Training in progress, step 1125, 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:b87fb5f16e86c333a7a579128323be3c8bfaef276b11d4ddac741d59eaf7bdf4
|
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 sha256:3296bee3d2f7bd2b087c3cc1e82fb1981750076032afa6f0e7a96726c5da29a8
|
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 sha256:48f5a8e138474e02659ec1fcd1a6c50eb579bc3c56b68a9bdae3c862c006c82f
|
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 sha256:966cc7c77768098c8af029c048689ac1de1ed0de938435a2a7a5601075335dbd
|
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": 375,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -5273,6 +5273,2639 @@
|
|
5273 |
"eval_samples_per_second": 16.654,
|
5274 |
"eval_steps_per_second": 8.327,
|
5275 |
"step": 750
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5276 |
}
|
5277 |
],
|
5278 |
"logging_steps": 1,
|
@@ -5292,7 +7925,7 @@
|
|
5292 |
"attributes": {}
|
5293 |
}
|
5294 |
},
|
5295 |
-
"total_flos": 1.
|
5296 |
"train_batch_size": 2,
|
5297 |
"trial_name": null,
|
5298 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.0735195010476529,
|
5 |
"eval_steps": 375,
|
6 |
+
"global_step": 1125,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
5273 |
"eval_samples_per_second": 16.654,
|
5274 |
"eval_steps_per_second": 8.327,
|
5275 |
"step": 750
|
5276 |
+
},
|
5277 |
+
{
|
5278 |
+
"epoch": 0.049078351366033175,
|
5279 |
+
"grad_norm": 21.074249267578125,
|
5280 |
+
"learning_rate": 0.00010084337058003303,
|
5281 |
+
"loss": 3.8955,
|
5282 |
+
"step": 751
|
5283 |
+
},
|
5284 |
+
{
|
5285 |
+
"epoch": 0.04914370203363109,
|
5286 |
+
"grad_norm": 18.855222702026367,
|
5287 |
+
"learning_rate": 0.00010063253121564868,
|
5288 |
+
"loss": 3.6844,
|
5289 |
+
"step": 752
|
5290 |
+
},
|
5291 |
+
{
|
5292 |
+
"epoch": 0.049209052701229,
|
5293 |
+
"grad_norm": 8.403053283691406,
|
5294 |
+
"learning_rate": 0.00010042168903930514,
|
5295 |
+
"loss": 3.4876,
|
5296 |
+
"step": 753
|
5297 |
+
},
|
5298 |
+
{
|
5299 |
+
"epoch": 0.049274403368826915,
|
5300 |
+
"grad_norm": 11.068482398986816,
|
5301 |
+
"learning_rate": 0.00010021084498831522,
|
5302 |
+
"loss": 3.5283,
|
5303 |
+
"step": 754
|
5304 |
+
},
|
5305 |
+
{
|
5306 |
+
"epoch": 0.04933975403642483,
|
5307 |
+
"grad_norm": 7.6804914474487305,
|
5308 |
+
"learning_rate": 0.0001,
|
5309 |
+
"loss": 3.1875,
|
5310 |
+
"step": 755
|
5311 |
+
},
|
5312 |
+
{
|
5313 |
+
"epoch": 0.04940510470402274,
|
5314 |
+
"grad_norm": 2.981393575668335,
|
5315 |
+
"learning_rate": 9.97891550116848e-05,
|
5316 |
+
"loss": 3.2444,
|
5317 |
+
"step": 756
|
5318 |
+
},
|
5319 |
+
{
|
5320 |
+
"epoch": 0.049470455371620654,
|
5321 |
+
"grad_norm": 4.217319488525391,
|
5322 |
+
"learning_rate": 9.957831096069488e-05,
|
5323 |
+
"loss": 3.2795,
|
5324 |
+
"step": 757
|
5325 |
+
},
|
5326 |
+
{
|
5327 |
+
"epoch": 0.04953580603921857,
|
5328 |
+
"grad_norm": 2.8416085243225098,
|
5329 |
+
"learning_rate": 9.936746878435136e-05,
|
5330 |
+
"loss": 3.2007,
|
5331 |
+
"step": 758
|
5332 |
+
},
|
5333 |
+
{
|
5334 |
+
"epoch": 0.04960115670681648,
|
5335 |
+
"grad_norm": 2.157623529434204,
|
5336 |
+
"learning_rate": 9.915662941996699e-05,
|
5337 |
+
"loss": 3.0093,
|
5338 |
+
"step": 759
|
5339 |
+
},
|
5340 |
+
{
|
5341 |
+
"epoch": 0.049666507374414394,
|
5342 |
+
"grad_norm": 2.7508373260498047,
|
5343 |
+
"learning_rate": 9.894579380484204e-05,
|
5344 |
+
"loss": 3.1868,
|
5345 |
+
"step": 760
|
5346 |
+
},
|
5347 |
+
{
|
5348 |
+
"epoch": 0.04973185804201231,
|
5349 |
+
"grad_norm": 2.0463340282440186,
|
5350 |
+
"learning_rate": 9.873496287626019e-05,
|
5351 |
+
"loss": 3.0286,
|
5352 |
+
"step": 761
|
5353 |
+
},
|
5354 |
+
{
|
5355 |
+
"epoch": 0.04979720870961023,
|
5356 |
+
"grad_norm": 1.8285865783691406,
|
5357 |
+
"learning_rate": 9.852413757148417e-05,
|
5358 |
+
"loss": 3.008,
|
5359 |
+
"step": 762
|
5360 |
+
},
|
5361 |
+
{
|
5362 |
+
"epoch": 0.04986255937720814,
|
5363 |
+
"grad_norm": 1.9253116846084595,
|
5364 |
+
"learning_rate": 9.831331882775178e-05,
|
5365 |
+
"loss": 3.0803,
|
5366 |
+
"step": 763
|
5367 |
+
},
|
5368 |
+
{
|
5369 |
+
"epoch": 0.049927910044806054,
|
5370 |
+
"grad_norm": 1.7544878721237183,
|
5371 |
+
"learning_rate": 9.81025075822716e-05,
|
5372 |
+
"loss": 2.9899,
|
5373 |
+
"step": 764
|
5374 |
+
},
|
5375 |
+
{
|
5376 |
+
"epoch": 0.04999326071240397,
|
5377 |
+
"grad_norm": 2.110278367996216,
|
5378 |
+
"learning_rate": 9.789170477221891e-05,
|
5379 |
+
"loss": 3.2229,
|
5380 |
+
"step": 765
|
5381 |
+
},
|
5382 |
+
{
|
5383 |
+
"epoch": 0.05005861138000188,
|
5384 |
+
"grad_norm": 1.698500633239746,
|
5385 |
+
"learning_rate": 9.76809113347315e-05,
|
5386 |
+
"loss": 2.943,
|
5387 |
+
"step": 766
|
5388 |
+
},
|
5389 |
+
{
|
5390 |
+
"epoch": 0.05012396204759979,
|
5391 |
+
"grad_norm": 1.850306749343872,
|
5392 |
+
"learning_rate": 9.747012820690543e-05,
|
5393 |
+
"loss": 3.178,
|
5394 |
+
"step": 767
|
5395 |
+
},
|
5396 |
+
{
|
5397 |
+
"epoch": 0.050189312715197706,
|
5398 |
+
"grad_norm": 1.7629265785217285,
|
5399 |
+
"learning_rate": 9.725935632579104e-05,
|
5400 |
+
"loss": 2.957,
|
5401 |
+
"step": 768
|
5402 |
+
},
|
5403 |
+
{
|
5404 |
+
"epoch": 0.05025466338279562,
|
5405 |
+
"grad_norm": 2.0044591426849365,
|
5406 |
+
"learning_rate": 9.704859662838855e-05,
|
5407 |
+
"loss": 3.1231,
|
5408 |
+
"step": 769
|
5409 |
+
},
|
5410 |
+
{
|
5411 |
+
"epoch": 0.05032001405039353,
|
5412 |
+
"grad_norm": 1.8281937837600708,
|
5413 |
+
"learning_rate": 9.683785005164411e-05,
|
5414 |
+
"loss": 2.944,
|
5415 |
+
"step": 770
|
5416 |
+
},
|
5417 |
+
{
|
5418 |
+
"epoch": 0.050385364717991446,
|
5419 |
+
"grad_norm": 2.0414679050445557,
|
5420 |
+
"learning_rate": 9.662711753244551e-05,
|
5421 |
+
"loss": 3.1581,
|
5422 |
+
"step": 771
|
5423 |
+
},
|
5424 |
+
{
|
5425 |
+
"epoch": 0.05045071538558936,
|
5426 |
+
"grad_norm": 1.8195239305496216,
|
5427 |
+
"learning_rate": 9.641640000761802e-05,
|
5428 |
+
"loss": 2.9227,
|
5429 |
+
"step": 772
|
5430 |
+
},
|
5431 |
+
{
|
5432 |
+
"epoch": 0.05051606605318727,
|
5433 |
+
"grad_norm": 1.9885085821151733,
|
5434 |
+
"learning_rate": 9.620569841392029e-05,
|
5435 |
+
"loss": 3.2418,
|
5436 |
+
"step": 773
|
5437 |
+
},
|
5438 |
+
{
|
5439 |
+
"epoch": 0.050581416720785186,
|
5440 |
+
"grad_norm": 2.3411426544189453,
|
5441 |
+
"learning_rate": 9.59950136880401e-05,
|
5442 |
+
"loss": 3.0666,
|
5443 |
+
"step": 774
|
5444 |
+
},
|
5445 |
+
{
|
5446 |
+
"epoch": 0.0506467673883831,
|
5447 |
+
"grad_norm": 1.9475857019424438,
|
5448 |
+
"learning_rate": 9.57843467665903e-05,
|
5449 |
+
"loss": 2.867,
|
5450 |
+
"step": 775
|
5451 |
+
},
|
5452 |
+
{
|
5453 |
+
"epoch": 0.05071211805598102,
|
5454 |
+
"grad_norm": 2.2501626014709473,
|
5455 |
+
"learning_rate": 9.557369858610453e-05,
|
5456 |
+
"loss": 3.0415,
|
5457 |
+
"step": 776
|
5458 |
+
},
|
5459 |
+
{
|
5460 |
+
"epoch": 0.05077746872357893,
|
5461 |
+
"grad_norm": 2.332026481628418,
|
5462 |
+
"learning_rate": 9.53630700830332e-05,
|
5463 |
+
"loss": 3.0397,
|
5464 |
+
"step": 777
|
5465 |
+
},
|
5466 |
+
{
|
5467 |
+
"epoch": 0.050842819391176845,
|
5468 |
+
"grad_norm": 2.243842601776123,
|
5469 |
+
"learning_rate": 9.51524621937391e-05,
|
5470 |
+
"loss": 2.9786,
|
5471 |
+
"step": 778
|
5472 |
+
},
|
5473 |
+
{
|
5474 |
+
"epoch": 0.05090817005877476,
|
5475 |
+
"grad_norm": 2.1287598609924316,
|
5476 |
+
"learning_rate": 9.494187585449358e-05,
|
5477 |
+
"loss": 2.9306,
|
5478 |
+
"step": 779
|
5479 |
+
},
|
5480 |
+
{
|
5481 |
+
"epoch": 0.05097352072637267,
|
5482 |
+
"grad_norm": 2.5319623947143555,
|
5483 |
+
"learning_rate": 9.473131200147205e-05,
|
5484 |
+
"loss": 3.0601,
|
5485 |
+
"step": 780
|
5486 |
+
},
|
5487 |
+
{
|
5488 |
+
"epoch": 0.051038871393970585,
|
5489 |
+
"grad_norm": 2.228027820587158,
|
5490 |
+
"learning_rate": 9.452077157074994e-05,
|
5491 |
+
"loss": 3.0304,
|
5492 |
+
"step": 781
|
5493 |
+
},
|
5494 |
+
{
|
5495 |
+
"epoch": 0.0511042220615685,
|
5496 |
+
"grad_norm": 2.4075875282287598,
|
5497 |
+
"learning_rate": 9.431025549829862e-05,
|
5498 |
+
"loss": 3.0306,
|
5499 |
+
"step": 782
|
5500 |
+
},
|
5501 |
+
{
|
5502 |
+
"epoch": 0.05116957272916641,
|
5503 |
+
"grad_norm": 2.6555421352386475,
|
5504 |
+
"learning_rate": 9.409976471998118e-05,
|
5505 |
+
"loss": 3.1512,
|
5506 |
+
"step": 783
|
5507 |
+
},
|
5508 |
+
{
|
5509 |
+
"epoch": 0.051234923396764324,
|
5510 |
+
"grad_norm": 2.4105820655822754,
|
5511 |
+
"learning_rate": 9.388930017154819e-05,
|
5512 |
+
"loss": 2.9704,
|
5513 |
+
"step": 784
|
5514 |
+
},
|
5515 |
+
{
|
5516 |
+
"epoch": 0.05130027406436224,
|
5517 |
+
"grad_norm": 2.718679428100586,
|
5518 |
+
"learning_rate": 9.367886278863366e-05,
|
5519 |
+
"loss": 3.2359,
|
5520 |
+
"step": 785
|
5521 |
+
},
|
5522 |
+
{
|
5523 |
+
"epoch": 0.05136562473196015,
|
5524 |
+
"grad_norm": 2.5873093605041504,
|
5525 |
+
"learning_rate": 9.346845350675088e-05,
|
5526 |
+
"loss": 2.8897,
|
5527 |
+
"step": 786
|
5528 |
+
},
|
5529 |
+
{
|
5530 |
+
"epoch": 0.051430975399558064,
|
5531 |
+
"grad_norm": 3.255326509475708,
|
5532 |
+
"learning_rate": 9.325807326128814e-05,
|
5533 |
+
"loss": 3.0515,
|
5534 |
+
"step": 787
|
5535 |
+
},
|
5536 |
+
{
|
5537 |
+
"epoch": 0.05149632606715598,
|
5538 |
+
"grad_norm": 2.9990360736846924,
|
5539 |
+
"learning_rate": 9.304772298750463e-05,
|
5540 |
+
"loss": 3.0665,
|
5541 |
+
"step": 788
|
5542 |
+
},
|
5543 |
+
{
|
5544 |
+
"epoch": 0.05156167673475389,
|
5545 |
+
"grad_norm": 2.877439022064209,
|
5546 |
+
"learning_rate": 9.283740362052642e-05,
|
5547 |
+
"loss": 3.0528,
|
5548 |
+
"step": 789
|
5549 |
+
},
|
5550 |
+
{
|
5551 |
+
"epoch": 0.05162702740235181,
|
5552 |
+
"grad_norm": 2.7675907611846924,
|
5553 |
+
"learning_rate": 9.26271160953421e-05,
|
5554 |
+
"loss": 2.9353,
|
5555 |
+
"step": 790
|
5556 |
+
},
|
5557 |
+
{
|
5558 |
+
"epoch": 0.051692378069949724,
|
5559 |
+
"grad_norm": 3.4324252605438232,
|
5560 |
+
"learning_rate": 9.241686134679867e-05,
|
5561 |
+
"loss": 2.8613,
|
5562 |
+
"step": 791
|
5563 |
+
},
|
5564 |
+
{
|
5565 |
+
"epoch": 0.05175772873754764,
|
5566 |
+
"grad_norm": 2.9301695823669434,
|
5567 |
+
"learning_rate": 9.220664030959749e-05,
|
5568 |
+
"loss": 2.6387,
|
5569 |
+
"step": 792
|
5570 |
+
},
|
5571 |
+
{
|
5572 |
+
"epoch": 0.05182307940514555,
|
5573 |
+
"grad_norm": 3.2582075595855713,
|
5574 |
+
"learning_rate": 9.199645391828999e-05,
|
5575 |
+
"loss": 3.0939,
|
5576 |
+
"step": 793
|
5577 |
+
},
|
5578 |
+
{
|
5579 |
+
"epoch": 0.05188843007274346,
|
5580 |
+
"grad_norm": 4.4022297859191895,
|
5581 |
+
"learning_rate": 9.178630310727365e-05,
|
5582 |
+
"loss": 2.8553,
|
5583 |
+
"step": 794
|
5584 |
+
},
|
5585 |
+
{
|
5586 |
+
"epoch": 0.051953780740341376,
|
5587 |
+
"grad_norm": 4.221108436584473,
|
5588 |
+
"learning_rate": 9.157618881078772e-05,
|
5589 |
+
"loss": 3.3631,
|
5590 |
+
"step": 795
|
5591 |
+
},
|
5592 |
+
{
|
5593 |
+
"epoch": 0.05201913140793929,
|
5594 |
+
"grad_norm": 4.237832069396973,
|
5595 |
+
"learning_rate": 9.136611196290915e-05,
|
5596 |
+
"loss": 3.1114,
|
5597 |
+
"step": 796
|
5598 |
+
},
|
5599 |
+
{
|
5600 |
+
"epoch": 0.0520844820755372,
|
5601 |
+
"grad_norm": 5.152820110321045,
|
5602 |
+
"learning_rate": 9.115607349754834e-05,
|
5603 |
+
"loss": 2.7191,
|
5604 |
+
"step": 797
|
5605 |
+
},
|
5606 |
+
{
|
5607 |
+
"epoch": 0.052149832743135116,
|
5608 |
+
"grad_norm": 4.774096965789795,
|
5609 |
+
"learning_rate": 9.094607434844523e-05,
|
5610 |
+
"loss": 2.9297,
|
5611 |
+
"step": 798
|
5612 |
+
},
|
5613 |
+
{
|
5614 |
+
"epoch": 0.05221518341073303,
|
5615 |
+
"grad_norm": 3.7902517318725586,
|
5616 |
+
"learning_rate": 9.07361154491648e-05,
|
5617 |
+
"loss": 2.9031,
|
5618 |
+
"step": 799
|
5619 |
+
},
|
5620 |
+
{
|
5621 |
+
"epoch": 0.05228053407833094,
|
5622 |
+
"grad_norm": 5.280699253082275,
|
5623 |
+
"learning_rate": 9.052619773309317e-05,
|
5624 |
+
"loss": 3.2022,
|
5625 |
+
"step": 800
|
5626 |
+
},
|
5627 |
+
{
|
5628 |
+
"epoch": 0.052345884745928856,
|
5629 |
+
"grad_norm": 35.736114501953125,
|
5630 |
+
"learning_rate": 9.031632213343339e-05,
|
5631 |
+
"loss": 5.3555,
|
5632 |
+
"step": 801
|
5633 |
+
},
|
5634 |
+
{
|
5635 |
+
"epoch": 0.05241123541352677,
|
5636 |
+
"grad_norm": 29.691875457763672,
|
5637 |
+
"learning_rate": 9.01064895832012e-05,
|
5638 |
+
"loss": 5.2932,
|
5639 |
+
"step": 802
|
5640 |
+
},
|
5641 |
+
{
|
5642 |
+
"epoch": 0.05247658608112468,
|
5643 |
+
"grad_norm": 23.267454147338867,
|
5644 |
+
"learning_rate": 8.98967010152211e-05,
|
5645 |
+
"loss": 4.7074,
|
5646 |
+
"step": 803
|
5647 |
+
},
|
5648 |
+
{
|
5649 |
+
"epoch": 0.052541936748722595,
|
5650 |
+
"grad_norm": 15.597977638244629,
|
5651 |
+
"learning_rate": 8.968695736212193e-05,
|
5652 |
+
"loss": 4.3194,
|
5653 |
+
"step": 804
|
5654 |
+
},
|
5655 |
+
{
|
5656 |
+
"epoch": 0.052607287416320515,
|
5657 |
+
"grad_norm": 10.932135581970215,
|
5658 |
+
"learning_rate": 8.947725955633294e-05,
|
5659 |
+
"loss": 3.974,
|
5660 |
+
"step": 805
|
5661 |
+
},
|
5662 |
+
{
|
5663 |
+
"epoch": 0.05267263808391843,
|
5664 |
+
"grad_norm": 7.222282886505127,
|
5665 |
+
"learning_rate": 8.926760853007946e-05,
|
5666 |
+
"loss": 3.5663,
|
5667 |
+
"step": 806
|
5668 |
+
},
|
5669 |
+
{
|
5670 |
+
"epoch": 0.05273798875151634,
|
5671 |
+
"grad_norm": 5.094066619873047,
|
5672 |
+
"learning_rate": 8.905800521537905e-05,
|
5673 |
+
"loss": 3.384,
|
5674 |
+
"step": 807
|
5675 |
+
},
|
5676 |
+
{
|
5677 |
+
"epoch": 0.052803339419114255,
|
5678 |
+
"grad_norm": 3.5329272747039795,
|
5679 |
+
"learning_rate": 8.884845054403699e-05,
|
5680 |
+
"loss": 3.2328,
|
5681 |
+
"step": 808
|
5682 |
+
},
|
5683 |
+
{
|
5684 |
+
"epoch": 0.05286869008671217,
|
5685 |
+
"grad_norm": 2.584484815597534,
|
5686 |
+
"learning_rate": 8.863894544764236e-05,
|
5687 |
+
"loss": 3.1267,
|
5688 |
+
"step": 809
|
5689 |
+
},
|
5690 |
+
{
|
5691 |
+
"epoch": 0.05293404075431008,
|
5692 |
+
"grad_norm": 1.915076494216919,
|
5693 |
+
"learning_rate": 8.84294908575639e-05,
|
5694 |
+
"loss": 2.7549,
|
5695 |
+
"step": 810
|
5696 |
+
},
|
5697 |
+
{
|
5698 |
+
"epoch": 0.052999391421907995,
|
5699 |
+
"grad_norm": 2.200808525085449,
|
5700 |
+
"learning_rate": 8.822008770494572e-05,
|
5701 |
+
"loss": 3.0228,
|
5702 |
+
"step": 811
|
5703 |
+
},
|
5704 |
+
{
|
5705 |
+
"epoch": 0.05306474208950591,
|
5706 |
+
"grad_norm": 2.0147557258605957,
|
5707 |
+
"learning_rate": 8.801073692070337e-05,
|
5708 |
+
"loss": 3.1289,
|
5709 |
+
"step": 812
|
5710 |
+
},
|
5711 |
+
{
|
5712 |
+
"epoch": 0.05313009275710382,
|
5713 |
+
"grad_norm": 2.382770299911499,
|
5714 |
+
"learning_rate": 8.780143943551954e-05,
|
5715 |
+
"loss": 2.8639,
|
5716 |
+
"step": 813
|
5717 |
+
},
|
5718 |
+
{
|
5719 |
+
"epoch": 0.053195443424701734,
|
5720 |
+
"grad_norm": 2.337371587753296,
|
5721 |
+
"learning_rate": 8.759219617983999e-05,
|
5722 |
+
"loss": 3.2999,
|
5723 |
+
"step": 814
|
5724 |
+
},
|
5725 |
+
{
|
5726 |
+
"epoch": 0.05326079409229965,
|
5727 |
+
"grad_norm": 2.0692298412323,
|
5728 |
+
"learning_rate": 8.738300808386935e-05,
|
5729 |
+
"loss": 2.8473,
|
5730 |
+
"step": 815
|
5731 |
+
},
|
5732 |
+
{
|
5733 |
+
"epoch": 0.05332614475989756,
|
5734 |
+
"grad_norm": 2.023470401763916,
|
5735 |
+
"learning_rate": 8.717387607756713e-05,
|
5736 |
+
"loss": 2.9644,
|
5737 |
+
"step": 816
|
5738 |
+
},
|
5739 |
+
{
|
5740 |
+
"epoch": 0.053391495427495474,
|
5741 |
+
"grad_norm": 1.8955703973770142,
|
5742 |
+
"learning_rate": 8.696480109064342e-05,
|
5743 |
+
"loss": 3.0004,
|
5744 |
+
"step": 817
|
5745 |
+
},
|
5746 |
+
{
|
5747 |
+
"epoch": 0.05345684609509339,
|
5748 |
+
"grad_norm": 2.153301477432251,
|
5749 |
+
"learning_rate": 8.675578405255485e-05,
|
5750 |
+
"loss": 3.1271,
|
5751 |
+
"step": 818
|
5752 |
+
},
|
5753 |
+
{
|
5754 |
+
"epoch": 0.05352219676269131,
|
5755 |
+
"grad_norm": 1.925044059753418,
|
5756 |
+
"learning_rate": 8.654682589250038e-05,
|
5757 |
+
"loss": 3.1745,
|
5758 |
+
"step": 819
|
5759 |
+
},
|
5760 |
+
{
|
5761 |
+
"epoch": 0.05358754743028922,
|
5762 |
+
"grad_norm": 2.142491102218628,
|
5763 |
+
"learning_rate": 8.633792753941733e-05,
|
5764 |
+
"loss": 3.0147,
|
5765 |
+
"step": 820
|
5766 |
+
},
|
5767 |
+
{
|
5768 |
+
"epoch": 0.05365289809788713,
|
5769 |
+
"grad_norm": 2.050201177597046,
|
5770 |
+
"learning_rate": 8.612908992197705e-05,
|
5771 |
+
"loss": 3.0532,
|
5772 |
+
"step": 821
|
5773 |
+
},
|
5774 |
+
{
|
5775 |
+
"epoch": 0.05371824876548505,
|
5776 |
+
"grad_norm": 2.1780121326446533,
|
5777 |
+
"learning_rate": 8.592031396858093e-05,
|
5778 |
+
"loss": 3.1273,
|
5779 |
+
"step": 822
|
5780 |
+
},
|
5781 |
+
{
|
5782 |
+
"epoch": 0.05378359943308296,
|
5783 |
+
"grad_norm": 2.17741060256958,
|
5784 |
+
"learning_rate": 8.571160060735624e-05,
|
5785 |
+
"loss": 2.9135,
|
5786 |
+
"step": 823
|
5787 |
+
},
|
5788 |
+
{
|
5789 |
+
"epoch": 0.05384895010068087,
|
5790 |
+
"grad_norm": 2.2485671043395996,
|
5791 |
+
"learning_rate": 8.550295076615188e-05,
|
5792 |
+
"loss": 3.0523,
|
5793 |
+
"step": 824
|
5794 |
+
},
|
5795 |
+
{
|
5796 |
+
"epoch": 0.053914300768278786,
|
5797 |
+
"grad_norm": 2.202855348587036,
|
5798 |
+
"learning_rate": 8.529436537253458e-05,
|
5799 |
+
"loss": 2.9131,
|
5800 |
+
"step": 825
|
5801 |
+
},
|
5802 |
+
{
|
5803 |
+
"epoch": 0.0539796514358767,
|
5804 |
+
"grad_norm": 2.4865453243255615,
|
5805 |
+
"learning_rate": 8.508584535378439e-05,
|
5806 |
+
"loss": 3.2252,
|
5807 |
+
"step": 826
|
5808 |
+
},
|
5809 |
+
{
|
5810 |
+
"epoch": 0.05404500210347461,
|
5811 |
+
"grad_norm": 2.1477367877960205,
|
5812 |
+
"learning_rate": 8.487739163689079e-05,
|
5813 |
+
"loss": 3.004,
|
5814 |
+
"step": 827
|
5815 |
+
},
|
5816 |
+
{
|
5817 |
+
"epoch": 0.054110352771072526,
|
5818 |
+
"grad_norm": 2.446580648422241,
|
5819 |
+
"learning_rate": 8.466900514854847e-05,
|
5820 |
+
"loss": 2.9681,
|
5821 |
+
"step": 828
|
5822 |
+
},
|
5823 |
+
{
|
5824 |
+
"epoch": 0.05417570343867044,
|
5825 |
+
"grad_norm": 2.8118631839752197,
|
5826 |
+
"learning_rate": 8.446068681515334e-05,
|
5827 |
+
"loss": 3.3683,
|
5828 |
+
"step": 829
|
5829 |
+
},
|
5830 |
+
{
|
5831 |
+
"epoch": 0.05424105410626835,
|
5832 |
+
"grad_norm": 2.5993924140930176,
|
5833 |
+
"learning_rate": 8.425243756279824e-05,
|
5834 |
+
"loss": 3.0367,
|
5835 |
+
"step": 830
|
5836 |
+
},
|
5837 |
+
{
|
5838 |
+
"epoch": 0.054306404773866265,
|
5839 |
+
"grad_norm": 2.7038724422454834,
|
5840 |
+
"learning_rate": 8.404425831726894e-05,
|
5841 |
+
"loss": 3.1722,
|
5842 |
+
"step": 831
|
5843 |
+
},
|
5844 |
+
{
|
5845 |
+
"epoch": 0.05437175544146418,
|
5846 |
+
"grad_norm": 2.772646427154541,
|
5847 |
+
"learning_rate": 8.383615000404e-05,
|
5848 |
+
"loss": 3.2879,
|
5849 |
+
"step": 832
|
5850 |
+
},
|
5851 |
+
{
|
5852 |
+
"epoch": 0.0544371061090621,
|
5853 |
+
"grad_norm": 2.631357431411743,
|
5854 |
+
"learning_rate": 8.362811354827059e-05,
|
5855 |
+
"loss": 3.0818,
|
5856 |
+
"step": 833
|
5857 |
+
},
|
5858 |
+
{
|
5859 |
+
"epoch": 0.05450245677666001,
|
5860 |
+
"grad_norm": 2.8192901611328125,
|
5861 |
+
"learning_rate": 8.342014987480047e-05,
|
5862 |
+
"loss": 3.0754,
|
5863 |
+
"step": 834
|
5864 |
+
},
|
5865 |
+
{
|
5866 |
+
"epoch": 0.054567807444257925,
|
5867 |
+
"grad_norm": 2.722759485244751,
|
5868 |
+
"learning_rate": 8.321225990814591e-05,
|
5869 |
+
"loss": 3.1893,
|
5870 |
+
"step": 835
|
5871 |
+
},
|
5872 |
+
{
|
5873 |
+
"epoch": 0.05463315811185584,
|
5874 |
+
"grad_norm": 2.6891770362854004,
|
5875 |
+
"learning_rate": 8.300444457249543e-05,
|
5876 |
+
"loss": 2.9677,
|
5877 |
+
"step": 836
|
5878 |
+
},
|
5879 |
+
{
|
5880 |
+
"epoch": 0.05469850877945375,
|
5881 |
+
"grad_norm": 2.8710973262786865,
|
5882 |
+
"learning_rate": 8.279670479170573e-05,
|
5883 |
+
"loss": 2.8906,
|
5884 |
+
"step": 837
|
5885 |
+
},
|
5886 |
+
{
|
5887 |
+
"epoch": 0.054763859447051665,
|
5888 |
+
"grad_norm": 3.082907199859619,
|
5889 |
+
"learning_rate": 8.258904148929775e-05,
|
5890 |
+
"loss": 3.3035,
|
5891 |
+
"step": 838
|
5892 |
+
},
|
5893 |
+
{
|
5894 |
+
"epoch": 0.05482921011464958,
|
5895 |
+
"grad_norm": 3.745654344558716,
|
5896 |
+
"learning_rate": 8.238145558845235e-05,
|
5897 |
+
"loss": 3.0103,
|
5898 |
+
"step": 839
|
5899 |
+
},
|
5900 |
+
{
|
5901 |
+
"epoch": 0.05489456078224749,
|
5902 |
+
"grad_norm": 3.0388245582580566,
|
5903 |
+
"learning_rate": 8.217394801200631e-05,
|
5904 |
+
"loss": 3.0215,
|
5905 |
+
"step": 840
|
5906 |
+
},
|
5907 |
+
{
|
5908 |
+
"epoch": 0.054959911449845404,
|
5909 |
+
"grad_norm": 2.8207526206970215,
|
5910 |
+
"learning_rate": 8.196651968244826e-05,
|
5911 |
+
"loss": 2.8506,
|
5912 |
+
"step": 841
|
5913 |
+
},
|
5914 |
+
{
|
5915 |
+
"epoch": 0.05502526211744332,
|
5916 |
+
"grad_norm": 3.4287781715393066,
|
5917 |
+
"learning_rate": 8.175917152191447e-05,
|
5918 |
+
"loss": 2.5894,
|
5919 |
+
"step": 842
|
5920 |
+
},
|
5921 |
+
{
|
5922 |
+
"epoch": 0.05509061278504123,
|
5923 |
+
"grad_norm": 3.6713480949401855,
|
5924 |
+
"learning_rate": 8.15519044521848e-05,
|
5925 |
+
"loss": 2.9109,
|
5926 |
+
"step": 843
|
5927 |
+
},
|
5928 |
+
{
|
5929 |
+
"epoch": 0.055155963452639144,
|
5930 |
+
"grad_norm": 3.62062668800354,
|
5931 |
+
"learning_rate": 8.134471939467874e-05,
|
5932 |
+
"loss": 3.3623,
|
5933 |
+
"step": 844
|
5934 |
+
},
|
5935 |
+
{
|
5936 |
+
"epoch": 0.05522131412023706,
|
5937 |
+
"grad_norm": 3.6668214797973633,
|
5938 |
+
"learning_rate": 8.113761727045105e-05,
|
5939 |
+
"loss": 2.6816,
|
5940 |
+
"step": 845
|
5941 |
+
},
|
5942 |
+
{
|
5943 |
+
"epoch": 0.05528666478783497,
|
5944 |
+
"grad_norm": 3.4973437786102295,
|
5945 |
+
"learning_rate": 8.093059900018792e-05,
|
5946 |
+
"loss": 2.7066,
|
5947 |
+
"step": 846
|
5948 |
+
},
|
5949 |
+
{
|
5950 |
+
"epoch": 0.05535201545543289,
|
5951 |
+
"grad_norm": 4.382777690887451,
|
5952 |
+
"learning_rate": 8.072366550420266e-05,
|
5953 |
+
"loss": 2.95,
|
5954 |
+
"step": 847
|
5955 |
+
},
|
5956 |
+
{
|
5957 |
+
"epoch": 0.0554173661230308,
|
5958 |
+
"grad_norm": 3.621400833129883,
|
5959 |
+
"learning_rate": 8.051681770243175e-05,
|
5960 |
+
"loss": 2.9411,
|
5961 |
+
"step": 848
|
5962 |
+
},
|
5963 |
+
{
|
5964 |
+
"epoch": 0.05548271679062872,
|
5965 |
+
"grad_norm": 4.142566680908203,
|
5966 |
+
"learning_rate": 8.031005651443073e-05,
|
5967 |
+
"loss": 2.5976,
|
5968 |
+
"step": 849
|
5969 |
+
},
|
5970 |
+
{
|
5971 |
+
"epoch": 0.05554806745822663,
|
5972 |
+
"grad_norm": 6.887070178985596,
|
5973 |
+
"learning_rate": 8.010338285937006e-05,
|
5974 |
+
"loss": 2.0695,
|
5975 |
+
"step": 850
|
5976 |
+
},
|
5977 |
+
{
|
5978 |
+
"epoch": 0.05561341812582454,
|
5979 |
+
"grad_norm": 8.775994300842285,
|
5980 |
+
"learning_rate": 7.989679765603108e-05,
|
5981 |
+
"loss": 3.5485,
|
5982 |
+
"step": 851
|
5983 |
+
},
|
5984 |
+
{
|
5985 |
+
"epoch": 0.055678768793422456,
|
5986 |
+
"grad_norm": 11.141053199768066,
|
5987 |
+
"learning_rate": 7.969030182280192e-05,
|
5988 |
+
"loss": 3.7084,
|
5989 |
+
"step": 852
|
5990 |
+
},
|
5991 |
+
{
|
5992 |
+
"epoch": 0.05574411946102037,
|
5993 |
+
"grad_norm": 10.11203670501709,
|
5994 |
+
"learning_rate": 7.948389627767343e-05,
|
5995 |
+
"loss": 3.5734,
|
5996 |
+
"step": 853
|
5997 |
+
},
|
5998 |
+
{
|
5999 |
+
"epoch": 0.05580947012861828,
|
6000 |
+
"grad_norm": 8.175284385681152,
|
6001 |
+
"learning_rate": 7.927758193823501e-05,
|
6002 |
+
"loss": 3.6809,
|
6003 |
+
"step": 854
|
6004 |
+
},
|
6005 |
+
{
|
6006 |
+
"epoch": 0.055874820796216196,
|
6007 |
+
"grad_norm": 6.074551582336426,
|
6008 |
+
"learning_rate": 7.907135972167069e-05,
|
6009 |
+
"loss": 3.4694,
|
6010 |
+
"step": 855
|
6011 |
+
},
|
6012 |
+
{
|
6013 |
+
"epoch": 0.05594017146381411,
|
6014 |
+
"grad_norm": 4.839896202087402,
|
6015 |
+
"learning_rate": 7.88652305447549e-05,
|
6016 |
+
"loss": 3.279,
|
6017 |
+
"step": 856
|
6018 |
+
},
|
6019 |
+
{
|
6020 |
+
"epoch": 0.05600552213141202,
|
6021 |
+
"grad_norm": 3.337378978729248,
|
6022 |
+
"learning_rate": 7.865919532384844e-05,
|
6023 |
+
"loss": 3.2873,
|
6024 |
+
"step": 857
|
6025 |
+
},
|
6026 |
+
{
|
6027 |
+
"epoch": 0.056070872799009935,
|
6028 |
+
"grad_norm": 2.713566780090332,
|
6029 |
+
"learning_rate": 7.845325497489449e-05,
|
6030 |
+
"loss": 3.2591,
|
6031 |
+
"step": 858
|
6032 |
+
},
|
6033 |
+
{
|
6034 |
+
"epoch": 0.05613622346660785,
|
6035 |
+
"grad_norm": 2.0323264598846436,
|
6036 |
+
"learning_rate": 7.82474104134144e-05,
|
6037 |
+
"loss": 2.8985,
|
6038 |
+
"step": 859
|
6039 |
+
},
|
6040 |
+
{
|
6041 |
+
"epoch": 0.05620157413420576,
|
6042 |
+
"grad_norm": 2.278287410736084,
|
6043 |
+
"learning_rate": 7.804166255450373e-05,
|
6044 |
+
"loss": 3.153,
|
6045 |
+
"step": 860
|
6046 |
+
},
|
6047 |
+
{
|
6048 |
+
"epoch": 0.05626692480180368,
|
6049 |
+
"grad_norm": 2.0815248489379883,
|
6050 |
+
"learning_rate": 7.783601231282812e-05,
|
6051 |
+
"loss": 3.1981,
|
6052 |
+
"step": 861
|
6053 |
+
},
|
6054 |
+
{
|
6055 |
+
"epoch": 0.056332275469401595,
|
6056 |
+
"grad_norm": 1.8382048606872559,
|
6057 |
+
"learning_rate": 7.763046060261932e-05,
|
6058 |
+
"loss": 3.0894,
|
6059 |
+
"step": 862
|
6060 |
+
},
|
6061 |
+
{
|
6062 |
+
"epoch": 0.05639762613699951,
|
6063 |
+
"grad_norm": 1.705528736114502,
|
6064 |
+
"learning_rate": 7.742500833767094e-05,
|
6065 |
+
"loss": 3.1151,
|
6066 |
+
"step": 863
|
6067 |
+
},
|
6068 |
+
{
|
6069 |
+
"epoch": 0.05646297680459742,
|
6070 |
+
"grad_norm": 1.7962243556976318,
|
6071 |
+
"learning_rate": 7.721965643133458e-05,
|
6072 |
+
"loss": 3.0139,
|
6073 |
+
"step": 864
|
6074 |
+
},
|
6075 |
+
{
|
6076 |
+
"epoch": 0.056528327472195335,
|
6077 |
+
"grad_norm": 1.7769712209701538,
|
6078 |
+
"learning_rate": 7.701440579651566e-05,
|
6079 |
+
"loss": 3.1339,
|
6080 |
+
"step": 865
|
6081 |
+
},
|
6082 |
+
{
|
6083 |
+
"epoch": 0.05659367813979325,
|
6084 |
+
"grad_norm": 1.7150388956069946,
|
6085 |
+
"learning_rate": 7.680925734566937e-05,
|
6086 |
+
"loss": 3.0084,
|
6087 |
+
"step": 866
|
6088 |
+
},
|
6089 |
+
{
|
6090 |
+
"epoch": 0.05665902880739116,
|
6091 |
+
"grad_norm": 2.3266444206237793,
|
6092 |
+
"learning_rate": 7.660421199079669e-05,
|
6093 |
+
"loss": 3.1235,
|
6094 |
+
"step": 867
|
6095 |
+
},
|
6096 |
+
{
|
6097 |
+
"epoch": 0.056724379474989074,
|
6098 |
+
"grad_norm": 1.864682912826538,
|
6099 |
+
"learning_rate": 7.639927064344022e-05,
|
6100 |
+
"loss": 3.1664,
|
6101 |
+
"step": 868
|
6102 |
+
},
|
6103 |
+
{
|
6104 |
+
"epoch": 0.05678973014258699,
|
6105 |
+
"grad_norm": 2.070551872253418,
|
6106 |
+
"learning_rate": 7.619443421468021e-05,
|
6107 |
+
"loss": 3.1723,
|
6108 |
+
"step": 869
|
6109 |
+
},
|
6110 |
+
{
|
6111 |
+
"epoch": 0.0568550808101849,
|
6112 |
+
"grad_norm": 1.8200955390930176,
|
6113 |
+
"learning_rate": 7.598970361513051e-05,
|
6114 |
+
"loss": 3.0291,
|
6115 |
+
"step": 870
|
6116 |
+
},
|
6117 |
+
{
|
6118 |
+
"epoch": 0.056920431477782814,
|
6119 |
+
"grad_norm": 1.9422317743301392,
|
6120 |
+
"learning_rate": 7.578507975493448e-05,
|
6121 |
+
"loss": 3.0344,
|
6122 |
+
"step": 871
|
6123 |
+
},
|
6124 |
+
{
|
6125 |
+
"epoch": 0.05698578214538073,
|
6126 |
+
"grad_norm": 1.9396344423294067,
|
6127 |
+
"learning_rate": 7.558056354376098e-05,
|
6128 |
+
"loss": 3.0946,
|
6129 |
+
"step": 872
|
6130 |
+
},
|
6131 |
+
{
|
6132 |
+
"epoch": 0.05705113281297864,
|
6133 |
+
"grad_norm": 2.011350393295288,
|
6134 |
+
"learning_rate": 7.537615589080027e-05,
|
6135 |
+
"loss": 3.1321,
|
6136 |
+
"step": 873
|
6137 |
+
},
|
6138 |
+
{
|
6139 |
+
"epoch": 0.05711648348057655,
|
6140 |
+
"grad_norm": 2.0479001998901367,
|
6141 |
+
"learning_rate": 7.517185770476006e-05,
|
6142 |
+
"loss": 3.0291,
|
6143 |
+
"step": 874
|
6144 |
+
},
|
6145 |
+
{
|
6146 |
+
"epoch": 0.05718183414817447,
|
6147 |
+
"grad_norm": 2.116241931915283,
|
6148 |
+
"learning_rate": 7.496766989386136e-05,
|
6149 |
+
"loss": 3.0737,
|
6150 |
+
"step": 875
|
6151 |
+
},
|
6152 |
+
{
|
6153 |
+
"epoch": 0.05724718481577239,
|
6154 |
+
"grad_norm": 1.9802380800247192,
|
6155 |
+
"learning_rate": 7.476359336583454e-05,
|
6156 |
+
"loss": 2.9693,
|
6157 |
+
"step": 876
|
6158 |
+
},
|
6159 |
+
{
|
6160 |
+
"epoch": 0.0573125354833703,
|
6161 |
+
"grad_norm": 2.047084093093872,
|
6162 |
+
"learning_rate": 7.455962902791522e-05,
|
6163 |
+
"loss": 3.0396,
|
6164 |
+
"step": 877
|
6165 |
+
},
|
6166 |
+
{
|
6167 |
+
"epoch": 0.05737788615096821,
|
6168 |
+
"grad_norm": 2.106415033340454,
|
6169 |
+
"learning_rate": 7.435577778684033e-05,
|
6170 |
+
"loss": 2.9766,
|
6171 |
+
"step": 878
|
6172 |
+
},
|
6173 |
+
{
|
6174 |
+
"epoch": 0.057443236818566126,
|
6175 |
+
"grad_norm": 2.0279815196990967,
|
6176 |
+
"learning_rate": 7.415204054884399e-05,
|
6177 |
+
"loss": 2.8485,
|
6178 |
+
"step": 879
|
6179 |
+
},
|
6180 |
+
{
|
6181 |
+
"epoch": 0.05750858748616404,
|
6182 |
+
"grad_norm": 2.28250789642334,
|
6183 |
+
"learning_rate": 7.394841821965345e-05,
|
6184 |
+
"loss": 2.9833,
|
6185 |
+
"step": 880
|
6186 |
+
},
|
6187 |
+
{
|
6188 |
+
"epoch": 0.05757393815376195,
|
6189 |
+
"grad_norm": 2.818068742752075,
|
6190 |
+
"learning_rate": 7.374491170448525e-05,
|
6191 |
+
"loss": 3.0896,
|
6192 |
+
"step": 881
|
6193 |
+
},
|
6194 |
+
{
|
6195 |
+
"epoch": 0.057639288821359866,
|
6196 |
+
"grad_norm": 3.2391390800476074,
|
6197 |
+
"learning_rate": 7.3541521908041e-05,
|
6198 |
+
"loss": 3.2677,
|
6199 |
+
"step": 882
|
6200 |
+
},
|
6201 |
+
{
|
6202 |
+
"epoch": 0.05770463948895778,
|
6203 |
+
"grad_norm": 2.3144891262054443,
|
6204 |
+
"learning_rate": 7.33382497345034e-05,
|
6205 |
+
"loss": 2.9707,
|
6206 |
+
"step": 883
|
6207 |
+
},
|
6208 |
+
{
|
6209 |
+
"epoch": 0.05776999015655569,
|
6210 |
+
"grad_norm": 2.2125191688537598,
|
6211 |
+
"learning_rate": 7.313509608753231e-05,
|
6212 |
+
"loss": 2.7935,
|
6213 |
+
"step": 884
|
6214 |
+
},
|
6215 |
+
{
|
6216 |
+
"epoch": 0.057835340824153605,
|
6217 |
+
"grad_norm": 2.3742804527282715,
|
6218 |
+
"learning_rate": 7.293206187026061e-05,
|
6219 |
+
"loss": 2.8455,
|
6220 |
+
"step": 885
|
6221 |
+
},
|
6222 |
+
{
|
6223 |
+
"epoch": 0.05790069149175152,
|
6224 |
+
"grad_norm": 2.8229916095733643,
|
6225 |
+
"learning_rate": 7.27291479852903e-05,
|
6226 |
+
"loss": 2.8573,
|
6227 |
+
"step": 886
|
6228 |
+
},
|
6229 |
+
{
|
6230 |
+
"epoch": 0.05796604215934943,
|
6231 |
+
"grad_norm": 3.4678986072540283,
|
6232 |
+
"learning_rate": 7.252635533468843e-05,
|
6233 |
+
"loss": 3.3535,
|
6234 |
+
"step": 887
|
6235 |
+
},
|
6236 |
+
{
|
6237 |
+
"epoch": 0.058031392826947345,
|
6238 |
+
"grad_norm": 2.905946731567383,
|
6239 |
+
"learning_rate": 7.232368481998309e-05,
|
6240 |
+
"loss": 2.8013,
|
6241 |
+
"step": 888
|
6242 |
+
},
|
6243 |
+
{
|
6244 |
+
"epoch": 0.05809674349454526,
|
6245 |
+
"grad_norm": 2.5806751251220703,
|
6246 |
+
"learning_rate": 7.212113734215932e-05,
|
6247 |
+
"loss": 3.0396,
|
6248 |
+
"step": 889
|
6249 |
+
},
|
6250 |
+
{
|
6251 |
+
"epoch": 0.05816209416214318,
|
6252 |
+
"grad_norm": 2.6748881340026855,
|
6253 |
+
"learning_rate": 7.191871380165538e-05,
|
6254 |
+
"loss": 3.1046,
|
6255 |
+
"step": 890
|
6256 |
+
},
|
6257 |
+
{
|
6258 |
+
"epoch": 0.05822744482974109,
|
6259 |
+
"grad_norm": 2.549471139907837,
|
6260 |
+
"learning_rate": 7.17164150983584e-05,
|
6261 |
+
"loss": 2.8889,
|
6262 |
+
"step": 891
|
6263 |
+
},
|
6264 |
+
{
|
6265 |
+
"epoch": 0.058292795497339005,
|
6266 |
+
"grad_norm": 2.936119556427002,
|
6267 |
+
"learning_rate": 7.151424213160061e-05,
|
6268 |
+
"loss": 2.914,
|
6269 |
+
"step": 892
|
6270 |
+
},
|
6271 |
+
{
|
6272 |
+
"epoch": 0.05835814616493692,
|
6273 |
+
"grad_norm": 2.6108150482177734,
|
6274 |
+
"learning_rate": 7.131219580015521e-05,
|
6275 |
+
"loss": 2.8157,
|
6276 |
+
"step": 893
|
6277 |
+
},
|
6278 |
+
{
|
6279 |
+
"epoch": 0.05842349683253483,
|
6280 |
+
"grad_norm": 3.079392194747925,
|
6281 |
+
"learning_rate": 7.11102770022325e-05,
|
6282 |
+
"loss": 2.8488,
|
6283 |
+
"step": 894
|
6284 |
+
},
|
6285 |
+
{
|
6286 |
+
"epoch": 0.058488847500132744,
|
6287 |
+
"grad_norm": 3.973539352416992,
|
6288 |
+
"learning_rate": 7.090848663547574e-05,
|
6289 |
+
"loss": 2.7098,
|
6290 |
+
"step": 895
|
6291 |
+
},
|
6292 |
+
{
|
6293 |
+
"epoch": 0.05855419816773066,
|
6294 |
+
"grad_norm": 3.3315682411193848,
|
6295 |
+
"learning_rate": 7.070682559695736e-05,
|
6296 |
+
"loss": 2.9506,
|
6297 |
+
"step": 896
|
6298 |
+
},
|
6299 |
+
{
|
6300 |
+
"epoch": 0.05861954883532857,
|
6301 |
+
"grad_norm": 4.045418739318848,
|
6302 |
+
"learning_rate": 7.050529478317476e-05,
|
6303 |
+
"loss": 2.9065,
|
6304 |
+
"step": 897
|
6305 |
+
},
|
6306 |
+
{
|
6307 |
+
"epoch": 0.058684899502926484,
|
6308 |
+
"grad_norm": 4.5215840339660645,
|
6309 |
+
"learning_rate": 7.03038950900464e-05,
|
6310 |
+
"loss": 2.7954,
|
6311 |
+
"step": 898
|
6312 |
+
},
|
6313 |
+
{
|
6314 |
+
"epoch": 0.0587502501705244,
|
6315 |
+
"grad_norm": 4.555988788604736,
|
6316 |
+
"learning_rate": 7.010262741290798e-05,
|
6317 |
+
"loss": 2.472,
|
6318 |
+
"step": 899
|
6319 |
+
},
|
6320 |
+
{
|
6321 |
+
"epoch": 0.05881560083812231,
|
6322 |
+
"grad_norm": 9.155795097351074,
|
6323 |
+
"learning_rate": 6.990149264650814e-05,
|
6324 |
+
"loss": 3.0061,
|
6325 |
+
"step": 900
|
6326 |
+
},
|
6327 |
+
{
|
6328 |
+
"epoch": 0.05888095150572022,
|
6329 |
+
"grad_norm": 11.21951675415039,
|
6330 |
+
"learning_rate": 6.970049168500474e-05,
|
6331 |
+
"loss": 3.5856,
|
6332 |
+
"step": 901
|
6333 |
+
},
|
6334 |
+
{
|
6335 |
+
"epoch": 0.05894630217331814,
|
6336 |
+
"grad_norm": 12.78548526763916,
|
6337 |
+
"learning_rate": 6.94996254219608e-05,
|
6338 |
+
"loss": 3.7137,
|
6339 |
+
"step": 902
|
6340 |
+
},
|
6341 |
+
{
|
6342 |
+
"epoch": 0.05901165284091605,
|
6343 |
+
"grad_norm": 11.96114444732666,
|
6344 |
+
"learning_rate": 6.929889475034048e-05,
|
6345 |
+
"loss": 3.5168,
|
6346 |
+
"step": 903
|
6347 |
+
},
|
6348 |
+
{
|
6349 |
+
"epoch": 0.05907700350851397,
|
6350 |
+
"grad_norm": 11.37401294708252,
|
6351 |
+
"learning_rate": 6.909830056250527e-05,
|
6352 |
+
"loss": 3.6274,
|
6353 |
+
"step": 904
|
6354 |
+
},
|
6355 |
+
{
|
6356 |
+
"epoch": 0.05914235417611188,
|
6357 |
+
"grad_norm": 9.569934844970703,
|
6358 |
+
"learning_rate": 6.889784375020978e-05,
|
6359 |
+
"loss": 3.5746,
|
6360 |
+
"step": 905
|
6361 |
+
},
|
6362 |
+
{
|
6363 |
+
"epoch": 0.059207704843709796,
|
6364 |
+
"grad_norm": 7.825098514556885,
|
6365 |
+
"learning_rate": 6.869752520459803e-05,
|
6366 |
+
"loss": 3.4076,
|
6367 |
+
"step": 906
|
6368 |
+
},
|
6369 |
+
{
|
6370 |
+
"epoch": 0.05927305551130771,
|
6371 |
+
"grad_norm": 5.357243537902832,
|
6372 |
+
"learning_rate": 6.849734581619918e-05,
|
6373 |
+
"loss": 3.275,
|
6374 |
+
"step": 907
|
6375 |
+
},
|
6376 |
+
{
|
6377 |
+
"epoch": 0.05933840617890562,
|
6378 |
+
"grad_norm": 4.018442630767822,
|
6379 |
+
"learning_rate": 6.829730647492404e-05,
|
6380 |
+
"loss": 3.1789,
|
6381 |
+
"step": 908
|
6382 |
+
},
|
6383 |
+
{
|
6384 |
+
"epoch": 0.059403756846503536,
|
6385 |
+
"grad_norm": 2.417752981185913,
|
6386 |
+
"learning_rate": 6.80974080700606e-05,
|
6387 |
+
"loss": 3.0707,
|
6388 |
+
"step": 909
|
6389 |
+
},
|
6390 |
+
{
|
6391 |
+
"epoch": 0.05946910751410145,
|
6392 |
+
"grad_norm": 1.9882200956344604,
|
6393 |
+
"learning_rate": 6.789765149027039e-05,
|
6394 |
+
"loss": 3.1577,
|
6395 |
+
"step": 910
|
6396 |
+
},
|
6397 |
+
{
|
6398 |
+
"epoch": 0.05953445818169936,
|
6399 |
+
"grad_norm": 1.722104549407959,
|
6400 |
+
"learning_rate": 6.769803762358443e-05,
|
6401 |
+
"loss": 3.1941,
|
6402 |
+
"step": 911
|
6403 |
+
},
|
6404 |
+
{
|
6405 |
+
"epoch": 0.059599808849297276,
|
6406 |
+
"grad_norm": 1.8540571928024292,
|
6407 |
+
"learning_rate": 6.749856735739928e-05,
|
6408 |
+
"loss": 3.2046,
|
6409 |
+
"step": 912
|
6410 |
+
},
|
6411 |
+
{
|
6412 |
+
"epoch": 0.05966515951689519,
|
6413 |
+
"grad_norm": 1.6410471200942993,
|
6414 |
+
"learning_rate": 6.729924157847323e-05,
|
6415 |
+
"loss": 3.1146,
|
6416 |
+
"step": 913
|
6417 |
+
},
|
6418 |
+
{
|
6419 |
+
"epoch": 0.0597305101844931,
|
6420 |
+
"grad_norm": 1.9086085557937622,
|
6421 |
+
"learning_rate": 6.710006117292209e-05,
|
6422 |
+
"loss": 3.0955,
|
6423 |
+
"step": 914
|
6424 |
+
},
|
6425 |
+
{
|
6426 |
+
"epoch": 0.059795860852091015,
|
6427 |
+
"grad_norm": 1.9486234188079834,
|
6428 |
+
"learning_rate": 6.690102702621548e-05,
|
6429 |
+
"loss": 3.1597,
|
6430 |
+
"step": 915
|
6431 |
+
},
|
6432 |
+
{
|
6433 |
+
"epoch": 0.05986121151968893,
|
6434 |
+
"grad_norm": 1.86729097366333,
|
6435 |
+
"learning_rate": 6.670214002317278e-05,
|
6436 |
+
"loss": 3.1814,
|
6437 |
+
"step": 916
|
6438 |
+
},
|
6439 |
+
{
|
6440 |
+
"epoch": 0.05992656218728684,
|
6441 |
+
"grad_norm": 1.9121336936950684,
|
6442 |
+
"learning_rate": 6.650340104795932e-05,
|
6443 |
+
"loss": 2.9355,
|
6444 |
+
"step": 917
|
6445 |
+
},
|
6446 |
+
{
|
6447 |
+
"epoch": 0.05999191285488476,
|
6448 |
+
"grad_norm": 1.899423360824585,
|
6449 |
+
"learning_rate": 6.630481098408228e-05,
|
6450 |
+
"loss": 2.8474,
|
6451 |
+
"step": 918
|
6452 |
+
},
|
6453 |
+
{
|
6454 |
+
"epoch": 0.060057263522482675,
|
6455 |
+
"grad_norm": 1.8932859897613525,
|
6456 |
+
"learning_rate": 6.610637071438686e-05,
|
6457 |
+
"loss": 2.9341,
|
6458 |
+
"step": 919
|
6459 |
+
},
|
6460 |
+
{
|
6461 |
+
"epoch": 0.06012261419008059,
|
6462 |
+
"grad_norm": 1.8576266765594482,
|
6463 |
+
"learning_rate": 6.590808112105232e-05,
|
6464 |
+
"loss": 3.0429,
|
6465 |
+
"step": 920
|
6466 |
+
},
|
6467 |
+
{
|
6468 |
+
"epoch": 0.0601879648576785,
|
6469 |
+
"grad_norm": 1.9361873865127563,
|
6470 |
+
"learning_rate": 6.570994308558812e-05,
|
6471 |
+
"loss": 3.0584,
|
6472 |
+
"step": 921
|
6473 |
+
},
|
6474 |
+
{
|
6475 |
+
"epoch": 0.060253315525276414,
|
6476 |
+
"grad_norm": 1.9576289653778076,
|
6477 |
+
"learning_rate": 6.551195748882997e-05,
|
6478 |
+
"loss": 3.0854,
|
6479 |
+
"step": 922
|
6480 |
+
},
|
6481 |
+
{
|
6482 |
+
"epoch": 0.06031866619287433,
|
6483 |
+
"grad_norm": 2.1195924282073975,
|
6484 |
+
"learning_rate": 6.531412521093586e-05,
|
6485 |
+
"loss": 3.2338,
|
6486 |
+
"step": 923
|
6487 |
+
},
|
6488 |
+
{
|
6489 |
+
"epoch": 0.06038401686047224,
|
6490 |
+
"grad_norm": 2.1367833614349365,
|
6491 |
+
"learning_rate": 6.51164471313822e-05,
|
6492 |
+
"loss": 2.7995,
|
6493 |
+
"step": 924
|
6494 |
+
},
|
6495 |
+
{
|
6496 |
+
"epoch": 0.060449367528070154,
|
6497 |
+
"grad_norm": 2.1398251056671143,
|
6498 |
+
"learning_rate": 6.491892412895995e-05,
|
6499 |
+
"loss": 3.0869,
|
6500 |
+
"step": 925
|
6501 |
+
},
|
6502 |
+
{
|
6503 |
+
"epoch": 0.06051471819566807,
|
6504 |
+
"grad_norm": 2.351515531539917,
|
6505 |
+
"learning_rate": 6.472155708177052e-05,
|
6506 |
+
"loss": 2.9483,
|
6507 |
+
"step": 926
|
6508 |
+
},
|
6509 |
+
{
|
6510 |
+
"epoch": 0.06058006886326598,
|
6511 |
+
"grad_norm": 2.108220100402832,
|
6512 |
+
"learning_rate": 6.452434686722224e-05,
|
6513 |
+
"loss": 2.9758,
|
6514 |
+
"step": 927
|
6515 |
+
},
|
6516 |
+
{
|
6517 |
+
"epoch": 0.060645419530863894,
|
6518 |
+
"grad_norm": 2.402775287628174,
|
6519 |
+
"learning_rate": 6.432729436202604e-05,
|
6520 |
+
"loss": 2.8292,
|
6521 |
+
"step": 928
|
6522 |
+
},
|
6523 |
+
{
|
6524 |
+
"epoch": 0.06071077019846181,
|
6525 |
+
"grad_norm": 2.2293272018432617,
|
6526 |
+
"learning_rate": 6.41304004421918e-05,
|
6527 |
+
"loss": 3.0568,
|
6528 |
+
"step": 929
|
6529 |
+
},
|
6530 |
+
{
|
6531 |
+
"epoch": 0.06077612086605972,
|
6532 |
+
"grad_norm": 2.485183000564575,
|
6533 |
+
"learning_rate": 6.393366598302446e-05,
|
6534 |
+
"loss": 3.2962,
|
6535 |
+
"step": 930
|
6536 |
+
},
|
6537 |
+
{
|
6538 |
+
"epoch": 0.06084147153365763,
|
6539 |
+
"grad_norm": 2.270014762878418,
|
6540 |
+
"learning_rate": 6.373709185911998e-05,
|
6541 |
+
"loss": 3.0106,
|
6542 |
+
"step": 931
|
6543 |
+
},
|
6544 |
+
{
|
6545 |
+
"epoch": 0.060906822201255546,
|
6546 |
+
"grad_norm": 2.5223422050476074,
|
6547 |
+
"learning_rate": 6.354067894436155e-05,
|
6548 |
+
"loss": 3.0954,
|
6549 |
+
"step": 932
|
6550 |
+
},
|
6551 |
+
{
|
6552 |
+
"epoch": 0.060972172868853466,
|
6553 |
+
"grad_norm": 3.3443124294281006,
|
6554 |
+
"learning_rate": 6.334442811191576e-05,
|
6555 |
+
"loss": 3.0791,
|
6556 |
+
"step": 933
|
6557 |
+
},
|
6558 |
+
{
|
6559 |
+
"epoch": 0.06103752353645138,
|
6560 |
+
"grad_norm": 2.9150595664978027,
|
6561 |
+
"learning_rate": 6.314834023422858e-05,
|
6562 |
+
"loss": 3.1297,
|
6563 |
+
"step": 934
|
6564 |
+
},
|
6565 |
+
{
|
6566 |
+
"epoch": 0.06110287420404929,
|
6567 |
+
"grad_norm": 2.2395308017730713,
|
6568 |
+
"learning_rate": 6.295241618302156e-05,
|
6569 |
+
"loss": 2.9412,
|
6570 |
+
"step": 935
|
6571 |
+
},
|
6572 |
+
{
|
6573 |
+
"epoch": 0.061168224871647206,
|
6574 |
+
"grad_norm": 2.848072052001953,
|
6575 |
+
"learning_rate": 6.275665682928803e-05,
|
6576 |
+
"loss": 3.2713,
|
6577 |
+
"step": 936
|
6578 |
+
},
|
6579 |
+
{
|
6580 |
+
"epoch": 0.06123357553924512,
|
6581 |
+
"grad_norm": 3.254633903503418,
|
6582 |
+
"learning_rate": 6.256106304328905e-05,
|
6583 |
+
"loss": 3.3262,
|
6584 |
+
"step": 937
|
6585 |
+
},
|
6586 |
+
{
|
6587 |
+
"epoch": 0.06129892620684303,
|
6588 |
+
"grad_norm": 3.0859248638153076,
|
6589 |
+
"learning_rate": 6.23656356945497e-05,
|
6590 |
+
"loss": 3.0816,
|
6591 |
+
"step": 938
|
6592 |
+
},
|
6593 |
+
{
|
6594 |
+
"epoch": 0.061364276874440946,
|
6595 |
+
"grad_norm": 6.151339530944824,
|
6596 |
+
"learning_rate": 6.21703756518551e-05,
|
6597 |
+
"loss": 3.4066,
|
6598 |
+
"step": 939
|
6599 |
+
},
|
6600 |
+
{
|
6601 |
+
"epoch": 0.06142962754203886,
|
6602 |
+
"grad_norm": 3.2941558361053467,
|
6603 |
+
"learning_rate": 6.197528378324665e-05,
|
6604 |
+
"loss": 2.8092,
|
6605 |
+
"step": 940
|
6606 |
+
},
|
6607 |
+
{
|
6608 |
+
"epoch": 0.06149497820963677,
|
6609 |
+
"grad_norm": 3.569110870361328,
|
6610 |
+
"learning_rate": 6.17803609560181e-05,
|
6611 |
+
"loss": 2.9905,
|
6612 |
+
"step": 941
|
6613 |
+
},
|
6614 |
+
{
|
6615 |
+
"epoch": 0.061560328877234685,
|
6616 |
+
"grad_norm": 3.397008180618286,
|
6617 |
+
"learning_rate": 6.158560803671168e-05,
|
6618 |
+
"loss": 2.7348,
|
6619 |
+
"step": 942
|
6620 |
+
},
|
6621 |
+
{
|
6622 |
+
"epoch": 0.0616256795448326,
|
6623 |
+
"grad_norm": 3.3827006816864014,
|
6624 |
+
"learning_rate": 6.139102589111435e-05,
|
6625 |
+
"loss": 3.0409,
|
6626 |
+
"step": 943
|
6627 |
+
},
|
6628 |
+
{
|
6629 |
+
"epoch": 0.06169103021243051,
|
6630 |
+
"grad_norm": 3.1896936893463135,
|
6631 |
+
"learning_rate": 6.119661538425381e-05,
|
6632 |
+
"loss": 3.2392,
|
6633 |
+
"step": 944
|
6634 |
+
},
|
6635 |
+
{
|
6636 |
+
"epoch": 0.061756380880028425,
|
6637 |
+
"grad_norm": 4.166941165924072,
|
6638 |
+
"learning_rate": 6.100237738039484e-05,
|
6639 |
+
"loss": 3.2631,
|
6640 |
+
"step": 945
|
6641 |
+
},
|
6642 |
+
{
|
6643 |
+
"epoch": 0.06182173154762634,
|
6644 |
+
"grad_norm": 4.260908126831055,
|
6645 |
+
"learning_rate": 6.0808312743035236e-05,
|
6646 |
+
"loss": 2.813,
|
6647 |
+
"step": 946
|
6648 |
+
},
|
6649 |
+
{
|
6650 |
+
"epoch": 0.06188708221522426,
|
6651 |
+
"grad_norm": 4.6422600746154785,
|
6652 |
+
"learning_rate": 6.061442233490211e-05,
|
6653 |
+
"loss": 3.279,
|
6654 |
+
"step": 947
|
6655 |
+
},
|
6656 |
+
{
|
6657 |
+
"epoch": 0.06195243288282217,
|
6658 |
+
"grad_norm": 5.281010627746582,
|
6659 |
+
"learning_rate": 6.042070701794806e-05,
|
6660 |
+
"loss": 3.049,
|
6661 |
+
"step": 948
|
6662 |
+
},
|
6663 |
+
{
|
6664 |
+
"epoch": 0.062017783550420084,
|
6665 |
+
"grad_norm": 5.087477684020996,
|
6666 |
+
"learning_rate": 6.0227167653347305e-05,
|
6667 |
+
"loss": 2.6435,
|
6668 |
+
"step": 949
|
6669 |
+
},
|
6670 |
+
{
|
6671 |
+
"epoch": 0.062083134218018,
|
6672 |
+
"grad_norm": 6.93611478805542,
|
6673 |
+
"learning_rate": 6.0033805101491794e-05,
|
6674 |
+
"loss": 3.1446,
|
6675 |
+
"step": 950
|
6676 |
+
},
|
6677 |
+
{
|
6678 |
+
"epoch": 0.06214848488561591,
|
6679 |
+
"grad_norm": 8.002715110778809,
|
6680 |
+
"learning_rate": 5.98406202219875e-05,
|
6681 |
+
"loss": 3.3335,
|
6682 |
+
"step": 951
|
6683 |
+
},
|
6684 |
+
{
|
6685 |
+
"epoch": 0.062213835553213824,
|
6686 |
+
"grad_norm": 9.21122932434082,
|
6687 |
+
"learning_rate": 5.964761387365052e-05,
|
6688 |
+
"loss": 3.3753,
|
6689 |
+
"step": 952
|
6690 |
+
},
|
6691 |
+
{
|
6692 |
+
"epoch": 0.06227918622081174,
|
6693 |
+
"grad_norm": 10.672674179077148,
|
6694 |
+
"learning_rate": 5.9454786914503255e-05,
|
6695 |
+
"loss": 3.4291,
|
6696 |
+
"step": 953
|
6697 |
+
},
|
6698 |
+
{
|
6699 |
+
"epoch": 0.06234453688840965,
|
6700 |
+
"grad_norm": 11.474349021911621,
|
6701 |
+
"learning_rate": 5.926214020177074e-05,
|
6702 |
+
"loss": 3.2926,
|
6703 |
+
"step": 954
|
6704 |
+
},
|
6705 |
+
{
|
6706 |
+
"epoch": 0.062409887556007564,
|
6707 |
+
"grad_norm": 9.654754638671875,
|
6708 |
+
"learning_rate": 5.9069674591876534e-05,
|
6709 |
+
"loss": 3.1907,
|
6710 |
+
"step": 955
|
6711 |
+
},
|
6712 |
+
{
|
6713 |
+
"epoch": 0.06247523822360548,
|
6714 |
+
"grad_norm": 8.261674880981445,
|
6715 |
+
"learning_rate": 5.887739094043923e-05,
|
6716 |
+
"loss": 3.3193,
|
6717 |
+
"step": 956
|
6718 |
+
},
|
6719 |
+
{
|
6720 |
+
"epoch": 0.0625405888912034,
|
6721 |
+
"grad_norm": 6.750918865203857,
|
6722 |
+
"learning_rate": 5.868529010226845e-05,
|
6723 |
+
"loss": 3.1375,
|
6724 |
+
"step": 957
|
6725 |
+
},
|
6726 |
+
{
|
6727 |
+
"epoch": 0.06260593955880131,
|
6728 |
+
"grad_norm": 4.5431342124938965,
|
6729 |
+
"learning_rate": 5.849337293136112e-05,
|
6730 |
+
"loss": 3.2364,
|
6731 |
+
"step": 958
|
6732 |
+
},
|
6733 |
+
{
|
6734 |
+
"epoch": 0.06267129022639922,
|
6735 |
+
"grad_norm": 3.82590389251709,
|
6736 |
+
"learning_rate": 5.830164028089766e-05,
|
6737 |
+
"loss": 3.2171,
|
6738 |
+
"step": 959
|
6739 |
+
},
|
6740 |
+
{
|
6741 |
+
"epoch": 0.06273664089399714,
|
6742 |
+
"grad_norm": 2.5566561222076416,
|
6743 |
+
"learning_rate": 5.811009300323818e-05,
|
6744 |
+
"loss": 3.0169,
|
6745 |
+
"step": 960
|
6746 |
+
},
|
6747 |
+
{
|
6748 |
+
"epoch": 0.06280199156159505,
|
6749 |
+
"grad_norm": 1.90986967086792,
|
6750 |
+
"learning_rate": 5.791873194991872e-05,
|
6751 |
+
"loss": 3.1772,
|
6752 |
+
"step": 961
|
6753 |
+
},
|
6754 |
+
{
|
6755 |
+
"epoch": 0.06286734222919296,
|
6756 |
+
"grad_norm": 1.7746076583862305,
|
6757 |
+
"learning_rate": 5.7727557971647427e-05,
|
6758 |
+
"loss": 2.9972,
|
6759 |
+
"step": 962
|
6760 |
+
},
|
6761 |
+
{
|
6762 |
+
"epoch": 0.06293269289679088,
|
6763 |
+
"grad_norm": 1.9239122867584229,
|
6764 |
+
"learning_rate": 5.7536571918300864e-05,
|
6765 |
+
"loss": 3.0812,
|
6766 |
+
"step": 963
|
6767 |
+
},
|
6768 |
+
{
|
6769 |
+
"epoch": 0.06299804356438879,
|
6770 |
+
"grad_norm": 2.131152391433716,
|
6771 |
+
"learning_rate": 5.734577463892008e-05,
|
6772 |
+
"loss": 2.9639,
|
6773 |
+
"step": 964
|
6774 |
+
},
|
6775 |
+
{
|
6776 |
+
"epoch": 0.0630633942319867,
|
6777 |
+
"grad_norm": 1.8903312683105469,
|
6778 |
+
"learning_rate": 5.7155166981706956e-05,
|
6779 |
+
"loss": 3.2021,
|
6780 |
+
"step": 965
|
6781 |
+
},
|
6782 |
+
{
|
6783 |
+
"epoch": 0.06312874489958462,
|
6784 |
+
"grad_norm": 2.311065673828125,
|
6785 |
+
"learning_rate": 5.6964749794020354e-05,
|
6786 |
+
"loss": 3.0328,
|
6787 |
+
"step": 966
|
6788 |
+
},
|
6789 |
+
{
|
6790 |
+
"epoch": 0.06319409556718253,
|
6791 |
+
"grad_norm": 1.8086532354354858,
|
6792 |
+
"learning_rate": 5.6774523922372394e-05,
|
6793 |
+
"loss": 2.9212,
|
6794 |
+
"step": 967
|
6795 |
+
},
|
6796 |
+
{
|
6797 |
+
"epoch": 0.06325944623478044,
|
6798 |
+
"grad_norm": 2.2068464756011963,
|
6799 |
+
"learning_rate": 5.6584490212424804e-05,
|
6800 |
+
"loss": 2.9809,
|
6801 |
+
"step": 968
|
6802 |
+
},
|
6803 |
+
{
|
6804 |
+
"epoch": 0.06332479690237836,
|
6805 |
+
"grad_norm": 1.7487001419067383,
|
6806 |
+
"learning_rate": 5.639464950898491e-05,
|
6807 |
+
"loss": 2.9219,
|
6808 |
+
"step": 969
|
6809 |
+
},
|
6810 |
+
{
|
6811 |
+
"epoch": 0.06339014756997627,
|
6812 |
+
"grad_norm": 2.115124225616455,
|
6813 |
+
"learning_rate": 5.620500265600206e-05,
|
6814 |
+
"loss": 2.9337,
|
6815 |
+
"step": 970
|
6816 |
+
},
|
6817 |
+
{
|
6818 |
+
"epoch": 0.06345549823757418,
|
6819 |
+
"grad_norm": 1.9892399311065674,
|
6820 |
+
"learning_rate": 5.601555049656382e-05,
|
6821 |
+
"loss": 3.1191,
|
6822 |
+
"step": 971
|
6823 |
+
},
|
6824 |
+
{
|
6825 |
+
"epoch": 0.0635208489051721,
|
6826 |
+
"grad_norm": 1.8472293615341187,
|
6827 |
+
"learning_rate": 5.58262938728922e-05,
|
6828 |
+
"loss": 3.0616,
|
6829 |
+
"step": 972
|
6830 |
+
},
|
6831 |
+
{
|
6832 |
+
"epoch": 0.06358619957277001,
|
6833 |
+
"grad_norm": 1.9462532997131348,
|
6834 |
+
"learning_rate": 5.563723362634008e-05,
|
6835 |
+
"loss": 3.0227,
|
6836 |
+
"step": 973
|
6837 |
+
},
|
6838 |
+
{
|
6839 |
+
"epoch": 0.06365155024036792,
|
6840 |
+
"grad_norm": 1.9503111839294434,
|
6841 |
+
"learning_rate": 5.544837059738719e-05,
|
6842 |
+
"loss": 3.0759,
|
6843 |
+
"step": 974
|
6844 |
+
},
|
6845 |
+
{
|
6846 |
+
"epoch": 0.06371690090796583,
|
6847 |
+
"grad_norm": 2.188789129257202,
|
6848 |
+
"learning_rate": 5.525970562563656e-05,
|
6849 |
+
"loss": 3.2502,
|
6850 |
+
"step": 975
|
6851 |
+
},
|
6852 |
+
{
|
6853 |
+
"epoch": 0.06378225157556375,
|
6854 |
+
"grad_norm": 1.924596905708313,
|
6855 |
+
"learning_rate": 5.507123954981073e-05,
|
6856 |
+
"loss": 2.8856,
|
6857 |
+
"step": 976
|
6858 |
+
},
|
6859 |
+
{
|
6860 |
+
"epoch": 0.06384760224316166,
|
6861 |
+
"grad_norm": 2.1890134811401367,
|
6862 |
+
"learning_rate": 5.488297320774807e-05,
|
6863 |
+
"loss": 2.9215,
|
6864 |
+
"step": 977
|
6865 |
+
},
|
6866 |
+
{
|
6867 |
+
"epoch": 0.06391295291075957,
|
6868 |
+
"grad_norm": 2.169776201248169,
|
6869 |
+
"learning_rate": 5.4694907436399e-05,
|
6870 |
+
"loss": 2.8289,
|
6871 |
+
"step": 978
|
6872 |
+
},
|
6873 |
+
{
|
6874 |
+
"epoch": 0.06397830357835749,
|
6875 |
+
"grad_norm": 2.278578758239746,
|
6876 |
+
"learning_rate": 5.4507043071822284e-05,
|
6877 |
+
"loss": 3.0919,
|
6878 |
+
"step": 979
|
6879 |
+
},
|
6880 |
+
{
|
6881 |
+
"epoch": 0.0640436542459554,
|
6882 |
+
"grad_norm": 2.17686128616333,
|
6883 |
+
"learning_rate": 5.431938094918132e-05,
|
6884 |
+
"loss": 2.8256,
|
6885 |
+
"step": 980
|
6886 |
+
},
|
6887 |
+
{
|
6888 |
+
"epoch": 0.06410900491355331,
|
6889 |
+
"grad_norm": 2.2082159519195557,
|
6890 |
+
"learning_rate": 5.41319219027404e-05,
|
6891 |
+
"loss": 3.1257,
|
6892 |
+
"step": 981
|
6893 |
+
},
|
6894 |
+
{
|
6895 |
+
"epoch": 0.06417435558115124,
|
6896 |
+
"grad_norm": 2.4352824687957764,
|
6897 |
+
"learning_rate": 5.394466676586114e-05,
|
6898 |
+
"loss": 3.1052,
|
6899 |
+
"step": 982
|
6900 |
+
},
|
6901 |
+
{
|
6902 |
+
"epoch": 0.06423970624874915,
|
6903 |
+
"grad_norm": 2.4740257263183594,
|
6904 |
+
"learning_rate": 5.375761637099854e-05,
|
6905 |
+
"loss": 2.921,
|
6906 |
+
"step": 983
|
6907 |
+
},
|
6908 |
+
{
|
6909 |
+
"epoch": 0.06430505691634707,
|
6910 |
+
"grad_norm": 2.5803582668304443,
|
6911 |
+
"learning_rate": 5.357077154969742e-05,
|
6912 |
+
"loss": 3.1491,
|
6913 |
+
"step": 984
|
6914 |
+
},
|
6915 |
+
{
|
6916 |
+
"epoch": 0.06437040758394498,
|
6917 |
+
"grad_norm": 2.5305521488189697,
|
6918 |
+
"learning_rate": 5.3384133132588784e-05,
|
6919 |
+
"loss": 3.0315,
|
6920 |
+
"step": 985
|
6921 |
+
},
|
6922 |
+
{
|
6923 |
+
"epoch": 0.0644357582515429,
|
6924 |
+
"grad_norm": 2.7055225372314453,
|
6925 |
+
"learning_rate": 5.3197701949386e-05,
|
6926 |
+
"loss": 2.9618,
|
6927 |
+
"step": 986
|
6928 |
+
},
|
6929 |
+
{
|
6930 |
+
"epoch": 0.0645011089191408,
|
6931 |
+
"grad_norm": 2.6253137588500977,
|
6932 |
+
"learning_rate": 5.301147882888116e-05,
|
6933 |
+
"loss": 3.2339,
|
6934 |
+
"step": 987
|
6935 |
+
},
|
6936 |
+
{
|
6937 |
+
"epoch": 0.06456645958673872,
|
6938 |
+
"grad_norm": 2.457127332687378,
|
6939 |
+
"learning_rate": 5.28254645989414e-05,
|
6940 |
+
"loss": 3.0656,
|
6941 |
+
"step": 988
|
6942 |
+
},
|
6943 |
+
{
|
6944 |
+
"epoch": 0.06463181025433663,
|
6945 |
+
"grad_norm": 2.733246088027954,
|
6946 |
+
"learning_rate": 5.2639660086505226e-05,
|
6947 |
+
"loss": 3.2059,
|
6948 |
+
"step": 989
|
6949 |
+
},
|
6950 |
+
{
|
6951 |
+
"epoch": 0.06469716092193455,
|
6952 |
+
"grad_norm": 3.0360665321350098,
|
6953 |
+
"learning_rate": 5.2454066117578815e-05,
|
6954 |
+
"loss": 3.0883,
|
6955 |
+
"step": 990
|
6956 |
+
},
|
6957 |
+
{
|
6958 |
+
"epoch": 0.06476251158953246,
|
6959 |
+
"grad_norm": 3.4953083992004395,
|
6960 |
+
"learning_rate": 5.226868351723244e-05,
|
6961 |
+
"loss": 3.1063,
|
6962 |
+
"step": 991
|
6963 |
+
},
|
6964 |
+
{
|
6965 |
+
"epoch": 0.06482786225713037,
|
6966 |
+
"grad_norm": 3.3253588676452637,
|
6967 |
+
"learning_rate": 5.2083513109596616e-05,
|
6968 |
+
"loss": 2.8817,
|
6969 |
+
"step": 992
|
6970 |
+
},
|
6971 |
+
{
|
6972 |
+
"epoch": 0.06489321292472829,
|
6973 |
+
"grad_norm": 2.7518532276153564,
|
6974 |
+
"learning_rate": 5.189855571785859e-05,
|
6975 |
+
"loss": 2.9094,
|
6976 |
+
"step": 993
|
6977 |
+
},
|
6978 |
+
{
|
6979 |
+
"epoch": 0.0649585635923262,
|
6980 |
+
"grad_norm": 3.39292049407959,
|
6981 |
+
"learning_rate": 5.171381216425863e-05,
|
6982 |
+
"loss": 3.1637,
|
6983 |
+
"step": 994
|
6984 |
+
},
|
6985 |
+
{
|
6986 |
+
"epoch": 0.06502391425992411,
|
6987 |
+
"grad_norm": 3.054955005645752,
|
6988 |
+
"learning_rate": 5.152928327008635e-05,
|
6989 |
+
"loss": 2.5573,
|
6990 |
+
"step": 995
|
6991 |
+
},
|
6992 |
+
{
|
6993 |
+
"epoch": 0.06508926492752203,
|
6994 |
+
"grad_norm": 4.050127983093262,
|
6995 |
+
"learning_rate": 5.134496985567714e-05,
|
6996 |
+
"loss": 2.6131,
|
6997 |
+
"step": 996
|
6998 |
+
},
|
6999 |
+
{
|
7000 |
+
"epoch": 0.06515461559511994,
|
7001 |
+
"grad_norm": 3.812195301055908,
|
7002 |
+
"learning_rate": 5.116087274040837e-05,
|
7003 |
+
"loss": 3.0002,
|
7004 |
+
"step": 997
|
7005 |
+
},
|
7006 |
+
{
|
7007 |
+
"epoch": 0.06521996626271785,
|
7008 |
+
"grad_norm": 4.295674800872803,
|
7009 |
+
"learning_rate": 5.0976992742695925e-05,
|
7010 |
+
"loss": 2.9748,
|
7011 |
+
"step": 998
|
7012 |
+
},
|
7013 |
+
{
|
7014 |
+
"epoch": 0.06528531693031576,
|
7015 |
+
"grad_norm": 5.213565349578857,
|
7016 |
+
"learning_rate": 5.07933306799904e-05,
|
7017 |
+
"loss": 2.7755,
|
7018 |
+
"step": 999
|
7019 |
+
},
|
7020 |
+
{
|
7021 |
+
"epoch": 0.06535066759791368,
|
7022 |
+
"grad_norm": 7.447957992553711,
|
7023 |
+
"learning_rate": 5.060988736877366e-05,
|
7024 |
+
"loss": 2.8446,
|
7025 |
+
"step": 1000
|
7026 |
+
},
|
7027 |
+
{
|
7028 |
+
"epoch": 0.06541601826551159,
|
7029 |
+
"grad_norm": 5.239957332611084,
|
7030 |
+
"learning_rate": 5.042666362455498e-05,
|
7031 |
+
"loss": 3.1544,
|
7032 |
+
"step": 1001
|
7033 |
+
},
|
7034 |
+
{
|
7035 |
+
"epoch": 0.0654813689331095,
|
7036 |
+
"grad_norm": 5.499394416809082,
|
7037 |
+
"learning_rate": 5.024366026186755e-05,
|
7038 |
+
"loss": 3.171,
|
7039 |
+
"step": 1002
|
7040 |
+
},
|
7041 |
+
{
|
7042 |
+
"epoch": 0.06554671960070742,
|
7043 |
+
"grad_norm": 5.459829330444336,
|
7044 |
+
"learning_rate": 5.006087809426496e-05,
|
7045 |
+
"loss": 3.1688,
|
7046 |
+
"step": 1003
|
7047 |
+
},
|
7048 |
+
{
|
7049 |
+
"epoch": 0.06561207026830533,
|
7050 |
+
"grad_norm": 5.655965805053711,
|
7051 |
+
"learning_rate": 4.987831793431731e-05,
|
7052 |
+
"loss": 3.1811,
|
7053 |
+
"step": 1004
|
7054 |
+
},
|
7055 |
+
{
|
7056 |
+
"epoch": 0.06567742093590324,
|
7057 |
+
"grad_norm": 5.825551986694336,
|
7058 |
+
"learning_rate": 4.9695980593607817e-05,
|
7059 |
+
"loss": 3.0934,
|
7060 |
+
"step": 1005
|
7061 |
+
},
|
7062 |
+
{
|
7063 |
+
"epoch": 0.06574277160350116,
|
7064 |
+
"grad_norm": 6.30647087097168,
|
7065 |
+
"learning_rate": 4.9513866882729146e-05,
|
7066 |
+
"loss": 3.1495,
|
7067 |
+
"step": 1006
|
7068 |
+
},
|
7069 |
+
{
|
7070 |
+
"epoch": 0.06580812227109907,
|
7071 |
+
"grad_norm": 5.474096298217773,
|
7072 |
+
"learning_rate": 4.9331977611279777e-05,
|
7073 |
+
"loss": 3.2107,
|
7074 |
+
"step": 1007
|
7075 |
+
},
|
7076 |
+
{
|
7077 |
+
"epoch": 0.06587347293869698,
|
7078 |
+
"grad_norm": 4.435100078582764,
|
7079 |
+
"learning_rate": 4.9150313587860433e-05,
|
7080 |
+
"loss": 3.1942,
|
7081 |
+
"step": 1008
|
7082 |
+
},
|
7083 |
+
{
|
7084 |
+
"epoch": 0.0659388236062949,
|
7085 |
+
"grad_norm": 3.5297582149505615,
|
7086 |
+
"learning_rate": 4.896887562007054e-05,
|
7087 |
+
"loss": 3.0878,
|
7088 |
+
"step": 1009
|
7089 |
+
},
|
7090 |
+
{
|
7091 |
+
"epoch": 0.06600417427389281,
|
7092 |
+
"grad_norm": 2.693669557571411,
|
7093 |
+
"learning_rate": 4.8787664514504504e-05,
|
7094 |
+
"loss": 3.1296,
|
7095 |
+
"step": 1010
|
7096 |
+
},
|
7097 |
+
{
|
7098 |
+
"epoch": 0.06606952494149074,
|
7099 |
+
"grad_norm": 2.1648197174072266,
|
7100 |
+
"learning_rate": 4.860668107674823e-05,
|
7101 |
+
"loss": 3.0485,
|
7102 |
+
"step": 1011
|
7103 |
+
},
|
7104 |
+
{
|
7105 |
+
"epoch": 0.06613487560908865,
|
7106 |
+
"grad_norm": 1.8515563011169434,
|
7107 |
+
"learning_rate": 4.8425926111375506e-05,
|
7108 |
+
"loss": 2.9758,
|
7109 |
+
"step": 1012
|
7110 |
+
},
|
7111 |
+
{
|
7112 |
+
"epoch": 0.06620022627668656,
|
7113 |
+
"grad_norm": 1.8367582559585571,
|
7114 |
+
"learning_rate": 4.824540042194443e-05,
|
7115 |
+
"loss": 2.9185,
|
7116 |
+
"step": 1013
|
7117 |
+
},
|
7118 |
+
{
|
7119 |
+
"epoch": 0.06626557694428448,
|
7120 |
+
"grad_norm": 1.9552206993103027,
|
7121 |
+
"learning_rate": 4.8065104810993856e-05,
|
7122 |
+
"loss": 3.2476,
|
7123 |
+
"step": 1014
|
7124 |
+
},
|
7125 |
+
{
|
7126 |
+
"epoch": 0.06633092761188239,
|
7127 |
+
"grad_norm": 1.8460843563079834,
|
7128 |
+
"learning_rate": 4.788504008003978e-05,
|
7129 |
+
"loss": 3.0385,
|
7130 |
+
"step": 1015
|
7131 |
+
},
|
7132 |
+
{
|
7133 |
+
"epoch": 0.0663962782794803,
|
7134 |
+
"grad_norm": 1.702373743057251,
|
7135 |
+
"learning_rate": 4.770520702957182e-05,
|
7136 |
+
"loss": 2.7663,
|
7137 |
+
"step": 1016
|
7138 |
+
},
|
7139 |
+
{
|
7140 |
+
"epoch": 0.06646162894707822,
|
7141 |
+
"grad_norm": 2.203813076019287,
|
7142 |
+
"learning_rate": 4.752560645904962e-05,
|
7143 |
+
"loss": 3.2382,
|
7144 |
+
"step": 1017
|
7145 |
+
},
|
7146 |
+
{
|
7147 |
+
"epoch": 0.06652697961467613,
|
7148 |
+
"grad_norm": 1.8665012121200562,
|
7149 |
+
"learning_rate": 4.734623916689941e-05,
|
7150 |
+
"loss": 3.1152,
|
7151 |
+
"step": 1018
|
7152 |
+
},
|
7153 |
+
{
|
7154 |
+
"epoch": 0.06659233028227404,
|
7155 |
+
"grad_norm": 1.8985743522644043,
|
7156 |
+
"learning_rate": 4.716710595051022e-05,
|
7157 |
+
"loss": 3.1682,
|
7158 |
+
"step": 1019
|
7159 |
+
},
|
7160 |
+
{
|
7161 |
+
"epoch": 0.06665768094987196,
|
7162 |
+
"grad_norm": 1.9103392362594604,
|
7163 |
+
"learning_rate": 4.698820760623064e-05,
|
7164 |
+
"loss": 3.2239,
|
7165 |
+
"step": 1020
|
7166 |
+
},
|
7167 |
+
{
|
7168 |
+
"epoch": 0.06672303161746987,
|
7169 |
+
"grad_norm": 1.9629900455474854,
|
7170 |
+
"learning_rate": 4.6809544929365004e-05,
|
7171 |
+
"loss": 2.9791,
|
7172 |
+
"step": 1021
|
7173 |
+
},
|
7174 |
+
{
|
7175 |
+
"epoch": 0.06678838228506778,
|
7176 |
+
"grad_norm": 1.8845793008804321,
|
7177 |
+
"learning_rate": 4.663111871417e-05,
|
7178 |
+
"loss": 3.0752,
|
7179 |
+
"step": 1022
|
7180 |
+
},
|
7181 |
+
{
|
7182 |
+
"epoch": 0.0668537329526657,
|
7183 |
+
"grad_norm": 2.029320001602173,
|
7184 |
+
"learning_rate": 4.645292975385111e-05,
|
7185 |
+
"loss": 2.8707,
|
7186 |
+
"step": 1023
|
7187 |
+
},
|
7188 |
+
{
|
7189 |
+
"epoch": 0.06691908362026361,
|
7190 |
+
"grad_norm": 2.452892303466797,
|
7191 |
+
"learning_rate": 4.627497884055912e-05,
|
7192 |
+
"loss": 3.2065,
|
7193 |
+
"step": 1024
|
7194 |
+
},
|
7195 |
+
{
|
7196 |
+
"epoch": 0.06698443428786152,
|
7197 |
+
"grad_norm": 2.150029182434082,
|
7198 |
+
"learning_rate": 4.609726676538652e-05,
|
7199 |
+
"loss": 3.2065,
|
7200 |
+
"step": 1025
|
7201 |
+
},
|
7202 |
+
{
|
7203 |
+
"epoch": 0.06704978495545943,
|
7204 |
+
"grad_norm": 2.1136484146118164,
|
7205 |
+
"learning_rate": 4.591979431836402e-05,
|
7206 |
+
"loss": 2.8519,
|
7207 |
+
"step": 1026
|
7208 |
+
},
|
7209 |
+
{
|
7210 |
+
"epoch": 0.06711513562305735,
|
7211 |
+
"grad_norm": 2.0389273166656494,
|
7212 |
+
"learning_rate": 4.574256228845706e-05,
|
7213 |
+
"loss": 2.9751,
|
7214 |
+
"step": 1027
|
7215 |
+
},
|
7216 |
+
{
|
7217 |
+
"epoch": 0.06718048629065526,
|
7218 |
+
"grad_norm": 2.3239119052886963,
|
7219 |
+
"learning_rate": 4.5565571463562365e-05,
|
7220 |
+
"loss": 3.0121,
|
7221 |
+
"step": 1028
|
7222 |
+
},
|
7223 |
+
{
|
7224 |
+
"epoch": 0.06724583695825317,
|
7225 |
+
"grad_norm": 2.3797855377197266,
|
7226 |
+
"learning_rate": 4.5388822630504256e-05,
|
7227 |
+
"loss": 3.0736,
|
7228 |
+
"step": 1029
|
7229 |
+
},
|
7230 |
+
{
|
7231 |
+
"epoch": 0.06731118762585109,
|
7232 |
+
"grad_norm": 2.2766530513763428,
|
7233 |
+
"learning_rate": 4.521231657503132e-05,
|
7234 |
+
"loss": 2.9345,
|
7235 |
+
"step": 1030
|
7236 |
+
},
|
7237 |
+
{
|
7238 |
+
"epoch": 0.067376538293449,
|
7239 |
+
"grad_norm": 2.233914375305176,
|
7240 |
+
"learning_rate": 4.503605408181286e-05,
|
7241 |
+
"loss": 2.9631,
|
7242 |
+
"step": 1031
|
7243 |
+
},
|
7244 |
+
{
|
7245 |
+
"epoch": 0.06744188896104691,
|
7246 |
+
"grad_norm": 2.228635311126709,
|
7247 |
+
"learning_rate": 4.486003593443537e-05,
|
7248 |
+
"loss": 2.9136,
|
7249 |
+
"step": 1032
|
7250 |
+
},
|
7251 |
+
{
|
7252 |
+
"epoch": 0.06750723962864483,
|
7253 |
+
"grad_norm": 2.3976728916168213,
|
7254 |
+
"learning_rate": 4.468426291539914e-05,
|
7255 |
+
"loss": 2.7201,
|
7256 |
+
"step": 1033
|
7257 |
+
},
|
7258 |
+
{
|
7259 |
+
"epoch": 0.06757259029624274,
|
7260 |
+
"grad_norm": 2.1901347637176514,
|
7261 |
+
"learning_rate": 4.4508735806114654e-05,
|
7262 |
+
"loss": 2.8576,
|
7263 |
+
"step": 1034
|
7264 |
+
},
|
7265 |
+
{
|
7266 |
+
"epoch": 0.06763794096384065,
|
7267 |
+
"grad_norm": 2.38387131690979,
|
7268 |
+
"learning_rate": 4.433345538689929e-05,
|
7269 |
+
"loss": 2.8988,
|
7270 |
+
"step": 1035
|
7271 |
+
},
|
7272 |
+
{
|
7273 |
+
"epoch": 0.06770329163143857,
|
7274 |
+
"grad_norm": 2.552875518798828,
|
7275 |
+
"learning_rate": 4.415842243697369e-05,
|
7276 |
+
"loss": 3.2502,
|
7277 |
+
"step": 1036
|
7278 |
+
},
|
7279 |
+
{
|
7280 |
+
"epoch": 0.06776864229903648,
|
7281 |
+
"grad_norm": 3.017981767654419,
|
7282 |
+
"learning_rate": 4.39836377344583e-05,
|
7283 |
+
"loss": 3.1039,
|
7284 |
+
"step": 1037
|
7285 |
+
},
|
7286 |
+
{
|
7287 |
+
"epoch": 0.0678339929666344,
|
7288 |
+
"grad_norm": 2.432250738143921,
|
7289 |
+
"learning_rate": 4.380910205637012e-05,
|
7290 |
+
"loss": 2.9834,
|
7291 |
+
"step": 1038
|
7292 |
+
},
|
7293 |
+
{
|
7294 |
+
"epoch": 0.06789934363423232,
|
7295 |
+
"grad_norm": 3.124380350112915,
|
7296 |
+
"learning_rate": 4.363481617861893e-05,
|
7297 |
+
"loss": 3.3313,
|
7298 |
+
"step": 1039
|
7299 |
+
},
|
7300 |
+
{
|
7301 |
+
"epoch": 0.06796469430183023,
|
7302 |
+
"grad_norm": 4.3898091316223145,
|
7303 |
+
"learning_rate": 4.346078087600412e-05,
|
7304 |
+
"loss": 3.1151,
|
7305 |
+
"step": 1040
|
7306 |
+
},
|
7307 |
+
{
|
7308 |
+
"epoch": 0.06803004496942815,
|
7309 |
+
"grad_norm": 3.011410713195801,
|
7310 |
+
"learning_rate": 4.3286996922211034e-05,
|
7311 |
+
"loss": 3.1023,
|
7312 |
+
"step": 1041
|
7313 |
+
},
|
7314 |
+
{
|
7315 |
+
"epoch": 0.06809539563702606,
|
7316 |
+
"grad_norm": 2.738132953643799,
|
7317 |
+
"learning_rate": 4.311346508980772e-05,
|
7318 |
+
"loss": 2.8665,
|
7319 |
+
"step": 1042
|
7320 |
+
},
|
7321 |
+
{
|
7322 |
+
"epoch": 0.06816074630462397,
|
7323 |
+
"grad_norm": 3.4573757648468018,
|
7324 |
+
"learning_rate": 4.2940186150241365e-05,
|
7325 |
+
"loss": 2.9487,
|
7326 |
+
"step": 1043
|
7327 |
+
},
|
7328 |
+
{
|
7329 |
+
"epoch": 0.06822609697222189,
|
7330 |
+
"grad_norm": 3.1583240032196045,
|
7331 |
+
"learning_rate": 4.27671608738349e-05,
|
7332 |
+
"loss": 2.9844,
|
7333 |
+
"step": 1044
|
7334 |
+
},
|
7335 |
+
{
|
7336 |
+
"epoch": 0.0682914476398198,
|
7337 |
+
"grad_norm": 3.2991983890533447,
|
7338 |
+
"learning_rate": 4.2594390029783534e-05,
|
7339 |
+
"loss": 3.1467,
|
7340 |
+
"step": 1045
|
7341 |
+
},
|
7342 |
+
{
|
7343 |
+
"epoch": 0.06835679830741771,
|
7344 |
+
"grad_norm": 3.0215463638305664,
|
7345 |
+
"learning_rate": 4.242187438615153e-05,
|
7346 |
+
"loss": 2.9524,
|
7347 |
+
"step": 1046
|
7348 |
+
},
|
7349 |
+
{
|
7350 |
+
"epoch": 0.06842214897501563,
|
7351 |
+
"grad_norm": 3.2807435989379883,
|
7352 |
+
"learning_rate": 4.224961470986849e-05,
|
7353 |
+
"loss": 2.8558,
|
7354 |
+
"step": 1047
|
7355 |
+
},
|
7356 |
+
{
|
7357 |
+
"epoch": 0.06848749964261354,
|
7358 |
+
"grad_norm": 4.612066268920898,
|
7359 |
+
"learning_rate": 4.207761176672614e-05,
|
7360 |
+
"loss": 3.1407,
|
7361 |
+
"step": 1048
|
7362 |
+
},
|
7363 |
+
{
|
7364 |
+
"epoch": 0.06855285031021145,
|
7365 |
+
"grad_norm": 4.230472087860107,
|
7366 |
+
"learning_rate": 4.190586632137491e-05,
|
7367 |
+
"loss": 2.8332,
|
7368 |
+
"step": 1049
|
7369 |
+
},
|
7370 |
+
{
|
7371 |
+
"epoch": 0.06861820097780937,
|
7372 |
+
"grad_norm": 8.014326095581055,
|
7373 |
+
"learning_rate": 4.173437913732048e-05,
|
7374 |
+
"loss": 2.8739,
|
7375 |
+
"step": 1050
|
7376 |
+
},
|
7377 |
+
{
|
7378 |
+
"epoch": 0.06868355164540728,
|
7379 |
+
"grad_norm": 4.5770087242126465,
|
7380 |
+
"learning_rate": 4.156315097692037e-05,
|
7381 |
+
"loss": 3.0627,
|
7382 |
+
"step": 1051
|
7383 |
+
},
|
7384 |
+
{
|
7385 |
+
"epoch": 0.06874890231300519,
|
7386 |
+
"grad_norm": 4.76311731338501,
|
7387 |
+
"learning_rate": 4.139218260138074e-05,
|
7388 |
+
"loss": 3.0922,
|
7389 |
+
"step": 1052
|
7390 |
+
},
|
7391 |
+
{
|
7392 |
+
"epoch": 0.0688142529806031,
|
7393 |
+
"grad_norm": 4.782283782958984,
|
7394 |
+
"learning_rate": 4.12214747707527e-05,
|
7395 |
+
"loss": 3.0974,
|
7396 |
+
"step": 1053
|
7397 |
+
},
|
7398 |
+
{
|
7399 |
+
"epoch": 0.06887960364820102,
|
7400 |
+
"grad_norm": 4.817895412445068,
|
7401 |
+
"learning_rate": 4.1051028243929125e-05,
|
7402 |
+
"loss": 3.2717,
|
7403 |
+
"step": 1054
|
7404 |
+
},
|
7405 |
+
{
|
7406 |
+
"epoch": 0.06894495431579893,
|
7407 |
+
"grad_norm": 4.34201717376709,
|
7408 |
+
"learning_rate": 4.088084377864135e-05,
|
7409 |
+
"loss": 3.161,
|
7410 |
+
"step": 1055
|
7411 |
+
},
|
7412 |
+
{
|
7413 |
+
"epoch": 0.06901030498339684,
|
7414 |
+
"grad_norm": 4.493038177490234,
|
7415 |
+
"learning_rate": 4.07109221314556e-05,
|
7416 |
+
"loss": 3.2462,
|
7417 |
+
"step": 1056
|
7418 |
+
},
|
7419 |
+
{
|
7420 |
+
"epoch": 0.06907565565099476,
|
7421 |
+
"grad_norm": 4.028043270111084,
|
7422 |
+
"learning_rate": 4.054126405776971e-05,
|
7423 |
+
"loss": 3.2795,
|
7424 |
+
"step": 1057
|
7425 |
+
},
|
7426 |
+
{
|
7427 |
+
"epoch": 0.06914100631859267,
|
7428 |
+
"grad_norm": 3.4543871879577637,
|
7429 |
+
"learning_rate": 4.037187031180985e-05,
|
7430 |
+
"loss": 3.0808,
|
7431 |
+
"step": 1058
|
7432 |
+
},
|
7433 |
+
{
|
7434 |
+
"epoch": 0.06920635698619058,
|
7435 |
+
"grad_norm": 2.844466209411621,
|
7436 |
+
"learning_rate": 4.020274164662707e-05,
|
7437 |
+
"loss": 3.1561,
|
7438 |
+
"step": 1059
|
7439 |
+
},
|
7440 |
+
{
|
7441 |
+
"epoch": 0.0692717076537885,
|
7442 |
+
"grad_norm": 2.4248085021972656,
|
7443 |
+
"learning_rate": 4.003387881409397e-05,
|
7444 |
+
"loss": 3.0333,
|
7445 |
+
"step": 1060
|
7446 |
+
},
|
7447 |
+
{
|
7448 |
+
"epoch": 0.06933705832138641,
|
7449 |
+
"grad_norm": 1.8686552047729492,
|
7450 |
+
"learning_rate": 3.986528256490141e-05,
|
7451 |
+
"loss": 3.0051,
|
7452 |
+
"step": 1061
|
7453 |
+
},
|
7454 |
+
{
|
7455 |
+
"epoch": 0.06940240898898432,
|
7456 |
+
"grad_norm": 1.6639132499694824,
|
7457 |
+
"learning_rate": 3.969695364855511e-05,
|
7458 |
+
"loss": 2.751,
|
7459 |
+
"step": 1062
|
7460 |
+
},
|
7461 |
+
{
|
7462 |
+
"epoch": 0.06946775965658224,
|
7463 |
+
"grad_norm": 1.6407933235168457,
|
7464 |
+
"learning_rate": 3.952889281337235e-05,
|
7465 |
+
"loss": 3.0891,
|
7466 |
+
"step": 1063
|
7467 |
+
},
|
7468 |
+
{
|
7469 |
+
"epoch": 0.06953311032418015,
|
7470 |
+
"grad_norm": 1.575294852256775,
|
7471 |
+
"learning_rate": 3.93611008064786e-05,
|
7472 |
+
"loss": 2.9513,
|
7473 |
+
"step": 1064
|
7474 |
+
},
|
7475 |
+
{
|
7476 |
+
"epoch": 0.06959846099177806,
|
7477 |
+
"grad_norm": 1.5433897972106934,
|
7478 |
+
"learning_rate": 3.9193578373804364e-05,
|
7479 |
+
"loss": 2.9242,
|
7480 |
+
"step": 1065
|
7481 |
+
},
|
7482 |
+
{
|
7483 |
+
"epoch": 0.06966381165937598,
|
7484 |
+
"grad_norm": 1.605277419090271,
|
7485 |
+
"learning_rate": 3.90263262600816e-05,
|
7486 |
+
"loss": 2.9362,
|
7487 |
+
"step": 1066
|
7488 |
+
},
|
7489 |
+
{
|
7490 |
+
"epoch": 0.06972916232697389,
|
7491 |
+
"grad_norm": 1.819908618927002,
|
7492 |
+
"learning_rate": 3.88593452088406e-05,
|
7493 |
+
"loss": 3.0245,
|
7494 |
+
"step": 1067
|
7495 |
+
},
|
7496 |
+
{
|
7497 |
+
"epoch": 0.06979451299457182,
|
7498 |
+
"grad_norm": 1.7380348443984985,
|
7499 |
+
"learning_rate": 3.869263596240661e-05,
|
7500 |
+
"loss": 2.8164,
|
7501 |
+
"step": 1068
|
7502 |
+
},
|
7503 |
+
{
|
7504 |
+
"epoch": 0.06985986366216973,
|
7505 |
+
"grad_norm": 1.6103025674819946,
|
7506 |
+
"learning_rate": 3.8526199261896544e-05,
|
7507 |
+
"loss": 2.733,
|
7508 |
+
"step": 1069
|
7509 |
+
},
|
7510 |
+
{
|
7511 |
+
"epoch": 0.06992521432976764,
|
7512 |
+
"grad_norm": 1.7424949407577515,
|
7513 |
+
"learning_rate": 3.836003584721577e-05,
|
7514 |
+
"loss": 2.9838,
|
7515 |
+
"step": 1070
|
7516 |
+
},
|
7517 |
+
{
|
7518 |
+
"epoch": 0.06999056499736556,
|
7519 |
+
"grad_norm": 1.7999818325042725,
|
7520 |
+
"learning_rate": 3.8194146457054655e-05,
|
7521 |
+
"loss": 2.931,
|
7522 |
+
"step": 1071
|
7523 |
+
},
|
7524 |
+
{
|
7525 |
+
"epoch": 0.07005591566496347,
|
7526 |
+
"grad_norm": 1.872725009918213,
|
7527 |
+
"learning_rate": 3.802853182888543e-05,
|
7528 |
+
"loss": 2.9594,
|
7529 |
+
"step": 1072
|
7530 |
+
},
|
7531 |
+
{
|
7532 |
+
"epoch": 0.07012126633256138,
|
7533 |
+
"grad_norm": 1.837857723236084,
|
7534 |
+
"learning_rate": 3.786319269895877e-05,
|
7535 |
+
"loss": 2.8684,
|
7536 |
+
"step": 1073
|
7537 |
+
},
|
7538 |
+
{
|
7539 |
+
"epoch": 0.0701866170001593,
|
7540 |
+
"grad_norm": 1.7518945932388306,
|
7541 |
+
"learning_rate": 3.769812980230074e-05,
|
7542 |
+
"loss": 2.7691,
|
7543 |
+
"step": 1074
|
7544 |
+
},
|
7545 |
+
{
|
7546 |
+
"epoch": 0.07025196766775721,
|
7547 |
+
"grad_norm": 1.7127944231033325,
|
7548 |
+
"learning_rate": 3.7533343872709294e-05,
|
7549 |
+
"loss": 3.0325,
|
7550 |
+
"step": 1075
|
7551 |
+
},
|
7552 |
+
{
|
7553 |
+
"epoch": 0.07031731833535512,
|
7554 |
+
"grad_norm": 2.217871904373169,
|
7555 |
+
"learning_rate": 3.736883564275112e-05,
|
7556 |
+
"loss": 3.2894,
|
7557 |
+
"step": 1076
|
7558 |
+
},
|
7559 |
+
{
|
7560 |
+
"epoch": 0.07038266900295304,
|
7561 |
+
"grad_norm": 2.020307779312134,
|
7562 |
+
"learning_rate": 3.7204605843758386e-05,
|
7563 |
+
"loss": 2.9408,
|
7564 |
+
"step": 1077
|
7565 |
+
},
|
7566 |
+
{
|
7567 |
+
"epoch": 0.07044801967055095,
|
7568 |
+
"grad_norm": 1.8516186475753784,
|
7569 |
+
"learning_rate": 3.704065520582549e-05,
|
7570 |
+
"loss": 2.9536,
|
7571 |
+
"step": 1078
|
7572 |
+
},
|
7573 |
+
{
|
7574 |
+
"epoch": 0.07051337033814886,
|
7575 |
+
"grad_norm": 2.386331081390381,
|
7576 |
+
"learning_rate": 3.6876984457805786e-05,
|
7577 |
+
"loss": 3.068,
|
7578 |
+
"step": 1079
|
7579 |
+
},
|
7580 |
+
{
|
7581 |
+
"epoch": 0.07057872100574678,
|
7582 |
+
"grad_norm": 2.0858514308929443,
|
7583 |
+
"learning_rate": 3.671359432730834e-05,
|
7584 |
+
"loss": 2.819,
|
7585 |
+
"step": 1080
|
7586 |
+
},
|
7587 |
+
{
|
7588 |
+
"epoch": 0.07064407167334469,
|
7589 |
+
"grad_norm": 2.173201322555542,
|
7590 |
+
"learning_rate": 3.655048554069478e-05,
|
7591 |
+
"loss": 2.8832,
|
7592 |
+
"step": 1081
|
7593 |
+
},
|
7594 |
+
{
|
7595 |
+
"epoch": 0.0707094223409426,
|
7596 |
+
"grad_norm": 2.1481897830963135,
|
7597 |
+
"learning_rate": 3.638765882307589e-05,
|
7598 |
+
"loss": 2.9347,
|
7599 |
+
"step": 1082
|
7600 |
+
},
|
7601 |
+
{
|
7602 |
+
"epoch": 0.07077477300854051,
|
7603 |
+
"grad_norm": 2.0271878242492676,
|
7604 |
+
"learning_rate": 3.6225114898308634e-05,
|
7605 |
+
"loss": 2.9075,
|
7606 |
+
"step": 1083
|
7607 |
+
},
|
7608 |
+
{
|
7609 |
+
"epoch": 0.07084012367613843,
|
7610 |
+
"grad_norm": 2.134906530380249,
|
7611 |
+
"learning_rate": 3.6062854488992714e-05,
|
7612 |
+
"loss": 3.1858,
|
7613 |
+
"step": 1084
|
7614 |
+
},
|
7615 |
+
{
|
7616 |
+
"epoch": 0.07090547434373634,
|
7617 |
+
"grad_norm": 2.5907504558563232,
|
7618 |
+
"learning_rate": 3.5900878316467454e-05,
|
7619 |
+
"loss": 3.0442,
|
7620 |
+
"step": 1085
|
7621 |
+
},
|
7622 |
+
{
|
7623 |
+
"epoch": 0.07097082501133425,
|
7624 |
+
"grad_norm": 2.3115131855010986,
|
7625 |
+
"learning_rate": 3.573918710080857e-05,
|
7626 |
+
"loss": 3.0885,
|
7627 |
+
"step": 1086
|
7628 |
+
},
|
7629 |
+
{
|
7630 |
+
"epoch": 0.07103617567893217,
|
7631 |
+
"grad_norm": 2.489321231842041,
|
7632 |
+
"learning_rate": 3.5577781560825066e-05,
|
7633 |
+
"loss": 3.0164,
|
7634 |
+
"step": 1087
|
7635 |
+
},
|
7636 |
+
{
|
7637 |
+
"epoch": 0.07110152634653008,
|
7638 |
+
"grad_norm": 2.4704771041870117,
|
7639 |
+
"learning_rate": 3.541666241405588e-05,
|
7640 |
+
"loss": 3.0861,
|
7641 |
+
"step": 1088
|
7642 |
+
},
|
7643 |
+
{
|
7644 |
+
"epoch": 0.071166877014128,
|
7645 |
+
"grad_norm": 2.6111903190612793,
|
7646 |
+
"learning_rate": 3.5255830376766764e-05,
|
7647 |
+
"loss": 3.1115,
|
7648 |
+
"step": 1089
|
7649 |
+
},
|
7650 |
+
{
|
7651 |
+
"epoch": 0.07123222768172591,
|
7652 |
+
"grad_norm": 3.0150208473205566,
|
7653 |
+
"learning_rate": 3.509528616394716e-05,
|
7654 |
+
"loss": 2.9542,
|
7655 |
+
"step": 1090
|
7656 |
+
},
|
7657 |
+
{
|
7658 |
+
"epoch": 0.07129757834932382,
|
7659 |
+
"grad_norm": 2.8812525272369385,
|
7660 |
+
"learning_rate": 3.4935030489306883e-05,
|
7661 |
+
"loss": 3.2642,
|
7662 |
+
"step": 1091
|
7663 |
+
},
|
7664 |
+
{
|
7665 |
+
"epoch": 0.07136292901692173,
|
7666 |
+
"grad_norm": 3.689237117767334,
|
7667 |
+
"learning_rate": 3.4775064065273165e-05,
|
7668 |
+
"loss": 2.8846,
|
7669 |
+
"step": 1092
|
7670 |
+
},
|
7671 |
+
{
|
7672 |
+
"epoch": 0.07142827968451965,
|
7673 |
+
"grad_norm": 2.6156420707702637,
|
7674 |
+
"learning_rate": 3.4615387602987236e-05,
|
7675 |
+
"loss": 2.9664,
|
7676 |
+
"step": 1093
|
7677 |
+
},
|
7678 |
+
{
|
7679 |
+
"epoch": 0.07149363035211756,
|
7680 |
+
"grad_norm": 2.7805891036987305,
|
7681 |
+
"learning_rate": 3.445600181230134e-05,
|
7682 |
+
"loss": 2.9319,
|
7683 |
+
"step": 1094
|
7684 |
+
},
|
7685 |
+
{
|
7686 |
+
"epoch": 0.07155898101971547,
|
7687 |
+
"grad_norm": 3.942396402359009,
|
7688 |
+
"learning_rate": 3.429690740177549e-05,
|
7689 |
+
"loss": 3.1113,
|
7690 |
+
"step": 1095
|
7691 |
+
},
|
7692 |
+
{
|
7693 |
+
"epoch": 0.0716243316873134,
|
7694 |
+
"grad_norm": 3.323171377182007,
|
7695 |
+
"learning_rate": 3.413810507867436e-05,
|
7696 |
+
"loss": 2.9402,
|
7697 |
+
"step": 1096
|
7698 |
+
},
|
7699 |
+
{
|
7700 |
+
"epoch": 0.07168968235491131,
|
7701 |
+
"grad_norm": 3.7846312522888184,
|
7702 |
+
"learning_rate": 3.397959554896415e-05,
|
7703 |
+
"loss": 2.9553,
|
7704 |
+
"step": 1097
|
7705 |
+
},
|
7706 |
+
{
|
7707 |
+
"epoch": 0.07175503302250923,
|
7708 |
+
"grad_norm": 3.664355516433716,
|
7709 |
+
"learning_rate": 3.3821379517309405e-05,
|
7710 |
+
"loss": 2.9913,
|
7711 |
+
"step": 1098
|
7712 |
+
},
|
7713 |
+
{
|
7714 |
+
"epoch": 0.07182038369010714,
|
7715 |
+
"grad_norm": 4.381472587585449,
|
7716 |
+
"learning_rate": 3.3663457687069924e-05,
|
7717 |
+
"loss": 2.6435,
|
7718 |
+
"step": 1099
|
7719 |
+
},
|
7720 |
+
{
|
7721 |
+
"epoch": 0.07188573435770505,
|
7722 |
+
"grad_norm": 7.457712173461914,
|
7723 |
+
"learning_rate": 3.350583076029754e-05,
|
7724 |
+
"loss": 2.7886,
|
7725 |
+
"step": 1100
|
7726 |
+
},
|
7727 |
+
{
|
7728 |
+
"epoch": 0.07195108502530297,
|
7729 |
+
"grad_norm": 2.3144917488098145,
|
7730 |
+
"learning_rate": 3.334849943773323e-05,
|
7731 |
+
"loss": 3.0036,
|
7732 |
+
"step": 1101
|
7733 |
+
},
|
7734 |
+
{
|
7735 |
+
"epoch": 0.07201643569290088,
|
7736 |
+
"grad_norm": 2.644132375717163,
|
7737 |
+
"learning_rate": 3.319146441880371e-05,
|
7738 |
+
"loss": 2.9626,
|
7739 |
+
"step": 1102
|
7740 |
+
},
|
7741 |
+
{
|
7742 |
+
"epoch": 0.07208178636049879,
|
7743 |
+
"grad_norm": 2.9737014770507812,
|
7744 |
+
"learning_rate": 3.3034726401618444e-05,
|
7745 |
+
"loss": 2.8726,
|
7746 |
+
"step": 1103
|
7747 |
+
},
|
7748 |
+
{
|
7749 |
+
"epoch": 0.0721471370280967,
|
7750 |
+
"grad_norm": 3.090719223022461,
|
7751 |
+
"learning_rate": 3.28782860829667e-05,
|
7752 |
+
"loss": 3.0025,
|
7753 |
+
"step": 1104
|
7754 |
+
},
|
7755 |
+
{
|
7756 |
+
"epoch": 0.07221248769569462,
|
7757 |
+
"grad_norm": 3.2808306217193604,
|
7758 |
+
"learning_rate": 3.272214415831418e-05,
|
7759 |
+
"loss": 3.0247,
|
7760 |
+
"step": 1105
|
7761 |
+
},
|
7762 |
+
{
|
7763 |
+
"epoch": 0.07227783836329253,
|
7764 |
+
"grad_norm": 3.4771127700805664,
|
7765 |
+
"learning_rate": 3.2566301321800085e-05,
|
7766 |
+
"loss": 3.0319,
|
7767 |
+
"step": 1106
|
7768 |
+
},
|
7769 |
+
{
|
7770 |
+
"epoch": 0.07234318903089045,
|
7771 |
+
"grad_norm": 3.013876438140869,
|
7772 |
+
"learning_rate": 3.241075826623401e-05,
|
7773 |
+
"loss": 3.0181,
|
7774 |
+
"step": 1107
|
7775 |
+
},
|
7776 |
+
{
|
7777 |
+
"epoch": 0.07240853969848836,
|
7778 |
+
"grad_norm": 2.801342010498047,
|
7779 |
+
"learning_rate": 3.225551568309284e-05,
|
7780 |
+
"loss": 3.011,
|
7781 |
+
"step": 1108
|
7782 |
+
},
|
7783 |
+
{
|
7784 |
+
"epoch": 0.07247389036608627,
|
7785 |
+
"grad_norm": 2.3741440773010254,
|
7786 |
+
"learning_rate": 3.210057426251773e-05,
|
7787 |
+
"loss": 3.1298,
|
7788 |
+
"step": 1109
|
7789 |
+
},
|
7790 |
+
{
|
7791 |
+
"epoch": 0.07253924103368418,
|
7792 |
+
"grad_norm": 2.518806219100952,
|
7793 |
+
"learning_rate": 3.1945934693310896e-05,
|
7794 |
+
"loss": 3.1574,
|
7795 |
+
"step": 1110
|
7796 |
+
},
|
7797 |
+
{
|
7798 |
+
"epoch": 0.0726045917012821,
|
7799 |
+
"grad_norm": 1.9327685832977295,
|
7800 |
+
"learning_rate": 3.179159766293282e-05,
|
7801 |
+
"loss": 3.0873,
|
7802 |
+
"step": 1111
|
7803 |
+
},
|
7804 |
+
{
|
7805 |
+
"epoch": 0.07266994236888001,
|
7806 |
+
"grad_norm": 1.7328757047653198,
|
7807 |
+
"learning_rate": 3.163756385749889e-05,
|
7808 |
+
"loss": 2.9867,
|
7809 |
+
"step": 1112
|
7810 |
+
},
|
7811 |
+
{
|
7812 |
+
"epoch": 0.07273529303647792,
|
7813 |
+
"grad_norm": 1.595176339149475,
|
7814 |
+
"learning_rate": 3.148383396177653e-05,
|
7815 |
+
"loss": 2.7885,
|
7816 |
+
"step": 1113
|
7817 |
+
},
|
7818 |
+
{
|
7819 |
+
"epoch": 0.07280064370407584,
|
7820 |
+
"grad_norm": 1.6727737188339233,
|
7821 |
+
"learning_rate": 3.133040865918213e-05,
|
7822 |
+
"loss": 3.0675,
|
7823 |
+
"step": 1114
|
7824 |
+
},
|
7825 |
+
{
|
7826 |
+
"epoch": 0.07286599437167375,
|
7827 |
+
"grad_norm": 1.6531821489334106,
|
7828 |
+
"learning_rate": 3.117728863177796e-05,
|
7829 |
+
"loss": 2.9855,
|
7830 |
+
"step": 1115
|
7831 |
+
},
|
7832 |
+
{
|
7833 |
+
"epoch": 0.07293134503927166,
|
7834 |
+
"grad_norm": 1.6658819913864136,
|
7835 |
+
"learning_rate": 3.102447456026919e-05,
|
7836 |
+
"loss": 2.9409,
|
7837 |
+
"step": 1116
|
7838 |
+
},
|
7839 |
+
{
|
7840 |
+
"epoch": 0.07299669570686958,
|
7841 |
+
"grad_norm": 1.7489627599716187,
|
7842 |
+
"learning_rate": 3.0871967124000834e-05,
|
7843 |
+
"loss": 2.9746,
|
7844 |
+
"step": 1117
|
7845 |
+
},
|
7846 |
+
{
|
7847 |
+
"epoch": 0.07306204637446749,
|
7848 |
+
"grad_norm": 1.6224987506866455,
|
7849 |
+
"learning_rate": 3.0719767000954714e-05,
|
7850 |
+
"loss": 2.8386,
|
7851 |
+
"step": 1118
|
7852 |
+
},
|
7853 |
+
{
|
7854 |
+
"epoch": 0.0731273970420654,
|
7855 |
+
"grad_norm": 1.608250379562378,
|
7856 |
+
"learning_rate": 3.056787486774656e-05,
|
7857 |
+
"loss": 2.9587,
|
7858 |
+
"step": 1119
|
7859 |
+
},
|
7860 |
+
{
|
7861 |
+
"epoch": 0.07319274770966332,
|
7862 |
+
"grad_norm": 1.9854613542556763,
|
7863 |
+
"learning_rate": 3.041629139962283e-05,
|
7864 |
+
"loss": 3.0895,
|
7865 |
+
"step": 1120
|
7866 |
+
},
|
7867 |
+
{
|
7868 |
+
"epoch": 0.07325809837726123,
|
7869 |
+
"grad_norm": 1.7103354930877686,
|
7870 |
+
"learning_rate": 3.0265017270457775e-05,
|
7871 |
+
"loss": 2.9426,
|
7872 |
+
"step": 1121
|
7873 |
+
},
|
7874 |
+
{
|
7875 |
+
"epoch": 0.07332344904485914,
|
7876 |
+
"grad_norm": 1.956892728805542,
|
7877 |
+
"learning_rate": 3.0114053152750556e-05,
|
7878 |
+
"loss": 2.9905,
|
7879 |
+
"step": 1122
|
7880 |
+
},
|
7881 |
+
{
|
7882 |
+
"epoch": 0.07338879971245706,
|
7883 |
+
"grad_norm": 2.029660940170288,
|
7884 |
+
"learning_rate": 2.9963399717622077e-05,
|
7885 |
+
"loss": 3.1778,
|
7886 |
+
"step": 1123
|
7887 |
+
},
|
7888 |
+
{
|
7889 |
+
"epoch": 0.07345415038005497,
|
7890 |
+
"grad_norm": 1.9287523031234741,
|
7891 |
+
"learning_rate": 2.98130576348121e-05,
|
7892 |
+
"loss": 2.8585,
|
7893 |
+
"step": 1124
|
7894 |
+
},
|
7895 |
+
{
|
7896 |
+
"epoch": 0.0735195010476529,
|
7897 |
+
"grad_norm": 1.9359210729599,
|
7898 |
+
"learning_rate": 2.966302757267625e-05,
|
7899 |
+
"loss": 3.0234,
|
7900 |
+
"step": 1125
|
7901 |
+
},
|
7902 |
+
{
|
7903 |
+
"epoch": 0.0735195010476529,
|
7904 |
+
"eval_loss": 3.029066801071167,
|
7905 |
+
"eval_runtime": 1548.6762,
|
7906 |
+
"eval_samples_per_second": 16.641,
|
7907 |
+
"eval_steps_per_second": 8.321,
|
7908 |
+
"step": 1125
|
7909 |
}
|
7910 |
],
|
7911 |
"logging_steps": 1,
|
|
|
7925 |
"attributes": {}
|
7926 |
}
|
7927 |
},
|
7928 |
+
"total_flos": 1.664607823331328e+18,
|
7929 |
"train_batch_size": 2,
|
7930 |
"trial_name": null,
|
7931 |
"trial_params": null
|