Training in progress, step 773, 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 1134088
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0dfe6177dd8e99f8251ddff2bd7c294f0cee74189477b02d858b7d75e148fab5
|
| 3 |
size 1134088
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 608186
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:45b3328583c18197b000ed09534cc64ce0599779b8718ca17d5cddb71214c0d7
|
| 3 |
size 608186
|
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:c8d25b65c8871ace8f29fc01c33af8c9535181702318707d9198298d9aa6cccf
|
| 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 sha256:45719461b132f33086549160c158249554ace1a4b2303f17dcf1d7d8dd8db700
|
| 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":
|
| 5 |
"eval_steps": 155,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -4113,6 +4113,1351 @@
|
|
| 4113 |
"learning_rate": 1.4681072873881312e-05,
|
| 4114 |
"loss": 1.0705,
|
| 4115 |
"step": 582
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4116 |
}
|
| 4117 |
],
|
| 4118 |
"logging_steps": 1,
|
|
@@ -4127,12 +5472,12 @@
|
|
| 4127 |
"should_evaluate": false,
|
| 4128 |
"should_log": false,
|
| 4129 |
"should_save": true,
|
| 4130 |
-
"should_training_stop":
|
| 4131 |
},
|
| 4132 |
"attributes": {}
|
| 4133 |
}
|
| 4134 |
},
|
| 4135 |
-
"total_flos":
|
| 4136 |
"train_batch_size": 2,
|
| 4137 |
"trial_name": null,
|
| 4138 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0009711880867596,
|
| 5 |
"eval_steps": 155,
|
| 6 |
+
"global_step": 773,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 4113 |
"learning_rate": 1.4681072873881312e-05,
|
| 4114 |
"loss": 1.0705,
|
| 4115 |
"step": 582
|
| 4116 |
+
},
|
| 4117 |
+
{
|
| 4118 |
+
"epoch": 0.7549368727743606,
|
| 4119 |
+
"grad_norm": 5.166491508483887,
|
| 4120 |
+
"learning_rate": 1.4535650064427003e-05,
|
| 4121 |
+
"loss": 0.6913,
|
| 4122 |
+
"step": 583
|
| 4123 |
+
},
|
| 4124 |
+
{
|
| 4125 |
+
"epoch": 0.7562317902233733,
|
| 4126 |
+
"grad_norm": 13.889178276062012,
|
| 4127 |
+
"learning_rate": 1.439082848693406e-05,
|
| 4128 |
+
"loss": 3.66,
|
| 4129 |
+
"step": 584
|
| 4130 |
+
},
|
| 4131 |
+
{
|
| 4132 |
+
"epoch": 0.7575267076723858,
|
| 4133 |
+
"grad_norm": 2.9717559814453125,
|
| 4134 |
+
"learning_rate": 1.4246610596582444e-05,
|
| 4135 |
+
"loss": 0.9647,
|
| 4136 |
+
"step": 585
|
| 4137 |
+
},
|
| 4138 |
+
{
|
| 4139 |
+
"epoch": 0.7588216251213985,
|
| 4140 |
+
"grad_norm": 3.951253652572632,
|
| 4141 |
+
"learning_rate": 1.41029988383177e-05,
|
| 4142 |
+
"loss": 0.7419,
|
| 4143 |
+
"step": 586
|
| 4144 |
+
},
|
| 4145 |
+
{
|
| 4146 |
+
"epoch": 0.7601165425704112,
|
| 4147 |
+
"grad_norm": 2.8635647296905518,
|
| 4148 |
+
"learning_rate": 1.3959995646809549e-05,
|
| 4149 |
+
"loss": 0.614,
|
| 4150 |
+
"step": 587
|
| 4151 |
+
},
|
| 4152 |
+
{
|
| 4153 |
+
"epoch": 0.7614114600194237,
|
| 4154 |
+
"grad_norm": 4.437415599822998,
|
| 4155 |
+
"learning_rate": 1.381760344641061e-05,
|
| 4156 |
+
"loss": 0.7816,
|
| 4157 |
+
"step": 588
|
| 4158 |
+
},
|
| 4159 |
+
{
|
| 4160 |
+
"epoch": 0.7627063774684364,
|
| 4161 |
+
"grad_norm": 3.6199607849121094,
|
| 4162 |
+
"learning_rate": 1.3675824651115276e-05,
|
| 4163 |
+
"loss": 0.659,
|
| 4164 |
+
"step": 589
|
| 4165 |
+
},
|
| 4166 |
+
{
|
| 4167 |
+
"epoch": 0.7640012949174491,
|
| 4168 |
+
"grad_norm": 6.909400939941406,
|
| 4169 |
+
"learning_rate": 1.3534661664518817e-05,
|
| 4170 |
+
"loss": 1.5943,
|
| 4171 |
+
"step": 590
|
| 4172 |
+
},
|
| 4173 |
+
{
|
| 4174 |
+
"epoch": 0.7652962123664616,
|
| 4175 |
+
"grad_norm": 6.746818542480469,
|
| 4176 |
+
"learning_rate": 1.339411687977657e-05,
|
| 4177 |
+
"loss": 1.505,
|
| 4178 |
+
"step": 591
|
| 4179 |
+
},
|
| 4180 |
+
{
|
| 4181 |
+
"epoch": 0.7665911298154743,
|
| 4182 |
+
"grad_norm": 4.21103048324585,
|
| 4183 |
+
"learning_rate": 1.325419267956346e-05,
|
| 4184 |
+
"loss": 0.4796,
|
| 4185 |
+
"step": 592
|
| 4186 |
+
},
|
| 4187 |
+
{
|
| 4188 |
+
"epoch": 0.7678860472644868,
|
| 4189 |
+
"grad_norm": 4.808207988739014,
|
| 4190 |
+
"learning_rate": 1.3114891436033522e-05,
|
| 4191 |
+
"loss": 0.7325,
|
| 4192 |
+
"step": 593
|
| 4193 |
+
},
|
| 4194 |
+
{
|
| 4195 |
+
"epoch": 0.7691809647134995,
|
| 4196 |
+
"grad_norm": 5.3486833572387695,
|
| 4197 |
+
"learning_rate": 1.2976215510779755e-05,
|
| 4198 |
+
"loss": 0.4682,
|
| 4199 |
+
"step": 594
|
| 4200 |
+
},
|
| 4201 |
+
{
|
| 4202 |
+
"epoch": 0.7704758821625122,
|
| 4203 |
+
"grad_norm": 7.150143623352051,
|
| 4204 |
+
"learning_rate": 1.2838167254794004e-05,
|
| 4205 |
+
"loss": 0.8242,
|
| 4206 |
+
"step": 595
|
| 4207 |
+
},
|
| 4208 |
+
{
|
| 4209 |
+
"epoch": 0.7717707996115247,
|
| 4210 |
+
"grad_norm": 7.019624710083008,
|
| 4211 |
+
"learning_rate": 1.2700749008427205e-05,
|
| 4212 |
+
"loss": 1.0202,
|
| 4213 |
+
"step": 596
|
| 4214 |
+
},
|
| 4215 |
+
{
|
| 4216 |
+
"epoch": 0.7730657170605374,
|
| 4217 |
+
"grad_norm": 5.221418857574463,
|
| 4218 |
+
"learning_rate": 1.2563963101349619e-05,
|
| 4219 |
+
"loss": 1.5878,
|
| 4220 |
+
"step": 597
|
| 4221 |
+
},
|
| 4222 |
+
{
|
| 4223 |
+
"epoch": 0.7743606345095501,
|
| 4224 |
+
"grad_norm": 2.91672682762146,
|
| 4225 |
+
"learning_rate": 1.2427811852511395e-05,
|
| 4226 |
+
"loss": 0.4305,
|
| 4227 |
+
"step": 598
|
| 4228 |
+
},
|
| 4229 |
+
{
|
| 4230 |
+
"epoch": 0.7756555519585626,
|
| 4231 |
+
"grad_norm": 9.53405475616455,
|
| 4232 |
+
"learning_rate": 1.2292297570103229e-05,
|
| 4233 |
+
"loss": 1.2377,
|
| 4234 |
+
"step": 599
|
| 4235 |
+
},
|
| 4236 |
+
{
|
| 4237 |
+
"epoch": 0.7769504694075753,
|
| 4238 |
+
"grad_norm": 4.454233646392822,
|
| 4239 |
+
"learning_rate": 1.2157422551517228e-05,
|
| 4240 |
+
"loss": 0.7973,
|
| 4241 |
+
"step": 600
|
| 4242 |
+
},
|
| 4243 |
+
{
|
| 4244 |
+
"epoch": 0.7782453868565878,
|
| 4245 |
+
"grad_norm": 6.495676040649414,
|
| 4246 |
+
"learning_rate": 1.202318908330795e-05,
|
| 4247 |
+
"loss": 1.1238,
|
| 4248 |
+
"step": 601
|
| 4249 |
+
},
|
| 4250 |
+
{
|
| 4251 |
+
"epoch": 0.7795403043056005,
|
| 4252 |
+
"grad_norm": 5.350216388702393,
|
| 4253 |
+
"learning_rate": 1.188959944115372e-05,
|
| 4254 |
+
"loss": 0.8506,
|
| 4255 |
+
"step": 602
|
| 4256 |
+
},
|
| 4257 |
+
{
|
| 4258 |
+
"epoch": 0.7808352217546132,
|
| 4259 |
+
"grad_norm": 3.3894073963165283,
|
| 4260 |
+
"learning_rate": 1.1756655889817953e-05,
|
| 4261 |
+
"loss": 0.9475,
|
| 4262 |
+
"step": 603
|
| 4263 |
+
},
|
| 4264 |
+
{
|
| 4265 |
+
"epoch": 0.7821301392036257,
|
| 4266 |
+
"grad_norm": 4.518886089324951,
|
| 4267 |
+
"learning_rate": 1.1624360683110819e-05,
|
| 4268 |
+
"loss": 1.1395,
|
| 4269 |
+
"step": 604
|
| 4270 |
+
},
|
| 4271 |
+
{
|
| 4272 |
+
"epoch": 0.7834250566526384,
|
| 4273 |
+
"grad_norm": 10.47011947631836,
|
| 4274 |
+
"learning_rate": 1.1492716063850973e-05,
|
| 4275 |
+
"loss": 1.136,
|
| 4276 |
+
"step": 605
|
| 4277 |
+
},
|
| 4278 |
+
{
|
| 4279 |
+
"epoch": 0.7847199741016511,
|
| 4280 |
+
"grad_norm": 5.102293491363525,
|
| 4281 |
+
"learning_rate": 1.1361724263827633e-05,
|
| 4282 |
+
"loss": 0.9341,
|
| 4283 |
+
"step": 606
|
| 4284 |
+
},
|
| 4285 |
+
{
|
| 4286 |
+
"epoch": 0.7860148915506636,
|
| 4287 |
+
"grad_norm": 4.892009258270264,
|
| 4288 |
+
"learning_rate": 1.123138750376262e-05,
|
| 4289 |
+
"loss": 1.017,
|
| 4290 |
+
"step": 607
|
| 4291 |
+
},
|
| 4292 |
+
{
|
| 4293 |
+
"epoch": 0.7873098089996763,
|
| 4294 |
+
"grad_norm": 5.155287265777588,
|
| 4295 |
+
"learning_rate": 1.1101707993272825e-05,
|
| 4296 |
+
"loss": 1.0451,
|
| 4297 |
+
"step": 608
|
| 4298 |
+
},
|
| 4299 |
+
{
|
| 4300 |
+
"epoch": 0.7886047264486888,
|
| 4301 |
+
"grad_norm": 8.183491706848145,
|
| 4302 |
+
"learning_rate": 1.097268793083266e-05,
|
| 4303 |
+
"loss": 1.3968,
|
| 4304 |
+
"step": 609
|
| 4305 |
+
},
|
| 4306 |
+
{
|
| 4307 |
+
"epoch": 0.7898996438977015,
|
| 4308 |
+
"grad_norm": 3.7006969451904297,
|
| 4309 |
+
"learning_rate": 1.084432950373685e-05,
|
| 4310 |
+
"loss": 0.9212,
|
| 4311 |
+
"step": 610
|
| 4312 |
+
},
|
| 4313 |
+
{
|
| 4314 |
+
"epoch": 0.7911945613467142,
|
| 4315 |
+
"grad_norm": 4.199628829956055,
|
| 4316 |
+
"learning_rate": 1.071663488806331e-05,
|
| 4317 |
+
"loss": 0.9636,
|
| 4318 |
+
"step": 611
|
| 4319 |
+
},
|
| 4320 |
+
{
|
| 4321 |
+
"epoch": 0.7924894787957267,
|
| 4322 |
+
"grad_norm": 3.8933045864105225,
|
| 4323 |
+
"learning_rate": 1.0589606248636292e-05,
|
| 4324 |
+
"loss": 0.4748,
|
| 4325 |
+
"step": 612
|
| 4326 |
+
},
|
| 4327 |
+
{
|
| 4328 |
+
"epoch": 0.7937843962447394,
|
| 4329 |
+
"grad_norm": 3.9218475818634033,
|
| 4330 |
+
"learning_rate": 1.0463245738989636e-05,
|
| 4331 |
+
"loss": 0.7236,
|
| 4332 |
+
"step": 613
|
| 4333 |
+
},
|
| 4334 |
+
{
|
| 4335 |
+
"epoch": 0.7950793136937521,
|
| 4336 |
+
"grad_norm": 5.374284744262695,
|
| 4337 |
+
"learning_rate": 1.0337555501330281e-05,
|
| 4338 |
+
"loss": 1.0386,
|
| 4339 |
+
"step": 614
|
| 4340 |
+
},
|
| 4341 |
+
{
|
| 4342 |
+
"epoch": 0.7963742311427646,
|
| 4343 |
+
"grad_norm": 5.497282028198242,
|
| 4344 |
+
"learning_rate": 1.0212537666501976e-05,
|
| 4345 |
+
"loss": 0.9231,
|
| 4346 |
+
"step": 615
|
| 4347 |
+
},
|
| 4348 |
+
{
|
| 4349 |
+
"epoch": 0.7976691485917773,
|
| 4350 |
+
"grad_norm": 7.081708908081055,
|
| 4351 |
+
"learning_rate": 1.0088194353949137e-05,
|
| 4352 |
+
"loss": 0.93,
|
| 4353 |
+
"step": 616
|
| 4354 |
+
},
|
| 4355 |
+
{
|
| 4356 |
+
"epoch": 0.7989640660407898,
|
| 4357 |
+
"grad_norm": 4.324736595153809,
|
| 4358 |
+
"learning_rate": 9.96452767168089e-06,
|
| 4359 |
+
"loss": 1.1612,
|
| 4360 |
+
"step": 617
|
| 4361 |
+
},
|
| 4362 |
+
{
|
| 4363 |
+
"epoch": 0.8002589834898025,
|
| 4364 |
+
"grad_norm": 5.2725605964660645,
|
| 4365 |
+
"learning_rate": 9.841539716235387e-06,
|
| 4366 |
+
"loss": 0.7365,
|
| 4367 |
+
"step": 618
|
| 4368 |
+
},
|
| 4369 |
+
{
|
| 4370 |
+
"epoch": 0.8015539009388152,
|
| 4371 |
+
"grad_norm": 6.695886135101318,
|
| 4372 |
+
"learning_rate": 9.719232572644187e-06,
|
| 4373 |
+
"loss": 2.2744,
|
| 4374 |
+
"step": 619
|
| 4375 |
+
},
|
| 4376 |
+
{
|
| 4377 |
+
"epoch": 0.8028488183878277,
|
| 4378 |
+
"grad_norm": 2.6052026748657227,
|
| 4379 |
+
"learning_rate": 9.597608314396978e-06,
|
| 4380 |
+
"loss": 0.4759,
|
| 4381 |
+
"step": 620
|
| 4382 |
+
},
|
| 4383 |
+
{
|
| 4384 |
+
"epoch": 0.8028488183878277,
|
| 4385 |
+
"eval_loss": 1.2390819787979126,
|
| 4386 |
+
"eval_runtime": 2.0056,
|
| 4387 |
+
"eval_samples_per_second": 162.545,
|
| 4388 |
+
"eval_steps_per_second": 81.272,
|
| 4389 |
+
"step": 620
|
| 4390 |
+
},
|
| 4391 |
+
{
|
| 4392 |
+
"epoch": 0.8041437358368404,
|
| 4393 |
+
"grad_norm": 4.534181594848633,
|
| 4394 |
+
"learning_rate": 9.476669003406403e-06,
|
| 4395 |
+
"loss": 1.3455,
|
| 4396 |
+
"step": 621
|
| 4397 |
+
},
|
| 4398 |
+
{
|
| 4399 |
+
"epoch": 0.8054386532858531,
|
| 4400 |
+
"grad_norm": 6.6253252029418945,
|
| 4401 |
+
"learning_rate": 9.356416689973108e-06,
|
| 4402 |
+
"loss": 1.6427,
|
| 4403 |
+
"step": 622
|
| 4404 |
+
},
|
| 4405 |
+
{
|
| 4406 |
+
"epoch": 0.8067335707348656,
|
| 4407 |
+
"grad_norm": 4.596927642822266,
|
| 4408 |
+
"learning_rate": 9.236853412750935e-06,
|
| 4409 |
+
"loss": 0.9381,
|
| 4410 |
+
"step": 623
|
| 4411 |
+
},
|
| 4412 |
+
{
|
| 4413 |
+
"epoch": 0.8080284881838783,
|
| 4414 |
+
"grad_norm": 12.092345237731934,
|
| 4415 |
+
"learning_rate": 9.11798119871245e-06,
|
| 4416 |
+
"loss": 3.2316,
|
| 4417 |
+
"step": 624
|
| 4418 |
+
},
|
| 4419 |
+
{
|
| 4420 |
+
"epoch": 0.809323405632891,
|
| 4421 |
+
"grad_norm": 3.8456172943115234,
|
| 4422 |
+
"learning_rate": 8.99980206311452e-06,
|
| 4423 |
+
"loss": 0.7777,
|
| 4424 |
+
"step": 625
|
| 4425 |
+
},
|
| 4426 |
+
{
|
| 4427 |
+
"epoch": 0.8106183230819035,
|
| 4428 |
+
"grad_norm": 4.333143711090088,
|
| 4429 |
+
"learning_rate": 8.882318009464125e-06,
|
| 4430 |
+
"loss": 0.8885,
|
| 4431 |
+
"step": 626
|
| 4432 |
+
},
|
| 4433 |
+
{
|
| 4434 |
+
"epoch": 0.8119132405309162,
|
| 4435 |
+
"grad_norm": 6.766184329986572,
|
| 4436 |
+
"learning_rate": 8.765531029484476e-06,
|
| 4437 |
+
"loss": 0.5351,
|
| 4438 |
+
"step": 627
|
| 4439 |
+
},
|
| 4440 |
+
{
|
| 4441 |
+
"epoch": 0.8132081579799287,
|
| 4442 |
+
"grad_norm": 5.541638374328613,
|
| 4443 |
+
"learning_rate": 8.64944310308114e-06,
|
| 4444 |
+
"loss": 1.9105,
|
| 4445 |
+
"step": 628
|
| 4446 |
+
},
|
| 4447 |
+
{
|
| 4448 |
+
"epoch": 0.8145030754289414,
|
| 4449 |
+
"grad_norm": 9.491156578063965,
|
| 4450 |
+
"learning_rate": 8.534056198308582e-06,
|
| 4451 |
+
"loss": 2.0113,
|
| 4452 |
+
"step": 629
|
| 4453 |
+
},
|
| 4454 |
+
{
|
| 4455 |
+
"epoch": 0.8157979928779541,
|
| 4456 |
+
"grad_norm": 4.046790599822998,
|
| 4457 |
+
"learning_rate": 8.419372271336745e-06,
|
| 4458 |
+
"loss": 0.7802,
|
| 4459 |
+
"step": 630
|
| 4460 |
+
},
|
| 4461 |
+
{
|
| 4462 |
+
"epoch": 0.8170929103269666,
|
| 4463 |
+
"grad_norm": 4.86843204498291,
|
| 4464 |
+
"learning_rate": 8.305393266417887e-06,
|
| 4465 |
+
"loss": 1.4356,
|
| 4466 |
+
"step": 631
|
| 4467 |
+
},
|
| 4468 |
+
{
|
| 4469 |
+
"epoch": 0.8183878277759793,
|
| 4470 |
+
"grad_norm": 5.277727127075195,
|
| 4471 |
+
"learning_rate": 8.192121115853602e-06,
|
| 4472 |
+
"loss": 0.5158,
|
| 4473 |
+
"step": 632
|
| 4474 |
+
},
|
| 4475 |
+
{
|
| 4476 |
+
"epoch": 0.819682745224992,
|
| 4477 |
+
"grad_norm": 4.455538272857666,
|
| 4478 |
+
"learning_rate": 8.079557739962128e-06,
|
| 4479 |
+
"loss": 0.6981,
|
| 4480 |
+
"step": 633
|
| 4481 |
+
},
|
| 4482 |
+
{
|
| 4483 |
+
"epoch": 0.8209776626740045,
|
| 4484 |
+
"grad_norm": 5.631432056427002,
|
| 4485 |
+
"learning_rate": 7.967705047045715e-06,
|
| 4486 |
+
"loss": 0.7477,
|
| 4487 |
+
"step": 634
|
| 4488 |
+
},
|
| 4489 |
+
{
|
| 4490 |
+
"epoch": 0.8222725801230172,
|
| 4491 |
+
"grad_norm": 5.253812789916992,
|
| 4492 |
+
"learning_rate": 7.856564933358324e-06,
|
| 4493 |
+
"loss": 0.9068,
|
| 4494 |
+
"step": 635
|
| 4495 |
+
},
|
| 4496 |
+
{
|
| 4497 |
+
"epoch": 0.8235674975720297,
|
| 4498 |
+
"grad_norm": 10.097738265991211,
|
| 4499 |
+
"learning_rate": 7.746139283073473e-06,
|
| 4500 |
+
"loss": 1.5746,
|
| 4501 |
+
"step": 636
|
| 4502 |
+
},
|
| 4503 |
+
{
|
| 4504 |
+
"epoch": 0.8248624150210424,
|
| 4505 |
+
"grad_norm": 4.023595333099365,
|
| 4506 |
+
"learning_rate": 7.636429968252257e-06,
|
| 4507 |
+
"loss": 0.9911,
|
| 4508 |
+
"step": 637
|
| 4509 |
+
},
|
| 4510 |
+
{
|
| 4511 |
+
"epoch": 0.8261573324700551,
|
| 4512 |
+
"grad_norm": 6.050657749176025,
|
| 4513 |
+
"learning_rate": 7.527438848811652e-06,
|
| 4514 |
+
"loss": 1.0283,
|
| 4515 |
+
"step": 638
|
| 4516 |
+
},
|
| 4517 |
+
{
|
| 4518 |
+
"epoch": 0.8274522499190676,
|
| 4519 |
+
"grad_norm": 6.443034648895264,
|
| 4520 |
+
"learning_rate": 7.4191677724929906e-06,
|
| 4521 |
+
"loss": 2.4053,
|
| 4522 |
+
"step": 639
|
| 4523 |
+
},
|
| 4524 |
+
{
|
| 4525 |
+
"epoch": 0.8287471673680803,
|
| 4526 |
+
"grad_norm": 9.725380897521973,
|
| 4527 |
+
"learning_rate": 7.31161857483057e-06,
|
| 4528 |
+
"loss": 1.0468,
|
| 4529 |
+
"step": 640
|
| 4530 |
+
},
|
| 4531 |
+
{
|
| 4532 |
+
"epoch": 0.830042084817093,
|
| 4533 |
+
"grad_norm": 5.43861722946167,
|
| 4534 |
+
"learning_rate": 7.204793079120636e-06,
|
| 4535 |
+
"loss": 1.044,
|
| 4536 |
+
"step": 641
|
| 4537 |
+
},
|
| 4538 |
+
{
|
| 4539 |
+
"epoch": 0.8313370022661055,
|
| 4540 |
+
"grad_norm": 4.303345680236816,
|
| 4541 |
+
"learning_rate": 7.0986930963903575e-06,
|
| 4542 |
+
"loss": 0.9896,
|
| 4543 |
+
"step": 642
|
| 4544 |
+
},
|
| 4545 |
+
{
|
| 4546 |
+
"epoch": 0.8326319197151182,
|
| 4547 |
+
"grad_norm": 5.471158981323242,
|
| 4548 |
+
"learning_rate": 6.993320425367222e-06,
|
| 4549 |
+
"loss": 1.501,
|
| 4550 |
+
"step": 643
|
| 4551 |
+
},
|
| 4552 |
+
{
|
| 4553 |
+
"epoch": 0.8339268371641307,
|
| 4554 |
+
"grad_norm": 4.747450828552246,
|
| 4555 |
+
"learning_rate": 6.8886768524485e-06,
|
| 4556 |
+
"loss": 0.6522,
|
| 4557 |
+
"step": 644
|
| 4558 |
+
},
|
| 4559 |
+
{
|
| 4560 |
+
"epoch": 0.8352217546131434,
|
| 4561 |
+
"grad_norm": 4.842288494110107,
|
| 4562 |
+
"learning_rate": 6.7847641516709635e-06,
|
| 4563 |
+
"loss": 0.819,
|
| 4564 |
+
"step": 645
|
| 4565 |
+
},
|
| 4566 |
+
{
|
| 4567 |
+
"epoch": 0.8365166720621561,
|
| 4568 |
+
"grad_norm": 8.534642219543457,
|
| 4569 |
+
"learning_rate": 6.681584084680787e-06,
|
| 4570 |
+
"loss": 0.7393,
|
| 4571 |
+
"step": 646
|
| 4572 |
+
},
|
| 4573 |
+
{
|
| 4574 |
+
"epoch": 0.8378115895111686,
|
| 4575 |
+
"grad_norm": 6.525353908538818,
|
| 4576 |
+
"learning_rate": 6.579138400703716e-06,
|
| 4577 |
+
"loss": 1.0387,
|
| 4578 |
+
"step": 647
|
| 4579 |
+
},
|
| 4580 |
+
{
|
| 4581 |
+
"epoch": 0.8391065069601813,
|
| 4582 |
+
"grad_norm": 6.289246082305908,
|
| 4583 |
+
"learning_rate": 6.4774288365154035e-06,
|
| 4584 |
+
"loss": 1.2105,
|
| 4585 |
+
"step": 648
|
| 4586 |
+
},
|
| 4587 |
+
{
|
| 4588 |
+
"epoch": 0.840401424409194,
|
| 4589 |
+
"grad_norm": 3.268260955810547,
|
| 4590 |
+
"learning_rate": 6.376457116411971e-06,
|
| 4591 |
+
"loss": 1.1416,
|
| 4592 |
+
"step": 649
|
| 4593 |
+
},
|
| 4594 |
+
{
|
| 4595 |
+
"epoch": 0.8416963418582065,
|
| 4596 |
+
"grad_norm": 4.153942108154297,
|
| 4597 |
+
"learning_rate": 6.2762249521807645e-06,
|
| 4598 |
+
"loss": 1.3022,
|
| 4599 |
+
"step": 650
|
| 4600 |
+
},
|
| 4601 |
+
{
|
| 4602 |
+
"epoch": 0.8429912593072192,
|
| 4603 |
+
"grad_norm": 4.653594493865967,
|
| 4604 |
+
"learning_rate": 6.17673404307132e-06,
|
| 4605 |
+
"loss": 1.1183,
|
| 4606 |
+
"step": 651
|
| 4607 |
+
},
|
| 4608 |
+
{
|
| 4609 |
+
"epoch": 0.8442861767562317,
|
| 4610 |
+
"grad_norm": 5.605881690979004,
|
| 4611 |
+
"learning_rate": 6.077986075766612e-06,
|
| 4612 |
+
"loss": 0.7487,
|
| 4613 |
+
"step": 652
|
| 4614 |
+
},
|
| 4615 |
+
{
|
| 4616 |
+
"epoch": 0.8455810942052444,
|
| 4617 |
+
"grad_norm": 4.79271936416626,
|
| 4618 |
+
"learning_rate": 5.979982724354366e-06,
|
| 4619 |
+
"loss": 1.7176,
|
| 4620 |
+
"step": 653
|
| 4621 |
+
},
|
| 4622 |
+
{
|
| 4623 |
+
"epoch": 0.8468760116542571,
|
| 4624 |
+
"grad_norm": 6.354664325714111,
|
| 4625 |
+
"learning_rate": 5.882725650298787e-06,
|
| 4626 |
+
"loss": 1.5522,
|
| 4627 |
+
"step": 654
|
| 4628 |
+
},
|
| 4629 |
+
{
|
| 4630 |
+
"epoch": 0.8481709291032696,
|
| 4631 |
+
"grad_norm": 4.5003862380981445,
|
| 4632 |
+
"learning_rate": 5.7862165024123175e-06,
|
| 4633 |
+
"loss": 1.0054,
|
| 4634 |
+
"step": 655
|
| 4635 |
+
},
|
| 4636 |
+
{
|
| 4637 |
+
"epoch": 0.8494658465522823,
|
| 4638 |
+
"grad_norm": 4.5881781578063965,
|
| 4639 |
+
"learning_rate": 5.690456916827691e-06,
|
| 4640 |
+
"loss": 0.8891,
|
| 4641 |
+
"step": 656
|
| 4642 |
+
},
|
| 4643 |
+
{
|
| 4644 |
+
"epoch": 0.850760764001295,
|
| 4645 |
+
"grad_norm": 4.843465328216553,
|
| 4646 |
+
"learning_rate": 5.5954485169702306e-06,
|
| 4647 |
+
"loss": 0.8689,
|
| 4648 |
+
"step": 657
|
| 4649 |
+
},
|
| 4650 |
+
{
|
| 4651 |
+
"epoch": 0.8520556814503075,
|
| 4652 |
+
"grad_norm": 4.317074775695801,
|
| 4653 |
+
"learning_rate": 5.501192913530301e-06,
|
| 4654 |
+
"loss": 0.963,
|
| 4655 |
+
"step": 658
|
| 4656 |
+
},
|
| 4657 |
+
{
|
| 4658 |
+
"epoch": 0.8533505988993202,
|
| 4659 |
+
"grad_norm": 14.343049049377441,
|
| 4660 |
+
"learning_rate": 5.407691704435991e-06,
|
| 4661 |
+
"loss": 1.5631,
|
| 4662 |
+
"step": 659
|
| 4663 |
+
},
|
| 4664 |
+
{
|
| 4665 |
+
"epoch": 0.8546455163483327,
|
| 4666 |
+
"grad_norm": 5.277613162994385,
|
| 4667 |
+
"learning_rate": 5.314946474826066e-06,
|
| 4668 |
+
"loss": 1.0714,
|
| 4669 |
+
"step": 660
|
| 4670 |
+
},
|
| 4671 |
+
{
|
| 4672 |
+
"epoch": 0.8559404337973454,
|
| 4673 |
+
"grad_norm": 5.00631046295166,
|
| 4674 |
+
"learning_rate": 5.222958797023036e-06,
|
| 4675 |
+
"loss": 0.4646,
|
| 4676 |
+
"step": 661
|
| 4677 |
+
},
|
| 4678 |
+
{
|
| 4679 |
+
"epoch": 0.8572353512463581,
|
| 4680 |
+
"grad_norm": 3.9715874195098877,
|
| 4681 |
+
"learning_rate": 5.13173023050656e-06,
|
| 4682 |
+
"loss": 0.5706,
|
| 4683 |
+
"step": 662
|
| 4684 |
+
},
|
| 4685 |
+
{
|
| 4686 |
+
"epoch": 0.8585302686953706,
|
| 4687 |
+
"grad_norm": 7.35955286026001,
|
| 4688 |
+
"learning_rate": 5.041262321886958e-06,
|
| 4689 |
+
"loss": 2.2692,
|
| 4690 |
+
"step": 663
|
| 4691 |
+
},
|
| 4692 |
+
{
|
| 4693 |
+
"epoch": 0.8598251861443833,
|
| 4694 |
+
"grad_norm": 3.029388666152954,
|
| 4695 |
+
"learning_rate": 4.951556604879048e-06,
|
| 4696 |
+
"loss": 0.6182,
|
| 4697 |
+
"step": 664
|
| 4698 |
+
},
|
| 4699 |
+
{
|
| 4700 |
+
"epoch": 0.861120103593396,
|
| 4701 |
+
"grad_norm": 5.505150318145752,
|
| 4702 |
+
"learning_rate": 4.862614600276061e-06,
|
| 4703 |
+
"loss": 0.4528,
|
| 4704 |
+
"step": 665
|
| 4705 |
+
},
|
| 4706 |
+
{
|
| 4707 |
+
"epoch": 0.8624150210424085,
|
| 4708 |
+
"grad_norm": 4.694687366485596,
|
| 4709 |
+
"learning_rate": 4.774437815923938e-06,
|
| 4710 |
+
"loss": 1.1174,
|
| 4711 |
+
"step": 666
|
| 4712 |
+
},
|
| 4713 |
+
{
|
| 4714 |
+
"epoch": 0.8637099384914212,
|
| 4715 |
+
"grad_norm": 5.113190650939941,
|
| 4716 |
+
"learning_rate": 4.687027746695727e-06,
|
| 4717 |
+
"loss": 1.0881,
|
| 4718 |
+
"step": 667
|
| 4719 |
+
},
|
| 4720 |
+
{
|
| 4721 |
+
"epoch": 0.8650048559404337,
|
| 4722 |
+
"grad_norm": 5.682245254516602,
|
| 4723 |
+
"learning_rate": 4.600385874466256e-06,
|
| 4724 |
+
"loss": 0.9992,
|
| 4725 |
+
"step": 668
|
| 4726 |
+
},
|
| 4727 |
+
{
|
| 4728 |
+
"epoch": 0.8662997733894464,
|
| 4729 |
+
"grad_norm": 5.2173285484313965,
|
| 4730 |
+
"learning_rate": 4.514513668087011e-06,
|
| 4731 |
+
"loss": 0.8637,
|
| 4732 |
+
"step": 669
|
| 4733 |
+
},
|
| 4734 |
+
{
|
| 4735 |
+
"epoch": 0.8675946908384591,
|
| 4736 |
+
"grad_norm": 6.744502067565918,
|
| 4737 |
+
"learning_rate": 4.429412583361209e-06,
|
| 4738 |
+
"loss": 0.8903,
|
| 4739 |
+
"step": 670
|
| 4740 |
+
},
|
| 4741 |
+
{
|
| 4742 |
+
"epoch": 0.8688896082874716,
|
| 4743 |
+
"grad_norm": 4.4934916496276855,
|
| 4744 |
+
"learning_rate": 4.34508406301915e-06,
|
| 4745 |
+
"loss": 0.8,
|
| 4746 |
+
"step": 671
|
| 4747 |
+
},
|
| 4748 |
+
{
|
| 4749 |
+
"epoch": 0.8701845257364843,
|
| 4750 |
+
"grad_norm": 5.831895351409912,
|
| 4751 |
+
"learning_rate": 4.261529536693737e-06,
|
| 4752 |
+
"loss": 1.1527,
|
| 4753 |
+
"step": 672
|
| 4754 |
+
},
|
| 4755 |
+
{
|
| 4756 |
+
"epoch": 0.871479443185497,
|
| 4757 |
+
"grad_norm": 6.142891883850098,
|
| 4758 |
+
"learning_rate": 4.178750420896255e-06,
|
| 4759 |
+
"loss": 0.6969,
|
| 4760 |
+
"step": 673
|
| 4761 |
+
},
|
| 4762 |
+
{
|
| 4763 |
+
"epoch": 0.8727743606345095,
|
| 4764 |
+
"grad_norm": 7.165826797485352,
|
| 4765 |
+
"learning_rate": 4.0967481189923384e-06,
|
| 4766 |
+
"loss": 1.0799,
|
| 4767 |
+
"step": 674
|
| 4768 |
+
},
|
| 4769 |
+
{
|
| 4770 |
+
"epoch": 0.8740692780835222,
|
| 4771 |
+
"grad_norm": 7.070796012878418,
|
| 4772 |
+
"learning_rate": 4.015524021178196e-06,
|
| 4773 |
+
"loss": 1.2246,
|
| 4774 |
+
"step": 675
|
| 4775 |
+
},
|
| 4776 |
+
{
|
| 4777 |
+
"epoch": 0.8753641955325348,
|
| 4778 |
+
"grad_norm": 5.325906276702881,
|
| 4779 |
+
"learning_rate": 3.935079504457034e-06,
|
| 4780 |
+
"loss": 0.6554,
|
| 4781 |
+
"step": 676
|
| 4782 |
+
},
|
| 4783 |
+
{
|
| 4784 |
+
"epoch": 0.8766591129815474,
|
| 4785 |
+
"grad_norm": 11.728653907775879,
|
| 4786 |
+
"learning_rate": 3.8554159326157304e-06,
|
| 4787 |
+
"loss": 2.4503,
|
| 4788 |
+
"step": 677
|
| 4789 |
+
},
|
| 4790 |
+
{
|
| 4791 |
+
"epoch": 0.8779540304305601,
|
| 4792 |
+
"grad_norm": 4.562655925750732,
|
| 4793 |
+
"learning_rate": 3.7765346562016744e-06,
|
| 4794 |
+
"loss": 1.2593,
|
| 4795 |
+
"step": 678
|
| 4796 |
+
},
|
| 4797 |
+
{
|
| 4798 |
+
"epoch": 0.8792489478795726,
|
| 4799 |
+
"grad_norm": 4.027012348175049,
|
| 4800 |
+
"learning_rate": 3.6984370124999058e-06,
|
| 4801 |
+
"loss": 0.9379,
|
| 4802 |
+
"step": 679
|
| 4803 |
+
},
|
| 4804 |
+
{
|
| 4805 |
+
"epoch": 0.8805438653285853,
|
| 4806 |
+
"grad_norm": 3.9469587802886963,
|
| 4807 |
+
"learning_rate": 3.621124325510422e-06,
|
| 4808 |
+
"loss": 1.204,
|
| 4809 |
+
"step": 680
|
| 4810 |
+
},
|
| 4811 |
+
{
|
| 4812 |
+
"epoch": 0.881838782777598,
|
| 4813 |
+
"grad_norm": 5.622085094451904,
|
| 4814 |
+
"learning_rate": 3.5445979059257505e-06,
|
| 4815 |
+
"loss": 1.4393,
|
| 4816 |
+
"step": 681
|
| 4817 |
+
},
|
| 4818 |
+
{
|
| 4819 |
+
"epoch": 0.8831337002266105,
|
| 4820 |
+
"grad_norm": 6.133405685424805,
|
| 4821 |
+
"learning_rate": 3.4688590511087304e-06,
|
| 4822 |
+
"loss": 1.0091,
|
| 4823 |
+
"step": 682
|
| 4824 |
+
},
|
| 4825 |
+
{
|
| 4826 |
+
"epoch": 0.8844286176756232,
|
| 4827 |
+
"grad_norm": 6.239793300628662,
|
| 4828 |
+
"learning_rate": 3.3939090450704925e-06,
|
| 4829 |
+
"loss": 1.4701,
|
| 4830 |
+
"step": 683
|
| 4831 |
+
},
|
| 4832 |
+
{
|
| 4833 |
+
"epoch": 0.8857235351246358,
|
| 4834 |
+
"grad_norm": 11.061779975891113,
|
| 4835 |
+
"learning_rate": 3.3197491584487093e-06,
|
| 4836 |
+
"loss": 1.2142,
|
| 4837 |
+
"step": 684
|
| 4838 |
+
},
|
| 4839 |
+
{
|
| 4840 |
+
"epoch": 0.8870184525736484,
|
| 4841 |
+
"grad_norm": 7.947810173034668,
|
| 4842 |
+
"learning_rate": 3.246380648486058e-06,
|
| 4843 |
+
"loss": 2.459,
|
| 4844 |
+
"step": 685
|
| 4845 |
+
},
|
| 4846 |
+
{
|
| 4847 |
+
"epoch": 0.8883133700226611,
|
| 4848 |
+
"grad_norm": 3.787471294403076,
|
| 4849 |
+
"learning_rate": 3.1738047590088803e-06,
|
| 4850 |
+
"loss": 0.5336,
|
| 4851 |
+
"step": 686
|
| 4852 |
+
},
|
| 4853 |
+
{
|
| 4854 |
+
"epoch": 0.8896082874716736,
|
| 4855 |
+
"grad_norm": 6.404763698577881,
|
| 4856 |
+
"learning_rate": 3.10202272040615e-06,
|
| 4857 |
+
"loss": 1.0348,
|
| 4858 |
+
"step": 687
|
| 4859 |
+
},
|
| 4860 |
+
{
|
| 4861 |
+
"epoch": 0.8909032049206863,
|
| 4862 |
+
"grad_norm": 3.7612321376800537,
|
| 4863 |
+
"learning_rate": 3.0310357496085405e-06,
|
| 4864 |
+
"loss": 0.5957,
|
| 4865 |
+
"step": 688
|
| 4866 |
+
},
|
| 4867 |
+
{
|
| 4868 |
+
"epoch": 0.892198122369699,
|
| 4869 |
+
"grad_norm": 4.630037784576416,
|
| 4870 |
+
"learning_rate": 2.9608450500678565e-06,
|
| 4871 |
+
"loss": 1.1506,
|
| 4872 |
+
"step": 689
|
| 4873 |
+
},
|
| 4874 |
+
{
|
| 4875 |
+
"epoch": 0.8934930398187115,
|
| 4876 |
+
"grad_norm": 4.205752372741699,
|
| 4877 |
+
"learning_rate": 2.8914518117366006e-06,
|
| 4878 |
+
"loss": 0.9329,
|
| 4879 |
+
"step": 690
|
| 4880 |
+
},
|
| 4881 |
+
{
|
| 4882 |
+
"epoch": 0.8947879572677242,
|
| 4883 |
+
"grad_norm": 3.5641722679138184,
|
| 4884 |
+
"learning_rate": 2.8228572110478133e-06,
|
| 4885 |
+
"loss": 0.7643,
|
| 4886 |
+
"step": 691
|
| 4887 |
+
},
|
| 4888 |
+
{
|
| 4889 |
+
"epoch": 0.8960828747167368,
|
| 4890 |
+
"grad_norm": 4.8683342933654785,
|
| 4891 |
+
"learning_rate": 2.755062410895104e-06,
|
| 4892 |
+
"loss": 1.3503,
|
| 4893 |
+
"step": 692
|
| 4894 |
+
},
|
| 4895 |
+
{
|
| 4896 |
+
"epoch": 0.8973777921657494,
|
| 4897 |
+
"grad_norm": 4.980081558227539,
|
| 4898 |
+
"learning_rate": 2.6880685606129664e-06,
|
| 4899 |
+
"loss": 0.9362,
|
| 4900 |
+
"step": 693
|
| 4901 |
+
},
|
| 4902 |
+
{
|
| 4903 |
+
"epoch": 0.8986727096147621,
|
| 4904 |
+
"grad_norm": 3.7096731662750244,
|
| 4905 |
+
"learning_rate": 2.62187679595729e-06,
|
| 4906 |
+
"loss": 0.735,
|
| 4907 |
+
"step": 694
|
| 4908 |
+
},
|
| 4909 |
+
{
|
| 4910 |
+
"epoch": 0.8999676270637746,
|
| 4911 |
+
"grad_norm": 6.497191905975342,
|
| 4912 |
+
"learning_rate": 2.55648823908608e-06,
|
| 4913 |
+
"loss": 1.1868,
|
| 4914 |
+
"step": 695
|
| 4915 |
+
},
|
| 4916 |
+
{
|
| 4917 |
+
"epoch": 0.9012625445127873,
|
| 4918 |
+
"grad_norm": 9.368871688842773,
|
| 4919 |
+
"learning_rate": 2.4919039985404626e-06,
|
| 4920 |
+
"loss": 0.6176,
|
| 4921 |
+
"step": 696
|
| 4922 |
+
},
|
| 4923 |
+
{
|
| 4924 |
+
"epoch": 0.9025574619618,
|
| 4925 |
+
"grad_norm": 2.9543919563293457,
|
| 4926 |
+
"learning_rate": 2.428125169225881e-06,
|
| 4927 |
+
"loss": 0.5762,
|
| 4928 |
+
"step": 697
|
| 4929 |
+
},
|
| 4930 |
+
{
|
| 4931 |
+
"epoch": 0.9038523794108125,
|
| 4932 |
+
"grad_norm": 5.527708530426025,
|
| 4933 |
+
"learning_rate": 2.36515283239353e-06,
|
| 4934 |
+
"loss": 0.8727,
|
| 4935 |
+
"step": 698
|
| 4936 |
+
},
|
| 4937 |
+
{
|
| 4938 |
+
"epoch": 0.9051472968598252,
|
| 4939 |
+
"grad_norm": 7.263698577880859,
|
| 4940 |
+
"learning_rate": 2.3029880556220074e-06,
|
| 4941 |
+
"loss": 0.5661,
|
| 4942 |
+
"step": 699
|
| 4943 |
+
},
|
| 4944 |
+
{
|
| 4945 |
+
"epoch": 0.9064422143088378,
|
| 4946 |
+
"grad_norm": 4.572330951690674,
|
| 4947 |
+
"learning_rate": 2.241631892799262e-06,
|
| 4948 |
+
"loss": 0.6105,
|
| 4949 |
+
"step": 700
|
| 4950 |
+
},
|
| 4951 |
+
{
|
| 4952 |
+
"epoch": 0.9077371317578504,
|
| 4953 |
+
"grad_norm": 7.1983795166015625,
|
| 4954 |
+
"learning_rate": 2.181085384104703e-06,
|
| 4955 |
+
"loss": 1.0509,
|
| 4956 |
+
"step": 701
|
| 4957 |
+
},
|
| 4958 |
+
{
|
| 4959 |
+
"epoch": 0.9090320492068631,
|
| 4960 |
+
"grad_norm": 4.944607734680176,
|
| 4961 |
+
"learning_rate": 2.121349555991525e-06,
|
| 4962 |
+
"loss": 1.3179,
|
| 4963 |
+
"step": 702
|
| 4964 |
+
},
|
| 4965 |
+
{
|
| 4966 |
+
"epoch": 0.9103269666558756,
|
| 4967 |
+
"grad_norm": 3.876669406890869,
|
| 4968 |
+
"learning_rate": 2.0624254211693894e-06,
|
| 4969 |
+
"loss": 0.5404,
|
| 4970 |
+
"step": 703
|
| 4971 |
+
},
|
| 4972 |
+
{
|
| 4973 |
+
"epoch": 0.9116218841048883,
|
| 4974 |
+
"grad_norm": 3.4817299842834473,
|
| 4975 |
+
"learning_rate": 2.004313978587186e-06,
|
| 4976 |
+
"loss": 0.881,
|
| 4977 |
+
"step": 704
|
| 4978 |
+
},
|
| 4979 |
+
{
|
| 4980 |
+
"epoch": 0.912916801553901,
|
| 4981 |
+
"grad_norm": 5.139196395874023,
|
| 4982 |
+
"learning_rate": 1.9470162134161143e-06,
|
| 4983 |
+
"loss": 0.7959,
|
| 4984 |
+
"step": 705
|
| 4985 |
+
},
|
| 4986 |
+
{
|
| 4987 |
+
"epoch": 0.9142117190029135,
|
| 4988 |
+
"grad_norm": 6.930759429931641,
|
| 4989 |
+
"learning_rate": 1.8905330970330259e-06,
|
| 4990 |
+
"loss": 1.6162,
|
| 4991 |
+
"step": 706
|
| 4992 |
+
},
|
| 4993 |
+
{
|
| 4994 |
+
"epoch": 0.9155066364519262,
|
| 4995 |
+
"grad_norm": 4.119734764099121,
|
| 4996 |
+
"learning_rate": 1.83486558700387e-06,
|
| 4997 |
+
"loss": 0.984,
|
| 4998 |
+
"step": 707
|
| 4999 |
+
},
|
| 5000 |
+
{
|
| 5001 |
+
"epoch": 0.9168015539009389,
|
| 5002 |
+
"grad_norm": 4.680385112762451,
|
| 5003 |
+
"learning_rate": 1.78001462706755e-06,
|
| 5004 |
+
"loss": 0.8123,
|
| 5005 |
+
"step": 708
|
| 5006 |
+
},
|
| 5007 |
+
{
|
| 5008 |
+
"epoch": 0.9180964713499514,
|
| 5009 |
+
"grad_norm": 6.069380283355713,
|
| 5010 |
+
"learning_rate": 1.7259811471198706e-06,
|
| 5011 |
+
"loss": 0.9954,
|
| 5012 |
+
"step": 709
|
| 5013 |
+
},
|
| 5014 |
+
{
|
| 5015 |
+
"epoch": 0.9193913887989641,
|
| 5016 |
+
"grad_norm": 3.990244150161743,
|
| 5017 |
+
"learning_rate": 1.6727660631977893e-06,
|
| 5018 |
+
"loss": 0.6567,
|
| 5019 |
+
"step": 710
|
| 5020 |
+
},
|
| 5021 |
+
{
|
| 5022 |
+
"epoch": 0.9206863062479766,
|
| 5023 |
+
"grad_norm": 4.472250461578369,
|
| 5024 |
+
"learning_rate": 1.620370277463884e-06,
|
| 5025 |
+
"loss": 1.8166,
|
| 5026 |
+
"step": 711
|
| 5027 |
+
},
|
| 5028 |
+
{
|
| 5029 |
+
"epoch": 0.9219812236969893,
|
| 5030 |
+
"grad_norm": 4.742919921875,
|
| 5031 |
+
"learning_rate": 1.5687946781910378e-06,
|
| 5032 |
+
"loss": 0.8593,
|
| 5033 |
+
"step": 712
|
| 5034 |
+
},
|
| 5035 |
+
{
|
| 5036 |
+
"epoch": 0.923276141146002,
|
| 5037 |
+
"grad_norm": 4.736976146697998,
|
| 5038 |
+
"learning_rate": 1.5180401397474343e-06,
|
| 5039 |
+
"loss": 1.4691,
|
| 5040 |
+
"step": 713
|
| 5041 |
+
},
|
| 5042 |
+
{
|
| 5043 |
+
"epoch": 0.9245710585950145,
|
| 5044 |
+
"grad_norm": 4.94817590713501,
|
| 5045 |
+
"learning_rate": 1.4681075225816854e-06,
|
| 5046 |
+
"loss": 1.0589,
|
| 5047 |
+
"step": 714
|
| 5048 |
+
},
|
| 5049 |
+
{
|
| 5050 |
+
"epoch": 0.9258659760440272,
|
| 5051 |
+
"grad_norm": 7.17746639251709,
|
| 5052 |
+
"learning_rate": 1.4189976732082666e-06,
|
| 5053 |
+
"loss": 2.4874,
|
| 5054 |
+
"step": 715
|
| 5055 |
+
},
|
| 5056 |
+
{
|
| 5057 |
+
"epoch": 0.9271608934930399,
|
| 5058 |
+
"grad_norm": 4.172902584075928,
|
| 5059 |
+
"learning_rate": 1.3707114241931328e-06,
|
| 5060 |
+
"loss": 0.9651,
|
| 5061 |
+
"step": 716
|
| 5062 |
+
},
|
| 5063 |
+
{
|
| 5064 |
+
"epoch": 0.9284558109420524,
|
| 5065 |
+
"grad_norm": 4.561225414276123,
|
| 5066 |
+
"learning_rate": 1.3232495941396639e-06,
|
| 5067 |
+
"loss": 0.8338,
|
| 5068 |
+
"step": 717
|
| 5069 |
+
},
|
| 5070 |
+
{
|
| 5071 |
+
"epoch": 0.9297507283910651,
|
| 5072 |
+
"grad_norm": 4.700865745544434,
|
| 5073 |
+
"learning_rate": 1.2766129876747413e-06,
|
| 5074 |
+
"loss": 0.6137,
|
| 5075 |
+
"step": 718
|
| 5076 |
+
},
|
| 5077 |
+
{
|
| 5078 |
+
"epoch": 0.9310456458400777,
|
| 5079 |
+
"grad_norm": 4.913844108581543,
|
| 5080 |
+
"learning_rate": 1.2308023954351043e-06,
|
| 5081 |
+
"loss": 1.0453,
|
| 5082 |
+
"step": 719
|
| 5083 |
+
},
|
| 5084 |
+
{
|
| 5085 |
+
"epoch": 0.9323405632890903,
|
| 5086 |
+
"grad_norm": 3.081576108932495,
|
| 5087 |
+
"learning_rate": 1.1858185940539779e-06,
|
| 5088 |
+
"loss": 0.738,
|
| 5089 |
+
"step": 720
|
| 5090 |
+
},
|
| 5091 |
+
{
|
| 5092 |
+
"epoch": 0.933635480738103,
|
| 5093 |
+
"grad_norm": 6.5089006423950195,
|
| 5094 |
+
"learning_rate": 1.1416623461478704e-06,
|
| 5095 |
+
"loss": 1.3315,
|
| 5096 |
+
"step": 721
|
| 5097 |
+
},
|
| 5098 |
+
{
|
| 5099 |
+
"epoch": 0.9349303981871155,
|
| 5100 |
+
"grad_norm": 4.095083236694336,
|
| 5101 |
+
"learning_rate": 1.0983344003036912e-06,
|
| 5102 |
+
"loss": 0.7988,
|
| 5103 |
+
"step": 722
|
| 5104 |
+
},
|
| 5105 |
+
{
|
| 5106 |
+
"epoch": 0.9362253156361282,
|
| 5107 |
+
"grad_norm": 4.421083927154541,
|
| 5108 |
+
"learning_rate": 1.055835491066004e-06,
|
| 5109 |
+
"loss": 0.6599,
|
| 5110 |
+
"step": 723
|
| 5111 |
+
},
|
| 5112 |
+
{
|
| 5113 |
+
"epoch": 0.9375202330851409,
|
| 5114 |
+
"grad_norm": 6.9717698097229,
|
| 5115 |
+
"learning_rate": 1.014166338924627e-06,
|
| 5116 |
+
"loss": 1.9028,
|
| 5117 |
+
"step": 724
|
| 5118 |
+
},
|
| 5119 |
+
{
|
| 5120 |
+
"epoch": 0.9388151505341534,
|
| 5121 |
+
"grad_norm": 5.759119033813477,
|
| 5122 |
+
"learning_rate": 9.733276503023692e-07,
|
| 5123 |
+
"loss": 1.0143,
|
| 5124 |
+
"step": 725
|
| 5125 |
+
},
|
| 5126 |
+
{
|
| 5127 |
+
"epoch": 0.9401100679831661,
|
| 5128 |
+
"grad_norm": 7.723138809204102,
|
| 5129 |
+
"learning_rate": 9.33320117543085e-07,
|
| 5130 |
+
"loss": 1.5553,
|
| 5131 |
+
"step": 726
|
| 5132 |
+
},
|
| 5133 |
+
{
|
| 5134 |
+
"epoch": 0.9414049854321787,
|
| 5135 |
+
"grad_norm": 5.182671070098877,
|
| 5136 |
+
"learning_rate": 8.941444188999393e-07,
|
| 5137 |
+
"loss": 1.2996,
|
| 5138 |
+
"step": 727
|
| 5139 |
+
},
|
| 5140 |
+
{
|
| 5141 |
+
"epoch": 0.9426999028811913,
|
| 5142 |
+
"grad_norm": 8.686777114868164,
|
| 5143 |
+
"learning_rate": 8.558012185238939e-07,
|
| 5144 |
+
"loss": 1.1213,
|
| 5145 |
+
"step": 728
|
| 5146 |
+
},
|
| 5147 |
+
{
|
| 5148 |
+
"epoch": 0.943994820330204,
|
| 5149 |
+
"grad_norm": 8.205387115478516,
|
| 5150 |
+
"learning_rate": 8.182911664524562e-07,
|
| 5151 |
+
"loss": 0.7117,
|
| 5152 |
+
"step": 729
|
| 5153 |
+
},
|
| 5154 |
+
{
|
| 5155 |
+
"epoch": 0.9452897377792165,
|
| 5156 |
+
"grad_norm": 5.972929000854492,
|
| 5157 |
+
"learning_rate": 7.816148985986483e-07,
|
| 5158 |
+
"loss": 0.9229,
|
| 5159 |
+
"step": 730
|
| 5160 |
+
},
|
| 5161 |
+
{
|
| 5162 |
+
"epoch": 0.9465846552282292,
|
| 5163 |
+
"grad_norm": 8.060487747192383,
|
| 5164 |
+
"learning_rate": 7.457730367402549e-07,
|
| 5165 |
+
"loss": 1.5945,
|
| 5166 |
+
"step": 731
|
| 5167 |
+
},
|
| 5168 |
+
{
|
| 5169 |
+
"epoch": 0.9478795726772419,
|
| 5170 |
+
"grad_norm": 4.760429382324219,
|
| 5171 |
+
"learning_rate": 7.107661885092321e-07,
|
| 5172 |
+
"loss": 0.9784,
|
| 5173 |
+
"step": 732
|
| 5174 |
+
},
|
| 5175 |
+
{
|
| 5176 |
+
"epoch": 0.9491744901262544,
|
| 5177 |
+
"grad_norm": 4.759274482727051,
|
| 5178 |
+
"learning_rate": 6.765949473814648e-07,
|
| 5179 |
+
"loss": 1.095,
|
| 5180 |
+
"step": 733
|
| 5181 |
+
},
|
| 5182 |
+
{
|
| 5183 |
+
"epoch": 0.9504694075752671,
|
| 5184 |
+
"grad_norm": 4.729125022888184,
|
| 5185 |
+
"learning_rate": 6.432598926666589e-07,
|
| 5186 |
+
"loss": 0.7815,
|
| 5187 |
+
"step": 734
|
| 5188 |
+
},
|
| 5189 |
+
{
|
| 5190 |
+
"epoch": 0.9517643250242797,
|
| 5191 |
+
"grad_norm": 5.148903846740723,
|
| 5192 |
+
"learning_rate": 6.107615894985375e-07,
|
| 5193 |
+
"loss": 1.2755,
|
| 5194 |
+
"step": 735
|
| 5195 |
+
},
|
| 5196 |
+
{
|
| 5197 |
+
"epoch": 0.9530592424732923,
|
| 5198 |
+
"grad_norm": 4.272920608520508,
|
| 5199 |
+
"learning_rate": 5.791005888252765e-07,
|
| 5200 |
+
"loss": 0.6422,
|
| 5201 |
+
"step": 736
|
| 5202 |
+
},
|
| 5203 |
+
{
|
| 5204 |
+
"epoch": 0.954354159922305,
|
| 5205 |
+
"grad_norm": 3.1463730335235596,
|
| 5206 |
+
"learning_rate": 5.482774274001401e-07,
|
| 5207 |
+
"loss": 0.7711,
|
| 5208 |
+
"step": 737
|
| 5209 |
+
},
|
| 5210 |
+
{
|
| 5211 |
+
"epoch": 0.9556490773713175,
|
| 5212 |
+
"grad_norm": 4.3048906326293945,
|
| 5213 |
+
"learning_rate": 5.18292627772382e-07,
|
| 5214 |
+
"loss": 0.6693,
|
| 5215 |
+
"step": 738
|
| 5216 |
+
},
|
| 5217 |
+
{
|
| 5218 |
+
"epoch": 0.9569439948203302,
|
| 5219 |
+
"grad_norm": 6.0783209800720215,
|
| 5220 |
+
"learning_rate": 4.891466982783977e-07,
|
| 5221 |
+
"loss": 0.8769,
|
| 5222 |
+
"step": 739
|
| 5223 |
+
},
|
| 5224 |
+
{
|
| 5225 |
+
"epoch": 0.9582389122693429,
|
| 5226 |
+
"grad_norm": 5.600512981414795,
|
| 5227 |
+
"learning_rate": 4.60840133033108e-07,
|
| 5228 |
+
"loss": 1.1665,
|
| 5229 |
+
"step": 740
|
| 5230 |
+
},
|
| 5231 |
+
{
|
| 5232 |
+
"epoch": 0.9595338297183554,
|
| 5233 |
+
"grad_norm": 9.313074111938477,
|
| 5234 |
+
"learning_rate": 4.3337341192157265e-07,
|
| 5235 |
+
"loss": 1.9236,
|
| 5236 |
+
"step": 741
|
| 5237 |
+
},
|
| 5238 |
+
{
|
| 5239 |
+
"epoch": 0.9608287471673681,
|
| 5240 |
+
"grad_norm": 7.646642208099365,
|
| 5241 |
+
"learning_rate": 4.067470005908625e-07,
|
| 5242 |
+
"loss": 0.8392,
|
| 5243 |
+
"step": 742
|
| 5244 |
+
},
|
| 5245 |
+
{
|
| 5246 |
+
"epoch": 0.9621236646163807,
|
| 5247 |
+
"grad_norm": 9.741737365722656,
|
| 5248 |
+
"learning_rate": 3.809613504421661e-07,
|
| 5249 |
+
"loss": 1.3232,
|
| 5250 |
+
"step": 743
|
| 5251 |
+
},
|
| 5252 |
+
{
|
| 5253 |
+
"epoch": 0.9634185820653933,
|
| 5254 |
+
"grad_norm": 6.090597152709961,
|
| 5255 |
+
"learning_rate": 3.5601689862311826e-07,
|
| 5256 |
+
"loss": 1.1143,
|
| 5257 |
+
"step": 744
|
| 5258 |
+
},
|
| 5259 |
+
{
|
| 5260 |
+
"epoch": 0.964713499514406,
|
| 5261 |
+
"grad_norm": 5.305496692657471,
|
| 5262 |
+
"learning_rate": 3.3191406802041693e-07,
|
| 5263 |
+
"loss": 0.7994,
|
| 5264 |
+
"step": 745
|
| 5265 |
+
},
|
| 5266 |
+
{
|
| 5267 |
+
"epoch": 0.9660084169634185,
|
| 5268 |
+
"grad_norm": 5.51359748840332,
|
| 5269 |
+
"learning_rate": 3.0865326725263435e-07,
|
| 5270 |
+
"loss": 2.0726,
|
| 5271 |
+
"step": 746
|
| 5272 |
+
},
|
| 5273 |
+
{
|
| 5274 |
+
"epoch": 0.9673033344124312,
|
| 5275 |
+
"grad_norm": 2.874615430831909,
|
| 5276 |
+
"learning_rate": 2.8623489066329503e-07,
|
| 5277 |
+
"loss": 0.6604,
|
| 5278 |
+
"step": 747
|
| 5279 |
+
},
|
| 5280 |
+
{
|
| 5281 |
+
"epoch": 0.9685982518614439,
|
| 5282 |
+
"grad_norm": 5.645188808441162,
|
| 5283 |
+
"learning_rate": 2.646593183142088e-07,
|
| 5284 |
+
"loss": 0.9807,
|
| 5285 |
+
"step": 748
|
| 5286 |
+
},
|
| 5287 |
+
{
|
| 5288 |
+
"epoch": 0.9698931693104564,
|
| 5289 |
+
"grad_norm": 3.8236138820648193,
|
| 5290 |
+
"learning_rate": 2.4392691597898143e-07,
|
| 5291 |
+
"loss": 0.923,
|
| 5292 |
+
"step": 749
|
| 5293 |
+
},
|
| 5294 |
+
{
|
| 5295 |
+
"epoch": 0.9711880867594691,
|
| 5296 |
+
"grad_norm": 4.602663993835449,
|
| 5297 |
+
"learning_rate": 2.2403803513686428e-07,
|
| 5298 |
+
"loss": 0.5873,
|
| 5299 |
+
"step": 750
|
| 5300 |
+
},
|
| 5301 |
+
{
|
| 5302 |
+
"epoch": 0.9724830042084817,
|
| 5303 |
+
"grad_norm": 3.112046718597412,
|
| 5304 |
+
"learning_rate": 2.0499301296676432e-07,
|
| 5305 |
+
"loss": 0.6582,
|
| 5306 |
+
"step": 751
|
| 5307 |
+
},
|
| 5308 |
+
{
|
| 5309 |
+
"epoch": 0.9737779216574943,
|
| 5310 |
+
"grad_norm": 5.2788920402526855,
|
| 5311 |
+
"learning_rate": 1.8679217234154334e-07,
|
| 5312 |
+
"loss": 1.0278,
|
| 5313 |
+
"step": 752
|
| 5314 |
+
},
|
| 5315 |
+
{
|
| 5316 |
+
"epoch": 0.975072839106507,
|
| 5317 |
+
"grad_norm": 6.206789493560791,
|
| 5318 |
+
"learning_rate": 1.6943582182253336e-07,
|
| 5319 |
+
"loss": 1.0729,
|
| 5320 |
+
"step": 753
|
| 5321 |
+
},
|
| 5322 |
+
{
|
| 5323 |
+
"epoch": 0.9763677565555196,
|
| 5324 |
+
"grad_norm": 7.300841331481934,
|
| 5325 |
+
"learning_rate": 1.5292425565430757e-07,
|
| 5326 |
+
"loss": 1.3212,
|
| 5327 |
+
"step": 754
|
| 5328 |
+
},
|
| 5329 |
+
{
|
| 5330 |
+
"epoch": 0.9776626740045322,
|
| 5331 |
+
"grad_norm": 4.0115275382995605,
|
| 5332 |
+
"learning_rate": 1.372577537597064e-07,
|
| 5333 |
+
"loss": 0.9395,
|
| 5334 |
+
"step": 755
|
| 5335 |
+
},
|
| 5336 |
+
{
|
| 5337 |
+
"epoch": 0.9789575914535449,
|
| 5338 |
+
"grad_norm": 7.623907089233398,
|
| 5339 |
+
"learning_rate": 1.224365817350692e-07,
|
| 5340 |
+
"loss": 1.3756,
|
| 5341 |
+
"step": 756
|
| 5342 |
+
},
|
| 5343 |
+
{
|
| 5344 |
+
"epoch": 0.9802525089025574,
|
| 5345 |
+
"grad_norm": 20.821199417114258,
|
| 5346 |
+
"learning_rate": 1.0846099084574346e-07,
|
| 5347 |
+
"loss": 3.2897,
|
| 5348 |
+
"step": 757
|
| 5349 |
+
},
|
| 5350 |
+
{
|
| 5351 |
+
"epoch": 0.9815474263515701,
|
| 5352 |
+
"grad_norm": 5.356776714324951,
|
| 5353 |
+
"learning_rate": 9.533121802183797e-08,
|
| 5354 |
+
"loss": 0.9017,
|
| 5355 |
+
"step": 758
|
| 5356 |
+
},
|
| 5357 |
+
{
|
| 5358 |
+
"epoch": 0.9828423438005827,
|
| 5359 |
+
"grad_norm": 6.1148505210876465,
|
| 5360 |
+
"learning_rate": 8.304748585417078e-08,
|
| 5361 |
+
"loss": 1.4205,
|
| 5362 |
+
"step": 759
|
| 5363 |
+
},
|
| 5364 |
+
{
|
| 5365 |
+
"epoch": 0.9841372612495953,
|
| 5366 |
+
"grad_norm": 6.684154033660889,
|
| 5367 |
+
"learning_rate": 7.161000259053308e-08,
|
| 5368 |
+
"loss": 1.3984,
|
| 5369 |
+
"step": 760
|
| 5370 |
+
},
|
| 5371 |
+
{
|
| 5372 |
+
"epoch": 0.985432178698608,
|
| 5373 |
+
"grad_norm": 2.9924120903015137,
|
| 5374 |
+
"learning_rate": 6.10189621321422e-08,
|
| 5375 |
+
"loss": 0.6213,
|
| 5376 |
+
"step": 761
|
| 5377 |
+
},
|
| 5378 |
+
{
|
| 5379 |
+
"epoch": 0.9867270961476206,
|
| 5380 |
+
"grad_norm": 4.251294136047363,
|
| 5381 |
+
"learning_rate": 5.127454403034415e-08,
|
| 5382 |
+
"loss": 1.0332,
|
| 5383 |
+
"step": 762
|
| 5384 |
+
},
|
| 5385 |
+
{
|
| 5386 |
+
"epoch": 0.9880220135966332,
|
| 5387 |
+
"grad_norm": 4.321407794952393,
|
| 5388 |
+
"learning_rate": 4.2376913483599404e-08,
|
| 5389 |
+
"loss": 1.1482,
|
| 5390 |
+
"step": 763
|
| 5391 |
+
},
|
| 5392 |
+
{
|
| 5393 |
+
"epoch": 0.9893169310456459,
|
| 5394 |
+
"grad_norm": 8.435138702392578,
|
| 5395 |
+
"learning_rate": 3.4326221334640695e-08,
|
| 5396 |
+
"loss": 2.5909,
|
| 5397 |
+
"step": 764
|
| 5398 |
+
},
|
| 5399 |
+
{
|
| 5400 |
+
"epoch": 0.9906118484946584,
|
| 5401 |
+
"grad_norm": 4.1094560623168945,
|
| 5402 |
+
"learning_rate": 2.712260406795286e-08,
|
| 5403 |
+
"loss": 0.4297,
|
| 5404 |
+
"step": 765
|
| 5405 |
+
},
|
| 5406 |
+
{
|
| 5407 |
+
"epoch": 0.9919067659436711,
|
| 5408 |
+
"grad_norm": 4.245459079742432,
|
| 5409 |
+
"learning_rate": 2.076618380744133e-08,
|
| 5410 |
+
"loss": 1.6027,
|
| 5411 |
+
"step": 766
|
| 5412 |
+
},
|
| 5413 |
+
{
|
| 5414 |
+
"epoch": 0.9932016833926837,
|
| 5415 |
+
"grad_norm": 4.5725297927856445,
|
| 5416 |
+
"learning_rate": 1.525706831437268e-08,
|
| 5417 |
+
"loss": 1.3605,
|
| 5418 |
+
"step": 767
|
| 5419 |
+
},
|
| 5420 |
+
{
|
| 5421 |
+
"epoch": 0.9944966008416963,
|
| 5422 |
+
"grad_norm": 3.8891658782958984,
|
| 5423 |
+
"learning_rate": 1.0595350985526109e-08,
|
| 5424 |
+
"loss": 0.8142,
|
| 5425 |
+
"step": 768
|
| 5426 |
+
},
|
| 5427 |
+
{
|
| 5428 |
+
"epoch": 0.995791518290709,
|
| 5429 |
+
"grad_norm": 4.054940223693848,
|
| 5430 |
+
"learning_rate": 6.781110851633576e-09,
|
| 5431 |
+
"loss": 1.1612,
|
| 5432 |
+
"step": 769
|
| 5433 |
+
},
|
| 5434 |
+
{
|
| 5435 |
+
"epoch": 0.9970864357397216,
|
| 5436 |
+
"grad_norm": 7.459210395812988,
|
| 5437 |
+
"learning_rate": 3.814412576025328e-09,
|
| 5438 |
+
"loss": 1.4391,
|
| 5439 |
+
"step": 770
|
| 5440 |
+
},
|
| 5441 |
+
{
|
| 5442 |
+
"epoch": 0.9983813531887342,
|
| 5443 |
+
"grad_norm": 4.039707183837891,
|
| 5444 |
+
"learning_rate": 1.6953064535474295e-09,
|
| 5445 |
+
"loss": 0.799,
|
| 5446 |
+
"step": 771
|
| 5447 |
+
},
|
| 5448 |
+
{
|
| 5449 |
+
"epoch": 0.9996762706377469,
|
| 5450 |
+
"grad_norm": 3.708043098449707,
|
| 5451 |
+
"learning_rate": 4.238284096902412e-10,
|
| 5452 |
+
"loss": 0.8702,
|
| 5453 |
+
"step": 772
|
| 5454 |
+
},
|
| 5455 |
+
{
|
| 5456 |
+
"epoch": 1.0009711880867596,
|
| 5457 |
+
"grad_norm": 27.973024368286133,
|
| 5458 |
+
"learning_rate": 0.0,
|
| 5459 |
+
"loss": 3.9517,
|
| 5460 |
+
"step": 773
|
| 5461 |
}
|
| 5462 |
],
|
| 5463 |
"logging_steps": 1,
|
|
|
|
| 5472 |
"should_evaluate": false,
|
| 5473 |
"should_log": false,
|
| 5474 |
"should_save": true,
|
| 5475 |
+
"should_training_stop": true
|
| 5476 |
},
|
| 5477 |
"attributes": {}
|
| 5478 |
}
|
| 5479 |
},
|
| 5480 |
+
"total_flos": 1084147192037376.0,
|
| 5481 |
"train_batch_size": 2,
|
| 5482 |
"trial_name": null,
|
| 5483 |
"trial_params": null
|