Training in progress, epoch 12
Browse files- last-checkpoint/optimizer.pt +1 -1
- last-checkpoint/pytorch_model.bin +1 -1
- last-checkpoint/rng_state.pth +1 -1
- last-checkpoint/scaler.pt +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +846 -3
- pytorch_model.bin +1 -1
- runs/Feb22_09-35-52_ubuntu-2004/events.out.tfevents.1677026198.ubuntu-2004.1870487.0 +2 -2
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 236470789
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c1533d6bddf3c8d95f6ff4932752f745542fa6faede711ac4e5f78630fd44fdb
|
| 3 |
size 236470789
|
last-checkpoint/pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 118243218
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e8e789cbff592c7abcbb70a0a1080069b2a642bdd41392da72c8a88aa5f570ae
|
| 3 |
size 118243218
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15597
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1555aa793e45869a50c0ea8fa30e8bcbcb3c7466ee570768b03040a237ce44ce
|
| 3 |
size 15597
|
last-checkpoint/scaler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 557
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c2004c6163970935b37a492dabc7d20258e92b44d334d59d43006d4c3533be13
|
| 3 |
size 557
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 627
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c877a115c1cbc8e43174987b4b039efb0e758136a482d45ffc9deb2c3305746a
|
| 3 |
size 627
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -9273,11 +9273,854 @@
|
|
| 9273 |
"eval_samples_per_second": 414.54,
|
| 9274 |
"eval_steps_per_second": 25.909,
|
| 9275 |
"step": 764203
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9276 |
}
|
| 9277 |
],
|
| 9278 |
"max_steps": 972622,
|
| 9279 |
"num_train_epochs": 14,
|
| 9280 |
-
"total_flos":
|
| 9281 |
"trial_name": null,
|
| 9282 |
"trial_params": null
|
| 9283 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 12.0,
|
| 5 |
+
"global_step": 833676,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 9273 |
"eval_samples_per_second": 414.54,
|
| 9274 |
"eval_steps_per_second": 25.909,
|
| 9275 |
"step": 764203
|
| 9276 |
+
},
|
| 9277 |
+
{
|
| 9278 |
+
"epoch": 11.0,
|
| 9279 |
+
"learning_rate": 2.165064726683595e-05,
|
| 9280 |
+
"loss": 1.7248,
|
| 9281 |
+
"step": 764500
|
| 9282 |
+
},
|
| 9283 |
+
{
|
| 9284 |
+
"epoch": 11.01,
|
| 9285 |
+
"learning_rate": 2.1598824378566722e-05,
|
| 9286 |
+
"loss": 1.7201,
|
| 9287 |
+
"step": 765000
|
| 9288 |
+
},
|
| 9289 |
+
{
|
| 9290 |
+
"epoch": 11.02,
|
| 9291 |
+
"learning_rate": 2.1546897636813983e-05,
|
| 9292 |
+
"loss": 1.7237,
|
| 9293 |
+
"step": 765500
|
| 9294 |
+
},
|
| 9295 |
+
{
|
| 9296 |
+
"epoch": 11.03,
|
| 9297 |
+
"learning_rate": 2.1494970895061248e-05,
|
| 9298 |
+
"loss": 1.724,
|
| 9299 |
+
"step": 766000
|
| 9300 |
+
},
|
| 9301 |
+
{
|
| 9302 |
+
"epoch": 11.03,
|
| 9303 |
+
"learning_rate": 2.1443044153308513e-05,
|
| 9304 |
+
"loss": 1.7209,
|
| 9305 |
+
"step": 766500
|
| 9306 |
+
},
|
| 9307 |
+
{
|
| 9308 |
+
"epoch": 11.04,
|
| 9309 |
+
"learning_rate": 2.1391117411555778e-05,
|
| 9310 |
+
"loss": 1.7219,
|
| 9311 |
+
"step": 767000
|
| 9312 |
+
},
|
| 9313 |
+
{
|
| 9314 |
+
"epoch": 11.05,
|
| 9315 |
+
"learning_rate": 2.1339190669803043e-05,
|
| 9316 |
+
"loss": 1.7215,
|
| 9317 |
+
"step": 767500
|
| 9318 |
+
},
|
| 9319 |
+
{
|
| 9320 |
+
"epoch": 11.05,
|
| 9321 |
+
"learning_rate": 2.1287263928050308e-05,
|
| 9322 |
+
"loss": 1.7206,
|
| 9323 |
+
"step": 768000
|
| 9324 |
+
},
|
| 9325 |
+
{
|
| 9326 |
+
"epoch": 11.06,
|
| 9327 |
+
"learning_rate": 2.1235441039781077e-05,
|
| 9328 |
+
"loss": 1.7198,
|
| 9329 |
+
"step": 768500
|
| 9330 |
+
},
|
| 9331 |
+
{
|
| 9332 |
+
"epoch": 11.07,
|
| 9333 |
+
"learning_rate": 2.1183514298028342e-05,
|
| 9334 |
+
"loss": 1.7244,
|
| 9335 |
+
"step": 769000
|
| 9336 |
+
},
|
| 9337 |
+
{
|
| 9338 |
+
"epoch": 11.08,
|
| 9339 |
+
"learning_rate": 2.1131587556275607e-05,
|
| 9340 |
+
"loss": 1.7213,
|
| 9341 |
+
"step": 769500
|
| 9342 |
+
},
|
| 9343 |
+
{
|
| 9344 |
+
"epoch": 11.08,
|
| 9345 |
+
"learning_rate": 2.1079660814522872e-05,
|
| 9346 |
+
"loss": 1.7236,
|
| 9347 |
+
"step": 770000
|
| 9348 |
+
},
|
| 9349 |
+
{
|
| 9350 |
+
"epoch": 11.09,
|
| 9351 |
+
"learning_rate": 2.1027837926253645e-05,
|
| 9352 |
+
"loss": 1.721,
|
| 9353 |
+
"step": 770500
|
| 9354 |
+
},
|
| 9355 |
+
{
|
| 9356 |
+
"epoch": 11.1,
|
| 9357 |
+
"learning_rate": 2.0975911184500907e-05,
|
| 9358 |
+
"loss": 1.7223,
|
| 9359 |
+
"step": 771000
|
| 9360 |
+
},
|
| 9361 |
+
{
|
| 9362 |
+
"epoch": 11.11,
|
| 9363 |
+
"learning_rate": 2.092398444274817e-05,
|
| 9364 |
+
"loss": 1.7238,
|
| 9365 |
+
"step": 771500
|
| 9366 |
+
},
|
| 9367 |
+
{
|
| 9368 |
+
"epoch": 11.11,
|
| 9369 |
+
"learning_rate": 2.0872057700995436e-05,
|
| 9370 |
+
"loss": 1.7226,
|
| 9371 |
+
"step": 772000
|
| 9372 |
+
},
|
| 9373 |
+
{
|
| 9374 |
+
"epoch": 11.12,
|
| 9375 |
+
"learning_rate": 2.08201309592427e-05,
|
| 9376 |
+
"loss": 1.7211,
|
| 9377 |
+
"step": 772500
|
| 9378 |
+
},
|
| 9379 |
+
{
|
| 9380 |
+
"epoch": 11.13,
|
| 9381 |
+
"learning_rate": 2.0768204217489966e-05,
|
| 9382 |
+
"loss": 1.7223,
|
| 9383 |
+
"step": 773000
|
| 9384 |
+
},
|
| 9385 |
+
{
|
| 9386 |
+
"epoch": 11.13,
|
| 9387 |
+
"learning_rate": 2.0716381329220736e-05,
|
| 9388 |
+
"loss": 1.7184,
|
| 9389 |
+
"step": 773500
|
| 9390 |
+
},
|
| 9391 |
+
{
|
| 9392 |
+
"epoch": 11.14,
|
| 9393 |
+
"learning_rate": 2.0664454587468e-05,
|
| 9394 |
+
"loss": 1.723,
|
| 9395 |
+
"step": 774000
|
| 9396 |
+
},
|
| 9397 |
+
{
|
| 9398 |
+
"epoch": 11.15,
|
| 9399 |
+
"learning_rate": 2.0612527845715266e-05,
|
| 9400 |
+
"loss": 1.7148,
|
| 9401 |
+
"step": 774500
|
| 9402 |
+
},
|
| 9403 |
+
{
|
| 9404 |
+
"epoch": 11.16,
|
| 9405 |
+
"learning_rate": 2.056060110396253e-05,
|
| 9406 |
+
"loss": 1.7249,
|
| 9407 |
+
"step": 775000
|
| 9408 |
+
},
|
| 9409 |
+
{
|
| 9410 |
+
"epoch": 11.16,
|
| 9411 |
+
"learning_rate": 2.05087782156933e-05,
|
| 9412 |
+
"loss": 1.7156,
|
| 9413 |
+
"step": 775500
|
| 9414 |
+
},
|
| 9415 |
+
{
|
| 9416 |
+
"epoch": 11.17,
|
| 9417 |
+
"learning_rate": 2.0456851473940565e-05,
|
| 9418 |
+
"loss": 1.7209,
|
| 9419 |
+
"step": 776000
|
| 9420 |
+
},
|
| 9421 |
+
{
|
| 9422 |
+
"epoch": 11.18,
|
| 9423 |
+
"learning_rate": 2.040492473218783e-05,
|
| 9424 |
+
"loss": 1.7263,
|
| 9425 |
+
"step": 776500
|
| 9426 |
+
},
|
| 9427 |
+
{
|
| 9428 |
+
"epoch": 11.18,
|
| 9429 |
+
"learning_rate": 2.0352997990435095e-05,
|
| 9430 |
+
"loss": 1.7217,
|
| 9431 |
+
"step": 777000
|
| 9432 |
+
},
|
| 9433 |
+
{
|
| 9434 |
+
"epoch": 11.19,
|
| 9435 |
+
"learning_rate": 2.0301175102165865e-05,
|
| 9436 |
+
"loss": 1.7213,
|
| 9437 |
+
"step": 777500
|
| 9438 |
+
},
|
| 9439 |
+
{
|
| 9440 |
+
"epoch": 11.2,
|
| 9441 |
+
"learning_rate": 2.024924836041313e-05,
|
| 9442 |
+
"loss": 1.7238,
|
| 9443 |
+
"step": 778000
|
| 9444 |
+
},
|
| 9445 |
+
{
|
| 9446 |
+
"epoch": 11.21,
|
| 9447 |
+
"learning_rate": 2.0197321618660394e-05,
|
| 9448 |
+
"loss": 1.7207,
|
| 9449 |
+
"step": 778500
|
| 9450 |
+
},
|
| 9451 |
+
{
|
| 9452 |
+
"epoch": 11.21,
|
| 9453 |
+
"learning_rate": 2.014539487690766e-05,
|
| 9454 |
+
"loss": 1.7236,
|
| 9455 |
+
"step": 779000
|
| 9456 |
+
},
|
| 9457 |
+
{
|
| 9458 |
+
"epoch": 11.22,
|
| 9459 |
+
"learning_rate": 2.0093468135154924e-05,
|
| 9460 |
+
"loss": 1.7216,
|
| 9461 |
+
"step": 779500
|
| 9462 |
+
},
|
| 9463 |
+
{
|
| 9464 |
+
"epoch": 11.23,
|
| 9465 |
+
"learning_rate": 2.0041645246885697e-05,
|
| 9466 |
+
"loss": 1.7229,
|
| 9467 |
+
"step": 780000
|
| 9468 |
+
},
|
| 9469 |
+
{
|
| 9470 |
+
"epoch": 11.23,
|
| 9471 |
+
"learning_rate": 1.998971850513296e-05,
|
| 9472 |
+
"loss": 1.7196,
|
| 9473 |
+
"step": 780500
|
| 9474 |
+
},
|
| 9475 |
+
{
|
| 9476 |
+
"epoch": 11.24,
|
| 9477 |
+
"learning_rate": 1.9937791763380224e-05,
|
| 9478 |
+
"loss": 1.7199,
|
| 9479 |
+
"step": 781000
|
| 9480 |
+
},
|
| 9481 |
+
{
|
| 9482 |
+
"epoch": 11.25,
|
| 9483 |
+
"learning_rate": 1.988586502162749e-05,
|
| 9484 |
+
"loss": 1.7186,
|
| 9485 |
+
"step": 781500
|
| 9486 |
+
},
|
| 9487 |
+
{
|
| 9488 |
+
"epoch": 11.26,
|
| 9489 |
+
"learning_rate": 1.9833938279874753e-05,
|
| 9490 |
+
"loss": 1.7208,
|
| 9491 |
+
"step": 782000
|
| 9492 |
+
},
|
| 9493 |
+
{
|
| 9494 |
+
"epoch": 11.26,
|
| 9495 |
+
"learning_rate": 1.9782011538122018e-05,
|
| 9496 |
+
"loss": 1.7177,
|
| 9497 |
+
"step": 782500
|
| 9498 |
+
},
|
| 9499 |
+
{
|
| 9500 |
+
"epoch": 11.27,
|
| 9501 |
+
"learning_rate": 1.9730084796369283e-05,
|
| 9502 |
+
"loss": 1.7212,
|
| 9503 |
+
"step": 783000
|
| 9504 |
+
},
|
| 9505 |
+
{
|
| 9506 |
+
"epoch": 11.28,
|
| 9507 |
+
"learning_rate": 1.9678158054616548e-05,
|
| 9508 |
+
"loss": 1.7219,
|
| 9509 |
+
"step": 783500
|
| 9510 |
+
},
|
| 9511 |
+
{
|
| 9512 |
+
"epoch": 11.28,
|
| 9513 |
+
"learning_rate": 1.9626335166347318e-05,
|
| 9514 |
+
"loss": 1.7191,
|
| 9515 |
+
"step": 784000
|
| 9516 |
+
},
|
| 9517 |
+
{
|
| 9518 |
+
"epoch": 11.29,
|
| 9519 |
+
"learning_rate": 1.9574512278078087e-05,
|
| 9520 |
+
"loss": 1.7213,
|
| 9521 |
+
"step": 784500
|
| 9522 |
+
},
|
| 9523 |
+
{
|
| 9524 |
+
"epoch": 11.3,
|
| 9525 |
+
"learning_rate": 1.9522585536325352e-05,
|
| 9526 |
+
"loss": 1.7227,
|
| 9527 |
+
"step": 785000
|
| 9528 |
+
},
|
| 9529 |
+
{
|
| 9530 |
+
"epoch": 11.31,
|
| 9531 |
+
"learning_rate": 1.9470658794572617e-05,
|
| 9532 |
+
"loss": 1.7204,
|
| 9533 |
+
"step": 785500
|
| 9534 |
+
},
|
| 9535 |
+
{
|
| 9536 |
+
"epoch": 11.31,
|
| 9537 |
+
"learning_rate": 1.9418732052819882e-05,
|
| 9538 |
+
"loss": 1.7223,
|
| 9539 |
+
"step": 786000
|
| 9540 |
+
},
|
| 9541 |
+
{
|
| 9542 |
+
"epoch": 11.32,
|
| 9543 |
+
"learning_rate": 1.9366805311067147e-05,
|
| 9544 |
+
"loss": 1.7194,
|
| 9545 |
+
"step": 786500
|
| 9546 |
+
},
|
| 9547 |
+
{
|
| 9548 |
+
"epoch": 11.33,
|
| 9549 |
+
"learning_rate": 1.9314878569314412e-05,
|
| 9550 |
+
"loss": 1.7205,
|
| 9551 |
+
"step": 787000
|
| 9552 |
+
},
|
| 9553 |
+
{
|
| 9554 |
+
"epoch": 11.34,
|
| 9555 |
+
"learning_rate": 1.9262951827561677e-05,
|
| 9556 |
+
"loss": 1.7195,
|
| 9557 |
+
"step": 787500
|
| 9558 |
+
},
|
| 9559 |
+
{
|
| 9560 |
+
"epoch": 11.34,
|
| 9561 |
+
"learning_rate": 1.9211128939292446e-05,
|
| 9562 |
+
"loss": 1.7208,
|
| 9563 |
+
"step": 788000
|
| 9564 |
+
},
|
| 9565 |
+
{
|
| 9566 |
+
"epoch": 11.35,
|
| 9567 |
+
"learning_rate": 1.915920219753971e-05,
|
| 9568 |
+
"loss": 1.7161,
|
| 9569 |
+
"step": 788500
|
| 9570 |
+
},
|
| 9571 |
+
{
|
| 9572 |
+
"epoch": 11.36,
|
| 9573 |
+
"learning_rate": 1.9107275455786976e-05,
|
| 9574 |
+
"loss": 1.7181,
|
| 9575 |
+
"step": 789000
|
| 9576 |
+
},
|
| 9577 |
+
{
|
| 9578 |
+
"epoch": 11.36,
|
| 9579 |
+
"learning_rate": 1.905534871403424e-05,
|
| 9580 |
+
"loss": 1.7177,
|
| 9581 |
+
"step": 789500
|
| 9582 |
+
},
|
| 9583 |
+
{
|
| 9584 |
+
"epoch": 11.37,
|
| 9585 |
+
"learning_rate": 1.9003421972281506e-05,
|
| 9586 |
+
"loss": 1.7162,
|
| 9587 |
+
"step": 790000
|
| 9588 |
+
},
|
| 9589 |
+
{
|
| 9590 |
+
"epoch": 11.38,
|
| 9591 |
+
"learning_rate": 1.8951599084012276e-05,
|
| 9592 |
+
"loss": 1.717,
|
| 9593 |
+
"step": 790500
|
| 9594 |
+
},
|
| 9595 |
+
{
|
| 9596 |
+
"epoch": 11.39,
|
| 9597 |
+
"learning_rate": 1.889967234225954e-05,
|
| 9598 |
+
"loss": 1.7229,
|
| 9599 |
+
"step": 791000
|
| 9600 |
+
},
|
| 9601 |
+
{
|
| 9602 |
+
"epoch": 11.39,
|
| 9603 |
+
"learning_rate": 1.8847745600506805e-05,
|
| 9604 |
+
"loss": 1.7197,
|
| 9605 |
+
"step": 791500
|
| 9606 |
+
},
|
| 9607 |
+
{
|
| 9608 |
+
"epoch": 11.4,
|
| 9609 |
+
"learning_rate": 1.879581885875407e-05,
|
| 9610 |
+
"loss": 1.7194,
|
| 9611 |
+
"step": 792000
|
| 9612 |
+
},
|
| 9613 |
+
{
|
| 9614 |
+
"epoch": 11.41,
|
| 9615 |
+
"learning_rate": 1.8743892117001335e-05,
|
| 9616 |
+
"loss": 1.7226,
|
| 9617 |
+
"step": 792500
|
| 9618 |
+
},
|
| 9619 |
+
{
|
| 9620 |
+
"epoch": 11.41,
|
| 9621 |
+
"learning_rate": 1.8692069228732105e-05,
|
| 9622 |
+
"loss": 1.7229,
|
| 9623 |
+
"step": 793000
|
| 9624 |
+
},
|
| 9625 |
+
{
|
| 9626 |
+
"epoch": 11.42,
|
| 9627 |
+
"learning_rate": 1.864014248697937e-05,
|
| 9628 |
+
"loss": 1.7167,
|
| 9629 |
+
"step": 793500
|
| 9630 |
+
},
|
| 9631 |
+
{
|
| 9632 |
+
"epoch": 11.43,
|
| 9633 |
+
"learning_rate": 1.8588215745226638e-05,
|
| 9634 |
+
"loss": 1.7203,
|
| 9635 |
+
"step": 794000
|
| 9636 |
+
},
|
| 9637 |
+
{
|
| 9638 |
+
"epoch": 11.44,
|
| 9639 |
+
"learning_rate": 1.85362890034739e-05,
|
| 9640 |
+
"loss": 1.723,
|
| 9641 |
+
"step": 794500
|
| 9642 |
+
},
|
| 9643 |
+
{
|
| 9644 |
+
"epoch": 11.44,
|
| 9645 |
+
"learning_rate": 1.8484362261721164e-05,
|
| 9646 |
+
"loss": 1.7216,
|
| 9647 |
+
"step": 795000
|
| 9648 |
+
},
|
| 9649 |
+
{
|
| 9650 |
+
"epoch": 11.45,
|
| 9651 |
+
"learning_rate": 1.8432539373451934e-05,
|
| 9652 |
+
"loss": 1.7197,
|
| 9653 |
+
"step": 795500
|
| 9654 |
+
},
|
| 9655 |
+
{
|
| 9656 |
+
"epoch": 11.46,
|
| 9657 |
+
"learning_rate": 1.83806126316992e-05,
|
| 9658 |
+
"loss": 1.7171,
|
| 9659 |
+
"step": 796000
|
| 9660 |
+
},
|
| 9661 |
+
{
|
| 9662 |
+
"epoch": 11.46,
|
| 9663 |
+
"learning_rate": 1.8328685889946464e-05,
|
| 9664 |
+
"loss": 1.7186,
|
| 9665 |
+
"step": 796500
|
| 9666 |
+
},
|
| 9667 |
+
{
|
| 9668 |
+
"epoch": 11.47,
|
| 9669 |
+
"learning_rate": 1.827675914819373e-05,
|
| 9670 |
+
"loss": 1.7188,
|
| 9671 |
+
"step": 797000
|
| 9672 |
+
},
|
| 9673 |
+
{
|
| 9674 |
+
"epoch": 11.48,
|
| 9675 |
+
"learning_rate": 1.82249362599245e-05,
|
| 9676 |
+
"loss": 1.7182,
|
| 9677 |
+
"step": 797500
|
| 9678 |
+
},
|
| 9679 |
+
{
|
| 9680 |
+
"epoch": 11.49,
|
| 9681 |
+
"learning_rate": 1.8173009518171763e-05,
|
| 9682 |
+
"loss": 1.7204,
|
| 9683 |
+
"step": 798000
|
| 9684 |
+
},
|
| 9685 |
+
{
|
| 9686 |
+
"epoch": 11.49,
|
| 9687 |
+
"learning_rate": 1.8121082776419028e-05,
|
| 9688 |
+
"loss": 1.7228,
|
| 9689 |
+
"step": 798500
|
| 9690 |
+
},
|
| 9691 |
+
{
|
| 9692 |
+
"epoch": 11.5,
|
| 9693 |
+
"learning_rate": 1.8069156034666293e-05,
|
| 9694 |
+
"loss": 1.7191,
|
| 9695 |
+
"step": 799000
|
| 9696 |
+
},
|
| 9697 |
+
{
|
| 9698 |
+
"epoch": 11.51,
|
| 9699 |
+
"learning_rate": 1.8017229292913558e-05,
|
| 9700 |
+
"loss": 1.7188,
|
| 9701 |
+
"step": 799500
|
| 9702 |
+
},
|
| 9703 |
+
{
|
| 9704 |
+
"epoch": 11.52,
|
| 9705 |
+
"learning_rate": 1.7965406404644328e-05,
|
| 9706 |
+
"loss": 1.7207,
|
| 9707 |
+
"step": 800000
|
| 9708 |
+
},
|
| 9709 |
+
{
|
| 9710 |
+
"epoch": 11.52,
|
| 9711 |
+
"learning_rate": 1.7913479662891593e-05,
|
| 9712 |
+
"loss": 1.7174,
|
| 9713 |
+
"step": 800500
|
| 9714 |
+
},
|
| 9715 |
+
{
|
| 9716 |
+
"epoch": 11.53,
|
| 9717 |
+
"learning_rate": 1.7861552921138858e-05,
|
| 9718 |
+
"loss": 1.7187,
|
| 9719 |
+
"step": 801000
|
| 9720 |
+
},
|
| 9721 |
+
{
|
| 9722 |
+
"epoch": 11.54,
|
| 9723 |
+
"learning_rate": 1.7809626179386122e-05,
|
| 9724 |
+
"loss": 1.7177,
|
| 9725 |
+
"step": 801500
|
| 9726 |
+
},
|
| 9727 |
+
{
|
| 9728 |
+
"epoch": 11.54,
|
| 9729 |
+
"learning_rate": 1.7757699437633387e-05,
|
| 9730 |
+
"loss": 1.7169,
|
| 9731 |
+
"step": 802000
|
| 9732 |
+
},
|
| 9733 |
+
{
|
| 9734 |
+
"epoch": 11.55,
|
| 9735 |
+
"learning_rate": 1.770587654936416e-05,
|
| 9736 |
+
"loss": 1.7187,
|
| 9737 |
+
"step": 802500
|
| 9738 |
+
},
|
| 9739 |
+
{
|
| 9740 |
+
"epoch": 11.56,
|
| 9741 |
+
"learning_rate": 1.7653949807611422e-05,
|
| 9742 |
+
"loss": 1.7142,
|
| 9743 |
+
"step": 803000
|
| 9744 |
+
},
|
| 9745 |
+
{
|
| 9746 |
+
"epoch": 11.57,
|
| 9747 |
+
"learning_rate": 1.760202306585869e-05,
|
| 9748 |
+
"loss": 1.7166,
|
| 9749 |
+
"step": 803500
|
| 9750 |
+
},
|
| 9751 |
+
{
|
| 9752 |
+
"epoch": 11.57,
|
| 9753 |
+
"learning_rate": 1.755009632410595e-05,
|
| 9754 |
+
"loss": 1.7169,
|
| 9755 |
+
"step": 804000
|
| 9756 |
+
},
|
| 9757 |
+
{
|
| 9758 |
+
"epoch": 11.58,
|
| 9759 |
+
"learning_rate": 1.7498273435836725e-05,
|
| 9760 |
+
"loss": 1.7156,
|
| 9761 |
+
"step": 804500
|
| 9762 |
+
},
|
| 9763 |
+
{
|
| 9764 |
+
"epoch": 11.59,
|
| 9765 |
+
"learning_rate": 1.7446346694083986e-05,
|
| 9766 |
+
"loss": 1.7156,
|
| 9767 |
+
"step": 805000
|
| 9768 |
+
},
|
| 9769 |
+
{
|
| 9770 |
+
"epoch": 11.59,
|
| 9771 |
+
"learning_rate": 1.739441995233125e-05,
|
| 9772 |
+
"loss": 1.7159,
|
| 9773 |
+
"step": 805500
|
| 9774 |
+
},
|
| 9775 |
+
{
|
| 9776 |
+
"epoch": 11.6,
|
| 9777 |
+
"learning_rate": 1.7342493210578516e-05,
|
| 9778 |
+
"loss": 1.717,
|
| 9779 |
+
"step": 806000
|
| 9780 |
+
},
|
| 9781 |
+
{
|
| 9782 |
+
"epoch": 11.61,
|
| 9783 |
+
"learning_rate": 1.729056646882578e-05,
|
| 9784 |
+
"loss": 1.716,
|
| 9785 |
+
"step": 806500
|
| 9786 |
+
},
|
| 9787 |
+
{
|
| 9788 |
+
"epoch": 11.62,
|
| 9789 |
+
"learning_rate": 1.723874358055655e-05,
|
| 9790 |
+
"loss": 1.7217,
|
| 9791 |
+
"step": 807000
|
| 9792 |
+
},
|
| 9793 |
+
{
|
| 9794 |
+
"epoch": 11.62,
|
| 9795 |
+
"learning_rate": 1.7186816838803815e-05,
|
| 9796 |
+
"loss": 1.7168,
|
| 9797 |
+
"step": 807500
|
| 9798 |
+
},
|
| 9799 |
+
{
|
| 9800 |
+
"epoch": 11.63,
|
| 9801 |
+
"learning_rate": 1.713489009705108e-05,
|
| 9802 |
+
"loss": 1.7147,
|
| 9803 |
+
"step": 808000
|
| 9804 |
+
},
|
| 9805 |
+
{
|
| 9806 |
+
"epoch": 11.64,
|
| 9807 |
+
"learning_rate": 1.7082963355298345e-05,
|
| 9808 |
+
"loss": 1.7156,
|
| 9809 |
+
"step": 808500
|
| 9810 |
+
},
|
| 9811 |
+
{
|
| 9812 |
+
"epoch": 11.64,
|
| 9813 |
+
"learning_rate": 1.7031140467029115e-05,
|
| 9814 |
+
"loss": 1.7187,
|
| 9815 |
+
"step": 809000
|
| 9816 |
+
},
|
| 9817 |
+
{
|
| 9818 |
+
"epoch": 11.65,
|
| 9819 |
+
"learning_rate": 1.697921372527638e-05,
|
| 9820 |
+
"loss": 1.7168,
|
| 9821 |
+
"step": 809500
|
| 9822 |
+
},
|
| 9823 |
+
{
|
| 9824 |
+
"epoch": 11.66,
|
| 9825 |
+
"learning_rate": 1.6927286983523648e-05,
|
| 9826 |
+
"loss": 1.7184,
|
| 9827 |
+
"step": 810000
|
| 9828 |
+
},
|
| 9829 |
+
{
|
| 9830 |
+
"epoch": 11.67,
|
| 9831 |
+
"learning_rate": 1.687536024177091e-05,
|
| 9832 |
+
"loss": 1.7142,
|
| 9833 |
+
"step": 810500
|
| 9834 |
+
},
|
| 9835 |
+
{
|
| 9836 |
+
"epoch": 11.67,
|
| 9837 |
+
"learning_rate": 1.6823433500018178e-05,
|
| 9838 |
+
"loss": 1.7171,
|
| 9839 |
+
"step": 811000
|
| 9840 |
+
},
|
| 9841 |
+
{
|
| 9842 |
+
"epoch": 11.68,
|
| 9843 |
+
"learning_rate": 1.6771610611748944e-05,
|
| 9844 |
+
"loss": 1.7163,
|
| 9845 |
+
"step": 811500
|
| 9846 |
+
},
|
| 9847 |
+
{
|
| 9848 |
+
"epoch": 11.69,
|
| 9849 |
+
"learning_rate": 1.6719683869996212e-05,
|
| 9850 |
+
"loss": 1.7208,
|
| 9851 |
+
"step": 812000
|
| 9852 |
+
},
|
| 9853 |
+
{
|
| 9854 |
+
"epoch": 11.7,
|
| 9855 |
+
"learning_rate": 1.6667757128243474e-05,
|
| 9856 |
+
"loss": 1.714,
|
| 9857 |
+
"step": 812500
|
| 9858 |
+
},
|
| 9859 |
+
{
|
| 9860 |
+
"epoch": 11.7,
|
| 9861 |
+
"learning_rate": 1.661583038649074e-05,
|
| 9862 |
+
"loss": 1.7158,
|
| 9863 |
+
"step": 813000
|
| 9864 |
+
},
|
| 9865 |
+
{
|
| 9866 |
+
"epoch": 11.71,
|
| 9867 |
+
"learning_rate": 1.656400749822151e-05,
|
| 9868 |
+
"loss": 1.7167,
|
| 9869 |
+
"step": 813500
|
| 9870 |
+
},
|
| 9871 |
+
{
|
| 9872 |
+
"epoch": 11.72,
|
| 9873 |
+
"learning_rate": 1.6512080756468777e-05,
|
| 9874 |
+
"loss": 1.7153,
|
| 9875 |
+
"step": 814000
|
| 9876 |
+
},
|
| 9877 |
+
{
|
| 9878 |
+
"epoch": 11.72,
|
| 9879 |
+
"learning_rate": 1.6460154014716038e-05,
|
| 9880 |
+
"loss": 1.7164,
|
| 9881 |
+
"step": 814500
|
| 9882 |
+
},
|
| 9883 |
+
{
|
| 9884 |
+
"epoch": 11.73,
|
| 9885 |
+
"learning_rate": 1.6408227272963303e-05,
|
| 9886 |
+
"loss": 1.7186,
|
| 9887 |
+
"step": 815000
|
| 9888 |
+
},
|
| 9889 |
+
{
|
| 9890 |
+
"epoch": 11.74,
|
| 9891 |
+
"learning_rate": 1.6356404384694073e-05,
|
| 9892 |
+
"loss": 1.7179,
|
| 9893 |
+
"step": 815500
|
| 9894 |
+
},
|
| 9895 |
+
{
|
| 9896 |
+
"epoch": 11.75,
|
| 9897 |
+
"learning_rate": 1.630447764294134e-05,
|
| 9898 |
+
"loss": 1.7159,
|
| 9899 |
+
"step": 816000
|
| 9900 |
+
},
|
| 9901 |
+
{
|
| 9902 |
+
"epoch": 11.75,
|
| 9903 |
+
"learning_rate": 1.6252550901188603e-05,
|
| 9904 |
+
"loss": 1.7159,
|
| 9905 |
+
"step": 816500
|
| 9906 |
+
},
|
| 9907 |
+
{
|
| 9908 |
+
"epoch": 11.76,
|
| 9909 |
+
"learning_rate": 1.6200624159435868e-05,
|
| 9910 |
+
"loss": 1.7179,
|
| 9911 |
+
"step": 817000
|
| 9912 |
+
},
|
| 9913 |
+
{
|
| 9914 |
+
"epoch": 11.77,
|
| 9915 |
+
"learning_rate": 1.6148801271166637e-05,
|
| 9916 |
+
"loss": 1.7141,
|
| 9917 |
+
"step": 817500
|
| 9918 |
+
},
|
| 9919 |
+
{
|
| 9920 |
+
"epoch": 11.77,
|
| 9921 |
+
"learning_rate": 1.6096874529413902e-05,
|
| 9922 |
+
"loss": 1.7203,
|
| 9923 |
+
"step": 818000
|
| 9924 |
+
},
|
| 9925 |
+
{
|
| 9926 |
+
"epoch": 11.78,
|
| 9927 |
+
"learning_rate": 1.604494778766117e-05,
|
| 9928 |
+
"loss": 1.7158,
|
| 9929 |
+
"step": 818500
|
| 9930 |
+
},
|
| 9931 |
+
{
|
| 9932 |
+
"epoch": 11.79,
|
| 9933 |
+
"learning_rate": 1.5993021045908432e-05,
|
| 9934 |
+
"loss": 1.7126,
|
| 9935 |
+
"step": 819000
|
| 9936 |
+
},
|
| 9937 |
+
{
|
| 9938 |
+
"epoch": 11.8,
|
| 9939 |
+
"learning_rate": 1.59410943041557e-05,
|
| 9940 |
+
"loss": 1.7146,
|
| 9941 |
+
"step": 819500
|
| 9942 |
+
},
|
| 9943 |
+
{
|
| 9944 |
+
"epoch": 11.8,
|
| 9945 |
+
"learning_rate": 1.5889271415886466e-05,
|
| 9946 |
+
"loss": 1.713,
|
| 9947 |
+
"step": 820000
|
| 9948 |
+
},
|
| 9949 |
+
{
|
| 9950 |
+
"epoch": 11.81,
|
| 9951 |
+
"learning_rate": 1.5837344674133735e-05,
|
| 9952 |
+
"loss": 1.715,
|
| 9953 |
+
"step": 820500
|
| 9954 |
+
},
|
| 9955 |
+
{
|
| 9956 |
+
"epoch": 11.82,
|
| 9957 |
+
"learning_rate": 1.5785417932380996e-05,
|
| 9958 |
+
"loss": 1.7166,
|
| 9959 |
+
"step": 821000
|
| 9960 |
+
},
|
| 9961 |
+
{
|
| 9962 |
+
"epoch": 11.82,
|
| 9963 |
+
"learning_rate": 1.5733491190628264e-05,
|
| 9964 |
+
"loss": 1.7166,
|
| 9965 |
+
"step": 821500
|
| 9966 |
+
},
|
| 9967 |
+
{
|
| 9968 |
+
"epoch": 11.83,
|
| 9969 |
+
"learning_rate": 1.5681564448875526e-05,
|
| 9970 |
+
"loss": 1.7136,
|
| 9971 |
+
"step": 822000
|
| 9972 |
+
},
|
| 9973 |
+
{
|
| 9974 |
+
"epoch": 11.84,
|
| 9975 |
+
"learning_rate": 1.562963770712279e-05,
|
| 9976 |
+
"loss": 1.7152,
|
| 9977 |
+
"step": 822500
|
| 9978 |
+
},
|
| 9979 |
+
{
|
| 9980 |
+
"epoch": 11.85,
|
| 9981 |
+
"learning_rate": 1.557781481885356e-05,
|
| 9982 |
+
"loss": 1.7161,
|
| 9983 |
+
"step": 823000
|
| 9984 |
+
},
|
| 9985 |
+
{
|
| 9986 |
+
"epoch": 11.85,
|
| 9987 |
+
"learning_rate": 1.552588807710083e-05,
|
| 9988 |
+
"loss": 1.7157,
|
| 9989 |
+
"step": 823500
|
| 9990 |
+
},
|
| 9991 |
+
{
|
| 9992 |
+
"epoch": 11.86,
|
| 9993 |
+
"learning_rate": 1.547396133534809e-05,
|
| 9994 |
+
"loss": 1.7162,
|
| 9995 |
+
"step": 824000
|
| 9996 |
+
},
|
| 9997 |
+
{
|
| 9998 |
+
"epoch": 11.87,
|
| 9999 |
+
"learning_rate": 1.5422034593595355e-05,
|
| 10000 |
+
"loss": 1.717,
|
| 10001 |
+
"step": 824500
|
| 10002 |
+
},
|
| 10003 |
+
{
|
| 10004 |
+
"epoch": 11.88,
|
| 10005 |
+
"learning_rate": 1.5370211705326125e-05,
|
| 10006 |
+
"loss": 1.7182,
|
| 10007 |
+
"step": 825000
|
| 10008 |
+
},
|
| 10009 |
+
{
|
| 10010 |
+
"epoch": 11.88,
|
| 10011 |
+
"learning_rate": 1.5318284963573393e-05,
|
| 10012 |
+
"loss": 1.7174,
|
| 10013 |
+
"step": 825500
|
| 10014 |
+
},
|
| 10015 |
+
{
|
| 10016 |
+
"epoch": 11.89,
|
| 10017 |
+
"learning_rate": 1.5266358221820655e-05,
|
| 10018 |
+
"loss": 1.7126,
|
| 10019 |
+
"step": 826000
|
| 10020 |
+
},
|
| 10021 |
+
{
|
| 10022 |
+
"epoch": 11.9,
|
| 10023 |
+
"learning_rate": 1.521443148006792e-05,
|
| 10024 |
+
"loss": 1.7112,
|
| 10025 |
+
"step": 826500
|
| 10026 |
+
},
|
| 10027 |
+
{
|
| 10028 |
+
"epoch": 11.9,
|
| 10029 |
+
"learning_rate": 1.5162608591798691e-05,
|
| 10030 |
+
"loss": 1.7143,
|
| 10031 |
+
"step": 827000
|
| 10032 |
+
},
|
| 10033 |
+
{
|
| 10034 |
+
"epoch": 11.91,
|
| 10035 |
+
"learning_rate": 1.5110681850045954e-05,
|
| 10036 |
+
"loss": 1.7127,
|
| 10037 |
+
"step": 827500
|
| 10038 |
+
},
|
| 10039 |
+
{
|
| 10040 |
+
"epoch": 11.92,
|
| 10041 |
+
"learning_rate": 1.505875510829322e-05,
|
| 10042 |
+
"loss": 1.72,
|
| 10043 |
+
"step": 828000
|
| 10044 |
+
},
|
| 10045 |
+
{
|
| 10046 |
+
"epoch": 11.93,
|
| 10047 |
+
"learning_rate": 1.5006828366540484e-05,
|
| 10048 |
+
"loss": 1.7146,
|
| 10049 |
+
"step": 828500
|
| 10050 |
+
},
|
| 10051 |
+
{
|
| 10052 |
+
"epoch": 11.93,
|
| 10053 |
+
"learning_rate": 1.4955005478271255e-05,
|
| 10054 |
+
"loss": 1.7156,
|
| 10055 |
+
"step": 829000
|
| 10056 |
+
},
|
| 10057 |
+
{
|
| 10058 |
+
"epoch": 11.94,
|
| 10059 |
+
"learning_rate": 1.4903078736518518e-05,
|
| 10060 |
+
"loss": 1.7144,
|
| 10061 |
+
"step": 829500
|
| 10062 |
+
},
|
| 10063 |
+
{
|
| 10064 |
+
"epoch": 11.95,
|
| 10065 |
+
"learning_rate": 1.4851151994765785e-05,
|
| 10066 |
+
"loss": 1.7136,
|
| 10067 |
+
"step": 830000
|
| 10068 |
+
},
|
| 10069 |
+
{
|
| 10070 |
+
"epoch": 11.95,
|
| 10071 |
+
"learning_rate": 1.479922525301305e-05,
|
| 10072 |
+
"loss": 1.713,
|
| 10073 |
+
"step": 830500
|
| 10074 |
+
},
|
| 10075 |
+
{
|
| 10076 |
+
"epoch": 11.96,
|
| 10077 |
+
"learning_rate": 1.474740236474382e-05,
|
| 10078 |
+
"loss": 1.7165,
|
| 10079 |
+
"step": 831000
|
| 10080 |
+
},
|
| 10081 |
+
{
|
| 10082 |
+
"epoch": 11.97,
|
| 10083 |
+
"learning_rate": 1.4695475622991084e-05,
|
| 10084 |
+
"loss": 1.7137,
|
| 10085 |
+
"step": 831500
|
| 10086 |
+
},
|
| 10087 |
+
{
|
| 10088 |
+
"epoch": 11.98,
|
| 10089 |
+
"learning_rate": 1.4643548881238351e-05,
|
| 10090 |
+
"loss": 1.7115,
|
| 10091 |
+
"step": 832000
|
| 10092 |
+
},
|
| 10093 |
+
{
|
| 10094 |
+
"epoch": 11.98,
|
| 10095 |
+
"learning_rate": 1.4591622139485614e-05,
|
| 10096 |
+
"loss": 1.7183,
|
| 10097 |
+
"step": 832500
|
| 10098 |
+
},
|
| 10099 |
+
{
|
| 10100 |
+
"epoch": 11.99,
|
| 10101 |
+
"learning_rate": 1.4539695397732881e-05,
|
| 10102 |
+
"loss": 1.7142,
|
| 10103 |
+
"step": 833000
|
| 10104 |
+
},
|
| 10105 |
+
{
|
| 10106 |
+
"epoch": 12.0,
|
| 10107 |
+
"learning_rate": 1.4487872509463649e-05,
|
| 10108 |
+
"loss": 1.7127,
|
| 10109 |
+
"step": 833500
|
| 10110 |
+
},
|
| 10111 |
+
{
|
| 10112 |
+
"epoch": 12.0,
|
| 10113 |
+
"eval_accuracy": 0.6658040615937678,
|
| 10114 |
+
"eval_loss": 1.5836162567138672,
|
| 10115 |
+
"eval_runtime": 1303.2983,
|
| 10116 |
+
"eval_samples_per_second": 413.518,
|
| 10117 |
+
"eval_steps_per_second": 25.845,
|
| 10118 |
+
"step": 833676
|
| 10119 |
}
|
| 10120 |
],
|
| 10121 |
"max_steps": 972622,
|
| 10122 |
"num_train_epochs": 14,
|
| 10123 |
+
"total_flos": 6.311350074383032e+18,
|
| 10124 |
"trial_name": null,
|
| 10125 |
"trial_params": null
|
| 10126 |
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 118243218
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e8e789cbff592c7abcbb70a0a1080069b2a642bdd41392da72c8a88aa5f570ae
|
| 3 |
size 118243218
|
runs/Feb22_09-35-52_ubuntu-2004/events.out.tfevents.1677026198.ubuntu-2004.1870487.0
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 sha256:54427ce9472c28a9802ca844bd62a2af3e4bc1d184b1bcde76bbe915e3c41367
|
| 3 |
+
size 274328
|