Training in progress, step 918, 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 4731640
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9fc08b3e12e93453c3437c5d80f8e665cbc6b1891b45b334267b216ac44adba1
|
3 |
size 4731640
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 2505850
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:baefa897c80cb28f962d6f6c10e561afc3a6b3a5b81bfbbabc2cf29bbb7eef05
|
3 |
size 2505850
|
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:c3510beba863a9deb50722a39a0fb21d656405f07be0052a6c1f899750afb4ec
|
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:ac062936ee06e2256ab1d2e8c72a29d901eb09b82eaa639f49140d904b17d3d4
|
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,
|
@@ -4869,6 +4869,1602 @@
|
|
4869 |
"eval_samples_per_second": 145.229,
|
4870 |
"eval_steps_per_second": 72.802,
|
4871 |
"step": 690
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4872 |
}
|
4873 |
],
|
4874 |
"logging_steps": 1,
|
@@ -4883,12 +6479,12 @@
|
|
4883 |
"should_evaluate": false,
|
4884 |
"should_log": false,
|
4885 |
"should_save": true,
|
4886 |
-
"should_training_stop":
|
4887 |
},
|
4888 |
"attributes": {}
|
4889 |
}
|
4890 |
},
|
4891 |
-
"total_flos":
|
4892 |
"train_batch_size": 2,
|
4893 |
"trial_name": null,
|
4894 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.0,
|
5 |
"eval_steps": 230,
|
6 |
+
"global_step": 918,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
4869 |
"eval_samples_per_second": 145.229,
|
4870 |
"eval_steps_per_second": 72.802,
|
4871 |
"step": 690
|
4872 |
+
},
|
4873 |
+
{
|
4874 |
+
"epoch": 0.7527233115468409,
|
4875 |
+
"grad_norm": 880.0890502929688,
|
4876 |
+
"learning_rate": 2.9289321881345254e-05,
|
4877 |
+
"loss": 12.8765,
|
4878 |
+
"step": 691
|
4879 |
+
},
|
4880 |
+
{
|
4881 |
+
"epoch": 0.7538126361655774,
|
4882 |
+
"grad_norm": 1300.8441162109375,
|
4883 |
+
"learning_rate": 2.9045093461312258e-05,
|
4884 |
+
"loss": 12.9364,
|
4885 |
+
"step": 692
|
4886 |
+
},
|
4887 |
+
{
|
4888 |
+
"epoch": 0.7549019607843137,
|
4889 |
+
"grad_norm": 837.1845092773438,
|
4890 |
+
"learning_rate": 2.8801714436958416e-05,
|
4891 |
+
"loss": 11.9844,
|
4892 |
+
"step": 693
|
4893 |
+
},
|
4894 |
+
{
|
4895 |
+
"epoch": 0.7559912854030502,
|
4896 |
+
"grad_norm": 837.4990844726562,
|
4897 |
+
"learning_rate": 2.855918772175522e-05,
|
4898 |
+
"loss": 12.274,
|
4899 |
+
"step": 694
|
4900 |
+
},
|
4901 |
+
{
|
4902 |
+
"epoch": 0.7570806100217865,
|
4903 |
+
"grad_norm": 1010.6368408203125,
|
4904 |
+
"learning_rate": 2.8317516218971073e-05,
|
4905 |
+
"loss": 14.0263,
|
4906 |
+
"step": 695
|
4907 |
+
},
|
4908 |
+
{
|
4909 |
+
"epoch": 0.7581699346405228,
|
4910 |
+
"grad_norm": 691.3543701171875,
|
4911 |
+
"learning_rate": 2.8076702821636867e-05,
|
4912 |
+
"loss": 12.8702,
|
4913 |
+
"step": 696
|
4914 |
+
},
|
4915 |
+
{
|
4916 |
+
"epoch": 0.7592592592592593,
|
4917 |
+
"grad_norm": 884.868896484375,
|
4918 |
+
"learning_rate": 2.7836750412511016e-05,
|
4919 |
+
"loss": 13.9273,
|
4920 |
+
"step": 697
|
4921 |
+
},
|
4922 |
+
{
|
4923 |
+
"epoch": 0.7603485838779956,
|
4924 |
+
"grad_norm": 1038.74462890625,
|
4925 |
+
"learning_rate": 2.7597661864045233e-05,
|
4926 |
+
"loss": 13.7905,
|
4927 |
+
"step": 698
|
4928 |
+
},
|
4929 |
+
{
|
4930 |
+
"epoch": 0.761437908496732,
|
4931 |
+
"grad_norm": 861.4619140625,
|
4932 |
+
"learning_rate": 2.735944003834997e-05,
|
4933 |
+
"loss": 13.3323,
|
4934 |
+
"step": 699
|
4935 |
+
},
|
4936 |
+
{
|
4937 |
+
"epoch": 0.7625272331154684,
|
4938 |
+
"grad_norm": 781.8521118164062,
|
4939 |
+
"learning_rate": 2.7122087787160166e-05,
|
4940 |
+
"loss": 13.0336,
|
4941 |
+
"step": 700
|
4942 |
+
},
|
4943 |
+
{
|
4944 |
+
"epoch": 0.7636165577342048,
|
4945 |
+
"grad_norm": 1335.78076171875,
|
4946 |
+
"learning_rate": 2.688560795180126e-05,
|
4947 |
+
"loss": 12.7611,
|
4948 |
+
"step": 701
|
4949 |
+
},
|
4950 |
+
{
|
4951 |
+
"epoch": 0.7647058823529411,
|
4952 |
+
"grad_norm": 704.1229248046875,
|
4953 |
+
"learning_rate": 2.6650003363154963e-05,
|
4954 |
+
"loss": 13.3398,
|
4955 |
+
"step": 702
|
4956 |
+
},
|
4957 |
+
{
|
4958 |
+
"epoch": 0.7657952069716776,
|
4959 |
+
"grad_norm": 814.8448486328125,
|
4960 |
+
"learning_rate": 2.641527684162556e-05,
|
4961 |
+
"loss": 12.0726,
|
4962 |
+
"step": 703
|
4963 |
+
},
|
4964 |
+
{
|
4965 |
+
"epoch": 0.7668845315904139,
|
4966 |
+
"grad_norm": 866.6876220703125,
|
4967 |
+
"learning_rate": 2.6181431197105998e-05,
|
4968 |
+
"loss": 12.4249,
|
4969 |
+
"step": 704
|
4970 |
+
},
|
4971 |
+
{
|
4972 |
+
"epoch": 0.7679738562091504,
|
4973 |
+
"grad_norm": 820.3846435546875,
|
4974 |
+
"learning_rate": 2.5948469228944318e-05,
|
4975 |
+
"loss": 12.3404,
|
4976 |
+
"step": 705
|
4977 |
+
},
|
4978 |
+
{
|
4979 |
+
"epoch": 0.7690631808278867,
|
4980 |
+
"grad_norm": 1044.1195068359375,
|
4981 |
+
"learning_rate": 2.5716393725910215e-05,
|
4982 |
+
"loss": 12.7415,
|
4983 |
+
"step": 706
|
4984 |
+
},
|
4985 |
+
{
|
4986 |
+
"epoch": 0.7701525054466231,
|
4987 |
+
"grad_norm": 731.3047485351562,
|
4988 |
+
"learning_rate": 2.5485207466161466e-05,
|
4989 |
+
"loss": 12.301,
|
4990 |
+
"step": 707
|
4991 |
+
},
|
4992 |
+
{
|
4993 |
+
"epoch": 0.7712418300653595,
|
4994 |
+
"grad_norm": 996.6677856445312,
|
4995 |
+
"learning_rate": 2.5254913217210886e-05,
|
4996 |
+
"loss": 12.2375,
|
4997 |
+
"step": 708
|
4998 |
+
},
|
4999 |
+
{
|
5000 |
+
"epoch": 0.7723311546840959,
|
5001 |
+
"grad_norm": 1071.778076171875,
|
5002 |
+
"learning_rate": 2.5025513735893014e-05,
|
5003 |
+
"loss": 12.9786,
|
5004 |
+
"step": 709
|
5005 |
+
},
|
5006 |
+
{
|
5007 |
+
"epoch": 0.7734204793028322,
|
5008 |
+
"grad_norm": 694.3305053710938,
|
5009 |
+
"learning_rate": 2.47970117683313e-05,
|
5010 |
+
"loss": 12.5304,
|
5011 |
+
"step": 710
|
5012 |
+
},
|
5013 |
+
{
|
5014 |
+
"epoch": 0.7745098039215687,
|
5015 |
+
"grad_norm": 568.67626953125,
|
5016 |
+
"learning_rate": 2.4569410049905016e-05,
|
5017 |
+
"loss": 13.4328,
|
5018 |
+
"step": 711
|
5019 |
+
},
|
5020 |
+
{
|
5021 |
+
"epoch": 0.775599128540305,
|
5022 |
+
"grad_norm": 781.4323120117188,
|
5023 |
+
"learning_rate": 2.434271130521666e-05,
|
5024 |
+
"loss": 13.7223,
|
5025 |
+
"step": 712
|
5026 |
+
},
|
5027 |
+
{
|
5028 |
+
"epoch": 0.7766884531590414,
|
5029 |
+
"grad_norm": 707.454345703125,
|
5030 |
+
"learning_rate": 2.411691824805934e-05,
|
5031 |
+
"loss": 13.2585,
|
5032 |
+
"step": 713
|
5033 |
+
},
|
5034 |
+
{
|
5035 |
+
"epoch": 0.7777777777777778,
|
5036 |
+
"grad_norm": 662.977783203125,
|
5037 |
+
"learning_rate": 2.389203358138419e-05,
|
5038 |
+
"loss": 13.1885,
|
5039 |
+
"step": 714
|
5040 |
+
},
|
5041 |
+
{
|
5042 |
+
"epoch": 0.7788671023965141,
|
5043 |
+
"grad_norm": 1106.654296875,
|
5044 |
+
"learning_rate": 2.3668059997268144e-05,
|
5045 |
+
"loss": 12.8549,
|
5046 |
+
"step": 715
|
5047 |
+
},
|
5048 |
+
{
|
5049 |
+
"epoch": 0.7799564270152506,
|
5050 |
+
"grad_norm": 809.3428955078125,
|
5051 |
+
"learning_rate": 2.3445000176881537e-05,
|
5052 |
+
"loss": 13.6922,
|
5053 |
+
"step": 716
|
5054 |
+
},
|
5055 |
+
{
|
5056 |
+
"epoch": 0.7810457516339869,
|
5057 |
+
"grad_norm": 1407.9627685546875,
|
5058 |
+
"learning_rate": 2.3222856790456226e-05,
|
5059 |
+
"loss": 12.3901,
|
5060 |
+
"step": 717
|
5061 |
+
},
|
5062 |
+
{
|
5063 |
+
"epoch": 0.7821350762527233,
|
5064 |
+
"grad_norm": 1036.4595947265625,
|
5065 |
+
"learning_rate": 2.3001632497253424e-05,
|
5066 |
+
"loss": 14.0636,
|
5067 |
+
"step": 718
|
5068 |
+
},
|
5069 |
+
{
|
5070 |
+
"epoch": 0.7832244008714597,
|
5071 |
+
"grad_norm": 713.0249633789062,
|
5072 |
+
"learning_rate": 2.2781329945531936e-05,
|
5073 |
+
"loss": 13.3694,
|
5074 |
+
"step": 719
|
5075 |
+
},
|
5076 |
+
{
|
5077 |
+
"epoch": 0.7843137254901961,
|
5078 |
+
"grad_norm": 946.2935180664062,
|
5079 |
+
"learning_rate": 2.2561951772516587e-05,
|
5080 |
+
"loss": 12.7461,
|
5081 |
+
"step": 720
|
5082 |
+
},
|
5083 |
+
{
|
5084 |
+
"epoch": 0.7854030501089324,
|
5085 |
+
"grad_norm": 1000.2520751953125,
|
5086 |
+
"learning_rate": 2.2343500604366374e-05,
|
5087 |
+
"loss": 14.3916,
|
5088 |
+
"step": 721
|
5089 |
+
},
|
5090 |
+
{
|
5091 |
+
"epoch": 0.7864923747276689,
|
5092 |
+
"grad_norm": 1008.6942749023438,
|
5093 |
+
"learning_rate": 2.2125979056143364e-05,
|
5094 |
+
"loss": 13.3091,
|
5095 |
+
"step": 722
|
5096 |
+
},
|
5097 |
+
{
|
5098 |
+
"epoch": 0.7875816993464052,
|
5099 |
+
"grad_norm": 720.97314453125,
|
5100 |
+
"learning_rate": 2.190938973178105e-05,
|
5101 |
+
"loss": 12.963,
|
5102 |
+
"step": 723
|
5103 |
+
},
|
5104 |
+
{
|
5105 |
+
"epoch": 0.7886710239651417,
|
5106 |
+
"grad_norm": 704.6624145507812,
|
5107 |
+
"learning_rate": 2.169373522405349e-05,
|
5108 |
+
"loss": 13.3057,
|
5109 |
+
"step": 724
|
5110 |
+
},
|
5111 |
+
{
|
5112 |
+
"epoch": 0.789760348583878,
|
5113 |
+
"grad_norm": 944.1021728515625,
|
5114 |
+
"learning_rate": 2.1479018114544026e-05,
|
5115 |
+
"loss": 12.621,
|
5116 |
+
"step": 725
|
5117 |
+
},
|
5118 |
+
{
|
5119 |
+
"epoch": 0.7908496732026143,
|
5120 |
+
"grad_norm": 552.0089111328125,
|
5121 |
+
"learning_rate": 2.1265240973614486e-05,
|
5122 |
+
"loss": 12.7888,
|
5123 |
+
"step": 726
|
5124 |
+
},
|
5125 |
+
{
|
5126 |
+
"epoch": 0.7919389978213508,
|
5127 |
+
"grad_norm": 685.5350341796875,
|
5128 |
+
"learning_rate": 2.105240636037449e-05,
|
5129 |
+
"loss": 13.6543,
|
5130 |
+
"step": 727
|
5131 |
+
},
|
5132 |
+
{
|
5133 |
+
"epoch": 0.7930283224400871,
|
5134 |
+
"grad_norm": 627.4608764648438,
|
5135 |
+
"learning_rate": 2.0840516822650614e-05,
|
5136 |
+
"loss": 12.6445,
|
5137 |
+
"step": 728
|
5138 |
+
},
|
5139 |
+
{
|
5140 |
+
"epoch": 0.7941176470588235,
|
5141 |
+
"grad_norm": 645.2318725585938,
|
5142 |
+
"learning_rate": 2.0629574896956126e-05,
|
5143 |
+
"loss": 12.6309,
|
5144 |
+
"step": 729
|
5145 |
+
},
|
5146 |
+
{
|
5147 |
+
"epoch": 0.7952069716775599,
|
5148 |
+
"grad_norm": 888.5296630859375,
|
5149 |
+
"learning_rate": 2.0419583108460418e-05,
|
5150 |
+
"loss": 11.886,
|
5151 |
+
"step": 730
|
5152 |
+
},
|
5153 |
+
{
|
5154 |
+
"epoch": 0.7962962962962963,
|
5155 |
+
"grad_norm": 1192.0579833984375,
|
5156 |
+
"learning_rate": 2.0210543970958872e-05,
|
5157 |
+
"loss": 12.3713,
|
5158 |
+
"step": 731
|
5159 |
+
},
|
5160 |
+
{
|
5161 |
+
"epoch": 0.7973856209150327,
|
5162 |
+
"grad_norm": 635.3908081054688,
|
5163 |
+
"learning_rate": 2.0002459986842825e-05,
|
5164 |
+
"loss": 13.2216,
|
5165 |
+
"step": 732
|
5166 |
+
},
|
5167 |
+
{
|
5168 |
+
"epoch": 0.7984749455337691,
|
5169 |
+
"grad_norm": 657.1572265625,
|
5170 |
+
"learning_rate": 1.9795333647069448e-05,
|
5171 |
+
"loss": 12.6068,
|
5172 |
+
"step": 733
|
5173 |
+
},
|
5174 |
+
{
|
5175 |
+
"epoch": 0.7995642701525054,
|
5176 |
+
"grad_norm": 1367.3870849609375,
|
5177 |
+
"learning_rate": 1.958916743113214e-05,
|
5178 |
+
"loss": 12.8324,
|
5179 |
+
"step": 734
|
5180 |
+
},
|
5181 |
+
{
|
5182 |
+
"epoch": 0.8006535947712419,
|
5183 |
+
"grad_norm": 1032.5328369140625,
|
5184 |
+
"learning_rate": 1.93839638070306e-05,
|
5185 |
+
"loss": 12.6028,
|
5186 |
+
"step": 735
|
5187 |
+
},
|
5188 |
+
{
|
5189 |
+
"epoch": 0.8017429193899782,
|
5190 |
+
"grad_norm": 762.1034545898438,
|
5191 |
+
"learning_rate": 1.9179725231241564e-05,
|
5192 |
+
"loss": 12.5233,
|
5193 |
+
"step": 736
|
5194 |
+
},
|
5195 |
+
{
|
5196 |
+
"epoch": 0.8028322440087146,
|
5197 |
+
"grad_norm": 1431.1048583984375,
|
5198 |
+
"learning_rate": 1.8976454148689127e-05,
|
5199 |
+
"loss": 12.9953,
|
5200 |
+
"step": 737
|
5201 |
+
},
|
5202 |
+
{
|
5203 |
+
"epoch": 0.803921568627451,
|
5204 |
+
"grad_norm": 1034.8011474609375,
|
5205 |
+
"learning_rate": 1.877415299271561e-05,
|
5206 |
+
"loss": 13.2373,
|
5207 |
+
"step": 738
|
5208 |
+
},
|
5209 |
+
{
|
5210 |
+
"epoch": 0.8050108932461874,
|
5211 |
+
"grad_norm": 752.5025024414062,
|
5212 |
+
"learning_rate": 1.857282418505253e-05,
|
5213 |
+
"loss": 13.1138,
|
5214 |
+
"step": 739
|
5215 |
+
},
|
5216 |
+
{
|
5217 |
+
"epoch": 0.8061002178649237,
|
5218 |
+
"grad_norm": 911.1949462890625,
|
5219 |
+
"learning_rate": 1.8372470135791344e-05,
|
5220 |
+
"loss": 12.7149,
|
5221 |
+
"step": 740
|
5222 |
+
},
|
5223 |
+
{
|
5224 |
+
"epoch": 0.8071895424836601,
|
5225 |
+
"grad_norm": 1050.7984619140625,
|
5226 |
+
"learning_rate": 1.8173093243354878e-05,
|
5227 |
+
"loss": 12.862,
|
5228 |
+
"step": 741
|
5229 |
+
},
|
5230 |
+
{
|
5231 |
+
"epoch": 0.8082788671023965,
|
5232 |
+
"grad_norm": 943.8112182617188,
|
5233 |
+
"learning_rate": 1.7974695894468384e-05,
|
5234 |
+
"loss": 12.6691,
|
5235 |
+
"step": 742
|
5236 |
+
},
|
5237 |
+
{
|
5238 |
+
"epoch": 0.809368191721133,
|
5239 |
+
"grad_norm": 860.751708984375,
|
5240 |
+
"learning_rate": 1.7777280464131197e-05,
|
5241 |
+
"loss": 12.9838,
|
5242 |
+
"step": 743
|
5243 |
+
},
|
5244 |
+
{
|
5245 |
+
"epoch": 0.8104575163398693,
|
5246 |
+
"grad_norm": 812.5382690429688,
|
5247 |
+
"learning_rate": 1.7580849315588068e-05,
|
5248 |
+
"loss": 12.5751,
|
5249 |
+
"step": 744
|
5250 |
+
},
|
5251 |
+
{
|
5252 |
+
"epoch": 0.8115468409586056,
|
5253 |
+
"grad_norm": 684.0803833007812,
|
5254 |
+
"learning_rate": 1.7385404800301007e-05,
|
5255 |
+
"loss": 12.6821,
|
5256 |
+
"step": 745
|
5257 |
+
},
|
5258 |
+
{
|
5259 |
+
"epoch": 0.8126361655773421,
|
5260 |
+
"grad_norm": 637.1691284179688,
|
5261 |
+
"learning_rate": 1.7190949257921196e-05,
|
5262 |
+
"loss": 12.704,
|
5263 |
+
"step": 746
|
5264 |
+
},
|
5265 |
+
{
|
5266 |
+
"epoch": 0.8137254901960784,
|
5267 |
+
"grad_norm": 1279.00146484375,
|
5268 |
+
"learning_rate": 1.6997485016260793e-05,
|
5269 |
+
"loss": 13.6321,
|
5270 |
+
"step": 747
|
5271 |
+
},
|
5272 |
+
{
|
5273 |
+
"epoch": 0.8148148148148148,
|
5274 |
+
"grad_norm": 1138.228515625,
|
5275 |
+
"learning_rate": 1.680501439126525e-05,
|
5276 |
+
"loss": 12.3677,
|
5277 |
+
"step": 748
|
5278 |
+
},
|
5279 |
+
{
|
5280 |
+
"epoch": 0.8159041394335512,
|
5281 |
+
"grad_norm": 718.772705078125,
|
5282 |
+
"learning_rate": 1.6613539686985458e-05,
|
5283 |
+
"loss": 13.3372,
|
5284 |
+
"step": 749
|
5285 |
+
},
|
5286 |
+
{
|
5287 |
+
"epoch": 0.8169934640522876,
|
5288 |
+
"grad_norm": 960.2557373046875,
|
5289 |
+
"learning_rate": 1.642306319555027e-05,
|
5290 |
+
"loss": 11.9433,
|
5291 |
+
"step": 750
|
5292 |
+
},
|
5293 |
+
{
|
5294 |
+
"epoch": 0.818082788671024,
|
5295 |
+
"grad_norm": 745.1720581054688,
|
5296 |
+
"learning_rate": 1.6233587197138968e-05,
|
5297 |
+
"loss": 11.8539,
|
5298 |
+
"step": 751
|
5299 |
+
},
|
5300 |
+
{
|
5301 |
+
"epoch": 0.8191721132897604,
|
5302 |
+
"grad_norm": 837.4060668945312,
|
5303 |
+
"learning_rate": 1.6045113959953985e-05,
|
5304 |
+
"loss": 12.3409,
|
5305 |
+
"step": 752
|
5306 |
+
},
|
5307 |
+
{
|
5308 |
+
"epoch": 0.8202614379084967,
|
5309 |
+
"grad_norm": 854.7937622070312,
|
5310 |
+
"learning_rate": 1.585764574019388e-05,
|
5311 |
+
"loss": 13.3537,
|
5312 |
+
"step": 753
|
5313 |
+
},
|
5314 |
+
{
|
5315 |
+
"epoch": 0.8213507625272332,
|
5316 |
+
"grad_norm": 879.7656860351562,
|
5317 |
+
"learning_rate": 1.5671184782026106e-05,
|
5318 |
+
"loss": 12.7348,
|
5319 |
+
"step": 754
|
5320 |
+
},
|
5321 |
+
{
|
5322 |
+
"epoch": 0.8224400871459695,
|
5323 |
+
"grad_norm": 941.7662353515625,
|
5324 |
+
"learning_rate": 1.548573331756038e-05,
|
5325 |
+
"loss": 12.8492,
|
5326 |
+
"step": 755
|
5327 |
+
},
|
5328 |
+
{
|
5329 |
+
"epoch": 0.8235294117647058,
|
5330 |
+
"grad_norm": 926.8936767578125,
|
5331 |
+
"learning_rate": 1.530129356682175e-05,
|
5332 |
+
"loss": 12.8661,
|
5333 |
+
"step": 756
|
5334 |
+
},
|
5335 |
+
{
|
5336 |
+
"epoch": 0.8246187363834423,
|
5337 |
+
"grad_norm": 1937.478271484375,
|
5338 |
+
"learning_rate": 1.5117867737724134e-05,
|
5339 |
+
"loss": 12.9826,
|
5340 |
+
"step": 757
|
5341 |
+
},
|
5342 |
+
{
|
5343 |
+
"epoch": 0.8257080610021786,
|
5344 |
+
"grad_norm": 909.1730346679688,
|
5345 |
+
"learning_rate": 1.4935458026043959e-05,
|
5346 |
+
"loss": 11.884,
|
5347 |
+
"step": 758
|
5348 |
+
},
|
5349 |
+
{
|
5350 |
+
"epoch": 0.826797385620915,
|
5351 |
+
"grad_norm": 833.2159423828125,
|
5352 |
+
"learning_rate": 1.4754066615393668e-05,
|
5353 |
+
"loss": 13.1729,
|
5354 |
+
"step": 759
|
5355 |
+
},
|
5356 |
+
{
|
5357 |
+
"epoch": 0.8278867102396514,
|
5358 |
+
"grad_norm": 1762.9833984375,
|
5359 |
+
"learning_rate": 1.457369567719581e-05,
|
5360 |
+
"loss": 12.5088,
|
5361 |
+
"step": 760
|
5362 |
+
},
|
5363 |
+
{
|
5364 |
+
"epoch": 0.8289760348583878,
|
5365 |
+
"grad_norm": 774.0321044921875,
|
5366 |
+
"learning_rate": 1.4394347370656836e-05,
|
5367 |
+
"loss": 12.6099,
|
5368 |
+
"step": 761
|
5369 |
+
},
|
5370 |
+
{
|
5371 |
+
"epoch": 0.8300653594771242,
|
5372 |
+
"grad_norm": 833.3447875976562,
|
5373 |
+
"learning_rate": 1.4216023842741455e-05,
|
5374 |
+
"loss": 11.8071,
|
5375 |
+
"step": 762
|
5376 |
+
},
|
5377 |
+
{
|
5378 |
+
"epoch": 0.8311546840958606,
|
5379 |
+
"grad_norm": 659.9131469726562,
|
5380 |
+
"learning_rate": 1.4038727228146753e-05,
|
5381 |
+
"loss": 12.0735,
|
5382 |
+
"step": 763
|
5383 |
+
},
|
5384 |
+
{
|
5385 |
+
"epoch": 0.8322440087145969,
|
5386 |
+
"grad_norm": 888.9996948242188,
|
5387 |
+
"learning_rate": 1.3862459649276715e-05,
|
5388 |
+
"loss": 12.47,
|
5389 |
+
"step": 764
|
5390 |
+
},
|
5391 |
+
{
|
5392 |
+
"epoch": 0.8333333333333334,
|
5393 |
+
"grad_norm": 759.5191040039062,
|
5394 |
+
"learning_rate": 1.3687223216216904e-05,
|
5395 |
+
"loss": 12.5613,
|
5396 |
+
"step": 765
|
5397 |
+
},
|
5398 |
+
{
|
5399 |
+
"epoch": 0.8344226579520697,
|
5400 |
+
"grad_norm": 805.9093627929688,
|
5401 |
+
"learning_rate": 1.3513020026709023e-05,
|
5402 |
+
"loss": 12.0561,
|
5403 |
+
"step": 766
|
5404 |
+
},
|
5405 |
+
{
|
5406 |
+
"epoch": 0.835511982570806,
|
5407 |
+
"grad_norm": 763.7495727539062,
|
5408 |
+
"learning_rate": 1.3339852166125954e-05,
|
5409 |
+
"loss": 12.959,
|
5410 |
+
"step": 767
|
5411 |
+
},
|
5412 |
+
{
|
5413 |
+
"epoch": 0.8366013071895425,
|
5414 |
+
"grad_norm": 610.1506958007812,
|
5415 |
+
"learning_rate": 1.3167721707446678e-05,
|
5416 |
+
"loss": 12.2513,
|
5417 |
+
"step": 768
|
5418 |
+
},
|
5419 |
+
{
|
5420 |
+
"epoch": 0.8376906318082789,
|
5421 |
+
"grad_norm": 958.7525634765625,
|
5422 |
+
"learning_rate": 1.2996630711231616e-05,
|
5423 |
+
"loss": 14.1049,
|
5424 |
+
"step": 769
|
5425 |
+
},
|
5426 |
+
{
|
5427 |
+
"epoch": 0.8387799564270153,
|
5428 |
+
"grad_norm": 1156.2576904296875,
|
5429 |
+
"learning_rate": 1.2826581225597767e-05,
|
5430 |
+
"loss": 12.8887,
|
5431 |
+
"step": 770
|
5432 |
+
},
|
5433 |
+
{
|
5434 |
+
"epoch": 0.8398692810457516,
|
5435 |
+
"grad_norm": 1074.29345703125,
|
5436 |
+
"learning_rate": 1.26575752861943e-05,
|
5437 |
+
"loss": 13.7254,
|
5438 |
+
"step": 771
|
5439 |
+
},
|
5440 |
+
{
|
5441 |
+
"epoch": 0.840958605664488,
|
5442 |
+
"grad_norm": 1287.4105224609375,
|
5443 |
+
"learning_rate": 1.248961491617826e-05,
|
5444 |
+
"loss": 13.033,
|
5445 |
+
"step": 772
|
5446 |
+
},
|
5447 |
+
{
|
5448 |
+
"epoch": 0.8420479302832244,
|
5449 |
+
"grad_norm": 945.987548828125,
|
5450 |
+
"learning_rate": 1.2322702126190156e-05,
|
5451 |
+
"loss": 11.7015,
|
5452 |
+
"step": 773
|
5453 |
+
},
|
5454 |
+
{
|
5455 |
+
"epoch": 0.8431372549019608,
|
5456 |
+
"grad_norm": 763.9169921875,
|
5457 |
+
"learning_rate": 1.2156838914330072e-05,
|
5458 |
+
"loss": 12.8775,
|
5459 |
+
"step": 774
|
5460 |
+
},
|
5461 |
+
{
|
5462 |
+
"epoch": 0.8442265795206971,
|
5463 |
+
"grad_norm": 869.27880859375,
|
5464 |
+
"learning_rate": 1.1992027266133598e-05,
|
5465 |
+
"loss": 12.8051,
|
5466 |
+
"step": 775
|
5467 |
+
},
|
5468 |
+
{
|
5469 |
+
"epoch": 0.8453159041394336,
|
5470 |
+
"grad_norm": 919.6107177734375,
|
5471 |
+
"learning_rate": 1.1828269154548244e-05,
|
5472 |
+
"loss": 12.7364,
|
5473 |
+
"step": 776
|
5474 |
+
},
|
5475 |
+
{
|
5476 |
+
"epoch": 0.8464052287581699,
|
5477 |
+
"grad_norm": 840.4449462890625,
|
5478 |
+
"learning_rate": 1.1665566539909623e-05,
|
5479 |
+
"loss": 12.2462,
|
5480 |
+
"step": 777
|
5481 |
+
},
|
5482 |
+
{
|
5483 |
+
"epoch": 0.8474945533769063,
|
5484 |
+
"grad_norm": 875.924072265625,
|
5485 |
+
"learning_rate": 1.1503921369918091e-05,
|
5486 |
+
"loss": 12.9855,
|
5487 |
+
"step": 778
|
5488 |
+
},
|
5489 |
+
{
|
5490 |
+
"epoch": 0.8485838779956427,
|
5491 |
+
"grad_norm": 2267.00927734375,
|
5492 |
+
"learning_rate": 1.1343335579615467e-05,
|
5493 |
+
"loss": 12.9955,
|
5494 |
+
"step": 779
|
5495 |
+
},
|
5496 |
+
{
|
5497 |
+
"epoch": 0.8496732026143791,
|
5498 |
+
"grad_norm": 934.5510864257812,
|
5499 |
+
"learning_rate": 1.118381109136174e-05,
|
5500 |
+
"loss": 12.6251,
|
5501 |
+
"step": 780
|
5502 |
+
},
|
5503 |
+
{
|
5504 |
+
"epoch": 0.8507625272331155,
|
5505 |
+
"grad_norm": 599.7080078125,
|
5506 |
+
"learning_rate": 1.1025349814812224e-05,
|
5507 |
+
"loss": 12.7546,
|
5508 |
+
"step": 781
|
5509 |
+
},
|
5510 |
+
{
|
5511 |
+
"epoch": 0.8518518518518519,
|
5512 |
+
"grad_norm": 1287.09912109375,
|
5513 |
+
"learning_rate": 1.0867953646894525e-05,
|
5514 |
+
"loss": 12.8024,
|
5515 |
+
"step": 782
|
5516 |
+
},
|
5517 |
+
{
|
5518 |
+
"epoch": 0.8529411764705882,
|
5519 |
+
"grad_norm": 924.6251831054688,
|
5520 |
+
"learning_rate": 1.0711624471785986e-05,
|
5521 |
+
"loss": 12.9911,
|
5522 |
+
"step": 783
|
5523 |
+
},
|
5524 |
+
{
|
5525 |
+
"epoch": 0.8540305010893247,
|
5526 |
+
"grad_norm": 904.363525390625,
|
5527 |
+
"learning_rate": 1.055636416089102e-05,
|
5528 |
+
"loss": 13.8365,
|
5529 |
+
"step": 784
|
5530 |
+
},
|
5531 |
+
{
|
5532 |
+
"epoch": 0.855119825708061,
|
5533 |
+
"grad_norm": 998.8102416992188,
|
5534 |
+
"learning_rate": 1.0402174572818723e-05,
|
5535 |
+
"loss": 11.9683,
|
5536 |
+
"step": 785
|
5537 |
+
},
|
5538 |
+
{
|
5539 |
+
"epoch": 0.8562091503267973,
|
5540 |
+
"grad_norm": 890.9308471679688,
|
5541 |
+
"learning_rate": 1.0249057553360742e-05,
|
5542 |
+
"loss": 12.8672,
|
5543 |
+
"step": 786
|
5544 |
+
},
|
5545 |
+
{
|
5546 |
+
"epoch": 0.8572984749455338,
|
5547 |
+
"grad_norm": 944.108642578125,
|
5548 |
+
"learning_rate": 1.0097014935468984e-05,
|
5549 |
+
"loss": 13.2105,
|
5550 |
+
"step": 787
|
5551 |
+
},
|
5552 |
+
{
|
5553 |
+
"epoch": 0.8583877995642701,
|
5554 |
+
"grad_norm": 956.6070556640625,
|
5555 |
+
"learning_rate": 9.946048539233865e-06,
|
5556 |
+
"loss": 12.8339,
|
5557 |
+
"step": 788
|
5558 |
+
},
|
5559 |
+
{
|
5560 |
+
"epoch": 0.8594771241830066,
|
5561 |
+
"grad_norm": 783.9725341796875,
|
5562 |
+
"learning_rate": 9.796160171862367e-06,
|
5563 |
+
"loss": 12.0513,
|
5564 |
+
"step": 789
|
5565 |
+
},
|
5566 |
+
{
|
5567 |
+
"epoch": 0.8605664488017429,
|
5568 |
+
"grad_norm": 678.079833984375,
|
5569 |
+
"learning_rate": 9.647351627656543e-06,
|
5570 |
+
"loss": 12.4566,
|
5571 |
+
"step": 790
|
5572 |
+
},
|
5573 |
+
{
|
5574 |
+
"epoch": 0.8616557734204793,
|
5575 |
+
"grad_norm": 795.08154296875,
|
5576 |
+
"learning_rate": 9.499624687991871e-06,
|
5577 |
+
"loss": 13.0777,
|
5578 |
+
"step": 791
|
5579 |
+
},
|
5580 |
+
{
|
5581 |
+
"epoch": 0.8627450980392157,
|
5582 |
+
"grad_norm": 1114.5704345703125,
|
5583 |
+
"learning_rate": 9.352981121296134e-06,
|
5584 |
+
"loss": 12.3807,
|
5585 |
+
"step": 792
|
5586 |
+
},
|
5587 |
+
{
|
5588 |
+
"epoch": 0.8638344226579521,
|
5589 |
+
"grad_norm": 810.4409790039062,
|
5590 |
+
"learning_rate": 9.207422683028066e-06,
|
5591 |
+
"loss": 12.0099,
|
5592 |
+
"step": 793
|
5593 |
+
},
|
5594 |
+
{
|
5595 |
+
"epoch": 0.8649237472766884,
|
5596 |
+
"grad_norm": 659.8233642578125,
|
5597 |
+
"learning_rate": 9.062951115656403e-06,
|
5598 |
+
"loss": 12.5349,
|
5599 |
+
"step": 794
|
5600 |
+
},
|
5601 |
+
{
|
5602 |
+
"epoch": 0.8660130718954249,
|
5603 |
+
"grad_norm": 858.1024780273438,
|
5604 |
+
"learning_rate": 8.919568148639123e-06,
|
5605 |
+
"loss": 12.1748,
|
5606 |
+
"step": 795
|
5607 |
+
},
|
5608 |
+
{
|
5609 |
+
"epoch": 0.8671023965141612,
|
5610 |
+
"grad_norm": 888.814453125,
|
5611 |
+
"learning_rate": 8.777275498402548e-06,
|
5612 |
+
"loss": 12.9867,
|
5613 |
+
"step": 796
|
5614 |
+
},
|
5615 |
+
{
|
5616 |
+
"epoch": 0.8681917211328976,
|
5617 |
+
"grad_norm": 845.94140625,
|
5618 |
+
"learning_rate": 8.636074868320987e-06,
|
5619 |
+
"loss": 13.1836,
|
5620 |
+
"step": 797
|
5621 |
+
},
|
5622 |
+
{
|
5623 |
+
"epoch": 0.869281045751634,
|
5624 |
+
"grad_norm": 970.3507080078125,
|
5625 |
+
"learning_rate": 8.495967948696192e-06,
|
5626 |
+
"loss": 11.6568,
|
5627 |
+
"step": 798
|
5628 |
+
},
|
5629 |
+
{
|
5630 |
+
"epoch": 0.8703703703703703,
|
5631 |
+
"grad_norm": 865.5740966796875,
|
5632 |
+
"learning_rate": 8.35695641673725e-06,
|
5633 |
+
"loss": 12.0325,
|
5634 |
+
"step": 799
|
5635 |
+
},
|
5636 |
+
{
|
5637 |
+
"epoch": 0.8714596949891068,
|
5638 |
+
"grad_norm": 916.9160766601562,
|
5639 |
+
"learning_rate": 8.219041936540395e-06,
|
5640 |
+
"loss": 13.141,
|
5641 |
+
"step": 800
|
5642 |
+
},
|
5643 |
+
{
|
5644 |
+
"epoch": 0.8725490196078431,
|
5645 |
+
"grad_norm": 746.8594970703125,
|
5646 |
+
"learning_rate": 8.082226159069196e-06,
|
5647 |
+
"loss": 13.3162,
|
5648 |
+
"step": 801
|
5649 |
+
},
|
5650 |
+
{
|
5651 |
+
"epoch": 0.8736383442265795,
|
5652 |
+
"grad_norm": 847.8311157226562,
|
5653 |
+
"learning_rate": 7.946510722134692e-06,
|
5654 |
+
"loss": 12.9385,
|
5655 |
+
"step": 802
|
5656 |
+
},
|
5657 |
+
{
|
5658 |
+
"epoch": 0.8747276688453159,
|
5659 |
+
"grad_norm": 1479.700439453125,
|
5660 |
+
"learning_rate": 7.811897250375833e-06,
|
5661 |
+
"loss": 13.0996,
|
5662 |
+
"step": 803
|
5663 |
+
},
|
5664 |
+
{
|
5665 |
+
"epoch": 0.8758169934640523,
|
5666 |
+
"grad_norm": 983.4318237304688,
|
5667 |
+
"learning_rate": 7.678387355240057e-06,
|
5668 |
+
"loss": 13.556,
|
5669 |
+
"step": 804
|
5670 |
+
},
|
5671 |
+
{
|
5672 |
+
"epoch": 0.8769063180827886,
|
5673 |
+
"grad_norm": 1585.536376953125,
|
5674 |
+
"learning_rate": 7.5459826349639436e-06,
|
5675 |
+
"loss": 13.0711,
|
5676 |
+
"step": 805
|
5677 |
+
},
|
5678 |
+
{
|
5679 |
+
"epoch": 0.8779956427015251,
|
5680 |
+
"grad_norm": 818.268310546875,
|
5681 |
+
"learning_rate": 7.4146846745541506e-06,
|
5682 |
+
"loss": 12.364,
|
5683 |
+
"step": 806
|
5684 |
+
},
|
5685 |
+
{
|
5686 |
+
"epoch": 0.8790849673202614,
|
5687 |
+
"grad_norm": 732.67626953125,
|
5688 |
+
"learning_rate": 7.284495045768325e-06,
|
5689 |
+
"loss": 11.9299,
|
5690 |
+
"step": 807
|
5691 |
+
},
|
5692 |
+
{
|
5693 |
+
"epoch": 0.8801742919389978,
|
5694 |
+
"grad_norm": 785.660888671875,
|
5695 |
+
"learning_rate": 7.155415307096458e-06,
|
5696 |
+
"loss": 13.0072,
|
5697 |
+
"step": 808
|
5698 |
+
},
|
5699 |
+
{
|
5700 |
+
"epoch": 0.8812636165577342,
|
5701 |
+
"grad_norm": 1017.672607421875,
|
5702 |
+
"learning_rate": 7.027447003742071e-06,
|
5703 |
+
"loss": 11.7492,
|
5704 |
+
"step": 809
|
5705 |
+
},
|
5706 |
+
{
|
5707 |
+
"epoch": 0.8823529411764706,
|
5708 |
+
"grad_norm": 961.2965087890625,
|
5709 |
+
"learning_rate": 6.900591667603751e-06,
|
5710 |
+
"loss": 12.7767,
|
5711 |
+
"step": 810
|
5712 |
+
},
|
5713 |
+
{
|
5714 |
+
"epoch": 0.883442265795207,
|
5715 |
+
"grad_norm": 742.8497314453125,
|
5716 |
+
"learning_rate": 6.774850817256939e-06,
|
5717 |
+
"loss": 11.4043,
|
5718 |
+
"step": 811
|
5719 |
+
},
|
5720 |
+
{
|
5721 |
+
"epoch": 0.8845315904139434,
|
5722 |
+
"grad_norm": 624.9981079101562,
|
5723 |
+
"learning_rate": 6.650225957935552e-06,
|
5724 |
+
"loss": 12.2672,
|
5725 |
+
"step": 812
|
5726 |
+
},
|
5727 |
+
{
|
5728 |
+
"epoch": 0.8856209150326797,
|
5729 |
+
"grad_norm": 779.7870483398438,
|
5730 |
+
"learning_rate": 6.5267185815141355e-06,
|
5731 |
+
"loss": 11.9513,
|
5732 |
+
"step": 813
|
5733 |
+
},
|
5734 |
+
{
|
5735 |
+
"epoch": 0.8867102396514162,
|
5736 |
+
"grad_norm": 1622.0540771484375,
|
5737 |
+
"learning_rate": 6.40433016648988e-06,
|
5738 |
+
"loss": 13.6973,
|
5739 |
+
"step": 814
|
5740 |
+
},
|
5741 |
+
{
|
5742 |
+
"epoch": 0.8877995642701525,
|
5743 |
+
"grad_norm": 1286.1015625,
|
5744 |
+
"learning_rate": 6.283062177965038e-06,
|
5745 |
+
"loss": 11.6033,
|
5746 |
+
"step": 815
|
5747 |
+
},
|
5748 |
+
{
|
5749 |
+
"epoch": 0.8888888888888888,
|
5750 |
+
"grad_norm": 1373.9736328125,
|
5751 |
+
"learning_rate": 6.162916067629254e-06,
|
5752 |
+
"loss": 13.2463,
|
5753 |
+
"step": 816
|
5754 |
+
},
|
5755 |
+
{
|
5756 |
+
"epoch": 0.8899782135076253,
|
5757 |
+
"grad_norm": 1360.0540771484375,
|
5758 |
+
"learning_rate": 6.043893273742329e-06,
|
5759 |
+
"loss": 11.335,
|
5760 |
+
"step": 817
|
5761 |
+
},
|
5762 |
+
{
|
5763 |
+
"epoch": 0.8910675381263616,
|
5764 |
+
"grad_norm": 1296.9820556640625,
|
5765 |
+
"learning_rate": 5.925995221116853e-06,
|
5766 |
+
"loss": 12.4607,
|
5767 |
+
"step": 818
|
5768 |
+
},
|
5769 |
+
{
|
5770 |
+
"epoch": 0.8921568627450981,
|
5771 |
+
"grad_norm": 1495.9398193359375,
|
5772 |
+
"learning_rate": 5.809223321101276e-06,
|
5773 |
+
"loss": 11.7732,
|
5774 |
+
"step": 819
|
5775 |
+
},
|
5776 |
+
{
|
5777 |
+
"epoch": 0.8932461873638344,
|
5778 |
+
"grad_norm": 880.9378662109375,
|
5779 |
+
"learning_rate": 5.693578971562963e-06,
|
5780 |
+
"loss": 12.8642,
|
5781 |
+
"step": 820
|
5782 |
+
},
|
5783 |
+
{
|
5784 |
+
"epoch": 0.8943355119825708,
|
5785 |
+
"grad_norm": 1063.22216796875,
|
5786 |
+
"learning_rate": 5.5790635568714224e-06,
|
5787 |
+
"loss": 12.4356,
|
5788 |
+
"step": 821
|
5789 |
+
},
|
5790 |
+
{
|
5791 |
+
"epoch": 0.8954248366013072,
|
5792 |
+
"grad_norm": 657.4075927734375,
|
5793 |
+
"learning_rate": 5.465678447881828e-06,
|
5794 |
+
"loss": 12.4044,
|
5795 |
+
"step": 822
|
5796 |
+
},
|
5797 |
+
{
|
5798 |
+
"epoch": 0.8965141612200436,
|
5799 |
+
"grad_norm": 1261.831787109375,
|
5800 |
+
"learning_rate": 5.3534250019184774e-06,
|
5801 |
+
"loss": 13.5235,
|
5802 |
+
"step": 823
|
5803 |
+
},
|
5804 |
+
{
|
5805 |
+
"epoch": 0.8976034858387799,
|
5806 |
+
"grad_norm": 601.0888061523438,
|
5807 |
+
"learning_rate": 5.242304562758704e-06,
|
5808 |
+
"loss": 11.9985,
|
5809 |
+
"step": 824
|
5810 |
+
},
|
5811 |
+
{
|
5812 |
+
"epoch": 0.8986928104575164,
|
5813 |
+
"grad_norm": 781.6649780273438,
|
5814 |
+
"learning_rate": 5.132318460616625e-06,
|
5815 |
+
"loss": 11.9329,
|
5816 |
+
"step": 825
|
5817 |
+
},
|
5818 |
+
{
|
5819 |
+
"epoch": 0.8997821350762527,
|
5820 |
+
"grad_norm": 1356.27294921875,
|
5821 |
+
"learning_rate": 5.023468012127364e-06,
|
5822 |
+
"loss": 12.6829,
|
5823 |
+
"step": 826
|
5824 |
+
},
|
5825 |
+
{
|
5826 |
+
"epoch": 0.900871459694989,
|
5827 |
+
"grad_norm": 929.7411499023438,
|
5828 |
+
"learning_rate": 4.915754520331173e-06,
|
5829 |
+
"loss": 11.9567,
|
5830 |
+
"step": 827
|
5831 |
+
},
|
5832 |
+
{
|
5833 |
+
"epoch": 0.9019607843137255,
|
5834 |
+
"grad_norm": 1020.9713745117188,
|
5835 |
+
"learning_rate": 4.8091792746578935e-06,
|
5836 |
+
"loss": 11.6725,
|
5837 |
+
"step": 828
|
5838 |
+
},
|
5839 |
+
{
|
5840 |
+
"epoch": 0.9030501089324618,
|
5841 |
+
"grad_norm": 741.8558349609375,
|
5842 |
+
"learning_rate": 4.703743550911543e-06,
|
5843 |
+
"loss": 12.448,
|
5844 |
+
"step": 829
|
5845 |
+
},
|
5846 |
+
{
|
5847 |
+
"epoch": 0.9041394335511983,
|
5848 |
+
"grad_norm": 637.9666137695312,
|
5849 |
+
"learning_rate": 4.599448611254964e-06,
|
5850 |
+
"loss": 12.7454,
|
5851 |
+
"step": 830
|
5852 |
+
},
|
5853 |
+
{
|
5854 |
+
"epoch": 0.9052287581699346,
|
5855 |
+
"grad_norm": 1003.8346557617188,
|
5856 |
+
"learning_rate": 4.496295704194819e-06,
|
5857 |
+
"loss": 13.0511,
|
5858 |
+
"step": 831
|
5859 |
+
},
|
5860 |
+
{
|
5861 |
+
"epoch": 0.906318082788671,
|
5862 |
+
"grad_norm": 819.62451171875,
|
5863 |
+
"learning_rate": 4.394286064566511e-06,
|
5864 |
+
"loss": 13.2029,
|
5865 |
+
"step": 832
|
5866 |
+
},
|
5867 |
+
{
|
5868 |
+
"epoch": 0.9074074074074074,
|
5869 |
+
"grad_norm": 1023.1488647460938,
|
5870 |
+
"learning_rate": 4.293420913519541e-06,
|
5871 |
+
"loss": 12.9151,
|
5872 |
+
"step": 833
|
5873 |
+
},
|
5874 |
+
{
|
5875 |
+
"epoch": 0.9084967320261438,
|
5876 |
+
"grad_norm": 788.2391967773438,
|
5877 |
+
"learning_rate": 4.193701458502807e-06,
|
5878 |
+
"loss": 12.7021,
|
5879 |
+
"step": 834
|
5880 |
+
},
|
5881 |
+
{
|
5882 |
+
"epoch": 0.9095860566448801,
|
5883 |
+
"grad_norm": 837.283447265625,
|
5884 |
+
"learning_rate": 4.095128893250156e-06,
|
5885 |
+
"loss": 13.0428,
|
5886 |
+
"step": 835
|
5887 |
+
},
|
5888 |
+
{
|
5889 |
+
"epoch": 0.9106753812636166,
|
5890 |
+
"grad_norm": 1206.6253662109375,
|
5891 |
+
"learning_rate": 3.997704397766122e-06,
|
5892 |
+
"loss": 12.3555,
|
5893 |
+
"step": 836
|
5894 |
+
},
|
5895 |
+
{
|
5896 |
+
"epoch": 0.9117647058823529,
|
5897 |
+
"grad_norm": 856.0335083007812,
|
5898 |
+
"learning_rate": 3.901429138311763e-06,
|
5899 |
+
"loss": 12.2161,
|
5900 |
+
"step": 837
|
5901 |
+
},
|
5902 |
+
{
|
5903 |
+
"epoch": 0.9128540305010894,
|
5904 |
+
"grad_norm": 840.4730834960938,
|
5905 |
+
"learning_rate": 3.80630426739077e-06,
|
5906 |
+
"loss": 12.3642,
|
5907 |
+
"step": 838
|
5908 |
+
},
|
5909 |
+
{
|
5910 |
+
"epoch": 0.9139433551198257,
|
5911 |
+
"grad_norm": 767.2365112304688,
|
5912 |
+
"learning_rate": 3.712330923735563e-06,
|
5913 |
+
"loss": 12.338,
|
5914 |
+
"step": 839
|
5915 |
+
},
|
5916 |
+
{
|
5917 |
+
"epoch": 0.9150326797385621,
|
5918 |
+
"grad_norm": 844.8609619140625,
|
5919 |
+
"learning_rate": 3.6195102322937545e-06,
|
5920 |
+
"loss": 12.4701,
|
5921 |
+
"step": 840
|
5922 |
+
},
|
5923 |
+
{
|
5924 |
+
"epoch": 0.9161220043572985,
|
5925 |
+
"grad_norm": 808.1879272460938,
|
5926 |
+
"learning_rate": 3.5278433042146397e-06,
|
5927 |
+
"loss": 12.4012,
|
5928 |
+
"step": 841
|
5929 |
+
},
|
5930 |
+
{
|
5931 |
+
"epoch": 0.9172113289760349,
|
5932 |
+
"grad_norm": 787.5524291992188,
|
5933 |
+
"learning_rate": 3.4373312368358944e-06,
|
5934 |
+
"loss": 12.1528,
|
5935 |
+
"step": 842
|
5936 |
+
},
|
5937 |
+
{
|
5938 |
+
"epoch": 0.9183006535947712,
|
5939 |
+
"grad_norm": 861.0357666015625,
|
5940 |
+
"learning_rate": 3.347975113670454e-06,
|
5941 |
+
"loss": 13.0016,
|
5942 |
+
"step": 843
|
5943 |
+
},
|
5944 |
+
{
|
5945 |
+
"epoch": 0.9193899782135077,
|
5946 |
+
"grad_norm": 1098.5736083984375,
|
5947 |
+
"learning_rate": 3.259776004393533e-06,
|
5948 |
+
"loss": 12.3925,
|
5949 |
+
"step": 844
|
5950 |
+
},
|
5951 |
+
{
|
5952 |
+
"epoch": 0.920479302832244,
|
5953 |
+
"grad_norm": 1017.1472778320312,
|
5954 |
+
"learning_rate": 3.1727349648298267e-06,
|
5955 |
+
"loss": 13.2196,
|
5956 |
+
"step": 845
|
5957 |
+
},
|
5958 |
+
{
|
5959 |
+
"epoch": 0.9215686274509803,
|
5960 |
+
"grad_norm": 729.2058715820312,
|
5961 |
+
"learning_rate": 3.086853036940862e-06,
|
5962 |
+
"loss": 11.8512,
|
5963 |
+
"step": 846
|
5964 |
+
},
|
5965 |
+
{
|
5966 |
+
"epoch": 0.9226579520697168,
|
5967 |
+
"grad_norm": 809.0467529296875,
|
5968 |
+
"learning_rate": 3.0021312488125454e-06,
|
5969 |
+
"loss": 11.8526,
|
5970 |
+
"step": 847
|
5971 |
+
},
|
5972 |
+
{
|
5973 |
+
"epoch": 0.9237472766884531,
|
5974 |
+
"grad_norm": 948.0686645507812,
|
5975 |
+
"learning_rate": 2.9185706146428017e-06,
|
5976 |
+
"loss": 12.4416,
|
5977 |
+
"step": 848
|
5978 |
+
},
|
5979 |
+
{
|
5980 |
+
"epoch": 0.9248366013071896,
|
5981 |
+
"grad_norm": 631.7981567382812,
|
5982 |
+
"learning_rate": 2.836172134729509e-06,
|
5983 |
+
"loss": 12.836,
|
5984 |
+
"step": 849
|
5985 |
+
},
|
5986 |
+
{
|
5987 |
+
"epoch": 0.9259259259259259,
|
5988 |
+
"grad_norm": 641.8875732421875,
|
5989 |
+
"learning_rate": 2.754936795458485e-06,
|
5990 |
+
"loss": 12.6921,
|
5991 |
+
"step": 850
|
5992 |
+
},
|
5993 |
+
{
|
5994 |
+
"epoch": 0.9270152505446623,
|
5995 |
+
"grad_norm": 860.4398803710938,
|
5996 |
+
"learning_rate": 2.674865569291651e-06,
|
5997 |
+
"loss": 13.6532,
|
5998 |
+
"step": 851
|
5999 |
+
},
|
6000 |
+
{
|
6001 |
+
"epoch": 0.9281045751633987,
|
6002 |
+
"grad_norm": 760.2114868164062,
|
6003 |
+
"learning_rate": 2.5959594147554667e-06,
|
6004 |
+
"loss": 13.178,
|
6005 |
+
"step": 852
|
6006 |
+
},
|
6007 |
+
{
|
6008 |
+
"epoch": 0.9291938997821351,
|
6009 |
+
"grad_norm": 1154.3988037109375,
|
6010 |
+
"learning_rate": 2.5182192764293567e-06,
|
6011 |
+
"loss": 12.716,
|
6012 |
+
"step": 853
|
6013 |
+
},
|
6014 |
+
{
|
6015 |
+
"epoch": 0.9302832244008714,
|
6016 |
+
"grad_norm": 762.8448486328125,
|
6017 |
+
"learning_rate": 2.4416460849345123e-06,
|
6018 |
+
"loss": 12.5605,
|
6019 |
+
"step": 854
|
6020 |
+
},
|
6021 |
+
{
|
6022 |
+
"epoch": 0.9313725490196079,
|
6023 |
+
"grad_norm": 1097.0977783203125,
|
6024 |
+
"learning_rate": 2.366240756922644e-06,
|
6025 |
+
"loss": 12.0626,
|
6026 |
+
"step": 855
|
6027 |
+
},
|
6028 |
+
{
|
6029 |
+
"epoch": 0.9324618736383442,
|
6030 |
+
"grad_norm": 989.987060546875,
|
6031 |
+
"learning_rate": 2.2920041950650783e-06,
|
6032 |
+
"loss": 12.1056,
|
6033 |
+
"step": 856
|
6034 |
+
},
|
6035 |
+
{
|
6036 |
+
"epoch": 0.9335511982570807,
|
6037 |
+
"grad_norm": 761.228271484375,
|
6038 |
+
"learning_rate": 2.218937288041956e-06,
|
6039 |
+
"loss": 12.0653,
|
6040 |
+
"step": 857
|
6041 |
+
},
|
6042 |
+
{
|
6043 |
+
"epoch": 0.934640522875817,
|
6044 |
+
"grad_norm": 698.48486328125,
|
6045 |
+
"learning_rate": 2.1470409105315283e-06,
|
6046 |
+
"loss": 12.689,
|
6047 |
+
"step": 858
|
6048 |
+
},
|
6049 |
+
{
|
6050 |
+
"epoch": 0.9357298474945533,
|
6051 |
+
"grad_norm": 1059.8203125,
|
6052 |
+
"learning_rate": 2.0763159231997674e-06,
|
6053 |
+
"loss": 12.268,
|
6054 |
+
"step": 859
|
6055 |
+
},
|
6056 |
+
{
|
6057 |
+
"epoch": 0.9368191721132898,
|
6058 |
+
"grad_norm": 1506.7076416015625,
|
6059 |
+
"learning_rate": 2.0067631726899962e-06,
|
6060 |
+
"loss": 13.0918,
|
6061 |
+
"step": 860
|
6062 |
+
},
|
6063 |
+
{
|
6064 |
+
"epoch": 0.9379084967320261,
|
6065 |
+
"grad_norm": 686.2244262695312,
|
6066 |
+
"learning_rate": 1.938383491612794e-06,
|
6067 |
+
"loss": 13.2276,
|
6068 |
+
"step": 861
|
6069 |
+
},
|
6070 |
+
{
|
6071 |
+
"epoch": 0.9389978213507625,
|
6072 |
+
"grad_norm": 1398.2288818359375,
|
6073 |
+
"learning_rate": 1.8711776985360308e-06,
|
6074 |
+
"loss": 12.4386,
|
6075 |
+
"step": 862
|
6076 |
+
},
|
6077 |
+
{
|
6078 |
+
"epoch": 0.9400871459694989,
|
6079 |
+
"grad_norm": 787.7633666992188,
|
6080 |
+
"learning_rate": 1.805146597975016e-06,
|
6081 |
+
"loss": 12.0259,
|
6082 |
+
"step": 863
|
6083 |
+
},
|
6084 |
+
{
|
6085 |
+
"epoch": 0.9411764705882353,
|
6086 |
+
"grad_norm": 1021.8810424804688,
|
6087 |
+
"learning_rate": 1.7402909803829525e-06,
|
6088 |
+
"loss": 13.2275,
|
6089 |
+
"step": 864
|
6090 |
+
},
|
6091 |
+
{
|
6092 |
+
"epoch": 0.9422657952069716,
|
6093 |
+
"grad_norm": 724.1331176757812,
|
6094 |
+
"learning_rate": 1.6766116221413774e-06,
|
6095 |
+
"loss": 12.9674,
|
6096 |
+
"step": 865
|
6097 |
+
},
|
6098 |
+
{
|
6099 |
+
"epoch": 0.9433551198257081,
|
6100 |
+
"grad_norm": 889.06201171875,
|
6101 |
+
"learning_rate": 1.61410928555098e-06,
|
6102 |
+
"loss": 12.9815,
|
6103 |
+
"step": 866
|
6104 |
+
},
|
6105 |
+
{
|
6106 |
+
"epoch": 0.9444444444444444,
|
6107 |
+
"grad_norm": 694.7861938476562,
|
6108 |
+
"learning_rate": 1.5527847188223644e-06,
|
6109 |
+
"loss": 11.78,
|
6110 |
+
"step": 867
|
6111 |
+
},
|
6112 |
+
{
|
6113 |
+
"epoch": 0.9455337690631809,
|
6114 |
+
"grad_norm": 1126.9815673828125,
|
6115 |
+
"learning_rate": 1.4926386560671358e-06,
|
6116 |
+
"loss": 13.2658,
|
6117 |
+
"step": 868
|
6118 |
+
},
|
6119 |
+
{
|
6120 |
+
"epoch": 0.9466230936819172,
|
6121 |
+
"grad_norm": 1251.4708251953125,
|
6122 |
+
"learning_rate": 1.433671817289184e-06,
|
6123 |
+
"loss": 13.5727,
|
6124 |
+
"step": 869
|
6125 |
+
},
|
6126 |
+
{
|
6127 |
+
"epoch": 0.9477124183006536,
|
6128 |
+
"grad_norm": 928.8302001953125,
|
6129 |
+
"learning_rate": 1.3758849083759352e-06,
|
6130 |
+
"loss": 12.5813,
|
6131 |
+
"step": 870
|
6132 |
+
},
|
6133 |
+
{
|
6134 |
+
"epoch": 0.94880174291939,
|
6135 |
+
"grad_norm": 906.8268432617188,
|
6136 |
+
"learning_rate": 1.3192786210900033e-06,
|
6137 |
+
"loss": 12.3939,
|
6138 |
+
"step": 871
|
6139 |
+
},
|
6140 |
+
{
|
6141 |
+
"epoch": 0.9498910675381264,
|
6142 |
+
"grad_norm": 1406.4046630859375,
|
6143 |
+
"learning_rate": 1.2638536330608408e-06,
|
6144 |
+
"loss": 12.5356,
|
6145 |
+
"step": 872
|
6146 |
+
},
|
6147 |
+
{
|
6148 |
+
"epoch": 0.9509803921568627,
|
6149 |
+
"grad_norm": 734.1050415039062,
|
6150 |
+
"learning_rate": 1.2096106077767011e-06,
|
6151 |
+
"loss": 12.0204,
|
6152 |
+
"step": 873
|
6153 |
+
},
|
6154 |
+
{
|
6155 |
+
"epoch": 0.9520697167755992,
|
6156 |
+
"grad_norm": 1201.427734375,
|
6157 |
+
"learning_rate": 1.1565501945766222e-06,
|
6158 |
+
"loss": 11.8001,
|
6159 |
+
"step": 874
|
6160 |
+
},
|
6161 |
+
{
|
6162 |
+
"epoch": 0.9531590413943355,
|
6163 |
+
"grad_norm": 964.8887329101562,
|
6164 |
+
"learning_rate": 1.1046730286426775e-06,
|
6165 |
+
"loss": 13.3203,
|
6166 |
+
"step": 875
|
6167 |
+
},
|
6168 |
+
{
|
6169 |
+
"epoch": 0.954248366013072,
|
6170 |
+
"grad_norm": 1066.5933837890625,
|
6171 |
+
"learning_rate": 1.053979730992416e-06,
|
6172 |
+
"loss": 12.9455,
|
6173 |
+
"step": 876
|
6174 |
+
},
|
6175 |
+
{
|
6176 |
+
"epoch": 0.9553376906318083,
|
6177 |
+
"grad_norm": 867.9620361328125,
|
6178 |
+
"learning_rate": 1.0044709084713554e-06,
|
6179 |
+
"loss": 12.5005,
|
6180 |
+
"step": 877
|
6181 |
+
},
|
6182 |
+
{
|
6183 |
+
"epoch": 0.9564270152505446,
|
6184 |
+
"grad_norm": 1209.2740478515625,
|
6185 |
+
"learning_rate": 9.56147153745779e-07,
|
6186 |
+
"loss": 12.6775,
|
6187 |
+
"step": 878
|
6188 |
+
},
|
6189 |
+
{
|
6190 |
+
"epoch": 0.9575163398692811,
|
6191 |
+
"grad_norm": 920.9111938476562,
|
6192 |
+
"learning_rate": 9.090090452955835e-07,
|
6193 |
+
"loss": 12.656,
|
6194 |
+
"step": 879
|
6195 |
+
},
|
6196 |
+
{
|
6197 |
+
"epoch": 0.9586056644880174,
|
6198 |
+
"grad_norm": 1178.7100830078125,
|
6199 |
+
"learning_rate": 8.630571474074311e-07,
|
6200 |
+
"loss": 11.8858,
|
6201 |
+
"step": 880
|
6202 |
+
},
|
6203 |
+
{
|
6204 |
+
"epoch": 0.9596949891067538,
|
6205 |
+
"grad_norm": 827.2933349609375,
|
6206 |
+
"learning_rate": 8.182920101679092e-07,
|
6207 |
+
"loss": 12.7875,
|
6208 |
+
"step": 881
|
6209 |
+
},
|
6210 |
+
{
|
6211 |
+
"epoch": 0.9607843137254902,
|
6212 |
+
"grad_norm": 970.2561645507812,
|
6213 |
+
"learning_rate": 7.747141694570026e-07,
|
6214 |
+
"loss": 12.1683,
|
6215 |
+
"step": 882
|
6216 |
+
},
|
6217 |
+
{
|
6218 |
+
"epoch": 0.9618736383442266,
|
6219 |
+
"grad_norm": 961.2227783203125,
|
6220 |
+
"learning_rate": 7.323241469416764e-07,
|
6221 |
+
"loss": 13.4533,
|
6222 |
+
"step": 883
|
6223 |
+
},
|
6224 |
+
{
|
6225 |
+
"epoch": 0.9629629629629629,
|
6226 |
+
"grad_norm": 1200.774658203125,
|
6227 |
+
"learning_rate": 6.911224500695702e-07,
|
6228 |
+
"loss": 13.4708,
|
6229 |
+
"step": 884
|
6230 |
+
},
|
6231 |
+
{
|
6232 |
+
"epoch": 0.9640522875816994,
|
6233 |
+
"grad_norm": 961.2836303710938,
|
6234 |
+
"learning_rate": 6.511095720630244e-07,
|
6235 |
+
"loss": 12.1307,
|
6236 |
+
"step": 885
|
6237 |
+
},
|
6238 |
+
{
|
6239 |
+
"epoch": 0.9651416122004357,
|
6240 |
+
"grad_norm": 949.4239501953125,
|
6241 |
+
"learning_rate": 6.122859919130974e-07,
|
6242 |
+
"loss": 11.6507,
|
6243 |
+
"step": 886
|
6244 |
+
},
|
6245 |
+
{
|
6246 |
+
"epoch": 0.9662309368191722,
|
6247 |
+
"grad_norm": 1107.519287109375,
|
6248 |
+
"learning_rate": 5.746521743738354e-07,
|
6249 |
+
"loss": 13.5597,
|
6250 |
+
"step": 887
|
6251 |
+
},
|
6252 |
+
{
|
6253 |
+
"epoch": 0.9673202614379085,
|
6254 |
+
"grad_norm": 713.34619140625,
|
6255 |
+
"learning_rate": 5.382085699567552e-07,
|
6256 |
+
"loss": 13.0859,
|
6257 |
+
"step": 888
|
6258 |
+
},
|
6259 |
+
{
|
6260 |
+
"epoch": 0.9684095860566448,
|
6261 |
+
"grad_norm": 763.1091918945312,
|
6262 |
+
"learning_rate": 5.029556149254266e-07,
|
6263 |
+
"loss": 12.9483,
|
6264 |
+
"step": 889
|
6265 |
+
},
|
6266 |
+
{
|
6267 |
+
"epoch": 0.9694989106753813,
|
6268 |
+
"grad_norm": 913.829345703125,
|
6269 |
+
"learning_rate": 4.6889373129022085e-07,
|
6270 |
+
"loss": 11.2009,
|
6271 |
+
"step": 890
|
6272 |
+
},
|
6273 |
+
{
|
6274 |
+
"epoch": 0.9705882352941176,
|
6275 |
+
"grad_norm": 1170.449462890625,
|
6276 |
+
"learning_rate": 4.3602332680331425e-07,
|
6277 |
+
"loss": 12.748,
|
6278 |
+
"step": 891
|
6279 |
+
},
|
6280 |
+
{
|
6281 |
+
"epoch": 0.971677559912854,
|
6282 |
+
"grad_norm": 877.0611572265625,
|
6283 |
+
"learning_rate": 4.0434479495378155e-07,
|
6284 |
+
"loss": 12.1814,
|
6285 |
+
"step": 892
|
6286 |
+
},
|
6287 |
+
{
|
6288 |
+
"epoch": 0.9727668845315904,
|
6289 |
+
"grad_norm": 1181.8524169921875,
|
6290 |
+
"learning_rate": 3.7385851496284374e-07,
|
6291 |
+
"loss": 12.2686,
|
6292 |
+
"step": 893
|
6293 |
+
},
|
6294 |
+
{
|
6295 |
+
"epoch": 0.9738562091503268,
|
6296 |
+
"grad_norm": 919.8130493164062,
|
6297 |
+
"learning_rate": 3.445648517793942e-07,
|
6298 |
+
"loss": 12.535,
|
6299 |
+
"step": 894
|
6300 |
+
},
|
6301 |
+
{
|
6302 |
+
"epoch": 0.9749455337690632,
|
6303 |
+
"grad_norm": 736.0946044921875,
|
6304 |
+
"learning_rate": 3.164641560756132e-07,
|
6305 |
+
"loss": 13.5968,
|
6306 |
+
"step": 895
|
6307 |
+
},
|
6308 |
+
{
|
6309 |
+
"epoch": 0.9760348583877996,
|
6310 |
+
"grad_norm": 1050.005126953125,
|
6311 |
+
"learning_rate": 2.895567642427488e-07,
|
6312 |
+
"loss": 13.5456,
|
6313 |
+
"step": 896
|
6314 |
+
},
|
6315 |
+
{
|
6316 |
+
"epoch": 0.9771241830065359,
|
6317 |
+
"grad_norm": 1240.2838134765625,
|
6318 |
+
"learning_rate": 2.638429983870983e-07,
|
6319 |
+
"loss": 12.4089,
|
6320 |
+
"step": 897
|
6321 |
+
},
|
6322 |
+
{
|
6323 |
+
"epoch": 0.9782135076252724,
|
6324 |
+
"grad_norm": 1174.1912841796875,
|
6325 |
+
"learning_rate": 2.3932316632614416e-07,
|
6326 |
+
"loss": 12.5838,
|
6327 |
+
"step": 898
|
6328 |
+
},
|
6329 |
+
{
|
6330 |
+
"epoch": 0.9793028322440087,
|
6331 |
+
"grad_norm": 900.91845703125,
|
6332 |
+
"learning_rate": 2.15997561584913e-07,
|
6333 |
+
"loss": 12.2009,
|
6334 |
+
"step": 899
|
6335 |
+
},
|
6336 |
+
{
|
6337 |
+
"epoch": 0.9803921568627451,
|
6338 |
+
"grad_norm": 1263.928466796875,
|
6339 |
+
"learning_rate": 1.9386646339238924e-07,
|
6340 |
+
"loss": 13.1211,
|
6341 |
+
"step": 900
|
6342 |
+
},
|
6343 |
+
{
|
6344 |
+
"epoch": 0.9814814814814815,
|
6345 |
+
"grad_norm": 1037.0902099609375,
|
6346 |
+
"learning_rate": 1.7293013667825098e-07,
|
6347 |
+
"loss": 12.676,
|
6348 |
+
"step": 901
|
6349 |
+
},
|
6350 |
+
{
|
6351 |
+
"epoch": 0.9825708061002179,
|
6352 |
+
"grad_norm": 1237.9251708984375,
|
6353 |
+
"learning_rate": 1.5318883206962842e-07,
|
6354 |
+
"loss": 12.2276,
|
6355 |
+
"step": 902
|
6356 |
+
},
|
6357 |
+
{
|
6358 |
+
"epoch": 0.9836601307189542,
|
6359 |
+
"grad_norm": 630.1217041015625,
|
6360 |
+
"learning_rate": 1.3464278588815048e-07,
|
6361 |
+
"loss": 12.1347,
|
6362 |
+
"step": 903
|
6363 |
+
},
|
6364 |
+
{
|
6365 |
+
"epoch": 0.9847494553376906,
|
6366 |
+
"grad_norm": 859.3702392578125,
|
6367 |
+
"learning_rate": 1.1729222014709162e-07,
|
6368 |
+
"loss": 12.4704,
|
6369 |
+
"step": 904
|
6370 |
+
},
|
6371 |
+
{
|
6372 |
+
"epoch": 0.985838779956427,
|
6373 |
+
"grad_norm": 1073.7818603515625,
|
6374 |
+
"learning_rate": 1.0113734254872942e-07,
|
6375 |
+
"loss": 12.5898,
|
6376 |
+
"step": 905
|
6377 |
+
},
|
6378 |
+
{
|
6379 |
+
"epoch": 0.9869281045751634,
|
6380 |
+
"grad_norm": 1251.8004150390625,
|
6381 |
+
"learning_rate": 8.617834648185774e-08,
|
6382 |
+
"loss": 13.251,
|
6383 |
+
"step": 906
|
6384 |
+
},
|
6385 |
+
{
|
6386 |
+
"epoch": 0.9880174291938998,
|
6387 |
+
"grad_norm": 1012.1276245117188,
|
6388 |
+
"learning_rate": 7.241541101945526e-08,
|
6389 |
+
"loss": 11.1789,
|
6390 |
+
"step": 907
|
6391 |
+
},
|
6392 |
+
{
|
6393 |
+
"epoch": 0.9891067538126361,
|
6394 |
+
"grad_norm": 662.820556640625,
|
6395 |
+
"learning_rate": 5.984870091654271e-08,
|
6396 |
+
"loss": 13.627,
|
6397 |
+
"step": 908
|
6398 |
+
},
|
6399 |
+
{
|
6400 |
+
"epoch": 0.9901960784313726,
|
6401 |
+
"grad_norm": 771.7218627929688,
|
6402 |
+
"learning_rate": 4.847836660824001e-08,
|
6403 |
+
"loss": 12.1929,
|
6404 |
+
"step": 909
|
6405 |
+
},
|
6406 |
+
{
|
6407 |
+
"epoch": 0.9912854030501089,
|
6408 |
+
"grad_norm": 654.5511474609375,
|
6409 |
+
"learning_rate": 3.8304544207945495e-08,
|
6410 |
+
"loss": 11.9869,
|
6411 |
+
"step": 910
|
6412 |
+
},
|
6413 |
+
{
|
6414 |
+
"epoch": 0.9923747276688453,
|
6415 |
+
"grad_norm": 1049.4482421875,
|
6416 |
+
"learning_rate": 2.9327355505681663e-08,
|
6417 |
+
"loss": 12.3512,
|
6418 |
+
"step": 911
|
6419 |
+
},
|
6420 |
+
{
|
6421 |
+
"epoch": 0.9934640522875817,
|
6422 |
+
"grad_norm": 861.9254150390625,
|
6423 |
+
"learning_rate": 2.1546907966685236e-08,
|
6424 |
+
"loss": 12.2443,
|
6425 |
+
"step": 912
|
6426 |
+
},
|
6427 |
+
{
|
6428 |
+
"epoch": 0.9945533769063181,
|
6429 |
+
"grad_norm": 1027.2105712890625,
|
6430 |
+
"learning_rate": 1.496329473008595e-08,
|
6431 |
+
"loss": 12.2233,
|
6432 |
+
"step": 913
|
6433 |
+
},
|
6434 |
+
{
|
6435 |
+
"epoch": 0.9956427015250545,
|
6436 |
+
"grad_norm": 783.8712158203125,
|
6437 |
+
"learning_rate": 9.576594607807465e-09,
|
6438 |
+
"loss": 13.0014,
|
6439 |
+
"step": 914
|
6440 |
+
},
|
6441 |
+
{
|
6442 |
+
"epoch": 0.9967320261437909,
|
6443 |
+
"grad_norm": 1203.1295166015625,
|
6444 |
+
"learning_rate": 5.3868720836236506e-09,
|
6445 |
+
"loss": 13.0101,
|
6446 |
+
"step": 915
|
6447 |
+
},
|
6448 |
+
{
|
6449 |
+
"epoch": 0.9978213507625272,
|
6450 |
+
"grad_norm": 702.3920288085938,
|
6451 |
+
"learning_rate": 2.3941773123814516e-09,
|
6452 |
+
"loss": 13.0705,
|
6453 |
+
"step": 916
|
6454 |
+
},
|
6455 |
+
{
|
6456 |
+
"epoch": 0.9989106753812637,
|
6457 |
+
"grad_norm": 858.6944580078125,
|
6458 |
+
"learning_rate": 5.985461193791509e-10,
|
6459 |
+
"loss": 12.3027,
|
6460 |
+
"step": 917
|
6461 |
+
},
|
6462 |
+
{
|
6463 |
+
"epoch": 1.0,
|
6464 |
+
"grad_norm": 1008.3307495117188,
|
6465 |
+
"learning_rate": 0.0,
|
6466 |
+
"loss": 12.5451,
|
6467 |
+
"step": 918
|
6468 |
}
|
6469 |
],
|
6470 |
"logging_steps": 1,
|
|
|
6479 |
"should_evaluate": false,
|
6480 |
"should_log": false,
|
6481 |
"should_save": true,
|
6482 |
+
"should_training_stop": true
|
6483 |
},
|
6484 |
"attributes": {}
|
6485 |
}
|
6486 |
},
|
6487 |
+
"total_flos": 3120952812503040.0,
|
6488 |
"train_batch_size": 2,
|
6489 |
"trial_name": null,
|
6490 |
"trial_params": null
|