Training in progress, step 1086, 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 83945296
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:88370438c3fb2fecf65fd750af4a03bf5f63eb1354b5481793ee4f9aeb92bd06
|
3 |
size 83945296
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 43123028
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9a3990a4c3e5ea0f10f28a5d04aa62ecb13d73ed3ed8c2181400ffff52acc448
|
3 |
size 43123028
|
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 sha256:41e93eafb8f356ab0d33451ccddeebc13394ebcc53d8a0bb20320cc9b3964fd6
|
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:d45b5f7992d5087f5d2cf4cd5a994915b8d54450a1e6c7da016ce466b6e8c4a1
|
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": 362,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -5099,6 +5099,2548 @@
|
|
5099 |
"eval_samples_per_second": 29.757,
|
5100 |
"eval_steps_per_second": 14.899,
|
5101 |
"step": 724
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5102 |
}
|
5103 |
],
|
5104 |
"logging_steps": 1,
|
@@ -5118,7 +7660,7 @@
|
|
5118 |
"attributes": {}
|
5119 |
}
|
5120 |
},
|
5121 |
-
"total_flos":
|
5122 |
"train_batch_size": 2,
|
5123 |
"trial_name": null,
|
5124 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.629109341057205,
|
5 |
"eval_steps": 362,
|
6 |
+
"global_step": 1086,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
5099 |
"eval_samples_per_second": 29.757,
|
5100 |
"eval_steps_per_second": 14.899,
|
5101 |
"step": 724
|
5102 |
+
},
|
5103 |
+
{
|
5104 |
+
"epoch": 0.41998551774076753,
|
5105 |
+
"grad_norm": 4.330077648162842,
|
5106 |
+
"learning_rate": 0.00010087386722348325,
|
5107 |
+
"loss": 0.6861,
|
5108 |
+
"step": 725
|
5109 |
+
},
|
5110 |
+
{
|
5111 |
+
"epoch": 0.4205648081100652,
|
5112 |
+
"grad_norm": 1.6572608947753906,
|
5113 |
+
"learning_rate": 0.00010065540406715483,
|
5114 |
+
"loss": 0.1448,
|
5115 |
+
"step": 726
|
5116 |
+
},
|
5117 |
+
{
|
5118 |
+
"epoch": 0.4211440984793628,
|
5119 |
+
"grad_norm": 1.9755557775497437,
|
5120 |
+
"learning_rate": 0.00010043693778265219,
|
5121 |
+
"loss": 0.2074,
|
5122 |
+
"step": 727
|
5123 |
+
},
|
5124 |
+
{
|
5125 |
+
"epoch": 0.4217233888486604,
|
5126 |
+
"grad_norm": 1.481980800628662,
|
5127 |
+
"learning_rate": 0.00010021846941269179,
|
5128 |
+
"loss": 0.1057,
|
5129 |
+
"step": 728
|
5130 |
+
},
|
5131 |
+
{
|
5132 |
+
"epoch": 0.422302679217958,
|
5133 |
+
"grad_norm": 1.5745717287063599,
|
5134 |
+
"learning_rate": 0.0001,
|
5135 |
+
"loss": 0.1283,
|
5136 |
+
"step": 729
|
5137 |
+
},
|
5138 |
+
{
|
5139 |
+
"epoch": 0.4228819695872556,
|
5140 |
+
"grad_norm": 2.2418360710144043,
|
5141 |
+
"learning_rate": 9.978153058730823e-05,
|
5142 |
+
"loss": 0.2794,
|
5143 |
+
"step": 730
|
5144 |
+
},
|
5145 |
+
{
|
5146 |
+
"epoch": 0.4234612599565532,
|
5147 |
+
"grad_norm": 2.879293441772461,
|
5148 |
+
"learning_rate": 9.956306221734782e-05,
|
5149 |
+
"loss": 0.1824,
|
5150 |
+
"step": 731
|
5151 |
+
},
|
5152 |
+
{
|
5153 |
+
"epoch": 0.4240405503258508,
|
5154 |
+
"grad_norm": 2.3594748973846436,
|
5155 |
+
"learning_rate": 9.934459593284517e-05,
|
5156 |
+
"loss": 0.2136,
|
5157 |
+
"step": 732
|
5158 |
+
},
|
5159 |
+
{
|
5160 |
+
"epoch": 0.42461984069514846,
|
5161 |
+
"grad_norm": 2.092688798904419,
|
5162 |
+
"learning_rate": 9.912613277651676e-05,
|
5163 |
+
"loss": 0.1288,
|
5164 |
+
"step": 733
|
5165 |
+
},
|
5166 |
+
{
|
5167 |
+
"epoch": 0.42519913106444607,
|
5168 |
+
"grad_norm": 3.732053756713867,
|
5169 |
+
"learning_rate": 9.89076737910641e-05,
|
5170 |
+
"loss": 0.4913,
|
5171 |
+
"step": 734
|
5172 |
+
},
|
5173 |
+
{
|
5174 |
+
"epoch": 0.42577842143374367,
|
5175 |
+
"grad_norm": 2.177419900894165,
|
5176 |
+
"learning_rate": 9.868922001916877e-05,
|
5177 |
+
"loss": 0.2712,
|
5178 |
+
"step": 735
|
5179 |
+
},
|
5180 |
+
{
|
5181 |
+
"epoch": 0.4263577118030413,
|
5182 |
+
"grad_norm": 1.897555947303772,
|
5183 |
+
"learning_rate": 9.847077250348749e-05,
|
5184 |
+
"loss": 0.267,
|
5185 |
+
"step": 736
|
5186 |
+
},
|
5187 |
+
{
|
5188 |
+
"epoch": 0.4269370021723389,
|
5189 |
+
"grad_norm": 1.8367071151733398,
|
5190 |
+
"learning_rate": 9.825233228664715e-05,
|
5191 |
+
"loss": 0.0813,
|
5192 |
+
"step": 737
|
5193 |
+
},
|
5194 |
+
{
|
5195 |
+
"epoch": 0.4275162925416365,
|
5196 |
+
"grad_norm": 2.8897194862365723,
|
5197 |
+
"learning_rate": 9.803390041123979e-05,
|
5198 |
+
"loss": 0.4995,
|
5199 |
+
"step": 738
|
5200 |
+
},
|
5201 |
+
{
|
5202 |
+
"epoch": 0.4280955829109341,
|
5203 |
+
"grad_norm": 3.664484739303589,
|
5204 |
+
"learning_rate": 9.781547791981758e-05,
|
5205 |
+
"loss": 1.0554,
|
5206 |
+
"step": 739
|
5207 |
+
},
|
5208 |
+
{
|
5209 |
+
"epoch": 0.42867487328023174,
|
5210 |
+
"grad_norm": 3.2397992610931396,
|
5211 |
+
"learning_rate": 9.759706585488797e-05,
|
5212 |
+
"loss": 0.5773,
|
5213 |
+
"step": 740
|
5214 |
+
},
|
5215 |
+
{
|
5216 |
+
"epoch": 0.42925416364952934,
|
5217 |
+
"grad_norm": 1.5391981601715088,
|
5218 |
+
"learning_rate": 9.737866525890863e-05,
|
5219 |
+
"loss": 0.2097,
|
5220 |
+
"step": 741
|
5221 |
+
},
|
5222 |
+
{
|
5223 |
+
"epoch": 0.42983345401882694,
|
5224 |
+
"grad_norm": 1.8281755447387695,
|
5225 |
+
"learning_rate": 9.716027717428243e-05,
|
5226 |
+
"loss": 0.2781,
|
5227 |
+
"step": 742
|
5228 |
+
},
|
5229 |
+
{
|
5230 |
+
"epoch": 0.43041274438812455,
|
5231 |
+
"grad_norm": 3.814570188522339,
|
5232 |
+
"learning_rate": 9.694190264335266e-05,
|
5233 |
+
"loss": 0.4264,
|
5234 |
+
"step": 743
|
5235 |
+
},
|
5236 |
+
{
|
5237 |
+
"epoch": 0.43099203475742215,
|
5238 |
+
"grad_norm": 1.9860742092132568,
|
5239 |
+
"learning_rate": 9.672354270839778e-05,
|
5240 |
+
"loss": 0.2837,
|
5241 |
+
"step": 744
|
5242 |
+
},
|
5243 |
+
{
|
5244 |
+
"epoch": 0.43157132512671975,
|
5245 |
+
"grad_norm": 2.617427110671997,
|
5246 |
+
"learning_rate": 9.650519841162666e-05,
|
5247 |
+
"loss": 0.4266,
|
5248 |
+
"step": 745
|
5249 |
+
},
|
5250 |
+
{
|
5251 |
+
"epoch": 0.43215061549601735,
|
5252 |
+
"grad_norm": 2.1865363121032715,
|
5253 |
+
"learning_rate": 9.628687079517348e-05,
|
5254 |
+
"loss": 0.3272,
|
5255 |
+
"step": 746
|
5256 |
+
},
|
5257 |
+
{
|
5258 |
+
"epoch": 0.432729905865315,
|
5259 |
+
"grad_norm": 2.159600257873535,
|
5260 |
+
"learning_rate": 9.60685609010929e-05,
|
5261 |
+
"loss": 0.1331,
|
5262 |
+
"step": 747
|
5263 |
+
},
|
5264 |
+
{
|
5265 |
+
"epoch": 0.4333091962346126,
|
5266 |
+
"grad_norm": 1.6754506826400757,
|
5267 |
+
"learning_rate": 9.585026977135487e-05,
|
5268 |
+
"loss": 0.1101,
|
5269 |
+
"step": 748
|
5270 |
+
},
|
5271 |
+
{
|
5272 |
+
"epoch": 0.4338884866039102,
|
5273 |
+
"grad_norm": 1.5430821180343628,
|
5274 |
+
"learning_rate": 9.563199844783987e-05,
|
5275 |
+
"loss": 0.1776,
|
5276 |
+
"step": 749
|
5277 |
+
},
|
5278 |
+
{
|
5279 |
+
"epoch": 0.4344677769732078,
|
5280 |
+
"grad_norm": 2.8259193897247314,
|
5281 |
+
"learning_rate": 9.541374797233381e-05,
|
5282 |
+
"loss": 0.4059,
|
5283 |
+
"step": 750
|
5284 |
+
},
|
5285 |
+
{
|
5286 |
+
"epoch": 0.4350470673425054,
|
5287 |
+
"grad_norm": 2.681009292602539,
|
5288 |
+
"learning_rate": 9.519551938652307e-05,
|
5289 |
+
"loss": 0.3884,
|
5290 |
+
"step": 751
|
5291 |
+
},
|
5292 |
+
{
|
5293 |
+
"epoch": 0.435626357711803,
|
5294 |
+
"grad_norm": 1.899491310119629,
|
5295 |
+
"learning_rate": 9.497731373198966e-05,
|
5296 |
+
"loss": 0.2038,
|
5297 |
+
"step": 752
|
5298 |
+
},
|
5299 |
+
{
|
5300 |
+
"epoch": 0.4362056480811006,
|
5301 |
+
"grad_norm": 3.2308342456817627,
|
5302 |
+
"learning_rate": 9.475913205020601e-05,
|
5303 |
+
"loss": 0.3621,
|
5304 |
+
"step": 753
|
5305 |
+
},
|
5306 |
+
{
|
5307 |
+
"epoch": 0.4367849384503983,
|
5308 |
+
"grad_norm": 3.4386329650878906,
|
5309 |
+
"learning_rate": 9.45409753825302e-05,
|
5310 |
+
"loss": 0.3119,
|
5311 |
+
"step": 754
|
5312 |
+
},
|
5313 |
+
{
|
5314 |
+
"epoch": 0.4373642288196959,
|
5315 |
+
"grad_norm": 2.312267303466797,
|
5316 |
+
"learning_rate": 9.432284477020086e-05,
|
5317 |
+
"loss": 0.2877,
|
5318 |
+
"step": 755
|
5319 |
+
},
|
5320 |
+
{
|
5321 |
+
"epoch": 0.4379435191889935,
|
5322 |
+
"grad_norm": 0.9429482817649841,
|
5323 |
+
"learning_rate": 9.41047412543324e-05,
|
5324 |
+
"loss": 0.0736,
|
5325 |
+
"step": 756
|
5326 |
+
},
|
5327 |
+
{
|
5328 |
+
"epoch": 0.4385228095582911,
|
5329 |
+
"grad_norm": 2.079658269882202,
|
5330 |
+
"learning_rate": 9.388666587590977e-05,
|
5331 |
+
"loss": 0.2138,
|
5332 |
+
"step": 757
|
5333 |
+
},
|
5334 |
+
{
|
5335 |
+
"epoch": 0.4391020999275887,
|
5336 |
+
"grad_norm": 2.0186920166015625,
|
5337 |
+
"learning_rate": 9.366861967578364e-05,
|
5338 |
+
"loss": 0.2668,
|
5339 |
+
"step": 758
|
5340 |
+
},
|
5341 |
+
{
|
5342 |
+
"epoch": 0.4396813902968863,
|
5343 |
+
"grad_norm": 1.2216030359268188,
|
5344 |
+
"learning_rate": 9.345060369466543e-05,
|
5345 |
+
"loss": 0.0946,
|
5346 |
+
"step": 759
|
5347 |
+
},
|
5348 |
+
{
|
5349 |
+
"epoch": 0.4402606806661839,
|
5350 |
+
"grad_norm": 2.7688989639282227,
|
5351 |
+
"learning_rate": 9.323261897312238e-05,
|
5352 |
+
"loss": 0.2123,
|
5353 |
+
"step": 760
|
5354 |
+
},
|
5355 |
+
{
|
5356 |
+
"epoch": 0.44083997103548156,
|
5357 |
+
"grad_norm": 2.861898422241211,
|
5358 |
+
"learning_rate": 9.301466655157247e-05,
|
5359 |
+
"loss": 0.4286,
|
5360 |
+
"step": 761
|
5361 |
+
},
|
5362 |
+
{
|
5363 |
+
"epoch": 0.44141926140477916,
|
5364 |
+
"grad_norm": 1.6773964166641235,
|
5365 |
+
"learning_rate": 9.279674747027952e-05,
|
5366 |
+
"loss": 0.2269,
|
5367 |
+
"step": 762
|
5368 |
+
},
|
5369 |
+
{
|
5370 |
+
"epoch": 0.44199855177407676,
|
5371 |
+
"grad_norm": 2.697532892227173,
|
5372 |
+
"learning_rate": 9.257886276934821e-05,
|
5373 |
+
"loss": 0.3912,
|
5374 |
+
"step": 763
|
5375 |
+
},
|
5376 |
+
{
|
5377 |
+
"epoch": 0.44257784214337437,
|
5378 |
+
"grad_norm": 2.0737850666046143,
|
5379 |
+
"learning_rate": 9.236101348871921e-05,
|
5380 |
+
"loss": 0.249,
|
5381 |
+
"step": 764
|
5382 |
+
},
|
5383 |
+
{
|
5384 |
+
"epoch": 0.44315713251267197,
|
5385 |
+
"grad_norm": 2.5787010192871094,
|
5386 |
+
"learning_rate": 9.214320066816403e-05,
|
5387 |
+
"loss": 0.3523,
|
5388 |
+
"step": 765
|
5389 |
+
},
|
5390 |
+
{
|
5391 |
+
"epoch": 0.44373642288196957,
|
5392 |
+
"grad_norm": 2.724083662033081,
|
5393 |
+
"learning_rate": 9.192542534728021e-05,
|
5394 |
+
"loss": 0.24,
|
5395 |
+
"step": 766
|
5396 |
+
},
|
5397 |
+
{
|
5398 |
+
"epoch": 0.4443157132512672,
|
5399 |
+
"grad_norm": 2.9584672451019287,
|
5400 |
+
"learning_rate": 9.170768856548627e-05,
|
5401 |
+
"loss": 0.2859,
|
5402 |
+
"step": 767
|
5403 |
+
},
|
5404 |
+
{
|
5405 |
+
"epoch": 0.44489500362056483,
|
5406 |
+
"grad_norm": 4.675621509552002,
|
5407 |
+
"learning_rate": 9.148999136201686e-05,
|
5408 |
+
"loss": 0.9754,
|
5409 |
+
"step": 768
|
5410 |
+
},
|
5411 |
+
{
|
5412 |
+
"epoch": 0.44547429398986244,
|
5413 |
+
"grad_norm": 2.282618522644043,
|
5414 |
+
"learning_rate": 9.127233477591766e-05,
|
5415 |
+
"loss": 0.1427,
|
5416 |
+
"step": 769
|
5417 |
+
},
|
5418 |
+
{
|
5419 |
+
"epoch": 0.44605358435916004,
|
5420 |
+
"grad_norm": 2.7693891525268555,
|
5421 |
+
"learning_rate": 9.105471984604055e-05,
|
5422 |
+
"loss": 0.2272,
|
5423 |
+
"step": 770
|
5424 |
+
},
|
5425 |
+
{
|
5426 |
+
"epoch": 0.44663287472845764,
|
5427 |
+
"grad_norm": 2.5014052391052246,
|
5428 |
+
"learning_rate": 9.08371476110385e-05,
|
5429 |
+
"loss": 0.2995,
|
5430 |
+
"step": 771
|
5431 |
+
},
|
5432 |
+
{
|
5433 |
+
"epoch": 0.44721216509775524,
|
5434 |
+
"grad_norm": 3.3536150455474854,
|
5435 |
+
"learning_rate": 9.061961910936078e-05,
|
5436 |
+
"loss": 1.0005,
|
5437 |
+
"step": 772
|
5438 |
+
},
|
5439 |
+
{
|
5440 |
+
"epoch": 0.44779145546705285,
|
5441 |
+
"grad_norm": 2.182767629623413,
|
5442 |
+
"learning_rate": 9.040213537924788e-05,
|
5443 |
+
"loss": 0.2987,
|
5444 |
+
"step": 773
|
5445 |
+
},
|
5446 |
+
{
|
5447 |
+
"epoch": 0.44837074583635045,
|
5448 |
+
"grad_norm": 3.0148394107818604,
|
5449 |
+
"learning_rate": 9.018469745872661e-05,
|
5450 |
+
"loss": 0.3368,
|
5451 |
+
"step": 774
|
5452 |
+
},
|
5453 |
+
{
|
5454 |
+
"epoch": 0.4489500362056481,
|
5455 |
+
"grad_norm": 3.7020537853240967,
|
5456 |
+
"learning_rate": 8.996730638560519e-05,
|
5457 |
+
"loss": 0.6397,
|
5458 |
+
"step": 775
|
5459 |
+
},
|
5460 |
+
{
|
5461 |
+
"epoch": 0.4495293265749457,
|
5462 |
+
"grad_norm": 1.8903841972351074,
|
5463 |
+
"learning_rate": 8.974996319746814e-05,
|
5464 |
+
"loss": 0.1667,
|
5465 |
+
"step": 776
|
5466 |
+
},
|
5467 |
+
{
|
5468 |
+
"epoch": 0.4501086169442433,
|
5469 |
+
"grad_norm": 2.55179762840271,
|
5470 |
+
"learning_rate": 8.953266893167153e-05,
|
5471 |
+
"loss": 0.4118,
|
5472 |
+
"step": 777
|
5473 |
+
},
|
5474 |
+
{
|
5475 |
+
"epoch": 0.4506879073135409,
|
5476 |
+
"grad_norm": 2.1008896827697754,
|
5477 |
+
"learning_rate": 8.931542462533783e-05,
|
5478 |
+
"loss": 0.3604,
|
5479 |
+
"step": 778
|
5480 |
+
},
|
5481 |
+
{
|
5482 |
+
"epoch": 0.4512671976828385,
|
5483 |
+
"grad_norm": 1.8134448528289795,
|
5484 |
+
"learning_rate": 8.909823131535117e-05,
|
5485 |
+
"loss": 0.2495,
|
5486 |
+
"step": 779
|
5487 |
+
},
|
5488 |
+
{
|
5489 |
+
"epoch": 0.4518464880521361,
|
5490 |
+
"grad_norm": 1.7121368646621704,
|
5491 |
+
"learning_rate": 8.88810900383522e-05,
|
5492 |
+
"loss": 0.265,
|
5493 |
+
"step": 780
|
5494 |
+
},
|
5495 |
+
{
|
5496 |
+
"epoch": 0.4524257784214337,
|
5497 |
+
"grad_norm": 2.0966012477874756,
|
5498 |
+
"learning_rate": 8.866400183073323e-05,
|
5499 |
+
"loss": 0.1536,
|
5500 |
+
"step": 781
|
5501 |
+
},
|
5502 |
+
{
|
5503 |
+
"epoch": 0.4530050687907314,
|
5504 |
+
"grad_norm": 2.6180903911590576,
|
5505 |
+
"learning_rate": 8.84469677286333e-05,
|
5506 |
+
"loss": 0.2901,
|
5507 |
+
"step": 782
|
5508 |
+
},
|
5509 |
+
{
|
5510 |
+
"epoch": 0.453584359160029,
|
5511 |
+
"grad_norm": 1.5925201177597046,
|
5512 |
+
"learning_rate": 8.822998876793322e-05,
|
5513 |
+
"loss": 0.1286,
|
5514 |
+
"step": 783
|
5515 |
+
},
|
5516 |
+
{
|
5517 |
+
"epoch": 0.4541636495293266,
|
5518 |
+
"grad_norm": 2.3173744678497314,
|
5519 |
+
"learning_rate": 8.801306598425059e-05,
|
5520 |
+
"loss": 0.1561,
|
5521 |
+
"step": 784
|
5522 |
+
},
|
5523 |
+
{
|
5524 |
+
"epoch": 0.4547429398986242,
|
5525 |
+
"grad_norm": 3.445146322250366,
|
5526 |
+
"learning_rate": 8.779620041293486e-05,
|
5527 |
+
"loss": 0.4528,
|
5528 |
+
"step": 785
|
5529 |
+
},
|
5530 |
+
{
|
5531 |
+
"epoch": 0.4553222302679218,
|
5532 |
+
"grad_norm": 3.779384136199951,
|
5533 |
+
"learning_rate": 8.757939308906248e-05,
|
5534 |
+
"loss": 0.2582,
|
5535 |
+
"step": 786
|
5536 |
+
},
|
5537 |
+
{
|
5538 |
+
"epoch": 0.4559015206372194,
|
5539 |
+
"grad_norm": 2.1341681480407715,
|
5540 |
+
"learning_rate": 8.73626450474318e-05,
|
5541 |
+
"loss": 0.2596,
|
5542 |
+
"step": 787
|
5543 |
+
},
|
5544 |
+
{
|
5545 |
+
"epoch": 0.456480811006517,
|
5546 |
+
"grad_norm": 1.8372859954833984,
|
5547 |
+
"learning_rate": 8.714595732255834e-05,
|
5548 |
+
"loss": 0.1585,
|
5549 |
+
"step": 788
|
5550 |
+
},
|
5551 |
+
{
|
5552 |
+
"epoch": 0.45706010137581465,
|
5553 |
+
"grad_norm": 1.5721983909606934,
|
5554 |
+
"learning_rate": 8.692933094866963e-05,
|
5555 |
+
"loss": 0.1551,
|
5556 |
+
"step": 789
|
5557 |
+
},
|
5558 |
+
{
|
5559 |
+
"epoch": 0.45763939174511226,
|
5560 |
+
"grad_norm": 2.166860580444336,
|
5561 |
+
"learning_rate": 8.671276695970043e-05,
|
5562 |
+
"loss": 0.3124,
|
5563 |
+
"step": 790
|
5564 |
+
},
|
5565 |
+
{
|
5566 |
+
"epoch": 0.45821868211440986,
|
5567 |
+
"grad_norm": 1.0890613794326782,
|
5568 |
+
"learning_rate": 8.649626638928766e-05,
|
5569 |
+
"loss": 0.0698,
|
5570 |
+
"step": 791
|
5571 |
+
},
|
5572 |
+
{
|
5573 |
+
"epoch": 0.45879797248370746,
|
5574 |
+
"grad_norm": 2.0683658123016357,
|
5575 |
+
"learning_rate": 8.627983027076573e-05,
|
5576 |
+
"loss": 0.3974,
|
5577 |
+
"step": 792
|
5578 |
+
},
|
5579 |
+
{
|
5580 |
+
"epoch": 0.45937726285300506,
|
5581 |
+
"grad_norm": 2.5842177867889404,
|
5582 |
+
"learning_rate": 8.606345963716126e-05,
|
5583 |
+
"loss": 0.2537,
|
5584 |
+
"step": 793
|
5585 |
+
},
|
5586 |
+
{
|
5587 |
+
"epoch": 0.45995655322230267,
|
5588 |
+
"grad_norm": 2.5562381744384766,
|
5589 |
+
"learning_rate": 8.584715552118842e-05,
|
5590 |
+
"loss": 0.5266,
|
5591 |
+
"step": 794
|
5592 |
+
},
|
5593 |
+
{
|
5594 |
+
"epoch": 0.46053584359160027,
|
5595 |
+
"grad_norm": 3.210618734359741,
|
5596 |
+
"learning_rate": 8.56309189552438e-05,
|
5597 |
+
"loss": 0.2998,
|
5598 |
+
"step": 795
|
5599 |
+
},
|
5600 |
+
{
|
5601 |
+
"epoch": 0.4611151339608979,
|
5602 |
+
"grad_norm": 1.4088205099105835,
|
5603 |
+
"learning_rate": 8.541475097140167e-05,
|
5604 |
+
"loss": 0.0904,
|
5605 |
+
"step": 796
|
5606 |
+
},
|
5607 |
+
{
|
5608 |
+
"epoch": 0.46169442433019553,
|
5609 |
+
"grad_norm": 1.8730751276016235,
|
5610 |
+
"learning_rate": 8.519865260140889e-05,
|
5611 |
+
"loss": 0.3828,
|
5612 |
+
"step": 797
|
5613 |
+
},
|
5614 |
+
{
|
5615 |
+
"epoch": 0.46227371469949313,
|
5616 |
+
"grad_norm": 2.921869993209839,
|
5617 |
+
"learning_rate": 8.498262487668015e-05,
|
5618 |
+
"loss": 0.223,
|
5619 |
+
"step": 798
|
5620 |
+
},
|
5621 |
+
{
|
5622 |
+
"epoch": 0.46285300506879073,
|
5623 |
+
"grad_norm": 2.3525073528289795,
|
5624 |
+
"learning_rate": 8.476666882829287e-05,
|
5625 |
+
"loss": 0.1754,
|
5626 |
+
"step": 799
|
5627 |
+
},
|
5628 |
+
{
|
5629 |
+
"epoch": 0.46343229543808834,
|
5630 |
+
"grad_norm": 2.6007490158081055,
|
5631 |
+
"learning_rate": 8.455078548698243e-05,
|
5632 |
+
"loss": 0.263,
|
5633 |
+
"step": 800
|
5634 |
+
},
|
5635 |
+
{
|
5636 |
+
"epoch": 0.46401158580738594,
|
5637 |
+
"grad_norm": 3.7595996856689453,
|
5638 |
+
"learning_rate": 8.433497588313714e-05,
|
5639 |
+
"loss": 0.3769,
|
5640 |
+
"step": 801
|
5641 |
+
},
|
5642 |
+
{
|
5643 |
+
"epoch": 0.46459087617668354,
|
5644 |
+
"grad_norm": 3.7927472591400146,
|
5645 |
+
"learning_rate": 8.411924104679339e-05,
|
5646 |
+
"loss": 0.3231,
|
5647 |
+
"step": 802
|
5648 |
+
},
|
5649 |
+
{
|
5650 |
+
"epoch": 0.4651701665459812,
|
5651 |
+
"grad_norm": 2.7741410732269287,
|
5652 |
+
"learning_rate": 8.390358200763076e-05,
|
5653 |
+
"loss": 0.2961,
|
5654 |
+
"step": 803
|
5655 |
+
},
|
5656 |
+
{
|
5657 |
+
"epoch": 0.4657494569152788,
|
5658 |
+
"grad_norm": 2.110931158065796,
|
5659 |
+
"learning_rate": 8.368799979496694e-05,
|
5660 |
+
"loss": 0.122,
|
5661 |
+
"step": 804
|
5662 |
+
},
|
5663 |
+
{
|
5664 |
+
"epoch": 0.4663287472845764,
|
5665 |
+
"grad_norm": 1.9475789070129395,
|
5666 |
+
"learning_rate": 8.347249543775303e-05,
|
5667 |
+
"loss": 0.2273,
|
5668 |
+
"step": 805
|
5669 |
+
},
|
5670 |
+
{
|
5671 |
+
"epoch": 0.466908037653874,
|
5672 |
+
"grad_norm": 1.9266232252120972,
|
5673 |
+
"learning_rate": 8.325706996456854e-05,
|
5674 |
+
"loss": 0.124,
|
5675 |
+
"step": 806
|
5676 |
+
},
|
5677 |
+
{
|
5678 |
+
"epoch": 0.4674873280231716,
|
5679 |
+
"grad_norm": 2.5791711807250977,
|
5680 |
+
"learning_rate": 8.304172440361642e-05,
|
5681 |
+
"loss": 0.5077,
|
5682 |
+
"step": 807
|
5683 |
+
},
|
5684 |
+
{
|
5685 |
+
"epoch": 0.4680666183924692,
|
5686 |
+
"grad_norm": 2.490081548690796,
|
5687 |
+
"learning_rate": 8.282645978271823e-05,
|
5688 |
+
"loss": 0.246,
|
5689 |
+
"step": 808
|
5690 |
+
},
|
5691 |
+
{
|
5692 |
+
"epoch": 0.4686459087617668,
|
5693 |
+
"grad_norm": 1.7749154567718506,
|
5694 |
+
"learning_rate": 8.261127712930923e-05,
|
5695 |
+
"loss": 0.333,
|
5696 |
+
"step": 809
|
5697 |
+
},
|
5698 |
+
{
|
5699 |
+
"epoch": 0.4692251991310644,
|
5700 |
+
"grad_norm": 1.92875337600708,
|
5701 |
+
"learning_rate": 8.23961774704334e-05,
|
5702 |
+
"loss": 0.4749,
|
5703 |
+
"step": 810
|
5704 |
+
},
|
5705 |
+
{
|
5706 |
+
"epoch": 0.4698044895003621,
|
5707 |
+
"grad_norm": 1.28525972366333,
|
5708 |
+
"learning_rate": 8.21811618327387e-05,
|
5709 |
+
"loss": 0.0999,
|
5710 |
+
"step": 811
|
5711 |
+
},
|
5712 |
+
{
|
5713 |
+
"epoch": 0.4703837798696597,
|
5714 |
+
"grad_norm": 1.602352499961853,
|
5715 |
+
"learning_rate": 8.196623124247197e-05,
|
5716 |
+
"loss": 0.0745,
|
5717 |
+
"step": 812
|
5718 |
+
},
|
5719 |
+
{
|
5720 |
+
"epoch": 0.4709630702389573,
|
5721 |
+
"grad_norm": 2.0801961421966553,
|
5722 |
+
"learning_rate": 8.175138672547415e-05,
|
5723 |
+
"loss": 0.2347,
|
5724 |
+
"step": 813
|
5725 |
+
},
|
5726 |
+
{
|
5727 |
+
"epoch": 0.4715423606082549,
|
5728 |
+
"grad_norm": 2.410165786743164,
|
5729 |
+
"learning_rate": 8.153662930717538e-05,
|
5730 |
+
"loss": 0.1434,
|
5731 |
+
"step": 814
|
5732 |
+
},
|
5733 |
+
{
|
5734 |
+
"epoch": 0.4721216509775525,
|
5735 |
+
"grad_norm": 1.8650652170181274,
|
5736 |
+
"learning_rate": 8.132196001259011e-05,
|
5737 |
+
"loss": 0.1641,
|
5738 |
+
"step": 815
|
5739 |
+
},
|
5740 |
+
{
|
5741 |
+
"epoch": 0.4727009413468501,
|
5742 |
+
"grad_norm": 3.301302194595337,
|
5743 |
+
"learning_rate": 8.110737986631209e-05,
|
5744 |
+
"loss": 0.7476,
|
5745 |
+
"step": 816
|
5746 |
+
},
|
5747 |
+
{
|
5748 |
+
"epoch": 0.4732802317161477,
|
5749 |
+
"grad_norm": 2.4830920696258545,
|
5750 |
+
"learning_rate": 8.089288989250972e-05,
|
5751 |
+
"loss": 0.3828,
|
5752 |
+
"step": 817
|
5753 |
+
},
|
5754 |
+
{
|
5755 |
+
"epoch": 0.47385952208544535,
|
5756 |
+
"grad_norm": 3.1074888706207275,
|
5757 |
+
"learning_rate": 8.06784911149209e-05,
|
5758 |
+
"loss": 0.3676,
|
5759 |
+
"step": 818
|
5760 |
+
},
|
5761 |
+
{
|
5762 |
+
"epoch": 0.47443881245474295,
|
5763 |
+
"grad_norm": 1.4939327239990234,
|
5764 |
+
"learning_rate": 8.046418455684827e-05,
|
5765 |
+
"loss": 0.0815,
|
5766 |
+
"step": 819
|
5767 |
+
},
|
5768 |
+
{
|
5769 |
+
"epoch": 0.47501810282404056,
|
5770 |
+
"grad_norm": 2.0348119735717773,
|
5771 |
+
"learning_rate": 8.024997124115437e-05,
|
5772 |
+
"loss": 0.2416,
|
5773 |
+
"step": 820
|
5774 |
+
},
|
5775 |
+
{
|
5776 |
+
"epoch": 0.47559739319333816,
|
5777 |
+
"grad_norm": 1.92427659034729,
|
5778 |
+
"learning_rate": 8.003585219025666e-05,
|
5779 |
+
"loss": 0.0939,
|
5780 |
+
"step": 821
|
5781 |
+
},
|
5782 |
+
{
|
5783 |
+
"epoch": 0.47617668356263576,
|
5784 |
+
"grad_norm": 1.8372902870178223,
|
5785 |
+
"learning_rate": 7.982182842612274e-05,
|
5786 |
+
"loss": 0.1303,
|
5787 |
+
"step": 822
|
5788 |
+
},
|
5789 |
+
{
|
5790 |
+
"epoch": 0.47675597393193336,
|
5791 |
+
"grad_norm": 1.8634800910949707,
|
5792 |
+
"learning_rate": 7.960790097026531e-05,
|
5793 |
+
"loss": 0.0993,
|
5794 |
+
"step": 823
|
5795 |
+
},
|
5796 |
+
{
|
5797 |
+
"epoch": 0.47733526430123097,
|
5798 |
+
"grad_norm": 2.8979063034057617,
|
5799 |
+
"learning_rate": 7.939407084373751e-05,
|
5800 |
+
"loss": 0.1954,
|
5801 |
+
"step": 824
|
5802 |
+
},
|
5803 |
+
{
|
5804 |
+
"epoch": 0.4779145546705286,
|
5805 |
+
"grad_norm": 3.0793609619140625,
|
5806 |
+
"learning_rate": 7.91803390671279e-05,
|
5807 |
+
"loss": 0.3532,
|
5808 |
+
"step": 825
|
5809 |
+
},
|
5810 |
+
{
|
5811 |
+
"epoch": 0.4784938450398262,
|
5812 |
+
"grad_norm": 2.3782129287719727,
|
5813 |
+
"learning_rate": 7.89667066605556e-05,
|
5814 |
+
"loss": 0.1652,
|
5815 |
+
"step": 826
|
5816 |
+
},
|
5817 |
+
{
|
5818 |
+
"epoch": 0.47907313540912383,
|
5819 |
+
"grad_norm": 1.3281863927841187,
|
5820 |
+
"learning_rate": 7.875317464366548e-05,
|
5821 |
+
"loss": 0.1078,
|
5822 |
+
"step": 827
|
5823 |
+
},
|
5824 |
+
{
|
5825 |
+
"epoch": 0.47965242577842143,
|
5826 |
+
"grad_norm": 0.7168323397636414,
|
5827 |
+
"learning_rate": 7.853974403562323e-05,
|
5828 |
+
"loss": 0.0436,
|
5829 |
+
"step": 828
|
5830 |
+
},
|
5831 |
+
{
|
5832 |
+
"epoch": 0.48023171614771903,
|
5833 |
+
"grad_norm": 4.326037406921387,
|
5834 |
+
"learning_rate": 7.832641585511054e-05,
|
5835 |
+
"loss": 0.3731,
|
5836 |
+
"step": 829
|
5837 |
+
},
|
5838 |
+
{
|
5839 |
+
"epoch": 0.48081100651701664,
|
5840 |
+
"grad_norm": 2.2524211406707764,
|
5841 |
+
"learning_rate": 7.811319112032027e-05,
|
5842 |
+
"loss": 0.1654,
|
5843 |
+
"step": 830
|
5844 |
+
},
|
5845 |
+
{
|
5846 |
+
"epoch": 0.48139029688631424,
|
5847 |
+
"grad_norm": 1.8576236963272095,
|
5848 |
+
"learning_rate": 7.790007084895148e-05,
|
5849 |
+
"loss": 0.2108,
|
5850 |
+
"step": 831
|
5851 |
+
},
|
5852 |
+
{
|
5853 |
+
"epoch": 0.4819695872556119,
|
5854 |
+
"grad_norm": 1.5746654272079468,
|
5855 |
+
"learning_rate": 7.768705605820466e-05,
|
5856 |
+
"loss": 0.1205,
|
5857 |
+
"step": 832
|
5858 |
+
},
|
5859 |
+
{
|
5860 |
+
"epoch": 0.4825488776249095,
|
5861 |
+
"grad_norm": 3.4119744300842285,
|
5862 |
+
"learning_rate": 7.747414776477684e-05,
|
5863 |
+
"loss": 0.2162,
|
5864 |
+
"step": 833
|
5865 |
+
},
|
5866 |
+
{
|
5867 |
+
"epoch": 0.4831281679942071,
|
5868 |
+
"grad_norm": 2.5503008365631104,
|
5869 |
+
"learning_rate": 7.726134698485679e-05,
|
5870 |
+
"loss": 0.1854,
|
5871 |
+
"step": 834
|
5872 |
+
},
|
5873 |
+
{
|
5874 |
+
"epoch": 0.4837074583635047,
|
5875 |
+
"grad_norm": 2.870460033416748,
|
5876 |
+
"learning_rate": 7.704865473412008e-05,
|
5877 |
+
"loss": 0.4803,
|
5878 |
+
"step": 835
|
5879 |
+
},
|
5880 |
+
{
|
5881 |
+
"epoch": 0.4842867487328023,
|
5882 |
+
"grad_norm": 3.056009292602539,
|
5883 |
+
"learning_rate": 7.683607202772433e-05,
|
5884 |
+
"loss": 0.5905,
|
5885 |
+
"step": 836
|
5886 |
+
},
|
5887 |
+
{
|
5888 |
+
"epoch": 0.4848660391020999,
|
5889 |
+
"grad_norm": 1.2388354539871216,
|
5890 |
+
"learning_rate": 7.662359988030425e-05,
|
5891 |
+
"loss": 0.117,
|
5892 |
+
"step": 837
|
5893 |
+
},
|
5894 |
+
{
|
5895 |
+
"epoch": 0.4854453294713975,
|
5896 |
+
"grad_norm": 3.5667262077331543,
|
5897 |
+
"learning_rate": 7.641123930596693e-05,
|
5898 |
+
"loss": 1.0978,
|
5899 |
+
"step": 838
|
5900 |
+
},
|
5901 |
+
{
|
5902 |
+
"epoch": 0.48602461984069517,
|
5903 |
+
"grad_norm": 2.819366693496704,
|
5904 |
+
"learning_rate": 7.619899131828692e-05,
|
5905 |
+
"loss": 0.1469,
|
5906 |
+
"step": 839
|
5907 |
+
},
|
5908 |
+
{
|
5909 |
+
"epoch": 0.4866039102099928,
|
5910 |
+
"grad_norm": 2.320831537246704,
|
5911 |
+
"learning_rate": 7.598685693030136e-05,
|
5912 |
+
"loss": 0.3762,
|
5913 |
+
"step": 840
|
5914 |
+
},
|
5915 |
+
{
|
5916 |
+
"epoch": 0.4871832005792904,
|
5917 |
+
"grad_norm": 2.2693238258361816,
|
5918 |
+
"learning_rate": 7.57748371545052e-05,
|
5919 |
+
"loss": 0.1933,
|
5920 |
+
"step": 841
|
5921 |
+
},
|
5922 |
+
{
|
5923 |
+
"epoch": 0.487762490948588,
|
5924 |
+
"grad_norm": 2.3183560371398926,
|
5925 |
+
"learning_rate": 7.556293300284647e-05,
|
5926 |
+
"loss": 0.3893,
|
5927 |
+
"step": 842
|
5928 |
+
},
|
5929 |
+
{
|
5930 |
+
"epoch": 0.4883417813178856,
|
5931 |
+
"grad_norm": 2.0969479084014893,
|
5932 |
+
"learning_rate": 7.535114548672119e-05,
|
5933 |
+
"loss": 0.1605,
|
5934 |
+
"step": 843
|
5935 |
+
},
|
5936 |
+
{
|
5937 |
+
"epoch": 0.4889210716871832,
|
5938 |
+
"grad_norm": 2.917405366897583,
|
5939 |
+
"learning_rate": 7.513947561696877e-05,
|
5940 |
+
"loss": 0.5537,
|
5941 |
+
"step": 844
|
5942 |
+
},
|
5943 |
+
{
|
5944 |
+
"epoch": 0.4895003620564808,
|
5945 |
+
"grad_norm": 2.151308536529541,
|
5946 |
+
"learning_rate": 7.492792440386709e-05,
|
5947 |
+
"loss": 0.2524,
|
5948 |
+
"step": 845
|
5949 |
+
},
|
5950 |
+
{
|
5951 |
+
"epoch": 0.49007965242577844,
|
5952 |
+
"grad_norm": 1.4771618843078613,
|
5953 |
+
"learning_rate": 7.471649285712768e-05,
|
5954 |
+
"loss": 0.067,
|
5955 |
+
"step": 846
|
5956 |
+
},
|
5957 |
+
{
|
5958 |
+
"epoch": 0.49065894279507605,
|
5959 |
+
"grad_norm": 3.089982032775879,
|
5960 |
+
"learning_rate": 7.450518198589092e-05,
|
5961 |
+
"loss": 0.3612,
|
5962 |
+
"step": 847
|
5963 |
+
},
|
5964 |
+
{
|
5965 |
+
"epoch": 0.49123823316437365,
|
5966 |
+
"grad_norm": 1.8487430810928345,
|
5967 |
+
"learning_rate": 7.429399279872122e-05,
|
5968 |
+
"loss": 0.2427,
|
5969 |
+
"step": 848
|
5970 |
+
},
|
5971 |
+
{
|
5972 |
+
"epoch": 0.49181752353367125,
|
5973 |
+
"grad_norm": 1.7247542142868042,
|
5974 |
+
"learning_rate": 7.408292630360225e-05,
|
5975 |
+
"loss": 0.1942,
|
5976 |
+
"step": 849
|
5977 |
+
},
|
5978 |
+
{
|
5979 |
+
"epoch": 0.49239681390296886,
|
5980 |
+
"grad_norm": 2.6909584999084473,
|
5981 |
+
"learning_rate": 7.387198350793201e-05,
|
5982 |
+
"loss": 0.5139,
|
5983 |
+
"step": 850
|
5984 |
+
},
|
5985 |
+
{
|
5986 |
+
"epoch": 0.49297610427226646,
|
5987 |
+
"grad_norm": 2.2064802646636963,
|
5988 |
+
"learning_rate": 7.366116541851814e-05,
|
5989 |
+
"loss": 0.3613,
|
5990 |
+
"step": 851
|
5991 |
+
},
|
5992 |
+
{
|
5993 |
+
"epoch": 0.49355539464156406,
|
5994 |
+
"grad_norm": 3.2793684005737305,
|
5995 |
+
"learning_rate": 7.345047304157304e-05,
|
5996 |
+
"loss": 0.3872,
|
5997 |
+
"step": 852
|
5998 |
+
},
|
5999 |
+
{
|
6000 |
+
"epoch": 0.4941346850108617,
|
6001 |
+
"grad_norm": 2.7883176803588867,
|
6002 |
+
"learning_rate": 7.323990738270917e-05,
|
6003 |
+
"loss": 0.2974,
|
6004 |
+
"step": 853
|
6005 |
+
},
|
6006 |
+
{
|
6007 |
+
"epoch": 0.4947139753801593,
|
6008 |
+
"grad_norm": 3.370043992996216,
|
6009 |
+
"learning_rate": 7.302946944693408e-05,
|
6010 |
+
"loss": 0.4254,
|
6011 |
+
"step": 854
|
6012 |
+
},
|
6013 |
+
{
|
6014 |
+
"epoch": 0.4952932657494569,
|
6015 |
+
"grad_norm": 2.4159679412841797,
|
6016 |
+
"learning_rate": 7.281916023864577e-05,
|
6017 |
+
"loss": 0.2242,
|
6018 |
+
"step": 855
|
6019 |
+
},
|
6020 |
+
{
|
6021 |
+
"epoch": 0.4958725561187545,
|
6022 |
+
"grad_norm": 1.912782907485962,
|
6023 |
+
"learning_rate": 7.260898076162782e-05,
|
6024 |
+
"loss": 0.1405,
|
6025 |
+
"step": 856
|
6026 |
+
},
|
6027 |
+
{
|
6028 |
+
"epoch": 0.49645184648805213,
|
6029 |
+
"grad_norm": 2.1396970748901367,
|
6030 |
+
"learning_rate": 7.239893201904462e-05,
|
6031 |
+
"loss": 0.1709,
|
6032 |
+
"step": 857
|
6033 |
+
},
|
6034 |
+
{
|
6035 |
+
"epoch": 0.49703113685734973,
|
6036 |
+
"grad_norm": 2.5621755123138428,
|
6037 |
+
"learning_rate": 7.218901501343658e-05,
|
6038 |
+
"loss": 0.4716,
|
6039 |
+
"step": 858
|
6040 |
+
},
|
6041 |
+
{
|
6042 |
+
"epoch": 0.49761042722664733,
|
6043 |
+
"grad_norm": 1.5708820819854736,
|
6044 |
+
"learning_rate": 7.197923074671536e-05,
|
6045 |
+
"loss": 0.1109,
|
6046 |
+
"step": 859
|
6047 |
+
},
|
6048 |
+
{
|
6049 |
+
"epoch": 0.498189717595945,
|
6050 |
+
"grad_norm": 3.326842784881592,
|
6051 |
+
"learning_rate": 7.176958022015902e-05,
|
6052 |
+
"loss": 0.1665,
|
6053 |
+
"step": 860
|
6054 |
+
},
|
6055 |
+
{
|
6056 |
+
"epoch": 0.4987690079652426,
|
6057 |
+
"grad_norm": 2.878673553466797,
|
6058 |
+
"learning_rate": 7.156006443440737e-05,
|
6059 |
+
"loss": 0.3088,
|
6060 |
+
"step": 861
|
6061 |
+
},
|
6062 |
+
{
|
6063 |
+
"epoch": 0.4993482983345402,
|
6064 |
+
"grad_norm": 1.5682700872421265,
|
6065 |
+
"learning_rate": 7.135068438945704e-05,
|
6066 |
+
"loss": 0.1415,
|
6067 |
+
"step": 862
|
6068 |
+
},
|
6069 |
+
{
|
6070 |
+
"epoch": 0.4999275887038378,
|
6071 |
+
"grad_norm": 2.8705825805664062,
|
6072 |
+
"learning_rate": 7.114144108465684e-05,
|
6073 |
+
"loss": 0.2013,
|
6074 |
+
"step": 863
|
6075 |
+
},
|
6076 |
+
{
|
6077 |
+
"epoch": 0.5005068790731354,
|
6078 |
+
"grad_norm": 0.8864466547966003,
|
6079 |
+
"learning_rate": 7.093233551870287e-05,
|
6080 |
+
"loss": 0.0533,
|
6081 |
+
"step": 864
|
6082 |
+
},
|
6083 |
+
{
|
6084 |
+
"epoch": 0.501086169442433,
|
6085 |
+
"grad_norm": 1.5986967086791992,
|
6086 |
+
"learning_rate": 7.072336868963387e-05,
|
6087 |
+
"loss": 0.1187,
|
6088 |
+
"step": 865
|
6089 |
+
},
|
6090 |
+
{
|
6091 |
+
"epoch": 0.5016654598117306,
|
6092 |
+
"grad_norm": 1.9611972570419312,
|
6093 |
+
"learning_rate": 7.05145415948264e-05,
|
6094 |
+
"loss": 0.1081,
|
6095 |
+
"step": 866
|
6096 |
+
},
|
6097 |
+
{
|
6098 |
+
"epoch": 0.5022447501810282,
|
6099 |
+
"grad_norm": 2.0408990383148193,
|
6100 |
+
"learning_rate": 7.03058552309901e-05,
|
6101 |
+
"loss": 0.786,
|
6102 |
+
"step": 867
|
6103 |
+
},
|
6104 |
+
{
|
6105 |
+
"epoch": 0.5028240405503258,
|
6106 |
+
"grad_norm": 2.0567972660064697,
|
6107 |
+
"learning_rate": 7.009731059416287e-05,
|
6108 |
+
"loss": 0.1873,
|
6109 |
+
"step": 868
|
6110 |
+
},
|
6111 |
+
{
|
6112 |
+
"epoch": 0.5034033309196234,
|
6113 |
+
"grad_norm": 2.1673219203948975,
|
6114 |
+
"learning_rate": 6.988890867970617e-05,
|
6115 |
+
"loss": 0.1091,
|
6116 |
+
"step": 869
|
6117 |
+
},
|
6118 |
+
{
|
6119 |
+
"epoch": 0.503982621288921,
|
6120 |
+
"grad_norm": 2.295874834060669,
|
6121 |
+
"learning_rate": 6.968065048230028e-05,
|
6122 |
+
"loss": 0.2526,
|
6123 |
+
"step": 870
|
6124 |
+
},
|
6125 |
+
{
|
6126 |
+
"epoch": 0.5045619116582187,
|
6127 |
+
"grad_norm": 1.535244107246399,
|
6128 |
+
"learning_rate": 6.947253699593951e-05,
|
6129 |
+
"loss": 0.1756,
|
6130 |
+
"step": 871
|
6131 |
+
},
|
6132 |
+
{
|
6133 |
+
"epoch": 0.5051412020275163,
|
6134 |
+
"grad_norm": 1.3690143823623657,
|
6135 |
+
"learning_rate": 6.926456921392753e-05,
|
6136 |
+
"loss": 0.0391,
|
6137 |
+
"step": 872
|
6138 |
+
},
|
6139 |
+
{
|
6140 |
+
"epoch": 0.5057204923968139,
|
6141 |
+
"grad_norm": 1.4958258867263794,
|
6142 |
+
"learning_rate": 6.905674812887253e-05,
|
6143 |
+
"loss": 0.0865,
|
6144 |
+
"step": 873
|
6145 |
+
},
|
6146 |
+
{
|
6147 |
+
"epoch": 0.5062997827661115,
|
6148 |
+
"grad_norm": 2.80315899848938,
|
6149 |
+
"learning_rate": 6.884907473268254e-05,
|
6150 |
+
"loss": 0.4071,
|
6151 |
+
"step": 874
|
6152 |
+
},
|
6153 |
+
{
|
6154 |
+
"epoch": 0.5068790731354091,
|
6155 |
+
"grad_norm": 1.5055872201919556,
|
6156 |
+
"learning_rate": 6.864155001656068e-05,
|
6157 |
+
"loss": 0.0666,
|
6158 |
+
"step": 875
|
6159 |
+
},
|
6160 |
+
{
|
6161 |
+
"epoch": 0.5074583635047067,
|
6162 |
+
"grad_norm": 4.0708112716674805,
|
6163 |
+
"learning_rate": 6.843417497100046e-05,
|
6164 |
+
"loss": 0.354,
|
6165 |
+
"step": 876
|
6166 |
+
},
|
6167 |
+
{
|
6168 |
+
"epoch": 0.5080376538740043,
|
6169 |
+
"grad_norm": 4.592182159423828,
|
6170 |
+
"learning_rate": 6.822695058578104e-05,
|
6171 |
+
"loss": 0.2587,
|
6172 |
+
"step": 877
|
6173 |
+
},
|
6174 |
+
{
|
6175 |
+
"epoch": 0.508616944243302,
|
6176 |
+
"grad_norm": 2.4305107593536377,
|
6177 |
+
"learning_rate": 6.801987784996244e-05,
|
6178 |
+
"loss": 0.2814,
|
6179 |
+
"step": 878
|
6180 |
+
},
|
6181 |
+
{
|
6182 |
+
"epoch": 0.5091962346125996,
|
6183 |
+
"grad_norm": 2.350832939147949,
|
6184 |
+
"learning_rate": 6.78129577518809e-05,
|
6185 |
+
"loss": 0.2766,
|
6186 |
+
"step": 879
|
6187 |
+
},
|
6188 |
+
{
|
6189 |
+
"epoch": 0.5097755249818972,
|
6190 |
+
"grad_norm": 1.6528229713439941,
|
6191 |
+
"learning_rate": 6.760619127914417e-05,
|
6192 |
+
"loss": 0.0857,
|
6193 |
+
"step": 880
|
6194 |
+
},
|
6195 |
+
{
|
6196 |
+
"epoch": 0.5103548153511948,
|
6197 |
+
"grad_norm": 2.696647882461548,
|
6198 |
+
"learning_rate": 6.739957941862671e-05,
|
6199 |
+
"loss": 0.2258,
|
6200 |
+
"step": 881
|
6201 |
+
},
|
6202 |
+
{
|
6203 |
+
"epoch": 0.5109341057204924,
|
6204 |
+
"grad_norm": 2.6145641803741455,
|
6205 |
+
"learning_rate": 6.719312315646507e-05,
|
6206 |
+
"loss": 0.1248,
|
6207 |
+
"step": 882
|
6208 |
+
},
|
6209 |
+
{
|
6210 |
+
"epoch": 0.51151339608979,
|
6211 |
+
"grad_norm": 2.0635499954223633,
|
6212 |
+
"learning_rate": 6.698682347805309e-05,
|
6213 |
+
"loss": 0.1407,
|
6214 |
+
"step": 883
|
6215 |
+
},
|
6216 |
+
{
|
6217 |
+
"epoch": 0.5120926864590876,
|
6218 |
+
"grad_norm": 1.5710378885269165,
|
6219 |
+
"learning_rate": 6.678068136803736e-05,
|
6220 |
+
"loss": 0.0424,
|
6221 |
+
"step": 884
|
6222 |
+
},
|
6223 |
+
{
|
6224 |
+
"epoch": 0.5126719768283853,
|
6225 |
+
"grad_norm": 1.969618320465088,
|
6226 |
+
"learning_rate": 6.657469781031229e-05,
|
6227 |
+
"loss": 0.1132,
|
6228 |
+
"step": 885
|
6229 |
+
},
|
6230 |
+
{
|
6231 |
+
"epoch": 0.5132512671976829,
|
6232 |
+
"grad_norm": 4.359981060028076,
|
6233 |
+
"learning_rate": 6.636887378801563e-05,
|
6234 |
+
"loss": 0.3641,
|
6235 |
+
"step": 886
|
6236 |
+
},
|
6237 |
+
{
|
6238 |
+
"epoch": 0.5138305575669805,
|
6239 |
+
"grad_norm": 3.112766981124878,
|
6240 |
+
"learning_rate": 6.616321028352361e-05,
|
6241 |
+
"loss": 0.3397,
|
6242 |
+
"step": 887
|
6243 |
+
},
|
6244 |
+
{
|
6245 |
+
"epoch": 0.5144098479362781,
|
6246 |
+
"grad_norm": 2.802192449569702,
|
6247 |
+
"learning_rate": 6.595770827844637e-05,
|
6248 |
+
"loss": 0.1709,
|
6249 |
+
"step": 888
|
6250 |
+
},
|
6251 |
+
{
|
6252 |
+
"epoch": 0.5149891383055757,
|
6253 |
+
"grad_norm": 2.710824966430664,
|
6254 |
+
"learning_rate": 6.575236875362325e-05,
|
6255 |
+
"loss": 0.4842,
|
6256 |
+
"step": 889
|
6257 |
+
},
|
6258 |
+
{
|
6259 |
+
"epoch": 0.5155684286748733,
|
6260 |
+
"grad_norm": 2.791417121887207,
|
6261 |
+
"learning_rate": 6.554719268911804e-05,
|
6262 |
+
"loss": 0.2808,
|
6263 |
+
"step": 890
|
6264 |
+
},
|
6265 |
+
{
|
6266 |
+
"epoch": 0.5161477190441709,
|
6267 |
+
"grad_norm": 2.1043267250061035,
|
6268 |
+
"learning_rate": 6.534218106421435e-05,
|
6269 |
+
"loss": 0.1768,
|
6270 |
+
"step": 891
|
6271 |
+
},
|
6272 |
+
{
|
6273 |
+
"epoch": 0.5167270094134685,
|
6274 |
+
"grad_norm": 1.4011709690093994,
|
6275 |
+
"learning_rate": 6.513733485741102e-05,
|
6276 |
+
"loss": 0.1623,
|
6277 |
+
"step": 892
|
6278 |
+
},
|
6279 |
+
{
|
6280 |
+
"epoch": 0.5173062997827661,
|
6281 |
+
"grad_norm": 1.3681217432022095,
|
6282 |
+
"learning_rate": 6.493265504641726e-05,
|
6283 |
+
"loss": 0.1039,
|
6284 |
+
"step": 893
|
6285 |
+
},
|
6286 |
+
{
|
6287 |
+
"epoch": 0.5178855901520637,
|
6288 |
+
"grad_norm": 2.42484712600708,
|
6289 |
+
"learning_rate": 6.472814260814814e-05,
|
6290 |
+
"loss": 0.1746,
|
6291 |
+
"step": 894
|
6292 |
+
},
|
6293 |
+
{
|
6294 |
+
"epoch": 0.5184648805213613,
|
6295 |
+
"grad_norm": 2.149044990539551,
|
6296 |
+
"learning_rate": 6.45237985187199e-05,
|
6297 |
+
"loss": 0.4645,
|
6298 |
+
"step": 895
|
6299 |
+
},
|
6300 |
+
{
|
6301 |
+
"epoch": 0.5190441708906589,
|
6302 |
+
"grad_norm": 2.7207820415496826,
|
6303 |
+
"learning_rate": 6.431962375344522e-05,
|
6304 |
+
"loss": 0.1451,
|
6305 |
+
"step": 896
|
6306 |
+
},
|
6307 |
+
{
|
6308 |
+
"epoch": 0.5196234612599565,
|
6309 |
+
"grad_norm": 2.014847755432129,
|
6310 |
+
"learning_rate": 6.411561928682866e-05,
|
6311 |
+
"loss": 0.0575,
|
6312 |
+
"step": 897
|
6313 |
+
},
|
6314 |
+
{
|
6315 |
+
"epoch": 0.5202027516292541,
|
6316 |
+
"grad_norm": 2.556762456893921,
|
6317 |
+
"learning_rate": 6.39117860925619e-05,
|
6318 |
+
"loss": 0.1226,
|
6319 |
+
"step": 898
|
6320 |
+
},
|
6321 |
+
{
|
6322 |
+
"epoch": 0.5207820419985518,
|
6323 |
+
"grad_norm": 2.2220590114593506,
|
6324 |
+
"learning_rate": 6.370812514351925e-05,
|
6325 |
+
"loss": 0.5027,
|
6326 |
+
"step": 899
|
6327 |
+
},
|
6328 |
+
{
|
6329 |
+
"epoch": 0.5213613323678494,
|
6330 |
+
"grad_norm": 2.47552490234375,
|
6331 |
+
"learning_rate": 6.350463741175281e-05,
|
6332 |
+
"loss": 0.4982,
|
6333 |
+
"step": 900
|
6334 |
+
},
|
6335 |
+
{
|
6336 |
+
"epoch": 0.521940622737147,
|
6337 |
+
"grad_norm": 2.882659435272217,
|
6338 |
+
"learning_rate": 6.3301323868488e-05,
|
6339 |
+
"loss": 0.3143,
|
6340 |
+
"step": 901
|
6341 |
+
},
|
6342 |
+
{
|
6343 |
+
"epoch": 0.5225199131064446,
|
6344 |
+
"grad_norm": 2.444570302963257,
|
6345 |
+
"learning_rate": 6.309818548411881e-05,
|
6346 |
+
"loss": 0.174,
|
6347 |
+
"step": 902
|
6348 |
+
},
|
6349 |
+
{
|
6350 |
+
"epoch": 0.5230992034757422,
|
6351 |
+
"grad_norm": 2.917492628097534,
|
6352 |
+
"learning_rate": 6.28952232282033e-05,
|
6353 |
+
"loss": 0.2331,
|
6354 |
+
"step": 903
|
6355 |
+
},
|
6356 |
+
{
|
6357 |
+
"epoch": 0.5236784938450398,
|
6358 |
+
"grad_norm": 1.820224404335022,
|
6359 |
+
"learning_rate": 6.269243806945878e-05,
|
6360 |
+
"loss": 0.1207,
|
6361 |
+
"step": 904
|
6362 |
+
},
|
6363 |
+
{
|
6364 |
+
"epoch": 0.5242577842143374,
|
6365 |
+
"grad_norm": 2.3333451747894287,
|
6366 |
+
"learning_rate": 6.248983097575734e-05,
|
6367 |
+
"loss": 0.1775,
|
6368 |
+
"step": 905
|
6369 |
+
},
|
6370 |
+
{
|
6371 |
+
"epoch": 0.524837074583635,
|
6372 |
+
"grad_norm": 2.552856922149658,
|
6373 |
+
"learning_rate": 6.228740291412121e-05,
|
6374 |
+
"loss": 0.3934,
|
6375 |
+
"step": 906
|
6376 |
+
},
|
6377 |
+
{
|
6378 |
+
"epoch": 0.5254163649529326,
|
6379 |
+
"grad_norm": 2.5968430042266846,
|
6380 |
+
"learning_rate": 6.208515485071806e-05,
|
6381 |
+
"loss": 0.1207,
|
6382 |
+
"step": 907
|
6383 |
+
},
|
6384 |
+
{
|
6385 |
+
"epoch": 0.5259956553222302,
|
6386 |
+
"grad_norm": 2.618913173675537,
|
6387 |
+
"learning_rate": 6.188308775085654e-05,
|
6388 |
+
"loss": 0.2191,
|
6389 |
+
"step": 908
|
6390 |
+
},
|
6391 |
+
{
|
6392 |
+
"epoch": 0.5265749456915279,
|
6393 |
+
"grad_norm": 2.2938358783721924,
|
6394 |
+
"learning_rate": 6.168120257898148e-05,
|
6395 |
+
"loss": 0.1433,
|
6396 |
+
"step": 909
|
6397 |
+
},
|
6398 |
+
{
|
6399 |
+
"epoch": 0.5271542360608255,
|
6400 |
+
"grad_norm": 3.4978723526000977,
|
6401 |
+
"learning_rate": 6.147950029866946e-05,
|
6402 |
+
"loss": 0.2251,
|
6403 |
+
"step": 910
|
6404 |
+
},
|
6405 |
+
{
|
6406 |
+
"epoch": 0.5277335264301231,
|
6407 |
+
"grad_norm": 1.9927196502685547,
|
6408 |
+
"learning_rate": 6.127798187262406e-05,
|
6409 |
+
"loss": 0.1476,
|
6410 |
+
"step": 911
|
6411 |
+
},
|
6412 |
+
{
|
6413 |
+
"epoch": 0.5283128167994207,
|
6414 |
+
"grad_norm": 3.5390841960906982,
|
6415 |
+
"learning_rate": 6.107664826267146e-05,
|
6416 |
+
"loss": 0.3075,
|
6417 |
+
"step": 912
|
6418 |
+
},
|
6419 |
+
{
|
6420 |
+
"epoch": 0.5288921071687184,
|
6421 |
+
"grad_norm": 2.5269696712493896,
|
6422 |
+
"learning_rate": 6.087550042975561e-05,
|
6423 |
+
"loss": 0.1361,
|
6424 |
+
"step": 913
|
6425 |
+
},
|
6426 |
+
{
|
6427 |
+
"epoch": 0.529471397538016,
|
6428 |
+
"grad_norm": 1.8151060342788696,
|
6429 |
+
"learning_rate": 6.067453933393383e-05,
|
6430 |
+
"loss": 0.1256,
|
6431 |
+
"step": 914
|
6432 |
+
},
|
6433 |
+
{
|
6434 |
+
"epoch": 0.5300506879073136,
|
6435 |
+
"grad_norm": 1.9455229043960571,
|
6436 |
+
"learning_rate": 6.047376593437214e-05,
|
6437 |
+
"loss": 0.0716,
|
6438 |
+
"step": 915
|
6439 |
+
},
|
6440 |
+
{
|
6441 |
+
"epoch": 0.5306299782766112,
|
6442 |
+
"grad_norm": 2.796208620071411,
|
6443 |
+
"learning_rate": 6.027318118934078e-05,
|
6444 |
+
"loss": 0.4193,
|
6445 |
+
"step": 916
|
6446 |
+
},
|
6447 |
+
{
|
6448 |
+
"epoch": 0.5312092686459088,
|
6449 |
+
"grad_norm": 1.6880676746368408,
|
6450 |
+
"learning_rate": 6.007278605620943e-05,
|
6451 |
+
"loss": 0.0714,
|
6452 |
+
"step": 917
|
6453 |
+
},
|
6454 |
+
{
|
6455 |
+
"epoch": 0.5317885590152064,
|
6456 |
+
"grad_norm": 3.125753879547119,
|
6457 |
+
"learning_rate": 5.987258149144288e-05,
|
6458 |
+
"loss": 0.1506,
|
6459 |
+
"step": 918
|
6460 |
+
},
|
6461 |
+
{
|
6462 |
+
"epoch": 0.532367849384504,
|
6463 |
+
"grad_norm": 1.9753881692886353,
|
6464 |
+
"learning_rate": 5.9672568450596335e-05,
|
6465 |
+
"loss": 0.279,
|
6466 |
+
"step": 919
|
6467 |
+
},
|
6468 |
+
{
|
6469 |
+
"epoch": 0.5329471397538016,
|
6470 |
+
"grad_norm": 1.4957164525985718,
|
6471 |
+
"learning_rate": 5.9472747888310834e-05,
|
6472 |
+
"loss": 0.1175,
|
6473 |
+
"step": 920
|
6474 |
+
},
|
6475 |
+
{
|
6476 |
+
"epoch": 0.5335264301230992,
|
6477 |
+
"grad_norm": 2.2510874271392822,
|
6478 |
+
"learning_rate": 5.927312075830874e-05,
|
6479 |
+
"loss": 0.1741,
|
6480 |
+
"step": 921
|
6481 |
+
},
|
6482 |
+
{
|
6483 |
+
"epoch": 0.5341057204923968,
|
6484 |
+
"grad_norm": 1.6291396617889404,
|
6485 |
+
"learning_rate": 5.907368801338927e-05,
|
6486 |
+
"loss": 0.2072,
|
6487 |
+
"step": 922
|
6488 |
+
},
|
6489 |
+
{
|
6490 |
+
"epoch": 0.5346850108616944,
|
6491 |
+
"grad_norm": 2.2468626499176025,
|
6492 |
+
"learning_rate": 5.887445060542377e-05,
|
6493 |
+
"loss": 0.1189,
|
6494 |
+
"step": 923
|
6495 |
+
},
|
6496 |
+
{
|
6497 |
+
"epoch": 0.535264301230992,
|
6498 |
+
"grad_norm": 3.759373188018799,
|
6499 |
+
"learning_rate": 5.86754094853513e-05,
|
6500 |
+
"loss": 0.2473,
|
6501 |
+
"step": 924
|
6502 |
+
},
|
6503 |
+
{
|
6504 |
+
"epoch": 0.5358435916002896,
|
6505 |
+
"grad_norm": 0.9833599328994751,
|
6506 |
+
"learning_rate": 5.8476565603174025e-05,
|
6507 |
+
"loss": 0.0406,
|
6508 |
+
"step": 925
|
6509 |
+
},
|
6510 |
+
{
|
6511 |
+
"epoch": 0.5364228819695872,
|
6512 |
+
"grad_norm": 3.8795347213745117,
|
6513 |
+
"learning_rate": 5.827791990795278e-05,
|
6514 |
+
"loss": 0.2467,
|
6515 |
+
"step": 926
|
6516 |
+
},
|
6517 |
+
{
|
6518 |
+
"epoch": 0.5370021723388849,
|
6519 |
+
"grad_norm": 1.0939834117889404,
|
6520 |
+
"learning_rate": 5.807947334780241e-05,
|
6521 |
+
"loss": 0.0357,
|
6522 |
+
"step": 927
|
6523 |
+
},
|
6524 |
+
{
|
6525 |
+
"epoch": 0.5375814627081825,
|
6526 |
+
"grad_norm": 1.7609702348709106,
|
6527 |
+
"learning_rate": 5.7881226869887314e-05,
|
6528 |
+
"loss": 0.0956,
|
6529 |
+
"step": 928
|
6530 |
+
},
|
6531 |
+
{
|
6532 |
+
"epoch": 0.5381607530774801,
|
6533 |
+
"grad_norm": 2.235502004623413,
|
6534 |
+
"learning_rate": 5.768318142041701e-05,
|
6535 |
+
"loss": 0.2473,
|
6536 |
+
"step": 929
|
6537 |
+
},
|
6538 |
+
{
|
6539 |
+
"epoch": 0.5387400434467777,
|
6540 |
+
"grad_norm": 1.577030062675476,
|
6541 |
+
"learning_rate": 5.748533794464142e-05,
|
6542 |
+
"loss": 0.0835,
|
6543 |
+
"step": 930
|
6544 |
+
},
|
6545 |
+
{
|
6546 |
+
"epoch": 0.5393193338160753,
|
6547 |
+
"grad_norm": 3.7965221405029297,
|
6548 |
+
"learning_rate": 5.7287697386846515e-05,
|
6549 |
+
"loss": 0.2135,
|
6550 |
+
"step": 931
|
6551 |
+
},
|
6552 |
+
{
|
6553 |
+
"epoch": 0.5398986241853729,
|
6554 |
+
"grad_norm": 2.143045663833618,
|
6555 |
+
"learning_rate": 5.709026069034973e-05,
|
6556 |
+
"loss": 0.1006,
|
6557 |
+
"step": 932
|
6558 |
+
},
|
6559 |
+
{
|
6560 |
+
"epoch": 0.5404779145546705,
|
6561 |
+
"grad_norm": 3.032989978790283,
|
6562 |
+
"learning_rate": 5.689302879749554e-05,
|
6563 |
+
"loss": 0.1924,
|
6564 |
+
"step": 933
|
6565 |
+
},
|
6566 |
+
{
|
6567 |
+
"epoch": 0.5410572049239681,
|
6568 |
+
"grad_norm": 5.1808271408081055,
|
6569 |
+
"learning_rate": 5.669600264965088e-05,
|
6570 |
+
"loss": 0.5963,
|
6571 |
+
"step": 934
|
6572 |
+
},
|
6573 |
+
{
|
6574 |
+
"epoch": 0.5416364952932657,
|
6575 |
+
"grad_norm": 3.114119052886963,
|
6576 |
+
"learning_rate": 5.649918318720069e-05,
|
6577 |
+
"loss": 0.2447,
|
6578 |
+
"step": 935
|
6579 |
+
},
|
6580 |
+
{
|
6581 |
+
"epoch": 0.5422157856625633,
|
6582 |
+
"grad_norm": 3.1728622913360596,
|
6583 |
+
"learning_rate": 5.630257134954338e-05,
|
6584 |
+
"loss": 0.2676,
|
6585 |
+
"step": 936
|
6586 |
+
},
|
6587 |
+
{
|
6588 |
+
"epoch": 0.542795076031861,
|
6589 |
+
"grad_norm": 1.5399963855743408,
|
6590 |
+
"learning_rate": 5.610616807508651e-05,
|
6591 |
+
"loss": 0.075,
|
6592 |
+
"step": 937
|
6593 |
+
},
|
6594 |
+
{
|
6595 |
+
"epoch": 0.5433743664011585,
|
6596 |
+
"grad_norm": 2.5393104553222656,
|
6597 |
+
"learning_rate": 5.59099743012421e-05,
|
6598 |
+
"loss": 0.3181,
|
6599 |
+
"step": 938
|
6600 |
+
},
|
6601 |
+
{
|
6602 |
+
"epoch": 0.5439536567704562,
|
6603 |
+
"grad_norm": 1.58083176612854,
|
6604 |
+
"learning_rate": 5.5713990964422235e-05,
|
6605 |
+
"loss": 0.0876,
|
6606 |
+
"step": 939
|
6607 |
+
},
|
6608 |
+
{
|
6609 |
+
"epoch": 0.5445329471397538,
|
6610 |
+
"grad_norm": 3.9276978969573975,
|
6611 |
+
"learning_rate": 5.551821900003461e-05,
|
6612 |
+
"loss": 0.2772,
|
6613 |
+
"step": 940
|
6614 |
+
},
|
6615 |
+
{
|
6616 |
+
"epoch": 0.5451122375090515,
|
6617 |
+
"grad_norm": 1.7526675462722778,
|
6618 |
+
"learning_rate": 5.5322659342478154e-05,
|
6619 |
+
"loss": 0.1198,
|
6620 |
+
"step": 941
|
6621 |
+
},
|
6622 |
+
{
|
6623 |
+
"epoch": 0.5456915278783491,
|
6624 |
+
"grad_norm": 2.6656346321105957,
|
6625 |
+
"learning_rate": 5.512731292513835e-05,
|
6626 |
+
"loss": 0.2505,
|
6627 |
+
"step": 942
|
6628 |
+
},
|
6629 |
+
{
|
6630 |
+
"epoch": 0.5462708182476467,
|
6631 |
+
"grad_norm": 1.0102094411849976,
|
6632 |
+
"learning_rate": 5.4932180680383004e-05,
|
6633 |
+
"loss": 0.079,
|
6634 |
+
"step": 943
|
6635 |
+
},
|
6636 |
+
{
|
6637 |
+
"epoch": 0.5468501086169443,
|
6638 |
+
"grad_norm": 0.9727888107299805,
|
6639 |
+
"learning_rate": 5.473726353955764e-05,
|
6640 |
+
"loss": 0.059,
|
6641 |
+
"step": 944
|
6642 |
+
},
|
6643 |
+
{
|
6644 |
+
"epoch": 0.5474293989862419,
|
6645 |
+
"grad_norm": 0.9393841624259949,
|
6646 |
+
"learning_rate": 5.454256243298112e-05,
|
6647 |
+
"loss": 0.0599,
|
6648 |
+
"step": 945
|
6649 |
+
},
|
6650 |
+
{
|
6651 |
+
"epoch": 0.5480086893555395,
|
6652 |
+
"grad_norm": 3.4037275314331055,
|
6653 |
+
"learning_rate": 5.434807828994123e-05,
|
6654 |
+
"loss": 0.3032,
|
6655 |
+
"step": 946
|
6656 |
+
},
|
6657 |
+
{
|
6658 |
+
"epoch": 0.5485879797248371,
|
6659 |
+
"grad_norm": 1.1161447763442993,
|
6660 |
+
"learning_rate": 5.4153812038690143e-05,
|
6661 |
+
"loss": 0.0603,
|
6662 |
+
"step": 947
|
6663 |
+
},
|
6664 |
+
{
|
6665 |
+
"epoch": 0.5491672700941347,
|
6666 |
+
"grad_norm": 1.1687431335449219,
|
6667 |
+
"learning_rate": 5.39597646064402e-05,
|
6668 |
+
"loss": 0.1083,
|
6669 |
+
"step": 948
|
6670 |
+
},
|
6671 |
+
{
|
6672 |
+
"epoch": 0.5497465604634323,
|
6673 |
+
"grad_norm": 2.019043207168579,
|
6674 |
+
"learning_rate": 5.3765936919359205e-05,
|
6675 |
+
"loss": 0.2099,
|
6676 |
+
"step": 949
|
6677 |
+
},
|
6678 |
+
{
|
6679 |
+
"epoch": 0.5503258508327299,
|
6680 |
+
"grad_norm": 2.151170015335083,
|
6681 |
+
"learning_rate": 5.357232990256618e-05,
|
6682 |
+
"loss": 0.1215,
|
6683 |
+
"step": 950
|
6684 |
+
},
|
6685 |
+
{
|
6686 |
+
"epoch": 0.5509051412020275,
|
6687 |
+
"grad_norm": 2.3641302585601807,
|
6688 |
+
"learning_rate": 5.337894448012692e-05,
|
6689 |
+
"loss": 0.216,
|
6690 |
+
"step": 951
|
6691 |
+
},
|
6692 |
+
{
|
6693 |
+
"epoch": 0.5514844315713251,
|
6694 |
+
"grad_norm": 1.8099020719528198,
|
6695 |
+
"learning_rate": 5.318578157504959e-05,
|
6696 |
+
"loss": 0.1586,
|
6697 |
+
"step": 952
|
6698 |
+
},
|
6699 |
+
{
|
6700 |
+
"epoch": 0.5520637219406227,
|
6701 |
+
"grad_norm": 1.4914263486862183,
|
6702 |
+
"learning_rate": 5.299284210928025e-05,
|
6703 |
+
"loss": 0.0594,
|
6704 |
+
"step": 953
|
6705 |
+
},
|
6706 |
+
{
|
6707 |
+
"epoch": 0.5526430123099203,
|
6708 |
+
"grad_norm": 1.684990644454956,
|
6709 |
+
"learning_rate": 5.280012700369856e-05,
|
6710 |
+
"loss": 0.0594,
|
6711 |
+
"step": 954
|
6712 |
+
},
|
6713 |
+
{
|
6714 |
+
"epoch": 0.5532223026792179,
|
6715 |
+
"grad_norm": 2.4913032054901123,
|
6716 |
+
"learning_rate": 5.260763717811328e-05,
|
6717 |
+
"loss": 0.0889,
|
6718 |
+
"step": 955
|
6719 |
+
},
|
6720 |
+
{
|
6721 |
+
"epoch": 0.5538015930485156,
|
6722 |
+
"grad_norm": 2.953730821609497,
|
6723 |
+
"learning_rate": 5.2415373551258054e-05,
|
6724 |
+
"loss": 0.2089,
|
6725 |
+
"step": 956
|
6726 |
+
},
|
6727 |
+
{
|
6728 |
+
"epoch": 0.5543808834178132,
|
6729 |
+
"grad_norm": 2.1618032455444336,
|
6730 |
+
"learning_rate": 5.2223337040786744e-05,
|
6731 |
+
"loss": 0.1158,
|
6732 |
+
"step": 957
|
6733 |
+
},
|
6734 |
+
{
|
6735 |
+
"epoch": 0.5549601737871108,
|
6736 |
+
"grad_norm": 1.2927886247634888,
|
6737 |
+
"learning_rate": 5.203152856326933e-05,
|
6738 |
+
"loss": 0.055,
|
6739 |
+
"step": 958
|
6740 |
+
},
|
6741 |
+
{
|
6742 |
+
"epoch": 0.5555394641564084,
|
6743 |
+
"grad_norm": 1.9999315738677979,
|
6744 |
+
"learning_rate": 5.183994903418732e-05,
|
6745 |
+
"loss": 0.3154,
|
6746 |
+
"step": 959
|
6747 |
+
},
|
6748 |
+
{
|
6749 |
+
"epoch": 0.556118754525706,
|
6750 |
+
"grad_norm": 2.866997003555298,
|
6751 |
+
"learning_rate": 5.164859936792955e-05,
|
6752 |
+
"loss": 0.2854,
|
6753 |
+
"step": 960
|
6754 |
+
},
|
6755 |
+
{
|
6756 |
+
"epoch": 0.5566980448950036,
|
6757 |
+
"grad_norm": 1.4728211164474487,
|
6758 |
+
"learning_rate": 5.145748047778771e-05,
|
6759 |
+
"loss": 0.0761,
|
6760 |
+
"step": 961
|
6761 |
+
},
|
6762 |
+
{
|
6763 |
+
"epoch": 0.5572773352643012,
|
6764 |
+
"grad_norm": 1.7411283254623413,
|
6765 |
+
"learning_rate": 5.1266593275952034e-05,
|
6766 |
+
"loss": 0.1866,
|
6767 |
+
"step": 962
|
6768 |
+
},
|
6769 |
+
{
|
6770 |
+
"epoch": 0.5578566256335988,
|
6771 |
+
"grad_norm": 1.9727611541748047,
|
6772 |
+
"learning_rate": 5.107593867350692e-05,
|
6773 |
+
"loss": 0.1024,
|
6774 |
+
"step": 963
|
6775 |
+
},
|
6776 |
+
{
|
6777 |
+
"epoch": 0.5584359160028964,
|
6778 |
+
"grad_norm": 1.5096935033798218,
|
6779 |
+
"learning_rate": 5.088551758042655e-05,
|
6780 |
+
"loss": 0.0958,
|
6781 |
+
"step": 964
|
6782 |
+
},
|
6783 |
+
{
|
6784 |
+
"epoch": 0.559015206372194,
|
6785 |
+
"grad_norm": 1.9016562700271606,
|
6786 |
+
"learning_rate": 5.0695330905570735e-05,
|
6787 |
+
"loss": 0.1431,
|
6788 |
+
"step": 965
|
6789 |
+
},
|
6790 |
+
{
|
6791 |
+
"epoch": 0.5595944967414916,
|
6792 |
+
"grad_norm": 1.6201834678649902,
|
6793 |
+
"learning_rate": 5.050537955668027e-05,
|
6794 |
+
"loss": 0.1043,
|
6795 |
+
"step": 966
|
6796 |
+
},
|
6797 |
+
{
|
6798 |
+
"epoch": 0.5601737871107892,
|
6799 |
+
"grad_norm": 3.7371792793273926,
|
6800 |
+
"learning_rate": 5.031566444037288e-05,
|
6801 |
+
"loss": 0.2303,
|
6802 |
+
"step": 967
|
6803 |
+
},
|
6804 |
+
{
|
6805 |
+
"epoch": 0.5607530774800868,
|
6806 |
+
"grad_norm": 3.2157890796661377,
|
6807 |
+
"learning_rate": 5.012618646213869e-05,
|
6808 |
+
"loss": 0.1548,
|
6809 |
+
"step": 968
|
6810 |
+
},
|
6811 |
+
{
|
6812 |
+
"epoch": 0.5613323678493845,
|
6813 |
+
"grad_norm": 1.6408613920211792,
|
6814 |
+
"learning_rate": 4.993694652633606e-05,
|
6815 |
+
"loss": 0.1083,
|
6816 |
+
"step": 969
|
6817 |
+
},
|
6818 |
+
{
|
6819 |
+
"epoch": 0.5619116582186822,
|
6820 |
+
"grad_norm": 5.066162109375,
|
6821 |
+
"learning_rate": 4.9747945536187145e-05,
|
6822 |
+
"loss": 0.7082,
|
6823 |
+
"step": 970
|
6824 |
+
},
|
6825 |
+
{
|
6826 |
+
"epoch": 0.5624909485879798,
|
6827 |
+
"grad_norm": 2.278359889984131,
|
6828 |
+
"learning_rate": 4.955918439377366e-05,
|
6829 |
+
"loss": 0.2712,
|
6830 |
+
"step": 971
|
6831 |
+
},
|
6832 |
+
{
|
6833 |
+
"epoch": 0.5630702389572774,
|
6834 |
+
"grad_norm": 2.0341436862945557,
|
6835 |
+
"learning_rate": 4.937066400003256e-05,
|
6836 |
+
"loss": 0.1048,
|
6837 |
+
"step": 972
|
6838 |
+
},
|
6839 |
+
{
|
6840 |
+
"epoch": 0.563649529326575,
|
6841 |
+
"grad_norm": 1.8281821012496948,
|
6842 |
+
"learning_rate": 4.918238525475173e-05,
|
6843 |
+
"loss": 0.0627,
|
6844 |
+
"step": 973
|
6845 |
+
},
|
6846 |
+
{
|
6847 |
+
"epoch": 0.5642288196958726,
|
6848 |
+
"grad_norm": 2.182070016860962,
|
6849 |
+
"learning_rate": 4.899434905656563e-05,
|
6850 |
+
"loss": 0.136,
|
6851 |
+
"step": 974
|
6852 |
+
},
|
6853 |
+
{
|
6854 |
+
"epoch": 0.5648081100651702,
|
6855 |
+
"grad_norm": 2.1413326263427734,
|
6856 |
+
"learning_rate": 4.880655630295122e-05,
|
6857 |
+
"loss": 0.2805,
|
6858 |
+
"step": 975
|
6859 |
+
},
|
6860 |
+
{
|
6861 |
+
"epoch": 0.5653874004344678,
|
6862 |
+
"grad_norm": 2.208822011947632,
|
6863 |
+
"learning_rate": 4.861900789022339e-05,
|
6864 |
+
"loss": 0.1767,
|
6865 |
+
"step": 976
|
6866 |
+
},
|
6867 |
+
{
|
6868 |
+
"epoch": 0.5659666908037654,
|
6869 |
+
"grad_norm": 2.2816905975341797,
|
6870 |
+
"learning_rate": 4.8431704713530834e-05,
|
6871 |
+
"loss": 0.247,
|
6872 |
+
"step": 977
|
6873 |
+
},
|
6874 |
+
{
|
6875 |
+
"epoch": 0.566545981173063,
|
6876 |
+
"grad_norm": 1.551398515701294,
|
6877 |
+
"learning_rate": 4.824464766685182e-05,
|
6878 |
+
"loss": 0.1369,
|
6879 |
+
"step": 978
|
6880 |
+
},
|
6881 |
+
{
|
6882 |
+
"epoch": 0.5671252715423606,
|
6883 |
+
"grad_norm": 1.2330127954483032,
|
6884 |
+
"learning_rate": 4.805783764298979e-05,
|
6885 |
+
"loss": 0.0566,
|
6886 |
+
"step": 979
|
6887 |
+
},
|
6888 |
+
{
|
6889 |
+
"epoch": 0.5677045619116582,
|
6890 |
+
"grad_norm": 2.6698801517486572,
|
6891 |
+
"learning_rate": 4.7871275533569223e-05,
|
6892 |
+
"loss": 0.1718,
|
6893 |
+
"step": 980
|
6894 |
+
},
|
6895 |
+
{
|
6896 |
+
"epoch": 0.5682838522809558,
|
6897 |
+
"grad_norm": 1.4910578727722168,
|
6898 |
+
"learning_rate": 4.768496222903134e-05,
|
6899 |
+
"loss": 0.1398,
|
6900 |
+
"step": 981
|
6901 |
+
},
|
6902 |
+
{
|
6903 |
+
"epoch": 0.5688631426502534,
|
6904 |
+
"grad_norm": 3.600367307662964,
|
6905 |
+
"learning_rate": 4.749889861862977e-05,
|
6906 |
+
"loss": 0.7276,
|
6907 |
+
"step": 982
|
6908 |
+
},
|
6909 |
+
{
|
6910 |
+
"epoch": 0.569442433019551,
|
6911 |
+
"grad_norm": 1.2307556867599487,
|
6912 |
+
"learning_rate": 4.731308559042643e-05,
|
6913 |
+
"loss": 0.0613,
|
6914 |
+
"step": 983
|
6915 |
+
},
|
6916 |
+
{
|
6917 |
+
"epoch": 0.5700217233888487,
|
6918 |
+
"grad_norm": 1.406229019165039,
|
6919 |
+
"learning_rate": 4.712752403128729e-05,
|
6920 |
+
"loss": 0.0437,
|
6921 |
+
"step": 984
|
6922 |
+
},
|
6923 |
+
{
|
6924 |
+
"epoch": 0.5706010137581463,
|
6925 |
+
"grad_norm": 1.7106623649597168,
|
6926 |
+
"learning_rate": 4.694221482687797e-05,
|
6927 |
+
"loss": 0.1914,
|
6928 |
+
"step": 985
|
6929 |
+
},
|
6930 |
+
{
|
6931 |
+
"epoch": 0.5711803041274439,
|
6932 |
+
"grad_norm": 1.2789785861968994,
|
6933 |
+
"learning_rate": 4.6757158861659714e-05,
|
6934 |
+
"loss": 0.0471,
|
6935 |
+
"step": 986
|
6936 |
+
},
|
6937 |
+
{
|
6938 |
+
"epoch": 0.5717595944967415,
|
6939 |
+
"grad_norm": 2.6956300735473633,
|
6940 |
+
"learning_rate": 4.657235701888504e-05,
|
6941 |
+
"loss": 0.6837,
|
6942 |
+
"step": 987
|
6943 |
+
},
|
6944 |
+
{
|
6945 |
+
"epoch": 0.5723388848660391,
|
6946 |
+
"grad_norm": 3.1540396213531494,
|
6947 |
+
"learning_rate": 4.638781018059356e-05,
|
6948 |
+
"loss": 0.2424,
|
6949 |
+
"step": 988
|
6950 |
+
},
|
6951 |
+
{
|
6952 |
+
"epoch": 0.5729181752353367,
|
6953 |
+
"grad_norm": 1.4880813360214233,
|
6954 |
+
"learning_rate": 4.620351922760781e-05,
|
6955 |
+
"loss": 0.0921,
|
6956 |
+
"step": 989
|
6957 |
+
},
|
6958 |
+
{
|
6959 |
+
"epoch": 0.5734974656046343,
|
6960 |
+
"grad_norm": 0.772277295589447,
|
6961 |
+
"learning_rate": 4.601948503952896e-05,
|
6962 |
+
"loss": 0.0245,
|
6963 |
+
"step": 990
|
6964 |
+
},
|
6965 |
+
{
|
6966 |
+
"epoch": 0.5740767559739319,
|
6967 |
+
"grad_norm": 2.9450745582580566,
|
6968 |
+
"learning_rate": 4.5835708494732775e-05,
|
6969 |
+
"loss": 0.1315,
|
6970 |
+
"step": 991
|
6971 |
+
},
|
6972 |
+
{
|
6973 |
+
"epoch": 0.5746560463432295,
|
6974 |
+
"grad_norm": 3.9312920570373535,
|
6975 |
+
"learning_rate": 4.565219047036521e-05,
|
6976 |
+
"loss": 0.4472,
|
6977 |
+
"step": 992
|
6978 |
+
},
|
6979 |
+
{
|
6980 |
+
"epoch": 0.5752353367125271,
|
6981 |
+
"grad_norm": 1.2942410707473755,
|
6982 |
+
"learning_rate": 4.5468931842338324e-05,
|
6983 |
+
"loss": 0.0608,
|
6984 |
+
"step": 993
|
6985 |
+
},
|
6986 |
+
{
|
6987 |
+
"epoch": 0.5758146270818247,
|
6988 |
+
"grad_norm": 1.3082633018493652,
|
6989 |
+
"learning_rate": 4.5285933485326226e-05,
|
6990 |
+
"loss": 0.0774,
|
6991 |
+
"step": 994
|
6992 |
+
},
|
6993 |
+
{
|
6994 |
+
"epoch": 0.5763939174511223,
|
6995 |
+
"grad_norm": 2.2955095767974854,
|
6996 |
+
"learning_rate": 4.510319627276066e-05,
|
6997 |
+
"loss": 0.1052,
|
6998 |
+
"step": 995
|
6999 |
+
},
|
7000 |
+
{
|
7001 |
+
"epoch": 0.5769732078204199,
|
7002 |
+
"grad_norm": 3.623216152191162,
|
7003 |
+
"learning_rate": 4.492072107682699e-05,
|
7004 |
+
"loss": 0.3283,
|
7005 |
+
"step": 996
|
7006 |
+
},
|
7007 |
+
{
|
7008 |
+
"epoch": 0.5775524981897175,
|
7009 |
+
"grad_norm": 2.389986038208008,
|
7010 |
+
"learning_rate": 4.473850876846001e-05,
|
7011 |
+
"loss": 0.1185,
|
7012 |
+
"step": 997
|
7013 |
+
},
|
7014 |
+
{
|
7015 |
+
"epoch": 0.5781317885590153,
|
7016 |
+
"grad_norm": 2.296396493911743,
|
7017 |
+
"learning_rate": 4.4556560217339774e-05,
|
7018 |
+
"loss": 0.2062,
|
7019 |
+
"step": 998
|
7020 |
+
},
|
7021 |
+
{
|
7022 |
+
"epoch": 0.5787110789283129,
|
7023 |
+
"grad_norm": 1.7258268594741821,
|
7024 |
+
"learning_rate": 4.4374876291887454e-05,
|
7025 |
+
"loss": 0.0945,
|
7026 |
+
"step": 999
|
7027 |
+
},
|
7028 |
+
{
|
7029 |
+
"epoch": 0.5792903692976105,
|
7030 |
+
"grad_norm": 2.0098509788513184,
|
7031 |
+
"learning_rate": 4.419345785926119e-05,
|
7032 |
+
"loss": 0.1731,
|
7033 |
+
"step": 1000
|
7034 |
+
},
|
7035 |
+
{
|
7036 |
+
"epoch": 0.5798696596669081,
|
7037 |
+
"grad_norm": 3.1916115283966064,
|
7038 |
+
"learning_rate": 4.401230578535191e-05,
|
7039 |
+
"loss": 0.1093,
|
7040 |
+
"step": 1001
|
7041 |
+
},
|
7042 |
+
{
|
7043 |
+
"epoch": 0.5804489500362057,
|
7044 |
+
"grad_norm": 0.8169403076171875,
|
7045 |
+
"learning_rate": 4.383142093477938e-05,
|
7046 |
+
"loss": 0.0621,
|
7047 |
+
"step": 1002
|
7048 |
+
},
|
7049 |
+
{
|
7050 |
+
"epoch": 0.5810282404055033,
|
7051 |
+
"grad_norm": 1.7748020887374878,
|
7052 |
+
"learning_rate": 4.365080417088776e-05,
|
7053 |
+
"loss": 0.0903,
|
7054 |
+
"step": 1003
|
7055 |
+
},
|
7056 |
+
{
|
7057 |
+
"epoch": 0.5816075307748009,
|
7058 |
+
"grad_norm": 2.6781952381134033,
|
7059 |
+
"learning_rate": 4.3470456355741786e-05,
|
7060 |
+
"loss": 0.3334,
|
7061 |
+
"step": 1004
|
7062 |
+
},
|
7063 |
+
{
|
7064 |
+
"epoch": 0.5821868211440985,
|
7065 |
+
"grad_norm": 1.354478120803833,
|
7066 |
+
"learning_rate": 4.329037835012245e-05,
|
7067 |
+
"loss": 0.0851,
|
7068 |
+
"step": 1005
|
7069 |
+
},
|
7070 |
+
{
|
7071 |
+
"epoch": 0.5827661115133961,
|
7072 |
+
"grad_norm": 2.0575757026672363,
|
7073 |
+
"learning_rate": 4.311057101352304e-05,
|
7074 |
+
"loss": 0.1127,
|
7075 |
+
"step": 1006
|
7076 |
+
},
|
7077 |
+
{
|
7078 |
+
"epoch": 0.5833454018826937,
|
7079 |
+
"grad_norm": 1.2931239604949951,
|
7080 |
+
"learning_rate": 4.293103520414491e-05,
|
7081 |
+
"loss": 0.0811,
|
7082 |
+
"step": 1007
|
7083 |
+
},
|
7084 |
+
{
|
7085 |
+
"epoch": 0.5839246922519913,
|
7086 |
+
"grad_norm": 4.230047702789307,
|
7087 |
+
"learning_rate": 4.275177177889347e-05,
|
7088 |
+
"loss": 0.2677,
|
7089 |
+
"step": 1008
|
7090 |
+
},
|
7091 |
+
{
|
7092 |
+
"epoch": 0.5845039826212889,
|
7093 |
+
"grad_norm": 1.209717869758606,
|
7094 |
+
"learning_rate": 4.2572781593374086e-05,
|
7095 |
+
"loss": 0.0734,
|
7096 |
+
"step": 1009
|
7097 |
+
},
|
7098 |
+
{
|
7099 |
+
"epoch": 0.5850832729905865,
|
7100 |
+
"grad_norm": 1.9632679224014282,
|
7101 |
+
"learning_rate": 4.239406550188791e-05,
|
7102 |
+
"loss": 0.1276,
|
7103 |
+
"step": 1010
|
7104 |
+
},
|
7105 |
+
{
|
7106 |
+
"epoch": 0.5856625633598841,
|
7107 |
+
"grad_norm": 1.337821125984192,
|
7108 |
+
"learning_rate": 4.221562435742802e-05,
|
7109 |
+
"loss": 0.0591,
|
7110 |
+
"step": 1011
|
7111 |
+
},
|
7112 |
+
{
|
7113 |
+
"epoch": 0.5862418537291818,
|
7114 |
+
"grad_norm": 1.3360164165496826,
|
7115 |
+
"learning_rate": 4.203745901167504e-05,
|
7116 |
+
"loss": 0.0924,
|
7117 |
+
"step": 1012
|
7118 |
+
},
|
7119 |
+
{
|
7120 |
+
"epoch": 0.5868211440984794,
|
7121 |
+
"grad_norm": 3.3772501945495605,
|
7122 |
+
"learning_rate": 4.1859570314993324e-05,
|
7123 |
+
"loss": 0.2445,
|
7124 |
+
"step": 1013
|
7125 |
+
},
|
7126 |
+
{
|
7127 |
+
"epoch": 0.587400434467777,
|
7128 |
+
"grad_norm": 1.5914647579193115,
|
7129 |
+
"learning_rate": 4.168195911642677e-05,
|
7130 |
+
"loss": 0.058,
|
7131 |
+
"step": 1014
|
7132 |
+
},
|
7133 |
+
{
|
7134 |
+
"epoch": 0.5879797248370746,
|
7135 |
+
"grad_norm": 1.4137589931488037,
|
7136 |
+
"learning_rate": 4.15046262636948e-05,
|
7137 |
+
"loss": 0.0644,
|
7138 |
+
"step": 1015
|
7139 |
+
},
|
7140 |
+
{
|
7141 |
+
"epoch": 0.5885590152063722,
|
7142 |
+
"grad_norm": 2.6447296142578125,
|
7143 |
+
"learning_rate": 4.1327572603188415e-05,
|
7144 |
+
"loss": 0.1097,
|
7145 |
+
"step": 1016
|
7146 |
+
},
|
7147 |
+
{
|
7148 |
+
"epoch": 0.5891383055756698,
|
7149 |
+
"grad_norm": 1.6296364068984985,
|
7150 |
+
"learning_rate": 4.115079897996595e-05,
|
7151 |
+
"loss": 0.0572,
|
7152 |
+
"step": 1017
|
7153 |
+
},
|
7154 |
+
{
|
7155 |
+
"epoch": 0.5897175959449674,
|
7156 |
+
"grad_norm": 2.1439051628112793,
|
7157 |
+
"learning_rate": 4.097430623774921e-05,
|
7158 |
+
"loss": 0.2172,
|
7159 |
+
"step": 1018
|
7160 |
+
},
|
7161 |
+
{
|
7162 |
+
"epoch": 0.590296886314265,
|
7163 |
+
"grad_norm": 2.0244362354278564,
|
7164 |
+
"learning_rate": 4.079809521891933e-05,
|
7165 |
+
"loss": 0.1482,
|
7166 |
+
"step": 1019
|
7167 |
+
},
|
7168 |
+
{
|
7169 |
+
"epoch": 0.5908761766835626,
|
7170 |
+
"grad_norm": 1.860674500465393,
|
7171 |
+
"learning_rate": 4.062216676451285e-05,
|
7172 |
+
"loss": 0.0855,
|
7173 |
+
"step": 1020
|
7174 |
+
},
|
7175 |
+
{
|
7176 |
+
"epoch": 0.5914554670528602,
|
7177 |
+
"grad_norm": 1.556273102760315,
|
7178 |
+
"learning_rate": 4.04465217142177e-05,
|
7179 |
+
"loss": 0.0782,
|
7180 |
+
"step": 1021
|
7181 |
+
},
|
7182 |
+
{
|
7183 |
+
"epoch": 0.5920347574221578,
|
7184 |
+
"grad_norm": 0.7192368507385254,
|
7185 |
+
"learning_rate": 4.027116090636909e-05,
|
7186 |
+
"loss": 0.0263,
|
7187 |
+
"step": 1022
|
7188 |
+
},
|
7189 |
+
{
|
7190 |
+
"epoch": 0.5926140477914554,
|
7191 |
+
"grad_norm": 2.22908353805542,
|
7192 |
+
"learning_rate": 4.0096085177945586e-05,
|
7193 |
+
"loss": 0.0979,
|
7194 |
+
"step": 1023
|
7195 |
+
},
|
7196 |
+
{
|
7197 |
+
"epoch": 0.593193338160753,
|
7198 |
+
"grad_norm": 2.0853970050811768,
|
7199 |
+
"learning_rate": 3.9921295364565094e-05,
|
7200 |
+
"loss": 0.1206,
|
7201 |
+
"step": 1024
|
7202 |
+
},
|
7203 |
+
{
|
7204 |
+
"epoch": 0.5937726285300506,
|
7205 |
+
"grad_norm": 2.7167630195617676,
|
7206 |
+
"learning_rate": 3.9746792300480894e-05,
|
7207 |
+
"loss": 0.1885,
|
7208 |
+
"step": 1025
|
7209 |
+
},
|
7210 |
+
{
|
7211 |
+
"epoch": 0.5943519188993484,
|
7212 |
+
"grad_norm": 2.2852001190185547,
|
7213 |
+
"learning_rate": 3.9572576818577637e-05,
|
7214 |
+
"loss": 0.1774,
|
7215 |
+
"step": 1026
|
7216 |
+
},
|
7217 |
+
{
|
7218 |
+
"epoch": 0.594931209268646,
|
7219 |
+
"grad_norm": 1.4725953340530396,
|
7220 |
+
"learning_rate": 3.9398649750367365e-05,
|
7221 |
+
"loss": 0.125,
|
7222 |
+
"step": 1027
|
7223 |
+
},
|
7224 |
+
{
|
7225 |
+
"epoch": 0.5955104996379436,
|
7226 |
+
"grad_norm": 1.077785849571228,
|
7227 |
+
"learning_rate": 3.922501192598557e-05,
|
7228 |
+
"loss": 0.0584,
|
7229 |
+
"step": 1028
|
7230 |
+
},
|
7231 |
+
{
|
7232 |
+
"epoch": 0.5960897900072412,
|
7233 |
+
"grad_norm": 2.285592555999756,
|
7234 |
+
"learning_rate": 3.9051664174187154e-05,
|
7235 |
+
"loss": 0.1603,
|
7236 |
+
"step": 1029
|
7237 |
+
},
|
7238 |
+
{
|
7239 |
+
"epoch": 0.5966690803765388,
|
7240 |
+
"grad_norm": 1.5708166360855103,
|
7241 |
+
"learning_rate": 3.8878607322342674e-05,
|
7242 |
+
"loss": 0.1087,
|
7243 |
+
"step": 1030
|
7244 |
+
},
|
7245 |
+
{
|
7246 |
+
"epoch": 0.5972483707458364,
|
7247 |
+
"grad_norm": 3.158360481262207,
|
7248 |
+
"learning_rate": 3.87058421964341e-05,
|
7249 |
+
"loss": 0.0728,
|
7250 |
+
"step": 1031
|
7251 |
+
},
|
7252 |
+
{
|
7253 |
+
"epoch": 0.597827661115134,
|
7254 |
+
"grad_norm": 2.05562162399292,
|
7255 |
+
"learning_rate": 3.853336962105112e-05,
|
7256 |
+
"loss": 0.1444,
|
7257 |
+
"step": 1032
|
7258 |
+
},
|
7259 |
+
{
|
7260 |
+
"epoch": 0.5984069514844316,
|
7261 |
+
"grad_norm": 1.0003705024719238,
|
7262 |
+
"learning_rate": 3.836119041938704e-05,
|
7263 |
+
"loss": 0.0551,
|
7264 |
+
"step": 1033
|
7265 |
+
},
|
7266 |
+
{
|
7267 |
+
"epoch": 0.5989862418537292,
|
7268 |
+
"grad_norm": 2.253399133682251,
|
7269 |
+
"learning_rate": 3.8189305413234975e-05,
|
7270 |
+
"loss": 0.1458,
|
7271 |
+
"step": 1034
|
7272 |
+
},
|
7273 |
+
{
|
7274 |
+
"epoch": 0.5995655322230268,
|
7275 |
+
"grad_norm": 1.7917815446853638,
|
7276 |
+
"learning_rate": 3.801771542298387e-05,
|
7277 |
+
"loss": 0.083,
|
7278 |
+
"step": 1035
|
7279 |
+
},
|
7280 |
+
{
|
7281 |
+
"epoch": 0.6001448225923244,
|
7282 |
+
"grad_norm": 1.4274410009384155,
|
7283 |
+
"learning_rate": 3.784642126761457e-05,
|
7284 |
+
"loss": 0.0903,
|
7285 |
+
"step": 1036
|
7286 |
+
},
|
7287 |
+
{
|
7288 |
+
"epoch": 0.600724112961622,
|
7289 |
+
"grad_norm": 1.5352691411972046,
|
7290 |
+
"learning_rate": 3.767542376469591e-05,
|
7291 |
+
"loss": 0.0471,
|
7292 |
+
"step": 1037
|
7293 |
+
},
|
7294 |
+
{
|
7295 |
+
"epoch": 0.6013034033309196,
|
7296 |
+
"grad_norm": 1.7276948690414429,
|
7297 |
+
"learning_rate": 3.750472373038093e-05,
|
7298 |
+
"loss": 0.0623,
|
7299 |
+
"step": 1038
|
7300 |
+
},
|
7301 |
+
{
|
7302 |
+
"epoch": 0.6018826937002172,
|
7303 |
+
"grad_norm": 2.8179798126220703,
|
7304 |
+
"learning_rate": 3.733432197940277e-05,
|
7305 |
+
"loss": 0.1997,
|
7306 |
+
"step": 1039
|
7307 |
+
},
|
7308 |
+
{
|
7309 |
+
"epoch": 0.6024619840695149,
|
7310 |
+
"grad_norm": 1.1214691400527954,
|
7311 |
+
"learning_rate": 3.7164219325070995e-05,
|
7312 |
+
"loss": 0.041,
|
7313 |
+
"step": 1040
|
7314 |
+
},
|
7315 |
+
{
|
7316 |
+
"epoch": 0.6030412744388125,
|
7317 |
+
"grad_norm": 1.1556260585784912,
|
7318 |
+
"learning_rate": 3.699441657926755e-05,
|
7319 |
+
"loss": 0.047,
|
7320 |
+
"step": 1041
|
7321 |
+
},
|
7322 |
+
{
|
7323 |
+
"epoch": 0.6036205648081101,
|
7324 |
+
"grad_norm": 1.069880485534668,
|
7325 |
+
"learning_rate": 3.682491455244294e-05,
|
7326 |
+
"loss": 0.0731,
|
7327 |
+
"step": 1042
|
7328 |
+
},
|
7329 |
+
{
|
7330 |
+
"epoch": 0.6041998551774077,
|
7331 |
+
"grad_norm": 1.0165038108825684,
|
7332 |
+
"learning_rate": 3.665571405361239e-05,
|
7333 |
+
"loss": 0.0565,
|
7334 |
+
"step": 1043
|
7335 |
+
},
|
7336 |
+
{
|
7337 |
+
"epoch": 0.6047791455467053,
|
7338 |
+
"grad_norm": 1.1815634965896606,
|
7339 |
+
"learning_rate": 3.6486815890352e-05,
|
7340 |
+
"loss": 0.0541,
|
7341 |
+
"step": 1044
|
7342 |
+
},
|
7343 |
+
{
|
7344 |
+
"epoch": 0.6053584359160029,
|
7345 |
+
"grad_norm": 1.7292768955230713,
|
7346 |
+
"learning_rate": 3.6318220868794784e-05,
|
7347 |
+
"loss": 0.1484,
|
7348 |
+
"step": 1045
|
7349 |
+
},
|
7350 |
+
{
|
7351 |
+
"epoch": 0.6059377262853005,
|
7352 |
+
"grad_norm": 1.3332797288894653,
|
7353 |
+
"learning_rate": 3.6149929793626954e-05,
|
7354 |
+
"loss": 0.0422,
|
7355 |
+
"step": 1046
|
7356 |
+
},
|
7357 |
+
{
|
7358 |
+
"epoch": 0.6065170166545981,
|
7359 |
+
"grad_norm": 2.8485829830169678,
|
7360 |
+
"learning_rate": 3.598194346808399e-05,
|
7361 |
+
"loss": 0.1431,
|
7362 |
+
"step": 1047
|
7363 |
+
},
|
7364 |
+
{
|
7365 |
+
"epoch": 0.6070963070238957,
|
7366 |
+
"grad_norm": 2.2351937294006348,
|
7367 |
+
"learning_rate": 3.581426269394679e-05,
|
7368 |
+
"loss": 0.1693,
|
7369 |
+
"step": 1048
|
7370 |
+
},
|
7371 |
+
{
|
7372 |
+
"epoch": 0.6076755973931933,
|
7373 |
+
"grad_norm": 1.5097788572311401,
|
7374 |
+
"learning_rate": 3.5646888271538014e-05,
|
7375 |
+
"loss": 0.0572,
|
7376 |
+
"step": 1049
|
7377 |
+
},
|
7378 |
+
{
|
7379 |
+
"epoch": 0.6082548877624909,
|
7380 |
+
"grad_norm": 1.114818811416626,
|
7381 |
+
"learning_rate": 3.5479820999718036e-05,
|
7382 |
+
"loss": 0.0772,
|
7383 |
+
"step": 1050
|
7384 |
+
},
|
7385 |
+
{
|
7386 |
+
"epoch": 0.6088341781317885,
|
7387 |
+
"grad_norm": 5.17358922958374,
|
7388 |
+
"learning_rate": 3.531306167588123e-05,
|
7389 |
+
"loss": 0.4743,
|
7390 |
+
"step": 1051
|
7391 |
+
},
|
7392 |
+
{
|
7393 |
+
"epoch": 0.6094134685010861,
|
7394 |
+
"grad_norm": 0.682974100112915,
|
7395 |
+
"learning_rate": 3.514661109595222e-05,
|
7396 |
+
"loss": 0.0337,
|
7397 |
+
"step": 1052
|
7398 |
+
},
|
7399 |
+
{
|
7400 |
+
"epoch": 0.6099927588703837,
|
7401 |
+
"grad_norm": 1.268609642982483,
|
7402 |
+
"learning_rate": 3.4980470054381975e-05,
|
7403 |
+
"loss": 0.0553,
|
7404 |
+
"step": 1053
|
7405 |
+
},
|
7406 |
+
{
|
7407 |
+
"epoch": 0.6105720492396813,
|
7408 |
+
"grad_norm": 2.0010392665863037,
|
7409 |
+
"learning_rate": 3.481463934414412e-05,
|
7410 |
+
"loss": 0.091,
|
7411 |
+
"step": 1054
|
7412 |
+
},
|
7413 |
+
{
|
7414 |
+
"epoch": 0.611151339608979,
|
7415 |
+
"grad_norm": 1.4881786108016968,
|
7416 |
+
"learning_rate": 3.4649119756731055e-05,
|
7417 |
+
"loss": 0.0445,
|
7418 |
+
"step": 1055
|
7419 |
+
},
|
7420 |
+
{
|
7421 |
+
"epoch": 0.6117306299782767,
|
7422 |
+
"grad_norm": 3.6607894897460938,
|
7423 |
+
"learning_rate": 3.4483912082150225e-05,
|
7424 |
+
"loss": 0.3531,
|
7425 |
+
"step": 1056
|
7426 |
+
},
|
7427 |
+
{
|
7428 |
+
"epoch": 0.6123099203475743,
|
7429 |
+
"grad_norm": 1.652235984802246,
|
7430 |
+
"learning_rate": 3.4319017108920435e-05,
|
7431 |
+
"loss": 0.0932,
|
7432 |
+
"step": 1057
|
7433 |
+
},
|
7434 |
+
{
|
7435 |
+
"epoch": 0.6128892107168719,
|
7436 |
+
"grad_norm": 0.7218719124794006,
|
7437 |
+
"learning_rate": 3.4154435624067904e-05,
|
7438 |
+
"loss": 0.0407,
|
7439 |
+
"step": 1058
|
7440 |
+
},
|
7441 |
+
{
|
7442 |
+
"epoch": 0.6134685010861695,
|
7443 |
+
"grad_norm": 1.194688081741333,
|
7444 |
+
"learning_rate": 3.3990168413122625e-05,
|
7445 |
+
"loss": 0.0394,
|
7446 |
+
"step": 1059
|
7447 |
+
},
|
7448 |
+
{
|
7449 |
+
"epoch": 0.6140477914554671,
|
7450 |
+
"grad_norm": 1.181309461593628,
|
7451 |
+
"learning_rate": 3.3826216260114604e-05,
|
7452 |
+
"loss": 0.052,
|
7453 |
+
"step": 1060
|
7454 |
+
},
|
7455 |
+
{
|
7456 |
+
"epoch": 0.6146270818247647,
|
7457 |
+
"grad_norm": 1.4079760313034058,
|
7458 |
+
"learning_rate": 3.366257994757013e-05,
|
7459 |
+
"loss": 0.0879,
|
7460 |
+
"step": 1061
|
7461 |
+
},
|
7462 |
+
{
|
7463 |
+
"epoch": 0.6152063721940623,
|
7464 |
+
"grad_norm": 0.7758262157440186,
|
7465 |
+
"learning_rate": 3.349926025650796e-05,
|
7466 |
+
"loss": 0.0339,
|
7467 |
+
"step": 1062
|
7468 |
+
},
|
7469 |
+
{
|
7470 |
+
"epoch": 0.6157856625633599,
|
7471 |
+
"grad_norm": 4.113604545593262,
|
7472 |
+
"learning_rate": 3.333625796643572e-05,
|
7473 |
+
"loss": 0.1483,
|
7474 |
+
"step": 1063
|
7475 |
+
},
|
7476 |
+
{
|
7477 |
+
"epoch": 0.6163649529326575,
|
7478 |
+
"grad_norm": 3.244152784347534,
|
7479 |
+
"learning_rate": 3.3173573855346043e-05,
|
7480 |
+
"loss": 0.137,
|
7481 |
+
"step": 1064
|
7482 |
+
},
|
7483 |
+
{
|
7484 |
+
"epoch": 0.6169442433019551,
|
7485 |
+
"grad_norm": 1.1724659204483032,
|
7486 |
+
"learning_rate": 3.3011208699713015e-05,
|
7487 |
+
"loss": 0.0713,
|
7488 |
+
"step": 1065
|
7489 |
+
},
|
7490 |
+
{
|
7491 |
+
"epoch": 0.6175235336712527,
|
7492 |
+
"grad_norm": 2.5028889179229736,
|
7493 |
+
"learning_rate": 3.2849163274488314e-05,
|
7494 |
+
"loss": 0.1607,
|
7495 |
+
"step": 1066
|
7496 |
+
},
|
7497 |
+
{
|
7498 |
+
"epoch": 0.6181028240405503,
|
7499 |
+
"grad_norm": 3.646338939666748,
|
7500 |
+
"learning_rate": 3.268743835309757e-05,
|
7501 |
+
"loss": 0.3208,
|
7502 |
+
"step": 1067
|
7503 |
+
},
|
7504 |
+
{
|
7505 |
+
"epoch": 0.6186821144098479,
|
7506 |
+
"grad_norm": 2.7392494678497314,
|
7507 |
+
"learning_rate": 3.252603470743676e-05,
|
7508 |
+
"loss": 0.1198,
|
7509 |
+
"step": 1068
|
7510 |
+
},
|
7511 |
+
{
|
7512 |
+
"epoch": 0.6192614047791456,
|
7513 |
+
"grad_norm": 1.6429647207260132,
|
7514 |
+
"learning_rate": 3.2364953107868376e-05,
|
7515 |
+
"loss": 0.0745,
|
7516 |
+
"step": 1069
|
7517 |
+
},
|
7518 |
+
{
|
7519 |
+
"epoch": 0.6198406951484432,
|
7520 |
+
"grad_norm": 1.132997989654541,
|
7521 |
+
"learning_rate": 3.220419432321783e-05,
|
7522 |
+
"loss": 0.026,
|
7523 |
+
"step": 1070
|
7524 |
+
},
|
7525 |
+
{
|
7526 |
+
"epoch": 0.6204199855177408,
|
7527 |
+
"grad_norm": 1.1884487867355347,
|
7528 |
+
"learning_rate": 3.204375912076979e-05,
|
7529 |
+
"loss": 0.0835,
|
7530 |
+
"step": 1071
|
7531 |
+
},
|
7532 |
+
{
|
7533 |
+
"epoch": 0.6209992758870384,
|
7534 |
+
"grad_norm": 1.6805810928344727,
|
7535 |
+
"learning_rate": 3.188364826626449e-05,
|
7536 |
+
"loss": 0.0588,
|
7537 |
+
"step": 1072
|
7538 |
+
},
|
7539 |
+
{
|
7540 |
+
"epoch": 0.621578566256336,
|
7541 |
+
"grad_norm": 4.5599846839904785,
|
7542 |
+
"learning_rate": 3.172386252389408e-05,
|
7543 |
+
"loss": 0.3706,
|
7544 |
+
"step": 1073
|
7545 |
+
},
|
7546 |
+
{
|
7547 |
+
"epoch": 0.6221578566256336,
|
7548 |
+
"grad_norm": 1.351007103919983,
|
7549 |
+
"learning_rate": 3.156440265629899e-05,
|
7550 |
+
"loss": 0.0848,
|
7551 |
+
"step": 1074
|
7552 |
+
},
|
7553 |
+
{
|
7554 |
+
"epoch": 0.6227371469949312,
|
7555 |
+
"grad_norm": 4.964895725250244,
|
7556 |
+
"learning_rate": 3.1405269424564244e-05,
|
7557 |
+
"loss": 0.2099,
|
7558 |
+
"step": 1075
|
7559 |
+
},
|
7560 |
+
{
|
7561 |
+
"epoch": 0.6233164373642288,
|
7562 |
+
"grad_norm": 2.0441789627075195,
|
7563 |
+
"learning_rate": 3.124646358821599e-05,
|
7564 |
+
"loss": 0.0642,
|
7565 |
+
"step": 1076
|
7566 |
+
},
|
7567 |
+
{
|
7568 |
+
"epoch": 0.6238957277335264,
|
7569 |
+
"grad_norm": 0.7961364984512329,
|
7570 |
+
"learning_rate": 3.108798590521764e-05,
|
7571 |
+
"loss": 0.0386,
|
7572 |
+
"step": 1077
|
7573 |
+
},
|
7574 |
+
{
|
7575 |
+
"epoch": 0.624475018102824,
|
7576 |
+
"grad_norm": 1.0998642444610596,
|
7577 |
+
"learning_rate": 3.092983713196639e-05,
|
7578 |
+
"loss": 0.0583,
|
7579 |
+
"step": 1078
|
7580 |
+
},
|
7581 |
+
{
|
7582 |
+
"epoch": 0.6250543084721216,
|
7583 |
+
"grad_norm": 1.602994441986084,
|
7584 |
+
"learning_rate": 3.077201802328958e-05,
|
7585 |
+
"loss": 0.0845,
|
7586 |
+
"step": 1079
|
7587 |
+
},
|
7588 |
+
{
|
7589 |
+
"epoch": 0.6256335988414192,
|
7590 |
+
"grad_norm": 3.1479015350341797,
|
7591 |
+
"learning_rate": 3.061452933244112e-05,
|
7592 |
+
"loss": 0.3006,
|
7593 |
+
"step": 1080
|
7594 |
+
},
|
7595 |
+
{
|
7596 |
+
"epoch": 0.6262128892107168,
|
7597 |
+
"grad_norm": 0.6543728709220886,
|
7598 |
+
"learning_rate": 3.0457371811097857e-05,
|
7599 |
+
"loss": 0.0303,
|
7600 |
+
"step": 1081
|
7601 |
+
},
|
7602 |
+
{
|
7603 |
+
"epoch": 0.6267921795800144,
|
7604 |
+
"grad_norm": 0.7708796858787537,
|
7605 |
+
"learning_rate": 3.0300546209355995e-05,
|
7606 |
+
"loss": 0.049,
|
7607 |
+
"step": 1082
|
7608 |
+
},
|
7609 |
+
{
|
7610 |
+
"epoch": 0.6273714699493121,
|
7611 |
+
"grad_norm": 1.3549633026123047,
|
7612 |
+
"learning_rate": 3.0144053275727533e-05,
|
7613 |
+
"loss": 0.0686,
|
7614 |
+
"step": 1083
|
7615 |
+
},
|
7616 |
+
{
|
7617 |
+
"epoch": 0.6279507603186097,
|
7618 |
+
"grad_norm": 1.6762343645095825,
|
7619 |
+
"learning_rate": 2.998789375713663e-05,
|
7620 |
+
"loss": 0.0832,
|
7621 |
+
"step": 1084
|
7622 |
+
},
|
7623 |
+
{
|
7624 |
+
"epoch": 0.6285300506879073,
|
7625 |
+
"grad_norm": 0.504910409450531,
|
7626 |
+
"learning_rate": 2.9832068398916212e-05,
|
7627 |
+
"loss": 0.0185,
|
7628 |
+
"step": 1085
|
7629 |
+
},
|
7630 |
+
{
|
7631 |
+
"epoch": 0.629109341057205,
|
7632 |
+
"grad_norm": 2.428058385848999,
|
7633 |
+
"learning_rate": 2.9676577944804175e-05,
|
7634 |
+
"loss": 0.0913,
|
7635 |
+
"step": 1086
|
7636 |
+
},
|
7637 |
+
{
|
7638 |
+
"epoch": 0.629109341057205,
|
7639 |
+
"eval_loss": 0.10047201067209244,
|
7640 |
+
"eval_runtime": 24.4476,
|
7641 |
+
"eval_samples_per_second": 29.737,
|
7642 |
+
"eval_steps_per_second": 14.889,
|
7643 |
+
"step": 1086
|
7644 |
}
|
7645 |
],
|
7646 |
"logging_steps": 1,
|
|
|
7660 |
"attributes": {}
|
7661 |
}
|
7662 |
},
|
7663 |
+
"total_flos": 2.0405317567669862e+17,
|
7664 |
"train_batch_size": 2,
|
7665 |
"trial_name": null,
|
7666 |
"trial_params": null
|