Hugging Face's logo

LucaFrat
/
groot-bs64

Robotics
Safetensors
Gr00tN1d7
gr00t
isaac
humanoid
Model card Files
xet
 Community
LucaFrat commited on
Commit
ddda7ed
·
verified ·
1 Parent(s): 56e7aa4

Upload folder using huggingface_hub

Browse files
Files changed (19) hide show
  1. checkpoint-6000/config.json +88 -0
  2. checkpoint-6000/embodiment_id.json +57 -0
  3. checkpoint-6000/experiment_cfg/conf.yaml +251 -0
  4. checkpoint-6000/experiment_cfg/config.yaml +277 -0
  5. checkpoint-6000/experiment_cfg/dataset_statistics.json +907 -0
  6. checkpoint-6000/experiment_cfg/final_model_config.json +55 -0
  7. checkpoint-6000/experiment_cfg/final_processor_config.json +0 -0
  8. checkpoint-6000/model-00001-of-00003.safetensors +3 -0
  9. checkpoint-6000/model-00002-of-00003.safetensors +3 -0
  10. checkpoint-6000/model-00003-of-00003.safetensors +3 -0
  11. checkpoint-6000/model.safetensors.index.json +0 -0
  12. checkpoint-6000/optimizer.pt +3 -0
  13. checkpoint-6000/processor_config.json +1159 -0
  14. checkpoint-6000/rng_state.pth +3 -0
  15. checkpoint-6000/scheduler.pt +3 -0
  16. checkpoint-6000/statistics.json +0 -0
  17. checkpoint-6000/trainer_state.json +3634 -0
  18. checkpoint-6000/training_args.bin +3 -0
  19. checkpoint-6000/wandb_config.json +1 -0
checkpoint-6000/config.json ADDED
@@ -0,0 +1,88 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "action_horizon": 40,
3
+ "add_pos_embed": true,
4
+ "apply_sincos_state_encoding": false,
5
+ "architectures": [
6
+ "Gr00tN1d7"
7
+ ],
8
+ "attn_dropout": 0.2,
9
+ "attn_implementation": null,
10
+ "backbone_embedding_dim": 2048,
11
+ "backbone_trainable_params_fp32": true,
12
+ "color_jitter_params": {
13
+ "brightness": 0.3,
14
+ "contrast": 0.4,
15
+ "hue": 0.08,
16
+ "saturation": 0.5
17
+ },
18
+ "crop_fraction": 0.95,
19
+ "diffusion_model_cfg": {
20
+ "attention_head_dim": 48,
21
+ "dropout": 0.2,
22
+ "final_dropout": true,
23
+ "interleave_self_attention": true,
24
+ "norm_type": "ada_norm",
25
+ "num_attention_heads": 32,
26
+ "num_layers": 32,
27
+ "output_dim": 1024,
28
+ "positional_embeddings": null
29
+ },
30
+ "dtype": "float32",
31
+ "exclude_state": false,
32
+ "formalize_language": true,
33
+ "hidden_size": 1024,
34
+ "image_crop_size": [
35
+ 230,
36
+ 230
37
+ ],
38
+ "image_target_size": [
39
+ 256,
40
+ 256
41
+ ],
42
+ "letter_box_transform": false,
43
+ "load_bf16": false,
44
+ "max_action_dim": 132,
45
+ "max_num_embodiments": 32,
46
+ "max_seq_len": 1024,
47
+ "max_state_dim": 132,
48
+ "model_dtype": "bfloat16",
49
+ "model_name": "nvidia/Cosmos-Reason2-2B",
50
+ "model_type": "Gr00tN1d7",
51
+ "noise_beta_alpha": 1.5,
52
+ "noise_beta_beta": 1.0,
53
+ "noise_s": 0.999,
54
+ "num_inference_timesteps": 4,
55
+ "num_timestep_buckets": 1000,
56
+ "random_history_crop": true,
57
+ "random_rotation_angle": 0,
58
+ "reproject_vision": false,
59
+ "rtc_ramp_rate": 6.0,
60
+ "select_layer": 16,
61
+ "shortest_image_edge": 256,
62
+ "state_dropout_prob": 0.2,
63
+ "state_gaussian_noise_std": 0.0,
64
+ "transformers_version": "4.57.3",
65
+ "tune_diffusion_model": true,
66
+ "tune_linear": true,
67
+ "tune_llm": false,
68
+ "tune_projector": true,
69
+ "tune_top_llm_layers": 0,
70
+ "tune_visual": false,
71
+ "tune_vlln": true,
72
+ "use_albumentations": true,
73
+ "use_alternate_vl_dit": true,
74
+ "use_flash_attention": true,
75
+ "use_future_tokens": false,
76
+ "use_mean_std": false,
77
+ "use_percentiles": true,
78
+ "use_vl_self_attention": true,
79
+ "use_vlln": true,
80
+ "vl_self_attention_cfg": {
81
+ "attention_head_dim": 64,
82
+ "dropout": 0.2,
83
+ "final_dropout": true,
84
+ "num_attention_heads": 32,
85
+ "num_layers": 4,
86
+ "positional_embeddings": null
87
+ }
88
+ }
checkpoint-6000/embodiment_id.json ADDED
@@ -0,0 +1,57 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "robocasa_panda_omron": 13,
3
+ "oxe_droid": 17,
4
+ "oxe_fractal": 18,
5
+ "oxe_language_table": 19,
6
+ "oxe_bridge": 20,
7
+ "unknown": 22,
8
+ "gr1_unified": 20,
9
+ "agibot": 26,
10
+ "sim_behavior_r1_pro": 23,
11
+ "xdof": 24,
12
+ "xdof_oss_data": 25,
13
+ "unitree_g1_full_body_with_waist_height_nav_cmd": 25,
14
+ "real_r1_pro_sharpa": 27,
15
+ "real_r1_pro_sharpa_add_view": 27,
16
+ "real_r1_pro_sharpa_relative_arm_joint": 26,
17
+ "real_r1_pro_sharpa_delta_eef": 26,
18
+ "real_r1_pro_sharpa_absolute_eef": 26,
19
+ "real_r1_pro_sharpa_meanstd": 26,
20
+ "real_r1_pro_sharpa_relative_eef": 26,
21
+ "real_r1_pro_sharpa_relative_eef_add_view": 26,
22
+ "real_r1_pro_sharpa_relative_eef_relative_hand": 26,
23
+ "real_r1_pro_sharpa_relative_eef_human": 26,
24
+ "real_r1_pro_sharpa_relative_eef_human_add_view": 26,
25
+ "real_r1_pro_sharpa_relative_eef_human_relative_hand": 26,
26
+ "real_r1_pro_sharpa_relative_eef_egodex": 26,
27
+ "real_r1_pro_sharpa_relative_eef_egodex_relative_hand": 26,
28
+ "real_r1_pro_sharpa_relative_eef_egodex_wrist_only": 26,
29
+ "real_r1_pro_sharpa_relative_eef_maxinsights": 26,
30
+ "real_r1_pro_sharpa_relative_eef_maxinsights_relative_hand": 26,
31
+ "real_r1_pro_sharpa_relative_eef_mecka": 26,
32
+ "real_r1_pro_sharpa_relative_eef_mecka_relative_hand": 26,
33
+ "real_g1_relative_eef_absolute_joints": 25,
34
+ "real_g1_relative_eef_absolute_joints_wrist_cam": 25,
35
+ "real_g1_relative_eef_relative_joints": 25,
36
+ "real_r1_pro_sharpa_relative_eef_relative_hand_relative_joint": 26,
37
+ "real_r1_pro_sharpa_relative_joint": 29,
38
+ "oxe_droid_relative_eef_relative_joint": 24,
39
+ "oxe_droid_relative_eef_relative_joint_swapped": 24,
40
+ "oxe_droid_relative_eef_relative_joint_upweight_z": 24,
41
+ "oxe_droid_relative_eef_relative_joint_upweight_z_swapped": 24,
42
+ "oxe_droid_relative_eef_relative_joint_3view": 24,
43
+ "oxe_droid_relative_eef_relative_joint_3view_swapped": 24,
44
+ "oxe_droid_relative_eef": 24,
45
+ "oxe_droid_joint_position_relative": 24,
46
+ "xdof_relative_eef_relative_joint": 27,
47
+ "xdof_relative_eef_relative_joint_subtask": 27,
48
+ "xdof_relative_eef": 27,
49
+ "xdof_relative_joint": 28,
50
+ "simpler_env_google": 0,
51
+ "simpler_env_widowx": 1,
52
+ "libero_sim": 2,
53
+ "droid_sim": 3,
54
+ "unitree_g1_sonic": 11,
55
+ "new_embodiment": 10,
56
+ "robocasa_gr1_tabletop": 10
57
+ }
checkpoint-6000/experiment_cfg/conf.yaml ADDED
@@ -0,0 +1,251 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ load_config_path: null
2
+ model:
3
+ model_type: Gr00tN1d7
4
+ model_dtype: bfloat16
5
+ model_name: nvidia/Cosmos-Reason2-2B
6
+ backbone_model_type: qwen
7
+ model_revision: null
8
+ tune_top_llm_layers: 0
9
+ backbone_embedding_dim: 2048
10
+ tune_llm: false
11
+ tune_visual: false
12
+ select_layer: 12
13
+ reproject_vision: false
14
+ use_flash_attention: true
15
+ load_bf16: false
16
+ backbone_trainable_params_fp32: true
17
+ image_crop_size:
18
+ - 230
19
+ - 230
20
+ image_target_size:
21
+ - 256
22
+ - 256
23
+ shortest_image_edge: null
24
+ crop_fraction: null
25
+ random_rotation_angle: null
26
+ color_jitter_params:
27
+ brightness: 0.3
28
+ contrast: 0.4
29
+ saturation: 0.5
30
+ hue: 0.08
31
+ use_albumentations_transforms: true
32
+ extra_augmentation_config: null
33
+ formalize_language: true
34
+ apply_sincos_state_encoding: false
35
+ use_percentiles: true
36
+ use_relative_action: true
37
+ max_state_dim: 132
38
+ max_action_dim: 132
39
+ action_horizon: 40
40
+ hidden_size: 1024
41
+ input_embedding_dim: 1536
42
+ state_history_length: 1
43
+ add_pos_embed: true
44
+ attn_dropout: 0.2
45
+ use_vlln: true
46
+ max_seq_len: 1024
47
+ use_alternate_vl_dit: true
48
+ attend_text_every_n_blocks: 2
49
+ diffusion_model_cfg:
50
+ positional_embeddings: null
51
+ num_layers: 16
52
+ num_attention_heads: 32
53
+ attention_head_dim: 48
54
+ norm_type: ada_norm
55
+ dropout: 0.2
56
+ final_dropout: true
57
+ output_dim: 1024
58
+ interleave_self_attention: true
59
+ num_inference_timesteps: 4
60
+ noise_beta_alpha: 1.5
61
+ noise_beta_beta: 1.0
62
+ noise_s: 0.999
63
+ num_timestep_buckets: 1000
64
+ tune_projector: true
65
+ tune_diffusion_model: true
66
+ tune_vlln: true
67
+ state_dropout_prob: 0.2
68
+ exclude_state: false
69
+ use_mean_std: false
70
+ max_num_embodiments: 32
71
+ data:
72
+ datasets:
73
+ - dataset_paths:
74
+ - /home/ubuntu/groot-files/dataset
75
+ embodiment_tag: unitree_g1_sonic
76
+ mix_ratio: 1.0
77
+ dataset_type: physical_embodiment
78
+ val_dataset_path: null
79
+ modality_configs:
80
+ unitree_g1_sonic:
81
+ video:
82
+ delta_indices:
83
+ - 0
84
+ modality_keys:
85
+ - ego_view
86
+ sin_cos_embedding_keys: null
87
+ mean_std_embedding_keys: null
88
+ action_configs: null
89
+ state:
90
+ delta_indices:
91
+ - 0
92
+ modality_keys:
93
+ - left_leg
94
+ - right_leg
95
+ - waist
96
+ - left_arm
97
+ - right_arm
98
+ - left_hand
99
+ - right_hand
100
+ - projected_gravity
101
+ sin_cos_embedding_keys: null
102
+ mean_std_embedding_keys: null
103
+ action_configs: null
104
+ action:
105
+ delta_indices:
106
+ - 0
107
+ - 1
108
+ - 2
109
+ - 3
110
+ - 4
111
+ - 5
112
+ - 6
113
+ - 7
114
+ - 8
115
+ - 9
116
+ - 10
117
+ - 11
118
+ - 12
119
+ - 13
120
+ - 14
121
+ - 15
122
+ - 16
123
+ - 17
124
+ - 18
125
+ - 19
126
+ - 20
127
+ - 21
128
+ - 22
129
+ - 23
130
+ - 24
131
+ - 25
132
+ - 26
133
+ - 27
134
+ - 28
135
+ - 29
136
+ - 30
137
+ - 31
138
+ - 32
139
+ - 33
140
+ - 34
141
+ - 35
142
+ - 36
143
+ - 37
144
+ - 38
145
+ - 39
146
+ modality_keys:
147
+ - motion_token
148
+ - left_hand_joints
149
+ - right_hand_joints
150
+ sin_cos_embedding_keys: null
151
+ mean_std_embedding_keys: null
152
+ action_configs:
153
+ - rep: ABSOLUTE
154
+ type: NON_EEF
155
+ format: DEFAULT
156
+ state_key: null
157
+ - rep: ABSOLUTE
158
+ type: NON_EEF
159
+ format: DEFAULT
160
+ state_key: null
161
+ - rep: ABSOLUTE
162
+ type: NON_EEF
163
+ format: DEFAULT
164
+ state_key: null
165
+ language:
166
+ delta_indices:
167
+ - 0
168
+ modality_keys:
169
+ - annotation.human.task_description
170
+ sin_cos_embedding_keys: null
171
+ mean_std_embedding_keys: null
172
+ action_configs: null
173
+ download_cache: false
174
+ shard_size: 1024
175
+ episode_sampling_rate: 0.1
176
+ num_shards_per_epoch: 100000
177
+ override_pretraining_statistics: true
178
+ mode: single_turn
179
+ random_chop: 0.0
180
+ mock_dataset_mode: false
181
+ shuffle: true
182
+ seed: 42
183
+ multiprocessing_context: fork
184
+ allow_padding: false
185
+ subsample_ratio: 1.0
186
+ image_crop_size:
187
+ - 244
188
+ - 244
189
+ image_target_size:
190
+ - 224
191
+ - 224
192
+ video_backend: torchcodec
193
+ training:
194
+ output_dir: /home/ubuntu/groot-files/checkpoints/g1_finetune
195
+ experiment_name: null
196
+ max_steps: 15000
197
+ global_batch_size: 64
198
+ batch_size: null
199
+ gradient_accumulation_steps: 1
200
+ learning_rate: 0.0001
201
+ lr_scheduler_type: cosine
202
+ weight_decay: 1.0e-05
203
+ warmup_ratio: 0.05
204
+ warmup_steps: 0
205
+ max_grad_norm: 1.0
206
+ optim: adamw_torch
207
+ start_from_checkpoint: nvidia/GR00T-N1.7-3B
208
+ skip_weight_loading: false
209
+ tf32: true
210
+ fp16: false
211
+ bf16: true
212
+ eval_bf16: true
213
+ logging_steps: 10
214
+ save_steps: 3000
215
+ save_total_limit: 10
216
+ save_vl_model: false
217
+ save_only_model: false
218
+ upload_checkpoints: false
219
+ upload_every: 1000
220
+ upload_last_n_checkpoints: 5
221
+ max_concurrent_uploads: 2
222
+ eval_strategy: 'no'
223
+ eval_steps: 500
224
+ eval_set_split_ratio: 0.1
225
+ eval_batch_size: 2
226
+ save_best_eval_metric_name: ''
227
+ save_best_eval_metric_greater_is_better: true
228
+ deepspeed_stage: 2
229
+ gradient_checkpointing: false
230
+ transformers_trust_remote_code: true
231
+ transformers_local_files_only: false
232
+ transformers_cache_dir: null
233
+ transformers_access_token: null
234
+ use_ddp: false
235
+ ddp_bucket_cap_mb: 100
236
+ num_gpus: 1
237
+ dataloader_num_workers: 6
238
+ remove_unused_columns: false
239
+ use_wandb: true
240
+ wandb_project: groot-finetune
241
+ enable_profiling: false
242
+ max_retries: 3
243
+ assert_loss_less_than: null
244
+ add_rl_callback: false
245
+ enable_open_loop_eval: false
246
+ open_loop_eval_traj_ids:
247
+ - 0
248
+ open_loop_eval_steps_per_traj: 100
249
+ open_loop_eval_plot_indices: null
250
+ max_steps: 15000
251
+ save_steps: 3000
checkpoint-6000/experiment_cfg/config.yaml ADDED
@@ -0,0 +1,277 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ !!python/object:gr00t.configs.base_config.Config
2
+ data: !!python/object:gr00t.configs.data.data_config.DataConfig
3
+ allow_padding: false
4
+ datasets:
5
+ - !!python/object:gr00t.configs.data.data_config.SingleDatasetConfig
6
+ dataset_paths:
7
+ - /home/ubuntu/groot-files/dataset
8
+ dataset_type: physical_embodiment
9
+ embodiment_tag: unitree_g1_sonic
10
+ mix_ratio: 1.0
11
+ val_dataset_path: null
12
+ download_cache: false
13
+ episode_sampling_rate: 0.1
14
+ image_crop_size:
15
+ - 244
16
+ - 244
17
+ image_target_size:
18
+ - 224
19
+ - 224
20
+ mock_dataset_mode: false
21
+ modality_configs:
22
+ unitree_g1_sonic:
23
+ action: !!python/object:gr00t.data.types.ModalityConfig
24
+ action_configs:
25
+ - !!python/object:gr00t.data.types.ActionConfig
26
+ format: &id001 !!python/object/apply:gr00t.data.types.ActionFormat
27
+ - default
28
+ rep: &id002 !!python/object/apply:gr00t.data.types.ActionRepresentation
29
+ - absolute
30
+ state_key: null
31
+ type: &id003 !!python/object/apply:gr00t.data.types.ActionType
32
+ - non_eef
33
+ - !!python/object:gr00t.data.types.ActionConfig
34
+ format: *id001
35
+ rep: *id002
36
+ state_key: null
37
+ type: *id003
38
+ - !!python/object:gr00t.data.types.ActionConfig
39
+ format: *id001
40
+ rep: *id002
41
+ state_key: null
42
+ type: *id003
43
+ delta_indices:
44
+ - 0
45
+ - 1
46
+ - 2
47
+ - 3
48
+ - 4
49
+ - 5
50
+ - 6
51
+ - 7
52
+ - 8
53
+ - 9
54
+ - 10
55
+ - 11
56
+ - 12
57
+ - 13
58
+ - 14
59
+ - 15
60
+ - 16
61
+ - 17
62
+ - 18
63
+ - 19
64
+ - 20
65
+ - 21
66
+ - 22
67
+ - 23
68
+ - 24
69
+ - 25
70
+ - 26
71
+ - 27
72
+ - 28
73
+ - 29
74
+ - 30
75
+ - 31
76
+ - 32
77
+ - 33
78
+ - 34
79
+ - 35
80
+ - 36
81
+ - 37
82
+ - 38
83
+ - 39
84
+ mean_std_embedding_keys: null
85
+ modality_keys:
86
+ - motion_token
87
+ - left_hand_joints
88
+ - right_hand_joints
89
+ sin_cos_embedding_keys: null
90
+ language: !!python/object:gr00t.data.types.ModalityConfig
91
+ action_configs: null
92
+ delta_indices:
93
+ - 0
94
+ mean_std_embedding_keys: null
95
+ modality_keys:
96
+ - annotation.human.task_description
97
+ sin_cos_embedding_keys: null
98
+ state: !!python/object:gr00t.data.types.ModalityConfig
99
+ action_configs: null
100
+ delta_indices:
101
+ - 0
102
+ mean_std_embedding_keys: null
103
+ modality_keys:
104
+ - left_leg
105
+ - right_leg
106
+ - waist
107
+ - left_arm
108
+ - right_arm
109
+ - left_hand
110
+ - right_hand
111
+ - projected_gravity
112
+ sin_cos_embedding_keys: null
113
+ video: !!python/object:gr00t.data.types.ModalityConfig
114
+ action_configs: null
115
+ delta_indices:
116
+ - 0
117
+ mean_std_embedding_keys: null
118
+ modality_keys:
119
+ - ego_view
120
+ sin_cos_embedding_keys: null
121
+ mode: single_turn
122
+ multiprocessing_context: fork
123
+ num_shards_per_epoch: 100000
124
+ override_pretraining_statistics: true
125
+ random_chop: 0.0
126
+ seed: 42
127
+ shard_size: 1024
128
+ shuffle: true
129
+ subsample_ratio: 1.0
130
+ video_backend: torchcodec
131
+ load_config_path: null
132
+ model: !!python/object:gr00t.configs.model.gr00t_n1d7.Gr00tN1d7Config
133
+ _attn_implementation_internal: null
134
+ _commit_hash: null
135
+ _name_or_path: ''
136
+ _output_attentions: false
137
+ add_cross_attention: false
138
+ architectures: null
139
+ backbone_trainable_params_fp32: true
140
+ bad_words_ids: null
141
+ begin_suppress_tokens: null
142
+ bos_token_id: null
143
+ chunk_size_feed_forward: 0
144
+ color_jitter_params:
145
+ brightness: 0.3
146
+ contrast: 0.4
147
+ hue: 0.08
148
+ saturation: 0.5
149
+ cross_attention_hidden_size: null
150
+ decoder_start_token_id: null
151
+ diffusion_model_cfg:
152
+ attention_head_dim: 48
153
+ dropout: 0.2
154
+ final_dropout: true
155
+ interleave_self_attention: true
156
+ norm_type: ada_norm
157
+ num_attention_heads: 32
158
+ num_layers: 16
159
+ output_dim: 1024
160
+ positional_embeddings: null
161
+ diversity_penalty: 0.0
162
+ do_sample: false
163
+ dtype: null
164
+ early_stopping: false
165
+ encoder_no_repeat_ngram_size: 0
166
+ eos_token_id: null
167
+ exponential_decay_length_penalty: null
168
+ extra_augmentation_config: null
169
+ finetuning_task: null
170
+ forced_bos_token_id: null
171
+ forced_eos_token_id: null
172
+ id2label:
173
+ 0: LABEL_0
174
+ 1: LABEL_1
175
+ is_decoder: false
176
+ is_encoder_decoder: false
177
+ label2id:
178
+ LABEL_0: 0
179
+ LABEL_1: 1
180
+ length_penalty: 1.0
181
+ load_bf16: false
182
+ max_length: 20
183
+ min_length: 0
184
+ model_name: nvidia/Cosmos-Reason2-2B
185
+ no_repeat_ngram_size: 0
186
+ num_beam_groups: 1
187
+ num_beams: 1
188
+ num_return_sequences: 1
189
+ output_hidden_states: false
190
+ output_scores: false
191
+ pad_token_id: null
192
+ prefix: null
193
+ problem_type: null
194
+ pruned_heads: {}
195
+ random_rotation_angle: null
196
+ remove_invalid_values: false
197
+ repetition_penalty: 1.0
198
+ reproject_vision: false
199
+ return_dict: true
200
+ return_dict_in_generate: false
201
+ sep_token_id: null
202
+ state_dropout_prob: 0.2
203
+ suppress_tokens: null
204
+ task_specific_params: null
205
+ temperature: 1.0
206
+ tf_legacy_loss: false
207
+ tie_encoder_decoder: false
208
+ tie_word_embeddings: true
209
+ tokenizer_class: null
210
+ top_k: 50
211
+ top_p: 1.0
212
+ torchscript: false
213
+ transformers_version: null
214
+ tune_diffusion_model: true
215
+ tune_llm: false
216
+ tune_projector: true
217
+ tune_visual: false
218
+ typical_p: 1.0
219
+ use_bfloat16: false
220
+ use_relative_action: true
221
+ training: !!python/object:gr00t.configs.training.training_config.TrainingConfig
222
+ add_rl_callback: false
223
+ assert_loss_less_than: null
224
+ batch_size: null
225
+ bf16: true
226
+ dataloader_num_workers: 6
227
+ ddp_bucket_cap_mb: 100
228
+ deepspeed_stage: 2
229
+ enable_open_loop_eval: false
230
+ enable_profiling: false
231
+ eval_batch_size: 2
232
+ eval_bf16: true
233
+ eval_set_split_ratio: 0.1
234
+ eval_steps: 500
235
+ eval_strategy: 'no'
236
+ experiment_name: null
237
+ fp16: false
238
+ global_batch_size: 64
239
+ gradient_accumulation_steps: 1
240
+ gradient_checkpointing: false
241
+ learning_rate: 0.0001
242
+ logging_steps: 10
243
+ lr_scheduler_type: cosine
244
+ max_concurrent_uploads: 2
245
+ max_grad_norm: 1.0
246
+ max_retries: 3
247
+ max_steps: 15000
248
+ num_gpus: 1
249
+ open_loop_eval_plot_indices: null
250
+ open_loop_eval_steps_per_traj: 100
251
+ open_loop_eval_traj_ids:
252
+ - 0
253
+ optim: adamw_torch
254
+ output_dir: /home/ubuntu/groot-files/checkpoints/g1_finetune
255
+ remove_unused_columns: false
256
+ save_best_eval_metric_greater_is_better: true
257
+ save_best_eval_metric_name: ''
258
+ save_only_model: false
259
+ save_steps: 3000
260
+ save_total_limit: 10
261
+ save_vl_model: false
262
+ skip_weight_loading: false
263
+ start_from_checkpoint: nvidia/GR00T-N1.7-3B
264
+ tf32: true
265
+ transformers_access_token: null
266
+ transformers_cache_dir: null
267
+ transformers_local_files_only: false
268
+ transformers_trust_remote_code: true
269
+ upload_checkpoints: false
270
+ upload_every: 1000
271
+ upload_last_n_checkpoints: 5
272
+ use_ddp: false
273
+ use_wandb: true
274
+ wandb_project: groot-finetune
275
+ warmup_ratio: 0.05
276
+ warmup_steps: 0
277
+ weight_decay: 1.0e-05
checkpoint-6000/experiment_cfg/dataset_statistics.json ADDED
@@ -0,0 +1,907 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "unitree_g1_sonic": {
3
+ "state": {
4
+ "left_leg": {
5
+ "min": [
6
+ -0.4857776165008545,
7
+ -0.24391146004199982,
8
+ -0.8776780366897583,
9
+ 0.03597185015678406,
10
+ -0.4930363595485687,
11
+ -0.20340518653392792
12
+ ],
13
+ "max": [
14
+ 0.3237466514110565,
15
+ 0.31727835536003113,
16
+ 0.7570706605911255,
17
+ 0.8986315131187439,
18
+ 0.1892659068107605,
19
+ 0.15561887621879578
20
+ ],
21
+ "mean": [
22
+ 0.038367897272109985,
23
+ 0.017137423157691956,
24
+ 0.10202965885400772,
25
+ 0.21319514513015747,
26
+ -0.20034761726856232,
27
+ -0.009860903024673462
28
+ ],
29
+ "std": [
30
+ 0.12473291903734207,
31
+ 0.0657326877117157,
32
+ 0.16789446771144867,
33
+ 0.0855027362704277,
34
+ 0.05454540252685547,
35
+ 0.043583426624536514
36
+ ],
37
+ "q01": [
38
+ -0.3429580569267273,
39
+ -0.14271177425980566,
40
+ -0.41709126979112626,
41
+ 0.12328944355249405,
42
+ -0.2934864178299904,
43
+ -0.11147186532616615
44
+ ],
45
+ "q99": [
46
+ 0.18860293924808502,
47
+ 0.17139848843216907,
48
+ 0.4404219493269922,
49
+ 0.5360466212034226,
50
+ -0.019520661234855623,
51
+ 0.09359118081629283
52
+ ]
53
+ },
54
+ "right_leg": {
55
+ "min": [
56
+ -0.5335952043533325,
57
+ -0.350761741399765,
58
+ -1.3307929039001465,
59
+ 0.07445771992206573,
60
+ -0.6092288494110107,
61
+ -0.1984633058309555
62
+ ],
63
+ "max": [
64
+ 0.25397607684135437,
65
+ 0.15654273331165314,
66
+ 0.24837948381900787,
67
+ 0.9999935626983643,
68
+ 0.1089707612991333,
69
+ 0.17701147496700287
70
+ ],
71
+ "mean": [
72
+ 0.05046115070581436,
73
+ -0.023679343983530998,
74
+ -0.213065966963768,
75
+ 0.20844903588294983,
76
+ -0.19808240234851837,
77
+ -0.0055496226996183395
78
+ ],
79
+ "std": [
80
+ 0.07887373864650726,
81
+ 0.06217808276414871,
82
+ 0.18437089025974274,
83
+ 0.10042016953229904,
84
+ 0.06798264384269714,
85
+ 0.04407079145312309
86
+ ],
87
+ "q01": [
88
+ -0.193911362439394,
89
+ -0.1981988400220871,
90
+ -0.90549436211586,
91
+ 0.127213454246521,
92
+ -0.4343456655740738,
93
+ -0.11320190876722336
94
+ ],
95
+ "q99": [
96
+ 0.1967100210487844,
97
+ 0.11513377726078033,
98
+ 0.15384408906102207,
99
+ 0.7395615458488474,
100
+ -0.05514147877693175,
101
+ 0.08568341806530994
102
+ ]
103
+ },
104
+ "waist": {
105
+ "min": [
106
+ -0.3889274299144745,
107
+ -0.3032904267311096,
108
+ -0.04627600684762001
109
+ ],
110
+ "max": [
111
+ 0.4404728412628174,
112
+ 0.11908156424760818,
113
+ 0.05046159774065018
114
+ ],
115
+ "mean": [
116
+ 0.1003476083278656,
117
+ -0.03575156629085541,
118
+ -0.0004057084152009338
119
+ ],
120
+ "std": [
121
+ 0.11155267059803009,
122
+ 0.044356729835271835,
123
+ 0.007639772724360228
124
+ ],
125
+ "q01": [
126
+ -0.09896008186042309,
127
+ -0.16525592356920243,
128
+ -0.019172759167850016
129
+ ],
130
+ "q99": [
131
+ 0.3729535743594171,
132
+ 0.04077914115041498,
133
+ 0.024438940081745388
134
+ ]
135
+ },
136
+ "left_arm": {
137
+ "min": [
138
+ -0.2003283053636551,
139
+ 0.09816279262304306,
140
+ -0.5412195920944214,
141
+ 0.7723593711853027,
142
+ -0.7412483096122742,
143
+ -0.09287774562835693,
144
+ -0.31815722584724426
145
+ ],
146
+ "max": [
147
+ 0.4134317934513092,
148
+ 0.38915175199508667,
149
+ 0.3470511734485626,
150
+ 1.363792896270752,
151
+ 0.38266828656196594,
152
+ 0.5152138471603394,
153
+ 0.42312705516815186
154
+ ],
155
+ "mean": [
156
+ 0.17467822134494781,
157
+ 0.20962199568748474,
158
+ 0.03295444697141647,
159
+ 1.0894439220428467,
160
+ -0.11893314123153687,
161
+ 0.20652393996715546,
162
+ 0.01763514056801796
163
+ ],
164
+ "std": [
165
+ 0.08956529945135117,
166
+ 0.03943527489900589,
167
+ 0.09209521114826202,
168
+ 0.07066802680492401,
169
+ 0.19965490698814392,
170
+ 0.11541580408811569,
171
+ 0.10361754149198532
172
+ ],
173
+ "q01": [
174
+ -0.06361526139080524,
175
+ 0.1356973797082901,
176
+ -0.1961050637066364,
177
+ 0.9089213967323303,
178
+ -0.5769511580467224,
179
+ -0.03977684192359447,
180
+ -0.22587568387389184
181
+ ],
182
+ "q99": [
183
+ 0.3310546398162843,
184
+ 0.31673349142074597,
185
+ 0.2289490304887296,
186
+ 1.2593449354171753,
187
+ 0.2899013936519628,
188
+ 0.46442466974258423,
189
+ 0.2727669641375545
190
+ ]
191
+ },
192
+ "right_arm": {
193
+ "min": [
194
+ -0.9788835048675537,
195
+ -0.7953690886497498,
196
+ -0.49778875708580017,
197
+ -0.7156979441642761,
198
+ -0.9786917567253113,
199
+ -0.1273084282875061,
200
+ -0.9962846040725708
201
+ ],
202
+ "max": [
203
+ 0.3628104329109192,
204
+ 0.08640626072883606,
205
+ 0.8718883395195007,
206
+ 1.3758729696273804,
207
+ 1.0651459693908691,
208
+ 0.6638182401657104,
209
+ 0.9581388235092163
210
+ ],
211
+ "mean": [
212
+ -0.14295516908168793,
213
+ -0.20207007229328156,
214
+ 0.164081871509552,
215
+ 0.32978397607803345,
216
+ 0.15659268200397491,
217
+ 0.21257737278938293,
218
+ 0.0897224023938179
219
+ ],
220
+ "std": [
221
+ 0.2832476496696472,
222
+ 0.07377105951309204,
223
+ 0.18590903282165527,
224
+ 0.6193384528160095,
225
+ 0.20538125932216644,
226
+ 0.12802571058273315,
227
+ 0.21542473137378693
228
+ ],
229
+ "q01": [
230
+ -0.7704071193933487,
231
+ -0.376973994076252,
232
+ -0.19678157344460487,
233
+ -0.5173931628465652,
234
+ -0.39670541584491725,
235
+ -0.03838966768234967,
236
+ -0.3572006195783615
237
+ ],
238
+ "q99": [
239
+ 0.24905617535114288,
240
+ 0.0024351945263335014,
241
+ 0.586448073387146,
242
+ 1.304707604646683,
243
+ 0.5929710775613787,
244
+ 0.5161845684051514,
245
+ 0.6641693621873858
246
+ ]
247
+ },
248
+ "left_hand": {
249
+ "min": [
250
+ 0.0,
251
+ 0.0,
252
+ 0.0,
253
+ 0.0,
254
+ 0.0,
255
+ 0.0,
256
+ 0.0
257
+ ],
258
+ "max": [
259
+ 0.0,
260
+ 0.0,
261
+ 0.0,
262
+ 0.0,
263
+ 0.0,
264
+ 0.0,
265
+ 0.0
266
+ ],
267
+ "mean": [
268
+ 0.0,
269
+ 0.0,
270
+ 0.0,
271
+ 0.0,
272
+ 0.0,
273
+ 0.0,
274
+ 0.0
275
+ ],
276
+ "std": [
277
+ 0.0,
278
+ 0.0,
279
+ 0.0,
280
+ 0.0,
281
+ 0.0,
282
+ 0.0,
283
+ 0.0
284
+ ],
285
+ "q01": [
286
+ 0.0,
287
+ 0.0,
288
+ 0.0,
289
+ 0.0,
290
+ 0.0,
291
+ 0.0,
292
+ 0.0
293
+ ],
294
+ "q99": [
295
+ 0.0,
296
+ 0.0,
297
+ 0.0,
298
+ 0.0,
299
+ 0.0,
300
+ 0.0,
301
+ 0.0
302
+ ]
303
+ },
304
+ "right_hand": {
305
+ "min": [
306
+ 0.0,
307
+ 0.0,
308
+ 0.0,
309
+ 0.0,
310
+ 0.0,
311
+ 0.0,
312
+ 0.0
313
+ ],
314
+ "max": [
315
+ 0.0,
316
+ 0.0,
317
+ 0.0,
318
+ 0.0,
319
+ 0.0,
320
+ 0.0,
321
+ 0.0
322
+ ],
323
+ "mean": [
324
+ 0.0,
325
+ 0.0,
326
+ 0.0,
327
+ 0.0,
328
+ 0.0,
329
+ 0.0,
330
+ 0.0
331
+ ],
332
+ "std": [
333
+ 0.0,
334
+ 0.0,
335
+ 0.0,
336
+ 0.0,
337
+ 0.0,
338
+ 0.0,
339
+ 0.0
340
+ ],
341
+ "q01": [
342
+ 0.0,
343
+ 0.0,
344
+ 0.0,
345
+ 0.0,
346
+ 0.0,
347
+ 0.0,
348
+ 0.0
349
+ ],
350
+ "q99": [
351
+ 0.0,
352
+ 0.0,
353
+ 0.0,
354
+ 0.0,
355
+ 0.0,
356
+ 0.0,
357
+ 0.0
358
+ ]
359
+ },
360
+ "projected_gravity": {
361
+ "min": [
362
+ -0.12354099005460739,
363
+ -0.13017292320728302,
364
+ -1.0
365
+ ],
366
+ "max": [
367
+ 0.22223307192325592,
368
+ 0.10705921798944473,
369
+ -0.9749471545219421
370
+ ],
371
+ "mean": [
372
+ -0.01818661577999592,
373
+ 0.0029902660753577948,
374
+ -0.9988934993743896
375
+ ],
376
+ "std": [
377
+ 0.03578517585992813,
378
+ 0.032206468284130096,
379
+ 0.0014198371209307285
380
+ ],
381
+ "q01": [
382
+ -0.09425989575684071,
383
+ -0.08990333564579486,
384
+ -0.9999815225601196
385
+ ],
386
+ "q99": [
387
+ 0.06742736026644708,
388
+ 0.06235775817185641,
389
+ -0.9940253734588623
390
+ ]
391
+ }
392
+ },
393
+ "action": {
394
+ "motion_token": {
395
+ "min": [
396
+ -0.3125,
397
+ -0.5,
398
+ -0.5,
399
+ -0.4375,
400
+ -0.625,
401
+ -0.375,
402
+ -0.25,
403
+ -0.0625,
404
+ -0.4375,
405
+ -0.6875,
406
+ -0.375,
407
+ -0.375,
408
+ -0.3125,
409
+ -0.375,
410
+ -0.6875,
411
+ -0.25,
412
+ -0.375,
413
+ -0.25,
414
+ -0.375,
415
+ -0.5,
416
+ -0.5,
417
+ -0.5,
418
+ -0.625,
419
+ -0.5,
420
+ -0.375,
421
+ -0.5625,
422
+ -0.125,
423
+ -0.5,
424
+ -0.3125,
425
+ -0.3125,
426
+ -0.125,
427
+ -0.375,
428
+ 0.0625,
429
+ -0.1875,
430
+ -0.1875,
431
+ -0.5625,
432
+ -0.6875,
433
+ -0.6875,
434
+ -0.125,
435
+ -0.125,
436
+ -0.4375,
437
+ -0.5625,
438
+ -0.3125,
439
+ -0.375,
440
+ -0.5,
441
+ -0.4375,
442
+ -0.125,
443
+ -0.3125,
444
+ -0.5,
445
+ -0.25,
446
+ -0.375,
447
+ -0.625,
448
+ -0.0625,
449
+ -0.4375,
450
+ -0.0625,
451
+ -0.4375,
452
+ -0.5,
453
+ 0.0,
454
+ -0.25,
455
+ -0.5,
456
+ -0.375,
457
+ -0.1875,
458
+ 0.0,
459
+ -0.5
460
+ ],
461
+ "max": [
462
+ 0.125,
463
+ 0.25,
464
+ 0.25,
465
+ 0.125,
466
+ 0.1875,
467
+ 0.1875,
468
+ 0.5,
469
+ 0.4375,
470
+ 0.25,
471
+ 0.125,
472
+ 0.1875,
473
+ 0.0625,
474
+ 0.125,
475
+ 0.3125,
476
+ -0.0625,
477
+ 0.1875,
478
+ 0.25,
479
+ 0.25,
480
+ 0.125,
481
+ 0.0625,
482
+ 0.3125,
483
+ 0.125,
484
+ 0.25,
485
+ 0.25,
486
+ 0.3125,
487
+ 0.125,
488
+ 0.375,
489
+ 0.1875,
490
+ 0.375,
491
+ 0.375,
492
+ 0.375,
493
+ 0.25,
494
+ 0.4375,
495
+ 0.5,
496
+ 0.5,
497
+ 0.5625,
498
+ 0.25,
499
+ -0.0625,
500
+ 0.3125,
501
+ 0.4375,
502
+ 0.125,
503
+ 0.4375,
504
+ 0.4375,
505
+ 0.375,
506
+ -0.0625,
507
+ 0.1875,
508
+ 0.5625,
509
+ 0.1875,
510
+ 0.3125,
511
+ 0.1875,
512
+ 0.25,
513
+ 0.1875,
514
+ 0.375,
515
+ 0.3125,
516
+ 0.5625,
517
+ 0.1875,
518
+ 0.4375,
519
+ 0.5625,
520
+ 0.1875,
521
+ 0.375,
522
+ 0.3125,
523
+ 0.4375,
524
+ 0.5625,
525
+ 0.0
526
+ ],
527
+ "mean": [
528
+ -0.05732722207903862,
529
+ -0.12596048414707184,
530
+ -0.015900524333119392,
531
+ -0.16434414684772491,
532
+ -0.18957528471946716,
533
+ -0.08797170221805573,
534
+ 0.2548595666885376,
535
+ 0.19856274127960205,
536
+ -0.025066815316677094,
537
+ -0.21372610330581665,
538
+ -0.10314688086509705,
539
+ -0.21981850266456604,
540
+ -0.12050355225801468,
541
+ -0.08165144175291061,
542
+ -0.38714757561683655,
543
+ -8.805885590845719e-05,
544
+ -0.049404650926589966,
545
+ -0.055493421852588654,
546
+ -0.09600593894720078,
547
+ -0.2977478802204132,
548
+ -0.029102997854351997,
549
+ -0.2329292893409729,
550
+ -0.19594638049602509,
551
+ -0.23360289633274078,
552
+ -0.095408596098423,
553
+ -0.17069709300994873,
554
+ 0.1292250156402588,
555
+ -0.2315811812877655,
556
+ 0.07302621006965637,
557
+ 0.06577451527118683,
558
+ 0.1414906084537506,
559
+ -0.11845822632312775,
560
+ 0.2995853126049042,
561
+ 0.17635828256607056,
562
+ 0.14573286473751068,
563
+ 0.0782189592719078,
564
+ -0.1727832704782486,
565
+ -0.3186105489730835,
566
+ 0.11981905251741409,
567
+ 0.13200204074382782,
568
+ -0.08226785063743591,
569
+ -0.09514441341161728,
570
+ 0.060834143310785294,
571
+ -0.006165935657918453,
572
+ -0.3161884844303131,
573
+ -0.02333650551736355,
574
+ 0.21719853579998016,
575
+ -0.10866008698940277,
576
+ -0.01649424061179161,
577
+ -0.007006580010056496,
578
+ -0.06802137941122055,
579
+ -0.14377468824386597,
580
+ 0.21333755552768707,
581
+ 0.015348567627370358,
582
+ 0.22719819843769073,
583
+ -0.1495230346918106,
584
+ 0.030986731871962547,
585
+ 0.296153724193573,
586
+ -0.023153124377131462,
587
+ -0.1389232873916626,
588
+ 0.035359714180231094,
589
+ 0.10143018513917923,
590
+ 0.30849015712738037,
591
+ -0.22524820268154144
592
+ ],
593
+ "std": [
594
+ 0.06542618572711945,
595
+ 0.09994574636220932,
596
+ 0.07207678258419037,
597
+ 0.08702709525823593,
598
+ 0.17209544777870178,
599
+ 0.0581701286137104,
600
+ 0.09559116512537003,
601
+ 0.07802454382181168,
602
+ 0.13620688021183014,
603
+ 0.1569388061761856,
604
+ 0.0967717096209526,
605
+ 0.0680033266544342,
606
+ 0.05837923660874367,
607
+ 0.11557621508836746,
608
+ 0.11890240013599396,
609
+ 0.08077684044837952,
610
+ 0.06398067623376846,
611
+ 0.06826013326644897,
612
+ 0.09755037724971771,
613
+ 0.10724268853664398,
614
+ 0.1992810219526291,
615
+ 0.12710419297218323,
616
+ 0.21719561517238617,
617
+ 0.14369657635688782,
618
+ 0.08655858784914017,
619
+ 0.11142221838235855,
620
+ 0.057688891887664795,
621
+ 0.18031662702560425,
622
+ 0.0692613422870636,
623
+ 0.16206243634223938,
624
+ 0.11068906635046005,
625
+ 0.06650221347808838,
626
+ 0.06952458620071411,
627
+ 0.10304083675146103,
628
+ 0.12724509835243225,
629
+ 0.3227654695510864,
630
+ 0.14636477828025818,
631
+ 0.08603093028068542,
632
+ 0.08780314773321152,
633
+ 0.06147885322570801,
634
+ 0.07425742596387863,
635
+ 0.27255958318710327,
636
+ 0.13721545040607452,
637
+ 0.10124049335718155,
638
+ 0.06286311149597168,
639
+ 0.0642034038901329,
640
+ 0.08693595230579376,
641
+ 0.06350599974393845,
642
+ 0.11969305574893951,
643
+ 0.06820058822631836,
644
+ 0.09139743447303772,
645
+ 0.1545047014951706,
646
+ 0.06951460987329483,
647
+ 0.08810964971780777,
648
+ 0.0868336483836174,
649
+ 0.0749349296092987,
650
+ 0.09778603166341782,
651
+ 0.11456362158060074,
652
+ 0.06560879200696945,
653
+ 0.2043108344078064,
654
+ 0.07238639891147614,
655
+ 0.0965481698513031,
656
+ 0.11412639170885086,
657
+ 0.11945699900388718
658
+ ],
659
+ "q01": [
660
+ -0.1875,
661
+ -0.375,
662
+ -0.1875,
663
+ -0.375,
664
+ -0.5625,
665
+ -0.1875,
666
+ 0.0625,
667
+ 0.0,
668
+ -0.3125,
669
+ -0.625,
670
+ -0.3125,
671
+ -0.3125,
672
+ -0.25,
673
+ -0.3125,
674
+ -0.625,
675
+ -0.1875,
676
+ -0.25,
677
+ -0.1875,
678
+ -0.3125,
679
+ -0.5,
680
+ -0.375,
681
+ -0.4375,
682
+ -0.5625,
683
+ -0.4375,
684
+ -0.3125,
685
+ -0.5,
686
+ -0.0625,
687
+ -0.5,
688
+ -0.0625,
689
+ -0.1875,
690
+ -0.0625,
691
+ -0.25,
692
+ 0.125,
693
+ -0.0625,
694
+ -0.0625,
695
+ -0.5,
696
+ -0.5625,
697
+ -0.5625,
698
+ -0.0625,
699
+ 0.0,
700
+ -0.25,
701
+ -0.5625,
702
+ -0.1875,
703
+ -0.25,
704
+ -0.4375,
705
+ -0.1875,
706
+ 0.0625,
707
+ -0.25,
708
+ -0.375,
709
+ -0.1875,
710
+ -0.3125,
711
+ -0.5,
712
+ 0.0625,
713
+ -0.1875,
714
+ 0.0625,
715
+ -0.375,
716
+ -0.375,
717
+ 0.0625,
718
+ -0.1875,
719
+ -0.4375,
720
+ -0.1875,
721
+ -0.0625,
722
+ 0.125,
723
+ -0.4375
724
+ ],
725
+ "q99": [
726
+ 0.0625,
727
+ 0.0625,
728
+ 0.1875,
729
+ 0.0,
730
+ 0.125,
731
+ 0.0625,
732
+ 0.375,
733
+ 0.375,
734
+ 0.1875,
735
+ 0.0625,
736
+ 0.0625,
737
+ -0.0625,
738
+ 0.0625,
739
+ 0.1875,
740
+ -0.125,
741
+ 0.125,
742
+ 0.125,
743
+ 0.125,
744
+ 0.0625,
745
+ -0.0625,
746
+ 0.25,
747
+ 0.0,
748
+ 0.1875,
749
+ 0.125,
750
+ 0.0625,
751
+ 0.0,
752
+ 0.25,
753
+ 0.125,
754
+ 0.25,
755
+ 0.3125,
756
+ 0.3125,
757
+ 0.0625,
758
+ 0.375,
759
+ 0.375,
760
+ 0.4375,
761
+ 0.5,
762
+ 0.125,
763
+ -0.125,
764
+ 0.3125,
765
+ 0.25,
766
+ 0.0625,
767
+ 0.3125,
768
+ 0.3125,
769
+ 0.25,
770
+ -0.125,
771
+ 0.125,
772
+ 0.4375,
773
+ 0.0,
774
+ 0.1875,
775
+ 0.125,
776
+ 0.125,
777
+ 0.125,
778
+ 0.375,
779
+ 0.1875,
780
+ 0.5,
781
+ 0.0625,
782
+ 0.25,
783
+ 0.5,
784
+ 0.125,
785
+ 0.1875,
786
+ 0.25,
787
+ 0.3125,
788
+ 0.5,
789
+ 0.0
790
+ ]
791
+ },
792
+ "left_hand_joints": {
793
+ "min": [
794
+ 0.0,
795
+ 0.0,
796
+ 0.0,
797
+ 0.0,
798
+ 0.0,
799
+ 0.0,
800
+ 0.0
801
+ ],
802
+ "max": [
803
+ 0.0,
804
+ 0.0,
805
+ 0.0,
806
+ 0.0,
807
+ 0.0,
808
+ 0.0,
809
+ 0.0
810
+ ],
811
+ "mean": [
812
+ 0.0,
813
+ 0.0,
814
+ 0.0,
815
+ 0.0,
816
+ 0.0,
817
+ 0.0,
818
+ 0.0
819
+ ],
820
+ "std": [
821
+ 0.0,
822
+ 0.0,
823
+ 0.0,
824
+ 0.0,
825
+ 0.0,
826
+ 0.0,
827
+ 0.0
828
+ ],
829
+ "q01": [
830
+ 0.0,
831
+ 0.0,
832
+ 0.0,
833
+ 0.0,
834
+ 0.0,
835
+ 0.0,
836
+ 0.0
837
+ ],
838
+ "q99": [
839
+ 0.0,
840
+ 0.0,
841
+ 0.0,
842
+ 0.0,
843
+ 0.0,
844
+ 0.0,
845
+ 0.0
846
+ ]
847
+ },
848
+ "right_hand_joints": {
849
+ "min": [
850
+ 0.0,
851
+ 0.0,
852
+ 0.0,
853
+ 0.0,
854
+ 0.0,
855
+ 0.0,
856
+ 0.0
857
+ ],
858
+ "max": [
859
+ 0.0,
860
+ 0.0,
861
+ 0.0,
862
+ 0.0,
863
+ 0.0,
864
+ 0.0,
865
+ 0.0
866
+ ],
867
+ "mean": [
868
+ 0.0,
869
+ 0.0,
870
+ 0.0,
871
+ 0.0,
872
+ 0.0,
873
+ 0.0,
874
+ 0.0
875
+ ],
876
+ "std": [
877
+ 0.0,
878
+ 0.0,
879
+ 0.0,
880
+ 0.0,
881
+ 0.0,
882
+ 0.0,
883
+ 0.0
884
+ ],
885
+ "q01": [
886
+ 0.0,
887
+ 0.0,
888
+ 0.0,
889
+ 0.0,
890
+ 0.0,
891
+ 0.0,
892
+ 0.0
893
+ ],
894
+ "q99": [
895
+ 0.0,
896
+ 0.0,
897
+ 0.0,
898
+ 0.0,
899
+ 0.0,
900
+ 0.0,
901
+ 0.0
902
+ ]
903
+ }
904
+ },
905
+ "relative_action": {}
906
+ }
907
+ }
checkpoint-6000/experiment_cfg/final_model_config.json ADDED
@@ -0,0 +1,55 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "Gr00tN1d7",
3
+ "model_dtype": "bfloat16",
4
+ "model_name": "nvidia/Cosmos-Reason2-2B",
5
+ "backbone_model_type": "qwen",
6
+ "model_revision": null,
7
+ "tune_top_llm_layers": 0,
8
+ "backbone_embedding_dim": 2048,
9
+ "tune_llm": false,
10
+ "tune_visual": false,
11
+ "select_layer": 16,
12
+ "reproject_vision": false,
13
+ "use_flash_attention": true,
14
+ "load_bf16": false,
15
+ "backbone_trainable_params_fp32": true,
16
+ "extra_augmentation_config": null,
17
+ "apply_sincos_state_encoding": false,
18
+ "use_percentiles": true,
19
+ "use_relative_action": false,
20
+ "max_state_dim": 132,
21
+ "max_action_dim": 132,
22
+ "action_horizon": 40,
23
+ "hidden_size": 1024,
24
+ "input_embedding_dim": 1536,
25
+ "state_history_length": 1,
26
+ "add_pos_embed": true,
27
+ "attn_dropout": 0.2,
28
+ "use_vlln": true,
29
+ "max_seq_len": 1024,
30
+ "use_alternate_vl_dit": true,
31
+ "attend_text_every_n_blocks": 2,
32
+ "diffusion_model_cfg": {
33
+ "attention_head_dim": 48,
34
+ "dropout": 0.2,
35
+ "final_dropout": true,
36
+ "interleave_self_attention": true,
37
+ "norm_type": "ada_norm",
38
+ "num_attention_heads": 32,
39
+ "num_layers": 32,
40
+ "output_dim": 1024,
41
+ "positional_embeddings": null
42
+ },
43
+ "num_inference_timesteps": 4,
44
+ "noise_beta_alpha": 1.5,
45
+ "noise_beta_beta": 1.0,
46
+ "noise_s": 0.999,
47
+ "num_timestep_buckets": 1000,
48
+ "tune_projector": true,
49
+ "tune_diffusion_model": true,
50
+ "tune_vlln": true,
51
+ "state_dropout_prob": 0.2,
52
+ "exclude_state": false,
53
+ "use_mean_std": false,
54
+ "max_num_embodiments": 32
55
+ }
checkpoint-6000/experiment_cfg/final_processor_config.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-6000/model-00001-of-00003.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cbcaea5ee88f1e0f1465043920a2647c67e7de17d24adfd1c477742a6168edec
3
+ size 4986649584
checkpoint-6000/model-00002-of-00003.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e9315683ceb4468a1054fbd63e6ee08d2da8437771471ad05fdecc4c629ea33a
3
+ size 4970792616
checkpoint-6000/model-00003-of-00003.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:830c25b165506025f1eb50e1f6c768b56029b65d46c2df30e478b5168348a821
3
+ size 2618758696
checkpoint-6000/model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-6000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e5c6ba58dd164c1fabace511d960401c084a8007a963c364b8e8430e6120e952
3
+ size 12964594710
checkpoint-6000/processor_config.json ADDED
@@ -0,0 +1,1159 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "processor_class": "Gr00tN1d7Processor",
3
+ "processor_kwargs": {
4
+ "modality_configs": {
5
+ "real_g1_relative_eef_relative_joints": {
6
+ "video": {
7
+ "delta_indices": [
8
+ -20,
9
+ 0
10
+ ],
11
+ "modality_keys": [
12
+ "ego_view"
13
+ ],
14
+ "sin_cos_embedding_keys": null,
15
+ "mean_std_embedding_keys": null,
16
+ "action_configs": null
17
+ },
18
+ "state": {
19
+ "delta_indices": [
20
+ 0
21
+ ],
22
+ "modality_keys": [
23
+ "left_wrist_eef_9d",
24
+ "right_wrist_eef_9d",
25
+ "left_hand",
26
+ "right_hand",
27
+ "left_arm",
28
+ "right_arm",
29
+ "waist"
30
+ ],
31
+ "sin_cos_embedding_keys": null,
32
+ "mean_std_embedding_keys": null,
33
+ "action_configs": null
34
+ },
35
+ "action": {
36
+ "delta_indices": [
37
+ 0,
38
+ 1,
39
+ 2,
40
+ 3,
41
+ 4,
42
+ 5,
43
+ 6,
44
+ 7,
45
+ 8,
46
+ 9,
47
+ 10,
48
+ 11,
49
+ 12,
50
+ 13,
51
+ 14,
52
+ 15,
53
+ 16,
54
+ 17,
55
+ 18,
56
+ 19,
57
+ 20,
58
+ 21,
59
+ 22,
60
+ 23,
61
+ 24,
62
+ 25,
63
+ 26,
64
+ 27,
65
+ 28,
66
+ 29,
67
+ 30,
68
+ 31,
69
+ 32,
70
+ 33,
71
+ 34,
72
+ 35,
73
+ 36,
74
+ 37,
75
+ 38,
76
+ 39
77
+ ],
78
+ "modality_keys": [
79
+ "left_wrist_eef_9d",
80
+ "right_wrist_eef_9d",
81
+ "left_hand",
82
+ "right_hand",
83
+ "left_arm",
84
+ "right_arm",
85
+ "waist",
86
+ "base_height_command",
87
+ "navigate_command"
88
+ ],
89
+ "sin_cos_embedding_keys": null,
90
+ "mean_std_embedding_keys": null,
91
+ "action_configs": [
92
+ {
93
+ "rep": "RELATIVE",
94
+ "type": "EEF",
95
+ "format": "XYZ_ROT6D",
96
+ "state_key": "left_wrist_eef_9d"
97
+ },
98
+ {
99
+ "rep": "RELATIVE",
100
+ "type": "EEF",
101
+ "format": "XYZ_ROT6D",
102
+ "state_key": "right_wrist_eef_9d"
103
+ },
104
+ {
105
+ "rep": "ABSOLUTE",
106
+ "type": "NON_EEF",
107
+ "format": "DEFAULT",
108
+ "state_key": "left_hand"
109
+ },
110
+ {
111
+ "rep": "ABSOLUTE",
112
+ "type": "NON_EEF",
113
+ "format": "DEFAULT",
114
+ "state_key": "right_hand"
115
+ },
116
+ {
117
+ "rep": "RELATIVE",
118
+ "type": "NON_EEF",
119
+ "format": "DEFAULT",
120
+ "state_key": "left_arm"
121
+ },
122
+ {
123
+ "rep": "RELATIVE",
124
+ "type": "NON_EEF",
125
+ "format": "DEFAULT",
126
+ "state_key": "right_arm"
127
+ },
128
+ {
129
+ "rep": "ABSOLUTE",
130
+ "type": "NON_EEF",
131
+ "format": "DEFAULT",
132
+ "state_key": "waist"
133
+ },
134
+ {
135
+ "rep": "ABSOLUTE",
136
+ "type": "NON_EEF",
137
+ "format": "DEFAULT",
138
+ "state_key": "base_height_command"
139
+ },
140
+ {
141
+ "rep": "ABSOLUTE",
142
+ "type": "NON_EEF",
143
+ "format": "DEFAULT",
144
+ "state_key": "navigate_command"
145
+ }
146
+ ]
147
+ },
148
+ "language": {
149
+ "delta_indices": [
150
+ 0
151
+ ],
152
+ "modality_keys": [
153
+ "annotation.human.task_description"
154
+ ],
155
+ "sin_cos_embedding_keys": null,
156
+ "mean_std_embedding_keys": null,
157
+ "action_configs": null
158
+ }
159
+ },
160
+ "real_r1_pro_sharpa_relative_eef_mecka": {
161
+ "video": {
162
+ "delta_indices": [
163
+ -30,
164
+ 0
165
+ ],
166
+ "modality_keys": [
167
+ "ego_view_cropratio_res320x240_freq30"
168
+ ],
169
+ "sin_cos_embedding_keys": null,
170
+ "mean_std_embedding_keys": null,
171
+ "action_configs": null
172
+ },
173
+ "state": {
174
+ "delta_indices": [
175
+ 0
176
+ ],
177
+ "modality_keys": [
178
+ "left_wrist_eef",
179
+ "right_wrist_eef",
180
+ "left_hand_joints",
181
+ "right_hand_joints"
182
+ ],
183
+ "sin_cos_embedding_keys": null,
184
+ "mean_std_embedding_keys": null,
185
+ "action_configs": null
186
+ },
187
+ "action": {
188
+ "delta_indices": [
189
+ 0,
190
+ 1,
191
+ 2,
192
+ 3,
193
+ 4,
194
+ 5,
195
+ 6,
196
+ 7,
197
+ 8,
198
+ 9,
199
+ 10,
200
+ 11,
201
+ 12,
202
+ 13,
203
+ 14,
204
+ 15,
205
+ 16,
206
+ 17,
207
+ 18,
208
+ 19,
209
+ 20,
210
+ 21,
211
+ 22,
212
+ 23,
213
+ 24,
214
+ 25,
215
+ 26,
216
+ 27,
217
+ 28,
218
+ 29,
219
+ 30,
220
+ 31,
221
+ 32,
222
+ 33,
223
+ 34,
224
+ 35,
225
+ 36,
226
+ 37,
227
+ 38,
228
+ 39
229
+ ],
230
+ "modality_keys": [
231
+ "left_wrist_eef",
232
+ "right_wrist_eef",
233
+ "left_hand_joints",
234
+ "right_hand_joints"
235
+ ],
236
+ "sin_cos_embedding_keys": null,
237
+ "mean_std_embedding_keys": null,
238
+ "action_configs": [
239
+ {
240
+ "rep": "RELATIVE",
241
+ "type": "EEF",
242
+ "format": "XYZ_ROT6D",
243
+ "state_key": "left_wrist_eef"
244
+ },
245
+ {
246
+ "rep": "RELATIVE",
247
+ "type": "EEF",
248
+ "format": "XYZ_ROT6D",
249
+ "state_key": "right_wrist_eef"
250
+ },
251
+ {
252
+ "rep": "ABSOLUTE",
253
+ "type": "NON_EEF",
254
+ "format": "DEFAULT",
255
+ "state_key": "left_hand_joints"
256
+ },
257
+ {
258
+ "rep": "ABSOLUTE",
259
+ "type": "NON_EEF",
260
+ "format": "DEFAULT",
261
+ "state_key": "right_hand_joints"
262
+ }
263
+ ]
264
+ },
265
+ "language": {
266
+ "delta_indices": [
267
+ 0
268
+ ],
269
+ "modality_keys": [
270
+ "annotation.human.coarse_action"
271
+ ],
272
+ "sin_cos_embedding_keys": null,
273
+ "mean_std_embedding_keys": null,
274
+ "action_configs": null
275
+ }
276
+ },
277
+ "real_r1_pro_sharpa_relative_eef_human": {
278
+ "video": {
279
+ "delta_indices": [
280
+ -20,
281
+ 0
282
+ ],
283
+ "modality_keys": [
284
+ "ego_view_res320x240_freq20",
285
+ "left_wrist_view_res320x240_freq20",
286
+ "right_wrist_view_res320x240_freq20"
287
+ ],
288
+ "sin_cos_embedding_keys": null,
289
+ "mean_std_embedding_keys": null,
290
+ "action_configs": null
291
+ },
292
+ "state": {
293
+ "delta_indices": [
294
+ 0
295
+ ],
296
+ "modality_keys": [
297
+ "left_wrist_eef",
298
+ "right_wrist_eef",
299
+ "left_hand_joints",
300
+ "right_hand_joints"
301
+ ],
302
+ "sin_cos_embedding_keys": null,
303
+ "mean_std_embedding_keys": null,
304
+ "action_configs": null
305
+ },
306
+ "action": {
307
+ "delta_indices": [
308
+ 0,
309
+ 1,
310
+ 2,
311
+ 3,
312
+ 4,
313
+ 5,
314
+ 6,
315
+ 7,
316
+ 8,
317
+ 9,
318
+ 10,
319
+ 11,
320
+ 12,
321
+ 13,
322
+ 14,
323
+ 15,
324
+ 16,
325
+ 17,
326
+ 18,
327
+ 19,
328
+ 20,
329
+ 21,
330
+ 22,
331
+ 23,
332
+ 24,
333
+ 25,
334
+ 26,
335
+ 27,
336
+ 28,
337
+ 29,
338
+ 30,
339
+ 31,
340
+ 32,
341
+ 33,
342
+ 34,
343
+ 35,
344
+ 36,
345
+ 37,
346
+ 38,
347
+ 39
348
+ ],
349
+ "modality_keys": [
350
+ "left_wrist_eef",
351
+ "right_wrist_eef",
352
+ "left_hand_joints",
353
+ "right_hand_joints"
354
+ ],
355
+ "sin_cos_embedding_keys": null,
356
+ "mean_std_embedding_keys": null,
357
+ "action_configs": [
358
+ {
359
+ "rep": "RELATIVE",
360
+ "type": "EEF",
361
+ "format": "XYZ_ROT6D",
362
+ "state_key": "left_wrist_eef"
363
+ },
364
+ {
365
+ "rep": "RELATIVE",
366
+ "type": "EEF",
367
+ "format": "XYZ_ROT6D",
368
+ "state_key": "right_wrist_eef"
369
+ },
370
+ {
371
+ "rep": "ABSOLUTE",
372
+ "type": "NON_EEF",
373
+ "format": "DEFAULT",
374
+ "state_key": "left_hand_joints"
375
+ },
376
+ {
377
+ "rep": "ABSOLUTE",
378
+ "type": "NON_EEF",
379
+ "format": "DEFAULT",
380
+ "state_key": "right_hand_joints"
381
+ }
382
+ ]
383
+ },
384
+ "language": {
385
+ "delta_indices": [
386
+ 0
387
+ ],
388
+ "modality_keys": [
389
+ "annotation.human.coarse_action"
390
+ ],
391
+ "sin_cos_embedding_keys": null,
392
+ "mean_std_embedding_keys": null,
393
+ "action_configs": null
394
+ }
395
+ },
396
+ "real_r1_pro_sharpa_relative_eef": {
397
+ "video": {
398
+ "delta_indices": [
399
+ -20,
400
+ 0
401
+ ],
402
+ "modality_keys": [
403
+ "ego_view_res320x240_freq20",
404
+ "left_wrist_view_res320x240_freq20",
405
+ "right_wrist_view_res320x240_freq20"
406
+ ],
407
+ "sin_cos_embedding_keys": null,
408
+ "mean_std_embedding_keys": null,
409
+ "action_configs": null
410
+ },
411
+ "state": {
412
+ "delta_indices": [
413
+ 0
414
+ ],
415
+ "modality_keys": [
416
+ "left_wrist_eef",
417
+ "right_wrist_eef",
418
+ "left_hand_joints",
419
+ "right_hand_joints"
420
+ ],
421
+ "sin_cos_embedding_keys": null,
422
+ "mean_std_embedding_keys": null,
423
+ "action_configs": null
424
+ },
425
+ "action": {
426
+ "delta_indices": [
427
+ 0,
428
+ 1,
429
+ 2,
430
+ 3,
431
+ 4,
432
+ 5,
433
+ 6,
434
+ 7,
435
+ 8,
436
+ 9,
437
+ 10,
438
+ 11,
439
+ 12,
440
+ 13,
441
+ 14,
442
+ 15,
443
+ 16,
444
+ 17,
445
+ 18,
446
+ 19,
447
+ 20,
448
+ 21,
449
+ 22,
450
+ 23,
451
+ 24,
452
+ 25,
453
+ 26,
454
+ 27,
455
+ 28,
456
+ 29,
457
+ 30,
458
+ 31,
459
+ 32,
460
+ 33,
461
+ 34,
462
+ 35,
463
+ 36,
464
+ 37,
465
+ 38,
466
+ 39
467
+ ],
468
+ "modality_keys": [
469
+ "left_wrist_eef",
470
+ "right_wrist_eef",
471
+ "left_hand_joints",
472
+ "right_hand_joints"
473
+ ],
474
+ "sin_cos_embedding_keys": null,
475
+ "mean_std_embedding_keys": null,
476
+ "action_configs": [
477
+ {
478
+ "rep": "RELATIVE",
479
+ "type": "EEF",
480
+ "format": "XYZ_ROT6D",
481
+ "state_key": "left_wrist_eef"
482
+ },
483
+ {
484
+ "rep": "RELATIVE",
485
+ "type": "EEF",
486
+ "format": "XYZ_ROT6D",
487
+ "state_key": "right_wrist_eef"
488
+ },
489
+ {
490
+ "rep": "ABSOLUTE",
491
+ "type": "NON_EEF",
492
+ "format": "DEFAULT",
493
+ "state_key": "left_hand_joints"
494
+ },
495
+ {
496
+ "rep": "ABSOLUTE",
497
+ "type": "NON_EEF",
498
+ "format": "DEFAULT",
499
+ "state_key": "right_hand_joints"
500
+ }
501
+ ]
502
+ },
503
+ "language": {
504
+ "delta_indices": [
505
+ 0
506
+ ],
507
+ "modality_keys": [
508
+ "annotation.human.coarse_action"
509
+ ],
510
+ "sin_cos_embedding_keys": null,
511
+ "mean_std_embedding_keys": null,
512
+ "action_configs": null
513
+ }
514
+ },
515
+ "xdof_relative_eef_relative_joint": {
516
+ "video": {
517
+ "delta_indices": [
518
+ -30,
519
+ 0
520
+ ],
521
+ "modality_keys": [
522
+ "top_camera-images-rgb_320_240",
523
+ "left_camera-images-rgb_320_240",
524
+ "right_camera-images-rgb_320_240"
525
+ ],
526
+ "sin_cos_embedding_keys": null,
527
+ "mean_std_embedding_keys": null,
528
+ "action_configs": null
529
+ },
530
+ "state": {
531
+ "delta_indices": [
532
+ 0
533
+ ],
534
+ "modality_keys": [
535
+ "left_wrist_eef",
536
+ "right_wrist_eef",
537
+ "left_gripper_pos",
538
+ "right_gripper_pos",
539
+ "left_joint_pos",
540
+ "right_joint_pos"
541
+ ],
542
+ "sin_cos_embedding_keys": null,
543
+ "mean_std_embedding_keys": null,
544
+ "action_configs": null
545
+ },
546
+ "action": {
547
+ "delta_indices": [
548
+ 0,
549
+ 1,
550
+ 2,
551
+ 3,
552
+ 4,
553
+ 5,
554
+ 6,
555
+ 7,
556
+ 8,
557
+ 9,
558
+ 10,
559
+ 11,
560
+ 12,
561
+ 13,
562
+ 14,
563
+ 15,
564
+ 16,
565
+ 17,
566
+ 18,
567
+ 19,
568
+ 20,
569
+ 21,
570
+ 22,
571
+ 23,
572
+ 24,
573
+ 25,
574
+ 26,
575
+ 27,
576
+ 28,
577
+ 29,
578
+ 30,
579
+ 31,
580
+ 32,
581
+ 33,
582
+ 34,
583
+ 35,
584
+ 36,
585
+ 37,
586
+ 38,
587
+ 39
588
+ ],
589
+ "modality_keys": [
590
+ "left_wrist_eef",
591
+ "right_wrist_eef",
592
+ "left_gripper_pos",
593
+ "right_gripper_pos",
594
+ "left_joint_pos",
595
+ "right_joint_pos"
596
+ ],
597
+ "sin_cos_embedding_keys": null,
598
+ "mean_std_embedding_keys": null,
599
+ "action_configs": [
600
+ {
601
+ "rep": "RELATIVE",
602
+ "type": "EEF",
603
+ "format": "XYZ_ROT6D",
604
+ "state_key": "left_wrist_eef"
605
+ },
606
+ {
607
+ "rep": "RELATIVE",
608
+ "type": "EEF",
609
+ "format": "XYZ_ROT6D",
610
+ "state_key": "right_wrist_eef"
611
+ },
612
+ {
613
+ "rep": "ABSOLUTE",
614
+ "type": "NON_EEF",
615
+ "format": "DEFAULT",
616
+ "state_key": "left_gripper_pos"
617
+ },
618
+ {
619
+ "rep": "ABSOLUTE",
620
+ "type": "NON_EEF",
621
+ "format": "DEFAULT",
622
+ "state_key": "right_gripper_pos"
623
+ },
624
+ {
625
+ "rep": "RELATIVE",
626
+ "type": "NON_EEF",
627
+ "format": "DEFAULT",
628
+ "state_key": "left_joint_pos"
629
+ },
630
+ {
631
+ "rep": "RELATIVE",
632
+ "type": "NON_EEF",
633
+ "format": "DEFAULT",
634
+ "state_key": "right_joint_pos"
635
+ }
636
+ ]
637
+ },
638
+ "language": {
639
+ "delta_indices": [
640
+ 0
641
+ ],
642
+ "modality_keys": [
643
+ "annotation.task"
644
+ ],
645
+ "sin_cos_embedding_keys": null,
646
+ "mean_std_embedding_keys": null,
647
+ "action_configs": null
648
+ }
649
+ },
650
+ "real_r1_pro_sharpa_relative_eef_maxinsights": {
651
+ "video": {
652
+ "delta_indices": [
653
+ -30,
654
+ 0
655
+ ],
656
+ "modality_keys": [
657
+ "ego_view_cropratio_res320x240_freq30"
658
+ ],
659
+ "sin_cos_embedding_keys": null,
660
+ "mean_std_embedding_keys": null,
661
+ "action_configs": null
662
+ },
663
+ "state": {
664
+ "delta_indices": [
665
+ 0
666
+ ],
667
+ "modality_keys": [
668
+ "left_wrist_eef",
669
+ "right_wrist_eef",
670
+ "left_hand_joints",
671
+ "right_hand_joints"
672
+ ],
673
+ "sin_cos_embedding_keys": null,
674
+ "mean_std_embedding_keys": null,
675
+ "action_configs": null
676
+ },
677
+ "action": {
678
+ "delta_indices": [
679
+ 0,
680
+ 1,
681
+ 2,
682
+ 3,
683
+ 4,
684
+ 5,
685
+ 6,
686
+ 7,
687
+ 8,
688
+ 9,
689
+ 10,
690
+ 11,
691
+ 12,
692
+ 13,
693
+ 14,
694
+ 15,
695
+ 16,
696
+ 17,
697
+ 18,
698
+ 19,
699
+ 20,
700
+ 21,
701
+ 22,
702
+ 23,
703
+ 24,
704
+ 25,
705
+ 26,
706
+ 27,
707
+ 28,
708
+ 29,
709
+ 30,
710
+ 31,
711
+ 32,
712
+ 33,
713
+ 34,
714
+ 35,
715
+ 36,
716
+ 37,
717
+ 38,
718
+ 39
719
+ ],
720
+ "modality_keys": [
721
+ "left_wrist_eef",
722
+ "right_wrist_eef",
723
+ "left_hand_joints",
724
+ "right_hand_joints"
725
+ ],
726
+ "sin_cos_embedding_keys": null,
727
+ "mean_std_embedding_keys": null,
728
+ "action_configs": [
729
+ {
730
+ "rep": "RELATIVE",
731
+ "type": "EEF",
732
+ "format": "XYZ_ROT6D",
733
+ "state_key": "left_wrist_eef"
734
+ },
735
+ {
736
+ "rep": "RELATIVE",
737
+ "type": "EEF",
738
+ "format": "XYZ_ROT6D",
739
+ "state_key": "right_wrist_eef"
740
+ },
741
+ {
742
+ "rep": "ABSOLUTE",
743
+ "type": "NON_EEF",
744
+ "format": "DEFAULT",
745
+ "state_key": "left_hand_joints"
746
+ },
747
+ {
748
+ "rep": "ABSOLUTE",
749
+ "type": "NON_EEF",
750
+ "format": "DEFAULT",
751
+ "state_key": "right_hand_joints"
752
+ }
753
+ ]
754
+ },
755
+ "language": {
756
+ "delta_indices": [
757
+ 0
758
+ ],
759
+ "modality_keys": [
760
+ "annotation.human.coarse_action"
761
+ ],
762
+ "sin_cos_embedding_keys": null,
763
+ "mean_std_embedding_keys": null,
764
+ "action_configs": null
765
+ }
766
+ },
767
+ "xdof_relative_eef_relative_joint_subtask": {
768
+ "video": {
769
+ "delta_indices": [
770
+ -30,
771
+ 0
772
+ ],
773
+ "modality_keys": [
774
+ "top_camera-images-rgb_320_240",
775
+ "left_camera-images-rgb_320_240",
776
+ "right_camera-images-rgb_320_240"
777
+ ],
778
+ "sin_cos_embedding_keys": null,
779
+ "mean_std_embedding_keys": null,
780
+ "action_configs": null
781
+ },
782
+ "state": {
783
+ "delta_indices": [
784
+ 0
785
+ ],
786
+ "modality_keys": [
787
+ "left_wrist_eef",
788
+ "right_wrist_eef",
789
+ "left_gripper_pos",
790
+ "right_gripper_pos",
791
+ "left_joint_pos",
792
+ "right_joint_pos"
793
+ ],
794
+ "sin_cos_embedding_keys": null,
795
+ "mean_std_embedding_keys": null,
796
+ "action_configs": null
797
+ },
798
+ "action": {
799
+ "delta_indices": [
800
+ 0,
801
+ 1,
802
+ 2,
803
+ 3,
804
+ 4,
805
+ 5,
806
+ 6,
807
+ 7,
808
+ 8,
809
+ 9,
810
+ 10,
811
+ 11,
812
+ 12,
813
+ 13,
814
+ 14,
815
+ 15,
816
+ 16,
817
+ 17,
818
+ 18,
819
+ 19,
820
+ 20,
821
+ 21,
822
+ 22,
823
+ 23,
824
+ 24,
825
+ 25,
826
+ 26,
827
+ 27,
828
+ 28,
829
+ 29,
830
+ 30,
831
+ 31,
832
+ 32,
833
+ 33,
834
+ 34,
835
+ 35,
836
+ 36,
837
+ 37,
838
+ 38,
839
+ 39
840
+ ],
841
+ "modality_keys": [
842
+ "left_wrist_eef",
843
+ "right_wrist_eef",
844
+ "left_gripper_pos",
845
+ "right_gripper_pos",
846
+ "left_joint_pos",
847
+ "right_joint_pos"
848
+ ],
849
+ "sin_cos_embedding_keys": null,
850
+ "mean_std_embedding_keys": null,
851
+ "action_configs": [
852
+ {
853
+ "rep": "RELATIVE",
854
+ "type": "EEF",
855
+ "format": "XYZ_ROT6D",
856
+ "state_key": "left_wrist_eef"
857
+ },
858
+ {
859
+ "rep": "RELATIVE",
860
+ "type": "EEF",
861
+ "format": "XYZ_ROT6D",
862
+ "state_key": "right_wrist_eef"
863
+ },
864
+ {
865
+ "rep": "ABSOLUTE",
866
+ "type": "NON_EEF",
867
+ "format": "DEFAULT",
868
+ "state_key": "left_gripper_pos"
869
+ },
870
+ {
871
+ "rep": "ABSOLUTE",
872
+ "type": "NON_EEF",
873
+ "format": "DEFAULT",
874
+ "state_key": "right_gripper_pos"
875
+ },
876
+ {
877
+ "rep": "RELATIVE",
878
+ "type": "NON_EEF",
879
+ "format": "DEFAULT",
880
+ "state_key": "left_joint_pos"
881
+ },
882
+ {
883
+ "rep": "RELATIVE",
884
+ "type": "NON_EEF",
885
+ "format": "DEFAULT",
886
+ "state_key": "right_joint_pos"
887
+ }
888
+ ]
889
+ },
890
+ "language": {
891
+ "delta_indices": [
892
+ 0
893
+ ],
894
+ "modality_keys": [
895
+ "annotation.sub_task"
896
+ ],
897
+ "sin_cos_embedding_keys": null,
898
+ "mean_std_embedding_keys": null,
899
+ "action_configs": null
900
+ }
901
+ },
902
+ "oxe_droid_relative_eef_relative_joint": {
903
+ "video": {
904
+ "delta_indices": [
905
+ -15,
906
+ 0
907
+ ],
908
+ "modality_keys": [
909
+ "exterior_image_1_left",
910
+ "wrist_image_left"
911
+ ],
912
+ "sin_cos_embedding_keys": null,
913
+ "mean_std_embedding_keys": null,
914
+ "action_configs": null
915
+ },
916
+ "state": {
917
+ "delta_indices": [
918
+ 0
919
+ ],
920
+ "modality_keys": [
921
+ "eef_9d",
922
+ "gripper_position",
923
+ "joint_position"
924
+ ],
925
+ "sin_cos_embedding_keys": null,
926
+ "mean_std_embedding_keys": null,
927
+ "action_configs": null
928
+ },
929
+ "action": {
930
+ "delta_indices": [
931
+ 0,
932
+ 1,
933
+ 2,
934
+ 3,
935
+ 4,
936
+ 5,
937
+ 6,
938
+ 7,
939
+ 8,
940
+ 9,
941
+ 10,
942
+ 11,
943
+ 12,
944
+ 13,
945
+ 14,
946
+ 15,
947
+ 16,
948
+ 17,
949
+ 18,
950
+ 19,
951
+ 20,
952
+ 21,
953
+ 22,
954
+ 23,
955
+ 24,
956
+ 25,
957
+ 26,
958
+ 27,
959
+ 28,
960
+ 29,
961
+ 30,
962
+ 31,
963
+ 32,
964
+ 33,
965
+ 34,
966
+ 35,
967
+ 36,
968
+ 37,
969
+ 38,
970
+ 39
971
+ ],
972
+ "modality_keys": [
973
+ "eef_9d",
974
+ "gripper_position",
975
+ "joint_position"
976
+ ],
977
+ "sin_cos_embedding_keys": null,
978
+ "mean_std_embedding_keys": null,
979
+ "action_configs": [
980
+ {
981
+ "rep": "RELATIVE",
982
+ "type": "EEF",
983
+ "format": "XYZ_ROT6D",
984
+ "state_key": "eef_9d"
985
+ },
986
+ {
987
+ "rep": "ABSOLUTE",
988
+ "type": "NON_EEF",
989
+ "format": "DEFAULT",
990
+ "state_key": "gripper_position"
991
+ },
992
+ {
993
+ "rep": "RELATIVE",
994
+ "type": "NON_EEF",
995
+ "format": "DEFAULT",
996
+ "state_key": "joint_position"
997
+ }
998
+ ]
999
+ },
1000
+ "language": {
1001
+ "delta_indices": [
1002
+ 0
1003
+ ],
1004
+ "modality_keys": [
1005
+ "annotation.language.language_instruction"
1006
+ ],
1007
+ "sin_cos_embedding_keys": null,
1008
+ "mean_std_embedding_keys": null,
1009
+ "action_configs": null
1010
+ }
1011
+ },
1012
+ "unitree_g1_sonic": {
1013
+ "video": {
1014
+ "delta_indices": [
1015
+ 0
1016
+ ],
1017
+ "modality_keys": [
1018
+ "ego_view"
1019
+ ],
1020
+ "sin_cos_embedding_keys": null,
1021
+ "mean_std_embedding_keys": null,
1022
+ "action_configs": null
1023
+ },
1024
+ "state": {
1025
+ "delta_indices": [
1026
+ 0
1027
+ ],
1028
+ "modality_keys": [
1029
+ "left_leg",
1030
+ "right_leg",
1031
+ "waist",
1032
+ "left_arm",
1033
+ "right_arm",
1034
+ "left_hand",
1035
+ "right_hand",
1036
+ "projected_gravity"
1037
+ ],
1038
+ "sin_cos_embedding_keys": null,
1039
+ "mean_std_embedding_keys": null,
1040
+ "action_configs": null
1041
+ },
1042
+ "action": {
1043
+ "delta_indices": [
1044
+ 0,
1045
+ 1,
1046
+ 2,
1047
+ 3,
1048
+ 4,
1049
+ 5,
1050
+ 6,
1051
+ 7,
1052
+ 8,
1053
+ 9,
1054
+ 10,
1055
+ 11,
1056
+ 12,
1057
+ 13,
1058
+ 14,
1059
+ 15,
1060
+ 16,
1061
+ 17,
1062
+ 18,
1063
+ 19,
1064
+ 20,
1065
+ 21,
1066
+ 22,
1067
+ 23,
1068
+ 24,
1069
+ 25,
1070
+ 26,
1071
+ 27,
1072
+ 28,
1073
+ 29,
1074
+ 30,
1075
+ 31,
1076
+ 32,
1077
+ 33,
1078
+ 34,
1079
+ 35,
1080
+ 36,
1081
+ 37,
1082
+ 38,
1083
+ 39
1084
+ ],
1085
+ "modality_keys": [
1086
+ "motion_token",
1087
+ "left_hand_joints",
1088
+ "right_hand_joints"
1089
+ ],
1090
+ "sin_cos_embedding_keys": null,
1091
+ "mean_std_embedding_keys": null,
1092
+ "action_configs": [
1093
+ {
1094
+ "rep": "ABSOLUTE",
1095
+ "type": "NON_EEF",
1096
+ "format": "DEFAULT",
1097
+ "state_key": null
1098
+ },
1099
+ {
1100
+ "rep": "ABSOLUTE",
1101
+ "type": "NON_EEF",
1102
+ "format": "DEFAULT",
1103
+ "state_key": null
1104
+ },
1105
+ {
1106
+ "rep": "ABSOLUTE",
1107
+ "type": "NON_EEF",
1108
+ "format": "DEFAULT",
1109
+ "state_key": null
1110
+ }
1111
+ ]
1112
+ },
1113
+ "language": {
1114
+ "delta_indices": [
1115
+ 0
1116
+ ],
1117
+ "modality_keys": [
1118
+ "annotation.human.task_description"
1119
+ ],
1120
+ "sin_cos_embedding_keys": null,
1121
+ "mean_std_embedding_keys": null,
1122
+ "action_configs": null
1123
+ }
1124
+ }
1125
+ },
1126
+ "image_crop_size": [
1127
+ 230,
1128
+ 230
1129
+ ],
1130
+ "image_target_size": [
1131
+ 256,
1132
+ 256
1133
+ ],
1134
+ "use_albumentations": true,
1135
+ "random_rotation_angle": 0,
1136
+ "color_jitter_params": {
1137
+ "brightness": 0.3,
1138
+ "contrast": 0.4,
1139
+ "saturation": 0.5,
1140
+ "hue": 0.08
1141
+ },
1142
+ "shortest_image_edge": 256,
1143
+ "crop_fraction": 0.95,
1144
+ "letter_box_transform": false,
1145
+ "model_name": "nvidia/Cosmos-Reason2-2B",
1146
+ "model_type": "qwen",
1147
+ "formalize_language": true,
1148
+ "max_state_dim": 132,
1149
+ "max_action_dim": 132,
1150
+ "max_action_horizon": 40,
1151
+ "use_percentiles": true,
1152
+ "use_mean_std": false,
1153
+ "clip_outliers": true,
1154
+ "apply_sincos_state_encoding": false,
1155
+ "use_relative_action": true,
1156
+ "exclude_state": false,
1157
+ "state_dropout_prob": 0.2
1158
+ }
1159
+ }
checkpoint-6000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:18f6895ab2d041ed8ff15e9173f478e29b71a52aa4e137a5a40ea937833b62e1
3
+ size 14645
checkpoint-6000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:11636c231f0243f3c3f2c2dd1368f406a6866d4a4b79306f2a4f9325be633cc2
3
+ size 1465
checkpoint-6000/statistics.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-6000/trainer_state.json ADDED
@@ -0,0 +1,3634 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 0.4,
6
+ "eval_steps": 500,
7
+ "global_step": 6000,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "grad_norm": 0.1528221219778061,
14
+ "learning_rate": 1.2000000000000002e-06,
15
+ "loss": 1.2068,
16
+ "step": 10
17
+ },
18
+ {
19
+ "grad_norm": 0.12877728044986725,
20
+ "learning_rate": 2.5333333333333334e-06,
21
+ "loss": 1.2134,
22
+ "step": 20
23
+ },
24
+ {
25
+ "grad_norm": 0.17161273956298828,
26
+ "learning_rate": 3.866666666666667e-06,
27
+ "loss": 1.206,
28
+ "step": 30
29
+ },
30
+ {
31
+ "grad_norm": 0.2625730037689209,
32
+ "learning_rate": 5.2e-06,
33
+ "loss": 1.1817,
34
+ "step": 40
35
+ },
36
+ {
37
+ "grad_norm": 0.2874051034450531,
38
+ "learning_rate": 6.533333333333333e-06,
39
+ "loss": 1.1488,
40
+ "step": 50
41
+ },
42
+ {
43
+ "grad_norm": 0.34452179074287415,
44
+ "learning_rate": 7.866666666666667e-06,
45
+ "loss": 1.1238,
46
+ "step": 60
47
+ },
48
+ {
49
+ "grad_norm": 0.3187244236469269,
50
+ "learning_rate": 9.2e-06,
51
+ "loss": 1.1072,
52
+ "step": 70
53
+ },
54
+ {
55
+ "grad_norm": 0.41171619296073914,
56
+ "learning_rate": 1.0533333333333335e-05,
57
+ "loss": 1.0936,
58
+ "step": 80
59
+ },
60
+ {
61
+ "grad_norm": 0.22929424047470093,
62
+ "learning_rate": 1.1866666666666668e-05,
63
+ "loss": 1.0933,
64
+ "step": 90
65
+ },
66
+ {
67
+ "grad_norm": 0.5159542560577393,
68
+ "learning_rate": 1.32e-05,
69
+ "loss": 1.0919,
70
+ "step": 100
71
+ },
72
+ {
73
+ "grad_norm": 0.4722374975681305,
74
+ "learning_rate": 1.4533333333333335e-05,
75
+ "loss": 1.0945,
76
+ "step": 110
77
+ },
78
+ {
79
+ "grad_norm": 0.2221572995185852,
80
+ "learning_rate": 1.586666666666667e-05,
81
+ "loss": 1.0911,
82
+ "step": 120
83
+ },
84
+ {
85
+ "grad_norm": 0.5312994122505188,
86
+ "learning_rate": 1.7199999999999998e-05,
87
+ "loss": 1.0815,
88
+ "step": 130
89
+ },
90
+ {
91
+ "grad_norm": 0.3242526650428772,
92
+ "learning_rate": 1.8533333333333334e-05,
93
+ "loss": 1.0778,
94
+ "step": 140
95
+ },
96
+ {
97
+ "grad_norm": 0.28936564922332764,
98
+ "learning_rate": 1.9866666666666667e-05,
99
+ "loss": 1.0505,
100
+ "step": 150
101
+ },
102
+ {
103
+ "grad_norm": 0.3376358151435852,
104
+ "learning_rate": 2.12e-05,
105
+ "loss": 1.02,
106
+ "step": 160
107
+ },
108
+ {
109
+ "grad_norm": 0.5211641192436218,
110
+ "learning_rate": 2.2533333333333333e-05,
111
+ "loss": 0.9858,
112
+ "step": 170
113
+ },
114
+ {
115
+ "grad_norm": 0.41194382309913635,
116
+ "learning_rate": 2.3866666666666666e-05,
117
+ "loss": 0.9434,
118
+ "step": 180
119
+ },
120
+ {
121
+ "grad_norm": 0.5447150468826294,
122
+ "learning_rate": 2.5200000000000003e-05,
123
+ "loss": 0.9057,
124
+ "step": 190
125
+ },
126
+ {
127
+ "grad_norm": 0.8995847702026367,
128
+ "learning_rate": 2.6533333333333332e-05,
129
+ "loss": 0.8692,
130
+ "step": 200
131
+ },
132
+ {
133
+ "grad_norm": 0.7821958065032959,
134
+ "learning_rate": 2.786666666666667e-05,
135
+ "loss": 0.8394,
136
+ "step": 210
137
+ },
138
+ {
139
+ "grad_norm": 1.4184938669204712,
140
+ "learning_rate": 2.9199999999999998e-05,
141
+ "loss": 0.8041,
142
+ "step": 220
143
+ },
144
+ {
145
+ "grad_norm": 0.9261109828948975,
146
+ "learning_rate": 3.0533333333333335e-05,
147
+ "loss": 0.7759,
148
+ "step": 230
149
+ },
150
+ {
151
+ "grad_norm": 1.2726421356201172,
152
+ "learning_rate": 3.1866666666666664e-05,
153
+ "loss": 0.7407,
154
+ "step": 240
155
+ },
156
+ {
157
+ "grad_norm": 1.1038875579833984,
158
+ "learning_rate": 3.32e-05,
159
+ "loss": 0.7069,
160
+ "step": 250
161
+ },
162
+ {
163
+ "grad_norm": 1.1673762798309326,
164
+ "learning_rate": 3.453333333333334e-05,
165
+ "loss": 0.6797,
166
+ "step": 260
167
+ },
168
+ {
169
+ "grad_norm": 1.0007288455963135,
170
+ "learning_rate": 3.586666666666667e-05,
171
+ "loss": 0.6563,
172
+ "step": 270
173
+ },
174
+ {
175
+ "grad_norm": 0.8429192304611206,
176
+ "learning_rate": 3.72e-05,
177
+ "loss": 0.6337,
178
+ "step": 280
179
+ },
180
+ {
181
+ "grad_norm": 0.9413348436355591,
182
+ "learning_rate": 3.853333333333334e-05,
183
+ "loss": 0.6113,
184
+ "step": 290
185
+ },
186
+ {
187
+ "grad_norm": 0.99355548620224,
188
+ "learning_rate": 3.986666666666667e-05,
189
+ "loss": 0.6006,
190
+ "step": 300
191
+ },
192
+ {
193
+ "grad_norm": 1.1705503463745117,
194
+ "learning_rate": 4.12e-05,
195
+ "loss": 0.5865,
196
+ "step": 310
197
+ },
198
+ {
199
+ "grad_norm": 1.3885726928710938,
200
+ "learning_rate": 4.2533333333333335e-05,
201
+ "loss": 0.5685,
202
+ "step": 320
203
+ },
204
+ {
205
+ "grad_norm": 1.224961280822754,
206
+ "learning_rate": 4.3866666666666665e-05,
207
+ "loss": 0.5633,
208
+ "step": 330
209
+ },
210
+ {
211
+ "grad_norm": 1.2367912530899048,
212
+ "learning_rate": 4.52e-05,
213
+ "loss": 0.5547,
214
+ "step": 340
215
+ },
216
+ {
217
+ "grad_norm": 1.1392102241516113,
218
+ "learning_rate": 4.653333333333334e-05,
219
+ "loss": 0.5403,
220
+ "step": 350
221
+ },
222
+ {
223
+ "grad_norm": 1.213355541229248,
224
+ "learning_rate": 4.7866666666666674e-05,
225
+ "loss": 0.5357,
226
+ "step": 360
227
+ },
228
+ {
229
+ "grad_norm": 1.2761033773422241,
230
+ "learning_rate": 4.92e-05,
231
+ "loss": 0.5058,
232
+ "step": 370
233
+ },
234
+ {
235
+ "grad_norm": 1.1698124408721924,
236
+ "learning_rate": 5.053333333333333e-05,
237
+ "loss": 0.4966,
238
+ "step": 380
239
+ },
240
+ {
241
+ "grad_norm": 1.201072096824646,
242
+ "learning_rate": 5.1866666666666676e-05,
243
+ "loss": 0.4828,
244
+ "step": 390
245
+ },
246
+ {
247
+ "grad_norm": 1.0158417224884033,
248
+ "learning_rate": 5.3200000000000006e-05,
249
+ "loss": 0.4651,
250
+ "step": 400
251
+ },
252
+ {
253
+ "grad_norm": 1.346627116203308,
254
+ "learning_rate": 5.4533333333333335e-05,
255
+ "loss": 0.4594,
256
+ "step": 410
257
+ },
258
+ {
259
+ "grad_norm": 1.0970842838287354,
260
+ "learning_rate": 5.5866666666666665e-05,
261
+ "loss": 0.4445,
262
+ "step": 420
263
+ },
264
+ {
265
+ "grad_norm": 1.2192479372024536,
266
+ "learning_rate": 5.72e-05,
267
+ "loss": 0.4299,
268
+ "step": 430
269
+ },
270
+ {
271
+ "grad_norm": 1.2104856967926025,
272
+ "learning_rate": 5.853333333333334e-05,
273
+ "loss": 0.4117,
274
+ "step": 440
275
+ },
276
+ {
277
+ "grad_norm": 1.2335385084152222,
278
+ "learning_rate": 5.9866666666666674e-05,
279
+ "loss": 0.4031,
280
+ "step": 450
281
+ },
282
+ {
283
+ "grad_norm": 0.9111726880073547,
284
+ "learning_rate": 6.12e-05,
285
+ "loss": 0.3901,
286
+ "step": 460
287
+ },
288
+ {
289
+ "grad_norm": 1.1724517345428467,
290
+ "learning_rate": 6.253333333333333e-05,
291
+ "loss": 0.3773,
292
+ "step": 470
293
+ },
294
+ {
295
+ "grad_norm": 1.2332345247268677,
296
+ "learning_rate": 6.386666666666667e-05,
297
+ "loss": 0.3737,
298
+ "step": 480
299
+ },
300
+ {
301
+ "grad_norm": 1.1842234134674072,
302
+ "learning_rate": 6.52e-05,
303
+ "loss": 0.3628,
304
+ "step": 490
305
+ },
306
+ {
307
+ "grad_norm": 1.2130239009857178,
308
+ "learning_rate": 6.653333333333334e-05,
309
+ "loss": 0.3544,
310
+ "step": 500
311
+ },
312
+ {
313
+ "grad_norm": 1.0411441326141357,
314
+ "learning_rate": 6.786666666666667e-05,
315
+ "loss": 0.3326,
316
+ "step": 510
317
+ },
318
+ {
319
+ "grad_norm": 1.2490897178649902,
320
+ "learning_rate": 6.92e-05,
321
+ "loss": 0.3279,
322
+ "step": 520
323
+ },
324
+ {
325
+ "grad_norm": 1.33174467086792,
326
+ "learning_rate": 7.053333333333334e-05,
327
+ "loss": 0.3177,
328
+ "step": 530
329
+ },
330
+ {
331
+ "grad_norm": 1.1573432683944702,
332
+ "learning_rate": 7.186666666666667e-05,
333
+ "loss": 0.3106,
334
+ "step": 540
335
+ },
336
+ {
337
+ "grad_norm": 1.2820322513580322,
338
+ "learning_rate": 7.32e-05,
339
+ "loss": 0.2829,
340
+ "step": 550
341
+ },
342
+ {
343
+ "grad_norm": 1.2154183387756348,
344
+ "learning_rate": 7.453333333333333e-05,
345
+ "loss": 0.2846,
346
+ "step": 560
347
+ },
348
+ {
349
+ "grad_norm": 1.0811824798583984,
350
+ "learning_rate": 7.586666666666668e-05,
351
+ "loss": 0.274,
352
+ "step": 570
353
+ },
354
+ {
355
+ "grad_norm": 1.2624622583389282,
356
+ "learning_rate": 7.72e-05,
357
+ "loss": 0.2581,
358
+ "step": 580
359
+ },
360
+ {
361
+ "grad_norm": 1.4608895778656006,
362
+ "learning_rate": 7.853333333333334e-05,
363
+ "loss": 0.2436,
364
+ "step": 590
365
+ },
366
+ {
367
+ "grad_norm": 1.3445744514465332,
368
+ "learning_rate": 7.986666666666667e-05,
369
+ "loss": 0.2339,
370
+ "step": 600
371
+ },
372
+ {
373
+ "grad_norm": 1.2945458889007568,
374
+ "learning_rate": 8.120000000000001e-05,
375
+ "loss": 0.22,
376
+ "step": 610
377
+ },
378
+ {
379
+ "grad_norm": 2.7877917289733887,
380
+ "learning_rate": 8.253333333333334e-05,
381
+ "loss": 0.206,
382
+ "step": 620
383
+ },
384
+ {
385
+ "grad_norm": 1.1054668426513672,
386
+ "learning_rate": 8.386666666666667e-05,
387
+ "loss": 0.2005,
388
+ "step": 630
389
+ },
390
+ {
391
+ "grad_norm": 1.2020183801651,
392
+ "learning_rate": 8.52e-05,
393
+ "loss": 0.1906,
394
+ "step": 640
395
+ },
396
+ {
397
+ "grad_norm": 1.3809555768966675,
398
+ "learning_rate": 8.653333333333333e-05,
399
+ "loss": 0.1761,
400
+ "step": 650
401
+ },
402
+ {
403
+ "grad_norm": 1.2968473434448242,
404
+ "learning_rate": 8.786666666666667e-05,
405
+ "loss": 0.17,
406
+ "step": 660
407
+ },
408
+ {
409
+ "grad_norm": 1.5829375982284546,
410
+ "learning_rate": 8.92e-05,
411
+ "loss": 0.1582,
412
+ "step": 670
413
+ },
414
+ {
415
+ "grad_norm": 1.4462990760803223,
416
+ "learning_rate": 9.053333333333334e-05,
417
+ "loss": 0.1532,
418
+ "step": 680
419
+ },
420
+ {
421
+ "grad_norm": 1.4937175512313843,
422
+ "learning_rate": 9.186666666666667e-05,
423
+ "loss": 0.1436,
424
+ "step": 690
425
+ },
426
+ {
427
+ "grad_norm": 1.0529409646987915,
428
+ "learning_rate": 9.320000000000002e-05,
429
+ "loss": 0.146,
430
+ "step": 700
431
+ },
432
+ {
433
+ "grad_norm": 1.4069396257400513,
434
+ "learning_rate": 9.453333333333335e-05,
435
+ "loss": 0.1401,
436
+ "step": 710
437
+ },
438
+ {
439
+ "grad_norm": 1.448398232460022,
440
+ "learning_rate": 9.586666666666667e-05,
441
+ "loss": 0.1399,
442
+ "step": 720
443
+ },
444
+ {
445
+ "grad_norm": 1.3302212953567505,
446
+ "learning_rate": 9.72e-05,
447
+ "loss": 0.1397,
448
+ "step": 730
449
+ },
450
+ {
451
+ "grad_norm": 1.2735353708267212,
452
+ "learning_rate": 9.853333333333333e-05,
453
+ "loss": 0.1411,
454
+ "step": 740
455
+ },
456
+ {
457
+ "grad_norm": 1.265570044517517,
458
+ "learning_rate": 9.986666666666668e-05,
459
+ "loss": 0.1434,
460
+ "step": 750
461
+ },
462
+ {
463
+ "grad_norm": 1.3887313604354858,
464
+ "learning_rate": 9.999990157738453e-05,
465
+ "loss": 0.1386,
466
+ "step": 760
467
+ },
468
+ {
469
+ "grad_norm": 1.2708746194839478,
470
+ "learning_rate": 9.999956135155687e-05,
471
+ "loss": 0.1394,
472
+ "step": 770
473
+ },
474
+ {
475
+ "grad_norm": 1.3188905715942383,
476
+ "learning_rate": 9.99989781090763e-05,
477
+ "loss": 0.1362,
478
+ "step": 780
479
+ },
480
+ {
481
+ "grad_norm": 2.1023690700531006,
482
+ "learning_rate": 9.999815185277755e-05,
483
+ "loss": 0.132,
484
+ "step": 790
485
+ },
486
+ {
487
+ "grad_norm": 1.0511329174041748,
488
+ "learning_rate": 9.999708258667652e-05,
489
+ "loss": 0.1393,
490
+ "step": 800
491
+ },
492
+ {
493
+ "grad_norm": 1.076863169670105,
494
+ "learning_rate": 9.999577031597029e-05,
495
+ "loss": 0.1219,
496
+ "step": 810
497
+ },
498
+ {
499
+ "grad_norm": 1.0692609548568726,
500
+ "learning_rate": 9.999421504703696e-05,
501
+ "loss": 0.1249,
502
+ "step": 820
503
+ },
504
+ {
505
+ "grad_norm": 1.0652865171432495,
506
+ "learning_rate": 9.999241678743574e-05,
507
+ "loss": 0.124,
508
+ "step": 830
509
+ },
510
+ {
511
+ "grad_norm": 1.1164456605911255,
512
+ "learning_rate": 9.999037554590683e-05,
513
+ "loss": 0.1301,
514
+ "step": 840
515
+ },
516
+ {
517
+ "grad_norm": 1.4576336145401,
518
+ "learning_rate": 9.998809133237143e-05,
519
+ "loss": 0.1263,
520
+ "step": 850
521
+ },
522
+ {
523
+ "grad_norm": 1.0527737140655518,
524
+ "learning_rate": 9.998556415793169e-05,
525
+ "loss": 0.1283,
526
+ "step": 860
527
+ },
528
+ {
529
+ "grad_norm": 1.2238246202468872,
530
+ "learning_rate": 9.998279403487062e-05,
531
+ "loss": 0.1271,
532
+ "step": 870
533
+ },
534
+ {
535
+ "grad_norm": 1.146423578262329,
536
+ "learning_rate": 9.997978097665205e-05,
537
+ "loss": 0.1269,
538
+ "step": 880
539
+ },
540
+ {
541
+ "grad_norm": 1.0656620264053345,
542
+ "learning_rate": 9.99765249979206e-05,
543
+ "loss": 0.1239,
544
+ "step": 890
545
+ },
546
+ {
547
+ "grad_norm": 1.1446648836135864,
548
+ "learning_rate": 9.997302611450154e-05,
549
+ "loss": 0.1237,
550
+ "step": 900
551
+ },
552
+ {
553
+ "grad_norm": 1.1433049440383911,
554
+ "learning_rate": 9.996928434340073e-05,
555
+ "loss": 0.1147,
556
+ "step": 910
557
+ },
558
+ {
559
+ "grad_norm": 0.9724878072738647,
560
+ "learning_rate": 9.996529970280462e-05,
561
+ "loss": 0.1113,
562
+ "step": 920
563
+ },
564
+ {
565
+ "grad_norm": 1.074139952659607,
566
+ "learning_rate": 9.996107221208004e-05,
567
+ "loss": 0.1198,
568
+ "step": 930
569
+ },
570
+ {
571
+ "grad_norm": 0.9398661851882935,
572
+ "learning_rate": 9.995660189177419e-05,
573
+ "loss": 0.1149,
574
+ "step": 940
575
+ },
576
+ {
577
+ "grad_norm": 0.8259074687957764,
578
+ "learning_rate": 9.995188876361451e-05,
579
+ "loss": 0.1155,
580
+ "step": 950
581
+ },
582
+ {
583
+ "grad_norm": 1.1398017406463623,
584
+ "learning_rate": 9.994693285050857e-05,
585
+ "loss": 0.1192,
586
+ "step": 960
587
+ },
588
+ {
589
+ "grad_norm": 0.9386220574378967,
590
+ "learning_rate": 9.994173417654395e-05,
591
+ "loss": 0.1186,
592
+ "step": 970
593
+ },
594
+ {
595
+ "grad_norm": 0.9809224605560303,
596
+ "learning_rate": 9.993629276698821e-05,
597
+ "loss": 0.1135,
598
+ "step": 980
599
+ },
600
+ {
601
+ "grad_norm": 1.196757197380066,
602
+ "learning_rate": 9.993060864828858e-05,
603
+ "loss": 0.1247,
604
+ "step": 990
605
+ },
606
+ {
607
+ "grad_norm": 0.9535070061683655,
608
+ "learning_rate": 9.992468184807206e-05,
609
+ "loss": 0.1098,
610
+ "step": 1000
611
+ },
612
+ {
613
+ "grad_norm": 1.011692762374878,
614
+ "learning_rate": 9.991851239514511e-05,
615
+ "loss": 0.1093,
616
+ "step": 1010
617
+ },
618
+ {
619
+ "grad_norm": 0.9049733877182007,
620
+ "learning_rate": 9.991210031949359e-05,
621
+ "loss": 0.1171,
622
+ "step": 1020
623
+ },
624
+ {
625
+ "grad_norm": 0.9988471269607544,
626
+ "learning_rate": 9.990544565228259e-05,
627
+ "loss": 0.1085,
628
+ "step": 1030
629
+ },
630
+ {
631
+ "grad_norm": 0.9829862713813782,
632
+ "learning_rate": 9.989854842585631e-05,
633
+ "loss": 0.1041,
634
+ "step": 1040
635
+ },
636
+ {
637
+ "grad_norm": 0.9640742540359497,
638
+ "learning_rate": 9.989140867373783e-05,
639
+ "loss": 0.1053,
640
+ "step": 1050
641
+ },
642
+ {
643
+ "grad_norm": 0.9070046544075012,
644
+ "learning_rate": 9.988402643062907e-05,
645
+ "loss": 0.1035,
646
+ "step": 1060
647
+ },
648
+ {
649
+ "grad_norm": 1.0155551433563232,
650
+ "learning_rate": 9.987640173241046e-05,
651
+ "loss": 0.1075,
652
+ "step": 1070
653
+ },
654
+ {
655
+ "grad_norm": 1.1102415323257446,
656
+ "learning_rate": 9.986853461614093e-05,
657
+ "loss": 0.1132,
658
+ "step": 1080
659
+ },
660
+ {
661
+ "grad_norm": 1.4231481552124023,
662
+ "learning_rate": 9.986042512005763e-05,
663
+ "loss": 0.1107,
664
+ "step": 1090
665
+ },
666
+ {
667
+ "grad_norm": 0.9319804310798645,
668
+ "learning_rate": 9.985207328357573e-05,
669
+ "loss": 0.1026,
670
+ "step": 1100
671
+ },
672
+ {
673
+ "grad_norm": 1.4864873886108398,
674
+ "learning_rate": 9.984347914728829e-05,
675
+ "loss": 0.1011,
676
+ "step": 1110
677
+ },
678
+ {
679
+ "grad_norm": 1.0301569700241089,
680
+ "learning_rate": 9.983464275296605e-05,
681
+ "loss": 0.1114,
682
+ "step": 1120
683
+ },
684
+ {
685
+ "grad_norm": 0.9580268859863281,
686
+ "learning_rate": 9.982556414355724e-05,
687
+ "loss": 0.1099,
688
+ "step": 1130
689
+ },
690
+ {
691
+ "grad_norm": 1.0523960590362549,
692
+ "learning_rate": 9.981624336318726e-05,
693
+ "loss": 0.1053,
694
+ "step": 1140
695
+ },
696
+ {
697
+ "grad_norm": 1.060106635093689,
698
+ "learning_rate": 9.980668045715864e-05,
699
+ "loss": 0.1,
700
+ "step": 1150
701
+ },
702
+ {
703
+ "grad_norm": 0.9312297701835632,
704
+ "learning_rate": 9.979687547195066e-05,
705
+ "loss": 0.1061,
706
+ "step": 1160
707
+ },
708
+ {
709
+ "grad_norm": 0.9556792974472046,
710
+ "learning_rate": 9.978682845521927e-05,
711
+ "loss": 0.107,
712
+ "step": 1170
713
+ },
714
+ {
715
+ "grad_norm": 1.0657126903533936,
716
+ "learning_rate": 9.977653945579673e-05,
717
+ "loss": 0.1049,
718
+ "step": 1180
719
+ },
720
+ {
721
+ "grad_norm": 0.8762414455413818,
722
+ "learning_rate": 9.976600852369144e-05,
723
+ "loss": 0.1001,
724
+ "step": 1190
725
+ },
726
+ {
727
+ "grad_norm": 0.861569344997406,
728
+ "learning_rate": 9.975523571008769e-05,
729
+ "loss": 0.1064,
730
+ "step": 1200
731
+ },
732
+ {
733
+ "grad_norm": 0.9219115376472473,
734
+ "learning_rate": 9.97442210673454e-05,
735
+ "loss": 0.0983,
736
+ "step": 1210
737
+ },
738
+ {
739
+ "grad_norm": 1.0116729736328125,
740
+ "learning_rate": 9.973296464899988e-05,
741
+ "loss": 0.1018,
742
+ "step": 1220
743
+ },
744
+ {
745
+ "grad_norm": 0.9443844556808472,
746
+ "learning_rate": 9.972146650976154e-05,
747
+ "loss": 0.1033,
748
+ "step": 1230
749
+ },
750
+ {
751
+ "grad_norm": 0.9114744067192078,
752
+ "learning_rate": 9.970972670551566e-05,
753
+ "loss": 0.1034,
754
+ "step": 1240
755
+ },
756
+ {
757
+ "grad_norm": 0.8281648755073547,
758
+ "learning_rate": 9.969774529332212e-05,
759
+ "loss": 0.0995,
760
+ "step": 1250
761
+ },
762
+ {
763
+ "grad_norm": 0.8310883045196533,
764
+ "learning_rate": 9.968552233141504e-05,
765
+ "loss": 0.0992,
766
+ "step": 1260
767
+ },
768
+ {
769
+ "grad_norm": 0.948841392993927,
770
+ "learning_rate": 9.967305787920264e-05,
771
+ "loss": 0.0985,
772
+ "step": 1270
773
+ },
774
+ {
775
+ "grad_norm": 0.8757680654525757,
776
+ "learning_rate": 9.966035199726684e-05,
777
+ "loss": 0.1062,
778
+ "step": 1280
779
+ },
780
+ {
781
+ "grad_norm": 0.7957766056060791,
782
+ "learning_rate": 9.9647404747363e-05,
783
+ "loss": 0.0949,
784
+ "step": 1290
785
+ },
786
+ {
787
+ "grad_norm": 0.9588861465454102,
788
+ "learning_rate": 9.96342161924196e-05,
789
+ "loss": 0.0952,
790
+ "step": 1300
791
+ },
792
+ {
793
+ "grad_norm": 0.8227360248565674,
794
+ "learning_rate": 9.962078639653797e-05,
795
+ "loss": 0.0931,
796
+ "step": 1310
797
+ },
798
+ {
799
+ "grad_norm": 1.0412708520889282,
800
+ "learning_rate": 9.960711542499202e-05,
801
+ "loss": 0.0922,
802
+ "step": 1320
803
+ },
804
+ {
805
+ "grad_norm": 0.8027946352958679,
806
+ "learning_rate": 9.959320334422772e-05,
807
+ "loss": 0.1075,
808
+ "step": 1330
809
+ },
810
+ {
811
+ "grad_norm": 0.9406609535217285,
812
+ "learning_rate": 9.957905022186309e-05,
813
+ "loss": 0.0984,
814
+ "step": 1340
815
+ },
816
+ {
817
+ "grad_norm": 0.8434228301048279,
818
+ "learning_rate": 9.956465612668757e-05,
819
+ "loss": 0.0979,
820
+ "step": 1350
821
+ },
822
+ {
823
+ "grad_norm": 0.9157196879386902,
824
+ "learning_rate": 9.95500211286619e-05,
825
+ "loss": 0.0941,
826
+ "step": 1360
827
+ },
828
+ {
829
+ "grad_norm": 0.8689035177230835,
830
+ "learning_rate": 9.953514529891763e-05,
831
+ "loss": 0.0949,
832
+ "step": 1370
833
+ },
834
+ {
835
+ "grad_norm": 0.7697312831878662,
836
+ "learning_rate": 9.952002870975693e-05,
837
+ "loss": 0.0914,
838
+ "step": 1380
839
+ },
840
+ {
841
+ "grad_norm": 0.8707885146141052,
842
+ "learning_rate": 9.950467143465207e-05,
843
+ "loss": 0.0926,
844
+ "step": 1390
845
+ },
846
+ {
847
+ "grad_norm": 1.025254487991333,
848
+ "learning_rate": 9.94890735482452e-05,
849
+ "loss": 0.0912,
850
+ "step": 1400
851
+ },
852
+ {
853
+ "grad_norm": 0.6527482271194458,
854
+ "learning_rate": 9.947323512634788e-05,
855
+ "loss": 0.0873,
856
+ "step": 1410
857
+ },
858
+ {
859
+ "grad_norm": 0.8045205473899841,
860
+ "learning_rate": 9.945715624594081e-05,
861
+ "loss": 0.09,
862
+ "step": 1420
863
+ },
864
+ {
865
+ "grad_norm": 0.7984917163848877,
866
+ "learning_rate": 9.944083698517339e-05,
867
+ "loss": 0.0907,
868
+ "step": 1430
869
+ },
870
+ {
871
+ "grad_norm": 0.7339720726013184,
872
+ "learning_rate": 9.942427742336334e-05,
873
+ "loss": 0.0923,
874
+ "step": 1440
875
+ },
876
+ {
877
+ "grad_norm": 0.8728718757629395,
878
+ "learning_rate": 9.940747764099638e-05,
879
+ "loss": 0.0891,
880
+ "step": 1450
881
+ },
882
+ {
883
+ "grad_norm": 0.8001891374588013,
884
+ "learning_rate": 9.939043771972574e-05,
885
+ "loss": 0.0999,
886
+ "step": 1460
887
+ },
888
+ {
889
+ "grad_norm": 0.861858606338501,
890
+ "learning_rate": 9.937315774237186e-05,
891
+ "loss": 0.0892,
892
+ "step": 1470
893
+ },
894
+ {
895
+ "grad_norm": 0.9286336898803711,
896
+ "learning_rate": 9.93556377929219e-05,
897
+ "loss": 0.0946,
898
+ "step": 1480
899
+ },
900
+ {
901
+ "grad_norm": 0.9956574440002441,
902
+ "learning_rate": 9.933787795652942e-05,
903
+ "loss": 0.0913,
904
+ "step": 1490
905
+ },
906
+ {
907
+ "grad_norm": 0.8465809226036072,
908
+ "learning_rate": 9.931987831951386e-05,
909
+ "loss": 0.0912,
910
+ "step": 1500
911
+ },
912
+ {
913
+ "grad_norm": 0.8478185534477234,
914
+ "learning_rate": 9.930163896936027e-05,
915
+ "loss": 0.0958,
916
+ "step": 1510
917
+ },
918
+ {
919
+ "grad_norm": 0.7136925458908081,
920
+ "learning_rate": 9.92831599947187e-05,
921
+ "loss": 0.0848,
922
+ "step": 1520
923
+ },
924
+ {
925
+ "grad_norm": 0.8562847375869751,
926
+ "learning_rate": 9.926444148540393e-05,
927
+ "loss": 0.0952,
928
+ "step": 1530
929
+ },
930
+ {
931
+ "grad_norm": 0.8773907423019409,
932
+ "learning_rate": 9.924548353239495e-05,
933
+ "loss": 0.0972,
934
+ "step": 1540
935
+ },
936
+ {
937
+ "grad_norm": 0.770634114742279,
938
+ "learning_rate": 9.922628622783451e-05,
939
+ "loss": 0.0913,
940
+ "step": 1550
941
+ },
942
+ {
943
+ "grad_norm": 0.6977461576461792,
944
+ "learning_rate": 9.920684966502878e-05,
945
+ "loss": 0.0914,
946
+ "step": 1560
947
+ },
948
+ {
949
+ "grad_norm": 0.7688239216804504,
950
+ "learning_rate": 9.918717393844669e-05,
951
+ "loss": 0.0927,
952
+ "step": 1570
953
+ },
954
+ {
955
+ "grad_norm": 0.7350327372550964,
956
+ "learning_rate": 9.916725914371969e-05,
957
+ "loss": 0.0884,
958
+ "step": 1580
959
+ },
960
+ {
961
+ "grad_norm": 0.7852026224136353,
962
+ "learning_rate": 9.914710537764117e-05,
963
+ "loss": 0.0869,
964
+ "step": 1590
965
+ },
966
+ {
967
+ "grad_norm": 0.6127405166625977,
968
+ "learning_rate": 9.912671273816601e-05,
969
+ "loss": 0.0928,
970
+ "step": 1600
971
+ },
972
+ {
973
+ "grad_norm": 0.8078952431678772,
974
+ "learning_rate": 9.910608132441008e-05,
975
+ "loss": 0.0863,
976
+ "step": 1610
977
+ },
978
+ {
979
+ "grad_norm": 0.6758414506912231,
980
+ "learning_rate": 9.908521123664981e-05,
981
+ "loss": 0.0797,
982
+ "step": 1620
983
+ },
984
+ {
985
+ "grad_norm": 0.5950756072998047,
986
+ "learning_rate": 9.906410257632168e-05,
987
+ "loss": 0.0876,
988
+ "step": 1630
989
+ },
990
+ {
991
+ "grad_norm": 0.9129499793052673,
992
+ "learning_rate": 9.904275544602169e-05,
993
+ "loss": 0.0842,
994
+ "step": 1640
995
+ },
996
+ {
997
+ "grad_norm": 0.6972604990005493,
998
+ "learning_rate": 9.902116994950493e-05,
999
+ "loss": 0.0842,
1000
+ "step": 1650
1001
+ },
1002
+ {
1003
+ "grad_norm": 0.7520588040351868,
1004
+ "learning_rate": 9.899934619168501e-05,
1005
+ "loss": 0.0791,
1006
+ "step": 1660
1007
+ },
1008
+ {
1009
+ "grad_norm": 0.7351466417312622,
1010
+ "learning_rate": 9.89772842786336e-05,
1011
+ "loss": 0.0831,
1012
+ "step": 1670
1013
+ },
1014
+ {
1015
+ "grad_norm": 0.7723562717437744,
1016
+ "learning_rate": 9.895498431757989e-05,
1017
+ "loss": 0.0868,
1018
+ "step": 1680
1019
+ },
1020
+ {
1021
+ "grad_norm": 0.7624236345291138,
1022
+ "learning_rate": 9.893244641691006e-05,
1023
+ "loss": 0.0911,
1024
+ "step": 1690
1025
+ },
1026
+ {
1027
+ "grad_norm": 0.6789768934249878,
1028
+ "learning_rate": 9.890967068616677e-05,
1029
+ "loss": 0.0889,
1030
+ "step": 1700
1031
+ },
1032
+ {
1033
+ "grad_norm": 0.6409830451011658,
1034
+ "learning_rate": 9.888665723604864e-05,
1035
+ "loss": 0.0853,
1036
+ "step": 1710
1037
+ },
1038
+ {
1039
+ "grad_norm": 0.653907835483551,
1040
+ "learning_rate": 9.886340617840968e-05,
1041
+ "loss": 0.0814,
1042
+ "step": 1720
1043
+ },
1044
+ {
1045
+ "grad_norm": 0.767878532409668,
1046
+ "learning_rate": 9.883991762625876e-05,
1047
+ "loss": 0.0825,
1048
+ "step": 1730
1049
+ },
1050
+ {
1051
+ "grad_norm": 0.7938691973686218,
1052
+ "learning_rate": 9.881619169375908e-05,
1053
+ "loss": 0.0797,
1054
+ "step": 1740
1055
+ },
1056
+ {
1057
+ "grad_norm": 0.863423764705658,
1058
+ "learning_rate": 9.879222849622758e-05,
1059
+ "loss": 0.0868,
1060
+ "step": 1750
1061
+ },
1062
+ {
1063
+ "grad_norm": 0.7077615261077881,
1064
+ "learning_rate": 9.876802815013439e-05,
1065
+ "loss": 0.083,
1066
+ "step": 1760
1067
+ },
1068
+ {
1069
+ "grad_norm": 0.7735344171524048,
1070
+ "learning_rate": 9.87435907731023e-05,
1071
+ "loss": 0.0877,
1072
+ "step": 1770
1073
+ },
1074
+ {
1075
+ "grad_norm": 0.7485953569412231,
1076
+ "learning_rate": 9.871891648390614e-05,
1077
+ "loss": 0.0865,
1078
+ "step": 1780
1079
+ },
1080
+ {
1081
+ "grad_norm": 0.9190590381622314,
1082
+ "learning_rate": 9.869400540247223e-05,
1083
+ "loss": 0.0893,
1084
+ "step": 1790
1085
+ },
1086
+ {
1087
+ "grad_norm": 0.7002912759780884,
1088
+ "learning_rate": 9.866885764987776e-05,
1089
+ "loss": 0.0821,
1090
+ "step": 1800
1091
+ },
1092
+ {
1093
+ "grad_norm": 0.7234448790550232,
1094
+ "learning_rate": 9.86434733483503e-05,
1095
+ "loss": 0.0805,
1096
+ "step": 1810
1097
+ },
1098
+ {
1099
+ "grad_norm": 0.6578029990196228,
1100
+ "learning_rate": 9.861785262126705e-05,
1101
+ "loss": 0.0853,
1102
+ "step": 1820
1103
+ },
1104
+ {
1105
+ "grad_norm": 0.7728158831596375,
1106
+ "learning_rate": 9.85919955931544e-05,
1107
+ "loss": 0.0788,
1108
+ "step": 1830
1109
+ },
1110
+ {
1111
+ "grad_norm": 0.6882961392402649,
1112
+ "learning_rate": 9.856590238968721e-05,
1113
+ "loss": 0.0891,
1114
+ "step": 1840
1115
+ },
1116
+ {
1117
+ "grad_norm": 0.7271476984024048,
1118
+ "learning_rate": 9.853957313768824e-05,
1119
+ "loss": 0.0859,
1120
+ "step": 1850
1121
+ },
1122
+ {
1123
+ "grad_norm": 0.7411231398582458,
1124
+ "learning_rate": 9.851300796512755e-05,
1125
+ "loss": 0.0844,
1126
+ "step": 1860
1127
+ },
1128
+ {
1129
+ "grad_norm": 0.7105385661125183,
1130
+ "learning_rate": 9.848620700112188e-05,
1131
+ "loss": 0.078,
1132
+ "step": 1870
1133
+ },
1134
+ {
1135
+ "grad_norm": 0.6703158617019653,
1136
+ "learning_rate": 9.845917037593396e-05,
1137
+ "loss": 0.0796,
1138
+ "step": 1880
1139
+ },
1140
+ {
1141
+ "grad_norm": 0.6238304376602173,
1142
+ "learning_rate": 9.843189822097196e-05,
1143
+ "loss": 0.0871,
1144
+ "step": 1890
1145
+ },
1146
+ {
1147
+ "grad_norm": 0.5811266899108887,
1148
+ "learning_rate": 9.84043906687888e-05,
1149
+ "loss": 0.0779,
1150
+ "step": 1900
1151
+ },
1152
+ {
1153
+ "grad_norm": 0.6616613864898682,
1154
+ "learning_rate": 9.837664785308149e-05,
1155
+ "loss": 0.0777,
1156
+ "step": 1910
1157
+ },
1158
+ {
1159
+ "grad_norm": 0.7105765342712402,
1160
+ "learning_rate": 9.834866990869059e-05,
1161
+ "loss": 0.0826,
1162
+ "step": 1920
1163
+ },
1164
+ {
1165
+ "grad_norm": 0.7052018642425537,
1166
+ "learning_rate": 9.832045697159938e-05,
1167
+ "loss": 0.0817,
1168
+ "step": 1930
1169
+ },
1170
+ {
1171
+ "grad_norm": 0.6906279921531677,
1172
+ "learning_rate": 9.829200917893334e-05,
1173
+ "loss": 0.0815,
1174
+ "step": 1940
1175
+ },
1176
+ {
1177
+ "grad_norm": 0.7250940203666687,
1178
+ "learning_rate": 9.826332666895944e-05,
1179
+ "loss": 0.0791,
1180
+ "step": 1950
1181
+ },
1182
+ {
1183
+ "grad_norm": 0.7513362765312195,
1184
+ "learning_rate": 9.823440958108545e-05,
1185
+ "loss": 0.0793,
1186
+ "step": 1960
1187
+ },
1188
+ {
1189
+ "grad_norm": 0.7912622690200806,
1190
+ "learning_rate": 9.820525805585927e-05,
1191
+ "loss": 0.0818,
1192
+ "step": 1970
1193
+ },
1194
+ {
1195
+ "grad_norm": 0.8074571490287781,
1196
+ "learning_rate": 9.81758722349683e-05,
1197
+ "loss": 0.0828,
1198
+ "step": 1980
1199
+ },
1200
+ {
1201
+ "grad_norm": 0.7632444500923157,
1202
+ "learning_rate": 9.814625226123862e-05,
1203
+ "loss": 0.0869,
1204
+ "step": 1990
1205
+ },
1206
+ {
1207
+ "grad_norm": 0.7208681702613831,
1208
+ "learning_rate": 9.811639827863449e-05,
1209
+ "loss": 0.0828,
1210
+ "step": 2000
1211
+ },
1212
+ {
1213
+ "grad_norm": 0.763584554195404,
1214
+ "learning_rate": 9.808631043225741e-05,
1215
+ "loss": 0.0807,
1216
+ "step": 2010
1217
+ },
1218
+ {
1219
+ "grad_norm": 0.7192198634147644,
1220
+ "learning_rate": 9.805598886834567e-05,
1221
+ "loss": 0.0845,
1222
+ "step": 2020
1223
+ },
1224
+ {
1225
+ "grad_norm": 0.6376525163650513,
1226
+ "learning_rate": 9.802543373427344e-05,
1227
+ "loss": 0.0807,
1228
+ "step": 2030
1229
+ },
1230
+ {
1231
+ "grad_norm": 0.737170934677124,
1232
+ "learning_rate": 9.799464517855018e-05,
1233
+ "loss": 0.0769,
1234
+ "step": 2040
1235
+ },
1236
+ {
1237
+ "grad_norm": 0.6961826086044312,
1238
+ "learning_rate": 9.79636233508198e-05,
1239
+ "loss": 0.082,
1240
+ "step": 2050
1241
+ },
1242
+ {
1243
+ "grad_norm": 0.8413031101226807,
1244
+ "learning_rate": 9.793236840186005e-05,
1245
+ "loss": 0.0806,
1246
+ "step": 2060
1247
+ },
1248
+ {
1249
+ "grad_norm": 0.7423960566520691,
1250
+ "learning_rate": 9.790088048358175e-05,
1251
+ "loss": 0.0819,
1252
+ "step": 2070
1253
+ },
1254
+ {
1255
+ "grad_norm": 0.8054006695747375,
1256
+ "learning_rate": 9.786915974902798e-05,
1257
+ "loss": 0.0733,
1258
+ "step": 2080
1259
+ },
1260
+ {
1261
+ "grad_norm": 0.674681544303894,
1262
+ "learning_rate": 9.783720635237343e-05,
1263
+ "loss": 0.0802,
1264
+ "step": 2090
1265
+ },
1266
+ {
1267
+ "grad_norm": 0.6383704543113708,
1268
+ "learning_rate": 9.780502044892362e-05,
1269
+ "loss": 0.079,
1270
+ "step": 2100
1271
+ },
1272
+ {
1273
+ "grad_norm": 0.6757469773292542,
1274
+ "learning_rate": 9.777260219511415e-05,
1275
+ "loss": 0.076,
1276
+ "step": 2110
1277
+ },
1278
+ {
1279
+ "grad_norm": 0.823462724685669,
1280
+ "learning_rate": 9.773995174850989e-05,
1281
+ "loss": 0.0786,
1282
+ "step": 2120
1283
+ },
1284
+ {
1285
+ "grad_norm": 0.718417227268219,
1286
+ "learning_rate": 9.770706926780428e-05,
1287
+ "loss": 0.0854,
1288
+ "step": 2130
1289
+ },
1290
+ {
1291
+ "grad_norm": 0.7160131931304932,
1292
+ "learning_rate": 9.767395491281855e-05,
1293
+ "loss": 0.0773,
1294
+ "step": 2140
1295
+ },
1296
+ {
1297
+ "grad_norm": 0.7102839350700378,
1298
+ "learning_rate": 9.764060884450086e-05,
1299
+ "loss": 0.0741,
1300
+ "step": 2150
1301
+ },
1302
+ {
1303
+ "grad_norm": 0.6298885345458984,
1304
+ "learning_rate": 9.76070312249257e-05,
1305
+ "loss": 0.0726,
1306
+ "step": 2160
1307
+ },
1308
+ {
1309
+ "grad_norm": 0.7110960483551025,
1310
+ "learning_rate": 9.757322221729283e-05,
1311
+ "loss": 0.0722,
1312
+ "step": 2170
1313
+ },
1314
+ {
1315
+ "grad_norm": 0.6902167797088623,
1316
+ "learning_rate": 9.753918198592682e-05,
1317
+ "loss": 0.0737,
1318
+ "step": 2180
1319
+ },
1320
+ {
1321
+ "grad_norm": 0.721842885017395,
1322
+ "learning_rate": 9.750491069627593e-05,
1323
+ "loss": 0.0741,
1324
+ "step": 2190
1325
+ },
1326
+ {
1327
+ "grad_norm": 0.6490733027458191,
1328
+ "learning_rate": 9.747040851491149e-05,
1329
+ "loss": 0.0869,
1330
+ "step": 2200
1331
+ },
1332
+ {
1333
+ "grad_norm": 0.6896973252296448,
1334
+ "learning_rate": 9.743567560952711e-05,
1335
+ "loss": 0.081,
1336
+ "step": 2210
1337
+ },
1338
+ {
1339
+ "grad_norm": 0.6303186416625977,
1340
+ "learning_rate": 9.740071214893773e-05,
1341
+ "loss": 0.0791,
1342
+ "step": 2220
1343
+ },
1344
+ {
1345
+ "grad_norm": 0.5973021388053894,
1346
+ "learning_rate": 9.736551830307892e-05,
1347
+ "loss": 0.0726,
1348
+ "step": 2230
1349
+ },
1350
+ {
1351
+ "grad_norm": 0.6787769198417664,
1352
+ "learning_rate": 9.733009424300597e-05,
1353
+ "loss": 0.0776,
1354
+ "step": 2240
1355
+ },
1356
+ {
1357
+ "grad_norm": 0.8184981346130371,
1358
+ "learning_rate": 9.729444014089314e-05,
1359
+ "loss": 0.0771,
1360
+ "step": 2250
1361
+ },
1362
+ {
1363
+ "grad_norm": 0.6981377601623535,
1364
+ "learning_rate": 9.725855617003275e-05,
1365
+ "loss": 0.0834,
1366
+ "step": 2260
1367
+ },
1368
+ {
1369
+ "grad_norm": 0.6165905594825745,
1370
+ "learning_rate": 9.72224425048344e-05,
1371
+ "loss": 0.0767,
1372
+ "step": 2270
1373
+ },
1374
+ {
1375
+ "grad_norm": 0.6394922137260437,
1376
+ "learning_rate": 9.718609932082405e-05,
1377
+ "loss": 0.07,
1378
+ "step": 2280
1379
+ },
1380
+ {
1381
+ "grad_norm": 0.6943788528442383,
1382
+ "learning_rate": 9.714952679464323e-05,
1383
+ "loss": 0.0756,
1384
+ "step": 2290
1385
+ },
1386
+ {
1387
+ "grad_norm": 0.7439070343971252,
1388
+ "learning_rate": 9.711272510404816e-05,
1389
+ "loss": 0.0762,
1390
+ "step": 2300
1391
+ },
1392
+ {
1393
+ "grad_norm": 0.6924915909767151,
1394
+ "learning_rate": 9.70756944279089e-05,
1395
+ "loss": 0.0801,
1396
+ "step": 2310
1397
+ },
1398
+ {
1399
+ "grad_norm": 0.6574571132659912,
1400
+ "learning_rate": 9.70384349462084e-05,
1401
+ "loss": 0.0756,
1402
+ "step": 2320
1403
+ },
1404
+ {
1405
+ "grad_norm": 0.6405468583106995,
1406
+ "learning_rate": 9.700094684004182e-05,
1407
+ "loss": 0.0827,
1408
+ "step": 2330
1409
+ },
1410
+ {
1411
+ "grad_norm": 0.6328576803207397,
1412
+ "learning_rate": 9.696323029161535e-05,
1413
+ "loss": 0.071,
1414
+ "step": 2340
1415
+ },
1416
+ {
1417
+ "grad_norm": 0.6095026135444641,
1418
+ "learning_rate": 9.692528548424567e-05,
1419
+ "loss": 0.0788,
1420
+ "step": 2350
1421
+ },
1422
+ {
1423
+ "grad_norm": 0.6449398398399353,
1424
+ "learning_rate": 9.688711260235872e-05,
1425
+ "loss": 0.0772,
1426
+ "step": 2360
1427
+ },
1428
+ {
1429
+ "grad_norm": 0.6582117676734924,
1430
+ "learning_rate": 9.684871183148912e-05,
1431
+ "loss": 0.0723,
1432
+ "step": 2370
1433
+ },
1434
+ {
1435
+ "grad_norm": 0.7571628093719482,
1436
+ "learning_rate": 9.681008335827898e-05,
1437
+ "loss": 0.0766,
1438
+ "step": 2380
1439
+ },
1440
+ {
1441
+ "grad_norm": 0.7709692716598511,
1442
+ "learning_rate": 9.677122737047724e-05,
1443
+ "loss": 0.0794,
1444
+ "step": 2390
1445
+ },
1446
+ {
1447
+ "grad_norm": 0.6195625066757202,
1448
+ "learning_rate": 9.673214405693857e-05,
1449
+ "loss": 0.0736,
1450
+ "step": 2400
1451
+ },
1452
+ {
1453
+ "grad_norm": 0.7350249290466309,
1454
+ "learning_rate": 9.669283360762258e-05,
1455
+ "loss": 0.0753,
1456
+ "step": 2410
1457
+ },
1458
+ {
1459
+ "grad_norm": 0.6809076070785522,
1460
+ "learning_rate": 9.66532962135928e-05,
1461
+ "loss": 0.086,
1462
+ "step": 2420
1463
+ },
1464
+ {
1465
+ "grad_norm": 0.7334204912185669,
1466
+ "learning_rate": 9.661353206701582e-05,
1467
+ "loss": 0.0782,
1468
+ "step": 2430
1469
+ },
1470
+ {
1471
+ "grad_norm": 0.6254956722259521,
1472
+ "learning_rate": 9.657354136116035e-05,
1473
+ "loss": 0.0777,
1474
+ "step": 2440
1475
+ },
1476
+ {
1477
+ "grad_norm": 0.9059262871742249,
1478
+ "learning_rate": 9.653332429039625e-05,
1479
+ "loss": 0.0798,
1480
+ "step": 2450
1481
+ },
1482
+ {
1483
+ "grad_norm": 0.70732182264328,
1484
+ "learning_rate": 9.649288105019356e-05,
1485
+ "loss": 0.0726,
1486
+ "step": 2460
1487
+ },
1488
+ {
1489
+ "grad_norm": 0.7314443588256836,
1490
+ "learning_rate": 9.645221183712165e-05,
1491
+ "loss": 0.0779,
1492
+ "step": 2470
1493
+ },
1494
+ {
1495
+ "grad_norm": 0.7098656296730042,
1496
+ "learning_rate": 9.641131684884817e-05,
1497
+ "loss": 0.0743,
1498
+ "step": 2480
1499
+ },
1500
+ {
1501
+ "grad_norm": 0.6610057950019836,
1502
+ "learning_rate": 9.637019628413813e-05,
1503
+ "loss": 0.082,
1504
+ "step": 2490
1505
+ },
1506
+ {
1507
+ "grad_norm": 0.5614057183265686,
1508
+ "learning_rate": 9.632885034285291e-05,
1509
+ "loss": 0.0767,
1510
+ "step": 2500
1511
+ },
1512
+ {
1513
+ "grad_norm": 0.6637583374977112,
1514
+ "learning_rate": 9.628727922594931e-05,
1515
+ "loss": 0.0733,
1516
+ "step": 2510
1517
+ },
1518
+ {
1519
+ "grad_norm": 0.595269501209259,
1520
+ "learning_rate": 9.624548313547862e-05,
1521
+ "loss": 0.0764,
1522
+ "step": 2520
1523
+ },
1524
+ {
1525
+ "grad_norm": 0.8634958863258362,
1526
+ "learning_rate": 9.620346227458547e-05,
1527
+ "loss": 0.0761,
1528
+ "step": 2530
1529
+ },
1530
+ {
1531
+ "grad_norm": 0.8040565252304077,
1532
+ "learning_rate": 9.616121684750712e-05,
1533
+ "loss": 0.0709,
1534
+ "step": 2540
1535
+ },
1536
+ {
1537
+ "grad_norm": 0.5635712742805481,
1538
+ "learning_rate": 9.611874705957215e-05,
1539
+ "loss": 0.079,
1540
+ "step": 2550
1541
+ },
1542
+ {
1543
+ "grad_norm": 0.6693586707115173,
1544
+ "learning_rate": 9.607605311719972e-05,
1545
+ "loss": 0.0805,
1546
+ "step": 2560
1547
+ },
1548
+ {
1549
+ "grad_norm": 0.6621934771537781,
1550
+ "learning_rate": 9.603313522789841e-05,
1551
+ "loss": 0.0706,
1552
+ "step": 2570
1553
+ },
1554
+ {
1555
+ "grad_norm": 0.662166953086853,
1556
+ "learning_rate": 9.598999360026529e-05,
1557
+ "loss": 0.0764,
1558
+ "step": 2580
1559
+ },
1560
+ {
1561
+ "grad_norm": 0.682948887348175,
1562
+ "learning_rate": 9.59466284439849e-05,
1563
+ "loss": 0.0671,
1564
+ "step": 2590
1565
+ },
1566
+ {
1567
+ "grad_norm": 0.6689608097076416,
1568
+ "learning_rate": 9.590303996982815e-05,
1569
+ "loss": 0.0762,
1570
+ "step": 2600
1571
+ },
1572
+ {
1573
+ "grad_norm": 0.6426261067390442,
1574
+ "learning_rate": 9.585922838965145e-05,
1575
+ "loss": 0.0774,
1576
+ "step": 2610
1577
+ },
1578
+ {
1579
+ "grad_norm": 0.6558887362480164,
1580
+ "learning_rate": 9.581519391639549e-05,
1581
+ "loss": 0.0759,
1582
+ "step": 2620
1583
+ },
1584
+ {
1585
+ "grad_norm": 0.6026521921157837,
1586
+ "learning_rate": 9.577093676408439e-05,
1587
+ "loss": 0.0701,
1588
+ "step": 2630
1589
+ },
1590
+ {
1591
+ "grad_norm": 0.6800580620765686,
1592
+ "learning_rate": 9.572645714782453e-05,
1593
+ "loss": 0.0748,
1594
+ "step": 2640
1595
+ },
1596
+ {
1597
+ "grad_norm": 0.6971288323402405,
1598
+ "learning_rate": 9.568175528380354e-05,
1599
+ "loss": 0.075,
1600
+ "step": 2650
1601
+ },
1602
+ {
1603
+ "grad_norm": 0.6518421769142151,
1604
+ "learning_rate": 9.56368313892893e-05,
1605
+ "loss": 0.0752,
1606
+ "step": 2660
1607
+ },
1608
+ {
1609
+ "grad_norm": 0.5340580344200134,
1610
+ "learning_rate": 9.55916856826288e-05,
1611
+ "loss": 0.073,
1612
+ "step": 2670
1613
+ },
1614
+ {
1615
+ "grad_norm": 0.7508995532989502,
1616
+ "learning_rate": 9.554631838324713e-05,
1617
+ "loss": 0.0757,
1618
+ "step": 2680
1619
+ },
1620
+ {
1621
+ "grad_norm": 0.5836358666419983,
1622
+ "learning_rate": 9.55007297116464e-05,
1623
+ "loss": 0.0743,
1624
+ "step": 2690
1625
+ },
1626
+ {
1627
+ "grad_norm": 0.7291355133056641,
1628
+ "learning_rate": 9.545491988940472e-05,
1629
+ "loss": 0.0726,
1630
+ "step": 2700
1631
+ },
1632
+ {
1633
+ "grad_norm": 0.7412474751472473,
1634
+ "learning_rate": 9.540888913917501e-05,
1635
+ "loss": 0.0692,
1636
+ "step": 2710
1637
+ },
1638
+ {
1639
+ "grad_norm": 0.5670576691627502,
1640
+ "learning_rate": 9.536263768468401e-05,
1641
+ "loss": 0.0744,
1642
+ "step": 2720
1643
+ },
1644
+ {
1645
+ "grad_norm": 0.8577015995979309,
1646
+ "learning_rate": 9.531616575073117e-05,
1647
+ "loss": 0.0793,
1648
+ "step": 2730
1649
+ },
1650
+ {
1651
+ "grad_norm": 0.6365728378295898,
1652
+ "learning_rate": 9.526947356318754e-05,
1653
+ "loss": 0.07,
1654
+ "step": 2740
1655
+ },
1656
+ {
1657
+ "grad_norm": 0.5782703757286072,
1658
+ "learning_rate": 9.52225613489947e-05,
1659
+ "loss": 0.0678,
1660
+ "step": 2750
1661
+ },
1662
+ {
1663
+ "grad_norm": 0.6387004852294922,
1664
+ "learning_rate": 9.517542933616365e-05,
1665
+ "loss": 0.0686,
1666
+ "step": 2760
1667
+ },
1668
+ {
1669
+ "grad_norm": 0.7190784215927124,
1670
+ "learning_rate": 9.512807775377366e-05,
1671
+ "loss": 0.0762,
1672
+ "step": 2770
1673
+ },
1674
+ {
1675
+ "grad_norm": 0.679921567440033,
1676
+ "learning_rate": 9.508050683197121e-05,
1677
+ "loss": 0.072,
1678
+ "step": 2780
1679
+ },
1680
+ {
1681
+ "grad_norm": 0.5193501114845276,
1682
+ "learning_rate": 9.503271680196888e-05,
1683
+ "loss": 0.0737,
1684
+ "step": 2790
1685
+ },
1686
+ {
1687
+ "grad_norm": 0.5788854956626892,
1688
+ "learning_rate": 9.498470789604413e-05,
1689
+ "loss": 0.0784,
1690
+ "step": 2800
1691
+ },
1692
+ {
1693
+ "grad_norm": 0.5486730933189392,
1694
+ "learning_rate": 9.49364803475383e-05,
1695
+ "loss": 0.0767,
1696
+ "step": 2810
1697
+ },
1698
+ {
1699
+ "grad_norm": 0.5417566299438477,
1700
+ "learning_rate": 9.48880343908554e-05,
1701
+ "loss": 0.0732,
1702
+ "step": 2820
1703
+ },
1704
+ {
1705
+ "grad_norm": 0.6178622245788574,
1706
+ "learning_rate": 9.4839370261461e-05,
1707
+ "loss": 0.0714,
1708
+ "step": 2830
1709
+ },
1710
+ {
1711
+ "grad_norm": 0.5790642499923706,
1712
+ "learning_rate": 9.479048819588098e-05,
1713
+ "loss": 0.0721,
1714
+ "step": 2840
1715
+ },
1716
+ {
1717
+ "grad_norm": 0.5813261270523071,
1718
+ "learning_rate": 9.474138843170063e-05,
1719
+ "loss": 0.0713,
1720
+ "step": 2850
1721
+ },
1722
+ {
1723
+ "grad_norm": 0.6615369915962219,
1724
+ "learning_rate": 9.46920712075632e-05,
1725
+ "loss": 0.0811,
1726
+ "step": 2860
1727
+ },
1728
+ {
1729
+ "grad_norm": 0.6264895796775818,
1730
+ "learning_rate": 9.464253676316893e-05,
1731
+ "loss": 0.0736,
1732
+ "step": 2870
1733
+ },
1734
+ {
1735
+ "grad_norm": 0.6997485160827637,
1736
+ "learning_rate": 9.459278533927384e-05,
1737
+ "loss": 0.0713,
1738
+ "step": 2880
1739
+ },
1740
+ {
1741
+ "grad_norm": 0.619152307510376,
1742
+ "learning_rate": 9.454281717768854e-05,
1743
+ "loss": 0.0741,
1744
+ "step": 2890
1745
+ },
1746
+ {
1747
+ "grad_norm": 0.6027862429618835,
1748
+ "learning_rate": 9.449263252127708e-05,
1749
+ "loss": 0.0664,
1750
+ "step": 2900
1751
+ },
1752
+ {
1753
+ "grad_norm": 0.8105010986328125,
1754
+ "learning_rate": 9.444223161395573e-05,
1755
+ "loss": 0.0671,
1756
+ "step": 2910
1757
+ },
1758
+ {
1759
+ "grad_norm": 0.6472406387329102,
1760
+ "learning_rate": 9.439161470069184e-05,
1761
+ "loss": 0.0756,
1762
+ "step": 2920
1763
+ },
1764
+ {
1765
+ "grad_norm": 0.6002207398414612,
1766
+ "learning_rate": 9.43407820275026e-05,
1767
+ "loss": 0.0767,
1768
+ "step": 2930
1769
+ },
1770
+ {
1771
+ "grad_norm": 0.644810140132904,
1772
+ "learning_rate": 9.428973384145396e-05,
1773
+ "loss": 0.0795,
1774
+ "step": 2940
1775
+ },
1776
+ {
1777
+ "grad_norm": 0.6397400498390198,
1778
+ "learning_rate": 9.423847039065922e-05,
1779
+ "loss": 0.0705,
1780
+ "step": 2950
1781
+ },
1782
+ {
1783
+ "grad_norm": 0.6125301718711853,
1784
+ "learning_rate": 9.418699192427805e-05,
1785
+ "loss": 0.0722,
1786
+ "step": 2960
1787
+ },
1788
+ {
1789
+ "grad_norm": 0.6119952201843262,
1790
+ "learning_rate": 9.41352986925151e-05,
1791
+ "loss": 0.069,
1792
+ "step": 2970
1793
+ },
1794
+ {
1795
+ "grad_norm": 0.619590163230896,
1796
+ "learning_rate": 9.408339094661895e-05,
1797
+ "loss": 0.0658,
1798
+ "step": 2980
1799
+ },
1800
+ {
1801
+ "grad_norm": 0.6107400059700012,
1802
+ "learning_rate": 9.40312689388807e-05,
1803
+ "loss": 0.0673,
1804
+ "step": 2990
1805
+ },
1806
+ {
1807
+ "grad_norm": 0.6754778623580933,
1808
+ "learning_rate": 9.397893292263292e-05,
1809
+ "loss": 0.0681,
1810
+ "step": 3000
1811
+ },
1812
+ {
1813
+ "grad_norm": 0.7630986571311951,
1814
+ "learning_rate": 9.392638315224829e-05,
1815
+ "loss": 0.0673,
1816
+ "step": 3010
1817
+ },
1818
+ {
1819
+ "grad_norm": 0.6023984551429749,
1820
+ "learning_rate": 9.387361988313846e-05,
1821
+ "loss": 0.0684,
1822
+ "step": 3020
1823
+ },
1824
+ {
1825
+ "grad_norm": 0.6075767874717712,
1826
+ "learning_rate": 9.38206433717527e-05,
1827
+ "loss": 0.0649,
1828
+ "step": 3030
1829
+ },
1830
+ {
1831
+ "grad_norm": 0.6851189136505127,
1832
+ "learning_rate": 9.376745387557681e-05,
1833
+ "loss": 0.0706,
1834
+ "step": 3040
1835
+ },
1836
+ {
1837
+ "grad_norm": 0.5601722598075867,
1838
+ "learning_rate": 9.371405165313169e-05,
1839
+ "loss": 0.0639,
1840
+ "step": 3050
1841
+ },
1842
+ {
1843
+ "grad_norm": 0.6507348418235779,
1844
+ "learning_rate": 9.366043696397222e-05,
1845
+ "loss": 0.0676,
1846
+ "step": 3060
1847
+ },
1848
+ {
1849
+ "grad_norm": 0.6262578964233398,
1850
+ "learning_rate": 9.360661006868592e-05,
1851
+ "loss": 0.0658,
1852
+ "step": 3070
1853
+ },
1854
+ {
1855
+ "grad_norm": 0.5408547520637512,
1856
+ "learning_rate": 9.355257122889173e-05,
1857
+ "loss": 0.0664,
1858
+ "step": 3080
1859
+ },
1860
+ {
1861
+ "grad_norm": 0.5795503854751587,
1862
+ "learning_rate": 9.349832070723871e-05,
1863
+ "loss": 0.0758,
1864
+ "step": 3090
1865
+ },
1866
+ {
1867
+ "grad_norm": 0.6226188540458679,
1868
+ "learning_rate": 9.34438587674048e-05,
1869
+ "loss": 0.0755,
1870
+ "step": 3100
1871
+ },
1872
+ {
1873
+ "grad_norm": 0.6497605443000793,
1874
+ "learning_rate": 9.338918567409545e-05,
1875
+ "loss": 0.0695,
1876
+ "step": 3110
1877
+ },
1878
+ {
1879
+ "grad_norm": 0.6185932755470276,
1880
+ "learning_rate": 9.333430169304247e-05,
1881
+ "loss": 0.0707,
1882
+ "step": 3120
1883
+ },
1884
+ {
1885
+ "grad_norm": 0.5427372455596924,
1886
+ "learning_rate": 9.327920709100259e-05,
1887
+ "loss": 0.0703,
1888
+ "step": 3130
1889
+ },
1890
+ {
1891
+ "grad_norm": 0.5989561080932617,
1892
+ "learning_rate": 9.322390213575631e-05,
1893
+ "loss": 0.0712,
1894
+ "step": 3140
1895
+ },
1896
+ {
1897
+ "grad_norm": 0.5449648499488831,
1898
+ "learning_rate": 9.316838709610648e-05,
1899
+ "loss": 0.0711,
1900
+ "step": 3150
1901
+ },
1902
+ {
1903
+ "grad_norm": 0.654105007648468,
1904
+ "learning_rate": 9.311266224187706e-05,
1905
+ "loss": 0.0672,
1906
+ "step": 3160
1907
+ },
1908
+ {
1909
+ "grad_norm": 0.5506020188331604,
1910
+ "learning_rate": 9.305672784391175e-05,
1911
+ "loss": 0.0672,
1912
+ "step": 3170
1913
+ },
1914
+ {
1915
+ "grad_norm": 0.503079354763031,
1916
+ "learning_rate": 9.300058417407276e-05,
1917
+ "loss": 0.0657,
1918
+ "step": 3180
1919
+ },
1920
+ {
1921
+ "grad_norm": 0.5641545653343201,
1922
+ "learning_rate": 9.29442315052394e-05,
1923
+ "loss": 0.0663,
1924
+ "step": 3190
1925
+ },
1926
+ {
1927
+ "grad_norm": 0.5445094108581543,
1928
+ "learning_rate": 9.288767011130684e-05,
1929
+ "loss": 0.0707,
1930
+ "step": 3200
1931
+ },
1932
+ {
1933
+ "grad_norm": 0.6540117263793945,
1934
+ "learning_rate": 9.283090026718466e-05,
1935
+ "loss": 0.068,
1936
+ "step": 3210
1937
+ },
1938
+ {
1939
+ "grad_norm": 0.7890759110450745,
1940
+ "learning_rate": 9.277392224879568e-05,
1941
+ "loss": 0.0637,
1942
+ "step": 3220
1943
+ },
1944
+ {
1945
+ "grad_norm": 0.5481688976287842,
1946
+ "learning_rate": 9.271673633307445e-05,
1947
+ "loss": 0.0618,
1948
+ "step": 3230
1949
+ },
1950
+ {
1951
+ "grad_norm": 0.5165878534317017,
1952
+ "learning_rate": 9.265934279796602e-05,
1953
+ "loss": 0.0686,
1954
+ "step": 3240
1955
+ },
1956
+ {
1957
+ "grad_norm": 0.6548216342926025,
1958
+ "learning_rate": 9.260174192242453e-05,
1959
+ "loss": 0.0701,
1960
+ "step": 3250
1961
+ },
1962
+ {
1963
+ "grad_norm": 0.8453028798103333,
1964
+ "learning_rate": 9.254393398641185e-05,
1965
+ "loss": 0.075,
1966
+ "step": 3260
1967
+ },
1968
+ {
1969
+ "grad_norm": 0.6304769515991211,
1970
+ "learning_rate": 9.248591927089628e-05,
1971
+ "loss": 0.0709,
1972
+ "step": 3270
1973
+ },
1974
+ {
1975
+ "grad_norm": 0.6732708215713501,
1976
+ "learning_rate": 9.242769805785115e-05,
1977
+ "loss": 0.0709,
1978
+ "step": 3280
1979
+ },
1980
+ {
1981
+ "grad_norm": 0.6626738905906677,
1982
+ "learning_rate": 9.236927063025342e-05,
1983
+ "loss": 0.0838,
1984
+ "step": 3290
1985
+ },
1986
+ {
1987
+ "grad_norm": 0.5738134384155273,
1988
+ "learning_rate": 9.231063727208234e-05,
1989
+ "loss": 0.0718,
1990
+ "step": 3300
1991
+ },
1992
+ {
1993
+ "grad_norm": 0.5651624202728271,
1994
+ "learning_rate": 9.225179826831807e-05,
1995
+ "loss": 0.0708,
1996
+ "step": 3310
1997
+ },
1998
+ {
1999
+ "grad_norm": 0.6571042537689209,
2000
+ "learning_rate": 9.219275390494024e-05,
2001
+ "loss": 0.0752,
2002
+ "step": 3320
2003
+ },
2004
+ {
2005
+ "grad_norm": 0.5358458757400513,
2006
+ "learning_rate": 9.213350446892668e-05,
2007
+ "loss": 0.0683,
2008
+ "step": 3330
2009
+ },
2010
+ {
2011
+ "grad_norm": 0.6220871806144714,
2012
+ "learning_rate": 9.207405024825186e-05,
2013
+ "loss": 0.0713,
2014
+ "step": 3340
2015
+ },
2016
+ {
2017
+ "grad_norm": 0.5692563056945801,
2018
+ "learning_rate": 9.201439153188569e-05,
2019
+ "loss": 0.0698,
2020
+ "step": 3350
2021
+ },
2022
+ {
2023
+ "grad_norm": 0.644374430179596,
2024
+ "learning_rate": 9.19545286097919e-05,
2025
+ "loss": 0.0682,
2026
+ "step": 3360
2027
+ },
2028
+ {
2029
+ "grad_norm": 0.5985931754112244,
2030
+ "learning_rate": 9.189446177292679e-05,
2031
+ "loss": 0.0744,
2032
+ "step": 3370
2033
+ },
2034
+ {
2035
+ "grad_norm": 0.6280204653739929,
2036
+ "learning_rate": 9.183419131323778e-05,
2037
+ "loss": 0.0667,
2038
+ "step": 3380
2039
+ },
2040
+ {
2041
+ "grad_norm": 0.6723644137382507,
2042
+ "learning_rate": 9.177371752366191e-05,
2043
+ "loss": 0.0648,
2044
+ "step": 3390
2045
+ },
2046
+ {
2047
+ "grad_norm": 0.6714949011802673,
2048
+ "learning_rate": 9.171304069812454e-05,
2049
+ "loss": 0.0767,
2050
+ "step": 3400
2051
+ },
2052
+ {
2053
+ "grad_norm": 0.7225590944290161,
2054
+ "learning_rate": 9.165216113153782e-05,
2055
+ "loss": 0.0644,
2056
+ "step": 3410
2057
+ },
2058
+ {
2059
+ "grad_norm": 0.5510388612747192,
2060
+ "learning_rate": 9.159107911979936e-05,
2061
+ "loss": 0.0637,
2062
+ "step": 3420
2063
+ },
2064
+ {
2065
+ "grad_norm": 0.47715097665786743,
2066
+ "learning_rate": 9.152979495979063e-05,
2067
+ "loss": 0.0694,
2068
+ "step": 3430
2069
+ },
2070
+ {
2071
+ "grad_norm": 0.6667302846908569,
2072
+ "learning_rate": 9.146830894937571e-05,
2073
+ "loss": 0.0739,
2074
+ "step": 3440
2075
+ },
2076
+ {
2077
+ "grad_norm": 0.5869659781455994,
2078
+ "learning_rate": 9.140662138739969e-05,
2079
+ "loss": 0.0708,
2080
+ "step": 3450
2081
+ },
2082
+ {
2083
+ "grad_norm": 0.5377545952796936,
2084
+ "learning_rate": 9.134473257368732e-05,
2085
+ "loss": 0.0744,
2086
+ "step": 3460
2087
+ },
2088
+ {
2089
+ "grad_norm": 0.6936116218566895,
2090
+ "learning_rate": 9.128264280904145e-05,
2091
+ "loss": 0.0606,
2092
+ "step": 3470
2093
+ },
2094
+ {
2095
+ "grad_norm": 0.5792559385299683,
2096
+ "learning_rate": 9.122035239524169e-05,
2097
+ "loss": 0.0711,
2098
+ "step": 3480
2099
+ },
2100
+ {
2101
+ "grad_norm": 0.5261418223381042,
2102
+ "learning_rate": 9.115786163504285e-05,
2103
+ "loss": 0.0628,
2104
+ "step": 3490
2105
+ },
2106
+ {
2107
+ "grad_norm": 0.6396857500076294,
2108
+ "learning_rate": 9.10951708321735e-05,
2109
+ "loss": 0.0648,
2110
+ "step": 3500
2111
+ },
2112
+ {
2113
+ "grad_norm": 0.5282208919525146,
2114
+ "learning_rate": 9.10322802913345e-05,
2115
+ "loss": 0.0645,
2116
+ "step": 3510
2117
+ },
2118
+ {
2119
+ "grad_norm": 0.7117626667022705,
2120
+ "learning_rate": 9.096919031819751e-05,
2121
+ "loss": 0.0702,
2122
+ "step": 3520
2123
+ },
2124
+ {
2125
+ "grad_norm": 0.6305633187294006,
2126
+ "learning_rate": 9.090590121940348e-05,
2127
+ "loss": 0.0706,
2128
+ "step": 3530
2129
+ },
2130
+ {
2131
+ "grad_norm": 0.5667370557785034,
2132
+ "learning_rate": 9.084241330256121e-05,
2133
+ "loss": 0.0639,
2134
+ "step": 3540
2135
+ },
2136
+ {
2137
+ "grad_norm": 0.608974277973175,
2138
+ "learning_rate": 9.077872687624586e-05,
2139
+ "loss": 0.0647,
2140
+ "step": 3550
2141
+ },
2142
+ {
2143
+ "grad_norm": 0.6044687032699585,
2144
+ "learning_rate": 9.071484224999735e-05,
2145
+ "loss": 0.067,
2146
+ "step": 3560
2147
+ },
2148
+ {
2149
+ "grad_norm": 0.5174039006233215,
2150
+ "learning_rate": 9.0650759734319e-05,
2151
+ "loss": 0.0661,
2152
+ "step": 3570
2153
+ },
2154
+ {
2155
+ "grad_norm": 0.6718793511390686,
2156
+ "learning_rate": 9.05864796406759e-05,
2157
+ "loss": 0.0697,
2158
+ "step": 3580
2159
+ },
2160
+ {
2161
+ "grad_norm": 0.6483207941055298,
2162
+ "learning_rate": 9.052200228149343e-05,
2163
+ "loss": 0.0715,
2164
+ "step": 3590
2165
+ },
2166
+ {
2167
+ "grad_norm": 0.5423234105110168,
2168
+ "learning_rate": 9.04573279701558e-05,
2169
+ "loss": 0.0719,
2170
+ "step": 3600
2171
+ },
2172
+ {
2173
+ "grad_norm": 0.5089258551597595,
2174
+ "learning_rate": 9.039245702100448e-05,
2175
+ "loss": 0.064,
2176
+ "step": 3610
2177
+ },
2178
+ {
2179
+ "grad_norm": 0.6893640756607056,
2180
+ "learning_rate": 9.032738974933664e-05,
2181
+ "loss": 0.0638,
2182
+ "step": 3620
2183
+ },
2184
+ {
2185
+ "grad_norm": 0.8346790671348572,
2186
+ "learning_rate": 9.026212647140365e-05,
2187
+ "loss": 0.0648,
2188
+ "step": 3630
2189
+ },
2190
+ {
2191
+ "grad_norm": 0.49496138095855713,
2192
+ "learning_rate": 9.019666750440956e-05,
2193
+ "loss": 0.0652,
2194
+ "step": 3640
2195
+ },
2196
+ {
2197
+ "grad_norm": 0.5753434300422668,
2198
+ "learning_rate": 9.013101316650956e-05,
2199
+ "loss": 0.069,
2200
+ "step": 3650
2201
+ },
2202
+ {
2203
+ "grad_norm": 0.4407230615615845,
2204
+ "learning_rate": 9.00651637768084e-05,
2205
+ "loss": 0.0636,
2206
+ "step": 3660
2207
+ },
2208
+ {
2209
+ "grad_norm": 0.4615444839000702,
2210
+ "learning_rate": 8.999911965535885e-05,
2211
+ "loss": 0.0684,
2212
+ "step": 3670
2213
+ },
2214
+ {
2215
+ "grad_norm": 0.4918903708457947,
2216
+ "learning_rate": 8.993288112316014e-05,
2217
+ "loss": 0.0621,
2218
+ "step": 3680
2219
+ },
2220
+ {
2221
+ "grad_norm": 0.5506250858306885,
2222
+ "learning_rate": 8.986644850215644e-05,
2223
+ "loss": 0.0671,
2224
+ "step": 3690
2225
+ },
2226
+ {
2227
+ "grad_norm": 0.5882449150085449,
2228
+ "learning_rate": 8.979982211523523e-05,
2229
+ "loss": 0.0665,
2230
+ "step": 3700
2231
+ },
2232
+ {
2233
+ "grad_norm": 0.5929926633834839,
2234
+ "learning_rate": 8.97330022862258e-05,
2235
+ "loss": 0.0658,
2236
+ "step": 3710
2237
+ },
2238
+ {
2239
+ "grad_norm": 0.5656607747077942,
2240
+ "learning_rate": 8.96659893398976e-05,
2241
+ "loss": 0.063,
2242
+ "step": 3720
2243
+ },
2244
+ {
2245
+ "grad_norm": 0.5676708817481995,
2246
+ "learning_rate": 8.959878360195876e-05,
2247
+ "loss": 0.0658,
2248
+ "step": 3730
2249
+ },
2250
+ {
2251
+ "grad_norm": 0.5738428235054016,
2252
+ "learning_rate": 8.953138539905438e-05,
2253
+ "loss": 0.0648,
2254
+ "step": 3740
2255
+ },
2256
+ {
2257
+ "grad_norm": 0.5268324613571167,
2258
+ "learning_rate": 8.946379505876506e-05,
2259
+ "loss": 0.0695,
2260
+ "step": 3750
2261
+ },
2262
+ {
2263
+ "grad_norm": 0.7279912233352661,
2264
+ "learning_rate": 8.939601290960527e-05,
2265
+ "loss": 0.0662,
2266
+ "step": 3760
2267
+ },
2268
+ {
2269
+ "grad_norm": 0.5215412974357605,
2270
+ "learning_rate": 8.932803928102167e-05,
2271
+ "loss": 0.0679,
2272
+ "step": 3770
2273
+ },
2274
+ {
2275
+ "grad_norm": 0.6495506763458252,
2276
+ "learning_rate": 8.925987450339168e-05,
2277
+ "loss": 0.0636,
2278
+ "step": 3780
2279
+ },
2280
+ {
2281
+ "grad_norm": 0.6008911728858948,
2282
+ "learning_rate": 8.919151890802172e-05,
2283
+ "loss": 0.0677,
2284
+ "step": 3790
2285
+ },
2286
+ {
2287
+ "grad_norm": 0.6184180378913879,
2288
+ "learning_rate": 8.912297282714564e-05,
2289
+ "loss": 0.0704,
2290
+ "step": 3800
2291
+ },
2292
+ {
2293
+ "grad_norm": 0.5512374639511108,
2294
+ "learning_rate": 8.905423659392316e-05,
2295
+ "loss": 0.0607,
2296
+ "step": 3810
2297
+ },
2298
+ {
2299
+ "grad_norm": 0.5496158599853516,
2300
+ "learning_rate": 8.898531054243822e-05,
2301
+ "loss": 0.0645,
2302
+ "step": 3820
2303
+ },
2304
+ {
2305
+ "grad_norm": 0.5281623005867004,
2306
+ "learning_rate": 8.891619500769729e-05,
2307
+ "loss": 0.0663,
2308
+ "step": 3830
2309
+ },
2310
+ {
2311
+ "grad_norm": 0.5163659453392029,
2312
+ "learning_rate": 8.884689032562785e-05,
2313
+ "loss": 0.0631,
2314
+ "step": 3840
2315
+ },
2316
+ {
2317
+ "grad_norm": 0.61015385389328,
2318
+ "learning_rate": 8.87773968330767e-05,
2319
+ "loss": 0.0621,
2320
+ "step": 3850
2321
+ },
2322
+ {
2323
+ "grad_norm": 0.5940917134284973,
2324
+ "learning_rate": 8.870771486780832e-05,
2325
+ "loss": 0.0654,
2326
+ "step": 3860
2327
+ },
2328
+ {
2329
+ "grad_norm": 0.4781402051448822,
2330
+ "learning_rate": 8.863784476850322e-05,
2331
+ "loss": 0.0656,
2332
+ "step": 3870
2333
+ },
2334
+ {
2335
+ "grad_norm": 0.5629350543022156,
2336
+ "learning_rate": 8.856778687475635e-05,
2337
+ "loss": 0.0631,
2338
+ "step": 3880
2339
+ },
2340
+ {
2341
+ "grad_norm": 0.49277231097221375,
2342
+ "learning_rate": 8.849754152707541e-05,
2343
+ "loss": 0.0608,
2344
+ "step": 3890
2345
+ },
2346
+ {
2347
+ "grad_norm": 0.4736921191215515,
2348
+ "learning_rate": 8.842710906687916e-05,
2349
+ "loss": 0.0597,
2350
+ "step": 3900
2351
+ },
2352
+ {
2353
+ "grad_norm": 0.5930362939834595,
2354
+ "learning_rate": 8.83564898364958e-05,
2355
+ "loss": 0.0694,
2356
+ "step": 3910
2357
+ },
2358
+ {
2359
+ "grad_norm": 0.609360933303833,
2360
+ "learning_rate": 8.828568417916136e-05,
2361
+ "loss": 0.072,
2362
+ "step": 3920
2363
+ },
2364
+ {
2365
+ "grad_norm": 0.5193405151367188,
2366
+ "learning_rate": 8.821469243901794e-05,
2367
+ "loss": 0.0603,
2368
+ "step": 3930
2369
+ },
2370
+ {
2371
+ "grad_norm": 0.6480633020401001,
2372
+ "learning_rate": 8.814351496111201e-05,
2373
+ "loss": 0.0652,
2374
+ "step": 3940
2375
+ },
2376
+ {
2377
+ "grad_norm": 0.5353022217750549,
2378
+ "learning_rate": 8.807215209139293e-05,
2379
+ "loss": 0.0618,
2380
+ "step": 3950
2381
+ },
2382
+ {
2383
+ "grad_norm": 0.6103038787841797,
2384
+ "learning_rate": 8.8000604176711e-05,
2385
+ "loss": 0.0668,
2386
+ "step": 3960
2387
+ },
2388
+ {
2389
+ "grad_norm": 0.5791677832603455,
2390
+ "learning_rate": 8.792887156481598e-05,
2391
+ "loss": 0.0667,
2392
+ "step": 3970
2393
+ },
2394
+ {
2395
+ "grad_norm": 0.5314576029777527,
2396
+ "learning_rate": 8.785695460435534e-05,
2397
+ "loss": 0.0739,
2398
+ "step": 3980
2399
+ },
2400
+ {
2401
+ "grad_norm": 0.6145054697990417,
2402
+ "learning_rate": 8.778485364487248e-05,
2403
+ "loss": 0.0719,
2404
+ "step": 3990
2405
+ },
2406
+ {
2407
+ "grad_norm": 0.528359055519104,
2408
+ "learning_rate": 8.771256903680519e-05,
2409
+ "loss": 0.0662,
2410
+ "step": 4000
2411
+ },
2412
+ {
2413
+ "grad_norm": 0.45334354043006897,
2414
+ "learning_rate": 8.764010113148382e-05,
2415
+ "loss": 0.0672,
2416
+ "step": 4010
2417
+ },
2418
+ {
2419
+ "grad_norm": 0.543138861656189,
2420
+ "learning_rate": 8.756745028112959e-05,
2421
+ "loss": 0.0631,
2422
+ "step": 4020
2423
+ },
2424
+ {
2425
+ "grad_norm": 0.5311511158943176,
2426
+ "learning_rate": 8.749461683885296e-05,
2427
+ "loss": 0.0619,
2428
+ "step": 4030
2429
+ },
2430
+ {
2431
+ "grad_norm": 0.5041663646697998,
2432
+ "learning_rate": 8.742160115865179e-05,
2433
+ "loss": 0.0615,
2434
+ "step": 4040
2435
+ },
2436
+ {
2437
+ "grad_norm": 0.44580286741256714,
2438
+ "learning_rate": 8.734840359540974e-05,
2439
+ "loss": 0.0559,
2440
+ "step": 4050
2441
+ },
2442
+ {
2443
+ "grad_norm": 0.5962316393852234,
2444
+ "learning_rate": 8.727502450489446e-05,
2445
+ "loss": 0.063,
2446
+ "step": 4060
2447
+ },
2448
+ {
2449
+ "grad_norm": 0.5856122374534607,
2450
+ "learning_rate": 8.720146424375591e-05,
2451
+ "loss": 0.0605,
2452
+ "step": 4070
2453
+ },
2454
+ {
2455
+ "grad_norm": 0.564177930355072,
2456
+ "learning_rate": 8.712772316952458e-05,
2457
+ "loss": 0.0656,
2458
+ "step": 4080
2459
+ },
2460
+ {
2461
+ "grad_norm": 0.5987861156463623,
2462
+ "learning_rate": 8.705380164060982e-05,
2463
+ "loss": 0.0637,
2464
+ "step": 4090
2465
+ },
2466
+ {
2467
+ "grad_norm": 0.5117592811584473,
2468
+ "learning_rate": 8.697970001629799e-05,
2469
+ "loss": 0.0668,
2470
+ "step": 4100
2471
+ },
2472
+ {
2473
+ "grad_norm": 0.5237540602684021,
2474
+ "learning_rate": 8.690541865675084e-05,
2475
+ "loss": 0.0656,
2476
+ "step": 4110
2477
+ },
2478
+ {
2479
+ "grad_norm": 0.6092665791511536,
2480
+ "learning_rate": 8.68309579230037e-05,
2481
+ "loss": 0.0647,
2482
+ "step": 4120
2483
+ },
2484
+ {
2485
+ "grad_norm": 0.5419392585754395,
2486
+ "learning_rate": 8.675631817696372e-05,
2487
+ "loss": 0.0611,
2488
+ "step": 4130
2489
+ },
2490
+ {
2491
+ "grad_norm": 0.4946824014186859,
2492
+ "learning_rate": 8.668149978140808e-05,
2493
+ "loss": 0.0706,
2494
+ "step": 4140
2495
+ },
2496
+ {
2497
+ "grad_norm": 0.6564487814903259,
2498
+ "learning_rate": 8.66065030999823e-05,
2499
+ "loss": 0.0596,
2500
+ "step": 4150
2501
+ },
2502
+ {
2503
+ "grad_norm": 0.497086763381958,
2504
+ "learning_rate": 8.653132849719845e-05,
2505
+ "loss": 0.0663,
2506
+ "step": 4160
2507
+ },
2508
+ {
2509
+ "grad_norm": 0.5801936388015747,
2510
+ "learning_rate": 8.64559763384333e-05,
2511
+ "loss": 0.0622,
2512
+ "step": 4170
2513
+ },
2514
+ {
2515
+ "grad_norm": 0.5763471722602844,
2516
+ "learning_rate": 8.638044698992669e-05,
2517
+ "loss": 0.0652,
2518
+ "step": 4180
2519
+ },
2520
+ {
2521
+ "grad_norm": 0.5335597395896912,
2522
+ "learning_rate": 8.630474081877959e-05,
2523
+ "loss": 0.0583,
2524
+ "step": 4190
2525
+ },
2526
+ {
2527
+ "grad_norm": 0.5130449533462524,
2528
+ "learning_rate": 8.62288581929525e-05,
2529
+ "loss": 0.059,
2530
+ "step": 4200
2531
+ },
2532
+ {
2533
+ "grad_norm": 0.6195492744445801,
2534
+ "learning_rate": 8.615279948126343e-05,
2535
+ "loss": 0.0621,
2536
+ "step": 4210
2537
+ },
2538
+ {
2539
+ "grad_norm": 0.5328609943389893,
2540
+ "learning_rate": 8.60765650533863e-05,
2541
+ "loss": 0.0595,
2542
+ "step": 4220
2543
+ },
2544
+ {
2545
+ "grad_norm": 0.6657204031944275,
2546
+ "learning_rate": 8.60001552798491e-05,
2547
+ "loss": 0.0592,
2548
+ "step": 4230
2549
+ },
2550
+ {
2551
+ "grad_norm": 0.5727778077125549,
2552
+ "learning_rate": 8.592357053203202e-05,
2553
+ "loss": 0.0649,
2554
+ "step": 4240
2555
+ },
2556
+ {
2557
+ "grad_norm": 0.5315781235694885,
2558
+ "learning_rate": 8.58468111821657e-05,
2559
+ "loss": 0.0662,
2560
+ "step": 4250
2561
+ },
2562
+ {
2563
+ "grad_norm": 0.5774106979370117,
2564
+ "learning_rate": 8.576987760332943e-05,
2565
+ "loss": 0.0666,
2566
+ "step": 4260
2567
+ },
2568
+ {
2569
+ "grad_norm": 0.5133674144744873,
2570
+ "learning_rate": 8.56927701694493e-05,
2571
+ "loss": 0.0599,
2572
+ "step": 4270
2573
+ },
2574
+ {
2575
+ "grad_norm": 0.5090412497520447,
2576
+ "learning_rate": 8.561548925529643e-05,
2577
+ "loss": 0.064,
2578
+ "step": 4280
2579
+ },
2580
+ {
2581
+ "grad_norm": 0.5547593235969543,
2582
+ "learning_rate": 8.553803523648506e-05,
2583
+ "loss": 0.0617,
2584
+ "step": 4290
2585
+ },
2586
+ {
2587
+ "grad_norm": 0.5150351524353027,
2588
+ "learning_rate": 8.546040848947086e-05,
2589
+ "loss": 0.0643,
2590
+ "step": 4300
2591
+ },
2592
+ {
2593
+ "grad_norm": 0.5246621370315552,
2594
+ "learning_rate": 8.538260939154894e-05,
2595
+ "loss": 0.0639,
2596
+ "step": 4310
2597
+ },
2598
+ {
2599
+ "grad_norm": 0.5461138486862183,
2600
+ "learning_rate": 8.530463832085218e-05,
2601
+ "loss": 0.0582,
2602
+ "step": 4320
2603
+ },
2604
+ {
2605
+ "grad_norm": 0.49885794520378113,
2606
+ "learning_rate": 8.522649565634927e-05,
2607
+ "loss": 0.0618,
2608
+ "step": 4330
2609
+ },
2610
+ {
2611
+ "grad_norm": 0.48420804738998413,
2612
+ "learning_rate": 8.51481817778429e-05,
2613
+ "loss": 0.0607,
2614
+ "step": 4340
2615
+ },
2616
+ {
2617
+ "grad_norm": 0.5489194393157959,
2618
+ "learning_rate": 8.506969706596797e-05,
2619
+ "loss": 0.0576,
2620
+ "step": 4350
2621
+ },
2622
+ {
2623
+ "grad_norm": 0.6276731491088867,
2624
+ "learning_rate": 8.499104190218964e-05,
2625
+ "loss": 0.057,
2626
+ "step": 4360
2627
+ },
2628
+ {
2629
+ "grad_norm": 0.4996776878833771,
2630
+ "learning_rate": 8.49122166688016e-05,
2631
+ "loss": 0.0626,
2632
+ "step": 4370
2633
+ },
2634
+ {
2635
+ "grad_norm": 0.49057215452194214,
2636
+ "learning_rate": 8.483322174892404e-05,
2637
+ "loss": 0.0658,
2638
+ "step": 4380
2639
+ },
2640
+ {
2641
+ "grad_norm": 0.491436630487442,
2642
+ "learning_rate": 8.475405752650199e-05,
2643
+ "loss": 0.0614,
2644
+ "step": 4390
2645
+ },
2646
+ {
2647
+ "grad_norm": 0.49525755643844604,
2648
+ "learning_rate": 8.467472438630328e-05,
2649
+ "loss": 0.0658,
2650
+ "step": 4400
2651
+ },
2652
+ {
2653
+ "grad_norm": 0.4709268808364868,
2654
+ "learning_rate": 8.459522271391682e-05,
2655
+ "loss": 0.0647,
2656
+ "step": 4410
2657
+ },
2658
+ {
2659
+ "grad_norm": 0.5747040510177612,
2660
+ "learning_rate": 8.451555289575057e-05,
2661
+ "loss": 0.0582,
2662
+ "step": 4420
2663
+ },
2664
+ {
2665
+ "grad_norm": 0.5364845395088196,
2666
+ "learning_rate": 8.443571531902981e-05,
2667
+ "loss": 0.0668,
2668
+ "step": 4430
2669
+ },
2670
+ {
2671
+ "grad_norm": 0.4799308180809021,
2672
+ "learning_rate": 8.435571037179512e-05,
2673
+ "loss": 0.0609,
2674
+ "step": 4440
2675
+ },
2676
+ {
2677
+ "grad_norm": 0.525842010974884,
2678
+ "learning_rate": 8.427553844290062e-05,
2679
+ "loss": 0.0624,
2680
+ "step": 4450
2681
+ },
2682
+ {
2683
+ "grad_norm": 0.6582574844360352,
2684
+ "learning_rate": 8.419519992201201e-05,
2685
+ "loss": 0.0607,
2686
+ "step": 4460
2687
+ },
2688
+ {
2689
+ "grad_norm": 0.4682302176952362,
2690
+ "learning_rate": 8.411469519960469e-05,
2691
+ "loss": 0.0599,
2692
+ "step": 4470
2693
+ },
2694
+ {
2695
+ "grad_norm": 0.6320589184761047,
2696
+ "learning_rate": 8.403402466696182e-05,
2697
+ "loss": 0.0641,
2698
+ "step": 4480
2699
+ },
2700
+ {
2701
+ "grad_norm": 0.6417620778083801,
2702
+ "learning_rate": 8.395318871617255e-05,
2703
+ "loss": 0.0569,
2704
+ "step": 4490
2705
+ },
2706
+ {
2707
+ "grad_norm": 0.6075902581214905,
2708
+ "learning_rate": 8.387218774012992e-05,
2709
+ "loss": 0.0666,
2710
+ "step": 4500
2711
+ },
2712
+ {
2713
+ "grad_norm": 0.4782496392726898,
2714
+ "learning_rate": 8.379102213252915e-05,
2715
+ "loss": 0.0556,
2716
+ "step": 4510
2717
+ },
2718
+ {
2719
+ "grad_norm": 0.5786132216453552,
2720
+ "learning_rate": 8.370969228786556e-05,
2721
+ "loss": 0.0584,
2722
+ "step": 4520
2723
+ },
2724
+ {
2725
+ "grad_norm": 0.502007007598877,
2726
+ "learning_rate": 8.362819860143275e-05,
2727
+ "loss": 0.0592,
2728
+ "step": 4530
2729
+ },
2730
+ {
2731
+ "grad_norm": 0.5026768445968628,
2732
+ "learning_rate": 8.354654146932066e-05,
2733
+ "loss": 0.0606,
2734
+ "step": 4540
2735
+ },
2736
+ {
2737
+ "grad_norm": 0.6542195677757263,
2738
+ "learning_rate": 8.346472128841364e-05,
2739
+ "loss": 0.0625,
2740
+ "step": 4550
2741
+ },
2742
+ {
2743
+ "grad_norm": 0.4475997984409332,
2744
+ "learning_rate": 8.338273845638848e-05,
2745
+ "loss": 0.0655,
2746
+ "step": 4560
2747
+ },
2748
+ {
2749
+ "grad_norm": 0.5038512349128723,
2750
+ "learning_rate": 8.330059337171258e-05,
2751
+ "loss": 0.0561,
2752
+ "step": 4570
2753
+ },
2754
+ {
2755
+ "grad_norm": 0.5571680068969727,
2756
+ "learning_rate": 8.32182864336419e-05,
2757
+ "loss": 0.064,
2758
+ "step": 4580
2759
+ },
2760
+ {
2761
+ "grad_norm": 0.5078977942466736,
2762
+ "learning_rate": 8.313581804221908e-05,
2763
+ "loss": 0.0656,
2764
+ "step": 4590
2765
+ },
2766
+ {
2767
+ "grad_norm": 0.5355502963066101,
2768
+ "learning_rate": 8.305318859827147e-05,
2769
+ "loss": 0.0586,
2770
+ "step": 4600
2771
+ },
2772
+ {
2773
+ "grad_norm": 0.6439716219902039,
2774
+ "learning_rate": 8.297039850340923e-05,
2775
+ "loss": 0.0638,
2776
+ "step": 4610
2777
+ },
2778
+ {
2779
+ "grad_norm": 0.4819484353065491,
2780
+ "learning_rate": 8.288744816002331e-05,
2781
+ "loss": 0.0602,
2782
+ "step": 4620
2783
+ },
2784
+ {
2785
+ "grad_norm": 0.5139064788818359,
2786
+ "learning_rate": 8.280433797128357e-05,
2787
+ "loss": 0.0597,
2788
+ "step": 4630
2789
+ },
2790
+ {
2791
+ "grad_norm": 0.523800253868103,
2792
+ "learning_rate": 8.272106834113674e-05,
2793
+ "loss": 0.0567,
2794
+ "step": 4640
2795
+ },
2796
+ {
2797
+ "grad_norm": 0.5478535890579224,
2798
+ "learning_rate": 8.26376396743045e-05,
2799
+ "loss": 0.0583,
2800
+ "step": 4650
2801
+ },
2802
+ {
2803
+ "grad_norm": 0.4601856768131256,
2804
+ "learning_rate": 8.25540523762815e-05,
2805
+ "loss": 0.0567,
2806
+ "step": 4660
2807
+ },
2808
+ {
2809
+ "grad_norm": 0.5561063885688782,
2810
+ "learning_rate": 8.247030685333346e-05,
2811
+ "loss": 0.0623,
2812
+ "step": 4670
2813
+ },
2814
+ {
2815
+ "grad_norm": 0.5215331315994263,
2816
+ "learning_rate": 8.238640351249503e-05,
2817
+ "loss": 0.0597,
2818
+ "step": 4680
2819
+ },
2820
+ {
2821
+ "grad_norm": 0.49608728289604187,
2822
+ "learning_rate": 8.2302342761568e-05,
2823
+ "loss": 0.0649,
2824
+ "step": 4690
2825
+ },
2826
+ {
2827
+ "grad_norm": 0.5840057134628296,
2828
+ "learning_rate": 8.221812500911919e-05,
2829
+ "loss": 0.0582,
2830
+ "step": 4700
2831
+ },
2832
+ {
2833
+ "grad_norm": 0.485944926738739,
2834
+ "learning_rate": 8.213375066447853e-05,
2835
+ "loss": 0.0588,
2836
+ "step": 4710
2837
+ },
2838
+ {
2839
+ "grad_norm": 0.5371300578117371,
2840
+ "learning_rate": 8.204922013773702e-05,
2841
+ "loss": 0.0644,
2842
+ "step": 4720
2843
+ },
2844
+ {
2845
+ "grad_norm": 0.5104807615280151,
2846
+ "learning_rate": 8.196453383974478e-05,
2847
+ "loss": 0.0588,
2848
+ "step": 4730
2849
+ },
2850
+ {
2851
+ "grad_norm": 0.47362270951271057,
2852
+ "learning_rate": 8.187969218210904e-05,
2853
+ "loss": 0.0596,
2854
+ "step": 4740
2855
+ },
2856
+ {
2857
+ "grad_norm": 0.45317813754081726,
2858
+ "learning_rate": 8.179469557719213e-05,
2859
+ "loss": 0.0633,
2860
+ "step": 4750
2861
+ },
2862
+ {
2863
+ "grad_norm": 0.5414328575134277,
2864
+ "learning_rate": 8.170954443810948e-05,
2865
+ "loss": 0.0586,
2866
+ "step": 4760
2867
+ },
2868
+ {
2869
+ "grad_norm": 0.4773976504802704,
2870
+ "learning_rate": 8.162423917872764e-05,
2871
+ "loss": 0.0632,
2872
+ "step": 4770
2873
+ },
2874
+ {
2875
+ "grad_norm": 0.4718090891838074,
2876
+ "learning_rate": 8.153878021366217e-05,
2877
+ "loss": 0.0639,
2878
+ "step": 4780
2879
+ },
2880
+ {
2881
+ "grad_norm": 0.5311549305915833,
2882
+ "learning_rate": 8.14531679582758e-05,
2883
+ "loss": 0.0648,
2884
+ "step": 4790
2885
+ },
2886
+ {
2887
+ "grad_norm": 0.6562144756317139,
2888
+ "learning_rate": 8.136740282867621e-05,
2889
+ "loss": 0.0566,
2890
+ "step": 4800
2891
+ },
2892
+ {
2893
+ "grad_norm": 0.5640397667884827,
2894
+ "learning_rate": 8.128148524171418e-05,
2895
+ "loss": 0.0696,
2896
+ "step": 4810
2897
+ },
2898
+ {
2899
+ "grad_norm": 0.5281029343605042,
2900
+ "learning_rate": 8.119541561498146e-05,
2901
+ "loss": 0.0597,
2902
+ "step": 4820
2903
+ },
2904
+ {
2905
+ "grad_norm": 0.5679964423179626,
2906
+ "learning_rate": 8.110919436680877e-05,
2907
+ "loss": 0.0605,
2908
+ "step": 4830
2909
+ },
2910
+ {
2911
+ "grad_norm": 0.5190762281417847,
2912
+ "learning_rate": 8.102282191626378e-05,
2913
+ "loss": 0.0606,
2914
+ "step": 4840
2915
+ },
2916
+ {
2917
+ "grad_norm": 0.5041296482086182,
2918
+ "learning_rate": 8.0936298683149e-05,
2919
+ "loss": 0.056,
2920
+ "step": 4850
2921
+ },
2922
+ {
2923
+ "grad_norm": 0.49754947423934937,
2924
+ "learning_rate": 8.084962508799991e-05,
2925
+ "loss": 0.0582,
2926
+ "step": 4860
2927
+ },
2928
+ {
2929
+ "grad_norm": 0.45711421966552734,
2930
+ "learning_rate": 8.076280155208273e-05,
2931
+ "loss": 0.0602,
2932
+ "step": 4870
2933
+ },
2934
+ {
2935
+ "grad_norm": 0.5648583173751831,
2936
+ "learning_rate": 8.067582849739245e-05,
2937
+ "loss": 0.0578,
2938
+ "step": 4880
2939
+ },
2940
+ {
2941
+ "grad_norm": 0.5352514982223511,
2942
+ "learning_rate": 8.058870634665079e-05,
2943
+ "loss": 0.056,
2944
+ "step": 4890
2945
+ },
2946
+ {
2947
+ "grad_norm": 0.5403103828430176,
2948
+ "learning_rate": 8.050143552330414e-05,
2949
+ "loss": 0.0566,
2950
+ "step": 4900
2951
+ },
2952
+ {
2953
+ "grad_norm": 0.6360242962837219,
2954
+ "learning_rate": 8.041401645152151e-05,
2955
+ "loss": 0.057,
2956
+ "step": 4910
2957
+ },
2958
+ {
2959
+ "grad_norm": 0.6194398403167725,
2960
+ "learning_rate": 8.032644955619239e-05,
2961
+ "loss": 0.0674,
2962
+ "step": 4920
2963
+ },
2964
+ {
2965
+ "grad_norm": 0.5398883819580078,
2966
+ "learning_rate": 8.023873526292483e-05,
2967
+ "loss": 0.0675,
2968
+ "step": 4930
2969
+ },
2970
+ {
2971
+ "grad_norm": 0.44168609380722046,
2972
+ "learning_rate": 8.015087399804322e-05,
2973
+ "loss": 0.0566,
2974
+ "step": 4940
2975
+ },
2976
+ {
2977
+ "grad_norm": 0.42431899905204773,
2978
+ "learning_rate": 8.006286618858635e-05,
2979
+ "loss": 0.0656,
2980
+ "step": 4950
2981
+ },
2982
+ {
2983
+ "grad_norm": 0.4777362048625946,
2984
+ "learning_rate": 7.99747122623052e-05,
2985
+ "loss": 0.0591,
2986
+ "step": 4960
2987
+ },
2988
+ {
2989
+ "grad_norm": 0.5099514722824097,
2990
+ "learning_rate": 7.988641264766097e-05,
2991
+ "loss": 0.0593,
2992
+ "step": 4970
2993
+ },
2994
+ {
2995
+ "grad_norm": 0.5469658374786377,
2996
+ "learning_rate": 7.9797967773823e-05,
2997
+ "loss": 0.0621,
2998
+ "step": 4980
2999
+ },
3000
+ {
3001
+ "grad_norm": 0.3929344117641449,
3002
+ "learning_rate": 7.970937807066659e-05,
3003
+ "loss": 0.0588,
3004
+ "step": 4990
3005
+ },
3006
+ {
3007
+ "grad_norm": 0.433919221162796,
3008
+ "learning_rate": 7.962064396877098e-05,
3009
+ "loss": 0.0615,
3010
+ "step": 5000
3011
+ },
3012
+ {
3013
+ "grad_norm": 0.45964470505714417,
3014
+ "learning_rate": 7.953176589941722e-05,
3015
+ "loss": 0.0603,
3016
+ "step": 5010
3017
+ },
3018
+ {
3019
+ "grad_norm": 0.4343070089817047,
3020
+ "learning_rate": 7.944274429458614e-05,
3021
+ "loss": 0.0637,
3022
+ "step": 5020
3023
+ },
3024
+ {
3025
+ "grad_norm": 0.4875974953174591,
3026
+ "learning_rate": 7.93535795869562e-05,
3027
+ "loss": 0.059,
3028
+ "step": 5030
3029
+ },
3030
+ {
3031
+ "grad_norm": 0.4838794767856598,
3032
+ "learning_rate": 7.926427220990134e-05,
3033
+ "loss": 0.0559,
3034
+ "step": 5040
3035
+ },
3036
+ {
3037
+ "grad_norm": 0.44954898953437805,
3038
+ "learning_rate": 7.9174822597489e-05,
3039
+ "loss": 0.0572,
3040
+ "step": 5050
3041
+ },
3042
+ {
3043
+ "grad_norm": 0.47687119245529175,
3044
+ "learning_rate": 7.908523118447789e-05,
3045
+ "loss": 0.0563,
3046
+ "step": 5060
3047
+ },
3048
+ {
3049
+ "grad_norm": 0.5028277635574341,
3050
+ "learning_rate": 7.89954984063159e-05,
3051
+ "loss": 0.0535,
3052
+ "step": 5070
3053
+ },
3054
+ {
3055
+ "grad_norm": 0.6060270667076111,
3056
+ "learning_rate": 7.890562469913811e-05,
3057
+ "loss": 0.056,
3058
+ "step": 5080
3059
+ },
3060
+ {
3061
+ "grad_norm": 0.48532041907310486,
3062
+ "learning_rate": 7.881561049976447e-05,
3063
+ "loss": 0.0592,
3064
+ "step": 5090
3065
+ },
3066
+ {
3067
+ "grad_norm": 0.5645678043365479,
3068
+ "learning_rate": 7.872545624569779e-05,
3069
+ "loss": 0.0606,
3070
+ "step": 5100
3071
+ },
3072
+ {
3073
+ "grad_norm": 0.5222822427749634,
3074
+ "learning_rate": 7.863516237512164e-05,
3075
+ "loss": 0.0584,
3076
+ "step": 5110
3077
+ },
3078
+ {
3079
+ "grad_norm": 0.4604170024394989,
3080
+ "learning_rate": 7.854472932689815e-05,
3081
+ "loss": 0.0577,
3082
+ "step": 5120
3083
+ },
3084
+ {
3085
+ "grad_norm": 0.44878771901130676,
3086
+ "learning_rate": 7.845415754056591e-05,
3087
+ "loss": 0.0559,
3088
+ "step": 5130
3089
+ },
3090
+ {
3091
+ "grad_norm": 0.4780261516571045,
3092
+ "learning_rate": 7.836344745633783e-05,
3093
+ "loss": 0.0642,
3094
+ "step": 5140
3095
+ },
3096
+ {
3097
+ "grad_norm": 0.4232221245765686,
3098
+ "learning_rate": 7.8272599515099e-05,
3099
+ "loss": 0.0594,
3100
+ "step": 5150
3101
+ },
3102
+ {
3103
+ "grad_norm": 0.556639552116394,
3104
+ "learning_rate": 7.818161415840453e-05,
3105
+ "loss": 0.0591,
3106
+ "step": 5160
3107
+ },
3108
+ {
3109
+ "grad_norm": 0.508721113204956,
3110
+ "learning_rate": 7.809049182847745e-05,
3111
+ "loss": 0.056,
3112
+ "step": 5170
3113
+ },
3114
+ {
3115
+ "grad_norm": 0.5749722123146057,
3116
+ "learning_rate": 7.799923296820653e-05,
3117
+ "loss": 0.0538,
3118
+ "step": 5180
3119
+ },
3120
+ {
3121
+ "grad_norm": 0.47284308075904846,
3122
+ "learning_rate": 7.790783802114408e-05,
3123
+ "loss": 0.0604,
3124
+ "step": 5190
3125
+ },
3126
+ {
3127
+ "grad_norm": 0.5093716382980347,
3128
+ "learning_rate": 7.781630743150392e-05,
3129
+ "loss": 0.057,
3130
+ "step": 5200
3131
+ },
3132
+ {
3133
+ "grad_norm": 0.5395658612251282,
3134
+ "learning_rate": 7.772464164415907e-05,
3135
+ "loss": 0.0585,
3136
+ "step": 5210
3137
+ },
3138
+ {
3139
+ "grad_norm": 0.5696783065795898,
3140
+ "learning_rate": 7.763284110463973e-05,
3141
+ "loss": 0.061,
3142
+ "step": 5220
3143
+ },
3144
+ {
3145
+ "grad_norm": 0.46128687262535095,
3146
+ "learning_rate": 7.754090625913099e-05,
3147
+ "loss": 0.0577,
3148
+ "step": 5230
3149
+ },
3150
+ {
3151
+ "grad_norm": 0.4267308712005615,
3152
+ "learning_rate": 7.744883755447075e-05,
3153
+ "loss": 0.0565,
3154
+ "step": 5240
3155
+ },
3156
+ {
3157
+ "grad_norm": 0.5390599966049194,
3158
+ "learning_rate": 7.735663543814749e-05,
3159
+ "loss": 0.0536,
3160
+ "step": 5250
3161
+ },
3162
+ {
3163
+ "grad_norm": 0.5402194261550903,
3164
+ "learning_rate": 7.726430035829813e-05,
3165
+ "loss": 0.0609,
3166
+ "step": 5260
3167
+ },
3168
+ {
3169
+ "grad_norm": 0.5593826174736023,
3170
+ "learning_rate": 7.717183276370586e-05,
3171
+ "loss": 0.0604,
3172
+ "step": 5270
3173
+ },
3174
+ {
3175
+ "grad_norm": 0.48568156361579895,
3176
+ "learning_rate": 7.707923310379794e-05,
3177
+ "loss": 0.056,
3178
+ "step": 5280
3179
+ },
3180
+ {
3181
+ "grad_norm": 0.5259646773338318,
3182
+ "learning_rate": 7.698650182864351e-05,
3183
+ "loss": 0.0549,
3184
+ "step": 5290
3185
+ },
3186
+ {
3187
+ "grad_norm": 0.4322700798511505,
3188
+ "learning_rate": 7.689363938895138e-05,
3189
+ "loss": 0.0575,
3190
+ "step": 5300
3191
+ },
3192
+ {
3193
+ "grad_norm": 0.46357619762420654,
3194
+ "learning_rate": 7.680064623606791e-05,
3195
+ "loss": 0.0535,
3196
+ "step": 5310
3197
+ },
3198
+ {
3199
+ "grad_norm": 0.4487699270248413,
3200
+ "learning_rate": 7.670752282197476e-05,
3201
+ "loss": 0.0557,
3202
+ "step": 5320
3203
+ },
3204
+ {
3205
+ "grad_norm": 0.5622449517250061,
3206
+ "learning_rate": 7.66142695992867e-05,
3207
+ "loss": 0.0521,
3208
+ "step": 5330
3209
+ },
3210
+ {
3211
+ "grad_norm": 0.5403662919998169,
3212
+ "learning_rate": 7.652088702124944e-05,
3213
+ "loss": 0.0548,
3214
+ "step": 5340
3215
+ },
3216
+ {
3217
+ "grad_norm": 0.46101441979408264,
3218
+ "learning_rate": 7.64273755417374e-05,
3219
+ "loss": 0.0547,
3220
+ "step": 5350
3221
+ },
3222
+ {
3223
+ "grad_norm": 0.48611223697662354,
3224
+ "learning_rate": 7.633373561525148e-05,
3225
+ "loss": 0.0553,
3226
+ "step": 5360
3227
+ },
3228
+ {
3229
+ "grad_norm": 0.5463878512382507,
3230
+ "learning_rate": 7.623996769691691e-05,
3231
+ "loss": 0.0547,
3232
+ "step": 5370
3233
+ },
3234
+ {
3235
+ "grad_norm": 0.5002387762069702,
3236
+ "learning_rate": 7.614607224248103e-05,
3237
+ "loss": 0.0575,
3238
+ "step": 5380
3239
+ },
3240
+ {
3241
+ "grad_norm": 0.4558352530002594,
3242
+ "learning_rate": 7.605204970831096e-05,
3243
+ "loss": 0.0563,
3244
+ "step": 5390
3245
+ },
3246
+ {
3247
+ "grad_norm": 0.5165241360664368,
3248
+ "learning_rate": 7.595790055139163e-05,
3249
+ "loss": 0.0544,
3250
+ "step": 5400
3251
+ },
3252
+ {
3253
+ "grad_norm": 0.49622392654418945,
3254
+ "learning_rate": 7.586362522932323e-05,
3255
+ "loss": 0.0556,
3256
+ "step": 5410
3257
+ },
3258
+ {
3259
+ "grad_norm": 0.5313038229942322,
3260
+ "learning_rate": 7.576922420031929e-05,
3261
+ "loss": 0.0596,
3262
+ "step": 5420
3263
+ },
3264
+ {
3265
+ "grad_norm": 0.5192410945892334,
3266
+ "learning_rate": 7.567469792320428e-05,
3267
+ "loss": 0.0536,
3268
+ "step": 5430
3269
+ },
3270
+ {
3271
+ "grad_norm": 0.44219714403152466,
3272
+ "learning_rate": 7.558004685741137e-05,
3273
+ "loss": 0.0541,
3274
+ "step": 5440
3275
+ },
3276
+ {
3277
+ "grad_norm": 0.552193284034729,
3278
+ "learning_rate": 7.548527146298036e-05,
3279
+ "loss": 0.0525,
3280
+ "step": 5450
3281
+ },
3282
+ {
3283
+ "grad_norm": 0.4243243336677551,
3284
+ "learning_rate": 7.539037220055527e-05,
3285
+ "loss": 0.0527,
3286
+ "step": 5460
3287
+ },
3288
+ {
3289
+ "grad_norm": 0.5068473219871521,
3290
+ "learning_rate": 7.529534953138213e-05,
3291
+ "loss": 0.058,
3292
+ "step": 5470
3293
+ },
3294
+ {
3295
+ "grad_norm": 0.46479758620262146,
3296
+ "learning_rate": 7.520020391730684e-05,
3297
+ "loss": 0.0563,
3298
+ "step": 5480
3299
+ },
3300
+ {
3301
+ "grad_norm": 0.4513914883136749,
3302
+ "learning_rate": 7.510493582077281e-05,
3303
+ "loss": 0.0553,
3304
+ "step": 5490
3305
+ },
3306
+ {
3307
+ "grad_norm": 0.4353752136230469,
3308
+ "learning_rate": 7.500954570481882e-05,
3309
+ "loss": 0.063,
3310
+ "step": 5500
3311
+ },
3312
+ {
3313
+ "grad_norm": 0.4668118953704834,
3314
+ "learning_rate": 7.491403403307662e-05,
3315
+ "loss": 0.0549,
3316
+ "step": 5510
3317
+ },
3318
+ {
3319
+ "grad_norm": 0.40962520241737366,
3320
+ "learning_rate": 7.481840126976885e-05,
3321
+ "loss": 0.0556,
3322
+ "step": 5520
3323
+ },
3324
+ {
3325
+ "grad_norm": 0.4395747184753418,
3326
+ "learning_rate": 7.472264787970666e-05,
3327
+ "loss": 0.0503,
3328
+ "step": 5530
3329
+ },
3330
+ {
3331
+ "grad_norm": 0.5224431157112122,
3332
+ "learning_rate": 7.462677432828751e-05,
3333
+ "loss": 0.058,
3334
+ "step": 5540
3335
+ },
3336
+ {
3337
+ "grad_norm": 0.5443442463874817,
3338
+ "learning_rate": 7.453078108149287e-05,
3339
+ "loss": 0.057,
3340
+ "step": 5550
3341
+ },
3342
+ {
3343
+ "grad_norm": 0.4638959765434265,
3344
+ "learning_rate": 7.443466860588599e-05,
3345
+ "loss": 0.0592,
3346
+ "step": 5560
3347
+ },
3348
+ {
3349
+ "grad_norm": 0.5321382284164429,
3350
+ "learning_rate": 7.43384373686096e-05,
3351
+ "loss": 0.0546,
3352
+ "step": 5570
3353
+ },
3354
+ {
3355
+ "grad_norm": 0.44946324825286865,
3356
+ "learning_rate": 7.424208783738367e-05,
3357
+ "loss": 0.0622,
3358
+ "step": 5580
3359
+ },
3360
+ {
3361
+ "grad_norm": 0.5622340440750122,
3362
+ "learning_rate": 7.414562048050315e-05,
3363
+ "loss": 0.0574,
3364
+ "step": 5590
3365
+ },
3366
+ {
3367
+ "grad_norm": 0.44878771901130676,
3368
+ "learning_rate": 7.404903576683559e-05,
3369
+ "loss": 0.0621,
3370
+ "step": 5600
3371
+ },
3372
+ {
3373
+ "grad_norm": 0.4351768493652344,
3374
+ "learning_rate": 7.3952334165819e-05,
3375
+ "loss": 0.0587,
3376
+ "step": 5610
3377
+ },
3378
+ {
3379
+ "grad_norm": 0.48724520206451416,
3380
+ "learning_rate": 7.385551614745952e-05,
3381
+ "loss": 0.0564,
3382
+ "step": 5620
3383
+ },
3384
+ {
3385
+ "grad_norm": 0.41412195563316345,
3386
+ "learning_rate": 7.375858218232905e-05,
3387
+ "loss": 0.0596,
3388
+ "step": 5630
3389
+ },
3390
+ {
3391
+ "grad_norm": 0.4886058270931244,
3392
+ "learning_rate": 7.366153274156312e-05,
3393
+ "loss": 0.0542,
3394
+ "step": 5640
3395
+ },
3396
+ {
3397
+ "grad_norm": 0.5508632063865662,
3398
+ "learning_rate": 7.356436829685844e-05,
3399
+ "loss": 0.0594,
3400
+ "step": 5650
3401
+ },
3402
+ {
3403
+ "grad_norm": 0.48630625009536743,
3404
+ "learning_rate": 7.346708932047074e-05,
3405
+ "loss": 0.0533,
3406
+ "step": 5660
3407
+ },
3408
+ {
3409
+ "grad_norm": 0.44918766617774963,
3410
+ "learning_rate": 7.336969628521237e-05,
3411
+ "loss": 0.0618,
3412
+ "step": 5670
3413
+ },
3414
+ {
3415
+ "grad_norm": 0.47114333510398865,
3416
+ "learning_rate": 7.32721896644501e-05,
3417
+ "loss": 0.0531,
3418
+ "step": 5680
3419
+ },
3420
+ {
3421
+ "grad_norm": 0.48327508568763733,
3422
+ "learning_rate": 7.317456993210272e-05,
3423
+ "loss": 0.0566,
3424
+ "step": 5690
3425
+ },
3426
+ {
3427
+ "grad_norm": 0.40669572353363037,
3428
+ "learning_rate": 7.307683756263881e-05,
3429
+ "loss": 0.0528,
3430
+ "step": 5700
3431
+ },
3432
+ {
3433
+ "grad_norm": 0.5093392729759216,
3434
+ "learning_rate": 7.297899303107441e-05,
3435
+ "loss": 0.0596,
3436
+ "step": 5710
3437
+ },
3438
+ {
3439
+ "grad_norm": 0.3939994275569916,
3440
+ "learning_rate": 7.288103681297068e-05,
3441
+ "loss": 0.0538,
3442
+ "step": 5720
3443
+ },
3444
+ {
3445
+ "grad_norm": 0.48547250032424927,
3446
+ "learning_rate": 7.278296938443166e-05,
3447
+ "loss": 0.0639,
3448
+ "step": 5730
3449
+ },
3450
+ {
3451
+ "grad_norm": 0.5965085029602051,
3452
+ "learning_rate": 7.26847912221019e-05,
3453
+ "loss": 0.057,
3454
+ "step": 5740
3455
+ },
3456
+ {
3457
+ "grad_norm": 0.5538206696510315,
3458
+ "learning_rate": 7.258650280316415e-05,
3459
+ "loss": 0.054,
3460
+ "step": 5750
3461
+ },
3462
+ {
3463
+ "grad_norm": 0.47953662276268005,
3464
+ "learning_rate": 7.248810460533706e-05,
3465
+ "loss": 0.0527,
3466
+ "step": 5760
3467
+ },
3468
+ {
3469
+ "grad_norm": 0.40855997800827026,
3470
+ "learning_rate": 7.238959710687282e-05,
3471
+ "loss": 0.0546,
3472
+ "step": 5770
3473
+ },
3474
+ {
3475
+ "grad_norm": 0.49721816182136536,
3476
+ "learning_rate": 7.229098078655489e-05,
3477
+ "loss": 0.0581,
3478
+ "step": 5780
3479
+ },
3480
+ {
3481
+ "grad_norm": 0.4714704155921936,
3482
+ "learning_rate": 7.219225612369565e-05,
3483
+ "loss": 0.0523,
3484
+ "step": 5790
3485
+ },
3486
+ {
3487
+ "grad_norm": 0.37301206588745117,
3488
+ "learning_rate": 7.209342359813404e-05,
3489
+ "loss": 0.0558,
3490
+ "step": 5800
3491
+ },
3492
+ {
3493
+ "grad_norm": 0.49109941720962524,
3494
+ "learning_rate": 7.199448369023327e-05,
3495
+ "loss": 0.0568,
3496
+ "step": 5810
3497
+ },
3498
+ {
3499
+ "grad_norm": 0.5443580150604248,
3500
+ "learning_rate": 7.189543688087845e-05,
3501
+ "loss": 0.0523,
3502
+ "step": 5820
3503
+ },
3504
+ {
3505
+ "grad_norm": 0.5693491697311401,
3506
+ "learning_rate": 7.17962836514743e-05,
3507
+ "loss": 0.0525,
3508
+ "step": 5830
3509
+ },
3510
+ {
3511
+ "grad_norm": 0.609158992767334,
3512
+ "learning_rate": 7.169702448394279e-05,
3513
+ "loss": 0.0536,
3514
+ "step": 5840
3515
+ },
3516
+ {
3517
+ "grad_norm": 0.5174346566200256,
3518
+ "learning_rate": 7.159765986072071e-05,
3519
+ "loss": 0.0555,
3520
+ "step": 5850
3521
+ },
3522
+ {
3523
+ "grad_norm": 0.5700979828834534,
3524
+ "learning_rate": 7.149819026475751e-05,
3525
+ "loss": 0.059,
3526
+ "step": 5860
3527
+ },
3528
+ {
3529
+ "grad_norm": 0.47033268213272095,
3530
+ "learning_rate": 7.139861617951275e-05,
3531
+ "loss": 0.0541,
3532
+ "step": 5870
3533
+ },
3534
+ {
3535
+ "grad_norm": 0.45670151710510254,
3536
+ "learning_rate": 7.129893808895395e-05,
3537
+ "loss": 0.056,
3538
+ "step": 5880
3539
+ },
3540
+ {
3541
+ "grad_norm": 0.41775184869766235,
3542
+ "learning_rate": 7.119915647755404e-05,
3543
+ "loss": 0.0504,
3544
+ "step": 5890
3545
+ },
3546
+ {
3547
+ "grad_norm": 0.43254759907722473,
3548
+ "learning_rate": 7.109927183028914e-05,
3549
+ "loss": 0.054,
3550
+ "step": 5900
3551
+ },
3552
+ {
3553
+ "grad_norm": 0.4849887192249298,
3554
+ "learning_rate": 7.099928463263619e-05,
3555
+ "loss": 0.0561,
3556
+ "step": 5910
3557
+ },
3558
+ {
3559
+ "grad_norm": 0.4883417785167694,
3560
+ "learning_rate": 7.08991953705705e-05,
3561
+ "loss": 0.0554,
3562
+ "step": 5920
3563
+ },
3564
+ {
3565
+ "grad_norm": 0.4409164488315582,
3566
+ "learning_rate": 7.07990045305635e-05,
3567
+ "loss": 0.0587,
3568
+ "step": 5930
3569
+ },
3570
+ {
3571
+ "grad_norm": 0.5324891209602356,
3572
+ "learning_rate": 7.069871259958034e-05,
3573
+ "loss": 0.0629,
3574
+ "step": 5940
3575
+ },
3576
+ {
3577
+ "grad_norm": 0.41858968138694763,
3578
+ "learning_rate": 7.059832006507745e-05,
3579
+ "loss": 0.0579,
3580
+ "step": 5950
3581
+ },
3582
+ {
3583
+ "grad_norm": 0.490377813577652,
3584
+ "learning_rate": 7.049782741500028e-05,
3585
+ "loss": 0.054,
3586
+ "step": 5960
3587
+ },
3588
+ {
3589
+ "grad_norm": 0.4170517921447754,
3590
+ "learning_rate": 7.039723513778087e-05,
3591
+ "loss": 0.0519,
3592
+ "step": 5970
3593
+ },
3594
+ {
3595
+ "grad_norm": 0.4487491548061371,
3596
+ "learning_rate": 7.029654372233544e-05,
3597
+ "loss": 0.0543,
3598
+ "step": 5980
3599
+ },
3600
+ {
3601
+ "grad_norm": 0.4388402998447418,
3602
+ "learning_rate": 7.019575365806215e-05,
3603
+ "loss": 0.0563,
3604
+ "step": 5990
3605
+ },
3606
+ {
3607
+ "grad_norm": 0.43077412247657776,
3608
+ "learning_rate": 7.009486543483858e-05,
3609
+ "loss": 0.0536,
3610
+ "step": 6000
3611
+ }
3612
+ ],
3613
+ "logging_steps": 10,
3614
+ "max_steps": 15000,
3615
+ "num_input_tokens_seen": 0,
3616
+ "num_train_epochs": 9223372036854775807,
3617
+ "save_steps": 3000,
3618
+ "stateful_callbacks": {
3619
+ "TrainerControl": {
3620
+ "args": {
3621
+ "should_epoch_stop": false,
3622
+ "should_evaluate": false,
3623
+ "should_log": false,
3624
+ "should_save": true,
3625
+ "should_training_stop": false
3626
+ },
3627
+ "attributes": {}
3628
+ }
3629
+ },
3630
+ "total_flos": 0.0,
3631
+ "train_batch_size": 64,
3632
+ "trial_name": null,
3633
+ "trial_params": null
3634
+ }
checkpoint-6000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e96fd6983006fc2b7d9d72a31598f10898f35edea7d6032196a16e3894c26712
3
+ size 5905
checkpoint-6000/wandb_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"project": "groot-finetune", "run_id": "g1_finetune"}