330 + }, + { + "epoch": 0.3898748908932208, + "flow_matching_loss": 0.015627908799797297, + "grad_norm": 0.09564722329378128, + "learning_rate": 2.678e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1340 + }, + { + "epoch": 0.3927844050043643, + "flow_matching_loss": 0.016911721508949994, + "grad_norm": 0.07986468076705933, + "learning_rate": 2.698e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1350 + }, + { + "epoch": 0.3956939191155077, + "flow_matching_loss": 0.015991545375436546, + "grad_norm": 0.09491477906703949, + "learning_rate": 2.718e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1360 + }, + { + "epoch": 0.39860343322665115, + "flow_matching_loss": 0.015804121643304823, + "grad_norm": 0.09121556580066681, + "learning_rate": 2.738e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1370 + }, + { + "epoch": 0.4015129473377946, + "flow_matching_loss": 0.013968724571168423, + "grad_norm": 0.0767732486128807, + "learning_rate": 2.758e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1380 + }, + { + "epoch": 0.404422461448938, + "flow_matching_loss": 0.01600760743021965, + "grad_norm": 0.06543821096420288, + "learning_rate": 2.778e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1390 + }, + { + "epoch": 0.4073319755600815, + "flow_matching_loss": 0.015721397288143635, + "grad_norm": 0.09912458062171936, + "learning_rate": 2.798e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1400 + }, + { + "epoch": 0.4102414896712249, + "flow_matching_loss": 0.017925867810845375, + "grad_norm": 0.0783025249838829, + "learning_rate": 2.818e-05, + "loss": 0.0179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1410 + }, + { + "epoch": 0.41315100378236835, + "flow_matching_loss": 0.014763960614800453, + "grad_norm": 0.07134222984313965, + "learning_rate": 2.838e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1420 + }, + { + "epoch": 0.4160605178935118, + "flow_matching_loss": 0.0166135735809803, + "grad_norm": 0.11627771705389023, + "learning_rate": 2.858e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1430 + }, + { + "epoch": 0.4189700320046552, + "flow_matching_loss": 0.015861354023218154, + "grad_norm": 0.0854114443063736, + "learning_rate": 2.8780000000000002e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1440 + }, + { + "epoch": 0.42187954611579864, + "flow_matching_loss": 0.01644874569028616, + "grad_norm": 0.06866839528083801, + "learning_rate": 2.898e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1450 + }, + { + "epoch": 0.42478906022694213, + "flow_matching_loss": 0.01662604510784149, + "grad_norm": 0.09363388270139694, + "learning_rate": 2.9180000000000002e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1460 + }, + { + "epoch": 0.42769857433808556, + "flow_matching_loss": 0.018401786219328642, + "grad_norm": 0.08369138836860657, + "learning_rate": 2.938e-05, + "loss": 0.0184, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1470 + }, + { + "epoch": 0.430608088449229, + "flow_matching_loss": 0.0159255750477314, + "grad_norm": 0.09205451607704163, + "learning_rate": 2.958e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1480 + }, + { + "epoch": 0.4335176025603724, + "flow_matching_loss": 0.015114751923829318, + "grad_norm": 0.10235147923231125, + "learning_rate": 2.9780000000000003e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1490 + }, + { + "epoch": 0.43642711667151585, + "flow_matching_loss": 0.015605824161320924, + "grad_norm": 0.07863368839025497, + "learning_rate": 2.998e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1500 + }, + { + "epoch": 0.4393366307826593, + "flow_matching_loss": 0.01585973547771573, + "grad_norm": 0.10935433208942413, + "learning_rate": 2.9999992618302022e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1510 + }, + { + "epoch": 0.44224614489380276, + "flow_matching_loss": 0.01584339402616024, + "grad_norm": 0.0744730532169342, + "learning_rate": 2.999996710133069e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1520 + }, + { + "epoch": 0.4451556590049462, + "flow_matching_loss": 0.015457820892333985, + "grad_norm": 0.0926228016614914, + "learning_rate": 2.9999923357984923e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1530 + }, + { + "epoch": 0.4480651731160896, + "flow_matching_loss": 0.015662801824510096, + "grad_norm": 0.08962888270616531, + "learning_rate": 2.9999861388317874e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1540 + }, + { + "epoch": 0.45097468722723305, + "flow_matching_loss": 0.017265825439244507, + "grad_norm": 0.09763578325510025, + "learning_rate": 2.9999781192404847e-05, + "loss": 0.0173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1550 + }, + { + "epoch": 0.4538842013383765, + "flow_matching_loss": 0.01535323103889823, + "grad_norm": 0.07710423320531845, + "learning_rate": 2.9999682770343283e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1560 + }, + { + "epoch": 0.4567937154495199, + "flow_matching_loss": 0.01681101433932781, + "grad_norm": 0.10248290002346039, + "learning_rate": 2.9999566122252777e-05, + "loss": 0.0168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1570 + }, + { + "epoch": 0.4597032295606634, + "flow_matching_loss": 0.01509981220588088, + "grad_norm": 0.08196043968200684, + "learning_rate": 2.9999431248275065e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1580 + }, + { + "epoch": 0.4626127436718068, + "flow_matching_loss": 0.014915278647094965, + "grad_norm": 0.08630470186471939, + "learning_rate": 2.9999278148574027e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1590 + }, + { + "epoch": 0.46552225778295026, + "flow_matching_loss": 0.016337312944233417, + "grad_norm": 0.10081274807453156, + "learning_rate": 2.9999106823335705e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1600 + }, + { + "epoch": 0.4684317718940937, + "flow_matching_loss": 0.015847948379814625, + "grad_norm": 0.11917604506015778, + "learning_rate": 2.999891727276827e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1610 + }, + { + "epoch": 0.4713412860052371, + "flow_matching_loss": 0.014048641081899405, + "grad_norm": 0.09099290519952774, + "learning_rate": 2.9998709497102037e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1620 + }, + { + "epoch": 0.47425080011638054, + "flow_matching_loss": 0.017282144539058207, + "grad_norm": 0.11449512839317322, + "learning_rate": 2.9998483496589485e-05, + "loss": 0.0173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1630 + }, + { + "epoch": 0.47716031422752403, + "flow_matching_loss": 0.01644410751760006, + "grad_norm": 0.09334636479616165, + "learning_rate": 2.9998239271505217e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1640 + }, + { + "epoch": 0.48006982833866746, + "flow_matching_loss": 0.017910709511488675, + "grad_norm": 0.11116024851799011, + "learning_rate": 2.9997976822145995e-05, + "loss": 0.0179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1650 + }, + { + "epoch": 0.4829793424498109, + "flow_matching_loss": 0.016419069189578295, + "grad_norm": 0.09047851711511612, + "learning_rate": 2.9997696148830714e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1660 + }, + { + "epoch": 0.4858888565609543, + "flow_matching_loss": 0.016025992669165136, + "grad_norm": 0.11612973362207413, + "learning_rate": 2.9997397251900423e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1670 + }, + { + "epoch": 0.48879837067209775, + "flow_matching_loss": 0.015530321653932333, + "grad_norm": 0.08912108838558197, + "learning_rate": 2.999708013171831e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1680 + }, + { + "epoch": 0.4917078847832412, + "flow_matching_loss": 0.01637124689295888, + "grad_norm": 0.08805261552333832, + "learning_rate": 2.9996744788669705e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1690 + }, + { + "epoch": 0.49461739889438466, + "flow_matching_loss": 0.01684948680922389, + "grad_norm": 0.10708291083574295, + "learning_rate": 2.999639122316208e-05, + "loss": 0.0168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1700 + }, + { + "epoch": 0.4975269130055281, + "flow_matching_loss": 0.013820651825517416, + "grad_norm": 0.09772846847772598, + "learning_rate": 2.999601943562505e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1710 + }, + { + "epoch": 0.5004364271166715, + "flow_matching_loss": 0.017831390164792536, + "grad_norm": 0.10578726977109909, + "learning_rate": 2.9995629426510377e-05, + "loss": 0.0178, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1720 + }, + { + "epoch": 0.503345941227815, + "flow_matching_loss": 0.017720865644514562, + "grad_norm": 0.07313293218612671, + "learning_rate": 2.9995221196291952e-05, + "loss": 0.0177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1730 + }, + { + "epoch": 0.5062554553389584, + "flow_matching_loss": 0.015306872129440308, + "grad_norm": 0.08817677944898605, + "learning_rate": 2.999479474546582e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1740 + }, + { + "epoch": 0.5091649694501018, + "flow_matching_loss": 0.016216257214546205, + "grad_norm": 0.09979268908500671, + "learning_rate": 2.9994350074550153e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1750 + }, + { + "epoch": 0.5120744835612453, + "flow_matching_loss": 0.015187506657093763, + "grad_norm": 0.08311764150857925, + "learning_rate": 2.9993887184085273e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1760 + }, + { + "epoch": 0.5149839976723887, + "flow_matching_loss": 0.016869840770959856, + "grad_norm": 0.11061535775661469, + "learning_rate": 2.9993406074633628e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1770 + }, + { + "epoch": 0.5178935117835322, + "flow_matching_loss": 0.015945746283978223, + "grad_norm": 0.11239200085401535, + "learning_rate": 2.9992906746779824e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1780 + }, + { + "epoch": 0.5208030258946756, + "flow_matching_loss": 0.016041725780814885, + "grad_norm": 0.08242582529783249, + "learning_rate": 2.9992389201130576e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1790 + }, + { + "epoch": 0.523712540005819, + "flow_matching_loss": 0.016799916047602892, + "grad_norm": 0.08927048742771149, + "learning_rate": 2.999185343831476e-05, + "loss": 0.0168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1800 + }, + { + "epoch": 0.5266220541169625, + "flow_matching_loss": 0.016636504512280225, + "grad_norm": 0.0918080061674118, + "learning_rate": 2.999129945898338e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1810 + }, + { + "epoch": 0.5295315682281059, + "flow_matching_loss": 0.015211990289390086, + "grad_norm": 0.07234266400337219, + "learning_rate": 2.999072726380957e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1820 + }, + { + "epoch": 0.5324410823392494, + "flow_matching_loss": 0.01712953606620431, + "grad_norm": 0.08822014182806015, + "learning_rate": 2.9990136853488603e-05, + "loss": 0.0171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1830 + }, + { + "epoch": 0.5353505964503927, + "flow_matching_loss": 0.015452528931200504, + "grad_norm": 0.08645249158143997, + "learning_rate": 2.9989528228737878e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1840 + }, + { + "epoch": 0.5382601105615362, + "flow_matching_loss": 0.01622134502977133, + "grad_norm": 0.08115624636411667, + "learning_rate": 2.9988901390296935e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1850 + }, + { + "epoch": 0.5411696246726797, + "flow_matching_loss": 0.017270419839769603, + "grad_norm": 0.11149939149618149, + "learning_rate": 2.9988256338927442e-05, + "loss": 0.0173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1860 + }, + { + "epoch": 0.5440791387838231, + "flow_matching_loss": 0.015267344750463962, + "grad_norm": 0.09169314801692963, + "learning_rate": 2.9987593075413198e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1870 + }, + { + "epoch": 0.5469886528949666, + "flow_matching_loss": 0.016400597430765627, + "grad_norm": 0.07823923230171204, + "learning_rate": 2.9986911600560128e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1880 + }, + { + "epoch": 0.5498981670061099, + "flow_matching_loss": 0.015292754769325257, + "grad_norm": 0.08637567609548569, + "learning_rate": 2.998621191519629e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1890 + }, + { + "epoch": 0.5528076811172534, + "flow_matching_loss": 0.01694622850045562, + "grad_norm": 0.10091260075569153, + "learning_rate": 2.9985494020171873e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1900 + }, + { + "epoch": 0.5557171952283969, + "flow_matching_loss": 0.015246470738202334, + "grad_norm": 0.08579451590776443, + "learning_rate": 2.9984757916359176e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1910 + }, + { + "epoch": 0.5586267093395403, + "flow_matching_loss": 0.015100692491978407, + "grad_norm": 0.09607203304767609, + "learning_rate": 2.998400360465264e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1920 + }, + { + "epoch": 0.5615362234506838, + "flow_matching_loss": 0.015255189593881368, + "grad_norm": 0.08345182240009308, + "learning_rate": 2.998323108596883e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1930 + }, + { + "epoch": 0.5644457375618271, + "flow_matching_loss": 0.01583172185346484, + "grad_norm": 0.10705272108316422, + "learning_rate": 2.9982440361246422e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1940 + }, + { + "epoch": 0.5673552516729706, + "flow_matching_loss": 0.015219814889132977, + "grad_norm": 0.10308173298835754, + "learning_rate": 2.998163143144622e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1950 + }, + { + "epoch": 0.570264765784114, + "flow_matching_loss": 0.019253619853407143, + "grad_norm": 0.09773002564907074, + "learning_rate": 2.9980804297551155e-05, + "loss": 0.0193, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1960 + }, + { + "epoch": 0.5731742798952575, + "flow_matching_loss": 0.01508047441020608, + "grad_norm": 0.08537430316209793, + "learning_rate": 2.997995896056627e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1970 + }, + { + "epoch": 0.576083794006401, + "flow_matching_loss": 0.0168718627654016, + "grad_norm": 0.08941356837749481, + "learning_rate": 2.997909542151873e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1980 + }, + { + "epoch": 0.5789933081175443, + "flow_matching_loss": 0.018896566051989794, + "grad_norm": 0.09971275180578232, + "learning_rate": 2.9978213681457823e-05, + "loss": 0.0189, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1990 + }, + { + "epoch": 0.5819028222286878, + "flow_matching_loss": 0.017373247910290955, + "grad_norm": 0.11923934519290924, + "learning_rate": 2.9977313741454933e-05, + "loss": 0.0174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2000 + }, + { + "epoch": 0.5848123363398312, + "flow_matching_loss": 0.016128742042928933, + "grad_norm": 0.06426210701465607, + "learning_rate": 2.997639560260358e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2010 + }, + { + "epoch": 0.5877218504509747, + "flow_matching_loss": 0.015490683726966381, + "grad_norm": 0.08987841010093689, + "learning_rate": 2.9975459266019387e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2020 + }, + { + "epoch": 0.5906313645621182, + "flow_matching_loss": 0.015146133676171302, + "grad_norm": 0.09874237328767776, + "learning_rate": 2.9974504732840092e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2030 + }, + { + "epoch": 0.5935408786732616, + "flow_matching_loss": 0.01619178894907236, + "grad_norm": 0.08761755377054214, + "learning_rate": 2.997353200422555e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2040 + }, + { + "epoch": 0.596450392784405, + "flow_matching_loss": 0.01603592224419117, + "grad_norm": 0.08184168487787247, + "learning_rate": 2.99725410813577e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2050 + }, + { + "epoch": 0.5993599068955484, + "flow_matching_loss": 0.01620795261114836, + "grad_norm": 0.10693944990634918, + "learning_rate": 2.997153196544063e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2060 + }, + { + "epoch": 0.6022694210066919, + "flow_matching_loss": 0.016392103023827077, + "grad_norm": 0.08448222279548645, + "learning_rate": 2.997050465770049e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2070 + }, + { + "epoch": 0.6051789351178353, + "flow_matching_loss": 0.016469165869057178, + "grad_norm": 0.07597212493419647, + "learning_rate": 2.9969459159385562e-05, + "loss": 0.0165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2080 + }, + { + "epoch": 0.6080884492289788, + "flow_matching_loss": 0.01636964697390795, + "grad_norm": 0.09906696528196335, + "learning_rate": 2.9968395471766237e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2090 + }, + { + "epoch": 0.6109979633401222, + "flow_matching_loss": 0.014984338171780109, + "grad_norm": 0.06699585914611816, + "learning_rate": 2.996731359613498e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2100 + }, + { + "epoch": 0.6139074774512656, + "flow_matching_loss": 0.015583224780857562, + "grad_norm": 0.09187520295381546, + "learning_rate": 2.996621353380638e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2110 + }, + { + "epoch": 0.6168169915624091, + "flow_matching_loss": 0.016640197206288575, + "grad_norm": 0.09415948390960693, + "learning_rate": 2.9965095286117113e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2120 + }, + { + "epoch": 0.6197265056735525, + "flow_matching_loss": 0.015109818242490292, + "grad_norm": 0.07334505021572113, + "learning_rate": 2.996395885442596e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2130 + }, + { + "epoch": 0.622636019784696, + "flow_matching_loss": 0.01574248615652323, + "grad_norm": 0.08964814990758896, + "learning_rate": 2.996280424011379e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2140 + }, + { + "epoch": 0.6255455338958394, + "flow_matching_loss": 0.016912514064460992, + "grad_norm": 0.10822220891714096, + "learning_rate": 2.9961631444583565e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2150 + }, + { + "epoch": 0.6284550480069828, + "flow_matching_loss": 0.01608455879613757, + "grad_norm": 0.09133346378803253, + "learning_rate": 2.9960440469260347e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2160 + }, + { + "epoch": 0.6313645621181263, + "flow_matching_loss": 0.016654092725366353, + "grad_norm": 0.07134044170379639, + "learning_rate": 2.995923131559128e-05, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2170 + }, + { + "epoch": 0.6342740762292697, + "flow_matching_loss": 0.015461845509707927, + "grad_norm": 0.08459217101335526, + "learning_rate": 2.99580039850456e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2180 + }, + { + "epoch": 0.6371835903404132, + "flow_matching_loss": 0.01697914069518447, + "grad_norm": 0.08910060673952103, + "learning_rate": 2.9956758479114635e-05, + "loss": 0.017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2190 + }, + { + "epoch": 0.6400931044515565, + "flow_matching_loss": 0.016714745108038186, + "grad_norm": 0.0933796763420105, + "learning_rate": 2.9955494799311787e-05, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2200 + }, + { + "epoch": 0.6430026185627, + "flow_matching_loss": 0.016644869279116393, + "grad_norm": 0.09966463595628738, + "learning_rate": 2.995421294717254e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2210 + }, + { + "epoch": 0.6459121326738435, + "flow_matching_loss": 0.016202248726040126, + "grad_norm": 0.11431661248207092, + "learning_rate": 2.995291292425448e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2220 + }, + { + "epoch": 0.6488216467849869, + "flow_matching_loss": 0.01475918861106038, + "grad_norm": 0.09202208369970322, + "learning_rate": 2.9951594732137248e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2230 + }, + { + "epoch": 0.6517311608961304, + "flow_matching_loss": 0.015611295774579047, + "grad_norm": 0.09436698257923126, + "learning_rate": 2.995025837242257e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2240 + }, + { + "epoch": 0.6546406750072737, + "flow_matching_loss": 0.016679289005696773, + "grad_norm": 0.10409297794103622, + "learning_rate": 2.994890384673426e-05, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2250 + }, + { + "epoch": 0.6575501891184172, + "flow_matching_loss": 0.013993483781814576, + "grad_norm": 0.10225618630647659, + "learning_rate": 2.9947531156718174e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2260 + }, + { + "epoch": 0.6604597032295607, + "flow_matching_loss": 0.015739514771848918, + "grad_norm": 0.10408040881156921, + "learning_rate": 2.994614030404228e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2270 + }, + { + "epoch": 0.6633692173407041, + "flow_matching_loss": 0.016490384191274642, + "grad_norm": 0.07608851045370102, + "learning_rate": 2.994473129039659e-05, + "loss": 0.0165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2280 + }, + { + "epoch": 0.6662787314518476, + "flow_matching_loss": 0.015639153216034173, + "grad_norm": 0.07670115679502487, + "learning_rate": 2.9943304117493183e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2290 + }, + { + "epoch": 0.669188245562991, + "flow_matching_loss": 0.016838644444942475, + "grad_norm": 0.08951390534639359, + "learning_rate": 2.9941858787066207e-05, + "loss": 0.0168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2300 + }, + { + "epoch": 0.6720977596741344, + "flow_matching_loss": 0.016990699619054795, + "grad_norm": 0.08044449239969254, + "learning_rate": 2.9940395300871888e-05, + "loss": 0.017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2310 + }, + { + "epoch": 0.6750072737852778, + "flow_matching_loss": 0.01551761869341135, + "grad_norm": 0.08063614368438721, + "learning_rate": 2.993891366068849e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2320 + }, + { + "epoch": 0.6779167878964213, + "flow_matching_loss": 0.014976986590772868, + "grad_norm": 0.08054924011230469, + "learning_rate": 2.993741386831634e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2330 + }, + { + "epoch": 0.6808263020075648, + "flow_matching_loss": 0.015969914756715296, + "grad_norm": 0.08964931219816208, + "learning_rate": 2.9935895925577837e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2340 + }, + { + "epoch": 0.6837358161187082, + "flow_matching_loss": 0.017782678082585334, + "grad_norm": 0.08368004113435745, + "learning_rate": 2.9934359834317425e-05, + "loss": 0.0178, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2350 + }, + { + "epoch": 0.6866453302298516, + "flow_matching_loss": 0.016935631167143583, + "grad_norm": 0.09442535042762756, + "learning_rate": 2.993280559640159e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2360 + }, + { + "epoch": 0.689554844340995, + "flow_matching_loss": 0.016029003262519836, + "grad_norm": 0.1001591756939888, + "learning_rate": 2.9931233213718887e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2370 + }, + { + "epoch": 0.6924643584521385, + "flow_matching_loss": 0.017900536954402923, + "grad_norm": 0.09709059447050095, + "learning_rate": 2.992964268817991e-05, + "loss": 0.0179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2380 + }, + { + "epoch": 0.695373872563282, + "flow_matching_loss": 0.017416270822286604, + "grad_norm": 0.10440242290496826, + "learning_rate": 2.9928034021717286e-05, + "loss": 0.0174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2390 + }, + { + "epoch": 0.6982833866744254, + "flow_matching_loss": 0.0157083828933537, + "grad_norm": 0.09669389575719833, + "learning_rate": 2.9926407216285707e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2400 + }, + { + "epoch": 0.7011929007855688, + "flow_matching_loss": 0.018429565988481045, + "grad_norm": 0.08342131972312927, + "learning_rate": 2.9924762273861894e-05, + "loss": 0.0184, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2410 + }, + { + "epoch": 0.7041024148967122, + "flow_matching_loss": 0.01691278014332056, + "grad_norm": 0.09969466179609299, + "learning_rate": 2.992309919644461e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2420 + }, + { + "epoch": 0.7070119290078557, + "flow_matching_loss": 0.01615206366404891, + "grad_norm": 0.08412396162748337, + "learning_rate": 2.9921417986054633e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2430 + }, + { + "epoch": 0.7099214431189991, + "flow_matching_loss": 0.016608106344938277, + "grad_norm": 0.08501747995615005, + "learning_rate": 2.9919718644734815e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2440 + }, + { + "epoch": 0.7128309572301426, + "flow_matching_loss": 0.015919825248420238, + "grad_norm": 0.07168353348970413, + "learning_rate": 2.991800117455e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2450 + }, + { + "epoch": 0.715740471341286, + "flow_matching_loss": 0.01701553277671337, + "grad_norm": 0.07371684163808823, + "learning_rate": 2.9916265577587085e-05, + "loss": 0.017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2460 + }, + { + "epoch": 0.7186499854524294, + "flow_matching_loss": 0.01670596282929182, + "grad_norm": 0.10919386893510818, + "learning_rate": 2.9914511855954986e-05, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2470 + }, + { + "epoch": 0.7215594995635729, + "flow_matching_loss": 0.016396622359752654, + "grad_norm": 0.06361756473779678, + "learning_rate": 2.9912740011784626e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2480 + }, + { + "epoch": 0.7244690136747163, + "flow_matching_loss": 0.01685114810243249, + "grad_norm": 0.07896296679973602, + "learning_rate": 2.991095004722898e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2490 + }, + { + "epoch": 0.7273785277858598, + "flow_matching_loss": 0.016981018614023922, + "grad_norm": 0.11163397878408432, + "learning_rate": 2.9909141964463017e-05, + "loss": 0.017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2500 + }, + { + "epoch": 0.7302880418970032, + "flow_matching_loss": 0.01677221655845642, + "grad_norm": 0.09986890852451324, + "learning_rate": 2.990731576568372e-05, + "loss": 0.0168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2510 + }, + { + "epoch": 0.7331975560081466, + "flow_matching_loss": 0.015867187455296517, + "grad_norm": 0.09201356768608093, + "learning_rate": 2.99054714531101e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2520 + }, + { + "epoch": 0.7361070701192901, + "flow_matching_loss": 0.015338861662894488, + "grad_norm": 0.09386633336544037, + "learning_rate": 2.9903609028983178e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2530 + }, + { + "epoch": 0.7390165842304335, + "flow_matching_loss": 0.014985625725239515, + "grad_norm": 0.10667502135038376, + "learning_rate": 2.990172849556596e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2540 + }, + { + "epoch": 0.741926098341577, + "flow_matching_loss": 0.01717310706153512, + "grad_norm": 0.0933995470404625, + "learning_rate": 2.989982985514348e-05, + "loss": 0.0172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2550 + }, + { + "epoch": 0.7448356124527203, + "flow_matching_loss": 0.01582259703427553, + "grad_norm": 0.08809196203947067, + "learning_rate": 2.9897913110022766e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2560 + }, + { + "epoch": 0.7477451265638638, + "flow_matching_loss": 0.017111106775701046, + "grad_norm": 0.10787299275398254, + "learning_rate": 2.989597826253284e-05, + "loss": 0.0171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2570 + }, + { + "epoch": 0.7506546406750073, + "flow_matching_loss": 0.017232571076601744, + "grad_norm": 0.11479879915714264, + "learning_rate": 2.989402531502473e-05, + "loss": 0.0172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2580 + }, + { + "epoch": 0.7535641547861507, + "flow_matching_loss": 0.01467562112957239, + "grad_norm": 0.09450816363096237, + "learning_rate": 2.989205426987144e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2590 + }, + { + "epoch": 0.7564736688972942, + "flow_matching_loss": 0.016261705476790667, + "grad_norm": 0.09440387040376663, + "learning_rate": 2.9890065129467988e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2600 + }, + { + "epoch": 0.7593831830084375, + "flow_matching_loss": 0.015833113063126802, + "grad_norm": 0.08569105714559555, + "learning_rate": 2.9888057896231364e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2610 + }, + { + "epoch": 0.762292697119581, + "flow_matching_loss": 0.016328735928982496, + "grad_norm": 0.11266880482435226, + "learning_rate": 2.988603257260054e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2620 + }, + { + "epoch": 0.7652022112307245, + "flow_matching_loss": 0.01755548743531108, + "grad_norm": 0.10675384104251862, + "learning_rate": 2.988398916103648e-05, + "loss": 0.0176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2630 + }, + { + "epoch": 0.7681117253418679, + "flow_matching_loss": 0.016064578481018544, + "grad_norm": 0.09908812493085861, + "learning_rate": 2.988192766402212e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2640 + }, + { + "epoch": 0.7710212394530114, + "flow_matching_loss": 0.015885993465781213, + "grad_norm": 0.10399903357028961, + "learning_rate": 2.9879848084062378e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2650 + }, + { + "epoch": 0.7739307535641547, + "flow_matching_loss": 0.015781211759895088, + "grad_norm": 0.11218878626823425, + "learning_rate": 2.9877750423684132e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2660 + }, + { + "epoch": 0.7768402676752982, + "flow_matching_loss": 0.01544955037534237, + "grad_norm": 0.09387079626321793, + "learning_rate": 2.987563468543624e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2670 + }, + { + "epoch": 0.7797497817864416, + "flow_matching_loss": 0.0153374457731843, + "grad_norm": 0.0970073714852333, + "learning_rate": 2.9873500871889533e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2680 + }, + { + "epoch": 0.7826592958975851, + "flow_matching_loss": 0.015189025457948447, + "grad_norm": 0.108725905418396, + "learning_rate": 2.9871348985636778e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2690 + }, + { + "epoch": 0.7855688100087286, + "flow_matching_loss": 0.016608563624322413, + "grad_norm": 0.0726596787571907, + "learning_rate": 2.986917902929273e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2700 + }, + { + "epoch": 0.788478324119872, + "flow_matching_loss": 0.01712124515324831, + "grad_norm": 0.1065329760313034, + "learning_rate": 2.9866991005494095e-05, + "loss": 0.0171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2710 + }, + { + "epoch": 0.7913878382310154, + "flow_matching_loss": 0.014456420857459306, + "grad_norm": 0.08049347251653671, + "learning_rate": 2.986478491689952e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2720 + }, + { + "epoch": 0.7942973523421588, + "flow_matching_loss": 0.014657068345695735, + "grad_norm": 0.07452212274074554, + "learning_rate": 2.986256076618961e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2730 + }, + { + "epoch": 0.7972068664533023, + "flow_matching_loss": 0.01555299423635006, + "grad_norm": 0.10264317691326141, + "learning_rate": 2.986031855606692e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2740 + }, + { + "epoch": 0.8001163805644458, + "flow_matching_loss": 0.017379239294677974, + "grad_norm": 0.09825549274682999, + "learning_rate": 2.9858058289255954e-05, + "loss": 0.0174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2750 + }, + { + "epoch": 0.8030258946755892, + "flow_matching_loss": 0.016170600056648256, + "grad_norm": 0.1024656742811203, + "learning_rate": 2.985577996850314e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2760 + }, + { + "epoch": 0.8059354087867326, + "flow_matching_loss": 0.0155223336070776, + "grad_norm": 0.10024957358837128, + "learning_rate": 2.9853483596576856e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2770 + }, + { + "epoch": 0.808844922897876, + "flow_matching_loss": 0.017741239909082652, + "grad_norm": 0.12146308273077011, + "learning_rate": 2.9851169176267412e-05, + "loss": 0.0177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2780 + }, + { + "epoch": 0.8117544370090195, + "flow_matching_loss": 0.017440488375723363, + "grad_norm": 0.08758268505334854, + "learning_rate": 2.9848836710387044e-05, + "loss": 0.0174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2790 + }, + { + "epoch": 0.814663951120163, + "flow_matching_loss": 0.01788812056183815, + "grad_norm": 0.09241093695163727, + "learning_rate": 2.9846486201769917e-05, + "loss": 0.0179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2800 + }, + { + "epoch": 0.8175734652313064, + "flow_matching_loss": 0.01815031738951802, + "grad_norm": 0.08483022451400757, + "learning_rate": 2.9844117653272122e-05, + "loss": 0.0182, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2810 + }, + { + "epoch": 0.8204829793424498, + "flow_matching_loss": 0.015972456987947227, + "grad_norm": 0.09558390825986862, + "learning_rate": 2.9841731067771676e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2820 + }, + { + "epoch": 0.8233924934535932, + "flow_matching_loss": 0.015946618840098382, + "grad_norm": 0.0959169790148735, + "learning_rate": 2.9839326448168492e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2830 + }, + { + "epoch": 0.8263020075647367, + "flow_matching_loss": 0.016741358023136855, + "grad_norm": 0.0939500704407692, + "learning_rate": 2.983690379738442e-05, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2840 + }, + { + "epoch": 0.8292115216758801, + "flow_matching_loss": 0.015819682739675045, + "grad_norm": 0.07937254011631012, + "learning_rate": 2.9834463118363204e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2850 + }, + { + "epoch": 0.8321210357870236, + "flow_matching_loss": 0.01683386489748955, + "grad_norm": 0.10468815267086029, + "learning_rate": 2.98320044140705e-05, + "loss": 0.0168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2860 + }, + { + "epoch": 0.835030549898167, + "flow_matching_loss": 0.01750215943902731, + "grad_norm": 0.0836903527379036, + "learning_rate": 2.9829527687493868e-05, + "loss": 0.0175, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2870 + }, + { + "epoch": 0.8379400640093104, + "flow_matching_loss": 0.014626704342663287, + "grad_norm": 0.07783964276313782, + "learning_rate": 2.9827032941642765e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2880 + }, + { + "epoch": 0.8408495781204539, + "flow_matching_loss": 0.016271307412534953, + "grad_norm": 0.0985247865319252, + "learning_rate": 2.9824520179548535e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2890 + }, + { + "epoch": 0.8437590922315973, + "flow_matching_loss": 0.015171524416655303, + "grad_norm": 0.0955333337187767, + "learning_rate": 2.9821989404264428e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2900 + }, + { + "epoch": 0.8466686063427408, + "flow_matching_loss": 0.016922454349696636, + "grad_norm": 0.0749974176287651, + "learning_rate": 2.9819440618865572e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2910 + }, + { + "epoch": 0.8495781204538843, + "flow_matching_loss": 0.015287903137505054, + "grad_norm": 0.08082201331853867, + "learning_rate": 2.981687382644898e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2920 + }, + { + "epoch": 0.8524876345650276, + "flow_matching_loss": 0.015850541088730095, + "grad_norm": 0.10351870208978653, + "learning_rate": 2.9814289030133556e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2930 + }, + { + "epoch": 0.8553971486761711, + "flow_matching_loss": 0.015937585104256868, + "grad_norm": 0.08927464485168457, + "learning_rate": 2.9811686233060058e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2940 + }, + { + "epoch": 0.8583066627873145, + "flow_matching_loss": 0.014586389437317849, + "grad_norm": 0.06647220999002457, + "learning_rate": 2.9809065438391138e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2950 + }, + { + "epoch": 0.861216176898458, + "flow_matching_loss": 0.016487871948629617, + "grad_norm": 0.08481042087078094, + "learning_rate": 2.9806426649311304e-05, + "loss": 0.0165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2960 + }, + { + "epoch": 0.8641256910096013, + "flow_matching_loss": 0.017198460176587105, + "grad_norm": 0.10276798158884048, + "learning_rate": 2.980376986902694e-05, + "loss": 0.0172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2970 + }, + { + "epoch": 0.8670352051207448, + "flow_matching_loss": 0.015980089642107488, + "grad_norm": 0.09061210602521896, + "learning_rate": 2.9801095100766276e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2980 + }, + { + "epoch": 0.8699447192318883, + "flow_matching_loss": 0.016231342405080795, + "grad_norm": 0.09074253588914871, + "learning_rate": 2.979840234777942e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2990 + }, + { + "epoch": 0.8728542333430317, + "flow_matching_loss": 0.015892937127500773, + "grad_norm": 0.09270940721035004, + "learning_rate": 2.979569161333831e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3000 + }, + { + "epoch": 0.8757637474541752, + "flow_matching_loss": 0.016931360960006712, + "grad_norm": 0.1009821891784668, + "learning_rate": 2.9792962900736744e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3010 + }, + { + "epoch": 0.8786732615653186, + "flow_matching_loss": 0.0159324555657804, + "grad_norm": 0.07667790353298187, + "learning_rate": 2.9790216213290372e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3020 + }, + { + "epoch": 0.881582775676462, + "flow_matching_loss": 0.01596696684136987, + "grad_norm": 0.10454988479614258, + "learning_rate": 2.9787451554336672e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3030 + }, + { + "epoch": 0.8844922897876055, + "flow_matching_loss": 0.017317077610641718, + "grad_norm": 0.07537317276000977, + "learning_rate": 2.9784668927234972e-05, + "loss": 0.0173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3040 + }, + { + "epoch": 0.8874018038987489, + "flow_matching_loss": 0.015397998411208392, + "grad_norm": 0.10559376329183578, + "learning_rate": 2.9781868335366418e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3050 + }, + { + "epoch": 0.8903113180098924, + "flow_matching_loss": 0.018185536842793226, + "grad_norm": 0.09139693528413773, + "learning_rate": 2.9779049782134e-05, + "loss": 0.0182, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3060 + }, + { + "epoch": 0.8932208321210358, + "flow_matching_loss": 0.016809832584112883, + "grad_norm": 0.10013652592897415, + "learning_rate": 2.9776213270962523e-05, + "loss": 0.0168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3070 + }, + { + "epoch": 0.8961303462321792, + "flow_matching_loss": 0.015187861397862434, + "grad_norm": 0.0906466394662857, + "learning_rate": 2.977335880529861e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3080 + }, + { + "epoch": 0.8990398603433226, + "flow_matching_loss": 0.014146106224507094, + "grad_norm": 0.10253210365772247, + "learning_rate": 2.977048638861072e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3090 + }, + { + "epoch": 0.9019493744544661, + "flow_matching_loss": 0.01687113819643855, + "grad_norm": 0.07659277319908142, + "learning_rate": 2.97675960243891e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3100 + }, + { + "epoch": 0.9048588885656096, + "flow_matching_loss": 0.01613765601068735, + "grad_norm": 0.09557303786277771, + "learning_rate": 2.976468771614582e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3110 + }, + { + "epoch": 0.907768402676753, + "flow_matching_loss": 0.015961189568042756, + "grad_norm": 0.10221809148788452, + "learning_rate": 2.9761761467414736e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3120 + }, + { + "epoch": 0.9106779167878964, + "flow_matching_loss": 0.01444694334641099, + "grad_norm": 0.07659067958593369, + "learning_rate": 2.9758817281751538e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3130 + }, + { + "epoch": 0.9135874308990398, + "flow_matching_loss": 0.01441199630498886, + "grad_norm": 0.09965319186449051, + "learning_rate": 2.975585516273367e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3140 + }, + { + "epoch": 0.9164969450101833, + "flow_matching_loss": 0.016243280656635762, + "grad_norm": 0.0802665650844574, + "learning_rate": 2.975287511396039e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3150 + }, + { + "epoch": 0.9194064591213268, + "flow_matching_loss": 0.015320717450231313, + "grad_norm": 0.10038921982049942, + "learning_rate": 2.9749877139052745e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3160 + }, + { + "epoch": 0.9223159732324702, + "flow_matching_loss": 0.015933105349540712, + "grad_norm": 0.08445850759744644, + "learning_rate": 2.9746861241653555e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3170 + }, + { + "epoch": 0.9252254873436137, + "flow_matching_loss": 0.017772769462317227, + "grad_norm": 0.10432998836040497, + "learning_rate": 2.9743827425427416e-05, + "loss": 0.0178, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3180 + }, + { + "epoch": 0.928135001454757, + "flow_matching_loss": 0.015445035602897405, + "grad_norm": 0.07340362668037415, + "learning_rate": 2.9740775694060704e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3190 + }, + { + "epoch": 0.9310445155659005, + "flow_matching_loss": 0.01721729300916195, + "grad_norm": 0.09550003707408905, + "learning_rate": 2.973770605126156e-05, + "loss": 0.0172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3200 + }, + { + "epoch": 0.9339540296770439, + "flow_matching_loss": 0.01720297262072563, + "grad_norm": 0.07866744697093964, + "learning_rate": 2.9734618500759886e-05, + "loss": 0.0172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3210 + }, + { + "epoch": 0.9368635437881874, + "flow_matching_loss": 0.016104437969624998, + "grad_norm": 0.11032714694738388, + "learning_rate": 2.9731513046307356e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3220 + }, + { + "epoch": 0.9397730578993309, + "flow_matching_loss": 0.015968777425587177, + "grad_norm": 0.11408781260251999, + "learning_rate": 2.972838969167738e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3230 + }, + { + "epoch": 0.9426825720104742, + "flow_matching_loss": 0.014829024486243726, + "grad_norm": 0.09033020585775375, + "learning_rate": 2.972524844066513e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3240 + }, + { + "epoch": 0.9455920861216177, + "flow_matching_loss": 0.017127299308776857, + "grad_norm": 0.07750748097896576, + "learning_rate": 2.9722089297087524e-05, + "loss": 0.0171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3250 + }, + { + "epoch": 0.9485016002327611, + "flow_matching_loss": 0.01614375030621886, + "grad_norm": 0.07390369474887848, + "learning_rate": 2.971891226478322e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3260 + }, + { + "epoch": 0.9514111143439046, + "flow_matching_loss": 0.014631124678999186, + "grad_norm": 0.09828458726406097, + "learning_rate": 2.9715717347612615e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3270 + }, + { + "epoch": 0.9543206284550481, + "flow_matching_loss": 0.0166591739282012, + "grad_norm": 0.09566663205623627, + "learning_rate": 2.971250454945783e-05, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3280 + }, + { + "epoch": 0.9572301425661914, + "flow_matching_loss": 0.015363987069576979, + "grad_norm": 0.09057138860225677, + "learning_rate": 2.970927387422272e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3290 + }, + { + "epoch": 0.9601396566773349, + "flow_matching_loss": 0.017006424069404603, + "grad_norm": 0.09373276680707932, + "learning_rate": 2.9706025325832858e-05, + "loss": 0.017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3300 + }, + { + "epoch": 0.9630491707884783, + "flow_matching_loss": 0.017132984194904565, + "grad_norm": 0.09518782049417496, + "learning_rate": 2.9702758908235534e-05, + "loss": 0.0171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3310 + }, + { + "epoch": 0.9659586848996218, + "flow_matching_loss": 0.018417852465063334, + "grad_norm": 0.10393598675727844, + "learning_rate": 2.9699474625399765e-05, + "loss": 0.0184, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3320 + }, + { + "epoch": 0.9688681990107652, + "flow_matching_loss": 0.017365238256752492, + "grad_norm": 0.07102219015359879, + "learning_rate": 2.9696172481316246e-05, + "loss": 0.0174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3330 + }, + { + "epoch": 0.9717777131219086, + "flow_matching_loss": 0.016018005181103946, + "grad_norm": 0.08020409196615219, + "learning_rate": 2.9692852479997407e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3340 + }, + { + "epoch": 0.9746872272330521, + "flow_matching_loss": 0.015828883182257412, + "grad_norm": 0.08268134295940399, + "learning_rate": 2.9689514625477364e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3350 + }, + { + "epoch": 0.9775967413441955, + "flow_matching_loss": 0.01825162163004279, + "grad_norm": 0.08725753426551819, + "learning_rate": 2.968615892181192e-05, + "loss": 0.0183, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3360 + }, + { + "epoch": 0.980506255455339, + "flow_matching_loss": 0.015790660679340363, + "grad_norm": 0.08080431073904037, + "learning_rate": 2.9682785373078567e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3370 + }, + { + "epoch": 0.9834157695664824, + "flow_matching_loss": 0.01810668520629406, + "grad_norm": 0.08331674337387085, + "learning_rate": 2.9679393983376493e-05, + "loss": 0.0181, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3380 + }, + { + "epoch": 0.9863252836776258, + "flow_matching_loss": 0.01687465561553836, + "grad_norm": 0.09624046832323074, + "learning_rate": 2.9675984756826545e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3390 + }, + { + "epoch": 0.9892347977887693, + "flow_matching_loss": 0.01613456355407834, + "grad_norm": 0.08344940841197968, + "learning_rate": 2.9672557697571273e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3400 + }, + { + "epoch": 0.9921443118999127, + "flow_matching_loss": 0.01856667585670948, + "grad_norm": 0.09230127185583115, + "learning_rate": 2.966911280977486e-05, + "loss": 0.0186, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3410 + }, + { + "epoch": 0.9950538260110562, + "flow_matching_loss": 0.016099588945508003, + "grad_norm": 0.07966925203800201, + "learning_rate": 2.966565009762318e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3420 + }, + { + "epoch": 0.9979633401221996, + "flow_matching_loss": 0.016219480987638236, + "grad_norm": 0.08818753063678741, + "learning_rate": 2.966216956532375e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3430 + }, + { + "epoch": 1.000872854233343, + "flow_matching_loss": 0.016615560930222272, + "grad_norm": 0.09289617836475372, + "learning_rate": 2.9658671217105744e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3440 + }, + { + "epoch": 1.0037823683444864, + "flow_matching_loss": 0.016114189010113477, + "grad_norm": 0.10166028887033463, + "learning_rate": 2.965515505721999e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3450 + }, + { + "epoch": 1.00669188245563, + "flow_matching_loss": 0.017192125134170056, + "grad_norm": 0.08819559961557388, + "learning_rate": 2.9651621089938948e-05, + "loss": 0.0172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3460 + }, + { + "epoch": 1.0096013965667734, + "flow_matching_loss": 0.015334953460842371, + "grad_norm": 0.08512591570615768, + "learning_rate": 2.9648069319556725e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3470 + }, + { + "epoch": 1.0125109106779169, + "flow_matching_loss": 0.01617780402302742, + "grad_norm": 0.08543789386749268, + "learning_rate": 2.9644499750389057e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3480 + }, + { + "epoch": 1.0154204247890601, + "flow_matching_loss": 0.016725917905569078, + "grad_norm": 0.08488797396421432, + "learning_rate": 2.9640912386773305e-05, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3490 + }, + { + "epoch": 1.0183299389002036, + "flow_matching_loss": 0.014781213831156492, + "grad_norm": 0.08511478453874588, + "learning_rate": 2.9637307233068455e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3500 + }, + { + "epoch": 1.021239453011347, + "flow_matching_loss": 0.01575693255290389, + "grad_norm": 0.09099441021680832, + "learning_rate": 2.9633684293655108e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3510 + }, + { + "epoch": 1.0241489671224906, + "flow_matching_loss": 0.015436618588864803, + "grad_norm": 0.07233893871307373, + "learning_rate": 2.963004357293548e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3520 + }, + { + "epoch": 1.027058481233634, + "flow_matching_loss": 0.015026210620999337, + "grad_norm": 0.09412762522697449, + "learning_rate": 2.962638507533339e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3530 + }, + { + "epoch": 1.0299679953447773, + "flow_matching_loss": 0.016016967594623566, + "grad_norm": 0.06896831095218658, + "learning_rate": 2.962270880529425e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3540 + }, + { + "epoch": 1.0328775094559208, + "flow_matching_loss": 0.01578781148418784, + "grad_norm": 0.08025539666414261, + "learning_rate": 2.9619014767285084e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3550 + }, + { + "epoch": 1.0357870235670643, + "flow_matching_loss": 0.01560348253697157, + "grad_norm": 0.11114704608917236, + "learning_rate": 2.9615302965794488e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3560 + }, + { + "epoch": 1.0386965376782078, + "flow_matching_loss": 0.01617269292473793, + "grad_norm": 0.08603175729513168, + "learning_rate": 2.961157340533266e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3570 + }, + { + "epoch": 1.0416060517893513, + "flow_matching_loss": 0.015083748660981656, + "grad_norm": 0.08137199282646179, + "learning_rate": 2.9607826090431364e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3580 + }, + { + "epoch": 1.0445155659004945, + "flow_matching_loss": 0.01611310299485922, + "grad_norm": 0.08478740602731705, + "learning_rate": 2.9604061025643936e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3590 + }, + { + "epoch": 1.047425080011638, + "flow_matching_loss": 0.0167066796682775, + "grad_norm": 0.12670022249221802, + "learning_rate": 2.9600278215545292e-05, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3600 + }, + { + "epoch": 1.0503345941227815, + "flow_matching_loss": 0.014966047368943691, + "grad_norm": 0.0924183577299118, + "learning_rate": 2.95964776647319e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3610 + }, + { + "epoch": 1.053244108233925, + "flow_matching_loss": 0.01755269579589367, + "grad_norm": 0.0791274830698967, + "learning_rate": 2.9592659377821787e-05, + "loss": 0.0176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3620 + }, + { + "epoch": 1.0561536223450685, + "flow_matching_loss": 0.01568504637107253, + "grad_norm": 0.08403309434652328, + "learning_rate": 2.9588823359454528e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3630 + }, + { + "epoch": 1.0590631364562118, + "flow_matching_loss": 0.016365532670170067, + "grad_norm": 0.08693163096904755, + "learning_rate": 2.958496961429126e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3640 + }, + { + "epoch": 1.0619726505673552, + "flow_matching_loss": 0.016349803376942872, + "grad_norm": 0.08726410567760468, + "learning_rate": 2.9581098147014628e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3650 + }, + { + "epoch": 1.0648821646784987, + "flow_matching_loss": 0.016823199577629566, + "grad_norm": 0.10159081965684891, + "learning_rate": 2.9577208962328843e-05, + "loss": 0.0168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3660 + }, + { + "epoch": 1.0677916787896422, + "flow_matching_loss": 0.016076928377151488, + "grad_norm": 0.0970849022269249, + "learning_rate": 2.9573302064959633e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3670 + }, + { + "epoch": 1.0707011929007855, + "flow_matching_loss": 0.017742085643112658, + "grad_norm": 0.08658982813358307, + "learning_rate": 2.9569377459654235e-05, + "loss": 0.0177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3680 + }, + { + "epoch": 1.073610707011929, + "flow_matching_loss": 0.01723613552749157, + "grad_norm": 0.09223171323537827, + "learning_rate": 2.9565435151181424e-05, + "loss": 0.0172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3690 + }, + { + "epoch": 1.0765202211230724, + "flow_matching_loss": 0.01659483155235648, + "grad_norm": 0.08370445668697357, + "learning_rate": 2.9561475144331473e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3700 + }, + { + "epoch": 1.079429735234216, + "flow_matching_loss": 0.014574976917356253, + "grad_norm": 0.07852961868047714, + "learning_rate": 2.9557497443916162e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3710 + }, + { + "epoch": 1.0823392493453594, + "flow_matching_loss": 0.016037733014672994, + "grad_norm": 0.10911324620246887, + "learning_rate": 2.9553502054768774e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3720 + }, + { + "epoch": 1.0852487634565027, + "flow_matching_loss": 0.016255978122353555, + "grad_norm": 0.08708459883928299, + "learning_rate": 2.954948898174409e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3730 + }, + { + "epoch": 1.0881582775676462, + "flow_matching_loss": 0.01662727938964963, + "grad_norm": 0.07759413123130798, + "learning_rate": 2.954545822971836e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3740 + }, + { + "epoch": 1.0910677916787896, + "flow_matching_loss": 0.015681768022477626, + "grad_norm": 0.08765118569135666, + "learning_rate": 2.9541409803589335e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3750 + }, + { + "epoch": 1.0939773057899331, + "flow_matching_loss": 0.016492433194071055, + "grad_norm": 0.09213028103113174, + "learning_rate": 2.9537343708276235e-05, + "loss": 0.0165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3760 + }, + { + "epoch": 1.0968868199010766, + "flow_matching_loss": 0.015980620589107274, + "grad_norm": 0.08095767349004745, + "learning_rate": 2.953325994871975e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3770 + }, + { + "epoch": 1.0997963340122199, + "flow_matching_loss": 0.016639428120106457, + "grad_norm": 0.08973095566034317, + "learning_rate": 2.9529158529882023e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3780 + }, + { + "epoch": 1.1027058481233634, + "flow_matching_loss": 0.014321097265928983, + "grad_norm": 0.07980693876743317, + "learning_rate": 2.9525039456746677e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3790 + }, + { + "epoch": 1.1056153622345068, + "flow_matching_loss": 0.015039610117673874, + "grad_norm": 0.08762512356042862, + "learning_rate": 2.952090273431877e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3800 + }, + { + "epoch": 1.1085248763456503, + "flow_matching_loss": 0.015380711667239666, + "grad_norm": 0.09394770115613937, + "learning_rate": 2.9516748367624808e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3810 + }, + { + "epoch": 1.1114343904567938, + "flow_matching_loss": 0.016951782628893852, + "grad_norm": 0.08707533776760101, + "learning_rate": 2.951257636171274e-05, + "loss": 0.017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3820 + }, + { + "epoch": 1.114343904567937, + "flow_matching_loss": 0.018123923521488904, + "grad_norm": 0.07820697873830795, + "learning_rate": 2.9508386721651944e-05, + "loss": 0.0181, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3830 + }, + { + "epoch": 1.1172534186790806, + "flow_matching_loss": 0.016757317073643207, + "grad_norm": 0.088571697473526, + "learning_rate": 2.9504179452533228e-05, + "loss": 0.0168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3840 + }, + { + "epoch": 1.120162932790224, + "flow_matching_loss": 0.01632948648184538, + "grad_norm": 0.08047845214605331, + "learning_rate": 2.9499954559468826e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3850 + }, + { + "epoch": 1.1230724469013675, + "flow_matching_loss": 0.015580816939473153, + "grad_norm": 0.11495048552751541, + "learning_rate": 2.9495712047592378e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3860 + }, + { + "epoch": 1.1259819610125108, + "flow_matching_loss": 0.0163562698289752, + "grad_norm": 0.07969000935554504, + "learning_rate": 2.949145192205893e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3870 + }, + { + "epoch": 1.1288914751236543, + "flow_matching_loss": 0.01656046323478222, + "grad_norm": 0.0912156030535698, + "learning_rate": 2.9487174188044953e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3880 + }, + { + "epoch": 1.1318009892347978, + "flow_matching_loss": 0.017483649030327796, + "grad_norm": 0.0863465666770935, + "learning_rate": 2.9482878850748276e-05, + "loss": 0.0175, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3890 + }, + { + "epoch": 1.1347105033459413, + "flow_matching_loss": 0.016016273759305477, + "grad_norm": 0.09228784590959549, + "learning_rate": 2.9478565915388157e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3900 + }, + { + "epoch": 1.1376200174570847, + "flow_matching_loss": 0.014189983531832696, + "grad_norm": 0.07403957843780518, + "learning_rate": 2.9474235387205207e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3910 + }, + { + "epoch": 1.140529531568228, + "flow_matching_loss": 0.015946767292916775, + "grad_norm": 0.07962863892316818, + "learning_rate": 2.946988727146143e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3920 + }, + { + "epoch": 1.1434390456793715, + "flow_matching_loss": 0.015845112036913633, + "grad_norm": 0.09777285158634186, + "learning_rate": 2.9465521573440196e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3930 + }, + { + "epoch": 1.146348559790515, + "flow_matching_loss": 0.01585151217877865, + "grad_norm": 0.08608304709196091, + "learning_rate": 2.946113829844624e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3940 + }, + { + "epoch": 1.1492580739016585, + "flow_matching_loss": 0.01646250197663903, + "grad_norm": 0.06913810968399048, + "learning_rate": 2.945673745180566e-05, + "loss": 0.0165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3950 + }, + { + "epoch": 1.152167588012802, + "flow_matching_loss": 0.016105422098189593, + "grad_norm": 0.09965198487043381, + "learning_rate": 2.945231903886589e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3960 + }, + { + "epoch": 1.1550771021239452, + "flow_matching_loss": 0.016550790704786777, + "grad_norm": 0.08811536431312561, + "learning_rate": 2.9447883064995728e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3970 + }, + { + "epoch": 1.1579866162350887, + "flow_matching_loss": 0.017178644984960557, + "grad_norm": 0.08233539760112762, + "learning_rate": 2.9443429535585288e-05, + "loss": 0.0172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3980 + }, + { + "epoch": 1.1608961303462322, + "flow_matching_loss": 0.018567059841006994, + "grad_norm": 0.0870196744799614, + "learning_rate": 2.9438958456046044e-05, + "loss": 0.0186, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3990 + }, + { + "epoch": 1.1638056444573757, + "flow_matching_loss": 0.016117499582469462, + "grad_norm": 0.09531106799840927, + "learning_rate": 2.9434469831810765e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4000 + }, + { + "epoch": 1.1667151585685192, + "flow_matching_loss": 0.014664914272725583, + "grad_norm": 0.07927317172288895, + "learning_rate": 2.9429963668333566e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4010 + }, + { + "epoch": 1.1696246726796624, + "flow_matching_loss": 0.016252793930470945, + "grad_norm": 0.09430635720491409, + "learning_rate": 2.9425439971089845e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4020 + }, + { + "epoch": 1.172534186790806, + "flow_matching_loss": 0.015343031100928783, + "grad_norm": 0.08712145686149597, + "learning_rate": 2.9420898745576336e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4030 + }, + { + "epoch": 1.1754437009019494, + "flow_matching_loss": 0.016081035323441028, + "grad_norm": 0.08925417065620422, + "learning_rate": 2.9416339997311044e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4040 + }, + { + "epoch": 1.1783532150130929, + "flow_matching_loss": 0.016970966290682555, + "grad_norm": 0.11795789003372192, + "learning_rate": 2.941176373183329e-05, + "loss": 0.017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4050 + }, + { + "epoch": 1.1812627291242364, + "flow_matching_loss": 0.01656428724527359, + "grad_norm": 0.09292424470186234, + "learning_rate": 2.9407169954703653e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4060 + }, + { + "epoch": 1.1841722432353796, + "flow_matching_loss": 0.015590407978743314, + "grad_norm": 0.07924865931272507, + "learning_rate": 2.9402558671504018e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4070 + }, + { + "epoch": 1.187081757346523, + "flow_matching_loss": 0.017001086752861737, + "grad_norm": 0.07695787400007248, + "learning_rate": 2.9397929887837524e-05, + "loss": 0.017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4080 + }, + { + "epoch": 1.1899912714576666, + "flow_matching_loss": 0.015659332182258366, + "grad_norm": 0.10103918612003326, + "learning_rate": 2.939328360932858e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4090 + }, + { + "epoch": 1.19290078556881, + "flow_matching_loss": 0.017010574508458377, + "grad_norm": 0.08434239774942398, + "learning_rate": 2.938861984162285e-05, + "loss": 0.017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4100 + }, + { + "epoch": 1.1958102996799536, + "flow_matching_loss": 0.01499585248529911, + "grad_norm": 0.1128140464425087, + "learning_rate": 2.938393859038726e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4110 + }, + { + "epoch": 1.1987198137910968, + "flow_matching_loss": 0.015547191258519888, + "grad_norm": 0.08268184214830399, + "learning_rate": 2.9379239861309964e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4120 + }, + { + "epoch": 1.2016293279022403, + "flow_matching_loss": 0.015956207178533077, + "grad_norm": 0.08538118749856949, + "learning_rate": 2.937452366010036e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4130 + }, + { + "epoch": 1.2045388420133838, + "flow_matching_loss": 0.015788362361490726, + "grad_norm": 0.07652311027050018, + "learning_rate": 2.936978999248908e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4140 + }, + { + "epoch": 1.2074483561245273, + "flow_matching_loss": 0.016201620269566775, + "grad_norm": 0.08550862222909927, + "learning_rate": 2.936503886422798e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4150 + }, + { + "epoch": 1.2103578702356708, + "flow_matching_loss": 0.01528551634401083, + "grad_norm": 0.09913728386163712, + "learning_rate": 2.9360270281090122e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4160 + }, + { + "epoch": 1.213267384346814, + "flow_matching_loss": 0.016641767509281636, + "grad_norm": 0.07358705997467041, + "learning_rate": 2.935548424886979e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4170 + }, + { + "epoch": 1.2161768984579575, + "flow_matching_loss": 0.016561183892190457, + "grad_norm": 0.10114070773124695, + "learning_rate": 2.935068077338246e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4180 + }, + { + "epoch": 1.219086412569101, + "flow_matching_loss": 0.015250001288950443, + "grad_norm": 0.08409279584884644, + "learning_rate": 2.934585986046481e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4190 + }, + { + "epoch": 1.2219959266802445, + "flow_matching_loss": 0.016552151180803776, + "grad_norm": 0.0864701122045517, + "learning_rate": 2.93410215159747e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4200 + }, + { + "epoch": 1.2249054407913877, + "flow_matching_loss": 0.01600694814696908, + "grad_norm": 0.07716823369264603, + "learning_rate": 2.933616574579118e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4210 + }, + { + "epoch": 1.2278149549025312, + "flow_matching_loss": 0.015896876249462365, + "grad_norm": 0.09650634974241257, + "learning_rate": 2.933129255581447e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4220 + }, + { + "epoch": 1.2307244690136747, + "flow_matching_loss": 0.017721747793257237, + "grad_norm": 0.07677017152309418, + "learning_rate": 2.9326401951965954e-05, + "loss": 0.0177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4230 + }, + { + "epoch": 1.2336339831248182, + "flow_matching_loss": 0.015361812431365252, + "grad_norm": 0.089948371052742, + "learning_rate": 2.9321493940188165e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4240 + }, + { + "epoch": 1.2365434972359617, + "flow_matching_loss": 0.016257422976195812, + "grad_norm": 0.08698765188455582, + "learning_rate": 2.931656852644482e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4250 + }, + { + "epoch": 1.239453011347105, + "flow_matching_loss": 0.015607860311865807, + "grad_norm": 0.0890323594212532, + "learning_rate": 2.9311625716720748e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4260 + }, + { + "epoch": 1.2423625254582484, + "flow_matching_loss": 0.01657609352841973, + "grad_norm": 0.09483136981725693, + "learning_rate": 2.9306665517021927e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4270 + }, + { + "epoch": 1.245272039569392, + "flow_matching_loss": 0.015322017669677734, + "grad_norm": 0.12060000747442245, + "learning_rate": 2.9301687933375483e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4280 + }, + { + "epoch": 1.2481815536805354, + "flow_matching_loss": 0.018026558868587016, + "grad_norm": 0.08563448488712311, + "learning_rate": 2.929669297182963e-05, + "loss": 0.018, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4290 + }, + { + "epoch": 1.2510910677916787, + "flow_matching_loss": 0.016061354894191027, + "grad_norm": 0.10009265691041946, + "learning_rate": 2.929168063845373e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4300 + }, + { + "epoch": 1.2540005819028222, + "flow_matching_loss": 0.016318678855895996, + "grad_norm": 0.07046043872833252, + "learning_rate": 2.9286650939338236e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4310 + }, + { + "epoch": 1.2569100960139656, + "flow_matching_loss": 0.016283846646547317, + "grad_norm": 0.0837203711271286, + "learning_rate": 2.9281603880594706e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4320 + }, + { + "epoch": 1.2598196101251091, + "flow_matching_loss": 0.017209051735699178, + "grad_norm": 0.09586206823587418, + "learning_rate": 2.9276539468355796e-05, + "loss": 0.0172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4330 + }, + { + "epoch": 1.2627291242362526, + "flow_matching_loss": 0.01689931647852063, + "grad_norm": 0.09266902506351471, + "learning_rate": 2.9271457708775244e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4340 + }, + { + "epoch": 1.2656386383473959, + "flow_matching_loss": 0.014378159865736961, + "grad_norm": 0.10606838762760162, + "learning_rate": 2.9266358608027863e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4350 + }, + { + "epoch": 1.2685481524585394, + "flow_matching_loss": 0.01667276890948415, + "grad_norm": 0.09279535710811615, + "learning_rate": 2.9261242172309543e-05, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4360 + }, + { + "epoch": 1.2714576665696828, + "flow_matching_loss": 0.016001065168529748, + "grad_norm": 0.09450732916593552, + "learning_rate": 2.925610840783723e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4370 + }, + { + "epoch": 1.2743671806808263, + "flow_matching_loss": 0.01472763791680336, + "grad_norm": 0.06998295336961746, + "learning_rate": 2.9250957320848946e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4380 + }, + { + "epoch": 1.2772766947919698, + "flow_matching_loss": 0.01653761798515916, + "grad_norm": 0.09177134186029434, + "learning_rate": 2.924578891760374e-05, + "loss": 0.0165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4390 + }, + { + "epoch": 1.280186208903113, + "flow_matching_loss": 0.015977238956838846, + "grad_norm": 0.11563343554735184, + "learning_rate": 2.924060320438171e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4400 + }, + { + "epoch": 1.2830957230142566, + "flow_matching_loss": 0.016215786058455705, + "grad_norm": 0.0975961834192276, + "learning_rate": 2.9235400187483978e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4410 + }, + { + "epoch": 1.2860052371254, + "flow_matching_loss": 0.016983539145439864, + "grad_norm": 0.086777463555336, + "learning_rate": 2.9230179873232716e-05, + "loss": 0.017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4420 + }, + { + "epoch": 1.2889147512365435, + "flow_matching_loss": 0.015727082919329404, + "grad_norm": 0.10123638063669205, + "learning_rate": 2.922494226797108e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4430 + }, + { + "epoch": 1.291824265347687, + "flow_matching_loss": 0.01718115322291851, + "grad_norm": 0.18616802990436554, + "learning_rate": 2.9219687378063275e-05, + "loss": 0.0172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4440 + }, + { + "epoch": 1.2947337794588303, + "flow_matching_loss": 0.017085175681859254, + "grad_norm": 0.07660215348005295, + "learning_rate": 2.921441520989447e-05, + "loss": 0.0171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4450 + }, + { + "epoch": 1.2976432935699738, + "flow_matching_loss": 0.016543583106249572, + "grad_norm": 0.08019988983869553, + "learning_rate": 2.9209125769870854e-05, + "loss": 0.0165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4460 + }, + { + "epoch": 1.3005528076811173, + "flow_matching_loss": 0.018261649087071417, + "grad_norm": 0.06852743029594421, + "learning_rate": 2.92038190644196e-05, + "loss": 0.0183, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4470 + }, + { + "epoch": 1.3034623217922607, + "flow_matching_loss": 0.01691235974431038, + "grad_norm": 0.09524425119161606, + "learning_rate": 2.9198495099988843e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4480 + }, + { + "epoch": 1.3063718359034042, + "flow_matching_loss": 0.015737817622721195, + "grad_norm": 0.07411278039216995, + "learning_rate": 2.9193153883047712e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4490 + }, + { + "epoch": 1.3092813500145475, + "flow_matching_loss": 0.016882221214473248, + "grad_norm": 0.09614377468824387, + "learning_rate": 2.9187795420086287e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4500 + }, + { + "epoch": 1.312190864125691, + "flow_matching_loss": 0.018073664605617525, + "grad_norm": 0.0873216912150383, + "learning_rate": 2.9182419717615602e-05, + "loss": 0.0181, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4510 + }, + { + "epoch": 1.3151003782368345, + "flow_matching_loss": 0.01692214170470834, + "grad_norm": 0.11277292668819427, + "learning_rate": 2.917702678216765e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4520 + }, + { + "epoch": 1.318009892347978, + "flow_matching_loss": 0.015783969312906265, + "grad_norm": 0.0775379091501236, + "learning_rate": 2.9171616620295347e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4530 + }, + { + "epoch": 1.3209194064591214, + "flow_matching_loss": 0.01685552317649126, + "grad_norm": 0.11512119323015213, + "learning_rate": 2.9166189238572554e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4540 + }, + { + "epoch": 1.3238289205702647, + "flow_matching_loss": 0.016050492972135545, + "grad_norm": 0.08007847517728806, + "learning_rate": 2.9160744643594052e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4550 + }, + { + "epoch": 1.3267384346814082, + "flow_matching_loss": 0.015197829715907574, + "grad_norm": 0.07782476395368576, + "learning_rate": 2.915528284197554e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4560 + }, + { + "epoch": 1.3296479487925517, + "flow_matching_loss": 0.015608631819486619, + "grad_norm": 0.10626310855150223, + "learning_rate": 2.914980384035362e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4570 + }, + { + "epoch": 1.3325574629036951, + "flow_matching_loss": 0.01636894615367055, + "grad_norm": 0.07692534476518631, + "learning_rate": 2.9144307645385787e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4580 + }, + { + "epoch": 1.3354669770148386, + "flow_matching_loss": 0.01624302435666323, + "grad_norm": 0.0927756205201149, + "learning_rate": 2.9138794263750454e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4590 + }, + { + "epoch": 1.338376491125982, + "flow_matching_loss": 0.016355537902563812, + "grad_norm": 0.095472551882267, + "learning_rate": 2.913326370214688e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4600 + }, + { + "epoch": 1.3412860052371254, + "flow_matching_loss": 0.015983693487942218, + "grad_norm": 0.07845505326986313, + "learning_rate": 2.9127715967295234e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4610 + }, + { + "epoch": 1.3441955193482689, + "flow_matching_loss": 0.015842099208384752, + "grad_norm": 0.06729034334421158, + "learning_rate": 2.912215106593653e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4620 + }, + { + "epoch": 1.3471050334594123, + "flow_matching_loss": 0.016081006918102502, + "grad_norm": 0.07335224002599716, + "learning_rate": 2.911656900483265e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4630 + }, + { + "epoch": 1.3500145475705558, + "flow_matching_loss": 0.01721614245325327, + "grad_norm": 0.07650701701641083, + "learning_rate": 2.9110969790766322e-05, + "loss": 0.0172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4640 + }, + { + "epoch": 1.352924061681699, + "flow_matching_loss": 0.01602906724438071, + "grad_norm": 0.07117973268032074, + "learning_rate": 2.910535343054112e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4650 + }, + { + "epoch": 1.3558335757928426, + "flow_matching_loss": 0.015232315007597207, + "grad_norm": 0.07777894288301468, + "learning_rate": 2.909971993098145e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4660 + }, + { + "epoch": 1.358743089903986, + "flow_matching_loss": 0.017398544866591693, + "grad_norm": 0.08582234382629395, + "learning_rate": 2.9094069298932552e-05, + "loss": 0.0174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4670 + }, + { + "epoch": 1.3616526040151296, + "flow_matching_loss": 0.016356722358614205, + "grad_norm": 0.07970980554819107, + "learning_rate": 2.908840154126047e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4680 + }, + { + "epoch": 1.364562118126273, + "flow_matching_loss": 0.013808551616966725, + "grad_norm": 0.07420185208320618, + "learning_rate": 2.9082716664852063e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4690 + }, + { + "epoch": 1.3674716322374163, + "flow_matching_loss": 0.017076992336660622, + "grad_norm": 0.07849185913801193, + "learning_rate": 2.9077014676615e-05, + "loss": 0.0171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4700 + }, + { + "epoch": 1.3703811463485598, + "flow_matching_loss": 0.015073524881154298, + "grad_norm": 0.09612323343753815, + "learning_rate": 2.9071295583477724e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4710 + }, + { + "epoch": 1.3732906604597033, + "flow_matching_loss": 0.015045350696891547, + "grad_norm": 0.10138226300477982, + "learning_rate": 2.906555939238948e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4720 + }, + { + "epoch": 1.3762001745708465, + "flow_matching_loss": 0.01647364441305399, + "grad_norm": 0.07208335399627686, + "learning_rate": 2.905980611032028e-05, + "loss": 0.0165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4730 + }, + { + "epoch": 1.3791096886819902, + "flow_matching_loss": 0.016636391170322895, + "grad_norm": 0.07950835675001144, + "learning_rate": 2.9054035744260904e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4740 + }, + { + "epoch": 1.3820192027931335, + "flow_matching_loss": 0.014941525924950839, + "grad_norm": 0.07850739359855652, + "learning_rate": 2.9048248301222885e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4750 + }, + { + "epoch": 1.384928716904277, + "flow_matching_loss": 0.016652335971593858, + "grad_norm": 0.07829860597848892, + "learning_rate": 2.904244378823852e-05, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4760 + }, + { + "epoch": 1.3878382310154205, + "flow_matching_loss": 0.014547477103769779, + "grad_norm": 0.09473945200443268, + "learning_rate": 2.9036622212360836e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4770 + }, + { + "epoch": 1.3907477451265637, + "flow_matching_loss": 0.015528455376625061, + "grad_norm": 0.09827444702386856, + "learning_rate": 2.9030783580663596e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4780 + }, + { + "epoch": 1.3936572592377072, + "flow_matching_loss": 0.015842401329427956, + "grad_norm": 0.09119103848934174, + "learning_rate": 2.902492790024129e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4790 + }, + { + "epoch": 1.3965667733488507, + "flow_matching_loss": 0.017023582570254804, + "grad_norm": 0.0951448529958725, + "learning_rate": 2.901905517820912e-05, + "loss": 0.017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4800 + }, + { + "epoch": 1.3994762874599942, + "flow_matching_loss": 0.014893618784844876, + "grad_norm": 0.07921819388866425, + "learning_rate": 2.9013165421703e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4810 + }, + { + "epoch": 1.4023858015711377, + "flow_matching_loss": 0.01694817328825593, + "grad_norm": 0.08631028980016708, + "learning_rate": 2.9007258637879534e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4820 + }, + { + "epoch": 1.405295315682281, + "flow_matching_loss": 0.015435223188251256, + "grad_norm": 0.07903553545475006, + "learning_rate": 2.9001334833916025e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4830 + }, + { + "epoch": 1.4082048297934244, + "flow_matching_loss": 0.014874486904591321, + "grad_norm": 0.06930951774120331, + "learning_rate": 2.899539401701046e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4840 + }, + { + "epoch": 1.411114343904568, + "flow_matching_loss": 0.015872561372816563, + "grad_norm": 0.08773857355117798, + "learning_rate": 2.8989436194381474e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4850 + }, + { + "epoch": 1.4140238580157114, + "flow_matching_loss": 0.01586693720892072, + "grad_norm": 0.09104348719120026, + "learning_rate": 2.8983461373268404e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4860 + }, + { + "epoch": 1.4169333721268549, + "flow_matching_loss": 0.016287946701049806, + "grad_norm": 0.10981693863868713, + "learning_rate": 2.8977469560931205e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4870 + }, + { + "epoch": 1.4198428862379981, + "flow_matching_loss": 0.015404810011386872, + "grad_norm": 0.08373432606458664, + "learning_rate": 2.8971460764650502e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4880 + }, + { + "epoch": 1.4227524003491416, + "flow_matching_loss": 0.016361122950911522, + "grad_norm": 0.09643609076738358, + "learning_rate": 2.8965434991727545e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4890 + }, + { + "epoch": 1.4256619144602851, + "flow_matching_loss": 0.01556221153587103, + "grad_norm": 0.09353429824113846, + "learning_rate": 2.895939224948422e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4900 + }, + { + "epoch": 1.4285714285714286, + "flow_matching_loss": 0.014056280814111232, + "grad_norm": 0.07999526709318161, + "learning_rate": 2.895333254526303e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4910 + }, + { + "epoch": 1.431480942682572, + "flow_matching_loss": 0.016375268064439296, + "grad_norm": 0.09745524823665619, + "learning_rate": 2.8947255886427084e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4920 + }, + { + "epoch": 1.4343904567937154, + "flow_matching_loss": 0.01643962152302265, + "grad_norm": 0.060994040220975876, + "learning_rate": 2.8941162280360095e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4930 + }, + { + "epoch": 1.4372999709048588, + "flow_matching_loss": 0.016496640536934136, + "grad_norm": 0.10324964672327042, + "learning_rate": 2.8935051734466368e-05, + "loss": 0.0165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4940 + }, + { + "epoch": 1.4402094850160023, + "flow_matching_loss": 0.016603867523372173, + "grad_norm": 0.07160889357328415, + "learning_rate": 2.8928924256170792e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4950 + }, + { + "epoch": 1.4431189991271458, + "flow_matching_loss": 0.016408909019082783, + "grad_norm": 0.08719894289970398, + "learning_rate": 2.8922779852918836e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4960 + }, + { + "epoch": 1.4460285132382893, + "flow_matching_loss": 0.017528209928423168, + "grad_norm": 0.09885963052511215, + "learning_rate": 2.8916618532176526e-05, + "loss": 0.0175, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4970 + }, + { + "epoch": 1.4489380273494326, + "flow_matching_loss": 0.015418567694723606, + "grad_norm": 0.10588334500789642, + "learning_rate": 2.891044030143045e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4980 + }, + { + "epoch": 1.451847541460576, + "flow_matching_loss": 0.014610941428691149, + "grad_norm": 0.0859362781047821, + "learning_rate": 2.890424516818774e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4990 + }, + { + "epoch": 1.4547570555717195, + "flow_matching_loss": 0.015713444165885448, + "grad_norm": 0.09071598201990128, + "learning_rate": 2.8898033139976057e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5000 + }, + { + "epoch": 1.457666569682863, + "flow_matching_loss": 0.015443787910044193, + "grad_norm": 0.10394571721553802, + "learning_rate": 2.889180422434362e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5010 + }, + { + "epoch": 1.4605760837940065, + "flow_matching_loss": 0.015315514244139194, + "grad_norm": 0.08940823376178741, + "learning_rate": 2.8885558428859136e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5020 + }, + { + "epoch": 1.4634855979051498, + "flow_matching_loss": 0.014877471048384905, + "grad_norm": 0.07137171924114227, + "learning_rate": 2.8879295761111835e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5030 + }, + { + "epoch": 1.4663951120162932, + "flow_matching_loss": 0.016936951130628587, + "grad_norm": 0.10004060715436935, + "learning_rate": 2.8873016228711452e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5040 + }, + { + "epoch": 1.4693046261274367, + "flow_matching_loss": 0.017105033993721007, + "grad_norm": 0.10752733051776886, + "learning_rate": 2.8866719839288215e-05, + "loss": 0.0171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5050 + }, + { + "epoch": 1.4722141402385802, + "flow_matching_loss": 0.015662490949034692, + "grad_norm": 0.08214660733938217, + "learning_rate": 2.886040660049283e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5060 + }, + { + "epoch": 1.4751236543497237, + "flow_matching_loss": 0.015451643615961075, + "grad_norm": 0.11266424506902695, + "learning_rate": 2.8854076519996467e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5070 + }, + { + "epoch": 1.478033168460867, + "flow_matching_loss": 0.013717983942478895, + "grad_norm": 0.07119423151016235, + "learning_rate": 2.884772960549078e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5080 + }, + { + "epoch": 1.4809426825720104, + "flow_matching_loss": 0.016919405199587345, + "grad_norm": 0.0893496572971344, + "learning_rate": 2.8841365864687867e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5090 + }, + { + "epoch": 1.483852196683154, + "flow_matching_loss": 0.01465538628399372, + "grad_norm": 0.09526645392179489, + "learning_rate": 2.8834985305320276e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5100 + }, + { + "epoch": 1.4867617107942974, + "flow_matching_loss": 0.014769438467919827, + "grad_norm": 0.08719686418771744, + "learning_rate": 2.882858793514098e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5110 + }, + { + "epoch": 1.489671224905441, + "flow_matching_loss": 0.015690541267395018, + "grad_norm": 0.1233711913228035, + "learning_rate": 2.8822173761923398e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5120 + }, + { + "epoch": 1.4925807390165842, + "flow_matching_loss": 0.014503569155931473, + "grad_norm": 0.10820773988962173, + "learning_rate": 2.8815742793461343e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5130 + }, + { + "epoch": 1.4954902531277277, + "flow_matching_loss": 0.014981897827237845, + "grad_norm": 0.09791743755340576, + "learning_rate": 2.880929503756906e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5140 + }, + { + "epoch": 1.4983997672388711, + "flow_matching_loss": 0.015650481823831795, + "grad_norm": 0.09877199679613113, + "learning_rate": 2.880283050208117e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5150 + }, + { + "epoch": 1.5013092813500144, + "flow_matching_loss": 0.01483117425814271, + "grad_norm": 0.07136408239603043, + "learning_rate": 2.8796349194852704e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5160 + }, + { + "epoch": 1.504218795461158, + "flow_matching_loss": 0.015813184902071953, + "grad_norm": 0.07664794474840164, + "learning_rate": 2.8789851123759052e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5170 + }, + { + "epoch": 1.5071283095723014, + "flow_matching_loss": 0.016522925812751055, + "grad_norm": 0.07714493572711945, + "learning_rate": 2.8783336296695985e-05, + "loss": 0.0165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5180 + }, + { + "epoch": 1.5100378236834449, + "flow_matching_loss": 0.01478433720767498, + "grad_norm": 0.11953578889369965, + "learning_rate": 2.877680472157964e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5190 + }, + { + "epoch": 1.5129473377945883, + "flow_matching_loss": 0.01763611948117614, + "grad_norm": 0.08311837911605835, + "learning_rate": 2.8770256406346483e-05, + "loss": 0.0176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5200 + }, + { + "epoch": 1.5158568519057316, + "flow_matching_loss": 0.01508103609085083, + "grad_norm": 0.0800619050860405, + "learning_rate": 2.8763691358953348e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5210 + }, + { + "epoch": 1.5187663660168753, + "flow_matching_loss": 0.014713621325790882, + "grad_norm": 0.06680182367563248, + "learning_rate": 2.8757109587377377e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5220 + }, + { + "epoch": 1.5216758801280186, + "flow_matching_loss": 0.01486662207171321, + "grad_norm": 0.07303626835346222, + "learning_rate": 2.8750511099616047e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5230 + }, + { + "epoch": 1.524585394239162, + "flow_matching_loss": 0.015369233302772045, + "grad_norm": 0.07977619767189026, + "learning_rate": 2.8743895903687143e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5240 + }, + { + "epoch": 1.5274949083503055, + "flow_matching_loss": 0.01732122730463743, + "grad_norm": 0.09101596474647522, + "learning_rate": 2.8737264007628753e-05, + "loss": 0.0173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5250 + }, + { + "epoch": 1.5304044224614488, + "flow_matching_loss": 0.014194305054843426, + "grad_norm": 0.07872718572616577, + "learning_rate": 2.8730615419499247e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5260 + }, + { + "epoch": 1.5333139365725925, + "flow_matching_loss": 0.017575482558459043, + "grad_norm": 0.09878865629434586, + "learning_rate": 2.8723950147377303e-05, + "loss": 0.0176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5270 + }, + { + "epoch": 1.5362234506837358, + "flow_matching_loss": 0.01617153696715832, + "grad_norm": 0.09206444770097733, + "learning_rate": 2.871726819936184e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5280 + }, + { + "epoch": 1.5391329647948793, + "flow_matching_loss": 0.015195518173277378, + "grad_norm": 0.06519578397274017, + "learning_rate": 2.8710569583572068e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5290 + }, + { + "epoch": 1.5420424789060228, + "flow_matching_loss": 0.014991163928061723, + "grad_norm": 0.0689903199672699, + "learning_rate": 2.870385430814742e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5300 + }, + { + "epoch": 1.544951993017166, + "flow_matching_loss": 0.015407795272767543, + "grad_norm": 0.07341072708368301, + "learning_rate": 2.8697122381247605e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5310 + }, + { + "epoch": 1.5478615071283097, + "flow_matching_loss": 0.0172335266135633, + "grad_norm": 0.07857664674520493, + "learning_rate": 2.8690373811052545e-05, + "loss": 0.0172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5320 + }, + { + "epoch": 1.550771021239453, + "flow_matching_loss": 0.014543556701391935, + "grad_norm": 0.08424032479524612, + "learning_rate": 2.8683608605762378e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5330 + }, + { + "epoch": 1.5536805353505965, + "flow_matching_loss": 0.016409151162952183, + "grad_norm": 0.08729743212461472, + "learning_rate": 2.8676826773597485e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5340 + }, + { + "epoch": 1.55659004946174, + "flow_matching_loss": 0.016196963284164666, + "grad_norm": 0.12099381536245346, + "learning_rate": 2.8670028322798413e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5350 + }, + { + "epoch": 1.5594995635728832, + "flow_matching_loss": 0.015498653892427683, + "grad_norm": 0.08215081691741943, + "learning_rate": 2.8663213261625932e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5360 + }, + { + "epoch": 1.562409077684027, + "flow_matching_loss": 0.01651291213929653, + "grad_norm": 0.08108015358448029, + "learning_rate": 2.865638159836098e-05, + "loss": 0.0165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5370 + }, + { + "epoch": 1.5653185917951702, + "flow_matching_loss": 0.01613965295255184, + "grad_norm": 0.10379652678966522, + "learning_rate": 2.8649533341304677e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5380 + }, + { + "epoch": 1.5682281059063137, + "flow_matching_loss": 0.013635441567748785, + "grad_norm": 0.08426789194345474, + "learning_rate": 2.8642668498778297e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5390 + }, + { + "epoch": 1.5711376200174572, + "flow_matching_loss": 0.016640089359134434, + "grad_norm": 0.10675352811813354, + "learning_rate": 2.8635787079123268e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5400 + }, + { + "epoch": 1.5740471341286004, + "flow_matching_loss": 0.01561229843646288, + "grad_norm": 0.10257070511579514, + "learning_rate": 2.8628889090701175e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5410 + }, + { + "epoch": 1.576956648239744, + "flow_matching_loss": 0.014604907855391503, + "grad_norm": 0.07513626664876938, + "learning_rate": 2.862197454189371e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5420 + }, + { + "epoch": 1.5798661623508874, + "flow_matching_loss": 0.015314889419823886, + "grad_norm": 0.08876118808984756, + "learning_rate": 2.8615043441102708e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5430 + }, + { + "epoch": 1.5827756764620309, + "flow_matching_loss": 0.015870345383882524, + "grad_norm": 0.07377567142248154, + "learning_rate": 2.860809579675011e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5440 + }, + { + "epoch": 1.5856851905731744, + "flow_matching_loss": 0.015169703494757414, + "grad_norm": 0.09700503945350647, + "learning_rate": 2.860113161727796e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5450 + }, + { + "epoch": 1.5885947046843176, + "flow_matching_loss": 0.015230753272771836, + "grad_norm": 0.08589129149913788, + "learning_rate": 2.8594150911148393e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5460 + }, + { + "epoch": 1.5915042187954611, + "flow_matching_loss": 0.016704594343900682, + "grad_norm": 0.08909369260072708, + "learning_rate": 2.8587153686843626e-05, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5470 + }, + { + "epoch": 1.5944137329066046, + "flow_matching_loss": 0.014659065753221512, + "grad_norm": 0.07649695873260498, + "learning_rate": 2.858013995286594e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5480 + }, + { + "epoch": 1.597323247017748, + "flow_matching_loss": 0.016513466089963912, + "grad_norm": 0.08189503103494644, + "learning_rate": 2.857310971773769e-05, + "loss": 0.0165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5490 + }, + { + "epoch": 1.6002327611288916, + "flow_matching_loss": 0.016637129429727792, + "grad_norm": 0.09481627494096756, + "learning_rate": 2.8566062990001274e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5500 + }, + { + "epoch": 1.6031422752400348, + "flow_matching_loss": 0.01613973779603839, + "grad_norm": 0.08587683737277985, + "learning_rate": 2.8558999778219126e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5510 + }, + { + "epoch": 1.6060517893511783, + "flow_matching_loss": 0.015141188632696867, + "grad_norm": 0.08702556043863297, + "learning_rate": 2.8551920090973722e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5520 + }, + { + "epoch": 1.6089613034623218, + "flow_matching_loss": 0.01668097460642457, + "grad_norm": 0.0829073116183281, + "learning_rate": 2.854482393686754e-05, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5530 + }, + { + "epoch": 1.6118708175734653, + "flow_matching_loss": 0.016475233994424344, + "grad_norm": 0.09708651155233383, + "learning_rate": 2.8537711324523086e-05, + "loss": 0.0165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5540 + }, + { + "epoch": 1.6147803316846088, + "flow_matching_loss": 0.015358355920761823, + "grad_norm": 0.09666385501623154, + "learning_rate": 2.8530582262582845e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5550 + }, + { + "epoch": 1.617689845795752, + "flow_matching_loss": 0.01618912722915411, + "grad_norm": 0.07488424330949783, + "learning_rate": 2.8523436759709306e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5560 + }, + { + "epoch": 1.6205993599068955, + "flow_matching_loss": 0.01661118809133768, + "grad_norm": 0.1028510332107544, + "learning_rate": 2.8516274824584926e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5570 + }, + { + "epoch": 1.623508874018039, + "flow_matching_loss": 0.013824503216892482, + "grad_norm": 0.07971043139696121, + "learning_rate": 2.8509096465912136e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5580 + }, + { + "epoch": 1.6264183881291823, + "flow_matching_loss": 0.016239992063492536, + "grad_norm": 0.0687912181019783, + "learning_rate": 2.8501901692413315e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5590 + }, + { + "epoch": 1.629327902240326, + "flow_matching_loss": 0.016985088866204024, + "grad_norm": 0.08911406248807907, + "learning_rate": 2.8494690512830788e-05, + "loss": 0.017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5600 + }, + { + "epoch": 1.6322374163514692, + "flow_matching_loss": 0.015051559079438448, + "grad_norm": 0.07682628929615021, + "learning_rate": 2.8487462935926823e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5610 + }, + { + "epoch": 1.6351469304626127, + "flow_matching_loss": 0.017676770873367786, + "grad_norm": 0.11416757106781006, + "learning_rate": 2.8480218970483606e-05, + "loss": 0.0177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5620 + }, + { + "epoch": 1.6380564445737562, + "flow_matching_loss": 0.01582254162058234, + "grad_norm": 0.07659206539392471, + "learning_rate": 2.847295862530324e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5630 + }, + { + "epoch": 1.6409659586848995, + "flow_matching_loss": 0.014524009451270104, + "grad_norm": 0.07369969040155411, + "learning_rate": 2.846568190920773e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5640 + }, + { + "epoch": 1.6438754727960432, + "flow_matching_loss": 0.013302547112107276, + "grad_norm": 0.07408090680837631, + "learning_rate": 2.8458388831038965e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5650 + }, + { + "epoch": 1.6467849869071864, + "flow_matching_loss": 0.014749534241855145, + "grad_norm": 0.0903811827301979, + "learning_rate": 2.8451079399658725e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5660 + }, + { + "epoch": 1.64969450101833, + "flow_matching_loss": 0.015471343882381916, + "grad_norm": 0.08682271093130112, + "learning_rate": 2.8443753623948665e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5670 + }, + { + "epoch": 1.6526040151294734, + "flow_matching_loss": 0.018160777073353528, + "grad_norm": 0.09576351195573807, + "learning_rate": 2.8436411512810288e-05, + "loss": 0.0182, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5680 + }, + { + "epoch": 1.6555135292406167, + "flow_matching_loss": 0.017002014443278313, + "grad_norm": 0.08625398576259613, + "learning_rate": 2.842905307516495e-05, + "loss": 0.017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5690 + }, + { + "epoch": 1.6584230433517604, + "flow_matching_loss": 0.016575361508876084, + "grad_norm": 0.08176717907190323, + "learning_rate": 2.8421678319953854e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5700 + }, + { + "epoch": 1.6613325574629036, + "flow_matching_loss": 0.015381514281034469, + "grad_norm": 0.07230891287326813, + "learning_rate": 2.841428725613801e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5710 + }, + { + "epoch": 1.6642420715740471, + "flow_matching_loss": 0.014833253249526024, + "grad_norm": 0.06226344779133797, + "learning_rate": 2.8406879892698272e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5720 + }, + { + "epoch": 1.6671515856851906, + "flow_matching_loss": 0.015636601112782957, + "grad_norm": 0.07280813902616501, + "learning_rate": 2.8399456238635274e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5730 + }, + { + "epoch": 1.6700610997963339, + "flow_matching_loss": 0.014433745201677084, + "grad_norm": 0.10202544182538986, + "learning_rate": 2.8392016302969457e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5740 + }, + { + "epoch": 1.6729706139074776, + "flow_matching_loss": 0.014466193318367005, + "grad_norm": 0.08230479061603546, + "learning_rate": 2.8384560094741055e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5750 + }, + { + "epoch": 1.6758801280186209, + "flow_matching_loss": 0.01669081449508667, + "grad_norm": 0.09611468762159348, + "learning_rate": 2.8377087623010054e-05, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5760 + }, + { + "epoch": 1.6787896421297643, + "flow_matching_loss": 0.017568223271518947, + "grad_norm": 0.11159799247980118, + "learning_rate": 2.8369598896856213e-05, + "loss": 0.0176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5770 + }, + { + "epoch": 1.6816991562409078, + "flow_matching_loss": 0.01721295788884163, + "grad_norm": 0.08086202293634415, + "learning_rate": 2.836209392537904e-05, + "loss": 0.0172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5780 + }, + { + "epoch": 1.684608670352051, + "flow_matching_loss": 0.015738803520798685, + "grad_norm": 0.1044156402349472, + "learning_rate": 2.8354572717697787e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5790 + }, + { + "epoch": 1.6875181844631948, + "flow_matching_loss": 0.01612101849168539, + "grad_norm": 0.09347093105316162, + "learning_rate": 2.8347035282951428e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5800 + }, + { + "epoch": 1.690427698574338, + "flow_matching_loss": 0.018791591469198465, + "grad_norm": 0.07631350308656693, + "learning_rate": 2.8339481630298653e-05, + "loss": 0.0188, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5810 + }, + { + "epoch": 1.6933372126854815, + "flow_matching_loss": 0.017710244469344615, + "grad_norm": 0.09524185955524445, + "learning_rate": 2.833191176891787e-05, + "loss": 0.0177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5820 + }, + { + "epoch": 1.696246726796625, + "flow_matching_loss": 0.015903646219521762, + "grad_norm": 0.07232221215963364, + "learning_rate": 2.832432570800717e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5830 + }, + { + "epoch": 1.6991562409077683, + "flow_matching_loss": 0.014573631342500448, + "grad_norm": 0.07607472687959671, + "learning_rate": 2.8316723456784332e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5840 + }, + { + "epoch": 1.702065755018912, + "flow_matching_loss": 0.015786456875503063, + "grad_norm": 0.10239914059638977, + "learning_rate": 2.8309105024486808e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5850 + }, + { + "epoch": 1.7049752691300553, + "flow_matching_loss": 0.01651978986337781, + "grad_norm": 0.08844476193189621, + "learning_rate": 2.830147042037172e-05, + "loss": 0.0165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5860 + }, + { + "epoch": 1.7078847832411987, + "flow_matching_loss": 0.016718499176204205, + "grad_norm": 0.08537891507148743, + "learning_rate": 2.8293819653715816e-05, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5870 + }, + { + "epoch": 1.7107942973523422, + "flow_matching_loss": 0.016448468156158923, + "grad_norm": 0.090235136449337, + "learning_rate": 2.828615273381551e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5880 + }, + { + "epoch": 1.7137038114634855, + "flow_matching_loss": 0.01527847144752741, + "grad_norm": 0.08011393249034882, + "learning_rate": 2.827846966998683e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5890 + }, + { + "epoch": 1.7166133255746292, + "flow_matching_loss": 0.016249282751232386, + "grad_norm": 0.08811885118484497, + "learning_rate": 2.8270770471565422e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5900 + }, + { + "epoch": 1.7195228396857725, + "flow_matching_loss": 0.01599040850996971, + "grad_norm": 0.0969032272696495, + "learning_rate": 2.826305514790654e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5910 + }, + { + "epoch": 1.722432353796916, + "flow_matching_loss": 0.017700706236064435, + "grad_norm": 0.1147921234369278, + "learning_rate": 2.825532370838503e-05, + "loss": 0.0177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5920 + }, + { + "epoch": 1.7253418679080594, + "flow_matching_loss": 0.015780264046043158, + "grad_norm": 0.08115257322788239, + "learning_rate": 2.824757616239532e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5930 + }, + { + "epoch": 1.7282513820192027, + "flow_matching_loss": 0.01610931484028697, + "grad_norm": 0.07152234762907028, + "learning_rate": 2.8239812519351404e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5940 + }, + { + "epoch": 1.7311608961303462, + "flow_matching_loss": 0.016589197982102632, + "grad_norm": 0.08282575011253357, + "learning_rate": 2.823203278868685e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5950 + }, + { + "epoch": 1.7340704102414897, + "flow_matching_loss": 0.016260738018900156, + "grad_norm": 0.08305250853300095, + "learning_rate": 2.8224236979854752e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5960 + }, + { + "epoch": 1.7369799243526332, + "flow_matching_loss": 0.01851731687784195, + "grad_norm": 0.0799955427646637, + "learning_rate": 2.821642510232776e-05, + "loss": 0.0185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5970 + }, + { + "epoch": 1.7398894384637766, + "flow_matching_loss": 0.01735373567789793, + "grad_norm": 0.0997990146279335, + "learning_rate": 2.8208597165598046e-05, + "loss": 0.0174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5980 + }, + { + "epoch": 1.74279895257492, + "flow_matching_loss": 0.016305083595216274, + "grad_norm": 0.09508252143859863, + "learning_rate": 2.8200753179177285e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5990 + }, + { + "epoch": 1.7457084666860634, + "flow_matching_loss": 0.015854191966354847, + "grad_norm": 0.0899110659956932, + "learning_rate": 2.819289315259666e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6000 + }, + { + "epoch": 1.7486179807972069, + "flow_matching_loss": 0.017766028828918933, + "grad_norm": 0.06545819342136383, + "learning_rate": 2.8185017095406845e-05, + "loss": 0.0178, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6010 + }, + { + "epoch": 1.7515274949083504, + "flow_matching_loss": 0.015063455421477556, + "grad_norm": 0.07263263314962387, + "learning_rate": 2.8177125017177998e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6020 + }, + { + "epoch": 1.7544370090194938, + "flow_matching_loss": 0.01509133167564869, + "grad_norm": 0.07296773046255112, + "learning_rate": 2.8169216927499727e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6030 + }, + { + "epoch": 1.757346523130637, + "flow_matching_loss": 0.014907889626920223, + "grad_norm": 0.08038290590047836, + "learning_rate": 2.816129283598112e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6040 + }, + { + "epoch": 1.7602560372417806, + "flow_matching_loss": 0.015729624778032303, + "grad_norm": 0.07915506511926651, + "learning_rate": 2.815335275225068e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6050 + }, + { + "epoch": 1.763165551352924, + "flow_matching_loss": 0.015504982322454453, + "grad_norm": 0.0985044538974762, + "learning_rate": 2.8145396685956374e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6060 + }, + { + "epoch": 1.7660750654640673, + "flow_matching_loss": 0.01632728148251772, + "grad_norm": 0.07785921543836594, + "learning_rate": 2.813742464676556e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6070 + }, + { + "epoch": 1.768984579575211, + "flow_matching_loss": 0.015218623168766498, + "grad_norm": 0.09563960880041122, + "learning_rate": 2.8129436644365025e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6080 + }, + { + "epoch": 1.7718940936863543, + "flow_matching_loss": 0.015406671911478043, + "grad_norm": 0.08362890034914017, + "learning_rate": 2.8121432688460938e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6090 + }, + { + "epoch": 1.7748036077974978, + "flow_matching_loss": 0.01670229248702526, + "grad_norm": 0.08071569353342056, + "learning_rate": 2.8113412788778867e-05, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6100 + }, + { + "epoch": 1.7777131219086413, + "flow_matching_loss": 0.015596924908459186, + "grad_norm": 0.08416204154491425, + "learning_rate": 2.8105376955063744e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6110 + }, + { + "epoch": 1.7806226360197845, + "flow_matching_loss": 0.015709175169467925, + "grad_norm": 0.08399364352226257, + "learning_rate": 2.8097325197079862e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6120 + }, + { + "epoch": 1.7835321501309283, + "flow_matching_loss": 0.015265069622546434, + "grad_norm": 0.07635396718978882, + "learning_rate": 2.808925752461087e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6130 + }, + { + "epoch": 1.7864416642420715, + "flow_matching_loss": 0.015560891013592482, + "grad_norm": 0.07111521810293198, + "learning_rate": 2.808117394745975e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6140 + }, + { + "epoch": 1.789351178353215, + "flow_matching_loss": 0.01582809556275606, + "grad_norm": 0.0976509377360344, + "learning_rate": 2.807307447544881e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6150 + }, + { + "epoch": 1.7922606924643585, + "flow_matching_loss": 0.016230324003845453, + "grad_norm": 0.06775064021348953, + "learning_rate": 2.806495911841967e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6160 + }, + { + "epoch": 1.7951702065755017, + "flow_matching_loss": 0.016257251054048537, + "grad_norm": 0.07692408561706543, + "learning_rate": 2.8056827886233255e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6170 + }, + { + "epoch": 1.7980797206866455, + "flow_matching_loss": 0.01621348997578025, + "grad_norm": 0.08354286849498749, + "learning_rate": 2.8048680788769783e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6180 + }, + { + "epoch": 1.8009892347977887, + "flow_matching_loss": 0.014351067412644625, + "grad_norm": 0.08376000821590424, + "learning_rate": 2.8040517835928742e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6190 + }, + { + "epoch": 1.8038987489089322, + "flow_matching_loss": 0.014863592479377985, + "grad_norm": 0.09383218735456467, + "learning_rate": 2.803233903762889e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6200 + }, + { + "epoch": 1.8068082630200757, + "flow_matching_loss": 0.01356146326288581, + "grad_norm": 0.08505436778068542, + "learning_rate": 2.8024144403808235e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6210 + }, + { + "epoch": 1.809717777131219, + "flow_matching_loss": 0.01622277870774269, + "grad_norm": 0.08028239011764526, + "learning_rate": 2.8015933944424037e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6220 + }, + { + "epoch": 1.8126272912423627, + "flow_matching_loss": 0.01490049185231328, + "grad_norm": 0.08047781139612198, + "learning_rate": 2.800770766945277e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6230 + }, + { + "epoch": 1.815536805353506, + "flow_matching_loss": 0.015668503381311892, + "grad_norm": 0.07818225026130676, + "learning_rate": 2.7999465588890135e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6240 + }, + { + "epoch": 1.8184463194646494, + "flow_matching_loss": 0.01592912571504712, + "grad_norm": 0.07926668226718903, + "learning_rate": 2.799120771275105e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6250 + }, + { + "epoch": 1.821355833575793, + "flow_matching_loss": 0.015243298746645451, + "grad_norm": 0.08820843696594238, + "learning_rate": 2.7982934051069597e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6260 + }, + { + "epoch": 1.8242653476869362, + "flow_matching_loss": 0.015750868059694768, + "grad_norm": 0.08638021349906921, + "learning_rate": 2.7974644613899063e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6270 + }, + { + "epoch": 1.8271748617980799, + "flow_matching_loss": 0.015649950597435236, + "grad_norm": 0.09678694605827332, + "learning_rate": 2.796633941131189e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6280 + }, + { + "epoch": 1.8300843759092231, + "flow_matching_loss": 0.015900793205946683, + "grad_norm": 0.09273884445428848, + "learning_rate": 2.795801845339969e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6290 + }, + { + "epoch": 1.8329938900203666, + "flow_matching_loss": 0.014910607412457467, + "grad_norm": 0.06879711896181107, + "learning_rate": 2.7949681750273208e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6300 + }, + { + "epoch": 1.83590340413151, + "flow_matching_loss": 0.01430373014882207, + "grad_norm": 0.07506968080997467, + "learning_rate": 2.7941329312062322e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6310 + }, + { + "epoch": 1.8388129182426534, + "flow_matching_loss": 0.015711913350969553, + "grad_norm": 0.0785423144698143, + "learning_rate": 2.7932961148916037e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6320 + }, + { + "epoch": 1.841722432353797, + "flow_matching_loss": 0.014200373366475106, + "grad_norm": 0.09327113628387451, + "learning_rate": 2.7924577271002456e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6330 + }, + { + "epoch": 1.8446319464649403, + "flow_matching_loss": 0.015215928945690394, + "grad_norm": 0.06317546963691711, + "learning_rate": 2.7916177688508782e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6340 + }, + { + "epoch": 1.8475414605760838, + "flow_matching_loss": 0.01567941140383482, + "grad_norm": 0.07583561539649963, + "learning_rate": 2.7907762411641302e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6350 + }, + { + "epoch": 1.8504509746872273, + "flow_matching_loss": 0.016836245730519294, + "grad_norm": 0.10079456865787506, + "learning_rate": 2.7899331450625367e-05, + "loss": 0.0168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6360 + }, + { + "epoch": 1.8533604887983706, + "flow_matching_loss": 0.01636140150949359, + "grad_norm": 0.09548244625329971, + "learning_rate": 2.7890884815705397e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6370 + }, + { + "epoch": 1.8562700029095143, + "flow_matching_loss": 0.017304816097021104, + "grad_norm": 0.08922358602285385, + "learning_rate": 2.788242251714484e-05, + "loss": 0.0173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6380 + }, + { + "epoch": 1.8591795170206575, + "flow_matching_loss": 0.015922659635543825, + "grad_norm": 0.09003841131925583, + "learning_rate": 2.787394456522619e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6390 + }, + { + "epoch": 1.862089031131801, + "flow_matching_loss": 0.017016487754881383, + "grad_norm": 0.09746212512254715, + "learning_rate": 2.7865450970250965e-05, + "loss": 0.017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6400 + }, + { + "epoch": 1.8649985452429445, + "flow_matching_loss": 0.015059055574238301, + "grad_norm": 0.09863533079624176, + "learning_rate": 2.7856941742539676e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6410 + }, + { + "epoch": 1.8679080593540878, + "flow_matching_loss": 0.01469008456915617, + "grad_norm": 0.08602042496204376, + "learning_rate": 2.784841689243184e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6420 + }, + { + "epoch": 1.8708175734652313, + "flow_matching_loss": 0.016509552393108605, + "grad_norm": 0.09672215580940247, + "learning_rate": 2.783987643028595e-05, + "loss": 0.0165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6430 + }, + { + "epoch": 1.8737270875763747, + "flow_matching_loss": 0.015484316367655992, + "grad_norm": 0.08779115229845047, + "learning_rate": 2.783132036647948e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6440 + }, + { + "epoch": 1.8766366016875182, + "flow_matching_loss": 0.015585191361606122, + "grad_norm": 0.07672059535980225, + "learning_rate": 2.7822748711408852e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6450 + }, + { + "epoch": 1.8795461157986617, + "flow_matching_loss": 0.014168043807148933, + "grad_norm": 0.10832560062408447, + "learning_rate": 2.7814161475489428e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6460 + }, + { + "epoch": 1.882455629909805, + "flow_matching_loss": 0.01571791395545006, + "grad_norm": 0.07998276501893997, + "learning_rate": 2.780555866915552e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6470 + }, + { + "epoch": 1.8853651440209485, + "flow_matching_loss": 0.016742059495300055, + "grad_norm": 0.07867638021707535, + "learning_rate": 2.7796940302860338e-05, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6480 + }, + { + "epoch": 1.888274658132092, + "flow_matching_loss": 0.014464244060218335, + "grad_norm": 0.08337190747261047, + "learning_rate": 2.7788306387076018e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6490 + }, + { + "epoch": 1.8911841722432354, + "flow_matching_loss": 0.01575704701244831, + "grad_norm": 0.07325877249240875, + "learning_rate": 2.7779656932293584e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6500 + }, + { + "epoch": 1.894093686354379, + "flow_matching_loss": 0.015483652148395777, + "grad_norm": 0.08800304681062698, + "learning_rate": 2.777099194902293e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6510 + }, + { + "epoch": 1.8970032004655222, + "flow_matching_loss": 0.014887611381709576, + "grad_norm": 0.08375705033540726, + "learning_rate": 2.776231144779283e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6520 + }, + { + "epoch": 1.8999127145766657, + "flow_matching_loss": 0.014849958010017872, + "grad_norm": 0.07960055023431778, + "learning_rate": 2.775361543915091e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6530 + }, + { + "epoch": 1.9028222286878091, + "flow_matching_loss": 0.01557031376287341, + "grad_norm": 0.08123334497213364, + "learning_rate": 2.774490393366365e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6540 + }, + { + "epoch": 1.9057317427989524, + "flow_matching_loss": 0.015818411204963922, + "grad_norm": 0.08479071408510208, + "learning_rate": 2.7736176941916346e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6550 + }, + { + "epoch": 1.9086412569100961, + "flow_matching_loss": 0.01599762151017785, + "grad_norm": 0.07876954972743988, + "learning_rate": 2.772743447451311e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6560 + }, + { + "epoch": 1.9115507710212394, + "flow_matching_loss": 0.01427690926939249, + "grad_norm": 0.10444455593824387, + "learning_rate": 2.7718676542076872e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6570 + }, + { + "epoch": 1.9144602851323829, + "flow_matching_loss": 0.014861188363283873, + "grad_norm": 0.06001394987106323, + "learning_rate": 2.770990315524934e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6580 + }, + { + "epoch": 1.9173697992435264, + "flow_matching_loss": 0.016089879628270864, + "grad_norm": 0.0998990386724472, + "learning_rate": 2.7701114324691015e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6590 + }, + { + "epoch": 1.9202793133546696, + "flow_matching_loss": 0.014735466428101062, + "grad_norm": 0.08763571083545685, + "learning_rate": 2.7692310061081145e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6600 + }, + { + "epoch": 1.9231888274658133, + "flow_matching_loss": 0.015196149237453938, + "grad_norm": 0.08724116533994675, + "learning_rate": 2.7683490375117742e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6610 + }, + { + "epoch": 1.9260983415769566, + "flow_matching_loss": 0.015487586986273527, + "grad_norm": 0.0793830081820488, + "learning_rate": 2.7674655277517556e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6620 + }, + { + "epoch": 1.9290078556881, + "flow_matching_loss": 0.015851244516670703, + "grad_norm": 0.12098550796508789, + "learning_rate": 2.7665804779016072e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6630 + }, + { + "epoch": 1.9319173697992436, + "flow_matching_loss": 0.014281152468174695, + "grad_norm": 0.0753508135676384, + "learning_rate": 2.765693889036747e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6640 + }, + { + "epoch": 1.9348268839103868, + "flow_matching_loss": 0.015518523007631301, + "grad_norm": 0.08056000620126724, + "learning_rate": 2.764805762234465e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6650 + }, + { + "epoch": 1.9377363980215305, + "flow_matching_loss": 0.01690604565665126, + "grad_norm": 0.08665195107460022, + "learning_rate": 2.7639160985739182e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6660 + }, + { + "epoch": 1.9406459121326738, + "flow_matching_loss": 0.01597912358120084, + "grad_norm": 0.07973069697618484, + "learning_rate": 2.7630248991361317e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6670 + }, + { + "epoch": 1.9435554262438173, + "flow_matching_loss": 0.017060137540102004, + "grad_norm": 0.10332511365413666, + "learning_rate": 2.762132165003998e-05, + "loss": 0.0171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6680 + }, + { + "epoch": 1.9464649403549608, + "flow_matching_loss": 0.014676640089601279, + "grad_norm": 0.08036736398935318, + "learning_rate": 2.7612378972622722e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6690 + }, + { + "epoch": 1.949374454466104, + "flow_matching_loss": 0.015904964972287418, + "grad_norm": 0.09946553409099579, + "learning_rate": 2.7603420969975743e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6700 + }, + { + "epoch": 1.9522839685772477, + "flow_matching_loss": 0.01660562166944146, + "grad_norm": 0.08789105713367462, + "learning_rate": 2.7594447652983862e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6710 + }, + { + "epoch": 1.955193482688391, + "flow_matching_loss": 0.016902447491884232, + "grad_norm": 0.09346222132444382, + "learning_rate": 2.7585459032550497e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6720 + }, + { + "epoch": 1.9581029967995345, + "flow_matching_loss": 0.014576209709048271, + "grad_norm": 0.08222950249910355, + "learning_rate": 2.7576455119597684e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6730 + }, + { + "epoch": 1.961012510910678, + "flow_matching_loss": 0.015688614826649426, + "grad_norm": 0.09301169216632843, + "learning_rate": 2.7567435925066012e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6740 + }, + { + "epoch": 1.9639220250218212, + "flow_matching_loss": 0.016456412523984908, + "grad_norm": 0.09235094487667084, + "learning_rate": 2.7558401459914665e-05, + "loss": 0.0165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6750 + }, + { + "epoch": 1.966831539132965, + "flow_matching_loss": 0.01513043474406004, + "grad_norm": 0.07681809365749359, + "learning_rate": 2.7549351735121362e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6760 + }, + { + "epoch": 1.9697410532441082, + "flow_matching_loss": 0.015971683897078037, + "grad_norm": 0.11988931149244308, + "learning_rate": 2.7540286761682377e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6770 + }, + { + "epoch": 1.9726505673552517, + "flow_matching_loss": 0.016900921519845725, + "grad_norm": 0.10359017550945282, + "learning_rate": 2.75312065506125e-05, + "loss": 0.0169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6780 + }, + { + "epoch": 1.9755600814663952, + "flow_matching_loss": 0.01682116324082017, + "grad_norm": 0.09477204829454422, + "learning_rate": 2.7522111112945052e-05, + "loss": 0.0168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6790 + }, + { + "epoch": 1.9784695955775384, + "flow_matching_loss": 0.016294421348720788, + "grad_norm": 0.09410135447978973, + "learning_rate": 2.751300045973184e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6800 + }, + { + "epoch": 1.9813791096886821, + "flow_matching_loss": 0.014512344636023045, + "grad_norm": 0.0634453296661377, + "learning_rate": 2.750387460204317e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6810 + }, + { + "epoch": 1.9842886237998254, + "flow_matching_loss": 0.014738522097468376, + "grad_norm": 0.07060014456510544, + "learning_rate": 2.7494733550967827e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6820 + }, + { + "epoch": 1.9871981379109689, + "flow_matching_loss": 0.01492313127964735, + "grad_norm": 0.07897765934467316, + "learning_rate": 2.748557731761304e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6830 + }, + { + "epoch": 1.9901076520221124, + "flow_matching_loss": 0.01640050280839205, + "grad_norm": 0.0850127637386322, + "learning_rate": 2.74764059131045e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6840 + }, + { + "epoch": 1.9930171661332556, + "flow_matching_loss": 0.015845713764429094, + "grad_norm": 0.08383449912071228, + "learning_rate": 2.7467219348586322e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6850 + }, + { + "epoch": 1.9959266802443993, + "flow_matching_loss": 0.015400558151304722, + "grad_norm": 0.11133845895528793, + "learning_rate": 2.7458017635221056e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6860 + }, + { + "epoch": 1.9988361943555426, + "flow_matching_loss": 0.016801132541149854, + "grad_norm": 0.07000070065259933, + "learning_rate": 2.7448800784189652e-05, + "loss": 0.0168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6870 + }, + { + "epoch": 2.001745708466686, + "flow_matching_loss": 0.016687260847538708, + "grad_norm": 0.0891856923699379, + "learning_rate": 2.7439568806691448e-05, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6880 + }, + { + "epoch": 2.0046552225778296, + "flow_matching_loss": 0.014006034284830094, + "grad_norm": 0.08002018928527832, + "learning_rate": 2.7430321713944167e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6890 + }, + { + "epoch": 2.007564736688973, + "flow_matching_loss": 0.014053787477314473, + "grad_norm": 0.0852690264582634, + "learning_rate": 2.7421059517183906e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6900 + }, + { + "epoch": 2.0104742508001165, + "flow_matching_loss": 0.01553097702562809, + "grad_norm": 0.08341170102357864, + "learning_rate": 2.7411782227665097e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6910 + }, + { + "epoch": 2.01338376491126, + "flow_matching_loss": 0.01718756426125765, + "grad_norm": 0.0738573744893074, + "learning_rate": 2.7402489856660528e-05, + "loss": 0.0172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6920 + }, + { + "epoch": 2.016293279022403, + "flow_matching_loss": 0.014074402302503586, + "grad_norm": 0.08655723929405212, + "learning_rate": 2.7393182415461307e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6930 + }, + { + "epoch": 2.019202793133547, + "flow_matching_loss": 0.016056020744144917, + "grad_norm": 0.09171625971794128, + "learning_rate": 2.7383859915376853e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6940 + }, + { + "epoch": 2.02211230724469, + "flow_matching_loss": 0.015660077054053544, + "grad_norm": 0.11456367373466492, + "learning_rate": 2.7374522367734875e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6950 + }, + { + "epoch": 2.0250218213558338, + "flow_matching_loss": 0.01604231307283044, + "grad_norm": 0.08371385931968689, + "learning_rate": 2.7365169783881378e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6960 + }, + { + "epoch": 2.027931335466977, + "flow_matching_loss": 0.01372207524254918, + "grad_norm": 0.08558692038059235, + "learning_rate": 2.735580217518064e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6970 + }, + { + "epoch": 2.0308408495781203, + "flow_matching_loss": 0.015791908744722603, + "grad_norm": 0.08813557773828506, + "learning_rate": 2.7346419553015175e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6980 + }, + { + "epoch": 2.033750363689264, + "flow_matching_loss": 0.016560510639101265, + "grad_norm": 0.10822415351867676, + "learning_rate": 2.733702192878577e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6990 + }, + { + "epoch": 2.0366598778004072, + "flow_matching_loss": 0.01458071144297719, + "grad_norm": 0.08485976606607437, + "learning_rate": 2.732760931391141e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7000 + }, + { + "epoch": 2.039569391911551, + "flow_matching_loss": 0.01554012754932046, + "grad_norm": 0.09552517533302307, + "learning_rate": 2.731818171982932e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7010 + }, + { + "epoch": 2.042478906022694, + "flow_matching_loss": 0.014248100202530622, + "grad_norm": 0.09029877185821533, + "learning_rate": 2.730873915799491e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7020 + }, + { + "epoch": 2.0453884201338375, + "flow_matching_loss": 0.015845361165702343, + "grad_norm": 0.09911608695983887, + "learning_rate": 2.7299281639881783e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7030 + }, + { + "epoch": 2.048297934244981, + "flow_matching_loss": 0.013787336554378271, + "grad_norm": 0.08773685991764069, + "learning_rate": 2.728980917698172e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7040 + }, + { + "epoch": 2.0512074483561245, + "flow_matching_loss": 0.014498281199485064, + "grad_norm": 0.0953591912984848, + "learning_rate": 2.7280321780804655e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7050 + }, + { + "epoch": 2.054116962467268, + "flow_matching_loss": 0.016197563707828523, + "grad_norm": 0.06624216586351395, + "learning_rate": 2.7270819462878666e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7060 + }, + { + "epoch": 2.0570264765784114, + "flow_matching_loss": 0.014696091134101152, + "grad_norm": 0.08875321596860886, + "learning_rate": 2.7261302234749965e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7070 + }, + { + "epoch": 2.0599359906895547, + "flow_matching_loss": 0.014034735411405564, + "grad_norm": 0.10042005032300949, + "learning_rate": 2.7251770107982888e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7080 + }, + { + "epoch": 2.0628455048006984, + "flow_matching_loss": 0.015017147082835436, + "grad_norm": 0.0831792801618576, + "learning_rate": 2.7242223094159856e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7090 + }, + { + "epoch": 2.0657550189118417, + "flow_matching_loss": 0.015427783224731683, + "grad_norm": 0.0804431214928627, + "learning_rate": 2.7232661204881408e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7100 + }, + { + "epoch": 2.0686645330229854, + "flow_matching_loss": 0.01380243692547083, + "grad_norm": 0.07207554578781128, + "learning_rate": 2.7223084451766123e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7110 + }, + { + "epoch": 2.0715740471341286, + "flow_matching_loss": 0.01547434451058507, + "grad_norm": 0.07813042402267456, + "learning_rate": 2.721349284645067e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7120 + }, + { + "epoch": 2.074483561245272, + "flow_matching_loss": 0.015500139817595482, + "grad_norm": 0.07897771894931793, + "learning_rate": 2.7203886400589752e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7130 + }, + { + "epoch": 2.0773930753564156, + "flow_matching_loss": 0.016100645996630192, + "grad_norm": 0.091176837682724, + "learning_rate": 2.7194265125856105e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7140 + }, + { + "epoch": 2.080302589467559, + "flow_matching_loss": 0.015365500748157502, + "grad_norm": 0.0965656116604805, + "learning_rate": 2.718462903394049e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7150 + }, + { + "epoch": 2.0832121035787026, + "flow_matching_loss": 0.01598641909658909, + "grad_norm": 0.09177626669406891, + "learning_rate": 2.7174978136551662e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7160 + }, + { + "epoch": 2.086121617689846, + "flow_matching_loss": 0.016294710151851178, + "grad_norm": 0.1031472235918045, + "learning_rate": 2.716531244541637e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7170 + }, + { + "epoch": 2.089031131800989, + "flow_matching_loss": 0.014871057402342559, + "grad_norm": 0.08283116668462753, + "learning_rate": 2.715563197227934e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7180 + }, + { + "epoch": 2.091940645912133, + "flow_matching_loss": 0.013976508751511574, + "grad_norm": 0.0825747475028038, + "learning_rate": 2.714593672890327e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7190 + }, + { + "epoch": 2.094850160023276, + "flow_matching_loss": 0.016043178178369998, + "grad_norm": 0.08355184644460678, + "learning_rate": 2.7136226727068783e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7200 + }, + { + "epoch": 2.0977596741344193, + "flow_matching_loss": 0.014136609341949224, + "grad_norm": 0.08419857919216156, + "learning_rate": 2.712650197857445e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7210 + }, + { + "epoch": 2.100669188245563, + "flow_matching_loss": 0.01576266102492809, + "grad_norm": 0.06763660907745361, + "learning_rate": 2.7116762495236763e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7220 + }, + { + "epoch": 2.1035787023567063, + "flow_matching_loss": 0.014387829415500163, + "grad_norm": 0.09047682583332062, + "learning_rate": 2.7107008288890106e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7230 + }, + { + "epoch": 2.10648821646785, + "flow_matching_loss": 0.01397844571620226, + "grad_norm": 0.06911168992519379, + "learning_rate": 2.709723937138676e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7240 + }, + { + "epoch": 2.1093977305789933, + "flow_matching_loss": 0.01510941917076707, + "grad_norm": 0.07068928331136703, + "learning_rate": 2.7087455754596883e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7250 + }, + { + "epoch": 2.112307244690137, + "flow_matching_loss": 0.014819025062024593, + "grad_norm": 0.06833461672067642, + "learning_rate": 2.7077657450408498e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7260 + }, + { + "epoch": 2.1152167588012802, + "flow_matching_loss": 0.014621153939515352, + "grad_norm": 0.06500514596700668, + "learning_rate": 2.7067844470727456e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7270 + }, + { + "epoch": 2.1181262729124235, + "flow_matching_loss": 0.017874925490468742, + "grad_norm": 0.08606477826833725, + "learning_rate": 2.7058016827477455e-05, + "loss": 0.0179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7280 + }, + { + "epoch": 2.121035787023567, + "flow_matching_loss": 0.01445076558738947, + "grad_norm": 0.10393314808607101, + "learning_rate": 2.7048174532600018e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7290 + }, + { + "epoch": 2.1239453011347105, + "flow_matching_loss": 0.014968509040772914, + "grad_norm": 0.0914638563990593, + "learning_rate": 2.7038317598054455e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7300 + }, + { + "epoch": 2.1268548152458537, + "flow_matching_loss": 0.01510093230754137, + "grad_norm": 0.08944635093212128, + "learning_rate": 2.702844603581787e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7310 + }, + { + "epoch": 2.1297643293569974, + "flow_matching_loss": 0.015475153271108866, + "grad_norm": 0.07951075583696365, + "learning_rate": 2.701855985788514e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7320 + }, + { + "epoch": 2.1326738434681407, + "flow_matching_loss": 0.014861342683434486, + "grad_norm": 0.09105315059423447, + "learning_rate": 2.7008659076268912e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7330 + }, + { + "epoch": 2.1355833575792844, + "flow_matching_loss": 0.016039444133639336, + "grad_norm": 0.08273877203464508, + "learning_rate": 2.6998743702999562e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7340 + }, + { + "epoch": 2.1384928716904277, + "flow_matching_loss": 0.015093087404966354, + "grad_norm": 0.1295512169599533, + "learning_rate": 2.6988813750125202e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7350 + }, + { + "epoch": 2.141402385801571, + "flow_matching_loss": 0.015736409556120635, + "grad_norm": 0.0872517004609108, + "learning_rate": 2.6978869229711668e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7360 + }, + { + "epoch": 2.1443118999127146, + "flow_matching_loss": 0.01604844620451331, + "grad_norm": 0.07430954277515411, + "learning_rate": 2.6968910153842484e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7370 + }, + { + "epoch": 2.147221414023858, + "flow_matching_loss": 0.014687509927898645, + "grad_norm": 0.08853308856487274, + "learning_rate": 2.6958936534618874e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7380 + }, + { + "epoch": 2.1501309281350016, + "flow_matching_loss": 0.014773161429911853, + "grad_norm": 0.08287356793880463, + "learning_rate": 2.694894838415971e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7390 + }, + { + "epoch": 2.153040442246145, + "flow_matching_loss": 0.014464683085680007, + "grad_norm": 0.07335192710161209, + "learning_rate": 2.6938945714601555e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7400 + }, + { + "epoch": 2.155949956357288, + "flow_matching_loss": 0.01399767016991973, + "grad_norm": 0.09033546596765518, + "learning_rate": 2.6928928538098587e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7410 + }, + { + "epoch": 2.158859470468432, + "flow_matching_loss": 0.016279870457947255, + "grad_norm": 0.09891233593225479, + "learning_rate": 2.6918896866822615e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7420 + }, + { + "epoch": 2.161768984579575, + "flow_matching_loss": 0.014593266043812036, + "grad_norm": 0.08760672062635422, + "learning_rate": 2.690885071296308e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7430 + }, + { + "epoch": 2.164678498690719, + "flow_matching_loss": 0.016445314139127733, + "grad_norm": 0.10370869189500809, + "learning_rate": 2.6898790088726993e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7440 + }, + { + "epoch": 2.167588012801862, + "flow_matching_loss": 0.015940096974372864, + "grad_norm": 0.11285644769668579, + "learning_rate": 2.6888715006338968e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7450 + }, + { + "epoch": 2.1704975269130053, + "flow_matching_loss": 0.015677055064588784, + "grad_norm": 0.09518235176801682, + "learning_rate": 2.6878625478041172e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7460 + }, + { + "epoch": 2.173407041024149, + "flow_matching_loss": 0.018114459700882435, + "grad_norm": 0.11226440221071243, + "learning_rate": 2.6868521516093342e-05, + "loss": 0.0181, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7470 + }, + { + "epoch": 2.1763165551352923, + "flow_matching_loss": 0.01352197015658021, + "grad_norm": 0.09003955125808716, + "learning_rate": 2.6858403132772745e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7480 + }, + { + "epoch": 2.179226069246436, + "flow_matching_loss": 0.015590622462332249, + "grad_norm": 0.09867899864912033, + "learning_rate": 2.6848270340374162e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7490 + }, + { + "epoch": 2.1821355833575793, + "flow_matching_loss": 0.013827286195009947, + "grad_norm": 0.07696697115898132, + "learning_rate": 2.68381231512099e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7500 + }, + { + "epoch": 2.1850450974687226, + "flow_matching_loss": 0.015621577575802802, + "grad_norm": 0.07376869767904282, + "learning_rate": 2.6827961577609745e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7510 + }, + { + "epoch": 2.1879546115798663, + "flow_matching_loss": 0.014935098495334387, + "grad_norm": 0.06218399107456207, + "learning_rate": 2.681778563192097e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7520 + }, + { + "epoch": 2.1908641256910095, + "flow_matching_loss": 0.014841482043266296, + "grad_norm": 0.08519314974546432, + "learning_rate": 2.6807595326508308e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7530 + }, + { + "epoch": 2.1937736398021532, + "flow_matching_loss": 0.016958463098853828, + "grad_norm": 0.1046217828989029, + "learning_rate": 2.679739067375394e-05, + "loss": 0.017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7540 + }, + { + "epoch": 2.1966831539132965, + "flow_matching_loss": 0.016764956526458263, + "grad_norm": 0.08223437517881393, + "learning_rate": 2.6787171686057487e-05, + "loss": 0.0168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7550 + }, + { + "epoch": 2.1995926680244398, + "flow_matching_loss": 0.015263914130628109, + "grad_norm": 0.08804715424776077, + "learning_rate": 2.677693837583597e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7560 + }, + { + "epoch": 2.2025021821355835, + "flow_matching_loss": 0.0149296959862113, + "grad_norm": 0.08889400213956833, + "learning_rate": 2.6766690755523836e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7570 + }, + { + "epoch": 2.2054116962467267, + "flow_matching_loss": 0.014331843890249728, + "grad_norm": 0.09210185706615448, + "learning_rate": 2.675642883757291e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7580 + }, + { + "epoch": 2.2083212103578704, + "flow_matching_loss": 0.015383264608681202, + "grad_norm": 0.10091467946767807, + "learning_rate": 2.674615263445239e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7590 + }, + { + "epoch": 2.2112307244690137, + "flow_matching_loss": 0.016516190208494663, + "grad_norm": 0.06578724831342697, + "learning_rate": 2.673586215864882e-05, + "loss": 0.0165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7600 + }, + { + "epoch": 2.214140238580157, + "flow_matching_loss": 0.014943863544613124, + "grad_norm": 0.09725803881883621, + "learning_rate": 2.6725557422666113e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7610 + }, + { + "epoch": 2.2170497526913007, + "flow_matching_loss": 0.014704861491918565, + "grad_norm": 0.08103442192077637, + "learning_rate": 2.6715238439025493e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7620 + }, + { + "epoch": 2.219959266802444, + "flow_matching_loss": 0.015780897252261637, + "grad_norm": 0.0989278182387352, + "learning_rate": 2.6704905220265493e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7630 + }, + { + "epoch": 2.2228687809135876, + "flow_matching_loss": 0.01592103336006403, + "grad_norm": 0.09688801318407059, + "learning_rate": 2.6694557778941947e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7640 + }, + { + "epoch": 2.225778295024731, + "flow_matching_loss": 0.015585344191640616, + "grad_norm": 0.07995352894067764, + "learning_rate": 2.6684196127627977e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7650 + }, + { + "epoch": 2.228687809135874, + "flow_matching_loss": 0.014154172409325838, + "grad_norm": 0.07513858377933502, + "learning_rate": 2.6673820278913963e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7660 + }, + { + "epoch": 2.231597323247018, + "flow_matching_loss": 0.014201079774647952, + "grad_norm": 0.09032312035560608, + "learning_rate": 2.6663430245407544e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7670 + }, + { + "epoch": 2.234506837358161, + "flow_matching_loss": 0.016297676600515842, + "grad_norm": 0.08806836605072021, + "learning_rate": 2.6653026039733585e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7680 + }, + { + "epoch": 2.237416351469305, + "flow_matching_loss": 0.016115516796708106, + "grad_norm": 0.09478949755430222, + "learning_rate": 2.6642607674534176e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7690 + }, + { + "epoch": 2.240325865580448, + "flow_matching_loss": 0.016303994785994293, + "grad_norm": 0.0809672400355339, + "learning_rate": 2.663217516246862e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7700 + }, + { + "epoch": 2.2432353796915914, + "flow_matching_loss": 0.015085996501147747, + "grad_norm": 0.09144578129053116, + "learning_rate": 2.66217285162134e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7710 + }, + { + "epoch": 2.246144893802735, + "flow_matching_loss": 0.01600094707682729, + "grad_norm": 0.10902480781078339, + "learning_rate": 2.6611267748462175e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7720 + }, + { + "epoch": 2.2490544079138783, + "flow_matching_loss": 0.014814405608922243, + "grad_norm": 0.09534116089344025, + "learning_rate": 2.6600792871925764e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7730 + }, + { + "epoch": 2.2519639220250216, + "flow_matching_loss": 0.014572432916611432, + "grad_norm": 0.09606514871120453, + "learning_rate": 2.659030389933213e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7740 + }, + { + "epoch": 2.2548734361361653, + "flow_matching_loss": 0.016095179785043, + "grad_norm": 0.08236677944660187, + "learning_rate": 2.6579800843426358e-05, + "loss": 0.0161, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7750 + }, + { + "epoch": 2.2577829502473086, + "flow_matching_loss": 0.015430446807295084, + "grad_norm": 0.08431357890367508, + "learning_rate": 2.656928371697067e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7760 + }, + { + "epoch": 2.2606924643584523, + "flow_matching_loss": 0.015206343960016966, + "grad_norm": 0.08057689666748047, + "learning_rate": 2.6558752532744346e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7770 + }, + { + "epoch": 2.2636019784695955, + "flow_matching_loss": 0.016308887116611003, + "grad_norm": 0.09380348771810532, + "learning_rate": 2.654820730354377e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7780 + }, + { + "epoch": 2.2665114925807393, + "flow_matching_loss": 0.016628941241651773, + "grad_norm": 0.08369313180446625, + "learning_rate": 2.6537648042182406e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7790 + }, + { + "epoch": 2.2694210066918825, + "flow_matching_loss": 0.01440786998718977, + "grad_norm": 0.09615826606750488, + "learning_rate": 2.6527074761490735e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7800 + }, + { + "epoch": 2.2723305208030258, + "flow_matching_loss": 0.014325074385851622, + "grad_norm": 0.07481158524751663, + "learning_rate": 2.6516487474316303e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7810 + }, + { + "epoch": 2.2752400349141695, + "flow_matching_loss": 0.013842524588108062, + "grad_norm": 0.07807376980781555, + "learning_rate": 2.650588619352365e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7820 + }, + { + "epoch": 2.2781495490253127, + "flow_matching_loss": 0.014688007719814777, + "grad_norm": 0.08896870166063309, + "learning_rate": 2.649527093199434e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7830 + }, + { + "epoch": 2.281059063136456, + "flow_matching_loss": 0.013472345285117626, + "grad_norm": 0.08293180167675018, + "learning_rate": 2.6484641702626914e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7840 + }, + { + "epoch": 2.2839685772475997, + "flow_matching_loss": 0.01448190351948142, + "grad_norm": 0.08825221657752991, + "learning_rate": 2.647399851833689e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7850 + }, + { + "epoch": 2.286878091358743, + "flow_matching_loss": 0.014990883599966764, + "grad_norm": 0.07351382076740265, + "learning_rate": 2.6463341392056735e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7860 + }, + { + "epoch": 2.2897876054698867, + "flow_matching_loss": 0.013465344719588757, + "grad_norm": 0.0657515898346901, + "learning_rate": 2.6452670336735867e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7870 + }, + { + "epoch": 2.29269711958103, + "flow_matching_loss": 0.015872781723737718, + "grad_norm": 0.0919480100274086, + "learning_rate": 2.644198536534063e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7880 + }, + { + "epoch": 2.295606633692173, + "flow_matching_loss": 0.01592162987217307, + "grad_norm": 0.11093220114707947, + "learning_rate": 2.6431286490854255e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7890 + }, + { + "epoch": 2.298516147803317, + "flow_matching_loss": 0.01680972445756197, + "grad_norm": 0.10801055282354355, + "learning_rate": 2.6420573726276903e-05, + "loss": 0.0168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7900 + }, + { + "epoch": 2.30142566191446, + "flow_matching_loss": 0.015878868103027344, + "grad_norm": 0.09607215225696564, + "learning_rate": 2.6409847084625575e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7910 + }, + { + "epoch": 2.304335176025604, + "flow_matching_loss": 0.014640671852976083, + "grad_norm": 0.08455350995063782, + "learning_rate": 2.6399106578934167e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7920 + }, + { + "epoch": 2.307244690136747, + "flow_matching_loss": 0.016739309299737215, + "grad_norm": 0.11892305314540863, + "learning_rate": 2.63883522222534e-05, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7930 + }, + { + "epoch": 2.3101542042478904, + "flow_matching_loss": 0.01469262707978487, + "grad_norm": 0.0727696567773819, + "learning_rate": 2.6377584027650827e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7940 + }, + { + "epoch": 2.313063718359034, + "flow_matching_loss": 0.015909372083842754, + "grad_norm": 0.09612101316452026, + "learning_rate": 2.636680200821083e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7950 + }, + { + "epoch": 2.3159732324701774, + "flow_matching_loss": 0.0152977978810668, + "grad_norm": 0.08419929444789886, + "learning_rate": 2.635600617703457e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7960 + }, + { + "epoch": 2.318882746581321, + "flow_matching_loss": 0.015046276245266199, + "grad_norm": 0.07879979908466339, + "learning_rate": 2.6345196547240007e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7970 + }, + { + "epoch": 2.3217922606924644, + "flow_matching_loss": 0.015192751865833997, + "grad_norm": 0.08083745092153549, + "learning_rate": 2.6334373131961854e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7980 + }, + { + "epoch": 2.3247017748036076, + "flow_matching_loss": 0.014290974661707877, + "grad_norm": 0.0841483324766159, + "learning_rate": 2.6323535944351583e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7990 + }, + { + "epoch": 2.3276112889147513, + "flow_matching_loss": 0.014214409328997136, + "grad_norm": 0.073344387114048, + "learning_rate": 2.63126849975774e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8000 + }, + { + "epoch": 2.3305208030258946, + "flow_matching_loss": 0.015752159990370274, + "grad_norm": 0.0951995998620987, + "learning_rate": 2.6301820304824234e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8010 + }, + { + "epoch": 2.3334303171370383, + "flow_matching_loss": 0.013621991500258446, + "grad_norm": 0.09052250534296036, + "learning_rate": 2.6290941879293703e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8020 + }, + { + "epoch": 2.3363398312481816, + "flow_matching_loss": 0.015892947558313608, + "grad_norm": 0.07912550121545792, + "learning_rate": 2.628004973420412e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8030 + }, + { + "epoch": 2.339249345359325, + "flow_matching_loss": 0.015727121382951736, + "grad_norm": 0.08996255695819855, + "learning_rate": 2.626914388279047e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8040 + }, + { + "epoch": 2.3421588594704685, + "flow_matching_loss": 0.013611921668052673, + "grad_norm": 0.10643363744020462, + "learning_rate": 2.6258224338304394e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8050 + }, + { + "epoch": 2.345068373581612, + "flow_matching_loss": 0.014662514347583056, + "grad_norm": 0.08409671485424042, + "learning_rate": 2.6247291114014166e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8060 + }, + { + "epoch": 2.347977887692755, + "flow_matching_loss": 0.01479412978515029, + "grad_norm": 0.07956776767969131, + "learning_rate": 2.6236344223204683e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8070 + }, + { + "epoch": 2.3508874018038988, + "flow_matching_loss": 0.015393637865781785, + "grad_norm": 0.16266778111457825, + "learning_rate": 2.6225383679177453e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8080 + }, + { + "epoch": 2.353796915915042, + "flow_matching_loss": 0.014476249366998673, + "grad_norm": 0.09052246063947678, + "learning_rate": 2.6214409495250568e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8090 + }, + { + "epoch": 2.3567064300261857, + "flow_matching_loss": 0.017747414764016867, + "grad_norm": 0.07860356569290161, + "learning_rate": 2.620342168475869e-05, + "loss": 0.0177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8100 + }, + { + "epoch": 2.359615944137329, + "flow_matching_loss": 0.01629470931366086, + "grad_norm": 0.08898527920246124, + "learning_rate": 2.619242026105306e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8110 + }, + { + "epoch": 2.3625254582484727, + "flow_matching_loss": 0.01437996355816722, + "grad_norm": 0.06860119849443436, + "learning_rate": 2.618140523750143e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8120 + }, + { + "epoch": 2.365434972359616, + "flow_matching_loss": 0.015572782140225172, + "grad_norm": 0.08905408531427383, + "learning_rate": 2.6170376627488103e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8130 + }, + { + "epoch": 2.3683444864707592, + "flow_matching_loss": 0.015980213601142167, + "grad_norm": 0.11391662806272507, + "learning_rate": 2.6159334444413873e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8140 + }, + { + "epoch": 2.371254000581903, + "flow_matching_loss": 0.017882842849940062, + "grad_norm": 0.08316270262002945, + "learning_rate": 2.614827870169604e-05, + "loss": 0.0179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8150 + }, + { + "epoch": 2.374163514693046, + "flow_matching_loss": 0.013557769171893597, + "grad_norm": 0.07775083929300308, + "learning_rate": 2.6137209412768367e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8160 + }, + { + "epoch": 2.3770730288041895, + "flow_matching_loss": 0.01619083471596241, + "grad_norm": 0.08014574646949768, + "learning_rate": 2.612612659108109e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8170 + }, + { + "epoch": 2.379982542915333, + "flow_matching_loss": 0.014824150502681733, + "grad_norm": 0.09454503655433655, + "learning_rate": 2.611503025010088e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8180 + }, + { + "epoch": 2.3828920570264764, + "flow_matching_loss": 0.014401414152234792, + "grad_norm": 0.07687917351722717, + "learning_rate": 2.6103920403310832e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8190 + }, + { + "epoch": 2.38580157113762, + "flow_matching_loss": 0.01455121422186494, + "grad_norm": 0.08798868954181671, + "learning_rate": 2.6092797064210465e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8200 + }, + { + "epoch": 2.3887110852487634, + "flow_matching_loss": 0.015816065296530724, + "grad_norm": 0.09440571069717407, + "learning_rate": 2.6081660246315685e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8210 + }, + { + "epoch": 2.391620599359907, + "flow_matching_loss": 0.013730627857148648, + "grad_norm": 0.08165957033634186, + "learning_rate": 2.6070509963158774e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8220 + }, + { + "epoch": 2.3945301134710504, + "flow_matching_loss": 0.014089516922831535, + "grad_norm": 0.10680492967367172, + "learning_rate": 2.6059346228288375e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8230 + }, + { + "epoch": 2.3974396275821936, + "flow_matching_loss": 0.01524124788120389, + "grad_norm": 0.10596584528684616, + "learning_rate": 2.604816905526948e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8240 + }, + { + "epoch": 2.4003491416933374, + "flow_matching_loss": 0.015080444421619177, + "grad_norm": 0.08267712593078613, + "learning_rate": 2.603697845768341e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8250 + }, + { + "epoch": 2.4032586558044806, + "flow_matching_loss": 0.015079002361744642, + "grad_norm": 0.07737356424331665, + "learning_rate": 2.6025774449127792e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8260 + }, + { + "epoch": 2.406168169915624, + "flow_matching_loss": 0.014232567977160215, + "grad_norm": 0.0765300989151001, + "learning_rate": 2.6014557043216562e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8270 + }, + { + "epoch": 2.4090776840267676, + "flow_matching_loss": 0.01404778277501464, + "grad_norm": 0.10327775031328201, + "learning_rate": 2.6003326253579913e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8280 + }, + { + "epoch": 2.411987198137911, + "flow_matching_loss": 0.014587139990180732, + "grad_norm": 0.07276114076375961, + "learning_rate": 2.599208209386432e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8290 + }, + { + "epoch": 2.4148967122490546, + "flow_matching_loss": 0.01571178436279297, + "grad_norm": 0.12033865600824356, + "learning_rate": 2.5980824577732502e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8300 + }, + { + "epoch": 2.417806226360198, + "flow_matching_loss": 0.014257180877029896, + "grad_norm": 0.07614407688379288, + "learning_rate": 2.5969553718863395e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8310 + }, + { + "epoch": 2.4207157404713415, + "flow_matching_loss": 0.015670361742377282, + "grad_norm": 0.08799499273300171, + "learning_rate": 2.595826953095216e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8320 + }, + { + "epoch": 2.423625254582485, + "flow_matching_loss": 0.014637295994907618, + "grad_norm": 0.07357558608055115, + "learning_rate": 2.594697202771014e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8330 + }, + { + "epoch": 2.426534768693628, + "flow_matching_loss": 0.015156016871333122, + "grad_norm": 0.06836707144975662, + "learning_rate": 2.5935661222864888e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8340 + }, + { + "epoch": 2.4294442828047718, + "flow_matching_loss": 0.016472499072551727, + "grad_norm": 0.07921077311038971, + "learning_rate": 2.592433713016008e-05, + "loss": 0.0165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8350 + }, + { + "epoch": 2.432353796915915, + "flow_matching_loss": 0.013238575868308544, + "grad_norm": 0.07178080826997757, + "learning_rate": 2.5912999763355553e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8360 + }, + { + "epoch": 2.4352633110270583, + "flow_matching_loss": 0.013974263332784176, + "grad_norm": 0.08763539046049118, + "learning_rate": 2.590164913622729e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8370 + }, + { + "epoch": 2.438172825138202, + "flow_matching_loss": 0.015604240540415048, + "grad_norm": 0.10790900141000748, + "learning_rate": 2.589028526256737e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8380 + }, + { + "epoch": 2.4410823392493453, + "flow_matching_loss": 0.01542277168482542, + "grad_norm": 0.09170646220445633, + "learning_rate": 2.587890815618397e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8390 + }, + { + "epoch": 2.443991853360489, + "flow_matching_loss": 0.01595898037776351, + "grad_norm": 0.09875811636447906, + "learning_rate": 2.5867517830901343e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8400 + }, + { + "epoch": 2.4469013674716322, + "flow_matching_loss": 0.014763825759291648, + "grad_norm": 0.08344326913356781, + "learning_rate": 2.5856114300559812e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8410 + }, + { + "epoch": 2.4498108815827755, + "flow_matching_loss": 0.015589122287929058, + "grad_norm": 0.08689700812101364, + "learning_rate": 2.584469757901574e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8420 + }, + { + "epoch": 2.452720395693919, + "flow_matching_loss": 0.013565236702561378, + "grad_norm": 0.06180119886994362, + "learning_rate": 2.5833267680141526e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8430 + }, + { + "epoch": 2.4556299098050625, + "flow_matching_loss": 0.014716735389083625, + "grad_norm": 0.06673333048820496, + "learning_rate": 2.5821824617825564e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8440 + }, + { + "epoch": 2.458539423916206, + "flow_matching_loss": 0.01458737561479211, + "grad_norm": 0.080013208091259, + "learning_rate": 2.5810368405972265e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8450 + }, + { + "epoch": 2.4614489380273494, + "flow_matching_loss": 0.014875032380223275, + "grad_norm": 0.07870429009199142, + "learning_rate": 2.5798899058501996e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8460 + }, + { + "epoch": 2.4643584521384927, + "flow_matching_loss": 0.013179139886051416, + "grad_norm": 0.07162608206272125, + "learning_rate": 2.578741658935111e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8470 + }, + { + "epoch": 2.4672679662496364, + "flow_matching_loss": 0.013810235634446144, + "grad_norm": 0.08705882728099823, + "learning_rate": 2.5775921012471875e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8480 + }, + { + "epoch": 2.4701774803607797, + "flow_matching_loss": 0.013810858782380819, + "grad_norm": 0.08789811283349991, + "learning_rate": 2.5764412341832505e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8490 + }, + { + "epoch": 2.4730869944719234, + "flow_matching_loss": 0.015221449267119169, + "grad_norm": 0.08810198307037354, + "learning_rate": 2.5752890591417125e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8500 + }, + { + "epoch": 2.4759965085830666, + "flow_matching_loss": 0.01652440829202533, + "grad_norm": 0.07161926478147507, + "learning_rate": 2.5741355775225744e-05, + "loss": 0.0165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8510 + }, + { + "epoch": 2.47890602269421, + "flow_matching_loss": 0.015106333885341883, + "grad_norm": 0.08192957192659378, + "learning_rate": 2.5729807907274254e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8520 + }, + { + "epoch": 2.4818155368053536, + "flow_matching_loss": 0.015226961392909289, + "grad_norm": 0.11699581891298294, + "learning_rate": 2.5718247001594403e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8530 + }, + { + "epoch": 2.484725050916497, + "flow_matching_loss": 0.013792990520596505, + "grad_norm": 0.09758847951889038, + "learning_rate": 2.5706673072233776e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8540 + }, + { + "epoch": 2.4876345650276406, + "flow_matching_loss": 0.014674542658030986, + "grad_norm": 0.0841912180185318, + "learning_rate": 2.5695086133255795e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8550 + }, + { + "epoch": 2.490544079138784, + "flow_matching_loss": 0.015669130813330413, + "grad_norm": 0.0948745608329773, + "learning_rate": 2.5683486198739687e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8560 + }, + { + "epoch": 2.493453593249927, + "flow_matching_loss": 0.015192539989948272, + "grad_norm": 0.07057670503854752, + "learning_rate": 2.5671873282780452e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8570 + }, + { + "epoch": 2.496363107361071, + "flow_matching_loss": 0.015471458621323108, + "grad_norm": 0.08338472992181778, + "learning_rate": 2.5660247399488887e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8580 + }, + { + "epoch": 2.499272621472214, + "flow_matching_loss": 0.01494266428053379, + "grad_norm": 0.08621294051408768, + "learning_rate": 2.564860856299154e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8590 + }, + { + "epoch": 2.5021821355833573, + "flow_matching_loss": 0.014327356778085231, + "grad_norm": 0.0887235775589943, + "learning_rate": 2.5636956787430693e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8600 + }, + { + "epoch": 2.505091649694501, + "flow_matching_loss": 0.015590361785143614, + "grad_norm": 0.06827245652675629, + "learning_rate": 2.5625292086964346e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8610 + }, + { + "epoch": 2.5080011638056443, + "flow_matching_loss": 0.015681112371385097, + "grad_norm": 0.07660006731748581, + "learning_rate": 2.5613614475766212e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8620 + }, + { + "epoch": 2.510910677916788, + "flow_matching_loss": 0.015190454758703708, + "grad_norm": 0.09373155981302261, + "learning_rate": 2.5601923968025694e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8630 + }, + { + "epoch": 2.5138201920279313, + "flow_matching_loss": 0.013883538823574782, + "grad_norm": 0.080729179084301, + "learning_rate": 2.5590220577947862e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8640 + }, + { + "epoch": 2.516729706139075, + "flow_matching_loss": 0.01511918418109417, + "grad_norm": 0.09426675736904144, + "learning_rate": 2.5578504319753427e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8650 + }, + { + "epoch": 2.5196392202502182, + "flow_matching_loss": 0.015135084092617036, + "grad_norm": 0.0800924226641655, + "learning_rate": 2.556677520767876e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8660 + }, + { + "epoch": 2.5225487343613615, + "flow_matching_loss": 0.014140944555401802, + "grad_norm": 0.089115671813488, + "learning_rate": 2.5555033255975835e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8670 + }, + { + "epoch": 2.525458248472505, + "flow_matching_loss": 0.01410072548314929, + "grad_norm": 0.0645175352692604, + "learning_rate": 2.5543278478912228e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8680 + }, + { + "epoch": 2.5283677625836485, + "flow_matching_loss": 0.015604391507804394, + "grad_norm": 0.07841598987579346, + "learning_rate": 2.55315108907711e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8690 + }, + { + "epoch": 2.5312772766947917, + "flow_matching_loss": 0.016033988911658525, + "grad_norm": 0.08887864649295807, + "learning_rate": 2.551973050585119e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8700 + }, + { + "epoch": 2.5341867908059355, + "flow_matching_loss": 0.014126046467572451, + "grad_norm": 0.08311211317777634, + "learning_rate": 2.5507937338466768e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8710 + }, + { + "epoch": 2.5370963049170787, + "flow_matching_loss": 0.013353736605495214, + "grad_norm": 0.07672525197267532, + "learning_rate": 2.549613140294764e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8720 + }, + { + "epoch": 2.5400058190282224, + "flow_matching_loss": 0.014103832095861435, + "grad_norm": 0.09786020964384079, + "learning_rate": 2.548431271363914e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8730 + }, + { + "epoch": 2.5429153331393657, + "flow_matching_loss": 0.01573379701003432, + "grad_norm": 0.07810751348733902, + "learning_rate": 2.5472481284902092e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8740 + }, + { + "epoch": 2.5458248472505094, + "flow_matching_loss": 0.015071121603250503, + "grad_norm": 0.09533120691776276, + "learning_rate": 2.5460637131112788e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8750 + }, + { + "epoch": 2.5487343613616527, + "flow_matching_loss": 0.01508156880736351, + "grad_norm": 0.08026155084371567, + "learning_rate": 2.5448780266662995e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8760 + }, + { + "epoch": 2.551643875472796, + "flow_matching_loss": 0.01627356242388487, + "grad_norm": 0.07603893429040909, + "learning_rate": 2.5436910705959926e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8770 + }, + { + "epoch": 2.5545533895839396, + "flow_matching_loss": 0.015440229885280132, + "grad_norm": 0.08555714040994644, + "learning_rate": 2.542502846342621e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8780 + }, + { + "epoch": 2.557462903695083, + "flow_matching_loss": 0.015677285846322775, + "grad_norm": 0.09372563660144806, + "learning_rate": 2.5413133553499892e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8790 + }, + { + "epoch": 2.560372417806226, + "flow_matching_loss": 0.016188987717032432, + "grad_norm": 0.06541736423969269, + "learning_rate": 2.540122599063442e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8800 + }, + { + "epoch": 2.56328193191737, + "flow_matching_loss": 0.014195558149367571, + "grad_norm": 0.07932368665933609, + "learning_rate": 2.538930578929859e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8810 + }, + { + "epoch": 2.566191446028513, + "flow_matching_loss": 0.01722688591107726, + "grad_norm": 0.075489342212677, + "learning_rate": 2.537737296397658e-05, + "loss": 0.0172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8820 + }, + { + "epoch": 2.569100960139657, + "flow_matching_loss": 0.017213862389326096, + "grad_norm": 0.10038011521100998, + "learning_rate": 2.5365427529167903e-05, + "loss": 0.0172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8830 + }, + { + "epoch": 2.5720104742508, + "flow_matching_loss": 0.013752058241516351, + "grad_norm": 0.07622182369232178, + "learning_rate": 2.5353469499387375e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8840 + }, + { + "epoch": 2.574919988361944, + "flow_matching_loss": 0.015986564569175243, + "grad_norm": 0.08039705455303192, + "learning_rate": 2.534149888916514e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8850 + }, + { + "epoch": 2.577829502473087, + "flow_matching_loss": 0.013063894957304001, + "grad_norm": 0.0773366317152977, + "learning_rate": 2.5329515713046616e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8860 + }, + { + "epoch": 2.5807390165842303, + "flow_matching_loss": 0.014240353927016258, + "grad_norm": 0.078423410654068, + "learning_rate": 2.531751998559249e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8870 + }, + { + "epoch": 2.583648530695374, + "flow_matching_loss": 0.014171305764466523, + "grad_norm": 0.45336470007896423, + "learning_rate": 2.5305511721378703e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8880 + }, + { + "epoch": 2.5865580448065173, + "flow_matching_loss": 0.014396458491683006, + "grad_norm": 0.07042349129915237, + "learning_rate": 2.529349093499642e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8890 + }, + { + "epoch": 2.5894675589176606, + "flow_matching_loss": 0.014758321270346642, + "grad_norm": 0.08278340846300125, + "learning_rate": 2.528145764105205e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8900 + }, + { + "epoch": 2.5923770730288043, + "flow_matching_loss": 0.015135197434574366, + "grad_norm": 0.1263064742088318, + "learning_rate": 2.5269411854167166e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8910 + }, + { + "epoch": 2.5952865871399475, + "flow_matching_loss": 0.017482858244329692, + "grad_norm": 0.09446372091770172, + "learning_rate": 2.5257353588978534e-05, + "loss": 0.0175, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8920 + }, + { + "epoch": 2.598196101251091, + "flow_matching_loss": 0.015310948900878429, + "grad_norm": 0.08655768632888794, + "learning_rate": 2.5245282860138083e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8930 + }, + { + "epoch": 2.6011056153622345, + "flow_matching_loss": 0.014121008664369583, + "grad_norm": 0.08149401098489761, + "learning_rate": 2.5233199682312894e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8940 + }, + { + "epoch": 2.604015129473378, + "flow_matching_loss": 0.013860475923866033, + "grad_norm": 0.06332944333553314, + "learning_rate": 2.5221104070185163e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8950 + }, + { + "epoch": 2.6069246435845215, + "flow_matching_loss": 0.015083832107484341, + "grad_norm": 0.06994985789060593, + "learning_rate": 2.5208996038452204e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8960 + }, + { + "epoch": 2.6098341576956647, + "flow_matching_loss": 0.013820645306259394, + "grad_norm": 0.09050416946411133, + "learning_rate": 2.5196875601826403e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8970 + }, + { + "epoch": 2.6127436718068084, + "flow_matching_loss": 0.01480758311226964, + "grad_norm": 0.06598400324583054, + "learning_rate": 2.518474277503525e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8980 + }, + { + "epoch": 2.6156531859179517, + "flow_matching_loss": 0.01495195021852851, + "grad_norm": 0.07590681314468384, + "learning_rate": 2.5172597572821266e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8990 + }, + { + "epoch": 2.618562700029095, + "flow_matching_loss": 0.013923635985702277, + "grad_norm": 0.06900059431791306, + "learning_rate": 2.5160440009942015e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9000 + }, + { + "epoch": 2.6214722141402387, + "flow_matching_loss": 0.01410939246416092, + "grad_norm": 0.09965262562036514, + "learning_rate": 2.514827010117009e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9010 + }, + { + "epoch": 2.624381728251382, + "flow_matching_loss": 0.014234842732548714, + "grad_norm": 0.07886447757482529, + "learning_rate": 2.5136087861293068e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9020 + }, + { + "epoch": 2.627291242362525, + "flow_matching_loss": 0.014508087188005447, + "grad_norm": 0.0791783332824707, + "learning_rate": 2.5123893305113525e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9030 + }, + { + "epoch": 2.630200756473669, + "flow_matching_loss": 0.014552861824631691, + "grad_norm": 0.06941908597946167, + "learning_rate": 2.5111686447448997e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9040 + }, + { + "epoch": 2.633110270584812, + "flow_matching_loss": 0.015062520559877158, + "grad_norm": 0.08891478180885315, + "learning_rate": 2.5099467303131966e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9050 + }, + { + "epoch": 2.636019784695956, + "flow_matching_loss": 0.01400592690333724, + "grad_norm": 0.10666070878505707, + "learning_rate": 2.508723588700984e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9060 + }, + { + "epoch": 2.638929298807099, + "flow_matching_loss": 0.01524353725835681, + "grad_norm": 0.08375947177410126, + "learning_rate": 2.5074992213944952e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9070 + }, + { + "epoch": 2.641838812918243, + "flow_matching_loss": 0.014101412612944841, + "grad_norm": 0.08040738850831985, + "learning_rate": 2.506273629881451e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9080 + }, + { + "epoch": 2.644748327029386, + "flow_matching_loss": 0.01634820122271776, + "grad_norm": 0.10277904570102692, + "learning_rate": 2.5050468156510617e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9090 + }, + { + "epoch": 2.6476578411405294, + "flow_matching_loss": 0.013861185032874345, + "grad_norm": 0.08796647936105728, + "learning_rate": 2.5038187801940215e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9100 + }, + { + "epoch": 2.650567355251673, + "flow_matching_loss": 0.015140210930258035, + "grad_norm": 0.07140561193227768, + "learning_rate": 2.5025895250025088e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9110 + }, + { + "epoch": 2.6534768693628163, + "flow_matching_loss": 0.015777440555393697, + "grad_norm": 0.07920173555612564, + "learning_rate": 2.501359051570186e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9120 + }, + { + "epoch": 2.6563863834739596, + "flow_matching_loss": 0.015452532563358545, + "grad_norm": 0.08947692811489105, + "learning_rate": 2.500127361392193e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9130 + }, + { + "epoch": 2.6592958975851033, + "flow_matching_loss": 0.014702870696783065, + "grad_norm": 0.0782092809677124, + "learning_rate": 2.49889445596515e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9140 + }, + { + "epoch": 2.6622054116962466, + "flow_matching_loss": 0.014917513821274042, + "grad_norm": 0.07520106434822083, + "learning_rate": 2.497660336787154e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9150 + }, + { + "epoch": 2.6651149258073903, + "flow_matching_loss": 0.014381734095513821, + "grad_norm": 0.12027426064014435, + "learning_rate": 2.496425005357775e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9160 + }, + { + "epoch": 2.6680244399185336, + "flow_matching_loss": 0.016249839588999747, + "grad_norm": 0.09485580772161484, + "learning_rate": 2.4951884631780578e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9170 + }, + { + "epoch": 2.6709339540296773, + "flow_matching_loss": 0.01433224705979228, + "grad_norm": 0.06823644787073135, + "learning_rate": 2.493950711750518e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9180 + }, + { + "epoch": 2.6738434681408205, + "flow_matching_loss": 0.016413773223757745, + "grad_norm": 0.0904158279299736, + "learning_rate": 2.49271175257914e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9190 + }, + { + "epoch": 2.676752982251964, + "flow_matching_loss": 0.01591698592528701, + "grad_norm": 0.10777058452367783, + "learning_rate": 2.4914715871693757e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9200 + }, + { + "epoch": 2.6796624963631075, + "flow_matching_loss": 0.01602168856188655, + "grad_norm": 0.07185259461402893, + "learning_rate": 2.4902302170281438e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9210 + }, + { + "epoch": 2.6825720104742508, + "flow_matching_loss": 0.015309410728514194, + "grad_norm": 0.08148767054080963, + "learning_rate": 2.4889876436638252e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9220 + }, + { + "epoch": 2.685481524585394, + "flow_matching_loss": 0.01475420631468296, + "grad_norm": 0.08113230764865875, + "learning_rate": 2.4877438685862643e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9230 + }, + { + "epoch": 2.6883910386965377, + "flow_matching_loss": 0.013987391162663699, + "grad_norm": 0.10750900208950043, + "learning_rate": 2.486498893306765e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9240 + }, + { + "epoch": 2.691300552807681, + "flow_matching_loss": 0.014568882435560227, + "grad_norm": 0.10093829780817032, + "learning_rate": 2.485252719338089e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9250 + }, + { + "epoch": 2.6942100669188247, + "flow_matching_loss": 0.015334756392985583, + "grad_norm": 0.09519622474908829, + "learning_rate": 2.4840053481944563e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9260 + }, + { + "epoch": 2.697119581029968, + "flow_matching_loss": 0.014644651673734188, + "grad_norm": 0.08228091150522232, + "learning_rate": 2.4827567813915398e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9270 + }, + { + "epoch": 2.7000290951411117, + "flow_matching_loss": 0.015604792349040508, + "grad_norm": 0.09405926614999771, + "learning_rate": 2.481507020446466e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9280 + }, + { + "epoch": 2.702938609252255, + "flow_matching_loss": 0.015297127608209848, + "grad_norm": 0.0843990296125412, + "learning_rate": 2.4802560668778122e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9290 + }, + { + "epoch": 2.705848123363398, + "flow_matching_loss": 0.014320077281445265, + "grad_norm": 0.09214337170124054, + "learning_rate": 2.479003922205605e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9300 + }, + { + "epoch": 2.708757637474542, + "flow_matching_loss": 0.015202712081372739, + "grad_norm": 0.08523458242416382, + "learning_rate": 2.4777505879513183e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9310 + }, + { + "epoch": 2.711667151585685, + "flow_matching_loss": 0.014136172737926245, + "grad_norm": 0.08671223372220993, + "learning_rate": 2.4764960656378714e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9320 + }, + { + "epoch": 2.7145766656968284, + "flow_matching_loss": 0.012762954086065292, + "grad_norm": 0.09493093937635422, + "learning_rate": 2.4752403567896274e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9330 + }, + { + "epoch": 2.717486179807972, + "flow_matching_loss": 0.013288697693496943, + "grad_norm": 0.08485761284828186, + "learning_rate": 2.47398346293239e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9340 + }, + { + "epoch": 2.7203956939191154, + "flow_matching_loss": 0.015046989638358354, + "grad_norm": 0.09492143988609314, + "learning_rate": 2.4727253855934055e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9350 + }, + { + "epoch": 2.723305208030259, + "flow_matching_loss": 0.014578027743846178, + "grad_norm": 0.07102907449007034, + "learning_rate": 2.4714661263013548e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9360 + }, + { + "epoch": 2.7262147221414024, + "flow_matching_loss": 0.014701435714960099, + "grad_norm": 0.08296357095241547, + "learning_rate": 2.4702056865863575e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9370 + }, + { + "epoch": 2.729124236252546, + "flow_matching_loss": 0.014682969078421593, + "grad_norm": 0.09403949975967407, + "learning_rate": 2.4689440679799672e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9380 + }, + { + "epoch": 2.7320337503636893, + "flow_matching_loss": 0.01388571560382843, + "grad_norm": 0.08650879561901093, + "learning_rate": 2.4676812720151678e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9390 + }, + { + "epoch": 2.7349432644748326, + "flow_matching_loss": 0.014299247227609158, + "grad_norm": 0.08959274739027023, + "learning_rate": 2.466417300226377e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9400 + }, + { + "epoch": 2.7378527785859763, + "flow_matching_loss": 0.014586058165878057, + "grad_norm": 0.07473376393318176, + "learning_rate": 2.4651521541494394e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9410 + }, + { + "epoch": 2.7407622926971196, + "flow_matching_loss": 0.015568195655941964, + "grad_norm": 0.08922772109508514, + "learning_rate": 2.4638858353216267e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9420 + }, + { + "epoch": 2.743671806808263, + "flow_matching_loss": 0.015417752973735332, + "grad_norm": 0.09589357674121857, + "learning_rate": 2.4626183452816355e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9430 + }, + { + "epoch": 2.7465813209194065, + "flow_matching_loss": 0.013476703222841025, + "grad_norm": 0.07322211563587189, + "learning_rate": 2.4613496855695855e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9440 + }, + { + "epoch": 2.74949083503055, + "flow_matching_loss": 0.013088272698223592, + "grad_norm": 0.09172949939966202, + "learning_rate": 2.4600798577270183e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9450 + }, + { + "epoch": 2.752400349141693, + "flow_matching_loss": 0.01503771347925067, + "grad_norm": 0.0895153060555458, + "learning_rate": 2.4588088632968944e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9460 + }, + { + "epoch": 2.7553098632528368, + "flow_matching_loss": 0.0146324060857296, + "grad_norm": 0.07478678226470947, + "learning_rate": 2.4575367038235916e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9470 + }, + { + "epoch": 2.7582193773639805, + "flow_matching_loss": 0.014388056006282568, + "grad_norm": 0.08206956088542938, + "learning_rate": 2.4562633808529043e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9480 + }, + { + "epoch": 2.7611288914751237, + "flow_matching_loss": 0.015332716517150402, + "grad_norm": 0.09772808849811554, + "learning_rate": 2.4549888959320396e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9490 + }, + { + "epoch": 2.764038405586267, + "flow_matching_loss": 0.015038052201271057, + "grad_norm": 0.06888318061828613, + "learning_rate": 2.4537132506096162e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9500 + }, + { + "epoch": 2.7669479196974107, + "flow_matching_loss": 0.013249221816658973, + "grad_norm": 0.10542184859514236, + "learning_rate": 2.452436446435664e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9510 + }, + { + "epoch": 2.769857433808554, + "flow_matching_loss": 0.01617000736296177, + "grad_norm": 0.09840209782123566, + "learning_rate": 2.451158484961621e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9520 + }, + { + "epoch": 2.7727669479196972, + "flow_matching_loss": 0.01621273374184966, + "grad_norm": 0.08364176005125046, + "learning_rate": 2.4498793677403304e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9530 + }, + { + "epoch": 2.775676462030841, + "flow_matching_loss": 0.016367369797080755, + "grad_norm": 0.070923812687397, + "learning_rate": 2.4485990963260398e-05, + "loss": 0.0164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9540 + }, + { + "epoch": 2.778585976141984, + "flow_matching_loss": 0.014970184117555619, + "grad_norm": 0.10200533270835876, + "learning_rate": 2.4473176722744e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9550 + }, + { + "epoch": 2.7814954902531275, + "flow_matching_loss": 0.016712818574160338, + "grad_norm": 0.08484915643930435, + "learning_rate": 2.446035097142463e-05, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9560 + }, + { + "epoch": 2.784405004364271, + "flow_matching_loss": 0.015175776835530997, + "grad_norm": 0.07625625282526016, + "learning_rate": 2.444751372488677e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9570 + }, + { + "epoch": 2.7873145184754144, + "flow_matching_loss": 0.017295822128653528, + "grad_norm": 0.0790567547082901, + "learning_rate": 2.443466499872889e-05, + "loss": 0.0173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9580 + }, + { + "epoch": 2.790224032586558, + "flow_matching_loss": 0.012869433406740427, + "grad_norm": 0.07763973623514175, + "learning_rate": 2.4421804808563413e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9590 + }, + { + "epoch": 2.7931335466977014, + "flow_matching_loss": 0.017448163591325283, + "grad_norm": 0.0970844253897667, + "learning_rate": 2.440893317001667e-05, + "loss": 0.0174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9600 + }, + { + "epoch": 2.796043060808845, + "flow_matching_loss": 0.013342344388365745, + "grad_norm": 0.0960264503955841, + "learning_rate": 2.4396050098728925e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9610 + }, + { + "epoch": 2.7989525749199884, + "flow_matching_loss": 0.014558465499430896, + "grad_norm": 0.07064047455787659, + "learning_rate": 2.4383155610354314e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9620 + }, + { + "epoch": 2.8018620890311317, + "flow_matching_loss": 0.01712046144530177, + "grad_norm": 0.08277909457683563, + "learning_rate": 2.437024972056087e-05, + "loss": 0.0171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9630 + }, + { + "epoch": 2.8047716031422754, + "flow_matching_loss": 0.014680145494639873, + "grad_norm": 0.09542010724544525, + "learning_rate": 2.4357332445030452e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9640 + }, + { + "epoch": 2.8076811172534186, + "flow_matching_loss": 0.015482053719460964, + "grad_norm": 0.08334318548440933, + "learning_rate": 2.434440379945878e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9650 + }, + { + "epoch": 2.810590631364562, + "flow_matching_loss": 0.01563199469819665, + "grad_norm": 0.1075943112373352, + "learning_rate": 2.4331463799555366e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9660 + }, + { + "epoch": 2.8135001454757056, + "flow_matching_loss": 0.013681411743164062, + "grad_norm": 0.09816986322402954, + "learning_rate": 2.4318512461043544e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9670 + }, + { + "epoch": 2.816409659586849, + "flow_matching_loss": 0.013915338274091481, + "grad_norm": 0.09930470585823059, + "learning_rate": 2.4305549799660397e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9680 + }, + { + "epoch": 2.8193191736979926, + "flow_matching_loss": 0.015183210000395774, + "grad_norm": 0.08749457448720932, + "learning_rate": 2.429257583115679e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9690 + }, + { + "epoch": 2.822228687809136, + "flow_matching_loss": 0.015697335451841356, + "grad_norm": 0.09840525686740875, + "learning_rate": 2.4279590571297316e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9700 + }, + { + "epoch": 2.8251382019202795, + "flow_matching_loss": 0.014781638886779547, + "grad_norm": 0.07759588956832886, + "learning_rate": 2.4266594035860288e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9710 + }, + { + "epoch": 2.828047716031423, + "flow_matching_loss": 0.014787136297672986, + "grad_norm": 0.09437765181064606, + "learning_rate": 2.425358624063773e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9720 + }, + { + "epoch": 2.830957230142566, + "flow_matching_loss": 0.012360894866287709, + "grad_norm": 0.08918270468711853, + "learning_rate": 2.424056720143532e-05, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9730 + }, + { + "epoch": 2.8338667442537098, + "flow_matching_loss": 0.013945755362510682, + "grad_norm": 0.08569581806659698, + "learning_rate": 2.422753693407244e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9740 + }, + { + "epoch": 2.836776258364853, + "flow_matching_loss": 0.013708339910954237, + "grad_norm": 0.08064734935760498, + "learning_rate": 2.421449545438209e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9750 + }, + { + "epoch": 2.8396857724759963, + "flow_matching_loss": 0.016157397907227276, + "grad_norm": 0.08180519193410873, + "learning_rate": 2.4201442778210887e-05, + "loss": 0.0162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9760 + }, + { + "epoch": 2.84259528658714, + "flow_matching_loss": 0.015459568984806538, + "grad_norm": 0.07879240810871124, + "learning_rate": 2.418837892141907e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9770 + }, + { + "epoch": 2.8455048006982833, + "flow_matching_loss": 0.014630671683698892, + "grad_norm": 0.07521244883537292, + "learning_rate": 2.4175303899880458e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9780 + }, + { + "epoch": 2.848414314809427, + "flow_matching_loss": 0.014722216222435237, + "grad_norm": 0.06849487870931625, + "learning_rate": 2.4162217729482427e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9790 + }, + { + "epoch": 2.8513238289205702, + "flow_matching_loss": 0.013697145506739616, + "grad_norm": 0.10694468766450882, + "learning_rate": 2.4149120426125918e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9800 + }, + { + "epoch": 2.854233343031714, + "flow_matching_loss": 0.013488055393099784, + "grad_norm": 0.07034002989530563, + "learning_rate": 2.413601200572538e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9810 + }, + { + "epoch": 2.857142857142857, + "flow_matching_loss": 0.01455203564837575, + "grad_norm": 0.08084866404533386, + "learning_rate": 2.4122892484208794e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9820 + }, + { + "epoch": 2.8600523712540005, + "flow_matching_loss": 0.015107475686818361, + "grad_norm": 0.08621092885732651, + "learning_rate": 2.4109761877517603e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9830 + }, + { + "epoch": 2.862961885365144, + "flow_matching_loss": 0.015463987737894059, + "grad_norm": 0.06951122730970383, + "learning_rate": 2.4096620201606737e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9840 + }, + { + "epoch": 2.8658713994762874, + "flow_matching_loss": 0.013006847817450762, + "grad_norm": 0.06366241723299026, + "learning_rate": 2.4083467472444573e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9850 + }, + { + "epoch": 2.8687809135874307, + "flow_matching_loss": 0.013556351698935033, + "grad_norm": 0.0836540013551712, + "learning_rate": 2.4070303706012912e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9860 + }, + { + "epoch": 2.8716904276985744, + "flow_matching_loss": 0.014256934728473425, + "grad_norm": 0.07913332432508469, + "learning_rate": 2.4057128918306975e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9870 + }, + { + "epoch": 2.8745999418097177, + "flow_matching_loss": 0.013728173449635506, + "grad_norm": 0.07458609342575073, + "learning_rate": 2.4043943125335373e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9880 + }, + { + "epoch": 2.8775094559208614, + "flow_matching_loss": 0.015425722859799863, + "grad_norm": 0.07548735290765762, + "learning_rate": 2.403074634312008e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9890 + }, + { + "epoch": 2.8804189700320046, + "flow_matching_loss": 0.01499072564765811, + "grad_norm": 0.07985015213489532, + "learning_rate": 2.4017538587696442e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9900 + }, + { + "epoch": 2.8833284841431484, + "flow_matching_loss": 0.013503898959606886, + "grad_norm": 0.07222063094377518, + "learning_rate": 2.4004319875113118e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9910 + }, + { + "epoch": 2.8862379982542916, + "flow_matching_loss": 0.014529114030301571, + "grad_norm": 0.08436278253793716, + "learning_rate": 2.3991090221432096e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9920 + }, + { + "epoch": 2.889147512365435, + "flow_matching_loss": 0.014713047351688147, + "grad_norm": 0.07220755517482758, + "learning_rate": 2.397784964272865e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9930 + }, + { + "epoch": 2.8920570264765786, + "flow_matching_loss": 0.014631029963493348, + "grad_norm": 0.0783870741724968, + "learning_rate": 2.3964598155091335e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9940 + }, + { + "epoch": 2.894966540587722, + "flow_matching_loss": 0.014845400676131248, + "grad_norm": 0.0637926384806633, + "learning_rate": 2.3951335774621952e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9950 + }, + { + "epoch": 2.897876054698865, + "flow_matching_loss": 0.015225117467343808, + "grad_norm": 0.07077085971832275, + "learning_rate": 2.3938062517435548e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9960 + }, + { + "epoch": 2.900785568810009, + "flow_matching_loss": 0.01664031147956848, + "grad_norm": 0.0749434232711792, + "learning_rate": 2.392477839966038e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9970 + }, + { + "epoch": 2.903695082921152, + "flow_matching_loss": 0.014667832944542169, + "grad_norm": 0.07881926745176315, + "learning_rate": 2.3911483437437904e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9980 + }, + { + "epoch": 2.9066045970322953, + "flow_matching_loss": 0.014025404211133718, + "grad_norm": 0.09098528325557709, + "learning_rate": 2.3898177646922757e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9990 + }, + { + "epoch": 2.909514111143439, + "flow_matching_loss": 0.012857913039624691, + "grad_norm": 0.07983743399381638, + "learning_rate": 2.3884861044282722e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10000 + }, + { + "epoch": 2.9124236252545828, + "flow_matching_loss": 0.013353279512375592, + "grad_norm": 0.09507381916046143, + "learning_rate": 2.3871533645698732e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10010 + }, + { + "epoch": 2.915333139365726, + "flow_matching_loss": 0.013720424473285675, + "grad_norm": 0.06656941026449203, + "learning_rate": 2.3858195467364833e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10020 + }, + { + "epoch": 2.9182426534768693, + "flow_matching_loss": 0.012723658699542284, + "grad_norm": 0.07606414705514908, + "learning_rate": 2.3844846525488166e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10030 + }, + { + "epoch": 2.921152167588013, + "flow_matching_loss": 0.013288330845534801, + "grad_norm": 0.08039414137601852, + "learning_rate": 2.3831486836288957e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10040 + }, + { + "epoch": 2.9240616816991563, + "flow_matching_loss": 0.01392119387164712, + "grad_norm": 0.0882871225476265, + "learning_rate": 2.381811641600048e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10050 + }, + { + "epoch": 2.9269711958102995, + "flow_matching_loss": 0.014963586535304784, + "grad_norm": 0.09771090000867844, + "learning_rate": 2.3804735280869073e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10060 + }, + { + "epoch": 2.9298807099214432, + "flow_matching_loss": 0.013999245502054692, + "grad_norm": 0.07588636130094528, + "learning_rate": 2.3791343447154064e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10070 + }, + { + "epoch": 2.9327902240325865, + "flow_matching_loss": 0.0142815631814301, + "grad_norm": 0.07045506685972214, + "learning_rate": 2.37779409311278e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10080 + }, + { + "epoch": 2.9356997381437298, + "flow_matching_loss": 0.015190581511706113, + "grad_norm": 0.08992680162191391, + "learning_rate": 2.3764527749075596e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10090 + }, + { + "epoch": 2.9386092522548735, + "flow_matching_loss": 0.014614090975373983, + "grad_norm": 0.07164852321147919, + "learning_rate": 2.3751103917295735e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10100 + }, + { + "epoch": 2.9415187663660167, + "flow_matching_loss": 0.015667822677642107, + "grad_norm": 0.07501152157783508, + "learning_rate": 2.373766945209944e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10110 + }, + { + "epoch": 2.9444282804771604, + "flow_matching_loss": 0.015528970584273338, + "grad_norm": 0.10167042911052704, + "learning_rate": 2.3724224369810856e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10120 + }, + { + "epoch": 2.9473377945883037, + "flow_matching_loss": 0.01495260763913393, + "grad_norm": 0.09525524824857712, + "learning_rate": 2.3710768686767015e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10130 + }, + { + "epoch": 2.9502473086994474, + "flow_matching_loss": 0.013934884872287511, + "grad_norm": 0.06943190097808838, + "learning_rate": 2.3697302419317856e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10140 + }, + { + "epoch": 2.9531568228105907, + "flow_matching_loss": 0.014350321516394616, + "grad_norm": 0.12464538216590881, + "learning_rate": 2.3683825583826146e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10150 + }, + { + "epoch": 2.956066336921734, + "flow_matching_loss": 0.01361681092530489, + "grad_norm": 0.0837177112698555, + "learning_rate": 2.3670338196667528e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10160 + }, + { + "epoch": 2.9589758510328776, + "flow_matching_loss": 0.013217182550579309, + "grad_norm": 0.08760584890842438, + "learning_rate": 2.3656840274230443e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10170 + }, + { + "epoch": 2.961885365144021, + "flow_matching_loss": 0.014024127647280693, + "grad_norm": 0.09452732652425766, + "learning_rate": 2.3643331832916133e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10180 + }, + { + "epoch": 2.964794879255164, + "flow_matching_loss": 0.016653187572956085, + "grad_norm": 0.11739503592252731, + "learning_rate": 2.362981288913863e-05, + "loss": 0.0167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10190 + }, + { + "epoch": 2.967704393366308, + "flow_matching_loss": 0.01424887329339981, + "grad_norm": 0.08051102608442307, + "learning_rate": 2.3616283459324737e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10200 + }, + { + "epoch": 2.970613907477451, + "flow_matching_loss": 0.012909305281937122, + "grad_norm": 0.0736803337931633, + "learning_rate": 2.3602743559913973e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10210 + }, + { + "epoch": 2.973523421588595, + "flow_matching_loss": 0.013517744652926923, + "grad_norm": 0.09868516772985458, + "learning_rate": 2.35891932073586e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10220 + }, + { + "epoch": 2.976432935699738, + "flow_matching_loss": 0.014657100290060043, + "grad_norm": 0.10416056960821152, + "learning_rate": 2.357563241812357e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10230 + }, + { + "epoch": 2.979342449810882, + "flow_matching_loss": 0.014155320543795823, + "grad_norm": 0.08894718438386917, + "learning_rate": 2.3562061208686522e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10240 + }, + { + "epoch": 2.982251963922025, + "flow_matching_loss": 0.014703120570629835, + "grad_norm": 0.09746775776147842, + "learning_rate": 2.354847959553776e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10250 + }, + { + "epoch": 2.9851614780331683, + "flow_matching_loss": 0.013981319218873977, + "grad_norm": 0.11957371979951859, + "learning_rate": 2.353488759518022e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10260 + }, + { + "epoch": 2.988070992144312, + "flow_matching_loss": 0.01410656301304698, + "grad_norm": 0.07225397974252701, + "learning_rate": 2.352128522412946e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10270 + }, + { + "epoch": 2.9909805062554553, + "flow_matching_loss": 0.013904337491840123, + "grad_norm": 0.0940210223197937, + "learning_rate": 2.3507672498913653e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10280 + }, + { + "epoch": 2.9938900203665986, + "flow_matching_loss": 0.014623110461980105, + "grad_norm": 0.1564757078886032, + "learning_rate": 2.3494049436073537e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10290 + }, + { + "epoch": 2.9967995344777423, + "flow_matching_loss": 0.014573990181088448, + "grad_norm": 0.0868067741394043, + "learning_rate": 2.348041605216242e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10300 + }, + { + "epoch": 2.9997090485888855, + "flow_matching_loss": 0.014331679511815309, + "grad_norm": 0.07495748996734619, + "learning_rate": 2.3466772363746146e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10310 + }, + { + "epoch": 3.0026185627000292, + "flow_matching_loss": 0.013898134231567383, + "grad_norm": 0.08442819118499756, + "learning_rate": 2.345311838740309e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10320 + }, + { + "epoch": 3.0055280768111725, + "flow_matching_loss": 0.013119673915207386, + "grad_norm": 0.11349183320999146, + "learning_rate": 2.343945413972411e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10330 + }, + { + "epoch": 3.0084375909223158, + "flow_matching_loss": 0.013838296383619308, + "grad_norm": 0.08836444467306137, + "learning_rate": 2.3425779637312568e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10340 + }, + { + "epoch": 3.0113471050334595, + "flow_matching_loss": 0.015083427261561155, + "grad_norm": 0.082401342689991, + "learning_rate": 2.3412094896784264e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10350 + }, + { + "epoch": 3.0142566191446027, + "flow_matching_loss": 0.01543148821219802, + "grad_norm": 0.09044619649648666, + "learning_rate": 2.339839993476745e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10360 + }, + { + "epoch": 3.0171661332557465, + "flow_matching_loss": 0.014356298744678498, + "grad_norm": 0.08719345182180405, + "learning_rate": 2.3384694767902804e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10370 + }, + { + "epoch": 3.0200756473668897, + "flow_matching_loss": 0.012503887433558703, + "grad_norm": 0.07855990529060364, + "learning_rate": 2.3370979412843384e-05, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10380 + }, + { + "epoch": 3.022985161478033, + "flow_matching_loss": 0.013710315339267254, + "grad_norm": 0.06917387247085571, + "learning_rate": 2.3357253886254648e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10390 + }, + { + "epoch": 3.0258946755891767, + "flow_matching_loss": 0.012411353643983603, + "grad_norm": 0.09441634267568588, + "learning_rate": 2.3343518204814407e-05, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10400 + }, + { + "epoch": 3.02880418970032, + "flow_matching_loss": 0.013876492436975241, + "grad_norm": 0.08281663060188293, + "learning_rate": 2.3329772385212798e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10410 + }, + { + "epoch": 3.0317137038114637, + "flow_matching_loss": 0.013756609614938497, + "grad_norm": 0.06574780493974686, + "learning_rate": 2.3316016444152296e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10420 + }, + { + "epoch": 3.034623217922607, + "flow_matching_loss": 0.014160884171724319, + "grad_norm": 0.07503997534513474, + "learning_rate": 2.3302250398347668e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10430 + }, + { + "epoch": 3.03753273203375, + "flow_matching_loss": 0.013907440099865198, + "grad_norm": 0.07462980598211288, + "learning_rate": 2.328847426452595e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10440 + }, + { + "epoch": 3.040442246144894, + "flow_matching_loss": 0.01395064676180482, + "grad_norm": 0.09544000029563904, + "learning_rate": 2.3274688059426456e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10450 + }, + { + "epoch": 3.043351760256037, + "flow_matching_loss": 0.015249690786004067, + "grad_norm": 0.08389827609062195, + "learning_rate": 2.3260891799800715e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10460 + }, + { + "epoch": 3.046261274367181, + "flow_matching_loss": 0.014854027703404427, + "grad_norm": 0.09345325827598572, + "learning_rate": 2.3247085502412482e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10470 + }, + { + "epoch": 3.049170788478324, + "flow_matching_loss": 0.013707280438393355, + "grad_norm": 0.07022792100906372, + "learning_rate": 2.3233269184037717e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10480 + }, + { + "epoch": 3.0520803025894674, + "flow_matching_loss": 0.015676969196647406, + "grad_norm": 0.08451235294342041, + "learning_rate": 2.3219442861464547e-05, + "loss": 0.0157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10490 + }, + { + "epoch": 3.054989816700611, + "flow_matching_loss": 0.01353941299021244, + "grad_norm": 0.07470034807920456, + "learning_rate": 2.3205606551493255e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10500 + }, + { + "epoch": 3.0578993308117544, + "flow_matching_loss": 0.01466761203482747, + "grad_norm": 0.07730909436941147, + "learning_rate": 2.319176027093627e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10510 + }, + { + "epoch": 3.060808844922898, + "flow_matching_loss": 0.015139707364141941, + "grad_norm": 0.09449778497219086, + "learning_rate": 2.3177904036618117e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10520 + }, + { + "epoch": 3.0637183590340413, + "flow_matching_loss": 0.013531203288584948, + "grad_norm": 0.1277318000793457, + "learning_rate": 2.316403786537544e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10530 + }, + { + "epoch": 3.0666278731451846, + "flow_matching_loss": 0.014829190447926521, + "grad_norm": 0.09834638983011246, + "learning_rate": 2.3150161774056933e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10540 + }, + { + "epoch": 3.0695373872563283, + "flow_matching_loss": 0.013278994336724281, + "grad_norm": 0.10686323046684265, + "learning_rate": 2.313627577952336e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10550 + }, + { + "epoch": 3.0724469013674716, + "flow_matching_loss": 0.013387982361018657, + "grad_norm": 0.07265585660934448, + "learning_rate": 2.312237989864752e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10560 + }, + { + "epoch": 3.0753564154786153, + "flow_matching_loss": 0.01514973696321249, + "grad_norm": 0.06913629919290543, + "learning_rate": 2.3108474148314205e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10570 + }, + { + "epoch": 3.0782659295897585, + "flow_matching_loss": 0.013448316138237715, + "grad_norm": 0.08079886436462402, + "learning_rate": 2.309455854542023e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10580 + }, + { + "epoch": 3.081175443700902, + "flow_matching_loss": 0.013715208042412996, + "grad_norm": 0.06823073327541351, + "learning_rate": 2.308063310687435e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10590 + }, + { + "epoch": 3.0840849578120455, + "flow_matching_loss": 0.01587248258292675, + "grad_norm": 0.07951945066452026, + "learning_rate": 2.306669784959729e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10600 + }, + { + "epoch": 3.0869944719231888, + "flow_matching_loss": 0.013497806712985038, + "grad_norm": 0.08932384848594666, + "learning_rate": 2.3052752790521703e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10610 + }, + { + "epoch": 3.0899039860343325, + "flow_matching_loss": 0.014177237264811992, + "grad_norm": 0.1252143383026123, + "learning_rate": 2.3038797946592152e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10620 + }, + { + "epoch": 3.0928135001454757, + "flow_matching_loss": 0.013957583997398614, + "grad_norm": 0.07429233193397522, + "learning_rate": 2.3024833334765087e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10630 + }, + { + "epoch": 3.095723014256619, + "flow_matching_loss": 0.012820713128894568, + "grad_norm": 0.06698773801326752, + "learning_rate": 2.301085897200883e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10640 + }, + { + "epoch": 3.0986325283677627, + "flow_matching_loss": 0.014030440896749496, + "grad_norm": 0.06427395343780518, + "learning_rate": 2.299687487530354e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10650 + }, + { + "epoch": 3.101542042478906, + "flow_matching_loss": 0.013940904475748538, + "grad_norm": 0.07613202184438705, + "learning_rate": 2.2982881061641228e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10660 + }, + { + "epoch": 3.1044515565900497, + "flow_matching_loss": 0.014922557026147842, + "grad_norm": 0.09453538805246353, + "learning_rate": 2.296887754802569e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10670 + }, + { + "epoch": 3.107361070701193, + "flow_matching_loss": 0.014631688594818115, + "grad_norm": 0.08028886467218399, + "learning_rate": 2.295486435147251e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10680 + }, + { + "epoch": 3.110270584812336, + "flow_matching_loss": 0.014200839400291442, + "grad_norm": 0.09312410652637482, + "learning_rate": 2.294084148900905e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10690 + }, + { + "epoch": 3.11318009892348, + "flow_matching_loss": 0.013879347778856754, + "grad_norm": 0.09993400424718857, + "learning_rate": 2.292680897767441e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10700 + }, + { + "epoch": 3.116089613034623, + "flow_matching_loss": 0.014725468773394822, + "grad_norm": 0.08879311382770538, + "learning_rate": 2.291276683451941e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10710 + }, + { + "epoch": 3.1189991271457664, + "flow_matching_loss": 0.014042444992810488, + "grad_norm": 0.0804884061217308, + "learning_rate": 2.2898715076606584e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10720 + }, + { + "epoch": 3.12190864125691, + "flow_matching_loss": 0.011860214360058307, + "grad_norm": 0.0807395800948143, + "learning_rate": 2.2884653721010135e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10730 + }, + { + "epoch": 3.1248181553680534, + "flow_matching_loss": 0.015622120536863803, + "grad_norm": 0.07151219993829727, + "learning_rate": 2.2870582784815946e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10740 + }, + { + "epoch": 3.127727669479197, + "flow_matching_loss": 0.013118054810911417, + "grad_norm": 0.07250900566577911, + "learning_rate": 2.2856502285121525e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10750 + }, + { + "epoch": 3.1306371835903404, + "flow_matching_loss": 0.01494838111102581, + "grad_norm": 0.08584984391927719, + "learning_rate": 2.2842412239036004e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10760 + }, + { + "epoch": 3.133546697701484, + "flow_matching_loss": 0.013828353118151426, + "grad_norm": 0.09076587855815887, + "learning_rate": 2.2828312663680125e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10770 + }, + { + "epoch": 3.1364562118126273, + "flow_matching_loss": 0.014520090166479349, + "grad_norm": 0.0996817797422409, + "learning_rate": 2.2814203576186194e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10780 + }, + { + "epoch": 3.1393657259237706, + "flow_matching_loss": 0.013529853243380786, + "grad_norm": 0.08522464334964752, + "learning_rate": 2.2800084993698093e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10790 + }, + { + "epoch": 3.1422752400349143, + "flow_matching_loss": 0.013756954204291105, + "grad_norm": 0.10148672759532928, + "learning_rate": 2.2785956933371224e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10800 + }, + { + "epoch": 3.1451847541460576, + "flow_matching_loss": 0.014472505450248719, + "grad_norm": 0.06309657543897629, + "learning_rate": 2.2771819412372505e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10810 + }, + { + "epoch": 3.148094268257201, + "flow_matching_loss": 0.013321712333709, + "grad_norm": 0.0855652317404747, + "learning_rate": 2.2757672447880374e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10820 + }, + { + "epoch": 3.1510037823683446, + "flow_matching_loss": 0.01369466194882989, + "grad_norm": 0.09337219595909119, + "learning_rate": 2.2743516057084712e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10830 + }, + { + "epoch": 3.153913296479488, + "flow_matching_loss": 0.013721063220873474, + "grad_norm": 0.08653906732797623, + "learning_rate": 2.2729350257186877e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10840 + }, + { + "epoch": 3.1568228105906315, + "flow_matching_loss": 0.013329101260751486, + "grad_norm": 0.09091076999902725, + "learning_rate": 2.2715175065399655e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10850 + }, + { + "epoch": 3.159732324701775, + "flow_matching_loss": 0.013469808362424373, + "grad_norm": 0.09407981485128403, + "learning_rate": 2.270099049894723e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10860 + }, + { + "epoch": 3.162641838812918, + "flow_matching_loss": 0.014145991764962673, + "grad_norm": 0.07109639793634415, + "learning_rate": 2.268679657506519e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10870 + }, + { + "epoch": 3.1655513529240618, + "flow_matching_loss": 0.013825444784015418, + "grad_norm": 0.10215814411640167, + "learning_rate": 2.2672593311000498e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10880 + }, + { + "epoch": 3.168460867035205, + "flow_matching_loss": 0.013589336071163416, + "grad_norm": 0.07786770164966583, + "learning_rate": 2.2658380724011452e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10890 + }, + { + "epoch": 3.1713703811463487, + "flow_matching_loss": 0.013472939748317004, + "grad_norm": 0.07568488270044327, + "learning_rate": 2.264415883136769e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10900 + }, + { + "epoch": 3.174279895257492, + "flow_matching_loss": 0.014562903344631195, + "grad_norm": 0.0773816928267479, + "learning_rate": 2.262992765035015e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10910 + }, + { + "epoch": 3.1771894093686353, + "flow_matching_loss": 0.013792482111603022, + "grad_norm": 0.08568373322486877, + "learning_rate": 2.2615687198251062e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10920 + }, + { + "epoch": 3.180098923479779, + "flow_matching_loss": 0.014794853329658509, + "grad_norm": 0.07479433715343475, + "learning_rate": 2.2601437492373914e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10930 + }, + { + "epoch": 3.1830084375909222, + "flow_matching_loss": 0.015869879722595216, + "grad_norm": 0.07617916911840439, + "learning_rate": 2.2587178550033444e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10940 + }, + { + "epoch": 3.185917951702066, + "flow_matching_loss": 0.013798694964498282, + "grad_norm": 0.08530832082033157, + "learning_rate": 2.2572910388555615e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10950 + }, + { + "epoch": 3.188827465813209, + "flow_matching_loss": 0.013558348454535008, + "grad_norm": 0.07614919543266296, + "learning_rate": 2.255863302527758e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10960 + }, + { + "epoch": 3.1917369799243525, + "flow_matching_loss": 0.015636995900422335, + "grad_norm": 0.08498238772153854, + "learning_rate": 2.2544346477547685e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10970 + }, + { + "epoch": 3.194646494035496, + "flow_matching_loss": 0.011756140738725662, + "grad_norm": 0.07512717694044113, + "learning_rate": 2.253005076272544e-05, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10980 + }, + { + "epoch": 3.1975560081466394, + "flow_matching_loss": 0.015300061367452145, + "grad_norm": 0.09346916526556015, + "learning_rate": 2.2515745898181477e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10990 + }, + { + "epoch": 3.200465522257783, + "flow_matching_loss": 0.01538235079497099, + "grad_norm": 0.08709216117858887, + "learning_rate": 2.250143190129756e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11000 + }, + { + "epoch": 3.2033750363689264, + "flow_matching_loss": 0.014125407487154008, + "grad_norm": 0.0865541324019432, + "learning_rate": 2.2487108789466547e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11010 + }, + { + "epoch": 3.2062845504800697, + "flow_matching_loss": 0.014140241220593453, + "grad_norm": 0.09022865444421768, + "learning_rate": 2.247277658009236e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11020 + }, + { + "epoch": 3.2091940645912134, + "flow_matching_loss": 0.012698497204110027, + "grad_norm": 0.0780833289027214, + "learning_rate": 2.245843529059e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11030 + }, + { + "epoch": 3.2121035787023566, + "flow_matching_loss": 0.013327318336814642, + "grad_norm": 0.09386253356933594, + "learning_rate": 2.244408493838547e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11040 + }, + { + "epoch": 3.2150130928135003, + "flow_matching_loss": 0.013207444269210101, + "grad_norm": 0.06977652758359909, + "learning_rate": 2.242972554091581e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11050 + }, + { + "epoch": 3.2179226069246436, + "flow_matching_loss": 0.013354569859802722, + "grad_norm": 0.08579529076814651, + "learning_rate": 2.2415357115629045e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11060 + }, + { + "epoch": 3.220832121035787, + "flow_matching_loss": 0.014178430661559105, + "grad_norm": 0.07672044634819031, + "learning_rate": 2.2400979679984162e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11070 + }, + { + "epoch": 3.2237416351469306, + "flow_matching_loss": 0.01632453305646777, + "grad_norm": 0.08430980890989304, + "learning_rate": 2.2386593251451096e-05, + "loss": 0.0163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11080 + }, + { + "epoch": 3.226651149258074, + "flow_matching_loss": 0.014158994238823653, + "grad_norm": 0.08130483329296112, + "learning_rate": 2.2372197847510722e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11090 + }, + { + "epoch": 3.2295606633692175, + "flow_matching_loss": 0.01317956978455186, + "grad_norm": 0.09883489459753036, + "learning_rate": 2.2357793485654813e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11100 + }, + { + "epoch": 3.232470177480361, + "flow_matching_loss": 0.014091561082750559, + "grad_norm": 0.0816628709435463, + "learning_rate": 2.2343380183386017e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11110 + }, + { + "epoch": 3.235379691591504, + "flow_matching_loss": 0.013459017872810364, + "grad_norm": 0.07271486520767212, + "learning_rate": 2.232895795821787e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11120 + }, + { + "epoch": 3.2382892057026478, + "flow_matching_loss": 0.013015996851027011, + "grad_norm": 0.10214908421039581, + "learning_rate": 2.2314526827674723e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11130 + }, + { + "epoch": 3.241198719813791, + "flow_matching_loss": 0.013851211871951818, + "grad_norm": 0.10515151917934418, + "learning_rate": 2.230008680929177e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11140 + }, + { + "epoch": 3.2441082339249343, + "flow_matching_loss": 0.015372201800346375, + "grad_norm": 0.08537808805704117, + "learning_rate": 2.2285637920614987e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11150 + }, + { + "epoch": 3.247017748036078, + "flow_matching_loss": 0.013339553028345108, + "grad_norm": 0.07967277616262436, + "learning_rate": 2.2271180179201143e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11160 + }, + { + "epoch": 3.2499272621472213, + "flow_matching_loss": 0.015230979491025209, + "grad_norm": 0.07238580286502838, + "learning_rate": 2.225671360261775e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11170 + }, + { + "epoch": 3.252836776258365, + "flow_matching_loss": 0.014225568436086179, + "grad_norm": 0.06568647921085358, + "learning_rate": 2.2242238208443068e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11180 + }, + { + "epoch": 3.2557462903695082, + "flow_matching_loss": 0.013780818227678538, + "grad_norm": 0.06033068522810936, + "learning_rate": 2.2227754014266063e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11190 + }, + { + "epoch": 3.258655804480652, + "flow_matching_loss": 0.01444034157320857, + "grad_norm": 0.06597112864255905, + "learning_rate": 2.22132610376864e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11200 + }, + { + "epoch": 3.261565318591795, + "flow_matching_loss": 0.015333375707268716, + "grad_norm": 0.0825604721903801, + "learning_rate": 2.219875929631441e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11210 + }, + { + "epoch": 3.2644748327029385, + "flow_matching_loss": 0.01329101249575615, + "grad_norm": 0.08325021713972092, + "learning_rate": 2.2184248807771074e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11220 + }, + { + "epoch": 3.267384346814082, + "flow_matching_loss": 0.013590831495821475, + "grad_norm": 0.07430235296487808, + "learning_rate": 2.216972958968801e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11230 + }, + { + "epoch": 3.2702938609252254, + "flow_matching_loss": 0.013354403711855412, + "grad_norm": 0.06958994269371033, + "learning_rate": 2.215520165970743e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11240 + }, + { + "epoch": 3.2732033750363687, + "flow_matching_loss": 0.012267412804067134, + "grad_norm": 0.0903317853808403, + "learning_rate": 2.2140665035482143e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11250 + }, + { + "epoch": 3.2761128891475124, + "flow_matching_loss": 0.013987416494637727, + "grad_norm": 0.08798271417617798, + "learning_rate": 2.2126119734675518e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11260 + }, + { + "epoch": 3.2790224032586557, + "flow_matching_loss": 0.01343520162627101, + "grad_norm": 0.09454398602247238, + "learning_rate": 2.2111565774961463e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11270 + }, + { + "epoch": 3.2819319173697994, + "flow_matching_loss": 0.01348086092621088, + "grad_norm": 0.06913752853870392, + "learning_rate": 2.2097003174024417e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11280 + }, + { + "epoch": 3.2848414314809427, + "flow_matching_loss": 0.013708162214607, + "grad_norm": 0.08407393097877502, + "learning_rate": 2.2082431949559315e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11290 + }, + { + "epoch": 3.2877509455920864, + "flow_matching_loss": 0.013959156349301338, + "grad_norm": 0.08854733407497406, + "learning_rate": 2.2067852119271558e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11300 + }, + { + "epoch": 3.2906604597032296, + "flow_matching_loss": 0.014404922723770142, + "grad_norm": 0.09436383843421936, + "learning_rate": 2.2053263700877026e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11310 + }, + { + "epoch": 3.293569973814373, + "flow_matching_loss": 0.014572698716074228, + "grad_norm": 0.06549359858036041, + "learning_rate": 2.2038666712102015e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11320 + }, + { + "epoch": 3.2964794879255166, + "flow_matching_loss": 0.014291741233319045, + "grad_norm": 0.07289101928472519, + "learning_rate": 2.2024061170683246e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11330 + }, + { + "epoch": 3.29938900203666, + "flow_matching_loss": 0.015382790472358465, + "grad_norm": 0.10202056914567947, + "learning_rate": 2.200944709436783e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11340 + }, + { + "epoch": 3.302298516147803, + "flow_matching_loss": 0.014433322846889496, + "grad_norm": 0.10120470821857452, + "learning_rate": 2.1994824500913246e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11350 + }, + { + "epoch": 3.305208030258947, + "flow_matching_loss": 0.015438137110322714, + "grad_norm": 0.06376376748085022, + "learning_rate": 2.1980193408087322e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11360 + }, + { + "epoch": 3.30811754437009, + "flow_matching_loss": 0.014355490170419215, + "grad_norm": 0.07907161116600037, + "learning_rate": 2.1965553833668224e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11370 + }, + { + "epoch": 3.311027058481234, + "flow_matching_loss": 0.01393259409815073, + "grad_norm": 0.07198983430862427, + "learning_rate": 2.195090579544441e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11380 + }, + { + "epoch": 3.313936572592377, + "flow_matching_loss": 0.013784467615187168, + "grad_norm": 0.08532877266407013, + "learning_rate": 2.193624931121462e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11390 + }, + { + "epoch": 3.3168460867035203, + "flow_matching_loss": 0.013446565438061952, + "grad_norm": 0.08343909680843353, + "learning_rate": 2.1921584398787872e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11400 + }, + { + "epoch": 3.319755600814664, + "flow_matching_loss": 0.013643037900328637, + "grad_norm": 0.08577539026737213, + "learning_rate": 2.1906911075983418e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11410 + }, + { + "epoch": 3.3226651149258073, + "flow_matching_loss": 0.011828726436942816, + "grad_norm": 0.07530295848846436, + "learning_rate": 2.1892229360630728e-05, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11420 + }, + { + "epoch": 3.325574629036951, + "flow_matching_loss": 0.013280759006738663, + "grad_norm": 0.09595679491758347, + "learning_rate": 2.187753927056947e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11430 + }, + { + "epoch": 3.3284841431480943, + "flow_matching_loss": 0.014527872018516064, + "grad_norm": 0.0939149484038353, + "learning_rate": 2.186284082364949e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11440 + }, + { + "epoch": 3.3313936572592375, + "flow_matching_loss": 0.013915737066417933, + "grad_norm": 0.0866771712899208, + "learning_rate": 2.1848134037730786e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11450 + }, + { + "epoch": 3.3343031713703812, + "flow_matching_loss": 0.01376499729231, + "grad_norm": 0.08940687775611877, + "learning_rate": 2.183341893068349e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11460 + }, + { + "epoch": 3.3372126854815245, + "flow_matching_loss": 0.013918669056147337, + "grad_norm": 0.07216468453407288, + "learning_rate": 2.1818695520387842e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11470 + }, + { + "epoch": 3.340122199592668, + "flow_matching_loss": 0.012257633544504643, + "grad_norm": 0.06399736553430557, + "learning_rate": 2.1803963824734173e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11480 + }, + { + "epoch": 3.3430317137038115, + "flow_matching_loss": 0.013238680548965931, + "grad_norm": 0.09579338133335114, + "learning_rate": 2.1789223861622887e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11490 + }, + { + "epoch": 3.3459412278149547, + "flow_matching_loss": 0.01398373544216156, + "grad_norm": 0.10460855811834335, + "learning_rate": 2.1774475648964428e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11500 + }, + { + "epoch": 3.3488507419260984, + "flow_matching_loss": 0.013530689477920532, + "grad_norm": 0.07868190109729767, + "learning_rate": 2.1759719204679267e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11510 + }, + { + "epoch": 3.3517602560372417, + "flow_matching_loss": 0.01412569424137473, + "grad_norm": 0.08569879829883575, + "learning_rate": 2.174495454669787e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11520 + }, + { + "epoch": 3.3546697701483854, + "flow_matching_loss": 0.01528460429981351, + "grad_norm": 0.07405132800340652, + "learning_rate": 2.1730181692960692e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11530 + }, + { + "epoch": 3.3575792842595287, + "flow_matching_loss": 0.012878464348614217, + "grad_norm": 0.1015937551856041, + "learning_rate": 2.171540066141814e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11540 + }, + { + "epoch": 3.360488798370672, + "flow_matching_loss": 0.01447745794430375, + "grad_norm": 0.11258327960968018, + "learning_rate": 2.170061147003056e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11550 + }, + { + "epoch": 3.3633983124818156, + "flow_matching_loss": 0.015626346878707408, + "grad_norm": 0.09761753678321838, + "learning_rate": 2.1685814136768225e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11560 + }, + { + "epoch": 3.366307826592959, + "flow_matching_loss": 0.012681196350604296, + "grad_norm": 0.06500336527824402, + "learning_rate": 2.167100867961128e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11570 + }, + { + "epoch": 3.369217340704102, + "flow_matching_loss": 0.015879432018846274, + "grad_norm": 0.08665972203016281, + "learning_rate": 2.1656195116549756e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11580 + }, + { + "epoch": 3.372126854815246, + "flow_matching_loss": 0.013555391505360604, + "grad_norm": 0.06950483471155167, + "learning_rate": 2.1641373465583527e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11590 + }, + { + "epoch": 3.375036368926389, + "flow_matching_loss": 0.01382079003378749, + "grad_norm": 0.07887714356184006, + "learning_rate": 2.1626543744722292e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11600 + }, + { + "epoch": 3.377945883037533, + "flow_matching_loss": 0.012747853249311446, + "grad_norm": 0.08868804574012756, + "learning_rate": 2.161170597198557e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11610 + }, + { + "epoch": 3.380855397148676, + "flow_matching_loss": 0.013464865554124118, + "grad_norm": 0.09149467945098877, + "learning_rate": 2.1596860165402645e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11620 + }, + { + "epoch": 3.38376491125982, + "flow_matching_loss": 0.014430978894233703, + "grad_norm": 0.09185919165611267, + "learning_rate": 2.1582006343012576e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11630 + }, + { + "epoch": 3.386674425370963, + "flow_matching_loss": 0.014054027292877436, + "grad_norm": 0.06894227117300034, + "learning_rate": 2.156714452286416e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11640 + }, + { + "epoch": 3.3895839394821063, + "flow_matching_loss": 0.014840505085885525, + "grad_norm": 0.08407308906316757, + "learning_rate": 2.155227472301591e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11650 + }, + { + "epoch": 3.39249345359325, + "flow_matching_loss": 0.01428353488445282, + "grad_norm": 0.0880606472492218, + "learning_rate": 2.1537396961536027e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11660 + }, + { + "epoch": 3.3954029677043933, + "flow_matching_loss": 0.014282704517245292, + "grad_norm": 0.0842691957950592, + "learning_rate": 2.1522511256502403e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11670 + }, + { + "epoch": 3.3983124818155366, + "flow_matching_loss": 0.013892639614641666, + "grad_norm": 0.10023491084575653, + "learning_rate": 2.1507617626002566e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11680 + }, + { + "epoch": 3.4012219959266803, + "flow_matching_loss": 0.014638228993862867, + "grad_norm": 0.06712905317544937, + "learning_rate": 2.1492716088133685e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11690 + }, + { + "epoch": 3.4041315100378235, + "flow_matching_loss": 0.013100909627974033, + "grad_norm": 0.06686848402023315, + "learning_rate": 2.1477806661002536e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11700 + }, + { + "epoch": 3.4070410241489673, + "flow_matching_loss": 0.014871411677449941, + "grad_norm": 0.07408462464809418, + "learning_rate": 2.146288936272548e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11710 + }, + { + "epoch": 3.4099505382601105, + "flow_matching_loss": 0.013905200082808733, + "grad_norm": 0.08191309869289398, + "learning_rate": 2.1447964211428433e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11720 + }, + { + "epoch": 3.4128600523712542, + "flow_matching_loss": 0.012521026749163866, + "grad_norm": 0.07456589490175247, + "learning_rate": 2.1433031225246865e-05, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11730 + }, + { + "epoch": 3.4157695664823975, + "flow_matching_loss": 0.016621217597275972, + "grad_norm": 0.08563211560249329, + "learning_rate": 2.1418090422325766e-05, + "loss": 0.0166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11740 + }, + { + "epoch": 3.4186790805935408, + "flow_matching_loss": 0.013378184381872415, + "grad_norm": 0.07568282634019852, + "learning_rate": 2.1403141820819623e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11750 + }, + { + "epoch": 3.4215885947046845, + "flow_matching_loss": 0.013905648794025182, + "grad_norm": 0.0975804403424263, + "learning_rate": 2.1388185438892377e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11760 + }, + { + "epoch": 3.4244981088158277, + "flow_matching_loss": 0.014023152925074101, + "grad_norm": 0.08300169557332993, + "learning_rate": 2.1373221294717465e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11770 + }, + { + "epoch": 3.427407622926971, + "flow_matching_loss": 0.014977919124066829, + "grad_norm": 0.14450213313102722, + "learning_rate": 2.1358249406477724e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11780 + }, + { + "epoch": 3.4303171370381147, + "flow_matching_loss": 0.014297822397202254, + "grad_norm": 0.08028066903352737, + "learning_rate": 2.1343269792365408e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11790 + }, + { + "epoch": 3.433226651149258, + "flow_matching_loss": 0.014985198806971311, + "grad_norm": 0.07794385403394699, + "learning_rate": 2.132828247058217e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11800 + }, + { + "epoch": 3.4361361652604017, + "flow_matching_loss": 0.013978570885956287, + "grad_norm": 0.08615535497665405, + "learning_rate": 2.1313287459339004e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11810 + }, + { + "epoch": 3.439045679371545, + "flow_matching_loss": 0.013246810343116523, + "grad_norm": 0.07310637831687927, + "learning_rate": 2.1298284776856276e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11820 + }, + { + "epoch": 3.4419551934826886, + "flow_matching_loss": 0.014454820845276117, + "grad_norm": 0.07885109633207321, + "learning_rate": 2.128327444136365e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11830 + }, + { + "epoch": 3.444864707593832, + "flow_matching_loss": 0.014562357496470214, + "grad_norm": 0.08522068709135056, + "learning_rate": 2.1268256471100105e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11840 + }, + { + "epoch": 3.447774221704975, + "flow_matching_loss": 0.012812746409326791, + "grad_norm": 0.07779952883720398, + "learning_rate": 2.1253230884313893e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11850 + }, + { + "epoch": 3.450683735816119, + "flow_matching_loss": 0.014360243547707796, + "grad_norm": 0.07574598491191864, + "learning_rate": 2.123819769926251e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11860 + }, + { + "epoch": 3.453593249927262, + "flow_matching_loss": 0.013374896068125964, + "grad_norm": 0.07058771699666977, + "learning_rate": 2.1223156934212702e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11870 + }, + { + "epoch": 3.4565027640384054, + "flow_matching_loss": 0.014003729540854692, + "grad_norm": 0.06990931183099747, + "learning_rate": 2.1208108607440418e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11880 + }, + { + "epoch": 3.459412278149549, + "flow_matching_loss": 0.013965758495032787, + "grad_norm": 0.10352489352226257, + "learning_rate": 2.119305273723078e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11890 + }, + { + "epoch": 3.4623217922606924, + "flow_matching_loss": 0.012862569000571967, + "grad_norm": 0.08729322999715805, + "learning_rate": 2.1177989341878106e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11900 + }, + { + "epoch": 3.465231306371836, + "flow_matching_loss": 0.013949212618172169, + "grad_norm": 0.07107767462730408, + "learning_rate": 2.116291843968583e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11910 + }, + { + "epoch": 3.4681408204829793, + "flow_matching_loss": 0.01448727333918214, + "grad_norm": 0.09886957705020905, + "learning_rate": 2.114784004896653e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11920 + }, + { + "epoch": 3.4710503345941226, + "flow_matching_loss": 0.014675617218017578, + "grad_norm": 0.07118137180805206, + "learning_rate": 2.113275418804187e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11930 + }, + { + "epoch": 3.4739598487052663, + "flow_matching_loss": 0.013335551787167787, + "grad_norm": 0.07729373127222061, + "learning_rate": 2.1117660875242583e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11940 + }, + { + "epoch": 3.4768693628164096, + "flow_matching_loss": 0.013712746743112802, + "grad_norm": 0.07678516954183578, + "learning_rate": 2.1102560128908478e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11950 + }, + { + "epoch": 3.4797788769275533, + "flow_matching_loss": 0.01289765303954482, + "grad_norm": 0.08976370841264725, + "learning_rate": 2.1087451967388384e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11960 + }, + { + "epoch": 3.4826883910386965, + "flow_matching_loss": 0.013323894422501325, + "grad_norm": 0.08615870773792267, + "learning_rate": 2.1072336409040135e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11970 + }, + { + "epoch": 3.48559790514984, + "flow_matching_loss": 0.014861629996448755, + "grad_norm": 0.0837198793888092, + "learning_rate": 2.1057213472230573e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11980 + }, + { + "epoch": 3.4885074192609835, + "flow_matching_loss": 0.015989808924496175, + "grad_norm": 0.10160668194293976, + "learning_rate": 2.1042083175335478e-05, + "loss": 0.016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11990 + }, + { + "epoch": 3.4914169333721268, + "flow_matching_loss": 0.012794542452320456, + "grad_norm": 0.08483412116765976, + "learning_rate": 2.1026945536739595e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12000 + }, + { + "epoch": 3.4943264474832705, + "flow_matching_loss": 0.014505849871784448, + "grad_norm": 0.07303226739168167, + "learning_rate": 2.1011800574836586e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12010 + }, + { + "epoch": 3.4972359615944137, + "flow_matching_loss": 0.013082619942724704, + "grad_norm": 0.0851777046918869, + "learning_rate": 2.0996648308028992e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12020 + }, + { + "epoch": 3.500145475705557, + "flow_matching_loss": 0.012974491529166698, + "grad_norm": 0.08859257400035858, + "learning_rate": 2.0981488754728266e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12030 + }, + { + "epoch": 3.5030549898167007, + "flow_matching_loss": 0.013555928599089384, + "grad_norm": 0.08258696645498276, + "learning_rate": 2.0966321933354685e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12040 + }, + { + "epoch": 3.505964503927844, + "flow_matching_loss": 0.013127357233315706, + "grad_norm": 0.0800786167383194, + "learning_rate": 2.095114786233736e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12050 + }, + { + "epoch": 3.5088740180389877, + "flow_matching_loss": 0.013519828207790852, + "grad_norm": 0.09467620402574539, + "learning_rate": 2.0935966560114236e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12060 + }, + { + "epoch": 3.511783532150131, + "flow_matching_loss": 0.013164359796792268, + "grad_norm": 0.07896920293569565, + "learning_rate": 2.092077804513201e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12070 + }, + { + "epoch": 3.514693046261274, + "flow_matching_loss": 0.013677843939512968, + "grad_norm": 0.07285791635513306, + "learning_rate": 2.0905582335846167e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12080 + }, + { + "epoch": 3.517602560372418, + "flow_matching_loss": 0.01412924388423562, + "grad_norm": 0.08013705909252167, + "learning_rate": 2.0890379450720927e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12090 + }, + { + "epoch": 3.520512074483561, + "flow_matching_loss": 0.01398180890828371, + "grad_norm": 0.07646368443965912, + "learning_rate": 2.087516940822923e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12100 + }, + { + "epoch": 3.5234215885947044, + "flow_matching_loss": 0.01407829038798809, + "grad_norm": 0.0881524458527565, + "learning_rate": 2.085995222685271e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12110 + }, + { + "epoch": 3.526331102705848, + "flow_matching_loss": 0.012942145951092244, + "grad_norm": 0.0603376142680645, + "learning_rate": 2.084472792508167e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12120 + }, + { + "epoch": 3.5292406168169914, + "flow_matching_loss": 0.012942001502960921, + "grad_norm": 0.07741373032331467, + "learning_rate": 2.082949652141508e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12130 + }, + { + "epoch": 3.532150130928135, + "flow_matching_loss": 0.013252376578748226, + "grad_norm": 0.08245828747749329, + "learning_rate": 2.081425803436054e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12140 + }, + { + "epoch": 3.5350596450392784, + "flow_matching_loss": 0.01431446047499776, + "grad_norm": 0.07975295186042786, + "learning_rate": 2.079901248243423e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12150 + }, + { + "epoch": 3.537969159150422, + "flow_matching_loss": 0.014932156633585691, + "grad_norm": 0.07865902781486511, + "learning_rate": 2.078375988416095e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12160 + }, + { + "epoch": 3.5408786732615654, + "flow_matching_loss": 0.012793684564530849, + "grad_norm": 0.05991195887327194, + "learning_rate": 2.076850025807404e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12170 + }, + { + "epoch": 3.5437881873727086, + "flow_matching_loss": 0.015136685501784087, + "grad_norm": 0.0877213329076767, + "learning_rate": 2.0753233622715374e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12180 + }, + { + "epoch": 3.5466977014838523, + "flow_matching_loss": 0.01280636191368103, + "grad_norm": 0.06043557450175285, + "learning_rate": 2.073795999663536e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12190 + }, + { + "epoch": 3.5496072155949956, + "flow_matching_loss": 0.013569577783346175, + "grad_norm": 0.07114829868078232, + "learning_rate": 2.07226793983929e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12200 + }, + { + "epoch": 3.552516729706139, + "flow_matching_loss": 0.015359111316502094, + "grad_norm": 0.08261126279830933, + "learning_rate": 2.070739184655536e-05, + "loss": 0.0154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12210 + }, + { + "epoch": 3.5554262438172826, + "flow_matching_loss": 0.01442403495311737, + "grad_norm": 0.09409912675619125, + "learning_rate": 2.069209735969855e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12220 + }, + { + "epoch": 3.558335757928426, + "flow_matching_loss": 0.015308712888509036, + "grad_norm": 0.11313486844301224, + "learning_rate": 2.067679595640672e-05, + "loss": 0.0153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12230 + }, + { + "epoch": 3.5612452720395695, + "flow_matching_loss": 0.012718404829502105, + "grad_norm": 0.10423050820827484, + "learning_rate": 2.066148765527252e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12240 + }, + { + "epoch": 3.564154786150713, + "flow_matching_loss": 0.01330976914614439, + "grad_norm": 0.10366906970739365, + "learning_rate": 2.0646172474896976e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12250 + }, + { + "epoch": 3.5670643002618565, + "flow_matching_loss": 0.013184835761785507, + "grad_norm": 0.09372839331626892, + "learning_rate": 2.063085043388947e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12260 + }, + { + "epoch": 3.5699738143729998, + "flow_matching_loss": 0.012950847763568163, + "grad_norm": 0.08443190902471542, + "learning_rate": 2.0615521550867732e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12270 + }, + { + "epoch": 3.572883328484143, + "flow_matching_loss": 0.01290441770106554, + "grad_norm": 0.0730128139257431, + "learning_rate": 2.0600185844457803e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12280 + }, + { + "epoch": 3.5757928425952867, + "flow_matching_loss": 0.013228435069322586, + "grad_norm": 0.07542489469051361, + "learning_rate": 2.058484333329401e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12290 + }, + { + "epoch": 3.57870235670643, + "flow_matching_loss": 0.013590777292847633, + "grad_norm": 0.09360262751579285, + "learning_rate": 2.0569494036018955e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12300 + }, + { + "epoch": 3.5816118708175733, + "flow_matching_loss": 0.013028672803193331, + "grad_norm": 0.09828761965036392, + "learning_rate": 2.055413797128347e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12310 + }, + { + "epoch": 3.584521384928717, + "flow_matching_loss": 0.012644825875759125, + "grad_norm": 0.07563616335391998, + "learning_rate": 2.0538775157746634e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12320 + }, + { + "epoch": 3.5874308990398602, + "flow_matching_loss": 0.012345390114933252, + "grad_norm": 0.06654352694749832, + "learning_rate": 2.0523405614075706e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12330 + }, + { + "epoch": 3.590340413151004, + "flow_matching_loss": 0.01282039675861597, + "grad_norm": 0.053753986954689026, + "learning_rate": 2.0508029358946132e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12340 + }, + { + "epoch": 3.593249927262147, + "flow_matching_loss": 0.013241526205092669, + "grad_norm": 0.06339375674724579, + "learning_rate": 2.0492646411041515e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12350 + }, + { + "epoch": 3.596159441373291, + "flow_matching_loss": 0.014204883389174939, + "grad_norm": 0.0750812366604805, + "learning_rate": 2.0477256789053587e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12360 + }, + { + "epoch": 3.599068955484434, + "flow_matching_loss": 0.013086938764899969, + "grad_norm": 0.10559969395399094, + "learning_rate": 2.0461860511682194e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12370 + }, + { + "epoch": 3.6019784695955774, + "flow_matching_loss": 0.013042191136628389, + "grad_norm": 0.09249567985534668, + "learning_rate": 2.0446457597635254e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12380 + }, + { + "epoch": 3.604887983706721, + "flow_matching_loss": 0.013475617486983537, + "grad_norm": 0.11489147692918777, + "learning_rate": 2.0431048065628763e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12390 + }, + { + "epoch": 3.6077974978178644, + "flow_matching_loss": 0.014174532610923052, + "grad_norm": 0.08195182681083679, + "learning_rate": 2.0415631934386764e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12400 + }, + { + "epoch": 3.6107070119290077, + "flow_matching_loss": 0.01383599154651165, + "grad_norm": 0.06786122918128967, + "learning_rate": 2.04002092226413e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12410 + }, + { + "epoch": 3.6136165260401514, + "flow_matching_loss": 0.013117772899568081, + "grad_norm": 0.09911109507083893, + "learning_rate": 2.038477994913242e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12420 + }, + { + "epoch": 3.6165260401512946, + "flow_matching_loss": 0.014558166824281215, + "grad_norm": 0.07203217595815659, + "learning_rate": 2.036934413260816e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12430 + }, + { + "epoch": 3.619435554262438, + "flow_matching_loss": 0.014575838390737773, + "grad_norm": 0.09684504568576813, + "learning_rate": 2.0353901791824473e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12440 + }, + { + "epoch": 3.6223450683735816, + "flow_matching_loss": 0.012725744396448135, + "grad_norm": 0.06816349923610687, + "learning_rate": 2.0338452945545275e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12450 + }, + { + "epoch": 3.6252545824847253, + "flow_matching_loss": 0.014372638799250126, + "grad_norm": 0.0615629144012928, + "learning_rate": 2.0322997612542366e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12460 + }, + { + "epoch": 3.6281640965958686, + "flow_matching_loss": 0.013994574546813965, + "grad_norm": 0.08781814575195312, + "learning_rate": 2.0307535811595427e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12470 + }, + { + "epoch": 3.631073610707012, + "flow_matching_loss": 0.013817962538450957, + "grad_norm": 0.07229514420032501, + "learning_rate": 2.0292067561492013e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12480 + }, + { + "epoch": 3.6339831248181556, + "flow_matching_loss": 0.014479922875761986, + "grad_norm": 0.1114216297864914, + "learning_rate": 2.02765928810275e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12490 + }, + { + "epoch": 3.636892638929299, + "flow_matching_loss": 0.0135107746347785, + "grad_norm": 0.08942416310310364, + "learning_rate": 2.0261111789005087e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12500 + }, + { + "epoch": 3.639802153040442, + "flow_matching_loss": 0.014093794953078031, + "grad_norm": 0.09742499142885208, + "learning_rate": 2.024562430423576e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12510 + }, + { + "epoch": 3.642711667151586, + "flow_matching_loss": 0.013716481626033783, + "grad_norm": 0.08961618691682816, + "learning_rate": 2.0230130445538273e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12520 + }, + { + "epoch": 3.645621181262729, + "flow_matching_loss": 0.013738163094967604, + "grad_norm": 0.0752427950501442, + "learning_rate": 2.0214630231739127e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12530 + }, + { + "epoch": 3.6485306953738723, + "flow_matching_loss": 0.013756055757403374, + "grad_norm": 0.08537019044160843, + "learning_rate": 2.0199123681672543e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12540 + }, + { + "epoch": 3.651440209485016, + "flow_matching_loss": 0.012845492549240589, + "grad_norm": 0.08816975355148315, + "learning_rate": 2.0183610814180437e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12550 + }, + { + "epoch": 3.6543497235961593, + "flow_matching_loss": 0.015161129273474216, + "grad_norm": 0.0802336186170578, + "learning_rate": 2.016809164811241e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12560 + }, + { + "epoch": 3.657259237707303, + "flow_matching_loss": 0.013891558162868023, + "grad_norm": 0.06539139151573181, + "learning_rate": 2.015256620232571e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12570 + }, + { + "epoch": 3.6601687518184463, + "flow_matching_loss": 0.013804935105144977, + "grad_norm": 0.07553073018789291, + "learning_rate": 2.0137034495685222e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12580 + }, + { + "epoch": 3.66307826592959, + "flow_matching_loss": 0.013182393927127122, + "grad_norm": 0.0659603625535965, + "learning_rate": 2.012149654706343e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12590 + }, + { + "epoch": 3.6659877800407332, + "flow_matching_loss": 0.014421148411929607, + "grad_norm": 0.07936687767505646, + "learning_rate": 2.010595237534041e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12600 + }, + { + "epoch": 3.6688972941518765, + "flow_matching_loss": 0.014028301741927863, + "grad_norm": 0.07632703334093094, + "learning_rate": 2.0090401999403793e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12610 + }, + { + "epoch": 3.67180680826302, + "flow_matching_loss": 0.0139091270044446, + "grad_norm": 0.07469461113214493, + "learning_rate": 2.007484543814875e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12620 + }, + { + "epoch": 3.6747163223741635, + "flow_matching_loss": 0.014007576555013657, + "grad_norm": 0.08954912424087524, + "learning_rate": 2.0059282710477972e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12630 + }, + { + "epoch": 3.6776258364853067, + "flow_matching_loss": 0.012581683043390513, + "grad_norm": 0.07171216607093811, + "learning_rate": 2.0043713835301647e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12640 + }, + { + "epoch": 3.6805353505964504, + "flow_matching_loss": 0.013361520506441594, + "grad_norm": 0.08367510139942169, + "learning_rate": 2.0028138831537417e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12650 + }, + { + "epoch": 3.6834448647075937, + "flow_matching_loss": 0.012909241765737534, + "grad_norm": 0.08248331397771835, + "learning_rate": 2.0012557718110387e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12660 + }, + { + "epoch": 3.6863543788187374, + "flow_matching_loss": 0.013311466574668885, + "grad_norm": 0.07762980461120605, + "learning_rate": 1.9996970513953077e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12670 + }, + { + "epoch": 3.6892638929298807, + "flow_matching_loss": 0.013588461931794882, + "grad_norm": 0.0857536718249321, + "learning_rate": 1.9981377238005405e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12680 + }, + { + "epoch": 3.6921734070410244, + "flow_matching_loss": 0.013012713100761176, + "grad_norm": 0.11582468450069427, + "learning_rate": 1.9965777909214674e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12690 + }, + { + "epoch": 3.6950829211521676, + "flow_matching_loss": 0.012130504194647073, + "grad_norm": 0.08447328209877014, + "learning_rate": 1.9950172546535545e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12700 + }, + { + "epoch": 3.697992435263311, + "flow_matching_loss": 0.014041427802294493, + "grad_norm": 0.09219472855329514, + "learning_rate": 1.9934561168930002e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12710 + }, + { + "epoch": 3.7009019493744546, + "flow_matching_loss": 0.012833046913146972, + "grad_norm": 0.0838807001709938, + "learning_rate": 1.991894379536734e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12720 + }, + { + "epoch": 3.703811463485598, + "flow_matching_loss": 0.01341736325994134, + "grad_norm": 0.12184175103902817, + "learning_rate": 1.9903320444824137e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12730 + }, + { + "epoch": 3.706720977596741, + "flow_matching_loss": 0.014325376972556115, + "grad_norm": 0.09022035449743271, + "learning_rate": 1.9887691136284245e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12740 + }, + { + "epoch": 3.709630491707885, + "flow_matching_loss": 0.014908552914857865, + "grad_norm": 0.08230815082788467, + "learning_rate": 1.9872055888738745e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12750 + }, + { + "epoch": 3.712540005819028, + "flow_matching_loss": 0.013171110954135656, + "grad_norm": 0.08059772104024887, + "learning_rate": 1.985641472118593e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12760 + }, + { + "epoch": 3.715449519930172, + "flow_matching_loss": 0.014002270996570587, + "grad_norm": 0.06471262127161026, + "learning_rate": 1.9840767652631294e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12770 + }, + { + "epoch": 3.718359034041315, + "flow_matching_loss": 0.014380245935171843, + "grad_norm": 0.06665792316198349, + "learning_rate": 1.9825114702087513e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12780 + }, + { + "epoch": 3.7212685481524588, + "flow_matching_loss": 0.013130496488884091, + "grad_norm": 0.07262258231639862, + "learning_rate": 1.9809455888574388e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12790 + }, + { + "epoch": 3.724178062263602, + "flow_matching_loss": 0.015459670778363943, + "grad_norm": 0.0958927571773529, + "learning_rate": 1.9793791231118864e-05, + "loss": 0.0155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12800 + }, + { + "epoch": 3.7270875763747453, + "flow_matching_loss": 0.013671725243330001, + "grad_norm": 0.07262185961008072, + "learning_rate": 1.9778120748754958e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12810 + }, + { + "epoch": 3.729997090485889, + "flow_matching_loss": 0.013093068730086088, + "grad_norm": 0.09027400612831116, + "learning_rate": 1.97624444605238e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12820 + }, + { + "epoch": 3.7329066045970323, + "flow_matching_loss": 0.012876763567328453, + "grad_norm": 0.09797899425029755, + "learning_rate": 1.9746762385473552e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12830 + }, + { + "epoch": 3.7358161187081755, + "flow_matching_loss": 0.013698756136000157, + "grad_norm": 0.08110892027616501, + "learning_rate": 1.9731074542659414e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12840 + }, + { + "epoch": 3.7387256328193192, + "flow_matching_loss": 0.014641696214675903, + "grad_norm": 0.08983185887336731, + "learning_rate": 1.9715380951143603e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12850 + }, + { + "epoch": 3.7416351469304625, + "flow_matching_loss": 0.013514551799744368, + "grad_norm": 0.09181129932403564, + "learning_rate": 1.9699681629995295e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12860 + }, + { + "epoch": 3.744544661041606, + "flow_matching_loss": 0.012733886111527681, + "grad_norm": 0.0729174092411995, + "learning_rate": 1.968397659829066e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12870 + }, + { + "epoch": 3.7474541751527495, + "flow_matching_loss": 0.012498815637081861, + "grad_norm": 0.05964187532663345, + "learning_rate": 1.966826587511279e-05, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12880 + }, + { + "epoch": 3.750363689263893, + "flow_matching_loss": 0.013329425919800996, + "grad_norm": 0.0848255604505539, + "learning_rate": 1.965254947955169e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12890 + }, + { + "epoch": 3.7532732033750364, + "flow_matching_loss": 0.014826099667698146, + "grad_norm": 0.07249825447797775, + "learning_rate": 1.963682743070427e-05, + "loss": 0.0148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12900 + }, + { + "epoch": 3.7561827174861797, + "flow_matching_loss": 0.012528755841776729, + "grad_norm": 0.08432704955339432, + "learning_rate": 1.962109974767429e-05, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12910 + }, + { + "epoch": 3.7590922315973234, + "flow_matching_loss": 0.013220042642205953, + "grad_norm": 0.0864679291844368, + "learning_rate": 1.960536644957238e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12920 + }, + { + "epoch": 3.7620017457084667, + "flow_matching_loss": 0.013387324428185821, + "grad_norm": 0.0785110667347908, + "learning_rate": 1.9589627555515974e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12930 + }, + { + "epoch": 3.76491125981961, + "flow_matching_loss": 0.014117594808340073, + "grad_norm": 0.06989198923110962, + "learning_rate": 1.9573883084629317e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12940 + }, + { + "epoch": 3.7678207739307537, + "flow_matching_loss": 0.013768939301371574, + "grad_norm": 0.06176136061549187, + "learning_rate": 1.9558133056043422e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12950 + }, + { + "epoch": 3.770730288041897, + "flow_matching_loss": 0.01223098235204816, + "grad_norm": 0.067729651927948, + "learning_rate": 1.9542377488896062e-05, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12960 + }, + { + "epoch": 3.77363980215304, + "flow_matching_loss": 0.013233457878232002, + "grad_norm": 0.0552489347755909, + "learning_rate": 1.9526616402331733e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12970 + }, + { + "epoch": 3.776549316264184, + "flow_matching_loss": 0.013202749751508236, + "grad_norm": 0.09192093461751938, + "learning_rate": 1.9510849815501638e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12980 + }, + { + "epoch": 3.7794588303753276, + "flow_matching_loss": 0.013159954734146596, + "grad_norm": 0.08059640973806381, + "learning_rate": 1.949507774756368e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12990 + }, + { + "epoch": 3.782368344486471, + "flow_matching_loss": 0.013973183371126652, + "grad_norm": 0.07471483200788498, + "learning_rate": 1.9479300217682394e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13000 + }, + { + "epoch": 3.785277858597614, + "flow_matching_loss": 0.013536117970943451, + "grad_norm": 0.08003319054841995, + "learning_rate": 1.946351724502898e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13010 + }, + { + "epoch": 3.788187372708758, + "flow_matching_loss": 0.013417511153966189, + "grad_norm": 0.07406677305698395, + "learning_rate": 1.9447728848781227e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13020 + }, + { + "epoch": 3.791096886819901, + "flow_matching_loss": 0.013119605462998151, + "grad_norm": 0.06957615166902542, + "learning_rate": 1.943193504812353e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13030 + }, + { + "epoch": 3.7940064009310444, + "flow_matching_loss": 0.012736596260219812, + "grad_norm": 0.07025441527366638, + "learning_rate": 1.9416135862246846e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13040 + }, + { + "epoch": 3.796915915042188, + "flow_matching_loss": 0.012157263141125441, + "grad_norm": 0.08600366860628128, + "learning_rate": 1.9400331310348673e-05, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13050 + }, + { + "epoch": 3.7998254291533313, + "flow_matching_loss": 0.014212573040276765, + "grad_norm": 0.09650292992591858, + "learning_rate": 1.9384521411633033e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13060 + }, + { + "epoch": 3.8027349432644746, + "flow_matching_loss": 0.013793570641428231, + "grad_norm": 0.09634295850992203, + "learning_rate": 1.936870618531045e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13070 + }, + { + "epoch": 3.8056444573756183, + "flow_matching_loss": 0.01352057820186019, + "grad_norm": 0.08541912585496902, + "learning_rate": 1.9352885650597904e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13080 + }, + { + "epoch": 3.8085539714867616, + "flow_matching_loss": 0.013310058135539293, + "grad_norm": 0.14029870927333832, + "learning_rate": 1.9337059826718847e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13090 + }, + { + "epoch": 3.8114634855979053, + "flow_matching_loss": 0.011908973380923272, + "grad_norm": 0.09734877198934555, + "learning_rate": 1.932122873290314e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13100 + }, + { + "epoch": 3.8143729997090485, + "flow_matching_loss": 0.0134675950743258, + "grad_norm": 0.0709296241402626, + "learning_rate": 1.9305392388387054e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13110 + }, + { + "epoch": 3.8172825138201922, + "flow_matching_loss": 0.014204933866858482, + "grad_norm": 0.07292674481868744, + "learning_rate": 1.928955081241325e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13120 + }, + { + "epoch": 3.8201920279313355, + "flow_matching_loss": 0.013463352434337139, + "grad_norm": 0.07284799963235855, + "learning_rate": 1.9273704024230725e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13130 + }, + { + "epoch": 3.8231015420424788, + "flow_matching_loss": 0.013654156681150198, + "grad_norm": 0.09299653768539429, + "learning_rate": 1.925785204309483e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13140 + }, + { + "epoch": 3.8260110561536225, + "flow_matching_loss": 0.012812197860330343, + "grad_norm": 0.06668690592050552, + "learning_rate": 1.9241994888267212e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13150 + }, + { + "epoch": 3.8289205702647657, + "flow_matching_loss": 0.014092878624796867, + "grad_norm": 0.09057392179965973, + "learning_rate": 1.9226132579015815e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13160 + }, + { + "epoch": 3.831830084375909, + "flow_matching_loss": 0.015829661395400763, + "grad_norm": 0.09452877938747406, + "learning_rate": 1.9210265134614832e-05, + "loss": 0.0158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13170 + }, + { + "epoch": 3.8347395984870527, + "flow_matching_loss": 0.013601618353277445, + "grad_norm": 0.10009738057851791, + "learning_rate": 1.9194392574344712e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13180 + }, + { + "epoch": 3.837649112598196, + "flow_matching_loss": 0.01286322670057416, + "grad_norm": 0.07003331929445267, + "learning_rate": 1.9178514917492103e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13190 + }, + { + "epoch": 3.8405586267093397, + "flow_matching_loss": 0.015030455123633147, + "grad_norm": 0.06343729048967361, + "learning_rate": 1.9162632183349866e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13200 + }, + { + "epoch": 3.843468140820483, + "flow_matching_loss": 0.013366334512829781, + "grad_norm": 0.08005347102880478, + "learning_rate": 1.914674439121701e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13210 + }, + { + "epoch": 3.8463776549316266, + "flow_matching_loss": 0.013528196699917316, + "grad_norm": 0.07812158763408661, + "learning_rate": 1.9130851560398715e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13220 + }, + { + "epoch": 3.84928716904277, + "flow_matching_loss": 0.012141352333128453, + "grad_norm": 0.06918546557426453, + "learning_rate": 1.911495371020625e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13230 + }, + { + "epoch": 3.852196683153913, + "flow_matching_loss": 0.013035605940967799, + "grad_norm": 0.08588917553424835, + "learning_rate": 1.909905085995702e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13240 + }, + { + "epoch": 3.855106197265057, + "flow_matching_loss": 0.013643583562225104, + "grad_norm": 0.07156587392091751, + "learning_rate": 1.908314302897448e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13250 + }, + { + "epoch": 3.8580157113762, + "flow_matching_loss": 0.014088161010295152, + "grad_norm": 0.08464937657117844, + "learning_rate": 1.9067230236588142e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13260 + }, + { + "epoch": 3.8609252254873434, + "flow_matching_loss": 0.012484131660312415, + "grad_norm": 0.09070228785276413, + "learning_rate": 1.905131250213355e-05, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13270 + }, + { + "epoch": 3.863834739598487, + "flow_matching_loss": 0.013097086129710079, + "grad_norm": 0.10769976675510406, + "learning_rate": 1.9035389844952256e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13280 + }, + { + "epoch": 3.8667442537096304, + "flow_matching_loss": 0.015893912594765424, + "grad_norm": 0.09047401696443558, + "learning_rate": 1.901946228439179e-05, + "loss": 0.0159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13290 + }, + { + "epoch": 3.869653767820774, + "flow_matching_loss": 0.013930309843271971, + "grad_norm": 0.09334196150302887, + "learning_rate": 1.9003529839805637e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13300 + }, + { + "epoch": 3.8725632819319173, + "flow_matching_loss": 0.01219948180951178, + "grad_norm": 0.059327948838472366, + "learning_rate": 1.8987592530553217e-05, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13310 + }, + { + "epoch": 3.875472796043061, + "flow_matching_loss": 0.015636242274194957, + "grad_norm": 0.152840256690979, + "learning_rate": 1.897165037599987e-05, + "loss": 0.0156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13320 + }, + { + "epoch": 3.8783823101542043, + "flow_matching_loss": 0.012895544059574604, + "grad_norm": 0.0694490596652031, + "learning_rate": 1.8955703395516814e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13330 + }, + { + "epoch": 3.8812918242653476, + "flow_matching_loss": 0.014068061392754317, + "grad_norm": 0.07667513936758041, + "learning_rate": 1.8939751608481134e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13340 + }, + { + "epoch": 3.8842013383764913, + "flow_matching_loss": 0.01327336048707366, + "grad_norm": 0.06139807403087616, + "learning_rate": 1.892379503427575e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13350 + }, + { + "epoch": 3.8871108524876345, + "flow_matching_loss": 0.013360910769551992, + "grad_norm": 0.07801999151706696, + "learning_rate": 1.8907833692289412e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13360 + }, + { + "epoch": 3.890020366598778, + "flow_matching_loss": 0.013383571151643991, + "grad_norm": 0.06326794624328613, + "learning_rate": 1.889186760191666e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13370 + }, + { + "epoch": 3.8929298807099215, + "flow_matching_loss": 0.013862076308578252, + "grad_norm": 0.11652218550443649, + "learning_rate": 1.887589678255779e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13380 + }, + { + "epoch": 3.895839394821065, + "flow_matching_loss": 0.013266316521912814, + "grad_norm": 0.14528010785579681, + "learning_rate": 1.8859921253618852e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13390 + }, + { + "epoch": 3.898748908932208, + "flow_matching_loss": 0.014595527667552233, + "grad_norm": 0.08563335984945297, + "learning_rate": 1.884394103451163e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13400 + }, + { + "epoch": 3.9016584230433518, + "flow_matching_loss": 0.012718370743095874, + "grad_norm": 0.06606749445199966, + "learning_rate": 1.8827956144653583e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13410 + }, + { + "epoch": 3.9045679371544955, + "flow_matching_loss": 0.012819773610681295, + "grad_norm": 0.09856653958559036, + "learning_rate": 1.8811966603467874e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13420 + }, + { + "epoch": 3.9074774512656387, + "flow_matching_loss": 0.01451029134914279, + "grad_norm": 0.1267450898885727, + "learning_rate": 1.8795972430383293e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13430 + }, + { + "epoch": 3.910386965376782, + "flow_matching_loss": 0.012920024245977402, + "grad_norm": 0.0911572203040123, + "learning_rate": 1.8779973644834273e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13440 + }, + { + "epoch": 3.9132964794879257, + "flow_matching_loss": 0.011956243310123682, + "grad_norm": 0.07271770387887955, + "learning_rate": 1.876397026626085e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13450 + }, + { + "epoch": 3.916205993599069, + "flow_matching_loss": 0.013166602328419685, + "grad_norm": 0.0672391802072525, + "learning_rate": 1.8747962314108633e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13460 + }, + { + "epoch": 3.919115507710212, + "flow_matching_loss": 0.013288868498057127, + "grad_norm": 0.09722218662500381, + "learning_rate": 1.873194980782879e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13470 + }, + { + "epoch": 3.922025021821356, + "flow_matching_loss": 0.01238437332212925, + "grad_norm": 0.07356017082929611, + "learning_rate": 1.871593276687804e-05, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13480 + }, + { + "epoch": 3.924934535932499, + "flow_matching_loss": 0.012321622762829065, + "grad_norm": 0.08978978544473648, + "learning_rate": 1.869991121071858e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13490 + }, + { + "epoch": 3.9278440500436425, + "flow_matching_loss": 0.014082365296781063, + "grad_norm": 0.08041387796401978, + "learning_rate": 1.8683885158818117e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13500 + }, + { + "epoch": 3.930753564154786, + "flow_matching_loss": 0.012040334288030863, + "grad_norm": 0.09647246450185776, + "learning_rate": 1.8667854630649825e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13510 + }, + { + "epoch": 3.9336630782659294, + "flow_matching_loss": 0.014091809745877981, + "grad_norm": 0.07276143878698349, + "learning_rate": 1.86518196456923e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13520 + }, + { + "epoch": 3.936572592377073, + "flow_matching_loss": 0.013915763329714536, + "grad_norm": 0.09075218439102173, + "learning_rate": 1.8635780223429565e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13530 + }, + { + "epoch": 3.9394821064882164, + "flow_matching_loss": 0.014537962805479764, + "grad_norm": 0.07167861610651016, + "learning_rate": 1.8619736383351017e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13540 + }, + { + "epoch": 3.94239162059936, + "flow_matching_loss": 0.012598059512674809, + "grad_norm": 0.05977954715490341, + "learning_rate": 1.8603688144951454e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13550 + }, + { + "epoch": 3.9453011347105034, + "flow_matching_loss": 0.013491497095674276, + "grad_norm": 0.11465038359165192, + "learning_rate": 1.8587635527730983e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13560 + }, + { + "epoch": 3.9482106488216466, + "flow_matching_loss": 0.012980656139552594, + "grad_norm": 0.08565167337656021, + "learning_rate": 1.8571578551195056e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13570 + }, + { + "epoch": 3.9511201629327903, + "flow_matching_loss": 0.011298711039125919, + "grad_norm": 0.07304073125123978, + "learning_rate": 1.855551723485441e-05, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13580 + }, + { + "epoch": 3.9540296770439336, + "flow_matching_loss": 0.012767414096742868, + "grad_norm": 0.06416001170873642, + "learning_rate": 1.8539451598225066e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13590 + }, + { + "epoch": 3.956939191155077, + "flow_matching_loss": 0.013489914126694202, + "grad_norm": 0.06624443084001541, + "learning_rate": 1.8523381660828276e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13600 + }, + { + "epoch": 3.9598487052662206, + "flow_matching_loss": 0.013690053950995207, + "grad_norm": 0.08360245823860168, + "learning_rate": 1.850730744219054e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13610 + }, + { + "epoch": 3.962758219377364, + "flow_matching_loss": 0.012434764020144939, + "grad_norm": 0.10242841392755508, + "learning_rate": 1.849122896184354e-05, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13620 + }, + { + "epoch": 3.9656677334885075, + "flow_matching_loss": 0.014299271814525127, + "grad_norm": 0.09506838023662567, + "learning_rate": 1.8475146239324146e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13630 + }, + { + "epoch": 3.968577247599651, + "flow_matching_loss": 0.011957266088575124, + "grad_norm": 0.07601627707481384, + "learning_rate": 1.8459059294174396e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13640 + }, + { + "epoch": 3.9714867617107945, + "flow_matching_loss": 0.013555850740522147, + "grad_norm": 0.06865059584379196, + "learning_rate": 1.844296814594143e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13650 + }, + { + "epoch": 3.9743962758219378, + "flow_matching_loss": 0.012974749319255352, + "grad_norm": 0.09249426424503326, + "learning_rate": 1.8426872814177516e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13660 + }, + { + "epoch": 3.977305789933081, + "flow_matching_loss": 0.01309784734621644, + "grad_norm": 0.09914877265691757, + "learning_rate": 1.8410773318440003e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13670 + }, + { + "epoch": 3.9802153040442247, + "flow_matching_loss": 0.012266931124031544, + "grad_norm": 0.09227374941110611, + "learning_rate": 1.839466967829129e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13680 + }, + { + "epoch": 3.983124818155368, + "flow_matching_loss": 0.012975198216736317, + "grad_norm": 0.08162607252597809, + "learning_rate": 1.8378561913298822e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13690 + }, + { + "epoch": 3.9860343322665113, + "flow_matching_loss": 0.012021734658628703, + "grad_norm": 0.07775233685970306, + "learning_rate": 1.8362450043035053e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13700 + }, + { + "epoch": 3.988943846377655, + "flow_matching_loss": 0.01341935396194458, + "grad_norm": 0.09304619580507278, + "learning_rate": 1.834633408707742e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13710 + }, + { + "epoch": 3.9918533604887982, + "flow_matching_loss": 0.01402414245530963, + "grad_norm": 0.09358681738376617, + "learning_rate": 1.8330214065008336e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13720 + }, + { + "epoch": 3.994762874599942, + "flow_matching_loss": 0.013797796610742808, + "grad_norm": 0.08154137432575226, + "learning_rate": 1.831408999641514e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13730 + }, + { + "epoch": 3.997672388711085, + "flow_matching_loss": 0.013244094699621201, + "grad_norm": 0.09857168048620224, + "learning_rate": 1.8297961900890107e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13740 + }, + { + "epoch": 4.000581902822229, + "flow_matching_loss": 0.012185734650120138, + "grad_norm": 0.08140819519758224, + "learning_rate": 1.8281829798030385e-05, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13750 + }, + { + "epoch": 4.003491416933372, + "flow_matching_loss": 0.013475547824054957, + "grad_norm": 0.0897757038474083, + "learning_rate": 1.8265693707438002e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13760 + }, + { + "epoch": 4.0064009310445154, + "flow_matching_loss": 0.012993837427347898, + "grad_norm": 0.08950677514076233, + "learning_rate": 1.824955364871982e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13770 + }, + { + "epoch": 4.009310445155659, + "flow_matching_loss": 0.013397919479757547, + "grad_norm": 0.09171722084283829, + "learning_rate": 1.8233409641487553e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13780 + }, + { + "epoch": 4.012219959266803, + "flow_matching_loss": 0.013768240716308356, + "grad_norm": 0.061938874423503876, + "learning_rate": 1.8217261705357677e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13790 + }, + { + "epoch": 4.015129473377946, + "flow_matching_loss": 0.013141484186053275, + "grad_norm": 0.08389876782894135, + "learning_rate": 1.8201109859951462e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13800 + }, + { + "epoch": 4.018038987489089, + "flow_matching_loss": 0.01170337563380599, + "grad_norm": 0.08223447948694229, + "learning_rate": 1.818495412489492e-05, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13810 + }, + { + "epoch": 4.020948501600233, + "flow_matching_loss": 0.01420569820329547, + "grad_norm": 0.07102543115615845, + "learning_rate": 1.8168794519818802e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13820 + }, + { + "epoch": 4.023858015711376, + "flow_matching_loss": 0.013760650530457497, + "grad_norm": 0.08861371129751205, + "learning_rate": 1.8152631064358537e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13830 + }, + { + "epoch": 4.02676752982252, + "flow_matching_loss": 0.013236703164875508, + "grad_norm": 0.09065357595682144, + "learning_rate": 1.8136463778154258e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13840 + }, + { + "epoch": 4.029677043933663, + "flow_matching_loss": 0.013624904677271843, + "grad_norm": 0.09067126363515854, + "learning_rate": 1.812029268085074e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13850 + }, + { + "epoch": 4.032586558044806, + "flow_matching_loss": 0.014408345986157655, + "grad_norm": 0.0720585286617279, + "learning_rate": 1.8104117792097387e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13860 + }, + { + "epoch": 4.03549607215595, + "flow_matching_loss": 0.014201872888952494, + "grad_norm": 0.09290245175361633, + "learning_rate": 1.808793913154822e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13870 + }, + { + "epoch": 4.038405586267094, + "flow_matching_loss": 0.013689570687711239, + "grad_norm": 0.06894000619649887, + "learning_rate": 1.8071756718861838e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13880 + }, + { + "epoch": 4.041315100378237, + "flow_matching_loss": 0.013159841392189264, + "grad_norm": 0.08071831613779068, + "learning_rate": 1.805557057370139e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13890 + }, + { + "epoch": 4.04422461448938, + "flow_matching_loss": 0.013944162521511316, + "grad_norm": 0.06824926286935806, + "learning_rate": 1.8039380715734574e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13900 + }, + { + "epoch": 4.047134128600524, + "flow_matching_loss": 0.013296370301395655, + "grad_norm": 0.09991374611854553, + "learning_rate": 1.802318716463359e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13910 + }, + { + "epoch": 4.0500436427116675, + "flow_matching_loss": 0.013330803997814656, + "grad_norm": 0.07369635999202728, + "learning_rate": 1.8006989940075138e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13920 + }, + { + "epoch": 4.05295315682281, + "flow_matching_loss": 0.012316454388201236, + "grad_norm": 0.059281397610902786, + "learning_rate": 1.7990789061740368e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13930 + }, + { + "epoch": 4.055862670933954, + "flow_matching_loss": 0.014529878366738557, + "grad_norm": 0.08116316795349121, + "learning_rate": 1.7974584549314875e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13940 + }, + { + "epoch": 4.058772185045098, + "flow_matching_loss": 0.014193316362798214, + "grad_norm": 0.07655535638332367, + "learning_rate": 1.7958376422488672e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13950 + }, + { + "epoch": 4.0616816991562406, + "flow_matching_loss": 0.014563153311610223, + "grad_norm": 0.05924784392118454, + "learning_rate": 1.7942164700956156e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13960 + }, + { + "epoch": 4.064591213267384, + "flow_matching_loss": 0.012030802201479673, + "grad_norm": 0.06633547693490982, + "learning_rate": 1.79259494044161e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13970 + }, + { + "epoch": 4.067500727378528, + "flow_matching_loss": 0.012955999281257392, + "grad_norm": 0.07310877740383148, + "learning_rate": 1.7909730552571617e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13980 + }, + { + "epoch": 4.070410241489671, + "flow_matching_loss": 0.013943110499531031, + "grad_norm": 0.0931544378399849, + "learning_rate": 1.7893508165130143e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13990 + }, + { + "epoch": 4.0733197556008145, + "flow_matching_loss": 0.01379026509821415, + "grad_norm": 0.06892651319503784, + "learning_rate": 1.787728226180341e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14000 + }, + { + "epoch": 4.076229269711958, + "flow_matching_loss": 0.01308284243568778, + "grad_norm": 0.06965580582618713, + "learning_rate": 1.786105286230742e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14010 + }, + { + "epoch": 4.079138783823102, + "flow_matching_loss": 0.013307677954435349, + "grad_norm": 0.07565286010503769, + "learning_rate": 1.784481998636242e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14020 + }, + { + "epoch": 4.082048297934245, + "flow_matching_loss": 0.014854741469025611, + "grad_norm": 0.08101504296064377, + "learning_rate": 1.782858365369289e-05, + "loss": 0.0149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14030 + }, + { + "epoch": 4.084957812045388, + "flow_matching_loss": 0.01314920112490654, + "grad_norm": 0.08974573761224747, + "learning_rate": 1.78123438840275e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14040 + }, + { + "epoch": 4.087867326156532, + "flow_matching_loss": 0.013855258747935296, + "grad_norm": 0.09265938401222229, + "learning_rate": 1.779610069709911e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14050 + }, + { + "epoch": 4.090776840267675, + "flow_matching_loss": 0.014261961914598942, + "grad_norm": 0.07948890328407288, + "learning_rate": 1.7779854112644713e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14060 + }, + { + "epoch": 4.093686354378819, + "flow_matching_loss": 0.013796891551464796, + "grad_norm": 0.06853088736534119, + "learning_rate": 1.7763604150405446e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14070 + }, + { + "epoch": 4.096595868489962, + "flow_matching_loss": 0.013727440126240253, + "grad_norm": 0.07795567810535431, + "learning_rate": 1.7747350830126548e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14080 + }, + { + "epoch": 4.099505382601105, + "flow_matching_loss": 0.013592680357396603, + "grad_norm": 0.0729692354798317, + "learning_rate": 1.7731094171557332e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14090 + }, + { + "epoch": 4.102414896712249, + "flow_matching_loss": 0.012596772238612174, + "grad_norm": 0.06830330938100815, + "learning_rate": 1.771483419445117e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14100 + }, + { + "epoch": 4.105324410823393, + "flow_matching_loss": 0.011137231532484294, + "grad_norm": 0.0615106001496315, + "learning_rate": 1.7698570918565474e-05, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14110 + }, + { + "epoch": 4.108233924934536, + "flow_matching_loss": 0.011404656805098057, + "grad_norm": 0.07868806272745132, + "learning_rate": 1.768230436366165e-05, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14120 + }, + { + "epoch": 4.111143439045679, + "flow_matching_loss": 0.014362907595932483, + "grad_norm": 0.06674683839082718, + "learning_rate": 1.7666034549505102e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14130 + }, + { + "epoch": 4.114052953156823, + "flow_matching_loss": 0.014127088058739901, + "grad_norm": 0.073395736515522, + "learning_rate": 1.7649761495865185e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14140 + }, + { + "epoch": 4.1169624672679666, + "flow_matching_loss": 0.012772728689014912, + "grad_norm": 0.06815199553966522, + "learning_rate": 1.7633485222515188e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14150 + }, + { + "epoch": 4.119871981379109, + "flow_matching_loss": 0.013888988457620145, + "grad_norm": 0.07039857655763626, + "learning_rate": 1.761720574923233e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14160 + }, + { + "epoch": 4.122781495490253, + "flow_matching_loss": 0.013739665038883686, + "grad_norm": 0.07584759593009949, + "learning_rate": 1.7600923095797704e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14170 + }, + { + "epoch": 4.125691009601397, + "flow_matching_loss": 0.012844882253557444, + "grad_norm": 0.10435619950294495, + "learning_rate": 1.758463728199626e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14180 + }, + { + "epoch": 4.12860052371254, + "flow_matching_loss": 0.012083445116877556, + "grad_norm": 0.11643065512180328, + "learning_rate": 1.7568348327616803e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14190 + }, + { + "epoch": 4.131510037823683, + "flow_matching_loss": 0.01314323078840971, + "grad_norm": 0.0657379999756813, + "learning_rate": 1.755205625245195e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14200 + }, + { + "epoch": 4.134419551934827, + "flow_matching_loss": 0.012117819674313069, + "grad_norm": 0.0692550465464592, + "learning_rate": 1.7535761076298104e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14210 + }, + { + "epoch": 4.137329066045971, + "flow_matching_loss": 0.011587505135685206, + "grad_norm": 0.07152283191680908, + "learning_rate": 1.7519462818955448e-05, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14220 + }, + { + "epoch": 4.1402385801571135, + "flow_matching_loss": 0.012818929459899663, + "grad_norm": 0.0650169849395752, + "learning_rate": 1.7503161500227896e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14230 + }, + { + "epoch": 4.143148094268257, + "flow_matching_loss": 0.01259844433516264, + "grad_norm": 0.07668087631464005, + "learning_rate": 1.7486857139923093e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14240 + }, + { + "epoch": 4.146057608379401, + "flow_matching_loss": 0.012851606030017138, + "grad_norm": 0.07089436054229736, + "learning_rate": 1.747054975785236e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14250 + }, + { + "epoch": 4.148967122490544, + "flow_matching_loss": 0.013998648058623076, + "grad_norm": 0.0781492069363594, + "learning_rate": 1.745423937383072e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14260 + }, + { + "epoch": 4.1518766366016875, + "flow_matching_loss": 0.011480747628957034, + "grad_norm": 0.07102370262145996, + "learning_rate": 1.7437926007676825e-05, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14270 + }, + { + "epoch": 4.154786150712831, + "flow_matching_loss": 0.012715253792703152, + "grad_norm": 0.08081667125225067, + "learning_rate": 1.742160967921294e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14280 + }, + { + "epoch": 4.157695664823974, + "flow_matching_loss": 0.014011422265321016, + "grad_norm": 0.1137520968914032, + "learning_rate": 1.7405290408264954e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14290 + }, + { + "epoch": 4.160605178935118, + "flow_matching_loss": 0.0128421182744205, + "grad_norm": 0.06502557545900345, + "learning_rate": 1.738896821466232e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14300 + }, + { + "epoch": 4.163514693046261, + "flow_matching_loss": 0.01202319860458374, + "grad_norm": 0.06452043354511261, + "learning_rate": 1.737264311823804e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14310 + }, + { + "epoch": 4.166424207157405, + "flow_matching_loss": 0.012616479583084584, + "grad_norm": 0.08131839334964752, + "learning_rate": 1.7356315138828645e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14320 + }, + { + "epoch": 4.169333721268548, + "flow_matching_loss": 0.01378798931837082, + "grad_norm": 0.07215756922960281, + "learning_rate": 1.7339984296274174e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14330 + }, + { + "epoch": 4.172243235379692, + "flow_matching_loss": 0.013161023333668708, + "grad_norm": 0.0858314260840416, + "learning_rate": 1.7323650610418134e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14340 + }, + { + "epoch": 4.175152749490835, + "flow_matching_loss": 0.012835447024554015, + "grad_norm": 0.09707736223936081, + "learning_rate": 1.730731410110749e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14350 + }, + { + "epoch": 4.178062263601978, + "flow_matching_loss": 0.012850960157811642, + "grad_norm": 0.08490154892206192, + "learning_rate": 1.7290974788192655e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14360 + }, + { + "epoch": 4.180971777713122, + "flow_matching_loss": 0.013499628473073243, + "grad_norm": 0.07044310122728348, + "learning_rate": 1.727463269152743e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14370 + }, + { + "epoch": 4.183881291824266, + "flow_matching_loss": 0.013662632927298546, + "grad_norm": 0.07536959648132324, + "learning_rate": 1.7258287830968996e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14380 + }, + { + "epoch": 4.186790805935408, + "flow_matching_loss": 0.013354901596903801, + "grad_norm": 0.10865319520235062, + "learning_rate": 1.7241940226377908e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14390 + }, + { + "epoch": 4.189700320046552, + "flow_matching_loss": 0.012736053485423327, + "grad_norm": 0.0861864760518074, + "learning_rate": 1.7225589897618043e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14400 + }, + { + "epoch": 4.192609834157696, + "flow_matching_loss": 0.013450188282877206, + "grad_norm": 0.12796209752559662, + "learning_rate": 1.7209236864556588e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14410 + }, + { + "epoch": 4.195519348268839, + "flow_matching_loss": 0.012149886041879655, + "grad_norm": 0.07372619211673737, + "learning_rate": 1.719288114706403e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14420 + }, + { + "epoch": 4.198428862379982, + "flow_matching_loss": 0.014104901254177094, + "grad_norm": 0.0877663642168045, + "learning_rate": 1.7176522765014104e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14430 + }, + { + "epoch": 4.201338376491126, + "flow_matching_loss": 0.01443738592788577, + "grad_norm": 0.07436210662126541, + "learning_rate": 1.7160161738283782e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14440 + }, + { + "epoch": 4.20424789060227, + "flow_matching_loss": 0.013630623929202557, + "grad_norm": 0.0652228593826294, + "learning_rate": 1.714379808675326e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14450 + }, + { + "epoch": 4.207157404713413, + "flow_matching_loss": 0.01471006330102682, + "grad_norm": 0.0735245794057846, + "learning_rate": 1.712743183030592e-05, + "loss": 0.0147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14460 + }, + { + "epoch": 4.210066918824556, + "flow_matching_loss": 0.01293271379545331, + "grad_norm": 0.1030760183930397, + "learning_rate": 1.7111062988828294e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14470 + }, + { + "epoch": 4.2129764329357, + "flow_matching_loss": 0.013685056660324334, + "grad_norm": 0.05829928070306778, + "learning_rate": 1.709469158221008e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14480 + }, + { + "epoch": 4.215885947046843, + "flow_matching_loss": 0.01330594252794981, + "grad_norm": 0.06969162076711655, + "learning_rate": 1.7078317630344083e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14490 + }, + { + "epoch": 4.2187954611579865, + "flow_matching_loss": 0.013248216826468706, + "grad_norm": 0.10398519039154053, + "learning_rate": 1.706194115312618e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14500 + }, + { + "epoch": 4.22170497526913, + "flow_matching_loss": 0.011707460414618254, + "grad_norm": 0.06809897720813751, + "learning_rate": 1.704556217045536e-05, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14510 + }, + { + "epoch": 4.224614489380274, + "flow_matching_loss": 0.01328309327363968, + "grad_norm": 0.07800958305597305, + "learning_rate": 1.7029180702233612e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14520 + }, + { + "epoch": 4.227524003491417, + "flow_matching_loss": 0.013675974495708943, + "grad_norm": 0.1104070171713829, + "learning_rate": 1.7012796768365975e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14530 + }, + { + "epoch": 4.2304335176025605, + "flow_matching_loss": 0.012954907119274139, + "grad_norm": 0.07732236385345459, + "learning_rate": 1.6996410388760465e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14540 + }, + { + "epoch": 4.233343031713704, + "flow_matching_loss": 0.012878537829965352, + "grad_norm": 0.10249853134155273, + "learning_rate": 1.698002158332809e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14550 + }, + { + "epoch": 4.236252545824847, + "flow_matching_loss": 0.013365665450692177, + "grad_norm": 0.07973042130470276, + "learning_rate": 1.6963630371982783e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14560 + }, + { + "epoch": 4.239162059935991, + "flow_matching_loss": 0.013593169301748276, + "grad_norm": 0.08239489793777466, + "learning_rate": 1.6947236774641416e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14570 + }, + { + "epoch": 4.242071574047134, + "flow_matching_loss": 0.013721154909580946, + "grad_norm": 0.07668357342481613, + "learning_rate": 1.6930840811223755e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14580 + }, + { + "epoch": 4.244981088158277, + "flow_matching_loss": 0.01378275929018855, + "grad_norm": 0.11770035326480865, + "learning_rate": 1.6914442501652448e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14590 + }, + { + "epoch": 4.247890602269421, + "flow_matching_loss": 0.012145843636244536, + "grad_norm": 0.07493524253368378, + "learning_rate": 1.6898041865852974e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14600 + }, + { + "epoch": 4.250800116380565, + "flow_matching_loss": 0.012324384786188603, + "grad_norm": 0.06796936690807343, + "learning_rate": 1.6881638923753663e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14610 + }, + { + "epoch": 4.2537096304917075, + "flow_matching_loss": 0.013082257099449634, + "grad_norm": 0.06792694330215454, + "learning_rate": 1.6865233695285632e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14620 + }, + { + "epoch": 4.256619144602851, + "flow_matching_loss": 0.012893185624852777, + "grad_norm": 0.07590830326080322, + "learning_rate": 1.6848826200382782e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14630 + }, + { + "epoch": 4.259528658713995, + "flow_matching_loss": 0.012582777813076974, + "grad_norm": 0.07559715956449509, + "learning_rate": 1.6832416458981762e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14640 + }, + { + "epoch": 4.262438172825139, + "flow_matching_loss": 0.012934844009578228, + "grad_norm": 0.07545854151248932, + "learning_rate": 1.681600449102196e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14650 + }, + { + "epoch": 4.265347686936281, + "flow_matching_loss": 0.01264470973983407, + "grad_norm": 0.06703189015388489, + "learning_rate": 1.6799590316445468e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14660 + }, + { + "epoch": 4.268257201047425, + "flow_matching_loss": 0.012006103061139584, + "grad_norm": 0.06985065340995789, + "learning_rate": 1.678317395519705e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14670 + }, + { + "epoch": 4.271166715158569, + "flow_matching_loss": 0.014234405010938644, + "grad_norm": 0.07231118530035019, + "learning_rate": 1.676675542722413e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14680 + }, + { + "epoch": 4.274076229269712, + "flow_matching_loss": 0.012120848149061203, + "grad_norm": 0.07499941438436508, + "learning_rate": 1.6750334752476776e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14690 + }, + { + "epoch": 4.276985743380855, + "flow_matching_loss": 0.012349982280284166, + "grad_norm": 0.08177611976861954, + "learning_rate": 1.673391195090765e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14700 + }, + { + "epoch": 4.279895257491999, + "flow_matching_loss": 0.012384305521845818, + "grad_norm": 0.08334437012672424, + "learning_rate": 1.6717487042472006e-05, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14710 + }, + { + "epoch": 4.282804771603142, + "flow_matching_loss": 0.014221071824431419, + "grad_norm": 0.08832553774118423, + "learning_rate": 1.6701060047127655e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14720 + }, + { + "epoch": 4.285714285714286, + "flow_matching_loss": 0.012264149729162455, + "grad_norm": 0.08744592219591141, + "learning_rate": 1.6684630984834947e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14730 + }, + { + "epoch": 4.288623799825429, + "flow_matching_loss": 0.013468694221228362, + "grad_norm": 0.06361344456672668, + "learning_rate": 1.666819987555675e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14740 + }, + { + "epoch": 4.291533313936573, + "flow_matching_loss": 0.014036297239363194, + "grad_norm": 0.08484221994876862, + "learning_rate": 1.6651766739258392e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14750 + }, + { + "epoch": 4.294442828047716, + "flow_matching_loss": 0.013716990128159523, + "grad_norm": 0.10232914239168167, + "learning_rate": 1.6635331595907704e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14760 + }, + { + "epoch": 4.2973523421588595, + "flow_matching_loss": 0.012240540143102408, + "grad_norm": 0.0819251537322998, + "learning_rate": 1.6618894465474923e-05, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14770 + }, + { + "epoch": 4.300261856270003, + "flow_matching_loss": 0.012960456311702728, + "grad_norm": 0.08708617836236954, + "learning_rate": 1.660245536793271e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14780 + }, + { + "epoch": 4.303171370381146, + "flow_matching_loss": 0.013042466901242732, + "grad_norm": 0.09111921489238739, + "learning_rate": 1.6586014323256126e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14790 + }, + { + "epoch": 4.30608088449229, + "flow_matching_loss": 0.012854738160967827, + "grad_norm": 0.06998693197965622, + "learning_rate": 1.656957135142259e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14800 + }, + { + "epoch": 4.3089903986034335, + "flow_matching_loss": 0.012149122264236211, + "grad_norm": 0.09512016922235489, + "learning_rate": 1.655312647241186e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14810 + }, + { + "epoch": 4.311899912714576, + "flow_matching_loss": 0.01461009131744504, + "grad_norm": 0.08755578100681305, + "learning_rate": 1.6536679706206015e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14820 + }, + { + "epoch": 4.31480942682572, + "flow_matching_loss": 0.012556344829499721, + "grad_norm": 0.05915209278464317, + "learning_rate": 1.6520231072789427e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14830 + }, + { + "epoch": 4.317718940936864, + "flow_matching_loss": 0.012240388989448547, + "grad_norm": 0.06916706264019012, + "learning_rate": 1.6503780592148737e-05, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14840 + }, + { + "epoch": 4.320628455048007, + "flow_matching_loss": 0.013448153715580701, + "grad_norm": 0.08357773721218109, + "learning_rate": 1.648732828427283e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14850 + }, + { + "epoch": 4.32353796915915, + "flow_matching_loss": 0.012153565604239702, + "grad_norm": 0.09181908518075943, + "learning_rate": 1.647087416915281e-05, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14860 + }, + { + "epoch": 4.326447483270294, + "flow_matching_loss": 0.012073616776615382, + "grad_norm": 0.06641252338886261, + "learning_rate": 1.6454418266781987e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14870 + }, + { + "epoch": 4.329356997381438, + "flow_matching_loss": 0.01335873333737254, + "grad_norm": 0.08126815408468246, + "learning_rate": 1.6437960597155824e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14880 + }, + { + "epoch": 4.3322665114925805, + "flow_matching_loss": 0.013181941211223602, + "grad_norm": 0.06776075065135956, + "learning_rate": 1.642150118027194e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14890 + }, + { + "epoch": 4.335176025603724, + "flow_matching_loss": 0.013488463871181012, + "grad_norm": 0.0780283510684967, + "learning_rate": 1.640504003613009e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14900 + }, + { + "epoch": 4.338085539714868, + "flow_matching_loss": 0.013083068653941154, + "grad_norm": 0.06862565875053406, + "learning_rate": 1.6388577184732107e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14910 + }, + { + "epoch": 4.340995053826011, + "flow_matching_loss": 0.012684880290180445, + "grad_norm": 0.07387526333332062, + "learning_rate": 1.6372112646081907e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14920 + }, + { + "epoch": 4.343904567937154, + "flow_matching_loss": 0.01353787574917078, + "grad_norm": 0.06698490679264069, + "learning_rate": 1.6355646440185464e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14930 + }, + { + "epoch": 4.346814082048298, + "flow_matching_loss": 0.013243423029780387, + "grad_norm": 0.09623456001281738, + "learning_rate": 1.6339178587050764e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14940 + }, + { + "epoch": 4.349723596159441, + "flow_matching_loss": 0.013103078491985798, + "grad_norm": 0.09819866716861725, + "learning_rate": 1.632270910668781e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14950 + }, + { + "epoch": 4.352633110270585, + "flow_matching_loss": 0.012865010183304548, + "grad_norm": 0.07459087669849396, + "learning_rate": 1.6306238019108562e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14960 + }, + { + "epoch": 4.355542624381728, + "flow_matching_loss": 0.014327965117990971, + "grad_norm": 0.09881847351789474, + "learning_rate": 1.6289765344326952e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14970 + }, + { + "epoch": 4.358452138492872, + "flow_matching_loss": 0.013816046994179487, + "grad_norm": 0.0650724247097969, + "learning_rate": 1.6273291102358833e-05, + "loss": 0.0138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14980 + }, + { + "epoch": 4.361361652604015, + "flow_matching_loss": 0.011996990907937289, + "grad_norm": 0.07786169648170471, + "learning_rate": 1.6256815313221963e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14990 + }, + { + "epoch": 4.364271166715159, + "flow_matching_loss": 0.011855941265821457, + "grad_norm": 0.07519581913948059, + "learning_rate": 1.6240337996935972e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15000 + }, + { + "epoch": 4.367180680826302, + "flow_matching_loss": 0.012758752331137657, + "grad_norm": 0.07350607961416245, + "learning_rate": 1.6223859173522367e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15010 + }, + { + "epoch": 4.370090194937445, + "flow_matching_loss": 0.0132432971149683, + "grad_norm": 0.06551969796419144, + "learning_rate": 1.6207378863004455e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15020 + }, + { + "epoch": 4.372999709048589, + "flow_matching_loss": 0.01201934814453125, + "grad_norm": 0.07329625636339188, + "learning_rate": 1.6190897085407382e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15030 + }, + { + "epoch": 4.3759092231597325, + "flow_matching_loss": 0.014398761000484228, + "grad_norm": 0.0787457674741745, + "learning_rate": 1.617441386075805e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15040 + }, + { + "epoch": 4.378818737270876, + "flow_matching_loss": 0.014037020783871413, + "grad_norm": 0.10531104356050491, + "learning_rate": 1.6157929209085142e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15050 + }, + { + "epoch": 4.381728251382019, + "flow_matching_loss": 0.012312994431704282, + "grad_norm": 0.06373909115791321, + "learning_rate": 1.614144315041906e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15060 + }, + { + "epoch": 4.384637765493163, + "flow_matching_loss": 0.015107828564941883, + "grad_norm": 0.08691934496164322, + "learning_rate": 1.612495570479192e-05, + "loss": 0.0151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15070 + }, + { + "epoch": 4.3875472796043065, + "flow_matching_loss": 0.014330282621085644, + "grad_norm": 0.06912688165903091, + "learning_rate": 1.6108466892237517e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15080 + }, + { + "epoch": 4.390456793715449, + "flow_matching_loss": 0.011961134895682336, + "grad_norm": 0.0845758244395256, + "learning_rate": 1.609197673279132e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15090 + }, + { + "epoch": 4.393366307826593, + "flow_matching_loss": 0.0121073835529387, + "grad_norm": 0.08701580762863159, + "learning_rate": 1.6075485246490428e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15100 + }, + { + "epoch": 4.396275821937737, + "flow_matching_loss": 0.012086448166519404, + "grad_norm": 0.07747369259595871, + "learning_rate": 1.605899245337355e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15110 + }, + { + "epoch": 4.3991853360488795, + "flow_matching_loss": 0.012871245387941599, + "grad_norm": 0.07847914844751358, + "learning_rate": 1.6042498373480983e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15120 + }, + { + "epoch": 4.402094850160023, + "flow_matching_loss": 0.012607232853770255, + "grad_norm": 0.08183800429105759, + "learning_rate": 1.602600302685459e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15130 + }, + { + "epoch": 4.405004364271167, + "flow_matching_loss": 0.01359317395836115, + "grad_norm": 0.09445607662200928, + "learning_rate": 1.6009506433537774e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15140 + }, + { + "epoch": 4.40791387838231, + "flow_matching_loss": 0.012262370344251395, + "grad_norm": 0.09681873023509979, + "learning_rate": 1.599300861357545e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15150 + }, + { + "epoch": 4.4108233924934535, + "flow_matching_loss": 0.012517038267105818, + "grad_norm": 0.07235462963581085, + "learning_rate": 1.5976509587014032e-05, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15160 + }, + { + "epoch": 4.413732906604597, + "flow_matching_loss": 0.012939321342855692, + "grad_norm": 0.049344003200531006, + "learning_rate": 1.5960009373901384e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15170 + }, + { + "epoch": 4.416642420715741, + "flow_matching_loss": 0.012780095357447863, + "grad_norm": 0.08251571655273438, + "learning_rate": 1.5943507994286823e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15180 + }, + { + "epoch": 4.419551934826884, + "flow_matching_loss": 0.012469058483839035, + "grad_norm": 0.08705936372280121, + "learning_rate": 1.5927005468221082e-05, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15190 + }, + { + "epoch": 4.422461448938027, + "flow_matching_loss": 0.01247761007398367, + "grad_norm": 0.08349733799695969, + "learning_rate": 1.5910501815756285e-05, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15200 + }, + { + "epoch": 4.425370963049171, + "flow_matching_loss": 0.015196060761809349, + "grad_norm": 0.08210425823926926, + "learning_rate": 1.589399705694593e-05, + "loss": 0.0152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15210 + }, + { + "epoch": 4.428280477160314, + "flow_matching_loss": 0.011878652125597, + "grad_norm": 0.059526607394218445, + "learning_rate": 1.587749121184485e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15220 + }, + { + "epoch": 4.431189991271458, + "flow_matching_loss": 0.012112385593354702, + "grad_norm": 0.07624858617782593, + "learning_rate": 1.5860984300509206e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15230 + }, + { + "epoch": 4.434099505382601, + "flow_matching_loss": 0.01101467665284872, + "grad_norm": 0.06283028423786163, + "learning_rate": 1.5844476342996453e-05, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15240 + }, + { + "epoch": 4.437009019493744, + "flow_matching_loss": 0.011086340621113777, + "grad_norm": 0.07091137766838074, + "learning_rate": 1.5827967359365306e-05, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15250 + }, + { + "epoch": 4.439918533604888, + "flow_matching_loss": 0.01197655387222767, + "grad_norm": 0.09176140278577805, + "learning_rate": 1.5811457369675758e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15260 + }, + { + "epoch": 4.442828047716032, + "flow_matching_loss": 0.012591863982379436, + "grad_norm": 0.07717109471559525, + "learning_rate": 1.5794946393988982e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15270 + }, + { + "epoch": 4.445737561827175, + "flow_matching_loss": 0.012726345658302307, + "grad_norm": 0.07776373624801636, + "learning_rate": 1.5778434452367382e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15280 + }, + { + "epoch": 4.448647075938318, + "flow_matching_loss": 0.012412574980407954, + "grad_norm": 0.0671505406498909, + "learning_rate": 1.576192156487452e-05, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15290 + }, + { + "epoch": 4.451556590049462, + "flow_matching_loss": 0.012236558459699154, + "grad_norm": 0.07206964492797852, + "learning_rate": 1.574540775157511e-05, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15300 + }, + { + "epoch": 4.4544661041606055, + "flow_matching_loss": 0.01335085667669773, + "grad_norm": 0.10014209151268005, + "learning_rate": 1.5728893032534998e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15310 + }, + { + "epoch": 4.457375618271748, + "flow_matching_loss": 0.011033355863764882, + "grad_norm": 0.09252733737230301, + "learning_rate": 1.571237742782112e-05, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15320 + }, + { + "epoch": 4.460285132382892, + "flow_matching_loss": 0.011884657479822636, + "grad_norm": 0.07106080651283264, + "learning_rate": 1.569586095750149e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15330 + }, + { + "epoch": 4.463194646494036, + "flow_matching_loss": 0.012885538395494223, + "grad_norm": 0.07921677082777023, + "learning_rate": 1.5679343641645193e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15340 + }, + { + "epoch": 4.466104160605179, + "flow_matching_loss": 0.012712163385003805, + "grad_norm": 0.07105975598096848, + "learning_rate": 1.5662825500322304e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15350 + }, + { + "epoch": 4.469013674716322, + "flow_matching_loss": 0.014271360822021961, + "grad_norm": 0.12264672666788101, + "learning_rate": 1.5646306553603932e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15360 + }, + { + "epoch": 4.471923188827466, + "flow_matching_loss": 0.012135438714176416, + "grad_norm": 0.07690636813640594, + "learning_rate": 1.5629786821562154e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15370 + }, + { + "epoch": 4.47483270293861, + "flow_matching_loss": 0.013688251096755265, + "grad_norm": 0.08515297621488571, + "learning_rate": 1.561326632427001e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15380 + }, + { + "epoch": 4.4777422170497525, + "flow_matching_loss": 0.012281844206154347, + "grad_norm": 0.06215362623333931, + "learning_rate": 1.5596745081801447e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15390 + }, + { + "epoch": 4.480651731160896, + "flow_matching_loss": 0.011205715034157038, + "grad_norm": 0.08203870058059692, + "learning_rate": 1.5580223114231348e-05, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15400 + }, + { + "epoch": 4.48356124527204, + "flow_matching_loss": 0.013183740247040988, + "grad_norm": 0.09901644289493561, + "learning_rate": 1.5563700441635445e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15410 + }, + { + "epoch": 4.486470759383183, + "flow_matching_loss": 0.012531279399991036, + "grad_norm": 0.09847154468297958, + "learning_rate": 1.5547177084090358e-05, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15420 + }, + { + "epoch": 4.4893802734943264, + "flow_matching_loss": 0.010797617863863707, + "grad_norm": 0.058534007519483566, + "learning_rate": 1.5530653061673514e-05, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15430 + }, + { + "epoch": 4.49228978760547, + "flow_matching_loss": 0.01138164266012609, + "grad_norm": 0.0754517912864685, + "learning_rate": 1.5514128394463167e-05, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15440 + }, + { + "epoch": 4.495199301716613, + "flow_matching_loss": 0.013933042343705893, + "grad_norm": 0.09574995934963226, + "learning_rate": 1.549760310253834e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15450 + }, + { + "epoch": 4.498108815827757, + "flow_matching_loss": 0.011564979469403625, + "grad_norm": 0.08063501864671707, + "learning_rate": 1.5481077205978816e-05, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15460 + }, + { + "epoch": 4.5010183299389, + "flow_matching_loss": 0.012142060790210962, + "grad_norm": 0.07179874181747437, + "learning_rate": 1.5464550724865135e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15470 + }, + { + "epoch": 4.503927844050043, + "flow_matching_loss": 0.013060390017926693, + "grad_norm": 0.07868154346942902, + "learning_rate": 1.5448023679278517e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15480 + }, + { + "epoch": 4.506837358161187, + "flow_matching_loss": 0.011501577403396367, + "grad_norm": 0.0773618072271347, + "learning_rate": 1.543149608930088e-05, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15490 + }, + { + "epoch": 4.509746872272331, + "flow_matching_loss": 0.013009783066809177, + "grad_norm": 0.08147561550140381, + "learning_rate": 1.5414967975014805e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15500 + }, + { + "epoch": 4.512656386383474, + "flow_matching_loss": 0.012808497436344624, + "grad_norm": 0.0829196497797966, + "learning_rate": 1.539843935650351e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15510 + }, + { + "epoch": 4.515565900494617, + "flow_matching_loss": 0.012502628099173307, + "grad_norm": 0.08821693807840347, + "learning_rate": 1.5381910253850834e-05, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15520 + }, + { + "epoch": 4.518475414605761, + "flow_matching_loss": 0.013640617299824954, + "grad_norm": 0.07170029729604721, + "learning_rate": 1.5365380687141183e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15530 + }, + { + "epoch": 4.521384928716905, + "flow_matching_loss": 0.013000825792551041, + "grad_norm": 0.07757821679115295, + "learning_rate": 1.5348850676459544e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15540 + }, + { + "epoch": 4.524294442828047, + "flow_matching_loss": 0.011901533138006926, + "grad_norm": 0.06159313768148422, + "learning_rate": 1.5332320241891438e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15550 + }, + { + "epoch": 4.527203956939191, + "flow_matching_loss": 0.013679082132875919, + "grad_norm": 0.08817652612924576, + "learning_rate": 1.5315789403522908e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15560 + }, + { + "epoch": 4.530113471050335, + "flow_matching_loss": 0.012568485457450151, + "grad_norm": 0.08264432847499847, + "learning_rate": 1.529925818144047e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15570 + }, + { + "epoch": 4.5330229851614785, + "flow_matching_loss": 0.012043741904199124, + "grad_norm": 0.07542986422777176, + "learning_rate": 1.5282726595731126e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15580 + }, + { + "epoch": 4.535932499272621, + "flow_matching_loss": 0.012247579218819738, + "grad_norm": 0.08619364351034164, + "learning_rate": 1.5266194666482304e-05, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15590 + }, + { + "epoch": 4.538842013383765, + "flow_matching_loss": 0.013496265467256307, + "grad_norm": 0.08903425186872482, + "learning_rate": 1.5249662413781866e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15600 + }, + { + "epoch": 4.541751527494909, + "flow_matching_loss": 0.012299573421478272, + "grad_norm": 0.10172484070062637, + "learning_rate": 1.5233129857718054e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15610 + }, + { + "epoch": 4.5446610416060516, + "flow_matching_loss": 0.010798867139965296, + "grad_norm": 0.07196561247110367, + "learning_rate": 1.521659701837948e-05, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15620 + }, + { + "epoch": 4.547570555717195, + "flow_matching_loss": 0.01256173336878419, + "grad_norm": 0.09014187753200531, + "learning_rate": 1.5200063915855107e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15630 + }, + { + "epoch": 4.550480069828339, + "flow_matching_loss": 0.012694302573800087, + "grad_norm": 0.06992723047733307, + "learning_rate": 1.5183530570234212e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15640 + }, + { + "epoch": 4.553389583939482, + "flow_matching_loss": 0.012546891067177057, + "grad_norm": 0.0583966001868248, + "learning_rate": 1.5166997001606361e-05, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15650 + }, + { + "epoch": 4.5562990980506255, + "flow_matching_loss": 0.011437196470797061, + "grad_norm": 0.07136863470077515, + "learning_rate": 1.5150463230061417e-05, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15660 + }, + { + "epoch": 4.559208612161769, + "flow_matching_loss": 0.012684909533709288, + "grad_norm": 0.06930498033761978, + "learning_rate": 1.5133929275689455e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15670 + }, + { + "epoch": 4.562118126272912, + "flow_matching_loss": 0.012251008115708829, + "grad_norm": 0.17243419587612152, + "learning_rate": 1.5117395158580795e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15680 + }, + { + "epoch": 4.565027640384056, + "flow_matching_loss": 0.012261008843779564, + "grad_norm": 0.08264695107936859, + "learning_rate": 1.510086089882595e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15690 + }, + { + "epoch": 4.567937154495199, + "flow_matching_loss": 0.012582523003220557, + "grad_norm": 0.08001397550106049, + "learning_rate": 1.5084326516515599e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15700 + }, + { + "epoch": 4.570846668606343, + "flow_matching_loss": 0.013150589680299163, + "grad_norm": 0.07949017733335495, + "learning_rate": 1.5067792031740582e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15710 + }, + { + "epoch": 4.573756182717486, + "flow_matching_loss": 0.013072742987424136, + "grad_norm": 0.10667234659194946, + "learning_rate": 1.505125746459185e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15720 + }, + { + "epoch": 4.57666569682863, + "flow_matching_loss": 0.01229816023260355, + "grad_norm": 0.09014542400836945, + "learning_rate": 1.5034722835160466e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15730 + }, + { + "epoch": 4.579575210939773, + "flow_matching_loss": 0.01121084913611412, + "grad_norm": 0.05719510465860367, + "learning_rate": 1.5018188163537566e-05, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15740 + }, + { + "epoch": 4.582484725050916, + "flow_matching_loss": 0.011877719312906265, + "grad_norm": 0.08557165414094925, + "learning_rate": 1.5001653469814331e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15750 + }, + { + "epoch": 4.58539423916206, + "flow_matching_loss": 0.011838355660438537, + "grad_norm": 0.0691252052783966, + "learning_rate": 1.4985118774081978e-05, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15760 + }, + { + "epoch": 4.588303753273204, + "flow_matching_loss": 0.01257286611944437, + "grad_norm": 0.09567476063966751, + "learning_rate": 1.4968584096431708e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15770 + }, + { + "epoch": 4.591213267384346, + "flow_matching_loss": 0.011806933581829071, + "grad_norm": 0.07718085497617722, + "learning_rate": 1.4952049456954731e-05, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15780 + }, + { + "epoch": 4.59412278149549, + "flow_matching_loss": 0.012832982745021582, + "grad_norm": 0.08107318729162216, + "learning_rate": 1.4935514875742182e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15790 + }, + { + "epoch": 4.597032295606634, + "flow_matching_loss": 0.011349229281768203, + "grad_norm": 0.08873460441827774, + "learning_rate": 1.4918980372885143e-05, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15800 + }, + { + "epoch": 4.599941809717777, + "flow_matching_loss": 0.012858840310946107, + "grad_norm": 0.06627368927001953, + "learning_rate": 1.4902445968474591e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15810 + }, + { + "epoch": 4.60285132382892, + "flow_matching_loss": 0.01297425664961338, + "grad_norm": 0.07444310188293457, + "learning_rate": 1.4885911682601378e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15820 + }, + { + "epoch": 4.605760837940064, + "flow_matching_loss": 0.012289527617394925, + "grad_norm": 0.07035349309444427, + "learning_rate": 1.4869377535356235e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15830 + }, + { + "epoch": 4.608670352051208, + "flow_matching_loss": 0.011564027424901724, + "grad_norm": 0.08386295288801193, + "learning_rate": 1.4852843546829704e-05, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15840 + }, + { + "epoch": 4.611579866162351, + "flow_matching_loss": 0.012603189796209335, + "grad_norm": 0.11729341000318527, + "learning_rate": 1.483630973711214e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15850 + }, + { + "epoch": 4.614489380273494, + "flow_matching_loss": 0.012322545610368251, + "grad_norm": 0.07474634796380997, + "learning_rate": 1.4819776126293677e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15860 + }, + { + "epoch": 4.617398894384638, + "flow_matching_loss": 0.012000572308897972, + "grad_norm": 0.2205224484205246, + "learning_rate": 1.4803242734464214e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15870 + }, + { + "epoch": 4.620308408495781, + "flow_matching_loss": 0.012104008998721837, + "grad_norm": 0.07134654372930527, + "learning_rate": 1.4786709581713383e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15880 + }, + { + "epoch": 4.6232179226069245, + "flow_matching_loss": 0.013284977246075868, + "grad_norm": 0.08667807281017303, + "learning_rate": 1.4770176688130524e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15890 + }, + { + "epoch": 4.626127436718068, + "flow_matching_loss": 0.014649080485105515, + "grad_norm": 0.06288594752550125, + "learning_rate": 1.475364407380466e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15900 + }, + { + "epoch": 4.629036950829212, + "flow_matching_loss": 0.012180530838668347, + "grad_norm": 0.08020012825727463, + "learning_rate": 1.4737111758824476e-05, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15910 + }, + { + "epoch": 4.631946464940355, + "flow_matching_loss": 0.012130334321409465, + "grad_norm": 0.07652346789836884, + "learning_rate": 1.4720579763278289e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15920 + }, + { + "epoch": 4.6348559790514985, + "flow_matching_loss": 0.013530004117637873, + "grad_norm": 0.06811786442995071, + "learning_rate": 1.4704048107254042e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15930 + }, + { + "epoch": 4.637765493162642, + "flow_matching_loss": 0.013287310395389796, + "grad_norm": 0.06377219408750534, + "learning_rate": 1.4687516810839246e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15940 + }, + { + "epoch": 4.640675007273785, + "flow_matching_loss": 0.013738627079874277, + "grad_norm": 0.07774698734283447, + "learning_rate": 1.4670985894120997e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15950 + }, + { + "epoch": 4.643584521384929, + "flow_matching_loss": 0.012452361360192298, + "grad_norm": 0.06666287779808044, + "learning_rate": 1.465445537718591e-05, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15960 + }, + { + "epoch": 4.646494035496072, + "flow_matching_loss": 0.011505126953125, + "grad_norm": 0.07167236506938934, + "learning_rate": 1.463792528012012e-05, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15970 + }, + { + "epoch": 4.649403549607215, + "flow_matching_loss": 0.015017636958509684, + "grad_norm": 0.059824537485837936, + "learning_rate": 1.4621395623009252e-05, + "loss": 0.015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15980 + }, + { + "epoch": 4.652313063718359, + "flow_matching_loss": 0.0132248611189425, + "grad_norm": 0.0776844248175621, + "learning_rate": 1.4604866425938407e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15990 + }, + { + "epoch": 4.655222577829503, + "flow_matching_loss": 0.012037350516766309, + "grad_norm": 0.07422196865081787, + "learning_rate": 1.4588337708992115e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16000 + }, + { + "epoch": 4.6581320919406455, + "flow_matching_loss": 0.012420498020946979, + "grad_norm": 0.07354327291250229, + "learning_rate": 1.4571809492254318e-05, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16010 + }, + { + "epoch": 4.661041606051789, + "flow_matching_loss": 0.014133196696639061, + "grad_norm": 0.0866200178861618, + "learning_rate": 1.4555281795808371e-05, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16020 + }, + { + "epoch": 4.663951120162933, + "flow_matching_loss": 0.012847883673384786, + "grad_norm": 0.0976148322224617, + "learning_rate": 1.4538754639736966e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16030 + }, + { + "epoch": 4.666860634274077, + "flow_matching_loss": 0.010705915279686451, + "grad_norm": 0.079405777156353, + "learning_rate": 1.4522228044122174e-05, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16040 + }, + { + "epoch": 4.669770148385219, + "flow_matching_loss": 0.01337145660072565, + "grad_norm": 0.09269309043884277, + "learning_rate": 1.4505702029045359e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16050 + }, + { + "epoch": 4.672679662496363, + "flow_matching_loss": 0.013205702975392341, + "grad_norm": 0.08850090950727463, + "learning_rate": 1.4489176614587189e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16060 + }, + { + "epoch": 4.675589176607507, + "flow_matching_loss": 0.01240725452080369, + "grad_norm": 0.08423317223787308, + "learning_rate": 1.4472651820827595e-05, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16070 + }, + { + "epoch": 4.67849869071865, + "flow_matching_loss": 0.01294075045734644, + "grad_norm": 0.07585937529802322, + "learning_rate": 1.4456127667845766e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16080 + }, + { + "epoch": 4.681408204829793, + "flow_matching_loss": 0.012344443984329701, + "grad_norm": 0.08080869913101196, + "learning_rate": 1.4439604175720104e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16090 + }, + { + "epoch": 4.684317718940937, + "flow_matching_loss": 0.011890137381851674, + "grad_norm": 0.06768543273210526, + "learning_rate": 1.4423081364528212e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16100 + }, + { + "epoch": 4.687227233052081, + "flow_matching_loss": 0.012262205127626657, + "grad_norm": 0.10445857793092728, + "learning_rate": 1.4406559254346863e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16110 + }, + { + "epoch": 4.690136747163224, + "flow_matching_loss": 0.012657944392412902, + "grad_norm": 0.09400296211242676, + "learning_rate": 1.4390037865251978e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16120 + }, + { + "epoch": 4.693046261274367, + "flow_matching_loss": 0.012206628825515509, + "grad_norm": 0.09506051987409592, + "learning_rate": 1.4373517217318597e-05, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16130 + }, + { + "epoch": 4.69595577538551, + "flow_matching_loss": 0.011909222649410366, + "grad_norm": 0.09117375314235687, + "learning_rate": 1.4356997330620875e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16140 + }, + { + "epoch": 4.698865289496654, + "flow_matching_loss": 0.011821175087243318, + "grad_norm": 0.06501957774162292, + "learning_rate": 1.434047822523203e-05, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16150 + }, + { + "epoch": 4.7017748036077975, + "flow_matching_loss": 0.014337362535297871, + "grad_norm": 0.08555147051811218, + "learning_rate": 1.4323959921224322e-05, + "loss": 0.0143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16160 + }, + { + "epoch": 4.704684317718941, + "flow_matching_loss": 0.01184965753927827, + "grad_norm": 0.08942795544862747, + "learning_rate": 1.4307442438669066e-05, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16170 + }, + { + "epoch": 4.707593831830084, + "flow_matching_loss": 0.01243455521762371, + "grad_norm": 0.06565604358911514, + "learning_rate": 1.4290925797636546e-05, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16180 + }, + { + "epoch": 4.710503345941228, + "flow_matching_loss": 0.012574842013418675, + "grad_norm": 0.08150700479745865, + "learning_rate": 1.4274410018196039e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16190 + }, + { + "epoch": 4.7134128600523715, + "flow_matching_loss": 0.013000651728361845, + "grad_norm": 0.06769342720508575, + "learning_rate": 1.4257895120415783e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16200 + }, + { + "epoch": 4.716322374163514, + "flow_matching_loss": 0.013695928640663624, + "grad_norm": 0.1103365495800972, + "learning_rate": 1.4241381124362933e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16210 + }, + { + "epoch": 4.719231888274658, + "flow_matching_loss": 0.01417354391887784, + "grad_norm": 0.07604174315929413, + "learning_rate": 1.4224868050103546e-05, + "loss": 0.0142, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16220 + }, + { + "epoch": 4.722141402385802, + "flow_matching_loss": 0.012482588551938534, + "grad_norm": 0.08411265909671783, + "learning_rate": 1.4208355917702573e-05, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16230 + }, + { + "epoch": 4.725050916496945, + "flow_matching_loss": 0.012355509214103222, + "grad_norm": 0.08365694433450699, + "learning_rate": 1.41918447472238e-05, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16240 + }, + { + "epoch": 4.727960430608088, + "flow_matching_loss": 0.012671943288296461, + "grad_norm": 0.08838405460119247, + "learning_rate": 1.4175334558729869e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16250 + }, + { + "epoch": 4.730869944719232, + "flow_matching_loss": 0.012633681762963533, + "grad_norm": 0.09989693015813828, + "learning_rate": 1.4158825372282211e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16260 + }, + { + "epoch": 4.733779458830376, + "flow_matching_loss": 0.012312524858862162, + "grad_norm": 0.07314862310886383, + "learning_rate": 1.4142317207941042e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16270 + }, + { + "epoch": 4.7366889729415185, + "flow_matching_loss": 0.012486561387777328, + "grad_norm": 0.05561090260744095, + "learning_rate": 1.4125810085765339e-05, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16280 + }, + { + "epoch": 4.739598487052662, + "flow_matching_loss": 0.013929995615035296, + "grad_norm": 0.07209901511669159, + "learning_rate": 1.410930402581281e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16290 + }, + { + "epoch": 4.742508001163806, + "flow_matching_loss": 0.012384308595210313, + "grad_norm": 0.09763772040605545, + "learning_rate": 1.4092799048139876e-05, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16300 + }, + { + "epoch": 4.745417515274949, + "flow_matching_loss": 0.012296232907101513, + "grad_norm": 0.0725608542561531, + "learning_rate": 1.4076295172801647e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16310 + }, + { + "epoch": 4.748327029386092, + "flow_matching_loss": 0.011416221223771571, + "grad_norm": 0.06041688844561577, + "learning_rate": 1.4059792419851884e-05, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16320 + }, + { + "epoch": 4.751236543497236, + "flow_matching_loss": 0.013597904238849878, + "grad_norm": 0.13167792558670044, + "learning_rate": 1.4043290809342984e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16330 + }, + { + "epoch": 4.754146057608379, + "flow_matching_loss": 0.011813896894454955, + "grad_norm": 0.08761502802371979, + "learning_rate": 1.4026790361325958e-05, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16340 + }, + { + "epoch": 4.757055571719523, + "flow_matching_loss": 0.012427508272230625, + "grad_norm": 0.07843992114067078, + "learning_rate": 1.4010291095850415e-05, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16350 + }, + { + "epoch": 4.759965085830666, + "flow_matching_loss": 0.012177314143627882, + "grad_norm": 0.07808153331279755, + "learning_rate": 1.3993793032964515e-05, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16360 + }, + { + "epoch": 4.76287459994181, + "flow_matching_loss": 0.013030512630939484, + "grad_norm": 0.08940644562244415, + "learning_rate": 1.3977296192714956e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16370 + }, + { + "epoch": 4.765784114052953, + "flow_matching_loss": 0.011706584692001342, + "grad_norm": 0.07734738290309906, + "learning_rate": 1.396080059514696e-05, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16380 + }, + { + "epoch": 4.768693628164097, + "flow_matching_loss": 0.01388099193572998, + "grad_norm": 0.08830130845308304, + "learning_rate": 1.3944306260304232e-05, + "loss": 0.0139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16390 + }, + { + "epoch": 4.77160314227524, + "flow_matching_loss": 0.01207869928330183, + "grad_norm": 0.08138339966535568, + "learning_rate": 1.3927813208228936e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16400 + }, + { + "epoch": 4.774512656386383, + "flow_matching_loss": 0.01151073444634676, + "grad_norm": 0.060227468609809875, + "learning_rate": 1.39113214589617e-05, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16410 + }, + { + "epoch": 4.777422170497527, + "flow_matching_loss": 0.01131387148052454, + "grad_norm": 0.07584517449140549, + "learning_rate": 1.3894831032541545e-05, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16420 + }, + { + "epoch": 4.7803316846086705, + "flow_matching_loss": 0.011964324302971363, + "grad_norm": 0.05621517077088356, + "learning_rate": 1.3878341949005895e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16430 + }, + { + "epoch": 4.783241198719814, + "flow_matching_loss": 0.012444240413606167, + "grad_norm": 0.06668926030397415, + "learning_rate": 1.3861854228390546e-05, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16440 + }, + { + "epoch": 4.786150712830957, + "flow_matching_loss": 0.012572192493826151, + "grad_norm": 0.07286295294761658, + "learning_rate": 1.3845367890729621e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16450 + }, + { + "epoch": 4.789060226942101, + "flow_matching_loss": 0.013085719291120767, + "grad_norm": 0.08814653754234314, + "learning_rate": 1.3828882956055591e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16460 + }, + { + "epoch": 4.7919697410532445, + "flow_matching_loss": 0.011512646079063415, + "grad_norm": 0.0692005306482315, + "learning_rate": 1.38123994443992e-05, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16470 + }, + { + "epoch": 4.794879255164387, + "flow_matching_loss": 0.010908878967165947, + "grad_norm": 0.07274910062551498, + "learning_rate": 1.3795917375789465e-05, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16480 + }, + { + "epoch": 4.797788769275531, + "flow_matching_loss": 0.012529705977067351, + "grad_norm": 0.07832375168800354, + "learning_rate": 1.3779436770253658e-05, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16490 + }, + { + "epoch": 4.800698283386675, + "flow_matching_loss": 0.01198336249217391, + "grad_norm": 0.09047137945890427, + "learning_rate": 1.3762957647817258e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16500 + }, + { + "epoch": 4.8036077974978175, + "flow_matching_loss": 0.011419107997789979, + "grad_norm": 0.06290780007839203, + "learning_rate": 1.374648002850397e-05, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16510 + }, + { + "epoch": 4.806517311608961, + "flow_matching_loss": 0.011220346111804247, + "grad_norm": 0.07637317478656769, + "learning_rate": 1.3730003932335642e-05, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16520 + }, + { + "epoch": 4.809426825720105, + "flow_matching_loss": 0.011459696060046553, + "grad_norm": 0.06977890431880951, + "learning_rate": 1.3713529379332292e-05, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16530 + }, + { + "epoch": 4.812336339831248, + "flow_matching_loss": 0.011751768738031387, + "grad_norm": 0.09027493745088577, + "learning_rate": 1.3697056389512054e-05, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16540 + }, + { + "epoch": 4.8152458539423915, + "flow_matching_loss": 0.012285243812948465, + "grad_norm": 0.08203183114528656, + "learning_rate": 1.3680584982891162e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16550 + }, + { + "epoch": 4.818155368053535, + "flow_matching_loss": 0.010516463406383991, + "grad_norm": 0.08351758122444153, + "learning_rate": 1.3664115179483921e-05, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16560 + }, + { + "epoch": 4.821064882164679, + "flow_matching_loss": 0.011831153929233552, + "grad_norm": 0.06982949376106262, + "learning_rate": 1.364764699930271e-05, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16570 + }, + { + "epoch": 4.823974396275822, + "flow_matching_loss": 0.010909044928848744, + "grad_norm": 0.05918826535344124, + "learning_rate": 1.363118046235791e-05, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16580 + }, + { + "epoch": 4.826883910386965, + "flow_matching_loss": 0.01131079699844122, + "grad_norm": 0.07791944593191147, + "learning_rate": 1.3614715588657921e-05, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16590 + }, + { + "epoch": 4.829793424498109, + "flow_matching_loss": 0.012093797326087952, + "grad_norm": 0.10392001271247864, + "learning_rate": 1.3598252398209118e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16600 + }, + { + "epoch": 4.832702938609252, + "flow_matching_loss": 0.013428058102726936, + "grad_norm": 0.07275225222110748, + "learning_rate": 1.358179091101582e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16610 + }, + { + "epoch": 4.835612452720396, + "flow_matching_loss": 0.01218098048120737, + "grad_norm": 0.06704240292310715, + "learning_rate": 1.3565331147080299e-05, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16620 + }, + { + "epoch": 4.838521966831539, + "flow_matching_loss": 0.013977711834013463, + "grad_norm": 0.07477643340826035, + "learning_rate": 1.3548873126402716e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16630 + }, + { + "epoch": 4.841431480942683, + "flow_matching_loss": 0.01030165757983923, + "grad_norm": 0.06740088760852814, + "learning_rate": 1.3532416868981117e-05, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16640 + }, + { + "epoch": 4.844340995053826, + "flow_matching_loss": 0.01276254327967763, + "grad_norm": 0.08300517499446869, + "learning_rate": 1.3515962394811401e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16650 + }, + { + "epoch": 4.84725050916497, + "flow_matching_loss": 0.012979493848979473, + "grad_norm": 0.12983554601669312, + "learning_rate": 1.3499509723887313e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16660 + }, + { + "epoch": 4.850160023276112, + "flow_matching_loss": 0.011959092039614917, + "grad_norm": 0.07564710825681686, + "learning_rate": 1.3483058876200396e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16670 + }, + { + "epoch": 4.853069537387256, + "flow_matching_loss": 0.012355373334139585, + "grad_norm": 0.060724761337041855, + "learning_rate": 1.3466609871739987e-05, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16680 + }, + { + "epoch": 4.8559790514984, + "flow_matching_loss": 0.011939580645412206, + "grad_norm": 0.07905786484479904, + "learning_rate": 1.3450162730493172e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16690 + }, + { + "epoch": 4.8588885656095435, + "flow_matching_loss": 0.010344566591084004, + "grad_norm": 0.06123218685388565, + "learning_rate": 1.3433717472444781e-05, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16700 + }, + { + "epoch": 4.861798079720686, + "flow_matching_loss": 0.011699367640540003, + "grad_norm": 0.06818483769893646, + "learning_rate": 1.3417274117577343e-05, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16710 + }, + { + "epoch": 4.86470759383183, + "flow_matching_loss": 0.01287763761356473, + "grad_norm": 0.07721693813800812, + "learning_rate": 1.3400832685871102e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16720 + }, + { + "epoch": 4.867617107942974, + "flow_matching_loss": 0.013252853509038686, + "grad_norm": 0.07477588951587677, + "learning_rate": 1.3384393197303936e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16730 + }, + { + "epoch": 4.870526622054117, + "flow_matching_loss": 0.011977143213152885, + "grad_norm": 0.0672844722867012, + "learning_rate": 1.336795567185138e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16740 + }, + { + "epoch": 4.87343613616526, + "flow_matching_loss": 0.01199538353830576, + "grad_norm": 0.08044341206550598, + "learning_rate": 1.3351520129486573e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16750 + }, + { + "epoch": 4.876345650276404, + "flow_matching_loss": 0.012667759973555803, + "grad_norm": 0.08840233087539673, + "learning_rate": 1.3335086590180252e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16760 + }, + { + "epoch": 4.879255164387548, + "flow_matching_loss": 0.012835921626538039, + "grad_norm": 0.09496687352657318, + "learning_rate": 1.3318655073900708e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16770 + }, + { + "epoch": 4.8821646784986905, + "flow_matching_loss": 0.01188533278182149, + "grad_norm": 0.0687289834022522, + "learning_rate": 1.3302225600613796e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16780 + }, + { + "epoch": 4.885074192609834, + "flow_matching_loss": 0.011546964850276708, + "grad_norm": 0.08181528002023697, + "learning_rate": 1.3285798190282869e-05, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16790 + }, + { + "epoch": 4.887983706720978, + "flow_matching_loss": 0.011988608352839946, + "grad_norm": 0.08311502635478973, + "learning_rate": 1.3269372862868776e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16800 + }, + { + "epoch": 4.890893220832121, + "flow_matching_loss": 0.011848117504268885, + "grad_norm": 0.09048815071582794, + "learning_rate": 1.3252949638329845e-05, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16810 + }, + { + "epoch": 4.8938027349432645, + "flow_matching_loss": 0.013079285062849522, + "grad_norm": 0.07154589891433716, + "learning_rate": 1.3236528536621834e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16820 + }, + { + "epoch": 4.896712249054408, + "flow_matching_loss": 0.012494944129139185, + "grad_norm": 0.07550466060638428, + "learning_rate": 1.3220109577697943e-05, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16830 + }, + { + "epoch": 4.899621763165551, + "flow_matching_loss": 0.01259272675961256, + "grad_norm": 0.07618536055088043, + "learning_rate": 1.3203692781508745e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16840 + }, + { + "epoch": 4.902531277276695, + "flow_matching_loss": 0.01261768965050578, + "grad_norm": 0.09654698520898819, + "learning_rate": 1.3187278168002198e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16850 + }, + { + "epoch": 4.905440791387838, + "flow_matching_loss": 0.012442667875438929, + "grad_norm": 0.11612565815448761, + "learning_rate": 1.3170865757123606e-05, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16860 + }, + { + "epoch": 4.908350305498981, + "flow_matching_loss": 0.012215909222140909, + "grad_norm": 0.07590119540691376, + "learning_rate": 1.3154455568815592e-05, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16870 + }, + { + "epoch": 4.911259819610125, + "flow_matching_loss": 0.012035066960379481, + "grad_norm": 0.07679706811904907, + "learning_rate": 1.3138047623018087e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16880 + }, + { + "epoch": 4.914169333721269, + "flow_matching_loss": 0.011901527922600508, + "grad_norm": 0.08221637457609177, + "learning_rate": 1.3121641939668293e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16890 + }, + { + "epoch": 4.917078847832412, + "flow_matching_loss": 0.012287847511470319, + "grad_norm": 0.1049543246626854, + "learning_rate": 1.3105238538700659e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16900 + }, + { + "epoch": 4.919988361943555, + "flow_matching_loss": 0.011158622056245803, + "grad_norm": 0.06939084082841873, + "learning_rate": 1.3088837440046866e-05, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16910 + }, + { + "epoch": 4.922897876054699, + "flow_matching_loss": 0.011962550133466721, + "grad_norm": 0.0703931450843811, + "learning_rate": 1.3072438663635787e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16920 + }, + { + "epoch": 4.925807390165843, + "flow_matching_loss": 0.012377348449081182, + "grad_norm": 0.07799417525529861, + "learning_rate": 1.3056042229393497e-05, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16930 + }, + { + "epoch": 4.928716904276985, + "flow_matching_loss": 0.012450177036225796, + "grad_norm": 0.07827797532081604, + "learning_rate": 1.30396481572432e-05, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16940 + }, + { + "epoch": 4.931626418388129, + "flow_matching_loss": 0.014614650141447782, + "grad_norm": 0.08025112003087997, + "learning_rate": 1.3023256467105237e-05, + "loss": 0.0146, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16950 + }, + { + "epoch": 4.934535932499273, + "flow_matching_loss": 0.01108954343944788, + "grad_norm": 0.08296813070774078, + "learning_rate": 1.3006867178897067e-05, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16960 + }, + { + "epoch": 4.9374454466104165, + "flow_matching_loss": 0.01168946186080575, + "grad_norm": 0.08076810091733932, + "learning_rate": 1.2990480312533213e-05, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16970 + }, + { + "epoch": 4.940354960721559, + "flow_matching_loss": 0.013539561163634061, + "grad_norm": 0.07518918812274933, + "learning_rate": 1.2974095887925258e-05, + "loss": 0.0135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16980 + }, + { + "epoch": 4.943264474832703, + "flow_matching_loss": 0.012019907590001822, + "grad_norm": 0.07719980925321579, + "learning_rate": 1.2957713924981837e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16990 + }, + { + "epoch": 4.946173988943847, + "flow_matching_loss": 0.011035476857796312, + "grad_norm": 0.06962855160236359, + "learning_rate": 1.2941334443608572e-05, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17000 + }, + { + "epoch": 4.94908350305499, + "flow_matching_loss": 0.011901284940540791, + "grad_norm": 0.0739063024520874, + "learning_rate": 1.2924957463708073e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17010 + }, + { + "epoch": 4.951993017166133, + "flow_matching_loss": 0.012005923269316553, + "grad_norm": 0.07401658594608307, + "learning_rate": 1.2908583005179924e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17020 + }, + { + "epoch": 4.954902531277277, + "flow_matching_loss": 0.013611771818250417, + "grad_norm": 0.07569872587919235, + "learning_rate": 1.2892211087920624e-05, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17030 + }, + { + "epoch": 4.95781204538842, + "flow_matching_loss": 0.00980057385750115, + "grad_norm": 0.06541982293128967, + "learning_rate": 1.287584173182361e-05, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17040 + }, + { + "epoch": 4.9607215594995635, + "flow_matching_loss": 0.01166467871516943, + "grad_norm": 0.062475744634866714, + "learning_rate": 1.2859474956779188e-05, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17050 + }, + { + "epoch": 4.963631073610707, + "flow_matching_loss": 0.011932212114334106, + "grad_norm": 0.09282271564006805, + "learning_rate": 1.2843110782674532e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17060 + }, + { + "epoch": 4.96654058772185, + "flow_matching_loss": 0.011475703492760659, + "grad_norm": 0.08792504668235779, + "learning_rate": 1.2826749229393655e-05, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17070 + }, + { + "epoch": 4.969450101832994, + "flow_matching_loss": 0.013097549602389336, + "grad_norm": 0.07064969092607498, + "learning_rate": 1.2810390316817381e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17080 + }, + { + "epoch": 4.9723596159441374, + "flow_matching_loss": 0.01162692504003644, + "grad_norm": 0.09198804199695587, + "learning_rate": 1.2794034064823346e-05, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17090 + }, + { + "epoch": 4.975269130055281, + "flow_matching_loss": 0.011591194849461316, + "grad_norm": 0.09173092991113663, + "learning_rate": 1.277768049328593e-05, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17100 + }, + { + "epoch": 4.978178644166424, + "flow_matching_loss": 0.01266457187011838, + "grad_norm": 0.07951425760984421, + "learning_rate": 1.2761329622076269e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17110 + }, + { + "epoch": 4.981088158277568, + "flow_matching_loss": 0.011850544065237046, + "grad_norm": 0.07000786811113358, + "learning_rate": 1.274498147106221e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17120 + }, + { + "epoch": 4.983997672388711, + "flow_matching_loss": 0.010836872365325689, + "grad_norm": 0.06852371990680695, + "learning_rate": 1.272863606010829e-05, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17130 + }, + { + "epoch": 4.986907186499854, + "flow_matching_loss": 0.012704213336110114, + "grad_norm": 0.08025939017534256, + "learning_rate": 1.2712293409075739e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17140 + }, + { + "epoch": 4.989816700610998, + "flow_matching_loss": 0.0126849633641541, + "grad_norm": 0.08470860868692398, + "learning_rate": 1.2695953537822413e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17150 + }, + { + "epoch": 4.992726214722142, + "flow_matching_loss": 0.013015072047710418, + "grad_norm": 0.1333530992269516, + "learning_rate": 1.2679616466202792e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17160 + }, + { + "epoch": 4.995635728833285, + "flow_matching_loss": 0.014422891195863485, + "grad_norm": 0.10126349329948425, + "learning_rate": 1.2663282214067967e-05, + "loss": 0.0144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17170 + }, + { + "epoch": 4.998545242944428, + "flow_matching_loss": 0.01452076192945242, + "grad_norm": 0.08485343307256699, + "learning_rate": 1.264695080126559e-05, + "loss": 0.0145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17180 + }, + { + "epoch": 5.001454757055572, + "flow_matching_loss": 0.010962570924311877, + "grad_norm": 0.07162106037139893, + "learning_rate": 1.263062224763986e-05, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17190 + }, + { + "epoch": 5.004364271166716, + "flow_matching_loss": 0.012299869023263455, + "grad_norm": 0.11704385280609131, + "learning_rate": 1.2614296573031522e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17200 + }, + { + "epoch": 5.007273785277858, + "flow_matching_loss": 0.012476841360330582, + "grad_norm": 0.06714934855699539, + "learning_rate": 1.259797379727781e-05, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17210 + }, + { + "epoch": 5.010183299389002, + "flow_matching_loss": 0.011112476792186498, + "grad_norm": 0.06658918410539627, + "learning_rate": 1.2581653940212423e-05, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17220 + }, + { + "epoch": 5.013092813500146, + "flow_matching_loss": 0.01206516269594431, + "grad_norm": 0.07689537853002548, + "learning_rate": 1.2565337021665537e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17230 + }, + { + "epoch": 5.016002327611289, + "flow_matching_loss": 0.012471249047666787, + "grad_norm": 0.08702100813388824, + "learning_rate": 1.2549023061463739e-05, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17240 + }, + { + "epoch": 5.018911841722432, + "flow_matching_loss": 0.012056828755885363, + "grad_norm": 0.09069876372814178, + "learning_rate": 1.2532712079430038e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17250 + }, + { + "epoch": 5.021821355833576, + "flow_matching_loss": 0.013704250566661358, + "grad_norm": 0.0922384038567543, + "learning_rate": 1.2516404095383813e-05, + "loss": 0.0137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17260 + }, + { + "epoch": 5.024730869944719, + "flow_matching_loss": 0.013116658059880137, + "grad_norm": 0.09576570987701416, + "learning_rate": 1.2500099129140796e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17270 + }, + { + "epoch": 5.0276403840558626, + "flow_matching_loss": 0.011779945529997348, + "grad_norm": 0.08006936311721802, + "learning_rate": 1.2483797200513065e-05, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17280 + }, + { + "epoch": 5.030549898167006, + "flow_matching_loss": 0.012610200699418784, + "grad_norm": 0.15134473145008087, + "learning_rate": 1.2467498329308992e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17290 + }, + { + "epoch": 5.03345941227815, + "flow_matching_loss": 0.012678340543061495, + "grad_norm": 0.09347978234291077, + "learning_rate": 1.2451202535333253e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17300 + }, + { + "epoch": 5.036368926389293, + "flow_matching_loss": 0.013337821047753096, + "grad_norm": 0.0740213692188263, + "learning_rate": 1.2434909838386768e-05, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17310 + }, + { + "epoch": 5.0392784405004365, + "flow_matching_loss": 0.01072750212624669, + "grad_norm": 0.05548504367470741, + "learning_rate": 1.2418620258266703e-05, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17320 + }, + { + "epoch": 5.04218795461158, + "flow_matching_loss": 0.01263734782114625, + "grad_norm": 0.0841054767370224, + "learning_rate": 1.2402333814766432e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17330 + }, + { + "epoch": 5.045097468722723, + "flow_matching_loss": 0.010595139674842357, + "grad_norm": 0.12616822123527527, + "learning_rate": 1.2386050527675515e-05, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17340 + }, + { + "epoch": 5.048006982833867, + "flow_matching_loss": 0.010795990098267793, + "grad_norm": 0.08536268770694733, + "learning_rate": 1.2369770416779693e-05, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17350 + }, + { + "epoch": 5.05091649694501, + "flow_matching_loss": 0.011298159696161747, + "grad_norm": 0.06659945100545883, + "learning_rate": 1.2353493501860828e-05, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17360 + }, + { + "epoch": 5.053826011056153, + "flow_matching_loss": 0.012864857167005538, + "grad_norm": 0.09351226687431335, + "learning_rate": 1.2337219802696905e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17370 + }, + { + "epoch": 5.056735525167297, + "flow_matching_loss": 0.012750377506017685, + "grad_norm": 0.07490944117307663, + "learning_rate": 1.2320949339062008e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17380 + }, + { + "epoch": 5.059645039278441, + "flow_matching_loss": 0.012740754708647727, + "grad_norm": 0.11211248487234116, + "learning_rate": 1.2304682130726281e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17390 + }, + { + "epoch": 5.0625545533895835, + "flow_matching_loss": 0.012704232707619667, + "grad_norm": 0.08162230998277664, + "learning_rate": 1.2288418197455913e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17400 + }, + { + "epoch": 5.065464067500727, + "flow_matching_loss": 0.01277845874428749, + "grad_norm": 0.08044886589050293, + "learning_rate": 1.2272157559013126e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17410 + }, + { + "epoch": 5.068373581611871, + "flow_matching_loss": 0.011728811636567115, + "grad_norm": 0.07762512564659119, + "learning_rate": 1.2255900235156123e-05, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17420 + }, + { + "epoch": 5.071283095723015, + "flow_matching_loss": 0.011340196803212165, + "grad_norm": 0.10361719876527786, + "learning_rate": 1.2239646245639087e-05, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17430 + }, + { + "epoch": 5.074192609834157, + "flow_matching_loss": 0.011761241033673287, + "grad_norm": 0.07218340039253235, + "learning_rate": 1.2223395610212142e-05, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17440 + }, + { + "epoch": 5.077102123945301, + "flow_matching_loss": 0.011872072145342827, + "grad_norm": 0.06217770278453827, + "learning_rate": 1.2207148348621348e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17450 + }, + { + "epoch": 5.080011638056445, + "flow_matching_loss": 0.011863450892269612, + "grad_norm": 0.07579617947340012, + "learning_rate": 1.219090448060866e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17460 + }, + { + "epoch": 5.082921152167588, + "flow_matching_loss": 0.012771452311426401, + "grad_norm": 0.06419515609741211, + "learning_rate": 1.217466402591191e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17470 + }, + { + "epoch": 5.085830666278731, + "flow_matching_loss": 0.011470094602555037, + "grad_norm": 0.06896653026342392, + "learning_rate": 1.2158427004264785e-05, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17480 + }, + { + "epoch": 5.088740180389875, + "flow_matching_loss": 0.012736793886870146, + "grad_norm": 0.0759667456150055, + "learning_rate": 1.2142193435396788e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17490 + }, + { + "epoch": 5.091649694501018, + "flow_matching_loss": 0.011560899950563908, + "grad_norm": 0.06611312925815582, + "learning_rate": 1.2125963339033234e-05, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17500 + }, + { + "epoch": 5.094559208612162, + "flow_matching_loss": 0.01150133293122053, + "grad_norm": 0.07653578370809555, + "learning_rate": 1.2109736734895237e-05, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17510 + }, + { + "epoch": 5.097468722723305, + "flow_matching_loss": 0.012196579761803151, + "grad_norm": 0.06881516426801682, + "learning_rate": 1.2093513642699634e-05, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17520 + }, + { + "epoch": 5.100378236834449, + "flow_matching_loss": 0.012292442657053471, + "grad_norm": 0.08057117462158203, + "learning_rate": 1.2077294082159024e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17530 + }, + { + "epoch": 5.103287750945592, + "flow_matching_loss": 0.012072954140603543, + "grad_norm": 0.089126817882061, + "learning_rate": 1.2061078072981697e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17540 + }, + { + "epoch": 5.1061972650567355, + "flow_matching_loss": 0.011691882275044918, + "grad_norm": 0.08647347241640091, + "learning_rate": 1.2044865634871626e-05, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17550 + }, + { + "epoch": 5.109106779167879, + "flow_matching_loss": 0.011881912220269441, + "grad_norm": 0.0650336965918541, + "learning_rate": 1.2028656787528468e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17560 + }, + { + "epoch": 5.112016293279022, + "flow_matching_loss": 0.010626645106822253, + "grad_norm": 0.08583660423755646, + "learning_rate": 1.2012451550647491e-05, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17570 + }, + { + "epoch": 5.114925807390166, + "flow_matching_loss": 0.010671341791749, + "grad_norm": 0.06952552497386932, + "learning_rate": 1.1996249943919589e-05, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17580 + }, + { + "epoch": 5.1178353215013095, + "flow_matching_loss": 0.010517854616045952, + "grad_norm": 0.06776769459247589, + "learning_rate": 1.1980051987031237e-05, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17590 + }, + { + "epoch": 5.120744835612452, + "flow_matching_loss": 0.010395173635333776, + "grad_norm": 0.06907788664102554, + "learning_rate": 1.1963857699664483e-05, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17600 + }, + { + "epoch": 5.123654349723596, + "flow_matching_loss": 0.012743244133889674, + "grad_norm": 0.07014105468988419, + "learning_rate": 1.1947667101496911e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17610 + }, + { + "epoch": 5.12656386383474, + "flow_matching_loss": 0.012119691167026759, + "grad_norm": 0.0655781552195549, + "learning_rate": 1.1931480212201633e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17620 + }, + { + "epoch": 5.129473377945883, + "flow_matching_loss": 0.010345116630196572, + "grad_norm": 0.07366937398910522, + "learning_rate": 1.1915297051447237e-05, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17630 + }, + { + "epoch": 5.132382892057026, + "flow_matching_loss": 0.010812005260959268, + "grad_norm": 0.06687987595796585, + "learning_rate": 1.1899117638897793e-05, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17640 + }, + { + "epoch": 5.13529240616817, + "flow_matching_loss": 0.012150796968489886, + "grad_norm": 0.07726689428091049, + "learning_rate": 1.1882941994212811e-05, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17650 + }, + { + "epoch": 5.138201920279314, + "flow_matching_loss": 0.011970588378608227, + "grad_norm": 0.0743100643157959, + "learning_rate": 1.1866770137047221e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17660 + }, + { + "epoch": 5.1411114343904565, + "flow_matching_loss": 0.012645612843334675, + "grad_norm": 0.07312867790460587, + "learning_rate": 1.1850602087051363e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17670 + }, + { + "epoch": 5.1440209485016, + "flow_matching_loss": 0.011369618587195873, + "grad_norm": 0.06729795038700104, + "learning_rate": 1.183443786387094e-05, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17680 + }, + { + "epoch": 5.146930462612744, + "flow_matching_loss": 0.012272614520043135, + "grad_norm": 0.09332886338233948, + "learning_rate": 1.1818277487147005e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17690 + }, + { + "epoch": 5.149839976723887, + "flow_matching_loss": 0.011800751183182, + "grad_norm": 0.0707053691148758, + "learning_rate": 1.180212097651594e-05, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17700 + }, + { + "epoch": 5.15274949083503, + "flow_matching_loss": 0.013121196627616882, + "grad_norm": 0.07238209992647171, + "learning_rate": 1.1785968351609422e-05, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17710 + }, + { + "epoch": 5.155659004946174, + "flow_matching_loss": 0.011328371195122599, + "grad_norm": 0.08886493742465973, + "learning_rate": 1.1769819632054427e-05, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17720 + }, + { + "epoch": 5.158568519057318, + "flow_matching_loss": 0.011914453469216824, + "grad_norm": 0.07839754968881607, + "learning_rate": 1.1753674837473165e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17730 + }, + { + "epoch": 5.161478033168461, + "flow_matching_loss": 0.010856694541871548, + "grad_norm": 0.06135198101401329, + "learning_rate": 1.1737533987483081e-05, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17740 + }, + { + "epoch": 5.164387547279604, + "flow_matching_loss": 0.011383802257478237, + "grad_norm": 0.059958476573228836, + "learning_rate": 1.1721397101696835e-05, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17750 + }, + { + "epoch": 5.167297061390748, + "flow_matching_loss": 0.011663253093138337, + "grad_norm": 0.07363416999578476, + "learning_rate": 1.1705264199722257e-05, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17760 + }, + { + "epoch": 5.170206575501891, + "flow_matching_loss": 0.012709229532629252, + "grad_norm": 0.06833469867706299, + "learning_rate": 1.1689135301162356e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17770 + }, + { + "epoch": 5.173116089613035, + "flow_matching_loss": 0.011283752648159862, + "grad_norm": 0.09060972929000854, + "learning_rate": 1.1673010425615258e-05, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17780 + }, + { + "epoch": 5.176025603724178, + "flow_matching_loss": 0.012925700657069683, + "grad_norm": 0.08208466321229935, + "learning_rate": 1.1656889592674208e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17790 + }, + { + "epoch": 5.178935117835321, + "flow_matching_loss": 0.012862254492938518, + "grad_norm": 0.08336563408374786, + "learning_rate": 1.1640772821927533e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17800 + }, + { + "epoch": 5.181844631946465, + "flow_matching_loss": 0.011402813624590636, + "grad_norm": 0.07709521800279617, + "learning_rate": 1.162466013295864e-05, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17810 + }, + { + "epoch": 5.1847541460576085, + "flow_matching_loss": 0.01156217735260725, + "grad_norm": 0.07093322277069092, + "learning_rate": 1.160855154534595e-05, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17820 + }, + { + "epoch": 5.187663660168752, + "flow_matching_loss": 0.011170644173398614, + "grad_norm": 0.06082039698958397, + "learning_rate": 1.1592447078662931e-05, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17830 + }, + { + "epoch": 5.190573174279895, + "flow_matching_loss": 0.013353346846997738, + "grad_norm": 0.07187207043170929, + "learning_rate": 1.1576346752478025e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17840 + }, + { + "epoch": 5.193482688391039, + "flow_matching_loss": 0.011321915965527295, + "grad_norm": 0.0865040197968483, + "learning_rate": 1.1560250586354643e-05, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17850 + }, + { + "epoch": 5.1963922025021825, + "flow_matching_loss": 0.010683856904506683, + "grad_norm": 0.07699424773454666, + "learning_rate": 1.1544158599851146e-05, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17860 + }, + { + "epoch": 5.199301716613325, + "flow_matching_loss": 0.011543207429349422, + "grad_norm": 0.0842663124203682, + "learning_rate": 1.1528070812520809e-05, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17870 + }, + { + "epoch": 5.202211230724469, + "flow_matching_loss": 0.011042265500873328, + "grad_norm": 0.07627267390489578, + "learning_rate": 1.1511987243911821e-05, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17880 + }, + { + "epoch": 5.205120744835613, + "flow_matching_loss": 0.011169884447008371, + "grad_norm": 0.0669635459780693, + "learning_rate": 1.1495907913567228e-05, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17890 + }, + { + "epoch": 5.2080302589467555, + "flow_matching_loss": 0.010559007804840804, + "grad_norm": 0.08505043387413025, + "learning_rate": 1.1479832841024936e-05, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17900 + }, + { + "epoch": 5.210939773057899, + "flow_matching_loss": 0.0125642747618258, + "grad_norm": 0.08552000671625137, + "learning_rate": 1.1463762045817675e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17910 + }, + { + "epoch": 5.213849287169043, + "flow_matching_loss": 0.011370225809514523, + "grad_norm": 0.06786242127418518, + "learning_rate": 1.1447695547472964e-05, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17920 + }, + { + "epoch": 5.216758801280186, + "flow_matching_loss": 0.011933476012200118, + "grad_norm": 0.0697856992483139, + "learning_rate": 1.1431633365513132e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17930 + }, + { + "epoch": 5.2196683153913295, + "flow_matching_loss": 0.011911987978965043, + "grad_norm": 0.10280648618936539, + "learning_rate": 1.1415575519455235e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17940 + }, + { + "epoch": 5.222577829502473, + "flow_matching_loss": 0.010949828010052443, + "grad_norm": 0.07509279251098633, + "learning_rate": 1.1399522028811069e-05, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17950 + }, + { + "epoch": 5.225487343613617, + "flow_matching_loss": 0.01198321464471519, + "grad_norm": 0.06167471036314964, + "learning_rate": 1.1383472913087143e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17960 + }, + { + "epoch": 5.22839685772476, + "flow_matching_loss": 0.01079658092930913, + "grad_norm": 0.0748199075460434, + "learning_rate": 1.1367428191784643e-05, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17970 + }, + { + "epoch": 5.231306371835903, + "flow_matching_loss": 0.011788365617394447, + "grad_norm": 0.07384754717350006, + "learning_rate": 1.1351387884399413e-05, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17980 + }, + { + "epoch": 5.234215885947047, + "flow_matching_loss": 0.012269437871873378, + "grad_norm": 0.08557983487844467, + "learning_rate": 1.1335352010421953e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17990 + }, + { + "epoch": 5.23712540005819, + "flow_matching_loss": 0.012435451243072748, + "grad_norm": 0.08306587487459183, + "learning_rate": 1.1319320589337352e-05, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18000 + }, + { + "epoch": 5.240034914169334, + "flow_matching_loss": 0.011580742243677378, + "grad_norm": 0.07238753139972687, + "learning_rate": 1.1303293640625299e-05, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18010 + }, + { + "epoch": 5.242944428280477, + "flow_matching_loss": 0.011312491912394761, + "grad_norm": 0.07475938647985458, + "learning_rate": 1.1287271183760052e-05, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18020 + }, + { + "epoch": 5.24585394239162, + "flow_matching_loss": 0.011437356285750867, + "grad_norm": 0.06085975095629692, + "learning_rate": 1.1271253238210398e-05, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18030 + }, + { + "epoch": 5.248763456502764, + "flow_matching_loss": 0.011326830461621284, + "grad_norm": 0.07510018348693848, + "learning_rate": 1.1255239823439663e-05, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18040 + }, + { + "epoch": 5.251672970613908, + "flow_matching_loss": 0.011614539101719857, + "grad_norm": 0.08231355994939804, + "learning_rate": 1.1239230958905652e-05, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18050 + }, + { + "epoch": 5.254582484725051, + "flow_matching_loss": 0.011806299258023501, + "grad_norm": 0.07514537125825882, + "learning_rate": 1.1223226664060641e-05, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18060 + }, + { + "epoch": 5.257491998836194, + "flow_matching_loss": 0.009994440106675028, + "grad_norm": 0.06433602422475815, + "learning_rate": 1.1207226958351362e-05, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18070 + }, + { + "epoch": 5.260401512947338, + "flow_matching_loss": 0.010489428788423539, + "grad_norm": 0.0635056346654892, + "learning_rate": 1.1191231861218955e-05, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18080 + }, + { + "epoch": 5.2633110270584815, + "flow_matching_loss": 0.010933890473097563, + "grad_norm": 0.07518002390861511, + "learning_rate": 1.1175241392098988e-05, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18090 + }, + { + "epoch": 5.266220541169624, + "flow_matching_loss": 0.011207816563546658, + "grad_norm": 0.09186786413192749, + "learning_rate": 1.1159255570421376e-05, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18100 + }, + { + "epoch": 5.269130055280768, + "flow_matching_loss": 0.01172122647985816, + "grad_norm": 0.06652490794658661, + "learning_rate": 1.1143274415610408e-05, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18110 + }, + { + "epoch": 5.272039569391912, + "flow_matching_loss": 0.012351293116807938, + "grad_norm": 0.08424998074769974, + "learning_rate": 1.1127297947084692e-05, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18120 + }, + { + "epoch": 5.274949083503055, + "flow_matching_loss": 0.012126821652054786, + "grad_norm": 0.06853222101926804, + "learning_rate": 1.1111326184257135e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18130 + }, + { + "epoch": 5.277858597614198, + "flow_matching_loss": 0.012684888206422329, + "grad_norm": 0.08395498991012573, + "learning_rate": 1.109535914653495e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18140 + }, + { + "epoch": 5.280768111725342, + "flow_matching_loss": 0.011722197476774453, + "grad_norm": 0.07331757247447968, + "learning_rate": 1.1079396853319582e-05, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18150 + }, + { + "epoch": 5.283677625836486, + "flow_matching_loss": 0.011970724258571864, + "grad_norm": 0.08292757719755173, + "learning_rate": 1.1063439324006724e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18160 + }, + { + "epoch": 5.2865871399476285, + "flow_matching_loss": 0.011962595861405134, + "grad_norm": 0.06109929829835892, + "learning_rate": 1.1047486577986285e-05, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18170 + }, + { + "epoch": 5.289496654058772, + "flow_matching_loss": 0.010854368982836603, + "grad_norm": 0.06872623413801193, + "learning_rate": 1.1031538634642353e-05, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18180 + }, + { + "epoch": 5.292406168169916, + "flow_matching_loss": 0.011572402156889438, + "grad_norm": 0.08004187047481537, + "learning_rate": 1.1015595513353172e-05, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18190 + }, + { + "epoch": 5.295315682281059, + "flow_matching_loss": 0.012576309219002723, + "grad_norm": 0.08515944331884384, + "learning_rate": 1.0999657233491156e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18200 + }, + { + "epoch": 5.2982251963922025, + "flow_matching_loss": 0.0125924838706851, + "grad_norm": 0.11053749918937683, + "learning_rate": 1.0983723814422808e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18210 + }, + { + "epoch": 5.301134710503346, + "flow_matching_loss": 0.011213307548314334, + "grad_norm": 0.08075515925884247, + "learning_rate": 1.0967795275508735e-05, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18220 + }, + { + "epoch": 5.304044224614489, + "flow_matching_loss": 0.012591231567785143, + "grad_norm": 0.10467074811458588, + "learning_rate": 1.095187163610361e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18230 + }, + { + "epoch": 5.306953738725633, + "flow_matching_loss": 0.011044947290793061, + "grad_norm": 0.08713023364543915, + "learning_rate": 1.0935952915556163e-05, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18240 + }, + { + "epoch": 5.309863252836776, + "flow_matching_loss": 0.012683343328535556, + "grad_norm": 0.07289326936006546, + "learning_rate": 1.0920039133209137e-05, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18250 + }, + { + "epoch": 5.31277276694792, + "flow_matching_loss": 0.01189479986205697, + "grad_norm": 0.07590097188949585, + "learning_rate": 1.0904130308399283e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18260 + }, + { + "epoch": 5.315682281059063, + "flow_matching_loss": 0.012604832462966443, + "grad_norm": 0.06920190155506134, + "learning_rate": 1.0888226460457318e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18270 + }, + { + "epoch": 5.318591795170207, + "flow_matching_loss": 0.011476304661482573, + "grad_norm": 0.07602523267269135, + "learning_rate": 1.087232760870792e-05, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18280 + }, + { + "epoch": 5.32150130928135, + "flow_matching_loss": 0.01129322312772274, + "grad_norm": 0.08746964484453201, + "learning_rate": 1.0856433772469686e-05, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18290 + }, + { + "epoch": 5.324410823392493, + "flow_matching_loss": 0.011626800196245313, + "grad_norm": 0.07271472364664078, + "learning_rate": 1.0840544971055135e-05, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18300 + }, + { + "epoch": 5.327320337503637, + "flow_matching_loss": 0.010057366546243429, + "grad_norm": 0.10160083323717117, + "learning_rate": 1.0824661223770655e-05, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18310 + }, + { + "epoch": 5.330229851614781, + "flow_matching_loss": 0.013961958512663842, + "grad_norm": 0.09330134838819504, + "learning_rate": 1.0808782549916498e-05, + "loss": 0.014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18320 + }, + { + "epoch": 5.333139365725923, + "flow_matching_loss": 0.011931604519486428, + "grad_norm": 0.07204294204711914, + "learning_rate": 1.079290896878675e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18330 + }, + { + "epoch": 5.336048879837067, + "flow_matching_loss": 0.012255429569631816, + "grad_norm": 0.07761840522289276, + "learning_rate": 1.07770404996693e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18340 + }, + { + "epoch": 5.338958393948211, + "flow_matching_loss": 0.010194361489266158, + "grad_norm": 0.17250646650791168, + "learning_rate": 1.0761177161845848e-05, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18350 + }, + { + "epoch": 5.3418679080593545, + "flow_matching_loss": 0.012601855304092168, + "grad_norm": 0.06943673640489578, + "learning_rate": 1.0745318974591841e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18360 + }, + { + "epoch": 5.344777422170497, + "flow_matching_loss": 0.01220246236771345, + "grad_norm": 0.08148182183504105, + "learning_rate": 1.0729465957176469e-05, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18370 + }, + { + "epoch": 5.347686936281641, + "flow_matching_loss": 0.010931172361597418, + "grad_norm": 0.0642298012971878, + "learning_rate": 1.0713618128862642e-05, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18380 + }, + { + "epoch": 5.350596450392785, + "flow_matching_loss": 0.011793471779674292, + "grad_norm": 0.06026188284158707, + "learning_rate": 1.0697775508906969e-05, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18390 + }, + { + "epoch": 5.353505964503928, + "flow_matching_loss": 0.011361818807199598, + "grad_norm": 0.05399364233016968, + "learning_rate": 1.068193811655972e-05, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18400 + }, + { + "epoch": 5.356415478615071, + "flow_matching_loss": 0.012093829456716776, + "grad_norm": 0.07873965054750443, + "learning_rate": 1.0666105971064835e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18410 + }, + { + "epoch": 5.359324992726215, + "flow_matching_loss": 0.011521439254283904, + "grad_norm": 0.07970715314149857, + "learning_rate": 1.0650279091659852e-05, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18420 + }, + { + "epoch": 5.362234506837358, + "flow_matching_loss": 0.010785855632275344, + "grad_norm": 0.0669216513633728, + "learning_rate": 1.0634457497575926e-05, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18430 + }, + { + "epoch": 5.3651440209485015, + "flow_matching_loss": 0.011806409619748593, + "grad_norm": 0.07279303669929504, + "learning_rate": 1.061864120803778e-05, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18440 + }, + { + "epoch": 5.368053535059645, + "flow_matching_loss": 0.011535592610016465, + "grad_norm": 0.0693967193365097, + "learning_rate": 1.0602830242263701e-05, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18450 + }, + { + "epoch": 5.370963049170788, + "flow_matching_loss": 0.011305190389975905, + "grad_norm": 0.07558847963809967, + "learning_rate": 1.0587024619465506e-05, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18460 + }, + { + "epoch": 5.373872563281932, + "flow_matching_loss": 0.01125871310941875, + "grad_norm": 0.057846374809741974, + "learning_rate": 1.0571224358848517e-05, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18470 + }, + { + "epoch": 5.3767820773930755, + "flow_matching_loss": 0.012297738390043379, + "grad_norm": 0.0651174932718277, + "learning_rate": 1.0555429479611537e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18480 + }, + { + "epoch": 5.379691591504219, + "flow_matching_loss": 0.011461163870990276, + "grad_norm": 0.07161073386669159, + "learning_rate": 1.0539640000946835e-05, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18490 + }, + { + "epoch": 5.382601105615362, + "flow_matching_loss": 0.010299050621688366, + "grad_norm": 0.07844138145446777, + "learning_rate": 1.0523855942040111e-05, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18500 + }, + { + "epoch": 5.385510619726506, + "flow_matching_loss": 0.011108068423345685, + "grad_norm": 0.06860329210758209, + "learning_rate": 1.0508077322070496e-05, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18510 + }, + { + "epoch": 5.388420133837649, + "flow_matching_loss": 0.011680190498009325, + "grad_norm": 0.10027361661195755, + "learning_rate": 1.0492304160210495e-05, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18520 + }, + { + "epoch": 5.391329647948792, + "flow_matching_loss": 0.012750179599970579, + "grad_norm": 0.08211226016283035, + "learning_rate": 1.0476536475625985e-05, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18530 + }, + { + "epoch": 5.394239162059936, + "flow_matching_loss": 0.013208509795367718, + "grad_norm": 0.08732141554355621, + "learning_rate": 1.0460774287476195e-05, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18540 + }, + { + "epoch": 5.39714867617108, + "flow_matching_loss": 0.010924434754997492, + "grad_norm": 0.05572027340531349, + "learning_rate": 1.044501761491366e-05, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18550 + }, + { + "epoch": 5.400058190282223, + "flow_matching_loss": 0.01038495423272252, + "grad_norm": 0.06920024007558823, + "learning_rate": 1.0429266477084235e-05, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18560 + }, + { + "epoch": 5.402967704393366, + "flow_matching_loss": 0.012635991536080837, + "grad_norm": 0.08104867488145828, + "learning_rate": 1.0413520893127034e-05, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18570 + }, + { + "epoch": 5.40587721850451, + "flow_matching_loss": 0.010294884350150824, + "grad_norm": 0.06467127799987793, + "learning_rate": 1.0397780882174424e-05, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18580 + }, + { + "epoch": 5.408786732615654, + "flow_matching_loss": 0.01192594701424241, + "grad_norm": 0.07633065432310104, + "learning_rate": 1.0382046463351998e-05, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18590 + }, + { + "epoch": 5.411696246726796, + "flow_matching_loss": 0.012864426244050264, + "grad_norm": 0.07439026236534119, + "learning_rate": 1.0366317655778568e-05, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18600 + }, + { + "epoch": 5.41460576083794, + "flow_matching_loss": 0.01086192955262959, + "grad_norm": 0.11956924945116043, + "learning_rate": 1.035059447856611e-05, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18610 + }, + { + "epoch": 5.417515274949084, + "flow_matching_loss": 0.011548298504203558, + "grad_norm": 0.07354191690683365, + "learning_rate": 1.0334876950819777e-05, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18620 + }, + { + "epoch": 5.420424789060227, + "flow_matching_loss": 0.011371590383350848, + "grad_norm": 0.06958974152803421, + "learning_rate": 1.0319165091637837e-05, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18630 + }, + { + "epoch": 5.42333430317137, + "flow_matching_loss": 0.011797898542135954, + "grad_norm": 0.07793737202882767, + "learning_rate": 1.0303458920111686e-05, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18640 + }, + { + "epoch": 5.426243817282514, + "flow_matching_loss": 0.012525248574092984, + "grad_norm": 0.053296808153390884, + "learning_rate": 1.0287758455325803e-05, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18650 + }, + { + "epoch": 5.429153331393657, + "flow_matching_loss": 0.011618616711348295, + "grad_norm": 0.07424590736627579, + "learning_rate": 1.0272063716357729e-05, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18660 + }, + { + "epoch": 5.432062845504801, + "flow_matching_loss": 0.010834738425910474, + "grad_norm": 0.06820913404226303, + "learning_rate": 1.0256374722278057e-05, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18670 + }, + { + "epoch": 5.434972359615944, + "flow_matching_loss": 0.012085391022264958, + "grad_norm": 0.07360082119703293, + "learning_rate": 1.0240691492150388e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18680 + }, + { + "epoch": 5.437881873727088, + "flow_matching_loss": 0.010586920008063317, + "grad_norm": 0.0621594674885273, + "learning_rate": 1.0225014045031334e-05, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18690 + }, + { + "epoch": 5.440791387838231, + "flow_matching_loss": 0.010883260471746325, + "grad_norm": 0.07514277845621109, + "learning_rate": 1.0209342399970468e-05, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18700 + }, + { + "epoch": 5.4437009019493745, + "flow_matching_loss": 0.011308075953274965, + "grad_norm": 0.06492136418819427, + "learning_rate": 1.019367657601031e-05, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18710 + }, + { + "epoch": 5.446610416060518, + "flow_matching_loss": 0.0111792903393507, + "grad_norm": 0.09121126681566238, + "learning_rate": 1.0178016592186323e-05, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18720 + }, + { + "epoch": 5.449519930171661, + "flow_matching_loss": 0.013037227932363749, + "grad_norm": 0.06699007004499435, + "learning_rate": 1.0162362467526861e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18730 + }, + { + "epoch": 5.452429444282805, + "flow_matching_loss": 0.012181120552122593, + "grad_norm": 0.06842909753322601, + "learning_rate": 1.0146714221053159e-05, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18740 + }, + { + "epoch": 5.4553389583939484, + "flow_matching_loss": 0.012268290249630808, + "grad_norm": 0.1367279291152954, + "learning_rate": 1.0131071871779314e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18750 + }, + { + "epoch": 5.458248472505091, + "flow_matching_loss": 0.013034654129296542, + "grad_norm": 0.06908149272203445, + "learning_rate": 1.0115435438712249e-05, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18760 + }, + { + "epoch": 5.461157986616235, + "flow_matching_loss": 0.013366857916116715, + "grad_norm": 0.09550720453262329, + "learning_rate": 1.0099804940851716e-05, + "loss": 0.0134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18770 + }, + { + "epoch": 5.464067500727379, + "flow_matching_loss": 0.011105112405493856, + "grad_norm": 0.07526471465826035, + "learning_rate": 1.0084180397190238e-05, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18780 + }, + { + "epoch": 5.4669770148385215, + "flow_matching_loss": 0.012297646701335907, + "grad_norm": 0.08188744634389877, + "learning_rate": 1.006856182671311e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18790 + }, + { + "epoch": 5.469886528949665, + "flow_matching_loss": 0.012259099632501602, + "grad_norm": 0.10960537195205688, + "learning_rate": 1.0052949248398367e-05, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18800 + }, + { + "epoch": 5.472796043060809, + "flow_matching_loss": 0.01212914977222681, + "grad_norm": 0.07234770059585571, + "learning_rate": 1.003734268121676e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18810 + }, + { + "epoch": 5.475705557171953, + "flow_matching_loss": 0.011520995199680329, + "grad_norm": 0.08045609295368195, + "learning_rate": 1.0021742144131744e-05, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18820 + }, + { + "epoch": 5.478615071283095, + "flow_matching_loss": 0.01214699842967093, + "grad_norm": 0.09413844347000122, + "learning_rate": 1.000614765609944e-05, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18830 + }, + { + "epoch": 5.481524585394239, + "flow_matching_loss": 0.010825816122815014, + "grad_norm": 0.0857827216386795, + "learning_rate": 9.990559236068628e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18840 + }, + { + "epoch": 5.484434099505383, + "flow_matching_loss": 0.011232878779992461, + "grad_norm": 0.061456479132175446, + "learning_rate": 9.974976902980703e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18850 + }, + { + "epoch": 5.487343613616526, + "flow_matching_loss": 0.01247717160731554, + "grad_norm": 0.07643360644578934, + "learning_rate": 9.959400675769668e-06, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18860 + }, + { + "epoch": 5.490253127727669, + "flow_matching_loss": 0.012456386722624302, + "grad_norm": 0.08476980030536652, + "learning_rate": 9.943830573362106e-06, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18870 + }, + { + "epoch": 5.493162641838813, + "flow_matching_loss": 0.013100736262276768, + "grad_norm": 0.08060884475708008, + "learning_rate": 9.928266614677168e-06, + "loss": 0.0131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18880 + }, + { + "epoch": 5.496072155949957, + "flow_matching_loss": 0.011740508675575256, + "grad_norm": 0.07038524746894836, + "learning_rate": 9.912708818626524e-06, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18890 + }, + { + "epoch": 5.4989816700611, + "flow_matching_loss": 0.011188269220292568, + "grad_norm": 0.08043045550584793, + "learning_rate": 9.897157204114369e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18900 + }, + { + "epoch": 5.501891184172243, + "flow_matching_loss": 0.01084269555285573, + "grad_norm": 0.06532034277915955, + "learning_rate": 9.881611790037375e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18910 + }, + { + "epoch": 5.504800698283387, + "flow_matching_loss": 0.011944981291890144, + "grad_norm": 0.08853818476200104, + "learning_rate": 9.866072595284684e-06, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18920 + }, + { + "epoch": 5.50771021239453, + "flow_matching_loss": 0.011664533149451017, + "grad_norm": 0.05814778432250023, + "learning_rate": 9.850539638737894e-06, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18930 + }, + { + "epoch": 5.5106197265056736, + "flow_matching_loss": 0.01303050359711051, + "grad_norm": 0.09376116842031479, + "learning_rate": 9.835012939271006e-06, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18940 + }, + { + "epoch": 5.513529240616817, + "flow_matching_loss": 0.010886568296700715, + "grad_norm": 0.08036026358604431, + "learning_rate": 9.819492515750418e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18950 + }, + { + "epoch": 5.51643875472796, + "flow_matching_loss": 0.010884758550673723, + "grad_norm": 0.08272328972816467, + "learning_rate": 9.803978387034918e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18960 + }, + { + "epoch": 5.519348268839104, + "flow_matching_loss": 0.014130535069853067, + "grad_norm": 0.07905006408691406, + "learning_rate": 9.788470571975624e-06, + "loss": 0.0141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18970 + }, + { + "epoch": 5.5222577829502475, + "flow_matching_loss": 0.010905354842543602, + "grad_norm": 0.06435848027467728, + "learning_rate": 9.772969089416008e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18980 + }, + { + "epoch": 5.52516729706139, + "flow_matching_loss": 0.011284787580370903, + "grad_norm": 0.05433003604412079, + "learning_rate": 9.757473958191822e-06, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18990 + }, + { + "epoch": 5.528076811172534, + "flow_matching_loss": 0.010640338063240051, + "grad_norm": 0.0661027804017067, + "learning_rate": 9.741985197131118e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19000 + }, + { + "epoch": 5.530986325283678, + "flow_matching_loss": 0.009868400637060404, + "grad_norm": 0.059163644909858704, + "learning_rate": 9.726502825054195e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19010 + }, + { + "epoch": 5.533895839394821, + "flow_matching_loss": 0.01030971584841609, + "grad_norm": 0.08164586871862411, + "learning_rate": 9.711026860773595e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19020 + }, + { + "epoch": 5.536805353505964, + "flow_matching_loss": 0.010863913036882878, + "grad_norm": 0.07804456353187561, + "learning_rate": 9.69555732309408e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19030 + }, + { + "epoch": 5.539714867617108, + "flow_matching_loss": 0.012613009568303824, + "grad_norm": 0.0977586880326271, + "learning_rate": 9.68009423081259e-06, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19040 + }, + { + "epoch": 5.542624381728252, + "flow_matching_loss": 0.01124730696901679, + "grad_norm": 0.07436170428991318, + "learning_rate": 9.664637602718249e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19050 + }, + { + "epoch": 5.5455338958393945, + "flow_matching_loss": 0.010952197108417749, + "grad_norm": 0.07767249643802643, + "learning_rate": 9.649187457592313e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19060 + }, + { + "epoch": 5.548443409950538, + "flow_matching_loss": 0.011083949776366352, + "grad_norm": 0.07166114449501038, + "learning_rate": 9.63374381420816e-06, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19070 + }, + { + "epoch": 5.551352924061682, + "flow_matching_loss": 0.011910467920824886, + "grad_norm": 0.0703168734908104, + "learning_rate": 9.618306691331279e-06, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19080 + }, + { + "epoch": 5.554262438172826, + "flow_matching_loss": 0.012333684507757426, + "grad_norm": 0.07082399725914001, + "learning_rate": 9.602876107719228e-06, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19090 + }, + { + "epoch": 5.557171952283968, + "flow_matching_loss": 0.011330625368282199, + "grad_norm": 0.07121128588914871, + "learning_rate": 9.587452082121621e-06, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19100 + }, + { + "epoch": 5.560081466395112, + "flow_matching_loss": 0.011623441614210606, + "grad_norm": 0.07441448420286179, + "learning_rate": 9.57203463328011e-06, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19110 + }, + { + "epoch": 5.562990980506255, + "flow_matching_loss": 0.010829730005934835, + "grad_norm": 0.0928175151348114, + "learning_rate": 9.556623779928341e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19120 + }, + { + "epoch": 5.565900494617399, + "flow_matching_loss": 0.010288135893642903, + "grad_norm": 0.06904781609773636, + "learning_rate": 9.541219540791949e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19130 + }, + { + "epoch": 5.568810008728542, + "flow_matching_loss": 0.012010021787136792, + "grad_norm": 0.059832535684108734, + "learning_rate": 9.525821934588551e-06, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19140 + }, + { + "epoch": 5.571719522839686, + "flow_matching_loss": 0.011647351132705808, + "grad_norm": 0.08134745061397552, + "learning_rate": 9.510430980027684e-06, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19150 + }, + { + "epoch": 5.574629036950829, + "flow_matching_loss": 0.011056415271013975, + "grad_norm": 0.04812154546380043, + "learning_rate": 9.495046695810809e-06, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19160 + }, + { + "epoch": 5.577538551061973, + "flow_matching_loss": 0.011389570543542504, + "grad_norm": 0.08642741292715073, + "learning_rate": 9.47966910063128e-06, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19170 + }, + { + "epoch": 5.580448065173116, + "flow_matching_loss": 0.011468818690627814, + "grad_norm": 0.09058577567338943, + "learning_rate": 9.464298213174325e-06, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19180 + }, + { + "epoch": 5.583357579284259, + "flow_matching_loss": 0.010112845385447145, + "grad_norm": 0.05700043588876724, + "learning_rate": 9.448934052117024e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19190 + }, + { + "epoch": 5.586267093395403, + "flow_matching_loss": 0.01170550538226962, + "grad_norm": 0.07695615291595459, + "learning_rate": 9.433576636128284e-06, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19200 + }, + { + "epoch": 5.5891766075065465, + "flow_matching_loss": 0.010654229205101728, + "grad_norm": 0.09646564722061157, + "learning_rate": 9.41822598386881e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19210 + }, + { + "epoch": 5.59208612161769, + "flow_matching_loss": 0.012464951304718853, + "grad_norm": 0.061472173780202866, + "learning_rate": 9.402882113991094e-06, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19220 + }, + { + "epoch": 5.594995635728833, + "flow_matching_loss": 0.012614634819328785, + "grad_norm": 0.08319879323244095, + "learning_rate": 9.387545045139381e-06, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19230 + }, + { + "epoch": 5.597905149839977, + "flow_matching_loss": 0.01068396382033825, + "grad_norm": 0.06771589070558548, + "learning_rate": 9.372214795949655e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19240 + }, + { + "epoch": 5.6008146639511205, + "flow_matching_loss": 0.012142210081219674, + "grad_norm": 0.08157192915678024, + "learning_rate": 9.356891385049618e-06, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19250 + }, + { + "epoch": 5.603724178062263, + "flow_matching_loss": 0.011899612983688712, + "grad_norm": 0.08263374865055084, + "learning_rate": 9.341574831058664e-06, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19260 + }, + { + "epoch": 5.606633692173407, + "flow_matching_loss": 0.011175725143402815, + "grad_norm": 0.06920837610960007, + "learning_rate": 9.326265152587844e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19270 + }, + { + "epoch": 5.609543206284551, + "flow_matching_loss": 0.011165313608944415, + "grad_norm": 0.07952592521905899, + "learning_rate": 9.310962368239861e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19280 + }, + { + "epoch": 5.6124527203956935, + "flow_matching_loss": 0.011944698449224234, + "grad_norm": 0.1002712994813919, + "learning_rate": 9.295666496609036e-06, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19290 + }, + { + "epoch": 5.615362234506837, + "flow_matching_loss": 0.01104344828054309, + "grad_norm": 0.09078894555568695, + "learning_rate": 9.280377556281306e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19300 + }, + { + "epoch": 5.618271748617981, + "flow_matching_loss": 0.013566870335489511, + "grad_norm": 0.10232536494731903, + "learning_rate": 9.265095565834172e-06, + "loss": 0.0136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19310 + }, + { + "epoch": 5.621181262729124, + "flow_matching_loss": 0.009896254632622004, + "grad_norm": 0.10680439323186874, + "learning_rate": 9.249820543836688e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19320 + }, + { + "epoch": 5.6240907768402675, + "flow_matching_loss": 0.011762357223778963, + "grad_norm": 0.0725361630320549, + "learning_rate": 9.234552508849454e-06, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19330 + }, + { + "epoch": 5.627000290951411, + "flow_matching_loss": 0.010983104165643454, + "grad_norm": 0.06744968146085739, + "learning_rate": 9.21929147942456e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19340 + }, + { + "epoch": 5.629909805062555, + "flow_matching_loss": 0.010454003373160959, + "grad_norm": 0.06561741232872009, + "learning_rate": 9.204037474105611e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19350 + }, + { + "epoch": 5.632819319173698, + "flow_matching_loss": 0.011413695523515344, + "grad_norm": 0.058325909078121185, + "learning_rate": 9.188790511427653e-06, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19360 + }, + { + "epoch": 5.635728833284841, + "flow_matching_loss": 0.011141166463494301, + "grad_norm": 0.07359929382801056, + "learning_rate": 9.173550609917186e-06, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19370 + }, + { + "epoch": 5.638638347395985, + "flow_matching_loss": 0.012361673079431057, + "grad_norm": 0.08007652312517166, + "learning_rate": 9.158317788092124e-06, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19380 + }, + { + "epoch": 5.641547861507128, + "flow_matching_loss": 0.011356354597955943, + "grad_norm": 0.05834180489182472, + "learning_rate": 9.143092064461783e-06, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19390 + }, + { + "epoch": 5.644457375618272, + "flow_matching_loss": 0.01026191576384008, + "grad_norm": 0.04776396229863167, + "learning_rate": 9.127873457526855e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19400 + }, + { + "epoch": 5.647366889729415, + "flow_matching_loss": 0.011236571054905654, + "grad_norm": 0.061666905879974365, + "learning_rate": 9.112661985779382e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19410 + }, + { + "epoch": 5.650276403840559, + "flow_matching_loss": 0.012604963686317205, + "grad_norm": 0.08646787703037262, + "learning_rate": 9.09745766770274e-06, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19420 + }, + { + "epoch": 5.653185917951702, + "flow_matching_loss": 0.011574326222762466, + "grad_norm": 0.07236456871032715, + "learning_rate": 9.082260521771601e-06, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19430 + }, + { + "epoch": 5.656095432062846, + "flow_matching_loss": 0.011506495578214527, + "grad_norm": 0.16817724704742432, + "learning_rate": 9.067070566451936e-06, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19440 + }, + { + "epoch": 5.659004946173989, + "flow_matching_loss": 0.012144005624577403, + "grad_norm": 0.08417174965143204, + "learning_rate": 9.051887820200966e-06, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19450 + }, + { + "epoch": 5.661914460285132, + "flow_matching_loss": 0.010614433884620666, + "grad_norm": 0.07051511853933334, + "learning_rate": 9.03671230146717e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19460 + }, + { + "epoch": 5.664823974396276, + "flow_matching_loss": 0.011232553701847791, + "grad_norm": 0.06401766091585159, + "learning_rate": 9.021544028690228e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19470 + }, + { + "epoch": 5.6677334885074195, + "flow_matching_loss": 0.010852543637156487, + "grad_norm": 0.08051423728466034, + "learning_rate": 9.006383020301024e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19480 + }, + { + "epoch": 5.670643002618562, + "flow_matching_loss": 0.010283898888155818, + "grad_norm": 0.06778524816036224, + "learning_rate": 8.991229294721613e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19490 + }, + { + "epoch": 5.673552516729706, + "flow_matching_loss": 0.01221234854310751, + "grad_norm": 0.06862425804138184, + "learning_rate": 8.976082870365191e-06, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19500 + }, + { + "epoch": 5.67646203084085, + "flow_matching_loss": 0.010388723388314248, + "grad_norm": 0.059825073927640915, + "learning_rate": 8.960943765636105e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19510 + }, + { + "epoch": 5.679371544951993, + "flow_matching_loss": 0.010819693095982075, + "grad_norm": 0.06469836086034775, + "learning_rate": 8.945811998929792e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19520 + }, + { + "epoch": 5.682281059063136, + "flow_matching_loss": 0.010455615259706974, + "grad_norm": 0.06503365188837051, + "learning_rate": 8.93068758863277e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19530 + }, + { + "epoch": 5.68519057317428, + "flow_matching_loss": 0.010795040335506201, + "grad_norm": 0.07059607654809952, + "learning_rate": 8.915570553122628e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19540 + }, + { + "epoch": 5.688100087285424, + "flow_matching_loss": 0.010589738143607974, + "grad_norm": 0.05741535499691963, + "learning_rate": 8.900460910767982e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19550 + }, + { + "epoch": 5.6910096013965665, + "flow_matching_loss": 0.011015067528933286, + "grad_norm": 0.1036759689450264, + "learning_rate": 8.885358679928486e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19560 + }, + { + "epoch": 5.69391911550771, + "flow_matching_loss": 0.010016590403392912, + "grad_norm": 0.05179066210985184, + "learning_rate": 8.870263878954767e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19570 + }, + { + "epoch": 5.696828629618854, + "flow_matching_loss": 0.011242946051061154, + "grad_norm": 0.08664411306381226, + "learning_rate": 8.855176526188428e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19580 + }, + { + "epoch": 5.699738143729997, + "flow_matching_loss": 0.011486923042684793, + "grad_norm": 0.11537020653486252, + "learning_rate": 8.840096639962033e-06, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19590 + }, + { + "epoch": 5.7026476578411405, + "flow_matching_loss": 0.01098584714345634, + "grad_norm": 0.06616611033678055, + "learning_rate": 8.825024238599056e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19600 + }, + { + "epoch": 5.705557171952284, + "flow_matching_loss": 0.011968967132270336, + "grad_norm": 0.06699541956186295, + "learning_rate": 8.809959340413891e-06, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19610 + }, + { + "epoch": 5.708466686063428, + "flow_matching_loss": 0.01124974680133164, + "grad_norm": 0.0751543715596199, + "learning_rate": 8.79490196371181e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19620 + }, + { + "epoch": 5.711376200174571, + "flow_matching_loss": 0.01184061700478196, + "grad_norm": 0.09834926575422287, + "learning_rate": 8.779852126788949e-06, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19630 + }, + { + "epoch": 5.714285714285714, + "flow_matching_loss": 0.012009629420936108, + "grad_norm": 0.0682937428355217, + "learning_rate": 8.764809847932277e-06, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19640 + }, + { + "epoch": 5.717195228396857, + "flow_matching_loss": 0.01088506644591689, + "grad_norm": 0.0800178050994873, + "learning_rate": 8.749775145419577e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19650 + }, + { + "epoch": 5.720104742508001, + "flow_matching_loss": 0.013342038355767727, + "grad_norm": 0.07751539349555969, + "learning_rate": 8.734748037519427e-06, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19660 + }, + { + "epoch": 5.723014256619145, + "flow_matching_loss": 0.011605213675647975, + "grad_norm": 0.06714365631341934, + "learning_rate": 8.719728542491195e-06, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19670 + }, + { + "epoch": 5.725923770730288, + "flow_matching_loss": 0.012003041803836823, + "grad_norm": 0.06746961921453476, + "learning_rate": 8.704716678584976e-06, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19680 + }, + { + "epoch": 5.728833284841431, + "flow_matching_loss": 0.012189948279410601, + "grad_norm": 0.08432503044605255, + "learning_rate": 8.689712464041596e-06, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19690 + }, + { + "epoch": 5.731742798952575, + "flow_matching_loss": 0.011745250970125198, + "grad_norm": 0.07424993067979813, + "learning_rate": 8.674715917092592e-06, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19700 + }, + { + "epoch": 5.734652313063719, + "flow_matching_loss": 0.011729348823428154, + "grad_norm": 0.07730931788682938, + "learning_rate": 8.65972705596018e-06, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19710 + }, + { + "epoch": 5.737561827174861, + "flow_matching_loss": 0.010874348320066929, + "grad_norm": 0.057908255606889725, + "learning_rate": 8.64474589885725e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19720 + }, + { + "epoch": 5.740471341286005, + "flow_matching_loss": 0.011832883534952997, + "grad_norm": 0.09001870453357697, + "learning_rate": 8.629772463987315e-06, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19730 + }, + { + "epoch": 5.743380855397149, + "flow_matching_loss": 0.010855992417782545, + "grad_norm": 0.08942126482725143, + "learning_rate": 8.614806769544507e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19740 + }, + { + "epoch": 5.7462903695082925, + "flow_matching_loss": 0.009340540552511812, + "grad_norm": 0.08977235853672028, + "learning_rate": 8.59984883371356e-06, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19750 + }, + { + "epoch": 5.749199883619435, + "flow_matching_loss": 0.011183062475174666, + "grad_norm": 0.059828244149684906, + "learning_rate": 8.584898674669772e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19760 + }, + { + "epoch": 5.752109397730579, + "flow_matching_loss": 0.011507123429328203, + "grad_norm": 0.09627729654312134, + "learning_rate": 8.569956310579001e-06, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19770 + }, + { + "epoch": 5.755018911841723, + "flow_matching_loss": 0.011049815267324448, + "grad_norm": 0.07817646116018295, + "learning_rate": 8.555021759597627e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19780 + }, + { + "epoch": 5.757928425952866, + "flow_matching_loss": 0.012948402063921094, + "grad_norm": 0.07850238680839539, + "learning_rate": 8.54009503987253e-06, + "loss": 0.0129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19790 + }, + { + "epoch": 5.760837940064009, + "flow_matching_loss": 0.010007133753970266, + "grad_norm": 0.07961800694465637, + "learning_rate": 8.525176169541095e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19800 + }, + { + "epoch": 5.763747454175153, + "flow_matching_loss": 0.011526440968737006, + "grad_norm": 0.07565491646528244, + "learning_rate": 8.510265166731151e-06, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19810 + }, + { + "epoch": 5.766656968286296, + "flow_matching_loss": 0.011534356651827693, + "grad_norm": 0.08416198939085007, + "learning_rate": 8.495362049560964e-06, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19820 + }, + { + "epoch": 5.7695664823974395, + "flow_matching_loss": 0.011052086483687162, + "grad_norm": 0.08066686242818832, + "learning_rate": 8.480466836139241e-06, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19830 + }, + { + "epoch": 5.772475996508583, + "flow_matching_loss": 0.010839247377589345, + "grad_norm": 0.0814267247915268, + "learning_rate": 8.465579544565066e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19840 + }, + { + "epoch": 5.775385510619726, + "flow_matching_loss": 0.012842402514070272, + "grad_norm": 0.0997985452413559, + "learning_rate": 8.450700192927897e-06, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19850 + }, + { + "epoch": 5.77829502473087, + "flow_matching_loss": 0.010820881556719541, + "grad_norm": 0.0760592371225357, + "learning_rate": 8.435828799307551e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19860 + }, + { + "epoch": 5.7812045388420135, + "flow_matching_loss": 0.010806372668594123, + "grad_norm": 0.07391087710857391, + "learning_rate": 8.420965381774172e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19870 + }, + { + "epoch": 5.784114052953157, + "flow_matching_loss": 0.011398011306300759, + "grad_norm": 0.134784996509552, + "learning_rate": 8.406109958388222e-06, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19880 + }, + { + "epoch": 5.7870235670643, + "flow_matching_loss": 0.01090796971693635, + "grad_norm": 0.06589692085981369, + "learning_rate": 8.391262547200434e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19890 + }, + { + "epoch": 5.789933081175444, + "flow_matching_loss": 0.012390551716089248, + "grad_norm": 0.07478359341621399, + "learning_rate": 8.376423166251816e-06, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19900 + }, + { + "epoch": 5.792842595286587, + "flow_matching_loss": 0.01178338248282671, + "grad_norm": 0.08356941491365433, + "learning_rate": 8.361591833573612e-06, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19910 + }, + { + "epoch": 5.79575210939773, + "flow_matching_loss": 0.011947001703083516, + "grad_norm": 0.07178604602813721, + "learning_rate": 8.346768567187287e-06, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19920 + }, + { + "epoch": 5.798661623508874, + "flow_matching_loss": 0.010911382175981999, + "grad_norm": 0.059174951165914536, + "learning_rate": 8.331953385104516e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19930 + }, + { + "epoch": 5.801571137620018, + "flow_matching_loss": 0.010693568922579288, + "grad_norm": 0.06974352896213531, + "learning_rate": 8.317146305327134e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19940 + }, + { + "epoch": 5.804480651731161, + "flow_matching_loss": 0.01164575289003551, + "grad_norm": 0.06405431777238846, + "learning_rate": 8.302347345847146e-06, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19950 + }, + { + "epoch": 5.807390165842304, + "flow_matching_loss": 0.011915654595941304, + "grad_norm": 0.082095667719841, + "learning_rate": 8.287556524646683e-06, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19960 + }, + { + "epoch": 5.810299679953448, + "flow_matching_loss": 0.010392249515280127, + "grad_norm": 0.06115087494254112, + "learning_rate": 8.272773859697978e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19970 + }, + { + "epoch": 5.813209194064592, + "flow_matching_loss": 0.010640692804008723, + "grad_norm": 0.13063421845436096, + "learning_rate": 8.257999368963374e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19980 + }, + { + "epoch": 5.816118708175734, + "flow_matching_loss": 0.010899664834141731, + "grad_norm": 0.06463823467493057, + "learning_rate": 8.243233070395266e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19990 + }, + { + "epoch": 5.819028222286878, + "flow_matching_loss": 0.01156119778752327, + "grad_norm": 0.07096574455499649, + "learning_rate": 8.2284749819361e-06, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20000 + }, + { + "epoch": 5.821937736398022, + "flow_matching_loss": 0.012178653571754693, + "grad_norm": 0.07939102500677109, + "learning_rate": 8.213725121518346e-06, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20010 + }, + { + "epoch": 5.824847250509165, + "flow_matching_loss": 0.012234797747805715, + "grad_norm": 0.07596246898174286, + "learning_rate": 8.198983507064469e-06, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20020 + }, + { + "epoch": 5.827756764620308, + "flow_matching_loss": 0.012032615672796964, + "grad_norm": 0.07449611276388168, + "learning_rate": 8.18425015648692e-06, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20030 + }, + { + "epoch": 5.830666278731452, + "flow_matching_loss": 0.011346068233251572, + "grad_norm": 0.06979726999998093, + "learning_rate": 8.169525087688122e-06, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20040 + }, + { + "epoch": 5.833575792842595, + "flow_matching_loss": 0.012315630819648505, + "grad_norm": 0.08037898689508438, + "learning_rate": 8.154808318560413e-06, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20050 + }, + { + "epoch": 5.836485306953739, + "flow_matching_loss": 0.010880677541717888, + "grad_norm": 0.06258638948202133, + "learning_rate": 8.140099866986058e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20060 + }, + { + "epoch": 5.839394821064882, + "flow_matching_loss": 0.010187493031844496, + "grad_norm": 0.07000651210546494, + "learning_rate": 8.125399750837202e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20070 + }, + { + "epoch": 5.842304335176026, + "flow_matching_loss": 0.011359978467226028, + "grad_norm": 0.08838701993227005, + "learning_rate": 8.110707987975888e-06, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20080 + }, + { + "epoch": 5.845213849287169, + "flow_matching_loss": 0.012047475902363658, + "grad_norm": 0.07127758115530014, + "learning_rate": 8.096024596253982e-06, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20090 + }, + { + "epoch": 5.8481233633983125, + "flow_matching_loss": 0.010615934804081916, + "grad_norm": 0.10863480716943741, + "learning_rate": 8.081349593513197e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20100 + }, + { + "epoch": 5.851032877509456, + "flow_matching_loss": 0.010698739904910326, + "grad_norm": 0.06919271498918533, + "learning_rate": 8.066682997585045e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20110 + }, + { + "epoch": 5.853942391620599, + "flow_matching_loss": 0.01074387999251485, + "grad_norm": 0.1049172654747963, + "learning_rate": 8.052024826290819e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20120 + }, + { + "epoch": 5.856851905731743, + "flow_matching_loss": 0.01321049123071134, + "grad_norm": 0.07360939681529999, + "learning_rate": 8.037375097441574e-06, + "loss": 0.0132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20130 + }, + { + "epoch": 5.8597614198428865, + "flow_matching_loss": 0.011336944438517094, + "grad_norm": 0.08202734589576721, + "learning_rate": 8.022733828838124e-06, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20140 + }, + { + "epoch": 5.86267093395403, + "flow_matching_loss": 0.012053031427785754, + "grad_norm": 0.0762871503829956, + "learning_rate": 8.008101038270988e-06, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20150 + }, + { + "epoch": 5.865580448065173, + "flow_matching_loss": 0.01188072538934648, + "grad_norm": 0.09545823931694031, + "learning_rate": 7.99347674352038e-06, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20160 + }, + { + "epoch": 5.868489962176317, + "flow_matching_loss": 0.011128063825890423, + "grad_norm": 0.08695333451032639, + "learning_rate": 7.978860962356202e-06, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20170 + }, + { + "epoch": 5.8713994762874595, + "flow_matching_loss": 0.011047853715717792, + "grad_norm": 0.10273630917072296, + "learning_rate": 7.964253712537997e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20180 + }, + { + "epoch": 5.874308990398603, + "flow_matching_loss": 0.01102027427405119, + "grad_norm": 0.07291309535503387, + "learning_rate": 7.949655011814968e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20190 + }, + { + "epoch": 5.877218504509747, + "flow_matching_loss": 0.010640031145885586, + "grad_norm": 0.066339410841465, + "learning_rate": 7.9350648779259e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20200 + }, + { + "epoch": 5.880128018620891, + "flow_matching_loss": 0.011019555432721973, + "grad_norm": 0.07737704366445541, + "learning_rate": 7.92048332859919e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20210 + }, + { + "epoch": 5.8830375327320334, + "flow_matching_loss": 0.01188174420967698, + "grad_norm": 0.07599838078022003, + "learning_rate": 7.90591038155278e-06, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20220 + }, + { + "epoch": 5.885947046843177, + "flow_matching_loss": 0.01021438492462039, + "grad_norm": 0.07128357887268066, + "learning_rate": 7.891346054494194e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20230 + }, + { + "epoch": 5.888856560954321, + "flow_matching_loss": 0.011097271414473652, + "grad_norm": 0.0763610452413559, + "learning_rate": 7.876790365120447e-06, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20240 + }, + { + "epoch": 5.891766075065464, + "flow_matching_loss": 0.01123825558461249, + "grad_norm": 0.1078815683722496, + "learning_rate": 7.862243331118088e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20250 + }, + { + "epoch": 5.894675589176607, + "flow_matching_loss": 0.010783718712627888, + "grad_norm": 0.06757218390703201, + "learning_rate": 7.847704970163127e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20260 + }, + { + "epoch": 5.897585103287751, + "flow_matching_loss": 0.011002293275669216, + "grad_norm": 0.09307634830474854, + "learning_rate": 7.833175299921046e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20270 + }, + { + "epoch": 5.900494617398895, + "flow_matching_loss": 0.010574291599914432, + "grad_norm": 0.07937256246805191, + "learning_rate": 7.81865433804676e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20280 + }, + { + "epoch": 5.903404131510038, + "flow_matching_loss": 0.011194739397615195, + "grad_norm": 0.09514323621988297, + "learning_rate": 7.80414210218461e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20290 + }, + { + "epoch": 5.906313645621181, + "flow_matching_loss": 0.01077665649354458, + "grad_norm": 0.07572256773710251, + "learning_rate": 7.789638609968332e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20300 + }, + { + "epoch": 5.909223159732325, + "flow_matching_loss": 0.012069091480225325, + "grad_norm": 0.05658813938498497, + "learning_rate": 7.775143879021037e-06, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20310 + }, + { + "epoch": 5.912132673843468, + "flow_matching_loss": 0.011520847817882896, + "grad_norm": 0.08322026580572128, + "learning_rate": 7.760657926955187e-06, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20320 + }, + { + "epoch": 5.915042187954612, + "flow_matching_loss": 0.010552659258246421, + "grad_norm": 0.10419327765703201, + "learning_rate": 7.746180771372584e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20330 + }, + { + "epoch": 5.917951702065755, + "flow_matching_loss": 0.010958691406995059, + "grad_norm": 0.07440722733736038, + "learning_rate": 7.731712429864325e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20340 + }, + { + "epoch": 5.920861216176898, + "flow_matching_loss": 0.011033537238836289, + "grad_norm": 0.07654019445180893, + "learning_rate": 7.717252920010828e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20350 + }, + { + "epoch": 5.923770730288042, + "flow_matching_loss": 0.012391898222267628, + "grad_norm": 0.08847720921039581, + "learning_rate": 7.702802259381742e-06, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20360 + }, + { + "epoch": 5.9266802443991855, + "flow_matching_loss": 0.012169025652110577, + "grad_norm": 0.07113845646381378, + "learning_rate": 7.688360465535999e-06, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20370 + }, + { + "epoch": 5.929589758510328, + "flow_matching_loss": 0.011172793991863727, + "grad_norm": 0.07121194899082184, + "learning_rate": 7.673927556021734e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20380 + }, + { + "epoch": 5.932499272621472, + "flow_matching_loss": 0.012459228467196225, + "grad_norm": 0.08170035481452942, + "learning_rate": 7.659503548376284e-06, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20390 + }, + { + "epoch": 5.935408786732616, + "flow_matching_loss": 0.010532141756266355, + "grad_norm": 0.10214098542928696, + "learning_rate": 7.645088460126193e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20400 + }, + { + "epoch": 5.9383183008437594, + "flow_matching_loss": 0.011230035033077002, + "grad_norm": 0.09386662393808365, + "learning_rate": 7.630682308787145e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20410 + }, + { + "epoch": 5.941227814954902, + "flow_matching_loss": 0.011531223449856043, + "grad_norm": 0.07038909941911697, + "learning_rate": 7.6162851118639754e-06, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20420 + }, + { + "epoch": 5.944137329066046, + "flow_matching_loss": 0.010410458128899337, + "grad_norm": 0.0740143209695816, + "learning_rate": 7.601896886850632e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20430 + }, + { + "epoch": 5.94704684317719, + "flow_matching_loss": 0.012359043303877115, + "grad_norm": 0.14804397523403168, + "learning_rate": 7.587517651230168e-06, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20440 + }, + { + "epoch": 5.9499563572883325, + "flow_matching_loss": 0.011768782418221236, + "grad_norm": 0.09739631414413452, + "learning_rate": 7.573147422474702e-06, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20450 + }, + { + "epoch": 5.952865871399476, + "flow_matching_loss": 0.010531966853886843, + "grad_norm": 0.057607073336839676, + "learning_rate": 7.55878621804543e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20460 + }, + { + "epoch": 5.95577538551062, + "flow_matching_loss": 0.011244722362607717, + "grad_norm": 0.06080159917473793, + "learning_rate": 7.544434055392564e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20470 + }, + { + "epoch": 5.958684899621764, + "flow_matching_loss": 0.011070734355598688, + "grad_norm": 0.07381041347980499, + "learning_rate": 7.530090951955334e-06, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20480 + }, + { + "epoch": 5.961594413732906, + "flow_matching_loss": 0.011247471487149597, + "grad_norm": 0.07484675198793411, + "learning_rate": 7.515756925161959e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20490 + }, + { + "epoch": 5.96450392784405, + "flow_matching_loss": 0.01108003156259656, + "grad_norm": 0.12106864899396896, + "learning_rate": 7.501431992429643e-06, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20500 + }, + { + "epoch": 5.967413441955194, + "flow_matching_loss": 0.011164610739797354, + "grad_norm": 0.06545185297727585, + "learning_rate": 7.487116171164519e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20510 + }, + { + "epoch": 5.970322956066337, + "flow_matching_loss": 0.011563959717750549, + "grad_norm": 0.0688479095697403, + "learning_rate": 7.472809478761672e-06, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20520 + }, + { + "epoch": 5.97323247017748, + "flow_matching_loss": 0.010828087478876114, + "grad_norm": 0.07090461254119873, + "learning_rate": 7.4585119326050775e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20530 + }, + { + "epoch": 5.976141984288624, + "flow_matching_loss": 0.01055160965770483, + "grad_norm": 0.08773346245288849, + "learning_rate": 7.444223550067599e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20540 + }, + { + "epoch": 5.979051498399767, + "flow_matching_loss": 0.010193952592089772, + "grad_norm": 0.05737272650003433, + "learning_rate": 7.429944348510964e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20550 + }, + { + "epoch": 5.981961012510911, + "flow_matching_loss": 0.012010680045932531, + "grad_norm": 0.12124756723642349, + "learning_rate": 7.415674345285763e-06, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20560 + }, + { + "epoch": 5.984870526622054, + "flow_matching_loss": 0.010757897235453128, + "grad_norm": 0.08028239011764526, + "learning_rate": 7.401413557731387e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20570 + }, + { + "epoch": 5.987780040733197, + "flow_matching_loss": 0.010165120847523213, + "grad_norm": 0.07422264665365219, + "learning_rate": 7.387162003176037e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20580 + }, + { + "epoch": 5.990689554844341, + "flow_matching_loss": 0.01063588671386242, + "grad_norm": 0.07565312087535858, + "learning_rate": 7.372919698936697e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20590 + }, + { + "epoch": 5.9935990689554846, + "flow_matching_loss": 0.012636345298960804, + "grad_norm": 0.10137388855218887, + "learning_rate": 7.358686662319104e-06, + "loss": 0.0126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20600 + }, + { + "epoch": 5.996508583066628, + "flow_matching_loss": 0.010601635742932557, + "grad_norm": 0.06304995715618134, + "learning_rate": 7.344462910617752e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20610 + }, + { + "epoch": 5.999418097177771, + "flow_matching_loss": 0.010392461251467467, + "grad_norm": 0.06667391210794449, + "learning_rate": 7.330248461115832e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20620 + }, + { + "epoch": 6.002327611288915, + "flow_matching_loss": 0.010663287527859212, + "grad_norm": 0.06637736409902573, + "learning_rate": 7.316043331085242e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20630 + }, + { + "epoch": 6.0052371254000585, + "flow_matching_loss": 0.011702483985573053, + "grad_norm": 0.08889845758676529, + "learning_rate": 7.30184753778655e-06, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20640 + }, + { + "epoch": 6.008146639511201, + "flow_matching_loss": 0.012042763968929648, + "grad_norm": 0.0670565515756607, + "learning_rate": 7.2876610984689935e-06, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20650 + }, + { + "epoch": 6.011056153622345, + "flow_matching_loss": 0.010892810113728047, + "grad_norm": 0.06516526639461517, + "learning_rate": 7.273484030370423e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20660 + }, + { + "epoch": 6.013965667733489, + "flow_matching_loss": 0.011693122051656245, + "grad_norm": 0.08497115224599838, + "learning_rate": 7.259316350717324e-06, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20670 + }, + { + "epoch": 6.0168751818446315, + "flow_matching_loss": 0.010576114524155855, + "grad_norm": 0.06585021317005157, + "learning_rate": 7.245158076724757e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20680 + }, + { + "epoch": 6.019784695955775, + "flow_matching_loss": 0.011085361940786242, + "grad_norm": 0.05636778101325035, + "learning_rate": 7.23100922559636e-06, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20690 + }, + { + "epoch": 6.022694210066919, + "flow_matching_loss": 0.011907130759209395, + "grad_norm": 0.07944143563508987, + "learning_rate": 7.216869814524323e-06, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20700 + }, + { + "epoch": 6.025603724178063, + "flow_matching_loss": 0.011239294987171888, + "grad_norm": 0.06200149282813072, + "learning_rate": 7.202739860689357e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20710 + }, + { + "epoch": 6.0285132382892055, + "flow_matching_loss": 0.010258430847898126, + "grad_norm": 0.06713312119245529, + "learning_rate": 7.188619381260698e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20720 + }, + { + "epoch": 6.031422752400349, + "flow_matching_loss": 0.011400162428617477, + "grad_norm": 0.07403846085071564, + "learning_rate": 7.174508393396057e-06, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20730 + }, + { + "epoch": 6.034332266511493, + "flow_matching_loss": 0.011106984969228507, + "grad_norm": 0.08690936863422394, + "learning_rate": 7.160406914241615e-06, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20740 + }, + { + "epoch": 6.037241780622636, + "flow_matching_loss": 0.010485982894897461, + "grad_norm": 0.0749373808503151, + "learning_rate": 7.146314960931996e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20750 + }, + { + "epoch": 6.040151294733779, + "flow_matching_loss": 0.010782324057072402, + "grad_norm": 0.07682894915342331, + "learning_rate": 7.132232550590247e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20760 + }, + { + "epoch": 6.043060808844923, + "flow_matching_loss": 0.011183286830782891, + "grad_norm": 0.06427881866693497, + "learning_rate": 7.118159700327839e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20770 + }, + { + "epoch": 6.045970322956066, + "flow_matching_loss": 0.010873616393655538, + "grad_norm": 0.07115310430526733, + "learning_rate": 7.104096427244603e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20780 + }, + { + "epoch": 6.04887983706721, + "flow_matching_loss": 0.010959853790700436, + "grad_norm": 0.08635660260915756, + "learning_rate": 7.090042748428737e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20790 + }, + { + "epoch": 6.051789351178353, + "flow_matching_loss": 0.010358949657529593, + "grad_norm": 0.1394442766904831, + "learning_rate": 7.075998680956797e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20800 + }, + { + "epoch": 6.054698865289497, + "flow_matching_loss": 0.009894221276044845, + "grad_norm": 0.06864023953676224, + "learning_rate": 7.061964241893639e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20810 + }, + { + "epoch": 6.05760837940064, + "flow_matching_loss": 0.011982854176312684, + "grad_norm": 0.07662144303321838, + "learning_rate": 7.047939448292442e-06, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20820 + }, + { + "epoch": 6.060517893511784, + "flow_matching_loss": 0.01041275099851191, + "grad_norm": 0.08358865976333618, + "learning_rate": 7.033924317194644e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20830 + }, + { + "epoch": 6.063427407622927, + "flow_matching_loss": 0.010910544404760003, + "grad_norm": 0.06877101212739944, + "learning_rate": 7.0199188656299524e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20840 + }, + { + "epoch": 6.06633692173407, + "flow_matching_loss": 0.010248178616166115, + "grad_norm": 0.09128780663013458, + "learning_rate": 7.005923110616313e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20850 + }, + { + "epoch": 6.069246435845214, + "flow_matching_loss": 0.010862370580434799, + "grad_norm": 0.08068026602268219, + "learning_rate": 6.9919370691598855e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20860 + }, + { + "epoch": 6.0721559499563575, + "flow_matching_loss": 0.010383547190576792, + "grad_norm": 0.07767166942358017, + "learning_rate": 6.977960758255024e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20870 + }, + { + "epoch": 6.0750654640675, + "flow_matching_loss": 0.010849595721811055, + "grad_norm": 0.07012787461280823, + "learning_rate": 6.963994194884276e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20880 + }, + { + "epoch": 6.077974978178644, + "flow_matching_loss": 0.011930942069739103, + "grad_norm": 0.08078636974096298, + "learning_rate": 6.950037396018328e-06, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20890 + }, + { + "epoch": 6.080884492289788, + "flow_matching_loss": 0.011428523808717728, + "grad_norm": 0.08088317513465881, + "learning_rate": 6.9360903786160056e-06, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20900 + }, + { + "epoch": 6.083794006400931, + "flow_matching_loss": 0.010337079828605055, + "grad_norm": 0.06802760064601898, + "learning_rate": 6.9221531596242485e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20910 + }, + { + "epoch": 6.086703520512074, + "flow_matching_loss": 0.011455736542120575, + "grad_norm": 0.05675613135099411, + "learning_rate": 6.908225755978089e-06, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20920 + }, + { + "epoch": 6.089613034623218, + "flow_matching_loss": 0.010576165094971657, + "grad_norm": 0.0753822922706604, + "learning_rate": 6.8943081846006466e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20930 + }, + { + "epoch": 6.092522548734362, + "flow_matching_loss": 0.011610766779631377, + "grad_norm": 0.10630965232849121, + "learning_rate": 6.88040046240307e-06, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20940 + }, + { + "epoch": 6.0954320628455045, + "flow_matching_loss": 0.011019970709457993, + "grad_norm": 0.10661514848470688, + "learning_rate": 6.866502606284566e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20950 + }, + { + "epoch": 6.098341576956648, + "flow_matching_loss": 0.011172134475782514, + "grad_norm": 0.06322585046291351, + "learning_rate": 6.852614633132335e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20960 + }, + { + "epoch": 6.101251091067792, + "flow_matching_loss": 0.011362717673182487, + "grad_norm": 0.12778615951538086, + "learning_rate": 6.8387365598215665e-06, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20970 + }, + { + "epoch": 6.104160605178935, + "flow_matching_loss": 0.011613547150045634, + "grad_norm": 0.06719117611646652, + "learning_rate": 6.82486840321544e-06, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20980 + }, + { + "epoch": 6.1070701192900785, + "flow_matching_loss": 0.01233734292909503, + "grad_norm": 0.08278451859951019, + "learning_rate": 6.811010180165069e-06, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20990 + }, + { + "epoch": 6.109979633401222, + "flow_matching_loss": 0.01004662485793233, + "grad_norm": 0.0796784833073616, + "learning_rate": 6.797161907509499e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21000 + }, + { + "epoch": 6.112889147512365, + "flow_matching_loss": 0.010015471186488867, + "grad_norm": 0.07802578806877136, + "learning_rate": 6.783323602075688e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21010 + }, + { + "epoch": 6.115798661623509, + "flow_matching_loss": 0.009400669019669294, + "grad_norm": 0.08061515539884567, + "learning_rate": 6.7694952806784784e-06, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21020 + }, + { + "epoch": 6.118708175734652, + "flow_matching_loss": 0.011088346224278211, + "grad_norm": 0.06804656237363815, + "learning_rate": 6.7556769601205944e-06, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21030 + }, + { + "epoch": 6.121617689845796, + "flow_matching_loss": 0.012988526793196797, + "grad_norm": 0.06754665076732635, + "learning_rate": 6.741868657192594e-06, + "loss": 0.013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21040 + }, + { + "epoch": 6.124527203956939, + "flow_matching_loss": 0.011074876738712191, + "grad_norm": 0.06141876056790352, + "learning_rate": 6.728070388672864e-06, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21050 + }, + { + "epoch": 6.127436718068083, + "flow_matching_loss": 0.010380812454968691, + "grad_norm": 0.07437661290168762, + "learning_rate": 6.714282171327604e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21060 + }, + { + "epoch": 6.130346232179226, + "flow_matching_loss": 0.010679511306807398, + "grad_norm": 0.0665203407406807, + "learning_rate": 6.700504021910791e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21070 + }, + { + "epoch": 6.133255746290369, + "flow_matching_loss": 0.010095590306445957, + "grad_norm": 0.04625299945473671, + "learning_rate": 6.68673595716419e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21080 + }, + { + "epoch": 6.136165260401513, + "flow_matching_loss": 0.011219768971204757, + "grad_norm": 0.0704159066081047, + "learning_rate": 6.6729779938172856e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21090 + }, + { + "epoch": 6.139074774512657, + "flow_matching_loss": 0.010358972707763314, + "grad_norm": 0.08027205616235733, + "learning_rate": 6.659230148587311e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21100 + }, + { + "epoch": 6.141984288623799, + "flow_matching_loss": 0.010228440305218101, + "grad_norm": 0.08345980197191238, + "learning_rate": 6.645492438179188e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21110 + }, + { + "epoch": 6.144893802734943, + "flow_matching_loss": 0.010350653855130076, + "grad_norm": 0.058234237134456635, + "learning_rate": 6.631764879285531e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21120 + }, + { + "epoch": 6.147803316846087, + "flow_matching_loss": 0.010546751832589508, + "grad_norm": 0.07736813277006149, + "learning_rate": 6.618047488586614e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21130 + }, + { + "epoch": 6.1507128309572305, + "flow_matching_loss": 0.01162839364260435, + "grad_norm": 0.08231203258037567, + "learning_rate": 6.604340282750371e-06, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21140 + }, + { + "epoch": 6.153622345068373, + "flow_matching_loss": 0.010685052443295717, + "grad_norm": 0.05953112617135048, + "learning_rate": 6.590643278432342e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21150 + }, + { + "epoch": 6.156531859179517, + "flow_matching_loss": 0.010723003931343556, + "grad_norm": 0.062108300626277924, + "learning_rate": 6.576956492275682e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21160 + }, + { + "epoch": 6.159441373290661, + "flow_matching_loss": 0.011569471098482609, + "grad_norm": 0.07178016006946564, + "learning_rate": 6.563279940911124e-06, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21170 + }, + { + "epoch": 6.162350887401804, + "flow_matching_loss": 0.009567476343363523, + "grad_norm": 0.06373847275972366, + "learning_rate": 6.549613640956963e-06, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21180 + }, + { + "epoch": 6.165260401512947, + "flow_matching_loss": 0.01165118683129549, + "grad_norm": 0.07484735548496246, + "learning_rate": 6.535957609019053e-06, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21190 + }, + { + "epoch": 6.168169915624091, + "flow_matching_loss": 0.011050342442467808, + "grad_norm": 0.06160948425531387, + "learning_rate": 6.5223118616907545e-06, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21200 + }, + { + "epoch": 6.171079429735234, + "flow_matching_loss": 0.011219744896516204, + "grad_norm": 0.09122089296579361, + "learning_rate": 6.508676415552939e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21210 + }, + { + "epoch": 6.1739889438463775, + "flow_matching_loss": 0.012103589251637459, + "grad_norm": 0.0787062868475914, + "learning_rate": 6.495051287173952e-06, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21220 + }, + { + "epoch": 6.176898457957521, + "flow_matching_loss": 0.010589081095531583, + "grad_norm": 0.07680193334817886, + "learning_rate": 6.481436493109621e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21230 + }, + { + "epoch": 6.179807972068665, + "flow_matching_loss": 0.011485238699242473, + "grad_norm": 0.07089872658252716, + "learning_rate": 6.467832049903194e-06, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21240 + }, + { + "epoch": 6.182717486179808, + "flow_matching_loss": 0.012352759204804897, + "grad_norm": 0.08758343756198883, + "learning_rate": 6.454237974085361e-06, + "loss": 0.0124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21250 + }, + { + "epoch": 6.1856270002909515, + "flow_matching_loss": 0.009964472195133568, + "grad_norm": 0.08690211176872253, + "learning_rate": 6.440654282174203e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21260 + }, + { + "epoch": 6.188536514402095, + "flow_matching_loss": 0.011456110421568156, + "grad_norm": 0.0761273130774498, + "learning_rate": 6.427080990675187e-06, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21270 + }, + { + "epoch": 6.191446028513238, + "flow_matching_loss": 0.011597252357751131, + "grad_norm": 0.06558848172426224, + "learning_rate": 6.41351811608114e-06, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21280 + }, + { + "epoch": 6.194355542624382, + "flow_matching_loss": 0.010389401623979212, + "grad_norm": 0.0669664666056633, + "learning_rate": 6.399965674872228e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21290 + }, + { + "epoch": 6.197265056735525, + "flow_matching_loss": 0.011976351588964462, + "grad_norm": 0.07744435220956802, + "learning_rate": 6.386423683515957e-06, + "loss": 0.012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21300 + }, + { + "epoch": 6.200174570846668, + "flow_matching_loss": 0.011225101910531521, + "grad_norm": 0.09681689739227295, + "learning_rate": 6.372892158467119e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21310 + }, + { + "epoch": 6.203084084957812, + "flow_matching_loss": 0.011385663691908122, + "grad_norm": 0.08484099805355072, + "learning_rate": 6.359371116167791e-06, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21320 + }, + { + "epoch": 6.205993599068956, + "flow_matching_loss": 0.010131639149039983, + "grad_norm": 0.07834558188915253, + "learning_rate": 6.345860573047314e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21330 + }, + { + "epoch": 6.208903113180099, + "flow_matching_loss": 0.011933671589940787, + "grad_norm": 0.08157046884298325, + "learning_rate": 6.332360545522268e-06, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21340 + }, + { + "epoch": 6.211812627291242, + "flow_matching_loss": 0.011551132332533598, + "grad_norm": 0.13751651346683502, + "learning_rate": 6.31887104999647e-06, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21350 + }, + { + "epoch": 6.214722141402386, + "flow_matching_loss": 0.009993914514780045, + "grad_norm": 0.07756341248750687, + "learning_rate": 6.305392102860924e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21360 + }, + { + "epoch": 6.21763165551353, + "flow_matching_loss": 0.010253155464306473, + "grad_norm": 0.0850810706615448, + "learning_rate": 6.291923720493814e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21370 + }, + { + "epoch": 6.220541169624672, + "flow_matching_loss": 0.011219762032851577, + "grad_norm": 0.06890425831079483, + "learning_rate": 6.278465919260507e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21380 + }, + { + "epoch": 6.223450683735816, + "flow_matching_loss": 0.01003874302841723, + "grad_norm": 0.091398686170578, + "learning_rate": 6.26501871551349e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21390 + }, + { + "epoch": 6.22636019784696, + "flow_matching_loss": 0.010506662493571638, + "grad_norm": 0.11380790174007416, + "learning_rate": 6.251582125592394e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21400 + }, + { + "epoch": 6.229269711958103, + "flow_matching_loss": 0.010712612280622124, + "grad_norm": 0.08580267429351807, + "learning_rate": 6.2381561658239386e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21410 + }, + { + "epoch": 6.232179226069246, + "flow_matching_loss": 0.010027607111260295, + "grad_norm": 0.15101276338100433, + "learning_rate": 6.2247408525219276e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21420 + }, + { + "epoch": 6.23508874018039, + "flow_matching_loss": 0.010827125376090407, + "grad_norm": 0.07079869508743286, + "learning_rate": 6.2113362019872335e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21430 + }, + { + "epoch": 6.237998254291533, + "flow_matching_loss": 0.010066774906590581, + "grad_norm": 0.07632629573345184, + "learning_rate": 6.197942230507772e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21440 + }, + { + "epoch": 6.240907768402677, + "flow_matching_loss": 0.011055329535156488, + "grad_norm": 0.0778394564986229, + "learning_rate": 6.184558954358475e-06, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21450 + }, + { + "epoch": 6.24381728251382, + "flow_matching_loss": 0.011319106770679354, + "grad_norm": 0.08335502445697784, + "learning_rate": 6.1711863898012965e-06, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21460 + }, + { + "epoch": 6.246726796624964, + "flow_matching_loss": 0.010425083339214325, + "grad_norm": 0.07916446030139923, + "learning_rate": 6.157824553085155e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21470 + }, + { + "epoch": 6.249636310736107, + "flow_matching_loss": 0.010811488097533583, + "grad_norm": 0.0646044909954071, + "learning_rate": 6.144473460445945e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21480 + }, + { + "epoch": 6.2525458248472505, + "flow_matching_loss": 0.010660661105066537, + "grad_norm": 0.09230724722146988, + "learning_rate": 6.131133128106503e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21490 + }, + { + "epoch": 6.255455338958394, + "flow_matching_loss": 0.010965155530720949, + "grad_norm": 0.07004731148481369, + "learning_rate": 6.117803572276584e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21500 + }, + { + "epoch": 6.258364853069537, + "flow_matching_loss": 0.010591701650992037, + "grad_norm": 0.06602638214826584, + "learning_rate": 6.104484809152866e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21510 + }, + { + "epoch": 6.261274367180681, + "flow_matching_loss": 0.011861305497586728, + "grad_norm": 0.08156034350395203, + "learning_rate": 6.091176854918892e-06, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21520 + }, + { + "epoch": 6.2641838812918245, + "flow_matching_loss": 0.011516683083027602, + "grad_norm": 0.06498407572507858, + "learning_rate": 6.077879725745092e-06, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21530 + }, + { + "epoch": 6.267093395402968, + "flow_matching_loss": 0.011212326446548104, + "grad_norm": 0.05270262062549591, + "learning_rate": 6.064593437788727e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21540 + }, + { + "epoch": 6.270002909514111, + "flow_matching_loss": 0.01022253641858697, + "grad_norm": 0.0756925716996193, + "learning_rate": 6.051318007193881e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21550 + }, + { + "epoch": 6.272912423625255, + "flow_matching_loss": 0.010359746869653463, + "grad_norm": 0.08031808584928513, + "learning_rate": 6.038053450091469e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21560 + }, + { + "epoch": 6.275821937736398, + "flow_matching_loss": 0.011216681264340878, + "grad_norm": 0.06457854062318802, + "learning_rate": 6.0247997825991705e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21570 + }, + { + "epoch": 6.278731451847541, + "flow_matching_loss": 0.010934583097696304, + "grad_norm": 0.08099361509084702, + "learning_rate": 6.011557020821444e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21580 + }, + { + "epoch": 6.281640965958685, + "flow_matching_loss": 0.010817733267322183, + "grad_norm": 0.055122628808021545, + "learning_rate": 5.99832518084949e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21590 + }, + { + "epoch": 6.284550480069829, + "flow_matching_loss": 0.011441019549965858, + "grad_norm": 0.08492949604988098, + "learning_rate": 5.98510427876124e-06, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21600 + }, + { + "epoch": 6.2874599941809715, + "flow_matching_loss": 0.010967879183590412, + "grad_norm": 0.07515063136816025, + "learning_rate": 5.971894330621344e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21610 + }, + { + "epoch": 6.290369508292115, + "flow_matching_loss": 0.012118155788630246, + "grad_norm": 0.0689619705080986, + "learning_rate": 5.958695352481134e-06, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21620 + }, + { + "epoch": 6.293279022403259, + "flow_matching_loss": 0.010037971939891577, + "grad_norm": 0.07190941274166107, + "learning_rate": 5.945507360378608e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21630 + }, + { + "epoch": 6.296188536514402, + "flow_matching_loss": 0.012255936767905951, + "grad_norm": 0.08562084287405014, + "learning_rate": 5.9323303703384225e-06, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21640 + }, + { + "epoch": 6.299098050625545, + "flow_matching_loss": 0.010752036795020103, + "grad_norm": 0.08304533362388611, + "learning_rate": 5.919164398371857e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21650 + }, + { + "epoch": 6.302007564736689, + "flow_matching_loss": 0.01052395598962903, + "grad_norm": 0.0763356015086174, + "learning_rate": 5.906009460476818e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21660 + }, + { + "epoch": 6.304917078847833, + "flow_matching_loss": 0.011372087895870209, + "grad_norm": 0.08185163140296936, + "learning_rate": 5.892865572637789e-06, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21670 + }, + { + "epoch": 6.307826592958976, + "flow_matching_loss": 0.010932360542938113, + "grad_norm": 0.07062315195798874, + "learning_rate": 5.8797327508258396e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21680 + }, + { + "epoch": 6.310736107070119, + "flow_matching_loss": 0.011263726558536292, + "grad_norm": 0.08606491982936859, + "learning_rate": 5.866611010998584e-06, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21690 + }, + { + "epoch": 6.313645621181263, + "flow_matching_loss": 0.011200494412332774, + "grad_norm": 0.06816782802343369, + "learning_rate": 5.853500369100172e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21700 + }, + { + "epoch": 6.316555135292406, + "flow_matching_loss": 0.009988046484068036, + "grad_norm": 0.07421620190143585, + "learning_rate": 5.840400841061265e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21710 + }, + { + "epoch": 6.31946464940355, + "flow_matching_loss": 0.010100199887529016, + "grad_norm": 0.07357862591743469, + "learning_rate": 5.827312442799034e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21720 + }, + { + "epoch": 6.322374163514693, + "flow_matching_loss": 0.009615933429449797, + "grad_norm": 0.0664716586470604, + "learning_rate": 5.814235190217114e-06, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21730 + }, + { + "epoch": 6.325283677625836, + "flow_matching_loss": 0.01060992032289505, + "grad_norm": 0.07665088772773743, + "learning_rate": 5.801169099205599e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21740 + }, + { + "epoch": 6.32819319173698, + "flow_matching_loss": 0.010357389412820339, + "grad_norm": 0.0920429676771164, + "learning_rate": 5.788114185641022e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21750 + }, + { + "epoch": 6.3311027058481235, + "flow_matching_loss": 0.01068288297392428, + "grad_norm": 0.06713535636663437, + "learning_rate": 5.775070465386328e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21760 + }, + { + "epoch": 6.334012219959266, + "flow_matching_loss": 0.011191779747605324, + "grad_norm": 0.09228601306676865, + "learning_rate": 5.7620379542908765e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21770 + }, + { + "epoch": 6.33692173407041, + "flow_matching_loss": 0.011201866995543241, + "grad_norm": 0.08220358192920685, + "learning_rate": 5.749016668190395e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21780 + }, + { + "epoch": 6.339831248181554, + "flow_matching_loss": 0.011589777003973723, + "grad_norm": 0.06629833579063416, + "learning_rate": 5.736006622906973e-06, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21790 + }, + { + "epoch": 6.3427407622926975, + "flow_matching_loss": 0.011847598385065794, + "grad_norm": 0.05659361928701401, + "learning_rate": 5.723007834249034e-06, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21800 + }, + { + "epoch": 6.34565027640384, + "flow_matching_loss": 0.010994887631386518, + "grad_norm": 0.06633307039737701, + "learning_rate": 5.710020318011346e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21810 + }, + { + "epoch": 6.348559790514984, + "flow_matching_loss": 0.011517756804823875, + "grad_norm": 0.06053311005234718, + "learning_rate": 5.697044089974955e-06, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21820 + }, + { + "epoch": 6.351469304626128, + "flow_matching_loss": 0.011366192065179349, + "grad_norm": 0.07258746773004532, + "learning_rate": 5.68407916590721e-06, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21830 + }, + { + "epoch": 6.3543788187372705, + "flow_matching_loss": 0.00983262937515974, + "grad_norm": 0.06605750322341919, + "learning_rate": 5.671125561561713e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21840 + }, + { + "epoch": 6.357288332848414, + "flow_matching_loss": 0.01165546989068389, + "grad_norm": 0.07290425151586533, + "learning_rate": 5.658183292678315e-06, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21850 + }, + { + "epoch": 6.360197846959558, + "flow_matching_loss": 0.010079888300970197, + "grad_norm": 0.04417240247130394, + "learning_rate": 5.645252374983091e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21860 + }, + { + "epoch": 6.363107361070702, + "flow_matching_loss": 0.011624017683789133, + "grad_norm": 0.054813679307699203, + "learning_rate": 5.6323328241883224e-06, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21870 + }, + { + "epoch": 6.3660168751818444, + "flow_matching_loss": 0.011113485833629966, + "grad_norm": 0.09586159139871597, + "learning_rate": 5.61942465599249e-06, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21880 + }, + { + "epoch": 6.368926389292988, + "flow_matching_loss": 0.011198474280536174, + "grad_norm": 0.11232902109622955, + "learning_rate": 5.60652788608023e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21890 + }, + { + "epoch": 6.371835903404132, + "flow_matching_loss": 0.010034953895956277, + "grad_norm": 0.06250927597284317, + "learning_rate": 5.593642530122336e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21900 + }, + { + "epoch": 6.374745417515275, + "flow_matching_loss": 0.012817969964817166, + "grad_norm": 0.08254998177289963, + "learning_rate": 5.58076860377573e-06, + "loss": 0.0128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21910 + }, + { + "epoch": 6.377654931626418, + "flow_matching_loss": 0.010385392373427749, + "grad_norm": 0.08434879779815674, + "learning_rate": 5.567906122683439e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21920 + }, + { + "epoch": 6.380564445737562, + "flow_matching_loss": 0.012085475493222475, + "grad_norm": 0.10374966263771057, + "learning_rate": 5.555055102474604e-06, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21930 + }, + { + "epoch": 6.383473959848705, + "flow_matching_loss": 0.010676835663616657, + "grad_norm": 0.05979422107338905, + "learning_rate": 5.542215558764421e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21940 + }, + { + "epoch": 6.386383473959849, + "flow_matching_loss": 0.011003722716122866, + "grad_norm": 0.08038497716188431, + "learning_rate": 5.529387507154138e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21950 + }, + { + "epoch": 6.389292988070992, + "flow_matching_loss": 0.011487421439960599, + "grad_norm": 0.07323332130908966, + "learning_rate": 5.516570963231058e-06, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21960 + }, + { + "epoch": 6.392202502182135, + "flow_matching_loss": 0.010729688731953502, + "grad_norm": 0.08265280723571777, + "learning_rate": 5.503765942568484e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21970 + }, + { + "epoch": 6.395112016293279, + "flow_matching_loss": 0.010483292303979397, + "grad_norm": 0.06391476094722748, + "learning_rate": 5.49097246072573e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21980 + }, + { + "epoch": 6.398021530404423, + "flow_matching_loss": 0.010926993563771248, + "grad_norm": 0.08631955087184906, + "learning_rate": 5.4781905332480805e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21990 + }, + { + "epoch": 6.400931044515566, + "flow_matching_loss": 0.010688553098589182, + "grad_norm": 0.06727775186300278, + "learning_rate": 5.4654201756667795e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22000 + }, + { + "epoch": 6.403840558626709, + "flow_matching_loss": 0.009563035937026143, + "grad_norm": 0.07356077432632446, + "learning_rate": 5.452661403499017e-06, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22010 + }, + { + "epoch": 6.406750072737853, + "flow_matching_loss": 0.010094000911340117, + "grad_norm": 0.0705229863524437, + "learning_rate": 5.4399142322479005e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22020 + }, + { + "epoch": 6.4096595868489965, + "flow_matching_loss": 0.010911641968414187, + "grad_norm": 0.07008513063192368, + "learning_rate": 5.427178677402452e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22030 + }, + { + "epoch": 6.412569100960139, + "flow_matching_loss": 0.0109323360491544, + "grad_norm": 0.08673612773418427, + "learning_rate": 5.41445475443757e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22040 + }, + { + "epoch": 6.415478615071283, + "flow_matching_loss": 0.010691702971234918, + "grad_norm": 0.07221484184265137, + "learning_rate": 5.4017424788140194e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22050 + }, + { + "epoch": 6.418388129182427, + "flow_matching_loss": 0.011637268261983991, + "grad_norm": 0.06431622803211212, + "learning_rate": 5.389041865978414e-06, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22060 + }, + { + "epoch": 6.4212976432935704, + "flow_matching_loss": 0.010877575632184744, + "grad_norm": 0.06743451952934265, + "learning_rate": 5.37635293136319e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22070 + }, + { + "epoch": 6.424207157404713, + "flow_matching_loss": 0.012151591014117002, + "grad_norm": 0.1176953837275505, + "learning_rate": 5.363675690386609e-06, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22080 + }, + { + "epoch": 6.427116671515857, + "flow_matching_loss": 0.011562235048040747, + "grad_norm": 0.06455793976783752, + "learning_rate": 5.351010158452708e-06, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22090 + }, + { + "epoch": 6.430026185627001, + "flow_matching_loss": 0.010754018602892757, + "grad_norm": 0.12407859414815903, + "learning_rate": 5.338356350951308e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22100 + }, + { + "epoch": 6.4329356997381435, + "flow_matching_loss": 0.00986883887089789, + "grad_norm": 0.08221182972192764, + "learning_rate": 5.325714283257974e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22110 + }, + { + "epoch": 6.435845213849287, + "flow_matching_loss": 0.01133245201781392, + "grad_norm": 0.054071709513664246, + "learning_rate": 5.313083970734013e-06, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22120 + }, + { + "epoch": 6.438754727960431, + "flow_matching_loss": 0.013281904021278023, + "grad_norm": 0.09594108909368515, + "learning_rate": 5.300465428726439e-06, + "loss": 0.0133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22130 + }, + { + "epoch": 6.441664242071574, + "flow_matching_loss": 0.010846563382074237, + "grad_norm": 0.05618395283818245, + "learning_rate": 5.287858672567979e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22140 + }, + { + "epoch": 6.444573756182717, + "flow_matching_loss": 0.011091956123709678, + "grad_norm": 0.061346665024757385, + "learning_rate": 5.275263717577028e-06, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22150 + }, + { + "epoch": 6.447483270293861, + "flow_matching_loss": 0.010006561363115906, + "grad_norm": 0.08235504478216171, + "learning_rate": 5.2626805790576435e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22160 + }, + { + "epoch": 6.450392784405004, + "flow_matching_loss": 0.010834379913285374, + "grad_norm": 0.07146710902452469, + "learning_rate": 5.250109272299523e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22170 + }, + { + "epoch": 6.453302298516148, + "flow_matching_loss": 0.010057849949225784, + "grad_norm": 0.12720942497253418, + "learning_rate": 5.237549812577987e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22180 + }, + { + "epoch": 6.456211812627291, + "flow_matching_loss": 0.011908175842836499, + "grad_norm": 0.0781310573220253, + "learning_rate": 5.225002215153973e-06, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22190 + }, + { + "epoch": 6.459121326738435, + "flow_matching_loss": 0.010320193832740188, + "grad_norm": 0.055062245577573776, + "learning_rate": 5.21246649527399e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22200 + }, + { + "epoch": 6.462030840849578, + "flow_matching_loss": 0.009886367619037629, + "grad_norm": 0.07134661078453064, + "learning_rate": 5.199942668170118e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22210 + }, + { + "epoch": 6.464940354960722, + "flow_matching_loss": 0.010603540483862162, + "grad_norm": 0.07056201994419098, + "learning_rate": 5.187430749059986e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22220 + }, + { + "epoch": 6.467849869071865, + "flow_matching_loss": 0.011477174703031778, + "grad_norm": 0.0758262574672699, + "learning_rate": 5.174930753146763e-06, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22230 + }, + { + "epoch": 6.470759383183008, + "flow_matching_loss": 0.009568921942263841, + "grad_norm": 0.05852721631526947, + "learning_rate": 5.162442695619114e-06, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22240 + }, + { + "epoch": 6.473668897294152, + "flow_matching_loss": 0.010492715518921614, + "grad_norm": 0.08952372521162033, + "learning_rate": 5.149966591651215e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22250 + }, + { + "epoch": 6.4765784114052956, + "flow_matching_loss": 0.011256443616002799, + "grad_norm": 0.07964839786291122, + "learning_rate": 5.137502456402705e-06, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22260 + }, + { + "epoch": 6.479487925516438, + "flow_matching_loss": 0.011308955401182175, + "grad_norm": 0.06593308597803116, + "learning_rate": 5.125050305018683e-06, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22270 + }, + { + "epoch": 6.482397439627582, + "flow_matching_loss": 0.01087953094393015, + "grad_norm": 0.06621571630239487, + "learning_rate": 5.112610152629687e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22280 + }, + { + "epoch": 6.485306953738726, + "flow_matching_loss": 0.011715941783040761, + "grad_norm": 0.053480278700590134, + "learning_rate": 5.10018201435167e-06, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22290 + }, + { + "epoch": 6.488216467849869, + "flow_matching_loss": 0.009501193976029753, + "grad_norm": 0.06221068277955055, + "learning_rate": 5.087765905286004e-06, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22300 + }, + { + "epoch": 6.491125981961012, + "flow_matching_loss": 0.01030484982766211, + "grad_norm": 0.1009124144911766, + "learning_rate": 5.0753618405194245e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22310 + }, + { + "epoch": 6.494035496072156, + "flow_matching_loss": 0.010425991425290704, + "grad_norm": 0.05749412626028061, + "learning_rate": 5.062969835124042e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22320 + }, + { + "epoch": 6.4969450101833, + "flow_matching_loss": 0.010454594995826483, + "grad_norm": 0.07880470901727676, + "learning_rate": 5.05058990415731e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22330 + }, + { + "epoch": 6.4998545242944425, + "flow_matching_loss": 0.010786056146025657, + "grad_norm": 0.08968978375196457, + "learning_rate": 5.038222062662007e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22340 + }, + { + "epoch": 6.502764038405586, + "flow_matching_loss": 0.012493227235972881, + "grad_norm": 0.09091076999902725, + "learning_rate": 5.025866325666237e-06, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22350 + }, + { + "epoch": 6.50567355251673, + "flow_matching_loss": 0.011661929450929165, + "grad_norm": 0.07243366539478302, + "learning_rate": 5.013522708183383e-06, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22360 + }, + { + "epoch": 6.508583066627873, + "flow_matching_loss": 0.011898537632077932, + "grad_norm": 0.06952078640460968, + "learning_rate": 5.001191225212096e-06, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22370 + }, + { + "epoch": 6.5114925807390165, + "flow_matching_loss": 0.01033204752020538, + "grad_norm": 0.11891891807317734, + "learning_rate": 4.9888718917363e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22380 + }, + { + "epoch": 6.51440209485016, + "flow_matching_loss": 0.009530191216617823, + "grad_norm": 0.07505469769239426, + "learning_rate": 4.9765647227251395e-06, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22390 + }, + { + "epoch": 6.517311608961304, + "flow_matching_loss": 0.00975820147432387, + "grad_norm": 0.0514066219329834, + "learning_rate": 4.964269733132992e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22400 + }, + { + "epoch": 6.520221123072447, + "flow_matching_loss": 0.01072551216930151, + "grad_norm": 0.06097044795751572, + "learning_rate": 4.951986937899428e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22410 + }, + { + "epoch": 6.52313063718359, + "flow_matching_loss": 0.010733660962432622, + "grad_norm": 0.08633723855018616, + "learning_rate": 4.9397163519492e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22420 + }, + { + "epoch": 6.526040151294734, + "flow_matching_loss": 0.010533998953178525, + "grad_norm": 0.05600294470787048, + "learning_rate": 4.9274579901922275e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22430 + }, + { + "epoch": 6.528949665405877, + "flow_matching_loss": 0.009210717538371683, + "grad_norm": 0.06775280088186264, + "learning_rate": 4.9152118675235705e-06, + "loss": 0.0092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22440 + }, + { + "epoch": 6.531859179517021, + "flow_matching_loss": 0.010882122442126274, + "grad_norm": 0.07598288357257843, + "learning_rate": 4.902977998823432e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22450 + }, + { + "epoch": 6.534768693628164, + "flow_matching_loss": 0.010904509946703911, + "grad_norm": 0.058775097131729126, + "learning_rate": 4.890756398957114e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22460 + }, + { + "epoch": 6.537678207739307, + "flow_matching_loss": 0.010515925753861666, + "grad_norm": 0.06739635020494461, + "learning_rate": 4.8785470827750075e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22470 + }, + { + "epoch": 6.540587721850451, + "flow_matching_loss": 0.010730902990326285, + "grad_norm": 0.0598277784883976, + "learning_rate": 4.866350065112588e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22480 + }, + { + "epoch": 6.543497235961595, + "flow_matching_loss": 0.009074293496087193, + "grad_norm": 0.10104523599147797, + "learning_rate": 4.854165360790381e-06, + "loss": 0.0091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22490 + }, + { + "epoch": 6.546406750072737, + "flow_matching_loss": 0.009560061618685722, + "grad_norm": 0.06868252903223038, + "learning_rate": 4.8419929846139425e-06, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22500 + }, + { + "epoch": 6.549316264183881, + "flow_matching_loss": 0.011800178140401841, + "grad_norm": 0.08853071182966232, + "learning_rate": 4.829832951373872e-06, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22510 + }, + { + "epoch": 6.552225778295025, + "flow_matching_loss": 0.0105248237028718, + "grad_norm": 0.07924090325832367, + "learning_rate": 4.817685275845746e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22520 + }, + { + "epoch": 6.5551352924061685, + "flow_matching_loss": 0.011466531362384558, + "grad_norm": 0.07439959049224854, + "learning_rate": 4.805549972790143e-06, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22530 + }, + { + "epoch": 6.558044806517311, + "flow_matching_loss": 0.010533817717805505, + "grad_norm": 0.0646279901266098, + "learning_rate": 4.7934270569525966e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22540 + }, + { + "epoch": 6.560954320628455, + "flow_matching_loss": 0.010119153186678886, + "grad_norm": 0.09347734600305557, + "learning_rate": 4.7813165430635874e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22550 + }, + { + "epoch": 6.563863834739599, + "flow_matching_loss": 0.01114663016051054, + "grad_norm": 0.08622638136148453, + "learning_rate": 4.769218445838544e-06, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22560 + }, + { + "epoch": 6.566773348850742, + "flow_matching_loss": 0.010855477768927813, + "grad_norm": 0.08330854773521423, + "learning_rate": 4.757132779977787e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22570 + }, + { + "epoch": 6.569682862961885, + "flow_matching_loss": 0.009843257628381253, + "grad_norm": 0.0543922483921051, + "learning_rate": 4.745059560166542e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22580 + }, + { + "epoch": 6.572592377073029, + "flow_matching_loss": 0.009868529113009571, + "grad_norm": 0.057622503489255905, + "learning_rate": 4.732998801074911e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22590 + }, + { + "epoch": 6.575501891184173, + "flow_matching_loss": 0.009872040571644902, + "grad_norm": 0.06450524926185608, + "learning_rate": 4.7209505173578445e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22600 + }, + { + "epoch": 6.5784114052953155, + "flow_matching_loss": 0.011769250128418207, + "grad_norm": 0.0932685062289238, + "learning_rate": 4.7089147236551555e-06, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22610 + }, + { + "epoch": 6.581320919406459, + "flow_matching_loss": 0.010160102462396025, + "grad_norm": 0.06843835860490799, + "learning_rate": 4.696891434591464e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22620 + }, + { + "epoch": 6.584230433517602, + "flow_matching_loss": 0.011439660377800465, + "grad_norm": 0.11798267811536789, + "learning_rate": 4.684880664776199e-06, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22630 + }, + { + "epoch": 6.587139947628746, + "flow_matching_loss": 0.011657760851085186, + "grad_norm": 0.0785459652543068, + "learning_rate": 4.6728824288035784e-06, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22640 + }, + { + "epoch": 6.5900494617398895, + "flow_matching_loss": 0.011420137714594603, + "grad_norm": 0.07147801667451859, + "learning_rate": 4.660896741252586e-06, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22650 + }, + { + "epoch": 6.592958975851033, + "flow_matching_loss": 0.011084096506237984, + "grad_norm": 0.06738278269767761, + "learning_rate": 4.6489236166869714e-06, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22660 + }, + { + "epoch": 6.595868489962176, + "flow_matching_loss": 0.008881580643355846, + "grad_norm": 0.059052422642707825, + "learning_rate": 4.6369630696552e-06, + "loss": 0.0089, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22670 + }, + { + "epoch": 6.59877800407332, + "flow_matching_loss": 0.01043472378514707, + "grad_norm": 0.06335704028606415, + "learning_rate": 4.625015114690474e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22680 + }, + { + "epoch": 6.601687518184463, + "flow_matching_loss": 0.00969835021533072, + "grad_norm": 0.06263366341590881, + "learning_rate": 4.613079766310682e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22690 + }, + { + "epoch": 6.604597032295606, + "flow_matching_loss": 0.009989737952128052, + "grad_norm": 0.0729808360338211, + "learning_rate": 4.601157039018396e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22700 + }, + { + "epoch": 6.60750654640675, + "flow_matching_loss": 0.009245872544124723, + "grad_norm": 0.06711620092391968, + "learning_rate": 4.589246947300849e-06, + "loss": 0.0092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22710 + }, + { + "epoch": 6.610416060517894, + "flow_matching_loss": 0.009881186950951814, + "grad_norm": 0.06401615589857101, + "learning_rate": 4.577349505629936e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22720 + }, + { + "epoch": 6.613325574629037, + "flow_matching_loss": 0.010883102333173156, + "grad_norm": 0.07894618064165115, + "learning_rate": 4.565464728462165e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22730 + }, + { + "epoch": 6.61623508874018, + "flow_matching_loss": 0.011653543077409267, + "grad_norm": 0.0919647365808487, + "learning_rate": 4.553592630238663e-06, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22740 + }, + { + "epoch": 6.619144602851324, + "flow_matching_loss": 0.010131051996722817, + "grad_norm": 0.0923100933432579, + "learning_rate": 4.541733225385146e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22750 + }, + { + "epoch": 6.622054116962468, + "flow_matching_loss": 0.010121889505535364, + "grad_norm": 0.08470270782709122, + "learning_rate": 4.529886528311905e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22760 + }, + { + "epoch": 6.62496363107361, + "flow_matching_loss": 0.011360310250893235, + "grad_norm": 0.07121415436267853, + "learning_rate": 4.5180525534138045e-06, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22770 + }, + { + "epoch": 6.627873145184754, + "flow_matching_loss": 0.011360930278897285, + "grad_norm": 0.06399358808994293, + "learning_rate": 4.506231315070235e-06, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22780 + }, + { + "epoch": 6.630782659295898, + "flow_matching_loss": 0.010545431589707732, + "grad_norm": 0.10425735265016556, + "learning_rate": 4.494422827645115e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22790 + }, + { + "epoch": 6.633692173407041, + "flow_matching_loss": 0.00980397528037429, + "grad_norm": 0.08156358450651169, + "learning_rate": 4.482627105486865e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22800 + }, + { + "epoch": 6.636601687518184, + "flow_matching_loss": 0.009571790602058173, + "grad_norm": 0.06583532691001892, + "learning_rate": 4.470844162928412e-06, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22810 + }, + { + "epoch": 6.639511201629328, + "flow_matching_loss": 0.01030798051506281, + "grad_norm": 0.07296961545944214, + "learning_rate": 4.459074014287129e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22820 + }, + { + "epoch": 6.642420715740471, + "flow_matching_loss": 0.010872604418545961, + "grad_norm": 0.06850424408912659, + "learning_rate": 4.447316673864869e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22830 + }, + { + "epoch": 6.645330229851615, + "flow_matching_loss": 0.010051537724211811, + "grad_norm": 0.07025720179080963, + "learning_rate": 4.435572155947903e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22840 + }, + { + "epoch": 6.648239743962758, + "flow_matching_loss": 0.01045102309435606, + "grad_norm": 0.06941697001457214, + "learning_rate": 4.423840474806929e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22850 + }, + { + "epoch": 6.651149258073902, + "flow_matching_loss": 0.01077987183816731, + "grad_norm": 0.06750548630952835, + "learning_rate": 4.412121644697043e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22860 + }, + { + "epoch": 6.654058772185045, + "flow_matching_loss": 0.009622130542993546, + "grad_norm": 0.0688510313630104, + "learning_rate": 4.400415679857727e-06, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22870 + }, + { + "epoch": 6.6569682862961885, + "flow_matching_loss": 0.010894953133538365, + "grad_norm": 0.08124823868274689, + "learning_rate": 4.388722594512838e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22880 + }, + { + "epoch": 6.659877800407332, + "flow_matching_loss": 0.010655834525823592, + "grad_norm": 0.06763538718223572, + "learning_rate": 4.377042402870577e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22890 + }, + { + "epoch": 6.662787314518475, + "flow_matching_loss": 0.009487154381349682, + "grad_norm": 0.06200428679585457, + "learning_rate": 4.365375119123476e-06, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22900 + }, + { + "epoch": 6.665696828629619, + "flow_matching_loss": 0.009581807302311063, + "grad_norm": 0.05460108444094658, + "learning_rate": 4.353720757448383e-06, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22910 + }, + { + "epoch": 6.6686063427407625, + "flow_matching_loss": 0.009828021144494414, + "grad_norm": 0.09868600964546204, + "learning_rate": 4.342079332006445e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22920 + }, + { + "epoch": 6.671515856851906, + "flow_matching_loss": 0.01045312792994082, + "grad_norm": 0.07399923354387283, + "learning_rate": 4.330450856943101e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22930 + }, + { + "epoch": 6.674425370963049, + "flow_matching_loss": 0.010338955651968717, + "grad_norm": 0.07668739557266235, + "learning_rate": 4.318835346388036e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22940 + }, + { + "epoch": 6.677334885074193, + "flow_matching_loss": 0.00991935795173049, + "grad_norm": 0.07127862423658371, + "learning_rate": 4.307232814455189e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22950 + }, + { + "epoch": 6.680244399185336, + "flow_matching_loss": 0.010477295937016606, + "grad_norm": 0.07653055340051651, + "learning_rate": 4.29564327524274e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22960 + }, + { + "epoch": 6.683153913296479, + "flow_matching_loss": 0.01054423050954938, + "grad_norm": 0.06092218682169914, + "learning_rate": 4.284066742833061e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22970 + }, + { + "epoch": 6.686063427407623, + "flow_matching_loss": 0.010357377119362354, + "grad_norm": 0.089095838367939, + "learning_rate": 4.272503231292741e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22980 + }, + { + "epoch": 6.688972941518767, + "flow_matching_loss": 0.011586513184010983, + "grad_norm": 0.08466538786888123, + "learning_rate": 4.260952754672532e-06, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22990 + }, + { + "epoch": 6.6918824556299095, + "flow_matching_loss": 0.01179595272988081, + "grad_norm": 0.10141734778881073, + "learning_rate": 4.249415327007353e-06, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23000 + }, + { + "epoch": 6.694791969741053, + "flow_matching_loss": 0.01069610440172255, + "grad_norm": 0.07811164855957031, + "learning_rate": 4.237890962316264e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23010 + }, + { + "epoch": 6.697701483852197, + "flow_matching_loss": 0.010570137249305844, + "grad_norm": 0.08986081928014755, + "learning_rate": 4.226379674602452e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23020 + }, + { + "epoch": 6.70061099796334, + "flow_matching_loss": 0.010777468839660288, + "grad_norm": 0.0918043851852417, + "learning_rate": 4.214881477853227e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23030 + }, + { + "epoch": 6.703520512074483, + "flow_matching_loss": 0.01043098010122776, + "grad_norm": 0.06447724997997284, + "learning_rate": 4.2033963860399745e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23040 + }, + { + "epoch": 6.706430026185627, + "flow_matching_loss": 0.009053063625469805, + "grad_norm": 0.06389632821083069, + "learning_rate": 4.1919244131181696e-06, + "loss": 0.0091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23050 + }, + { + "epoch": 6.709339540296771, + "flow_matching_loss": 0.01048198975622654, + "grad_norm": 0.06244286149740219, + "learning_rate": 4.180465573027335e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23060 + }, + { + "epoch": 6.712249054407914, + "flow_matching_loss": 0.008938416466116905, + "grad_norm": 0.08081894367933273, + "learning_rate": 4.169019879691046e-06, + "loss": 0.0089, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23070 + }, + { + "epoch": 6.715158568519057, + "flow_matching_loss": 0.010657793283462525, + "grad_norm": 0.0769881084561348, + "learning_rate": 4.157587347016891e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23080 + }, + { + "epoch": 6.718068082630201, + "flow_matching_loss": 0.01055358317680657, + "grad_norm": 0.07440828531980515, + "learning_rate": 4.14616798889649e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23090 + }, + { + "epoch": 6.720977596741344, + "flow_matching_loss": 0.009020025422796608, + "grad_norm": 0.0628008171916008, + "learning_rate": 4.134761819205425e-06, + "loss": 0.009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23100 + }, + { + "epoch": 6.723887110852488, + "flow_matching_loss": 0.009856946580111981, + "grad_norm": 0.06592578440904617, + "learning_rate": 4.1233688518032794e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23110 + }, + { + "epoch": 6.726796624963631, + "flow_matching_loss": 0.009562835516408085, + "grad_norm": 0.07065927237272263, + "learning_rate": 4.1119891005335795e-06, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23120 + }, + { + "epoch": 6.729706139074775, + "flow_matching_loss": 0.009507758682593704, + "grad_norm": 0.07437610626220703, + "learning_rate": 4.1006225792237895e-06, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23130 + }, + { + "epoch": 6.732615653185918, + "flow_matching_loss": 0.010040757432579994, + "grad_norm": 0.08320998400449753, + "learning_rate": 4.089269301685314e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23140 + }, + { + "epoch": 6.7355251672970615, + "flow_matching_loss": 0.009321223897859453, + "grad_norm": 0.07336831092834473, + "learning_rate": 4.077929281713453e-06, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23150 + }, + { + "epoch": 6.738434681408204, + "flow_matching_loss": 0.010534630250185728, + "grad_norm": 0.08516070246696472, + "learning_rate": 4.066602533087398e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23160 + }, + { + "epoch": 6.741344195519348, + "flow_matching_loss": 0.010406833281740547, + "grad_norm": 0.07715325802564621, + "learning_rate": 4.0552890695702175e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23170 + }, + { + "epoch": 6.744253709630492, + "flow_matching_loss": 0.010581648349761963, + "grad_norm": 0.08299023658037186, + "learning_rate": 4.04398890490883e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23180 + }, + { + "epoch": 6.7471632237416355, + "flow_matching_loss": 0.010566968005150557, + "grad_norm": 0.08972060680389404, + "learning_rate": 4.032702052834013e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23190 + }, + { + "epoch": 6.750072737852778, + "flow_matching_loss": 0.010292152408510447, + "grad_norm": 0.0592804029583931, + "learning_rate": 4.021428527060352e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23200 + }, + { + "epoch": 6.752982251963922, + "flow_matching_loss": 0.010884643672034145, + "grad_norm": 0.046798981726169586, + "learning_rate": 4.010168341286241e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23210 + }, + { + "epoch": 6.755891766075066, + "flow_matching_loss": 0.010775156132876874, + "grad_norm": 0.1010979413986206, + "learning_rate": 3.998921509193868e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23220 + }, + { + "epoch": 6.7588012801862085, + "flow_matching_loss": 0.010676543693989515, + "grad_norm": 0.08143346756696701, + "learning_rate": 3.987688044449193e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23230 + }, + { + "epoch": 6.761710794297352, + "flow_matching_loss": 0.012298999540507793, + "grad_norm": 0.081698477268219, + "learning_rate": 3.9764679607019385e-06, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23240 + }, + { + "epoch": 6.764620308408496, + "flow_matching_loss": 0.010494666127488016, + "grad_norm": 0.07197803258895874, + "learning_rate": 3.965261271585561e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23250 + }, + { + "epoch": 6.76752982251964, + "flow_matching_loss": 0.010061824787408113, + "grad_norm": 0.08602410554885864, + "learning_rate": 3.954067990717251e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23260 + }, + { + "epoch": 6.7704393366307825, + "flow_matching_loss": 0.009937513340264558, + "grad_norm": 0.07883789390325546, + "learning_rate": 3.9428881316978935e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23270 + }, + { + "epoch": 6.773348850741926, + "flow_matching_loss": 0.011452178005129099, + "grad_norm": 0.06957551836967468, + "learning_rate": 3.931721708112075e-06, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23280 + }, + { + "epoch": 6.77625836485307, + "flow_matching_loss": 0.00971499881707132, + "grad_norm": 0.05582930147647858, + "learning_rate": 3.920568733528047e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23290 + }, + { + "epoch": 6.779167878964213, + "flow_matching_loss": 0.010538758104667067, + "grad_norm": 0.05959146097302437, + "learning_rate": 3.909429221497734e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23300 + }, + { + "epoch": 6.782077393075356, + "flow_matching_loss": 0.01018094141036272, + "grad_norm": 0.06995944678783417, + "learning_rate": 3.898303185556689e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23310 + }, + { + "epoch": 6.7849869071865, + "flow_matching_loss": 0.011185329873114824, + "grad_norm": 0.09174073487520218, + "learning_rate": 3.887190639224097e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23320 + }, + { + "epoch": 6.787896421297643, + "flow_matching_loss": 0.00953395334072411, + "grad_norm": 0.05992875248193741, + "learning_rate": 3.876091596002746e-06, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23330 + }, + { + "epoch": 6.790805935408787, + "flow_matching_loss": 0.011250666156411172, + "grad_norm": 0.07425006479024887, + "learning_rate": 3.865006069379019e-06, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23340 + }, + { + "epoch": 6.79371544951993, + "flow_matching_loss": 0.009658182179555297, + "grad_norm": 0.06913809478282928, + "learning_rate": 3.853934072822881e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23350 + }, + { + "epoch": 6.796624963631073, + "flow_matching_loss": 0.009963947581127286, + "grad_norm": 0.10095816850662231, + "learning_rate": 3.8428756197878514e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23360 + }, + { + "epoch": 6.799534477742217, + "flow_matching_loss": 0.011216903291642665, + "grad_norm": 0.06643474102020264, + "learning_rate": 3.831830723710992e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23370 + }, + { + "epoch": 6.802443991853361, + "flow_matching_loss": 0.009686031937599182, + "grad_norm": 0.08692451566457748, + "learning_rate": 3.820799398012889e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23380 + }, + { + "epoch": 6.805353505964504, + "flow_matching_loss": 0.011046546837314963, + "grad_norm": 0.10616673529148102, + "learning_rate": 3.80978165609765e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23390 + }, + { + "epoch": 6.808263020075647, + "flow_matching_loss": 0.011266785487532616, + "grad_norm": 0.1233898401260376, + "learning_rate": 3.798777511352864e-06, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23400 + }, + { + "epoch": 6.811172534186791, + "flow_matching_loss": 0.010757653694599866, + "grad_norm": 0.07024506479501724, + "learning_rate": 3.787786977149613e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23410 + }, + { + "epoch": 6.8140820482979345, + "flow_matching_loss": 0.010365914972499012, + "grad_norm": 0.07635378837585449, + "learning_rate": 3.776810066842427e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23420 + }, + { + "epoch": 6.816991562409077, + "flow_matching_loss": 0.009669589390978218, + "grad_norm": 0.07103187590837479, + "learning_rate": 3.7658467937692866e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23430 + }, + { + "epoch": 6.819901076520221, + "flow_matching_loss": 0.009713065717369318, + "grad_norm": 0.06050186604261398, + "learning_rate": 3.7548971712516e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23440 + }, + { + "epoch": 6.822810590631365, + "flow_matching_loss": 0.010560292471200227, + "grad_norm": 0.07495636492967606, + "learning_rate": 3.743961212594197e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23450 + }, + { + "epoch": 6.8257201047425085, + "flow_matching_loss": 0.009897005651146174, + "grad_norm": 0.06100299954414368, + "learning_rate": 3.7330389310852975e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23460 + }, + { + "epoch": 6.828629618853651, + "flow_matching_loss": 0.01182216675952077, + "grad_norm": 0.07334741204977036, + "learning_rate": 3.7221303399965008e-06, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23470 + }, + { + "epoch": 6.831539132964795, + "flow_matching_loss": 0.009887227602303029, + "grad_norm": 0.07577802240848541, + "learning_rate": 3.711235452582775e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23480 + }, + { + "epoch": 6.834448647075939, + "flow_matching_loss": 0.009769758395850658, + "grad_norm": 0.059921927750110626, + "learning_rate": 3.7003542820824347e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23490 + }, + { + "epoch": 6.8373581611870815, + "flow_matching_loss": 0.011194508336484433, + "grad_norm": 0.1482708901166916, + "learning_rate": 3.6894868417171252e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23500 + }, + { + "epoch": 6.840267675298225, + "flow_matching_loss": 0.010463344817981123, + "grad_norm": 0.0721730962395668, + "learning_rate": 3.678633144691818e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23510 + }, + { + "epoch": 6.843177189409369, + "flow_matching_loss": 0.010488907666876912, + "grad_norm": 0.06635210663080215, + "learning_rate": 3.667793204194776e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23520 + }, + { + "epoch": 6.846086703520512, + "flow_matching_loss": 0.008065061178058386, + "grad_norm": 0.0637611597776413, + "learning_rate": 3.656967033397544e-06, + "loss": 0.0081, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23530 + }, + { + "epoch": 6.8489962176316554, + "flow_matching_loss": 0.011277193017303944, + "grad_norm": 0.15559256076812744, + "learning_rate": 3.646154645454948e-06, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23540 + }, + { + "epoch": 6.851905731742799, + "flow_matching_loss": 0.009765047393739224, + "grad_norm": 0.06072736531496048, + "learning_rate": 3.6353560535050524e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23550 + }, + { + "epoch": 6.854815245853942, + "flow_matching_loss": 0.009745443193241953, + "grad_norm": 0.06371792405843735, + "learning_rate": 3.624571270669172e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23560 + }, + { + "epoch": 6.857724759965086, + "flow_matching_loss": 0.010239339014515281, + "grad_norm": 0.06948916614055634, + "learning_rate": 3.6138003100518296e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23570 + }, + { + "epoch": 6.860634274076229, + "flow_matching_loss": 0.01017552292905748, + "grad_norm": 0.054157745093107224, + "learning_rate": 3.6030431847407595e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23580 + }, + { + "epoch": 6.863543788187373, + "flow_matching_loss": 0.011453427374362946, + "grad_norm": 0.08837827295064926, + "learning_rate": 3.592299907806881e-06, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23590 + }, + { + "epoch": 6.866453302298516, + "flow_matching_loss": 0.009489061264321207, + "grad_norm": 0.11396949738264084, + "learning_rate": 3.5815704923042892e-06, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23600 + }, + { + "epoch": 6.86936281640966, + "flow_matching_loss": 0.010791440634056926, + "grad_norm": 0.07918751984834671, + "learning_rate": 3.5708549512702392e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23610 + }, + { + "epoch": 6.872272330520803, + "flow_matching_loss": 0.01088286442682147, + "grad_norm": 0.06546202301979065, + "learning_rate": 3.5601532977251217e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23620 + }, + { + "epoch": 6.875181844631946, + "flow_matching_loss": 0.009692893410101534, + "grad_norm": 0.13435028493404388, + "learning_rate": 3.5494655446724544e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23630 + }, + { + "epoch": 6.87809135874309, + "flow_matching_loss": 0.01098232800140977, + "grad_norm": 0.12443574517965317, + "learning_rate": 3.538791705098866e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23640 + }, + { + "epoch": 6.881000872854234, + "flow_matching_loss": 0.010603776946663857, + "grad_norm": 0.08257017284631729, + "learning_rate": 3.5281317919740718e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23650 + }, + { + "epoch": 6.883910386965377, + "flow_matching_loss": 0.008804046222940088, + "grad_norm": 0.11285407841205597, + "learning_rate": 3.5174858182508825e-06, + "loss": 0.0088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23660 + }, + { + "epoch": 6.88681990107652, + "flow_matching_loss": 0.011675117630511522, + "grad_norm": 0.0673433169722557, + "learning_rate": 3.506853796865153e-06, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23670 + }, + { + "epoch": 6.889729415187664, + "flow_matching_loss": 0.00992878614924848, + "grad_norm": 0.06687384843826294, + "learning_rate": 3.496235740735789e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23680 + }, + { + "epoch": 6.892638929298807, + "flow_matching_loss": 0.010511466907337308, + "grad_norm": 0.0807017907500267, + "learning_rate": 3.485631662764738e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23690 + }, + { + "epoch": 6.89554844340995, + "flow_matching_loss": 0.010554137732833624, + "grad_norm": 0.062443897128105164, + "learning_rate": 3.475041575836952e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23700 + }, + { + "epoch": 6.898457957521094, + "flow_matching_loss": 0.009983940701931714, + "grad_norm": 0.06739271432161331, + "learning_rate": 3.4644654928203796e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23710 + }, + { + "epoch": 6.901367471632238, + "flow_matching_loss": 0.009476414788514376, + "grad_norm": 0.06153979152441025, + "learning_rate": 3.4539034265659663e-06, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23720 + }, + { + "epoch": 6.904276985743381, + "flow_matching_loss": 0.010164432041347028, + "grad_norm": 0.07636187970638275, + "learning_rate": 3.4433553899076163e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23730 + }, + { + "epoch": 6.907186499854524, + "flow_matching_loss": 0.01074972739443183, + "grad_norm": 0.06144774705171585, + "learning_rate": 3.432821395662189e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23740 + }, + { + "epoch": 6.910096013965668, + "flow_matching_loss": 0.010972604621201754, + "grad_norm": 0.07866502553224564, + "learning_rate": 3.4223014566294785e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23750 + }, + { + "epoch": 6.913005528076811, + "flow_matching_loss": 0.011558174062520266, + "grad_norm": 0.06729863584041595, + "learning_rate": 3.4117955855922e-06, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23760 + }, + { + "epoch": 6.9159150421879545, + "flow_matching_loss": 0.010206640092656016, + "grad_norm": 0.04430825635790825, + "learning_rate": 3.4013037953159855e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23770 + }, + { + "epoch": 6.918824556299098, + "flow_matching_loss": 0.011412893328815699, + "grad_norm": 0.08783931285142899, + "learning_rate": 3.390826098549345e-06, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23780 + }, + { + "epoch": 6.921734070410242, + "flow_matching_loss": 0.009285690868273378, + "grad_norm": 0.07740875333547592, + "learning_rate": 3.3803625080236684e-06, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23790 + }, + { + "epoch": 6.924643584521385, + "flow_matching_loss": 0.01272791475057602, + "grad_norm": 0.13259850442409515, + "learning_rate": 3.369913036453205e-06, + "loss": 0.0127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23800 + }, + { + "epoch": 6.927553098632528, + "flow_matching_loss": 0.010371743654832243, + "grad_norm": 0.0597398541867733, + "learning_rate": 3.3594776965350434e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23810 + }, + { + "epoch": 6.930462612743672, + "flow_matching_loss": 0.0109394870698452, + "grad_norm": 0.058630023151636124, + "learning_rate": 3.34905650094911e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23820 + }, + { + "epoch": 6.933372126854815, + "flow_matching_loss": 0.009573278855532408, + "grad_norm": 0.06854598224163055, + "learning_rate": 3.338649462358145e-06, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23830 + }, + { + "epoch": 6.936281640965959, + "flow_matching_loss": 0.01015215590596199, + "grad_norm": 0.06377987563610077, + "learning_rate": 3.3282565934076765e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23840 + }, + { + "epoch": 6.939191155077102, + "flow_matching_loss": 0.01040540263056755, + "grad_norm": 0.06857599318027496, + "learning_rate": 3.31787790672602e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23850 + }, + { + "epoch": 6.942100669188245, + "flow_matching_loss": 0.009387770760804414, + "grad_norm": 0.07975255697965622, + "learning_rate": 3.307513414924255e-06, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23860 + }, + { + "epoch": 6.945010183299389, + "flow_matching_loss": 0.01116835786961019, + "grad_norm": 0.06634720414876938, + "learning_rate": 3.2971631305962246e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23870 + }, + { + "epoch": 6.947919697410533, + "flow_matching_loss": 0.010060734441503882, + "grad_norm": 0.05505410581827164, + "learning_rate": 3.2868270663184964e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23880 + }, + { + "epoch": 6.950829211521675, + "flow_matching_loss": 0.011325017455965281, + "grad_norm": 0.08290955424308777, + "learning_rate": 3.276505234650362e-06, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23890 + }, + { + "epoch": 6.953738725632819, + "flow_matching_loss": 0.010312958154827356, + "grad_norm": 0.0869620144367218, + "learning_rate": 3.266197648133821e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23900 + }, + { + "epoch": 6.956648239743963, + "flow_matching_loss": 0.01071281167678535, + "grad_norm": 0.09247194975614548, + "learning_rate": 3.255904319293562e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23910 + }, + { + "epoch": 6.9595577538551066, + "flow_matching_loss": 0.009962941939011215, + "grad_norm": 0.06322045624256134, + "learning_rate": 3.2456252606369485e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23920 + }, + { + "epoch": 6.962467267966249, + "flow_matching_loss": 0.009138092771172524, + "grad_norm": 0.06261929869651794, + "learning_rate": 3.235360484654011e-06, + "loss": 0.0091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23930 + }, + { + "epoch": 6.965376782077393, + "flow_matching_loss": 0.010426340671256185, + "grad_norm": 0.05590621754527092, + "learning_rate": 3.225110003817418e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23940 + }, + { + "epoch": 6.968286296188537, + "flow_matching_loss": 0.011203574016690254, + "grad_norm": 0.08401056379079819, + "learning_rate": 3.214873830582467e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23950 + }, + { + "epoch": 6.97119581029968, + "flow_matching_loss": 0.010644776001572608, + "grad_norm": 0.06261520087718964, + "learning_rate": 3.2046519773870803e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23960 + }, + { + "epoch": 6.974105324410823, + "flow_matching_loss": 0.010970491217449307, + "grad_norm": 0.06402667611837387, + "learning_rate": 3.1944444566517654e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23970 + }, + { + "epoch": 6.977014838521967, + "flow_matching_loss": 0.010725788911804557, + "grad_norm": 0.06849075853824615, + "learning_rate": 3.18425128077963e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23980 + }, + { + "epoch": 6.979924352633111, + "flow_matching_loss": 0.010638514626771211, + "grad_norm": 0.09845274686813354, + "learning_rate": 3.174072462156342e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23990 + }, + { + "epoch": 6.9828338667442535, + "flow_matching_loss": 0.008881258592009545, + "grad_norm": 0.06848648935556412, + "learning_rate": 3.163908013150124e-06, + "loss": 0.0089, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24000 + }, + { + "epoch": 6.985743380855397, + "flow_matching_loss": 0.00959054077975452, + "grad_norm": 0.06233704835176468, + "learning_rate": 3.153757946111741e-06, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24010 + }, + { + "epoch": 6.988652894966541, + "flow_matching_loss": 0.009690765477716922, + "grad_norm": 0.0745784118771553, + "learning_rate": 3.1436222733744776e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24020 + }, + { + "epoch": 6.991562409077684, + "flow_matching_loss": 0.009287129063159228, + "grad_norm": 0.058895912021398544, + "learning_rate": 3.133501007254138e-06, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24030 + }, + { + "epoch": 6.9944719231888275, + "flow_matching_loss": 0.010977535322308541, + "grad_norm": 0.07630622386932373, + "learning_rate": 3.1233941600490123e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24040 + }, + { + "epoch": 6.997381437299971, + "flow_matching_loss": 0.010491259908303619, + "grad_norm": 0.06621526926755905, + "learning_rate": 3.113301744039875e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24050 + }, + { + "epoch": 7.000290951411114, + "flow_matching_loss": 0.01035543237812817, + "grad_norm": 0.08303680270910263, + "learning_rate": 3.103223771489959e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24060 + }, + { + "epoch": 7.003200465522258, + "flow_matching_loss": 0.009140671137720347, + "grad_norm": 0.07548404484987259, + "learning_rate": 3.09316025464495e-06, + "loss": 0.0091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24070 + }, + { + "epoch": 7.006109979633401, + "flow_matching_loss": 0.010758683225139976, + "grad_norm": 0.07222791761159897, + "learning_rate": 3.0831112057329775e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24080 + }, + { + "epoch": 7.009019493744544, + "flow_matching_loss": 0.010277127055451274, + "grad_norm": 0.06976478546857834, + "learning_rate": 3.073076636964579e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24090 + }, + { + "epoch": 7.011929007855688, + "flow_matching_loss": 0.00933602023869753, + "grad_norm": 0.05668822303414345, + "learning_rate": 3.063056560532698e-06, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24100 + }, + { + "epoch": 7.014838521966832, + "flow_matching_loss": 0.009241723967716099, + "grad_norm": 0.07933174818754196, + "learning_rate": 3.0530509886126807e-06, + "loss": 0.0092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24110 + }, + { + "epoch": 7.017748036077975, + "flow_matching_loss": 0.0094625947996974, + "grad_norm": 0.10549532622098923, + "learning_rate": 3.043059933362236e-06, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24120 + }, + { + "epoch": 7.020657550189118, + "flow_matching_loss": 0.010684310644865035, + "grad_norm": 0.09598326683044434, + "learning_rate": 3.0330834069214337e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24130 + }, + { + "epoch": 7.023567064300262, + "flow_matching_loss": 0.010365666635334492, + "grad_norm": 0.09003135561943054, + "learning_rate": 3.023121421412703e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24140 + }, + { + "epoch": 7.026476578411406, + "flow_matching_loss": 0.01089885737746954, + "grad_norm": 0.07456701248884201, + "learning_rate": 3.0131739889407896e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24150 + }, + { + "epoch": 7.029386092522548, + "flow_matching_loss": 0.010368928592652083, + "grad_norm": 0.06540676206350327, + "learning_rate": 3.0032411215927645e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24160 + }, + { + "epoch": 7.032295606633692, + "flow_matching_loss": 0.010262905340641736, + "grad_norm": 0.07065881043672562, + "learning_rate": 2.9933228314379975e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24170 + }, + { + "epoch": 7.035205120744836, + "flow_matching_loss": 0.00965582779608667, + "grad_norm": 0.08713869005441666, + "learning_rate": 2.9834191305281405e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24180 + }, + { + "epoch": 7.038114634855979, + "flow_matching_loss": 0.010376376193016767, + "grad_norm": 0.06729117035865784, + "learning_rate": 2.973530030897135e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24190 + }, + { + "epoch": 7.041024148967122, + "flow_matching_loss": 0.009297521645203232, + "grad_norm": 0.09476396441459656, + "learning_rate": 2.9636555445611646e-06, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24200 + }, + { + "epoch": 7.043933663078266, + "flow_matching_loss": 0.010140687040984631, + "grad_norm": 0.08083068579435349, + "learning_rate": 2.953795683518659e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24210 + }, + { + "epoch": 7.04684317718941, + "flow_matching_loss": 0.010335349291563035, + "grad_norm": 0.08916924893856049, + "learning_rate": 2.9439504597502815e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24220 + }, + { + "epoch": 7.049752691300553, + "flow_matching_loss": 0.010937897395342589, + "grad_norm": 0.06992794573307037, + "learning_rate": 2.934119885218902e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24230 + }, + { + "epoch": 7.052662205411696, + "flow_matching_loss": 0.009001356549561024, + "grad_norm": 0.07075667381286621, + "learning_rate": 2.9243039718696045e-06, + "loss": 0.009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24240 + }, + { + "epoch": 7.05557171952284, + "flow_matching_loss": 0.009139121929183602, + "grad_norm": 0.06286873668432236, + "learning_rate": 2.914502731629641e-06, + "loss": 0.0091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24250 + }, + { + "epoch": 7.058481233633983, + "flow_matching_loss": 0.011486322386190295, + "grad_norm": 0.08058319240808487, + "learning_rate": 2.904716176408451e-06, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24260 + }, + { + "epoch": 7.0613907477451265, + "flow_matching_loss": 0.010269630793482065, + "grad_norm": 0.06219126284122467, + "learning_rate": 2.8949443180976172e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24270 + }, + { + "epoch": 7.06430026185627, + "flow_matching_loss": 0.010556462593376637, + "grad_norm": 0.06910209357738495, + "learning_rate": 2.885187168570864e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24280 + }, + { + "epoch": 7.067209775967413, + "flow_matching_loss": 0.011401872895658017, + "grad_norm": 0.06471602618694305, + "learning_rate": 2.875444739684058e-06, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24290 + }, + { + "epoch": 7.070119290078557, + "flow_matching_loss": 0.009715716121718287, + "grad_norm": 0.10029185563325882, + "learning_rate": 2.865717043275165e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24300 + }, + { + "epoch": 7.0730288041897005, + "flow_matching_loss": 0.01138574993237853, + "grad_norm": 0.07054685801267624, + "learning_rate": 2.8560040911642528e-06, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24310 + }, + { + "epoch": 7.075938318300844, + "flow_matching_loss": 0.011130404891446233, + "grad_norm": 0.0701863095164299, + "learning_rate": 2.846305895153472e-06, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24320 + }, + { + "epoch": 7.078847832411987, + "flow_matching_loss": 0.011275877803564071, + "grad_norm": 0.07447267323732376, + "learning_rate": 2.8366224670270487e-06, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24330 + }, + { + "epoch": 7.081757346523131, + "flow_matching_loss": 0.008811836829409003, + "grad_norm": 0.05420175567269325, + "learning_rate": 2.8269538185512535e-06, + "loss": 0.0088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24340 + }, + { + "epoch": 7.084666860634274, + "flow_matching_loss": 0.009845378110185266, + "grad_norm": 0.06912465393543243, + "learning_rate": 2.817299961474417e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24350 + }, + { + "epoch": 7.087576374745417, + "flow_matching_loss": 0.00991758517920971, + "grad_norm": 0.06082571670413017, + "learning_rate": 2.8076609075268816e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24360 + }, + { + "epoch": 7.090485888856561, + "flow_matching_loss": 0.011394975753501058, + "grad_norm": 0.07213622331619263, + "learning_rate": 2.798036668421006e-06, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24370 + }, + { + "epoch": 7.093395402967705, + "flow_matching_loss": 0.009753242740407586, + "grad_norm": 0.10523683577775955, + "learning_rate": 2.7884272558511433e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24380 + }, + { + "epoch": 7.0963049170788475, + "flow_matching_loss": 0.010757237067446112, + "grad_norm": 0.0601082481443882, + "learning_rate": 2.7788326814936456e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24390 + }, + { + "epoch": 7.099214431189991, + "flow_matching_loss": 0.011497257370501756, + "grad_norm": 0.05862673744559288, + "learning_rate": 2.7692529570068176e-06, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24400 + }, + { + "epoch": 7.102123945301135, + "flow_matching_loss": 0.012057247292250394, + "grad_norm": 0.08381319046020508, + "learning_rate": 2.759688094030933e-06, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24410 + }, + { + "epoch": 7.105033459412278, + "flow_matching_loss": 0.010467425175011157, + "grad_norm": 0.07310295104980469, + "learning_rate": 2.750138104188201e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24420 + }, + { + "epoch": 7.107942973523421, + "flow_matching_loss": 0.009877457469701766, + "grad_norm": 0.07564050704240799, + "learning_rate": 2.740602999082759e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24430 + }, + { + "epoch": 7.110852487634565, + "flow_matching_loss": 0.009642329812049866, + "grad_norm": 0.0631658211350441, + "learning_rate": 2.731082790300654e-06, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24440 + }, + { + "epoch": 7.113762001745709, + "flow_matching_loss": 0.01055844109505415, + "grad_norm": 0.04973910376429558, + "learning_rate": 2.7215774894098474e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24450 + }, + { + "epoch": 7.116671515856852, + "flow_matching_loss": 0.008901829272508622, + "grad_norm": 0.06574303656816483, + "learning_rate": 2.7120871079601696e-06, + "loss": 0.0089, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24460 + }, + { + "epoch": 7.119581029967995, + "flow_matching_loss": 0.010570053244009614, + "grad_norm": 0.07553192228078842, + "learning_rate": 2.7026116574833314e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24470 + }, + { + "epoch": 7.122490544079139, + "flow_matching_loss": 0.00977697903290391, + "grad_norm": 0.08560416847467422, + "learning_rate": 2.6931511494928953e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24480 + }, + { + "epoch": 7.125400058190282, + "flow_matching_loss": 0.009887263737618923, + "grad_norm": 0.06702516227960587, + "learning_rate": 2.683705595484275e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24490 + }, + { + "epoch": 7.128309572301426, + "flow_matching_loss": 0.010930808633565903, + "grad_norm": 0.06500335782766342, + "learning_rate": 2.6742750069347014e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24500 + }, + { + "epoch": 7.131219086412569, + "flow_matching_loss": 0.010952184675261378, + "grad_norm": 0.07608911395072937, + "learning_rate": 2.664859395303239e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24510 + }, + { + "epoch": 7.134128600523712, + "flow_matching_loss": 0.012191849807277323, + "grad_norm": 0.09180646389722824, + "learning_rate": 2.655458772030739e-06, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24520 + }, + { + "epoch": 7.137038114634856, + "flow_matching_loss": 0.009367563156411051, + "grad_norm": 0.09987214207649231, + "learning_rate": 2.6460731485398403e-06, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24530 + }, + { + "epoch": 7.1399476287459995, + "flow_matching_loss": 0.011349564837291836, + "grad_norm": 0.0801633894443512, + "learning_rate": 2.6367025362349695e-06, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24540 + }, + { + "epoch": 7.142857142857143, + "flow_matching_loss": 0.010607979074120522, + "grad_norm": 0.09070931375026703, + "learning_rate": 2.6273469465022965e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24550 + }, + { + "epoch": 7.145766656968286, + "flow_matching_loss": 0.010243060113862158, + "grad_norm": 0.0714074969291687, + "learning_rate": 2.618006390709753e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24560 + }, + { + "epoch": 7.14867617107943, + "flow_matching_loss": 0.009760004235431552, + "grad_norm": 0.052160173654556274, + "learning_rate": 2.60868088020699e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24570 + }, + { + "epoch": 7.1515856851905735, + "flow_matching_loss": 0.009900157991796731, + "grad_norm": 0.06292671710252762, + "learning_rate": 2.599370426325382e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24580 + }, + { + "epoch": 7.154495199301716, + "flow_matching_loss": 0.011241854121908545, + "grad_norm": 0.0675458163022995, + "learning_rate": 2.5900750403780093e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24590 + }, + { + "epoch": 7.15740471341286, + "flow_matching_loss": 0.009768071165308357, + "grad_norm": 0.06440471857786179, + "learning_rate": 2.580794733659639e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24600 + }, + { + "epoch": 7.160314227524004, + "flow_matching_loss": 0.011334720766171813, + "grad_norm": 0.0672619566321373, + "learning_rate": 2.5715295174467236e-06, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24610 + }, + { + "epoch": 7.1632237416351465, + "flow_matching_loss": 0.009523514471948148, + "grad_norm": 0.07075752317905426, + "learning_rate": 2.5622794029973746e-06, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24620 + }, + { + "epoch": 7.16613325574629, + "flow_matching_loss": 0.010643337992951274, + "grad_norm": 0.06719788163900375, + "learning_rate": 2.553044401551351e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24630 + }, + { + "epoch": 7.169042769857434, + "flow_matching_loss": 0.011546381749212742, + "grad_norm": 0.0768919363617897, + "learning_rate": 2.54382452433005e-06, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24640 + }, + { + "epoch": 7.171952283968578, + "flow_matching_loss": 0.01013157581910491, + "grad_norm": 0.09108243137598038, + "learning_rate": 2.53461978253649e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24650 + }, + { + "epoch": 7.1748617980797205, + "flow_matching_loss": 0.009426108002662659, + "grad_norm": 0.07831895351409912, + "learning_rate": 2.5254301873553065e-06, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24660 + }, + { + "epoch": 7.177771312190864, + "flow_matching_loss": 0.010255171358585358, + "grad_norm": 0.06395012885332108, + "learning_rate": 2.5162557499527184e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24670 + }, + { + "epoch": 7.180680826302008, + "flow_matching_loss": 0.009747090796008707, + "grad_norm": 0.07311573624610901, + "learning_rate": 2.5070964814765335e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24680 + }, + { + "epoch": 7.183590340413151, + "flow_matching_loss": 0.010674394201487303, + "grad_norm": 0.10208986699581146, + "learning_rate": 2.4979523930561295e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24690 + }, + { + "epoch": 7.186499854524294, + "flow_matching_loss": 0.009112580912187696, + "grad_norm": 0.07840323448181152, + "learning_rate": 2.4888234958024333e-06, + "loss": 0.0091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24700 + }, + { + "epoch": 7.189409368635438, + "flow_matching_loss": 0.011052297754213214, + "grad_norm": 0.0854983851313591, + "learning_rate": 2.479709800807915e-06, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24710 + }, + { + "epoch": 7.192318882746581, + "flow_matching_loss": 0.010216060699895025, + "grad_norm": 0.07867439091205597, + "learning_rate": 2.4706113191465767e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24720 + }, + { + "epoch": 7.195228396857725, + "flow_matching_loss": 0.010005846899002791, + "grad_norm": 0.07361030578613281, + "learning_rate": 2.46152806187393e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24730 + }, + { + "epoch": 7.198137910968868, + "flow_matching_loss": 0.009296417934820056, + "grad_norm": 0.07206222414970398, + "learning_rate": 2.452460040026989e-06, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24740 + }, + { + "epoch": 7.201047425080012, + "flow_matching_loss": 0.010903284000232816, + "grad_norm": 0.08609401434659958, + "learning_rate": 2.4434072646242555e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24750 + }, + { + "epoch": 7.203956939191155, + "flow_matching_loss": 0.009266528859734535, + "grad_norm": 0.07369790226221085, + "learning_rate": 2.4343697466657015e-06, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24760 + }, + { + "epoch": 7.206866453302299, + "flow_matching_loss": 0.008983802143484354, + "grad_norm": 0.08808577060699463, + "learning_rate": 2.42534749713277e-06, + "loss": 0.009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24770 + }, + { + "epoch": 7.209775967413442, + "flow_matching_loss": 0.010598273761570454, + "grad_norm": 0.052231959998607635, + "learning_rate": 2.4163405269883425e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24780 + }, + { + "epoch": 7.212685481524585, + "flow_matching_loss": 0.010745943570509554, + "grad_norm": 0.06367671489715576, + "learning_rate": 2.407348847176735e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24790 + }, + { + "epoch": 7.215594995635729, + "flow_matching_loss": 0.010628305189311505, + "grad_norm": 0.07489947974681854, + "learning_rate": 2.398372468623688e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24800 + }, + { + "epoch": 7.2185045097468725, + "flow_matching_loss": 0.009393510688096286, + "grad_norm": 0.05768919363617897, + "learning_rate": 2.3894114022363446e-06, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24810 + }, + { + "epoch": 7.221414023858015, + "flow_matching_loss": 0.009744117502123117, + "grad_norm": 0.08409295231103897, + "learning_rate": 2.3804656589032514e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24820 + }, + { + "epoch": 7.224323537969159, + "flow_matching_loss": 0.009927761694416403, + "grad_norm": 0.08216265588998795, + "learning_rate": 2.3715352494943237e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24830 + }, + { + "epoch": 7.227233052080303, + "flow_matching_loss": 0.010088860429823399, + "grad_norm": 0.07893350720405579, + "learning_rate": 2.3626201848608582e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24840 + }, + { + "epoch": 7.2301425661914465, + "flow_matching_loss": 0.010238938871771098, + "grad_norm": 0.07961757481098175, + "learning_rate": 2.353720475835494e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24850 + }, + { + "epoch": 7.233052080302589, + "flow_matching_loss": 0.009140241472050548, + "grad_norm": 0.07044759392738342, + "learning_rate": 2.344836133232215e-06, + "loss": 0.0091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24860 + }, + { + "epoch": 7.235961594413733, + "flow_matching_loss": 0.01075037051923573, + "grad_norm": 0.07178587466478348, + "learning_rate": 2.335967167846342e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24870 + }, + { + "epoch": 7.238871108524877, + "flow_matching_loss": 0.009798020217567682, + "grad_norm": 0.07600636780261993, + "learning_rate": 2.3271135904544993e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24880 + }, + { + "epoch": 7.2417806226360195, + "flow_matching_loss": 0.009211605321615935, + "grad_norm": 0.07996444404125214, + "learning_rate": 2.3182754118146206e-06, + "loss": 0.0092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24890 + }, + { + "epoch": 7.244690136747163, + "flow_matching_loss": 0.010249456437304617, + "grad_norm": 0.061552103608846664, + "learning_rate": 2.309452642665923e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24900 + }, + { + "epoch": 7.247599650858307, + "flow_matching_loss": 0.010604459140449763, + "grad_norm": 0.09355847537517548, + "learning_rate": 2.3006452937289065e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24910 + }, + { + "epoch": 7.25050916496945, + "flow_matching_loss": 0.010665127635002136, + "grad_norm": 0.07144126296043396, + "learning_rate": 2.291853375705327e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24920 + }, + { + "epoch": 7.2534186790805935, + "flow_matching_loss": 0.010348321823403239, + "grad_norm": 0.07466218620538712, + "learning_rate": 2.283076899278197e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24930 + }, + { + "epoch": 7.256328193191737, + "flow_matching_loss": 0.01058195298537612, + "grad_norm": 0.07314074784517288, + "learning_rate": 2.274315875111763e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24940 + }, + { + "epoch": 7.25923770730288, + "flow_matching_loss": 0.009761977847665549, + "grad_norm": 0.05763005092740059, + "learning_rate": 2.265570313851493e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24950 + }, + { + "epoch": 7.262147221414024, + "flow_matching_loss": 0.010001564351841807, + "grad_norm": 0.0700284093618393, + "learning_rate": 2.2568402261240685e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24960 + }, + { + "epoch": 7.265056735525167, + "flow_matching_loss": 0.010041861142963172, + "grad_norm": 0.06883878260850906, + "learning_rate": 2.248125622537373e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24970 + }, + { + "epoch": 7.267966249636311, + "flow_matching_loss": 0.012343378644436598, + "grad_norm": 0.07509274780750275, + "learning_rate": 2.239426513680467e-06, + "loss": 0.0123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24980 + }, + { + "epoch": 7.270875763747454, + "flow_matching_loss": 0.008436440024524927, + "grad_norm": 0.09596317261457443, + "learning_rate": 2.2307429101235915e-06, + "loss": 0.0084, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24990 + }, + { + "epoch": 7.273785277858598, + "flow_matching_loss": 0.010935846390202642, + "grad_norm": 0.07942938059568405, + "learning_rate": 2.222074822418145e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25000 + }, + { + "epoch": 7.276694791969741, + "flow_matching_loss": 0.010035132570192219, + "grad_norm": 0.07490196079015732, + "learning_rate": 2.213422261096668e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25010 + }, + { + "epoch": 7.279604306080884, + "flow_matching_loss": 0.009496475523337721, + "grad_norm": 0.08925238996744156, + "learning_rate": 2.204785236672835e-06, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25020 + }, + { + "epoch": 7.282513820192028, + "flow_matching_loss": 0.009912334755063057, + "grad_norm": 0.07054001092910767, + "learning_rate": 2.1961637596414536e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25030 + }, + { + "epoch": 7.285423334303172, + "flow_matching_loss": 0.011685167020186783, + "grad_norm": 0.07590518891811371, + "learning_rate": 2.1875578404784275e-06, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25040 + }, + { + "epoch": 7.288332848414315, + "flow_matching_loss": 0.009778102859854698, + "grad_norm": 0.06455443054437637, + "learning_rate": 2.1789674896407583e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25050 + }, + { + "epoch": 7.291242362525458, + "flow_matching_loss": 0.010060043027624487, + "grad_norm": 0.07595211267471313, + "learning_rate": 2.170392717566534e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25060 + }, + { + "epoch": 7.294151876636602, + "flow_matching_loss": 0.010209771804511547, + "grad_norm": 0.05872350186109543, + "learning_rate": 2.161833534674909e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25070 + }, + { + "epoch": 7.2970613907477455, + "flow_matching_loss": 0.010780793055891991, + "grad_norm": 0.0636904388666153, + "learning_rate": 2.1532899513661015e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25080 + }, + { + "epoch": 7.299970904858888, + "flow_matching_loss": 0.010154520440846682, + "grad_norm": 0.09404293447732925, + "learning_rate": 2.1447619780213682e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25090 + }, + { + "epoch": 7.302880418970032, + "flow_matching_loss": 0.009703642595559359, + "grad_norm": 0.05623862147331238, + "learning_rate": 2.136249625003001e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25100 + }, + { + "epoch": 7.305789933081176, + "flow_matching_loss": 0.010057944152504206, + "grad_norm": 0.05800950154662132, + "learning_rate": 2.1277529026543107e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25110 + }, + { + "epoch": 7.308699447192319, + "flow_matching_loss": 0.010545112565159798, + "grad_norm": 0.08040273934602737, + "learning_rate": 2.1192718212996188e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25120 + }, + { + "epoch": 7.311608961303462, + "flow_matching_loss": 0.008269529696553946, + "grad_norm": 0.08461315184831619, + "learning_rate": 2.110806391244235e-06, + "loss": 0.0083, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25130 + }, + { + "epoch": 7.314518475414606, + "flow_matching_loss": 0.01070059509947896, + "grad_norm": 0.07671215385198593, + "learning_rate": 2.1023566227744596e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25140 + }, + { + "epoch": 7.317427989525749, + "flow_matching_loss": 0.010666103940457106, + "grad_norm": 0.09083399176597595, + "learning_rate": 2.0939225261575563e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25150 + }, + { + "epoch": 7.3203375036368925, + "flow_matching_loss": 0.00945567972958088, + "grad_norm": 0.07892517745494843, + "learning_rate": 2.0855041116417473e-06, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25160 + }, + { + "epoch": 7.323247017748036, + "flow_matching_loss": 0.009432456037029625, + "grad_norm": 0.0829935297369957, + "learning_rate": 2.0771013894562003e-06, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25170 + }, + { + "epoch": 7.32615653185918, + "flow_matching_loss": 0.009781568124890328, + "grad_norm": 0.0714675709605217, + "learning_rate": 2.0687143698110113e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25180 + }, + { + "epoch": 7.329066045970323, + "flow_matching_loss": 0.010330332303419709, + "grad_norm": 0.07644734531641006, + "learning_rate": 2.060343062897208e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25190 + }, + { + "epoch": 7.3319755600814664, + "flow_matching_loss": 0.009753537643700839, + "grad_norm": 0.07324396073818207, + "learning_rate": 2.0519874788867127e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25200 + }, + { + "epoch": 7.33488507419261, + "flow_matching_loss": 0.01139302896335721, + "grad_norm": 0.1025027334690094, + "learning_rate": 2.043647627932347e-06, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25210 + }, + { + "epoch": 7.337794588303753, + "flow_matching_loss": 0.010001893900334836, + "grad_norm": 0.059781692922115326, + "learning_rate": 2.0353235201678193e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25220 + }, + { + "epoch": 7.340704102414897, + "flow_matching_loss": 0.009544132743030787, + "grad_norm": 0.06997289508581161, + "learning_rate": 2.0270151657076995e-06, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25230 + }, + { + "epoch": 7.34361361652604, + "flow_matching_loss": 0.011614193441346288, + "grad_norm": 0.06533019244670868, + "learning_rate": 2.0187225746474286e-06, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25240 + }, + { + "epoch": 7.346523130637183, + "flow_matching_loss": 0.009821071336045862, + "grad_norm": 0.0729142278432846, + "learning_rate": 2.0104457570632845e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25250 + }, + { + "epoch": 7.349432644748327, + "flow_matching_loss": 0.009436420910060406, + "grad_norm": 0.07212337106466293, + "learning_rate": 2.0021847230123775e-06, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25260 + }, + { + "epoch": 7.352342158859471, + "flow_matching_loss": 0.008815545635297895, + "grad_norm": 0.05086681619286537, + "learning_rate": 1.9939394825326494e-06, + "loss": 0.0088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25270 + }, + { + "epoch": 7.355251672970613, + "flow_matching_loss": 0.00930657093413174, + "grad_norm": 0.04929738491773605, + "learning_rate": 1.9857100456428385e-06, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25280 + }, + { + "epoch": 7.358161187081757, + "flow_matching_loss": 0.01012486764229834, + "grad_norm": 0.06992840021848679, + "learning_rate": 1.977496422342492e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25290 + }, + { + "epoch": 7.361070701192901, + "flow_matching_loss": 0.01017886190675199, + "grad_norm": 0.1028137058019638, + "learning_rate": 1.9692986226119345e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25300 + }, + { + "epoch": 7.363980215304045, + "flow_matching_loss": 0.01145113045349717, + "grad_norm": 0.10074733197689056, + "learning_rate": 1.9611166564122658e-06, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25310 + }, + { + "epoch": 7.366889729415187, + "flow_matching_loss": 0.010159315774217248, + "grad_norm": 0.08158684521913528, + "learning_rate": 1.9529505336853456e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25320 + }, + { + "epoch": 7.369799243526331, + "flow_matching_loss": 0.011658118246123195, + "grad_norm": 0.0576140396296978, + "learning_rate": 1.9448002643537833e-06, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25330 + }, + { + "epoch": 7.372708757637475, + "flow_matching_loss": 0.009017424518242478, + "grad_norm": 0.05918961018323898, + "learning_rate": 1.9366658583209205e-06, + "loss": 0.009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25340 + }, + { + "epoch": 7.375618271748618, + "flow_matching_loss": 0.009417259087786078, + "grad_norm": 0.07628355920314789, + "learning_rate": 1.9285473254708363e-06, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25350 + }, + { + "epoch": 7.378527785859761, + "flow_matching_loss": 0.01087144147604704, + "grad_norm": 0.0639435276389122, + "learning_rate": 1.9204446756683092e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25360 + }, + { + "epoch": 7.381437299970905, + "flow_matching_loss": 0.009983518160879612, + "grad_norm": 0.06778324395418167, + "learning_rate": 1.912357918758822e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25370 + }, + { + "epoch": 7.384346814082049, + "flow_matching_loss": 0.009571711765602231, + "grad_norm": 0.06916384398937225, + "learning_rate": 1.9042870645685478e-06, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25380 + }, + { + "epoch": 7.387256328193192, + "flow_matching_loss": 0.010623767552897335, + "grad_norm": 0.06961469352245331, + "learning_rate": 1.8962321229043322e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25390 + }, + { + "epoch": 7.390165842304335, + "flow_matching_loss": 0.011254333425313234, + "grad_norm": 0.06863929331302643, + "learning_rate": 1.8881931035536943e-06, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25400 + }, + { + "epoch": 7.393075356415479, + "flow_matching_loss": 0.009972949000075459, + "grad_norm": 0.07521859556436539, + "learning_rate": 1.8801700162847952e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25410 + }, + { + "epoch": 7.395984870526622, + "flow_matching_loss": 0.01064867302775383, + "grad_norm": 0.07025227695703506, + "learning_rate": 1.87216287084645e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25420 + }, + { + "epoch": 7.3988943846377655, + "flow_matching_loss": 0.010471457242965698, + "grad_norm": 0.11901270598173141, + "learning_rate": 1.8641716769680895e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25430 + }, + { + "epoch": 7.401803898748909, + "flow_matching_loss": 0.009352154843509198, + "grad_norm": 0.061213862150907516, + "learning_rate": 1.8561964443597662e-06, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25440 + }, + { + "epoch": 7.404713412860052, + "flow_matching_loss": 0.010258167469874024, + "grad_norm": 0.08280056715011597, + "learning_rate": 1.848237182712147e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25450 + }, + { + "epoch": 7.407622926971196, + "flow_matching_loss": 0.009729246702045202, + "grad_norm": 0.06491003930568695, + "learning_rate": 1.8402939016964788e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25460 + }, + { + "epoch": 7.410532441082339, + "flow_matching_loss": 0.010328479250892997, + "grad_norm": 0.054865770041942596, + "learning_rate": 1.832366610964601e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25470 + }, + { + "epoch": 7.413441955193482, + "flow_matching_loss": 0.009701339527964592, + "grad_norm": 0.11661362648010254, + "learning_rate": 1.8244553201489166e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25480 + }, + { + "epoch": 7.416351469304626, + "flow_matching_loss": 0.009435597574338317, + "grad_norm": 0.06790923327207565, + "learning_rate": 1.8165600388623882e-06, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25490 + }, + { + "epoch": 7.41926098341577, + "flow_matching_loss": 0.0102243279106915, + "grad_norm": 0.08551516383886337, + "learning_rate": 1.8086807766985331e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25500 + }, + { + "epoch": 7.422170497526913, + "flow_matching_loss": 0.009357222449034452, + "grad_norm": 0.06223909556865692, + "learning_rate": 1.8008175432313946e-06, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25510 + }, + { + "epoch": 7.425080011638056, + "flow_matching_loss": 0.009108431916683913, + "grad_norm": 0.07515736669301987, + "learning_rate": 1.7929703480155429e-06, + "loss": 0.0091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25520 + }, + { + "epoch": 7.4279895257492, + "flow_matching_loss": 0.01082046264782548, + "grad_norm": 0.09650807827711105, + "learning_rate": 1.7851392005860595e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25530 + }, + { + "epoch": 7.430899039860344, + "flow_matching_loss": 0.009948767814785242, + "grad_norm": 0.060923922806978226, + "learning_rate": 1.7773241104585243e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25540 + }, + { + "epoch": 7.433808553971486, + "flow_matching_loss": 0.009292423818260431, + "grad_norm": 0.08448682725429535, + "learning_rate": 1.7695250871290108e-06, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25550 + }, + { + "epoch": 7.43671806808263, + "flow_matching_loss": 0.010434665670618414, + "grad_norm": 0.06722834706306458, + "learning_rate": 1.761742140074074e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25560 + }, + { + "epoch": 7.439627582193774, + "flow_matching_loss": 0.009703850513324142, + "grad_norm": 0.06925749033689499, + "learning_rate": 1.7539752787507208e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25570 + }, + { + "epoch": 7.442537096304917, + "flow_matching_loss": 0.01077431538142264, + "grad_norm": 0.06824279576539993, + "learning_rate": 1.7462245125964233e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25580 + }, + { + "epoch": 7.44544661041606, + "flow_matching_loss": 0.011612507048994303, + "grad_norm": 0.0748504176735878, + "learning_rate": 1.7384898510290931e-06, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25590 + }, + { + "epoch": 7.448356124527204, + "flow_matching_loss": 0.010346911055967212, + "grad_norm": 0.06176278367638588, + "learning_rate": 1.730771303447069e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25600 + }, + { + "epoch": 7.451265638638348, + "flow_matching_loss": 0.00990656572394073, + "grad_norm": 0.055678531527519226, + "learning_rate": 1.7230688792291204e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25610 + }, + { + "epoch": 7.454175152749491, + "flow_matching_loss": 0.011770316772162915, + "grad_norm": 0.08738914132118225, + "learning_rate": 1.7153825877344175e-06, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25620 + }, + { + "epoch": 7.457084666860634, + "flow_matching_loss": 0.00963967926800251, + "grad_norm": 0.05805916339159012, + "learning_rate": 1.7077124383025278e-06, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25630 + }, + { + "epoch": 7.459994180971778, + "flow_matching_loss": 0.00934621193446219, + "grad_norm": 0.07340722531080246, + "learning_rate": 1.7000584402534058e-06, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25640 + }, + { + "epoch": 7.462903695082921, + "flow_matching_loss": 0.009750072471797466, + "grad_norm": 0.05694437026977539, + "learning_rate": 1.6924206028873774e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25650 + }, + { + "epoch": 7.4658132091940645, + "flow_matching_loss": 0.010377070540562272, + "grad_norm": 0.058469243347644806, + "learning_rate": 1.6847989354851434e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25660 + }, + { + "epoch": 7.468722723305208, + "flow_matching_loss": 0.008863866794854403, + "grad_norm": 0.05467579886317253, + "learning_rate": 1.6771934473077443e-06, + "loss": 0.0089, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25670 + }, + { + "epoch": 7.471632237416351, + "flow_matching_loss": 0.00982497688382864, + "grad_norm": 0.08236885070800781, + "learning_rate": 1.669604147596563e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25680 + }, + { + "epoch": 7.474541751527495, + "flow_matching_loss": 0.009791796142235399, + "grad_norm": 0.07751026749610901, + "learning_rate": 1.662031045573319e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25690 + }, + { + "epoch": 7.4774512656386385, + "flow_matching_loss": 0.010253849066793918, + "grad_norm": 0.07491634041070938, + "learning_rate": 1.6544741504400402e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25700 + }, + { + "epoch": 7.480360779749782, + "flow_matching_loss": 0.008955058362334966, + "grad_norm": 0.10619127750396729, + "learning_rate": 1.6469334713790712e-06, + "loss": 0.009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25710 + }, + { + "epoch": 7.483270293860925, + "flow_matching_loss": 0.01055750884115696, + "grad_norm": 0.10607615113258362, + "learning_rate": 1.639409017553048e-06, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25720 + }, + { + "epoch": 7.486179807972069, + "flow_matching_loss": 0.009650717815384268, + "grad_norm": 0.04987390339374542, + "learning_rate": 1.6319007981048884e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25730 + }, + { + "epoch": 7.489089322083212, + "flow_matching_loss": 0.00940116373822093, + "grad_norm": 0.06237965077161789, + "learning_rate": 1.6244088221577896e-06, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25740 + }, + { + "epoch": 7.491998836194355, + "flow_matching_loss": 0.010984189203009009, + "grad_norm": 0.07307201623916626, + "learning_rate": 1.6169330988152081e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25750 + }, + { + "epoch": 7.494908350305499, + "flow_matching_loss": 0.009841401968151331, + "grad_norm": 0.07166663557291031, + "learning_rate": 1.6094736371608475e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25760 + }, + { + "epoch": 7.497817864416643, + "flow_matching_loss": 0.00980448378250003, + "grad_norm": 0.08344807475805283, + "learning_rate": 1.602030446258666e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25770 + }, + { + "epoch": 7.5007273785277855, + "flow_matching_loss": 0.009982823440805078, + "grad_norm": 0.07501685619354248, + "learning_rate": 1.5946035351528388e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25780 + }, + { + "epoch": 7.503636892638929, + "flow_matching_loss": 0.008916236413642763, + "grad_norm": 0.07062829285860062, + "learning_rate": 1.5871929128677621e-06, + "loss": 0.0089, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25790 + }, + { + "epoch": 7.506546406750073, + "flow_matching_loss": 0.01016015661880374, + "grad_norm": 0.07724898308515549, + "learning_rate": 1.5797985884080412e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25800 + }, + { + "epoch": 7.509455920861216, + "flow_matching_loss": 0.010300452122464776, + "grad_norm": 0.06938227266073227, + "learning_rate": 1.572420570758476e-06, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25810 + }, + { + "epoch": 7.512365434972359, + "flow_matching_loss": 0.011675127409398555, + "grad_norm": 0.06234145537018776, + "learning_rate": 1.5650588688840567e-06, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25820 + }, + { + "epoch": 7.515274949083503, + "flow_matching_loss": 0.011230405187234282, + "grad_norm": 0.08129651844501495, + "learning_rate": 1.5577134917299413e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25830 + }, + { + "epoch": 7.518184463194647, + "flow_matching_loss": 0.009578860830515623, + "grad_norm": 0.09260963648557663, + "learning_rate": 1.5503844482214618e-06, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25840 + }, + { + "epoch": 7.52109397730579, + "flow_matching_loss": 0.009312378615140915, + "grad_norm": 0.06343889236450195, + "learning_rate": 1.5430717472640932e-06, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25850 + }, + { + "epoch": 7.524003491416933, + "flow_matching_loss": 0.010227884678170086, + "grad_norm": 0.07722092419862747, + "learning_rate": 1.5357753977434541e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25860 + }, + { + "epoch": 7.526913005528077, + "flow_matching_loss": 0.010432990081608296, + "grad_norm": 0.08217858523130417, + "learning_rate": 1.5284954085253043e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25870 + }, + { + "epoch": 7.52982251963922, + "flow_matching_loss": 0.010010290937498213, + "grad_norm": 0.05672130361199379, + "learning_rate": 1.5212317884555115e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25880 + }, + { + "epoch": 7.532732033750364, + "flow_matching_loss": 0.00957383019849658, + "grad_norm": 0.07402228564023972, + "learning_rate": 1.5139845463600605e-06, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25890 + }, + { + "epoch": 7.535641547861507, + "flow_matching_loss": 0.00936931255273521, + "grad_norm": 0.07390867173671722, + "learning_rate": 1.5067536910450352e-06, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25900 + }, + { + "epoch": 7.538551061972651, + "flow_matching_loss": 0.007981835724785924, + "grad_norm": 0.058439042419195175, + "learning_rate": 1.4995392312966044e-06, + "loss": 0.008, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25910 + }, + { + "epoch": 7.541460576083794, + "flow_matching_loss": 0.009504101518541575, + "grad_norm": 0.07010823488235474, + "learning_rate": 1.492341175881014e-06, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25920 + }, + { + "epoch": 7.5443700901949375, + "flow_matching_loss": 0.010413714265450836, + "grad_norm": 0.07156502455472946, + "learning_rate": 1.4851595335445895e-06, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25930 + }, + { + "epoch": 7.547279604306081, + "flow_matching_loss": 0.010847284784540534, + "grad_norm": 0.07980712503194809, + "learning_rate": 1.4779943130136953e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25940 + }, + { + "epoch": 7.550189118417224, + "flow_matching_loss": 0.008907918957993389, + "grad_norm": 0.0643768385052681, + "learning_rate": 1.4708455229947537e-06, + "loss": 0.0089, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25950 + }, + { + "epoch": 7.553098632528368, + "flow_matching_loss": 0.009571415232494473, + "grad_norm": 0.07092766463756561, + "learning_rate": 1.4637131721742148e-06, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25960 + }, + { + "epoch": 7.5560081466395115, + "flow_matching_loss": 0.009881845815107227, + "grad_norm": 0.06024826690554619, + "learning_rate": 1.4565972692185626e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25970 + }, + { + "epoch": 7.558917660750654, + "flow_matching_loss": 0.009659718489274383, + "grad_norm": 0.09420550614595413, + "learning_rate": 1.4494978227742855e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25980 + }, + { + "epoch": 7.561827174861798, + "flow_matching_loss": 0.010735076572746038, + "grad_norm": 0.04849692061543465, + "learning_rate": 1.442414841467885e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25990 + }, + { + "epoch": 7.564736688972942, + "flow_matching_loss": 0.010487677063792944, + "grad_norm": 0.09542535245418549, + "learning_rate": 1.4353483339058498e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26000 + }, + { + "epoch": 7.5676462030840845, + "flow_matching_loss": 0.009410757385194302, + "grad_norm": 0.058832138776779175, + "learning_rate": 1.4282983086746498e-06, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26010 + }, + { + "epoch": 7.570555717195228, + "flow_matching_loss": 0.008793372008949519, + "grad_norm": 0.05546681582927704, + "learning_rate": 1.4212647743407325e-06, + "loss": 0.0088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26020 + }, + { + "epoch": 7.573465231306372, + "flow_matching_loss": 0.01003286479972303, + "grad_norm": 0.06521807610988617, + "learning_rate": 1.4142477394505065e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26030 + }, + { + "epoch": 7.576374745417516, + "flow_matching_loss": 0.009989116434007883, + "grad_norm": 0.07820935547351837, + "learning_rate": 1.4072472125303299e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26040 + }, + { + "epoch": 7.5792842595286585, + "flow_matching_loss": 0.009758471557870508, + "grad_norm": 0.08272764086723328, + "learning_rate": 1.4002632020865031e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26050 + }, + { + "epoch": 7.582193773639802, + "flow_matching_loss": 0.009589159348979592, + "grad_norm": 0.07693710178136826, + "learning_rate": 1.393295716605258e-06, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26060 + }, + { + "epoch": 7.585103287750946, + "flow_matching_loss": 0.010201762337237597, + "grad_norm": 0.0680125430226326, + "learning_rate": 1.386344764552742e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26070 + }, + { + "epoch": 7.588012801862089, + "flow_matching_loss": 0.008933221269398928, + "grad_norm": 0.07082618027925491, + "learning_rate": 1.3794103543750246e-06, + "loss": 0.0089, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26080 + }, + { + "epoch": 7.590922315973232, + "flow_matching_loss": 0.00907397917471826, + "grad_norm": 0.058743707835674286, + "learning_rate": 1.3724924944980637e-06, + "loss": 0.0091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26090 + }, + { + "epoch": 7.593831830084376, + "flow_matching_loss": 0.00875417129136622, + "grad_norm": 0.05365953966975212, + "learning_rate": 1.365591193327712e-06, + "loss": 0.0088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26100 + }, + { + "epoch": 7.59674134419552, + "flow_matching_loss": 0.010974767291918397, + "grad_norm": 0.0593426376581192, + "learning_rate": 1.3587064592496984e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26110 + }, + { + "epoch": 7.599650858306663, + "flow_matching_loss": 0.011386786587536335, + "grad_norm": 0.08127536624670029, + "learning_rate": 1.3518383006296292e-06, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26120 + }, + { + "epoch": 7.602560372417806, + "flow_matching_loss": 0.008994166972115636, + "grad_norm": 0.07823975384235382, + "learning_rate": 1.344986725812959e-06, + "loss": 0.009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26130 + }, + { + "epoch": 7.605469886528949, + "flow_matching_loss": 0.0092685226816684, + "grad_norm": 0.12044209241867065, + "learning_rate": 1.3381517431250046e-06, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26140 + }, + { + "epoch": 7.608379400640093, + "flow_matching_loss": 0.009854571847245096, + "grad_norm": 0.05409906059503555, + "learning_rate": 1.3313333608709095e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26150 + }, + { + "epoch": 7.611288914751237, + "flow_matching_loss": 0.010222713556140661, + "grad_norm": 0.0848807543516159, + "learning_rate": 1.3245315873356528e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26160 + }, + { + "epoch": 7.61419842886238, + "flow_matching_loss": 0.009488109033554792, + "grad_norm": 0.06383969634771347, + "learning_rate": 1.3177464307840304e-06, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26170 + }, + { + "epoch": 7.617107942973523, + "flow_matching_loss": 0.00937518123537302, + "grad_norm": 0.06259596347808838, + "learning_rate": 1.310977899460642e-06, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26180 + }, + { + "epoch": 7.620017457084667, + "flow_matching_loss": 0.01007348638959229, + "grad_norm": 0.05703268200159073, + "learning_rate": 1.3042260015899006e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26190 + }, + { + "epoch": 7.6229269711958105, + "flow_matching_loss": 0.010906883236020803, + "grad_norm": 0.08795327693223953, + "learning_rate": 1.2974907453759948e-06, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26200 + }, + { + "epoch": 7.625836485306953, + "flow_matching_loss": 0.010974465217441321, + "grad_norm": 0.08677050471305847, + "learning_rate": 1.2907721390028937e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26210 + }, + { + "epoch": 7.628745999418097, + "flow_matching_loss": 0.010465735988691448, + "grad_norm": 0.08169075101613998, + "learning_rate": 1.2840701906343417e-06, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26220 + }, + { + "epoch": 7.631655513529241, + "flow_matching_loss": 0.010755849909037352, + "grad_norm": 0.078217513859272, + "learning_rate": 1.2773849084138312e-06, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26230 + }, + { + "epoch": 7.6345650276403845, + "flow_matching_loss": 0.008884300850331784, + "grad_norm": 0.06748068332672119, + "learning_rate": 1.2707163004646177e-06, + "loss": 0.0089, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26240 + }, + { + "epoch": 7.637474541751527, + "flow_matching_loss": 0.0096301996614784, + "grad_norm": 0.07132003456354141, + "learning_rate": 1.2640643748896879e-06, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26250 + }, + { + "epoch": 7.640384055862671, + "flow_matching_loss": 0.00960582559928298, + "grad_norm": 0.05631144717335701, + "learning_rate": 1.257429139771753e-06, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26260 + }, + { + "epoch": 7.643293569973815, + "flow_matching_loss": 0.010030721873044967, + "grad_norm": 0.0837254524230957, + "learning_rate": 1.2508106031732553e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26270 + }, + { + "epoch": 7.6462030840849575, + "flow_matching_loss": 0.009425458731129766, + "grad_norm": 0.05717676877975464, + "learning_rate": 1.2442087731363366e-06, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26280 + }, + { + "epoch": 7.649112598196101, + "flow_matching_loss": 0.01023601358756423, + "grad_norm": 0.08134057372808456, + "learning_rate": 1.237623657682847e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26290 + }, + { + "epoch": 7.652022112307245, + "flow_matching_loss": 0.010048351436853408, + "grad_norm": 0.08659128844738007, + "learning_rate": 1.2310552648143208e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26300 + }, + { + "epoch": 7.654931626418388, + "flow_matching_loss": 0.009271671855822205, + "grad_norm": 0.0713469386100769, + "learning_rate": 1.2245036025119738e-06, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26310 + }, + { + "epoch": 7.6578411405295315, + "flow_matching_loss": 0.009694171976298094, + "grad_norm": 0.07888574153184891, + "learning_rate": 1.2179686787366962e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26320 + }, + { + "epoch": 7.660750654640675, + "flow_matching_loss": 0.011051851464435459, + "grad_norm": 0.08173272013664246, + "learning_rate": 1.2114505014290329e-06, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26330 + }, + { + "epoch": 7.663660168751818, + "flow_matching_loss": 0.011144749075174331, + "grad_norm": 0.06337563693523407, + "learning_rate": 1.2049490785091833e-06, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26340 + }, + { + "epoch": 7.666569682862962, + "flow_matching_loss": 0.011171266855672001, + "grad_norm": 0.06142672523856163, + "learning_rate": 1.1984644178769949e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26350 + }, + { + "epoch": 7.669479196974105, + "flow_matching_loss": 0.009701981442049146, + "grad_norm": 0.06182142347097397, + "learning_rate": 1.1919965274119381e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26360 + }, + { + "epoch": 7.672388711085249, + "flow_matching_loss": 0.009547146409749985, + "grad_norm": 0.06847891956567764, + "learning_rate": 1.18554541497311e-06, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26370 + }, + { + "epoch": 7.675298225196392, + "flow_matching_loss": 0.009726633317768574, + "grad_norm": 0.06491076946258545, + "learning_rate": 1.1791110883992213e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26380 + }, + { + "epoch": 7.678207739307536, + "flow_matching_loss": 0.00945090726017952, + "grad_norm": 0.061868228018283844, + "learning_rate": 1.1726935555085817e-06, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26390 + }, + { + "epoch": 7.681117253418679, + "flow_matching_loss": 0.010131673701107503, + "grad_norm": 0.07049606740474701, + "learning_rate": 1.166292824099105e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26400 + }, + { + "epoch": 7.684026767529822, + "flow_matching_loss": 0.00919355652295053, + "grad_norm": 0.06347928196191788, + "learning_rate": 1.1599089019482768e-06, + "loss": 0.0092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26410 + }, + { + "epoch": 7.686936281640966, + "flow_matching_loss": 0.01073927991092205, + "grad_norm": 0.05818722024559975, + "learning_rate": 1.1535417968131718e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26420 + }, + { + "epoch": 7.68984579575211, + "flow_matching_loss": 0.01006984682753682, + "grad_norm": 0.07978253066539764, + "learning_rate": 1.147191516430418e-06, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26430 + }, + { + "epoch": 7.692755309863253, + "flow_matching_loss": 0.00908002844080329, + "grad_norm": 0.06418143212795258, + "learning_rate": 1.140858068516205e-06, + "loss": 0.0091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26440 + }, + { + "epoch": 7.695664823974396, + "flow_matching_loss": 0.009853215469047428, + "grad_norm": 0.06959497183561325, + "learning_rate": 1.1345414607662736e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26450 + }, + { + "epoch": 7.69857433808554, + "flow_matching_loss": 0.011003611609339714, + "grad_norm": 0.06348992139101028, + "learning_rate": 1.1282417008558937e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26460 + }, + { + "epoch": 7.7014838521966835, + "flow_matching_loss": 0.008562624221667647, + "grad_norm": 0.08227016776800156, + "learning_rate": 1.1219587964398674e-06, + "loss": 0.0086, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26470 + }, + { + "epoch": 7.704393366307826, + "flow_matching_loss": 0.011008466780185699, + "grad_norm": 0.0714375376701355, + "learning_rate": 1.1156927551525198e-06, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26480 + }, + { + "epoch": 7.70730288041897, + "flow_matching_loss": 0.009611369529739022, + "grad_norm": 0.08908835053443909, + "learning_rate": 1.109443584607675e-06, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26490 + }, + { + "epoch": 7.710212394530114, + "flow_matching_loss": 0.009950731322169304, + "grad_norm": 0.07369742542505264, + "learning_rate": 1.1032112923986708e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26500 + }, + { + "epoch": 7.713121908641257, + "flow_matching_loss": 0.008422464225441217, + "grad_norm": 0.07307665795087814, + "learning_rate": 1.0969958860983299e-06, + "loss": 0.0084, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26510 + }, + { + "epoch": 7.7160314227524, + "flow_matching_loss": 0.01018064389936626, + "grad_norm": 0.06449777632951736, + "learning_rate": 1.0907973732589537e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26520 + }, + { + "epoch": 7.718940936863544, + "flow_matching_loss": 0.011689583957195281, + "grad_norm": 0.08869370818138123, + "learning_rate": 1.0846157614123215e-06, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26530 + }, + { + "epoch": 7.721850450974687, + "flow_matching_loss": 0.00999866663478315, + "grad_norm": 0.07440363615751266, + "learning_rate": 1.078451058069671e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26540 + }, + { + "epoch": 7.7247599650858305, + "flow_matching_loss": 0.008960275771096348, + "grad_norm": 0.0694521814584732, + "learning_rate": 1.0723032707217056e-06, + "loss": 0.009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26550 + }, + { + "epoch": 7.727669479196974, + "flow_matching_loss": 0.010038303397595882, + "grad_norm": 0.08308684825897217, + "learning_rate": 1.066172406838558e-06, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26560 + }, + { + "epoch": 7.730578993308118, + "flow_matching_loss": 0.01065055588260293, + "grad_norm": 0.08395369350910187, + "learning_rate": 1.0600584738698127e-06, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26570 + }, + { + "epoch": 7.733488507419261, + "flow_matching_loss": 0.00916396975517273, + "grad_norm": 0.061812225729227066, + "learning_rate": 1.0539614792444735e-06, + "loss": 0.0092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26580 + }, + { + "epoch": 7.7363980215304045, + "flow_matching_loss": 0.009880801662802696, + "grad_norm": 0.07498515397310257, + "learning_rate": 1.0478814303709627e-06, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26590 + }, + { + "epoch": 7.739307535641548, + "flow_matching_loss": 0.009788538096472621, + "grad_norm": 0.07655282318592072, + "learning_rate": 1.041818334637109e-06, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26600 + }, + { + "epoch": 7.742217049752691, + "flow_matching_loss": 0.010219514556229114, + "grad_norm": 0.07203924655914307, + "learning_rate": 1.0357721994101526e-06, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26610 + }, + { + "epoch": 7.745126563863835, + "flow_matching_loss": 0.00901625449769199, + "grad_norm": 0.06788650900125504, + "learning_rate": 1.0297430320367135e-06, + "loss": 0.009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26620 + }, + { + "epoch": 7.748036077974978, + "flow_matching_loss": 0.011196648050099612, + "grad_norm": 0.06300666183233261, + "learning_rate": 1.0237308398428014e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26630 + }, + { + "epoch": 7.750945592086122, + "flow_matching_loss": 0.009742135182023048, + "grad_norm": 0.07064953446388245, + "learning_rate": 1.0177356301337925e-06, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26640 + }, + { + "epoch": 7.753855106197265, + "flow_matching_loss": 0.009576794132590294, + "grad_norm": 0.13109442591667175, + "learning_rate": 1.0117574101944348e-06, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26650 + }, + { + "epoch": 7.756764620308409, + "flow_matching_loss": 0.01117850742302835, + "grad_norm": 0.09482118487358093, + "learning_rate": 1.0057961872888293e-06, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26660 + }, + { + "epoch": 7.7596741344195515, + "flow_matching_loss": 0.011212746985256672, + "grad_norm": 0.07110675424337387, + "learning_rate": 9.998519686604235e-07, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26670 + }, + { + "epoch": 7.762583648530695, + "flow_matching_loss": 0.010211790073662996, + "grad_norm": 0.07877244055271149, + "learning_rate": 9.939247615320047e-07, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26680 + }, + { + "epoch": 7.765493162641839, + "flow_matching_loss": 0.010433230269700288, + "grad_norm": 0.07209515571594238, + "learning_rate": 9.880145731056834e-07, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26690 + }, + { + "epoch": 7.768402676752983, + "flow_matching_loss": 0.009144902788102626, + "grad_norm": 0.07155667990446091, + "learning_rate": 9.821214105629034e-07, + "loss": 0.0091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26700 + }, + { + "epoch": 7.771312190864125, + "flow_matching_loss": 0.01029562745243311, + "grad_norm": 0.06331732869148254, + "learning_rate": 9.762452810644068e-07, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26710 + }, + { + "epoch": 7.774221704975269, + "flow_matching_loss": 0.009817787911742926, + "grad_norm": 0.06679579615592957, + "learning_rate": 9.703861917502487e-07, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26720 + }, + { + "epoch": 7.777131219086413, + "flow_matching_loss": 0.0096424940507859, + "grad_norm": 0.06255993992090225, + "learning_rate": 9.645441497397744e-07, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26730 + }, + { + "epoch": 7.780040733197556, + "flow_matching_loss": 0.009906693687662482, + "grad_norm": 0.10890898108482361, + "learning_rate": 9.587191621316154e-07, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26740 + }, + { + "epoch": 7.782950247308699, + "flow_matching_loss": 0.00938876885920763, + "grad_norm": 0.06641503423452377, + "learning_rate": 9.529112360036802e-07, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26750 + }, + { + "epoch": 7.785859761419843, + "flow_matching_loss": 0.009836552292108535, + "grad_norm": 0.08941333740949631, + "learning_rate": 9.471203784131449e-07, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26760 + }, + { + "epoch": 7.788769275530987, + "flow_matching_loss": 0.009304779907688498, + "grad_norm": 0.059894170612096786, + "learning_rate": 9.41346596396453e-07, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26770 + }, + { + "epoch": 7.79167878964213, + "flow_matching_loss": 0.010583551647141577, + "grad_norm": 0.08981550484895706, + "learning_rate": 9.355898969692889e-07, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26780 + }, + { + "epoch": 7.794588303753273, + "flow_matching_loss": 0.010345546109601856, + "grad_norm": 0.0494038388133049, + "learning_rate": 9.29850287126589e-07, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26790 + }, + { + "epoch": 7.797497817864417, + "flow_matching_loss": 0.010322661511600018, + "grad_norm": 0.07071777433156967, + "learning_rate": 9.24127773842518e-07, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26800 + }, + { + "epoch": 7.80040733197556, + "flow_matching_loss": 0.00962583776563406, + "grad_norm": 0.07406158000230789, + "learning_rate": 9.18422364070467e-07, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26810 + }, + { + "epoch": 7.8033168460867035, + "flow_matching_loss": 0.01019678576849401, + "grad_norm": 0.12371325492858887, + "learning_rate": 9.127340647430538e-07, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26820 + }, + { + "epoch": 7.806226360197847, + "flow_matching_loss": 0.009055807115510106, + "grad_norm": 0.06982080638408661, + "learning_rate": 9.070628827720929e-07, + "loss": 0.0091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26830 + }, + { + "epoch": 7.80913587430899, + "flow_matching_loss": 0.010490858042612673, + "grad_norm": 0.0881710946559906, + "learning_rate": 9.014088250486052e-07, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26840 + }, + { + "epoch": 7.812045388420134, + "flow_matching_loss": 0.009288632357493044, + "grad_norm": 0.051720235496759415, + "learning_rate": 8.95771898442807e-07, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26850 + }, + { + "epoch": 7.8149549025312774, + "flow_matching_loss": 0.00945050772279501, + "grad_norm": 0.0715634822845459, + "learning_rate": 8.90152109804091e-07, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26860 + }, + { + "epoch": 7.81786441664242, + "flow_matching_loss": 0.009584809513762593, + "grad_norm": 0.06635976582765579, + "learning_rate": 8.845494659610365e-07, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26870 + }, + { + "epoch": 7.820773930753564, + "flow_matching_loss": 0.009282743977382778, + "grad_norm": 0.05366949737071991, + "learning_rate": 8.789639737213812e-07, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26880 + }, + { + "epoch": 7.823683444864708, + "flow_matching_loss": 0.01100431764498353, + "grad_norm": 0.12047780305147171, + "learning_rate": 8.733956398720244e-07, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26890 + }, + { + "epoch": 7.826592958975851, + "flow_matching_loss": 0.008710878994315862, + "grad_norm": 0.052503883838653564, + "learning_rate": 8.678444711790157e-07, + "loss": 0.0087, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26900 + }, + { + "epoch": 7.829502473086994, + "flow_matching_loss": 0.010400314815342426, + "grad_norm": 0.08320474624633789, + "learning_rate": 8.623104743875493e-07, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26910 + }, + { + "epoch": 7.832411987198138, + "flow_matching_loss": 0.008755639754235744, + "grad_norm": 0.07085058838129044, + "learning_rate": 8.567936562219547e-07, + "loss": 0.0088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26920 + }, + { + "epoch": 7.835321501309282, + "flow_matching_loss": 0.00959904259070754, + "grad_norm": 0.05510720610618591, + "learning_rate": 8.512940233856847e-07, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26930 + }, + { + "epoch": 7.838231015420424, + "flow_matching_loss": 0.010503334878012537, + "grad_norm": 0.07539393752813339, + "learning_rate": 8.458115825613122e-07, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26940 + }, + { + "epoch": 7.841140529531568, + "flow_matching_loss": 0.010271371807903051, + "grad_norm": 0.07495416700839996, + "learning_rate": 8.403463404105199e-07, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26950 + }, + { + "epoch": 7.844050043642712, + "flow_matching_loss": 0.010769420675933361, + "grad_norm": 0.10015255957841873, + "learning_rate": 8.348983035740926e-07, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26960 + }, + { + "epoch": 7.846959557753856, + "flow_matching_loss": 0.010784434201195837, + "grad_norm": 0.0591096468269825, + "learning_rate": 8.294674786719064e-07, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26970 + }, + { + "epoch": 7.849869071864998, + "flow_matching_loss": 0.00963052837178111, + "grad_norm": 0.06626389920711517, + "learning_rate": 8.240538723029295e-07, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26980 + }, + { + "epoch": 7.852778585976142, + "flow_matching_loss": 0.011111474689096212, + "grad_norm": 0.08680056035518646, + "learning_rate": 8.186574910451983e-07, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26990 + }, + { + "epoch": 7.855688100087286, + "flow_matching_loss": 0.010594869265332818, + "grad_norm": 0.08167365938425064, + "learning_rate": 8.132783414558326e-07, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27000 + }, + { + "epoch": 7.858597614198429, + "flow_matching_loss": 0.00997072565369308, + "grad_norm": 0.04843444749712944, + "learning_rate": 8.079164300710023e-07, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27010 + }, + { + "epoch": 7.861507128309572, + "flow_matching_loss": 0.010315426625311375, + "grad_norm": 0.07608775794506073, + "learning_rate": 8.025717634059326e-07, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27020 + }, + { + "epoch": 7.864416642420716, + "flow_matching_loss": 0.009616878954693675, + "grad_norm": 0.05759797617793083, + "learning_rate": 7.972443479549035e-07, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27030 + }, + { + "epoch": 7.867326156531859, + "flow_matching_loss": 0.01029429929330945, + "grad_norm": 0.08525585383176804, + "learning_rate": 7.919341901912236e-07, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27040 + }, + { + "epoch": 7.870235670643003, + "flow_matching_loss": 0.01012773923575878, + "grad_norm": 0.07299192994832993, + "learning_rate": 7.866412965672365e-07, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27050 + }, + { + "epoch": 7.873145184754146, + "flow_matching_loss": 0.01027255766093731, + "grad_norm": 0.055269721895456314, + "learning_rate": 7.813656735143044e-07, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27060 + }, + { + "epoch": 7.876054698865289, + "flow_matching_loss": 0.011831075605005025, + "grad_norm": 0.06918061524629593, + "learning_rate": 7.761073274428076e-07, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27070 + }, + { + "epoch": 7.878964212976433, + "flow_matching_loss": 0.009191901935264469, + "grad_norm": 0.1523800641298294, + "learning_rate": 7.708662647421355e-07, + "loss": 0.0092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27080 + }, + { + "epoch": 7.8818737270875765, + "flow_matching_loss": 0.01048777843825519, + "grad_norm": 0.05704738572239876, + "learning_rate": 7.65642491780672e-07, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27090 + }, + { + "epoch": 7.88478324119872, + "flow_matching_loss": 0.00879450049251318, + "grad_norm": 0.09375312179327011, + "learning_rate": 7.604360149057916e-07, + "loss": 0.0088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27100 + }, + { + "epoch": 7.887692755309863, + "flow_matching_loss": 0.009774689516052603, + "grad_norm": 0.06643969565629959, + "learning_rate": 7.552468404438568e-07, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27110 + }, + { + "epoch": 7.890602269421007, + "flow_matching_loss": 0.00949265849776566, + "grad_norm": 0.10279186815023422, + "learning_rate": 7.500749747002022e-07, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27120 + }, + { + "epoch": 7.89351178353215, + "flow_matching_loss": 0.011054556025192142, + "grad_norm": 0.05971432849764824, + "learning_rate": 7.449204239591345e-07, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27130 + }, + { + "epoch": 7.896421297643293, + "flow_matching_loss": 0.009596019750460982, + "grad_norm": 0.053944241255521774, + "learning_rate": 7.397831944839167e-07, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27140 + }, + { + "epoch": 7.899330811754437, + "flow_matching_loss": 0.00934325447306037, + "grad_norm": 0.07364506274461746, + "learning_rate": 7.346632925167706e-07, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27150 + }, + { + "epoch": 7.902240325865581, + "flow_matching_loss": 0.00950344312004745, + "grad_norm": 0.06551103293895721, + "learning_rate": 7.295607242788566e-07, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27160 + }, + { + "epoch": 7.905149839976724, + "flow_matching_loss": 0.01086734221316874, + "grad_norm": 0.10161653906106949, + "learning_rate": 7.244754959702765e-07, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27170 + }, + { + "epoch": 7.908059354087867, + "flow_matching_loss": 0.010578262899070979, + "grad_norm": 0.08264730125665665, + "learning_rate": 7.194076137700628e-07, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27180 + }, + { + "epoch": 7.910968868199011, + "flow_matching_loss": 0.009304359834641219, + "grad_norm": 0.07819482684135437, + "learning_rate": 7.143570838361696e-07, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27190 + }, + { + "epoch": 7.913878382310154, + "flow_matching_loss": 0.008889743499457836, + "grad_norm": 0.08480924367904663, + "learning_rate": 7.093239123054696e-07, + "loss": 0.0089, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27200 + }, + { + "epoch": 7.916787896421297, + "flow_matching_loss": 0.00976791032589972, + "grad_norm": 0.07728677988052368, + "learning_rate": 7.043081052937389e-07, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27210 + }, + { + "epoch": 7.919697410532441, + "flow_matching_loss": 0.00963947493582964, + "grad_norm": 0.0649641677737236, + "learning_rate": 6.993096688956558e-07, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27220 + }, + { + "epoch": 7.922606924643585, + "flow_matching_loss": 0.010857795644551516, + "grad_norm": 0.05789034441113472, + "learning_rate": 6.943286091847884e-07, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27230 + }, + { + "epoch": 7.925516438754728, + "flow_matching_loss": 0.009746998501941561, + "grad_norm": 0.0808810368180275, + "learning_rate": 6.89364932213602e-07, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27240 + }, + { + "epoch": 7.928425952865871, + "flow_matching_loss": 0.010203489940613509, + "grad_norm": 0.05639386922121048, + "learning_rate": 6.844186440134253e-07, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27250 + }, + { + "epoch": 7.931335466977015, + "flow_matching_loss": 0.009050596924498677, + "grad_norm": 0.061302948743104935, + "learning_rate": 6.794897505944692e-07, + "loss": 0.0091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27260 + }, + { + "epoch": 7.934244981088158, + "flow_matching_loss": 0.00892077460885048, + "grad_norm": 0.06614714860916138, + "learning_rate": 6.745782579457993e-07, + "loss": 0.0089, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27270 + }, + { + "epoch": 7.937154495199302, + "flow_matching_loss": 0.010318469209596514, + "grad_norm": 0.09129539132118225, + "learning_rate": 6.696841720353453e-07, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27280 + }, + { + "epoch": 7.940064009310445, + "flow_matching_loss": 0.009613360278308392, + "grad_norm": 0.06305792182683945, + "learning_rate": 6.648074988098834e-07, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27290 + }, + { + "epoch": 7.942973523421589, + "flow_matching_loss": 0.009354098560288548, + "grad_norm": 0.08672972768545151, + "learning_rate": 6.599482441950322e-07, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27300 + }, + { + "epoch": 7.945883037532732, + "flow_matching_loss": 0.010716384649276734, + "grad_norm": 0.06843575835227966, + "learning_rate": 6.551064140952417e-07, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27310 + }, + { + "epoch": 7.9487925516438755, + "flow_matching_loss": 0.0097413401119411, + "grad_norm": 0.09427965432405472, + "learning_rate": 6.502820143937943e-07, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27320 + }, + { + "epoch": 7.951702065755019, + "flow_matching_loss": 0.010341684333980083, + "grad_norm": 0.08961600810289383, + "learning_rate": 6.454750509527874e-07, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27330 + }, + { + "epoch": 7.954611579866162, + "flow_matching_loss": 0.009552499651908875, + "grad_norm": 0.09004835784435272, + "learning_rate": 6.406855296131387e-07, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27340 + }, + { + "epoch": 7.957521093977306, + "flow_matching_loss": 0.009897935204207897, + "grad_norm": 0.06829670071601868, + "learning_rate": 6.359134561945662e-07, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27350 + }, + { + "epoch": 7.9604306080884495, + "flow_matching_loss": 0.009063611319288612, + "grad_norm": 0.05771292746067047, + "learning_rate": 6.311588364955883e-07, + "loss": 0.0091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27360 + }, + { + "epoch": 7.963340122199592, + "flow_matching_loss": 0.009516797866672278, + "grad_norm": 0.08483880758285522, + "learning_rate": 6.264216762935187e-07, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27370 + }, + { + "epoch": 7.966249636310736, + "flow_matching_loss": 0.010078936768695711, + "grad_norm": 0.07545648515224457, + "learning_rate": 6.21701981344453e-07, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27380 + }, + { + "epoch": 7.96915915042188, + "flow_matching_loss": 0.01102995458059013, + "grad_norm": 0.05255739763379097, + "learning_rate": 6.16999757383262e-07, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27390 + }, + { + "epoch": 7.9720686645330225, + "flow_matching_loss": 0.008618461294099688, + "grad_norm": 0.0486694872379303, + "learning_rate": 6.123150101235952e-07, + "loss": 0.0086, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27400 + }, + { + "epoch": 7.974978178644166, + "flow_matching_loss": 0.010117668099701405, + "grad_norm": 0.08390704542398453, + "learning_rate": 6.076477452578594e-07, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27410 + }, + { + "epoch": 7.97788769275531, + "flow_matching_loss": 0.010710039548575878, + "grad_norm": 0.10561958700418472, + "learning_rate": 6.029979684572262e-07, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27420 + }, + { + "epoch": 7.980797206866454, + "flow_matching_loss": 0.009123465232551097, + "grad_norm": 0.06180242821574211, + "learning_rate": 5.983656853716097e-07, + "loss": 0.0091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27430 + }, + { + "epoch": 7.9837067209775965, + "flow_matching_loss": 0.010502588422968983, + "grad_norm": 0.08040227741003036, + "learning_rate": 5.937509016296689e-07, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27440 + }, + { + "epoch": 7.98661623508874, + "flow_matching_loss": 0.010482975468039512, + "grad_norm": 0.07001315802335739, + "learning_rate": 5.89153622838805e-07, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27450 + }, + { + "epoch": 7.989525749199884, + "flow_matching_loss": 0.009196651168167591, + "grad_norm": 0.05149224400520325, + "learning_rate": 5.845738545851431e-07, + "loss": 0.0092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27460 + }, + { + "epoch": 7.992435263311027, + "flow_matching_loss": 0.01044569374062121, + "grad_norm": 0.07738276571035385, + "learning_rate": 5.800116024335334e-07, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27470 + }, + { + "epoch": 7.99534477742217, + "flow_matching_loss": 0.010204731533303858, + "grad_norm": 0.08293994516134262, + "learning_rate": 5.75466871927543e-07, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27480 + }, + { + "epoch": 7.998254291533314, + "flow_matching_loss": 0.009533475618809462, + "grad_norm": 0.07692577689886093, + "learning_rate": 5.709396685894463e-07, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27490 + }, + { + "epoch": 8.001163805644458, + "flow_matching_loss": 0.008271088963374496, + "grad_norm": 0.07523611187934875, + "learning_rate": 5.664299979202264e-07, + "loss": 0.0083, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27500 + }, + { + "epoch": 8.004073319755602, + "flow_matching_loss": 0.010615805303677917, + "grad_norm": 0.0719403475522995, + "learning_rate": 5.61937865399555e-07, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27510 + }, + { + "epoch": 8.006982833866743, + "flow_matching_loss": 0.010439946549013257, + "grad_norm": 0.0567501038312912, + "learning_rate": 5.574632764857974e-07, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27520 + }, + { + "epoch": 8.009892347977887, + "flow_matching_loss": 0.010276829730719328, + "grad_norm": 0.06371169537305832, + "learning_rate": 5.53006236616001e-07, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27530 + }, + { + "epoch": 8.012801862089031, + "flow_matching_loss": 0.010000711632892489, + "grad_norm": 0.0915832668542862, + "learning_rate": 5.485667512058873e-07, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27540 + }, + { + "epoch": 8.015711376200175, + "flow_matching_loss": 0.010451217647641897, + "grad_norm": 0.059863217175006866, + "learning_rate": 5.441448256498544e-07, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27550 + }, + { + "epoch": 8.018620890311318, + "flow_matching_loss": 0.009114786004647612, + "grad_norm": 0.06254053860902786, + "learning_rate": 5.397404653209559e-07, + "loss": 0.0091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27560 + }, + { + "epoch": 8.021530404422462, + "flow_matching_loss": 0.010947692254558206, + "grad_norm": 0.06637128442525864, + "learning_rate": 5.35353675570906e-07, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27570 + }, + { + "epoch": 8.024439918533606, + "flow_matching_loss": 0.010156094375997782, + "grad_norm": 0.07641180604696274, + "learning_rate": 5.309844617300674e-07, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27580 + }, + { + "epoch": 8.027349432644748, + "flow_matching_loss": 0.009482092736288906, + "grad_norm": 0.06970449537038803, + "learning_rate": 5.266328291074468e-07, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27590 + }, + { + "epoch": 8.030258946755891, + "flow_matching_loss": 0.011088270833715797, + "grad_norm": 0.09411448240280151, + "learning_rate": 5.222987829906861e-07, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27600 + }, + { + "epoch": 8.033168460867035, + "flow_matching_loss": 0.009775972878560423, + "grad_norm": 0.11240684241056442, + "learning_rate": 5.179823286460644e-07, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27610 + }, + { + "epoch": 8.036077974978179, + "flow_matching_loss": 0.010549916978925466, + "grad_norm": 0.05637681856751442, + "learning_rate": 5.13683471318478e-07, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27620 + }, + { + "epoch": 8.038987489089322, + "flow_matching_loss": 0.009223851887509227, + "grad_norm": 0.06291762739419937, + "learning_rate": 5.094022162314449e-07, + "loss": 0.0092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27630 + }, + { + "epoch": 8.041897003200466, + "flow_matching_loss": 0.008691467065364122, + "grad_norm": 0.058046210557222366, + "learning_rate": 5.051385685870924e-07, + "loss": 0.0087, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27640 + }, + { + "epoch": 8.044806517311608, + "flow_matching_loss": 0.01009858506731689, + "grad_norm": 0.07875633984804153, + "learning_rate": 5.008925335661546e-07, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27650 + }, + { + "epoch": 8.047716031422752, + "flow_matching_loss": 0.011172257317230105, + "grad_norm": 0.08776925504207611, + "learning_rate": 4.966641163279662e-07, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27660 + }, + { + "epoch": 8.050625545533896, + "flow_matching_loss": 0.009474982880055905, + "grad_norm": 0.07101555168628693, + "learning_rate": 4.924533220104521e-07, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27670 + }, + { + "epoch": 8.05353505964504, + "flow_matching_loss": 0.0102107850369066, + "grad_norm": 0.08287035673856735, + "learning_rate": 4.882601557301231e-07, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27680 + }, + { + "epoch": 8.056444573756183, + "flow_matching_loss": 0.009883608762174844, + "grad_norm": 0.05463339388370514, + "learning_rate": 4.840846225820717e-07, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27690 + }, + { + "epoch": 8.059354087867327, + "flow_matching_loss": 0.010574099654331804, + "grad_norm": 0.07237759232521057, + "learning_rate": 4.79926727639966e-07, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27700 + }, + { + "epoch": 8.06226360197847, + "flow_matching_loss": 0.011578458501026035, + "grad_norm": 0.07007847726345062, + "learning_rate": 4.7578647595603817e-07, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27710 + }, + { + "epoch": 8.065173116089612, + "flow_matching_loss": 0.009067208133637905, + "grad_norm": 0.057634372264146805, + "learning_rate": 4.716638725610889e-07, + "loss": 0.0091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27720 + }, + { + "epoch": 8.068082630200756, + "flow_matching_loss": 0.008702394599094988, + "grad_norm": 0.10011287778615952, + "learning_rate": 4.6755892246446606e-07, + "loss": 0.0087, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27730 + }, + { + "epoch": 8.0709921443119, + "flow_matching_loss": 0.00982276895083487, + "grad_norm": 0.068878173828125, + "learning_rate": 4.634716306540715e-07, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27740 + }, + { + "epoch": 8.073901658423043, + "flow_matching_loss": 0.010090920422226191, + "grad_norm": 0.08403272181749344, + "learning_rate": 4.5940200209635244e-07, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27750 + }, + { + "epoch": 8.076811172534187, + "flow_matching_loss": 0.010676303552463651, + "grad_norm": 0.07824612408876419, + "learning_rate": 4.5535004173628657e-07, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27760 + }, + { + "epoch": 8.07972068664533, + "flow_matching_loss": 0.00983977927826345, + "grad_norm": 0.057407740503549576, + "learning_rate": 4.5131575449739225e-07, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27770 + }, + { + "epoch": 8.082630200756475, + "flow_matching_loss": 0.009453547419980168, + "grad_norm": 0.08923652768135071, + "learning_rate": 4.472991452817099e-07, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27780 + }, + { + "epoch": 8.085539714867616, + "flow_matching_loss": 0.010327611397951842, + "grad_norm": 0.07860283553600311, + "learning_rate": 4.433002189697938e-07, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27790 + }, + { + "epoch": 8.08844922897876, + "flow_matching_loss": 0.009938256861642004, + "grad_norm": 0.07508209347724915, + "learning_rate": 4.393189804207187e-07, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27800 + }, + { + "epoch": 8.091358743089904, + "flow_matching_loss": 0.011067236587405205, + "grad_norm": 0.06917927414178848, + "learning_rate": 4.353554344720651e-07, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27810 + }, + { + "epoch": 8.094268257201048, + "flow_matching_loss": 0.01133133927360177, + "grad_norm": 0.07961548119783401, + "learning_rate": 4.314095859399153e-07, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27820 + }, + { + "epoch": 8.097177771312191, + "flow_matching_loss": 0.009609205275774002, + "grad_norm": 0.09707125276327133, + "learning_rate": 4.274814396188459e-07, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27830 + }, + { + "epoch": 8.100087285423335, + "flow_matching_loss": 0.009973609587177634, + "grad_norm": 0.06264534592628479, + "learning_rate": 4.235710002819271e-07, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27840 + }, + { + "epoch": 8.102996799534477, + "flow_matching_loss": 0.010195560660213232, + "grad_norm": 0.0629933550953865, + "learning_rate": 4.196782726807097e-07, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27850 + }, + { + "epoch": 8.10590631364562, + "flow_matching_loss": 0.009961808333173394, + "grad_norm": 0.0642850250005722, + "learning_rate": 4.158032615452251e-07, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27860 + }, + { + "epoch": 8.108815827756764, + "flow_matching_loss": 0.011015108367428184, + "grad_norm": 0.06748133897781372, + "learning_rate": 4.1194597158397853e-07, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27870 + }, + { + "epoch": 8.111725341867908, + "flow_matching_loss": 0.010126016521826386, + "grad_norm": 0.05319735035300255, + "learning_rate": 4.081064074839408e-07, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27880 + }, + { + "epoch": 8.114634855979052, + "flow_matching_loss": 0.010083307791501284, + "grad_norm": 0.08372007310390472, + "learning_rate": 4.042845739105433e-07, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27890 + }, + { + "epoch": 8.117544370090195, + "flow_matching_loss": 0.009911464527249336, + "grad_norm": 0.09235431253910065, + "learning_rate": 4.004804755076746e-07, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27900 + }, + { + "epoch": 8.12045388420134, + "flow_matching_loss": 0.008856439776718617, + "grad_norm": 0.08024214953184128, + "learning_rate": 3.9669411689767063e-07, + "loss": 0.0089, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27910 + }, + { + "epoch": 8.123363398312481, + "flow_matching_loss": 0.009848667168989777, + "grad_norm": 0.06748820096254349, + "learning_rate": 3.929255026813194e-07, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27920 + }, + { + "epoch": 8.126272912423625, + "flow_matching_loss": 0.009334944421425462, + "grad_norm": 0.06121765077114105, + "learning_rate": 3.8917463743783797e-07, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27930 + }, + { + "epoch": 8.129182426534769, + "flow_matching_loss": 0.010301115922629833, + "grad_norm": 0.0703229010105133, + "learning_rate": 3.854415257248839e-07, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27940 + }, + { + "epoch": 8.132091940645912, + "flow_matching_loss": 0.008847353514283895, + "grad_norm": 0.06479807943105698, + "learning_rate": 3.8172617207853864e-07, + "loss": 0.0088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27950 + }, + { + "epoch": 8.135001454757056, + "flow_matching_loss": 0.009653720259666442, + "grad_norm": 0.07960417866706848, + "learning_rate": 3.7802858101330773e-07, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27960 + }, + { + "epoch": 8.1379109688682, + "flow_matching_loss": 0.009965751087293029, + "grad_norm": 0.11866356432437897, + "learning_rate": 3.743487570221121e-07, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27970 + }, + { + "epoch": 8.140820482979342, + "flow_matching_loss": 0.009560779901221395, + "grad_norm": 0.08715460449457169, + "learning_rate": 3.7068670457628517e-07, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27980 + }, + { + "epoch": 8.143729997090485, + "flow_matching_loss": 0.010161380050703883, + "grad_norm": 0.06236015260219574, + "learning_rate": 3.6704242812556576e-07, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27990 + }, + { + "epoch": 8.146639511201629, + "flow_matching_loss": 0.00992299714125693, + "grad_norm": 0.09047369658946991, + "learning_rate": 3.634159320980951e-07, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28000 + }, + { + "epoch": 8.149549025312773, + "flow_matching_loss": 0.010044115409255028, + "grad_norm": 0.07175258547067642, + "learning_rate": 3.5980722090040663e-07, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28010 + }, + { + "epoch": 8.152458539423916, + "flow_matching_loss": 0.009549460839480162, + "grad_norm": 0.10134624689817429, + "learning_rate": 3.5621629891742447e-07, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28020 + }, + { + "epoch": 8.15536805353506, + "flow_matching_loss": 0.011459081433713436, + "grad_norm": 0.07084327936172485, + "learning_rate": 3.526431705124583e-07, + "loss": 0.0115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28030 + }, + { + "epoch": 8.158277567646204, + "flow_matching_loss": 0.01156274862587452, + "grad_norm": 0.0828588679432869, + "learning_rate": 3.4908784002719517e-07, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28040 + }, + { + "epoch": 8.161187081757346, + "flow_matching_loss": 0.010245710611343384, + "grad_norm": 0.07527806609869003, + "learning_rate": 3.4555031178169773e-07, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28050 + }, + { + "epoch": 8.16409659586849, + "flow_matching_loss": 0.010711404401808977, + "grad_norm": 0.09134647995233536, + "learning_rate": 3.420305900743942e-07, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28060 + }, + { + "epoch": 8.167006109979633, + "flow_matching_loss": 0.010462304577231407, + "grad_norm": 0.05616521090269089, + "learning_rate": 3.385286791820802e-07, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28070 + }, + { + "epoch": 8.169915624090777, + "flow_matching_loss": 0.010846586199477315, + "grad_norm": 0.09387663751840591, + "learning_rate": 3.350445833599086e-07, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28080 + }, + { + "epoch": 8.17282513820192, + "flow_matching_loss": 0.00850522550754249, + "grad_norm": 0.06936273723840714, + "learning_rate": 3.3157830684138303e-07, + "loss": 0.0085, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28090 + }, + { + "epoch": 8.175734652313064, + "flow_matching_loss": 0.0101880535017699, + "grad_norm": 0.06504557281732559, + "learning_rate": 3.281298538383559e-07, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28100 + }, + { + "epoch": 8.178644166424208, + "flow_matching_loss": 0.010393839329481125, + "grad_norm": 0.07052182406187057, + "learning_rate": 3.2469922854102387e-07, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28110 + }, + { + "epoch": 8.18155368053535, + "flow_matching_loss": 0.011385597893968225, + "grad_norm": 0.07141392678022385, + "learning_rate": 3.21286435117919e-07, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28120 + }, + { + "epoch": 8.184463194646494, + "flow_matching_loss": 0.009691843623295426, + "grad_norm": 0.07726869732141495, + "learning_rate": 3.1789147771590754e-07, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28130 + }, + { + "epoch": 8.187372708757637, + "flow_matching_loss": 0.01083978209644556, + "grad_norm": 0.05136396363377571, + "learning_rate": 3.14514360460183e-07, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28140 + }, + { + "epoch": 8.190282222868781, + "flow_matching_loss": 0.009577336348593235, + "grad_norm": 0.0692293643951416, + "learning_rate": 3.1115508745426293e-07, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28150 + }, + { + "epoch": 8.193191736979925, + "flow_matching_loss": 0.010869679460301995, + "grad_norm": 0.1134125217795372, + "learning_rate": 3.0781366277997716e-07, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28160 + }, + { + "epoch": 8.196101251091068, + "flow_matching_loss": 0.01141199511475861, + "grad_norm": 0.05821368470788002, + "learning_rate": 3.044900904974762e-07, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28170 + }, + { + "epoch": 8.19901076520221, + "flow_matching_loss": 0.009776703361421824, + "grad_norm": 0.06868702173233032, + "learning_rate": 3.011843746452081e-07, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28180 + }, + { + "epoch": 8.201920279313354, + "flow_matching_loss": 0.011942344438284635, + "grad_norm": 0.06264328956604004, + "learning_rate": 2.9789651923993464e-07, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28190 + }, + { + "epoch": 8.204829793424498, + "flow_matching_loss": 0.009007089864462614, + "grad_norm": 0.08107494562864304, + "learning_rate": 2.9462652827670855e-07, + "loss": 0.009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28200 + }, + { + "epoch": 8.207739307535642, + "flow_matching_loss": 0.01033234540373087, + "grad_norm": 0.11991221457719803, + "learning_rate": 2.913744057288764e-07, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28210 + }, + { + "epoch": 8.210648821646785, + "flow_matching_loss": 0.011358285741880535, + "grad_norm": 0.13977499306201935, + "learning_rate": 2.8814015554807396e-07, + "loss": 0.0114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28220 + }, + { + "epoch": 8.213558335757929, + "flow_matching_loss": 0.010037331376224756, + "grad_norm": 0.05727997422218323, + "learning_rate": 2.849237816642192e-07, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28230 + }, + { + "epoch": 8.216467849869073, + "flow_matching_loss": 0.01005111839622259, + "grad_norm": 0.06447912007570267, + "learning_rate": 2.817252879855109e-07, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28240 + }, + { + "epoch": 8.219377363980215, + "flow_matching_loss": 0.009127595694735646, + "grad_norm": 0.06789831817150116, + "learning_rate": 2.785446783984219e-07, + "loss": 0.0091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28250 + }, + { + "epoch": 8.222286878091358, + "flow_matching_loss": 0.008264537760987877, + "grad_norm": 0.06983085721731186, + "learning_rate": 2.7538195676769076e-07, + "loss": 0.0083, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28260 + }, + { + "epoch": 8.225196392202502, + "flow_matching_loss": 0.01061343695037067, + "grad_norm": 0.06599225103855133, + "learning_rate": 2.722371269363216e-07, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28270 + }, + { + "epoch": 8.228105906313646, + "flow_matching_loss": 0.010274713207036257, + "grad_norm": 0.06708740442991257, + "learning_rate": 2.6911019272558125e-07, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28280 + }, + { + "epoch": 8.23101542042479, + "flow_matching_loss": 0.009049435053020716, + "grad_norm": 0.07760927081108093, + "learning_rate": 2.6600115793498857e-07, + "loss": 0.009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28290 + }, + { + "epoch": 8.233924934535933, + "flow_matching_loss": 0.01043965038843453, + "grad_norm": 0.08237320929765701, + "learning_rate": 2.629100263423151e-07, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28300 + }, + { + "epoch": 8.236834448647077, + "flow_matching_loss": 0.008583317138254642, + "grad_norm": 0.05629745125770569, + "learning_rate": 2.598368017035763e-07, + "loss": 0.0086, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28310 + }, + { + "epoch": 8.239743962758219, + "flow_matching_loss": 0.010426041577011347, + "grad_norm": 0.06720198690891266, + "learning_rate": 2.5678148775303e-07, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28320 + }, + { + "epoch": 8.242653476869362, + "flow_matching_loss": 0.00946067674085498, + "grad_norm": 0.13470865786075592, + "learning_rate": 2.537440882031683e-07, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28330 + }, + { + "epoch": 8.245562990980506, + "flow_matching_loss": 0.010503829922527075, + "grad_norm": 0.07407888025045395, + "learning_rate": 2.5072460674472385e-07, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28340 + }, + { + "epoch": 8.24847250509165, + "flow_matching_loss": 0.008834488736465574, + "grad_norm": 0.10312820971012115, + "learning_rate": 2.4772304704664683e-07, + "loss": 0.0088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28350 + }, + { + "epoch": 8.251382019202794, + "flow_matching_loss": 0.009589826036244632, + "grad_norm": 0.08528553694486618, + "learning_rate": 2.447394127561181e-07, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28360 + }, + { + "epoch": 8.254291533313937, + "flow_matching_loss": 0.010484333615750074, + "grad_norm": 0.06905576586723328, + "learning_rate": 2.41773707498531e-07, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28370 + }, + { + "epoch": 8.25720104742508, + "flow_matching_loss": 0.009791735466569662, + "grad_norm": 0.07389391958713531, + "learning_rate": 2.388259348774996e-07, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28380 + }, + { + "epoch": 8.260110561536223, + "flow_matching_loss": 0.009879294643178582, + "grad_norm": 0.06411775946617126, + "learning_rate": 2.358960984748454e-07, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28390 + }, + { + "epoch": 8.263020075647367, + "flow_matching_loss": 0.008751374529674649, + "grad_norm": 0.06585448235273361, + "learning_rate": 2.3298420185059565e-07, + "loss": 0.0088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28400 + }, + { + "epoch": 8.26592958975851, + "flow_matching_loss": 0.010219198698177934, + "grad_norm": 0.07016962766647339, + "learning_rate": 2.3009024854297844e-07, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28410 + }, + { + "epoch": 8.268839103869654, + "flow_matching_loss": 0.009187044482678175, + "grad_norm": 0.06904853880405426, + "learning_rate": 2.272142420684209e-07, + "loss": 0.0092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28420 + }, + { + "epoch": 8.271748617980798, + "flow_matching_loss": 0.010627025365829467, + "grad_norm": 0.07659046351909637, + "learning_rate": 2.243561859215426e-07, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28430 + }, + { + "epoch": 8.274658132091941, + "flow_matching_loss": 0.009682474145665764, + "grad_norm": 0.06283754110336304, + "learning_rate": 2.2151608357514897e-07, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28440 + }, + { + "epoch": 8.277567646203083, + "flow_matching_loss": 0.009579898975789546, + "grad_norm": 0.05972142517566681, + "learning_rate": 2.186939384802361e-07, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28450 + }, + { + "epoch": 8.280477160314227, + "flow_matching_loss": 0.009956642892211676, + "grad_norm": 0.07060263305902481, + "learning_rate": 2.1588975406597266e-07, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28460 + }, + { + "epoch": 8.28338667442537, + "flow_matching_loss": 0.009765761252492667, + "grad_norm": 0.08753261715173721, + "learning_rate": 2.131035337397097e-07, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28470 + }, + { + "epoch": 8.286296188536515, + "flow_matching_loss": 0.01068481830880046, + "grad_norm": 0.08077938109636307, + "learning_rate": 2.1033528088696575e-07, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28480 + }, + { + "epoch": 8.289205702647658, + "flow_matching_loss": 0.009539148304611444, + "grad_norm": 0.1098211333155632, + "learning_rate": 2.0758499887143012e-07, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28490 + }, + { + "epoch": 8.292115216758802, + "flow_matching_loss": 0.00983520494773984, + "grad_norm": 0.06406718492507935, + "learning_rate": 2.0485269103495462e-07, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28500 + }, + { + "epoch": 8.295024730869944, + "flow_matching_loss": 0.009845503885298968, + "grad_norm": 0.07864520698785782, + "learning_rate": 2.021383606975502e-07, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28510 + }, + { + "epoch": 8.297934244981088, + "flow_matching_loss": 0.009287477796897292, + "grad_norm": 0.06624917685985565, + "learning_rate": 1.9944201115738524e-07, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28520 + }, + { + "epoch": 8.300843759092231, + "flow_matching_loss": 0.009711061464622616, + "grad_norm": 0.07243185490369797, + "learning_rate": 1.9676364569077732e-07, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28530 + }, + { + "epoch": 8.303753273203375, + "flow_matching_loss": 0.009763498976826667, + "grad_norm": 0.08726829290390015, + "learning_rate": 1.9410326755219477e-07, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28540 + }, + { + "epoch": 8.306662787314519, + "flow_matching_loss": 0.009227848518639803, + "grad_norm": 0.10282786190509796, + "learning_rate": 1.9146087997424677e-07, + "loss": 0.0092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28550 + }, + { + "epoch": 8.309572301425662, + "flow_matching_loss": 0.010517702158540487, + "grad_norm": 0.06703747808933258, + "learning_rate": 1.8883648616768335e-07, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28560 + }, + { + "epoch": 8.312481815536806, + "flow_matching_loss": 0.01001697857864201, + "grad_norm": 0.05903391167521477, + "learning_rate": 1.862300893213903e-07, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28570 + }, + { + "epoch": 8.315391329647948, + "flow_matching_loss": 0.009117203392088413, + "grad_norm": 0.087901271879673, + "learning_rate": 1.8364169260238595e-07, + "loss": 0.0091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28580 + }, + { + "epoch": 8.318300843759092, + "flow_matching_loss": 0.01083039427176118, + "grad_norm": 0.08491784334182739, + "learning_rate": 1.8107129915581777e-07, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28590 + }, + { + "epoch": 8.321210357870235, + "flow_matching_loss": 0.009248446812853217, + "grad_norm": 0.07268477976322174, + "learning_rate": 1.785189121049541e-07, + "loss": 0.0092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28600 + }, + { + "epoch": 8.32411987198138, + "flow_matching_loss": 0.009871915262192488, + "grad_norm": 0.07594834268093109, + "learning_rate": 1.7598453455118567e-07, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28610 + }, + { + "epoch": 8.327029386092523, + "flow_matching_loss": 0.010996569134294986, + "grad_norm": 0.08957212418317795, + "learning_rate": 1.7346816957402256e-07, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28620 + }, + { + "epoch": 8.329938900203667, + "flow_matching_loss": 0.009467951441183686, + "grad_norm": 0.08114203065633774, + "learning_rate": 1.7096982023108389e-07, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28630 + }, + { + "epoch": 8.33284841431481, + "flow_matching_loss": 0.009662577230483294, + "grad_norm": 0.05601830035448074, + "learning_rate": 1.6848948955809963e-07, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28640 + }, + { + "epoch": 8.335757928425952, + "flow_matching_loss": 0.01036135908216238, + "grad_norm": 0.08814041316509247, + "learning_rate": 1.66027180568904e-07, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28650 + }, + { + "epoch": 8.338667442537096, + "flow_matching_loss": 0.010139564890414477, + "grad_norm": 0.05854123458266258, + "learning_rate": 1.6358289625543865e-07, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28660 + }, + { + "epoch": 8.34157695664824, + "flow_matching_loss": 0.009630043152719736, + "grad_norm": 0.058322496712207794, + "learning_rate": 1.6115663958773618e-07, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28670 + }, + { + "epoch": 8.344486470759383, + "flow_matching_loss": 0.01028409879654646, + "grad_norm": 0.0689372643828392, + "learning_rate": 1.5874841351392832e-07, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28680 + }, + { + "epoch": 8.347395984870527, + "flow_matching_loss": 0.012249072454869748, + "grad_norm": 0.08257762342691422, + "learning_rate": 1.56358220960236e-07, + "loss": 0.0122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28690 + }, + { + "epoch": 8.35030549898167, + "flow_matching_loss": 0.00933807590045035, + "grad_norm": 0.06203928589820862, + "learning_rate": 1.5398606483096932e-07, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28700 + }, + { + "epoch": 8.353215013092813, + "flow_matching_loss": 0.011037145089358092, + "grad_norm": 0.09655436873435974, + "learning_rate": 1.5163194800852265e-07, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28710 + }, + { + "epoch": 8.356124527203956, + "flow_matching_loss": 0.011700251372531056, + "grad_norm": 0.083184614777565, + "learning_rate": 1.4929587335336614e-07, + "loss": 0.0117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28720 + }, + { + "epoch": 8.3590340413151, + "flow_matching_loss": 0.009862778661772608, + "grad_norm": 0.08188007026910782, + "learning_rate": 1.4697784370405587e-07, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28730 + }, + { + "epoch": 8.361943555426244, + "flow_matching_loss": 0.009671272663399576, + "grad_norm": 0.11854954808950424, + "learning_rate": 1.4467786187721376e-07, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28740 + }, + { + "epoch": 8.364853069537387, + "flow_matching_loss": 0.009600053681060672, + "grad_norm": 0.078000009059906, + "learning_rate": 1.4239593066753265e-07, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28750 + }, + { + "epoch": 8.367762583648531, + "flow_matching_loss": 0.009405265329405665, + "grad_norm": 0.06951086223125458, + "learning_rate": 1.4013205284777953e-07, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28760 + }, + { + "epoch": 8.370672097759675, + "flow_matching_loss": 0.00967941414564848, + "grad_norm": 0.0683078020811081, + "learning_rate": 1.3788623116877397e-07, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28770 + }, + { + "epoch": 8.373581611870817, + "flow_matching_loss": 0.009579418087378144, + "grad_norm": 0.06338600814342499, + "learning_rate": 1.3565846835940477e-07, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28780 + }, + { + "epoch": 8.37649112598196, + "flow_matching_loss": 0.00969677660614252, + "grad_norm": 0.12940239906311035, + "learning_rate": 1.334487671266116e-07, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28790 + }, + { + "epoch": 8.379400640093104, + "flow_matching_loss": 0.00911537865176797, + "grad_norm": 0.0814804956316948, + "learning_rate": 1.3125713015538998e-07, + "loss": 0.0091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28800 + }, + { + "epoch": 8.382310154204248, + "flow_matching_loss": 0.009882763121277094, + "grad_norm": 0.07049933820962906, + "learning_rate": 1.290835601087864e-07, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28810 + }, + { + "epoch": 8.385219668315392, + "flow_matching_loss": 0.009624071605503558, + "grad_norm": 0.061815809458494186, + "learning_rate": 1.2692805962789323e-07, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28820 + }, + { + "epoch": 8.388129182426535, + "flow_matching_loss": 0.009496138291433453, + "grad_norm": 0.05314493551850319, + "learning_rate": 1.2479063133184698e-07, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28830 + }, + { + "epoch": 8.391038696537677, + "flow_matching_loss": 0.009209095453843474, + "grad_norm": 0.07868126034736633, + "learning_rate": 1.2267127781782516e-07, + "loss": 0.0092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28840 + }, + { + "epoch": 8.393948210648821, + "flow_matching_loss": 0.010227313870564103, + "grad_norm": 0.09371959418058395, + "learning_rate": 1.205700016610395e-07, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28850 + }, + { + "epoch": 8.396857724759965, + "flow_matching_loss": 0.009748604707419872, + "grad_norm": 0.058257583528757095, + "learning_rate": 1.1848680541474088e-07, + "loss": 0.0097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28860 + }, + { + "epoch": 8.399767238871108, + "flow_matching_loss": 0.009758405573666095, + "grad_norm": 0.06912808865308762, + "learning_rate": 1.1642169161020955e-07, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28870 + }, + { + "epoch": 8.402676752982252, + "flow_matching_loss": 0.009308374440297484, + "grad_norm": 0.07247412949800491, + "learning_rate": 1.1437466275675156e-07, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28880 + }, + { + "epoch": 8.405586267093396, + "flow_matching_loss": 0.010054950881749392, + "grad_norm": 0.06972804665565491, + "learning_rate": 1.1234572134170229e-07, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28890 + }, + { + "epoch": 8.40849578120454, + "flow_matching_loss": 0.008981030434370041, + "grad_norm": 0.1327265352010727, + "learning_rate": 1.1033486983041463e-07, + "loss": 0.009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28900 + }, + { + "epoch": 8.411405295315681, + "flow_matching_loss": 0.009869582671672105, + "grad_norm": 0.07330531626939774, + "learning_rate": 1.0834211066626409e-07, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28910 + }, + { + "epoch": 8.414314809426825, + "flow_matching_loss": 0.00876918639987707, + "grad_norm": 0.0594681091606617, + "learning_rate": 1.0636744627064044e-07, + "loss": 0.0088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28920 + }, + { + "epoch": 8.417224323537969, + "flow_matching_loss": 0.011060327803716064, + "grad_norm": 0.06841545552015305, + "learning_rate": 1.0441087904294766e-07, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28930 + }, + { + "epoch": 8.420133837649113, + "flow_matching_loss": 0.00828237747773528, + "grad_norm": 0.06346837431192398, + "learning_rate": 1.0247241136059904e-07, + "loss": 0.0083, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28940 + }, + { + "epoch": 8.423043351760256, + "flow_matching_loss": 0.01005834024399519, + "grad_norm": 0.06837088614702225, + "learning_rate": 1.0055204557901709e-07, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28950 + }, + { + "epoch": 8.4259528658714, + "flow_matching_loss": 0.009919754602015019, + "grad_norm": 0.07451081275939941, + "learning_rate": 9.864978403162361e-08, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28960 + }, + { + "epoch": 8.428862379982544, + "flow_matching_loss": 0.010051714023575187, + "grad_norm": 0.05816582962870598, + "learning_rate": 9.676562902984965e-08, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28970 + }, + { + "epoch": 8.431771894093686, + "flow_matching_loss": 0.00919510037638247, + "grad_norm": 0.062317363917827606, + "learning_rate": 9.489958286311883e-08, + "loss": 0.0092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28980 + }, + { + "epoch": 8.43468140820483, + "flow_matching_loss": 0.009243406122550368, + "grad_norm": 0.05976438894867897, + "learning_rate": 9.30516477988541e-08, + "loss": 0.0092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28990 + }, + { + "epoch": 8.437590922315973, + "flow_matching_loss": 0.009395627677440644, + "grad_norm": 0.061699483543634415, + "learning_rate": 9.122182608247098e-08, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29000 + }, + { + "epoch": 8.440500436427117, + "flow_matching_loss": 0.00942104049026966, + "grad_norm": 0.06844773143529892, + "learning_rate": 8.94101199373759e-08, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29010 + }, + { + "epoch": 8.44340995053826, + "flow_matching_loss": 0.0090442537330091, + "grad_norm": 0.06330887973308563, + "learning_rate": 8.761653156495964e-08, + "loss": 0.009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29020 + }, + { + "epoch": 8.446319464649404, + "flow_matching_loss": 0.010128476098179818, + "grad_norm": 0.09039199352264404, + "learning_rate": 8.584106314460549e-08, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29030 + }, + { + "epoch": 8.449228978760548, + "flow_matching_loss": 0.010580373322591185, + "grad_norm": 0.07135152816772461, + "learning_rate": 8.408371683367111e-08, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29040 + }, + { + "epoch": 8.45213849287169, + "flow_matching_loss": 0.009180137095972896, + "grad_norm": 0.06061941012740135, + "learning_rate": 8.234449476750172e-08, + "loss": 0.0092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29050 + }, + { + "epoch": 8.455048006982834, + "flow_matching_loss": 0.00991600388661027, + "grad_norm": 0.060185015201568604, + "learning_rate": 8.062339905941351e-08, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29060 + }, + { + "epoch": 8.457957521093977, + "flow_matching_loss": 0.009998669475317001, + "grad_norm": 0.06262557953596115, + "learning_rate": 7.89204318007003e-08, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29070 + }, + { + "epoch": 8.460867035205121, + "flow_matching_loss": 0.009100728249177337, + "grad_norm": 0.05462110415101051, + "learning_rate": 7.723559506062849e-08, + "loss": 0.0091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29080 + }, + { + "epoch": 8.463776549316265, + "flow_matching_loss": 0.010296799754723907, + "grad_norm": 0.04802414029836655, + "learning_rate": 7.556889088643381e-08, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29090 + }, + { + "epoch": 8.466686063427408, + "flow_matching_loss": 0.010817027278244495, + "grad_norm": 0.09395008534193039, + "learning_rate": 7.392032130331793e-08, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29100 + }, + { + "epoch": 8.46959557753855, + "flow_matching_loss": 0.012506189988926053, + "grad_norm": 0.08471909910440445, + "learning_rate": 7.228988831445015e-08, + "loss": 0.0125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29110 + }, + { + "epoch": 8.472505091649694, + "flow_matching_loss": 0.010260558873414993, + "grad_norm": 0.0718117207288742, + "learning_rate": 7.06775939009574e-08, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29120 + }, + { + "epoch": 8.475414605760838, + "flow_matching_loss": 0.010077983606606722, + "grad_norm": 0.043804191052913666, + "learning_rate": 6.908344002193256e-08, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29130 + }, + { + "epoch": 8.478324119871981, + "flow_matching_loss": 0.00862187179736793, + "grad_norm": 0.05153152719140053, + "learning_rate": 6.750742861442116e-08, + "loss": 0.0086, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29140 + }, + { + "epoch": 8.481233633983125, + "flow_matching_loss": 0.009293228387832642, + "grad_norm": 0.06202733889222145, + "learning_rate": 6.594956159342636e-08, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29150 + }, + { + "epoch": 8.484143148094269, + "flow_matching_loss": 0.011020161025226117, + "grad_norm": 0.05297967419028282, + "learning_rate": 6.440984085190394e-08, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29160 + }, + { + "epoch": 8.487052662205413, + "flow_matching_loss": 0.00977969579398632, + "grad_norm": 0.06492245942354202, + "learning_rate": 6.288826826075733e-08, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29170 + }, + { + "epoch": 8.489962176316554, + "flow_matching_loss": 0.01016886909492314, + "grad_norm": 0.0654323399066925, + "learning_rate": 6.138484566884261e-08, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29180 + }, + { + "epoch": 8.492871690427698, + "flow_matching_loss": 0.009400333277881145, + "grad_norm": 0.06419173628091812, + "learning_rate": 5.989957490296183e-08, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29190 + }, + { + "epoch": 8.495781204538842, + "flow_matching_loss": 0.00899043339304626, + "grad_norm": 0.05082686245441437, + "learning_rate": 5.843245776785466e-08, + "loss": 0.009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29200 + }, + { + "epoch": 8.498690718649986, + "flow_matching_loss": 0.009460867242887616, + "grad_norm": 0.0631447285413742, + "learning_rate": 5.6983496046210115e-08, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29210 + }, + { + "epoch": 8.50160023276113, + "flow_matching_loss": 0.009627141570672394, + "grad_norm": 0.06712007522583008, + "learning_rate": 5.5552691498653165e-08, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29220 + }, + { + "epoch": 8.504509746872273, + "flow_matching_loss": 0.009267876902595163, + "grad_norm": 0.06442270427942276, + "learning_rate": 5.4140045863743106e-08, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29230 + }, + { + "epoch": 8.507419260983415, + "flow_matching_loss": 0.010597721533849835, + "grad_norm": 0.06009647622704506, + "learning_rate": 5.274556085798188e-08, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29240 + }, + { + "epoch": 8.510328775094559, + "flow_matching_loss": 0.009520055912435054, + "grad_norm": 0.0645410344004631, + "learning_rate": 5.136923817579742e-08, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29250 + }, + { + "epoch": 8.513238289205702, + "flow_matching_loss": 0.009491727082058787, + "grad_norm": 0.08374840021133423, + "learning_rate": 5.001107948955197e-08, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29260 + }, + { + "epoch": 8.516147803316846, + "flow_matching_loss": 0.009523494029417634, + "grad_norm": 0.07592754065990448, + "learning_rate": 4.867108644953711e-08, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29270 + }, + { + "epoch": 8.51905731742799, + "flow_matching_loss": 0.011862742342054844, + "grad_norm": 0.08753779530525208, + "learning_rate": 4.734926068397038e-08, + "loss": 0.0119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29280 + }, + { + "epoch": 8.521966831539133, + "flow_matching_loss": 0.011116762086749077, + "grad_norm": 0.05929351970553398, + "learning_rate": 4.604560379899536e-08, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29290 + }, + { + "epoch": 8.524876345650277, + "flow_matching_loss": 0.009971285983920098, + "grad_norm": 0.06375187635421753, + "learning_rate": 4.476011737867658e-08, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29300 + }, + { + "epoch": 8.52778585976142, + "flow_matching_loss": 0.010381284635514021, + "grad_norm": 0.07766076922416687, + "learning_rate": 4.349280298500291e-08, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29310 + }, + { + "epoch": 8.530695373872563, + "flow_matching_loss": 0.009967388026416302, + "grad_norm": 0.10301542282104492, + "learning_rate": 4.224366215787923e-08, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29320 + }, + { + "epoch": 8.533604887983707, + "flow_matching_loss": 0.00845358455553651, + "grad_norm": 0.09004109352827072, + "learning_rate": 4.101269641513139e-08, + "loss": 0.0085, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29330 + }, + { + "epoch": 8.53651440209485, + "flow_matching_loss": 0.011152794305235147, + "grad_norm": 0.06410057842731476, + "learning_rate": 3.979990725249627e-08, + "loss": 0.0112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29340 + }, + { + "epoch": 8.539423916205994, + "flow_matching_loss": 0.011034189583733678, + "grad_norm": 0.0836617574095726, + "learning_rate": 3.860529614363006e-08, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29350 + }, + { + "epoch": 8.542333430317138, + "flow_matching_loss": 0.010036887181922794, + "grad_norm": 0.05851885303854942, + "learning_rate": 3.742886454009664e-08, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29360 + }, + { + "epoch": 8.545242944428281, + "flow_matching_loss": 0.0088160939514637, + "grad_norm": 0.06534448266029358, + "learning_rate": 3.627061387137087e-08, + "loss": 0.0088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29370 + }, + { + "epoch": 8.548152458539423, + "flow_matching_loss": 0.010594023671001197, + "grad_norm": 0.08545251190662384, + "learning_rate": 3.5130545544835305e-08, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29380 + }, + { + "epoch": 8.551061972650567, + "flow_matching_loss": 0.010958983702585101, + "grad_norm": 0.08254723995923996, + "learning_rate": 3.400866094578348e-08, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29390 + }, + { + "epoch": 8.55397148676171, + "flow_matching_loss": 0.010043667117133736, + "grad_norm": 0.05503053963184357, + "learning_rate": 3.290496143740829e-08, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29400 + }, + { + "epoch": 8.556881000872854, + "flow_matching_loss": 0.012116063851863146, + "grad_norm": 0.07917556166648865, + "learning_rate": 3.18194483608103e-08, + "loss": 0.0121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29410 + }, + { + "epoch": 8.559790514983998, + "flow_matching_loss": 0.009304636204615236, + "grad_norm": 0.10559462755918503, + "learning_rate": 3.075212303498942e-08, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29420 + }, + { + "epoch": 8.562700029095142, + "flow_matching_loss": 0.009938055742532014, + "grad_norm": 0.0757661685347557, + "learning_rate": 2.9702986756846574e-08, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29430 + }, + { + "epoch": 8.565609543206284, + "flow_matching_loss": 0.007690348569303751, + "grad_norm": 0.07824914157390594, + "learning_rate": 2.867204080118202e-08, + "loss": 0.0077, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29440 + }, + { + "epoch": 8.568519057317427, + "flow_matching_loss": 0.010812568617984653, + "grad_norm": 0.08363974094390869, + "learning_rate": 2.7659286420693708e-08, + "loss": 0.0108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29450 + }, + { + "epoch": 8.571428571428571, + "flow_matching_loss": 0.009889220213517546, + "grad_norm": 0.07288940250873566, + "learning_rate": 2.6664724845968934e-08, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29460 + }, + { + "epoch": 8.574338085539715, + "flow_matching_loss": 0.010148403374478221, + "grad_norm": 0.07193626463413239, + "learning_rate": 2.5688357285499343e-08, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29470 + }, + { + "epoch": 8.577247599650859, + "flow_matching_loss": 0.011148522701114416, + "grad_norm": 0.09155832976102829, + "learning_rate": 2.473018492565926e-08, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29480 + }, + { + "epoch": 8.580157113762002, + "flow_matching_loss": 0.008546976279467345, + "grad_norm": 0.07545860856771469, + "learning_rate": 2.3790208930720702e-08, + "loss": 0.0085, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29490 + }, + { + "epoch": 8.583066627873146, + "flow_matching_loss": 0.009334194520488382, + "grad_norm": 0.07865756005048752, + "learning_rate": 2.2868430442843368e-08, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29500 + }, + { + "epoch": 8.585976141984288, + "flow_matching_loss": 0.008878777781501413, + "grad_norm": 0.07473579049110413, + "learning_rate": 2.196485058207298e-08, + "loss": 0.0089, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29510 + }, + { + "epoch": 8.588885656095432, + "flow_matching_loss": 0.00929998173378408, + "grad_norm": 0.06098777428269386, + "learning_rate": 2.1079470446346282e-08, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29520 + }, + { + "epoch": 8.591795170206575, + "flow_matching_loss": 0.010999703826382757, + "grad_norm": 0.048986516892910004, + "learning_rate": 2.0212291111482707e-08, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29530 + }, + { + "epoch": 8.594704684317719, + "flow_matching_loss": 0.009822556050494313, + "grad_norm": 0.07778673619031906, + "learning_rate": 1.9363313631187707e-08, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29540 + }, + { + "epoch": 8.597614198428863, + "flow_matching_loss": 0.011320974491536618, + "grad_norm": 0.08133607357740402, + "learning_rate": 1.853253903704777e-08, + "loss": 0.0113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29550 + }, + { + "epoch": 8.600523712540006, + "flow_matching_loss": 0.008920188527554274, + "grad_norm": 0.06641609221696854, + "learning_rate": 1.7719968338533733e-08, + "loss": 0.0089, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29560 + }, + { + "epoch": 8.603433226651148, + "flow_matching_loss": 0.008805545279756188, + "grad_norm": 0.06057284027338028, + "learning_rate": 1.6925602522992467e-08, + "loss": 0.0088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29570 + }, + { + "epoch": 8.606342740762292, + "flow_matching_loss": 0.009194182744249702, + "grad_norm": 0.08973537385463715, + "learning_rate": 1.6149442555658533e-08, + "loss": 0.0092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29580 + }, + { + "epoch": 8.609252254873436, + "flow_matching_loss": 0.01050971751101315, + "grad_norm": 0.0634363442659378, + "learning_rate": 1.5391489379635858e-08, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29590 + }, + { + "epoch": 8.61216176898458, + "flow_matching_loss": 0.009779860964044929, + "grad_norm": 0.0722789317369461, + "learning_rate": 1.4651743915909399e-08, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29600 + }, + { + "epoch": 8.615071283095723, + "flow_matching_loss": 0.011644937237724662, + "grad_norm": 0.08816242218017578, + "learning_rate": 1.3930207063340139e-08, + "loss": 0.0116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29610 + }, + { + "epoch": 8.617980797206867, + "flow_matching_loss": 0.009380505187436938, + "grad_norm": 0.05989018455147743, + "learning_rate": 1.322687969866343e-08, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29620 + }, + { + "epoch": 8.62089031131801, + "flow_matching_loss": 0.008532678755000234, + "grad_norm": 0.07026875019073486, + "learning_rate": 1.2541762676488989e-08, + "loss": 0.0085, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29630 + }, + { + "epoch": 8.623799825429153, + "flow_matching_loss": 0.008829163433983923, + "grad_norm": 0.0816655084490776, + "learning_rate": 1.1874856829297564e-08, + "loss": 0.0088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29640 + }, + { + "epoch": 8.626709339540296, + "flow_matching_loss": 0.010632879426702857, + "grad_norm": 0.06775835156440735, + "learning_rate": 1.1226162967442611e-08, + "loss": 0.0106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29650 + }, + { + "epoch": 8.62961885365144, + "flow_matching_loss": 0.008672183053568006, + "grad_norm": 0.07282351702451706, + "learning_rate": 1.059568187914861e-08, + "loss": 0.0087, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29660 + }, + { + "epoch": 8.632528367762584, + "flow_matching_loss": 0.009407793451100589, + "grad_norm": 0.07060790807008743, + "learning_rate": 9.983414330511087e-09, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29670 + }, + { + "epoch": 8.635437881873727, + "flow_matching_loss": 0.010255587426945567, + "grad_norm": 0.07762458175420761, + "learning_rate": 9.389361065489932e-09, + "loss": 0.0103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29680 + }, + { + "epoch": 8.638347395984871, + "flow_matching_loss": 0.009856095910072327, + "grad_norm": 0.09145714342594147, + "learning_rate": 8.81352280591774e-09, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29690 + }, + { + "epoch": 8.641256910096015, + "flow_matching_loss": 0.00820228261873126, + "grad_norm": 0.059180259704589844, + "learning_rate": 8.255900251491478e-09, + "loss": 0.0082, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29700 + }, + { + "epoch": 8.644166424207157, + "flow_matching_loss": 0.011752921156585217, + "grad_norm": 0.08370531350374222, + "learning_rate": 7.716494079775816e-09, + "loss": 0.0118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29710 + }, + { + "epoch": 8.6470759383183, + "flow_matching_loss": 0.009541962807998062, + "grad_norm": 0.06585140526294708, + "learning_rate": 7.195304946199799e-09, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29720 + }, + { + "epoch": 8.649985452429444, + "flow_matching_loss": 0.010012993542477488, + "grad_norm": 0.06074671447277069, + "learning_rate": 6.692333484056845e-09, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29730 + }, + { + "epoch": 8.652894966540588, + "flow_matching_loss": 0.01067270953208208, + "grad_norm": 0.07678768038749695, + "learning_rate": 6.207580304506411e-09, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29740 + }, + { + "epoch": 8.655804480651732, + "flow_matching_loss": 0.01005560220219195, + "grad_norm": 0.09038334339857101, + "learning_rate": 5.741045996567329e-09, + "loss": 0.0101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29750 + }, + { + "epoch": 8.658713994762875, + "flow_matching_loss": 0.009510599356144667, + "grad_norm": 0.07134902477264404, + "learning_rate": 5.2927311271244725e-09, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29760 + }, + { + "epoch": 8.661623508874017, + "flow_matching_loss": 0.010937798442319036, + "grad_norm": 0.07753702998161316, + "learning_rate": 4.8626362409204265e-09, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29770 + }, + { + "epoch": 8.664533022985161, + "flow_matching_loss": 0.011035944987088441, + "grad_norm": 0.06986797600984573, + "learning_rate": 4.4507618605638126e-09, + "loss": 0.011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29780 + }, + { + "epoch": 8.667442537096305, + "flow_matching_loss": 0.011066671274602414, + "grad_norm": 0.08310453593730927, + "learning_rate": 4.057108486519301e-09, + "loss": 0.0111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29790 + }, + { + "epoch": 8.670352051207448, + "flow_matching_loss": 0.009976267628371716, + "grad_norm": 0.08631128817796707, + "learning_rate": 3.6816765971126043e-09, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29800 + }, + { + "epoch": 8.673261565318592, + "flow_matching_loss": 0.009956712555140256, + "grad_norm": 0.06346580386161804, + "learning_rate": 3.324466648528812e-09, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29810 + }, + { + "epoch": 8.676171079429736, + "flow_matching_loss": 0.010682950168848038, + "grad_norm": 0.05988394841551781, + "learning_rate": 2.9854790748123917e-09, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29820 + }, + { + "epoch": 8.67908059354088, + "flow_matching_loss": 0.009537119790911675, + "grad_norm": 0.051916614174842834, + "learning_rate": 2.664714287865522e-09, + "loss": 0.0095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29830 + }, + { + "epoch": 8.681990107652021, + "flow_matching_loss": 0.009553043823689222, + "grad_norm": 0.0652647539973259, + "learning_rate": 2.362172677446428e-09, + "loss": 0.0096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29840 + }, + { + "epoch": 8.684899621763165, + "flow_matching_loss": 0.008735716296359896, + "grad_norm": 0.06468670070171356, + "learning_rate": 2.0778546111727137e-09, + "loss": 0.0087, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29850 + }, + { + "epoch": 8.687809135874309, + "flow_matching_loss": 0.009299062564969063, + "grad_norm": 0.08514440059661865, + "learning_rate": 1.811760434516363e-09, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29860 + }, + { + "epoch": 8.690718649985453, + "flow_matching_loss": 0.009395268885418773, + "grad_norm": 0.056820739060640335, + "learning_rate": 1.5638904708087377e-09, + "loss": 0.0094, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29870 + }, + { + "epoch": 8.693628164096596, + "flow_matching_loss": 0.009179613413289189, + "grad_norm": 0.061355527490377426, + "learning_rate": 1.3342450212355805e-09, + "loss": 0.0092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29880 + }, + { + "epoch": 8.69653767820774, + "flow_matching_loss": 0.010685388185083865, + "grad_norm": 0.07539530843496323, + "learning_rate": 1.122824364835351e-09, + "loss": 0.0107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29890 + }, + { + "epoch": 8.699447192318882, + "flow_matching_loss": 0.009321595029905438, + "grad_norm": 0.06827625632286072, + "learning_rate": 9.296287585042196e-10, + "loss": 0.0093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29900 + }, + { + "epoch": 8.702356706430026, + "flow_matching_loss": 0.009120131796225905, + "grad_norm": 0.06553196161985397, + "learning_rate": 7.546584369960697e-10, + "loss": 0.0091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29910 + }, + { + "epoch": 8.70526622054117, + "flow_matching_loss": 0.010548067931085824, + "grad_norm": 0.06549860537052155, + "learning_rate": 5.97913612912504e-10, + "loss": 0.0105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29920 + }, + { + "epoch": 8.708175734652313, + "flow_matching_loss": 0.009003273118287325, + "grad_norm": 0.05257369577884674, + "learning_rate": 4.593944767161684e-10, + "loss": 0.009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29930 + }, + { + "epoch": 8.711085248763457, + "flow_matching_loss": 0.010854490799829364, + "grad_norm": 0.07248915731906891, + "learning_rate": 3.391011967207591e-10, + "loss": 0.0109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29940 + }, + { + "epoch": 8.7139947628746, + "flow_matching_loss": 0.008839306142181157, + "grad_norm": 0.06360756605863571, + "learning_rate": 2.370339190910231e-10, + "loss": 0.0088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29950 + }, + { + "epoch": 8.716904276985744, + "flow_matching_loss": 0.009980500163510441, + "grad_norm": 0.0518195815384388, + "learning_rate": 1.531927678510847e-10, + "loss": 0.01, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29960 + }, + { + "epoch": 8.719813791096886, + "flow_matching_loss": 0.010403203032910823, + "grad_norm": 0.06835220754146576, + "learning_rate": 8.757784487611886e-11, + "loss": 0.0104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29970 + }, + { + "epoch": 8.72272330520803, + "flow_matching_loss": 0.010211311792954803, + "grad_norm": 0.0712846890091896, + "learning_rate": 4.018922989235119e-11, + "loss": 0.0102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29980 + }, + { + "epoch": 8.725632819319173, + "flow_matching_loss": 0.009935767715796828, + "grad_norm": 0.06613810360431671, + "learning_rate": 1.1026980482053972e-11, + "loss": 0.0099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29990 + }, + { + "epoch": 8.728542333430317, + "flow_matching_loss": 0.009836738603189588, + "grad_norm": 0.048139479011297226, + "learning_rate": 9.113208188082922e-14, + "loss": 0.0098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0, + "mgd_loss_weight": 0.0, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 30000 + }, + { + "epoch": 8.728542333430317, + "step": 30000, + "total_flos": 0.0, + "train_loss": 0.013192495517681042, + "train_runtime": 51008.1968, + "train_samples_per_second": 75.282, + "train_steps_per_second": 0.588 + } + ], + "logging_steps": 10, + "max_steps": 30000, + "num_input_tokens_seen": 0, + "num_train_epochs": 9, + "save_steps": 1, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": false, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 0.0, + "train_batch_size": 128, + "trial_name": null, + "trial_params": null +} diff --git a/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/phase3/training_args.bin b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/phase3/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..c0f07a840b416c6bafa97037cd3ba06202fadbfc --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/phase3/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:61c201c0590a7e7402ef8a9b4ff47ca5c6c23b33ee6de020cd86228b8fe323fc +size 6033 diff --git a/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/phase3/model-00001-of-00002.safetensors b/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/phase3/model-00001-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..fdec019ef49989af14dd5df486d225da6b9c9d8a --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/phase3/model-00001-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e21a56d8a7835b42feb6fd8b3b56244b789ab9db60e2c92cc693313b8caeb135 +size 4999367032 diff --git a/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/phase3/model-00002-of-00002.safetensors b/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/phase3/model-00002-of-00002.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..1e45dfe35587fa17308fbc00f637619fe9ceedc1 --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/phase3/model-00002-of-00002.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bc40e7f4662cfce43a74aa8f08b3a1a6bfed19257136e545babb21a7e73a915f +size 2586705312 diff --git a/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/phase3/training_args.bin b/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/phase3/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..17f55ad17fcf82437f3eb0740fac7969edb199af --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/phase3/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a7c8fdab5414ab2211294b97d317d76ceadfc61a731e2598b6fc2bfce4c26ef4 +size 6033