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 80013120
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4bbb33dd7312ba82afadbb41371ec5be3d881c71c1d2b60065ca3dc84f21e2a0
|
3 |
size 80013120
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 41120084
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:039a14cff66e4f6390c923d456c75465022ff6298ff6613b6e16234513eb7caa
|
3 |
size 41120084
|
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 sha256:71831c93a847d268aa853fc7a7696d5981f22052ee44d02cc36b70988787ff77
|
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": 17.584,
|
5274 |
"eval_steps_per_second": 8.794,
|
5275 |
"step": 750
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5276 |
}
|
5277 |
],
|
5278 |
"logging_steps": 1,
|
@@ -5292,7 +7925,7 @@
|
|
5292 |
"attributes": {}
|
5293 |
}
|
5294 |
},
|
5295 |
-
"total_flos":
|
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": 0.3327048907618942,
|
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": 17.584,
|
5274 |
"eval_steps_per_second": 8.794,
|
5275 |
"step": 750
|
5276 |
+
},
|
5277 |
+
{
|
5278 |
+
"epoch": 0.2220989981886067,
|
5279 |
+
"grad_norm": 0.115847647190094,
|
5280 |
+
"learning_rate": 0.00010084337058003303,
|
5281 |
+
"loss": 0.1349,
|
5282 |
+
"step": 751
|
5283 |
+
},
|
5284 |
+
{
|
5285 |
+
"epoch": 0.22239473586928393,
|
5286 |
+
"grad_norm": 0.13945455849170685,
|
5287 |
+
"learning_rate": 0.00010063253121564868,
|
5288 |
+
"loss": 0.1831,
|
5289 |
+
"step": 752
|
5290 |
+
},
|
5291 |
+
{
|
5292 |
+
"epoch": 0.2226904735499612,
|
5293 |
+
"grad_norm": 0.1549556851387024,
|
5294 |
+
"learning_rate": 0.00010042168903930514,
|
5295 |
+
"loss": 0.1844,
|
5296 |
+
"step": 753
|
5297 |
+
},
|
5298 |
+
{
|
5299 |
+
"epoch": 0.22298621123063841,
|
5300 |
+
"grad_norm": 0.1443348079919815,
|
5301 |
+
"learning_rate": 0.00010021084498831522,
|
5302 |
+
"loss": 0.1744,
|
5303 |
+
"step": 754
|
5304 |
+
},
|
5305 |
+
{
|
5306 |
+
"epoch": 0.22328194891131567,
|
5307 |
+
"grad_norm": 0.17714332044124603,
|
5308 |
+
"learning_rate": 0.0001,
|
5309 |
+
"loss": 0.2593,
|
5310 |
+
"step": 755
|
5311 |
+
},
|
5312 |
+
{
|
5313 |
+
"epoch": 0.2235776865919929,
|
5314 |
+
"grad_norm": 0.16148702800273895,
|
5315 |
+
"learning_rate": 9.97891550116848e-05,
|
5316 |
+
"loss": 0.2155,
|
5317 |
+
"step": 756
|
5318 |
+
},
|
5319 |
+
{
|
5320 |
+
"epoch": 0.22387342427267015,
|
5321 |
+
"grad_norm": 0.16650299727916718,
|
5322 |
+
"learning_rate": 9.957831096069488e-05,
|
5323 |
+
"loss": 0.2036,
|
5324 |
+
"step": 757
|
5325 |
+
},
|
5326 |
+
{
|
5327 |
+
"epoch": 0.22416916195334738,
|
5328 |
+
"grad_norm": 0.15004287660121918,
|
5329 |
+
"learning_rate": 9.936746878435136e-05,
|
5330 |
+
"loss": 0.1661,
|
5331 |
+
"step": 758
|
5332 |
+
},
|
5333 |
+
{
|
5334 |
+
"epoch": 0.22446489963402463,
|
5335 |
+
"grad_norm": 0.15887637436389923,
|
5336 |
+
"learning_rate": 9.915662941996699e-05,
|
5337 |
+
"loss": 0.1838,
|
5338 |
+
"step": 759
|
5339 |
+
},
|
5340 |
+
{
|
5341 |
+
"epoch": 0.22476063731470186,
|
5342 |
+
"grad_norm": 0.17734317481517792,
|
5343 |
+
"learning_rate": 9.894579380484204e-05,
|
5344 |
+
"loss": 0.2173,
|
5345 |
+
"step": 760
|
5346 |
+
},
|
5347 |
+
{
|
5348 |
+
"epoch": 0.2250563749953791,
|
5349 |
+
"grad_norm": 0.17889343202114105,
|
5350 |
+
"learning_rate": 9.873496287626019e-05,
|
5351 |
+
"loss": 0.1961,
|
5352 |
+
"step": 761
|
5353 |
+
},
|
5354 |
+
{
|
5355 |
+
"epoch": 0.22535211267605634,
|
5356 |
+
"grad_norm": 0.1788489818572998,
|
5357 |
+
"learning_rate": 9.852413757148417e-05,
|
5358 |
+
"loss": 0.2122,
|
5359 |
+
"step": 762
|
5360 |
+
},
|
5361 |
+
{
|
5362 |
+
"epoch": 0.22564785035673357,
|
5363 |
+
"grad_norm": 0.1814776510000229,
|
5364 |
+
"learning_rate": 9.831331882775178e-05,
|
5365 |
+
"loss": 0.2257,
|
5366 |
+
"step": 763
|
5367 |
+
},
|
5368 |
+
{
|
5369 |
+
"epoch": 0.22594358803741083,
|
5370 |
+
"grad_norm": 0.2010183185338974,
|
5371 |
+
"learning_rate": 9.81025075822716e-05,
|
5372 |
+
"loss": 0.2178,
|
5373 |
+
"step": 764
|
5374 |
+
},
|
5375 |
+
{
|
5376 |
+
"epoch": 0.22623932571808805,
|
5377 |
+
"grad_norm": 0.1668553501367569,
|
5378 |
+
"learning_rate": 9.789170477221891e-05,
|
5379 |
+
"loss": 0.1643,
|
5380 |
+
"step": 765
|
5381 |
+
},
|
5382 |
+
{
|
5383 |
+
"epoch": 0.2265350633987653,
|
5384 |
+
"grad_norm": 0.18353386223316193,
|
5385 |
+
"learning_rate": 9.76809113347315e-05,
|
5386 |
+
"loss": 0.191,
|
5387 |
+
"step": 766
|
5388 |
+
},
|
5389 |
+
{
|
5390 |
+
"epoch": 0.22683080107944253,
|
5391 |
+
"grad_norm": 0.19008232653141022,
|
5392 |
+
"learning_rate": 9.747012820690543e-05,
|
5393 |
+
"loss": 0.2259,
|
5394 |
+
"step": 767
|
5395 |
+
},
|
5396 |
+
{
|
5397 |
+
"epoch": 0.22712653876011976,
|
5398 |
+
"grad_norm": 0.1797637939453125,
|
5399 |
+
"learning_rate": 9.725935632579104e-05,
|
5400 |
+
"loss": 0.1842,
|
5401 |
+
"step": 768
|
5402 |
+
},
|
5403 |
+
{
|
5404 |
+
"epoch": 0.22742227644079702,
|
5405 |
+
"grad_norm": 0.1997433602809906,
|
5406 |
+
"learning_rate": 9.704859662838855e-05,
|
5407 |
+
"loss": 0.2003,
|
5408 |
+
"step": 769
|
5409 |
+
},
|
5410 |
+
{
|
5411 |
+
"epoch": 0.22771801412147424,
|
5412 |
+
"grad_norm": 0.17852815985679626,
|
5413 |
+
"learning_rate": 9.683785005164411e-05,
|
5414 |
+
"loss": 0.1665,
|
5415 |
+
"step": 770
|
5416 |
+
},
|
5417 |
+
{
|
5418 |
+
"epoch": 0.2280137518021515,
|
5419 |
+
"grad_norm": 0.19264471530914307,
|
5420 |
+
"learning_rate": 9.662711753244551e-05,
|
5421 |
+
"loss": 0.2015,
|
5422 |
+
"step": 771
|
5423 |
+
},
|
5424 |
+
{
|
5425 |
+
"epoch": 0.22830948948282873,
|
5426 |
+
"grad_norm": 0.20208404958248138,
|
5427 |
+
"learning_rate": 9.641640000761802e-05,
|
5428 |
+
"loss": 0.1836,
|
5429 |
+
"step": 772
|
5430 |
+
},
|
5431 |
+
{
|
5432 |
+
"epoch": 0.22860522716350598,
|
5433 |
+
"grad_norm": 0.20002898573875427,
|
5434 |
+
"learning_rate": 9.620569841392029e-05,
|
5435 |
+
"loss": 0.1956,
|
5436 |
+
"step": 773
|
5437 |
+
},
|
5438 |
+
{
|
5439 |
+
"epoch": 0.2289009648441832,
|
5440 |
+
"grad_norm": 0.20128233730793,
|
5441 |
+
"learning_rate": 9.59950136880401e-05,
|
5442 |
+
"loss": 0.218,
|
5443 |
+
"step": 774
|
5444 |
+
},
|
5445 |
+
{
|
5446 |
+
"epoch": 0.22919670252486044,
|
5447 |
+
"grad_norm": 0.1997966766357422,
|
5448 |
+
"learning_rate": 9.57843467665903e-05,
|
5449 |
+
"loss": 0.1771,
|
5450 |
+
"step": 775
|
5451 |
+
},
|
5452 |
+
{
|
5453 |
+
"epoch": 0.2294924402055377,
|
5454 |
+
"grad_norm": 0.22592507302761078,
|
5455 |
+
"learning_rate": 9.557369858610453e-05,
|
5456 |
+
"loss": 0.2222,
|
5457 |
+
"step": 776
|
5458 |
+
},
|
5459 |
+
{
|
5460 |
+
"epoch": 0.22978817788621492,
|
5461 |
+
"grad_norm": 0.18361607193946838,
|
5462 |
+
"learning_rate": 9.53630700830332e-05,
|
5463 |
+
"loss": 0.1476,
|
5464 |
+
"step": 777
|
5465 |
+
},
|
5466 |
+
{
|
5467 |
+
"epoch": 0.23008391556689217,
|
5468 |
+
"grad_norm": 0.17344044148921967,
|
5469 |
+
"learning_rate": 9.51524621937391e-05,
|
5470 |
+
"loss": 0.1298,
|
5471 |
+
"step": 778
|
5472 |
+
},
|
5473 |
+
{
|
5474 |
+
"epoch": 0.2303796532475694,
|
5475 |
+
"grad_norm": 0.20982904732227325,
|
5476 |
+
"learning_rate": 9.494187585449358e-05,
|
5477 |
+
"loss": 0.2031,
|
5478 |
+
"step": 779
|
5479 |
+
},
|
5480 |
+
{
|
5481 |
+
"epoch": 0.23067539092824665,
|
5482 |
+
"grad_norm": 0.20879507064819336,
|
5483 |
+
"learning_rate": 9.473131200147205e-05,
|
5484 |
+
"loss": 0.1622,
|
5485 |
+
"step": 780
|
5486 |
+
},
|
5487 |
+
{
|
5488 |
+
"epoch": 0.23097112860892388,
|
5489 |
+
"grad_norm": 0.24040478467941284,
|
5490 |
+
"learning_rate": 9.452077157074994e-05,
|
5491 |
+
"loss": 0.2284,
|
5492 |
+
"step": 781
|
5493 |
+
},
|
5494 |
+
{
|
5495 |
+
"epoch": 0.23126686628960114,
|
5496 |
+
"grad_norm": 0.2203570157289505,
|
5497 |
+
"learning_rate": 9.431025549829862e-05,
|
5498 |
+
"loss": 0.1829,
|
5499 |
+
"step": 782
|
5500 |
+
},
|
5501 |
+
{
|
5502 |
+
"epoch": 0.23156260397027836,
|
5503 |
+
"grad_norm": 0.2384481579065323,
|
5504 |
+
"learning_rate": 9.409976471998118e-05,
|
5505 |
+
"loss": 0.178,
|
5506 |
+
"step": 783
|
5507 |
+
},
|
5508 |
+
{
|
5509 |
+
"epoch": 0.2318583416509556,
|
5510 |
+
"grad_norm": 0.2443898469209671,
|
5511 |
+
"learning_rate": 9.388930017154819e-05,
|
5512 |
+
"loss": 0.169,
|
5513 |
+
"step": 784
|
5514 |
+
},
|
5515 |
+
{
|
5516 |
+
"epoch": 0.23215407933163285,
|
5517 |
+
"grad_norm": 0.27730923891067505,
|
5518 |
+
"learning_rate": 9.367886278863366e-05,
|
5519 |
+
"loss": 0.2722,
|
5520 |
+
"step": 785
|
5521 |
+
},
|
5522 |
+
{
|
5523 |
+
"epoch": 0.23244981701231007,
|
5524 |
+
"grad_norm": 0.2513563632965088,
|
5525 |
+
"learning_rate": 9.346845350675088e-05,
|
5526 |
+
"loss": 0.1804,
|
5527 |
+
"step": 786
|
5528 |
+
},
|
5529 |
+
{
|
5530 |
+
"epoch": 0.23274555469298733,
|
5531 |
+
"grad_norm": 0.25606629252433777,
|
5532 |
+
"learning_rate": 9.325807326128814e-05,
|
5533 |
+
"loss": 0.2188,
|
5534 |
+
"step": 787
|
5535 |
+
},
|
5536 |
+
{
|
5537 |
+
"epoch": 0.23304129237366455,
|
5538 |
+
"grad_norm": 0.24177873134613037,
|
5539 |
+
"learning_rate": 9.304772298750463e-05,
|
5540 |
+
"loss": 0.2054,
|
5541 |
+
"step": 788
|
5542 |
+
},
|
5543 |
+
{
|
5544 |
+
"epoch": 0.2333370300543418,
|
5545 |
+
"grad_norm": 0.23143790662288666,
|
5546 |
+
"learning_rate": 9.283740362052642e-05,
|
5547 |
+
"loss": 0.1496,
|
5548 |
+
"step": 789
|
5549 |
+
},
|
5550 |
+
{
|
5551 |
+
"epoch": 0.23363276773501904,
|
5552 |
+
"grad_norm": 0.25107666850090027,
|
5553 |
+
"learning_rate": 9.26271160953421e-05,
|
5554 |
+
"loss": 0.1276,
|
5555 |
+
"step": 790
|
5556 |
+
},
|
5557 |
+
{
|
5558 |
+
"epoch": 0.23392850541569626,
|
5559 |
+
"grad_norm": 0.25497153401374817,
|
5560 |
+
"learning_rate": 9.241686134679867e-05,
|
5561 |
+
"loss": 0.1567,
|
5562 |
+
"step": 791
|
5563 |
+
},
|
5564 |
+
{
|
5565 |
+
"epoch": 0.23422424309637352,
|
5566 |
+
"grad_norm": 0.3068770170211792,
|
5567 |
+
"learning_rate": 9.220664030959749e-05,
|
5568 |
+
"loss": 0.1859,
|
5569 |
+
"step": 792
|
5570 |
+
},
|
5571 |
+
{
|
5572 |
+
"epoch": 0.23451998077705075,
|
5573 |
+
"grad_norm": 0.2575652301311493,
|
5574 |
+
"learning_rate": 9.199645391828999e-05,
|
5575 |
+
"loss": 0.1626,
|
5576 |
+
"step": 793
|
5577 |
+
},
|
5578 |
+
{
|
5579 |
+
"epoch": 0.234815718457728,
|
5580 |
+
"grad_norm": 0.2657597064971924,
|
5581 |
+
"learning_rate": 9.178630310727365e-05,
|
5582 |
+
"loss": 0.1557,
|
5583 |
+
"step": 794
|
5584 |
+
},
|
5585 |
+
{
|
5586 |
+
"epoch": 0.23511145613840523,
|
5587 |
+
"grad_norm": 0.3128422200679779,
|
5588 |
+
"learning_rate": 9.157618881078772e-05,
|
5589 |
+
"loss": 0.2013,
|
5590 |
+
"step": 795
|
5591 |
+
},
|
5592 |
+
{
|
5593 |
+
"epoch": 0.23540719381908248,
|
5594 |
+
"grad_norm": 0.280213326215744,
|
5595 |
+
"learning_rate": 9.136611196290915e-05,
|
5596 |
+
"loss": 0.1513,
|
5597 |
+
"step": 796
|
5598 |
+
},
|
5599 |
+
{
|
5600 |
+
"epoch": 0.2357029314997597,
|
5601 |
+
"grad_norm": 0.32494768500328064,
|
5602 |
+
"learning_rate": 9.115607349754834e-05,
|
5603 |
+
"loss": 0.164,
|
5604 |
+
"step": 797
|
5605 |
+
},
|
5606 |
+
{
|
5607 |
+
"epoch": 0.23599866918043697,
|
5608 |
+
"grad_norm": 0.31203532218933105,
|
5609 |
+
"learning_rate": 9.094607434844523e-05,
|
5610 |
+
"loss": 0.1492,
|
5611 |
+
"step": 798
|
5612 |
+
},
|
5613 |
+
{
|
5614 |
+
"epoch": 0.2362944068611142,
|
5615 |
+
"grad_norm": 0.44517087936401367,
|
5616 |
+
"learning_rate": 9.07361154491648e-05,
|
5617 |
+
"loss": 0.2025,
|
5618 |
+
"step": 799
|
5619 |
+
},
|
5620 |
+
{
|
5621 |
+
"epoch": 0.23659014454179142,
|
5622 |
+
"grad_norm": 0.5797000527381897,
|
5623 |
+
"learning_rate": 9.052619773309317e-05,
|
5624 |
+
"loss": 0.2612,
|
5625 |
+
"step": 800
|
5626 |
+
},
|
5627 |
+
{
|
5628 |
+
"epoch": 0.23688588222246867,
|
5629 |
+
"grad_norm": 0.11432012170553207,
|
5630 |
+
"learning_rate": 9.031632213343339e-05,
|
5631 |
+
"loss": 0.1212,
|
5632 |
+
"step": 801
|
5633 |
+
},
|
5634 |
+
{
|
5635 |
+
"epoch": 0.2371816199031459,
|
5636 |
+
"grad_norm": 0.15115061402320862,
|
5637 |
+
"learning_rate": 9.01064895832012e-05,
|
5638 |
+
"loss": 0.1791,
|
5639 |
+
"step": 802
|
5640 |
+
},
|
5641 |
+
{
|
5642 |
+
"epoch": 0.23747735758382316,
|
5643 |
+
"grad_norm": 0.1600162535905838,
|
5644 |
+
"learning_rate": 8.98967010152211e-05,
|
5645 |
+
"loss": 0.1696,
|
5646 |
+
"step": 803
|
5647 |
+
},
|
5648 |
+
{
|
5649 |
+
"epoch": 0.23777309526450038,
|
5650 |
+
"grad_norm": 0.1539337933063507,
|
5651 |
+
"learning_rate": 8.968695736212193e-05,
|
5652 |
+
"loss": 0.1813,
|
5653 |
+
"step": 804
|
5654 |
+
},
|
5655 |
+
{
|
5656 |
+
"epoch": 0.23806883294517764,
|
5657 |
+
"grad_norm": 0.15929153561592102,
|
5658 |
+
"learning_rate": 8.947725955633294e-05,
|
5659 |
+
"loss": 0.1551,
|
5660 |
+
"step": 805
|
5661 |
+
},
|
5662 |
+
{
|
5663 |
+
"epoch": 0.23836457062585487,
|
5664 |
+
"grad_norm": 0.14739802479743958,
|
5665 |
+
"learning_rate": 8.926760853007946e-05,
|
5666 |
+
"loss": 0.1555,
|
5667 |
+
"step": 806
|
5668 |
+
},
|
5669 |
+
{
|
5670 |
+
"epoch": 0.2386603083065321,
|
5671 |
+
"grad_norm": 0.16396670043468475,
|
5672 |
+
"learning_rate": 8.905800521537905e-05,
|
5673 |
+
"loss": 0.1776,
|
5674 |
+
"step": 807
|
5675 |
+
},
|
5676 |
+
{
|
5677 |
+
"epoch": 0.23895604598720935,
|
5678 |
+
"grad_norm": 0.16819187998771667,
|
5679 |
+
"learning_rate": 8.884845054403699e-05,
|
5680 |
+
"loss": 0.1925,
|
5681 |
+
"step": 808
|
5682 |
+
},
|
5683 |
+
{
|
5684 |
+
"epoch": 0.23925178366788658,
|
5685 |
+
"grad_norm": 0.17342273890972137,
|
5686 |
+
"learning_rate": 8.863894544764236e-05,
|
5687 |
+
"loss": 0.183,
|
5688 |
+
"step": 809
|
5689 |
+
},
|
5690 |
+
{
|
5691 |
+
"epoch": 0.23954752134856383,
|
5692 |
+
"grad_norm": 0.18809068202972412,
|
5693 |
+
"learning_rate": 8.84294908575639e-05,
|
5694 |
+
"loss": 0.2289,
|
5695 |
+
"step": 810
|
5696 |
+
},
|
5697 |
+
{
|
5698 |
+
"epoch": 0.23984325902924106,
|
5699 |
+
"grad_norm": 0.18252570927143097,
|
5700 |
+
"learning_rate": 8.822008770494572e-05,
|
5701 |
+
"loss": 0.2011,
|
5702 |
+
"step": 811
|
5703 |
+
},
|
5704 |
+
{
|
5705 |
+
"epoch": 0.2401389967099183,
|
5706 |
+
"grad_norm": 0.1622556895017624,
|
5707 |
+
"learning_rate": 8.801073692070337e-05,
|
5708 |
+
"loss": 0.1416,
|
5709 |
+
"step": 812
|
5710 |
+
},
|
5711 |
+
{
|
5712 |
+
"epoch": 0.24043473439059554,
|
5713 |
+
"grad_norm": 0.17876408994197845,
|
5714 |
+
"learning_rate": 8.780143943551954e-05,
|
5715 |
+
"loss": 0.2014,
|
5716 |
+
"step": 813
|
5717 |
+
},
|
5718 |
+
{
|
5719 |
+
"epoch": 0.2407304720712728,
|
5720 |
+
"grad_norm": 0.18106475472450256,
|
5721 |
+
"learning_rate": 8.759219617983999e-05,
|
5722 |
+
"loss": 0.2002,
|
5723 |
+
"step": 814
|
5724 |
+
},
|
5725 |
+
{
|
5726 |
+
"epoch": 0.24102620975195002,
|
5727 |
+
"grad_norm": 0.20041576027870178,
|
5728 |
+
"learning_rate": 8.738300808386935e-05,
|
5729 |
+
"loss": 0.2289,
|
5730 |
+
"step": 815
|
5731 |
+
},
|
5732 |
+
{
|
5733 |
+
"epoch": 0.24132194743262725,
|
5734 |
+
"grad_norm": 0.17808635532855988,
|
5735 |
+
"learning_rate": 8.717387607756713e-05,
|
5736 |
+
"loss": 0.1851,
|
5737 |
+
"step": 816
|
5738 |
+
},
|
5739 |
+
{
|
5740 |
+
"epoch": 0.2416176851133045,
|
5741 |
+
"grad_norm": 0.22498534619808197,
|
5742 |
+
"learning_rate": 8.696480109064342e-05,
|
5743 |
+
"loss": 0.2488,
|
5744 |
+
"step": 817
|
5745 |
+
},
|
5746 |
+
{
|
5747 |
+
"epoch": 0.24191342279398173,
|
5748 |
+
"grad_norm": 0.19668036699295044,
|
5749 |
+
"learning_rate": 8.675578405255485e-05,
|
5750 |
+
"loss": 0.1995,
|
5751 |
+
"step": 818
|
5752 |
+
},
|
5753 |
+
{
|
5754 |
+
"epoch": 0.24220916047465899,
|
5755 |
+
"grad_norm": 0.19187040627002716,
|
5756 |
+
"learning_rate": 8.654682589250038e-05,
|
5757 |
+
"loss": 0.1759,
|
5758 |
+
"step": 819
|
5759 |
+
},
|
5760 |
+
{
|
5761 |
+
"epoch": 0.2425048981553362,
|
5762 |
+
"grad_norm": 0.208050936460495,
|
5763 |
+
"learning_rate": 8.633792753941733e-05,
|
5764 |
+
"loss": 0.2253,
|
5765 |
+
"step": 820
|
5766 |
+
},
|
5767 |
+
{
|
5768 |
+
"epoch": 0.24280063583601347,
|
5769 |
+
"grad_norm": 0.18446066975593567,
|
5770 |
+
"learning_rate": 8.612908992197705e-05,
|
5771 |
+
"loss": 0.1743,
|
5772 |
+
"step": 821
|
5773 |
+
},
|
5774 |
+
{
|
5775 |
+
"epoch": 0.2430963735166907,
|
5776 |
+
"grad_norm": 0.19328705966472626,
|
5777 |
+
"learning_rate": 8.592031396858093e-05,
|
5778 |
+
"loss": 0.1572,
|
5779 |
+
"step": 822
|
5780 |
+
},
|
5781 |
+
{
|
5782 |
+
"epoch": 0.24339211119736792,
|
5783 |
+
"grad_norm": 0.21649695932865143,
|
5784 |
+
"learning_rate": 8.571160060735624e-05,
|
5785 |
+
"loss": 0.2124,
|
5786 |
+
"step": 823
|
5787 |
+
},
|
5788 |
+
{
|
5789 |
+
"epoch": 0.24368784887804518,
|
5790 |
+
"grad_norm": 0.1915416270494461,
|
5791 |
+
"learning_rate": 8.550295076615188e-05,
|
5792 |
+
"loss": 0.169,
|
5793 |
+
"step": 824
|
5794 |
+
},
|
5795 |
+
{
|
5796 |
+
"epoch": 0.2439835865587224,
|
5797 |
+
"grad_norm": 0.20172414183616638,
|
5798 |
+
"learning_rate": 8.529436537253458e-05,
|
5799 |
+
"loss": 0.1644,
|
5800 |
+
"step": 825
|
5801 |
+
},
|
5802 |
+
{
|
5803 |
+
"epoch": 0.24427932423939966,
|
5804 |
+
"grad_norm": 0.19948998093605042,
|
5805 |
+
"learning_rate": 8.508584535378439e-05,
|
5806 |
+
"loss": 0.2082,
|
5807 |
+
"step": 826
|
5808 |
+
},
|
5809 |
+
{
|
5810 |
+
"epoch": 0.24457506192007689,
|
5811 |
+
"grad_norm": 0.22914430499076843,
|
5812 |
+
"learning_rate": 8.487739163689079e-05,
|
5813 |
+
"loss": 0.205,
|
5814 |
+
"step": 827
|
5815 |
+
},
|
5816 |
+
{
|
5817 |
+
"epoch": 0.24487079960075414,
|
5818 |
+
"grad_norm": 0.21713916957378387,
|
5819 |
+
"learning_rate": 8.466900514854847e-05,
|
5820 |
+
"loss": 0.2202,
|
5821 |
+
"step": 828
|
5822 |
+
},
|
5823 |
+
{
|
5824 |
+
"epoch": 0.24516653728143137,
|
5825 |
+
"grad_norm": 0.2027435600757599,
|
5826 |
+
"learning_rate": 8.446068681515334e-05,
|
5827 |
+
"loss": 0.1789,
|
5828 |
+
"step": 829
|
5829 |
+
},
|
5830 |
+
{
|
5831 |
+
"epoch": 0.24546227496210862,
|
5832 |
+
"grad_norm": 0.24722711741924286,
|
5833 |
+
"learning_rate": 8.425243756279824e-05,
|
5834 |
+
"loss": 0.1915,
|
5835 |
+
"step": 830
|
5836 |
+
},
|
5837 |
+
{
|
5838 |
+
"epoch": 0.24575801264278585,
|
5839 |
+
"grad_norm": 0.2104775607585907,
|
5840 |
+
"learning_rate": 8.404425831726894e-05,
|
5841 |
+
"loss": 0.1414,
|
5842 |
+
"step": 831
|
5843 |
+
},
|
5844 |
+
{
|
5845 |
+
"epoch": 0.24605375032346308,
|
5846 |
+
"grad_norm": 0.22853389382362366,
|
5847 |
+
"learning_rate": 8.383615000404e-05,
|
5848 |
+
"loss": 0.1666,
|
5849 |
+
"step": 832
|
5850 |
+
},
|
5851 |
+
{
|
5852 |
+
"epoch": 0.24634948800414033,
|
5853 |
+
"grad_norm": 0.22614425420761108,
|
5854 |
+
"learning_rate": 8.362811354827059e-05,
|
5855 |
+
"loss": 0.1459,
|
5856 |
+
"step": 833
|
5857 |
+
},
|
5858 |
+
{
|
5859 |
+
"epoch": 0.24664522568481756,
|
5860 |
+
"grad_norm": 0.23260444402694702,
|
5861 |
+
"learning_rate": 8.342014987480047e-05,
|
5862 |
+
"loss": 0.1829,
|
5863 |
+
"step": 834
|
5864 |
+
},
|
5865 |
+
{
|
5866 |
+
"epoch": 0.24694096336549481,
|
5867 |
+
"grad_norm": 0.27205580472946167,
|
5868 |
+
"learning_rate": 8.321225990814591e-05,
|
5869 |
+
"loss": 0.1895,
|
5870 |
+
"step": 835
|
5871 |
+
},
|
5872 |
+
{
|
5873 |
+
"epoch": 0.24723670104617204,
|
5874 |
+
"grad_norm": 0.27459388971328735,
|
5875 |
+
"learning_rate": 8.300444457249543e-05,
|
5876 |
+
"loss": 0.222,
|
5877 |
+
"step": 836
|
5878 |
+
},
|
5879 |
+
{
|
5880 |
+
"epoch": 0.2475324387268493,
|
5881 |
+
"grad_norm": 0.25127366185188293,
|
5882 |
+
"learning_rate": 8.279670479170573e-05,
|
5883 |
+
"loss": 0.1232,
|
5884 |
+
"step": 837
|
5885 |
+
},
|
5886 |
+
{
|
5887 |
+
"epoch": 0.24782817640752652,
|
5888 |
+
"grad_norm": 0.244548961520195,
|
5889 |
+
"learning_rate": 8.258904148929775e-05,
|
5890 |
+
"loss": 0.2147,
|
5891 |
+
"step": 838
|
5892 |
+
},
|
5893 |
+
{
|
5894 |
+
"epoch": 0.24812391408820375,
|
5895 |
+
"grad_norm": 0.24851387739181519,
|
5896 |
+
"learning_rate": 8.238145558845235e-05,
|
5897 |
+
"loss": 0.2079,
|
5898 |
+
"step": 839
|
5899 |
+
},
|
5900 |
+
{
|
5901 |
+
"epoch": 0.248419651768881,
|
5902 |
+
"grad_norm": 0.29407623410224915,
|
5903 |
+
"learning_rate": 8.217394801200631e-05,
|
5904 |
+
"loss": 0.2305,
|
5905 |
+
"step": 840
|
5906 |
+
},
|
5907 |
+
{
|
5908 |
+
"epoch": 0.24871538944955823,
|
5909 |
+
"grad_norm": 0.2977339029312134,
|
5910 |
+
"learning_rate": 8.196651968244826e-05,
|
5911 |
+
"loss": 0.2305,
|
5912 |
+
"step": 841
|
5913 |
+
},
|
5914 |
+
{
|
5915 |
+
"epoch": 0.2490111271302355,
|
5916 |
+
"grad_norm": 0.3162849247455597,
|
5917 |
+
"learning_rate": 8.175917152191447e-05,
|
5918 |
+
"loss": 0.2099,
|
5919 |
+
"step": 842
|
5920 |
+
},
|
5921 |
+
{
|
5922 |
+
"epoch": 0.24930686481091272,
|
5923 |
+
"grad_norm": 0.315044105052948,
|
5924 |
+
"learning_rate": 8.15519044521848e-05,
|
5925 |
+
"loss": 0.2464,
|
5926 |
+
"step": 843
|
5927 |
+
},
|
5928 |
+
{
|
5929 |
+
"epoch": 0.24960260249158997,
|
5930 |
+
"grad_norm": 0.23404821753501892,
|
5931 |
+
"learning_rate": 8.134471939467874e-05,
|
5932 |
+
"loss": 0.1401,
|
5933 |
+
"step": 844
|
5934 |
+
},
|
5935 |
+
{
|
5936 |
+
"epoch": 0.2498983401722672,
|
5937 |
+
"grad_norm": 0.28799015283584595,
|
5938 |
+
"learning_rate": 8.113761727045105e-05,
|
5939 |
+
"loss": 0.1694,
|
5940 |
+
"step": 845
|
5941 |
+
},
|
5942 |
+
{
|
5943 |
+
"epoch": 0.25019407785294445,
|
5944 |
+
"grad_norm": 0.3503269553184509,
|
5945 |
+
"learning_rate": 8.093059900018792e-05,
|
5946 |
+
"loss": 0.2245,
|
5947 |
+
"step": 846
|
5948 |
+
},
|
5949 |
+
{
|
5950 |
+
"epoch": 0.25048981553362165,
|
5951 |
+
"grad_norm": 0.3673860430717468,
|
5952 |
+
"learning_rate": 8.072366550420266e-05,
|
5953 |
+
"loss": 0.224,
|
5954 |
+
"step": 847
|
5955 |
+
},
|
5956 |
+
{
|
5957 |
+
"epoch": 0.2507855532142989,
|
5958 |
+
"grad_norm": 0.4725538492202759,
|
5959 |
+
"learning_rate": 8.051681770243175e-05,
|
5960 |
+
"loss": 0.1999,
|
5961 |
+
"step": 848
|
5962 |
+
},
|
5963 |
+
{
|
5964 |
+
"epoch": 0.25108129089497616,
|
5965 |
+
"grad_norm": 0.454391211271286,
|
5966 |
+
"learning_rate": 8.031005651443073e-05,
|
5967 |
+
"loss": 0.2453,
|
5968 |
+
"step": 849
|
5969 |
+
},
|
5970 |
+
{
|
5971 |
+
"epoch": 0.2513770285756534,
|
5972 |
+
"grad_norm": 0.6510867476463318,
|
5973 |
+
"learning_rate": 8.010338285937006e-05,
|
5974 |
+
"loss": 0.3214,
|
5975 |
+
"step": 850
|
5976 |
+
},
|
5977 |
+
{
|
5978 |
+
"epoch": 0.2516727662563306,
|
5979 |
+
"grad_norm": 0.1124342530965805,
|
5980 |
+
"learning_rate": 7.989679765603108e-05,
|
5981 |
+
"loss": 0.1517,
|
5982 |
+
"step": 851
|
5983 |
+
},
|
5984 |
+
{
|
5985 |
+
"epoch": 0.25196850393700787,
|
5986 |
+
"grad_norm": 0.1325543373823166,
|
5987 |
+
"learning_rate": 7.969030182280192e-05,
|
5988 |
+
"loss": 0.177,
|
5989 |
+
"step": 852
|
5990 |
+
},
|
5991 |
+
{
|
5992 |
+
"epoch": 0.2522642416176851,
|
5993 |
+
"grad_norm": 0.14178207516670227,
|
5994 |
+
"learning_rate": 7.948389627767343e-05,
|
5995 |
+
"loss": 0.1601,
|
5996 |
+
"step": 853
|
5997 |
+
},
|
5998 |
+
{
|
5999 |
+
"epoch": 0.2525599792983623,
|
6000 |
+
"grad_norm": 0.15501843392848969,
|
6001 |
+
"learning_rate": 7.927758193823501e-05,
|
6002 |
+
"loss": 0.161,
|
6003 |
+
"step": 854
|
6004 |
+
},
|
6005 |
+
{
|
6006 |
+
"epoch": 0.2528557169790396,
|
6007 |
+
"grad_norm": 0.15361559391021729,
|
6008 |
+
"learning_rate": 7.907135972167069e-05,
|
6009 |
+
"loss": 0.1775,
|
6010 |
+
"step": 855
|
6011 |
+
},
|
6012 |
+
{
|
6013 |
+
"epoch": 0.25315145465971683,
|
6014 |
+
"grad_norm": 0.1672465205192566,
|
6015 |
+
"learning_rate": 7.88652305447549e-05,
|
6016 |
+
"loss": 0.1796,
|
6017 |
+
"step": 856
|
6018 |
+
},
|
6019 |
+
{
|
6020 |
+
"epoch": 0.2534471923403941,
|
6021 |
+
"grad_norm": 0.1553036868572235,
|
6022 |
+
"learning_rate": 7.865919532384844e-05,
|
6023 |
+
"loss": 0.1647,
|
6024 |
+
"step": 857
|
6025 |
+
},
|
6026 |
+
{
|
6027 |
+
"epoch": 0.2537429300210713,
|
6028 |
+
"grad_norm": 0.16316907107830048,
|
6029 |
+
"learning_rate": 7.845325497489449e-05,
|
6030 |
+
"loss": 0.1431,
|
6031 |
+
"step": 858
|
6032 |
+
},
|
6033 |
+
{
|
6034 |
+
"epoch": 0.25403866770174854,
|
6035 |
+
"grad_norm": 0.1651161015033722,
|
6036 |
+
"learning_rate": 7.82474104134144e-05,
|
6037 |
+
"loss": 0.1587,
|
6038 |
+
"step": 859
|
6039 |
+
},
|
6040 |
+
{
|
6041 |
+
"epoch": 0.2543344053824258,
|
6042 |
+
"grad_norm": 0.18129472434520721,
|
6043 |
+
"learning_rate": 7.804166255450373e-05,
|
6044 |
+
"loss": 0.1635,
|
6045 |
+
"step": 860
|
6046 |
+
},
|
6047 |
+
{
|
6048 |
+
"epoch": 0.25463014306310305,
|
6049 |
+
"grad_norm": 0.18252769112586975,
|
6050 |
+
"learning_rate": 7.783601231282812e-05,
|
6051 |
+
"loss": 0.2157,
|
6052 |
+
"step": 861
|
6053 |
+
},
|
6054 |
+
{
|
6055 |
+
"epoch": 0.25492588074378025,
|
6056 |
+
"grad_norm": 0.17854921519756317,
|
6057 |
+
"learning_rate": 7.763046060261932e-05,
|
6058 |
+
"loss": 0.1804,
|
6059 |
+
"step": 862
|
6060 |
+
},
|
6061 |
+
{
|
6062 |
+
"epoch": 0.2552216184244575,
|
6063 |
+
"grad_norm": 0.1713121086359024,
|
6064 |
+
"learning_rate": 7.742500833767094e-05,
|
6065 |
+
"loss": 0.1704,
|
6066 |
+
"step": 863
|
6067 |
+
},
|
6068 |
+
{
|
6069 |
+
"epoch": 0.25551735610513476,
|
6070 |
+
"grad_norm": 0.18181180953979492,
|
6071 |
+
"learning_rate": 7.721965643133458e-05,
|
6072 |
+
"loss": 0.192,
|
6073 |
+
"step": 864
|
6074 |
+
},
|
6075 |
+
{
|
6076 |
+
"epoch": 0.25581309378581196,
|
6077 |
+
"grad_norm": 0.18746164441108704,
|
6078 |
+
"learning_rate": 7.701440579651566e-05,
|
6079 |
+
"loss": 0.2068,
|
6080 |
+
"step": 865
|
6081 |
+
},
|
6082 |
+
{
|
6083 |
+
"epoch": 0.2561088314664892,
|
6084 |
+
"grad_norm": 0.19459117949008942,
|
6085 |
+
"learning_rate": 7.680925734566937e-05,
|
6086 |
+
"loss": 0.1758,
|
6087 |
+
"step": 866
|
6088 |
+
},
|
6089 |
+
{
|
6090 |
+
"epoch": 0.2564045691471665,
|
6091 |
+
"grad_norm": 0.18460817635059357,
|
6092 |
+
"learning_rate": 7.660421199079669e-05,
|
6093 |
+
"loss": 0.1814,
|
6094 |
+
"step": 867
|
6095 |
+
},
|
6096 |
+
{
|
6097 |
+
"epoch": 0.2567003068278437,
|
6098 |
+
"grad_norm": 0.20752295851707458,
|
6099 |
+
"learning_rate": 7.639927064344022e-05,
|
6100 |
+
"loss": 0.2329,
|
6101 |
+
"step": 868
|
6102 |
+
},
|
6103 |
+
{
|
6104 |
+
"epoch": 0.2569960445085209,
|
6105 |
+
"grad_norm": 0.2058822363615036,
|
6106 |
+
"learning_rate": 7.619443421468021e-05,
|
6107 |
+
"loss": 0.2083,
|
6108 |
+
"step": 869
|
6109 |
+
},
|
6110 |
+
{
|
6111 |
+
"epoch": 0.2572917821891982,
|
6112 |
+
"grad_norm": 0.2240809202194214,
|
6113 |
+
"learning_rate": 7.598970361513051e-05,
|
6114 |
+
"loss": 0.2273,
|
6115 |
+
"step": 870
|
6116 |
+
},
|
6117 |
+
{
|
6118 |
+
"epoch": 0.25758751986987544,
|
6119 |
+
"grad_norm": 0.20064809918403625,
|
6120 |
+
"learning_rate": 7.578507975493448e-05,
|
6121 |
+
"loss": 0.1672,
|
6122 |
+
"step": 871
|
6123 |
+
},
|
6124 |
+
{
|
6125 |
+
"epoch": 0.25788325755055264,
|
6126 |
+
"grad_norm": 0.2004321813583374,
|
6127 |
+
"learning_rate": 7.558056354376098e-05,
|
6128 |
+
"loss": 0.1797,
|
6129 |
+
"step": 872
|
6130 |
+
},
|
6131 |
+
{
|
6132 |
+
"epoch": 0.2581789952312299,
|
6133 |
+
"grad_norm": 0.21047692000865936,
|
6134 |
+
"learning_rate": 7.537615589080027e-05,
|
6135 |
+
"loss": 0.1897,
|
6136 |
+
"step": 873
|
6137 |
+
},
|
6138 |
+
{
|
6139 |
+
"epoch": 0.25847473291190715,
|
6140 |
+
"grad_norm": 0.2111281305551529,
|
6141 |
+
"learning_rate": 7.517185770476006e-05,
|
6142 |
+
"loss": 0.1866,
|
6143 |
+
"step": 874
|
6144 |
+
},
|
6145 |
+
{
|
6146 |
+
"epoch": 0.2587704705925844,
|
6147 |
+
"grad_norm": 0.208242729306221,
|
6148 |
+
"learning_rate": 7.496766989386136e-05,
|
6149 |
+
"loss": 0.2176,
|
6150 |
+
"step": 875
|
6151 |
+
},
|
6152 |
+
{
|
6153 |
+
"epoch": 0.2590662082732616,
|
6154 |
+
"grad_norm": 0.24142082035541534,
|
6155 |
+
"learning_rate": 7.476359336583454e-05,
|
6156 |
+
"loss": 0.2557,
|
6157 |
+
"step": 876
|
6158 |
+
},
|
6159 |
+
{
|
6160 |
+
"epoch": 0.25936194595393885,
|
6161 |
+
"grad_norm": 0.19397041201591492,
|
6162 |
+
"learning_rate": 7.455962902791522e-05,
|
6163 |
+
"loss": 0.1439,
|
6164 |
+
"step": 877
|
6165 |
+
},
|
6166 |
+
{
|
6167 |
+
"epoch": 0.2596576836346161,
|
6168 |
+
"grad_norm": 0.24474111199378967,
|
6169 |
+
"learning_rate": 7.435577778684033e-05,
|
6170 |
+
"loss": 0.2145,
|
6171 |
+
"step": 878
|
6172 |
+
},
|
6173 |
+
{
|
6174 |
+
"epoch": 0.2599534213152933,
|
6175 |
+
"grad_norm": 0.2113002985715866,
|
6176 |
+
"learning_rate": 7.415204054884399e-05,
|
6177 |
+
"loss": 0.1806,
|
6178 |
+
"step": 879
|
6179 |
+
},
|
6180 |
+
{
|
6181 |
+
"epoch": 0.26024915899597056,
|
6182 |
+
"grad_norm": 0.23416808247566223,
|
6183 |
+
"learning_rate": 7.394841821965345e-05,
|
6184 |
+
"loss": 0.2005,
|
6185 |
+
"step": 880
|
6186 |
+
},
|
6187 |
+
{
|
6188 |
+
"epoch": 0.2605448966766478,
|
6189 |
+
"grad_norm": 0.21154069900512695,
|
6190 |
+
"learning_rate": 7.374491170448525e-05,
|
6191 |
+
"loss": 0.1623,
|
6192 |
+
"step": 881
|
6193 |
+
},
|
6194 |
+
{
|
6195 |
+
"epoch": 0.2608406343573251,
|
6196 |
+
"grad_norm": 0.20221436023712158,
|
6197 |
+
"learning_rate": 7.3541521908041e-05,
|
6198 |
+
"loss": 0.1428,
|
6199 |
+
"step": 882
|
6200 |
+
},
|
6201 |
+
{
|
6202 |
+
"epoch": 0.2611363720380023,
|
6203 |
+
"grad_norm": 0.24770595133304596,
|
6204 |
+
"learning_rate": 7.33382497345034e-05,
|
6205 |
+
"loss": 0.1745,
|
6206 |
+
"step": 883
|
6207 |
+
},
|
6208 |
+
{
|
6209 |
+
"epoch": 0.26143210971867953,
|
6210 |
+
"grad_norm": 0.21654701232910156,
|
6211 |
+
"learning_rate": 7.313509608753231e-05,
|
6212 |
+
"loss": 0.1651,
|
6213 |
+
"step": 884
|
6214 |
+
},
|
6215 |
+
{
|
6216 |
+
"epoch": 0.2617278473993568,
|
6217 |
+
"grad_norm": 0.21652589738368988,
|
6218 |
+
"learning_rate": 7.293206187026061e-05,
|
6219 |
+
"loss": 0.1332,
|
6220 |
+
"step": 885
|
6221 |
+
},
|
6222 |
+
{
|
6223 |
+
"epoch": 0.262023585080034,
|
6224 |
+
"grad_norm": 0.29497137665748596,
|
6225 |
+
"learning_rate": 7.27291479852903e-05,
|
6226 |
+
"loss": 0.1639,
|
6227 |
+
"step": 886
|
6228 |
+
},
|
6229 |
+
{
|
6230 |
+
"epoch": 0.26231932276071124,
|
6231 |
+
"grad_norm": 0.2427954226732254,
|
6232 |
+
"learning_rate": 7.252635533468843e-05,
|
6233 |
+
"loss": 0.1723,
|
6234 |
+
"step": 887
|
6235 |
+
},
|
6236 |
+
{
|
6237 |
+
"epoch": 0.2626150604413885,
|
6238 |
+
"grad_norm": 0.2615757882595062,
|
6239 |
+
"learning_rate": 7.232368481998309e-05,
|
6240 |
+
"loss": 0.1891,
|
6241 |
+
"step": 888
|
6242 |
+
},
|
6243 |
+
{
|
6244 |
+
"epoch": 0.26291079812206575,
|
6245 |
+
"grad_norm": 0.21479550004005432,
|
6246 |
+
"learning_rate": 7.212113734215932e-05,
|
6247 |
+
"loss": 0.1295,
|
6248 |
+
"step": 889
|
6249 |
+
},
|
6250 |
+
{
|
6251 |
+
"epoch": 0.26320653580274295,
|
6252 |
+
"grad_norm": 0.29835671186447144,
|
6253 |
+
"learning_rate": 7.191871380165538e-05,
|
6254 |
+
"loss": 0.2239,
|
6255 |
+
"step": 890
|
6256 |
+
},
|
6257 |
+
{
|
6258 |
+
"epoch": 0.2635022734834202,
|
6259 |
+
"grad_norm": 0.304277241230011,
|
6260 |
+
"learning_rate": 7.17164150983584e-05,
|
6261 |
+
"loss": 0.1899,
|
6262 |
+
"step": 891
|
6263 |
+
},
|
6264 |
+
{
|
6265 |
+
"epoch": 0.26379801116409746,
|
6266 |
+
"grad_norm": 0.28030019998550415,
|
6267 |
+
"learning_rate": 7.151424213160061e-05,
|
6268 |
+
"loss": 0.1829,
|
6269 |
+
"step": 892
|
6270 |
+
},
|
6271 |
+
{
|
6272 |
+
"epoch": 0.2640937488447747,
|
6273 |
+
"grad_norm": 0.26866456866264343,
|
6274 |
+
"learning_rate": 7.131219580015521e-05,
|
6275 |
+
"loss": 0.1586,
|
6276 |
+
"step": 893
|
6277 |
+
},
|
6278 |
+
{
|
6279 |
+
"epoch": 0.2643894865254519,
|
6280 |
+
"grad_norm": 0.2647477984428406,
|
6281 |
+
"learning_rate": 7.11102770022325e-05,
|
6282 |
+
"loss": 0.1242,
|
6283 |
+
"step": 894
|
6284 |
+
},
|
6285 |
+
{
|
6286 |
+
"epoch": 0.26468522420612917,
|
6287 |
+
"grad_norm": 0.27126333117485046,
|
6288 |
+
"learning_rate": 7.090848663547574e-05,
|
6289 |
+
"loss": 0.1559,
|
6290 |
+
"step": 895
|
6291 |
+
},
|
6292 |
+
{
|
6293 |
+
"epoch": 0.2649809618868064,
|
6294 |
+
"grad_norm": 0.2967953681945801,
|
6295 |
+
"learning_rate": 7.070682559695736e-05,
|
6296 |
+
"loss": 0.1913,
|
6297 |
+
"step": 896
|
6298 |
+
},
|
6299 |
+
{
|
6300 |
+
"epoch": 0.2652766995674836,
|
6301 |
+
"grad_norm": 0.3403211236000061,
|
6302 |
+
"learning_rate": 7.050529478317476e-05,
|
6303 |
+
"loss": 0.181,
|
6304 |
+
"step": 897
|
6305 |
+
},
|
6306 |
+
{
|
6307 |
+
"epoch": 0.2655724372481609,
|
6308 |
+
"grad_norm": 0.3107486963272095,
|
6309 |
+
"learning_rate": 7.03038950900464e-05,
|
6310 |
+
"loss": 0.1367,
|
6311 |
+
"step": 898
|
6312 |
+
},
|
6313 |
+
{
|
6314 |
+
"epoch": 0.26586817492883813,
|
6315 |
+
"grad_norm": 0.4827352464199066,
|
6316 |
+
"learning_rate": 7.010262741290798e-05,
|
6317 |
+
"loss": 0.2186,
|
6318 |
+
"step": 899
|
6319 |
+
},
|
6320 |
+
{
|
6321 |
+
"epoch": 0.2661639126095154,
|
6322 |
+
"grad_norm": 0.5701566934585571,
|
6323 |
+
"learning_rate": 6.990149264650814e-05,
|
6324 |
+
"loss": 0.2813,
|
6325 |
+
"step": 900
|
6326 |
+
},
|
6327 |
+
{
|
6328 |
+
"epoch": 0.2664596502901926,
|
6329 |
+
"grad_norm": 0.10672613233327866,
|
6330 |
+
"learning_rate": 6.970049168500474e-05,
|
6331 |
+
"loss": 0.1362,
|
6332 |
+
"step": 901
|
6333 |
+
},
|
6334 |
+
{
|
6335 |
+
"epoch": 0.26675538797086984,
|
6336 |
+
"grad_norm": 0.123167484998703,
|
6337 |
+
"learning_rate": 6.94996254219608e-05,
|
6338 |
+
"loss": 0.1512,
|
6339 |
+
"step": 902
|
6340 |
+
},
|
6341 |
+
{
|
6342 |
+
"epoch": 0.2670511256515471,
|
6343 |
+
"grad_norm": 0.14451348781585693,
|
6344 |
+
"learning_rate": 6.929889475034048e-05,
|
6345 |
+
"loss": 0.2042,
|
6346 |
+
"step": 903
|
6347 |
+
},
|
6348 |
+
{
|
6349 |
+
"epoch": 0.2673468633322243,
|
6350 |
+
"grad_norm": 0.15311995148658752,
|
6351 |
+
"learning_rate": 6.909830056250527e-05,
|
6352 |
+
"loss": 0.1774,
|
6353 |
+
"step": 904
|
6354 |
+
},
|
6355 |
+
{
|
6356 |
+
"epoch": 0.26764260101290155,
|
6357 |
+
"grad_norm": 0.1696614921092987,
|
6358 |
+
"learning_rate": 6.889784375020978e-05,
|
6359 |
+
"loss": 0.2346,
|
6360 |
+
"step": 905
|
6361 |
+
},
|
6362 |
+
{
|
6363 |
+
"epoch": 0.2679383386935788,
|
6364 |
+
"grad_norm": 0.158672496676445,
|
6365 |
+
"learning_rate": 6.869752520459803e-05,
|
6366 |
+
"loss": 0.1889,
|
6367 |
+
"step": 906
|
6368 |
+
},
|
6369 |
+
{
|
6370 |
+
"epoch": 0.26823407637425606,
|
6371 |
+
"grad_norm": 0.1748230755329132,
|
6372 |
+
"learning_rate": 6.849734581619918e-05,
|
6373 |
+
"loss": 0.1925,
|
6374 |
+
"step": 907
|
6375 |
+
},
|
6376 |
+
{
|
6377 |
+
"epoch": 0.26852981405493326,
|
6378 |
+
"grad_norm": 0.1830466389656067,
|
6379 |
+
"learning_rate": 6.829730647492404e-05,
|
6380 |
+
"loss": 0.2056,
|
6381 |
+
"step": 908
|
6382 |
+
},
|
6383 |
+
{
|
6384 |
+
"epoch": 0.2688255517356105,
|
6385 |
+
"grad_norm": 0.17888867855072021,
|
6386 |
+
"learning_rate": 6.80974080700606e-05,
|
6387 |
+
"loss": 0.194,
|
6388 |
+
"step": 909
|
6389 |
+
},
|
6390 |
+
{
|
6391 |
+
"epoch": 0.26912128941628777,
|
6392 |
+
"grad_norm": 0.1619090586900711,
|
6393 |
+
"learning_rate": 6.789765149027039e-05,
|
6394 |
+
"loss": 0.1591,
|
6395 |
+
"step": 910
|
6396 |
+
},
|
6397 |
+
{
|
6398 |
+
"epoch": 0.26941702709696497,
|
6399 |
+
"grad_norm": 0.17818139493465424,
|
6400 |
+
"learning_rate": 6.769803762358443e-05,
|
6401 |
+
"loss": 0.2072,
|
6402 |
+
"step": 911
|
6403 |
+
},
|
6404 |
+
{
|
6405 |
+
"epoch": 0.2697127647776422,
|
6406 |
+
"grad_norm": 0.17115117609500885,
|
6407 |
+
"learning_rate": 6.749856735739928e-05,
|
6408 |
+
"loss": 0.1528,
|
6409 |
+
"step": 912
|
6410 |
+
},
|
6411 |
+
{
|
6412 |
+
"epoch": 0.2700085024583195,
|
6413 |
+
"grad_norm": 0.18800678849220276,
|
6414 |
+
"learning_rate": 6.729924157847323e-05,
|
6415 |
+
"loss": 0.2115,
|
6416 |
+
"step": 913
|
6417 |
+
},
|
6418 |
+
{
|
6419 |
+
"epoch": 0.27030424013899673,
|
6420 |
+
"grad_norm": 0.19144469499588013,
|
6421 |
+
"learning_rate": 6.710006117292209e-05,
|
6422 |
+
"loss": 0.2246,
|
6423 |
+
"step": 914
|
6424 |
+
},
|
6425 |
+
{
|
6426 |
+
"epoch": 0.27059997781967393,
|
6427 |
+
"grad_norm": 0.17849388718605042,
|
6428 |
+
"learning_rate": 6.690102702621548e-05,
|
6429 |
+
"loss": 0.1629,
|
6430 |
+
"step": 915
|
6431 |
+
},
|
6432 |
+
{
|
6433 |
+
"epoch": 0.2708957155003512,
|
6434 |
+
"grad_norm": 0.18855208158493042,
|
6435 |
+
"learning_rate": 6.670214002317278e-05,
|
6436 |
+
"loss": 0.1774,
|
6437 |
+
"step": 916
|
6438 |
+
},
|
6439 |
+
{
|
6440 |
+
"epoch": 0.27119145318102844,
|
6441 |
+
"grad_norm": 0.1926363706588745,
|
6442 |
+
"learning_rate": 6.650340104795932e-05,
|
6443 |
+
"loss": 0.201,
|
6444 |
+
"step": 917
|
6445 |
+
},
|
6446 |
+
{
|
6447 |
+
"epoch": 0.27148719086170564,
|
6448 |
+
"grad_norm": 0.21725712716579437,
|
6449 |
+
"learning_rate": 6.630481098408228e-05,
|
6450 |
+
"loss": 0.2316,
|
6451 |
+
"step": 918
|
6452 |
+
},
|
6453 |
+
{
|
6454 |
+
"epoch": 0.2717829285423829,
|
6455 |
+
"grad_norm": 0.2055988311767578,
|
6456 |
+
"learning_rate": 6.610637071438686e-05,
|
6457 |
+
"loss": 0.1953,
|
6458 |
+
"step": 919
|
6459 |
+
},
|
6460 |
+
{
|
6461 |
+
"epoch": 0.27207866622306015,
|
6462 |
+
"grad_norm": 0.186083123087883,
|
6463 |
+
"learning_rate": 6.590808112105232e-05,
|
6464 |
+
"loss": 0.1441,
|
6465 |
+
"step": 920
|
6466 |
+
},
|
6467 |
+
{
|
6468 |
+
"epoch": 0.2723744039037374,
|
6469 |
+
"grad_norm": 0.2097381055355072,
|
6470 |
+
"learning_rate": 6.570994308558812e-05,
|
6471 |
+
"loss": 0.2146,
|
6472 |
+
"step": 921
|
6473 |
+
},
|
6474 |
+
{
|
6475 |
+
"epoch": 0.2726701415844146,
|
6476 |
+
"grad_norm": 0.18082082271575928,
|
6477 |
+
"learning_rate": 6.551195748882997e-05,
|
6478 |
+
"loss": 0.1919,
|
6479 |
+
"step": 922
|
6480 |
+
},
|
6481 |
+
{
|
6482 |
+
"epoch": 0.27296587926509186,
|
6483 |
+
"grad_norm": 0.19363375008106232,
|
6484 |
+
"learning_rate": 6.531412521093586e-05,
|
6485 |
+
"loss": 0.1879,
|
6486 |
+
"step": 923
|
6487 |
+
},
|
6488 |
+
{
|
6489 |
+
"epoch": 0.2732616169457691,
|
6490 |
+
"grad_norm": 0.2286785989999771,
|
6491 |
+
"learning_rate": 6.51164471313822e-05,
|
6492 |
+
"loss": 0.2411,
|
6493 |
+
"step": 924
|
6494 |
+
},
|
6495 |
+
{
|
6496 |
+
"epoch": 0.27355735462644637,
|
6497 |
+
"grad_norm": 0.21459421515464783,
|
6498 |
+
"learning_rate": 6.491892412895995e-05,
|
6499 |
+
"loss": 0.1957,
|
6500 |
+
"step": 925
|
6501 |
+
},
|
6502 |
+
{
|
6503 |
+
"epoch": 0.27385309230712357,
|
6504 |
+
"grad_norm": 0.18594589829444885,
|
6505 |
+
"learning_rate": 6.472155708177052e-05,
|
6506 |
+
"loss": 0.1569,
|
6507 |
+
"step": 926
|
6508 |
+
},
|
6509 |
+
{
|
6510 |
+
"epoch": 0.2741488299878008,
|
6511 |
+
"grad_norm": 0.20744723081588745,
|
6512 |
+
"learning_rate": 6.452434686722224e-05,
|
6513 |
+
"loss": 0.1756,
|
6514 |
+
"step": 927
|
6515 |
+
},
|
6516 |
+
{
|
6517 |
+
"epoch": 0.2744445676684781,
|
6518 |
+
"grad_norm": 0.2012493759393692,
|
6519 |
+
"learning_rate": 6.432729436202604e-05,
|
6520 |
+
"loss": 0.1642,
|
6521 |
+
"step": 928
|
6522 |
+
},
|
6523 |
+
{
|
6524 |
+
"epoch": 0.2747403053491553,
|
6525 |
+
"grad_norm": 0.2502315938472748,
|
6526 |
+
"learning_rate": 6.41304004421918e-05,
|
6527 |
+
"loss": 0.2156,
|
6528 |
+
"step": 929
|
6529 |
+
},
|
6530 |
+
{
|
6531 |
+
"epoch": 0.27503604302983253,
|
6532 |
+
"grad_norm": 0.22481419146060944,
|
6533 |
+
"learning_rate": 6.393366598302446e-05,
|
6534 |
+
"loss": 0.1745,
|
6535 |
+
"step": 930
|
6536 |
+
},
|
6537 |
+
{
|
6538 |
+
"epoch": 0.2753317807105098,
|
6539 |
+
"grad_norm": 0.21398960053920746,
|
6540 |
+
"learning_rate": 6.373709185911998e-05,
|
6541 |
+
"loss": 0.1837,
|
6542 |
+
"step": 931
|
6543 |
+
},
|
6544 |
+
{
|
6545 |
+
"epoch": 0.27562751839118704,
|
6546 |
+
"grad_norm": 0.2531358301639557,
|
6547 |
+
"learning_rate": 6.354067894436155e-05,
|
6548 |
+
"loss": 0.2479,
|
6549 |
+
"step": 932
|
6550 |
+
},
|
6551 |
+
{
|
6552 |
+
"epoch": 0.27592325607186424,
|
6553 |
+
"grad_norm": 0.2133578509092331,
|
6554 |
+
"learning_rate": 6.334442811191576e-05,
|
6555 |
+
"loss": 0.1834,
|
6556 |
+
"step": 933
|
6557 |
+
},
|
6558 |
+
{
|
6559 |
+
"epoch": 0.2762189937525415,
|
6560 |
+
"grad_norm": 0.24796530604362488,
|
6561 |
+
"learning_rate": 6.314834023422858e-05,
|
6562 |
+
"loss": 0.1974,
|
6563 |
+
"step": 934
|
6564 |
+
},
|
6565 |
+
{
|
6566 |
+
"epoch": 0.27651473143321875,
|
6567 |
+
"grad_norm": 0.2332259565591812,
|
6568 |
+
"learning_rate": 6.295241618302156e-05,
|
6569 |
+
"loss": 0.1923,
|
6570 |
+
"step": 935
|
6571 |
+
},
|
6572 |
+
{
|
6573 |
+
"epoch": 0.27681046911389595,
|
6574 |
+
"grad_norm": 0.221559077501297,
|
6575 |
+
"learning_rate": 6.275665682928803e-05,
|
6576 |
+
"loss": 0.154,
|
6577 |
+
"step": 936
|
6578 |
+
},
|
6579 |
+
{
|
6580 |
+
"epoch": 0.2771062067945732,
|
6581 |
+
"grad_norm": 0.24034371972084045,
|
6582 |
+
"learning_rate": 6.256106304328905e-05,
|
6583 |
+
"loss": 0.187,
|
6584 |
+
"step": 937
|
6585 |
+
},
|
6586 |
+
{
|
6587 |
+
"epoch": 0.27740194447525046,
|
6588 |
+
"grad_norm": 0.25638851523399353,
|
6589 |
+
"learning_rate": 6.23656356945497e-05,
|
6590 |
+
"loss": 0.1833,
|
6591 |
+
"step": 938
|
6592 |
+
},
|
6593 |
+
{
|
6594 |
+
"epoch": 0.2776976821559277,
|
6595 |
+
"grad_norm": 0.25576722621917725,
|
6596 |
+
"learning_rate": 6.21703756518551e-05,
|
6597 |
+
"loss": 0.1888,
|
6598 |
+
"step": 939
|
6599 |
+
},
|
6600 |
+
{
|
6601 |
+
"epoch": 0.2779934198366049,
|
6602 |
+
"grad_norm": 0.2722214162349701,
|
6603 |
+
"learning_rate": 6.197528378324665e-05,
|
6604 |
+
"loss": 0.1869,
|
6605 |
+
"step": 940
|
6606 |
+
},
|
6607 |
+
{
|
6608 |
+
"epoch": 0.27828915751728217,
|
6609 |
+
"grad_norm": 0.2651073634624481,
|
6610 |
+
"learning_rate": 6.17803609560181e-05,
|
6611 |
+
"loss": 0.1871,
|
6612 |
+
"step": 941
|
6613 |
+
},
|
6614 |
+
{
|
6615 |
+
"epoch": 0.2785848951979594,
|
6616 |
+
"grad_norm": 0.24269293248653412,
|
6617 |
+
"learning_rate": 6.158560803671168e-05,
|
6618 |
+
"loss": 0.157,
|
6619 |
+
"step": 942
|
6620 |
+
},
|
6621 |
+
{
|
6622 |
+
"epoch": 0.2788806328786366,
|
6623 |
+
"grad_norm": 0.24029861390590668,
|
6624 |
+
"learning_rate": 6.139102589111435e-05,
|
6625 |
+
"loss": 0.1162,
|
6626 |
+
"step": 943
|
6627 |
+
},
|
6628 |
+
{
|
6629 |
+
"epoch": 0.2791763705593139,
|
6630 |
+
"grad_norm": 0.30216261744499207,
|
6631 |
+
"learning_rate": 6.119661538425381e-05,
|
6632 |
+
"loss": 0.1928,
|
6633 |
+
"step": 944
|
6634 |
+
},
|
6635 |
+
{
|
6636 |
+
"epoch": 0.27947210823999113,
|
6637 |
+
"grad_norm": 0.29744547605514526,
|
6638 |
+
"learning_rate": 6.100237738039484e-05,
|
6639 |
+
"loss": 0.195,
|
6640 |
+
"step": 945
|
6641 |
+
},
|
6642 |
+
{
|
6643 |
+
"epoch": 0.2797678459206684,
|
6644 |
+
"grad_norm": 0.3087689280509949,
|
6645 |
+
"learning_rate": 6.0808312743035236e-05,
|
6646 |
+
"loss": 0.2018,
|
6647 |
+
"step": 946
|
6648 |
+
},
|
6649 |
+
{
|
6650 |
+
"epoch": 0.2800635836013456,
|
6651 |
+
"grad_norm": 0.30813395977020264,
|
6652 |
+
"learning_rate": 6.061442233490211e-05,
|
6653 |
+
"loss": 0.1712,
|
6654 |
+
"step": 947
|
6655 |
+
},
|
6656 |
+
{
|
6657 |
+
"epoch": 0.28035932128202284,
|
6658 |
+
"grad_norm": 0.2932427227497101,
|
6659 |
+
"learning_rate": 6.042070701794806e-05,
|
6660 |
+
"loss": 0.1527,
|
6661 |
+
"step": 948
|
6662 |
+
},
|
6663 |
+
{
|
6664 |
+
"epoch": 0.2806550589627001,
|
6665 |
+
"grad_norm": 0.3830631673336029,
|
6666 |
+
"learning_rate": 6.0227167653347305e-05,
|
6667 |
+
"loss": 0.1975,
|
6668 |
+
"step": 949
|
6669 |
+
},
|
6670 |
+
{
|
6671 |
+
"epoch": 0.2809507966433773,
|
6672 |
+
"grad_norm": 0.5851044654846191,
|
6673 |
+
"learning_rate": 6.0033805101491794e-05,
|
6674 |
+
"loss": 0.283,
|
6675 |
+
"step": 950
|
6676 |
+
},
|
6677 |
+
{
|
6678 |
+
"epoch": 0.28124653432405455,
|
6679 |
+
"grad_norm": 0.12433775514364243,
|
6680 |
+
"learning_rate": 5.98406202219875e-05,
|
6681 |
+
"loss": 0.176,
|
6682 |
+
"step": 951
|
6683 |
+
},
|
6684 |
+
{
|
6685 |
+
"epoch": 0.2815422720047318,
|
6686 |
+
"grad_norm": 0.13331016898155212,
|
6687 |
+
"learning_rate": 5.964761387365052e-05,
|
6688 |
+
"loss": 0.1551,
|
6689 |
+
"step": 952
|
6690 |
+
},
|
6691 |
+
{
|
6692 |
+
"epoch": 0.28183800968540906,
|
6693 |
+
"grad_norm": 0.15550710260868073,
|
6694 |
+
"learning_rate": 5.9454786914503255e-05,
|
6695 |
+
"loss": 0.1739,
|
6696 |
+
"step": 953
|
6697 |
+
},
|
6698 |
+
{
|
6699 |
+
"epoch": 0.28213374736608626,
|
6700 |
+
"grad_norm": 0.1473347395658493,
|
6701 |
+
"learning_rate": 5.926214020177074e-05,
|
6702 |
+
"loss": 0.1637,
|
6703 |
+
"step": 954
|
6704 |
+
},
|
6705 |
+
{
|
6706 |
+
"epoch": 0.2824294850467635,
|
6707 |
+
"grad_norm": 0.17270837724208832,
|
6708 |
+
"learning_rate": 5.9069674591876534e-05,
|
6709 |
+
"loss": 0.1727,
|
6710 |
+
"step": 955
|
6711 |
+
},
|
6712 |
+
{
|
6713 |
+
"epoch": 0.2827252227274408,
|
6714 |
+
"grad_norm": 0.14710725843906403,
|
6715 |
+
"learning_rate": 5.887739094043923e-05,
|
6716 |
+
"loss": 0.1256,
|
6717 |
+
"step": 956
|
6718 |
+
},
|
6719 |
+
{
|
6720 |
+
"epoch": 0.28302096040811797,
|
6721 |
+
"grad_norm": 0.1642570048570633,
|
6722 |
+
"learning_rate": 5.868529010226845e-05,
|
6723 |
+
"loss": 0.1671,
|
6724 |
+
"step": 957
|
6725 |
+
},
|
6726 |
+
{
|
6727 |
+
"epoch": 0.2833166980887952,
|
6728 |
+
"grad_norm": 0.1761314868927002,
|
6729 |
+
"learning_rate": 5.849337293136112e-05,
|
6730 |
+
"loss": 0.1987,
|
6731 |
+
"step": 958
|
6732 |
+
},
|
6733 |
+
{
|
6734 |
+
"epoch": 0.2836124357694725,
|
6735 |
+
"grad_norm": 0.15199066698551178,
|
6736 |
+
"learning_rate": 5.830164028089766e-05,
|
6737 |
+
"loss": 0.1716,
|
6738 |
+
"step": 959
|
6739 |
+
},
|
6740 |
+
{
|
6741 |
+
"epoch": 0.28390817345014974,
|
6742 |
+
"grad_norm": 0.18991252779960632,
|
6743 |
+
"learning_rate": 5.811009300323818e-05,
|
6744 |
+
"loss": 0.2306,
|
6745 |
+
"step": 960
|
6746 |
+
},
|
6747 |
+
{
|
6748 |
+
"epoch": 0.28420391113082694,
|
6749 |
+
"grad_norm": 0.18644598126411438,
|
6750 |
+
"learning_rate": 5.791873194991872e-05,
|
6751 |
+
"loss": 0.2177,
|
6752 |
+
"step": 961
|
6753 |
+
},
|
6754 |
+
{
|
6755 |
+
"epoch": 0.2844996488115042,
|
6756 |
+
"grad_norm": 0.17034782469272614,
|
6757 |
+
"learning_rate": 5.7727557971647427e-05,
|
6758 |
+
"loss": 0.1614,
|
6759 |
+
"step": 962
|
6760 |
+
},
|
6761 |
+
{
|
6762 |
+
"epoch": 0.28479538649218145,
|
6763 |
+
"grad_norm": 0.17145271599292755,
|
6764 |
+
"learning_rate": 5.7536571918300864e-05,
|
6765 |
+
"loss": 0.1853,
|
6766 |
+
"step": 963
|
6767 |
+
},
|
6768 |
+
{
|
6769 |
+
"epoch": 0.2850911241728587,
|
6770 |
+
"grad_norm": 0.20625253021717072,
|
6771 |
+
"learning_rate": 5.734577463892008e-05,
|
6772 |
+
"loss": 0.2303,
|
6773 |
+
"step": 964
|
6774 |
+
},
|
6775 |
+
{
|
6776 |
+
"epoch": 0.2853868618535359,
|
6777 |
+
"grad_norm": 0.17384910583496094,
|
6778 |
+
"learning_rate": 5.7155166981706956e-05,
|
6779 |
+
"loss": 0.1725,
|
6780 |
+
"step": 965
|
6781 |
+
},
|
6782 |
+
{
|
6783 |
+
"epoch": 0.28568259953421316,
|
6784 |
+
"grad_norm": 0.1765337586402893,
|
6785 |
+
"learning_rate": 5.6964749794020354e-05,
|
6786 |
+
"loss": 0.1495,
|
6787 |
+
"step": 966
|
6788 |
+
},
|
6789 |
+
{
|
6790 |
+
"epoch": 0.2859783372148904,
|
6791 |
+
"grad_norm": 0.22681960463523865,
|
6792 |
+
"learning_rate": 5.6774523922372394e-05,
|
6793 |
+
"loss": 0.2106,
|
6794 |
+
"step": 967
|
6795 |
+
},
|
6796 |
+
{
|
6797 |
+
"epoch": 0.2862740748955676,
|
6798 |
+
"grad_norm": 0.21203817427158356,
|
6799 |
+
"learning_rate": 5.6584490212424804e-05,
|
6800 |
+
"loss": 0.2547,
|
6801 |
+
"step": 968
|
6802 |
+
},
|
6803 |
+
{
|
6804 |
+
"epoch": 0.28656981257624486,
|
6805 |
+
"grad_norm": 0.1752469837665558,
|
6806 |
+
"learning_rate": 5.639464950898491e-05,
|
6807 |
+
"loss": 0.1504,
|
6808 |
+
"step": 969
|
6809 |
+
},
|
6810 |
+
{
|
6811 |
+
"epoch": 0.2868655502569221,
|
6812 |
+
"grad_norm": 0.18426285684108734,
|
6813 |
+
"learning_rate": 5.620500265600206e-05,
|
6814 |
+
"loss": 0.1627,
|
6815 |
+
"step": 970
|
6816 |
+
},
|
6817 |
+
{
|
6818 |
+
"epoch": 0.2871612879375994,
|
6819 |
+
"grad_norm": 0.20485304296016693,
|
6820 |
+
"learning_rate": 5.601555049656382e-05,
|
6821 |
+
"loss": 0.2002,
|
6822 |
+
"step": 971
|
6823 |
+
},
|
6824 |
+
{
|
6825 |
+
"epoch": 0.2874570256182766,
|
6826 |
+
"grad_norm": 0.21289728581905365,
|
6827 |
+
"learning_rate": 5.58262938728922e-05,
|
6828 |
+
"loss": 0.2021,
|
6829 |
+
"step": 972
|
6830 |
+
},
|
6831 |
+
{
|
6832 |
+
"epoch": 0.28775276329895383,
|
6833 |
+
"grad_norm": 0.20441678166389465,
|
6834 |
+
"learning_rate": 5.563723362634008e-05,
|
6835 |
+
"loss": 0.1861,
|
6836 |
+
"step": 973
|
6837 |
+
},
|
6838 |
+
{
|
6839 |
+
"epoch": 0.2880485009796311,
|
6840 |
+
"grad_norm": 0.22460561990737915,
|
6841 |
+
"learning_rate": 5.544837059738719e-05,
|
6842 |
+
"loss": 0.2161,
|
6843 |
+
"step": 974
|
6844 |
+
},
|
6845 |
+
{
|
6846 |
+
"epoch": 0.2883442386603083,
|
6847 |
+
"grad_norm": 0.2056421935558319,
|
6848 |
+
"learning_rate": 5.525970562563656e-05,
|
6849 |
+
"loss": 0.1906,
|
6850 |
+
"step": 975
|
6851 |
+
},
|
6852 |
+
{
|
6853 |
+
"epoch": 0.28863997634098554,
|
6854 |
+
"grad_norm": 0.22834764420986176,
|
6855 |
+
"learning_rate": 5.507123954981073e-05,
|
6856 |
+
"loss": 0.1751,
|
6857 |
+
"step": 976
|
6858 |
+
},
|
6859 |
+
{
|
6860 |
+
"epoch": 0.2889357140216628,
|
6861 |
+
"grad_norm": 0.19000335037708282,
|
6862 |
+
"learning_rate": 5.488297320774807e-05,
|
6863 |
+
"loss": 0.1728,
|
6864 |
+
"step": 977
|
6865 |
+
},
|
6866 |
+
{
|
6867 |
+
"epoch": 0.28923145170234005,
|
6868 |
+
"grad_norm": 0.209343820810318,
|
6869 |
+
"learning_rate": 5.4694907436399e-05,
|
6870 |
+
"loss": 0.1645,
|
6871 |
+
"step": 978
|
6872 |
+
},
|
6873 |
+
{
|
6874 |
+
"epoch": 0.28952718938301725,
|
6875 |
+
"grad_norm": 0.2307262420654297,
|
6876 |
+
"learning_rate": 5.4507043071822284e-05,
|
6877 |
+
"loss": 0.2139,
|
6878 |
+
"step": 979
|
6879 |
+
},
|
6880 |
+
{
|
6881 |
+
"epoch": 0.2898229270636945,
|
6882 |
+
"grad_norm": 0.2119540274143219,
|
6883 |
+
"learning_rate": 5.431938094918132e-05,
|
6884 |
+
"loss": 0.1719,
|
6885 |
+
"step": 980
|
6886 |
+
},
|
6887 |
+
{
|
6888 |
+
"epoch": 0.29011866474437176,
|
6889 |
+
"grad_norm": 0.24757258594036102,
|
6890 |
+
"learning_rate": 5.41319219027404e-05,
|
6891 |
+
"loss": 0.2469,
|
6892 |
+
"step": 981
|
6893 |
+
},
|
6894 |
+
{
|
6895 |
+
"epoch": 0.29041440242504896,
|
6896 |
+
"grad_norm": 0.23471477627754211,
|
6897 |
+
"learning_rate": 5.394466676586114e-05,
|
6898 |
+
"loss": 0.1832,
|
6899 |
+
"step": 982
|
6900 |
+
},
|
6901 |
+
{
|
6902 |
+
"epoch": 0.2907101401057262,
|
6903 |
+
"grad_norm": 0.22288459539413452,
|
6904 |
+
"learning_rate": 5.375761637099854e-05,
|
6905 |
+
"loss": 0.1564,
|
6906 |
+
"step": 983
|
6907 |
+
},
|
6908 |
+
{
|
6909 |
+
"epoch": 0.29100587778640347,
|
6910 |
+
"grad_norm": 0.22171498835086823,
|
6911 |
+
"learning_rate": 5.357077154969742e-05,
|
6912 |
+
"loss": 0.1845,
|
6913 |
+
"step": 984
|
6914 |
+
},
|
6915 |
+
{
|
6916 |
+
"epoch": 0.2913016154670807,
|
6917 |
+
"grad_norm": 0.2288697361946106,
|
6918 |
+
"learning_rate": 5.3384133132588784e-05,
|
6919 |
+
"loss": 0.1793,
|
6920 |
+
"step": 985
|
6921 |
+
},
|
6922 |
+
{
|
6923 |
+
"epoch": 0.2915973531477579,
|
6924 |
+
"grad_norm": 0.30814412236213684,
|
6925 |
+
"learning_rate": 5.3197701949386e-05,
|
6926 |
+
"loss": 0.2074,
|
6927 |
+
"step": 986
|
6928 |
+
},
|
6929 |
+
{
|
6930 |
+
"epoch": 0.2918930908284352,
|
6931 |
+
"grad_norm": 0.22758153080940247,
|
6932 |
+
"learning_rate": 5.301147882888116e-05,
|
6933 |
+
"loss": 0.1599,
|
6934 |
+
"step": 987
|
6935 |
+
},
|
6936 |
+
{
|
6937 |
+
"epoch": 0.29218882850911243,
|
6938 |
+
"grad_norm": 0.23016326129436493,
|
6939 |
+
"learning_rate": 5.28254645989414e-05,
|
6940 |
+
"loss": 0.1528,
|
6941 |
+
"step": 988
|
6942 |
+
},
|
6943 |
+
{
|
6944 |
+
"epoch": 0.29248456618978963,
|
6945 |
+
"grad_norm": 0.2873390316963196,
|
6946 |
+
"learning_rate": 5.2639660086505226e-05,
|
6947 |
+
"loss": 0.2101,
|
6948 |
+
"step": 989
|
6949 |
+
},
|
6950 |
+
{
|
6951 |
+
"epoch": 0.2927803038704669,
|
6952 |
+
"grad_norm": 0.2666926681995392,
|
6953 |
+
"learning_rate": 5.2454066117578815e-05,
|
6954 |
+
"loss": 0.1964,
|
6955 |
+
"step": 990
|
6956 |
+
},
|
6957 |
+
{
|
6958 |
+
"epoch": 0.29307604155114414,
|
6959 |
+
"grad_norm": 0.26695266366004944,
|
6960 |
+
"learning_rate": 5.226868351723244e-05,
|
6961 |
+
"loss": 0.1716,
|
6962 |
+
"step": 991
|
6963 |
+
},
|
6964 |
+
{
|
6965 |
+
"epoch": 0.2933717792318214,
|
6966 |
+
"grad_norm": 0.3416760265827179,
|
6967 |
+
"learning_rate": 5.2083513109596616e-05,
|
6968 |
+
"loss": 0.2446,
|
6969 |
+
"step": 992
|
6970 |
+
},
|
6971 |
+
{
|
6972 |
+
"epoch": 0.2936675169124986,
|
6973 |
+
"grad_norm": 0.2576669156551361,
|
6974 |
+
"learning_rate": 5.189855571785859e-05,
|
6975 |
+
"loss": 0.111,
|
6976 |
+
"step": 993
|
6977 |
+
},
|
6978 |
+
{
|
6979 |
+
"epoch": 0.29396325459317585,
|
6980 |
+
"grad_norm": 0.2883356213569641,
|
6981 |
+
"learning_rate": 5.171381216425863e-05,
|
6982 |
+
"loss": 0.1702,
|
6983 |
+
"step": 994
|
6984 |
+
},
|
6985 |
+
{
|
6986 |
+
"epoch": 0.2942589922738531,
|
6987 |
+
"grad_norm": 0.3241150379180908,
|
6988 |
+
"learning_rate": 5.152928327008635e-05,
|
6989 |
+
"loss": 0.2342,
|
6990 |
+
"step": 995
|
6991 |
+
},
|
6992 |
+
{
|
6993 |
+
"epoch": 0.29455472995453036,
|
6994 |
+
"grad_norm": 0.3817709684371948,
|
6995 |
+
"learning_rate": 5.134496985567714e-05,
|
6996 |
+
"loss": 0.2512,
|
6997 |
+
"step": 996
|
6998 |
+
},
|
6999 |
+
{
|
7000 |
+
"epoch": 0.29485046763520756,
|
7001 |
+
"grad_norm": 0.34549957513809204,
|
7002 |
+
"learning_rate": 5.116087274040837e-05,
|
7003 |
+
"loss": 0.2041,
|
7004 |
+
"step": 997
|
7005 |
+
},
|
7006 |
+
{
|
7007 |
+
"epoch": 0.2951462053158848,
|
7008 |
+
"grad_norm": 0.39972737431526184,
|
7009 |
+
"learning_rate": 5.0976992742695925e-05,
|
7010 |
+
"loss": 0.2526,
|
7011 |
+
"step": 998
|
7012 |
+
},
|
7013 |
+
{
|
7014 |
+
"epoch": 0.29544194299656207,
|
7015 |
+
"grad_norm": 0.4173913598060608,
|
7016 |
+
"learning_rate": 5.07933306799904e-05,
|
7017 |
+
"loss": 0.2467,
|
7018 |
+
"step": 999
|
7019 |
+
},
|
7020 |
+
{
|
7021 |
+
"epoch": 0.29573768067723927,
|
7022 |
+
"grad_norm": 0.58672034740448,
|
7023 |
+
"learning_rate": 5.060988736877366e-05,
|
7024 |
+
"loss": 0.2481,
|
7025 |
+
"step": 1000
|
7026 |
+
},
|
7027 |
+
{
|
7028 |
+
"epoch": 0.2960334183579165,
|
7029 |
+
"grad_norm": 0.10767298191785812,
|
7030 |
+
"learning_rate": 5.042666362455498e-05,
|
7031 |
+
"loss": 0.1372,
|
7032 |
+
"step": 1001
|
7033 |
+
},
|
7034 |
+
{
|
7035 |
+
"epoch": 0.2963291560385938,
|
7036 |
+
"grad_norm": 0.14959672093391418,
|
7037 |
+
"learning_rate": 5.024366026186755e-05,
|
7038 |
+
"loss": 0.2057,
|
7039 |
+
"step": 1002
|
7040 |
+
},
|
7041 |
+
{
|
7042 |
+
"epoch": 0.29662489371927103,
|
7043 |
+
"grad_norm": 0.14518089592456818,
|
7044 |
+
"learning_rate": 5.006087809426496e-05,
|
7045 |
+
"loss": 0.1744,
|
7046 |
+
"step": 1003
|
7047 |
+
},
|
7048 |
+
{
|
7049 |
+
"epoch": 0.29692063139994823,
|
7050 |
+
"grad_norm": 0.16306771337985992,
|
7051 |
+
"learning_rate": 4.987831793431731e-05,
|
7052 |
+
"loss": 0.1904,
|
7053 |
+
"step": 1004
|
7054 |
+
},
|
7055 |
+
{
|
7056 |
+
"epoch": 0.2972163690806255,
|
7057 |
+
"grad_norm": 0.18029610812664032,
|
7058 |
+
"learning_rate": 4.9695980593607817e-05,
|
7059 |
+
"loss": 0.2223,
|
7060 |
+
"step": 1005
|
7061 |
+
},
|
7062 |
+
{
|
7063 |
+
"epoch": 0.29751210676130274,
|
7064 |
+
"grad_norm": 0.18586061894893646,
|
7065 |
+
"learning_rate": 4.9513866882729146e-05,
|
7066 |
+
"loss": 0.2306,
|
7067 |
+
"step": 1006
|
7068 |
+
},
|
7069 |
+
{
|
7070 |
+
"epoch": 0.29780784444197994,
|
7071 |
+
"grad_norm": 0.17091114819049835,
|
7072 |
+
"learning_rate": 4.9331977611279777e-05,
|
7073 |
+
"loss": 0.1717,
|
7074 |
+
"step": 1007
|
7075 |
+
},
|
7076 |
+
{
|
7077 |
+
"epoch": 0.2981035821226572,
|
7078 |
+
"grad_norm": 0.15413694083690643,
|
7079 |
+
"learning_rate": 4.9150313587860433e-05,
|
7080 |
+
"loss": 0.1638,
|
7081 |
+
"step": 1008
|
7082 |
+
},
|
7083 |
+
{
|
7084 |
+
"epoch": 0.29839931980333445,
|
7085 |
+
"grad_norm": 0.15122149884700775,
|
7086 |
+
"learning_rate": 4.896887562007054e-05,
|
7087 |
+
"loss": 0.1489,
|
7088 |
+
"step": 1009
|
7089 |
+
},
|
7090 |
+
{
|
7091 |
+
"epoch": 0.2986950574840117,
|
7092 |
+
"grad_norm": 0.16240714490413666,
|
7093 |
+
"learning_rate": 4.8787664514504504e-05,
|
7094 |
+
"loss": 0.1718,
|
7095 |
+
"step": 1010
|
7096 |
+
},
|
7097 |
+
{
|
7098 |
+
"epoch": 0.2989907951646889,
|
7099 |
+
"grad_norm": 0.17468984425067902,
|
7100 |
+
"learning_rate": 4.860668107674823e-05,
|
7101 |
+
"loss": 0.1577,
|
7102 |
+
"step": 1011
|
7103 |
+
},
|
7104 |
+
{
|
7105 |
+
"epoch": 0.29928653284536616,
|
7106 |
+
"grad_norm": 0.19555851817131042,
|
7107 |
+
"learning_rate": 4.8425926111375506e-05,
|
7108 |
+
"loss": 0.2611,
|
7109 |
+
"step": 1012
|
7110 |
+
},
|
7111 |
+
{
|
7112 |
+
"epoch": 0.2995822705260434,
|
7113 |
+
"grad_norm": 0.20301313698291779,
|
7114 |
+
"learning_rate": 4.824540042194443e-05,
|
7115 |
+
"loss": 0.2266,
|
7116 |
+
"step": 1013
|
7117 |
+
},
|
7118 |
+
{
|
7119 |
+
"epoch": 0.2998780082067206,
|
7120 |
+
"grad_norm": 0.1904042363166809,
|
7121 |
+
"learning_rate": 4.8065104810993856e-05,
|
7122 |
+
"loss": 0.1847,
|
7123 |
+
"step": 1014
|
7124 |
+
},
|
7125 |
+
{
|
7126 |
+
"epoch": 0.30017374588739787,
|
7127 |
+
"grad_norm": 0.18135389685630798,
|
7128 |
+
"learning_rate": 4.788504008003978e-05,
|
7129 |
+
"loss": 0.1712,
|
7130 |
+
"step": 1015
|
7131 |
+
},
|
7132 |
+
{
|
7133 |
+
"epoch": 0.3004694835680751,
|
7134 |
+
"grad_norm": 0.2035091519355774,
|
7135 |
+
"learning_rate": 4.770520702957182e-05,
|
7136 |
+
"loss": 0.2459,
|
7137 |
+
"step": 1016
|
7138 |
+
},
|
7139 |
+
{
|
7140 |
+
"epoch": 0.3007652212487524,
|
7141 |
+
"grad_norm": 0.1759434938430786,
|
7142 |
+
"learning_rate": 4.752560645904962e-05,
|
7143 |
+
"loss": 0.1494,
|
7144 |
+
"step": 1017
|
7145 |
+
},
|
7146 |
+
{
|
7147 |
+
"epoch": 0.3010609589294296,
|
7148 |
+
"grad_norm": 0.21353545784950256,
|
7149 |
+
"learning_rate": 4.734623916689941e-05,
|
7150 |
+
"loss": 0.2468,
|
7151 |
+
"step": 1018
|
7152 |
+
},
|
7153 |
+
{
|
7154 |
+
"epoch": 0.30135669661010683,
|
7155 |
+
"grad_norm": 0.19691097736358643,
|
7156 |
+
"learning_rate": 4.716710595051022e-05,
|
7157 |
+
"loss": 0.1913,
|
7158 |
+
"step": 1019
|
7159 |
+
},
|
7160 |
+
{
|
7161 |
+
"epoch": 0.3016524342907841,
|
7162 |
+
"grad_norm": 0.20307397842407227,
|
7163 |
+
"learning_rate": 4.698820760623064e-05,
|
7164 |
+
"loss": 0.2042,
|
7165 |
+
"step": 1020
|
7166 |
+
},
|
7167 |
+
{
|
7168 |
+
"epoch": 0.3019481719714613,
|
7169 |
+
"grad_norm": 0.1996135711669922,
|
7170 |
+
"learning_rate": 4.6809544929365004e-05,
|
7171 |
+
"loss": 0.1734,
|
7172 |
+
"step": 1021
|
7173 |
+
},
|
7174 |
+
{
|
7175 |
+
"epoch": 0.30224390965213854,
|
7176 |
+
"grad_norm": 0.1997922658920288,
|
7177 |
+
"learning_rate": 4.663111871417e-05,
|
7178 |
+
"loss": 0.1601,
|
7179 |
+
"step": 1022
|
7180 |
+
},
|
7181 |
+
{
|
7182 |
+
"epoch": 0.3025396473328158,
|
7183 |
+
"grad_norm": 0.2047778218984604,
|
7184 |
+
"learning_rate": 4.645292975385111e-05,
|
7185 |
+
"loss": 0.1823,
|
7186 |
+
"step": 1023
|
7187 |
+
},
|
7188 |
+
{
|
7189 |
+
"epoch": 0.30283538501349305,
|
7190 |
+
"grad_norm": 0.21905602514743805,
|
7191 |
+
"learning_rate": 4.627497884055912e-05,
|
7192 |
+
"loss": 0.1889,
|
7193 |
+
"step": 1024
|
7194 |
+
},
|
7195 |
+
{
|
7196 |
+
"epoch": 0.30313112269417025,
|
7197 |
+
"grad_norm": 0.22655650973320007,
|
7198 |
+
"learning_rate": 4.609726676538652e-05,
|
7199 |
+
"loss": 0.204,
|
7200 |
+
"step": 1025
|
7201 |
+
},
|
7202 |
+
{
|
7203 |
+
"epoch": 0.3034268603748475,
|
7204 |
+
"grad_norm": 0.2180885523557663,
|
7205 |
+
"learning_rate": 4.591979431836402e-05,
|
7206 |
+
"loss": 0.2277,
|
7207 |
+
"step": 1026
|
7208 |
+
},
|
7209 |
+
{
|
7210 |
+
"epoch": 0.30372259805552476,
|
7211 |
+
"grad_norm": 0.2357274740934372,
|
7212 |
+
"learning_rate": 4.574256228845706e-05,
|
7213 |
+
"loss": 0.1818,
|
7214 |
+
"step": 1027
|
7215 |
+
},
|
7216 |
+
{
|
7217 |
+
"epoch": 0.30401833573620196,
|
7218 |
+
"grad_norm": 0.24533364176750183,
|
7219 |
+
"learning_rate": 4.5565571463562365e-05,
|
7220 |
+
"loss": 0.209,
|
7221 |
+
"step": 1028
|
7222 |
+
},
|
7223 |
+
{
|
7224 |
+
"epoch": 0.3043140734168792,
|
7225 |
+
"grad_norm": 0.22792395949363708,
|
7226 |
+
"learning_rate": 4.5388822630504256e-05,
|
7227 |
+
"loss": 0.2153,
|
7228 |
+
"step": 1029
|
7229 |
+
},
|
7230 |
+
{
|
7231 |
+
"epoch": 0.30460981109755647,
|
7232 |
+
"grad_norm": 0.2345706820487976,
|
7233 |
+
"learning_rate": 4.521231657503132e-05,
|
7234 |
+
"loss": 0.1802,
|
7235 |
+
"step": 1030
|
7236 |
+
},
|
7237 |
+
{
|
7238 |
+
"epoch": 0.3049055487782337,
|
7239 |
+
"grad_norm": 0.23721234500408173,
|
7240 |
+
"learning_rate": 4.503605408181286e-05,
|
7241 |
+
"loss": 0.2189,
|
7242 |
+
"step": 1031
|
7243 |
+
},
|
7244 |
+
{
|
7245 |
+
"epoch": 0.3052012864589109,
|
7246 |
+
"grad_norm": 0.24193932116031647,
|
7247 |
+
"learning_rate": 4.486003593443537e-05,
|
7248 |
+
"loss": 0.2247,
|
7249 |
+
"step": 1032
|
7250 |
+
},
|
7251 |
+
{
|
7252 |
+
"epoch": 0.3054970241395882,
|
7253 |
+
"grad_norm": 0.24470211565494537,
|
7254 |
+
"learning_rate": 4.468426291539914e-05,
|
7255 |
+
"loss": 0.1972,
|
7256 |
+
"step": 1033
|
7257 |
+
},
|
7258 |
+
{
|
7259 |
+
"epoch": 0.30579276182026544,
|
7260 |
+
"grad_norm": 0.22956788539886475,
|
7261 |
+
"learning_rate": 4.4508735806114654e-05,
|
7262 |
+
"loss": 0.1631,
|
7263 |
+
"step": 1034
|
7264 |
+
},
|
7265 |
+
{
|
7266 |
+
"epoch": 0.3060884995009427,
|
7267 |
+
"grad_norm": 0.2537655532360077,
|
7268 |
+
"learning_rate": 4.433345538689929e-05,
|
7269 |
+
"loss": 0.1794,
|
7270 |
+
"step": 1035
|
7271 |
+
},
|
7272 |
+
{
|
7273 |
+
"epoch": 0.3063842371816199,
|
7274 |
+
"grad_norm": 0.25624555349349976,
|
7275 |
+
"learning_rate": 4.415842243697369e-05,
|
7276 |
+
"loss": 0.2303,
|
7277 |
+
"step": 1036
|
7278 |
+
},
|
7279 |
+
{
|
7280 |
+
"epoch": 0.30667997486229714,
|
7281 |
+
"grad_norm": 0.2514943480491638,
|
7282 |
+
"learning_rate": 4.39836377344583e-05,
|
7283 |
+
"loss": 0.2273,
|
7284 |
+
"step": 1037
|
7285 |
+
},
|
7286 |
+
{
|
7287 |
+
"epoch": 0.3069757125429744,
|
7288 |
+
"grad_norm": 0.28939127922058105,
|
7289 |
+
"learning_rate": 4.380910205637012e-05,
|
7290 |
+
"loss": 0.2312,
|
7291 |
+
"step": 1038
|
7292 |
+
},
|
7293 |
+
{
|
7294 |
+
"epoch": 0.3072714502236516,
|
7295 |
+
"grad_norm": 0.26121294498443604,
|
7296 |
+
"learning_rate": 4.363481617861893e-05,
|
7297 |
+
"loss": 0.1898,
|
7298 |
+
"step": 1039
|
7299 |
+
},
|
7300 |
+
{
|
7301 |
+
"epoch": 0.30756718790432885,
|
7302 |
+
"grad_norm": 0.295492023229599,
|
7303 |
+
"learning_rate": 4.346078087600412e-05,
|
7304 |
+
"loss": 0.2393,
|
7305 |
+
"step": 1040
|
7306 |
+
},
|
7307 |
+
{
|
7308 |
+
"epoch": 0.3078629255850061,
|
7309 |
+
"grad_norm": 0.2641611099243164,
|
7310 |
+
"learning_rate": 4.3286996922211034e-05,
|
7311 |
+
"loss": 0.183,
|
7312 |
+
"step": 1041
|
7313 |
+
},
|
7314 |
+
{
|
7315 |
+
"epoch": 0.30815866326568336,
|
7316 |
+
"grad_norm": 0.2948877513408661,
|
7317 |
+
"learning_rate": 4.311346508980772e-05,
|
7318 |
+
"loss": 0.2088,
|
7319 |
+
"step": 1042
|
7320 |
+
},
|
7321 |
+
{
|
7322 |
+
"epoch": 0.30845440094636056,
|
7323 |
+
"grad_norm": 0.28656068444252014,
|
7324 |
+
"learning_rate": 4.2940186150241365e-05,
|
7325 |
+
"loss": 0.2029,
|
7326 |
+
"step": 1043
|
7327 |
+
},
|
7328 |
+
{
|
7329 |
+
"epoch": 0.3087501386270378,
|
7330 |
+
"grad_norm": 0.2820681929588318,
|
7331 |
+
"learning_rate": 4.27671608738349e-05,
|
7332 |
+
"loss": 0.1992,
|
7333 |
+
"step": 1044
|
7334 |
+
},
|
7335 |
+
{
|
7336 |
+
"epoch": 0.3090458763077151,
|
7337 |
+
"grad_norm": 0.34954750537872314,
|
7338 |
+
"learning_rate": 4.2594390029783534e-05,
|
7339 |
+
"loss": 0.186,
|
7340 |
+
"step": 1045
|
7341 |
+
},
|
7342 |
+
{
|
7343 |
+
"epoch": 0.30934161398839227,
|
7344 |
+
"grad_norm": 0.3151059150695801,
|
7345 |
+
"learning_rate": 4.242187438615153e-05,
|
7346 |
+
"loss": 0.1745,
|
7347 |
+
"step": 1046
|
7348 |
+
},
|
7349 |
+
{
|
7350 |
+
"epoch": 0.3096373516690695,
|
7351 |
+
"grad_norm": 0.3306736350059509,
|
7352 |
+
"learning_rate": 4.224961470986849e-05,
|
7353 |
+
"loss": 0.1497,
|
7354 |
+
"step": 1047
|
7355 |
+
},
|
7356 |
+
{
|
7357 |
+
"epoch": 0.3099330893497468,
|
7358 |
+
"grad_norm": 0.3289806842803955,
|
7359 |
+
"learning_rate": 4.207761176672614e-05,
|
7360 |
+
"loss": 0.1829,
|
7361 |
+
"step": 1048
|
7362 |
+
},
|
7363 |
+
{
|
7364 |
+
"epoch": 0.31022882703042404,
|
7365 |
+
"grad_norm": 0.363712340593338,
|
7366 |
+
"learning_rate": 4.190586632137491e-05,
|
7367 |
+
"loss": 0.2085,
|
7368 |
+
"step": 1049
|
7369 |
+
},
|
7370 |
+
{
|
7371 |
+
"epoch": 0.31052456471110124,
|
7372 |
+
"grad_norm": 0.4890635311603546,
|
7373 |
+
"learning_rate": 4.173437913732048e-05,
|
7374 |
+
"loss": 0.2581,
|
7375 |
+
"step": 1050
|
7376 |
+
},
|
7377 |
+
{
|
7378 |
+
"epoch": 0.3108203023917785,
|
7379 |
+
"grad_norm": 0.10311943292617798,
|
7380 |
+
"learning_rate": 4.156315097692037e-05,
|
7381 |
+
"loss": 0.1254,
|
7382 |
+
"step": 1051
|
7383 |
+
},
|
7384 |
+
{
|
7385 |
+
"epoch": 0.31111604007245575,
|
7386 |
+
"grad_norm": 0.12436298280954361,
|
7387 |
+
"learning_rate": 4.139218260138074e-05,
|
7388 |
+
"loss": 0.1637,
|
7389 |
+
"step": 1052
|
7390 |
+
},
|
7391 |
+
{
|
7392 |
+
"epoch": 0.31141177775313295,
|
7393 |
+
"grad_norm": 0.1472722887992859,
|
7394 |
+
"learning_rate": 4.12214747707527e-05,
|
7395 |
+
"loss": 0.2009,
|
7396 |
+
"step": 1053
|
7397 |
+
},
|
7398 |
+
{
|
7399 |
+
"epoch": 0.3117075154338102,
|
7400 |
+
"grad_norm": 0.1530962437391281,
|
7401 |
+
"learning_rate": 4.1051028243929125e-05,
|
7402 |
+
"loss": 0.1974,
|
7403 |
+
"step": 1054
|
7404 |
+
},
|
7405 |
+
{
|
7406 |
+
"epoch": 0.31200325311448746,
|
7407 |
+
"grad_norm": 0.15057165920734406,
|
7408 |
+
"learning_rate": 4.088084377864135e-05,
|
7409 |
+
"loss": 0.1616,
|
7410 |
+
"step": 1055
|
7411 |
+
},
|
7412 |
+
{
|
7413 |
+
"epoch": 0.3122989907951647,
|
7414 |
+
"grad_norm": 0.17087000608444214,
|
7415 |
+
"learning_rate": 4.07109221314556e-05,
|
7416 |
+
"loss": 0.1957,
|
7417 |
+
"step": 1056
|
7418 |
+
},
|
7419 |
+
{
|
7420 |
+
"epoch": 0.3125947284758419,
|
7421 |
+
"grad_norm": 0.17251181602478027,
|
7422 |
+
"learning_rate": 4.054126405776971e-05,
|
7423 |
+
"loss": 0.1952,
|
7424 |
+
"step": 1057
|
7425 |
+
},
|
7426 |
+
{
|
7427 |
+
"epoch": 0.31289046615651916,
|
7428 |
+
"grad_norm": 0.17809037864208221,
|
7429 |
+
"learning_rate": 4.037187031180985e-05,
|
7430 |
+
"loss": 0.22,
|
7431 |
+
"step": 1058
|
7432 |
+
},
|
7433 |
+
{
|
7434 |
+
"epoch": 0.3131862038371964,
|
7435 |
+
"grad_norm": 0.17678867280483246,
|
7436 |
+
"learning_rate": 4.020274164662707e-05,
|
7437 |
+
"loss": 0.1695,
|
7438 |
+
"step": 1059
|
7439 |
+
},
|
7440 |
+
{
|
7441 |
+
"epoch": 0.3134819415178736,
|
7442 |
+
"grad_norm": 0.17473168671131134,
|
7443 |
+
"learning_rate": 4.003387881409397e-05,
|
7444 |
+
"loss": 0.1947,
|
7445 |
+
"step": 1060
|
7446 |
+
},
|
7447 |
+
{
|
7448 |
+
"epoch": 0.3137776791985509,
|
7449 |
+
"grad_norm": 0.16512729227542877,
|
7450 |
+
"learning_rate": 3.986528256490141e-05,
|
7451 |
+
"loss": 0.1541,
|
7452 |
+
"step": 1061
|
7453 |
+
},
|
7454 |
+
{
|
7455 |
+
"epoch": 0.31407341687922813,
|
7456 |
+
"grad_norm": 0.1978088766336441,
|
7457 |
+
"learning_rate": 3.969695364855511e-05,
|
7458 |
+
"loss": 0.2243,
|
7459 |
+
"step": 1062
|
7460 |
+
},
|
7461 |
+
{
|
7462 |
+
"epoch": 0.3143691545599054,
|
7463 |
+
"grad_norm": 0.2224223017692566,
|
7464 |
+
"learning_rate": 3.952889281337235e-05,
|
7465 |
+
"loss": 0.2512,
|
7466 |
+
"step": 1063
|
7467 |
+
},
|
7468 |
+
{
|
7469 |
+
"epoch": 0.3146648922405826,
|
7470 |
+
"grad_norm": 0.19392409920692444,
|
7471 |
+
"learning_rate": 3.93611008064786e-05,
|
7472 |
+
"loss": 0.1894,
|
7473 |
+
"step": 1064
|
7474 |
+
},
|
7475 |
+
{
|
7476 |
+
"epoch": 0.31496062992125984,
|
7477 |
+
"grad_norm": 0.20266327261924744,
|
7478 |
+
"learning_rate": 3.9193578373804364e-05,
|
7479 |
+
"loss": 0.2205,
|
7480 |
+
"step": 1065
|
7481 |
+
},
|
7482 |
+
{
|
7483 |
+
"epoch": 0.3152563676019371,
|
7484 |
+
"grad_norm": 0.1855594664812088,
|
7485 |
+
"learning_rate": 3.90263262600816e-05,
|
7486 |
+
"loss": 0.189,
|
7487 |
+
"step": 1066
|
7488 |
+
},
|
7489 |
+
{
|
7490 |
+
"epoch": 0.31555210528261435,
|
7491 |
+
"grad_norm": 0.1787533313035965,
|
7492 |
+
"learning_rate": 3.88593452088406e-05,
|
7493 |
+
"loss": 0.1558,
|
7494 |
+
"step": 1067
|
7495 |
+
},
|
7496 |
+
{
|
7497 |
+
"epoch": 0.31584784296329155,
|
7498 |
+
"grad_norm": 0.18954086303710938,
|
7499 |
+
"learning_rate": 3.869263596240661e-05,
|
7500 |
+
"loss": 0.1614,
|
7501 |
+
"step": 1068
|
7502 |
+
},
|
7503 |
+
{
|
7504 |
+
"epoch": 0.3161435806439688,
|
7505 |
+
"grad_norm": 0.1829981952905655,
|
7506 |
+
"learning_rate": 3.8526199261896544e-05,
|
7507 |
+
"loss": 0.1646,
|
7508 |
+
"step": 1069
|
7509 |
+
},
|
7510 |
+
{
|
7511 |
+
"epoch": 0.31643931832464606,
|
7512 |
+
"grad_norm": 0.1618681699037552,
|
7513 |
+
"learning_rate": 3.836003584721577e-05,
|
7514 |
+
"loss": 0.1343,
|
7515 |
+
"step": 1070
|
7516 |
+
},
|
7517 |
+
{
|
7518 |
+
"epoch": 0.31673505600532326,
|
7519 |
+
"grad_norm": 0.1964295357465744,
|
7520 |
+
"learning_rate": 3.8194146457054655e-05,
|
7521 |
+
"loss": 0.1592,
|
7522 |
+
"step": 1071
|
7523 |
+
},
|
7524 |
+
{
|
7525 |
+
"epoch": 0.3170307936860005,
|
7526 |
+
"grad_norm": 0.2202250063419342,
|
7527 |
+
"learning_rate": 3.802853182888543e-05,
|
7528 |
+
"loss": 0.1966,
|
7529 |
+
"step": 1072
|
7530 |
+
},
|
7531 |
+
{
|
7532 |
+
"epoch": 0.31732653136667777,
|
7533 |
+
"grad_norm": 0.18789294362068176,
|
7534 |
+
"learning_rate": 3.786319269895877e-05,
|
7535 |
+
"loss": 0.1788,
|
7536 |
+
"step": 1073
|
7537 |
+
},
|
7538 |
+
{
|
7539 |
+
"epoch": 0.317622269047355,
|
7540 |
+
"grad_norm": 0.21173900365829468,
|
7541 |
+
"learning_rate": 3.769812980230074e-05,
|
7542 |
+
"loss": 0.1972,
|
7543 |
+
"step": 1074
|
7544 |
+
},
|
7545 |
+
{
|
7546 |
+
"epoch": 0.3179180067280322,
|
7547 |
+
"grad_norm": 0.19276420772075653,
|
7548 |
+
"learning_rate": 3.7533343872709294e-05,
|
7549 |
+
"loss": 0.1512,
|
7550 |
+
"step": 1075
|
7551 |
+
},
|
7552 |
+
{
|
7553 |
+
"epoch": 0.3182137444087095,
|
7554 |
+
"grad_norm": 0.20392374694347382,
|
7555 |
+
"learning_rate": 3.736883564275112e-05,
|
7556 |
+
"loss": 0.17,
|
7557 |
+
"step": 1076
|
7558 |
+
},
|
7559 |
+
{
|
7560 |
+
"epoch": 0.31850948208938673,
|
7561 |
+
"grad_norm": 0.2147340625524521,
|
7562 |
+
"learning_rate": 3.7204605843758386e-05,
|
7563 |
+
"loss": 0.1964,
|
7564 |
+
"step": 1077
|
7565 |
+
},
|
7566 |
+
{
|
7567 |
+
"epoch": 0.31880521977006393,
|
7568 |
+
"grad_norm": 0.2067640870809555,
|
7569 |
+
"learning_rate": 3.704065520582549e-05,
|
7570 |
+
"loss": 0.1778,
|
7571 |
+
"step": 1078
|
7572 |
+
},
|
7573 |
+
{
|
7574 |
+
"epoch": 0.3191009574507412,
|
7575 |
+
"grad_norm": 0.2175614833831787,
|
7576 |
+
"learning_rate": 3.6876984457805786e-05,
|
7577 |
+
"loss": 0.1743,
|
7578 |
+
"step": 1079
|
7579 |
+
},
|
7580 |
+
{
|
7581 |
+
"epoch": 0.31939669513141844,
|
7582 |
+
"grad_norm": 0.24874760210514069,
|
7583 |
+
"learning_rate": 3.671359432730834e-05,
|
7584 |
+
"loss": 0.2407,
|
7585 |
+
"step": 1080
|
7586 |
+
},
|
7587 |
+
{
|
7588 |
+
"epoch": 0.3196924328120957,
|
7589 |
+
"grad_norm": 0.2241019457578659,
|
7590 |
+
"learning_rate": 3.655048554069478e-05,
|
7591 |
+
"loss": 0.1543,
|
7592 |
+
"step": 1081
|
7593 |
+
},
|
7594 |
+
{
|
7595 |
+
"epoch": 0.3199881704927729,
|
7596 |
+
"grad_norm": 0.21091891825199127,
|
7597 |
+
"learning_rate": 3.638765882307589e-05,
|
7598 |
+
"loss": 0.1665,
|
7599 |
+
"step": 1082
|
7600 |
+
},
|
7601 |
+
{
|
7602 |
+
"epoch": 0.32028390817345015,
|
7603 |
+
"grad_norm": 0.2465236932039261,
|
7604 |
+
"learning_rate": 3.6225114898308634e-05,
|
7605 |
+
"loss": 0.1946,
|
7606 |
+
"step": 1083
|
7607 |
+
},
|
7608 |
+
{
|
7609 |
+
"epoch": 0.3205796458541274,
|
7610 |
+
"grad_norm": 0.21476277709007263,
|
7611 |
+
"learning_rate": 3.6062854488992714e-05,
|
7612 |
+
"loss": 0.1506,
|
7613 |
+
"step": 1084
|
7614 |
+
},
|
7615 |
+
{
|
7616 |
+
"epoch": 0.3208753835348046,
|
7617 |
+
"grad_norm": 0.21354220807552338,
|
7618 |
+
"learning_rate": 3.5900878316467454e-05,
|
7619 |
+
"loss": 0.1596,
|
7620 |
+
"step": 1085
|
7621 |
+
},
|
7622 |
+
{
|
7623 |
+
"epoch": 0.32117112121548186,
|
7624 |
+
"grad_norm": 0.22661133110523224,
|
7625 |
+
"learning_rate": 3.573918710080857e-05,
|
7626 |
+
"loss": 0.1826,
|
7627 |
+
"step": 1086
|
7628 |
+
},
|
7629 |
+
{
|
7630 |
+
"epoch": 0.3214668588961591,
|
7631 |
+
"grad_norm": 0.27455446124076843,
|
7632 |
+
"learning_rate": 3.5577781560825066e-05,
|
7633 |
+
"loss": 0.2736,
|
7634 |
+
"step": 1087
|
7635 |
+
},
|
7636 |
+
{
|
7637 |
+
"epoch": 0.32176259657683637,
|
7638 |
+
"grad_norm": 0.2345777302980423,
|
7639 |
+
"learning_rate": 3.541666241405588e-05,
|
7640 |
+
"loss": 0.1832,
|
7641 |
+
"step": 1088
|
7642 |
+
},
|
7643 |
+
{
|
7644 |
+
"epoch": 0.32205833425751357,
|
7645 |
+
"grad_norm": 0.2570793628692627,
|
7646 |
+
"learning_rate": 3.5255830376766764e-05,
|
7647 |
+
"loss": 0.1677,
|
7648 |
+
"step": 1089
|
7649 |
+
},
|
7650 |
+
{
|
7651 |
+
"epoch": 0.3223540719381908,
|
7652 |
+
"grad_norm": 0.2677852213382721,
|
7653 |
+
"learning_rate": 3.509528616394716e-05,
|
7654 |
+
"loss": 0.1459,
|
7655 |
+
"step": 1090
|
7656 |
+
},
|
7657 |
+
{
|
7658 |
+
"epoch": 0.3226498096188681,
|
7659 |
+
"grad_norm": 0.264311283826828,
|
7660 |
+
"learning_rate": 3.4935030489306883e-05,
|
7661 |
+
"loss": 0.153,
|
7662 |
+
"step": 1091
|
7663 |
+
},
|
7664 |
+
{
|
7665 |
+
"epoch": 0.3229455472995453,
|
7666 |
+
"grad_norm": 0.23239494860172272,
|
7667 |
+
"learning_rate": 3.4775064065273165e-05,
|
7668 |
+
"loss": 0.1069,
|
7669 |
+
"step": 1092
|
7670 |
+
},
|
7671 |
+
{
|
7672 |
+
"epoch": 0.32324128498022253,
|
7673 |
+
"grad_norm": 0.2711178660392761,
|
7674 |
+
"learning_rate": 3.4615387602987236e-05,
|
7675 |
+
"loss": 0.1556,
|
7676 |
+
"step": 1093
|
7677 |
+
},
|
7678 |
+
{
|
7679 |
+
"epoch": 0.3235370226608998,
|
7680 |
+
"grad_norm": 0.2906439006328583,
|
7681 |
+
"learning_rate": 3.445600181230134e-05,
|
7682 |
+
"loss": 0.1422,
|
7683 |
+
"step": 1094
|
7684 |
+
},
|
7685 |
+
{
|
7686 |
+
"epoch": 0.32383276034157704,
|
7687 |
+
"grad_norm": 0.29046574234962463,
|
7688 |
+
"learning_rate": 3.429690740177549e-05,
|
7689 |
+
"loss": 0.1434,
|
7690 |
+
"step": 1095
|
7691 |
+
},
|
7692 |
+
{
|
7693 |
+
"epoch": 0.32412849802225424,
|
7694 |
+
"grad_norm": 0.3531874418258667,
|
7695 |
+
"learning_rate": 3.413810507867436e-05,
|
7696 |
+
"loss": 0.208,
|
7697 |
+
"step": 1096
|
7698 |
+
},
|
7699 |
+
{
|
7700 |
+
"epoch": 0.3244242357029315,
|
7701 |
+
"grad_norm": 0.409625381231308,
|
7702 |
+
"learning_rate": 3.397959554896415e-05,
|
7703 |
+
"loss": 0.3359,
|
7704 |
+
"step": 1097
|
7705 |
+
},
|
7706 |
+
{
|
7707 |
+
"epoch": 0.32471997338360875,
|
7708 |
+
"grad_norm": 0.3413349390029907,
|
7709 |
+
"learning_rate": 3.3821379517309405e-05,
|
7710 |
+
"loss": 0.1659,
|
7711 |
+
"step": 1098
|
7712 |
+
},
|
7713 |
+
{
|
7714 |
+
"epoch": 0.325015711064286,
|
7715 |
+
"grad_norm": 0.40019068121910095,
|
7716 |
+
"learning_rate": 3.3663457687069924e-05,
|
7717 |
+
"loss": 0.1975,
|
7718 |
+
"step": 1099
|
7719 |
+
},
|
7720 |
+
{
|
7721 |
+
"epoch": 0.3253114487449632,
|
7722 |
+
"grad_norm": 0.44681063294410706,
|
7723 |
+
"learning_rate": 3.350583076029754e-05,
|
7724 |
+
"loss": 0.2325,
|
7725 |
+
"step": 1100
|
7726 |
+
},
|
7727 |
+
{
|
7728 |
+
"epoch": 0.32560718642564046,
|
7729 |
+
"grad_norm": 0.10489065200090408,
|
7730 |
+
"learning_rate": 3.334849943773323e-05,
|
7731 |
+
"loss": 0.1379,
|
7732 |
+
"step": 1101
|
7733 |
+
},
|
7734 |
+
{
|
7735 |
+
"epoch": 0.3259029241063177,
|
7736 |
+
"grad_norm": 0.13087670505046844,
|
7737 |
+
"learning_rate": 3.319146441880371e-05,
|
7738 |
+
"loss": 0.1632,
|
7739 |
+
"step": 1102
|
7740 |
+
},
|
7741 |
+
{
|
7742 |
+
"epoch": 0.3261986617869949,
|
7743 |
+
"grad_norm": 0.14324639737606049,
|
7744 |
+
"learning_rate": 3.3034726401618444e-05,
|
7745 |
+
"loss": 0.1943,
|
7746 |
+
"step": 1103
|
7747 |
+
},
|
7748 |
+
{
|
7749 |
+
"epoch": 0.32649439946767217,
|
7750 |
+
"grad_norm": 0.13718920946121216,
|
7751 |
+
"learning_rate": 3.28782860829667e-05,
|
7752 |
+
"loss": 0.1442,
|
7753 |
+
"step": 1104
|
7754 |
+
},
|
7755 |
+
{
|
7756 |
+
"epoch": 0.3267901371483494,
|
7757 |
+
"grad_norm": 0.17169281840324402,
|
7758 |
+
"learning_rate": 3.272214415831418e-05,
|
7759 |
+
"loss": 0.1917,
|
7760 |
+
"step": 1105
|
7761 |
+
},
|
7762 |
+
{
|
7763 |
+
"epoch": 0.3270858748290267,
|
7764 |
+
"grad_norm": 0.15564316511154175,
|
7765 |
+
"learning_rate": 3.2566301321800085e-05,
|
7766 |
+
"loss": 0.1558,
|
7767 |
+
"step": 1106
|
7768 |
+
},
|
7769 |
+
{
|
7770 |
+
"epoch": 0.3273816125097039,
|
7771 |
+
"grad_norm": 0.1527709662914276,
|
7772 |
+
"learning_rate": 3.241075826623401e-05,
|
7773 |
+
"loss": 0.2016,
|
7774 |
+
"step": 1107
|
7775 |
+
},
|
7776 |
+
{
|
7777 |
+
"epoch": 0.32767735019038113,
|
7778 |
+
"grad_norm": 0.18245702981948853,
|
7779 |
+
"learning_rate": 3.225551568309284e-05,
|
7780 |
+
"loss": 0.1997,
|
7781 |
+
"step": 1108
|
7782 |
+
},
|
7783 |
+
{
|
7784 |
+
"epoch": 0.3279730878710584,
|
7785 |
+
"grad_norm": 0.14350612461566925,
|
7786 |
+
"learning_rate": 3.210057426251773e-05,
|
7787 |
+
"loss": 0.1301,
|
7788 |
+
"step": 1109
|
7789 |
+
},
|
7790 |
+
{
|
7791 |
+
"epoch": 0.3282688255517356,
|
7792 |
+
"grad_norm": 0.1906651109457016,
|
7793 |
+
"learning_rate": 3.1945934693310896e-05,
|
7794 |
+
"loss": 0.1982,
|
7795 |
+
"step": 1110
|
7796 |
+
},
|
7797 |
+
{
|
7798 |
+
"epoch": 0.32856456323241284,
|
7799 |
+
"grad_norm": 0.21341390907764435,
|
7800 |
+
"learning_rate": 3.179159766293282e-05,
|
7801 |
+
"loss": 0.2742,
|
7802 |
+
"step": 1111
|
7803 |
+
},
|
7804 |
+
{
|
7805 |
+
"epoch": 0.3288603009130901,
|
7806 |
+
"grad_norm": 0.18190768361091614,
|
7807 |
+
"learning_rate": 3.163756385749889e-05,
|
7808 |
+
"loss": 0.1947,
|
7809 |
+
"step": 1112
|
7810 |
+
},
|
7811 |
+
{
|
7812 |
+
"epoch": 0.32915603859376735,
|
7813 |
+
"grad_norm": 0.1745702177286148,
|
7814 |
+
"learning_rate": 3.148383396177653e-05,
|
7815 |
+
"loss": 0.171,
|
7816 |
+
"step": 1113
|
7817 |
+
},
|
7818 |
+
{
|
7819 |
+
"epoch": 0.32945177627444455,
|
7820 |
+
"grad_norm": 0.18562626838684082,
|
7821 |
+
"learning_rate": 3.133040865918213e-05,
|
7822 |
+
"loss": 0.1952,
|
7823 |
+
"step": 1114
|
7824 |
+
},
|
7825 |
+
{
|
7826 |
+
"epoch": 0.3297475139551218,
|
7827 |
+
"grad_norm": 0.18852019309997559,
|
7828 |
+
"learning_rate": 3.117728863177796e-05,
|
7829 |
+
"loss": 0.1815,
|
7830 |
+
"step": 1115
|
7831 |
+
},
|
7832 |
+
{
|
7833 |
+
"epoch": 0.33004325163579906,
|
7834 |
+
"grad_norm": 0.194679394364357,
|
7835 |
+
"learning_rate": 3.102447456026919e-05,
|
7836 |
+
"loss": 0.1826,
|
7837 |
+
"step": 1116
|
7838 |
+
},
|
7839 |
+
{
|
7840 |
+
"epoch": 0.33033898931647626,
|
7841 |
+
"grad_norm": 0.21376194059848785,
|
7842 |
+
"learning_rate": 3.0871967124000834e-05,
|
7843 |
+
"loss": 0.2233,
|
7844 |
+
"step": 1117
|
7845 |
+
},
|
7846 |
+
{
|
7847 |
+
"epoch": 0.3306347269971535,
|
7848 |
+
"grad_norm": 0.2095218151807785,
|
7849 |
+
"learning_rate": 3.0719767000954714e-05,
|
7850 |
+
"loss": 0.1997,
|
7851 |
+
"step": 1118
|
7852 |
+
},
|
7853 |
+
{
|
7854 |
+
"epoch": 0.33093046467783077,
|
7855 |
+
"grad_norm": 0.19964182376861572,
|
7856 |
+
"learning_rate": 3.056787486774656e-05,
|
7857 |
+
"loss": 0.1715,
|
7858 |
+
"step": 1119
|
7859 |
+
},
|
7860 |
+
{
|
7861 |
+
"epoch": 0.331226202358508,
|
7862 |
+
"grad_norm": 0.19055882096290588,
|
7863 |
+
"learning_rate": 3.041629139962283e-05,
|
7864 |
+
"loss": 0.1629,
|
7865 |
+
"step": 1120
|
7866 |
+
},
|
7867 |
+
{
|
7868 |
+
"epoch": 0.3315219400391852,
|
7869 |
+
"grad_norm": 0.17358700931072235,
|
7870 |
+
"learning_rate": 3.0265017270457775e-05,
|
7871 |
+
"loss": 0.11,
|
7872 |
+
"step": 1121
|
7873 |
+
},
|
7874 |
+
{
|
7875 |
+
"epoch": 0.3318176777198625,
|
7876 |
+
"grad_norm": 0.1570686399936676,
|
7877 |
+
"learning_rate": 3.0114053152750556e-05,
|
7878 |
+
"loss": 0.1093,
|
7879 |
+
"step": 1122
|
7880 |
+
},
|
7881 |
+
{
|
7882 |
+
"epoch": 0.33211341540053974,
|
7883 |
+
"grad_norm": 0.2210582047700882,
|
7884 |
+
"learning_rate": 2.9963399717622077e-05,
|
7885 |
+
"loss": 0.1951,
|
7886 |
+
"step": 1123
|
7887 |
+
},
|
7888 |
+
{
|
7889 |
+
"epoch": 0.33240915308121693,
|
7890 |
+
"grad_norm": 0.19304311275482178,
|
7891 |
+
"learning_rate": 2.98130576348121e-05,
|
7892 |
+
"loss": 0.1603,
|
7893 |
+
"step": 1124
|
7894 |
+
},
|
7895 |
+
{
|
7896 |
+
"epoch": 0.3327048907618942,
|
7897 |
+
"grad_norm": 0.23825351893901825,
|
7898 |
+
"learning_rate": 2.966302757267625e-05,
|
7899 |
+
"loss": 0.2385,
|
7900 |
+
"step": 1125
|
7901 |
+
},
|
7902 |
+
{
|
7903 |
+
"epoch": 0.3327048907618942,
|
7904 |
+
"eval_loss": 0.18209633231163025,
|
7905 |
+
"eval_runtime": 323.5083,
|
7906 |
+
"eval_samples_per_second": 17.604,
|
7907 |
+
"eval_steps_per_second": 8.803,
|
7908 |
+
"step": 1125
|
7909 |
}
|
7910 |
],
|
7911 |
"logging_steps": 1,
|
|
|
7925 |
"attributes": {}
|
7926 |
}
|
7927 |
},
|
7928 |
+
"total_flos": 1.4886246305696317e+18,
|
7929 |
"train_batch_size": 2,
|
7930 |
"trial_name": null,
|
7931 |
"trial_params": null
|