checkpoint-5000 (16D, hybrid crop, batch128)
Browse files- 16d_30k/checkpoint-5000/config.json +88 -0
- 16d_30k/checkpoint-5000/embodiment_id.json +57 -0
- 16d_30k/checkpoint-5000/experiment_cfg/conf.yaml +214 -0
- 16d_30k/checkpoint-5000/experiment_cfg/config.yaml +239 -0
- 16d_30k/checkpoint-5000/experiment_cfg/dataset_statistics.json +257 -0
- 16d_30k/checkpoint-5000/experiment_cfg/final_model_config.json +55 -0
- 16d_30k/checkpoint-5000/experiment_cfg/final_processor_config.json +0 -0
- 16d_30k/checkpoint-5000/model-00001-of-00002.safetensors +3 -0
- 16d_30k/checkpoint-5000/model-00002-of-00002.safetensors +3 -0
- 16d_30k/checkpoint-5000/model.safetensors.index.json +0 -0
- 16d_30k/checkpoint-5000/processor_config.json +1124 -0
- 16d_30k/checkpoint-5000/statistics.json +0 -0
- 16d_30k/checkpoint-5000/trainer_state.json +3034 -0
- 16d_30k/checkpoint-5000/training_args.bin +3 -0
- 16d_30k/checkpoint-5000/wandb_config.json +1 -0
16d_30k/checkpoint-5000/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": 1.0,
|
| 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": "bfloat16",
|
| 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": false,
|
| 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 |
+
}
|
16d_30k/checkpoint-5000/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 |
+
}
|
16d_30k/checkpoint-5000/experiment_cfg/conf.yaml
ADDED
|
@@ -0,0 +1,214 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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: null
|
| 27 |
+
use_albumentations_transforms: true
|
| 28 |
+
extra_augmentation_config: null
|
| 29 |
+
formalize_language: true
|
| 30 |
+
apply_sincos_state_encoding: false
|
| 31 |
+
use_percentiles: true
|
| 32 |
+
use_relative_action: true
|
| 33 |
+
max_state_dim: 132
|
| 34 |
+
max_action_dim: 132
|
| 35 |
+
action_horizon: 40
|
| 36 |
+
hidden_size: 1024
|
| 37 |
+
input_embedding_dim: 1536
|
| 38 |
+
state_history_length: 1
|
| 39 |
+
add_pos_embed: true
|
| 40 |
+
attn_dropout: 0.2
|
| 41 |
+
use_vlln: true
|
| 42 |
+
max_seq_len: 1024
|
| 43 |
+
use_alternate_vl_dit: true
|
| 44 |
+
attend_text_every_n_blocks: 2
|
| 45 |
+
diffusion_model_cfg:
|
| 46 |
+
positional_embeddings: null
|
| 47 |
+
num_layers: 16
|
| 48 |
+
num_attention_heads: 32
|
| 49 |
+
attention_head_dim: 48
|
| 50 |
+
norm_type: ada_norm
|
| 51 |
+
dropout: 0.2
|
| 52 |
+
final_dropout: true
|
| 53 |
+
output_dim: 1024
|
| 54 |
+
interleave_self_attention: true
|
| 55 |
+
num_inference_timesteps: 4
|
| 56 |
+
noise_beta_alpha: 1.5
|
| 57 |
+
noise_beta_beta: 1.0
|
| 58 |
+
noise_s: 0.999
|
| 59 |
+
num_timestep_buckets: 1000
|
| 60 |
+
tune_projector: true
|
| 61 |
+
tune_diffusion_model: true
|
| 62 |
+
tune_vlln: true
|
| 63 |
+
state_dropout_prob: 0.2
|
| 64 |
+
exclude_state: false
|
| 65 |
+
use_mean_std: false
|
| 66 |
+
max_num_embodiments: 32
|
| 67 |
+
data:
|
| 68 |
+
datasets:
|
| 69 |
+
- dataset_paths:
|
| 70 |
+
- /root/screwing_train/datasets/screwing35_follower_train
|
| 71 |
+
embodiment_tag: new_embodiment
|
| 72 |
+
mix_ratio: 1.0
|
| 73 |
+
dataset_type: physical_embodiment
|
| 74 |
+
val_dataset_path: null
|
| 75 |
+
modality_configs:
|
| 76 |
+
new_embodiment:
|
| 77 |
+
video:
|
| 78 |
+
delta_indices:
|
| 79 |
+
- 0
|
| 80 |
+
modality_keys:
|
| 81 |
+
- cam_left_head
|
| 82 |
+
- cam_left_wrist
|
| 83 |
+
- cam_right_wrist
|
| 84 |
+
sin_cos_embedding_keys: null
|
| 85 |
+
mean_std_embedding_keys: null
|
| 86 |
+
action_configs: null
|
| 87 |
+
state:
|
| 88 |
+
delta_indices:
|
| 89 |
+
- 0
|
| 90 |
+
modality_keys:
|
| 91 |
+
- arm_left
|
| 92 |
+
- arm_right
|
| 93 |
+
sin_cos_embedding_keys: null
|
| 94 |
+
mean_std_embedding_keys: null
|
| 95 |
+
action_configs: null
|
| 96 |
+
action:
|
| 97 |
+
delta_indices:
|
| 98 |
+
- 0
|
| 99 |
+
- 1
|
| 100 |
+
- 2
|
| 101 |
+
- 3
|
| 102 |
+
- 4
|
| 103 |
+
- 5
|
| 104 |
+
- 6
|
| 105 |
+
- 7
|
| 106 |
+
- 8
|
| 107 |
+
- 9
|
| 108 |
+
- 10
|
| 109 |
+
- 11
|
| 110 |
+
- 12
|
| 111 |
+
- 13
|
| 112 |
+
- 14
|
| 113 |
+
- 15
|
| 114 |
+
modality_keys:
|
| 115 |
+
- arm_left
|
| 116 |
+
- arm_right
|
| 117 |
+
sin_cos_embedding_keys: null
|
| 118 |
+
mean_std_embedding_keys: null
|
| 119 |
+
action_configs:
|
| 120 |
+
- rep: ABSOLUTE
|
| 121 |
+
type: NON_EEF
|
| 122 |
+
format: DEFAULT
|
| 123 |
+
state_key: null
|
| 124 |
+
- rep: ABSOLUTE
|
| 125 |
+
type: NON_EEF
|
| 126 |
+
format: DEFAULT
|
| 127 |
+
state_key: null
|
| 128 |
+
language:
|
| 129 |
+
delta_indices:
|
| 130 |
+
- 0
|
| 131 |
+
modality_keys:
|
| 132 |
+
- annotation.human.primitive_instruction
|
| 133 |
+
sin_cos_embedding_keys: null
|
| 134 |
+
mean_std_embedding_keys: null
|
| 135 |
+
action_configs: null
|
| 136 |
+
download_cache: false
|
| 137 |
+
shard_size: 1024
|
| 138 |
+
episode_sampling_rate: 0.1
|
| 139 |
+
num_shards_per_epoch: 100000
|
| 140 |
+
override_pretraining_statistics: true
|
| 141 |
+
mode: single_turn
|
| 142 |
+
random_chop: 0.0
|
| 143 |
+
mock_dataset_mode: false
|
| 144 |
+
shuffle: true
|
| 145 |
+
seed: 42
|
| 146 |
+
multiprocessing_context: fork
|
| 147 |
+
allow_padding: false
|
| 148 |
+
subsample_ratio: 1.0
|
| 149 |
+
image_crop_size:
|
| 150 |
+
- 244
|
| 151 |
+
- 244
|
| 152 |
+
image_target_size:
|
| 153 |
+
- 224
|
| 154 |
+
- 224
|
| 155 |
+
video_backend: torchcodec
|
| 156 |
+
training:
|
| 157 |
+
output_dir: /root/screwing_train/runs/groot_16d_30k
|
| 158 |
+
experiment_name: null
|
| 159 |
+
max_steps: 30000
|
| 160 |
+
global_batch_size: 128
|
| 161 |
+
batch_size: null
|
| 162 |
+
gradient_accumulation_steps: 1
|
| 163 |
+
learning_rate: 0.0001
|
| 164 |
+
lr_scheduler_type: cosine
|
| 165 |
+
weight_decay: 1.0e-05
|
| 166 |
+
warmup_ratio: 0.05
|
| 167 |
+
warmup_steps: 0
|
| 168 |
+
max_grad_norm: 1.0
|
| 169 |
+
optim: adamw_torch
|
| 170 |
+
start_from_checkpoint: /root/screwing_train/base_model_patched
|
| 171 |
+
skip_weight_loading: false
|
| 172 |
+
tf32: true
|
| 173 |
+
fp16: false
|
| 174 |
+
bf16: true
|
| 175 |
+
eval_bf16: true
|
| 176 |
+
logging_steps: 10
|
| 177 |
+
save_steps: 2500
|
| 178 |
+
save_total_limit: 30
|
| 179 |
+
save_vl_model: false
|
| 180 |
+
save_only_model: true
|
| 181 |
+
upload_checkpoints: false
|
| 182 |
+
upload_every: 1000
|
| 183 |
+
upload_last_n_checkpoints: 5
|
| 184 |
+
max_concurrent_uploads: 2
|
| 185 |
+
eval_strategy: 'no'
|
| 186 |
+
eval_steps: 500
|
| 187 |
+
eval_set_split_ratio: 0.1
|
| 188 |
+
eval_batch_size: 2
|
| 189 |
+
save_best_eval_metric_name: ''
|
| 190 |
+
save_best_eval_metric_greater_is_better: true
|
| 191 |
+
deepspeed_stage: 2
|
| 192 |
+
gradient_checkpointing: false
|
| 193 |
+
transformers_trust_remote_code: true
|
| 194 |
+
transformers_local_files_only: false
|
| 195 |
+
transformers_cache_dir: null
|
| 196 |
+
transformers_access_token: null
|
| 197 |
+
use_ddp: false
|
| 198 |
+
ddp_bucket_cap_mb: 100
|
| 199 |
+
num_gpus: 2
|
| 200 |
+
dataloader_num_workers: 24
|
| 201 |
+
remove_unused_columns: false
|
| 202 |
+
use_wandb: true
|
| 203 |
+
wandb_project: screwing_hil
|
| 204 |
+
enable_profiling: false
|
| 205 |
+
max_retries: 3
|
| 206 |
+
assert_loss_less_than: null
|
| 207 |
+
add_rl_callback: false
|
| 208 |
+
enable_open_loop_eval: false
|
| 209 |
+
open_loop_eval_traj_ids:
|
| 210 |
+
- 0
|
| 211 |
+
open_loop_eval_steps_per_traj: 100
|
| 212 |
+
open_loop_eval_plot_indices: null
|
| 213 |
+
max_steps: 30000
|
| 214 |
+
save_steps: 2500
|
16d_30k/checkpoint-5000/experiment_cfg/config.yaml
ADDED
|
@@ -0,0 +1,239 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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 |
+
- /root/screwing_train/datasets/screwing35_follower_train
|
| 8 |
+
dataset_type: physical_embodiment
|
| 9 |
+
embodiment_tag: new_embodiment
|
| 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 |
+
new_embodiment:
|
| 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 |
+
delta_indices:
|
| 39 |
+
- 0
|
| 40 |
+
- 1
|
| 41 |
+
- 2
|
| 42 |
+
- 3
|
| 43 |
+
- 4
|
| 44 |
+
- 5
|
| 45 |
+
- 6
|
| 46 |
+
- 7
|
| 47 |
+
- 8
|
| 48 |
+
- 9
|
| 49 |
+
- 10
|
| 50 |
+
- 11
|
| 51 |
+
- 12
|
| 52 |
+
- 13
|
| 53 |
+
- 14
|
| 54 |
+
- 15
|
| 55 |
+
mean_std_embedding_keys: null
|
| 56 |
+
modality_keys:
|
| 57 |
+
- arm_left
|
| 58 |
+
- arm_right
|
| 59 |
+
sin_cos_embedding_keys: null
|
| 60 |
+
language: !!python/object:gr00t.data.types.ModalityConfig
|
| 61 |
+
action_configs: null
|
| 62 |
+
delta_indices:
|
| 63 |
+
- 0
|
| 64 |
+
mean_std_embedding_keys: null
|
| 65 |
+
modality_keys:
|
| 66 |
+
- annotation.human.primitive_instruction
|
| 67 |
+
sin_cos_embedding_keys: null
|
| 68 |
+
state: !!python/object:gr00t.data.types.ModalityConfig
|
| 69 |
+
action_configs: null
|
| 70 |
+
delta_indices:
|
| 71 |
+
- 0
|
| 72 |
+
mean_std_embedding_keys: null
|
| 73 |
+
modality_keys:
|
| 74 |
+
- arm_left
|
| 75 |
+
- arm_right
|
| 76 |
+
sin_cos_embedding_keys: null
|
| 77 |
+
video: !!python/object:gr00t.data.types.ModalityConfig
|
| 78 |
+
action_configs: null
|
| 79 |
+
delta_indices:
|
| 80 |
+
- 0
|
| 81 |
+
mean_std_embedding_keys: null
|
| 82 |
+
modality_keys:
|
| 83 |
+
- cam_left_head
|
| 84 |
+
- cam_left_wrist
|
| 85 |
+
- cam_right_wrist
|
| 86 |
+
sin_cos_embedding_keys: null
|
| 87 |
+
mode: single_turn
|
| 88 |
+
multiprocessing_context: fork
|
| 89 |
+
num_shards_per_epoch: 100000
|
| 90 |
+
override_pretraining_statistics: true
|
| 91 |
+
random_chop: 0.0
|
| 92 |
+
seed: 42
|
| 93 |
+
shard_size: 1024
|
| 94 |
+
shuffle: true
|
| 95 |
+
subsample_ratio: 1.0
|
| 96 |
+
video_backend: torchcodec
|
| 97 |
+
load_config_path: null
|
| 98 |
+
model: !!python/object:gr00t.configs.model.gr00t_n1d7.Gr00tN1d7Config
|
| 99 |
+
_attn_implementation_internal: null
|
| 100 |
+
_commit_hash: null
|
| 101 |
+
_name_or_path: ''
|
| 102 |
+
_output_attentions: false
|
| 103 |
+
add_cross_attention: false
|
| 104 |
+
architectures: null
|
| 105 |
+
backbone_trainable_params_fp32: true
|
| 106 |
+
bad_words_ids: null
|
| 107 |
+
begin_suppress_tokens: null
|
| 108 |
+
bos_token_id: null
|
| 109 |
+
chunk_size_feed_forward: 0
|
| 110 |
+
color_jitter_params: null
|
| 111 |
+
cross_attention_hidden_size: null
|
| 112 |
+
decoder_start_token_id: null
|
| 113 |
+
diffusion_model_cfg:
|
| 114 |
+
attention_head_dim: 48
|
| 115 |
+
dropout: 0.2
|
| 116 |
+
final_dropout: true
|
| 117 |
+
interleave_self_attention: true
|
| 118 |
+
norm_type: ada_norm
|
| 119 |
+
num_attention_heads: 32
|
| 120 |
+
num_layers: 16
|
| 121 |
+
output_dim: 1024
|
| 122 |
+
positional_embeddings: null
|
| 123 |
+
diversity_penalty: 0.0
|
| 124 |
+
do_sample: false
|
| 125 |
+
dtype: null
|
| 126 |
+
early_stopping: false
|
| 127 |
+
encoder_no_repeat_ngram_size: 0
|
| 128 |
+
eos_token_id: null
|
| 129 |
+
exponential_decay_length_penalty: null
|
| 130 |
+
extra_augmentation_config: null
|
| 131 |
+
finetuning_task: null
|
| 132 |
+
forced_bos_token_id: null
|
| 133 |
+
forced_eos_token_id: null
|
| 134 |
+
id2label:
|
| 135 |
+
0: LABEL_0
|
| 136 |
+
1: LABEL_1
|
| 137 |
+
is_decoder: false
|
| 138 |
+
is_encoder_decoder: false
|
| 139 |
+
label2id:
|
| 140 |
+
LABEL_0: 0
|
| 141 |
+
LABEL_1: 1
|
| 142 |
+
length_penalty: 1.0
|
| 143 |
+
load_bf16: false
|
| 144 |
+
max_length: 20
|
| 145 |
+
min_length: 0
|
| 146 |
+
model_name: nvidia/Cosmos-Reason2-2B
|
| 147 |
+
no_repeat_ngram_size: 0
|
| 148 |
+
num_beam_groups: 1
|
| 149 |
+
num_beams: 1
|
| 150 |
+
num_return_sequences: 1
|
| 151 |
+
output_hidden_states: false
|
| 152 |
+
output_scores: false
|
| 153 |
+
pad_token_id: null
|
| 154 |
+
prefix: null
|
| 155 |
+
problem_type: null
|
| 156 |
+
pruned_heads: {}
|
| 157 |
+
random_rotation_angle: null
|
| 158 |
+
remove_invalid_values: false
|
| 159 |
+
repetition_penalty: 1.0
|
| 160 |
+
reproject_vision: false
|
| 161 |
+
return_dict: true
|
| 162 |
+
return_dict_in_generate: false
|
| 163 |
+
sep_token_id: null
|
| 164 |
+
state_dropout_prob: 0.2
|
| 165 |
+
suppress_tokens: null
|
| 166 |
+
task_specific_params: null
|
| 167 |
+
temperature: 1.0
|
| 168 |
+
tf_legacy_loss: false
|
| 169 |
+
tie_encoder_decoder: false
|
| 170 |
+
tie_word_embeddings: true
|
| 171 |
+
tokenizer_class: null
|
| 172 |
+
top_k: 50
|
| 173 |
+
top_p: 1.0
|
| 174 |
+
torchscript: false
|
| 175 |
+
transformers_version: null
|
| 176 |
+
tune_diffusion_model: true
|
| 177 |
+
tune_llm: false
|
| 178 |
+
tune_projector: true
|
| 179 |
+
tune_visual: false
|
| 180 |
+
typical_p: 1.0
|
| 181 |
+
use_bfloat16: false
|
| 182 |
+
use_relative_action: true
|
| 183 |
+
training: !!python/object:gr00t.configs.training.training_config.TrainingConfig
|
| 184 |
+
add_rl_callback: false
|
| 185 |
+
assert_loss_less_than: null
|
| 186 |
+
batch_size: null
|
| 187 |
+
bf16: true
|
| 188 |
+
dataloader_num_workers: 24
|
| 189 |
+
ddp_bucket_cap_mb: 100
|
| 190 |
+
deepspeed_stage: 2
|
| 191 |
+
enable_open_loop_eval: false
|
| 192 |
+
enable_profiling: false
|
| 193 |
+
eval_batch_size: 2
|
| 194 |
+
eval_bf16: true
|
| 195 |
+
eval_set_split_ratio: 0.1
|
| 196 |
+
eval_steps: 500
|
| 197 |
+
eval_strategy: 'no'
|
| 198 |
+
experiment_name: null
|
| 199 |
+
fp16: false
|
| 200 |
+
global_batch_size: 128
|
| 201 |
+
gradient_accumulation_steps: 1
|
| 202 |
+
gradient_checkpointing: false
|
| 203 |
+
learning_rate: 0.0001
|
| 204 |
+
logging_steps: 10
|
| 205 |
+
lr_scheduler_type: cosine
|
| 206 |
+
max_concurrent_uploads: 2
|
| 207 |
+
max_grad_norm: 1.0
|
| 208 |
+
max_retries: 3
|
| 209 |
+
max_steps: 30000
|
| 210 |
+
num_gpus: 2
|
| 211 |
+
open_loop_eval_plot_indices: null
|
| 212 |
+
open_loop_eval_steps_per_traj: 100
|
| 213 |
+
open_loop_eval_traj_ids:
|
| 214 |
+
- 0
|
| 215 |
+
optim: adamw_torch
|
| 216 |
+
output_dir: /root/screwing_train/runs/groot_16d_30k
|
| 217 |
+
remove_unused_columns: false
|
| 218 |
+
save_best_eval_metric_greater_is_better: true
|
| 219 |
+
save_best_eval_metric_name: ''
|
| 220 |
+
save_only_model: true
|
| 221 |
+
save_steps: 2500
|
| 222 |
+
save_total_limit: 30
|
| 223 |
+
save_vl_model: false
|
| 224 |
+
skip_weight_loading: false
|
| 225 |
+
start_from_checkpoint: /root/screwing_train/base_model_patched
|
| 226 |
+
tf32: true
|
| 227 |
+
transformers_access_token: null
|
| 228 |
+
transformers_cache_dir: null
|
| 229 |
+
transformers_local_files_only: false
|
| 230 |
+
transformers_trust_remote_code: true
|
| 231 |
+
upload_checkpoints: false
|
| 232 |
+
upload_every: 1000
|
| 233 |
+
upload_last_n_checkpoints: 5
|
| 234 |
+
use_ddp: false
|
| 235 |
+
use_wandb: true
|
| 236 |
+
wandb_project: screwing_hil
|
| 237 |
+
warmup_ratio: 0.05
|
| 238 |
+
warmup_steps: 0
|
| 239 |
+
weight_decay: 1.0e-05
|
16d_30k/checkpoint-5000/experiment_cfg/dataset_statistics.json
ADDED
|
@@ -0,0 +1,257 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"new_embodiment": {
|
| 3 |
+
"state": {
|
| 4 |
+
"arm_left": {
|
| 5 |
+
"min": [
|
| 6 |
+
-0.3512935936450958,
|
| 7 |
+
-0.0003355582884978503,
|
| 8 |
+
-0.4693501889705658,
|
| 9 |
+
-2.0985336303710938,
|
| 10 |
+
-0.30987611413002014,
|
| 11 |
+
-0.390062540769577,
|
| 12 |
+
-0.756526529788971,
|
| 13 |
+
-0.0015339808305725455
|
| 14 |
+
],
|
| 15 |
+
"max": [
|
| 16 |
+
0.4587441384792328,
|
| 17 |
+
0.4674566686153412,
|
| 18 |
+
0.2403915673494339,
|
| 19 |
+
-0.8097261190414429,
|
| 20 |
+
0.30678418278694153,
|
| 21 |
+
0.7219297289848328,
|
| 22 |
+
0.09214304387569427,
|
| 23 |
+
1.0001554489135742
|
| 24 |
+
],
|
| 25 |
+
"mean": [
|
| 26 |
+
0.045786481350660324,
|
| 27 |
+
0.1028616651892662,
|
| 28 |
+
-0.08757976442575455,
|
| 29 |
+
-1.7214374542236328,
|
| 30 |
+
0.016063803806900978,
|
| 31 |
+
0.22246477007865906,
|
| 32 |
+
-0.3082488775253296,
|
| 33 |
+
0.23499731719493866
|
| 34 |
+
],
|
| 35 |
+
"std": [
|
| 36 |
+
0.20334461331367493,
|
| 37 |
+
0.0962500348687172,
|
| 38 |
+
0.16793721914291382,
|
| 39 |
+
0.24059325456619263,
|
| 40 |
+
0.12082602083683014,
|
| 41 |
+
0.2635464370250702,
|
| 42 |
+
0.22770243883132935,
|
| 43 |
+
0.38598382472991943
|
| 44 |
+
],
|
| 45 |
+
"q01": [
|
| 46 |
+
-0.28486502170562744,
|
| 47 |
+
0.0,
|
| 48 |
+
-0.4279566705226898,
|
| 49 |
+
-2.013697385787964,
|
| 50 |
+
-0.2523278594017029,
|
| 51 |
+
-0.2879210114479065,
|
| 52 |
+
-0.7209579944610596,
|
| 53 |
+
0.0
|
| 54 |
+
],
|
| 55 |
+
"q99": [
|
| 56 |
+
0.4115143120288849,
|
| 57 |
+
0.4564431607723236,
|
| 58 |
+
0.18591128289699554,
|
| 59 |
+
-0.9433622360229492,
|
| 60 |
+
0.2618792951107025,
|
| 61 |
+
0.7118869423866272,
|
| 62 |
+
0.06595534831285477,
|
| 63 |
+
0.9970875382423401
|
| 64 |
+
]
|
| 65 |
+
},
|
| 66 |
+
"arm_right": {
|
| 67 |
+
"min": [
|
| 68 |
+
-0.9125747680664062,
|
| 69 |
+
-0.46813976764678955,
|
| 70 |
+
-0.08915065228939056,
|
| 71 |
+
-2.0727076530456543,
|
| 72 |
+
-0.9404261112213135,
|
| 73 |
+
-0.15235544741153717,
|
| 74 |
+
-0.053947482258081436,
|
| 75 |
+
0.0
|
| 76 |
+
],
|
| 77 |
+
"max": [
|
| 78 |
+
0.3974807858467102,
|
| 79 |
+
-0.031806133687496185,
|
| 80 |
+
1.0656253099441528,
|
| 81 |
+
-1.032512903213501,
|
| 82 |
+
0.13881327211856842,
|
| 83 |
+
1.1778095960617065,
|
| 84 |
+
0.8715579509735107,
|
| 85 |
+
1.277806043624878
|
| 86 |
+
],
|
| 87 |
+
"mean": [
|
| 88 |
+
-0.25570186972618103,
|
| 89 |
+
-0.2676304280757904,
|
| 90 |
+
0.3035861551761627,
|
| 91 |
+
-1.7064963579177856,
|
| 92 |
+
-0.25106674432754517,
|
| 93 |
+
0.5091767311096191,
|
| 94 |
+
0.43237197399139404,
|
| 95 |
+
0.3860172629356384
|
| 96 |
+
],
|
| 97 |
+
"std": [
|
| 98 |
+
0.25905025005340576,
|
| 99 |
+
0.0594896525144577,
|
| 100 |
+
0.17308898270130157,
|
| 101 |
+
0.18367764353752136,
|
| 102 |
+
0.19315017759799957,
|
| 103 |
+
0.2880626618862152,
|
| 104 |
+
0.1807917058467865,
|
| 105 |
+
0.33443471789360046
|
| 106 |
+
],
|
| 107 |
+
"q01": [
|
| 108 |
+
-0.8353484272956848,
|
| 109 |
+
-0.43082091212272644,
|
| 110 |
+
-0.01853959634900093,
|
| 111 |
+
-2.0081846714019775,
|
| 112 |
+
-0.7463775277137756,
|
| 113 |
+
-0.06663229316473007,
|
| 114 |
+
0.07988695055246353,
|
| 115 |
+
0.0015339808305725455
|
| 116 |
+
],
|
| 117 |
+
"q99": [
|
| 118 |
+
0.28748956322669983,
|
| 119 |
+
-0.14874820411205292,
|
| 120 |
+
0.8437014222145081,
|
| 121 |
+
-1.2486603260040283,
|
| 122 |
+
0.12267052382230759,
|
| 123 |
+
1.0415489673614502,
|
| 124 |
+
0.8113635182380676,
|
| 125 |
+
0.8682331442832947
|
| 126 |
+
]
|
| 127 |
+
}
|
| 128 |
+
},
|
| 129 |
+
"action": {
|
| 130 |
+
"arm_left": {
|
| 131 |
+
"min": [
|
| 132 |
+
-0.3512935936450958,
|
| 133 |
+
-0.0003355582884978503,
|
| 134 |
+
-0.4693501889705658,
|
| 135 |
+
-2.0985336303710938,
|
| 136 |
+
-0.30987611413002014,
|
| 137 |
+
-0.390062540769577,
|
| 138 |
+
-0.756526529788971,
|
| 139 |
+
-0.0015339808305725455
|
| 140 |
+
],
|
| 141 |
+
"max": [
|
| 142 |
+
0.4587441384792328,
|
| 143 |
+
0.4674566686153412,
|
| 144 |
+
0.2403915673494339,
|
| 145 |
+
-0.8097261190414429,
|
| 146 |
+
0.30678418278694153,
|
| 147 |
+
0.7219297289848328,
|
| 148 |
+
0.09214304387569427,
|
| 149 |
+
1.0001554489135742
|
| 150 |
+
],
|
| 151 |
+
"mean": [
|
| 152 |
+
0.04590541869401932,
|
| 153 |
+
0.10304273664951324,
|
| 154 |
+
-0.08747139573097229,
|
| 155 |
+
-1.721885085105896,
|
| 156 |
+
0.016054483130574226,
|
| 157 |
+
0.22288048267364502,
|
| 158 |
+
-0.3089201748371124,
|
| 159 |
+
0.23478840291500092
|
| 160 |
+
],
|
| 161 |
+
"std": [
|
| 162 |
+
0.20350636541843414,
|
| 163 |
+
0.09618024528026581,
|
| 164 |
+
0.16804173588752747,
|
| 165 |
+
0.24068261682987166,
|
| 166 |
+
0.1208517923951149,
|
| 167 |
+
0.26317518949508667,
|
| 168 |
+
0.22766965627670288,
|
| 169 |
+
0.3860923647880554
|
| 170 |
+
],
|
| 171 |
+
"q01": [
|
| 172 |
+
-0.28486502170562744,
|
| 173 |
+
0.0,
|
| 174 |
+
-0.4279566705226898,
|
| 175 |
+
-2.013697385787964,
|
| 176 |
+
-0.2523278594017029,
|
| 177 |
+
-0.2877771854400635,
|
| 178 |
+
-0.7209579944610596,
|
| 179 |
+
0.0
|
| 180 |
+
],
|
| 181 |
+
"q99": [
|
| 182 |
+
0.41167011857032776,
|
| 183 |
+
0.4564431607723236,
|
| 184 |
+
0.1860191375017166,
|
| 185 |
+
-0.9433622360229492,
|
| 186 |
+
0.2618792951107025,
|
| 187 |
+
0.7118869423866272,
|
| 188 |
+
0.06595534831285477,
|
| 189 |
+
0.9970875382423401
|
| 190 |
+
]
|
| 191 |
+
},
|
| 192 |
+
"arm_right": {
|
| 193 |
+
"min": [
|
| 194 |
+
-0.9125747680664062,
|
| 195 |
+
-0.46813976764678955,
|
| 196 |
+
-0.08915065228939056,
|
| 197 |
+
-2.0727076530456543,
|
| 198 |
+
-0.9404261112213135,
|
| 199 |
+
-0.15235544741153717,
|
| 200 |
+
-0.053947482258081436,
|
| 201 |
+
0.0
|
| 202 |
+
],
|
| 203 |
+
"max": [
|
| 204 |
+
0.3974807858467102,
|
| 205 |
+
-0.031806133687496185,
|
| 206 |
+
1.0656253099441528,
|
| 207 |
+
-1.032512903213501,
|
| 208 |
+
0.13881327211856842,
|
| 209 |
+
1.1778095960617065,
|
| 210 |
+
0.8715579509735107,
|
| 211 |
+
1.277806043624878
|
| 212 |
+
],
|
| 213 |
+
"mean": [
|
| 214 |
+
-0.255352258682251,
|
| 215 |
+
-0.2675997316837311,
|
| 216 |
+
0.3033587634563446,
|
| 217 |
+
-1.7069568634033203,
|
| 218 |
+
-0.2512829005718231,
|
| 219 |
+
0.5092591047286987,
|
| 220 |
+
0.43265601992607117,
|
| 221 |
+
0.3857439160346985
|
| 222 |
+
],
|
| 223 |
+
"std": [
|
| 224 |
+
0.25960078835487366,
|
| 225 |
+
0.059564266353845596,
|
| 226 |
+
0.17328619956970215,
|
| 227 |
+
0.1838947683572763,
|
| 228 |
+
0.19315995275974274,
|
| 229 |
+
0.2880189120769501,
|
| 230 |
+
0.18076331913471222,
|
| 231 |
+
0.3346700072288513
|
| 232 |
+
],
|
| 233 |
+
"q01": [
|
| 234 |
+
-0.8353484272956848,
|
| 235 |
+
-0.43082091212272644,
|
| 236 |
+
-0.01853959634900093,
|
| 237 |
+
-2.0086400508880615,
|
| 238 |
+
-0.7463775277137756,
|
| 239 |
+
-0.06663229316473007,
|
| 240 |
+
0.07988695055246353,
|
| 241 |
+
0.0015339808305725455
|
| 242 |
+
],
|
| 243 |
+
"q99": [
|
| 244 |
+
0.288448303937912,
|
| 245 |
+
-0.14868827164173126,
|
| 246 |
+
0.8437014222145081,
|
| 247 |
+
-1.2486603260040283,
|
| 248 |
+
0.12267052382230759,
|
| 249 |
+
1.0415489673614502,
|
| 250 |
+
0.8113635182380676,
|
| 251 |
+
0.8682331442832947
|
| 252 |
+
]
|
| 253 |
+
}
|
| 254 |
+
},
|
| 255 |
+
"relative_action": {}
|
| 256 |
+
}
|
| 257 |
+
}
|
16d_30k/checkpoint-5000/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": false,
|
| 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 |
+
}
|
16d_30k/checkpoint-5000/experiment_cfg/final_processor_config.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
16d_30k/checkpoint-5000/model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:841f25e80d40ffdf3aff3d2d48fc007d94459d37f2726fb39b8aa4211babf347
|
| 3 |
+
size 4990519232
|
16d_30k/checkpoint-5000/model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c8ab8ee3801047af13ee0f7ef482c489cf13504e8730a5a701c919daf2c85c0a
|
| 3 |
+
size 1919980184
|
16d_30k/checkpoint-5000/model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
16d_30k/checkpoint-5000/processor_config.json
ADDED
|
@@ -0,0 +1,1124 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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 |
+
"new_embodiment": {
|
| 1013 |
+
"video": {
|
| 1014 |
+
"delta_indices": [
|
| 1015 |
+
0
|
| 1016 |
+
],
|
| 1017 |
+
"modality_keys": [
|
| 1018 |
+
"cam_left_head",
|
| 1019 |
+
"cam_left_wrist",
|
| 1020 |
+
"cam_right_wrist"
|
| 1021 |
+
],
|
| 1022 |
+
"sin_cos_embedding_keys": null,
|
| 1023 |
+
"mean_std_embedding_keys": null,
|
| 1024 |
+
"action_configs": null
|
| 1025 |
+
},
|
| 1026 |
+
"state": {
|
| 1027 |
+
"delta_indices": [
|
| 1028 |
+
0
|
| 1029 |
+
],
|
| 1030 |
+
"modality_keys": [
|
| 1031 |
+
"arm_left",
|
| 1032 |
+
"arm_right"
|
| 1033 |
+
],
|
| 1034 |
+
"sin_cos_embedding_keys": null,
|
| 1035 |
+
"mean_std_embedding_keys": null,
|
| 1036 |
+
"action_configs": null
|
| 1037 |
+
},
|
| 1038 |
+
"action": {
|
| 1039 |
+
"delta_indices": [
|
| 1040 |
+
0,
|
| 1041 |
+
1,
|
| 1042 |
+
2,
|
| 1043 |
+
3,
|
| 1044 |
+
4,
|
| 1045 |
+
5,
|
| 1046 |
+
6,
|
| 1047 |
+
7,
|
| 1048 |
+
8,
|
| 1049 |
+
9,
|
| 1050 |
+
10,
|
| 1051 |
+
11,
|
| 1052 |
+
12,
|
| 1053 |
+
13,
|
| 1054 |
+
14,
|
| 1055 |
+
15
|
| 1056 |
+
],
|
| 1057 |
+
"modality_keys": [
|
| 1058 |
+
"arm_left",
|
| 1059 |
+
"arm_right"
|
| 1060 |
+
],
|
| 1061 |
+
"sin_cos_embedding_keys": null,
|
| 1062 |
+
"mean_std_embedding_keys": null,
|
| 1063 |
+
"action_configs": [
|
| 1064 |
+
{
|
| 1065 |
+
"rep": "ABSOLUTE",
|
| 1066 |
+
"type": "NON_EEF",
|
| 1067 |
+
"format": "DEFAULT",
|
| 1068 |
+
"state_key": null
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"rep": "ABSOLUTE",
|
| 1072 |
+
"type": "NON_EEF",
|
| 1073 |
+
"format": "DEFAULT",
|
| 1074 |
+
"state_key": null
|
| 1075 |
+
}
|
| 1076 |
+
]
|
| 1077 |
+
},
|
| 1078 |
+
"language": {
|
| 1079 |
+
"delta_indices": [
|
| 1080 |
+
0
|
| 1081 |
+
],
|
| 1082 |
+
"modality_keys": [
|
| 1083 |
+
"annotation.human.primitive_instruction"
|
| 1084 |
+
],
|
| 1085 |
+
"sin_cos_embedding_keys": null,
|
| 1086 |
+
"mean_std_embedding_keys": null,
|
| 1087 |
+
"action_configs": null
|
| 1088 |
+
}
|
| 1089 |
+
}
|
| 1090 |
+
},
|
| 1091 |
+
"image_crop_size": [
|
| 1092 |
+
230,
|
| 1093 |
+
230
|
| 1094 |
+
],
|
| 1095 |
+
"image_target_size": [
|
| 1096 |
+
256,
|
| 1097 |
+
256
|
| 1098 |
+
],
|
| 1099 |
+
"use_albumentations": true,
|
| 1100 |
+
"random_rotation_angle": 0,
|
| 1101 |
+
"color_jitter_params": {
|
| 1102 |
+
"brightness": 0.3,
|
| 1103 |
+
"contrast": 0.4,
|
| 1104 |
+
"saturation": 0.5,
|
| 1105 |
+
"hue": 0.08
|
| 1106 |
+
},
|
| 1107 |
+
"shortest_image_edge": 256,
|
| 1108 |
+
"crop_fraction": 1.0,
|
| 1109 |
+
"letter_box_transform": false,
|
| 1110 |
+
"model_name": "nvidia/Cosmos-Reason2-2B",
|
| 1111 |
+
"model_type": "qwen",
|
| 1112 |
+
"formalize_language": true,
|
| 1113 |
+
"max_state_dim": 132,
|
| 1114 |
+
"max_action_dim": 132,
|
| 1115 |
+
"max_action_horizon": 40,
|
| 1116 |
+
"use_percentiles": false,
|
| 1117 |
+
"use_mean_std": false,
|
| 1118 |
+
"clip_outliers": true,
|
| 1119 |
+
"apply_sincos_state_encoding": false,
|
| 1120 |
+
"use_relative_action": true,
|
| 1121 |
+
"exclude_state": false,
|
| 1122 |
+
"state_dropout_prob": 0.2
|
| 1123 |
+
}
|
| 1124 |
+
}
|
16d_30k/checkpoint-5000/statistics.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
16d_30k/checkpoint-5000/trainer_state.json
ADDED
|
@@ -0,0 +1,3034 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.16666666666666666,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 5000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"grad_norm": 0.6208589673042297,
|
| 14 |
+
"learning_rate": 6.000000000000001e-07,
|
| 15 |
+
"loss": 1.2746,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"grad_norm": 0.6478228569030762,
|
| 20 |
+
"learning_rate": 1.2666666666666667e-06,
|
| 21 |
+
"loss": 1.282,
|
| 22 |
+
"step": 20
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"grad_norm": 0.5199011564254761,
|
| 26 |
+
"learning_rate": 1.9333333333333336e-06,
|
| 27 |
+
"loss": 1.2766,
|
| 28 |
+
"step": 30
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"grad_norm": 0.6404500603675842,
|
| 32 |
+
"learning_rate": 2.6e-06,
|
| 33 |
+
"loss": 1.2254,
|
| 34 |
+
"step": 40
|
| 35 |
+
},
|
| 36 |
+
{
|
| 37 |
+
"grad_norm": 0.5328770279884338,
|
| 38 |
+
"learning_rate": 3.2666666666666666e-06,
|
| 39 |
+
"loss": 1.1938,
|
| 40 |
+
"step": 50
|
| 41 |
+
},
|
| 42 |
+
{
|
| 43 |
+
"grad_norm": 0.5413777232170105,
|
| 44 |
+
"learning_rate": 3.9333333333333335e-06,
|
| 45 |
+
"loss": 1.1359,
|
| 46 |
+
"step": 60
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"grad_norm": 0.49827340245246887,
|
| 50 |
+
"learning_rate": 4.6e-06,
|
| 51 |
+
"loss": 1.0902,
|
| 52 |
+
"step": 70
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"grad_norm": 0.46059373021125793,
|
| 56 |
+
"learning_rate": 5.266666666666667e-06,
|
| 57 |
+
"loss": 1.0633,
|
| 58 |
+
"step": 80
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"grad_norm": 0.4852186441421509,
|
| 62 |
+
"learning_rate": 5.933333333333334e-06,
|
| 63 |
+
"loss": 1.048,
|
| 64 |
+
"step": 90
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"grad_norm": 0.644527018070221,
|
| 68 |
+
"learning_rate": 6.6e-06,
|
| 69 |
+
"loss": 1.0359,
|
| 70 |
+
"step": 100
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"grad_norm": 0.6214720606803894,
|
| 74 |
+
"learning_rate": 7.266666666666668e-06,
|
| 75 |
+
"loss": 1.0328,
|
| 76 |
+
"step": 110
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"grad_norm": 0.38992390036582947,
|
| 80 |
+
"learning_rate": 7.933333333333334e-06,
|
| 81 |
+
"loss": 1.0285,
|
| 82 |
+
"step": 120
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"grad_norm": 0.5468058586120605,
|
| 86 |
+
"learning_rate": 8.599999999999999e-06,
|
| 87 |
+
"loss": 1.0145,
|
| 88 |
+
"step": 130
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"grad_norm": 0.743439257144928,
|
| 92 |
+
"learning_rate": 9.266666666666667e-06,
|
| 93 |
+
"loss": 1.0,
|
| 94 |
+
"step": 140
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"grad_norm": 0.9581028819084167,
|
| 98 |
+
"learning_rate": 9.933333333333334e-06,
|
| 99 |
+
"loss": 0.9418,
|
| 100 |
+
"step": 150
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"grad_norm": 1.0746005773544312,
|
| 104 |
+
"learning_rate": 1.06e-05,
|
| 105 |
+
"loss": 0.8699,
|
| 106 |
+
"step": 160
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"grad_norm": 1.3099654912948608,
|
| 110 |
+
"learning_rate": 1.1266666666666667e-05,
|
| 111 |
+
"loss": 0.7877,
|
| 112 |
+
"step": 170
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"grad_norm": 1.502665400505066,
|
| 116 |
+
"learning_rate": 1.1933333333333333e-05,
|
| 117 |
+
"loss": 0.7059,
|
| 118 |
+
"step": 180
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"grad_norm": 1.873738408088684,
|
| 122 |
+
"learning_rate": 1.2600000000000001e-05,
|
| 123 |
+
"loss": 0.6436,
|
| 124 |
+
"step": 190
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"grad_norm": 2.2341554164886475,
|
| 128 |
+
"learning_rate": 1.3266666666666666e-05,
|
| 129 |
+
"loss": 0.5734,
|
| 130 |
+
"step": 200
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"grad_norm": 1.5233055353164673,
|
| 134 |
+
"learning_rate": 1.3933333333333334e-05,
|
| 135 |
+
"loss": 0.5299,
|
| 136 |
+
"step": 210
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"grad_norm": 1.5989943742752075,
|
| 140 |
+
"learning_rate": 1.4599999999999999e-05,
|
| 141 |
+
"loss": 0.482,
|
| 142 |
+
"step": 220
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"grad_norm": 1.6445683240890503,
|
| 146 |
+
"learning_rate": 1.5266666666666667e-05,
|
| 147 |
+
"loss": 0.4228,
|
| 148 |
+
"step": 230
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"grad_norm": 1.5990145206451416,
|
| 152 |
+
"learning_rate": 1.5933333333333332e-05,
|
| 153 |
+
"loss": 0.3652,
|
| 154 |
+
"step": 240
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"grad_norm": 1.929441213607788,
|
| 158 |
+
"learning_rate": 1.66e-05,
|
| 159 |
+
"loss": 0.3081,
|
| 160 |
+
"step": 250
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"grad_norm": 2.241440773010254,
|
| 164 |
+
"learning_rate": 1.726666666666667e-05,
|
| 165 |
+
"loss": 0.2768,
|
| 166 |
+
"step": 260
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"grad_norm": 1.6238600015640259,
|
| 170 |
+
"learning_rate": 1.7933333333333337e-05,
|
| 171 |
+
"loss": 0.2393,
|
| 172 |
+
"step": 270
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"grad_norm": 1.9927496910095215,
|
| 176 |
+
"learning_rate": 1.86e-05,
|
| 177 |
+
"loss": 0.1933,
|
| 178 |
+
"step": 280
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"grad_norm": 2.0286266803741455,
|
| 182 |
+
"learning_rate": 1.926666666666667e-05,
|
| 183 |
+
"loss": 0.1622,
|
| 184 |
+
"step": 290
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"grad_norm": 1.611731767654419,
|
| 188 |
+
"learning_rate": 1.9933333333333334e-05,
|
| 189 |
+
"loss": 0.1558,
|
| 190 |
+
"step": 300
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"grad_norm": 1.767244815826416,
|
| 194 |
+
"learning_rate": 2.06e-05,
|
| 195 |
+
"loss": 0.1541,
|
| 196 |
+
"step": 310
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"grad_norm": 1.6201092004776,
|
| 200 |
+
"learning_rate": 2.1266666666666667e-05,
|
| 201 |
+
"loss": 0.1285,
|
| 202 |
+
"step": 320
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"grad_norm": 1.7983145713806152,
|
| 206 |
+
"learning_rate": 2.1933333333333332e-05,
|
| 207 |
+
"loss": 0.1288,
|
| 208 |
+
"step": 330
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"grad_norm": 2.1310415267944336,
|
| 212 |
+
"learning_rate": 2.26e-05,
|
| 213 |
+
"loss": 0.1355,
|
| 214 |
+
"step": 340
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"grad_norm": 1.7930408716201782,
|
| 218 |
+
"learning_rate": 2.326666666666667e-05,
|
| 219 |
+
"loss": 0.124,
|
| 220 |
+
"step": 350
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"grad_norm": 1.5554205179214478,
|
| 224 |
+
"learning_rate": 2.3933333333333337e-05,
|
| 225 |
+
"loss": 0.1252,
|
| 226 |
+
"step": 360
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"grad_norm": 1.9930179119110107,
|
| 230 |
+
"learning_rate": 2.46e-05,
|
| 231 |
+
"loss": 0.1045,
|
| 232 |
+
"step": 370
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"grad_norm": 1.8832621574401855,
|
| 236 |
+
"learning_rate": 2.5266666666666666e-05,
|
| 237 |
+
"loss": 0.1069,
|
| 238 |
+
"step": 380
|
| 239 |
+
},
|
| 240 |
+
{
|
| 241 |
+
"grad_norm": 1.9651679992675781,
|
| 242 |
+
"learning_rate": 2.5933333333333338e-05,
|
| 243 |
+
"loss": 0.1077,
|
| 244 |
+
"step": 390
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"grad_norm": 1.5526353120803833,
|
| 248 |
+
"learning_rate": 2.6600000000000003e-05,
|
| 249 |
+
"loss": 0.1046,
|
| 250 |
+
"step": 400
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"grad_norm": 1.75920569896698,
|
| 254 |
+
"learning_rate": 2.7266666666666668e-05,
|
| 255 |
+
"loss": 0.1031,
|
| 256 |
+
"step": 410
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"grad_norm": 1.6312711238861084,
|
| 260 |
+
"learning_rate": 2.7933333333333332e-05,
|
| 261 |
+
"loss": 0.0985,
|
| 262 |
+
"step": 420
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"grad_norm": 1.9239977598190308,
|
| 266 |
+
"learning_rate": 2.86e-05,
|
| 267 |
+
"loss": 0.1029,
|
| 268 |
+
"step": 430
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"grad_norm": 1.6071902513504028,
|
| 272 |
+
"learning_rate": 2.926666666666667e-05,
|
| 273 |
+
"loss": 0.092,
|
| 274 |
+
"step": 440
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"grad_norm": 1.3962411880493164,
|
| 278 |
+
"learning_rate": 2.9933333333333337e-05,
|
| 279 |
+
"loss": 0.0927,
|
| 280 |
+
"step": 450
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"grad_norm": 1.8652002811431885,
|
| 284 |
+
"learning_rate": 3.06e-05,
|
| 285 |
+
"loss": 0.0897,
|
| 286 |
+
"step": 460
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"grad_norm": 1.4422093629837036,
|
| 290 |
+
"learning_rate": 3.126666666666666e-05,
|
| 291 |
+
"loss": 0.0838,
|
| 292 |
+
"step": 470
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"grad_norm": 1.1935328245162964,
|
| 296 |
+
"learning_rate": 3.1933333333333335e-05,
|
| 297 |
+
"loss": 0.0873,
|
| 298 |
+
"step": 480
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"grad_norm": 1.549795389175415,
|
| 302 |
+
"learning_rate": 3.26e-05,
|
| 303 |
+
"loss": 0.0846,
|
| 304 |
+
"step": 490
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"grad_norm": 1.3405718803405762,
|
| 308 |
+
"learning_rate": 3.326666666666667e-05,
|
| 309 |
+
"loss": 0.0866,
|
| 310 |
+
"step": 500
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"grad_norm": 1.535412311553955,
|
| 314 |
+
"learning_rate": 3.3933333333333336e-05,
|
| 315 |
+
"loss": 0.0806,
|
| 316 |
+
"step": 510
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"grad_norm": 1.7111966609954834,
|
| 320 |
+
"learning_rate": 3.46e-05,
|
| 321 |
+
"loss": 0.084,
|
| 322 |
+
"step": 520
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"grad_norm": 1.424939513206482,
|
| 326 |
+
"learning_rate": 3.526666666666667e-05,
|
| 327 |
+
"loss": 0.0822,
|
| 328 |
+
"step": 530
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"grad_norm": 2.0077829360961914,
|
| 332 |
+
"learning_rate": 3.593333333333334e-05,
|
| 333 |
+
"loss": 0.0831,
|
| 334 |
+
"step": 540
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"grad_norm": 1.394214391708374,
|
| 338 |
+
"learning_rate": 3.66e-05,
|
| 339 |
+
"loss": 0.0741,
|
| 340 |
+
"step": 550
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"grad_norm": 1.670121431350708,
|
| 344 |
+
"learning_rate": 3.726666666666667e-05,
|
| 345 |
+
"loss": 0.0842,
|
| 346 |
+
"step": 560
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"grad_norm": 1.5677956342697144,
|
| 350 |
+
"learning_rate": 3.793333333333334e-05,
|
| 351 |
+
"loss": 0.0877,
|
| 352 |
+
"step": 570
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"grad_norm": 1.7964739799499512,
|
| 356 |
+
"learning_rate": 3.86e-05,
|
| 357 |
+
"loss": 0.0815,
|
| 358 |
+
"step": 580
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"grad_norm": 1.4517905712127686,
|
| 362 |
+
"learning_rate": 3.926666666666667e-05,
|
| 363 |
+
"loss": 0.0792,
|
| 364 |
+
"step": 590
|
| 365 |
+
},
|
| 366 |
+
{
|
| 367 |
+
"grad_norm": 1.4527804851531982,
|
| 368 |
+
"learning_rate": 3.993333333333333e-05,
|
| 369 |
+
"loss": 0.0793,
|
| 370 |
+
"step": 600
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"grad_norm": 1.163316011428833,
|
| 374 |
+
"learning_rate": 4.0600000000000004e-05,
|
| 375 |
+
"loss": 0.0745,
|
| 376 |
+
"step": 610
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"grad_norm": 1.5932477712631226,
|
| 380 |
+
"learning_rate": 4.126666666666667e-05,
|
| 381 |
+
"loss": 0.0754,
|
| 382 |
+
"step": 620
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"grad_norm": 1.437664270401001,
|
| 386 |
+
"learning_rate": 4.1933333333333334e-05,
|
| 387 |
+
"loss": 0.0756,
|
| 388 |
+
"step": 630
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"grad_norm": 1.3479504585266113,
|
| 392 |
+
"learning_rate": 4.26e-05,
|
| 393 |
+
"loss": 0.0778,
|
| 394 |
+
"step": 640
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"grad_norm": 1.1809881925582886,
|
| 398 |
+
"learning_rate": 4.3266666666666664e-05,
|
| 399 |
+
"loss": 0.0791,
|
| 400 |
+
"step": 650
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"grad_norm": 1.592273235321045,
|
| 404 |
+
"learning_rate": 4.3933333333333335e-05,
|
| 405 |
+
"loss": 0.0726,
|
| 406 |
+
"step": 660
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"grad_norm": 1.2641804218292236,
|
| 410 |
+
"learning_rate": 4.46e-05,
|
| 411 |
+
"loss": 0.0692,
|
| 412 |
+
"step": 670
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"grad_norm": 1.5784950256347656,
|
| 416 |
+
"learning_rate": 4.526666666666667e-05,
|
| 417 |
+
"loss": 0.0671,
|
| 418 |
+
"step": 680
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"grad_norm": 1.8079859018325806,
|
| 422 |
+
"learning_rate": 4.5933333333333336e-05,
|
| 423 |
+
"loss": 0.0648,
|
| 424 |
+
"step": 690
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"grad_norm": 1.452563762664795,
|
| 428 |
+
"learning_rate": 4.660000000000001e-05,
|
| 429 |
+
"loss": 0.0725,
|
| 430 |
+
"step": 700
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"grad_norm": 1.6882253885269165,
|
| 434 |
+
"learning_rate": 4.726666666666667e-05,
|
| 435 |
+
"loss": 0.0678,
|
| 436 |
+
"step": 710
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"grad_norm": 1.1656196117401123,
|
| 440 |
+
"learning_rate": 4.793333333333334e-05,
|
| 441 |
+
"loss": 0.0683,
|
| 442 |
+
"step": 720
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"grad_norm": 1.4506734609603882,
|
| 446 |
+
"learning_rate": 4.86e-05,
|
| 447 |
+
"loss": 0.071,
|
| 448 |
+
"step": 730
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"grad_norm": 1.171644687652588,
|
| 452 |
+
"learning_rate": 4.926666666666667e-05,
|
| 453 |
+
"loss": 0.0742,
|
| 454 |
+
"step": 740
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"grad_norm": 1.3164993524551392,
|
| 458 |
+
"learning_rate": 4.993333333333334e-05,
|
| 459 |
+
"loss": 0.0717,
|
| 460 |
+
"step": 750
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"grad_norm": 1.38760244846344,
|
| 464 |
+
"learning_rate": 5.0600000000000003e-05,
|
| 465 |
+
"loss": 0.0654,
|
| 466 |
+
"step": 760
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"grad_norm": 1.2424696683883667,
|
| 470 |
+
"learning_rate": 5.1266666666666675e-05,
|
| 471 |
+
"loss": 0.0722,
|
| 472 |
+
"step": 770
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"grad_norm": 1.150290608406067,
|
| 476 |
+
"learning_rate": 5.193333333333333e-05,
|
| 477 |
+
"loss": 0.0657,
|
| 478 |
+
"step": 780
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"grad_norm": 0.8061581254005432,
|
| 482 |
+
"learning_rate": 5.2600000000000005e-05,
|
| 483 |
+
"loss": 0.0622,
|
| 484 |
+
"step": 790
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"grad_norm": 1.1061843633651733,
|
| 488 |
+
"learning_rate": 5.326666666666666e-05,
|
| 489 |
+
"loss": 0.0637,
|
| 490 |
+
"step": 800
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"grad_norm": 1.2182507514953613,
|
| 494 |
+
"learning_rate": 5.3933333333333334e-05,
|
| 495 |
+
"loss": 0.0589,
|
| 496 |
+
"step": 810
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"grad_norm": 1.4238909482955933,
|
| 500 |
+
"learning_rate": 5.4600000000000006e-05,
|
| 501 |
+
"loss": 0.0614,
|
| 502 |
+
"step": 820
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"grad_norm": 0.9367620944976807,
|
| 506 |
+
"learning_rate": 5.5266666666666664e-05,
|
| 507 |
+
"loss": 0.0622,
|
| 508 |
+
"step": 830
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"grad_norm": 1.3980387449264526,
|
| 512 |
+
"learning_rate": 5.5933333333333335e-05,
|
| 513 |
+
"loss": 0.065,
|
| 514 |
+
"step": 840
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"grad_norm": 1.3358958959579468,
|
| 518 |
+
"learning_rate": 5.66e-05,
|
| 519 |
+
"loss": 0.068,
|
| 520 |
+
"step": 850
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"grad_norm": 0.9675244092941284,
|
| 524 |
+
"learning_rate": 5.726666666666667e-05,
|
| 525 |
+
"loss": 0.0634,
|
| 526 |
+
"step": 860
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"grad_norm": 0.9360224008560181,
|
| 530 |
+
"learning_rate": 5.7933333333333337e-05,
|
| 531 |
+
"loss": 0.0586,
|
| 532 |
+
"step": 870
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"grad_norm": 1.0572683811187744,
|
| 536 |
+
"learning_rate": 5.86e-05,
|
| 537 |
+
"loss": 0.0641,
|
| 538 |
+
"step": 880
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"grad_norm": 1.307563066482544,
|
| 542 |
+
"learning_rate": 5.926666666666667e-05,
|
| 543 |
+
"loss": 0.0593,
|
| 544 |
+
"step": 890
|
| 545 |
+
},
|
| 546 |
+
{
|
| 547 |
+
"grad_norm": 0.8857334852218628,
|
| 548 |
+
"learning_rate": 5.9933333333333345e-05,
|
| 549 |
+
"loss": 0.0634,
|
| 550 |
+
"step": 900
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"grad_norm": 1.2183889150619507,
|
| 554 |
+
"learning_rate": 6.06e-05,
|
| 555 |
+
"loss": 0.0571,
|
| 556 |
+
"step": 910
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"grad_norm": 1.2149014472961426,
|
| 560 |
+
"learning_rate": 6.126666666666667e-05,
|
| 561 |
+
"loss": 0.0586,
|
| 562 |
+
"step": 920
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"grad_norm": 0.971682071685791,
|
| 566 |
+
"learning_rate": 6.193333333333333e-05,
|
| 567 |
+
"loss": 0.0609,
|
| 568 |
+
"step": 930
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"grad_norm": 1.2368853092193604,
|
| 572 |
+
"learning_rate": 6.26e-05,
|
| 573 |
+
"loss": 0.0595,
|
| 574 |
+
"step": 940
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"grad_norm": 0.9683165550231934,
|
| 578 |
+
"learning_rate": 6.326666666666667e-05,
|
| 579 |
+
"loss": 0.0616,
|
| 580 |
+
"step": 950
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"grad_norm": 0.8244102001190186,
|
| 584 |
+
"learning_rate": 6.393333333333333e-05,
|
| 585 |
+
"loss": 0.0575,
|
| 586 |
+
"step": 960
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"grad_norm": 0.970955491065979,
|
| 590 |
+
"learning_rate": 6.460000000000001e-05,
|
| 591 |
+
"loss": 0.0581,
|
| 592 |
+
"step": 970
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"grad_norm": 1.0662046670913696,
|
| 596 |
+
"learning_rate": 6.526666666666666e-05,
|
| 597 |
+
"loss": 0.0557,
|
| 598 |
+
"step": 980
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"grad_norm": 0.9161382913589478,
|
| 602 |
+
"learning_rate": 6.593333333333334e-05,
|
| 603 |
+
"loss": 0.0622,
|
| 604 |
+
"step": 990
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"grad_norm": 0.8752316832542419,
|
| 608 |
+
"learning_rate": 6.66e-05,
|
| 609 |
+
"loss": 0.0539,
|
| 610 |
+
"step": 1000
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"grad_norm": 0.7521458268165588,
|
| 614 |
+
"learning_rate": 6.726666666666667e-05,
|
| 615 |
+
"loss": 0.0506,
|
| 616 |
+
"step": 1010
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"grad_norm": 0.8169746994972229,
|
| 620 |
+
"learning_rate": 6.793333333333334e-05,
|
| 621 |
+
"loss": 0.0603,
|
| 622 |
+
"step": 1020
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"grad_norm": 0.8718522191047668,
|
| 626 |
+
"learning_rate": 6.860000000000001e-05,
|
| 627 |
+
"loss": 0.0516,
|
| 628 |
+
"step": 1030
|
| 629 |
+
},
|
| 630 |
+
{
|
| 631 |
+
"grad_norm": 0.7131385207176208,
|
| 632 |
+
"learning_rate": 6.926666666666667e-05,
|
| 633 |
+
"loss": 0.0492,
|
| 634 |
+
"step": 1040
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"grad_norm": 0.8163769841194153,
|
| 638 |
+
"learning_rate": 6.993333333333334e-05,
|
| 639 |
+
"loss": 0.0475,
|
| 640 |
+
"step": 1050
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"grad_norm": 0.6880462169647217,
|
| 644 |
+
"learning_rate": 7.06e-05,
|
| 645 |
+
"loss": 0.0457,
|
| 646 |
+
"step": 1060
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"grad_norm": 1.1300956010818481,
|
| 650 |
+
"learning_rate": 7.126666666666667e-05,
|
| 651 |
+
"loss": 0.0508,
|
| 652 |
+
"step": 1070
|
| 653 |
+
},
|
| 654 |
+
{
|
| 655 |
+
"grad_norm": 1.0228790044784546,
|
| 656 |
+
"learning_rate": 7.193333333333334e-05,
|
| 657 |
+
"loss": 0.0524,
|
| 658 |
+
"step": 1080
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"grad_norm": 1.0921752452850342,
|
| 662 |
+
"learning_rate": 7.26e-05,
|
| 663 |
+
"loss": 0.0499,
|
| 664 |
+
"step": 1090
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"grad_norm": 0.5290265679359436,
|
| 668 |
+
"learning_rate": 7.326666666666667e-05,
|
| 669 |
+
"loss": 0.0495,
|
| 670 |
+
"step": 1100
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"grad_norm": 1.0428764820098877,
|
| 674 |
+
"learning_rate": 7.393333333333333e-05,
|
| 675 |
+
"loss": 0.0481,
|
| 676 |
+
"step": 1110
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"grad_norm": 0.6707554459571838,
|
| 680 |
+
"learning_rate": 7.46e-05,
|
| 681 |
+
"loss": 0.0487,
|
| 682 |
+
"step": 1120
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"grad_norm": 0.7440798282623291,
|
| 686 |
+
"learning_rate": 7.526666666666668e-05,
|
| 687 |
+
"loss": 0.0509,
|
| 688 |
+
"step": 1130
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"grad_norm": 0.9219576120376587,
|
| 692 |
+
"learning_rate": 7.593333333333334e-05,
|
| 693 |
+
"loss": 0.0501,
|
| 694 |
+
"step": 1140
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"grad_norm": 1.2103952169418335,
|
| 698 |
+
"learning_rate": 7.66e-05,
|
| 699 |
+
"loss": 0.0483,
|
| 700 |
+
"step": 1150
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"grad_norm": 0.8884477615356445,
|
| 704 |
+
"learning_rate": 7.726666666666667e-05,
|
| 705 |
+
"loss": 0.0512,
|
| 706 |
+
"step": 1160
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"grad_norm": 0.9153897762298584,
|
| 710 |
+
"learning_rate": 7.793333333333333e-05,
|
| 711 |
+
"loss": 0.0508,
|
| 712 |
+
"step": 1170
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"grad_norm": 0.6855103969573975,
|
| 716 |
+
"learning_rate": 7.860000000000001e-05,
|
| 717 |
+
"loss": 0.0479,
|
| 718 |
+
"step": 1180
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"grad_norm": 0.6617424488067627,
|
| 722 |
+
"learning_rate": 7.926666666666666e-05,
|
| 723 |
+
"loss": 0.0438,
|
| 724 |
+
"step": 1190
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"grad_norm": 0.8509991765022278,
|
| 728 |
+
"learning_rate": 7.993333333333334e-05,
|
| 729 |
+
"loss": 0.0502,
|
| 730 |
+
"step": 1200
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"grad_norm": 0.8399519920349121,
|
| 734 |
+
"learning_rate": 8.060000000000001e-05,
|
| 735 |
+
"loss": 0.0466,
|
| 736 |
+
"step": 1210
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"grad_norm": 0.6967006325721741,
|
| 740 |
+
"learning_rate": 8.126666666666667e-05,
|
| 741 |
+
"loss": 0.0458,
|
| 742 |
+
"step": 1220
|
| 743 |
+
},
|
| 744 |
+
{
|
| 745 |
+
"grad_norm": 0.7278540134429932,
|
| 746 |
+
"learning_rate": 8.193333333333334e-05,
|
| 747 |
+
"loss": 0.047,
|
| 748 |
+
"step": 1230
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"grad_norm": 0.8349273204803467,
|
| 752 |
+
"learning_rate": 8.26e-05,
|
| 753 |
+
"loss": 0.0448,
|
| 754 |
+
"step": 1240
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"grad_norm": 0.8303181529045105,
|
| 758 |
+
"learning_rate": 8.326666666666667e-05,
|
| 759 |
+
"loss": 0.0431,
|
| 760 |
+
"step": 1250
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"grad_norm": 0.9701292514801025,
|
| 764 |
+
"learning_rate": 8.393333333333335e-05,
|
| 765 |
+
"loss": 0.0469,
|
| 766 |
+
"step": 1260
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"grad_norm": 0.8359465003013611,
|
| 770 |
+
"learning_rate": 8.46e-05,
|
| 771 |
+
"loss": 0.0444,
|
| 772 |
+
"step": 1270
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"grad_norm": 0.6769126057624817,
|
| 776 |
+
"learning_rate": 8.526666666666667e-05,
|
| 777 |
+
"loss": 0.0483,
|
| 778 |
+
"step": 1280
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"grad_norm": 0.7515086531639099,
|
| 782 |
+
"learning_rate": 8.593333333333333e-05,
|
| 783 |
+
"loss": 0.0463,
|
| 784 |
+
"step": 1290
|
| 785 |
+
},
|
| 786 |
+
{
|
| 787 |
+
"grad_norm": 0.8276355266571045,
|
| 788 |
+
"learning_rate": 8.66e-05,
|
| 789 |
+
"loss": 0.0438,
|
| 790 |
+
"step": 1300
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"grad_norm": 0.5742467045783997,
|
| 794 |
+
"learning_rate": 8.726666666666667e-05,
|
| 795 |
+
"loss": 0.0426,
|
| 796 |
+
"step": 1310
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"grad_norm": 0.8865119814872742,
|
| 800 |
+
"learning_rate": 8.793333333333333e-05,
|
| 801 |
+
"loss": 0.0389,
|
| 802 |
+
"step": 1320
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"grad_norm": 0.6985632181167603,
|
| 806 |
+
"learning_rate": 8.86e-05,
|
| 807 |
+
"loss": 0.0482,
|
| 808 |
+
"step": 1330
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"grad_norm": 0.7202398180961609,
|
| 812 |
+
"learning_rate": 8.926666666666668e-05,
|
| 813 |
+
"loss": 0.0491,
|
| 814 |
+
"step": 1340
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"grad_norm": 0.6817957758903503,
|
| 818 |
+
"learning_rate": 8.993333333333334e-05,
|
| 819 |
+
"loss": 0.0467,
|
| 820 |
+
"step": 1350
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"grad_norm": 0.6327383518218994,
|
| 824 |
+
"learning_rate": 9.06e-05,
|
| 825 |
+
"loss": 0.0445,
|
| 826 |
+
"step": 1360
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"grad_norm": 0.4578998386859894,
|
| 830 |
+
"learning_rate": 9.126666666666667e-05,
|
| 831 |
+
"loss": 0.0437,
|
| 832 |
+
"step": 1370
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"grad_norm": 0.7874047160148621,
|
| 836 |
+
"learning_rate": 9.193333333333334e-05,
|
| 837 |
+
"loss": 0.0381,
|
| 838 |
+
"step": 1380
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"grad_norm": 0.6632944345474243,
|
| 842 |
+
"learning_rate": 9.260000000000001e-05,
|
| 843 |
+
"loss": 0.0412,
|
| 844 |
+
"step": 1390
|
| 845 |
+
},
|
| 846 |
+
{
|
| 847 |
+
"grad_norm": 0.7300984263420105,
|
| 848 |
+
"learning_rate": 9.326666666666667e-05,
|
| 849 |
+
"loss": 0.0386,
|
| 850 |
+
"step": 1400
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"grad_norm": 0.7691074013710022,
|
| 854 |
+
"learning_rate": 9.393333333333334e-05,
|
| 855 |
+
"loss": 0.0412,
|
| 856 |
+
"step": 1410
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"grad_norm": 0.9861852526664734,
|
| 860 |
+
"learning_rate": 9.46e-05,
|
| 861 |
+
"loss": 0.0404,
|
| 862 |
+
"step": 1420
|
| 863 |
+
},
|
| 864 |
+
{
|
| 865 |
+
"grad_norm": 0.8763015866279602,
|
| 866 |
+
"learning_rate": 9.526666666666667e-05,
|
| 867 |
+
"loss": 0.0415,
|
| 868 |
+
"step": 1430
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"grad_norm": 0.7889045476913452,
|
| 872 |
+
"learning_rate": 9.593333333333334e-05,
|
| 873 |
+
"loss": 0.0401,
|
| 874 |
+
"step": 1440
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"grad_norm": 0.7392305731773376,
|
| 878 |
+
"learning_rate": 9.66e-05,
|
| 879 |
+
"loss": 0.0391,
|
| 880 |
+
"step": 1450
|
| 881 |
+
},
|
| 882 |
+
{
|
| 883 |
+
"grad_norm": 0.7114829421043396,
|
| 884 |
+
"learning_rate": 9.726666666666667e-05,
|
| 885 |
+
"loss": 0.0453,
|
| 886 |
+
"step": 1460
|
| 887 |
+
},
|
| 888 |
+
{
|
| 889 |
+
"grad_norm": 0.5600028038024902,
|
| 890 |
+
"learning_rate": 9.793333333333333e-05,
|
| 891 |
+
"loss": 0.0385,
|
| 892 |
+
"step": 1470
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"grad_norm": 0.7662580013275146,
|
| 896 |
+
"learning_rate": 9.86e-05,
|
| 897 |
+
"loss": 0.0399,
|
| 898 |
+
"step": 1480
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"grad_norm": 0.7539925575256348,
|
| 902 |
+
"learning_rate": 9.926666666666668e-05,
|
| 903 |
+
"loss": 0.0407,
|
| 904 |
+
"step": 1490
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"grad_norm": 0.7231246829032898,
|
| 908 |
+
"learning_rate": 9.993333333333334e-05,
|
| 909 |
+
"loss": 0.04,
|
| 910 |
+
"step": 1500
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"grad_norm": 0.6617307662963867,
|
| 914 |
+
"learning_rate": 9.999997539434007e-05,
|
| 915 |
+
"loss": 0.0429,
|
| 916 |
+
"step": 1510
|
| 917 |
+
},
|
| 918 |
+
{
|
| 919 |
+
"grad_norm": 0.7075979709625244,
|
| 920 |
+
"learning_rate": 9.999989033776898e-05,
|
| 921 |
+
"loss": 0.0377,
|
| 922 |
+
"step": 1520
|
| 923 |
+
},
|
| 924 |
+
{
|
| 925 |
+
"grad_norm": 0.6477856636047363,
|
| 926 |
+
"learning_rate": 9.999974452661641e-05,
|
| 927 |
+
"loss": 0.0417,
|
| 928 |
+
"step": 1530
|
| 929 |
+
},
|
| 930 |
+
{
|
| 931 |
+
"grad_norm": 0.5997700691223145,
|
| 932 |
+
"learning_rate": 9.999953796105959e-05,
|
| 933 |
+
"loss": 0.0394,
|
| 934 |
+
"step": 1540
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"grad_norm": 0.5923762917518616,
|
| 938 |
+
"learning_rate": 9.999927064134949e-05,
|
| 939 |
+
"loss": 0.0372,
|
| 940 |
+
"step": 1550
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"grad_norm": 0.46595168113708496,
|
| 944 |
+
"learning_rate": 9.999894256781095e-05,
|
| 945 |
+
"loss": 0.0385,
|
| 946 |
+
"step": 1560
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"grad_norm": 0.583235502243042,
|
| 950 |
+
"learning_rate": 9.99985537408426e-05,
|
| 951 |
+
"loss": 0.0388,
|
| 952 |
+
"step": 1570
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"grad_norm": 0.7547820806503296,
|
| 956 |
+
"learning_rate": 9.999810416091688e-05,
|
| 957 |
+
"loss": 0.0378,
|
| 958 |
+
"step": 1580
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"grad_norm": 0.49323832988739014,
|
| 962 |
+
"learning_rate": 9.99975938285801e-05,
|
| 963 |
+
"loss": 0.0386,
|
| 964 |
+
"step": 1590
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"grad_norm": 0.6207570433616638,
|
| 968 |
+
"learning_rate": 9.999702274445236e-05,
|
| 969 |
+
"loss": 0.0412,
|
| 970 |
+
"step": 1600
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"grad_norm": 0.7229306697845459,
|
| 974 |
+
"learning_rate": 9.999639090922756e-05,
|
| 975 |
+
"loss": 0.038,
|
| 976 |
+
"step": 1610
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"grad_norm": 0.6029353141784668,
|
| 980 |
+
"learning_rate": 9.999569832367346e-05,
|
| 981 |
+
"loss": 0.0358,
|
| 982 |
+
"step": 1620
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"grad_norm": 0.5695170164108276,
|
| 986 |
+
"learning_rate": 9.999494498863162e-05,
|
| 987 |
+
"loss": 0.038,
|
| 988 |
+
"step": 1630
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"grad_norm": 0.4581107795238495,
|
| 992 |
+
"learning_rate": 9.99941309050174e-05,
|
| 993 |
+
"loss": 0.0336,
|
| 994 |
+
"step": 1640
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"grad_norm": 0.5888658761978149,
|
| 998 |
+
"learning_rate": 9.999325607381999e-05,
|
| 999 |
+
"loss": 0.0363,
|
| 1000 |
+
"step": 1650
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"grad_norm": 0.42355862259864807,
|
| 1004 |
+
"learning_rate": 9.999232049610238e-05,
|
| 1005 |
+
"loss": 0.0329,
|
| 1006 |
+
"step": 1660
|
| 1007 |
+
},
|
| 1008 |
+
{
|
| 1009 |
+
"grad_norm": 0.44192200899124146,
|
| 1010 |
+
"learning_rate": 9.999132417300142e-05,
|
| 1011 |
+
"loss": 0.0318,
|
| 1012 |
+
"step": 1670
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"grad_norm": 0.582292377948761,
|
| 1016 |
+
"learning_rate": 9.99902671057277e-05,
|
| 1017 |
+
"loss": 0.037,
|
| 1018 |
+
"step": 1680
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"grad_norm": 0.640484631061554,
|
| 1022 |
+
"learning_rate": 9.998914929556569e-05,
|
| 1023 |
+
"loss": 0.0415,
|
| 1024 |
+
"step": 1690
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"grad_norm": 0.5324658155441284,
|
| 1028 |
+
"learning_rate": 9.998797074387361e-05,
|
| 1029 |
+
"loss": 0.0406,
|
| 1030 |
+
"step": 1700
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"grad_norm": 0.4396641254425049,
|
| 1034 |
+
"learning_rate": 9.99867314520835e-05,
|
| 1035 |
+
"loss": 0.0358,
|
| 1036 |
+
"step": 1710
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"grad_norm": 0.5642094016075134,
|
| 1040 |
+
"learning_rate": 9.998543142170126e-05,
|
| 1041 |
+
"loss": 0.0355,
|
| 1042 |
+
"step": 1720
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"grad_norm": 0.4254641830921173,
|
| 1046 |
+
"learning_rate": 9.99840706543065e-05,
|
| 1047 |
+
"loss": 0.0347,
|
| 1048 |
+
"step": 1730
|
| 1049 |
+
},
|
| 1050 |
+
{
|
| 1051 |
+
"grad_norm": 0.6551932096481323,
|
| 1052 |
+
"learning_rate": 9.998264915155274e-05,
|
| 1053 |
+
"loss": 0.0331,
|
| 1054 |
+
"step": 1740
|
| 1055 |
+
},
|
| 1056 |
+
{
|
| 1057 |
+
"grad_norm": 0.6991862654685974,
|
| 1058 |
+
"learning_rate": 9.998116691516718e-05,
|
| 1059 |
+
"loss": 0.0369,
|
| 1060 |
+
"step": 1750
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"grad_norm": 0.5255552530288696,
|
| 1064 |
+
"learning_rate": 9.997962394695091e-05,
|
| 1065 |
+
"loss": 0.0357,
|
| 1066 |
+
"step": 1760
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"grad_norm": 0.5381750464439392,
|
| 1070 |
+
"learning_rate": 9.997802024877875e-05,
|
| 1071 |
+
"loss": 0.0369,
|
| 1072 |
+
"step": 1770
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"grad_norm": 0.46699705719947815,
|
| 1076 |
+
"learning_rate": 9.99763558225994e-05,
|
| 1077 |
+
"loss": 0.0332,
|
| 1078 |
+
"step": 1780
|
| 1079 |
+
},
|
| 1080 |
+
{
|
| 1081 |
+
"grad_norm": 0.5546779632568359,
|
| 1082 |
+
"learning_rate": 9.997463067043526e-05,
|
| 1083 |
+
"loss": 0.035,
|
| 1084 |
+
"step": 1790
|
| 1085 |
+
},
|
| 1086 |
+
{
|
| 1087 |
+
"grad_norm": 0.5206839442253113,
|
| 1088 |
+
"learning_rate": 9.997284479438253e-05,
|
| 1089 |
+
"loss": 0.0332,
|
| 1090 |
+
"step": 1800
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"grad_norm": 0.5379693508148193,
|
| 1094 |
+
"learning_rate": 9.997099819661127e-05,
|
| 1095 |
+
"loss": 0.0309,
|
| 1096 |
+
"step": 1810
|
| 1097 |
+
},
|
| 1098 |
+
{
|
| 1099 |
+
"grad_norm": 0.4955759346485138,
|
| 1100 |
+
"learning_rate": 9.996909087936524e-05,
|
| 1101 |
+
"loss": 0.0353,
|
| 1102 |
+
"step": 1820
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"grad_norm": 0.42540910840034485,
|
| 1106 |
+
"learning_rate": 9.996712284496201e-05,
|
| 1107 |
+
"loss": 0.0327,
|
| 1108 |
+
"step": 1830
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"grad_norm": 0.3907061219215393,
|
| 1112 |
+
"learning_rate": 9.996509409579293e-05,
|
| 1113 |
+
"loss": 0.0323,
|
| 1114 |
+
"step": 1840
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"grad_norm": 0.6000369787216187,
|
| 1118 |
+
"learning_rate": 9.996300463432312e-05,
|
| 1119 |
+
"loss": 0.0336,
|
| 1120 |
+
"step": 1850
|
| 1121 |
+
},
|
| 1122 |
+
{
|
| 1123 |
+
"grad_norm": 0.5673982501029968,
|
| 1124 |
+
"learning_rate": 9.996085446309148e-05,
|
| 1125 |
+
"loss": 0.0369,
|
| 1126 |
+
"step": 1860
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"grad_norm": 0.49209660291671753,
|
| 1130 |
+
"learning_rate": 9.995864358471066e-05,
|
| 1131 |
+
"loss": 0.0341,
|
| 1132 |
+
"step": 1870
|
| 1133 |
+
},
|
| 1134 |
+
{
|
| 1135 |
+
"grad_norm": 0.7095867991447449,
|
| 1136 |
+
"learning_rate": 9.99563720018671e-05,
|
| 1137 |
+
"loss": 0.0316,
|
| 1138 |
+
"step": 1880
|
| 1139 |
+
},
|
| 1140 |
+
{
|
| 1141 |
+
"grad_norm": 0.6636500358581543,
|
| 1142 |
+
"learning_rate": 9.995403971732098e-05,
|
| 1143 |
+
"loss": 0.035,
|
| 1144 |
+
"step": 1890
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"grad_norm": 0.4743417799472809,
|
| 1148 |
+
"learning_rate": 9.995164673390625e-05,
|
| 1149 |
+
"loss": 0.0311,
|
| 1150 |
+
"step": 1900
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"grad_norm": 0.5287648439407349,
|
| 1154 |
+
"learning_rate": 9.994919305453059e-05,
|
| 1155 |
+
"loss": 0.031,
|
| 1156 |
+
"step": 1910
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"grad_norm": 0.45866692066192627,
|
| 1160 |
+
"learning_rate": 9.994667868217548e-05,
|
| 1161 |
+
"loss": 0.032,
|
| 1162 |
+
"step": 1920
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"grad_norm": 0.5198220610618591,
|
| 1166 |
+
"learning_rate": 9.99441036198961e-05,
|
| 1167 |
+
"loss": 0.0326,
|
| 1168 |
+
"step": 1930
|
| 1169 |
+
},
|
| 1170 |
+
{
|
| 1171 |
+
"grad_norm": 0.5405005812644958,
|
| 1172 |
+
"learning_rate": 9.99414678708214e-05,
|
| 1173 |
+
"loss": 0.0322,
|
| 1174 |
+
"step": 1940
|
| 1175 |
+
},
|
| 1176 |
+
{
|
| 1177 |
+
"grad_norm": 0.42117366194725037,
|
| 1178 |
+
"learning_rate": 9.993877143815407e-05,
|
| 1179 |
+
"loss": 0.0314,
|
| 1180 |
+
"step": 1950
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"grad_norm": 0.4138369858264923,
|
| 1184 |
+
"learning_rate": 9.993601432517053e-05,
|
| 1185 |
+
"loss": 0.0311,
|
| 1186 |
+
"step": 1960
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"grad_norm": 0.5241675972938538,
|
| 1190 |
+
"learning_rate": 9.993319653522091e-05,
|
| 1191 |
+
"loss": 0.0322,
|
| 1192 |
+
"step": 1970
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"grad_norm": 0.48230016231536865,
|
| 1196 |
+
"learning_rate": 9.993031807172911e-05,
|
| 1197 |
+
"loss": 0.0321,
|
| 1198 |
+
"step": 1980
|
| 1199 |
+
},
|
| 1200 |
+
{
|
| 1201 |
+
"grad_norm": 0.4062332808971405,
|
| 1202 |
+
"learning_rate": 9.992737893819273e-05,
|
| 1203 |
+
"loss": 0.0317,
|
| 1204 |
+
"step": 1990
|
| 1205 |
+
},
|
| 1206 |
+
{
|
| 1207 |
+
"grad_norm": 0.5321586728096008,
|
| 1208 |
+
"learning_rate": 9.992437913818312e-05,
|
| 1209 |
+
"loss": 0.0307,
|
| 1210 |
+
"step": 2000
|
| 1211 |
+
},
|
| 1212 |
+
{
|
| 1213 |
+
"grad_norm": 0.4237384498119354,
|
| 1214 |
+
"learning_rate": 9.992131867534526e-05,
|
| 1215 |
+
"loss": 0.0288,
|
| 1216 |
+
"step": 2010
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"grad_norm": 0.5707708597183228,
|
| 1220 |
+
"learning_rate": 9.991819755339796e-05,
|
| 1221 |
+
"loss": 0.0318,
|
| 1222 |
+
"step": 2020
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"grad_norm": 0.3146398961544037,
|
| 1226 |
+
"learning_rate": 9.991501577613365e-05,
|
| 1227 |
+
"loss": 0.0312,
|
| 1228 |
+
"step": 2030
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"grad_norm": 0.5046992897987366,
|
| 1232 |
+
"learning_rate": 9.99117733474185e-05,
|
| 1233 |
+
"loss": 0.0305,
|
| 1234 |
+
"step": 2040
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"grad_norm": 0.4503960609436035,
|
| 1238 |
+
"learning_rate": 9.990847027119234e-05,
|
| 1239 |
+
"loss": 0.032,
|
| 1240 |
+
"step": 2050
|
| 1241 |
+
},
|
| 1242 |
+
{
|
| 1243 |
+
"grad_norm": 0.6307269930839539,
|
| 1244 |
+
"learning_rate": 9.990510655146877e-05,
|
| 1245 |
+
"loss": 0.0314,
|
| 1246 |
+
"step": 2060
|
| 1247 |
+
},
|
| 1248 |
+
{
|
| 1249 |
+
"grad_norm": 0.5946677327156067,
|
| 1250 |
+
"learning_rate": 9.990168219233496e-05,
|
| 1251 |
+
"loss": 0.031,
|
| 1252 |
+
"step": 2070
|
| 1253 |
+
},
|
| 1254 |
+
{
|
| 1255 |
+
"grad_norm": 0.478351891040802,
|
| 1256 |
+
"learning_rate": 9.989819719795188e-05,
|
| 1257 |
+
"loss": 0.0289,
|
| 1258 |
+
"step": 2080
|
| 1259 |
+
},
|
| 1260 |
+
{
|
| 1261 |
+
"grad_norm": 0.5508173108100891,
|
| 1262 |
+
"learning_rate": 9.989465157255412e-05,
|
| 1263 |
+
"loss": 0.0334,
|
| 1264 |
+
"step": 2090
|
| 1265 |
+
},
|
| 1266 |
+
{
|
| 1267 |
+
"grad_norm": 0.4113427996635437,
|
| 1268 |
+
"learning_rate": 9.989104532044994e-05,
|
| 1269 |
+
"loss": 0.0293,
|
| 1270 |
+
"step": 2100
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"grad_norm": 0.47436055541038513,
|
| 1274 |
+
"learning_rate": 9.988737844602128e-05,
|
| 1275 |
+
"loss": 0.0264,
|
| 1276 |
+
"step": 2110
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"grad_norm": 0.6576429009437561,
|
| 1280 |
+
"learning_rate": 9.988365095372372e-05,
|
| 1281 |
+
"loss": 0.029,
|
| 1282 |
+
"step": 2120
|
| 1283 |
+
},
|
| 1284 |
+
{
|
| 1285 |
+
"grad_norm": 0.4389994740486145,
|
| 1286 |
+
"learning_rate": 9.987986284808654e-05,
|
| 1287 |
+
"loss": 0.0313,
|
| 1288 |
+
"step": 2130
|
| 1289 |
+
},
|
| 1290 |
+
{
|
| 1291 |
+
"grad_norm": 0.5478070378303528,
|
| 1292 |
+
"learning_rate": 9.987601413371264e-05,
|
| 1293 |
+
"loss": 0.0303,
|
| 1294 |
+
"step": 2140
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"grad_norm": 0.4985005557537079,
|
| 1298 |
+
"learning_rate": 9.987210481527855e-05,
|
| 1299 |
+
"loss": 0.0284,
|
| 1300 |
+
"step": 2150
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"grad_norm": 0.4611583650112152,
|
| 1304 |
+
"learning_rate": 9.98681348975345e-05,
|
| 1305 |
+
"loss": 0.0283,
|
| 1306 |
+
"step": 2160
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"grad_norm": 0.43860509991645813,
|
| 1310 |
+
"learning_rate": 9.986410438530427e-05,
|
| 1311 |
+
"loss": 0.0269,
|
| 1312 |
+
"step": 2170
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"grad_norm": 0.541072428226471,
|
| 1316 |
+
"learning_rate": 9.986001328348534e-05,
|
| 1317 |
+
"loss": 0.0273,
|
| 1318 |
+
"step": 2180
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"grad_norm": 0.4160540997982025,
|
| 1322 |
+
"learning_rate": 9.985586159704878e-05,
|
| 1323 |
+
"loss": 0.0267,
|
| 1324 |
+
"step": 2190
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"grad_norm": 0.5684108138084412,
|
| 1328 |
+
"learning_rate": 9.985164933103929e-05,
|
| 1329 |
+
"loss": 0.0331,
|
| 1330 |
+
"step": 2200
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"grad_norm": 0.60076904296875,
|
| 1334 |
+
"learning_rate": 9.984737649057513e-05,
|
| 1335 |
+
"loss": 0.0315,
|
| 1336 |
+
"step": 2210
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"grad_norm": 0.4548479914665222,
|
| 1340 |
+
"learning_rate": 9.984304308084827e-05,
|
| 1341 |
+
"loss": 0.0332,
|
| 1342 |
+
"step": 2220
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"grad_norm": 0.32628098130226135,
|
| 1346 |
+
"learning_rate": 9.983864910712416e-05,
|
| 1347 |
+
"loss": 0.0277,
|
| 1348 |
+
"step": 2230
|
| 1349 |
+
},
|
| 1350 |
+
{
|
| 1351 |
+
"grad_norm": 0.4243181645870209,
|
| 1352 |
+
"learning_rate": 9.98341945747419e-05,
|
| 1353 |
+
"loss": 0.0271,
|
| 1354 |
+
"step": 2240
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"grad_norm": 0.4711628258228302,
|
| 1358 |
+
"learning_rate": 9.98296794891142e-05,
|
| 1359 |
+
"loss": 0.0283,
|
| 1360 |
+
"step": 2250
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"grad_norm": 0.3072936534881592,
|
| 1364 |
+
"learning_rate": 9.982510385572725e-05,
|
| 1365 |
+
"loss": 0.0294,
|
| 1366 |
+
"step": 2260
|
| 1367 |
+
},
|
| 1368 |
+
{
|
| 1369 |
+
"grad_norm": 0.37349826097488403,
|
| 1370 |
+
"learning_rate": 9.982046768014094e-05,
|
| 1371 |
+
"loss": 0.0271,
|
| 1372 |
+
"step": 2270
|
| 1373 |
+
},
|
| 1374 |
+
{
|
| 1375 |
+
"grad_norm": 0.41922029852867126,
|
| 1376 |
+
"learning_rate": 9.981577096798863e-05,
|
| 1377 |
+
"loss": 0.0255,
|
| 1378 |
+
"step": 2280
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"grad_norm": 0.41333553194999695,
|
| 1382 |
+
"learning_rate": 9.981101372497727e-05,
|
| 1383 |
+
"loss": 0.0276,
|
| 1384 |
+
"step": 2290
|
| 1385 |
+
},
|
| 1386 |
+
{
|
| 1387 |
+
"grad_norm": 0.49039316177368164,
|
| 1388 |
+
"learning_rate": 9.980619595688737e-05,
|
| 1389 |
+
"loss": 0.0264,
|
| 1390 |
+
"step": 2300
|
| 1391 |
+
},
|
| 1392 |
+
{
|
| 1393 |
+
"grad_norm": 0.45840826630592346,
|
| 1394 |
+
"learning_rate": 9.980131766957295e-05,
|
| 1395 |
+
"loss": 0.0289,
|
| 1396 |
+
"step": 2310
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"grad_norm": 0.43983516097068787,
|
| 1400 |
+
"learning_rate": 9.979637886896163e-05,
|
| 1401 |
+
"loss": 0.0295,
|
| 1402 |
+
"step": 2320
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"grad_norm": 0.519990861415863,
|
| 1406 |
+
"learning_rate": 9.979137956105447e-05,
|
| 1407 |
+
"loss": 0.0351,
|
| 1408 |
+
"step": 2330
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"grad_norm": 0.37182801961898804,
|
| 1412 |
+
"learning_rate": 9.978631975192613e-05,
|
| 1413 |
+
"loss": 0.0272,
|
| 1414 |
+
"step": 2340
|
| 1415 |
+
},
|
| 1416 |
+
{
|
| 1417 |
+
"grad_norm": 0.3941374123096466,
|
| 1418 |
+
"learning_rate": 9.978119944772475e-05,
|
| 1419 |
+
"loss": 0.0338,
|
| 1420 |
+
"step": 2350
|
| 1421 |
+
},
|
| 1422 |
+
{
|
| 1423 |
+
"grad_norm": 0.49252980947494507,
|
| 1424 |
+
"learning_rate": 9.977601865467197e-05,
|
| 1425 |
+
"loss": 0.029,
|
| 1426 |
+
"step": 2360
|
| 1427 |
+
},
|
| 1428 |
+
{
|
| 1429 |
+
"grad_norm": 0.36538147926330566,
|
| 1430 |
+
"learning_rate": 9.977077737906297e-05,
|
| 1431 |
+
"loss": 0.0271,
|
| 1432 |
+
"step": 2370
|
| 1433 |
+
},
|
| 1434 |
+
{
|
| 1435 |
+
"grad_norm": 0.48212969303131104,
|
| 1436 |
+
"learning_rate": 9.976547562726636e-05,
|
| 1437 |
+
"loss": 0.0286,
|
| 1438 |
+
"step": 2380
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"grad_norm": 0.3857809603214264,
|
| 1442 |
+
"learning_rate": 9.976011340572429e-05,
|
| 1443 |
+
"loss": 0.0299,
|
| 1444 |
+
"step": 2390
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"grad_norm": 0.3532734215259552,
|
| 1448 |
+
"learning_rate": 9.975469072095237e-05,
|
| 1449 |
+
"loss": 0.029,
|
| 1450 |
+
"step": 2400
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"grad_norm": 0.3491647243499756,
|
| 1454 |
+
"learning_rate": 9.974920757953965e-05,
|
| 1455 |
+
"loss": 0.0246,
|
| 1456 |
+
"step": 2410
|
| 1457 |
+
},
|
| 1458 |
+
{
|
| 1459 |
+
"grad_norm": 0.40819358825683594,
|
| 1460 |
+
"learning_rate": 9.97436639881487e-05,
|
| 1461 |
+
"loss": 0.0322,
|
| 1462 |
+
"step": 2420
|
| 1463 |
+
},
|
| 1464 |
+
{
|
| 1465 |
+
"grad_norm": 0.4187079071998596,
|
| 1466 |
+
"learning_rate": 9.973805995351545e-05,
|
| 1467 |
+
"loss": 0.0284,
|
| 1468 |
+
"step": 2430
|
| 1469 |
+
},
|
| 1470 |
+
{
|
| 1471 |
+
"grad_norm": 0.47072187066078186,
|
| 1472 |
+
"learning_rate": 9.973239548244939e-05,
|
| 1473 |
+
"loss": 0.0287,
|
| 1474 |
+
"step": 2440
|
| 1475 |
+
},
|
| 1476 |
+
{
|
| 1477 |
+
"grad_norm": 0.4486375153064728,
|
| 1478 |
+
"learning_rate": 9.972667058183333e-05,
|
| 1479 |
+
"loss": 0.0274,
|
| 1480 |
+
"step": 2450
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"grad_norm": 0.387708455324173,
|
| 1484 |
+
"learning_rate": 9.972088525862362e-05,
|
| 1485 |
+
"loss": 0.0246,
|
| 1486 |
+
"step": 2460
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"grad_norm": 0.45445823669433594,
|
| 1490 |
+
"learning_rate": 9.971503951984995e-05,
|
| 1491 |
+
"loss": 0.0262,
|
| 1492 |
+
"step": 2470
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"grad_norm": 0.5244600176811218,
|
| 1496 |
+
"learning_rate": 9.970913337261543e-05,
|
| 1497 |
+
"loss": 0.026,
|
| 1498 |
+
"step": 2480
|
| 1499 |
+
},
|
| 1500 |
+
{
|
| 1501 |
+
"grad_norm": 0.33774614334106445,
|
| 1502 |
+
"learning_rate": 9.97031668240966e-05,
|
| 1503 |
+
"loss": 0.0327,
|
| 1504 |
+
"step": 2490
|
| 1505 |
+
},
|
| 1506 |
+
{
|
| 1507 |
+
"grad_norm": 0.45767754316329956,
|
| 1508 |
+
"learning_rate": 9.969713988154339e-05,
|
| 1509 |
+
"loss": 0.0278,
|
| 1510 |
+
"step": 2500
|
| 1511 |
+
},
|
| 1512 |
+
{
|
| 1513 |
+
"grad_norm": 0.452639639377594,
|
| 1514 |
+
"learning_rate": 9.969105255227906e-05,
|
| 1515 |
+
"loss": 0.0281,
|
| 1516 |
+
"step": 2510
|
| 1517 |
+
},
|
| 1518 |
+
{
|
| 1519 |
+
"grad_norm": 0.4824736416339874,
|
| 1520 |
+
"learning_rate": 9.968490484370035e-05,
|
| 1521 |
+
"loss": 0.0268,
|
| 1522 |
+
"step": 2520
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"grad_norm": 0.42104530334472656,
|
| 1526 |
+
"learning_rate": 9.967869676327726e-05,
|
| 1527 |
+
"loss": 0.0277,
|
| 1528 |
+
"step": 2530
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"grad_norm": 0.3387914299964905,
|
| 1532 |
+
"learning_rate": 9.96724283185532e-05,
|
| 1533 |
+
"loss": 0.0256,
|
| 1534 |
+
"step": 2540
|
| 1535 |
+
},
|
| 1536 |
+
{
|
| 1537 |
+
"grad_norm": 0.46041038632392883,
|
| 1538 |
+
"learning_rate": 9.966609951714494e-05,
|
| 1539 |
+
"loss": 0.0283,
|
| 1540 |
+
"step": 2550
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"grad_norm": 0.482858806848526,
|
| 1544 |
+
"learning_rate": 9.965971036674255e-05,
|
| 1545 |
+
"loss": 0.0302,
|
| 1546 |
+
"step": 2560
|
| 1547 |
+
},
|
| 1548 |
+
{
|
| 1549 |
+
"grad_norm": 0.4021144509315491,
|
| 1550 |
+
"learning_rate": 9.965326087510947e-05,
|
| 1551 |
+
"loss": 0.025,
|
| 1552 |
+
"step": 2570
|
| 1553 |
+
},
|
| 1554 |
+
{
|
| 1555 |
+
"grad_norm": 0.4823286533355713,
|
| 1556 |
+
"learning_rate": 9.964675105008244e-05,
|
| 1557 |
+
"loss": 0.0258,
|
| 1558 |
+
"step": 2580
|
| 1559 |
+
},
|
| 1560 |
+
{
|
| 1561 |
+
"grad_norm": 0.4102964401245117,
|
| 1562 |
+
"learning_rate": 9.964018089957147e-05,
|
| 1563 |
+
"loss": 0.0236,
|
| 1564 |
+
"step": 2590
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"grad_norm": 0.4453105330467224,
|
| 1568 |
+
"learning_rate": 9.963355043155996e-05,
|
| 1569 |
+
"loss": 0.0293,
|
| 1570 |
+
"step": 2600
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"grad_norm": 0.42528659105300903,
|
| 1574 |
+
"learning_rate": 9.962685965410455e-05,
|
| 1575 |
+
"loss": 0.0284,
|
| 1576 |
+
"step": 2610
|
| 1577 |
+
},
|
| 1578 |
+
{
|
| 1579 |
+
"grad_norm": 0.4291960597038269,
|
| 1580 |
+
"learning_rate": 9.962010857533514e-05,
|
| 1581 |
+
"loss": 0.0271,
|
| 1582 |
+
"step": 2620
|
| 1583 |
+
},
|
| 1584 |
+
{
|
| 1585 |
+
"grad_norm": 0.38245514035224915,
|
| 1586 |
+
"learning_rate": 9.961329720345493e-05,
|
| 1587 |
+
"loss": 0.0258,
|
| 1588 |
+
"step": 2630
|
| 1589 |
+
},
|
| 1590 |
+
{
|
| 1591 |
+
"grad_norm": 0.5179672241210938,
|
| 1592 |
+
"learning_rate": 9.96064255467404e-05,
|
| 1593 |
+
"loss": 0.0281,
|
| 1594 |
+
"step": 2640
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"grad_norm": 0.38057994842529297,
|
| 1598 |
+
"learning_rate": 9.959949361354126e-05,
|
| 1599 |
+
"loss": 0.0281,
|
| 1600 |
+
"step": 2650
|
| 1601 |
+
},
|
| 1602 |
+
{
|
| 1603 |
+
"grad_norm": 0.42559677362442017,
|
| 1604 |
+
"learning_rate": 9.959250141228045e-05,
|
| 1605 |
+
"loss": 0.0262,
|
| 1606 |
+
"step": 2660
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"grad_norm": 0.3978665769100189,
|
| 1610 |
+
"learning_rate": 9.958544895145414e-05,
|
| 1611 |
+
"loss": 0.0265,
|
| 1612 |
+
"step": 2670
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"grad_norm": 0.42062318325042725,
|
| 1616 |
+
"learning_rate": 9.957833623963177e-05,
|
| 1617 |
+
"loss": 0.0272,
|
| 1618 |
+
"step": 2680
|
| 1619 |
+
},
|
| 1620 |
+
{
|
| 1621 |
+
"grad_norm": 0.36069512367248535,
|
| 1622 |
+
"learning_rate": 9.957116328545593e-05,
|
| 1623 |
+
"loss": 0.0279,
|
| 1624 |
+
"step": 2690
|
| 1625 |
+
},
|
| 1626 |
+
{
|
| 1627 |
+
"grad_norm": 0.3319280743598938,
|
| 1628 |
+
"learning_rate": 9.956393009764244e-05,
|
| 1629 |
+
"loss": 0.0244,
|
| 1630 |
+
"step": 2700
|
| 1631 |
+
},
|
| 1632 |
+
{
|
| 1633 |
+
"grad_norm": 0.38565364480018616,
|
| 1634 |
+
"learning_rate": 9.955663668498032e-05,
|
| 1635 |
+
"loss": 0.0242,
|
| 1636 |
+
"step": 2710
|
| 1637 |
+
},
|
| 1638 |
+
{
|
| 1639 |
+
"grad_norm": 0.44830113649368286,
|
| 1640 |
+
"learning_rate": 9.954928305633173e-05,
|
| 1641 |
+
"loss": 0.025,
|
| 1642 |
+
"step": 2720
|
| 1643 |
+
},
|
| 1644 |
+
{
|
| 1645 |
+
"grad_norm": 0.3731986880302429,
|
| 1646 |
+
"learning_rate": 9.954186922063204e-05,
|
| 1647 |
+
"loss": 0.0254,
|
| 1648 |
+
"step": 2730
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"grad_norm": 0.36890801787376404,
|
| 1652 |
+
"learning_rate": 9.953439518688974e-05,
|
| 1653 |
+
"loss": 0.0242,
|
| 1654 |
+
"step": 2740
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"grad_norm": 0.47049516439437866,
|
| 1658 |
+
"learning_rate": 9.952686096418652e-05,
|
| 1659 |
+
"loss": 0.0223,
|
| 1660 |
+
"step": 2750
|
| 1661 |
+
},
|
| 1662 |
+
{
|
| 1663 |
+
"grad_norm": 0.2939774990081787,
|
| 1664 |
+
"learning_rate": 9.951926656167714e-05,
|
| 1665 |
+
"loss": 0.023,
|
| 1666 |
+
"step": 2760
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"grad_norm": 0.4406392574310303,
|
| 1670 |
+
"learning_rate": 9.951161198858953e-05,
|
| 1671 |
+
"loss": 0.0264,
|
| 1672 |
+
"step": 2770
|
| 1673 |
+
},
|
| 1674 |
+
{
|
| 1675 |
+
"grad_norm": 0.41060757637023926,
|
| 1676 |
+
"learning_rate": 9.950389725422471e-05,
|
| 1677 |
+
"loss": 0.0226,
|
| 1678 |
+
"step": 2780
|
| 1679 |
+
},
|
| 1680 |
+
{
|
| 1681 |
+
"grad_norm": 0.3619372844696045,
|
| 1682 |
+
"learning_rate": 9.949612236795682e-05,
|
| 1683 |
+
"loss": 0.0233,
|
| 1684 |
+
"step": 2790
|
| 1685 |
+
},
|
| 1686 |
+
{
|
| 1687 |
+
"grad_norm": 0.36144009232521057,
|
| 1688 |
+
"learning_rate": 9.948828733923305e-05,
|
| 1689 |
+
"loss": 0.0252,
|
| 1690 |
+
"step": 2800
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"grad_norm": 0.4079873561859131,
|
| 1694 |
+
"learning_rate": 9.948039217757374e-05,
|
| 1695 |
+
"loss": 0.0262,
|
| 1696 |
+
"step": 2810
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"grad_norm": 0.47260719537734985,
|
| 1700 |
+
"learning_rate": 9.947243689257225e-05,
|
| 1701 |
+
"loss": 0.0272,
|
| 1702 |
+
"step": 2820
|
| 1703 |
+
},
|
| 1704 |
+
{
|
| 1705 |
+
"grad_norm": 0.3860573172569275,
|
| 1706 |
+
"learning_rate": 9.946442149389497e-05,
|
| 1707 |
+
"loss": 0.0241,
|
| 1708 |
+
"step": 2830
|
| 1709 |
+
},
|
| 1710 |
+
{
|
| 1711 |
+
"grad_norm": 0.3537745773792267,
|
| 1712 |
+
"learning_rate": 9.945634599128139e-05,
|
| 1713 |
+
"loss": 0.0239,
|
| 1714 |
+
"step": 2840
|
| 1715 |
+
},
|
| 1716 |
+
{
|
| 1717 |
+
"grad_norm": 0.37954181432724,
|
| 1718 |
+
"learning_rate": 9.944821039454402e-05,
|
| 1719 |
+
"loss": 0.0247,
|
| 1720 |
+
"step": 2850
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"grad_norm": 0.46963128447532654,
|
| 1724 |
+
"learning_rate": 9.944001471356835e-05,
|
| 1725 |
+
"loss": 0.0287,
|
| 1726 |
+
"step": 2860
|
| 1727 |
+
},
|
| 1728 |
+
{
|
| 1729 |
+
"grad_norm": 0.37109649181365967,
|
| 1730 |
+
"learning_rate": 9.94317589583129e-05,
|
| 1731 |
+
"loss": 0.0281,
|
| 1732 |
+
"step": 2870
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"grad_norm": 0.2578738033771515,
|
| 1736 |
+
"learning_rate": 9.942344313880922e-05,
|
| 1737 |
+
"loss": 0.0236,
|
| 1738 |
+
"step": 2880
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"grad_norm": 0.33481988310813904,
|
| 1742 |
+
"learning_rate": 9.941506726516179e-05,
|
| 1743 |
+
"loss": 0.0246,
|
| 1744 |
+
"step": 2890
|
| 1745 |
+
},
|
| 1746 |
+
{
|
| 1747 |
+
"grad_norm": 0.2952929437160492,
|
| 1748 |
+
"learning_rate": 9.94066313475481e-05,
|
| 1749 |
+
"loss": 0.024,
|
| 1750 |
+
"step": 2900
|
| 1751 |
+
},
|
| 1752 |
+
{
|
| 1753 |
+
"grad_norm": 0.33066704869270325,
|
| 1754 |
+
"learning_rate": 9.939813539621857e-05,
|
| 1755 |
+
"loss": 0.0227,
|
| 1756 |
+
"step": 2910
|
| 1757 |
+
},
|
| 1758 |
+
{
|
| 1759 |
+
"grad_norm": 0.4288365840911865,
|
| 1760 |
+
"learning_rate": 9.93895794214966e-05,
|
| 1761 |
+
"loss": 0.0268,
|
| 1762 |
+
"step": 2920
|
| 1763 |
+
},
|
| 1764 |
+
{
|
| 1765 |
+
"grad_norm": 0.37286025285720825,
|
| 1766 |
+
"learning_rate": 9.938096343377852e-05,
|
| 1767 |
+
"loss": 0.0284,
|
| 1768 |
+
"step": 2930
|
| 1769 |
+
},
|
| 1770 |
+
{
|
| 1771 |
+
"grad_norm": 0.35089603066444397,
|
| 1772 |
+
"learning_rate": 9.937228744353353e-05,
|
| 1773 |
+
"loss": 0.0285,
|
| 1774 |
+
"step": 2940
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"grad_norm": 0.49157941341400146,
|
| 1778 |
+
"learning_rate": 9.936355146130379e-05,
|
| 1779 |
+
"loss": 0.0254,
|
| 1780 |
+
"step": 2950
|
| 1781 |
+
},
|
| 1782 |
+
{
|
| 1783 |
+
"grad_norm": 0.5023393630981445,
|
| 1784 |
+
"learning_rate": 9.935475549770435e-05,
|
| 1785 |
+
"loss": 0.0258,
|
| 1786 |
+
"step": 2960
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"grad_norm": 0.4146648645401001,
|
| 1790 |
+
"learning_rate": 9.934589956342315e-05,
|
| 1791 |
+
"loss": 0.0244,
|
| 1792 |
+
"step": 2970
|
| 1793 |
+
},
|
| 1794 |
+
{
|
| 1795 |
+
"grad_norm": 0.4460250735282898,
|
| 1796 |
+
"learning_rate": 9.933698366922093e-05,
|
| 1797 |
+
"loss": 0.0237,
|
| 1798 |
+
"step": 2980
|
| 1799 |
+
},
|
| 1800 |
+
{
|
| 1801 |
+
"grad_norm": 0.44778910279273987,
|
| 1802 |
+
"learning_rate": 9.93280078259314e-05,
|
| 1803 |
+
"loss": 0.025,
|
| 1804 |
+
"step": 2990
|
| 1805 |
+
},
|
| 1806 |
+
{
|
| 1807 |
+
"grad_norm": 0.3062107563018799,
|
| 1808 |
+
"learning_rate": 9.931897204446104e-05,
|
| 1809 |
+
"loss": 0.0241,
|
| 1810 |
+
"step": 3000
|
| 1811 |
+
},
|
| 1812 |
+
{
|
| 1813 |
+
"grad_norm": 0.3562763035297394,
|
| 1814 |
+
"learning_rate": 9.930987633578915e-05,
|
| 1815 |
+
"loss": 0.0235,
|
| 1816 |
+
"step": 3010
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"grad_norm": 0.37201783061027527,
|
| 1820 |
+
"learning_rate": 9.93007207109679e-05,
|
| 1821 |
+
"loss": 0.0231,
|
| 1822 |
+
"step": 3020
|
| 1823 |
+
},
|
| 1824 |
+
{
|
| 1825 |
+
"grad_norm": 0.3234574496746063,
|
| 1826 |
+
"learning_rate": 9.929150518112224e-05,
|
| 1827 |
+
"loss": 0.0215,
|
| 1828 |
+
"step": 3030
|
| 1829 |
+
},
|
| 1830 |
+
{
|
| 1831 |
+
"grad_norm": 0.4681793451309204,
|
| 1832 |
+
"learning_rate": 9.928222975744991e-05,
|
| 1833 |
+
"loss": 0.0244,
|
| 1834 |
+
"step": 3040
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"grad_norm": 0.4070403277873993,
|
| 1838 |
+
"learning_rate": 9.92728944512214e-05,
|
| 1839 |
+
"loss": 0.0225,
|
| 1840 |
+
"step": 3050
|
| 1841 |
+
},
|
| 1842 |
+
{
|
| 1843 |
+
"grad_norm": 0.44380709528923035,
|
| 1844 |
+
"learning_rate": 9.926349927378001e-05,
|
| 1845 |
+
"loss": 0.0222,
|
| 1846 |
+
"step": 3060
|
| 1847 |
+
},
|
| 1848 |
+
{
|
| 1849 |
+
"grad_norm": 0.43774229288101196,
|
| 1850 |
+
"learning_rate": 9.925404423654174e-05,
|
| 1851 |
+
"loss": 0.023,
|
| 1852 |
+
"step": 3070
|
| 1853 |
+
},
|
| 1854 |
+
{
|
| 1855 |
+
"grad_norm": 0.35820233821868896,
|
| 1856 |
+
"learning_rate": 9.924452935099537e-05,
|
| 1857 |
+
"loss": 0.0238,
|
| 1858 |
+
"step": 3080
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"grad_norm": 0.29745739698410034,
|
| 1862 |
+
"learning_rate": 9.92349546287024e-05,
|
| 1863 |
+
"loss": 0.024,
|
| 1864 |
+
"step": 3090
|
| 1865 |
+
},
|
| 1866 |
+
{
|
| 1867 |
+
"grad_norm": 0.46426740288734436,
|
| 1868 |
+
"learning_rate": 9.9225320081297e-05,
|
| 1869 |
+
"loss": 0.0245,
|
| 1870 |
+
"step": 3100
|
| 1871 |
+
},
|
| 1872 |
+
{
|
| 1873 |
+
"grad_norm": 0.3944103419780731,
|
| 1874 |
+
"learning_rate": 9.921562572048606e-05,
|
| 1875 |
+
"loss": 0.0238,
|
| 1876 |
+
"step": 3110
|
| 1877 |
+
},
|
| 1878 |
+
{
|
| 1879 |
+
"grad_norm": 0.35884836316108704,
|
| 1880 |
+
"learning_rate": 9.920587155804913e-05,
|
| 1881 |
+
"loss": 0.0247,
|
| 1882 |
+
"step": 3120
|
| 1883 |
+
},
|
| 1884 |
+
{
|
| 1885 |
+
"grad_norm": 0.3041065037250519,
|
| 1886 |
+
"learning_rate": 9.919605760583845e-05,
|
| 1887 |
+
"loss": 0.0246,
|
| 1888 |
+
"step": 3130
|
| 1889 |
+
},
|
| 1890 |
+
{
|
| 1891 |
+
"grad_norm": 0.263491690158844,
|
| 1892 |
+
"learning_rate": 9.91861838757789e-05,
|
| 1893 |
+
"loss": 0.0227,
|
| 1894 |
+
"step": 3140
|
| 1895 |
+
},
|
| 1896 |
+
{
|
| 1897 |
+
"grad_norm": 0.24135278165340424,
|
| 1898 |
+
"learning_rate": 9.917625037986798e-05,
|
| 1899 |
+
"loss": 0.0231,
|
| 1900 |
+
"step": 3150
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"grad_norm": 0.42098352313041687,
|
| 1904 |
+
"learning_rate": 9.916625713017583e-05,
|
| 1905 |
+
"loss": 0.021,
|
| 1906 |
+
"step": 3160
|
| 1907 |
+
},
|
| 1908 |
+
{
|
| 1909 |
+
"grad_norm": 0.3001592457294464,
|
| 1910 |
+
"learning_rate": 9.915620413884519e-05,
|
| 1911 |
+
"loss": 0.0208,
|
| 1912 |
+
"step": 3170
|
| 1913 |
+
},
|
| 1914 |
+
{
|
| 1915 |
+
"grad_norm": 0.3589301109313965,
|
| 1916 |
+
"learning_rate": 9.914609141809139e-05,
|
| 1917 |
+
"loss": 0.021,
|
| 1918 |
+
"step": 3180
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"grad_norm": 0.4449957311153412,
|
| 1922 |
+
"learning_rate": 9.913591898020235e-05,
|
| 1923 |
+
"loss": 0.0231,
|
| 1924 |
+
"step": 3190
|
| 1925 |
+
},
|
| 1926 |
+
{
|
| 1927 |
+
"grad_norm": 0.49094122648239136,
|
| 1928 |
+
"learning_rate": 9.912568683753853e-05,
|
| 1929 |
+
"loss": 0.0242,
|
| 1930 |
+
"step": 3200
|
| 1931 |
+
},
|
| 1932 |
+
{
|
| 1933 |
+
"grad_norm": 0.31530943512916565,
|
| 1934 |
+
"learning_rate": 9.911539500253295e-05,
|
| 1935 |
+
"loss": 0.0253,
|
| 1936 |
+
"step": 3210
|
| 1937 |
+
},
|
| 1938 |
+
{
|
| 1939 |
+
"grad_norm": 0.3312493860721588,
|
| 1940 |
+
"learning_rate": 9.910504348769118e-05,
|
| 1941 |
+
"loss": 0.0219,
|
| 1942 |
+
"step": 3220
|
| 1943 |
+
},
|
| 1944 |
+
{
|
| 1945 |
+
"grad_norm": 0.25662603974342346,
|
| 1946 |
+
"learning_rate": 9.909463230559127e-05,
|
| 1947 |
+
"loss": 0.0202,
|
| 1948 |
+
"step": 3230
|
| 1949 |
+
},
|
| 1950 |
+
{
|
| 1951 |
+
"grad_norm": 0.35309165716171265,
|
| 1952 |
+
"learning_rate": 9.908416146888376e-05,
|
| 1953 |
+
"loss": 0.0226,
|
| 1954 |
+
"step": 3240
|
| 1955 |
+
},
|
| 1956 |
+
{
|
| 1957 |
+
"grad_norm": 0.3551497161388397,
|
| 1958 |
+
"learning_rate": 9.907363099029175e-05,
|
| 1959 |
+
"loss": 0.0218,
|
| 1960 |
+
"step": 3250
|
| 1961 |
+
},
|
| 1962 |
+
{
|
| 1963 |
+
"grad_norm": 0.3065674602985382,
|
| 1964 |
+
"learning_rate": 9.906304088261073e-05,
|
| 1965 |
+
"loss": 0.0259,
|
| 1966 |
+
"step": 3260
|
| 1967 |
+
},
|
| 1968 |
+
{
|
| 1969 |
+
"grad_norm": 0.34306737780570984,
|
| 1970 |
+
"learning_rate": 9.905239115870872e-05,
|
| 1971 |
+
"loss": 0.0246,
|
| 1972 |
+
"step": 3270
|
| 1973 |
+
},
|
| 1974 |
+
{
|
| 1975 |
+
"grad_norm": 0.25299543142318726,
|
| 1976 |
+
"learning_rate": 9.90416818315261e-05,
|
| 1977 |
+
"loss": 0.022,
|
| 1978 |
+
"step": 3280
|
| 1979 |
+
},
|
| 1980 |
+
{
|
| 1981 |
+
"grad_norm": 0.3820056915283203,
|
| 1982 |
+
"learning_rate": 9.903091291407573e-05,
|
| 1983 |
+
"loss": 0.027,
|
| 1984 |
+
"step": 3290
|
| 1985 |
+
},
|
| 1986 |
+
{
|
| 1987 |
+
"grad_norm": 0.3862016201019287,
|
| 1988 |
+
"learning_rate": 9.902008441944286e-05,
|
| 1989 |
+
"loss": 0.0231,
|
| 1990 |
+
"step": 3300
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"grad_norm": 0.3469122648239136,
|
| 1994 |
+
"learning_rate": 9.900919636078512e-05,
|
| 1995 |
+
"loss": 0.0226,
|
| 1996 |
+
"step": 3310
|
| 1997 |
+
},
|
| 1998 |
+
{
|
| 1999 |
+
"grad_norm": 0.3083968758583069,
|
| 2000 |
+
"learning_rate": 9.899824875133255e-05,
|
| 2001 |
+
"loss": 0.0242,
|
| 2002 |
+
"step": 3320
|
| 2003 |
+
},
|
| 2004 |
+
{
|
| 2005 |
+
"grad_norm": 0.46409106254577637,
|
| 2006 |
+
"learning_rate": 9.898724160438749e-05,
|
| 2007 |
+
"loss": 0.0219,
|
| 2008 |
+
"step": 3330
|
| 2009 |
+
},
|
| 2010 |
+
{
|
| 2011 |
+
"grad_norm": 0.2895025908946991,
|
| 2012 |
+
"learning_rate": 9.89761749333247e-05,
|
| 2013 |
+
"loss": 0.0232,
|
| 2014 |
+
"step": 3340
|
| 2015 |
+
},
|
| 2016 |
+
{
|
| 2017 |
+
"grad_norm": 0.33076900243759155,
|
| 2018 |
+
"learning_rate": 9.896504875159122e-05,
|
| 2019 |
+
"loss": 0.0216,
|
| 2020 |
+
"step": 3350
|
| 2021 |
+
},
|
| 2022 |
+
{
|
| 2023 |
+
"grad_norm": 0.30149585008621216,
|
| 2024 |
+
"learning_rate": 9.89538630727064e-05,
|
| 2025 |
+
"loss": 0.0212,
|
| 2026 |
+
"step": 3360
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"grad_norm": 0.26640868186950684,
|
| 2030 |
+
"learning_rate": 9.894261791026189e-05,
|
| 2031 |
+
"loss": 0.0231,
|
| 2032 |
+
"step": 3370
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"grad_norm": 0.43421223759651184,
|
| 2036 |
+
"learning_rate": 9.893131327792165e-05,
|
| 2037 |
+
"loss": 0.0211,
|
| 2038 |
+
"step": 3380
|
| 2039 |
+
},
|
| 2040 |
+
{
|
| 2041 |
+
"grad_norm": 0.34027573466300964,
|
| 2042 |
+
"learning_rate": 9.891994918942182e-05,
|
| 2043 |
+
"loss": 0.0202,
|
| 2044 |
+
"step": 3390
|
| 2045 |
+
},
|
| 2046 |
+
{
|
| 2047 |
+
"grad_norm": 0.3347245156764984,
|
| 2048 |
+
"learning_rate": 9.890852565857092e-05,
|
| 2049 |
+
"loss": 0.0229,
|
| 2050 |
+
"step": 3400
|
| 2051 |
+
},
|
| 2052 |
+
{
|
| 2053 |
+
"grad_norm": 0.35026970505714417,
|
| 2054 |
+
"learning_rate": 9.889704269924954e-05,
|
| 2055 |
+
"loss": 0.0206,
|
| 2056 |
+
"step": 3410
|
| 2057 |
+
},
|
| 2058 |
+
{
|
| 2059 |
+
"grad_norm": 0.3942737579345703,
|
| 2060 |
+
"learning_rate": 9.888550032541059e-05,
|
| 2061 |
+
"loss": 0.022,
|
| 2062 |
+
"step": 3420
|
| 2063 |
+
},
|
| 2064 |
+
{
|
| 2065 |
+
"grad_norm": 0.39353707432746887,
|
| 2066 |
+
"learning_rate": 9.887389855107916e-05,
|
| 2067 |
+
"loss": 0.0226,
|
| 2068 |
+
"step": 3430
|
| 2069 |
+
},
|
| 2070 |
+
{
|
| 2071 |
+
"grad_norm": 0.3677908182144165,
|
| 2072 |
+
"learning_rate": 9.886223739035248e-05,
|
| 2073 |
+
"loss": 0.0272,
|
| 2074 |
+
"step": 3440
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"grad_norm": 0.37933892011642456,
|
| 2078 |
+
"learning_rate": 9.885051685739997e-05,
|
| 2079 |
+
"loss": 0.0243,
|
| 2080 |
+
"step": 3450
|
| 2081 |
+
},
|
| 2082 |
+
{
|
| 2083 |
+
"grad_norm": 0.4893406629562378,
|
| 2084 |
+
"learning_rate": 9.883873696646316e-05,
|
| 2085 |
+
"loss": 0.027,
|
| 2086 |
+
"step": 3460
|
| 2087 |
+
},
|
| 2088 |
+
{
|
| 2089 |
+
"grad_norm": 0.300083190202713,
|
| 2090 |
+
"learning_rate": 9.882689773185575e-05,
|
| 2091 |
+
"loss": 0.0207,
|
| 2092 |
+
"step": 3470
|
| 2093 |
+
},
|
| 2094 |
+
{
|
| 2095 |
+
"grad_norm": 0.40003201365470886,
|
| 2096 |
+
"learning_rate": 9.881499916796353e-05,
|
| 2097 |
+
"loss": 0.0243,
|
| 2098 |
+
"step": 3480
|
| 2099 |
+
},
|
| 2100 |
+
{
|
| 2101 |
+
"grad_norm": 0.3844235837459564,
|
| 2102 |
+
"learning_rate": 9.880304128924434e-05,
|
| 2103 |
+
"loss": 0.0225,
|
| 2104 |
+
"step": 3490
|
| 2105 |
+
},
|
| 2106 |
+
{
|
| 2107 |
+
"grad_norm": 0.31904521584510803,
|
| 2108 |
+
"learning_rate": 9.879102411022817e-05,
|
| 2109 |
+
"loss": 0.0223,
|
| 2110 |
+
"step": 3500
|
| 2111 |
+
},
|
| 2112 |
+
{
|
| 2113 |
+
"grad_norm": 0.34493687748908997,
|
| 2114 |
+
"learning_rate": 9.877894764551703e-05,
|
| 2115 |
+
"loss": 0.0201,
|
| 2116 |
+
"step": 3510
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"grad_norm": 0.41117796301841736,
|
| 2120 |
+
"learning_rate": 9.876681190978494e-05,
|
| 2121 |
+
"loss": 0.024,
|
| 2122 |
+
"step": 3520
|
| 2123 |
+
},
|
| 2124 |
+
{
|
| 2125 |
+
"grad_norm": 0.3274706304073334,
|
| 2126 |
+
"learning_rate": 9.875461691777797e-05,
|
| 2127 |
+
"loss": 0.0258,
|
| 2128 |
+
"step": 3530
|
| 2129 |
+
},
|
| 2130 |
+
{
|
| 2131 |
+
"grad_norm": 0.28383010625839233,
|
| 2132 |
+
"learning_rate": 9.874236268431417e-05,
|
| 2133 |
+
"loss": 0.0219,
|
| 2134 |
+
"step": 3540
|
| 2135 |
+
},
|
| 2136 |
+
{
|
| 2137 |
+
"grad_norm": 0.3705231845378876,
|
| 2138 |
+
"learning_rate": 9.873004922428361e-05,
|
| 2139 |
+
"loss": 0.0213,
|
| 2140 |
+
"step": 3550
|
| 2141 |
+
},
|
| 2142 |
+
{
|
| 2143 |
+
"grad_norm": 0.3192824423313141,
|
| 2144 |
+
"learning_rate": 9.871767655264829e-05,
|
| 2145 |
+
"loss": 0.022,
|
| 2146 |
+
"step": 3560
|
| 2147 |
+
},
|
| 2148 |
+
{
|
| 2149 |
+
"grad_norm": 0.25980308651924133,
|
| 2150 |
+
"learning_rate": 9.87052446844422e-05,
|
| 2151 |
+
"loss": 0.0211,
|
| 2152 |
+
"step": 3570
|
| 2153 |
+
},
|
| 2154 |
+
{
|
| 2155 |
+
"grad_norm": 0.2518915832042694,
|
| 2156 |
+
"learning_rate": 9.869275363477122e-05,
|
| 2157 |
+
"loss": 0.024,
|
| 2158 |
+
"step": 3580
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"grad_norm": 0.3379889726638794,
|
| 2162 |
+
"learning_rate": 9.868020341881312e-05,
|
| 2163 |
+
"loss": 0.0209,
|
| 2164 |
+
"step": 3590
|
| 2165 |
+
},
|
| 2166 |
+
{
|
| 2167 |
+
"grad_norm": 0.39600417017936707,
|
| 2168 |
+
"learning_rate": 9.866759405181765e-05,
|
| 2169 |
+
"loss": 0.0237,
|
| 2170 |
+
"step": 3600
|
| 2171 |
+
},
|
| 2172 |
+
{
|
| 2173 |
+
"grad_norm": 0.33133968710899353,
|
| 2174 |
+
"learning_rate": 9.865492554910633e-05,
|
| 2175 |
+
"loss": 0.0209,
|
| 2176 |
+
"step": 3610
|
| 2177 |
+
},
|
| 2178 |
+
{
|
| 2179 |
+
"grad_norm": 0.32698678970336914,
|
| 2180 |
+
"learning_rate": 9.864219792607262e-05,
|
| 2181 |
+
"loss": 0.0196,
|
| 2182 |
+
"step": 3620
|
| 2183 |
+
},
|
| 2184 |
+
{
|
| 2185 |
+
"grad_norm": 0.3060948848724365,
|
| 2186 |
+
"learning_rate": 9.862941119818177e-05,
|
| 2187 |
+
"loss": 0.0192,
|
| 2188 |
+
"step": 3630
|
| 2189 |
+
},
|
| 2190 |
+
{
|
| 2191 |
+
"grad_norm": 0.3851982355117798,
|
| 2192 |
+
"learning_rate": 9.861656538097086e-05,
|
| 2193 |
+
"loss": 0.0208,
|
| 2194 |
+
"step": 3640
|
| 2195 |
+
},
|
| 2196 |
+
{
|
| 2197 |
+
"grad_norm": 0.49290308356285095,
|
| 2198 |
+
"learning_rate": 9.860366049004877e-05,
|
| 2199 |
+
"loss": 0.022,
|
| 2200 |
+
"step": 3650
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"grad_norm": 0.3673649728298187,
|
| 2204 |
+
"learning_rate": 9.859069654109615e-05,
|
| 2205 |
+
"loss": 0.0233,
|
| 2206 |
+
"step": 3660
|
| 2207 |
+
},
|
| 2208 |
+
{
|
| 2209 |
+
"grad_norm": 0.2948131561279297,
|
| 2210 |
+
"learning_rate": 9.857767354986545e-05,
|
| 2211 |
+
"loss": 0.023,
|
| 2212 |
+
"step": 3670
|
| 2213 |
+
},
|
| 2214 |
+
{
|
| 2215 |
+
"grad_norm": 0.3540187478065491,
|
| 2216 |
+
"learning_rate": 9.856459153218078e-05,
|
| 2217 |
+
"loss": 0.0196,
|
| 2218 |
+
"step": 3680
|
| 2219 |
+
},
|
| 2220 |
+
{
|
| 2221 |
+
"grad_norm": 0.27228233218193054,
|
| 2222 |
+
"learning_rate": 9.855145050393808e-05,
|
| 2223 |
+
"loss": 0.0211,
|
| 2224 |
+
"step": 3690
|
| 2225 |
+
},
|
| 2226 |
+
{
|
| 2227 |
+
"grad_norm": 0.24771757423877716,
|
| 2228 |
+
"learning_rate": 9.85382504811049e-05,
|
| 2229 |
+
"loss": 0.021,
|
| 2230 |
+
"step": 3700
|
| 2231 |
+
},
|
| 2232 |
+
{
|
| 2233 |
+
"grad_norm": 0.26397937536239624,
|
| 2234 |
+
"learning_rate": 9.852499147972054e-05,
|
| 2235 |
+
"loss": 0.0201,
|
| 2236 |
+
"step": 3710
|
| 2237 |
+
},
|
| 2238 |
+
{
|
| 2239 |
+
"grad_norm": 0.28861305117607117,
|
| 2240 |
+
"learning_rate": 9.851167351589592e-05,
|
| 2241 |
+
"loss": 0.0215,
|
| 2242 |
+
"step": 3720
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"grad_norm": 0.3233491778373718,
|
| 2246 |
+
"learning_rate": 9.849829660581363e-05,
|
| 2247 |
+
"loss": 0.0201,
|
| 2248 |
+
"step": 3730
|
| 2249 |
+
},
|
| 2250 |
+
{
|
| 2251 |
+
"grad_norm": 0.3898829519748688,
|
| 2252 |
+
"learning_rate": 9.848486076572787e-05,
|
| 2253 |
+
"loss": 0.0221,
|
| 2254 |
+
"step": 3740
|
| 2255 |
+
},
|
| 2256 |
+
{
|
| 2257 |
+
"grad_norm": 0.38374295830726624,
|
| 2258 |
+
"learning_rate": 9.847136601196446e-05,
|
| 2259 |
+
"loss": 0.0223,
|
| 2260 |
+
"step": 3750
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"grad_norm": 0.2840521037578583,
|
| 2264 |
+
"learning_rate": 9.845781236092078e-05,
|
| 2265 |
+
"loss": 0.0221,
|
| 2266 |
+
"step": 3760
|
| 2267 |
+
},
|
| 2268 |
+
{
|
| 2269 |
+
"grad_norm": 0.3588821589946747,
|
| 2270 |
+
"learning_rate": 9.844419982906583e-05,
|
| 2271 |
+
"loss": 0.0212,
|
| 2272 |
+
"step": 3770
|
| 2273 |
+
},
|
| 2274 |
+
{
|
| 2275 |
+
"grad_norm": 0.34918320178985596,
|
| 2276 |
+
"learning_rate": 9.843052843294008e-05,
|
| 2277 |
+
"loss": 0.019,
|
| 2278 |
+
"step": 3780
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"grad_norm": 0.3414880633354187,
|
| 2282 |
+
"learning_rate": 9.841679818915559e-05,
|
| 2283 |
+
"loss": 0.0216,
|
| 2284 |
+
"step": 3790
|
| 2285 |
+
},
|
| 2286 |
+
{
|
| 2287 |
+
"grad_norm": 0.35619089007377625,
|
| 2288 |
+
"learning_rate": 9.840300911439591e-05,
|
| 2289 |
+
"loss": 0.0213,
|
| 2290 |
+
"step": 3800
|
| 2291 |
+
},
|
| 2292 |
+
{
|
| 2293 |
+
"grad_norm": 0.3069482147693634,
|
| 2294 |
+
"learning_rate": 9.838916122541603e-05,
|
| 2295 |
+
"loss": 0.019,
|
| 2296 |
+
"step": 3810
|
| 2297 |
+
},
|
| 2298 |
+
{
|
| 2299 |
+
"grad_norm": 0.31180906295776367,
|
| 2300 |
+
"learning_rate": 9.837525453904246e-05,
|
| 2301 |
+
"loss": 0.0196,
|
| 2302 |
+
"step": 3820
|
| 2303 |
+
},
|
| 2304 |
+
{
|
| 2305 |
+
"grad_norm": 0.29071903228759766,
|
| 2306 |
+
"learning_rate": 9.836128907217314e-05,
|
| 2307 |
+
"loss": 0.0201,
|
| 2308 |
+
"step": 3830
|
| 2309 |
+
},
|
| 2310 |
+
{
|
| 2311 |
+
"grad_norm": 0.336658239364624,
|
| 2312 |
+
"learning_rate": 9.834726484177743e-05,
|
| 2313 |
+
"loss": 0.0198,
|
| 2314 |
+
"step": 3840
|
| 2315 |
+
},
|
| 2316 |
+
{
|
| 2317 |
+
"grad_norm": 0.3188108503818512,
|
| 2318 |
+
"learning_rate": 9.833318186489609e-05,
|
| 2319 |
+
"loss": 0.0205,
|
| 2320 |
+
"step": 3850
|
| 2321 |
+
},
|
| 2322 |
+
{
|
| 2323 |
+
"grad_norm": 0.2784384787082672,
|
| 2324 |
+
"learning_rate": 9.831904015864126e-05,
|
| 2325 |
+
"loss": 0.0196,
|
| 2326 |
+
"step": 3860
|
| 2327 |
+
},
|
| 2328 |
+
{
|
| 2329 |
+
"grad_norm": 0.3234076201915741,
|
| 2330 |
+
"learning_rate": 9.830483974019645e-05,
|
| 2331 |
+
"loss": 0.0211,
|
| 2332 |
+
"step": 3870
|
| 2333 |
+
},
|
| 2334 |
+
{
|
| 2335 |
+
"grad_norm": 0.3331151008605957,
|
| 2336 |
+
"learning_rate": 9.82905806268165e-05,
|
| 2337 |
+
"loss": 0.0191,
|
| 2338 |
+
"step": 3880
|
| 2339 |
+
},
|
| 2340 |
+
{
|
| 2341 |
+
"grad_norm": 0.31715860962867737,
|
| 2342 |
+
"learning_rate": 9.82762628358276e-05,
|
| 2343 |
+
"loss": 0.0173,
|
| 2344 |
+
"step": 3890
|
| 2345 |
+
},
|
| 2346 |
+
{
|
| 2347 |
+
"grad_norm": 0.3170963227748871,
|
| 2348 |
+
"learning_rate": 9.826188638462718e-05,
|
| 2349 |
+
"loss": 0.018,
|
| 2350 |
+
"step": 3900
|
| 2351 |
+
},
|
| 2352 |
+
{
|
| 2353 |
+
"grad_norm": 0.2912008762359619,
|
| 2354 |
+
"learning_rate": 9.824745129068402e-05,
|
| 2355 |
+
"loss": 0.022,
|
| 2356 |
+
"step": 3910
|
| 2357 |
+
},
|
| 2358 |
+
{
|
| 2359 |
+
"grad_norm": 0.48704785108566284,
|
| 2360 |
+
"learning_rate": 9.82329575715381e-05,
|
| 2361 |
+
"loss": 0.0242,
|
| 2362 |
+
"step": 3920
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"grad_norm": 0.28424951434135437,
|
| 2366 |
+
"learning_rate": 9.821840524480066e-05,
|
| 2367 |
+
"loss": 0.0203,
|
| 2368 |
+
"step": 3930
|
| 2369 |
+
},
|
| 2370 |
+
{
|
| 2371 |
+
"grad_norm": 0.2560309171676636,
|
| 2372 |
+
"learning_rate": 9.820379432815414e-05,
|
| 2373 |
+
"loss": 0.0203,
|
| 2374 |
+
"step": 3940
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"grad_norm": 0.3647140562534332,
|
| 2378 |
+
"learning_rate": 9.81891248393522e-05,
|
| 2379 |
+
"loss": 0.0187,
|
| 2380 |
+
"step": 3950
|
| 2381 |
+
},
|
| 2382 |
+
{
|
| 2383 |
+
"grad_norm": 0.31988921761512756,
|
| 2384 |
+
"learning_rate": 9.817439679621963e-05,
|
| 2385 |
+
"loss": 0.0207,
|
| 2386 |
+
"step": 3960
|
| 2387 |
+
},
|
| 2388 |
+
{
|
| 2389 |
+
"grad_norm": 0.3253921866416931,
|
| 2390 |
+
"learning_rate": 9.815961021665243e-05,
|
| 2391 |
+
"loss": 0.0212,
|
| 2392 |
+
"step": 3970
|
| 2393 |
+
},
|
| 2394 |
+
{
|
| 2395 |
+
"grad_norm": 0.29170772433280945,
|
| 2396 |
+
"learning_rate": 9.814476511861763e-05,
|
| 2397 |
+
"loss": 0.0263,
|
| 2398 |
+
"step": 3980
|
| 2399 |
+
},
|
| 2400 |
+
{
|
| 2401 |
+
"grad_norm": 0.4017032980918884,
|
| 2402 |
+
"learning_rate": 9.812986152015348e-05,
|
| 2403 |
+
"loss": 0.0265,
|
| 2404 |
+
"step": 3990
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"grad_norm": 0.28655076026916504,
|
| 2408 |
+
"learning_rate": 9.811489943936922e-05,
|
| 2409 |
+
"loss": 0.0205,
|
| 2410 |
+
"step": 4000
|
| 2411 |
+
},
|
| 2412 |
+
{
|
| 2413 |
+
"grad_norm": 0.23081740736961365,
|
| 2414 |
+
"learning_rate": 9.809987889444522e-05,
|
| 2415 |
+
"loss": 0.021,
|
| 2416 |
+
"step": 4010
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"grad_norm": 0.34190860390663147,
|
| 2420 |
+
"learning_rate": 9.808479990363282e-05,
|
| 2421 |
+
"loss": 0.0203,
|
| 2422 |
+
"step": 4020
|
| 2423 |
+
},
|
| 2424 |
+
{
|
| 2425 |
+
"grad_norm": 0.2937399744987488,
|
| 2426 |
+
"learning_rate": 9.806966248525445e-05,
|
| 2427 |
+
"loss": 0.0188,
|
| 2428 |
+
"step": 4030
|
| 2429 |
+
},
|
| 2430 |
+
{
|
| 2431 |
+
"grad_norm": 0.2845223546028137,
|
| 2432 |
+
"learning_rate": 9.805446665770348e-05,
|
| 2433 |
+
"loss": 0.0191,
|
| 2434 |
+
"step": 4040
|
| 2435 |
+
},
|
| 2436 |
+
{
|
| 2437 |
+
"grad_norm": 0.2973966896533966,
|
| 2438 |
+
"learning_rate": 9.803921243944429e-05,
|
| 2439 |
+
"loss": 0.0175,
|
| 2440 |
+
"step": 4050
|
| 2441 |
+
},
|
| 2442 |
+
{
|
| 2443 |
+
"grad_norm": 0.3171347379684448,
|
| 2444 |
+
"learning_rate": 9.802389984901218e-05,
|
| 2445 |
+
"loss": 0.0195,
|
| 2446 |
+
"step": 4060
|
| 2447 |
+
},
|
| 2448 |
+
{
|
| 2449 |
+
"grad_norm": 0.2669548988342285,
|
| 2450 |
+
"learning_rate": 9.80085289050134e-05,
|
| 2451 |
+
"loss": 0.0183,
|
| 2452 |
+
"step": 4070
|
| 2453 |
+
},
|
| 2454 |
+
{
|
| 2455 |
+
"grad_norm": 0.42913228273391724,
|
| 2456 |
+
"learning_rate": 9.799309962612508e-05,
|
| 2457 |
+
"loss": 0.0196,
|
| 2458 |
+
"step": 4080
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"grad_norm": 0.4162735342979431,
|
| 2462 |
+
"learning_rate": 9.797761203109527e-05,
|
| 2463 |
+
"loss": 0.0215,
|
| 2464 |
+
"step": 4090
|
| 2465 |
+
},
|
| 2466 |
+
{
|
| 2467 |
+
"grad_norm": 0.31903430819511414,
|
| 2468 |
+
"learning_rate": 9.796206613874283e-05,
|
| 2469 |
+
"loss": 0.0195,
|
| 2470 |
+
"step": 4100
|
| 2471 |
+
},
|
| 2472 |
+
{
|
| 2473 |
+
"grad_norm": 0.3554883301258087,
|
| 2474 |
+
"learning_rate": 9.794646196795754e-05,
|
| 2475 |
+
"loss": 0.0201,
|
| 2476 |
+
"step": 4110
|
| 2477 |
+
},
|
| 2478 |
+
{
|
| 2479 |
+
"grad_norm": 0.31217727065086365,
|
| 2480 |
+
"learning_rate": 9.793079953769987e-05,
|
| 2481 |
+
"loss": 0.019,
|
| 2482 |
+
"step": 4120
|
| 2483 |
+
},
|
| 2484 |
+
{
|
| 2485 |
+
"grad_norm": 0.28474947810173035,
|
| 2486 |
+
"learning_rate": 9.79150788670012e-05,
|
| 2487 |
+
"loss": 0.0184,
|
| 2488 |
+
"step": 4130
|
| 2489 |
+
},
|
| 2490 |
+
{
|
| 2491 |
+
"grad_norm": 0.32388654351234436,
|
| 2492 |
+
"learning_rate": 9.78992999749636e-05,
|
| 2493 |
+
"loss": 0.0216,
|
| 2494 |
+
"step": 4140
|
| 2495 |
+
},
|
| 2496 |
+
{
|
| 2497 |
+
"grad_norm": 0.20659787952899933,
|
| 2498 |
+
"learning_rate": 9.788346288075994e-05,
|
| 2499 |
+
"loss": 0.0181,
|
| 2500 |
+
"step": 4150
|
| 2501 |
+
},
|
| 2502 |
+
{
|
| 2503 |
+
"grad_norm": 0.2864404022693634,
|
| 2504 |
+
"learning_rate": 9.786756760363373e-05,
|
| 2505 |
+
"loss": 0.0207,
|
| 2506 |
+
"step": 4160
|
| 2507 |
+
},
|
| 2508 |
+
{
|
| 2509 |
+
"grad_norm": 0.2995869815349579,
|
| 2510 |
+
"learning_rate": 9.78516141628993e-05,
|
| 2511 |
+
"loss": 0.0182,
|
| 2512 |
+
"step": 4170
|
| 2513 |
+
},
|
| 2514 |
+
{
|
| 2515 |
+
"grad_norm": 0.31789812445640564,
|
| 2516 |
+
"learning_rate": 9.783560257794154e-05,
|
| 2517 |
+
"loss": 0.0188,
|
| 2518 |
+
"step": 4180
|
| 2519 |
+
},
|
| 2520 |
+
{
|
| 2521 |
+
"grad_norm": 0.28194430470466614,
|
| 2522 |
+
"learning_rate": 9.781953286821603e-05,
|
| 2523 |
+
"loss": 0.0168,
|
| 2524 |
+
"step": 4190
|
| 2525 |
+
},
|
| 2526 |
+
{
|
| 2527 |
+
"grad_norm": 0.31114333868026733,
|
| 2528 |
+
"learning_rate": 9.780340505324901e-05,
|
| 2529 |
+
"loss": 0.0172,
|
| 2530 |
+
"step": 4200
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"grad_norm": 0.2011519819498062,
|
| 2534 |
+
"learning_rate": 9.778721915263727e-05,
|
| 2535 |
+
"loss": 0.0177,
|
| 2536 |
+
"step": 4210
|
| 2537 |
+
},
|
| 2538 |
+
{
|
| 2539 |
+
"grad_norm": 0.22189873456954956,
|
| 2540 |
+
"learning_rate": 9.777097518604824e-05,
|
| 2541 |
+
"loss": 0.0167,
|
| 2542 |
+
"step": 4220
|
| 2543 |
+
},
|
| 2544 |
+
{
|
| 2545 |
+
"grad_norm": 0.30673474073410034,
|
| 2546 |
+
"learning_rate": 9.775467317321984e-05,
|
| 2547 |
+
"loss": 0.0161,
|
| 2548 |
+
"step": 4230
|
| 2549 |
+
},
|
| 2550 |
+
{
|
| 2551 |
+
"grad_norm": 0.3615068197250366,
|
| 2552 |
+
"learning_rate": 9.773831313396055e-05,
|
| 2553 |
+
"loss": 0.0201,
|
| 2554 |
+
"step": 4240
|
| 2555 |
+
},
|
| 2556 |
+
{
|
| 2557 |
+
"grad_norm": 0.25624963641166687,
|
| 2558 |
+
"learning_rate": 9.77218950881494e-05,
|
| 2559 |
+
"loss": 0.0176,
|
| 2560 |
+
"step": 4250
|
| 2561 |
+
},
|
| 2562 |
+
{
|
| 2563 |
+
"grad_norm": 0.32759591937065125,
|
| 2564 |
+
"learning_rate": 9.770541905573583e-05,
|
| 2565 |
+
"loss": 0.0182,
|
| 2566 |
+
"step": 4260
|
| 2567 |
+
},
|
| 2568 |
+
{
|
| 2569 |
+
"grad_norm": 0.23994171619415283,
|
| 2570 |
+
"learning_rate": 9.768888505673976e-05,
|
| 2571 |
+
"loss": 0.0157,
|
| 2572 |
+
"step": 4270
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"grad_norm": 0.29931554198265076,
|
| 2576 |
+
"learning_rate": 9.767229311125162e-05,
|
| 2577 |
+
"loss": 0.0173,
|
| 2578 |
+
"step": 4280
|
| 2579 |
+
},
|
| 2580 |
+
{
|
| 2581 |
+
"grad_norm": 0.3035125732421875,
|
| 2582 |
+
"learning_rate": 9.765564323943211e-05,
|
| 2583 |
+
"loss": 0.0181,
|
| 2584 |
+
"step": 4290
|
| 2585 |
+
},
|
| 2586 |
+
{
|
| 2587 |
+
"grad_norm": 0.3667411804199219,
|
| 2588 |
+
"learning_rate": 9.763893546151244e-05,
|
| 2589 |
+
"loss": 0.0192,
|
| 2590 |
+
"step": 4300
|
| 2591 |
+
},
|
| 2592 |
+
{
|
| 2593 |
+
"grad_norm": 0.2947080731391907,
|
| 2594 |
+
"learning_rate": 9.762216979779412e-05,
|
| 2595 |
+
"loss": 0.0203,
|
| 2596 |
+
"step": 4310
|
| 2597 |
+
},
|
| 2598 |
+
{
|
| 2599 |
+
"grad_norm": 0.2096947431564331,
|
| 2600 |
+
"learning_rate": 9.760534626864902e-05,
|
| 2601 |
+
"loss": 0.0185,
|
| 2602 |
+
"step": 4320
|
| 2603 |
+
},
|
| 2604 |
+
{
|
| 2605 |
+
"grad_norm": 0.301518976688385,
|
| 2606 |
+
"learning_rate": 9.758846489451931e-05,
|
| 2607 |
+
"loss": 0.0203,
|
| 2608 |
+
"step": 4330
|
| 2609 |
+
},
|
| 2610 |
+
{
|
| 2611 |
+
"grad_norm": 0.3711138367652893,
|
| 2612 |
+
"learning_rate": 9.757152569591748e-05,
|
| 2613 |
+
"loss": 0.0199,
|
| 2614 |
+
"step": 4340
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"grad_norm": 0.23832212388515472,
|
| 2618 |
+
"learning_rate": 9.75545286934262e-05,
|
| 2619 |
+
"loss": 0.0184,
|
| 2620 |
+
"step": 4350
|
| 2621 |
+
},
|
| 2622 |
+
{
|
| 2623 |
+
"grad_norm": 0.3737817108631134,
|
| 2624 |
+
"learning_rate": 9.753747390769847e-05,
|
| 2625 |
+
"loss": 0.018,
|
| 2626 |
+
"step": 4360
|
| 2627 |
+
},
|
| 2628 |
+
{
|
| 2629 |
+
"grad_norm": 0.2761482000350952,
|
| 2630 |
+
"learning_rate": 9.752036135945744e-05,
|
| 2631 |
+
"loss": 0.0194,
|
| 2632 |
+
"step": 4370
|
| 2633 |
+
},
|
| 2634 |
+
{
|
| 2635 |
+
"grad_norm": 0.40149354934692383,
|
| 2636 |
+
"learning_rate": 9.750319106949649e-05,
|
| 2637 |
+
"loss": 0.0199,
|
| 2638 |
+
"step": 4380
|
| 2639 |
+
},
|
| 2640 |
+
{
|
| 2641 |
+
"grad_norm": 0.2389083057641983,
|
| 2642 |
+
"learning_rate": 9.748596305867913e-05,
|
| 2643 |
+
"loss": 0.0193,
|
| 2644 |
+
"step": 4390
|
| 2645 |
+
},
|
| 2646 |
+
{
|
| 2647 |
+
"grad_norm": 0.2904573380947113,
|
| 2648 |
+
"learning_rate": 9.746867734793903e-05,
|
| 2649 |
+
"loss": 0.0194,
|
| 2650 |
+
"step": 4400
|
| 2651 |
+
},
|
| 2652 |
+
{
|
| 2653 |
+
"grad_norm": 0.28680405020713806,
|
| 2654 |
+
"learning_rate": 9.745133395827993e-05,
|
| 2655 |
+
"loss": 0.0213,
|
| 2656 |
+
"step": 4410
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"grad_norm": 0.3064373731613159,
|
| 2660 |
+
"learning_rate": 9.743393291077572e-05,
|
| 2661 |
+
"loss": 0.0178,
|
| 2662 |
+
"step": 4420
|
| 2663 |
+
},
|
| 2664 |
+
{
|
| 2665 |
+
"grad_norm": 0.23364102840423584,
|
| 2666 |
+
"learning_rate": 9.741647422657028e-05,
|
| 2667 |
+
"loss": 0.02,
|
| 2668 |
+
"step": 4430
|
| 2669 |
+
},
|
| 2670 |
+
{
|
| 2671 |
+
"grad_norm": 0.2293742299079895,
|
| 2672 |
+
"learning_rate": 9.739895792687758e-05,
|
| 2673 |
+
"loss": 0.0173,
|
| 2674 |
+
"step": 4440
|
| 2675 |
+
},
|
| 2676 |
+
{
|
| 2677 |
+
"grad_norm": 0.4201240539550781,
|
| 2678 |
+
"learning_rate": 9.738138403298157e-05,
|
| 2679 |
+
"loss": 0.0203,
|
| 2680 |
+
"step": 4450
|
| 2681 |
+
},
|
| 2682 |
+
{
|
| 2683 |
+
"grad_norm": 0.30462443828582764,
|
| 2684 |
+
"learning_rate": 9.736375256623619e-05,
|
| 2685 |
+
"loss": 0.0173,
|
| 2686 |
+
"step": 4460
|
| 2687 |
+
},
|
| 2688 |
+
{
|
| 2689 |
+
"grad_norm": 0.26824820041656494,
|
| 2690 |
+
"learning_rate": 9.734606354806533e-05,
|
| 2691 |
+
"loss": 0.0163,
|
| 2692 |
+
"step": 4470
|
| 2693 |
+
},
|
| 2694 |
+
{
|
| 2695 |
+
"grad_norm": 0.3436148762702942,
|
| 2696 |
+
"learning_rate": 9.73283169999628e-05,
|
| 2697 |
+
"loss": 0.0196,
|
| 2698 |
+
"step": 4480
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"grad_norm": 0.2987440526485443,
|
| 2702 |
+
"learning_rate": 9.731051294349238e-05,
|
| 2703 |
+
"loss": 0.0185,
|
| 2704 |
+
"step": 4490
|
| 2705 |
+
},
|
| 2706 |
+
{
|
| 2707 |
+
"grad_norm": 0.3387744724750519,
|
| 2708 |
+
"learning_rate": 9.729265140028762e-05,
|
| 2709 |
+
"loss": 0.0208,
|
| 2710 |
+
"step": 4500
|
| 2711 |
+
},
|
| 2712 |
+
{
|
| 2713 |
+
"grad_norm": 0.22033269703388214,
|
| 2714 |
+
"learning_rate": 9.727473239205201e-05,
|
| 2715 |
+
"loss": 0.0154,
|
| 2716 |
+
"step": 4510
|
| 2717 |
+
},
|
| 2718 |
+
{
|
| 2719 |
+
"grad_norm": 0.3045775890350342,
|
| 2720 |
+
"learning_rate": 9.725675594055883e-05,
|
| 2721 |
+
"loss": 0.0156,
|
| 2722 |
+
"step": 4520
|
| 2723 |
+
},
|
| 2724 |
+
{
|
| 2725 |
+
"grad_norm": 0.3046495020389557,
|
| 2726 |
+
"learning_rate": 9.723872206765116e-05,
|
| 2727 |
+
"loss": 0.0172,
|
| 2728 |
+
"step": 4530
|
| 2729 |
+
},
|
| 2730 |
+
{
|
| 2731 |
+
"grad_norm": 0.28411006927490234,
|
| 2732 |
+
"learning_rate": 9.722063079524185e-05,
|
| 2733 |
+
"loss": 0.0176,
|
| 2734 |
+
"step": 4540
|
| 2735 |
+
},
|
| 2736 |
+
{
|
| 2737 |
+
"grad_norm": 0.2794414758682251,
|
| 2738 |
+
"learning_rate": 9.720248214531351e-05,
|
| 2739 |
+
"loss": 0.017,
|
| 2740 |
+
"step": 4550
|
| 2741 |
+
},
|
| 2742 |
+
{
|
| 2743 |
+
"grad_norm": 0.28699204325675964,
|
| 2744 |
+
"learning_rate": 9.718427613991848e-05,
|
| 2745 |
+
"loss": 0.0183,
|
| 2746 |
+
"step": 4560
|
| 2747 |
+
},
|
| 2748 |
+
{
|
| 2749 |
+
"grad_norm": 0.23602908849716187,
|
| 2750 |
+
"learning_rate": 9.716601280117873e-05,
|
| 2751 |
+
"loss": 0.015,
|
| 2752 |
+
"step": 4570
|
| 2753 |
+
},
|
| 2754 |
+
{
|
| 2755 |
+
"grad_norm": 0.3002392053604126,
|
| 2756 |
+
"learning_rate": 9.714769215128596e-05,
|
| 2757 |
+
"loss": 0.0187,
|
| 2758 |
+
"step": 4580
|
| 2759 |
+
},
|
| 2760 |
+
{
|
| 2761 |
+
"grad_norm": 0.2982144355773926,
|
| 2762 |
+
"learning_rate": 9.712931421250152e-05,
|
| 2763 |
+
"loss": 0.0204,
|
| 2764 |
+
"step": 4590
|
| 2765 |
+
},
|
| 2766 |
+
{
|
| 2767 |
+
"grad_norm": 0.33681154251098633,
|
| 2768 |
+
"learning_rate": 9.711087900715627e-05,
|
| 2769 |
+
"loss": 0.0186,
|
| 2770 |
+
"step": 4600
|
| 2771 |
+
},
|
| 2772 |
+
{
|
| 2773 |
+
"grad_norm": 0.2547285854816437,
|
| 2774 |
+
"learning_rate": 9.709238655765078e-05,
|
| 2775 |
+
"loss": 0.0193,
|
| 2776 |
+
"step": 4610
|
| 2777 |
+
},
|
| 2778 |
+
{
|
| 2779 |
+
"grad_norm": 0.2880988121032715,
|
| 2780 |
+
"learning_rate": 9.707383688645511e-05,
|
| 2781 |
+
"loss": 0.0169,
|
| 2782 |
+
"step": 4620
|
| 2783 |
+
},
|
| 2784 |
+
{
|
| 2785 |
+
"grad_norm": 0.2253848910331726,
|
| 2786 |
+
"learning_rate": 9.705523001610883e-05,
|
| 2787 |
+
"loss": 0.0173,
|
| 2788 |
+
"step": 4630
|
| 2789 |
+
},
|
| 2790 |
+
{
|
| 2791 |
+
"grad_norm": 0.24737682938575745,
|
| 2792 |
+
"learning_rate": 9.703656596922107e-05,
|
| 2793 |
+
"loss": 0.0152,
|
| 2794 |
+
"step": 4640
|
| 2795 |
+
},
|
| 2796 |
+
{
|
| 2797 |
+
"grad_norm": 0.26355868577957153,
|
| 2798 |
+
"learning_rate": 9.70178447684704e-05,
|
| 2799 |
+
"loss": 0.0171,
|
| 2800 |
+
"step": 4650
|
| 2801 |
+
},
|
| 2802 |
+
{
|
| 2803 |
+
"grad_norm": 0.33829301595687866,
|
| 2804 |
+
"learning_rate": 9.699906643660483e-05,
|
| 2805 |
+
"loss": 0.0177,
|
| 2806 |
+
"step": 4660
|
| 2807 |
+
},
|
| 2808 |
+
{
|
| 2809 |
+
"grad_norm": 0.27919113636016846,
|
| 2810 |
+
"learning_rate": 9.698023099644185e-05,
|
| 2811 |
+
"loss": 0.018,
|
| 2812 |
+
"step": 4670
|
| 2813 |
+
},
|
| 2814 |
+
{
|
| 2815 |
+
"grad_norm": 0.3479491174221039,
|
| 2816 |
+
"learning_rate": 9.696133847086823e-05,
|
| 2817 |
+
"loss": 0.0165,
|
| 2818 |
+
"step": 4680
|
| 2819 |
+
},
|
| 2820 |
+
{
|
| 2821 |
+
"grad_norm": 0.23865260183811188,
|
| 2822 |
+
"learning_rate": 9.694238888284022e-05,
|
| 2823 |
+
"loss": 0.0206,
|
| 2824 |
+
"step": 4690
|
| 2825 |
+
},
|
| 2826 |
+
{
|
| 2827 |
+
"grad_norm": 0.2442196160554886,
|
| 2828 |
+
"learning_rate": 9.692338225538333e-05,
|
| 2829 |
+
"loss": 0.0168,
|
| 2830 |
+
"step": 4700
|
| 2831 |
+
},
|
| 2832 |
+
{
|
| 2833 |
+
"grad_norm": 0.2947649657726288,
|
| 2834 |
+
"learning_rate": 9.690431861159241e-05,
|
| 2835 |
+
"loss": 0.0169,
|
| 2836 |
+
"step": 4710
|
| 2837 |
+
},
|
| 2838 |
+
{
|
| 2839 |
+
"grad_norm": 0.336140900850296,
|
| 2840 |
+
"learning_rate": 9.688519797463161e-05,
|
| 2841 |
+
"loss": 0.0184,
|
| 2842 |
+
"step": 4720
|
| 2843 |
+
},
|
| 2844 |
+
{
|
| 2845 |
+
"grad_norm": 0.39535579085350037,
|
| 2846 |
+
"learning_rate": 9.686602036773426e-05,
|
| 2847 |
+
"loss": 0.018,
|
| 2848 |
+
"step": 4730
|
| 2849 |
+
},
|
| 2850 |
+
{
|
| 2851 |
+
"grad_norm": 0.28196772933006287,
|
| 2852 |
+
"learning_rate": 9.684678581420302e-05,
|
| 2853 |
+
"loss": 0.0175,
|
| 2854 |
+
"step": 4740
|
| 2855 |
+
},
|
| 2856 |
+
{
|
| 2857 |
+
"grad_norm": 0.33065393567085266,
|
| 2858 |
+
"learning_rate": 9.682749433740962e-05,
|
| 2859 |
+
"loss": 0.0201,
|
| 2860 |
+
"step": 4750
|
| 2861 |
+
},
|
| 2862 |
+
{
|
| 2863 |
+
"grad_norm": 0.298825740814209,
|
| 2864 |
+
"learning_rate": 9.680814596079507e-05,
|
| 2865 |
+
"loss": 0.0194,
|
| 2866 |
+
"step": 4760
|
| 2867 |
+
},
|
| 2868 |
+
{
|
| 2869 |
+
"grad_norm": 0.28574132919311523,
|
| 2870 |
+
"learning_rate": 9.678874070786945e-05,
|
| 2871 |
+
"loss": 0.018,
|
| 2872 |
+
"step": 4770
|
| 2873 |
+
},
|
| 2874 |
+
{
|
| 2875 |
+
"grad_norm": 0.2778089940547943,
|
| 2876 |
+
"learning_rate": 9.676927860221199e-05,
|
| 2877 |
+
"loss": 0.0191,
|
| 2878 |
+
"step": 4780
|
| 2879 |
+
},
|
| 2880 |
+
{
|
| 2881 |
+
"grad_norm": 0.2691532373428345,
|
| 2882 |
+
"learning_rate": 9.674975966747097e-05,
|
| 2883 |
+
"loss": 0.018,
|
| 2884 |
+
"step": 4790
|
| 2885 |
+
},
|
| 2886 |
+
{
|
| 2887 |
+
"grad_norm": 0.28231796622276306,
|
| 2888 |
+
"learning_rate": 9.673018392736374e-05,
|
| 2889 |
+
"loss": 0.0161,
|
| 2890 |
+
"step": 4800
|
| 2891 |
+
},
|
| 2892 |
+
{
|
| 2893 |
+
"grad_norm": 0.3621669113636017,
|
| 2894 |
+
"learning_rate": 9.671055140567667e-05,
|
| 2895 |
+
"loss": 0.0189,
|
| 2896 |
+
"step": 4810
|
| 2897 |
+
},
|
| 2898 |
+
{
|
| 2899 |
+
"grad_norm": 0.3124409019947052,
|
| 2900 |
+
"learning_rate": 9.669086212626511e-05,
|
| 2901 |
+
"loss": 0.0182,
|
| 2902 |
+
"step": 4820
|
| 2903 |
+
},
|
| 2904 |
+
{
|
| 2905 |
+
"grad_norm": 0.30056077241897583,
|
| 2906 |
+
"learning_rate": 9.667111611305341e-05,
|
| 2907 |
+
"loss": 0.0183,
|
| 2908 |
+
"step": 4830
|
| 2909 |
+
},
|
| 2910 |
+
{
|
| 2911 |
+
"grad_norm": 0.3483680486679077,
|
| 2912 |
+
"learning_rate": 9.665131339003486e-05,
|
| 2913 |
+
"loss": 0.018,
|
| 2914 |
+
"step": 4840
|
| 2915 |
+
},
|
| 2916 |
+
{
|
| 2917 |
+
"grad_norm": 0.2676101624965668,
|
| 2918 |
+
"learning_rate": 9.663145398127158e-05,
|
| 2919 |
+
"loss": 0.0156,
|
| 2920 |
+
"step": 4850
|
| 2921 |
+
},
|
| 2922 |
+
{
|
| 2923 |
+
"grad_norm": 0.23675771057605743,
|
| 2924 |
+
"learning_rate": 9.661153791089467e-05,
|
| 2925 |
+
"loss": 0.0185,
|
| 2926 |
+
"step": 4860
|
| 2927 |
+
},
|
| 2928 |
+
{
|
| 2929 |
+
"grad_norm": 0.2168475240468979,
|
| 2930 |
+
"learning_rate": 9.659156520310402e-05,
|
| 2931 |
+
"loss": 0.0173,
|
| 2932 |
+
"step": 4870
|
| 2933 |
+
},
|
| 2934 |
+
{
|
| 2935 |
+
"grad_norm": 0.2815901041030884,
|
| 2936 |
+
"learning_rate": 9.657153588216835e-05,
|
| 2937 |
+
"loss": 0.0161,
|
| 2938 |
+
"step": 4880
|
| 2939 |
+
},
|
| 2940 |
+
{
|
| 2941 |
+
"grad_norm": 0.2556694746017456,
|
| 2942 |
+
"learning_rate": 9.655144997242516e-05,
|
| 2943 |
+
"loss": 0.0156,
|
| 2944 |
+
"step": 4890
|
| 2945 |
+
},
|
| 2946 |
+
{
|
| 2947 |
+
"grad_norm": 0.21819503605365753,
|
| 2948 |
+
"learning_rate": 9.653130749828075e-05,
|
| 2949 |
+
"loss": 0.0157,
|
| 2950 |
+
"step": 4900
|
| 2951 |
+
},
|
| 2952 |
+
{
|
| 2953 |
+
"grad_norm": 0.2848703861236572,
|
| 2954 |
+
"learning_rate": 9.65111084842101e-05,
|
| 2955 |
+
"loss": 0.0164,
|
| 2956 |
+
"step": 4910
|
| 2957 |
+
},
|
| 2958 |
+
{
|
| 2959 |
+
"grad_norm": 0.3069203794002533,
|
| 2960 |
+
"learning_rate": 9.649085295475695e-05,
|
| 2961 |
+
"loss": 0.0201,
|
| 2962 |
+
"step": 4920
|
| 2963 |
+
},
|
| 2964 |
+
{
|
| 2965 |
+
"grad_norm": 0.29833701252937317,
|
| 2966 |
+
"learning_rate": 9.647054093453365e-05,
|
| 2967 |
+
"loss": 0.0199,
|
| 2968 |
+
"step": 4930
|
| 2969 |
+
},
|
| 2970 |
+
{
|
| 2971 |
+
"grad_norm": 0.20307645201683044,
|
| 2972 |
+
"learning_rate": 9.645017244822123e-05,
|
| 2973 |
+
"loss": 0.0167,
|
| 2974 |
+
"step": 4940
|
| 2975 |
+
},
|
| 2976 |
+
{
|
| 2977 |
+
"grad_norm": 0.21525900065898895,
|
| 2978 |
+
"learning_rate": 9.642974752056931e-05,
|
| 2979 |
+
"loss": 0.018,
|
| 2980 |
+
"step": 4950
|
| 2981 |
+
},
|
| 2982 |
+
{
|
| 2983 |
+
"grad_norm": 0.33719441294670105,
|
| 2984 |
+
"learning_rate": 9.640926617639613e-05,
|
| 2985 |
+
"loss": 0.019,
|
| 2986 |
+
"step": 4960
|
| 2987 |
+
},
|
| 2988 |
+
{
|
| 2989 |
+
"grad_norm": 0.3345821797847748,
|
| 2990 |
+
"learning_rate": 9.638872844058843e-05,
|
| 2991 |
+
"loss": 0.0174,
|
| 2992 |
+
"step": 4970
|
| 2993 |
+
},
|
| 2994 |
+
{
|
| 2995 |
+
"grad_norm": 0.29934078454971313,
|
| 2996 |
+
"learning_rate": 9.63681343381015e-05,
|
| 2997 |
+
"loss": 0.019,
|
| 2998 |
+
"step": 4980
|
| 2999 |
+
},
|
| 3000 |
+
{
|
| 3001 |
+
"grad_norm": 0.31826502084732056,
|
| 3002 |
+
"learning_rate": 9.634748389395914e-05,
|
| 3003 |
+
"loss": 0.017,
|
| 3004 |
+
"step": 4990
|
| 3005 |
+
},
|
| 3006 |
+
{
|
| 3007 |
+
"grad_norm": 0.3035266101360321,
|
| 3008 |
+
"learning_rate": 9.632677713325353e-05,
|
| 3009 |
+
"loss": 0.018,
|
| 3010 |
+
"step": 5000
|
| 3011 |
+
}
|
| 3012 |
+
],
|
| 3013 |
+
"logging_steps": 10,
|
| 3014 |
+
"max_steps": 30000,
|
| 3015 |
+
"num_input_tokens_seen": 0,
|
| 3016 |
+
"num_train_epochs": 9223372036854775807,
|
| 3017 |
+
"save_steps": 2500,
|
| 3018 |
+
"stateful_callbacks": {
|
| 3019 |
+
"TrainerControl": {
|
| 3020 |
+
"args": {
|
| 3021 |
+
"should_epoch_stop": false,
|
| 3022 |
+
"should_evaluate": false,
|
| 3023 |
+
"should_log": false,
|
| 3024 |
+
"should_save": true,
|
| 3025 |
+
"should_training_stop": false
|
| 3026 |
+
},
|
| 3027 |
+
"attributes": {}
|
| 3028 |
+
}
|
| 3029 |
+
},
|
| 3030 |
+
"total_flos": 0.0,
|
| 3031 |
+
"train_batch_size": 64,
|
| 3032 |
+
"trial_name": null,
|
| 3033 |
+
"trial_params": null
|
| 3034 |
+
}
|
16d_30k/checkpoint-5000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9e929ad874fd372cf09ea10ff49872f55bb6b39bb208fe0f9ef27cb89015e7be
|
| 3 |
+
size 7761
|
16d_30k/checkpoint-5000/wandb_config.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"project": "screwing_hil", "run_id": "groot_16d_30k"}
|