Upload folder using huggingface_hub
Browse files- adapter_model.safetensors +1 -1
- optimizer.pt +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +1053 -3
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3705288
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:824a420a29fbca187fa0b6e671b0317cd1adf88bba812da28f574cf5fa0d02a3
|
| 3 |
size 3705288
|
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2213690
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:73ce32829785e98d533baacbe635c9f455ccb83097a89bb1320503636d55672c
|
| 3 |
size 2213690
|
rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:20c684281fc59eb041b568ca6312fe070f3ba7bbd2cfa4e1743a07104a2ddde8
|
| 3 |
size 14244
|
scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cda7ce864d0e71320eb0cb58ba5c2f7aa8429fdb142b72675c9c8df9d5b58d17
|
| 3 |
size 1064
|
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -5257,6 +5257,1056 @@
|
|
| 5257 |
"learning_rate": 5.320197044334976e-05,
|
| 5258 |
"loss": 0.2625,
|
| 5259 |
"step": 15000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5260 |
}
|
| 5261 |
],
|
| 5262 |
"logging_steps": 20,
|
|
@@ -5276,7 +6326,7 @@
|
|
| 5276 |
"attributes": {}
|
| 5277 |
}
|
| 5278 |
},
|
| 5279 |
-
"total_flos":
|
| 5280 |
"train_batch_size": 1,
|
| 5281 |
"trial_name": null,
|
| 5282 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 87.85845027455765,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 18000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 5257 |
"learning_rate": 5.320197044334976e-05,
|
| 5258 |
"loss": 0.2625,
|
| 5259 |
"step": 15000
|
| 5260 |
+
},
|
| 5261 |
+
{
|
| 5262 |
+
"epoch": 73.31299572910311,
|
| 5263 |
+
"grad_norm": 3.005783796310425,
|
| 5264 |
+
"learning_rate": 5.3004926108374385e-05,
|
| 5265 |
+
"loss": 0.2589,
|
| 5266 |
+
"step": 15020
|
| 5267 |
+
},
|
| 5268 |
+
{
|
| 5269 |
+
"epoch": 73.41061622940818,
|
| 5270 |
+
"grad_norm": 2.7578892707824707,
|
| 5271 |
+
"learning_rate": 5.280788177339901e-05,
|
| 5272 |
+
"loss": 0.2519,
|
| 5273 |
+
"step": 15040
|
| 5274 |
+
},
|
| 5275 |
+
{
|
| 5276 |
+
"epoch": 73.50823672971325,
|
| 5277 |
+
"grad_norm": 3.286733627319336,
|
| 5278 |
+
"learning_rate": 5.2610837438423646e-05,
|
| 5279 |
+
"loss": 0.2603,
|
| 5280 |
+
"step": 15060
|
| 5281 |
+
},
|
| 5282 |
+
{
|
| 5283 |
+
"epoch": 73.6058572300183,
|
| 5284 |
+
"grad_norm": 2.323225975036621,
|
| 5285 |
+
"learning_rate": 5.241379310344828e-05,
|
| 5286 |
+
"loss": 0.2576,
|
| 5287 |
+
"step": 15080
|
| 5288 |
+
},
|
| 5289 |
+
{
|
| 5290 |
+
"epoch": 73.70347773032337,
|
| 5291 |
+
"grad_norm": 2.407222032546997,
|
| 5292 |
+
"learning_rate": 5.2216748768472914e-05,
|
| 5293 |
+
"loss": 0.253,
|
| 5294 |
+
"step": 15100
|
| 5295 |
+
},
|
| 5296 |
+
{
|
| 5297 |
+
"epoch": 73.80109823062843,
|
| 5298 |
+
"grad_norm": 3.0755960941314697,
|
| 5299 |
+
"learning_rate": 5.201970443349754e-05,
|
| 5300 |
+
"loss": 0.261,
|
| 5301 |
+
"step": 15120
|
| 5302 |
+
},
|
| 5303 |
+
{
|
| 5304 |
+
"epoch": 73.8987187309335,
|
| 5305 |
+
"grad_norm": 1.9469565153121948,
|
| 5306 |
+
"learning_rate": 5.182266009852217e-05,
|
| 5307 |
+
"loss": 0.2556,
|
| 5308 |
+
"step": 15140
|
| 5309 |
+
},
|
| 5310 |
+
{
|
| 5311 |
+
"epoch": 73.99633923123857,
|
| 5312 |
+
"grad_norm": 3.5689964294433594,
|
| 5313 |
+
"learning_rate": 5.1625615763546795e-05,
|
| 5314 |
+
"loss": 0.2718,
|
| 5315 |
+
"step": 15160
|
| 5316 |
+
},
|
| 5317 |
+
{
|
| 5318 |
+
"epoch": 74.09395973154362,
|
| 5319 |
+
"grad_norm": 1.9299124479293823,
|
| 5320 |
+
"learning_rate": 5.142857142857143e-05,
|
| 5321 |
+
"loss": 0.2497,
|
| 5322 |
+
"step": 15180
|
| 5323 |
+
},
|
| 5324 |
+
{
|
| 5325 |
+
"epoch": 74.19158023184869,
|
| 5326 |
+
"grad_norm": 2.1597163677215576,
|
| 5327 |
+
"learning_rate": 5.123152709359606e-05,
|
| 5328 |
+
"loss": 0.2526,
|
| 5329 |
+
"step": 15200
|
| 5330 |
+
},
|
| 5331 |
+
{
|
| 5332 |
+
"epoch": 74.28920073215376,
|
| 5333 |
+
"grad_norm": 2.4359443187713623,
|
| 5334 |
+
"learning_rate": 5.10344827586207e-05,
|
| 5335 |
+
"loss": 0.2557,
|
| 5336 |
+
"step": 15220
|
| 5337 |
+
},
|
| 5338 |
+
{
|
| 5339 |
+
"epoch": 74.38682123245881,
|
| 5340 |
+
"grad_norm": 2.449601411819458,
|
| 5341 |
+
"learning_rate": 5.0837438423645324e-05,
|
| 5342 |
+
"loss": 0.2628,
|
| 5343 |
+
"step": 15240
|
| 5344 |
+
},
|
| 5345 |
+
{
|
| 5346 |
+
"epoch": 74.48444173276388,
|
| 5347 |
+
"grad_norm": 2.5450046062469482,
|
| 5348 |
+
"learning_rate": 5.064039408866995e-05,
|
| 5349 |
+
"loss": 0.2683,
|
| 5350 |
+
"step": 15260
|
| 5351 |
+
},
|
| 5352 |
+
{
|
| 5353 |
+
"epoch": 74.58206223306894,
|
| 5354 |
+
"grad_norm": 2.499568462371826,
|
| 5355 |
+
"learning_rate": 5.044334975369458e-05,
|
| 5356 |
+
"loss": 0.2456,
|
| 5357 |
+
"step": 15280
|
| 5358 |
+
},
|
| 5359 |
+
{
|
| 5360 |
+
"epoch": 74.67968273337401,
|
| 5361 |
+
"grad_norm": 2.276536703109741,
|
| 5362 |
+
"learning_rate": 5.024630541871922e-05,
|
| 5363 |
+
"loss": 0.2613,
|
| 5364 |
+
"step": 15300
|
| 5365 |
+
},
|
| 5366 |
+
{
|
| 5367 |
+
"epoch": 74.77730323367908,
|
| 5368 |
+
"grad_norm": 6.047021865844727,
|
| 5369 |
+
"learning_rate": 5.0049261083743846e-05,
|
| 5370 |
+
"loss": 0.2591,
|
| 5371 |
+
"step": 15320
|
| 5372 |
+
},
|
| 5373 |
+
{
|
| 5374 |
+
"epoch": 74.87492373398413,
|
| 5375 |
+
"grad_norm": 2.7853705883026123,
|
| 5376 |
+
"learning_rate": 4.985221674876848e-05,
|
| 5377 |
+
"loss": 0.2584,
|
| 5378 |
+
"step": 15340
|
| 5379 |
+
},
|
| 5380 |
+
{
|
| 5381 |
+
"epoch": 74.9725442342892,
|
| 5382 |
+
"grad_norm": 2.658870220184326,
|
| 5383 |
+
"learning_rate": 4.9655172413793107e-05,
|
| 5384 |
+
"loss": 0.2485,
|
| 5385 |
+
"step": 15360
|
| 5386 |
+
},
|
| 5387 |
+
{
|
| 5388 |
+
"epoch": 75.07016473459427,
|
| 5389 |
+
"grad_norm": 1.9290242195129395,
|
| 5390 |
+
"learning_rate": 4.9458128078817734e-05,
|
| 5391 |
+
"loss": 0.2456,
|
| 5392 |
+
"step": 15380
|
| 5393 |
+
},
|
| 5394 |
+
{
|
| 5395 |
+
"epoch": 75.16778523489933,
|
| 5396 |
+
"grad_norm": 2.4340288639068604,
|
| 5397 |
+
"learning_rate": 4.926108374384237e-05,
|
| 5398 |
+
"loss": 0.2517,
|
| 5399 |
+
"step": 15400
|
| 5400 |
+
},
|
| 5401 |
+
{
|
| 5402 |
+
"epoch": 75.2654057352044,
|
| 5403 |
+
"grad_norm": 1.7368818521499634,
|
| 5404 |
+
"learning_rate": 4.9064039408866995e-05,
|
| 5405 |
+
"loss": 0.245,
|
| 5406 |
+
"step": 15420
|
| 5407 |
+
},
|
| 5408 |
+
{
|
| 5409 |
+
"epoch": 75.36302623550945,
|
| 5410 |
+
"grad_norm": 3.224472999572754,
|
| 5411 |
+
"learning_rate": 4.886699507389163e-05,
|
| 5412 |
+
"loss": 0.2512,
|
| 5413 |
+
"step": 15440
|
| 5414 |
+
},
|
| 5415 |
+
{
|
| 5416 |
+
"epoch": 75.46064673581452,
|
| 5417 |
+
"grad_norm": 2.9347827434539795,
|
| 5418 |
+
"learning_rate": 4.866995073891626e-05,
|
| 5419 |
+
"loss": 0.252,
|
| 5420 |
+
"step": 15460
|
| 5421 |
+
},
|
| 5422 |
+
{
|
| 5423 |
+
"epoch": 75.55826723611959,
|
| 5424 |
+
"grad_norm": 3.1281368732452393,
|
| 5425 |
+
"learning_rate": 4.847290640394089e-05,
|
| 5426 |
+
"loss": 0.2662,
|
| 5427 |
+
"step": 15480
|
| 5428 |
+
},
|
| 5429 |
+
{
|
| 5430 |
+
"epoch": 75.65588773642465,
|
| 5431 |
+
"grad_norm": 2.1834158897399902,
|
| 5432 |
+
"learning_rate": 4.827586206896552e-05,
|
| 5433 |
+
"loss": 0.2549,
|
| 5434 |
+
"step": 15500
|
| 5435 |
+
},
|
| 5436 |
+
{
|
| 5437 |
+
"epoch": 75.75350823672972,
|
| 5438 |
+
"grad_norm": 2.4959053993225098,
|
| 5439 |
+
"learning_rate": 4.807881773399015e-05,
|
| 5440 |
+
"loss": 0.2489,
|
| 5441 |
+
"step": 15520
|
| 5442 |
+
},
|
| 5443 |
+
{
|
| 5444 |
+
"epoch": 75.85112873703477,
|
| 5445 |
+
"grad_norm": 1.9630552530288696,
|
| 5446 |
+
"learning_rate": 4.788177339901478e-05,
|
| 5447 |
+
"loss": 0.2685,
|
| 5448 |
+
"step": 15540
|
| 5449 |
+
},
|
| 5450 |
+
{
|
| 5451 |
+
"epoch": 75.94874923733984,
|
| 5452 |
+
"grad_norm": 2.9730660915374756,
|
| 5453 |
+
"learning_rate": 4.768472906403941e-05,
|
| 5454 |
+
"loss": 0.2568,
|
| 5455 |
+
"step": 15560
|
| 5456 |
+
},
|
| 5457 |
+
{
|
| 5458 |
+
"epoch": 76.04636973764491,
|
| 5459 |
+
"grad_norm": 2.492307186126709,
|
| 5460 |
+
"learning_rate": 4.748768472906404e-05,
|
| 5461 |
+
"loss": 0.254,
|
| 5462 |
+
"step": 15580
|
| 5463 |
+
},
|
| 5464 |
+
{
|
| 5465 |
+
"epoch": 76.14399023794996,
|
| 5466 |
+
"grad_norm": 2.1463494300842285,
|
| 5467 |
+
"learning_rate": 4.729064039408867e-05,
|
| 5468 |
+
"loss": 0.2623,
|
| 5469 |
+
"step": 15600
|
| 5470 |
+
},
|
| 5471 |
+
{
|
| 5472 |
+
"epoch": 76.24161073825503,
|
| 5473 |
+
"grad_norm": 2.957017421722412,
|
| 5474 |
+
"learning_rate": 4.7093596059113306e-05,
|
| 5475 |
+
"loss": 0.2576,
|
| 5476 |
+
"step": 15620
|
| 5477 |
+
},
|
| 5478 |
+
{
|
| 5479 |
+
"epoch": 76.3392312385601,
|
| 5480 |
+
"grad_norm": 2.1611711978912354,
|
| 5481 |
+
"learning_rate": 4.689655172413793e-05,
|
| 5482 |
+
"loss": 0.2447,
|
| 5483 |
+
"step": 15640
|
| 5484 |
+
},
|
| 5485 |
+
{
|
| 5486 |
+
"epoch": 76.43685173886516,
|
| 5487 |
+
"grad_norm": 3.1399998664855957,
|
| 5488 |
+
"learning_rate": 4.669950738916256e-05,
|
| 5489 |
+
"loss": 0.2586,
|
| 5490 |
+
"step": 15660
|
| 5491 |
+
},
|
| 5492 |
+
{
|
| 5493 |
+
"epoch": 76.53447223917023,
|
| 5494 |
+
"grad_norm": 2.817157030105591,
|
| 5495 |
+
"learning_rate": 4.6502463054187194e-05,
|
| 5496 |
+
"loss": 0.2439,
|
| 5497 |
+
"step": 15680
|
| 5498 |
+
},
|
| 5499 |
+
{
|
| 5500 |
+
"epoch": 76.63209273947528,
|
| 5501 |
+
"grad_norm": 1.3343191146850586,
|
| 5502 |
+
"learning_rate": 4.630541871921182e-05,
|
| 5503 |
+
"loss": 0.2522,
|
| 5504 |
+
"step": 15700
|
| 5505 |
+
},
|
| 5506 |
+
{
|
| 5507 |
+
"epoch": 76.72971323978035,
|
| 5508 |
+
"grad_norm": 2.9455504417419434,
|
| 5509 |
+
"learning_rate": 4.6108374384236455e-05,
|
| 5510 |
+
"loss": 0.2606,
|
| 5511 |
+
"step": 15720
|
| 5512 |
+
},
|
| 5513 |
+
{
|
| 5514 |
+
"epoch": 76.82733374008542,
|
| 5515 |
+
"grad_norm": 2.981264352798462,
|
| 5516 |
+
"learning_rate": 4.591133004926109e-05,
|
| 5517 |
+
"loss": 0.2482,
|
| 5518 |
+
"step": 15740
|
| 5519 |
+
},
|
| 5520 |
+
{
|
| 5521 |
+
"epoch": 76.92495424039048,
|
| 5522 |
+
"grad_norm": 2.9296011924743652,
|
| 5523 |
+
"learning_rate": 4.5714285714285716e-05,
|
| 5524 |
+
"loss": 0.2578,
|
| 5525 |
+
"step": 15760
|
| 5526 |
+
},
|
| 5527 |
+
{
|
| 5528 |
+
"epoch": 77.02257474069555,
|
| 5529 |
+
"grad_norm": 2.8159282207489014,
|
| 5530 |
+
"learning_rate": 4.551724137931035e-05,
|
| 5531 |
+
"loss": 0.2571,
|
| 5532 |
+
"step": 15780
|
| 5533 |
+
},
|
| 5534 |
+
{
|
| 5535 |
+
"epoch": 77.12019524100062,
|
| 5536 |
+
"grad_norm": 2.184053421020508,
|
| 5537 |
+
"learning_rate": 4.532019704433498e-05,
|
| 5538 |
+
"loss": 0.2548,
|
| 5539 |
+
"step": 15800
|
| 5540 |
+
},
|
| 5541 |
+
{
|
| 5542 |
+
"epoch": 77.21781574130567,
|
| 5543 |
+
"grad_norm": 2.1801810264587402,
|
| 5544 |
+
"learning_rate": 4.5123152709359604e-05,
|
| 5545 |
+
"loss": 0.2367,
|
| 5546 |
+
"step": 15820
|
| 5547 |
+
},
|
| 5548 |
+
{
|
| 5549 |
+
"epoch": 77.31543624161074,
|
| 5550 |
+
"grad_norm": 2.510050058364868,
|
| 5551 |
+
"learning_rate": 4.492610837438424e-05,
|
| 5552 |
+
"loss": 0.2522,
|
| 5553 |
+
"step": 15840
|
| 5554 |
+
},
|
| 5555 |
+
{
|
| 5556 |
+
"epoch": 77.4130567419158,
|
| 5557 |
+
"grad_norm": 2.849837303161621,
|
| 5558 |
+
"learning_rate": 4.472906403940887e-05,
|
| 5559 |
+
"loss": 0.2524,
|
| 5560 |
+
"step": 15860
|
| 5561 |
+
},
|
| 5562 |
+
{
|
| 5563 |
+
"epoch": 77.51067724222086,
|
| 5564 |
+
"grad_norm": 3.769998788833618,
|
| 5565 |
+
"learning_rate": 4.45320197044335e-05,
|
| 5566 |
+
"loss": 0.2568,
|
| 5567 |
+
"step": 15880
|
| 5568 |
+
},
|
| 5569 |
+
{
|
| 5570 |
+
"epoch": 77.60829774252593,
|
| 5571 |
+
"grad_norm": 3.2575082778930664,
|
| 5572 |
+
"learning_rate": 4.433497536945813e-05,
|
| 5573 |
+
"loss": 0.2565,
|
| 5574 |
+
"step": 15900
|
| 5575 |
+
},
|
| 5576 |
+
{
|
| 5577 |
+
"epoch": 77.70591824283099,
|
| 5578 |
+
"grad_norm": 2.199042797088623,
|
| 5579 |
+
"learning_rate": 4.413793103448276e-05,
|
| 5580 |
+
"loss": 0.2508,
|
| 5581 |
+
"step": 15920
|
| 5582 |
+
},
|
| 5583 |
+
{
|
| 5584 |
+
"epoch": 77.80353874313606,
|
| 5585 |
+
"grad_norm": 1.9908735752105713,
|
| 5586 |
+
"learning_rate": 4.394088669950739e-05,
|
| 5587 |
+
"loss": 0.2612,
|
| 5588 |
+
"step": 15940
|
| 5589 |
+
},
|
| 5590 |
+
{
|
| 5591 |
+
"epoch": 77.90115924344113,
|
| 5592 |
+
"grad_norm": 2.091723680496216,
|
| 5593 |
+
"learning_rate": 4.374384236453202e-05,
|
| 5594 |
+
"loss": 0.2491,
|
| 5595 |
+
"step": 15960
|
| 5596 |
+
},
|
| 5597 |
+
{
|
| 5598 |
+
"epoch": 77.99877974374618,
|
| 5599 |
+
"grad_norm": 2.705829381942749,
|
| 5600 |
+
"learning_rate": 4.3546798029556655e-05,
|
| 5601 |
+
"loss": 0.2596,
|
| 5602 |
+
"step": 15980
|
| 5603 |
+
},
|
| 5604 |
+
{
|
| 5605 |
+
"epoch": 78.09640024405125,
|
| 5606 |
+
"grad_norm": 2.6604998111724854,
|
| 5607 |
+
"learning_rate": 4.334975369458129e-05,
|
| 5608 |
+
"loss": 0.2475,
|
| 5609 |
+
"step": 16000
|
| 5610 |
+
},
|
| 5611 |
+
{
|
| 5612 |
+
"epoch": 78.19402074435631,
|
| 5613 |
+
"grad_norm": 2.4286489486694336,
|
| 5614 |
+
"learning_rate": 4.3152709359605916e-05,
|
| 5615 |
+
"loss": 0.2509,
|
| 5616 |
+
"step": 16020
|
| 5617 |
+
},
|
| 5618 |
+
{
|
| 5619 |
+
"epoch": 78.29164124466138,
|
| 5620 |
+
"grad_norm": 3.3478493690490723,
|
| 5621 |
+
"learning_rate": 4.295566502463054e-05,
|
| 5622 |
+
"loss": 0.2491,
|
| 5623 |
+
"step": 16040
|
| 5624 |
+
},
|
| 5625 |
+
{
|
| 5626 |
+
"epoch": 78.38926174496645,
|
| 5627 |
+
"grad_norm": 2.9512908458709717,
|
| 5628 |
+
"learning_rate": 4.275862068965518e-05,
|
| 5629 |
+
"loss": 0.2362,
|
| 5630 |
+
"step": 16060
|
| 5631 |
+
},
|
| 5632 |
+
{
|
| 5633 |
+
"epoch": 78.4868822452715,
|
| 5634 |
+
"grad_norm": 2.0870890617370605,
|
| 5635 |
+
"learning_rate": 4.2561576354679804e-05,
|
| 5636 |
+
"loss": 0.2546,
|
| 5637 |
+
"step": 16080
|
| 5638 |
+
},
|
| 5639 |
+
{
|
| 5640 |
+
"epoch": 78.58450274557657,
|
| 5641 |
+
"grad_norm": 2.3549749851226807,
|
| 5642 |
+
"learning_rate": 4.236453201970443e-05,
|
| 5643 |
+
"loss": 0.2544,
|
| 5644 |
+
"step": 16100
|
| 5645 |
+
},
|
| 5646 |
+
{
|
| 5647 |
+
"epoch": 78.68212324588164,
|
| 5648 |
+
"grad_norm": 2.296377658843994,
|
| 5649 |
+
"learning_rate": 4.2167487684729065e-05,
|
| 5650 |
+
"loss": 0.2524,
|
| 5651 |
+
"step": 16120
|
| 5652 |
+
},
|
| 5653 |
+
{
|
| 5654 |
+
"epoch": 78.7797437461867,
|
| 5655 |
+
"grad_norm": 2.9563801288604736,
|
| 5656 |
+
"learning_rate": 4.19704433497537e-05,
|
| 5657 |
+
"loss": 0.2534,
|
| 5658 |
+
"step": 16140
|
| 5659 |
+
},
|
| 5660 |
+
{
|
| 5661 |
+
"epoch": 78.87736424649177,
|
| 5662 |
+
"grad_norm": 3.3844058513641357,
|
| 5663 |
+
"learning_rate": 4.1773399014778326e-05,
|
| 5664 |
+
"loss": 0.2629,
|
| 5665 |
+
"step": 16160
|
| 5666 |
+
},
|
| 5667 |
+
{
|
| 5668 |
+
"epoch": 78.97498474679682,
|
| 5669 |
+
"grad_norm": 1.9131345748901367,
|
| 5670 |
+
"learning_rate": 4.157635467980296e-05,
|
| 5671 |
+
"loss": 0.2478,
|
| 5672 |
+
"step": 16180
|
| 5673 |
+
},
|
| 5674 |
+
{
|
| 5675 |
+
"epoch": 79.07260524710189,
|
| 5676 |
+
"grad_norm": 3.4866435527801514,
|
| 5677 |
+
"learning_rate": 4.1379310344827587e-05,
|
| 5678 |
+
"loss": 0.2464,
|
| 5679 |
+
"step": 16200
|
| 5680 |
+
},
|
| 5681 |
+
{
|
| 5682 |
+
"epoch": 79.17022574740696,
|
| 5683 |
+
"grad_norm": 2.0751941204071045,
|
| 5684 |
+
"learning_rate": 4.1182266009852214e-05,
|
| 5685 |
+
"loss": 0.2435,
|
| 5686 |
+
"step": 16220
|
| 5687 |
+
},
|
| 5688 |
+
{
|
| 5689 |
+
"epoch": 79.26784624771201,
|
| 5690 |
+
"grad_norm": 1.776879072189331,
|
| 5691 |
+
"learning_rate": 4.098522167487685e-05,
|
| 5692 |
+
"loss": 0.2501,
|
| 5693 |
+
"step": 16240
|
| 5694 |
+
},
|
| 5695 |
+
{
|
| 5696 |
+
"epoch": 79.36546674801708,
|
| 5697 |
+
"grad_norm": 3.9006545543670654,
|
| 5698 |
+
"learning_rate": 4.078817733990148e-05,
|
| 5699 |
+
"loss": 0.2586,
|
| 5700 |
+
"step": 16260
|
| 5701 |
+
},
|
| 5702 |
+
{
|
| 5703 |
+
"epoch": 79.46308724832215,
|
| 5704 |
+
"grad_norm": 2.390000581741333,
|
| 5705 |
+
"learning_rate": 4.0591133004926115e-05,
|
| 5706 |
+
"loss": 0.2408,
|
| 5707 |
+
"step": 16280
|
| 5708 |
+
},
|
| 5709 |
+
{
|
| 5710 |
+
"epoch": 79.56070774862721,
|
| 5711 |
+
"grad_norm": 3.1795706748962402,
|
| 5712 |
+
"learning_rate": 4.039408866995074e-05,
|
| 5713 |
+
"loss": 0.2469,
|
| 5714 |
+
"step": 16300
|
| 5715 |
+
},
|
| 5716 |
+
{
|
| 5717 |
+
"epoch": 79.65832824893228,
|
| 5718 |
+
"grad_norm": 2.6821188926696777,
|
| 5719 |
+
"learning_rate": 4.019704433497537e-05,
|
| 5720 |
+
"loss": 0.2429,
|
| 5721 |
+
"step": 16320
|
| 5722 |
+
},
|
| 5723 |
+
{
|
| 5724 |
+
"epoch": 79.75594874923733,
|
| 5725 |
+
"grad_norm": 3.01457142829895,
|
| 5726 |
+
"learning_rate": 4e-05,
|
| 5727 |
+
"loss": 0.2598,
|
| 5728 |
+
"step": 16340
|
| 5729 |
+
},
|
| 5730 |
+
{
|
| 5731 |
+
"epoch": 79.8535692495424,
|
| 5732 |
+
"grad_norm": 2.8440592288970947,
|
| 5733 |
+
"learning_rate": 3.980295566502463e-05,
|
| 5734 |
+
"loss": 0.2494,
|
| 5735 |
+
"step": 16360
|
| 5736 |
+
},
|
| 5737 |
+
{
|
| 5738 |
+
"epoch": 79.95118974984747,
|
| 5739 |
+
"grad_norm": 3.210845708847046,
|
| 5740 |
+
"learning_rate": 3.9605911330049264e-05,
|
| 5741 |
+
"loss": 0.2521,
|
| 5742 |
+
"step": 16380
|
| 5743 |
+
},
|
| 5744 |
+
{
|
| 5745 |
+
"epoch": 80.04881025015253,
|
| 5746 |
+
"grad_norm": 3.9740731716156006,
|
| 5747 |
+
"learning_rate": 3.94088669950739e-05,
|
| 5748 |
+
"loss": 0.2537,
|
| 5749 |
+
"step": 16400
|
| 5750 |
+
},
|
| 5751 |
+
{
|
| 5752 |
+
"epoch": 80.1464307504576,
|
| 5753 |
+
"grad_norm": 2.3433115482330322,
|
| 5754 |
+
"learning_rate": 3.9211822660098525e-05,
|
| 5755 |
+
"loss": 0.2441,
|
| 5756 |
+
"step": 16420
|
| 5757 |
+
},
|
| 5758 |
+
{
|
| 5759 |
+
"epoch": 80.24405125076267,
|
| 5760 |
+
"grad_norm": 2.5279314517974854,
|
| 5761 |
+
"learning_rate": 3.901477832512315e-05,
|
| 5762 |
+
"loss": 0.2564,
|
| 5763 |
+
"step": 16440
|
| 5764 |
+
},
|
| 5765 |
+
{
|
| 5766 |
+
"epoch": 80.34167175106772,
|
| 5767 |
+
"grad_norm": 2.8062689304351807,
|
| 5768 |
+
"learning_rate": 3.8817733990147786e-05,
|
| 5769 |
+
"loss": 0.245,
|
| 5770 |
+
"step": 16460
|
| 5771 |
+
},
|
| 5772 |
+
{
|
| 5773 |
+
"epoch": 80.43929225137279,
|
| 5774 |
+
"grad_norm": 1.9689416885375977,
|
| 5775 |
+
"learning_rate": 3.862068965517241e-05,
|
| 5776 |
+
"loss": 0.2497,
|
| 5777 |
+
"step": 16480
|
| 5778 |
+
},
|
| 5779 |
+
{
|
| 5780 |
+
"epoch": 80.53691275167785,
|
| 5781 |
+
"grad_norm": 2.462744951248169,
|
| 5782 |
+
"learning_rate": 3.842364532019704e-05,
|
| 5783 |
+
"loss": 0.2525,
|
| 5784 |
+
"step": 16500
|
| 5785 |
+
},
|
| 5786 |
+
{
|
| 5787 |
+
"epoch": 80.63453325198292,
|
| 5788 |
+
"grad_norm": 1.9201568365097046,
|
| 5789 |
+
"learning_rate": 3.822660098522168e-05,
|
| 5790 |
+
"loss": 0.2495,
|
| 5791 |
+
"step": 16520
|
| 5792 |
+
},
|
| 5793 |
+
{
|
| 5794 |
+
"epoch": 80.73215375228799,
|
| 5795 |
+
"grad_norm": 1.7118130922317505,
|
| 5796 |
+
"learning_rate": 3.802955665024631e-05,
|
| 5797 |
+
"loss": 0.2415,
|
| 5798 |
+
"step": 16540
|
| 5799 |
+
},
|
| 5800 |
+
{
|
| 5801 |
+
"epoch": 80.82977425259304,
|
| 5802 |
+
"grad_norm": 2.311931848526001,
|
| 5803 |
+
"learning_rate": 3.783251231527094e-05,
|
| 5804 |
+
"loss": 0.247,
|
| 5805 |
+
"step": 16560
|
| 5806 |
+
},
|
| 5807 |
+
{
|
| 5808 |
+
"epoch": 80.92739475289811,
|
| 5809 |
+
"grad_norm": 2.030750274658203,
|
| 5810 |
+
"learning_rate": 3.763546798029557e-05,
|
| 5811 |
+
"loss": 0.2415,
|
| 5812 |
+
"step": 16580
|
| 5813 |
+
},
|
| 5814 |
+
{
|
| 5815 |
+
"epoch": 81.02501525320318,
|
| 5816 |
+
"grad_norm": 1.949194312095642,
|
| 5817 |
+
"learning_rate": 3.7438423645320196e-05,
|
| 5818 |
+
"loss": 0.2555,
|
| 5819 |
+
"step": 16600
|
| 5820 |
+
},
|
| 5821 |
+
{
|
| 5822 |
+
"epoch": 81.12263575350823,
|
| 5823 |
+
"grad_norm": 1.8409544229507446,
|
| 5824 |
+
"learning_rate": 3.724137931034483e-05,
|
| 5825 |
+
"loss": 0.2412,
|
| 5826 |
+
"step": 16620
|
| 5827 |
+
},
|
| 5828 |
+
{
|
| 5829 |
+
"epoch": 81.2202562538133,
|
| 5830 |
+
"grad_norm": 2.5164377689361572,
|
| 5831 |
+
"learning_rate": 3.704433497536946e-05,
|
| 5832 |
+
"loss": 0.2326,
|
| 5833 |
+
"step": 16640
|
| 5834 |
+
},
|
| 5835 |
+
{
|
| 5836 |
+
"epoch": 81.31787675411836,
|
| 5837 |
+
"grad_norm": 2.3859026432037354,
|
| 5838 |
+
"learning_rate": 3.684729064039409e-05,
|
| 5839 |
+
"loss": 0.2499,
|
| 5840 |
+
"step": 16660
|
| 5841 |
+
},
|
| 5842 |
+
{
|
| 5843 |
+
"epoch": 81.41549725442343,
|
| 5844 |
+
"grad_norm": 2.753124713897705,
|
| 5845 |
+
"learning_rate": 3.6650246305418725e-05,
|
| 5846 |
+
"loss": 0.2504,
|
| 5847 |
+
"step": 16680
|
| 5848 |
+
},
|
| 5849 |
+
{
|
| 5850 |
+
"epoch": 81.5131177547285,
|
| 5851 |
+
"grad_norm": 2.294701099395752,
|
| 5852 |
+
"learning_rate": 3.645320197044335e-05,
|
| 5853 |
+
"loss": 0.2433,
|
| 5854 |
+
"step": 16700
|
| 5855 |
+
},
|
| 5856 |
+
{
|
| 5857 |
+
"epoch": 81.61073825503355,
|
| 5858 |
+
"grad_norm": 2.179985761642456,
|
| 5859 |
+
"learning_rate": 3.625615763546798e-05,
|
| 5860 |
+
"loss": 0.2511,
|
| 5861 |
+
"step": 16720
|
| 5862 |
+
},
|
| 5863 |
+
{
|
| 5864 |
+
"epoch": 81.70835875533862,
|
| 5865 |
+
"grad_norm": 2.242023229598999,
|
| 5866 |
+
"learning_rate": 3.605911330049261e-05,
|
| 5867 |
+
"loss": 0.2558,
|
| 5868 |
+
"step": 16740
|
| 5869 |
+
},
|
| 5870 |
+
{
|
| 5871 |
+
"epoch": 81.80597925564369,
|
| 5872 |
+
"grad_norm": 2.9500415325164795,
|
| 5873 |
+
"learning_rate": 3.586206896551724e-05,
|
| 5874 |
+
"loss": 0.2423,
|
| 5875 |
+
"step": 16760
|
| 5876 |
+
},
|
| 5877 |
+
{
|
| 5878 |
+
"epoch": 81.90359975594875,
|
| 5879 |
+
"grad_norm": 2.372332811355591,
|
| 5880 |
+
"learning_rate": 3.5665024630541874e-05,
|
| 5881 |
+
"loss": 0.2503,
|
| 5882 |
+
"step": 16780
|
| 5883 |
+
},
|
| 5884 |
+
{
|
| 5885 |
+
"epoch": 82.00122025625382,
|
| 5886 |
+
"grad_norm": 2.8338615894317627,
|
| 5887 |
+
"learning_rate": 3.546798029556651e-05,
|
| 5888 |
+
"loss": 0.2442,
|
| 5889 |
+
"step": 16800
|
| 5890 |
+
},
|
| 5891 |
+
{
|
| 5892 |
+
"epoch": 82.09884075655887,
|
| 5893 |
+
"grad_norm": 2.5122156143188477,
|
| 5894 |
+
"learning_rate": 3.5270935960591135e-05,
|
| 5895 |
+
"loss": 0.2386,
|
| 5896 |
+
"step": 16820
|
| 5897 |
+
},
|
| 5898 |
+
{
|
| 5899 |
+
"epoch": 82.19646125686394,
|
| 5900 |
+
"grad_norm": 2.6733508110046387,
|
| 5901 |
+
"learning_rate": 3.507389162561577e-05,
|
| 5902 |
+
"loss": 0.2376,
|
| 5903 |
+
"step": 16840
|
| 5904 |
+
},
|
| 5905 |
+
{
|
| 5906 |
+
"epoch": 82.29408175716901,
|
| 5907 |
+
"grad_norm": 1.9639496803283691,
|
| 5908 |
+
"learning_rate": 3.4876847290640396e-05,
|
| 5909 |
+
"loss": 0.2366,
|
| 5910 |
+
"step": 16860
|
| 5911 |
+
},
|
| 5912 |
+
{
|
| 5913 |
+
"epoch": 82.39170225747407,
|
| 5914 |
+
"grad_norm": 2.2403128147125244,
|
| 5915 |
+
"learning_rate": 3.467980295566502e-05,
|
| 5916 |
+
"loss": 0.2478,
|
| 5917 |
+
"step": 16880
|
| 5918 |
+
},
|
| 5919 |
+
{
|
| 5920 |
+
"epoch": 82.48932275777914,
|
| 5921 |
+
"grad_norm": 2.3874387741088867,
|
| 5922 |
+
"learning_rate": 3.4482758620689657e-05,
|
| 5923 |
+
"loss": 0.2561,
|
| 5924 |
+
"step": 16900
|
| 5925 |
+
},
|
| 5926 |
+
{
|
| 5927 |
+
"epoch": 82.5869432580842,
|
| 5928 |
+
"grad_norm": 3.6774182319641113,
|
| 5929 |
+
"learning_rate": 3.428571428571429e-05,
|
| 5930 |
+
"loss": 0.2448,
|
| 5931 |
+
"step": 16920
|
| 5932 |
+
},
|
| 5933 |
+
{
|
| 5934 |
+
"epoch": 82.68456375838926,
|
| 5935 |
+
"grad_norm": 1.8325834274291992,
|
| 5936 |
+
"learning_rate": 3.408866995073892e-05,
|
| 5937 |
+
"loss": 0.2515,
|
| 5938 |
+
"step": 16940
|
| 5939 |
+
},
|
| 5940 |
+
{
|
| 5941 |
+
"epoch": 82.78218425869433,
|
| 5942 |
+
"grad_norm": 2.846112012863159,
|
| 5943 |
+
"learning_rate": 3.389162561576355e-05,
|
| 5944 |
+
"loss": 0.2481,
|
| 5945 |
+
"step": 16960
|
| 5946 |
+
},
|
| 5947 |
+
{
|
| 5948 |
+
"epoch": 82.87980475899938,
|
| 5949 |
+
"grad_norm": 3.7636115550994873,
|
| 5950 |
+
"learning_rate": 3.369458128078818e-05,
|
| 5951 |
+
"loss": 0.2529,
|
| 5952 |
+
"step": 16980
|
| 5953 |
+
},
|
| 5954 |
+
{
|
| 5955 |
+
"epoch": 82.97742525930445,
|
| 5956 |
+
"grad_norm": 2.4501962661743164,
|
| 5957 |
+
"learning_rate": 3.3497536945812806e-05,
|
| 5958 |
+
"loss": 0.2344,
|
| 5959 |
+
"step": 17000
|
| 5960 |
+
},
|
| 5961 |
+
{
|
| 5962 |
+
"epoch": 83.07504575960952,
|
| 5963 |
+
"grad_norm": 2.4377410411834717,
|
| 5964 |
+
"learning_rate": 3.330049261083744e-05,
|
| 5965 |
+
"loss": 0.2373,
|
| 5966 |
+
"step": 17020
|
| 5967 |
+
},
|
| 5968 |
+
{
|
| 5969 |
+
"epoch": 83.17266625991458,
|
| 5970 |
+
"grad_norm": 2.180765151977539,
|
| 5971 |
+
"learning_rate": 3.310344827586207e-05,
|
| 5972 |
+
"loss": 0.2395,
|
| 5973 |
+
"step": 17040
|
| 5974 |
+
},
|
| 5975 |
+
{
|
| 5976 |
+
"epoch": 83.27028676021965,
|
| 5977 |
+
"grad_norm": 3.2704169750213623,
|
| 5978 |
+
"learning_rate": 3.29064039408867e-05,
|
| 5979 |
+
"loss": 0.2407,
|
| 5980 |
+
"step": 17060
|
| 5981 |
+
},
|
| 5982 |
+
{
|
| 5983 |
+
"epoch": 83.36790726052472,
|
| 5984 |
+
"grad_norm": 2.74991512298584,
|
| 5985 |
+
"learning_rate": 3.2709359605911334e-05,
|
| 5986 |
+
"loss": 0.2351,
|
| 5987 |
+
"step": 17080
|
| 5988 |
+
},
|
| 5989 |
+
{
|
| 5990 |
+
"epoch": 83.46552776082977,
|
| 5991 |
+
"grad_norm": 1.780633807182312,
|
| 5992 |
+
"learning_rate": 3.251231527093596e-05,
|
| 5993 |
+
"loss": 0.2379,
|
| 5994 |
+
"step": 17100
|
| 5995 |
+
},
|
| 5996 |
+
{
|
| 5997 |
+
"epoch": 83.56314826113484,
|
| 5998 |
+
"grad_norm": 2.352802038192749,
|
| 5999 |
+
"learning_rate": 3.2315270935960595e-05,
|
| 6000 |
+
"loss": 0.244,
|
| 6001 |
+
"step": 17120
|
| 6002 |
+
},
|
| 6003 |
+
{
|
| 6004 |
+
"epoch": 83.6607687614399,
|
| 6005 |
+
"grad_norm": 3.505608320236206,
|
| 6006 |
+
"learning_rate": 3.211822660098522e-05,
|
| 6007 |
+
"loss": 0.2443,
|
| 6008 |
+
"step": 17140
|
| 6009 |
+
},
|
| 6010 |
+
{
|
| 6011 |
+
"epoch": 83.75838926174497,
|
| 6012 |
+
"grad_norm": 2.568233013153076,
|
| 6013 |
+
"learning_rate": 3.192118226600985e-05,
|
| 6014 |
+
"loss": 0.2499,
|
| 6015 |
+
"step": 17160
|
| 6016 |
+
},
|
| 6017 |
+
{
|
| 6018 |
+
"epoch": 83.85600976205004,
|
| 6019 |
+
"grad_norm": 1.864367961883545,
|
| 6020 |
+
"learning_rate": 3.172413793103448e-05,
|
| 6021 |
+
"loss": 0.2543,
|
| 6022 |
+
"step": 17180
|
| 6023 |
+
},
|
| 6024 |
+
{
|
| 6025 |
+
"epoch": 83.95363026235509,
|
| 6026 |
+
"grad_norm": 2.386052370071411,
|
| 6027 |
+
"learning_rate": 3.152709359605912e-05,
|
| 6028 |
+
"loss": 0.2505,
|
| 6029 |
+
"step": 17200
|
| 6030 |
+
},
|
| 6031 |
+
{
|
| 6032 |
+
"epoch": 84.05125076266016,
|
| 6033 |
+
"grad_norm": 4.361128330230713,
|
| 6034 |
+
"learning_rate": 3.1330049261083744e-05,
|
| 6035 |
+
"loss": 0.2505,
|
| 6036 |
+
"step": 17220
|
| 6037 |
+
},
|
| 6038 |
+
{
|
| 6039 |
+
"epoch": 84.14887126296523,
|
| 6040 |
+
"grad_norm": 1.4861139059066772,
|
| 6041 |
+
"learning_rate": 3.113300492610838e-05,
|
| 6042 |
+
"loss": 0.2314,
|
| 6043 |
+
"step": 17240
|
| 6044 |
+
},
|
| 6045 |
+
{
|
| 6046 |
+
"epoch": 84.24649176327028,
|
| 6047 |
+
"grad_norm": 1.9692414999008179,
|
| 6048 |
+
"learning_rate": 3.0935960591133005e-05,
|
| 6049 |
+
"loss": 0.2499,
|
| 6050 |
+
"step": 17260
|
| 6051 |
+
},
|
| 6052 |
+
{
|
| 6053 |
+
"epoch": 84.34411226357535,
|
| 6054 |
+
"grad_norm": 2.245277166366577,
|
| 6055 |
+
"learning_rate": 3.073891625615763e-05,
|
| 6056 |
+
"loss": 0.243,
|
| 6057 |
+
"step": 17280
|
| 6058 |
+
},
|
| 6059 |
+
{
|
| 6060 |
+
"epoch": 84.44173276388041,
|
| 6061 |
+
"grad_norm": 2.0669002532958984,
|
| 6062 |
+
"learning_rate": 3.0541871921182266e-05,
|
| 6063 |
+
"loss": 0.2388,
|
| 6064 |
+
"step": 17300
|
| 6065 |
+
},
|
| 6066 |
+
{
|
| 6067 |
+
"epoch": 84.53935326418548,
|
| 6068 |
+
"grad_norm": 2.377110004425049,
|
| 6069 |
+
"learning_rate": 3.0344827586206897e-05,
|
| 6070 |
+
"loss": 0.2431,
|
| 6071 |
+
"step": 17320
|
| 6072 |
+
},
|
| 6073 |
+
{
|
| 6074 |
+
"epoch": 84.63697376449055,
|
| 6075 |
+
"grad_norm": 2.4260573387145996,
|
| 6076 |
+
"learning_rate": 3.0147783251231527e-05,
|
| 6077 |
+
"loss": 0.2393,
|
| 6078 |
+
"step": 17340
|
| 6079 |
+
},
|
| 6080 |
+
{
|
| 6081 |
+
"epoch": 84.7345942647956,
|
| 6082 |
+
"grad_norm": 1.7577930688858032,
|
| 6083 |
+
"learning_rate": 2.995073891625616e-05,
|
| 6084 |
+
"loss": 0.2444,
|
| 6085 |
+
"step": 17360
|
| 6086 |
+
},
|
| 6087 |
+
{
|
| 6088 |
+
"epoch": 84.83221476510067,
|
| 6089 |
+
"grad_norm": 2.4844295978546143,
|
| 6090 |
+
"learning_rate": 2.9753694581280788e-05,
|
| 6091 |
+
"loss": 0.2474,
|
| 6092 |
+
"step": 17380
|
| 6093 |
+
},
|
| 6094 |
+
{
|
| 6095 |
+
"epoch": 84.92983526540573,
|
| 6096 |
+
"grad_norm": 2.7530508041381836,
|
| 6097 |
+
"learning_rate": 2.9556650246305422e-05,
|
| 6098 |
+
"loss": 0.2459,
|
| 6099 |
+
"step": 17400
|
| 6100 |
+
},
|
| 6101 |
+
{
|
| 6102 |
+
"epoch": 85.0274557657108,
|
| 6103 |
+
"grad_norm": 1.6418040990829468,
|
| 6104 |
+
"learning_rate": 2.9359605911330052e-05,
|
| 6105 |
+
"loss": 0.2491,
|
| 6106 |
+
"step": 17420
|
| 6107 |
+
},
|
| 6108 |
+
{
|
| 6109 |
+
"epoch": 85.12507626601587,
|
| 6110 |
+
"grad_norm": 2.0329489707946777,
|
| 6111 |
+
"learning_rate": 2.916256157635468e-05,
|
| 6112 |
+
"loss": 0.2426,
|
| 6113 |
+
"step": 17440
|
| 6114 |
+
},
|
| 6115 |
+
{
|
| 6116 |
+
"epoch": 85.22269676632092,
|
| 6117 |
+
"grad_norm": 1.6439207792282104,
|
| 6118 |
+
"learning_rate": 2.8965517241379313e-05,
|
| 6119 |
+
"loss": 0.2351,
|
| 6120 |
+
"step": 17460
|
| 6121 |
+
},
|
| 6122 |
+
{
|
| 6123 |
+
"epoch": 85.32031726662599,
|
| 6124 |
+
"grad_norm": 1.6182892322540283,
|
| 6125 |
+
"learning_rate": 2.8768472906403944e-05,
|
| 6126 |
+
"loss": 0.2468,
|
| 6127 |
+
"step": 17480
|
| 6128 |
+
},
|
| 6129 |
+
{
|
| 6130 |
+
"epoch": 85.41793776693106,
|
| 6131 |
+
"grad_norm": 3.263887882232666,
|
| 6132 |
+
"learning_rate": 2.857142857142857e-05,
|
| 6133 |
+
"loss": 0.2426,
|
| 6134 |
+
"step": 17500
|
| 6135 |
+
},
|
| 6136 |
+
{
|
| 6137 |
+
"epoch": 85.51555826723612,
|
| 6138 |
+
"grad_norm": 3.062742233276367,
|
| 6139 |
+
"learning_rate": 2.8374384236453205e-05,
|
| 6140 |
+
"loss": 0.2386,
|
| 6141 |
+
"step": 17520
|
| 6142 |
+
},
|
| 6143 |
+
{
|
| 6144 |
+
"epoch": 85.61317876754119,
|
| 6145 |
+
"grad_norm": 2.8203582763671875,
|
| 6146 |
+
"learning_rate": 2.8177339901477835e-05,
|
| 6147 |
+
"loss": 0.2407,
|
| 6148 |
+
"step": 17540
|
| 6149 |
+
},
|
| 6150 |
+
{
|
| 6151 |
+
"epoch": 85.71079926784624,
|
| 6152 |
+
"grad_norm": 2.3993334770202637,
|
| 6153 |
+
"learning_rate": 2.7980295566502462e-05,
|
| 6154 |
+
"loss": 0.2418,
|
| 6155 |
+
"step": 17560
|
| 6156 |
+
},
|
| 6157 |
+
{
|
| 6158 |
+
"epoch": 85.80841976815131,
|
| 6159 |
+
"grad_norm": 1.7914482355117798,
|
| 6160 |
+
"learning_rate": 2.7783251231527096e-05,
|
| 6161 |
+
"loss": 0.2377,
|
| 6162 |
+
"step": 17580
|
| 6163 |
+
},
|
| 6164 |
+
{
|
| 6165 |
+
"epoch": 85.90604026845638,
|
| 6166 |
+
"grad_norm": 3.20501971244812,
|
| 6167 |
+
"learning_rate": 2.7586206896551727e-05,
|
| 6168 |
+
"loss": 0.2398,
|
| 6169 |
+
"step": 17600
|
| 6170 |
+
},
|
| 6171 |
+
{
|
| 6172 |
+
"epoch": 86.00366076876143,
|
| 6173 |
+
"grad_norm": 1.6623684167861938,
|
| 6174 |
+
"learning_rate": 2.7389162561576354e-05,
|
| 6175 |
+
"loss": 0.2442,
|
| 6176 |
+
"step": 17620
|
| 6177 |
+
},
|
| 6178 |
+
{
|
| 6179 |
+
"epoch": 86.1012812690665,
|
| 6180 |
+
"grad_norm": 2.3433034420013428,
|
| 6181 |
+
"learning_rate": 2.7192118226600988e-05,
|
| 6182 |
+
"loss": 0.2358,
|
| 6183 |
+
"step": 17640
|
| 6184 |
+
},
|
| 6185 |
+
{
|
| 6186 |
+
"epoch": 86.19890176937157,
|
| 6187 |
+
"grad_norm": 2.6188597679138184,
|
| 6188 |
+
"learning_rate": 2.6995073891625615e-05,
|
| 6189 |
+
"loss": 0.2336,
|
| 6190 |
+
"step": 17660
|
| 6191 |
+
},
|
| 6192 |
+
{
|
| 6193 |
+
"epoch": 86.29652226967663,
|
| 6194 |
+
"grad_norm": 3.1089391708374023,
|
| 6195 |
+
"learning_rate": 2.6798029556650245e-05,
|
| 6196 |
+
"loss": 0.239,
|
| 6197 |
+
"step": 17680
|
| 6198 |
+
},
|
| 6199 |
+
{
|
| 6200 |
+
"epoch": 86.3941427699817,
|
| 6201 |
+
"grad_norm": 2.378998041152954,
|
| 6202 |
+
"learning_rate": 2.660098522167488e-05,
|
| 6203 |
+
"loss": 0.2336,
|
| 6204 |
+
"step": 17700
|
| 6205 |
+
},
|
| 6206 |
+
{
|
| 6207 |
+
"epoch": 86.49176327028675,
|
| 6208 |
+
"grad_norm": 2.4956347942352295,
|
| 6209 |
+
"learning_rate": 2.6403940886699506e-05,
|
| 6210 |
+
"loss": 0.2497,
|
| 6211 |
+
"step": 17720
|
| 6212 |
+
},
|
| 6213 |
+
{
|
| 6214 |
+
"epoch": 86.58938377059182,
|
| 6215 |
+
"grad_norm": 2.529139757156372,
|
| 6216 |
+
"learning_rate": 2.620689655172414e-05,
|
| 6217 |
+
"loss": 0.2436,
|
| 6218 |
+
"step": 17740
|
| 6219 |
+
},
|
| 6220 |
+
{
|
| 6221 |
+
"epoch": 86.68700427089689,
|
| 6222 |
+
"grad_norm": 2.6899948120117188,
|
| 6223 |
+
"learning_rate": 2.600985221674877e-05,
|
| 6224 |
+
"loss": 0.2445,
|
| 6225 |
+
"step": 17760
|
| 6226 |
+
},
|
| 6227 |
+
{
|
| 6228 |
+
"epoch": 86.78462477120195,
|
| 6229 |
+
"grad_norm": 1.8922455310821533,
|
| 6230 |
+
"learning_rate": 2.5812807881773398e-05,
|
| 6231 |
+
"loss": 0.2366,
|
| 6232 |
+
"step": 17780
|
| 6233 |
+
},
|
| 6234 |
+
{
|
| 6235 |
+
"epoch": 86.88224527150702,
|
| 6236 |
+
"grad_norm": 1.9104729890823364,
|
| 6237 |
+
"learning_rate": 2.561576354679803e-05,
|
| 6238 |
+
"loss": 0.2345,
|
| 6239 |
+
"step": 17800
|
| 6240 |
+
},
|
| 6241 |
+
{
|
| 6242 |
+
"epoch": 86.97986577181209,
|
| 6243 |
+
"grad_norm": 3.2369461059570312,
|
| 6244 |
+
"learning_rate": 2.5418719211822662e-05,
|
| 6245 |
+
"loss": 0.2515,
|
| 6246 |
+
"step": 17820
|
| 6247 |
+
},
|
| 6248 |
+
{
|
| 6249 |
+
"epoch": 87.07748627211714,
|
| 6250 |
+
"grad_norm": 2.2592508792877197,
|
| 6251 |
+
"learning_rate": 2.522167487684729e-05,
|
| 6252 |
+
"loss": 0.2333,
|
| 6253 |
+
"step": 17840
|
| 6254 |
+
},
|
| 6255 |
+
{
|
| 6256 |
+
"epoch": 87.17510677242221,
|
| 6257 |
+
"grad_norm": 2.302445888519287,
|
| 6258 |
+
"learning_rate": 2.5024630541871923e-05,
|
| 6259 |
+
"loss": 0.2308,
|
| 6260 |
+
"step": 17860
|
| 6261 |
+
},
|
| 6262 |
+
{
|
| 6263 |
+
"epoch": 87.27272727272727,
|
| 6264 |
+
"grad_norm": 2.0607619285583496,
|
| 6265 |
+
"learning_rate": 2.4827586206896553e-05,
|
| 6266 |
+
"loss": 0.2323,
|
| 6267 |
+
"step": 17880
|
| 6268 |
+
},
|
| 6269 |
+
{
|
| 6270 |
+
"epoch": 87.37034777303234,
|
| 6271 |
+
"grad_norm": 2.4503376483917236,
|
| 6272 |
+
"learning_rate": 2.4630541871921184e-05,
|
| 6273 |
+
"loss": 0.2399,
|
| 6274 |
+
"step": 17900
|
| 6275 |
+
},
|
| 6276 |
+
{
|
| 6277 |
+
"epoch": 87.4679682733374,
|
| 6278 |
+
"grad_norm": 1.7061033248901367,
|
| 6279 |
+
"learning_rate": 2.4433497536945814e-05,
|
| 6280 |
+
"loss": 0.249,
|
| 6281 |
+
"step": 17920
|
| 6282 |
+
},
|
| 6283 |
+
{
|
| 6284 |
+
"epoch": 87.56558877364246,
|
| 6285 |
+
"grad_norm": 2.1557867527008057,
|
| 6286 |
+
"learning_rate": 2.4236453201970445e-05,
|
| 6287 |
+
"loss": 0.243,
|
| 6288 |
+
"step": 17940
|
| 6289 |
+
},
|
| 6290 |
+
{
|
| 6291 |
+
"epoch": 87.66320927394753,
|
| 6292 |
+
"grad_norm": 2.0752928256988525,
|
| 6293 |
+
"learning_rate": 2.4039408866995075e-05,
|
| 6294 |
+
"loss": 0.236,
|
| 6295 |
+
"step": 17960
|
| 6296 |
+
},
|
| 6297 |
+
{
|
| 6298 |
+
"epoch": 87.7608297742526,
|
| 6299 |
+
"grad_norm": 1.9939770698547363,
|
| 6300 |
+
"learning_rate": 2.3842364532019706e-05,
|
| 6301 |
+
"loss": 0.24,
|
| 6302 |
+
"step": 17980
|
| 6303 |
+
},
|
| 6304 |
+
{
|
| 6305 |
+
"epoch": 87.85845027455765,
|
| 6306 |
+
"grad_norm": 2.043842315673828,
|
| 6307 |
+
"learning_rate": 2.3645320197044336e-05,
|
| 6308 |
+
"loss": 0.2438,
|
| 6309 |
+
"step": 18000
|
| 6310 |
}
|
| 6311 |
],
|
| 6312 |
"logging_steps": 20,
|
|
|
|
| 6326 |
"attributes": {}
|
| 6327 |
}
|
| 6328 |
},
|
| 6329 |
+
"total_flos": 3.586273126839091e+17,
|
| 6330 |
"train_batch_size": 1,
|
| 6331 |
"trial_name": null,
|
| 6332 |
"trial_params": null
|