Training in progress, epoch 2, checkpoint
Browse files- last-checkpoint/optimizer.pt +1 -1
- last-checkpoint/rng_state.pth +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +2251 -3
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 295765866
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8898176c9ee551ed32dd529cc843abd1840164f3645656fdebe1c239a3d87ae1
|
| 3 |
size 295765866
|
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 sha256:3895045b743f09572b379a470906b3b61fec60b0f952eeaa286c3266a3af0c82
|
| 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:dd3e72f9db43bec86b839573da68f94e1b5aee7b8c1a2d328ad343d08e7ad7dc
|
| 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": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -2248,6 +2248,2254 @@
|
|
| 2248 |
"eval_samples_per_second": 35.534,
|
| 2249 |
"eval_steps_per_second": 8.9,
|
| 2250 |
"step": 319
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2251 |
}
|
| 2252 |
],
|
| 2253 |
"logging_steps": 1,
|
|
@@ -2267,7 +4515,7 @@
|
|
| 2267 |
"attributes": {}
|
| 2268 |
}
|
| 2269 |
},
|
| 2270 |
-
"total_flos":
|
| 2271 |
"train_batch_size": 4,
|
| 2272 |
"trial_name": null,
|
| 2273 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.0,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 639,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 2248 |
"eval_samples_per_second": 35.534,
|
| 2249 |
"eval_steps_per_second": 8.9,
|
| 2250 |
"step": 319
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 1.001564945226917,
|
| 2254 |
+
"grad_norm": NaN,
|
| 2255 |
+
"learning_rate": 7.580954190982213e-05,
|
| 2256 |
+
"loss": 0.0,
|
| 2257 |
+
"step": 320
|
| 2258 |
+
},
|
| 2259 |
+
{
|
| 2260 |
+
"epoch": 1.0046948356807512,
|
| 2261 |
+
"grad_norm": NaN,
|
| 2262 |
+
"learning_rate": 7.566733620348596e-05,
|
| 2263 |
+
"loss": 0.0,
|
| 2264 |
+
"step": 321
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 1.0078247261345852,
|
| 2268 |
+
"grad_norm": NaN,
|
| 2269 |
+
"learning_rate": 7.552484802203858e-05,
|
| 2270 |
+
"loss": 0.0,
|
| 2271 |
+
"step": 322
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 1.0109546165884193,
|
| 2275 |
+
"grad_norm": NaN,
|
| 2276 |
+
"learning_rate": 7.538207893359611e-05,
|
| 2277 |
+
"loss": 0.0,
|
| 2278 |
+
"step": 323
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 1.0140845070422535,
|
| 2282 |
+
"grad_norm": NaN,
|
| 2283 |
+
"learning_rate": 7.523903050936621e-05,
|
| 2284 |
+
"loss": 0.0,
|
| 2285 |
+
"step": 324
|
| 2286 |
+
},
|
| 2287 |
+
{
|
| 2288 |
+
"epoch": 1.0172143974960877,
|
| 2289 |
+
"grad_norm": NaN,
|
| 2290 |
+
"learning_rate": 7.509570432363061e-05,
|
| 2291 |
+
"loss": 0.0,
|
| 2292 |
+
"step": 325
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 1.0203442879499218,
|
| 2296 |
+
"grad_norm": NaN,
|
| 2297 |
+
"learning_rate": 7.495210195372795e-05,
|
| 2298 |
+
"loss": 0.0,
|
| 2299 |
+
"step": 326
|
| 2300 |
+
},
|
| 2301 |
+
{
|
| 2302 |
+
"epoch": 1.0234741784037558,
|
| 2303 |
+
"grad_norm": NaN,
|
| 2304 |
+
"learning_rate": 7.48082249800363e-05,
|
| 2305 |
+
"loss": 0.0,
|
| 2306 |
+
"step": 327
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 1.02660406885759,
|
| 2310 |
+
"grad_norm": NaN,
|
| 2311 |
+
"learning_rate": 7.46640749859558e-05,
|
| 2312 |
+
"loss": 0.0,
|
| 2313 |
+
"step": 328
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 1.029733959311424,
|
| 2317 |
+
"grad_norm": NaN,
|
| 2318 |
+
"learning_rate": 7.451965355789129e-05,
|
| 2319 |
+
"loss": 0.0,
|
| 2320 |
+
"step": 329
|
| 2321 |
+
},
|
| 2322 |
+
{
|
| 2323 |
+
"epoch": 1.0328638497652582,
|
| 2324 |
+
"grad_norm": NaN,
|
| 2325 |
+
"learning_rate": 7.437496228523473e-05,
|
| 2326 |
+
"loss": 0.0,
|
| 2327 |
+
"step": 330
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 1.0359937402190924,
|
| 2331 |
+
"grad_norm": NaN,
|
| 2332 |
+
"learning_rate": 7.423000276034786e-05,
|
| 2333 |
+
"loss": 0.0,
|
| 2334 |
+
"step": 331
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 1.0391236306729263,
|
| 2338 |
+
"grad_norm": NaN,
|
| 2339 |
+
"learning_rate": 7.408477657854458e-05,
|
| 2340 |
+
"loss": 0.0,
|
| 2341 |
+
"step": 332
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 1.0422535211267605,
|
| 2345 |
+
"grad_norm": NaN,
|
| 2346 |
+
"learning_rate": 7.393928533807334e-05,
|
| 2347 |
+
"loss": 0.0,
|
| 2348 |
+
"step": 333
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 1.0453834115805947,
|
| 2352 |
+
"grad_norm": NaN,
|
| 2353 |
+
"learning_rate": 7.379353064009976e-05,
|
| 2354 |
+
"loss": 0.0,
|
| 2355 |
+
"step": 334
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 1.0485133020344288,
|
| 2359 |
+
"grad_norm": NaN,
|
| 2360 |
+
"learning_rate": 7.364751408868877e-05,
|
| 2361 |
+
"loss": 0.0,
|
| 2362 |
+
"step": 335
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 1.051643192488263,
|
| 2366 |
+
"grad_norm": NaN,
|
| 2367 |
+
"learning_rate": 7.350123729078705e-05,
|
| 2368 |
+
"loss": 0.0,
|
| 2369 |
+
"step": 336
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 1.054773082942097,
|
| 2373 |
+
"grad_norm": NaN,
|
| 2374 |
+
"learning_rate": 7.335470185620543e-05,
|
| 2375 |
+
"loss": 0.0,
|
| 2376 |
+
"step": 337
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 1.057902973395931,
|
| 2380 |
+
"grad_norm": NaN,
|
| 2381 |
+
"learning_rate": 7.320790939760105e-05,
|
| 2382 |
+
"loss": 0.0,
|
| 2383 |
+
"step": 338
|
| 2384 |
+
},
|
| 2385 |
+
{
|
| 2386 |
+
"epoch": 1.0610328638497653,
|
| 2387 |
+
"grad_norm": NaN,
|
| 2388 |
+
"learning_rate": 7.306086153045965e-05,
|
| 2389 |
+
"loss": 0.0,
|
| 2390 |
+
"step": 339
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 1.0641627543035994,
|
| 2394 |
+
"grad_norm": NaN,
|
| 2395 |
+
"learning_rate": 7.291355987307784e-05,
|
| 2396 |
+
"loss": 0.0,
|
| 2397 |
+
"step": 340
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 1.0672926447574336,
|
| 2401 |
+
"grad_norm": NaN,
|
| 2402 |
+
"learning_rate": 7.276600604654522e-05,
|
| 2403 |
+
"loss": 0.0,
|
| 2404 |
+
"step": 341
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"epoch": 1.0704225352112675,
|
| 2408 |
+
"grad_norm": NaN,
|
| 2409 |
+
"learning_rate": 7.261820167472658e-05,
|
| 2410 |
+
"loss": 0.0,
|
| 2411 |
+
"step": 342
|
| 2412 |
+
},
|
| 2413 |
+
{
|
| 2414 |
+
"epoch": 1.0735524256651017,
|
| 2415 |
+
"grad_norm": NaN,
|
| 2416 |
+
"learning_rate": 7.247014838424404e-05,
|
| 2417 |
+
"loss": 0.0,
|
| 2418 |
+
"step": 343
|
| 2419 |
+
},
|
| 2420 |
+
{
|
| 2421 |
+
"epoch": 1.0766823161189358,
|
| 2422 |
+
"grad_norm": NaN,
|
| 2423 |
+
"learning_rate": 7.232184780445907e-05,
|
| 2424 |
+
"loss": 0.0,
|
| 2425 |
+
"step": 344
|
| 2426 |
+
},
|
| 2427 |
+
{
|
| 2428 |
+
"epoch": 1.07981220657277,
|
| 2429 |
+
"grad_norm": NaN,
|
| 2430 |
+
"learning_rate": 7.217330156745471e-05,
|
| 2431 |
+
"loss": 0.0,
|
| 2432 |
+
"step": 345
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 1.0829420970266042,
|
| 2436 |
+
"grad_norm": NaN,
|
| 2437 |
+
"learning_rate": 7.202451130801742e-05,
|
| 2438 |
+
"loss": 0.0,
|
| 2439 |
+
"step": 346
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 1.086071987480438,
|
| 2443 |
+
"grad_norm": NaN,
|
| 2444 |
+
"learning_rate": 7.187547866361925e-05,
|
| 2445 |
+
"loss": 0.0,
|
| 2446 |
+
"step": 347
|
| 2447 |
+
},
|
| 2448 |
+
{
|
| 2449 |
+
"epoch": 1.0892018779342723,
|
| 2450 |
+
"grad_norm": NaN,
|
| 2451 |
+
"learning_rate": 7.172620527439976e-05,
|
| 2452 |
+
"loss": 0.0,
|
| 2453 |
+
"step": 348
|
| 2454 |
+
},
|
| 2455 |
+
{
|
| 2456 |
+
"epoch": 1.0923317683881064,
|
| 2457 |
+
"grad_norm": NaN,
|
| 2458 |
+
"learning_rate": 7.157669278314789e-05,
|
| 2459 |
+
"loss": 0.0,
|
| 2460 |
+
"step": 349
|
| 2461 |
+
},
|
| 2462 |
+
{
|
| 2463 |
+
"epoch": 1.0954616588419406,
|
| 2464 |
+
"grad_norm": NaN,
|
| 2465 |
+
"learning_rate": 7.1426942835284e-05,
|
| 2466 |
+
"loss": 0.0,
|
| 2467 |
+
"step": 350
|
| 2468 |
+
},
|
| 2469 |
+
{
|
| 2470 |
+
"epoch": 1.0985915492957747,
|
| 2471 |
+
"grad_norm": NaN,
|
| 2472 |
+
"learning_rate": 7.127695707884176e-05,
|
| 2473 |
+
"loss": 0.0,
|
| 2474 |
+
"step": 351
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 1.1017214397496087,
|
| 2478 |
+
"grad_norm": NaN,
|
| 2479 |
+
"learning_rate": 7.112673716444988e-05,
|
| 2480 |
+
"loss": 0.0,
|
| 2481 |
+
"step": 352
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 1.1048513302034428,
|
| 2485 |
+
"grad_norm": NaN,
|
| 2486 |
+
"learning_rate": 7.09762847453141e-05,
|
| 2487 |
+
"loss": 0.0,
|
| 2488 |
+
"step": 353
|
| 2489 |
+
},
|
| 2490 |
+
{
|
| 2491 |
+
"epoch": 1.107981220657277,
|
| 2492 |
+
"grad_norm": NaN,
|
| 2493 |
+
"learning_rate": 7.082560147719893e-05,
|
| 2494 |
+
"loss": 0.0,
|
| 2495 |
+
"step": 354
|
| 2496 |
+
},
|
| 2497 |
+
{
|
| 2498 |
+
"epoch": 1.1111111111111112,
|
| 2499 |
+
"grad_norm": NaN,
|
| 2500 |
+
"learning_rate": 7.067468901840937e-05,
|
| 2501 |
+
"loss": 0.0,
|
| 2502 |
+
"step": 355
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"epoch": 1.1142410015649453,
|
| 2506 |
+
"grad_norm": NaN,
|
| 2507 |
+
"learning_rate": 7.052354902977278e-05,
|
| 2508 |
+
"loss": 0.0,
|
| 2509 |
+
"step": 356
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 1.1173708920187793,
|
| 2513 |
+
"grad_norm": NaN,
|
| 2514 |
+
"learning_rate": 7.037218317462051e-05,
|
| 2515 |
+
"loss": 0.0,
|
| 2516 |
+
"step": 357
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 1.1205007824726134,
|
| 2520 |
+
"grad_norm": NaN,
|
| 2521 |
+
"learning_rate": 7.022059311876962e-05,
|
| 2522 |
+
"loss": 0.0,
|
| 2523 |
+
"step": 358
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 1.1236306729264476,
|
| 2527 |
+
"grad_norm": NaN,
|
| 2528 |
+
"learning_rate": 7.006878053050459e-05,
|
| 2529 |
+
"loss": 0.0,
|
| 2530 |
+
"step": 359
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 1.1267605633802817,
|
| 2534 |
+
"grad_norm": NaN,
|
| 2535 |
+
"learning_rate": 6.991674708055885e-05,
|
| 2536 |
+
"loss": 0.0,
|
| 2537 |
+
"step": 360
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 1.1298904538341157,
|
| 2541 |
+
"grad_norm": NaN,
|
| 2542 |
+
"learning_rate": 6.976449444209654e-05,
|
| 2543 |
+
"loss": 0.0,
|
| 2544 |
+
"step": 361
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 1.1330203442879498,
|
| 2548 |
+
"grad_norm": NaN,
|
| 2549 |
+
"learning_rate": 6.961202429069396e-05,
|
| 2550 |
+
"loss": 0.0,
|
| 2551 |
+
"step": 362
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 1.136150234741784,
|
| 2555 |
+
"grad_norm": NaN,
|
| 2556 |
+
"learning_rate": 6.945933830432127e-05,
|
| 2557 |
+
"loss": 0.0,
|
| 2558 |
+
"step": 363
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 1.1392801251956182,
|
| 2562 |
+
"grad_norm": NaN,
|
| 2563 |
+
"learning_rate": 6.930643816332385e-05,
|
| 2564 |
+
"loss": 0.0,
|
| 2565 |
+
"step": 364
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 1.1424100156494523,
|
| 2569 |
+
"grad_norm": NaN,
|
| 2570 |
+
"learning_rate": 6.915332555040396e-05,
|
| 2571 |
+
"loss": 0.0,
|
| 2572 |
+
"step": 365
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 1.1455399061032865,
|
| 2576 |
+
"grad_norm": NaN,
|
| 2577 |
+
"learning_rate": 6.900000215060216e-05,
|
| 2578 |
+
"loss": 0.0,
|
| 2579 |
+
"step": 366
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 1.1486697965571204,
|
| 2583 |
+
"grad_norm": NaN,
|
| 2584 |
+
"learning_rate": 6.884646965127876e-05,
|
| 2585 |
+
"loss": 0.0,
|
| 2586 |
+
"step": 367
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 1.1517996870109546,
|
| 2590 |
+
"grad_norm": NaN,
|
| 2591 |
+
"learning_rate": 6.869272974209525e-05,
|
| 2592 |
+
"loss": 0.0,
|
| 2593 |
+
"step": 368
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 1.1549295774647887,
|
| 2597 |
+
"grad_norm": NaN,
|
| 2598 |
+
"learning_rate": 6.853878411499574e-05,
|
| 2599 |
+
"loss": 0.0,
|
| 2600 |
+
"step": 369
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 1.158059467918623,
|
| 2604 |
+
"grad_norm": NaN,
|
| 2605 |
+
"learning_rate": 6.83846344641883e-05,
|
| 2606 |
+
"loss": 0.0,
|
| 2607 |
+
"step": 370
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 1.1611893583724568,
|
| 2611 |
+
"grad_norm": NaN,
|
| 2612 |
+
"learning_rate": 6.823028248612632e-05,
|
| 2613 |
+
"loss": 0.0,
|
| 2614 |
+
"step": 371
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 1.164319248826291,
|
| 2618 |
+
"grad_norm": NaN,
|
| 2619 |
+
"learning_rate": 6.807572987948986e-05,
|
| 2620 |
+
"loss": 0.0,
|
| 2621 |
+
"step": 372
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 1.1674491392801252,
|
| 2625 |
+
"grad_norm": NaN,
|
| 2626 |
+
"learning_rate": 6.792097834516697e-05,
|
| 2627 |
+
"loss": 0.0,
|
| 2628 |
+
"step": 373
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 1.1705790297339593,
|
| 2632 |
+
"grad_norm": NaN,
|
| 2633 |
+
"learning_rate": 6.776602958623487e-05,
|
| 2634 |
+
"loss": 0.0,
|
| 2635 |
+
"step": 374
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 1.1737089201877935,
|
| 2639 |
+
"grad_norm": NaN,
|
| 2640 |
+
"learning_rate": 6.761088530794136e-05,
|
| 2641 |
+
"loss": 0.0,
|
| 2642 |
+
"step": 375
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 1.1768388106416277,
|
| 2646 |
+
"grad_norm": NaN,
|
| 2647 |
+
"learning_rate": 6.745554721768597e-05,
|
| 2648 |
+
"loss": 0.0,
|
| 2649 |
+
"step": 376
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 1.1799687010954616,
|
| 2653 |
+
"grad_norm": NaN,
|
| 2654 |
+
"learning_rate": 6.730001702500113e-05,
|
| 2655 |
+
"loss": 0.0,
|
| 2656 |
+
"step": 377
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"epoch": 1.1830985915492958,
|
| 2660 |
+
"grad_norm": NaN,
|
| 2661 |
+
"learning_rate": 6.714429644153344e-05,
|
| 2662 |
+
"loss": 0.0,
|
| 2663 |
+
"step": 378
|
| 2664 |
+
},
|
| 2665 |
+
{
|
| 2666 |
+
"epoch": 1.18622848200313,
|
| 2667 |
+
"grad_norm": NaN,
|
| 2668 |
+
"learning_rate": 6.698838718102478e-05,
|
| 2669 |
+
"loss": 0.0,
|
| 2670 |
+
"step": 379
|
| 2671 |
+
},
|
| 2672 |
+
{
|
| 2673 |
+
"epoch": 1.189358372456964,
|
| 2674 |
+
"grad_norm": NaN,
|
| 2675 |
+
"learning_rate": 6.683229095929347e-05,
|
| 2676 |
+
"loss": 0.0,
|
| 2677 |
+
"step": 380
|
| 2678 |
+
},
|
| 2679 |
+
{
|
| 2680 |
+
"epoch": 1.192488262910798,
|
| 2681 |
+
"grad_norm": NaN,
|
| 2682 |
+
"learning_rate": 6.66760094942154e-05,
|
| 2683 |
+
"loss": 0.0,
|
| 2684 |
+
"step": 381
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 1.1956181533646322,
|
| 2688 |
+
"grad_norm": NaN,
|
| 2689 |
+
"learning_rate": 6.651954450570508e-05,
|
| 2690 |
+
"loss": 0.0,
|
| 2691 |
+
"step": 382
|
| 2692 |
+
},
|
| 2693 |
+
{
|
| 2694 |
+
"epoch": 1.1987480438184663,
|
| 2695 |
+
"grad_norm": NaN,
|
| 2696 |
+
"learning_rate": 6.636289771569673e-05,
|
| 2697 |
+
"loss": 0.0,
|
| 2698 |
+
"step": 383
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"epoch": 1.2018779342723005,
|
| 2702 |
+
"grad_norm": NaN,
|
| 2703 |
+
"learning_rate": 6.620607084812538e-05,
|
| 2704 |
+
"loss": 0.0,
|
| 2705 |
+
"step": 384
|
| 2706 |
+
},
|
| 2707 |
+
{
|
| 2708 |
+
"epoch": 1.2050078247261347,
|
| 2709 |
+
"grad_norm": NaN,
|
| 2710 |
+
"learning_rate": 6.604906562890783e-05,
|
| 2711 |
+
"loss": 0.0,
|
| 2712 |
+
"step": 385
|
| 2713 |
+
},
|
| 2714 |
+
{
|
| 2715 |
+
"epoch": 1.2081377151799688,
|
| 2716 |
+
"grad_norm": NaN,
|
| 2717 |
+
"learning_rate": 6.589188378592369e-05,
|
| 2718 |
+
"loss": 0.0,
|
| 2719 |
+
"step": 386
|
| 2720 |
+
},
|
| 2721 |
+
{
|
| 2722 |
+
"epoch": 1.2112676056338028,
|
| 2723 |
+
"grad_norm": NaN,
|
| 2724 |
+
"learning_rate": 6.573452704899633e-05,
|
| 2725 |
+
"loss": 0.0,
|
| 2726 |
+
"step": 387
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 1.214397496087637,
|
| 2730 |
+
"grad_norm": NaN,
|
| 2731 |
+
"learning_rate": 6.557699714987393e-05,
|
| 2732 |
+
"loss": 0.0,
|
| 2733 |
+
"step": 388
|
| 2734 |
+
},
|
| 2735 |
+
{
|
| 2736 |
+
"epoch": 1.217527386541471,
|
| 2737 |
+
"grad_norm": NaN,
|
| 2738 |
+
"learning_rate": 6.541929582221027e-05,
|
| 2739 |
+
"loss": 0.0,
|
| 2740 |
+
"step": 389
|
| 2741 |
+
},
|
| 2742 |
+
{
|
| 2743 |
+
"epoch": 1.2206572769953052,
|
| 2744 |
+
"grad_norm": NaN,
|
| 2745 |
+
"learning_rate": 6.526142480154584e-05,
|
| 2746 |
+
"loss": 0.0,
|
| 2747 |
+
"step": 390
|
| 2748 |
+
},
|
| 2749 |
+
{
|
| 2750 |
+
"epoch": 1.2237871674491392,
|
| 2751 |
+
"grad_norm": NaN,
|
| 2752 |
+
"learning_rate": 6.51033858252886e-05,
|
| 2753 |
+
"loss": 0.0,
|
| 2754 |
+
"step": 391
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 1.2269170579029733,
|
| 2758 |
+
"grad_norm": NaN,
|
| 2759 |
+
"learning_rate": 6.494518063269486e-05,
|
| 2760 |
+
"loss": 0.0,
|
| 2761 |
+
"step": 392
|
| 2762 |
+
},
|
| 2763 |
+
{
|
| 2764 |
+
"epoch": 1.2300469483568075,
|
| 2765 |
+
"grad_norm": NaN,
|
| 2766 |
+
"learning_rate": 6.478681096485022e-05,
|
| 2767 |
+
"loss": 0.0,
|
| 2768 |
+
"step": 393
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 1.2331768388106417,
|
| 2772 |
+
"grad_norm": NaN,
|
| 2773 |
+
"learning_rate": 6.46282785646504e-05,
|
| 2774 |
+
"loss": 0.0,
|
| 2775 |
+
"step": 394
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 1.2363067292644758,
|
| 2779 |
+
"grad_norm": NaN,
|
| 2780 |
+
"learning_rate": 6.446958517678197e-05,
|
| 2781 |
+
"loss": 0.0,
|
| 2782 |
+
"step": 395
|
| 2783 |
+
},
|
| 2784 |
+
{
|
| 2785 |
+
"epoch": 1.2394366197183098,
|
| 2786 |
+
"grad_norm": NaN,
|
| 2787 |
+
"learning_rate": 6.431073254770323e-05,
|
| 2788 |
+
"loss": 0.0,
|
| 2789 |
+
"step": 396
|
| 2790 |
+
},
|
| 2791 |
+
{
|
| 2792 |
+
"epoch": 1.242566510172144,
|
| 2793 |
+
"grad_norm": NaN,
|
| 2794 |
+
"learning_rate": 6.415172242562497e-05,
|
| 2795 |
+
"loss": 0.0,
|
| 2796 |
+
"step": 397
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"epoch": 1.245696400625978,
|
| 2800 |
+
"grad_norm": NaN,
|
| 2801 |
+
"learning_rate": 6.399255656049125e-05,
|
| 2802 |
+
"loss": 0.0,
|
| 2803 |
+
"step": 398
|
| 2804 |
+
},
|
| 2805 |
+
{
|
| 2806 |
+
"epoch": 1.2488262910798122,
|
| 2807 |
+
"grad_norm": NaN,
|
| 2808 |
+
"learning_rate": 6.383323670396006e-05,
|
| 2809 |
+
"loss": 0.0,
|
| 2810 |
+
"step": 399
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 1.2519561815336462,
|
| 2814 |
+
"grad_norm": NaN,
|
| 2815 |
+
"learning_rate": 6.367376460938416e-05,
|
| 2816 |
+
"loss": 0.0,
|
| 2817 |
+
"step": 400
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 1.2550860719874803,
|
| 2821 |
+
"grad_norm": NaN,
|
| 2822 |
+
"learning_rate": 6.35141420317917e-05,
|
| 2823 |
+
"loss": 0.0,
|
| 2824 |
+
"step": 401
|
| 2825 |
+
},
|
| 2826 |
+
{
|
| 2827 |
+
"epoch": 1.2582159624413145,
|
| 2828 |
+
"grad_norm": NaN,
|
| 2829 |
+
"learning_rate": 6.335437072786692e-05,
|
| 2830 |
+
"loss": 0.0,
|
| 2831 |
+
"step": 402
|
| 2832 |
+
},
|
| 2833 |
+
{
|
| 2834 |
+
"epoch": 1.2613458528951487,
|
| 2835 |
+
"grad_norm": NaN,
|
| 2836 |
+
"learning_rate": 6.319445245593087e-05,
|
| 2837 |
+
"loss": 0.0,
|
| 2838 |
+
"step": 403
|
| 2839 |
+
},
|
| 2840 |
+
{
|
| 2841 |
+
"epoch": 1.2644757433489828,
|
| 2842 |
+
"grad_norm": NaN,
|
| 2843 |
+
"learning_rate": 6.303438897592197e-05,
|
| 2844 |
+
"loss": 0.0,
|
| 2845 |
+
"step": 404
|
| 2846 |
+
},
|
| 2847 |
+
{
|
| 2848 |
+
"epoch": 1.267605633802817,
|
| 2849 |
+
"grad_norm": NaN,
|
| 2850 |
+
"learning_rate": 6.28741820493767e-05,
|
| 2851 |
+
"loss": 0.0,
|
| 2852 |
+
"step": 405
|
| 2853 |
+
},
|
| 2854 |
+
{
|
| 2855 |
+
"epoch": 1.2707355242566511,
|
| 2856 |
+
"grad_norm": NaN,
|
| 2857 |
+
"learning_rate": 6.271383343941024e-05,
|
| 2858 |
+
"loss": 0.0,
|
| 2859 |
+
"step": 406
|
| 2860 |
+
},
|
| 2861 |
+
{
|
| 2862 |
+
"epoch": 1.273865414710485,
|
| 2863 |
+
"grad_norm": NaN,
|
| 2864 |
+
"learning_rate": 6.2553344910697e-05,
|
| 2865 |
+
"loss": 0.0,
|
| 2866 |
+
"step": 407
|
| 2867 |
+
},
|
| 2868 |
+
{
|
| 2869 |
+
"epoch": 1.2769953051643192,
|
| 2870 |
+
"grad_norm": NaN,
|
| 2871 |
+
"learning_rate": 6.23927182294512e-05,
|
| 2872 |
+
"loss": 0.0,
|
| 2873 |
+
"step": 408
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 1.2801251956181534,
|
| 2877 |
+
"grad_norm": NaN,
|
| 2878 |
+
"learning_rate": 6.223195516340752e-05,
|
| 2879 |
+
"loss": 0.0,
|
| 2880 |
+
"step": 409
|
| 2881 |
+
},
|
| 2882 |
+
{
|
| 2883 |
+
"epoch": 1.2832550860719873,
|
| 2884 |
+
"grad_norm": NaN,
|
| 2885 |
+
"learning_rate": 6.207105748180158e-05,
|
| 2886 |
+
"loss": 0.0,
|
| 2887 |
+
"step": 410
|
| 2888 |
+
},
|
| 2889 |
+
{
|
| 2890 |
+
"epoch": 1.2863849765258215,
|
| 2891 |
+
"grad_norm": NaN,
|
| 2892 |
+
"learning_rate": 6.19100269553504e-05,
|
| 2893 |
+
"loss": 0.0,
|
| 2894 |
+
"step": 411
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 1.2895148669796557,
|
| 2898 |
+
"grad_norm": NaN,
|
| 2899 |
+
"learning_rate": 6.174886535623307e-05,
|
| 2900 |
+
"loss": 0.0,
|
| 2901 |
+
"step": 412
|
| 2902 |
+
},
|
| 2903 |
+
{
|
| 2904 |
+
"epoch": 1.2926447574334898,
|
| 2905 |
+
"grad_norm": NaN,
|
| 2906 |
+
"learning_rate": 6.158757445807118e-05,
|
| 2907 |
+
"loss": 0.0,
|
| 2908 |
+
"step": 413
|
| 2909 |
+
},
|
| 2910 |
+
{
|
| 2911 |
+
"epoch": 1.295774647887324,
|
| 2912 |
+
"grad_norm": NaN,
|
| 2913 |
+
"learning_rate": 6.142615603590919e-05,
|
| 2914 |
+
"loss": 0.0,
|
| 2915 |
+
"step": 414
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 1.2989045383411582,
|
| 2919 |
+
"grad_norm": NaN,
|
| 2920 |
+
"learning_rate": 6.126461186619508e-05,
|
| 2921 |
+
"loss": 0.0,
|
| 2922 |
+
"step": 415
|
| 2923 |
+
},
|
| 2924 |
+
{
|
| 2925 |
+
"epoch": 1.302034428794992,
|
| 2926 |
+
"grad_norm": NaN,
|
| 2927 |
+
"learning_rate": 6.110294372676065e-05,
|
| 2928 |
+
"loss": 0.0,
|
| 2929 |
+
"step": 416
|
| 2930 |
+
},
|
| 2931 |
+
{
|
| 2932 |
+
"epoch": 1.3051643192488263,
|
| 2933 |
+
"grad_norm": NaN,
|
| 2934 |
+
"learning_rate": 6.094115339680209e-05,
|
| 2935 |
+
"loss": 0.0,
|
| 2936 |
+
"step": 417
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 1.3082942097026604,
|
| 2940 |
+
"grad_norm": NaN,
|
| 2941 |
+
"learning_rate": 6.077924265686027e-05,
|
| 2942 |
+
"loss": 0.0,
|
| 2943 |
+
"step": 418
|
| 2944 |
+
},
|
| 2945 |
+
{
|
| 2946 |
+
"epoch": 1.3114241001564946,
|
| 2947 |
+
"grad_norm": NaN,
|
| 2948 |
+
"learning_rate": 6.061721328880119e-05,
|
| 2949 |
+
"loss": 0.0,
|
| 2950 |
+
"step": 419
|
| 2951 |
+
},
|
| 2952 |
+
{
|
| 2953 |
+
"epoch": 1.3145539906103285,
|
| 2954 |
+
"grad_norm": NaN,
|
| 2955 |
+
"learning_rate": 6.0455067075796424e-05,
|
| 2956 |
+
"loss": 0.0,
|
| 2957 |
+
"step": 420
|
| 2958 |
+
},
|
| 2959 |
+
{
|
| 2960 |
+
"epoch": 1.3176838810641627,
|
| 2961 |
+
"grad_norm": NaN,
|
| 2962 |
+
"learning_rate": 6.029280580230342e-05,
|
| 2963 |
+
"loss": 0.0,
|
| 2964 |
+
"step": 421
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 1.3208137715179968,
|
| 2968 |
+
"grad_norm": NaN,
|
| 2969 |
+
"learning_rate": 6.0130431254045904e-05,
|
| 2970 |
+
"loss": 0.0,
|
| 2971 |
+
"step": 422
|
| 2972 |
+
},
|
| 2973 |
+
{
|
| 2974 |
+
"epoch": 1.323943661971831,
|
| 2975 |
+
"grad_norm": NaN,
|
| 2976 |
+
"learning_rate": 5.996794521799425e-05,
|
| 2977 |
+
"loss": 0.0,
|
| 2978 |
+
"step": 423
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 1.3270735524256652,
|
| 2982 |
+
"grad_norm": NaN,
|
| 2983 |
+
"learning_rate": 5.9805349482345706e-05,
|
| 2984 |
+
"loss": 0.0,
|
| 2985 |
+
"step": 424
|
| 2986 |
+
},
|
| 2987 |
+
{
|
| 2988 |
+
"epoch": 1.3302034428794993,
|
| 2989 |
+
"grad_norm": NaN,
|
| 2990 |
+
"learning_rate": 5.964264583650486e-05,
|
| 2991 |
+
"loss": 0.0,
|
| 2992 |
+
"step": 425
|
| 2993 |
+
},
|
| 2994 |
+
{
|
| 2995 |
+
"epoch": 1.3333333333333333,
|
| 2996 |
+
"grad_norm": NaN,
|
| 2997 |
+
"learning_rate": 5.947983607106385e-05,
|
| 2998 |
+
"loss": 0.0,
|
| 2999 |
+
"step": 426
|
| 3000 |
+
},
|
| 3001 |
+
{
|
| 3002 |
+
"epoch": 1.3364632237871674,
|
| 3003 |
+
"grad_norm": NaN,
|
| 3004 |
+
"learning_rate": 5.931692197778269e-05,
|
| 3005 |
+
"loss": 0.0,
|
| 3006 |
+
"step": 427
|
| 3007 |
+
},
|
| 3008 |
+
{
|
| 3009 |
+
"epoch": 1.3395931142410016,
|
| 3010 |
+
"grad_norm": NaN,
|
| 3011 |
+
"learning_rate": 5.9153905349569525e-05,
|
| 3012 |
+
"loss": 0.0,
|
| 3013 |
+
"step": 428
|
| 3014 |
+
},
|
| 3015 |
+
{
|
| 3016 |
+
"epoch": 1.3427230046948357,
|
| 3017 |
+
"grad_norm": NaN,
|
| 3018 |
+
"learning_rate": 5.8990787980460974e-05,
|
| 3019 |
+
"loss": 0.0,
|
| 3020 |
+
"step": 429
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 1.3458528951486697,
|
| 3024 |
+
"grad_norm": NaN,
|
| 3025 |
+
"learning_rate": 5.882757166560226e-05,
|
| 3026 |
+
"loss": 0.0,
|
| 3027 |
+
"step": 430
|
| 3028 |
+
},
|
| 3029 |
+
{
|
| 3030 |
+
"epoch": 1.3489827856025038,
|
| 3031 |
+
"grad_norm": NaN,
|
| 3032 |
+
"learning_rate": 5.866425820122758e-05,
|
| 3033 |
+
"loss": 0.0,
|
| 3034 |
+
"step": 431
|
| 3035 |
+
},
|
| 3036 |
+
{
|
| 3037 |
+
"epoch": 1.352112676056338,
|
| 3038 |
+
"grad_norm": NaN,
|
| 3039 |
+
"learning_rate": 5.8500849384640285e-05,
|
| 3040 |
+
"loss": 0.0,
|
| 3041 |
+
"step": 432
|
| 3042 |
+
},
|
| 3043 |
+
{
|
| 3044 |
+
"epoch": 1.3552425665101722,
|
| 3045 |
+
"grad_norm": NaN,
|
| 3046 |
+
"learning_rate": 5.833734701419308e-05,
|
| 3047 |
+
"loss": 0.0,
|
| 3048 |
+
"step": 433
|
| 3049 |
+
},
|
| 3050 |
+
{
|
| 3051 |
+
"epoch": 1.3583724569640063,
|
| 3052 |
+
"grad_norm": NaN,
|
| 3053 |
+
"learning_rate": 5.817375288926825e-05,
|
| 3054 |
+
"loss": 0.0,
|
| 3055 |
+
"step": 434
|
| 3056 |
+
},
|
| 3057 |
+
{
|
| 3058 |
+
"epoch": 1.3615023474178405,
|
| 3059 |
+
"grad_norm": NaN,
|
| 3060 |
+
"learning_rate": 5.801006881025788e-05,
|
| 3061 |
+
"loss": 0.0,
|
| 3062 |
+
"step": 435
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 1.3646322378716744,
|
| 3066 |
+
"grad_norm": NaN,
|
| 3067 |
+
"learning_rate": 5.784629657854399e-05,
|
| 3068 |
+
"loss": 0.0,
|
| 3069 |
+
"step": 436
|
| 3070 |
+
},
|
| 3071 |
+
{
|
| 3072 |
+
"epoch": 1.3677621283255086,
|
| 3073 |
+
"grad_norm": NaN,
|
| 3074 |
+
"learning_rate": 5.768243799647879e-05,
|
| 3075 |
+
"loss": 0.0,
|
| 3076 |
+
"step": 437
|
| 3077 |
+
},
|
| 3078 |
+
{
|
| 3079 |
+
"epoch": 1.3708920187793427,
|
| 3080 |
+
"grad_norm": NaN,
|
| 3081 |
+
"learning_rate": 5.7518494867364725e-05,
|
| 3082 |
+
"loss": 0.0,
|
| 3083 |
+
"step": 438
|
| 3084 |
+
},
|
| 3085 |
+
{
|
| 3086 |
+
"epoch": 1.374021909233177,
|
| 3087 |
+
"grad_norm": NaN,
|
| 3088 |
+
"learning_rate": 5.7354468995434794e-05,
|
| 3089 |
+
"loss": 0.0,
|
| 3090 |
+
"step": 439
|
| 3091 |
+
},
|
| 3092 |
+
{
|
| 3093 |
+
"epoch": 1.3771517996870108,
|
| 3094 |
+
"grad_norm": NaN,
|
| 3095 |
+
"learning_rate": 5.71903621858325e-05,
|
| 3096 |
+
"loss": 0.0,
|
| 3097 |
+
"step": 440
|
| 3098 |
+
},
|
| 3099 |
+
{
|
| 3100 |
+
"epoch": 1.380281690140845,
|
| 3101 |
+
"grad_norm": NaN,
|
| 3102 |
+
"learning_rate": 5.7026176244592155e-05,
|
| 3103 |
+
"loss": 0.0,
|
| 3104 |
+
"step": 441
|
| 3105 |
+
},
|
| 3106 |
+
{
|
| 3107 |
+
"epoch": 1.3834115805946792,
|
| 3108 |
+
"grad_norm": NaN,
|
| 3109 |
+
"learning_rate": 5.686191297861892e-05,
|
| 3110 |
+
"loss": 0.0,
|
| 3111 |
+
"step": 442
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 1.3865414710485133,
|
| 3115 |
+
"grad_norm": NaN,
|
| 3116 |
+
"learning_rate": 5.66975741956689e-05,
|
| 3117 |
+
"loss": 0.0,
|
| 3118 |
+
"step": 443
|
| 3119 |
+
},
|
| 3120 |
+
{
|
| 3121 |
+
"epoch": 1.3896713615023475,
|
| 3122 |
+
"grad_norm": NaN,
|
| 3123 |
+
"learning_rate": 5.653316170432932e-05,
|
| 3124 |
+
"loss": 0.0,
|
| 3125 |
+
"step": 444
|
| 3126 |
+
},
|
| 3127 |
+
{
|
| 3128 |
+
"epoch": 1.3928012519561817,
|
| 3129 |
+
"grad_norm": NaN,
|
| 3130 |
+
"learning_rate": 5.636867731399856e-05,
|
| 3131 |
+
"loss": 0.0,
|
| 3132 |
+
"step": 445
|
| 3133 |
+
},
|
| 3134 |
+
{
|
| 3135 |
+
"epoch": 1.3959311424100156,
|
| 3136 |
+
"grad_norm": NaN,
|
| 3137 |
+
"learning_rate": 5.620412283486629e-05,
|
| 3138 |
+
"loss": 0.0,
|
| 3139 |
+
"step": 446
|
| 3140 |
+
},
|
| 3141 |
+
{
|
| 3142 |
+
"epoch": 1.3990610328638498,
|
| 3143 |
+
"grad_norm": NaN,
|
| 3144 |
+
"learning_rate": 5.603950007789349e-05,
|
| 3145 |
+
"loss": 0.0,
|
| 3146 |
+
"step": 447
|
| 3147 |
+
},
|
| 3148 |
+
{
|
| 3149 |
+
"epoch": 1.402190923317684,
|
| 3150 |
+
"grad_norm": NaN,
|
| 3151 |
+
"learning_rate": 5.5874810854792606e-05,
|
| 3152 |
+
"loss": 0.0,
|
| 3153 |
+
"step": 448
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 1.405320813771518,
|
| 3157 |
+
"grad_norm": NaN,
|
| 3158 |
+
"learning_rate": 5.571005697800748e-05,
|
| 3159 |
+
"loss": 0.0,
|
| 3160 |
+
"step": 449
|
| 3161 |
+
},
|
| 3162 |
+
{
|
| 3163 |
+
"epoch": 1.408450704225352,
|
| 3164 |
+
"grad_norm": NaN,
|
| 3165 |
+
"learning_rate": 5.554524026069354e-05,
|
| 3166 |
+
"loss": 0.0,
|
| 3167 |
+
"step": 450
|
| 3168 |
+
},
|
| 3169 |
+
{
|
| 3170 |
+
"epoch": 1.4115805946791862,
|
| 3171 |
+
"grad_norm": NaN,
|
| 3172 |
+
"learning_rate": 5.5380362516697794e-05,
|
| 3173 |
+
"loss": 0.0,
|
| 3174 |
+
"step": 451
|
| 3175 |
+
},
|
| 3176 |
+
{
|
| 3177 |
+
"epoch": 1.4147104851330203,
|
| 3178 |
+
"grad_norm": NaN,
|
| 3179 |
+
"learning_rate": 5.521542556053885e-05,
|
| 3180 |
+
"loss": 0.0,
|
| 3181 |
+
"step": 452
|
| 3182 |
+
},
|
| 3183 |
+
{
|
| 3184 |
+
"epoch": 1.4178403755868545,
|
| 3185 |
+
"grad_norm": NaN,
|
| 3186 |
+
"learning_rate": 5.505043120738693e-05,
|
| 3187 |
+
"loss": 0.0,
|
| 3188 |
+
"step": 453
|
| 3189 |
+
},
|
| 3190 |
+
{
|
| 3191 |
+
"epoch": 1.4209702660406887,
|
| 3192 |
+
"grad_norm": NaN,
|
| 3193 |
+
"learning_rate": 5.488538127304399e-05,
|
| 3194 |
+
"loss": 0.0,
|
| 3195 |
+
"step": 454
|
| 3196 |
+
},
|
| 3197 |
+
{
|
| 3198 |
+
"epoch": 1.4241001564945228,
|
| 3199 |
+
"grad_norm": NaN,
|
| 3200 |
+
"learning_rate": 5.4720277573923595e-05,
|
| 3201 |
+
"loss": 0.0,
|
| 3202 |
+
"step": 455
|
| 3203 |
+
},
|
| 3204 |
+
{
|
| 3205 |
+
"epoch": 1.4272300469483568,
|
| 3206 |
+
"grad_norm": NaN,
|
| 3207 |
+
"learning_rate": 5.455512192703105e-05,
|
| 3208 |
+
"loss": 0.0,
|
| 3209 |
+
"step": 456
|
| 3210 |
+
},
|
| 3211 |
+
{
|
| 3212 |
+
"epoch": 1.430359937402191,
|
| 3213 |
+
"grad_norm": NaN,
|
| 3214 |
+
"learning_rate": 5.4389916149943354e-05,
|
| 3215 |
+
"loss": 0.0,
|
| 3216 |
+
"step": 457
|
| 3217 |
+
},
|
| 3218 |
+
{
|
| 3219 |
+
"epoch": 1.433489827856025,
|
| 3220 |
+
"grad_norm": NaN,
|
| 3221 |
+
"learning_rate": 5.422466206078919e-05,
|
| 3222 |
+
"loss": 0.0,
|
| 3223 |
+
"step": 458
|
| 3224 |
+
},
|
| 3225 |
+
{
|
| 3226 |
+
"epoch": 1.436619718309859,
|
| 3227 |
+
"grad_norm": NaN,
|
| 3228 |
+
"learning_rate": 5.405936147822892e-05,
|
| 3229 |
+
"loss": 0.0,
|
| 3230 |
+
"step": 459
|
| 3231 |
+
},
|
| 3232 |
+
{
|
| 3233 |
+
"epoch": 1.4397496087636932,
|
| 3234 |
+
"grad_norm": NaN,
|
| 3235 |
+
"learning_rate": 5.3894016221434605e-05,
|
| 3236 |
+
"loss": 0.0,
|
| 3237 |
+
"step": 460
|
| 3238 |
+
},
|
| 3239 |
+
{
|
| 3240 |
+
"epoch": 1.4428794992175273,
|
| 3241 |
+
"grad_norm": NaN,
|
| 3242 |
+
"learning_rate": 5.372862811006992e-05,
|
| 3243 |
+
"loss": 0.0,
|
| 3244 |
+
"step": 461
|
| 3245 |
+
},
|
| 3246 |
+
{
|
| 3247 |
+
"epoch": 1.4460093896713615,
|
| 3248 |
+
"grad_norm": NaN,
|
| 3249 |
+
"learning_rate": 5.35631989642702e-05,
|
| 3250 |
+
"loss": 0.0,
|
| 3251 |
+
"step": 462
|
| 3252 |
+
},
|
| 3253 |
+
{
|
| 3254 |
+
"epoch": 1.4491392801251957,
|
| 3255 |
+
"grad_norm": NaN,
|
| 3256 |
+
"learning_rate": 5.3397730604622344e-05,
|
| 3257 |
+
"loss": 0.0,
|
| 3258 |
+
"step": 463
|
| 3259 |
+
},
|
| 3260 |
+
{
|
| 3261 |
+
"epoch": 1.4522691705790298,
|
| 3262 |
+
"grad_norm": NaN,
|
| 3263 |
+
"learning_rate": 5.323222485214484e-05,
|
| 3264 |
+
"loss": 0.0,
|
| 3265 |
+
"step": 464
|
| 3266 |
+
},
|
| 3267 |
+
{
|
| 3268 |
+
"epoch": 1.455399061032864,
|
| 3269 |
+
"grad_norm": NaN,
|
| 3270 |
+
"learning_rate": 5.306668352826765e-05,
|
| 3271 |
+
"loss": 0.0,
|
| 3272 |
+
"step": 465
|
| 3273 |
+
},
|
| 3274 |
+
{
|
| 3275 |
+
"epoch": 1.458528951486698,
|
| 3276 |
+
"grad_norm": NaN,
|
| 3277 |
+
"learning_rate": 5.290110845481224e-05,
|
| 3278 |
+
"loss": 0.0,
|
| 3279 |
+
"step": 466
|
| 3280 |
+
},
|
| 3281 |
+
{
|
| 3282 |
+
"epoch": 1.461658841940532,
|
| 3283 |
+
"grad_norm": NaN,
|
| 3284 |
+
"learning_rate": 5.2735501453971516e-05,
|
| 3285 |
+
"loss": 0.0,
|
| 3286 |
+
"step": 467
|
| 3287 |
+
},
|
| 3288 |
+
{
|
| 3289 |
+
"epoch": 1.4647887323943662,
|
| 3290 |
+
"grad_norm": NaN,
|
| 3291 |
+
"learning_rate": 5.256986434828969e-05,
|
| 3292 |
+
"loss": 0.0,
|
| 3293 |
+
"step": 468
|
| 3294 |
+
},
|
| 3295 |
+
{
|
| 3296 |
+
"epoch": 1.4679186228482002,
|
| 3297 |
+
"grad_norm": NaN,
|
| 3298 |
+
"learning_rate": 5.240419896064235e-05,
|
| 3299 |
+
"loss": 0.0,
|
| 3300 |
+
"step": 469
|
| 3301 |
+
},
|
| 3302 |
+
{
|
| 3303 |
+
"epoch": 1.4710485133020343,
|
| 3304 |
+
"grad_norm": NaN,
|
| 3305 |
+
"learning_rate": 5.2238507114216285e-05,
|
| 3306 |
+
"loss": 0.0,
|
| 3307 |
+
"step": 470
|
| 3308 |
+
},
|
| 3309 |
+
{
|
| 3310 |
+
"epoch": 1.4741784037558685,
|
| 3311 |
+
"grad_norm": NaN,
|
| 3312 |
+
"learning_rate": 5.2072790632489497e-05,
|
| 3313 |
+
"loss": 0.0,
|
| 3314 |
+
"step": 471
|
| 3315 |
+
},
|
| 3316 |
+
{
|
| 3317 |
+
"epoch": 1.4773082942097027,
|
| 3318 |
+
"grad_norm": NaN,
|
| 3319 |
+
"learning_rate": 5.19070513392111e-05,
|
| 3320 |
+
"loss": 0.0,
|
| 3321 |
+
"step": 472
|
| 3322 |
+
},
|
| 3323 |
+
{
|
| 3324 |
+
"epoch": 1.4804381846635368,
|
| 3325 |
+
"grad_norm": NaN,
|
| 3326 |
+
"learning_rate": 5.174129105838127e-05,
|
| 3327 |
+
"loss": 0.0,
|
| 3328 |
+
"step": 473
|
| 3329 |
+
},
|
| 3330 |
+
{
|
| 3331 |
+
"epoch": 1.483568075117371,
|
| 3332 |
+
"grad_norm": NaN,
|
| 3333 |
+
"learning_rate": 5.1575511614231096e-05,
|
| 3334 |
+
"loss": 0.0,
|
| 3335 |
+
"step": 474
|
| 3336 |
+
},
|
| 3337 |
+
{
|
| 3338 |
+
"epoch": 1.486697965571205,
|
| 3339 |
+
"grad_norm": NaN,
|
| 3340 |
+
"learning_rate": 5.140971483120265e-05,
|
| 3341 |
+
"loss": 0.0,
|
| 3342 |
+
"step": 475
|
| 3343 |
+
},
|
| 3344 |
+
{
|
| 3345 |
+
"epoch": 1.489827856025039,
|
| 3346 |
+
"grad_norm": NaN,
|
| 3347 |
+
"learning_rate": 5.1243902533928754e-05,
|
| 3348 |
+
"loss": 0.0,
|
| 3349 |
+
"step": 476
|
| 3350 |
+
},
|
| 3351 |
+
{
|
| 3352 |
+
"epoch": 1.4929577464788732,
|
| 3353 |
+
"grad_norm": NaN,
|
| 3354 |
+
"learning_rate": 5.1078076547212994e-05,
|
| 3355 |
+
"loss": 0.0,
|
| 3356 |
+
"step": 477
|
| 3357 |
+
},
|
| 3358 |
+
{
|
| 3359 |
+
"epoch": 1.4960876369327074,
|
| 3360 |
+
"grad_norm": NaN,
|
| 3361 |
+
"learning_rate": 5.0912238696009616e-05,
|
| 3362 |
+
"loss": 0.0,
|
| 3363 |
+
"step": 478
|
| 3364 |
+
},
|
| 3365 |
+
{
|
| 3366 |
+
"epoch": 1.4992175273865413,
|
| 3367 |
+
"grad_norm": NaN,
|
| 3368 |
+
"learning_rate": 5.0746390805403445e-05,
|
| 3369 |
+
"loss": 0.0,
|
| 3370 |
+
"step": 479
|
| 3371 |
+
},
|
| 3372 |
+
{
|
| 3373 |
+
"epoch": 1.5023474178403755,
|
| 3374 |
+
"grad_norm": NaN,
|
| 3375 |
+
"learning_rate": 5.058053470058974e-05,
|
| 3376 |
+
"loss": 0.0,
|
| 3377 |
+
"step": 480
|
| 3378 |
+
},
|
| 3379 |
+
{
|
| 3380 |
+
"epoch": 1.5054773082942097,
|
| 3381 |
+
"grad_norm": NaN,
|
| 3382 |
+
"learning_rate": 5.041467220685424e-05,
|
| 3383 |
+
"loss": 0.0,
|
| 3384 |
+
"step": 481
|
| 3385 |
+
},
|
| 3386 |
+
{
|
| 3387 |
+
"epoch": 1.5086071987480438,
|
| 3388 |
+
"grad_norm": NaN,
|
| 3389 |
+
"learning_rate": 5.024880514955292e-05,
|
| 3390 |
+
"loss": 0.0,
|
| 3391 |
+
"step": 482
|
| 3392 |
+
},
|
| 3393 |
+
{
|
| 3394 |
+
"epoch": 1.511737089201878,
|
| 3395 |
+
"grad_norm": NaN,
|
| 3396 |
+
"learning_rate": 5.0082935354092044e-05,
|
| 3397 |
+
"loss": 0.0,
|
| 3398 |
+
"step": 483
|
| 3399 |
+
},
|
| 3400 |
+
{
|
| 3401 |
+
"epoch": 1.5148669796557122,
|
| 3402 |
+
"grad_norm": NaN,
|
| 3403 |
+
"learning_rate": 4.991706464590797e-05,
|
| 3404 |
+
"loss": 0.0,
|
| 3405 |
+
"step": 484
|
| 3406 |
+
},
|
| 3407 |
+
{
|
| 3408 |
+
"epoch": 1.5179968701095463,
|
| 3409 |
+
"grad_norm": NaN,
|
| 3410 |
+
"learning_rate": 4.975119485044709e-05,
|
| 3411 |
+
"loss": 0.0,
|
| 3412 |
+
"step": 485
|
| 3413 |
+
},
|
| 3414 |
+
{
|
| 3415 |
+
"epoch": 1.5211267605633803,
|
| 3416 |
+
"grad_norm": NaN,
|
| 3417 |
+
"learning_rate": 4.958532779314578e-05,
|
| 3418 |
+
"loss": 0.0,
|
| 3419 |
+
"step": 486
|
| 3420 |
+
},
|
| 3421 |
+
{
|
| 3422 |
+
"epoch": 1.5242566510172144,
|
| 3423 |
+
"grad_norm": NaN,
|
| 3424 |
+
"learning_rate": 4.9419465299410264e-05,
|
| 3425 |
+
"loss": 0.0,
|
| 3426 |
+
"step": 487
|
| 3427 |
+
},
|
| 3428 |
+
{
|
| 3429 |
+
"epoch": 1.5273865414710484,
|
| 3430 |
+
"grad_norm": NaN,
|
| 3431 |
+
"learning_rate": 4.9253609194596574e-05,
|
| 3432 |
+
"loss": 0.0,
|
| 3433 |
+
"step": 488
|
| 3434 |
+
},
|
| 3435 |
+
{
|
| 3436 |
+
"epoch": 1.5305164319248825,
|
| 3437 |
+
"grad_norm": NaN,
|
| 3438 |
+
"learning_rate": 4.908776130399039e-05,
|
| 3439 |
+
"loss": 0.0,
|
| 3440 |
+
"step": 489
|
| 3441 |
+
},
|
| 3442 |
+
{
|
| 3443 |
+
"epoch": 1.5336463223787167,
|
| 3444 |
+
"grad_norm": NaN,
|
| 3445 |
+
"learning_rate": 4.8921923452787025e-05,
|
| 3446 |
+
"loss": 0.0,
|
| 3447 |
+
"step": 490
|
| 3448 |
+
},
|
| 3449 |
+
{
|
| 3450 |
+
"epoch": 1.5367762128325508,
|
| 3451 |
+
"grad_norm": NaN,
|
| 3452 |
+
"learning_rate": 4.875609746607127e-05,
|
| 3453 |
+
"loss": 0.0,
|
| 3454 |
+
"step": 491
|
| 3455 |
+
},
|
| 3456 |
+
{
|
| 3457 |
+
"epoch": 1.539906103286385,
|
| 3458 |
+
"grad_norm": NaN,
|
| 3459 |
+
"learning_rate": 4.859028516879737e-05,
|
| 3460 |
+
"loss": 0.0,
|
| 3461 |
+
"step": 492
|
| 3462 |
+
},
|
| 3463 |
+
{
|
| 3464 |
+
"epoch": 1.5430359937402192,
|
| 3465 |
+
"grad_norm": NaN,
|
| 3466 |
+
"learning_rate": 4.842448838576891e-05,
|
| 3467 |
+
"loss": 0.0,
|
| 3468 |
+
"step": 493
|
| 3469 |
+
},
|
| 3470 |
+
{
|
| 3471 |
+
"epoch": 1.5461658841940533,
|
| 3472 |
+
"grad_norm": NaN,
|
| 3473 |
+
"learning_rate": 4.825870894161874e-05,
|
| 3474 |
+
"loss": 0.0,
|
| 3475 |
+
"step": 494
|
| 3476 |
+
},
|
| 3477 |
+
{
|
| 3478 |
+
"epoch": 1.5492957746478875,
|
| 3479 |
+
"grad_norm": NaN,
|
| 3480 |
+
"learning_rate": 4.8092948660788894e-05,
|
| 3481 |
+
"loss": 0.0,
|
| 3482 |
+
"step": 495
|
| 3483 |
+
},
|
| 3484 |
+
{
|
| 3485 |
+
"epoch": 1.5524256651017214,
|
| 3486 |
+
"grad_norm": NaN,
|
| 3487 |
+
"learning_rate": 4.79272093675105e-05,
|
| 3488 |
+
"loss": 0.0,
|
| 3489 |
+
"step": 496
|
| 3490 |
+
},
|
| 3491 |
+
{
|
| 3492 |
+
"epoch": 1.5555555555555556,
|
| 3493 |
+
"grad_norm": NaN,
|
| 3494 |
+
"learning_rate": 4.7761492885783734e-05,
|
| 3495 |
+
"loss": 0.0,
|
| 3496 |
+
"step": 497
|
| 3497 |
+
},
|
| 3498 |
+
{
|
| 3499 |
+
"epoch": 1.5586854460093895,
|
| 3500 |
+
"grad_norm": NaN,
|
| 3501 |
+
"learning_rate": 4.759580103935767e-05,
|
| 3502 |
+
"loss": 0.0,
|
| 3503 |
+
"step": 498
|
| 3504 |
+
},
|
| 3505 |
+
{
|
| 3506 |
+
"epoch": 1.5618153364632237,
|
| 3507 |
+
"grad_norm": NaN,
|
| 3508 |
+
"learning_rate": 4.7430135651710315e-05,
|
| 3509 |
+
"loss": 0.0,
|
| 3510 |
+
"step": 499
|
| 3511 |
+
},
|
| 3512 |
+
{
|
| 3513 |
+
"epoch": 1.5649452269170578,
|
| 3514 |
+
"grad_norm": NaN,
|
| 3515 |
+
"learning_rate": 4.7264498546028496e-05,
|
| 3516 |
+
"loss": 0.0,
|
| 3517 |
+
"step": 500
|
| 3518 |
+
},
|
| 3519 |
+
{
|
| 3520 |
+
"epoch": 1.568075117370892,
|
| 3521 |
+
"grad_norm": NaN,
|
| 3522 |
+
"learning_rate": 4.7098891545187755e-05,
|
| 3523 |
+
"loss": 0.0,
|
| 3524 |
+
"step": 501
|
| 3525 |
+
},
|
| 3526 |
+
{
|
| 3527 |
+
"epoch": 1.5712050078247262,
|
| 3528 |
+
"grad_norm": NaN,
|
| 3529 |
+
"learning_rate": 4.693331647173235e-05,
|
| 3530 |
+
"loss": 0.0,
|
| 3531 |
+
"step": 502
|
| 3532 |
+
},
|
| 3533 |
+
{
|
| 3534 |
+
"epoch": 1.5743348982785603,
|
| 3535 |
+
"grad_norm": NaN,
|
| 3536 |
+
"learning_rate": 4.6767775147855186e-05,
|
| 3537 |
+
"loss": 0.0,
|
| 3538 |
+
"step": 503
|
| 3539 |
+
},
|
| 3540 |
+
{
|
| 3541 |
+
"epoch": 1.5774647887323945,
|
| 3542 |
+
"grad_norm": NaN,
|
| 3543 |
+
"learning_rate": 4.660226939537767e-05,
|
| 3544 |
+
"loss": 0.0,
|
| 3545 |
+
"step": 504
|
| 3546 |
+
},
|
| 3547 |
+
{
|
| 3548 |
+
"epoch": 1.5805946791862286,
|
| 3549 |
+
"grad_norm": NaN,
|
| 3550 |
+
"learning_rate": 4.643680103572981e-05,
|
| 3551 |
+
"loss": 0.0,
|
| 3552 |
+
"step": 505
|
| 3553 |
+
},
|
| 3554 |
+
{
|
| 3555 |
+
"epoch": 1.5837245696400626,
|
| 3556 |
+
"grad_norm": NaN,
|
| 3557 |
+
"learning_rate": 4.627137188993009e-05,
|
| 3558 |
+
"loss": 0.0,
|
| 3559 |
+
"step": 506
|
| 3560 |
+
},
|
| 3561 |
+
{
|
| 3562 |
+
"epoch": 1.5868544600938967,
|
| 3563 |
+
"grad_norm": NaN,
|
| 3564 |
+
"learning_rate": 4.6105983778565406e-05,
|
| 3565 |
+
"loss": 0.0,
|
| 3566 |
+
"step": 507
|
| 3567 |
+
},
|
| 3568 |
+
{
|
| 3569 |
+
"epoch": 1.5899843505477307,
|
| 3570 |
+
"grad_norm": NaN,
|
| 3571 |
+
"learning_rate": 4.594063852177108e-05,
|
| 3572 |
+
"loss": 0.0,
|
| 3573 |
+
"step": 508
|
| 3574 |
+
},
|
| 3575 |
+
{
|
| 3576 |
+
"epoch": 1.5931142410015648,
|
| 3577 |
+
"grad_norm": NaN,
|
| 3578 |
+
"learning_rate": 4.577533793921083e-05,
|
| 3579 |
+
"loss": 0.0,
|
| 3580 |
+
"step": 509
|
| 3581 |
+
},
|
| 3582 |
+
{
|
| 3583 |
+
"epoch": 1.596244131455399,
|
| 3584 |
+
"grad_norm": NaN,
|
| 3585 |
+
"learning_rate": 4.5610083850056664e-05,
|
| 3586 |
+
"loss": 0.0,
|
| 3587 |
+
"step": 510
|
| 3588 |
+
},
|
| 3589 |
+
{
|
| 3590 |
+
"epoch": 1.5993740219092332,
|
| 3591 |
+
"grad_norm": NaN,
|
| 3592 |
+
"learning_rate": 4.5444878072968963e-05,
|
| 3593 |
+
"loss": 0.0,
|
| 3594 |
+
"step": 511
|
| 3595 |
+
},
|
| 3596 |
+
{
|
| 3597 |
+
"epoch": 1.6025039123630673,
|
| 3598 |
+
"grad_norm": NaN,
|
| 3599 |
+
"learning_rate": 4.527972242607642e-05,
|
| 3600 |
+
"loss": 0.0,
|
| 3601 |
+
"step": 512
|
| 3602 |
+
},
|
| 3603 |
+
{
|
| 3604 |
+
"epoch": 1.6056338028169015,
|
| 3605 |
+
"grad_norm": NaN,
|
| 3606 |
+
"learning_rate": 4.511461872695602e-05,
|
| 3607 |
+
"loss": 0.0,
|
| 3608 |
+
"step": 513
|
| 3609 |
+
},
|
| 3610 |
+
{
|
| 3611 |
+
"epoch": 1.6087636932707357,
|
| 3612 |
+
"grad_norm": NaN,
|
| 3613 |
+
"learning_rate": 4.4949568792613066e-05,
|
| 3614 |
+
"loss": 0.0,
|
| 3615 |
+
"step": 514
|
| 3616 |
+
},
|
| 3617 |
+
{
|
| 3618 |
+
"epoch": 1.6118935837245696,
|
| 3619 |
+
"grad_norm": NaN,
|
| 3620 |
+
"learning_rate": 4.478457443946117e-05,
|
| 3621 |
+
"loss": 0.0,
|
| 3622 |
+
"step": 515
|
| 3623 |
+
},
|
| 3624 |
+
{
|
| 3625 |
+
"epoch": 1.6150234741784038,
|
| 3626 |
+
"grad_norm": NaN,
|
| 3627 |
+
"learning_rate": 4.461963748330221e-05,
|
| 3628 |
+
"loss": 0.0,
|
| 3629 |
+
"step": 516
|
| 3630 |
+
},
|
| 3631 |
+
{
|
| 3632 |
+
"epoch": 1.618153364632238,
|
| 3633 |
+
"grad_norm": NaN,
|
| 3634 |
+
"learning_rate": 4.445475973930647e-05,
|
| 3635 |
+
"loss": 0.0,
|
| 3636 |
+
"step": 517
|
| 3637 |
+
},
|
| 3638 |
+
{
|
| 3639 |
+
"epoch": 1.6212832550860719,
|
| 3640 |
+
"grad_norm": NaN,
|
| 3641 |
+
"learning_rate": 4.428994302199254e-05,
|
| 3642 |
+
"loss": 0.0,
|
| 3643 |
+
"step": 518
|
| 3644 |
+
},
|
| 3645 |
+
{
|
| 3646 |
+
"epoch": 1.624413145539906,
|
| 3647 |
+
"grad_norm": NaN,
|
| 3648 |
+
"learning_rate": 4.412518914520741e-05,
|
| 3649 |
+
"loss": 0.0,
|
| 3650 |
+
"step": 519
|
| 3651 |
+
},
|
| 3652 |
+
{
|
| 3653 |
+
"epoch": 1.6275430359937402,
|
| 3654 |
+
"grad_norm": NaN,
|
| 3655 |
+
"learning_rate": 4.396049992210652e-05,
|
| 3656 |
+
"loss": 0.0,
|
| 3657 |
+
"step": 520
|
| 3658 |
+
},
|
| 3659 |
+
{
|
| 3660 |
+
"epoch": 1.6306729264475743,
|
| 3661 |
+
"grad_norm": NaN,
|
| 3662 |
+
"learning_rate": 4.379587716513372e-05,
|
| 3663 |
+
"loss": 0.0,
|
| 3664 |
+
"step": 521
|
| 3665 |
+
},
|
| 3666 |
+
{
|
| 3667 |
+
"epoch": 1.6338028169014085,
|
| 3668 |
+
"grad_norm": NaN,
|
| 3669 |
+
"learning_rate": 4.363132268600145e-05,
|
| 3670 |
+
"loss": 0.0,
|
| 3671 |
+
"step": 522
|
| 3672 |
+
},
|
| 3673 |
+
{
|
| 3674 |
+
"epoch": 1.6369327073552427,
|
| 3675 |
+
"grad_norm": NaN,
|
| 3676 |
+
"learning_rate": 4.34668382956707e-05,
|
| 3677 |
+
"loss": 0.0,
|
| 3678 |
+
"step": 523
|
| 3679 |
+
},
|
| 3680 |
+
{
|
| 3681 |
+
"epoch": 1.6400625978090768,
|
| 3682 |
+
"grad_norm": NaN,
|
| 3683 |
+
"learning_rate": 4.330242580433111e-05,
|
| 3684 |
+
"loss": 0.0,
|
| 3685 |
+
"step": 524
|
| 3686 |
+
},
|
| 3687 |
+
{
|
| 3688 |
+
"epoch": 1.6431924882629108,
|
| 3689 |
+
"grad_norm": NaN,
|
| 3690 |
+
"learning_rate": 4.313808702138109e-05,
|
| 3691 |
+
"loss": 0.0,
|
| 3692 |
+
"step": 525
|
| 3693 |
+
},
|
| 3694 |
+
{
|
| 3695 |
+
"epoch": 1.646322378716745,
|
| 3696 |
+
"grad_norm": NaN,
|
| 3697 |
+
"learning_rate": 4.297382375540784e-05,
|
| 3698 |
+
"loss": 0.0,
|
| 3699 |
+
"step": 526
|
| 3700 |
+
},
|
| 3701 |
+
{
|
| 3702 |
+
"epoch": 1.649452269170579,
|
| 3703 |
+
"grad_norm": NaN,
|
| 3704 |
+
"learning_rate": 4.2809637814167505e-05,
|
| 3705 |
+
"loss": 0.0,
|
| 3706 |
+
"step": 527
|
| 3707 |
+
},
|
| 3708 |
+
{
|
| 3709 |
+
"epoch": 1.652582159624413,
|
| 3710 |
+
"grad_norm": NaN,
|
| 3711 |
+
"learning_rate": 4.264553100456523e-05,
|
| 3712 |
+
"loss": 0.0,
|
| 3713 |
+
"step": 528
|
| 3714 |
+
},
|
| 3715 |
+
{
|
| 3716 |
+
"epoch": 1.6557120500782472,
|
| 3717 |
+
"grad_norm": NaN,
|
| 3718 |
+
"learning_rate": 4.248150513263528e-05,
|
| 3719 |
+
"loss": 0.0,
|
| 3720 |
+
"step": 529
|
| 3721 |
+
},
|
| 3722 |
+
{
|
| 3723 |
+
"epoch": 1.6588419405320813,
|
| 3724 |
+
"grad_norm": NaN,
|
| 3725 |
+
"learning_rate": 4.231756200352123e-05,
|
| 3726 |
+
"loss": 0.0,
|
| 3727 |
+
"step": 530
|
| 3728 |
+
},
|
| 3729 |
+
{
|
| 3730 |
+
"epoch": 1.6619718309859155,
|
| 3731 |
+
"grad_norm": NaN,
|
| 3732 |
+
"learning_rate": 4.215370342145601e-05,
|
| 3733 |
+
"loss": 0.0,
|
| 3734 |
+
"step": 531
|
| 3735 |
+
},
|
| 3736 |
+
{
|
| 3737 |
+
"epoch": 1.6651017214397497,
|
| 3738 |
+
"grad_norm": NaN,
|
| 3739 |
+
"learning_rate": 4.198993118974212e-05,
|
| 3740 |
+
"loss": 0.0,
|
| 3741 |
+
"step": 532
|
| 3742 |
+
},
|
| 3743 |
+
{
|
| 3744 |
+
"epoch": 1.6682316118935838,
|
| 3745 |
+
"grad_norm": NaN,
|
| 3746 |
+
"learning_rate": 4.1826247110731755e-05,
|
| 3747 |
+
"loss": 0.0,
|
| 3748 |
+
"step": 533
|
| 3749 |
+
},
|
| 3750 |
+
{
|
| 3751 |
+
"epoch": 1.671361502347418,
|
| 3752 |
+
"grad_norm": NaN,
|
| 3753 |
+
"learning_rate": 4.166265298580694e-05,
|
| 3754 |
+
"loss": 0.0,
|
| 3755 |
+
"step": 534
|
| 3756 |
+
},
|
| 3757 |
+
{
|
| 3758 |
+
"epoch": 1.674491392801252,
|
| 3759 |
+
"grad_norm": NaN,
|
| 3760 |
+
"learning_rate": 4.1499150615359726e-05,
|
| 3761 |
+
"loss": 0.0,
|
| 3762 |
+
"step": 535
|
| 3763 |
+
},
|
| 3764 |
+
{
|
| 3765 |
+
"epoch": 1.677621283255086,
|
| 3766 |
+
"grad_norm": NaN,
|
| 3767 |
+
"learning_rate": 4.133574179877243e-05,
|
| 3768 |
+
"loss": 0.0,
|
| 3769 |
+
"step": 536
|
| 3770 |
+
},
|
| 3771 |
+
{
|
| 3772 |
+
"epoch": 1.6807511737089202,
|
| 3773 |
+
"grad_norm": NaN,
|
| 3774 |
+
"learning_rate": 4.1172428334397754e-05,
|
| 3775 |
+
"loss": 0.0,
|
| 3776 |
+
"step": 537
|
| 3777 |
+
},
|
| 3778 |
+
{
|
| 3779 |
+
"epoch": 1.6838810641627542,
|
| 3780 |
+
"grad_norm": NaN,
|
| 3781 |
+
"learning_rate": 4.1009212019539044e-05,
|
| 3782 |
+
"loss": 0.0,
|
| 3783 |
+
"step": 538
|
| 3784 |
+
},
|
| 3785 |
+
{
|
| 3786 |
+
"epoch": 1.6870109546165883,
|
| 3787 |
+
"grad_norm": NaN,
|
| 3788 |
+
"learning_rate": 4.084609465043047e-05,
|
| 3789 |
+
"loss": 0.0,
|
| 3790 |
+
"step": 539
|
| 3791 |
+
},
|
| 3792 |
+
{
|
| 3793 |
+
"epoch": 1.6901408450704225,
|
| 3794 |
+
"grad_norm": NaN,
|
| 3795 |
+
"learning_rate": 4.068307802221734e-05,
|
| 3796 |
+
"loss": 0.0,
|
| 3797 |
+
"step": 540
|
| 3798 |
+
},
|
| 3799 |
+
{
|
| 3800 |
+
"epoch": 1.6932707355242567,
|
| 3801 |
+
"grad_norm": NaN,
|
| 3802 |
+
"learning_rate": 4.052016392893616e-05,
|
| 3803 |
+
"loss": 0.0,
|
| 3804 |
+
"step": 541
|
| 3805 |
+
},
|
| 3806 |
+
{
|
| 3807 |
+
"epoch": 1.6964006259780908,
|
| 3808 |
+
"grad_norm": NaN,
|
| 3809 |
+
"learning_rate": 4.035735416349515e-05,
|
| 3810 |
+
"loss": 0.0,
|
| 3811 |
+
"step": 542
|
| 3812 |
+
},
|
| 3813 |
+
{
|
| 3814 |
+
"epoch": 1.699530516431925,
|
| 3815 |
+
"grad_norm": NaN,
|
| 3816 |
+
"learning_rate": 4.0194650517654306e-05,
|
| 3817 |
+
"loss": 0.0,
|
| 3818 |
+
"step": 543
|
| 3819 |
+
},
|
| 3820 |
+
{
|
| 3821 |
+
"epoch": 1.7026604068857591,
|
| 3822 |
+
"grad_norm": NaN,
|
| 3823 |
+
"learning_rate": 4.003205478200576e-05,
|
| 3824 |
+
"loss": 0.0,
|
| 3825 |
+
"step": 544
|
| 3826 |
+
},
|
| 3827 |
+
{
|
| 3828 |
+
"epoch": 1.705790297339593,
|
| 3829 |
+
"grad_norm": NaN,
|
| 3830 |
+
"learning_rate": 3.986956874595409e-05,
|
| 3831 |
+
"loss": 0.0,
|
| 3832 |
+
"step": 545
|
| 3833 |
+
},
|
| 3834 |
+
{
|
| 3835 |
+
"epoch": 1.7089201877934272,
|
| 3836 |
+
"grad_norm": NaN,
|
| 3837 |
+
"learning_rate": 3.9707194197696584e-05,
|
| 3838 |
+
"loss": 0.0,
|
| 3839 |
+
"step": 546
|
| 3840 |
+
},
|
| 3841 |
+
{
|
| 3842 |
+
"epoch": 1.7120500782472612,
|
| 3843 |
+
"grad_norm": NaN,
|
| 3844 |
+
"learning_rate": 3.9544932924203594e-05,
|
| 3845 |
+
"loss": 0.0,
|
| 3846 |
+
"step": 547
|
| 3847 |
+
},
|
| 3848 |
+
{
|
| 3849 |
+
"epoch": 1.7151799687010953,
|
| 3850 |
+
"grad_norm": NaN,
|
| 3851 |
+
"learning_rate": 3.9382786711198814e-05,
|
| 3852 |
+
"loss": 0.0,
|
| 3853 |
+
"step": 548
|
| 3854 |
+
},
|
| 3855 |
+
{
|
| 3856 |
+
"epoch": 1.7183098591549295,
|
| 3857 |
+
"grad_norm": NaN,
|
| 3858 |
+
"learning_rate": 3.922075734313974e-05,
|
| 3859 |
+
"loss": 0.0,
|
| 3860 |
+
"step": 549
|
| 3861 |
+
},
|
| 3862 |
+
{
|
| 3863 |
+
"epoch": 1.7214397496087637,
|
| 3864 |
+
"grad_norm": NaN,
|
| 3865 |
+
"learning_rate": 3.905884660319792e-05,
|
| 3866 |
+
"loss": 0.0,
|
| 3867 |
+
"step": 550
|
| 3868 |
+
},
|
| 3869 |
+
{
|
| 3870 |
+
"epoch": 1.7245696400625978,
|
| 3871 |
+
"grad_norm": NaN,
|
| 3872 |
+
"learning_rate": 3.8897056273239354e-05,
|
| 3873 |
+
"loss": 0.0,
|
| 3874 |
+
"step": 551
|
| 3875 |
+
},
|
| 3876 |
+
{
|
| 3877 |
+
"epoch": 1.727699530516432,
|
| 3878 |
+
"grad_norm": NaN,
|
| 3879 |
+
"learning_rate": 3.8735388133804936e-05,
|
| 3880 |
+
"loss": 0.0,
|
| 3881 |
+
"step": 552
|
| 3882 |
+
},
|
| 3883 |
+
{
|
| 3884 |
+
"epoch": 1.7308294209702662,
|
| 3885 |
+
"grad_norm": NaN,
|
| 3886 |
+
"learning_rate": 3.857384396409083e-05,
|
| 3887 |
+
"loss": 0.0,
|
| 3888 |
+
"step": 553
|
| 3889 |
+
},
|
| 3890 |
+
{
|
| 3891 |
+
"epoch": 1.7339593114241003,
|
| 3892 |
+
"grad_norm": NaN,
|
| 3893 |
+
"learning_rate": 3.841242554192884e-05,
|
| 3894 |
+
"loss": 0.0,
|
| 3895 |
+
"step": 554
|
| 3896 |
+
},
|
| 3897 |
+
{
|
| 3898 |
+
"epoch": 1.7370892018779343,
|
| 3899 |
+
"grad_norm": NaN,
|
| 3900 |
+
"learning_rate": 3.825113464376693e-05,
|
| 3901 |
+
"loss": 0.0,
|
| 3902 |
+
"step": 555
|
| 3903 |
+
},
|
| 3904 |
+
{
|
| 3905 |
+
"epoch": 1.7402190923317684,
|
| 3906 |
+
"grad_norm": NaN,
|
| 3907 |
+
"learning_rate": 3.808997304464961e-05,
|
| 3908 |
+
"loss": 0.0,
|
| 3909 |
+
"step": 556
|
| 3910 |
+
},
|
| 3911 |
+
{
|
| 3912 |
+
"epoch": 1.7433489827856024,
|
| 3913 |
+
"grad_norm": NaN,
|
| 3914 |
+
"learning_rate": 3.7928942518198444e-05,
|
| 3915 |
+
"loss": 0.0,
|
| 3916 |
+
"step": 557
|
| 3917 |
+
},
|
| 3918 |
+
{
|
| 3919 |
+
"epoch": 1.7464788732394365,
|
| 3920 |
+
"grad_norm": NaN,
|
| 3921 |
+
"learning_rate": 3.7768044836592475e-05,
|
| 3922 |
+
"loss": 0.0,
|
| 3923 |
+
"step": 558
|
| 3924 |
+
},
|
| 3925 |
+
{
|
| 3926 |
+
"epoch": 1.7496087636932707,
|
| 3927 |
+
"grad_norm": NaN,
|
| 3928 |
+
"learning_rate": 3.7607281770548816e-05,
|
| 3929 |
+
"loss": 0.0,
|
| 3930 |
+
"step": 559
|
| 3931 |
+
},
|
| 3932 |
+
{
|
| 3933 |
+
"epoch": 1.7527386541471048,
|
| 3934 |
+
"grad_norm": NaN,
|
| 3935 |
+
"learning_rate": 3.744665508930303e-05,
|
| 3936 |
+
"loss": 0.0,
|
| 3937 |
+
"step": 560
|
| 3938 |
+
},
|
| 3939 |
+
{
|
| 3940 |
+
"epoch": 1.755868544600939,
|
| 3941 |
+
"grad_norm": NaN,
|
| 3942 |
+
"learning_rate": 3.728616656058977e-05,
|
| 3943 |
+
"loss": 0.0,
|
| 3944 |
+
"step": 561
|
| 3945 |
+
},
|
| 3946 |
+
{
|
| 3947 |
+
"epoch": 1.7589984350547732,
|
| 3948 |
+
"grad_norm": NaN,
|
| 3949 |
+
"learning_rate": 3.712581795062331e-05,
|
| 3950 |
+
"loss": 0.0,
|
| 3951 |
+
"step": 562
|
| 3952 |
+
},
|
| 3953 |
+
{
|
| 3954 |
+
"epoch": 1.7621283255086073,
|
| 3955 |
+
"grad_norm": NaN,
|
| 3956 |
+
"learning_rate": 3.696561102407805e-05,
|
| 3957 |
+
"loss": 0.0,
|
| 3958 |
+
"step": 563
|
| 3959 |
+
},
|
| 3960 |
+
{
|
| 3961 |
+
"epoch": 1.7652582159624415,
|
| 3962 |
+
"grad_norm": NaN,
|
| 3963 |
+
"learning_rate": 3.6805547544069144e-05,
|
| 3964 |
+
"loss": 0.0,
|
| 3965 |
+
"step": 564
|
| 3966 |
+
},
|
| 3967 |
+
{
|
| 3968 |
+
"epoch": 1.7683881064162754,
|
| 3969 |
+
"grad_norm": NaN,
|
| 3970 |
+
"learning_rate": 3.664562927213308e-05,
|
| 3971 |
+
"loss": 0.0,
|
| 3972 |
+
"step": 565
|
| 3973 |
+
},
|
| 3974 |
+
{
|
| 3975 |
+
"epoch": 1.7715179968701096,
|
| 3976 |
+
"grad_norm": NaN,
|
| 3977 |
+
"learning_rate": 3.648585796820833e-05,
|
| 3978 |
+
"loss": 0.0,
|
| 3979 |
+
"step": 566
|
| 3980 |
+
},
|
| 3981 |
+
{
|
| 3982 |
+
"epoch": 1.7746478873239435,
|
| 3983 |
+
"grad_norm": NaN,
|
| 3984 |
+
"learning_rate": 3.632623539061585e-05,
|
| 3985 |
+
"loss": 0.0,
|
| 3986 |
+
"step": 567
|
| 3987 |
+
},
|
| 3988 |
+
{
|
| 3989 |
+
"epoch": 1.7777777777777777,
|
| 3990 |
+
"grad_norm": NaN,
|
| 3991 |
+
"learning_rate": 3.616676329603995e-05,
|
| 3992 |
+
"loss": 0.0,
|
| 3993 |
+
"step": 568
|
| 3994 |
+
},
|
| 3995 |
+
{
|
| 3996 |
+
"epoch": 1.7809076682316118,
|
| 3997 |
+
"grad_norm": NaN,
|
| 3998 |
+
"learning_rate": 3.600744343950876e-05,
|
| 3999 |
+
"loss": 0.0,
|
| 4000 |
+
"step": 569
|
| 4001 |
+
},
|
| 4002 |
+
{
|
| 4003 |
+
"epoch": 1.784037558685446,
|
| 4004 |
+
"grad_norm": NaN,
|
| 4005 |
+
"learning_rate": 3.5848277574375024e-05,
|
| 4006 |
+
"loss": 0.0,
|
| 4007 |
+
"step": 570
|
| 4008 |
+
},
|
| 4009 |
+
{
|
| 4010 |
+
"epoch": 1.7871674491392802,
|
| 4011 |
+
"grad_norm": NaN,
|
| 4012 |
+
"learning_rate": 3.568926745229677e-05,
|
| 4013 |
+
"loss": 0.0,
|
| 4014 |
+
"step": 571
|
| 4015 |
+
},
|
| 4016 |
+
{
|
| 4017 |
+
"epoch": 1.7902973395931143,
|
| 4018 |
+
"grad_norm": NaN,
|
| 4019 |
+
"learning_rate": 3.553041482321805e-05,
|
| 4020 |
+
"loss": 0.0,
|
| 4021 |
+
"step": 572
|
| 4022 |
+
},
|
| 4023 |
+
{
|
| 4024 |
+
"epoch": 1.7934272300469485,
|
| 4025 |
+
"grad_norm": NaN,
|
| 4026 |
+
"learning_rate": 3.5371721435349615e-05,
|
| 4027 |
+
"loss": 0.0,
|
| 4028 |
+
"step": 573
|
| 4029 |
+
},
|
| 4030 |
+
{
|
| 4031 |
+
"epoch": 1.7965571205007824,
|
| 4032 |
+
"grad_norm": NaN,
|
| 4033 |
+
"learning_rate": 3.521318903514979e-05,
|
| 4034 |
+
"loss": 0.0,
|
| 4035 |
+
"step": 574
|
| 4036 |
+
},
|
| 4037 |
+
{
|
| 4038 |
+
"epoch": 1.7996870109546166,
|
| 4039 |
+
"grad_norm": NaN,
|
| 4040 |
+
"learning_rate": 3.505481936730516e-05,
|
| 4041 |
+
"loss": 0.0,
|
| 4042 |
+
"step": 575
|
| 4043 |
+
},
|
| 4044 |
+
{
|
| 4045 |
+
"epoch": 1.8028169014084507,
|
| 4046 |
+
"grad_norm": NaN,
|
| 4047 |
+
"learning_rate": 3.489661417471142e-05,
|
| 4048 |
+
"loss": 0.0,
|
| 4049 |
+
"step": 576
|
| 4050 |
+
},
|
| 4051 |
+
{
|
| 4052 |
+
"epoch": 1.8059467918622847,
|
| 4053 |
+
"grad_norm": NaN,
|
| 4054 |
+
"learning_rate": 3.473857519845415e-05,
|
| 4055 |
+
"loss": 0.0,
|
| 4056 |
+
"step": 577
|
| 4057 |
+
},
|
| 4058 |
+
{
|
| 4059 |
+
"epoch": 1.8090766823161188,
|
| 4060 |
+
"grad_norm": NaN,
|
| 4061 |
+
"learning_rate": 3.458070417778974e-05,
|
| 4062 |
+
"loss": 0.0,
|
| 4063 |
+
"step": 578
|
| 4064 |
+
},
|
| 4065 |
+
{
|
| 4066 |
+
"epoch": 1.812206572769953,
|
| 4067 |
+
"grad_norm": NaN,
|
| 4068 |
+
"learning_rate": 3.442300285012609e-05,
|
| 4069 |
+
"loss": 0.0,
|
| 4070 |
+
"step": 579
|
| 4071 |
+
},
|
| 4072 |
+
{
|
| 4073 |
+
"epoch": 1.8153364632237872,
|
| 4074 |
+
"grad_norm": NaN,
|
| 4075 |
+
"learning_rate": 3.4265472951003676e-05,
|
| 4076 |
+
"loss": 0.0,
|
| 4077 |
+
"step": 580
|
| 4078 |
+
},
|
| 4079 |
+
{
|
| 4080 |
+
"epoch": 1.8184663536776213,
|
| 4081 |
+
"grad_norm": NaN,
|
| 4082 |
+
"learning_rate": 3.410811621407633e-05,
|
| 4083 |
+
"loss": 0.0,
|
| 4084 |
+
"step": 581
|
| 4085 |
+
},
|
| 4086 |
+
{
|
| 4087 |
+
"epoch": 1.8215962441314555,
|
| 4088 |
+
"grad_norm": NaN,
|
| 4089 |
+
"learning_rate": 3.395093437109219e-05,
|
| 4090 |
+
"loss": 0.0,
|
| 4091 |
+
"step": 582
|
| 4092 |
+
},
|
| 4093 |
+
{
|
| 4094 |
+
"epoch": 1.8247261345852896,
|
| 4095 |
+
"grad_norm": NaN,
|
| 4096 |
+
"learning_rate": 3.3793929151874635e-05,
|
| 4097 |
+
"loss": 0.0,
|
| 4098 |
+
"step": 583
|
| 4099 |
+
},
|
| 4100 |
+
{
|
| 4101 |
+
"epoch": 1.8278560250391236,
|
| 4102 |
+
"grad_norm": NaN,
|
| 4103 |
+
"learning_rate": 3.363710228430329e-05,
|
| 4104 |
+
"loss": 0.0,
|
| 4105 |
+
"step": 584
|
| 4106 |
+
},
|
| 4107 |
+
{
|
| 4108 |
+
"epoch": 1.8309859154929577,
|
| 4109 |
+
"grad_norm": NaN,
|
| 4110 |
+
"learning_rate": 3.348045549429495e-05,
|
| 4111 |
+
"loss": 0.0,
|
| 4112 |
+
"step": 585
|
| 4113 |
+
},
|
| 4114 |
+
{
|
| 4115 |
+
"epoch": 1.834115805946792,
|
| 4116 |
+
"grad_norm": NaN,
|
| 4117 |
+
"learning_rate": 3.332399050578462e-05,
|
| 4118 |
+
"loss": 0.0,
|
| 4119 |
+
"step": 586
|
| 4120 |
+
},
|
| 4121 |
+
{
|
| 4122 |
+
"epoch": 1.8372456964006258,
|
| 4123 |
+
"grad_norm": NaN,
|
| 4124 |
+
"learning_rate": 3.3167709040706535e-05,
|
| 4125 |
+
"loss": 0.0,
|
| 4126 |
+
"step": 587
|
| 4127 |
+
},
|
| 4128 |
+
{
|
| 4129 |
+
"epoch": 1.84037558685446,
|
| 4130 |
+
"grad_norm": NaN,
|
| 4131 |
+
"learning_rate": 3.301161281897523e-05,
|
| 4132 |
+
"loss": 0.0,
|
| 4133 |
+
"step": 588
|
| 4134 |
+
},
|
| 4135 |
+
{
|
| 4136 |
+
"epoch": 1.8435054773082942,
|
| 4137 |
+
"grad_norm": NaN,
|
| 4138 |
+
"learning_rate": 3.285570355846657e-05,
|
| 4139 |
+
"loss": 0.0,
|
| 4140 |
+
"step": 589
|
| 4141 |
+
},
|
| 4142 |
+
{
|
| 4143 |
+
"epoch": 1.8466353677621283,
|
| 4144 |
+
"grad_norm": NaN,
|
| 4145 |
+
"learning_rate": 3.269998297499887e-05,
|
| 4146 |
+
"loss": 0.0,
|
| 4147 |
+
"step": 590
|
| 4148 |
+
},
|
| 4149 |
+
{
|
| 4150 |
+
"epoch": 1.8497652582159625,
|
| 4151 |
+
"grad_norm": NaN,
|
| 4152 |
+
"learning_rate": 3.254445278231405e-05,
|
| 4153 |
+
"loss": 0.0,
|
| 4154 |
+
"step": 591
|
| 4155 |
+
},
|
| 4156 |
+
{
|
| 4157 |
+
"epoch": 1.8528951486697967,
|
| 4158 |
+
"grad_norm": NaN,
|
| 4159 |
+
"learning_rate": 3.238911469205865e-05,
|
| 4160 |
+
"loss": 0.0,
|
| 4161 |
+
"step": 592
|
| 4162 |
+
},
|
| 4163 |
+
{
|
| 4164 |
+
"epoch": 1.8560250391236308,
|
| 4165 |
+
"grad_norm": NaN,
|
| 4166 |
+
"learning_rate": 3.223397041376515e-05,
|
| 4167 |
+
"loss": 0.0,
|
| 4168 |
+
"step": 593
|
| 4169 |
+
},
|
| 4170 |
+
{
|
| 4171 |
+
"epoch": 1.8591549295774648,
|
| 4172 |
+
"grad_norm": NaN,
|
| 4173 |
+
"learning_rate": 3.207902165483305e-05,
|
| 4174 |
+
"loss": 0.0,
|
| 4175 |
+
"step": 594
|
| 4176 |
+
},
|
| 4177 |
+
{
|
| 4178 |
+
"epoch": 1.862284820031299,
|
| 4179 |
+
"grad_norm": NaN,
|
| 4180 |
+
"learning_rate": 3.1924270120510135e-05,
|
| 4181 |
+
"loss": 0.0,
|
| 4182 |
+
"step": 595
|
| 4183 |
+
},
|
| 4184 |
+
{
|
| 4185 |
+
"epoch": 1.8654147104851329,
|
| 4186 |
+
"grad_norm": NaN,
|
| 4187 |
+
"learning_rate": 3.176971751387368e-05,
|
| 4188 |
+
"loss": 0.0,
|
| 4189 |
+
"step": 596
|
| 4190 |
+
},
|
| 4191 |
+
{
|
| 4192 |
+
"epoch": 1.868544600938967,
|
| 4193 |
+
"grad_norm": NaN,
|
| 4194 |
+
"learning_rate": 3.161536553581172e-05,
|
| 4195 |
+
"loss": 0.0,
|
| 4196 |
+
"step": 597
|
| 4197 |
+
},
|
| 4198 |
+
{
|
| 4199 |
+
"epoch": 1.8716744913928012,
|
| 4200 |
+
"grad_norm": NaN,
|
| 4201 |
+
"learning_rate": 3.1461215885004266e-05,
|
| 4202 |
+
"loss": 0.0,
|
| 4203 |
+
"step": 598
|
| 4204 |
+
},
|
| 4205 |
+
{
|
| 4206 |
+
"epoch": 1.8748043818466353,
|
| 4207 |
+
"grad_norm": NaN,
|
| 4208 |
+
"learning_rate": 3.1307270257904763e-05,
|
| 4209 |
+
"loss": 0.0,
|
| 4210 |
+
"step": 599
|
| 4211 |
+
},
|
| 4212 |
+
{
|
| 4213 |
+
"epoch": 1.8779342723004695,
|
| 4214 |
+
"grad_norm": NaN,
|
| 4215 |
+
"learning_rate": 3.1153530348721257e-05,
|
| 4216 |
+
"loss": 0.0,
|
| 4217 |
+
"step": 600
|
| 4218 |
+
},
|
| 4219 |
+
{
|
| 4220 |
+
"epoch": 1.8810641627543037,
|
| 4221 |
+
"grad_norm": NaN,
|
| 4222 |
+
"learning_rate": 3.099999784939784e-05,
|
| 4223 |
+
"loss": 0.0,
|
| 4224 |
+
"step": 601
|
| 4225 |
+
},
|
| 4226 |
+
{
|
| 4227 |
+
"epoch": 1.8841940532081378,
|
| 4228 |
+
"grad_norm": NaN,
|
| 4229 |
+
"learning_rate": 3.084667444959605e-05,
|
| 4230 |
+
"loss": 0.0,
|
| 4231 |
+
"step": 602
|
| 4232 |
+
},
|
| 4233 |
+
{
|
| 4234 |
+
"epoch": 1.887323943661972,
|
| 4235 |
+
"grad_norm": NaN,
|
| 4236 |
+
"learning_rate": 3.0693561836676174e-05,
|
| 4237 |
+
"loss": 0.0,
|
| 4238 |
+
"step": 603
|
| 4239 |
+
},
|
| 4240 |
+
{
|
| 4241 |
+
"epoch": 1.890453834115806,
|
| 4242 |
+
"grad_norm": NaN,
|
| 4243 |
+
"learning_rate": 3.0540661695678755e-05,
|
| 4244 |
+
"loss": 0.0,
|
| 4245 |
+
"step": 604
|
| 4246 |
+
},
|
| 4247 |
+
{
|
| 4248 |
+
"epoch": 1.89358372456964,
|
| 4249 |
+
"grad_norm": NaN,
|
| 4250 |
+
"learning_rate": 3.0387975709306043e-05,
|
| 4251 |
+
"loss": 0.0,
|
| 4252 |
+
"step": 605
|
| 4253 |
+
},
|
| 4254 |
+
{
|
| 4255 |
+
"epoch": 1.896713615023474,
|
| 4256 |
+
"grad_norm": NaN,
|
| 4257 |
+
"learning_rate": 3.0235505557903478e-05,
|
| 4258 |
+
"loss": 0.0,
|
| 4259 |
+
"step": 606
|
| 4260 |
+
},
|
| 4261 |
+
{
|
| 4262 |
+
"epoch": 1.8998435054773082,
|
| 4263 |
+
"grad_norm": NaN,
|
| 4264 |
+
"learning_rate": 3.0083252919441158e-05,
|
| 4265 |
+
"loss": 0.0,
|
| 4266 |
+
"step": 607
|
| 4267 |
+
},
|
| 4268 |
+
{
|
| 4269 |
+
"epoch": 1.9029733959311423,
|
| 4270 |
+
"grad_norm": NaN,
|
| 4271 |
+
"learning_rate": 2.993121946949542e-05,
|
| 4272 |
+
"loss": 0.0,
|
| 4273 |
+
"step": 608
|
| 4274 |
+
},
|
| 4275 |
+
{
|
| 4276 |
+
"epoch": 1.9061032863849765,
|
| 4277 |
+
"grad_norm": NaN,
|
| 4278 |
+
"learning_rate": 2.9779406881230383e-05,
|
| 4279 |
+
"loss": 0.0,
|
| 4280 |
+
"step": 609
|
| 4281 |
+
},
|
| 4282 |
+
{
|
| 4283 |
+
"epoch": 1.9092331768388107,
|
| 4284 |
+
"grad_norm": NaN,
|
| 4285 |
+
"learning_rate": 2.9627816825379497e-05,
|
| 4286 |
+
"loss": 0.0,
|
| 4287 |
+
"step": 610
|
| 4288 |
+
},
|
| 4289 |
+
{
|
| 4290 |
+
"epoch": 1.9123630672926448,
|
| 4291 |
+
"grad_norm": NaN,
|
| 4292 |
+
"learning_rate": 2.9476450970227233e-05,
|
| 4293 |
+
"loss": 0.0,
|
| 4294 |
+
"step": 611
|
| 4295 |
+
},
|
| 4296 |
+
{
|
| 4297 |
+
"epoch": 1.915492957746479,
|
| 4298 |
+
"grad_norm": NaN,
|
| 4299 |
+
"learning_rate": 2.9325310981590642e-05,
|
| 4300 |
+
"loss": 0.0,
|
| 4301 |
+
"step": 612
|
| 4302 |
+
},
|
| 4303 |
+
{
|
| 4304 |
+
"epoch": 1.9186228482003131,
|
| 4305 |
+
"grad_norm": NaN,
|
| 4306 |
+
"learning_rate": 2.917439852280108e-05,
|
| 4307 |
+
"loss": 0.0,
|
| 4308 |
+
"step": 613
|
| 4309 |
+
},
|
| 4310 |
+
{
|
| 4311 |
+
"epoch": 1.921752738654147,
|
| 4312 |
+
"grad_norm": NaN,
|
| 4313 |
+
"learning_rate": 2.9023715254685903e-05,
|
| 4314 |
+
"loss": 0.0,
|
| 4315 |
+
"step": 614
|
| 4316 |
+
},
|
| 4317 |
+
{
|
| 4318 |
+
"epoch": 1.9248826291079812,
|
| 4319 |
+
"grad_norm": NaN,
|
| 4320 |
+
"learning_rate": 2.8873262835550118e-05,
|
| 4321 |
+
"loss": 0.0,
|
| 4322 |
+
"step": 615
|
| 4323 |
+
},
|
| 4324 |
+
{
|
| 4325 |
+
"epoch": 1.9280125195618152,
|
| 4326 |
+
"grad_norm": NaN,
|
| 4327 |
+
"learning_rate": 2.872304292115828e-05,
|
| 4328 |
+
"loss": 0.0,
|
| 4329 |
+
"step": 616
|
| 4330 |
+
},
|
| 4331 |
+
{
|
| 4332 |
+
"epoch": 1.9311424100156493,
|
| 4333 |
+
"grad_norm": NaN,
|
| 4334 |
+
"learning_rate": 2.857305716471601e-05,
|
| 4335 |
+
"loss": 0.0,
|
| 4336 |
+
"step": 617
|
| 4337 |
+
},
|
| 4338 |
+
{
|
| 4339 |
+
"epoch": 1.9342723004694835,
|
| 4340 |
+
"grad_norm": NaN,
|
| 4341 |
+
"learning_rate": 2.8423307216852142e-05,
|
| 4342 |
+
"loss": 0.0,
|
| 4343 |
+
"step": 618
|
| 4344 |
+
},
|
| 4345 |
+
{
|
| 4346 |
+
"epoch": 1.9374021909233177,
|
| 4347 |
+
"grad_norm": NaN,
|
| 4348 |
+
"learning_rate": 2.8273794725600255e-05,
|
| 4349 |
+
"loss": 0.0,
|
| 4350 |
+
"step": 619
|
| 4351 |
+
},
|
| 4352 |
+
{
|
| 4353 |
+
"epoch": 1.9405320813771518,
|
| 4354 |
+
"grad_norm": NaN,
|
| 4355 |
+
"learning_rate": 2.812452133638075e-05,
|
| 4356 |
+
"loss": 0.0,
|
| 4357 |
+
"step": 620
|
| 4358 |
+
},
|
| 4359 |
+
{
|
| 4360 |
+
"epoch": 1.943661971830986,
|
| 4361 |
+
"grad_norm": NaN,
|
| 4362 |
+
"learning_rate": 2.7975488691982578e-05,
|
| 4363 |
+
"loss": 0.0,
|
| 4364 |
+
"step": 621
|
| 4365 |
+
},
|
| 4366 |
+
{
|
| 4367 |
+
"epoch": 1.9467918622848202,
|
| 4368 |
+
"grad_norm": NaN,
|
| 4369 |
+
"learning_rate": 2.7826698432545317e-05,
|
| 4370 |
+
"loss": 0.0,
|
| 4371 |
+
"step": 622
|
| 4372 |
+
},
|
| 4373 |
+
{
|
| 4374 |
+
"epoch": 1.9499217527386543,
|
| 4375 |
+
"grad_norm": NaN,
|
| 4376 |
+
"learning_rate": 2.767815219554094e-05,
|
| 4377 |
+
"loss": 0.0,
|
| 4378 |
+
"step": 623
|
| 4379 |
+
},
|
| 4380 |
+
{
|
| 4381 |
+
"epoch": 1.9530516431924883,
|
| 4382 |
+
"grad_norm": NaN,
|
| 4383 |
+
"learning_rate": 2.7529851615755993e-05,
|
| 4384 |
+
"loss": 0.0,
|
| 4385 |
+
"step": 624
|
| 4386 |
+
},
|
| 4387 |
+
{
|
| 4388 |
+
"epoch": 1.9561815336463224,
|
| 4389 |
+
"grad_norm": NaN,
|
| 4390 |
+
"learning_rate": 2.738179832527343e-05,
|
| 4391 |
+
"loss": 0.0,
|
| 4392 |
+
"step": 625
|
| 4393 |
+
},
|
| 4394 |
+
{
|
| 4395 |
+
"epoch": 1.9593114241001564,
|
| 4396 |
+
"grad_norm": NaN,
|
| 4397 |
+
"learning_rate": 2.7233993953454795e-05,
|
| 4398 |
+
"loss": 0.0,
|
| 4399 |
+
"step": 626
|
| 4400 |
+
},
|
| 4401 |
+
{
|
| 4402 |
+
"epoch": 1.9624413145539905,
|
| 4403 |
+
"grad_norm": NaN,
|
| 4404 |
+
"learning_rate": 2.7086440126922163e-05,
|
| 4405 |
+
"loss": 0.0,
|
| 4406 |
+
"step": 627
|
| 4407 |
+
},
|
| 4408 |
+
{
|
| 4409 |
+
"epoch": 1.9655712050078247,
|
| 4410 |
+
"grad_norm": NaN,
|
| 4411 |
+
"learning_rate": 2.693913846954036e-05,
|
| 4412 |
+
"loss": 0.0,
|
| 4413 |
+
"step": 628
|
| 4414 |
+
},
|
| 4415 |
+
{
|
| 4416 |
+
"epoch": 1.9687010954616588,
|
| 4417 |
+
"grad_norm": NaN,
|
| 4418 |
+
"learning_rate": 2.6792090602398966e-05,
|
| 4419 |
+
"loss": 0.0,
|
| 4420 |
+
"step": 629
|
| 4421 |
+
},
|
| 4422 |
+
{
|
| 4423 |
+
"epoch": 1.971830985915493,
|
| 4424 |
+
"grad_norm": NaN,
|
| 4425 |
+
"learning_rate": 2.664529814379457e-05,
|
| 4426 |
+
"loss": 0.0,
|
| 4427 |
+
"step": 630
|
| 4428 |
+
},
|
| 4429 |
+
{
|
| 4430 |
+
"epoch": 1.9749608763693272,
|
| 4431 |
+
"grad_norm": NaN,
|
| 4432 |
+
"learning_rate": 2.6498762709212956e-05,
|
| 4433 |
+
"loss": 0.0,
|
| 4434 |
+
"step": 631
|
| 4435 |
+
},
|
| 4436 |
+
{
|
| 4437 |
+
"epoch": 1.9780907668231613,
|
| 4438 |
+
"grad_norm": NaN,
|
| 4439 |
+
"learning_rate": 2.6352485911311253e-05,
|
| 4440 |
+
"loss": 0.0,
|
| 4441 |
+
"step": 632
|
| 4442 |
+
},
|
| 4443 |
+
{
|
| 4444 |
+
"epoch": 1.9812206572769953,
|
| 4445 |
+
"grad_norm": NaN,
|
| 4446 |
+
"learning_rate": 2.6206469359900236e-05,
|
| 4447 |
+
"loss": 0.0,
|
| 4448 |
+
"step": 633
|
| 4449 |
+
},
|
| 4450 |
+
{
|
| 4451 |
+
"epoch": 1.9843505477308294,
|
| 4452 |
+
"grad_norm": NaN,
|
| 4453 |
+
"learning_rate": 2.6060714661926654e-05,
|
| 4454 |
+
"loss": 0.0,
|
| 4455 |
+
"step": 634
|
| 4456 |
+
},
|
| 4457 |
+
{
|
| 4458 |
+
"epoch": 1.9874804381846636,
|
| 4459 |
+
"grad_norm": NaN,
|
| 4460 |
+
"learning_rate": 2.5915223421455448e-05,
|
| 4461 |
+
"loss": 0.0,
|
| 4462 |
+
"step": 635
|
| 4463 |
+
},
|
| 4464 |
+
{
|
| 4465 |
+
"epoch": 1.9906103286384975,
|
| 4466 |
+
"grad_norm": NaN,
|
| 4467 |
+
"learning_rate": 2.5769997239652137e-05,
|
| 4468 |
+
"loss": 0.0,
|
| 4469 |
+
"step": 636
|
| 4470 |
+
},
|
| 4471 |
+
{
|
| 4472 |
+
"epoch": 1.9937402190923317,
|
| 4473 |
+
"grad_norm": NaN,
|
| 4474 |
+
"learning_rate": 2.5625037714765277e-05,
|
| 4475 |
+
"loss": 0.0,
|
| 4476 |
+
"step": 637
|
| 4477 |
+
},
|
| 4478 |
+
{
|
| 4479 |
+
"epoch": 1.9968701095461658,
|
| 4480 |
+
"grad_norm": NaN,
|
| 4481 |
+
"learning_rate": 2.5480346442108714e-05,
|
| 4482 |
+
"loss": 0.0,
|
| 4483 |
+
"step": 638
|
| 4484 |
+
},
|
| 4485 |
+
{
|
| 4486 |
+
"epoch": 2.0,
|
| 4487 |
+
"grad_norm": NaN,
|
| 4488 |
+
"learning_rate": 2.5335925014044193e-05,
|
| 4489 |
+
"loss": 0.0,
|
| 4490 |
+
"step": 639
|
| 4491 |
+
},
|
| 4492 |
+
{
|
| 4493 |
+
"epoch": 2.0,
|
| 4494 |
+
"eval_loss": NaN,
|
| 4495 |
+
"eval_runtime": 15.1515,
|
| 4496 |
+
"eval_samples_per_second": 35.574,
|
| 4497 |
+
"eval_steps_per_second": 8.91,
|
| 4498 |
+
"step": 639
|
| 4499 |
}
|
| 4500 |
],
|
| 4501 |
"logging_steps": 1,
|
|
|
|
| 4515 |
"attributes": {}
|
| 4516 |
}
|
| 4517 |
},
|
| 4518 |
+
"total_flos": 1.6924437727779226e+17,
|
| 4519 |
"train_batch_size": 4,
|
| 4520 |
"trial_name": null,
|
| 4521 |
"trial_params": null
|