Training in progress, step 958, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 48679352
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4f5a00a78e423620e021fbd8a144680909f3b3b316b4db72cc2433120e14140f
|
3 |
size 48679352
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 25152884
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:00d91609100f05aa7550e6c90fd1e83339d641e8c40058a4b6a728b366749b3a
|
3 |
size 25152884
|
last-checkpoint/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:aa34e95f3b762e82210d69d9159af8a97768bdd66bdbac85961d502cc41f3157
|
3 |
size 14244
|
last-checkpoint/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:fafd9a7c6dbc3f71076a7b5b864e1667067c311eadec3f4837b9340812da01f0
|
3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 0
|
5 |
"eval_steps": 240,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -5079,6 +5079,1672 @@
|
|
5079 |
"eval_samples_per_second": 19.864,
|
5080 |
"eval_steps_per_second": 9.932,
|
5081 |
"step": 720
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5082 |
}
|
5083 |
],
|
5084 |
"logging_steps": 1,
|
@@ -5093,12 +6759,12 @@
|
|
5093 |
"should_evaluate": false,
|
5094 |
"should_log": false,
|
5095 |
"should_save": true,
|
5096 |
-
"should_training_stop":
|
5097 |
},
|
5098 |
"attributes": {}
|
5099 |
}
|
5100 |
},
|
5101 |
-
"total_flos":
|
5102 |
"train_batch_size": 2,
|
5103 |
"trial_name": null,
|
5104 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.0,
|
5 |
"eval_steps": 240,
|
6 |
+
"global_step": 958,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
5079 |
"eval_samples_per_second": 19.864,
|
5080 |
"eval_steps_per_second": 9.932,
|
5081 |
"step": 720
|
5082 |
+
},
|
5083 |
+
{
|
5084 |
+
"epoch": 0.7526096033402923,
|
5085 |
+
"grad_norm": 0.43510791659355164,
|
5086 |
+
"learning_rate": 2.9289321881345254e-05,
|
5087 |
+
"loss": 0.1719,
|
5088 |
+
"step": 721
|
5089 |
+
},
|
5090 |
+
{
|
5091 |
+
"epoch": 0.7536534446764092,
|
5092 |
+
"grad_norm": 0.4775535762310028,
|
5093 |
+
"learning_rate": 2.905538131479376e-05,
|
5094 |
+
"loss": 0.1874,
|
5095 |
+
"step": 722
|
5096 |
+
},
|
5097 |
+
{
|
5098 |
+
"epoch": 0.7546972860125261,
|
5099 |
+
"grad_norm": 0.5398250818252563,
|
5100 |
+
"learning_rate": 2.8822219864258272e-05,
|
5101 |
+
"loss": 0.1667,
|
5102 |
+
"step": 723
|
5103 |
+
},
|
5104 |
+
{
|
5105 |
+
"epoch": 0.755741127348643,
|
5106 |
+
"grad_norm": 0.4013484716415405,
|
5107 |
+
"learning_rate": 2.8589840090325027e-05,
|
5108 |
+
"loss": 0.1252,
|
5109 |
+
"step": 724
|
5110 |
+
},
|
5111 |
+
{
|
5112 |
+
"epoch": 0.7567849686847599,
|
5113 |
+
"grad_norm": 0.8314927816390991,
|
5114 |
+
"learning_rate": 2.8358244544996038e-05,
|
5115 |
+
"loss": 0.3924,
|
5116 |
+
"step": 725
|
5117 |
+
},
|
5118 |
+
{
|
5119 |
+
"epoch": 0.7578288100208769,
|
5120 |
+
"grad_norm": 0.8482892513275146,
|
5121 |
+
"learning_rate": 2.8127435771660747e-05,
|
5122 |
+
"loss": 0.3675,
|
5123 |
+
"step": 726
|
5124 |
+
},
|
5125 |
+
{
|
5126 |
+
"epoch": 0.7588726513569938,
|
5127 |
+
"grad_norm": 0.6500058770179749,
|
5128 |
+
"learning_rate": 2.7897416305068323e-05,
|
5129 |
+
"loss": 0.3016,
|
5130 |
+
"step": 727
|
5131 |
+
},
|
5132 |
+
{
|
5133 |
+
"epoch": 0.7599164926931107,
|
5134 |
+
"grad_norm": 0.6976576447486877,
|
5135 |
+
"learning_rate": 2.7668188671299755e-05,
|
5136 |
+
"loss": 0.3133,
|
5137 |
+
"step": 728
|
5138 |
+
},
|
5139 |
+
{
|
5140 |
+
"epoch": 0.7609603340292276,
|
5141 |
+
"grad_norm": 0.7735400795936584,
|
5142 |
+
"learning_rate": 2.743975538774002e-05,
|
5143 |
+
"loss": 0.3557,
|
5144 |
+
"step": 729
|
5145 |
+
},
|
5146 |
+
{
|
5147 |
+
"epoch": 0.7620041753653445,
|
5148 |
+
"grad_norm": 0.7905307412147522,
|
5149 |
+
"learning_rate": 2.7212118963050592e-05,
|
5150 |
+
"loss": 0.3316,
|
5151 |
+
"step": 730
|
5152 |
+
},
|
5153 |
+
{
|
5154 |
+
"epoch": 0.7630480167014614,
|
5155 |
+
"grad_norm": 0.7098974585533142,
|
5156 |
+
"learning_rate": 2.6985281897141812e-05,
|
5157 |
+
"loss": 0.2877,
|
5158 |
+
"step": 731
|
5159 |
+
},
|
5160 |
+
{
|
5161 |
+
"epoch": 0.7640918580375783,
|
5162 |
+
"grad_norm": 0.6374404430389404,
|
5163 |
+
"learning_rate": 2.675924668114537e-05,
|
5164 |
+
"loss": 0.2589,
|
5165 |
+
"step": 732
|
5166 |
+
},
|
5167 |
+
{
|
5168 |
+
"epoch": 0.7651356993736952,
|
5169 |
+
"grad_norm": 0.7561594247817993,
|
5170 |
+
"learning_rate": 2.65340157973871e-05,
|
5171 |
+
"loss": 0.2953,
|
5172 |
+
"step": 733
|
5173 |
+
},
|
5174 |
+
{
|
5175 |
+
"epoch": 0.7661795407098121,
|
5176 |
+
"grad_norm": 0.7232580780982971,
|
5177 |
+
"learning_rate": 2.630959171935956e-05,
|
5178 |
+
"loss": 0.3002,
|
5179 |
+
"step": 734
|
5180 |
+
},
|
5181 |
+
{
|
5182 |
+
"epoch": 0.767223382045929,
|
5183 |
+
"grad_norm": 0.7140358686447144,
|
5184 |
+
"learning_rate": 2.6085976911694987e-05,
|
5185 |
+
"loss": 0.2545,
|
5186 |
+
"step": 735
|
5187 |
+
},
|
5188 |
+
{
|
5189 |
+
"epoch": 0.7682672233820459,
|
5190 |
+
"grad_norm": 0.749450147151947,
|
5191 |
+
"learning_rate": 2.586317383013821e-05,
|
5192 |
+
"loss": 0.25,
|
5193 |
+
"step": 736
|
5194 |
+
},
|
5195 |
+
{
|
5196 |
+
"epoch": 0.7693110647181628,
|
5197 |
+
"grad_norm": 0.847427248954773,
|
5198 |
+
"learning_rate": 2.564118492151957e-05,
|
5199 |
+
"loss": 0.2887,
|
5200 |
+
"step": 737
|
5201 |
+
},
|
5202 |
+
{
|
5203 |
+
"epoch": 0.7703549060542797,
|
5204 |
+
"grad_norm": 0.5772815346717834,
|
5205 |
+
"learning_rate": 2.5420012623728208e-05,
|
5206 |
+
"loss": 0.2101,
|
5207 |
+
"step": 738
|
5208 |
+
},
|
5209 |
+
{
|
5210 |
+
"epoch": 0.7713987473903967,
|
5211 |
+
"grad_norm": 0.6374946236610413,
|
5212 |
+
"learning_rate": 2.5199659365685235e-05,
|
5213 |
+
"loss": 0.2193,
|
5214 |
+
"step": 739
|
5215 |
+
},
|
5216 |
+
{
|
5217 |
+
"epoch": 0.7724425887265136,
|
5218 |
+
"grad_norm": 0.642776370048523,
|
5219 |
+
"learning_rate": 2.4980127567316948e-05,
|
5220 |
+
"loss": 0.2168,
|
5221 |
+
"step": 740
|
5222 |
+
},
|
5223 |
+
{
|
5224 |
+
"epoch": 0.7734864300626305,
|
5225 |
+
"grad_norm": 0.6233210563659668,
|
5226 |
+
"learning_rate": 2.4761419639528437e-05,
|
5227 |
+
"loss": 0.1986,
|
5228 |
+
"step": 741
|
5229 |
+
},
|
5230 |
+
{
|
5231 |
+
"epoch": 0.7745302713987474,
|
5232 |
+
"grad_norm": 0.47856077551841736,
|
5233 |
+
"learning_rate": 2.4543537984176978e-05,
|
5234 |
+
"loss": 0.1681,
|
5235 |
+
"step": 742
|
5236 |
+
},
|
5237 |
+
{
|
5238 |
+
"epoch": 0.7755741127348643,
|
5239 |
+
"grad_norm": 0.5664119720458984,
|
5240 |
+
"learning_rate": 2.4326484994045752e-05,
|
5241 |
+
"loss": 0.1847,
|
5242 |
+
"step": 743
|
5243 |
+
},
|
5244 |
+
{
|
5245 |
+
"epoch": 0.7766179540709812,
|
5246 |
+
"grad_norm": 0.5382654666900635,
|
5247 |
+
"learning_rate": 2.4110263052817394e-05,
|
5248 |
+
"loss": 0.1765,
|
5249 |
+
"step": 744
|
5250 |
+
},
|
5251 |
+
{
|
5252 |
+
"epoch": 0.7776617954070981,
|
5253 |
+
"grad_norm": 0.46047693490982056,
|
5254 |
+
"learning_rate": 2.3894874535048063e-05,
|
5255 |
+
"loss": 0.1487,
|
5256 |
+
"step": 745
|
5257 |
+
},
|
5258 |
+
{
|
5259 |
+
"epoch": 0.778705636743215,
|
5260 |
+
"grad_norm": 0.5149843096733093,
|
5261 |
+
"learning_rate": 2.368032180614118e-05,
|
5262 |
+
"loss": 0.1817,
|
5263 |
+
"step": 746
|
5264 |
+
},
|
5265 |
+
{
|
5266 |
+
"epoch": 0.7797494780793319,
|
5267 |
+
"grad_norm": 0.4655948579311371,
|
5268 |
+
"learning_rate": 2.346660722232148e-05,
|
5269 |
+
"loss": 0.1548,
|
5270 |
+
"step": 747
|
5271 |
+
},
|
5272 |
+
{
|
5273 |
+
"epoch": 0.7807933194154488,
|
5274 |
+
"grad_norm": 0.5313979387283325,
|
5275 |
+
"learning_rate": 2.325373313060919e-05,
|
5276 |
+
"loss": 0.1504,
|
5277 |
+
"step": 748
|
5278 |
+
},
|
5279 |
+
{
|
5280 |
+
"epoch": 0.7818371607515657,
|
5281 |
+
"grad_norm": 0.5336787700653076,
|
5282 |
+
"learning_rate": 2.3041701868794287e-05,
|
5283 |
+
"loss": 0.1306,
|
5284 |
+
"step": 749
|
5285 |
+
},
|
5286 |
+
{
|
5287 |
+
"epoch": 0.7828810020876826,
|
5288 |
+
"grad_norm": 0.7393002510070801,
|
5289 |
+
"learning_rate": 2.2830515765410622e-05,
|
5290 |
+
"loss": 0.1137,
|
5291 |
+
"step": 750
|
5292 |
+
},
|
5293 |
+
{
|
5294 |
+
"epoch": 0.7839248434237995,
|
5295 |
+
"grad_norm": 0.5598475933074951,
|
5296 |
+
"learning_rate": 2.262017713971063e-05,
|
5297 |
+
"loss": 0.3354,
|
5298 |
+
"step": 751
|
5299 |
+
},
|
5300 |
+
{
|
5301 |
+
"epoch": 0.7849686847599165,
|
5302 |
+
"grad_norm": 0.5945207476615906,
|
5303 |
+
"learning_rate": 2.2410688301639616e-05,
|
5304 |
+
"loss": 0.2804,
|
5305 |
+
"step": 752
|
5306 |
+
},
|
5307 |
+
{
|
5308 |
+
"epoch": 0.7860125260960334,
|
5309 |
+
"grad_norm": 0.4905988574028015,
|
5310 |
+
"learning_rate": 2.2202051551810565e-05,
|
5311 |
+
"loss": 0.2229,
|
5312 |
+
"step": 753
|
5313 |
+
},
|
5314 |
+
{
|
5315 |
+
"epoch": 0.7870563674321504,
|
5316 |
+
"grad_norm": 0.607475996017456,
|
5317 |
+
"learning_rate": 2.19942691814788e-05,
|
5318 |
+
"loss": 0.289,
|
5319 |
+
"step": 754
|
5320 |
+
},
|
5321 |
+
{
|
5322 |
+
"epoch": 0.7881002087682673,
|
5323 |
+
"grad_norm": 0.6597141027450562,
|
5324 |
+
"learning_rate": 2.178734347251673e-05,
|
5325 |
+
"loss": 0.2965,
|
5326 |
+
"step": 755
|
5327 |
+
},
|
5328 |
+
{
|
5329 |
+
"epoch": 0.7891440501043842,
|
5330 |
+
"grad_norm": 0.6289554238319397,
|
5331 |
+
"learning_rate": 2.1581276697388975e-05,
|
5332 |
+
"loss": 0.2582,
|
5333 |
+
"step": 756
|
5334 |
+
},
|
5335 |
+
{
|
5336 |
+
"epoch": 0.7901878914405011,
|
5337 |
+
"grad_norm": 0.6147776246070862,
|
5338 |
+
"learning_rate": 2.1376071119127338e-05,
|
5339 |
+
"loss": 0.2848,
|
5340 |
+
"step": 757
|
5341 |
+
},
|
5342 |
+
{
|
5343 |
+
"epoch": 0.791231732776618,
|
5344 |
+
"grad_norm": 0.5135255455970764,
|
5345 |
+
"learning_rate": 2.1171728991305795e-05,
|
5346 |
+
"loss": 0.2306,
|
5347 |
+
"step": 758
|
5348 |
+
},
|
5349 |
+
{
|
5350 |
+
"epoch": 0.7922755741127349,
|
5351 |
+
"grad_norm": 0.5475291013717651,
|
5352 |
+
"learning_rate": 2.0968252558016055e-05,
|
5353 |
+
"loss": 0.2394,
|
5354 |
+
"step": 759
|
5355 |
+
},
|
5356 |
+
{
|
5357 |
+
"epoch": 0.7933194154488518,
|
5358 |
+
"grad_norm": 0.6322019696235657,
|
5359 |
+
"learning_rate": 2.076564405384258e-05,
|
5360 |
+
"loss": 0.2066,
|
5361 |
+
"step": 760
|
5362 |
+
},
|
5363 |
+
{
|
5364 |
+
"epoch": 0.7943632567849687,
|
5365 |
+
"grad_norm": 0.5838301181793213,
|
5366 |
+
"learning_rate": 2.0563905703838316e-05,
|
5367 |
+
"loss": 0.2321,
|
5368 |
+
"step": 761
|
5369 |
+
},
|
5370 |
+
{
|
5371 |
+
"epoch": 0.7954070981210856,
|
5372 |
+
"grad_norm": 0.47901853919029236,
|
5373 |
+
"learning_rate": 2.0363039723500156e-05,
|
5374 |
+
"loss": 0.1445,
|
5375 |
+
"step": 762
|
5376 |
+
},
|
5377 |
+
{
|
5378 |
+
"epoch": 0.7964509394572025,
|
5379 |
+
"grad_norm": 0.6136653423309326,
|
5380 |
+
"learning_rate": 2.0163048318744493e-05,
|
5381 |
+
"loss": 0.2168,
|
5382 |
+
"step": 763
|
5383 |
+
},
|
5384 |
+
{
|
5385 |
+
"epoch": 0.7974947807933194,
|
5386 |
+
"grad_norm": 0.49966001510620117,
|
5387 |
+
"learning_rate": 1.9963933685883253e-05,
|
5388 |
+
"loss": 0.1414,
|
5389 |
+
"step": 764
|
5390 |
+
},
|
5391 |
+
{
|
5392 |
+
"epoch": 0.7985386221294363,
|
5393 |
+
"grad_norm": 0.5253435373306274,
|
5394 |
+
"learning_rate": 1.9765698011599466e-05,
|
5395 |
+
"loss": 0.1513,
|
5396 |
+
"step": 765
|
5397 |
+
},
|
5398 |
+
{
|
5399 |
+
"epoch": 0.7995824634655533,
|
5400 |
+
"grad_norm": 0.3400777578353882,
|
5401 |
+
"learning_rate": 1.9568343472923524e-05,
|
5402 |
+
"loss": 0.1112,
|
5403 |
+
"step": 766
|
5404 |
+
},
|
5405 |
+
{
|
5406 |
+
"epoch": 0.8006263048016702,
|
5407 |
+
"grad_norm": 0.5851226449012756,
|
5408 |
+
"learning_rate": 1.9371872237209165e-05,
|
5409 |
+
"loss": 0.1619,
|
5410 |
+
"step": 767
|
5411 |
+
},
|
5412 |
+
{
|
5413 |
+
"epoch": 0.8016701461377871,
|
5414 |
+
"grad_norm": 0.30553382635116577,
|
5415 |
+
"learning_rate": 1.917628646210957e-05,
|
5416 |
+
"loss": 0.0872,
|
5417 |
+
"step": 768
|
5418 |
+
},
|
5419 |
+
{
|
5420 |
+
"epoch": 0.802713987473904,
|
5421 |
+
"grad_norm": 0.5676819682121277,
|
5422 |
+
"learning_rate": 1.8981588295553853e-05,
|
5423 |
+
"loss": 0.2251,
|
5424 |
+
"step": 769
|
5425 |
+
},
|
5426 |
+
{
|
5427 |
+
"epoch": 0.8037578288100209,
|
5428 |
+
"grad_norm": 0.5648460984230042,
|
5429 |
+
"learning_rate": 1.878777987572339e-05,
|
5430 |
+
"loss": 0.2181,
|
5431 |
+
"step": 770
|
5432 |
+
},
|
5433 |
+
{
|
5434 |
+
"epoch": 0.8048016701461378,
|
5435 |
+
"grad_norm": 0.607913076877594,
|
5436 |
+
"learning_rate": 1.8594863331028224e-05,
|
5437 |
+
"loss": 0.2216,
|
5438 |
+
"step": 771
|
5439 |
+
},
|
5440 |
+
{
|
5441 |
+
"epoch": 0.8058455114822547,
|
5442 |
+
"grad_norm": 0.49821627140045166,
|
5443 |
+
"learning_rate": 1.840284078008393e-05,
|
5444 |
+
"loss": 0.1845,
|
5445 |
+
"step": 772
|
5446 |
+
},
|
5447 |
+
{
|
5448 |
+
"epoch": 0.8068893528183716,
|
5449 |
+
"grad_norm": 0.42891865968704224,
|
5450 |
+
"learning_rate": 1.821171433168809e-05,
|
5451 |
+
"loss": 0.1302,
|
5452 |
+
"step": 773
|
5453 |
+
},
|
5454 |
+
{
|
5455 |
+
"epoch": 0.8079331941544885,
|
5456 |
+
"grad_norm": 0.8176518678665161,
|
5457 |
+
"learning_rate": 1.8021486084797368e-05,
|
5458 |
+
"loss": 0.3043,
|
5459 |
+
"step": 774
|
5460 |
+
},
|
5461 |
+
{
|
5462 |
+
"epoch": 0.8089770354906054,
|
5463 |
+
"grad_norm": 0.6164413690567017,
|
5464 |
+
"learning_rate": 1.7832158128504328e-05,
|
5465 |
+
"loss": 0.2472,
|
5466 |
+
"step": 775
|
5467 |
+
},
|
5468 |
+
{
|
5469 |
+
"epoch": 0.8100208768267223,
|
5470 |
+
"grad_norm": 0.7422143220901489,
|
5471 |
+
"learning_rate": 1.7643732542014434e-05,
|
5472 |
+
"loss": 0.3395,
|
5473 |
+
"step": 776
|
5474 |
+
},
|
5475 |
+
{
|
5476 |
+
"epoch": 0.8110647181628392,
|
5477 |
+
"grad_norm": 0.7129305601119995,
|
5478 |
+
"learning_rate": 1.7456211394623378e-05,
|
5479 |
+
"loss": 0.2934,
|
5480 |
+
"step": 777
|
5481 |
+
},
|
5482 |
+
{
|
5483 |
+
"epoch": 0.8121085594989561,
|
5484 |
+
"grad_norm": 0.7638242840766907,
|
5485 |
+
"learning_rate": 1.7269596745694295e-05,
|
5486 |
+
"loss": 0.3399,
|
5487 |
+
"step": 778
|
5488 |
+
},
|
5489 |
+
{
|
5490 |
+
"epoch": 0.8131524008350731,
|
5491 |
+
"grad_norm": 0.624290943145752,
|
5492 |
+
"learning_rate": 1.7083890644635014e-05,
|
5493 |
+
"loss": 0.2525,
|
5494 |
+
"step": 779
|
5495 |
+
},
|
5496 |
+
{
|
5497 |
+
"epoch": 0.81419624217119,
|
5498 |
+
"grad_norm": 0.6761390566825867,
|
5499 |
+
"learning_rate": 1.6899095130875774e-05,
|
5500 |
+
"loss": 0.2753,
|
5501 |
+
"step": 780
|
5502 |
+
},
|
5503 |
+
{
|
5504 |
+
"epoch": 0.8152400835073069,
|
5505 |
+
"grad_norm": 0.8017570972442627,
|
5506 |
+
"learning_rate": 1.6715212233846655e-05,
|
5507 |
+
"loss": 0.3149,
|
5508 |
+
"step": 781
|
5509 |
+
},
|
5510 |
+
{
|
5511 |
+
"epoch": 0.8162839248434238,
|
5512 |
+
"grad_norm": 0.6984527111053467,
|
5513 |
+
"learning_rate": 1.6532243972955398e-05,
|
5514 |
+
"loss": 0.3103,
|
5515 |
+
"step": 782
|
5516 |
+
},
|
5517 |
+
{
|
5518 |
+
"epoch": 0.8173277661795407,
|
5519 |
+
"grad_norm": 0.6139991283416748,
|
5520 |
+
"learning_rate": 1.635019235756511e-05,
|
5521 |
+
"loss": 0.2274,
|
5522 |
+
"step": 783
|
5523 |
+
},
|
5524 |
+
{
|
5525 |
+
"epoch": 0.8183716075156576,
|
5526 |
+
"grad_norm": 0.7596770524978638,
|
5527 |
+
"learning_rate": 1.616905938697234e-05,
|
5528 |
+
"loss": 0.2612,
|
5529 |
+
"step": 784
|
5530 |
+
},
|
5531 |
+
{
|
5532 |
+
"epoch": 0.8194154488517745,
|
5533 |
+
"grad_norm": 0.685114324092865,
|
5534 |
+
"learning_rate": 1.5988847050385037e-05,
|
5535 |
+
"loss": 0.2542,
|
5536 |
+
"step": 785
|
5537 |
+
},
|
5538 |
+
{
|
5539 |
+
"epoch": 0.8204592901878914,
|
5540 |
+
"grad_norm": 0.6803342700004578,
|
5541 |
+
"learning_rate": 1.5809557326900647e-05,
|
5542 |
+
"loss": 0.2267,
|
5543 |
+
"step": 786
|
5544 |
+
},
|
5545 |
+
{
|
5546 |
+
"epoch": 0.8215031315240083,
|
5547 |
+
"grad_norm": 0.6194560527801514,
|
5548 |
+
"learning_rate": 1.5631192185484554e-05,
|
5549 |
+
"loss": 0.2029,
|
5550 |
+
"step": 787
|
5551 |
+
},
|
5552 |
+
{
|
5553 |
+
"epoch": 0.8225469728601252,
|
5554 |
+
"grad_norm": 0.8063942193984985,
|
5555 |
+
"learning_rate": 1.5453753584948328e-05,
|
5556 |
+
"loss": 0.2735,
|
5557 |
+
"step": 788
|
5558 |
+
},
|
5559 |
+
{
|
5560 |
+
"epoch": 0.8235908141962421,
|
5561 |
+
"grad_norm": 0.6718615293502808,
|
5562 |
+
"learning_rate": 1.527724347392815e-05,
|
5563 |
+
"loss": 0.2278,
|
5564 |
+
"step": 789
|
5565 |
+
},
|
5566 |
+
{
|
5567 |
+
"epoch": 0.824634655532359,
|
5568 |
+
"grad_norm": 0.6082893013954163,
|
5569 |
+
"learning_rate": 1.5101663790863596e-05,
|
5570 |
+
"loss": 0.1969,
|
5571 |
+
"step": 790
|
5572 |
+
},
|
5573 |
+
{
|
5574 |
+
"epoch": 0.8256784968684759,
|
5575 |
+
"grad_norm": 0.7025957703590393,
|
5576 |
+
"learning_rate": 1.4927016463976263e-05,
|
5577 |
+
"loss": 0.2071,
|
5578 |
+
"step": 791
|
5579 |
+
},
|
5580 |
+
{
|
5581 |
+
"epoch": 0.826722338204593,
|
5582 |
+
"grad_norm": 0.5405511260032654,
|
5583 |
+
"learning_rate": 1.4753303411248475e-05,
|
5584 |
+
"loss": 0.191,
|
5585 |
+
"step": 792
|
5586 |
+
},
|
5587 |
+
{
|
5588 |
+
"epoch": 0.8277661795407099,
|
5589 |
+
"grad_norm": 0.6413715481758118,
|
5590 |
+
"learning_rate": 1.4580526540402461e-05,
|
5591 |
+
"loss": 0.185,
|
5592 |
+
"step": 793
|
5593 |
+
},
|
5594 |
+
{
|
5595 |
+
"epoch": 0.8288100208768268,
|
5596 |
+
"grad_norm": 0.6367815136909485,
|
5597 |
+
"learning_rate": 1.4408687748879156e-05,
|
5598 |
+
"loss": 0.221,
|
5599 |
+
"step": 794
|
5600 |
+
},
|
5601 |
+
{
|
5602 |
+
"epoch": 0.8298538622129437,
|
5603 |
+
"grad_norm": 0.4684351086616516,
|
5604 |
+
"learning_rate": 1.4237788923817553e-05,
|
5605 |
+
"loss": 0.1415,
|
5606 |
+
"step": 795
|
5607 |
+
},
|
5608 |
+
{
|
5609 |
+
"epoch": 0.8308977035490606,
|
5610 |
+
"grad_norm": 0.5200543403625488,
|
5611 |
+
"learning_rate": 1.4067831942033904e-05,
|
5612 |
+
"loss": 0.1648,
|
5613 |
+
"step": 796
|
5614 |
+
},
|
5615 |
+
{
|
5616 |
+
"epoch": 0.8319415448851775,
|
5617 |
+
"grad_norm": 0.44216129183769226,
|
5618 |
+
"learning_rate": 1.3898818670001034e-05,
|
5619 |
+
"loss": 0.1608,
|
5620 |
+
"step": 797
|
5621 |
+
},
|
5622 |
+
{
|
5623 |
+
"epoch": 0.8329853862212944,
|
5624 |
+
"grad_norm": 0.5650377869606018,
|
5625 |
+
"learning_rate": 1.3730750963828032e-05,
|
5626 |
+
"loss": 0.1585,
|
5627 |
+
"step": 798
|
5628 |
+
},
|
5629 |
+
{
|
5630 |
+
"epoch": 0.8340292275574113,
|
5631 |
+
"grad_norm": 0.5171220898628235,
|
5632 |
+
"learning_rate": 1.3563630669239624e-05,
|
5633 |
+
"loss": 0.1575,
|
5634 |
+
"step": 799
|
5635 |
+
},
|
5636 |
+
{
|
5637 |
+
"epoch": 0.8350730688935282,
|
5638 |
+
"grad_norm": 0.5441738367080688,
|
5639 |
+
"learning_rate": 1.339745962155613e-05,
|
5640 |
+
"loss": 0.0949,
|
5641 |
+
"step": 800
|
5642 |
+
},
|
5643 |
+
{
|
5644 |
+
"epoch": 0.8361169102296451,
|
5645 |
+
"grad_norm": 0.6055110096931458,
|
5646 |
+
"learning_rate": 1.3232239645673217e-05,
|
5647 |
+
"loss": 0.4181,
|
5648 |
+
"step": 801
|
5649 |
+
},
|
5650 |
+
{
|
5651 |
+
"epoch": 0.837160751565762,
|
5652 |
+
"grad_norm": 0.6455709338188171,
|
5653 |
+
"learning_rate": 1.3067972556041752e-05,
|
5654 |
+
"loss": 0.3383,
|
5655 |
+
"step": 802
|
5656 |
+
},
|
5657 |
+
{
|
5658 |
+
"epoch": 0.8382045929018789,
|
5659 |
+
"grad_norm": 0.5849418640136719,
|
5660 |
+
"learning_rate": 1.2904660156648074e-05,
|
5661 |
+
"loss": 0.2803,
|
5662 |
+
"step": 803
|
5663 |
+
},
|
5664 |
+
{
|
5665 |
+
"epoch": 0.8392484342379958,
|
5666 |
+
"grad_norm": 0.538429856300354,
|
5667 |
+
"learning_rate": 1.2742304240994053e-05,
|
5668 |
+
"loss": 0.2402,
|
5669 |
+
"step": 804
|
5670 |
+
},
|
5671 |
+
{
|
5672 |
+
"epoch": 0.8402922755741128,
|
5673 |
+
"grad_norm": 0.6266717314720154,
|
5674 |
+
"learning_rate": 1.2580906592077402e-05,
|
5675 |
+
"loss": 0.3009,
|
5676 |
+
"step": 805
|
5677 |
+
},
|
5678 |
+
{
|
5679 |
+
"epoch": 0.8413361169102297,
|
5680 |
+
"grad_norm": 0.5892521142959595,
|
5681 |
+
"learning_rate": 1.2420468982372158e-05,
|
5682 |
+
"loss": 0.249,
|
5683 |
+
"step": 806
|
5684 |
+
},
|
5685 |
+
{
|
5686 |
+
"epoch": 0.8423799582463466,
|
5687 |
+
"grad_norm": 0.5853463411331177,
|
5688 |
+
"learning_rate": 1.226099317380912e-05,
|
5689 |
+
"loss": 0.2465,
|
5690 |
+
"step": 807
|
5691 |
+
},
|
5692 |
+
{
|
5693 |
+
"epoch": 0.8434237995824635,
|
5694 |
+
"grad_norm": 0.640910267829895,
|
5695 |
+
"learning_rate": 1.210248091775663e-05,
|
5696 |
+
"loss": 0.1872,
|
5697 |
+
"step": 808
|
5698 |
+
},
|
5699 |
+
{
|
5700 |
+
"epoch": 0.8444676409185804,
|
5701 |
+
"grad_norm": 0.6560840606689453,
|
5702 |
+
"learning_rate": 1.1944933955001225e-05,
|
5703 |
+
"loss": 0.2542,
|
5704 |
+
"step": 809
|
5705 |
+
},
|
5706 |
+
{
|
5707 |
+
"epoch": 0.8455114822546973,
|
5708 |
+
"grad_norm": 0.5866312980651855,
|
5709 |
+
"learning_rate": 1.1788354015728543e-05,
|
5710 |
+
"loss": 0.1911,
|
5711 |
+
"step": 810
|
5712 |
+
},
|
5713 |
+
{
|
5714 |
+
"epoch": 0.8465553235908142,
|
5715 |
+
"grad_norm": 0.6191656589508057,
|
5716 |
+
"learning_rate": 1.1632742819504405e-05,
|
5717 |
+
"loss": 0.2215,
|
5718 |
+
"step": 811
|
5719 |
+
},
|
5720 |
+
{
|
5721 |
+
"epoch": 0.8475991649269311,
|
5722 |
+
"grad_norm": 0.5251643657684326,
|
5723 |
+
"learning_rate": 1.147810207525577e-05,
|
5724 |
+
"loss": 0.1583,
|
5725 |
+
"step": 812
|
5726 |
+
},
|
5727 |
+
{
|
5728 |
+
"epoch": 0.848643006263048,
|
5729 |
+
"grad_norm": 0.4927600622177124,
|
5730 |
+
"learning_rate": 1.132443348125214e-05,
|
5731 |
+
"loss": 0.1512,
|
5732 |
+
"step": 813
|
5733 |
+
},
|
5734 |
+
{
|
5735 |
+
"epoch": 0.8496868475991649,
|
5736 |
+
"grad_norm": 0.4139147698879242,
|
5737 |
+
"learning_rate": 1.1171738725086833e-05,
|
5738 |
+
"loss": 0.1172,
|
5739 |
+
"step": 814
|
5740 |
+
},
|
5741 |
+
{
|
5742 |
+
"epoch": 0.8507306889352818,
|
5743 |
+
"grad_norm": 0.5602164268493652,
|
5744 |
+
"learning_rate": 1.1020019483658384e-05,
|
5745 |
+
"loss": 0.1821,
|
5746 |
+
"step": 815
|
5747 |
+
},
|
5748 |
+
{
|
5749 |
+
"epoch": 0.8517745302713987,
|
5750 |
+
"grad_norm": 0.4220430850982666,
|
5751 |
+
"learning_rate": 1.0869277423152246e-05,
|
5752 |
+
"loss": 0.1263,
|
5753 |
+
"step": 816
|
5754 |
+
},
|
5755 |
+
{
|
5756 |
+
"epoch": 0.8528183716075156,
|
5757 |
+
"grad_norm": 0.38943833112716675,
|
5758 |
+
"learning_rate": 1.0719514199022473e-05,
|
5759 |
+
"loss": 0.1141,
|
5760 |
+
"step": 817
|
5761 |
+
},
|
5762 |
+
{
|
5763 |
+
"epoch": 0.8538622129436325,
|
5764 |
+
"grad_norm": 0.5523675084114075,
|
5765 |
+
"learning_rate": 1.0570731455973414e-05,
|
5766 |
+
"loss": 0.1578,
|
5767 |
+
"step": 818
|
5768 |
+
},
|
5769 |
+
{
|
5770 |
+
"epoch": 0.8549060542797495,
|
5771 |
+
"grad_norm": 0.6071298718452454,
|
5772 |
+
"learning_rate": 1.04229308279418e-05,
|
5773 |
+
"loss": 0.2323,
|
5774 |
+
"step": 819
|
5775 |
+
},
|
5776 |
+
{
|
5777 |
+
"epoch": 0.8559498956158664,
|
5778 |
+
"grad_norm": 0.5393807291984558,
|
5779 |
+
"learning_rate": 1.0276113938078769e-05,
|
5780 |
+
"loss": 0.1967,
|
5781 |
+
"step": 820
|
5782 |
+
},
|
5783 |
+
{
|
5784 |
+
"epoch": 0.8569937369519833,
|
5785 |
+
"grad_norm": 0.6537972688674927,
|
5786 |
+
"learning_rate": 1.0130282398731982e-05,
|
5787 |
+
"loss": 0.1953,
|
5788 |
+
"step": 821
|
5789 |
+
},
|
5790 |
+
{
|
5791 |
+
"epoch": 0.8580375782881002,
|
5792 |
+
"grad_norm": 0.4417877197265625,
|
5793 |
+
"learning_rate": 9.985437811427933e-06,
|
5794 |
+
"loss": 0.1453,
|
5795 |
+
"step": 822
|
5796 |
+
},
|
5797 |
+
{
|
5798 |
+
"epoch": 0.8590814196242171,
|
5799 |
+
"grad_norm": 0.6434723734855652,
|
5800 |
+
"learning_rate": 9.841581766854401e-06,
|
5801 |
+
"loss": 0.1683,
|
5802 |
+
"step": 823
|
5803 |
+
},
|
5804 |
+
{
|
5805 |
+
"epoch": 0.860125260960334,
|
5806 |
+
"grad_norm": 0.7254697680473328,
|
5807 |
+
"learning_rate": 9.698715844842988e-06,
|
5808 |
+
"loss": 0.3499,
|
5809 |
+
"step": 824
|
5810 |
+
},
|
5811 |
+
{
|
5812 |
+
"epoch": 0.8611691022964509,
|
5813 |
+
"grad_norm": 0.6909394264221191,
|
5814 |
+
"learning_rate": 9.556841614351664e-06,
|
5815 |
+
"loss": 0.2757,
|
5816 |
+
"step": 825
|
5817 |
+
},
|
5818 |
+
{
|
5819 |
+
"epoch": 0.8622129436325678,
|
5820 |
+
"grad_norm": 0.7137805223464966,
|
5821 |
+
"learning_rate": 9.415960633447674e-06,
|
5822 |
+
"loss": 0.3122,
|
5823 |
+
"step": 826
|
5824 |
+
},
|
5825 |
+
{
|
5826 |
+
"epoch": 0.8632567849686847,
|
5827 |
+
"grad_norm": 0.7450171709060669,
|
5828 |
+
"learning_rate": 9.276074449290361e-06,
|
5829 |
+
"loss": 0.2928,
|
5830 |
+
"step": 827
|
5831 |
+
},
|
5832 |
+
{
|
5833 |
+
"epoch": 0.8643006263048016,
|
5834 |
+
"grad_norm": 0.6502891778945923,
|
5835 |
+
"learning_rate": 9.137184598114134e-06,
|
5836 |
+
"loss": 0.2517,
|
5837 |
+
"step": 828
|
5838 |
+
},
|
5839 |
+
{
|
5840 |
+
"epoch": 0.8653444676409185,
|
5841 |
+
"grad_norm": 0.6984295845031738,
|
5842 |
+
"learning_rate": 8.999292605211695e-06,
|
5843 |
+
"loss": 0.3062,
|
5844 |
+
"step": 829
|
5845 |
+
},
|
5846 |
+
{
|
5847 |
+
"epoch": 0.8663883089770354,
|
5848 |
+
"grad_norm": 0.7169867753982544,
|
5849 |
+
"learning_rate": 8.862399984917213e-06,
|
5850 |
+
"loss": 0.2748,
|
5851 |
+
"step": 830
|
5852 |
+
},
|
5853 |
+
{
|
5854 |
+
"epoch": 0.8674321503131524,
|
5855 |
+
"grad_norm": 0.7259141802787781,
|
5856 |
+
"learning_rate": 8.726508240589692e-06,
|
5857 |
+
"loss": 0.3033,
|
5858 |
+
"step": 831
|
5859 |
+
},
|
5860 |
+
{
|
5861 |
+
"epoch": 0.8684759916492694,
|
5862 |
+
"grad_norm": 0.7949566841125488,
|
5863 |
+
"learning_rate": 8.59161886459654e-06,
|
5864 |
+
"loss": 0.3149,
|
5865 |
+
"step": 832
|
5866 |
+
},
|
5867 |
+
{
|
5868 |
+
"epoch": 0.8695198329853863,
|
5869 |
+
"grad_norm": 0.7562083005905151,
|
5870 |
+
"learning_rate": 8.457733338297069e-06,
|
5871 |
+
"loss": 0.3192,
|
5872 |
+
"step": 833
|
5873 |
+
},
|
5874 |
+
{
|
5875 |
+
"epoch": 0.8705636743215032,
|
5876 |
+
"grad_norm": 0.6112555265426636,
|
5877 |
+
"learning_rate": 8.3248531320263e-06,
|
5878 |
+
"loss": 0.2175,
|
5879 |
+
"step": 834
|
5880 |
+
},
|
5881 |
+
{
|
5882 |
+
"epoch": 0.8716075156576201,
|
5883 |
+
"grad_norm": 0.7050125598907471,
|
5884 |
+
"learning_rate": 8.192979705078852e-06,
|
5885 |
+
"loss": 0.243,
|
5886 |
+
"step": 835
|
5887 |
+
},
|
5888 |
+
{
|
5889 |
+
"epoch": 0.872651356993737,
|
5890 |
+
"grad_norm": 0.6470485925674438,
|
5891 |
+
"learning_rate": 8.062114505692742e-06,
|
5892 |
+
"loss": 0.2384,
|
5893 |
+
"step": 836
|
5894 |
+
},
|
5895 |
+
{
|
5896 |
+
"epoch": 0.8736951983298539,
|
5897 |
+
"grad_norm": 0.7082952260971069,
|
5898 |
+
"learning_rate": 7.932258971033746e-06,
|
5899 |
+
"loss": 0.2795,
|
5900 |
+
"step": 837
|
5901 |
+
},
|
5902 |
+
{
|
5903 |
+
"epoch": 0.8747390396659708,
|
5904 |
+
"grad_norm": 0.843268096446991,
|
5905 |
+
"learning_rate": 7.803414527179343e-06,
|
5906 |
+
"loss": 0.2991,
|
5907 |
+
"step": 838
|
5908 |
+
},
|
5909 |
+
{
|
5910 |
+
"epoch": 0.8757828810020877,
|
5911 |
+
"grad_norm": 0.6356431245803833,
|
5912 |
+
"learning_rate": 7.675582589103247e-06,
|
5913 |
+
"loss": 0.1963,
|
5914 |
+
"step": 839
|
5915 |
+
},
|
5916 |
+
{
|
5917 |
+
"epoch": 0.8768267223382046,
|
5918 |
+
"grad_norm": 0.569520890712738,
|
5919 |
+
"learning_rate": 7.548764560659816e-06,
|
5920 |
+
"loss": 0.1703,
|
5921 |
+
"step": 840
|
5922 |
+
},
|
5923 |
+
{
|
5924 |
+
"epoch": 0.8778705636743215,
|
5925 |
+
"grad_norm": 0.6984921097755432,
|
5926 |
+
"learning_rate": 7.422961834568565e-06,
|
5927 |
+
"loss": 0.2231,
|
5928 |
+
"step": 841
|
5929 |
+
},
|
5930 |
+
{
|
5931 |
+
"epoch": 0.8789144050104384,
|
5932 |
+
"grad_norm": 0.6111634969711304,
|
5933 |
+
"learning_rate": 7.2981757923989755e-06,
|
5934 |
+
"loss": 0.1825,
|
5935 |
+
"step": 842
|
5936 |
+
},
|
5937 |
+
{
|
5938 |
+
"epoch": 0.8799582463465553,
|
5939 |
+
"grad_norm": 0.666388213634491,
|
5940 |
+
"learning_rate": 7.174407804555261e-06,
|
5941 |
+
"loss": 0.1775,
|
5942 |
+
"step": 843
|
5943 |
+
},
|
5944 |
+
{
|
5945 |
+
"epoch": 0.8810020876826722,
|
5946 |
+
"grad_norm": 0.7088585495948792,
|
5947 |
+
"learning_rate": 7.051659230261298e-06,
|
5948 |
+
"loss": 0.1873,
|
5949 |
+
"step": 844
|
5950 |
+
},
|
5951 |
+
{
|
5952 |
+
"epoch": 0.8820459290187892,
|
5953 |
+
"grad_norm": 0.6127867102622986,
|
5954 |
+
"learning_rate": 6.929931417545788e-06,
|
5955 |
+
"loss": 0.1732,
|
5956 |
+
"step": 845
|
5957 |
+
},
|
5958 |
+
{
|
5959 |
+
"epoch": 0.8830897703549061,
|
5960 |
+
"grad_norm": 0.637973964214325,
|
5961 |
+
"learning_rate": 6.809225703227351e-06,
|
5962 |
+
"loss": 0.1856,
|
5963 |
+
"step": 846
|
5964 |
+
},
|
5965 |
+
{
|
5966 |
+
"epoch": 0.884133611691023,
|
5967 |
+
"grad_norm": 0.5888153910636902,
|
5968 |
+
"learning_rate": 6.689543412899913e-06,
|
5969 |
+
"loss": 0.1872,
|
5970 |
+
"step": 847
|
5971 |
+
},
|
5972 |
+
{
|
5973 |
+
"epoch": 0.8851774530271399,
|
5974 |
+
"grad_norm": 0.6536146402359009,
|
5975 |
+
"learning_rate": 6.57088586091813e-06,
|
5976 |
+
"loss": 0.2259,
|
5977 |
+
"step": 848
|
5978 |
+
},
|
5979 |
+
{
|
5980 |
+
"epoch": 0.8862212943632568,
|
5981 |
+
"grad_norm": 0.4231550693511963,
|
5982 |
+
"learning_rate": 6.45325435038292e-06,
|
5983 |
+
"loss": 0.1388,
|
5984 |
+
"step": 849
|
5985 |
+
},
|
5986 |
+
{
|
5987 |
+
"epoch": 0.8872651356993737,
|
5988 |
+
"grad_norm": 0.8168404698371887,
|
5989 |
+
"learning_rate": 6.336650173127223e-06,
|
5990 |
+
"loss": 0.1743,
|
5991 |
+
"step": 850
|
5992 |
+
},
|
5993 |
+
{
|
5994 |
+
"epoch": 0.8883089770354906,
|
5995 |
+
"grad_norm": 0.5078144073486328,
|
5996 |
+
"learning_rate": 6.221074609701738e-06,
|
5997 |
+
"loss": 0.2793,
|
5998 |
+
"step": 851
|
5999 |
+
},
|
6000 |
+
{
|
6001 |
+
"epoch": 0.8893528183716075,
|
6002 |
+
"grad_norm": 0.7190085053443909,
|
6003 |
+
"learning_rate": 6.106528929360911e-06,
|
6004 |
+
"loss": 0.4109,
|
6005 |
+
"step": 852
|
6006 |
+
},
|
6007 |
+
{
|
6008 |
+
"epoch": 0.8903966597077244,
|
6009 |
+
"grad_norm": 0.5939377546310425,
|
6010 |
+
"learning_rate": 5.99301439004899e-06,
|
6011 |
+
"loss": 0.2797,
|
6012 |
+
"step": 853
|
6013 |
+
},
|
6014 |
+
{
|
6015 |
+
"epoch": 0.8914405010438413,
|
6016 |
+
"grad_norm": 0.7328153848648071,
|
6017 |
+
"learning_rate": 5.880532238386161e-06,
|
6018 |
+
"loss": 0.3652,
|
6019 |
+
"step": 854
|
6020 |
+
},
|
6021 |
+
{
|
6022 |
+
"epoch": 0.8924843423799582,
|
6023 |
+
"grad_norm": 0.6011344194412231,
|
6024 |
+
"learning_rate": 5.769083709654932e-06,
|
6025 |
+
"loss": 0.2699,
|
6026 |
+
"step": 855
|
6027 |
+
},
|
6028 |
+
{
|
6029 |
+
"epoch": 0.8935281837160751,
|
6030 |
+
"grad_norm": 0.7004411220550537,
|
6031 |
+
"learning_rate": 5.658670027786561e-06,
|
6032 |
+
"loss": 0.3191,
|
6033 |
+
"step": 856
|
6034 |
+
},
|
6035 |
+
{
|
6036 |
+
"epoch": 0.894572025052192,
|
6037 |
+
"grad_norm": 0.5985621809959412,
|
6038 |
+
"learning_rate": 5.549292405347495e-06,
|
6039 |
+
"loss": 0.2593,
|
6040 |
+
"step": 857
|
6041 |
+
},
|
6042 |
+
{
|
6043 |
+
"epoch": 0.8956158663883089,
|
6044 |
+
"grad_norm": 0.6498935222625732,
|
6045 |
+
"learning_rate": 5.440952043526215e-06,
|
6046 |
+
"loss": 0.3027,
|
6047 |
+
"step": 858
|
6048 |
+
},
|
6049 |
+
{
|
6050 |
+
"epoch": 0.8966597077244259,
|
6051 |
+
"grad_norm": 0.6431671380996704,
|
6052 |
+
"learning_rate": 5.3336501321199714e-06,
|
6053 |
+
"loss": 0.237,
|
6054 |
+
"step": 859
|
6055 |
+
},
|
6056 |
+
{
|
6057 |
+
"epoch": 0.8977035490605428,
|
6058 |
+
"grad_norm": 0.6603933572769165,
|
6059 |
+
"learning_rate": 5.22738784952167e-06,
|
6060 |
+
"loss": 0.2533,
|
6061 |
+
"step": 860
|
6062 |
+
},
|
6063 |
+
{
|
6064 |
+
"epoch": 0.8987473903966597,
|
6065 |
+
"grad_norm": 0.5379349589347839,
|
6066 |
+
"learning_rate": 5.1221663627070485e-06,
|
6067 |
+
"loss": 0.1883,
|
6068 |
+
"step": 861
|
6069 |
+
},
|
6070 |
+
{
|
6071 |
+
"epoch": 0.8997912317327766,
|
6072 |
+
"grad_norm": 0.4264977276325226,
|
6073 |
+
"learning_rate": 5.017986827221733e-06,
|
6074 |
+
"loss": 0.1174,
|
6075 |
+
"step": 862
|
6076 |
+
},
|
6077 |
+
{
|
6078 |
+
"epoch": 0.9008350730688935,
|
6079 |
+
"grad_norm": 0.5029094815254211,
|
6080 |
+
"learning_rate": 4.914850387168657e-06,
|
6081 |
+
"loss": 0.1564,
|
6082 |
+
"step": 863
|
6083 |
+
},
|
6084 |
+
{
|
6085 |
+
"epoch": 0.9018789144050104,
|
6086 |
+
"grad_norm": 0.5162425637245178,
|
6087 |
+
"learning_rate": 4.812758175195397e-06,
|
6088 |
+
"loss": 0.1547,
|
6089 |
+
"step": 864
|
6090 |
+
},
|
6091 |
+
{
|
6092 |
+
"epoch": 0.9029227557411273,
|
6093 |
+
"grad_norm": 0.6308012008666992,
|
6094 |
+
"learning_rate": 4.711711312481815e-06,
|
6095 |
+
"loss": 0.1844,
|
6096 |
+
"step": 865
|
6097 |
+
},
|
6098 |
+
{
|
6099 |
+
"epoch": 0.9039665970772442,
|
6100 |
+
"grad_norm": 0.46069368720054626,
|
6101 |
+
"learning_rate": 4.61171090872774e-06,
|
6102 |
+
"loss": 0.1363,
|
6103 |
+
"step": 866
|
6104 |
+
},
|
6105 |
+
{
|
6106 |
+
"epoch": 0.9050104384133612,
|
6107 |
+
"grad_norm": 0.40711909532546997,
|
6108 |
+
"learning_rate": 4.512758062140687e-06,
|
6109 |
+
"loss": 0.1252,
|
6110 |
+
"step": 867
|
6111 |
+
},
|
6112 |
+
{
|
6113 |
+
"epoch": 0.906054279749478,
|
6114 |
+
"grad_norm": 0.5450437664985657,
|
6115 |
+
"learning_rate": 4.4148538594239174e-06,
|
6116 |
+
"loss": 0.1885,
|
6117 |
+
"step": 868
|
6118 |
+
},
|
6119 |
+
{
|
6120 |
+
"epoch": 0.907098121085595,
|
6121 |
+
"grad_norm": 0.5693588852882385,
|
6122 |
+
"learning_rate": 4.317999375764459e-06,
|
6123 |
+
"loss": 0.2161,
|
6124 |
+
"step": 869
|
6125 |
+
},
|
6126 |
+
{
|
6127 |
+
"epoch": 0.9081419624217119,
|
6128 |
+
"grad_norm": 0.45915868878364563,
|
6129 |
+
"learning_rate": 4.2221956748212384e-06,
|
6130 |
+
"loss": 0.1642,
|
6131 |
+
"step": 870
|
6132 |
+
},
|
6133 |
+
{
|
6134 |
+
"epoch": 0.9091858037578288,
|
6135 |
+
"grad_norm": 0.3877635896205902,
|
6136 |
+
"learning_rate": 4.127443808713527e-06,
|
6137 |
+
"loss": 0.1424,
|
6138 |
+
"step": 871
|
6139 |
+
},
|
6140 |
+
{
|
6141 |
+
"epoch": 0.9102296450939458,
|
6142 |
+
"grad_norm": 0.5209342241287231,
|
6143 |
+
"learning_rate": 4.033744818009244e-06,
|
6144 |
+
"loss": 0.1703,
|
6145 |
+
"step": 872
|
6146 |
+
},
|
6147 |
+
{
|
6148 |
+
"epoch": 0.9112734864300627,
|
6149 |
+
"grad_norm": 0.4651091396808624,
|
6150 |
+
"learning_rate": 3.941099731713637e-06,
|
6151 |
+
"loss": 0.1584,
|
6152 |
+
"step": 873
|
6153 |
+
},
|
6154 |
+
{
|
6155 |
+
"epoch": 0.9123173277661796,
|
6156 |
+
"grad_norm": 0.8967810869216919,
|
6157 |
+
"learning_rate": 3.849509567257959e-06,
|
6158 |
+
"loss": 0.3558,
|
6159 |
+
"step": 874
|
6160 |
+
},
|
6161 |
+
{
|
6162 |
+
"epoch": 0.9133611691022965,
|
6163 |
+
"grad_norm": 0.8381048440933228,
|
6164 |
+
"learning_rate": 3.7589753304882124e-06,
|
6165 |
+
"loss": 0.3886,
|
6166 |
+
"step": 875
|
6167 |
+
},
|
6168 |
+
{
|
6169 |
+
"epoch": 0.9144050104384134,
|
6170 |
+
"grad_norm": 0.6149895787239075,
|
6171 |
+
"learning_rate": 3.669498015654249e-06,
|
6172 |
+
"loss": 0.31,
|
6173 |
+
"step": 876
|
6174 |
+
},
|
6175 |
+
{
|
6176 |
+
"epoch": 0.9154488517745303,
|
6177 |
+
"grad_norm": 0.7612007856369019,
|
6178 |
+
"learning_rate": 3.5810786053987023e-06,
|
6179 |
+
"loss": 0.359,
|
6180 |
+
"step": 877
|
6181 |
+
},
|
6182 |
+
{
|
6183 |
+
"epoch": 0.9164926931106472,
|
6184 |
+
"grad_norm": 0.6727755069732666,
|
6185 |
+
"learning_rate": 3.493718070746299e-06,
|
6186 |
+
"loss": 0.2748,
|
6187 |
+
"step": 878
|
6188 |
+
},
|
6189 |
+
{
|
6190 |
+
"epoch": 0.9175365344467641,
|
6191 |
+
"grad_norm": 0.734786868095398,
|
6192 |
+
"learning_rate": 3.40741737109318e-06,
|
6193 |
+
"loss": 0.3066,
|
6194 |
+
"step": 879
|
6195 |
+
},
|
6196 |
+
{
|
6197 |
+
"epoch": 0.918580375782881,
|
6198 |
+
"grad_norm": 0.6576768159866333,
|
6199 |
+
"learning_rate": 3.3221774541962845e-06,
|
6200 |
+
"loss": 0.2677,
|
6201 |
+
"step": 880
|
6202 |
+
},
|
6203 |
+
{
|
6204 |
+
"epoch": 0.9196242171189979,
|
6205 |
+
"grad_norm": 0.6196028590202332,
|
6206 |
+
"learning_rate": 3.2379992561630712e-06,
|
6207 |
+
"loss": 0.2484,
|
6208 |
+
"step": 881
|
6209 |
+
},
|
6210 |
+
{
|
6211 |
+
"epoch": 0.9206680584551148,
|
6212 |
+
"grad_norm": 0.7527311444282532,
|
6213 |
+
"learning_rate": 3.1548837014411357e-06,
|
6214 |
+
"loss": 0.2787,
|
6215 |
+
"step": 882
|
6216 |
+
},
|
6217 |
+
{
|
6218 |
+
"epoch": 0.9217118997912317,
|
6219 |
+
"grad_norm": 0.7063425779342651,
|
6220 |
+
"learning_rate": 3.0728317028080657e-06,
|
6221 |
+
"loss": 0.303,
|
6222 |
+
"step": 883
|
6223 |
+
},
|
6224 |
+
{
|
6225 |
+
"epoch": 0.9227557411273486,
|
6226 |
+
"grad_norm": 0.6437200307846069,
|
6227 |
+
"learning_rate": 2.9918441613615123e-06,
|
6228 |
+
"loss": 0.2501,
|
6229 |
+
"step": 884
|
6230 |
+
},
|
6231 |
+
{
|
6232 |
+
"epoch": 0.9237995824634656,
|
6233 |
+
"grad_norm": 0.8236239552497864,
|
6234 |
+
"learning_rate": 2.9119219665091344e-06,
|
6235 |
+
"loss": 0.3234,
|
6236 |
+
"step": 885
|
6237 |
+
},
|
6238 |
+
{
|
6239 |
+
"epoch": 0.9248434237995825,
|
6240 |
+
"grad_norm": 0.686543881893158,
|
6241 |
+
"learning_rate": 2.8330659959589946e-06,
|
6242 |
+
"loss": 0.2407,
|
6243 |
+
"step": 886
|
6244 |
+
},
|
6245 |
+
{
|
6246 |
+
"epoch": 0.9258872651356994,
|
6247 |
+
"grad_norm": 0.7093439698219299,
|
6248 |
+
"learning_rate": 2.755277115709842e-06,
|
6249 |
+
"loss": 0.246,
|
6250 |
+
"step": 887
|
6251 |
+
},
|
6252 |
+
{
|
6253 |
+
"epoch": 0.9269311064718163,
|
6254 |
+
"grad_norm": 0.7115840315818787,
|
6255 |
+
"learning_rate": 2.678556180041547e-06,
|
6256 |
+
"loss": 0.2551,
|
6257 |
+
"step": 888
|
6258 |
+
},
|
6259 |
+
{
|
6260 |
+
"epoch": 0.9279749478079332,
|
6261 |
+
"grad_norm": 0.6106806993484497,
|
6262 |
+
"learning_rate": 2.6029040315058485e-06,
|
6263 |
+
"loss": 0.2137,
|
6264 |
+
"step": 889
|
6265 |
+
},
|
6266 |
+
{
|
6267 |
+
"epoch": 0.9290187891440501,
|
6268 |
+
"grad_norm": 0.5461225509643555,
|
6269 |
+
"learning_rate": 2.5283215009169857e-06,
|
6270 |
+
"loss": 0.1847,
|
6271 |
+
"step": 890
|
6272 |
+
},
|
6273 |
+
{
|
6274 |
+
"epoch": 0.930062630480167,
|
6275 |
+
"grad_norm": 0.6097748875617981,
|
6276 |
+
"learning_rate": 2.4548094073426398e-06,
|
6277 |
+
"loss": 0.209,
|
6278 |
+
"step": 891
|
6279 |
+
},
|
6280 |
+
{
|
6281 |
+
"epoch": 0.9311064718162839,
|
6282 |
+
"grad_norm": 0.6483787298202515,
|
6283 |
+
"learning_rate": 2.3823685580949273e-06,
|
6284 |
+
"loss": 0.2244,
|
6285 |
+
"step": 892
|
6286 |
+
},
|
6287 |
+
{
|
6288 |
+
"epoch": 0.9321503131524008,
|
6289 |
+
"grad_norm": 0.5993359088897705,
|
6290 |
+
"learning_rate": 2.3109997487214983e-06,
|
6291 |
+
"loss": 0.189,
|
6292 |
+
"step": 893
|
6293 |
+
},
|
6294 |
+
{
|
6295 |
+
"epoch": 0.9331941544885177,
|
6296 |
+
"grad_norm": 0.8005963563919067,
|
6297 |
+
"learning_rate": 2.240703762996843e-06,
|
6298 |
+
"loss": 0.2385,
|
6299 |
+
"step": 894
|
6300 |
+
},
|
6301 |
+
{
|
6302 |
+
"epoch": 0.9342379958246346,
|
6303 |
+
"grad_norm": 0.4756294786930084,
|
6304 |
+
"learning_rate": 2.1714813729136975e-06,
|
6305 |
+
"loss": 0.1531,
|
6306 |
+
"step": 895
|
6307 |
+
},
|
6308 |
+
{
|
6309 |
+
"epoch": 0.9352818371607515,
|
6310 |
+
"grad_norm": 0.577684760093689,
|
6311 |
+
"learning_rate": 2.1033333386744846e-06,
|
6312 |
+
"loss": 0.1722,
|
6313 |
+
"step": 896
|
6314 |
+
},
|
6315 |
+
{
|
6316 |
+
"epoch": 0.9363256784968684,
|
6317 |
+
"grad_norm": 0.5431109666824341,
|
6318 |
+
"learning_rate": 2.036260408683033e-06,
|
6319 |
+
"loss": 0.1796,
|
6320 |
+
"step": 897
|
6321 |
+
},
|
6322 |
+
{
|
6323 |
+
"epoch": 0.9373695198329853,
|
6324 |
+
"grad_norm": 0.5910576581954956,
|
6325 |
+
"learning_rate": 1.9702633195363917e-06,
|
6326 |
+
"loss": 0.1689,
|
6327 |
+
"step": 898
|
6328 |
+
},
|
6329 |
+
{
|
6330 |
+
"epoch": 0.9384133611691023,
|
6331 |
+
"grad_norm": 0.6282055377960205,
|
6332 |
+
"learning_rate": 1.9053427960166182e-06,
|
6333 |
+
"loss": 0.1681,
|
6334 |
+
"step": 899
|
6335 |
+
},
|
6336 |
+
{
|
6337 |
+
"epoch": 0.9394572025052192,
|
6338 |
+
"grad_norm": 0.46189793944358826,
|
6339 |
+
"learning_rate": 1.8414995510829368e-06,
|
6340 |
+
"loss": 0.0832,
|
6341 |
+
"step": 900
|
6342 |
+
},
|
6343 |
+
{
|
6344 |
+
"epoch": 0.9405010438413361,
|
6345 |
+
"grad_norm": 0.7074732780456543,
|
6346 |
+
"learning_rate": 1.778734285863859e-06,
|
6347 |
+
"loss": 0.3342,
|
6348 |
+
"step": 901
|
6349 |
+
},
|
6350 |
+
{
|
6351 |
+
"epoch": 0.941544885177453,
|
6352 |
+
"grad_norm": 0.6299365162849426,
|
6353 |
+
"learning_rate": 1.717047689649487e-06,
|
6354 |
+
"loss": 0.3618,
|
6355 |
+
"step": 902
|
6356 |
+
},
|
6357 |
+
{
|
6358 |
+
"epoch": 0.94258872651357,
|
6359 |
+
"grad_norm": 0.6098471879959106,
|
6360 |
+
"learning_rate": 1.6564404398839439e-06,
|
6361 |
+
"loss": 0.3235,
|
6362 |
+
"step": 903
|
6363 |
+
},
|
6364 |
+
{
|
6365 |
+
"epoch": 0.9436325678496869,
|
6366 |
+
"grad_norm": 0.5733586549758911,
|
6367 |
+
"learning_rate": 1.5969132021579347e-06,
|
6368 |
+
"loss": 0.2646,
|
6369 |
+
"step": 904
|
6370 |
+
},
|
6371 |
+
{
|
6372 |
+
"epoch": 0.9446764091858038,
|
6373 |
+
"grad_norm": 0.6624009609222412,
|
6374 |
+
"learning_rate": 1.5384666302014406e-06,
|
6375 |
+
"loss": 0.3138,
|
6376 |
+
"step": 905
|
6377 |
+
},
|
6378 |
+
{
|
6379 |
+
"epoch": 0.9457202505219207,
|
6380 |
+
"grad_norm": 0.7231829762458801,
|
6381 |
+
"learning_rate": 1.481101365876547e-06,
|
6382 |
+
"loss": 0.3016,
|
6383 |
+
"step": 906
|
6384 |
+
},
|
6385 |
+
{
|
6386 |
+
"epoch": 0.9467640918580376,
|
6387 |
+
"grad_norm": 0.634404718875885,
|
6388 |
+
"learning_rate": 1.4248180391703614e-06,
|
6389 |
+
"loss": 0.2247,
|
6390 |
+
"step": 907
|
6391 |
+
},
|
6392 |
+
{
|
6393 |
+
"epoch": 0.9478079331941545,
|
6394 |
+
"grad_norm": 0.664415180683136,
|
6395 |
+
"learning_rate": 1.3696172681881503e-06,
|
6396 |
+
"loss": 0.3176,
|
6397 |
+
"step": 908
|
6398 |
+
},
|
6399 |
+
{
|
6400 |
+
"epoch": 0.9488517745302714,
|
6401 |
+
"grad_norm": 0.5892297625541687,
|
6402 |
+
"learning_rate": 1.3154996591464908e-06,
|
6403 |
+
"loss": 0.2327,
|
6404 |
+
"step": 909
|
6405 |
+
},
|
6406 |
+
{
|
6407 |
+
"epoch": 0.9498956158663883,
|
6408 |
+
"grad_norm": 0.5772663354873657,
|
6409 |
+
"learning_rate": 1.2624658063666639e-06,
|
6410 |
+
"loss": 0.2104,
|
6411 |
+
"step": 910
|
6412 |
+
},
|
6413 |
+
{
|
6414 |
+
"epoch": 0.9509394572025052,
|
6415 |
+
"grad_norm": 0.5238860845565796,
|
6416 |
+
"learning_rate": 1.2105162922680824e-06,
|
6417 |
+
"loss": 0.1807,
|
6418 |
+
"step": 911
|
6419 |
+
},
|
6420 |
+
{
|
6421 |
+
"epoch": 0.9519832985386222,
|
6422 |
+
"grad_norm": 0.5960121750831604,
|
6423 |
+
"learning_rate": 1.15965168736194e-06,
|
6424 |
+
"loss": 0.2065,
|
6425 |
+
"step": 912
|
6426 |
+
},
|
6427 |
+
{
|
6428 |
+
"epoch": 0.9530271398747391,
|
6429 |
+
"grad_norm": 0.47030001878738403,
|
6430 |
+
"learning_rate": 1.109872550244917e-06,
|
6431 |
+
"loss": 0.1519,
|
6432 |
+
"step": 913
|
6433 |
+
},
|
6434 |
+
{
|
6435 |
+
"epoch": 0.954070981210856,
|
6436 |
+
"grad_norm": 0.5340988636016846,
|
6437 |
+
"learning_rate": 1.0611794275930399e-06,
|
6438 |
+
"loss": 0.1603,
|
6439 |
+
"step": 914
|
6440 |
+
},
|
6441 |
+
{
|
6442 |
+
"epoch": 0.9551148225469729,
|
6443 |
+
"grad_norm": 0.3915995955467224,
|
6444 |
+
"learning_rate": 1.01357285415572e-06,
|
6445 |
+
"loss": 0.1179,
|
6446 |
+
"step": 915
|
6447 |
+
},
|
6448 |
+
{
|
6449 |
+
"epoch": 0.9561586638830898,
|
6450 |
+
"grad_norm": 0.3927573561668396,
|
6451 |
+
"learning_rate": 9.670533527498137e-07,
|
6452 |
+
"loss": 0.1151,
|
6453 |
+
"step": 916
|
6454 |
+
},
|
6455 |
+
{
|
6456 |
+
"epoch": 0.9572025052192067,
|
6457 |
+
"grad_norm": 0.3486252725124359,
|
6458 |
+
"learning_rate": 9.216214342539386e-07,
|
6459 |
+
"loss": 0.0942,
|
6460 |
+
"step": 917
|
6461 |
+
},
|
6462 |
+
{
|
6463 |
+
"epoch": 0.9582463465553236,
|
6464 |
+
"grad_norm": 0.3841801583766937,
|
6465 |
+
"learning_rate": 8.772775976028546e-07,
|
6466 |
+
"loss": 0.1185,
|
6467 |
+
"step": 918
|
6468 |
+
},
|
6469 |
+
{
|
6470 |
+
"epoch": 0.9592901878914405,
|
6471 |
+
"grad_norm": 0.49043968319892883,
|
6472 |
+
"learning_rate": 8.340223297819471e-07,
|
6473 |
+
"loss": 0.1571,
|
6474 |
+
"step": 919
|
6475 |
+
},
|
6476 |
+
{
|
6477 |
+
"epoch": 0.9603340292275574,
|
6478 |
+
"grad_norm": 0.5569294691085815,
|
6479 |
+
"learning_rate": 7.918561058219198e-07,
|
6480 |
+
"loss": 0.1835,
|
6481 |
+
"step": 920
|
6482 |
+
},
|
6483 |
+
{
|
6484 |
+
"epoch": 0.9613778705636743,
|
6485 |
+
"grad_norm": 0.39632856845855713,
|
6486 |
+
"learning_rate": 7.507793887935654e-07,
|
6487 |
+
"loss": 0.1649,
|
6488 |
+
"step": 921
|
6489 |
+
},
|
6490 |
+
{
|
6491 |
+
"epoch": 0.9624217118997912,
|
6492 |
+
"grad_norm": 0.46687746047973633,
|
6493 |
+
"learning_rate": 7.10792629802659e-07,
|
6494 |
+
"loss": 0.1609,
|
6495 |
+
"step": 922
|
6496 |
+
},
|
6497 |
+
{
|
6498 |
+
"epoch": 0.9634655532359081,
|
6499 |
+
"grad_norm": 0.3641255795955658,
|
6500 |
+
"learning_rate": 6.718962679850505e-07,
|
6501 |
+
"loss": 0.1387,
|
6502 |
+
"step": 923
|
6503 |
+
},
|
6504 |
+
{
|
6505 |
+
"epoch": 0.964509394572025,
|
6506 |
+
"grad_norm": 0.5133196115493774,
|
6507 |
+
"learning_rate": 6.340907305017907e-07,
|
6508 |
+
"loss": 0.1752,
|
6509 |
+
"step": 924
|
6510 |
+
},
|
6511 |
+
{
|
6512 |
+
"epoch": 0.965553235908142,
|
6513 |
+
"grad_norm": 0.7964520454406738,
|
6514 |
+
"learning_rate": 5.973764325344688e-07,
|
6515 |
+
"loss": 0.3131,
|
6516 |
+
"step": 925
|
6517 |
+
},
|
6518 |
+
{
|
6519 |
+
"epoch": 0.9665970772442589,
|
6520 |
+
"grad_norm": 0.7928893566131592,
|
6521 |
+
"learning_rate": 5.617537772806602e-07,
|
6522 |
+
"loss": 0.3533,
|
6523 |
+
"step": 926
|
6524 |
+
},
|
6525 |
+
{
|
6526 |
+
"epoch": 0.9676409185803758,
|
6527 |
+
"grad_norm": 0.7127247452735901,
|
6528 |
+
"learning_rate": 5.272231559494634e-07,
|
6529 |
+
"loss": 0.3272,
|
6530 |
+
"step": 927
|
6531 |
+
},
|
6532 |
+
{
|
6533 |
+
"epoch": 0.9686847599164927,
|
6534 |
+
"grad_norm": 0.6839384436607361,
|
6535 |
+
"learning_rate": 4.937849477572587e-07,
|
6536 |
+
"loss": 0.3335,
|
6537 |
+
"step": 928
|
6538 |
+
},
|
6539 |
+
{
|
6540 |
+
"epoch": 0.9697286012526096,
|
6541 |
+
"grad_norm": 0.6431897878646851,
|
6542 |
+
"learning_rate": 4.614395199234678e-07,
|
6543 |
+
"loss": 0.279,
|
6544 |
+
"step": 929
|
6545 |
+
},
|
6546 |
+
{
|
6547 |
+
"epoch": 0.9707724425887265,
|
6548 |
+
"grad_norm": 0.6917023658752441,
|
6549 |
+
"learning_rate": 4.3018722766661193e-07,
|
6550 |
+
"loss": 0.3009,
|
6551 |
+
"step": 930
|
6552 |
+
},
|
6553 |
+
{
|
6554 |
+
"epoch": 0.9718162839248434,
|
6555 |
+
"grad_norm": 0.6915965676307678,
|
6556 |
+
"learning_rate": 4.0002841420032634e-07,
|
6557 |
+
"loss": 0.2842,
|
6558 |
+
"step": 931
|
6559 |
+
},
|
6560 |
+
{
|
6561 |
+
"epoch": 0.9728601252609603,
|
6562 |
+
"grad_norm": 0.6639567017555237,
|
6563 |
+
"learning_rate": 3.7096341072964113e-07,
|
6564 |
+
"loss": 0.2613,
|
6565 |
+
"step": 932
|
6566 |
+
},
|
6567 |
+
{
|
6568 |
+
"epoch": 0.9739039665970772,
|
6569 |
+
"grad_norm": 0.5585394501686096,
|
6570 |
+
"learning_rate": 3.4299253644732855e-07,
|
6571 |
+
"loss": 0.2097,
|
6572 |
+
"step": 933
|
6573 |
+
},
|
6574 |
+
{
|
6575 |
+
"epoch": 0.9749478079331941,
|
6576 |
+
"grad_norm": 0.7092443704605103,
|
6577 |
+
"learning_rate": 3.161160985304168e-07,
|
6578 |
+
"loss": 0.2731,
|
6579 |
+
"step": 934
|
6580 |
+
},
|
6581 |
+
{
|
6582 |
+
"epoch": 0.975991649269311,
|
6583 |
+
"grad_norm": 0.8628230094909668,
|
6584 |
+
"learning_rate": 2.903343921367707e-07,
|
6585 |
+
"loss": 0.3337,
|
6586 |
+
"step": 935
|
6587 |
+
},
|
6588 |
+
{
|
6589 |
+
"epoch": 0.9770354906054279,
|
6590 |
+
"grad_norm": 0.7503064870834351,
|
6591 |
+
"learning_rate": 2.6564770040190535e-07,
|
6592 |
+
"loss": 0.2528,
|
6593 |
+
"step": 936
|
6594 |
+
},
|
6595 |
+
{
|
6596 |
+
"epoch": 0.9780793319415448,
|
6597 |
+
"grad_norm": 0.6571477651596069,
|
6598 |
+
"learning_rate": 2.420562944358329e-07,
|
6599 |
+
"loss": 0.2097,
|
6600 |
+
"step": 937
|
6601 |
+
},
|
6602 |
+
{
|
6603 |
+
"epoch": 0.9791231732776617,
|
6604 |
+
"grad_norm": 0.7250248193740845,
|
6605 |
+
"learning_rate": 2.1956043332010955e-07,
|
6606 |
+
"loss": 0.2418,
|
6607 |
+
"step": 938
|
6608 |
+
},
|
6609 |
+
{
|
6610 |
+
"epoch": 0.9801670146137788,
|
6611 |
+
"grad_norm": 0.5199112892150879,
|
6612 |
+
"learning_rate": 1.9816036410499338e-07,
|
6613 |
+
"loss": 0.1988,
|
6614 |
+
"step": 939
|
6615 |
+
},
|
6616 |
+
{
|
6617 |
+
"epoch": 0.9812108559498957,
|
6618 |
+
"grad_norm": 0.6716915369033813,
|
6619 |
+
"learning_rate": 1.7785632180670198e-07,
|
6620 |
+
"loss": 0.2049,
|
6621 |
+
"step": 940
|
6622 |
+
},
|
6623 |
+
{
|
6624 |
+
"epoch": 0.9822546972860126,
|
6625 |
+
"grad_norm": 0.6581385731697083,
|
6626 |
+
"learning_rate": 1.5864852940485898e-07,
|
6627 |
+
"loss": 0.2065,
|
6628 |
+
"step": 941
|
6629 |
+
},
|
6630 |
+
{
|
6631 |
+
"epoch": 0.9832985386221295,
|
6632 |
+
"grad_norm": 0.5942964553833008,
|
6633 |
+
"learning_rate": 1.405371978400516e-07,
|
6634 |
+
"loss": 0.188,
|
6635 |
+
"step": 942
|
6636 |
+
},
|
6637 |
+
{
|
6638 |
+
"epoch": 0.9843423799582464,
|
6639 |
+
"grad_norm": 0.6144523620605469,
|
6640 |
+
"learning_rate": 1.2352252601147697e-07,
|
6641 |
+
"loss": 0.1848,
|
6642 |
+
"step": 943
|
6643 |
+
},
|
6644 |
+
{
|
6645 |
+
"epoch": 0.9853862212943633,
|
6646 |
+
"grad_norm": 0.6265504956245422,
|
6647 |
+
"learning_rate": 1.0760470077479934e-07,
|
6648 |
+
"loss": 0.2032,
|
6649 |
+
"step": 944
|
6650 |
+
},
|
6651 |
+
{
|
6652 |
+
"epoch": 0.9864300626304802,
|
6653 |
+
"grad_norm": 0.5609498023986816,
|
6654 |
+
"learning_rate": 9.278389694006296e-08,
|
6655 |
+
"loss": 0.1657,
|
6656 |
+
"step": 945
|
6657 |
+
},
|
6658 |
+
{
|
6659 |
+
"epoch": 0.9874739039665971,
|
6660 |
+
"grad_norm": 0.5039533972740173,
|
6661 |
+
"learning_rate": 7.906027726981568e-08,
|
6662 |
+
"loss": 0.158,
|
6663 |
+
"step": 946
|
6664 |
+
},
|
6665 |
+
{
|
6666 |
+
"epoch": 0.988517745302714,
|
6667 |
+
"grad_norm": 0.5719618201255798,
|
6668 |
+
"learning_rate": 6.643399247725502e-08,
|
6669 |
+
"loss": 0.151,
|
6670 |
+
"step": 947
|
6671 |
+
},
|
6672 |
+
{
|
6673 |
+
"epoch": 0.9895615866388309,
|
6674 |
+
"grad_norm": 0.5313311815261841,
|
6675 |
+
"learning_rate": 5.490518122465149e-08,
|
6676 |
+
"loss": 0.1563,
|
6677 |
+
"step": 948
|
6678 |
+
},
|
6679 |
+
{
|
6680 |
+
"epoch": 0.9906054279749478,
|
6681 |
+
"grad_norm": 0.48120567202568054,
|
6682 |
+
"learning_rate": 4.447397012177224e-08,
|
6683 |
+
"loss": 0.1376,
|
6684 |
+
"step": 949
|
6685 |
+
},
|
6686 |
+
{
|
6687 |
+
"epoch": 0.9916492693110647,
|
6688 |
+
"grad_norm": 0.44713348150253296,
|
6689 |
+
"learning_rate": 3.5140473724482034e-08,
|
6690 |
+
"loss": 0.1198,
|
6691 |
+
"step": 950
|
6692 |
+
},
|
6693 |
+
{
|
6694 |
+
"epoch": 0.9926931106471816,
|
6695 |
+
"grad_norm": 0.6056554913520813,
|
6696 |
+
"learning_rate": 2.6904794533544332e-08,
|
6697 |
+
"loss": 0.2989,
|
6698 |
+
"step": 951
|
6699 |
+
},
|
6700 |
+
{
|
6701 |
+
"epoch": 0.9937369519832986,
|
6702 |
+
"grad_norm": 0.6082270741462708,
|
6703 |
+
"learning_rate": 1.976702299344435e-08,
|
6704 |
+
"loss": 0.2356,
|
6705 |
+
"step": 952
|
6706 |
+
},
|
6707 |
+
{
|
6708 |
+
"epoch": 0.9947807933194155,
|
6709 |
+
"grad_norm": 0.46945539116859436,
|
6710 |
+
"learning_rate": 1.3727237491412137e-08,
|
6711 |
+
"loss": 0.1697,
|
6712 |
+
"step": 953
|
6713 |
+
},
|
6714 |
+
{
|
6715 |
+
"epoch": 0.9958246346555324,
|
6716 |
+
"grad_norm": 0.5799762010574341,
|
6717 |
+
"learning_rate": 8.785504356556563e-09,
|
6718 |
+
"loss": 0.2196,
|
6719 |
+
"step": 954
|
6720 |
+
},
|
6721 |
+
{
|
6722 |
+
"epoch": 0.9968684759916493,
|
6723 |
+
"grad_norm": 0.6873183846473694,
|
6724 |
+
"learning_rate": 4.941877859143684e-09,
|
6725 |
+
"loss": 0.2819,
|
6726 |
+
"step": 955
|
6727 |
+
},
|
6728 |
+
{
|
6729 |
+
"epoch": 0.9979123173277662,
|
6730 |
+
"grad_norm": 0.6708718538284302,
|
6731 |
+
"learning_rate": 2.1964002100083312e-09,
|
6732 |
+
"loss": 0.2342,
|
6733 |
+
"step": 956
|
6734 |
+
},
|
6735 |
+
{
|
6736 |
+
"epoch": 0.9989561586638831,
|
6737 |
+
"grad_norm": 0.6889147758483887,
|
6738 |
+
"learning_rate": 5.491015600656013e-10,
|
6739 |
+
"loss": 0.2048,
|
6740 |
+
"step": 957
|
6741 |
+
},
|
6742 |
+
{
|
6743 |
+
"epoch": 1.0,
|
6744 |
+
"grad_norm": 0.7045766115188599,
|
6745 |
+
"learning_rate": 0.0,
|
6746 |
+
"loss": 0.1947,
|
6747 |
+
"step": 958
|
6748 |
}
|
6749 |
],
|
6750 |
"logging_steps": 1,
|
|
|
6759 |
"should_evaluate": false,
|
6760 |
"should_log": false,
|
6761 |
"should_save": true,
|
6762 |
+
"should_training_stop": true
|
6763 |
},
|
6764 |
"attributes": {}
|
6765 |
}
|
6766 |
},
|
6767 |
+
"total_flos": 5.355154213385011e+17,
|
6768 |
"train_batch_size": 2,
|
6769 |
"trial_name": null,
|
6770 |
"trial_params": null
|