Training in progress, step 1378, 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 109009752
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8e594ca116971a4e4ab3a88569e01a56bcd5e81da6ac110c770d8071f9df3bd5
|
3 |
size 109009752
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 56088788
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:fba6a3e83839be616217177d8872707bce691805eaaee12021cb863a70babc1f
|
3 |
size 56088788
|
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 sha256:8d065c5a8080e93829770dda5e134034bcc876ac05a53d0fe996c233149b9da3
|
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:c109da43816272dd893db12818af67423aca35f0abeca5da7f9166176ccc48e4
|
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": 345,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -7284,6 +7284,2407 @@
|
|
7284 |
"eval_samples_per_second": 18.874,
|
7285 |
"eval_steps_per_second": 9.439,
|
7286 |
"step": 1035
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7287 |
}
|
7288 |
],
|
7289 |
"logging_steps": 1,
|
@@ -7298,12 +9699,12 @@
|
|
7298 |
"should_evaluate": false,
|
7299 |
"should_log": false,
|
7300 |
"should_save": true,
|
7301 |
-
"should_training_stop":
|
7302 |
},
|
7303 |
"attributes": {}
|
7304 |
}
|
7305 |
},
|
7306 |
-
"total_flos":
|
7307 |
"train_batch_size": 2,
|
7308 |
"trial_name": null,
|
7309 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.0978085351787774,
|
5 |
"eval_steps": 345,
|
6 |
+
"global_step": 1378,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
7284 |
"eval_samples_per_second": 18.874,
|
7285 |
"eval_steps_per_second": 9.439,
|
7286 |
"step": 1035
|
7287 |
+
},
|
7288 |
+
{
|
7289 |
+
"epoch": 0.0735338479283116,
|
7290 |
+
"grad_norm": 0.5463880896568298,
|
7291 |
+
"learning_rate": 2.9289321881345254e-05,
|
7292 |
+
"loss": 1.0284,
|
7293 |
+
"step": 1036
|
7294 |
+
},
|
7295 |
+
{
|
7296 |
+
"epoch": 0.07360482654600302,
|
7297 |
+
"grad_norm": 0.5464505553245544,
|
7298 |
+
"learning_rate": 2.9127122410354647e-05,
|
7299 |
+
"loss": 1.1125,
|
7300 |
+
"step": 1037
|
7301 |
+
},
|
7302 |
+
{
|
7303 |
+
"epoch": 0.07367580516369444,
|
7304 |
+
"grad_norm": 0.5442793965339661,
|
7305 |
+
"learning_rate": 2.89652967119336e-05,
|
7306 |
+
"loss": 1.2379,
|
7307 |
+
"step": 1038
|
7308 |
+
},
|
7309 |
+
{
|
7310 |
+
"epoch": 0.07374678378138586,
|
7311 |
+
"grad_norm": 0.5177668929100037,
|
7312 |
+
"learning_rate": 2.8803845639525728e-05,
|
7313 |
+
"loss": 0.9388,
|
7314 |
+
"step": 1039
|
7315 |
+
},
|
7316 |
+
{
|
7317 |
+
"epoch": 0.07381776239907728,
|
7318 |
+
"grad_norm": 0.4273097813129425,
|
7319 |
+
"learning_rate": 2.8642770044598966e-05,
|
7320 |
+
"loss": 0.8004,
|
7321 |
+
"step": 1040
|
7322 |
+
},
|
7323 |
+
{
|
7324 |
+
"epoch": 0.0738887410167687,
|
7325 |
+
"grad_norm": 0.8218151926994324,
|
7326 |
+
"learning_rate": 2.848207077664099e-05,
|
7327 |
+
"loss": 1.2082,
|
7328 |
+
"step": 1041
|
7329 |
+
},
|
7330 |
+
{
|
7331 |
+
"epoch": 0.07395971963446012,
|
7332 |
+
"grad_norm": 0.490460067987442,
|
7333 |
+
"learning_rate": 2.8321748683154893e-05,
|
7334 |
+
"loss": 0.8117,
|
7335 |
+
"step": 1042
|
7336 |
+
},
|
7337 |
+
{
|
7338 |
+
"epoch": 0.07403069825215154,
|
7339 |
+
"grad_norm": 0.690358579158783,
|
7340 |
+
"learning_rate": 2.8161804609654484e-05,
|
7341 |
+
"loss": 0.9035,
|
7342 |
+
"step": 1043
|
7343 |
+
},
|
7344 |
+
{
|
7345 |
+
"epoch": 0.07410167686984295,
|
7346 |
+
"grad_norm": 0.498638778924942,
|
7347 |
+
"learning_rate": 2.800223939966007e-05,
|
7348 |
+
"loss": 0.9094,
|
7349 |
+
"step": 1044
|
7350 |
+
},
|
7351 |
+
{
|
7352 |
+
"epoch": 0.07417265548753438,
|
7353 |
+
"grad_norm": 0.7770102620124817,
|
7354 |
+
"learning_rate": 2.7843053894693803e-05,
|
7355 |
+
"loss": 1.151,
|
7356 |
+
"step": 1045
|
7357 |
+
},
|
7358 |
+
{
|
7359 |
+
"epoch": 0.0742436341052258,
|
7360 |
+
"grad_norm": 0.6199392676353455,
|
7361 |
+
"learning_rate": 2.7684248934275325e-05,
|
7362 |
+
"loss": 1.1426,
|
7363 |
+
"step": 1046
|
7364 |
+
},
|
7365 |
+
{
|
7366 |
+
"epoch": 0.07431461272291721,
|
7367 |
+
"grad_norm": 0.5609120726585388,
|
7368 |
+
"learning_rate": 2.7525825355917446e-05,
|
7369 |
+
"loss": 0.9465,
|
7370 |
+
"step": 1047
|
7371 |
+
},
|
7372 |
+
{
|
7373 |
+
"epoch": 0.07438559134060864,
|
7374 |
+
"grad_norm": 0.5529163479804993,
|
7375 |
+
"learning_rate": 2.73677839951215e-05,
|
7376 |
+
"loss": 0.8544,
|
7377 |
+
"step": 1048
|
7378 |
+
},
|
7379 |
+
{
|
7380 |
+
"epoch": 0.07445656995830006,
|
7381 |
+
"grad_norm": 0.4198525846004486,
|
7382 |
+
"learning_rate": 2.721012568537318e-05,
|
7383 |
+
"loss": 1.0313,
|
7384 |
+
"step": 1049
|
7385 |
+
},
|
7386 |
+
{
|
7387 |
+
"epoch": 0.07452754857599148,
|
7388 |
+
"grad_norm": 1.1126874685287476,
|
7389 |
+
"learning_rate": 2.7052851258137935e-05,
|
7390 |
+
"loss": 1.2214,
|
7391 |
+
"step": 1050
|
7392 |
+
},
|
7393 |
+
{
|
7394 |
+
"epoch": 0.0745985271936829,
|
7395 |
+
"grad_norm": 0.5191254615783691,
|
7396 |
+
"learning_rate": 2.689596154285672e-05,
|
7397 |
+
"loss": 0.7111,
|
7398 |
+
"step": 1051
|
7399 |
+
},
|
7400 |
+
{
|
7401 |
+
"epoch": 0.07466950581137433,
|
7402 |
+
"grad_norm": 0.5139110088348389,
|
7403 |
+
"learning_rate": 2.6739457366941543e-05,
|
7404 |
+
"loss": 1.0698,
|
7405 |
+
"step": 1052
|
7406 |
+
},
|
7407 |
+
{
|
7408 |
+
"epoch": 0.07474048442906574,
|
7409 |
+
"grad_norm": 0.6746183633804321,
|
7410 |
+
"learning_rate": 2.6583339555771192e-05,
|
7411 |
+
"loss": 1.1982,
|
7412 |
+
"step": 1053
|
7413 |
+
},
|
7414 |
+
{
|
7415 |
+
"epoch": 0.07481146304675716,
|
7416 |
+
"grad_norm": 0.528917133808136,
|
7417 |
+
"learning_rate": 2.6427608932686843e-05,
|
7418 |
+
"loss": 1.128,
|
7419 |
+
"step": 1054
|
7420 |
+
},
|
7421 |
+
{
|
7422 |
+
"epoch": 0.07488244166444859,
|
7423 |
+
"grad_norm": 0.49486279487609863,
|
7424 |
+
"learning_rate": 2.6272266318987603e-05,
|
7425 |
+
"loss": 0.7873,
|
7426 |
+
"step": 1055
|
7427 |
+
},
|
7428 |
+
{
|
7429 |
+
"epoch": 0.07495342028214,
|
7430 |
+
"grad_norm": 0.4257170557975769,
|
7431 |
+
"learning_rate": 2.6117312533926362e-05,
|
7432 |
+
"loss": 0.761,
|
7433 |
+
"step": 1056
|
7434 |
+
},
|
7435 |
+
{
|
7436 |
+
"epoch": 0.07502439889983142,
|
7437 |
+
"grad_norm": 0.6389133334159851,
|
7438 |
+
"learning_rate": 2.5962748394705328e-05,
|
7439 |
+
"loss": 0.8278,
|
7440 |
+
"step": 1057
|
7441 |
+
},
|
7442 |
+
{
|
7443 |
+
"epoch": 0.07509537751752285,
|
7444 |
+
"grad_norm": 0.5689634084701538,
|
7445 |
+
"learning_rate": 2.5808574716471856e-05,
|
7446 |
+
"loss": 0.9675,
|
7447 |
+
"step": 1058
|
7448 |
+
},
|
7449 |
+
{
|
7450 |
+
"epoch": 0.07516635613521426,
|
7451 |
+
"grad_norm": 0.4619472026824951,
|
7452 |
+
"learning_rate": 2.565479231231398e-05,
|
7453 |
+
"loss": 0.6782,
|
7454 |
+
"step": 1059
|
7455 |
+
},
|
7456 |
+
{
|
7457 |
+
"epoch": 0.07523733475290569,
|
7458 |
+
"grad_norm": 0.45457616448402405,
|
7459 |
+
"learning_rate": 2.55014019932563e-05,
|
7460 |
+
"loss": 0.8111,
|
7461 |
+
"step": 1060
|
7462 |
+
},
|
7463 |
+
{
|
7464 |
+
"epoch": 0.07530831337059711,
|
7465 |
+
"grad_norm": 0.6634262800216675,
|
7466 |
+
"learning_rate": 2.534840456825558e-05,
|
7467 |
+
"loss": 0.8143,
|
7468 |
+
"step": 1061
|
7469 |
+
},
|
7470 |
+
{
|
7471 |
+
"epoch": 0.07537929198828852,
|
7472 |
+
"grad_norm": 0.6119337677955627,
|
7473 |
+
"learning_rate": 2.519580084419646e-05,
|
7474 |
+
"loss": 0.8083,
|
7475 |
+
"step": 1062
|
7476 |
+
},
|
7477 |
+
{
|
7478 |
+
"epoch": 0.07545027060597995,
|
7479 |
+
"grad_norm": 0.3665156960487366,
|
7480 |
+
"learning_rate": 2.5043591625887407e-05,
|
7481 |
+
"loss": 0.6853,
|
7482 |
+
"step": 1063
|
7483 |
+
},
|
7484 |
+
{
|
7485 |
+
"epoch": 0.07552124922367137,
|
7486 |
+
"grad_norm": 0.7194316983222961,
|
7487 |
+
"learning_rate": 2.4891777716056176e-05,
|
7488 |
+
"loss": 1.0921,
|
7489 |
+
"step": 1064
|
7490 |
+
},
|
7491 |
+
{
|
7492 |
+
"epoch": 0.07559222784136278,
|
7493 |
+
"grad_norm": 0.3780640661716461,
|
7494 |
+
"learning_rate": 2.4740359915345835e-05,
|
7495 |
+
"loss": 0.9959,
|
7496 |
+
"step": 1065
|
7497 |
+
},
|
7498 |
+
{
|
7499 |
+
"epoch": 0.07566320645905421,
|
7500 |
+
"grad_norm": 0.5528492331504822,
|
7501 |
+
"learning_rate": 2.4589339022310386e-05,
|
7502 |
+
"loss": 0.7746,
|
7503 |
+
"step": 1066
|
7504 |
+
},
|
7505 |
+
{
|
7506 |
+
"epoch": 0.07573418507674563,
|
7507 |
+
"grad_norm": 0.532551109790802,
|
7508 |
+
"learning_rate": 2.4438715833410595e-05,
|
7509 |
+
"loss": 1.1754,
|
7510 |
+
"step": 1067
|
7511 |
+
},
|
7512 |
+
{
|
7513 |
+
"epoch": 0.07580516369443704,
|
7514 |
+
"grad_norm": 0.4763934016227722,
|
7515 |
+
"learning_rate": 2.4288491143009795e-05,
|
7516 |
+
"loss": 0.9457,
|
7517 |
+
"step": 1068
|
7518 |
+
},
|
7519 |
+
{
|
7520 |
+
"epoch": 0.07587614231212847,
|
7521 |
+
"grad_norm": 0.4264293611049652,
|
7522 |
+
"learning_rate": 2.413866574336975e-05,
|
7523 |
+
"loss": 0.7942,
|
7524 |
+
"step": 1069
|
7525 |
+
},
|
7526 |
+
{
|
7527 |
+
"epoch": 0.0759471209298199,
|
7528 |
+
"grad_norm": 0.5233449935913086,
|
7529 |
+
"learning_rate": 2.3989240424646355e-05,
|
7530 |
+
"loss": 0.9646,
|
7531 |
+
"step": 1070
|
7532 |
+
},
|
7533 |
+
{
|
7534 |
+
"epoch": 0.0760180995475113,
|
7535 |
+
"grad_norm": 0.4843657314777374,
|
7536 |
+
"learning_rate": 2.3840215974885615e-05,
|
7537 |
+
"loss": 0.8398,
|
7538 |
+
"step": 1071
|
7539 |
+
},
|
7540 |
+
{
|
7541 |
+
"epoch": 0.07608907816520273,
|
7542 |
+
"grad_norm": 0.43439170718193054,
|
7543 |
+
"learning_rate": 2.3691593180019366e-05,
|
7544 |
+
"loss": 0.7895,
|
7545 |
+
"step": 1072
|
7546 |
+
},
|
7547 |
+
{
|
7548 |
+
"epoch": 0.07616005678289416,
|
7549 |
+
"grad_norm": 0.513238787651062,
|
7550 |
+
"learning_rate": 2.3543372823861154e-05,
|
7551 |
+
"loss": 0.9543,
|
7552 |
+
"step": 1073
|
7553 |
+
},
|
7554 |
+
{
|
7555 |
+
"epoch": 0.07623103540058557,
|
7556 |
+
"grad_norm": 0.5333254337310791,
|
7557 |
+
"learning_rate": 2.339555568810221e-05,
|
7558 |
+
"loss": 1.0294,
|
7559 |
+
"step": 1074
|
7560 |
+
},
|
7561 |
+
{
|
7562 |
+
"epoch": 0.07630201401827699,
|
7563 |
+
"grad_norm": 0.44629040360450745,
|
7564 |
+
"learning_rate": 2.324814255230714e-05,
|
7565 |
+
"loss": 0.81,
|
7566 |
+
"step": 1075
|
7567 |
+
},
|
7568 |
+
{
|
7569 |
+
"epoch": 0.07637299263596842,
|
7570 |
+
"grad_norm": 0.5009654760360718,
|
7571 |
+
"learning_rate": 2.3101134193910024e-05,
|
7572 |
+
"loss": 0.95,
|
7573 |
+
"step": 1076
|
7574 |
+
},
|
7575 |
+
{
|
7576 |
+
"epoch": 0.07644397125365983,
|
7577 |
+
"grad_norm": 0.45047643780708313,
|
7578 |
+
"learning_rate": 2.295453138821013e-05,
|
7579 |
+
"loss": 0.904,
|
7580 |
+
"step": 1077
|
7581 |
+
},
|
7582 |
+
{
|
7583 |
+
"epoch": 0.07651494987135125,
|
7584 |
+
"grad_norm": 0.7342311143875122,
|
7585 |
+
"learning_rate": 2.2808334908367914e-05,
|
7586 |
+
"loss": 1.3786,
|
7587 |
+
"step": 1078
|
7588 |
+
},
|
7589 |
+
{
|
7590 |
+
"epoch": 0.07658592848904268,
|
7591 |
+
"grad_norm": 0.42931681871414185,
|
7592 |
+
"learning_rate": 2.266254552540097e-05,
|
7593 |
+
"loss": 1.0701,
|
7594 |
+
"step": 1079
|
7595 |
+
},
|
7596 |
+
{
|
7597 |
+
"epoch": 0.07665690710673409,
|
7598 |
+
"grad_norm": 0.49413037300109863,
|
7599 |
+
"learning_rate": 2.2517164008179882e-05,
|
7600 |
+
"loss": 1.0956,
|
7601 |
+
"step": 1080
|
7602 |
+
},
|
7603 |
+
{
|
7604 |
+
"epoch": 0.07672788572442552,
|
7605 |
+
"grad_norm": 0.3613120913505554,
|
7606 |
+
"learning_rate": 2.237219112342426e-05,
|
7607 |
+
"loss": 0.5434,
|
7608 |
+
"step": 1081
|
7609 |
+
},
|
7610 |
+
{
|
7611 |
+
"epoch": 0.07679886434211694,
|
7612 |
+
"grad_norm": 0.4425318241119385,
|
7613 |
+
"learning_rate": 2.222762763569862e-05,
|
7614 |
+
"loss": 0.9588,
|
7615 |
+
"step": 1082
|
7616 |
+
},
|
7617 |
+
{
|
7618 |
+
"epoch": 0.07686984295980835,
|
7619 |
+
"grad_norm": 0.5189344882965088,
|
7620 |
+
"learning_rate": 2.208347430740837e-05,
|
7621 |
+
"loss": 0.8712,
|
7622 |
+
"step": 1083
|
7623 |
+
},
|
7624 |
+
{
|
7625 |
+
"epoch": 0.07694082157749978,
|
7626 |
+
"grad_norm": 0.5996214747428894,
|
7627 |
+
"learning_rate": 2.1939731898795802e-05,
|
7628 |
+
"loss": 0.99,
|
7629 |
+
"step": 1084
|
7630 |
+
},
|
7631 |
+
{
|
7632 |
+
"epoch": 0.0770118001951912,
|
7633 |
+
"grad_norm": 0.6045960783958435,
|
7634 |
+
"learning_rate": 2.1796401167936143e-05,
|
7635 |
+
"loss": 1.0449,
|
7636 |
+
"step": 1085
|
7637 |
+
},
|
7638 |
+
{
|
7639 |
+
"epoch": 0.07708277881288261,
|
7640 |
+
"grad_norm": 0.5072668790817261,
|
7641 |
+
"learning_rate": 2.165348287073339e-05,
|
7642 |
+
"loss": 1.1151,
|
7643 |
+
"step": 1086
|
7644 |
+
},
|
7645 |
+
{
|
7646 |
+
"epoch": 0.07715375743057404,
|
7647 |
+
"grad_norm": 0.7324336767196655,
|
7648 |
+
"learning_rate": 2.151097776091654e-05,
|
7649 |
+
"loss": 1.2269,
|
7650 |
+
"step": 1087
|
7651 |
+
},
|
7652 |
+
{
|
7653 |
+
"epoch": 0.07722473604826546,
|
7654 |
+
"grad_norm": 0.5092812180519104,
|
7655 |
+
"learning_rate": 2.1368886590035443e-05,
|
7656 |
+
"loss": 0.8989,
|
7657 |
+
"step": 1088
|
7658 |
+
},
|
7659 |
+
{
|
7660 |
+
"epoch": 0.07729571466595687,
|
7661 |
+
"grad_norm": 0.42177918553352356,
|
7662 |
+
"learning_rate": 2.1227210107456852e-05,
|
7663 |
+
"loss": 0.9729,
|
7664 |
+
"step": 1089
|
7665 |
+
},
|
7666 |
+
{
|
7667 |
+
"epoch": 0.0773666932836483,
|
7668 |
+
"grad_norm": 0.38119325041770935,
|
7669 |
+
"learning_rate": 2.1085949060360654e-05,
|
7670 |
+
"loss": 0.9433,
|
7671 |
+
"step": 1090
|
7672 |
+
},
|
7673 |
+
{
|
7674 |
+
"epoch": 0.07743767190133972,
|
7675 |
+
"grad_norm": 0.35730451345443726,
|
7676 |
+
"learning_rate": 2.0945104193735644e-05,
|
7677 |
+
"loss": 0.5919,
|
7678 |
+
"step": 1091
|
7679 |
+
},
|
7680 |
+
{
|
7681 |
+
"epoch": 0.07750865051903114,
|
7682 |
+
"grad_norm": 0.4657171964645386,
|
7683 |
+
"learning_rate": 2.0804676250375867e-05,
|
7684 |
+
"loss": 0.8516,
|
7685 |
+
"step": 1092
|
7686 |
+
},
|
7687 |
+
{
|
7688 |
+
"epoch": 0.07757962913672256,
|
7689 |
+
"grad_norm": 0.42287111282348633,
|
7690 |
+
"learning_rate": 2.0664665970876496e-05,
|
7691 |
+
"loss": 0.7939,
|
7692 |
+
"step": 1093
|
7693 |
+
},
|
7694 |
+
{
|
7695 |
+
"epoch": 0.07765060775441399,
|
7696 |
+
"grad_norm": 0.4931599795818329,
|
7697 |
+
"learning_rate": 2.0525074093630036e-05,
|
7698 |
+
"loss": 0.7928,
|
7699 |
+
"step": 1094
|
7700 |
+
},
|
7701 |
+
{
|
7702 |
+
"epoch": 0.0777215863721054,
|
7703 |
+
"grad_norm": 0.4923013746738434,
|
7704 |
+
"learning_rate": 2.0385901354822434e-05,
|
7705 |
+
"loss": 1.3041,
|
7706 |
+
"step": 1095
|
7707 |
+
},
|
7708 |
+
{
|
7709 |
+
"epoch": 0.07779256498979682,
|
7710 |
+
"grad_norm": 0.5996274948120117,
|
7711 |
+
"learning_rate": 2.02471484884291e-05,
|
7712 |
+
"loss": 0.981,
|
7713 |
+
"step": 1096
|
7714 |
+
},
|
7715 |
+
{
|
7716 |
+
"epoch": 0.07786354360748825,
|
7717 |
+
"grad_norm": 0.429426908493042,
|
7718 |
+
"learning_rate": 2.01088162262112e-05,
|
7719 |
+
"loss": 0.7224,
|
7720 |
+
"step": 1097
|
7721 |
+
},
|
7722 |
+
{
|
7723 |
+
"epoch": 0.07793452222517966,
|
7724 |
+
"grad_norm": 0.617928147315979,
|
7725 |
+
"learning_rate": 1.9970905297711606e-05,
|
7726 |
+
"loss": 0.909,
|
7727 |
+
"step": 1098
|
7728 |
+
},
|
7729 |
+
{
|
7730 |
+
"epoch": 0.07800550084287108,
|
7731 |
+
"grad_norm": 0.5632895231246948,
|
7732 |
+
"learning_rate": 1.9833416430251174e-05,
|
7733 |
+
"loss": 0.9735,
|
7734 |
+
"step": 1099
|
7735 |
+
},
|
7736 |
+
{
|
7737 |
+
"epoch": 0.07807647946056251,
|
7738 |
+
"grad_norm": 0.4692310690879822,
|
7739 |
+
"learning_rate": 1.969635034892485e-05,
|
7740 |
+
"loss": 1.0158,
|
7741 |
+
"step": 1100
|
7742 |
+
},
|
7743 |
+
{
|
7744 |
+
"epoch": 0.07814745807825392,
|
7745 |
+
"grad_norm": 0.39456284046173096,
|
7746 |
+
"learning_rate": 1.9559707776597957e-05,
|
7747 |
+
"loss": 0.8835,
|
7748 |
+
"step": 1101
|
7749 |
+
},
|
7750 |
+
{
|
7751 |
+
"epoch": 0.07821843669594535,
|
7752 |
+
"grad_norm": 0.6546867489814758,
|
7753 |
+
"learning_rate": 1.9423489433902186e-05,
|
7754 |
+
"loss": 1.0172,
|
7755 |
+
"step": 1102
|
7756 |
+
},
|
7757 |
+
{
|
7758 |
+
"epoch": 0.07828941531363677,
|
7759 |
+
"grad_norm": 0.778107225894928,
|
7760 |
+
"learning_rate": 1.9287696039232005e-05,
|
7761 |
+
"loss": 1.5264,
|
7762 |
+
"step": 1103
|
7763 |
+
},
|
7764 |
+
{
|
7765 |
+
"epoch": 0.07836039393132818,
|
7766 |
+
"grad_norm": 0.3635626435279846,
|
7767 |
+
"learning_rate": 1.9152328308740707e-05,
|
7768 |
+
"loss": 0.6865,
|
7769 |
+
"step": 1104
|
7770 |
+
},
|
7771 |
+
{
|
7772 |
+
"epoch": 0.0784313725490196,
|
7773 |
+
"grad_norm": 0.6078665256500244,
|
7774 |
+
"learning_rate": 1.901738695633669e-05,
|
7775 |
+
"loss": 1.0725,
|
7776 |
+
"step": 1105
|
7777 |
+
},
|
7778 |
+
{
|
7779 |
+
"epoch": 0.07850235116671103,
|
7780 |
+
"grad_norm": 0.45997220277786255,
|
7781 |
+
"learning_rate": 1.888287269367979e-05,
|
7782 |
+
"loss": 0.7352,
|
7783 |
+
"step": 1106
|
7784 |
+
},
|
7785 |
+
{
|
7786 |
+
"epoch": 0.07857332978440244,
|
7787 |
+
"grad_norm": 0.3929154574871063,
|
7788 |
+
"learning_rate": 1.8748786230177294e-05,
|
7789 |
+
"loss": 0.7464,
|
7790 |
+
"step": 1107
|
7791 |
+
},
|
7792 |
+
{
|
7793 |
+
"epoch": 0.07864430840209387,
|
7794 |
+
"grad_norm": 0.48282280564308167,
|
7795 |
+
"learning_rate": 1.861512827298051e-05,
|
7796 |
+
"loss": 0.7389,
|
7797 |
+
"step": 1108
|
7798 |
+
},
|
7799 |
+
{
|
7800 |
+
"epoch": 0.0787152870197853,
|
7801 |
+
"grad_norm": 0.5808085203170776,
|
7802 |
+
"learning_rate": 1.848189952698074e-05,
|
7803 |
+
"loss": 1.1251,
|
7804 |
+
"step": 1109
|
7805 |
+
},
|
7806 |
+
{
|
7807 |
+
"epoch": 0.0787862656374767,
|
7808 |
+
"grad_norm": 0.404216468334198,
|
7809 |
+
"learning_rate": 1.834910069480571e-05,
|
7810 |
+
"loss": 0.7776,
|
7811 |
+
"step": 1110
|
7812 |
+
},
|
7813 |
+
{
|
7814 |
+
"epoch": 0.07885724425516813,
|
7815 |
+
"grad_norm": 0.3583569824695587,
|
7816 |
+
"learning_rate": 1.8216732476815933e-05,
|
7817 |
+
"loss": 0.7038,
|
7818 |
+
"step": 1111
|
7819 |
+
},
|
7820 |
+
{
|
7821 |
+
"epoch": 0.07892822287285955,
|
7822 |
+
"grad_norm": 0.7691454291343689,
|
7823 |
+
"learning_rate": 1.808479557110081e-05,
|
7824 |
+
"loss": 1.1151,
|
7825 |
+
"step": 1112
|
7826 |
+
},
|
7827 |
+
{
|
7828 |
+
"epoch": 0.07899920149055097,
|
7829 |
+
"grad_norm": 0.42444515228271484,
|
7830 |
+
"learning_rate": 1.7953290673475197e-05,
|
7831 |
+
"loss": 0.8253,
|
7832 |
+
"step": 1113
|
7833 |
+
},
|
7834 |
+
{
|
7835 |
+
"epoch": 0.07907018010824239,
|
7836 |
+
"grad_norm": 0.3662635087966919,
|
7837 |
+
"learning_rate": 1.7822218477475494e-05,
|
7838 |
+
"loss": 0.8391,
|
7839 |
+
"step": 1114
|
7840 |
+
},
|
7841 |
+
{
|
7842 |
+
"epoch": 0.07914115872593382,
|
7843 |
+
"grad_norm": 0.5567343831062317,
|
7844 |
+
"learning_rate": 1.769157967435615e-05,
|
7845 |
+
"loss": 0.9544,
|
7846 |
+
"step": 1115
|
7847 |
+
},
|
7848 |
+
{
|
7849 |
+
"epoch": 0.07921213734362523,
|
7850 |
+
"grad_norm": 0.6112503409385681,
|
7851 |
+
"learning_rate": 1.756137495308594e-05,
|
7852 |
+
"loss": 1.0489,
|
7853 |
+
"step": 1116
|
7854 |
+
},
|
7855 |
+
{
|
7856 |
+
"epoch": 0.07928311596131665,
|
7857 |
+
"grad_norm": 0.4554908275604248,
|
7858 |
+
"learning_rate": 1.7431605000344432e-05,
|
7859 |
+
"loss": 0.8538,
|
7860 |
+
"step": 1117
|
7861 |
+
},
|
7862 |
+
{
|
7863 |
+
"epoch": 0.07935409457900808,
|
7864 |
+
"grad_norm": 0.4239465594291687,
|
7865 |
+
"learning_rate": 1.7302270500518182e-05,
|
7866 |
+
"loss": 0.917,
|
7867 |
+
"step": 1118
|
7868 |
+
},
|
7869 |
+
{
|
7870 |
+
"epoch": 0.07942507319669949,
|
7871 |
+
"grad_norm": 0.4315805435180664,
|
7872 |
+
"learning_rate": 1.717337213569733e-05,
|
7873 |
+
"loss": 0.9517,
|
7874 |
+
"step": 1119
|
7875 |
+
},
|
7876 |
+
{
|
7877 |
+
"epoch": 0.07949605181439091,
|
7878 |
+
"grad_norm": 0.48045405745506287,
|
7879 |
+
"learning_rate": 1.704491058567187e-05,
|
7880 |
+
"loss": 1.022,
|
7881 |
+
"step": 1120
|
7882 |
+
},
|
7883 |
+
{
|
7884 |
+
"epoch": 0.07956703043208234,
|
7885 |
+
"grad_norm": 0.5694467425346375,
|
7886 |
+
"learning_rate": 1.6916886527928066e-05,
|
7887 |
+
"loss": 0.9376,
|
7888 |
+
"step": 1121
|
7889 |
+
},
|
7890 |
+
{
|
7891 |
+
"epoch": 0.07963800904977375,
|
7892 |
+
"grad_norm": 0.6039645075798035,
|
7893 |
+
"learning_rate": 1.6789300637645e-05,
|
7894 |
+
"loss": 1.1573,
|
7895 |
+
"step": 1122
|
7896 |
+
},
|
7897 |
+
{
|
7898 |
+
"epoch": 0.07970898766746518,
|
7899 |
+
"grad_norm": 0.5177348852157593,
|
7900 |
+
"learning_rate": 1.666215358769084e-05,
|
7901 |
+
"loss": 1.0799,
|
7902 |
+
"step": 1123
|
7903 |
+
},
|
7904 |
+
{
|
7905 |
+
"epoch": 0.0797799662851566,
|
7906 |
+
"grad_norm": 0.3842318058013916,
|
7907 |
+
"learning_rate": 1.653544604861945e-05,
|
7908 |
+
"loss": 0.8841,
|
7909 |
+
"step": 1124
|
7910 |
+
},
|
7911 |
+
{
|
7912 |
+
"epoch": 0.07985094490284801,
|
7913 |
+
"grad_norm": 0.57900071144104,
|
7914 |
+
"learning_rate": 1.6409178688666726e-05,
|
7915 |
+
"loss": 1.0456,
|
7916 |
+
"step": 1125
|
7917 |
+
},
|
7918 |
+
{
|
7919 |
+
"epoch": 0.07992192352053944,
|
7920 |
+
"grad_norm": 0.6663162112236023,
|
7921 |
+
"learning_rate": 1.6283352173747145e-05,
|
7922 |
+
"loss": 1.1332,
|
7923 |
+
"step": 1126
|
7924 |
+
},
|
7925 |
+
{
|
7926 |
+
"epoch": 0.07999290213823086,
|
7927 |
+
"grad_norm": 0.5811548233032227,
|
7928 |
+
"learning_rate": 1.615796716745026e-05,
|
7929 |
+
"loss": 0.9705,
|
7930 |
+
"step": 1127
|
7931 |
+
},
|
7932 |
+
{
|
7933 |
+
"epoch": 0.08006388075592227,
|
7934 |
+
"grad_norm": 0.512183666229248,
|
7935 |
+
"learning_rate": 1.6033024331037138e-05,
|
7936 |
+
"loss": 1.1737,
|
7937 |
+
"step": 1128
|
7938 |
+
},
|
7939 |
+
{
|
7940 |
+
"epoch": 0.0801348593736137,
|
7941 |
+
"grad_norm": 0.3928144872188568,
|
7942 |
+
"learning_rate": 1.5908524323436957e-05,
|
7943 |
+
"loss": 0.8263,
|
7944 |
+
"step": 1129
|
7945 |
+
},
|
7946 |
+
{
|
7947 |
+
"epoch": 0.08020583799130512,
|
7948 |
+
"grad_norm": 0.47604188323020935,
|
7949 |
+
"learning_rate": 1.578446780124344e-05,
|
7950 |
+
"loss": 0.9866,
|
7951 |
+
"step": 1130
|
7952 |
+
},
|
7953 |
+
{
|
7954 |
+
"epoch": 0.08027681660899653,
|
7955 |
+
"grad_norm": 0.609736979007721,
|
7956 |
+
"learning_rate": 1.566085541871145e-05,
|
7957 |
+
"loss": 0.9433,
|
7958 |
+
"step": 1131
|
7959 |
+
},
|
7960 |
+
{
|
7961 |
+
"epoch": 0.08034779522668796,
|
7962 |
+
"grad_norm": 0.4574166238307953,
|
7963 |
+
"learning_rate": 1.553768782775351e-05,
|
7964 |
+
"loss": 0.9907,
|
7965 |
+
"step": 1132
|
7966 |
+
},
|
7967 |
+
{
|
7968 |
+
"epoch": 0.08041877384437938,
|
7969 |
+
"grad_norm": 0.4988664388656616,
|
7970 |
+
"learning_rate": 1.5414965677936445e-05,
|
7971 |
+
"loss": 0.7608,
|
7972 |
+
"step": 1133
|
7973 |
+
},
|
7974 |
+
{
|
7975 |
+
"epoch": 0.0804897524620708,
|
7976 |
+
"grad_norm": 0.7940763235092163,
|
7977 |
+
"learning_rate": 1.5292689616477806e-05,
|
7978 |
+
"loss": 1.2404,
|
7979 |
+
"step": 1134
|
7980 |
+
},
|
7981 |
+
{
|
7982 |
+
"epoch": 0.08056073107976222,
|
7983 |
+
"grad_norm": 0.36464962363243103,
|
7984 |
+
"learning_rate": 1.5170860288242638e-05,
|
7985 |
+
"loss": 0.8057,
|
7986 |
+
"step": 1135
|
7987 |
+
},
|
7988 |
+
{
|
7989 |
+
"epoch": 0.08063170969745365,
|
7990 |
+
"grad_norm": 0.7205368876457214,
|
7991 |
+
"learning_rate": 1.5049478335739886e-05,
|
7992 |
+
"loss": 0.9051,
|
7993 |
+
"step": 1136
|
7994 |
+
},
|
7995 |
+
{
|
7996 |
+
"epoch": 0.08070268831514506,
|
7997 |
+
"grad_norm": 0.5451712012290955,
|
7998 |
+
"learning_rate": 1.4928544399119148e-05,
|
7999 |
+
"loss": 1.0423,
|
8000 |
+
"step": 1137
|
8001 |
+
},
|
8002 |
+
{
|
8003 |
+
"epoch": 0.08077366693283648,
|
8004 |
+
"grad_norm": 0.8497772812843323,
|
8005 |
+
"learning_rate": 1.4808059116167305e-05,
|
8006 |
+
"loss": 1.2004,
|
8007 |
+
"step": 1138
|
8008 |
+
},
|
8009 |
+
{
|
8010 |
+
"epoch": 0.08084464555052791,
|
8011 |
+
"grad_norm": 0.4573586881160736,
|
8012 |
+
"learning_rate": 1.4688023122305005e-05,
|
8013 |
+
"loss": 0.8203,
|
8014 |
+
"step": 1139
|
8015 |
+
},
|
8016 |
+
{
|
8017 |
+
"epoch": 0.08091562416821932,
|
8018 |
+
"grad_norm": 0.5086919665336609,
|
8019 |
+
"learning_rate": 1.4568437050583517e-05,
|
8020 |
+
"loss": 1.0065,
|
8021 |
+
"step": 1140
|
8022 |
+
},
|
8023 |
+
{
|
8024 |
+
"epoch": 0.08098660278591074,
|
8025 |
+
"grad_norm": 0.5985661745071411,
|
8026 |
+
"learning_rate": 1.4449301531681226e-05,
|
8027 |
+
"loss": 1.0,
|
8028 |
+
"step": 1141
|
8029 |
+
},
|
8030 |
+
{
|
8031 |
+
"epoch": 0.08105758140360217,
|
8032 |
+
"grad_norm": 0.5257460474967957,
|
8033 |
+
"learning_rate": 1.4330617193900364e-05,
|
8034 |
+
"loss": 1.0537,
|
8035 |
+
"step": 1142
|
8036 |
+
},
|
8037 |
+
{
|
8038 |
+
"epoch": 0.08112856002129358,
|
8039 |
+
"grad_norm": 0.5764384865760803,
|
8040 |
+
"learning_rate": 1.4212384663163791e-05,
|
8041 |
+
"loss": 1.2133,
|
8042 |
+
"step": 1143
|
8043 |
+
},
|
8044 |
+
{
|
8045 |
+
"epoch": 0.081199538638985,
|
8046 |
+
"grad_norm": 0.36415451765060425,
|
8047 |
+
"learning_rate": 1.4094604563011472e-05,
|
8048 |
+
"loss": 0.7653,
|
8049 |
+
"step": 1144
|
8050 |
+
},
|
8051 |
+
{
|
8052 |
+
"epoch": 0.08127051725667643,
|
8053 |
+
"grad_norm": 0.3863047957420349,
|
8054 |
+
"learning_rate": 1.3977277514597464e-05,
|
8055 |
+
"loss": 0.7643,
|
8056 |
+
"step": 1145
|
8057 |
+
},
|
8058 |
+
{
|
8059 |
+
"epoch": 0.08134149587436784,
|
8060 |
+
"grad_norm": 0.4379030168056488,
|
8061 |
+
"learning_rate": 1.3860404136686411e-05,
|
8062 |
+
"loss": 0.6285,
|
8063 |
+
"step": 1146
|
8064 |
+
},
|
8065 |
+
{
|
8066 |
+
"epoch": 0.08141247449205927,
|
8067 |
+
"grad_norm": 0.4873761534690857,
|
8068 |
+
"learning_rate": 1.3743985045650364e-05,
|
8069 |
+
"loss": 0.688,
|
8070 |
+
"step": 1147
|
8071 |
+
},
|
8072 |
+
{
|
8073 |
+
"epoch": 0.08148345310975069,
|
8074 |
+
"grad_norm": 0.8107748031616211,
|
8075 |
+
"learning_rate": 1.3628020855465572e-05,
|
8076 |
+
"loss": 1.2016,
|
8077 |
+
"step": 1148
|
8078 |
+
},
|
8079 |
+
{
|
8080 |
+
"epoch": 0.0815544317274421,
|
8081 |
+
"grad_norm": 0.6649096012115479,
|
8082 |
+
"learning_rate": 1.3512512177709224e-05,
|
8083 |
+
"loss": 1.1479,
|
8084 |
+
"step": 1149
|
8085 |
+
},
|
8086 |
+
{
|
8087 |
+
"epoch": 0.08162541034513353,
|
8088 |
+
"grad_norm": 0.41352081298828125,
|
8089 |
+
"learning_rate": 1.339745962155613e-05,
|
8090 |
+
"loss": 0.8386,
|
8091 |
+
"step": 1150
|
8092 |
+
},
|
8093 |
+
{
|
8094 |
+
"epoch": 0.08169638896282495,
|
8095 |
+
"grad_norm": 0.4705093502998352,
|
8096 |
+
"learning_rate": 1.3282863793775702e-05,
|
8097 |
+
"loss": 1.1097,
|
8098 |
+
"step": 1151
|
8099 |
+
},
|
8100 |
+
{
|
8101 |
+
"epoch": 0.08176736758051636,
|
8102 |
+
"grad_norm": 0.4707050919532776,
|
8103 |
+
"learning_rate": 1.3168725298728524e-05,
|
8104 |
+
"loss": 0.857,
|
8105 |
+
"step": 1152
|
8106 |
+
},
|
8107 |
+
{
|
8108 |
+
"epoch": 0.08183834619820779,
|
8109 |
+
"grad_norm": 0.4865884482860565,
|
8110 |
+
"learning_rate": 1.305504473836331e-05,
|
8111 |
+
"loss": 1.1221,
|
8112 |
+
"step": 1153
|
8113 |
+
},
|
8114 |
+
{
|
8115 |
+
"epoch": 0.08190932481589921,
|
8116 |
+
"grad_norm": 0.39382314682006836,
|
8117 |
+
"learning_rate": 1.294182271221377e-05,
|
8118 |
+
"loss": 0.5837,
|
8119 |
+
"step": 1154
|
8120 |
+
},
|
8121 |
+
{
|
8122 |
+
"epoch": 0.08198030343359063,
|
8123 |
+
"grad_norm": 0.5165771245956421,
|
8124 |
+
"learning_rate": 1.2829059817395261e-05,
|
8125 |
+
"loss": 1.2069,
|
8126 |
+
"step": 1155
|
8127 |
+
},
|
8128 |
+
{
|
8129 |
+
"epoch": 0.08205128205128205,
|
8130 |
+
"grad_norm": 0.4368021786212921,
|
8131 |
+
"learning_rate": 1.2716756648601857e-05,
|
8132 |
+
"loss": 0.8342,
|
8133 |
+
"step": 1156
|
8134 |
+
},
|
8135 |
+
{
|
8136 |
+
"epoch": 0.08212226066897348,
|
8137 |
+
"grad_norm": 0.5267868638038635,
|
8138 |
+
"learning_rate": 1.2604913798103034e-05,
|
8139 |
+
"loss": 0.9511,
|
8140 |
+
"step": 1157
|
8141 |
+
},
|
8142 |
+
{
|
8143 |
+
"epoch": 0.08219323928666489,
|
8144 |
+
"grad_norm": 0.43068644404411316,
|
8145 |
+
"learning_rate": 1.2493531855740625e-05,
|
8146 |
+
"loss": 1.0738,
|
8147 |
+
"step": 1158
|
8148 |
+
},
|
8149 |
+
{
|
8150 |
+
"epoch": 0.08226421790435631,
|
8151 |
+
"grad_norm": 0.43189457058906555,
|
8152 |
+
"learning_rate": 1.2382611408925793e-05,
|
8153 |
+
"loss": 0.7604,
|
8154 |
+
"step": 1159
|
8155 |
+
},
|
8156 |
+
{
|
8157 |
+
"epoch": 0.08233519652204774,
|
8158 |
+
"grad_norm": 0.5125055313110352,
|
8159 |
+
"learning_rate": 1.2272153042635704e-05,
|
8160 |
+
"loss": 0.7189,
|
8161 |
+
"step": 1160
|
8162 |
+
},
|
8163 |
+
{
|
8164 |
+
"epoch": 0.08240617513973915,
|
8165 |
+
"grad_norm": 0.47541847825050354,
|
8166 |
+
"learning_rate": 1.2162157339410718e-05,
|
8167 |
+
"loss": 0.9183,
|
8168 |
+
"step": 1161
|
8169 |
+
},
|
8170 |
+
{
|
8171 |
+
"epoch": 0.08247715375743057,
|
8172 |
+
"grad_norm": 0.5350788235664368,
|
8173 |
+
"learning_rate": 1.2052624879351104e-05,
|
8174 |
+
"loss": 1.1197,
|
8175 |
+
"step": 1162
|
8176 |
+
},
|
8177 |
+
{
|
8178 |
+
"epoch": 0.082548132375122,
|
8179 |
+
"grad_norm": 0.4992624521255493,
|
8180 |
+
"learning_rate": 1.1943556240114062e-05,
|
8181 |
+
"loss": 0.9298,
|
8182 |
+
"step": 1163
|
8183 |
+
},
|
8184 |
+
{
|
8185 |
+
"epoch": 0.08261911099281341,
|
8186 |
+
"grad_norm": 0.44900640845298767,
|
8187 |
+
"learning_rate": 1.183495199691068e-05,
|
8188 |
+
"loss": 0.9439,
|
8189 |
+
"step": 1164
|
8190 |
+
},
|
8191 |
+
{
|
8192 |
+
"epoch": 0.08269008961050484,
|
8193 |
+
"grad_norm": 0.4040171205997467,
|
8194 |
+
"learning_rate": 1.1726812722502945e-05,
|
8195 |
+
"loss": 0.9444,
|
8196 |
+
"step": 1165
|
8197 |
+
},
|
8198 |
+
{
|
8199 |
+
"epoch": 0.08276106822819626,
|
8200 |
+
"grad_norm": 0.3812856078147888,
|
8201 |
+
"learning_rate": 1.1619138987200562e-05,
|
8202 |
+
"loss": 0.9681,
|
8203 |
+
"step": 1166
|
8204 |
+
},
|
8205 |
+
{
|
8206 |
+
"epoch": 0.08283204684588767,
|
8207 |
+
"grad_norm": 0.4886763095855713,
|
8208 |
+
"learning_rate": 1.1511931358858185e-05,
|
8209 |
+
"loss": 0.9581,
|
8210 |
+
"step": 1167
|
8211 |
+
},
|
8212 |
+
{
|
8213 |
+
"epoch": 0.0829030254635791,
|
8214 |
+
"grad_norm": 0.5425056219100952,
|
8215 |
+
"learning_rate": 1.1405190402872202e-05,
|
8216 |
+
"loss": 0.7195,
|
8217 |
+
"step": 1168
|
8218 |
+
},
|
8219 |
+
{
|
8220 |
+
"epoch": 0.08297400408127052,
|
8221 |
+
"grad_norm": 0.48632103204727173,
|
8222 |
+
"learning_rate": 1.129891668217783e-05,
|
8223 |
+
"loss": 1.0468,
|
8224 |
+
"step": 1169
|
8225 |
+
},
|
8226 |
+
{
|
8227 |
+
"epoch": 0.08304498269896193,
|
8228 |
+
"grad_norm": 0.4761428236961365,
|
8229 |
+
"learning_rate": 1.119311075724625e-05,
|
8230 |
+
"loss": 1.111,
|
8231 |
+
"step": 1170
|
8232 |
+
},
|
8233 |
+
{
|
8234 |
+
"epoch": 0.08311596131665336,
|
8235 |
+
"grad_norm": 0.5897856950759888,
|
8236 |
+
"learning_rate": 1.1087773186081473e-05,
|
8237 |
+
"loss": 1.1972,
|
8238 |
+
"step": 1171
|
8239 |
+
},
|
8240 |
+
{
|
8241 |
+
"epoch": 0.08318693993434478,
|
8242 |
+
"grad_norm": 0.4703165590763092,
|
8243 |
+
"learning_rate": 1.0982904524217551e-05,
|
8244 |
+
"loss": 0.9719,
|
8245 |
+
"step": 1172
|
8246 |
+
},
|
8247 |
+
{
|
8248 |
+
"epoch": 0.0832579185520362,
|
8249 |
+
"grad_norm": 0.41847726702690125,
|
8250 |
+
"learning_rate": 1.0878505324715538e-05,
|
8251 |
+
"loss": 0.6459,
|
8252 |
+
"step": 1173
|
8253 |
+
},
|
8254 |
+
{
|
8255 |
+
"epoch": 0.08332889716972762,
|
8256 |
+
"grad_norm": 0.5312420129776001,
|
8257 |
+
"learning_rate": 1.0774576138160597e-05,
|
8258 |
+
"loss": 0.9006,
|
8259 |
+
"step": 1174
|
8260 |
+
},
|
8261 |
+
{
|
8262 |
+
"epoch": 0.08339987578741904,
|
8263 |
+
"grad_norm": 0.49165722727775574,
|
8264 |
+
"learning_rate": 1.067111751265919e-05,
|
8265 |
+
"loss": 0.7813,
|
8266 |
+
"step": 1175
|
8267 |
+
},
|
8268 |
+
{
|
8269 |
+
"epoch": 0.08347085440511046,
|
8270 |
+
"grad_norm": 0.4631386399269104,
|
8271 |
+
"learning_rate": 1.056812999383604e-05,
|
8272 |
+
"loss": 0.8755,
|
8273 |
+
"step": 1176
|
8274 |
+
},
|
8275 |
+
{
|
8276 |
+
"epoch": 0.08354183302280188,
|
8277 |
+
"grad_norm": 0.4755702316761017,
|
8278 |
+
"learning_rate": 1.0465614124831379e-05,
|
8279 |
+
"loss": 0.8196,
|
8280 |
+
"step": 1177
|
8281 |
+
},
|
8282 |
+
{
|
8283 |
+
"epoch": 0.0836128116404933,
|
8284 |
+
"grad_norm": 0.49447572231292725,
|
8285 |
+
"learning_rate": 1.0363570446297999e-05,
|
8286 |
+
"loss": 1.1008,
|
8287 |
+
"step": 1178
|
8288 |
+
},
|
8289 |
+
{
|
8290 |
+
"epoch": 0.08368379025818472,
|
8291 |
+
"grad_norm": 0.5133283138275146,
|
8292 |
+
"learning_rate": 1.0261999496398434e-05,
|
8293 |
+
"loss": 1.117,
|
8294 |
+
"step": 1179
|
8295 |
+
},
|
8296 |
+
{
|
8297 |
+
"epoch": 0.08375476887587614,
|
8298 |
+
"grad_norm": 1.6076154708862305,
|
8299 |
+
"learning_rate": 1.0160901810802115e-05,
|
8300 |
+
"loss": 1.458,
|
8301 |
+
"step": 1180
|
8302 |
+
},
|
8303 |
+
{
|
8304 |
+
"epoch": 0.08382574749356757,
|
8305 |
+
"grad_norm": 0.39624008536338806,
|
8306 |
+
"learning_rate": 1.0060277922682605e-05,
|
8307 |
+
"loss": 0.6493,
|
8308 |
+
"step": 1181
|
8309 |
+
},
|
8310 |
+
{
|
8311 |
+
"epoch": 0.08389672611125898,
|
8312 |
+
"grad_norm": 0.565704882144928,
|
8313 |
+
"learning_rate": 9.960128362714637e-06,
|
8314 |
+
"loss": 0.941,
|
8315 |
+
"step": 1182
|
8316 |
+
},
|
8317 |
+
{
|
8318 |
+
"epoch": 0.0839677047289504,
|
8319 |
+
"grad_norm": 0.39284053444862366,
|
8320 |
+
"learning_rate": 9.860453659071533e-06,
|
8321 |
+
"loss": 0.5983,
|
8322 |
+
"step": 1183
|
8323 |
+
},
|
8324 |
+
{
|
8325 |
+
"epoch": 0.08403868334664183,
|
8326 |
+
"grad_norm": 0.67037433385849,
|
8327 |
+
"learning_rate": 9.761254337422176e-06,
|
8328 |
+
"loss": 1.0344,
|
8329 |
+
"step": 1184
|
8330 |
+
},
|
8331 |
+
{
|
8332 |
+
"epoch": 0.08410966196433324,
|
8333 |
+
"grad_norm": 0.5679614543914795,
|
8334 |
+
"learning_rate": 9.662530920928403e-06,
|
8335 |
+
"loss": 0.9806,
|
8336 |
+
"step": 1185
|
8337 |
+
},
|
8338 |
+
{
|
8339 |
+
"epoch": 0.08418064058202467,
|
8340 |
+
"grad_norm": 0.4572247564792633,
|
8341 |
+
"learning_rate": 9.564283930242257e-06,
|
8342 |
+
"loss": 0.9108,
|
8343 |
+
"step": 1186
|
8344 |
+
},
|
8345 |
+
{
|
8346 |
+
"epoch": 0.08425161919971609,
|
8347 |
+
"grad_norm": 0.6498078107833862,
|
8348 |
+
"learning_rate": 9.466513883503091e-06,
|
8349 |
+
"loss": 1.0266,
|
8350 |
+
"step": 1187
|
8351 |
+
},
|
8352 |
+
{
|
8353 |
+
"epoch": 0.0843225978174075,
|
8354 |
+
"grad_norm": 0.4459131956100464,
|
8355 |
+
"learning_rate": 9.369221296335006e-06,
|
8356 |
+
"loss": 0.7574,
|
8357 |
+
"step": 1188
|
8358 |
+
},
|
8359 |
+
{
|
8360 |
+
"epoch": 0.08439357643509893,
|
8361 |
+
"grad_norm": 0.5533161759376526,
|
8362 |
+
"learning_rate": 9.272406681844015e-06,
|
8363 |
+
"loss": 1.1075,
|
8364 |
+
"step": 1189
|
8365 |
+
},
|
8366 |
+
{
|
8367 |
+
"epoch": 0.08446455505279035,
|
8368 |
+
"grad_norm": 0.6843585968017578,
|
8369 |
+
"learning_rate": 9.176070550615378e-06,
|
8370 |
+
"loss": 1.0174,
|
8371 |
+
"step": 1190
|
8372 |
+
},
|
8373 |
+
{
|
8374 |
+
"epoch": 0.08453553367048176,
|
8375 |
+
"grad_norm": 0.6016431450843811,
|
8376 |
+
"learning_rate": 9.080213410710913e-06,
|
8377 |
+
"loss": 0.8994,
|
8378 |
+
"step": 1191
|
8379 |
+
},
|
8380 |
+
{
|
8381 |
+
"epoch": 0.08460651228817319,
|
8382 |
+
"grad_norm": 0.4497312605381012,
|
8383 |
+
"learning_rate": 8.98483576766631e-06,
|
8384 |
+
"loss": 0.9091,
|
8385 |
+
"step": 1192
|
8386 |
+
},
|
8387 |
+
{
|
8388 |
+
"epoch": 0.08467749090586461,
|
8389 |
+
"grad_norm": 0.4260040819644928,
|
8390 |
+
"learning_rate": 8.889938124488517e-06,
|
8391 |
+
"loss": 0.686,
|
8392 |
+
"step": 1193
|
8393 |
+
},
|
8394 |
+
{
|
8395 |
+
"epoch": 0.08474846952355602,
|
8396 |
+
"grad_norm": 0.48108816146850586,
|
8397 |
+
"learning_rate": 8.795520981652961e-06,
|
8398 |
+
"loss": 0.9346,
|
8399 |
+
"step": 1194
|
8400 |
+
},
|
8401 |
+
{
|
8402 |
+
"epoch": 0.08481944814124745,
|
8403 |
+
"grad_norm": 0.45303186774253845,
|
8404 |
+
"learning_rate": 8.701584837101018e-06,
|
8405 |
+
"loss": 0.816,
|
8406 |
+
"step": 1195
|
8407 |
+
},
|
8408 |
+
{
|
8409 |
+
"epoch": 0.08489042675893888,
|
8410 |
+
"grad_norm": 0.5208687782287598,
|
8411 |
+
"learning_rate": 8.608130186237329e-06,
|
8412 |
+
"loss": 0.7334,
|
8413 |
+
"step": 1196
|
8414 |
+
},
|
8415 |
+
{
|
8416 |
+
"epoch": 0.08496140537663029,
|
8417 |
+
"grad_norm": 0.7175068855285645,
|
8418 |
+
"learning_rate": 8.515157521927286e-06,
|
8419 |
+
"loss": 1.2355,
|
8420 |
+
"step": 1197
|
8421 |
+
},
|
8422 |
+
{
|
8423 |
+
"epoch": 0.08503238399432171,
|
8424 |
+
"grad_norm": 0.5213837027549744,
|
8425 |
+
"learning_rate": 8.422667334494249e-06,
|
8426 |
+
"loss": 1.0409,
|
8427 |
+
"step": 1198
|
8428 |
+
},
|
8429 |
+
{
|
8430 |
+
"epoch": 0.08510336261201314,
|
8431 |
+
"grad_norm": 0.4475990831851959,
|
8432 |
+
"learning_rate": 8.330660111717192e-06,
|
8433 |
+
"loss": 1.0196,
|
8434 |
+
"step": 1199
|
8435 |
+
},
|
8436 |
+
{
|
8437 |
+
"epoch": 0.08517434122970455,
|
8438 |
+
"grad_norm": 0.523715615272522,
|
8439 |
+
"learning_rate": 8.239136338827903e-06,
|
8440 |
+
"loss": 0.8783,
|
8441 |
+
"step": 1200
|
8442 |
+
},
|
8443 |
+
{
|
8444 |
+
"epoch": 0.08524531984739597,
|
8445 |
+
"grad_norm": 1.0819703340530396,
|
8446 |
+
"learning_rate": 8.148096498508573e-06,
|
8447 |
+
"loss": 1.6188,
|
8448 |
+
"step": 1201
|
8449 |
+
},
|
8450 |
+
{
|
8451 |
+
"epoch": 0.0853162984650874,
|
8452 |
+
"grad_norm": 0.3292709290981293,
|
8453 |
+
"learning_rate": 8.05754107088923e-06,
|
8454 |
+
"loss": 0.5635,
|
8455 |
+
"step": 1202
|
8456 |
+
},
|
8457 |
+
{
|
8458 |
+
"epoch": 0.08538727708277881,
|
8459 |
+
"grad_norm": 0.4222264289855957,
|
8460 |
+
"learning_rate": 7.96747053354513e-06,
|
8461 |
+
"loss": 0.8573,
|
8462 |
+
"step": 1203
|
8463 |
+
},
|
8464 |
+
{
|
8465 |
+
"epoch": 0.08545825570047023,
|
8466 |
+
"grad_norm": 0.4970225691795349,
|
8467 |
+
"learning_rate": 7.877885361494353e-06,
|
8468 |
+
"loss": 0.7876,
|
8469 |
+
"step": 1204
|
8470 |
+
},
|
8471 |
+
{
|
8472 |
+
"epoch": 0.08552923431816166,
|
8473 |
+
"grad_norm": 0.5125081539154053,
|
8474 |
+
"learning_rate": 7.788786027195195e-06,
|
8475 |
+
"loss": 0.8243,
|
8476 |
+
"step": 1205
|
8477 |
+
},
|
8478 |
+
{
|
8479 |
+
"epoch": 0.08560021293585307,
|
8480 |
+
"grad_norm": 0.5454389452934265,
|
8481 |
+
"learning_rate": 7.700173000543742e-06,
|
8482 |
+
"loss": 0.8166,
|
8483 |
+
"step": 1206
|
8484 |
+
},
|
8485 |
+
{
|
8486 |
+
"epoch": 0.0856711915535445,
|
8487 |
+
"grad_norm": 0.4455251395702362,
|
8488 |
+
"learning_rate": 7.612046748871327e-06,
|
8489 |
+
"loss": 0.7249,
|
8490 |
+
"step": 1207
|
8491 |
+
},
|
8492 |
+
{
|
8493 |
+
"epoch": 0.08574217017123592,
|
8494 |
+
"grad_norm": 0.461681604385376,
|
8495 |
+
"learning_rate": 7.524407736942174e-06,
|
8496 |
+
"loss": 0.8663,
|
8497 |
+
"step": 1208
|
8498 |
+
},
|
8499 |
+
{
|
8500 |
+
"epoch": 0.08581314878892733,
|
8501 |
+
"grad_norm": 0.429550439119339,
|
8502 |
+
"learning_rate": 7.4372564269508445e-06,
|
8503 |
+
"loss": 0.9401,
|
8504 |
+
"step": 1209
|
8505 |
+
},
|
8506 |
+
{
|
8507 |
+
"epoch": 0.08588412740661876,
|
8508 |
+
"grad_norm": 0.44344857335090637,
|
8509 |
+
"learning_rate": 7.350593278519824e-06,
|
8510 |
+
"loss": 0.9269,
|
8511 |
+
"step": 1210
|
8512 |
+
},
|
8513 |
+
{
|
8514 |
+
"epoch": 0.08595510602431018,
|
8515 |
+
"grad_norm": 0.45307299494743347,
|
8516 |
+
"learning_rate": 7.264418748697144e-06,
|
8517 |
+
"loss": 0.854,
|
8518 |
+
"step": 1211
|
8519 |
+
},
|
8520 |
+
{
|
8521 |
+
"epoch": 0.0860260846420016,
|
8522 |
+
"grad_norm": 0.5577502250671387,
|
8523 |
+
"learning_rate": 7.178733291953865e-06,
|
8524 |
+
"loss": 0.8163,
|
8525 |
+
"step": 1212
|
8526 |
+
},
|
8527 |
+
{
|
8528 |
+
"epoch": 0.08609706325969302,
|
8529 |
+
"grad_norm": 0.3416275084018707,
|
8530 |
+
"learning_rate": 7.093537360181868e-06,
|
8531 |
+
"loss": 0.8937,
|
8532 |
+
"step": 1213
|
8533 |
+
},
|
8534 |
+
{
|
8535 |
+
"epoch": 0.08616804187738444,
|
8536 |
+
"grad_norm": 0.3705112338066101,
|
8537 |
+
"learning_rate": 7.00883140269123e-06,
|
8538 |
+
"loss": 0.7082,
|
8539 |
+
"step": 1214
|
8540 |
+
},
|
8541 |
+
{
|
8542 |
+
"epoch": 0.08623902049507585,
|
8543 |
+
"grad_norm": 0.5228879451751709,
|
8544 |
+
"learning_rate": 6.924615866208095e-06,
|
8545 |
+
"loss": 1.0063,
|
8546 |
+
"step": 1215
|
8547 |
+
},
|
8548 |
+
{
|
8549 |
+
"epoch": 0.08630999911276728,
|
8550 |
+
"grad_norm": 0.4439268708229065,
|
8551 |
+
"learning_rate": 6.840891194872112e-06,
|
8552 |
+
"loss": 0.9436,
|
8553 |
+
"step": 1216
|
8554 |
+
},
|
8555 |
+
{
|
8556 |
+
"epoch": 0.0863809777304587,
|
8557 |
+
"grad_norm": 0.5648981928825378,
|
8558 |
+
"learning_rate": 6.75765783023421e-06,
|
8559 |
+
"loss": 0.7849,
|
8560 |
+
"step": 1217
|
8561 |
+
},
|
8562 |
+
{
|
8563 |
+
"epoch": 0.08645195634815012,
|
8564 |
+
"grad_norm": 0.5214237570762634,
|
8565 |
+
"learning_rate": 6.674916211254289e-06,
|
8566 |
+
"loss": 0.8858,
|
8567 |
+
"step": 1218
|
8568 |
+
},
|
8569 |
+
{
|
8570 |
+
"epoch": 0.08652293496584154,
|
8571 |
+
"grad_norm": 0.5296366810798645,
|
8572 |
+
"learning_rate": 6.592666774298784e-06,
|
8573 |
+
"loss": 0.7019,
|
8574 |
+
"step": 1219
|
8575 |
+
},
|
8576 |
+
{
|
8577 |
+
"epoch": 0.08659391358353297,
|
8578 |
+
"grad_norm": 0.43214577436447144,
|
8579 |
+
"learning_rate": 6.510909953138511e-06,
|
8580 |
+
"loss": 0.6289,
|
8581 |
+
"step": 1220
|
8582 |
+
},
|
8583 |
+
{
|
8584 |
+
"epoch": 0.08666489220122438,
|
8585 |
+
"grad_norm": 0.4352208375930786,
|
8586 |
+
"learning_rate": 6.429646178946247e-06,
|
8587 |
+
"loss": 0.9008,
|
8588 |
+
"step": 1221
|
8589 |
+
},
|
8590 |
+
{
|
8591 |
+
"epoch": 0.0867358708189158,
|
8592 |
+
"grad_norm": 0.36285680532455444,
|
8593 |
+
"learning_rate": 6.3488758802945354e-06,
|
8594 |
+
"loss": 0.8594,
|
8595 |
+
"step": 1222
|
8596 |
+
},
|
8597 |
+
{
|
8598 |
+
"epoch": 0.08680684943660723,
|
8599 |
+
"grad_norm": 0.728671133518219,
|
8600 |
+
"learning_rate": 6.268599483153403e-06,
|
8601 |
+
"loss": 1.052,
|
8602 |
+
"step": 1223
|
8603 |
+
},
|
8604 |
+
{
|
8605 |
+
"epoch": 0.08687782805429864,
|
8606 |
+
"grad_norm": 0.5141593813896179,
|
8607 |
+
"learning_rate": 6.188817410888148e-06,
|
8608 |
+
"loss": 0.9049,
|
8609 |
+
"step": 1224
|
8610 |
+
},
|
8611 |
+
{
|
8612 |
+
"epoch": 0.08694880667199006,
|
8613 |
+
"grad_norm": 0.46414846181869507,
|
8614 |
+
"learning_rate": 6.109530084257042e-06,
|
8615 |
+
"loss": 1.0772,
|
8616 |
+
"step": 1225
|
8617 |
+
},
|
8618 |
+
{
|
8619 |
+
"epoch": 0.08701978528968149,
|
8620 |
+
"grad_norm": 0.46595925092697144,
|
8621 |
+
"learning_rate": 6.030737921409169e-06,
|
8622 |
+
"loss": 0.9155,
|
8623 |
+
"step": 1226
|
8624 |
+
},
|
8625 |
+
{
|
8626 |
+
"epoch": 0.0870907639073729,
|
8627 |
+
"grad_norm": 0.4655408263206482,
|
8628 |
+
"learning_rate": 5.952441337882153e-06,
|
8629 |
+
"loss": 0.9474,
|
8630 |
+
"step": 1227
|
8631 |
+
},
|
8632 |
+
{
|
8633 |
+
"epoch": 0.08716174252506433,
|
8634 |
+
"grad_norm": 0.4467454254627228,
|
8635 |
+
"learning_rate": 5.8746407466000464e-06,
|
8636 |
+
"loss": 0.7343,
|
8637 |
+
"step": 1228
|
8638 |
+
},
|
8639 |
+
{
|
8640 |
+
"epoch": 0.08723272114275575,
|
8641 |
+
"grad_norm": 0.49474817514419556,
|
8642 |
+
"learning_rate": 5.797336557871125e-06,
|
8643 |
+
"loss": 1.0227,
|
8644 |
+
"step": 1229
|
8645 |
+
},
|
8646 |
+
{
|
8647 |
+
"epoch": 0.08730369976044716,
|
8648 |
+
"grad_norm": 0.43683508038520813,
|
8649 |
+
"learning_rate": 5.720529179385659e-06,
|
8650 |
+
"loss": 0.7119,
|
8651 |
+
"step": 1230
|
8652 |
+
},
|
8653 |
+
{
|
8654 |
+
"epoch": 0.08737467837813859,
|
8655 |
+
"grad_norm": 0.47792723774909973,
|
8656 |
+
"learning_rate": 5.644219016213903e-06,
|
8657 |
+
"loss": 0.8374,
|
8658 |
+
"step": 1231
|
8659 |
+
},
|
8660 |
+
{
|
8661 |
+
"epoch": 0.08744565699583001,
|
8662 |
+
"grad_norm": 0.5540241599082947,
|
8663 |
+
"learning_rate": 5.568406470803799e-06,
|
8664 |
+
"loss": 0.9396,
|
8665 |
+
"step": 1232
|
8666 |
+
},
|
8667 |
+
{
|
8668 |
+
"epoch": 0.08751663561352142,
|
8669 |
+
"grad_norm": 0.4485909342765808,
|
8670 |
+
"learning_rate": 5.493091942978956e-06,
|
8671 |
+
"loss": 0.8863,
|
8672 |
+
"step": 1233
|
8673 |
+
},
|
8674 |
+
{
|
8675 |
+
"epoch": 0.08758761423121285,
|
8676 |
+
"grad_norm": 0.6412665843963623,
|
8677 |
+
"learning_rate": 5.418275829936537e-06,
|
8678 |
+
"loss": 0.9413,
|
8679 |
+
"step": 1234
|
8680 |
+
},
|
8681 |
+
{
|
8682 |
+
"epoch": 0.08765859284890427,
|
8683 |
+
"grad_norm": 0.46158745884895325,
|
8684 |
+
"learning_rate": 5.343958526245096e-06,
|
8685 |
+
"loss": 0.9195,
|
8686 |
+
"step": 1235
|
8687 |
+
},
|
8688 |
+
{
|
8689 |
+
"epoch": 0.08772957146659568,
|
8690 |
+
"grad_norm": 0.4636899530887604,
|
8691 |
+
"learning_rate": 5.270140423842607e-06,
|
8692 |
+
"loss": 0.7886,
|
8693 |
+
"step": 1236
|
8694 |
+
},
|
8695 |
+
{
|
8696 |
+
"epoch": 0.08780055008428711,
|
8697 |
+
"grad_norm": 0.6638062000274658,
|
8698 |
+
"learning_rate": 5.1968219120342775e-06,
|
8699 |
+
"loss": 1.1065,
|
8700 |
+
"step": 1237
|
8701 |
+
},
|
8702 |
+
{
|
8703 |
+
"epoch": 0.08787152870197854,
|
8704 |
+
"grad_norm": 0.4253000319004059,
|
8705 |
+
"learning_rate": 5.124003377490582e-06,
|
8706 |
+
"loss": 1.1271,
|
8707 |
+
"step": 1238
|
8708 |
+
},
|
8709 |
+
{
|
8710 |
+
"epoch": 0.08794250731966995,
|
8711 |
+
"grad_norm": 0.5168247222900391,
|
8712 |
+
"learning_rate": 5.0516852042451605e-06,
|
8713 |
+
"loss": 0.7275,
|
8714 |
+
"step": 1239
|
8715 |
+
},
|
8716 |
+
{
|
8717 |
+
"epoch": 0.08801348593736137,
|
8718 |
+
"grad_norm": 0.5668867230415344,
|
8719 |
+
"learning_rate": 4.979867773692881e-06,
|
8720 |
+
"loss": 0.8994,
|
8721 |
+
"step": 1240
|
8722 |
+
},
|
8723 |
+
{
|
8724 |
+
"epoch": 0.0880844645550528,
|
8725 |
+
"grad_norm": 0.5159637331962585,
|
8726 |
+
"learning_rate": 4.908551464587707e-06,
|
8727 |
+
"loss": 1.2211,
|
8728 |
+
"step": 1241
|
8729 |
+
},
|
8730 |
+
{
|
8731 |
+
"epoch": 0.08815544317274421,
|
8732 |
+
"grad_norm": 0.9431582689285278,
|
8733 |
+
"learning_rate": 4.8377366530408254e-06,
|
8734 |
+
"loss": 1.2378,
|
8735 |
+
"step": 1242
|
8736 |
+
},
|
8737 |
+
{
|
8738 |
+
"epoch": 0.08822642179043563,
|
8739 |
+
"grad_norm": 0.423022985458374,
|
8740 |
+
"learning_rate": 4.767423712518559e-06,
|
8741 |
+
"loss": 0.8821,
|
8742 |
+
"step": 1243
|
8743 |
+
},
|
8744 |
+
{
|
8745 |
+
"epoch": 0.08829740040812706,
|
8746 |
+
"grad_norm": 0.5624905824661255,
|
8747 |
+
"learning_rate": 4.697613013840441e-06,
|
8748 |
+
"loss": 0.7666,
|
8749 |
+
"step": 1244
|
8750 |
+
},
|
8751 |
+
{
|
8752 |
+
"epoch": 0.08836837902581847,
|
8753 |
+
"grad_norm": 0.7056260704994202,
|
8754 |
+
"learning_rate": 4.628304925177318e-06,
|
8755 |
+
"loss": 1.6976,
|
8756 |
+
"step": 1245
|
8757 |
+
},
|
8758 |
+
{
|
8759 |
+
"epoch": 0.0884393576435099,
|
8760 |
+
"grad_norm": 0.6922840476036072,
|
8761 |
+
"learning_rate": 4.559499812049251e-06,
|
8762 |
+
"loss": 1.151,
|
8763 |
+
"step": 1246
|
8764 |
+
},
|
8765 |
+
{
|
8766 |
+
"epoch": 0.08851033626120132,
|
8767 |
+
"grad_norm": 0.5164980888366699,
|
8768 |
+
"learning_rate": 4.491198037323796e-06,
|
8769 |
+
"loss": 0.9296,
|
8770 |
+
"step": 1247
|
8771 |
+
},
|
8772 |
+
{
|
8773 |
+
"epoch": 0.08858131487889273,
|
8774 |
+
"grad_norm": 0.5316797494888306,
|
8775 |
+
"learning_rate": 4.423399961213892e-06,
|
8776 |
+
"loss": 0.8836,
|
8777 |
+
"step": 1248
|
8778 |
+
},
|
8779 |
+
{
|
8780 |
+
"epoch": 0.08865229349658416,
|
8781 |
+
"grad_norm": 0.45848792791366577,
|
8782 |
+
"learning_rate": 4.3561059412760675e-06,
|
8783 |
+
"loss": 1.0931,
|
8784 |
+
"step": 1249
|
8785 |
+
},
|
8786 |
+
{
|
8787 |
+
"epoch": 0.08872327211427558,
|
8788 |
+
"grad_norm": 0.5291265845298767,
|
8789 |
+
"learning_rate": 4.2893163324085885e-06,
|
8790 |
+
"loss": 0.9476,
|
8791 |
+
"step": 1250
|
8792 |
+
},
|
8793 |
+
{
|
8794 |
+
"epoch": 0.08879425073196699,
|
8795 |
+
"grad_norm": 0.6027399301528931,
|
8796 |
+
"learning_rate": 4.223031486849472e-06,
|
8797 |
+
"loss": 0.967,
|
8798 |
+
"step": 1251
|
8799 |
+
},
|
8800 |
+
{
|
8801 |
+
"epoch": 0.08886522934965842,
|
8802 |
+
"grad_norm": 0.8059013485908508,
|
8803 |
+
"learning_rate": 4.1572517541747294e-06,
|
8804 |
+
"loss": 1.1932,
|
8805 |
+
"step": 1252
|
8806 |
+
},
|
8807 |
+
{
|
8808 |
+
"epoch": 0.08893620796734984,
|
8809 |
+
"grad_norm": 0.4496321976184845,
|
8810 |
+
"learning_rate": 4.091977481296483e-06,
|
8811 |
+
"loss": 0.7402,
|
8812 |
+
"step": 1253
|
8813 |
+
},
|
8814 |
+
{
|
8815 |
+
"epoch": 0.08900718658504125,
|
8816 |
+
"grad_norm": 0.5405556559562683,
|
8817 |
+
"learning_rate": 4.027209012461108e-06,
|
8818 |
+
"loss": 1.0949,
|
8819 |
+
"step": 1254
|
8820 |
+
},
|
8821 |
+
{
|
8822 |
+
"epoch": 0.08907816520273268,
|
8823 |
+
"grad_norm": 0.39425739645957947,
|
8824 |
+
"learning_rate": 3.962946689247471e-06,
|
8825 |
+
"loss": 0.7058,
|
8826 |
+
"step": 1255
|
8827 |
+
},
|
8828 |
+
{
|
8829 |
+
"epoch": 0.0891491438204241,
|
8830 |
+
"grad_norm": 0.4992437958717346,
|
8831 |
+
"learning_rate": 3.899190850565115e-06,
|
8832 |
+
"loss": 0.9825,
|
8833 |
+
"step": 1256
|
8834 |
+
},
|
8835 |
+
{
|
8836 |
+
"epoch": 0.08922012243811552,
|
8837 |
+
"grad_norm": 0.31615936756134033,
|
8838 |
+
"learning_rate": 3.835941832652434e-06,
|
8839 |
+
"loss": 0.6421,
|
8840 |
+
"step": 1257
|
8841 |
+
},
|
8842 |
+
{
|
8843 |
+
"epoch": 0.08929110105580694,
|
8844 |
+
"grad_norm": 0.5412048101425171,
|
8845 |
+
"learning_rate": 3.7731999690749585e-06,
|
8846 |
+
"loss": 1.0389,
|
8847 |
+
"step": 1258
|
8848 |
+
},
|
8849 |
+
{
|
8850 |
+
"epoch": 0.08936207967349837,
|
8851 |
+
"grad_norm": 0.4242056906223297,
|
8852 |
+
"learning_rate": 3.7109655907235584e-06,
|
8853 |
+
"loss": 1.0086,
|
8854 |
+
"step": 1259
|
8855 |
+
},
|
8856 |
+
{
|
8857 |
+
"epoch": 0.08943305829118978,
|
8858 |
+
"grad_norm": 0.4553348124027252,
|
8859 |
+
"learning_rate": 3.6492390258126673e-06,
|
8860 |
+
"loss": 0.73,
|
8861 |
+
"step": 1260
|
8862 |
+
},
|
8863 |
+
{
|
8864 |
+
"epoch": 0.0895040369088812,
|
8865 |
+
"grad_norm": 0.4538853168487549,
|
8866 |
+
"learning_rate": 3.5880205998786388e-06,
|
8867 |
+
"loss": 0.8733,
|
8868 |
+
"step": 1261
|
8869 |
+
},
|
8870 |
+
{
|
8871 |
+
"epoch": 0.08957501552657263,
|
8872 |
+
"grad_norm": 0.42450854182243347,
|
8873 |
+
"learning_rate": 3.5273106357779585e-06,
|
8874 |
+
"loss": 0.8259,
|
8875 |
+
"step": 1262
|
8876 |
+
},
|
8877 |
+
{
|
8878 |
+
"epoch": 0.08964599414426404,
|
8879 |
+
"grad_norm": 0.5689201354980469,
|
8880 |
+
"learning_rate": 3.4671094536855687e-06,
|
8881 |
+
"loss": 0.9794,
|
8882 |
+
"step": 1263
|
8883 |
+
},
|
8884 |
+
{
|
8885 |
+
"epoch": 0.08971697276195546,
|
8886 |
+
"grad_norm": 0.46215319633483887,
|
8887 |
+
"learning_rate": 3.40741737109318e-06,
|
8888 |
+
"loss": 1.3214,
|
8889 |
+
"step": 1264
|
8890 |
+
},
|
8891 |
+
{
|
8892 |
+
"epoch": 0.08978795137964689,
|
8893 |
+
"grad_norm": 0.6304516196250916,
|
8894 |
+
"learning_rate": 3.3482347028075513e-06,
|
8895 |
+
"loss": 1.1341,
|
8896 |
+
"step": 1265
|
8897 |
+
},
|
8898 |
+
{
|
8899 |
+
"epoch": 0.0898589299973383,
|
8900 |
+
"grad_norm": 0.5463753938674927,
|
8901 |
+
"learning_rate": 3.2895617609489336e-06,
|
8902 |
+
"loss": 0.8568,
|
8903 |
+
"step": 1266
|
8904 |
+
},
|
8905 |
+
{
|
8906 |
+
"epoch": 0.08992990861502972,
|
8907 |
+
"grad_norm": 0.7814898490905762,
|
8908 |
+
"learning_rate": 3.231398854949297e-06,
|
8909 |
+
"loss": 1.2902,
|
8910 |
+
"step": 1267
|
8911 |
+
},
|
8912 |
+
{
|
8913 |
+
"epoch": 0.09000088723272115,
|
8914 |
+
"grad_norm": 0.395901083946228,
|
8915 |
+
"learning_rate": 3.1737462915508277e-06,
|
8916 |
+
"loss": 0.9164,
|
8917 |
+
"step": 1268
|
8918 |
+
},
|
8919 |
+
{
|
8920 |
+
"epoch": 0.09007186585041256,
|
8921 |
+
"grad_norm": 0.47108373045921326,
|
8922 |
+
"learning_rate": 3.1166043748041772e-06,
|
8923 |
+
"loss": 0.9783,
|
8924 |
+
"step": 1269
|
8925 |
+
},
|
8926 |
+
{
|
8927 |
+
"epoch": 0.09014284446810399,
|
8928 |
+
"grad_norm": 0.6641191244125366,
|
8929 |
+
"learning_rate": 3.059973406066963e-06,
|
8930 |
+
"loss": 1.2843,
|
8931 |
+
"step": 1270
|
8932 |
+
},
|
8933 |
+
{
|
8934 |
+
"epoch": 0.09021382308579541,
|
8935 |
+
"grad_norm": 0.44633758068084717,
|
8936 |
+
"learning_rate": 3.0038536840021335e-06,
|
8937 |
+
"loss": 0.8454,
|
8938 |
+
"step": 1271
|
8939 |
+
},
|
8940 |
+
{
|
8941 |
+
"epoch": 0.09028480170348682,
|
8942 |
+
"grad_norm": 0.49430206418037415,
|
8943 |
+
"learning_rate": 2.948245504576419e-06,
|
8944 |
+
"loss": 1.2095,
|
8945 |
+
"step": 1272
|
8946 |
+
},
|
8947 |
+
{
|
8948 |
+
"epoch": 0.09035578032117825,
|
8949 |
+
"grad_norm": 0.4891290068626404,
|
8950 |
+
"learning_rate": 2.893149161058717e-06,
|
8951 |
+
"loss": 0.7466,
|
8952 |
+
"step": 1273
|
8953 |
+
},
|
8954 |
+
{
|
8955 |
+
"epoch": 0.09042675893886967,
|
8956 |
+
"grad_norm": 0.48959529399871826,
|
8957 |
+
"learning_rate": 2.838564944018618e-06,
|
8958 |
+
"loss": 0.8947,
|
8959 |
+
"step": 1274
|
8960 |
+
},
|
8961 |
+
{
|
8962 |
+
"epoch": 0.09049773755656108,
|
8963 |
+
"grad_norm": 0.4747793972492218,
|
8964 |
+
"learning_rate": 2.784493141324818e-06,
|
8965 |
+
"loss": 1.1404,
|
8966 |
+
"step": 1275
|
8967 |
+
},
|
8968 |
+
{
|
8969 |
+
"epoch": 0.09056871617425251,
|
8970 |
+
"grad_norm": 0.41771140694618225,
|
8971 |
+
"learning_rate": 2.730934038143607e-06,
|
8972 |
+
"loss": 0.9322,
|
8973 |
+
"step": 1276
|
8974 |
+
},
|
8975 |
+
{
|
8976 |
+
"epoch": 0.09063969479194392,
|
8977 |
+
"grad_norm": 0.46313968300819397,
|
8978 |
+
"learning_rate": 2.677887916937416e-06,
|
8979 |
+
"loss": 0.9321,
|
8980 |
+
"step": 1277
|
8981 |
+
},
|
8982 |
+
{
|
8983 |
+
"epoch": 0.09071067340963535,
|
8984 |
+
"grad_norm": 0.6357395052909851,
|
8985 |
+
"learning_rate": 2.6253550574632303e-06,
|
8986 |
+
"loss": 1.2126,
|
8987 |
+
"step": 1278
|
8988 |
+
},
|
8989 |
+
{
|
8990 |
+
"epoch": 0.09078165202732677,
|
8991 |
+
"grad_norm": 0.48085835576057434,
|
8992 |
+
"learning_rate": 2.573335736771254e-06,
|
8993 |
+
"loss": 0.7637,
|
8994 |
+
"step": 1279
|
8995 |
+
},
|
8996 |
+
{
|
8997 |
+
"epoch": 0.09085263064501818,
|
8998 |
+
"grad_norm": 0.7907678484916687,
|
8999 |
+
"learning_rate": 2.5218302292032816e-06,
|
9000 |
+
"loss": 1.1955,
|
9001 |
+
"step": 1280
|
9002 |
+
},
|
9003 |
+
{
|
9004 |
+
"epoch": 0.0909236092627096,
|
9005 |
+
"grad_norm": 0.6866028308868408,
|
9006 |
+
"learning_rate": 2.4708388063913844e-06,
|
9007 |
+
"loss": 1.1785,
|
9008 |
+
"step": 1281
|
9009 |
+
},
|
9010 |
+
{
|
9011 |
+
"epoch": 0.09099458788040103,
|
9012 |
+
"grad_norm": 0.41570764780044556,
|
9013 |
+
"learning_rate": 2.420361737256438e-06,
|
9014 |
+
"loss": 0.94,
|
9015 |
+
"step": 1282
|
9016 |
+
},
|
9017 |
+
{
|
9018 |
+
"epoch": 0.09106556649809244,
|
9019 |
+
"grad_norm": 0.4501184821128845,
|
9020 |
+
"learning_rate": 2.3703992880066638e-06,
|
9021 |
+
"loss": 1.0563,
|
9022 |
+
"step": 1283
|
9023 |
+
},
|
9024 |
+
{
|
9025 |
+
"epoch": 0.09113654511578387,
|
9026 |
+
"grad_norm": 0.47688695788383484,
|
9027 |
+
"learning_rate": 2.3209517221362777e-06,
|
9028 |
+
"loss": 0.8763,
|
9029 |
+
"step": 1284
|
9030 |
+
},
|
9031 |
+
{
|
9032 |
+
"epoch": 0.0912075237334753,
|
9033 |
+
"grad_norm": 0.467288076877594,
|
9034 |
+
"learning_rate": 2.2720193004240774e-06,
|
9035 |
+
"loss": 0.9097,
|
9036 |
+
"step": 1285
|
9037 |
+
},
|
9038 |
+
{
|
9039 |
+
"epoch": 0.0912785023511667,
|
9040 |
+
"grad_norm": 0.4979704022407532,
|
9041 |
+
"learning_rate": 2.22360228093208e-06,
|
9042 |
+
"loss": 1.069,
|
9043 |
+
"step": 1286
|
9044 |
+
},
|
9045 |
+
{
|
9046 |
+
"epoch": 0.09134948096885813,
|
9047 |
+
"grad_norm": 0.6934332847595215,
|
9048 |
+
"learning_rate": 2.1757009190041088e-06,
|
9049 |
+
"loss": 0.8044,
|
9050 |
+
"step": 1287
|
9051 |
+
},
|
9052 |
+
{
|
9053 |
+
"epoch": 0.09142045958654955,
|
9054 |
+
"grad_norm": 1.4915975332260132,
|
9055 |
+
"learning_rate": 2.128315467264552e-06,
|
9056 |
+
"loss": 0.9586,
|
9057 |
+
"step": 1288
|
9058 |
+
},
|
9059 |
+
{
|
9060 |
+
"epoch": 0.09149143820424097,
|
9061 |
+
"grad_norm": 0.3834763765335083,
|
9062 |
+
"learning_rate": 2.08144617561693e-06,
|
9063 |
+
"loss": 0.7679,
|
9064 |
+
"step": 1289
|
9065 |
+
},
|
9066 |
+
{
|
9067 |
+
"epoch": 0.09156241682193239,
|
9068 |
+
"grad_norm": 0.36206838488578796,
|
9069 |
+
"learning_rate": 2.035093291242607e-06,
|
9070 |
+
"loss": 0.7367,
|
9071 |
+
"step": 1290
|
9072 |
+
},
|
9073 |
+
{
|
9074 |
+
"epoch": 0.09163339543962382,
|
9075 |
+
"grad_norm": 0.5193326473236084,
|
9076 |
+
"learning_rate": 1.9892570585995363e-06,
|
9077 |
+
"loss": 0.9015,
|
9078 |
+
"step": 1291
|
9079 |
+
},
|
9080 |
+
{
|
9081 |
+
"epoch": 0.09170437405731523,
|
9082 |
+
"grad_norm": 0.46295395493507385,
|
9083 |
+
"learning_rate": 1.943937719420863e-06,
|
9084 |
+
"loss": 0.8407,
|
9085 |
+
"step": 1292
|
9086 |
+
},
|
9087 |
+
{
|
9088 |
+
"epoch": 0.09177535267500665,
|
9089 |
+
"grad_norm": 0.4065030813217163,
|
9090 |
+
"learning_rate": 1.899135512713801e-06,
|
9091 |
+
"loss": 0.9282,
|
9092 |
+
"step": 1293
|
9093 |
+
},
|
9094 |
+
{
|
9095 |
+
"epoch": 0.09184633129269808,
|
9096 |
+
"grad_norm": 0.4650251865386963,
|
9097 |
+
"learning_rate": 1.8548506747582129e-06,
|
9098 |
+
"loss": 1.0284,
|
9099 |
+
"step": 1294
|
9100 |
+
},
|
9101 |
+
{
|
9102 |
+
"epoch": 0.09191730991038949,
|
9103 |
+
"grad_norm": 0.43935665488243103,
|
9104 |
+
"learning_rate": 1.8110834391054764e-06,
|
9105 |
+
"loss": 0.8786,
|
9106 |
+
"step": 1295
|
9107 |
+
},
|
9108 |
+
{
|
9109 |
+
"epoch": 0.09198828852808091,
|
9110 |
+
"grad_norm": 0.3632609248161316,
|
9111 |
+
"learning_rate": 1.7678340365772206e-06,
|
9112 |
+
"loss": 0.8486,
|
9113 |
+
"step": 1296
|
9114 |
+
},
|
9115 |
+
{
|
9116 |
+
"epoch": 0.09205926714577234,
|
9117 |
+
"grad_norm": 0.5031712055206299,
|
9118 |
+
"learning_rate": 1.725102695264058e-06,
|
9119 |
+
"loss": 0.8989,
|
9120 |
+
"step": 1297
|
9121 |
+
},
|
9122 |
+
{
|
9123 |
+
"epoch": 0.09213024576346375,
|
9124 |
+
"grad_norm": 0.5061722993850708,
|
9125 |
+
"learning_rate": 1.6828896405244988e-06,
|
9126 |
+
"loss": 0.8698,
|
9127 |
+
"step": 1298
|
9128 |
+
},
|
9129 |
+
{
|
9130 |
+
"epoch": 0.09220122438115518,
|
9131 |
+
"grad_norm": 0.6190900802612305,
|
9132 |
+
"learning_rate": 1.6411950949836164e-06,
|
9133 |
+
"loss": 1.1324,
|
9134 |
+
"step": 1299
|
9135 |
+
},
|
9136 |
+
{
|
9137 |
+
"epoch": 0.0922722029988466,
|
9138 |
+
"grad_norm": 0.4328751266002655,
|
9139 |
+
"learning_rate": 1.6000192785320057e-06,
|
9140 |
+
"loss": 0.9834,
|
9141 |
+
"step": 1300
|
9142 |
+
},
|
9143 |
+
{
|
9144 |
+
"epoch": 0.09234318161653801,
|
9145 |
+
"grad_norm": 0.45743778347969055,
|
9146 |
+
"learning_rate": 1.5593624083245162e-06,
|
9147 |
+
"loss": 0.8712,
|
9148 |
+
"step": 1301
|
9149 |
+
},
|
9150 |
+
{
|
9151 |
+
"epoch": 0.09241416023422944,
|
9152 |
+
"grad_norm": 0.4247590899467468,
|
9153 |
+
"learning_rate": 1.5192246987791981e-06,
|
9154 |
+
"loss": 0.7241,
|
9155 |
+
"step": 1302
|
9156 |
+
},
|
9157 |
+
{
|
9158 |
+
"epoch": 0.09248513885192086,
|
9159 |
+
"grad_norm": 0.5770044326782227,
|
9160 |
+
"learning_rate": 1.4796063615760913e-06,
|
9161 |
+
"loss": 0.8319,
|
9162 |
+
"step": 1303
|
9163 |
+
},
|
9164 |
+
{
|
9165 |
+
"epoch": 0.09255611746961227,
|
9166 |
+
"grad_norm": 0.4635300934314728,
|
9167 |
+
"learning_rate": 1.4405076056561828e-06,
|
9168 |
+
"loss": 1.0233,
|
9169 |
+
"step": 1304
|
9170 |
+
},
|
9171 |
+
{
|
9172 |
+
"epoch": 0.0926270960873037,
|
9173 |
+
"grad_norm": 0.42207738757133484,
|
9174 |
+
"learning_rate": 1.40192863722024e-06,
|
9175 |
+
"loss": 0.7407,
|
9176 |
+
"step": 1305
|
9177 |
+
},
|
9178 |
+
{
|
9179 |
+
"epoch": 0.09269807470499512,
|
9180 |
+
"grad_norm": 0.5023476481437683,
|
9181 |
+
"learning_rate": 1.3638696597277679e-06,
|
9182 |
+
"loss": 0.9832,
|
9183 |
+
"step": 1306
|
9184 |
+
},
|
9185 |
+
{
|
9186 |
+
"epoch": 0.09276905332268653,
|
9187 |
+
"grad_norm": 0.5583863258361816,
|
9188 |
+
"learning_rate": 1.326330873895909e-06,
|
9189 |
+
"loss": 1.1022,
|
9190 |
+
"step": 1307
|
9191 |
+
},
|
9192 |
+
{
|
9193 |
+
"epoch": 0.09284003194037796,
|
9194 |
+
"grad_norm": 0.4288213551044464,
|
9195 |
+
"learning_rate": 1.28931247769839e-06,
|
9196 |
+
"loss": 0.9897,
|
9197 |
+
"step": 1308
|
9198 |
+
},
|
9199 |
+
{
|
9200 |
+
"epoch": 0.09291101055806938,
|
9201 |
+
"grad_norm": 0.5217162370681763,
|
9202 |
+
"learning_rate": 1.2528146663645102e-06,
|
9203 |
+
"loss": 0.8312,
|
9204 |
+
"step": 1309
|
9205 |
+
},
|
9206 |
+
{
|
9207 |
+
"epoch": 0.0929819891757608,
|
9208 |
+
"grad_norm": 0.6898660063743591,
|
9209 |
+
"learning_rate": 1.216837632378065e-06,
|
9210 |
+
"loss": 1.0515,
|
9211 |
+
"step": 1310
|
9212 |
+
},
|
9213 |
+
{
|
9214 |
+
"epoch": 0.09305296779345222,
|
9215 |
+
"grad_norm": 0.48667794466018677,
|
9216 |
+
"learning_rate": 1.1813815654763472e-06,
|
9217 |
+
"loss": 0.8842,
|
9218 |
+
"step": 1311
|
9219 |
+
},
|
9220 |
+
{
|
9221 |
+
"epoch": 0.09312394641114365,
|
9222 |
+
"grad_norm": 0.4529171884059906,
|
9223 |
+
"learning_rate": 1.146446652649169e-06,
|
9224 |
+
"loss": 1.3358,
|
9225 |
+
"step": 1312
|
9226 |
+
},
|
9227 |
+
{
|
9228 |
+
"epoch": 0.09319492502883506,
|
9229 |
+
"grad_norm": 0.6250870227813721,
|
9230 |
+
"learning_rate": 1.1120330781378307e-06,
|
9231 |
+
"loss": 0.8719,
|
9232 |
+
"step": 1313
|
9233 |
+
},
|
9234 |
+
{
|
9235 |
+
"epoch": 0.09326590364652648,
|
9236 |
+
"grad_norm": 0.4203828275203705,
|
9237 |
+
"learning_rate": 1.0781410234342094e-06,
|
9238 |
+
"loss": 0.7693,
|
9239 |
+
"step": 1314
|
9240 |
+
},
|
9241 |
+
{
|
9242 |
+
"epoch": 0.09333688226421791,
|
9243 |
+
"grad_norm": 0.40615299344062805,
|
9244 |
+
"learning_rate": 1.0447706672797264e-06,
|
9245 |
+
"loss": 0.7339,
|
9246 |
+
"step": 1315
|
9247 |
+
},
|
9248 |
+
{
|
9249 |
+
"epoch": 0.09340786088190932,
|
9250 |
+
"grad_norm": 0.4049832224845886,
|
9251 |
+
"learning_rate": 1.0119221856644712e-06,
|
9252 |
+
"loss": 0.7479,
|
9253 |
+
"step": 1316
|
9254 |
+
},
|
9255 |
+
{
|
9256 |
+
"epoch": 0.09347883949960074,
|
9257 |
+
"grad_norm": 0.41221579909324646,
|
9258 |
+
"learning_rate": 9.795957518262566e-07,
|
9259 |
+
"loss": 0.713,
|
9260 |
+
"step": 1317
|
9261 |
+
},
|
9262 |
+
{
|
9263 |
+
"epoch": 0.09354981811729217,
|
9264 |
+
"grad_norm": 0.42666563391685486,
|
9265 |
+
"learning_rate": 9.477915362496758e-07,
|
9266 |
+
"loss": 1.06,
|
9267 |
+
"step": 1318
|
9268 |
+
},
|
9269 |
+
{
|
9270 |
+
"epoch": 0.09362079673498358,
|
9271 |
+
"grad_norm": 0.4528525173664093,
|
9272 |
+
"learning_rate": 9.16509706665225e-07,
|
9273 |
+
"loss": 0.7928,
|
9274 |
+
"step": 1319
|
9275 |
+
},
|
9276 |
+
{
|
9277 |
+
"epoch": 0.093691775352675,
|
9278 |
+
"grad_norm": 0.5852678418159485,
|
9279 |
+
"learning_rate": 8.857504280484375e-07,
|
9280 |
+
"loss": 0.9635,
|
9281 |
+
"step": 1320
|
9282 |
+
},
|
9283 |
+
{
|
9284 |
+
"epoch": 0.09376275397036643,
|
9285 |
+
"grad_norm": 0.47171440720558167,
|
9286 |
+
"learning_rate": 8.555138626189618e-07,
|
9287 |
+
"loss": 1.0355,
|
9288 |
+
"step": 1321
|
9289 |
+
},
|
9290 |
+
{
|
9291 |
+
"epoch": 0.09383373258805784,
|
9292 |
+
"grad_norm": 0.5309885740280151,
|
9293 |
+
"learning_rate": 8.258001698397744e-07,
|
9294 |
+
"loss": 1.0872,
|
9295 |
+
"step": 1322
|
9296 |
+
},
|
9297 |
+
{
|
9298 |
+
"epoch": 0.09390471120574927,
|
9299 |
+
"grad_norm": 0.6979671716690063,
|
9300 |
+
"learning_rate": 7.966095064162682e-07,
|
9301 |
+
"loss": 1.2462,
|
9302 |
+
"step": 1323
|
9303 |
+
},
|
9304 |
+
{
|
9305 |
+
"epoch": 0.09397568982344069,
|
9306 |
+
"grad_norm": 0.5324621796607971,
|
9307 |
+
"learning_rate": 7.679420262954984e-07,
|
9308 |
+
"loss": 0.9156,
|
9309 |
+
"step": 1324
|
9310 |
+
},
|
9311 |
+
{
|
9312 |
+
"epoch": 0.0940466684411321,
|
9313 |
+
"grad_norm": 0.42955929040908813,
|
9314 |
+
"learning_rate": 7.397978806652939e-07,
|
9315 |
+
"loss": 0.636,
|
9316 |
+
"step": 1325
|
9317 |
+
},
|
9318 |
+
{
|
9319 |
+
"epoch": 0.09411764705882353,
|
9320 |
+
"grad_norm": 0.4044051468372345,
|
9321 |
+
"learning_rate": 7.121772179535135e-07,
|
9322 |
+
"loss": 0.8698,
|
9323 |
+
"step": 1326
|
9324 |
+
},
|
9325 |
+
{
|
9326 |
+
"epoch": 0.09418862567651495,
|
9327 |
+
"grad_norm": 0.5710899829864502,
|
9328 |
+
"learning_rate": 6.850801838272691e-07,
|
9329 |
+
"loss": 1.2614,
|
9330 |
+
"step": 1327
|
9331 |
+
},
|
9332 |
+
{
|
9333 |
+
"epoch": 0.09425960429420636,
|
9334 |
+
"grad_norm": 0.5085283517837524,
|
9335 |
+
"learning_rate": 6.585069211921035e-07,
|
9336 |
+
"loss": 0.9378,
|
9337 |
+
"step": 1328
|
9338 |
+
},
|
9339 |
+
{
|
9340 |
+
"epoch": 0.09433058291189779,
|
9341 |
+
"grad_norm": 0.4785238802433014,
|
9342 |
+
"learning_rate": 6.324575701912805e-07,
|
9343 |
+
"loss": 1.1437,
|
9344 |
+
"step": 1329
|
9345 |
+
},
|
9346 |
+
{
|
9347 |
+
"epoch": 0.09440156152958921,
|
9348 |
+
"grad_norm": 0.6342504024505615,
|
9349 |
+
"learning_rate": 6.069322682050516e-07,
|
9350 |
+
"loss": 0.9806,
|
9351 |
+
"step": 1330
|
9352 |
+
},
|
9353 |
+
{
|
9354 |
+
"epoch": 0.09447254014728063,
|
9355 |
+
"grad_norm": 0.5013599395751953,
|
9356 |
+
"learning_rate": 5.819311498499013e-07,
|
9357 |
+
"loss": 1.2397,
|
9358 |
+
"step": 1331
|
9359 |
+
},
|
9360 |
+
{
|
9361 |
+
"epoch": 0.09454351876497205,
|
9362 |
+
"grad_norm": 0.36016663908958435,
|
9363 |
+
"learning_rate": 5.57454346977837e-07,
|
9364 |
+
"loss": 0.7081,
|
9365 |
+
"step": 1332
|
9366 |
+
},
|
9367 |
+
{
|
9368 |
+
"epoch": 0.09461449738266348,
|
9369 |
+
"grad_norm": 0.3618755340576172,
|
9370 |
+
"learning_rate": 5.335019886757442e-07,
|
9371 |
+
"loss": 0.7656,
|
9372 |
+
"step": 1333
|
9373 |
+
},
|
9374 |
+
{
|
9375 |
+
"epoch": 0.09468547600035489,
|
9376 |
+
"grad_norm": 0.37141433358192444,
|
9377 |
+
"learning_rate": 5.10074201264632e-07,
|
9378 |
+
"loss": 0.8844,
|
9379 |
+
"step": 1334
|
9380 |
+
},
|
9381 |
+
{
|
9382 |
+
"epoch": 0.09475645461804631,
|
9383 |
+
"grad_norm": 0.4491944909095764,
|
9384 |
+
"learning_rate": 4.871711082990227e-07,
|
9385 |
+
"loss": 0.8487,
|
9386 |
+
"step": 1335
|
9387 |
+
},
|
9388 |
+
{
|
9389 |
+
"epoch": 0.09482743323573774,
|
9390 |
+
"grad_norm": 0.4374077320098877,
|
9391 |
+
"learning_rate": 4.647928305662852e-07,
|
9392 |
+
"loss": 0.8902,
|
9393 |
+
"step": 1336
|
9394 |
+
},
|
9395 |
+
{
|
9396 |
+
"epoch": 0.09489841185342915,
|
9397 |
+
"grad_norm": 0.4476800262928009,
|
9398 |
+
"learning_rate": 4.4293948608596925e-07,
|
9399 |
+
"loss": 1.0808,
|
9400 |
+
"step": 1337
|
9401 |
+
},
|
9402 |
+
{
|
9403 |
+
"epoch": 0.09496939047112057,
|
9404 |
+
"grad_norm": 0.623502790927887,
|
9405 |
+
"learning_rate": 4.216111901092501e-07,
|
9406 |
+
"loss": 1.0005,
|
9407 |
+
"step": 1338
|
9408 |
+
},
|
9409 |
+
{
|
9410 |
+
"epoch": 0.095040369088812,
|
9411 |
+
"grad_norm": 0.4662952125072479,
|
9412 |
+
"learning_rate": 4.008080551182403e-07,
|
9413 |
+
"loss": 0.7606,
|
9414 |
+
"step": 1339
|
9415 |
+
},
|
9416 |
+
{
|
9417 |
+
"epoch": 0.09511134770650341,
|
9418 |
+
"grad_norm": 0.5216524004936218,
|
9419 |
+
"learning_rate": 3.805301908254455e-07,
|
9420 |
+
"loss": 0.9909,
|
9421 |
+
"step": 1340
|
9422 |
+
},
|
9423 |
+
{
|
9424 |
+
"epoch": 0.09518232632419484,
|
9425 |
+
"grad_norm": 0.4408697485923767,
|
9426 |
+
"learning_rate": 3.6077770417318744e-07,
|
9427 |
+
"loss": 0.7013,
|
9428 |
+
"step": 1341
|
9429 |
+
},
|
9430 |
+
{
|
9431 |
+
"epoch": 0.09525330494188626,
|
9432 |
+
"grad_norm": 0.4359828233718872,
|
9433 |
+
"learning_rate": 3.415506993330153e-07,
|
9434 |
+
"loss": 1.0082,
|
9435 |
+
"step": 1342
|
9436 |
+
},
|
9437 |
+
{
|
9438 |
+
"epoch": 0.09532428355957767,
|
9439 |
+
"grad_norm": 0.3965587019920349,
|
9440 |
+
"learning_rate": 3.228492777051728e-07,
|
9441 |
+
"loss": 1.0,
|
9442 |
+
"step": 1343
|
9443 |
+
},
|
9444 |
+
{
|
9445 |
+
"epoch": 0.0953952621772691,
|
9446 |
+
"grad_norm": 0.41036057472229004,
|
9447 |
+
"learning_rate": 3.046735379180543e-07,
|
9448 |
+
"loss": 0.8307,
|
9449 |
+
"step": 1344
|
9450 |
+
},
|
9451 |
+
{
|
9452 |
+
"epoch": 0.09546624079496052,
|
9453 |
+
"grad_norm": 0.8139565587043762,
|
9454 |
+
"learning_rate": 2.87023575827694e-07,
|
9455 |
+
"loss": 1.0802,
|
9456 |
+
"step": 1345
|
9457 |
+
},
|
9458 |
+
{
|
9459 |
+
"epoch": 0.09553721941265193,
|
9460 |
+
"grad_norm": 0.9029821753501892,
|
9461 |
+
"learning_rate": 2.6989948451726643e-07,
|
9462 |
+
"loss": 2.0635,
|
9463 |
+
"step": 1346
|
9464 |
+
},
|
9465 |
+
{
|
9466 |
+
"epoch": 0.09560819803034336,
|
9467 |
+
"grad_norm": 0.5215340852737427,
|
9468 |
+
"learning_rate": 2.5330135429656457e-07,
|
9469 |
+
"loss": 0.9462,
|
9470 |
+
"step": 1347
|
9471 |
+
},
|
9472 |
+
{
|
9473 |
+
"epoch": 0.09567917664803478,
|
9474 |
+
"grad_norm": 0.4245148003101349,
|
9475 |
+
"learning_rate": 2.372292727015557e-07,
|
9476 |
+
"loss": 0.7983,
|
9477 |
+
"step": 1348
|
9478 |
+
},
|
9479 |
+
{
|
9480 |
+
"epoch": 0.0957501552657262,
|
9481 |
+
"grad_norm": 0.4835399389266968,
|
9482 |
+
"learning_rate": 2.2168332449390427e-07,
|
9483 |
+
"loss": 1.0793,
|
9484 |
+
"step": 1349
|
9485 |
+
},
|
9486 |
+
{
|
9487 |
+
"epoch": 0.09582113388341762,
|
9488 |
+
"grad_norm": 0.5569635629653931,
|
9489 |
+
"learning_rate": 2.066635916605386e-07,
|
9490 |
+
"loss": 0.9318,
|
9491 |
+
"step": 1350
|
9492 |
+
},
|
9493 |
+
{
|
9494 |
+
"epoch": 0.09589211250110904,
|
9495 |
+
"grad_norm": 0.4609260559082031,
|
9496 |
+
"learning_rate": 1.921701534131848e-07,
|
9497 |
+
"loss": 1.0345,
|
9498 |
+
"step": 1351
|
9499 |
+
},
|
9500 |
+
{
|
9501 |
+
"epoch": 0.09596309111880046,
|
9502 |
+
"grad_norm": 0.389445424079895,
|
9503 |
+
"learning_rate": 1.782030861880113e-07,
|
9504 |
+
"loss": 0.7788,
|
9505 |
+
"step": 1352
|
9506 |
+
},
|
9507 |
+
{
|
9508 |
+
"epoch": 0.09603406973649188,
|
9509 |
+
"grad_norm": 0.3875621557235718,
|
9510 |
+
"learning_rate": 1.6476246364515169e-07,
|
9511 |
+
"loss": 0.7591,
|
9512 |
+
"step": 1353
|
9513 |
+
},
|
9514 |
+
{
|
9515 |
+
"epoch": 0.0961050483541833,
|
9516 |
+
"grad_norm": 0.5420302152633667,
|
9517 |
+
"learning_rate": 1.518483566683826e-07,
|
9518 |
+
"loss": 1.1174,
|
9519 |
+
"step": 1354
|
9520 |
+
},
|
9521 |
+
{
|
9522 |
+
"epoch": 0.09617602697187472,
|
9523 |
+
"grad_norm": 0.5549187660217285,
|
9524 |
+
"learning_rate": 1.3946083336467964e-07,
|
9525 |
+
"loss": 0.9157,
|
9526 |
+
"step": 1355
|
9527 |
+
},
|
9528 |
+
{
|
9529 |
+
"epoch": 0.09624700558956614,
|
9530 |
+
"grad_norm": 0.6199358701705933,
|
9531 |
+
"learning_rate": 1.2759995906392874e-07,
|
9532 |
+
"loss": 0.8482,
|
9533 |
+
"step": 1356
|
9534 |
+
},
|
9535 |
+
{
|
9536 |
+
"epoch": 0.09631798420725757,
|
9537 |
+
"grad_norm": 0.646535336971283,
|
9538 |
+
"learning_rate": 1.1626579631853762e-07,
|
9539 |
+
"loss": 1.086,
|
9540 |
+
"step": 1357
|
9541 |
+
},
|
9542 |
+
{
|
9543 |
+
"epoch": 0.09638896282494898,
|
9544 |
+
"grad_norm": 0.4827463924884796,
|
9545 |
+
"learning_rate": 1.0545840490313596e-07,
|
9546 |
+
"loss": 0.8859,
|
9547 |
+
"step": 1358
|
9548 |
+
},
|
9549 |
+
{
|
9550 |
+
"epoch": 0.0964599414426404,
|
9551 |
+
"grad_norm": 0.3394274413585663,
|
9552 |
+
"learning_rate": 9.517784181422019e-08,
|
9553 |
+
"loss": 0.8065,
|
9554 |
+
"step": 1359
|
9555 |
+
},
|
9556 |
+
{
|
9557 |
+
"epoch": 0.09653092006033183,
|
9558 |
+
"grad_norm": 0.5661752223968506,
|
9559 |
+
"learning_rate": 8.542416126989805e-08,
|
9560 |
+
"loss": 0.9616,
|
9561 |
+
"step": 1360
|
9562 |
+
},
|
9563 |
+
{
|
9564 |
+
"epoch": 0.09660189867802324,
|
9565 |
+
"grad_norm": 0.3854394853115082,
|
9566 |
+
"learning_rate": 7.619741470955566e-08,
|
9567 |
+
"loss": 0.7678,
|
9568 |
+
"step": 1361
|
9569 |
+
},
|
9570 |
+
{
|
9571 |
+
"epoch": 0.09667287729571467,
|
9572 |
+
"grad_norm": 0.6302008032798767,
|
9573 |
+
"learning_rate": 6.749765079363534e-08,
|
9574 |
+
"loss": 1.1516,
|
9575 |
+
"step": 1362
|
9576 |
+
},
|
9577 |
+
{
|
9578 |
+
"epoch": 0.09674385591340609,
|
9579 |
+
"grad_norm": 0.4585036635398865,
|
9580 |
+
"learning_rate": 5.9324915403324855e-08,
|
9581 |
+
"loss": 0.6375,
|
9582 |
+
"step": 1363
|
9583 |
+
},
|
9584 |
+
{
|
9585 |
+
"epoch": 0.0968148345310975,
|
9586 |
+
"grad_norm": 0.493896484375,
|
9587 |
+
"learning_rate": 5.167925164037968e-08,
|
9588 |
+
"loss": 0.7289,
|
9589 |
+
"step": 1364
|
9590 |
+
},
|
9591 |
+
{
|
9592 |
+
"epoch": 0.09688581314878893,
|
9593 |
+
"grad_norm": 0.42399123311042786,
|
9594 |
+
"learning_rate": 4.456069982684552e-08,
|
9595 |
+
"loss": 1.0201,
|
9596 |
+
"step": 1365
|
9597 |
+
},
|
9598 |
+
{
|
9599 |
+
"epoch": 0.09695679176648035,
|
9600 |
+
"grad_norm": 0.6302375793457031,
|
9601 |
+
"learning_rate": 3.796929750485845e-08,
|
9602 |
+
"loss": 1.1759,
|
9603 |
+
"step": 1366
|
9604 |
+
},
|
9605 |
+
{
|
9606 |
+
"epoch": 0.09702777038417176,
|
9607 |
+
"grad_norm": 0.4702526926994324,
|
9608 |
+
"learning_rate": 3.190507943644505e-08,
|
9609 |
+
"loss": 0.8664,
|
9610 |
+
"step": 1367
|
9611 |
+
},
|
9612 |
+
{
|
9613 |
+
"epoch": 0.09709874900186319,
|
9614 |
+
"grad_norm": 0.5638194680213928,
|
9615 |
+
"learning_rate": 2.6368077603367015e-08,
|
9616 |
+
"loss": 1.117,
|
9617 |
+
"step": 1368
|
9618 |
+
},
|
9619 |
+
{
|
9620 |
+
"epoch": 0.09716972761955461,
|
9621 |
+
"grad_norm": 0.5271203517913818,
|
9622 |
+
"learning_rate": 2.135832120689907e-08,
|
9623 |
+
"loss": 0.8312,
|
9624 |
+
"step": 1369
|
9625 |
+
},
|
9626 |
+
{
|
9627 |
+
"epoch": 0.09724070623724602,
|
9628 |
+
"grad_norm": 0.43867847323417664,
|
9629 |
+
"learning_rate": 1.687583666772907e-08,
|
9630 |
+
"loss": 0.7566,
|
9631 |
+
"step": 1370
|
9632 |
+
},
|
9633 |
+
{
|
9634 |
+
"epoch": 0.09731168485493745,
|
9635 |
+
"grad_norm": 0.45883628726005554,
|
9636 |
+
"learning_rate": 1.2920647625769277e-08,
|
9637 |
+
"loss": 0.9586,
|
9638 |
+
"step": 1371
|
9639 |
+
},
|
9640 |
+
{
|
9641 |
+
"epoch": 0.09738266347262887,
|
9642 |
+
"grad_norm": 0.542040228843689,
|
9643 |
+
"learning_rate": 9.49277494008971e-09,
|
9644 |
+
"loss": 0.9555,
|
9645 |
+
"step": 1372
|
9646 |
+
},
|
9647 |
+
{
|
9648 |
+
"epoch": 0.09745364209032029,
|
9649 |
+
"grad_norm": 0.45585867762565613,
|
9650 |
+
"learning_rate": 6.592236688762743e-09,
|
9651 |
+
"loss": 0.8365,
|
9652 |
+
"step": 1373
|
9653 |
+
},
|
9654 |
+
{
|
9655 |
+
"epoch": 0.09752462070801171,
|
9656 |
+
"grad_norm": 0.5066502094268799,
|
9657 |
+
"learning_rate": 4.219048168763174e-09,
|
9658 |
+
"loss": 0.8912,
|
9659 |
+
"step": 1374
|
9660 |
+
},
|
9661 |
+
{
|
9662 |
+
"epoch": 0.09759559932570314,
|
9663 |
+
"grad_norm": 0.38993802666664124,
|
9664 |
+
"learning_rate": 2.3732218959349185e-09,
|
9665 |
+
"loss": 0.6505,
|
9666 |
+
"step": 1375
|
9667 |
+
},
|
9668 |
+
{
|
9669 |
+
"epoch": 0.09766657794339455,
|
9670 |
+
"grad_norm": 0.6296946406364441,
|
9671 |
+
"learning_rate": 1.0547676048688892e-09,
|
9672 |
+
"loss": 0.8125,
|
9673 |
+
"step": 1376
|
9674 |
+
},
|
9675 |
+
{
|
9676 |
+
"epoch": 0.09773755656108597,
|
9677 |
+
"grad_norm": 0.6637731194496155,
|
9678 |
+
"learning_rate": 2.6369224888078693e-10,
|
9679 |
+
"loss": 0.8864,
|
9680 |
+
"step": 1377
|
9681 |
+
},
|
9682 |
+
{
|
9683 |
+
"epoch": 0.0978085351787774,
|
9684 |
+
"grad_norm": 0.5165863037109375,
|
9685 |
+
"learning_rate": 0.0,
|
9686 |
+
"loss": 1.3353,
|
9687 |
+
"step": 1378
|
9688 |
}
|
9689 |
],
|
9690 |
"logging_steps": 1,
|
|
|
9699 |
"should_evaluate": false,
|
9700 |
"should_log": false,
|
9701 |
"should_save": true,
|
9702 |
+
"should_training_stop": true
|
9703 |
},
|
9704 |
"attributes": {}
|
9705 |
}
|
9706 |
},
|
9707 |
+
"total_flos": 4.028988515155968e+17,
|
9708 |
"train_batch_size": 2,
|
9709 |
"trial_name": null,
|
9710 |
"trial_params": null
|