Commit
•
8a25ec7
1
Parent(s):
4d4eea4
Model save
Browse files- last-checkpoint/config.json +0 -27
- last-checkpoint/optimizer.pt +0 -3
- last-checkpoint/pytorch_model.bin +0 -3
- last-checkpoint/rng_state_0.pth +0 -3
- last-checkpoint/rng_state_1.pth +0 -3
- last-checkpoint/rng_state_2.pth +0 -3
- last-checkpoint/rng_state_3.pth +0 -3
- last-checkpoint/rng_state_4.pth +0 -3
- last-checkpoint/rng_state_5.pth +0 -3
- last-checkpoint/rng_state_6.pth +0 -3
- last-checkpoint/rng_state_7.pth +0 -3
- last-checkpoint/scheduler.pt +0 -3
- last-checkpoint/special_tokens_map.json +0 -9
- last-checkpoint/tokenizer.json +0 -0
- last-checkpoint/tokenizer_config.json +0 -13
- last-checkpoint/trainer_state.json +0 -1248
- last-checkpoint/training_args.bin +0 -3
- runs/Feb06_00-07-48_t1v-n-58269bec-w-0/events.out.tfevents.1675642086.t1v-n-58269bec-w-0.23209.0 +2 -2
last-checkpoint/config.json
DELETED
|
@@ -1,27 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"_name_or_path": "data/plms/legal-czech-roberta-base",
|
| 3 |
-
"architectures": [
|
| 4 |
-
"RobertaForMaskedLM"
|
| 5 |
-
],
|
| 6 |
-
"attention_probs_dropout_prob": 0.1,
|
| 7 |
-
"bos_token_id": 1,
|
| 8 |
-
"classifier_dropout": null,
|
| 9 |
-
"eos_token_id": 2,
|
| 10 |
-
"hidden_act": "gelu",
|
| 11 |
-
"hidden_dropout_prob": 0.1,
|
| 12 |
-
"hidden_size": 768,
|
| 13 |
-
"initializer_range": 0.02,
|
| 14 |
-
"intermediate_size": 3072,
|
| 15 |
-
"layer_norm_eps": 1e-05,
|
| 16 |
-
"max_position_embeddings": 514,
|
| 17 |
-
"model_type": "roberta",
|
| 18 |
-
"num_attention_heads": 12,
|
| 19 |
-
"num_hidden_layers": 12,
|
| 20 |
-
"pad_token_id": 0,
|
| 21 |
-
"position_embedding_type": "absolute",
|
| 22 |
-
"torch_dtype": "float32",
|
| 23 |
-
"transformers_version": "4.20.1",
|
| 24 |
-
"type_vocab_size": 1,
|
| 25 |
-
"use_cache": true,
|
| 26 |
-
"vocab_size": 32000
|
| 27 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/optimizer.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:48440da5808134a6321409c9e1af932e241921cb0b391da7a1322846cff6854c
|
| 3 |
-
size 885325017
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/pytorch_model.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:ce072253c3b54cde4896b4ca3333404136e19caa6969b0ababd35d2c876d2c5e
|
| 3 |
-
size 442675755
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/rng_state_0.pth
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:66740bfa71631c8b49432c0222efa25cfa829a668716d033e26db3b852d01971
|
| 3 |
-
size 13611
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/rng_state_1.pth
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:66740bfa71631c8b49432c0222efa25cfa829a668716d033e26db3b852d01971
|
| 3 |
-
size 13611
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/rng_state_2.pth
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:66740bfa71631c8b49432c0222efa25cfa829a668716d033e26db3b852d01971
|
| 3 |
-
size 13611
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/rng_state_3.pth
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:66740bfa71631c8b49432c0222efa25cfa829a668716d033e26db3b852d01971
|
| 3 |
-
size 13611
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/rng_state_4.pth
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:66740bfa71631c8b49432c0222efa25cfa829a668716d033e26db3b852d01971
|
| 3 |
-
size 13611
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/rng_state_5.pth
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:66740bfa71631c8b49432c0222efa25cfa829a668716d033e26db3b852d01971
|
| 3 |
-
size 13611
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/rng_state_6.pth
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:66740bfa71631c8b49432c0222efa25cfa829a668716d033e26db3b852d01971
|
| 3 |
-
size 13611
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/rng_state_7.pth
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:66740bfa71631c8b49432c0222efa25cfa829a668716d033e26db3b852d01971
|
| 3 |
-
size 13611
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/scheduler.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:34c7791bbb10178054552ff1c1aa9bf08a101a8199906b7bf72dd42f5c977109
|
| 3 |
-
size 623
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/special_tokens_map.json
DELETED
|
@@ -1,9 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"bos_token": "<s>",
|
| 3 |
-
"cls_token": "<s>",
|
| 4 |
-
"eos_token": "</s>",
|
| 5 |
-
"mask_token": "<mask>",
|
| 6 |
-
"pad_token": "<pad>",
|
| 7 |
-
"sep_token": "</s>",
|
| 8 |
-
"unk_token": "<unk>"
|
| 9 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/tokenizer.json
DELETED
|
The diff for this file is too large to render.
See raw diff
|
|
|
last-checkpoint/tokenizer_config.json
DELETED
|
@@ -1,13 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"bos_token": "<s>",
|
| 3 |
-
"cls_token": "<s>",
|
| 4 |
-
"eos_token": "</s>",
|
| 5 |
-
"mask_token": "<mask>",
|
| 6 |
-
"model_max_length": 512,
|
| 7 |
-
"name_or_path": "data/plms/legal-czech-roberta-base",
|
| 8 |
-
"pad_token": "<pad>",
|
| 9 |
-
"sep_token": "</s>",
|
| 10 |
-
"special_tokens_map_file": "data/plms/legal-czech-roberta-base_32k/special_tokens_map.json",
|
| 11 |
-
"tokenizer_class": "PreTrainedTokenizerFast",
|
| 12 |
-
"unk_token": "<unk>"
|
| 13 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/trainer_state.json
DELETED
|
@@ -1,1248 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_metric": null,
|
| 3 |
-
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 2.21021,
|
| 5 |
-
"global_step": 200000,
|
| 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": 1e-05,
|
| 13 |
-
"loss": 6.5639,
|
| 14 |
-
"step": 1000
|
| 15 |
-
},
|
| 16 |
-
{
|
| 17 |
-
"epoch": 0.01,
|
| 18 |
-
"learning_rate": 2e-05,
|
| 19 |
-
"loss": 4.3433,
|
| 20 |
-
"step": 2000
|
| 21 |
-
},
|
| 22 |
-
{
|
| 23 |
-
"epoch": 0.01,
|
| 24 |
-
"learning_rate": 3e-05,
|
| 25 |
-
"loss": 4.0026,
|
| 26 |
-
"step": 3000
|
| 27 |
-
},
|
| 28 |
-
{
|
| 29 |
-
"epoch": 0.02,
|
| 30 |
-
"learning_rate": 4e-05,
|
| 31 |
-
"loss": 3.1046,
|
| 32 |
-
"step": 4000
|
| 33 |
-
},
|
| 34 |
-
{
|
| 35 |
-
"epoch": 0.03,
|
| 36 |
-
"learning_rate": 5e-05,
|
| 37 |
-
"loss": 2.78,
|
| 38 |
-
"step": 5000
|
| 39 |
-
},
|
| 40 |
-
{
|
| 41 |
-
"epoch": 0.03,
|
| 42 |
-
"learning_rate": 6e-05,
|
| 43 |
-
"loss": 2.2622,
|
| 44 |
-
"step": 6000
|
| 45 |
-
},
|
| 46 |
-
{
|
| 47 |
-
"epoch": 0.04,
|
| 48 |
-
"learning_rate": 7e-05,
|
| 49 |
-
"loss": 1.7388,
|
| 50 |
-
"step": 7000
|
| 51 |
-
},
|
| 52 |
-
{
|
| 53 |
-
"epoch": 0.04,
|
| 54 |
-
"learning_rate": 8e-05,
|
| 55 |
-
"loss": 1.3481,
|
| 56 |
-
"step": 8000
|
| 57 |
-
},
|
| 58 |
-
{
|
| 59 |
-
"epoch": 0.04,
|
| 60 |
-
"learning_rate": 9e-05,
|
| 61 |
-
"loss": 1.3416,
|
| 62 |
-
"step": 9000
|
| 63 |
-
},
|
| 64 |
-
{
|
| 65 |
-
"epoch": 0.05,
|
| 66 |
-
"learning_rate": 0.0001,
|
| 67 |
-
"loss": 1.0433,
|
| 68 |
-
"step": 10000
|
| 69 |
-
},
|
| 70 |
-
{
|
| 71 |
-
"epoch": 0.06,
|
| 72 |
-
"learning_rate": 9.999316524962345e-05,
|
| 73 |
-
"loss": 1.1522,
|
| 74 |
-
"step": 11000
|
| 75 |
-
},
|
| 76 |
-
{
|
| 77 |
-
"epoch": 0.06,
|
| 78 |
-
"learning_rate": 9.997266286704631e-05,
|
| 79 |
-
"loss": 1.4435,
|
| 80 |
-
"step": 12000
|
| 81 |
-
},
|
| 82 |
-
{
|
| 83 |
-
"epoch": 0.07,
|
| 84 |
-
"learning_rate": 9.993849845741524e-05,
|
| 85 |
-
"loss": 1.122,
|
| 86 |
-
"step": 13000
|
| 87 |
-
},
|
| 88 |
-
{
|
| 89 |
-
"epoch": 0.07,
|
| 90 |
-
"learning_rate": 9.989068136093873e-05,
|
| 91 |
-
"loss": 0.9974,
|
| 92 |
-
"step": 14000
|
| 93 |
-
},
|
| 94 |
-
{
|
| 95 |
-
"epoch": 0.07,
|
| 96 |
-
"learning_rate": 9.98292246503335e-05,
|
| 97 |
-
"loss": 0.9886,
|
| 98 |
-
"step": 15000
|
| 99 |
-
},
|
| 100 |
-
{
|
| 101 |
-
"epoch": 0.08,
|
| 102 |
-
"learning_rate": 9.975414512725057e-05,
|
| 103 |
-
"loss": 0.9671,
|
| 104 |
-
"step": 16000
|
| 105 |
-
},
|
| 106 |
-
{
|
| 107 |
-
"epoch": 0.09,
|
| 108 |
-
"learning_rate": 9.966546331768191e-05,
|
| 109 |
-
"loss": 1.0903,
|
| 110 |
-
"step": 17000
|
| 111 |
-
},
|
| 112 |
-
{
|
| 113 |
-
"epoch": 0.09,
|
| 114 |
-
"learning_rate": 9.956320346634876e-05,
|
| 115 |
-
"loss": 1.1959,
|
| 116 |
-
"step": 18000
|
| 117 |
-
},
|
| 118 |
-
{
|
| 119 |
-
"epoch": 0.1,
|
| 120 |
-
"learning_rate": 9.944739353007344e-05,
|
| 121 |
-
"loss": 0.9674,
|
| 122 |
-
"step": 19000
|
| 123 |
-
},
|
| 124 |
-
{
|
| 125 |
-
"epoch": 0.1,
|
| 126 |
-
"learning_rate": 9.931806517013612e-05,
|
| 127 |
-
"loss": 0.833,
|
| 128 |
-
"step": 20000
|
| 129 |
-
},
|
| 130 |
-
{
|
| 131 |
-
"epoch": 0.1,
|
| 132 |
-
"learning_rate": 9.917525374361912e-05,
|
| 133 |
-
"loss": 0.9286,
|
| 134 |
-
"step": 21000
|
| 135 |
-
},
|
| 136 |
-
{
|
| 137 |
-
"epoch": 0.11,
|
| 138 |
-
"learning_rate": 9.901899829374047e-05,
|
| 139 |
-
"loss": 0.9765,
|
| 140 |
-
"step": 22000
|
| 141 |
-
},
|
| 142 |
-
{
|
| 143 |
-
"epoch": 0.12,
|
| 144 |
-
"learning_rate": 9.884934153917997e-05,
|
| 145 |
-
"loss": 1.0718,
|
| 146 |
-
"step": 23000
|
| 147 |
-
},
|
| 148 |
-
{
|
| 149 |
-
"epoch": 0.12,
|
| 150 |
-
"learning_rate": 9.86663298624003e-05,
|
| 151 |
-
"loss": 1.0438,
|
| 152 |
-
"step": 24000
|
| 153 |
-
},
|
| 154 |
-
{
|
| 155 |
-
"epoch": 0.12,
|
| 156 |
-
"learning_rate": 9.847001329696653e-05,
|
| 157 |
-
"loss": 0.8762,
|
| 158 |
-
"step": 25000
|
| 159 |
-
},
|
| 160 |
-
{
|
| 161 |
-
"epoch": 0.13,
|
| 162 |
-
"learning_rate": 9.826044551386744e-05,
|
| 163 |
-
"loss": 0.7131,
|
| 164 |
-
"step": 26000
|
| 165 |
-
},
|
| 166 |
-
{
|
| 167 |
-
"epoch": 0.14,
|
| 168 |
-
"learning_rate": 9.803768380684242e-05,
|
| 169 |
-
"loss": 0.9814,
|
| 170 |
-
"step": 27000
|
| 171 |
-
},
|
| 172 |
-
{
|
| 173 |
-
"epoch": 0.14,
|
| 174 |
-
"learning_rate": 9.780178907671789e-05,
|
| 175 |
-
"loss": 1.0355,
|
| 176 |
-
"step": 28000
|
| 177 |
-
},
|
| 178 |
-
{
|
| 179 |
-
"epoch": 0.14,
|
| 180 |
-
"learning_rate": 9.755282581475769e-05,
|
| 181 |
-
"loss": 0.9196,
|
| 182 |
-
"step": 29000
|
| 183 |
-
},
|
| 184 |
-
{
|
| 185 |
-
"epoch": 0.15,
|
| 186 |
-
"learning_rate": 9.729086208503174e-05,
|
| 187 |
-
"loss": 0.9795,
|
| 188 |
-
"step": 30000
|
| 189 |
-
},
|
| 190 |
-
{
|
| 191 |
-
"epoch": 0.15,
|
| 192 |
-
"learning_rate": 9.701596950580806e-05,
|
| 193 |
-
"loss": 0.763,
|
| 194 |
-
"step": 31000
|
| 195 |
-
},
|
| 196 |
-
{
|
| 197 |
-
"epoch": 0.16,
|
| 198 |
-
"learning_rate": 9.672822322997305e-05,
|
| 199 |
-
"loss": 0.731,
|
| 200 |
-
"step": 32000
|
| 201 |
-
},
|
| 202 |
-
{
|
| 203 |
-
"epoch": 0.17,
|
| 204 |
-
"learning_rate": 9.642770192448536e-05,
|
| 205 |
-
"loss": 1.0306,
|
| 206 |
-
"step": 33000
|
| 207 |
-
},
|
| 208 |
-
{
|
| 209 |
-
"epoch": 0.17,
|
| 210 |
-
"learning_rate": 9.611448774886924e-05,
|
| 211 |
-
"loss": 1.0033,
|
| 212 |
-
"step": 34000
|
| 213 |
-
},
|
| 214 |
-
{
|
| 215 |
-
"epoch": 0.17,
|
| 216 |
-
"learning_rate": 9.578866633275288e-05,
|
| 217 |
-
"loss": 0.8266,
|
| 218 |
-
"step": 35000
|
| 219 |
-
},
|
| 220 |
-
{
|
| 221 |
-
"epoch": 0.18,
|
| 222 |
-
"learning_rate": 9.545032675245813e-05,
|
| 223 |
-
"loss": 0.9093,
|
| 224 |
-
"step": 36000
|
| 225 |
-
},
|
| 226 |
-
{
|
| 227 |
-
"epoch": 0.18,
|
| 228 |
-
"learning_rate": 9.509956150664796e-05,
|
| 229 |
-
"loss": 0.6842,
|
| 230 |
-
"step": 37000
|
| 231 |
-
},
|
| 232 |
-
{
|
| 233 |
-
"epoch": 0.19,
|
| 234 |
-
"learning_rate": 9.473646649103818e-05,
|
| 235 |
-
"loss": 0.8017,
|
| 236 |
-
"step": 38000
|
| 237 |
-
},
|
| 238 |
-
{
|
| 239 |
-
"epoch": 0.2,
|
| 240 |
-
"learning_rate": 9.43611409721806e-05,
|
| 241 |
-
"loss": 1.1086,
|
| 242 |
-
"step": 39000
|
| 243 |
-
},
|
| 244 |
-
{
|
| 245 |
-
"epoch": 0.2,
|
| 246 |
-
"learning_rate": 9.397368756032445e-05,
|
| 247 |
-
"loss": 0.8712,
|
| 248 |
-
"step": 40000
|
| 249 |
-
},
|
| 250 |
-
{
|
| 251 |
-
"epoch": 0.2,
|
| 252 |
-
"learning_rate": 9.357421218136386e-05,
|
| 253 |
-
"loss": 0.785,
|
| 254 |
-
"step": 41000
|
| 255 |
-
},
|
| 256 |
-
{
|
| 257 |
-
"epoch": 0.21,
|
| 258 |
-
"learning_rate": 9.316282404787871e-05,
|
| 259 |
-
"loss": 0.8121,
|
| 260 |
-
"step": 42000
|
| 261 |
-
},
|
| 262 |
-
{
|
| 263 |
-
"epoch": 0.21,
|
| 264 |
-
"learning_rate": 9.273963562927695e-05,
|
| 265 |
-
"loss": 0.7343,
|
| 266 |
-
"step": 43000
|
| 267 |
-
},
|
| 268 |
-
{
|
| 269 |
-
"epoch": 0.22,
|
| 270 |
-
"learning_rate": 9.230476262104677e-05,
|
| 271 |
-
"loss": 0.8619,
|
| 272 |
-
"step": 44000
|
| 273 |
-
},
|
| 274 |
-
{
|
| 275 |
-
"epoch": 0.23,
|
| 276 |
-
"learning_rate": 9.185832391312644e-05,
|
| 277 |
-
"loss": 1.0451,
|
| 278 |
-
"step": 45000
|
| 279 |
-
},
|
| 280 |
-
{
|
| 281 |
-
"epoch": 0.23,
|
| 282 |
-
"learning_rate": 9.140044155740101e-05,
|
| 283 |
-
"loss": 0.8101,
|
| 284 |
-
"step": 46000
|
| 285 |
-
},
|
| 286 |
-
{
|
| 287 |
-
"epoch": 0.23,
|
| 288 |
-
"learning_rate": 9.093124073433463e-05,
|
| 289 |
-
"loss": 0.719,
|
| 290 |
-
"step": 47000
|
| 291 |
-
},
|
| 292 |
-
{
|
| 293 |
-
"epoch": 0.24,
|
| 294 |
-
"learning_rate": 9.045084971874738e-05,
|
| 295 |
-
"loss": 0.765,
|
| 296 |
-
"step": 48000
|
| 297 |
-
},
|
| 298 |
-
{
|
| 299 |
-
"epoch": 0.24,
|
| 300 |
-
"learning_rate": 8.995939984474624e-05,
|
| 301 |
-
"loss": 0.8051,
|
| 302 |
-
"step": 49000
|
| 303 |
-
},
|
| 304 |
-
{
|
| 305 |
-
"epoch": 0.25,
|
| 306 |
-
"learning_rate": 8.945702546981969e-05,
|
| 307 |
-
"loss": 0.9425,
|
| 308 |
-
"step": 50000
|
| 309 |
-
},
|
| 310 |
-
{
|
| 311 |
-
"epoch": 0.25,
|
| 312 |
-
"eval_loss": 0.7593458890914917,
|
| 313 |
-
"eval_runtime": 16.7311,
|
| 314 |
-
"eval_samples_per_second": 298.845,
|
| 315 |
-
"eval_steps_per_second": 2.391,
|
| 316 |
-
"step": 50000
|
| 317 |
-
},
|
| 318 |
-
{
|
| 319 |
-
"epoch": 0.26,
|
| 320 |
-
"learning_rate": 8.894386393810563e-05,
|
| 321 |
-
"loss": 0.929,
|
| 322 |
-
"step": 51000
|
| 323 |
-
},
|
| 324 |
-
{
|
| 325 |
-
"epoch": 0.26,
|
| 326 |
-
"learning_rate": 8.842005554284296e-05,
|
| 327 |
-
"loss": 0.7752,
|
| 328 |
-
"step": 52000
|
| 329 |
-
},
|
| 330 |
-
{
|
| 331 |
-
"epoch": 0.27,
|
| 332 |
-
"learning_rate": 8.788574348801675e-05,
|
| 333 |
-
"loss": 0.6349,
|
| 334 |
-
"step": 53000
|
| 335 |
-
},
|
| 336 |
-
{
|
| 337 |
-
"epoch": 0.27,
|
| 338 |
-
"learning_rate": 8.73410738492077e-05,
|
| 339 |
-
"loss": 0.8323,
|
| 340 |
-
"step": 54000
|
| 341 |
-
},
|
| 342 |
-
{
|
| 343 |
-
"epoch": 0.28,
|
| 344 |
-
"learning_rate": 8.678619553365659e-05,
|
| 345 |
-
"loss": 0.8745,
|
| 346 |
-
"step": 55000
|
| 347 |
-
},
|
| 348 |
-
{
|
| 349 |
-
"epoch": 0.28,
|
| 350 |
-
"learning_rate": 8.622126023955446e-05,
|
| 351 |
-
"loss": 0.8607,
|
| 352 |
-
"step": 56000
|
| 353 |
-
},
|
| 354 |
-
{
|
| 355 |
-
"epoch": 0.28,
|
| 356 |
-
"learning_rate": 8.564642241456986e-05,
|
| 357 |
-
"loss": 0.8832,
|
| 358 |
-
"step": 57000
|
| 359 |
-
},
|
| 360 |
-
{
|
| 361 |
-
"epoch": 0.29,
|
| 362 |
-
"learning_rate": 8.506183921362443e-05,
|
| 363 |
-
"loss": 0.7021,
|
| 364 |
-
"step": 58000
|
| 365 |
-
},
|
| 366 |
-
{
|
| 367 |
-
"epoch": 0.29,
|
| 368 |
-
"learning_rate": 8.44676704559283e-05,
|
| 369 |
-
"loss": 0.6244,
|
| 370 |
-
"step": 59000
|
| 371 |
-
},
|
| 372 |
-
{
|
| 373 |
-
"epoch": 0.3,
|
| 374 |
-
"learning_rate": 8.386407858128706e-05,
|
| 375 |
-
"loss": 0.8984,
|
| 376 |
-
"step": 60000
|
| 377 |
-
},
|
| 378 |
-
{
|
| 379 |
-
"epoch": 0.3,
|
| 380 |
-
"learning_rate": 8.32512286056924e-05,
|
| 381 |
-
"loss": 0.9368,
|
| 382 |
-
"step": 61000
|
| 383 |
-
},
|
| 384 |
-
{
|
| 385 |
-
"epoch": 0.31,
|
| 386 |
-
"learning_rate": 8.262928807620843e-05,
|
| 387 |
-
"loss": 0.7612,
|
| 388 |
-
"step": 62000
|
| 389 |
-
},
|
| 390 |
-
{
|
| 391 |
-
"epoch": 0.32,
|
| 392 |
-
"learning_rate": 8.199842702516583e-05,
|
| 393 |
-
"loss": 0.8501,
|
| 394 |
-
"step": 63000
|
| 395 |
-
},
|
| 396 |
-
{
|
| 397 |
-
"epoch": 0.32,
|
| 398 |
-
"learning_rate": 8.135881792367686e-05,
|
| 399 |
-
"loss": 0.6263,
|
| 400 |
-
"step": 64000
|
| 401 |
-
},
|
| 402 |
-
{
|
| 403 |
-
"epoch": 0.33,
|
| 404 |
-
"learning_rate": 8.07106356344834e-05,
|
| 405 |
-
"loss": 0.6996,
|
| 406 |
-
"step": 65000
|
| 407 |
-
},
|
| 408 |
-
{
|
| 409 |
-
"epoch": 0.33,
|
| 410 |
-
"learning_rate": 8.005405736415126e-05,
|
| 411 |
-
"loss": 0.9828,
|
| 412 |
-
"step": 66000
|
| 413 |
-
},
|
| 414 |
-
{
|
| 415 |
-
"epoch": 0.34,
|
| 416 |
-
"learning_rate": 7.938926261462366e-05,
|
| 417 |
-
"loss": 0.8341,
|
| 418 |
-
"step": 67000
|
| 419 |
-
},
|
| 420 |
-
{
|
| 421 |
-
"epoch": 0.34,
|
| 422 |
-
"learning_rate": 7.871643313414718e-05,
|
| 423 |
-
"loss": 0.7325,
|
| 424 |
-
"step": 68000
|
| 425 |
-
},
|
| 426 |
-
{
|
| 427 |
-
"epoch": 0.34,
|
| 428 |
-
"learning_rate": 7.803575286758364e-05,
|
| 429 |
-
"loss": 0.7716,
|
| 430 |
-
"step": 69000
|
| 431 |
-
},
|
| 432 |
-
{
|
| 433 |
-
"epoch": 0.35,
|
| 434 |
-
"learning_rate": 7.734740790612136e-05,
|
| 435 |
-
"loss": 0.642,
|
| 436 |
-
"step": 70000
|
| 437 |
-
},
|
| 438 |
-
{
|
| 439 |
-
"epoch": 0.35,
|
| 440 |
-
"learning_rate": 7.66515864363997e-05,
|
| 441 |
-
"loss": 0.7692,
|
| 442 |
-
"step": 71000
|
| 443 |
-
},
|
| 444 |
-
{
|
| 445 |
-
"epoch": 0.36,
|
| 446 |
-
"learning_rate": 7.594847868906076e-05,
|
| 447 |
-
"loss": 1.0059,
|
| 448 |
-
"step": 72000
|
| 449 |
-
},
|
| 450 |
-
{
|
| 451 |
-
"epoch": 0.36,
|
| 452 |
-
"learning_rate": 7.52382768867422e-05,
|
| 453 |
-
"loss": 0.7512,
|
| 454 |
-
"step": 73000
|
| 455 |
-
},
|
| 456 |
-
{
|
| 457 |
-
"epoch": 0.37,
|
| 458 |
-
"learning_rate": 7.452117519152542e-05,
|
| 459 |
-
"loss": 0.6939,
|
| 460 |
-
"step": 74000
|
| 461 |
-
},
|
| 462 |
-
{
|
| 463 |
-
"epoch": 0.38,
|
| 464 |
-
"learning_rate": 7.379736965185368e-05,
|
| 465 |
-
"loss": 0.7031,
|
| 466 |
-
"step": 75000
|
| 467 |
-
},
|
| 468 |
-
{
|
| 469 |
-
"epoch": 0.38,
|
| 470 |
-
"learning_rate": 7.30670581489344e-05,
|
| 471 |
-
"loss": 0.7218,
|
| 472 |
-
"step": 76000
|
| 473 |
-
},
|
| 474 |
-
{
|
| 475 |
-
"epoch": 0.39,
|
| 476 |
-
"learning_rate": 7.233044034264034e-05,
|
| 477 |
-
"loss": 0.8535,
|
| 478 |
-
"step": 77000
|
| 479 |
-
},
|
| 480 |
-
{
|
| 481 |
-
"epoch": 0.39,
|
| 482 |
-
"learning_rate": 7.158771761692464e-05,
|
| 483 |
-
"loss": 0.8998,
|
| 484 |
-
"step": 78000
|
| 485 |
-
},
|
| 486 |
-
{
|
| 487 |
-
"epoch": 1.0,
|
| 488 |
-
"learning_rate": 7.083909302476453e-05,
|
| 489 |
-
"loss": 0.7264,
|
| 490 |
-
"step": 79000
|
| 491 |
-
},
|
| 492 |
-
{
|
| 493 |
-
"epoch": 1.01,
|
| 494 |
-
"learning_rate": 7.008477123264848e-05,
|
| 495 |
-
"loss": 0.8991,
|
| 496 |
-
"step": 80000
|
| 497 |
-
},
|
| 498 |
-
{
|
| 499 |
-
"epoch": 1.01,
|
| 500 |
-
"learning_rate": 6.932495846462261e-05,
|
| 501 |
-
"loss": 0.7313,
|
| 502 |
-
"step": 81000
|
| 503 |
-
},
|
| 504 |
-
{
|
| 505 |
-
"epoch": 1.02,
|
| 506 |
-
"learning_rate": 6.855986244591104e-05,
|
| 507 |
-
"loss": 0.8268,
|
| 508 |
-
"step": 82000
|
| 509 |
-
},
|
| 510 |
-
{
|
| 511 |
-
"epoch": 1.02,
|
| 512 |
-
"learning_rate": 6.778969234612584e-05,
|
| 513 |
-
"loss": 0.6143,
|
| 514 |
-
"step": 83000
|
| 515 |
-
},
|
| 516 |
-
{
|
| 517 |
-
"epoch": 1.03,
|
| 518 |
-
"learning_rate": 6.701465872208216e-05,
|
| 519 |
-
"loss": 0.6512,
|
| 520 |
-
"step": 84000
|
| 521 |
-
},
|
| 522 |
-
{
|
| 523 |
-
"epoch": 1.03,
|
| 524 |
-
"learning_rate": 6.623497346023418e-05,
|
| 525 |
-
"loss": 0.9238,
|
| 526 |
-
"step": 85000
|
| 527 |
-
},
|
| 528 |
-
{
|
| 529 |
-
"epoch": 1.04,
|
| 530 |
-
"learning_rate": 6.545084971874738e-05,
|
| 531 |
-
"loss": 0.8349,
|
| 532 |
-
"step": 86000
|
| 533 |
-
},
|
| 534 |
-
{
|
| 535 |
-
"epoch": 1.04,
|
| 536 |
-
"learning_rate": 6.466250186922325e-05,
|
| 537 |
-
"loss": 0.6955,
|
| 538 |
-
"step": 87000
|
| 539 |
-
},
|
| 540 |
-
{
|
| 541 |
-
"epoch": 1.05,
|
| 542 |
-
"learning_rate": 6.387014543809223e-05,
|
| 543 |
-
"loss": 0.7628,
|
| 544 |
-
"step": 88000
|
| 545 |
-
},
|
| 546 |
-
{
|
| 547 |
-
"epoch": 1.05,
|
| 548 |
-
"learning_rate": 6.307399704769099e-05,
|
| 549 |
-
"loss": 0.6049,
|
| 550 |
-
"step": 89000
|
| 551 |
-
},
|
| 552 |
-
{
|
| 553 |
-
"epoch": 1.06,
|
| 554 |
-
"learning_rate": 6.227427435703997e-05,
|
| 555 |
-
"loss": 0.7122,
|
| 556 |
-
"step": 90000
|
| 557 |
-
},
|
| 558 |
-
{
|
| 559 |
-
"epoch": 1.06,
|
| 560 |
-
"learning_rate": 6.147119600233758e-05,
|
| 561 |
-
"loss": 0.9885,
|
| 562 |
-
"step": 91000
|
| 563 |
-
},
|
| 564 |
-
{
|
| 565 |
-
"epoch": 1.07,
|
| 566 |
-
"learning_rate": 6.066498153718735e-05,
|
| 567 |
-
"loss": 0.7389,
|
| 568 |
-
"step": 92000
|
| 569 |
-
},
|
| 570 |
-
{
|
| 571 |
-
"epoch": 1.07,
|
| 572 |
-
"learning_rate": 5.985585137257401e-05,
|
| 573 |
-
"loss": 0.6712,
|
| 574 |
-
"step": 93000
|
| 575 |
-
},
|
| 576 |
-
{
|
| 577 |
-
"epoch": 1.08,
|
| 578 |
-
"learning_rate": 5.90440267166055e-05,
|
| 579 |
-
"loss": 0.6921,
|
| 580 |
-
"step": 94000
|
| 581 |
-
},
|
| 582 |
-
{
|
| 583 |
-
"epoch": 1.08,
|
| 584 |
-
"learning_rate": 5.8229729514036705e-05,
|
| 585 |
-
"loss": 0.6788,
|
| 586 |
-
"step": 95000
|
| 587 |
-
},
|
| 588 |
-
{
|
| 589 |
-
"epoch": 1.09,
|
| 590 |
-
"learning_rate": 5.74131823855921e-05,
|
| 591 |
-
"loss": 0.7947,
|
| 592 |
-
"step": 96000
|
| 593 |
-
},
|
| 594 |
-
{
|
| 595 |
-
"epoch": 1.09,
|
| 596 |
-
"learning_rate": 5.6594608567103456e-05,
|
| 597 |
-
"loss": 0.8967,
|
| 598 |
-
"step": 97000
|
| 599 |
-
},
|
| 600 |
-
{
|
| 601 |
-
"epoch": 1.1,
|
| 602 |
-
"learning_rate": 5.577423184847932e-05,
|
| 603 |
-
"loss": 0.7084,
|
| 604 |
-
"step": 98000
|
| 605 |
-
},
|
| 606 |
-
{
|
| 607 |
-
"epoch": 1.1,
|
| 608 |
-
"learning_rate": 5.495227651252315e-05,
|
| 609 |
-
"loss": 0.6064,
|
| 610 |
-
"step": 99000
|
| 611 |
-
},
|
| 612 |
-
{
|
| 613 |
-
"epoch": 1.11,
|
| 614 |
-
"learning_rate": 5.4128967273616625e-05,
|
| 615 |
-
"loss": 0.712,
|
| 616 |
-
"step": 100000
|
| 617 |
-
},
|
| 618 |
-
{
|
| 619 |
-
"epoch": 1.11,
|
| 620 |
-
"eval_loss": 0.675422191619873,
|
| 621 |
-
"eval_runtime": 10.1951,
|
| 622 |
-
"eval_samples_per_second": 490.432,
|
| 623 |
-
"eval_steps_per_second": 3.923,
|
| 624 |
-
"step": 100000
|
| 625 |
-
},
|
| 626 |
-
{
|
| 627 |
-
"epoch": 1.11,
|
| 628 |
-
"learning_rate": 5.330452921628497e-05,
|
| 629 |
-
"loss": 0.7451,
|
| 630 |
-
"step": 101000
|
| 631 |
-
},
|
| 632 |
-
{
|
| 633 |
-
"epoch": 1.12,
|
| 634 |
-
"learning_rate": 5.247918773366112e-05,
|
| 635 |
-
"loss": 0.8311,
|
| 636 |
-
"step": 102000
|
| 637 |
-
},
|
| 638 |
-
{
|
| 639 |
-
"epoch": 1.12,
|
| 640 |
-
"learning_rate": 5.165316846586541e-05,
|
| 641 |
-
"loss": 0.8221,
|
| 642 |
-
"step": 103000
|
| 643 |
-
},
|
| 644 |
-
{
|
| 645 |
-
"epoch": 1.13,
|
| 646 |
-
"learning_rate": 5.0826697238317935e-05,
|
| 647 |
-
"loss": 0.6741,
|
| 648 |
-
"step": 104000
|
| 649 |
-
},
|
| 650 |
-
{
|
| 651 |
-
"epoch": 1.13,
|
| 652 |
-
"learning_rate": 5e-05,
|
| 653 |
-
"loss": 0.5447,
|
| 654 |
-
"step": 105000
|
| 655 |
-
},
|
| 656 |
-
{
|
| 657 |
-
"epoch": 1.14,
|
| 658 |
-
"learning_rate": 4.917330276168208e-05,
|
| 659 |
-
"loss": 0.7913,
|
| 660 |
-
"step": 106000
|
| 661 |
-
},
|
| 662 |
-
{
|
| 663 |
-
"epoch": 1.14,
|
| 664 |
-
"learning_rate": 4.834683153413459e-05,
|
| 665 |
-
"loss": 0.8348,
|
| 666 |
-
"step": 107000
|
| 667 |
-
},
|
| 668 |
-
{
|
| 669 |
-
"epoch": 1.15,
|
| 670 |
-
"learning_rate": 4.7520812266338885e-05,
|
| 671 |
-
"loss": 0.7295,
|
| 672 |
-
"step": 108000
|
| 673 |
-
},
|
| 674 |
-
{
|
| 675 |
-
"epoch": 1.15,
|
| 676 |
-
"learning_rate": 4.669547078371504e-05,
|
| 677 |
-
"loss": 0.8008,
|
| 678 |
-
"step": 109000
|
| 679 |
-
},
|
| 680 |
-
{
|
| 681 |
-
"epoch": 1.16,
|
| 682 |
-
"learning_rate": 4.5871032726383386e-05,
|
| 683 |
-
"loss": 0.6031,
|
| 684 |
-
"step": 110000
|
| 685 |
-
},
|
| 686 |
-
{
|
| 687 |
-
"epoch": 1.16,
|
| 688 |
-
"learning_rate": 4.504772348747687e-05,
|
| 689 |
-
"loss": 0.5856,
|
| 690 |
-
"step": 111000
|
| 691 |
-
},
|
| 692 |
-
{
|
| 693 |
-
"epoch": 1.17,
|
| 694 |
-
"learning_rate": 4.4225768151520694e-05,
|
| 695 |
-
"loss": 0.8584,
|
| 696 |
-
"step": 112000
|
| 697 |
-
},
|
| 698 |
-
{
|
| 699 |
-
"epoch": 1.17,
|
| 700 |
-
"learning_rate": 4.3405391432896555e-05,
|
| 701 |
-
"loss": 0.8269,
|
| 702 |
-
"step": 113000
|
| 703 |
-
},
|
| 704 |
-
{
|
| 705 |
-
"epoch": 1.18,
|
| 706 |
-
"learning_rate": 4.2586817614407895e-05,
|
| 707 |
-
"loss": 0.6727,
|
| 708 |
-
"step": 114000
|
| 709 |
-
},
|
| 710 |
-
{
|
| 711 |
-
"epoch": 1.18,
|
| 712 |
-
"learning_rate": 4.17702704859633e-05,
|
| 713 |
-
"loss": 0.7562,
|
| 714 |
-
"step": 115000
|
| 715 |
-
},
|
| 716 |
-
{
|
| 717 |
-
"epoch": 1.19,
|
| 718 |
-
"learning_rate": 4.095597328339452e-05,
|
| 719 |
-
"loss": 0.5582,
|
| 720 |
-
"step": 116000
|
| 721 |
-
},
|
| 722 |
-
{
|
| 723 |
-
"epoch": 1.19,
|
| 724 |
-
"learning_rate": 4.0144148627425993e-05,
|
| 725 |
-
"loss": 0.6643,
|
| 726 |
-
"step": 117000
|
| 727 |
-
},
|
| 728 |
-
{
|
| 729 |
-
"epoch": 1.2,
|
| 730 |
-
"learning_rate": 3.933501846281267e-05,
|
| 731 |
-
"loss": 0.9486,
|
| 732 |
-
"step": 118000
|
| 733 |
-
},
|
| 734 |
-
{
|
| 735 |
-
"epoch": 1.2,
|
| 736 |
-
"learning_rate": 3.852880399766243e-05,
|
| 737 |
-
"loss": 0.7244,
|
| 738 |
-
"step": 119000
|
| 739 |
-
},
|
| 740 |
-
{
|
| 741 |
-
"epoch": 1.21,
|
| 742 |
-
"learning_rate": 3.772572564296005e-05,
|
| 743 |
-
"loss": 0.6545,
|
| 744 |
-
"step": 120000
|
| 745 |
-
},
|
| 746 |
-
{
|
| 747 |
-
"epoch": 1.21,
|
| 748 |
-
"learning_rate": 3.6926002952309016e-05,
|
| 749 |
-
"loss": 0.6842,
|
| 750 |
-
"step": 121000
|
| 751 |
-
},
|
| 752 |
-
{
|
| 753 |
-
"epoch": 1.22,
|
| 754 |
-
"learning_rate": 3.612985456190778e-05,
|
| 755 |
-
"loss": 0.6159,
|
| 756 |
-
"step": 122000
|
| 757 |
-
},
|
| 758 |
-
{
|
| 759 |
-
"epoch": 1.22,
|
| 760 |
-
"learning_rate": 3.533749813077677e-05,
|
| 761 |
-
"loss": 0.7326,
|
| 762 |
-
"step": 123000
|
| 763 |
-
},
|
| 764 |
-
{
|
| 765 |
-
"epoch": 1.23,
|
| 766 |
-
"learning_rate": 3.4549150281252636e-05,
|
| 767 |
-
"loss": 0.9025,
|
| 768 |
-
"step": 124000
|
| 769 |
-
},
|
| 770 |
-
{
|
| 771 |
-
"epoch": 1.23,
|
| 772 |
-
"learning_rate": 3.3765026539765834e-05,
|
| 773 |
-
"loss": 0.6844,
|
| 774 |
-
"step": 125000
|
| 775 |
-
},
|
| 776 |
-
{
|
| 777 |
-
"epoch": 1.24,
|
| 778 |
-
"learning_rate": 3.298534127791785e-05,
|
| 779 |
-
"loss": 0.6053,
|
| 780 |
-
"step": 126000
|
| 781 |
-
},
|
| 782 |
-
{
|
| 783 |
-
"epoch": 1.24,
|
| 784 |
-
"learning_rate": 3.221030765387417e-05,
|
| 785 |
-
"loss": 0.6597,
|
| 786 |
-
"step": 127000
|
| 787 |
-
},
|
| 788 |
-
{
|
| 789 |
-
"epoch": 1.25,
|
| 790 |
-
"learning_rate": 3.144013755408895e-05,
|
| 791 |
-
"loss": 0.6888,
|
| 792 |
-
"step": 128000
|
| 793 |
-
},
|
| 794 |
-
{
|
| 795 |
-
"epoch": 1.25,
|
| 796 |
-
"learning_rate": 3.0675041535377405e-05,
|
| 797 |
-
"loss": 0.8143,
|
| 798 |
-
"step": 129000
|
| 799 |
-
},
|
| 800 |
-
{
|
| 801 |
-
"epoch": 1.26,
|
| 802 |
-
"learning_rate": 2.991522876735154e-05,
|
| 803 |
-
"loss": 0.8078,
|
| 804 |
-
"step": 130000
|
| 805 |
-
},
|
| 806 |
-
{
|
| 807 |
-
"epoch": 1.26,
|
| 808 |
-
"learning_rate": 2.916090697523549e-05,
|
| 809 |
-
"loss": 0.6652,
|
| 810 |
-
"step": 131000
|
| 811 |
-
},
|
| 812 |
-
{
|
| 813 |
-
"epoch": 1.27,
|
| 814 |
-
"learning_rate": 2.8412282383075363e-05,
|
| 815 |
-
"loss": 0.5361,
|
| 816 |
-
"step": 132000
|
| 817 |
-
},
|
| 818 |
-
{
|
| 819 |
-
"epoch": 1.27,
|
| 820 |
-
"learning_rate": 2.766955965735968e-05,
|
| 821 |
-
"loss": 0.7314,
|
| 822 |
-
"step": 133000
|
| 823 |
-
},
|
| 824 |
-
{
|
| 825 |
-
"epoch": 1.28,
|
| 826 |
-
"learning_rate": 2.693294185106562e-05,
|
| 827 |
-
"loss": 0.7631,
|
| 828 |
-
"step": 134000
|
| 829 |
-
},
|
| 830 |
-
{
|
| 831 |
-
"epoch": 1.28,
|
| 832 |
-
"learning_rate": 2.6202630348146324e-05,
|
| 833 |
-
"loss": 0.7479,
|
| 834 |
-
"step": 135000
|
| 835 |
-
},
|
| 836 |
-
{
|
| 837 |
-
"epoch": 1.29,
|
| 838 |
-
"learning_rate": 2.547882480847461e-05,
|
| 839 |
-
"loss": 0.7767,
|
| 840 |
-
"step": 136000
|
| 841 |
-
},
|
| 842 |
-
{
|
| 843 |
-
"epoch": 1.29,
|
| 844 |
-
"learning_rate": 2.476172311325783e-05,
|
| 845 |
-
"loss": 0.6033,
|
| 846 |
-
"step": 137000
|
| 847 |
-
},
|
| 848 |
-
{
|
| 849 |
-
"epoch": 1.3,
|
| 850 |
-
"learning_rate": 2.405152131093926e-05,
|
| 851 |
-
"loss": 0.5424,
|
| 852 |
-
"step": 138000
|
| 853 |
-
},
|
| 854 |
-
{
|
| 855 |
-
"epoch": 1.3,
|
| 856 |
-
"learning_rate": 2.3348413563600325e-05,
|
| 857 |
-
"loss": 0.7977,
|
| 858 |
-
"step": 139000
|
| 859 |
-
},
|
| 860 |
-
{
|
| 861 |
-
"epoch": 1.31,
|
| 862 |
-
"learning_rate": 2.2652592093878666e-05,
|
| 863 |
-
"loss": 0.8251,
|
| 864 |
-
"step": 140000
|
| 865 |
-
},
|
| 866 |
-
{
|
| 867 |
-
"epoch": 1.31,
|
| 868 |
-
"learning_rate": 2.196424713241637e-05,
|
| 869 |
-
"loss": 0.6668,
|
| 870 |
-
"step": 141000
|
| 871 |
-
},
|
| 872 |
-
{
|
| 873 |
-
"epoch": 1.32,
|
| 874 |
-
"learning_rate": 2.128356686585282e-05,
|
| 875 |
-
"loss": 0.7531,
|
| 876 |
-
"step": 142000
|
| 877 |
-
},
|
| 878 |
-
{
|
| 879 |
-
"epoch": 1.32,
|
| 880 |
-
"learning_rate": 2.061073738537635e-05,
|
| 881 |
-
"loss": 0.5407,
|
| 882 |
-
"step": 143000
|
| 883 |
-
},
|
| 884 |
-
{
|
| 885 |
-
"epoch": 1.33,
|
| 886 |
-
"learning_rate": 1.9945942635848748e-05,
|
| 887 |
-
"loss": 0.6199,
|
| 888 |
-
"step": 144000
|
| 889 |
-
},
|
| 890 |
-
{
|
| 891 |
-
"epoch": 1.33,
|
| 892 |
-
"learning_rate": 1.928936436551661e-05,
|
| 893 |
-
"loss": 0.8869,
|
| 894 |
-
"step": 145000
|
| 895 |
-
},
|
| 896 |
-
{
|
| 897 |
-
"epoch": 1.34,
|
| 898 |
-
"learning_rate": 1.8641182076323148e-05,
|
| 899 |
-
"loss": 0.7356,
|
| 900 |
-
"step": 146000
|
| 901 |
-
},
|
| 902 |
-
{
|
| 903 |
-
"epoch": 1.34,
|
| 904 |
-
"learning_rate": 1.800157297483417e-05,
|
| 905 |
-
"loss": 0.6482,
|
| 906 |
-
"step": 147000
|
| 907 |
-
},
|
| 908 |
-
{
|
| 909 |
-
"epoch": 1.35,
|
| 910 |
-
"learning_rate": 1.7370711923791567e-05,
|
| 911 |
-
"loss": 0.6856,
|
| 912 |
-
"step": 148000
|
| 913 |
-
},
|
| 914 |
-
{
|
| 915 |
-
"epoch": 1.35,
|
| 916 |
-
"learning_rate": 1.6748771394307585e-05,
|
| 917 |
-
"loss": 0.5664,
|
| 918 |
-
"step": 149000
|
| 919 |
-
},
|
| 920 |
-
{
|
| 921 |
-
"epoch": 1.36,
|
| 922 |
-
"learning_rate": 1.6135921418712956e-05,
|
| 923 |
-
"loss": 0.6889,
|
| 924 |
-
"step": 150000
|
| 925 |
-
},
|
| 926 |
-
{
|
| 927 |
-
"epoch": 1.36,
|
| 928 |
-
"eval_loss": 0.619526207447052,
|
| 929 |
-
"eval_runtime": 17.5567,
|
| 930 |
-
"eval_samples_per_second": 284.792,
|
| 931 |
-
"eval_steps_per_second": 2.278,
|
| 932 |
-
"step": 150000
|
| 933 |
-
},
|
| 934 |
-
{
|
| 935 |
-
"epoch": 1.36,
|
| 936 |
-
"learning_rate": 1.553232954407171e-05,
|
| 937 |
-
"loss": 0.9116,
|
| 938 |
-
"step": 151000
|
| 939 |
-
},
|
| 940 |
-
{
|
| 941 |
-
"epoch": 1.37,
|
| 942 |
-
"learning_rate": 1.4938160786375572e-05,
|
| 943 |
-
"loss": 0.6653,
|
| 944 |
-
"step": 152000
|
| 945 |
-
},
|
| 946 |
-
{
|
| 947 |
-
"epoch": 1.37,
|
| 948 |
-
"learning_rate": 1.435357758543015e-05,
|
| 949 |
-
"loss": 0.6182,
|
| 950 |
-
"step": 153000
|
| 951 |
-
},
|
| 952 |
-
{
|
| 953 |
-
"epoch": 1.38,
|
| 954 |
-
"learning_rate": 1.3778739760445552e-05,
|
| 955 |
-
"loss": 0.6328,
|
| 956 |
-
"step": 154000
|
| 957 |
-
},
|
| 958 |
-
{
|
| 959 |
-
"epoch": 1.38,
|
| 960 |
-
"learning_rate": 1.3213804466343421e-05,
|
| 961 |
-
"loss": 0.6461,
|
| 962 |
-
"step": 155000
|
| 963 |
-
},
|
| 964 |
-
{
|
| 965 |
-
"epoch": 1.39,
|
| 966 |
-
"learning_rate": 1.2658926150792322e-05,
|
| 967 |
-
"loss": 0.7774,
|
| 968 |
-
"step": 156000
|
| 969 |
-
},
|
| 970 |
-
{
|
| 971 |
-
"epoch": 1.39,
|
| 972 |
-
"learning_rate": 1.2114256511983274e-05,
|
| 973 |
-
"loss": 0.8142,
|
| 974 |
-
"step": 157000
|
| 975 |
-
},
|
| 976 |
-
{
|
| 977 |
-
"epoch": 2.0,
|
| 978 |
-
"learning_rate": 1.157994445715706e-05,
|
| 979 |
-
"loss": 0.6521,
|
| 980 |
-
"step": 158000
|
| 981 |
-
},
|
| 982 |
-
{
|
| 983 |
-
"epoch": 2.01,
|
| 984 |
-
"learning_rate": 1.1056136061894384e-05,
|
| 985 |
-
"loss": 0.8228,
|
| 986 |
-
"step": 159000
|
| 987 |
-
},
|
| 988 |
-
{
|
| 989 |
-
"epoch": 2.01,
|
| 990 |
-
"learning_rate": 1.0542974530180327e-05,
|
| 991 |
-
"loss": 0.6581,
|
| 992 |
-
"step": 160000
|
| 993 |
-
},
|
| 994 |
-
{
|
| 995 |
-
"epoch": 2.02,
|
| 996 |
-
"learning_rate": 1.0040600155253765e-05,
|
| 997 |
-
"loss": 0.7555,
|
| 998 |
-
"step": 161000
|
| 999 |
-
},
|
| 1000 |
-
{
|
| 1001 |
-
"epoch": 2.02,
|
| 1002 |
-
"learning_rate": 9.549150281252633e-06,
|
| 1003 |
-
"loss": 0.5475,
|
| 1004 |
-
"step": 162000
|
| 1005 |
-
},
|
| 1006 |
-
{
|
| 1007 |
-
"epoch": 2.03,
|
| 1008 |
-
"learning_rate": 9.068759265665384e-06,
|
| 1009 |
-
"loss": 0.5882,
|
| 1010 |
-
"step": 163000
|
| 1011 |
-
},
|
| 1012 |
-
{
|
| 1013 |
-
"epoch": 2.03,
|
| 1014 |
-
"learning_rate": 8.599558442598998e-06,
|
| 1015 |
-
"loss": 0.8543,
|
| 1016 |
-
"step": 164000
|
| 1017 |
-
},
|
| 1018 |
-
{
|
| 1019 |
-
"epoch": 2.04,
|
| 1020 |
-
"learning_rate": 8.141676086873572e-06,
|
| 1021 |
-
"loss": 0.7606,
|
| 1022 |
-
"step": 165000
|
| 1023 |
-
},
|
| 1024 |
-
{
|
| 1025 |
-
"epoch": 2.04,
|
| 1026 |
-
"learning_rate": 7.695237378953223e-06,
|
| 1027 |
-
"loss": 0.6321,
|
| 1028 |
-
"step": 166000
|
| 1029 |
-
},
|
| 1030 |
-
{
|
| 1031 |
-
"epoch": 2.05,
|
| 1032 |
-
"learning_rate": 7.260364370723044e-06,
|
| 1033 |
-
"loss": 0.7002,
|
| 1034 |
-
"step": 167000
|
| 1035 |
-
},
|
| 1036 |
-
{
|
| 1037 |
-
"epoch": 2.05,
|
| 1038 |
-
"learning_rate": 6.837175952121306e-06,
|
| 1039 |
-
"loss": 0.5491,
|
| 1040 |
-
"step": 168000
|
| 1041 |
-
},
|
| 1042 |
-
{
|
| 1043 |
-
"epoch": 2.06,
|
| 1044 |
-
"learning_rate": 6.425787818636131e-06,
|
| 1045 |
-
"loss": 0.6536,
|
| 1046 |
-
"step": 169000
|
| 1047 |
-
},
|
| 1048 |
-
{
|
| 1049 |
-
"epoch": 2.06,
|
| 1050 |
-
"learning_rate": 6.026312439675552e-06,
|
| 1051 |
-
"loss": 0.917,
|
| 1052 |
-
"step": 170000
|
| 1053 |
-
},
|
| 1054 |
-
{
|
| 1055 |
-
"epoch": 2.07,
|
| 1056 |
-
"learning_rate": 5.6388590278194096e-06,
|
| 1057 |
-
"loss": 0.6754,
|
| 1058 |
-
"step": 171000
|
| 1059 |
-
},
|
| 1060 |
-
{
|
| 1061 |
-
"epoch": 2.07,
|
| 1062 |
-
"learning_rate": 5.263533508961827e-06,
|
| 1063 |
-
"loss": 0.613,
|
| 1064 |
-
"step": 172000
|
| 1065 |
-
},
|
| 1066 |
-
{
|
| 1067 |
-
"epoch": 2.08,
|
| 1068 |
-
"learning_rate": 4.900438493352055e-06,
|
| 1069 |
-
"loss": 0.6372,
|
| 1070 |
-
"step": 173000
|
| 1071 |
-
},
|
| 1072 |
-
{
|
| 1073 |
-
"epoch": 2.08,
|
| 1074 |
-
"learning_rate": 4.549673247541875e-06,
|
| 1075 |
-
"loss": 0.6282,
|
| 1076 |
-
"step": 174000
|
| 1077 |
-
},
|
| 1078 |
-
{
|
| 1079 |
-
"epoch": 2.09,
|
| 1080 |
-
"learning_rate": 4.2113336672471245e-06,
|
| 1081 |
-
"loss": 0.7406,
|
| 1082 |
-
"step": 175000
|
| 1083 |
-
},
|
| 1084 |
-
{
|
| 1085 |
-
"epoch": 2.09,
|
| 1086 |
-
"learning_rate": 3.885512251130763e-06,
|
| 1087 |
-
"loss": 0.8337,
|
| 1088 |
-
"step": 176000
|
| 1089 |
-
},
|
| 1090 |
-
{
|
| 1091 |
-
"epoch": 2.1,
|
| 1092 |
-
"learning_rate": 3.5722980755146517e-06,
|
| 1093 |
-
"loss": 0.6544,
|
| 1094 |
-
"step": 177000
|
| 1095 |
-
},
|
| 1096 |
-
{
|
| 1097 |
-
"epoch": 2.1,
|
| 1098 |
-
"learning_rate": 3.271776770026963e-06,
|
| 1099 |
-
"loss": 0.5552,
|
| 1100 |
-
"step": 178000
|
| 1101 |
-
},
|
| 1102 |
-
{
|
| 1103 |
-
"epoch": 2.11,
|
| 1104 |
-
"learning_rate": 2.9840304941919415e-06,
|
| 1105 |
-
"loss": 0.6663,
|
| 1106 |
-
"step": 179000
|
| 1107 |
-
},
|
| 1108 |
-
{
|
| 1109 |
-
"epoch": 2.11,
|
| 1110 |
-
"learning_rate": 2.7091379149682685e-06,
|
| 1111 |
-
"loss": 0.6946,
|
| 1112 |
-
"step": 180000
|
| 1113 |
-
},
|
| 1114 |
-
{
|
| 1115 |
-
"epoch": 2.12,
|
| 1116 |
-
"learning_rate": 2.4471741852423237e-06,
|
| 1117 |
-
"loss": 0.7745,
|
| 1118 |
-
"step": 181000
|
| 1119 |
-
},
|
| 1120 |
-
{
|
| 1121 |
-
"epoch": 2.12,
|
| 1122 |
-
"learning_rate": 2.1982109232821178e-06,
|
| 1123 |
-
"loss": 0.7707,
|
| 1124 |
-
"step": 182000
|
| 1125 |
-
},
|
| 1126 |
-
{
|
| 1127 |
-
"epoch": 2.13,
|
| 1128 |
-
"learning_rate": 1.962316193157593e-06,
|
| 1129 |
-
"loss": 0.6257,
|
| 1130 |
-
"step": 183000
|
| 1131 |
-
},
|
| 1132 |
-
{
|
| 1133 |
-
"epoch": 2.13,
|
| 1134 |
-
"learning_rate": 1.7395544861325718e-06,
|
| 1135 |
-
"loss": 0.5028,
|
| 1136 |
-
"step": 184000
|
| 1137 |
-
},
|
| 1138 |
-
{
|
| 1139 |
-
"epoch": 2.14,
|
| 1140 |
-
"learning_rate": 1.5299867030334814e-06,
|
| 1141 |
-
"loss": 0.7479,
|
| 1142 |
-
"step": 185000
|
| 1143 |
-
},
|
| 1144 |
-
{
|
| 1145 |
-
"epoch": 2.14,
|
| 1146 |
-
"learning_rate": 1.333670137599713e-06,
|
| 1147 |
-
"loss": 0.7893,
|
| 1148 |
-
"step": 186000
|
| 1149 |
-
},
|
| 1150 |
-
{
|
| 1151 |
-
"epoch": 2.15,
|
| 1152 |
-
"learning_rate": 1.1506584608200367e-06,
|
| 1153 |
-
"loss": 0.681,
|
| 1154 |
-
"step": 187000
|
| 1155 |
-
},
|
| 1156 |
-
{
|
| 1157 |
-
"epoch": 2.15,
|
| 1158 |
-
"learning_rate": 9.810017062595322e-07,
|
| 1159 |
-
"loss": 0.7565,
|
| 1160 |
-
"step": 188000
|
| 1161 |
-
},
|
| 1162 |
-
{
|
| 1163 |
-
"epoch": 2.16,
|
| 1164 |
-
"learning_rate": 8.247462563808817e-07,
|
| 1165 |
-
"loss": 0.5622,
|
| 1166 |
-
"step": 189000
|
| 1167 |
-
},
|
| 1168 |
-
{
|
| 1169 |
-
"epoch": 2.16,
|
| 1170 |
-
"learning_rate": 6.819348298638839e-07,
|
| 1171 |
-
"loss": 0.5524,
|
| 1172 |
-
"step": 190000
|
| 1173 |
-
},
|
| 1174 |
-
{
|
| 1175 |
-
"epoch": 2.17,
|
| 1176 |
-
"learning_rate": 5.526064699265753e-07,
|
| 1177 |
-
"loss": 0.8182,
|
| 1178 |
-
"step": 191000
|
| 1179 |
-
},
|
| 1180 |
-
{
|
| 1181 |
-
"epoch": 2.17,
|
| 1182 |
-
"learning_rate": 4.367965336512403e-07,
|
| 1183 |
-
"loss": 0.7816,
|
| 1184 |
-
"step": 192000
|
| 1185 |
-
},
|
| 1186 |
-
{
|
| 1187 |
-
"epoch": 2.18,
|
| 1188 |
-
"learning_rate": 3.3453668231809286e-07,
|
| 1189 |
-
"loss": 0.635,
|
| 1190 |
-
"step": 193000
|
| 1191 |
-
},
|
| 1192 |
-
{
|
| 1193 |
-
"epoch": 2.18,
|
| 1194 |
-
"learning_rate": 2.458548727494292e-07,
|
| 1195 |
-
"loss": 0.7174,
|
| 1196 |
-
"step": 194000
|
| 1197 |
-
},
|
| 1198 |
-
{
|
| 1199 |
-
"epoch": 2.19,
|
| 1200 |
-
"learning_rate": 1.7077534966650766e-07,
|
| 1201 |
-
"loss": 0.5278,
|
| 1202 |
-
"step": 195000
|
| 1203 |
-
},
|
| 1204 |
-
{
|
| 1205 |
-
"epoch": 2.19,
|
| 1206 |
-
"learning_rate": 1.0931863906127327e-07,
|
| 1207 |
-
"loss": 0.6336,
|
| 1208 |
-
"step": 196000
|
| 1209 |
-
},
|
| 1210 |
-
{
|
| 1211 |
-
"epoch": 2.2,
|
| 1212 |
-
"learning_rate": 6.150154258476315e-08,
|
| 1213 |
-
"loss": 0.9099,
|
| 1214 |
-
"step": 197000
|
| 1215 |
-
},
|
| 1216 |
-
{
|
| 1217 |
-
"epoch": 2.2,
|
| 1218 |
-
"learning_rate": 2.7337132953697554e-08,
|
| 1219 |
-
"loss": 0.6854,
|
| 1220 |
-
"step": 198000
|
| 1221 |
-
},
|
| 1222 |
-
{
|
| 1223 |
-
"epoch": 2.21,
|
| 1224 |
-
"learning_rate": 6.834750376549792e-09,
|
| 1225 |
-
"loss": 0.6256,
|
| 1226 |
-
"step": 199000
|
| 1227 |
-
},
|
| 1228 |
-
{
|
| 1229 |
-
"epoch": 2.21,
|
| 1230 |
-
"learning_rate": 0.0,
|
| 1231 |
-
"loss": 0.6506,
|
| 1232 |
-
"step": 200000
|
| 1233 |
-
},
|
| 1234 |
-
{
|
| 1235 |
-
"epoch": 2.21,
|
| 1236 |
-
"eval_loss": 0.6097196936607361,
|
| 1237 |
-
"eval_runtime": 10.7865,
|
| 1238 |
-
"eval_samples_per_second": 463.544,
|
| 1239 |
-
"eval_steps_per_second": 3.708,
|
| 1240 |
-
"step": 200000
|
| 1241 |
-
}
|
| 1242 |
-
],
|
| 1243 |
-
"max_steps": 200000,
|
| 1244 |
-
"num_train_epochs": 9223372036854775807,
|
| 1245 |
-
"total_flos": 3.369088143864103e+18,
|
| 1246 |
-
"trial_name": null,
|
| 1247 |
-
"trial_params": null
|
| 1248 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/training_args.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:b6c3a49ce7ceae30b6c2ffeb5a04bb56e7a988a399456fba3edbdc11896fdd34
|
| 3 |
-
size 3439
|
|
|
|
|
|
|
|
|
|
|
|
runs/Feb06_00-07-48_t1v-n-58269bec-w-0/events.out.tfevents.1675642086.t1v-n-58269bec-w-0.23209.0
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 sha256:68d365763a0cc79b68348c8eb4e2ff2d490487e84d8ab10d605d69597549beb4
|
| 3 |
+
size 37186
|