Upload folder using huggingface_hub
Browse files- config.json +64 -0
- experiment_cfg/metadata.json +387 -0
- model-00001-of-00002.safetensors +3 -0
- model-00002-of-00002.safetensors +3 -0
- model.safetensors.index.json +0 -0
- trainer_state.json +883 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,64 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"action_dim": 32,
|
| 3 |
+
"action_head_cfg": {
|
| 4 |
+
"action_dim": 32,
|
| 5 |
+
"action_horizon": 16,
|
| 6 |
+
"add_pos_embed": true,
|
| 7 |
+
"backbone_embedding_dim": 2048,
|
| 8 |
+
"diffusion_model_cfg": {
|
| 9 |
+
"attention_head_dim": 48,
|
| 10 |
+
"cross_attention_dim": 2048,
|
| 11 |
+
"dropout": 0.2,
|
| 12 |
+
"final_dropout": true,
|
| 13 |
+
"interleave_self_attention": true,
|
| 14 |
+
"norm_type": "ada_norm",
|
| 15 |
+
"num_attention_heads": 32,
|
| 16 |
+
"num_layers": 16,
|
| 17 |
+
"output_dim": 1024,
|
| 18 |
+
"positional_embeddings": null
|
| 19 |
+
},
|
| 20 |
+
"hidden_size": 1024,
|
| 21 |
+
"input_embedding_dim": 1536,
|
| 22 |
+
"max_action_dim": 32,
|
| 23 |
+
"max_state_dim": 64,
|
| 24 |
+
"model_dtype": "float32",
|
| 25 |
+
"noise_beta_alpha": 1.5,
|
| 26 |
+
"noise_beta_beta": 1.0,
|
| 27 |
+
"noise_s": 0.999,
|
| 28 |
+
"num_inference_timesteps": 4,
|
| 29 |
+
"num_target_vision_tokens": 32,
|
| 30 |
+
"num_timestep_buckets": 1000,
|
| 31 |
+
"tune_diffusion_model": true,
|
| 32 |
+
"tune_projector": true,
|
| 33 |
+
"use_vlln": true,
|
| 34 |
+
"vl_self_attention_cfg": {
|
| 35 |
+
"attention_head_dim": 64,
|
| 36 |
+
"dropout": 0.2,
|
| 37 |
+
"final_dropout": true,
|
| 38 |
+
"num_attention_heads": 32,
|
| 39 |
+
"num_layers": 4,
|
| 40 |
+
"positional_embeddings": null
|
| 41 |
+
}
|
| 42 |
+
},
|
| 43 |
+
"action_horizon": 16,
|
| 44 |
+
"architectures": [
|
| 45 |
+
"GR00T_N1_5"
|
| 46 |
+
],
|
| 47 |
+
"attn_implementation": null,
|
| 48 |
+
"backbone_cfg": {
|
| 49 |
+
"eagle_path": "NVEagle/eagle_er-qwen3_1_7B-Siglip2_400M_stage1_5_128gpu_er_v7_1mlp_nops",
|
| 50 |
+
"load_bf16": false,
|
| 51 |
+
"project_to_dim": null,
|
| 52 |
+
"reproject_vision": false,
|
| 53 |
+
"select_layer": 12,
|
| 54 |
+
"tune_llm": false,
|
| 55 |
+
"tune_visual": true,
|
| 56 |
+
"use_flash_attention": true
|
| 57 |
+
},
|
| 58 |
+
"compute_dtype": "bfloat16",
|
| 59 |
+
"hidden_size": 2048,
|
| 60 |
+
"model_dtype": "float32",
|
| 61 |
+
"model_type": "gr00t_n1_5",
|
| 62 |
+
"torch_dtype": "bfloat16",
|
| 63 |
+
"transformers_version": "4.52.1"
|
| 64 |
+
}
|
experiment_cfg/metadata.json
ADDED
|
@@ -0,0 +1,387 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"new_embodiment": {
|
| 3 |
+
"statistics": {
|
| 4 |
+
"state": {
|
| 5 |
+
"left_arm": {
|
| 6 |
+
"max": [
|
| 7 |
+
0.23685863614082336,
|
| 8 |
+
2.63502836227417,
|
| 9 |
+
0.03330088034272194,
|
| 10 |
+
1.4575419425964355,
|
| 11 |
+
0.8281587362289429,
|
| 12 |
+
1.9787147045135498
|
| 13 |
+
],
|
| 14 |
+
"min": [
|
| 15 |
+
-0.797720193862915,
|
| 16 |
+
-0.012758356519043446,
|
| 17 |
+
-2.2290821075439453,
|
| 18 |
+
-1.7361313104629517,
|
| 19 |
+
-0.532290518283844,
|
| 20 |
+
-1.4025291204452515
|
| 21 |
+
],
|
| 22 |
+
"mean": [
|
| 23 |
+
-0.09467598050832748,
|
| 24 |
+
0.29610729217529297,
|
| 25 |
+
-0.1754428744316101,
|
| 26 |
+
-0.0011706651421263814,
|
| 27 |
+
0.08008287101984024,
|
| 28 |
+
0.004690642934292555
|
| 29 |
+
],
|
| 30 |
+
"std": [
|
| 31 |
+
0.11690603941679001,
|
| 32 |
+
0.6777294278144836,
|
| 33 |
+
0.4925391376018524,
|
| 34 |
+
0.22012004256248474,
|
| 35 |
+
0.08415775001049042,
|
| 36 |
+
0.2379513680934906
|
| 37 |
+
],
|
| 38 |
+
"q01": [
|
| 39 |
+
-0.6201645332574844,
|
| 40 |
+
-0.012758356519043446,
|
| 41 |
+
-1.8302751040458678,
|
| 42 |
+
-0.8233977997303008,
|
| 43 |
+
-0.25266467064619064,
|
| 44 |
+
-0.6677455306053162
|
| 45 |
+
],
|
| 46 |
+
"q99": [
|
| 47 |
+
0.0558854416012764,
|
| 48 |
+
2.305639197826385,
|
| 49 |
+
0.03330088034272194,
|
| 50 |
+
0.8909350645542136,
|
| 51 |
+
0.378433117866516,
|
| 52 |
+
0.9695222181081766
|
| 53 |
+
]
|
| 54 |
+
},
|
| 55 |
+
"left_gripper": {
|
| 56 |
+
"max": [
|
| 57 |
+
0.054858461022377014
|
| 58 |
+
],
|
| 59 |
+
"min": [
|
| 60 |
+
-0.0037799999117851257
|
| 61 |
+
],
|
| 62 |
+
"mean": [
|
| 63 |
+
0.0005394071922637522
|
| 64 |
+
],
|
| 65 |
+
"std": [
|
| 66 |
+
0.008630749769508839
|
| 67 |
+
],
|
| 68 |
+
"q01": [
|
| 69 |
+
-0.0037799999117851257
|
| 70 |
+
],
|
| 71 |
+
"q99": [
|
| 72 |
+
0.04758923128247261
|
| 73 |
+
]
|
| 74 |
+
},
|
| 75 |
+
"right_arm": {
|
| 76 |
+
"max": [
|
| 77 |
+
0.8744099736213684,
|
| 78 |
+
2.716465473175049,
|
| 79 |
+
0.010594148188829422,
|
| 80 |
+
1.7382432222366333,
|
| 81 |
+
1.2609829902648926,
|
| 82 |
+
0.7371398210525513
|
| 83 |
+
],
|
| 84 |
+
"min": [
|
| 85 |
+
-0.30108675360679626,
|
| 86 |
+
0.0,
|
| 87 |
+
-2.7580041885375977,
|
| 88 |
+
-1.2717341184616089,
|
| 89 |
+
-0.7101046442985535,
|
| 90 |
+
-2.1718528270721436
|
| 91 |
+
],
|
| 92 |
+
"mean": [
|
| 93 |
+
0.13320903480052948,
|
| 94 |
+
1.0849677324295044,
|
| 95 |
+
-0.8214209079742432,
|
| 96 |
+
0.08245757967233658,
|
| 97 |
+
0.34770581126213074,
|
| 98 |
+
-0.34033045172691345
|
| 99 |
+
],
|
| 100 |
+
"std": [
|
| 101 |
+
0.23607806861400604,
|
| 102 |
+
0.958350658416748,
|
| 103 |
+
0.8432449102401733,
|
| 104 |
+
0.26838839054107666,
|
| 105 |
+
0.4480270445346832,
|
| 106 |
+
0.4256483316421509
|
| 107 |
+
],
|
| 108 |
+
"q01": [
|
| 109 |
+
-0.1667721477150917,
|
| 110 |
+
0.0,
|
| 111 |
+
-2.3997578620910645,
|
| 112 |
+
-0.7350239264965057,
|
| 113 |
+
-0.21033382192254066,
|
| 114 |
+
-1.8293345880508423
|
| 115 |
+
],
|
| 116 |
+
"q99": [
|
| 117 |
+
0.734721313714981,
|
| 118 |
+
2.5196237874031064,
|
| 119 |
+
0.009965830482542515,
|
| 120 |
+
1.0604445934295654,
|
| 121 |
+
1.2147316932678223,
|
| 122 |
+
0.37224382162094116
|
| 123 |
+
]
|
| 124 |
+
},
|
| 125 |
+
"right_gripper": {
|
| 126 |
+
"max": [
|
| 127 |
+
0.08273077011108398
|
| 128 |
+
],
|
| 129 |
+
"min": [
|
| 130 |
+
-0.0005769230774603784
|
| 131 |
+
],
|
| 132 |
+
"mean": [
|
| 133 |
+
0.017459724098443985
|
| 134 |
+
],
|
| 135 |
+
"std": [
|
| 136 |
+
0.02439081110060215
|
| 137 |
+
],
|
| 138 |
+
"q01": [
|
| 139 |
+
-0.0004615384677890688
|
| 140 |
+
],
|
| 141 |
+
"q99": [
|
| 142 |
+
0.07165384292602539
|
| 143 |
+
]
|
| 144 |
+
}
|
| 145 |
+
},
|
| 146 |
+
"action": {
|
| 147 |
+
"left_arm": {
|
| 148 |
+
"max": [
|
| 149 |
+
0.23685863614082336,
|
| 150 |
+
2.63502836227417,
|
| 151 |
+
0.03330088034272194,
|
| 152 |
+
1.4575419425964355,
|
| 153 |
+
0.8281587362289429,
|
| 154 |
+
1.9787147045135498
|
| 155 |
+
],
|
| 156 |
+
"min": [
|
| 157 |
+
-0.797720193862915,
|
| 158 |
+
-0.012758356519043446,
|
| 159 |
+
-2.2290821075439453,
|
| 160 |
+
-1.7361313104629517,
|
| 161 |
+
-0.532290518283844,
|
| 162 |
+
-1.4025291204452515
|
| 163 |
+
],
|
| 164 |
+
"mean": [
|
| 165 |
+
-0.09467556327581406,
|
| 166 |
+
0.2961057424545288,
|
| 167 |
+
-0.1754431128501892,
|
| 168 |
+
-0.0011543770087882876,
|
| 169 |
+
0.08008351922035217,
|
| 170 |
+
0.0046795387752354145
|
| 171 |
+
],
|
| 172 |
+
"std": [
|
| 173 |
+
0.11690639704465866,
|
| 174 |
+
0.6777272820472717,
|
| 175 |
+
0.4925388991832733,
|
| 176 |
+
0.2201227992773056,
|
| 177 |
+
0.08415870368480682,
|
| 178 |
+
0.23795734345912933
|
| 179 |
+
],
|
| 180 |
+
"q01": [
|
| 181 |
+
-0.6201645332574844,
|
| 182 |
+
-0.012758356519043446,
|
| 183 |
+
-1.8302751040458678,
|
| 184 |
+
-0.8233977997303008,
|
| 185 |
+
-0.25266467064619064,
|
| 186 |
+
-0.6677455306053162
|
| 187 |
+
],
|
| 188 |
+
"q99": [
|
| 189 |
+
0.0558854416012764,
|
| 190 |
+
2.305639197826385,
|
| 191 |
+
0.03330088034272194,
|
| 192 |
+
0.8909350645542136,
|
| 193 |
+
0.378433117866516,
|
| 194 |
+
0.9695222181081766
|
| 195 |
+
]
|
| 196 |
+
},
|
| 197 |
+
"left_gripper": {
|
| 198 |
+
"max": [
|
| 199 |
+
0.054858461022377014
|
| 200 |
+
],
|
| 201 |
+
"min": [
|
| 202 |
+
-0.0037799999117851257
|
| 203 |
+
],
|
| 204 |
+
"mean": [
|
| 205 |
+
0.0005394072504714131
|
| 206 |
+
],
|
| 207 |
+
"std": [
|
| 208 |
+
0.008630749769508839
|
| 209 |
+
],
|
| 210 |
+
"q01": [
|
| 211 |
+
-0.0037799999117851257
|
| 212 |
+
],
|
| 213 |
+
"q99": [
|
| 214 |
+
0.04758923128247261
|
| 215 |
+
]
|
| 216 |
+
},
|
| 217 |
+
"right_arm": {
|
| 218 |
+
"max": [
|
| 219 |
+
0.8744099736213684,
|
| 220 |
+
2.716465473175049,
|
| 221 |
+
0.010594148188829422,
|
| 222 |
+
1.7382432222366333,
|
| 223 |
+
1.2609829902648926,
|
| 224 |
+
0.7371398210525513
|
| 225 |
+
],
|
| 226 |
+
"min": [
|
| 227 |
+
-0.30108675360679626,
|
| 228 |
+
0.0,
|
| 229 |
+
-2.7580041885375977,
|
| 230 |
+
-1.2717341184616089,
|
| 231 |
+
-0.7101046442985535,
|
| 232 |
+
-2.1718528270721436
|
| 233 |
+
],
|
| 234 |
+
"mean": [
|
| 235 |
+
0.13320539891719818,
|
| 236 |
+
1.084963083267212,
|
| 237 |
+
-0.8214203119277954,
|
| 238 |
+
0.08245743811130524,
|
| 239 |
+
0.3477059602737427,
|
| 240 |
+
-0.3403414785861969
|
| 241 |
+
],
|
| 242 |
+
"std": [
|
| 243 |
+
0.23608171939849854,
|
| 244 |
+
0.958354115486145,
|
| 245 |
+
0.8432462811470032,
|
| 246 |
+
0.2683871388435364,
|
| 247 |
+
0.44802600145339966,
|
| 248 |
+
0.4256395101547241
|
| 249 |
+
],
|
| 250 |
+
"q01": [
|
| 251 |
+
-0.16684684187173843,
|
| 252 |
+
0.0,
|
| 253 |
+
-2.3997578620910645,
|
| 254 |
+
-0.7350239264965057,
|
| 255 |
+
-0.21033382192254066,
|
| 256 |
+
-1.8293345880508423
|
| 257 |
+
],
|
| 258 |
+
"q99": [
|
| 259 |
+
0.734721313714981,
|
| 260 |
+
2.5196237874031064,
|
| 261 |
+
0.009965830482542515,
|
| 262 |
+
1.0604445934295654,
|
| 263 |
+
1.2147316932678223,
|
| 264 |
+
0.37224382162094116
|
| 265 |
+
]
|
| 266 |
+
},
|
| 267 |
+
"right_gripper": {
|
| 268 |
+
"max": [
|
| 269 |
+
0.08273077011108398
|
| 270 |
+
],
|
| 271 |
+
"min": [
|
| 272 |
+
-0.0005769230774603784
|
| 273 |
+
],
|
| 274 |
+
"mean": [
|
| 275 |
+
0.01745966635644436
|
| 276 |
+
],
|
| 277 |
+
"std": [
|
| 278 |
+
0.024390874430537224
|
| 279 |
+
],
|
| 280 |
+
"q01": [
|
| 281 |
+
-0.0004615384677890688
|
| 282 |
+
],
|
| 283 |
+
"q99": [
|
| 284 |
+
0.07165384292602539
|
| 285 |
+
]
|
| 286 |
+
}
|
| 287 |
+
}
|
| 288 |
+
},
|
| 289 |
+
"modalities": {
|
| 290 |
+
"video": {
|
| 291 |
+
"cam_high": {
|
| 292 |
+
"resolution": [
|
| 293 |
+
640,
|
| 294 |
+
480
|
| 295 |
+
],
|
| 296 |
+
"channels": 3,
|
| 297 |
+
"fps": 30.0
|
| 298 |
+
},
|
| 299 |
+
"cam_left_wrist": {
|
| 300 |
+
"resolution": [
|
| 301 |
+
640,
|
| 302 |
+
480
|
| 303 |
+
],
|
| 304 |
+
"channels": 3,
|
| 305 |
+
"fps": 30.0
|
| 306 |
+
},
|
| 307 |
+
"cam_right_wrist": {
|
| 308 |
+
"resolution": [
|
| 309 |
+
640,
|
| 310 |
+
480
|
| 311 |
+
],
|
| 312 |
+
"channels": 3,
|
| 313 |
+
"fps": 30.0
|
| 314 |
+
}
|
| 315 |
+
},
|
| 316 |
+
"state": {
|
| 317 |
+
"left_arm": {
|
| 318 |
+
"absolute": true,
|
| 319 |
+
"rotation_type": null,
|
| 320 |
+
"shape": [
|
| 321 |
+
6
|
| 322 |
+
],
|
| 323 |
+
"continuous": true
|
| 324 |
+
},
|
| 325 |
+
"left_gripper": {
|
| 326 |
+
"absolute": true,
|
| 327 |
+
"rotation_type": null,
|
| 328 |
+
"shape": [
|
| 329 |
+
1
|
| 330 |
+
],
|
| 331 |
+
"continuous": true
|
| 332 |
+
},
|
| 333 |
+
"right_arm": {
|
| 334 |
+
"absolute": true,
|
| 335 |
+
"rotation_type": null,
|
| 336 |
+
"shape": [
|
| 337 |
+
6
|
| 338 |
+
],
|
| 339 |
+
"continuous": true
|
| 340 |
+
},
|
| 341 |
+
"right_gripper": {
|
| 342 |
+
"absolute": true,
|
| 343 |
+
"rotation_type": null,
|
| 344 |
+
"shape": [
|
| 345 |
+
1
|
| 346 |
+
],
|
| 347 |
+
"continuous": true
|
| 348 |
+
}
|
| 349 |
+
},
|
| 350 |
+
"action": {
|
| 351 |
+
"left_arm": {
|
| 352 |
+
"absolute": true,
|
| 353 |
+
"rotation_type": null,
|
| 354 |
+
"shape": [
|
| 355 |
+
6
|
| 356 |
+
],
|
| 357 |
+
"continuous": true
|
| 358 |
+
},
|
| 359 |
+
"left_gripper": {
|
| 360 |
+
"absolute": true,
|
| 361 |
+
"rotation_type": null,
|
| 362 |
+
"shape": [
|
| 363 |
+
1
|
| 364 |
+
],
|
| 365 |
+
"continuous": true
|
| 366 |
+
},
|
| 367 |
+
"right_arm": {
|
| 368 |
+
"absolute": true,
|
| 369 |
+
"rotation_type": null,
|
| 370 |
+
"shape": [
|
| 371 |
+
6
|
| 372 |
+
],
|
| 373 |
+
"continuous": true
|
| 374 |
+
},
|
| 375 |
+
"right_gripper": {
|
| 376 |
+
"absolute": true,
|
| 377 |
+
"rotation_type": null,
|
| 378 |
+
"shape": [
|
| 379 |
+
1
|
| 380 |
+
],
|
| 381 |
+
"continuous": true
|
| 382 |
+
}
|
| 383 |
+
}
|
| 384 |
+
},
|
| 385 |
+
"embodiment_tag": "new_embodiment"
|
| 386 |
+
}
|
| 387 |
+
}
|
model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d91b3b4e00662607b0966ad86755610d96ccbc0b52043c833074172e5516dc16
|
| 3 |
+
size 4999367032
|
model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fb7ca664e98f9202c0fdf3bc2b348a08eeef09b5a410fc55700fd5ad4fc006b7
|
| 3 |
+
size 2586705312
|
model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
trainer_state.json
ADDED
|
@@ -0,0 +1,883 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 6.382978723404255,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 1200,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.05319148936170213,
|
| 14 |
+
"grad_norm": 3.2709779739379883,
|
| 15 |
+
"learning_rate": 1.5e-05,
|
| 16 |
+
"loss": 0.8282,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.10638297872340426,
|
| 21 |
+
"grad_norm": 0.862433910369873,
|
| 22 |
+
"learning_rate": 3.1666666666666666e-05,
|
| 23 |
+
"loss": 0.1893,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.1595744680851064,
|
| 28 |
+
"grad_norm": 0.6577978134155273,
|
| 29 |
+
"learning_rate": 4.8333333333333334e-05,
|
| 30 |
+
"loss": 0.1087,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.2127659574468085,
|
| 35 |
+
"grad_norm": 0.39728644490242004,
|
| 36 |
+
"learning_rate": 6.500000000000001e-05,
|
| 37 |
+
"loss": 0.0913,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.26595744680851063,
|
| 42 |
+
"grad_norm": 0.5268833041191101,
|
| 43 |
+
"learning_rate": 8.166666666666667e-05,
|
| 44 |
+
"loss": 0.0749,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.3191489361702128,
|
| 49 |
+
"grad_norm": 0.45948076248168945,
|
| 50 |
+
"learning_rate": 9.833333333333333e-05,
|
| 51 |
+
"loss": 0.063,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.3723404255319149,
|
| 56 |
+
"grad_norm": 0.5164420008659363,
|
| 57 |
+
"learning_rate": 9.998462224960175e-05,
|
| 58 |
+
"loss": 0.0608,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.425531914893617,
|
| 63 |
+
"grad_norm": 0.3698779046535492,
|
| 64 |
+
"learning_rate": 9.99314767377287e-05,
|
| 65 |
+
"loss": 0.0525,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.4787234042553192,
|
| 70 |
+
"grad_norm": 0.4313788115978241,
|
| 71 |
+
"learning_rate": 9.984041396395343e-05,
|
| 72 |
+
"loss": 0.0473,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.5319148936170213,
|
| 77 |
+
"grad_norm": 0.4923841655254364,
|
| 78 |
+
"learning_rate": 9.97115030800669e-05,
|
| 79 |
+
"loss": 0.0418,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.5851063829787234,
|
| 84 |
+
"grad_norm": 0.36365053057670593,
|
| 85 |
+
"learning_rate": 9.954484197918315e-05,
|
| 86 |
+
"loss": 0.0372,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.6382978723404256,
|
| 91 |
+
"grad_norm": 0.3355150520801544,
|
| 92 |
+
"learning_rate": 9.934055722140061e-05,
|
| 93 |
+
"loss": 0.0332,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.6914893617021277,
|
| 98 |
+
"grad_norm": 0.30868247151374817,
|
| 99 |
+
"learning_rate": 9.90988039376942e-05,
|
| 100 |
+
"loss": 0.0296,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.7446808510638298,
|
| 105 |
+
"grad_norm": 0.20192113518714905,
|
| 106 |
+
"learning_rate": 9.88197657121109e-05,
|
| 107 |
+
"loss": 0.0312,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.7978723404255319,
|
| 112 |
+
"grad_norm": 0.2799043357372284,
|
| 113 |
+
"learning_rate": 9.85036544423588e-05,
|
| 114 |
+
"loss": 0.0271,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.851063829787234,
|
| 119 |
+
"grad_norm": 0.2039630264043808,
|
| 120 |
+
"learning_rate": 9.815071017889482e-05,
|
| 121 |
+
"loss": 0.0239,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.9042553191489362,
|
| 126 |
+
"grad_norm": 0.2750301957130432,
|
| 127 |
+
"learning_rate": 9.776120094263376e-05,
|
| 128 |
+
"loss": 0.0232,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.9574468085106383,
|
| 133 |
+
"grad_norm": 0.27556082606315613,
|
| 134 |
+
"learning_rate": 9.733542252141711e-05,
|
| 135 |
+
"loss": 0.0238,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 1.0106382978723405,
|
| 140 |
+
"grad_norm": 0.33584487438201904,
|
| 141 |
+
"learning_rate": 9.687369824539577e-05,
|
| 142 |
+
"loss": 0.0255,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 1.0638297872340425,
|
| 147 |
+
"grad_norm": 0.283274382352829,
|
| 148 |
+
"learning_rate": 9.637637874149779e-05,
|
| 149 |
+
"loss": 0.0249,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 1.1170212765957448,
|
| 154 |
+
"grad_norm": 0.2790728211402893,
|
| 155 |
+
"learning_rate": 9.584384166716714e-05,
|
| 156 |
+
"loss": 0.022,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 1.1702127659574468,
|
| 161 |
+
"grad_norm": 0.2873629927635193,
|
| 162 |
+
"learning_rate": 9.527649142357596e-05,
|
| 163 |
+
"loss": 0.0204,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 1.2234042553191489,
|
| 168 |
+
"grad_norm": 0.3249283730983734,
|
| 169 |
+
"learning_rate": 9.4674758848528e-05,
|
| 170 |
+
"loss": 0.0197,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 1.2765957446808511,
|
| 175 |
+
"grad_norm": 0.21401825547218323,
|
| 176 |
+
"learning_rate": 9.403910088928651e-05,
|
| 177 |
+
"loss": 0.0207,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 1.3297872340425532,
|
| 182 |
+
"grad_norm": 0.30631324648857117,
|
| 183 |
+
"learning_rate": 9.337000025557476e-05,
|
| 184 |
+
"loss": 0.0193,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 1.3829787234042552,
|
| 189 |
+
"grad_norm": 0.2569330334663391,
|
| 190 |
+
"learning_rate": 9.266796505301322e-05,
|
| 191 |
+
"loss": 0.0173,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 1.4361702127659575,
|
| 196 |
+
"grad_norm": 0.2321569174528122,
|
| 197 |
+
"learning_rate": 9.193352839727121e-05,
|
| 198 |
+
"loss": 0.0208,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 1.4893617021276595,
|
| 203 |
+
"grad_norm": 0.1602529138326645,
|
| 204 |
+
"learning_rate": 9.116724800922629e-05,
|
| 205 |
+
"loss": 0.0185,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 1.5425531914893615,
|
| 210 |
+
"grad_norm": 0.18745878338813782,
|
| 211 |
+
"learning_rate": 9.0369705791439e-05,
|
| 212 |
+
"loss": 0.0173,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 1.5957446808510638,
|
| 217 |
+
"grad_norm": 0.177002415060997,
|
| 218 |
+
"learning_rate": 8.954150738626414e-05,
|
| 219 |
+
"loss": 0.0209,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 1.648936170212766,
|
| 224 |
+
"grad_norm": 0.22678908705711365,
|
| 225 |
+
"learning_rate": 8.868328171593448e-05,
|
| 226 |
+
"loss": 0.0166,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 1.702127659574468,
|
| 231 |
+
"grad_norm": 0.18619535863399506,
|
| 232 |
+
"learning_rate": 8.77956805049661e-05,
|
| 233 |
+
"loss": 0.0158,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 1.7553191489361701,
|
| 238 |
+
"grad_norm": 0.23058553040027618,
|
| 239 |
+
"learning_rate": 8.687937778524786e-05,
|
| 240 |
+
"loss": 0.0174,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 1.8085106382978724,
|
| 245 |
+
"grad_norm": 0.22281178832054138,
|
| 246 |
+
"learning_rate": 8.59350693841912e-05,
|
| 247 |
+
"loss": 0.0169,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 1.8617021276595744,
|
| 252 |
+
"grad_norm": 0.19580762088298798,
|
| 253 |
+
"learning_rate": 8.49634723963284e-05,
|
| 254 |
+
"loss": 0.0149,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 1.9148936170212765,
|
| 259 |
+
"grad_norm": 0.21648390591144562,
|
| 260 |
+
"learning_rate": 8.396532463876124e-05,
|
| 261 |
+
"loss": 0.0143,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 1.9680851063829787,
|
| 266 |
+
"grad_norm": 0.26432013511657715,
|
| 267 |
+
"learning_rate": 8.29413840908729e-05,
|
| 268 |
+
"loss": 0.0152,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 2.021276595744681,
|
| 273 |
+
"grad_norm": 0.2518749237060547,
|
| 274 |
+
"learning_rate": 8.18924283187292e-05,
|
| 275 |
+
"loss": 0.0159,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 2.074468085106383,
|
| 280 |
+
"grad_norm": 0.20243488252162933,
|
| 281 |
+
"learning_rate": 8.081925388460578e-05,
|
| 282 |
+
"loss": 0.0176,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 2.127659574468085,
|
| 287 |
+
"grad_norm": 0.19186046719551086,
|
| 288 |
+
"learning_rate": 7.972267574208991e-05,
|
| 289 |
+
"loss": 0.0142,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 2.1808510638297873,
|
| 294 |
+
"grad_norm": 0.16403602063655853,
|
| 295 |
+
"learning_rate": 7.860352661721619e-05,
|
| 296 |
+
"loss": 0.0134,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 2.2340425531914896,
|
| 301 |
+
"grad_norm": 0.15480686724185944,
|
| 302 |
+
"learning_rate": 7.746265637610613e-05,
|
| 303 |
+
"loss": 0.0131,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 2.2872340425531914,
|
| 308 |
+
"grad_norm": 0.18170325458049774,
|
| 309 |
+
"learning_rate": 7.630093137959171e-05,
|
| 310 |
+
"loss": 0.0169,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 2.3404255319148937,
|
| 315 |
+
"grad_norm": 0.241190105676651,
|
| 316 |
+
"learning_rate": 7.511923382531317e-05,
|
| 317 |
+
"loss": 0.0149,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 2.393617021276596,
|
| 322 |
+
"grad_norm": 0.2991958558559418,
|
| 323 |
+
"learning_rate": 7.391846107779047e-05,
|
| 324 |
+
"loss": 0.0166,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 2.4468085106382977,
|
| 329 |
+
"grad_norm": 0.21235980093479156,
|
| 330 |
+
"learning_rate": 7.269952498697734e-05,
|
| 331 |
+
"loss": 0.0149,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 2.5,
|
| 336 |
+
"grad_norm": 0.2054467499256134,
|
| 337 |
+
"learning_rate": 7.146335119581523e-05,
|
| 338 |
+
"loss": 0.0138,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 2.5531914893617023,
|
| 343 |
+
"grad_norm": 0.2078186273574829,
|
| 344 |
+
"learning_rate": 7.021087843731302e-05,
|
| 345 |
+
"loss": 0.0136,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 2.6063829787234045,
|
| 350 |
+
"grad_norm": 0.24719053506851196,
|
| 351 |
+
"learning_rate": 6.894305782168638e-05,
|
| 352 |
+
"loss": 0.0117,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 2.6595744680851063,
|
| 357 |
+
"grad_norm": 0.22392000257968903,
|
| 358 |
+
"learning_rate": 6.76608521140981e-05,
|
| 359 |
+
"loss": 0.0118,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 2.7127659574468086,
|
| 364 |
+
"grad_norm": 0.21282769739627838,
|
| 365 |
+
"learning_rate": 6.636523500354759e-05,
|
| 366 |
+
"loss": 0.0137,
|
| 367 |
+
"step": 510
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 2.7659574468085104,
|
| 371 |
+
"grad_norm": 0.24844494462013245,
|
| 372 |
+
"learning_rate": 6.505719036346539e-05,
|
| 373 |
+
"loss": 0.0154,
|
| 374 |
+
"step": 520
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 2.8191489361702127,
|
| 378 |
+
"grad_norm": 0.292816162109375,
|
| 379 |
+
"learning_rate": 6.37377115045732e-05,
|
| 380 |
+
"loss": 0.0124,
|
| 381 |
+
"step": 530
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 2.872340425531915,
|
| 385 |
+
"grad_norm": 0.25602397322654724,
|
| 386 |
+
"learning_rate": 6.240780042057796e-05,
|
| 387 |
+
"loss": 0.0133,
|
| 388 |
+
"step": 540
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 2.925531914893617,
|
| 392 |
+
"grad_norm": 0.18833638727664948,
|
| 393 |
+
"learning_rate": 6.106846702727172e-05,
|
| 394 |
+
"loss": 0.014,
|
| 395 |
+
"step": 550
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 2.978723404255319,
|
| 399 |
+
"grad_norm": 0.15500977635383606,
|
| 400 |
+
"learning_rate": 5.9720728395615875e-05,
|
| 401 |
+
"loss": 0.012,
|
| 402 |
+
"step": 560
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 3.0319148936170213,
|
| 406 |
+
"grad_norm": 0.17720523476600647,
|
| 407 |
+
"learning_rate": 5.8365607979391666e-05,
|
| 408 |
+
"loss": 0.0104,
|
| 409 |
+
"step": 570
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 3.0851063829787235,
|
| 413 |
+
"grad_norm": 0.1569543480873108,
|
| 414 |
+
"learning_rate": 5.70041348380039e-05,
|
| 415 |
+
"loss": 0.0127,
|
| 416 |
+
"step": 580
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 3.1382978723404253,
|
| 420 |
+
"grad_norm": 0.20494183897972107,
|
| 421 |
+
"learning_rate": 5.5637342855027554e-05,
|
| 422 |
+
"loss": 0.0106,
|
| 423 |
+
"step": 590
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 3.1914893617021276,
|
| 427 |
+
"grad_norm": 0.14022426307201385,
|
| 428 |
+
"learning_rate": 5.426626995309123e-05,
|
| 429 |
+
"loss": 0.0104,
|
| 430 |
+
"step": 600
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 3.24468085106383,
|
| 434 |
+
"grad_norm": 0.13388872146606445,
|
| 435 |
+
"learning_rate": 5.2891957305693205e-05,
|
| 436 |
+
"loss": 0.0115,
|
| 437 |
+
"step": 610
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 3.297872340425532,
|
| 441 |
+
"grad_norm": 0.21251560747623444,
|
| 442 |
+
"learning_rate": 5.151544854654895e-05,
|
| 443 |
+
"loss": 0.0092,
|
| 444 |
+
"step": 620
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 3.351063829787234,
|
| 448 |
+
"grad_norm": 0.16313986480236053,
|
| 449 |
+
"learning_rate": 5.0137788977070353e-05,
|
| 450 |
+
"loss": 0.0111,
|
| 451 |
+
"step": 630
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 3.404255319148936,
|
| 455 |
+
"grad_norm": 0.14321237802505493,
|
| 456 |
+
"learning_rate": 4.87600247725785e-05,
|
| 457 |
+
"loss": 0.0119,
|
| 458 |
+
"step": 640
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 3.4574468085106385,
|
| 462 |
+
"grad_norm": 0.14059048891067505,
|
| 463 |
+
"learning_rate": 4.738320218785281e-05,
|
| 464 |
+
"loss": 0.0095,
|
| 465 |
+
"step": 650
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 3.5106382978723403,
|
| 469 |
+
"grad_norm": 0.18556830286979675,
|
| 470 |
+
"learning_rate": 4.6008366762619926e-05,
|
| 471 |
+
"loss": 0.0105,
|
| 472 |
+
"step": 660
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 3.5638297872340425,
|
| 476 |
+
"grad_norm": 0.16162021458148956,
|
| 477 |
+
"learning_rate": 4.463656252758542e-05,
|
| 478 |
+
"loss": 0.0097,
|
| 479 |
+
"step": 670
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 3.617021276595745,
|
| 483 |
+
"grad_norm": 0.16970565915107727,
|
| 484 |
+
"learning_rate": 4.3268831211611707e-05,
|
| 485 |
+
"loss": 0.0123,
|
| 486 |
+
"step": 680
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 3.670212765957447,
|
| 490 |
+
"grad_norm": 0.1753172129392624,
|
| 491 |
+
"learning_rate": 4.190621145064363e-05,
|
| 492 |
+
"loss": 0.0103,
|
| 493 |
+
"step": 690
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 3.723404255319149,
|
| 497 |
+
"grad_norm": 0.1282825917005539,
|
| 498 |
+
"learning_rate": 4.0549737998983e-05,
|
| 499 |
+
"loss": 0.0083,
|
| 500 |
+
"step": 700
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 3.776595744680851,
|
| 504 |
+
"grad_norm": 0.18947535753250122,
|
| 505 |
+
"learning_rate": 3.9200440943510665e-05,
|
| 506 |
+
"loss": 0.0101,
|
| 507 |
+
"step": 710
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 3.829787234042553,
|
| 511 |
+
"grad_norm": 0.14048849046230316,
|
| 512 |
+
"learning_rate": 3.7859344921453064e-05,
|
| 513 |
+
"loss": 0.0101,
|
| 514 |
+
"step": 720
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 3.882978723404255,
|
| 518 |
+
"grad_norm": 0.1769946664571762,
|
| 519 |
+
"learning_rate": 3.65274683422871e-05,
|
| 520 |
+
"loss": 0.0134,
|
| 521 |
+
"step": 730
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 3.9361702127659575,
|
| 525 |
+
"grad_norm": 0.13211847841739655,
|
| 526 |
+
"learning_rate": 3.5205822614374255e-05,
|
| 527 |
+
"loss": 0.0097,
|
| 528 |
+
"step": 740
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 3.9893617021276597,
|
| 532 |
+
"grad_norm": 0.17846064269542694,
|
| 533 |
+
"learning_rate": 3.389541137691129e-05,
|
| 534 |
+
"loss": 0.0099,
|
| 535 |
+
"step": 750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 4.042553191489362,
|
| 539 |
+
"grad_norm": 0.15190204977989197,
|
| 540 |
+
"learning_rate": 3.2597229737780774e-05,
|
| 541 |
+
"loss": 0.0093,
|
| 542 |
+
"step": 760
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 4.095744680851064,
|
| 546 |
+
"grad_norm": 0.18606022000312805,
|
| 547 |
+
"learning_rate": 3.131226351788018e-05,
|
| 548 |
+
"loss": 0.0103,
|
| 549 |
+
"step": 770
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 4.148936170212766,
|
| 553 |
+
"grad_norm": 0.1476437896490097,
|
| 554 |
+
"learning_rate": 3.004148850250323e-05,
|
| 555 |
+
"loss": 0.0092,
|
| 556 |
+
"step": 780
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 4.202127659574468,
|
| 560 |
+
"grad_norm": 0.12311765551567078,
|
| 561 |
+
"learning_rate": 2.878586970034232e-05,
|
| 562 |
+
"loss": 0.0103,
|
| 563 |
+
"step": 790
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 4.25531914893617,
|
| 567 |
+
"grad_norm": 0.1675160676240921,
|
| 568 |
+
"learning_rate": 2.7546360610674493e-05,
|
| 569 |
+
"loss": 0.0109,
|
| 570 |
+
"step": 800
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 4.308510638297872,
|
| 574 |
+
"grad_norm": 0.09888270497322083,
|
| 575 |
+
"learning_rate": 2.6323902499287473e-05,
|
| 576 |
+
"loss": 0.0089,
|
| 577 |
+
"step": 810
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 4.361702127659575,
|
| 581 |
+
"grad_norm": 0.15007852017879486,
|
| 582 |
+
"learning_rate": 2.511942368369566e-05,
|
| 583 |
+
"loss": 0.0101,
|
| 584 |
+
"step": 820
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 4.414893617021277,
|
| 588 |
+
"grad_norm": 0.1582082360982895,
|
| 589 |
+
"learning_rate": 2.3933838828188733e-05,
|
| 590 |
+
"loss": 0.0104,
|
| 591 |
+
"step": 830
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 4.468085106382979,
|
| 595 |
+
"grad_norm": 0.10153388231992722,
|
| 596 |
+
"learning_rate": 2.2768048249248648e-05,
|
| 597 |
+
"loss": 0.009,
|
| 598 |
+
"step": 840
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 4.5212765957446805,
|
| 602 |
+
"grad_norm": 0.13839617371559143,
|
| 603 |
+
"learning_rate": 2.1622937231861822e-05,
|
| 604 |
+
"loss": 0.0089,
|
| 605 |
+
"step": 850
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 4.574468085106383,
|
| 609 |
+
"grad_norm": 0.1278831511735916,
|
| 610 |
+
"learning_rate": 2.0499375357246148e-05,
|
| 611 |
+
"loss": 0.0083,
|
| 612 |
+
"step": 860
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 4.627659574468085,
|
| 616 |
+
"grad_norm": 0.19818226993083954,
|
| 617 |
+
"learning_rate": 1.939821584250326e-05,
|
| 618 |
+
"loss": 0.0084,
|
| 619 |
+
"step": 870
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 4.680851063829787,
|
| 623 |
+
"grad_norm": 0.13401909172534943,
|
| 624 |
+
"learning_rate": 1.8320294892697478e-05,
|
| 625 |
+
"loss": 0.0069,
|
| 626 |
+
"step": 880
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 4.73404255319149,
|
| 630 |
+
"grad_norm": 0.08135835826396942,
|
| 631 |
+
"learning_rate": 1.7266431065853416e-05,
|
| 632 |
+
"loss": 0.0072,
|
| 633 |
+
"step": 890
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 4.787234042553192,
|
| 637 |
+
"grad_norm": 0.09311998635530472,
|
| 638 |
+
"learning_rate": 1.623742465135449e-05,
|
| 639 |
+
"loss": 0.008,
|
| 640 |
+
"step": 900
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 4.840425531914894,
|
| 644 |
+
"grad_norm": 0.1272549033164978,
|
| 645 |
+
"learning_rate": 1.5234057062214402e-05,
|
| 646 |
+
"loss": 0.0095,
|
| 647 |
+
"step": 910
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 4.8936170212765955,
|
| 651 |
+
"grad_norm": 0.0637887716293335,
|
| 652 |
+
"learning_rate": 1.4257090241683152e-05,
|
| 653 |
+
"loss": 0.0084,
|
| 654 |
+
"step": 920
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 4.946808510638298,
|
| 658 |
+
"grad_norm": 0.09416159242391586,
|
| 659 |
+
"learning_rate": 1.330726608463797e-05,
|
| 660 |
+
"loss": 0.0087,
|
| 661 |
+
"step": 930
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 5.0,
|
| 665 |
+
"grad_norm": 0.12856261432170868,
|
| 666 |
+
"learning_rate": 1.2385305874198776e-05,
|
| 667 |
+
"loss": 0.0082,
|
| 668 |
+
"step": 940
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 5.053191489361702,
|
| 672 |
+
"grad_norm": 0.10889431834220886,
|
| 673 |
+
"learning_rate": 1.1491909733995898e-05,
|
| 674 |
+
"loss": 0.0097,
|
| 675 |
+
"step": 950
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 5.1063829787234045,
|
| 679 |
+
"grad_norm": 0.12346695363521576,
|
| 680 |
+
"learning_rate": 1.0627756096506102e-05,
|
| 681 |
+
"loss": 0.0098,
|
| 682 |
+
"step": 960
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 5.159574468085107,
|
| 686 |
+
"grad_norm": 0.10313884168863297,
|
| 687 |
+
"learning_rate": 9.793501187860432e-06,
|
| 688 |
+
"loss": 0.0078,
|
| 689 |
+
"step": 970
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 5.212765957446808,
|
| 693 |
+
"grad_norm": 0.10387715697288513,
|
| 694 |
+
"learning_rate": 8.989778529515298e-06,
|
| 695 |
+
"loss": 0.008,
|
| 696 |
+
"step": 980
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 5.26595744680851,
|
| 700 |
+
"grad_norm": 0.10801278799772263,
|
| 701 |
+
"learning_rate": 8.21719845716521e-06,
|
| 702 |
+
"loss": 0.0107,
|
| 703 |
+
"step": 990
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 5.319148936170213,
|
| 707 |
+
"grad_norm": 0.07922052592039108,
|
| 708 |
+
"learning_rate": 7.476347657262456e-06,
|
| 709 |
+
"loss": 0.0082,
|
| 710 |
+
"step": 1000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 5.372340425531915,
|
| 714 |
+
"grad_norm": 0.08209902048110962,
|
| 715 |
+
"learning_rate": 6.7677887214956e-06,
|
| 716 |
+
"loss": 0.0076,
|
| 717 |
+
"step": 1010
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 5.425531914893617,
|
| 721 |
+
"grad_norm": 0.07317307591438293,
|
| 722 |
+
"learning_rate": 6.092059719565274e-06,
|
| 723 |
+
"loss": 0.0077,
|
| 724 |
+
"step": 1020
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 5.4787234042553195,
|
| 728 |
+
"grad_norm": 0.06752776354551315,
|
| 729 |
+
"learning_rate": 5.449673790581611e-06,
|
| 730 |
+
"loss": 0.0077,
|
| 731 |
+
"step": 1030
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 5.531914893617021,
|
| 735 |
+
"grad_norm": 0.05100482329726219,
|
| 736 |
+
"learning_rate": 4.8411187533936195e-06,
|
| 737 |
+
"loss": 0.0069,
|
| 738 |
+
"step": 1040
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 5.585106382978723,
|
| 742 |
+
"grad_norm": 0.07704073190689087,
|
| 743 |
+
"learning_rate": 4.266856736146424e-06,
|
| 744 |
+
"loss": 0.0069,
|
| 745 |
+
"step": 1050
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 5.638297872340425,
|
| 749 |
+
"grad_norm": 0.060355961322784424,
|
| 750 |
+
"learning_rate": 3.7273238253475785e-06,
|
| 751 |
+
"loss": 0.0083,
|
| 752 |
+
"step": 1060
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 5.691489361702128,
|
| 756 |
+
"grad_norm": 0.05728999152779579,
|
| 757 |
+
"learning_rate": 3.2229297347091514e-06,
|
| 758 |
+
"loss": 0.007,
|
| 759 |
+
"step": 1070
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 5.74468085106383,
|
| 763 |
+
"grad_norm": 0.05493322014808655,
|
| 764 |
+
"learning_rate": 2.754057494016904e-06,
|
| 765 |
+
"loss": 0.0089,
|
| 766 |
+
"step": 1080
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 5.797872340425532,
|
| 770 |
+
"grad_norm": 0.049081362783908844,
|
| 771 |
+
"learning_rate": 2.321063158262793e-06,
|
| 772 |
+
"loss": 0.0068,
|
| 773 |
+
"step": 1090
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 5.851063829787234,
|
| 777 |
+
"grad_norm": 0.10410504788160324,
|
| 778 |
+
"learning_rate": 1.924275537261877e-06,
|
| 779 |
+
"loss": 0.0082,
|
| 780 |
+
"step": 1100
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 5.904255319148936,
|
| 784 |
+
"grad_norm": 0.06362517923116684,
|
| 785 |
+
"learning_rate": 1.5639959459586962e-06,
|
| 786 |
+
"loss": 0.0083,
|
| 787 |
+
"step": 1110
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 5.957446808510638,
|
| 791 |
+
"grad_norm": 0.0873829796910286,
|
| 792 |
+
"learning_rate": 1.2404979756130142e-06,
|
| 793 |
+
"loss": 0.0086,
|
| 794 |
+
"step": 1120
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 6.01063829787234,
|
| 798 |
+
"grad_norm": 0.07028144598007202,
|
| 799 |
+
"learning_rate": 9.540272860383837e-07,
|
| 800 |
+
"loss": 0.0079,
|
| 801 |
+
"step": 1130
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 6.0638297872340425,
|
| 805 |
+
"grad_norm": 0.043298330157995224,
|
| 806 |
+
"learning_rate": 7.048014190515872e-07,
|
| 807 |
+
"loss": 0.008,
|
| 808 |
+
"step": 1140
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 6.117021276595745,
|
| 812 |
+
"grad_norm": 0.09406614303588867,
|
| 813 |
+
"learning_rate": 4.930096332744105e-07,
|
| 814 |
+
"loss": 0.0108,
|
| 815 |
+
"step": 1150
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 6.170212765957447,
|
| 819 |
+
"grad_norm": 0.08257335424423218,
|
| 820 |
+
"learning_rate": 3.188127604133284e-07,
|
| 821 |
+
"loss": 0.0074,
|
| 822 |
+
"step": 1160
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 6.223404255319149,
|
| 826 |
+
"grad_norm": 0.06063027307391167,
|
| 827 |
+
"learning_rate": 1.8234308312612968e-07,
|
| 828 |
+
"loss": 0.0067,
|
| 829 |
+
"step": 1170
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 6.276595744680851,
|
| 833 |
+
"grad_norm": 0.08021627366542816,
|
| 834 |
+
"learning_rate": 8.37042345683714e-08,
|
| 835 |
+
"loss": 0.0065,
|
| 836 |
+
"step": 1180
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 6.329787234042553,
|
| 840 |
+
"grad_norm": 0.06200169771909714,
|
| 841 |
+
"learning_rate": 2.2971119695774925e-08,
|
| 842 |
+
"loss": 0.0081,
|
| 843 |
+
"step": 1190
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 6.382978723404255,
|
| 847 |
+
"grad_norm": 0.06996981799602509,
|
| 848 |
+
"learning_rate": 1.8985838248219978e-10,
|
| 849 |
+
"loss": 0.0079,
|
| 850 |
+
"step": 1200
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 6.382978723404255,
|
| 854 |
+
"step": 1200,
|
| 855 |
+
"total_flos": 0.0,
|
| 856 |
+
"train_loss": 0.024982591159641743,
|
| 857 |
+
"train_runtime": 2390.4851,
|
| 858 |
+
"train_samples_per_second": 200.796,
|
| 859 |
+
"train_steps_per_second": 0.502
|
| 860 |
+
}
|
| 861 |
+
],
|
| 862 |
+
"logging_steps": 10,
|
| 863 |
+
"max_steps": 1200,
|
| 864 |
+
"num_input_tokens_seen": 0,
|
| 865 |
+
"num_train_epochs": 7,
|
| 866 |
+
"save_steps": 800,
|
| 867 |
+
"stateful_callbacks": {
|
| 868 |
+
"TrainerControl": {
|
| 869 |
+
"args": {
|
| 870 |
+
"should_epoch_stop": false,
|
| 871 |
+
"should_evaluate": false,
|
| 872 |
+
"should_log": false,
|
| 873 |
+
"should_save": true,
|
| 874 |
+
"should_training_stop": true
|
| 875 |
+
},
|
| 876 |
+
"attributes": {}
|
| 877 |
+
}
|
| 878 |
+
},
|
| 879 |
+
"total_flos": 0.0,
|
| 880 |
+
"train_batch_size": 50,
|
| 881 |
+
"trial_name": null,
|
| 882 |
+
"trial_params": null
|
| 883 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:071492e7dcc463f7ccce7628156854d9fa443e26fb41901ec4d684d7aaf33629
|
| 3 |
+
size 5713
|