End of training
Browse files- last-checkpoint/config.json +0 -75
- last-checkpoint/optimizer.pt +0 -3
- last-checkpoint/pytorch_model.bin +0 -3
- last-checkpoint/rng_state.pth +0 -3
- last-checkpoint/scheduler.pt +0 -3
- last-checkpoint/trainer_state.json +0 -1816
- last-checkpoint/training_args.bin +0 -3
- runs/Jan03_01-47-25_21bb7eba274c/events.out.tfevents.1672710460.21bb7eba274c.23.0 +2 -2
last-checkpoint/config.json
DELETED
|
@@ -1,75 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"_name_or_path": "facebook/bart-base",
|
| 3 |
-
"activation_dropout": 0.1,
|
| 4 |
-
"activation_function": "gelu",
|
| 5 |
-
"add_bias_logits": false,
|
| 6 |
-
"add_final_layer_norm": false,
|
| 7 |
-
"architectures": [
|
| 8 |
-
"BartForConditionalGeneration"
|
| 9 |
-
],
|
| 10 |
-
"attention_dropout": 0.1,
|
| 11 |
-
"bos_token_id": 0,
|
| 12 |
-
"classif_dropout": 0.1,
|
| 13 |
-
"classifier_dropout": 0.0,
|
| 14 |
-
"d_model": 768,
|
| 15 |
-
"decoder_attention_heads": 12,
|
| 16 |
-
"decoder_ffn_dim": 3072,
|
| 17 |
-
"decoder_layerdrop": 0.0,
|
| 18 |
-
"decoder_layers": 6,
|
| 19 |
-
"decoder_start_token_id": 2,
|
| 20 |
-
"dropout": 0.1,
|
| 21 |
-
"early_stopping": true,
|
| 22 |
-
"encoder_attention_heads": 12,
|
| 23 |
-
"encoder_ffn_dim": 3072,
|
| 24 |
-
"encoder_layerdrop": 0.0,
|
| 25 |
-
"encoder_layers": 6,
|
| 26 |
-
"eos_token_id": 2,
|
| 27 |
-
"forced_bos_token_id": 0,
|
| 28 |
-
"forced_eos_token_id": 2,
|
| 29 |
-
"gradient_checkpointing": false,
|
| 30 |
-
"id2label": {
|
| 31 |
-
"0": "LABEL_0",
|
| 32 |
-
"1": "LABEL_1",
|
| 33 |
-
"2": "LABEL_2"
|
| 34 |
-
},
|
| 35 |
-
"init_std": 0.02,
|
| 36 |
-
"is_encoder_decoder": true,
|
| 37 |
-
"label2id": {
|
| 38 |
-
"LABEL_0": 0,
|
| 39 |
-
"LABEL_1": 1,
|
| 40 |
-
"LABEL_2": 2
|
| 41 |
-
},
|
| 42 |
-
"max_position_embeddings": 1024,
|
| 43 |
-
"model_type": "bart",
|
| 44 |
-
"no_repeat_ngram_size": 3,
|
| 45 |
-
"normalize_before": false,
|
| 46 |
-
"normalize_embedding": true,
|
| 47 |
-
"num_beams": 4,
|
| 48 |
-
"num_hidden_layers": 6,
|
| 49 |
-
"pad_token_id": 1,
|
| 50 |
-
"scale_embedding": false,
|
| 51 |
-
"task_specific_params": {
|
| 52 |
-
"summarization": {
|
| 53 |
-
"length_penalty": 1.0,
|
| 54 |
-
"max_length": 128,
|
| 55 |
-
"min_length": 12,
|
| 56 |
-
"num_beams": 4
|
| 57 |
-
},
|
| 58 |
-
"summarization_cnn": {
|
| 59 |
-
"length_penalty": 2.0,
|
| 60 |
-
"max_length": 142,
|
| 61 |
-
"min_length": 56,
|
| 62 |
-
"num_beams": 4
|
| 63 |
-
},
|
| 64 |
-
"summarization_xsum": {
|
| 65 |
-
"length_penalty": 1.0,
|
| 66 |
-
"max_length": 62,
|
| 67 |
-
"min_length": 11,
|
| 68 |
-
"num_beams": 6
|
| 69 |
-
}
|
| 70 |
-
},
|
| 71 |
-
"torch_dtype": "float32",
|
| 72 |
-
"transformers_version": "4.20.1",
|
| 73 |
-
"use_cache": true,
|
| 74 |
-
"vocab_size": 50265
|
| 75 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/optimizer.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:f5c2293da2b83801179dc88cb4ea8a83d0ded5a86ced9d169c06cc11594bea02
|
| 3 |
-
size 1115513717
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/pytorch_model.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:6ac15d1c531a2537f2ed5a5ca9b2c15212a32f681269a48c059475929253d03c
|
| 3 |
-
size 557969145
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/rng_state.pth
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:6d67f718eac66f9d6c498cfa6f6b1be1064b53488a37cdf5e844b2fef7504e4c
|
| 3 |
-
size 15523
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/scheduler.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:e065087a5984c84e0b749307b78503c25e808e39abb0844af949250ee8a8bed8
|
| 3 |
-
size 623
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/trainer_state.json
DELETED
|
@@ -1,1816 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_metric": null,
|
| 3 |
-
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 1.6816143497757847,
|
| 5 |
-
"global_step": 30000,
|
| 6 |
-
"is_hyper_param_search": false,
|
| 7 |
-
"is_local_process_zero": true,
|
| 8 |
-
"is_world_process_zero": true,
|
| 9 |
-
"log_history": [
|
| 10 |
-
{
|
| 11 |
-
"epoch": 0.01,
|
| 12 |
-
"learning_rate": 4.966367713004485e-05,
|
| 13 |
-
"loss": 1.0008,
|
| 14 |
-
"step": 100
|
| 15 |
-
},
|
| 16 |
-
{
|
| 17 |
-
"epoch": 0.01,
|
| 18 |
-
"learning_rate": 4.9352267065271554e-05,
|
| 19 |
-
"loss": 0.9438,
|
| 20 |
-
"step": 200
|
| 21 |
-
},
|
| 22 |
-
{
|
| 23 |
-
"epoch": 0.02,
|
| 24 |
-
"learning_rate": 4.904085700049826e-05,
|
| 25 |
-
"loss": 0.912,
|
| 26 |
-
"step": 300
|
| 27 |
-
},
|
| 28 |
-
{
|
| 29 |
-
"epoch": 0.02,
|
| 30 |
-
"learning_rate": 4.872944693572497e-05,
|
| 31 |
-
"loss": 0.8963,
|
| 32 |
-
"step": 400
|
| 33 |
-
},
|
| 34 |
-
{
|
| 35 |
-
"epoch": 0.03,
|
| 36 |
-
"learning_rate": 4.8418036870951675e-05,
|
| 37 |
-
"loss": 0.8806,
|
| 38 |
-
"step": 500
|
| 39 |
-
},
|
| 40 |
-
{
|
| 41 |
-
"epoch": 0.03,
|
| 42 |
-
"learning_rate": 4.8106626806178375e-05,
|
| 43 |
-
"loss": 0.8355,
|
| 44 |
-
"step": 600
|
| 45 |
-
},
|
| 46 |
-
{
|
| 47 |
-
"epoch": 0.04,
|
| 48 |
-
"learning_rate": 4.779521674140508e-05,
|
| 49 |
-
"loss": 0.8538,
|
| 50 |
-
"step": 700
|
| 51 |
-
},
|
| 52 |
-
{
|
| 53 |
-
"epoch": 0.04,
|
| 54 |
-
"learning_rate": 4.748380667663179e-05,
|
| 55 |
-
"loss": 0.8787,
|
| 56 |
-
"step": 800
|
| 57 |
-
},
|
| 58 |
-
{
|
| 59 |
-
"epoch": 0.05,
|
| 60 |
-
"learning_rate": 4.7172396611858496e-05,
|
| 61 |
-
"loss": 0.8817,
|
| 62 |
-
"step": 900
|
| 63 |
-
},
|
| 64 |
-
{
|
| 65 |
-
"epoch": 0.06,
|
| 66 |
-
"learning_rate": 4.6860986547085203e-05,
|
| 67 |
-
"loss": 0.8483,
|
| 68 |
-
"step": 1000
|
| 69 |
-
},
|
| 70 |
-
{
|
| 71 |
-
"epoch": 0.06,
|
| 72 |
-
"learning_rate": 4.654957648231191e-05,
|
| 73 |
-
"loss": 0.8328,
|
| 74 |
-
"step": 1100
|
| 75 |
-
},
|
| 76 |
-
{
|
| 77 |
-
"epoch": 0.07,
|
| 78 |
-
"learning_rate": 4.623816641753862e-05,
|
| 79 |
-
"loss": 0.9183,
|
| 80 |
-
"step": 1200
|
| 81 |
-
},
|
| 82 |
-
{
|
| 83 |
-
"epoch": 0.07,
|
| 84 |
-
"learning_rate": 4.5926756352765324e-05,
|
| 85 |
-
"loss": 0.8237,
|
| 86 |
-
"step": 1300
|
| 87 |
-
},
|
| 88 |
-
{
|
| 89 |
-
"epoch": 0.08,
|
| 90 |
-
"learning_rate": 4.5615346287992025e-05,
|
| 91 |
-
"loss": 0.8074,
|
| 92 |
-
"step": 1400
|
| 93 |
-
},
|
| 94 |
-
{
|
| 95 |
-
"epoch": 0.08,
|
| 96 |
-
"learning_rate": 4.530393622321873e-05,
|
| 97 |
-
"loss": 0.881,
|
| 98 |
-
"step": 1500
|
| 99 |
-
},
|
| 100 |
-
{
|
| 101 |
-
"epoch": 0.09,
|
| 102 |
-
"learning_rate": 4.499252615844544e-05,
|
| 103 |
-
"loss": 0.85,
|
| 104 |
-
"step": 1600
|
| 105 |
-
},
|
| 106 |
-
{
|
| 107 |
-
"epoch": 0.1,
|
| 108 |
-
"learning_rate": 4.4681116093672146e-05,
|
| 109 |
-
"loss": 0.8827,
|
| 110 |
-
"step": 1700
|
| 111 |
-
},
|
| 112 |
-
{
|
| 113 |
-
"epoch": 0.1,
|
| 114 |
-
"learning_rate": 4.436970602889885e-05,
|
| 115 |
-
"loss": 0.9127,
|
| 116 |
-
"step": 1800
|
| 117 |
-
},
|
| 118 |
-
{
|
| 119 |
-
"epoch": 0.11,
|
| 120 |
-
"learning_rate": 4.405829596412556e-05,
|
| 121 |
-
"loss": 0.9172,
|
| 122 |
-
"step": 1900
|
| 123 |
-
},
|
| 124 |
-
{
|
| 125 |
-
"epoch": 0.11,
|
| 126 |
-
"learning_rate": 4.374688589935227e-05,
|
| 127 |
-
"loss": 0.8163,
|
| 128 |
-
"step": 2000
|
| 129 |
-
},
|
| 130 |
-
{
|
| 131 |
-
"epoch": 0.12,
|
| 132 |
-
"learning_rate": 4.3435475834578974e-05,
|
| 133 |
-
"loss": 0.8268,
|
| 134 |
-
"step": 2100
|
| 135 |
-
},
|
| 136 |
-
{
|
| 137 |
-
"epoch": 0.12,
|
| 138 |
-
"learning_rate": 4.312406576980568e-05,
|
| 139 |
-
"loss": 0.7457,
|
| 140 |
-
"step": 2200
|
| 141 |
-
},
|
| 142 |
-
{
|
| 143 |
-
"epoch": 0.13,
|
| 144 |
-
"learning_rate": 4.281265570503239e-05,
|
| 145 |
-
"loss": 0.8587,
|
| 146 |
-
"step": 2300
|
| 147 |
-
},
|
| 148 |
-
{
|
| 149 |
-
"epoch": 0.13,
|
| 150 |
-
"learning_rate": 4.2501245640259095e-05,
|
| 151 |
-
"loss": 0.8008,
|
| 152 |
-
"step": 2400
|
| 153 |
-
},
|
| 154 |
-
{
|
| 155 |
-
"epoch": 0.14,
|
| 156 |
-
"learning_rate": 4.21898355754858e-05,
|
| 157 |
-
"loss": 0.7494,
|
| 158 |
-
"step": 2500
|
| 159 |
-
},
|
| 160 |
-
{
|
| 161 |
-
"epoch": 0.15,
|
| 162 |
-
"learning_rate": 4.187842551071251e-05,
|
| 163 |
-
"loss": 0.7831,
|
| 164 |
-
"step": 2600
|
| 165 |
-
},
|
| 166 |
-
{
|
| 167 |
-
"epoch": 0.15,
|
| 168 |
-
"learning_rate": 4.1567015445939216e-05,
|
| 169 |
-
"loss": 0.8458,
|
| 170 |
-
"step": 2700
|
| 171 |
-
},
|
| 172 |
-
{
|
| 173 |
-
"epoch": 0.16,
|
| 174 |
-
"learning_rate": 4.125560538116592e-05,
|
| 175 |
-
"loss": 0.8128,
|
| 176 |
-
"step": 2800
|
| 177 |
-
},
|
| 178 |
-
{
|
| 179 |
-
"epoch": 0.16,
|
| 180 |
-
"learning_rate": 4.094419531639262e-05,
|
| 181 |
-
"loss": 0.8544,
|
| 182 |
-
"step": 2900
|
| 183 |
-
},
|
| 184 |
-
{
|
| 185 |
-
"epoch": 0.17,
|
| 186 |
-
"learning_rate": 4.063278525161933e-05,
|
| 187 |
-
"loss": 0.8352,
|
| 188 |
-
"step": 3000
|
| 189 |
-
},
|
| 190 |
-
{
|
| 191 |
-
"epoch": 0.17,
|
| 192 |
-
"learning_rate": 4.032137518684604e-05,
|
| 193 |
-
"loss": 0.8429,
|
| 194 |
-
"step": 3100
|
| 195 |
-
},
|
| 196 |
-
{
|
| 197 |
-
"epoch": 0.18,
|
| 198 |
-
"learning_rate": 4.0009965122072744e-05,
|
| 199 |
-
"loss": 0.7716,
|
| 200 |
-
"step": 3200
|
| 201 |
-
},
|
| 202 |
-
{
|
| 203 |
-
"epoch": 0.18,
|
| 204 |
-
"learning_rate": 3.969855505729945e-05,
|
| 205 |
-
"loss": 0.7816,
|
| 206 |
-
"step": 3300
|
| 207 |
-
},
|
| 208 |
-
{
|
| 209 |
-
"epoch": 0.19,
|
| 210 |
-
"learning_rate": 3.938714499252616e-05,
|
| 211 |
-
"loss": 0.7572,
|
| 212 |
-
"step": 3400
|
| 213 |
-
},
|
| 214 |
-
{
|
| 215 |
-
"epoch": 0.2,
|
| 216 |
-
"learning_rate": 3.9075734927752865e-05,
|
| 217 |
-
"loss": 0.83,
|
| 218 |
-
"step": 3500
|
| 219 |
-
},
|
| 220 |
-
{
|
| 221 |
-
"epoch": 0.2,
|
| 222 |
-
"learning_rate": 3.876432486297957e-05,
|
| 223 |
-
"loss": 0.785,
|
| 224 |
-
"step": 3600
|
| 225 |
-
},
|
| 226 |
-
{
|
| 227 |
-
"epoch": 0.21,
|
| 228 |
-
"learning_rate": 3.845291479820628e-05,
|
| 229 |
-
"loss": 0.8274,
|
| 230 |
-
"step": 3700
|
| 231 |
-
},
|
| 232 |
-
{
|
| 233 |
-
"epoch": 0.21,
|
| 234 |
-
"learning_rate": 3.8141504733432986e-05,
|
| 235 |
-
"loss": 0.73,
|
| 236 |
-
"step": 3800
|
| 237 |
-
},
|
| 238 |
-
{
|
| 239 |
-
"epoch": 0.22,
|
| 240 |
-
"learning_rate": 3.783009466865969e-05,
|
| 241 |
-
"loss": 0.8378,
|
| 242 |
-
"step": 3900
|
| 243 |
-
},
|
| 244 |
-
{
|
| 245 |
-
"epoch": 0.22,
|
| 246 |
-
"learning_rate": 3.75186846038864e-05,
|
| 247 |
-
"loss": 0.7528,
|
| 248 |
-
"step": 4000
|
| 249 |
-
},
|
| 250 |
-
{
|
| 251 |
-
"epoch": 0.23,
|
| 252 |
-
"learning_rate": 3.720727453911311e-05,
|
| 253 |
-
"loss": 0.7561,
|
| 254 |
-
"step": 4100
|
| 255 |
-
},
|
| 256 |
-
{
|
| 257 |
-
"epoch": 0.24,
|
| 258 |
-
"learning_rate": 3.6895864474339814e-05,
|
| 259 |
-
"loss": 0.7529,
|
| 260 |
-
"step": 4200
|
| 261 |
-
},
|
| 262 |
-
{
|
| 263 |
-
"epoch": 0.24,
|
| 264 |
-
"learning_rate": 3.658445440956652e-05,
|
| 265 |
-
"loss": 0.7632,
|
| 266 |
-
"step": 4300
|
| 267 |
-
},
|
| 268 |
-
{
|
| 269 |
-
"epoch": 0.25,
|
| 270 |
-
"learning_rate": 3.627304434479323e-05,
|
| 271 |
-
"loss": 0.7931,
|
| 272 |
-
"step": 4400
|
| 273 |
-
},
|
| 274 |
-
{
|
| 275 |
-
"epoch": 0.25,
|
| 276 |
-
"learning_rate": 3.5961634280019935e-05,
|
| 277 |
-
"loss": 0.7812,
|
| 278 |
-
"step": 4500
|
| 279 |
-
},
|
| 280 |
-
{
|
| 281 |
-
"epoch": 0.26,
|
| 282 |
-
"learning_rate": 3.5650224215246636e-05,
|
| 283 |
-
"loss": 0.7401,
|
| 284 |
-
"step": 4600
|
| 285 |
-
},
|
| 286 |
-
{
|
| 287 |
-
"epoch": 0.26,
|
| 288 |
-
"learning_rate": 3.533881415047334e-05,
|
| 289 |
-
"loss": 0.823,
|
| 290 |
-
"step": 4700
|
| 291 |
-
},
|
| 292 |
-
{
|
| 293 |
-
"epoch": 0.27,
|
| 294 |
-
"learning_rate": 3.502740408570005e-05,
|
| 295 |
-
"loss": 0.7879,
|
| 296 |
-
"step": 4800
|
| 297 |
-
},
|
| 298 |
-
{
|
| 299 |
-
"epoch": 0.27,
|
| 300 |
-
"learning_rate": 3.4715994020926757e-05,
|
| 301 |
-
"loss": 0.7881,
|
| 302 |
-
"step": 4900
|
| 303 |
-
},
|
| 304 |
-
{
|
| 305 |
-
"epoch": 0.28,
|
| 306 |
-
"learning_rate": 3.4404583956153464e-05,
|
| 307 |
-
"loss": 0.7804,
|
| 308 |
-
"step": 5000
|
| 309 |
-
},
|
| 310 |
-
{
|
| 311 |
-
"epoch": 0.29,
|
| 312 |
-
"learning_rate": 3.409317389138017e-05,
|
| 313 |
-
"loss": 0.7806,
|
| 314 |
-
"step": 5100
|
| 315 |
-
},
|
| 316 |
-
{
|
| 317 |
-
"epoch": 0.29,
|
| 318 |
-
"learning_rate": 3.378176382660688e-05,
|
| 319 |
-
"loss": 0.8504,
|
| 320 |
-
"step": 5200
|
| 321 |
-
},
|
| 322 |
-
{
|
| 323 |
-
"epoch": 0.3,
|
| 324 |
-
"learning_rate": 3.3470353761833585e-05,
|
| 325 |
-
"loss": 0.7554,
|
| 326 |
-
"step": 5300
|
| 327 |
-
},
|
| 328 |
-
{
|
| 329 |
-
"epoch": 0.3,
|
| 330 |
-
"learning_rate": 3.315894369706029e-05,
|
| 331 |
-
"loss": 0.8439,
|
| 332 |
-
"step": 5400
|
| 333 |
-
},
|
| 334 |
-
{
|
| 335 |
-
"epoch": 0.31,
|
| 336 |
-
"learning_rate": 3.2847533632287e-05,
|
| 337 |
-
"loss": 0.7792,
|
| 338 |
-
"step": 5500
|
| 339 |
-
},
|
| 340 |
-
{
|
| 341 |
-
"epoch": 0.31,
|
| 342 |
-
"learning_rate": 3.2536123567513706e-05,
|
| 343 |
-
"loss": 0.8368,
|
| 344 |
-
"step": 5600
|
| 345 |
-
},
|
| 346 |
-
{
|
| 347 |
-
"epoch": 0.32,
|
| 348 |
-
"learning_rate": 3.222471350274041e-05,
|
| 349 |
-
"loss": 0.7269,
|
| 350 |
-
"step": 5700
|
| 351 |
-
},
|
| 352 |
-
{
|
| 353 |
-
"epoch": 0.33,
|
| 354 |
-
"learning_rate": 3.191330343796712e-05,
|
| 355 |
-
"loss": 0.8092,
|
| 356 |
-
"step": 5800
|
| 357 |
-
},
|
| 358 |
-
{
|
| 359 |
-
"epoch": 0.33,
|
| 360 |
-
"learning_rate": 3.160189337319383e-05,
|
| 361 |
-
"loss": 0.7401,
|
| 362 |
-
"step": 5900
|
| 363 |
-
},
|
| 364 |
-
{
|
| 365 |
-
"epoch": 0.34,
|
| 366 |
-
"learning_rate": 3.1290483308420534e-05,
|
| 367 |
-
"loss": 0.8564,
|
| 368 |
-
"step": 6000
|
| 369 |
-
},
|
| 370 |
-
{
|
| 371 |
-
"epoch": 0.34,
|
| 372 |
-
"learning_rate": 3.097907324364724e-05,
|
| 373 |
-
"loss": 0.7721,
|
| 374 |
-
"step": 6100
|
| 375 |
-
},
|
| 376 |
-
{
|
| 377 |
-
"epoch": 0.35,
|
| 378 |
-
"learning_rate": 3.066766317887394e-05,
|
| 379 |
-
"loss": 0.6934,
|
| 380 |
-
"step": 6200
|
| 381 |
-
},
|
| 382 |
-
{
|
| 383 |
-
"epoch": 0.35,
|
| 384 |
-
"learning_rate": 3.0356253114100648e-05,
|
| 385 |
-
"loss": 0.7353,
|
| 386 |
-
"step": 6300
|
| 387 |
-
},
|
| 388 |
-
{
|
| 389 |
-
"epoch": 0.36,
|
| 390 |
-
"learning_rate": 3.0044843049327355e-05,
|
| 391 |
-
"loss": 0.8003,
|
| 392 |
-
"step": 6400
|
| 393 |
-
},
|
| 394 |
-
{
|
| 395 |
-
"epoch": 0.36,
|
| 396 |
-
"learning_rate": 2.9733432984554062e-05,
|
| 397 |
-
"loss": 0.8693,
|
| 398 |
-
"step": 6500
|
| 399 |
-
},
|
| 400 |
-
{
|
| 401 |
-
"epoch": 0.37,
|
| 402 |
-
"learning_rate": 2.9422022919780766e-05,
|
| 403 |
-
"loss": 0.8033,
|
| 404 |
-
"step": 6600
|
| 405 |
-
},
|
| 406 |
-
{
|
| 407 |
-
"epoch": 0.38,
|
| 408 |
-
"learning_rate": 2.9110612855007473e-05,
|
| 409 |
-
"loss": 0.8024,
|
| 410 |
-
"step": 6700
|
| 411 |
-
},
|
| 412 |
-
{
|
| 413 |
-
"epoch": 0.38,
|
| 414 |
-
"learning_rate": 2.879920279023418e-05,
|
| 415 |
-
"loss": 0.7088,
|
| 416 |
-
"step": 6800
|
| 417 |
-
},
|
| 418 |
-
{
|
| 419 |
-
"epoch": 0.39,
|
| 420 |
-
"learning_rate": 2.8487792725460887e-05,
|
| 421 |
-
"loss": 0.7556,
|
| 422 |
-
"step": 6900
|
| 423 |
-
},
|
| 424 |
-
{
|
| 425 |
-
"epoch": 0.39,
|
| 426 |
-
"learning_rate": 2.8176382660687594e-05,
|
| 427 |
-
"loss": 0.8773,
|
| 428 |
-
"step": 7000
|
| 429 |
-
},
|
| 430 |
-
{
|
| 431 |
-
"epoch": 0.4,
|
| 432 |
-
"learning_rate": 2.78649725959143e-05,
|
| 433 |
-
"loss": 0.8164,
|
| 434 |
-
"step": 7100
|
| 435 |
-
},
|
| 436 |
-
{
|
| 437 |
-
"epoch": 0.4,
|
| 438 |
-
"learning_rate": 2.7553562531141008e-05,
|
| 439 |
-
"loss": 0.8011,
|
| 440 |
-
"step": 7200
|
| 441 |
-
},
|
| 442 |
-
{
|
| 443 |
-
"epoch": 0.41,
|
| 444 |
-
"learning_rate": 2.7242152466367715e-05,
|
| 445 |
-
"loss": 0.7086,
|
| 446 |
-
"step": 7300
|
| 447 |
-
},
|
| 448 |
-
{
|
| 449 |
-
"epoch": 0.41,
|
| 450 |
-
"learning_rate": 2.6930742401594422e-05,
|
| 451 |
-
"loss": 0.811,
|
| 452 |
-
"step": 7400
|
| 453 |
-
},
|
| 454 |
-
{
|
| 455 |
-
"epoch": 0.42,
|
| 456 |
-
"learning_rate": 2.661933233682113e-05,
|
| 457 |
-
"loss": 0.7833,
|
| 458 |
-
"step": 7500
|
| 459 |
-
},
|
| 460 |
-
{
|
| 461 |
-
"epoch": 0.43,
|
| 462 |
-
"learning_rate": 2.6307922272047836e-05,
|
| 463 |
-
"loss": 0.7085,
|
| 464 |
-
"step": 7600
|
| 465 |
-
},
|
| 466 |
-
{
|
| 467 |
-
"epoch": 0.43,
|
| 468 |
-
"learning_rate": 2.5996512207274543e-05,
|
| 469 |
-
"loss": 0.8229,
|
| 470 |
-
"step": 7700
|
| 471 |
-
},
|
| 472 |
-
{
|
| 473 |
-
"epoch": 0.44,
|
| 474 |
-
"learning_rate": 2.568510214250125e-05,
|
| 475 |
-
"loss": 0.8108,
|
| 476 |
-
"step": 7800
|
| 477 |
-
},
|
| 478 |
-
{
|
| 479 |
-
"epoch": 0.44,
|
| 480 |
-
"learning_rate": 2.537369207772795e-05,
|
| 481 |
-
"loss": 0.744,
|
| 482 |
-
"step": 7900
|
| 483 |
-
},
|
| 484 |
-
{
|
| 485 |
-
"epoch": 0.45,
|
| 486 |
-
"learning_rate": 2.5062282012954657e-05,
|
| 487 |
-
"loss": 0.749,
|
| 488 |
-
"step": 8000
|
| 489 |
-
},
|
| 490 |
-
{
|
| 491 |
-
"epoch": 0.45,
|
| 492 |
-
"learning_rate": 2.4750871948181367e-05,
|
| 493 |
-
"loss": 0.705,
|
| 494 |
-
"step": 8100
|
| 495 |
-
},
|
| 496 |
-
{
|
| 497 |
-
"epoch": 0.46,
|
| 498 |
-
"learning_rate": 2.4439461883408075e-05,
|
| 499 |
-
"loss": 0.7805,
|
| 500 |
-
"step": 8200
|
| 501 |
-
},
|
| 502 |
-
{
|
| 503 |
-
"epoch": 0.47,
|
| 504 |
-
"learning_rate": 2.4128051818634778e-05,
|
| 505 |
-
"loss": 0.6936,
|
| 506 |
-
"step": 8300
|
| 507 |
-
},
|
| 508 |
-
{
|
| 509 |
-
"epoch": 0.47,
|
| 510 |
-
"learning_rate": 2.3816641753861485e-05,
|
| 511 |
-
"loss": 0.8277,
|
| 512 |
-
"step": 8400
|
| 513 |
-
},
|
| 514 |
-
{
|
| 515 |
-
"epoch": 0.48,
|
| 516 |
-
"learning_rate": 2.3505231689088192e-05,
|
| 517 |
-
"loss": 0.7361,
|
| 518 |
-
"step": 8500
|
| 519 |
-
},
|
| 520 |
-
{
|
| 521 |
-
"epoch": 0.48,
|
| 522 |
-
"learning_rate": 2.31938216243149e-05,
|
| 523 |
-
"loss": 0.6989,
|
| 524 |
-
"step": 8600
|
| 525 |
-
},
|
| 526 |
-
{
|
| 527 |
-
"epoch": 0.49,
|
| 528 |
-
"learning_rate": 2.2882411559541606e-05,
|
| 529 |
-
"loss": 0.8185,
|
| 530 |
-
"step": 8700
|
| 531 |
-
},
|
| 532 |
-
{
|
| 533 |
-
"epoch": 0.49,
|
| 534 |
-
"learning_rate": 2.2571001494768313e-05,
|
| 535 |
-
"loss": 0.6851,
|
| 536 |
-
"step": 8800
|
| 537 |
-
},
|
| 538 |
-
{
|
| 539 |
-
"epoch": 0.5,
|
| 540 |
-
"learning_rate": 2.225959142999502e-05,
|
| 541 |
-
"loss": 0.7143,
|
| 542 |
-
"step": 8900
|
| 543 |
-
},
|
| 544 |
-
{
|
| 545 |
-
"epoch": 0.5,
|
| 546 |
-
"learning_rate": 2.1948181365221727e-05,
|
| 547 |
-
"loss": 0.8171,
|
| 548 |
-
"step": 9000
|
| 549 |
-
},
|
| 550 |
-
{
|
| 551 |
-
"epoch": 0.51,
|
| 552 |
-
"learning_rate": 2.163677130044843e-05,
|
| 553 |
-
"loss": 0.7185,
|
| 554 |
-
"step": 9100
|
| 555 |
-
},
|
| 556 |
-
{
|
| 557 |
-
"epoch": 0.52,
|
| 558 |
-
"learning_rate": 2.1325361235675138e-05,
|
| 559 |
-
"loss": 0.755,
|
| 560 |
-
"step": 9200
|
| 561 |
-
},
|
| 562 |
-
{
|
| 563 |
-
"epoch": 0.52,
|
| 564 |
-
"learning_rate": 2.1013951170901845e-05,
|
| 565 |
-
"loss": 0.7207,
|
| 566 |
-
"step": 9300
|
| 567 |
-
},
|
| 568 |
-
{
|
| 569 |
-
"epoch": 0.53,
|
| 570 |
-
"learning_rate": 2.0702541106128552e-05,
|
| 571 |
-
"loss": 0.7663,
|
| 572 |
-
"step": 9400
|
| 573 |
-
},
|
| 574 |
-
{
|
| 575 |
-
"epoch": 0.53,
|
| 576 |
-
"learning_rate": 2.039113104135526e-05,
|
| 577 |
-
"loss": 0.6746,
|
| 578 |
-
"step": 9500
|
| 579 |
-
},
|
| 580 |
-
{
|
| 581 |
-
"epoch": 0.54,
|
| 582 |
-
"learning_rate": 2.0079720976581966e-05,
|
| 583 |
-
"loss": 0.7107,
|
| 584 |
-
"step": 9600
|
| 585 |
-
},
|
| 586 |
-
{
|
| 587 |
-
"epoch": 0.54,
|
| 588 |
-
"learning_rate": 1.976831091180867e-05,
|
| 589 |
-
"loss": 0.6593,
|
| 590 |
-
"step": 9700
|
| 591 |
-
},
|
| 592 |
-
{
|
| 593 |
-
"epoch": 0.55,
|
| 594 |
-
"learning_rate": 1.9456900847035377e-05,
|
| 595 |
-
"loss": 0.7946,
|
| 596 |
-
"step": 9800
|
| 597 |
-
},
|
| 598 |
-
{
|
| 599 |
-
"epoch": 0.55,
|
| 600 |
-
"learning_rate": 1.9145490782262084e-05,
|
| 601 |
-
"loss": 0.7825,
|
| 602 |
-
"step": 9900
|
| 603 |
-
},
|
| 604 |
-
{
|
| 605 |
-
"epoch": 0.56,
|
| 606 |
-
"learning_rate": 1.883408071748879e-05,
|
| 607 |
-
"loss": 0.6866,
|
| 608 |
-
"step": 10000
|
| 609 |
-
},
|
| 610 |
-
{
|
| 611 |
-
"epoch": 0.57,
|
| 612 |
-
"learning_rate": 1.8522670652715494e-05,
|
| 613 |
-
"loss": 0.7075,
|
| 614 |
-
"step": 10100
|
| 615 |
-
},
|
| 616 |
-
{
|
| 617 |
-
"epoch": 0.57,
|
| 618 |
-
"learning_rate": 1.82112605879422e-05,
|
| 619 |
-
"loss": 0.7549,
|
| 620 |
-
"step": 10200
|
| 621 |
-
},
|
| 622 |
-
{
|
| 623 |
-
"epoch": 0.58,
|
| 624 |
-
"learning_rate": 1.7899850523168908e-05,
|
| 625 |
-
"loss": 0.7435,
|
| 626 |
-
"step": 10300
|
| 627 |
-
},
|
| 628 |
-
{
|
| 629 |
-
"epoch": 0.58,
|
| 630 |
-
"learning_rate": 1.7588440458395615e-05,
|
| 631 |
-
"loss": 0.7484,
|
| 632 |
-
"step": 10400
|
| 633 |
-
},
|
| 634 |
-
{
|
| 635 |
-
"epoch": 0.59,
|
| 636 |
-
"learning_rate": 1.7277030393622322e-05,
|
| 637 |
-
"loss": 0.7309,
|
| 638 |
-
"step": 10500
|
| 639 |
-
},
|
| 640 |
-
{
|
| 641 |
-
"epoch": 0.59,
|
| 642 |
-
"learning_rate": 1.696562032884903e-05,
|
| 643 |
-
"loss": 0.8038,
|
| 644 |
-
"step": 10600
|
| 645 |
-
},
|
| 646 |
-
{
|
| 647 |
-
"epoch": 0.6,
|
| 648 |
-
"learning_rate": 1.6654210264075736e-05,
|
| 649 |
-
"loss": 0.7695,
|
| 650 |
-
"step": 10700
|
| 651 |
-
},
|
| 652 |
-
{
|
| 653 |
-
"epoch": 0.61,
|
| 654 |
-
"learning_rate": 1.634280019930244e-05,
|
| 655 |
-
"loss": 0.7406,
|
| 656 |
-
"step": 10800
|
| 657 |
-
},
|
| 658 |
-
{
|
| 659 |
-
"epoch": 0.61,
|
| 660 |
-
"learning_rate": 1.6031390134529147e-05,
|
| 661 |
-
"loss": 0.6867,
|
| 662 |
-
"step": 10900
|
| 663 |
-
},
|
| 664 |
-
{
|
| 665 |
-
"epoch": 0.62,
|
| 666 |
-
"learning_rate": 1.5719980069755854e-05,
|
| 667 |
-
"loss": 0.6952,
|
| 668 |
-
"step": 11000
|
| 669 |
-
},
|
| 670 |
-
{
|
| 671 |
-
"epoch": 0.62,
|
| 672 |
-
"learning_rate": 1.540857000498256e-05,
|
| 673 |
-
"loss": 0.6863,
|
| 674 |
-
"step": 11100
|
| 675 |
-
},
|
| 676 |
-
{
|
| 677 |
-
"epoch": 0.63,
|
| 678 |
-
"learning_rate": 1.5097159940209268e-05,
|
| 679 |
-
"loss": 0.7765,
|
| 680 |
-
"step": 11200
|
| 681 |
-
},
|
| 682 |
-
{
|
| 683 |
-
"epoch": 0.63,
|
| 684 |
-
"learning_rate": 1.4785749875435975e-05,
|
| 685 |
-
"loss": 0.7468,
|
| 686 |
-
"step": 11300
|
| 687 |
-
},
|
| 688 |
-
{
|
| 689 |
-
"epoch": 0.64,
|
| 690 |
-
"learning_rate": 1.4474339810662682e-05,
|
| 691 |
-
"loss": 0.8029,
|
| 692 |
-
"step": 11400
|
| 693 |
-
},
|
| 694 |
-
{
|
| 695 |
-
"epoch": 0.64,
|
| 696 |
-
"learning_rate": 1.4162929745889389e-05,
|
| 697 |
-
"loss": 0.7224,
|
| 698 |
-
"step": 11500
|
| 699 |
-
},
|
| 700 |
-
{
|
| 701 |
-
"epoch": 0.65,
|
| 702 |
-
"learning_rate": 1.3851519681116093e-05,
|
| 703 |
-
"loss": 0.7289,
|
| 704 |
-
"step": 11600
|
| 705 |
-
},
|
| 706 |
-
{
|
| 707 |
-
"epoch": 0.66,
|
| 708 |
-
"learning_rate": 1.35401096163428e-05,
|
| 709 |
-
"loss": 0.758,
|
| 710 |
-
"step": 11700
|
| 711 |
-
},
|
| 712 |
-
{
|
| 713 |
-
"epoch": 0.66,
|
| 714 |
-
"learning_rate": 1.3228699551569507e-05,
|
| 715 |
-
"loss": 0.7118,
|
| 716 |
-
"step": 11800
|
| 717 |
-
},
|
| 718 |
-
{
|
| 719 |
-
"epoch": 0.67,
|
| 720 |
-
"learning_rate": 1.2917289486796214e-05,
|
| 721 |
-
"loss": 0.7478,
|
| 722 |
-
"step": 11900
|
| 723 |
-
},
|
| 724 |
-
{
|
| 725 |
-
"epoch": 0.67,
|
| 726 |
-
"learning_rate": 1.260587942202292e-05,
|
| 727 |
-
"loss": 0.7617,
|
| 728 |
-
"step": 12000
|
| 729 |
-
},
|
| 730 |
-
{
|
| 731 |
-
"epoch": 0.68,
|
| 732 |
-
"learning_rate": 1.2294469357249628e-05,
|
| 733 |
-
"loss": 0.7288,
|
| 734 |
-
"step": 12100
|
| 735 |
-
},
|
| 736 |
-
{
|
| 737 |
-
"epoch": 0.68,
|
| 738 |
-
"learning_rate": 1.1983059292476335e-05,
|
| 739 |
-
"loss": 0.7643,
|
| 740 |
-
"step": 12200
|
| 741 |
-
},
|
| 742 |
-
{
|
| 743 |
-
"epoch": 0.69,
|
| 744 |
-
"learning_rate": 1.167164922770304e-05,
|
| 745 |
-
"loss": 0.7059,
|
| 746 |
-
"step": 12300
|
| 747 |
-
},
|
| 748 |
-
{
|
| 749 |
-
"epoch": 0.7,
|
| 750 |
-
"learning_rate": 1.1360239162929747e-05,
|
| 751 |
-
"loss": 0.7185,
|
| 752 |
-
"step": 12400
|
| 753 |
-
},
|
| 754 |
-
{
|
| 755 |
-
"epoch": 0.7,
|
| 756 |
-
"learning_rate": 1.1048829098156454e-05,
|
| 757 |
-
"loss": 0.6894,
|
| 758 |
-
"step": 12500
|
| 759 |
-
},
|
| 760 |
-
{
|
| 761 |
-
"epoch": 0.71,
|
| 762 |
-
"learning_rate": 1.073741903338316e-05,
|
| 763 |
-
"loss": 0.7713,
|
| 764 |
-
"step": 12600
|
| 765 |
-
},
|
| 766 |
-
{
|
| 767 |
-
"epoch": 0.71,
|
| 768 |
-
"learning_rate": 1.0426008968609865e-05,
|
| 769 |
-
"loss": 0.6758,
|
| 770 |
-
"step": 12700
|
| 771 |
-
},
|
| 772 |
-
{
|
| 773 |
-
"epoch": 0.72,
|
| 774 |
-
"learning_rate": 1.0114598903836572e-05,
|
| 775 |
-
"loss": 0.7704,
|
| 776 |
-
"step": 12800
|
| 777 |
-
},
|
| 778 |
-
{
|
| 779 |
-
"epoch": 0.72,
|
| 780 |
-
"learning_rate": 9.803188839063279e-06,
|
| 781 |
-
"loss": 0.7496,
|
| 782 |
-
"step": 12900
|
| 783 |
-
},
|
| 784 |
-
{
|
| 785 |
-
"epoch": 0.73,
|
| 786 |
-
"learning_rate": 9.491778774289986e-06,
|
| 787 |
-
"loss": 0.7154,
|
| 788 |
-
"step": 13000
|
| 789 |
-
},
|
| 790 |
-
{
|
| 791 |
-
"epoch": 0.73,
|
| 792 |
-
"learning_rate": 9.180368709516691e-06,
|
| 793 |
-
"loss": 0.7938,
|
| 794 |
-
"step": 13100
|
| 795 |
-
},
|
| 796 |
-
{
|
| 797 |
-
"epoch": 0.74,
|
| 798 |
-
"learning_rate": 8.868958644743398e-06,
|
| 799 |
-
"loss": 0.6646,
|
| 800 |
-
"step": 13200
|
| 801 |
-
},
|
| 802 |
-
{
|
| 803 |
-
"epoch": 0.75,
|
| 804 |
-
"learning_rate": 8.557548579970105e-06,
|
| 805 |
-
"loss": 0.6602,
|
| 806 |
-
"step": 13300
|
| 807 |
-
},
|
| 808 |
-
{
|
| 809 |
-
"epoch": 0.75,
|
| 810 |
-
"learning_rate": 8.246138515196812e-06,
|
| 811 |
-
"loss": 0.7505,
|
| 812 |
-
"step": 13400
|
| 813 |
-
},
|
| 814 |
-
{
|
| 815 |
-
"epoch": 0.76,
|
| 816 |
-
"learning_rate": 7.934728450423517e-06,
|
| 817 |
-
"loss": 0.766,
|
| 818 |
-
"step": 13500
|
| 819 |
-
},
|
| 820 |
-
{
|
| 821 |
-
"epoch": 0.76,
|
| 822 |
-
"learning_rate": 7.6233183856502244e-06,
|
| 823 |
-
"loss": 0.7222,
|
| 824 |
-
"step": 13600
|
| 825 |
-
},
|
| 826 |
-
{
|
| 827 |
-
"epoch": 0.77,
|
| 828 |
-
"learning_rate": 7.3119083208769314e-06,
|
| 829 |
-
"loss": 0.7841,
|
| 830 |
-
"step": 13700
|
| 831 |
-
},
|
| 832 |
-
{
|
| 833 |
-
"epoch": 0.77,
|
| 834 |
-
"learning_rate": 7.0004982561036385e-06,
|
| 835 |
-
"loss": 0.7688,
|
| 836 |
-
"step": 13800
|
| 837 |
-
},
|
| 838 |
-
{
|
| 839 |
-
"epoch": 0.78,
|
| 840 |
-
"learning_rate": 6.689088191330344e-06,
|
| 841 |
-
"loss": 0.6692,
|
| 842 |
-
"step": 13900
|
| 843 |
-
},
|
| 844 |
-
{
|
| 845 |
-
"epoch": 0.78,
|
| 846 |
-
"learning_rate": 6.377678126557051e-06,
|
| 847 |
-
"loss": 0.7331,
|
| 848 |
-
"step": 14000
|
| 849 |
-
},
|
| 850 |
-
{
|
| 851 |
-
"epoch": 0.79,
|
| 852 |
-
"learning_rate": 6.066268061783757e-06,
|
| 853 |
-
"loss": 0.792,
|
| 854 |
-
"step": 14100
|
| 855 |
-
},
|
| 856 |
-
{
|
| 857 |
-
"epoch": 0.8,
|
| 858 |
-
"learning_rate": 5.754857997010463e-06,
|
| 859 |
-
"loss": 0.7135,
|
| 860 |
-
"step": 14200
|
| 861 |
-
},
|
| 862 |
-
{
|
| 863 |
-
"epoch": 0.8,
|
| 864 |
-
"learning_rate": 5.44344793223717e-06,
|
| 865 |
-
"loss": 0.7422,
|
| 866 |
-
"step": 14300
|
| 867 |
-
},
|
| 868 |
-
{
|
| 869 |
-
"epoch": 0.81,
|
| 870 |
-
"learning_rate": 5.132037867463876e-06,
|
| 871 |
-
"loss": 0.7497,
|
| 872 |
-
"step": 14400
|
| 873 |
-
},
|
| 874 |
-
{
|
| 875 |
-
"epoch": 0.81,
|
| 876 |
-
"learning_rate": 4.820627802690583e-06,
|
| 877 |
-
"loss": 0.7462,
|
| 878 |
-
"step": 14500
|
| 879 |
-
},
|
| 880 |
-
{
|
| 881 |
-
"epoch": 0.82,
|
| 882 |
-
"learning_rate": 4.5092177379172895e-06,
|
| 883 |
-
"loss": 0.7303,
|
| 884 |
-
"step": 14600
|
| 885 |
-
},
|
| 886 |
-
{
|
| 887 |
-
"epoch": 0.82,
|
| 888 |
-
"learning_rate": 4.1978076731439965e-06,
|
| 889 |
-
"loss": 0.7504,
|
| 890 |
-
"step": 14700
|
| 891 |
-
},
|
| 892 |
-
{
|
| 893 |
-
"epoch": 0.83,
|
| 894 |
-
"learning_rate": 3.886397608370703e-06,
|
| 895 |
-
"loss": 0.7655,
|
| 896 |
-
"step": 14800
|
| 897 |
-
},
|
| 898 |
-
{
|
| 899 |
-
"epoch": 0.84,
|
| 900 |
-
"learning_rate": 3.5749875435974093e-06,
|
| 901 |
-
"loss": 0.7503,
|
| 902 |
-
"step": 14900
|
| 903 |
-
},
|
| 904 |
-
{
|
| 905 |
-
"epoch": 0.84,
|
| 906 |
-
"learning_rate": 3.2635774788241163e-06,
|
| 907 |
-
"loss": 0.625,
|
| 908 |
-
"step": 15000
|
| 909 |
-
},
|
| 910 |
-
{
|
| 911 |
-
"epoch": 0.85,
|
| 912 |
-
"learning_rate": 2.952167414050822e-06,
|
| 913 |
-
"loss": 0.8014,
|
| 914 |
-
"step": 15100
|
| 915 |
-
},
|
| 916 |
-
{
|
| 917 |
-
"epoch": 0.85,
|
| 918 |
-
"learning_rate": 2.6407573492775286e-06,
|
| 919 |
-
"loss": 0.7089,
|
| 920 |
-
"step": 15200
|
| 921 |
-
},
|
| 922 |
-
{
|
| 923 |
-
"epoch": 0.86,
|
| 924 |
-
"learning_rate": 2.329347284504235e-06,
|
| 925 |
-
"loss": 0.74,
|
| 926 |
-
"step": 15300
|
| 927 |
-
},
|
| 928 |
-
{
|
| 929 |
-
"epoch": 0.86,
|
| 930 |
-
"learning_rate": 2.0179372197309418e-06,
|
| 931 |
-
"loss": 0.7056,
|
| 932 |
-
"step": 15400
|
| 933 |
-
},
|
| 934 |
-
{
|
| 935 |
-
"epoch": 0.87,
|
| 936 |
-
"learning_rate": 1.7065271549576482e-06,
|
| 937 |
-
"loss": 0.7139,
|
| 938 |
-
"step": 15500
|
| 939 |
-
},
|
| 940 |
-
{
|
| 941 |
-
"epoch": 0.87,
|
| 942 |
-
"learning_rate": 1.395117090184355e-06,
|
| 943 |
-
"loss": 0.7094,
|
| 944 |
-
"step": 15600
|
| 945 |
-
},
|
| 946 |
-
{
|
| 947 |
-
"epoch": 0.88,
|
| 948 |
-
"learning_rate": 1.0837070254110613e-06,
|
| 949 |
-
"loss": 0.7352,
|
| 950 |
-
"step": 15700
|
| 951 |
-
},
|
| 952 |
-
{
|
| 953 |
-
"epoch": 0.89,
|
| 954 |
-
"learning_rate": 7.722969606377678e-07,
|
| 955 |
-
"loss": 0.6215,
|
| 956 |
-
"step": 15800
|
| 957 |
-
},
|
| 958 |
-
{
|
| 959 |
-
"epoch": 0.89,
|
| 960 |
-
"learning_rate": 4.608868958644744e-07,
|
| 961 |
-
"loss": 0.72,
|
| 962 |
-
"step": 15900
|
| 963 |
-
},
|
| 964 |
-
{
|
| 965 |
-
"epoch": 0.9,
|
| 966 |
-
"learning_rate": 1.4947683109118088e-07,
|
| 967 |
-
"loss": 0.844,
|
| 968 |
-
"step": 16000
|
| 969 |
-
},
|
| 970 |
-
{
|
| 971 |
-
"epoch": 0.9,
|
| 972 |
-
"learning_rate": 0.0,
|
| 973 |
-
"loss": 0.7329,
|
| 974 |
-
"step": 16100
|
| 975 |
-
},
|
| 976 |
-
{
|
| 977 |
-
"epoch": 0.91,
|
| 978 |
-
"learning_rate": 0.0,
|
| 979 |
-
"loss": 0.7173,
|
| 980 |
-
"step": 16200
|
| 981 |
-
},
|
| 982 |
-
{
|
| 983 |
-
"epoch": 0.91,
|
| 984 |
-
"learning_rate": 0.0,
|
| 985 |
-
"loss": 0.7362,
|
| 986 |
-
"step": 16300
|
| 987 |
-
},
|
| 988 |
-
{
|
| 989 |
-
"epoch": 0.92,
|
| 990 |
-
"learning_rate": 0.0,
|
| 991 |
-
"loss": 0.7832,
|
| 992 |
-
"step": 16400
|
| 993 |
-
},
|
| 994 |
-
{
|
| 995 |
-
"epoch": 0.92,
|
| 996 |
-
"learning_rate": 0.0,
|
| 997 |
-
"loss": 0.6208,
|
| 998 |
-
"step": 16500
|
| 999 |
-
},
|
| 1000 |
-
{
|
| 1001 |
-
"epoch": 0.93,
|
| 1002 |
-
"learning_rate": 0.0,
|
| 1003 |
-
"loss": 0.7005,
|
| 1004 |
-
"step": 16600
|
| 1005 |
-
},
|
| 1006 |
-
{
|
| 1007 |
-
"epoch": 0.94,
|
| 1008 |
-
"learning_rate": 0.0,
|
| 1009 |
-
"loss": 0.6966,
|
| 1010 |
-
"step": 16700
|
| 1011 |
-
},
|
| 1012 |
-
{
|
| 1013 |
-
"epoch": 0.94,
|
| 1014 |
-
"learning_rate": 0.0,
|
| 1015 |
-
"loss": 0.7328,
|
| 1016 |
-
"step": 16800
|
| 1017 |
-
},
|
| 1018 |
-
{
|
| 1019 |
-
"epoch": 0.95,
|
| 1020 |
-
"learning_rate": 0.0,
|
| 1021 |
-
"loss": 0.7344,
|
| 1022 |
-
"step": 16900
|
| 1023 |
-
},
|
| 1024 |
-
{
|
| 1025 |
-
"epoch": 0.95,
|
| 1026 |
-
"learning_rate": 0.0,
|
| 1027 |
-
"loss": 0.6683,
|
| 1028 |
-
"step": 17000
|
| 1029 |
-
},
|
| 1030 |
-
{
|
| 1031 |
-
"epoch": 0.96,
|
| 1032 |
-
"learning_rate": 0.0,
|
| 1033 |
-
"loss": 0.7593,
|
| 1034 |
-
"step": 17100
|
| 1035 |
-
},
|
| 1036 |
-
{
|
| 1037 |
-
"epoch": 0.96,
|
| 1038 |
-
"learning_rate": 0.0,
|
| 1039 |
-
"loss": 0.6735,
|
| 1040 |
-
"step": 17200
|
| 1041 |
-
},
|
| 1042 |
-
{
|
| 1043 |
-
"epoch": 0.97,
|
| 1044 |
-
"learning_rate": 0.0,
|
| 1045 |
-
"loss": 0.6199,
|
| 1046 |
-
"step": 17300
|
| 1047 |
-
},
|
| 1048 |
-
{
|
| 1049 |
-
"epoch": 0.98,
|
| 1050 |
-
"learning_rate": 0.0,
|
| 1051 |
-
"loss": 0.7717,
|
| 1052 |
-
"step": 17400
|
| 1053 |
-
},
|
| 1054 |
-
{
|
| 1055 |
-
"epoch": 0.98,
|
| 1056 |
-
"learning_rate": 0.0,
|
| 1057 |
-
"loss": 0.7252,
|
| 1058 |
-
"step": 17500
|
| 1059 |
-
},
|
| 1060 |
-
{
|
| 1061 |
-
"epoch": 0.99,
|
| 1062 |
-
"learning_rate": 0.0,
|
| 1063 |
-
"loss": 0.8161,
|
| 1064 |
-
"step": 17600
|
| 1065 |
-
},
|
| 1066 |
-
{
|
| 1067 |
-
"epoch": 0.99,
|
| 1068 |
-
"learning_rate": 0.0,
|
| 1069 |
-
"loss": 0.7526,
|
| 1070 |
-
"step": 17700
|
| 1071 |
-
},
|
| 1072 |
-
{
|
| 1073 |
-
"epoch": 1.0,
|
| 1074 |
-
"learning_rate": 0.0,
|
| 1075 |
-
"loss": 0.7201,
|
| 1076 |
-
"step": 17800
|
| 1077 |
-
},
|
| 1078 |
-
{
|
| 1079 |
-
"epoch": 1.0,
|
| 1080 |
-
"learning_rate": 0.0,
|
| 1081 |
-
"loss": 0.7642,
|
| 1082 |
-
"step": 17900
|
| 1083 |
-
},
|
| 1084 |
-
{
|
| 1085 |
-
"epoch": 1.01,
|
| 1086 |
-
"learning_rate": 0.0,
|
| 1087 |
-
"loss": 0.6952,
|
| 1088 |
-
"step": 18000
|
| 1089 |
-
},
|
| 1090 |
-
{
|
| 1091 |
-
"epoch": 1.01,
|
| 1092 |
-
"learning_rate": 0.0,
|
| 1093 |
-
"loss": 0.6005,
|
| 1094 |
-
"step": 18100
|
| 1095 |
-
},
|
| 1096 |
-
{
|
| 1097 |
-
"epoch": 1.02,
|
| 1098 |
-
"learning_rate": 0.0,
|
| 1099 |
-
"loss": 0.7199,
|
| 1100 |
-
"step": 18200
|
| 1101 |
-
},
|
| 1102 |
-
{
|
| 1103 |
-
"epoch": 1.03,
|
| 1104 |
-
"learning_rate": 0.0,
|
| 1105 |
-
"loss": 0.7147,
|
| 1106 |
-
"step": 18300
|
| 1107 |
-
},
|
| 1108 |
-
{
|
| 1109 |
-
"epoch": 1.03,
|
| 1110 |
-
"learning_rate": 0.0,
|
| 1111 |
-
"loss": 0.6447,
|
| 1112 |
-
"step": 18400
|
| 1113 |
-
},
|
| 1114 |
-
{
|
| 1115 |
-
"epoch": 1.04,
|
| 1116 |
-
"learning_rate": 0.0,
|
| 1117 |
-
"loss": 0.679,
|
| 1118 |
-
"step": 18500
|
| 1119 |
-
},
|
| 1120 |
-
{
|
| 1121 |
-
"epoch": 1.04,
|
| 1122 |
-
"learning_rate": 0.0,
|
| 1123 |
-
"loss": 0.6931,
|
| 1124 |
-
"step": 18600
|
| 1125 |
-
},
|
| 1126 |
-
{
|
| 1127 |
-
"epoch": 1.05,
|
| 1128 |
-
"learning_rate": 0.0,
|
| 1129 |
-
"loss": 0.6616,
|
| 1130 |
-
"step": 18700
|
| 1131 |
-
},
|
| 1132 |
-
{
|
| 1133 |
-
"epoch": 1.05,
|
| 1134 |
-
"learning_rate": 0.0,
|
| 1135 |
-
"loss": 0.7168,
|
| 1136 |
-
"step": 18800
|
| 1137 |
-
},
|
| 1138 |
-
{
|
| 1139 |
-
"epoch": 1.06,
|
| 1140 |
-
"learning_rate": 0.0,
|
| 1141 |
-
"loss": 0.7334,
|
| 1142 |
-
"step": 18900
|
| 1143 |
-
},
|
| 1144 |
-
{
|
| 1145 |
-
"epoch": 1.07,
|
| 1146 |
-
"learning_rate": 0.0,
|
| 1147 |
-
"loss": 0.727,
|
| 1148 |
-
"step": 19000
|
| 1149 |
-
},
|
| 1150 |
-
{
|
| 1151 |
-
"epoch": 1.07,
|
| 1152 |
-
"learning_rate": 0.0,
|
| 1153 |
-
"loss": 0.6829,
|
| 1154 |
-
"step": 19100
|
| 1155 |
-
},
|
| 1156 |
-
{
|
| 1157 |
-
"epoch": 1.08,
|
| 1158 |
-
"learning_rate": 0.0,
|
| 1159 |
-
"loss": 0.6817,
|
| 1160 |
-
"step": 19200
|
| 1161 |
-
},
|
| 1162 |
-
{
|
| 1163 |
-
"epoch": 1.08,
|
| 1164 |
-
"learning_rate": 0.0,
|
| 1165 |
-
"loss": 0.6693,
|
| 1166 |
-
"step": 19300
|
| 1167 |
-
},
|
| 1168 |
-
{
|
| 1169 |
-
"epoch": 1.09,
|
| 1170 |
-
"learning_rate": 0.0,
|
| 1171 |
-
"loss": 0.6882,
|
| 1172 |
-
"step": 19400
|
| 1173 |
-
},
|
| 1174 |
-
{
|
| 1175 |
-
"epoch": 1.09,
|
| 1176 |
-
"learning_rate": 0.0,
|
| 1177 |
-
"loss": 0.6243,
|
| 1178 |
-
"step": 19500
|
| 1179 |
-
},
|
| 1180 |
-
{
|
| 1181 |
-
"epoch": 1.1,
|
| 1182 |
-
"learning_rate": 0.0,
|
| 1183 |
-
"loss": 0.7041,
|
| 1184 |
-
"step": 19600
|
| 1185 |
-
},
|
| 1186 |
-
{
|
| 1187 |
-
"epoch": 1.1,
|
| 1188 |
-
"learning_rate": 0.0,
|
| 1189 |
-
"loss": 0.7011,
|
| 1190 |
-
"step": 19700
|
| 1191 |
-
},
|
| 1192 |
-
{
|
| 1193 |
-
"epoch": 1.11,
|
| 1194 |
-
"learning_rate": 0.0,
|
| 1195 |
-
"loss": 0.6356,
|
| 1196 |
-
"step": 19800
|
| 1197 |
-
},
|
| 1198 |
-
{
|
| 1199 |
-
"epoch": 1.12,
|
| 1200 |
-
"learning_rate": 0.0,
|
| 1201 |
-
"loss": 0.7251,
|
| 1202 |
-
"step": 19900
|
| 1203 |
-
},
|
| 1204 |
-
{
|
| 1205 |
-
"epoch": 1.12,
|
| 1206 |
-
"learning_rate": 0.0,
|
| 1207 |
-
"loss": 0.665,
|
| 1208 |
-
"step": 20000
|
| 1209 |
-
},
|
| 1210 |
-
{
|
| 1211 |
-
"epoch": 1.13,
|
| 1212 |
-
"learning_rate": 0.0,
|
| 1213 |
-
"loss": 0.7389,
|
| 1214 |
-
"step": 20100
|
| 1215 |
-
},
|
| 1216 |
-
{
|
| 1217 |
-
"epoch": 1.13,
|
| 1218 |
-
"learning_rate": 0.0,
|
| 1219 |
-
"loss": 0.7405,
|
| 1220 |
-
"step": 20200
|
| 1221 |
-
},
|
| 1222 |
-
{
|
| 1223 |
-
"epoch": 1.14,
|
| 1224 |
-
"learning_rate": 0.0,
|
| 1225 |
-
"loss": 0.701,
|
| 1226 |
-
"step": 20300
|
| 1227 |
-
},
|
| 1228 |
-
{
|
| 1229 |
-
"epoch": 1.14,
|
| 1230 |
-
"learning_rate": 0.0,
|
| 1231 |
-
"loss": 0.6543,
|
| 1232 |
-
"step": 20400
|
| 1233 |
-
},
|
| 1234 |
-
{
|
| 1235 |
-
"epoch": 1.15,
|
| 1236 |
-
"learning_rate": 0.0,
|
| 1237 |
-
"loss": 0.65,
|
| 1238 |
-
"step": 20500
|
| 1239 |
-
},
|
| 1240 |
-
{
|
| 1241 |
-
"epoch": 1.15,
|
| 1242 |
-
"learning_rate": 0.0,
|
| 1243 |
-
"loss": 0.7009,
|
| 1244 |
-
"step": 20600
|
| 1245 |
-
},
|
| 1246 |
-
{
|
| 1247 |
-
"epoch": 1.16,
|
| 1248 |
-
"learning_rate": 0.0,
|
| 1249 |
-
"loss": 0.6925,
|
| 1250 |
-
"step": 20700
|
| 1251 |
-
},
|
| 1252 |
-
{
|
| 1253 |
-
"epoch": 1.17,
|
| 1254 |
-
"learning_rate": 0.0,
|
| 1255 |
-
"loss": 0.6334,
|
| 1256 |
-
"step": 20800
|
| 1257 |
-
},
|
| 1258 |
-
{
|
| 1259 |
-
"epoch": 1.17,
|
| 1260 |
-
"learning_rate": 0.0,
|
| 1261 |
-
"loss": 0.6791,
|
| 1262 |
-
"step": 20900
|
| 1263 |
-
},
|
| 1264 |
-
{
|
| 1265 |
-
"epoch": 1.18,
|
| 1266 |
-
"learning_rate": 0.0,
|
| 1267 |
-
"loss": 0.6704,
|
| 1268 |
-
"step": 21000
|
| 1269 |
-
},
|
| 1270 |
-
{
|
| 1271 |
-
"epoch": 1.18,
|
| 1272 |
-
"learning_rate": 0.0,
|
| 1273 |
-
"loss": 0.7143,
|
| 1274 |
-
"step": 21100
|
| 1275 |
-
},
|
| 1276 |
-
{
|
| 1277 |
-
"epoch": 1.19,
|
| 1278 |
-
"learning_rate": 0.0,
|
| 1279 |
-
"loss": 0.6845,
|
| 1280 |
-
"step": 21200
|
| 1281 |
-
},
|
| 1282 |
-
{
|
| 1283 |
-
"epoch": 1.19,
|
| 1284 |
-
"learning_rate": 0.0,
|
| 1285 |
-
"loss": 0.6935,
|
| 1286 |
-
"step": 21300
|
| 1287 |
-
},
|
| 1288 |
-
{
|
| 1289 |
-
"epoch": 1.2,
|
| 1290 |
-
"learning_rate": 0.0,
|
| 1291 |
-
"loss": 0.6093,
|
| 1292 |
-
"step": 21400
|
| 1293 |
-
},
|
| 1294 |
-
{
|
| 1295 |
-
"epoch": 1.21,
|
| 1296 |
-
"learning_rate": 0.0,
|
| 1297 |
-
"loss": 0.6883,
|
| 1298 |
-
"step": 21500
|
| 1299 |
-
},
|
| 1300 |
-
{
|
| 1301 |
-
"epoch": 1.21,
|
| 1302 |
-
"learning_rate": 0.0,
|
| 1303 |
-
"loss": 0.6498,
|
| 1304 |
-
"step": 21600
|
| 1305 |
-
},
|
| 1306 |
-
{
|
| 1307 |
-
"epoch": 1.22,
|
| 1308 |
-
"learning_rate": 0.0,
|
| 1309 |
-
"loss": 0.6822,
|
| 1310 |
-
"step": 21700
|
| 1311 |
-
},
|
| 1312 |
-
{
|
| 1313 |
-
"epoch": 1.22,
|
| 1314 |
-
"learning_rate": 0.0,
|
| 1315 |
-
"loss": 0.6887,
|
| 1316 |
-
"step": 21800
|
| 1317 |
-
},
|
| 1318 |
-
{
|
| 1319 |
-
"epoch": 1.23,
|
| 1320 |
-
"learning_rate": 0.0,
|
| 1321 |
-
"loss": 0.6755,
|
| 1322 |
-
"step": 21900
|
| 1323 |
-
},
|
| 1324 |
-
{
|
| 1325 |
-
"epoch": 1.23,
|
| 1326 |
-
"learning_rate": 0.0,
|
| 1327 |
-
"loss": 0.6265,
|
| 1328 |
-
"step": 22000
|
| 1329 |
-
},
|
| 1330 |
-
{
|
| 1331 |
-
"epoch": 1.24,
|
| 1332 |
-
"learning_rate": 0.0,
|
| 1333 |
-
"loss": 0.7267,
|
| 1334 |
-
"step": 22100
|
| 1335 |
-
},
|
| 1336 |
-
{
|
| 1337 |
-
"epoch": 1.24,
|
| 1338 |
-
"learning_rate": 0.0,
|
| 1339 |
-
"loss": 0.6801,
|
| 1340 |
-
"step": 22200
|
| 1341 |
-
},
|
| 1342 |
-
{
|
| 1343 |
-
"epoch": 1.25,
|
| 1344 |
-
"learning_rate": 0.0,
|
| 1345 |
-
"loss": 0.6961,
|
| 1346 |
-
"step": 22300
|
| 1347 |
-
},
|
| 1348 |
-
{
|
| 1349 |
-
"epoch": 1.26,
|
| 1350 |
-
"learning_rate": 0.0,
|
| 1351 |
-
"loss": 0.7783,
|
| 1352 |
-
"step": 22400
|
| 1353 |
-
},
|
| 1354 |
-
{
|
| 1355 |
-
"epoch": 1.26,
|
| 1356 |
-
"learning_rate": 0.0,
|
| 1357 |
-
"loss": 0.7396,
|
| 1358 |
-
"step": 22500
|
| 1359 |
-
},
|
| 1360 |
-
{
|
| 1361 |
-
"epoch": 1.27,
|
| 1362 |
-
"learning_rate": 0.0,
|
| 1363 |
-
"loss": 0.7497,
|
| 1364 |
-
"step": 22600
|
| 1365 |
-
},
|
| 1366 |
-
{
|
| 1367 |
-
"epoch": 1.27,
|
| 1368 |
-
"learning_rate": 0.0,
|
| 1369 |
-
"loss": 0.6458,
|
| 1370 |
-
"step": 22700
|
| 1371 |
-
},
|
| 1372 |
-
{
|
| 1373 |
-
"epoch": 1.28,
|
| 1374 |
-
"learning_rate": 0.0,
|
| 1375 |
-
"loss": 0.6969,
|
| 1376 |
-
"step": 22800
|
| 1377 |
-
},
|
| 1378 |
-
{
|
| 1379 |
-
"epoch": 1.28,
|
| 1380 |
-
"learning_rate": 0.0,
|
| 1381 |
-
"loss": 0.7066,
|
| 1382 |
-
"step": 22900
|
| 1383 |
-
},
|
| 1384 |
-
{
|
| 1385 |
-
"epoch": 1.29,
|
| 1386 |
-
"learning_rate": 0.0,
|
| 1387 |
-
"loss": 0.6013,
|
| 1388 |
-
"step": 23000
|
| 1389 |
-
},
|
| 1390 |
-
{
|
| 1391 |
-
"epoch": 1.29,
|
| 1392 |
-
"learning_rate": 0.0,
|
| 1393 |
-
"loss": 0.6545,
|
| 1394 |
-
"step": 23100
|
| 1395 |
-
},
|
| 1396 |
-
{
|
| 1397 |
-
"epoch": 1.3,
|
| 1398 |
-
"learning_rate": 0.0,
|
| 1399 |
-
"loss": 0.6699,
|
| 1400 |
-
"step": 23200
|
| 1401 |
-
},
|
| 1402 |
-
{
|
| 1403 |
-
"epoch": 1.31,
|
| 1404 |
-
"learning_rate": 0.0,
|
| 1405 |
-
"loss": 0.66,
|
| 1406 |
-
"step": 23300
|
| 1407 |
-
},
|
| 1408 |
-
{
|
| 1409 |
-
"epoch": 1.31,
|
| 1410 |
-
"learning_rate": 0.0,
|
| 1411 |
-
"loss": 0.6678,
|
| 1412 |
-
"step": 23400
|
| 1413 |
-
},
|
| 1414 |
-
{
|
| 1415 |
-
"epoch": 1.32,
|
| 1416 |
-
"learning_rate": 0.0,
|
| 1417 |
-
"loss": 0.7262,
|
| 1418 |
-
"step": 23500
|
| 1419 |
-
},
|
| 1420 |
-
{
|
| 1421 |
-
"epoch": 1.32,
|
| 1422 |
-
"learning_rate": 0.0,
|
| 1423 |
-
"loss": 0.6492,
|
| 1424 |
-
"step": 23600
|
| 1425 |
-
},
|
| 1426 |
-
{
|
| 1427 |
-
"epoch": 1.33,
|
| 1428 |
-
"learning_rate": 0.0,
|
| 1429 |
-
"loss": 0.6833,
|
| 1430 |
-
"step": 23700
|
| 1431 |
-
},
|
| 1432 |
-
{
|
| 1433 |
-
"epoch": 1.33,
|
| 1434 |
-
"learning_rate": 0.0,
|
| 1435 |
-
"loss": 0.6783,
|
| 1436 |
-
"step": 23800
|
| 1437 |
-
},
|
| 1438 |
-
{
|
| 1439 |
-
"epoch": 1.34,
|
| 1440 |
-
"learning_rate": 0.0,
|
| 1441 |
-
"loss": 0.6809,
|
| 1442 |
-
"step": 23900
|
| 1443 |
-
},
|
| 1444 |
-
{
|
| 1445 |
-
"epoch": 1.35,
|
| 1446 |
-
"learning_rate": 0.0,
|
| 1447 |
-
"loss": 0.6651,
|
| 1448 |
-
"step": 24000
|
| 1449 |
-
},
|
| 1450 |
-
{
|
| 1451 |
-
"epoch": 1.35,
|
| 1452 |
-
"learning_rate": 0.0,
|
| 1453 |
-
"loss": 0.6928,
|
| 1454 |
-
"step": 24100
|
| 1455 |
-
},
|
| 1456 |
-
{
|
| 1457 |
-
"epoch": 1.36,
|
| 1458 |
-
"learning_rate": 0.0,
|
| 1459 |
-
"loss": 0.6163,
|
| 1460 |
-
"step": 24200
|
| 1461 |
-
},
|
| 1462 |
-
{
|
| 1463 |
-
"epoch": 1.36,
|
| 1464 |
-
"learning_rate": 0.0,
|
| 1465 |
-
"loss": 0.6045,
|
| 1466 |
-
"step": 24300
|
| 1467 |
-
},
|
| 1468 |
-
{
|
| 1469 |
-
"epoch": 1.37,
|
| 1470 |
-
"learning_rate": 0.0,
|
| 1471 |
-
"loss": 0.723,
|
| 1472 |
-
"step": 24400
|
| 1473 |
-
},
|
| 1474 |
-
{
|
| 1475 |
-
"epoch": 1.37,
|
| 1476 |
-
"learning_rate": 0.0,
|
| 1477 |
-
"loss": 0.644,
|
| 1478 |
-
"step": 24500
|
| 1479 |
-
},
|
| 1480 |
-
{
|
| 1481 |
-
"epoch": 1.38,
|
| 1482 |
-
"learning_rate": 0.0,
|
| 1483 |
-
"loss": 0.7337,
|
| 1484 |
-
"step": 24600
|
| 1485 |
-
},
|
| 1486 |
-
{
|
| 1487 |
-
"epoch": 1.38,
|
| 1488 |
-
"learning_rate": 0.0,
|
| 1489 |
-
"loss": 0.7065,
|
| 1490 |
-
"step": 24700
|
| 1491 |
-
},
|
| 1492 |
-
{
|
| 1493 |
-
"epoch": 1.39,
|
| 1494 |
-
"learning_rate": 0.0,
|
| 1495 |
-
"loss": 0.6687,
|
| 1496 |
-
"step": 24800
|
| 1497 |
-
},
|
| 1498 |
-
{
|
| 1499 |
-
"epoch": 1.4,
|
| 1500 |
-
"learning_rate": 0.0,
|
| 1501 |
-
"loss": 0.7265,
|
| 1502 |
-
"step": 24900
|
| 1503 |
-
},
|
| 1504 |
-
{
|
| 1505 |
-
"epoch": 1.4,
|
| 1506 |
-
"learning_rate": 0.0,
|
| 1507 |
-
"loss": 0.7165,
|
| 1508 |
-
"step": 25000
|
| 1509 |
-
},
|
| 1510 |
-
{
|
| 1511 |
-
"epoch": 1.41,
|
| 1512 |
-
"learning_rate": 0.0,
|
| 1513 |
-
"loss": 0.7042,
|
| 1514 |
-
"step": 25100
|
| 1515 |
-
},
|
| 1516 |
-
{
|
| 1517 |
-
"epoch": 1.41,
|
| 1518 |
-
"learning_rate": 0.0,
|
| 1519 |
-
"loss": 0.6953,
|
| 1520 |
-
"step": 25200
|
| 1521 |
-
},
|
| 1522 |
-
{
|
| 1523 |
-
"epoch": 1.42,
|
| 1524 |
-
"learning_rate": 0.0,
|
| 1525 |
-
"loss": 0.7518,
|
| 1526 |
-
"step": 25300
|
| 1527 |
-
},
|
| 1528 |
-
{
|
| 1529 |
-
"epoch": 1.42,
|
| 1530 |
-
"learning_rate": 0.0,
|
| 1531 |
-
"loss": 0.671,
|
| 1532 |
-
"step": 25400
|
| 1533 |
-
},
|
| 1534 |
-
{
|
| 1535 |
-
"epoch": 1.43,
|
| 1536 |
-
"learning_rate": 0.0,
|
| 1537 |
-
"loss": 0.7909,
|
| 1538 |
-
"step": 25500
|
| 1539 |
-
},
|
| 1540 |
-
{
|
| 1541 |
-
"epoch": 1.43,
|
| 1542 |
-
"learning_rate": 0.0,
|
| 1543 |
-
"loss": 0.6001,
|
| 1544 |
-
"step": 25600
|
| 1545 |
-
},
|
| 1546 |
-
{
|
| 1547 |
-
"epoch": 1.44,
|
| 1548 |
-
"learning_rate": 0.0,
|
| 1549 |
-
"loss": 0.663,
|
| 1550 |
-
"step": 25700
|
| 1551 |
-
},
|
| 1552 |
-
{
|
| 1553 |
-
"epoch": 1.45,
|
| 1554 |
-
"learning_rate": 0.0,
|
| 1555 |
-
"loss": 0.6896,
|
| 1556 |
-
"step": 25800
|
| 1557 |
-
},
|
| 1558 |
-
{
|
| 1559 |
-
"epoch": 1.45,
|
| 1560 |
-
"learning_rate": 0.0,
|
| 1561 |
-
"loss": 0.6647,
|
| 1562 |
-
"step": 25900
|
| 1563 |
-
},
|
| 1564 |
-
{
|
| 1565 |
-
"epoch": 1.46,
|
| 1566 |
-
"learning_rate": 0.0,
|
| 1567 |
-
"loss": 0.6437,
|
| 1568 |
-
"step": 26000
|
| 1569 |
-
},
|
| 1570 |
-
{
|
| 1571 |
-
"epoch": 1.46,
|
| 1572 |
-
"learning_rate": 0.0,
|
| 1573 |
-
"loss": 0.6731,
|
| 1574 |
-
"step": 26100
|
| 1575 |
-
},
|
| 1576 |
-
{
|
| 1577 |
-
"epoch": 1.47,
|
| 1578 |
-
"learning_rate": 0.0,
|
| 1579 |
-
"loss": 0.6877,
|
| 1580 |
-
"step": 26200
|
| 1581 |
-
},
|
| 1582 |
-
{
|
| 1583 |
-
"epoch": 1.47,
|
| 1584 |
-
"learning_rate": 0.0,
|
| 1585 |
-
"loss": 0.6896,
|
| 1586 |
-
"step": 26300
|
| 1587 |
-
},
|
| 1588 |
-
{
|
| 1589 |
-
"epoch": 1.48,
|
| 1590 |
-
"learning_rate": 0.0,
|
| 1591 |
-
"loss": 0.6735,
|
| 1592 |
-
"step": 26400
|
| 1593 |
-
},
|
| 1594 |
-
{
|
| 1595 |
-
"epoch": 1.49,
|
| 1596 |
-
"learning_rate": 0.0,
|
| 1597 |
-
"loss": 0.7419,
|
| 1598 |
-
"step": 26500
|
| 1599 |
-
},
|
| 1600 |
-
{
|
| 1601 |
-
"epoch": 1.49,
|
| 1602 |
-
"learning_rate": 0.0,
|
| 1603 |
-
"loss": 0.6452,
|
| 1604 |
-
"step": 26600
|
| 1605 |
-
},
|
| 1606 |
-
{
|
| 1607 |
-
"epoch": 1.5,
|
| 1608 |
-
"learning_rate": 0.0,
|
| 1609 |
-
"loss": 0.6525,
|
| 1610 |
-
"step": 26700
|
| 1611 |
-
},
|
| 1612 |
-
{
|
| 1613 |
-
"epoch": 1.5,
|
| 1614 |
-
"learning_rate": 0.0,
|
| 1615 |
-
"loss": 0.6998,
|
| 1616 |
-
"step": 26800
|
| 1617 |
-
},
|
| 1618 |
-
{
|
| 1619 |
-
"epoch": 1.51,
|
| 1620 |
-
"learning_rate": 0.0,
|
| 1621 |
-
"loss": 0.6136,
|
| 1622 |
-
"step": 26900
|
| 1623 |
-
},
|
| 1624 |
-
{
|
| 1625 |
-
"epoch": 1.51,
|
| 1626 |
-
"learning_rate": 0.0,
|
| 1627 |
-
"loss": 0.688,
|
| 1628 |
-
"step": 27000
|
| 1629 |
-
},
|
| 1630 |
-
{
|
| 1631 |
-
"epoch": 1.52,
|
| 1632 |
-
"learning_rate": 0.0,
|
| 1633 |
-
"loss": 0.6384,
|
| 1634 |
-
"step": 27100
|
| 1635 |
-
},
|
| 1636 |
-
{
|
| 1637 |
-
"epoch": 1.52,
|
| 1638 |
-
"learning_rate": 0.0,
|
| 1639 |
-
"loss": 0.6077,
|
| 1640 |
-
"step": 27200
|
| 1641 |
-
},
|
| 1642 |
-
{
|
| 1643 |
-
"epoch": 1.53,
|
| 1644 |
-
"learning_rate": 0.0,
|
| 1645 |
-
"loss": 0.6896,
|
| 1646 |
-
"step": 27300
|
| 1647 |
-
},
|
| 1648 |
-
{
|
| 1649 |
-
"epoch": 1.54,
|
| 1650 |
-
"learning_rate": 0.0,
|
| 1651 |
-
"loss": 0.6546,
|
| 1652 |
-
"step": 27400
|
| 1653 |
-
},
|
| 1654 |
-
{
|
| 1655 |
-
"epoch": 1.54,
|
| 1656 |
-
"learning_rate": 0.0,
|
| 1657 |
-
"loss": 0.6993,
|
| 1658 |
-
"step": 27500
|
| 1659 |
-
},
|
| 1660 |
-
{
|
| 1661 |
-
"epoch": 1.55,
|
| 1662 |
-
"learning_rate": 0.0,
|
| 1663 |
-
"loss": 0.6705,
|
| 1664 |
-
"step": 27600
|
| 1665 |
-
},
|
| 1666 |
-
{
|
| 1667 |
-
"epoch": 1.55,
|
| 1668 |
-
"learning_rate": 0.0,
|
| 1669 |
-
"loss": 0.6845,
|
| 1670 |
-
"step": 27700
|
| 1671 |
-
},
|
| 1672 |
-
{
|
| 1673 |
-
"epoch": 1.56,
|
| 1674 |
-
"learning_rate": 0.0,
|
| 1675 |
-
"loss": 0.6766,
|
| 1676 |
-
"step": 27800
|
| 1677 |
-
},
|
| 1678 |
-
{
|
| 1679 |
-
"epoch": 1.56,
|
| 1680 |
-
"learning_rate": 0.0,
|
| 1681 |
-
"loss": 0.6294,
|
| 1682 |
-
"step": 27900
|
| 1683 |
-
},
|
| 1684 |
-
{
|
| 1685 |
-
"epoch": 1.57,
|
| 1686 |
-
"learning_rate": 0.0,
|
| 1687 |
-
"loss": 0.6676,
|
| 1688 |
-
"step": 28000
|
| 1689 |
-
},
|
| 1690 |
-
{
|
| 1691 |
-
"epoch": 1.58,
|
| 1692 |
-
"learning_rate": 0.0,
|
| 1693 |
-
"loss": 0.652,
|
| 1694 |
-
"step": 28100
|
| 1695 |
-
},
|
| 1696 |
-
{
|
| 1697 |
-
"epoch": 1.58,
|
| 1698 |
-
"learning_rate": 0.0,
|
| 1699 |
-
"loss": 0.7111,
|
| 1700 |
-
"step": 28200
|
| 1701 |
-
},
|
| 1702 |
-
{
|
| 1703 |
-
"epoch": 1.59,
|
| 1704 |
-
"learning_rate": 0.0,
|
| 1705 |
-
"loss": 0.6388,
|
| 1706 |
-
"step": 28300
|
| 1707 |
-
},
|
| 1708 |
-
{
|
| 1709 |
-
"epoch": 1.59,
|
| 1710 |
-
"learning_rate": 0.0,
|
| 1711 |
-
"loss": 0.7184,
|
| 1712 |
-
"step": 28400
|
| 1713 |
-
},
|
| 1714 |
-
{
|
| 1715 |
-
"epoch": 1.6,
|
| 1716 |
-
"learning_rate": 0.0,
|
| 1717 |
-
"loss": 0.6623,
|
| 1718 |
-
"step": 28500
|
| 1719 |
-
},
|
| 1720 |
-
{
|
| 1721 |
-
"epoch": 1.6,
|
| 1722 |
-
"learning_rate": 0.0,
|
| 1723 |
-
"loss": 0.6997,
|
| 1724 |
-
"step": 28600
|
| 1725 |
-
},
|
| 1726 |
-
{
|
| 1727 |
-
"epoch": 1.61,
|
| 1728 |
-
"learning_rate": 0.0,
|
| 1729 |
-
"loss": 0.6616,
|
| 1730 |
-
"step": 28700
|
| 1731 |
-
},
|
| 1732 |
-
{
|
| 1733 |
-
"epoch": 1.61,
|
| 1734 |
-
"learning_rate": 0.0,
|
| 1735 |
-
"loss": 0.6768,
|
| 1736 |
-
"step": 28800
|
| 1737 |
-
},
|
| 1738 |
-
{
|
| 1739 |
-
"epoch": 1.62,
|
| 1740 |
-
"learning_rate": 0.0,
|
| 1741 |
-
"loss": 0.7443,
|
| 1742 |
-
"step": 28900
|
| 1743 |
-
},
|
| 1744 |
-
{
|
| 1745 |
-
"epoch": 1.63,
|
| 1746 |
-
"learning_rate": 0.0,
|
| 1747 |
-
"loss": 0.6635,
|
| 1748 |
-
"step": 29000
|
| 1749 |
-
},
|
| 1750 |
-
{
|
| 1751 |
-
"epoch": 1.63,
|
| 1752 |
-
"learning_rate": 0.0,
|
| 1753 |
-
"loss": 0.6982,
|
| 1754 |
-
"step": 29100
|
| 1755 |
-
},
|
| 1756 |
-
{
|
| 1757 |
-
"epoch": 1.64,
|
| 1758 |
-
"learning_rate": 0.0,
|
| 1759 |
-
"loss": 0.7079,
|
| 1760 |
-
"step": 29200
|
| 1761 |
-
},
|
| 1762 |
-
{
|
| 1763 |
-
"epoch": 1.64,
|
| 1764 |
-
"learning_rate": 0.0,
|
| 1765 |
-
"loss": 0.6674,
|
| 1766 |
-
"step": 29300
|
| 1767 |
-
},
|
| 1768 |
-
{
|
| 1769 |
-
"epoch": 1.65,
|
| 1770 |
-
"learning_rate": 0.0,
|
| 1771 |
-
"loss": 0.718,
|
| 1772 |
-
"step": 29400
|
| 1773 |
-
},
|
| 1774 |
-
{
|
| 1775 |
-
"epoch": 1.65,
|
| 1776 |
-
"learning_rate": 0.0,
|
| 1777 |
-
"loss": 0.693,
|
| 1778 |
-
"step": 29500
|
| 1779 |
-
},
|
| 1780 |
-
{
|
| 1781 |
-
"epoch": 1.66,
|
| 1782 |
-
"learning_rate": 0.0,
|
| 1783 |
-
"loss": 0.7557,
|
| 1784 |
-
"step": 29600
|
| 1785 |
-
},
|
| 1786 |
-
{
|
| 1787 |
-
"epoch": 1.66,
|
| 1788 |
-
"learning_rate": 0.0,
|
| 1789 |
-
"loss": 0.7186,
|
| 1790 |
-
"step": 29700
|
| 1791 |
-
},
|
| 1792 |
-
{
|
| 1793 |
-
"epoch": 1.67,
|
| 1794 |
-
"learning_rate": 0.0,
|
| 1795 |
-
"loss": 0.6966,
|
| 1796 |
-
"step": 29800
|
| 1797 |
-
},
|
| 1798 |
-
{
|
| 1799 |
-
"epoch": 1.68,
|
| 1800 |
-
"learning_rate": 0.0,
|
| 1801 |
-
"loss": 0.7333,
|
| 1802 |
-
"step": 29900
|
| 1803 |
-
},
|
| 1804 |
-
{
|
| 1805 |
-
"epoch": 1.68,
|
| 1806 |
-
"learning_rate": 0.0,
|
| 1807 |
-
"loss": 0.6626,
|
| 1808 |
-
"step": 30000
|
| 1809 |
-
}
|
| 1810 |
-
],
|
| 1811 |
-
"max_steps": 32112,
|
| 1812 |
-
"num_train_epochs": 2,
|
| 1813 |
-
"total_flos": 1.384048497581568e+17,
|
| 1814 |
-
"trial_name": null,
|
| 1815 |
-
"trial_params": null
|
| 1816 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/training_args.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:55ba5b3f42228ed1cc9f39879296610a57bbd6ad0d8088954907f54e07dac886
|
| 3 |
-
size 3439
|
|
|
|
|
|
|
|
|
|
|
|
runs/Jan03_01-47-25_21bb7eba274c/events.out.tfevents.1672710460.21bb7eba274c.23.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0db0ff0415c34d9624f4448adaf7f5a5c762cf5eab86d0040d66365b867733b9
|
| 3 |
+
size 60870
|