Upload folder using huggingface_hub
Browse files- drawer_20251223_5k/checkpoint-5000/config.json +64 -0
- drawer_20251223_5k/checkpoint-5000/experiment_cfg/metadata.json +807 -0
- drawer_20251223_5k/checkpoint-5000/model-00001-of-00002.safetensors +3 -0
- drawer_20251223_5k/checkpoint-5000/model-00002-of-00002.safetensors +3 -0
- drawer_20251223_5k/checkpoint-5000/model.safetensors.index.json +0 -0
- drawer_20251223_5k/checkpoint-5000/optimizer.pt +3 -0
- drawer_20251223_5k/checkpoint-5000/rng_state.pth +3 -0
- drawer_20251223_5k/checkpoint-5000/scheduler.pt +3 -0
- drawer_20251223_5k/checkpoint-5000/trainer_state.json +3534 -0
- drawer_20251223_5k/config.json +64 -0
- drawer_20251223_5k/experiment_cfg/metadata.json +807 -0
- drawer_20251223_5k/model-00001-of-00002.safetensors +3 -0
- drawer_20251223_5k/model-00002-of-00002.safetensors +3 -0
- drawer_20251223_5k/model.safetensors.index.json +0 -0
- drawer_20251223_5k/runs/Dec17_05-56-02_computeinstance-e00t0xnm5mmbva0n8n/events.out.tfevents.1765950964.computeinstance-e00t0xnm5mmbva0n8n.2615708.0 +3 -0
- drawer_20251223_5k/runs/Dec17_05-56-44_computeinstance-e00t0xnm5mmbva0n8n/events.out.tfevents.1765951006.computeinstance-e00t0xnm5mmbva0n8n.2643153.0 +3 -0
- drawer_20251223_5k/trainer_state.json +3543 -0
- drawer_20251223_5k/training_args.bin +3 -0
drawer_20251223_5k/checkpoint-5000/config.json
ADDED
|
@@ -0,0 +1,64 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"action_dim": 32,
|
| 3 |
+
"action_head_cfg": {
|
| 4 |
+
"action_dim": 32,
|
| 5 |
+
"action_horizon": 16,
|
| 6 |
+
"add_pos_embed": true,
|
| 7 |
+
"backbone_embedding_dim": 2048,
|
| 8 |
+
"diffusion_model_cfg": {
|
| 9 |
+
"attention_head_dim": 48,
|
| 10 |
+
"cross_attention_dim": 2048,
|
| 11 |
+
"dropout": 0.2,
|
| 12 |
+
"final_dropout": true,
|
| 13 |
+
"interleave_self_attention": true,
|
| 14 |
+
"norm_type": "ada_norm",
|
| 15 |
+
"num_attention_heads": 32,
|
| 16 |
+
"num_layers": 16,
|
| 17 |
+
"output_dim": 1024,
|
| 18 |
+
"positional_embeddings": null
|
| 19 |
+
},
|
| 20 |
+
"hidden_size": 1024,
|
| 21 |
+
"input_embedding_dim": 1536,
|
| 22 |
+
"max_action_dim": 32,
|
| 23 |
+
"max_state_dim": 64,
|
| 24 |
+
"model_dtype": "float32",
|
| 25 |
+
"noise_beta_alpha": 1.5,
|
| 26 |
+
"noise_beta_beta": 1.0,
|
| 27 |
+
"noise_s": 0.999,
|
| 28 |
+
"num_inference_timesteps": 4,
|
| 29 |
+
"num_target_vision_tokens": 32,
|
| 30 |
+
"num_timestep_buckets": 1000,
|
| 31 |
+
"tune_diffusion_model": true,
|
| 32 |
+
"tune_projector": true,
|
| 33 |
+
"use_vlln": true,
|
| 34 |
+
"vl_self_attention_cfg": {
|
| 35 |
+
"attention_head_dim": 64,
|
| 36 |
+
"dropout": 0.2,
|
| 37 |
+
"final_dropout": true,
|
| 38 |
+
"num_attention_heads": 32,
|
| 39 |
+
"num_layers": 4,
|
| 40 |
+
"positional_embeddings": null
|
| 41 |
+
}
|
| 42 |
+
},
|
| 43 |
+
"action_horizon": 16,
|
| 44 |
+
"architectures": [
|
| 45 |
+
"GR00T_N1_5"
|
| 46 |
+
],
|
| 47 |
+
"attn_implementation": null,
|
| 48 |
+
"backbone_cfg": {
|
| 49 |
+
"eagle_path": "NVEagle/eagle_er-qwen3_1_7B-Siglip2_400M_stage1_5_128gpu_er_v7_1mlp_nops",
|
| 50 |
+
"load_bf16": false,
|
| 51 |
+
"project_to_dim": null,
|
| 52 |
+
"reproject_vision": false,
|
| 53 |
+
"select_layer": 12,
|
| 54 |
+
"tune_llm": false,
|
| 55 |
+
"tune_visual": true,
|
| 56 |
+
"use_flash_attention": true
|
| 57 |
+
},
|
| 58 |
+
"compute_dtype": "bfloat16",
|
| 59 |
+
"hidden_size": 2048,
|
| 60 |
+
"model_dtype": "float32",
|
| 61 |
+
"model_type": "gr00t_n1_5",
|
| 62 |
+
"torch_dtype": "bfloat16",
|
| 63 |
+
"transformers_version": "4.51.3"
|
| 64 |
+
}
|
drawer_20251223_5k/checkpoint-5000/experiment_cfg/metadata.json
ADDED
|
@@ -0,0 +1,807 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"new_embodiment": {
|
| 3 |
+
"statistics": {
|
| 4 |
+
"state": {
|
| 5 |
+
"base": {
|
| 6 |
+
"max": [
|
| 7 |
+
0.06309662759304047,
|
| 8 |
+
0.17190714180469513,
|
| 9 |
+
0.14714741706848145
|
| 10 |
+
],
|
| 11 |
+
"min": [
|
| 12 |
+
-0.08241941034793854,
|
| 13 |
+
-0.04592813551425934,
|
| 14 |
+
-0.0676141232252121
|
| 15 |
+
],
|
| 16 |
+
"mean": [
|
| 17 |
+
0.0005052018095739186,
|
| 18 |
+
3.843669765046798e-05,
|
| 19 |
+
0.0004570714954752475
|
| 20 |
+
],
|
| 21 |
+
"std": [
|
| 22 |
+
0.006058344151824713,
|
| 23 |
+
0.008796663954854012,
|
| 24 |
+
0.008676460944116116
|
| 25 |
+
],
|
| 26 |
+
"q01": [
|
| 27 |
+
-0.017743217926472425,
|
| 28 |
+
-0.021156907770782708,
|
| 29 |
+
-0.019517241157591344
|
| 30 |
+
],
|
| 31 |
+
"q99": [
|
| 32 |
+
0.024395816065371036,
|
| 33 |
+
0.03118515063077203,
|
| 34 |
+
0.0335056135058403
|
| 35 |
+
]
|
| 36 |
+
},
|
| 37 |
+
"torso": {
|
| 38 |
+
"max": [
|
| 39 |
+
0.21035662293434143
|
| 40 |
+
],
|
| 41 |
+
"min": [
|
| 42 |
+
0.0007334263646043837
|
| 43 |
+
],
|
| 44 |
+
"mean": [
|
| 45 |
+
0.1093309074640274
|
| 46 |
+
],
|
| 47 |
+
"std": [
|
| 48 |
+
0.10130645334720612
|
| 49 |
+
],
|
| 50 |
+
"q01": [
|
| 51 |
+
0.0009143962111556902
|
| 52 |
+
],
|
| 53 |
+
"q99": [
|
| 54 |
+
0.21021610468626023
|
| 55 |
+
]
|
| 56 |
+
},
|
| 57 |
+
"left_arm": {
|
| 58 |
+
"max": [
|
| 59 |
+
1.3228962421417236,
|
| 60 |
+
0.9760400056838989,
|
| 61 |
+
1.048154592514038,
|
| 62 |
+
-0.45797696709632874,
|
| 63 |
+
0.3519979417324066,
|
| 64 |
+
0.01325496006757021,
|
| 65 |
+
1.5758417844772339
|
| 66 |
+
],
|
| 67 |
+
"min": [
|
| 68 |
+
-0.5555572509765625,
|
| 69 |
+
-0.002730769105255604,
|
| 70 |
+
-0.7607836723327637,
|
| 71 |
+
-2.2513277530670166,
|
| 72 |
+
-0.3550925850868225,
|
| 73 |
+
-0.3537217676639557,
|
| 74 |
+
-0.8624185919761658
|
| 75 |
+
],
|
| 76 |
+
"mean": [
|
| 77 |
+
0.1697605848312378,
|
| 78 |
+
0.11266966909170151,
|
| 79 |
+
0.27108195424079895,
|
| 80 |
+
-1.643442988395691,
|
| 81 |
+
-0.13571397960186005,
|
| 82 |
+
-0.09654814004898071,
|
| 83 |
+
0.31360459327697754
|
| 84 |
+
],
|
| 85 |
+
"std": [
|
| 86 |
+
0.1941879391670227,
|
| 87 |
+
0.21234647929668427,
|
| 88 |
+
0.18590451776981354,
|
| 89 |
+
0.2731393277645111,
|
| 90 |
+
0.0512852668762207,
|
| 91 |
+
0.05550851300358772,
|
| 92 |
+
0.44211140275001526
|
| 93 |
+
],
|
| 94 |
+
"q01": [
|
| 95 |
+
-0.4777324840426445,
|
| 96 |
+
-0.000706656820839271,
|
| 97 |
+
-0.1646677777171135,
|
| 98 |
+
-1.987500010728836,
|
| 99 |
+
-0.35049458891153334,
|
| 100 |
+
-0.34794992953538895,
|
| 101 |
+
-0.6912282472848892
|
| 102 |
+
],
|
| 103 |
+
"q99": [
|
| 104 |
+
0.8433255749940849,
|
| 105 |
+
0.7800812005996703,
|
| 106 |
+
0.934328988790512,
|
| 107 |
+
-0.6880783206224441,
|
| 108 |
+
0.05031383018940674,
|
| 109 |
+
-0.0052628690563142405,
|
| 110 |
+
1.570149998664856
|
| 111 |
+
]
|
| 112 |
+
},
|
| 113 |
+
"right_arm": {
|
| 114 |
+
"max": [
|
| 115 |
+
1.3470864295959473,
|
| 116 |
+
0.003337543224915862,
|
| 117 |
+
0.8092557787895203,
|
| 118 |
+
-0.37025120854377747,
|
| 119 |
+
0.3554441034793854,
|
| 120 |
+
0.07210186868906021,
|
| 121 |
+
0.7510986328125
|
| 122 |
+
],
|
| 123 |
+
"min": [
|
| 124 |
+
-0.6000412106513977,
|
| 125 |
+
-0.9089406132698059,
|
| 126 |
+
-1.0145564079284668,
|
| 127 |
+
-2.252852201461792,
|
| 128 |
+
-0.3539217412471771,
|
| 129 |
+
-0.3530242145061493,
|
| 130 |
+
-1.5733824968338013
|
| 131 |
+
],
|
| 132 |
+
"mean": [
|
| 133 |
+
0.1856241077184677,
|
| 134 |
+
-0.12809492647647858,
|
| 135 |
+
-0.24569183588027954,
|
| 136 |
+
-1.6810379028320312,
|
| 137 |
+
0.17203257977962494,
|
| 138 |
+
-0.05976219102740288,
|
| 139 |
+
-0.45409324765205383
|
| 140 |
+
],
|
| 141 |
+
"std": [
|
| 142 |
+
0.2284369170665741,
|
| 143 |
+
0.2075410932302475,
|
| 144 |
+
0.28148508071899414,
|
| 145 |
+
0.29220107197761536,
|
| 146 |
+
0.0935073271393776,
|
| 147 |
+
0.06213328242301941,
|
| 148 |
+
0.46751895546913147
|
| 149 |
+
],
|
| 150 |
+
"q01": [
|
| 151 |
+
-0.4613877633213997,
|
| 152 |
+
-0.8093644535541534,
|
| 153 |
+
-0.9522596222162246,
|
| 154 |
+
-2.1474283027648924,
|
| 155 |
+
-0.26031066954135895,
|
| 156 |
+
-0.3504389730095863,
|
| 157 |
+
-1.570744615793228
|
| 158 |
+
],
|
| 159 |
+
"q99": [
|
| 160 |
+
1.0315392100811003,
|
| 161 |
+
0.000740842941449955,
|
| 162 |
+
0.5001527267694472,
|
| 163 |
+
-0.7204233664274217,
|
| 164 |
+
0.3507135462760925,
|
| 165 |
+
0.03258553277701139,
|
| 166 |
+
0.65115398645401
|
| 167 |
+
]
|
| 168 |
+
},
|
| 169 |
+
"left_hand": {
|
| 170 |
+
"max": [
|
| 171 |
+
0.7979256510734558
|
| 172 |
+
],
|
| 173 |
+
"min": [
|
| 174 |
+
0.308659166097641
|
| 175 |
+
],
|
| 176 |
+
"mean": [
|
| 177 |
+
0.7642289400100708
|
| 178 |
+
],
|
| 179 |
+
"std": [
|
| 180 |
+
0.09134768694639206
|
| 181 |
+
],
|
| 182 |
+
"q01": [
|
| 183 |
+
0.4216100376844406
|
| 184 |
+
],
|
| 185 |
+
"q99": [
|
| 186 |
+
0.7934369444847107
|
| 187 |
+
]
|
| 188 |
+
},
|
| 189 |
+
"left_gripper_qpos": {
|
| 190 |
+
"max": [
|
| 191 |
+
0.7979256510734558
|
| 192 |
+
],
|
| 193 |
+
"min": [
|
| 194 |
+
0.308659166097641
|
| 195 |
+
],
|
| 196 |
+
"mean": [
|
| 197 |
+
0.7642289400100708
|
| 198 |
+
],
|
| 199 |
+
"std": [
|
| 200 |
+
0.09134768694639206
|
| 201 |
+
],
|
| 202 |
+
"q01": [
|
| 203 |
+
0.4216100376844406
|
| 204 |
+
],
|
| 205 |
+
"q99": [
|
| 206 |
+
0.7934369444847107
|
| 207 |
+
]
|
| 208 |
+
},
|
| 209 |
+
"right_hand": {
|
| 210 |
+
"max": [
|
| 211 |
+
0.7982689142227173
|
| 212 |
+
],
|
| 213 |
+
"min": [
|
| 214 |
+
0.31247764825820923
|
| 215 |
+
],
|
| 216 |
+
"mean": [
|
| 217 |
+
0.7446423172950745
|
| 218 |
+
],
|
| 219 |
+
"std": [
|
| 220 |
+
0.11271625757217407
|
| 221 |
+
],
|
| 222 |
+
"q01": [
|
| 223 |
+
0.3954943937063217
|
| 224 |
+
],
|
| 225 |
+
"q99": [
|
| 226 |
+
0.7934401035308838
|
| 227 |
+
]
|
| 228 |
+
},
|
| 229 |
+
"right_gripper_qpos": {
|
| 230 |
+
"max": [
|
| 231 |
+
0.7982689142227173
|
| 232 |
+
],
|
| 233 |
+
"min": [
|
| 234 |
+
0.31247764825820923
|
| 235 |
+
],
|
| 236 |
+
"mean": [
|
| 237 |
+
0.7446423172950745
|
| 238 |
+
],
|
| 239 |
+
"std": [
|
| 240 |
+
0.11271625757217407
|
| 241 |
+
],
|
| 242 |
+
"q01": [
|
| 243 |
+
0.3954943937063217
|
| 244 |
+
],
|
| 245 |
+
"q99": [
|
| 246 |
+
0.7934401035308838
|
| 247 |
+
]
|
| 248 |
+
},
|
| 249 |
+
"left_arm_eef_pos": {
|
| 250 |
+
"max": [
|
| 251 |
+
5.777933597564697,
|
| 252 |
+
-0.46024391055107117,
|
| 253 |
+
1.008889079093933
|
| 254 |
+
],
|
| 255 |
+
"min": [
|
| 256 |
+
0.616549015045166,
|
| 257 |
+
-2.8081092834472656,
|
| 258 |
+
0.6963161826133728
|
| 259 |
+
],
|
| 260 |
+
"mean": [
|
| 261 |
+
3.3723342418670654,
|
| 262 |
+
-0.9225618839263916,
|
| 263 |
+
0.8467212319374084
|
| 264 |
+
],
|
| 265 |
+
"std": [
|
| 266 |
+
1.4625983238220215,
|
| 267 |
+
0.5607711672782898,
|
| 268 |
+
0.09720194339752197
|
| 269 |
+
],
|
| 270 |
+
"q01": [
|
| 271 |
+
0.6381882685422897,
|
| 272 |
+
-2.802727987766266,
|
| 273 |
+
0.7077158093452454
|
| 274 |
+
],
|
| 275 |
+
"q99": [
|
| 276 |
+
5.752657957077027,
|
| 277 |
+
-0.49149866312742235,
|
| 278 |
+
0.9674001401662826
|
| 279 |
+
]
|
| 280 |
+
},
|
| 281 |
+
"left_arm_eef_quat": {
|
| 282 |
+
"max": [
|
| 283 |
+
0.774299681186676,
|
| 284 |
+
0.7777989506721497,
|
| 285 |
+
0.7498747706413269,
|
| 286 |
+
0.8196669220924377
|
| 287 |
+
],
|
| 288 |
+
"min": [
|
| 289 |
+
-0.739790141582489,
|
| 290 |
+
-0.7685531973838806,
|
| 291 |
+
-0.7645168304443359,
|
| 292 |
+
7.452100544469431e-05
|
| 293 |
+
],
|
| 294 |
+
"mean": [
|
| 295 |
+
0.3311719298362732,
|
| 296 |
+
-0.43145933747291565,
|
| 297 |
+
0.23162278532981873,
|
| 298 |
+
0.4911676049232483
|
| 299 |
+
],
|
| 300 |
+
"std": [
|
| 301 |
+
0.35719844698905945,
|
| 302 |
+
0.3199194669723511,
|
| 303 |
+
0.3855322599411011,
|
| 304 |
+
0.1750989556312561
|
| 305 |
+
],
|
| 306 |
+
"q01": [
|
| 307 |
+
-0.6739041733741761,
|
| 308 |
+
-0.7349927252531052,
|
| 309 |
+
-0.7179731333255768,
|
| 310 |
+
0.012307872828096152
|
| 311 |
+
],
|
| 312 |
+
"q99": [
|
| 313 |
+
0.7349334359169006,
|
| 314 |
+
0.6683141338825225,
|
| 315 |
+
0.6888063704967499,
|
| 316 |
+
0.762659209370613
|
| 317 |
+
]
|
| 318 |
+
},
|
| 319 |
+
"right_arm_eef_pos": {
|
| 320 |
+
"max": [
|
| 321 |
+
5.781275272369385,
|
| 322 |
+
-0.4482811987400055,
|
| 323 |
+
1.0768160820007324
|
| 324 |
+
],
|
| 325 |
+
"min": [
|
| 326 |
+
0.6298854947090149,
|
| 327 |
+
-2.4375555515289307,
|
| 328 |
+
0.6910211443901062
|
| 329 |
+
],
|
| 330 |
+
"mean": [
|
| 331 |
+
3.7318050861358643,
|
| 332 |
+
-0.9764854907989502,
|
| 333 |
+
0.8498363494873047
|
| 334 |
+
],
|
| 335 |
+
"std": [
|
| 336 |
+
1.4858821630477905,
|
| 337 |
+
0.5990075469017029,
|
| 338 |
+
0.0975065678358078
|
| 339 |
+
],
|
| 340 |
+
"q01": [
|
| 341 |
+
0.658288916349411,
|
| 342 |
+
-2.4192774152755736,
|
| 343 |
+
0.7090645414590836
|
| 344 |
+
],
|
| 345 |
+
"q99": [
|
| 346 |
+
5.773581142425537,
|
| 347 |
+
-0.5025484305620194,
|
| 348 |
+
1.0039407348632812
|
| 349 |
+
]
|
| 350 |
+
},
|
| 351 |
+
"right_arm_eef_quat": {
|
| 352 |
+
"max": [
|
| 353 |
+
0.727423906326294,
|
| 354 |
+
0.706809401512146,
|
| 355 |
+
0.8224709630012512,
|
| 356 |
+
0.8125941753387451
|
| 357 |
+
],
|
| 358 |
+
"min": [
|
| 359 |
+
-0.746112585067749,
|
| 360 |
+
-0.7207116484642029,
|
| 361 |
+
-0.8228012919425964,
|
| 362 |
+
4.271203579264693e-06
|
| 363 |
+
],
|
| 364 |
+
"mean": [
|
| 365 |
+
0.31211209297180176,
|
| 366 |
+
-0.3976847529411316,
|
| 367 |
+
0.22617658972740173,
|
| 368 |
+
0.4546644687652588
|
| 369 |
+
],
|
| 370 |
+
"std": [
|
| 371 |
+
0.43644431233406067,
|
| 372 |
+
0.24810448288917542,
|
| 373 |
+
0.4512842893600464,
|
| 374 |
+
0.17576341331005096
|
| 375 |
+
],
|
| 376 |
+
"q01": [
|
| 377 |
+
-0.7195372068881989,
|
| 378 |
+
-0.7195094227790833,
|
| 379 |
+
-0.6934988576173783,
|
| 380 |
+
0.00853458046913147
|
| 381 |
+
],
|
| 382 |
+
"q99": [
|
| 383 |
+
0.6808225351572037,
|
| 384 |
+
0.48683623820543287,
|
| 385 |
+
0.7670918434858323,
|
| 386 |
+
0.7782849752902985
|
| 387 |
+
]
|
| 388 |
+
}
|
| 389 |
+
},
|
| 390 |
+
"action": {
|
| 391 |
+
"base": {
|
| 392 |
+
"max": [
|
| 393 |
+
0.0,
|
| 394 |
+
0.0,
|
| 395 |
+
0.0
|
| 396 |
+
],
|
| 397 |
+
"min": [
|
| 398 |
+
0.0,
|
| 399 |
+
0.0,
|
| 400 |
+
0.0
|
| 401 |
+
],
|
| 402 |
+
"mean": [
|
| 403 |
+
0.0,
|
| 404 |
+
0.0,
|
| 405 |
+
0.0
|
| 406 |
+
],
|
| 407 |
+
"std": [
|
| 408 |
+
0.0,
|
| 409 |
+
0.0,
|
| 410 |
+
0.0
|
| 411 |
+
],
|
| 412 |
+
"q01": [
|
| 413 |
+
0.0,
|
| 414 |
+
0.0,
|
| 415 |
+
0.0
|
| 416 |
+
],
|
| 417 |
+
"q99": [
|
| 418 |
+
0.0,
|
| 419 |
+
0.0,
|
| 420 |
+
0.0
|
| 421 |
+
]
|
| 422 |
+
},
|
| 423 |
+
"torso": {
|
| 424 |
+
"max": [
|
| 425 |
+
0.0
|
| 426 |
+
],
|
| 427 |
+
"min": [
|
| 428 |
+
0.0
|
| 429 |
+
],
|
| 430 |
+
"mean": [
|
| 431 |
+
0.0
|
| 432 |
+
],
|
| 433 |
+
"std": [
|
| 434 |
+
0.0
|
| 435 |
+
],
|
| 436 |
+
"q01": [
|
| 437 |
+
0.0
|
| 438 |
+
],
|
| 439 |
+
"q99": [
|
| 440 |
+
0.0
|
| 441 |
+
]
|
| 442 |
+
},
|
| 443 |
+
"left_arm_eef_pos": {
|
| 444 |
+
"max": [
|
| 445 |
+
1.0,
|
| 446 |
+
1.0,
|
| 447 |
+
1.0
|
| 448 |
+
],
|
| 449 |
+
"min": [
|
| 450 |
+
-1.0,
|
| 451 |
+
-1.0,
|
| 452 |
+
-1.0
|
| 453 |
+
],
|
| 454 |
+
"mean": [
|
| 455 |
+
-0.029605818912386894,
|
| 456 |
+
0.008585687726736069,
|
| 457 |
+
-0.012815090827643871
|
| 458 |
+
],
|
| 459 |
+
"std": [
|
| 460 |
+
0.2323869913816452,
|
| 461 |
+
0.20034699141979218,
|
| 462 |
+
0.14409661293029785
|
| 463 |
+
],
|
| 464 |
+
"q01": [
|
| 465 |
+
-1.0,
|
| 466 |
+
-1.0,
|
| 467 |
+
-1.0
|
| 468 |
+
],
|
| 469 |
+
"q99": [
|
| 470 |
+
1.0,
|
| 471 |
+
1.0,
|
| 472 |
+
0.0
|
| 473 |
+
]
|
| 474 |
+
},
|
| 475 |
+
"left_arm_eef_rot": {
|
| 476 |
+
"max": [
|
| 477 |
+
1.0,
|
| 478 |
+
0.0,
|
| 479 |
+
0.44999998807907104
|
| 480 |
+
],
|
| 481 |
+
"min": [
|
| 482 |
+
-0.44999998807907104,
|
| 483 |
+
0.0,
|
| 484 |
+
-0.44999998807907104
|
| 485 |
+
],
|
| 486 |
+
"mean": [
|
| 487 |
+
0.0015839255647733808,
|
| 488 |
+
0.0,
|
| 489 |
+
-0.003083235351368785
|
| 490 |
+
],
|
| 491 |
+
"std": [
|
| 492 |
+
0.09480395913124084,
|
| 493 |
+
0.0,
|
| 494 |
+
0.060461949557065964
|
| 495 |
+
],
|
| 496 |
+
"q01": [
|
| 497 |
+
-0.44999998807907104,
|
| 498 |
+
0.0,
|
| 499 |
+
-0.44999998807907104
|
| 500 |
+
],
|
| 501 |
+
"q99": [
|
| 502 |
+
0.44999998807907104,
|
| 503 |
+
0.0,
|
| 504 |
+
-0.0
|
| 505 |
+
]
|
| 506 |
+
},
|
| 507 |
+
"right_arm_eef_pos": {
|
| 508 |
+
"max": [
|
| 509 |
+
1.0,
|
| 510 |
+
1.0,
|
| 511 |
+
1.0
|
| 512 |
+
],
|
| 513 |
+
"min": [
|
| 514 |
+
-1.0,
|
| 515 |
+
-1.0,
|
| 516 |
+
-1.0
|
| 517 |
+
],
|
| 518 |
+
"mean": [
|
| 519 |
+
-0.029140586033463478,
|
| 520 |
+
0.01636778935790062,
|
| 521 |
+
-0.008501099422574043
|
| 522 |
+
],
|
| 523 |
+
"std": [
|
| 524 |
+
0.2782858908176422,
|
| 525 |
+
0.267070472240448,
|
| 526 |
+
0.19106145203113556
|
| 527 |
+
],
|
| 528 |
+
"q01": [
|
| 529 |
+
-1.0,
|
| 530 |
+
-1.0,
|
| 531 |
+
-1.0
|
| 532 |
+
],
|
| 533 |
+
"q99": [
|
| 534 |
+
1.0,
|
| 535 |
+
1.0,
|
| 536 |
+
1.0
|
| 537 |
+
]
|
| 538 |
+
},
|
| 539 |
+
"right_arm_eef_rot": {
|
| 540 |
+
"max": [
|
| 541 |
+
0.44999998807907104,
|
| 542 |
+
0.0,
|
| 543 |
+
0.44999998807907104
|
| 544 |
+
],
|
| 545 |
+
"min": [
|
| 546 |
+
-0.8999999761581421,
|
| 547 |
+
0.0,
|
| 548 |
+
-0.44999998807907104
|
| 549 |
+
],
|
| 550 |
+
"mean": [
|
| 551 |
+
0.0026645150501281023,
|
| 552 |
+
0.0,
|
| 553 |
+
0.007765152957290411
|
| 554 |
+
],
|
| 555 |
+
"std": [
|
| 556 |
+
0.0954006016254425,
|
| 557 |
+
0.0,
|
| 558 |
+
0.09971662610769272
|
| 559 |
+
],
|
| 560 |
+
"q01": [
|
| 561 |
+
-0.44999998807907104,
|
| 562 |
+
0.0,
|
| 563 |
+
-0.44999998807907104
|
| 564 |
+
],
|
| 565 |
+
"q99": [
|
| 566 |
+
0.44999998807907104,
|
| 567 |
+
0.0,
|
| 568 |
+
0.44999998807907104
|
| 569 |
+
]
|
| 570 |
+
},
|
| 571 |
+
"left_gripper_close": {
|
| 572 |
+
"max": [
|
| 573 |
+
1.0
|
| 574 |
+
],
|
| 575 |
+
"min": [
|
| 576 |
+
-1.0
|
| 577 |
+
],
|
| 578 |
+
"mean": [
|
| 579 |
+
0.7750803828239441
|
| 580 |
+
],
|
| 581 |
+
"std": [
|
| 582 |
+
0.631952166557312
|
| 583 |
+
],
|
| 584 |
+
"q01": [
|
| 585 |
+
-1.0
|
| 586 |
+
],
|
| 587 |
+
"q99": [
|
| 588 |
+
1.0
|
| 589 |
+
]
|
| 590 |
+
},
|
| 591 |
+
"right_gripper_close": {
|
| 592 |
+
"max": [
|
| 593 |
+
1.0
|
| 594 |
+
],
|
| 595 |
+
"min": [
|
| 596 |
+
-1.0
|
| 597 |
+
],
|
| 598 |
+
"mean": [
|
| 599 |
+
0.6190999746322632
|
| 600 |
+
],
|
| 601 |
+
"std": [
|
| 602 |
+
0.7853696942329407
|
| 603 |
+
],
|
| 604 |
+
"q01": [
|
| 605 |
+
-1.0
|
| 606 |
+
],
|
| 607 |
+
"q99": [
|
| 608 |
+
1.0
|
| 609 |
+
]
|
| 610 |
+
}
|
| 611 |
+
}
|
| 612 |
+
},
|
| 613 |
+
"modalities": {
|
| 614 |
+
"video": {
|
| 615 |
+
"cam_robot_0:agentview_center_rgb": {
|
| 616 |
+
"resolution": [
|
| 617 |
+
256,
|
| 618 |
+
256
|
| 619 |
+
],
|
| 620 |
+
"channels": 3,
|
| 621 |
+
"fps": 20.0
|
| 622 |
+
},
|
| 623 |
+
"cam_robot_0:agentview_left_rgb": {
|
| 624 |
+
"resolution": [
|
| 625 |
+
256,
|
| 626 |
+
256
|
| 627 |
+
],
|
| 628 |
+
"channels": 3,
|
| 629 |
+
"fps": 20.0
|
| 630 |
+
},
|
| 631 |
+
"cam_robot_0:agentview_right_rgb": {
|
| 632 |
+
"resolution": [
|
| 633 |
+
256,
|
| 634 |
+
256
|
| 635 |
+
],
|
| 636 |
+
"channels": 3,
|
| 637 |
+
"fps": 20.0
|
| 638 |
+
}
|
| 639 |
+
},
|
| 640 |
+
"state": {
|
| 641 |
+
"base": {
|
| 642 |
+
"absolute": true,
|
| 643 |
+
"rotation_type": null,
|
| 644 |
+
"shape": [
|
| 645 |
+
3
|
| 646 |
+
],
|
| 647 |
+
"continuous": true
|
| 648 |
+
},
|
| 649 |
+
"torso": {
|
| 650 |
+
"absolute": true,
|
| 651 |
+
"rotation_type": null,
|
| 652 |
+
"shape": [
|
| 653 |
+
1
|
| 654 |
+
],
|
| 655 |
+
"continuous": true
|
| 656 |
+
},
|
| 657 |
+
"left_arm": {
|
| 658 |
+
"absolute": true,
|
| 659 |
+
"rotation_type": null,
|
| 660 |
+
"shape": [
|
| 661 |
+
7
|
| 662 |
+
],
|
| 663 |
+
"continuous": true
|
| 664 |
+
},
|
| 665 |
+
"right_arm": {
|
| 666 |
+
"absolute": true,
|
| 667 |
+
"rotation_type": null,
|
| 668 |
+
"shape": [
|
| 669 |
+
7
|
| 670 |
+
],
|
| 671 |
+
"continuous": true
|
| 672 |
+
},
|
| 673 |
+
"left_hand": {
|
| 674 |
+
"absolute": true,
|
| 675 |
+
"rotation_type": null,
|
| 676 |
+
"shape": [
|
| 677 |
+
1
|
| 678 |
+
],
|
| 679 |
+
"continuous": true
|
| 680 |
+
},
|
| 681 |
+
"left_gripper_qpos": {
|
| 682 |
+
"absolute": true,
|
| 683 |
+
"rotation_type": null,
|
| 684 |
+
"shape": [
|
| 685 |
+
1
|
| 686 |
+
],
|
| 687 |
+
"continuous": true
|
| 688 |
+
},
|
| 689 |
+
"right_hand": {
|
| 690 |
+
"absolute": true,
|
| 691 |
+
"rotation_type": null,
|
| 692 |
+
"shape": [
|
| 693 |
+
1
|
| 694 |
+
],
|
| 695 |
+
"continuous": true
|
| 696 |
+
},
|
| 697 |
+
"right_gripper_qpos": {
|
| 698 |
+
"absolute": true,
|
| 699 |
+
"rotation_type": null,
|
| 700 |
+
"shape": [
|
| 701 |
+
1
|
| 702 |
+
],
|
| 703 |
+
"continuous": true
|
| 704 |
+
},
|
| 705 |
+
"left_arm_eef_pos": {
|
| 706 |
+
"absolute": true,
|
| 707 |
+
"rotation_type": null,
|
| 708 |
+
"shape": [
|
| 709 |
+
3
|
| 710 |
+
],
|
| 711 |
+
"continuous": true
|
| 712 |
+
},
|
| 713 |
+
"left_arm_eef_quat": {
|
| 714 |
+
"absolute": true,
|
| 715 |
+
"rotation_type": "quaternion",
|
| 716 |
+
"shape": [
|
| 717 |
+
4
|
| 718 |
+
],
|
| 719 |
+
"continuous": true
|
| 720 |
+
},
|
| 721 |
+
"right_arm_eef_pos": {
|
| 722 |
+
"absolute": true,
|
| 723 |
+
"rotation_type": null,
|
| 724 |
+
"shape": [
|
| 725 |
+
3
|
| 726 |
+
],
|
| 727 |
+
"continuous": true
|
| 728 |
+
},
|
| 729 |
+
"right_arm_eef_quat": {
|
| 730 |
+
"absolute": true,
|
| 731 |
+
"rotation_type": "quaternion",
|
| 732 |
+
"shape": [
|
| 733 |
+
4
|
| 734 |
+
],
|
| 735 |
+
"continuous": true
|
| 736 |
+
}
|
| 737 |
+
},
|
| 738 |
+
"action": {
|
| 739 |
+
"base": {
|
| 740 |
+
"absolute": true,
|
| 741 |
+
"rotation_type": null,
|
| 742 |
+
"shape": [
|
| 743 |
+
3
|
| 744 |
+
],
|
| 745 |
+
"continuous": true
|
| 746 |
+
},
|
| 747 |
+
"torso": {
|
| 748 |
+
"absolute": true,
|
| 749 |
+
"rotation_type": null,
|
| 750 |
+
"shape": [
|
| 751 |
+
1
|
| 752 |
+
],
|
| 753 |
+
"continuous": true
|
| 754 |
+
},
|
| 755 |
+
"left_arm_eef_pos": {
|
| 756 |
+
"absolute": true,
|
| 757 |
+
"rotation_type": null,
|
| 758 |
+
"shape": [
|
| 759 |
+
3
|
| 760 |
+
],
|
| 761 |
+
"continuous": true
|
| 762 |
+
},
|
| 763 |
+
"left_arm_eef_rot": {
|
| 764 |
+
"absolute": true,
|
| 765 |
+
"rotation_type": null,
|
| 766 |
+
"shape": [
|
| 767 |
+
3
|
| 768 |
+
],
|
| 769 |
+
"continuous": true
|
| 770 |
+
},
|
| 771 |
+
"right_arm_eef_pos": {
|
| 772 |
+
"absolute": true,
|
| 773 |
+
"rotation_type": null,
|
| 774 |
+
"shape": [
|
| 775 |
+
3
|
| 776 |
+
],
|
| 777 |
+
"continuous": true
|
| 778 |
+
},
|
| 779 |
+
"right_arm_eef_rot": {
|
| 780 |
+
"absolute": true,
|
| 781 |
+
"rotation_type": null,
|
| 782 |
+
"shape": [
|
| 783 |
+
3
|
| 784 |
+
],
|
| 785 |
+
"continuous": true
|
| 786 |
+
},
|
| 787 |
+
"left_gripper_close": {
|
| 788 |
+
"absolute": true,
|
| 789 |
+
"rotation_type": null,
|
| 790 |
+
"shape": [
|
| 791 |
+
1
|
| 792 |
+
],
|
| 793 |
+
"continuous": true
|
| 794 |
+
},
|
| 795 |
+
"right_gripper_close": {
|
| 796 |
+
"absolute": true,
|
| 797 |
+
"rotation_type": null,
|
| 798 |
+
"shape": [
|
| 799 |
+
1
|
| 800 |
+
],
|
| 801 |
+
"continuous": true
|
| 802 |
+
}
|
| 803 |
+
}
|
| 804 |
+
},
|
| 805 |
+
"embodiment_tag": "new_embodiment"
|
| 806 |
+
}
|
| 807 |
+
}
|
drawer_20251223_5k/checkpoint-5000/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:acbfe35063048b4ca278ee680b282ac260e6c8c211431119a3c7b2a34edda778
|
| 3 |
+
size 4999367032
|
drawer_20251223_5k/checkpoint-5000/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:132ed8db684729787d62f74f99204828f719b125d6406369286000523fa4f471
|
| 3 |
+
size 2586705312
|
drawer_20251223_5k/checkpoint-5000/model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
drawer_20251223_5k/checkpoint-5000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b1aff09dd964b43f5142a25d1b0d17f741971e40386627da43b34b4c33290674
|
| 3 |
+
size 8550720062
|
drawer_20251223_5k/checkpoint-5000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:61789216e07ea10cef9449bb6525adc7d4c2681079592f62ef34ba0a2845a005
|
| 3 |
+
size 14244
|
drawer_20251223_5k/checkpoint-5000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:50875594b892047c8e46a61b82a18becc1a9e9c9af6ce75ec5a292af0d5a8cd2
|
| 3 |
+
size 1064
|
drawer_20251223_5k/checkpoint-5000/trainer_state.json
ADDED
|
@@ -0,0 +1,3534 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 6.7658998646820026,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 5000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.013531799729364006,
|
| 14 |
+
"grad_norm": 7.038280963897705,
|
| 15 |
+
"learning_rate": 3.6e-06,
|
| 16 |
+
"loss": 0.6356,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.02706359945872801,
|
| 21 |
+
"grad_norm": 1.3748748302459717,
|
| 22 |
+
"learning_rate": 7.6e-06,
|
| 23 |
+
"loss": 0.3383,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.04059539918809202,
|
| 28 |
+
"grad_norm": 1.3761250972747803,
|
| 29 |
+
"learning_rate": 1.16e-05,
|
| 30 |
+
"loss": 0.2068,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.05412719891745602,
|
| 35 |
+
"grad_norm": 0.7148615717887878,
|
| 36 |
+
"learning_rate": 1.56e-05,
|
| 37 |
+
"loss": 0.1672,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.06765899864682003,
|
| 42 |
+
"grad_norm": 0.6142189502716064,
|
| 43 |
+
"learning_rate": 1.9600000000000002e-05,
|
| 44 |
+
"loss": 0.1437,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.08119079837618404,
|
| 49 |
+
"grad_norm": 0.6323403120040894,
|
| 50 |
+
"learning_rate": 2.36e-05,
|
| 51 |
+
"loss": 0.1566,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.09472259810554803,
|
| 56 |
+
"grad_norm": 0.7742477059364319,
|
| 57 |
+
"learning_rate": 2.7600000000000003e-05,
|
| 58 |
+
"loss": 0.1181,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.10825439783491204,
|
| 63 |
+
"grad_norm": 0.7195658683776855,
|
| 64 |
+
"learning_rate": 3.16e-05,
|
| 65 |
+
"loss": 0.1254,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.12178619756427606,
|
| 70 |
+
"grad_norm": 0.4564816951751709,
|
| 71 |
+
"learning_rate": 3.56e-05,
|
| 72 |
+
"loss": 0.1064,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.13531799729364005,
|
| 77 |
+
"grad_norm": 0.6378539204597473,
|
| 78 |
+
"learning_rate": 3.960000000000001e-05,
|
| 79 |
+
"loss": 0.1191,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.14884979702300405,
|
| 84 |
+
"grad_norm": 0.41418567299842834,
|
| 85 |
+
"learning_rate": 4.36e-05,
|
| 86 |
+
"loss": 0.1022,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.16238159675236807,
|
| 91 |
+
"grad_norm": 0.5222947001457214,
|
| 92 |
+
"learning_rate": 4.76e-05,
|
| 93 |
+
"loss": 0.0977,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.17591339648173207,
|
| 98 |
+
"grad_norm": 0.40328437089920044,
|
| 99 |
+
"learning_rate": 5.16e-05,
|
| 100 |
+
"loss": 0.0966,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.18944519621109607,
|
| 105 |
+
"grad_norm": 0.45488178730010986,
|
| 106 |
+
"learning_rate": 5.560000000000001e-05,
|
| 107 |
+
"loss": 0.0977,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.2029769959404601,
|
| 112 |
+
"grad_norm": 0.48504000902175903,
|
| 113 |
+
"learning_rate": 5.96e-05,
|
| 114 |
+
"loss": 0.0918,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.2165087956698241,
|
| 119 |
+
"grad_norm": 0.6960858106613159,
|
| 120 |
+
"learning_rate": 6.36e-05,
|
| 121 |
+
"loss": 0.0865,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.23004059539918809,
|
| 126 |
+
"grad_norm": 0.39376893639564514,
|
| 127 |
+
"learning_rate": 6.76e-05,
|
| 128 |
+
"loss": 0.0825,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.2435723951285521,
|
| 133 |
+
"grad_norm": 0.6831949949264526,
|
| 134 |
+
"learning_rate": 7.16e-05,
|
| 135 |
+
"loss": 0.083,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.2571041948579161,
|
| 140 |
+
"grad_norm": 0.4721453785896301,
|
| 141 |
+
"learning_rate": 7.560000000000001e-05,
|
| 142 |
+
"loss": 0.0766,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.2706359945872801,
|
| 147 |
+
"grad_norm": 0.5577224493026733,
|
| 148 |
+
"learning_rate": 7.960000000000001e-05,
|
| 149 |
+
"loss": 0.0808,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.28416779431664413,
|
| 154 |
+
"grad_norm": 0.42680981755256653,
|
| 155 |
+
"learning_rate": 8.36e-05,
|
| 156 |
+
"loss": 0.0821,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.2976995940460081,
|
| 161 |
+
"grad_norm": 0.40946292877197266,
|
| 162 |
+
"learning_rate": 8.76e-05,
|
| 163 |
+
"loss": 0.0714,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.3112313937753721,
|
| 168 |
+
"grad_norm": 0.3602572977542877,
|
| 169 |
+
"learning_rate": 9.16e-05,
|
| 170 |
+
"loss": 0.0775,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.32476319350473615,
|
| 175 |
+
"grad_norm": 0.41732338070869446,
|
| 176 |
+
"learning_rate": 9.56e-05,
|
| 177 |
+
"loss": 0.0715,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.3382949932341001,
|
| 182 |
+
"grad_norm": 0.37370163202285767,
|
| 183 |
+
"learning_rate": 9.960000000000001e-05,
|
| 184 |
+
"loss": 0.071,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.35182679296346414,
|
| 189 |
+
"grad_norm": 0.5187098383903503,
|
| 190 |
+
"learning_rate": 9.999911419878559e-05,
|
| 191 |
+
"loss": 0.071,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.36535859269282817,
|
| 196 |
+
"grad_norm": 0.3645962178707123,
|
| 197 |
+
"learning_rate": 9.999605221019081e-05,
|
| 198 |
+
"loss": 0.0751,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.37889039242219213,
|
| 203 |
+
"grad_norm": 0.480901837348938,
|
| 204 |
+
"learning_rate": 9.999080323230761e-05,
|
| 205 |
+
"loss": 0.0714,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.39242219215155616,
|
| 210 |
+
"grad_norm": 0.40216612815856934,
|
| 211 |
+
"learning_rate": 9.998336749474329e-05,
|
| 212 |
+
"loss": 0.0794,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.4059539918809202,
|
| 217 |
+
"grad_norm": 0.5251531600952148,
|
| 218 |
+
"learning_rate": 9.997374532276107e-05,
|
| 219 |
+
"loss": 0.0651,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.41948579161028415,
|
| 224 |
+
"grad_norm": 0.404252290725708,
|
| 225 |
+
"learning_rate": 9.996193713726596e-05,
|
| 226 |
+
"loss": 0.0694,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.4330175913396482,
|
| 231 |
+
"grad_norm": 0.37963226437568665,
|
| 232 |
+
"learning_rate": 9.994794345478624e-05,
|
| 233 |
+
"loss": 0.0691,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.4465493910690122,
|
| 238 |
+
"grad_norm": 0.38739103078842163,
|
| 239 |
+
"learning_rate": 9.99317648874509e-05,
|
| 240 |
+
"loss": 0.0653,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.46008119079837617,
|
| 245 |
+
"grad_norm": 0.3632943630218506,
|
| 246 |
+
"learning_rate": 9.991340214296292e-05,
|
| 247 |
+
"loss": 0.0623,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.4736129905277402,
|
| 252 |
+
"grad_norm": 0.39061322808265686,
|
| 253 |
+
"learning_rate": 9.989285602456819e-05,
|
| 254 |
+
"loss": 0.0607,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.4871447902571042,
|
| 259 |
+
"grad_norm": 0.3611036241054535,
|
| 260 |
+
"learning_rate": 9.98701274310205e-05,
|
| 261 |
+
"loss": 0.0629,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.5006765899864682,
|
| 266 |
+
"grad_norm": 0.43176841735839844,
|
| 267 |
+
"learning_rate": 9.984521735654218e-05,
|
| 268 |
+
"loss": 0.0639,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.5142083897158322,
|
| 273 |
+
"grad_norm": 0.4912489056587219,
|
| 274 |
+
"learning_rate": 9.981812689078057e-05,
|
| 275 |
+
"loss": 0.0625,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.5277401894451962,
|
| 280 |
+
"grad_norm": 0.33789411187171936,
|
| 281 |
+
"learning_rate": 9.978885721876041e-05,
|
| 282 |
+
"loss": 0.0658,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.5412719891745602,
|
| 287 |
+
"grad_norm": 0.30039742588996887,
|
| 288 |
+
"learning_rate": 9.975740962083198e-05,
|
| 289 |
+
"loss": 0.0567,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.5548037889039242,
|
| 294 |
+
"grad_norm": 0.35562002658843994,
|
| 295 |
+
"learning_rate": 9.972378547261504e-05,
|
| 296 |
+
"loss": 0.0658,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.5683355886332883,
|
| 301 |
+
"grad_norm": 0.37655043601989746,
|
| 302 |
+
"learning_rate": 9.968798624493885e-05,
|
| 303 |
+
"loss": 0.0582,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.5818673883626523,
|
| 308 |
+
"grad_norm": 0.35913047194480896,
|
| 309 |
+
"learning_rate": 9.965001350377753e-05,
|
| 310 |
+
"loss": 0.0618,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.5953991880920162,
|
| 315 |
+
"grad_norm": 0.4832420349121094,
|
| 316 |
+
"learning_rate": 9.960986891018183e-05,
|
| 317 |
+
"loss": 0.061,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.6089309878213802,
|
| 322 |
+
"grad_norm": 0.5011168122291565,
|
| 323 |
+
"learning_rate": 9.95675542202063e-05,
|
| 324 |
+
"loss": 0.0615,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.6224627875507442,
|
| 329 |
+
"grad_norm": 0.26562678813934326,
|
| 330 |
+
"learning_rate": 9.952307128483256e-05,
|
| 331 |
+
"loss": 0.0603,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.6359945872801083,
|
| 336 |
+
"grad_norm": 0.3474351167678833,
|
| 337 |
+
"learning_rate": 9.947642204988835e-05,
|
| 338 |
+
"loss": 0.058,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.6495263870094723,
|
| 343 |
+
"grad_norm": 0.29512283205986023,
|
| 344 |
+
"learning_rate": 9.942760855596226e-05,
|
| 345 |
+
"loss": 0.0585,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.6630581867388363,
|
| 350 |
+
"grad_norm": 0.29113370180130005,
|
| 351 |
+
"learning_rate": 9.937663293831471e-05,
|
| 352 |
+
"loss": 0.0588,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.6765899864682002,
|
| 357 |
+
"grad_norm": 0.2804959714412689,
|
| 358 |
+
"learning_rate": 9.932349742678433e-05,
|
| 359 |
+
"loss": 0.0523,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.6901217861975643,
|
| 364 |
+
"grad_norm": 0.30944955348968506,
|
| 365 |
+
"learning_rate": 9.926820434569051e-05,
|
| 366 |
+
"loss": 0.0572,
|
| 367 |
+
"step": 510
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.7036535859269283,
|
| 371 |
+
"grad_norm": 0.3345446288585663,
|
| 372 |
+
"learning_rate": 9.921075611373179e-05,
|
| 373 |
+
"loss": 0.055,
|
| 374 |
+
"step": 520
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.7171853856562923,
|
| 378 |
+
"grad_norm": 0.32225489616394043,
|
| 379 |
+
"learning_rate": 9.915115524387988e-05,
|
| 380 |
+
"loss": 0.0555,
|
| 381 |
+
"step": 530
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.7307171853856563,
|
| 385 |
+
"grad_norm": 0.5840979218482971,
|
| 386 |
+
"learning_rate": 9.908940434326997e-05,
|
| 387 |
+
"loss": 0.056,
|
| 388 |
+
"step": 540
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.7442489851150202,
|
| 392 |
+
"grad_norm": 0.2597893476486206,
|
| 393 |
+
"learning_rate": 9.902550611308645e-05,
|
| 394 |
+
"loss": 0.0504,
|
| 395 |
+
"step": 550
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.7577807848443843,
|
| 399 |
+
"grad_norm": 0.31497254967689514,
|
| 400 |
+
"learning_rate": 9.895946334844494e-05,
|
| 401 |
+
"loss": 0.052,
|
| 402 |
+
"step": 560
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.7713125845737483,
|
| 406 |
+
"grad_norm": 0.2690199017524719,
|
| 407 |
+
"learning_rate": 9.889127893826989e-05,
|
| 408 |
+
"loss": 0.0538,
|
| 409 |
+
"step": 570
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.7848443843031123,
|
| 413 |
+
"grad_norm": 0.3441496789455414,
|
| 414 |
+
"learning_rate": 9.882095586516831e-05,
|
| 415 |
+
"loss": 0.0542,
|
| 416 |
+
"step": 580
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.7983761840324763,
|
| 420 |
+
"grad_norm": 0.339771568775177,
|
| 421 |
+
"learning_rate": 9.874849720529921e-05,
|
| 422 |
+
"loss": 0.0579,
|
| 423 |
+
"step": 590
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.8119079837618404,
|
| 427 |
+
"grad_norm": 0.26893261075019836,
|
| 428 |
+
"learning_rate": 9.867390612823914e-05,
|
| 429 |
+
"loss": 0.0549,
|
| 430 |
+
"step": 600
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.8254397834912043,
|
| 434 |
+
"grad_norm": 0.2606127858161926,
|
| 435 |
+
"learning_rate": 9.859718589684344e-05,
|
| 436 |
+
"loss": 0.0556,
|
| 437 |
+
"step": 610
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.8389715832205683,
|
| 441 |
+
"grad_norm": 0.29616016149520874,
|
| 442 |
+
"learning_rate": 9.851833986710353e-05,
|
| 443 |
+
"loss": 0.0525,
|
| 444 |
+
"step": 620
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.8525033829499323,
|
| 448 |
+
"grad_norm": 0.2588672339916229,
|
| 449 |
+
"learning_rate": 9.843737148800023e-05,
|
| 450 |
+
"loss": 0.0473,
|
| 451 |
+
"step": 630
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.8660351826792964,
|
| 455 |
+
"grad_norm": 0.23314815759658813,
|
| 456 |
+
"learning_rate": 9.835428430135271e-05,
|
| 457 |
+
"loss": 0.045,
|
| 458 |
+
"step": 640
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.8795669824086604,
|
| 462 |
+
"grad_norm": 0.2904779613018036,
|
| 463 |
+
"learning_rate": 9.82690819416637e-05,
|
| 464 |
+
"loss": 0.0549,
|
| 465 |
+
"step": 650
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.8930987821380244,
|
| 469 |
+
"grad_norm": 0.27372926473617554,
|
| 470 |
+
"learning_rate": 9.818176813596041e-05,
|
| 471 |
+
"loss": 0.0498,
|
| 472 |
+
"step": 660
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.9066305818673883,
|
| 476 |
+
"grad_norm": 0.37094977498054504,
|
| 477 |
+
"learning_rate": 9.809234670363159e-05,
|
| 478 |
+
"loss": 0.0544,
|
| 479 |
+
"step": 670
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.9201623815967523,
|
| 483 |
+
"grad_norm": 0.34140118956565857,
|
| 484 |
+
"learning_rate": 9.800082155626034e-05,
|
| 485 |
+
"loss": 0.0546,
|
| 486 |
+
"step": 680
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.9336941813261164,
|
| 490 |
+
"grad_norm": 0.2957614064216614,
|
| 491 |
+
"learning_rate": 9.790719669745312e-05,
|
| 492 |
+
"loss": 0.0508,
|
| 493 |
+
"step": 690
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.9472259810554804,
|
| 497 |
+
"grad_norm": 0.2524469494819641,
|
| 498 |
+
"learning_rate": 9.781147622266455e-05,
|
| 499 |
+
"loss": 0.0521,
|
| 500 |
+
"step": 700
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.9607577807848444,
|
| 504 |
+
"grad_norm": 0.26907259225845337,
|
| 505 |
+
"learning_rate": 9.771366431901831e-05,
|
| 506 |
+
"loss": 0.0552,
|
| 507 |
+
"step": 710
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.9742895805142084,
|
| 511 |
+
"grad_norm": 0.2906481921672821,
|
| 512 |
+
"learning_rate": 9.761376526512394e-05,
|
| 513 |
+
"loss": 0.0516,
|
| 514 |
+
"step": 720
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.9878213802435724,
|
| 518 |
+
"grad_norm": 0.21193251013755798,
|
| 519 |
+
"learning_rate": 9.751178343088963e-05,
|
| 520 |
+
"loss": 0.0523,
|
| 521 |
+
"step": 730
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 1.0013531799729365,
|
| 525 |
+
"grad_norm": 0.23670238256454468,
|
| 526 |
+
"learning_rate": 9.740772327733123e-05,
|
| 527 |
+
"loss": 0.0466,
|
| 528 |
+
"step": 740
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 1.0148849797023005,
|
| 532 |
+
"grad_norm": 0.23938466608524323,
|
| 533 |
+
"learning_rate": 9.730158935637697e-05,
|
| 534 |
+
"loss": 0.056,
|
| 535 |
+
"step": 750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 1.0284167794316643,
|
| 539 |
+
"grad_norm": 0.23213449120521545,
|
| 540 |
+
"learning_rate": 9.719338631066834e-05,
|
| 541 |
+
"loss": 0.0488,
|
| 542 |
+
"step": 760
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 1.0419485791610283,
|
| 546 |
+
"grad_norm": 0.35314860939979553,
|
| 547 |
+
"learning_rate": 9.708311887335713e-05,
|
| 548 |
+
"loss": 0.0519,
|
| 549 |
+
"step": 770
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.0554803788903924,
|
| 553 |
+
"grad_norm": 0.3346068859100342,
|
| 554 |
+
"learning_rate": 9.697079186789823e-05,
|
| 555 |
+
"loss": 0.0499,
|
| 556 |
+
"step": 780
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 1.0690121786197564,
|
| 560 |
+
"grad_norm": 0.3368820250034332,
|
| 561 |
+
"learning_rate": 9.685641020783876e-05,
|
| 562 |
+
"loss": 0.0506,
|
| 563 |
+
"step": 790
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 1.0825439783491204,
|
| 567 |
+
"grad_norm": 0.25122860074043274,
|
| 568 |
+
"learning_rate": 9.67399788966031e-05,
|
| 569 |
+
"loss": 0.0492,
|
| 570 |
+
"step": 800
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 1.0960757780784844,
|
| 574 |
+
"grad_norm": 0.20261575281620026,
|
| 575 |
+
"learning_rate": 9.662150302727395e-05,
|
| 576 |
+
"loss": 0.0505,
|
| 577 |
+
"step": 810
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 1.1096075778078485,
|
| 581 |
+
"grad_norm": 0.2698753774166107,
|
| 582 |
+
"learning_rate": 9.650098778236968e-05,
|
| 583 |
+
"loss": 0.0521,
|
| 584 |
+
"step": 820
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 1.1231393775372125,
|
| 588 |
+
"grad_norm": 0.29095539450645447,
|
| 589 |
+
"learning_rate": 9.637843843361749e-05,
|
| 590 |
+
"loss": 0.0457,
|
| 591 |
+
"step": 830
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 1.1366711772665765,
|
| 595 |
+
"grad_norm": 0.2353201061487198,
|
| 596 |
+
"learning_rate": 9.62538603417229e-05,
|
| 597 |
+
"loss": 0.044,
|
| 598 |
+
"step": 840
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 1.1502029769959405,
|
| 602 |
+
"grad_norm": 0.2477278709411621,
|
| 603 |
+
"learning_rate": 9.612725895613526e-05,
|
| 604 |
+
"loss": 0.0531,
|
| 605 |
+
"step": 850
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 1.1637347767253043,
|
| 609 |
+
"grad_norm": 0.31642410159111023,
|
| 610 |
+
"learning_rate": 9.599863981480926e-05,
|
| 611 |
+
"loss": 0.0523,
|
| 612 |
+
"step": 860
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 1.1772665764546684,
|
| 616 |
+
"grad_norm": 0.305105984210968,
|
| 617 |
+
"learning_rate": 9.586800854396283e-05,
|
| 618 |
+
"loss": 0.0496,
|
| 619 |
+
"step": 870
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 1.1907983761840324,
|
| 623 |
+
"grad_norm": 0.44601723551750183,
|
| 624 |
+
"learning_rate": 9.573537085783095e-05,
|
| 625 |
+
"loss": 0.0507,
|
| 626 |
+
"step": 880
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 1.2043301759133964,
|
| 630 |
+
"grad_norm": 0.4266583323478699,
|
| 631 |
+
"learning_rate": 9.560073255841571e-05,
|
| 632 |
+
"loss": 0.0485,
|
| 633 |
+
"step": 890
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 1.2178619756427604,
|
| 637 |
+
"grad_norm": 0.19282500445842743,
|
| 638 |
+
"learning_rate": 9.546409953523247e-05,
|
| 639 |
+
"loss": 0.0497,
|
| 640 |
+
"step": 900
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 1.2313937753721245,
|
| 644 |
+
"grad_norm": 0.23939521610736847,
|
| 645 |
+
"learning_rate": 9.532547776505229e-05,
|
| 646 |
+
"loss": 0.0481,
|
| 647 |
+
"step": 910
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 1.2449255751014885,
|
| 651 |
+
"grad_norm": 0.24894139170646667,
|
| 652 |
+
"learning_rate": 9.518487331164048e-05,
|
| 653 |
+
"loss": 0.0483,
|
| 654 |
+
"step": 920
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 1.2584573748308525,
|
| 658 |
+
"grad_norm": 0.3222191631793976,
|
| 659 |
+
"learning_rate": 9.504229232549134e-05,
|
| 660 |
+
"loss": 0.0472,
|
| 661 |
+
"step": 930
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 1.2719891745602165,
|
| 665 |
+
"grad_norm": 0.2381894290447235,
|
| 666 |
+
"learning_rate": 9.489774104355909e-05,
|
| 667 |
+
"loss": 0.0434,
|
| 668 |
+
"step": 940
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 1.2855209742895806,
|
| 672 |
+
"grad_norm": 0.3593835234642029,
|
| 673 |
+
"learning_rate": 9.475122578898507e-05,
|
| 674 |
+
"loss": 0.0509,
|
| 675 |
+
"step": 950
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 1.2990527740189446,
|
| 679 |
+
"grad_norm": 0.33093661069869995,
|
| 680 |
+
"learning_rate": 9.460275297082119e-05,
|
| 681 |
+
"loss": 0.0445,
|
| 682 |
+
"step": 960
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 1.3125845737483086,
|
| 686 |
+
"grad_norm": 0.242259219288826,
|
| 687 |
+
"learning_rate": 9.445232908374948e-05,
|
| 688 |
+
"loss": 0.0498,
|
| 689 |
+
"step": 970
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 1.3261163734776726,
|
| 693 |
+
"grad_norm": 0.29638198018074036,
|
| 694 |
+
"learning_rate": 9.429996070779808e-05,
|
| 695 |
+
"loss": 0.0428,
|
| 696 |
+
"step": 980
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 1.3396481732070367,
|
| 700 |
+
"grad_norm": 0.2383440136909485,
|
| 701 |
+
"learning_rate": 9.414565450805333e-05,
|
| 702 |
+
"loss": 0.0479,
|
| 703 |
+
"step": 990
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 1.3531799729364005,
|
| 707 |
+
"grad_norm": 0.2849849760532379,
|
| 708 |
+
"learning_rate": 9.398941723436831e-05,
|
| 709 |
+
"loss": 0.0452,
|
| 710 |
+
"step": 1000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 1.3667117726657645,
|
| 714 |
+
"grad_norm": 0.26978209614753723,
|
| 715 |
+
"learning_rate": 9.383125572106752e-05,
|
| 716 |
+
"loss": 0.0435,
|
| 717 |
+
"step": 1010
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 1.3802435723951285,
|
| 721 |
+
"grad_norm": 0.2569567561149597,
|
| 722 |
+
"learning_rate": 9.367117688664791e-05,
|
| 723 |
+
"loss": 0.049,
|
| 724 |
+
"step": 1020
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 1.3937753721244925,
|
| 728 |
+
"grad_norm": 0.2809486389160156,
|
| 729 |
+
"learning_rate": 9.35091877334763e-05,
|
| 730 |
+
"loss": 0.0454,
|
| 731 |
+
"step": 1030
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 1.4073071718538566,
|
| 735 |
+
"grad_norm": 0.28595587611198425,
|
| 736 |
+
"learning_rate": 9.334529534748297e-05,
|
| 737 |
+
"loss": 0.0518,
|
| 738 |
+
"step": 1040
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 1.4208389715832206,
|
| 742 |
+
"grad_norm": 0.27410170435905457,
|
| 743 |
+
"learning_rate": 9.317950689785188e-05,
|
| 744 |
+
"loss": 0.0435,
|
| 745 |
+
"step": 1050
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 1.4343707713125846,
|
| 749 |
+
"grad_norm": 0.23330658674240112,
|
| 750 |
+
"learning_rate": 9.301182963670688e-05,
|
| 751 |
+
"loss": 0.0498,
|
| 752 |
+
"step": 1060
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 1.4479025710419486,
|
| 756 |
+
"grad_norm": 0.2517768144607544,
|
| 757 |
+
"learning_rate": 9.284227089879456e-05,
|
| 758 |
+
"loss": 0.0479,
|
| 759 |
+
"step": 1070
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 1.4614343707713127,
|
| 763 |
+
"grad_norm": 0.23287087678909302,
|
| 764 |
+
"learning_rate": 9.26708381011634e-05,
|
| 765 |
+
"loss": 0.0478,
|
| 766 |
+
"step": 1080
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 1.4749661705006765,
|
| 770 |
+
"grad_norm": 0.20903582870960236,
|
| 771 |
+
"learning_rate": 9.249753874283937e-05,
|
| 772 |
+
"loss": 0.0456,
|
| 773 |
+
"step": 1090
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.4884979702300405,
|
| 777 |
+
"grad_norm": 0.21921710669994354,
|
| 778 |
+
"learning_rate": 9.232238040449779e-05,
|
| 779 |
+
"loss": 0.0461,
|
| 780 |
+
"step": 1100
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.5020297699594045,
|
| 784 |
+
"grad_norm": 0.258622407913208,
|
| 785 |
+
"learning_rate": 9.214537074813181e-05,
|
| 786 |
+
"loss": 0.0427,
|
| 787 |
+
"step": 1110
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.5155615696887685,
|
| 791 |
+
"grad_norm": 0.25410041213035583,
|
| 792 |
+
"learning_rate": 9.196651751671724e-05,
|
| 793 |
+
"loss": 0.0483,
|
| 794 |
+
"step": 1120
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.5290933694181326,
|
| 798 |
+
"grad_norm": 0.24682465195655823,
|
| 799 |
+
"learning_rate": 9.178582853387384e-05,
|
| 800 |
+
"loss": 0.0491,
|
| 801 |
+
"step": 1130
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 1.5426251691474966,
|
| 805 |
+
"grad_norm": 0.24718515574932098,
|
| 806 |
+
"learning_rate": 9.160331170352304e-05,
|
| 807 |
+
"loss": 0.0488,
|
| 808 |
+
"step": 1140
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 1.5561569688768606,
|
| 812 |
+
"grad_norm": 0.2638084292411804,
|
| 813 |
+
"learning_rate": 9.141897500954229e-05,
|
| 814 |
+
"loss": 0.0486,
|
| 815 |
+
"step": 1150
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 1.5696887686062246,
|
| 819 |
+
"grad_norm": 0.2443169504404068,
|
| 820 |
+
"learning_rate": 9.123282651541576e-05,
|
| 821 |
+
"loss": 0.0449,
|
| 822 |
+
"step": 1160
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 1.5832205683355887,
|
| 826 |
+
"grad_norm": 0.24525044858455658,
|
| 827 |
+
"learning_rate": 9.104487436388161e-05,
|
| 828 |
+
"loss": 0.044,
|
| 829 |
+
"step": 1170
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 1.5967523680649527,
|
| 833 |
+
"grad_norm": 0.225367471575737,
|
| 834 |
+
"learning_rate": 9.085512677657582e-05,
|
| 835 |
+
"loss": 0.0401,
|
| 836 |
+
"step": 1180
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 1.6102841677943167,
|
| 840 |
+
"grad_norm": 0.2916682958602905,
|
| 841 |
+
"learning_rate": 9.066359205367258e-05,
|
| 842 |
+
"loss": 0.0412,
|
| 843 |
+
"step": 1190
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 1.6238159675236807,
|
| 847 |
+
"grad_norm": 0.21963126957416534,
|
| 848 |
+
"learning_rate": 9.047027857352112e-05,
|
| 849 |
+
"loss": 0.0482,
|
| 850 |
+
"step": 1200
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 1.6373477672530448,
|
| 854 |
+
"grad_norm": 0.25542739033699036,
|
| 855 |
+
"learning_rate": 9.027519479227935e-05,
|
| 856 |
+
"loss": 0.0465,
|
| 857 |
+
"step": 1210
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 1.6508795669824088,
|
| 861 |
+
"grad_norm": 0.1971822828054428,
|
| 862 |
+
"learning_rate": 9.007834924354383e-05,
|
| 863 |
+
"loss": 0.0443,
|
| 864 |
+
"step": 1220
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 1.6644113667117728,
|
| 868 |
+
"grad_norm": 0.1578320562839508,
|
| 869 |
+
"learning_rate": 8.987975053797655e-05,
|
| 870 |
+
"loss": 0.0443,
|
| 871 |
+
"step": 1230
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 1.6779431664411368,
|
| 875 |
+
"grad_norm": 0.25263530015945435,
|
| 876 |
+
"learning_rate": 8.967940736292825e-05,
|
| 877 |
+
"loss": 0.0434,
|
| 878 |
+
"step": 1240
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 1.6914749661705006,
|
| 882 |
+
"grad_norm": 0.2303190976381302,
|
| 883 |
+
"learning_rate": 8.947732848205846e-05,
|
| 884 |
+
"loss": 0.0491,
|
| 885 |
+
"step": 1250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 1.7050067658998647,
|
| 889 |
+
"grad_norm": 0.27885767817497253,
|
| 890 |
+
"learning_rate": 8.927352273495204e-05,
|
| 891 |
+
"loss": 0.041,
|
| 892 |
+
"step": 1260
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 1.7185385656292287,
|
| 896 |
+
"grad_norm": 0.2695358991622925,
|
| 897 |
+
"learning_rate": 8.906799903673265e-05,
|
| 898 |
+
"loss": 0.0434,
|
| 899 |
+
"step": 1270
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 1.7320703653585927,
|
| 903 |
+
"grad_norm": 0.24871905148029327,
|
| 904 |
+
"learning_rate": 8.88607663776726e-05,
|
| 905 |
+
"loss": 0.0439,
|
| 906 |
+
"step": 1280
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 1.7456021650879567,
|
| 910 |
+
"grad_norm": 0.2683444023132324,
|
| 911 |
+
"learning_rate": 8.865183382279978e-05,
|
| 912 |
+
"loss": 0.0442,
|
| 913 |
+
"step": 1290
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 1.7591339648173205,
|
| 917 |
+
"grad_norm": 0.26836928725242615,
|
| 918 |
+
"learning_rate": 8.844121051150096e-05,
|
| 919 |
+
"loss": 0.0425,
|
| 920 |
+
"step": 1300
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 1.7726657645466846,
|
| 924 |
+
"grad_norm": 0.18367311358451843,
|
| 925 |
+
"learning_rate": 8.822890565712211e-05,
|
| 926 |
+
"loss": 0.0429,
|
| 927 |
+
"step": 1310
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 1.7861975642760486,
|
| 931 |
+
"grad_norm": 0.249272882938385,
|
| 932 |
+
"learning_rate": 8.801492854656536e-05,
|
| 933 |
+
"loss": 0.046,
|
| 934 |
+
"step": 1320
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 1.7997293640054126,
|
| 938 |
+
"grad_norm": 0.2543359100818634,
|
| 939 |
+
"learning_rate": 8.779928853988268e-05,
|
| 940 |
+
"loss": 0.0434,
|
| 941 |
+
"step": 1330
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 1.8132611637347766,
|
| 945 |
+
"grad_norm": 0.27556613087654114,
|
| 946 |
+
"learning_rate": 8.758199506986655e-05,
|
| 947 |
+
"loss": 0.0391,
|
| 948 |
+
"step": 1340
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 1.8267929634641407,
|
| 952 |
+
"grad_norm": 0.24716362357139587,
|
| 953 |
+
"learning_rate": 8.73630576416373e-05,
|
| 954 |
+
"loss": 0.0381,
|
| 955 |
+
"step": 1350
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 1.8403247631935047,
|
| 959 |
+
"grad_norm": 0.29948118329048157,
|
| 960 |
+
"learning_rate": 8.714248583222726e-05,
|
| 961 |
+
"loss": 0.043,
|
| 962 |
+
"step": 1360
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 1.8538565629228687,
|
| 966 |
+
"grad_norm": 0.3200075626373291,
|
| 967 |
+
"learning_rate": 8.692028929016196e-05,
|
| 968 |
+
"loss": 0.0462,
|
| 969 |
+
"step": 1370
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 1.8673883626522327,
|
| 973 |
+
"grad_norm": 0.23841191828250885,
|
| 974 |
+
"learning_rate": 8.669647773503797e-05,
|
| 975 |
+
"loss": 0.0411,
|
| 976 |
+
"step": 1380
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 1.8809201623815968,
|
| 980 |
+
"grad_norm": 0.21782785654067993,
|
| 981 |
+
"learning_rate": 8.647106095709773e-05,
|
| 982 |
+
"loss": 0.0435,
|
| 983 |
+
"step": 1390
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 1.8944519621109608,
|
| 987 |
+
"grad_norm": 0.22446033358573914,
|
| 988 |
+
"learning_rate": 8.624404881680139e-05,
|
| 989 |
+
"loss": 0.041,
|
| 990 |
+
"step": 1400
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 1.9079837618403248,
|
| 994 |
+
"grad_norm": 0.21831940114498138,
|
| 995 |
+
"learning_rate": 8.601545124439535e-05,
|
| 996 |
+
"loss": 0.0422,
|
| 997 |
+
"step": 1410
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 1.9215155615696888,
|
| 1001 |
+
"grad_norm": 0.3092878460884094,
|
| 1002 |
+
"learning_rate": 8.5785278239478e-05,
|
| 1003 |
+
"loss": 0.046,
|
| 1004 |
+
"step": 1420
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 1.9350473612990529,
|
| 1008 |
+
"grad_norm": 0.27636203169822693,
|
| 1009 |
+
"learning_rate": 8.555353987056224e-05,
|
| 1010 |
+
"loss": 0.0457,
|
| 1011 |
+
"step": 1430
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 1.9485791610284169,
|
| 1015 |
+
"grad_norm": 0.2327750325202942,
|
| 1016 |
+
"learning_rate": 8.532024627463505e-05,
|
| 1017 |
+
"loss": 0.0431,
|
| 1018 |
+
"step": 1440
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 1.962110960757781,
|
| 1022 |
+
"grad_norm": 0.20582954585552216,
|
| 1023 |
+
"learning_rate": 8.508540765671407e-05,
|
| 1024 |
+
"loss": 0.0408,
|
| 1025 |
+
"step": 1450
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 1.975642760487145,
|
| 1029 |
+
"grad_norm": 0.2168937623500824,
|
| 1030 |
+
"learning_rate": 8.484903428940121e-05,
|
| 1031 |
+
"loss": 0.0405,
|
| 1032 |
+
"step": 1460
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 1.989174560216509,
|
| 1036 |
+
"grad_norm": 0.2852031886577606,
|
| 1037 |
+
"learning_rate": 8.461113651243334e-05,
|
| 1038 |
+
"loss": 0.044,
|
| 1039 |
+
"step": 1470
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 2.002706359945873,
|
| 1043 |
+
"grad_norm": 0.25714772939682007,
|
| 1044 |
+
"learning_rate": 8.437172473222987e-05,
|
| 1045 |
+
"loss": 0.0458,
|
| 1046 |
+
"step": 1480
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 2.016238159675237,
|
| 1050 |
+
"grad_norm": 0.21723440289497375,
|
| 1051 |
+
"learning_rate": 8.413080942143767e-05,
|
| 1052 |
+
"loss": 0.0419,
|
| 1053 |
+
"step": 1490
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 2.029769959404601,
|
| 1057 |
+
"grad_norm": 0.3050743043422699,
|
| 1058 |
+
"learning_rate": 8.388840111847288e-05,
|
| 1059 |
+
"loss": 0.0488,
|
| 1060 |
+
"step": 1500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 2.0433017591339646,
|
| 1064 |
+
"grad_norm": 0.251375287771225,
|
| 1065 |
+
"learning_rate": 8.364451042705998e-05,
|
| 1066 |
+
"loss": 0.0413,
|
| 1067 |
+
"step": 1510
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 2.0568335588633286,
|
| 1071 |
+
"grad_norm": 0.2977745831012726,
|
| 1072 |
+
"learning_rate": 8.33991480157679e-05,
|
| 1073 |
+
"loss": 0.0419,
|
| 1074 |
+
"step": 1520
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 2.0703653585926927,
|
| 1078 |
+
"grad_norm": 0.22888727486133575,
|
| 1079 |
+
"learning_rate": 8.315232461754338e-05,
|
| 1080 |
+
"loss": 0.0432,
|
| 1081 |
+
"step": 1530
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 2.0838971583220567,
|
| 1085 |
+
"grad_norm": 0.2524541914463043,
|
| 1086 |
+
"learning_rate": 8.290405102924144e-05,
|
| 1087 |
+
"loss": 0.0465,
|
| 1088 |
+
"step": 1540
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 2.0974289580514207,
|
| 1092 |
+
"grad_norm": 0.2784097492694855,
|
| 1093 |
+
"learning_rate": 8.265433811115316e-05,
|
| 1094 |
+
"loss": 0.04,
|
| 1095 |
+
"step": 1550
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 2.1109607577807847,
|
| 1099 |
+
"grad_norm": 0.1700204312801361,
|
| 1100 |
+
"learning_rate": 8.240319678653049e-05,
|
| 1101 |
+
"loss": 0.0426,
|
| 1102 |
+
"step": 1560
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 2.1244925575101488,
|
| 1106 |
+
"grad_norm": 0.27274981141090393,
|
| 1107 |
+
"learning_rate": 8.215063804110857e-05,
|
| 1108 |
+
"loss": 0.0427,
|
| 1109 |
+
"step": 1570
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 2.138024357239513,
|
| 1113 |
+
"grad_norm": 0.19321367144584656,
|
| 1114 |
+
"learning_rate": 8.189667292262512e-05,
|
| 1115 |
+
"loss": 0.0401,
|
| 1116 |
+
"step": 1580
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 2.151556156968877,
|
| 1120 |
+
"grad_norm": 0.2078326791524887,
|
| 1121 |
+
"learning_rate": 8.164131254033716e-05,
|
| 1122 |
+
"loss": 0.042,
|
| 1123 |
+
"step": 1590
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 2.165087956698241,
|
| 1127 |
+
"grad_norm": 0.2133825272321701,
|
| 1128 |
+
"learning_rate": 8.138456806453503e-05,
|
| 1129 |
+
"loss": 0.0393,
|
| 1130 |
+
"step": 1600
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 2.178619756427605,
|
| 1134 |
+
"grad_norm": 0.2624984383583069,
|
| 1135 |
+
"learning_rate": 8.112645072605386e-05,
|
| 1136 |
+
"loss": 0.0428,
|
| 1137 |
+
"step": 1610
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 2.192151556156969,
|
| 1141 |
+
"grad_norm": 0.23092296719551086,
|
| 1142 |
+
"learning_rate": 8.086697181578222e-05,
|
| 1143 |
+
"loss": 0.0376,
|
| 1144 |
+
"step": 1620
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 2.205683355886333,
|
| 1148 |
+
"grad_norm": 0.2146901935338974,
|
| 1149 |
+
"learning_rate": 8.060614268416823e-05,
|
| 1150 |
+
"loss": 0.0419,
|
| 1151 |
+
"step": 1630
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 2.219215155615697,
|
| 1155 |
+
"grad_norm": 0.2366856336593628,
|
| 1156 |
+
"learning_rate": 8.034397474072309e-05,
|
| 1157 |
+
"loss": 0.0389,
|
| 1158 |
+
"step": 1640
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 2.232746955345061,
|
| 1162 |
+
"grad_norm": 0.19228318333625793,
|
| 1163 |
+
"learning_rate": 8.008047945352193e-05,
|
| 1164 |
+
"loss": 0.0427,
|
| 1165 |
+
"step": 1650
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 2.246278755074425,
|
| 1169 |
+
"grad_norm": 0.25969430804252625,
|
| 1170 |
+
"learning_rate": 7.981566834870225e-05,
|
| 1171 |
+
"loss": 0.035,
|
| 1172 |
+
"step": 1660
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 2.259810554803789,
|
| 1176 |
+
"grad_norm": 0.23067066073417664,
|
| 1177 |
+
"learning_rate": 7.954955300995961e-05,
|
| 1178 |
+
"loss": 0.0363,
|
| 1179 |
+
"step": 1670
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 2.273342354533153,
|
| 1183 |
+
"grad_norm": 0.20263032615184784,
|
| 1184 |
+
"learning_rate": 7.928214507804104e-05,
|
| 1185 |
+
"loss": 0.0392,
|
| 1186 |
+
"step": 1680
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 2.286874154262517,
|
| 1190 |
+
"grad_norm": 0.2349165827035904,
|
| 1191 |
+
"learning_rate": 7.901345625023576e-05,
|
| 1192 |
+
"loss": 0.0429,
|
| 1193 |
+
"step": 1690
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 2.300405953991881,
|
| 1197 |
+
"grad_norm": 0.2644764482975006,
|
| 1198 |
+
"learning_rate": 7.874349827986354e-05,
|
| 1199 |
+
"loss": 0.0376,
|
| 1200 |
+
"step": 1700
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 2.313937753721245,
|
| 1204 |
+
"grad_norm": 0.1891845464706421,
|
| 1205 |
+
"learning_rate": 7.847228297576053e-05,
|
| 1206 |
+
"loss": 0.0394,
|
| 1207 |
+
"step": 1710
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 2.3274695534506087,
|
| 1211 |
+
"grad_norm": 0.2525619566440582,
|
| 1212 |
+
"learning_rate": 7.819982220176276e-05,
|
| 1213 |
+
"loss": 0.0416,
|
| 1214 |
+
"step": 1720
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 2.3410013531799727,
|
| 1218 |
+
"grad_norm": 0.1973452866077423,
|
| 1219 |
+
"learning_rate": 7.792612787618714e-05,
|
| 1220 |
+
"loss": 0.041,
|
| 1221 |
+
"step": 1730
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 2.3545331529093367,
|
| 1225 |
+
"grad_norm": 0.2173379361629486,
|
| 1226 |
+
"learning_rate": 7.765121197131009e-05,
|
| 1227 |
+
"loss": 0.0389,
|
| 1228 |
+
"step": 1740
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 2.3680649526387008,
|
| 1232 |
+
"grad_norm": 0.2490306794643402,
|
| 1233 |
+
"learning_rate": 7.737508651284391e-05,
|
| 1234 |
+
"loss": 0.0441,
|
| 1235 |
+
"step": 1750
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 2.381596752368065,
|
| 1239 |
+
"grad_norm": 0.26197293400764465,
|
| 1240 |
+
"learning_rate": 7.709776357941069e-05,
|
| 1241 |
+
"loss": 0.0451,
|
| 1242 |
+
"step": 1760
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 2.395128552097429,
|
| 1246 |
+
"grad_norm": 0.2130267322063446,
|
| 1247 |
+
"learning_rate": 7.681925530201392e-05,
|
| 1248 |
+
"loss": 0.0422,
|
| 1249 |
+
"step": 1770
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 2.408660351826793,
|
| 1253 |
+
"grad_norm": 0.19408588111400604,
|
| 1254 |
+
"learning_rate": 7.65395738635079e-05,
|
| 1255 |
+
"loss": 0.0409,
|
| 1256 |
+
"step": 1780
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 2.422192151556157,
|
| 1260 |
+
"grad_norm": 0.3232887387275696,
|
| 1261 |
+
"learning_rate": 7.62587314980648e-05,
|
| 1262 |
+
"loss": 0.0457,
|
| 1263 |
+
"step": 1790
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 2.435723951285521,
|
| 1267 |
+
"grad_norm": 0.21439595520496368,
|
| 1268 |
+
"learning_rate": 7.597674049063947e-05,
|
| 1269 |
+
"loss": 0.0401,
|
| 1270 |
+
"step": 1800
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 2.449255751014885,
|
| 1274 |
+
"grad_norm": 0.20550991594791412,
|
| 1275 |
+
"learning_rate": 7.569361317643211e-05,
|
| 1276 |
+
"loss": 0.0435,
|
| 1277 |
+
"step": 1810
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 2.462787550744249,
|
| 1281 |
+
"grad_norm": 0.1518939584493637,
|
| 1282 |
+
"learning_rate": 7.540936194034865e-05,
|
| 1283 |
+
"loss": 0.0382,
|
| 1284 |
+
"step": 1820
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 2.476319350473613,
|
| 1288 |
+
"grad_norm": 0.22963009774684906,
|
| 1289 |
+
"learning_rate": 7.512399921645901e-05,
|
| 1290 |
+
"loss": 0.0358,
|
| 1291 |
+
"step": 1830
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 2.489851150202977,
|
| 1295 |
+
"grad_norm": 0.17325811088085175,
|
| 1296 |
+
"learning_rate": 7.483753748745317e-05,
|
| 1297 |
+
"loss": 0.0457,
|
| 1298 |
+
"step": 1840
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 2.503382949932341,
|
| 1302 |
+
"grad_norm": 0.18786922097206116,
|
| 1303 |
+
"learning_rate": 7.454998928409516e-05,
|
| 1304 |
+
"loss": 0.0372,
|
| 1305 |
+
"step": 1850
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 2.516914749661705,
|
| 1309 |
+
"grad_norm": 0.18113826215267181,
|
| 1310 |
+
"learning_rate": 7.426136718467493e-05,
|
| 1311 |
+
"loss": 0.0395,
|
| 1312 |
+
"step": 1860
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 2.530446549391069,
|
| 1316 |
+
"grad_norm": 0.21239681541919708,
|
| 1317 |
+
"learning_rate": 7.397168381445812e-05,
|
| 1318 |
+
"loss": 0.0385,
|
| 1319 |
+
"step": 1870
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 2.543978349120433,
|
| 1323 |
+
"grad_norm": 0.17999856173992157,
|
| 1324 |
+
"learning_rate": 7.368095184513377e-05,
|
| 1325 |
+
"loss": 0.0383,
|
| 1326 |
+
"step": 1880
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 2.557510148849797,
|
| 1330 |
+
"grad_norm": 0.21162152290344238,
|
| 1331 |
+
"learning_rate": 7.338918399426005e-05,
|
| 1332 |
+
"loss": 0.0378,
|
| 1333 |
+
"step": 1890
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 2.571041948579161,
|
| 1337 |
+
"grad_norm": 0.2367720603942871,
|
| 1338 |
+
"learning_rate": 7.309639302470801e-05,
|
| 1339 |
+
"loss": 0.0469,
|
| 1340 |
+
"step": 1900
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 2.584573748308525,
|
| 1344 |
+
"grad_norm": 0.21758848428726196,
|
| 1345 |
+
"learning_rate": 7.280259174410312e-05,
|
| 1346 |
+
"loss": 0.0405,
|
| 1347 |
+
"step": 1910
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 2.598105548037889,
|
| 1351 |
+
"grad_norm": 0.1995786428451538,
|
| 1352 |
+
"learning_rate": 7.250779300426517e-05,
|
| 1353 |
+
"loss": 0.0399,
|
| 1354 |
+
"step": 1920
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 2.611637347767253,
|
| 1358 |
+
"grad_norm": 0.19054250419139862,
|
| 1359 |
+
"learning_rate": 7.22120097006461e-05,
|
| 1360 |
+
"loss": 0.0428,
|
| 1361 |
+
"step": 1930
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 2.6251691474966172,
|
| 1365 |
+
"grad_norm": 0.25428929924964905,
|
| 1366 |
+
"learning_rate": 7.191525477176577e-05,
|
| 1367 |
+
"loss": 0.0417,
|
| 1368 |
+
"step": 1940
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 2.6387009472259813,
|
| 1372 |
+
"grad_norm": 0.2825187146663666,
|
| 1373 |
+
"learning_rate": 7.161754119864616e-05,
|
| 1374 |
+
"loss": 0.0412,
|
| 1375 |
+
"step": 1950
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 2.6522327469553453,
|
| 1379 |
+
"grad_norm": 0.17652887105941772,
|
| 1380 |
+
"learning_rate": 7.131888200424339e-05,
|
| 1381 |
+
"loss": 0.0372,
|
| 1382 |
+
"step": 1960
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 2.6657645466847093,
|
| 1386 |
+
"grad_norm": 0.2632055878639221,
|
| 1387 |
+
"learning_rate": 7.101929025287816e-05,
|
| 1388 |
+
"loss": 0.0389,
|
| 1389 |
+
"step": 1970
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 2.6792963464140733,
|
| 1393 |
+
"grad_norm": 0.3633202910423279,
|
| 1394 |
+
"learning_rate": 7.071877904966423e-05,
|
| 1395 |
+
"loss": 0.0476,
|
| 1396 |
+
"step": 1980
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 2.6928281461434374,
|
| 1400 |
+
"grad_norm": 0.21020017564296722,
|
| 1401 |
+
"learning_rate": 7.04173615399351e-05,
|
| 1402 |
+
"loss": 0.041,
|
| 1403 |
+
"step": 1990
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 2.706359945872801,
|
| 1407 |
+
"grad_norm": 0.19143259525299072,
|
| 1408 |
+
"learning_rate": 7.011505090866913e-05,
|
| 1409 |
+
"loss": 0.0351,
|
| 1410 |
+
"step": 2000
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 2.719891745602165,
|
| 1414 |
+
"grad_norm": 0.2139059454202652,
|
| 1415 |
+
"learning_rate": 6.981186037991271e-05,
|
| 1416 |
+
"loss": 0.0388,
|
| 1417 |
+
"step": 2010
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 2.733423545331529,
|
| 1421 |
+
"grad_norm": 0.17334319651126862,
|
| 1422 |
+
"learning_rate": 6.950780321620174e-05,
|
| 1423 |
+
"loss": 0.0344,
|
| 1424 |
+
"step": 2020
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 2.746955345060893,
|
| 1428 |
+
"grad_norm": 0.2146347612142563,
|
| 1429 |
+
"learning_rate": 6.920289271798157e-05,
|
| 1430 |
+
"loss": 0.0449,
|
| 1431 |
+
"step": 2030
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 2.760487144790257,
|
| 1435 |
+
"grad_norm": 0.24033454060554504,
|
| 1436 |
+
"learning_rate": 6.889714222302517e-05,
|
| 1437 |
+
"loss": 0.0401,
|
| 1438 |
+
"step": 2040
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 2.774018944519621,
|
| 1442 |
+
"grad_norm": 0.19013287127017975,
|
| 1443 |
+
"learning_rate": 6.85905651058497e-05,
|
| 1444 |
+
"loss": 0.0375,
|
| 1445 |
+
"step": 2050
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 2.787550744248985,
|
| 1449 |
+
"grad_norm": 0.15871116518974304,
|
| 1450 |
+
"learning_rate": 6.82831747771314e-05,
|
| 1451 |
+
"loss": 0.0368,
|
| 1452 |
+
"step": 2060
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 2.801082543978349,
|
| 1456 |
+
"grad_norm": 0.17981383204460144,
|
| 1457 |
+
"learning_rate": 6.797498468311907e-05,
|
| 1458 |
+
"loss": 0.0388,
|
| 1459 |
+
"step": 2070
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 2.814614343707713,
|
| 1463 |
+
"grad_norm": 0.232260063290596,
|
| 1464 |
+
"learning_rate": 6.766600830504585e-05,
|
| 1465 |
+
"loss": 0.0368,
|
| 1466 |
+
"step": 2080
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 2.828146143437077,
|
| 1470 |
+
"grad_norm": 0.1559726893901825,
|
| 1471 |
+
"learning_rate": 6.735625915853942e-05,
|
| 1472 |
+
"loss": 0.037,
|
| 1473 |
+
"step": 2090
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 2.841677943166441,
|
| 1477 |
+
"grad_norm": 0.1821858137845993,
|
| 1478 |
+
"learning_rate": 6.70457507930309e-05,
|
| 1479 |
+
"loss": 0.035,
|
| 1480 |
+
"step": 2100
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 2.855209742895805,
|
| 1484 |
+
"grad_norm": 0.19013968110084534,
|
| 1485 |
+
"learning_rate": 6.673449679116215e-05,
|
| 1486 |
+
"loss": 0.0385,
|
| 1487 |
+
"step": 2110
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 2.8687415426251692,
|
| 1491 |
+
"grad_norm": 0.169216126203537,
|
| 1492 |
+
"learning_rate": 6.642251076819148e-05,
|
| 1493 |
+
"loss": 0.0322,
|
| 1494 |
+
"step": 2120
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 2.8822733423545333,
|
| 1498 |
+
"grad_norm": 0.16981689631938934,
|
| 1499 |
+
"learning_rate": 6.610980637139827e-05,
|
| 1500 |
+
"loss": 0.0351,
|
| 1501 |
+
"step": 2130
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 2.8958051420838973,
|
| 1505 |
+
"grad_norm": 0.14609754085540771,
|
| 1506 |
+
"learning_rate": 6.579639727948583e-05,
|
| 1507 |
+
"loss": 0.0388,
|
| 1508 |
+
"step": 2140
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 2.9093369418132613,
|
| 1512 |
+
"grad_norm": 0.1752074807882309,
|
| 1513 |
+
"learning_rate": 6.548229720198315e-05,
|
| 1514 |
+
"loss": 0.0381,
|
| 1515 |
+
"step": 2150
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 2.9228687415426253,
|
| 1519 |
+
"grad_norm": 0.20120173692703247,
|
| 1520 |
+
"learning_rate": 6.516751987864517e-05,
|
| 1521 |
+
"loss": 0.0354,
|
| 1522 |
+
"step": 2160
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 2.936400541271989,
|
| 1526 |
+
"grad_norm": 0.15326353907585144,
|
| 1527 |
+
"learning_rate": 6.485207907885175e-05,
|
| 1528 |
+
"loss": 0.0368,
|
| 1529 |
+
"step": 2170
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 2.949932341001353,
|
| 1533 |
+
"grad_norm": 0.22607815265655518,
|
| 1534 |
+
"learning_rate": 6.453598860100536e-05,
|
| 1535 |
+
"loss": 0.0348,
|
| 1536 |
+
"step": 2180
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 2.963464140730717,
|
| 1540 |
+
"grad_norm": 0.21510228514671326,
|
| 1541 |
+
"learning_rate": 6.421926227192749e-05,
|
| 1542 |
+
"loss": 0.0379,
|
| 1543 |
+
"step": 2190
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 2.976995940460081,
|
| 1547 |
+
"grad_norm": 0.16515401005744934,
|
| 1548 |
+
"learning_rate": 6.390191394625381e-05,
|
| 1549 |
+
"loss": 0.0424,
|
| 1550 |
+
"step": 2200
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 2.990527740189445,
|
| 1554 |
+
"grad_norm": 0.19766052067279816,
|
| 1555 |
+
"learning_rate": 6.358395750582817e-05,
|
| 1556 |
+
"loss": 0.0365,
|
| 1557 |
+
"step": 2210
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 3.004059539918809,
|
| 1561 |
+
"grad_norm": 0.13713599741458893,
|
| 1562 |
+
"learning_rate": 6.326540685909532e-05,
|
| 1563 |
+
"loss": 0.0333,
|
| 1564 |
+
"step": 2220
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 3.017591339648173,
|
| 1568 |
+
"grad_norm": 0.2925935685634613,
|
| 1569 |
+
"learning_rate": 6.294627594049249e-05,
|
| 1570 |
+
"loss": 0.0342,
|
| 1571 |
+
"step": 2230
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 3.031123139377537,
|
| 1575 |
+
"grad_norm": 0.2056179940700531,
|
| 1576 |
+
"learning_rate": 6.262657870983989e-05,
|
| 1577 |
+
"loss": 0.0341,
|
| 1578 |
+
"step": 2240
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 3.044654939106901,
|
| 1582 |
+
"grad_norm": 0.34839096665382385,
|
| 1583 |
+
"learning_rate": 6.230632915173009e-05,
|
| 1584 |
+
"loss": 0.0432,
|
| 1585 |
+
"step": 2250
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 3.058186738836265,
|
| 1589 |
+
"grad_norm": 0.24220605194568634,
|
| 1590 |
+
"learning_rate": 6.198554127491622e-05,
|
| 1591 |
+
"loss": 0.0348,
|
| 1592 |
+
"step": 2260
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 3.071718538565629,
|
| 1596 |
+
"grad_norm": 0.17123287916183472,
|
| 1597 |
+
"learning_rate": 6.166422911169923e-05,
|
| 1598 |
+
"loss": 0.0335,
|
| 1599 |
+
"step": 2270
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 3.085250338294993,
|
| 1603 |
+
"grad_norm": 0.2569161653518677,
|
| 1604 |
+
"learning_rate": 6.1342406717314e-05,
|
| 1605 |
+
"loss": 0.0384,
|
| 1606 |
+
"step": 2280
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 3.098782138024357,
|
| 1610 |
+
"grad_norm": 0.15369683504104614,
|
| 1611 |
+
"learning_rate": 6.102008816931466e-05,
|
| 1612 |
+
"loss": 0.0336,
|
| 1613 |
+
"step": 2290
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 3.1123139377537212,
|
| 1617 |
+
"grad_norm": 0.20889005064964294,
|
| 1618 |
+
"learning_rate": 6.069728756695866e-05,
|
| 1619 |
+
"loss": 0.034,
|
| 1620 |
+
"step": 2300
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 3.1258457374830853,
|
| 1624 |
+
"grad_norm": 0.2171514481306076,
|
| 1625 |
+
"learning_rate": 6.037401903059008e-05,
|
| 1626 |
+
"loss": 0.0395,
|
| 1627 |
+
"step": 2310
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 3.1393775372124493,
|
| 1631 |
+
"grad_norm": 0.2569717764854431,
|
| 1632 |
+
"learning_rate": 6.005029670102195e-05,
|
| 1633 |
+
"loss": 0.0315,
|
| 1634 |
+
"step": 2320
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 3.1529093369418133,
|
| 1638 |
+
"grad_norm": 0.17308172583580017,
|
| 1639 |
+
"learning_rate": 5.972613473891766e-05,
|
| 1640 |
+
"loss": 0.0394,
|
| 1641 |
+
"step": 2330
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 3.1664411366711773,
|
| 1645 |
+
"grad_norm": 0.18107521533966064,
|
| 1646 |
+
"learning_rate": 5.940154732417158e-05,
|
| 1647 |
+
"loss": 0.0365,
|
| 1648 |
+
"step": 2340
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 3.1799729364005414,
|
| 1652 |
+
"grad_norm": 0.2191227376461029,
|
| 1653 |
+
"learning_rate": 5.907654865528876e-05,
|
| 1654 |
+
"loss": 0.0374,
|
| 1655 |
+
"step": 2350
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 3.1935047361299054,
|
| 1659 |
+
"grad_norm": 0.22716890275478363,
|
| 1660 |
+
"learning_rate": 5.875115294876381e-05,
|
| 1661 |
+
"loss": 0.035,
|
| 1662 |
+
"step": 2360
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 3.2070365358592694,
|
| 1666 |
+
"grad_norm": 0.22536906599998474,
|
| 1667 |
+
"learning_rate": 5.842537443845908e-05,
|
| 1668 |
+
"loss": 0.0373,
|
| 1669 |
+
"step": 2370
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 3.2205683355886334,
|
| 1673 |
+
"grad_norm": 0.16970916092395782,
|
| 1674 |
+
"learning_rate": 5.809922737498198e-05,
|
| 1675 |
+
"loss": 0.0316,
|
| 1676 |
+
"step": 2380
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 3.2341001353179974,
|
| 1680 |
+
"grad_norm": 0.21783339977264404,
|
| 1681 |
+
"learning_rate": 5.777272602506165e-05,
|
| 1682 |
+
"loss": 0.0343,
|
| 1683 |
+
"step": 2390
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 3.2476319350473615,
|
| 1687 |
+
"grad_norm": 0.34707897901535034,
|
| 1688 |
+
"learning_rate": 5.744588467092483e-05,
|
| 1689 |
+
"loss": 0.038,
|
| 1690 |
+
"step": 2400
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 3.2611637347767255,
|
| 1694 |
+
"grad_norm": 0.2520757019519806,
|
| 1695 |
+
"learning_rate": 5.7118717609671194e-05,
|
| 1696 |
+
"loss": 0.035,
|
| 1697 |
+
"step": 2410
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 3.2746955345060895,
|
| 1701 |
+
"grad_norm": 0.3899773061275482,
|
| 1702 |
+
"learning_rate": 5.679123915264786e-05,
|
| 1703 |
+
"loss": 0.038,
|
| 1704 |
+
"step": 2420
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 3.2882273342354535,
|
| 1708 |
+
"grad_norm": 0.2649787664413452,
|
| 1709 |
+
"learning_rate": 5.646346362482342e-05,
|
| 1710 |
+
"loss": 0.0394,
|
| 1711 |
+
"step": 2430
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 3.301759133964817,
|
| 1715 |
+
"grad_norm": 0.15509116649627686,
|
| 1716 |
+
"learning_rate": 5.613540536416132e-05,
|
| 1717 |
+
"loss": 0.0346,
|
| 1718 |
+
"step": 2440
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 3.315290933694181,
|
| 1722 |
+
"grad_norm": 0.14768008887767792,
|
| 1723 |
+
"learning_rate": 5.5807078720992645e-05,
|
| 1724 |
+
"loss": 0.0335,
|
| 1725 |
+
"step": 2450
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 3.328822733423545,
|
| 1729 |
+
"grad_norm": 0.18915897607803345,
|
| 1730 |
+
"learning_rate": 5.547849805738836e-05,
|
| 1731 |
+
"loss": 0.0376,
|
| 1732 |
+
"step": 2460
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 3.342354533152909,
|
| 1736 |
+
"grad_norm": 0.17673052847385406,
|
| 1737 |
+
"learning_rate": 5.514967774653118e-05,
|
| 1738 |
+
"loss": 0.0347,
|
| 1739 |
+
"step": 2470
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 3.3558863328822732,
|
| 1743 |
+
"grad_norm": 0.18610958755016327,
|
| 1744 |
+
"learning_rate": 5.482063217208674e-05,
|
| 1745 |
+
"loss": 0.0324,
|
| 1746 |
+
"step": 2480
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 3.3694181326116373,
|
| 1750 |
+
"grad_norm": 0.18916338682174683,
|
| 1751 |
+
"learning_rate": 5.449137572757439e-05,
|
| 1752 |
+
"loss": 0.0363,
|
| 1753 |
+
"step": 2490
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 3.3829499323410013,
|
| 1757 |
+
"grad_norm": 0.1974906623363495,
|
| 1758 |
+
"learning_rate": 5.4161922815737696e-05,
|
| 1759 |
+
"loss": 0.0382,
|
| 1760 |
+
"step": 2500
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 3.3964817320703653,
|
| 1764 |
+
"grad_norm": 0.22988206148147583,
|
| 1765 |
+
"learning_rate": 5.3832287847914276e-05,
|
| 1766 |
+
"loss": 0.0364,
|
| 1767 |
+
"step": 2510
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 3.4100135317997293,
|
| 1771 |
+
"grad_norm": 0.19624315202236176,
|
| 1772 |
+
"learning_rate": 5.35024852434055e-05,
|
| 1773 |
+
"loss": 0.0367,
|
| 1774 |
+
"step": 2520
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 3.4235453315290933,
|
| 1778 |
+
"grad_norm": 0.2123284786939621,
|
| 1779 |
+
"learning_rate": 5.317252942884567e-05,
|
| 1780 |
+
"loss": 0.0341,
|
| 1781 |
+
"step": 2530
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 3.4370771312584574,
|
| 1785 |
+
"grad_norm": 0.21080602705478668,
|
| 1786 |
+
"learning_rate": 5.284243483757109e-05,
|
| 1787 |
+
"loss": 0.0327,
|
| 1788 |
+
"step": 2540
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 3.4506089309878214,
|
| 1792 |
+
"grad_norm": 0.16273650527000427,
|
| 1793 |
+
"learning_rate": 5.2512215908988484e-05,
|
| 1794 |
+
"loss": 0.0364,
|
| 1795 |
+
"step": 2550
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 3.4641407307171854,
|
| 1799 |
+
"grad_norm": 0.25207382440567017,
|
| 1800 |
+
"learning_rate": 5.218188708794357e-05,
|
| 1801 |
+
"loss": 0.0353,
|
| 1802 |
+
"step": 2560
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 3.4776725304465494,
|
| 1806 |
+
"grad_norm": 0.19061563909053802,
|
| 1807 |
+
"learning_rate": 5.18514628240891e-05,
|
| 1808 |
+
"loss": 0.0335,
|
| 1809 |
+
"step": 2570
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 3.4912043301759135,
|
| 1813 |
+
"grad_norm": 0.1704767942428589,
|
| 1814 |
+
"learning_rate": 5.1520957571252795e-05,
|
| 1815 |
+
"loss": 0.0378,
|
| 1816 |
+
"step": 2580
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 3.5047361299052775,
|
| 1820 |
+
"grad_norm": 0.15847642719745636,
|
| 1821 |
+
"learning_rate": 5.1190385786805106e-05,
|
| 1822 |
+
"loss": 0.0323,
|
| 1823 |
+
"step": 2590
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 3.5182679296346415,
|
| 1827 |
+
"grad_norm": 0.19395560026168823,
|
| 1828 |
+
"learning_rate": 5.085976193102677e-05,
|
| 1829 |
+
"loss": 0.0337,
|
| 1830 |
+
"step": 2600
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 3.5317997293640055,
|
| 1834 |
+
"grad_norm": 0.2881505787372589,
|
| 1835 |
+
"learning_rate": 5.052910046647634e-05,
|
| 1836 |
+
"loss": 0.0353,
|
| 1837 |
+
"step": 2610
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 3.5453315290933696,
|
| 1841 |
+
"grad_norm": 0.2774830162525177,
|
| 1842 |
+
"learning_rate": 5.0198415857357464e-05,
|
| 1843 |
+
"loss": 0.037,
|
| 1844 |
+
"step": 2620
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 3.558863328822733,
|
| 1848 |
+
"grad_norm": 0.24088598787784576,
|
| 1849 |
+
"learning_rate": 4.9867722568886223e-05,
|
| 1850 |
+
"loss": 0.0333,
|
| 1851 |
+
"step": 2630
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 3.572395128552097,
|
| 1855 |
+
"grad_norm": 0.2002314329147339,
|
| 1856 |
+
"learning_rate": 4.9537035066658314e-05,
|
| 1857 |
+
"loss": 0.0361,
|
| 1858 |
+
"step": 2640
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 3.585926928281461,
|
| 1862 |
+
"grad_norm": 0.3165316581726074,
|
| 1863 |
+
"learning_rate": 4.920636781601638e-05,
|
| 1864 |
+
"loss": 0.0345,
|
| 1865 |
+
"step": 2650
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 3.5994587280108252,
|
| 1869 |
+
"grad_norm": 0.2521885633468628,
|
| 1870 |
+
"learning_rate": 4.88757352814172e-05,
|
| 1871 |
+
"loss": 0.0389,
|
| 1872 |
+
"step": 2660
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 3.6129905277401893,
|
| 1876 |
+
"grad_norm": 0.18471549451351166,
|
| 1877 |
+
"learning_rate": 4.8545151925798924e-05,
|
| 1878 |
+
"loss": 0.0382,
|
| 1879 |
+
"step": 2670
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 3.6265223274695533,
|
| 1883 |
+
"grad_norm": 0.21062928438186646,
|
| 1884 |
+
"learning_rate": 4.821463220994848e-05,
|
| 1885 |
+
"loss": 0.0364,
|
| 1886 |
+
"step": 2680
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 3.6400541271989173,
|
| 1890 |
+
"grad_norm": 0.2001008540391922,
|
| 1891 |
+
"learning_rate": 4.788419059186895e-05,
|
| 1892 |
+
"loss": 0.0362,
|
| 1893 |
+
"step": 2690
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 3.6535859269282813,
|
| 1897 |
+
"grad_norm": 0.16223043203353882,
|
| 1898 |
+
"learning_rate": 4.7553841526147205e-05,
|
| 1899 |
+
"loss": 0.0338,
|
| 1900 |
+
"step": 2700
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 3.6671177266576453,
|
| 1904 |
+
"grad_norm": 0.22189919650554657,
|
| 1905 |
+
"learning_rate": 4.722359946332156e-05,
|
| 1906 |
+
"loss": 0.0369,
|
| 1907 |
+
"step": 2710
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 3.6806495263870094,
|
| 1911 |
+
"grad_norm": 0.22588606178760529,
|
| 1912 |
+
"learning_rate": 4.6893478849249654e-05,
|
| 1913 |
+
"loss": 0.0343,
|
| 1914 |
+
"step": 2720
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 3.6941813261163734,
|
| 1918 |
+
"grad_norm": 0.19541437923908234,
|
| 1919 |
+
"learning_rate": 4.656349412447664e-05,
|
| 1920 |
+
"loss": 0.0345,
|
| 1921 |
+
"step": 2730
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 3.7077131258457374,
|
| 1925 |
+
"grad_norm": 0.2428983449935913,
|
| 1926 |
+
"learning_rate": 4.623365972360337e-05,
|
| 1927 |
+
"loss": 0.0376,
|
| 1928 |
+
"step": 2740
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 3.7212449255751014,
|
| 1932 |
+
"grad_norm": 0.19307422637939453,
|
| 1933 |
+
"learning_rate": 4.590399007465503e-05,
|
| 1934 |
+
"loss": 0.0359,
|
| 1935 |
+
"step": 2750
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 3.7347767253044655,
|
| 1939 |
+
"grad_norm": 0.17713914811611176,
|
| 1940 |
+
"learning_rate": 4.557449959845005e-05,
|
| 1941 |
+
"loss": 0.0333,
|
| 1942 |
+
"step": 2760
|
| 1943 |
+
},
|
| 1944 |
+
{
|
| 1945 |
+
"epoch": 3.7483085250338295,
|
| 1946 |
+
"grad_norm": 0.16902363300323486,
|
| 1947 |
+
"learning_rate": 4.524520270796927e-05,
|
| 1948 |
+
"loss": 0.0308,
|
| 1949 |
+
"step": 2770
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 3.7618403247631935,
|
| 1953 |
+
"grad_norm": 0.162404403090477,
|
| 1954 |
+
"learning_rate": 4.491611380772545e-05,
|
| 1955 |
+
"loss": 0.0324,
|
| 1956 |
+
"step": 2780
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 3.7753721244925575,
|
| 1960 |
+
"grad_norm": 0.17956501245498657,
|
| 1961 |
+
"learning_rate": 4.458724729313318e-05,
|
| 1962 |
+
"loss": 0.031,
|
| 1963 |
+
"step": 2790
|
| 1964 |
+
},
|
| 1965 |
+
{
|
| 1966 |
+
"epoch": 3.7889039242219216,
|
| 1967 |
+
"grad_norm": 0.1946847140789032,
|
| 1968 |
+
"learning_rate": 4.42586175498792e-05,
|
| 1969 |
+
"loss": 0.0353,
|
| 1970 |
+
"step": 2800
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 3.8024357239512856,
|
| 1974 |
+
"grad_norm": 0.16929283738136292,
|
| 1975 |
+
"learning_rate": 4.3930238953293094e-05,
|
| 1976 |
+
"loss": 0.0341,
|
| 1977 |
+
"step": 2810
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 3.8159675236806496,
|
| 1981 |
+
"grad_norm": 0.2041793018579483,
|
| 1982 |
+
"learning_rate": 4.360212586771847e-05,
|
| 1983 |
+
"loss": 0.0328,
|
| 1984 |
+
"step": 2820
|
| 1985 |
+
},
|
| 1986 |
+
{
|
| 1987 |
+
"epoch": 3.8294993234100136,
|
| 1988 |
+
"grad_norm": 0.18276740610599518,
|
| 1989 |
+
"learning_rate": 4.327429264588463e-05,
|
| 1990 |
+
"loss": 0.0344,
|
| 1991 |
+
"step": 2830
|
| 1992 |
+
},
|
| 1993 |
+
{
|
| 1994 |
+
"epoch": 3.8430311231393777,
|
| 1995 |
+
"grad_norm": 0.15672671794891357,
|
| 1996 |
+
"learning_rate": 4.2946753628278725e-05,
|
| 1997 |
+
"loss": 0.0331,
|
| 1998 |
+
"step": 2840
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 3.8565629228687417,
|
| 2002 |
+
"grad_norm": 0.22230131924152374,
|
| 2003 |
+
"learning_rate": 4.2619523142518474e-05,
|
| 2004 |
+
"loss": 0.0335,
|
| 2005 |
+
"step": 2850
|
| 2006 |
+
},
|
| 2007 |
+
{
|
| 2008 |
+
"epoch": 3.8700947225981057,
|
| 2009 |
+
"grad_norm": 0.2010042667388916,
|
| 2010 |
+
"learning_rate": 4.229261550272539e-05,
|
| 2011 |
+
"loss": 0.033,
|
| 2012 |
+
"step": 2860
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 3.8836265223274697,
|
| 2016 |
+
"grad_norm": 0.15846724808216095,
|
| 2017 |
+
"learning_rate": 4.196604500889868e-05,
|
| 2018 |
+
"loss": 0.0344,
|
| 2019 |
+
"step": 2870
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 3.8971583220568338,
|
| 2023 |
+
"grad_norm": 0.2011309415102005,
|
| 2024 |
+
"learning_rate": 4.163982594628969e-05,
|
| 2025 |
+
"loss": 0.0328,
|
| 2026 |
+
"step": 2880
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 3.910690121786198,
|
| 2030 |
+
"grad_norm": 0.17424169182777405,
|
| 2031 |
+
"learning_rate": 4.131397258477702e-05,
|
| 2032 |
+
"loss": 0.0317,
|
| 2033 |
+
"step": 2890
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 3.924221921515562,
|
| 2037 |
+
"grad_norm": 0.1555200070142746,
|
| 2038 |
+
"learning_rate": 4.0988499178242315e-05,
|
| 2039 |
+
"loss": 0.0332,
|
| 2040 |
+
"step": 2900
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 3.937753721244926,
|
| 2044 |
+
"grad_norm": 0.20862902700901031,
|
| 2045 |
+
"learning_rate": 4.066341996394678e-05,
|
| 2046 |
+
"loss": 0.0381,
|
| 2047 |
+
"step": 2910
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 3.9512855209742894,
|
| 2051 |
+
"grad_norm": 0.1553603708744049,
|
| 2052 |
+
"learning_rate": 4.033874916190833e-05,
|
| 2053 |
+
"loss": 0.0349,
|
| 2054 |
+
"step": 2920
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 3.9648173207036534,
|
| 2058 |
+
"grad_norm": 0.13140486180782318,
|
| 2059 |
+
"learning_rate": 4.001450097427966e-05,
|
| 2060 |
+
"loss": 0.0322,
|
| 2061 |
+
"step": 2930
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 3.9783491204330175,
|
| 2065 |
+
"grad_norm": 0.17590387165546417,
|
| 2066 |
+
"learning_rate": 3.9690689584726894e-05,
|
| 2067 |
+
"loss": 0.0336,
|
| 2068 |
+
"step": 2940
|
| 2069 |
+
},
|
| 2070 |
+
{
|
| 2071 |
+
"epoch": 3.9918809201623815,
|
| 2072 |
+
"grad_norm": 0.16788561642169952,
|
| 2073 |
+
"learning_rate": 3.936732915780923e-05,
|
| 2074 |
+
"loss": 0.0305,
|
| 2075 |
+
"step": 2950
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 4.005412719891746,
|
| 2079 |
+
"grad_norm": 0.1186370700597763,
|
| 2080 |
+
"learning_rate": 3.904443383835929e-05,
|
| 2081 |
+
"loss": 0.0318,
|
| 2082 |
+
"step": 2960
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"epoch": 4.01894451962111,
|
| 2086 |
+
"grad_norm": 0.16949886083602905,
|
| 2087 |
+
"learning_rate": 3.872201775086437e-05,
|
| 2088 |
+
"loss": 0.0336,
|
| 2089 |
+
"step": 2970
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"epoch": 4.032476319350474,
|
| 2093 |
+
"grad_norm": 0.18508851528167725,
|
| 2094 |
+
"learning_rate": 3.8400094998848616e-05,
|
| 2095 |
+
"loss": 0.034,
|
| 2096 |
+
"step": 2980
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 4.046008119079838,
|
| 2100 |
+
"grad_norm": 0.26010963320732117,
|
| 2101 |
+
"learning_rate": 3.807867966425611e-05,
|
| 2102 |
+
"loss": 0.0311,
|
| 2103 |
+
"step": 2990
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 4.059539918809202,
|
| 2107 |
+
"grad_norm": 0.23572185635566711,
|
| 2108 |
+
"learning_rate": 3.775778580683481e-05,
|
| 2109 |
+
"loss": 0.0321,
|
| 2110 |
+
"step": 3000
|
| 2111 |
+
},
|
| 2112 |
+
{
|
| 2113 |
+
"epoch": 4.073071718538565,
|
| 2114 |
+
"grad_norm": 0.15540273487567902,
|
| 2115 |
+
"learning_rate": 3.743742746352156e-05,
|
| 2116 |
+
"loss": 0.0325,
|
| 2117 |
+
"step": 3010
|
| 2118 |
+
},
|
| 2119 |
+
{
|
| 2120 |
+
"epoch": 4.086603518267929,
|
| 2121 |
+
"grad_norm": 0.1744704693555832,
|
| 2122 |
+
"learning_rate": 3.711761864782817e-05,
|
| 2123 |
+
"loss": 0.036,
|
| 2124 |
+
"step": 3020
|
| 2125 |
+
},
|
| 2126 |
+
{
|
| 2127 |
+
"epoch": 4.100135317997293,
|
| 2128 |
+
"grad_norm": 0.31342482566833496,
|
| 2129 |
+
"learning_rate": 3.679837334922825e-05,
|
| 2130 |
+
"loss": 0.0316,
|
| 2131 |
+
"step": 3030
|
| 2132 |
+
},
|
| 2133 |
+
{
|
| 2134 |
+
"epoch": 4.113667117726657,
|
| 2135 |
+
"grad_norm": 0.14255890250205994,
|
| 2136 |
+
"learning_rate": 3.647970553254538e-05,
|
| 2137 |
+
"loss": 0.0336,
|
| 2138 |
+
"step": 3040
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 4.127198917456021,
|
| 2142 |
+
"grad_norm": 0.20257729291915894,
|
| 2143 |
+
"learning_rate": 3.61616291373422e-05,
|
| 2144 |
+
"loss": 0.0359,
|
| 2145 |
+
"step": 3050
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 4.140730717185385,
|
| 2149 |
+
"grad_norm": 0.15252985060214996,
|
| 2150 |
+
"learning_rate": 3.584415807731065e-05,
|
| 2151 |
+
"loss": 0.031,
|
| 2152 |
+
"step": 3060
|
| 2153 |
+
},
|
| 2154 |
+
{
|
| 2155 |
+
"epoch": 4.154262516914749,
|
| 2156 |
+
"grad_norm": 0.14593230187892914,
|
| 2157 |
+
"learning_rate": 3.552730623966337e-05,
|
| 2158 |
+
"loss": 0.033,
|
| 2159 |
+
"step": 3070
|
| 2160 |
+
},
|
| 2161 |
+
{
|
| 2162 |
+
"epoch": 4.167794316644113,
|
| 2163 |
+
"grad_norm": 0.19335772097110748,
|
| 2164 |
+
"learning_rate": 3.521108748452617e-05,
|
| 2165 |
+
"loss": 0.0309,
|
| 2166 |
+
"step": 3080
|
| 2167 |
+
},
|
| 2168 |
+
{
|
| 2169 |
+
"epoch": 4.181326116373477,
|
| 2170 |
+
"grad_norm": 0.1618116796016693,
|
| 2171 |
+
"learning_rate": 3.489551564433186e-05,
|
| 2172 |
+
"loss": 0.0333,
|
| 2173 |
+
"step": 3090
|
| 2174 |
+
},
|
| 2175 |
+
{
|
| 2176 |
+
"epoch": 4.194857916102841,
|
| 2177 |
+
"grad_norm": 0.16830849647521973,
|
| 2178 |
+
"learning_rate": 3.4580604523215006e-05,
|
| 2179 |
+
"loss": 0.0314,
|
| 2180 |
+
"step": 3100
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 4.2083897158322054,
|
| 2184 |
+
"grad_norm": 0.15059009194374084,
|
| 2185 |
+
"learning_rate": 3.4266367896408216e-05,
|
| 2186 |
+
"loss": 0.0333,
|
| 2187 |
+
"step": 3110
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 4.2219215155615695,
|
| 2191 |
+
"grad_norm": 0.1731686294078827,
|
| 2192 |
+
"learning_rate": 3.3952819509639534e-05,
|
| 2193 |
+
"loss": 0.0328,
|
| 2194 |
+
"step": 3120
|
| 2195 |
+
},
|
| 2196 |
+
{
|
| 2197 |
+
"epoch": 4.2354533152909335,
|
| 2198 |
+
"grad_norm": 0.13689175248146057,
|
| 2199 |
+
"learning_rate": 3.3639973078531165e-05,
|
| 2200 |
+
"loss": 0.0322,
|
| 2201 |
+
"step": 3130
|
| 2202 |
+
},
|
| 2203 |
+
{
|
| 2204 |
+
"epoch": 4.2489851150202975,
|
| 2205 |
+
"grad_norm": 0.1830020397901535,
|
| 2206 |
+
"learning_rate": 3.332784228799947e-05,
|
| 2207 |
+
"loss": 0.0349,
|
| 2208 |
+
"step": 3140
|
| 2209 |
+
},
|
| 2210 |
+
{
|
| 2211 |
+
"epoch": 4.2625169147496615,
|
| 2212 |
+
"grad_norm": 0.1898087114095688,
|
| 2213 |
+
"learning_rate": 3.301644079165638e-05,
|
| 2214 |
+
"loss": 0.0311,
|
| 2215 |
+
"step": 3150
|
| 2216 |
+
},
|
| 2217 |
+
{
|
| 2218 |
+
"epoch": 4.276048714479026,
|
| 2219 |
+
"grad_norm": 0.14833958446979523,
|
| 2220 |
+
"learning_rate": 3.27057822112122e-05,
|
| 2221 |
+
"loss": 0.0344,
|
| 2222 |
+
"step": 3160
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 4.28958051420839,
|
| 2226 |
+
"grad_norm": 0.232608363032341,
|
| 2227 |
+
"learning_rate": 3.239588013587958e-05,
|
| 2228 |
+
"loss": 0.0325,
|
| 2229 |
+
"step": 3170
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 4.303112313937754,
|
| 2233 |
+
"grad_norm": 0.31661397218704224,
|
| 2234 |
+
"learning_rate": 3.208674812177926e-05,
|
| 2235 |
+
"loss": 0.0314,
|
| 2236 |
+
"step": 3180
|
| 2237 |
+
},
|
| 2238 |
+
{
|
| 2239 |
+
"epoch": 4.316644113667118,
|
| 2240 |
+
"grad_norm": 0.18021564185619354,
|
| 2241 |
+
"learning_rate": 3.177839969134698e-05,
|
| 2242 |
+
"loss": 0.0332,
|
| 2243 |
+
"step": 3190
|
| 2244 |
+
},
|
| 2245 |
+
{
|
| 2246 |
+
"epoch": 4.330175913396482,
|
| 2247 |
+
"grad_norm": 0.16829122602939606,
|
| 2248 |
+
"learning_rate": 3.1470848332742e-05,
|
| 2249 |
+
"loss": 0.0327,
|
| 2250 |
+
"step": 3200
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 4.343707713125846,
|
| 2254 |
+
"grad_norm": 0.209175243973732,
|
| 2255 |
+
"learning_rate": 3.116410749925708e-05,
|
| 2256 |
+
"loss": 0.0287,
|
| 2257 |
+
"step": 3210
|
| 2258 |
+
},
|
| 2259 |
+
{
|
| 2260 |
+
"epoch": 4.35723951285521,
|
| 2261 |
+
"grad_norm": 0.14701929688453674,
|
| 2262 |
+
"learning_rate": 3.085819060872995e-05,
|
| 2263 |
+
"loss": 0.0321,
|
| 2264 |
+
"step": 3220
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 4.370771312584574,
|
| 2268 |
+
"grad_norm": 0.16794687509536743,
|
| 2269 |
+
"learning_rate": 3.055311104295648e-05,
|
| 2270 |
+
"loss": 0.033,
|
| 2271 |
+
"step": 3230
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 4.384303112313938,
|
| 2275 |
+
"grad_norm": 0.1298779845237732,
|
| 2276 |
+
"learning_rate": 3.024888214710517e-05,
|
| 2277 |
+
"loss": 0.0314,
|
| 2278 |
+
"step": 3240
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 4.397834912043302,
|
| 2282 |
+
"grad_norm": 0.1175541877746582,
|
| 2283 |
+
"learning_rate": 2.994551722913349e-05,
|
| 2284 |
+
"loss": 0.033,
|
| 2285 |
+
"step": 3250
|
| 2286 |
+
},
|
| 2287 |
+
{
|
| 2288 |
+
"epoch": 4.411366711772666,
|
| 2289 |
+
"grad_norm": 0.29746726155281067,
|
| 2290 |
+
"learning_rate": 2.9643029559205727e-05,
|
| 2291 |
+
"loss": 0.0358,
|
| 2292 |
+
"step": 3260
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 4.42489851150203,
|
| 2296 |
+
"grad_norm": 0.15811285376548767,
|
| 2297 |
+
"learning_rate": 2.934143236911248e-05,
|
| 2298 |
+
"loss": 0.0314,
|
| 2299 |
+
"step": 3270
|
| 2300 |
+
},
|
| 2301 |
+
{
|
| 2302 |
+
"epoch": 4.438430311231394,
|
| 2303 |
+
"grad_norm": 0.1696750819683075,
|
| 2304 |
+
"learning_rate": 2.90407388516919e-05,
|
| 2305 |
+
"loss": 0.0342,
|
| 2306 |
+
"step": 3280
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 4.451962110960758,
|
| 2310 |
+
"grad_norm": 0.13138838112354279,
|
| 2311 |
+
"learning_rate": 2.8740962160252495e-05,
|
| 2312 |
+
"loss": 0.031,
|
| 2313 |
+
"step": 3290
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 4.465493910690122,
|
| 2317 |
+
"grad_norm": 0.1903950721025467,
|
| 2318 |
+
"learning_rate": 2.844211540799797e-05,
|
| 2319 |
+
"loss": 0.0305,
|
| 2320 |
+
"step": 3300
|
| 2321 |
+
},
|
| 2322 |
+
{
|
| 2323 |
+
"epoch": 4.479025710419486,
|
| 2324 |
+
"grad_norm": 0.2284688502550125,
|
| 2325 |
+
"learning_rate": 2.8144211667453368e-05,
|
| 2326 |
+
"loss": 0.0303,
|
| 2327 |
+
"step": 3310
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 4.49255751014885,
|
| 2331 |
+
"grad_norm": 0.2390712946653366,
|
| 2332 |
+
"learning_rate": 2.7847263969893344e-05,
|
| 2333 |
+
"loss": 0.0299,
|
| 2334 |
+
"step": 3320
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 4.506089309878214,
|
| 2338 |
+
"grad_norm": 0.1737690418958664,
|
| 2339 |
+
"learning_rate": 2.7551285304772206e-05,
|
| 2340 |
+
"loss": 0.0296,
|
| 2341 |
+
"step": 3330
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 4.519621109607578,
|
| 2345 |
+
"grad_norm": 0.14434020221233368,
|
| 2346 |
+
"learning_rate": 2.7256288619155567e-05,
|
| 2347 |
+
"loss": 0.0341,
|
| 2348 |
+
"step": 3340
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 4.533152909336942,
|
| 2352 |
+
"grad_norm": 0.19870182871818542,
|
| 2353 |
+
"learning_rate": 2.6962286817154158e-05,
|
| 2354 |
+
"loss": 0.0349,
|
| 2355 |
+
"step": 3350
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 4.546684709066306,
|
| 2359 |
+
"grad_norm": 0.12394054234027863,
|
| 2360 |
+
"learning_rate": 2.6669292759359166e-05,
|
| 2361 |
+
"loss": 0.0303,
|
| 2362 |
+
"step": 3360
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 4.56021650879567,
|
| 2366 |
+
"grad_norm": 0.13529327511787415,
|
| 2367 |
+
"learning_rate": 2.637731926227993e-05,
|
| 2368 |
+
"loss": 0.033,
|
| 2369 |
+
"step": 3370
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 4.573748308525034,
|
| 2373 |
+
"grad_norm": 0.23153837025165558,
|
| 2374 |
+
"learning_rate": 2.6086379097783033e-05,
|
| 2375 |
+
"loss": 0.0314,
|
| 2376 |
+
"step": 3380
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 4.587280108254398,
|
| 2380 |
+
"grad_norm": 0.1868782937526703,
|
| 2381 |
+
"learning_rate": 2.579648499253377e-05,
|
| 2382 |
+
"loss": 0.03,
|
| 2383 |
+
"step": 3390
|
| 2384 |
+
},
|
| 2385 |
+
{
|
| 2386 |
+
"epoch": 4.600811907983762,
|
| 2387 |
+
"grad_norm": 0.20445260405540466,
|
| 2388 |
+
"learning_rate": 2.5507649627439466e-05,
|
| 2389 |
+
"loss": 0.0343,
|
| 2390 |
+
"step": 3400
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 4.614343707713126,
|
| 2394 |
+
"grad_norm": 0.1350487917661667,
|
| 2395 |
+
"learning_rate": 2.5219885637094653e-05,
|
| 2396 |
+
"loss": 0.0295,
|
| 2397 |
+
"step": 3410
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 4.62787550744249,
|
| 2401 |
+
"grad_norm": 0.18250331282615662,
|
| 2402 |
+
"learning_rate": 2.4933205609228533e-05,
|
| 2403 |
+
"loss": 0.0317,
|
| 2404 |
+
"step": 3420
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"epoch": 4.641407307171854,
|
| 2408 |
+
"grad_norm": 0.36254093050956726,
|
| 2409 |
+
"learning_rate": 2.464762208415419e-05,
|
| 2410 |
+
"loss": 0.0318,
|
| 2411 |
+
"step": 3430
|
| 2412 |
+
},
|
| 2413 |
+
{
|
| 2414 |
+
"epoch": 4.654939106901217,
|
| 2415 |
+
"grad_norm": 0.17919518053531647,
|
| 2416 |
+
"learning_rate": 2.4363147554220213e-05,
|
| 2417 |
+
"loss": 0.0345,
|
| 2418 |
+
"step": 3440
|
| 2419 |
+
},
|
| 2420 |
+
{
|
| 2421 |
+
"epoch": 4.668470906630581,
|
| 2422 |
+
"grad_norm": 0.15475888550281525,
|
| 2423 |
+
"learning_rate": 2.407979446326411e-05,
|
| 2424 |
+
"loss": 0.0288,
|
| 2425 |
+
"step": 3450
|
| 2426 |
+
},
|
| 2427 |
+
{
|
| 2428 |
+
"epoch": 4.682002706359945,
|
| 2429 |
+
"grad_norm": 0.18196256458759308,
|
| 2430 |
+
"learning_rate": 2.379757520606799e-05,
|
| 2431 |
+
"loss": 0.0341,
|
| 2432 |
+
"step": 3460
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 4.6955345060893094,
|
| 2436 |
+
"grad_norm": 0.11349283903837204,
|
| 2437 |
+
"learning_rate": 2.3516502127816455e-05,
|
| 2438 |
+
"loss": 0.0275,
|
| 2439 |
+
"step": 3470
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 4.7090663058186735,
|
| 2443 |
+
"grad_norm": 0.12871690094470978,
|
| 2444 |
+
"learning_rate": 2.323658752355647e-05,
|
| 2445 |
+
"loss": 0.0294,
|
| 2446 |
+
"step": 3480
|
| 2447 |
+
},
|
| 2448 |
+
{
|
| 2449 |
+
"epoch": 4.7225981055480375,
|
| 2450 |
+
"grad_norm": 0.136393204331398,
|
| 2451 |
+
"learning_rate": 2.2957843637659654e-05,
|
| 2452 |
+
"loss": 0.0305,
|
| 2453 |
+
"step": 3490
|
| 2454 |
+
},
|
| 2455 |
+
{
|
| 2456 |
+
"epoch": 4.7361299052774015,
|
| 2457 |
+
"grad_norm": 0.1588805764913559,
|
| 2458 |
+
"learning_rate": 2.2680282663286552e-05,
|
| 2459 |
+
"loss": 0.0328,
|
| 2460 |
+
"step": 3500
|
| 2461 |
+
},
|
| 2462 |
+
{
|
| 2463 |
+
"epoch": 4.7496617050067655,
|
| 2464 |
+
"grad_norm": 0.16840600967407227,
|
| 2465 |
+
"learning_rate": 2.2403916741853364e-05,
|
| 2466 |
+
"loss": 0.0306,
|
| 2467 |
+
"step": 3510
|
| 2468 |
+
},
|
| 2469 |
+
{
|
| 2470 |
+
"epoch": 4.76319350473613,
|
| 2471 |
+
"grad_norm": 0.10826493799686432,
|
| 2472 |
+
"learning_rate": 2.2128757962500817e-05,
|
| 2473 |
+
"loss": 0.0268,
|
| 2474 |
+
"step": 3520
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 4.776725304465494,
|
| 2478 |
+
"grad_norm": 0.223701611161232,
|
| 2479 |
+
"learning_rate": 2.1854818361565275e-05,
|
| 2480 |
+
"loss": 0.0319,
|
| 2481 |
+
"step": 3530
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 4.790257104194858,
|
| 2485 |
+
"grad_norm": 0.19489619135856628,
|
| 2486 |
+
"learning_rate": 2.1582109922052364e-05,
|
| 2487 |
+
"loss": 0.0321,
|
| 2488 |
+
"step": 3540
|
| 2489 |
+
},
|
| 2490 |
+
{
|
| 2491 |
+
"epoch": 4.803788903924222,
|
| 2492 |
+
"grad_norm": 0.16030816733837128,
|
| 2493 |
+
"learning_rate": 2.1310644573112635e-05,
|
| 2494 |
+
"loss": 0.0306,
|
| 2495 |
+
"step": 3550
|
| 2496 |
+
},
|
| 2497 |
+
{
|
| 2498 |
+
"epoch": 4.817320703653586,
|
| 2499 |
+
"grad_norm": 0.16815383732318878,
|
| 2500 |
+
"learning_rate": 2.1040434189519924e-05,
|
| 2501 |
+
"loss": 0.033,
|
| 2502 |
+
"step": 3560
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"epoch": 4.83085250338295,
|
| 2506 |
+
"grad_norm": 0.2008376121520996,
|
| 2507 |
+
"learning_rate": 2.0771490591151733e-05,
|
| 2508 |
+
"loss": 0.0323,
|
| 2509 |
+
"step": 3570
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 4.844384303112314,
|
| 2513 |
+
"grad_norm": 0.12762121856212616,
|
| 2514 |
+
"learning_rate": 2.0503825542472317e-05,
|
| 2515 |
+
"loss": 0.0361,
|
| 2516 |
+
"step": 3580
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 4.857916102841678,
|
| 2520 |
+
"grad_norm": 0.1627681702375412,
|
| 2521 |
+
"learning_rate": 2.023745075201805e-05,
|
| 2522 |
+
"loss": 0.0281,
|
| 2523 |
+
"step": 3590
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 4.871447902571042,
|
| 2527 |
+
"grad_norm": 0.1745183914899826,
|
| 2528 |
+
"learning_rate": 1.9972377871885157e-05,
|
| 2529 |
+
"loss": 0.0302,
|
| 2530 |
+
"step": 3600
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 4.884979702300406,
|
| 2534 |
+
"grad_norm": 0.1831778585910797,
|
| 2535 |
+
"learning_rate": 1.970861849722017e-05,
|
| 2536 |
+
"loss": 0.0297,
|
| 2537 |
+
"step": 3610
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 4.89851150202977,
|
| 2541 |
+
"grad_norm": 0.2168680876493454,
|
| 2542 |
+
"learning_rate": 1.9446184165712587e-05,
|
| 2543 |
+
"loss": 0.0304,
|
| 2544 |
+
"step": 3620
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 4.912043301759134,
|
| 2548 |
+
"grad_norm": 0.1703914850950241,
|
| 2549 |
+
"learning_rate": 1.9185086357090214e-05,
|
| 2550 |
+
"loss": 0.0301,
|
| 2551 |
+
"step": 3630
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 4.925575101488498,
|
| 2555 |
+
"grad_norm": 0.22563602030277252,
|
| 2556 |
+
"learning_rate": 1.8925336492617057e-05,
|
| 2557 |
+
"loss": 0.0294,
|
| 2558 |
+
"step": 3640
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 4.939106901217862,
|
| 2562 |
+
"grad_norm": 0.16377097368240356,
|
| 2563 |
+
"learning_rate": 1.8666945934593666e-05,
|
| 2564 |
+
"loss": 0.0334,
|
| 2565 |
+
"step": 3650
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 4.952638700947226,
|
| 2569 |
+
"grad_norm": 0.11018290370702744,
|
| 2570 |
+
"learning_rate": 1.8409925985860126e-05,
|
| 2571 |
+
"loss": 0.0252,
|
| 2572 |
+
"step": 3660
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 4.96617050067659,
|
| 2576 |
+
"grad_norm": 0.15742255747318268,
|
| 2577 |
+
"learning_rate": 1.8154287889301603e-05,
|
| 2578 |
+
"loss": 0.0295,
|
| 2579 |
+
"step": 3670
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 4.979702300405954,
|
| 2583 |
+
"grad_norm": 0.1602838933467865,
|
| 2584 |
+
"learning_rate": 1.7900042827356612e-05,
|
| 2585 |
+
"loss": 0.0326,
|
| 2586 |
+
"step": 3680
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 4.993234100135318,
|
| 2590 |
+
"grad_norm": 0.2006329745054245,
|
| 2591 |
+
"learning_rate": 1.76472019215278e-05,
|
| 2592 |
+
"loss": 0.0306,
|
| 2593 |
+
"step": 3690
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 5.006765899864682,
|
| 2597 |
+
"grad_norm": 0.2930007874965668,
|
| 2598 |
+
"learning_rate": 1.739577623189545e-05,
|
| 2599 |
+
"loss": 0.0293,
|
| 2600 |
+
"step": 3700
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 5.020297699594046,
|
| 2604 |
+
"grad_norm": 0.3344183564186096,
|
| 2605 |
+
"learning_rate": 1.7145776756633768e-05,
|
| 2606 |
+
"loss": 0.0331,
|
| 2607 |
+
"step": 3710
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 5.03382949932341,
|
| 2611 |
+
"grad_norm": 0.25694382190704346,
|
| 2612 |
+
"learning_rate": 1.6897214431529646e-05,
|
| 2613 |
+
"loss": 0.0325,
|
| 2614 |
+
"step": 3720
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 5.047361299052774,
|
| 2618 |
+
"grad_norm": 0.11163618415594101,
|
| 2619 |
+
"learning_rate": 1.6650100129504475e-05,
|
| 2620 |
+
"loss": 0.0298,
|
| 2621 |
+
"step": 3730
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 5.060893098782138,
|
| 2625 |
+
"grad_norm": 0.3061096668243408,
|
| 2626 |
+
"learning_rate": 1.6404444660138335e-05,
|
| 2627 |
+
"loss": 0.0313,
|
| 2628 |
+
"step": 3740
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 5.074424898511502,
|
| 2632 |
+
"grad_norm": 0.16697120666503906,
|
| 2633 |
+
"learning_rate": 1.616025876919725e-05,
|
| 2634 |
+
"loss": 0.0293,
|
| 2635 |
+
"step": 3750
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 5.087956698240866,
|
| 2639 |
+
"grad_norm": 0.11484449356794357,
|
| 2640 |
+
"learning_rate": 1.5917553138163172e-05,
|
| 2641 |
+
"loss": 0.0284,
|
| 2642 |
+
"step": 3760
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 5.10148849797023,
|
| 2646 |
+
"grad_norm": 0.3616161048412323,
|
| 2647 |
+
"learning_rate": 1.5676338383766632e-05,
|
| 2648 |
+
"loss": 0.0318,
|
| 2649 |
+
"step": 3770
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 5.115020297699594,
|
| 2653 |
+
"grad_norm": 0.18025486171245575,
|
| 2654 |
+
"learning_rate": 1.5436625057522447e-05,
|
| 2655 |
+
"loss": 0.0316,
|
| 2656 |
+
"step": 3780
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"epoch": 5.128552097428958,
|
| 2660 |
+
"grad_norm": 0.14067353308200836,
|
| 2661 |
+
"learning_rate": 1.519842364526804e-05,
|
| 2662 |
+
"loss": 0.0322,
|
| 2663 |
+
"step": 3790
|
| 2664 |
+
},
|
| 2665 |
+
{
|
| 2666 |
+
"epoch": 5.142083897158322,
|
| 2667 |
+
"grad_norm": 0.18656225502490997,
|
| 2668 |
+
"learning_rate": 1.4961744566704855e-05,
|
| 2669 |
+
"loss": 0.0297,
|
| 2670 |
+
"step": 3800
|
| 2671 |
+
},
|
| 2672 |
+
{
|
| 2673 |
+
"epoch": 5.155615696887686,
|
| 2674 |
+
"grad_norm": 0.14053630828857422,
|
| 2675 |
+
"learning_rate": 1.4726598174942551e-05,
|
| 2676 |
+
"loss": 0.0275,
|
| 2677 |
+
"step": 3810
|
| 2678 |
+
},
|
| 2679 |
+
{
|
| 2680 |
+
"epoch": 5.16914749661705,
|
| 2681 |
+
"grad_norm": 0.1318153440952301,
|
| 2682 |
+
"learning_rate": 1.4492994756046035e-05,
|
| 2683 |
+
"loss": 0.0307,
|
| 2684 |
+
"step": 3820
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 5.182679296346414,
|
| 2688 |
+
"grad_norm": 0.21349585056304932,
|
| 2689 |
+
"learning_rate": 1.4260944528585645e-05,
|
| 2690 |
+
"loss": 0.0313,
|
| 2691 |
+
"step": 3830
|
| 2692 |
+
},
|
| 2693 |
+
{
|
| 2694 |
+
"epoch": 5.196211096075778,
|
| 2695 |
+
"grad_norm": 0.143988698720932,
|
| 2696 |
+
"learning_rate": 1.4030457643190048e-05,
|
| 2697 |
+
"loss": 0.0299,
|
| 2698 |
+
"step": 3840
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"epoch": 5.209742895805142,
|
| 2702 |
+
"grad_norm": 0.1565936654806137,
|
| 2703 |
+
"learning_rate": 1.3801544182102311e-05,
|
| 2704 |
+
"loss": 0.0312,
|
| 2705 |
+
"step": 3850
|
| 2706 |
+
},
|
| 2707 |
+
{
|
| 2708 |
+
"epoch": 5.223274695534506,
|
| 2709 |
+
"grad_norm": 0.18127118051052094,
|
| 2710 |
+
"learning_rate": 1.3574214158738763e-05,
|
| 2711 |
+
"loss": 0.0301,
|
| 2712 |
+
"step": 3860
|
| 2713 |
+
},
|
| 2714 |
+
{
|
| 2715 |
+
"epoch": 5.23680649526387,
|
| 2716 |
+
"grad_norm": 0.12334998697042465,
|
| 2717 |
+
"learning_rate": 1.3348477517251101e-05,
|
| 2718 |
+
"loss": 0.0271,
|
| 2719 |
+
"step": 3870
|
| 2720 |
+
},
|
| 2721 |
+
{
|
| 2722 |
+
"epoch": 5.2503382949932345,
|
| 2723 |
+
"grad_norm": 0.16843830049037933,
|
| 2724 |
+
"learning_rate": 1.312434413209131e-05,
|
| 2725 |
+
"loss": 0.0266,
|
| 2726 |
+
"step": 3880
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 5.2638700947225985,
|
| 2730 |
+
"grad_norm": 0.1459604948759079,
|
| 2731 |
+
"learning_rate": 1.2901823807579727e-05,
|
| 2732 |
+
"loss": 0.0281,
|
| 2733 |
+
"step": 3890
|
| 2734 |
+
},
|
| 2735 |
+
{
|
| 2736 |
+
"epoch": 5.2774018944519625,
|
| 2737 |
+
"grad_norm": 0.25701919198036194,
|
| 2738 |
+
"learning_rate": 1.2680926277476245e-05,
|
| 2739 |
+
"loss": 0.0264,
|
| 2740 |
+
"step": 3900
|
| 2741 |
+
},
|
| 2742 |
+
{
|
| 2743 |
+
"epoch": 5.2909336941813265,
|
| 2744 |
+
"grad_norm": 0.1085154339671135,
|
| 2745 |
+
"learning_rate": 1.2461661204554397e-05,
|
| 2746 |
+
"loss": 0.0297,
|
| 2747 |
+
"step": 3910
|
| 2748 |
+
},
|
| 2749 |
+
{
|
| 2750 |
+
"epoch": 5.3044654939106906,
|
| 2751 |
+
"grad_norm": 0.09449544548988342,
|
| 2752 |
+
"learning_rate": 1.2244038180178835e-05,
|
| 2753 |
+
"loss": 0.0277,
|
| 2754 |
+
"step": 3920
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 5.317997293640055,
|
| 2758 |
+
"grad_norm": 0.10733329504728317,
|
| 2759 |
+
"learning_rate": 1.2028066723885612e-05,
|
| 2760 |
+
"loss": 0.0291,
|
| 2761 |
+
"step": 3930
|
| 2762 |
+
},
|
| 2763 |
+
{
|
| 2764 |
+
"epoch": 5.331529093369419,
|
| 2765 |
+
"grad_norm": 0.14141711592674255,
|
| 2766 |
+
"learning_rate": 1.1813756282965888e-05,
|
| 2767 |
+
"loss": 0.0324,
|
| 2768 |
+
"step": 3940
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 5.345060893098782,
|
| 2772 |
+
"grad_norm": 0.1509343385696411,
|
| 2773 |
+
"learning_rate": 1.1601116232052638e-05,
|
| 2774 |
+
"loss": 0.0293,
|
| 2775 |
+
"step": 3950
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 5.358592692828146,
|
| 2779 |
+
"grad_norm": 0.24892765283584595,
|
| 2780 |
+
"learning_rate": 1.1390155872710517e-05,
|
| 2781 |
+
"loss": 0.0287,
|
| 2782 |
+
"step": 3960
|
| 2783 |
+
},
|
| 2784 |
+
{
|
| 2785 |
+
"epoch": 5.37212449255751,
|
| 2786 |
+
"grad_norm": 0.28979232907295227,
|
| 2787 |
+
"learning_rate": 1.1180884433029087e-05,
|
| 2788 |
+
"loss": 0.0263,
|
| 2789 |
+
"step": 3970
|
| 2790 |
+
},
|
| 2791 |
+
{
|
| 2792 |
+
"epoch": 5.385656292286874,
|
| 2793 |
+
"grad_norm": 0.1439746767282486,
|
| 2794 |
+
"learning_rate": 1.097331106721904e-05,
|
| 2795 |
+
"loss": 0.0319,
|
| 2796 |
+
"step": 3980
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"epoch": 5.399188092016238,
|
| 2800 |
+
"grad_norm": 0.12562195956707,
|
| 2801 |
+
"learning_rate": 1.0767444855211862e-05,
|
| 2802 |
+
"loss": 0.0274,
|
| 2803 |
+
"step": 3990
|
| 2804 |
+
},
|
| 2805 |
+
{
|
| 2806 |
+
"epoch": 5.412719891745602,
|
| 2807 |
+
"grad_norm": 0.14074930548667908,
|
| 2808 |
+
"learning_rate": 1.0563294802262558e-05,
|
| 2809 |
+
"loss": 0.0273,
|
| 2810 |
+
"step": 4000
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 5.426251691474966,
|
| 2814 |
+
"grad_norm": 0.10954784601926804,
|
| 2815 |
+
"learning_rate": 1.0360869838555809e-05,
|
| 2816 |
+
"loss": 0.0285,
|
| 2817 |
+
"step": 4010
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 5.43978349120433,
|
| 2821 |
+
"grad_norm": 0.147128164768219,
|
| 2822 |
+
"learning_rate": 1.0160178818815313e-05,
|
| 2823 |
+
"loss": 0.0296,
|
| 2824 |
+
"step": 4020
|
| 2825 |
+
},
|
| 2826 |
+
{
|
| 2827 |
+
"epoch": 5.453315290933694,
|
| 2828 |
+
"grad_norm": 0.10633907467126846,
|
| 2829 |
+
"learning_rate": 9.961230521916387e-06,
|
| 2830 |
+
"loss": 0.031,
|
| 2831 |
+
"step": 4030
|
| 2832 |
+
},
|
| 2833 |
+
{
|
| 2834 |
+
"epoch": 5.466847090663058,
|
| 2835 |
+
"grad_norm": 0.15704360604286194,
|
| 2836 |
+
"learning_rate": 9.764033650502074e-06,
|
| 2837 |
+
"loss": 0.0277,
|
| 2838 |
+
"step": 4040
|
| 2839 |
+
},
|
| 2840 |
+
{
|
| 2841 |
+
"epoch": 5.480378890392422,
|
| 2842 |
+
"grad_norm": 0.10788650810718536,
|
| 2843 |
+
"learning_rate": 9.568596830602344e-06,
|
| 2844 |
+
"loss": 0.028,
|
| 2845 |
+
"step": 4050
|
| 2846 |
+
},
|
| 2847 |
+
{
|
| 2848 |
+
"epoch": 5.493910690121786,
|
| 2849 |
+
"grad_norm": 0.14843221008777618,
|
| 2850 |
+
"learning_rate": 9.37492861125681e-06,
|
| 2851 |
+
"loss": 0.0303,
|
| 2852 |
+
"step": 4060
|
| 2853 |
+
},
|
| 2854 |
+
{
|
| 2855 |
+
"epoch": 5.50744248985115,
|
| 2856 |
+
"grad_norm": 0.1297934353351593,
|
| 2857 |
+
"learning_rate": 9.183037464140804e-06,
|
| 2858 |
+
"loss": 0.0293,
|
| 2859 |
+
"step": 4070
|
| 2860 |
+
},
|
| 2861 |
+
{
|
| 2862 |
+
"epoch": 5.520974289580514,
|
| 2863 |
+
"grad_norm": 0.13509418070316315,
|
| 2864 |
+
"learning_rate": 8.992931783194735e-06,
|
| 2865 |
+
"loss": 0.0294,
|
| 2866 |
+
"step": 4080
|
| 2867 |
+
},
|
| 2868 |
+
{
|
| 2869 |
+
"epoch": 5.534506089309878,
|
| 2870 |
+
"grad_norm": 0.1298404484987259,
|
| 2871 |
+
"learning_rate": 8.80461988425696e-06,
|
| 2872 |
+
"loss": 0.0329,
|
| 2873 |
+
"step": 4090
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 5.548037889039242,
|
| 2877 |
+
"grad_norm": 0.12505508959293365,
|
| 2878 |
+
"learning_rate": 8.618110004699974e-06,
|
| 2879 |
+
"loss": 0.0272,
|
| 2880 |
+
"step": 4100
|
| 2881 |
+
},
|
| 2882 |
+
{
|
| 2883 |
+
"epoch": 5.561569688768606,
|
| 2884 |
+
"grad_norm": 0.09546499699354172,
|
| 2885 |
+
"learning_rate": 8.4334103030701e-06,
|
| 2886 |
+
"loss": 0.0273,
|
| 2887 |
+
"step": 4110
|
| 2888 |
+
},
|
| 2889 |
+
{
|
| 2890 |
+
"epoch": 5.57510148849797,
|
| 2891 |
+
"grad_norm": 0.15529732406139374,
|
| 2892 |
+
"learning_rate": 8.25052885873066e-06,
|
| 2893 |
+
"loss": 0.0293,
|
| 2894 |
+
"step": 4120
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 5.588633288227334,
|
| 2898 |
+
"grad_norm": 0.12811461091041565,
|
| 2899 |
+
"learning_rate": 8.06947367150846e-06,
|
| 2900 |
+
"loss": 0.0274,
|
| 2901 |
+
"step": 4130
|
| 2902 |
+
},
|
| 2903 |
+
{
|
| 2904 |
+
"epoch": 5.602165087956698,
|
| 2905 |
+
"grad_norm": 0.18342335522174835,
|
| 2906 |
+
"learning_rate": 7.890252661343938e-06,
|
| 2907 |
+
"loss": 0.028,
|
| 2908 |
+
"step": 4140
|
| 2909 |
+
},
|
| 2910 |
+
{
|
| 2911 |
+
"epoch": 5.615696887686062,
|
| 2912 |
+
"grad_norm": 0.11130325496196747,
|
| 2913 |
+
"learning_rate": 7.712873667944681e-06,
|
| 2914 |
+
"loss": 0.0286,
|
| 2915 |
+
"step": 4150
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 5.629228687415426,
|
| 2919 |
+
"grad_norm": 0.14142289757728577,
|
| 2920 |
+
"learning_rate": 7.537344450442469e-06,
|
| 2921 |
+
"loss": 0.03,
|
| 2922 |
+
"step": 4160
|
| 2923 |
+
},
|
| 2924 |
+
{
|
| 2925 |
+
"epoch": 5.64276048714479,
|
| 2926 |
+
"grad_norm": 0.19619788229465485,
|
| 2927 |
+
"learning_rate": 7.36367268705393e-06,
|
| 2928 |
+
"loss": 0.0335,
|
| 2929 |
+
"step": 4170
|
| 2930 |
+
},
|
| 2931 |
+
{
|
| 2932 |
+
"epoch": 5.656292286874154,
|
| 2933 |
+
"grad_norm": 0.13693951070308685,
|
| 2934 |
+
"learning_rate": 7.1918659747446e-06,
|
| 2935 |
+
"loss": 0.0287,
|
| 2936 |
+
"step": 4180
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 5.669824086603518,
|
| 2940 |
+
"grad_norm": 0.10885664820671082,
|
| 2941 |
+
"learning_rate": 7.021931828896666e-06,
|
| 2942 |
+
"loss": 0.0327,
|
| 2943 |
+
"step": 4190
|
| 2944 |
+
},
|
| 2945 |
+
{
|
| 2946 |
+
"epoch": 5.683355886332882,
|
| 2947 |
+
"grad_norm": 0.14536768198013306,
|
| 2948 |
+
"learning_rate": 6.8538776829801584e-06,
|
| 2949 |
+
"loss": 0.0278,
|
| 2950 |
+
"step": 4200
|
| 2951 |
+
},
|
| 2952 |
+
{
|
| 2953 |
+
"epoch": 5.696887686062246,
|
| 2954 |
+
"grad_norm": 0.15630270540714264,
|
| 2955 |
+
"learning_rate": 6.687710888227849e-06,
|
| 2956 |
+
"loss": 0.0288,
|
| 2957 |
+
"step": 4210
|
| 2958 |
+
},
|
| 2959 |
+
{
|
| 2960 |
+
"epoch": 5.71041948579161,
|
| 2961 |
+
"grad_norm": 0.1373508870601654,
|
| 2962 |
+
"learning_rate": 6.5234387133136565e-06,
|
| 2963 |
+
"loss": 0.0296,
|
| 2964 |
+
"step": 4220
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 5.723951285520974,
|
| 2968 |
+
"grad_norm": 0.16177499294281006,
|
| 2969 |
+
"learning_rate": 6.361068344034665e-06,
|
| 2970 |
+
"loss": 0.028,
|
| 2971 |
+
"step": 4230
|
| 2972 |
+
},
|
| 2973 |
+
{
|
| 2974 |
+
"epoch": 5.7374830852503385,
|
| 2975 |
+
"grad_norm": 0.11304031312465668,
|
| 2976 |
+
"learning_rate": 6.200606882996846e-06,
|
| 2977 |
+
"loss": 0.0283,
|
| 2978 |
+
"step": 4240
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 5.7510148849797025,
|
| 2982 |
+
"grad_norm": 0.27458086609840393,
|
| 2983 |
+
"learning_rate": 6.042061349304312e-06,
|
| 2984 |
+
"loss": 0.03,
|
| 2985 |
+
"step": 4250
|
| 2986 |
+
},
|
| 2987 |
+
{
|
| 2988 |
+
"epoch": 5.7645466847090665,
|
| 2989 |
+
"grad_norm": 0.24669907987117767,
|
| 2990 |
+
"learning_rate": 5.885438678252342e-06,
|
| 2991 |
+
"loss": 0.0317,
|
| 2992 |
+
"step": 4260
|
| 2993 |
+
},
|
| 2994 |
+
{
|
| 2995 |
+
"epoch": 5.7780784844384305,
|
| 2996 |
+
"grad_norm": 0.14182038605213165,
|
| 2997 |
+
"learning_rate": 5.730745721023939e-06,
|
| 2998 |
+
"loss": 0.0302,
|
| 2999 |
+
"step": 4270
|
| 3000 |
+
},
|
| 3001 |
+
{
|
| 3002 |
+
"epoch": 5.7916102841677946,
|
| 3003 |
+
"grad_norm": 0.1324010193347931,
|
| 3004 |
+
"learning_rate": 5.577989244390192e-06,
|
| 3005 |
+
"loss": 0.0288,
|
| 3006 |
+
"step": 4280
|
| 3007 |
+
},
|
| 3008 |
+
{
|
| 3009 |
+
"epoch": 5.805142083897159,
|
| 3010 |
+
"grad_norm": 0.2429128736257553,
|
| 3011 |
+
"learning_rate": 5.4271759304142635e-06,
|
| 3012 |
+
"loss": 0.0268,
|
| 3013 |
+
"step": 4290
|
| 3014 |
+
},
|
| 3015 |
+
{
|
| 3016 |
+
"epoch": 5.818673883626523,
|
| 3017 |
+
"grad_norm": 0.20134113729000092,
|
| 3018 |
+
"learning_rate": 5.278312376159051e-06,
|
| 3019 |
+
"loss": 0.0282,
|
| 3020 |
+
"step": 4300
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 5.832205683355887,
|
| 3024 |
+
"grad_norm": 0.18878605961799622,
|
| 3025 |
+
"learning_rate": 5.1314050933986944e-06,
|
| 3026 |
+
"loss": 0.0265,
|
| 3027 |
+
"step": 4310
|
| 3028 |
+
},
|
| 3029 |
+
{
|
| 3030 |
+
"epoch": 5.845737483085251,
|
| 3031 |
+
"grad_norm": 0.1928325742483139,
|
| 3032 |
+
"learning_rate": 4.986460508333634e-06,
|
| 3033 |
+
"loss": 0.0262,
|
| 3034 |
+
"step": 4320
|
| 3035 |
+
},
|
| 3036 |
+
{
|
| 3037 |
+
"epoch": 5.859269282814615,
|
| 3038 |
+
"grad_norm": 0.11430050432682037,
|
| 3039 |
+
"learning_rate": 4.843484961309597e-06,
|
| 3040 |
+
"loss": 0.0278,
|
| 3041 |
+
"step": 4330
|
| 3042 |
+
},
|
| 3043 |
+
{
|
| 3044 |
+
"epoch": 5.872801082543979,
|
| 3045 |
+
"grad_norm": 0.1418386846780777,
|
| 3046 |
+
"learning_rate": 4.702484706540161e-06,
|
| 3047 |
+
"loss": 0.0249,
|
| 3048 |
+
"step": 4340
|
| 3049 |
+
},
|
| 3050 |
+
{
|
| 3051 |
+
"epoch": 5.886332882273343,
|
| 3052 |
+
"grad_norm": 0.1781853586435318,
|
| 3053 |
+
"learning_rate": 4.563465911833259e-06,
|
| 3054 |
+
"loss": 0.0289,
|
| 3055 |
+
"step": 4350
|
| 3056 |
+
},
|
| 3057 |
+
{
|
| 3058 |
+
"epoch": 5.899864682002706,
|
| 3059 |
+
"grad_norm": 0.16866528987884521,
|
| 3060 |
+
"learning_rate": 4.426434658321344e-06,
|
| 3061 |
+
"loss": 0.0248,
|
| 3062 |
+
"step": 4360
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 5.91339648173207,
|
| 3066 |
+
"grad_norm": 0.25363948941230774,
|
| 3067 |
+
"learning_rate": 4.2913969401953466e-06,
|
| 3068 |
+
"loss": 0.0272,
|
| 3069 |
+
"step": 4370
|
| 3070 |
+
},
|
| 3071 |
+
{
|
| 3072 |
+
"epoch": 5.926928281461434,
|
| 3073 |
+
"grad_norm": 0.1106293648481369,
|
| 3074 |
+
"learning_rate": 4.15835866444253e-06,
|
| 3075 |
+
"loss": 0.0252,
|
| 3076 |
+
"step": 4380
|
| 3077 |
+
},
|
| 3078 |
+
{
|
| 3079 |
+
"epoch": 5.940460081190798,
|
| 3080 |
+
"grad_norm": 0.1236589327454567,
|
| 3081 |
+
"learning_rate": 4.027325650588043e-06,
|
| 3082 |
+
"loss": 0.0281,
|
| 3083 |
+
"step": 4390
|
| 3084 |
+
},
|
| 3085 |
+
{
|
| 3086 |
+
"epoch": 5.953991880920162,
|
| 3087 |
+
"grad_norm": 0.14823655784130096,
|
| 3088 |
+
"learning_rate": 3.898303630440419e-06,
|
| 3089 |
+
"loss": 0.0306,
|
| 3090 |
+
"step": 4400
|
| 3091 |
+
},
|
| 3092 |
+
{
|
| 3093 |
+
"epoch": 5.967523680649526,
|
| 3094 |
+
"grad_norm": 0.16370555758476257,
|
| 3095 |
+
"learning_rate": 3.7712982478407877e-06,
|
| 3096 |
+
"loss": 0.0294,
|
| 3097 |
+
"step": 4410
|
| 3098 |
+
},
|
| 3099 |
+
{
|
| 3100 |
+
"epoch": 5.98105548037889,
|
| 3101 |
+
"grad_norm": 0.2005997598171234,
|
| 3102 |
+
"learning_rate": 3.6463150584160053e-06,
|
| 3103 |
+
"loss": 0.0275,
|
| 3104 |
+
"step": 4420
|
| 3105 |
+
},
|
| 3106 |
+
{
|
| 3107 |
+
"epoch": 5.994587280108254,
|
| 3108 |
+
"grad_norm": 0.2858164310455322,
|
| 3109 |
+
"learning_rate": 3.5233595293356957e-06,
|
| 3110 |
+
"loss": 0.029,
|
| 3111 |
+
"step": 4430
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 6.008119079837618,
|
| 3115 |
+
"grad_norm": 0.16128398478031158,
|
| 3116 |
+
"learning_rate": 3.4024370390730033e-06,
|
| 3117 |
+
"loss": 0.0316,
|
| 3118 |
+
"step": 4440
|
| 3119 |
+
},
|
| 3120 |
+
{
|
| 3121 |
+
"epoch": 6.021650879566982,
|
| 3122 |
+
"grad_norm": 0.12964995205402374,
|
| 3123 |
+
"learning_rate": 3.2835528771693992e-06,
|
| 3124 |
+
"loss": 0.0261,
|
| 3125 |
+
"step": 4450
|
| 3126 |
+
},
|
| 3127 |
+
{
|
| 3128 |
+
"epoch": 6.035182679296346,
|
| 3129 |
+
"grad_norm": 0.16563497483730316,
|
| 3130 |
+
"learning_rate": 3.1667122440032505e-06,
|
| 3131 |
+
"loss": 0.0263,
|
| 3132 |
+
"step": 4460
|
| 3133 |
+
},
|
| 3134 |
+
{
|
| 3135 |
+
"epoch": 6.04871447902571,
|
| 3136 |
+
"grad_norm": 0.1846737265586853,
|
| 3137 |
+
"learning_rate": 3.051920250562351e-06,
|
| 3138 |
+
"loss": 0.0286,
|
| 3139 |
+
"step": 4470
|
| 3140 |
+
},
|
| 3141 |
+
{
|
| 3142 |
+
"epoch": 6.062246278755074,
|
| 3143 |
+
"grad_norm": 0.23904356360435486,
|
| 3144 |
+
"learning_rate": 2.939181918220385e-06,
|
| 3145 |
+
"loss": 0.0284,
|
| 3146 |
+
"step": 4480
|
| 3147 |
+
},
|
| 3148 |
+
{
|
| 3149 |
+
"epoch": 6.075778078484438,
|
| 3150 |
+
"grad_norm": 0.22299540042877197,
|
| 3151 |
+
"learning_rate": 2.8285021785172226e-06,
|
| 3152 |
+
"loss": 0.0268,
|
| 3153 |
+
"step": 4490
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 6.089309878213802,
|
| 3157 |
+
"grad_norm": 0.36245062947273254,
|
| 3158 |
+
"learning_rate": 2.7198858729432288e-06,
|
| 3159 |
+
"loss": 0.0302,
|
| 3160 |
+
"step": 4500
|
| 3161 |
+
},
|
| 3162 |
+
{
|
| 3163 |
+
"epoch": 6.102841677943166,
|
| 3164 |
+
"grad_norm": 0.1929812878370285,
|
| 3165 |
+
"learning_rate": 2.6133377527274905e-06,
|
| 3166 |
+
"loss": 0.029,
|
| 3167 |
+
"step": 4510
|
| 3168 |
+
},
|
| 3169 |
+
{
|
| 3170 |
+
"epoch": 6.11637347767253,
|
| 3171 |
+
"grad_norm": 0.1011333093047142,
|
| 3172 |
+
"learning_rate": 2.5088624786299366e-06,
|
| 3173 |
+
"loss": 0.0252,
|
| 3174 |
+
"step": 4520
|
| 3175 |
+
},
|
| 3176 |
+
{
|
| 3177 |
+
"epoch": 6.129905277401894,
|
| 3178 |
+
"grad_norm": 0.19274236261844635,
|
| 3179 |
+
"learning_rate": 2.406464620737531e-06,
|
| 3180 |
+
"loss": 0.0271,
|
| 3181 |
+
"step": 4530
|
| 3182 |
+
},
|
| 3183 |
+
{
|
| 3184 |
+
"epoch": 6.143437077131258,
|
| 3185 |
+
"grad_norm": 0.11799866706132889,
|
| 3186 |
+
"learning_rate": 2.3061486582642734e-06,
|
| 3187 |
+
"loss": 0.0241,
|
| 3188 |
+
"step": 4540
|
| 3189 |
+
},
|
| 3190 |
+
{
|
| 3191 |
+
"epoch": 6.156968876860622,
|
| 3192 |
+
"grad_norm": 0.09267459064722061,
|
| 3193 |
+
"learning_rate": 2.2079189793553667e-06,
|
| 3194 |
+
"loss": 0.0248,
|
| 3195 |
+
"step": 4550
|
| 3196 |
+
},
|
| 3197 |
+
{
|
| 3198 |
+
"epoch": 6.170500676589986,
|
| 3199 |
+
"grad_norm": 0.15937362611293793,
|
| 3200 |
+
"learning_rate": 2.111779880895165e-06,
|
| 3201 |
+
"loss": 0.0284,
|
| 3202 |
+
"step": 4560
|
| 3203 |
+
},
|
| 3204 |
+
{
|
| 3205 |
+
"epoch": 6.18403247631935,
|
| 3206 |
+
"grad_norm": 0.12410353124141693,
|
| 3207 |
+
"learning_rate": 2.01773556831929e-06,
|
| 3208 |
+
"loss": 0.0292,
|
| 3209 |
+
"step": 4570
|
| 3210 |
+
},
|
| 3211 |
+
{
|
| 3212 |
+
"epoch": 6.197564276048714,
|
| 3213 |
+
"grad_norm": 0.20976869761943817,
|
| 3214 |
+
"learning_rate": 1.9257901554306513e-06,
|
| 3215 |
+
"loss": 0.0278,
|
| 3216 |
+
"step": 4580
|
| 3217 |
+
},
|
| 3218 |
+
{
|
| 3219 |
+
"epoch": 6.211096075778078,
|
| 3220 |
+
"grad_norm": 0.13598106801509857,
|
| 3221 |
+
"learning_rate": 1.835947664219445e-06,
|
| 3222 |
+
"loss": 0.0265,
|
| 3223 |
+
"step": 4590
|
| 3224 |
+
},
|
| 3225 |
+
{
|
| 3226 |
+
"epoch": 6.2246278755074425,
|
| 3227 |
+
"grad_norm": 0.10610091686248779,
|
| 3228 |
+
"learning_rate": 1.748212024687307e-06,
|
| 3229 |
+
"loss": 0.027,
|
| 3230 |
+
"step": 4600
|
| 3231 |
+
},
|
| 3232 |
+
{
|
| 3233 |
+
"epoch": 6.2381596752368065,
|
| 3234 |
+
"grad_norm": 0.14155301451683044,
|
| 3235 |
+
"learning_rate": 1.6625870746753147e-06,
|
| 3236 |
+
"loss": 0.026,
|
| 3237 |
+
"step": 4610
|
| 3238 |
+
},
|
| 3239 |
+
{
|
| 3240 |
+
"epoch": 6.2516914749661705,
|
| 3241 |
+
"grad_norm": 0.12627993524074554,
|
| 3242 |
+
"learning_rate": 1.5790765596961853e-06,
|
| 3243 |
+
"loss": 0.0276,
|
| 3244 |
+
"step": 4620
|
| 3245 |
+
},
|
| 3246 |
+
{
|
| 3247 |
+
"epoch": 6.2652232746955345,
|
| 3248 |
+
"grad_norm": 0.2003839612007141,
|
| 3249 |
+
"learning_rate": 1.4976841327703717e-06,
|
| 3250 |
+
"loss": 0.0272,
|
| 3251 |
+
"step": 4630
|
| 3252 |
+
},
|
| 3253 |
+
{
|
| 3254 |
+
"epoch": 6.2787550744248986,
|
| 3255 |
+
"grad_norm": 0.18389511108398438,
|
| 3256 |
+
"learning_rate": 1.4184133542663014e-06,
|
| 3257 |
+
"loss": 0.0303,
|
| 3258 |
+
"step": 4640
|
| 3259 |
+
},
|
| 3260 |
+
{
|
| 3261 |
+
"epoch": 6.292286874154263,
|
| 3262 |
+
"grad_norm": 0.1884324848651886,
|
| 3263 |
+
"learning_rate": 1.341267691744641e-06,
|
| 3264 |
+
"loss": 0.0308,
|
| 3265 |
+
"step": 4650
|
| 3266 |
+
},
|
| 3267 |
+
{
|
| 3268 |
+
"epoch": 6.305818673883627,
|
| 3269 |
+
"grad_norm": 0.13071194291114807,
|
| 3270 |
+
"learning_rate": 1.2662505198065666e-06,
|
| 3271 |
+
"loss": 0.0316,
|
| 3272 |
+
"step": 4660
|
| 3273 |
+
},
|
| 3274 |
+
{
|
| 3275 |
+
"epoch": 6.319350473612991,
|
| 3276 |
+
"grad_norm": 0.09089189767837524,
|
| 3277 |
+
"learning_rate": 1.193365119946216e-06,
|
| 3278 |
+
"loss": 0.0279,
|
| 3279 |
+
"step": 4670
|
| 3280 |
+
},
|
| 3281 |
+
{
|
| 3282 |
+
"epoch": 6.332882273342355,
|
| 3283 |
+
"grad_norm": 0.1488385945558548,
|
| 3284 |
+
"learning_rate": 1.1226146804070859e-06,
|
| 3285 |
+
"loss": 0.027,
|
| 3286 |
+
"step": 4680
|
| 3287 |
+
},
|
| 3288 |
+
{
|
| 3289 |
+
"epoch": 6.346414073071719,
|
| 3290 |
+
"grad_norm": 0.12004260718822479,
|
| 3291 |
+
"learning_rate": 1.0540022960426111e-06,
|
| 3292 |
+
"loss": 0.0272,
|
| 3293 |
+
"step": 4690
|
| 3294 |
+
},
|
| 3295 |
+
{
|
| 3296 |
+
"epoch": 6.359945872801083,
|
| 3297 |
+
"grad_norm": 0.08131157606840134,
|
| 3298 |
+
"learning_rate": 9.875309681807443e-07,
|
| 3299 |
+
"loss": 0.0288,
|
| 3300 |
+
"step": 4700
|
| 3301 |
+
},
|
| 3302 |
+
{
|
| 3303 |
+
"epoch": 6.373477672530447,
|
| 3304 |
+
"grad_norm": 0.13074468076229095,
|
| 3305 |
+
"learning_rate": 9.232036044927061e-07,
|
| 3306 |
+
"loss": 0.0282,
|
| 3307 |
+
"step": 4710
|
| 3308 |
+
},
|
| 3309 |
+
{
|
| 3310 |
+
"epoch": 6.387009472259811,
|
| 3311 |
+
"grad_norm": 0.2715424597263336,
|
| 3312 |
+
"learning_rate": 8.610230188657919e-07,
|
| 3313 |
+
"loss": 0.0268,
|
| 3314 |
+
"step": 4720
|
| 3315 |
+
},
|
| 3316 |
+
{
|
| 3317 |
+
"epoch": 6.400541271989175,
|
| 3318 |
+
"grad_norm": 0.11489482969045639,
|
| 3319 |
+
"learning_rate": 8.009919312802372e-07,
|
| 3320 |
+
"loss": 0.0288,
|
| 3321 |
+
"step": 4730
|
| 3322 |
+
},
|
| 3323 |
+
{
|
| 3324 |
+
"epoch": 6.414073071718539,
|
| 3325 |
+
"grad_norm": 0.30984801054000854,
|
| 3326 |
+
"learning_rate": 7.431129676902904e-07,
|
| 3327 |
+
"loss": 0.0312,
|
| 3328 |
+
"step": 4740
|
| 3329 |
+
},
|
| 3330 |
+
{
|
| 3331 |
+
"epoch": 6.427604871447903,
|
| 3332 |
+
"grad_norm": 0.26709073781967163,
|
| 3333 |
+
"learning_rate": 6.873886599093215e-07,
|
| 3334 |
+
"loss": 0.026,
|
| 3335 |
+
"step": 4750
|
| 3336 |
+
},
|
| 3337 |
+
{
|
| 3338 |
+
"epoch": 6.441136671177267,
|
| 3339 |
+
"grad_norm": 0.16023863852024078,
|
| 3340 |
+
"learning_rate": 6.338214454990776e-07,
|
| 3341 |
+
"loss": 0.0289,
|
| 3342 |
+
"step": 4760
|
| 3343 |
+
},
|
| 3344 |
+
{
|
| 3345 |
+
"epoch": 6.454668470906631,
|
| 3346 |
+
"grad_norm": 0.09971076250076294,
|
| 3347 |
+
"learning_rate": 5.82413667663051e-07,
|
| 3348 |
+
"loss": 0.028,
|
| 3349 |
+
"step": 4770
|
| 3350 |
+
},
|
| 3351 |
+
{
|
| 3352 |
+
"epoch": 6.468200270635995,
|
| 3353 |
+
"grad_norm": 0.10005897283554077,
|
| 3354 |
+
"learning_rate": 5.331675751439725e-07,
|
| 3355 |
+
"loss": 0.0274,
|
| 3356 |
+
"step": 4780
|
| 3357 |
+
},
|
| 3358 |
+
{
|
| 3359 |
+
"epoch": 6.481732070365359,
|
| 3360 |
+
"grad_norm": 0.132291778922081,
|
| 3361 |
+
"learning_rate": 4.86085322125479e-07,
|
| 3362 |
+
"loss": 0.0257,
|
| 3363 |
+
"step": 4790
|
| 3364 |
+
},
|
| 3365 |
+
{
|
| 3366 |
+
"epoch": 6.495263870094723,
|
| 3367 |
+
"grad_norm": 0.11425412446260452,
|
| 3368 |
+
"learning_rate": 4.411689681378284e-07,
|
| 3369 |
+
"loss": 0.0296,
|
| 3370 |
+
"step": 4800
|
| 3371 |
+
},
|
| 3372 |
+
{
|
| 3373 |
+
"epoch": 6.508795669824087,
|
| 3374 |
+
"grad_norm": 0.10741697996854782,
|
| 3375 |
+
"learning_rate": 3.9842047796786466e-07,
|
| 3376 |
+
"loss": 0.0263,
|
| 3377 |
+
"step": 4810
|
| 3378 |
+
},
|
| 3379 |
+
{
|
| 3380 |
+
"epoch": 6.522327469553451,
|
| 3381 |
+
"grad_norm": 0.13160759210586548,
|
| 3382 |
+
"learning_rate": 3.578417215730323e-07,
|
| 3383 |
+
"loss": 0.0278,
|
| 3384 |
+
"step": 4820
|
| 3385 |
+
},
|
| 3386 |
+
{
|
| 3387 |
+
"epoch": 6.535859269282815,
|
| 3388 |
+
"grad_norm": 0.14389033615589142,
|
| 3389 |
+
"learning_rate": 3.1943447399958027e-07,
|
| 3390 |
+
"loss": 0.0252,
|
| 3391 |
+
"step": 4830
|
| 3392 |
+
},
|
| 3393 |
+
{
|
| 3394 |
+
"epoch": 6.549391069012179,
|
| 3395 |
+
"grad_norm": 0.4230325222015381,
|
| 3396 |
+
"learning_rate": 2.8320041530495724e-07,
|
| 3397 |
+
"loss": 0.0356,
|
| 3398 |
+
"step": 4840
|
| 3399 |
+
},
|
| 3400 |
+
{
|
| 3401 |
+
"epoch": 6.562922868741543,
|
| 3402 |
+
"grad_norm": 0.5279276371002197,
|
| 3403 |
+
"learning_rate": 2.491411304842539e-07,
|
| 3404 |
+
"loss": 0.0289,
|
| 3405 |
+
"step": 4850
|
| 3406 |
+
},
|
| 3407 |
+
{
|
| 3408 |
+
"epoch": 6.576454668470907,
|
| 3409 |
+
"grad_norm": 0.12740273773670197,
|
| 3410 |
+
"learning_rate": 2.1725810940094183e-07,
|
| 3411 |
+
"loss": 0.0278,
|
| 3412 |
+
"step": 4860
|
| 3413 |
+
},
|
| 3414 |
+
{
|
| 3415 |
+
"epoch": 6.589986468200271,
|
| 3416 |
+
"grad_norm": 0.1272871196269989,
|
| 3417 |
+
"learning_rate": 1.8755274672164202e-07,
|
| 3418 |
+
"loss": 0.0256,
|
| 3419 |
+
"step": 4870
|
| 3420 |
+
},
|
| 3421 |
+
{
|
| 3422 |
+
"epoch": 6.603518267929634,
|
| 3423 |
+
"grad_norm": 0.1781158745288849,
|
| 3424 |
+
"learning_rate": 1.600263418551573e-07,
|
| 3425 |
+
"loss": 0.0263,
|
| 3426 |
+
"step": 4880
|
| 3427 |
+
},
|
| 3428 |
+
{
|
| 3429 |
+
"epoch": 6.617050067658998,
|
| 3430 |
+
"grad_norm": 0.0960567444562912,
|
| 3431 |
+
"learning_rate": 1.346800988955954e-07,
|
| 3432 |
+
"loss": 0.0268,
|
| 3433 |
+
"step": 4890
|
| 3434 |
+
},
|
| 3435 |
+
{
|
| 3436 |
+
"epoch": 6.630581867388362,
|
| 3437 |
+
"grad_norm": 0.22953420877456665,
|
| 3438 |
+
"learning_rate": 1.1151512656975005e-07,
|
| 3439 |
+
"loss": 0.027,
|
| 3440 |
+
"step": 4900
|
| 3441 |
+
},
|
| 3442 |
+
{
|
| 3443 |
+
"epoch": 6.644113667117726,
|
| 3444 |
+
"grad_norm": 0.22247491776943207,
|
| 3445 |
+
"learning_rate": 9.053243818853973e-08,
|
| 3446 |
+
"loss": 0.0291,
|
| 3447 |
+
"step": 4910
|
| 3448 |
+
},
|
| 3449 |
+
{
|
| 3450 |
+
"epoch": 6.65764546684709,
|
| 3451 |
+
"grad_norm": 0.13925011456012726,
|
| 3452 |
+
"learning_rate": 7.173295160273763e-08,
|
| 3453 |
+
"loss": 0.0256,
|
| 3454 |
+
"step": 4920
|
| 3455 |
+
},
|
| 3456 |
+
{
|
| 3457 |
+
"epoch": 6.671177266576454,
|
| 3458 |
+
"grad_norm": 0.4142376780509949,
|
| 3459 |
+
"learning_rate": 5.511748916279258e-08,
|
| 3460 |
+
"loss": 0.0275,
|
| 3461 |
+
"step": 4930
|
| 3462 |
+
},
|
| 3463 |
+
{
|
| 3464 |
+
"epoch": 6.684709066305818,
|
| 3465 |
+
"grad_norm": 0.2592661678791046,
|
| 3466 |
+
"learning_rate": 4.068677768285234e-08,
|
| 3467 |
+
"loss": 0.0256,
|
| 3468 |
+
"step": 4940
|
| 3469 |
+
},
|
| 3470 |
+
{
|
| 3471 |
+
"epoch": 6.698240866035182,
|
| 3472 |
+
"grad_norm": 0.1092233955860138,
|
| 3473 |
+
"learning_rate": 2.844144840898344e-08,
|
| 3474 |
+
"loss": 0.0253,
|
| 3475 |
+
"step": 4950
|
| 3476 |
+
},
|
| 3477 |
+
{
|
| 3478 |
+
"epoch": 6.7117726657645465,
|
| 3479 |
+
"grad_norm": 0.09404899179935455,
|
| 3480 |
+
"learning_rate": 1.8382036991559936e-08,
|
| 3481 |
+
"loss": 0.0255,
|
| 3482 |
+
"step": 4960
|
| 3483 |
+
},
|
| 3484 |
+
{
|
| 3485 |
+
"epoch": 6.7253044654939105,
|
| 3486 |
+
"grad_norm": 0.18313178420066833,
|
| 3487 |
+
"learning_rate": 1.0508983461832156e-08,
|
| 3488 |
+
"loss": 0.0286,
|
| 3489 |
+
"step": 4970
|
| 3490 |
+
},
|
| 3491 |
+
{
|
| 3492 |
+
"epoch": 6.7388362652232745,
|
| 3493 |
+
"grad_norm": 0.0930226519703865,
|
| 3494 |
+
"learning_rate": 4.822632212653222e-09,
|
| 3495 |
+
"loss": 0.0284,
|
| 3496 |
+
"step": 4980
|
| 3497 |
+
},
|
| 3498 |
+
{
|
| 3499 |
+
"epoch": 6.7523680649526385,
|
| 3500 |
+
"grad_norm": 0.12706884741783142,
|
| 3501 |
+
"learning_rate": 1.3232319834632912e-09,
|
| 3502 |
+
"loss": 0.0255,
|
| 3503 |
+
"step": 4990
|
| 3504 |
+
},
|
| 3505 |
+
{
|
| 3506 |
+
"epoch": 6.7658998646820026,
|
| 3507 |
+
"grad_norm": 0.264944463968277,
|
| 3508 |
+
"learning_rate": 1.0935849353854721e-11,
|
| 3509 |
+
"loss": 0.0264,
|
| 3510 |
+
"step": 5000
|
| 3511 |
+
}
|
| 3512 |
+
],
|
| 3513 |
+
"logging_steps": 10,
|
| 3514 |
+
"max_steps": 5000,
|
| 3515 |
+
"num_input_tokens_seen": 0,
|
| 3516 |
+
"num_train_epochs": 7,
|
| 3517 |
+
"save_steps": 1000,
|
| 3518 |
+
"stateful_callbacks": {
|
| 3519 |
+
"TrainerControl": {
|
| 3520 |
+
"args": {
|
| 3521 |
+
"should_epoch_stop": false,
|
| 3522 |
+
"should_evaluate": false,
|
| 3523 |
+
"should_log": false,
|
| 3524 |
+
"should_save": true,
|
| 3525 |
+
"should_training_stop": true
|
| 3526 |
+
},
|
| 3527 |
+
"attributes": {}
|
| 3528 |
+
}
|
| 3529 |
+
},
|
| 3530 |
+
"total_flos": 0.0,
|
| 3531 |
+
"train_batch_size": 32,
|
| 3532 |
+
"trial_name": null,
|
| 3533 |
+
"trial_params": null
|
| 3534 |
+
}
|
drawer_20251223_5k/config.json
ADDED
|
@@ -0,0 +1,64 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"action_dim": 32,
|
| 3 |
+
"action_head_cfg": {
|
| 4 |
+
"action_dim": 32,
|
| 5 |
+
"action_horizon": 16,
|
| 6 |
+
"add_pos_embed": true,
|
| 7 |
+
"backbone_embedding_dim": 2048,
|
| 8 |
+
"diffusion_model_cfg": {
|
| 9 |
+
"attention_head_dim": 48,
|
| 10 |
+
"cross_attention_dim": 2048,
|
| 11 |
+
"dropout": 0.2,
|
| 12 |
+
"final_dropout": true,
|
| 13 |
+
"interleave_self_attention": true,
|
| 14 |
+
"norm_type": "ada_norm",
|
| 15 |
+
"num_attention_heads": 32,
|
| 16 |
+
"num_layers": 16,
|
| 17 |
+
"output_dim": 1024,
|
| 18 |
+
"positional_embeddings": null
|
| 19 |
+
},
|
| 20 |
+
"hidden_size": 1024,
|
| 21 |
+
"input_embedding_dim": 1536,
|
| 22 |
+
"max_action_dim": 32,
|
| 23 |
+
"max_state_dim": 64,
|
| 24 |
+
"model_dtype": "float32",
|
| 25 |
+
"noise_beta_alpha": 1.5,
|
| 26 |
+
"noise_beta_beta": 1.0,
|
| 27 |
+
"noise_s": 0.999,
|
| 28 |
+
"num_inference_timesteps": 4,
|
| 29 |
+
"num_target_vision_tokens": 32,
|
| 30 |
+
"num_timestep_buckets": 1000,
|
| 31 |
+
"tune_diffusion_model": true,
|
| 32 |
+
"tune_projector": true,
|
| 33 |
+
"use_vlln": true,
|
| 34 |
+
"vl_self_attention_cfg": {
|
| 35 |
+
"attention_head_dim": 64,
|
| 36 |
+
"dropout": 0.2,
|
| 37 |
+
"final_dropout": true,
|
| 38 |
+
"num_attention_heads": 32,
|
| 39 |
+
"num_layers": 4,
|
| 40 |
+
"positional_embeddings": null
|
| 41 |
+
}
|
| 42 |
+
},
|
| 43 |
+
"action_horizon": 16,
|
| 44 |
+
"architectures": [
|
| 45 |
+
"GR00T_N1_5"
|
| 46 |
+
],
|
| 47 |
+
"attn_implementation": null,
|
| 48 |
+
"backbone_cfg": {
|
| 49 |
+
"eagle_path": "NVEagle/eagle_er-qwen3_1_7B-Siglip2_400M_stage1_5_128gpu_er_v7_1mlp_nops",
|
| 50 |
+
"load_bf16": false,
|
| 51 |
+
"project_to_dim": null,
|
| 52 |
+
"reproject_vision": false,
|
| 53 |
+
"select_layer": 12,
|
| 54 |
+
"tune_llm": false,
|
| 55 |
+
"tune_visual": true,
|
| 56 |
+
"use_flash_attention": true
|
| 57 |
+
},
|
| 58 |
+
"compute_dtype": "bfloat16",
|
| 59 |
+
"hidden_size": 2048,
|
| 60 |
+
"model_dtype": "float32",
|
| 61 |
+
"model_type": "gr00t_n1_5",
|
| 62 |
+
"torch_dtype": "bfloat16",
|
| 63 |
+
"transformers_version": "4.51.3"
|
| 64 |
+
}
|
drawer_20251223_5k/experiment_cfg/metadata.json
ADDED
|
@@ -0,0 +1,807 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"new_embodiment": {
|
| 3 |
+
"statistics": {
|
| 4 |
+
"state": {
|
| 5 |
+
"base": {
|
| 6 |
+
"max": [
|
| 7 |
+
0.06309662759304047,
|
| 8 |
+
0.17190714180469513,
|
| 9 |
+
0.14714741706848145
|
| 10 |
+
],
|
| 11 |
+
"min": [
|
| 12 |
+
-0.08241941034793854,
|
| 13 |
+
-0.04592813551425934,
|
| 14 |
+
-0.0676141232252121
|
| 15 |
+
],
|
| 16 |
+
"mean": [
|
| 17 |
+
0.0005052018095739186,
|
| 18 |
+
3.843669765046798e-05,
|
| 19 |
+
0.0004570714954752475
|
| 20 |
+
],
|
| 21 |
+
"std": [
|
| 22 |
+
0.006058344151824713,
|
| 23 |
+
0.008796663954854012,
|
| 24 |
+
0.008676460944116116
|
| 25 |
+
],
|
| 26 |
+
"q01": [
|
| 27 |
+
-0.017743217926472425,
|
| 28 |
+
-0.021156907770782708,
|
| 29 |
+
-0.019517241157591344
|
| 30 |
+
],
|
| 31 |
+
"q99": [
|
| 32 |
+
0.024395816065371036,
|
| 33 |
+
0.03118515063077203,
|
| 34 |
+
0.0335056135058403
|
| 35 |
+
]
|
| 36 |
+
},
|
| 37 |
+
"torso": {
|
| 38 |
+
"max": [
|
| 39 |
+
0.21035662293434143
|
| 40 |
+
],
|
| 41 |
+
"min": [
|
| 42 |
+
0.0007334263646043837
|
| 43 |
+
],
|
| 44 |
+
"mean": [
|
| 45 |
+
0.1093309074640274
|
| 46 |
+
],
|
| 47 |
+
"std": [
|
| 48 |
+
0.10130645334720612
|
| 49 |
+
],
|
| 50 |
+
"q01": [
|
| 51 |
+
0.0009143962111556902
|
| 52 |
+
],
|
| 53 |
+
"q99": [
|
| 54 |
+
0.21021610468626023
|
| 55 |
+
]
|
| 56 |
+
},
|
| 57 |
+
"left_arm": {
|
| 58 |
+
"max": [
|
| 59 |
+
1.3228962421417236,
|
| 60 |
+
0.9760400056838989,
|
| 61 |
+
1.048154592514038,
|
| 62 |
+
-0.45797696709632874,
|
| 63 |
+
0.3519979417324066,
|
| 64 |
+
0.01325496006757021,
|
| 65 |
+
1.5758417844772339
|
| 66 |
+
],
|
| 67 |
+
"min": [
|
| 68 |
+
-0.5555572509765625,
|
| 69 |
+
-0.002730769105255604,
|
| 70 |
+
-0.7607836723327637,
|
| 71 |
+
-2.2513277530670166,
|
| 72 |
+
-0.3550925850868225,
|
| 73 |
+
-0.3537217676639557,
|
| 74 |
+
-0.8624185919761658
|
| 75 |
+
],
|
| 76 |
+
"mean": [
|
| 77 |
+
0.1697605848312378,
|
| 78 |
+
0.11266966909170151,
|
| 79 |
+
0.27108195424079895,
|
| 80 |
+
-1.643442988395691,
|
| 81 |
+
-0.13571397960186005,
|
| 82 |
+
-0.09654814004898071,
|
| 83 |
+
0.31360459327697754
|
| 84 |
+
],
|
| 85 |
+
"std": [
|
| 86 |
+
0.1941879391670227,
|
| 87 |
+
0.21234647929668427,
|
| 88 |
+
0.18590451776981354,
|
| 89 |
+
0.2731393277645111,
|
| 90 |
+
0.0512852668762207,
|
| 91 |
+
0.05550851300358772,
|
| 92 |
+
0.44211140275001526
|
| 93 |
+
],
|
| 94 |
+
"q01": [
|
| 95 |
+
-0.4777324840426445,
|
| 96 |
+
-0.000706656820839271,
|
| 97 |
+
-0.1646677777171135,
|
| 98 |
+
-1.987500010728836,
|
| 99 |
+
-0.35049458891153334,
|
| 100 |
+
-0.34794992953538895,
|
| 101 |
+
-0.6912282472848892
|
| 102 |
+
],
|
| 103 |
+
"q99": [
|
| 104 |
+
0.8433255749940849,
|
| 105 |
+
0.7800812005996703,
|
| 106 |
+
0.934328988790512,
|
| 107 |
+
-0.6880783206224441,
|
| 108 |
+
0.05031383018940674,
|
| 109 |
+
-0.0052628690563142405,
|
| 110 |
+
1.570149998664856
|
| 111 |
+
]
|
| 112 |
+
},
|
| 113 |
+
"right_arm": {
|
| 114 |
+
"max": [
|
| 115 |
+
1.3470864295959473,
|
| 116 |
+
0.003337543224915862,
|
| 117 |
+
0.8092557787895203,
|
| 118 |
+
-0.37025120854377747,
|
| 119 |
+
0.3554441034793854,
|
| 120 |
+
0.07210186868906021,
|
| 121 |
+
0.7510986328125
|
| 122 |
+
],
|
| 123 |
+
"min": [
|
| 124 |
+
-0.6000412106513977,
|
| 125 |
+
-0.9089406132698059,
|
| 126 |
+
-1.0145564079284668,
|
| 127 |
+
-2.252852201461792,
|
| 128 |
+
-0.3539217412471771,
|
| 129 |
+
-0.3530242145061493,
|
| 130 |
+
-1.5733824968338013
|
| 131 |
+
],
|
| 132 |
+
"mean": [
|
| 133 |
+
0.1856241077184677,
|
| 134 |
+
-0.12809492647647858,
|
| 135 |
+
-0.24569183588027954,
|
| 136 |
+
-1.6810379028320312,
|
| 137 |
+
0.17203257977962494,
|
| 138 |
+
-0.05976219102740288,
|
| 139 |
+
-0.45409324765205383
|
| 140 |
+
],
|
| 141 |
+
"std": [
|
| 142 |
+
0.2284369170665741,
|
| 143 |
+
0.2075410932302475,
|
| 144 |
+
0.28148508071899414,
|
| 145 |
+
0.29220107197761536,
|
| 146 |
+
0.0935073271393776,
|
| 147 |
+
0.06213328242301941,
|
| 148 |
+
0.46751895546913147
|
| 149 |
+
],
|
| 150 |
+
"q01": [
|
| 151 |
+
-0.4613877633213997,
|
| 152 |
+
-0.8093644535541534,
|
| 153 |
+
-0.9522596222162246,
|
| 154 |
+
-2.1474283027648924,
|
| 155 |
+
-0.26031066954135895,
|
| 156 |
+
-0.3504389730095863,
|
| 157 |
+
-1.570744615793228
|
| 158 |
+
],
|
| 159 |
+
"q99": [
|
| 160 |
+
1.0315392100811003,
|
| 161 |
+
0.000740842941449955,
|
| 162 |
+
0.5001527267694472,
|
| 163 |
+
-0.7204233664274217,
|
| 164 |
+
0.3507135462760925,
|
| 165 |
+
0.03258553277701139,
|
| 166 |
+
0.65115398645401
|
| 167 |
+
]
|
| 168 |
+
},
|
| 169 |
+
"left_hand": {
|
| 170 |
+
"max": [
|
| 171 |
+
0.7979256510734558
|
| 172 |
+
],
|
| 173 |
+
"min": [
|
| 174 |
+
0.308659166097641
|
| 175 |
+
],
|
| 176 |
+
"mean": [
|
| 177 |
+
0.7642289400100708
|
| 178 |
+
],
|
| 179 |
+
"std": [
|
| 180 |
+
0.09134768694639206
|
| 181 |
+
],
|
| 182 |
+
"q01": [
|
| 183 |
+
0.4216100376844406
|
| 184 |
+
],
|
| 185 |
+
"q99": [
|
| 186 |
+
0.7934369444847107
|
| 187 |
+
]
|
| 188 |
+
},
|
| 189 |
+
"left_gripper_qpos": {
|
| 190 |
+
"max": [
|
| 191 |
+
0.7979256510734558
|
| 192 |
+
],
|
| 193 |
+
"min": [
|
| 194 |
+
0.308659166097641
|
| 195 |
+
],
|
| 196 |
+
"mean": [
|
| 197 |
+
0.7642289400100708
|
| 198 |
+
],
|
| 199 |
+
"std": [
|
| 200 |
+
0.09134768694639206
|
| 201 |
+
],
|
| 202 |
+
"q01": [
|
| 203 |
+
0.4216100376844406
|
| 204 |
+
],
|
| 205 |
+
"q99": [
|
| 206 |
+
0.7934369444847107
|
| 207 |
+
]
|
| 208 |
+
},
|
| 209 |
+
"right_hand": {
|
| 210 |
+
"max": [
|
| 211 |
+
0.7982689142227173
|
| 212 |
+
],
|
| 213 |
+
"min": [
|
| 214 |
+
0.31247764825820923
|
| 215 |
+
],
|
| 216 |
+
"mean": [
|
| 217 |
+
0.7446423172950745
|
| 218 |
+
],
|
| 219 |
+
"std": [
|
| 220 |
+
0.11271625757217407
|
| 221 |
+
],
|
| 222 |
+
"q01": [
|
| 223 |
+
0.3954943937063217
|
| 224 |
+
],
|
| 225 |
+
"q99": [
|
| 226 |
+
0.7934401035308838
|
| 227 |
+
]
|
| 228 |
+
},
|
| 229 |
+
"right_gripper_qpos": {
|
| 230 |
+
"max": [
|
| 231 |
+
0.7982689142227173
|
| 232 |
+
],
|
| 233 |
+
"min": [
|
| 234 |
+
0.31247764825820923
|
| 235 |
+
],
|
| 236 |
+
"mean": [
|
| 237 |
+
0.7446423172950745
|
| 238 |
+
],
|
| 239 |
+
"std": [
|
| 240 |
+
0.11271625757217407
|
| 241 |
+
],
|
| 242 |
+
"q01": [
|
| 243 |
+
0.3954943937063217
|
| 244 |
+
],
|
| 245 |
+
"q99": [
|
| 246 |
+
0.7934401035308838
|
| 247 |
+
]
|
| 248 |
+
},
|
| 249 |
+
"left_arm_eef_pos": {
|
| 250 |
+
"max": [
|
| 251 |
+
5.777933597564697,
|
| 252 |
+
-0.46024391055107117,
|
| 253 |
+
1.008889079093933
|
| 254 |
+
],
|
| 255 |
+
"min": [
|
| 256 |
+
0.616549015045166,
|
| 257 |
+
-2.8081092834472656,
|
| 258 |
+
0.6963161826133728
|
| 259 |
+
],
|
| 260 |
+
"mean": [
|
| 261 |
+
3.3723342418670654,
|
| 262 |
+
-0.9225618839263916,
|
| 263 |
+
0.8467212319374084
|
| 264 |
+
],
|
| 265 |
+
"std": [
|
| 266 |
+
1.4625983238220215,
|
| 267 |
+
0.5607711672782898,
|
| 268 |
+
0.09720194339752197
|
| 269 |
+
],
|
| 270 |
+
"q01": [
|
| 271 |
+
0.6381882685422897,
|
| 272 |
+
-2.802727987766266,
|
| 273 |
+
0.7077158093452454
|
| 274 |
+
],
|
| 275 |
+
"q99": [
|
| 276 |
+
5.752657957077027,
|
| 277 |
+
-0.49149866312742235,
|
| 278 |
+
0.9674001401662826
|
| 279 |
+
]
|
| 280 |
+
},
|
| 281 |
+
"left_arm_eef_quat": {
|
| 282 |
+
"max": [
|
| 283 |
+
0.774299681186676,
|
| 284 |
+
0.7777989506721497,
|
| 285 |
+
0.7498747706413269,
|
| 286 |
+
0.8196669220924377
|
| 287 |
+
],
|
| 288 |
+
"min": [
|
| 289 |
+
-0.739790141582489,
|
| 290 |
+
-0.7685531973838806,
|
| 291 |
+
-0.7645168304443359,
|
| 292 |
+
7.452100544469431e-05
|
| 293 |
+
],
|
| 294 |
+
"mean": [
|
| 295 |
+
0.3311719298362732,
|
| 296 |
+
-0.43145933747291565,
|
| 297 |
+
0.23162278532981873,
|
| 298 |
+
0.4911676049232483
|
| 299 |
+
],
|
| 300 |
+
"std": [
|
| 301 |
+
0.35719844698905945,
|
| 302 |
+
0.3199194669723511,
|
| 303 |
+
0.3855322599411011,
|
| 304 |
+
0.1750989556312561
|
| 305 |
+
],
|
| 306 |
+
"q01": [
|
| 307 |
+
-0.6739041733741761,
|
| 308 |
+
-0.7349927252531052,
|
| 309 |
+
-0.7179731333255768,
|
| 310 |
+
0.012307872828096152
|
| 311 |
+
],
|
| 312 |
+
"q99": [
|
| 313 |
+
0.7349334359169006,
|
| 314 |
+
0.6683141338825225,
|
| 315 |
+
0.6888063704967499,
|
| 316 |
+
0.762659209370613
|
| 317 |
+
]
|
| 318 |
+
},
|
| 319 |
+
"right_arm_eef_pos": {
|
| 320 |
+
"max": [
|
| 321 |
+
5.781275272369385,
|
| 322 |
+
-0.4482811987400055,
|
| 323 |
+
1.0768160820007324
|
| 324 |
+
],
|
| 325 |
+
"min": [
|
| 326 |
+
0.6298854947090149,
|
| 327 |
+
-2.4375555515289307,
|
| 328 |
+
0.6910211443901062
|
| 329 |
+
],
|
| 330 |
+
"mean": [
|
| 331 |
+
3.7318050861358643,
|
| 332 |
+
-0.9764854907989502,
|
| 333 |
+
0.8498363494873047
|
| 334 |
+
],
|
| 335 |
+
"std": [
|
| 336 |
+
1.4858821630477905,
|
| 337 |
+
0.5990075469017029,
|
| 338 |
+
0.0975065678358078
|
| 339 |
+
],
|
| 340 |
+
"q01": [
|
| 341 |
+
0.658288916349411,
|
| 342 |
+
-2.4192774152755736,
|
| 343 |
+
0.7090645414590836
|
| 344 |
+
],
|
| 345 |
+
"q99": [
|
| 346 |
+
5.773581142425537,
|
| 347 |
+
-0.5025484305620194,
|
| 348 |
+
1.0039407348632812
|
| 349 |
+
]
|
| 350 |
+
},
|
| 351 |
+
"right_arm_eef_quat": {
|
| 352 |
+
"max": [
|
| 353 |
+
0.727423906326294,
|
| 354 |
+
0.706809401512146,
|
| 355 |
+
0.8224709630012512,
|
| 356 |
+
0.8125941753387451
|
| 357 |
+
],
|
| 358 |
+
"min": [
|
| 359 |
+
-0.746112585067749,
|
| 360 |
+
-0.7207116484642029,
|
| 361 |
+
-0.8228012919425964,
|
| 362 |
+
4.271203579264693e-06
|
| 363 |
+
],
|
| 364 |
+
"mean": [
|
| 365 |
+
0.31211209297180176,
|
| 366 |
+
-0.3976847529411316,
|
| 367 |
+
0.22617658972740173,
|
| 368 |
+
0.4546644687652588
|
| 369 |
+
],
|
| 370 |
+
"std": [
|
| 371 |
+
0.43644431233406067,
|
| 372 |
+
0.24810448288917542,
|
| 373 |
+
0.4512842893600464,
|
| 374 |
+
0.17576341331005096
|
| 375 |
+
],
|
| 376 |
+
"q01": [
|
| 377 |
+
-0.7195372068881989,
|
| 378 |
+
-0.7195094227790833,
|
| 379 |
+
-0.6934988576173783,
|
| 380 |
+
0.00853458046913147
|
| 381 |
+
],
|
| 382 |
+
"q99": [
|
| 383 |
+
0.6808225351572037,
|
| 384 |
+
0.48683623820543287,
|
| 385 |
+
0.7670918434858323,
|
| 386 |
+
0.7782849752902985
|
| 387 |
+
]
|
| 388 |
+
}
|
| 389 |
+
},
|
| 390 |
+
"action": {
|
| 391 |
+
"base": {
|
| 392 |
+
"max": [
|
| 393 |
+
0.0,
|
| 394 |
+
0.0,
|
| 395 |
+
0.0
|
| 396 |
+
],
|
| 397 |
+
"min": [
|
| 398 |
+
0.0,
|
| 399 |
+
0.0,
|
| 400 |
+
0.0
|
| 401 |
+
],
|
| 402 |
+
"mean": [
|
| 403 |
+
0.0,
|
| 404 |
+
0.0,
|
| 405 |
+
0.0
|
| 406 |
+
],
|
| 407 |
+
"std": [
|
| 408 |
+
0.0,
|
| 409 |
+
0.0,
|
| 410 |
+
0.0
|
| 411 |
+
],
|
| 412 |
+
"q01": [
|
| 413 |
+
0.0,
|
| 414 |
+
0.0,
|
| 415 |
+
0.0
|
| 416 |
+
],
|
| 417 |
+
"q99": [
|
| 418 |
+
0.0,
|
| 419 |
+
0.0,
|
| 420 |
+
0.0
|
| 421 |
+
]
|
| 422 |
+
},
|
| 423 |
+
"torso": {
|
| 424 |
+
"max": [
|
| 425 |
+
0.0
|
| 426 |
+
],
|
| 427 |
+
"min": [
|
| 428 |
+
0.0
|
| 429 |
+
],
|
| 430 |
+
"mean": [
|
| 431 |
+
0.0
|
| 432 |
+
],
|
| 433 |
+
"std": [
|
| 434 |
+
0.0
|
| 435 |
+
],
|
| 436 |
+
"q01": [
|
| 437 |
+
0.0
|
| 438 |
+
],
|
| 439 |
+
"q99": [
|
| 440 |
+
0.0
|
| 441 |
+
]
|
| 442 |
+
},
|
| 443 |
+
"left_arm_eef_pos": {
|
| 444 |
+
"max": [
|
| 445 |
+
1.0,
|
| 446 |
+
1.0,
|
| 447 |
+
1.0
|
| 448 |
+
],
|
| 449 |
+
"min": [
|
| 450 |
+
-1.0,
|
| 451 |
+
-1.0,
|
| 452 |
+
-1.0
|
| 453 |
+
],
|
| 454 |
+
"mean": [
|
| 455 |
+
-0.029605818912386894,
|
| 456 |
+
0.008585687726736069,
|
| 457 |
+
-0.012815090827643871
|
| 458 |
+
],
|
| 459 |
+
"std": [
|
| 460 |
+
0.2323869913816452,
|
| 461 |
+
0.20034699141979218,
|
| 462 |
+
0.14409661293029785
|
| 463 |
+
],
|
| 464 |
+
"q01": [
|
| 465 |
+
-1.0,
|
| 466 |
+
-1.0,
|
| 467 |
+
-1.0
|
| 468 |
+
],
|
| 469 |
+
"q99": [
|
| 470 |
+
1.0,
|
| 471 |
+
1.0,
|
| 472 |
+
0.0
|
| 473 |
+
]
|
| 474 |
+
},
|
| 475 |
+
"left_arm_eef_rot": {
|
| 476 |
+
"max": [
|
| 477 |
+
1.0,
|
| 478 |
+
0.0,
|
| 479 |
+
0.44999998807907104
|
| 480 |
+
],
|
| 481 |
+
"min": [
|
| 482 |
+
-0.44999998807907104,
|
| 483 |
+
0.0,
|
| 484 |
+
-0.44999998807907104
|
| 485 |
+
],
|
| 486 |
+
"mean": [
|
| 487 |
+
0.0015839255647733808,
|
| 488 |
+
0.0,
|
| 489 |
+
-0.003083235351368785
|
| 490 |
+
],
|
| 491 |
+
"std": [
|
| 492 |
+
0.09480395913124084,
|
| 493 |
+
0.0,
|
| 494 |
+
0.060461949557065964
|
| 495 |
+
],
|
| 496 |
+
"q01": [
|
| 497 |
+
-0.44999998807907104,
|
| 498 |
+
0.0,
|
| 499 |
+
-0.44999998807907104
|
| 500 |
+
],
|
| 501 |
+
"q99": [
|
| 502 |
+
0.44999998807907104,
|
| 503 |
+
0.0,
|
| 504 |
+
-0.0
|
| 505 |
+
]
|
| 506 |
+
},
|
| 507 |
+
"right_arm_eef_pos": {
|
| 508 |
+
"max": [
|
| 509 |
+
1.0,
|
| 510 |
+
1.0,
|
| 511 |
+
1.0
|
| 512 |
+
],
|
| 513 |
+
"min": [
|
| 514 |
+
-1.0,
|
| 515 |
+
-1.0,
|
| 516 |
+
-1.0
|
| 517 |
+
],
|
| 518 |
+
"mean": [
|
| 519 |
+
-0.029140586033463478,
|
| 520 |
+
0.01636778935790062,
|
| 521 |
+
-0.008501099422574043
|
| 522 |
+
],
|
| 523 |
+
"std": [
|
| 524 |
+
0.2782858908176422,
|
| 525 |
+
0.267070472240448,
|
| 526 |
+
0.19106145203113556
|
| 527 |
+
],
|
| 528 |
+
"q01": [
|
| 529 |
+
-1.0,
|
| 530 |
+
-1.0,
|
| 531 |
+
-1.0
|
| 532 |
+
],
|
| 533 |
+
"q99": [
|
| 534 |
+
1.0,
|
| 535 |
+
1.0,
|
| 536 |
+
1.0
|
| 537 |
+
]
|
| 538 |
+
},
|
| 539 |
+
"right_arm_eef_rot": {
|
| 540 |
+
"max": [
|
| 541 |
+
0.44999998807907104,
|
| 542 |
+
0.0,
|
| 543 |
+
0.44999998807907104
|
| 544 |
+
],
|
| 545 |
+
"min": [
|
| 546 |
+
-0.8999999761581421,
|
| 547 |
+
0.0,
|
| 548 |
+
-0.44999998807907104
|
| 549 |
+
],
|
| 550 |
+
"mean": [
|
| 551 |
+
0.0026645150501281023,
|
| 552 |
+
0.0,
|
| 553 |
+
0.007765152957290411
|
| 554 |
+
],
|
| 555 |
+
"std": [
|
| 556 |
+
0.0954006016254425,
|
| 557 |
+
0.0,
|
| 558 |
+
0.09971662610769272
|
| 559 |
+
],
|
| 560 |
+
"q01": [
|
| 561 |
+
-0.44999998807907104,
|
| 562 |
+
0.0,
|
| 563 |
+
-0.44999998807907104
|
| 564 |
+
],
|
| 565 |
+
"q99": [
|
| 566 |
+
0.44999998807907104,
|
| 567 |
+
0.0,
|
| 568 |
+
0.44999998807907104
|
| 569 |
+
]
|
| 570 |
+
},
|
| 571 |
+
"left_gripper_close": {
|
| 572 |
+
"max": [
|
| 573 |
+
1.0
|
| 574 |
+
],
|
| 575 |
+
"min": [
|
| 576 |
+
-1.0
|
| 577 |
+
],
|
| 578 |
+
"mean": [
|
| 579 |
+
0.7750803828239441
|
| 580 |
+
],
|
| 581 |
+
"std": [
|
| 582 |
+
0.631952166557312
|
| 583 |
+
],
|
| 584 |
+
"q01": [
|
| 585 |
+
-1.0
|
| 586 |
+
],
|
| 587 |
+
"q99": [
|
| 588 |
+
1.0
|
| 589 |
+
]
|
| 590 |
+
},
|
| 591 |
+
"right_gripper_close": {
|
| 592 |
+
"max": [
|
| 593 |
+
1.0
|
| 594 |
+
],
|
| 595 |
+
"min": [
|
| 596 |
+
-1.0
|
| 597 |
+
],
|
| 598 |
+
"mean": [
|
| 599 |
+
0.6190999746322632
|
| 600 |
+
],
|
| 601 |
+
"std": [
|
| 602 |
+
0.7853696942329407
|
| 603 |
+
],
|
| 604 |
+
"q01": [
|
| 605 |
+
-1.0
|
| 606 |
+
],
|
| 607 |
+
"q99": [
|
| 608 |
+
1.0
|
| 609 |
+
]
|
| 610 |
+
}
|
| 611 |
+
}
|
| 612 |
+
},
|
| 613 |
+
"modalities": {
|
| 614 |
+
"video": {
|
| 615 |
+
"cam_robot_0:agentview_center_rgb": {
|
| 616 |
+
"resolution": [
|
| 617 |
+
256,
|
| 618 |
+
256
|
| 619 |
+
],
|
| 620 |
+
"channels": 3,
|
| 621 |
+
"fps": 20.0
|
| 622 |
+
},
|
| 623 |
+
"cam_robot_0:agentview_left_rgb": {
|
| 624 |
+
"resolution": [
|
| 625 |
+
256,
|
| 626 |
+
256
|
| 627 |
+
],
|
| 628 |
+
"channels": 3,
|
| 629 |
+
"fps": 20.0
|
| 630 |
+
},
|
| 631 |
+
"cam_robot_0:agentview_right_rgb": {
|
| 632 |
+
"resolution": [
|
| 633 |
+
256,
|
| 634 |
+
256
|
| 635 |
+
],
|
| 636 |
+
"channels": 3,
|
| 637 |
+
"fps": 20.0
|
| 638 |
+
}
|
| 639 |
+
},
|
| 640 |
+
"state": {
|
| 641 |
+
"base": {
|
| 642 |
+
"absolute": true,
|
| 643 |
+
"rotation_type": null,
|
| 644 |
+
"shape": [
|
| 645 |
+
3
|
| 646 |
+
],
|
| 647 |
+
"continuous": true
|
| 648 |
+
},
|
| 649 |
+
"torso": {
|
| 650 |
+
"absolute": true,
|
| 651 |
+
"rotation_type": null,
|
| 652 |
+
"shape": [
|
| 653 |
+
1
|
| 654 |
+
],
|
| 655 |
+
"continuous": true
|
| 656 |
+
},
|
| 657 |
+
"left_arm": {
|
| 658 |
+
"absolute": true,
|
| 659 |
+
"rotation_type": null,
|
| 660 |
+
"shape": [
|
| 661 |
+
7
|
| 662 |
+
],
|
| 663 |
+
"continuous": true
|
| 664 |
+
},
|
| 665 |
+
"right_arm": {
|
| 666 |
+
"absolute": true,
|
| 667 |
+
"rotation_type": null,
|
| 668 |
+
"shape": [
|
| 669 |
+
7
|
| 670 |
+
],
|
| 671 |
+
"continuous": true
|
| 672 |
+
},
|
| 673 |
+
"left_hand": {
|
| 674 |
+
"absolute": true,
|
| 675 |
+
"rotation_type": null,
|
| 676 |
+
"shape": [
|
| 677 |
+
1
|
| 678 |
+
],
|
| 679 |
+
"continuous": true
|
| 680 |
+
},
|
| 681 |
+
"left_gripper_qpos": {
|
| 682 |
+
"absolute": true,
|
| 683 |
+
"rotation_type": null,
|
| 684 |
+
"shape": [
|
| 685 |
+
1
|
| 686 |
+
],
|
| 687 |
+
"continuous": true
|
| 688 |
+
},
|
| 689 |
+
"right_hand": {
|
| 690 |
+
"absolute": true,
|
| 691 |
+
"rotation_type": null,
|
| 692 |
+
"shape": [
|
| 693 |
+
1
|
| 694 |
+
],
|
| 695 |
+
"continuous": true
|
| 696 |
+
},
|
| 697 |
+
"right_gripper_qpos": {
|
| 698 |
+
"absolute": true,
|
| 699 |
+
"rotation_type": null,
|
| 700 |
+
"shape": [
|
| 701 |
+
1
|
| 702 |
+
],
|
| 703 |
+
"continuous": true
|
| 704 |
+
},
|
| 705 |
+
"left_arm_eef_pos": {
|
| 706 |
+
"absolute": true,
|
| 707 |
+
"rotation_type": null,
|
| 708 |
+
"shape": [
|
| 709 |
+
3
|
| 710 |
+
],
|
| 711 |
+
"continuous": true
|
| 712 |
+
},
|
| 713 |
+
"left_arm_eef_quat": {
|
| 714 |
+
"absolute": true,
|
| 715 |
+
"rotation_type": "quaternion",
|
| 716 |
+
"shape": [
|
| 717 |
+
4
|
| 718 |
+
],
|
| 719 |
+
"continuous": true
|
| 720 |
+
},
|
| 721 |
+
"right_arm_eef_pos": {
|
| 722 |
+
"absolute": true,
|
| 723 |
+
"rotation_type": null,
|
| 724 |
+
"shape": [
|
| 725 |
+
3
|
| 726 |
+
],
|
| 727 |
+
"continuous": true
|
| 728 |
+
},
|
| 729 |
+
"right_arm_eef_quat": {
|
| 730 |
+
"absolute": true,
|
| 731 |
+
"rotation_type": "quaternion",
|
| 732 |
+
"shape": [
|
| 733 |
+
4
|
| 734 |
+
],
|
| 735 |
+
"continuous": true
|
| 736 |
+
}
|
| 737 |
+
},
|
| 738 |
+
"action": {
|
| 739 |
+
"base": {
|
| 740 |
+
"absolute": true,
|
| 741 |
+
"rotation_type": null,
|
| 742 |
+
"shape": [
|
| 743 |
+
3
|
| 744 |
+
],
|
| 745 |
+
"continuous": true
|
| 746 |
+
},
|
| 747 |
+
"torso": {
|
| 748 |
+
"absolute": true,
|
| 749 |
+
"rotation_type": null,
|
| 750 |
+
"shape": [
|
| 751 |
+
1
|
| 752 |
+
],
|
| 753 |
+
"continuous": true
|
| 754 |
+
},
|
| 755 |
+
"left_arm_eef_pos": {
|
| 756 |
+
"absolute": true,
|
| 757 |
+
"rotation_type": null,
|
| 758 |
+
"shape": [
|
| 759 |
+
3
|
| 760 |
+
],
|
| 761 |
+
"continuous": true
|
| 762 |
+
},
|
| 763 |
+
"left_arm_eef_rot": {
|
| 764 |
+
"absolute": true,
|
| 765 |
+
"rotation_type": null,
|
| 766 |
+
"shape": [
|
| 767 |
+
3
|
| 768 |
+
],
|
| 769 |
+
"continuous": true
|
| 770 |
+
},
|
| 771 |
+
"right_arm_eef_pos": {
|
| 772 |
+
"absolute": true,
|
| 773 |
+
"rotation_type": null,
|
| 774 |
+
"shape": [
|
| 775 |
+
3
|
| 776 |
+
],
|
| 777 |
+
"continuous": true
|
| 778 |
+
},
|
| 779 |
+
"right_arm_eef_rot": {
|
| 780 |
+
"absolute": true,
|
| 781 |
+
"rotation_type": null,
|
| 782 |
+
"shape": [
|
| 783 |
+
3
|
| 784 |
+
],
|
| 785 |
+
"continuous": true
|
| 786 |
+
},
|
| 787 |
+
"left_gripper_close": {
|
| 788 |
+
"absolute": true,
|
| 789 |
+
"rotation_type": null,
|
| 790 |
+
"shape": [
|
| 791 |
+
1
|
| 792 |
+
],
|
| 793 |
+
"continuous": true
|
| 794 |
+
},
|
| 795 |
+
"right_gripper_close": {
|
| 796 |
+
"absolute": true,
|
| 797 |
+
"rotation_type": null,
|
| 798 |
+
"shape": [
|
| 799 |
+
1
|
| 800 |
+
],
|
| 801 |
+
"continuous": true
|
| 802 |
+
}
|
| 803 |
+
}
|
| 804 |
+
},
|
| 805 |
+
"embodiment_tag": "new_embodiment"
|
| 806 |
+
}
|
| 807 |
+
}
|
drawer_20251223_5k/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:acbfe35063048b4ca278ee680b282ac260e6c8c211431119a3c7b2a34edda778
|
| 3 |
+
size 4999367032
|
drawer_20251223_5k/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:132ed8db684729787d62f74f99204828f719b125d6406369286000523fa4f471
|
| 3 |
+
size 2586705312
|
drawer_20251223_5k/model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
drawer_20251223_5k/runs/Dec17_05-56-02_computeinstance-e00t0xnm5mmbva0n8n/events.out.tfevents.1765950964.computeinstance-e00t0xnm5mmbva0n8n.2615708.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:26933e3a6534e3ead4341e5e90bcf3f6afc79853979115d91a6d30ba2f209b0c
|
| 3 |
+
size 6104
|
drawer_20251223_5k/runs/Dec17_05-56-44_computeinstance-e00t0xnm5mmbva0n8n/events.out.tfevents.1765951006.computeinstance-e00t0xnm5mmbva0n8n.2643153.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5d6d128f0c30ce206d4df8a4d6d25c76833b1c8953eb50830f60c12f8792efaf
|
| 3 |
+
size 111910
|
drawer_20251223_5k/trainer_state.json
ADDED
|
@@ -0,0 +1,3543 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 6.7658998646820026,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 5000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.013531799729364006,
|
| 14 |
+
"grad_norm": 7.038280963897705,
|
| 15 |
+
"learning_rate": 3.6e-06,
|
| 16 |
+
"loss": 0.6356,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.02706359945872801,
|
| 21 |
+
"grad_norm": 1.3748748302459717,
|
| 22 |
+
"learning_rate": 7.6e-06,
|
| 23 |
+
"loss": 0.3383,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.04059539918809202,
|
| 28 |
+
"grad_norm": 1.3761250972747803,
|
| 29 |
+
"learning_rate": 1.16e-05,
|
| 30 |
+
"loss": 0.2068,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.05412719891745602,
|
| 35 |
+
"grad_norm": 0.7148615717887878,
|
| 36 |
+
"learning_rate": 1.56e-05,
|
| 37 |
+
"loss": 0.1672,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.06765899864682003,
|
| 42 |
+
"grad_norm": 0.6142189502716064,
|
| 43 |
+
"learning_rate": 1.9600000000000002e-05,
|
| 44 |
+
"loss": 0.1437,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.08119079837618404,
|
| 49 |
+
"grad_norm": 0.6323403120040894,
|
| 50 |
+
"learning_rate": 2.36e-05,
|
| 51 |
+
"loss": 0.1566,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.09472259810554803,
|
| 56 |
+
"grad_norm": 0.7742477059364319,
|
| 57 |
+
"learning_rate": 2.7600000000000003e-05,
|
| 58 |
+
"loss": 0.1181,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.10825439783491204,
|
| 63 |
+
"grad_norm": 0.7195658683776855,
|
| 64 |
+
"learning_rate": 3.16e-05,
|
| 65 |
+
"loss": 0.1254,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.12178619756427606,
|
| 70 |
+
"grad_norm": 0.4564816951751709,
|
| 71 |
+
"learning_rate": 3.56e-05,
|
| 72 |
+
"loss": 0.1064,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.13531799729364005,
|
| 77 |
+
"grad_norm": 0.6378539204597473,
|
| 78 |
+
"learning_rate": 3.960000000000001e-05,
|
| 79 |
+
"loss": 0.1191,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.14884979702300405,
|
| 84 |
+
"grad_norm": 0.41418567299842834,
|
| 85 |
+
"learning_rate": 4.36e-05,
|
| 86 |
+
"loss": 0.1022,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.16238159675236807,
|
| 91 |
+
"grad_norm": 0.5222947001457214,
|
| 92 |
+
"learning_rate": 4.76e-05,
|
| 93 |
+
"loss": 0.0977,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.17591339648173207,
|
| 98 |
+
"grad_norm": 0.40328437089920044,
|
| 99 |
+
"learning_rate": 5.16e-05,
|
| 100 |
+
"loss": 0.0966,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.18944519621109607,
|
| 105 |
+
"grad_norm": 0.45488178730010986,
|
| 106 |
+
"learning_rate": 5.560000000000001e-05,
|
| 107 |
+
"loss": 0.0977,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.2029769959404601,
|
| 112 |
+
"grad_norm": 0.48504000902175903,
|
| 113 |
+
"learning_rate": 5.96e-05,
|
| 114 |
+
"loss": 0.0918,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.2165087956698241,
|
| 119 |
+
"grad_norm": 0.6960858106613159,
|
| 120 |
+
"learning_rate": 6.36e-05,
|
| 121 |
+
"loss": 0.0865,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.23004059539918809,
|
| 126 |
+
"grad_norm": 0.39376893639564514,
|
| 127 |
+
"learning_rate": 6.76e-05,
|
| 128 |
+
"loss": 0.0825,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.2435723951285521,
|
| 133 |
+
"grad_norm": 0.6831949949264526,
|
| 134 |
+
"learning_rate": 7.16e-05,
|
| 135 |
+
"loss": 0.083,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.2571041948579161,
|
| 140 |
+
"grad_norm": 0.4721453785896301,
|
| 141 |
+
"learning_rate": 7.560000000000001e-05,
|
| 142 |
+
"loss": 0.0766,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.2706359945872801,
|
| 147 |
+
"grad_norm": 0.5577224493026733,
|
| 148 |
+
"learning_rate": 7.960000000000001e-05,
|
| 149 |
+
"loss": 0.0808,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.28416779431664413,
|
| 154 |
+
"grad_norm": 0.42680981755256653,
|
| 155 |
+
"learning_rate": 8.36e-05,
|
| 156 |
+
"loss": 0.0821,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.2976995940460081,
|
| 161 |
+
"grad_norm": 0.40946292877197266,
|
| 162 |
+
"learning_rate": 8.76e-05,
|
| 163 |
+
"loss": 0.0714,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.3112313937753721,
|
| 168 |
+
"grad_norm": 0.3602572977542877,
|
| 169 |
+
"learning_rate": 9.16e-05,
|
| 170 |
+
"loss": 0.0775,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.32476319350473615,
|
| 175 |
+
"grad_norm": 0.41732338070869446,
|
| 176 |
+
"learning_rate": 9.56e-05,
|
| 177 |
+
"loss": 0.0715,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.3382949932341001,
|
| 182 |
+
"grad_norm": 0.37370163202285767,
|
| 183 |
+
"learning_rate": 9.960000000000001e-05,
|
| 184 |
+
"loss": 0.071,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.35182679296346414,
|
| 189 |
+
"grad_norm": 0.5187098383903503,
|
| 190 |
+
"learning_rate": 9.999911419878559e-05,
|
| 191 |
+
"loss": 0.071,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.36535859269282817,
|
| 196 |
+
"grad_norm": 0.3645962178707123,
|
| 197 |
+
"learning_rate": 9.999605221019081e-05,
|
| 198 |
+
"loss": 0.0751,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.37889039242219213,
|
| 203 |
+
"grad_norm": 0.480901837348938,
|
| 204 |
+
"learning_rate": 9.999080323230761e-05,
|
| 205 |
+
"loss": 0.0714,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.39242219215155616,
|
| 210 |
+
"grad_norm": 0.40216612815856934,
|
| 211 |
+
"learning_rate": 9.998336749474329e-05,
|
| 212 |
+
"loss": 0.0794,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.4059539918809202,
|
| 217 |
+
"grad_norm": 0.5251531600952148,
|
| 218 |
+
"learning_rate": 9.997374532276107e-05,
|
| 219 |
+
"loss": 0.0651,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.41948579161028415,
|
| 224 |
+
"grad_norm": 0.404252290725708,
|
| 225 |
+
"learning_rate": 9.996193713726596e-05,
|
| 226 |
+
"loss": 0.0694,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.4330175913396482,
|
| 231 |
+
"grad_norm": 0.37963226437568665,
|
| 232 |
+
"learning_rate": 9.994794345478624e-05,
|
| 233 |
+
"loss": 0.0691,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.4465493910690122,
|
| 238 |
+
"grad_norm": 0.38739103078842163,
|
| 239 |
+
"learning_rate": 9.99317648874509e-05,
|
| 240 |
+
"loss": 0.0653,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.46008119079837617,
|
| 245 |
+
"grad_norm": 0.3632943630218506,
|
| 246 |
+
"learning_rate": 9.991340214296292e-05,
|
| 247 |
+
"loss": 0.0623,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.4736129905277402,
|
| 252 |
+
"grad_norm": 0.39061322808265686,
|
| 253 |
+
"learning_rate": 9.989285602456819e-05,
|
| 254 |
+
"loss": 0.0607,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.4871447902571042,
|
| 259 |
+
"grad_norm": 0.3611036241054535,
|
| 260 |
+
"learning_rate": 9.98701274310205e-05,
|
| 261 |
+
"loss": 0.0629,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.5006765899864682,
|
| 266 |
+
"grad_norm": 0.43176841735839844,
|
| 267 |
+
"learning_rate": 9.984521735654218e-05,
|
| 268 |
+
"loss": 0.0639,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.5142083897158322,
|
| 273 |
+
"grad_norm": 0.4912489056587219,
|
| 274 |
+
"learning_rate": 9.981812689078057e-05,
|
| 275 |
+
"loss": 0.0625,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.5277401894451962,
|
| 280 |
+
"grad_norm": 0.33789411187171936,
|
| 281 |
+
"learning_rate": 9.978885721876041e-05,
|
| 282 |
+
"loss": 0.0658,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.5412719891745602,
|
| 287 |
+
"grad_norm": 0.30039742588996887,
|
| 288 |
+
"learning_rate": 9.975740962083198e-05,
|
| 289 |
+
"loss": 0.0567,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.5548037889039242,
|
| 294 |
+
"grad_norm": 0.35562002658843994,
|
| 295 |
+
"learning_rate": 9.972378547261504e-05,
|
| 296 |
+
"loss": 0.0658,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.5683355886332883,
|
| 301 |
+
"grad_norm": 0.37655043601989746,
|
| 302 |
+
"learning_rate": 9.968798624493885e-05,
|
| 303 |
+
"loss": 0.0582,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.5818673883626523,
|
| 308 |
+
"grad_norm": 0.35913047194480896,
|
| 309 |
+
"learning_rate": 9.965001350377753e-05,
|
| 310 |
+
"loss": 0.0618,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.5953991880920162,
|
| 315 |
+
"grad_norm": 0.4832420349121094,
|
| 316 |
+
"learning_rate": 9.960986891018183e-05,
|
| 317 |
+
"loss": 0.061,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.6089309878213802,
|
| 322 |
+
"grad_norm": 0.5011168122291565,
|
| 323 |
+
"learning_rate": 9.95675542202063e-05,
|
| 324 |
+
"loss": 0.0615,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.6224627875507442,
|
| 329 |
+
"grad_norm": 0.26562678813934326,
|
| 330 |
+
"learning_rate": 9.952307128483256e-05,
|
| 331 |
+
"loss": 0.0603,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.6359945872801083,
|
| 336 |
+
"grad_norm": 0.3474351167678833,
|
| 337 |
+
"learning_rate": 9.947642204988835e-05,
|
| 338 |
+
"loss": 0.058,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.6495263870094723,
|
| 343 |
+
"grad_norm": 0.29512283205986023,
|
| 344 |
+
"learning_rate": 9.942760855596226e-05,
|
| 345 |
+
"loss": 0.0585,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.6630581867388363,
|
| 350 |
+
"grad_norm": 0.29113370180130005,
|
| 351 |
+
"learning_rate": 9.937663293831471e-05,
|
| 352 |
+
"loss": 0.0588,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.6765899864682002,
|
| 357 |
+
"grad_norm": 0.2804959714412689,
|
| 358 |
+
"learning_rate": 9.932349742678433e-05,
|
| 359 |
+
"loss": 0.0523,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.6901217861975643,
|
| 364 |
+
"grad_norm": 0.30944955348968506,
|
| 365 |
+
"learning_rate": 9.926820434569051e-05,
|
| 366 |
+
"loss": 0.0572,
|
| 367 |
+
"step": 510
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.7036535859269283,
|
| 371 |
+
"grad_norm": 0.3345446288585663,
|
| 372 |
+
"learning_rate": 9.921075611373179e-05,
|
| 373 |
+
"loss": 0.055,
|
| 374 |
+
"step": 520
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.7171853856562923,
|
| 378 |
+
"grad_norm": 0.32225489616394043,
|
| 379 |
+
"learning_rate": 9.915115524387988e-05,
|
| 380 |
+
"loss": 0.0555,
|
| 381 |
+
"step": 530
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.7307171853856563,
|
| 385 |
+
"grad_norm": 0.5840979218482971,
|
| 386 |
+
"learning_rate": 9.908940434326997e-05,
|
| 387 |
+
"loss": 0.056,
|
| 388 |
+
"step": 540
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.7442489851150202,
|
| 392 |
+
"grad_norm": 0.2597893476486206,
|
| 393 |
+
"learning_rate": 9.902550611308645e-05,
|
| 394 |
+
"loss": 0.0504,
|
| 395 |
+
"step": 550
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.7577807848443843,
|
| 399 |
+
"grad_norm": 0.31497254967689514,
|
| 400 |
+
"learning_rate": 9.895946334844494e-05,
|
| 401 |
+
"loss": 0.052,
|
| 402 |
+
"step": 560
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.7713125845737483,
|
| 406 |
+
"grad_norm": 0.2690199017524719,
|
| 407 |
+
"learning_rate": 9.889127893826989e-05,
|
| 408 |
+
"loss": 0.0538,
|
| 409 |
+
"step": 570
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.7848443843031123,
|
| 413 |
+
"grad_norm": 0.3441496789455414,
|
| 414 |
+
"learning_rate": 9.882095586516831e-05,
|
| 415 |
+
"loss": 0.0542,
|
| 416 |
+
"step": 580
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.7983761840324763,
|
| 420 |
+
"grad_norm": 0.339771568775177,
|
| 421 |
+
"learning_rate": 9.874849720529921e-05,
|
| 422 |
+
"loss": 0.0579,
|
| 423 |
+
"step": 590
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.8119079837618404,
|
| 427 |
+
"grad_norm": 0.26893261075019836,
|
| 428 |
+
"learning_rate": 9.867390612823914e-05,
|
| 429 |
+
"loss": 0.0549,
|
| 430 |
+
"step": 600
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.8254397834912043,
|
| 434 |
+
"grad_norm": 0.2606127858161926,
|
| 435 |
+
"learning_rate": 9.859718589684344e-05,
|
| 436 |
+
"loss": 0.0556,
|
| 437 |
+
"step": 610
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.8389715832205683,
|
| 441 |
+
"grad_norm": 0.29616016149520874,
|
| 442 |
+
"learning_rate": 9.851833986710353e-05,
|
| 443 |
+
"loss": 0.0525,
|
| 444 |
+
"step": 620
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.8525033829499323,
|
| 448 |
+
"grad_norm": 0.2588672339916229,
|
| 449 |
+
"learning_rate": 9.843737148800023e-05,
|
| 450 |
+
"loss": 0.0473,
|
| 451 |
+
"step": 630
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.8660351826792964,
|
| 455 |
+
"grad_norm": 0.23314815759658813,
|
| 456 |
+
"learning_rate": 9.835428430135271e-05,
|
| 457 |
+
"loss": 0.045,
|
| 458 |
+
"step": 640
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.8795669824086604,
|
| 462 |
+
"grad_norm": 0.2904779613018036,
|
| 463 |
+
"learning_rate": 9.82690819416637e-05,
|
| 464 |
+
"loss": 0.0549,
|
| 465 |
+
"step": 650
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.8930987821380244,
|
| 469 |
+
"grad_norm": 0.27372926473617554,
|
| 470 |
+
"learning_rate": 9.818176813596041e-05,
|
| 471 |
+
"loss": 0.0498,
|
| 472 |
+
"step": 660
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.9066305818673883,
|
| 476 |
+
"grad_norm": 0.37094977498054504,
|
| 477 |
+
"learning_rate": 9.809234670363159e-05,
|
| 478 |
+
"loss": 0.0544,
|
| 479 |
+
"step": 670
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.9201623815967523,
|
| 483 |
+
"grad_norm": 0.34140118956565857,
|
| 484 |
+
"learning_rate": 9.800082155626034e-05,
|
| 485 |
+
"loss": 0.0546,
|
| 486 |
+
"step": 680
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.9336941813261164,
|
| 490 |
+
"grad_norm": 0.2957614064216614,
|
| 491 |
+
"learning_rate": 9.790719669745312e-05,
|
| 492 |
+
"loss": 0.0508,
|
| 493 |
+
"step": 690
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.9472259810554804,
|
| 497 |
+
"grad_norm": 0.2524469494819641,
|
| 498 |
+
"learning_rate": 9.781147622266455e-05,
|
| 499 |
+
"loss": 0.0521,
|
| 500 |
+
"step": 700
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.9607577807848444,
|
| 504 |
+
"grad_norm": 0.26907259225845337,
|
| 505 |
+
"learning_rate": 9.771366431901831e-05,
|
| 506 |
+
"loss": 0.0552,
|
| 507 |
+
"step": 710
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.9742895805142084,
|
| 511 |
+
"grad_norm": 0.2906481921672821,
|
| 512 |
+
"learning_rate": 9.761376526512394e-05,
|
| 513 |
+
"loss": 0.0516,
|
| 514 |
+
"step": 720
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.9878213802435724,
|
| 518 |
+
"grad_norm": 0.21193251013755798,
|
| 519 |
+
"learning_rate": 9.751178343088963e-05,
|
| 520 |
+
"loss": 0.0523,
|
| 521 |
+
"step": 730
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 1.0013531799729365,
|
| 525 |
+
"grad_norm": 0.23670238256454468,
|
| 526 |
+
"learning_rate": 9.740772327733123e-05,
|
| 527 |
+
"loss": 0.0466,
|
| 528 |
+
"step": 740
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 1.0148849797023005,
|
| 532 |
+
"grad_norm": 0.23938466608524323,
|
| 533 |
+
"learning_rate": 9.730158935637697e-05,
|
| 534 |
+
"loss": 0.056,
|
| 535 |
+
"step": 750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 1.0284167794316643,
|
| 539 |
+
"grad_norm": 0.23213449120521545,
|
| 540 |
+
"learning_rate": 9.719338631066834e-05,
|
| 541 |
+
"loss": 0.0488,
|
| 542 |
+
"step": 760
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 1.0419485791610283,
|
| 546 |
+
"grad_norm": 0.35314860939979553,
|
| 547 |
+
"learning_rate": 9.708311887335713e-05,
|
| 548 |
+
"loss": 0.0519,
|
| 549 |
+
"step": 770
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.0554803788903924,
|
| 553 |
+
"grad_norm": 0.3346068859100342,
|
| 554 |
+
"learning_rate": 9.697079186789823e-05,
|
| 555 |
+
"loss": 0.0499,
|
| 556 |
+
"step": 780
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 1.0690121786197564,
|
| 560 |
+
"grad_norm": 0.3368820250034332,
|
| 561 |
+
"learning_rate": 9.685641020783876e-05,
|
| 562 |
+
"loss": 0.0506,
|
| 563 |
+
"step": 790
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 1.0825439783491204,
|
| 567 |
+
"grad_norm": 0.25122860074043274,
|
| 568 |
+
"learning_rate": 9.67399788966031e-05,
|
| 569 |
+
"loss": 0.0492,
|
| 570 |
+
"step": 800
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 1.0960757780784844,
|
| 574 |
+
"grad_norm": 0.20261575281620026,
|
| 575 |
+
"learning_rate": 9.662150302727395e-05,
|
| 576 |
+
"loss": 0.0505,
|
| 577 |
+
"step": 810
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 1.1096075778078485,
|
| 581 |
+
"grad_norm": 0.2698753774166107,
|
| 582 |
+
"learning_rate": 9.650098778236968e-05,
|
| 583 |
+
"loss": 0.0521,
|
| 584 |
+
"step": 820
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 1.1231393775372125,
|
| 588 |
+
"grad_norm": 0.29095539450645447,
|
| 589 |
+
"learning_rate": 9.637843843361749e-05,
|
| 590 |
+
"loss": 0.0457,
|
| 591 |
+
"step": 830
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 1.1366711772665765,
|
| 595 |
+
"grad_norm": 0.2353201061487198,
|
| 596 |
+
"learning_rate": 9.62538603417229e-05,
|
| 597 |
+
"loss": 0.044,
|
| 598 |
+
"step": 840
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 1.1502029769959405,
|
| 602 |
+
"grad_norm": 0.2477278709411621,
|
| 603 |
+
"learning_rate": 9.612725895613526e-05,
|
| 604 |
+
"loss": 0.0531,
|
| 605 |
+
"step": 850
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 1.1637347767253043,
|
| 609 |
+
"grad_norm": 0.31642410159111023,
|
| 610 |
+
"learning_rate": 9.599863981480926e-05,
|
| 611 |
+
"loss": 0.0523,
|
| 612 |
+
"step": 860
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 1.1772665764546684,
|
| 616 |
+
"grad_norm": 0.305105984210968,
|
| 617 |
+
"learning_rate": 9.586800854396283e-05,
|
| 618 |
+
"loss": 0.0496,
|
| 619 |
+
"step": 870
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 1.1907983761840324,
|
| 623 |
+
"grad_norm": 0.44601723551750183,
|
| 624 |
+
"learning_rate": 9.573537085783095e-05,
|
| 625 |
+
"loss": 0.0507,
|
| 626 |
+
"step": 880
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 1.2043301759133964,
|
| 630 |
+
"grad_norm": 0.4266583323478699,
|
| 631 |
+
"learning_rate": 9.560073255841571e-05,
|
| 632 |
+
"loss": 0.0485,
|
| 633 |
+
"step": 890
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 1.2178619756427604,
|
| 637 |
+
"grad_norm": 0.19282500445842743,
|
| 638 |
+
"learning_rate": 9.546409953523247e-05,
|
| 639 |
+
"loss": 0.0497,
|
| 640 |
+
"step": 900
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 1.2313937753721245,
|
| 644 |
+
"grad_norm": 0.23939521610736847,
|
| 645 |
+
"learning_rate": 9.532547776505229e-05,
|
| 646 |
+
"loss": 0.0481,
|
| 647 |
+
"step": 910
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 1.2449255751014885,
|
| 651 |
+
"grad_norm": 0.24894139170646667,
|
| 652 |
+
"learning_rate": 9.518487331164048e-05,
|
| 653 |
+
"loss": 0.0483,
|
| 654 |
+
"step": 920
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 1.2584573748308525,
|
| 658 |
+
"grad_norm": 0.3222191631793976,
|
| 659 |
+
"learning_rate": 9.504229232549134e-05,
|
| 660 |
+
"loss": 0.0472,
|
| 661 |
+
"step": 930
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 1.2719891745602165,
|
| 665 |
+
"grad_norm": 0.2381894290447235,
|
| 666 |
+
"learning_rate": 9.489774104355909e-05,
|
| 667 |
+
"loss": 0.0434,
|
| 668 |
+
"step": 940
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 1.2855209742895806,
|
| 672 |
+
"grad_norm": 0.3593835234642029,
|
| 673 |
+
"learning_rate": 9.475122578898507e-05,
|
| 674 |
+
"loss": 0.0509,
|
| 675 |
+
"step": 950
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 1.2990527740189446,
|
| 679 |
+
"grad_norm": 0.33093661069869995,
|
| 680 |
+
"learning_rate": 9.460275297082119e-05,
|
| 681 |
+
"loss": 0.0445,
|
| 682 |
+
"step": 960
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 1.3125845737483086,
|
| 686 |
+
"grad_norm": 0.242259219288826,
|
| 687 |
+
"learning_rate": 9.445232908374948e-05,
|
| 688 |
+
"loss": 0.0498,
|
| 689 |
+
"step": 970
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 1.3261163734776726,
|
| 693 |
+
"grad_norm": 0.29638198018074036,
|
| 694 |
+
"learning_rate": 9.429996070779808e-05,
|
| 695 |
+
"loss": 0.0428,
|
| 696 |
+
"step": 980
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 1.3396481732070367,
|
| 700 |
+
"grad_norm": 0.2383440136909485,
|
| 701 |
+
"learning_rate": 9.414565450805333e-05,
|
| 702 |
+
"loss": 0.0479,
|
| 703 |
+
"step": 990
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 1.3531799729364005,
|
| 707 |
+
"grad_norm": 0.2849849760532379,
|
| 708 |
+
"learning_rate": 9.398941723436831e-05,
|
| 709 |
+
"loss": 0.0452,
|
| 710 |
+
"step": 1000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 1.3667117726657645,
|
| 714 |
+
"grad_norm": 0.26978209614753723,
|
| 715 |
+
"learning_rate": 9.383125572106752e-05,
|
| 716 |
+
"loss": 0.0435,
|
| 717 |
+
"step": 1010
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 1.3802435723951285,
|
| 721 |
+
"grad_norm": 0.2569567561149597,
|
| 722 |
+
"learning_rate": 9.367117688664791e-05,
|
| 723 |
+
"loss": 0.049,
|
| 724 |
+
"step": 1020
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 1.3937753721244925,
|
| 728 |
+
"grad_norm": 0.2809486389160156,
|
| 729 |
+
"learning_rate": 9.35091877334763e-05,
|
| 730 |
+
"loss": 0.0454,
|
| 731 |
+
"step": 1030
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 1.4073071718538566,
|
| 735 |
+
"grad_norm": 0.28595587611198425,
|
| 736 |
+
"learning_rate": 9.334529534748297e-05,
|
| 737 |
+
"loss": 0.0518,
|
| 738 |
+
"step": 1040
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 1.4208389715832206,
|
| 742 |
+
"grad_norm": 0.27410170435905457,
|
| 743 |
+
"learning_rate": 9.317950689785188e-05,
|
| 744 |
+
"loss": 0.0435,
|
| 745 |
+
"step": 1050
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 1.4343707713125846,
|
| 749 |
+
"grad_norm": 0.23330658674240112,
|
| 750 |
+
"learning_rate": 9.301182963670688e-05,
|
| 751 |
+
"loss": 0.0498,
|
| 752 |
+
"step": 1060
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 1.4479025710419486,
|
| 756 |
+
"grad_norm": 0.2517768144607544,
|
| 757 |
+
"learning_rate": 9.284227089879456e-05,
|
| 758 |
+
"loss": 0.0479,
|
| 759 |
+
"step": 1070
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 1.4614343707713127,
|
| 763 |
+
"grad_norm": 0.23287087678909302,
|
| 764 |
+
"learning_rate": 9.26708381011634e-05,
|
| 765 |
+
"loss": 0.0478,
|
| 766 |
+
"step": 1080
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 1.4749661705006765,
|
| 770 |
+
"grad_norm": 0.20903582870960236,
|
| 771 |
+
"learning_rate": 9.249753874283937e-05,
|
| 772 |
+
"loss": 0.0456,
|
| 773 |
+
"step": 1090
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.4884979702300405,
|
| 777 |
+
"grad_norm": 0.21921710669994354,
|
| 778 |
+
"learning_rate": 9.232238040449779e-05,
|
| 779 |
+
"loss": 0.0461,
|
| 780 |
+
"step": 1100
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.5020297699594045,
|
| 784 |
+
"grad_norm": 0.258622407913208,
|
| 785 |
+
"learning_rate": 9.214537074813181e-05,
|
| 786 |
+
"loss": 0.0427,
|
| 787 |
+
"step": 1110
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.5155615696887685,
|
| 791 |
+
"grad_norm": 0.25410041213035583,
|
| 792 |
+
"learning_rate": 9.196651751671724e-05,
|
| 793 |
+
"loss": 0.0483,
|
| 794 |
+
"step": 1120
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.5290933694181326,
|
| 798 |
+
"grad_norm": 0.24682465195655823,
|
| 799 |
+
"learning_rate": 9.178582853387384e-05,
|
| 800 |
+
"loss": 0.0491,
|
| 801 |
+
"step": 1130
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 1.5426251691474966,
|
| 805 |
+
"grad_norm": 0.24718515574932098,
|
| 806 |
+
"learning_rate": 9.160331170352304e-05,
|
| 807 |
+
"loss": 0.0488,
|
| 808 |
+
"step": 1140
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 1.5561569688768606,
|
| 812 |
+
"grad_norm": 0.2638084292411804,
|
| 813 |
+
"learning_rate": 9.141897500954229e-05,
|
| 814 |
+
"loss": 0.0486,
|
| 815 |
+
"step": 1150
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 1.5696887686062246,
|
| 819 |
+
"grad_norm": 0.2443169504404068,
|
| 820 |
+
"learning_rate": 9.123282651541576e-05,
|
| 821 |
+
"loss": 0.0449,
|
| 822 |
+
"step": 1160
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 1.5832205683355887,
|
| 826 |
+
"grad_norm": 0.24525044858455658,
|
| 827 |
+
"learning_rate": 9.104487436388161e-05,
|
| 828 |
+
"loss": 0.044,
|
| 829 |
+
"step": 1170
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 1.5967523680649527,
|
| 833 |
+
"grad_norm": 0.225367471575737,
|
| 834 |
+
"learning_rate": 9.085512677657582e-05,
|
| 835 |
+
"loss": 0.0401,
|
| 836 |
+
"step": 1180
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 1.6102841677943167,
|
| 840 |
+
"grad_norm": 0.2916682958602905,
|
| 841 |
+
"learning_rate": 9.066359205367258e-05,
|
| 842 |
+
"loss": 0.0412,
|
| 843 |
+
"step": 1190
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 1.6238159675236807,
|
| 847 |
+
"grad_norm": 0.21963126957416534,
|
| 848 |
+
"learning_rate": 9.047027857352112e-05,
|
| 849 |
+
"loss": 0.0482,
|
| 850 |
+
"step": 1200
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 1.6373477672530448,
|
| 854 |
+
"grad_norm": 0.25542739033699036,
|
| 855 |
+
"learning_rate": 9.027519479227935e-05,
|
| 856 |
+
"loss": 0.0465,
|
| 857 |
+
"step": 1210
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 1.6508795669824088,
|
| 861 |
+
"grad_norm": 0.1971822828054428,
|
| 862 |
+
"learning_rate": 9.007834924354383e-05,
|
| 863 |
+
"loss": 0.0443,
|
| 864 |
+
"step": 1220
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 1.6644113667117728,
|
| 868 |
+
"grad_norm": 0.1578320562839508,
|
| 869 |
+
"learning_rate": 8.987975053797655e-05,
|
| 870 |
+
"loss": 0.0443,
|
| 871 |
+
"step": 1230
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 1.6779431664411368,
|
| 875 |
+
"grad_norm": 0.25263530015945435,
|
| 876 |
+
"learning_rate": 8.967940736292825e-05,
|
| 877 |
+
"loss": 0.0434,
|
| 878 |
+
"step": 1240
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 1.6914749661705006,
|
| 882 |
+
"grad_norm": 0.2303190976381302,
|
| 883 |
+
"learning_rate": 8.947732848205846e-05,
|
| 884 |
+
"loss": 0.0491,
|
| 885 |
+
"step": 1250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 1.7050067658998647,
|
| 889 |
+
"grad_norm": 0.27885767817497253,
|
| 890 |
+
"learning_rate": 8.927352273495204e-05,
|
| 891 |
+
"loss": 0.041,
|
| 892 |
+
"step": 1260
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 1.7185385656292287,
|
| 896 |
+
"grad_norm": 0.2695358991622925,
|
| 897 |
+
"learning_rate": 8.906799903673265e-05,
|
| 898 |
+
"loss": 0.0434,
|
| 899 |
+
"step": 1270
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 1.7320703653585927,
|
| 903 |
+
"grad_norm": 0.24871905148029327,
|
| 904 |
+
"learning_rate": 8.88607663776726e-05,
|
| 905 |
+
"loss": 0.0439,
|
| 906 |
+
"step": 1280
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 1.7456021650879567,
|
| 910 |
+
"grad_norm": 0.2683444023132324,
|
| 911 |
+
"learning_rate": 8.865183382279978e-05,
|
| 912 |
+
"loss": 0.0442,
|
| 913 |
+
"step": 1290
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 1.7591339648173205,
|
| 917 |
+
"grad_norm": 0.26836928725242615,
|
| 918 |
+
"learning_rate": 8.844121051150096e-05,
|
| 919 |
+
"loss": 0.0425,
|
| 920 |
+
"step": 1300
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 1.7726657645466846,
|
| 924 |
+
"grad_norm": 0.18367311358451843,
|
| 925 |
+
"learning_rate": 8.822890565712211e-05,
|
| 926 |
+
"loss": 0.0429,
|
| 927 |
+
"step": 1310
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 1.7861975642760486,
|
| 931 |
+
"grad_norm": 0.249272882938385,
|
| 932 |
+
"learning_rate": 8.801492854656536e-05,
|
| 933 |
+
"loss": 0.046,
|
| 934 |
+
"step": 1320
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 1.7997293640054126,
|
| 938 |
+
"grad_norm": 0.2543359100818634,
|
| 939 |
+
"learning_rate": 8.779928853988268e-05,
|
| 940 |
+
"loss": 0.0434,
|
| 941 |
+
"step": 1330
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 1.8132611637347766,
|
| 945 |
+
"grad_norm": 0.27556613087654114,
|
| 946 |
+
"learning_rate": 8.758199506986655e-05,
|
| 947 |
+
"loss": 0.0391,
|
| 948 |
+
"step": 1340
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 1.8267929634641407,
|
| 952 |
+
"grad_norm": 0.24716362357139587,
|
| 953 |
+
"learning_rate": 8.73630576416373e-05,
|
| 954 |
+
"loss": 0.0381,
|
| 955 |
+
"step": 1350
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 1.8403247631935047,
|
| 959 |
+
"grad_norm": 0.29948118329048157,
|
| 960 |
+
"learning_rate": 8.714248583222726e-05,
|
| 961 |
+
"loss": 0.043,
|
| 962 |
+
"step": 1360
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 1.8538565629228687,
|
| 966 |
+
"grad_norm": 0.3200075626373291,
|
| 967 |
+
"learning_rate": 8.692028929016196e-05,
|
| 968 |
+
"loss": 0.0462,
|
| 969 |
+
"step": 1370
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 1.8673883626522327,
|
| 973 |
+
"grad_norm": 0.23841191828250885,
|
| 974 |
+
"learning_rate": 8.669647773503797e-05,
|
| 975 |
+
"loss": 0.0411,
|
| 976 |
+
"step": 1380
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 1.8809201623815968,
|
| 980 |
+
"grad_norm": 0.21782785654067993,
|
| 981 |
+
"learning_rate": 8.647106095709773e-05,
|
| 982 |
+
"loss": 0.0435,
|
| 983 |
+
"step": 1390
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 1.8944519621109608,
|
| 987 |
+
"grad_norm": 0.22446033358573914,
|
| 988 |
+
"learning_rate": 8.624404881680139e-05,
|
| 989 |
+
"loss": 0.041,
|
| 990 |
+
"step": 1400
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 1.9079837618403248,
|
| 994 |
+
"grad_norm": 0.21831940114498138,
|
| 995 |
+
"learning_rate": 8.601545124439535e-05,
|
| 996 |
+
"loss": 0.0422,
|
| 997 |
+
"step": 1410
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 1.9215155615696888,
|
| 1001 |
+
"grad_norm": 0.3092878460884094,
|
| 1002 |
+
"learning_rate": 8.5785278239478e-05,
|
| 1003 |
+
"loss": 0.046,
|
| 1004 |
+
"step": 1420
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 1.9350473612990529,
|
| 1008 |
+
"grad_norm": 0.27636203169822693,
|
| 1009 |
+
"learning_rate": 8.555353987056224e-05,
|
| 1010 |
+
"loss": 0.0457,
|
| 1011 |
+
"step": 1430
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 1.9485791610284169,
|
| 1015 |
+
"grad_norm": 0.2327750325202942,
|
| 1016 |
+
"learning_rate": 8.532024627463505e-05,
|
| 1017 |
+
"loss": 0.0431,
|
| 1018 |
+
"step": 1440
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 1.962110960757781,
|
| 1022 |
+
"grad_norm": 0.20582954585552216,
|
| 1023 |
+
"learning_rate": 8.508540765671407e-05,
|
| 1024 |
+
"loss": 0.0408,
|
| 1025 |
+
"step": 1450
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 1.975642760487145,
|
| 1029 |
+
"grad_norm": 0.2168937623500824,
|
| 1030 |
+
"learning_rate": 8.484903428940121e-05,
|
| 1031 |
+
"loss": 0.0405,
|
| 1032 |
+
"step": 1460
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 1.989174560216509,
|
| 1036 |
+
"grad_norm": 0.2852031886577606,
|
| 1037 |
+
"learning_rate": 8.461113651243334e-05,
|
| 1038 |
+
"loss": 0.044,
|
| 1039 |
+
"step": 1470
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 2.002706359945873,
|
| 1043 |
+
"grad_norm": 0.25714772939682007,
|
| 1044 |
+
"learning_rate": 8.437172473222987e-05,
|
| 1045 |
+
"loss": 0.0458,
|
| 1046 |
+
"step": 1480
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 2.016238159675237,
|
| 1050 |
+
"grad_norm": 0.21723440289497375,
|
| 1051 |
+
"learning_rate": 8.413080942143767e-05,
|
| 1052 |
+
"loss": 0.0419,
|
| 1053 |
+
"step": 1490
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 2.029769959404601,
|
| 1057 |
+
"grad_norm": 0.3050743043422699,
|
| 1058 |
+
"learning_rate": 8.388840111847288e-05,
|
| 1059 |
+
"loss": 0.0488,
|
| 1060 |
+
"step": 1500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 2.0433017591339646,
|
| 1064 |
+
"grad_norm": 0.251375287771225,
|
| 1065 |
+
"learning_rate": 8.364451042705998e-05,
|
| 1066 |
+
"loss": 0.0413,
|
| 1067 |
+
"step": 1510
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 2.0568335588633286,
|
| 1071 |
+
"grad_norm": 0.2977745831012726,
|
| 1072 |
+
"learning_rate": 8.33991480157679e-05,
|
| 1073 |
+
"loss": 0.0419,
|
| 1074 |
+
"step": 1520
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 2.0703653585926927,
|
| 1078 |
+
"grad_norm": 0.22888727486133575,
|
| 1079 |
+
"learning_rate": 8.315232461754338e-05,
|
| 1080 |
+
"loss": 0.0432,
|
| 1081 |
+
"step": 1530
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 2.0838971583220567,
|
| 1085 |
+
"grad_norm": 0.2524541914463043,
|
| 1086 |
+
"learning_rate": 8.290405102924144e-05,
|
| 1087 |
+
"loss": 0.0465,
|
| 1088 |
+
"step": 1540
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 2.0974289580514207,
|
| 1092 |
+
"grad_norm": 0.2784097492694855,
|
| 1093 |
+
"learning_rate": 8.265433811115316e-05,
|
| 1094 |
+
"loss": 0.04,
|
| 1095 |
+
"step": 1550
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 2.1109607577807847,
|
| 1099 |
+
"grad_norm": 0.1700204312801361,
|
| 1100 |
+
"learning_rate": 8.240319678653049e-05,
|
| 1101 |
+
"loss": 0.0426,
|
| 1102 |
+
"step": 1560
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 2.1244925575101488,
|
| 1106 |
+
"grad_norm": 0.27274981141090393,
|
| 1107 |
+
"learning_rate": 8.215063804110857e-05,
|
| 1108 |
+
"loss": 0.0427,
|
| 1109 |
+
"step": 1570
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 2.138024357239513,
|
| 1113 |
+
"grad_norm": 0.19321367144584656,
|
| 1114 |
+
"learning_rate": 8.189667292262512e-05,
|
| 1115 |
+
"loss": 0.0401,
|
| 1116 |
+
"step": 1580
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 2.151556156968877,
|
| 1120 |
+
"grad_norm": 0.2078326791524887,
|
| 1121 |
+
"learning_rate": 8.164131254033716e-05,
|
| 1122 |
+
"loss": 0.042,
|
| 1123 |
+
"step": 1590
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 2.165087956698241,
|
| 1127 |
+
"grad_norm": 0.2133825272321701,
|
| 1128 |
+
"learning_rate": 8.138456806453503e-05,
|
| 1129 |
+
"loss": 0.0393,
|
| 1130 |
+
"step": 1600
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 2.178619756427605,
|
| 1134 |
+
"grad_norm": 0.2624984383583069,
|
| 1135 |
+
"learning_rate": 8.112645072605386e-05,
|
| 1136 |
+
"loss": 0.0428,
|
| 1137 |
+
"step": 1610
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 2.192151556156969,
|
| 1141 |
+
"grad_norm": 0.23092296719551086,
|
| 1142 |
+
"learning_rate": 8.086697181578222e-05,
|
| 1143 |
+
"loss": 0.0376,
|
| 1144 |
+
"step": 1620
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 2.205683355886333,
|
| 1148 |
+
"grad_norm": 0.2146901935338974,
|
| 1149 |
+
"learning_rate": 8.060614268416823e-05,
|
| 1150 |
+
"loss": 0.0419,
|
| 1151 |
+
"step": 1630
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 2.219215155615697,
|
| 1155 |
+
"grad_norm": 0.2366856336593628,
|
| 1156 |
+
"learning_rate": 8.034397474072309e-05,
|
| 1157 |
+
"loss": 0.0389,
|
| 1158 |
+
"step": 1640
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 2.232746955345061,
|
| 1162 |
+
"grad_norm": 0.19228318333625793,
|
| 1163 |
+
"learning_rate": 8.008047945352193e-05,
|
| 1164 |
+
"loss": 0.0427,
|
| 1165 |
+
"step": 1650
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 2.246278755074425,
|
| 1169 |
+
"grad_norm": 0.25969430804252625,
|
| 1170 |
+
"learning_rate": 7.981566834870225e-05,
|
| 1171 |
+
"loss": 0.035,
|
| 1172 |
+
"step": 1660
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 2.259810554803789,
|
| 1176 |
+
"grad_norm": 0.23067066073417664,
|
| 1177 |
+
"learning_rate": 7.954955300995961e-05,
|
| 1178 |
+
"loss": 0.0363,
|
| 1179 |
+
"step": 1670
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 2.273342354533153,
|
| 1183 |
+
"grad_norm": 0.20263032615184784,
|
| 1184 |
+
"learning_rate": 7.928214507804104e-05,
|
| 1185 |
+
"loss": 0.0392,
|
| 1186 |
+
"step": 1680
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 2.286874154262517,
|
| 1190 |
+
"grad_norm": 0.2349165827035904,
|
| 1191 |
+
"learning_rate": 7.901345625023576e-05,
|
| 1192 |
+
"loss": 0.0429,
|
| 1193 |
+
"step": 1690
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 2.300405953991881,
|
| 1197 |
+
"grad_norm": 0.2644764482975006,
|
| 1198 |
+
"learning_rate": 7.874349827986354e-05,
|
| 1199 |
+
"loss": 0.0376,
|
| 1200 |
+
"step": 1700
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 2.313937753721245,
|
| 1204 |
+
"grad_norm": 0.1891845464706421,
|
| 1205 |
+
"learning_rate": 7.847228297576053e-05,
|
| 1206 |
+
"loss": 0.0394,
|
| 1207 |
+
"step": 1710
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 2.3274695534506087,
|
| 1211 |
+
"grad_norm": 0.2525619566440582,
|
| 1212 |
+
"learning_rate": 7.819982220176276e-05,
|
| 1213 |
+
"loss": 0.0416,
|
| 1214 |
+
"step": 1720
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 2.3410013531799727,
|
| 1218 |
+
"grad_norm": 0.1973452866077423,
|
| 1219 |
+
"learning_rate": 7.792612787618714e-05,
|
| 1220 |
+
"loss": 0.041,
|
| 1221 |
+
"step": 1730
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 2.3545331529093367,
|
| 1225 |
+
"grad_norm": 0.2173379361629486,
|
| 1226 |
+
"learning_rate": 7.765121197131009e-05,
|
| 1227 |
+
"loss": 0.0389,
|
| 1228 |
+
"step": 1740
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 2.3680649526387008,
|
| 1232 |
+
"grad_norm": 0.2490306794643402,
|
| 1233 |
+
"learning_rate": 7.737508651284391e-05,
|
| 1234 |
+
"loss": 0.0441,
|
| 1235 |
+
"step": 1750
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 2.381596752368065,
|
| 1239 |
+
"grad_norm": 0.26197293400764465,
|
| 1240 |
+
"learning_rate": 7.709776357941069e-05,
|
| 1241 |
+
"loss": 0.0451,
|
| 1242 |
+
"step": 1760
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 2.395128552097429,
|
| 1246 |
+
"grad_norm": 0.2130267322063446,
|
| 1247 |
+
"learning_rate": 7.681925530201392e-05,
|
| 1248 |
+
"loss": 0.0422,
|
| 1249 |
+
"step": 1770
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 2.408660351826793,
|
| 1253 |
+
"grad_norm": 0.19408588111400604,
|
| 1254 |
+
"learning_rate": 7.65395738635079e-05,
|
| 1255 |
+
"loss": 0.0409,
|
| 1256 |
+
"step": 1780
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 2.422192151556157,
|
| 1260 |
+
"grad_norm": 0.3232887387275696,
|
| 1261 |
+
"learning_rate": 7.62587314980648e-05,
|
| 1262 |
+
"loss": 0.0457,
|
| 1263 |
+
"step": 1790
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 2.435723951285521,
|
| 1267 |
+
"grad_norm": 0.21439595520496368,
|
| 1268 |
+
"learning_rate": 7.597674049063947e-05,
|
| 1269 |
+
"loss": 0.0401,
|
| 1270 |
+
"step": 1800
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 2.449255751014885,
|
| 1274 |
+
"grad_norm": 0.20550991594791412,
|
| 1275 |
+
"learning_rate": 7.569361317643211e-05,
|
| 1276 |
+
"loss": 0.0435,
|
| 1277 |
+
"step": 1810
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 2.462787550744249,
|
| 1281 |
+
"grad_norm": 0.1518939584493637,
|
| 1282 |
+
"learning_rate": 7.540936194034865e-05,
|
| 1283 |
+
"loss": 0.0382,
|
| 1284 |
+
"step": 1820
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 2.476319350473613,
|
| 1288 |
+
"grad_norm": 0.22963009774684906,
|
| 1289 |
+
"learning_rate": 7.512399921645901e-05,
|
| 1290 |
+
"loss": 0.0358,
|
| 1291 |
+
"step": 1830
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 2.489851150202977,
|
| 1295 |
+
"grad_norm": 0.17325811088085175,
|
| 1296 |
+
"learning_rate": 7.483753748745317e-05,
|
| 1297 |
+
"loss": 0.0457,
|
| 1298 |
+
"step": 1840
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 2.503382949932341,
|
| 1302 |
+
"grad_norm": 0.18786922097206116,
|
| 1303 |
+
"learning_rate": 7.454998928409516e-05,
|
| 1304 |
+
"loss": 0.0372,
|
| 1305 |
+
"step": 1850
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 2.516914749661705,
|
| 1309 |
+
"grad_norm": 0.18113826215267181,
|
| 1310 |
+
"learning_rate": 7.426136718467493e-05,
|
| 1311 |
+
"loss": 0.0395,
|
| 1312 |
+
"step": 1860
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 2.530446549391069,
|
| 1316 |
+
"grad_norm": 0.21239681541919708,
|
| 1317 |
+
"learning_rate": 7.397168381445812e-05,
|
| 1318 |
+
"loss": 0.0385,
|
| 1319 |
+
"step": 1870
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 2.543978349120433,
|
| 1323 |
+
"grad_norm": 0.17999856173992157,
|
| 1324 |
+
"learning_rate": 7.368095184513377e-05,
|
| 1325 |
+
"loss": 0.0383,
|
| 1326 |
+
"step": 1880
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 2.557510148849797,
|
| 1330 |
+
"grad_norm": 0.21162152290344238,
|
| 1331 |
+
"learning_rate": 7.338918399426005e-05,
|
| 1332 |
+
"loss": 0.0378,
|
| 1333 |
+
"step": 1890
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 2.571041948579161,
|
| 1337 |
+
"grad_norm": 0.2367720603942871,
|
| 1338 |
+
"learning_rate": 7.309639302470801e-05,
|
| 1339 |
+
"loss": 0.0469,
|
| 1340 |
+
"step": 1900
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 2.584573748308525,
|
| 1344 |
+
"grad_norm": 0.21758848428726196,
|
| 1345 |
+
"learning_rate": 7.280259174410312e-05,
|
| 1346 |
+
"loss": 0.0405,
|
| 1347 |
+
"step": 1910
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 2.598105548037889,
|
| 1351 |
+
"grad_norm": 0.1995786428451538,
|
| 1352 |
+
"learning_rate": 7.250779300426517e-05,
|
| 1353 |
+
"loss": 0.0399,
|
| 1354 |
+
"step": 1920
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 2.611637347767253,
|
| 1358 |
+
"grad_norm": 0.19054250419139862,
|
| 1359 |
+
"learning_rate": 7.22120097006461e-05,
|
| 1360 |
+
"loss": 0.0428,
|
| 1361 |
+
"step": 1930
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 2.6251691474966172,
|
| 1365 |
+
"grad_norm": 0.25428929924964905,
|
| 1366 |
+
"learning_rate": 7.191525477176577e-05,
|
| 1367 |
+
"loss": 0.0417,
|
| 1368 |
+
"step": 1940
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 2.6387009472259813,
|
| 1372 |
+
"grad_norm": 0.2825187146663666,
|
| 1373 |
+
"learning_rate": 7.161754119864616e-05,
|
| 1374 |
+
"loss": 0.0412,
|
| 1375 |
+
"step": 1950
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 2.6522327469553453,
|
| 1379 |
+
"grad_norm": 0.17652887105941772,
|
| 1380 |
+
"learning_rate": 7.131888200424339e-05,
|
| 1381 |
+
"loss": 0.0372,
|
| 1382 |
+
"step": 1960
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 2.6657645466847093,
|
| 1386 |
+
"grad_norm": 0.2632055878639221,
|
| 1387 |
+
"learning_rate": 7.101929025287816e-05,
|
| 1388 |
+
"loss": 0.0389,
|
| 1389 |
+
"step": 1970
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 2.6792963464140733,
|
| 1393 |
+
"grad_norm": 0.3633202910423279,
|
| 1394 |
+
"learning_rate": 7.071877904966423e-05,
|
| 1395 |
+
"loss": 0.0476,
|
| 1396 |
+
"step": 1980
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 2.6928281461434374,
|
| 1400 |
+
"grad_norm": 0.21020017564296722,
|
| 1401 |
+
"learning_rate": 7.04173615399351e-05,
|
| 1402 |
+
"loss": 0.041,
|
| 1403 |
+
"step": 1990
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 2.706359945872801,
|
| 1407 |
+
"grad_norm": 0.19143259525299072,
|
| 1408 |
+
"learning_rate": 7.011505090866913e-05,
|
| 1409 |
+
"loss": 0.0351,
|
| 1410 |
+
"step": 2000
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 2.719891745602165,
|
| 1414 |
+
"grad_norm": 0.2139059454202652,
|
| 1415 |
+
"learning_rate": 6.981186037991271e-05,
|
| 1416 |
+
"loss": 0.0388,
|
| 1417 |
+
"step": 2010
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 2.733423545331529,
|
| 1421 |
+
"grad_norm": 0.17334319651126862,
|
| 1422 |
+
"learning_rate": 6.950780321620174e-05,
|
| 1423 |
+
"loss": 0.0344,
|
| 1424 |
+
"step": 2020
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 2.746955345060893,
|
| 1428 |
+
"grad_norm": 0.2146347612142563,
|
| 1429 |
+
"learning_rate": 6.920289271798157e-05,
|
| 1430 |
+
"loss": 0.0449,
|
| 1431 |
+
"step": 2030
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 2.760487144790257,
|
| 1435 |
+
"grad_norm": 0.24033454060554504,
|
| 1436 |
+
"learning_rate": 6.889714222302517e-05,
|
| 1437 |
+
"loss": 0.0401,
|
| 1438 |
+
"step": 2040
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 2.774018944519621,
|
| 1442 |
+
"grad_norm": 0.19013287127017975,
|
| 1443 |
+
"learning_rate": 6.85905651058497e-05,
|
| 1444 |
+
"loss": 0.0375,
|
| 1445 |
+
"step": 2050
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 2.787550744248985,
|
| 1449 |
+
"grad_norm": 0.15871116518974304,
|
| 1450 |
+
"learning_rate": 6.82831747771314e-05,
|
| 1451 |
+
"loss": 0.0368,
|
| 1452 |
+
"step": 2060
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 2.801082543978349,
|
| 1456 |
+
"grad_norm": 0.17981383204460144,
|
| 1457 |
+
"learning_rate": 6.797498468311907e-05,
|
| 1458 |
+
"loss": 0.0388,
|
| 1459 |
+
"step": 2070
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 2.814614343707713,
|
| 1463 |
+
"grad_norm": 0.232260063290596,
|
| 1464 |
+
"learning_rate": 6.766600830504585e-05,
|
| 1465 |
+
"loss": 0.0368,
|
| 1466 |
+
"step": 2080
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 2.828146143437077,
|
| 1470 |
+
"grad_norm": 0.1559726893901825,
|
| 1471 |
+
"learning_rate": 6.735625915853942e-05,
|
| 1472 |
+
"loss": 0.037,
|
| 1473 |
+
"step": 2090
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 2.841677943166441,
|
| 1477 |
+
"grad_norm": 0.1821858137845993,
|
| 1478 |
+
"learning_rate": 6.70457507930309e-05,
|
| 1479 |
+
"loss": 0.035,
|
| 1480 |
+
"step": 2100
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 2.855209742895805,
|
| 1484 |
+
"grad_norm": 0.19013968110084534,
|
| 1485 |
+
"learning_rate": 6.673449679116215e-05,
|
| 1486 |
+
"loss": 0.0385,
|
| 1487 |
+
"step": 2110
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 2.8687415426251692,
|
| 1491 |
+
"grad_norm": 0.169216126203537,
|
| 1492 |
+
"learning_rate": 6.642251076819148e-05,
|
| 1493 |
+
"loss": 0.0322,
|
| 1494 |
+
"step": 2120
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 2.8822733423545333,
|
| 1498 |
+
"grad_norm": 0.16981689631938934,
|
| 1499 |
+
"learning_rate": 6.610980637139827e-05,
|
| 1500 |
+
"loss": 0.0351,
|
| 1501 |
+
"step": 2130
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 2.8958051420838973,
|
| 1505 |
+
"grad_norm": 0.14609754085540771,
|
| 1506 |
+
"learning_rate": 6.579639727948583e-05,
|
| 1507 |
+
"loss": 0.0388,
|
| 1508 |
+
"step": 2140
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 2.9093369418132613,
|
| 1512 |
+
"grad_norm": 0.1752074807882309,
|
| 1513 |
+
"learning_rate": 6.548229720198315e-05,
|
| 1514 |
+
"loss": 0.0381,
|
| 1515 |
+
"step": 2150
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 2.9228687415426253,
|
| 1519 |
+
"grad_norm": 0.20120173692703247,
|
| 1520 |
+
"learning_rate": 6.516751987864517e-05,
|
| 1521 |
+
"loss": 0.0354,
|
| 1522 |
+
"step": 2160
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 2.936400541271989,
|
| 1526 |
+
"grad_norm": 0.15326353907585144,
|
| 1527 |
+
"learning_rate": 6.485207907885175e-05,
|
| 1528 |
+
"loss": 0.0368,
|
| 1529 |
+
"step": 2170
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 2.949932341001353,
|
| 1533 |
+
"grad_norm": 0.22607815265655518,
|
| 1534 |
+
"learning_rate": 6.453598860100536e-05,
|
| 1535 |
+
"loss": 0.0348,
|
| 1536 |
+
"step": 2180
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 2.963464140730717,
|
| 1540 |
+
"grad_norm": 0.21510228514671326,
|
| 1541 |
+
"learning_rate": 6.421926227192749e-05,
|
| 1542 |
+
"loss": 0.0379,
|
| 1543 |
+
"step": 2190
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 2.976995940460081,
|
| 1547 |
+
"grad_norm": 0.16515401005744934,
|
| 1548 |
+
"learning_rate": 6.390191394625381e-05,
|
| 1549 |
+
"loss": 0.0424,
|
| 1550 |
+
"step": 2200
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 2.990527740189445,
|
| 1554 |
+
"grad_norm": 0.19766052067279816,
|
| 1555 |
+
"learning_rate": 6.358395750582817e-05,
|
| 1556 |
+
"loss": 0.0365,
|
| 1557 |
+
"step": 2210
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 3.004059539918809,
|
| 1561 |
+
"grad_norm": 0.13713599741458893,
|
| 1562 |
+
"learning_rate": 6.326540685909532e-05,
|
| 1563 |
+
"loss": 0.0333,
|
| 1564 |
+
"step": 2220
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 3.017591339648173,
|
| 1568 |
+
"grad_norm": 0.2925935685634613,
|
| 1569 |
+
"learning_rate": 6.294627594049249e-05,
|
| 1570 |
+
"loss": 0.0342,
|
| 1571 |
+
"step": 2230
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 3.031123139377537,
|
| 1575 |
+
"grad_norm": 0.2056179940700531,
|
| 1576 |
+
"learning_rate": 6.262657870983989e-05,
|
| 1577 |
+
"loss": 0.0341,
|
| 1578 |
+
"step": 2240
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 3.044654939106901,
|
| 1582 |
+
"grad_norm": 0.34839096665382385,
|
| 1583 |
+
"learning_rate": 6.230632915173009e-05,
|
| 1584 |
+
"loss": 0.0432,
|
| 1585 |
+
"step": 2250
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 3.058186738836265,
|
| 1589 |
+
"grad_norm": 0.24220605194568634,
|
| 1590 |
+
"learning_rate": 6.198554127491622e-05,
|
| 1591 |
+
"loss": 0.0348,
|
| 1592 |
+
"step": 2260
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 3.071718538565629,
|
| 1596 |
+
"grad_norm": 0.17123287916183472,
|
| 1597 |
+
"learning_rate": 6.166422911169923e-05,
|
| 1598 |
+
"loss": 0.0335,
|
| 1599 |
+
"step": 2270
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 3.085250338294993,
|
| 1603 |
+
"grad_norm": 0.2569161653518677,
|
| 1604 |
+
"learning_rate": 6.1342406717314e-05,
|
| 1605 |
+
"loss": 0.0384,
|
| 1606 |
+
"step": 2280
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 3.098782138024357,
|
| 1610 |
+
"grad_norm": 0.15369683504104614,
|
| 1611 |
+
"learning_rate": 6.102008816931466e-05,
|
| 1612 |
+
"loss": 0.0336,
|
| 1613 |
+
"step": 2290
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 3.1123139377537212,
|
| 1617 |
+
"grad_norm": 0.20889005064964294,
|
| 1618 |
+
"learning_rate": 6.069728756695866e-05,
|
| 1619 |
+
"loss": 0.034,
|
| 1620 |
+
"step": 2300
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 3.1258457374830853,
|
| 1624 |
+
"grad_norm": 0.2171514481306076,
|
| 1625 |
+
"learning_rate": 6.037401903059008e-05,
|
| 1626 |
+
"loss": 0.0395,
|
| 1627 |
+
"step": 2310
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 3.1393775372124493,
|
| 1631 |
+
"grad_norm": 0.2569717764854431,
|
| 1632 |
+
"learning_rate": 6.005029670102195e-05,
|
| 1633 |
+
"loss": 0.0315,
|
| 1634 |
+
"step": 2320
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 3.1529093369418133,
|
| 1638 |
+
"grad_norm": 0.17308172583580017,
|
| 1639 |
+
"learning_rate": 5.972613473891766e-05,
|
| 1640 |
+
"loss": 0.0394,
|
| 1641 |
+
"step": 2330
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 3.1664411366711773,
|
| 1645 |
+
"grad_norm": 0.18107521533966064,
|
| 1646 |
+
"learning_rate": 5.940154732417158e-05,
|
| 1647 |
+
"loss": 0.0365,
|
| 1648 |
+
"step": 2340
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 3.1799729364005414,
|
| 1652 |
+
"grad_norm": 0.2191227376461029,
|
| 1653 |
+
"learning_rate": 5.907654865528876e-05,
|
| 1654 |
+
"loss": 0.0374,
|
| 1655 |
+
"step": 2350
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 3.1935047361299054,
|
| 1659 |
+
"grad_norm": 0.22716890275478363,
|
| 1660 |
+
"learning_rate": 5.875115294876381e-05,
|
| 1661 |
+
"loss": 0.035,
|
| 1662 |
+
"step": 2360
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 3.2070365358592694,
|
| 1666 |
+
"grad_norm": 0.22536906599998474,
|
| 1667 |
+
"learning_rate": 5.842537443845908e-05,
|
| 1668 |
+
"loss": 0.0373,
|
| 1669 |
+
"step": 2370
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 3.2205683355886334,
|
| 1673 |
+
"grad_norm": 0.16970916092395782,
|
| 1674 |
+
"learning_rate": 5.809922737498198e-05,
|
| 1675 |
+
"loss": 0.0316,
|
| 1676 |
+
"step": 2380
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 3.2341001353179974,
|
| 1680 |
+
"grad_norm": 0.21783339977264404,
|
| 1681 |
+
"learning_rate": 5.777272602506165e-05,
|
| 1682 |
+
"loss": 0.0343,
|
| 1683 |
+
"step": 2390
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 3.2476319350473615,
|
| 1687 |
+
"grad_norm": 0.34707897901535034,
|
| 1688 |
+
"learning_rate": 5.744588467092483e-05,
|
| 1689 |
+
"loss": 0.038,
|
| 1690 |
+
"step": 2400
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 3.2611637347767255,
|
| 1694 |
+
"grad_norm": 0.2520757019519806,
|
| 1695 |
+
"learning_rate": 5.7118717609671194e-05,
|
| 1696 |
+
"loss": 0.035,
|
| 1697 |
+
"step": 2410
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 3.2746955345060895,
|
| 1701 |
+
"grad_norm": 0.3899773061275482,
|
| 1702 |
+
"learning_rate": 5.679123915264786e-05,
|
| 1703 |
+
"loss": 0.038,
|
| 1704 |
+
"step": 2420
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 3.2882273342354535,
|
| 1708 |
+
"grad_norm": 0.2649787664413452,
|
| 1709 |
+
"learning_rate": 5.646346362482342e-05,
|
| 1710 |
+
"loss": 0.0394,
|
| 1711 |
+
"step": 2430
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 3.301759133964817,
|
| 1715 |
+
"grad_norm": 0.15509116649627686,
|
| 1716 |
+
"learning_rate": 5.613540536416132e-05,
|
| 1717 |
+
"loss": 0.0346,
|
| 1718 |
+
"step": 2440
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 3.315290933694181,
|
| 1722 |
+
"grad_norm": 0.14768008887767792,
|
| 1723 |
+
"learning_rate": 5.5807078720992645e-05,
|
| 1724 |
+
"loss": 0.0335,
|
| 1725 |
+
"step": 2450
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 3.328822733423545,
|
| 1729 |
+
"grad_norm": 0.18915897607803345,
|
| 1730 |
+
"learning_rate": 5.547849805738836e-05,
|
| 1731 |
+
"loss": 0.0376,
|
| 1732 |
+
"step": 2460
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 3.342354533152909,
|
| 1736 |
+
"grad_norm": 0.17673052847385406,
|
| 1737 |
+
"learning_rate": 5.514967774653118e-05,
|
| 1738 |
+
"loss": 0.0347,
|
| 1739 |
+
"step": 2470
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 3.3558863328822732,
|
| 1743 |
+
"grad_norm": 0.18610958755016327,
|
| 1744 |
+
"learning_rate": 5.482063217208674e-05,
|
| 1745 |
+
"loss": 0.0324,
|
| 1746 |
+
"step": 2480
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 3.3694181326116373,
|
| 1750 |
+
"grad_norm": 0.18916338682174683,
|
| 1751 |
+
"learning_rate": 5.449137572757439e-05,
|
| 1752 |
+
"loss": 0.0363,
|
| 1753 |
+
"step": 2490
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 3.3829499323410013,
|
| 1757 |
+
"grad_norm": 0.1974906623363495,
|
| 1758 |
+
"learning_rate": 5.4161922815737696e-05,
|
| 1759 |
+
"loss": 0.0382,
|
| 1760 |
+
"step": 2500
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 3.3964817320703653,
|
| 1764 |
+
"grad_norm": 0.22988206148147583,
|
| 1765 |
+
"learning_rate": 5.3832287847914276e-05,
|
| 1766 |
+
"loss": 0.0364,
|
| 1767 |
+
"step": 2510
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 3.4100135317997293,
|
| 1771 |
+
"grad_norm": 0.19624315202236176,
|
| 1772 |
+
"learning_rate": 5.35024852434055e-05,
|
| 1773 |
+
"loss": 0.0367,
|
| 1774 |
+
"step": 2520
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 3.4235453315290933,
|
| 1778 |
+
"grad_norm": 0.2123284786939621,
|
| 1779 |
+
"learning_rate": 5.317252942884567e-05,
|
| 1780 |
+
"loss": 0.0341,
|
| 1781 |
+
"step": 2530
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 3.4370771312584574,
|
| 1785 |
+
"grad_norm": 0.21080602705478668,
|
| 1786 |
+
"learning_rate": 5.284243483757109e-05,
|
| 1787 |
+
"loss": 0.0327,
|
| 1788 |
+
"step": 2540
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 3.4506089309878214,
|
| 1792 |
+
"grad_norm": 0.16273650527000427,
|
| 1793 |
+
"learning_rate": 5.2512215908988484e-05,
|
| 1794 |
+
"loss": 0.0364,
|
| 1795 |
+
"step": 2550
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 3.4641407307171854,
|
| 1799 |
+
"grad_norm": 0.25207382440567017,
|
| 1800 |
+
"learning_rate": 5.218188708794357e-05,
|
| 1801 |
+
"loss": 0.0353,
|
| 1802 |
+
"step": 2560
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 3.4776725304465494,
|
| 1806 |
+
"grad_norm": 0.19061563909053802,
|
| 1807 |
+
"learning_rate": 5.18514628240891e-05,
|
| 1808 |
+
"loss": 0.0335,
|
| 1809 |
+
"step": 2570
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 3.4912043301759135,
|
| 1813 |
+
"grad_norm": 0.1704767942428589,
|
| 1814 |
+
"learning_rate": 5.1520957571252795e-05,
|
| 1815 |
+
"loss": 0.0378,
|
| 1816 |
+
"step": 2580
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 3.5047361299052775,
|
| 1820 |
+
"grad_norm": 0.15847642719745636,
|
| 1821 |
+
"learning_rate": 5.1190385786805106e-05,
|
| 1822 |
+
"loss": 0.0323,
|
| 1823 |
+
"step": 2590
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 3.5182679296346415,
|
| 1827 |
+
"grad_norm": 0.19395560026168823,
|
| 1828 |
+
"learning_rate": 5.085976193102677e-05,
|
| 1829 |
+
"loss": 0.0337,
|
| 1830 |
+
"step": 2600
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 3.5317997293640055,
|
| 1834 |
+
"grad_norm": 0.2881505787372589,
|
| 1835 |
+
"learning_rate": 5.052910046647634e-05,
|
| 1836 |
+
"loss": 0.0353,
|
| 1837 |
+
"step": 2610
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 3.5453315290933696,
|
| 1841 |
+
"grad_norm": 0.2774830162525177,
|
| 1842 |
+
"learning_rate": 5.0198415857357464e-05,
|
| 1843 |
+
"loss": 0.037,
|
| 1844 |
+
"step": 2620
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 3.558863328822733,
|
| 1848 |
+
"grad_norm": 0.24088598787784576,
|
| 1849 |
+
"learning_rate": 4.9867722568886223e-05,
|
| 1850 |
+
"loss": 0.0333,
|
| 1851 |
+
"step": 2630
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 3.572395128552097,
|
| 1855 |
+
"grad_norm": 0.2002314329147339,
|
| 1856 |
+
"learning_rate": 4.9537035066658314e-05,
|
| 1857 |
+
"loss": 0.0361,
|
| 1858 |
+
"step": 2640
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 3.585926928281461,
|
| 1862 |
+
"grad_norm": 0.3165316581726074,
|
| 1863 |
+
"learning_rate": 4.920636781601638e-05,
|
| 1864 |
+
"loss": 0.0345,
|
| 1865 |
+
"step": 2650
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 3.5994587280108252,
|
| 1869 |
+
"grad_norm": 0.2521885633468628,
|
| 1870 |
+
"learning_rate": 4.88757352814172e-05,
|
| 1871 |
+
"loss": 0.0389,
|
| 1872 |
+
"step": 2660
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 3.6129905277401893,
|
| 1876 |
+
"grad_norm": 0.18471549451351166,
|
| 1877 |
+
"learning_rate": 4.8545151925798924e-05,
|
| 1878 |
+
"loss": 0.0382,
|
| 1879 |
+
"step": 2670
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 3.6265223274695533,
|
| 1883 |
+
"grad_norm": 0.21062928438186646,
|
| 1884 |
+
"learning_rate": 4.821463220994848e-05,
|
| 1885 |
+
"loss": 0.0364,
|
| 1886 |
+
"step": 2680
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 3.6400541271989173,
|
| 1890 |
+
"grad_norm": 0.2001008540391922,
|
| 1891 |
+
"learning_rate": 4.788419059186895e-05,
|
| 1892 |
+
"loss": 0.0362,
|
| 1893 |
+
"step": 2690
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 3.6535859269282813,
|
| 1897 |
+
"grad_norm": 0.16223043203353882,
|
| 1898 |
+
"learning_rate": 4.7553841526147205e-05,
|
| 1899 |
+
"loss": 0.0338,
|
| 1900 |
+
"step": 2700
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 3.6671177266576453,
|
| 1904 |
+
"grad_norm": 0.22189919650554657,
|
| 1905 |
+
"learning_rate": 4.722359946332156e-05,
|
| 1906 |
+
"loss": 0.0369,
|
| 1907 |
+
"step": 2710
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 3.6806495263870094,
|
| 1911 |
+
"grad_norm": 0.22588606178760529,
|
| 1912 |
+
"learning_rate": 4.6893478849249654e-05,
|
| 1913 |
+
"loss": 0.0343,
|
| 1914 |
+
"step": 2720
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 3.6941813261163734,
|
| 1918 |
+
"grad_norm": 0.19541437923908234,
|
| 1919 |
+
"learning_rate": 4.656349412447664e-05,
|
| 1920 |
+
"loss": 0.0345,
|
| 1921 |
+
"step": 2730
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 3.7077131258457374,
|
| 1925 |
+
"grad_norm": 0.2428983449935913,
|
| 1926 |
+
"learning_rate": 4.623365972360337e-05,
|
| 1927 |
+
"loss": 0.0376,
|
| 1928 |
+
"step": 2740
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 3.7212449255751014,
|
| 1932 |
+
"grad_norm": 0.19307422637939453,
|
| 1933 |
+
"learning_rate": 4.590399007465503e-05,
|
| 1934 |
+
"loss": 0.0359,
|
| 1935 |
+
"step": 2750
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 3.7347767253044655,
|
| 1939 |
+
"grad_norm": 0.17713914811611176,
|
| 1940 |
+
"learning_rate": 4.557449959845005e-05,
|
| 1941 |
+
"loss": 0.0333,
|
| 1942 |
+
"step": 2760
|
| 1943 |
+
},
|
| 1944 |
+
{
|
| 1945 |
+
"epoch": 3.7483085250338295,
|
| 1946 |
+
"grad_norm": 0.16902363300323486,
|
| 1947 |
+
"learning_rate": 4.524520270796927e-05,
|
| 1948 |
+
"loss": 0.0308,
|
| 1949 |
+
"step": 2770
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 3.7618403247631935,
|
| 1953 |
+
"grad_norm": 0.162404403090477,
|
| 1954 |
+
"learning_rate": 4.491611380772545e-05,
|
| 1955 |
+
"loss": 0.0324,
|
| 1956 |
+
"step": 2780
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 3.7753721244925575,
|
| 1960 |
+
"grad_norm": 0.17956501245498657,
|
| 1961 |
+
"learning_rate": 4.458724729313318e-05,
|
| 1962 |
+
"loss": 0.031,
|
| 1963 |
+
"step": 2790
|
| 1964 |
+
},
|
| 1965 |
+
{
|
| 1966 |
+
"epoch": 3.7889039242219216,
|
| 1967 |
+
"grad_norm": 0.1946847140789032,
|
| 1968 |
+
"learning_rate": 4.42586175498792e-05,
|
| 1969 |
+
"loss": 0.0353,
|
| 1970 |
+
"step": 2800
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 3.8024357239512856,
|
| 1974 |
+
"grad_norm": 0.16929283738136292,
|
| 1975 |
+
"learning_rate": 4.3930238953293094e-05,
|
| 1976 |
+
"loss": 0.0341,
|
| 1977 |
+
"step": 2810
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 3.8159675236806496,
|
| 1981 |
+
"grad_norm": 0.2041793018579483,
|
| 1982 |
+
"learning_rate": 4.360212586771847e-05,
|
| 1983 |
+
"loss": 0.0328,
|
| 1984 |
+
"step": 2820
|
| 1985 |
+
},
|
| 1986 |
+
{
|
| 1987 |
+
"epoch": 3.8294993234100136,
|
| 1988 |
+
"grad_norm": 0.18276740610599518,
|
| 1989 |
+
"learning_rate": 4.327429264588463e-05,
|
| 1990 |
+
"loss": 0.0344,
|
| 1991 |
+
"step": 2830
|
| 1992 |
+
},
|
| 1993 |
+
{
|
| 1994 |
+
"epoch": 3.8430311231393777,
|
| 1995 |
+
"grad_norm": 0.15672671794891357,
|
| 1996 |
+
"learning_rate": 4.2946753628278725e-05,
|
| 1997 |
+
"loss": 0.0331,
|
| 1998 |
+
"step": 2840
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 3.8565629228687417,
|
| 2002 |
+
"grad_norm": 0.22230131924152374,
|
| 2003 |
+
"learning_rate": 4.2619523142518474e-05,
|
| 2004 |
+
"loss": 0.0335,
|
| 2005 |
+
"step": 2850
|
| 2006 |
+
},
|
| 2007 |
+
{
|
| 2008 |
+
"epoch": 3.8700947225981057,
|
| 2009 |
+
"grad_norm": 0.2010042667388916,
|
| 2010 |
+
"learning_rate": 4.229261550272539e-05,
|
| 2011 |
+
"loss": 0.033,
|
| 2012 |
+
"step": 2860
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 3.8836265223274697,
|
| 2016 |
+
"grad_norm": 0.15846724808216095,
|
| 2017 |
+
"learning_rate": 4.196604500889868e-05,
|
| 2018 |
+
"loss": 0.0344,
|
| 2019 |
+
"step": 2870
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 3.8971583220568338,
|
| 2023 |
+
"grad_norm": 0.2011309415102005,
|
| 2024 |
+
"learning_rate": 4.163982594628969e-05,
|
| 2025 |
+
"loss": 0.0328,
|
| 2026 |
+
"step": 2880
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 3.910690121786198,
|
| 2030 |
+
"grad_norm": 0.17424169182777405,
|
| 2031 |
+
"learning_rate": 4.131397258477702e-05,
|
| 2032 |
+
"loss": 0.0317,
|
| 2033 |
+
"step": 2890
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 3.924221921515562,
|
| 2037 |
+
"grad_norm": 0.1555200070142746,
|
| 2038 |
+
"learning_rate": 4.0988499178242315e-05,
|
| 2039 |
+
"loss": 0.0332,
|
| 2040 |
+
"step": 2900
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 3.937753721244926,
|
| 2044 |
+
"grad_norm": 0.20862902700901031,
|
| 2045 |
+
"learning_rate": 4.066341996394678e-05,
|
| 2046 |
+
"loss": 0.0381,
|
| 2047 |
+
"step": 2910
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 3.9512855209742894,
|
| 2051 |
+
"grad_norm": 0.1553603708744049,
|
| 2052 |
+
"learning_rate": 4.033874916190833e-05,
|
| 2053 |
+
"loss": 0.0349,
|
| 2054 |
+
"step": 2920
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 3.9648173207036534,
|
| 2058 |
+
"grad_norm": 0.13140486180782318,
|
| 2059 |
+
"learning_rate": 4.001450097427966e-05,
|
| 2060 |
+
"loss": 0.0322,
|
| 2061 |
+
"step": 2930
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 3.9783491204330175,
|
| 2065 |
+
"grad_norm": 0.17590387165546417,
|
| 2066 |
+
"learning_rate": 3.9690689584726894e-05,
|
| 2067 |
+
"loss": 0.0336,
|
| 2068 |
+
"step": 2940
|
| 2069 |
+
},
|
| 2070 |
+
{
|
| 2071 |
+
"epoch": 3.9918809201623815,
|
| 2072 |
+
"grad_norm": 0.16788561642169952,
|
| 2073 |
+
"learning_rate": 3.936732915780923e-05,
|
| 2074 |
+
"loss": 0.0305,
|
| 2075 |
+
"step": 2950
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 4.005412719891746,
|
| 2079 |
+
"grad_norm": 0.1186370700597763,
|
| 2080 |
+
"learning_rate": 3.904443383835929e-05,
|
| 2081 |
+
"loss": 0.0318,
|
| 2082 |
+
"step": 2960
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"epoch": 4.01894451962111,
|
| 2086 |
+
"grad_norm": 0.16949886083602905,
|
| 2087 |
+
"learning_rate": 3.872201775086437e-05,
|
| 2088 |
+
"loss": 0.0336,
|
| 2089 |
+
"step": 2970
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"epoch": 4.032476319350474,
|
| 2093 |
+
"grad_norm": 0.18508851528167725,
|
| 2094 |
+
"learning_rate": 3.8400094998848616e-05,
|
| 2095 |
+
"loss": 0.034,
|
| 2096 |
+
"step": 2980
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 4.046008119079838,
|
| 2100 |
+
"grad_norm": 0.26010963320732117,
|
| 2101 |
+
"learning_rate": 3.807867966425611e-05,
|
| 2102 |
+
"loss": 0.0311,
|
| 2103 |
+
"step": 2990
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 4.059539918809202,
|
| 2107 |
+
"grad_norm": 0.23572185635566711,
|
| 2108 |
+
"learning_rate": 3.775778580683481e-05,
|
| 2109 |
+
"loss": 0.0321,
|
| 2110 |
+
"step": 3000
|
| 2111 |
+
},
|
| 2112 |
+
{
|
| 2113 |
+
"epoch": 4.073071718538565,
|
| 2114 |
+
"grad_norm": 0.15540273487567902,
|
| 2115 |
+
"learning_rate": 3.743742746352156e-05,
|
| 2116 |
+
"loss": 0.0325,
|
| 2117 |
+
"step": 3010
|
| 2118 |
+
},
|
| 2119 |
+
{
|
| 2120 |
+
"epoch": 4.086603518267929,
|
| 2121 |
+
"grad_norm": 0.1744704693555832,
|
| 2122 |
+
"learning_rate": 3.711761864782817e-05,
|
| 2123 |
+
"loss": 0.036,
|
| 2124 |
+
"step": 3020
|
| 2125 |
+
},
|
| 2126 |
+
{
|
| 2127 |
+
"epoch": 4.100135317997293,
|
| 2128 |
+
"grad_norm": 0.31342482566833496,
|
| 2129 |
+
"learning_rate": 3.679837334922825e-05,
|
| 2130 |
+
"loss": 0.0316,
|
| 2131 |
+
"step": 3030
|
| 2132 |
+
},
|
| 2133 |
+
{
|
| 2134 |
+
"epoch": 4.113667117726657,
|
| 2135 |
+
"grad_norm": 0.14255890250205994,
|
| 2136 |
+
"learning_rate": 3.647970553254538e-05,
|
| 2137 |
+
"loss": 0.0336,
|
| 2138 |
+
"step": 3040
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 4.127198917456021,
|
| 2142 |
+
"grad_norm": 0.20257729291915894,
|
| 2143 |
+
"learning_rate": 3.61616291373422e-05,
|
| 2144 |
+
"loss": 0.0359,
|
| 2145 |
+
"step": 3050
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 4.140730717185385,
|
| 2149 |
+
"grad_norm": 0.15252985060214996,
|
| 2150 |
+
"learning_rate": 3.584415807731065e-05,
|
| 2151 |
+
"loss": 0.031,
|
| 2152 |
+
"step": 3060
|
| 2153 |
+
},
|
| 2154 |
+
{
|
| 2155 |
+
"epoch": 4.154262516914749,
|
| 2156 |
+
"grad_norm": 0.14593230187892914,
|
| 2157 |
+
"learning_rate": 3.552730623966337e-05,
|
| 2158 |
+
"loss": 0.033,
|
| 2159 |
+
"step": 3070
|
| 2160 |
+
},
|
| 2161 |
+
{
|
| 2162 |
+
"epoch": 4.167794316644113,
|
| 2163 |
+
"grad_norm": 0.19335772097110748,
|
| 2164 |
+
"learning_rate": 3.521108748452617e-05,
|
| 2165 |
+
"loss": 0.0309,
|
| 2166 |
+
"step": 3080
|
| 2167 |
+
},
|
| 2168 |
+
{
|
| 2169 |
+
"epoch": 4.181326116373477,
|
| 2170 |
+
"grad_norm": 0.1618116796016693,
|
| 2171 |
+
"learning_rate": 3.489551564433186e-05,
|
| 2172 |
+
"loss": 0.0333,
|
| 2173 |
+
"step": 3090
|
| 2174 |
+
},
|
| 2175 |
+
{
|
| 2176 |
+
"epoch": 4.194857916102841,
|
| 2177 |
+
"grad_norm": 0.16830849647521973,
|
| 2178 |
+
"learning_rate": 3.4580604523215006e-05,
|
| 2179 |
+
"loss": 0.0314,
|
| 2180 |
+
"step": 3100
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 4.2083897158322054,
|
| 2184 |
+
"grad_norm": 0.15059009194374084,
|
| 2185 |
+
"learning_rate": 3.4266367896408216e-05,
|
| 2186 |
+
"loss": 0.0333,
|
| 2187 |
+
"step": 3110
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 4.2219215155615695,
|
| 2191 |
+
"grad_norm": 0.1731686294078827,
|
| 2192 |
+
"learning_rate": 3.3952819509639534e-05,
|
| 2193 |
+
"loss": 0.0328,
|
| 2194 |
+
"step": 3120
|
| 2195 |
+
},
|
| 2196 |
+
{
|
| 2197 |
+
"epoch": 4.2354533152909335,
|
| 2198 |
+
"grad_norm": 0.13689175248146057,
|
| 2199 |
+
"learning_rate": 3.3639973078531165e-05,
|
| 2200 |
+
"loss": 0.0322,
|
| 2201 |
+
"step": 3130
|
| 2202 |
+
},
|
| 2203 |
+
{
|
| 2204 |
+
"epoch": 4.2489851150202975,
|
| 2205 |
+
"grad_norm": 0.1830020397901535,
|
| 2206 |
+
"learning_rate": 3.332784228799947e-05,
|
| 2207 |
+
"loss": 0.0349,
|
| 2208 |
+
"step": 3140
|
| 2209 |
+
},
|
| 2210 |
+
{
|
| 2211 |
+
"epoch": 4.2625169147496615,
|
| 2212 |
+
"grad_norm": 0.1898087114095688,
|
| 2213 |
+
"learning_rate": 3.301644079165638e-05,
|
| 2214 |
+
"loss": 0.0311,
|
| 2215 |
+
"step": 3150
|
| 2216 |
+
},
|
| 2217 |
+
{
|
| 2218 |
+
"epoch": 4.276048714479026,
|
| 2219 |
+
"grad_norm": 0.14833958446979523,
|
| 2220 |
+
"learning_rate": 3.27057822112122e-05,
|
| 2221 |
+
"loss": 0.0344,
|
| 2222 |
+
"step": 3160
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 4.28958051420839,
|
| 2226 |
+
"grad_norm": 0.232608363032341,
|
| 2227 |
+
"learning_rate": 3.239588013587958e-05,
|
| 2228 |
+
"loss": 0.0325,
|
| 2229 |
+
"step": 3170
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 4.303112313937754,
|
| 2233 |
+
"grad_norm": 0.31661397218704224,
|
| 2234 |
+
"learning_rate": 3.208674812177926e-05,
|
| 2235 |
+
"loss": 0.0314,
|
| 2236 |
+
"step": 3180
|
| 2237 |
+
},
|
| 2238 |
+
{
|
| 2239 |
+
"epoch": 4.316644113667118,
|
| 2240 |
+
"grad_norm": 0.18021564185619354,
|
| 2241 |
+
"learning_rate": 3.177839969134698e-05,
|
| 2242 |
+
"loss": 0.0332,
|
| 2243 |
+
"step": 3190
|
| 2244 |
+
},
|
| 2245 |
+
{
|
| 2246 |
+
"epoch": 4.330175913396482,
|
| 2247 |
+
"grad_norm": 0.16829122602939606,
|
| 2248 |
+
"learning_rate": 3.1470848332742e-05,
|
| 2249 |
+
"loss": 0.0327,
|
| 2250 |
+
"step": 3200
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 4.343707713125846,
|
| 2254 |
+
"grad_norm": 0.209175243973732,
|
| 2255 |
+
"learning_rate": 3.116410749925708e-05,
|
| 2256 |
+
"loss": 0.0287,
|
| 2257 |
+
"step": 3210
|
| 2258 |
+
},
|
| 2259 |
+
{
|
| 2260 |
+
"epoch": 4.35723951285521,
|
| 2261 |
+
"grad_norm": 0.14701929688453674,
|
| 2262 |
+
"learning_rate": 3.085819060872995e-05,
|
| 2263 |
+
"loss": 0.0321,
|
| 2264 |
+
"step": 3220
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 4.370771312584574,
|
| 2268 |
+
"grad_norm": 0.16794687509536743,
|
| 2269 |
+
"learning_rate": 3.055311104295648e-05,
|
| 2270 |
+
"loss": 0.033,
|
| 2271 |
+
"step": 3230
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 4.384303112313938,
|
| 2275 |
+
"grad_norm": 0.1298779845237732,
|
| 2276 |
+
"learning_rate": 3.024888214710517e-05,
|
| 2277 |
+
"loss": 0.0314,
|
| 2278 |
+
"step": 3240
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 4.397834912043302,
|
| 2282 |
+
"grad_norm": 0.1175541877746582,
|
| 2283 |
+
"learning_rate": 2.994551722913349e-05,
|
| 2284 |
+
"loss": 0.033,
|
| 2285 |
+
"step": 3250
|
| 2286 |
+
},
|
| 2287 |
+
{
|
| 2288 |
+
"epoch": 4.411366711772666,
|
| 2289 |
+
"grad_norm": 0.29746726155281067,
|
| 2290 |
+
"learning_rate": 2.9643029559205727e-05,
|
| 2291 |
+
"loss": 0.0358,
|
| 2292 |
+
"step": 3260
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 4.42489851150203,
|
| 2296 |
+
"grad_norm": 0.15811285376548767,
|
| 2297 |
+
"learning_rate": 2.934143236911248e-05,
|
| 2298 |
+
"loss": 0.0314,
|
| 2299 |
+
"step": 3270
|
| 2300 |
+
},
|
| 2301 |
+
{
|
| 2302 |
+
"epoch": 4.438430311231394,
|
| 2303 |
+
"grad_norm": 0.1696750819683075,
|
| 2304 |
+
"learning_rate": 2.90407388516919e-05,
|
| 2305 |
+
"loss": 0.0342,
|
| 2306 |
+
"step": 3280
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 4.451962110960758,
|
| 2310 |
+
"grad_norm": 0.13138838112354279,
|
| 2311 |
+
"learning_rate": 2.8740962160252495e-05,
|
| 2312 |
+
"loss": 0.031,
|
| 2313 |
+
"step": 3290
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 4.465493910690122,
|
| 2317 |
+
"grad_norm": 0.1903950721025467,
|
| 2318 |
+
"learning_rate": 2.844211540799797e-05,
|
| 2319 |
+
"loss": 0.0305,
|
| 2320 |
+
"step": 3300
|
| 2321 |
+
},
|
| 2322 |
+
{
|
| 2323 |
+
"epoch": 4.479025710419486,
|
| 2324 |
+
"grad_norm": 0.2284688502550125,
|
| 2325 |
+
"learning_rate": 2.8144211667453368e-05,
|
| 2326 |
+
"loss": 0.0303,
|
| 2327 |
+
"step": 3310
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 4.49255751014885,
|
| 2331 |
+
"grad_norm": 0.2390712946653366,
|
| 2332 |
+
"learning_rate": 2.7847263969893344e-05,
|
| 2333 |
+
"loss": 0.0299,
|
| 2334 |
+
"step": 3320
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 4.506089309878214,
|
| 2338 |
+
"grad_norm": 0.1737690418958664,
|
| 2339 |
+
"learning_rate": 2.7551285304772206e-05,
|
| 2340 |
+
"loss": 0.0296,
|
| 2341 |
+
"step": 3330
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 4.519621109607578,
|
| 2345 |
+
"grad_norm": 0.14434020221233368,
|
| 2346 |
+
"learning_rate": 2.7256288619155567e-05,
|
| 2347 |
+
"loss": 0.0341,
|
| 2348 |
+
"step": 3340
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 4.533152909336942,
|
| 2352 |
+
"grad_norm": 0.19870182871818542,
|
| 2353 |
+
"learning_rate": 2.6962286817154158e-05,
|
| 2354 |
+
"loss": 0.0349,
|
| 2355 |
+
"step": 3350
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 4.546684709066306,
|
| 2359 |
+
"grad_norm": 0.12394054234027863,
|
| 2360 |
+
"learning_rate": 2.6669292759359166e-05,
|
| 2361 |
+
"loss": 0.0303,
|
| 2362 |
+
"step": 3360
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 4.56021650879567,
|
| 2366 |
+
"grad_norm": 0.13529327511787415,
|
| 2367 |
+
"learning_rate": 2.637731926227993e-05,
|
| 2368 |
+
"loss": 0.033,
|
| 2369 |
+
"step": 3370
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 4.573748308525034,
|
| 2373 |
+
"grad_norm": 0.23153837025165558,
|
| 2374 |
+
"learning_rate": 2.6086379097783033e-05,
|
| 2375 |
+
"loss": 0.0314,
|
| 2376 |
+
"step": 3380
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 4.587280108254398,
|
| 2380 |
+
"grad_norm": 0.1868782937526703,
|
| 2381 |
+
"learning_rate": 2.579648499253377e-05,
|
| 2382 |
+
"loss": 0.03,
|
| 2383 |
+
"step": 3390
|
| 2384 |
+
},
|
| 2385 |
+
{
|
| 2386 |
+
"epoch": 4.600811907983762,
|
| 2387 |
+
"grad_norm": 0.20445260405540466,
|
| 2388 |
+
"learning_rate": 2.5507649627439466e-05,
|
| 2389 |
+
"loss": 0.0343,
|
| 2390 |
+
"step": 3400
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 4.614343707713126,
|
| 2394 |
+
"grad_norm": 0.1350487917661667,
|
| 2395 |
+
"learning_rate": 2.5219885637094653e-05,
|
| 2396 |
+
"loss": 0.0295,
|
| 2397 |
+
"step": 3410
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 4.62787550744249,
|
| 2401 |
+
"grad_norm": 0.18250331282615662,
|
| 2402 |
+
"learning_rate": 2.4933205609228533e-05,
|
| 2403 |
+
"loss": 0.0317,
|
| 2404 |
+
"step": 3420
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"epoch": 4.641407307171854,
|
| 2408 |
+
"grad_norm": 0.36254093050956726,
|
| 2409 |
+
"learning_rate": 2.464762208415419e-05,
|
| 2410 |
+
"loss": 0.0318,
|
| 2411 |
+
"step": 3430
|
| 2412 |
+
},
|
| 2413 |
+
{
|
| 2414 |
+
"epoch": 4.654939106901217,
|
| 2415 |
+
"grad_norm": 0.17919518053531647,
|
| 2416 |
+
"learning_rate": 2.4363147554220213e-05,
|
| 2417 |
+
"loss": 0.0345,
|
| 2418 |
+
"step": 3440
|
| 2419 |
+
},
|
| 2420 |
+
{
|
| 2421 |
+
"epoch": 4.668470906630581,
|
| 2422 |
+
"grad_norm": 0.15475888550281525,
|
| 2423 |
+
"learning_rate": 2.407979446326411e-05,
|
| 2424 |
+
"loss": 0.0288,
|
| 2425 |
+
"step": 3450
|
| 2426 |
+
},
|
| 2427 |
+
{
|
| 2428 |
+
"epoch": 4.682002706359945,
|
| 2429 |
+
"grad_norm": 0.18196256458759308,
|
| 2430 |
+
"learning_rate": 2.379757520606799e-05,
|
| 2431 |
+
"loss": 0.0341,
|
| 2432 |
+
"step": 3460
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 4.6955345060893094,
|
| 2436 |
+
"grad_norm": 0.11349283903837204,
|
| 2437 |
+
"learning_rate": 2.3516502127816455e-05,
|
| 2438 |
+
"loss": 0.0275,
|
| 2439 |
+
"step": 3470
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 4.7090663058186735,
|
| 2443 |
+
"grad_norm": 0.12871690094470978,
|
| 2444 |
+
"learning_rate": 2.323658752355647e-05,
|
| 2445 |
+
"loss": 0.0294,
|
| 2446 |
+
"step": 3480
|
| 2447 |
+
},
|
| 2448 |
+
{
|
| 2449 |
+
"epoch": 4.7225981055480375,
|
| 2450 |
+
"grad_norm": 0.136393204331398,
|
| 2451 |
+
"learning_rate": 2.2957843637659654e-05,
|
| 2452 |
+
"loss": 0.0305,
|
| 2453 |
+
"step": 3490
|
| 2454 |
+
},
|
| 2455 |
+
{
|
| 2456 |
+
"epoch": 4.7361299052774015,
|
| 2457 |
+
"grad_norm": 0.1588805764913559,
|
| 2458 |
+
"learning_rate": 2.2680282663286552e-05,
|
| 2459 |
+
"loss": 0.0328,
|
| 2460 |
+
"step": 3500
|
| 2461 |
+
},
|
| 2462 |
+
{
|
| 2463 |
+
"epoch": 4.7496617050067655,
|
| 2464 |
+
"grad_norm": 0.16840600967407227,
|
| 2465 |
+
"learning_rate": 2.2403916741853364e-05,
|
| 2466 |
+
"loss": 0.0306,
|
| 2467 |
+
"step": 3510
|
| 2468 |
+
},
|
| 2469 |
+
{
|
| 2470 |
+
"epoch": 4.76319350473613,
|
| 2471 |
+
"grad_norm": 0.10826493799686432,
|
| 2472 |
+
"learning_rate": 2.2128757962500817e-05,
|
| 2473 |
+
"loss": 0.0268,
|
| 2474 |
+
"step": 3520
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 4.776725304465494,
|
| 2478 |
+
"grad_norm": 0.223701611161232,
|
| 2479 |
+
"learning_rate": 2.1854818361565275e-05,
|
| 2480 |
+
"loss": 0.0319,
|
| 2481 |
+
"step": 3530
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 4.790257104194858,
|
| 2485 |
+
"grad_norm": 0.19489619135856628,
|
| 2486 |
+
"learning_rate": 2.1582109922052364e-05,
|
| 2487 |
+
"loss": 0.0321,
|
| 2488 |
+
"step": 3540
|
| 2489 |
+
},
|
| 2490 |
+
{
|
| 2491 |
+
"epoch": 4.803788903924222,
|
| 2492 |
+
"grad_norm": 0.16030816733837128,
|
| 2493 |
+
"learning_rate": 2.1310644573112635e-05,
|
| 2494 |
+
"loss": 0.0306,
|
| 2495 |
+
"step": 3550
|
| 2496 |
+
},
|
| 2497 |
+
{
|
| 2498 |
+
"epoch": 4.817320703653586,
|
| 2499 |
+
"grad_norm": 0.16815383732318878,
|
| 2500 |
+
"learning_rate": 2.1040434189519924e-05,
|
| 2501 |
+
"loss": 0.033,
|
| 2502 |
+
"step": 3560
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"epoch": 4.83085250338295,
|
| 2506 |
+
"grad_norm": 0.2008376121520996,
|
| 2507 |
+
"learning_rate": 2.0771490591151733e-05,
|
| 2508 |
+
"loss": 0.0323,
|
| 2509 |
+
"step": 3570
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 4.844384303112314,
|
| 2513 |
+
"grad_norm": 0.12762121856212616,
|
| 2514 |
+
"learning_rate": 2.0503825542472317e-05,
|
| 2515 |
+
"loss": 0.0361,
|
| 2516 |
+
"step": 3580
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 4.857916102841678,
|
| 2520 |
+
"grad_norm": 0.1627681702375412,
|
| 2521 |
+
"learning_rate": 2.023745075201805e-05,
|
| 2522 |
+
"loss": 0.0281,
|
| 2523 |
+
"step": 3590
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 4.871447902571042,
|
| 2527 |
+
"grad_norm": 0.1745183914899826,
|
| 2528 |
+
"learning_rate": 1.9972377871885157e-05,
|
| 2529 |
+
"loss": 0.0302,
|
| 2530 |
+
"step": 3600
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 4.884979702300406,
|
| 2534 |
+
"grad_norm": 0.1831778585910797,
|
| 2535 |
+
"learning_rate": 1.970861849722017e-05,
|
| 2536 |
+
"loss": 0.0297,
|
| 2537 |
+
"step": 3610
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 4.89851150202977,
|
| 2541 |
+
"grad_norm": 0.2168680876493454,
|
| 2542 |
+
"learning_rate": 1.9446184165712587e-05,
|
| 2543 |
+
"loss": 0.0304,
|
| 2544 |
+
"step": 3620
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 4.912043301759134,
|
| 2548 |
+
"grad_norm": 0.1703914850950241,
|
| 2549 |
+
"learning_rate": 1.9185086357090214e-05,
|
| 2550 |
+
"loss": 0.0301,
|
| 2551 |
+
"step": 3630
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 4.925575101488498,
|
| 2555 |
+
"grad_norm": 0.22563602030277252,
|
| 2556 |
+
"learning_rate": 1.8925336492617057e-05,
|
| 2557 |
+
"loss": 0.0294,
|
| 2558 |
+
"step": 3640
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 4.939106901217862,
|
| 2562 |
+
"grad_norm": 0.16377097368240356,
|
| 2563 |
+
"learning_rate": 1.8666945934593666e-05,
|
| 2564 |
+
"loss": 0.0334,
|
| 2565 |
+
"step": 3650
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 4.952638700947226,
|
| 2569 |
+
"grad_norm": 0.11018290370702744,
|
| 2570 |
+
"learning_rate": 1.8409925985860126e-05,
|
| 2571 |
+
"loss": 0.0252,
|
| 2572 |
+
"step": 3660
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 4.96617050067659,
|
| 2576 |
+
"grad_norm": 0.15742255747318268,
|
| 2577 |
+
"learning_rate": 1.8154287889301603e-05,
|
| 2578 |
+
"loss": 0.0295,
|
| 2579 |
+
"step": 3670
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 4.979702300405954,
|
| 2583 |
+
"grad_norm": 0.1602838933467865,
|
| 2584 |
+
"learning_rate": 1.7900042827356612e-05,
|
| 2585 |
+
"loss": 0.0326,
|
| 2586 |
+
"step": 3680
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 4.993234100135318,
|
| 2590 |
+
"grad_norm": 0.2006329745054245,
|
| 2591 |
+
"learning_rate": 1.76472019215278e-05,
|
| 2592 |
+
"loss": 0.0306,
|
| 2593 |
+
"step": 3690
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 5.006765899864682,
|
| 2597 |
+
"grad_norm": 0.2930007874965668,
|
| 2598 |
+
"learning_rate": 1.739577623189545e-05,
|
| 2599 |
+
"loss": 0.0293,
|
| 2600 |
+
"step": 3700
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 5.020297699594046,
|
| 2604 |
+
"grad_norm": 0.3344183564186096,
|
| 2605 |
+
"learning_rate": 1.7145776756633768e-05,
|
| 2606 |
+
"loss": 0.0331,
|
| 2607 |
+
"step": 3710
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 5.03382949932341,
|
| 2611 |
+
"grad_norm": 0.25694382190704346,
|
| 2612 |
+
"learning_rate": 1.6897214431529646e-05,
|
| 2613 |
+
"loss": 0.0325,
|
| 2614 |
+
"step": 3720
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 5.047361299052774,
|
| 2618 |
+
"grad_norm": 0.11163618415594101,
|
| 2619 |
+
"learning_rate": 1.6650100129504475e-05,
|
| 2620 |
+
"loss": 0.0298,
|
| 2621 |
+
"step": 3730
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 5.060893098782138,
|
| 2625 |
+
"grad_norm": 0.3061096668243408,
|
| 2626 |
+
"learning_rate": 1.6404444660138335e-05,
|
| 2627 |
+
"loss": 0.0313,
|
| 2628 |
+
"step": 3740
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 5.074424898511502,
|
| 2632 |
+
"grad_norm": 0.16697120666503906,
|
| 2633 |
+
"learning_rate": 1.616025876919725e-05,
|
| 2634 |
+
"loss": 0.0293,
|
| 2635 |
+
"step": 3750
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 5.087956698240866,
|
| 2639 |
+
"grad_norm": 0.11484449356794357,
|
| 2640 |
+
"learning_rate": 1.5917553138163172e-05,
|
| 2641 |
+
"loss": 0.0284,
|
| 2642 |
+
"step": 3760
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 5.10148849797023,
|
| 2646 |
+
"grad_norm": 0.3616161048412323,
|
| 2647 |
+
"learning_rate": 1.5676338383766632e-05,
|
| 2648 |
+
"loss": 0.0318,
|
| 2649 |
+
"step": 3770
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 5.115020297699594,
|
| 2653 |
+
"grad_norm": 0.18025486171245575,
|
| 2654 |
+
"learning_rate": 1.5436625057522447e-05,
|
| 2655 |
+
"loss": 0.0316,
|
| 2656 |
+
"step": 3780
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"epoch": 5.128552097428958,
|
| 2660 |
+
"grad_norm": 0.14067353308200836,
|
| 2661 |
+
"learning_rate": 1.519842364526804e-05,
|
| 2662 |
+
"loss": 0.0322,
|
| 2663 |
+
"step": 3790
|
| 2664 |
+
},
|
| 2665 |
+
{
|
| 2666 |
+
"epoch": 5.142083897158322,
|
| 2667 |
+
"grad_norm": 0.18656225502490997,
|
| 2668 |
+
"learning_rate": 1.4961744566704855e-05,
|
| 2669 |
+
"loss": 0.0297,
|
| 2670 |
+
"step": 3800
|
| 2671 |
+
},
|
| 2672 |
+
{
|
| 2673 |
+
"epoch": 5.155615696887686,
|
| 2674 |
+
"grad_norm": 0.14053630828857422,
|
| 2675 |
+
"learning_rate": 1.4726598174942551e-05,
|
| 2676 |
+
"loss": 0.0275,
|
| 2677 |
+
"step": 3810
|
| 2678 |
+
},
|
| 2679 |
+
{
|
| 2680 |
+
"epoch": 5.16914749661705,
|
| 2681 |
+
"grad_norm": 0.1318153440952301,
|
| 2682 |
+
"learning_rate": 1.4492994756046035e-05,
|
| 2683 |
+
"loss": 0.0307,
|
| 2684 |
+
"step": 3820
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 5.182679296346414,
|
| 2688 |
+
"grad_norm": 0.21349585056304932,
|
| 2689 |
+
"learning_rate": 1.4260944528585645e-05,
|
| 2690 |
+
"loss": 0.0313,
|
| 2691 |
+
"step": 3830
|
| 2692 |
+
},
|
| 2693 |
+
{
|
| 2694 |
+
"epoch": 5.196211096075778,
|
| 2695 |
+
"grad_norm": 0.143988698720932,
|
| 2696 |
+
"learning_rate": 1.4030457643190048e-05,
|
| 2697 |
+
"loss": 0.0299,
|
| 2698 |
+
"step": 3840
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"epoch": 5.209742895805142,
|
| 2702 |
+
"grad_norm": 0.1565936654806137,
|
| 2703 |
+
"learning_rate": 1.3801544182102311e-05,
|
| 2704 |
+
"loss": 0.0312,
|
| 2705 |
+
"step": 3850
|
| 2706 |
+
},
|
| 2707 |
+
{
|
| 2708 |
+
"epoch": 5.223274695534506,
|
| 2709 |
+
"grad_norm": 0.18127118051052094,
|
| 2710 |
+
"learning_rate": 1.3574214158738763e-05,
|
| 2711 |
+
"loss": 0.0301,
|
| 2712 |
+
"step": 3860
|
| 2713 |
+
},
|
| 2714 |
+
{
|
| 2715 |
+
"epoch": 5.23680649526387,
|
| 2716 |
+
"grad_norm": 0.12334998697042465,
|
| 2717 |
+
"learning_rate": 1.3348477517251101e-05,
|
| 2718 |
+
"loss": 0.0271,
|
| 2719 |
+
"step": 3870
|
| 2720 |
+
},
|
| 2721 |
+
{
|
| 2722 |
+
"epoch": 5.2503382949932345,
|
| 2723 |
+
"grad_norm": 0.16843830049037933,
|
| 2724 |
+
"learning_rate": 1.312434413209131e-05,
|
| 2725 |
+
"loss": 0.0266,
|
| 2726 |
+
"step": 3880
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 5.2638700947225985,
|
| 2730 |
+
"grad_norm": 0.1459604948759079,
|
| 2731 |
+
"learning_rate": 1.2901823807579727e-05,
|
| 2732 |
+
"loss": 0.0281,
|
| 2733 |
+
"step": 3890
|
| 2734 |
+
},
|
| 2735 |
+
{
|
| 2736 |
+
"epoch": 5.2774018944519625,
|
| 2737 |
+
"grad_norm": 0.25701919198036194,
|
| 2738 |
+
"learning_rate": 1.2680926277476245e-05,
|
| 2739 |
+
"loss": 0.0264,
|
| 2740 |
+
"step": 3900
|
| 2741 |
+
},
|
| 2742 |
+
{
|
| 2743 |
+
"epoch": 5.2909336941813265,
|
| 2744 |
+
"grad_norm": 0.1085154339671135,
|
| 2745 |
+
"learning_rate": 1.2461661204554397e-05,
|
| 2746 |
+
"loss": 0.0297,
|
| 2747 |
+
"step": 3910
|
| 2748 |
+
},
|
| 2749 |
+
{
|
| 2750 |
+
"epoch": 5.3044654939106906,
|
| 2751 |
+
"grad_norm": 0.09449544548988342,
|
| 2752 |
+
"learning_rate": 1.2244038180178835e-05,
|
| 2753 |
+
"loss": 0.0277,
|
| 2754 |
+
"step": 3920
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 5.317997293640055,
|
| 2758 |
+
"grad_norm": 0.10733329504728317,
|
| 2759 |
+
"learning_rate": 1.2028066723885612e-05,
|
| 2760 |
+
"loss": 0.0291,
|
| 2761 |
+
"step": 3930
|
| 2762 |
+
},
|
| 2763 |
+
{
|
| 2764 |
+
"epoch": 5.331529093369419,
|
| 2765 |
+
"grad_norm": 0.14141711592674255,
|
| 2766 |
+
"learning_rate": 1.1813756282965888e-05,
|
| 2767 |
+
"loss": 0.0324,
|
| 2768 |
+
"step": 3940
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 5.345060893098782,
|
| 2772 |
+
"grad_norm": 0.1509343385696411,
|
| 2773 |
+
"learning_rate": 1.1601116232052638e-05,
|
| 2774 |
+
"loss": 0.0293,
|
| 2775 |
+
"step": 3950
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 5.358592692828146,
|
| 2779 |
+
"grad_norm": 0.24892765283584595,
|
| 2780 |
+
"learning_rate": 1.1390155872710517e-05,
|
| 2781 |
+
"loss": 0.0287,
|
| 2782 |
+
"step": 3960
|
| 2783 |
+
},
|
| 2784 |
+
{
|
| 2785 |
+
"epoch": 5.37212449255751,
|
| 2786 |
+
"grad_norm": 0.28979232907295227,
|
| 2787 |
+
"learning_rate": 1.1180884433029087e-05,
|
| 2788 |
+
"loss": 0.0263,
|
| 2789 |
+
"step": 3970
|
| 2790 |
+
},
|
| 2791 |
+
{
|
| 2792 |
+
"epoch": 5.385656292286874,
|
| 2793 |
+
"grad_norm": 0.1439746767282486,
|
| 2794 |
+
"learning_rate": 1.097331106721904e-05,
|
| 2795 |
+
"loss": 0.0319,
|
| 2796 |
+
"step": 3980
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"epoch": 5.399188092016238,
|
| 2800 |
+
"grad_norm": 0.12562195956707,
|
| 2801 |
+
"learning_rate": 1.0767444855211862e-05,
|
| 2802 |
+
"loss": 0.0274,
|
| 2803 |
+
"step": 3990
|
| 2804 |
+
},
|
| 2805 |
+
{
|
| 2806 |
+
"epoch": 5.412719891745602,
|
| 2807 |
+
"grad_norm": 0.14074930548667908,
|
| 2808 |
+
"learning_rate": 1.0563294802262558e-05,
|
| 2809 |
+
"loss": 0.0273,
|
| 2810 |
+
"step": 4000
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 5.426251691474966,
|
| 2814 |
+
"grad_norm": 0.10954784601926804,
|
| 2815 |
+
"learning_rate": 1.0360869838555809e-05,
|
| 2816 |
+
"loss": 0.0285,
|
| 2817 |
+
"step": 4010
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 5.43978349120433,
|
| 2821 |
+
"grad_norm": 0.147128164768219,
|
| 2822 |
+
"learning_rate": 1.0160178818815313e-05,
|
| 2823 |
+
"loss": 0.0296,
|
| 2824 |
+
"step": 4020
|
| 2825 |
+
},
|
| 2826 |
+
{
|
| 2827 |
+
"epoch": 5.453315290933694,
|
| 2828 |
+
"grad_norm": 0.10633907467126846,
|
| 2829 |
+
"learning_rate": 9.961230521916387e-06,
|
| 2830 |
+
"loss": 0.031,
|
| 2831 |
+
"step": 4030
|
| 2832 |
+
},
|
| 2833 |
+
{
|
| 2834 |
+
"epoch": 5.466847090663058,
|
| 2835 |
+
"grad_norm": 0.15704360604286194,
|
| 2836 |
+
"learning_rate": 9.764033650502074e-06,
|
| 2837 |
+
"loss": 0.0277,
|
| 2838 |
+
"step": 4040
|
| 2839 |
+
},
|
| 2840 |
+
{
|
| 2841 |
+
"epoch": 5.480378890392422,
|
| 2842 |
+
"grad_norm": 0.10788650810718536,
|
| 2843 |
+
"learning_rate": 9.568596830602344e-06,
|
| 2844 |
+
"loss": 0.028,
|
| 2845 |
+
"step": 4050
|
| 2846 |
+
},
|
| 2847 |
+
{
|
| 2848 |
+
"epoch": 5.493910690121786,
|
| 2849 |
+
"grad_norm": 0.14843221008777618,
|
| 2850 |
+
"learning_rate": 9.37492861125681e-06,
|
| 2851 |
+
"loss": 0.0303,
|
| 2852 |
+
"step": 4060
|
| 2853 |
+
},
|
| 2854 |
+
{
|
| 2855 |
+
"epoch": 5.50744248985115,
|
| 2856 |
+
"grad_norm": 0.1297934353351593,
|
| 2857 |
+
"learning_rate": 9.183037464140804e-06,
|
| 2858 |
+
"loss": 0.0293,
|
| 2859 |
+
"step": 4070
|
| 2860 |
+
},
|
| 2861 |
+
{
|
| 2862 |
+
"epoch": 5.520974289580514,
|
| 2863 |
+
"grad_norm": 0.13509418070316315,
|
| 2864 |
+
"learning_rate": 8.992931783194735e-06,
|
| 2865 |
+
"loss": 0.0294,
|
| 2866 |
+
"step": 4080
|
| 2867 |
+
},
|
| 2868 |
+
{
|
| 2869 |
+
"epoch": 5.534506089309878,
|
| 2870 |
+
"grad_norm": 0.1298404484987259,
|
| 2871 |
+
"learning_rate": 8.80461988425696e-06,
|
| 2872 |
+
"loss": 0.0329,
|
| 2873 |
+
"step": 4090
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 5.548037889039242,
|
| 2877 |
+
"grad_norm": 0.12505508959293365,
|
| 2878 |
+
"learning_rate": 8.618110004699974e-06,
|
| 2879 |
+
"loss": 0.0272,
|
| 2880 |
+
"step": 4100
|
| 2881 |
+
},
|
| 2882 |
+
{
|
| 2883 |
+
"epoch": 5.561569688768606,
|
| 2884 |
+
"grad_norm": 0.09546499699354172,
|
| 2885 |
+
"learning_rate": 8.4334103030701e-06,
|
| 2886 |
+
"loss": 0.0273,
|
| 2887 |
+
"step": 4110
|
| 2888 |
+
},
|
| 2889 |
+
{
|
| 2890 |
+
"epoch": 5.57510148849797,
|
| 2891 |
+
"grad_norm": 0.15529732406139374,
|
| 2892 |
+
"learning_rate": 8.25052885873066e-06,
|
| 2893 |
+
"loss": 0.0293,
|
| 2894 |
+
"step": 4120
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 5.588633288227334,
|
| 2898 |
+
"grad_norm": 0.12811461091041565,
|
| 2899 |
+
"learning_rate": 8.06947367150846e-06,
|
| 2900 |
+
"loss": 0.0274,
|
| 2901 |
+
"step": 4130
|
| 2902 |
+
},
|
| 2903 |
+
{
|
| 2904 |
+
"epoch": 5.602165087956698,
|
| 2905 |
+
"grad_norm": 0.18342335522174835,
|
| 2906 |
+
"learning_rate": 7.890252661343938e-06,
|
| 2907 |
+
"loss": 0.028,
|
| 2908 |
+
"step": 4140
|
| 2909 |
+
},
|
| 2910 |
+
{
|
| 2911 |
+
"epoch": 5.615696887686062,
|
| 2912 |
+
"grad_norm": 0.11130325496196747,
|
| 2913 |
+
"learning_rate": 7.712873667944681e-06,
|
| 2914 |
+
"loss": 0.0286,
|
| 2915 |
+
"step": 4150
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 5.629228687415426,
|
| 2919 |
+
"grad_norm": 0.14142289757728577,
|
| 2920 |
+
"learning_rate": 7.537344450442469e-06,
|
| 2921 |
+
"loss": 0.03,
|
| 2922 |
+
"step": 4160
|
| 2923 |
+
},
|
| 2924 |
+
{
|
| 2925 |
+
"epoch": 5.64276048714479,
|
| 2926 |
+
"grad_norm": 0.19619788229465485,
|
| 2927 |
+
"learning_rate": 7.36367268705393e-06,
|
| 2928 |
+
"loss": 0.0335,
|
| 2929 |
+
"step": 4170
|
| 2930 |
+
},
|
| 2931 |
+
{
|
| 2932 |
+
"epoch": 5.656292286874154,
|
| 2933 |
+
"grad_norm": 0.13693951070308685,
|
| 2934 |
+
"learning_rate": 7.1918659747446e-06,
|
| 2935 |
+
"loss": 0.0287,
|
| 2936 |
+
"step": 4180
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 5.669824086603518,
|
| 2940 |
+
"grad_norm": 0.10885664820671082,
|
| 2941 |
+
"learning_rate": 7.021931828896666e-06,
|
| 2942 |
+
"loss": 0.0327,
|
| 2943 |
+
"step": 4190
|
| 2944 |
+
},
|
| 2945 |
+
{
|
| 2946 |
+
"epoch": 5.683355886332882,
|
| 2947 |
+
"grad_norm": 0.14536768198013306,
|
| 2948 |
+
"learning_rate": 6.8538776829801584e-06,
|
| 2949 |
+
"loss": 0.0278,
|
| 2950 |
+
"step": 4200
|
| 2951 |
+
},
|
| 2952 |
+
{
|
| 2953 |
+
"epoch": 5.696887686062246,
|
| 2954 |
+
"grad_norm": 0.15630270540714264,
|
| 2955 |
+
"learning_rate": 6.687710888227849e-06,
|
| 2956 |
+
"loss": 0.0288,
|
| 2957 |
+
"step": 4210
|
| 2958 |
+
},
|
| 2959 |
+
{
|
| 2960 |
+
"epoch": 5.71041948579161,
|
| 2961 |
+
"grad_norm": 0.1373508870601654,
|
| 2962 |
+
"learning_rate": 6.5234387133136565e-06,
|
| 2963 |
+
"loss": 0.0296,
|
| 2964 |
+
"step": 4220
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 5.723951285520974,
|
| 2968 |
+
"grad_norm": 0.16177499294281006,
|
| 2969 |
+
"learning_rate": 6.361068344034665e-06,
|
| 2970 |
+
"loss": 0.028,
|
| 2971 |
+
"step": 4230
|
| 2972 |
+
},
|
| 2973 |
+
{
|
| 2974 |
+
"epoch": 5.7374830852503385,
|
| 2975 |
+
"grad_norm": 0.11304031312465668,
|
| 2976 |
+
"learning_rate": 6.200606882996846e-06,
|
| 2977 |
+
"loss": 0.0283,
|
| 2978 |
+
"step": 4240
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 5.7510148849797025,
|
| 2982 |
+
"grad_norm": 0.27458086609840393,
|
| 2983 |
+
"learning_rate": 6.042061349304312e-06,
|
| 2984 |
+
"loss": 0.03,
|
| 2985 |
+
"step": 4250
|
| 2986 |
+
},
|
| 2987 |
+
{
|
| 2988 |
+
"epoch": 5.7645466847090665,
|
| 2989 |
+
"grad_norm": 0.24669907987117767,
|
| 2990 |
+
"learning_rate": 5.885438678252342e-06,
|
| 2991 |
+
"loss": 0.0317,
|
| 2992 |
+
"step": 4260
|
| 2993 |
+
},
|
| 2994 |
+
{
|
| 2995 |
+
"epoch": 5.7780784844384305,
|
| 2996 |
+
"grad_norm": 0.14182038605213165,
|
| 2997 |
+
"learning_rate": 5.730745721023939e-06,
|
| 2998 |
+
"loss": 0.0302,
|
| 2999 |
+
"step": 4270
|
| 3000 |
+
},
|
| 3001 |
+
{
|
| 3002 |
+
"epoch": 5.7916102841677946,
|
| 3003 |
+
"grad_norm": 0.1324010193347931,
|
| 3004 |
+
"learning_rate": 5.577989244390192e-06,
|
| 3005 |
+
"loss": 0.0288,
|
| 3006 |
+
"step": 4280
|
| 3007 |
+
},
|
| 3008 |
+
{
|
| 3009 |
+
"epoch": 5.805142083897159,
|
| 3010 |
+
"grad_norm": 0.2429128736257553,
|
| 3011 |
+
"learning_rate": 5.4271759304142635e-06,
|
| 3012 |
+
"loss": 0.0268,
|
| 3013 |
+
"step": 4290
|
| 3014 |
+
},
|
| 3015 |
+
{
|
| 3016 |
+
"epoch": 5.818673883626523,
|
| 3017 |
+
"grad_norm": 0.20134113729000092,
|
| 3018 |
+
"learning_rate": 5.278312376159051e-06,
|
| 3019 |
+
"loss": 0.0282,
|
| 3020 |
+
"step": 4300
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 5.832205683355887,
|
| 3024 |
+
"grad_norm": 0.18878605961799622,
|
| 3025 |
+
"learning_rate": 5.1314050933986944e-06,
|
| 3026 |
+
"loss": 0.0265,
|
| 3027 |
+
"step": 4310
|
| 3028 |
+
},
|
| 3029 |
+
{
|
| 3030 |
+
"epoch": 5.845737483085251,
|
| 3031 |
+
"grad_norm": 0.1928325742483139,
|
| 3032 |
+
"learning_rate": 4.986460508333634e-06,
|
| 3033 |
+
"loss": 0.0262,
|
| 3034 |
+
"step": 4320
|
| 3035 |
+
},
|
| 3036 |
+
{
|
| 3037 |
+
"epoch": 5.859269282814615,
|
| 3038 |
+
"grad_norm": 0.11430050432682037,
|
| 3039 |
+
"learning_rate": 4.843484961309597e-06,
|
| 3040 |
+
"loss": 0.0278,
|
| 3041 |
+
"step": 4330
|
| 3042 |
+
},
|
| 3043 |
+
{
|
| 3044 |
+
"epoch": 5.872801082543979,
|
| 3045 |
+
"grad_norm": 0.1418386846780777,
|
| 3046 |
+
"learning_rate": 4.702484706540161e-06,
|
| 3047 |
+
"loss": 0.0249,
|
| 3048 |
+
"step": 4340
|
| 3049 |
+
},
|
| 3050 |
+
{
|
| 3051 |
+
"epoch": 5.886332882273343,
|
| 3052 |
+
"grad_norm": 0.1781853586435318,
|
| 3053 |
+
"learning_rate": 4.563465911833259e-06,
|
| 3054 |
+
"loss": 0.0289,
|
| 3055 |
+
"step": 4350
|
| 3056 |
+
},
|
| 3057 |
+
{
|
| 3058 |
+
"epoch": 5.899864682002706,
|
| 3059 |
+
"grad_norm": 0.16866528987884521,
|
| 3060 |
+
"learning_rate": 4.426434658321344e-06,
|
| 3061 |
+
"loss": 0.0248,
|
| 3062 |
+
"step": 4360
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 5.91339648173207,
|
| 3066 |
+
"grad_norm": 0.25363948941230774,
|
| 3067 |
+
"learning_rate": 4.2913969401953466e-06,
|
| 3068 |
+
"loss": 0.0272,
|
| 3069 |
+
"step": 4370
|
| 3070 |
+
},
|
| 3071 |
+
{
|
| 3072 |
+
"epoch": 5.926928281461434,
|
| 3073 |
+
"grad_norm": 0.1106293648481369,
|
| 3074 |
+
"learning_rate": 4.15835866444253e-06,
|
| 3075 |
+
"loss": 0.0252,
|
| 3076 |
+
"step": 4380
|
| 3077 |
+
},
|
| 3078 |
+
{
|
| 3079 |
+
"epoch": 5.940460081190798,
|
| 3080 |
+
"grad_norm": 0.1236589327454567,
|
| 3081 |
+
"learning_rate": 4.027325650588043e-06,
|
| 3082 |
+
"loss": 0.0281,
|
| 3083 |
+
"step": 4390
|
| 3084 |
+
},
|
| 3085 |
+
{
|
| 3086 |
+
"epoch": 5.953991880920162,
|
| 3087 |
+
"grad_norm": 0.14823655784130096,
|
| 3088 |
+
"learning_rate": 3.898303630440419e-06,
|
| 3089 |
+
"loss": 0.0306,
|
| 3090 |
+
"step": 4400
|
| 3091 |
+
},
|
| 3092 |
+
{
|
| 3093 |
+
"epoch": 5.967523680649526,
|
| 3094 |
+
"grad_norm": 0.16370555758476257,
|
| 3095 |
+
"learning_rate": 3.7712982478407877e-06,
|
| 3096 |
+
"loss": 0.0294,
|
| 3097 |
+
"step": 4410
|
| 3098 |
+
},
|
| 3099 |
+
{
|
| 3100 |
+
"epoch": 5.98105548037889,
|
| 3101 |
+
"grad_norm": 0.2005997598171234,
|
| 3102 |
+
"learning_rate": 3.6463150584160053e-06,
|
| 3103 |
+
"loss": 0.0275,
|
| 3104 |
+
"step": 4420
|
| 3105 |
+
},
|
| 3106 |
+
{
|
| 3107 |
+
"epoch": 5.994587280108254,
|
| 3108 |
+
"grad_norm": 0.2858164310455322,
|
| 3109 |
+
"learning_rate": 3.5233595293356957e-06,
|
| 3110 |
+
"loss": 0.029,
|
| 3111 |
+
"step": 4430
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 6.008119079837618,
|
| 3115 |
+
"grad_norm": 0.16128398478031158,
|
| 3116 |
+
"learning_rate": 3.4024370390730033e-06,
|
| 3117 |
+
"loss": 0.0316,
|
| 3118 |
+
"step": 4440
|
| 3119 |
+
},
|
| 3120 |
+
{
|
| 3121 |
+
"epoch": 6.021650879566982,
|
| 3122 |
+
"grad_norm": 0.12964995205402374,
|
| 3123 |
+
"learning_rate": 3.2835528771693992e-06,
|
| 3124 |
+
"loss": 0.0261,
|
| 3125 |
+
"step": 4450
|
| 3126 |
+
},
|
| 3127 |
+
{
|
| 3128 |
+
"epoch": 6.035182679296346,
|
| 3129 |
+
"grad_norm": 0.16563497483730316,
|
| 3130 |
+
"learning_rate": 3.1667122440032505e-06,
|
| 3131 |
+
"loss": 0.0263,
|
| 3132 |
+
"step": 4460
|
| 3133 |
+
},
|
| 3134 |
+
{
|
| 3135 |
+
"epoch": 6.04871447902571,
|
| 3136 |
+
"grad_norm": 0.1846737265586853,
|
| 3137 |
+
"learning_rate": 3.051920250562351e-06,
|
| 3138 |
+
"loss": 0.0286,
|
| 3139 |
+
"step": 4470
|
| 3140 |
+
},
|
| 3141 |
+
{
|
| 3142 |
+
"epoch": 6.062246278755074,
|
| 3143 |
+
"grad_norm": 0.23904356360435486,
|
| 3144 |
+
"learning_rate": 2.939181918220385e-06,
|
| 3145 |
+
"loss": 0.0284,
|
| 3146 |
+
"step": 4480
|
| 3147 |
+
},
|
| 3148 |
+
{
|
| 3149 |
+
"epoch": 6.075778078484438,
|
| 3150 |
+
"grad_norm": 0.22299540042877197,
|
| 3151 |
+
"learning_rate": 2.8285021785172226e-06,
|
| 3152 |
+
"loss": 0.0268,
|
| 3153 |
+
"step": 4490
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 6.089309878213802,
|
| 3157 |
+
"grad_norm": 0.36245062947273254,
|
| 3158 |
+
"learning_rate": 2.7198858729432288e-06,
|
| 3159 |
+
"loss": 0.0302,
|
| 3160 |
+
"step": 4500
|
| 3161 |
+
},
|
| 3162 |
+
{
|
| 3163 |
+
"epoch": 6.102841677943166,
|
| 3164 |
+
"grad_norm": 0.1929812878370285,
|
| 3165 |
+
"learning_rate": 2.6133377527274905e-06,
|
| 3166 |
+
"loss": 0.029,
|
| 3167 |
+
"step": 4510
|
| 3168 |
+
},
|
| 3169 |
+
{
|
| 3170 |
+
"epoch": 6.11637347767253,
|
| 3171 |
+
"grad_norm": 0.1011333093047142,
|
| 3172 |
+
"learning_rate": 2.5088624786299366e-06,
|
| 3173 |
+
"loss": 0.0252,
|
| 3174 |
+
"step": 4520
|
| 3175 |
+
},
|
| 3176 |
+
{
|
| 3177 |
+
"epoch": 6.129905277401894,
|
| 3178 |
+
"grad_norm": 0.19274236261844635,
|
| 3179 |
+
"learning_rate": 2.406464620737531e-06,
|
| 3180 |
+
"loss": 0.0271,
|
| 3181 |
+
"step": 4530
|
| 3182 |
+
},
|
| 3183 |
+
{
|
| 3184 |
+
"epoch": 6.143437077131258,
|
| 3185 |
+
"grad_norm": 0.11799866706132889,
|
| 3186 |
+
"learning_rate": 2.3061486582642734e-06,
|
| 3187 |
+
"loss": 0.0241,
|
| 3188 |
+
"step": 4540
|
| 3189 |
+
},
|
| 3190 |
+
{
|
| 3191 |
+
"epoch": 6.156968876860622,
|
| 3192 |
+
"grad_norm": 0.09267459064722061,
|
| 3193 |
+
"learning_rate": 2.2079189793553667e-06,
|
| 3194 |
+
"loss": 0.0248,
|
| 3195 |
+
"step": 4550
|
| 3196 |
+
},
|
| 3197 |
+
{
|
| 3198 |
+
"epoch": 6.170500676589986,
|
| 3199 |
+
"grad_norm": 0.15937362611293793,
|
| 3200 |
+
"learning_rate": 2.111779880895165e-06,
|
| 3201 |
+
"loss": 0.0284,
|
| 3202 |
+
"step": 4560
|
| 3203 |
+
},
|
| 3204 |
+
{
|
| 3205 |
+
"epoch": 6.18403247631935,
|
| 3206 |
+
"grad_norm": 0.12410353124141693,
|
| 3207 |
+
"learning_rate": 2.01773556831929e-06,
|
| 3208 |
+
"loss": 0.0292,
|
| 3209 |
+
"step": 4570
|
| 3210 |
+
},
|
| 3211 |
+
{
|
| 3212 |
+
"epoch": 6.197564276048714,
|
| 3213 |
+
"grad_norm": 0.20976869761943817,
|
| 3214 |
+
"learning_rate": 1.9257901554306513e-06,
|
| 3215 |
+
"loss": 0.0278,
|
| 3216 |
+
"step": 4580
|
| 3217 |
+
},
|
| 3218 |
+
{
|
| 3219 |
+
"epoch": 6.211096075778078,
|
| 3220 |
+
"grad_norm": 0.13598106801509857,
|
| 3221 |
+
"learning_rate": 1.835947664219445e-06,
|
| 3222 |
+
"loss": 0.0265,
|
| 3223 |
+
"step": 4590
|
| 3224 |
+
},
|
| 3225 |
+
{
|
| 3226 |
+
"epoch": 6.2246278755074425,
|
| 3227 |
+
"grad_norm": 0.10610091686248779,
|
| 3228 |
+
"learning_rate": 1.748212024687307e-06,
|
| 3229 |
+
"loss": 0.027,
|
| 3230 |
+
"step": 4600
|
| 3231 |
+
},
|
| 3232 |
+
{
|
| 3233 |
+
"epoch": 6.2381596752368065,
|
| 3234 |
+
"grad_norm": 0.14155301451683044,
|
| 3235 |
+
"learning_rate": 1.6625870746753147e-06,
|
| 3236 |
+
"loss": 0.026,
|
| 3237 |
+
"step": 4610
|
| 3238 |
+
},
|
| 3239 |
+
{
|
| 3240 |
+
"epoch": 6.2516914749661705,
|
| 3241 |
+
"grad_norm": 0.12627993524074554,
|
| 3242 |
+
"learning_rate": 1.5790765596961853e-06,
|
| 3243 |
+
"loss": 0.0276,
|
| 3244 |
+
"step": 4620
|
| 3245 |
+
},
|
| 3246 |
+
{
|
| 3247 |
+
"epoch": 6.2652232746955345,
|
| 3248 |
+
"grad_norm": 0.2003839612007141,
|
| 3249 |
+
"learning_rate": 1.4976841327703717e-06,
|
| 3250 |
+
"loss": 0.0272,
|
| 3251 |
+
"step": 4630
|
| 3252 |
+
},
|
| 3253 |
+
{
|
| 3254 |
+
"epoch": 6.2787550744248986,
|
| 3255 |
+
"grad_norm": 0.18389511108398438,
|
| 3256 |
+
"learning_rate": 1.4184133542663014e-06,
|
| 3257 |
+
"loss": 0.0303,
|
| 3258 |
+
"step": 4640
|
| 3259 |
+
},
|
| 3260 |
+
{
|
| 3261 |
+
"epoch": 6.292286874154263,
|
| 3262 |
+
"grad_norm": 0.1884324848651886,
|
| 3263 |
+
"learning_rate": 1.341267691744641e-06,
|
| 3264 |
+
"loss": 0.0308,
|
| 3265 |
+
"step": 4650
|
| 3266 |
+
},
|
| 3267 |
+
{
|
| 3268 |
+
"epoch": 6.305818673883627,
|
| 3269 |
+
"grad_norm": 0.13071194291114807,
|
| 3270 |
+
"learning_rate": 1.2662505198065666e-06,
|
| 3271 |
+
"loss": 0.0316,
|
| 3272 |
+
"step": 4660
|
| 3273 |
+
},
|
| 3274 |
+
{
|
| 3275 |
+
"epoch": 6.319350473612991,
|
| 3276 |
+
"grad_norm": 0.09089189767837524,
|
| 3277 |
+
"learning_rate": 1.193365119946216e-06,
|
| 3278 |
+
"loss": 0.0279,
|
| 3279 |
+
"step": 4670
|
| 3280 |
+
},
|
| 3281 |
+
{
|
| 3282 |
+
"epoch": 6.332882273342355,
|
| 3283 |
+
"grad_norm": 0.1488385945558548,
|
| 3284 |
+
"learning_rate": 1.1226146804070859e-06,
|
| 3285 |
+
"loss": 0.027,
|
| 3286 |
+
"step": 4680
|
| 3287 |
+
},
|
| 3288 |
+
{
|
| 3289 |
+
"epoch": 6.346414073071719,
|
| 3290 |
+
"grad_norm": 0.12004260718822479,
|
| 3291 |
+
"learning_rate": 1.0540022960426111e-06,
|
| 3292 |
+
"loss": 0.0272,
|
| 3293 |
+
"step": 4690
|
| 3294 |
+
},
|
| 3295 |
+
{
|
| 3296 |
+
"epoch": 6.359945872801083,
|
| 3297 |
+
"grad_norm": 0.08131157606840134,
|
| 3298 |
+
"learning_rate": 9.875309681807443e-07,
|
| 3299 |
+
"loss": 0.0288,
|
| 3300 |
+
"step": 4700
|
| 3301 |
+
},
|
| 3302 |
+
{
|
| 3303 |
+
"epoch": 6.373477672530447,
|
| 3304 |
+
"grad_norm": 0.13074468076229095,
|
| 3305 |
+
"learning_rate": 9.232036044927061e-07,
|
| 3306 |
+
"loss": 0.0282,
|
| 3307 |
+
"step": 4710
|
| 3308 |
+
},
|
| 3309 |
+
{
|
| 3310 |
+
"epoch": 6.387009472259811,
|
| 3311 |
+
"grad_norm": 0.2715424597263336,
|
| 3312 |
+
"learning_rate": 8.610230188657919e-07,
|
| 3313 |
+
"loss": 0.0268,
|
| 3314 |
+
"step": 4720
|
| 3315 |
+
},
|
| 3316 |
+
{
|
| 3317 |
+
"epoch": 6.400541271989175,
|
| 3318 |
+
"grad_norm": 0.11489482969045639,
|
| 3319 |
+
"learning_rate": 8.009919312802372e-07,
|
| 3320 |
+
"loss": 0.0288,
|
| 3321 |
+
"step": 4730
|
| 3322 |
+
},
|
| 3323 |
+
{
|
| 3324 |
+
"epoch": 6.414073071718539,
|
| 3325 |
+
"grad_norm": 0.30984801054000854,
|
| 3326 |
+
"learning_rate": 7.431129676902904e-07,
|
| 3327 |
+
"loss": 0.0312,
|
| 3328 |
+
"step": 4740
|
| 3329 |
+
},
|
| 3330 |
+
{
|
| 3331 |
+
"epoch": 6.427604871447903,
|
| 3332 |
+
"grad_norm": 0.26709073781967163,
|
| 3333 |
+
"learning_rate": 6.873886599093215e-07,
|
| 3334 |
+
"loss": 0.026,
|
| 3335 |
+
"step": 4750
|
| 3336 |
+
},
|
| 3337 |
+
{
|
| 3338 |
+
"epoch": 6.441136671177267,
|
| 3339 |
+
"grad_norm": 0.16023863852024078,
|
| 3340 |
+
"learning_rate": 6.338214454990776e-07,
|
| 3341 |
+
"loss": 0.0289,
|
| 3342 |
+
"step": 4760
|
| 3343 |
+
},
|
| 3344 |
+
{
|
| 3345 |
+
"epoch": 6.454668470906631,
|
| 3346 |
+
"grad_norm": 0.09971076250076294,
|
| 3347 |
+
"learning_rate": 5.82413667663051e-07,
|
| 3348 |
+
"loss": 0.028,
|
| 3349 |
+
"step": 4770
|
| 3350 |
+
},
|
| 3351 |
+
{
|
| 3352 |
+
"epoch": 6.468200270635995,
|
| 3353 |
+
"grad_norm": 0.10005897283554077,
|
| 3354 |
+
"learning_rate": 5.331675751439725e-07,
|
| 3355 |
+
"loss": 0.0274,
|
| 3356 |
+
"step": 4780
|
| 3357 |
+
},
|
| 3358 |
+
{
|
| 3359 |
+
"epoch": 6.481732070365359,
|
| 3360 |
+
"grad_norm": 0.132291778922081,
|
| 3361 |
+
"learning_rate": 4.86085322125479e-07,
|
| 3362 |
+
"loss": 0.0257,
|
| 3363 |
+
"step": 4790
|
| 3364 |
+
},
|
| 3365 |
+
{
|
| 3366 |
+
"epoch": 6.495263870094723,
|
| 3367 |
+
"grad_norm": 0.11425412446260452,
|
| 3368 |
+
"learning_rate": 4.411689681378284e-07,
|
| 3369 |
+
"loss": 0.0296,
|
| 3370 |
+
"step": 4800
|
| 3371 |
+
},
|
| 3372 |
+
{
|
| 3373 |
+
"epoch": 6.508795669824087,
|
| 3374 |
+
"grad_norm": 0.10741697996854782,
|
| 3375 |
+
"learning_rate": 3.9842047796786466e-07,
|
| 3376 |
+
"loss": 0.0263,
|
| 3377 |
+
"step": 4810
|
| 3378 |
+
},
|
| 3379 |
+
{
|
| 3380 |
+
"epoch": 6.522327469553451,
|
| 3381 |
+
"grad_norm": 0.13160759210586548,
|
| 3382 |
+
"learning_rate": 3.578417215730323e-07,
|
| 3383 |
+
"loss": 0.0278,
|
| 3384 |
+
"step": 4820
|
| 3385 |
+
},
|
| 3386 |
+
{
|
| 3387 |
+
"epoch": 6.535859269282815,
|
| 3388 |
+
"grad_norm": 0.14389033615589142,
|
| 3389 |
+
"learning_rate": 3.1943447399958027e-07,
|
| 3390 |
+
"loss": 0.0252,
|
| 3391 |
+
"step": 4830
|
| 3392 |
+
},
|
| 3393 |
+
{
|
| 3394 |
+
"epoch": 6.549391069012179,
|
| 3395 |
+
"grad_norm": 0.4230325222015381,
|
| 3396 |
+
"learning_rate": 2.8320041530495724e-07,
|
| 3397 |
+
"loss": 0.0356,
|
| 3398 |
+
"step": 4840
|
| 3399 |
+
},
|
| 3400 |
+
{
|
| 3401 |
+
"epoch": 6.562922868741543,
|
| 3402 |
+
"grad_norm": 0.5279276371002197,
|
| 3403 |
+
"learning_rate": 2.491411304842539e-07,
|
| 3404 |
+
"loss": 0.0289,
|
| 3405 |
+
"step": 4850
|
| 3406 |
+
},
|
| 3407 |
+
{
|
| 3408 |
+
"epoch": 6.576454668470907,
|
| 3409 |
+
"grad_norm": 0.12740273773670197,
|
| 3410 |
+
"learning_rate": 2.1725810940094183e-07,
|
| 3411 |
+
"loss": 0.0278,
|
| 3412 |
+
"step": 4860
|
| 3413 |
+
},
|
| 3414 |
+
{
|
| 3415 |
+
"epoch": 6.589986468200271,
|
| 3416 |
+
"grad_norm": 0.1272871196269989,
|
| 3417 |
+
"learning_rate": 1.8755274672164202e-07,
|
| 3418 |
+
"loss": 0.0256,
|
| 3419 |
+
"step": 4870
|
| 3420 |
+
},
|
| 3421 |
+
{
|
| 3422 |
+
"epoch": 6.603518267929634,
|
| 3423 |
+
"grad_norm": 0.1781158745288849,
|
| 3424 |
+
"learning_rate": 1.600263418551573e-07,
|
| 3425 |
+
"loss": 0.0263,
|
| 3426 |
+
"step": 4880
|
| 3427 |
+
},
|
| 3428 |
+
{
|
| 3429 |
+
"epoch": 6.617050067658998,
|
| 3430 |
+
"grad_norm": 0.0960567444562912,
|
| 3431 |
+
"learning_rate": 1.346800988955954e-07,
|
| 3432 |
+
"loss": 0.0268,
|
| 3433 |
+
"step": 4890
|
| 3434 |
+
},
|
| 3435 |
+
{
|
| 3436 |
+
"epoch": 6.630581867388362,
|
| 3437 |
+
"grad_norm": 0.22953420877456665,
|
| 3438 |
+
"learning_rate": 1.1151512656975005e-07,
|
| 3439 |
+
"loss": 0.027,
|
| 3440 |
+
"step": 4900
|
| 3441 |
+
},
|
| 3442 |
+
{
|
| 3443 |
+
"epoch": 6.644113667117726,
|
| 3444 |
+
"grad_norm": 0.22247491776943207,
|
| 3445 |
+
"learning_rate": 9.053243818853973e-08,
|
| 3446 |
+
"loss": 0.0291,
|
| 3447 |
+
"step": 4910
|
| 3448 |
+
},
|
| 3449 |
+
{
|
| 3450 |
+
"epoch": 6.65764546684709,
|
| 3451 |
+
"grad_norm": 0.13925011456012726,
|
| 3452 |
+
"learning_rate": 7.173295160273763e-08,
|
| 3453 |
+
"loss": 0.0256,
|
| 3454 |
+
"step": 4920
|
| 3455 |
+
},
|
| 3456 |
+
{
|
| 3457 |
+
"epoch": 6.671177266576454,
|
| 3458 |
+
"grad_norm": 0.4142376780509949,
|
| 3459 |
+
"learning_rate": 5.511748916279258e-08,
|
| 3460 |
+
"loss": 0.0275,
|
| 3461 |
+
"step": 4930
|
| 3462 |
+
},
|
| 3463 |
+
{
|
| 3464 |
+
"epoch": 6.684709066305818,
|
| 3465 |
+
"grad_norm": 0.2592661678791046,
|
| 3466 |
+
"learning_rate": 4.068677768285234e-08,
|
| 3467 |
+
"loss": 0.0256,
|
| 3468 |
+
"step": 4940
|
| 3469 |
+
},
|
| 3470 |
+
{
|
| 3471 |
+
"epoch": 6.698240866035182,
|
| 3472 |
+
"grad_norm": 0.1092233955860138,
|
| 3473 |
+
"learning_rate": 2.844144840898344e-08,
|
| 3474 |
+
"loss": 0.0253,
|
| 3475 |
+
"step": 4950
|
| 3476 |
+
},
|
| 3477 |
+
{
|
| 3478 |
+
"epoch": 6.7117726657645465,
|
| 3479 |
+
"grad_norm": 0.09404899179935455,
|
| 3480 |
+
"learning_rate": 1.8382036991559936e-08,
|
| 3481 |
+
"loss": 0.0255,
|
| 3482 |
+
"step": 4960
|
| 3483 |
+
},
|
| 3484 |
+
{
|
| 3485 |
+
"epoch": 6.7253044654939105,
|
| 3486 |
+
"grad_norm": 0.18313178420066833,
|
| 3487 |
+
"learning_rate": 1.0508983461832156e-08,
|
| 3488 |
+
"loss": 0.0286,
|
| 3489 |
+
"step": 4970
|
| 3490 |
+
},
|
| 3491 |
+
{
|
| 3492 |
+
"epoch": 6.7388362652232745,
|
| 3493 |
+
"grad_norm": 0.0930226519703865,
|
| 3494 |
+
"learning_rate": 4.822632212653222e-09,
|
| 3495 |
+
"loss": 0.0284,
|
| 3496 |
+
"step": 4980
|
| 3497 |
+
},
|
| 3498 |
+
{
|
| 3499 |
+
"epoch": 6.7523680649526385,
|
| 3500 |
+
"grad_norm": 0.12706884741783142,
|
| 3501 |
+
"learning_rate": 1.3232319834632912e-09,
|
| 3502 |
+
"loss": 0.0255,
|
| 3503 |
+
"step": 4990
|
| 3504 |
+
},
|
| 3505 |
+
{
|
| 3506 |
+
"epoch": 6.7658998646820026,
|
| 3507 |
+
"grad_norm": 0.264944463968277,
|
| 3508 |
+
"learning_rate": 1.0935849353854721e-11,
|
| 3509 |
+
"loss": 0.0264,
|
| 3510 |
+
"step": 5000
|
| 3511 |
+
},
|
| 3512 |
+
{
|
| 3513 |
+
"epoch": 6.7658998646820026,
|
| 3514 |
+
"step": 5000,
|
| 3515 |
+
"total_flos": 0.0,
|
| 3516 |
+
"train_loss": 0.04250964983403683,
|
| 3517 |
+
"train_runtime": 2361.608,
|
| 3518 |
+
"train_samples_per_second": 67.75,
|
| 3519 |
+
"train_steps_per_second": 2.117
|
| 3520 |
+
}
|
| 3521 |
+
],
|
| 3522 |
+
"logging_steps": 10,
|
| 3523 |
+
"max_steps": 5000,
|
| 3524 |
+
"num_input_tokens_seen": 0,
|
| 3525 |
+
"num_train_epochs": 7,
|
| 3526 |
+
"save_steps": 1000,
|
| 3527 |
+
"stateful_callbacks": {
|
| 3528 |
+
"TrainerControl": {
|
| 3529 |
+
"args": {
|
| 3530 |
+
"should_epoch_stop": false,
|
| 3531 |
+
"should_evaluate": false,
|
| 3532 |
+
"should_log": false,
|
| 3533 |
+
"should_save": true,
|
| 3534 |
+
"should_training_stop": true
|
| 3535 |
+
},
|
| 3536 |
+
"attributes": {}
|
| 3537 |
+
}
|
| 3538 |
+
},
|
| 3539 |
+
"total_flos": 0.0,
|
| 3540 |
+
"train_batch_size": 32,
|
| 3541 |
+
"trial_name": null,
|
| 3542 |
+
"trial_params": null
|
| 3543 |
+
}
|
drawer_20251223_5k/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:576ba2168540c67be894f4a8a0cfa9afb1c457a2faa746a2dabd6bd4e9b6ceda
|
| 3 |
+
size 5368
|