Training in progress, step 919, 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 80792096
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1515b702eceb2f51b86a9fdf1a1ce0ea88ee1a6339e9a98a87eb3042e469511c
|
3 |
size 80792096
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 41460084
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d2f2298f7d9f22736b05dc0d3b31a935f9e586ee545c06b4aa4ae6dfb8c04c0e
|
3 |
size 41460084
|
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 sha256:028c7f31433a21c7b3b1e4d127a97a8c315b426a21e4b4211c587af6a44cbe05
|
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 sha256:35e84ece30fd0b4a6c153a215e13d9b4959be67da893dc5bfcddc6528f0c0d70
|
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": 230,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -4861,6 +4861,1609 @@
|
|
4861 |
"eval_samples_per_second": 15.867,
|
4862 |
"eval_steps_per_second": 7.954,
|
4863 |
"step": 690
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4864 |
}
|
4865 |
],
|
4866 |
"logging_steps": 1,
|
@@ -4875,12 +6478,12 @@
|
|
4875 |
"should_evaluate": false,
|
4876 |
"should_log": false,
|
4877 |
"should_save": true,
|
4878 |
-
"should_training_stop":
|
4879 |
},
|
4880 |
"attributes": {}
|
4881 |
}
|
4882 |
},
|
4883 |
-
"total_flos":
|
4884 |
"train_batch_size": 2,
|
4885 |
"trial_name": null,
|
4886 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.0,
|
5 |
"eval_steps": 230,
|
6 |
+
"global_step": 919,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
4861 |
"eval_samples_per_second": 15.867,
|
4862 |
"eval_steps_per_second": 7.954,
|
4863 |
"step": 690
|
4864 |
+
},
|
4865 |
+
{
|
4866 |
+
"epoch": 0.7519042437431991,
|
4867 |
+
"grad_norm": 0.21464985609054565,
|
4868 |
+
"learning_rate": 2.9472846472869298e-05,
|
4869 |
+
"loss": 0.0963,
|
4870 |
+
"step": 691
|
4871 |
+
},
|
4872 |
+
{
|
4873 |
+
"epoch": 0.7529923830250272,
|
4874 |
+
"grad_norm": 0.19391842186450958,
|
4875 |
+
"learning_rate": 2.922825253307947e-05,
|
4876 |
+
"loss": 0.078,
|
4877 |
+
"step": 692
|
4878 |
+
},
|
4879 |
+
{
|
4880 |
+
"epoch": 0.7540805223068553,
|
4881 |
+
"grad_norm": 0.5395727157592773,
|
4882 |
+
"learning_rate": 2.898450393337977e-05,
|
4883 |
+
"loss": 0.2525,
|
4884 |
+
"step": 693
|
4885 |
+
},
|
4886 |
+
{
|
4887 |
+
"epoch": 0.7551686615886833,
|
4888 |
+
"grad_norm": 0.25584131479263306,
|
4889 |
+
"learning_rate": 2.874160358524931e-05,
|
4890 |
+
"loss": 0.1305,
|
4891 |
+
"step": 694
|
4892 |
+
},
|
4893 |
+
{
|
4894 |
+
"epoch": 0.7562568008705114,
|
4895 |
+
"grad_norm": 0.37683942914009094,
|
4896 |
+
"learning_rate": 2.8499554390035143e-05,
|
4897 |
+
"loss": 0.1649,
|
4898 |
+
"step": 695
|
4899 |
+
},
|
4900 |
+
{
|
4901 |
+
"epoch": 0.7573449401523396,
|
4902 |
+
"grad_norm": 0.5291657447814941,
|
4903 |
+
"learning_rate": 2.8258359238917665e-05,
|
4904 |
+
"loss": 0.401,
|
4905 |
+
"step": 696
|
4906 |
+
},
|
4907 |
+
{
|
4908 |
+
"epoch": 0.7584330794341676,
|
4909 |
+
"grad_norm": 0.3864194452762604,
|
4910 |
+
"learning_rate": 2.8018021012875994e-05,
|
4911 |
+
"loss": 0.1678,
|
4912 |
+
"step": 697
|
4913 |
+
},
|
4914 |
+
{
|
4915 |
+
"epoch": 0.7595212187159956,
|
4916 |
+
"grad_norm": 0.582097589969635,
|
4917 |
+
"learning_rate": 2.7778542582653744e-05,
|
4918 |
+
"loss": 0.3378,
|
4919 |
+
"step": 698
|
4920 |
+
},
|
4921 |
+
{
|
4922 |
+
"epoch": 0.7606093579978237,
|
4923 |
+
"grad_norm": 0.5892650485038757,
|
4924 |
+
"learning_rate": 2.753992680872457e-05,
|
4925 |
+
"loss": 0.3441,
|
4926 |
+
"step": 699
|
4927 |
+
},
|
4928 |
+
{
|
4929 |
+
"epoch": 0.7616974972796517,
|
4930 |
+
"grad_norm": 0.5687892436981201,
|
4931 |
+
"learning_rate": 2.7302176541257986e-05,
|
4932 |
+
"loss": 0.1971,
|
4933 |
+
"step": 700
|
4934 |
+
},
|
4935 |
+
{
|
4936 |
+
"epoch": 0.7627856365614799,
|
4937 |
+
"grad_norm": 0.05355902388691902,
|
4938 |
+
"learning_rate": 2.7065294620085424e-05,
|
4939 |
+
"loss": 0.04,
|
4940 |
+
"step": 701
|
4941 |
+
},
|
4942 |
+
{
|
4943 |
+
"epoch": 0.763873775843308,
|
4944 |
+
"grad_norm": 0.06114435940980911,
|
4945 |
+
"learning_rate": 2.6829283874666233e-05,
|
4946 |
+
"loss": 0.0591,
|
4947 |
+
"step": 702
|
4948 |
+
},
|
4949 |
+
{
|
4950 |
+
"epoch": 0.764961915125136,
|
4951 |
+
"grad_norm": 0.07170981913805008,
|
4952 |
+
"learning_rate": 2.659414712405398e-05,
|
4953 |
+
"loss": 0.0786,
|
4954 |
+
"step": 703
|
4955 |
+
},
|
4956 |
+
{
|
4957 |
+
"epoch": 0.766050054406964,
|
4958 |
+
"grad_norm": 0.06378500908613205,
|
4959 |
+
"learning_rate": 2.6359887176862718e-05,
|
4960 |
+
"loss": 0.0583,
|
4961 |
+
"step": 704
|
4962 |
+
},
|
4963 |
+
{
|
4964 |
+
"epoch": 0.7671381936887922,
|
4965 |
+
"grad_norm": 0.07413233816623688,
|
4966 |
+
"learning_rate": 2.6126506831233344e-05,
|
4967 |
+
"loss": 0.0649,
|
4968 |
+
"step": 705
|
4969 |
+
},
|
4970 |
+
{
|
4971 |
+
"epoch": 0.7682263329706203,
|
4972 |
+
"grad_norm": 0.08822925388813019,
|
4973 |
+
"learning_rate": 2.5894008874800325e-05,
|
4974 |
+
"loss": 0.0867,
|
4975 |
+
"step": 706
|
4976 |
+
},
|
4977 |
+
{
|
4978 |
+
"epoch": 0.7693144722524483,
|
4979 |
+
"grad_norm": 0.08858999609947205,
|
4980 |
+
"learning_rate": 2.566239608465838e-05,
|
4981 |
+
"loss": 0.097,
|
4982 |
+
"step": 707
|
4983 |
+
},
|
4984 |
+
{
|
4985 |
+
"epoch": 0.7704026115342764,
|
4986 |
+
"grad_norm": 0.08912277966737747,
|
4987 |
+
"learning_rate": 2.543167122732918e-05,
|
4988 |
+
"loss": 0.0789,
|
4989 |
+
"step": 708
|
4990 |
+
},
|
4991 |
+
{
|
4992 |
+
"epoch": 0.7714907508161044,
|
4993 |
+
"grad_norm": 0.09791934490203857,
|
4994 |
+
"learning_rate": 2.5201837058728505e-05,
|
4995 |
+
"loss": 0.0694,
|
4996 |
+
"step": 709
|
4997 |
+
},
|
4998 |
+
{
|
4999 |
+
"epoch": 0.7725788900979326,
|
5000 |
+
"grad_norm": 0.08719997853040695,
|
5001 |
+
"learning_rate": 2.4972896324133144e-05,
|
5002 |
+
"loss": 0.0815,
|
5003 |
+
"step": 710
|
5004 |
+
},
|
5005 |
+
{
|
5006 |
+
"epoch": 0.7736670293797606,
|
5007 |
+
"grad_norm": 0.09913161396980286,
|
5008 |
+
"learning_rate": 2.4744851758148156e-05,
|
5009 |
+
"loss": 0.0781,
|
5010 |
+
"step": 711
|
5011 |
+
},
|
5012 |
+
{
|
5013 |
+
"epoch": 0.7747551686615887,
|
5014 |
+
"grad_norm": 0.09172407537698746,
|
5015 |
+
"learning_rate": 2.451770608467432e-05,
|
5016 |
+
"loss": 0.0924,
|
5017 |
+
"step": 712
|
5018 |
+
},
|
5019 |
+
{
|
5020 |
+
"epoch": 0.7758433079434167,
|
5021 |
+
"grad_norm": 0.08026967942714691,
|
5022 |
+
"learning_rate": 2.429146201687538e-05,
|
5023 |
+
"loss": 0.0716,
|
5024 |
+
"step": 713
|
5025 |
+
},
|
5026 |
+
{
|
5027 |
+
"epoch": 0.7769314472252449,
|
5028 |
+
"grad_norm": 0.11445032805204391,
|
5029 |
+
"learning_rate": 2.4066122257145894e-05,
|
5030 |
+
"loss": 0.1015,
|
5031 |
+
"step": 714
|
5032 |
+
},
|
5033 |
+
{
|
5034 |
+
"epoch": 0.7780195865070729,
|
5035 |
+
"grad_norm": 0.09224435687065125,
|
5036 |
+
"learning_rate": 2.3841689497078746e-05,
|
5037 |
+
"loss": 0.0694,
|
5038 |
+
"step": 715
|
5039 |
+
},
|
5040 |
+
{
|
5041 |
+
"epoch": 0.779107725788901,
|
5042 |
+
"grad_norm": 0.08717647194862366,
|
5043 |
+
"learning_rate": 2.361816641743303e-05,
|
5044 |
+
"loss": 0.0592,
|
5045 |
+
"step": 716
|
5046 |
+
},
|
5047 |
+
{
|
5048 |
+
"epoch": 0.780195865070729,
|
5049 |
+
"grad_norm": 0.0966147854924202,
|
5050 |
+
"learning_rate": 2.339555568810221e-05,
|
5051 |
+
"loss": 0.0786,
|
5052 |
+
"step": 717
|
5053 |
+
},
|
5054 |
+
{
|
5055 |
+
"epoch": 0.7812840043525572,
|
5056 |
+
"grad_norm": 0.1080668717622757,
|
5057 |
+
"learning_rate": 2.3173859968081944e-05,
|
5058 |
+
"loss": 0.0977,
|
5059 |
+
"step": 718
|
5060 |
+
},
|
5061 |
+
{
|
5062 |
+
"epoch": 0.7823721436343852,
|
5063 |
+
"grad_norm": 0.10351711511611938,
|
5064 |
+
"learning_rate": 2.295308190543859e-05,
|
5065 |
+
"loss": 0.0867,
|
5066 |
+
"step": 719
|
5067 |
+
},
|
5068 |
+
{
|
5069 |
+
"epoch": 0.7834602829162133,
|
5070 |
+
"grad_norm": 0.11572374403476715,
|
5071 |
+
"learning_rate": 2.2733224137277366e-05,
|
5072 |
+
"loss": 0.0998,
|
5073 |
+
"step": 720
|
5074 |
+
},
|
5075 |
+
{
|
5076 |
+
"epoch": 0.7845484221980413,
|
5077 |
+
"grad_norm": 0.09316647052764893,
|
5078 |
+
"learning_rate": 2.251428928971102e-05,
|
5079 |
+
"loss": 0.0702,
|
5080 |
+
"step": 721
|
5081 |
+
},
|
5082 |
+
{
|
5083 |
+
"epoch": 0.7856365614798694,
|
5084 |
+
"grad_norm": 0.10631789267063141,
|
5085 |
+
"learning_rate": 2.2296279977828337e-05,
|
5086 |
+
"loss": 0.0631,
|
5087 |
+
"step": 722
|
5088 |
+
},
|
5089 |
+
{
|
5090 |
+
"epoch": 0.7867247007616975,
|
5091 |
+
"grad_norm": 0.11438383907079697,
|
5092 |
+
"learning_rate": 2.2079198805662914e-05,
|
5093 |
+
"loss": 0.0969,
|
5094 |
+
"step": 723
|
5095 |
+
},
|
5096 |
+
{
|
5097 |
+
"epoch": 0.7878128400435256,
|
5098 |
+
"grad_norm": 0.11713221669197083,
|
5099 |
+
"learning_rate": 2.1863048366162208e-05,
|
5100 |
+
"loss": 0.0966,
|
5101 |
+
"step": 724
|
5102 |
+
},
|
5103 |
+
{
|
5104 |
+
"epoch": 0.7889009793253536,
|
5105 |
+
"grad_norm": 0.12945635616779327,
|
5106 |
+
"learning_rate": 2.1647831241156302e-05,
|
5107 |
+
"loss": 0.1063,
|
5108 |
+
"step": 725
|
5109 |
+
},
|
5110 |
+
{
|
5111 |
+
"epoch": 0.7899891186071817,
|
5112 |
+
"grad_norm": 0.10376941412687302,
|
5113 |
+
"learning_rate": 2.1433550001327373e-05,
|
5114 |
+
"loss": 0.0658,
|
5115 |
+
"step": 726
|
5116 |
+
},
|
5117 |
+
{
|
5118 |
+
"epoch": 0.7910772578890098,
|
5119 |
+
"grad_norm": 0.13715630769729614,
|
5120 |
+
"learning_rate": 2.1220207206178688e-05,
|
5121 |
+
"loss": 0.1085,
|
5122 |
+
"step": 727
|
5123 |
+
},
|
5124 |
+
{
|
5125 |
+
"epoch": 0.7921653971708379,
|
5126 |
+
"grad_norm": 0.12354642152786255,
|
5127 |
+
"learning_rate": 2.1007805404004242e-05,
|
5128 |
+
"loss": 0.0897,
|
5129 |
+
"step": 728
|
5130 |
+
},
|
5131 |
+
{
|
5132 |
+
"epoch": 0.7932535364526659,
|
5133 |
+
"grad_norm": 0.12347663193941116,
|
5134 |
+
"learning_rate": 2.0796347131858186e-05,
|
5135 |
+
"loss": 0.086,
|
5136 |
+
"step": 729
|
5137 |
+
},
|
5138 |
+
{
|
5139 |
+
"epoch": 0.794341675734494,
|
5140 |
+
"grad_norm": 0.12429869920015335,
|
5141 |
+
"learning_rate": 2.058583491552465e-05,
|
5142 |
+
"loss": 0.1008,
|
5143 |
+
"step": 730
|
5144 |
+
},
|
5145 |
+
{
|
5146 |
+
"epoch": 0.795429815016322,
|
5147 |
+
"grad_norm": 0.12943844497203827,
|
5148 |
+
"learning_rate": 2.0376271269487514e-05,
|
5149 |
+
"loss": 0.0773,
|
5150 |
+
"step": 731
|
5151 |
+
},
|
5152 |
+
{
|
5153 |
+
"epoch": 0.7965179542981502,
|
5154 |
+
"grad_norm": 0.1839335411787033,
|
5155 |
+
"learning_rate": 2.0167658696900317e-05,
|
5156 |
+
"loss": 0.1115,
|
5157 |
+
"step": 732
|
5158 |
+
},
|
5159 |
+
{
|
5160 |
+
"epoch": 0.7976060935799782,
|
5161 |
+
"grad_norm": 0.15224863588809967,
|
5162 |
+
"learning_rate": 1.995999968955641e-05,
|
5163 |
+
"loss": 0.0847,
|
5164 |
+
"step": 733
|
5165 |
+
},
|
5166 |
+
{
|
5167 |
+
"epoch": 0.7986942328618063,
|
5168 |
+
"grad_norm": 0.19822123646736145,
|
5169 |
+
"learning_rate": 1.9753296727859195e-05,
|
5170 |
+
"loss": 0.1225,
|
5171 |
+
"step": 734
|
5172 |
+
},
|
5173 |
+
{
|
5174 |
+
"epoch": 0.7997823721436343,
|
5175 |
+
"grad_norm": 0.20507045090198517,
|
5176 |
+
"learning_rate": 1.9547552280792524e-05,
|
5177 |
+
"loss": 0.104,
|
5178 |
+
"step": 735
|
5179 |
+
},
|
5180 |
+
{
|
5181 |
+
"epoch": 0.8008705114254625,
|
5182 |
+
"grad_norm": 0.1983303725719452,
|
5183 |
+
"learning_rate": 1.9342768805891178e-05,
|
5184 |
+
"loss": 0.1208,
|
5185 |
+
"step": 736
|
5186 |
+
},
|
5187 |
+
{
|
5188 |
+
"epoch": 0.8019586507072906,
|
5189 |
+
"grad_norm": 0.18087254464626312,
|
5190 |
+
"learning_rate": 1.9138948749211472e-05,
|
5191 |
+
"loss": 0.1085,
|
5192 |
+
"step": 737
|
5193 |
+
},
|
5194 |
+
{
|
5195 |
+
"epoch": 0.8030467899891186,
|
5196 |
+
"grad_norm": 0.20106813311576843,
|
5197 |
+
"learning_rate": 1.8936094545302095e-05,
|
5198 |
+
"loss": 0.106,
|
5199 |
+
"step": 738
|
5200 |
+
},
|
5201 |
+
{
|
5202 |
+
"epoch": 0.8041349292709467,
|
5203 |
+
"grad_norm": 0.2678329348564148,
|
5204 |
+
"learning_rate": 1.8734208617174988e-05,
|
5205 |
+
"loss": 0.1376,
|
5206 |
+
"step": 739
|
5207 |
+
},
|
5208 |
+
{
|
5209 |
+
"epoch": 0.8052230685527747,
|
5210 |
+
"grad_norm": 0.3087542951107025,
|
5211 |
+
"learning_rate": 1.8533293376276472e-05,
|
5212 |
+
"loss": 0.1685,
|
5213 |
+
"step": 740
|
5214 |
+
},
|
5215 |
+
{
|
5216 |
+
"epoch": 0.8063112078346029,
|
5217 |
+
"grad_norm": 0.21778535842895508,
|
5218 |
+
"learning_rate": 1.8333351222458407e-05,
|
5219 |
+
"loss": 0.1088,
|
5220 |
+
"step": 741
|
5221 |
+
},
|
5222 |
+
{
|
5223 |
+
"epoch": 0.8073993471164309,
|
5224 |
+
"grad_norm": 0.30875080823898315,
|
5225 |
+
"learning_rate": 1.8134384543949478e-05,
|
5226 |
+
"loss": 0.1272,
|
5227 |
+
"step": 742
|
5228 |
+
},
|
5229 |
+
{
|
5230 |
+
"epoch": 0.808487486398259,
|
5231 |
+
"grad_norm": 0.28981634974479675,
|
5232 |
+
"learning_rate": 1.7936395717326704e-05,
|
5233 |
+
"loss": 0.0876,
|
5234 |
+
"step": 743
|
5235 |
+
},
|
5236 |
+
{
|
5237 |
+
"epoch": 0.809575625680087,
|
5238 |
+
"grad_norm": 0.4402623474597931,
|
5239 |
+
"learning_rate": 1.773938710748706e-05,
|
5240 |
+
"loss": 0.1671,
|
5241 |
+
"step": 744
|
5242 |
+
},
|
5243 |
+
{
|
5244 |
+
"epoch": 0.8106637649619152,
|
5245 |
+
"grad_norm": 0.3520585894584656,
|
5246 |
+
"learning_rate": 1.754336106761927e-05,
|
5247 |
+
"loss": 0.1063,
|
5248 |
+
"step": 745
|
5249 |
+
},
|
5250 |
+
{
|
5251 |
+
"epoch": 0.8117519042437432,
|
5252 |
+
"grad_norm": 0.31854307651519775,
|
5253 |
+
"learning_rate": 1.7348319939175637e-05,
|
5254 |
+
"loss": 0.1521,
|
5255 |
+
"step": 746
|
5256 |
+
},
|
5257 |
+
{
|
5258 |
+
"epoch": 0.8128400435255713,
|
5259 |
+
"grad_norm": 0.4468978941440582,
|
5260 |
+
"learning_rate": 1.715426605184407e-05,
|
5261 |
+
"loss": 0.2484,
|
5262 |
+
"step": 747
|
5263 |
+
},
|
5264 |
+
{
|
5265 |
+
"epoch": 0.8139281828073993,
|
5266 |
+
"grad_norm": 0.385442316532135,
|
5267 |
+
"learning_rate": 1.696120172352025e-05,
|
5268 |
+
"loss": 0.2066,
|
5269 |
+
"step": 748
|
5270 |
+
},
|
5271 |
+
{
|
5272 |
+
"epoch": 0.8150163220892275,
|
5273 |
+
"grad_norm": 0.5463036894798279,
|
5274 |
+
"learning_rate": 1.676912926028007e-05,
|
5275 |
+
"loss": 0.2398,
|
5276 |
+
"step": 749
|
5277 |
+
},
|
5278 |
+
{
|
5279 |
+
"epoch": 0.8161044613710555,
|
5280 |
+
"grad_norm": 0.8237993717193604,
|
5281 |
+
"learning_rate": 1.6578050956351886e-05,
|
5282 |
+
"loss": 0.2319,
|
5283 |
+
"step": 750
|
5284 |
+
},
|
5285 |
+
{
|
5286 |
+
"epoch": 0.8171926006528836,
|
5287 |
+
"grad_norm": 0.056420858949422836,
|
5288 |
+
"learning_rate": 1.6387969094089316e-05,
|
5289 |
+
"loss": 0.0726,
|
5290 |
+
"step": 751
|
5291 |
+
},
|
5292 |
+
{
|
5293 |
+
"epoch": 0.8182807399347116,
|
5294 |
+
"grad_norm": 0.0559864416718483,
|
5295 |
+
"learning_rate": 1.619888594394382e-05,
|
5296 |
+
"loss": 0.0535,
|
5297 |
+
"step": 752
|
5298 |
+
},
|
5299 |
+
{
|
5300 |
+
"epoch": 0.8193688792165397,
|
5301 |
+
"grad_norm": 0.0695619061589241,
|
5302 |
+
"learning_rate": 1.601080376443763e-05,
|
5303 |
+
"loss": 0.0764,
|
5304 |
+
"step": 753
|
5305 |
+
},
|
5306 |
+
{
|
5307 |
+
"epoch": 0.8204570184983678,
|
5308 |
+
"grad_norm": 0.06585152447223663,
|
5309 |
+
"learning_rate": 1.5823724802136865e-05,
|
5310 |
+
"loss": 0.0583,
|
5311 |
+
"step": 754
|
5312 |
+
},
|
5313 |
+
{
|
5314 |
+
"epoch": 0.8215451577801959,
|
5315 |
+
"grad_norm": 0.09294595569372177,
|
5316 |
+
"learning_rate": 1.5637651291624523e-05,
|
5317 |
+
"loss": 0.0952,
|
5318 |
+
"step": 755
|
5319 |
+
},
|
5320 |
+
{
|
5321 |
+
"epoch": 0.8226332970620239,
|
5322 |
+
"grad_norm": 0.07608404755592346,
|
5323 |
+
"learning_rate": 1.5452585455473977e-05,
|
5324 |
+
"loss": 0.0665,
|
5325 |
+
"step": 756
|
5326 |
+
},
|
5327 |
+
{
|
5328 |
+
"epoch": 0.823721436343852,
|
5329 |
+
"grad_norm": 0.09707889705896378,
|
5330 |
+
"learning_rate": 1.526852950422226e-05,
|
5331 |
+
"loss": 0.101,
|
5332 |
+
"step": 757
|
5333 |
+
},
|
5334 |
+
{
|
5335 |
+
"epoch": 0.8248095756256801,
|
5336 |
+
"grad_norm": 0.086356520652771,
|
5337 |
+
"learning_rate": 1.5085485636343755e-05,
|
5338 |
+
"loss": 0.0713,
|
5339 |
+
"step": 758
|
5340 |
+
},
|
5341 |
+
{
|
5342 |
+
"epoch": 0.8258977149075082,
|
5343 |
+
"grad_norm": 0.10413988679647446,
|
5344 |
+
"learning_rate": 1.4903456038223939e-05,
|
5345 |
+
"loss": 0.0931,
|
5346 |
+
"step": 759
|
5347 |
+
},
|
5348 |
+
{
|
5349 |
+
"epoch": 0.8269858541893362,
|
5350 |
+
"grad_norm": 0.11106099933385849,
|
5351 |
+
"learning_rate": 1.4722442884133214e-05,
|
5352 |
+
"loss": 0.0637,
|
5353 |
+
"step": 760
|
5354 |
+
},
|
5355 |
+
{
|
5356 |
+
"epoch": 0.8280739934711643,
|
5357 |
+
"grad_norm": 0.07780591398477554,
|
5358 |
+
"learning_rate": 1.454244833620102e-05,
|
5359 |
+
"loss": 0.0736,
|
5360 |
+
"step": 761
|
5361 |
+
},
|
5362 |
+
{
|
5363 |
+
"epoch": 0.8291621327529923,
|
5364 |
+
"grad_norm": 0.08770725876092911,
|
5365 |
+
"learning_rate": 1.4363474544389877e-05,
|
5366 |
+
"loss": 0.0743,
|
5367 |
+
"step": 762
|
5368 |
+
},
|
5369 |
+
{
|
5370 |
+
"epoch": 0.8302502720348205,
|
5371 |
+
"grad_norm": 0.10190238058567047,
|
5372 |
+
"learning_rate": 1.4185523646469822e-05,
|
5373 |
+
"loss": 0.0983,
|
5374 |
+
"step": 763
|
5375 |
+
},
|
5376 |
+
{
|
5377 |
+
"epoch": 0.8313384113166485,
|
5378 |
+
"grad_norm": 0.09307550638914108,
|
5379 |
+
"learning_rate": 1.4008597767992871e-05,
|
5380 |
+
"loss": 0.0814,
|
5381 |
+
"step": 764
|
5382 |
+
},
|
5383 |
+
{
|
5384 |
+
"epoch": 0.8324265505984766,
|
5385 |
+
"grad_norm": 0.1155632883310318,
|
5386 |
+
"learning_rate": 1.3832699022267515e-05,
|
5387 |
+
"loss": 0.0876,
|
5388 |
+
"step": 765
|
5389 |
+
},
|
5390 |
+
{
|
5391 |
+
"epoch": 0.8335146898803046,
|
5392 |
+
"grad_norm": 0.09275174140930176,
|
5393 |
+
"learning_rate": 1.3657829510333654e-05,
|
5394 |
+
"loss": 0.0754,
|
5395 |
+
"step": 766
|
5396 |
+
},
|
5397 |
+
{
|
5398 |
+
"epoch": 0.8346028291621328,
|
5399 |
+
"grad_norm": 0.08752154558897018,
|
5400 |
+
"learning_rate": 1.3483991320937306e-05,
|
5401 |
+
"loss": 0.0775,
|
5402 |
+
"step": 767
|
5403 |
+
},
|
5404 |
+
{
|
5405 |
+
"epoch": 0.8356909684439608,
|
5406 |
+
"grad_norm": 0.10071904957294464,
|
5407 |
+
"learning_rate": 1.3311186530505838e-05,
|
5408 |
+
"loss": 0.0957,
|
5409 |
+
"step": 768
|
5410 |
+
},
|
5411 |
+
{
|
5412 |
+
"epoch": 0.8367791077257889,
|
5413 |
+
"grad_norm": 0.11828504502773285,
|
5414 |
+
"learning_rate": 1.3139417203123027e-05,
|
5415 |
+
"loss": 0.0915,
|
5416 |
+
"step": 769
|
5417 |
+
},
|
5418 |
+
{
|
5419 |
+
"epoch": 0.8378672470076169,
|
5420 |
+
"grad_norm": 0.1135001927614212,
|
5421 |
+
"learning_rate": 1.2968685390504465e-05,
|
5422 |
+
"loss": 0.0925,
|
5423 |
+
"step": 770
|
5424 |
+
},
|
5425 |
+
{
|
5426 |
+
"epoch": 0.8389553862894451,
|
5427 |
+
"grad_norm": 0.10750744491815567,
|
5428 |
+
"learning_rate": 1.2798993131973091e-05,
|
5429 |
+
"loss": 0.0955,
|
5430 |
+
"step": 771
|
5431 |
+
},
|
5432 |
+
{
|
5433 |
+
"epoch": 0.8400435255712732,
|
5434 |
+
"grad_norm": 0.1162073016166687,
|
5435 |
+
"learning_rate": 1.263034245443473e-05,
|
5436 |
+
"loss": 0.089,
|
5437 |
+
"step": 772
|
5438 |
+
},
|
5439 |
+
{
|
5440 |
+
"epoch": 0.8411316648531012,
|
5441 |
+
"grad_norm": 0.12537071108818054,
|
5442 |
+
"learning_rate": 1.2462735372353996e-05,
|
5443 |
+
"loss": 0.0955,
|
5444 |
+
"step": 773
|
5445 |
+
},
|
5446 |
+
{
|
5447 |
+
"epoch": 0.8422198041349293,
|
5448 |
+
"grad_norm": 0.13251665234565735,
|
5449 |
+
"learning_rate": 1.2296173887730123e-05,
|
5450 |
+
"loss": 0.1036,
|
5451 |
+
"step": 774
|
5452 |
+
},
|
5453 |
+
{
|
5454 |
+
"epoch": 0.8433079434167573,
|
5455 |
+
"grad_norm": 0.13233721256256104,
|
5456 |
+
"learning_rate": 1.2130659990073146e-05,
|
5457 |
+
"loss": 0.1139,
|
5458 |
+
"step": 775
|
5459 |
+
},
|
5460 |
+
{
|
5461 |
+
"epoch": 0.8443960826985855,
|
5462 |
+
"grad_norm": 0.1206088662147522,
|
5463 |
+
"learning_rate": 1.1966195656380031e-05,
|
5464 |
+
"loss": 0.0853,
|
5465 |
+
"step": 776
|
5466 |
+
},
|
5467 |
+
{
|
5468 |
+
"epoch": 0.8454842219804135,
|
5469 |
+
"grad_norm": 0.10946936160326004,
|
5470 |
+
"learning_rate": 1.1802782851111205e-05,
|
5471 |
+
"loss": 0.074,
|
5472 |
+
"step": 777
|
5473 |
+
},
|
5474 |
+
{
|
5475 |
+
"epoch": 0.8465723612622416,
|
5476 |
+
"grad_norm": 0.1602969467639923,
|
5477 |
+
"learning_rate": 1.1640423526166988e-05,
|
5478 |
+
"loss": 0.1153,
|
5479 |
+
"step": 778
|
5480 |
+
},
|
5481 |
+
{
|
5482 |
+
"epoch": 0.8476605005440696,
|
5483 |
+
"grad_norm": 0.1147325336933136,
|
5484 |
+
"learning_rate": 1.1479119620864276e-05,
|
5485 |
+
"loss": 0.0542,
|
5486 |
+
"step": 779
|
5487 |
+
},
|
5488 |
+
{
|
5489 |
+
"epoch": 0.8487486398258978,
|
5490 |
+
"grad_norm": 0.16215886175632477,
|
5491 |
+
"learning_rate": 1.1318873061913405e-05,
|
5492 |
+
"loss": 0.1048,
|
5493 |
+
"step": 780
|
5494 |
+
},
|
5495 |
+
{
|
5496 |
+
"epoch": 0.8498367791077258,
|
5497 |
+
"grad_norm": 0.15750151872634888,
|
5498 |
+
"learning_rate": 1.1159685763395111e-05,
|
5499 |
+
"loss": 0.1229,
|
5500 |
+
"step": 781
|
5501 |
+
},
|
5502 |
+
{
|
5503 |
+
"epoch": 0.8509249183895539,
|
5504 |
+
"grad_norm": 0.16287699341773987,
|
5505 |
+
"learning_rate": 1.1001559626737756e-05,
|
5506 |
+
"loss": 0.0973,
|
5507 |
+
"step": 782
|
5508 |
+
},
|
5509 |
+
{
|
5510 |
+
"epoch": 0.8520130576713819,
|
5511 |
+
"grad_norm": 0.14996270835399628,
|
5512 |
+
"learning_rate": 1.0844496540694515e-05,
|
5513 |
+
"loss": 0.0959,
|
5514 |
+
"step": 783
|
5515 |
+
},
|
5516 |
+
{
|
5517 |
+
"epoch": 0.85310119695321,
|
5518 |
+
"grad_norm": 0.2524786591529846,
|
5519 |
+
"learning_rate": 1.0688498381320855e-05,
|
5520 |
+
"loss": 0.255,
|
5521 |
+
"step": 784
|
5522 |
+
},
|
5523 |
+
{
|
5524 |
+
"epoch": 0.8541893362350381,
|
5525 |
+
"grad_norm": 0.19718289375305176,
|
5526 |
+
"learning_rate": 1.0533567011952094e-05,
|
5527 |
+
"loss": 0.1131,
|
5528 |
+
"step": 785
|
5529 |
+
},
|
5530 |
+
{
|
5531 |
+
"epoch": 0.8552774755168662,
|
5532 |
+
"grad_norm": 0.21403737366199493,
|
5533 |
+
"learning_rate": 1.0379704283181179e-05,
|
5534 |
+
"loss": 0.1055,
|
5535 |
+
"step": 786
|
5536 |
+
},
|
5537 |
+
{
|
5538 |
+
"epoch": 0.8563656147986942,
|
5539 |
+
"grad_norm": 0.2776722013950348,
|
5540 |
+
"learning_rate": 1.0226912032836611e-05,
|
5541 |
+
"loss": 0.1149,
|
5542 |
+
"step": 787
|
5543 |
+
},
|
5544 |
+
{
|
5545 |
+
"epoch": 0.8574537540805223,
|
5546 |
+
"grad_norm": 0.2056991308927536,
|
5547 |
+
"learning_rate": 1.007519208596045e-05,
|
5548 |
+
"loss": 0.078,
|
5549 |
+
"step": 788
|
5550 |
+
},
|
5551 |
+
{
|
5552 |
+
"epoch": 0.8585418933623504,
|
5553 |
+
"grad_norm": 0.3096133768558502,
|
5554 |
+
"learning_rate": 9.924546254786493e-06,
|
5555 |
+
"loss": 0.1237,
|
5556 |
+
"step": 789
|
5557 |
+
},
|
5558 |
+
{
|
5559 |
+
"epoch": 0.8596300326441785,
|
5560 |
+
"grad_norm": 0.38796815276145935,
|
5561 |
+
"learning_rate": 9.774976338718677e-06,
|
5562 |
+
"loss": 0.1631,
|
5563 |
+
"step": 790
|
5564 |
+
},
|
5565 |
+
{
|
5566 |
+
"epoch": 0.8607181719260065,
|
5567 |
+
"grad_norm": 0.2669233977794647,
|
5568 |
+
"learning_rate": 9.62648412430951e-06,
|
5569 |
+
"loss": 0.1011,
|
5570 |
+
"step": 791
|
5571 |
+
},
|
5572 |
+
{
|
5573 |
+
"epoch": 0.8618063112078346,
|
5574 |
+
"grad_norm": 0.4013719856739044,
|
5575 |
+
"learning_rate": 9.479071385238892e-06,
|
5576 |
+
"loss": 0.2422,
|
5577 |
+
"step": 792
|
5578 |
+
},
|
5579 |
+
{
|
5580 |
+
"epoch": 0.8628944504896626,
|
5581 |
+
"grad_norm": 0.35241514444351196,
|
5582 |
+
"learning_rate": 9.332739882292752e-06,
|
5583 |
+
"loss": 0.1604,
|
5584 |
+
"step": 793
|
5585 |
+
},
|
5586 |
+
{
|
5587 |
+
"epoch": 0.8639825897714908,
|
5588 |
+
"grad_norm": 0.3863401710987091,
|
5589 |
+
"learning_rate": 9.187491363342093e-06,
|
5590 |
+
"loss": 0.1977,
|
5591 |
+
"step": 794
|
5592 |
+
},
|
5593 |
+
{
|
5594 |
+
"epoch": 0.8650707290533188,
|
5595 |
+
"grad_norm": 0.4365151524543762,
|
5596 |
+
"learning_rate": 9.043327563322112e-06,
|
5597 |
+
"loss": 0.2164,
|
5598 |
+
"step": 795
|
5599 |
+
},
|
5600 |
+
{
|
5601 |
+
"epoch": 0.8661588683351469,
|
5602 |
+
"grad_norm": 0.4250886142253876,
|
5603 |
+
"learning_rate": 8.900250204211514e-06,
|
5604 |
+
"loss": 0.2607,
|
5605 |
+
"step": 796
|
5606 |
+
},
|
5607 |
+
{
|
5608 |
+
"epoch": 0.8672470076169749,
|
5609 |
+
"grad_norm": 0.6903502345085144,
|
5610 |
+
"learning_rate": 8.758260995011825e-06,
|
5611 |
+
"loss": 0.2994,
|
5612 |
+
"step": 797
|
5613 |
+
},
|
5614 |
+
{
|
5615 |
+
"epoch": 0.8683351468988031,
|
5616 |
+
"grad_norm": 0.6091783046722412,
|
5617 |
+
"learning_rate": 8.617361631727138e-06,
|
5618 |
+
"loss": 0.3278,
|
5619 |
+
"step": 798
|
5620 |
+
},
|
5621 |
+
{
|
5622 |
+
"epoch": 0.8694232861806311,
|
5623 |
+
"grad_norm": 0.7825955152511597,
|
5624 |
+
"learning_rate": 8.47755379734373e-06,
|
5625 |
+
"loss": 0.2998,
|
5626 |
+
"step": 799
|
5627 |
+
},
|
5628 |
+
{
|
5629 |
+
"epoch": 0.8705114254624592,
|
5630 |
+
"grad_norm": 0.5094212293624878,
|
5631 |
+
"learning_rate": 8.338839161809997e-06,
|
5632 |
+
"loss": 0.1437,
|
5633 |
+
"step": 800
|
5634 |
+
},
|
5635 |
+
{
|
5636 |
+
"epoch": 0.8715995647442872,
|
5637 |
+
"grad_norm": 0.047669120132923126,
|
5638 |
+
"learning_rate": 8.201219382016556e-06,
|
5639 |
+
"loss": 0.0489,
|
5640 |
+
"step": 801
|
5641 |
+
},
|
5642 |
+
{
|
5643 |
+
"epoch": 0.8726877040261154,
|
5644 |
+
"grad_norm": 0.06768417358398438,
|
5645 |
+
"learning_rate": 8.064696101776358e-06,
|
5646 |
+
"loss": 0.0689,
|
5647 |
+
"step": 802
|
5648 |
+
},
|
5649 |
+
{
|
5650 |
+
"epoch": 0.8737758433079434,
|
5651 |
+
"grad_norm": 0.06095738708972931,
|
5652 |
+
"learning_rate": 7.929270951805178e-06,
|
5653 |
+
"loss": 0.0639,
|
5654 |
+
"step": 803
|
5655 |
+
},
|
5656 |
+
{
|
5657 |
+
"epoch": 0.8748639825897715,
|
5658 |
+
"grad_norm": 0.07757483422756195,
|
5659 |
+
"learning_rate": 7.794945549701993e-06,
|
5660 |
+
"loss": 0.077,
|
5661 |
+
"step": 804
|
5662 |
+
},
|
5663 |
+
{
|
5664 |
+
"epoch": 0.8759521218715995,
|
5665 |
+
"grad_norm": 0.07621616870164871,
|
5666 |
+
"learning_rate": 7.661721499929753e-06,
|
5667 |
+
"loss": 0.0679,
|
5668 |
+
"step": 805
|
5669 |
+
},
|
5670 |
+
{
|
5671 |
+
"epoch": 0.8770402611534276,
|
5672 |
+
"grad_norm": 0.06622826308012009,
|
5673 |
+
"learning_rate": 7.529600393796232e-06,
|
5674 |
+
"loss": 0.0554,
|
5675 |
+
"step": 806
|
5676 |
+
},
|
5677 |
+
{
|
5678 |
+
"epoch": 0.8781284004352558,
|
5679 |
+
"grad_norm": 0.0733778178691864,
|
5680 |
+
"learning_rate": 7.3985838094349444e-06,
|
5681 |
+
"loss": 0.0443,
|
5682 |
+
"step": 807
|
5683 |
+
},
|
5684 |
+
{
|
5685 |
+
"epoch": 0.8792165397170838,
|
5686 |
+
"grad_norm": 0.0740467980504036,
|
5687 |
+
"learning_rate": 7.2686733117863784e-06,
|
5688 |
+
"loss": 0.0663,
|
5689 |
+
"step": 808
|
5690 |
+
},
|
5691 |
+
{
|
5692 |
+
"epoch": 0.8803046789989118,
|
5693 |
+
"grad_norm": 0.073309525847435,
|
5694 |
+
"learning_rate": 7.1398704525792e-06,
|
5695 |
+
"loss": 0.0549,
|
5696 |
+
"step": 809
|
5697 |
+
},
|
5698 |
+
{
|
5699 |
+
"epoch": 0.8813928182807399,
|
5700 |
+
"grad_norm": 0.08254940807819366,
|
5701 |
+
"learning_rate": 7.012176770311862e-06,
|
5702 |
+
"loss": 0.0849,
|
5703 |
+
"step": 810
|
5704 |
+
},
|
5705 |
+
{
|
5706 |
+
"epoch": 0.8824809575625681,
|
5707 |
+
"grad_norm": 0.0972483903169632,
|
5708 |
+
"learning_rate": 6.8855937902340576e-06,
|
5709 |
+
"loss": 0.0861,
|
5710 |
+
"step": 811
|
5711 |
+
},
|
5712 |
+
{
|
5713 |
+
"epoch": 0.8835690968443961,
|
5714 |
+
"grad_norm": 0.08929561078548431,
|
5715 |
+
"learning_rate": 6.760123024328624e-06,
|
5716 |
+
"loss": 0.0626,
|
5717 |
+
"step": 812
|
5718 |
+
},
|
5719 |
+
{
|
5720 |
+
"epoch": 0.8846572361262242,
|
5721 |
+
"grad_norm": 0.08776423335075378,
|
5722 |
+
"learning_rate": 6.635765971293484e-06,
|
5723 |
+
"loss": 0.0781,
|
5724 |
+
"step": 813
|
5725 |
+
},
|
5726 |
+
{
|
5727 |
+
"epoch": 0.8857453754080522,
|
5728 |
+
"grad_norm": 0.09048158675432205,
|
5729 |
+
"learning_rate": 6.512524116523633e-06,
|
5730 |
+
"loss": 0.0748,
|
5731 |
+
"step": 814
|
5732 |
+
},
|
5733 |
+
{
|
5734 |
+
"epoch": 0.8868335146898803,
|
5735 |
+
"grad_norm": 0.09002123028039932,
|
5736 |
+
"learning_rate": 6.390398932093555e-06,
|
5737 |
+
"loss": 0.0743,
|
5738 |
+
"step": 815
|
5739 |
+
},
|
5740 |
+
{
|
5741 |
+
"epoch": 0.8879216539717084,
|
5742 |
+
"grad_norm": 0.08896075189113617,
|
5743 |
+
"learning_rate": 6.269391876739495e-06,
|
5744 |
+
"loss": 0.0759,
|
5745 |
+
"step": 816
|
5746 |
+
},
|
5747 |
+
{
|
5748 |
+
"epoch": 0.8890097932535365,
|
5749 |
+
"grad_norm": 0.09824143350124359,
|
5750 |
+
"learning_rate": 6.149504395842087e-06,
|
5751 |
+
"loss": 0.0944,
|
5752 |
+
"step": 817
|
5753 |
+
},
|
5754 |
+
{
|
5755 |
+
"epoch": 0.8900979325353645,
|
5756 |
+
"grad_norm": 0.11370905488729477,
|
5757 |
+
"learning_rate": 6.030737921409169e-06,
|
5758 |
+
"loss": 0.1098,
|
5759 |
+
"step": 818
|
5760 |
+
},
|
5761 |
+
{
|
5762 |
+
"epoch": 0.8911860718171926,
|
5763 |
+
"grad_norm": 0.09926322102546692,
|
5764 |
+
"learning_rate": 5.913093872058528e-06,
|
5765 |
+
"loss": 0.0721,
|
5766 |
+
"step": 819
|
5767 |
+
},
|
5768 |
+
{
|
5769 |
+
"epoch": 0.8922742110990207,
|
5770 |
+
"grad_norm": 0.11143568158149719,
|
5771 |
+
"learning_rate": 5.7965736530010916e-06,
|
5772 |
+
"loss": 0.0838,
|
5773 |
+
"step": 820
|
5774 |
+
},
|
5775 |
+
{
|
5776 |
+
"epoch": 0.8933623503808488,
|
5777 |
+
"grad_norm": 0.10661139339208603,
|
5778 |
+
"learning_rate": 5.681178656024055e-06,
|
5779 |
+
"loss": 0.0828,
|
5780 |
+
"step": 821
|
5781 |
+
},
|
5782 |
+
{
|
5783 |
+
"epoch": 0.8944504896626768,
|
5784 |
+
"grad_norm": 0.09252527356147766,
|
5785 |
+
"learning_rate": 5.566910259474289e-06,
|
5786 |
+
"loss": 0.0607,
|
5787 |
+
"step": 822
|
5788 |
+
},
|
5789 |
+
{
|
5790 |
+
"epoch": 0.8955386289445049,
|
5791 |
+
"grad_norm": 0.11243504285812378,
|
5792 |
+
"learning_rate": 5.453769828241872e-06,
|
5793 |
+
"loss": 0.0883,
|
5794 |
+
"step": 823
|
5795 |
+
},
|
5796 |
+
{
|
5797 |
+
"epoch": 0.8966267682263329,
|
5798 |
+
"grad_norm": 0.08846960961818695,
|
5799 |
+
"learning_rate": 5.341758713743828e-06,
|
5800 |
+
"loss": 0.0726,
|
5801 |
+
"step": 824
|
5802 |
+
},
|
5803 |
+
{
|
5804 |
+
"epoch": 0.8977149075081611,
|
5805 |
+
"grad_norm": 0.09837763756513596,
|
5806 |
+
"learning_rate": 5.230878253907912e-06,
|
5807 |
+
"loss": 0.0761,
|
5808 |
+
"step": 825
|
5809 |
+
},
|
5810 |
+
{
|
5811 |
+
"epoch": 0.8988030467899891,
|
5812 |
+
"grad_norm": 0.0819426029920578,
|
5813 |
+
"learning_rate": 5.121129773156663e-06,
|
5814 |
+
"loss": 0.0521,
|
5815 |
+
"step": 826
|
5816 |
+
},
|
5817 |
+
{
|
5818 |
+
"epoch": 0.8998911860718172,
|
5819 |
+
"grad_norm": 0.11781435459852219,
|
5820 |
+
"learning_rate": 5.012514582391592e-06,
|
5821 |
+
"loss": 0.0843,
|
5822 |
+
"step": 827
|
5823 |
+
},
|
5824 |
+
{
|
5825 |
+
"epoch": 0.9009793253536452,
|
5826 |
+
"grad_norm": 0.1131996214389801,
|
5827 |
+
"learning_rate": 4.905033978977491e-06,
|
5828 |
+
"loss": 0.0716,
|
5829 |
+
"step": 828
|
5830 |
+
},
|
5831 |
+
{
|
5832 |
+
"epoch": 0.9020674646354734,
|
5833 |
+
"grad_norm": 0.11639894545078278,
|
5834 |
+
"learning_rate": 4.798689246727006e-06,
|
5835 |
+
"loss": 0.0872,
|
5836 |
+
"step": 829
|
5837 |
+
},
|
5838 |
+
{
|
5839 |
+
"epoch": 0.9031556039173014,
|
5840 |
+
"grad_norm": 0.12089676409959793,
|
5841 |
+
"learning_rate": 4.693481655885257e-06,
|
5842 |
+
"loss": 0.0776,
|
5843 |
+
"step": 830
|
5844 |
+
},
|
5845 |
+
{
|
5846 |
+
"epoch": 0.9042437431991295,
|
5847 |
+
"grad_norm": 0.13454534113407135,
|
5848 |
+
"learning_rate": 4.58941246311464e-06,
|
5849 |
+
"loss": 0.0824,
|
5850 |
+
"step": 831
|
5851 |
+
},
|
5852 |
+
{
|
5853 |
+
"epoch": 0.9053318824809575,
|
5854 |
+
"grad_norm": 0.13089610636234283,
|
5855 |
+
"learning_rate": 4.486482911479839e-06,
|
5856 |
+
"loss": 0.0841,
|
5857 |
+
"step": 832
|
5858 |
+
},
|
5859 |
+
{
|
5860 |
+
"epoch": 0.9064200217627857,
|
5861 |
+
"grad_norm": 0.12023388594388962,
|
5862 |
+
"learning_rate": 4.384694230432984e-06,
|
5863 |
+
"loss": 0.0799,
|
5864 |
+
"step": 833
|
5865 |
+
},
|
5866 |
+
{
|
5867 |
+
"epoch": 0.9075081610446137,
|
5868 |
+
"grad_norm": 0.17653657495975494,
|
5869 |
+
"learning_rate": 4.2840476357989825e-06,
|
5870 |
+
"loss": 0.1006,
|
5871 |
+
"step": 834
|
5872 |
+
},
|
5873 |
+
{
|
5874 |
+
"epoch": 0.9085963003264418,
|
5875 |
+
"grad_norm": 0.16812124848365784,
|
5876 |
+
"learning_rate": 4.184544329761009e-06,
|
5877 |
+
"loss": 0.0959,
|
5878 |
+
"step": 835
|
5879 |
+
},
|
5880 |
+
{
|
5881 |
+
"epoch": 0.9096844396082698,
|
5882 |
+
"grad_norm": 0.12245675921440125,
|
5883 |
+
"learning_rate": 4.0861855008460405e-06,
|
5884 |
+
"loss": 0.0525,
|
5885 |
+
"step": 836
|
5886 |
+
},
|
5887 |
+
{
|
5888 |
+
"epoch": 0.9107725788900979,
|
5889 |
+
"grad_norm": 0.17066197097301483,
|
5890 |
+
"learning_rate": 3.988972323910778e-06,
|
5891 |
+
"loss": 0.0974,
|
5892 |
+
"step": 837
|
5893 |
+
},
|
5894 |
+
{
|
5895 |
+
"epoch": 0.911860718171926,
|
5896 |
+
"grad_norm": 0.23829351365566254,
|
5897 |
+
"learning_rate": 3.892905960127546e-06,
|
5898 |
+
"loss": 0.1489,
|
5899 |
+
"step": 838
|
5900 |
+
},
|
5901 |
+
{
|
5902 |
+
"epoch": 0.9129488574537541,
|
5903 |
+
"grad_norm": 0.18741947412490845,
|
5904 |
+
"learning_rate": 3.797987556970495e-06,
|
5905 |
+
"loss": 0.0845,
|
5906 |
+
"step": 839
|
5907 |
+
},
|
5908 |
+
{
|
5909 |
+
"epoch": 0.9140369967355821,
|
5910 |
+
"grad_norm": 0.31613054871559143,
|
5911 |
+
"learning_rate": 3.7042182482018075e-06,
|
5912 |
+
"loss": 0.1522,
|
5913 |
+
"step": 840
|
5914 |
+
},
|
5915 |
+
{
|
5916 |
+
"epoch": 0.9151251360174102,
|
5917 |
+
"grad_norm": 0.30818799138069153,
|
5918 |
+
"learning_rate": 3.611599153858214e-06,
|
5919 |
+
"loss": 0.1807,
|
5920 |
+
"step": 841
|
5921 |
+
},
|
5922 |
+
{
|
5923 |
+
"epoch": 0.9162132752992383,
|
5924 |
+
"grad_norm": 0.19653384387493134,
|
5925 |
+
"learning_rate": 3.5201313802375456e-06,
|
5926 |
+
"loss": 0.073,
|
5927 |
+
"step": 842
|
5928 |
+
},
|
5929 |
+
{
|
5930 |
+
"epoch": 0.9173014145810664,
|
5931 |
+
"grad_norm": 0.311813622713089,
|
5932 |
+
"learning_rate": 3.4298160198856568e-06,
|
5933 |
+
"loss": 0.1273,
|
5934 |
+
"step": 843
|
5935 |
+
},
|
5936 |
+
{
|
5937 |
+
"epoch": 0.9183895538628944,
|
5938 |
+
"grad_norm": 0.3088919520378113,
|
5939 |
+
"learning_rate": 3.3406541515832003e-06,
|
5940 |
+
"loss": 0.1707,
|
5941 |
+
"step": 844
|
5942 |
+
},
|
5943 |
+
{
|
5944 |
+
"epoch": 0.9194776931447225,
|
5945 |
+
"grad_norm": 0.34374913573265076,
|
5946 |
+
"learning_rate": 3.252646840332918e-06,
|
5947 |
+
"loss": 0.1326,
|
5948 |
+
"step": 845
|
5949 |
+
},
|
5950 |
+
{
|
5951 |
+
"epoch": 0.9205658324265505,
|
5952 |
+
"grad_norm": 0.4254063069820404,
|
5953 |
+
"learning_rate": 3.1657951373467497e-06,
|
5954 |
+
"loss": 0.2334,
|
5955 |
+
"step": 846
|
5956 |
+
},
|
5957 |
+
{
|
5958 |
+
"epoch": 0.9216539717083787,
|
5959 |
+
"grad_norm": 0.3956216275691986,
|
5960 |
+
"learning_rate": 3.0801000800333877e-06,
|
5961 |
+
"loss": 0.2152,
|
5962 |
+
"step": 847
|
5963 |
+
},
|
5964 |
+
{
|
5965 |
+
"epoch": 0.9227421109902068,
|
5966 |
+
"grad_norm": 0.41808605194091797,
|
5967 |
+
"learning_rate": 2.995562691985898e-06,
|
5968 |
+
"loss": 0.1875,
|
5969 |
+
"step": 848
|
5970 |
+
},
|
5971 |
+
{
|
5972 |
+
"epoch": 0.9238302502720348,
|
5973 |
+
"grad_norm": 0.48160499334335327,
|
5974 |
+
"learning_rate": 2.912183982969385e-06,
|
5975 |
+
"loss": 0.1849,
|
5976 |
+
"step": 849
|
5977 |
+
},
|
5978 |
+
{
|
5979 |
+
"epoch": 0.9249183895538629,
|
5980 |
+
"grad_norm": 0.5654423832893372,
|
5981 |
+
"learning_rate": 2.8299649489090475e-06,
|
5982 |
+
"loss": 0.175,
|
5983 |
+
"step": 850
|
5984 |
+
},
|
5985 |
+
{
|
5986 |
+
"epoch": 0.926006528835691,
|
5987 |
+
"grad_norm": 0.04587412625551224,
|
5988 |
+
"learning_rate": 2.748906571878207e-06,
|
5989 |
+
"loss": 0.0362,
|
5990 |
+
"step": 851
|
5991 |
+
},
|
5992 |
+
{
|
5993 |
+
"epoch": 0.9270946681175191,
|
5994 |
+
"grad_norm": 0.05238286778330803,
|
5995 |
+
"learning_rate": 2.6690098200866098e-06,
|
5996 |
+
"loss": 0.0459,
|
5997 |
+
"step": 852
|
5998 |
+
},
|
5999 |
+
{
|
6000 |
+
"epoch": 0.9281828073993471,
|
6001 |
+
"grad_norm": 0.05864802002906799,
|
6002 |
+
"learning_rate": 2.590275647868867e-06,
|
6003 |
+
"loss": 0.0568,
|
6004 |
+
"step": 853
|
6005 |
+
},
|
6006 |
+
{
|
6007 |
+
"epoch": 0.9292709466811752,
|
6008 |
+
"grad_norm": 0.06772205978631973,
|
6009 |
+
"learning_rate": 2.5127049956730207e-06,
|
6010 |
+
"loss": 0.0671,
|
6011 |
+
"step": 854
|
6012 |
+
},
|
6013 |
+
{
|
6014 |
+
"epoch": 0.9303590859630033,
|
6015 |
+
"grad_norm": 0.07428263127803802,
|
6016 |
+
"learning_rate": 2.436298790049363e-06,
|
6017 |
+
"loss": 0.0653,
|
6018 |
+
"step": 855
|
6019 |
+
},
|
6020 |
+
{
|
6021 |
+
"epoch": 0.9314472252448314,
|
6022 |
+
"grad_norm": 0.06716316193342209,
|
6023 |
+
"learning_rate": 2.3610579436393e-06,
|
6024 |
+
"loss": 0.0608,
|
6025 |
+
"step": 856
|
6026 |
+
},
|
6027 |
+
{
|
6028 |
+
"epoch": 0.9325353645266594,
|
6029 |
+
"grad_norm": 0.07774016261100769,
|
6030 |
+
"learning_rate": 2.286983355164529e-06,
|
6031 |
+
"loss": 0.0641,
|
6032 |
+
"step": 857
|
6033 |
+
},
|
6034 |
+
{
|
6035 |
+
"epoch": 0.9336235038084875,
|
6036 |
+
"grad_norm": 0.07007763534784317,
|
6037 |
+
"learning_rate": 2.2140759094162467e-06,
|
6038 |
+
"loss": 0.0688,
|
6039 |
+
"step": 858
|
6040 |
+
},
|
6041 |
+
{
|
6042 |
+
"epoch": 0.9347116430903155,
|
6043 |
+
"grad_norm": 0.08392661809921265,
|
6044 |
+
"learning_rate": 2.1423364772445887e-06,
|
6045 |
+
"loss": 0.0862,
|
6046 |
+
"step": 859
|
6047 |
+
},
|
6048 |
+
{
|
6049 |
+
"epoch": 0.9357997823721437,
|
6050 |
+
"grad_norm": 0.08616143465042114,
|
6051 |
+
"learning_rate": 2.0717659155482738e-06,
|
6052 |
+
"loss": 0.084,
|
6053 |
+
"step": 860
|
6054 |
+
},
|
6055 |
+
{
|
6056 |
+
"epoch": 0.9368879216539717,
|
6057 |
+
"grad_norm": 0.08230069279670715,
|
6058 |
+
"learning_rate": 2.002365067264289e-06,
|
6059 |
+
"loss": 0.0671,
|
6060 |
+
"step": 861
|
6061 |
+
},
|
6062 |
+
{
|
6063 |
+
"epoch": 0.9379760609357998,
|
6064 |
+
"grad_norm": 0.10698749870061874,
|
6065 |
+
"learning_rate": 1.9341347613579087e-06,
|
6066 |
+
"loss": 0.0856,
|
6067 |
+
"step": 862
|
6068 |
+
},
|
6069 |
+
{
|
6070 |
+
"epoch": 0.9390642002176278,
|
6071 |
+
"grad_norm": 0.07910988479852676,
|
6072 |
+
"learning_rate": 1.8670758128126909e-06,
|
6073 |
+
"loss": 0.0643,
|
6074 |
+
"step": 863
|
6075 |
+
},
|
6076 |
+
{
|
6077 |
+
"epoch": 0.940152339499456,
|
6078 |
+
"grad_norm": 0.08783937990665436,
|
6079 |
+
"learning_rate": 1.8011890226208527e-06,
|
6080 |
+
"loss": 0.0921,
|
6081 |
+
"step": 864
|
6082 |
+
},
|
6083 |
+
{
|
6084 |
+
"epoch": 0.941240478781284,
|
6085 |
+
"grad_norm": 0.09200912714004517,
|
6086 |
+
"learning_rate": 1.7364751777736332e-06,
|
6087 |
+
"loss": 0.0853,
|
6088 |
+
"step": 865
|
6089 |
+
},
|
6090 |
+
{
|
6091 |
+
"epoch": 0.9423286180631121,
|
6092 |
+
"grad_norm": 0.0886380597949028,
|
6093 |
+
"learning_rate": 1.6729350512519005e-06,
|
6094 |
+
"loss": 0.0655,
|
6095 |
+
"step": 866
|
6096 |
+
},
|
6097 |
+
{
|
6098 |
+
"epoch": 0.9434167573449401,
|
6099 |
+
"grad_norm": 0.08538145571947098,
|
6100 |
+
"learning_rate": 1.6105694020169593e-06,
|
6101 |
+
"loss": 0.069,
|
6102 |
+
"step": 867
|
6103 |
+
},
|
6104 |
+
{
|
6105 |
+
"epoch": 0.9445048966267682,
|
6106 |
+
"grad_norm": 0.10413581132888794,
|
6107 |
+
"learning_rate": 1.5493789750014031e-06,
|
6108 |
+
"loss": 0.0789,
|
6109 |
+
"step": 868
|
6110 |
+
},
|
6111 |
+
{
|
6112 |
+
"epoch": 0.9455930359085963,
|
6113 |
+
"grad_norm": 0.09816232323646545,
|
6114 |
+
"learning_rate": 1.489364501100332e-06,
|
6115 |
+
"loss": 0.0749,
|
6116 |
+
"step": 869
|
6117 |
+
},
|
6118 |
+
{
|
6119 |
+
"epoch": 0.9466811751904244,
|
6120 |
+
"grad_norm": 0.10278590023517609,
|
6121 |
+
"learning_rate": 1.430526697162482e-06,
|
6122 |
+
"loss": 0.0895,
|
6123 |
+
"step": 870
|
6124 |
+
},
|
6125 |
+
{
|
6126 |
+
"epoch": 0.9477693144722524,
|
6127 |
+
"grad_norm": 0.10024615377187729,
|
6128 |
+
"learning_rate": 1.3728662659818204e-06,
|
6129 |
+
"loss": 0.0869,
|
6130 |
+
"step": 871
|
6131 |
+
},
|
6132 |
+
{
|
6133 |
+
"epoch": 0.9488574537540805,
|
6134 |
+
"grad_norm": 0.08774451911449432,
|
6135 |
+
"learning_rate": 1.3163838962890195e-06,
|
6136 |
+
"loss": 0.0631,
|
6137 |
+
"step": 872
|
6138 |
+
},
|
6139 |
+
{
|
6140 |
+
"epoch": 0.9499455930359086,
|
6141 |
+
"grad_norm": 0.08854610472917557,
|
6142 |
+
"learning_rate": 1.261080262743297e-06,
|
6143 |
+
"loss": 0.061,
|
6144 |
+
"step": 873
|
6145 |
+
},
|
6146 |
+
{
|
6147 |
+
"epoch": 0.9510337323177367,
|
6148 |
+
"grad_norm": 0.08709405362606049,
|
6149 |
+
"learning_rate": 1.2069560259243328e-06,
|
6150 |
+
"loss": 0.0547,
|
6151 |
+
"step": 874
|
6152 |
+
},
|
6153 |
+
{
|
6154 |
+
"epoch": 0.9521218715995647,
|
6155 |
+
"grad_norm": 0.11931779980659485,
|
6156 |
+
"learning_rate": 1.1540118323243865e-06,
|
6157 |
+
"loss": 0.0876,
|
6158 |
+
"step": 875
|
6159 |
+
},
|
6160 |
+
{
|
6161 |
+
"epoch": 0.9532100108813928,
|
6162 |
+
"grad_norm": 0.11160387098789215,
|
6163 |
+
"learning_rate": 1.1022483143405705e-06,
|
6164 |
+
"loss": 0.0894,
|
6165 |
+
"step": 876
|
6166 |
+
},
|
6167 |
+
{
|
6168 |
+
"epoch": 0.9542981501632208,
|
6169 |
+
"grad_norm": 0.1185893788933754,
|
6170 |
+
"learning_rate": 1.0516660902673448e-06,
|
6171 |
+
"loss": 0.0966,
|
6172 |
+
"step": 877
|
6173 |
+
},
|
6174 |
+
{
|
6175 |
+
"epoch": 0.955386289445049,
|
6176 |
+
"grad_norm": 0.1318797916173935,
|
6177 |
+
"learning_rate": 1.0022657642890231e-06,
|
6178 |
+
"loss": 0.1193,
|
6179 |
+
"step": 878
|
6180 |
+
},
|
6181 |
+
{
|
6182 |
+
"epoch": 0.956474428726877,
|
6183 |
+
"grad_norm": 0.11834029108285904,
|
6184 |
+
"learning_rate": 9.540479264726676e-07,
|
6185 |
+
"loss": 0.0869,
|
6186 |
+
"step": 879
|
6187 |
+
},
|
6188 |
+
{
|
6189 |
+
"epoch": 0.9575625680087051,
|
6190 |
+
"grad_norm": 0.11685867607593536,
|
6191 |
+
"learning_rate": 9.070131527609604e-07,
|
6192 |
+
"loss": 0.07,
|
6193 |
+
"step": 880
|
6194 |
+
},
|
6195 |
+
{
|
6196 |
+
"epoch": 0.9586507072905331,
|
6197 |
+
"grad_norm": 0.12375818938016891,
|
6198 |
+
"learning_rate": 8.611620049653879e-07,
|
6199 |
+
"loss": 0.0742,
|
6200 |
+
"step": 881
|
6201 |
+
},
|
6202 |
+
{
|
6203 |
+
"epoch": 0.9597388465723613,
|
6204 |
+
"grad_norm": 0.14064817130565643,
|
6205 |
+
"learning_rate": 8.16495030759501e-07,
|
6206 |
+
"loss": 0.1006,
|
6207 |
+
"step": 882
|
6208 |
+
},
|
6209 |
+
{
|
6210 |
+
"epoch": 0.9608269858541894,
|
6211 |
+
"grad_norm": 0.1763945370912552,
|
6212 |
+
"learning_rate": 7.730127636723539e-07,
|
6213 |
+
"loss": 0.0958,
|
6214 |
+
"step": 883
|
6215 |
+
},
|
6216 |
+
{
|
6217 |
+
"epoch": 0.9619151251360174,
|
6218 |
+
"grad_norm": 0.1824800968170166,
|
6219 |
+
"learning_rate": 7.307157230821426e-07,
|
6220 |
+
"loss": 0.0771,
|
6221 |
+
"step": 884
|
6222 |
+
},
|
6223 |
+
{
|
6224 |
+
"epoch": 0.9630032644178455,
|
6225 |
+
"grad_norm": 0.1460670381784439,
|
6226 |
+
"learning_rate": 6.896044142100433e-07,
|
6227 |
+
"loss": 0.0701,
|
6228 |
+
"step": 885
|
6229 |
+
},
|
6230 |
+
{
|
6231 |
+
"epoch": 0.9640914036996736,
|
6232 |
+
"grad_norm": 0.22308966517448425,
|
6233 |
+
"learning_rate": 6.496793281141056e-07,
|
6234 |
+
"loss": 0.1186,
|
6235 |
+
"step": 886
|
6236 |
+
},
|
6237 |
+
{
|
6238 |
+
"epoch": 0.9651795429815017,
|
6239 |
+
"grad_norm": 0.24287503957748413,
|
6240 |
+
"learning_rate": 6.109409416834688e-07,
|
6241 |
+
"loss": 0.1604,
|
6242 |
+
"step": 887
|
6243 |
+
},
|
6244 |
+
{
|
6245 |
+
"epoch": 0.9662676822633297,
|
6246 |
+
"grad_norm": 0.19692181050777435,
|
6247 |
+
"learning_rate": 5.733897176325665e-07,
|
6248 |
+
"loss": 0.0767,
|
6249 |
+
"step": 888
|
6250 |
+
},
|
6251 |
+
{
|
6252 |
+
"epoch": 0.9673558215451578,
|
6253 |
+
"grad_norm": 0.20672950148582458,
|
6254 |
+
"learning_rate": 5.370261044956971e-07,
|
6255 |
+
"loss": 0.0995,
|
6256 |
+
"step": 889
|
6257 |
+
},
|
6258 |
+
{
|
6259 |
+
"epoch": 0.9684439608269858,
|
6260 |
+
"grad_norm": 0.30662739276885986,
|
6261 |
+
"learning_rate": 5.018505366216175e-07,
|
6262 |
+
"loss": 0.1752,
|
6263 |
+
"step": 890
|
6264 |
+
},
|
6265 |
+
{
|
6266 |
+
"epoch": 0.969532100108814,
|
6267 |
+
"grad_norm": 0.30604445934295654,
|
6268 |
+
"learning_rate": 4.678634341683252e-07,
|
6269 |
+
"loss": 0.2206,
|
6270 |
+
"step": 891
|
6271 |
+
},
|
6272 |
+
{
|
6273 |
+
"epoch": 0.970620239390642,
|
6274 |
+
"grad_norm": 0.3338996469974518,
|
6275 |
+
"learning_rate": 4.3506520309813947e-07,
|
6276 |
+
"loss": 0.0772,
|
6277 |
+
"step": 892
|
6278 |
+
},
|
6279 |
+
{
|
6280 |
+
"epoch": 0.9717083786724701,
|
6281 |
+
"grad_norm": 0.40458595752716064,
|
6282 |
+
"learning_rate": 4.034562351727389e-07,
|
6283 |
+
"loss": 0.1714,
|
6284 |
+
"step": 893
|
6285 |
+
},
|
6286 |
+
{
|
6287 |
+
"epoch": 0.9727965179542981,
|
6288 |
+
"grad_norm": 0.20113840699195862,
|
6289 |
+
"learning_rate": 3.73036907948543e-07,
|
6290 |
+
"loss": 0.0493,
|
6291 |
+
"step": 894
|
6292 |
+
},
|
6293 |
+
{
|
6294 |
+
"epoch": 0.9738846572361263,
|
6295 |
+
"grad_norm": 0.2628382742404938,
|
6296 |
+
"learning_rate": 3.4380758477219333e-07,
|
6297 |
+
"loss": 0.0849,
|
6298 |
+
"step": 895
|
6299 |
+
},
|
6300 |
+
{
|
6301 |
+
"epoch": 0.9749727965179543,
|
6302 |
+
"grad_norm": 0.4046911597251892,
|
6303 |
+
"learning_rate": 3.1576861477621287e-07,
|
6304 |
+
"loss": 0.1658,
|
6305 |
+
"step": 896
|
6306 |
+
},
|
6307 |
+
{
|
6308 |
+
"epoch": 0.9760609357997824,
|
6309 |
+
"grad_norm": 0.5590918660163879,
|
6310 |
+
"learning_rate": 2.889203328748424e-07,
|
6311 |
+
"loss": 0.2407,
|
6312 |
+
"step": 897
|
6313 |
+
},
|
6314 |
+
{
|
6315 |
+
"epoch": 0.9771490750816104,
|
6316 |
+
"grad_norm": 0.42356160283088684,
|
6317 |
+
"learning_rate": 2.6326305976001055e-07,
|
6318 |
+
"loss": 0.1904,
|
6319 |
+
"step": 898
|
6320 |
+
},
|
6321 |
+
{
|
6322 |
+
"epoch": 0.9782372143634385,
|
6323 |
+
"grad_norm": 0.6852498650550842,
|
6324 |
+
"learning_rate": 2.3879710189753656e-07,
|
6325 |
+
"loss": 0.2016,
|
6326 |
+
"step": 899
|
6327 |
+
},
|
6328 |
+
{
|
6329 |
+
"epoch": 0.9793253536452666,
|
6330 |
+
"grad_norm": 0.913657546043396,
|
6331 |
+
"learning_rate": 2.15522751523467e-07,
|
6332 |
+
"loss": 0.3404,
|
6333 |
+
"step": 900
|
6334 |
+
},
|
6335 |
+
{
|
6336 |
+
"epoch": 0.9804134929270947,
|
6337 |
+
"grad_norm": 0.05817437916994095,
|
6338 |
+
"learning_rate": 1.9344028664056713e-07,
|
6339 |
+
"loss": 0.0545,
|
6340 |
+
"step": 901
|
6341 |
+
},
|
6342 |
+
{
|
6343 |
+
"epoch": 0.9815016322089227,
|
6344 |
+
"grad_norm": 0.06097998097538948,
|
6345 |
+
"learning_rate": 1.7254997101500137e-07,
|
6346 |
+
"loss": 0.0471,
|
6347 |
+
"step": 902
|
6348 |
+
},
|
6349 |
+
{
|
6350 |
+
"epoch": 0.9825897714907508,
|
6351 |
+
"grad_norm": 0.07131399214267731,
|
6352 |
+
"learning_rate": 1.5285205417319149e-07,
|
6353 |
+
"loss": 0.0588,
|
6354 |
+
"step": 903
|
6355 |
+
},
|
6356 |
+
{
|
6357 |
+
"epoch": 0.9836779107725789,
|
6358 |
+
"grad_norm": 0.09075633436441422,
|
6359 |
+
"learning_rate": 1.3434677139885222e-07,
|
6360 |
+
"loss": 0.0759,
|
6361 |
+
"step": 904
|
6362 |
+
},
|
6363 |
+
{
|
6364 |
+
"epoch": 0.984766050054407,
|
6365 |
+
"grad_norm": 0.08667346835136414,
|
6366 |
+
"learning_rate": 1.170343437301491e-07,
|
6367 |
+
"loss": 0.078,
|
6368 |
+
"step": 905
|
6369 |
+
},
|
6370 |
+
{
|
6371 |
+
"epoch": 0.985854189336235,
|
6372 |
+
"grad_norm": 0.0773538202047348,
|
6373 |
+
"learning_rate": 1.0091497795706728e-07,
|
6374 |
+
"loss": 0.0658,
|
6375 |
+
"step": 906
|
6376 |
+
},
|
6377 |
+
{
|
6378 |
+
"epoch": 0.9869423286180631,
|
6379 |
+
"grad_norm": 0.09016713500022888,
|
6380 |
+
"learning_rate": 8.598886661895788e-08,
|
6381 |
+
"loss": 0.0687,
|
6382 |
+
"step": 907
|
6383 |
+
},
|
6384 |
+
{
|
6385 |
+
"epoch": 0.9880304678998912,
|
6386 |
+
"grad_norm": 0.10080257803201675,
|
6387 |
+
"learning_rate": 7.225618800222877e-08,
|
6388 |
+
"loss": 0.0807,
|
6389 |
+
"step": 908
|
6390 |
+
},
|
6391 |
+
{
|
6392 |
+
"epoch": 0.9891186071817193,
|
6393 |
+
"grad_norm": 0.10645921528339386,
|
6394 |
+
"learning_rate": 5.971710613821291e-08,
|
6395 |
+
"loss": 0.0832,
|
6396 |
+
"step": 909
|
6397 |
+
},
|
6398 |
+
{
|
6399 |
+
"epoch": 0.9902067464635473,
|
6400 |
+
"grad_norm": 0.11274649947881699,
|
6401 |
+
"learning_rate": 4.837177080119215e-08,
|
6402 |
+
"loss": 0.0918,
|
6403 |
+
"step": 910
|
6404 |
+
},
|
6405 |
+
{
|
6406 |
+
"epoch": 0.9912948857453754,
|
6407 |
+
"grad_norm": 0.12319578975439072,
|
6408 |
+
"learning_rate": 3.8220317506654226e-08,
|
6409 |
+
"loss": 0.0858,
|
6410 |
+
"step": 911
|
6411 |
+
},
|
6412 |
+
{
|
6413 |
+
"epoch": 0.9923830250272034,
|
6414 |
+
"grad_norm": 0.12649929523468018,
|
6415 |
+
"learning_rate": 2.9262867509605163e-08,
|
6416 |
+
"loss": 0.0769,
|
6417 |
+
"step": 912
|
6418 |
+
},
|
6419 |
+
{
|
6420 |
+
"epoch": 0.9934711643090316,
|
6421 |
+
"grad_norm": 0.1557326763868332,
|
6422 |
+
"learning_rate": 2.1499527803214846e-08,
|
6423 |
+
"loss": 0.1158,
|
6424 |
+
"step": 913
|
6425 |
+
},
|
6426 |
+
{
|
6427 |
+
"epoch": 0.9945593035908596,
|
6428 |
+
"grad_norm": 0.1900101900100708,
|
6429 |
+
"learning_rate": 1.4930391117451426e-08,
|
6430 |
+
"loss": 0.1022,
|
6431 |
+
"step": 914
|
6432 |
+
},
|
6433 |
+
{
|
6434 |
+
"epoch": 0.9956474428726877,
|
6435 |
+
"grad_norm": 0.2401544153690338,
|
6436 |
+
"learning_rate": 9.555535917993297e-09,
|
6437 |
+
"loss": 0.1377,
|
6438 |
+
"step": 915
|
6439 |
+
},
|
6440 |
+
{
|
6441 |
+
"epoch": 0.9967355821545157,
|
6442 |
+
"grad_norm": 0.2791292071342468,
|
6443 |
+
"learning_rate": 5.375026405352035e-09,
|
6444 |
+
"loss": 0.1531,
|
6445 |
+
"step": 916
|
6446 |
+
},
|
6447 |
+
{
|
6448 |
+
"epoch": 0.9978237214363439,
|
6449 |
+
"grad_norm": 0.24386504292488098,
|
6450 |
+
"learning_rate": 2.388912514017516e-09,
|
6451 |
+
"loss": 0.1123,
|
6452 |
+
"step": 917
|
6453 |
+
},
|
6454 |
+
{
|
6455 |
+
"epoch": 0.998911860718172,
|
6456 |
+
"grad_norm": 0.312024861574173,
|
6457 |
+
"learning_rate": 5.972299119250125e-10,
|
6458 |
+
"loss": 0.1117,
|
6459 |
+
"step": 918
|
6460 |
+
},
|
6461 |
+
{
|
6462 |
+
"epoch": 1.0,
|
6463 |
+
"grad_norm": 0.4759993255138397,
|
6464 |
+
"learning_rate": 0.0,
|
6465 |
+
"loss": 0.2228,
|
6466 |
+
"step": 919
|
6467 |
}
|
6468 |
],
|
6469 |
"logging_steps": 1,
|
|
|
6478 |
"should_evaluate": false,
|
6479 |
"should_log": false,
|
6480 |
"should_save": true,
|
6481 |
+
"should_training_stop": true
|
6482 |
},
|
6483 |
"attributes": {}
|
6484 |
}
|
6485 |
},
|
6486 |
+
"total_flos": 3.65038685754753e+17,
|
6487 |
"train_batch_size": 2,
|
6488 |
"trial_name": null,
|
6489 |
"trial_params": null
|