"auto-commit"
Browse files- model-bin/finetune/base/{checkpoint-23890 → checkpoint-25631}/config.json +0 -0
- model-bin/finetune/base/{checkpoint-23890 → checkpoint-25631}/optimizer.pt +1 -1
- model-bin/finetune/base/{checkpoint-23890 → checkpoint-25631}/preprocessor_config.json +0 -0
- model-bin/finetune/base/{checkpoint-25132 → checkpoint-25631}/pytorch_model.bin +1 -1
- model-bin/finetune/base/{checkpoint-25132 → checkpoint-25631}/rng_state.pth +1 -1
- model-bin/finetune/base/{checkpoint-25132 → checkpoint-25631}/scaler.pt +1 -1
- model-bin/finetune/base/{checkpoint-23890 → checkpoint-25631}/scheduler.pt +1 -1
- model-bin/finetune/base/{checkpoint-25132 → checkpoint-25631}/trainer_state.json +341 -5
- model-bin/finetune/base/{checkpoint-23890 → checkpoint-25631}/training_args.bin +0 -0
- model-bin/finetune/base/{checkpoint-25132 → checkpoint-25756}/config.json +0 -0
- model-bin/finetune/base/{checkpoint-25132 → checkpoint-25756}/optimizer.pt +1 -1
- model-bin/finetune/base/{checkpoint-25132 → checkpoint-25756}/preprocessor_config.json +0 -0
- model-bin/finetune/base/{checkpoint-23890 → checkpoint-25756}/pytorch_model.bin +1 -1
- model-bin/finetune/base/{checkpoint-23890 → checkpoint-25756}/rng_state.pth +1 -1
- model-bin/finetune/base/{checkpoint-23890 → checkpoint-25756}/scaler.pt +1 -1
- model-bin/finetune/base/{checkpoint-25132 → checkpoint-25756}/scheduler.pt +1 -1
- model-bin/finetune/base/{checkpoint-23890 → checkpoint-25756}/trainer_state.json +1256 -5
- model-bin/finetune/base/{checkpoint-25132 → checkpoint-25756}/training_args.bin +0 -0
- model-bin/finetune/base/log/1629604642.4566/events.out.tfevents.1629604642.91f414e9573a.903.41 +3 -0
- model-bin/finetune/base/log/1629605296.7616844/events.out.tfevents.1629605296.91f414e9573a.903.43 +3 -0
- model-bin/finetune/base/log/1629605964.3344083/events.out.tfevents.1629605964.91f414e9573a.903.45 +3 -0
- model-bin/finetune/base/log/1629606623.202551/events.out.tfevents.1629606623.91f414e9573a.903.47 +3 -0
- model-bin/finetune/base/log/1629607273.070973/events.out.tfevents.1629607273.91f414e9573a.903.49 +3 -0
- model-bin/finetune/base/log/events.out.tfevents.1629604642.91f414e9573a.903.40 +3 -0
- model-bin/finetune/base/log/events.out.tfevents.1629605296.91f414e9573a.903.42 +3 -0
- model-bin/finetune/base/log/events.out.tfevents.1629605964.91f414e9573a.903.44 +3 -0
- model-bin/finetune/base/log/events.out.tfevents.1629606623.91f414e9573a.903.46 +3 -0
- model-bin/finetune/base/log/events.out.tfevents.1629607273.91f414e9573a.903.48 +3 -0
model-bin/finetune/base/{checkpoint-23890 → checkpoint-25631}/config.json
RENAMED
|
File without changes
|
model-bin/finetune/base/{checkpoint-23890 → checkpoint-25631}/optimizer.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 722165009
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8b9b3dbb5713e6f0a38862c6dac9ec64357790f68d4322482159dce1dae58dfc
|
| 3 |
size 722165009
|
model-bin/finetune/base/{checkpoint-23890 → checkpoint-25631}/preprocessor_config.json
RENAMED
|
File without changes
|
model-bin/finetune/base/{checkpoint-25132 → checkpoint-25631}/pytorch_model.bin
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 377909911
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:74beccf1d6607529f568afc5331682ee4dd3d4a0648a55431828907842ba5d67
|
| 3 |
size 377909911
|
model-bin/finetune/base/{checkpoint-25132 → checkpoint-25631}/rng_state.pth
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14503
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:85711aac419c4249eba31cb0dc1fe1f442e282dad0958d3ca567a143bc38f533
|
| 3 |
size 14503
|
model-bin/finetune/base/{checkpoint-25132 → checkpoint-25631}/scaler.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 559
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3d62cf03d4f6da9a8a6f2f90c77691b01ca5326238eb58c8d7e43b316390bac9
|
| 3 |
size 559
|
model-bin/finetune/base/{checkpoint-23890 → checkpoint-25631}/scheduler.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 623
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5786a1e0f0ca5a608e9385515a2dc37cee01907e81e88b9a48f4bc15fa0e463e
|
| 3 |
size 623
|
model-bin/finetune/base/{checkpoint-25132 → checkpoint-25631}/trainer_state.json
RENAMED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "./model-bin/finetune/base/checkpoint-
|
| 4 |
-
"epoch":
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -139182,11 +139182,347 @@
|
|
| 139182 |
"eval_steps_per_second": 0.652,
|
| 139183 |
"eval_wer": 0.23441176470588235,
|
| 139184 |
"step": 25132
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 139185 |
}
|
| 139186 |
],
|
| 139187 |
"max_steps": 625000,
|
| 139188 |
"num_train_epochs": 5000,
|
| 139189 |
-
"total_flos": 7.
|
| 139190 |
"trial_name": null,
|
| 139191 |
"trial_params": null
|
| 139192 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.2158154859967051,
|
| 3 |
+
"best_model_checkpoint": "./model-bin/finetune/base/checkpoint-25631",
|
| 4 |
+
"epoch": 205.0,
|
| 5 |
+
"global_step": 25631,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 139182 |
"eval_steps_per_second": 0.652,
|
| 139183 |
"eval_wer": 0.23441176470588235,
|
| 139184 |
"step": 25132
|
| 139185 |
+
},
|
| 139186 |
+
{
|
| 139187 |
+
"epoch": 202.06,
|
| 139188 |
+
"learning_rate": 9.610177705977384e-06,
|
| 139189 |
+
"loss": 0.4814,
|
| 139190 |
+
"step": 25140
|
| 139191 |
+
},
|
| 139192 |
+
{
|
| 139193 |
+
"epoch": 202.15,
|
| 139194 |
+
"learning_rate": 9.610016155088854e-06,
|
| 139195 |
+
"loss": 0.5008,
|
| 139196 |
+
"step": 25150
|
| 139197 |
+
},
|
| 139198 |
+
{
|
| 139199 |
+
"epoch": 202.23,
|
| 139200 |
+
"learning_rate": 9.609854604200324e-06,
|
| 139201 |
+
"loss": 1.0201,
|
| 139202 |
+
"step": 25160
|
| 139203 |
+
},
|
| 139204 |
+
{
|
| 139205 |
+
"epoch": 202.31,
|
| 139206 |
+
"learning_rate": 9.609693053311794e-06,
|
| 139207 |
+
"loss": 0.4608,
|
| 139208 |
+
"step": 25170
|
| 139209 |
+
},
|
| 139210 |
+
{
|
| 139211 |
+
"epoch": 202.39,
|
| 139212 |
+
"learning_rate": 9.609531502423264e-06,
|
| 139213 |
+
"loss": 0.7121,
|
| 139214 |
+
"step": 25180
|
| 139215 |
+
},
|
| 139216 |
+
{
|
| 139217 |
+
"epoch": 202.47,
|
| 139218 |
+
"learning_rate": 9.609369951534734e-06,
|
| 139219 |
+
"loss": 0.7111,
|
| 139220 |
+
"step": 25190
|
| 139221 |
+
},
|
| 139222 |
+
{
|
| 139223 |
+
"epoch": 202.55,
|
| 139224 |
+
"learning_rate": 9.609208400646204e-06,
|
| 139225 |
+
"loss": 0.6784,
|
| 139226 |
+
"step": 25200
|
| 139227 |
+
},
|
| 139228 |
+
{
|
| 139229 |
+
"epoch": 202.63,
|
| 139230 |
+
"learning_rate": 9.609046849757674e-06,
|
| 139231 |
+
"loss": 0.9925,
|
| 139232 |
+
"step": 25210
|
| 139233 |
+
},
|
| 139234 |
+
{
|
| 139235 |
+
"epoch": 202.71,
|
| 139236 |
+
"learning_rate": 9.608885298869144e-06,
|
| 139237 |
+
"loss": 0.3937,
|
| 139238 |
+
"step": 25220
|
| 139239 |
+
},
|
| 139240 |
+
{
|
| 139241 |
+
"epoch": 202.79,
|
| 139242 |
+
"learning_rate": 9.608723747980616e-06,
|
| 139243 |
+
"loss": 0.7348,
|
| 139244 |
+
"step": 25230
|
| 139245 |
+
},
|
| 139246 |
+
{
|
| 139247 |
+
"epoch": 202.87,
|
| 139248 |
+
"learning_rate": 9.608562197092086e-06,
|
| 139249 |
+
"loss": 0.7775,
|
| 139250 |
+
"step": 25240
|
| 139251 |
+
},
|
| 139252 |
+
{
|
| 139253 |
+
"epoch": 202.95,
|
| 139254 |
+
"learning_rate": 9.608400646203556e-06,
|
| 139255 |
+
"loss": 0.55,
|
| 139256 |
+
"step": 25250
|
| 139257 |
+
},
|
| 139258 |
+
{
|
| 139259 |
+
"epoch": 203.0,
|
| 139260 |
+
"eval_loss": 0.4302833378314972,
|
| 139261 |
+
"eval_runtime": 40.5265,
|
| 139262 |
+
"eval_samples_per_second": 20.851,
|
| 139263 |
+
"eval_steps_per_second": 0.666,
|
| 139264 |
+
"eval_wer": 0.2268192488262911,
|
| 139265 |
+
"step": 25256
|
| 139266 |
+
},
|
| 139267 |
+
{
|
| 139268 |
+
"epoch": 202.03,
|
| 139269 |
+
"learning_rate": 9.608239095315026e-06,
|
| 139270 |
+
"loss": 0.4492,
|
| 139271 |
+
"step": 25260
|
| 139272 |
+
},
|
| 139273 |
+
{
|
| 139274 |
+
"epoch": 202.11,
|
| 139275 |
+
"learning_rate": 9.608077544426496e-06,
|
| 139276 |
+
"loss": 0.4215,
|
| 139277 |
+
"step": 25270
|
| 139278 |
+
},
|
| 139279 |
+
{
|
| 139280 |
+
"epoch": 202.19,
|
| 139281 |
+
"learning_rate": 9.607915993537966e-06,
|
| 139282 |
+
"loss": 0.845,
|
| 139283 |
+
"step": 25280
|
| 139284 |
+
},
|
| 139285 |
+
{
|
| 139286 |
+
"epoch": 202.27,
|
| 139287 |
+
"learning_rate": 9.607754442649436e-06,
|
| 139288 |
+
"loss": 0.6781,
|
| 139289 |
+
"step": 25290
|
| 139290 |
+
},
|
| 139291 |
+
{
|
| 139292 |
+
"epoch": 202.35,
|
| 139293 |
+
"learning_rate": 9.607592891760906e-06,
|
| 139294 |
+
"loss": 0.6022,
|
| 139295 |
+
"step": 25300
|
| 139296 |
+
},
|
| 139297 |
+
{
|
| 139298 |
+
"epoch": 202.43,
|
| 139299 |
+
"learning_rate": 9.607431340872376e-06,
|
| 139300 |
+
"loss": 0.9881,
|
| 139301 |
+
"step": 25310
|
| 139302 |
+
},
|
| 139303 |
+
{
|
| 139304 |
+
"epoch": 202.51,
|
| 139305 |
+
"learning_rate": 9.607269789983846e-06,
|
| 139306 |
+
"loss": 0.414,
|
| 139307 |
+
"step": 25320
|
| 139308 |
+
},
|
| 139309 |
+
{
|
| 139310 |
+
"epoch": 202.59,
|
| 139311 |
+
"learning_rate": 9.607108239095316e-06,
|
| 139312 |
+
"loss": 0.8022,
|
| 139313 |
+
"step": 25330
|
| 139314 |
+
},
|
| 139315 |
+
{
|
| 139316 |
+
"epoch": 202.67,
|
| 139317 |
+
"learning_rate": 9.606946688206786e-06,
|
| 139318 |
+
"loss": 0.7133,
|
| 139319 |
+
"step": 25340
|
| 139320 |
+
},
|
| 139321 |
+
{
|
| 139322 |
+
"epoch": 202.75,
|
| 139323 |
+
"learning_rate": 9.606785137318256e-06,
|
| 139324 |
+
"loss": 0.445,
|
| 139325 |
+
"step": 25350
|
| 139326 |
+
},
|
| 139327 |
+
{
|
| 139328 |
+
"epoch": 202.83,
|
| 139329 |
+
"learning_rate": 9.606623586429727e-06,
|
| 139330 |
+
"loss": 0.9602,
|
| 139331 |
+
"step": 25360
|
| 139332 |
+
},
|
| 139333 |
+
{
|
| 139334 |
+
"epoch": 202.91,
|
| 139335 |
+
"learning_rate": 9.606462035541197e-06,
|
| 139336 |
+
"loss": 0.4239,
|
| 139337 |
+
"step": 25370
|
| 139338 |
+
},
|
| 139339 |
+
{
|
| 139340 |
+
"epoch": 202.99,
|
| 139341 |
+
"learning_rate": 9.606300484652667e-06,
|
| 139342 |
+
"loss": 0.9062,
|
| 139343 |
+
"step": 25380
|
| 139344 |
+
},
|
| 139345 |
+
{
|
| 139346 |
+
"epoch": 203.0,
|
| 139347 |
+
"eval_loss": 0.6219351887702942,
|
| 139348 |
+
"eval_runtime": 40.3017,
|
| 139349 |
+
"eval_samples_per_second": 20.967,
|
| 139350 |
+
"eval_steps_per_second": 0.67,
|
| 139351 |
+
"eval_wer": 0.22838964458095656,
|
| 139352 |
+
"step": 25381
|
| 139353 |
+
},
|
| 139354 |
+
{
|
| 139355 |
+
"epoch": 203.07,
|
| 139356 |
+
"learning_rate": 9.606138933764137e-06,
|
| 139357 |
+
"loss": 0.386,
|
| 139358 |
+
"step": 25390
|
| 139359 |
+
},
|
| 139360 |
+
{
|
| 139361 |
+
"epoch": 203.15,
|
| 139362 |
+
"learning_rate": 9.605977382875607e-06,
|
| 139363 |
+
"loss": 0.487,
|
| 139364 |
+
"step": 25400
|
| 139365 |
+
},
|
| 139366 |
+
{
|
| 139367 |
+
"epoch": 203.23,
|
| 139368 |
+
"learning_rate": 9.605815831987077e-06,
|
| 139369 |
+
"loss": 1.1331,
|
| 139370 |
+
"step": 25410
|
| 139371 |
+
},
|
| 139372 |
+
{
|
| 139373 |
+
"epoch": 203.31,
|
| 139374 |
+
"learning_rate": 9.605654281098547e-06,
|
| 139375 |
+
"loss": 0.4191,
|
| 139376 |
+
"step": 25420
|
| 139377 |
+
},
|
| 139378 |
+
{
|
| 139379 |
+
"epoch": 203.39,
|
| 139380 |
+
"learning_rate": 9.605492730210017e-06,
|
| 139381 |
+
"loss": 0.8739,
|
| 139382 |
+
"step": 25430
|
| 139383 |
+
},
|
| 139384 |
+
{
|
| 139385 |
+
"epoch": 203.47,
|
| 139386 |
+
"learning_rate": 9.605331179321487e-06,
|
| 139387 |
+
"loss": 0.6426,
|
| 139388 |
+
"step": 25440
|
| 139389 |
+
},
|
| 139390 |
+
{
|
| 139391 |
+
"epoch": 203.55,
|
| 139392 |
+
"learning_rate": 9.605169628432957e-06,
|
| 139393 |
+
"loss": 0.4751,
|
| 139394 |
+
"step": 25450
|
| 139395 |
+
},
|
| 139396 |
+
{
|
| 139397 |
+
"epoch": 203.63,
|
| 139398 |
+
"learning_rate": 9.605008077544427e-06,
|
| 139399 |
+
"loss": 0.9483,
|
| 139400 |
+
"step": 25460
|
| 139401 |
+
},
|
| 139402 |
+
{
|
| 139403 |
+
"epoch": 203.71,
|
| 139404 |
+
"learning_rate": 9.604846526655897e-06,
|
| 139405 |
+
"loss": 0.4071,
|
| 139406 |
+
"step": 25470
|
| 139407 |
+
},
|
| 139408 |
+
{
|
| 139409 |
+
"epoch": 203.79,
|
| 139410 |
+
"learning_rate": 9.604684975767367e-06,
|
| 139411 |
+
"loss": 0.8267,
|
| 139412 |
+
"step": 25480
|
| 139413 |
+
},
|
| 139414 |
+
{
|
| 139415 |
+
"epoch": 203.87,
|
| 139416 |
+
"learning_rate": 9.604523424878837e-06,
|
| 139417 |
+
"loss": 0.6425,
|
| 139418 |
+
"step": 25490
|
| 139419 |
+
},
|
| 139420 |
+
{
|
| 139421 |
+
"epoch": 203.95,
|
| 139422 |
+
"learning_rate": 9.604361873990309e-06,
|
| 139423 |
+
"loss": 0.5668,
|
| 139424 |
+
"step": 25500
|
| 139425 |
+
},
|
| 139426 |
+
{
|
| 139427 |
+
"epoch": 204.0,
|
| 139428 |
+
"eval_loss": 0.4339040219783783,
|
| 139429 |
+
"eval_runtime": 41.2578,
|
| 139430 |
+
"eval_samples_per_second": 20.481,
|
| 139431 |
+
"eval_steps_per_second": 0.654,
|
| 139432 |
+
"eval_wer": 0.2250612790611305,
|
| 139433 |
+
"step": 25506
|
| 139434 |
+
},
|
| 139435 |
+
{
|
| 139436 |
+
"epoch": 204.03,
|
| 139437 |
+
"learning_rate": 9.604200323101779e-06,
|
| 139438 |
+
"loss": 0.5052,
|
| 139439 |
+
"step": 25510
|
| 139440 |
+
},
|
| 139441 |
+
{
|
| 139442 |
+
"epoch": 204.11,
|
| 139443 |
+
"learning_rate": 9.604038772213248e-06,
|
| 139444 |
+
"loss": 0.4648,
|
| 139445 |
+
"step": 25520
|
| 139446 |
+
},
|
| 139447 |
+
{
|
| 139448 |
+
"epoch": 204.19,
|
| 139449 |
+
"learning_rate": 9.603877221324718e-06,
|
| 139450 |
+
"loss": 0.8878,
|
| 139451 |
+
"step": 25530
|
| 139452 |
+
},
|
| 139453 |
+
{
|
| 139454 |
+
"epoch": 204.27,
|
| 139455 |
+
"learning_rate": 9.603715670436188e-06,
|
| 139456 |
+
"loss": 0.6632,
|
| 139457 |
+
"step": 25540
|
| 139458 |
+
},
|
| 139459 |
+
{
|
| 139460 |
+
"epoch": 204.35,
|
| 139461 |
+
"learning_rate": 9.603554119547658e-06,
|
| 139462 |
+
"loss": 0.4286,
|
| 139463 |
+
"step": 25550
|
| 139464 |
+
},
|
| 139465 |
+
{
|
| 139466 |
+
"epoch": 204.43,
|
| 139467 |
+
"learning_rate": 9.603392568659128e-06,
|
| 139468 |
+
"loss": 0.9665,
|
| 139469 |
+
"step": 25560
|
| 139470 |
+
},
|
| 139471 |
+
{
|
| 139472 |
+
"epoch": 204.51,
|
| 139473 |
+
"learning_rate": 9.603231017770598e-06,
|
| 139474 |
+
"loss": 0.442,
|
| 139475 |
+
"step": 25570
|
| 139476 |
+
},
|
| 139477 |
+
{
|
| 139478 |
+
"epoch": 204.59,
|
| 139479 |
+
"learning_rate": 9.603069466882068e-06,
|
| 139480 |
+
"loss": 0.9,
|
| 139481 |
+
"step": 25580
|
| 139482 |
+
},
|
| 139483 |
+
{
|
| 139484 |
+
"epoch": 204.67,
|
| 139485 |
+
"learning_rate": 9.602907915993538e-06,
|
| 139486 |
+
"loss": 0.668,
|
| 139487 |
+
"step": 25590
|
| 139488 |
+
},
|
| 139489 |
+
{
|
| 139490 |
+
"epoch": 204.75,
|
| 139491 |
+
"learning_rate": 9.602746365105008e-06,
|
| 139492 |
+
"loss": 0.4957,
|
| 139493 |
+
"step": 25600
|
| 139494 |
+
},
|
| 139495 |
+
{
|
| 139496 |
+
"epoch": 204.83,
|
| 139497 |
+
"learning_rate": 9.602584814216478e-06,
|
| 139498 |
+
"loss": 1.0674,
|
| 139499 |
+
"step": 25610
|
| 139500 |
+
},
|
| 139501 |
+
{
|
| 139502 |
+
"epoch": 204.91,
|
| 139503 |
+
"learning_rate": 9.602423263327948e-06,
|
| 139504 |
+
"loss": 0.5895,
|
| 139505 |
+
"step": 25620
|
| 139506 |
+
},
|
| 139507 |
+
{
|
| 139508 |
+
"epoch": 204.99,
|
| 139509 |
+
"learning_rate": 9.60226171243942e-06,
|
| 139510 |
+
"loss": 0.8323,
|
| 139511 |
+
"step": 25630
|
| 139512 |
+
},
|
| 139513 |
+
{
|
| 139514 |
+
"epoch": 205.0,
|
| 139515 |
+
"eval_loss": 0.4558946192264557,
|
| 139516 |
+
"eval_runtime": 40.4839,
|
| 139517 |
+
"eval_samples_per_second": 20.872,
|
| 139518 |
+
"eval_steps_per_second": 0.667,
|
| 139519 |
+
"eval_wer": 0.2158154859967051,
|
| 139520 |
+
"step": 25631
|
| 139521 |
}
|
| 139522 |
],
|
| 139523 |
"max_steps": 625000,
|
| 139524 |
"num_train_epochs": 5000,
|
| 139525 |
+
"total_flos": 7.211969668193911e+19,
|
| 139526 |
"trial_name": null,
|
| 139527 |
"trial_params": null
|
| 139528 |
}
|
model-bin/finetune/base/{checkpoint-23890 → checkpoint-25631}/training_args.bin
RENAMED
|
File without changes
|
model-bin/finetune/base/{checkpoint-25132 → checkpoint-25756}/config.json
RENAMED
|
File without changes
|
model-bin/finetune/base/{checkpoint-25132 → checkpoint-25756}/optimizer.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 722165009
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e5ca39c13040af41e73aea212f5895eaf64539da8f2992d477d80e588b2c90de
|
| 3 |
size 722165009
|
model-bin/finetune/base/{checkpoint-25132 → checkpoint-25756}/preprocessor_config.json
RENAMED
|
File without changes
|
model-bin/finetune/base/{checkpoint-23890 → checkpoint-25756}/pytorch_model.bin
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 377909911
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4ae1a583da9187750c0586e2515bf52c0f8c116d8daf944035c92dc58ede52b7
|
| 3 |
size 377909911
|
model-bin/finetune/base/{checkpoint-23890 → checkpoint-25756}/rng_state.pth
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14567
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8b216daa002f839d5bc2269415695eadb2a48adc0c575b728653a6b08cdb8102
|
| 3 |
size 14567
|
model-bin/finetune/base/{checkpoint-23890 → checkpoint-25756}/scaler.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 559
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5f9452be965d9bfbc9a3a50be7654b408bafe2c05c1b389e5079b6c83781bf2c
|
| 3 |
size 559
|
model-bin/finetune/base/{checkpoint-25132 → checkpoint-25756}/scheduler.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 623
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d5753ebfd40f550bd6f2a5d782321d2dd31a354da6244533710debe87314e292
|
| 3 |
size 623
|
model-bin/finetune/base/{checkpoint-23890 → checkpoint-25756}/trainer_state.json
RENAMED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "./model-bin/finetune/base/checkpoint-
|
| 4 |
-
"epoch":
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -138348,11 +138348,1262 @@
|
|
| 138348 |
"eval_steps_per_second": 0.643,
|
| 138349 |
"eval_wer": 0.2172567645365573,
|
| 138350 |
"step": 23890
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 138351 |
}
|
| 138352 |
],
|
| 138353 |
"max_steps": 625000,
|
| 138354 |
"num_train_epochs": 5000,
|
| 138355 |
-
"total_flos":
|
| 138356 |
"trial_name": null,
|
| 138357 |
"trial_params": null
|
| 138358 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.2158154859967051,
|
| 3 |
+
"best_model_checkpoint": "./model-bin/finetune/base/checkpoint-25631",
|
| 4 |
+
"epoch": 206.0,
|
| 5 |
+
"global_step": 25756,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 138348 |
"eval_steps_per_second": 0.643,
|
| 138349 |
"eval_wer": 0.2172567645365573,
|
| 138350 |
"step": 23890
|
| 138351 |
+
},
|
| 138352 |
+
{
|
| 138353 |
+
"epoch": 191.08,
|
| 138354 |
+
"learning_rate": 9.630193861066237e-06,
|
| 138355 |
+
"loss": 0.4907,
|
| 138356 |
+
"step": 23900
|
| 138357 |
+
},
|
| 138358 |
+
{
|
| 138359 |
+
"epoch": 191.16,
|
| 138360 |
+
"learning_rate": 9.630032310177707e-06,
|
| 138361 |
+
"loss": 0.4867,
|
| 138362 |
+
"step": 23910
|
| 138363 |
+
},
|
| 138364 |
+
{
|
| 138365 |
+
"epoch": 191.24,
|
| 138366 |
+
"learning_rate": 9.629870759289177e-06,
|
| 138367 |
+
"loss": 1.1891,
|
| 138368 |
+
"step": 23920
|
| 138369 |
+
},
|
| 138370 |
+
{
|
| 138371 |
+
"epoch": 191.32,
|
| 138372 |
+
"learning_rate": 9.629709208400647e-06,
|
| 138373 |
+
"loss": 0.3992,
|
| 138374 |
+
"step": 23930
|
| 138375 |
+
},
|
| 138376 |
+
{
|
| 138377 |
+
"epoch": 191.4,
|
| 138378 |
+
"learning_rate": 9.629547657512117e-06,
|
| 138379 |
+
"loss": 1.0267,
|
| 138380 |
+
"step": 23940
|
| 138381 |
+
},
|
| 138382 |
+
{
|
| 138383 |
+
"epoch": 191.48,
|
| 138384 |
+
"learning_rate": 9.629386106623587e-06,
|
| 138385 |
+
"loss": 0.4272,
|
| 138386 |
+
"step": 23950
|
| 138387 |
+
},
|
| 138388 |
+
{
|
| 138389 |
+
"epoch": 191.56,
|
| 138390 |
+
"learning_rate": 9.629224555735057e-06,
|
| 138391 |
+
"loss": 0.5949,
|
| 138392 |
+
"step": 23960
|
| 138393 |
+
},
|
| 138394 |
+
{
|
| 138395 |
+
"epoch": 191.64,
|
| 138396 |
+
"learning_rate": 9.629063004846527e-06,
|
| 138397 |
+
"loss": 0.9722,
|
| 138398 |
+
"step": 23970
|
| 138399 |
+
},
|
| 138400 |
+
{
|
| 138401 |
+
"epoch": 191.72,
|
| 138402 |
+
"learning_rate": 9.628901453957997e-06,
|
| 138403 |
+
"loss": 0.5117,
|
| 138404 |
+
"step": 23980
|
| 138405 |
+
},
|
| 138406 |
+
{
|
| 138407 |
+
"epoch": 191.8,
|
| 138408 |
+
"learning_rate": 9.628739903069467e-06,
|
| 138409 |
+
"loss": 1.0096,
|
| 138410 |
+
"step": 23990
|
| 138411 |
+
},
|
| 138412 |
+
{
|
| 138413 |
+
"epoch": 191.88,
|
| 138414 |
+
"learning_rate": 9.628578352180937e-06,
|
| 138415 |
+
"loss": 0.4353,
|
| 138416 |
+
"step": 24000
|
| 138417 |
+
},
|
| 138418 |
+
{
|
| 138419 |
+
"epoch": 191.96,
|
| 138420 |
+
"learning_rate": 9.628416801292409e-06,
|
| 138421 |
+
"loss": 0.5322,
|
| 138422 |
+
"step": 24010
|
| 138423 |
+
},
|
| 138424 |
+
{
|
| 138425 |
+
"epoch": 192.0,
|
| 138426 |
+
"eval_loss": 0.47713422775268555,
|
| 138427 |
+
"eval_runtime": 40.8067,
|
| 138428 |
+
"eval_samples_per_second": 20.487,
|
| 138429 |
+
"eval_steps_per_second": 0.662,
|
| 138430 |
+
"eval_wer": 0.23508771929824562,
|
| 138431 |
+
"step": 24015
|
| 138432 |
+
},
|
| 138433 |
+
{
|
| 138434 |
+
"epoch": 193.04,
|
| 138435 |
+
"learning_rate": 9.62827140549273e-06,
|
| 138436 |
+
"loss": 0.4565,
|
| 138437 |
+
"step": 24020
|
| 138438 |
+
},
|
| 138439 |
+
{
|
| 138440 |
+
"epoch": 193.12,
|
| 138441 |
+
"learning_rate": 9.6281098546042e-06,
|
| 138442 |
+
"loss": 0.4753,
|
| 138443 |
+
"step": 24030
|
| 138444 |
+
},
|
| 138445 |
+
{
|
| 138446 |
+
"epoch": 193.2,
|
| 138447 |
+
"learning_rate": 9.62794830371567e-06,
|
| 138448 |
+
"loss": 1.1457,
|
| 138449 |
+
"step": 24040
|
| 138450 |
+
},
|
| 138451 |
+
{
|
| 138452 |
+
"epoch": 193.28,
|
| 138453 |
+
"learning_rate": 9.627786752827142e-06,
|
| 138454 |
+
"loss": 0.4221,
|
| 138455 |
+
"step": 24050
|
| 138456 |
+
},
|
| 138457 |
+
{
|
| 138458 |
+
"epoch": 193.36,
|
| 138459 |
+
"learning_rate": 9.627625201938612e-06,
|
| 138460 |
+
"loss": 0.5413,
|
| 138461 |
+
"step": 24060
|
| 138462 |
+
},
|
| 138463 |
+
{
|
| 138464 |
+
"epoch": 193.44,
|
| 138465 |
+
"learning_rate": 9.627463651050082e-06,
|
| 138466 |
+
"loss": 0.9784,
|
| 138467 |
+
"step": 24070
|
| 138468 |
+
},
|
| 138469 |
+
{
|
| 138470 |
+
"epoch": 193.52,
|
| 138471 |
+
"learning_rate": 9.627302100161552e-06,
|
| 138472 |
+
"loss": 0.4663,
|
| 138473 |
+
"step": 24080
|
| 138474 |
+
},
|
| 138475 |
+
{
|
| 138476 |
+
"epoch": 193.6,
|
| 138477 |
+
"learning_rate": 9.627140549273022e-06,
|
| 138478 |
+
"loss": 1.0874,
|
| 138479 |
+
"step": 24090
|
| 138480 |
+
},
|
| 138481 |
+
{
|
| 138482 |
+
"epoch": 193.68,
|
| 138483 |
+
"learning_rate": 9.626978998384492e-06,
|
| 138484 |
+
"loss": 0.4995,
|
| 138485 |
+
"step": 24100
|
| 138486 |
+
},
|
| 138487 |
+
{
|
| 138488 |
+
"epoch": 193.76,
|
| 138489 |
+
"learning_rate": 9.626817447495962e-06,
|
| 138490 |
+
"loss": 0.4804,
|
| 138491 |
+
"step": 24110
|
| 138492 |
+
},
|
| 138493 |
+
{
|
| 138494 |
+
"epoch": 193.84,
|
| 138495 |
+
"learning_rate": 9.626655896607432e-06,
|
| 138496 |
+
"loss": 0.874,
|
| 138497 |
+
"step": 24120
|
| 138498 |
+
},
|
| 138499 |
+
{
|
| 138500 |
+
"epoch": 193.92,
|
| 138501 |
+
"learning_rate": 9.626494345718902e-06,
|
| 138502 |
+
"loss": 0.5074,
|
| 138503 |
+
"step": 24130
|
| 138504 |
+
},
|
| 138505 |
+
{
|
| 138506 |
+
"epoch": 194.0,
|
| 138507 |
+
"eval_loss": 0.5371707677841187,
|
| 138508 |
+
"eval_runtime": 40.5936,
|
| 138509 |
+
"eval_samples_per_second": 20.619,
|
| 138510 |
+
"eval_steps_per_second": 0.665,
|
| 138511 |
+
"eval_wer": 0.22862439747868002,
|
| 138512 |
+
"step": 24139
|
| 138513 |
+
},
|
| 138514 |
+
{
|
| 138515 |
+
"epoch": 194.01,
|
| 138516 |
+
"learning_rate": 9.626332794830372e-06,
|
| 138517 |
+
"loss": 0.5268,
|
| 138518 |
+
"step": 24140
|
| 138519 |
+
},
|
| 138520 |
+
{
|
| 138521 |
+
"epoch": 194.09,
|
| 138522 |
+
"learning_rate": 9.626171243941842e-06,
|
| 138523 |
+
"loss": 0.4241,
|
| 138524 |
+
"step": 24150
|
| 138525 |
+
},
|
| 138526 |
+
{
|
| 138527 |
+
"epoch": 194.17,
|
| 138528 |
+
"learning_rate": 9.626009693053313e-06,
|
| 138529 |
+
"loss": 0.6227,
|
| 138530 |
+
"step": 24160
|
| 138531 |
+
},
|
| 138532 |
+
{
|
| 138533 |
+
"epoch": 194.25,
|
| 138534 |
+
"learning_rate": 9.625848142164783e-06,
|
| 138535 |
+
"loss": 0.9652,
|
| 138536 |
+
"step": 24170
|
| 138537 |
+
},
|
| 138538 |
+
{
|
| 138539 |
+
"epoch": 194.33,
|
| 138540 |
+
"learning_rate": 9.625686591276253e-06,
|
| 138541 |
+
"loss": 0.4683,
|
| 138542 |
+
"step": 24180
|
| 138543 |
+
},
|
| 138544 |
+
{
|
| 138545 |
+
"epoch": 194.41,
|
| 138546 |
+
"learning_rate": 9.625525040387723e-06,
|
| 138547 |
+
"loss": 1.1767,
|
| 138548 |
+
"step": 24190
|
| 138549 |
+
},
|
| 138550 |
+
{
|
| 138551 |
+
"epoch": 194.49,
|
| 138552 |
+
"learning_rate": 9.625363489499193e-06,
|
| 138553 |
+
"loss": 0.4377,
|
| 138554 |
+
"step": 24200
|
| 138555 |
+
},
|
| 138556 |
+
{
|
| 138557 |
+
"epoch": 194.57,
|
| 138558 |
+
"learning_rate": 9.625201938610663e-06,
|
| 138559 |
+
"loss": 0.6596,
|
| 138560 |
+
"step": 24210
|
| 138561 |
+
},
|
| 138562 |
+
{
|
| 138563 |
+
"epoch": 194.65,
|
| 138564 |
+
"learning_rate": 9.625040387722133e-06,
|
| 138565 |
+
"loss": 0.9457,
|
| 138566 |
+
"step": 24220
|
| 138567 |
+
},
|
| 138568 |
+
{
|
| 138569 |
+
"epoch": 194.73,
|
| 138570 |
+
"learning_rate": 9.624878836833603e-06,
|
| 138571 |
+
"loss": 0.4555,
|
| 138572 |
+
"step": 24230
|
| 138573 |
+
},
|
| 138574 |
+
{
|
| 138575 |
+
"epoch": 194.81,
|
| 138576 |
+
"learning_rate": 9.624717285945073e-06,
|
| 138577 |
+
"loss": 0.9888,
|
| 138578 |
+
"step": 24240
|
| 138579 |
+
},
|
| 138580 |
+
{
|
| 138581 |
+
"epoch": 194.89,
|
| 138582 |
+
"learning_rate": 9.624555735056543e-06,
|
| 138583 |
+
"loss": 0.4755,
|
| 138584 |
+
"step": 24250
|
| 138585 |
+
},
|
| 138586 |
+
{
|
| 138587 |
+
"epoch": 194.97,
|
| 138588 |
+
"learning_rate": 9.624394184168013e-06,
|
| 138589 |
+
"loss": 0.5908,
|
| 138590 |
+
"step": 24260
|
| 138591 |
+
},
|
| 138592 |
+
{
|
| 138593 |
+
"epoch": 195.0,
|
| 138594 |
+
"eval_loss": 0.5001394748687744,
|
| 138595 |
+
"eval_runtime": 40.3014,
|
| 138596 |
+
"eval_samples_per_second": 20.769,
|
| 138597 |
+
"eval_steps_per_second": 0.67,
|
| 138598 |
+
"eval_wer": 0.22985141828005404,
|
| 138599 |
+
"step": 24263
|
| 138600 |
+
},
|
| 138601 |
+
{
|
| 138602 |
+
"epoch": 194.06,
|
| 138603 |
+
"learning_rate": 9.624232633279483e-06,
|
| 138604 |
+
"loss": 0.4893,
|
| 138605 |
+
"step": 24270
|
| 138606 |
+
},
|
| 138607 |
+
{
|
| 138608 |
+
"epoch": 194.14,
|
| 138609 |
+
"learning_rate": 9.624071082390955e-06,
|
| 138610 |
+
"loss": 0.4607,
|
| 138611 |
+
"step": 24280
|
| 138612 |
+
},
|
| 138613 |
+
{
|
| 138614 |
+
"epoch": 194.22,
|
| 138615 |
+
"learning_rate": 9.623909531502425e-06,
|
| 138616 |
+
"loss": 1.0318,
|
| 138617 |
+
"step": 24290
|
| 138618 |
+
},
|
| 138619 |
+
{
|
| 138620 |
+
"epoch": 194.3,
|
| 138621 |
+
"learning_rate": 9.623747980613895e-06,
|
| 138622 |
+
"loss": 0.4608,
|
| 138623 |
+
"step": 24300
|
| 138624 |
+
},
|
| 138625 |
+
{
|
| 138626 |
+
"epoch": 194.38,
|
| 138627 |
+
"learning_rate": 9.623586429725365e-06,
|
| 138628 |
+
"loss": 0.6567,
|
| 138629 |
+
"step": 24310
|
| 138630 |
+
},
|
| 138631 |
+
{
|
| 138632 |
+
"epoch": 194.46,
|
| 138633 |
+
"learning_rate": 9.623424878836835e-06,
|
| 138634 |
+
"loss": 0.8709,
|
| 138635 |
+
"step": 24320
|
| 138636 |
+
},
|
| 138637 |
+
{
|
| 138638 |
+
"epoch": 194.54,
|
| 138639 |
+
"learning_rate": 9.623263327948305e-06,
|
| 138640 |
+
"loss": 0.4206,
|
| 138641 |
+
"step": 24330
|
| 138642 |
+
},
|
| 138643 |
+
{
|
| 138644 |
+
"epoch": 194.62,
|
| 138645 |
+
"learning_rate": 9.623101777059774e-06,
|
| 138646 |
+
"loss": 0.9308,
|
| 138647 |
+
"step": 24340
|
| 138648 |
+
},
|
| 138649 |
+
{
|
| 138650 |
+
"epoch": 194.7,
|
| 138651 |
+
"learning_rate": 9.622940226171244e-06,
|
| 138652 |
+
"loss": 0.3828,
|
| 138653 |
+
"step": 24350
|
| 138654 |
+
},
|
| 138655 |
+
{
|
| 138656 |
+
"epoch": 194.78,
|
| 138657 |
+
"learning_rate": 9.622778675282714e-06,
|
| 138658 |
+
"loss": 0.7304,
|
| 138659 |
+
"step": 24360
|
| 138660 |
+
},
|
| 138661 |
+
{
|
| 138662 |
+
"epoch": 194.86,
|
| 138663 |
+
"learning_rate": 9.622617124394184e-06,
|
| 138664 |
+
"loss": 0.8561,
|
| 138665 |
+
"step": 24370
|
| 138666 |
+
},
|
| 138667 |
+
{
|
| 138668 |
+
"epoch": 194.94,
|
| 138669 |
+
"learning_rate": 9.622455573505654e-06,
|
| 138670 |
+
"loss": 0.5146,
|
| 138671 |
+
"step": 24380
|
| 138672 |
+
},
|
| 138673 |
+
{
|
| 138674 |
+
"epoch": 195.0,
|
| 138675 |
+
"eval_loss": 0.6156412959098816,
|
| 138676 |
+
"eval_runtime": 39.0293,
|
| 138677 |
+
"eval_samples_per_second": 21.65,
|
| 138678 |
+
"eval_steps_per_second": 0.692,
|
| 138679 |
+
"eval_wer": 0.23226561357320463,
|
| 138680 |
+
"step": 24388
|
| 138681 |
+
},
|
| 138682 |
+
{
|
| 138683 |
+
"epoch": 196.02,
|
| 138684 |
+
"learning_rate": 9.622294022617124e-06,
|
| 138685 |
+
"loss": 0.4826,
|
| 138686 |
+
"step": 24390
|
| 138687 |
+
},
|
| 138688 |
+
{
|
| 138689 |
+
"epoch": 196.1,
|
| 138690 |
+
"learning_rate": 9.622132471728596e-06,
|
| 138691 |
+
"loss": 0.4913,
|
| 138692 |
+
"step": 24400
|
| 138693 |
+
},
|
| 138694 |
+
{
|
| 138695 |
+
"epoch": 196.18,
|
| 138696 |
+
"learning_rate": 9.621970920840066e-06,
|
| 138697 |
+
"loss": 0.6569,
|
| 138698 |
+
"step": 24410
|
| 138699 |
+
},
|
| 138700 |
+
{
|
| 138701 |
+
"epoch": 196.26,
|
| 138702 |
+
"learning_rate": 9.621809369951536e-06,
|
| 138703 |
+
"loss": 0.9062,
|
| 138704 |
+
"step": 24420
|
| 138705 |
+
},
|
| 138706 |
+
{
|
| 138707 |
+
"epoch": 196.34,
|
| 138708 |
+
"learning_rate": 9.621647819063006e-06,
|
| 138709 |
+
"loss": 0.4574,
|
| 138710 |
+
"step": 24430
|
| 138711 |
+
},
|
| 138712 |
+
{
|
| 138713 |
+
"epoch": 196.42,
|
| 138714 |
+
"learning_rate": 9.621486268174476e-06,
|
| 138715 |
+
"loss": 0.9443,
|
| 138716 |
+
"step": 24440
|
| 138717 |
+
},
|
| 138718 |
+
{
|
| 138719 |
+
"epoch": 196.5,
|
| 138720 |
+
"learning_rate": 9.621324717285946e-06,
|
| 138721 |
+
"loss": 0.4282,
|
| 138722 |
+
"step": 24450
|
| 138723 |
+
},
|
| 138724 |
+
{
|
| 138725 |
+
"epoch": 196.58,
|
| 138726 |
+
"learning_rate": 9.621163166397416e-06,
|
| 138727 |
+
"loss": 0.692,
|
| 138728 |
+
"step": 24460
|
| 138729 |
+
},
|
| 138730 |
+
{
|
| 138731 |
+
"epoch": 196.66,
|
| 138732 |
+
"learning_rate": 9.621001615508886e-06,
|
| 138733 |
+
"loss": 0.9446,
|
| 138734 |
+
"step": 24470
|
| 138735 |
+
},
|
| 138736 |
+
{
|
| 138737 |
+
"epoch": 196.74,
|
| 138738 |
+
"learning_rate": 9.620840064620356e-06,
|
| 138739 |
+
"loss": 0.4262,
|
| 138740 |
+
"step": 24480
|
| 138741 |
+
},
|
| 138742 |
+
{
|
| 138743 |
+
"epoch": 196.82,
|
| 138744 |
+
"learning_rate": 9.620678513731826e-06,
|
| 138745 |
+
"loss": 1.078,
|
| 138746 |
+
"step": 24490
|
| 138747 |
+
},
|
| 138748 |
+
{
|
| 138749 |
+
"epoch": 196.9,
|
| 138750 |
+
"learning_rate": 9.620516962843296e-06,
|
| 138751 |
+
"loss": 0.4244,
|
| 138752 |
+
"step": 24500
|
| 138753 |
+
},
|
| 138754 |
+
{
|
| 138755 |
+
"epoch": 196.98,
|
| 138756 |
+
"learning_rate": 9.620355411954767e-06,
|
| 138757 |
+
"loss": 0.725,
|
| 138758 |
+
"step": 24510
|
| 138759 |
+
},
|
| 138760 |
+
{
|
| 138761 |
+
"epoch": 197.0,
|
| 138762 |
+
"eval_loss": 0.4438294768333435,
|
| 138763 |
+
"eval_runtime": 40.621,
|
| 138764 |
+
"eval_samples_per_second": 20.802,
|
| 138765 |
+
"eval_steps_per_second": 0.665,
|
| 138766 |
+
"eval_wer": 0.22446705786228924,
|
| 138767 |
+
"step": 24512
|
| 138768 |
+
},
|
| 138769 |
+
{
|
| 138770 |
+
"epoch": 197.06,
|
| 138771 |
+
"learning_rate": 9.620193861066237e-06,
|
| 138772 |
+
"loss": 0.4979,
|
| 138773 |
+
"step": 24520
|
| 138774 |
+
},
|
| 138775 |
+
{
|
| 138776 |
+
"epoch": 197.14,
|
| 138777 |
+
"learning_rate": 9.620032310177707e-06,
|
| 138778 |
+
"loss": 0.4723,
|
| 138779 |
+
"step": 24530
|
| 138780 |
+
},
|
| 138781 |
+
{
|
| 138782 |
+
"epoch": 197.22,
|
| 138783 |
+
"learning_rate": 9.619870759289177e-06,
|
| 138784 |
+
"loss": 1.0379,
|
| 138785 |
+
"step": 24540
|
| 138786 |
+
},
|
| 138787 |
+
{
|
| 138788 |
+
"epoch": 197.31,
|
| 138789 |
+
"learning_rate": 9.619709208400647e-06,
|
| 138790 |
+
"loss": 0.3889,
|
| 138791 |
+
"step": 24550
|
| 138792 |
+
},
|
| 138793 |
+
{
|
| 138794 |
+
"epoch": 197.39,
|
| 138795 |
+
"learning_rate": 9.619547657512117e-06,
|
| 138796 |
+
"loss": 0.7348,
|
| 138797 |
+
"step": 24560
|
| 138798 |
+
},
|
| 138799 |
+
{
|
| 138800 |
+
"epoch": 197.47,
|
| 138801 |
+
"learning_rate": 9.619386106623587e-06,
|
| 138802 |
+
"loss": 0.7806,
|
| 138803 |
+
"step": 24570
|
| 138804 |
+
},
|
| 138805 |
+
{
|
| 138806 |
+
"epoch": 197.55,
|
| 138807 |
+
"learning_rate": 9.619224555735057e-06,
|
| 138808 |
+
"loss": 0.477,
|
| 138809 |
+
"step": 24580
|
| 138810 |
+
},
|
| 138811 |
+
{
|
| 138812 |
+
"epoch": 197.63,
|
| 138813 |
+
"learning_rate": 9.619063004846527e-06,
|
| 138814 |
+
"loss": 0.9734,
|
| 138815 |
+
"step": 24590
|
| 138816 |
+
},
|
| 138817 |
+
{
|
| 138818 |
+
"epoch": 197.71,
|
| 138819 |
+
"learning_rate": 9.618901453957997e-06,
|
| 138820 |
+
"loss": 0.4244,
|
| 138821 |
+
"step": 24600
|
| 138822 |
+
},
|
| 138823 |
+
{
|
| 138824 |
+
"epoch": 197.79,
|
| 138825 |
+
"learning_rate": 9.618739903069467e-06,
|
| 138826 |
+
"loss": 0.7103,
|
| 138827 |
+
"step": 24610
|
| 138828 |
+
},
|
| 138829 |
+
{
|
| 138830 |
+
"epoch": 197.87,
|
| 138831 |
+
"learning_rate": 9.618578352180937e-06,
|
| 138832 |
+
"loss": 0.7796,
|
| 138833 |
+
"step": 24620
|
| 138834 |
+
},
|
| 138835 |
+
{
|
| 138836 |
+
"epoch": 197.95,
|
| 138837 |
+
"learning_rate": 9.618416801292409e-06,
|
| 138838 |
+
"loss": 0.5144,
|
| 138839 |
+
"step": 24630
|
| 138840 |
+
},
|
| 138841 |
+
{
|
| 138842 |
+
"epoch": 198.0,
|
| 138843 |
+
"eval_loss": 0.46397992968559265,
|
| 138844 |
+
"eval_runtime": 40.8749,
|
| 138845 |
+
"eval_samples_per_second": 20.673,
|
| 138846 |
+
"eval_steps_per_second": 0.661,
|
| 138847 |
+
"eval_wer": 0.21918200701955448,
|
| 138848 |
+
"step": 24636
|
| 138849 |
+
},
|
| 138850 |
+
{
|
| 138851 |
+
"epoch": 198.03,
|
| 138852 |
+
"learning_rate": 9.618255250403879e-06,
|
| 138853 |
+
"loss": 0.4148,
|
| 138854 |
+
"step": 24640
|
| 138855 |
+
},
|
| 138856 |
+
{
|
| 138857 |
+
"epoch": 198.11,
|
| 138858 |
+
"learning_rate": 9.618093699515349e-06,
|
| 138859 |
+
"loss": 0.4704,
|
| 138860 |
+
"step": 24650
|
| 138861 |
+
},
|
| 138862 |
+
{
|
| 138863 |
+
"epoch": 198.19,
|
| 138864 |
+
"learning_rate": 9.617932148626819e-06,
|
| 138865 |
+
"loss": 0.9196,
|
| 138866 |
+
"step": 24660
|
| 138867 |
+
},
|
| 138868 |
+
{
|
| 138869 |
+
"epoch": 198.27,
|
| 138870 |
+
"learning_rate": 9.617770597738289e-06,
|
| 138871 |
+
"loss": 0.594,
|
| 138872 |
+
"step": 24670
|
| 138873 |
+
},
|
| 138874 |
+
{
|
| 138875 |
+
"epoch": 198.35,
|
| 138876 |
+
"learning_rate": 9.617609046849759e-06,
|
| 138877 |
+
"loss": 0.5186,
|
| 138878 |
+
"step": 24680
|
| 138879 |
+
},
|
| 138880 |
+
{
|
| 138881 |
+
"epoch": 198.43,
|
| 138882 |
+
"learning_rate": 9.617447495961229e-06,
|
| 138883 |
+
"loss": 1.0653,
|
| 138884 |
+
"step": 24690
|
| 138885 |
+
},
|
| 138886 |
+
{
|
| 138887 |
+
"epoch": 198.51,
|
| 138888 |
+
"learning_rate": 9.617285945072699e-06,
|
| 138889 |
+
"loss": 0.4178,
|
| 138890 |
+
"step": 24700
|
| 138891 |
+
},
|
| 138892 |
+
{
|
| 138893 |
+
"epoch": 198.59,
|
| 138894 |
+
"learning_rate": 9.617124394184169e-06,
|
| 138895 |
+
"loss": 0.9037,
|
| 138896 |
+
"step": 24710
|
| 138897 |
+
},
|
| 138898 |
+
{
|
| 138899 |
+
"epoch": 198.67,
|
| 138900 |
+
"learning_rate": 9.616962843295639e-06,
|
| 138901 |
+
"loss": 0.5858,
|
| 138902 |
+
"step": 24720
|
| 138903 |
+
},
|
| 138904 |
+
{
|
| 138905 |
+
"epoch": 198.76,
|
| 138906 |
+
"learning_rate": 9.616801292407109e-06,
|
| 138907 |
+
"loss": 0.5078,
|
| 138908 |
+
"step": 24730
|
| 138909 |
+
},
|
| 138910 |
+
{
|
| 138911 |
+
"epoch": 198.84,
|
| 138912 |
+
"learning_rate": 9.616639741518579e-06,
|
| 138913 |
+
"loss": 1.0187,
|
| 138914 |
+
"step": 24740
|
| 138915 |
+
},
|
| 138916 |
+
{
|
| 138917 |
+
"epoch": 198.92,
|
| 138918 |
+
"learning_rate": 9.61647819063005e-06,
|
| 138919 |
+
"loss": 0.4836,
|
| 138920 |
+
"step": 24750
|
| 138921 |
+
},
|
| 138922 |
+
{
|
| 138923 |
+
"epoch": 199.0,
|
| 138924 |
+
"learning_rate": 9.61631663974152e-06,
|
| 138925 |
+
"loss": 0.8794,
|
| 138926 |
+
"step": 24760
|
| 138927 |
+
},
|
| 138928 |
+
{
|
| 138929 |
+
"epoch": 199.0,
|
| 138930 |
+
"eval_loss": 0.5073041319847107,
|
| 138931 |
+
"eval_runtime": 40.7573,
|
| 138932 |
+
"eval_samples_per_second": 20.732,
|
| 138933 |
+
"eval_steps_per_second": 0.662,
|
| 138934 |
+
"eval_wer": 0.221430669800235,
|
| 138935 |
+
"step": 24760
|
| 138936 |
+
},
|
| 138937 |
+
{
|
| 138938 |
+
"epoch": 201.08,
|
| 138939 |
+
"learning_rate": 9.61615508885299e-06,
|
| 138940 |
+
"loss": 0.4919,
|
| 138941 |
+
"step": 24770
|
| 138942 |
+
},
|
| 138943 |
+
{
|
| 138944 |
+
"epoch": 201.16,
|
| 138945 |
+
"learning_rate": 9.61599353796446e-06,
|
| 138946 |
+
"loss": 0.5007,
|
| 138947 |
+
"step": 24780
|
| 138948 |
+
},
|
| 138949 |
+
{
|
| 138950 |
+
"epoch": 201.24,
|
| 138951 |
+
"learning_rate": 9.61583198707593e-06,
|
| 138952 |
+
"loss": 0.9708,
|
| 138953 |
+
"step": 24790
|
| 138954 |
+
},
|
| 138955 |
+
{
|
| 138956 |
+
"epoch": 201.32,
|
| 138957 |
+
"learning_rate": 9.6156704361874e-06,
|
| 138958 |
+
"loss": 0.4545,
|
| 138959 |
+
"step": 24800
|
| 138960 |
+
},
|
| 138961 |
+
{
|
| 138962 |
+
"epoch": 201.4,
|
| 138963 |
+
"learning_rate": 9.61550888529887e-06,
|
| 138964 |
+
"loss": 1.0798,
|
| 138965 |
+
"step": 24810
|
| 138966 |
+
},
|
| 138967 |
+
{
|
| 138968 |
+
"epoch": 201.49,
|
| 138969 |
+
"learning_rate": 9.61534733441034e-06,
|
| 138970 |
+
"loss": 0.461,
|
| 138971 |
+
"step": 24820
|
| 138972 |
+
},
|
| 138973 |
+
{
|
| 138974 |
+
"epoch": 201.57,
|
| 138975 |
+
"learning_rate": 9.61518578352181e-06,
|
| 138976 |
+
"loss": 0.5125,
|
| 138977 |
+
"step": 24830
|
| 138978 |
+
},
|
| 138979 |
+
{
|
| 138980 |
+
"epoch": 201.65,
|
| 138981 |
+
"learning_rate": 9.61502423263328e-06,
|
| 138982 |
+
"loss": 0.9715,
|
| 138983 |
+
"step": 24840
|
| 138984 |
+
},
|
| 138985 |
+
{
|
| 138986 |
+
"epoch": 201.73,
|
| 138987 |
+
"learning_rate": 9.61486268174475e-06,
|
| 138988 |
+
"loss": 0.4949,
|
| 138989 |
+
"step": 24850
|
| 138990 |
+
},
|
| 138991 |
+
{
|
| 138992 |
+
"epoch": 201.81,
|
| 138993 |
+
"learning_rate": 9.614701130856222e-06,
|
| 138994 |
+
"loss": 1.0068,
|
| 138995 |
+
"step": 24860
|
| 138996 |
+
},
|
| 138997 |
+
{
|
| 138998 |
+
"epoch": 201.89,
|
| 138999 |
+
"learning_rate": 9.614539579967692e-06,
|
| 139000 |
+
"loss": 0.4748,
|
| 139001 |
+
"step": 24870
|
| 139002 |
+
},
|
| 139003 |
+
{
|
| 139004 |
+
"epoch": 201.97,
|
| 139005 |
+
"learning_rate": 9.614378029079162e-06,
|
| 139006 |
+
"loss": 0.606,
|
| 139007 |
+
"step": 24880
|
| 139008 |
+
},
|
| 139009 |
+
{
|
| 139010 |
+
"epoch": 202.0,
|
| 139011 |
+
"eval_loss": 0.5188912749290466,
|
| 139012 |
+
"eval_runtime": 41.1958,
|
| 139013 |
+
"eval_samples_per_second": 20.512,
|
| 139014 |
+
"eval_steps_per_second": 0.655,
|
| 139015 |
+
"eval_wer": 0.23777341762953275,
|
| 139016 |
+
"step": 24883
|
| 139017 |
+
},
|
| 139018 |
+
{
|
| 139019 |
+
"epoch": 200.06,
|
| 139020 |
+
"learning_rate": 9.614216478190632e-06,
|
| 139021 |
+
"loss": 0.5252,
|
| 139022 |
+
"step": 24890
|
| 139023 |
+
},
|
| 139024 |
+
{
|
| 139025 |
+
"epoch": 200.14,
|
| 139026 |
+
"learning_rate": 9.614054927302102e-06,
|
| 139027 |
+
"loss": 0.564,
|
| 139028 |
+
"step": 24900
|
| 139029 |
+
},
|
| 139030 |
+
{
|
| 139031 |
+
"epoch": 200.22,
|
| 139032 |
+
"learning_rate": 9.613893376413572e-06,
|
| 139033 |
+
"loss": 0.9814,
|
| 139034 |
+
"step": 24910
|
| 139035 |
+
},
|
| 139036 |
+
{
|
| 139037 |
+
"epoch": 200.3,
|
| 139038 |
+
"learning_rate": 9.613731825525042e-06,
|
| 139039 |
+
"loss": 0.4039,
|
| 139040 |
+
"step": 24920
|
| 139041 |
+
},
|
| 139042 |
+
{
|
| 139043 |
+
"epoch": 200.38,
|
| 139044 |
+
"learning_rate": 9.613570274636511e-06,
|
| 139045 |
+
"loss": 0.5923,
|
| 139046 |
+
"step": 24930
|
| 139047 |
+
},
|
| 139048 |
+
{
|
| 139049 |
+
"epoch": 200.46,
|
| 139050 |
+
"learning_rate": 9.613408723747981e-06,
|
| 139051 |
+
"loss": 0.9005,
|
| 139052 |
+
"step": 24940
|
| 139053 |
+
},
|
| 139054 |
+
{
|
| 139055 |
+
"epoch": 200.54,
|
| 139056 |
+
"learning_rate": 9.613247172859451e-06,
|
| 139057 |
+
"loss": 0.4075,
|
| 139058 |
+
"step": 24950
|
| 139059 |
+
},
|
| 139060 |
+
{
|
| 139061 |
+
"epoch": 200.62,
|
| 139062 |
+
"learning_rate": 9.613085621970921e-06,
|
| 139063 |
+
"loss": 1.0365,
|
| 139064 |
+
"step": 24960
|
| 139065 |
+
},
|
| 139066 |
+
{
|
| 139067 |
+
"epoch": 200.7,
|
| 139068 |
+
"learning_rate": 9.612924071082391e-06,
|
| 139069 |
+
"loss": 0.4891,
|
| 139070 |
+
"step": 24970
|
| 139071 |
+
},
|
| 139072 |
+
{
|
| 139073 |
+
"epoch": 200.78,
|
| 139074 |
+
"learning_rate": 9.612762520193863e-06,
|
| 139075 |
+
"loss": 0.6625,
|
| 139076 |
+
"step": 24980
|
| 139077 |
+
},
|
| 139078 |
+
{
|
| 139079 |
+
"epoch": 200.86,
|
| 139080 |
+
"learning_rate": 9.612600969305333e-06,
|
| 139081 |
+
"loss": 0.8385,
|
| 139082 |
+
"step": 24990
|
| 139083 |
+
},
|
| 139084 |
+
{
|
| 139085 |
+
"epoch": 200.94,
|
| 139086 |
+
"learning_rate": 9.612439418416803e-06,
|
| 139087 |
+
"loss": 0.5504,
|
| 139088 |
+
"step": 25000
|
| 139089 |
+
},
|
| 139090 |
+
{
|
| 139091 |
+
"epoch": 201.0,
|
| 139092 |
+
"eval_loss": 0.5322402715682983,
|
| 139093 |
+
"eval_runtime": 39.5678,
|
| 139094 |
+
"eval_samples_per_second": 21.33,
|
| 139095 |
+
"eval_steps_per_second": 0.682,
|
| 139096 |
+
"eval_wer": 0.22988161246815525,
|
| 139097 |
+
"step": 25007
|
| 139098 |
+
},
|
| 139099 |
+
{
|
| 139100 |
+
"epoch": 200.02,
|
| 139101 |
+
"learning_rate": 9.612277867528273e-06,
|
| 139102 |
+
"loss": 0.4325,
|
| 139103 |
+
"step": 25010
|
| 139104 |
+
},
|
| 139105 |
+
{
|
| 139106 |
+
"epoch": 200.1,
|
| 139107 |
+
"learning_rate": 9.612116316639743e-06,
|
| 139108 |
+
"loss": 0.4277,
|
| 139109 |
+
"step": 25020
|
| 139110 |
+
},
|
| 139111 |
+
{
|
| 139112 |
+
"epoch": 200.18,
|
| 139113 |
+
"learning_rate": 9.611954765751213e-06,
|
| 139114 |
+
"loss": 0.6847,
|
| 139115 |
+
"step": 25030
|
| 139116 |
+
},
|
| 139117 |
+
{
|
| 139118 |
+
"epoch": 200.26,
|
| 139119 |
+
"learning_rate": 9.611793214862683e-06,
|
| 139120 |
+
"loss": 0.8183,
|
| 139121 |
+
"step": 25040
|
| 139122 |
+
},
|
| 139123 |
+
{
|
| 139124 |
+
"epoch": 200.34,
|
| 139125 |
+
"learning_rate": 9.611631663974153e-06,
|
| 139126 |
+
"loss": 0.4621,
|
| 139127 |
+
"step": 25050
|
| 139128 |
+
},
|
| 139129 |
+
{
|
| 139130 |
+
"epoch": 200.42,
|
| 139131 |
+
"learning_rate": 9.611470113085623e-06,
|
| 139132 |
+
"loss": 0.9832,
|
| 139133 |
+
"step": 25060
|
| 139134 |
+
},
|
| 139135 |
+
{
|
| 139136 |
+
"epoch": 200.5,
|
| 139137 |
+
"learning_rate": 9.611308562197093e-06,
|
| 139138 |
+
"loss": 0.4897,
|
| 139139 |
+
"step": 25070
|
| 139140 |
+
},
|
| 139141 |
+
{
|
| 139142 |
+
"epoch": 200.58,
|
| 139143 |
+
"learning_rate": 9.611147011308563e-06,
|
| 139144 |
+
"loss": 1.0011,
|
| 139145 |
+
"step": 25080
|
| 139146 |
+
},
|
| 139147 |
+
{
|
| 139148 |
+
"epoch": 200.66,
|
| 139149 |
+
"learning_rate": 9.610985460420033e-06,
|
| 139150 |
+
"loss": 0.7532,
|
| 139151 |
+
"step": 25090
|
| 139152 |
+
},
|
| 139153 |
+
{
|
| 139154 |
+
"epoch": 200.74,
|
| 139155 |
+
"learning_rate": 9.610823909531504e-06,
|
| 139156 |
+
"loss": 0.4999,
|
| 139157 |
+
"step": 25100
|
| 139158 |
+
},
|
| 139159 |
+
{
|
| 139160 |
+
"epoch": 200.82,
|
| 139161 |
+
"learning_rate": 9.610662358642974e-06,
|
| 139162 |
+
"loss": 1.0194,
|
| 139163 |
+
"step": 25110
|
| 139164 |
+
},
|
| 139165 |
+
{
|
| 139166 |
+
"epoch": 200.9,
|
| 139167 |
+
"learning_rate": 9.610500807754444e-06,
|
| 139168 |
+
"loss": 0.4002,
|
| 139169 |
+
"step": 25120
|
| 139170 |
+
},
|
| 139171 |
+
{
|
| 139172 |
+
"epoch": 200.98,
|
| 139173 |
+
"learning_rate": 9.610339256865914e-06,
|
| 139174 |
+
"loss": 0.7423,
|
| 139175 |
+
"step": 25130
|
| 139176 |
+
},
|
| 139177 |
+
{
|
| 139178 |
+
"epoch": 201.0,
|
| 139179 |
+
"eval_loss": 0.5457486510276794,
|
| 139180 |
+
"eval_runtime": 41.422,
|
| 139181 |
+
"eval_samples_per_second": 20.4,
|
| 139182 |
+
"eval_steps_per_second": 0.652,
|
| 139183 |
+
"eval_wer": 0.23441176470588235,
|
| 139184 |
+
"step": 25132
|
| 139185 |
+
},
|
| 139186 |
+
{
|
| 139187 |
+
"epoch": 202.06,
|
| 139188 |
+
"learning_rate": 9.610177705977384e-06,
|
| 139189 |
+
"loss": 0.4814,
|
| 139190 |
+
"step": 25140
|
| 139191 |
+
},
|
| 139192 |
+
{
|
| 139193 |
+
"epoch": 202.15,
|
| 139194 |
+
"learning_rate": 9.610016155088854e-06,
|
| 139195 |
+
"loss": 0.5008,
|
| 139196 |
+
"step": 25150
|
| 139197 |
+
},
|
| 139198 |
+
{
|
| 139199 |
+
"epoch": 202.23,
|
| 139200 |
+
"learning_rate": 9.609854604200324e-06,
|
| 139201 |
+
"loss": 1.0201,
|
| 139202 |
+
"step": 25160
|
| 139203 |
+
},
|
| 139204 |
+
{
|
| 139205 |
+
"epoch": 202.31,
|
| 139206 |
+
"learning_rate": 9.609693053311794e-06,
|
| 139207 |
+
"loss": 0.4608,
|
| 139208 |
+
"step": 25170
|
| 139209 |
+
},
|
| 139210 |
+
{
|
| 139211 |
+
"epoch": 202.39,
|
| 139212 |
+
"learning_rate": 9.609531502423264e-06,
|
| 139213 |
+
"loss": 0.7121,
|
| 139214 |
+
"step": 25180
|
| 139215 |
+
},
|
| 139216 |
+
{
|
| 139217 |
+
"epoch": 202.47,
|
| 139218 |
+
"learning_rate": 9.609369951534734e-06,
|
| 139219 |
+
"loss": 0.7111,
|
| 139220 |
+
"step": 25190
|
| 139221 |
+
},
|
| 139222 |
+
{
|
| 139223 |
+
"epoch": 202.55,
|
| 139224 |
+
"learning_rate": 9.609208400646204e-06,
|
| 139225 |
+
"loss": 0.6784,
|
| 139226 |
+
"step": 25200
|
| 139227 |
+
},
|
| 139228 |
+
{
|
| 139229 |
+
"epoch": 202.63,
|
| 139230 |
+
"learning_rate": 9.609046849757674e-06,
|
| 139231 |
+
"loss": 0.9925,
|
| 139232 |
+
"step": 25210
|
| 139233 |
+
},
|
| 139234 |
+
{
|
| 139235 |
+
"epoch": 202.71,
|
| 139236 |
+
"learning_rate": 9.608885298869144e-06,
|
| 139237 |
+
"loss": 0.3937,
|
| 139238 |
+
"step": 25220
|
| 139239 |
+
},
|
| 139240 |
+
{
|
| 139241 |
+
"epoch": 202.79,
|
| 139242 |
+
"learning_rate": 9.608723747980616e-06,
|
| 139243 |
+
"loss": 0.7348,
|
| 139244 |
+
"step": 25230
|
| 139245 |
+
},
|
| 139246 |
+
{
|
| 139247 |
+
"epoch": 202.87,
|
| 139248 |
+
"learning_rate": 9.608562197092086e-06,
|
| 139249 |
+
"loss": 0.7775,
|
| 139250 |
+
"step": 25240
|
| 139251 |
+
},
|
| 139252 |
+
{
|
| 139253 |
+
"epoch": 202.95,
|
| 139254 |
+
"learning_rate": 9.608400646203556e-06,
|
| 139255 |
+
"loss": 0.55,
|
| 139256 |
+
"step": 25250
|
| 139257 |
+
},
|
| 139258 |
+
{
|
| 139259 |
+
"epoch": 203.0,
|
| 139260 |
+
"eval_loss": 0.4302833378314972,
|
| 139261 |
+
"eval_runtime": 40.5265,
|
| 139262 |
+
"eval_samples_per_second": 20.851,
|
| 139263 |
+
"eval_steps_per_second": 0.666,
|
| 139264 |
+
"eval_wer": 0.2268192488262911,
|
| 139265 |
+
"step": 25256
|
| 139266 |
+
},
|
| 139267 |
+
{
|
| 139268 |
+
"epoch": 202.03,
|
| 139269 |
+
"learning_rate": 9.608239095315026e-06,
|
| 139270 |
+
"loss": 0.4492,
|
| 139271 |
+
"step": 25260
|
| 139272 |
+
},
|
| 139273 |
+
{
|
| 139274 |
+
"epoch": 202.11,
|
| 139275 |
+
"learning_rate": 9.608077544426496e-06,
|
| 139276 |
+
"loss": 0.4215,
|
| 139277 |
+
"step": 25270
|
| 139278 |
+
},
|
| 139279 |
+
{
|
| 139280 |
+
"epoch": 202.19,
|
| 139281 |
+
"learning_rate": 9.607915993537966e-06,
|
| 139282 |
+
"loss": 0.845,
|
| 139283 |
+
"step": 25280
|
| 139284 |
+
},
|
| 139285 |
+
{
|
| 139286 |
+
"epoch": 202.27,
|
| 139287 |
+
"learning_rate": 9.607754442649436e-06,
|
| 139288 |
+
"loss": 0.6781,
|
| 139289 |
+
"step": 25290
|
| 139290 |
+
},
|
| 139291 |
+
{
|
| 139292 |
+
"epoch": 202.35,
|
| 139293 |
+
"learning_rate": 9.607592891760906e-06,
|
| 139294 |
+
"loss": 0.6022,
|
| 139295 |
+
"step": 25300
|
| 139296 |
+
},
|
| 139297 |
+
{
|
| 139298 |
+
"epoch": 202.43,
|
| 139299 |
+
"learning_rate": 9.607431340872376e-06,
|
| 139300 |
+
"loss": 0.9881,
|
| 139301 |
+
"step": 25310
|
| 139302 |
+
},
|
| 139303 |
+
{
|
| 139304 |
+
"epoch": 202.51,
|
| 139305 |
+
"learning_rate": 9.607269789983846e-06,
|
| 139306 |
+
"loss": 0.414,
|
| 139307 |
+
"step": 25320
|
| 139308 |
+
},
|
| 139309 |
+
{
|
| 139310 |
+
"epoch": 202.59,
|
| 139311 |
+
"learning_rate": 9.607108239095316e-06,
|
| 139312 |
+
"loss": 0.8022,
|
| 139313 |
+
"step": 25330
|
| 139314 |
+
},
|
| 139315 |
+
{
|
| 139316 |
+
"epoch": 202.67,
|
| 139317 |
+
"learning_rate": 9.606946688206786e-06,
|
| 139318 |
+
"loss": 0.7133,
|
| 139319 |
+
"step": 25340
|
| 139320 |
+
},
|
| 139321 |
+
{
|
| 139322 |
+
"epoch": 202.75,
|
| 139323 |
+
"learning_rate": 9.606785137318256e-06,
|
| 139324 |
+
"loss": 0.445,
|
| 139325 |
+
"step": 25350
|
| 139326 |
+
},
|
| 139327 |
+
{
|
| 139328 |
+
"epoch": 202.83,
|
| 139329 |
+
"learning_rate": 9.606623586429727e-06,
|
| 139330 |
+
"loss": 0.9602,
|
| 139331 |
+
"step": 25360
|
| 139332 |
+
},
|
| 139333 |
+
{
|
| 139334 |
+
"epoch": 202.91,
|
| 139335 |
+
"learning_rate": 9.606462035541197e-06,
|
| 139336 |
+
"loss": 0.4239,
|
| 139337 |
+
"step": 25370
|
| 139338 |
+
},
|
| 139339 |
+
{
|
| 139340 |
+
"epoch": 202.99,
|
| 139341 |
+
"learning_rate": 9.606300484652667e-06,
|
| 139342 |
+
"loss": 0.9062,
|
| 139343 |
+
"step": 25380
|
| 139344 |
+
},
|
| 139345 |
+
{
|
| 139346 |
+
"epoch": 203.0,
|
| 139347 |
+
"eval_loss": 0.6219351887702942,
|
| 139348 |
+
"eval_runtime": 40.3017,
|
| 139349 |
+
"eval_samples_per_second": 20.967,
|
| 139350 |
+
"eval_steps_per_second": 0.67,
|
| 139351 |
+
"eval_wer": 0.22838964458095656,
|
| 139352 |
+
"step": 25381
|
| 139353 |
+
},
|
| 139354 |
+
{
|
| 139355 |
+
"epoch": 203.07,
|
| 139356 |
+
"learning_rate": 9.606138933764137e-06,
|
| 139357 |
+
"loss": 0.386,
|
| 139358 |
+
"step": 25390
|
| 139359 |
+
},
|
| 139360 |
+
{
|
| 139361 |
+
"epoch": 203.15,
|
| 139362 |
+
"learning_rate": 9.605977382875607e-06,
|
| 139363 |
+
"loss": 0.487,
|
| 139364 |
+
"step": 25400
|
| 139365 |
+
},
|
| 139366 |
+
{
|
| 139367 |
+
"epoch": 203.23,
|
| 139368 |
+
"learning_rate": 9.605815831987077e-06,
|
| 139369 |
+
"loss": 1.1331,
|
| 139370 |
+
"step": 25410
|
| 139371 |
+
},
|
| 139372 |
+
{
|
| 139373 |
+
"epoch": 203.31,
|
| 139374 |
+
"learning_rate": 9.605654281098547e-06,
|
| 139375 |
+
"loss": 0.4191,
|
| 139376 |
+
"step": 25420
|
| 139377 |
+
},
|
| 139378 |
+
{
|
| 139379 |
+
"epoch": 203.39,
|
| 139380 |
+
"learning_rate": 9.605492730210017e-06,
|
| 139381 |
+
"loss": 0.8739,
|
| 139382 |
+
"step": 25430
|
| 139383 |
+
},
|
| 139384 |
+
{
|
| 139385 |
+
"epoch": 203.47,
|
| 139386 |
+
"learning_rate": 9.605331179321487e-06,
|
| 139387 |
+
"loss": 0.6426,
|
| 139388 |
+
"step": 25440
|
| 139389 |
+
},
|
| 139390 |
+
{
|
| 139391 |
+
"epoch": 203.55,
|
| 139392 |
+
"learning_rate": 9.605169628432957e-06,
|
| 139393 |
+
"loss": 0.4751,
|
| 139394 |
+
"step": 25450
|
| 139395 |
+
},
|
| 139396 |
+
{
|
| 139397 |
+
"epoch": 203.63,
|
| 139398 |
+
"learning_rate": 9.605008077544427e-06,
|
| 139399 |
+
"loss": 0.9483,
|
| 139400 |
+
"step": 25460
|
| 139401 |
+
},
|
| 139402 |
+
{
|
| 139403 |
+
"epoch": 203.71,
|
| 139404 |
+
"learning_rate": 9.604846526655897e-06,
|
| 139405 |
+
"loss": 0.4071,
|
| 139406 |
+
"step": 25470
|
| 139407 |
+
},
|
| 139408 |
+
{
|
| 139409 |
+
"epoch": 203.79,
|
| 139410 |
+
"learning_rate": 9.604684975767367e-06,
|
| 139411 |
+
"loss": 0.8267,
|
| 139412 |
+
"step": 25480
|
| 139413 |
+
},
|
| 139414 |
+
{
|
| 139415 |
+
"epoch": 203.87,
|
| 139416 |
+
"learning_rate": 9.604523424878837e-06,
|
| 139417 |
+
"loss": 0.6425,
|
| 139418 |
+
"step": 25490
|
| 139419 |
+
},
|
| 139420 |
+
{
|
| 139421 |
+
"epoch": 203.95,
|
| 139422 |
+
"learning_rate": 9.604361873990309e-06,
|
| 139423 |
+
"loss": 0.5668,
|
| 139424 |
+
"step": 25500
|
| 139425 |
+
},
|
| 139426 |
+
{
|
| 139427 |
+
"epoch": 204.0,
|
| 139428 |
+
"eval_loss": 0.4339040219783783,
|
| 139429 |
+
"eval_runtime": 41.2578,
|
| 139430 |
+
"eval_samples_per_second": 20.481,
|
| 139431 |
+
"eval_steps_per_second": 0.654,
|
| 139432 |
+
"eval_wer": 0.2250612790611305,
|
| 139433 |
+
"step": 25506
|
| 139434 |
+
},
|
| 139435 |
+
{
|
| 139436 |
+
"epoch": 204.03,
|
| 139437 |
+
"learning_rate": 9.604200323101779e-06,
|
| 139438 |
+
"loss": 0.5052,
|
| 139439 |
+
"step": 25510
|
| 139440 |
+
},
|
| 139441 |
+
{
|
| 139442 |
+
"epoch": 204.11,
|
| 139443 |
+
"learning_rate": 9.604038772213248e-06,
|
| 139444 |
+
"loss": 0.4648,
|
| 139445 |
+
"step": 25520
|
| 139446 |
+
},
|
| 139447 |
+
{
|
| 139448 |
+
"epoch": 204.19,
|
| 139449 |
+
"learning_rate": 9.603877221324718e-06,
|
| 139450 |
+
"loss": 0.8878,
|
| 139451 |
+
"step": 25530
|
| 139452 |
+
},
|
| 139453 |
+
{
|
| 139454 |
+
"epoch": 204.27,
|
| 139455 |
+
"learning_rate": 9.603715670436188e-06,
|
| 139456 |
+
"loss": 0.6632,
|
| 139457 |
+
"step": 25540
|
| 139458 |
+
},
|
| 139459 |
+
{
|
| 139460 |
+
"epoch": 204.35,
|
| 139461 |
+
"learning_rate": 9.603554119547658e-06,
|
| 139462 |
+
"loss": 0.4286,
|
| 139463 |
+
"step": 25550
|
| 139464 |
+
},
|
| 139465 |
+
{
|
| 139466 |
+
"epoch": 204.43,
|
| 139467 |
+
"learning_rate": 9.603392568659128e-06,
|
| 139468 |
+
"loss": 0.9665,
|
| 139469 |
+
"step": 25560
|
| 139470 |
+
},
|
| 139471 |
+
{
|
| 139472 |
+
"epoch": 204.51,
|
| 139473 |
+
"learning_rate": 9.603231017770598e-06,
|
| 139474 |
+
"loss": 0.442,
|
| 139475 |
+
"step": 25570
|
| 139476 |
+
},
|
| 139477 |
+
{
|
| 139478 |
+
"epoch": 204.59,
|
| 139479 |
+
"learning_rate": 9.603069466882068e-06,
|
| 139480 |
+
"loss": 0.9,
|
| 139481 |
+
"step": 25580
|
| 139482 |
+
},
|
| 139483 |
+
{
|
| 139484 |
+
"epoch": 204.67,
|
| 139485 |
+
"learning_rate": 9.602907915993538e-06,
|
| 139486 |
+
"loss": 0.668,
|
| 139487 |
+
"step": 25590
|
| 139488 |
+
},
|
| 139489 |
+
{
|
| 139490 |
+
"epoch": 204.75,
|
| 139491 |
+
"learning_rate": 9.602746365105008e-06,
|
| 139492 |
+
"loss": 0.4957,
|
| 139493 |
+
"step": 25600
|
| 139494 |
+
},
|
| 139495 |
+
{
|
| 139496 |
+
"epoch": 204.83,
|
| 139497 |
+
"learning_rate": 9.602584814216478e-06,
|
| 139498 |
+
"loss": 1.0674,
|
| 139499 |
+
"step": 25610
|
| 139500 |
+
},
|
| 139501 |
+
{
|
| 139502 |
+
"epoch": 204.91,
|
| 139503 |
+
"learning_rate": 9.602423263327948e-06,
|
| 139504 |
+
"loss": 0.5895,
|
| 139505 |
+
"step": 25620
|
| 139506 |
+
},
|
| 139507 |
+
{
|
| 139508 |
+
"epoch": 204.99,
|
| 139509 |
+
"learning_rate": 9.60226171243942e-06,
|
| 139510 |
+
"loss": 0.8323,
|
| 139511 |
+
"step": 25630
|
| 139512 |
+
},
|
| 139513 |
+
{
|
| 139514 |
+
"epoch": 205.0,
|
| 139515 |
+
"eval_loss": 0.4558946192264557,
|
| 139516 |
+
"eval_runtime": 40.4839,
|
| 139517 |
+
"eval_samples_per_second": 20.872,
|
| 139518 |
+
"eval_steps_per_second": 0.667,
|
| 139519 |
+
"eval_wer": 0.2158154859967051,
|
| 139520 |
+
"step": 25631
|
| 139521 |
+
},
|
| 139522 |
+
{
|
| 139523 |
+
"epoch": 205.07,
|
| 139524 |
+
"learning_rate": 9.60210016155089e-06,
|
| 139525 |
+
"loss": 0.4653,
|
| 139526 |
+
"step": 25640
|
| 139527 |
+
},
|
| 139528 |
+
{
|
| 139529 |
+
"epoch": 205.15,
|
| 139530 |
+
"learning_rate": 9.60193861066236e-06,
|
| 139531 |
+
"loss": 0.4812,
|
| 139532 |
+
"step": 25650
|
| 139533 |
+
},
|
| 139534 |
+
{
|
| 139535 |
+
"epoch": 205.23,
|
| 139536 |
+
"learning_rate": 9.60177705977383e-06,
|
| 139537 |
+
"loss": 1.0351,
|
| 139538 |
+
"step": 25660
|
| 139539 |
+
},
|
| 139540 |
+
{
|
| 139541 |
+
"epoch": 205.31,
|
| 139542 |
+
"learning_rate": 9.6016155088853e-06,
|
| 139543 |
+
"loss": 0.4079,
|
| 139544 |
+
"step": 25670
|
| 139545 |
+
},
|
| 139546 |
+
{
|
| 139547 |
+
"epoch": 205.39,
|
| 139548 |
+
"learning_rate": 9.60145395799677e-06,
|
| 139549 |
+
"loss": 0.8295,
|
| 139550 |
+
"step": 25680
|
| 139551 |
+
},
|
| 139552 |
+
{
|
| 139553 |
+
"epoch": 205.47,
|
| 139554 |
+
"learning_rate": 9.60129240710824e-06,
|
| 139555 |
+
"loss": 0.5685,
|
| 139556 |
+
"step": 25690
|
| 139557 |
+
},
|
| 139558 |
+
{
|
| 139559 |
+
"epoch": 205.55,
|
| 139560 |
+
"learning_rate": 9.60113085621971e-06,
|
| 139561 |
+
"loss": 0.4887,
|
| 139562 |
+
"step": 25700
|
| 139563 |
+
},
|
| 139564 |
+
{
|
| 139565 |
+
"epoch": 205.63,
|
| 139566 |
+
"learning_rate": 9.60096930533118e-06,
|
| 139567 |
+
"loss": 0.9767,
|
| 139568 |
+
"step": 25710
|
| 139569 |
+
},
|
| 139570 |
+
{
|
| 139571 |
+
"epoch": 205.71,
|
| 139572 |
+
"learning_rate": 9.60080775444265e-06,
|
| 139573 |
+
"loss": 0.4089,
|
| 139574 |
+
"step": 25720
|
| 139575 |
+
},
|
| 139576 |
+
{
|
| 139577 |
+
"epoch": 205.79,
|
| 139578 |
+
"learning_rate": 9.60064620355412e-06,
|
| 139579 |
+
"loss": 0.8511,
|
| 139580 |
+
"step": 25730
|
| 139581 |
+
},
|
| 139582 |
+
{
|
| 139583 |
+
"epoch": 205.87,
|
| 139584 |
+
"learning_rate": 9.60048465266559e-06,
|
| 139585 |
+
"loss": 0.5963,
|
| 139586 |
+
"step": 25740
|
| 139587 |
+
},
|
| 139588 |
+
{
|
| 139589 |
+
"epoch": 205.95,
|
| 139590 |
+
"learning_rate": 9.60032310177706e-06,
|
| 139591 |
+
"loss": 0.5197,
|
| 139592 |
+
"step": 25750
|
| 139593 |
+
},
|
| 139594 |
+
{
|
| 139595 |
+
"epoch": 206.0,
|
| 139596 |
+
"eval_loss": 0.49196144938468933,
|
| 139597 |
+
"eval_runtime": 39.9888,
|
| 139598 |
+
"eval_samples_per_second": 21.131,
|
| 139599 |
+
"eval_steps_per_second": 0.675,
|
| 139600 |
+
"eval_wer": 0.2235605170387779,
|
| 139601 |
+
"step": 25756
|
| 139602 |
}
|
| 139603 |
],
|
| 139604 |
"max_steps": 625000,
|
| 139605 |
"num_train_epochs": 5000,
|
| 139606 |
+
"total_flos": 7.2468597998005535e+19,
|
| 139607 |
"trial_name": null,
|
| 139608 |
"trial_params": null
|
| 139609 |
}
|
model-bin/finetune/base/{checkpoint-25132 → checkpoint-25756}/training_args.bin
RENAMED
|
File without changes
|
model-bin/finetune/base/log/1629604642.4566/events.out.tfevents.1629604642.91f414e9573a.903.41
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:892c73bb87e32f00c0e990619d85616cc7d9b068480a568d73828bcf16c9ca47
|
| 3 |
+
size 4194
|
model-bin/finetune/base/log/1629605296.7616844/events.out.tfevents.1629605296.91f414e9573a.903.43
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2de73da195e0c14094a2f1722526ddb981001a578989430aaf10ef487cec0173
|
| 3 |
+
size 4194
|
model-bin/finetune/base/log/1629605964.3344083/events.out.tfevents.1629605964.91f414e9573a.903.45
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2370a890f483f11ce07c58cce4137f9b1e7b2553a16ed6ea12448587f2b64576
|
| 3 |
+
size 4194
|
model-bin/finetune/base/log/1629606623.202551/events.out.tfevents.1629606623.91f414e9573a.903.47
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:807fbed627f7e7b5eb22bf897145908ac871cae2b8c0d726d36ab190c804e680
|
| 3 |
+
size 4194
|
model-bin/finetune/base/log/1629607273.070973/events.out.tfevents.1629607273.91f414e9573a.903.49
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5a7a0406837854de3a29cde2ec9a572865ec3b95d5eeb4f044b340a2b3b3bf8c
|
| 3 |
+
size 4194
|
model-bin/finetune/base/log/events.out.tfevents.1629604642.91f414e9573a.903.40
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b35ffae2b01e3a178f9e3e78cb6ed57b13fa0b6a410acf22888283c60b110e07
|
| 3 |
+
size 6543
|
model-bin/finetune/base/log/events.out.tfevents.1629605296.91f414e9573a.903.42
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:196e089ea1876152ab62adce761730af3711bb738dd5b087169c9752b00be67a
|
| 3 |
+
size 6703
|
model-bin/finetune/base/log/events.out.tfevents.1629605964.91f414e9573a.903.44
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a91332f13d46fab96b3680c1b810c7553753db5f959d0cde3746dec624155476
|
| 3 |
+
size 6543
|
model-bin/finetune/base/log/events.out.tfevents.1629606623.91f414e9573a.903.46
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:49c8feb2c4c22bca25c0fdaea893e6377b461c6cd071b42afc8e8340c0ca7ce4
|
| 3 |
+
size 6703
|
model-bin/finetune/base/log/events.out.tfevents.1629607273.91f414e9573a.903.48
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e48963f23239f62f533a2c05ad52e7d03fdd4eab9ac17a43eb362501ef5bbe67
|
| 3 |
+
size 6543
|