Add files using upload-large-folder tool
Browse filesThis view is limited to 50 files because it contains too many changes. See raw diff
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/resolved_config.yaml +28 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/seed0_5k.yaml +27 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/train_5shot_seed0_fullactionhead_5k/checkpoint-1000/model-00001-of-00002.safetensors +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/train_5shot_seed0_fullactionhead_5k/checkpoint-1000/model-00002-of-00002.safetensors +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/train_5shot_seed0_fullactionhead_5k/checkpoint-1000/rng_state.pth +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/train_5shot_seed0_fullactionhead_5k/checkpoint-1000/scheduler.pt +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/train_5shot_seed0_fullactionhead_5k/checkpoint-3000/config.json +64 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/train_5shot_seed0_fullactionhead_5k/checkpoint-3000/model-00001-of-00002.safetensors +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/train_5shot_seed0_fullactionhead_5k/checkpoint-3000/model-00002-of-00002.safetensors +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/train_5shot_seed0_fullactionhead_5k/checkpoint-3000/rng_state.pth +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/train_5shot_seed0_fullactionhead_5k/checkpoint-3000/scheduler.pt +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/train_5shot_seed0_fullactionhead_5k/checkpoint-3000/trainer_state.json +2134 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/train_5shot_seed0_fullactionhead_5k/experiment_cfg/metadata.json +431 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/resolved_config.yaml +28 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/seed1_5k.yaml +27 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-1000/config.json +64 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-1000/experiment_cfg/metadata.json +431 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-1000/model-00001-of-00002.safetensors +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-1000/model-00002-of-00002.safetensors +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-1000/model.safetensors.index.json +0 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-1000/rng_state.pth +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-1000/scheduler.pt +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-1000/trainer_state.json +734 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-3000/config.json +64 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-3000/experiment_cfg/metadata.json +431 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-3000/model-00001-of-00002.safetensors +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-3000/model-00002-of-00002.safetensors +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-3000/model.safetensors.index.json +0 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-3000/optimizer.pt +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-3000/rng_state.pth +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-3000/scheduler.pt +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-3000/trainer_state.json +2134 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/experiment_cfg/metadata.json +431 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-1000/config.json +64 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-1000/model-00001-of-00002.safetensors +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-1000/model-00002-of-00002.safetensors +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-1000/model.safetensors.index.json +0 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-1000/optimizer.pt +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-1000/rng_state.pth +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-1000/scheduler.pt +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-3000/model-00001-of-00002.safetensors +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-3000/model-00002-of-00002.safetensors +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-3000/rng_state.pth +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-3000/scheduler.pt +3 -0
- FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-3000/trainer_state.json +2134 -0
- FewShot/39tasks/5shot/FromBaseline26_Processingline_5k/seed0/default/train_5shot_seed0_processingline_5k/checkpoint-1000/model-00001-of-00002.safetensors +3 -0
- FewShot/39tasks/5shot/FromBaseline26_Processingline_5k/seed0/default/train_5shot_seed0_processingline_5k/checkpoint-1000/model-00002-of-00002.safetensors +3 -0
- FewShot/39tasks/5shot/FromBaseline26_Processingline_5k/seed0/default/train_5shot_seed0_processingline_5k/checkpoint-1000/optimizer.pt +3 -0
- FewShot/39tasks/5shot/FromBaseline26_Processingline_5k/seed0/default/train_5shot_seed0_processingline_5k/checkpoint-1000/rng_state.pth +3 -0
- FewShot/39tasks/5shot/FromBaseline26_Processingline_5k/seed0/default/train_5shot_seed0_processingline_5k/checkpoint-1000/scheduler.pt +3 -0
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/resolved_config.yaml
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
experiment_name: fewshot_39tasks_5shot_from_baseline26_fullactionhead_seed0_5k
|
| 2 |
+
policy_type: baseline
|
| 3 |
+
base_model_path: ${HOME}/groot_robocasa/Atomic26_baseline/checkpoint-120000
|
| 4 |
+
output_root: ${HOME}/groot_robocasa/robocasa_v2/FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0
|
| 5 |
+
dataset:
|
| 6 |
+
dataset_soup: fewshot_heldout_atomic39_human_5shot
|
| 7 |
+
training:
|
| 8 |
+
num_gpus: 1
|
| 9 |
+
batch_size: 64
|
| 10 |
+
seed: 42
|
| 11 |
+
filter_key_seed: 0
|
| 12 |
+
manifest_path: ${HOME}/clvla/benchmarks/robocasa_v2/my_scripts/Fewshot/manifests/fewshot_heldout_atomic39/shot5_seed0.json
|
| 13 |
+
save_total_limit: 4
|
| 14 |
+
phases:
|
| 15 |
+
- name: train_5shot_seed0_fullactionhead_5k
|
| 16 |
+
max_steps: 5000
|
| 17 |
+
save_steps: 0
|
| 18 |
+
save_step_list:
|
| 19 |
+
- 1000
|
| 20 |
+
- 3000
|
| 21 |
+
- 5000
|
| 22 |
+
save_final_model: false
|
| 23 |
+
trainable:
|
| 24 |
+
tune_llm: false
|
| 25 |
+
tune_visual: false
|
| 26 |
+
tune_projector: true
|
| 27 |
+
tune_diffusion_model: true
|
| 28 |
+
resolved_sweep: {}
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/seed0_5k.yaml
ADDED
|
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
experiment_name: fewshot_39tasks_5shot_from_baseline26_fullactionhead_seed0_5k
|
| 2 |
+
policy_type: baseline
|
| 3 |
+
base_model_path: ${HOME}/groot_robocasa/Atomic26_baseline/checkpoint-120000
|
| 4 |
+
output_root: ${HOME}/groot_robocasa/robocasa_v2/FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0
|
| 5 |
+
dataset:
|
| 6 |
+
dataset_soup: fewshot_heldout_atomic39_human_5shot
|
| 7 |
+
training:
|
| 8 |
+
num_gpus: 1
|
| 9 |
+
batch_size: 64
|
| 10 |
+
seed: 42
|
| 11 |
+
filter_key_seed: 0
|
| 12 |
+
manifest_path: ${HOME}/clvla/benchmarks/robocasa_v2/my_scripts/Fewshot/manifests/fewshot_heldout_atomic39/shot5_seed0.json
|
| 13 |
+
save_total_limit: 4
|
| 14 |
+
phases:
|
| 15 |
+
- name: train_5shot_seed0_fullactionhead_5k
|
| 16 |
+
max_steps: 5000
|
| 17 |
+
save_steps: 0
|
| 18 |
+
save_step_list:
|
| 19 |
+
- 1000
|
| 20 |
+
- 3000
|
| 21 |
+
- 5000
|
| 22 |
+
save_final_model: false
|
| 23 |
+
trainable:
|
| 24 |
+
tune_llm: false
|
| 25 |
+
tune_visual: false
|
| 26 |
+
tune_projector: true
|
| 27 |
+
tune_diffusion_model: true
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/train_5shot_seed0_fullactionhead_5k/checkpoint-1000/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:61046d6d7e9c58345891562a0fe071dd9475ab479c8ad61ed6d04127525d516c
|
| 3 |
+
size 4999367032
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/train_5shot_seed0_fullactionhead_5k/checkpoint-1000/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1710a8ef3e45be4a163c4f962d41466205343411de6f144fc6a83a28b8f503bf
|
| 3 |
+
size 2586705312
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/train_5shot_seed0_fullactionhead_5k/checkpoint-1000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bbf47a3b7290c0b108a84d6a1be53d7485caf86c3052476f4f13b3b4f8067bef
|
| 3 |
+
size 14645
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/train_5shot_seed0_fullactionhead_5k/checkpoint-1000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6e19ead5779e3f8f2f666b7793de3a03d118f57f1b16a97eeb87bd732d18b445
|
| 3 |
+
size 1465
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/train_5shot_seed0_fullactionhead_5k/checkpoint-3000/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 |
+
}
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/train_5shot_seed0_fullactionhead_5k/checkpoint-3000/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e25f0b02d749bea2cca2fdfc5adf54ab4d1ebc771d7dee2a89d0784d30c57367
|
| 3 |
+
size 4999367032
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/train_5shot_seed0_fullactionhead_5k/checkpoint-3000/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0bb2f052e8c73380d13c16d9797f5516bcfbe2b88b40b1b94885f5ac80b6e0e6
|
| 3 |
+
size 2586705312
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/train_5shot_seed0_fullactionhead_5k/checkpoint-3000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2cfb9120e2de97798d6ff37a711f2ac1f7397b57a59789d33d9833d2d22d5cab
|
| 3 |
+
size 14645
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/train_5shot_seed0_fullactionhead_5k/checkpoint-3000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:42d8e6d8e53e6b837084d0cc18fdd63ca9e40bd44a65d38eb52695647a8e84e8
|
| 3 |
+
size 1465
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/train_5shot_seed0_fullactionhead_5k/checkpoint-3000/trainer_state.json
ADDED
|
@@ -0,0 +1,2134 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 5.033557046979865,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 3000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.016778523489932886,
|
| 14 |
+
"grad_norm": 1.0388541221618652,
|
| 15 |
+
"learning_rate": 1.08e-06,
|
| 16 |
+
"loss": 0.2251,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.03355704697986577,
|
| 21 |
+
"grad_norm": 2.1458616256713867,
|
| 22 |
+
"learning_rate": 2.28e-06,
|
| 23 |
+
"loss": 0.2286,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.050335570469798654,
|
| 28 |
+
"grad_norm": 1.0232809782028198,
|
| 29 |
+
"learning_rate": 3.48e-06,
|
| 30 |
+
"loss": 0.1901,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.06711409395973154,
|
| 35 |
+
"grad_norm": 0.6550730466842651,
|
| 36 |
+
"learning_rate": 4.68e-06,
|
| 37 |
+
"loss": 0.1701,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08389261744966443,
|
| 42 |
+
"grad_norm": 0.6718728542327881,
|
| 43 |
+
"learning_rate": 5.8800000000000005e-06,
|
| 44 |
+
"loss": 0.1338,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.10067114093959731,
|
| 49 |
+
"grad_norm": 0.5689961910247803,
|
| 50 |
+
"learning_rate": 7.08e-06,
|
| 51 |
+
"loss": 0.1274,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.1174496644295302,
|
| 56 |
+
"grad_norm": 0.4747137129306793,
|
| 57 |
+
"learning_rate": 8.28e-06,
|
| 58 |
+
"loss": 0.1026,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.1342281879194631,
|
| 63 |
+
"grad_norm": 0.3482891023159027,
|
| 64 |
+
"learning_rate": 9.48e-06,
|
| 65 |
+
"loss": 0.0929,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.15100671140939598,
|
| 70 |
+
"grad_norm": 0.3731693923473358,
|
| 71 |
+
"learning_rate": 1.068e-05,
|
| 72 |
+
"loss": 0.0835,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.16778523489932887,
|
| 77 |
+
"grad_norm": 0.3104197680950165,
|
| 78 |
+
"learning_rate": 1.1880000000000001e-05,
|
| 79 |
+
"loss": 0.0756,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.18456375838926176,
|
| 84 |
+
"grad_norm": 0.2531762421131134,
|
| 85 |
+
"learning_rate": 1.308e-05,
|
| 86 |
+
"loss": 0.0726,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.20134228187919462,
|
| 91 |
+
"grad_norm": 0.29734447598457336,
|
| 92 |
+
"learning_rate": 1.428e-05,
|
| 93 |
+
"loss": 0.0666,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.2181208053691275,
|
| 98 |
+
"grad_norm": 0.24068564176559448,
|
| 99 |
+
"learning_rate": 1.548e-05,
|
| 100 |
+
"loss": 0.0632,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.2348993288590604,
|
| 105 |
+
"grad_norm": 0.3159964382648468,
|
| 106 |
+
"learning_rate": 1.6680000000000003e-05,
|
| 107 |
+
"loss": 0.0647,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.2516778523489933,
|
| 112 |
+
"grad_norm": 0.23858052492141724,
|
| 113 |
+
"learning_rate": 1.7879999999999998e-05,
|
| 114 |
+
"loss": 0.065,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.2684563758389262,
|
| 119 |
+
"grad_norm": 0.2374449223279953,
|
| 120 |
+
"learning_rate": 1.908e-05,
|
| 121 |
+
"loss": 0.0601,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.28523489932885904,
|
| 126 |
+
"grad_norm": 0.20790916681289673,
|
| 127 |
+
"learning_rate": 2.0280000000000002e-05,
|
| 128 |
+
"loss": 0.0593,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.30201342281879195,
|
| 133 |
+
"grad_norm": 0.1952684223651886,
|
| 134 |
+
"learning_rate": 2.148e-05,
|
| 135 |
+
"loss": 0.0527,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.3187919463087248,
|
| 140 |
+
"grad_norm": 0.3542490601539612,
|
| 141 |
+
"learning_rate": 2.268e-05,
|
| 142 |
+
"loss": 0.0583,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.33557046979865773,
|
| 147 |
+
"grad_norm": 0.28306496143341064,
|
| 148 |
+
"learning_rate": 2.3880000000000002e-05,
|
| 149 |
+
"loss": 0.057,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.3523489932885906,
|
| 154 |
+
"grad_norm": 0.2882148325443268,
|
| 155 |
+
"learning_rate": 2.508e-05,
|
| 156 |
+
"loss": 0.0541,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.3691275167785235,
|
| 161 |
+
"grad_norm": 0.19627629220485687,
|
| 162 |
+
"learning_rate": 2.628e-05,
|
| 163 |
+
"loss": 0.0552,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.3859060402684564,
|
| 168 |
+
"grad_norm": 0.1990572214126587,
|
| 169 |
+
"learning_rate": 2.748e-05,
|
| 170 |
+
"loss": 0.0506,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.40268456375838924,
|
| 175 |
+
"grad_norm": 0.24517984688282013,
|
| 176 |
+
"learning_rate": 2.868e-05,
|
| 177 |
+
"loss": 0.0528,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.41946308724832215,
|
| 182 |
+
"grad_norm": 0.2502721846103668,
|
| 183 |
+
"learning_rate": 2.9880000000000002e-05,
|
| 184 |
+
"loss": 0.0494,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.436241610738255,
|
| 189 |
+
"grad_norm": 0.18276523053646088,
|
| 190 |
+
"learning_rate": 2.9999734259635676e-05,
|
| 191 |
+
"loss": 0.0451,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.45302013422818793,
|
| 196 |
+
"grad_norm": 0.2971280813217163,
|
| 197 |
+
"learning_rate": 2.9998815663057245e-05,
|
| 198 |
+
"loss": 0.0488,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.4697986577181208,
|
| 203 |
+
"grad_norm": 0.20258396863937378,
|
| 204 |
+
"learning_rate": 2.999724096969228e-05,
|
| 205 |
+
"loss": 0.0454,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.4865771812080537,
|
| 210 |
+
"grad_norm": 0.27414819598197937,
|
| 211 |
+
"learning_rate": 2.9995010248422984e-05,
|
| 212 |
+
"loss": 0.0436,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.5033557046979866,
|
| 217 |
+
"grad_norm": 0.2332964837551117,
|
| 218 |
+
"learning_rate": 2.999212359682832e-05,
|
| 219 |
+
"loss": 0.0459,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.5201342281879194,
|
| 224 |
+
"grad_norm": 0.23066847026348114,
|
| 225 |
+
"learning_rate": 2.9988581141179787e-05,
|
| 226 |
+
"loss": 0.0446,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.5369127516778524,
|
| 231 |
+
"grad_norm": 0.21959786117076874,
|
| 232 |
+
"learning_rate": 2.998438303643587e-05,
|
| 233 |
+
"loss": 0.0401,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.5536912751677853,
|
| 238 |
+
"grad_norm": 0.1958390325307846,
|
| 239 |
+
"learning_rate": 2.997952946623527e-05,
|
| 240 |
+
"loss": 0.0418,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.5704697986577181,
|
| 245 |
+
"grad_norm": 0.23207500576972961,
|
| 246 |
+
"learning_rate": 2.9974020642888875e-05,
|
| 247 |
+
"loss": 0.0455,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.587248322147651,
|
| 252 |
+
"grad_norm": 0.1922421157360077,
|
| 253 |
+
"learning_rate": 2.9967856807370458e-05,
|
| 254 |
+
"loss": 0.0413,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.6040268456375839,
|
| 259 |
+
"grad_norm": 0.20857904851436615,
|
| 260 |
+
"learning_rate": 2.996103822930615e-05,
|
| 261 |
+
"loss": 0.0445,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.6208053691275168,
|
| 266 |
+
"grad_norm": 0.20092910528182983,
|
| 267 |
+
"learning_rate": 2.9953565206962653e-05,
|
| 268 |
+
"loss": 0.0376,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.6375838926174496,
|
| 273 |
+
"grad_norm": 0.17054688930511475,
|
| 274 |
+
"learning_rate": 2.9945438067234172e-05,
|
| 275 |
+
"loss": 0.039,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.6543624161073825,
|
| 280 |
+
"grad_norm": 0.18088215589523315,
|
| 281 |
+
"learning_rate": 2.9936657165628124e-05,
|
| 282 |
+
"loss": 0.0385,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.6711409395973155,
|
| 287 |
+
"grad_norm": 0.22038601338863373,
|
| 288 |
+
"learning_rate": 2.992722288624959e-05,
|
| 289 |
+
"loss": 0.0381,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.6879194630872483,
|
| 294 |
+
"grad_norm": 0.16217142343521118,
|
| 295 |
+
"learning_rate": 2.9917135641784515e-05,
|
| 296 |
+
"loss": 0.0384,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.7046979865771812,
|
| 301 |
+
"grad_norm": 0.1879982352256775,
|
| 302 |
+
"learning_rate": 2.9906395873481655e-05,
|
| 303 |
+
"loss": 0.0375,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.7214765100671141,
|
| 308 |
+
"grad_norm": 0.1811523288488388,
|
| 309 |
+
"learning_rate": 2.989500405113326e-05,
|
| 310 |
+
"loss": 0.0391,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.738255033557047,
|
| 315 |
+
"grad_norm": 0.15471374988555908,
|
| 316 |
+
"learning_rate": 2.9882960673054547e-05,
|
| 317 |
+
"loss": 0.0347,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.7550335570469798,
|
| 322 |
+
"grad_norm": 0.20345190167427063,
|
| 323 |
+
"learning_rate": 2.9870266266061886e-05,
|
| 324 |
+
"loss": 0.034,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.7718120805369127,
|
| 329 |
+
"grad_norm": 0.22297805547714233,
|
| 330 |
+
"learning_rate": 2.9856921385449766e-05,
|
| 331 |
+
"loss": 0.0364,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.7885906040268457,
|
| 336 |
+
"grad_norm": 0.18335476517677307,
|
| 337 |
+
"learning_rate": 2.9842926614966502e-05,
|
| 338 |
+
"loss": 0.0355,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.8053691275167785,
|
| 343 |
+
"grad_norm": 0.21670755743980408,
|
| 344 |
+
"learning_rate": 2.982828256678868e-05,
|
| 345 |
+
"loss": 0.0356,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.8221476510067114,
|
| 350 |
+
"grad_norm": 0.22311744093894958,
|
| 351 |
+
"learning_rate": 2.9812989881494412e-05,
|
| 352 |
+
"loss": 0.0345,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.8389261744966443,
|
| 357 |
+
"grad_norm": 0.1998075544834137,
|
| 358 |
+
"learning_rate": 2.9797049228035296e-05,
|
| 359 |
+
"loss": 0.0377,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.8557046979865772,
|
| 364 |
+
"grad_norm": 0.19125983119010925,
|
| 365 |
+
"learning_rate": 2.978046130370715e-05,
|
| 366 |
+
"loss": 0.0334,
|
| 367 |
+
"step": 510
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.87248322147651,
|
| 371 |
+
"grad_norm": 0.2721741497516632,
|
| 372 |
+
"learning_rate": 2.9763226834119535e-05,
|
| 373 |
+
"loss": 0.0355,
|
| 374 |
+
"step": 520
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.889261744966443,
|
| 378 |
+
"grad_norm": 0.20220515131950378,
|
| 379 |
+
"learning_rate": 2.9745346573163963e-05,
|
| 380 |
+
"loss": 0.0345,
|
| 381 |
+
"step": 530
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.9060402684563759,
|
| 385 |
+
"grad_norm": 0.29605722427368164,
|
| 386 |
+
"learning_rate": 2.9726821302980987e-05,
|
| 387 |
+
"loss": 0.0338,
|
| 388 |
+
"step": 540
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.9228187919463087,
|
| 392 |
+
"grad_norm": 0.1908901482820511,
|
| 393 |
+
"learning_rate": 2.9707651833925935e-05,
|
| 394 |
+
"loss": 0.0306,
|
| 395 |
+
"step": 550
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.9395973154362416,
|
| 399 |
+
"grad_norm": 0.18491318821907043,
|
| 400 |
+
"learning_rate": 2.968783900453348e-05,
|
| 401 |
+
"loss": 0.034,
|
| 402 |
+
"step": 560
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.9563758389261745,
|
| 406 |
+
"grad_norm": 0.15446802973747253,
|
| 407 |
+
"learning_rate": 2.9667383681480968e-05,
|
| 408 |
+
"loss": 0.0365,
|
| 409 |
+
"step": 570
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.9731543624161074,
|
| 413 |
+
"grad_norm": 0.18752741813659668,
|
| 414 |
+
"learning_rate": 2.9646286759550493e-05,
|
| 415 |
+
"loss": 0.032,
|
| 416 |
+
"step": 580
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.9899328859060402,
|
| 420 |
+
"grad_norm": 0.4135258197784424,
|
| 421 |
+
"learning_rate": 2.9624549161589764e-05,
|
| 422 |
+
"loss": 0.031,
|
| 423 |
+
"step": 590
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 1.0067114093959733,
|
| 427 |
+
"grad_norm": 0.19306783378124237,
|
| 428 |
+
"learning_rate": 2.9602171838471745e-05,
|
| 429 |
+
"loss": 0.0314,
|
| 430 |
+
"step": 600
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 1.023489932885906,
|
| 434 |
+
"grad_norm": 0.1685028225183487,
|
| 435 |
+
"learning_rate": 2.957915576905303e-05,
|
| 436 |
+
"loss": 0.03,
|
| 437 |
+
"step": 610
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 1.0402684563758389,
|
| 441 |
+
"grad_norm": 0.15813031792640686,
|
| 442 |
+
"learning_rate": 2.9555501960131058e-05,
|
| 443 |
+
"loss": 0.0287,
|
| 444 |
+
"step": 620
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 1.0570469798657718,
|
| 448 |
+
"grad_norm": 0.16480985283851624,
|
| 449 |
+
"learning_rate": 2.9531211446400068e-05,
|
| 450 |
+
"loss": 0.0326,
|
| 451 |
+
"step": 630
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 1.0738255033557047,
|
| 455 |
+
"grad_norm": 0.16310042142868042,
|
| 456 |
+
"learning_rate": 2.9506285290405813e-05,
|
| 457 |
+
"loss": 0.0319,
|
| 458 |
+
"step": 640
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 1.0906040268456376,
|
| 462 |
+
"grad_norm": 0.28549784421920776,
|
| 463 |
+
"learning_rate": 2.948072458249911e-05,
|
| 464 |
+
"loss": 0.0308,
|
| 465 |
+
"step": 650
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 1.1073825503355705,
|
| 469 |
+
"grad_norm": 0.4056604206562042,
|
| 470 |
+
"learning_rate": 2.9454530440788125e-05,
|
| 471 |
+
"loss": 0.0298,
|
| 472 |
+
"step": 660
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 1.1241610738255035,
|
| 476 |
+
"grad_norm": 0.24711917340755463,
|
| 477 |
+
"learning_rate": 2.9427704011089474e-05,
|
| 478 |
+
"loss": 0.0287,
|
| 479 |
+
"step": 670
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 1.1409395973154361,
|
| 483 |
+
"grad_norm": 0.2376968264579773,
|
| 484 |
+
"learning_rate": 2.94002464668781e-05,
|
| 485 |
+
"loss": 0.0281,
|
| 486 |
+
"step": 680
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 1.157718120805369,
|
| 490 |
+
"grad_norm": 0.1800604611635208,
|
| 491 |
+
"learning_rate": 2.9372159009235936e-05,
|
| 492 |
+
"loss": 0.0269,
|
| 493 |
+
"step": 690
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 1.174496644295302,
|
| 497 |
+
"grad_norm": 0.15358570218086243,
|
| 498 |
+
"learning_rate": 2.9343442866799365e-05,
|
| 499 |
+
"loss": 0.0277,
|
| 500 |
+
"step": 700
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 1.191275167785235,
|
| 504 |
+
"grad_norm": 0.28675392270088196,
|
| 505 |
+
"learning_rate": 2.9314099295705495e-05,
|
| 506 |
+
"loss": 0.0297,
|
| 507 |
+
"step": 710
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 1.2080536912751678,
|
| 511 |
+
"grad_norm": 0.20398671925067902,
|
| 512 |
+
"learning_rate": 2.928412957953718e-05,
|
| 513 |
+
"loss": 0.0275,
|
| 514 |
+
"step": 720
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 1.2248322147651007,
|
| 518 |
+
"grad_norm": 0.2259288877248764,
|
| 519 |
+
"learning_rate": 2.925353502926689e-05,
|
| 520 |
+
"loss": 0.0296,
|
| 521 |
+
"step": 730
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 1.2416107382550337,
|
| 525 |
+
"grad_norm": 0.16415706276893616,
|
| 526 |
+
"learning_rate": 2.922231698319937e-05,
|
| 527 |
+
"loss": 0.0291,
|
| 528 |
+
"step": 740
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 1.2583892617449663,
|
| 532 |
+
"grad_norm": 0.19491496682167053,
|
| 533 |
+
"learning_rate": 2.9190476806913088e-05,
|
| 534 |
+
"loss": 0.0295,
|
| 535 |
+
"step": 750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 1.2751677852348993,
|
| 539 |
+
"grad_norm": 0.18139910697937012,
|
| 540 |
+
"learning_rate": 2.9158015893200504e-05,
|
| 541 |
+
"loss": 0.0285,
|
| 542 |
+
"step": 760
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 1.2919463087248322,
|
| 546 |
+
"grad_norm": 0.18600593507289886,
|
| 547 |
+
"learning_rate": 2.9124935662007137e-05,
|
| 548 |
+
"loss": 0.0307,
|
| 549 |
+
"step": 770
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.308724832214765,
|
| 553 |
+
"grad_norm": 0.16569755971431732,
|
| 554 |
+
"learning_rate": 2.9091237560369467e-05,
|
| 555 |
+
"loss": 0.0289,
|
| 556 |
+
"step": 780
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 1.325503355704698,
|
| 560 |
+
"grad_norm": 0.1949482262134552,
|
| 561 |
+
"learning_rate": 2.905692306235163e-05,
|
| 562 |
+
"loss": 0.0283,
|
| 563 |
+
"step": 790
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 1.342281879194631,
|
| 567 |
+
"grad_norm": 0.2368042767047882,
|
| 568 |
+
"learning_rate": 2.9021993668980925e-05,
|
| 569 |
+
"loss": 0.0279,
|
| 570 |
+
"step": 800
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 1.3590604026845639,
|
| 574 |
+
"grad_norm": 0.21733024716377258,
|
| 575 |
+
"learning_rate": 2.8986450908182185e-05,
|
| 576 |
+
"loss": 0.0257,
|
| 577 |
+
"step": 810
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 1.3758389261744965,
|
| 581 |
+
"grad_norm": 0.16328634321689606,
|
| 582 |
+
"learning_rate": 2.8950296334710902e-05,
|
| 583 |
+
"loss": 0.0281,
|
| 584 |
+
"step": 820
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 1.3926174496644295,
|
| 588 |
+
"grad_norm": 0.20662836730480194,
|
| 589 |
+
"learning_rate": 2.8913531530085248e-05,
|
| 590 |
+
"loss": 0.0272,
|
| 591 |
+
"step": 830
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 1.4093959731543624,
|
| 595 |
+
"grad_norm": 0.17186735570430756,
|
| 596 |
+
"learning_rate": 2.887615810251687e-05,
|
| 597 |
+
"loss": 0.0286,
|
| 598 |
+
"step": 840
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 1.4261744966442953,
|
| 602 |
+
"grad_norm": 0.187661275267601,
|
| 603 |
+
"learning_rate": 2.8838177686840577e-05,
|
| 604 |
+
"loss": 0.0284,
|
| 605 |
+
"step": 850
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 1.4429530201342282,
|
| 609 |
+
"grad_norm": 0.18711018562316895,
|
| 610 |
+
"learning_rate": 2.8799591944442776e-05,
|
| 611 |
+
"loss": 0.0289,
|
| 612 |
+
"step": 860
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 1.4597315436241611,
|
| 616 |
+
"grad_norm": 0.2038484513759613,
|
| 617 |
+
"learning_rate": 2.8760402563188848e-05,
|
| 618 |
+
"loss": 0.0279,
|
| 619 |
+
"step": 870
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 1.476510067114094,
|
| 623 |
+
"grad_norm": 0.20940718054771423,
|
| 624 |
+
"learning_rate": 2.8720611257349285e-05,
|
| 625 |
+
"loss": 0.0269,
|
| 626 |
+
"step": 880
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 1.4932885906040267,
|
| 630 |
+
"grad_norm": 0.19567424058914185,
|
| 631 |
+
"learning_rate": 2.868021976752471e-05,
|
| 632 |
+
"loss": 0.027,
|
| 633 |
+
"step": 890
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 1.5100671140939599,
|
| 637 |
+
"grad_norm": 0.16011153161525726,
|
| 638 |
+
"learning_rate": 2.8639229860569738e-05,
|
| 639 |
+
"loss": 0.0293,
|
| 640 |
+
"step": 900
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 1.5268456375838926,
|
| 644 |
+
"grad_norm": 0.13916991651058197,
|
| 645 |
+
"learning_rate": 2.8597643329515688e-05,
|
| 646 |
+
"loss": 0.0253,
|
| 647 |
+
"step": 910
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 1.5436241610738255,
|
| 651 |
+
"grad_norm": 0.12822005152702332,
|
| 652 |
+
"learning_rate": 2.855546199349214e-05,
|
| 653 |
+
"loss": 0.025,
|
| 654 |
+
"step": 920
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 1.5604026845637584,
|
| 658 |
+
"grad_norm": 0.18722663819789886,
|
| 659 |
+
"learning_rate": 2.85126876976474e-05,
|
| 660 |
+
"loss": 0.0277,
|
| 661 |
+
"step": 930
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 1.5771812080536913,
|
| 665 |
+
"grad_norm": 0.1512211114168167,
|
| 666 |
+
"learning_rate": 2.8469322313067726e-05,
|
| 667 |
+
"loss": 0.0253,
|
| 668 |
+
"step": 940
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 1.5939597315436242,
|
| 672 |
+
"grad_norm": 0.20754490792751312,
|
| 673 |
+
"learning_rate": 2.8425367736695522e-05,
|
| 674 |
+
"loss": 0.0254,
|
| 675 |
+
"step": 950
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 1.610738255033557,
|
| 679 |
+
"grad_norm": 0.238246351480484,
|
| 680 |
+
"learning_rate": 2.838082589124636e-05,
|
| 681 |
+
"loss": 0.028,
|
| 682 |
+
"step": 960
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 1.62751677852349,
|
| 686 |
+
"grad_norm": 0.14989012479782104,
|
| 687 |
+
"learning_rate": 2.8335698725124843e-05,
|
| 688 |
+
"loss": 0.0261,
|
| 689 |
+
"step": 970
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 1.6442953020134228,
|
| 693 |
+
"grad_norm": 0.15159383416175842,
|
| 694 |
+
"learning_rate": 2.8289988212339422e-05,
|
| 695 |
+
"loss": 0.0232,
|
| 696 |
+
"step": 980
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 1.6610738255033557,
|
| 700 |
+
"grad_norm": 0.17884178459644318,
|
| 701 |
+
"learning_rate": 2.8243696352415997e-05,
|
| 702 |
+
"loss": 0.0303,
|
| 703 |
+
"step": 990
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 1.6778523489932886,
|
| 707 |
+
"grad_norm": 0.1455809772014618,
|
| 708 |
+
"learning_rate": 2.8196825170310495e-05,
|
| 709 |
+
"loss": 0.0239,
|
| 710 |
+
"step": 1000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 1.6946308724832215,
|
| 714 |
+
"grad_norm": 0.15527848899364471,
|
| 715 |
+
"learning_rate": 2.8149376716320256e-05,
|
| 716 |
+
"loss": 0.0255,
|
| 717 |
+
"step": 1010
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 1.7114093959731544,
|
| 721 |
+
"grad_norm": 0.16052792966365814,
|
| 722 |
+
"learning_rate": 2.8101353065994373e-05,
|
| 723 |
+
"loss": 0.0276,
|
| 724 |
+
"step": 1020
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 1.7281879194630871,
|
| 728 |
+
"grad_norm": 0.12123282998800278,
|
| 729 |
+
"learning_rate": 2.8052756320042885e-05,
|
| 730 |
+
"loss": 0.025,
|
| 731 |
+
"step": 1030
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 1.7449664429530203,
|
| 735 |
+
"grad_norm": 0.16650134325027466,
|
| 736 |
+
"learning_rate": 2.8003588604244893e-05,
|
| 737 |
+
"loss": 0.0232,
|
| 738 |
+
"step": 1040
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 1.761744966442953,
|
| 742 |
+
"grad_norm": 0.14930953085422516,
|
| 743 |
+
"learning_rate": 2.7953852069355565e-05,
|
| 744 |
+
"loss": 0.0225,
|
| 745 |
+
"step": 1050
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 1.778523489932886,
|
| 749 |
+
"grad_norm": 0.21071627736091614,
|
| 750 |
+
"learning_rate": 2.790354889101206e-05,
|
| 751 |
+
"loss": 0.0212,
|
| 752 |
+
"step": 1060
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 1.7953020134228188,
|
| 756 |
+
"grad_norm": 0.1896669566631317,
|
| 757 |
+
"learning_rate": 2.7852681269638366e-05,
|
| 758 |
+
"loss": 0.0259,
|
| 759 |
+
"step": 1070
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 1.8120805369127517,
|
| 763 |
+
"grad_norm": 0.15259258449077606,
|
| 764 |
+
"learning_rate": 2.780125143034902e-05,
|
| 765 |
+
"loss": 0.0263,
|
| 766 |
+
"step": 1080
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 1.8288590604026846,
|
| 770 |
+
"grad_norm": 0.14170797169208527,
|
| 771 |
+
"learning_rate": 2.7749261622851812e-05,
|
| 772 |
+
"loss": 0.0228,
|
| 773 |
+
"step": 1090
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.8456375838926173,
|
| 777 |
+
"grad_norm": 0.14577420055866241,
|
| 778 |
+
"learning_rate": 2.7696714121349335e-05,
|
| 779 |
+
"loss": 0.0233,
|
| 780 |
+
"step": 1100
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.8624161073825505,
|
| 784 |
+
"grad_norm": 0.1250564157962799,
|
| 785 |
+
"learning_rate": 2.764361122443954e-05,
|
| 786 |
+
"loss": 0.0218,
|
| 787 |
+
"step": 1110
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.8791946308724832,
|
| 791 |
+
"grad_norm": 0.16452714800834656,
|
| 792 |
+
"learning_rate": 2.758995525501517e-05,
|
| 793 |
+
"loss": 0.023,
|
| 794 |
+
"step": 1120
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.895973154362416,
|
| 798 |
+
"grad_norm": 0.17396444082260132,
|
| 799 |
+
"learning_rate": 2.753574856016215e-05,
|
| 800 |
+
"loss": 0.0231,
|
| 801 |
+
"step": 1130
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 1.912751677852349,
|
| 805 |
+
"grad_norm": 0.15328766405582428,
|
| 806 |
+
"learning_rate": 2.7480993511056913e-05,
|
| 807 |
+
"loss": 0.024,
|
| 808 |
+
"step": 1140
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 1.929530201342282,
|
| 812 |
+
"grad_norm": 0.1501884162425995,
|
| 813 |
+
"learning_rate": 2.7425692502862684e-05,
|
| 814 |
+
"loss": 0.0211,
|
| 815 |
+
"step": 1150
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 1.9463087248322148,
|
| 819 |
+
"grad_norm": 0.1589355766773224,
|
| 820 |
+
"learning_rate": 2.7369847954624728e-05,
|
| 821 |
+
"loss": 0.0232,
|
| 822 |
+
"step": 1160
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 1.9630872483221475,
|
| 826 |
+
"grad_norm": 0.12628960609436035,
|
| 827 |
+
"learning_rate": 2.731346230916448e-05,
|
| 828 |
+
"loss": 0.0254,
|
| 829 |
+
"step": 1170
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 1.9798657718120807,
|
| 833 |
+
"grad_norm": 0.14850759506225586,
|
| 834 |
+
"learning_rate": 2.7256538032972746e-05,
|
| 835 |
+
"loss": 0.0231,
|
| 836 |
+
"step": 1180
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 1.9966442953020134,
|
| 840 |
+
"grad_norm": 0.12806259095668793,
|
| 841 |
+
"learning_rate": 2.7199077616101773e-05,
|
| 842 |
+
"loss": 0.0209,
|
| 843 |
+
"step": 1190
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 2.0134228187919465,
|
| 847 |
+
"grad_norm": 0.19582599401474,
|
| 848 |
+
"learning_rate": 2.714108357205634e-05,
|
| 849 |
+
"loss": 0.0238,
|
| 850 |
+
"step": 1200
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 2.030201342281879,
|
| 854 |
+
"grad_norm": 0.41255050897598267,
|
| 855 |
+
"learning_rate": 2.7082558437683803e-05,
|
| 856 |
+
"loss": 0.0218,
|
| 857 |
+
"step": 1210
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 2.046979865771812,
|
| 861 |
+
"grad_norm": 0.15883375704288483,
|
| 862 |
+
"learning_rate": 2.702350477306315e-05,
|
| 863 |
+
"loss": 0.0217,
|
| 864 |
+
"step": 1220
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 2.063758389261745,
|
| 868 |
+
"grad_norm": 0.21646834909915924,
|
| 869 |
+
"learning_rate": 2.6963925161392963e-05,
|
| 870 |
+
"loss": 0.0229,
|
| 871 |
+
"step": 1230
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 2.0805369127516777,
|
| 875 |
+
"grad_norm": 0.14378032088279724,
|
| 876 |
+
"learning_rate": 2.6903822208878476e-05,
|
| 877 |
+
"loss": 0.0225,
|
| 878 |
+
"step": 1240
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 2.097315436241611,
|
| 882 |
+
"grad_norm": 0.15373851358890533,
|
| 883 |
+
"learning_rate": 2.6843198544617535e-05,
|
| 884 |
+
"loss": 0.0241,
|
| 885 |
+
"step": 1250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 2.1140939597315436,
|
| 889 |
+
"grad_norm": 0.1561702936887741,
|
| 890 |
+
"learning_rate": 2.6782056820485612e-05,
|
| 891 |
+
"loss": 0.0228,
|
| 892 |
+
"step": 1260
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 2.1308724832214767,
|
| 896 |
+
"grad_norm": 0.15469935536384583,
|
| 897 |
+
"learning_rate": 2.672039971101979e-05,
|
| 898 |
+
"loss": 0.021,
|
| 899 |
+
"step": 1270
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 2.1476510067114094,
|
| 903 |
+
"grad_norm": 0.11616158485412598,
|
| 904 |
+
"learning_rate": 2.665822991330178e-05,
|
| 905 |
+
"loss": 0.0244,
|
| 906 |
+
"step": 1280
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 2.164429530201342,
|
| 910 |
+
"grad_norm": 0.18575182557106018,
|
| 911 |
+
"learning_rate": 2.6595550146839934e-05,
|
| 912 |
+
"loss": 0.0233,
|
| 913 |
+
"step": 1290
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 2.1812080536912752,
|
| 917 |
+
"grad_norm": 0.1657201200723648,
|
| 918 |
+
"learning_rate": 2.6532363153450287e-05,
|
| 919 |
+
"loss": 0.0219,
|
| 920 |
+
"step": 1300
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 2.197986577181208,
|
| 924 |
+
"grad_norm": 0.17577281594276428,
|
| 925 |
+
"learning_rate": 2.6468671697136634e-05,
|
| 926 |
+
"loss": 0.0206,
|
| 927 |
+
"step": 1310
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 2.214765100671141,
|
| 931 |
+
"grad_norm": 0.16942515969276428,
|
| 932 |
+
"learning_rate": 2.640447856396961e-05,
|
| 933 |
+
"loss": 0.0202,
|
| 934 |
+
"step": 1320
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 2.2315436241610738,
|
| 938 |
+
"grad_norm": 0.12942443788051605,
|
| 939 |
+
"learning_rate": 2.6339786561964805e-05,
|
| 940 |
+
"loss": 0.0235,
|
| 941 |
+
"step": 1330
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 2.248322147651007,
|
| 945 |
+
"grad_norm": 0.10895993560552597,
|
| 946 |
+
"learning_rate": 2.6274598520959966e-05,
|
| 947 |
+
"loss": 0.0217,
|
| 948 |
+
"step": 1340
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 2.2651006711409396,
|
| 952 |
+
"grad_norm": 0.16702544689178467,
|
| 953 |
+
"learning_rate": 2.6208917292491185e-05,
|
| 954 |
+
"loss": 0.0229,
|
| 955 |
+
"step": 1350
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 2.2818791946308723,
|
| 959 |
+
"grad_norm": 0.14049775898456573,
|
| 960 |
+
"learning_rate": 2.6142745749668176e-05,
|
| 961 |
+
"loss": 0.0213,
|
| 962 |
+
"step": 1360
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 2.2986577181208054,
|
| 966 |
+
"grad_norm": 0.17712368071079254,
|
| 967 |
+
"learning_rate": 2.607608678704859e-05,
|
| 968 |
+
"loss": 0.0222,
|
| 969 |
+
"step": 1370
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 2.315436241610738,
|
| 973 |
+
"grad_norm": 0.17040906846523285,
|
| 974 |
+
"learning_rate": 2.600894332051139e-05,
|
| 975 |
+
"loss": 0.0191,
|
| 976 |
+
"step": 1380
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 2.3322147651006713,
|
| 980 |
+
"grad_norm": 0.1793813854455948,
|
| 981 |
+
"learning_rate": 2.594131828712932e-05,
|
| 982 |
+
"loss": 0.0226,
|
| 983 |
+
"step": 1390
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 2.348993288590604,
|
| 987 |
+
"grad_norm": 0.17199549078941345,
|
| 988 |
+
"learning_rate": 2.5873214645040414e-05,
|
| 989 |
+
"loss": 0.0205,
|
| 990 |
+
"step": 1400
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 2.365771812080537,
|
| 994 |
+
"grad_norm": 0.2560262680053711,
|
| 995 |
+
"learning_rate": 2.5804635373318604e-05,
|
| 996 |
+
"loss": 0.0194,
|
| 997 |
+
"step": 1410
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 2.38255033557047,
|
| 1001 |
+
"grad_norm": 0.15049883723258972,
|
| 1002 |
+
"learning_rate": 2.57355834718434e-05,
|
| 1003 |
+
"loss": 0.0225,
|
| 1004 |
+
"step": 1420
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 2.3993288590604025,
|
| 1008 |
+
"grad_norm": 0.10687083005905151,
|
| 1009 |
+
"learning_rate": 2.566606196116867e-05,
|
| 1010 |
+
"loss": 0.0201,
|
| 1011 |
+
"step": 1430
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 2.4161073825503356,
|
| 1015 |
+
"grad_norm": 0.21092043817043304,
|
| 1016 |
+
"learning_rate": 2.559607388239051e-05,
|
| 1017 |
+
"loss": 0.0203,
|
| 1018 |
+
"step": 1440
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 2.4328859060402683,
|
| 1022 |
+
"grad_norm": 0.1304531693458557,
|
| 1023 |
+
"learning_rate": 2.552562229701422e-05,
|
| 1024 |
+
"loss": 0.0207,
|
| 1025 |
+
"step": 1450
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 2.4496644295302015,
|
| 1029 |
+
"grad_norm": 0.17584650218486786,
|
| 1030 |
+
"learning_rate": 2.545471028682036e-05,
|
| 1031 |
+
"loss": 0.0225,
|
| 1032 |
+
"step": 1460
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 2.466442953020134,
|
| 1036 |
+
"grad_norm": 0.28326791524887085,
|
| 1037 |
+
"learning_rate": 2.538334095373e-05,
|
| 1038 |
+
"loss": 0.0199,
|
| 1039 |
+
"step": 1470
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 2.4832214765100673,
|
| 1043 |
+
"grad_norm": 0.2687714695930481,
|
| 1044 |
+
"learning_rate": 2.531151741966896e-05,
|
| 1045 |
+
"loss": 0.021,
|
| 1046 |
+
"step": 1480
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 2.5,
|
| 1050 |
+
"grad_norm": 0.12673330307006836,
|
| 1051 |
+
"learning_rate": 2.5239242826431302e-05,
|
| 1052 |
+
"loss": 0.0212,
|
| 1053 |
+
"step": 1490
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 2.5167785234899327,
|
| 1057 |
+
"grad_norm": 0.14194728434085846,
|
| 1058 |
+
"learning_rate": 2.5166520335541865e-05,
|
| 1059 |
+
"loss": 0.0202,
|
| 1060 |
+
"step": 1500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 2.533557046979866,
|
| 1064 |
+
"grad_norm": 0.2310604751110077,
|
| 1065 |
+
"learning_rate": 2.5093353128117996e-05,
|
| 1066 |
+
"loss": 0.0213,
|
| 1067 |
+
"step": 1510
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 2.5503355704697985,
|
| 1071 |
+
"grad_norm": 0.13804496824741364,
|
| 1072 |
+
"learning_rate": 2.501974440473037e-05,
|
| 1073 |
+
"loss": 0.0193,
|
| 1074 |
+
"step": 1520
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 2.5671140939597317,
|
| 1078 |
+
"grad_norm": 0.14828719198703766,
|
| 1079 |
+
"learning_rate": 2.4945697385263015e-05,
|
| 1080 |
+
"loss": 0.0222,
|
| 1081 |
+
"step": 1530
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 2.5838926174496644,
|
| 1085 |
+
"grad_norm": 0.1583152413368225,
|
| 1086 |
+
"learning_rate": 2.487121530877243e-05,
|
| 1087 |
+
"loss": 0.0218,
|
| 1088 |
+
"step": 1540
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 2.600671140939597,
|
| 1092 |
+
"grad_norm": 0.1458803117275238,
|
| 1093 |
+
"learning_rate": 2.4796301433345943e-05,
|
| 1094 |
+
"loss": 0.0226,
|
| 1095 |
+
"step": 1550
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 2.61744966442953,
|
| 1099 |
+
"grad_norm": 0.15886026620864868,
|
| 1100 |
+
"learning_rate": 2.4720959035959148e-05,
|
| 1101 |
+
"loss": 0.0226,
|
| 1102 |
+
"step": 1560
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 2.6342281879194633,
|
| 1106 |
+
"grad_norm": 0.15409216284751892,
|
| 1107 |
+
"learning_rate": 2.464519141233257e-05,
|
| 1108 |
+
"loss": 0.0206,
|
| 1109 |
+
"step": 1570
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 2.651006711409396,
|
| 1113 |
+
"grad_norm": 0.13891032338142395,
|
| 1114 |
+
"learning_rate": 2.456900187678754e-05,
|
| 1115 |
+
"loss": 0.0183,
|
| 1116 |
+
"step": 1580
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 2.6677852348993287,
|
| 1120 |
+
"grad_norm": 0.13419780135154724,
|
| 1121 |
+
"learning_rate": 2.4492393762101147e-05,
|
| 1122 |
+
"loss": 0.0221,
|
| 1123 |
+
"step": 1590
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 2.684563758389262,
|
| 1127 |
+
"grad_norm": 0.124265655875206,
|
| 1128 |
+
"learning_rate": 2.4415370419360508e-05,
|
| 1129 |
+
"loss": 0.0216,
|
| 1130 |
+
"step": 1600
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 2.7013422818791946,
|
| 1134 |
+
"grad_norm": 0.1302771270275116,
|
| 1135 |
+
"learning_rate": 2.4337935217816157e-05,
|
| 1136 |
+
"loss": 0.0201,
|
| 1137 |
+
"step": 1610
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 2.7181208053691277,
|
| 1141 |
+
"grad_norm": 0.16515520215034485,
|
| 1142 |
+
"learning_rate": 2.4260091544734665e-05,
|
| 1143 |
+
"loss": 0.019,
|
| 1144 |
+
"step": 1620
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 2.7348993288590604,
|
| 1148 |
+
"grad_norm": 0.13658763468265533,
|
| 1149 |
+
"learning_rate": 2.4181842805250466e-05,
|
| 1150 |
+
"loss": 0.0205,
|
| 1151 |
+
"step": 1630
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 2.751677852348993,
|
| 1155 |
+
"grad_norm": 0.11953597515821457,
|
| 1156 |
+
"learning_rate": 2.4103192422216923e-05,
|
| 1157 |
+
"loss": 0.0199,
|
| 1158 |
+
"step": 1640
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 2.7684563758389262,
|
| 1162 |
+
"grad_norm": 0.16735540330410004,
|
| 1163 |
+
"learning_rate": 2.402414383605658e-05,
|
| 1164 |
+
"loss": 0.0192,
|
| 1165 |
+
"step": 1650
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 2.785234899328859,
|
| 1169 |
+
"grad_norm": 0.14770597219467163,
|
| 1170 |
+
"learning_rate": 2.3944700504610674e-05,
|
| 1171 |
+
"loss": 0.0197,
|
| 1172 |
+
"step": 1660
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 2.802013422818792,
|
| 1176 |
+
"grad_norm": 0.14873377978801727,
|
| 1177 |
+
"learning_rate": 2.3864865902987883e-05,
|
| 1178 |
+
"loss": 0.0186,
|
| 1179 |
+
"step": 1670
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 2.8187919463087248,
|
| 1183 |
+
"grad_norm": 0.22620314359664917,
|
| 1184 |
+
"learning_rate": 2.378464352341231e-05,
|
| 1185 |
+
"loss": 0.0214,
|
| 1186 |
+
"step": 1680
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 2.8355704697986575,
|
| 1190 |
+
"grad_norm": 0.20077335834503174,
|
| 1191 |
+
"learning_rate": 2.3704036875070728e-05,
|
| 1192 |
+
"loss": 0.0233,
|
| 1193 |
+
"step": 1690
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 2.8523489932885906,
|
| 1197 |
+
"grad_norm": 0.18293912708759308,
|
| 1198 |
+
"learning_rate": 2.362304948395906e-05,
|
| 1199 |
+
"loss": 0.0244,
|
| 1200 |
+
"step": 1700
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 2.8691275167785237,
|
| 1204 |
+
"grad_norm": 0.17799849808216095,
|
| 1205 |
+
"learning_rate": 2.3541684892728157e-05,
|
| 1206 |
+
"loss": 0.0195,
|
| 1207 |
+
"step": 1710
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 2.8859060402684564,
|
| 1211 |
+
"grad_norm": 0.15650226175785065,
|
| 1212 |
+
"learning_rate": 2.345994666052883e-05,
|
| 1213 |
+
"loss": 0.0183,
|
| 1214 |
+
"step": 1720
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 2.902684563758389,
|
| 1218 |
+
"grad_norm": 0.14934243261814117,
|
| 1219 |
+
"learning_rate": 2.337783836285614e-05,
|
| 1220 |
+
"loss": 0.0212,
|
| 1221 |
+
"step": 1730
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 2.9194630872483223,
|
| 1225 |
+
"grad_norm": 0.19469675421714783,
|
| 1226 |
+
"learning_rate": 2.3295363591393026e-05,
|
| 1227 |
+
"loss": 0.0191,
|
| 1228 |
+
"step": 1740
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 2.936241610738255,
|
| 1232 |
+
"grad_norm": 0.20920106768608093,
|
| 1233 |
+
"learning_rate": 2.3212525953853174e-05,
|
| 1234 |
+
"loss": 0.0202,
|
| 1235 |
+
"step": 1750
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 2.953020134228188,
|
| 1239 |
+
"grad_norm": 0.1367284506559372,
|
| 1240 |
+
"learning_rate": 2.3129329073823206e-05,
|
| 1241 |
+
"loss": 0.0204,
|
| 1242 |
+
"step": 1760
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 2.969798657718121,
|
| 1246 |
+
"grad_norm": 0.15733259916305542,
|
| 1247 |
+
"learning_rate": 2.3045776590604175e-05,
|
| 1248 |
+
"loss": 0.0192,
|
| 1249 |
+
"step": 1770
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 2.9865771812080535,
|
| 1253 |
+
"grad_norm": 0.1706557273864746,
|
| 1254 |
+
"learning_rate": 2.296187215905237e-05,
|
| 1255 |
+
"loss": 0.0206,
|
| 1256 |
+
"step": 1780
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 3.0033557046979866,
|
| 1260 |
+
"grad_norm": 0.20850840210914612,
|
| 1261 |
+
"learning_rate": 2.2877619449419437e-05,
|
| 1262 |
+
"loss": 0.022,
|
| 1263 |
+
"step": 1790
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 3.0201342281879193,
|
| 1267 |
+
"grad_norm": 0.21721336245536804,
|
| 1268 |
+
"learning_rate": 2.2793022147191843e-05,
|
| 1269 |
+
"loss": 0.0201,
|
| 1270 |
+
"step": 1800
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 3.0369127516778525,
|
| 1274 |
+
"grad_norm": 0.16258729994297028,
|
| 1275 |
+
"learning_rate": 2.2708083952929634e-05,
|
| 1276 |
+
"loss": 0.0185,
|
| 1277 |
+
"step": 1810
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 3.053691275167785,
|
| 1281 |
+
"grad_norm": 0.18179331719875336,
|
| 1282 |
+
"learning_rate": 2.2622808582104594e-05,
|
| 1283 |
+
"loss": 0.0215,
|
| 1284 |
+
"step": 1820
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 3.0704697986577183,
|
| 1288 |
+
"grad_norm": 0.11735052615404129,
|
| 1289 |
+
"learning_rate": 2.2537199764937703e-05,
|
| 1290 |
+
"loss": 0.019,
|
| 1291 |
+
"step": 1830
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 3.087248322147651,
|
| 1295 |
+
"grad_norm": 0.15188120305538177,
|
| 1296 |
+
"learning_rate": 2.245126124623595e-05,
|
| 1297 |
+
"loss": 0.0196,
|
| 1298 |
+
"step": 1840
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 3.1040268456375837,
|
| 1302 |
+
"grad_norm": 0.14142458140850067,
|
| 1303 |
+
"learning_rate": 2.2364996785228547e-05,
|
| 1304 |
+
"loss": 0.0188,
|
| 1305 |
+
"step": 1850
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 3.120805369127517,
|
| 1309 |
+
"grad_norm": 0.15011872351169586,
|
| 1310 |
+
"learning_rate": 2.227841015540248e-05,
|
| 1311 |
+
"loss": 0.0211,
|
| 1312 |
+
"step": 1860
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 3.1375838926174495,
|
| 1316 |
+
"grad_norm": 0.13913294672966003,
|
| 1317 |
+
"learning_rate": 2.2191505144337433e-05,
|
| 1318 |
+
"loss": 0.0203,
|
| 1319 |
+
"step": 1870
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 3.1543624161073827,
|
| 1323 |
+
"grad_norm": 0.1547662764787674,
|
| 1324 |
+
"learning_rate": 2.210428555354013e-05,
|
| 1325 |
+
"loss": 0.0198,
|
| 1326 |
+
"step": 1880
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 3.1711409395973154,
|
| 1330 |
+
"grad_norm": 0.16952672600746155,
|
| 1331 |
+
"learning_rate": 2.201675519827802e-05,
|
| 1332 |
+
"loss": 0.0207,
|
| 1333 |
+
"step": 1890
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 3.1879194630872485,
|
| 1337 |
+
"grad_norm": 0.16245362162590027,
|
| 1338 |
+
"learning_rate": 2.1928917907412403e-05,
|
| 1339 |
+
"loss": 0.0191,
|
| 1340 |
+
"step": 1900
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 3.204697986577181,
|
| 1344 |
+
"grad_norm": 0.20379698276519775,
|
| 1345 |
+
"learning_rate": 2.1840777523230937e-05,
|
| 1346 |
+
"loss": 0.0201,
|
| 1347 |
+
"step": 1910
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 3.221476510067114,
|
| 1351 |
+
"grad_norm": 0.12898804247379303,
|
| 1352 |
+
"learning_rate": 2.1752337901279552e-05,
|
| 1353 |
+
"loss": 0.0189,
|
| 1354 |
+
"step": 1920
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 3.238255033557047,
|
| 1358 |
+
"grad_norm": 0.22864995896816254,
|
| 1359 |
+
"learning_rate": 2.166360291019383e-05,
|
| 1360 |
+
"loss": 0.0189,
|
| 1361 |
+
"step": 1930
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 3.2550335570469797,
|
| 1365 |
+
"grad_norm": 0.10070180147886276,
|
| 1366 |
+
"learning_rate": 2.157457643152973e-05,
|
| 1367 |
+
"loss": 0.0191,
|
| 1368 |
+
"step": 1940
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 3.271812080536913,
|
| 1372 |
+
"grad_norm": 0.12235262989997864,
|
| 1373 |
+
"learning_rate": 2.148526235959385e-05,
|
| 1374 |
+
"loss": 0.0189,
|
| 1375 |
+
"step": 1950
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 3.2885906040268456,
|
| 1379 |
+
"grad_norm": 0.17240414023399353,
|
| 1380 |
+
"learning_rate": 2.1395664601273016e-05,
|
| 1381 |
+
"loss": 0.0185,
|
| 1382 |
+
"step": 1960
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 3.3053691275167787,
|
| 1386 |
+
"grad_norm": 0.1916498988866806,
|
| 1387 |
+
"learning_rate": 2.130578707586345e-05,
|
| 1388 |
+
"loss": 0.0204,
|
| 1389 |
+
"step": 1970
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 3.3221476510067114,
|
| 1393 |
+
"grad_norm": 0.13949474692344666,
|
| 1394 |
+
"learning_rate": 2.1215633714899264e-05,
|
| 1395 |
+
"loss": 0.0194,
|
| 1396 |
+
"step": 1980
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 3.3389261744966445,
|
| 1400 |
+
"grad_norm": 0.148747980594635,
|
| 1401 |
+
"learning_rate": 2.112520846198053e-05,
|
| 1402 |
+
"loss": 0.0187,
|
| 1403 |
+
"step": 1990
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 3.3557046979865772,
|
| 1407 |
+
"grad_norm": 0.14247365295886993,
|
| 1408 |
+
"learning_rate": 2.103451527260074e-05,
|
| 1409 |
+
"loss": 0.0179,
|
| 1410 |
+
"step": 2000
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 3.37248322147651,
|
| 1414 |
+
"grad_norm": 0.1270933598279953,
|
| 1415 |
+
"learning_rate": 2.094355811397381e-05,
|
| 1416 |
+
"loss": 0.0193,
|
| 1417 |
+
"step": 2010
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 3.389261744966443,
|
| 1421 |
+
"grad_norm": 0.16048407554626465,
|
| 1422 |
+
"learning_rate": 2.0852340964860522e-05,
|
| 1423 |
+
"loss": 0.0228,
|
| 1424 |
+
"step": 2020
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 3.4060402684563758,
|
| 1428 |
+
"grad_norm": 0.19520506262779236,
|
| 1429 |
+
"learning_rate": 2.076086781539447e-05,
|
| 1430 |
+
"loss": 0.0183,
|
| 1431 |
+
"step": 2030
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 3.422818791946309,
|
| 1435 |
+
"grad_norm": 0.1389211118221283,
|
| 1436 |
+
"learning_rate": 2.066914266690755e-05,
|
| 1437 |
+
"loss": 0.0187,
|
| 1438 |
+
"step": 2040
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 3.4395973154362416,
|
| 1442 |
+
"grad_norm": 0.15584081411361694,
|
| 1443 |
+
"learning_rate": 2.0577169531754906e-05,
|
| 1444 |
+
"loss": 0.0187,
|
| 1445 |
+
"step": 2050
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 3.4563758389261743,
|
| 1449 |
+
"grad_norm": 0.1802147477865219,
|
| 1450 |
+
"learning_rate": 2.048495243313942e-05,
|
| 1451 |
+
"loss": 0.0197,
|
| 1452 |
+
"step": 2060
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 3.4731543624161074,
|
| 1456 |
+
"grad_norm": 0.12090757489204407,
|
| 1457 |
+
"learning_rate": 2.039249540493572e-05,
|
| 1458 |
+
"loss": 0.0194,
|
| 1459 |
+
"step": 2070
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 3.48993288590604,
|
| 1463 |
+
"grad_norm": 0.18710780143737793,
|
| 1464 |
+
"learning_rate": 2.0299802491513755e-05,
|
| 1465 |
+
"loss": 0.0175,
|
| 1466 |
+
"step": 2080
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 3.5067114093959733,
|
| 1470 |
+
"grad_norm": 0.10997475683689117,
|
| 1471 |
+
"learning_rate": 2.0206877747561828e-05,
|
| 1472 |
+
"loss": 0.0195,
|
| 1473 |
+
"step": 2090
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 3.523489932885906,
|
| 1477 |
+
"grad_norm": 0.13992996513843536,
|
| 1478 |
+
"learning_rate": 2.011372523790927e-05,
|
| 1479 |
+
"loss": 0.0202,
|
| 1480 |
+
"step": 2100
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 3.540268456375839,
|
| 1484 |
+
"grad_norm": 0.11927074193954468,
|
| 1485 |
+
"learning_rate": 2.0020349037348643e-05,
|
| 1486 |
+
"loss": 0.0178,
|
| 1487 |
+
"step": 2110
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 3.557046979865772,
|
| 1491 |
+
"grad_norm": 0.18657186627388,
|
| 1492 |
+
"learning_rate": 1.9926753230457443e-05,
|
| 1493 |
+
"loss": 0.0182,
|
| 1494 |
+
"step": 2120
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 3.573825503355705,
|
| 1498 |
+
"grad_norm": 0.12504293024539948,
|
| 1499 |
+
"learning_rate": 1.983294191141948e-05,
|
| 1500 |
+
"loss": 0.0198,
|
| 1501 |
+
"step": 2130
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 3.5906040268456376,
|
| 1505 |
+
"grad_norm": 0.1450442671775818,
|
| 1506 |
+
"learning_rate": 1.973891918384575e-05,
|
| 1507 |
+
"loss": 0.0172,
|
| 1508 |
+
"step": 2140
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 3.6073825503355703,
|
| 1512 |
+
"grad_norm": 0.14081084728240967,
|
| 1513 |
+
"learning_rate": 1.9644689160594945e-05,
|
| 1514 |
+
"loss": 0.018,
|
| 1515 |
+
"step": 2150
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 3.6241610738255035,
|
| 1519 |
+
"grad_norm": 0.09649883955717087,
|
| 1520 |
+
"learning_rate": 1.9550255963593553e-05,
|
| 1521 |
+
"loss": 0.0156,
|
| 1522 |
+
"step": 2160
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 3.640939597315436,
|
| 1526 |
+
"grad_norm": 0.1074327751994133,
|
| 1527 |
+
"learning_rate": 1.9455623723655524e-05,
|
| 1528 |
+
"loss": 0.0157,
|
| 1529 |
+
"step": 2170
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 3.6577181208053693,
|
| 1533 |
+
"grad_norm": 0.1782575249671936,
|
| 1534 |
+
"learning_rate": 1.9360796580301606e-05,
|
| 1535 |
+
"loss": 0.0168,
|
| 1536 |
+
"step": 2180
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 3.674496644295302,
|
| 1540 |
+
"grad_norm": 0.16483958065509796,
|
| 1541 |
+
"learning_rate": 1.9265778681578245e-05,
|
| 1542 |
+
"loss": 0.0172,
|
| 1543 |
+
"step": 2190
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 3.6912751677852347,
|
| 1547 |
+
"grad_norm": 0.14613379538059235,
|
| 1548 |
+
"learning_rate": 1.9170574183876143e-05,
|
| 1549 |
+
"loss": 0.0204,
|
| 1550 |
+
"step": 2200
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 3.708053691275168,
|
| 1554 |
+
"grad_norm": 0.2648290693759918,
|
| 1555 |
+
"learning_rate": 1.9075187251748452e-05,
|
| 1556 |
+
"loss": 0.0202,
|
| 1557 |
+
"step": 2210
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 3.7248322147651005,
|
| 1561 |
+
"grad_norm": 0.15106196701526642,
|
| 1562 |
+
"learning_rate": 1.8979622057728596e-05,
|
| 1563 |
+
"loss": 0.0187,
|
| 1564 |
+
"step": 2220
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 3.7416107382550337,
|
| 1568 |
+
"grad_norm": 0.11246120184659958,
|
| 1569 |
+
"learning_rate": 1.8883882782147748e-05,
|
| 1570 |
+
"loss": 0.0169,
|
| 1571 |
+
"step": 2230
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 3.7583892617449663,
|
| 1575 |
+
"grad_norm": 0.1419929713010788,
|
| 1576 |
+
"learning_rate": 1.8787973612951967e-05,
|
| 1577 |
+
"loss": 0.0172,
|
| 1578 |
+
"step": 2240
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 3.7751677852348995,
|
| 1582 |
+
"grad_norm": 0.12565530836582184,
|
| 1583 |
+
"learning_rate": 1.8691898745519028e-05,
|
| 1584 |
+
"loss": 0.0192,
|
| 1585 |
+
"step": 2250
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 3.791946308724832,
|
| 1589 |
+
"grad_norm": 0.13186733424663544,
|
| 1590 |
+
"learning_rate": 1.8595662382474863e-05,
|
| 1591 |
+
"loss": 0.0206,
|
| 1592 |
+
"step": 2260
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 3.8087248322147653,
|
| 1596 |
+
"grad_norm": 0.16787581145763397,
|
| 1597 |
+
"learning_rate": 1.8499268733509766e-05,
|
| 1598 |
+
"loss": 0.0181,
|
| 1599 |
+
"step": 2270
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 3.825503355704698,
|
| 1603 |
+
"grad_norm": 0.1428891271352768,
|
| 1604 |
+
"learning_rate": 1.84027220151942e-05,
|
| 1605 |
+
"loss": 0.016,
|
| 1606 |
+
"step": 2280
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 3.8422818791946307,
|
| 1610 |
+
"grad_norm": 0.16047702729701996,
|
| 1611 |
+
"learning_rate": 1.83060264507944e-05,
|
| 1612 |
+
"loss": 0.0169,
|
| 1613 |
+
"step": 2290
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 3.859060402684564,
|
| 1617 |
+
"grad_norm": 0.13212737441062927,
|
| 1618 |
+
"learning_rate": 1.82091862700876e-05,
|
| 1619 |
+
"loss": 0.0168,
|
| 1620 |
+
"step": 2300
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 3.8758389261744965,
|
| 1624 |
+
"grad_norm": 0.10623549669981003,
|
| 1625 |
+
"learning_rate": 1.8112205709177023e-05,
|
| 1626 |
+
"loss": 0.0185,
|
| 1627 |
+
"step": 2310
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 3.8926174496644297,
|
| 1631 |
+
"grad_norm": 0.19981749355793,
|
| 1632 |
+
"learning_rate": 1.8015089010306585e-05,
|
| 1633 |
+
"loss": 0.0155,
|
| 1634 |
+
"step": 2320
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 3.9093959731543624,
|
| 1638 |
+
"grad_norm": 0.18881696462631226,
|
| 1639 |
+
"learning_rate": 1.7917840421675298e-05,
|
| 1640 |
+
"loss": 0.0209,
|
| 1641 |
+
"step": 2330
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 3.926174496644295,
|
| 1645 |
+
"grad_norm": 0.10692999511957169,
|
| 1646 |
+
"learning_rate": 1.7820464197251474e-05,
|
| 1647 |
+
"loss": 0.0161,
|
| 1648 |
+
"step": 2340
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 3.942953020134228,
|
| 1652 |
+
"grad_norm": 0.21753492951393127,
|
| 1653 |
+
"learning_rate": 1.7722964596586627e-05,
|
| 1654 |
+
"loss": 0.0189,
|
| 1655 |
+
"step": 2350
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 3.959731543624161,
|
| 1659 |
+
"grad_norm": 0.18070799112319946,
|
| 1660 |
+
"learning_rate": 1.7625345884629144e-05,
|
| 1661 |
+
"loss": 0.0184,
|
| 1662 |
+
"step": 2360
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 3.976510067114094,
|
| 1666 |
+
"grad_norm": 0.15308111906051636,
|
| 1667 |
+
"learning_rate": 1.7527612331537722e-05,
|
| 1668 |
+
"loss": 0.0156,
|
| 1669 |
+
"step": 2370
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 3.9932885906040267,
|
| 1673 |
+
"grad_norm": 0.14668720960617065,
|
| 1674 |
+
"learning_rate": 1.7429768212494593e-05,
|
| 1675 |
+
"loss": 0.0178,
|
| 1676 |
+
"step": 2380
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 4.010067114093959,
|
| 1680 |
+
"grad_norm": 0.3539591431617737,
|
| 1681 |
+
"learning_rate": 1.7331817807518497e-05,
|
| 1682 |
+
"loss": 0.0185,
|
| 1683 |
+
"step": 2390
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 4.026845637583893,
|
| 1687 |
+
"grad_norm": 0.1322038173675537,
|
| 1688 |
+
"learning_rate": 1.723376540127745e-05,
|
| 1689 |
+
"loss": 0.0171,
|
| 1690 |
+
"step": 2400
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 4.043624161073826,
|
| 1694 |
+
"grad_norm": 0.11787392944097519,
|
| 1695 |
+
"learning_rate": 1.7135615282901357e-05,
|
| 1696 |
+
"loss": 0.0163,
|
| 1697 |
+
"step": 2410
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 4.060402684563758,
|
| 1701 |
+
"grad_norm": 0.1422634720802307,
|
| 1702 |
+
"learning_rate": 1.7037371745794358e-05,
|
| 1703 |
+
"loss": 0.0206,
|
| 1704 |
+
"step": 2420
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 4.077181208053691,
|
| 1708 |
+
"grad_norm": 0.1516404151916504,
|
| 1709 |
+
"learning_rate": 1.6939039087447026e-05,
|
| 1710 |
+
"loss": 0.0166,
|
| 1711 |
+
"step": 2430
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 4.093959731543624,
|
| 1715 |
+
"grad_norm": 0.14106018841266632,
|
| 1716 |
+
"learning_rate": 1.6840621609248397e-05,
|
| 1717 |
+
"loss": 0.0197,
|
| 1718 |
+
"step": 2440
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 4.110738255033557,
|
| 1722 |
+
"grad_norm": 0.12483642995357513,
|
| 1723 |
+
"learning_rate": 1.6742123616297794e-05,
|
| 1724 |
+
"loss": 0.0175,
|
| 1725 |
+
"step": 2450
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 4.12751677852349,
|
| 1729 |
+
"grad_norm": 0.16260197758674622,
|
| 1730 |
+
"learning_rate": 1.664354941721651e-05,
|
| 1731 |
+
"loss": 0.0166,
|
| 1732 |
+
"step": 2460
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 4.144295302013423,
|
| 1736 |
+
"grad_norm": 0.1442679613828659,
|
| 1737 |
+
"learning_rate": 1.6544903323959355e-05,
|
| 1738 |
+
"loss": 0.016,
|
| 1739 |
+
"step": 2470
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 4.1610738255033555,
|
| 1743 |
+
"grad_norm": 0.11389831453561783,
|
| 1744 |
+
"learning_rate": 1.6446189651626023e-05,
|
| 1745 |
+
"loss": 0.0183,
|
| 1746 |
+
"step": 2480
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 4.177852348993288,
|
| 1750 |
+
"grad_norm": 0.14809614419937134,
|
| 1751 |
+
"learning_rate": 1.6347412718272316e-05,
|
| 1752 |
+
"loss": 0.0205,
|
| 1753 |
+
"step": 2490
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 4.194630872483222,
|
| 1757 |
+
"grad_norm": 0.17286059260368347,
|
| 1758 |
+
"learning_rate": 1.624857684472131e-05,
|
| 1759 |
+
"loss": 0.0168,
|
| 1760 |
+
"step": 2500
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 4.2114093959731544,
|
| 1764 |
+
"grad_norm": 0.14666445553302765,
|
| 1765 |
+
"learning_rate": 1.6149686354374283e-05,
|
| 1766 |
+
"loss": 0.0177,
|
| 1767 |
+
"step": 2510
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 4.228187919463087,
|
| 1771 |
+
"grad_norm": 0.15864591300487518,
|
| 1772 |
+
"learning_rate": 1.605074557302165e-05,
|
| 1773 |
+
"loss": 0.0175,
|
| 1774 |
+
"step": 2520
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 4.24496644295302,
|
| 1778 |
+
"grad_norm": 0.177460715174675,
|
| 1779 |
+
"learning_rate": 1.59517588286537e-05,
|
| 1780 |
+
"loss": 0.0175,
|
| 1781 |
+
"step": 2530
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 4.261744966442953,
|
| 1785 |
+
"grad_norm": 0.14390940964221954,
|
| 1786 |
+
"learning_rate": 1.5852730451271324e-05,
|
| 1787 |
+
"loss": 0.0156,
|
| 1788 |
+
"step": 2540
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 4.278523489932886,
|
| 1792 |
+
"grad_norm": 0.15486080944538116,
|
| 1793 |
+
"learning_rate": 1.5753664772696546e-05,
|
| 1794 |
+
"loss": 0.0182,
|
| 1795 |
+
"step": 2550
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 4.295302013422819,
|
| 1799 |
+
"grad_norm": 0.15260493755340576,
|
| 1800 |
+
"learning_rate": 1.565456612638307e-05,
|
| 1801 |
+
"loss": 0.0182,
|
| 1802 |
+
"step": 2560
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 4.3120805369127515,
|
| 1806 |
+
"grad_norm": 0.12060414999723434,
|
| 1807 |
+
"learning_rate": 1.5555438847226732e-05,
|
| 1808 |
+
"loss": 0.0167,
|
| 1809 |
+
"step": 2570
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 4.328859060402684,
|
| 1813 |
+
"grad_norm": 0.1500580757856369,
|
| 1814 |
+
"learning_rate": 1.545628727137584e-05,
|
| 1815 |
+
"loss": 0.0164,
|
| 1816 |
+
"step": 2580
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 4.345637583892618,
|
| 1820 |
+
"grad_norm": 0.15014661848545074,
|
| 1821 |
+
"learning_rate": 1.535711573604153e-05,
|
| 1822 |
+
"loss": 0.0161,
|
| 1823 |
+
"step": 2590
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 4.3624161073825505,
|
| 1827 |
+
"grad_norm": 0.16664306819438934,
|
| 1828 |
+
"learning_rate": 1.5257928579308032e-05,
|
| 1829 |
+
"loss": 0.0189,
|
| 1830 |
+
"step": 2600
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 4.379194630872483,
|
| 1834 |
+
"grad_norm": 0.155680850148201,
|
| 1835 |
+
"learning_rate": 1.5158730139942902e-05,
|
| 1836 |
+
"loss": 0.019,
|
| 1837 |
+
"step": 2610
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 4.395973154362416,
|
| 1841 |
+
"grad_norm": 0.16779354214668274,
|
| 1842 |
+
"learning_rate": 1.5059524757207239e-05,
|
| 1843 |
+
"loss": 0.0179,
|
| 1844 |
+
"step": 2620
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 4.412751677852349,
|
| 1848 |
+
"grad_norm": 0.1292392611503601,
|
| 1849 |
+
"learning_rate": 1.4960316770665866e-05,
|
| 1850 |
+
"loss": 0.0157,
|
| 1851 |
+
"step": 2630
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 4.429530201342282,
|
| 1855 |
+
"grad_norm": 0.1438266485929489,
|
| 1856 |
+
"learning_rate": 1.4861110519997495e-05,
|
| 1857 |
+
"loss": 0.0177,
|
| 1858 |
+
"step": 2640
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 4.446308724832215,
|
| 1862 |
+
"grad_norm": 0.15133166313171387,
|
| 1863 |
+
"learning_rate": 1.4761910344804913e-05,
|
| 1864 |
+
"loss": 0.0162,
|
| 1865 |
+
"step": 2650
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 4.4630872483221475,
|
| 1869 |
+
"grad_norm": 0.1657615602016449,
|
| 1870 |
+
"learning_rate": 1.466272058442516e-05,
|
| 1871 |
+
"loss": 0.0165,
|
| 1872 |
+
"step": 2660
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 4.47986577181208,
|
| 1876 |
+
"grad_norm": 0.1509656012058258,
|
| 1877 |
+
"learning_rate": 1.4563545577739676e-05,
|
| 1878 |
+
"loss": 0.0177,
|
| 1879 |
+
"step": 2670
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 4.496644295302014,
|
| 1883 |
+
"grad_norm": 0.1461293250322342,
|
| 1884 |
+
"learning_rate": 1.4464389662984543e-05,
|
| 1885 |
+
"loss": 0.0164,
|
| 1886 |
+
"step": 2680
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 4.5134228187919465,
|
| 1890 |
+
"grad_norm": 0.18428373336791992,
|
| 1891 |
+
"learning_rate": 1.4365257177560684e-05,
|
| 1892 |
+
"loss": 0.017,
|
| 1893 |
+
"step": 2690
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 4.530201342281879,
|
| 1897 |
+
"grad_norm": 0.13298265635967255,
|
| 1898 |
+
"learning_rate": 1.426615245784416e-05,
|
| 1899 |
+
"loss": 0.0158,
|
| 1900 |
+
"step": 2700
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 4.546979865771812,
|
| 1904 |
+
"grad_norm": 0.10208632797002792,
|
| 1905 |
+
"learning_rate": 1.4167079838996468e-05,
|
| 1906 |
+
"loss": 0.016,
|
| 1907 |
+
"step": 2710
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 4.563758389261745,
|
| 1911 |
+
"grad_norm": 0.16623789072036743,
|
| 1912 |
+
"learning_rate": 1.4068043654774897e-05,
|
| 1913 |
+
"loss": 0.0156,
|
| 1914 |
+
"step": 2720
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 4.580536912751678,
|
| 1918 |
+
"grad_norm": 0.14804600179195404,
|
| 1919 |
+
"learning_rate": 1.3969048237342992e-05,
|
| 1920 |
+
"loss": 0.0177,
|
| 1921 |
+
"step": 2730
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 4.597315436241611,
|
| 1925 |
+
"grad_norm": 0.16890902817249298,
|
| 1926 |
+
"learning_rate": 1.3870097917081011e-05,
|
| 1927 |
+
"loss": 0.0147,
|
| 1928 |
+
"step": 2740
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 4.614093959731544,
|
| 1932 |
+
"grad_norm": 0.21568931639194489,
|
| 1933 |
+
"learning_rate": 1.3771197022396508e-05,
|
| 1934 |
+
"loss": 0.0155,
|
| 1935 |
+
"step": 2750
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 4.630872483221476,
|
| 1939 |
+
"grad_norm": 0.14996445178985596,
|
| 1940 |
+
"learning_rate": 1.3672349879535015e-05,
|
| 1941 |
+
"loss": 0.0147,
|
| 1942 |
+
"step": 2760
|
| 1943 |
+
},
|
| 1944 |
+
{
|
| 1945 |
+
"epoch": 4.64765100671141,
|
| 1946 |
+
"grad_norm": 0.10208549350500107,
|
| 1947 |
+
"learning_rate": 1.3573560812390781e-05,
|
| 1948 |
+
"loss": 0.0166,
|
| 1949 |
+
"step": 2770
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 4.6644295302013425,
|
| 1953 |
+
"grad_norm": 0.11123307049274445,
|
| 1954 |
+
"learning_rate": 1.3474834142317635e-05,
|
| 1955 |
+
"loss": 0.0151,
|
| 1956 |
+
"step": 2780
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 4.681208053691275,
|
| 1960 |
+
"grad_norm": 0.11926154792308807,
|
| 1961 |
+
"learning_rate": 1.3376174187939955e-05,
|
| 1962 |
+
"loss": 0.0157,
|
| 1963 |
+
"step": 2790
|
| 1964 |
+
},
|
| 1965 |
+
{
|
| 1966 |
+
"epoch": 4.697986577181208,
|
| 1967 |
+
"grad_norm": 0.12520945072174072,
|
| 1968 |
+
"learning_rate": 1.3277585264963761e-05,
|
| 1969 |
+
"loss": 0.0164,
|
| 1970 |
+
"step": 2800
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 4.714765100671141,
|
| 1974 |
+
"grad_norm": 0.15269853174686432,
|
| 1975 |
+
"learning_rate": 1.3179071685987928e-05,
|
| 1976 |
+
"loss": 0.0187,
|
| 1977 |
+
"step": 2810
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 4.731543624161074,
|
| 1981 |
+
"grad_norm": 0.11586833000183105,
|
| 1982 |
+
"learning_rate": 1.308063776031554e-05,
|
| 1983 |
+
"loss": 0.0153,
|
| 1984 |
+
"step": 2820
|
| 1985 |
+
},
|
| 1986 |
+
{
|
| 1987 |
+
"epoch": 4.748322147651007,
|
| 1988 |
+
"grad_norm": 0.23391515016555786,
|
| 1989 |
+
"learning_rate": 1.2982287793765389e-05,
|
| 1990 |
+
"loss": 0.0158,
|
| 1991 |
+
"step": 2830
|
| 1992 |
+
},
|
| 1993 |
+
{
|
| 1994 |
+
"epoch": 4.76510067114094,
|
| 1995 |
+
"grad_norm": 0.13222748041152954,
|
| 1996 |
+
"learning_rate": 1.2884026088483616e-05,
|
| 1997 |
+
"loss": 0.0145,
|
| 1998 |
+
"step": 2840
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 4.781879194630872,
|
| 2002 |
+
"grad_norm": 0.1492713987827301,
|
| 2003 |
+
"learning_rate": 1.2785856942755543e-05,
|
| 2004 |
+
"loss": 0.0168,
|
| 2005 |
+
"step": 2850
|
| 2006 |
+
},
|
| 2007 |
+
{
|
| 2008 |
+
"epoch": 4.798657718120805,
|
| 2009 |
+
"grad_norm": 0.13140155375003815,
|
| 2010 |
+
"learning_rate": 1.2687784650817616e-05,
|
| 2011 |
+
"loss": 0.0172,
|
| 2012 |
+
"step": 2860
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 4.815436241610739,
|
| 2016 |
+
"grad_norm": 0.11021384596824646,
|
| 2017 |
+
"learning_rate": 1.2589813502669603e-05,
|
| 2018 |
+
"loss": 0.0172,
|
| 2019 |
+
"step": 2870
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 4.832214765100671,
|
| 2023 |
+
"grad_norm": 0.15835842490196228,
|
| 2024 |
+
"learning_rate": 1.2491947783886907e-05,
|
| 2025 |
+
"loss": 0.016,
|
| 2026 |
+
"step": 2880
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 4.848993288590604,
|
| 2030 |
+
"grad_norm": 0.13786916434764862,
|
| 2031 |
+
"learning_rate": 1.2394191775433106e-05,
|
| 2032 |
+
"loss": 0.0158,
|
| 2033 |
+
"step": 2890
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 4.865771812080537,
|
| 2037 |
+
"grad_norm": 0.1657201498746872,
|
| 2038 |
+
"learning_rate": 1.2296549753472696e-05,
|
| 2039 |
+
"loss": 0.0157,
|
| 2040 |
+
"step": 2900
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 4.882550335570469,
|
| 2044 |
+
"grad_norm": 0.18127189576625824,
|
| 2045 |
+
"learning_rate": 1.2199025989184032e-05,
|
| 2046 |
+
"loss": 0.0143,
|
| 2047 |
+
"step": 2910
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 4.899328859060403,
|
| 2051 |
+
"grad_norm": 0.14471322298049927,
|
| 2052 |
+
"learning_rate": 1.21016247485725e-05,
|
| 2053 |
+
"loss": 0.0171,
|
| 2054 |
+
"step": 2920
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 4.916107382550336,
|
| 2058 |
+
"grad_norm": 0.15776599943637848,
|
| 2059 |
+
"learning_rate": 1.2004350292283897e-05,
|
| 2060 |
+
"loss": 0.0175,
|
| 2061 |
+
"step": 2930
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 4.932885906040268,
|
| 2065 |
+
"grad_norm": 0.12908852100372314,
|
| 2066 |
+
"learning_rate": 1.1907206875418068e-05,
|
| 2067 |
+
"loss": 0.0171,
|
| 2068 |
+
"step": 2940
|
| 2069 |
+
},
|
| 2070 |
+
{
|
| 2071 |
+
"epoch": 4.949664429530201,
|
| 2072 |
+
"grad_norm": 0.19710037112236023,
|
| 2073 |
+
"learning_rate": 1.1810198747342769e-05,
|
| 2074 |
+
"loss": 0.0162,
|
| 2075 |
+
"step": 2950
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 4.966442953020135,
|
| 2079 |
+
"grad_norm": 0.1662857085466385,
|
| 2080 |
+
"learning_rate": 1.1713330151507786e-05,
|
| 2081 |
+
"loss": 0.016,
|
| 2082 |
+
"step": 2960
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"epoch": 4.983221476510067,
|
| 2086 |
+
"grad_norm": 0.1350361704826355,
|
| 2087 |
+
"learning_rate": 1.161660532525931e-05,
|
| 2088 |
+
"loss": 0.0156,
|
| 2089 |
+
"step": 2970
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"epoch": 5.0,
|
| 2093 |
+
"grad_norm": 0.15521599352359772,
|
| 2094 |
+
"learning_rate": 1.1520028499654584e-05,
|
| 2095 |
+
"loss": 0.0134,
|
| 2096 |
+
"step": 2980
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 5.016778523489933,
|
| 2100 |
+
"grad_norm": 0.14748677611351013,
|
| 2101 |
+
"learning_rate": 1.1423603899276833e-05,
|
| 2102 |
+
"loss": 0.0143,
|
| 2103 |
+
"step": 2990
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 5.033557046979865,
|
| 2107 |
+
"grad_norm": 0.11525747179985046,
|
| 2108 |
+
"learning_rate": 1.1327335742050442e-05,
|
| 2109 |
+
"loss": 0.0158,
|
| 2110 |
+
"step": 3000
|
| 2111 |
+
}
|
| 2112 |
+
],
|
| 2113 |
+
"logging_steps": 10,
|
| 2114 |
+
"max_steps": 5000,
|
| 2115 |
+
"num_input_tokens_seen": 0,
|
| 2116 |
+
"num_train_epochs": 9,
|
| 2117 |
+
"save_steps": 5001,
|
| 2118 |
+
"stateful_callbacks": {
|
| 2119 |
+
"TrainerControl": {
|
| 2120 |
+
"args": {
|
| 2121 |
+
"should_epoch_stop": false,
|
| 2122 |
+
"should_evaluate": false,
|
| 2123 |
+
"should_log": false,
|
| 2124 |
+
"should_save": true,
|
| 2125 |
+
"should_training_stop": false
|
| 2126 |
+
},
|
| 2127 |
+
"attributes": {}
|
| 2128 |
+
}
|
| 2129 |
+
},
|
| 2130 |
+
"total_flos": 0.0,
|
| 2131 |
+
"train_batch_size": 64,
|
| 2132 |
+
"trial_name": null,
|
| 2133 |
+
"trial_params": null
|
| 2134 |
+
}
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed0/default/train_5shot_seed0_fullactionhead_5k/experiment_cfg/metadata.json
ADDED
|
@@ -0,0 +1,431 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"new_embodiment": {
|
| 3 |
+
"statistics": {
|
| 4 |
+
"state": {
|
| 5 |
+
"base_position": {
|
| 6 |
+
"max": [
|
| 7 |
+
7.778976917266846,
|
| 8 |
+
0.33403685688972473,
|
| 9 |
+
0.7223060727119446
|
| 10 |
+
],
|
| 11 |
+
"min": [
|
| 12 |
+
-4.821934223175049,
|
| 13 |
+
-6.785110950469971,
|
| 14 |
+
0.6994727253913879
|
| 15 |
+
],
|
| 16 |
+
"mean": [
|
| 17 |
+
2.5886094736594543,
|
| 18 |
+
-2.1559664577316586,
|
| 19 |
+
0.7003580854548973
|
| 20 |
+
],
|
| 21 |
+
"std": [
|
| 22 |
+
1.6717149343420803,
|
| 23 |
+
1.3743695399555862,
|
| 24 |
+
0.000940069542359506
|
| 25 |
+
],
|
| 26 |
+
"q01": [
|
| 27 |
+
-0.7517815060249613,
|
| 28 |
+
-5.497849509089705,
|
| 29 |
+
0.6999776914424374
|
| 30 |
+
],
|
| 31 |
+
"q99": [
|
| 32 |
+
6.060945690807028,
|
| 33 |
+
-0.5331499932493624,
|
| 34 |
+
0.7028864260472257
|
| 35 |
+
]
|
| 36 |
+
},
|
| 37 |
+
"base_rotation": {
|
| 38 |
+
"max": [
|
| 39 |
+
0.0,
|
| 40 |
+
0.0,
|
| 41 |
+
1.0,
|
| 42 |
+
1.0
|
| 43 |
+
],
|
| 44 |
+
"min": [
|
| 45 |
+
0.0,
|
| 46 |
+
0.0,
|
| 47 |
+
-1.0,
|
| 48 |
+
0.0
|
| 49 |
+
],
|
| 50 |
+
"mean": [
|
| 51 |
+
0.0,
|
| 52 |
+
0.0,
|
| 53 |
+
0.24488939171811902,
|
| 54 |
+
0.5968549204077488
|
| 55 |
+
],
|
| 56 |
+
"std": [
|
| 57 |
+
0.0,
|
| 58 |
+
0.0,
|
| 59 |
+
0.6659970734507764,
|
| 60 |
+
0.37443298392262264
|
| 61 |
+
],
|
| 62 |
+
"q01": [
|
| 63 |
+
0.0,
|
| 64 |
+
0.0,
|
| 65 |
+
-0.9999999999999999,
|
| 66 |
+
4.149924358336069e-06
|
| 67 |
+
],
|
| 68 |
+
"q99": [
|
| 69 |
+
0.0,
|
| 70 |
+
0.0,
|
| 71 |
+
0.9999999999999999,
|
| 72 |
+
0.9999999999999999
|
| 73 |
+
]
|
| 74 |
+
},
|
| 75 |
+
"end_effector_position_relative": {
|
| 76 |
+
"max": [
|
| 77 |
+
0.8848381638526917,
|
| 78 |
+
0.8569459915161133,
|
| 79 |
+
1.053884744644165
|
| 80 |
+
],
|
| 81 |
+
"min": [
|
| 82 |
+
-0.5828549861907959,
|
| 83 |
+
-0.8486236333847046,
|
| 84 |
+
-0.3446122705936432
|
| 85 |
+
],
|
| 86 |
+
"mean": [
|
| 87 |
+
0.25087486209693133,
|
| 88 |
+
-0.005738920462036393,
|
| 89 |
+
0.43643948866361787
|
| 90 |
+
],
|
| 91 |
+
"std": [
|
| 92 |
+
0.15741544532518012,
|
| 93 |
+
0.2547603106793105,
|
| 94 |
+
0.2050434492495558
|
| 95 |
+
],
|
| 96 |
+
"q01": [
|
| 97 |
+
-0.013990057288082561,
|
| 98 |
+
-0.4824599092623036,
|
| 99 |
+
0.18825220353500022
|
| 100 |
+
],
|
| 101 |
+
"q99": [
|
| 102 |
+
0.5411989642841469,
|
| 103 |
+
0.4754923657245658,
|
| 104 |
+
0.6912323747882034
|
| 105 |
+
]
|
| 106 |
+
},
|
| 107 |
+
"end_effector_rotation_relative": {
|
| 108 |
+
"max": [
|
| 109 |
+
0.9999949932098389,
|
| 110 |
+
0.9997354745864868,
|
| 111 |
+
0.9900747537612915,
|
| 112 |
+
0.9735639095306396
|
| 113 |
+
],
|
| 114 |
+
"min": [
|
| 115 |
+
-0.9999914169311523,
|
| 116 |
+
-0.9999884366989136,
|
| 117 |
+
-0.9975369572639465,
|
| 118 |
+
6.754255821306288e-08
|
| 119 |
+
],
|
| 120 |
+
"mean": [
|
| 121 |
+
-0.22290331318740714,
|
| 122 |
+
0.04579037037503981,
|
| 123 |
+
-0.05804094906527823,
|
| 124 |
+
0.12347219024917105
|
| 125 |
+
],
|
| 126 |
+
"std": [
|
| 127 |
+
0.8519267931761344,
|
| 128 |
+
0.2949394732270141,
|
| 129 |
+
0.30026078832556913,
|
| 130 |
+
0.16333086331829588
|
| 131 |
+
],
|
| 132 |
+
"q01": [
|
| 133 |
+
-0.9957959450418518,
|
| 134 |
+
-0.5678869493126506,
|
| 135 |
+
-0.5318135475761875,
|
| 136 |
+
0.0014160465686576505
|
| 137 |
+
],
|
| 138 |
+
"q99": [
|
| 139 |
+
0.9900692296898317,
|
| 140 |
+
0.6357358195673684,
|
| 141 |
+
0.5009847868015971,
|
| 142 |
+
0.4287044482210845
|
| 143 |
+
]
|
| 144 |
+
},
|
| 145 |
+
"gripper_qpos": {
|
| 146 |
+
"max": [
|
| 147 |
+
0.05289580300450325,
|
| 148 |
+
0.00937709677964449
|
| 149 |
+
],
|
| 150 |
+
"min": [
|
| 151 |
+
-0.011669117026031017,
|
| 152 |
+
-0.06307756155729294
|
| 153 |
+
],
|
| 154 |
+
"mean": [
|
| 155 |
+
0.03203268930378034,
|
| 156 |
+
-0.032331639271968075
|
| 157 |
+
],
|
| 158 |
+
"std": [
|
| 159 |
+
0.011607049173153697,
|
| 160 |
+
0.011306286745872529
|
| 161 |
+
],
|
| 162 |
+
"q01": [
|
| 163 |
+
0.006429109393656672,
|
| 164 |
+
-0.04056586943958638
|
| 165 |
+
],
|
| 166 |
+
"q99": [
|
| 167 |
+
0.040377860142451545,
|
| 168 |
+
-0.006593889698673258
|
| 169 |
+
]
|
| 170 |
+
}
|
| 171 |
+
},
|
| 172 |
+
"action": {
|
| 173 |
+
"base_motion": {
|
| 174 |
+
"max": [
|
| 175 |
+
1.0,
|
| 176 |
+
1.0,
|
| 177 |
+
1.0,
|
| 178 |
+
0.0
|
| 179 |
+
],
|
| 180 |
+
"min": [
|
| 181 |
+
-1.0,
|
| 182 |
+
-1.0,
|
| 183 |
+
-1.0,
|
| 184 |
+
0.0
|
| 185 |
+
],
|
| 186 |
+
"mean": [
|
| 187 |
+
0.006050336852760097,
|
| 188 |
+
0.002728916466628117,
|
| 189 |
+
-0.0017236811393598838,
|
| 190 |
+
0.0
|
| 191 |
+
],
|
| 192 |
+
"std": [
|
| 193 |
+
0.11032167711449951,
|
| 194 |
+
0.09824140301615687,
|
| 195 |
+
0.07825208047515658,
|
| 196 |
+
0.0
|
| 197 |
+
],
|
| 198 |
+
"q01": [
|
| 199 |
+
-0.154721262094005,
|
| 200 |
+
-0.18955302988697728,
|
| 201 |
+
-0.14454065867754942,
|
| 202 |
+
0.0
|
| 203 |
+
],
|
| 204 |
+
"q99": [
|
| 205 |
+
0.24492955247744197,
|
| 206 |
+
0.2496797822190429,
|
| 207 |
+
0.11966742325336034,
|
| 208 |
+
0.0
|
| 209 |
+
]
|
| 210 |
+
},
|
| 211 |
+
"control_mode": {
|
| 212 |
+
"max": [
|
| 213 |
+
1.0
|
| 214 |
+
],
|
| 215 |
+
"min": [
|
| 216 |
+
-1.0
|
| 217 |
+
],
|
| 218 |
+
"mean": [
|
| 219 |
+
-0.9294448715752568
|
| 220 |
+
],
|
| 221 |
+
"std": [
|
| 222 |
+
0.36897903503358326
|
| 223 |
+
],
|
| 224 |
+
"q01": [
|
| 225 |
+
-1.0
|
| 226 |
+
],
|
| 227 |
+
"q99": [
|
| 228 |
+
-0.21627562513311385
|
| 229 |
+
]
|
| 230 |
+
},
|
| 231 |
+
"end_effector_position": {
|
| 232 |
+
"max": [
|
| 233 |
+
1.0,
|
| 234 |
+
1.0,
|
| 235 |
+
1.0
|
| 236 |
+
],
|
| 237 |
+
"min": [
|
| 238 |
+
-1.0,
|
| 239 |
+
-1.0,
|
| 240 |
+
-1.0
|
| 241 |
+
],
|
| 242 |
+
"mean": [
|
| 243 |
+
-0.00826780165926883,
|
| 244 |
+
0.003161997359379232,
|
| 245 |
+
-0.08157453651841007
|
| 246 |
+
],
|
| 247 |
+
"std": [
|
| 248 |
+
0.4545430944464769,
|
| 249 |
+
0.41040151434986594,
|
| 250 |
+
0.4102644251536544
|
| 251 |
+
],
|
| 252 |
+
"q01": [
|
| 253 |
+
-0.9653185255999107,
|
| 254 |
+
-0.9255635199318754,
|
| 255 |
+
-0.8208437313341718
|
| 256 |
+
],
|
| 257 |
+
"q99": [
|
| 258 |
+
0.7987781084592441,
|
| 259 |
+
0.9260844232330088,
|
| 260 |
+
0.8357421812622201
|
| 261 |
+
]
|
| 262 |
+
},
|
| 263 |
+
"end_effector_rotation": {
|
| 264 |
+
"max": [
|
| 265 |
+
1.0,
|
| 266 |
+
1.0,
|
| 267 |
+
1.0
|
| 268 |
+
],
|
| 269 |
+
"min": [
|
| 270 |
+
-1.0,
|
| 271 |
+
-1.0,
|
| 272 |
+
-1.0
|
| 273 |
+
],
|
| 274 |
+
"mean": [
|
| 275 |
+
0.007367024641295137,
|
| 276 |
+
-0.02060756343728361,
|
| 277 |
+
-0.001884833075998489
|
| 278 |
+
],
|
| 279 |
+
"std": [
|
| 280 |
+
0.1116768956941076,
|
| 281 |
+
0.13249042272165212,
|
| 282 |
+
0.12338908364725322
|
| 283 |
+
],
|
| 284 |
+
"q01": [
|
| 285 |
+
-0.26894658663246845,
|
| 286 |
+
-0.39865607740339326,
|
| 287 |
+
-0.34393735191302655
|
| 288 |
+
],
|
| 289 |
+
"q99": [
|
| 290 |
+
0.3191674780171587,
|
| 291 |
+
0.31133630397431267,
|
| 292 |
+
0.33968476423051097
|
| 293 |
+
]
|
| 294 |
+
},
|
| 295 |
+
"gripper_close": {
|
| 296 |
+
"max": [
|
| 297 |
+
1.0
|
| 298 |
+
],
|
| 299 |
+
"min": [
|
| 300 |
+
-1.0
|
| 301 |
+
],
|
| 302 |
+
"mean": [
|
| 303 |
+
-0.3098994894722959
|
| 304 |
+
],
|
| 305 |
+
"std": [
|
| 306 |
+
0.9507718287498876
|
| 307 |
+
],
|
| 308 |
+
"q01": [
|
| 309 |
+
-1.0
|
| 310 |
+
],
|
| 311 |
+
"q99": [
|
| 312 |
+
0.7545620899936146
|
| 313 |
+
]
|
| 314 |
+
}
|
| 315 |
+
}
|
| 316 |
+
},
|
| 317 |
+
"modalities": {
|
| 318 |
+
"video": {
|
| 319 |
+
"robot0_eye_in_hand": {
|
| 320 |
+
"resolution": [
|
| 321 |
+
256,
|
| 322 |
+
256
|
| 323 |
+
],
|
| 324 |
+
"channels": 3,
|
| 325 |
+
"fps": 20.0
|
| 326 |
+
},
|
| 327 |
+
"robot0_agentview_left": {
|
| 328 |
+
"resolution": [
|
| 329 |
+
256,
|
| 330 |
+
256
|
| 331 |
+
],
|
| 332 |
+
"channels": 3,
|
| 333 |
+
"fps": 20.0
|
| 334 |
+
},
|
| 335 |
+
"robot0_agentview_right": {
|
| 336 |
+
"resolution": [
|
| 337 |
+
256,
|
| 338 |
+
256
|
| 339 |
+
],
|
| 340 |
+
"channels": 3,
|
| 341 |
+
"fps": 20.0
|
| 342 |
+
}
|
| 343 |
+
},
|
| 344 |
+
"state": {
|
| 345 |
+
"base_position": {
|
| 346 |
+
"absolute": true,
|
| 347 |
+
"rotation_type": null,
|
| 348 |
+
"shape": [
|
| 349 |
+
3
|
| 350 |
+
],
|
| 351 |
+
"continuous": true
|
| 352 |
+
},
|
| 353 |
+
"base_rotation": {
|
| 354 |
+
"absolute": true,
|
| 355 |
+
"rotation_type": "quaternion",
|
| 356 |
+
"shape": [
|
| 357 |
+
4
|
| 358 |
+
],
|
| 359 |
+
"continuous": true
|
| 360 |
+
},
|
| 361 |
+
"end_effector_position_relative": {
|
| 362 |
+
"absolute": true,
|
| 363 |
+
"rotation_type": null,
|
| 364 |
+
"shape": [
|
| 365 |
+
3
|
| 366 |
+
],
|
| 367 |
+
"continuous": true
|
| 368 |
+
},
|
| 369 |
+
"end_effector_rotation_relative": {
|
| 370 |
+
"absolute": true,
|
| 371 |
+
"rotation_type": "quaternion",
|
| 372 |
+
"shape": [
|
| 373 |
+
4
|
| 374 |
+
],
|
| 375 |
+
"continuous": true
|
| 376 |
+
},
|
| 377 |
+
"gripper_qpos": {
|
| 378 |
+
"absolute": true,
|
| 379 |
+
"rotation_type": null,
|
| 380 |
+
"shape": [
|
| 381 |
+
2
|
| 382 |
+
],
|
| 383 |
+
"continuous": true
|
| 384 |
+
}
|
| 385 |
+
},
|
| 386 |
+
"action": {
|
| 387 |
+
"base_motion": {
|
| 388 |
+
"absolute": true,
|
| 389 |
+
"rotation_type": null,
|
| 390 |
+
"shape": [
|
| 391 |
+
4
|
| 392 |
+
],
|
| 393 |
+
"continuous": true
|
| 394 |
+
},
|
| 395 |
+
"control_mode": {
|
| 396 |
+
"absolute": true,
|
| 397 |
+
"rotation_type": null,
|
| 398 |
+
"shape": [
|
| 399 |
+
1
|
| 400 |
+
],
|
| 401 |
+
"continuous": true
|
| 402 |
+
},
|
| 403 |
+
"end_effector_position": {
|
| 404 |
+
"absolute": true,
|
| 405 |
+
"rotation_type": null,
|
| 406 |
+
"shape": [
|
| 407 |
+
3
|
| 408 |
+
],
|
| 409 |
+
"continuous": true
|
| 410 |
+
},
|
| 411 |
+
"end_effector_rotation": {
|
| 412 |
+
"absolute": true,
|
| 413 |
+
"rotation_type": "axis_angle",
|
| 414 |
+
"shape": [
|
| 415 |
+
3
|
| 416 |
+
],
|
| 417 |
+
"continuous": true
|
| 418 |
+
},
|
| 419 |
+
"gripper_close": {
|
| 420 |
+
"absolute": true,
|
| 421 |
+
"rotation_type": null,
|
| 422 |
+
"shape": [
|
| 423 |
+
1
|
| 424 |
+
],
|
| 425 |
+
"continuous": true
|
| 426 |
+
}
|
| 427 |
+
}
|
| 428 |
+
},
|
| 429 |
+
"embodiment_tag": "new_embodiment"
|
| 430 |
+
}
|
| 431 |
+
}
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/resolved_config.yaml
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
experiment_name: fewshot_39tasks_5shot_from_baseline26_fullactionhead_seed1_5k
|
| 2 |
+
policy_type: baseline
|
| 3 |
+
base_model_path: ${HOME}/groot_robocasa/Atomic26_baseline/checkpoint-120000
|
| 4 |
+
output_root: ${HOME}/groot_robocasa/robocasa_v2/FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1
|
| 5 |
+
dataset:
|
| 6 |
+
dataset_soup: fewshot_heldout_atomic39_human_5shot
|
| 7 |
+
training:
|
| 8 |
+
num_gpus: 1
|
| 9 |
+
batch_size: 64
|
| 10 |
+
seed: 42
|
| 11 |
+
filter_key_seed: 1
|
| 12 |
+
manifest_path: ${HOME}/clvla/benchmarks/robocasa_v2/my_scripts/Fewshot/manifests/fewshot_heldout_atomic39/shot5_seed1.json
|
| 13 |
+
save_total_limit: 4
|
| 14 |
+
phases:
|
| 15 |
+
- name: train_5shot_seed1_fullactionhead_5k
|
| 16 |
+
max_steps: 5000
|
| 17 |
+
save_steps: 0
|
| 18 |
+
save_step_list:
|
| 19 |
+
- 1000
|
| 20 |
+
- 3000
|
| 21 |
+
- 5000
|
| 22 |
+
save_final_model: false
|
| 23 |
+
trainable:
|
| 24 |
+
tune_llm: false
|
| 25 |
+
tune_visual: false
|
| 26 |
+
tune_projector: true
|
| 27 |
+
tune_diffusion_model: true
|
| 28 |
+
resolved_sweep: {}
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/seed1_5k.yaml
ADDED
|
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
experiment_name: fewshot_39tasks_5shot_from_baseline26_fullactionhead_seed1_5k
|
| 2 |
+
policy_type: baseline
|
| 3 |
+
base_model_path: ${HOME}/groot_robocasa/Atomic26_baseline/checkpoint-120000
|
| 4 |
+
output_root: ${HOME}/groot_robocasa/robocasa_v2/FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1
|
| 5 |
+
dataset:
|
| 6 |
+
dataset_soup: fewshot_heldout_atomic39_human_5shot
|
| 7 |
+
training:
|
| 8 |
+
num_gpus: 1
|
| 9 |
+
batch_size: 64
|
| 10 |
+
seed: 42
|
| 11 |
+
filter_key_seed: 1
|
| 12 |
+
manifest_path: ${HOME}/clvla/benchmarks/robocasa_v2/my_scripts/Fewshot/manifests/fewshot_heldout_atomic39/shot5_seed1.json
|
| 13 |
+
save_total_limit: 4
|
| 14 |
+
phases:
|
| 15 |
+
- name: train_5shot_seed1_fullactionhead_5k
|
| 16 |
+
max_steps: 5000
|
| 17 |
+
save_steps: 0
|
| 18 |
+
save_step_list:
|
| 19 |
+
- 1000
|
| 20 |
+
- 3000
|
| 21 |
+
- 5000
|
| 22 |
+
save_final_model: false
|
| 23 |
+
trainable:
|
| 24 |
+
tune_llm: false
|
| 25 |
+
tune_visual: false
|
| 26 |
+
tune_projector: true
|
| 27 |
+
tune_diffusion_model: true
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-1000/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 |
+
}
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-1000/experiment_cfg/metadata.json
ADDED
|
@@ -0,0 +1,431 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"new_embodiment": {
|
| 3 |
+
"statistics": {
|
| 4 |
+
"state": {
|
| 5 |
+
"base_position": {
|
| 6 |
+
"max": [
|
| 7 |
+
7.778976917266846,
|
| 8 |
+
0.33403685688972473,
|
| 9 |
+
0.7223060727119446
|
| 10 |
+
],
|
| 11 |
+
"min": [
|
| 12 |
+
-4.821934223175049,
|
| 13 |
+
-6.785110950469971,
|
| 14 |
+
0.6994727253913879
|
| 15 |
+
],
|
| 16 |
+
"mean": [
|
| 17 |
+
2.5719731235192254,
|
| 18 |
+
-2.1489371002267235,
|
| 19 |
+
0.7003612741268546
|
| 20 |
+
],
|
| 21 |
+
"std": [
|
| 22 |
+
1.669945884010784,
|
| 23 |
+
1.3739411381361726,
|
| 24 |
+
0.0009413282367754987
|
| 25 |
+
],
|
| 26 |
+
"q01": [
|
| 27 |
+
-0.7944286339399775,
|
| 28 |
+
-5.486611518471242,
|
| 29 |
+
0.699977555971739
|
| 30 |
+
],
|
| 31 |
+
"q99": [
|
| 32 |
+
6.019886905263648,
|
| 33 |
+
-0.5229184210026855,
|
| 34 |
+
0.702930326282967
|
| 35 |
+
]
|
| 36 |
+
},
|
| 37 |
+
"base_rotation": {
|
| 38 |
+
"max": [
|
| 39 |
+
0.0,
|
| 40 |
+
0.0,
|
| 41 |
+
1.0,
|
| 42 |
+
1.0
|
| 43 |
+
],
|
| 44 |
+
"min": [
|
| 45 |
+
0.0,
|
| 46 |
+
0.0,
|
| 47 |
+
-1.0,
|
| 48 |
+
0.0
|
| 49 |
+
],
|
| 50 |
+
"mean": [
|
| 51 |
+
0.0,
|
| 52 |
+
0.0,
|
| 53 |
+
0.248543484335273,
|
| 54 |
+
0.5959317405459531
|
| 55 |
+
],
|
| 56 |
+
"std": [
|
| 57 |
+
0.0,
|
| 58 |
+
0.0,
|
| 59 |
+
0.6655550240074092,
|
| 60 |
+
0.374283995143559
|
| 61 |
+
],
|
| 62 |
+
"q01": [
|
| 63 |
+
0.0,
|
| 64 |
+
0.0,
|
| 65 |
+
-1.0000000000000002,
|
| 66 |
+
4.503254968910557e-06
|
| 67 |
+
],
|
| 68 |
+
"q99": [
|
| 69 |
+
0.0,
|
| 70 |
+
0.0,
|
| 71 |
+
1.0000000000000002,
|
| 72 |
+
1.0000000000000002
|
| 73 |
+
]
|
| 74 |
+
},
|
| 75 |
+
"end_effector_position_relative": {
|
| 76 |
+
"max": [
|
| 77 |
+
0.8848381638526917,
|
| 78 |
+
0.8569459915161133,
|
| 79 |
+
1.053884744644165
|
| 80 |
+
],
|
| 81 |
+
"min": [
|
| 82 |
+
-0.5828549861907959,
|
| 83 |
+
-0.8486236333847046,
|
| 84 |
+
-0.3446122705936432
|
| 85 |
+
],
|
| 86 |
+
"mean": [
|
| 87 |
+
0.2465490874130862,
|
| 88 |
+
-0.004711658908481467,
|
| 89 |
+
0.43463835464312073
|
| 90 |
+
],
|
| 91 |
+
"std": [
|
| 92 |
+
0.15798784670671595,
|
| 93 |
+
0.25624102400497073,
|
| 94 |
+
0.20836028727974915
|
| 95 |
+
],
|
| 96 |
+
"q01": [
|
| 97 |
+
-0.020752051134640608,
|
| 98 |
+
-0.4837933887347875,
|
| 99 |
+
0.1802942952289915
|
| 100 |
+
],
|
| 101 |
+
"q99": [
|
| 102 |
+
0.5386784714844007,
|
| 103 |
+
0.4801424212024721,
|
| 104 |
+
0.695095732975766
|
| 105 |
+
]
|
| 106 |
+
},
|
| 107 |
+
"end_effector_rotation_relative": {
|
| 108 |
+
"max": [
|
| 109 |
+
0.9999949932098389,
|
| 110 |
+
0.9997354745864868,
|
| 111 |
+
0.9900747537612915,
|
| 112 |
+
0.9735639095306396
|
| 113 |
+
],
|
| 114 |
+
"min": [
|
| 115 |
+
-0.9999914169311523,
|
| 116 |
+
-0.9999884366989136,
|
| 117 |
+
-0.9975369572639465,
|
| 118 |
+
6.754255821306288e-08
|
| 119 |
+
],
|
| 120 |
+
"mean": [
|
| 121 |
+
-0.22531859714056926,
|
| 122 |
+
0.046973702821026614,
|
| 123 |
+
-0.0568176967878771,
|
| 124 |
+
0.12364307042955697
|
| 125 |
+
],
|
| 126 |
+
"std": [
|
| 127 |
+
0.8502144333612052,
|
| 128 |
+
0.29591250318147405,
|
| 129 |
+
0.30218053156837016,
|
| 130 |
+
0.1636031353412931
|
| 131 |
+
],
|
| 132 |
+
"q01": [
|
| 133 |
+
-0.9959057387172566,
|
| 134 |
+
-0.5665384745573048,
|
| 135 |
+
-0.5329334620834977,
|
| 136 |
+
0.0014372279999397553
|
| 137 |
+
],
|
| 138 |
+
"q99": [
|
| 139 |
+
0.9904107951565839,
|
| 140 |
+
0.6359623612021396,
|
| 141 |
+
0.5026076022439377,
|
| 142 |
+
0.42955559234879914
|
| 143 |
+
]
|
| 144 |
+
},
|
| 145 |
+
"gripper_qpos": {
|
| 146 |
+
"max": [
|
| 147 |
+
0.05289580300450325,
|
| 148 |
+
0.00937709677964449
|
| 149 |
+
],
|
| 150 |
+
"min": [
|
| 151 |
+
-0.011669117026031017,
|
| 152 |
+
-0.06307756155729294
|
| 153 |
+
],
|
| 154 |
+
"mean": [
|
| 155 |
+
0.03188757060429838,
|
| 156 |
+
-0.0321793333086053
|
| 157 |
+
],
|
| 158 |
+
"std": [
|
| 159 |
+
0.01171052313737148,
|
| 160 |
+
0.011427336801203189
|
| 161 |
+
],
|
| 162 |
+
"q01": [
|
| 163 |
+
0.006462901319459925,
|
| 164 |
+
-0.040557140599419166
|
| 165 |
+
],
|
| 166 |
+
"q99": [
|
| 167 |
+
0.04037619335388386,
|
| 168 |
+
-0.006523947835993347
|
| 169 |
+
]
|
| 170 |
+
}
|
| 171 |
+
},
|
| 172 |
+
"action": {
|
| 173 |
+
"base_motion": {
|
| 174 |
+
"max": [
|
| 175 |
+
1.0,
|
| 176 |
+
1.0,
|
| 177 |
+
1.0,
|
| 178 |
+
0.0
|
| 179 |
+
],
|
| 180 |
+
"min": [
|
| 181 |
+
-1.0,
|
| 182 |
+
-1.0,
|
| 183 |
+
-1.0,
|
| 184 |
+
0.0
|
| 185 |
+
],
|
| 186 |
+
"mean": [
|
| 187 |
+
0.006639007355857881,
|
| 188 |
+
0.0027330421795561874,
|
| 189 |
+
-0.0017774441970214892,
|
| 190 |
+
0.0
|
| 191 |
+
],
|
| 192 |
+
"std": [
|
| 193 |
+
0.11841528114040736,
|
| 194 |
+
0.1041359083946544,
|
| 195 |
+
0.08346265521208351,
|
| 196 |
+
0.0
|
| 197 |
+
],
|
| 198 |
+
"q01": [
|
| 199 |
+
-0.17562359675270237,
|
| 200 |
+
-0.21706402321529558,
|
| 201 |
+
-0.16273392585723104,
|
| 202 |
+
0.0
|
| 203 |
+
],
|
| 204 |
+
"q99": [
|
| 205 |
+
0.2572918701644887,
|
| 206 |
+
0.27022325575787115,
|
| 207 |
+
0.1369626751049278,
|
| 208 |
+
0.0
|
| 209 |
+
]
|
| 210 |
+
},
|
| 211 |
+
"control_mode": {
|
| 212 |
+
"max": [
|
| 213 |
+
1.0
|
| 214 |
+
],
|
| 215 |
+
"min": [
|
| 216 |
+
-1.0
|
| 217 |
+
],
|
| 218 |
+
"mean": [
|
| 219 |
+
-0.9205539608927732
|
| 220 |
+
],
|
| 221 |
+
"std": [
|
| 222 |
+
0.39063657242534666
|
| 223 |
+
],
|
| 224 |
+
"q01": [
|
| 225 |
+
-1.0
|
| 226 |
+
],
|
| 227 |
+
"q99": [
|
| 228 |
+
-0.17711863866562935
|
| 229 |
+
]
|
| 230 |
+
},
|
| 231 |
+
"end_effector_position": {
|
| 232 |
+
"max": [
|
| 233 |
+
1.0,
|
| 234 |
+
1.0,
|
| 235 |
+
1.0
|
| 236 |
+
],
|
| 237 |
+
"min": [
|
| 238 |
+
-1.0,
|
| 239 |
+
-1.0,
|
| 240 |
+
-1.0
|
| 241 |
+
],
|
| 242 |
+
"mean": [
|
| 243 |
+
-0.00924803964621179,
|
| 244 |
+
0.004053841761390575,
|
| 245 |
+
-0.08385115578460103
|
| 246 |
+
],
|
| 247 |
+
"std": [
|
| 248 |
+
0.45413072518223824,
|
| 249 |
+
0.40158612714634856,
|
| 250 |
+
0.41068634292849304
|
| 251 |
+
],
|
| 252 |
+
"q01": [
|
| 253 |
+
-0.9669817145875343,
|
| 254 |
+
-0.9241307799783718,
|
| 255 |
+
-0.8264079503068825
|
| 256 |
+
],
|
| 257 |
+
"q99": [
|
| 258 |
+
0.8021511975722166,
|
| 259 |
+
0.9241346582748478,
|
| 260 |
+
0.8348116781502298
|
| 261 |
+
]
|
| 262 |
+
},
|
| 263 |
+
"end_effector_rotation": {
|
| 264 |
+
"max": [
|
| 265 |
+
1.0,
|
| 266 |
+
1.0,
|
| 267 |
+
1.0
|
| 268 |
+
],
|
| 269 |
+
"min": [
|
| 270 |
+
-1.0,
|
| 271 |
+
-1.0,
|
| 272 |
+
-1.0
|
| 273 |
+
],
|
| 274 |
+
"mean": [
|
| 275 |
+
0.007644157504353809,
|
| 276 |
+
-0.019962543093886335,
|
| 277 |
+
-0.002047110452602617
|
| 278 |
+
],
|
| 279 |
+
"std": [
|
| 280 |
+
0.11076012154791087,
|
| 281 |
+
0.13176735140285795,
|
| 282 |
+
0.12286528665895614
|
| 283 |
+
],
|
| 284 |
+
"q01": [
|
| 285 |
+
-0.2659105439401192,
|
| 286 |
+
-0.3961694033813533,
|
| 287 |
+
-0.3434870832468936
|
| 288 |
+
],
|
| 289 |
+
"q99": [
|
| 290 |
+
0.3145786959266369,
|
| 291 |
+
0.31309430361227547,
|
| 292 |
+
0.34107754401942986
|
| 293 |
+
]
|
| 294 |
+
},
|
| 295 |
+
"gripper_close": {
|
| 296 |
+
"max": [
|
| 297 |
+
1.0
|
| 298 |
+
],
|
| 299 |
+
"min": [
|
| 300 |
+
-1.0
|
| 301 |
+
],
|
| 302 |
+
"mean": [
|
| 303 |
+
-0.29870612267533575
|
| 304 |
+
],
|
| 305 |
+
"std": [
|
| 306 |
+
0.9543487555375691
|
| 307 |
+
],
|
| 308 |
+
"q01": [
|
| 309 |
+
-1.0
|
| 310 |
+
],
|
| 311 |
+
"q99": [
|
| 312 |
+
0.7610674802484976
|
| 313 |
+
]
|
| 314 |
+
}
|
| 315 |
+
}
|
| 316 |
+
},
|
| 317 |
+
"modalities": {
|
| 318 |
+
"video": {
|
| 319 |
+
"robot0_eye_in_hand": {
|
| 320 |
+
"resolution": [
|
| 321 |
+
256,
|
| 322 |
+
256
|
| 323 |
+
],
|
| 324 |
+
"channels": 3,
|
| 325 |
+
"fps": 20.0
|
| 326 |
+
},
|
| 327 |
+
"robot0_agentview_left": {
|
| 328 |
+
"resolution": [
|
| 329 |
+
256,
|
| 330 |
+
256
|
| 331 |
+
],
|
| 332 |
+
"channels": 3,
|
| 333 |
+
"fps": 20.0
|
| 334 |
+
},
|
| 335 |
+
"robot0_agentview_right": {
|
| 336 |
+
"resolution": [
|
| 337 |
+
256,
|
| 338 |
+
256
|
| 339 |
+
],
|
| 340 |
+
"channels": 3,
|
| 341 |
+
"fps": 20.0
|
| 342 |
+
}
|
| 343 |
+
},
|
| 344 |
+
"state": {
|
| 345 |
+
"base_position": {
|
| 346 |
+
"absolute": true,
|
| 347 |
+
"rotation_type": null,
|
| 348 |
+
"shape": [
|
| 349 |
+
3
|
| 350 |
+
],
|
| 351 |
+
"continuous": true
|
| 352 |
+
},
|
| 353 |
+
"base_rotation": {
|
| 354 |
+
"absolute": true,
|
| 355 |
+
"rotation_type": "quaternion",
|
| 356 |
+
"shape": [
|
| 357 |
+
4
|
| 358 |
+
],
|
| 359 |
+
"continuous": true
|
| 360 |
+
},
|
| 361 |
+
"end_effector_position_relative": {
|
| 362 |
+
"absolute": true,
|
| 363 |
+
"rotation_type": null,
|
| 364 |
+
"shape": [
|
| 365 |
+
3
|
| 366 |
+
],
|
| 367 |
+
"continuous": true
|
| 368 |
+
},
|
| 369 |
+
"end_effector_rotation_relative": {
|
| 370 |
+
"absolute": true,
|
| 371 |
+
"rotation_type": "quaternion",
|
| 372 |
+
"shape": [
|
| 373 |
+
4
|
| 374 |
+
],
|
| 375 |
+
"continuous": true
|
| 376 |
+
},
|
| 377 |
+
"gripper_qpos": {
|
| 378 |
+
"absolute": true,
|
| 379 |
+
"rotation_type": null,
|
| 380 |
+
"shape": [
|
| 381 |
+
2
|
| 382 |
+
],
|
| 383 |
+
"continuous": true
|
| 384 |
+
}
|
| 385 |
+
},
|
| 386 |
+
"action": {
|
| 387 |
+
"base_motion": {
|
| 388 |
+
"absolute": true,
|
| 389 |
+
"rotation_type": null,
|
| 390 |
+
"shape": [
|
| 391 |
+
4
|
| 392 |
+
],
|
| 393 |
+
"continuous": true
|
| 394 |
+
},
|
| 395 |
+
"control_mode": {
|
| 396 |
+
"absolute": true,
|
| 397 |
+
"rotation_type": null,
|
| 398 |
+
"shape": [
|
| 399 |
+
1
|
| 400 |
+
],
|
| 401 |
+
"continuous": true
|
| 402 |
+
},
|
| 403 |
+
"end_effector_position": {
|
| 404 |
+
"absolute": true,
|
| 405 |
+
"rotation_type": null,
|
| 406 |
+
"shape": [
|
| 407 |
+
3
|
| 408 |
+
],
|
| 409 |
+
"continuous": true
|
| 410 |
+
},
|
| 411 |
+
"end_effector_rotation": {
|
| 412 |
+
"absolute": true,
|
| 413 |
+
"rotation_type": "axis_angle",
|
| 414 |
+
"shape": [
|
| 415 |
+
3
|
| 416 |
+
],
|
| 417 |
+
"continuous": true
|
| 418 |
+
},
|
| 419 |
+
"gripper_close": {
|
| 420 |
+
"absolute": true,
|
| 421 |
+
"rotation_type": null,
|
| 422 |
+
"shape": [
|
| 423 |
+
1
|
| 424 |
+
],
|
| 425 |
+
"continuous": true
|
| 426 |
+
}
|
| 427 |
+
}
|
| 428 |
+
},
|
| 429 |
+
"embodiment_tag": "new_embodiment"
|
| 430 |
+
}
|
| 431 |
+
}
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-1000/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3cad739162e0ee8414ba4511b7fcaa0c2ddb8fe047d46496aa03423aeaeee6e4
|
| 3 |
+
size 4999367032
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-1000/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:33b7fd8ca7c5411df32b3e3aca2c478b6ba6fb382b541607e176c33f60b106a0
|
| 3 |
+
size 2586705312
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-1000/model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-1000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:79d26d43acc5ab7bea94836e04a88ec871c236de452bb39c79dfe60e17ed0674
|
| 3 |
+
size 14645
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-1000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6e19ead5779e3f8f2f666b7793de3a03d118f57f1b16a97eeb87bd732d18b445
|
| 3 |
+
size 1465
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-1000/trainer_state.json
ADDED
|
@@ -0,0 +1,734 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.675041876046901,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 1000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.01675041876046901,
|
| 14 |
+
"grad_norm": 1.3394557237625122,
|
| 15 |
+
"learning_rate": 1.08e-06,
|
| 16 |
+
"loss": 0.2322,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.03350083752093802,
|
| 21 |
+
"grad_norm": 1.0503181219100952,
|
| 22 |
+
"learning_rate": 2.28e-06,
|
| 23 |
+
"loss": 0.2236,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.05025125628140704,
|
| 28 |
+
"grad_norm": 0.8409915566444397,
|
| 29 |
+
"learning_rate": 3.48e-06,
|
| 30 |
+
"loss": 0.1834,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.06700167504187604,
|
| 35 |
+
"grad_norm": 1.0835442543029785,
|
| 36 |
+
"learning_rate": 4.68e-06,
|
| 37 |
+
"loss": 0.1607,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08375209380234507,
|
| 42 |
+
"grad_norm": 0.6771818399429321,
|
| 43 |
+
"learning_rate": 5.8800000000000005e-06,
|
| 44 |
+
"loss": 0.1415,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.10050251256281408,
|
| 49 |
+
"grad_norm": 0.690451443195343,
|
| 50 |
+
"learning_rate": 7.08e-06,
|
| 51 |
+
"loss": 0.137,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.11725293132328309,
|
| 56 |
+
"grad_norm": 0.3838907480239868,
|
| 57 |
+
"learning_rate": 8.28e-06,
|
| 58 |
+
"loss": 0.1083,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.13400335008375208,
|
| 63 |
+
"grad_norm": 0.3522876799106598,
|
| 64 |
+
"learning_rate": 9.48e-06,
|
| 65 |
+
"loss": 0.0972,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.1507537688442211,
|
| 70 |
+
"grad_norm": 0.2768699526786804,
|
| 71 |
+
"learning_rate": 1.068e-05,
|
| 72 |
+
"loss": 0.0842,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.16750418760469013,
|
| 77 |
+
"grad_norm": 0.27654263377189636,
|
| 78 |
+
"learning_rate": 1.1880000000000001e-05,
|
| 79 |
+
"loss": 0.0762,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.18425460636515914,
|
| 84 |
+
"grad_norm": 0.33774033188819885,
|
| 85 |
+
"learning_rate": 1.308e-05,
|
| 86 |
+
"loss": 0.0713,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.20100502512562815,
|
| 91 |
+
"grad_norm": 0.2147975116968155,
|
| 92 |
+
"learning_rate": 1.428e-05,
|
| 93 |
+
"loss": 0.0672,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.21775544388609716,
|
| 98 |
+
"grad_norm": 0.25137603282928467,
|
| 99 |
+
"learning_rate": 1.548e-05,
|
| 100 |
+
"loss": 0.0625,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.23450586264656617,
|
| 105 |
+
"grad_norm": 0.3030863404273987,
|
| 106 |
+
"learning_rate": 1.6680000000000003e-05,
|
| 107 |
+
"loss": 0.0694,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.25125628140703515,
|
| 112 |
+
"grad_norm": 0.2430053949356079,
|
| 113 |
+
"learning_rate": 1.7879999999999998e-05,
|
| 114 |
+
"loss": 0.0638,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.26800670016750416,
|
| 119 |
+
"grad_norm": 0.24571731686592102,
|
| 120 |
+
"learning_rate": 1.908e-05,
|
| 121 |
+
"loss": 0.0658,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.2847571189279732,
|
| 126 |
+
"grad_norm": 0.27303019165992737,
|
| 127 |
+
"learning_rate": 2.0280000000000002e-05,
|
| 128 |
+
"loss": 0.0633,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.3015075376884422,
|
| 133 |
+
"grad_norm": 0.23230038583278656,
|
| 134 |
+
"learning_rate": 2.148e-05,
|
| 135 |
+
"loss": 0.0564,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.31825795644891125,
|
| 140 |
+
"grad_norm": 0.2652761936187744,
|
| 141 |
+
"learning_rate": 2.268e-05,
|
| 142 |
+
"loss": 0.0579,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.33500837520938026,
|
| 147 |
+
"grad_norm": 0.3008716106414795,
|
| 148 |
+
"learning_rate": 2.3880000000000002e-05,
|
| 149 |
+
"loss": 0.0581,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.35175879396984927,
|
| 154 |
+
"grad_norm": 0.25785622000694275,
|
| 155 |
+
"learning_rate": 2.508e-05,
|
| 156 |
+
"loss": 0.0614,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.3685092127303183,
|
| 161 |
+
"grad_norm": 0.26564037799835205,
|
| 162 |
+
"learning_rate": 2.628e-05,
|
| 163 |
+
"loss": 0.0522,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.3852596314907873,
|
| 168 |
+
"grad_norm": 0.2962818741798401,
|
| 169 |
+
"learning_rate": 2.748e-05,
|
| 170 |
+
"loss": 0.0521,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.4020100502512563,
|
| 175 |
+
"grad_norm": 0.28678590059280396,
|
| 176 |
+
"learning_rate": 2.868e-05,
|
| 177 |
+
"loss": 0.0555,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.4187604690117253,
|
| 182 |
+
"grad_norm": 0.29701972007751465,
|
| 183 |
+
"learning_rate": 2.9880000000000002e-05,
|
| 184 |
+
"loss": 0.0517,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.4355108877721943,
|
| 189 |
+
"grad_norm": 0.2258671224117279,
|
| 190 |
+
"learning_rate": 2.9999734259635676e-05,
|
| 191 |
+
"loss": 0.0523,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.45226130653266333,
|
| 196 |
+
"grad_norm": 0.2276943176984787,
|
| 197 |
+
"learning_rate": 2.9998815663057245e-05,
|
| 198 |
+
"loss": 0.0491,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.46901172529313234,
|
| 203 |
+
"grad_norm": 0.367814838886261,
|
| 204 |
+
"learning_rate": 2.999724096969228e-05,
|
| 205 |
+
"loss": 0.0472,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.48576214405360135,
|
| 210 |
+
"grad_norm": 0.19225119054317474,
|
| 211 |
+
"learning_rate": 2.9995010248422984e-05,
|
| 212 |
+
"loss": 0.0452,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.5025125628140703,
|
| 217 |
+
"grad_norm": 0.3076336681842804,
|
| 218 |
+
"learning_rate": 2.999212359682832e-05,
|
| 219 |
+
"loss": 0.0478,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.5192629815745393,
|
| 224 |
+
"grad_norm": 0.2448616325855255,
|
| 225 |
+
"learning_rate": 2.9988581141179787e-05,
|
| 226 |
+
"loss": 0.0471,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.5360134003350083,
|
| 231 |
+
"grad_norm": 0.20978757739067078,
|
| 232 |
+
"learning_rate": 2.998438303643587e-05,
|
| 233 |
+
"loss": 0.0383,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.5527638190954773,
|
| 238 |
+
"grad_norm": 0.24052295088768005,
|
| 239 |
+
"learning_rate": 2.997952946623527e-05,
|
| 240 |
+
"loss": 0.0435,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.5695142378559463,
|
| 245 |
+
"grad_norm": 0.22263383865356445,
|
| 246 |
+
"learning_rate": 2.9974020642888875e-05,
|
| 247 |
+
"loss": 0.0467,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.5862646566164154,
|
| 252 |
+
"grad_norm": 0.25700053572654724,
|
| 253 |
+
"learning_rate": 2.9967856807370458e-05,
|
| 254 |
+
"loss": 0.0448,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.6030150753768844,
|
| 259 |
+
"grad_norm": 0.19276398420333862,
|
| 260 |
+
"learning_rate": 2.996103822930615e-05,
|
| 261 |
+
"loss": 0.0436,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.6197654941373534,
|
| 266 |
+
"grad_norm": 0.19240586459636688,
|
| 267 |
+
"learning_rate": 2.9953565206962653e-05,
|
| 268 |
+
"loss": 0.0406,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.6365159128978225,
|
| 273 |
+
"grad_norm": 0.23333503305912018,
|
| 274 |
+
"learning_rate": 2.9945438067234172e-05,
|
| 275 |
+
"loss": 0.0397,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.6532663316582915,
|
| 280 |
+
"grad_norm": 0.26178690791130066,
|
| 281 |
+
"learning_rate": 2.9936657165628124e-05,
|
| 282 |
+
"loss": 0.0394,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.6700167504187605,
|
| 287 |
+
"grad_norm": 0.16855783760547638,
|
| 288 |
+
"learning_rate": 2.992722288624959e-05,
|
| 289 |
+
"loss": 0.0385,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.6867671691792295,
|
| 294 |
+
"grad_norm": 0.1688416302204132,
|
| 295 |
+
"learning_rate": 2.9917135641784515e-05,
|
| 296 |
+
"loss": 0.0422,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.7035175879396985,
|
| 301 |
+
"grad_norm": 0.18370051681995392,
|
| 302 |
+
"learning_rate": 2.9906395873481655e-05,
|
| 303 |
+
"loss": 0.0371,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.7202680067001676,
|
| 308 |
+
"grad_norm": 0.2349836379289627,
|
| 309 |
+
"learning_rate": 2.989500405113326e-05,
|
| 310 |
+
"loss": 0.0413,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.7370184254606366,
|
| 315 |
+
"grad_norm": 0.20748260617256165,
|
| 316 |
+
"learning_rate": 2.9882960673054547e-05,
|
| 317 |
+
"loss": 0.0367,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.7537688442211056,
|
| 322 |
+
"grad_norm": 0.1702982783317566,
|
| 323 |
+
"learning_rate": 2.9870266266061886e-05,
|
| 324 |
+
"loss": 0.0379,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.7705192629815746,
|
| 329 |
+
"grad_norm": 0.26680412888526917,
|
| 330 |
+
"learning_rate": 2.9856921385449766e-05,
|
| 331 |
+
"loss": 0.0367,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.7872696817420436,
|
| 336 |
+
"grad_norm": 0.21671664714813232,
|
| 337 |
+
"learning_rate": 2.9842926614966502e-05,
|
| 338 |
+
"loss": 0.0367,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.8040201005025126,
|
| 343 |
+
"grad_norm": 0.20597104728221893,
|
| 344 |
+
"learning_rate": 2.982828256678868e-05,
|
| 345 |
+
"loss": 0.0377,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.8207705192629816,
|
| 350 |
+
"grad_norm": 0.25721216201782227,
|
| 351 |
+
"learning_rate": 2.9812989881494412e-05,
|
| 352 |
+
"loss": 0.0376,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.8375209380234506,
|
| 357 |
+
"grad_norm": 0.20260390639305115,
|
| 358 |
+
"learning_rate": 2.9797049228035296e-05,
|
| 359 |
+
"loss": 0.0353,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.8542713567839196,
|
| 364 |
+
"grad_norm": 0.18940532207489014,
|
| 365 |
+
"learning_rate": 2.978046130370715e-05,
|
| 366 |
+
"loss": 0.0326,
|
| 367 |
+
"step": 510
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.8710217755443886,
|
| 371 |
+
"grad_norm": 0.14810627698898315,
|
| 372 |
+
"learning_rate": 2.9763226834119535e-05,
|
| 373 |
+
"loss": 0.0359,
|
| 374 |
+
"step": 520
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.8877721943048577,
|
| 378 |
+
"grad_norm": 0.21682192385196686,
|
| 379 |
+
"learning_rate": 2.9745346573163963e-05,
|
| 380 |
+
"loss": 0.0343,
|
| 381 |
+
"step": 530
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.9045226130653267,
|
| 385 |
+
"grad_norm": 0.2462719827890396,
|
| 386 |
+
"learning_rate": 2.9726821302980987e-05,
|
| 387 |
+
"loss": 0.0344,
|
| 388 |
+
"step": 540
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.9212730318257957,
|
| 392 |
+
"grad_norm": 0.20271016657352448,
|
| 393 |
+
"learning_rate": 2.9707651833925935e-05,
|
| 394 |
+
"loss": 0.0295,
|
| 395 |
+
"step": 550
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.9380234505862647,
|
| 399 |
+
"grad_norm": 0.18994300067424774,
|
| 400 |
+
"learning_rate": 2.968783900453348e-05,
|
| 401 |
+
"loss": 0.0331,
|
| 402 |
+
"step": 560
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.9547738693467337,
|
| 406 |
+
"grad_norm": 0.1817820817232132,
|
| 407 |
+
"learning_rate": 2.9667383681480968e-05,
|
| 408 |
+
"loss": 0.0344,
|
| 409 |
+
"step": 570
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.9715242881072027,
|
| 413 |
+
"grad_norm": 0.20372416079044342,
|
| 414 |
+
"learning_rate": 2.9646286759550493e-05,
|
| 415 |
+
"loss": 0.0333,
|
| 416 |
+
"step": 580
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.9882747068676717,
|
| 420 |
+
"grad_norm": 0.22303369641304016,
|
| 421 |
+
"learning_rate": 2.9624549161589764e-05,
|
| 422 |
+
"loss": 0.0316,
|
| 423 |
+
"step": 590
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 1.0050251256281406,
|
| 427 |
+
"grad_norm": 0.15984657406806946,
|
| 428 |
+
"learning_rate": 2.9602171838471745e-05,
|
| 429 |
+
"loss": 0.0325,
|
| 430 |
+
"step": 600
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 1.0217755443886096,
|
| 434 |
+
"grad_norm": 0.18722732365131378,
|
| 435 |
+
"learning_rate": 2.957915576905303e-05,
|
| 436 |
+
"loss": 0.0304,
|
| 437 |
+
"step": 610
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 1.0385259631490786,
|
| 441 |
+
"grad_norm": 0.18170492351055145,
|
| 442 |
+
"learning_rate": 2.9555501960131058e-05,
|
| 443 |
+
"loss": 0.0314,
|
| 444 |
+
"step": 620
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 1.0552763819095476,
|
| 448 |
+
"grad_norm": 0.19419005513191223,
|
| 449 |
+
"learning_rate": 2.9531211446400068e-05,
|
| 450 |
+
"loss": 0.0336,
|
| 451 |
+
"step": 630
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 1.0720268006700167,
|
| 455 |
+
"grad_norm": 0.21015776693820953,
|
| 456 |
+
"learning_rate": 2.9506285290405813e-05,
|
| 457 |
+
"loss": 0.0312,
|
| 458 |
+
"step": 640
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 1.0887772194304857,
|
| 462 |
+
"grad_norm": 0.13695307075977325,
|
| 463 |
+
"learning_rate": 2.948072458249911e-05,
|
| 464 |
+
"loss": 0.0341,
|
| 465 |
+
"step": 650
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 1.1055276381909547,
|
| 469 |
+
"grad_norm": 0.16071368753910065,
|
| 470 |
+
"learning_rate": 2.9454530440788125e-05,
|
| 471 |
+
"loss": 0.0295,
|
| 472 |
+
"step": 660
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 1.1222780569514237,
|
| 476 |
+
"grad_norm": 0.18829545378684998,
|
| 477 |
+
"learning_rate": 2.9427704011089474e-05,
|
| 478 |
+
"loss": 0.0303,
|
| 479 |
+
"step": 670
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 1.1390284757118927,
|
| 483 |
+
"grad_norm": 0.1615035980939865,
|
| 484 |
+
"learning_rate": 2.94002464668781e-05,
|
| 485 |
+
"loss": 0.0294,
|
| 486 |
+
"step": 680
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 1.1557788944723617,
|
| 490 |
+
"grad_norm": 0.1708756685256958,
|
| 491 |
+
"learning_rate": 2.9372159009235936e-05,
|
| 492 |
+
"loss": 0.0278,
|
| 493 |
+
"step": 690
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 1.1725293132328307,
|
| 497 |
+
"grad_norm": 0.1325443536043167,
|
| 498 |
+
"learning_rate": 2.9343442866799365e-05,
|
| 499 |
+
"loss": 0.0289,
|
| 500 |
+
"step": 700
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 1.1892797319932997,
|
| 504 |
+
"grad_norm": 0.18041647970676422,
|
| 505 |
+
"learning_rate": 2.9314099295705495e-05,
|
| 506 |
+
"loss": 0.0316,
|
| 507 |
+
"step": 710
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 1.2060301507537687,
|
| 511 |
+
"grad_norm": 0.15595586597919464,
|
| 512 |
+
"learning_rate": 2.928412957953718e-05,
|
| 513 |
+
"loss": 0.0294,
|
| 514 |
+
"step": 720
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 1.2227805695142377,
|
| 518 |
+
"grad_norm": 0.18720653653144836,
|
| 519 |
+
"learning_rate": 2.925353502926689e-05,
|
| 520 |
+
"loss": 0.0292,
|
| 521 |
+
"step": 730
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 1.2395309882747068,
|
| 525 |
+
"grad_norm": 0.1344250738620758,
|
| 526 |
+
"learning_rate": 2.922231698319937e-05,
|
| 527 |
+
"loss": 0.0299,
|
| 528 |
+
"step": 740
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 1.2562814070351758,
|
| 532 |
+
"grad_norm": 0.18409386277198792,
|
| 533 |
+
"learning_rate": 2.9190476806913088e-05,
|
| 534 |
+
"loss": 0.0305,
|
| 535 |
+
"step": 750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 1.2730318257956448,
|
| 539 |
+
"grad_norm": 0.16554060578346252,
|
| 540 |
+
"learning_rate": 2.9158015893200504e-05,
|
| 541 |
+
"loss": 0.0292,
|
| 542 |
+
"step": 760
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 1.2897822445561138,
|
| 546 |
+
"grad_norm": 0.1602775752544403,
|
| 547 |
+
"learning_rate": 2.9124935662007137e-05,
|
| 548 |
+
"loss": 0.03,
|
| 549 |
+
"step": 770
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.3065326633165828,
|
| 553 |
+
"grad_norm": 0.1876162886619568,
|
| 554 |
+
"learning_rate": 2.9091237560369467e-05,
|
| 555 |
+
"loss": 0.0284,
|
| 556 |
+
"step": 780
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 1.3232830820770518,
|
| 560 |
+
"grad_norm": 0.24121980369091034,
|
| 561 |
+
"learning_rate": 2.905692306235163e-05,
|
| 562 |
+
"loss": 0.0282,
|
| 563 |
+
"step": 790
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 1.3400335008375208,
|
| 567 |
+
"grad_norm": 0.15151315927505493,
|
| 568 |
+
"learning_rate": 2.9021993668980925e-05,
|
| 569 |
+
"loss": 0.0288,
|
| 570 |
+
"step": 800
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 1.3567839195979898,
|
| 574 |
+
"grad_norm": 0.32094353437423706,
|
| 575 |
+
"learning_rate": 2.8986450908182185e-05,
|
| 576 |
+
"loss": 0.0265,
|
| 577 |
+
"step": 810
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 1.3735343383584588,
|
| 581 |
+
"grad_norm": 0.20612798631191254,
|
| 582 |
+
"learning_rate": 2.8950296334710902e-05,
|
| 583 |
+
"loss": 0.0281,
|
| 584 |
+
"step": 820
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 1.3902847571189278,
|
| 588 |
+
"grad_norm": 0.16587623953819275,
|
| 589 |
+
"learning_rate": 2.8913531530085248e-05,
|
| 590 |
+
"loss": 0.0272,
|
| 591 |
+
"step": 830
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 1.4070351758793969,
|
| 595 |
+
"grad_norm": 0.1658754050731659,
|
| 596 |
+
"learning_rate": 2.887615810251687e-05,
|
| 597 |
+
"loss": 0.0279,
|
| 598 |
+
"step": 840
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 1.4237855946398659,
|
| 602 |
+
"grad_norm": 0.155488982796669,
|
| 603 |
+
"learning_rate": 2.8838177686840577e-05,
|
| 604 |
+
"loss": 0.0266,
|
| 605 |
+
"step": 850
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 1.4405360134003349,
|
| 609 |
+
"grad_norm": 0.2229546457529068,
|
| 610 |
+
"learning_rate": 2.8799591944442776e-05,
|
| 611 |
+
"loss": 0.0292,
|
| 612 |
+
"step": 860
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 1.457286432160804,
|
| 616 |
+
"grad_norm": 0.36190536618232727,
|
| 617 |
+
"learning_rate": 2.8760402563188848e-05,
|
| 618 |
+
"loss": 0.0277,
|
| 619 |
+
"step": 870
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 1.474036850921273,
|
| 623 |
+
"grad_norm": 0.18090955913066864,
|
| 624 |
+
"learning_rate": 2.8720611257349285e-05,
|
| 625 |
+
"loss": 0.0297,
|
| 626 |
+
"step": 880
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 1.490787269681742,
|
| 630 |
+
"grad_norm": 0.17235404253005981,
|
| 631 |
+
"learning_rate": 2.868021976752471e-05,
|
| 632 |
+
"loss": 0.0265,
|
| 633 |
+
"step": 890
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 1.507537688442211,
|
| 637 |
+
"grad_norm": 0.1455887258052826,
|
| 638 |
+
"learning_rate": 2.8639229860569738e-05,
|
| 639 |
+
"loss": 0.0298,
|
| 640 |
+
"step": 900
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 1.52428810720268,
|
| 644 |
+
"grad_norm": 0.1774197816848755,
|
| 645 |
+
"learning_rate": 2.8597643329515688e-05,
|
| 646 |
+
"loss": 0.0263,
|
| 647 |
+
"step": 910
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 1.541038525963149,
|
| 651 |
+
"grad_norm": 0.15991894900798798,
|
| 652 |
+
"learning_rate": 2.855546199349214e-05,
|
| 653 |
+
"loss": 0.0259,
|
| 654 |
+
"step": 920
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 1.557788944723618,
|
| 658 |
+
"grad_norm": 0.30203819274902344,
|
| 659 |
+
"learning_rate": 2.85126876976474e-05,
|
| 660 |
+
"loss": 0.0276,
|
| 661 |
+
"step": 930
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 1.574539363484087,
|
| 665 |
+
"grad_norm": 0.13825957477092743,
|
| 666 |
+
"learning_rate": 2.8469322313067726e-05,
|
| 667 |
+
"loss": 0.0262,
|
| 668 |
+
"step": 940
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 1.591289782244556,
|
| 672 |
+
"grad_norm": 0.19100086390972137,
|
| 673 |
+
"learning_rate": 2.8425367736695522e-05,
|
| 674 |
+
"loss": 0.0257,
|
| 675 |
+
"step": 950
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 1.608040201005025,
|
| 679 |
+
"grad_norm": 0.16253994405269623,
|
| 680 |
+
"learning_rate": 2.838082589124636e-05,
|
| 681 |
+
"loss": 0.0282,
|
| 682 |
+
"step": 960
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 1.624790619765494,
|
| 686 |
+
"grad_norm": 0.21166931092739105,
|
| 687 |
+
"learning_rate": 2.8335698725124843e-05,
|
| 688 |
+
"loss": 0.0278,
|
| 689 |
+
"step": 970
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 1.641541038525963,
|
| 693 |
+
"grad_norm": 0.13361522555351257,
|
| 694 |
+
"learning_rate": 2.8289988212339422e-05,
|
| 695 |
+
"loss": 0.0254,
|
| 696 |
+
"step": 980
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 1.658291457286432,
|
| 700 |
+
"grad_norm": 0.20177683234214783,
|
| 701 |
+
"learning_rate": 2.8243696352415997e-05,
|
| 702 |
+
"loss": 0.0317,
|
| 703 |
+
"step": 990
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 1.675041876046901,
|
| 707 |
+
"grad_norm": 0.17741645872592926,
|
| 708 |
+
"learning_rate": 2.8196825170310495e-05,
|
| 709 |
+
"loss": 0.0254,
|
| 710 |
+
"step": 1000
|
| 711 |
+
}
|
| 712 |
+
],
|
| 713 |
+
"logging_steps": 10,
|
| 714 |
+
"max_steps": 5000,
|
| 715 |
+
"num_input_tokens_seen": 0,
|
| 716 |
+
"num_train_epochs": 9,
|
| 717 |
+
"save_steps": 5001,
|
| 718 |
+
"stateful_callbacks": {
|
| 719 |
+
"TrainerControl": {
|
| 720 |
+
"args": {
|
| 721 |
+
"should_epoch_stop": false,
|
| 722 |
+
"should_evaluate": false,
|
| 723 |
+
"should_log": false,
|
| 724 |
+
"should_save": true,
|
| 725 |
+
"should_training_stop": false
|
| 726 |
+
},
|
| 727 |
+
"attributes": {}
|
| 728 |
+
}
|
| 729 |
+
},
|
| 730 |
+
"total_flos": 0.0,
|
| 731 |
+
"train_batch_size": 64,
|
| 732 |
+
"trial_name": null,
|
| 733 |
+
"trial_params": null
|
| 734 |
+
}
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-3000/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 |
+
}
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-3000/experiment_cfg/metadata.json
ADDED
|
@@ -0,0 +1,431 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"new_embodiment": {
|
| 3 |
+
"statistics": {
|
| 4 |
+
"state": {
|
| 5 |
+
"base_position": {
|
| 6 |
+
"max": [
|
| 7 |
+
7.778976917266846,
|
| 8 |
+
0.33403685688972473,
|
| 9 |
+
0.7223060727119446
|
| 10 |
+
],
|
| 11 |
+
"min": [
|
| 12 |
+
-4.821934223175049,
|
| 13 |
+
-6.785110950469971,
|
| 14 |
+
0.6994727253913879
|
| 15 |
+
],
|
| 16 |
+
"mean": [
|
| 17 |
+
2.5719731235192254,
|
| 18 |
+
-2.1489371002267235,
|
| 19 |
+
0.7003612741268546
|
| 20 |
+
],
|
| 21 |
+
"std": [
|
| 22 |
+
1.669945884010784,
|
| 23 |
+
1.3739411381361726,
|
| 24 |
+
0.0009413282367754987
|
| 25 |
+
],
|
| 26 |
+
"q01": [
|
| 27 |
+
-0.7944286339399775,
|
| 28 |
+
-5.486611518471242,
|
| 29 |
+
0.699977555971739
|
| 30 |
+
],
|
| 31 |
+
"q99": [
|
| 32 |
+
6.019886905263648,
|
| 33 |
+
-0.5229184210026855,
|
| 34 |
+
0.702930326282967
|
| 35 |
+
]
|
| 36 |
+
},
|
| 37 |
+
"base_rotation": {
|
| 38 |
+
"max": [
|
| 39 |
+
0.0,
|
| 40 |
+
0.0,
|
| 41 |
+
1.0,
|
| 42 |
+
1.0
|
| 43 |
+
],
|
| 44 |
+
"min": [
|
| 45 |
+
0.0,
|
| 46 |
+
0.0,
|
| 47 |
+
-1.0,
|
| 48 |
+
0.0
|
| 49 |
+
],
|
| 50 |
+
"mean": [
|
| 51 |
+
0.0,
|
| 52 |
+
0.0,
|
| 53 |
+
0.248543484335273,
|
| 54 |
+
0.5959317405459531
|
| 55 |
+
],
|
| 56 |
+
"std": [
|
| 57 |
+
0.0,
|
| 58 |
+
0.0,
|
| 59 |
+
0.6655550240074092,
|
| 60 |
+
0.374283995143559
|
| 61 |
+
],
|
| 62 |
+
"q01": [
|
| 63 |
+
0.0,
|
| 64 |
+
0.0,
|
| 65 |
+
-1.0000000000000002,
|
| 66 |
+
4.503254968910557e-06
|
| 67 |
+
],
|
| 68 |
+
"q99": [
|
| 69 |
+
0.0,
|
| 70 |
+
0.0,
|
| 71 |
+
1.0000000000000002,
|
| 72 |
+
1.0000000000000002
|
| 73 |
+
]
|
| 74 |
+
},
|
| 75 |
+
"end_effector_position_relative": {
|
| 76 |
+
"max": [
|
| 77 |
+
0.8848381638526917,
|
| 78 |
+
0.8569459915161133,
|
| 79 |
+
1.053884744644165
|
| 80 |
+
],
|
| 81 |
+
"min": [
|
| 82 |
+
-0.5828549861907959,
|
| 83 |
+
-0.8486236333847046,
|
| 84 |
+
-0.3446122705936432
|
| 85 |
+
],
|
| 86 |
+
"mean": [
|
| 87 |
+
0.2465490874130862,
|
| 88 |
+
-0.004711658908481467,
|
| 89 |
+
0.43463835464312073
|
| 90 |
+
],
|
| 91 |
+
"std": [
|
| 92 |
+
0.15798784670671595,
|
| 93 |
+
0.25624102400497073,
|
| 94 |
+
0.20836028727974915
|
| 95 |
+
],
|
| 96 |
+
"q01": [
|
| 97 |
+
-0.020752051134640608,
|
| 98 |
+
-0.4837933887347875,
|
| 99 |
+
0.1802942952289915
|
| 100 |
+
],
|
| 101 |
+
"q99": [
|
| 102 |
+
0.5386784714844007,
|
| 103 |
+
0.4801424212024721,
|
| 104 |
+
0.695095732975766
|
| 105 |
+
]
|
| 106 |
+
},
|
| 107 |
+
"end_effector_rotation_relative": {
|
| 108 |
+
"max": [
|
| 109 |
+
0.9999949932098389,
|
| 110 |
+
0.9997354745864868,
|
| 111 |
+
0.9900747537612915,
|
| 112 |
+
0.9735639095306396
|
| 113 |
+
],
|
| 114 |
+
"min": [
|
| 115 |
+
-0.9999914169311523,
|
| 116 |
+
-0.9999884366989136,
|
| 117 |
+
-0.9975369572639465,
|
| 118 |
+
6.754255821306288e-08
|
| 119 |
+
],
|
| 120 |
+
"mean": [
|
| 121 |
+
-0.22531859714056926,
|
| 122 |
+
0.046973702821026614,
|
| 123 |
+
-0.0568176967878771,
|
| 124 |
+
0.12364307042955697
|
| 125 |
+
],
|
| 126 |
+
"std": [
|
| 127 |
+
0.8502144333612052,
|
| 128 |
+
0.29591250318147405,
|
| 129 |
+
0.30218053156837016,
|
| 130 |
+
0.1636031353412931
|
| 131 |
+
],
|
| 132 |
+
"q01": [
|
| 133 |
+
-0.9959057387172566,
|
| 134 |
+
-0.5665384745573048,
|
| 135 |
+
-0.5329334620834977,
|
| 136 |
+
0.0014372279999397553
|
| 137 |
+
],
|
| 138 |
+
"q99": [
|
| 139 |
+
0.9904107951565839,
|
| 140 |
+
0.6359623612021396,
|
| 141 |
+
0.5026076022439377,
|
| 142 |
+
0.42955559234879914
|
| 143 |
+
]
|
| 144 |
+
},
|
| 145 |
+
"gripper_qpos": {
|
| 146 |
+
"max": [
|
| 147 |
+
0.05289580300450325,
|
| 148 |
+
0.00937709677964449
|
| 149 |
+
],
|
| 150 |
+
"min": [
|
| 151 |
+
-0.011669117026031017,
|
| 152 |
+
-0.06307756155729294
|
| 153 |
+
],
|
| 154 |
+
"mean": [
|
| 155 |
+
0.03188757060429838,
|
| 156 |
+
-0.0321793333086053
|
| 157 |
+
],
|
| 158 |
+
"std": [
|
| 159 |
+
0.01171052313737148,
|
| 160 |
+
0.011427336801203189
|
| 161 |
+
],
|
| 162 |
+
"q01": [
|
| 163 |
+
0.006462901319459925,
|
| 164 |
+
-0.040557140599419166
|
| 165 |
+
],
|
| 166 |
+
"q99": [
|
| 167 |
+
0.04037619335388386,
|
| 168 |
+
-0.006523947835993347
|
| 169 |
+
]
|
| 170 |
+
}
|
| 171 |
+
},
|
| 172 |
+
"action": {
|
| 173 |
+
"base_motion": {
|
| 174 |
+
"max": [
|
| 175 |
+
1.0,
|
| 176 |
+
1.0,
|
| 177 |
+
1.0,
|
| 178 |
+
0.0
|
| 179 |
+
],
|
| 180 |
+
"min": [
|
| 181 |
+
-1.0,
|
| 182 |
+
-1.0,
|
| 183 |
+
-1.0,
|
| 184 |
+
0.0
|
| 185 |
+
],
|
| 186 |
+
"mean": [
|
| 187 |
+
0.006639007355857881,
|
| 188 |
+
0.0027330421795561874,
|
| 189 |
+
-0.0017774441970214892,
|
| 190 |
+
0.0
|
| 191 |
+
],
|
| 192 |
+
"std": [
|
| 193 |
+
0.11841528114040736,
|
| 194 |
+
0.1041359083946544,
|
| 195 |
+
0.08346265521208351,
|
| 196 |
+
0.0
|
| 197 |
+
],
|
| 198 |
+
"q01": [
|
| 199 |
+
-0.17562359675270237,
|
| 200 |
+
-0.21706402321529558,
|
| 201 |
+
-0.16273392585723104,
|
| 202 |
+
0.0
|
| 203 |
+
],
|
| 204 |
+
"q99": [
|
| 205 |
+
0.2572918701644887,
|
| 206 |
+
0.27022325575787115,
|
| 207 |
+
0.1369626751049278,
|
| 208 |
+
0.0
|
| 209 |
+
]
|
| 210 |
+
},
|
| 211 |
+
"control_mode": {
|
| 212 |
+
"max": [
|
| 213 |
+
1.0
|
| 214 |
+
],
|
| 215 |
+
"min": [
|
| 216 |
+
-1.0
|
| 217 |
+
],
|
| 218 |
+
"mean": [
|
| 219 |
+
-0.9205539608927732
|
| 220 |
+
],
|
| 221 |
+
"std": [
|
| 222 |
+
0.39063657242534666
|
| 223 |
+
],
|
| 224 |
+
"q01": [
|
| 225 |
+
-1.0
|
| 226 |
+
],
|
| 227 |
+
"q99": [
|
| 228 |
+
-0.17711863866562935
|
| 229 |
+
]
|
| 230 |
+
},
|
| 231 |
+
"end_effector_position": {
|
| 232 |
+
"max": [
|
| 233 |
+
1.0,
|
| 234 |
+
1.0,
|
| 235 |
+
1.0
|
| 236 |
+
],
|
| 237 |
+
"min": [
|
| 238 |
+
-1.0,
|
| 239 |
+
-1.0,
|
| 240 |
+
-1.0
|
| 241 |
+
],
|
| 242 |
+
"mean": [
|
| 243 |
+
-0.00924803964621179,
|
| 244 |
+
0.004053841761390575,
|
| 245 |
+
-0.08385115578460103
|
| 246 |
+
],
|
| 247 |
+
"std": [
|
| 248 |
+
0.45413072518223824,
|
| 249 |
+
0.40158612714634856,
|
| 250 |
+
0.41068634292849304
|
| 251 |
+
],
|
| 252 |
+
"q01": [
|
| 253 |
+
-0.9669817145875343,
|
| 254 |
+
-0.9241307799783718,
|
| 255 |
+
-0.8264079503068825
|
| 256 |
+
],
|
| 257 |
+
"q99": [
|
| 258 |
+
0.8021511975722166,
|
| 259 |
+
0.9241346582748478,
|
| 260 |
+
0.8348116781502298
|
| 261 |
+
]
|
| 262 |
+
},
|
| 263 |
+
"end_effector_rotation": {
|
| 264 |
+
"max": [
|
| 265 |
+
1.0,
|
| 266 |
+
1.0,
|
| 267 |
+
1.0
|
| 268 |
+
],
|
| 269 |
+
"min": [
|
| 270 |
+
-1.0,
|
| 271 |
+
-1.0,
|
| 272 |
+
-1.0
|
| 273 |
+
],
|
| 274 |
+
"mean": [
|
| 275 |
+
0.007644157504353809,
|
| 276 |
+
-0.019962543093886335,
|
| 277 |
+
-0.002047110452602617
|
| 278 |
+
],
|
| 279 |
+
"std": [
|
| 280 |
+
0.11076012154791087,
|
| 281 |
+
0.13176735140285795,
|
| 282 |
+
0.12286528665895614
|
| 283 |
+
],
|
| 284 |
+
"q01": [
|
| 285 |
+
-0.2659105439401192,
|
| 286 |
+
-0.3961694033813533,
|
| 287 |
+
-0.3434870832468936
|
| 288 |
+
],
|
| 289 |
+
"q99": [
|
| 290 |
+
0.3145786959266369,
|
| 291 |
+
0.31309430361227547,
|
| 292 |
+
0.34107754401942986
|
| 293 |
+
]
|
| 294 |
+
},
|
| 295 |
+
"gripper_close": {
|
| 296 |
+
"max": [
|
| 297 |
+
1.0
|
| 298 |
+
],
|
| 299 |
+
"min": [
|
| 300 |
+
-1.0
|
| 301 |
+
],
|
| 302 |
+
"mean": [
|
| 303 |
+
-0.29870612267533575
|
| 304 |
+
],
|
| 305 |
+
"std": [
|
| 306 |
+
0.9543487555375691
|
| 307 |
+
],
|
| 308 |
+
"q01": [
|
| 309 |
+
-1.0
|
| 310 |
+
],
|
| 311 |
+
"q99": [
|
| 312 |
+
0.7610674802484976
|
| 313 |
+
]
|
| 314 |
+
}
|
| 315 |
+
}
|
| 316 |
+
},
|
| 317 |
+
"modalities": {
|
| 318 |
+
"video": {
|
| 319 |
+
"robot0_eye_in_hand": {
|
| 320 |
+
"resolution": [
|
| 321 |
+
256,
|
| 322 |
+
256
|
| 323 |
+
],
|
| 324 |
+
"channels": 3,
|
| 325 |
+
"fps": 20.0
|
| 326 |
+
},
|
| 327 |
+
"robot0_agentview_left": {
|
| 328 |
+
"resolution": [
|
| 329 |
+
256,
|
| 330 |
+
256
|
| 331 |
+
],
|
| 332 |
+
"channels": 3,
|
| 333 |
+
"fps": 20.0
|
| 334 |
+
},
|
| 335 |
+
"robot0_agentview_right": {
|
| 336 |
+
"resolution": [
|
| 337 |
+
256,
|
| 338 |
+
256
|
| 339 |
+
],
|
| 340 |
+
"channels": 3,
|
| 341 |
+
"fps": 20.0
|
| 342 |
+
}
|
| 343 |
+
},
|
| 344 |
+
"state": {
|
| 345 |
+
"base_position": {
|
| 346 |
+
"absolute": true,
|
| 347 |
+
"rotation_type": null,
|
| 348 |
+
"shape": [
|
| 349 |
+
3
|
| 350 |
+
],
|
| 351 |
+
"continuous": true
|
| 352 |
+
},
|
| 353 |
+
"base_rotation": {
|
| 354 |
+
"absolute": true,
|
| 355 |
+
"rotation_type": "quaternion",
|
| 356 |
+
"shape": [
|
| 357 |
+
4
|
| 358 |
+
],
|
| 359 |
+
"continuous": true
|
| 360 |
+
},
|
| 361 |
+
"end_effector_position_relative": {
|
| 362 |
+
"absolute": true,
|
| 363 |
+
"rotation_type": null,
|
| 364 |
+
"shape": [
|
| 365 |
+
3
|
| 366 |
+
],
|
| 367 |
+
"continuous": true
|
| 368 |
+
},
|
| 369 |
+
"end_effector_rotation_relative": {
|
| 370 |
+
"absolute": true,
|
| 371 |
+
"rotation_type": "quaternion",
|
| 372 |
+
"shape": [
|
| 373 |
+
4
|
| 374 |
+
],
|
| 375 |
+
"continuous": true
|
| 376 |
+
},
|
| 377 |
+
"gripper_qpos": {
|
| 378 |
+
"absolute": true,
|
| 379 |
+
"rotation_type": null,
|
| 380 |
+
"shape": [
|
| 381 |
+
2
|
| 382 |
+
],
|
| 383 |
+
"continuous": true
|
| 384 |
+
}
|
| 385 |
+
},
|
| 386 |
+
"action": {
|
| 387 |
+
"base_motion": {
|
| 388 |
+
"absolute": true,
|
| 389 |
+
"rotation_type": null,
|
| 390 |
+
"shape": [
|
| 391 |
+
4
|
| 392 |
+
],
|
| 393 |
+
"continuous": true
|
| 394 |
+
},
|
| 395 |
+
"control_mode": {
|
| 396 |
+
"absolute": true,
|
| 397 |
+
"rotation_type": null,
|
| 398 |
+
"shape": [
|
| 399 |
+
1
|
| 400 |
+
],
|
| 401 |
+
"continuous": true
|
| 402 |
+
},
|
| 403 |
+
"end_effector_position": {
|
| 404 |
+
"absolute": true,
|
| 405 |
+
"rotation_type": null,
|
| 406 |
+
"shape": [
|
| 407 |
+
3
|
| 408 |
+
],
|
| 409 |
+
"continuous": true
|
| 410 |
+
},
|
| 411 |
+
"end_effector_rotation": {
|
| 412 |
+
"absolute": true,
|
| 413 |
+
"rotation_type": "axis_angle",
|
| 414 |
+
"shape": [
|
| 415 |
+
3
|
| 416 |
+
],
|
| 417 |
+
"continuous": true
|
| 418 |
+
},
|
| 419 |
+
"gripper_close": {
|
| 420 |
+
"absolute": true,
|
| 421 |
+
"rotation_type": null,
|
| 422 |
+
"shape": [
|
| 423 |
+
1
|
| 424 |
+
],
|
| 425 |
+
"continuous": true
|
| 426 |
+
}
|
| 427 |
+
}
|
| 428 |
+
},
|
| 429 |
+
"embodiment_tag": "new_embodiment"
|
| 430 |
+
}
|
| 431 |
+
}
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-3000/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:198273fb376d0ef6bf0de2576f81b6e5621cfaa0811f0315876056eb703889b5
|
| 3 |
+
size 4999367032
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-3000/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6d1fc016c68b2a0b15538d66fc764cd9d97818706db8fdc808bc880466fd1ab1
|
| 3 |
+
size 2586705312
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-3000/model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-3000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c92d34f806bd357292f6ef538c5846496ee7354432a4e1a4328334bd4b95bc43
|
| 3 |
+
size 8550719959
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-3000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:85cf750ef17be819e7281f7355fdb6fe24e17d7b2f3f773c8f182329ad9e5dc9
|
| 3 |
+
size 14645
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-3000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:42d8e6d8e53e6b837084d0cc18fdd63ca9e40bd44a65d38eb52695647a8e84e8
|
| 3 |
+
size 1465
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/checkpoint-3000/trainer_state.json
ADDED
|
@@ -0,0 +1,2134 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 5.025125628140704,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 3000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.01675041876046901,
|
| 14 |
+
"grad_norm": 1.3394557237625122,
|
| 15 |
+
"learning_rate": 1.08e-06,
|
| 16 |
+
"loss": 0.2322,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.03350083752093802,
|
| 21 |
+
"grad_norm": 1.0503181219100952,
|
| 22 |
+
"learning_rate": 2.28e-06,
|
| 23 |
+
"loss": 0.2236,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.05025125628140704,
|
| 28 |
+
"grad_norm": 0.8409915566444397,
|
| 29 |
+
"learning_rate": 3.48e-06,
|
| 30 |
+
"loss": 0.1834,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.06700167504187604,
|
| 35 |
+
"grad_norm": 1.0835442543029785,
|
| 36 |
+
"learning_rate": 4.68e-06,
|
| 37 |
+
"loss": 0.1607,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08375209380234507,
|
| 42 |
+
"grad_norm": 0.6771818399429321,
|
| 43 |
+
"learning_rate": 5.8800000000000005e-06,
|
| 44 |
+
"loss": 0.1415,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.10050251256281408,
|
| 49 |
+
"grad_norm": 0.690451443195343,
|
| 50 |
+
"learning_rate": 7.08e-06,
|
| 51 |
+
"loss": 0.137,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.11725293132328309,
|
| 56 |
+
"grad_norm": 0.3838907480239868,
|
| 57 |
+
"learning_rate": 8.28e-06,
|
| 58 |
+
"loss": 0.1083,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.13400335008375208,
|
| 63 |
+
"grad_norm": 0.3522876799106598,
|
| 64 |
+
"learning_rate": 9.48e-06,
|
| 65 |
+
"loss": 0.0972,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.1507537688442211,
|
| 70 |
+
"grad_norm": 0.2768699526786804,
|
| 71 |
+
"learning_rate": 1.068e-05,
|
| 72 |
+
"loss": 0.0842,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.16750418760469013,
|
| 77 |
+
"grad_norm": 0.27654263377189636,
|
| 78 |
+
"learning_rate": 1.1880000000000001e-05,
|
| 79 |
+
"loss": 0.0762,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.18425460636515914,
|
| 84 |
+
"grad_norm": 0.33774033188819885,
|
| 85 |
+
"learning_rate": 1.308e-05,
|
| 86 |
+
"loss": 0.0713,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.20100502512562815,
|
| 91 |
+
"grad_norm": 0.2147975116968155,
|
| 92 |
+
"learning_rate": 1.428e-05,
|
| 93 |
+
"loss": 0.0672,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.21775544388609716,
|
| 98 |
+
"grad_norm": 0.25137603282928467,
|
| 99 |
+
"learning_rate": 1.548e-05,
|
| 100 |
+
"loss": 0.0625,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.23450586264656617,
|
| 105 |
+
"grad_norm": 0.3030863404273987,
|
| 106 |
+
"learning_rate": 1.6680000000000003e-05,
|
| 107 |
+
"loss": 0.0694,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.25125628140703515,
|
| 112 |
+
"grad_norm": 0.2430053949356079,
|
| 113 |
+
"learning_rate": 1.7879999999999998e-05,
|
| 114 |
+
"loss": 0.0638,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.26800670016750416,
|
| 119 |
+
"grad_norm": 0.24571731686592102,
|
| 120 |
+
"learning_rate": 1.908e-05,
|
| 121 |
+
"loss": 0.0658,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.2847571189279732,
|
| 126 |
+
"grad_norm": 0.27303019165992737,
|
| 127 |
+
"learning_rate": 2.0280000000000002e-05,
|
| 128 |
+
"loss": 0.0633,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.3015075376884422,
|
| 133 |
+
"grad_norm": 0.23230038583278656,
|
| 134 |
+
"learning_rate": 2.148e-05,
|
| 135 |
+
"loss": 0.0564,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.31825795644891125,
|
| 140 |
+
"grad_norm": 0.2652761936187744,
|
| 141 |
+
"learning_rate": 2.268e-05,
|
| 142 |
+
"loss": 0.0579,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.33500837520938026,
|
| 147 |
+
"grad_norm": 0.3008716106414795,
|
| 148 |
+
"learning_rate": 2.3880000000000002e-05,
|
| 149 |
+
"loss": 0.0581,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.35175879396984927,
|
| 154 |
+
"grad_norm": 0.25785622000694275,
|
| 155 |
+
"learning_rate": 2.508e-05,
|
| 156 |
+
"loss": 0.0614,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.3685092127303183,
|
| 161 |
+
"grad_norm": 0.26564037799835205,
|
| 162 |
+
"learning_rate": 2.628e-05,
|
| 163 |
+
"loss": 0.0522,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.3852596314907873,
|
| 168 |
+
"grad_norm": 0.2962818741798401,
|
| 169 |
+
"learning_rate": 2.748e-05,
|
| 170 |
+
"loss": 0.0521,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.4020100502512563,
|
| 175 |
+
"grad_norm": 0.28678590059280396,
|
| 176 |
+
"learning_rate": 2.868e-05,
|
| 177 |
+
"loss": 0.0555,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.4187604690117253,
|
| 182 |
+
"grad_norm": 0.29701972007751465,
|
| 183 |
+
"learning_rate": 2.9880000000000002e-05,
|
| 184 |
+
"loss": 0.0517,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.4355108877721943,
|
| 189 |
+
"grad_norm": 0.2258671224117279,
|
| 190 |
+
"learning_rate": 2.9999734259635676e-05,
|
| 191 |
+
"loss": 0.0523,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.45226130653266333,
|
| 196 |
+
"grad_norm": 0.2276943176984787,
|
| 197 |
+
"learning_rate": 2.9998815663057245e-05,
|
| 198 |
+
"loss": 0.0491,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.46901172529313234,
|
| 203 |
+
"grad_norm": 0.367814838886261,
|
| 204 |
+
"learning_rate": 2.999724096969228e-05,
|
| 205 |
+
"loss": 0.0472,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.48576214405360135,
|
| 210 |
+
"grad_norm": 0.19225119054317474,
|
| 211 |
+
"learning_rate": 2.9995010248422984e-05,
|
| 212 |
+
"loss": 0.0452,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.5025125628140703,
|
| 217 |
+
"grad_norm": 0.3076336681842804,
|
| 218 |
+
"learning_rate": 2.999212359682832e-05,
|
| 219 |
+
"loss": 0.0478,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.5192629815745393,
|
| 224 |
+
"grad_norm": 0.2448616325855255,
|
| 225 |
+
"learning_rate": 2.9988581141179787e-05,
|
| 226 |
+
"loss": 0.0471,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.5360134003350083,
|
| 231 |
+
"grad_norm": 0.20978757739067078,
|
| 232 |
+
"learning_rate": 2.998438303643587e-05,
|
| 233 |
+
"loss": 0.0383,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.5527638190954773,
|
| 238 |
+
"grad_norm": 0.24052295088768005,
|
| 239 |
+
"learning_rate": 2.997952946623527e-05,
|
| 240 |
+
"loss": 0.0435,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.5695142378559463,
|
| 245 |
+
"grad_norm": 0.22263383865356445,
|
| 246 |
+
"learning_rate": 2.9974020642888875e-05,
|
| 247 |
+
"loss": 0.0467,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.5862646566164154,
|
| 252 |
+
"grad_norm": 0.25700053572654724,
|
| 253 |
+
"learning_rate": 2.9967856807370458e-05,
|
| 254 |
+
"loss": 0.0448,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.6030150753768844,
|
| 259 |
+
"grad_norm": 0.19276398420333862,
|
| 260 |
+
"learning_rate": 2.996103822930615e-05,
|
| 261 |
+
"loss": 0.0436,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.6197654941373534,
|
| 266 |
+
"grad_norm": 0.19240586459636688,
|
| 267 |
+
"learning_rate": 2.9953565206962653e-05,
|
| 268 |
+
"loss": 0.0406,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.6365159128978225,
|
| 273 |
+
"grad_norm": 0.23333503305912018,
|
| 274 |
+
"learning_rate": 2.9945438067234172e-05,
|
| 275 |
+
"loss": 0.0397,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.6532663316582915,
|
| 280 |
+
"grad_norm": 0.26178690791130066,
|
| 281 |
+
"learning_rate": 2.9936657165628124e-05,
|
| 282 |
+
"loss": 0.0394,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.6700167504187605,
|
| 287 |
+
"grad_norm": 0.16855783760547638,
|
| 288 |
+
"learning_rate": 2.992722288624959e-05,
|
| 289 |
+
"loss": 0.0385,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.6867671691792295,
|
| 294 |
+
"grad_norm": 0.1688416302204132,
|
| 295 |
+
"learning_rate": 2.9917135641784515e-05,
|
| 296 |
+
"loss": 0.0422,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.7035175879396985,
|
| 301 |
+
"grad_norm": 0.18370051681995392,
|
| 302 |
+
"learning_rate": 2.9906395873481655e-05,
|
| 303 |
+
"loss": 0.0371,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.7202680067001676,
|
| 308 |
+
"grad_norm": 0.2349836379289627,
|
| 309 |
+
"learning_rate": 2.989500405113326e-05,
|
| 310 |
+
"loss": 0.0413,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.7370184254606366,
|
| 315 |
+
"grad_norm": 0.20748260617256165,
|
| 316 |
+
"learning_rate": 2.9882960673054547e-05,
|
| 317 |
+
"loss": 0.0367,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.7537688442211056,
|
| 322 |
+
"grad_norm": 0.1702982783317566,
|
| 323 |
+
"learning_rate": 2.9870266266061886e-05,
|
| 324 |
+
"loss": 0.0379,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.7705192629815746,
|
| 329 |
+
"grad_norm": 0.26680412888526917,
|
| 330 |
+
"learning_rate": 2.9856921385449766e-05,
|
| 331 |
+
"loss": 0.0367,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.7872696817420436,
|
| 336 |
+
"grad_norm": 0.21671664714813232,
|
| 337 |
+
"learning_rate": 2.9842926614966502e-05,
|
| 338 |
+
"loss": 0.0367,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.8040201005025126,
|
| 343 |
+
"grad_norm": 0.20597104728221893,
|
| 344 |
+
"learning_rate": 2.982828256678868e-05,
|
| 345 |
+
"loss": 0.0377,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.8207705192629816,
|
| 350 |
+
"grad_norm": 0.25721216201782227,
|
| 351 |
+
"learning_rate": 2.9812989881494412e-05,
|
| 352 |
+
"loss": 0.0376,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.8375209380234506,
|
| 357 |
+
"grad_norm": 0.20260390639305115,
|
| 358 |
+
"learning_rate": 2.9797049228035296e-05,
|
| 359 |
+
"loss": 0.0353,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.8542713567839196,
|
| 364 |
+
"grad_norm": 0.18940532207489014,
|
| 365 |
+
"learning_rate": 2.978046130370715e-05,
|
| 366 |
+
"loss": 0.0326,
|
| 367 |
+
"step": 510
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.8710217755443886,
|
| 371 |
+
"grad_norm": 0.14810627698898315,
|
| 372 |
+
"learning_rate": 2.9763226834119535e-05,
|
| 373 |
+
"loss": 0.0359,
|
| 374 |
+
"step": 520
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.8877721943048577,
|
| 378 |
+
"grad_norm": 0.21682192385196686,
|
| 379 |
+
"learning_rate": 2.9745346573163963e-05,
|
| 380 |
+
"loss": 0.0343,
|
| 381 |
+
"step": 530
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.9045226130653267,
|
| 385 |
+
"grad_norm": 0.2462719827890396,
|
| 386 |
+
"learning_rate": 2.9726821302980987e-05,
|
| 387 |
+
"loss": 0.0344,
|
| 388 |
+
"step": 540
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.9212730318257957,
|
| 392 |
+
"grad_norm": 0.20271016657352448,
|
| 393 |
+
"learning_rate": 2.9707651833925935e-05,
|
| 394 |
+
"loss": 0.0295,
|
| 395 |
+
"step": 550
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.9380234505862647,
|
| 399 |
+
"grad_norm": 0.18994300067424774,
|
| 400 |
+
"learning_rate": 2.968783900453348e-05,
|
| 401 |
+
"loss": 0.0331,
|
| 402 |
+
"step": 560
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.9547738693467337,
|
| 406 |
+
"grad_norm": 0.1817820817232132,
|
| 407 |
+
"learning_rate": 2.9667383681480968e-05,
|
| 408 |
+
"loss": 0.0344,
|
| 409 |
+
"step": 570
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.9715242881072027,
|
| 413 |
+
"grad_norm": 0.20372416079044342,
|
| 414 |
+
"learning_rate": 2.9646286759550493e-05,
|
| 415 |
+
"loss": 0.0333,
|
| 416 |
+
"step": 580
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.9882747068676717,
|
| 420 |
+
"grad_norm": 0.22303369641304016,
|
| 421 |
+
"learning_rate": 2.9624549161589764e-05,
|
| 422 |
+
"loss": 0.0316,
|
| 423 |
+
"step": 590
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 1.0050251256281406,
|
| 427 |
+
"grad_norm": 0.15984657406806946,
|
| 428 |
+
"learning_rate": 2.9602171838471745e-05,
|
| 429 |
+
"loss": 0.0325,
|
| 430 |
+
"step": 600
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 1.0217755443886096,
|
| 434 |
+
"grad_norm": 0.18722732365131378,
|
| 435 |
+
"learning_rate": 2.957915576905303e-05,
|
| 436 |
+
"loss": 0.0304,
|
| 437 |
+
"step": 610
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 1.0385259631490786,
|
| 441 |
+
"grad_norm": 0.18170492351055145,
|
| 442 |
+
"learning_rate": 2.9555501960131058e-05,
|
| 443 |
+
"loss": 0.0314,
|
| 444 |
+
"step": 620
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 1.0552763819095476,
|
| 448 |
+
"grad_norm": 0.19419005513191223,
|
| 449 |
+
"learning_rate": 2.9531211446400068e-05,
|
| 450 |
+
"loss": 0.0336,
|
| 451 |
+
"step": 630
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 1.0720268006700167,
|
| 455 |
+
"grad_norm": 0.21015776693820953,
|
| 456 |
+
"learning_rate": 2.9506285290405813e-05,
|
| 457 |
+
"loss": 0.0312,
|
| 458 |
+
"step": 640
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 1.0887772194304857,
|
| 462 |
+
"grad_norm": 0.13695307075977325,
|
| 463 |
+
"learning_rate": 2.948072458249911e-05,
|
| 464 |
+
"loss": 0.0341,
|
| 465 |
+
"step": 650
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 1.1055276381909547,
|
| 469 |
+
"grad_norm": 0.16071368753910065,
|
| 470 |
+
"learning_rate": 2.9454530440788125e-05,
|
| 471 |
+
"loss": 0.0295,
|
| 472 |
+
"step": 660
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 1.1222780569514237,
|
| 476 |
+
"grad_norm": 0.18829545378684998,
|
| 477 |
+
"learning_rate": 2.9427704011089474e-05,
|
| 478 |
+
"loss": 0.0303,
|
| 479 |
+
"step": 670
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 1.1390284757118927,
|
| 483 |
+
"grad_norm": 0.1615035980939865,
|
| 484 |
+
"learning_rate": 2.94002464668781e-05,
|
| 485 |
+
"loss": 0.0294,
|
| 486 |
+
"step": 680
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 1.1557788944723617,
|
| 490 |
+
"grad_norm": 0.1708756685256958,
|
| 491 |
+
"learning_rate": 2.9372159009235936e-05,
|
| 492 |
+
"loss": 0.0278,
|
| 493 |
+
"step": 690
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 1.1725293132328307,
|
| 497 |
+
"grad_norm": 0.1325443536043167,
|
| 498 |
+
"learning_rate": 2.9343442866799365e-05,
|
| 499 |
+
"loss": 0.0289,
|
| 500 |
+
"step": 700
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 1.1892797319932997,
|
| 504 |
+
"grad_norm": 0.18041647970676422,
|
| 505 |
+
"learning_rate": 2.9314099295705495e-05,
|
| 506 |
+
"loss": 0.0316,
|
| 507 |
+
"step": 710
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 1.2060301507537687,
|
| 511 |
+
"grad_norm": 0.15595586597919464,
|
| 512 |
+
"learning_rate": 2.928412957953718e-05,
|
| 513 |
+
"loss": 0.0294,
|
| 514 |
+
"step": 720
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 1.2227805695142377,
|
| 518 |
+
"grad_norm": 0.18720653653144836,
|
| 519 |
+
"learning_rate": 2.925353502926689e-05,
|
| 520 |
+
"loss": 0.0292,
|
| 521 |
+
"step": 730
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 1.2395309882747068,
|
| 525 |
+
"grad_norm": 0.1344250738620758,
|
| 526 |
+
"learning_rate": 2.922231698319937e-05,
|
| 527 |
+
"loss": 0.0299,
|
| 528 |
+
"step": 740
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 1.2562814070351758,
|
| 532 |
+
"grad_norm": 0.18409386277198792,
|
| 533 |
+
"learning_rate": 2.9190476806913088e-05,
|
| 534 |
+
"loss": 0.0305,
|
| 535 |
+
"step": 750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 1.2730318257956448,
|
| 539 |
+
"grad_norm": 0.16554060578346252,
|
| 540 |
+
"learning_rate": 2.9158015893200504e-05,
|
| 541 |
+
"loss": 0.0292,
|
| 542 |
+
"step": 760
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 1.2897822445561138,
|
| 546 |
+
"grad_norm": 0.1602775752544403,
|
| 547 |
+
"learning_rate": 2.9124935662007137e-05,
|
| 548 |
+
"loss": 0.03,
|
| 549 |
+
"step": 770
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.3065326633165828,
|
| 553 |
+
"grad_norm": 0.1876162886619568,
|
| 554 |
+
"learning_rate": 2.9091237560369467e-05,
|
| 555 |
+
"loss": 0.0284,
|
| 556 |
+
"step": 780
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 1.3232830820770518,
|
| 560 |
+
"grad_norm": 0.24121980369091034,
|
| 561 |
+
"learning_rate": 2.905692306235163e-05,
|
| 562 |
+
"loss": 0.0282,
|
| 563 |
+
"step": 790
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 1.3400335008375208,
|
| 567 |
+
"grad_norm": 0.15151315927505493,
|
| 568 |
+
"learning_rate": 2.9021993668980925e-05,
|
| 569 |
+
"loss": 0.0288,
|
| 570 |
+
"step": 800
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 1.3567839195979898,
|
| 574 |
+
"grad_norm": 0.32094353437423706,
|
| 575 |
+
"learning_rate": 2.8986450908182185e-05,
|
| 576 |
+
"loss": 0.0265,
|
| 577 |
+
"step": 810
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 1.3735343383584588,
|
| 581 |
+
"grad_norm": 0.20612798631191254,
|
| 582 |
+
"learning_rate": 2.8950296334710902e-05,
|
| 583 |
+
"loss": 0.0281,
|
| 584 |
+
"step": 820
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 1.3902847571189278,
|
| 588 |
+
"grad_norm": 0.16587623953819275,
|
| 589 |
+
"learning_rate": 2.8913531530085248e-05,
|
| 590 |
+
"loss": 0.0272,
|
| 591 |
+
"step": 830
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 1.4070351758793969,
|
| 595 |
+
"grad_norm": 0.1658754050731659,
|
| 596 |
+
"learning_rate": 2.887615810251687e-05,
|
| 597 |
+
"loss": 0.0279,
|
| 598 |
+
"step": 840
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 1.4237855946398659,
|
| 602 |
+
"grad_norm": 0.155488982796669,
|
| 603 |
+
"learning_rate": 2.8838177686840577e-05,
|
| 604 |
+
"loss": 0.0266,
|
| 605 |
+
"step": 850
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 1.4405360134003349,
|
| 609 |
+
"grad_norm": 0.2229546457529068,
|
| 610 |
+
"learning_rate": 2.8799591944442776e-05,
|
| 611 |
+
"loss": 0.0292,
|
| 612 |
+
"step": 860
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 1.457286432160804,
|
| 616 |
+
"grad_norm": 0.36190536618232727,
|
| 617 |
+
"learning_rate": 2.8760402563188848e-05,
|
| 618 |
+
"loss": 0.0277,
|
| 619 |
+
"step": 870
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 1.474036850921273,
|
| 623 |
+
"grad_norm": 0.18090955913066864,
|
| 624 |
+
"learning_rate": 2.8720611257349285e-05,
|
| 625 |
+
"loss": 0.0297,
|
| 626 |
+
"step": 880
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 1.490787269681742,
|
| 630 |
+
"grad_norm": 0.17235404253005981,
|
| 631 |
+
"learning_rate": 2.868021976752471e-05,
|
| 632 |
+
"loss": 0.0265,
|
| 633 |
+
"step": 890
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 1.507537688442211,
|
| 637 |
+
"grad_norm": 0.1455887258052826,
|
| 638 |
+
"learning_rate": 2.8639229860569738e-05,
|
| 639 |
+
"loss": 0.0298,
|
| 640 |
+
"step": 900
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 1.52428810720268,
|
| 644 |
+
"grad_norm": 0.1774197816848755,
|
| 645 |
+
"learning_rate": 2.8597643329515688e-05,
|
| 646 |
+
"loss": 0.0263,
|
| 647 |
+
"step": 910
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 1.541038525963149,
|
| 651 |
+
"grad_norm": 0.15991894900798798,
|
| 652 |
+
"learning_rate": 2.855546199349214e-05,
|
| 653 |
+
"loss": 0.0259,
|
| 654 |
+
"step": 920
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 1.557788944723618,
|
| 658 |
+
"grad_norm": 0.30203819274902344,
|
| 659 |
+
"learning_rate": 2.85126876976474e-05,
|
| 660 |
+
"loss": 0.0276,
|
| 661 |
+
"step": 930
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 1.574539363484087,
|
| 665 |
+
"grad_norm": 0.13825957477092743,
|
| 666 |
+
"learning_rate": 2.8469322313067726e-05,
|
| 667 |
+
"loss": 0.0262,
|
| 668 |
+
"step": 940
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 1.591289782244556,
|
| 672 |
+
"grad_norm": 0.19100086390972137,
|
| 673 |
+
"learning_rate": 2.8425367736695522e-05,
|
| 674 |
+
"loss": 0.0257,
|
| 675 |
+
"step": 950
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 1.608040201005025,
|
| 679 |
+
"grad_norm": 0.16253994405269623,
|
| 680 |
+
"learning_rate": 2.838082589124636e-05,
|
| 681 |
+
"loss": 0.0282,
|
| 682 |
+
"step": 960
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 1.624790619765494,
|
| 686 |
+
"grad_norm": 0.21166931092739105,
|
| 687 |
+
"learning_rate": 2.8335698725124843e-05,
|
| 688 |
+
"loss": 0.0278,
|
| 689 |
+
"step": 970
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 1.641541038525963,
|
| 693 |
+
"grad_norm": 0.13361522555351257,
|
| 694 |
+
"learning_rate": 2.8289988212339422e-05,
|
| 695 |
+
"loss": 0.0254,
|
| 696 |
+
"step": 980
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 1.658291457286432,
|
| 700 |
+
"grad_norm": 0.20177683234214783,
|
| 701 |
+
"learning_rate": 2.8243696352415997e-05,
|
| 702 |
+
"loss": 0.0317,
|
| 703 |
+
"step": 990
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 1.675041876046901,
|
| 707 |
+
"grad_norm": 0.17741645872592926,
|
| 708 |
+
"learning_rate": 2.8196825170310495e-05,
|
| 709 |
+
"loss": 0.0254,
|
| 710 |
+
"step": 1000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 1.69179229480737,
|
| 714 |
+
"grad_norm": 0.1586075723171234,
|
| 715 |
+
"learning_rate": 2.8149376716320256e-05,
|
| 716 |
+
"loss": 0.0264,
|
| 717 |
+
"step": 1010
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 1.708542713567839,
|
| 721 |
+
"grad_norm": 0.18498659133911133,
|
| 722 |
+
"learning_rate": 2.8101353065994373e-05,
|
| 723 |
+
"loss": 0.0289,
|
| 724 |
+
"step": 1020
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 1.725293132328308,
|
| 728 |
+
"grad_norm": 0.16134566068649292,
|
| 729 |
+
"learning_rate": 2.8052756320042885e-05,
|
| 730 |
+
"loss": 0.0262,
|
| 731 |
+
"step": 1030
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 1.742043551088777,
|
| 735 |
+
"grad_norm": 0.18623295426368713,
|
| 736 |
+
"learning_rate": 2.8003588604244893e-05,
|
| 737 |
+
"loss": 0.0242,
|
| 738 |
+
"step": 1040
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 1.758793969849246,
|
| 742 |
+
"grad_norm": 0.1394229531288147,
|
| 743 |
+
"learning_rate": 2.7953852069355565e-05,
|
| 744 |
+
"loss": 0.0236,
|
| 745 |
+
"step": 1050
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 1.775544388609715,
|
| 749 |
+
"grad_norm": 0.15025576949119568,
|
| 750 |
+
"learning_rate": 2.790354889101206e-05,
|
| 751 |
+
"loss": 0.0222,
|
| 752 |
+
"step": 1060
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 1.792294807370184,
|
| 756 |
+
"grad_norm": 0.13321664929389954,
|
| 757 |
+
"learning_rate": 2.7852681269638366e-05,
|
| 758 |
+
"loss": 0.0259,
|
| 759 |
+
"step": 1070
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 1.809045226130653,
|
| 763 |
+
"grad_norm": 0.16647320985794067,
|
| 764 |
+
"learning_rate": 2.780125143034902e-05,
|
| 765 |
+
"loss": 0.0268,
|
| 766 |
+
"step": 1080
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 1.8257956448911221,
|
| 770 |
+
"grad_norm": 0.21666789054870605,
|
| 771 |
+
"learning_rate": 2.7749261622851812e-05,
|
| 772 |
+
"loss": 0.0242,
|
| 773 |
+
"step": 1090
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.8425460636515911,
|
| 777 |
+
"grad_norm": 0.18027998507022858,
|
| 778 |
+
"learning_rate": 2.7696714121349335e-05,
|
| 779 |
+
"loss": 0.0253,
|
| 780 |
+
"step": 1100
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.8592964824120601,
|
| 784 |
+
"grad_norm": 0.17092131078243256,
|
| 785 |
+
"learning_rate": 2.764361122443954e-05,
|
| 786 |
+
"loss": 0.0236,
|
| 787 |
+
"step": 1110
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.8760469011725294,
|
| 791 |
+
"grad_norm": 0.1665954291820526,
|
| 792 |
+
"learning_rate": 2.758995525501517e-05,
|
| 793 |
+
"loss": 0.0226,
|
| 794 |
+
"step": 1120
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.8927973199329984,
|
| 798 |
+
"grad_norm": 0.1582082360982895,
|
| 799 |
+
"learning_rate": 2.753574856016215e-05,
|
| 800 |
+
"loss": 0.0257,
|
| 801 |
+
"step": 1130
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 1.9095477386934674,
|
| 805 |
+
"grad_norm": 0.18370600044727325,
|
| 806 |
+
"learning_rate": 2.7480993511056913e-05,
|
| 807 |
+
"loss": 0.0252,
|
| 808 |
+
"step": 1140
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 1.9262981574539364,
|
| 812 |
+
"grad_norm": 0.14998869597911835,
|
| 813 |
+
"learning_rate": 2.7425692502862684e-05,
|
| 814 |
+
"loss": 0.022,
|
| 815 |
+
"step": 1150
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 1.9430485762144054,
|
| 819 |
+
"grad_norm": 0.13657869398593903,
|
| 820 |
+
"learning_rate": 2.7369847954624728e-05,
|
| 821 |
+
"loss": 0.0248,
|
| 822 |
+
"step": 1160
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 1.9597989949748744,
|
| 826 |
+
"grad_norm": 0.15892188251018524,
|
| 827 |
+
"learning_rate": 2.731346230916448e-05,
|
| 828 |
+
"loss": 0.026,
|
| 829 |
+
"step": 1170
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 1.9765494137353434,
|
| 833 |
+
"grad_norm": 0.15034610033035278,
|
| 834 |
+
"learning_rate": 2.7256538032972746e-05,
|
| 835 |
+
"loss": 0.0232,
|
| 836 |
+
"step": 1180
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 1.9932998324958124,
|
| 840 |
+
"grad_norm": 0.1602204591035843,
|
| 841 |
+
"learning_rate": 2.7199077616101773e-05,
|
| 842 |
+
"loss": 0.0231,
|
| 843 |
+
"step": 1190
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 2.0100502512562812,
|
| 847 |
+
"grad_norm": 0.12914741039276123,
|
| 848 |
+
"learning_rate": 2.714108357205634e-05,
|
| 849 |
+
"loss": 0.0239,
|
| 850 |
+
"step": 1200
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 2.0268006700167502,
|
| 854 |
+
"grad_norm": 0.1998392939567566,
|
| 855 |
+
"learning_rate": 2.7082558437683803e-05,
|
| 856 |
+
"loss": 0.0224,
|
| 857 |
+
"step": 1210
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 2.0435510887772192,
|
| 861 |
+
"grad_norm": 0.18535244464874268,
|
| 862 |
+
"learning_rate": 2.702350477306315e-05,
|
| 863 |
+
"loss": 0.0233,
|
| 864 |
+
"step": 1220
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 2.0603015075376883,
|
| 868 |
+
"grad_norm": 0.14260677993297577,
|
| 869 |
+
"learning_rate": 2.6963925161392963e-05,
|
| 870 |
+
"loss": 0.0245,
|
| 871 |
+
"step": 1230
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 2.0770519262981573,
|
| 875 |
+
"grad_norm": 0.201503187417984,
|
| 876 |
+
"learning_rate": 2.6903822208878476e-05,
|
| 877 |
+
"loss": 0.0235,
|
| 878 |
+
"step": 1240
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 2.0938023450586263,
|
| 882 |
+
"grad_norm": 0.16170263290405273,
|
| 883 |
+
"learning_rate": 2.6843198544617535e-05,
|
| 884 |
+
"loss": 0.0242,
|
| 885 |
+
"step": 1250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 2.1105527638190953,
|
| 889 |
+
"grad_norm": 0.14339309930801392,
|
| 890 |
+
"learning_rate": 2.6782056820485612e-05,
|
| 891 |
+
"loss": 0.0229,
|
| 892 |
+
"step": 1260
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 2.1273031825795643,
|
| 896 |
+
"grad_norm": 0.23185782134532928,
|
| 897 |
+
"learning_rate": 2.672039971101979e-05,
|
| 898 |
+
"loss": 0.0229,
|
| 899 |
+
"step": 1270
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 2.1440536013400333,
|
| 903 |
+
"grad_norm": 0.2018624097108841,
|
| 904 |
+
"learning_rate": 2.665822991330178e-05,
|
| 905 |
+
"loss": 0.0263,
|
| 906 |
+
"step": 1280
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 2.1608040201005023,
|
| 910 |
+
"grad_norm": 0.22470778226852417,
|
| 911 |
+
"learning_rate": 2.6595550146839934e-05,
|
| 912 |
+
"loss": 0.0244,
|
| 913 |
+
"step": 1290
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 2.1775544388609713,
|
| 917 |
+
"grad_norm": 0.1919565051794052,
|
| 918 |
+
"learning_rate": 2.6532363153450287e-05,
|
| 919 |
+
"loss": 0.023,
|
| 920 |
+
"step": 1300
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 2.1943048576214403,
|
| 924 |
+
"grad_norm": 0.16625232994556427,
|
| 925 |
+
"learning_rate": 2.6468671697136634e-05,
|
| 926 |
+
"loss": 0.022,
|
| 927 |
+
"step": 1310
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 2.2110552763819094,
|
| 931 |
+
"grad_norm": 0.16590924561023712,
|
| 932 |
+
"learning_rate": 2.640447856396961e-05,
|
| 933 |
+
"loss": 0.0211,
|
| 934 |
+
"step": 1320
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 2.2278056951423784,
|
| 938 |
+
"grad_norm": 0.1491633653640747,
|
| 939 |
+
"learning_rate": 2.6339786561964805e-05,
|
| 940 |
+
"loss": 0.0265,
|
| 941 |
+
"step": 1330
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 2.2445561139028474,
|
| 945 |
+
"grad_norm": 0.16424474120140076,
|
| 946 |
+
"learning_rate": 2.6274598520959966e-05,
|
| 947 |
+
"loss": 0.0235,
|
| 948 |
+
"step": 1340
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 2.2613065326633164,
|
| 952 |
+
"grad_norm": 0.18205176293849945,
|
| 953 |
+
"learning_rate": 2.6208917292491185e-05,
|
| 954 |
+
"loss": 0.0242,
|
| 955 |
+
"step": 1350
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 2.2780569514237854,
|
| 959 |
+
"grad_norm": 0.1608763039112091,
|
| 960 |
+
"learning_rate": 2.6142745749668176e-05,
|
| 961 |
+
"loss": 0.0225,
|
| 962 |
+
"step": 1360
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 2.2948073701842544,
|
| 966 |
+
"grad_norm": 0.1522231101989746,
|
| 967 |
+
"learning_rate": 2.607608678704859e-05,
|
| 968 |
+
"loss": 0.0229,
|
| 969 |
+
"step": 1370
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 2.3115577889447234,
|
| 973 |
+
"grad_norm": 0.16214381158351898,
|
| 974 |
+
"learning_rate": 2.600894332051139e-05,
|
| 975 |
+
"loss": 0.0201,
|
| 976 |
+
"step": 1380
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 2.3283082077051924,
|
| 980 |
+
"grad_norm": 0.1418379843235016,
|
| 981 |
+
"learning_rate": 2.594131828712932e-05,
|
| 982 |
+
"loss": 0.0232,
|
| 983 |
+
"step": 1390
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 2.3450586264656614,
|
| 987 |
+
"grad_norm": 0.1545649766921997,
|
| 988 |
+
"learning_rate": 2.5873214645040414e-05,
|
| 989 |
+
"loss": 0.0224,
|
| 990 |
+
"step": 1400
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 2.3618090452261304,
|
| 994 |
+
"grad_norm": 0.16949930787086487,
|
| 995 |
+
"learning_rate": 2.5804635373318604e-05,
|
| 996 |
+
"loss": 0.0209,
|
| 997 |
+
"step": 1410
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 2.3785594639865995,
|
| 1001 |
+
"grad_norm": 0.1950673907995224,
|
| 1002 |
+
"learning_rate": 2.57355834718434e-05,
|
| 1003 |
+
"loss": 0.0235,
|
| 1004 |
+
"step": 1420
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 2.3953098827470685,
|
| 1008 |
+
"grad_norm": 0.15320082008838654,
|
| 1009 |
+
"learning_rate": 2.566606196116867e-05,
|
| 1010 |
+
"loss": 0.0209,
|
| 1011 |
+
"step": 1430
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 2.4120603015075375,
|
| 1015 |
+
"grad_norm": 0.14742597937583923,
|
| 1016 |
+
"learning_rate": 2.559607388239051e-05,
|
| 1017 |
+
"loss": 0.0206,
|
| 1018 |
+
"step": 1440
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 2.4288107202680065,
|
| 1022 |
+
"grad_norm": 0.15421146154403687,
|
| 1023 |
+
"learning_rate": 2.552562229701422e-05,
|
| 1024 |
+
"loss": 0.0228,
|
| 1025 |
+
"step": 1450
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 2.4455611390284755,
|
| 1029 |
+
"grad_norm": 0.16883830726146698,
|
| 1030 |
+
"learning_rate": 2.545471028682036e-05,
|
| 1031 |
+
"loss": 0.0252,
|
| 1032 |
+
"step": 1460
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 2.4623115577889445,
|
| 1036 |
+
"grad_norm": 0.143342524766922,
|
| 1037 |
+
"learning_rate": 2.538334095373e-05,
|
| 1038 |
+
"loss": 0.0207,
|
| 1039 |
+
"step": 1470
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 2.4790619765494135,
|
| 1043 |
+
"grad_norm": 0.17817887663841248,
|
| 1044 |
+
"learning_rate": 2.531151741966896e-05,
|
| 1045 |
+
"loss": 0.0212,
|
| 1046 |
+
"step": 1480
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 2.4958123953098825,
|
| 1050 |
+
"grad_norm": 0.1521480828523636,
|
| 1051 |
+
"learning_rate": 2.5239242826431302e-05,
|
| 1052 |
+
"loss": 0.0208,
|
| 1053 |
+
"step": 1490
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 2.5125628140703515,
|
| 1057 |
+
"grad_norm": 0.14518405497074127,
|
| 1058 |
+
"learning_rate": 2.5166520335541865e-05,
|
| 1059 |
+
"loss": 0.0204,
|
| 1060 |
+
"step": 1500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 2.5293132328308205,
|
| 1064 |
+
"grad_norm": 0.11578862369060516,
|
| 1065 |
+
"learning_rate": 2.5093353128117996e-05,
|
| 1066 |
+
"loss": 0.0237,
|
| 1067 |
+
"step": 1510
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 2.5460636515912896,
|
| 1071 |
+
"grad_norm": 0.1930747777223587,
|
| 1072 |
+
"learning_rate": 2.501974440473037e-05,
|
| 1073 |
+
"loss": 0.0205,
|
| 1074 |
+
"step": 1520
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 2.5628140703517586,
|
| 1078 |
+
"grad_norm": 0.1401190161705017,
|
| 1079 |
+
"learning_rate": 2.4945697385263015e-05,
|
| 1080 |
+
"loss": 0.0217,
|
| 1081 |
+
"step": 1530
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 2.5795644891122276,
|
| 1085 |
+
"grad_norm": 0.1486596018075943,
|
| 1086 |
+
"learning_rate": 2.487121530877243e-05,
|
| 1087 |
+
"loss": 0.0242,
|
| 1088 |
+
"step": 1540
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 2.5963149078726966,
|
| 1092 |
+
"grad_norm": 0.16599705815315247,
|
| 1093 |
+
"learning_rate": 2.4796301433345943e-05,
|
| 1094 |
+
"loss": 0.0219,
|
| 1095 |
+
"step": 1550
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 2.6130653266331656,
|
| 1099 |
+
"grad_norm": 0.1955460160970688,
|
| 1100 |
+
"learning_rate": 2.4720959035959148e-05,
|
| 1101 |
+
"loss": 0.022,
|
| 1102 |
+
"step": 1560
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 2.6298157453936346,
|
| 1106 |
+
"grad_norm": 0.16018950939178467,
|
| 1107 |
+
"learning_rate": 2.464519141233257e-05,
|
| 1108 |
+
"loss": 0.0219,
|
| 1109 |
+
"step": 1570
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 2.6465661641541036,
|
| 1113 |
+
"grad_norm": 0.15230196714401245,
|
| 1114 |
+
"learning_rate": 2.456900187678754e-05,
|
| 1115 |
+
"loss": 0.0211,
|
| 1116 |
+
"step": 1580
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 2.6633165829145726,
|
| 1120 |
+
"grad_norm": 0.1539147049188614,
|
| 1121 |
+
"learning_rate": 2.4492393762101147e-05,
|
| 1122 |
+
"loss": 0.0224,
|
| 1123 |
+
"step": 1590
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 2.6800670016750416,
|
| 1127 |
+
"grad_norm": 0.2199530005455017,
|
| 1128 |
+
"learning_rate": 2.4415370419360508e-05,
|
| 1129 |
+
"loss": 0.0216,
|
| 1130 |
+
"step": 1600
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 2.6968174204355106,
|
| 1134 |
+
"grad_norm": 0.11828820407390594,
|
| 1135 |
+
"learning_rate": 2.4337935217816157e-05,
|
| 1136 |
+
"loss": 0.0191,
|
| 1137 |
+
"step": 1610
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 2.7135678391959797,
|
| 1141 |
+
"grad_norm": 0.11301644891500473,
|
| 1142 |
+
"learning_rate": 2.4260091544734665e-05,
|
| 1143 |
+
"loss": 0.0209,
|
| 1144 |
+
"step": 1620
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 2.7303182579564487,
|
| 1148 |
+
"grad_norm": 0.13908027112483978,
|
| 1149 |
+
"learning_rate": 2.4181842805250466e-05,
|
| 1150 |
+
"loss": 0.0219,
|
| 1151 |
+
"step": 1630
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 2.7470686767169177,
|
| 1155 |
+
"grad_norm": 0.14333346486091614,
|
| 1156 |
+
"learning_rate": 2.4103192422216923e-05,
|
| 1157 |
+
"loss": 0.0179,
|
| 1158 |
+
"step": 1640
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 2.7638190954773867,
|
| 1162 |
+
"grad_norm": 0.15043054521083832,
|
| 1163 |
+
"learning_rate": 2.402414383605658e-05,
|
| 1164 |
+
"loss": 0.0209,
|
| 1165 |
+
"step": 1650
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 2.7805695142378557,
|
| 1169 |
+
"grad_norm": 0.1858556866645813,
|
| 1170 |
+
"learning_rate": 2.3944700504610674e-05,
|
| 1171 |
+
"loss": 0.0185,
|
| 1172 |
+
"step": 1660
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 2.7973199329983247,
|
| 1176 |
+
"grad_norm": 0.13898970186710358,
|
| 1177 |
+
"learning_rate": 2.3864865902987883e-05,
|
| 1178 |
+
"loss": 0.0206,
|
| 1179 |
+
"step": 1670
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 2.8140703517587937,
|
| 1183 |
+
"grad_norm": 0.17943225800991058,
|
| 1184 |
+
"learning_rate": 2.378464352341231e-05,
|
| 1185 |
+
"loss": 0.0212,
|
| 1186 |
+
"step": 1680
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 2.8308207705192627,
|
| 1190 |
+
"grad_norm": 0.1789410412311554,
|
| 1191 |
+
"learning_rate": 2.3704036875070728e-05,
|
| 1192 |
+
"loss": 0.0234,
|
| 1193 |
+
"step": 1690
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 2.8475711892797317,
|
| 1197 |
+
"grad_norm": 0.16943080723285675,
|
| 1198 |
+
"learning_rate": 2.362304948395906e-05,
|
| 1199 |
+
"loss": 0.0238,
|
| 1200 |
+
"step": 1700
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 2.8643216080402008,
|
| 1204 |
+
"grad_norm": 0.182373046875,
|
| 1205 |
+
"learning_rate": 2.3541684892728157e-05,
|
| 1206 |
+
"loss": 0.022,
|
| 1207 |
+
"step": 1710
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 2.8810720268006698,
|
| 1211 |
+
"grad_norm": 0.1499493569135666,
|
| 1212 |
+
"learning_rate": 2.345994666052883e-05,
|
| 1213 |
+
"loss": 0.02,
|
| 1214 |
+
"step": 1720
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 2.8978224455611388,
|
| 1218 |
+
"grad_norm": 0.17924560606479645,
|
| 1219 |
+
"learning_rate": 2.337783836285614e-05,
|
| 1220 |
+
"loss": 0.0208,
|
| 1221 |
+
"step": 1730
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 2.914572864321608,
|
| 1225 |
+
"grad_norm": 0.17541290819644928,
|
| 1226 |
+
"learning_rate": 2.3295363591393026e-05,
|
| 1227 |
+
"loss": 0.0183,
|
| 1228 |
+
"step": 1740
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 2.931323283082077,
|
| 1232 |
+
"grad_norm": 0.14497047662734985,
|
| 1233 |
+
"learning_rate": 2.3212525953853174e-05,
|
| 1234 |
+
"loss": 0.0219,
|
| 1235 |
+
"step": 1750
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 2.948073701842546,
|
| 1239 |
+
"grad_norm": 0.1496923714876175,
|
| 1240 |
+
"learning_rate": 2.3129329073823206e-05,
|
| 1241 |
+
"loss": 0.0196,
|
| 1242 |
+
"step": 1760
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 2.964824120603015,
|
| 1246 |
+
"grad_norm": 0.18601380288600922,
|
| 1247 |
+
"learning_rate": 2.3045776590604175e-05,
|
| 1248 |
+
"loss": 0.0228,
|
| 1249 |
+
"step": 1770
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 2.981574539363484,
|
| 1253 |
+
"grad_norm": 0.16404607892036438,
|
| 1254 |
+
"learning_rate": 2.296187215905237e-05,
|
| 1255 |
+
"loss": 0.0201,
|
| 1256 |
+
"step": 1780
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 2.998324958123953,
|
| 1260 |
+
"grad_norm": 0.16068114340305328,
|
| 1261 |
+
"learning_rate": 2.2877619449419437e-05,
|
| 1262 |
+
"loss": 0.0222,
|
| 1263 |
+
"step": 1790
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 3.0150753768844223,
|
| 1267 |
+
"grad_norm": 0.22023139894008636,
|
| 1268 |
+
"learning_rate": 2.2793022147191843e-05,
|
| 1269 |
+
"loss": 0.0196,
|
| 1270 |
+
"step": 1800
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 3.0318257956448913,
|
| 1274 |
+
"grad_norm": 0.22863224148750305,
|
| 1275 |
+
"learning_rate": 2.2708083952929634e-05,
|
| 1276 |
+
"loss": 0.0175,
|
| 1277 |
+
"step": 1810
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 3.0485762144053603,
|
| 1281 |
+
"grad_norm": 0.1506689339876175,
|
| 1282 |
+
"learning_rate": 2.2622808582104594e-05,
|
| 1283 |
+
"loss": 0.0195,
|
| 1284 |
+
"step": 1820
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 3.0653266331658293,
|
| 1288 |
+
"grad_norm": 0.12392555922269821,
|
| 1289 |
+
"learning_rate": 2.2537199764937703e-05,
|
| 1290 |
+
"loss": 0.0203,
|
| 1291 |
+
"step": 1830
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 3.0820770519262983,
|
| 1295 |
+
"grad_norm": 0.13800229132175446,
|
| 1296 |
+
"learning_rate": 2.245126124623595e-05,
|
| 1297 |
+
"loss": 0.0209,
|
| 1298 |
+
"step": 1840
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 3.0988274706867673,
|
| 1302 |
+
"grad_norm": 0.12618660926818848,
|
| 1303 |
+
"learning_rate": 2.2364996785228547e-05,
|
| 1304 |
+
"loss": 0.0206,
|
| 1305 |
+
"step": 1850
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 3.1155778894472363,
|
| 1309 |
+
"grad_norm": 0.1612847000360489,
|
| 1310 |
+
"learning_rate": 2.227841015540248e-05,
|
| 1311 |
+
"loss": 0.0212,
|
| 1312 |
+
"step": 1860
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 3.1323283082077054,
|
| 1316 |
+
"grad_norm": 0.11662063002586365,
|
| 1317 |
+
"learning_rate": 2.2191505144337433e-05,
|
| 1318 |
+
"loss": 0.0208,
|
| 1319 |
+
"step": 1870
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 3.1490787269681744,
|
| 1323 |
+
"grad_norm": 0.14581643044948578,
|
| 1324 |
+
"learning_rate": 2.210428555354013e-05,
|
| 1325 |
+
"loss": 0.0202,
|
| 1326 |
+
"step": 1880
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 3.1658291457286434,
|
| 1330 |
+
"grad_norm": 0.16268415749073029,
|
| 1331 |
+
"learning_rate": 2.201675519827802e-05,
|
| 1332 |
+
"loss": 0.0205,
|
| 1333 |
+
"step": 1890
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 3.1825795644891124,
|
| 1337 |
+
"grad_norm": 0.1655290424823761,
|
| 1338 |
+
"learning_rate": 2.1928917907412403e-05,
|
| 1339 |
+
"loss": 0.0188,
|
| 1340 |
+
"step": 1900
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 3.1993299832495814,
|
| 1344 |
+
"grad_norm": 0.1120857372879982,
|
| 1345 |
+
"learning_rate": 2.1840777523230937e-05,
|
| 1346 |
+
"loss": 0.0202,
|
| 1347 |
+
"step": 1910
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 3.2160804020100504,
|
| 1351 |
+
"grad_norm": 0.12592901289463043,
|
| 1352 |
+
"learning_rate": 2.1752337901279552e-05,
|
| 1353 |
+
"loss": 0.0192,
|
| 1354 |
+
"step": 1920
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 3.2328308207705194,
|
| 1358 |
+
"grad_norm": 0.22475941479206085,
|
| 1359 |
+
"learning_rate": 2.166360291019383e-05,
|
| 1360 |
+
"loss": 0.0198,
|
| 1361 |
+
"step": 1930
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 3.2495812395309884,
|
| 1365 |
+
"grad_norm": 0.11940304934978485,
|
| 1366 |
+
"learning_rate": 2.157457643152973e-05,
|
| 1367 |
+
"loss": 0.019,
|
| 1368 |
+
"step": 1940
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 3.2663316582914574,
|
| 1372 |
+
"grad_norm": 0.14816822111606598,
|
| 1373 |
+
"learning_rate": 2.148526235959385e-05,
|
| 1374 |
+
"loss": 0.0186,
|
| 1375 |
+
"step": 1950
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 3.2830820770519265,
|
| 1379 |
+
"grad_norm": 0.1205066591501236,
|
| 1380 |
+
"learning_rate": 2.1395664601273016e-05,
|
| 1381 |
+
"loss": 0.02,
|
| 1382 |
+
"step": 1960
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 3.2998324958123955,
|
| 1386 |
+
"grad_norm": 0.13585013151168823,
|
| 1387 |
+
"learning_rate": 2.130578707586345e-05,
|
| 1388 |
+
"loss": 0.0194,
|
| 1389 |
+
"step": 1970
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 3.3165829145728645,
|
| 1393 |
+
"grad_norm": 0.13307107985019684,
|
| 1394 |
+
"learning_rate": 2.1215633714899264e-05,
|
| 1395 |
+
"loss": 0.0205,
|
| 1396 |
+
"step": 1980
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 3.3333333333333335,
|
| 1400 |
+
"grad_norm": 0.1215062215924263,
|
| 1401 |
+
"learning_rate": 2.112520846198053e-05,
|
| 1402 |
+
"loss": 0.0202,
|
| 1403 |
+
"step": 1990
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 3.3500837520938025,
|
| 1407 |
+
"grad_norm": 0.14667080342769623,
|
| 1408 |
+
"learning_rate": 2.103451527260074e-05,
|
| 1409 |
+
"loss": 0.0205,
|
| 1410 |
+
"step": 2000
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 3.3668341708542715,
|
| 1414 |
+
"grad_norm": 0.19008496403694153,
|
| 1415 |
+
"learning_rate": 2.094355811397381e-05,
|
| 1416 |
+
"loss": 0.0196,
|
| 1417 |
+
"step": 2010
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 3.3835845896147405,
|
| 1421 |
+
"grad_norm": 0.1476626843214035,
|
| 1422 |
+
"learning_rate": 2.0852340964860522e-05,
|
| 1423 |
+
"loss": 0.0227,
|
| 1424 |
+
"step": 2020
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 3.4003350083752095,
|
| 1428 |
+
"grad_norm": 0.17680875957012177,
|
| 1429 |
+
"learning_rate": 2.076086781539447e-05,
|
| 1430 |
+
"loss": 0.0193,
|
| 1431 |
+
"step": 2030
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 3.4170854271356785,
|
| 1435 |
+
"grad_norm": 0.1655515730381012,
|
| 1436 |
+
"learning_rate": 2.066914266690755e-05,
|
| 1437 |
+
"loss": 0.0191,
|
| 1438 |
+
"step": 2040
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 3.4338358458961475,
|
| 1442 |
+
"grad_norm": 0.18550711870193481,
|
| 1443 |
+
"learning_rate": 2.0577169531754906e-05,
|
| 1444 |
+
"loss": 0.0198,
|
| 1445 |
+
"step": 2050
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 3.4505862646566166,
|
| 1449 |
+
"grad_norm": 0.11140353232622147,
|
| 1450 |
+
"learning_rate": 2.048495243313942e-05,
|
| 1451 |
+
"loss": 0.0186,
|
| 1452 |
+
"step": 2060
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 3.4673366834170856,
|
| 1456 |
+
"grad_norm": 0.15830710530281067,
|
| 1457 |
+
"learning_rate": 2.039249540493572e-05,
|
| 1458 |
+
"loss": 0.0208,
|
| 1459 |
+
"step": 2070
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 3.4840871021775546,
|
| 1463 |
+
"grad_norm": 0.15438948571681976,
|
| 1464 |
+
"learning_rate": 2.0299802491513755e-05,
|
| 1465 |
+
"loss": 0.017,
|
| 1466 |
+
"step": 2080
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 3.5008375209380236,
|
| 1470 |
+
"grad_norm": 0.17817939817905426,
|
| 1471 |
+
"learning_rate": 2.0206877747561828e-05,
|
| 1472 |
+
"loss": 0.0203,
|
| 1473 |
+
"step": 2090
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 3.5175879396984926,
|
| 1477 |
+
"grad_norm": 0.1366434544324875,
|
| 1478 |
+
"learning_rate": 2.011372523790927e-05,
|
| 1479 |
+
"loss": 0.0212,
|
| 1480 |
+
"step": 2100
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 3.5343383584589616,
|
| 1484 |
+
"grad_norm": 0.13273486495018005,
|
| 1485 |
+
"learning_rate": 2.0020349037348643e-05,
|
| 1486 |
+
"loss": 0.016,
|
| 1487 |
+
"step": 2110
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 3.5510887772194306,
|
| 1491 |
+
"grad_norm": 0.15586011111736298,
|
| 1492 |
+
"learning_rate": 1.9926753230457443e-05,
|
| 1493 |
+
"loss": 0.0192,
|
| 1494 |
+
"step": 2120
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 3.5678391959798996,
|
| 1498 |
+
"grad_norm": 0.10179942101240158,
|
| 1499 |
+
"learning_rate": 1.983294191141948e-05,
|
| 1500 |
+
"loss": 0.0194,
|
| 1501 |
+
"step": 2130
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 3.5845896147403686,
|
| 1505 |
+
"grad_norm": 0.18417240679264069,
|
| 1506 |
+
"learning_rate": 1.973891918384575e-05,
|
| 1507 |
+
"loss": 0.0186,
|
| 1508 |
+
"step": 2140
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 3.6013400335008376,
|
| 1512 |
+
"grad_norm": 0.14377598464488983,
|
| 1513 |
+
"learning_rate": 1.9644689160594945e-05,
|
| 1514 |
+
"loss": 0.0181,
|
| 1515 |
+
"step": 2150
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 3.6180904522613067,
|
| 1519 |
+
"grad_norm": 0.18220196664333344,
|
| 1520 |
+
"learning_rate": 1.9550255963593553e-05,
|
| 1521 |
+
"loss": 0.018,
|
| 1522 |
+
"step": 2160
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 3.6348408710217757,
|
| 1526 |
+
"grad_norm": 0.22227473556995392,
|
| 1527 |
+
"learning_rate": 1.9455623723655524e-05,
|
| 1528 |
+
"loss": 0.018,
|
| 1529 |
+
"step": 2170
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 3.6515912897822447,
|
| 1533 |
+
"grad_norm": 0.10854306817054749,
|
| 1534 |
+
"learning_rate": 1.9360796580301606e-05,
|
| 1535 |
+
"loss": 0.0171,
|
| 1536 |
+
"step": 2180
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 3.6683417085427137,
|
| 1540 |
+
"grad_norm": 0.15350358188152313,
|
| 1541 |
+
"learning_rate": 1.9265778681578245e-05,
|
| 1542 |
+
"loss": 0.0187,
|
| 1543 |
+
"step": 2190
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 3.6850921273031827,
|
| 1547 |
+
"grad_norm": 0.1715448796749115,
|
| 1548 |
+
"learning_rate": 1.9170574183876143e-05,
|
| 1549 |
+
"loss": 0.0211,
|
| 1550 |
+
"step": 2200
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 3.7018425460636517,
|
| 1554 |
+
"grad_norm": 0.15281938016414642,
|
| 1555 |
+
"learning_rate": 1.9075187251748452e-05,
|
| 1556 |
+
"loss": 0.0205,
|
| 1557 |
+
"step": 2210
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 3.7185929648241207,
|
| 1561 |
+
"grad_norm": 0.1549525409936905,
|
| 1562 |
+
"learning_rate": 1.8979622057728596e-05,
|
| 1563 |
+
"loss": 0.0193,
|
| 1564 |
+
"step": 2220
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 3.7353433835845897,
|
| 1568 |
+
"grad_norm": 0.1523311883211136,
|
| 1569 |
+
"learning_rate": 1.8883882782147748e-05,
|
| 1570 |
+
"loss": 0.0184,
|
| 1571 |
+
"step": 2230
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 3.7520938023450587,
|
| 1575 |
+
"grad_norm": 0.18689101934432983,
|
| 1576 |
+
"learning_rate": 1.8787973612951967e-05,
|
| 1577 |
+
"loss": 0.0183,
|
| 1578 |
+
"step": 2240
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 3.7688442211055277,
|
| 1582 |
+
"grad_norm": 0.16317549347877502,
|
| 1583 |
+
"learning_rate": 1.8691898745519028e-05,
|
| 1584 |
+
"loss": 0.0189,
|
| 1585 |
+
"step": 2250
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 3.7855946398659968,
|
| 1589 |
+
"grad_norm": 0.12337777763605118,
|
| 1590 |
+
"learning_rate": 1.8595662382474863e-05,
|
| 1591 |
+
"loss": 0.0197,
|
| 1592 |
+
"step": 2260
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 3.8023450586264658,
|
| 1596 |
+
"grad_norm": 0.1409161388874054,
|
| 1597 |
+
"learning_rate": 1.8499268733509766e-05,
|
| 1598 |
+
"loss": 0.0181,
|
| 1599 |
+
"step": 2270
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 3.819095477386935,
|
| 1603 |
+
"grad_norm": 0.15133973956108093,
|
| 1604 |
+
"learning_rate": 1.84027220151942e-05,
|
| 1605 |
+
"loss": 0.0158,
|
| 1606 |
+
"step": 2280
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 3.835845896147404,
|
| 1610 |
+
"grad_norm": 0.13350911438465118,
|
| 1611 |
+
"learning_rate": 1.83060264507944e-05,
|
| 1612 |
+
"loss": 0.0184,
|
| 1613 |
+
"step": 2290
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 3.852596314907873,
|
| 1617 |
+
"grad_norm": 0.1512780487537384,
|
| 1618 |
+
"learning_rate": 1.82091862700876e-05,
|
| 1619 |
+
"loss": 0.0174,
|
| 1620 |
+
"step": 2300
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 3.869346733668342,
|
| 1624 |
+
"grad_norm": 0.14480029046535492,
|
| 1625 |
+
"learning_rate": 1.8112205709177023e-05,
|
| 1626 |
+
"loss": 0.0209,
|
| 1627 |
+
"step": 2310
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 3.886097152428811,
|
| 1631 |
+
"grad_norm": 0.11316472291946411,
|
| 1632 |
+
"learning_rate": 1.8015089010306585e-05,
|
| 1633 |
+
"loss": 0.0177,
|
| 1634 |
+
"step": 2320
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 3.90284757118928,
|
| 1638 |
+
"grad_norm": 0.18374566733837128,
|
| 1639 |
+
"learning_rate": 1.7917840421675298e-05,
|
| 1640 |
+
"loss": 0.0211,
|
| 1641 |
+
"step": 2330
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 3.919597989949749,
|
| 1645 |
+
"grad_norm": 0.14135697484016418,
|
| 1646 |
+
"learning_rate": 1.7820464197251474e-05,
|
| 1647 |
+
"loss": 0.0187,
|
| 1648 |
+
"step": 2340
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 3.936348408710218,
|
| 1652 |
+
"grad_norm": 0.13300488889217377,
|
| 1653 |
+
"learning_rate": 1.7722964596586627e-05,
|
| 1654 |
+
"loss": 0.0188,
|
| 1655 |
+
"step": 2350
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 3.953098827470687,
|
| 1659 |
+
"grad_norm": 0.17664390802383423,
|
| 1660 |
+
"learning_rate": 1.7625345884629144e-05,
|
| 1661 |
+
"loss": 0.0189,
|
| 1662 |
+
"step": 2360
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 3.969849246231156,
|
| 1666 |
+
"grad_norm": 0.11432985216379166,
|
| 1667 |
+
"learning_rate": 1.7527612331537722e-05,
|
| 1668 |
+
"loss": 0.0168,
|
| 1669 |
+
"step": 2370
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 3.986599664991625,
|
| 1673 |
+
"grad_norm": 0.16970939934253693,
|
| 1674 |
+
"learning_rate": 1.7429768212494593e-05,
|
| 1675 |
+
"loss": 0.0194,
|
| 1676 |
+
"step": 2380
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 4.0033500837520934,
|
| 1680 |
+
"grad_norm": 0.21405430138111115,
|
| 1681 |
+
"learning_rate": 1.7331817807518497e-05,
|
| 1682 |
+
"loss": 0.021,
|
| 1683 |
+
"step": 2390
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 4.0201005025125625,
|
| 1687 |
+
"grad_norm": 0.14571137726306915,
|
| 1688 |
+
"learning_rate": 1.723376540127745e-05,
|
| 1689 |
+
"loss": 0.0189,
|
| 1690 |
+
"step": 2400
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 4.0368509212730315,
|
| 1694 |
+
"grad_norm": 0.14875146746635437,
|
| 1695 |
+
"learning_rate": 1.7135615282901357e-05,
|
| 1696 |
+
"loss": 0.0166,
|
| 1697 |
+
"step": 2410
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 4.0536013400335005,
|
| 1701 |
+
"grad_norm": 0.1627630889415741,
|
| 1702 |
+
"learning_rate": 1.7037371745794358e-05,
|
| 1703 |
+
"loss": 0.0206,
|
| 1704 |
+
"step": 2420
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 4.0703517587939695,
|
| 1708 |
+
"grad_norm": 0.1687653660774231,
|
| 1709 |
+
"learning_rate": 1.6939039087447026e-05,
|
| 1710 |
+
"loss": 0.0188,
|
| 1711 |
+
"step": 2430
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 4.0871021775544385,
|
| 1715 |
+
"grad_norm": 0.13051913678646088,
|
| 1716 |
+
"learning_rate": 1.6840621609248397e-05,
|
| 1717 |
+
"loss": 0.0172,
|
| 1718 |
+
"step": 2440
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 4.1038525963149075,
|
| 1722 |
+
"grad_norm": 0.1474522203207016,
|
| 1723 |
+
"learning_rate": 1.6742123616297794e-05,
|
| 1724 |
+
"loss": 0.0181,
|
| 1725 |
+
"step": 2450
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 4.1206030150753765,
|
| 1729 |
+
"grad_norm": 0.11590561270713806,
|
| 1730 |
+
"learning_rate": 1.664354941721651e-05,
|
| 1731 |
+
"loss": 0.0161,
|
| 1732 |
+
"step": 2460
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 4.1373534338358455,
|
| 1736 |
+
"grad_norm": 0.12784413993358612,
|
| 1737 |
+
"learning_rate": 1.6544903323959355e-05,
|
| 1738 |
+
"loss": 0.0167,
|
| 1739 |
+
"step": 2470
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 4.1541038525963145,
|
| 1743 |
+
"grad_norm": 0.14732132852077484,
|
| 1744 |
+
"learning_rate": 1.6446189651626023e-05,
|
| 1745 |
+
"loss": 0.0186,
|
| 1746 |
+
"step": 2480
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 4.1708542713567835,
|
| 1750 |
+
"grad_norm": 0.15262089669704437,
|
| 1751 |
+
"learning_rate": 1.6347412718272316e-05,
|
| 1752 |
+
"loss": 0.0189,
|
| 1753 |
+
"step": 2490
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 4.187604690117253,
|
| 1757 |
+
"grad_norm": 0.12261030077934265,
|
| 1758 |
+
"learning_rate": 1.624857684472131e-05,
|
| 1759 |
+
"loss": 0.0166,
|
| 1760 |
+
"step": 2500
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 4.204355108877722,
|
| 1764 |
+
"grad_norm": 0.16570910811424255,
|
| 1765 |
+
"learning_rate": 1.6149686354374283e-05,
|
| 1766 |
+
"loss": 0.0175,
|
| 1767 |
+
"step": 2510
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 4.221105527638191,
|
| 1771 |
+
"grad_norm": 0.13071776926517487,
|
| 1772 |
+
"learning_rate": 1.605074557302165e-05,
|
| 1773 |
+
"loss": 0.017,
|
| 1774 |
+
"step": 2520
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 4.23785594639866,
|
| 1778 |
+
"grad_norm": 0.13835109770298004,
|
| 1779 |
+
"learning_rate": 1.59517588286537e-05,
|
| 1780 |
+
"loss": 0.0181,
|
| 1781 |
+
"step": 2530
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 4.254606365159129,
|
| 1785 |
+
"grad_norm": 0.13597555458545685,
|
| 1786 |
+
"learning_rate": 1.5852730451271324e-05,
|
| 1787 |
+
"loss": 0.0174,
|
| 1788 |
+
"step": 2540
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 4.271356783919598,
|
| 1792 |
+
"grad_norm": 0.11292872577905655,
|
| 1793 |
+
"learning_rate": 1.5753664772696546e-05,
|
| 1794 |
+
"loss": 0.0184,
|
| 1795 |
+
"step": 2550
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 4.288107202680067,
|
| 1799 |
+
"grad_norm": 0.18765822052955627,
|
| 1800 |
+
"learning_rate": 1.565456612638307e-05,
|
| 1801 |
+
"loss": 0.0192,
|
| 1802 |
+
"step": 2560
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 4.304857621440536,
|
| 1806 |
+
"grad_norm": 0.1113753542304039,
|
| 1807 |
+
"learning_rate": 1.5555438847226732e-05,
|
| 1808 |
+
"loss": 0.0155,
|
| 1809 |
+
"step": 2570
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 4.321608040201005,
|
| 1813 |
+
"grad_norm": 0.15917973220348358,
|
| 1814 |
+
"learning_rate": 1.545628727137584e-05,
|
| 1815 |
+
"loss": 0.0178,
|
| 1816 |
+
"step": 2580
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 4.338358458961474,
|
| 1820 |
+
"grad_norm": 0.1233622133731842,
|
| 1821 |
+
"learning_rate": 1.535711573604153e-05,
|
| 1822 |
+
"loss": 0.015,
|
| 1823 |
+
"step": 2590
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 4.355108877721943,
|
| 1827 |
+
"grad_norm": 0.12452328950166702,
|
| 1828 |
+
"learning_rate": 1.5257928579308032e-05,
|
| 1829 |
+
"loss": 0.0187,
|
| 1830 |
+
"step": 2600
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 4.371859296482412,
|
| 1834 |
+
"grad_norm": 0.13558247685432434,
|
| 1835 |
+
"learning_rate": 1.5158730139942902e-05,
|
| 1836 |
+
"loss": 0.0174,
|
| 1837 |
+
"step": 2610
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 4.388609715242881,
|
| 1841 |
+
"grad_norm": 0.15536141395568848,
|
| 1842 |
+
"learning_rate": 1.5059524757207239e-05,
|
| 1843 |
+
"loss": 0.0165,
|
| 1844 |
+
"step": 2620
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 4.40536013400335,
|
| 1848 |
+
"grad_norm": 0.1601993888616562,
|
| 1849 |
+
"learning_rate": 1.4960316770665866e-05,
|
| 1850 |
+
"loss": 0.0164,
|
| 1851 |
+
"step": 2630
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 4.422110552763819,
|
| 1855 |
+
"grad_norm": 0.13969165086746216,
|
| 1856 |
+
"learning_rate": 1.4861110519997495e-05,
|
| 1857 |
+
"loss": 0.0188,
|
| 1858 |
+
"step": 2640
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 4.438860971524288,
|
| 1862 |
+
"grad_norm": 0.11961787194013596,
|
| 1863 |
+
"learning_rate": 1.4761910344804913e-05,
|
| 1864 |
+
"loss": 0.016,
|
| 1865 |
+
"step": 2650
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 4.455611390284757,
|
| 1869 |
+
"grad_norm": 0.12037152051925659,
|
| 1870 |
+
"learning_rate": 1.466272058442516e-05,
|
| 1871 |
+
"loss": 0.0179,
|
| 1872 |
+
"step": 2660
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 4.472361809045226,
|
| 1876 |
+
"grad_norm": 0.15370553731918335,
|
| 1877 |
+
"learning_rate": 1.4563545577739676e-05,
|
| 1878 |
+
"loss": 0.0166,
|
| 1879 |
+
"step": 2670
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 4.489112227805695,
|
| 1883 |
+
"grad_norm": 0.11886673420667648,
|
| 1884 |
+
"learning_rate": 1.4464389662984543e-05,
|
| 1885 |
+
"loss": 0.0182,
|
| 1886 |
+
"step": 2680
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 4.505862646566165,
|
| 1890 |
+
"grad_norm": 0.13794168829917908,
|
| 1891 |
+
"learning_rate": 1.4365257177560684e-05,
|
| 1892 |
+
"loss": 0.0187,
|
| 1893 |
+
"step": 2690
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 4.522613065326633,
|
| 1897 |
+
"grad_norm": 0.16981415450572968,
|
| 1898 |
+
"learning_rate": 1.426615245784416e-05,
|
| 1899 |
+
"loss": 0.0161,
|
| 1900 |
+
"step": 2700
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 4.539363484087103,
|
| 1904 |
+
"grad_norm": 0.14231683313846588,
|
| 1905 |
+
"learning_rate": 1.4167079838996468e-05,
|
| 1906 |
+
"loss": 0.0164,
|
| 1907 |
+
"step": 2710
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 4.556113902847571,
|
| 1911 |
+
"grad_norm": 0.12874573469161987,
|
| 1912 |
+
"learning_rate": 1.4068043654774897e-05,
|
| 1913 |
+
"loss": 0.0165,
|
| 1914 |
+
"step": 2720
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 4.572864321608041,
|
| 1918 |
+
"grad_norm": 0.1743222177028656,
|
| 1919 |
+
"learning_rate": 1.3969048237342992e-05,
|
| 1920 |
+
"loss": 0.0179,
|
| 1921 |
+
"step": 2730
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 4.589614740368509,
|
| 1925 |
+
"grad_norm": 0.26698336005210876,
|
| 1926 |
+
"learning_rate": 1.3870097917081011e-05,
|
| 1927 |
+
"loss": 0.0169,
|
| 1928 |
+
"step": 2740
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 4.606365159128979,
|
| 1932 |
+
"grad_norm": 0.12284237146377563,
|
| 1933 |
+
"learning_rate": 1.3771197022396508e-05,
|
| 1934 |
+
"loss": 0.0146,
|
| 1935 |
+
"step": 2750
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 4.623115577889447,
|
| 1939 |
+
"grad_norm": 0.11153502017259598,
|
| 1940 |
+
"learning_rate": 1.3672349879535015e-05,
|
| 1941 |
+
"loss": 0.0153,
|
| 1942 |
+
"step": 2760
|
| 1943 |
+
},
|
| 1944 |
+
{
|
| 1945 |
+
"epoch": 4.639865996649917,
|
| 1946 |
+
"grad_norm": 0.13498924672603607,
|
| 1947 |
+
"learning_rate": 1.3573560812390781e-05,
|
| 1948 |
+
"loss": 0.0185,
|
| 1949 |
+
"step": 2770
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 4.656616415410385,
|
| 1953 |
+
"grad_norm": 0.13696305453777313,
|
| 1954 |
+
"learning_rate": 1.3474834142317635e-05,
|
| 1955 |
+
"loss": 0.0149,
|
| 1956 |
+
"step": 2780
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 4.673366834170855,
|
| 1960 |
+
"grad_norm": 0.13542267680168152,
|
| 1961 |
+
"learning_rate": 1.3376174187939955e-05,
|
| 1962 |
+
"loss": 0.0147,
|
| 1963 |
+
"step": 2790
|
| 1964 |
+
},
|
| 1965 |
+
{
|
| 1966 |
+
"epoch": 4.690117252931323,
|
| 1967 |
+
"grad_norm": 0.14426903426647186,
|
| 1968 |
+
"learning_rate": 1.3277585264963761e-05,
|
| 1969 |
+
"loss": 0.0176,
|
| 1970 |
+
"step": 2800
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 4.706867671691793,
|
| 1974 |
+
"grad_norm": 0.12668757140636444,
|
| 1975 |
+
"learning_rate": 1.3179071685987928e-05,
|
| 1976 |
+
"loss": 0.0179,
|
| 1977 |
+
"step": 2810
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 4.723618090452261,
|
| 1981 |
+
"grad_norm": 0.14874869585037231,
|
| 1982 |
+
"learning_rate": 1.308063776031554e-05,
|
| 1983 |
+
"loss": 0.0161,
|
| 1984 |
+
"step": 2820
|
| 1985 |
+
},
|
| 1986 |
+
{
|
| 1987 |
+
"epoch": 4.740368509212731,
|
| 1988 |
+
"grad_norm": 0.1480618417263031,
|
| 1989 |
+
"learning_rate": 1.2982287793765389e-05,
|
| 1990 |
+
"loss": 0.0151,
|
| 1991 |
+
"step": 2830
|
| 1992 |
+
},
|
| 1993 |
+
{
|
| 1994 |
+
"epoch": 4.757118927973199,
|
| 1995 |
+
"grad_norm": 0.1443120837211609,
|
| 1996 |
+
"learning_rate": 1.2884026088483616e-05,
|
| 1997 |
+
"loss": 0.0165,
|
| 1998 |
+
"step": 2840
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 4.773869346733669,
|
| 2002 |
+
"grad_norm": 0.16240982711315155,
|
| 2003 |
+
"learning_rate": 1.2785856942755543e-05,
|
| 2004 |
+
"loss": 0.0166,
|
| 2005 |
+
"step": 2850
|
| 2006 |
+
},
|
| 2007 |
+
{
|
| 2008 |
+
"epoch": 4.790619765494137,
|
| 2009 |
+
"grad_norm": 0.1501774787902832,
|
| 2010 |
+
"learning_rate": 1.2687784650817616e-05,
|
| 2011 |
+
"loss": 0.0178,
|
| 2012 |
+
"step": 2860
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 4.807370184254607,
|
| 2016 |
+
"grad_norm": 0.12356068193912506,
|
| 2017 |
+
"learning_rate": 1.2589813502669603e-05,
|
| 2018 |
+
"loss": 0.019,
|
| 2019 |
+
"step": 2870
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 4.824120603015075,
|
| 2023 |
+
"grad_norm": 0.1650429666042328,
|
| 2024 |
+
"learning_rate": 1.2491947783886907e-05,
|
| 2025 |
+
"loss": 0.0172,
|
| 2026 |
+
"step": 2880
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 4.840871021775545,
|
| 2030 |
+
"grad_norm": 0.10088960081338882,
|
| 2031 |
+
"learning_rate": 1.2394191775433106e-05,
|
| 2032 |
+
"loss": 0.0159,
|
| 2033 |
+
"step": 2890
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 4.857621440536013,
|
| 2037 |
+
"grad_norm": 0.14617778360843658,
|
| 2038 |
+
"learning_rate": 1.2296549753472696e-05,
|
| 2039 |
+
"loss": 0.0169,
|
| 2040 |
+
"step": 2900
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 4.874371859296483,
|
| 2044 |
+
"grad_norm": 0.14807341992855072,
|
| 2045 |
+
"learning_rate": 1.2199025989184032e-05,
|
| 2046 |
+
"loss": 0.0155,
|
| 2047 |
+
"step": 2910
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 4.891122278056951,
|
| 2051 |
+
"grad_norm": 0.32588717341423035,
|
| 2052 |
+
"learning_rate": 1.21016247485725e-05,
|
| 2053 |
+
"loss": 0.0206,
|
| 2054 |
+
"step": 2920
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 4.907872696817421,
|
| 2058 |
+
"grad_norm": 0.12184573709964752,
|
| 2059 |
+
"learning_rate": 1.2004350292283897e-05,
|
| 2060 |
+
"loss": 0.0168,
|
| 2061 |
+
"step": 2930
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 4.924623115577889,
|
| 2065 |
+
"grad_norm": 0.20224563777446747,
|
| 2066 |
+
"learning_rate": 1.1907206875418068e-05,
|
| 2067 |
+
"loss": 0.0167,
|
| 2068 |
+
"step": 2940
|
| 2069 |
+
},
|
| 2070 |
+
{
|
| 2071 |
+
"epoch": 4.941373534338359,
|
| 2072 |
+
"grad_norm": 0.1300506293773651,
|
| 2073 |
+
"learning_rate": 1.1810198747342769e-05,
|
| 2074 |
+
"loss": 0.0181,
|
| 2075 |
+
"step": 2950
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 4.958123953098827,
|
| 2079 |
+
"grad_norm": 0.15161874890327454,
|
| 2080 |
+
"learning_rate": 1.1713330151507786e-05,
|
| 2081 |
+
"loss": 0.0157,
|
| 2082 |
+
"step": 2960
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"epoch": 4.974874371859297,
|
| 2086 |
+
"grad_norm": 0.14213404059410095,
|
| 2087 |
+
"learning_rate": 1.161660532525931e-05,
|
| 2088 |
+
"loss": 0.0167,
|
| 2089 |
+
"step": 2970
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"epoch": 4.991624790619765,
|
| 2093 |
+
"grad_norm": 0.17077548801898956,
|
| 2094 |
+
"learning_rate": 1.1520028499654584e-05,
|
| 2095 |
+
"loss": 0.0162,
|
| 2096 |
+
"step": 2980
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 5.008375209380235,
|
| 2100 |
+
"grad_norm": 0.13322734832763672,
|
| 2101 |
+
"learning_rate": 1.1423603899276833e-05,
|
| 2102 |
+
"loss": 0.0144,
|
| 2103 |
+
"step": 2990
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 5.025125628140704,
|
| 2107 |
+
"grad_norm": 0.17589686810970306,
|
| 2108 |
+
"learning_rate": 1.1327335742050442e-05,
|
| 2109 |
+
"loss": 0.0163,
|
| 2110 |
+
"step": 3000
|
| 2111 |
+
}
|
| 2112 |
+
],
|
| 2113 |
+
"logging_steps": 10,
|
| 2114 |
+
"max_steps": 5000,
|
| 2115 |
+
"num_input_tokens_seen": 0,
|
| 2116 |
+
"num_train_epochs": 9,
|
| 2117 |
+
"save_steps": 5001,
|
| 2118 |
+
"stateful_callbacks": {
|
| 2119 |
+
"TrainerControl": {
|
| 2120 |
+
"args": {
|
| 2121 |
+
"should_epoch_stop": false,
|
| 2122 |
+
"should_evaluate": false,
|
| 2123 |
+
"should_log": false,
|
| 2124 |
+
"should_save": true,
|
| 2125 |
+
"should_training_stop": false
|
| 2126 |
+
},
|
| 2127 |
+
"attributes": {}
|
| 2128 |
+
}
|
| 2129 |
+
},
|
| 2130 |
+
"total_flos": 0.0,
|
| 2131 |
+
"train_batch_size": 64,
|
| 2132 |
+
"trial_name": null,
|
| 2133 |
+
"trial_params": null
|
| 2134 |
+
}
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed1/default/train_5shot_seed1_fullactionhead_5k/experiment_cfg/metadata.json
ADDED
|
@@ -0,0 +1,431 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"new_embodiment": {
|
| 3 |
+
"statistics": {
|
| 4 |
+
"state": {
|
| 5 |
+
"base_position": {
|
| 6 |
+
"max": [
|
| 7 |
+
7.778976917266846,
|
| 8 |
+
0.33403685688972473,
|
| 9 |
+
0.7223060727119446
|
| 10 |
+
],
|
| 11 |
+
"min": [
|
| 12 |
+
-4.821934223175049,
|
| 13 |
+
-6.785110950469971,
|
| 14 |
+
0.6994727253913879
|
| 15 |
+
],
|
| 16 |
+
"mean": [
|
| 17 |
+
2.5719731235192254,
|
| 18 |
+
-2.1489371002267235,
|
| 19 |
+
0.7003612741268546
|
| 20 |
+
],
|
| 21 |
+
"std": [
|
| 22 |
+
1.669945884010784,
|
| 23 |
+
1.3739411381361726,
|
| 24 |
+
0.0009413282367754987
|
| 25 |
+
],
|
| 26 |
+
"q01": [
|
| 27 |
+
-0.7944286339399775,
|
| 28 |
+
-5.486611518471242,
|
| 29 |
+
0.699977555971739
|
| 30 |
+
],
|
| 31 |
+
"q99": [
|
| 32 |
+
6.019886905263648,
|
| 33 |
+
-0.5229184210026855,
|
| 34 |
+
0.702930326282967
|
| 35 |
+
]
|
| 36 |
+
},
|
| 37 |
+
"base_rotation": {
|
| 38 |
+
"max": [
|
| 39 |
+
0.0,
|
| 40 |
+
0.0,
|
| 41 |
+
1.0,
|
| 42 |
+
1.0
|
| 43 |
+
],
|
| 44 |
+
"min": [
|
| 45 |
+
0.0,
|
| 46 |
+
0.0,
|
| 47 |
+
-1.0,
|
| 48 |
+
0.0
|
| 49 |
+
],
|
| 50 |
+
"mean": [
|
| 51 |
+
0.0,
|
| 52 |
+
0.0,
|
| 53 |
+
0.248543484335273,
|
| 54 |
+
0.5959317405459531
|
| 55 |
+
],
|
| 56 |
+
"std": [
|
| 57 |
+
0.0,
|
| 58 |
+
0.0,
|
| 59 |
+
0.6655550240074092,
|
| 60 |
+
0.374283995143559
|
| 61 |
+
],
|
| 62 |
+
"q01": [
|
| 63 |
+
0.0,
|
| 64 |
+
0.0,
|
| 65 |
+
-1.0000000000000002,
|
| 66 |
+
4.503254968910557e-06
|
| 67 |
+
],
|
| 68 |
+
"q99": [
|
| 69 |
+
0.0,
|
| 70 |
+
0.0,
|
| 71 |
+
1.0000000000000002,
|
| 72 |
+
1.0000000000000002
|
| 73 |
+
]
|
| 74 |
+
},
|
| 75 |
+
"end_effector_position_relative": {
|
| 76 |
+
"max": [
|
| 77 |
+
0.8848381638526917,
|
| 78 |
+
0.8569459915161133,
|
| 79 |
+
1.053884744644165
|
| 80 |
+
],
|
| 81 |
+
"min": [
|
| 82 |
+
-0.5828549861907959,
|
| 83 |
+
-0.8486236333847046,
|
| 84 |
+
-0.3446122705936432
|
| 85 |
+
],
|
| 86 |
+
"mean": [
|
| 87 |
+
0.2465490874130862,
|
| 88 |
+
-0.004711658908481467,
|
| 89 |
+
0.43463835464312073
|
| 90 |
+
],
|
| 91 |
+
"std": [
|
| 92 |
+
0.15798784670671595,
|
| 93 |
+
0.25624102400497073,
|
| 94 |
+
0.20836028727974915
|
| 95 |
+
],
|
| 96 |
+
"q01": [
|
| 97 |
+
-0.020752051134640608,
|
| 98 |
+
-0.4837933887347875,
|
| 99 |
+
0.1802942952289915
|
| 100 |
+
],
|
| 101 |
+
"q99": [
|
| 102 |
+
0.5386784714844007,
|
| 103 |
+
0.4801424212024721,
|
| 104 |
+
0.695095732975766
|
| 105 |
+
]
|
| 106 |
+
},
|
| 107 |
+
"end_effector_rotation_relative": {
|
| 108 |
+
"max": [
|
| 109 |
+
0.9999949932098389,
|
| 110 |
+
0.9997354745864868,
|
| 111 |
+
0.9900747537612915,
|
| 112 |
+
0.9735639095306396
|
| 113 |
+
],
|
| 114 |
+
"min": [
|
| 115 |
+
-0.9999914169311523,
|
| 116 |
+
-0.9999884366989136,
|
| 117 |
+
-0.9975369572639465,
|
| 118 |
+
6.754255821306288e-08
|
| 119 |
+
],
|
| 120 |
+
"mean": [
|
| 121 |
+
-0.22531859714056926,
|
| 122 |
+
0.046973702821026614,
|
| 123 |
+
-0.0568176967878771,
|
| 124 |
+
0.12364307042955697
|
| 125 |
+
],
|
| 126 |
+
"std": [
|
| 127 |
+
0.8502144333612052,
|
| 128 |
+
0.29591250318147405,
|
| 129 |
+
0.30218053156837016,
|
| 130 |
+
0.1636031353412931
|
| 131 |
+
],
|
| 132 |
+
"q01": [
|
| 133 |
+
-0.9959057387172566,
|
| 134 |
+
-0.5665384745573048,
|
| 135 |
+
-0.5329334620834977,
|
| 136 |
+
0.0014372279999397553
|
| 137 |
+
],
|
| 138 |
+
"q99": [
|
| 139 |
+
0.9904107951565839,
|
| 140 |
+
0.6359623612021396,
|
| 141 |
+
0.5026076022439377,
|
| 142 |
+
0.42955559234879914
|
| 143 |
+
]
|
| 144 |
+
},
|
| 145 |
+
"gripper_qpos": {
|
| 146 |
+
"max": [
|
| 147 |
+
0.05289580300450325,
|
| 148 |
+
0.00937709677964449
|
| 149 |
+
],
|
| 150 |
+
"min": [
|
| 151 |
+
-0.011669117026031017,
|
| 152 |
+
-0.06307756155729294
|
| 153 |
+
],
|
| 154 |
+
"mean": [
|
| 155 |
+
0.03188757060429838,
|
| 156 |
+
-0.0321793333086053
|
| 157 |
+
],
|
| 158 |
+
"std": [
|
| 159 |
+
0.01171052313737148,
|
| 160 |
+
0.011427336801203189
|
| 161 |
+
],
|
| 162 |
+
"q01": [
|
| 163 |
+
0.006462901319459925,
|
| 164 |
+
-0.040557140599419166
|
| 165 |
+
],
|
| 166 |
+
"q99": [
|
| 167 |
+
0.04037619335388386,
|
| 168 |
+
-0.006523947835993347
|
| 169 |
+
]
|
| 170 |
+
}
|
| 171 |
+
},
|
| 172 |
+
"action": {
|
| 173 |
+
"base_motion": {
|
| 174 |
+
"max": [
|
| 175 |
+
1.0,
|
| 176 |
+
1.0,
|
| 177 |
+
1.0,
|
| 178 |
+
0.0
|
| 179 |
+
],
|
| 180 |
+
"min": [
|
| 181 |
+
-1.0,
|
| 182 |
+
-1.0,
|
| 183 |
+
-1.0,
|
| 184 |
+
0.0
|
| 185 |
+
],
|
| 186 |
+
"mean": [
|
| 187 |
+
0.006639007355857881,
|
| 188 |
+
0.0027330421795561874,
|
| 189 |
+
-0.0017774441970214892,
|
| 190 |
+
0.0
|
| 191 |
+
],
|
| 192 |
+
"std": [
|
| 193 |
+
0.11841528114040736,
|
| 194 |
+
0.1041359083946544,
|
| 195 |
+
0.08346265521208351,
|
| 196 |
+
0.0
|
| 197 |
+
],
|
| 198 |
+
"q01": [
|
| 199 |
+
-0.17562359675270237,
|
| 200 |
+
-0.21706402321529558,
|
| 201 |
+
-0.16273392585723104,
|
| 202 |
+
0.0
|
| 203 |
+
],
|
| 204 |
+
"q99": [
|
| 205 |
+
0.2572918701644887,
|
| 206 |
+
0.27022325575787115,
|
| 207 |
+
0.1369626751049278,
|
| 208 |
+
0.0
|
| 209 |
+
]
|
| 210 |
+
},
|
| 211 |
+
"control_mode": {
|
| 212 |
+
"max": [
|
| 213 |
+
1.0
|
| 214 |
+
],
|
| 215 |
+
"min": [
|
| 216 |
+
-1.0
|
| 217 |
+
],
|
| 218 |
+
"mean": [
|
| 219 |
+
-0.9205539608927732
|
| 220 |
+
],
|
| 221 |
+
"std": [
|
| 222 |
+
0.39063657242534666
|
| 223 |
+
],
|
| 224 |
+
"q01": [
|
| 225 |
+
-1.0
|
| 226 |
+
],
|
| 227 |
+
"q99": [
|
| 228 |
+
-0.17711863866562935
|
| 229 |
+
]
|
| 230 |
+
},
|
| 231 |
+
"end_effector_position": {
|
| 232 |
+
"max": [
|
| 233 |
+
1.0,
|
| 234 |
+
1.0,
|
| 235 |
+
1.0
|
| 236 |
+
],
|
| 237 |
+
"min": [
|
| 238 |
+
-1.0,
|
| 239 |
+
-1.0,
|
| 240 |
+
-1.0
|
| 241 |
+
],
|
| 242 |
+
"mean": [
|
| 243 |
+
-0.00924803964621179,
|
| 244 |
+
0.004053841761390575,
|
| 245 |
+
-0.08385115578460103
|
| 246 |
+
],
|
| 247 |
+
"std": [
|
| 248 |
+
0.45413072518223824,
|
| 249 |
+
0.40158612714634856,
|
| 250 |
+
0.41068634292849304
|
| 251 |
+
],
|
| 252 |
+
"q01": [
|
| 253 |
+
-0.9669817145875343,
|
| 254 |
+
-0.9241307799783718,
|
| 255 |
+
-0.8264079503068825
|
| 256 |
+
],
|
| 257 |
+
"q99": [
|
| 258 |
+
0.8021511975722166,
|
| 259 |
+
0.9241346582748478,
|
| 260 |
+
0.8348116781502298
|
| 261 |
+
]
|
| 262 |
+
},
|
| 263 |
+
"end_effector_rotation": {
|
| 264 |
+
"max": [
|
| 265 |
+
1.0,
|
| 266 |
+
1.0,
|
| 267 |
+
1.0
|
| 268 |
+
],
|
| 269 |
+
"min": [
|
| 270 |
+
-1.0,
|
| 271 |
+
-1.0,
|
| 272 |
+
-1.0
|
| 273 |
+
],
|
| 274 |
+
"mean": [
|
| 275 |
+
0.007644157504353809,
|
| 276 |
+
-0.019962543093886335,
|
| 277 |
+
-0.002047110452602617
|
| 278 |
+
],
|
| 279 |
+
"std": [
|
| 280 |
+
0.11076012154791087,
|
| 281 |
+
0.13176735140285795,
|
| 282 |
+
0.12286528665895614
|
| 283 |
+
],
|
| 284 |
+
"q01": [
|
| 285 |
+
-0.2659105439401192,
|
| 286 |
+
-0.3961694033813533,
|
| 287 |
+
-0.3434870832468936
|
| 288 |
+
],
|
| 289 |
+
"q99": [
|
| 290 |
+
0.3145786959266369,
|
| 291 |
+
0.31309430361227547,
|
| 292 |
+
0.34107754401942986
|
| 293 |
+
]
|
| 294 |
+
},
|
| 295 |
+
"gripper_close": {
|
| 296 |
+
"max": [
|
| 297 |
+
1.0
|
| 298 |
+
],
|
| 299 |
+
"min": [
|
| 300 |
+
-1.0
|
| 301 |
+
],
|
| 302 |
+
"mean": [
|
| 303 |
+
-0.29870612267533575
|
| 304 |
+
],
|
| 305 |
+
"std": [
|
| 306 |
+
0.9543487555375691
|
| 307 |
+
],
|
| 308 |
+
"q01": [
|
| 309 |
+
-1.0
|
| 310 |
+
],
|
| 311 |
+
"q99": [
|
| 312 |
+
0.7610674802484976
|
| 313 |
+
]
|
| 314 |
+
}
|
| 315 |
+
}
|
| 316 |
+
},
|
| 317 |
+
"modalities": {
|
| 318 |
+
"video": {
|
| 319 |
+
"robot0_eye_in_hand": {
|
| 320 |
+
"resolution": [
|
| 321 |
+
256,
|
| 322 |
+
256
|
| 323 |
+
],
|
| 324 |
+
"channels": 3,
|
| 325 |
+
"fps": 20.0
|
| 326 |
+
},
|
| 327 |
+
"robot0_agentview_left": {
|
| 328 |
+
"resolution": [
|
| 329 |
+
256,
|
| 330 |
+
256
|
| 331 |
+
],
|
| 332 |
+
"channels": 3,
|
| 333 |
+
"fps": 20.0
|
| 334 |
+
},
|
| 335 |
+
"robot0_agentview_right": {
|
| 336 |
+
"resolution": [
|
| 337 |
+
256,
|
| 338 |
+
256
|
| 339 |
+
],
|
| 340 |
+
"channels": 3,
|
| 341 |
+
"fps": 20.0
|
| 342 |
+
}
|
| 343 |
+
},
|
| 344 |
+
"state": {
|
| 345 |
+
"base_position": {
|
| 346 |
+
"absolute": true,
|
| 347 |
+
"rotation_type": null,
|
| 348 |
+
"shape": [
|
| 349 |
+
3
|
| 350 |
+
],
|
| 351 |
+
"continuous": true
|
| 352 |
+
},
|
| 353 |
+
"base_rotation": {
|
| 354 |
+
"absolute": true,
|
| 355 |
+
"rotation_type": "quaternion",
|
| 356 |
+
"shape": [
|
| 357 |
+
4
|
| 358 |
+
],
|
| 359 |
+
"continuous": true
|
| 360 |
+
},
|
| 361 |
+
"end_effector_position_relative": {
|
| 362 |
+
"absolute": true,
|
| 363 |
+
"rotation_type": null,
|
| 364 |
+
"shape": [
|
| 365 |
+
3
|
| 366 |
+
],
|
| 367 |
+
"continuous": true
|
| 368 |
+
},
|
| 369 |
+
"end_effector_rotation_relative": {
|
| 370 |
+
"absolute": true,
|
| 371 |
+
"rotation_type": "quaternion",
|
| 372 |
+
"shape": [
|
| 373 |
+
4
|
| 374 |
+
],
|
| 375 |
+
"continuous": true
|
| 376 |
+
},
|
| 377 |
+
"gripper_qpos": {
|
| 378 |
+
"absolute": true,
|
| 379 |
+
"rotation_type": null,
|
| 380 |
+
"shape": [
|
| 381 |
+
2
|
| 382 |
+
],
|
| 383 |
+
"continuous": true
|
| 384 |
+
}
|
| 385 |
+
},
|
| 386 |
+
"action": {
|
| 387 |
+
"base_motion": {
|
| 388 |
+
"absolute": true,
|
| 389 |
+
"rotation_type": null,
|
| 390 |
+
"shape": [
|
| 391 |
+
4
|
| 392 |
+
],
|
| 393 |
+
"continuous": true
|
| 394 |
+
},
|
| 395 |
+
"control_mode": {
|
| 396 |
+
"absolute": true,
|
| 397 |
+
"rotation_type": null,
|
| 398 |
+
"shape": [
|
| 399 |
+
1
|
| 400 |
+
],
|
| 401 |
+
"continuous": true
|
| 402 |
+
},
|
| 403 |
+
"end_effector_position": {
|
| 404 |
+
"absolute": true,
|
| 405 |
+
"rotation_type": null,
|
| 406 |
+
"shape": [
|
| 407 |
+
3
|
| 408 |
+
],
|
| 409 |
+
"continuous": true
|
| 410 |
+
},
|
| 411 |
+
"end_effector_rotation": {
|
| 412 |
+
"absolute": true,
|
| 413 |
+
"rotation_type": "axis_angle",
|
| 414 |
+
"shape": [
|
| 415 |
+
3
|
| 416 |
+
],
|
| 417 |
+
"continuous": true
|
| 418 |
+
},
|
| 419 |
+
"gripper_close": {
|
| 420 |
+
"absolute": true,
|
| 421 |
+
"rotation_type": null,
|
| 422 |
+
"shape": [
|
| 423 |
+
1
|
| 424 |
+
],
|
| 425 |
+
"continuous": true
|
| 426 |
+
}
|
| 427 |
+
}
|
| 428 |
+
},
|
| 429 |
+
"embodiment_tag": "new_embodiment"
|
| 430 |
+
}
|
| 431 |
+
}
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-1000/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 |
+
}
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-1000/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6609af3d3bd3bf14cadbe66a91687968c154549621c5df35b9966992f230416a
|
| 3 |
+
size 4999367032
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-1000/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:69171ad5963fab0bd3f27b20e600d60bafda65d5522e7dbf5c4619528f62628e
|
| 3 |
+
size 2586705312
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-1000/model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-1000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3e5ea787a79065bd33196ddfc4a157aadd4974fe449cec2f580e807e4a9a2600
|
| 3 |
+
size 8550719959
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-1000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e3b5547f9bceb2f4c2f2444aef8721d4c47e59d74bc07eb896351aab894daa96
|
| 3 |
+
size 14645
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-1000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6e19ead5779e3f8f2f666b7793de3a03d118f57f1b16a97eeb87bd732d18b445
|
| 3 |
+
size 1465
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-3000/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:47267e3f9c5a3cc4eff7aa426660c1ccc906d211cac5b9561adc60dc7b62d4d5
|
| 3 |
+
size 4999367032
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-3000/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:940b04c26d7de74a2e8db252b7ce90fac13dc3f4eb65005d3a43947ae267fcae
|
| 3 |
+
size 2586705312
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-3000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:64f3ef76c50e1cd0a2c91e4863b114c43241cc2cd478c6560d32cdc2ca6b23b8
|
| 3 |
+
size 14645
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-3000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:42d8e6d8e53e6b837084d0cc18fdd63ca9e40bd44a65d38eb52695647a8e84e8
|
| 3 |
+
size 1465
|
FewShot/39tasks/5shot/FromBaseline26_FullactionHead_5k/seed2/default/train_5shot_seed2_fullactionhead_5k/checkpoint-3000/trainer_state.json
ADDED
|
@@ -0,0 +1,2134 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 5.128205128205128,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 3000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.017094017094017096,
|
| 14 |
+
"grad_norm": 1.2046294212341309,
|
| 15 |
+
"learning_rate": 1.08e-06,
|
| 16 |
+
"loss": 0.2224,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.03418803418803419,
|
| 21 |
+
"grad_norm": 1.3672140836715698,
|
| 22 |
+
"learning_rate": 2.28e-06,
|
| 23 |
+
"loss": 0.2037,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.05128205128205128,
|
| 28 |
+
"grad_norm": 1.2484416961669922,
|
| 29 |
+
"learning_rate": 3.48e-06,
|
| 30 |
+
"loss": 0.193,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.06837606837606838,
|
| 35 |
+
"grad_norm": 0.7811980247497559,
|
| 36 |
+
"learning_rate": 4.68e-06,
|
| 37 |
+
"loss": 0.158,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08547008547008547,
|
| 42 |
+
"grad_norm": 1.1057628393173218,
|
| 43 |
+
"learning_rate": 5.8800000000000005e-06,
|
| 44 |
+
"loss": 0.1518,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.10256410256410256,
|
| 49 |
+
"grad_norm": 0.6198669672012329,
|
| 50 |
+
"learning_rate": 7.08e-06,
|
| 51 |
+
"loss": 0.1235,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.11965811965811966,
|
| 56 |
+
"grad_norm": 0.6399325728416443,
|
| 57 |
+
"learning_rate": 8.28e-06,
|
| 58 |
+
"loss": 0.1012,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.13675213675213677,
|
| 63 |
+
"grad_norm": 0.34900981187820435,
|
| 64 |
+
"learning_rate": 9.48e-06,
|
| 65 |
+
"loss": 0.0894,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.15384615384615385,
|
| 70 |
+
"grad_norm": 0.24374151229858398,
|
| 71 |
+
"learning_rate": 1.068e-05,
|
| 72 |
+
"loss": 0.0847,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.17094017094017094,
|
| 77 |
+
"grad_norm": 0.3181402087211609,
|
| 78 |
+
"learning_rate": 1.1880000000000001e-05,
|
| 79 |
+
"loss": 0.076,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.18803418803418803,
|
| 84 |
+
"grad_norm": 0.3558669984340668,
|
| 85 |
+
"learning_rate": 1.308e-05,
|
| 86 |
+
"loss": 0.0756,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.20512820512820512,
|
| 91 |
+
"grad_norm": 0.25770920515060425,
|
| 92 |
+
"learning_rate": 1.428e-05,
|
| 93 |
+
"loss": 0.0678,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.2222222222222222,
|
| 98 |
+
"grad_norm": 0.3094629645347595,
|
| 99 |
+
"learning_rate": 1.548e-05,
|
| 100 |
+
"loss": 0.0656,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.23931623931623933,
|
| 105 |
+
"grad_norm": 0.31149178743362427,
|
| 106 |
+
"learning_rate": 1.6680000000000003e-05,
|
| 107 |
+
"loss": 0.0661,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.2564102564102564,
|
| 112 |
+
"grad_norm": 0.28312113881111145,
|
| 113 |
+
"learning_rate": 1.7879999999999998e-05,
|
| 114 |
+
"loss": 0.0644,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.27350427350427353,
|
| 119 |
+
"grad_norm": 0.32937219738960266,
|
| 120 |
+
"learning_rate": 1.908e-05,
|
| 121 |
+
"loss": 0.0634,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.2905982905982906,
|
| 126 |
+
"grad_norm": 0.24805986881256104,
|
| 127 |
+
"learning_rate": 2.0280000000000002e-05,
|
| 128 |
+
"loss": 0.0604,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.3076923076923077,
|
| 133 |
+
"grad_norm": 0.3144904673099518,
|
| 134 |
+
"learning_rate": 2.148e-05,
|
| 135 |
+
"loss": 0.0559,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.3247863247863248,
|
| 140 |
+
"grad_norm": 0.22721275687217712,
|
| 141 |
+
"learning_rate": 2.268e-05,
|
| 142 |
+
"loss": 0.0559,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.3418803418803419,
|
| 147 |
+
"grad_norm": 0.24481946229934692,
|
| 148 |
+
"learning_rate": 2.3880000000000002e-05,
|
| 149 |
+
"loss": 0.0563,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.358974358974359,
|
| 154 |
+
"grad_norm": 0.29072335362434387,
|
| 155 |
+
"learning_rate": 2.508e-05,
|
| 156 |
+
"loss": 0.0543,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.37606837606837606,
|
| 161 |
+
"grad_norm": 0.2773635983467102,
|
| 162 |
+
"learning_rate": 2.628e-05,
|
| 163 |
+
"loss": 0.0533,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.39316239316239315,
|
| 168 |
+
"grad_norm": 0.28752514719963074,
|
| 169 |
+
"learning_rate": 2.748e-05,
|
| 170 |
+
"loss": 0.0531,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.41025641025641024,
|
| 175 |
+
"grad_norm": 0.2401578277349472,
|
| 176 |
+
"learning_rate": 2.868e-05,
|
| 177 |
+
"loss": 0.0543,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.42735042735042733,
|
| 182 |
+
"grad_norm": 0.22457115352153778,
|
| 183 |
+
"learning_rate": 2.9880000000000002e-05,
|
| 184 |
+
"loss": 0.0498,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.4444444444444444,
|
| 189 |
+
"grad_norm": 0.22619064152240753,
|
| 190 |
+
"learning_rate": 2.9999734259635676e-05,
|
| 191 |
+
"loss": 0.0495,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.46153846153846156,
|
| 196 |
+
"grad_norm": 0.23179255425930023,
|
| 197 |
+
"learning_rate": 2.9998815663057245e-05,
|
| 198 |
+
"loss": 0.0463,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.47863247863247865,
|
| 203 |
+
"grad_norm": 0.23389773070812225,
|
| 204 |
+
"learning_rate": 2.999724096969228e-05,
|
| 205 |
+
"loss": 0.0476,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.49572649572649574,
|
| 210 |
+
"grad_norm": 0.24128898978233337,
|
| 211 |
+
"learning_rate": 2.9995010248422984e-05,
|
| 212 |
+
"loss": 0.0437,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.5128205128205128,
|
| 217 |
+
"grad_norm": 0.2507748007774353,
|
| 218 |
+
"learning_rate": 2.999212359682832e-05,
|
| 219 |
+
"loss": 0.0474,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.5299145299145299,
|
| 224 |
+
"grad_norm": 0.21490685641765594,
|
| 225 |
+
"learning_rate": 2.9988581141179787e-05,
|
| 226 |
+
"loss": 0.0434,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.5470085470085471,
|
| 231 |
+
"grad_norm": 0.22923023998737335,
|
| 232 |
+
"learning_rate": 2.998438303643587e-05,
|
| 233 |
+
"loss": 0.044,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.5641025641025641,
|
| 238 |
+
"grad_norm": 0.22642545402050018,
|
| 239 |
+
"learning_rate": 2.997952946623527e-05,
|
| 240 |
+
"loss": 0.0429,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.5811965811965812,
|
| 245 |
+
"grad_norm": 0.221108540892601,
|
| 246 |
+
"learning_rate": 2.9974020642888875e-05,
|
| 247 |
+
"loss": 0.0465,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.5982905982905983,
|
| 252 |
+
"grad_norm": 0.19173994660377502,
|
| 253 |
+
"learning_rate": 2.9967856807370458e-05,
|
| 254 |
+
"loss": 0.0433,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.6153846153846154,
|
| 259 |
+
"grad_norm": 0.2417234629392624,
|
| 260 |
+
"learning_rate": 2.996103822930615e-05,
|
| 261 |
+
"loss": 0.0418,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.6324786324786325,
|
| 266 |
+
"grad_norm": 0.22277653217315674,
|
| 267 |
+
"learning_rate": 2.9953565206962653e-05,
|
| 268 |
+
"loss": 0.0388,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.6495726495726496,
|
| 273 |
+
"grad_norm": 0.16107745468616486,
|
| 274 |
+
"learning_rate": 2.9945438067234172e-05,
|
| 275 |
+
"loss": 0.0382,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.6666666666666666,
|
| 280 |
+
"grad_norm": 0.15353648364543915,
|
| 281 |
+
"learning_rate": 2.9936657165628124e-05,
|
| 282 |
+
"loss": 0.0406,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.6837606837606838,
|
| 287 |
+
"grad_norm": 0.18034681677818298,
|
| 288 |
+
"learning_rate": 2.992722288624959e-05,
|
| 289 |
+
"loss": 0.0415,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.7008547008547008,
|
| 294 |
+
"grad_norm": 0.23336061835289001,
|
| 295 |
+
"learning_rate": 2.9917135641784515e-05,
|
| 296 |
+
"loss": 0.0414,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.717948717948718,
|
| 301 |
+
"grad_norm": 0.19218440353870392,
|
| 302 |
+
"learning_rate": 2.9906395873481655e-05,
|
| 303 |
+
"loss": 0.0392,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.7350427350427351,
|
| 308 |
+
"grad_norm": 0.22765028476715088,
|
| 309 |
+
"learning_rate": 2.989500405113326e-05,
|
| 310 |
+
"loss": 0.0403,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.7521367521367521,
|
| 315 |
+
"grad_norm": 0.20929336547851562,
|
| 316 |
+
"learning_rate": 2.9882960673054547e-05,
|
| 317 |
+
"loss": 0.0369,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.7692307692307693,
|
| 322 |
+
"grad_norm": 0.17191247642040253,
|
| 323 |
+
"learning_rate": 2.9870266266061886e-05,
|
| 324 |
+
"loss": 0.0354,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.7863247863247863,
|
| 329 |
+
"grad_norm": 0.2489067018032074,
|
| 330 |
+
"learning_rate": 2.9856921385449766e-05,
|
| 331 |
+
"loss": 0.0365,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.8034188034188035,
|
| 336 |
+
"grad_norm": 0.25272417068481445,
|
| 337 |
+
"learning_rate": 2.9842926614966502e-05,
|
| 338 |
+
"loss": 0.0385,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.8205128205128205,
|
| 343 |
+
"grad_norm": 0.19717201590538025,
|
| 344 |
+
"learning_rate": 2.982828256678868e-05,
|
| 345 |
+
"loss": 0.0339,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.8376068376068376,
|
| 350 |
+
"grad_norm": 0.25095418095588684,
|
| 351 |
+
"learning_rate": 2.9812989881494412e-05,
|
| 352 |
+
"loss": 0.0352,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.8547008547008547,
|
| 357 |
+
"grad_norm": 0.2102588564157486,
|
| 358 |
+
"learning_rate": 2.9797049228035296e-05,
|
| 359 |
+
"loss": 0.0369,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.8717948717948718,
|
| 364 |
+
"grad_norm": 0.20112262666225433,
|
| 365 |
+
"learning_rate": 2.978046130370715e-05,
|
| 366 |
+
"loss": 0.0326,
|
| 367 |
+
"step": 510
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.8888888888888888,
|
| 371 |
+
"grad_norm": 0.19247573614120483,
|
| 372 |
+
"learning_rate": 2.9763226834119535e-05,
|
| 373 |
+
"loss": 0.0362,
|
| 374 |
+
"step": 520
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.905982905982906,
|
| 378 |
+
"grad_norm": 0.21424508094787598,
|
| 379 |
+
"learning_rate": 2.9745346573163963e-05,
|
| 380 |
+
"loss": 0.0359,
|
| 381 |
+
"step": 530
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.9230769230769231,
|
| 385 |
+
"grad_norm": 0.26557332277297974,
|
| 386 |
+
"learning_rate": 2.9726821302980987e-05,
|
| 387 |
+
"loss": 0.0345,
|
| 388 |
+
"step": 540
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.9401709401709402,
|
| 392 |
+
"grad_norm": 0.19497093558311462,
|
| 393 |
+
"learning_rate": 2.9707651833925935e-05,
|
| 394 |
+
"loss": 0.0298,
|
| 395 |
+
"step": 550
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.9572649572649573,
|
| 399 |
+
"grad_norm": 0.17632479965686798,
|
| 400 |
+
"learning_rate": 2.968783900453348e-05,
|
| 401 |
+
"loss": 0.0333,
|
| 402 |
+
"step": 560
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.9743589743589743,
|
| 406 |
+
"grad_norm": 0.2774340808391571,
|
| 407 |
+
"learning_rate": 2.9667383681480968e-05,
|
| 408 |
+
"loss": 0.0351,
|
| 409 |
+
"step": 570
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.9914529914529915,
|
| 413 |
+
"grad_norm": 0.17489512264728546,
|
| 414 |
+
"learning_rate": 2.9646286759550493e-05,
|
| 415 |
+
"loss": 0.0341,
|
| 416 |
+
"step": 580
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 1.0085470085470085,
|
| 420 |
+
"grad_norm": 0.18617698550224304,
|
| 421 |
+
"learning_rate": 2.9624549161589764e-05,
|
| 422 |
+
"loss": 0.031,
|
| 423 |
+
"step": 590
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 1.0256410256410255,
|
| 427 |
+
"grad_norm": 0.1862032115459442,
|
| 428 |
+
"learning_rate": 2.9602171838471745e-05,
|
| 429 |
+
"loss": 0.0327,
|
| 430 |
+
"step": 600
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 1.0427350427350428,
|
| 434 |
+
"grad_norm": 0.16319014132022858,
|
| 435 |
+
"learning_rate": 2.957915576905303e-05,
|
| 436 |
+
"loss": 0.0311,
|
| 437 |
+
"step": 610
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 1.0598290598290598,
|
| 441 |
+
"grad_norm": 0.18638090789318085,
|
| 442 |
+
"learning_rate": 2.9555501960131058e-05,
|
| 443 |
+
"loss": 0.0328,
|
| 444 |
+
"step": 620
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 1.0769230769230769,
|
| 448 |
+
"grad_norm": 0.159883052110672,
|
| 449 |
+
"learning_rate": 2.9531211446400068e-05,
|
| 450 |
+
"loss": 0.0319,
|
| 451 |
+
"step": 630
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 1.0940170940170941,
|
| 455 |
+
"grad_norm": 0.17638160288333893,
|
| 456 |
+
"learning_rate": 2.9506285290405813e-05,
|
| 457 |
+
"loss": 0.0324,
|
| 458 |
+
"step": 640
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 1.1111111111111112,
|
| 462 |
+
"grad_norm": 0.17149314284324646,
|
| 463 |
+
"learning_rate": 2.948072458249911e-05,
|
| 464 |
+
"loss": 0.033,
|
| 465 |
+
"step": 650
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 1.1282051282051282,
|
| 469 |
+
"grad_norm": 0.1854614019393921,
|
| 470 |
+
"learning_rate": 2.9454530440788125e-05,
|
| 471 |
+
"loss": 0.0321,
|
| 472 |
+
"step": 660
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 1.1452991452991452,
|
| 476 |
+
"grad_norm": 0.18628022074699402,
|
| 477 |
+
"learning_rate": 2.9427704011089474e-05,
|
| 478 |
+
"loss": 0.0308,
|
| 479 |
+
"step": 670
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 1.1623931623931625,
|
| 483 |
+
"grad_norm": 0.15781794488430023,
|
| 484 |
+
"learning_rate": 2.94002464668781e-05,
|
| 485 |
+
"loss": 0.0289,
|
| 486 |
+
"step": 680
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 1.1794871794871795,
|
| 490 |
+
"grad_norm": 0.18502387404441833,
|
| 491 |
+
"learning_rate": 2.9372159009235936e-05,
|
| 492 |
+
"loss": 0.0285,
|
| 493 |
+
"step": 690
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 1.1965811965811965,
|
| 497 |
+
"grad_norm": 0.13340029120445251,
|
| 498 |
+
"learning_rate": 2.9343442866799365e-05,
|
| 499 |
+
"loss": 0.03,
|
| 500 |
+
"step": 700
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 1.2136752136752136,
|
| 504 |
+
"grad_norm": 0.16018730401992798,
|
| 505 |
+
"learning_rate": 2.9314099295705495e-05,
|
| 506 |
+
"loss": 0.0283,
|
| 507 |
+
"step": 710
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 1.2307692307692308,
|
| 511 |
+
"grad_norm": 0.1523085981607437,
|
| 512 |
+
"learning_rate": 2.928412957953718e-05,
|
| 513 |
+
"loss": 0.0276,
|
| 514 |
+
"step": 720
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 1.2478632478632479,
|
| 518 |
+
"grad_norm": 0.1537986546754837,
|
| 519 |
+
"learning_rate": 2.925353502926689e-05,
|
| 520 |
+
"loss": 0.0291,
|
| 521 |
+
"step": 730
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 1.264957264957265,
|
| 525 |
+
"grad_norm": 0.1516842246055603,
|
| 526 |
+
"learning_rate": 2.922231698319937e-05,
|
| 527 |
+
"loss": 0.0315,
|
| 528 |
+
"step": 740
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 1.282051282051282,
|
| 532 |
+
"grad_norm": 0.1799113154411316,
|
| 533 |
+
"learning_rate": 2.9190476806913088e-05,
|
| 534 |
+
"loss": 0.03,
|
| 535 |
+
"step": 750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 1.2991452991452992,
|
| 539 |
+
"grad_norm": 0.2539559304714203,
|
| 540 |
+
"learning_rate": 2.9158015893200504e-05,
|
| 541 |
+
"loss": 0.0288,
|
| 542 |
+
"step": 760
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 1.3162393162393162,
|
| 546 |
+
"grad_norm": 0.1597651243209839,
|
| 547 |
+
"learning_rate": 2.9124935662007137e-05,
|
| 548 |
+
"loss": 0.0303,
|
| 549 |
+
"step": 770
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.3333333333333333,
|
| 553 |
+
"grad_norm": 0.18849721550941467,
|
| 554 |
+
"learning_rate": 2.9091237560369467e-05,
|
| 555 |
+
"loss": 0.0295,
|
| 556 |
+
"step": 780
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 1.3504273504273505,
|
| 560 |
+
"grad_norm": 0.13707464933395386,
|
| 561 |
+
"learning_rate": 2.905692306235163e-05,
|
| 562 |
+
"loss": 0.0261,
|
| 563 |
+
"step": 790
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 1.3675213675213675,
|
| 567 |
+
"grad_norm": 0.1797618865966797,
|
| 568 |
+
"learning_rate": 2.9021993668980925e-05,
|
| 569 |
+
"loss": 0.0266,
|
| 570 |
+
"step": 800
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 1.3846153846153846,
|
| 574 |
+
"grad_norm": 0.18061158061027527,
|
| 575 |
+
"learning_rate": 2.8986450908182185e-05,
|
| 576 |
+
"loss": 0.0277,
|
| 577 |
+
"step": 810
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 1.4017094017094016,
|
| 581 |
+
"grad_norm": 0.14432664215564728,
|
| 582 |
+
"learning_rate": 2.8950296334710902e-05,
|
| 583 |
+
"loss": 0.0268,
|
| 584 |
+
"step": 820
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 1.4188034188034189,
|
| 588 |
+
"grad_norm": 0.19661664962768555,
|
| 589 |
+
"learning_rate": 2.8913531530085248e-05,
|
| 590 |
+
"loss": 0.0275,
|
| 591 |
+
"step": 830
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 1.435897435897436,
|
| 595 |
+
"grad_norm": 0.16374897956848145,
|
| 596 |
+
"learning_rate": 2.887615810251687e-05,
|
| 597 |
+
"loss": 0.0272,
|
| 598 |
+
"step": 840
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 1.452991452991453,
|
| 602 |
+
"grad_norm": 0.16281157732009888,
|
| 603 |
+
"learning_rate": 2.8838177686840577e-05,
|
| 604 |
+
"loss": 0.0283,
|
| 605 |
+
"step": 850
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 1.4700854700854702,
|
| 609 |
+
"grad_norm": 0.1767149716615677,
|
| 610 |
+
"learning_rate": 2.8799591944442776e-05,
|
| 611 |
+
"loss": 0.0291,
|
| 612 |
+
"step": 860
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 1.4871794871794872,
|
| 616 |
+
"grad_norm": 0.2484145313501358,
|
| 617 |
+
"learning_rate": 2.8760402563188848e-05,
|
| 618 |
+
"loss": 0.0284,
|
| 619 |
+
"step": 870
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 1.5042735042735043,
|
| 623 |
+
"grad_norm": 0.21058519184589386,
|
| 624 |
+
"learning_rate": 2.8720611257349285e-05,
|
| 625 |
+
"loss": 0.0304,
|
| 626 |
+
"step": 880
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 1.5213675213675213,
|
| 630 |
+
"grad_norm": 0.19965742528438568,
|
| 631 |
+
"learning_rate": 2.868021976752471e-05,
|
| 632 |
+
"loss": 0.028,
|
| 633 |
+
"step": 890
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 1.5384615384615383,
|
| 637 |
+
"grad_norm": 0.1859794408082962,
|
| 638 |
+
"learning_rate": 2.8639229860569738e-05,
|
| 639 |
+
"loss": 0.0292,
|
| 640 |
+
"step": 900
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 1.5555555555555556,
|
| 644 |
+
"grad_norm": 0.18770064413547516,
|
| 645 |
+
"learning_rate": 2.8597643329515688e-05,
|
| 646 |
+
"loss": 0.0264,
|
| 647 |
+
"step": 910
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 1.5726495726495726,
|
| 651 |
+
"grad_norm": 0.18467789888381958,
|
| 652 |
+
"learning_rate": 2.855546199349214e-05,
|
| 653 |
+
"loss": 0.0253,
|
| 654 |
+
"step": 920
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 1.5897435897435899,
|
| 658 |
+
"grad_norm": 0.1546042114496231,
|
| 659 |
+
"learning_rate": 2.85126876976474e-05,
|
| 660 |
+
"loss": 0.0282,
|
| 661 |
+
"step": 930
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 1.606837606837607,
|
| 665 |
+
"grad_norm": 0.17473237216472626,
|
| 666 |
+
"learning_rate": 2.8469322313067726e-05,
|
| 667 |
+
"loss": 0.0262,
|
| 668 |
+
"step": 940
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 1.623931623931624,
|
| 672 |
+
"grad_norm": 0.17649267613887787,
|
| 673 |
+
"learning_rate": 2.8425367736695522e-05,
|
| 674 |
+
"loss": 0.026,
|
| 675 |
+
"step": 950
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 1.641025641025641,
|
| 679 |
+
"grad_norm": 0.15947958827018738,
|
| 680 |
+
"learning_rate": 2.838082589124636e-05,
|
| 681 |
+
"loss": 0.0258,
|
| 682 |
+
"step": 960
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 1.658119658119658,
|
| 686 |
+
"grad_norm": 0.18516002595424652,
|
| 687 |
+
"learning_rate": 2.8335698725124843e-05,
|
| 688 |
+
"loss": 0.0277,
|
| 689 |
+
"step": 970
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 1.6752136752136753,
|
| 693 |
+
"grad_norm": 0.1311817169189453,
|
| 694 |
+
"learning_rate": 2.8289988212339422e-05,
|
| 695 |
+
"loss": 0.0253,
|
| 696 |
+
"step": 980
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 1.6923076923076923,
|
| 700 |
+
"grad_norm": 0.18864114582538605,
|
| 701 |
+
"learning_rate": 2.8243696352415997e-05,
|
| 702 |
+
"loss": 0.03,
|
| 703 |
+
"step": 990
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 1.7094017094017095,
|
| 707 |
+
"grad_norm": 0.1583152711391449,
|
| 708 |
+
"learning_rate": 2.8196825170310495e-05,
|
| 709 |
+
"loss": 0.024,
|
| 710 |
+
"step": 1000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 1.7264957264957266,
|
| 714 |
+
"grad_norm": 0.19438093900680542,
|
| 715 |
+
"learning_rate": 2.8149376716320256e-05,
|
| 716 |
+
"loss": 0.025,
|
| 717 |
+
"step": 1010
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 1.7435897435897436,
|
| 721 |
+
"grad_norm": 0.17569419741630554,
|
| 722 |
+
"learning_rate": 2.8101353065994373e-05,
|
| 723 |
+
"loss": 0.0299,
|
| 724 |
+
"step": 1020
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 1.7606837606837606,
|
| 728 |
+
"grad_norm": 0.1977061629295349,
|
| 729 |
+
"learning_rate": 2.8052756320042885e-05,
|
| 730 |
+
"loss": 0.0235,
|
| 731 |
+
"step": 1030
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 1.7777777777777777,
|
| 735 |
+
"grad_norm": 0.14877791702747345,
|
| 736 |
+
"learning_rate": 2.8003588604244893e-05,
|
| 737 |
+
"loss": 0.0241,
|
| 738 |
+
"step": 1040
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 1.7948717948717947,
|
| 742 |
+
"grad_norm": 0.18093445897102356,
|
| 743 |
+
"learning_rate": 2.7953852069355565e-05,
|
| 744 |
+
"loss": 0.0215,
|
| 745 |
+
"step": 1050
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 1.811965811965812,
|
| 749 |
+
"grad_norm": 0.13836748898029327,
|
| 750 |
+
"learning_rate": 2.790354889101206e-05,
|
| 751 |
+
"loss": 0.0219,
|
| 752 |
+
"step": 1060
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 1.8290598290598292,
|
| 756 |
+
"grad_norm": 0.1955246478319168,
|
| 757 |
+
"learning_rate": 2.7852681269638366e-05,
|
| 758 |
+
"loss": 0.0262,
|
| 759 |
+
"step": 1070
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 1.8461538461538463,
|
| 763 |
+
"grad_norm": 0.13735222816467285,
|
| 764 |
+
"learning_rate": 2.780125143034902e-05,
|
| 765 |
+
"loss": 0.0234,
|
| 766 |
+
"step": 1080
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 1.8632478632478633,
|
| 770 |
+
"grad_norm": 0.20777004957199097,
|
| 771 |
+
"learning_rate": 2.7749261622851812e-05,
|
| 772 |
+
"loss": 0.0225,
|
| 773 |
+
"step": 1090
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.8803418803418803,
|
| 777 |
+
"grad_norm": 0.15912115573883057,
|
| 778 |
+
"learning_rate": 2.7696714121349335e-05,
|
| 779 |
+
"loss": 0.0235,
|
| 780 |
+
"step": 1100
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.8974358974358974,
|
| 784 |
+
"grad_norm": 0.14222097396850586,
|
| 785 |
+
"learning_rate": 2.764361122443954e-05,
|
| 786 |
+
"loss": 0.0217,
|
| 787 |
+
"step": 1110
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.9145299145299144,
|
| 791 |
+
"grad_norm": 0.12735839188098907,
|
| 792 |
+
"learning_rate": 2.758995525501517e-05,
|
| 793 |
+
"loss": 0.024,
|
| 794 |
+
"step": 1120
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.9316239316239316,
|
| 798 |
+
"grad_norm": 0.13369329273700714,
|
| 799 |
+
"learning_rate": 2.753574856016215e-05,
|
| 800 |
+
"loss": 0.0259,
|
| 801 |
+
"step": 1130
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 1.9487179487179487,
|
| 805 |
+
"grad_norm": 0.14417479932308197,
|
| 806 |
+
"learning_rate": 2.7480993511056913e-05,
|
| 807 |
+
"loss": 0.024,
|
| 808 |
+
"step": 1140
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 1.965811965811966,
|
| 812 |
+
"grad_norm": 0.13541226089000702,
|
| 813 |
+
"learning_rate": 2.7425692502862684e-05,
|
| 814 |
+
"loss": 0.021,
|
| 815 |
+
"step": 1150
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 1.982905982905983,
|
| 819 |
+
"grad_norm": 0.13426485657691956,
|
| 820 |
+
"learning_rate": 2.7369847954624728e-05,
|
| 821 |
+
"loss": 0.0249,
|
| 822 |
+
"step": 1160
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 2.0,
|
| 826 |
+
"grad_norm": 0.22447210550308228,
|
| 827 |
+
"learning_rate": 2.731346230916448e-05,
|
| 828 |
+
"loss": 0.0275,
|
| 829 |
+
"step": 1170
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 2.017094017094017,
|
| 833 |
+
"grad_norm": 0.1316627860069275,
|
| 834 |
+
"learning_rate": 2.7256538032972746e-05,
|
| 835 |
+
"loss": 0.0236,
|
| 836 |
+
"step": 1180
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 2.034188034188034,
|
| 840 |
+
"grad_norm": 0.1581067442893982,
|
| 841 |
+
"learning_rate": 2.7199077616101773e-05,
|
| 842 |
+
"loss": 0.0209,
|
| 843 |
+
"step": 1190
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 2.051282051282051,
|
| 847 |
+
"grad_norm": 0.1645020693540573,
|
| 848 |
+
"learning_rate": 2.714108357205634e-05,
|
| 849 |
+
"loss": 0.0254,
|
| 850 |
+
"step": 1200
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 2.0683760683760686,
|
| 854 |
+
"grad_norm": 0.15171298384666443,
|
| 855 |
+
"learning_rate": 2.7082558437683803e-05,
|
| 856 |
+
"loss": 0.0219,
|
| 857 |
+
"step": 1210
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 2.0854700854700856,
|
| 861 |
+
"grad_norm": 0.14786364138126373,
|
| 862 |
+
"learning_rate": 2.702350477306315e-05,
|
| 863 |
+
"loss": 0.0232,
|
| 864 |
+
"step": 1220
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 2.1025641025641026,
|
| 868 |
+
"grad_norm": 0.19275642931461334,
|
| 869 |
+
"learning_rate": 2.6963925161392963e-05,
|
| 870 |
+
"loss": 0.0261,
|
| 871 |
+
"step": 1230
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 2.1196581196581197,
|
| 875 |
+
"grad_norm": 0.21039193868637085,
|
| 876 |
+
"learning_rate": 2.6903822208878476e-05,
|
| 877 |
+
"loss": 0.0237,
|
| 878 |
+
"step": 1240
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 2.1367521367521367,
|
| 882 |
+
"grad_norm": 0.1635989099740982,
|
| 883 |
+
"learning_rate": 2.6843198544617535e-05,
|
| 884 |
+
"loss": 0.024,
|
| 885 |
+
"step": 1250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 2.1538461538461537,
|
| 889 |
+
"grad_norm": 0.14116357266902924,
|
| 890 |
+
"learning_rate": 2.6782056820485612e-05,
|
| 891 |
+
"loss": 0.0233,
|
| 892 |
+
"step": 1260
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 2.1709401709401708,
|
| 896 |
+
"grad_norm": 0.1337611824274063,
|
| 897 |
+
"learning_rate": 2.672039971101979e-05,
|
| 898 |
+
"loss": 0.0223,
|
| 899 |
+
"step": 1270
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 2.1880341880341883,
|
| 903 |
+
"grad_norm": 0.12742610275745392,
|
| 904 |
+
"learning_rate": 2.665822991330178e-05,
|
| 905 |
+
"loss": 0.0233,
|
| 906 |
+
"step": 1280
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 2.2051282051282053,
|
| 910 |
+
"grad_norm": 0.16495631635189056,
|
| 911 |
+
"learning_rate": 2.6595550146839934e-05,
|
| 912 |
+
"loss": 0.0244,
|
| 913 |
+
"step": 1290
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 2.2222222222222223,
|
| 917 |
+
"grad_norm": 0.13133522868156433,
|
| 918 |
+
"learning_rate": 2.6532363153450287e-05,
|
| 919 |
+
"loss": 0.0236,
|
| 920 |
+
"step": 1300
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 2.2393162393162394,
|
| 924 |
+
"grad_norm": 0.11399739980697632,
|
| 925 |
+
"learning_rate": 2.6468671697136634e-05,
|
| 926 |
+
"loss": 0.0222,
|
| 927 |
+
"step": 1310
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 2.2564102564102564,
|
| 931 |
+
"grad_norm": 0.12264499813318253,
|
| 932 |
+
"learning_rate": 2.640447856396961e-05,
|
| 933 |
+
"loss": 0.0205,
|
| 934 |
+
"step": 1320
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 2.2735042735042734,
|
| 938 |
+
"grad_norm": 0.1786639243364334,
|
| 939 |
+
"learning_rate": 2.6339786561964805e-05,
|
| 940 |
+
"loss": 0.0251,
|
| 941 |
+
"step": 1330
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 2.2905982905982905,
|
| 945 |
+
"grad_norm": 0.15574303269386292,
|
| 946 |
+
"learning_rate": 2.6274598520959966e-05,
|
| 947 |
+
"loss": 0.0226,
|
| 948 |
+
"step": 1340
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 2.3076923076923075,
|
| 952 |
+
"grad_norm": 0.15068724751472473,
|
| 953 |
+
"learning_rate": 2.6208917292491185e-05,
|
| 954 |
+
"loss": 0.0219,
|
| 955 |
+
"step": 1350
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 2.324786324786325,
|
| 959 |
+
"grad_norm": 0.16358910501003265,
|
| 960 |
+
"learning_rate": 2.6142745749668176e-05,
|
| 961 |
+
"loss": 0.0225,
|
| 962 |
+
"step": 1360
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 2.341880341880342,
|
| 966 |
+
"grad_norm": 0.21654905378818512,
|
| 967 |
+
"learning_rate": 2.607608678704859e-05,
|
| 968 |
+
"loss": 0.0232,
|
| 969 |
+
"step": 1370
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 2.358974358974359,
|
| 973 |
+
"grad_norm": 0.22834260761737823,
|
| 974 |
+
"learning_rate": 2.600894332051139e-05,
|
| 975 |
+
"loss": 0.0215,
|
| 976 |
+
"step": 1380
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 2.376068376068376,
|
| 980 |
+
"grad_norm": 0.17436617612838745,
|
| 981 |
+
"learning_rate": 2.594131828712932e-05,
|
| 982 |
+
"loss": 0.0217,
|
| 983 |
+
"step": 1390
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 2.393162393162393,
|
| 987 |
+
"grad_norm": 0.13975125551223755,
|
| 988 |
+
"learning_rate": 2.5873214645040414e-05,
|
| 989 |
+
"loss": 0.02,
|
| 990 |
+
"step": 1400
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 2.41025641025641,
|
| 994 |
+
"grad_norm": 0.18962903320789337,
|
| 995 |
+
"learning_rate": 2.5804635373318604e-05,
|
| 996 |
+
"loss": 0.0216,
|
| 997 |
+
"step": 1410
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 2.427350427350427,
|
| 1001 |
+
"grad_norm": 0.1517896205186844,
|
| 1002 |
+
"learning_rate": 2.57355834718434e-05,
|
| 1003 |
+
"loss": 0.0221,
|
| 1004 |
+
"step": 1420
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 2.4444444444444446,
|
| 1008 |
+
"grad_norm": 0.1567123383283615,
|
| 1009 |
+
"learning_rate": 2.566606196116867e-05,
|
| 1010 |
+
"loss": 0.0218,
|
| 1011 |
+
"step": 1430
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 2.4615384615384617,
|
| 1015 |
+
"grad_norm": 0.2237260341644287,
|
| 1016 |
+
"learning_rate": 2.559607388239051e-05,
|
| 1017 |
+
"loss": 0.0205,
|
| 1018 |
+
"step": 1440
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 2.4786324786324787,
|
| 1022 |
+
"grad_norm": 0.1526583880186081,
|
| 1023 |
+
"learning_rate": 2.552562229701422e-05,
|
| 1024 |
+
"loss": 0.0215,
|
| 1025 |
+
"step": 1450
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 2.4957264957264957,
|
| 1029 |
+
"grad_norm": 0.14554832875728607,
|
| 1030 |
+
"learning_rate": 2.545471028682036e-05,
|
| 1031 |
+
"loss": 0.0235,
|
| 1032 |
+
"step": 1460
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 2.5128205128205128,
|
| 1036 |
+
"grad_norm": 0.14237725734710693,
|
| 1037 |
+
"learning_rate": 2.538334095373e-05,
|
| 1038 |
+
"loss": 0.0209,
|
| 1039 |
+
"step": 1470
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 2.52991452991453,
|
| 1043 |
+
"grad_norm": 0.17106768488883972,
|
| 1044 |
+
"learning_rate": 2.531151741966896e-05,
|
| 1045 |
+
"loss": 0.0215,
|
| 1046 |
+
"step": 1480
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 2.547008547008547,
|
| 1050 |
+
"grad_norm": 0.12999463081359863,
|
| 1051 |
+
"learning_rate": 2.5239242826431302e-05,
|
| 1052 |
+
"loss": 0.0207,
|
| 1053 |
+
"step": 1490
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 2.564102564102564,
|
| 1057 |
+
"grad_norm": 0.23773175477981567,
|
| 1058 |
+
"learning_rate": 2.5166520335541865e-05,
|
| 1059 |
+
"loss": 0.0218,
|
| 1060 |
+
"step": 1500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 2.5811965811965814,
|
| 1064 |
+
"grad_norm": 0.11908961087465286,
|
| 1065 |
+
"learning_rate": 2.5093353128117996e-05,
|
| 1066 |
+
"loss": 0.0223,
|
| 1067 |
+
"step": 1510
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 2.5982905982905984,
|
| 1071 |
+
"grad_norm": 0.25520920753479004,
|
| 1072 |
+
"learning_rate": 2.501974440473037e-05,
|
| 1073 |
+
"loss": 0.0198,
|
| 1074 |
+
"step": 1520
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 2.6153846153846154,
|
| 1078 |
+
"grad_norm": 0.21974427998065948,
|
| 1079 |
+
"learning_rate": 2.4945697385263015e-05,
|
| 1080 |
+
"loss": 0.0223,
|
| 1081 |
+
"step": 1530
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 2.6324786324786325,
|
| 1085 |
+
"grad_norm": 0.1549292355775833,
|
| 1086 |
+
"learning_rate": 2.487121530877243e-05,
|
| 1087 |
+
"loss": 0.0208,
|
| 1088 |
+
"step": 1540
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 2.6495726495726495,
|
| 1092 |
+
"grad_norm": 0.17997530102729797,
|
| 1093 |
+
"learning_rate": 2.4796301433345943e-05,
|
| 1094 |
+
"loss": 0.0207,
|
| 1095 |
+
"step": 1550
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 2.6666666666666665,
|
| 1099 |
+
"grad_norm": 0.15226222574710846,
|
| 1100 |
+
"learning_rate": 2.4720959035959148e-05,
|
| 1101 |
+
"loss": 0.0218,
|
| 1102 |
+
"step": 1560
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 2.683760683760684,
|
| 1106 |
+
"grad_norm": 0.2047930359840393,
|
| 1107 |
+
"learning_rate": 2.464519141233257e-05,
|
| 1108 |
+
"loss": 0.0215,
|
| 1109 |
+
"step": 1570
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 2.700854700854701,
|
| 1113 |
+
"grad_norm": 0.16199080646038055,
|
| 1114 |
+
"learning_rate": 2.456900187678754e-05,
|
| 1115 |
+
"loss": 0.0214,
|
| 1116 |
+
"step": 1580
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 2.717948717948718,
|
| 1120 |
+
"grad_norm": 0.11895282566547394,
|
| 1121 |
+
"learning_rate": 2.4492393762101147e-05,
|
| 1122 |
+
"loss": 0.0202,
|
| 1123 |
+
"step": 1590
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 2.735042735042735,
|
| 1127 |
+
"grad_norm": 0.15882159769535065,
|
| 1128 |
+
"learning_rate": 2.4415370419360508e-05,
|
| 1129 |
+
"loss": 0.022,
|
| 1130 |
+
"step": 1600
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 2.752136752136752,
|
| 1134 |
+
"grad_norm": 0.14013583958148956,
|
| 1135 |
+
"learning_rate": 2.4337935217816157e-05,
|
| 1136 |
+
"loss": 0.022,
|
| 1137 |
+
"step": 1610
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 2.769230769230769,
|
| 1141 |
+
"grad_norm": 0.20299099385738373,
|
| 1142 |
+
"learning_rate": 2.4260091544734665e-05,
|
| 1143 |
+
"loss": 0.0207,
|
| 1144 |
+
"step": 1620
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 2.786324786324786,
|
| 1148 |
+
"grad_norm": 0.21985934674739838,
|
| 1149 |
+
"learning_rate": 2.4181842805250466e-05,
|
| 1150 |
+
"loss": 0.0214,
|
| 1151 |
+
"step": 1630
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 2.8034188034188032,
|
| 1155 |
+
"grad_norm": 0.14461225271224976,
|
| 1156 |
+
"learning_rate": 2.4103192422216923e-05,
|
| 1157 |
+
"loss": 0.0195,
|
| 1158 |
+
"step": 1640
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 2.8205128205128203,
|
| 1162 |
+
"grad_norm": 0.10911990702152252,
|
| 1163 |
+
"learning_rate": 2.402414383605658e-05,
|
| 1164 |
+
"loss": 0.0204,
|
| 1165 |
+
"step": 1650
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 2.8376068376068377,
|
| 1169 |
+
"grad_norm": 0.13800372183322906,
|
| 1170 |
+
"learning_rate": 2.3944700504610674e-05,
|
| 1171 |
+
"loss": 0.0192,
|
| 1172 |
+
"step": 1660
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 2.8547008547008548,
|
| 1176 |
+
"grad_norm": 0.14190635085105896,
|
| 1177 |
+
"learning_rate": 2.3864865902987883e-05,
|
| 1178 |
+
"loss": 0.0204,
|
| 1179 |
+
"step": 1670
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 2.871794871794872,
|
| 1183 |
+
"grad_norm": 0.14053791761398315,
|
| 1184 |
+
"learning_rate": 2.378464352341231e-05,
|
| 1185 |
+
"loss": 0.0199,
|
| 1186 |
+
"step": 1680
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 2.888888888888889,
|
| 1190 |
+
"grad_norm": 0.1504385620355606,
|
| 1191 |
+
"learning_rate": 2.3704036875070728e-05,
|
| 1192 |
+
"loss": 0.0233,
|
| 1193 |
+
"step": 1690
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 2.905982905982906,
|
| 1197 |
+
"grad_norm": 0.18149782717227936,
|
| 1198 |
+
"learning_rate": 2.362304948395906e-05,
|
| 1199 |
+
"loss": 0.0235,
|
| 1200 |
+
"step": 1700
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 2.9230769230769234,
|
| 1204 |
+
"grad_norm": 0.17068181931972504,
|
| 1205 |
+
"learning_rate": 2.3541684892728157e-05,
|
| 1206 |
+
"loss": 0.0212,
|
| 1207 |
+
"step": 1710
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 2.9401709401709404,
|
| 1211 |
+
"grad_norm": 0.1624104231595993,
|
| 1212 |
+
"learning_rate": 2.345994666052883e-05,
|
| 1213 |
+
"loss": 0.0177,
|
| 1214 |
+
"step": 1720
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 2.9572649572649574,
|
| 1218 |
+
"grad_norm": 0.19329632818698883,
|
| 1219 |
+
"learning_rate": 2.337783836285614e-05,
|
| 1220 |
+
"loss": 0.0214,
|
| 1221 |
+
"step": 1730
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 2.9743589743589745,
|
| 1225 |
+
"grad_norm": 0.13297595083713531,
|
| 1226 |
+
"learning_rate": 2.3295363591393026e-05,
|
| 1227 |
+
"loss": 0.0184,
|
| 1228 |
+
"step": 1740
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 2.9914529914529915,
|
| 1232 |
+
"grad_norm": 0.12240353226661682,
|
| 1233 |
+
"learning_rate": 2.3212525953853174e-05,
|
| 1234 |
+
"loss": 0.0212,
|
| 1235 |
+
"step": 1750
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 3.0085470085470085,
|
| 1239 |
+
"grad_norm": 0.130058154463768,
|
| 1240 |
+
"learning_rate": 2.3129329073823206e-05,
|
| 1241 |
+
"loss": 0.0205,
|
| 1242 |
+
"step": 1760
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 3.0256410256410255,
|
| 1246 |
+
"grad_norm": 0.11880940198898315,
|
| 1247 |
+
"learning_rate": 2.3045776590604175e-05,
|
| 1248 |
+
"loss": 0.0193,
|
| 1249 |
+
"step": 1770
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 3.0427350427350426,
|
| 1253 |
+
"grad_norm": 0.12890465557575226,
|
| 1254 |
+
"learning_rate": 2.296187215905237e-05,
|
| 1255 |
+
"loss": 0.0212,
|
| 1256 |
+
"step": 1780
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 3.0598290598290596,
|
| 1260 |
+
"grad_norm": 0.1741575449705124,
|
| 1261 |
+
"learning_rate": 2.2877619449419437e-05,
|
| 1262 |
+
"loss": 0.0212,
|
| 1263 |
+
"step": 1790
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 3.076923076923077,
|
| 1267 |
+
"grad_norm": 0.1907522976398468,
|
| 1268 |
+
"learning_rate": 2.2793022147191843e-05,
|
| 1269 |
+
"loss": 0.0203,
|
| 1270 |
+
"step": 1800
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 3.094017094017094,
|
| 1274 |
+
"grad_norm": 0.17696961760520935,
|
| 1275 |
+
"learning_rate": 2.2708083952929634e-05,
|
| 1276 |
+
"loss": 0.0181,
|
| 1277 |
+
"step": 1810
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 3.111111111111111,
|
| 1281 |
+
"grad_norm": 0.1593712717294693,
|
| 1282 |
+
"learning_rate": 2.2622808582104594e-05,
|
| 1283 |
+
"loss": 0.0197,
|
| 1284 |
+
"step": 1820
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 3.128205128205128,
|
| 1288 |
+
"grad_norm": 0.12099262326955795,
|
| 1289 |
+
"learning_rate": 2.2537199764937703e-05,
|
| 1290 |
+
"loss": 0.0188,
|
| 1291 |
+
"step": 1830
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 3.1452991452991452,
|
| 1295 |
+
"grad_norm": 0.16116322576999664,
|
| 1296 |
+
"learning_rate": 2.245126124623595e-05,
|
| 1297 |
+
"loss": 0.0196,
|
| 1298 |
+
"step": 1840
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 3.1623931623931623,
|
| 1302 |
+
"grad_norm": 0.1666380614042282,
|
| 1303 |
+
"learning_rate": 2.2364996785228547e-05,
|
| 1304 |
+
"loss": 0.0203,
|
| 1305 |
+
"step": 1850
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 3.1794871794871793,
|
| 1309 |
+
"grad_norm": 0.13607926666736603,
|
| 1310 |
+
"learning_rate": 2.227841015540248e-05,
|
| 1311 |
+
"loss": 0.0212,
|
| 1312 |
+
"step": 1860
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 3.1965811965811968,
|
| 1316 |
+
"grad_norm": 0.1656973510980606,
|
| 1317 |
+
"learning_rate": 2.2191505144337433e-05,
|
| 1318 |
+
"loss": 0.0195,
|
| 1319 |
+
"step": 1870
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 3.213675213675214,
|
| 1323 |
+
"grad_norm": 0.13743546605110168,
|
| 1324 |
+
"learning_rate": 2.210428555354013e-05,
|
| 1325 |
+
"loss": 0.0213,
|
| 1326 |
+
"step": 1880
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 3.230769230769231,
|
| 1330 |
+
"grad_norm": 0.15309397876262665,
|
| 1331 |
+
"learning_rate": 2.201675519827802e-05,
|
| 1332 |
+
"loss": 0.0215,
|
| 1333 |
+
"step": 1890
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 3.247863247863248,
|
| 1337 |
+
"grad_norm": 0.13153575360774994,
|
| 1338 |
+
"learning_rate": 2.1928917907412403e-05,
|
| 1339 |
+
"loss": 0.0189,
|
| 1340 |
+
"step": 1900
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 3.264957264957265,
|
| 1344 |
+
"grad_norm": 0.16302916407585144,
|
| 1345 |
+
"learning_rate": 2.1840777523230937e-05,
|
| 1346 |
+
"loss": 0.0189,
|
| 1347 |
+
"step": 1910
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 3.282051282051282,
|
| 1351 |
+
"grad_norm": 0.12375231087207794,
|
| 1352 |
+
"learning_rate": 2.1752337901279552e-05,
|
| 1353 |
+
"loss": 0.021,
|
| 1354 |
+
"step": 1920
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 3.299145299145299,
|
| 1358 |
+
"grad_norm": 0.18395335972309113,
|
| 1359 |
+
"learning_rate": 2.166360291019383e-05,
|
| 1360 |
+
"loss": 0.0187,
|
| 1361 |
+
"step": 1930
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 3.316239316239316,
|
| 1365 |
+
"grad_norm": 0.12572312355041504,
|
| 1366 |
+
"learning_rate": 2.157457643152973e-05,
|
| 1367 |
+
"loss": 0.0203,
|
| 1368 |
+
"step": 1940
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 3.3333333333333335,
|
| 1372 |
+
"grad_norm": 0.12790457904338837,
|
| 1373 |
+
"learning_rate": 2.148526235959385e-05,
|
| 1374 |
+
"loss": 0.0177,
|
| 1375 |
+
"step": 1950
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 3.3504273504273505,
|
| 1379 |
+
"grad_norm": 0.19870531558990479,
|
| 1380 |
+
"learning_rate": 2.1395664601273016e-05,
|
| 1381 |
+
"loss": 0.0204,
|
| 1382 |
+
"step": 1960
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 3.3675213675213675,
|
| 1386 |
+
"grad_norm": 0.18182598054409027,
|
| 1387 |
+
"learning_rate": 2.130578707586345e-05,
|
| 1388 |
+
"loss": 0.023,
|
| 1389 |
+
"step": 1970
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 3.3846153846153846,
|
| 1393 |
+
"grad_norm": 0.13019001483917236,
|
| 1394 |
+
"learning_rate": 2.1215633714899264e-05,
|
| 1395 |
+
"loss": 0.021,
|
| 1396 |
+
"step": 1980
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 3.4017094017094016,
|
| 1400 |
+
"grad_norm": 0.12526895105838776,
|
| 1401 |
+
"learning_rate": 2.112520846198053e-05,
|
| 1402 |
+
"loss": 0.0201,
|
| 1403 |
+
"step": 1990
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 3.4188034188034186,
|
| 1407 |
+
"grad_norm": 0.1366206407546997,
|
| 1408 |
+
"learning_rate": 2.103451527260074e-05,
|
| 1409 |
+
"loss": 0.0186,
|
| 1410 |
+
"step": 2000
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 3.435897435897436,
|
| 1414 |
+
"grad_norm": 0.13825932145118713,
|
| 1415 |
+
"learning_rate": 2.094355811397381e-05,
|
| 1416 |
+
"loss": 0.0193,
|
| 1417 |
+
"step": 2010
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 3.452991452991453,
|
| 1421 |
+
"grad_norm": 0.10611677914857864,
|
| 1422 |
+
"learning_rate": 2.0852340964860522e-05,
|
| 1423 |
+
"loss": 0.0203,
|
| 1424 |
+
"step": 2020
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 3.47008547008547,
|
| 1428 |
+
"grad_norm": 0.15153281390666962,
|
| 1429 |
+
"learning_rate": 2.076086781539447e-05,
|
| 1430 |
+
"loss": 0.02,
|
| 1431 |
+
"step": 2030
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 3.4871794871794872,
|
| 1435 |
+
"grad_norm": 0.12375279515981674,
|
| 1436 |
+
"learning_rate": 2.066914266690755e-05,
|
| 1437 |
+
"loss": 0.0182,
|
| 1438 |
+
"step": 2040
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 3.5042735042735043,
|
| 1442 |
+
"grad_norm": 0.12043378502130508,
|
| 1443 |
+
"learning_rate": 2.0577169531754906e-05,
|
| 1444 |
+
"loss": 0.0197,
|
| 1445 |
+
"step": 2050
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 3.5213675213675213,
|
| 1449 |
+
"grad_norm": 0.12067852169275284,
|
| 1450 |
+
"learning_rate": 2.048495243313942e-05,
|
| 1451 |
+
"loss": 0.0195,
|
| 1452 |
+
"step": 2060
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 3.5384615384615383,
|
| 1456 |
+
"grad_norm": 0.16918447613716125,
|
| 1457 |
+
"learning_rate": 2.039249540493572e-05,
|
| 1458 |
+
"loss": 0.0201,
|
| 1459 |
+
"step": 2070
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 3.5555555555555554,
|
| 1463 |
+
"grad_norm": 0.1681264340877533,
|
| 1464 |
+
"learning_rate": 2.0299802491513755e-05,
|
| 1465 |
+
"loss": 0.0172,
|
| 1466 |
+
"step": 2080
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 3.5726495726495724,
|
| 1470 |
+
"grad_norm": 0.14336660504341125,
|
| 1471 |
+
"learning_rate": 2.0206877747561828e-05,
|
| 1472 |
+
"loss": 0.0216,
|
| 1473 |
+
"step": 2090
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 3.58974358974359,
|
| 1477 |
+
"grad_norm": 0.15480872988700867,
|
| 1478 |
+
"learning_rate": 2.011372523790927e-05,
|
| 1479 |
+
"loss": 0.02,
|
| 1480 |
+
"step": 2100
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 3.606837606837607,
|
| 1484 |
+
"grad_norm": 0.15778888761997223,
|
| 1485 |
+
"learning_rate": 2.0020349037348643e-05,
|
| 1486 |
+
"loss": 0.0178,
|
| 1487 |
+
"step": 2110
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 3.623931623931624,
|
| 1491 |
+
"grad_norm": 0.1326507329940796,
|
| 1492 |
+
"learning_rate": 1.9926753230457443e-05,
|
| 1493 |
+
"loss": 0.0193,
|
| 1494 |
+
"step": 2120
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 3.641025641025641,
|
| 1498 |
+
"grad_norm": 0.1321679651737213,
|
| 1499 |
+
"learning_rate": 1.983294191141948e-05,
|
| 1500 |
+
"loss": 0.0214,
|
| 1501 |
+
"step": 2130
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 3.658119658119658,
|
| 1505 |
+
"grad_norm": 0.12508369982242584,
|
| 1506 |
+
"learning_rate": 1.973891918384575e-05,
|
| 1507 |
+
"loss": 0.0167,
|
| 1508 |
+
"step": 2140
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 3.6752136752136755,
|
| 1512 |
+
"grad_norm": 0.17087364196777344,
|
| 1513 |
+
"learning_rate": 1.9644689160594945e-05,
|
| 1514 |
+
"loss": 0.0199,
|
| 1515 |
+
"step": 2150
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 3.6923076923076925,
|
| 1519 |
+
"grad_norm": 0.12470774352550507,
|
| 1520 |
+
"learning_rate": 1.9550255963593553e-05,
|
| 1521 |
+
"loss": 0.0174,
|
| 1522 |
+
"step": 2160
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 3.7094017094017095,
|
| 1526 |
+
"grad_norm": 0.12396084517240524,
|
| 1527 |
+
"learning_rate": 1.9455623723655524e-05,
|
| 1528 |
+
"loss": 0.015,
|
| 1529 |
+
"step": 2170
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 3.7264957264957266,
|
| 1533 |
+
"grad_norm": 0.11109942197799683,
|
| 1534 |
+
"learning_rate": 1.9360796580301606e-05,
|
| 1535 |
+
"loss": 0.0177,
|
| 1536 |
+
"step": 2180
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 3.7435897435897436,
|
| 1540 |
+
"grad_norm": 0.12820954620838165,
|
| 1541 |
+
"learning_rate": 1.9265778681578245e-05,
|
| 1542 |
+
"loss": 0.0172,
|
| 1543 |
+
"step": 2190
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 3.7606837606837606,
|
| 1547 |
+
"grad_norm": 0.16764280200004578,
|
| 1548 |
+
"learning_rate": 1.9170574183876143e-05,
|
| 1549 |
+
"loss": 0.0213,
|
| 1550 |
+
"step": 2200
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 3.7777777777777777,
|
| 1554 |
+
"grad_norm": 0.13727493584156036,
|
| 1555 |
+
"learning_rate": 1.9075187251748452e-05,
|
| 1556 |
+
"loss": 0.0202,
|
| 1557 |
+
"step": 2210
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 3.7948717948717947,
|
| 1561 |
+
"grad_norm": 0.1443384438753128,
|
| 1562 |
+
"learning_rate": 1.8979622057728596e-05,
|
| 1563 |
+
"loss": 0.0191,
|
| 1564 |
+
"step": 2220
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 3.8119658119658117,
|
| 1568 |
+
"grad_norm": 0.13722945749759674,
|
| 1569 |
+
"learning_rate": 1.8883882782147748e-05,
|
| 1570 |
+
"loss": 0.0189,
|
| 1571 |
+
"step": 2230
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 3.8290598290598292,
|
| 1575 |
+
"grad_norm": 0.15205898880958557,
|
| 1576 |
+
"learning_rate": 1.8787973612951967e-05,
|
| 1577 |
+
"loss": 0.0178,
|
| 1578 |
+
"step": 2240
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 3.8461538461538463,
|
| 1582 |
+
"grad_norm": 0.1548609435558319,
|
| 1583 |
+
"learning_rate": 1.8691898745519028e-05,
|
| 1584 |
+
"loss": 0.0194,
|
| 1585 |
+
"step": 2250
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 3.8632478632478633,
|
| 1589 |
+
"grad_norm": 0.1382390409708023,
|
| 1590 |
+
"learning_rate": 1.8595662382474863e-05,
|
| 1591 |
+
"loss": 0.0195,
|
| 1592 |
+
"step": 2260
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 3.8803418803418803,
|
| 1596 |
+
"grad_norm": 0.12302783131599426,
|
| 1597 |
+
"learning_rate": 1.8499268733509766e-05,
|
| 1598 |
+
"loss": 0.0187,
|
| 1599 |
+
"step": 2270
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 3.8974358974358974,
|
| 1603 |
+
"grad_norm": 0.1820244938135147,
|
| 1604 |
+
"learning_rate": 1.84027220151942e-05,
|
| 1605 |
+
"loss": 0.0165,
|
| 1606 |
+
"step": 2280
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 3.9145299145299144,
|
| 1610 |
+
"grad_norm": 0.1326133906841278,
|
| 1611 |
+
"learning_rate": 1.83060264507944e-05,
|
| 1612 |
+
"loss": 0.017,
|
| 1613 |
+
"step": 2290
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 3.931623931623932,
|
| 1617 |
+
"grad_norm": 0.12418877333402634,
|
| 1618 |
+
"learning_rate": 1.82091862700876e-05,
|
| 1619 |
+
"loss": 0.0173,
|
| 1620 |
+
"step": 2300
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 3.948717948717949,
|
| 1624 |
+
"grad_norm": 0.12344951182603836,
|
| 1625 |
+
"learning_rate": 1.8112205709177023e-05,
|
| 1626 |
+
"loss": 0.0204,
|
| 1627 |
+
"step": 2310
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 3.965811965811966,
|
| 1631 |
+
"grad_norm": 0.20541010797023773,
|
| 1632 |
+
"learning_rate": 1.8015089010306585e-05,
|
| 1633 |
+
"loss": 0.0169,
|
| 1634 |
+
"step": 2320
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 3.982905982905983,
|
| 1638 |
+
"grad_norm": 0.20820003747940063,
|
| 1639 |
+
"learning_rate": 1.7917840421675298e-05,
|
| 1640 |
+
"loss": 0.0222,
|
| 1641 |
+
"step": 2330
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 4.0,
|
| 1645 |
+
"grad_norm": 0.12587814033031464,
|
| 1646 |
+
"learning_rate": 1.7820464197251474e-05,
|
| 1647 |
+
"loss": 0.0168,
|
| 1648 |
+
"step": 2340
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 4.017094017094017,
|
| 1652 |
+
"grad_norm": 0.12032312154769897,
|
| 1653 |
+
"learning_rate": 1.7722964596586627e-05,
|
| 1654 |
+
"loss": 0.02,
|
| 1655 |
+
"step": 2350
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 4.034188034188034,
|
| 1659 |
+
"grad_norm": 0.16112498939037323,
|
| 1660 |
+
"learning_rate": 1.7625345884629144e-05,
|
| 1661 |
+
"loss": 0.0173,
|
| 1662 |
+
"step": 2360
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 4.051282051282051,
|
| 1666 |
+
"grad_norm": 0.11967416107654572,
|
| 1667 |
+
"learning_rate": 1.7527612331537722e-05,
|
| 1668 |
+
"loss": 0.0175,
|
| 1669 |
+
"step": 2370
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 4.068376068376068,
|
| 1673 |
+
"grad_norm": 0.12133967131376266,
|
| 1674 |
+
"learning_rate": 1.7429768212494593e-05,
|
| 1675 |
+
"loss": 0.0179,
|
| 1676 |
+
"step": 2380
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 4.085470085470085,
|
| 1680 |
+
"grad_norm": 0.16594752669334412,
|
| 1681 |
+
"learning_rate": 1.7331817807518497e-05,
|
| 1682 |
+
"loss": 0.0195,
|
| 1683 |
+
"step": 2390
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 4.102564102564102,
|
| 1687 |
+
"grad_norm": 0.15331916511058807,
|
| 1688 |
+
"learning_rate": 1.723376540127745e-05,
|
| 1689 |
+
"loss": 0.0156,
|
| 1690 |
+
"step": 2400
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 4.119658119658119,
|
| 1694 |
+
"grad_norm": 0.13228918612003326,
|
| 1695 |
+
"learning_rate": 1.7135615282901357e-05,
|
| 1696 |
+
"loss": 0.0169,
|
| 1697 |
+
"step": 2410
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 4.136752136752137,
|
| 1701 |
+
"grad_norm": 0.13137143850326538,
|
| 1702 |
+
"learning_rate": 1.7037371745794358e-05,
|
| 1703 |
+
"loss": 0.0221,
|
| 1704 |
+
"step": 2420
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 4.153846153846154,
|
| 1708 |
+
"grad_norm": 0.1315498948097229,
|
| 1709 |
+
"learning_rate": 1.6939039087447026e-05,
|
| 1710 |
+
"loss": 0.0186,
|
| 1711 |
+
"step": 2430
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 4.170940170940171,
|
| 1715 |
+
"grad_norm": 0.1358824223279953,
|
| 1716 |
+
"learning_rate": 1.6840621609248397e-05,
|
| 1717 |
+
"loss": 0.0182,
|
| 1718 |
+
"step": 2440
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 4.188034188034188,
|
| 1722 |
+
"grad_norm": 0.13738547265529633,
|
| 1723 |
+
"learning_rate": 1.6742123616297794e-05,
|
| 1724 |
+
"loss": 0.0167,
|
| 1725 |
+
"step": 2450
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 4.205128205128205,
|
| 1729 |
+
"grad_norm": 0.12346506863832474,
|
| 1730 |
+
"learning_rate": 1.664354941721651e-05,
|
| 1731 |
+
"loss": 0.0162,
|
| 1732 |
+
"step": 2460
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 4.222222222222222,
|
| 1736 |
+
"grad_norm": 0.13007137179374695,
|
| 1737 |
+
"learning_rate": 1.6544903323959355e-05,
|
| 1738 |
+
"loss": 0.0166,
|
| 1739 |
+
"step": 2470
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 4.239316239316239,
|
| 1743 |
+
"grad_norm": 0.13907431066036224,
|
| 1744 |
+
"learning_rate": 1.6446189651626023e-05,
|
| 1745 |
+
"loss": 0.0174,
|
| 1746 |
+
"step": 2480
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 4.256410256410256,
|
| 1750 |
+
"grad_norm": 0.12497235089540482,
|
| 1751 |
+
"learning_rate": 1.6347412718272316e-05,
|
| 1752 |
+
"loss": 0.0193,
|
| 1753 |
+
"step": 2490
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 4.273504273504273,
|
| 1757 |
+
"grad_norm": 0.13083861768245697,
|
| 1758 |
+
"learning_rate": 1.624857684472131e-05,
|
| 1759 |
+
"loss": 0.0168,
|
| 1760 |
+
"step": 2500
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 4.2905982905982905,
|
| 1764 |
+
"grad_norm": 0.13517604768276215,
|
| 1765 |
+
"learning_rate": 1.6149686354374283e-05,
|
| 1766 |
+
"loss": 0.0197,
|
| 1767 |
+
"step": 2510
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 4.3076923076923075,
|
| 1771 |
+
"grad_norm": 0.15625643730163574,
|
| 1772 |
+
"learning_rate": 1.605074557302165e-05,
|
| 1773 |
+
"loss": 0.0177,
|
| 1774 |
+
"step": 2520
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 4.3247863247863245,
|
| 1778 |
+
"grad_norm": 0.20975004136562347,
|
| 1779 |
+
"learning_rate": 1.59517588286537e-05,
|
| 1780 |
+
"loss": 0.0189,
|
| 1781 |
+
"step": 2530
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 4.3418803418803416,
|
| 1785 |
+
"grad_norm": 0.16902174055576324,
|
| 1786 |
+
"learning_rate": 1.5852730451271324e-05,
|
| 1787 |
+
"loss": 0.0164,
|
| 1788 |
+
"step": 2540
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 4.358974358974359,
|
| 1792 |
+
"grad_norm": 0.14469504356384277,
|
| 1793 |
+
"learning_rate": 1.5753664772696546e-05,
|
| 1794 |
+
"loss": 0.0186,
|
| 1795 |
+
"step": 2550
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 4.3760683760683765,
|
| 1799 |
+
"grad_norm": 0.14689213037490845,
|
| 1800 |
+
"learning_rate": 1.565456612638307e-05,
|
| 1801 |
+
"loss": 0.0203,
|
| 1802 |
+
"step": 2560
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 4.3931623931623935,
|
| 1806 |
+
"grad_norm": 0.13142313063144684,
|
| 1807 |
+
"learning_rate": 1.5555438847226732e-05,
|
| 1808 |
+
"loss": 0.0163,
|
| 1809 |
+
"step": 2570
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 4.410256410256411,
|
| 1813 |
+
"grad_norm": 0.15418632328510284,
|
| 1814 |
+
"learning_rate": 1.545628727137584e-05,
|
| 1815 |
+
"loss": 0.0178,
|
| 1816 |
+
"step": 2580
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 4.427350427350428,
|
| 1820 |
+
"grad_norm": 0.15956638753414154,
|
| 1821 |
+
"learning_rate": 1.535711573604153e-05,
|
| 1822 |
+
"loss": 0.0144,
|
| 1823 |
+
"step": 2590
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 4.444444444444445,
|
| 1827 |
+
"grad_norm": 0.1449543535709381,
|
| 1828 |
+
"learning_rate": 1.5257928579308032e-05,
|
| 1829 |
+
"loss": 0.0177,
|
| 1830 |
+
"step": 2600
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 4.461538461538462,
|
| 1834 |
+
"grad_norm": 0.1384369283914566,
|
| 1835 |
+
"learning_rate": 1.5158730139942902e-05,
|
| 1836 |
+
"loss": 0.0174,
|
| 1837 |
+
"step": 2610
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 4.478632478632479,
|
| 1841 |
+
"grad_norm": 0.13507409393787384,
|
| 1842 |
+
"learning_rate": 1.5059524757207239e-05,
|
| 1843 |
+
"loss": 0.0173,
|
| 1844 |
+
"step": 2620
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 4.495726495726496,
|
| 1848 |
+
"grad_norm": 0.14348851144313812,
|
| 1849 |
+
"learning_rate": 1.4960316770665866e-05,
|
| 1850 |
+
"loss": 0.017,
|
| 1851 |
+
"step": 2630
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 4.512820512820513,
|
| 1855 |
+
"grad_norm": 0.10914012789726257,
|
| 1856 |
+
"learning_rate": 1.4861110519997495e-05,
|
| 1857 |
+
"loss": 0.0178,
|
| 1858 |
+
"step": 2640
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 4.52991452991453,
|
| 1862 |
+
"grad_norm": 0.21071846783161163,
|
| 1863 |
+
"learning_rate": 1.4761910344804913e-05,
|
| 1864 |
+
"loss": 0.0168,
|
| 1865 |
+
"step": 2650
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 4.547008547008547,
|
| 1869 |
+
"grad_norm": 0.11937859654426575,
|
| 1870 |
+
"learning_rate": 1.466272058442516e-05,
|
| 1871 |
+
"loss": 0.0171,
|
| 1872 |
+
"step": 2660
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 4.564102564102564,
|
| 1876 |
+
"grad_norm": 0.09995169937610626,
|
| 1877 |
+
"learning_rate": 1.4563545577739676e-05,
|
| 1878 |
+
"loss": 0.0174,
|
| 1879 |
+
"step": 2670
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 4.581196581196581,
|
| 1883 |
+
"grad_norm": 0.1595757007598877,
|
| 1884 |
+
"learning_rate": 1.4464389662984543e-05,
|
| 1885 |
+
"loss": 0.0173,
|
| 1886 |
+
"step": 2680
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 4.598290598290598,
|
| 1890 |
+
"grad_norm": 0.13326671719551086,
|
| 1891 |
+
"learning_rate": 1.4365257177560684e-05,
|
| 1892 |
+
"loss": 0.0177,
|
| 1893 |
+
"step": 2690
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 4.615384615384615,
|
| 1897 |
+
"grad_norm": 0.1444454789161682,
|
| 1898 |
+
"learning_rate": 1.426615245784416e-05,
|
| 1899 |
+
"loss": 0.0166,
|
| 1900 |
+
"step": 2700
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 4.632478632478632,
|
| 1904 |
+
"grad_norm": 0.11446025967597961,
|
| 1905 |
+
"learning_rate": 1.4167079838996468e-05,
|
| 1906 |
+
"loss": 0.0151,
|
| 1907 |
+
"step": 2710
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 4.64957264957265,
|
| 1911 |
+
"grad_norm": 0.13881082832813263,
|
| 1912 |
+
"learning_rate": 1.4068043654774897e-05,
|
| 1913 |
+
"loss": 0.0155,
|
| 1914 |
+
"step": 2720
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 4.666666666666667,
|
| 1918 |
+
"grad_norm": 0.20510493218898773,
|
| 1919 |
+
"learning_rate": 1.3969048237342992e-05,
|
| 1920 |
+
"loss": 0.0185,
|
| 1921 |
+
"step": 2730
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 4.683760683760684,
|
| 1925 |
+
"grad_norm": 0.14961980283260345,
|
| 1926 |
+
"learning_rate": 1.3870097917081011e-05,
|
| 1927 |
+
"loss": 0.0166,
|
| 1928 |
+
"step": 2740
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 4.700854700854701,
|
| 1932 |
+
"grad_norm": 0.12033514678478241,
|
| 1933 |
+
"learning_rate": 1.3771197022396508e-05,
|
| 1934 |
+
"loss": 0.0146,
|
| 1935 |
+
"step": 2750
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 4.717948717948718,
|
| 1939 |
+
"grad_norm": 0.13812503218650818,
|
| 1940 |
+
"learning_rate": 1.3672349879535015e-05,
|
| 1941 |
+
"loss": 0.0172,
|
| 1942 |
+
"step": 2760
|
| 1943 |
+
},
|
| 1944 |
+
{
|
| 1945 |
+
"epoch": 4.735042735042735,
|
| 1946 |
+
"grad_norm": 0.13671137392520905,
|
| 1947 |
+
"learning_rate": 1.3573560812390781e-05,
|
| 1948 |
+
"loss": 0.0177,
|
| 1949 |
+
"step": 2770
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 4.752136752136752,
|
| 1953 |
+
"grad_norm": 0.14928513765335083,
|
| 1954 |
+
"learning_rate": 1.3474834142317635e-05,
|
| 1955 |
+
"loss": 0.0142,
|
| 1956 |
+
"step": 2780
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 4.769230769230769,
|
| 1960 |
+
"grad_norm": 0.10949765145778656,
|
| 1961 |
+
"learning_rate": 1.3376174187939955e-05,
|
| 1962 |
+
"loss": 0.0161,
|
| 1963 |
+
"step": 2790
|
| 1964 |
+
},
|
| 1965 |
+
{
|
| 1966 |
+
"epoch": 4.786324786324786,
|
| 1967 |
+
"grad_norm": 0.10777349025011063,
|
| 1968 |
+
"learning_rate": 1.3277585264963761e-05,
|
| 1969 |
+
"loss": 0.0168,
|
| 1970 |
+
"step": 2800
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 4.803418803418803,
|
| 1974 |
+
"grad_norm": 0.17750239372253418,
|
| 1975 |
+
"learning_rate": 1.3179071685987928e-05,
|
| 1976 |
+
"loss": 0.0182,
|
| 1977 |
+
"step": 2810
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 4.82051282051282,
|
| 1981 |
+
"grad_norm": 0.13401111960411072,
|
| 1982 |
+
"learning_rate": 1.308063776031554e-05,
|
| 1983 |
+
"loss": 0.0164,
|
| 1984 |
+
"step": 2820
|
| 1985 |
+
},
|
| 1986 |
+
{
|
| 1987 |
+
"epoch": 4.837606837606837,
|
| 1988 |
+
"grad_norm": 0.1763419806957245,
|
| 1989 |
+
"learning_rate": 1.2982287793765389e-05,
|
| 1990 |
+
"loss": 0.0147,
|
| 1991 |
+
"step": 2830
|
| 1992 |
+
},
|
| 1993 |
+
{
|
| 1994 |
+
"epoch": 4.854700854700854,
|
| 1995 |
+
"grad_norm": 0.13647712767124176,
|
| 1996 |
+
"learning_rate": 1.2884026088483616e-05,
|
| 1997 |
+
"loss": 0.0162,
|
| 1998 |
+
"step": 2840
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 4.871794871794872,
|
| 2002 |
+
"grad_norm": 0.1311454325914383,
|
| 2003 |
+
"learning_rate": 1.2785856942755543e-05,
|
| 2004 |
+
"loss": 0.0174,
|
| 2005 |
+
"step": 2850
|
| 2006 |
+
},
|
| 2007 |
+
{
|
| 2008 |
+
"epoch": 4.888888888888889,
|
| 2009 |
+
"grad_norm": 0.12965312600135803,
|
| 2010 |
+
"learning_rate": 1.2687784650817616e-05,
|
| 2011 |
+
"loss": 0.0169,
|
| 2012 |
+
"step": 2860
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 4.905982905982906,
|
| 2016 |
+
"grad_norm": 0.14787934720516205,
|
| 2017 |
+
"learning_rate": 1.2589813502669603e-05,
|
| 2018 |
+
"loss": 0.0193,
|
| 2019 |
+
"step": 2870
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 4.923076923076923,
|
| 2023 |
+
"grad_norm": 0.1825832575559616,
|
| 2024 |
+
"learning_rate": 1.2491947783886907e-05,
|
| 2025 |
+
"loss": 0.0169,
|
| 2026 |
+
"step": 2880
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 4.94017094017094,
|
| 2030 |
+
"grad_norm": 0.1275591403245926,
|
| 2031 |
+
"learning_rate": 1.2394191775433106e-05,
|
| 2032 |
+
"loss": 0.017,
|
| 2033 |
+
"step": 2890
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 4.957264957264957,
|
| 2037 |
+
"grad_norm": 0.12773366272449493,
|
| 2038 |
+
"learning_rate": 1.2296549753472696e-05,
|
| 2039 |
+
"loss": 0.0176,
|
| 2040 |
+
"step": 2900
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 4.9743589743589745,
|
| 2044 |
+
"grad_norm": 0.15548300743103027,
|
| 2045 |
+
"learning_rate": 1.2199025989184032e-05,
|
| 2046 |
+
"loss": 0.0141,
|
| 2047 |
+
"step": 2910
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 4.9914529914529915,
|
| 2051 |
+
"grad_norm": 0.12168218195438385,
|
| 2052 |
+
"learning_rate": 1.21016247485725e-05,
|
| 2053 |
+
"loss": 0.0179,
|
| 2054 |
+
"step": 2920
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 5.0085470085470085,
|
| 2058 |
+
"grad_norm": 0.12161333858966827,
|
| 2059 |
+
"learning_rate": 1.2004350292283897e-05,
|
| 2060 |
+
"loss": 0.0171,
|
| 2061 |
+
"step": 2930
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 5.0256410256410255,
|
| 2065 |
+
"grad_norm": 0.2516013979911804,
|
| 2066 |
+
"learning_rate": 1.1907206875418068e-05,
|
| 2067 |
+
"loss": 0.0165,
|
| 2068 |
+
"step": 2940
|
| 2069 |
+
},
|
| 2070 |
+
{
|
| 2071 |
+
"epoch": 5.042735042735043,
|
| 2072 |
+
"grad_norm": 0.16508907079696655,
|
| 2073 |
+
"learning_rate": 1.1810198747342769e-05,
|
| 2074 |
+
"loss": 0.0186,
|
| 2075 |
+
"step": 2950
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 5.05982905982906,
|
| 2079 |
+
"grad_norm": 0.09363128989934921,
|
| 2080 |
+
"learning_rate": 1.1713330151507786e-05,
|
| 2081 |
+
"loss": 0.0145,
|
| 2082 |
+
"step": 2960
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"epoch": 5.076923076923077,
|
| 2086 |
+
"grad_norm": 0.1175096333026886,
|
| 2087 |
+
"learning_rate": 1.161660532525931e-05,
|
| 2088 |
+
"loss": 0.016,
|
| 2089 |
+
"step": 2970
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"epoch": 5.094017094017094,
|
| 2093 |
+
"grad_norm": 0.18064428865909576,
|
| 2094 |
+
"learning_rate": 1.1520028499654584e-05,
|
| 2095 |
+
"loss": 0.0147,
|
| 2096 |
+
"step": 2980
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 5.111111111111111,
|
| 2100 |
+
"grad_norm": 0.12403568625450134,
|
| 2101 |
+
"learning_rate": 1.1423603899276833e-05,
|
| 2102 |
+
"loss": 0.0144,
|
| 2103 |
+
"step": 2990
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 5.128205128205128,
|
| 2107 |
+
"grad_norm": 0.2672509551048279,
|
| 2108 |
+
"learning_rate": 1.1327335742050442e-05,
|
| 2109 |
+
"loss": 0.016,
|
| 2110 |
+
"step": 3000
|
| 2111 |
+
}
|
| 2112 |
+
],
|
| 2113 |
+
"logging_steps": 10,
|
| 2114 |
+
"max_steps": 5000,
|
| 2115 |
+
"num_input_tokens_seen": 0,
|
| 2116 |
+
"num_train_epochs": 9,
|
| 2117 |
+
"save_steps": 5001,
|
| 2118 |
+
"stateful_callbacks": {
|
| 2119 |
+
"TrainerControl": {
|
| 2120 |
+
"args": {
|
| 2121 |
+
"should_epoch_stop": false,
|
| 2122 |
+
"should_evaluate": false,
|
| 2123 |
+
"should_log": false,
|
| 2124 |
+
"should_save": true,
|
| 2125 |
+
"should_training_stop": false
|
| 2126 |
+
},
|
| 2127 |
+
"attributes": {}
|
| 2128 |
+
}
|
| 2129 |
+
},
|
| 2130 |
+
"total_flos": 0.0,
|
| 2131 |
+
"train_batch_size": 64,
|
| 2132 |
+
"trial_name": null,
|
| 2133 |
+
"trial_params": null
|
| 2134 |
+
}
|
FewShot/39tasks/5shot/FromBaseline26_Processingline_5k/seed0/default/train_5shot_seed0_processingline_5k/checkpoint-1000/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9fc9945eca28b94cc6398098b3d2092b0e9731d322853d46197bf3dc7a37a961
|
| 3 |
+
size 4999367032
|
FewShot/39tasks/5shot/FromBaseline26_Processingline_5k/seed0/default/train_5shot_seed0_processingline_5k/checkpoint-1000/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:31ed3ffe024dd151c65eb826083fb58083fbf503901ac040459c44795fb46581
|
| 3 |
+
size 2586705312
|
FewShot/39tasks/5shot/FromBaseline26_Processingline_5k/seed0/default/train_5shot_seed0_processingline_5k/checkpoint-1000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e782ba28ffec111c690d45ea5de37d10ecd8f620fb09fe4c2cd13af5b48280cf
|
| 3 |
+
size 1611553483
|
FewShot/39tasks/5shot/FromBaseline26_Processingline_5k/seed0/default/train_5shot_seed0_processingline_5k/checkpoint-1000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0361e2999e6ebf69b5475734034a7149e063bab21bdd95d90fc2dfbb338e183b
|
| 3 |
+
size 14645
|
FewShot/39tasks/5shot/FromBaseline26_Processingline_5k/seed0/default/train_5shot_seed0_processingline_5k/checkpoint-1000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6e19ead5779e3f8f2f666b7793de3a03d118f57f1b16a97eeb87bd732d18b445
|
| 3 |
+
size 1465
|