test-4-checkpoint-20000
Browse files- test-4-checkpoint-20000/config.json +133 -0
- test-4-checkpoint-20000/dataset_statistics.json +253 -0
- test-4-checkpoint-20000/model.safetensors +3 -0
- test-4-checkpoint-20000/optimizer.pt +3 -0
- test-4-checkpoint-20000/rng_state_0.pth +3 -0
- test-4-checkpoint-20000/rng_state_1.pth +3 -0
- test-4-checkpoint-20000/rng_state_2.pth +3 -0
- test-4-checkpoint-20000/rng_state_3.pth +3 -0
- test-4-checkpoint-20000/rng_state_4.pth +3 -0
- test-4-checkpoint-20000/rng_state_5.pth +3 -0
- test-4-checkpoint-20000/rng_state_6.pth +3 -0
- test-4-checkpoint-20000/rng_state_7.pth +3 -0
- test-4-checkpoint-20000/scheduler.pt +3 -0
- test-4-checkpoint-20000/trainer_state.json +1433 -0
test-4-checkpoint-20000/config.json
ADDED
|
@@ -0,0 +1,133 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"batch_size": 64,
|
| 3 |
+
"buffer_size": 64,
|
| 4 |
+
"from_pretrained": "oxe-g2-checkpoint-300000",
|
| 5 |
+
"frozen_keys": [
|
| 6 |
+
"*hf_model*"
|
| 7 |
+
],
|
| 8 |
+
"model": {
|
| 9 |
+
"heads": {
|
| 10 |
+
"action": {
|
| 11 |
+
"args": [],
|
| 12 |
+
"kwargs": {
|
| 13 |
+
"action_dim": 7,
|
| 14 |
+
"action_horizon": 4,
|
| 15 |
+
"dropout_rate": 0.0,
|
| 16 |
+
"n_diffusion_samples": 1,
|
| 17 |
+
"readout_key": "readout_action",
|
| 18 |
+
"token_embedding_size": 768,
|
| 19 |
+
"use_map": false
|
| 20 |
+
},
|
| 21 |
+
"module": "octo.model.components.action_heads",
|
| 22 |
+
"name": "DiffusionActionHead"
|
| 23 |
+
}
|
| 24 |
+
},
|
| 25 |
+
"max_horizon": 10,
|
| 26 |
+
"observation_tokenizers": {
|
| 27 |
+
"primary": {
|
| 28 |
+
"args": [],
|
| 29 |
+
"kwargs": {
|
| 30 |
+
"encoder": {
|
| 31 |
+
"args": [],
|
| 32 |
+
"kwargs": {
|
| 33 |
+
"in_features": 6
|
| 34 |
+
},
|
| 35 |
+
"module": "octo.model.components.vit_encoders",
|
| 36 |
+
"name": "SmallStem16"
|
| 37 |
+
},
|
| 38 |
+
"obs_stack_keys": [
|
| 39 |
+
"image_primary"
|
| 40 |
+
],
|
| 41 |
+
"task_stack_keys": [
|
| 42 |
+
"image_primary"
|
| 43 |
+
]
|
| 44 |
+
},
|
| 45 |
+
"module": "octo.model.components.tokenizers",
|
| 46 |
+
"name": "ImageTokenizer"
|
| 47 |
+
}
|
| 48 |
+
},
|
| 49 |
+
"readouts": {
|
| 50 |
+
"action": 1
|
| 51 |
+
},
|
| 52 |
+
"repeat_task_tokens": true,
|
| 53 |
+
"task_tokenizers": {
|
| 54 |
+
"language": {
|
| 55 |
+
"args": [],
|
| 56 |
+
"kwargs": {
|
| 57 |
+
"encoder": "t5-base",
|
| 58 |
+
"finetune_encoder": false
|
| 59 |
+
},
|
| 60 |
+
"module": "octo.model.components.tokenizers",
|
| 61 |
+
"name": "LanguageTokenizer"
|
| 62 |
+
}
|
| 63 |
+
},
|
| 64 |
+
"token_embedding_size": 768,
|
| 65 |
+
"transformer_kwargs": {
|
| 66 |
+
"add_position_embedding": false,
|
| 67 |
+
"attention_dropout_rate": 0.0,
|
| 68 |
+
"dropout_rate": 0.0,
|
| 69 |
+
"mlp_dim": 3072,
|
| 70 |
+
"num_attention_heads": 12,
|
| 71 |
+
"num_layers": 12
|
| 72 |
+
},
|
| 73 |
+
"use_correct_attention": true
|
| 74 |
+
},
|
| 75 |
+
"normalize_method": "sign",
|
| 76 |
+
"obs_token_nums": {
|
| 77 |
+
"primary": 256
|
| 78 |
+
},
|
| 79 |
+
"sampler_num_samples": 2560000,
|
| 80 |
+
"seed": 2,
|
| 81 |
+
"subsample_length": 99999,
|
| 82 |
+
"text_processor": {
|
| 83 |
+
"args": [],
|
| 84 |
+
"kwargs": {
|
| 85 |
+
"encode_with_model": false,
|
| 86 |
+
"tokenizer_kwargs": {
|
| 87 |
+
"max_length": 16,
|
| 88 |
+
"padding": "max_length",
|
| 89 |
+
"return_tensors": "np",
|
| 90 |
+
"truncation": true
|
| 91 |
+
},
|
| 92 |
+
"tokenizer_name": "t5-base"
|
| 93 |
+
},
|
| 94 |
+
"module": "octo.components.text_processing",
|
| 95 |
+
"name": "HFTokenizer"
|
| 96 |
+
},
|
| 97 |
+
"training_arguments": {
|
| 98 |
+
"bf16": true,
|
| 99 |
+
"dataloader_pin_memory": true,
|
| 100 |
+
"ddp_find_unused_parameters": true,
|
| 101 |
+
"eval_steps": 5000,
|
| 102 |
+
"gradient_accumulation_steps": 4,
|
| 103 |
+
"learning_rate": 5e-05,
|
| 104 |
+
"logging_nan_inf_filter": false,
|
| 105 |
+
"logging_steps": 100,
|
| 106 |
+
"lr_scheduler_kwargs": {
|
| 107 |
+
"decay_type": "cosine",
|
| 108 |
+
"min_lr_ratio": 0.05,
|
| 109 |
+
"num_decay_steps": 298000
|
| 110 |
+
},
|
| 111 |
+
"lr_scheduler_type": "constant",
|
| 112 |
+
"max_grad_norm": 1.0,
|
| 113 |
+
"max_steps": 300000,
|
| 114 |
+
"optim": "adamw_torch_fused",
|
| 115 |
+
"per_device_eval_batch_size": 1,
|
| 116 |
+
"per_device_train_batch_size": 1,
|
| 117 |
+
"report_to": "wandb",
|
| 118 |
+
"save_safetensors": true,
|
| 119 |
+
"save_steps": 500,
|
| 120 |
+
"torch_compile": true,
|
| 121 |
+
"warmup_steps": 2000,
|
| 122 |
+
"weight_decay": 0.1
|
| 123 |
+
},
|
| 124 |
+
"training_keys": [
|
| 125 |
+
"*action*",
|
| 126 |
+
"*task_projections*",
|
| 127 |
+
"*obs_projections*",
|
| 128 |
+
"*readout_embeddings*",
|
| 129 |
+
"*task_pos_embeddings*",
|
| 130 |
+
"*obs_pos_embeddings*",
|
| 131 |
+
"*observation_tokenizers*"
|
| 132 |
+
]
|
| 133 |
+
}
|
test-4-checkpoint-20000/dataset_statistics.json
ADDED
|
@@ -0,0 +1,253 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"observation.state": {
|
| 4 |
+
"min": [
|
| 5 |
+
-0.6913691163063049,
|
| 6 |
+
-0.25204867124557495,
|
| 7 |
+
-0.22021731734275818,
|
| 8 |
+
-3.140528440475464,
|
| 9 |
+
-1.213579535484314,
|
| 10 |
+
-3.1340115070343018,
|
| 11 |
+
0.0
|
| 12 |
+
],
|
| 13 |
+
"max": [
|
| 14 |
+
-0.26134854555130005,
|
| 15 |
+
0.40225377678871155,
|
| 16 |
+
0.4610653519630432,
|
| 17 |
+
3.1412761211395264,
|
| 18 |
+
0.644866406917572,
|
| 19 |
+
3.1401240825653076,
|
| 20 |
+
1.0
|
| 21 |
+
],
|
| 22 |
+
"mean": [
|
| 23 |
+
-0.4765559839921994,
|
| 24 |
+
0.03277433536791847,
|
| 25 |
+
0.01813845893630427,
|
| 26 |
+
1.4840563985447381,
|
| 27 |
+
-0.1872646863000659,
|
| 28 |
+
-1.201557804914303,
|
| 29 |
+
0.13328598546378098
|
| 30 |
+
],
|
| 31 |
+
"std": [
|
| 32 |
+
0.09873974979292631,
|
| 33 |
+
0.12291375325367411,
|
| 34 |
+
0.13182701724212956,
|
| 35 |
+
2.530463134923631,
|
| 36 |
+
0.39532953579234853,
|
| 37 |
+
1.3639962652103377,
|
| 38 |
+
0.33988356830665484
|
| 39 |
+
],
|
| 40 |
+
"count": [
|
| 41 |
+
12672
|
| 42 |
+
]
|
| 43 |
+
},
|
| 44 |
+
"task_index": {
|
| 45 |
+
"min": [
|
| 46 |
+
0
|
| 47 |
+
],
|
| 48 |
+
"max": [
|
| 49 |
+
3
|
| 50 |
+
],
|
| 51 |
+
"mean": [
|
| 52 |
+
1.6840277777777777
|
| 53 |
+
],
|
| 54 |
+
"std": [
|
| 55 |
+
1.1113384592218372
|
| 56 |
+
],
|
| 57 |
+
"count": [
|
| 58 |
+
12672
|
| 59 |
+
]
|
| 60 |
+
},
|
| 61 |
+
"episode_index": {
|
| 62 |
+
"min": [
|
| 63 |
+
0
|
| 64 |
+
],
|
| 65 |
+
"max": [
|
| 66 |
+
128
|
| 67 |
+
],
|
| 68 |
+
"mean": [
|
| 69 |
+
63.14946338383838
|
| 70 |
+
],
|
| 71 |
+
"std": [
|
| 72 |
+
35.1467891774593
|
| 73 |
+
],
|
| 74 |
+
"count": [
|
| 75 |
+
12672
|
| 76 |
+
]
|
| 77 |
+
},
|
| 78 |
+
"observation.images.main_camera": {
|
| 79 |
+
"min": [
|
| 80 |
+
[
|
| 81 |
+
[
|
| 82 |
+
0.0
|
| 83 |
+
]
|
| 84 |
+
],
|
| 85 |
+
[
|
| 86 |
+
[
|
| 87 |
+
0.0
|
| 88 |
+
]
|
| 89 |
+
],
|
| 90 |
+
[
|
| 91 |
+
[
|
| 92 |
+
0.0
|
| 93 |
+
]
|
| 94 |
+
]
|
| 95 |
+
],
|
| 96 |
+
"max": [
|
| 97 |
+
[
|
| 98 |
+
[
|
| 99 |
+
1.0
|
| 100 |
+
]
|
| 101 |
+
],
|
| 102 |
+
[
|
| 103 |
+
[
|
| 104 |
+
1.0
|
| 105 |
+
]
|
| 106 |
+
],
|
| 107 |
+
[
|
| 108 |
+
[
|
| 109 |
+
1.0
|
| 110 |
+
]
|
| 111 |
+
]
|
| 112 |
+
],
|
| 113 |
+
"mean": [
|
| 114 |
+
[
|
| 115 |
+
[
|
| 116 |
+
0.56009308397514
|
| 117 |
+
]
|
| 118 |
+
],
|
| 119 |
+
[
|
| 120 |
+
[
|
| 121 |
+
0.5088963212208076
|
| 122 |
+
]
|
| 123 |
+
],
|
| 124 |
+
[
|
| 125 |
+
[
|
| 126 |
+
0.47626117961949455
|
| 127 |
+
]
|
| 128 |
+
]
|
| 129 |
+
],
|
| 130 |
+
"std": [
|
| 131 |
+
[
|
| 132 |
+
[
|
| 133 |
+
0.27688289864943944
|
| 134 |
+
]
|
| 135 |
+
],
|
| 136 |
+
[
|
| 137 |
+
[
|
| 138 |
+
0.26883972589038224
|
| 139 |
+
]
|
| 140 |
+
],
|
| 141 |
+
[
|
| 142 |
+
[
|
| 143 |
+
0.29031452356480364
|
| 144 |
+
]
|
| 145 |
+
]
|
| 146 |
+
],
|
| 147 |
+
"count": [
|
| 148 |
+
11006
|
| 149 |
+
]
|
| 150 |
+
},
|
| 151 |
+
"action": {
|
| 152 |
+
"min": [
|
| 153 |
+
-0.009999999776482582,
|
| 154 |
+
-0.009999999776482582,
|
| 155 |
+
-0.009999999776482582,
|
| 156 |
+
-0.06283185631036758,
|
| 157 |
+
-0.06283185631036758,
|
| 158 |
+
-0.06283185631036758,
|
| 159 |
+
0.0
|
| 160 |
+
],
|
| 161 |
+
"max": [
|
| 162 |
+
0.009999999776482582,
|
| 163 |
+
0.009999999776482582,
|
| 164 |
+
0.009999999776482582,
|
| 165 |
+
0.06283185631036758,
|
| 166 |
+
0.06283185631036758,
|
| 167 |
+
0.06283185631036758,
|
| 168 |
+
1.0
|
| 169 |
+
],
|
| 170 |
+
"mean": [
|
| 171 |
+
-0.0010779668760702325,
|
| 172 |
+
-0.000870422678087375,
|
| 173 |
+
-0.0009524932320082003,
|
| 174 |
+
-0.00011404143052376722,
|
| 175 |
+
0.001646163140721359,
|
| 176 |
+
0.0006842485724881945,
|
| 177 |
+
0.14094065714126566
|
| 178 |
+
],
|
| 179 |
+
"std": [
|
| 180 |
+
0.0050928476793778805,
|
| 181 |
+
0.0052109778002382215,
|
| 182 |
+
0.0062727072406226275,
|
| 183 |
+
0.01305372613557191,
|
| 184 |
+
0.014969692486291531,
|
| 185 |
+
0.021519857205295197,
|
| 186 |
+
0.3479603303557627
|
| 187 |
+
],
|
| 188 |
+
"count": [
|
| 189 |
+
12672
|
| 190 |
+
],
|
| 191 |
+
"mask": [
|
| 192 |
+
true,
|
| 193 |
+
true,
|
| 194 |
+
true,
|
| 195 |
+
true,
|
| 196 |
+
true,
|
| 197 |
+
true,
|
| 198 |
+
false
|
| 199 |
+
]
|
| 200 |
+
},
|
| 201 |
+
"timestamp": {
|
| 202 |
+
"min": [
|
| 203 |
+
0.0
|
| 204 |
+
],
|
| 205 |
+
"max": [
|
| 206 |
+
22.5
|
| 207 |
+
],
|
| 208 |
+
"mean": [
|
| 209 |
+
5.516895517676768
|
| 210 |
+
],
|
| 211 |
+
"std": [
|
| 212 |
+
3.9518586666950704
|
| 213 |
+
],
|
| 214 |
+
"count": [
|
| 215 |
+
12672
|
| 216 |
+
]
|
| 217 |
+
},
|
| 218 |
+
"index": {
|
| 219 |
+
"min": [
|
| 220 |
+
0
|
| 221 |
+
],
|
| 222 |
+
"max": [
|
| 223 |
+
12671
|
| 224 |
+
],
|
| 225 |
+
"mean": [
|
| 226 |
+
6335.5
|
| 227 |
+
],
|
| 228 |
+
"std": [
|
| 229 |
+
3658.0912941951938
|
| 230 |
+
],
|
| 231 |
+
"count": [
|
| 232 |
+
12672
|
| 233 |
+
]
|
| 234 |
+
},
|
| 235 |
+
"frame_index": {
|
| 236 |
+
"min": [
|
| 237 |
+
0
|
| 238 |
+
],
|
| 239 |
+
"max": [
|
| 240 |
+
225
|
| 241 |
+
],
|
| 242 |
+
"mean": [
|
| 243 |
+
55.168955176767675
|
| 244 |
+
],
|
| 245 |
+
"std": [
|
| 246 |
+
39.5185866669507
|
| 247 |
+
],
|
| 248 |
+
"count": [
|
| 249 |
+
12672
|
| 250 |
+
]
|
| 251 |
+
}
|
| 252 |
+
}
|
| 253 |
+
]
|
test-4-checkpoint-20000/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:753949daf5c03b0b672f31f00aaceb797e4844c91b85479e603d8ae41b86759c
|
| 3 |
+
size 401078544
|
test-4-checkpoint-20000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:33ff289d68afb68096a2f7dfde8ab5a4bc2f67058f88158b2c33b86a345241e7
|
| 3 |
+
size 26356666
|
test-4-checkpoint-20000/rng_state_0.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:be008c18eb98044633d0664090f84604f0f1b0215d5cf9d9b37582c3bcf01389
|
| 3 |
+
size 15984
|
test-4-checkpoint-20000/rng_state_1.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:44408f37a9991e9c49c1a0d2b48e814078f97a4b017faaf5701a931f4cc61c47
|
| 3 |
+
size 15984
|
test-4-checkpoint-20000/rng_state_2.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b93675ed0e6963b9f94cec16e10771d9b77a3b0a6a8ed88d0679ff67f07f361b
|
| 3 |
+
size 15984
|
test-4-checkpoint-20000/rng_state_3.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:66a45184112bb15de265c44756224eb9235f8dfd0b57f64f2f520bc837d8fc92
|
| 3 |
+
size 15984
|
test-4-checkpoint-20000/rng_state_4.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:793d24d4aa67f33ffdc10732b9fcd0150457f620d5f50d28f166568df8c7fa2a
|
| 3 |
+
size 15984
|
test-4-checkpoint-20000/rng_state_5.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:11176b9ecdf7aed0622ae853fa2bc29ee0caebc518521f6dc5673d99b5d0cba6
|
| 3 |
+
size 15984
|
test-4-checkpoint-20000/rng_state_6.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f58a3f2f9ab9a43d2374025b9c0d17865bfe1426babbf5e7255c167a0b7938db
|
| 3 |
+
size 15984
|
test-4-checkpoint-20000/rng_state_7.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:703c9fd454cb589c58b050bb97cbaeca8eeaee6a208fc200d1cef991e0d016d3
|
| 3 |
+
size 15984
|
test-4-checkpoint-20000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2dc3ea032367762f4c7d201a5fcaa046fb82a3934e937afb888166934d8fde85
|
| 3 |
+
size 1064
|
test-4-checkpoint-20000/trainer_state.json
ADDED
|
@@ -0,0 +1,1433 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 16.0,
|
| 5 |
+
"eval_steps": 5000,
|
| 6 |
+
"global_step": 20000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.08,
|
| 13 |
+
"grad_norm": 3.25,
|
| 14 |
+
"learning_rate": 5e-05,
|
| 15 |
+
"loss": 0.5172,
|
| 16 |
+
"step": 100
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.16,
|
| 20 |
+
"grad_norm": 3.921875,
|
| 21 |
+
"learning_rate": 5e-05,
|
| 22 |
+
"loss": 0.3445,
|
| 23 |
+
"step": 200
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.24,
|
| 27 |
+
"grad_norm": 4.09375,
|
| 28 |
+
"learning_rate": 5e-05,
|
| 29 |
+
"loss": 0.3051,
|
| 30 |
+
"step": 300
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.32,
|
| 34 |
+
"grad_norm": 3.09375,
|
| 35 |
+
"learning_rate": 5e-05,
|
| 36 |
+
"loss": 0.2824,
|
| 37 |
+
"step": 400
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.4,
|
| 41 |
+
"grad_norm": 3.09375,
|
| 42 |
+
"learning_rate": 5e-05,
|
| 43 |
+
"loss": 0.265,
|
| 44 |
+
"step": 500
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.48,
|
| 48 |
+
"grad_norm": 2.484375,
|
| 49 |
+
"learning_rate": 5e-05,
|
| 50 |
+
"loss": 0.2517,
|
| 51 |
+
"step": 600
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.56,
|
| 55 |
+
"grad_norm": 3.109375,
|
| 56 |
+
"learning_rate": 5e-05,
|
| 57 |
+
"loss": 0.2448,
|
| 58 |
+
"step": 700
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.64,
|
| 62 |
+
"grad_norm": 2.5,
|
| 63 |
+
"learning_rate": 5e-05,
|
| 64 |
+
"loss": 0.2374,
|
| 65 |
+
"step": 800
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.72,
|
| 69 |
+
"grad_norm": 2.0625,
|
| 70 |
+
"learning_rate": 5e-05,
|
| 71 |
+
"loss": 0.2327,
|
| 72 |
+
"step": 900
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.8,
|
| 76 |
+
"grad_norm": 2.765625,
|
| 77 |
+
"learning_rate": 5e-05,
|
| 78 |
+
"loss": 0.2291,
|
| 79 |
+
"step": 1000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.88,
|
| 83 |
+
"grad_norm": 2.125,
|
| 84 |
+
"learning_rate": 5e-05,
|
| 85 |
+
"loss": 0.2243,
|
| 86 |
+
"step": 1100
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.96,
|
| 90 |
+
"grad_norm": 3.03125,
|
| 91 |
+
"learning_rate": 5e-05,
|
| 92 |
+
"loss": 0.2209,
|
| 93 |
+
"step": 1200
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 1.04,
|
| 97 |
+
"grad_norm": 2.265625,
|
| 98 |
+
"learning_rate": 5e-05,
|
| 99 |
+
"loss": 0.2167,
|
| 100 |
+
"step": 1300
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 1.12,
|
| 104 |
+
"grad_norm": 2.546875,
|
| 105 |
+
"learning_rate": 5e-05,
|
| 106 |
+
"loss": 0.2142,
|
| 107 |
+
"step": 1400
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 1.2,
|
| 111 |
+
"grad_norm": 2.171875,
|
| 112 |
+
"learning_rate": 5e-05,
|
| 113 |
+
"loss": 0.2124,
|
| 114 |
+
"step": 1500
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 1.28,
|
| 118 |
+
"grad_norm": 2.609375,
|
| 119 |
+
"learning_rate": 5e-05,
|
| 120 |
+
"loss": 0.2116,
|
| 121 |
+
"step": 1600
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 1.3599999999999999,
|
| 125 |
+
"grad_norm": 2.078125,
|
| 126 |
+
"learning_rate": 5e-05,
|
| 127 |
+
"loss": 0.2097,
|
| 128 |
+
"step": 1700
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 1.44,
|
| 132 |
+
"grad_norm": 2.296875,
|
| 133 |
+
"learning_rate": 5e-05,
|
| 134 |
+
"loss": 0.2069,
|
| 135 |
+
"step": 1800
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 1.52,
|
| 139 |
+
"grad_norm": 2.1875,
|
| 140 |
+
"learning_rate": 5e-05,
|
| 141 |
+
"loss": 0.2037,
|
| 142 |
+
"step": 1900
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 1.6,
|
| 146 |
+
"grad_norm": 2.21875,
|
| 147 |
+
"learning_rate": 5e-05,
|
| 148 |
+
"loss": 0.2025,
|
| 149 |
+
"step": 2000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 1.6800000000000002,
|
| 153 |
+
"grad_norm": 2.640625,
|
| 154 |
+
"learning_rate": 5e-05,
|
| 155 |
+
"loss": 0.2019,
|
| 156 |
+
"step": 2100
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 1.76,
|
| 160 |
+
"grad_norm": 3.015625,
|
| 161 |
+
"learning_rate": 5e-05,
|
| 162 |
+
"loss": 0.1997,
|
| 163 |
+
"step": 2200
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 1.8399999999999999,
|
| 167 |
+
"grad_norm": 2.515625,
|
| 168 |
+
"learning_rate": 5e-05,
|
| 169 |
+
"loss": 0.1987,
|
| 170 |
+
"step": 2300
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 1.92,
|
| 174 |
+
"grad_norm": 2.390625,
|
| 175 |
+
"learning_rate": 5e-05,
|
| 176 |
+
"loss": 0.1974,
|
| 177 |
+
"step": 2400
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 2.0,
|
| 181 |
+
"grad_norm": 1.9921875,
|
| 182 |
+
"learning_rate": 5e-05,
|
| 183 |
+
"loss": 0.1978,
|
| 184 |
+
"step": 2500
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 2.08,
|
| 188 |
+
"grad_norm": 2.4375,
|
| 189 |
+
"learning_rate": 5e-05,
|
| 190 |
+
"loss": 0.1949,
|
| 191 |
+
"step": 2600
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 2.16,
|
| 195 |
+
"grad_norm": 2.453125,
|
| 196 |
+
"learning_rate": 5e-05,
|
| 197 |
+
"loss": 0.1928,
|
| 198 |
+
"step": 2700
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 2.24,
|
| 202 |
+
"grad_norm": 3.1875,
|
| 203 |
+
"learning_rate": 5e-05,
|
| 204 |
+
"loss": 0.1935,
|
| 205 |
+
"step": 2800
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 2.32,
|
| 209 |
+
"grad_norm": 2.078125,
|
| 210 |
+
"learning_rate": 5e-05,
|
| 211 |
+
"loss": 0.1912,
|
| 212 |
+
"step": 2900
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 2.4,
|
| 216 |
+
"grad_norm": 2.0,
|
| 217 |
+
"learning_rate": 5e-05,
|
| 218 |
+
"loss": 0.1898,
|
| 219 |
+
"step": 3000
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 2.48,
|
| 223 |
+
"grad_norm": 2.234375,
|
| 224 |
+
"learning_rate": 5e-05,
|
| 225 |
+
"loss": 0.1881,
|
| 226 |
+
"step": 3100
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 2.56,
|
| 230 |
+
"grad_norm": 2.4375,
|
| 231 |
+
"learning_rate": 5e-05,
|
| 232 |
+
"loss": 0.1862,
|
| 233 |
+
"step": 3200
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 2.64,
|
| 237 |
+
"grad_norm": 1.9921875,
|
| 238 |
+
"learning_rate": 5e-05,
|
| 239 |
+
"loss": 0.1871,
|
| 240 |
+
"step": 3300
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 2.7199999999999998,
|
| 244 |
+
"grad_norm": 2.1875,
|
| 245 |
+
"learning_rate": 5e-05,
|
| 246 |
+
"loss": 0.1853,
|
| 247 |
+
"step": 3400
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 2.8,
|
| 251 |
+
"grad_norm": 2.34375,
|
| 252 |
+
"learning_rate": 5e-05,
|
| 253 |
+
"loss": 0.185,
|
| 254 |
+
"step": 3500
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 2.88,
|
| 258 |
+
"grad_norm": 2.078125,
|
| 259 |
+
"learning_rate": 5e-05,
|
| 260 |
+
"loss": 0.1853,
|
| 261 |
+
"step": 3600
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 2.96,
|
| 265 |
+
"grad_norm": 2.015625,
|
| 266 |
+
"learning_rate": 5e-05,
|
| 267 |
+
"loss": 0.1849,
|
| 268 |
+
"step": 3700
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 3.04,
|
| 272 |
+
"grad_norm": 2.8125,
|
| 273 |
+
"learning_rate": 5e-05,
|
| 274 |
+
"loss": 0.1846,
|
| 275 |
+
"step": 3800
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 3.12,
|
| 279 |
+
"grad_norm": 2.40625,
|
| 280 |
+
"learning_rate": 5e-05,
|
| 281 |
+
"loss": 0.1833,
|
| 282 |
+
"step": 3900
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 3.2,
|
| 286 |
+
"grad_norm": 2.140625,
|
| 287 |
+
"learning_rate": 5e-05,
|
| 288 |
+
"loss": 0.1828,
|
| 289 |
+
"step": 4000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 3.2800000000000002,
|
| 293 |
+
"grad_norm": 2.140625,
|
| 294 |
+
"learning_rate": 5e-05,
|
| 295 |
+
"loss": 0.1821,
|
| 296 |
+
"step": 4100
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 3.36,
|
| 300 |
+
"grad_norm": 2.203125,
|
| 301 |
+
"learning_rate": 5e-05,
|
| 302 |
+
"loss": 0.1816,
|
| 303 |
+
"step": 4200
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 3.44,
|
| 307 |
+
"grad_norm": 2.515625,
|
| 308 |
+
"learning_rate": 5e-05,
|
| 309 |
+
"loss": 0.181,
|
| 310 |
+
"step": 4300
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 3.52,
|
| 314 |
+
"grad_norm": 1.84375,
|
| 315 |
+
"learning_rate": 5e-05,
|
| 316 |
+
"loss": 0.1803,
|
| 317 |
+
"step": 4400
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 3.6,
|
| 321 |
+
"grad_norm": 2.046875,
|
| 322 |
+
"learning_rate": 5e-05,
|
| 323 |
+
"loss": 0.1798,
|
| 324 |
+
"step": 4500
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 3.68,
|
| 328 |
+
"grad_norm": 1.9296875,
|
| 329 |
+
"learning_rate": 5e-05,
|
| 330 |
+
"loss": 0.1778,
|
| 331 |
+
"step": 4600
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 3.76,
|
| 335 |
+
"grad_norm": 2.421875,
|
| 336 |
+
"learning_rate": 5e-05,
|
| 337 |
+
"loss": 0.1782,
|
| 338 |
+
"step": 4700
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 3.84,
|
| 342 |
+
"grad_norm": 2.25,
|
| 343 |
+
"learning_rate": 5e-05,
|
| 344 |
+
"loss": 0.1755,
|
| 345 |
+
"step": 4800
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 3.92,
|
| 349 |
+
"grad_norm": 2.296875,
|
| 350 |
+
"learning_rate": 5e-05,
|
| 351 |
+
"loss": 0.1771,
|
| 352 |
+
"step": 4900
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 4.0,
|
| 356 |
+
"grad_norm": 1.8203125,
|
| 357 |
+
"learning_rate": 5e-05,
|
| 358 |
+
"loss": 0.1751,
|
| 359 |
+
"step": 5000
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 4.08,
|
| 363 |
+
"grad_norm": 1.828125,
|
| 364 |
+
"learning_rate": 5e-05,
|
| 365 |
+
"loss": 0.1751,
|
| 366 |
+
"step": 5100
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 4.16,
|
| 370 |
+
"grad_norm": 2.515625,
|
| 371 |
+
"learning_rate": 5e-05,
|
| 372 |
+
"loss": 0.1753,
|
| 373 |
+
"step": 5200
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 4.24,
|
| 377 |
+
"grad_norm": 2.546875,
|
| 378 |
+
"learning_rate": 5e-05,
|
| 379 |
+
"loss": 0.1756,
|
| 380 |
+
"step": 5300
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 4.32,
|
| 384 |
+
"grad_norm": 1.609375,
|
| 385 |
+
"learning_rate": 5e-05,
|
| 386 |
+
"loss": 0.1752,
|
| 387 |
+
"step": 5400
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 4.4,
|
| 391 |
+
"grad_norm": 2.03125,
|
| 392 |
+
"learning_rate": 5e-05,
|
| 393 |
+
"loss": 0.1743,
|
| 394 |
+
"step": 5500
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 4.48,
|
| 398 |
+
"grad_norm": 3.40625,
|
| 399 |
+
"learning_rate": 5e-05,
|
| 400 |
+
"loss": 0.175,
|
| 401 |
+
"step": 5600
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 4.5600000000000005,
|
| 405 |
+
"grad_norm": 2.21875,
|
| 406 |
+
"learning_rate": 5e-05,
|
| 407 |
+
"loss": 0.1735,
|
| 408 |
+
"step": 5700
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 4.64,
|
| 412 |
+
"grad_norm": 1.8984375,
|
| 413 |
+
"learning_rate": 5e-05,
|
| 414 |
+
"loss": 0.1721,
|
| 415 |
+
"step": 5800
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 4.72,
|
| 419 |
+
"grad_norm": 2.46875,
|
| 420 |
+
"learning_rate": 5e-05,
|
| 421 |
+
"loss": 0.1729,
|
| 422 |
+
"step": 5900
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 4.8,
|
| 426 |
+
"grad_norm": 2.3125,
|
| 427 |
+
"learning_rate": 5e-05,
|
| 428 |
+
"loss": 0.1734,
|
| 429 |
+
"step": 6000
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 4.88,
|
| 433 |
+
"grad_norm": 1.8203125,
|
| 434 |
+
"learning_rate": 5e-05,
|
| 435 |
+
"loss": 0.1721,
|
| 436 |
+
"step": 6100
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 4.96,
|
| 440 |
+
"grad_norm": 2.140625,
|
| 441 |
+
"learning_rate": 5e-05,
|
| 442 |
+
"loss": 0.1719,
|
| 443 |
+
"step": 6200
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 5.04,
|
| 447 |
+
"grad_norm": 1.921875,
|
| 448 |
+
"learning_rate": 5e-05,
|
| 449 |
+
"loss": 0.171,
|
| 450 |
+
"step": 6300
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 5.12,
|
| 454 |
+
"grad_norm": 2.6875,
|
| 455 |
+
"learning_rate": 5e-05,
|
| 456 |
+
"loss": 0.1696,
|
| 457 |
+
"step": 6400
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 5.2,
|
| 461 |
+
"grad_norm": 2.15625,
|
| 462 |
+
"learning_rate": 5e-05,
|
| 463 |
+
"loss": 0.1713,
|
| 464 |
+
"step": 6500
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 5.28,
|
| 468 |
+
"grad_norm": 2.734375,
|
| 469 |
+
"learning_rate": 5e-05,
|
| 470 |
+
"loss": 0.1705,
|
| 471 |
+
"step": 6600
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 5.36,
|
| 475 |
+
"grad_norm": 1.8359375,
|
| 476 |
+
"learning_rate": 5e-05,
|
| 477 |
+
"loss": 0.1701,
|
| 478 |
+
"step": 6700
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 5.44,
|
| 482 |
+
"grad_norm": 1.890625,
|
| 483 |
+
"learning_rate": 5e-05,
|
| 484 |
+
"loss": 0.1707,
|
| 485 |
+
"step": 6800
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 5.52,
|
| 489 |
+
"grad_norm": 2.046875,
|
| 490 |
+
"learning_rate": 5e-05,
|
| 491 |
+
"loss": 0.1692,
|
| 492 |
+
"step": 6900
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 5.6,
|
| 496 |
+
"grad_norm": 1.9609375,
|
| 497 |
+
"learning_rate": 5e-05,
|
| 498 |
+
"loss": 0.1686,
|
| 499 |
+
"step": 7000
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 5.68,
|
| 503 |
+
"grad_norm": 2.0,
|
| 504 |
+
"learning_rate": 5e-05,
|
| 505 |
+
"loss": 0.1689,
|
| 506 |
+
"step": 7100
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 5.76,
|
| 510 |
+
"grad_norm": 2.59375,
|
| 511 |
+
"learning_rate": 5e-05,
|
| 512 |
+
"loss": 0.17,
|
| 513 |
+
"step": 7200
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 5.84,
|
| 517 |
+
"grad_norm": 1.9921875,
|
| 518 |
+
"learning_rate": 5e-05,
|
| 519 |
+
"loss": 0.1692,
|
| 520 |
+
"step": 7300
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 5.92,
|
| 524 |
+
"grad_norm": 2.1875,
|
| 525 |
+
"learning_rate": 5e-05,
|
| 526 |
+
"loss": 0.1672,
|
| 527 |
+
"step": 7400
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 6.0,
|
| 531 |
+
"grad_norm": 1.8828125,
|
| 532 |
+
"learning_rate": 5e-05,
|
| 533 |
+
"loss": 0.167,
|
| 534 |
+
"step": 7500
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 6.08,
|
| 538 |
+
"grad_norm": 2.0625,
|
| 539 |
+
"learning_rate": 5e-05,
|
| 540 |
+
"loss": 0.1676,
|
| 541 |
+
"step": 7600
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 6.16,
|
| 545 |
+
"grad_norm": 2.0625,
|
| 546 |
+
"learning_rate": 5e-05,
|
| 547 |
+
"loss": 0.1681,
|
| 548 |
+
"step": 7700
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 6.24,
|
| 552 |
+
"grad_norm": 1.71875,
|
| 553 |
+
"learning_rate": 5e-05,
|
| 554 |
+
"loss": 0.1671,
|
| 555 |
+
"step": 7800
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 6.32,
|
| 559 |
+
"grad_norm": 1.875,
|
| 560 |
+
"learning_rate": 5e-05,
|
| 561 |
+
"loss": 0.1681,
|
| 562 |
+
"step": 7900
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 6.4,
|
| 566 |
+
"grad_norm": 1.90625,
|
| 567 |
+
"learning_rate": 5e-05,
|
| 568 |
+
"loss": 0.1672,
|
| 569 |
+
"step": 8000
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 6.48,
|
| 573 |
+
"grad_norm": 2.1875,
|
| 574 |
+
"learning_rate": 5e-05,
|
| 575 |
+
"loss": 0.1679,
|
| 576 |
+
"step": 8100
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 6.5600000000000005,
|
| 580 |
+
"grad_norm": 1.9609375,
|
| 581 |
+
"learning_rate": 5e-05,
|
| 582 |
+
"loss": 0.1646,
|
| 583 |
+
"step": 8200
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 6.64,
|
| 587 |
+
"grad_norm": 1.984375,
|
| 588 |
+
"learning_rate": 5e-05,
|
| 589 |
+
"loss": 0.1663,
|
| 590 |
+
"step": 8300
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 6.72,
|
| 594 |
+
"grad_norm": 2.734375,
|
| 595 |
+
"learning_rate": 5e-05,
|
| 596 |
+
"loss": 0.1661,
|
| 597 |
+
"step": 8400
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 6.8,
|
| 601 |
+
"grad_norm": 2.078125,
|
| 602 |
+
"learning_rate": 5e-05,
|
| 603 |
+
"loss": 0.1649,
|
| 604 |
+
"step": 8500
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 6.88,
|
| 608 |
+
"grad_norm": 2.28125,
|
| 609 |
+
"learning_rate": 5e-05,
|
| 610 |
+
"loss": 0.1638,
|
| 611 |
+
"step": 8600
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 6.96,
|
| 615 |
+
"grad_norm": 2.21875,
|
| 616 |
+
"learning_rate": 5e-05,
|
| 617 |
+
"loss": 0.1663,
|
| 618 |
+
"step": 8700
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 7.04,
|
| 622 |
+
"grad_norm": 2.453125,
|
| 623 |
+
"learning_rate": 5e-05,
|
| 624 |
+
"loss": 0.1656,
|
| 625 |
+
"step": 8800
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 7.12,
|
| 629 |
+
"grad_norm": 2.109375,
|
| 630 |
+
"learning_rate": 5e-05,
|
| 631 |
+
"loss": 0.1645,
|
| 632 |
+
"step": 8900
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 7.2,
|
| 636 |
+
"grad_norm": 1.6875,
|
| 637 |
+
"learning_rate": 5e-05,
|
| 638 |
+
"loss": 0.1638,
|
| 639 |
+
"step": 9000
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 7.28,
|
| 643 |
+
"grad_norm": 2.984375,
|
| 644 |
+
"learning_rate": 5e-05,
|
| 645 |
+
"loss": 0.1639,
|
| 646 |
+
"step": 9100
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 7.36,
|
| 650 |
+
"grad_norm": 2.28125,
|
| 651 |
+
"learning_rate": 5e-05,
|
| 652 |
+
"loss": 0.1641,
|
| 653 |
+
"step": 9200
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 7.44,
|
| 657 |
+
"grad_norm": 2.03125,
|
| 658 |
+
"learning_rate": 5e-05,
|
| 659 |
+
"loss": 0.1645,
|
| 660 |
+
"step": 9300
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 7.52,
|
| 664 |
+
"grad_norm": 2.0625,
|
| 665 |
+
"learning_rate": 5e-05,
|
| 666 |
+
"loss": 0.1642,
|
| 667 |
+
"step": 9400
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 7.6,
|
| 671 |
+
"grad_norm": 2.53125,
|
| 672 |
+
"learning_rate": 5e-05,
|
| 673 |
+
"loss": 0.1649,
|
| 674 |
+
"step": 9500
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 7.68,
|
| 678 |
+
"grad_norm": 2.0625,
|
| 679 |
+
"learning_rate": 5e-05,
|
| 680 |
+
"loss": 0.1622,
|
| 681 |
+
"step": 9600
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 7.76,
|
| 685 |
+
"grad_norm": 2.046875,
|
| 686 |
+
"learning_rate": 5e-05,
|
| 687 |
+
"loss": 0.1646,
|
| 688 |
+
"step": 9700
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 7.84,
|
| 692 |
+
"grad_norm": 2.140625,
|
| 693 |
+
"learning_rate": 5e-05,
|
| 694 |
+
"loss": 0.1644,
|
| 695 |
+
"step": 9800
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 7.92,
|
| 699 |
+
"grad_norm": 2.046875,
|
| 700 |
+
"learning_rate": 5e-05,
|
| 701 |
+
"loss": 0.1634,
|
| 702 |
+
"step": 9900
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 8.0,
|
| 706 |
+
"grad_norm": 2.0625,
|
| 707 |
+
"learning_rate": 5e-05,
|
| 708 |
+
"loss": 0.1629,
|
| 709 |
+
"step": 10000
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 8.08,
|
| 713 |
+
"grad_norm": 1.7265625,
|
| 714 |
+
"learning_rate": 5e-05,
|
| 715 |
+
"loss": 0.1633,
|
| 716 |
+
"step": 10100
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 8.16,
|
| 720 |
+
"grad_norm": 2.140625,
|
| 721 |
+
"learning_rate": 5e-05,
|
| 722 |
+
"loss": 0.1614,
|
| 723 |
+
"step": 10200
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 8.24,
|
| 727 |
+
"grad_norm": 2.40625,
|
| 728 |
+
"learning_rate": 5e-05,
|
| 729 |
+
"loss": 0.1629,
|
| 730 |
+
"step": 10300
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 8.32,
|
| 734 |
+
"grad_norm": 2.234375,
|
| 735 |
+
"learning_rate": 5e-05,
|
| 736 |
+
"loss": 0.1617,
|
| 737 |
+
"step": 10400
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 8.4,
|
| 741 |
+
"grad_norm": 1.78125,
|
| 742 |
+
"learning_rate": 5e-05,
|
| 743 |
+
"loss": 0.1631,
|
| 744 |
+
"step": 10500
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 8.48,
|
| 748 |
+
"grad_norm": 1.796875,
|
| 749 |
+
"learning_rate": 5e-05,
|
| 750 |
+
"loss": 0.1624,
|
| 751 |
+
"step": 10600
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 8.56,
|
| 755 |
+
"grad_norm": 2.25,
|
| 756 |
+
"learning_rate": 5e-05,
|
| 757 |
+
"loss": 0.1615,
|
| 758 |
+
"step": 10700
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 8.64,
|
| 762 |
+
"grad_norm": 2.046875,
|
| 763 |
+
"learning_rate": 5e-05,
|
| 764 |
+
"loss": 0.1635,
|
| 765 |
+
"step": 10800
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 8.72,
|
| 769 |
+
"grad_norm": 2.0,
|
| 770 |
+
"learning_rate": 5e-05,
|
| 771 |
+
"loss": 0.1614,
|
| 772 |
+
"step": 10900
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 8.8,
|
| 776 |
+
"grad_norm": 2.078125,
|
| 777 |
+
"learning_rate": 5e-05,
|
| 778 |
+
"loss": 0.1625,
|
| 779 |
+
"step": 11000
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 8.88,
|
| 783 |
+
"grad_norm": 2.109375,
|
| 784 |
+
"learning_rate": 5e-05,
|
| 785 |
+
"loss": 0.1614,
|
| 786 |
+
"step": 11100
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 8.96,
|
| 790 |
+
"grad_norm": 2.4375,
|
| 791 |
+
"learning_rate": 5e-05,
|
| 792 |
+
"loss": 0.1621,
|
| 793 |
+
"step": 11200
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 9.04,
|
| 797 |
+
"grad_norm": 1.796875,
|
| 798 |
+
"learning_rate": 5e-05,
|
| 799 |
+
"loss": 0.1622,
|
| 800 |
+
"step": 11300
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 9.12,
|
| 804 |
+
"grad_norm": 2.4375,
|
| 805 |
+
"learning_rate": 5e-05,
|
| 806 |
+
"loss": 0.1623,
|
| 807 |
+
"step": 11400
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 9.2,
|
| 811 |
+
"grad_norm": 1.9453125,
|
| 812 |
+
"learning_rate": 5e-05,
|
| 813 |
+
"loss": 0.162,
|
| 814 |
+
"step": 11500
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 9.28,
|
| 818 |
+
"grad_norm": 1.6328125,
|
| 819 |
+
"learning_rate": 5e-05,
|
| 820 |
+
"loss": 0.1619,
|
| 821 |
+
"step": 11600
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 9.36,
|
| 825 |
+
"grad_norm": 1.84375,
|
| 826 |
+
"learning_rate": 5e-05,
|
| 827 |
+
"loss": 0.1616,
|
| 828 |
+
"step": 11700
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 9.44,
|
| 832 |
+
"grad_norm": 2.09375,
|
| 833 |
+
"learning_rate": 5e-05,
|
| 834 |
+
"loss": 0.162,
|
| 835 |
+
"step": 11800
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 9.52,
|
| 839 |
+
"grad_norm": 2.125,
|
| 840 |
+
"learning_rate": 5e-05,
|
| 841 |
+
"loss": 0.1613,
|
| 842 |
+
"step": 11900
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 9.6,
|
| 846 |
+
"grad_norm": 2.59375,
|
| 847 |
+
"learning_rate": 5e-05,
|
| 848 |
+
"loss": 0.1608,
|
| 849 |
+
"step": 12000
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 9.68,
|
| 853 |
+
"grad_norm": 1.734375,
|
| 854 |
+
"learning_rate": 5e-05,
|
| 855 |
+
"loss": 0.159,
|
| 856 |
+
"step": 12100
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 9.76,
|
| 860 |
+
"grad_norm": 1.8984375,
|
| 861 |
+
"learning_rate": 5e-05,
|
| 862 |
+
"loss": 0.1615,
|
| 863 |
+
"step": 12200
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 9.84,
|
| 867 |
+
"grad_norm": 1.6875,
|
| 868 |
+
"learning_rate": 5e-05,
|
| 869 |
+
"loss": 0.1614,
|
| 870 |
+
"step": 12300
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 9.92,
|
| 874 |
+
"grad_norm": 1.8125,
|
| 875 |
+
"learning_rate": 5e-05,
|
| 876 |
+
"loss": 0.1601,
|
| 877 |
+
"step": 12400
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 10.0,
|
| 881 |
+
"grad_norm": 1.828125,
|
| 882 |
+
"learning_rate": 5e-05,
|
| 883 |
+
"loss": 0.1598,
|
| 884 |
+
"step": 12500
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 10.08,
|
| 888 |
+
"grad_norm": 2.3125,
|
| 889 |
+
"learning_rate": 5e-05,
|
| 890 |
+
"loss": 0.1615,
|
| 891 |
+
"step": 12600
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 10.16,
|
| 895 |
+
"grad_norm": 1.5078125,
|
| 896 |
+
"learning_rate": 5e-05,
|
| 897 |
+
"loss": 0.1595,
|
| 898 |
+
"step": 12700
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 10.24,
|
| 902 |
+
"grad_norm": 2.453125,
|
| 903 |
+
"learning_rate": 5e-05,
|
| 904 |
+
"loss": 0.1603,
|
| 905 |
+
"step": 12800
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 10.32,
|
| 909 |
+
"grad_norm": 1.5390625,
|
| 910 |
+
"learning_rate": 5e-05,
|
| 911 |
+
"loss": 0.1599,
|
| 912 |
+
"step": 12900
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 10.4,
|
| 916 |
+
"grad_norm": 1.765625,
|
| 917 |
+
"learning_rate": 5e-05,
|
| 918 |
+
"loss": 0.1595,
|
| 919 |
+
"step": 13000
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 10.48,
|
| 923 |
+
"grad_norm": 1.9765625,
|
| 924 |
+
"learning_rate": 5e-05,
|
| 925 |
+
"loss": 0.1599,
|
| 926 |
+
"step": 13100
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 10.56,
|
| 930 |
+
"grad_norm": 2.140625,
|
| 931 |
+
"learning_rate": 5e-05,
|
| 932 |
+
"loss": 0.1578,
|
| 933 |
+
"step": 13200
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 10.64,
|
| 937 |
+
"grad_norm": 1.84375,
|
| 938 |
+
"learning_rate": 5e-05,
|
| 939 |
+
"loss": 0.1577,
|
| 940 |
+
"step": 13300
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 10.72,
|
| 944 |
+
"grad_norm": 1.90625,
|
| 945 |
+
"learning_rate": 5e-05,
|
| 946 |
+
"loss": 0.1599,
|
| 947 |
+
"step": 13400
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 10.8,
|
| 951 |
+
"grad_norm": 2.1875,
|
| 952 |
+
"learning_rate": 5e-05,
|
| 953 |
+
"loss": 0.1594,
|
| 954 |
+
"step": 13500
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 10.88,
|
| 958 |
+
"grad_norm": 1.7578125,
|
| 959 |
+
"learning_rate": 5e-05,
|
| 960 |
+
"loss": 0.1598,
|
| 961 |
+
"step": 13600
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 10.96,
|
| 965 |
+
"grad_norm": 1.609375,
|
| 966 |
+
"learning_rate": 5e-05,
|
| 967 |
+
"loss": 0.1595,
|
| 968 |
+
"step": 13700
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 11.04,
|
| 972 |
+
"grad_norm": 2.359375,
|
| 973 |
+
"learning_rate": 5e-05,
|
| 974 |
+
"loss": 0.16,
|
| 975 |
+
"step": 13800
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 11.12,
|
| 979 |
+
"grad_norm": 2.59375,
|
| 980 |
+
"learning_rate": 5e-05,
|
| 981 |
+
"loss": 0.1588,
|
| 982 |
+
"step": 13900
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 11.2,
|
| 986 |
+
"grad_norm": 2.046875,
|
| 987 |
+
"learning_rate": 5e-05,
|
| 988 |
+
"loss": 0.1596,
|
| 989 |
+
"step": 14000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 11.28,
|
| 993 |
+
"grad_norm": 1.609375,
|
| 994 |
+
"learning_rate": 5e-05,
|
| 995 |
+
"loss": 0.1595,
|
| 996 |
+
"step": 14100
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 11.36,
|
| 1000 |
+
"grad_norm": 1.5078125,
|
| 1001 |
+
"learning_rate": 5e-05,
|
| 1002 |
+
"loss": 0.1581,
|
| 1003 |
+
"step": 14200
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 11.44,
|
| 1007 |
+
"grad_norm": 1.421875,
|
| 1008 |
+
"learning_rate": 5e-05,
|
| 1009 |
+
"loss": 0.1584,
|
| 1010 |
+
"step": 14300
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 11.52,
|
| 1014 |
+
"grad_norm": 1.8515625,
|
| 1015 |
+
"learning_rate": 5e-05,
|
| 1016 |
+
"loss": 0.1589,
|
| 1017 |
+
"step": 14400
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 11.6,
|
| 1021 |
+
"grad_norm": 1.7578125,
|
| 1022 |
+
"learning_rate": 5e-05,
|
| 1023 |
+
"loss": 0.1578,
|
| 1024 |
+
"step": 14500
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 11.68,
|
| 1028 |
+
"grad_norm": 1.671875,
|
| 1029 |
+
"learning_rate": 5e-05,
|
| 1030 |
+
"loss": 0.1591,
|
| 1031 |
+
"step": 14600
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 11.76,
|
| 1035 |
+
"grad_norm": 1.65625,
|
| 1036 |
+
"learning_rate": 5e-05,
|
| 1037 |
+
"loss": 0.1582,
|
| 1038 |
+
"step": 14700
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 11.84,
|
| 1042 |
+
"grad_norm": 1.9296875,
|
| 1043 |
+
"learning_rate": 5e-05,
|
| 1044 |
+
"loss": 0.1581,
|
| 1045 |
+
"step": 14800
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 11.92,
|
| 1049 |
+
"grad_norm": 2.203125,
|
| 1050 |
+
"learning_rate": 5e-05,
|
| 1051 |
+
"loss": 0.1585,
|
| 1052 |
+
"step": 14900
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 12.0,
|
| 1056 |
+
"grad_norm": 1.7890625,
|
| 1057 |
+
"learning_rate": 5e-05,
|
| 1058 |
+
"loss": 0.1588,
|
| 1059 |
+
"step": 15000
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 12.08,
|
| 1063 |
+
"grad_norm": 1.8203125,
|
| 1064 |
+
"learning_rate": 5e-05,
|
| 1065 |
+
"loss": 0.1594,
|
| 1066 |
+
"step": 15100
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 12.16,
|
| 1070 |
+
"grad_norm": 2.125,
|
| 1071 |
+
"learning_rate": 5e-05,
|
| 1072 |
+
"loss": 0.158,
|
| 1073 |
+
"step": 15200
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 12.24,
|
| 1077 |
+
"grad_norm": 1.8203125,
|
| 1078 |
+
"learning_rate": 5e-05,
|
| 1079 |
+
"loss": 0.1577,
|
| 1080 |
+
"step": 15300
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 12.32,
|
| 1084 |
+
"grad_norm": 2.046875,
|
| 1085 |
+
"learning_rate": 5e-05,
|
| 1086 |
+
"loss": 0.1587,
|
| 1087 |
+
"step": 15400
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 12.4,
|
| 1091 |
+
"grad_norm": 1.4375,
|
| 1092 |
+
"learning_rate": 5e-05,
|
| 1093 |
+
"loss": 0.1579,
|
| 1094 |
+
"step": 15500
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 12.48,
|
| 1098 |
+
"grad_norm": 1.609375,
|
| 1099 |
+
"learning_rate": 5e-05,
|
| 1100 |
+
"loss": 0.1581,
|
| 1101 |
+
"step": 15600
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 12.56,
|
| 1105 |
+
"grad_norm": 1.8359375,
|
| 1106 |
+
"learning_rate": 5e-05,
|
| 1107 |
+
"loss": 0.1577,
|
| 1108 |
+
"step": 15700
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 12.64,
|
| 1112 |
+
"grad_norm": 2.4375,
|
| 1113 |
+
"learning_rate": 5e-05,
|
| 1114 |
+
"loss": 0.1566,
|
| 1115 |
+
"step": 15800
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 12.72,
|
| 1119 |
+
"grad_norm": 1.8046875,
|
| 1120 |
+
"learning_rate": 5e-05,
|
| 1121 |
+
"loss": 0.1594,
|
| 1122 |
+
"step": 15900
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 12.8,
|
| 1126 |
+
"grad_norm": 2.359375,
|
| 1127 |
+
"learning_rate": 5e-05,
|
| 1128 |
+
"loss": 0.1578,
|
| 1129 |
+
"step": 16000
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 12.88,
|
| 1133 |
+
"grad_norm": 1.921875,
|
| 1134 |
+
"learning_rate": 5e-05,
|
| 1135 |
+
"loss": 0.1576,
|
| 1136 |
+
"step": 16100
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 12.96,
|
| 1140 |
+
"grad_norm": 2.078125,
|
| 1141 |
+
"learning_rate": 5e-05,
|
| 1142 |
+
"loss": 0.1577,
|
| 1143 |
+
"step": 16200
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 13.04,
|
| 1147 |
+
"grad_norm": 1.953125,
|
| 1148 |
+
"learning_rate": 5e-05,
|
| 1149 |
+
"loss": 0.1561,
|
| 1150 |
+
"step": 16300
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 13.12,
|
| 1154 |
+
"grad_norm": 1.90625,
|
| 1155 |
+
"learning_rate": 5e-05,
|
| 1156 |
+
"loss": 0.157,
|
| 1157 |
+
"step": 16400
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 13.2,
|
| 1161 |
+
"grad_norm": 2.046875,
|
| 1162 |
+
"learning_rate": 5e-05,
|
| 1163 |
+
"loss": 0.1564,
|
| 1164 |
+
"step": 16500
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 13.28,
|
| 1168 |
+
"grad_norm": 1.7265625,
|
| 1169 |
+
"learning_rate": 5e-05,
|
| 1170 |
+
"loss": 0.1579,
|
| 1171 |
+
"step": 16600
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 13.36,
|
| 1175 |
+
"grad_norm": 1.9921875,
|
| 1176 |
+
"learning_rate": 5e-05,
|
| 1177 |
+
"loss": 0.1571,
|
| 1178 |
+
"step": 16700
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 13.44,
|
| 1182 |
+
"grad_norm": 1.7109375,
|
| 1183 |
+
"learning_rate": 5e-05,
|
| 1184 |
+
"loss": 0.1565,
|
| 1185 |
+
"step": 16800
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 13.52,
|
| 1189 |
+
"grad_norm": 2.0625,
|
| 1190 |
+
"learning_rate": 5e-05,
|
| 1191 |
+
"loss": 0.156,
|
| 1192 |
+
"step": 16900
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 13.6,
|
| 1196 |
+
"grad_norm": 1.625,
|
| 1197 |
+
"learning_rate": 5e-05,
|
| 1198 |
+
"loss": 0.1565,
|
| 1199 |
+
"step": 17000
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 13.68,
|
| 1203 |
+
"grad_norm": 2.171875,
|
| 1204 |
+
"learning_rate": 5e-05,
|
| 1205 |
+
"loss": 0.1565,
|
| 1206 |
+
"step": 17100
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 13.76,
|
| 1210 |
+
"grad_norm": 2.046875,
|
| 1211 |
+
"learning_rate": 5e-05,
|
| 1212 |
+
"loss": 0.156,
|
| 1213 |
+
"step": 17200
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 13.84,
|
| 1217 |
+
"grad_norm": 2.203125,
|
| 1218 |
+
"learning_rate": 5e-05,
|
| 1219 |
+
"loss": 0.1561,
|
| 1220 |
+
"step": 17300
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 13.92,
|
| 1224 |
+
"grad_norm": 1.7265625,
|
| 1225 |
+
"learning_rate": 5e-05,
|
| 1226 |
+
"loss": 0.1558,
|
| 1227 |
+
"step": 17400
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 14.0,
|
| 1231 |
+
"grad_norm": 1.7890625,
|
| 1232 |
+
"learning_rate": 5e-05,
|
| 1233 |
+
"loss": 0.157,
|
| 1234 |
+
"step": 17500
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 14.08,
|
| 1238 |
+
"grad_norm": 1.8125,
|
| 1239 |
+
"learning_rate": 5e-05,
|
| 1240 |
+
"loss": 0.1556,
|
| 1241 |
+
"step": 17600
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 14.16,
|
| 1245 |
+
"grad_norm": 1.84375,
|
| 1246 |
+
"learning_rate": 5e-05,
|
| 1247 |
+
"loss": 0.1576,
|
| 1248 |
+
"step": 17700
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 14.24,
|
| 1252 |
+
"grad_norm": 2.21875,
|
| 1253 |
+
"learning_rate": 5e-05,
|
| 1254 |
+
"loss": 0.1563,
|
| 1255 |
+
"step": 17800
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 14.32,
|
| 1259 |
+
"grad_norm": 1.796875,
|
| 1260 |
+
"learning_rate": 5e-05,
|
| 1261 |
+
"loss": 0.1553,
|
| 1262 |
+
"step": 17900
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 14.4,
|
| 1266 |
+
"grad_norm": 1.8671875,
|
| 1267 |
+
"learning_rate": 5e-05,
|
| 1268 |
+
"loss": 0.1577,
|
| 1269 |
+
"step": 18000
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 14.48,
|
| 1273 |
+
"grad_norm": 1.8671875,
|
| 1274 |
+
"learning_rate": 5e-05,
|
| 1275 |
+
"loss": 0.156,
|
| 1276 |
+
"step": 18100
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 14.56,
|
| 1280 |
+
"grad_norm": 1.921875,
|
| 1281 |
+
"learning_rate": 5e-05,
|
| 1282 |
+
"loss": 0.1564,
|
| 1283 |
+
"step": 18200
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 14.64,
|
| 1287 |
+
"grad_norm": 1.5859375,
|
| 1288 |
+
"learning_rate": 5e-05,
|
| 1289 |
+
"loss": 0.1549,
|
| 1290 |
+
"step": 18300
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 14.72,
|
| 1294 |
+
"grad_norm": 1.3828125,
|
| 1295 |
+
"learning_rate": 5e-05,
|
| 1296 |
+
"loss": 0.1565,
|
| 1297 |
+
"step": 18400
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 14.8,
|
| 1301 |
+
"grad_norm": 2.59375,
|
| 1302 |
+
"learning_rate": 5e-05,
|
| 1303 |
+
"loss": 0.156,
|
| 1304 |
+
"step": 18500
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 14.88,
|
| 1308 |
+
"grad_norm": 1.875,
|
| 1309 |
+
"learning_rate": 5e-05,
|
| 1310 |
+
"loss": 0.1574,
|
| 1311 |
+
"step": 18600
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 14.96,
|
| 1315 |
+
"grad_norm": 2.09375,
|
| 1316 |
+
"learning_rate": 5e-05,
|
| 1317 |
+
"loss": 0.1562,
|
| 1318 |
+
"step": 18700
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 15.04,
|
| 1322 |
+
"grad_norm": 1.953125,
|
| 1323 |
+
"learning_rate": 5e-05,
|
| 1324 |
+
"loss": 0.1555,
|
| 1325 |
+
"step": 18800
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 15.12,
|
| 1329 |
+
"grad_norm": 2.1875,
|
| 1330 |
+
"learning_rate": 5e-05,
|
| 1331 |
+
"loss": 0.1564,
|
| 1332 |
+
"step": 18900
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 15.2,
|
| 1336 |
+
"grad_norm": 2.015625,
|
| 1337 |
+
"learning_rate": 5e-05,
|
| 1338 |
+
"loss": 0.1549,
|
| 1339 |
+
"step": 19000
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 15.28,
|
| 1343 |
+
"grad_norm": 1.609375,
|
| 1344 |
+
"learning_rate": 5e-05,
|
| 1345 |
+
"loss": 0.1563,
|
| 1346 |
+
"step": 19100
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 15.36,
|
| 1350 |
+
"grad_norm": 2.203125,
|
| 1351 |
+
"learning_rate": 5e-05,
|
| 1352 |
+
"loss": 0.1559,
|
| 1353 |
+
"step": 19200
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 15.44,
|
| 1357 |
+
"grad_norm": 2.046875,
|
| 1358 |
+
"learning_rate": 5e-05,
|
| 1359 |
+
"loss": 0.1551,
|
| 1360 |
+
"step": 19300
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 15.52,
|
| 1364 |
+
"grad_norm": 1.46875,
|
| 1365 |
+
"learning_rate": 5e-05,
|
| 1366 |
+
"loss": 0.1571,
|
| 1367 |
+
"step": 19400
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 15.6,
|
| 1371 |
+
"grad_norm": 2.109375,
|
| 1372 |
+
"learning_rate": 5e-05,
|
| 1373 |
+
"loss": 0.1555,
|
| 1374 |
+
"step": 19500
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 15.68,
|
| 1378 |
+
"grad_norm": 1.8203125,
|
| 1379 |
+
"learning_rate": 5e-05,
|
| 1380 |
+
"loss": 0.157,
|
| 1381 |
+
"step": 19600
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 15.76,
|
| 1385 |
+
"grad_norm": 1.8515625,
|
| 1386 |
+
"learning_rate": 5e-05,
|
| 1387 |
+
"loss": 0.1564,
|
| 1388 |
+
"step": 19700
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 15.84,
|
| 1392 |
+
"grad_norm": 1.8125,
|
| 1393 |
+
"learning_rate": 5e-05,
|
| 1394 |
+
"loss": 0.1561,
|
| 1395 |
+
"step": 19800
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 15.92,
|
| 1399 |
+
"grad_norm": 1.96875,
|
| 1400 |
+
"learning_rate": 5e-05,
|
| 1401 |
+
"loss": 0.1559,
|
| 1402 |
+
"step": 19900
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 16.0,
|
| 1406 |
+
"grad_norm": 1.796875,
|
| 1407 |
+
"learning_rate": 5e-05,
|
| 1408 |
+
"loss": 0.1548,
|
| 1409 |
+
"step": 20000
|
| 1410 |
+
}
|
| 1411 |
+
],
|
| 1412 |
+
"logging_steps": 100,
|
| 1413 |
+
"max_steps": 300000,
|
| 1414 |
+
"num_input_tokens_seen": 0,
|
| 1415 |
+
"num_train_epochs": 240,
|
| 1416 |
+
"save_steps": 500,
|
| 1417 |
+
"stateful_callbacks": {
|
| 1418 |
+
"TrainerControl": {
|
| 1419 |
+
"args": {
|
| 1420 |
+
"should_epoch_stop": false,
|
| 1421 |
+
"should_evaluate": false,
|
| 1422 |
+
"should_log": false,
|
| 1423 |
+
"should_save": true,
|
| 1424 |
+
"should_training_stop": false
|
| 1425 |
+
},
|
| 1426 |
+
"attributes": {}
|
| 1427 |
+
}
|
| 1428 |
+
},
|
| 1429 |
+
"total_flos": 0.0,
|
| 1430 |
+
"train_batch_size": 1,
|
| 1431 |
+
"trial_name": null,
|
| 1432 |
+
"trial_params": null
|
| 1433 |
+
}
|