Training in progress, step 750, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 83945296
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:99a77a83a96075b809ebf9cc00a63acbc7957a42a478e066b93e59296ebacdda
|
| 3 |
size 83945296
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 43123028
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e05babd698fc1b664a2688fdd396bcd9fa52762a02688a6f2a49644fe2e508bc
|
| 3 |
size 43123028
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:66a278e168770b17c6200dfa59a55a1ce746c931ea0cd8f411aebbb607213cf7
|
| 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:ac9c79078a88e8cb4c64a0c9b0c0b44a2e00e7411388bd2b931c0d5e22790bf5
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 375,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -2640,6 +2640,2639 @@
|
|
| 2640 |
"eval_samples_per_second": 16.605,
|
| 2641 |
"eval_steps_per_second": 8.302,
|
| 2642 |
"step": 375
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2643 |
}
|
| 2644 |
],
|
| 2645 |
"logging_steps": 1,
|
|
@@ -2659,7 +5292,7 @@
|
|
| 2659 |
"attributes": {}
|
| 2660 |
}
|
| 2661 |
},
|
| 2662 |
-
"total_flos":
|
| 2663 |
"train_batch_size": 2,
|
| 2664 |
"trial_name": null,
|
| 2665 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.04901300069843526,
|
| 5 |
"eval_steps": 375,
|
| 6 |
+
"global_step": 750,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 2640 |
"eval_samples_per_second": 16.605,
|
| 2641 |
"eval_steps_per_second": 8.302,
|
| 2642 |
"step": 375
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 0.024571851016815544,
|
| 2646 |
+
"grad_norm": 2.339012384414673,
|
| 2647 |
+
"learning_rate": 0.00017167309246649297,
|
| 2648 |
+
"loss": 3.1042,
|
| 2649 |
+
"step": 376
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 0.024637201684413457,
|
| 2653 |
+
"grad_norm": 2.4127211570739746,
|
| 2654 |
+
"learning_rate": 0.0001715259004365791,
|
| 2655 |
+
"loss": 3.1541,
|
| 2656 |
+
"step": 377
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"epoch": 0.02470255235201137,
|
| 2660 |
+
"grad_norm": 2.5378224849700928,
|
| 2661 |
+
"learning_rate": 0.00017137839043356484,
|
| 2662 |
+
"loss": 3.1696,
|
| 2663 |
+
"step": 378
|
| 2664 |
+
},
|
| 2665 |
+
{
|
| 2666 |
+
"epoch": 0.024767903019609284,
|
| 2667 |
+
"grad_norm": 2.3195438385009766,
|
| 2668 |
+
"learning_rate": 0.00017123056311321562,
|
| 2669 |
+
"loss": 3.0877,
|
| 2670 |
+
"step": 379
|
| 2671 |
+
},
|
| 2672 |
+
{
|
| 2673 |
+
"epoch": 0.024833253687207197,
|
| 2674 |
+
"grad_norm": 2.556420087814331,
|
| 2675 |
+
"learning_rate": 0.0001710824191327075,
|
| 2676 |
+
"loss": 3.0469,
|
| 2677 |
+
"step": 380
|
| 2678 |
+
},
|
| 2679 |
+
{
|
| 2680 |
+
"epoch": 0.024898604354805114,
|
| 2681 |
+
"grad_norm": 2.2406117916107178,
|
| 2682 |
+
"learning_rate": 0.00017093395915062428,
|
| 2683 |
+
"loss": 3.0364,
|
| 2684 |
+
"step": 381
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 0.024963955022403027,
|
| 2688 |
+
"grad_norm": 2.573399782180786,
|
| 2689 |
+
"learning_rate": 0.00017078518382695465,
|
| 2690 |
+
"loss": 3.0017,
|
| 2691 |
+
"step": 382
|
| 2692 |
+
},
|
| 2693 |
+
{
|
| 2694 |
+
"epoch": 0.02502930569000094,
|
| 2695 |
+
"grad_norm": 2.368605375289917,
|
| 2696 |
+
"learning_rate": 0.00017063609382308908,
|
| 2697 |
+
"loss": 3.1007,
|
| 2698 |
+
"step": 383
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"epoch": 0.025094656357598853,
|
| 2702 |
+
"grad_norm": 2.4229259490966797,
|
| 2703 |
+
"learning_rate": 0.00017048668980181698,
|
| 2704 |
+
"loss": 2.9687,
|
| 2705 |
+
"step": 384
|
| 2706 |
+
},
|
| 2707 |
+
{
|
| 2708 |
+
"epoch": 0.025160007025196766,
|
| 2709 |
+
"grad_norm": 2.3166263103485107,
|
| 2710 |
+
"learning_rate": 0.00017033697242732377,
|
| 2711 |
+
"loss": 2.8533,
|
| 2712 |
+
"step": 385
|
| 2713 |
+
},
|
| 2714 |
+
{
|
| 2715 |
+
"epoch": 0.02522535769279468,
|
| 2716 |
+
"grad_norm": 3.2635836601257324,
|
| 2717 |
+
"learning_rate": 0.0001701869423651879,
|
| 2718 |
+
"loss": 3.0645,
|
| 2719 |
+
"step": 386
|
| 2720 |
+
},
|
| 2721 |
+
{
|
| 2722 |
+
"epoch": 0.025290708360392593,
|
| 2723 |
+
"grad_norm": 2.6406781673431396,
|
| 2724 |
+
"learning_rate": 0.00017003660028237793,
|
| 2725 |
+
"loss": 3.0866,
|
| 2726 |
+
"step": 387
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 0.02535605902799051,
|
| 2730 |
+
"grad_norm": 3.3586738109588623,
|
| 2731 |
+
"learning_rate": 0.00016988594684724947,
|
| 2732 |
+
"loss": 2.9727,
|
| 2733 |
+
"step": 388
|
| 2734 |
+
},
|
| 2735 |
+
{
|
| 2736 |
+
"epoch": 0.025421409695588423,
|
| 2737 |
+
"grad_norm": 3.0996286869049072,
|
| 2738 |
+
"learning_rate": 0.00016973498272954222,
|
| 2739 |
+
"loss": 3.0378,
|
| 2740 |
+
"step": 389
|
| 2741 |
+
},
|
| 2742 |
+
{
|
| 2743 |
+
"epoch": 0.025486760363186336,
|
| 2744 |
+
"grad_norm": 2.8624045848846436,
|
| 2745 |
+
"learning_rate": 0.00016958370860037717,
|
| 2746 |
+
"loss": 3.3147,
|
| 2747 |
+
"step": 390
|
| 2748 |
+
},
|
| 2749 |
+
{
|
| 2750 |
+
"epoch": 0.02555211103078425,
|
| 2751 |
+
"grad_norm": 3.3544201850891113,
|
| 2752 |
+
"learning_rate": 0.00016943212513225345,
|
| 2753 |
+
"loss": 3.1472,
|
| 2754 |
+
"step": 391
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 0.025617461698382162,
|
| 2758 |
+
"grad_norm": 4.079308986663818,
|
| 2759 |
+
"learning_rate": 0.00016928023299904533,
|
| 2760 |
+
"loss": 2.9491,
|
| 2761 |
+
"step": 392
|
| 2762 |
+
},
|
| 2763 |
+
{
|
| 2764 |
+
"epoch": 0.025682812365980075,
|
| 2765 |
+
"grad_norm": 3.312035083770752,
|
| 2766 |
+
"learning_rate": 0.0001691280328759992,
|
| 2767 |
+
"loss": 2.811,
|
| 2768 |
+
"step": 393
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 0.02574816303357799,
|
| 2772 |
+
"grad_norm": 3.138267755508423,
|
| 2773 |
+
"learning_rate": 0.00016897552543973084,
|
| 2774 |
+
"loss": 2.6726,
|
| 2775 |
+
"step": 394
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 0.025813513701175905,
|
| 2779 |
+
"grad_norm": 3.613313913345337,
|
| 2780 |
+
"learning_rate": 0.00016882271136822206,
|
| 2781 |
+
"loss": 2.8072,
|
| 2782 |
+
"step": 395
|
| 2783 |
+
},
|
| 2784 |
+
{
|
| 2785 |
+
"epoch": 0.02587886436877382,
|
| 2786 |
+
"grad_norm": 3.8998324871063232,
|
| 2787 |
+
"learning_rate": 0.0001686695913408179,
|
| 2788 |
+
"loss": 3.1871,
|
| 2789 |
+
"step": 396
|
| 2790 |
+
},
|
| 2791 |
+
{
|
| 2792 |
+
"epoch": 0.02594421503637173,
|
| 2793 |
+
"grad_norm": 6.394239902496338,
|
| 2794 |
+
"learning_rate": 0.0001685161660382235,
|
| 2795 |
+
"loss": 3.0556,
|
| 2796 |
+
"step": 397
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"epoch": 0.026009565703969645,
|
| 2800 |
+
"grad_norm": 5.196231842041016,
|
| 2801 |
+
"learning_rate": 0.00016836243614250113,
|
| 2802 |
+
"loss": 3.4508,
|
| 2803 |
+
"step": 398
|
| 2804 |
+
},
|
| 2805 |
+
{
|
| 2806 |
+
"epoch": 0.026074916371567558,
|
| 2807 |
+
"grad_norm": 5.16098165512085,
|
| 2808 |
+
"learning_rate": 0.00016820840233706719,
|
| 2809 |
+
"loss": 2.7991,
|
| 2810 |
+
"step": 399
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 0.02614026703916547,
|
| 2814 |
+
"grad_norm": 9.842926979064941,
|
| 2815 |
+
"learning_rate": 0.0001680540653066891,
|
| 2816 |
+
"loss": 2.9014,
|
| 2817 |
+
"step": 400
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 0.026205617706763384,
|
| 2821 |
+
"grad_norm": 18.941564559936523,
|
| 2822 |
+
"learning_rate": 0.00016789942573748232,
|
| 2823 |
+
"loss": 3.7021,
|
| 2824 |
+
"step": 401
|
| 2825 |
+
},
|
| 2826 |
+
{
|
| 2827 |
+
"epoch": 0.026270968374361298,
|
| 2828 |
+
"grad_norm": 5.617252826690674,
|
| 2829 |
+
"learning_rate": 0.0001677444843169072,
|
| 2830 |
+
"loss": 3.4405,
|
| 2831 |
+
"step": 402
|
| 2832 |
+
},
|
| 2833 |
+
{
|
| 2834 |
+
"epoch": 0.026336319041959214,
|
| 2835 |
+
"grad_norm": 6.610146522521973,
|
| 2836 |
+
"learning_rate": 0.00016758924173376603,
|
| 2837 |
+
"loss": 3.2698,
|
| 2838 |
+
"step": 403
|
| 2839 |
+
},
|
| 2840 |
+
{
|
| 2841 |
+
"epoch": 0.026401669709557127,
|
| 2842 |
+
"grad_norm": 3.9733502864837646,
|
| 2843 |
+
"learning_rate": 0.0001674336986781999,
|
| 2844 |
+
"loss": 3.2307,
|
| 2845 |
+
"step": 404
|
| 2846 |
+
},
|
| 2847 |
+
{
|
| 2848 |
+
"epoch": 0.02646702037715504,
|
| 2849 |
+
"grad_norm": 1.9799753427505493,
|
| 2850 |
+
"learning_rate": 0.00016727785584168581,
|
| 2851 |
+
"loss": 3.1771,
|
| 2852 |
+
"step": 405
|
| 2853 |
+
},
|
| 2854 |
+
{
|
| 2855 |
+
"epoch": 0.026532371044752954,
|
| 2856 |
+
"grad_norm": 2.117891311645508,
|
| 2857 |
+
"learning_rate": 0.0001671217139170333,
|
| 2858 |
+
"loss": 3.2574,
|
| 2859 |
+
"step": 406
|
| 2860 |
+
},
|
| 2861 |
+
{
|
| 2862 |
+
"epoch": 0.026597721712350867,
|
| 2863 |
+
"grad_norm": 2.028475522994995,
|
| 2864 |
+
"learning_rate": 0.00016696527359838154,
|
| 2865 |
+
"loss": 3.3172,
|
| 2866 |
+
"step": 407
|
| 2867 |
+
},
|
| 2868 |
+
{
|
| 2869 |
+
"epoch": 0.02666307237994878,
|
| 2870 |
+
"grad_norm": 1.6997264623641968,
|
| 2871 |
+
"learning_rate": 0.00016680853558119632,
|
| 2872 |
+
"loss": 3.119,
|
| 2873 |
+
"step": 408
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 0.026728423047546693,
|
| 2877 |
+
"grad_norm": 1.6663233041763306,
|
| 2878 |
+
"learning_rate": 0.0001666515005622668,
|
| 2879 |
+
"loss": 3.0839,
|
| 2880 |
+
"step": 409
|
| 2881 |
+
},
|
| 2882 |
+
{
|
| 2883 |
+
"epoch": 0.02679377371514461,
|
| 2884 |
+
"grad_norm": 1.6285985708236694,
|
| 2885 |
+
"learning_rate": 0.0001664941692397025,
|
| 2886 |
+
"loss": 3.0017,
|
| 2887 |
+
"step": 410
|
| 2888 |
+
},
|
| 2889 |
+
{
|
| 2890 |
+
"epoch": 0.026859124382742523,
|
| 2891 |
+
"grad_norm": 1.765998363494873,
|
| 2892 |
+
"learning_rate": 0.00016633654231293013,
|
| 2893 |
+
"loss": 2.9814,
|
| 2894 |
+
"step": 411
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 0.026924475050340436,
|
| 2898 |
+
"grad_norm": 1.6121286153793335,
|
| 2899 |
+
"learning_rate": 0.00016617862048269065,
|
| 2900 |
+
"loss": 2.8452,
|
| 2901 |
+
"step": 412
|
| 2902 |
+
},
|
| 2903 |
+
{
|
| 2904 |
+
"epoch": 0.02698982571793835,
|
| 2905 |
+
"grad_norm": 1.76937997341156,
|
| 2906 |
+
"learning_rate": 0.00016602040445103588,
|
| 2907 |
+
"loss": 3.1867,
|
| 2908 |
+
"step": 413
|
| 2909 |
+
},
|
| 2910 |
+
{
|
| 2911 |
+
"epoch": 0.027055176385536263,
|
| 2912 |
+
"grad_norm": 1.8129544258117676,
|
| 2913 |
+
"learning_rate": 0.00016586189492132566,
|
| 2914 |
+
"loss": 2.9669,
|
| 2915 |
+
"step": 414
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 0.027120527053134176,
|
| 2919 |
+
"grad_norm": 1.7040003538131714,
|
| 2920 |
+
"learning_rate": 0.00016570309259822453,
|
| 2921 |
+
"loss": 2.9403,
|
| 2922 |
+
"step": 415
|
| 2923 |
+
},
|
| 2924 |
+
{
|
| 2925 |
+
"epoch": 0.02718587772073209,
|
| 2926 |
+
"grad_norm": 1.8865150213241577,
|
| 2927 |
+
"learning_rate": 0.0001655439981876987,
|
| 2928 |
+
"loss": 3.1433,
|
| 2929 |
+
"step": 416
|
| 2930 |
+
},
|
| 2931 |
+
{
|
| 2932 |
+
"epoch": 0.027251228388330006,
|
| 2933 |
+
"grad_norm": 1.884498119354248,
|
| 2934 |
+
"learning_rate": 0.00016538461239701277,
|
| 2935 |
+
"loss": 3.0231,
|
| 2936 |
+
"step": 417
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 0.02731657905592792,
|
| 2940 |
+
"grad_norm": 1.9052989482879639,
|
| 2941 |
+
"learning_rate": 0.00016522493593472683,
|
| 2942 |
+
"loss": 2.8977,
|
| 2943 |
+
"step": 418
|
| 2944 |
+
},
|
| 2945 |
+
{
|
| 2946 |
+
"epoch": 0.027381929723525832,
|
| 2947 |
+
"grad_norm": 2.26375150680542,
|
| 2948 |
+
"learning_rate": 0.0001650649695106931,
|
| 2949 |
+
"loss": 3.1765,
|
| 2950 |
+
"step": 419
|
| 2951 |
+
},
|
| 2952 |
+
{
|
| 2953 |
+
"epoch": 0.027447280391123745,
|
| 2954 |
+
"grad_norm": 2.234987258911133,
|
| 2955 |
+
"learning_rate": 0.00016490471383605288,
|
| 2956 |
+
"loss": 3.2449,
|
| 2957 |
+
"step": 420
|
| 2958 |
+
},
|
| 2959 |
+
{
|
| 2960 |
+
"epoch": 0.02751263105872166,
|
| 2961 |
+
"grad_norm": 2.077850818634033,
|
| 2962 |
+
"learning_rate": 0.00016474416962323325,
|
| 2963 |
+
"loss": 2.838,
|
| 2964 |
+
"step": 421
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 0.027577981726319572,
|
| 2968 |
+
"grad_norm": 2.2941253185272217,
|
| 2969 |
+
"learning_rate": 0.00016458333758594414,
|
| 2970 |
+
"loss": 2.985,
|
| 2971 |
+
"step": 422
|
| 2972 |
+
},
|
| 2973 |
+
{
|
| 2974 |
+
"epoch": 0.027643332393917485,
|
| 2975 |
+
"grad_norm": 2.149695873260498,
|
| 2976 |
+
"learning_rate": 0.00016442221843917496,
|
| 2977 |
+
"loss": 3.2526,
|
| 2978 |
+
"step": 423
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 0.0277086830615154,
|
| 2982 |
+
"grad_norm": 2.061628580093384,
|
| 2983 |
+
"learning_rate": 0.00016426081289919143,
|
| 2984 |
+
"loss": 3.1972,
|
| 2985 |
+
"step": 424
|
| 2986 |
+
},
|
| 2987 |
+
{
|
| 2988 |
+
"epoch": 0.027774033729113315,
|
| 2989 |
+
"grad_norm": 2.2111194133758545,
|
| 2990 |
+
"learning_rate": 0.0001640991216835326,
|
| 2991 |
+
"loss": 3.1089,
|
| 2992 |
+
"step": 425
|
| 2993 |
+
},
|
| 2994 |
+
{
|
| 2995 |
+
"epoch": 0.027839384396711228,
|
| 2996 |
+
"grad_norm": 2.13586163520813,
|
| 2997 |
+
"learning_rate": 0.00016393714551100734,
|
| 2998 |
+
"loss": 3.099,
|
| 2999 |
+
"step": 426
|
| 3000 |
+
},
|
| 3001 |
+
{
|
| 3002 |
+
"epoch": 0.02790473506430914,
|
| 3003 |
+
"grad_norm": 2.1120922565460205,
|
| 3004 |
+
"learning_rate": 0.0001637748851016914,
|
| 3005 |
+
"loss": 3.1227,
|
| 3006 |
+
"step": 427
|
| 3007 |
+
},
|
| 3008 |
+
{
|
| 3009 |
+
"epoch": 0.027970085731907054,
|
| 3010 |
+
"grad_norm": 2.342458486557007,
|
| 3011 |
+
"learning_rate": 0.00016361234117692413,
|
| 3012 |
+
"loss": 3.3583,
|
| 3013 |
+
"step": 428
|
| 3014 |
+
},
|
| 3015 |
+
{
|
| 3016 |
+
"epoch": 0.028035436399504968,
|
| 3017 |
+
"grad_norm": 2.4300196170806885,
|
| 3018 |
+
"learning_rate": 0.00016344951445930526,
|
| 3019 |
+
"loss": 3.2421,
|
| 3020 |
+
"step": 429
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 0.02810078706710288,
|
| 3024 |
+
"grad_norm": 2.273008108139038,
|
| 3025 |
+
"learning_rate": 0.0001632864056726917,
|
| 3026 |
+
"loss": 3.0715,
|
| 3027 |
+
"step": 430
|
| 3028 |
+
},
|
| 3029 |
+
{
|
| 3030 |
+
"epoch": 0.028166137734700798,
|
| 3031 |
+
"grad_norm": 2.4410061836242676,
|
| 3032 |
+
"learning_rate": 0.00016312301554219426,
|
| 3033 |
+
"loss": 3.1462,
|
| 3034 |
+
"step": 431
|
| 3035 |
+
},
|
| 3036 |
+
{
|
| 3037 |
+
"epoch": 0.02823148840229871,
|
| 3038 |
+
"grad_norm": 2.5503828525543213,
|
| 3039 |
+
"learning_rate": 0.00016295934479417453,
|
| 3040 |
+
"loss": 3.0751,
|
| 3041 |
+
"step": 432
|
| 3042 |
+
},
|
| 3043 |
+
{
|
| 3044 |
+
"epoch": 0.028296839069896624,
|
| 3045 |
+
"grad_norm": 2.3271334171295166,
|
| 3046 |
+
"learning_rate": 0.00016279539415624164,
|
| 3047 |
+
"loss": 3.0947,
|
| 3048 |
+
"step": 433
|
| 3049 |
+
},
|
| 3050 |
+
{
|
| 3051 |
+
"epoch": 0.028362189737494537,
|
| 3052 |
+
"grad_norm": 2.718640089035034,
|
| 3053 |
+
"learning_rate": 0.0001626311643572489,
|
| 3054 |
+
"loss": 2.9781,
|
| 3055 |
+
"step": 434
|
| 3056 |
+
},
|
| 3057 |
+
{
|
| 3058 |
+
"epoch": 0.02842754040509245,
|
| 3059 |
+
"grad_norm": 3.1213088035583496,
|
| 3060 |
+
"learning_rate": 0.00016246665612729074,
|
| 3061 |
+
"loss": 3.1426,
|
| 3062 |
+
"step": 435
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 0.028492891072690364,
|
| 3066 |
+
"grad_norm": 2.5128207206726074,
|
| 3067 |
+
"learning_rate": 0.00016230187019769928,
|
| 3068 |
+
"loss": 2.8893,
|
| 3069 |
+
"step": 436
|
| 3070 |
+
},
|
| 3071 |
+
{
|
| 3072 |
+
"epoch": 0.028558241740288277,
|
| 3073 |
+
"grad_norm": 2.748035430908203,
|
| 3074 |
+
"learning_rate": 0.00016213680730104124,
|
| 3075 |
+
"loss": 2.9823,
|
| 3076 |
+
"step": 437
|
| 3077 |
+
},
|
| 3078 |
+
{
|
| 3079 |
+
"epoch": 0.028623592407886193,
|
| 3080 |
+
"grad_norm": 3.006138801574707,
|
| 3081 |
+
"learning_rate": 0.0001619714681711146,
|
| 3082 |
+
"loss": 2.9931,
|
| 3083 |
+
"step": 438
|
| 3084 |
+
},
|
| 3085 |
+
{
|
| 3086 |
+
"epoch": 0.028688943075484107,
|
| 3087 |
+
"grad_norm": 2.952946424484253,
|
| 3088 |
+
"learning_rate": 0.00016180585354294536,
|
| 3089 |
+
"loss": 2.8257,
|
| 3090 |
+
"step": 439
|
| 3091 |
+
},
|
| 3092 |
+
{
|
| 3093 |
+
"epoch": 0.02875429374308202,
|
| 3094 |
+
"grad_norm": 3.55165958404541,
|
| 3095 |
+
"learning_rate": 0.00016163996415278424,
|
| 3096 |
+
"loss": 3.0414,
|
| 3097 |
+
"step": 440
|
| 3098 |
+
},
|
| 3099 |
+
{
|
| 3100 |
+
"epoch": 0.028819644410679933,
|
| 3101 |
+
"grad_norm": 3.483384370803833,
|
| 3102 |
+
"learning_rate": 0.00016147380073810346,
|
| 3103 |
+
"loss": 3.0496,
|
| 3104 |
+
"step": 441
|
| 3105 |
+
},
|
| 3106 |
+
{
|
| 3107 |
+
"epoch": 0.028884995078277846,
|
| 3108 |
+
"grad_norm": 3.988264322280884,
|
| 3109 |
+
"learning_rate": 0.0001613073640375934,
|
| 3110 |
+
"loss": 3.0479,
|
| 3111 |
+
"step": 442
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 0.02895034574587576,
|
| 3115 |
+
"grad_norm": 3.7168514728546143,
|
| 3116 |
+
"learning_rate": 0.00016114065479115946,
|
| 3117 |
+
"loss": 2.7044,
|
| 3118 |
+
"step": 443
|
| 3119 |
+
},
|
| 3120 |
+
{
|
| 3121 |
+
"epoch": 0.029015696413473673,
|
| 3122 |
+
"grad_norm": 3.7768290042877197,
|
| 3123 |
+
"learning_rate": 0.00016097367373991842,
|
| 3124 |
+
"loss": 3.2466,
|
| 3125 |
+
"step": 444
|
| 3126 |
+
},
|
| 3127 |
+
{
|
| 3128 |
+
"epoch": 0.02908104708107159,
|
| 3129 |
+
"grad_norm": 4.519565582275391,
|
| 3130 |
+
"learning_rate": 0.00016080642162619565,
|
| 3131 |
+
"loss": 2.6545,
|
| 3132 |
+
"step": 445
|
| 3133 |
+
},
|
| 3134 |
+
{
|
| 3135 |
+
"epoch": 0.029146397748669502,
|
| 3136 |
+
"grad_norm": 3.5160961151123047,
|
| 3137 |
+
"learning_rate": 0.0001606388991935214,
|
| 3138 |
+
"loss": 3.0352,
|
| 3139 |
+
"step": 446
|
| 3140 |
+
},
|
| 3141 |
+
{
|
| 3142 |
+
"epoch": 0.029211748416267416,
|
| 3143 |
+
"grad_norm": 3.72017502784729,
|
| 3144 |
+
"learning_rate": 0.0001604711071866277,
|
| 3145 |
+
"loss": 2.7479,
|
| 3146 |
+
"step": 447
|
| 3147 |
+
},
|
| 3148 |
+
{
|
| 3149 |
+
"epoch": 0.02927709908386533,
|
| 3150 |
+
"grad_norm": 4.312921047210693,
|
| 3151 |
+
"learning_rate": 0.00016030304635144494,
|
| 3152 |
+
"loss": 3.0772,
|
| 3153 |
+
"step": 448
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 0.029342449751463242,
|
| 3157 |
+
"grad_norm": 5.077251434326172,
|
| 3158 |
+
"learning_rate": 0.00016013471743509862,
|
| 3159 |
+
"loss": 2.801,
|
| 3160 |
+
"step": 449
|
| 3161 |
+
},
|
| 3162 |
+
{
|
| 3163 |
+
"epoch": 0.029407800419061155,
|
| 3164 |
+
"grad_norm": 7.480358600616455,
|
| 3165 |
+
"learning_rate": 0.00015996612118590603,
|
| 3166 |
+
"loss": 3.4775,
|
| 3167 |
+
"step": 450
|
| 3168 |
+
},
|
| 3169 |
+
{
|
| 3170 |
+
"epoch": 0.02947315108665907,
|
| 3171 |
+
"grad_norm": 24.4322452545166,
|
| 3172 |
+
"learning_rate": 0.00015979725835337294,
|
| 3173 |
+
"loss": 4.0432,
|
| 3174 |
+
"step": 451
|
| 3175 |
+
},
|
| 3176 |
+
{
|
| 3177 |
+
"epoch": 0.029538501754256985,
|
| 3178 |
+
"grad_norm": 8.49026107788086,
|
| 3179 |
+
"learning_rate": 0.00015962812968819016,
|
| 3180 |
+
"loss": 3.4552,
|
| 3181 |
+
"step": 452
|
| 3182 |
+
},
|
| 3183 |
+
{
|
| 3184 |
+
"epoch": 0.029603852421854898,
|
| 3185 |
+
"grad_norm": 7.049969673156738,
|
| 3186 |
+
"learning_rate": 0.0001594587359422303,
|
| 3187 |
+
"loss": 3.3839,
|
| 3188 |
+
"step": 453
|
| 3189 |
+
},
|
| 3190 |
+
{
|
| 3191 |
+
"epoch": 0.02966920308945281,
|
| 3192 |
+
"grad_norm": 9.08895206451416,
|
| 3193 |
+
"learning_rate": 0.0001592890778685444,
|
| 3194 |
+
"loss": 3.4686,
|
| 3195 |
+
"step": 454
|
| 3196 |
+
},
|
| 3197 |
+
{
|
| 3198 |
+
"epoch": 0.029734553757050725,
|
| 3199 |
+
"grad_norm": 3.652604818344116,
|
| 3200 |
+
"learning_rate": 0.00015911915622135862,
|
| 3201 |
+
"loss": 3.1602,
|
| 3202 |
+
"step": 455
|
| 3203 |
+
},
|
| 3204 |
+
{
|
| 3205 |
+
"epoch": 0.029799904424648638,
|
| 3206 |
+
"grad_norm": 8.94747257232666,
|
| 3207 |
+
"learning_rate": 0.00015894897175607086,
|
| 3208 |
+
"loss": 3.283,
|
| 3209 |
+
"step": 456
|
| 3210 |
+
},
|
| 3211 |
+
{
|
| 3212 |
+
"epoch": 0.02986525509224655,
|
| 3213 |
+
"grad_norm": 3.347947597503662,
|
| 3214 |
+
"learning_rate": 0.00015877852522924732,
|
| 3215 |
+
"loss": 3.0169,
|
| 3216 |
+
"step": 457
|
| 3217 |
+
},
|
| 3218 |
+
{
|
| 3219 |
+
"epoch": 0.029930605759844464,
|
| 3220 |
+
"grad_norm": 2.9615442752838135,
|
| 3221 |
+
"learning_rate": 0.00015860781739861928,
|
| 3222 |
+
"loss": 3.105,
|
| 3223 |
+
"step": 458
|
| 3224 |
+
},
|
| 3225 |
+
{
|
| 3226 |
+
"epoch": 0.02999595642744238,
|
| 3227 |
+
"grad_norm": 1.7090089321136475,
|
| 3228 |
+
"learning_rate": 0.00015843684902307962,
|
| 3229 |
+
"loss": 2.9634,
|
| 3230 |
+
"step": 459
|
| 3231 |
+
},
|
| 3232 |
+
{
|
| 3233 |
+
"epoch": 0.030061307095040294,
|
| 3234 |
+
"grad_norm": 1.87732994556427,
|
| 3235 |
+
"learning_rate": 0.00015826562086267956,
|
| 3236 |
+
"loss": 3.0923,
|
| 3237 |
+
"step": 460
|
| 3238 |
+
},
|
| 3239 |
+
{
|
| 3240 |
+
"epoch": 0.030126657762638207,
|
| 3241 |
+
"grad_norm": 1.8423588275909424,
|
| 3242 |
+
"learning_rate": 0.00015809413367862512,
|
| 3243 |
+
"loss": 3.0317,
|
| 3244 |
+
"step": 461
|
| 3245 |
+
},
|
| 3246 |
+
{
|
| 3247 |
+
"epoch": 0.03019200843023612,
|
| 3248 |
+
"grad_norm": 1.9231833219528198,
|
| 3249 |
+
"learning_rate": 0.00015792238823327388,
|
| 3250 |
+
"loss": 3.0236,
|
| 3251 |
+
"step": 462
|
| 3252 |
+
},
|
| 3253 |
+
{
|
| 3254 |
+
"epoch": 0.030257359097834034,
|
| 3255 |
+
"grad_norm": 1.718591570854187,
|
| 3256 |
+
"learning_rate": 0.00015775038529013152,
|
| 3257 |
+
"loss": 3.0055,
|
| 3258 |
+
"step": 463
|
| 3259 |
+
},
|
| 3260 |
+
{
|
| 3261 |
+
"epoch": 0.030322709765431947,
|
| 3262 |
+
"grad_norm": 1.6554956436157227,
|
| 3263 |
+
"learning_rate": 0.0001575781256138485,
|
| 3264 |
+
"loss": 2.9719,
|
| 3265 |
+
"step": 464
|
| 3266 |
+
},
|
| 3267 |
+
{
|
| 3268 |
+
"epoch": 0.03038806043302986,
|
| 3269 |
+
"grad_norm": 1.7320899963378906,
|
| 3270 |
+
"learning_rate": 0.00015740560997021648,
|
| 3271 |
+
"loss": 3.048,
|
| 3272 |
+
"step": 465
|
| 3273 |
+
},
|
| 3274 |
+
{
|
| 3275 |
+
"epoch": 0.030453411100627773,
|
| 3276 |
+
"grad_norm": 2.028738260269165,
|
| 3277 |
+
"learning_rate": 0.00015723283912616513,
|
| 3278 |
+
"loss": 3.1429,
|
| 3279 |
+
"step": 466
|
| 3280 |
+
},
|
| 3281 |
+
{
|
| 3282 |
+
"epoch": 0.03051876176822569,
|
| 3283 |
+
"grad_norm": 1.883371114730835,
|
| 3284 |
+
"learning_rate": 0.00015705981384975866,
|
| 3285 |
+
"loss": 3.0832,
|
| 3286 |
+
"step": 467
|
| 3287 |
+
},
|
| 3288 |
+
{
|
| 3289 |
+
"epoch": 0.030584112435823603,
|
| 3290 |
+
"grad_norm": 1.8488166332244873,
|
| 3291 |
+
"learning_rate": 0.0001568865349101923,
|
| 3292 |
+
"loss": 2.9372,
|
| 3293 |
+
"step": 468
|
| 3294 |
+
},
|
| 3295 |
+
{
|
| 3296 |
+
"epoch": 0.030649463103421516,
|
| 3297 |
+
"grad_norm": 2.0402255058288574,
|
| 3298 |
+
"learning_rate": 0.00015671300307778898,
|
| 3299 |
+
"loss": 3.2902,
|
| 3300 |
+
"step": 469
|
| 3301 |
+
},
|
| 3302 |
+
{
|
| 3303 |
+
"epoch": 0.03071481377101943,
|
| 3304 |
+
"grad_norm": 1.9097765684127808,
|
| 3305 |
+
"learning_rate": 0.00015653921912399589,
|
| 3306 |
+
"loss": 3.0758,
|
| 3307 |
+
"step": 470
|
| 3308 |
+
},
|
| 3309 |
+
{
|
| 3310 |
+
"epoch": 0.030780164438617343,
|
| 3311 |
+
"grad_norm": 1.973686695098877,
|
| 3312 |
+
"learning_rate": 0.00015636518382138107,
|
| 3313 |
+
"loss": 2.8372,
|
| 3314 |
+
"step": 471
|
| 3315 |
+
},
|
| 3316 |
+
{
|
| 3317 |
+
"epoch": 0.030845515106215256,
|
| 3318 |
+
"grad_norm": 1.9181742668151855,
|
| 3319 |
+
"learning_rate": 0.0001561908979436299,
|
| 3320 |
+
"loss": 2.8792,
|
| 3321 |
+
"step": 472
|
| 3322 |
+
},
|
| 3323 |
+
{
|
| 3324 |
+
"epoch": 0.03091086577381317,
|
| 3325 |
+
"grad_norm": 1.8591138124465942,
|
| 3326 |
+
"learning_rate": 0.00015601636226554168,
|
| 3327 |
+
"loss": 3.0248,
|
| 3328 |
+
"step": 473
|
| 3329 |
+
},
|
| 3330 |
+
{
|
| 3331 |
+
"epoch": 0.030976216441411086,
|
| 3332 |
+
"grad_norm": 1.9952205419540405,
|
| 3333 |
+
"learning_rate": 0.00015584157756302634,
|
| 3334 |
+
"loss": 2.8397,
|
| 3335 |
+
"step": 474
|
| 3336 |
+
},
|
| 3337 |
+
{
|
| 3338 |
+
"epoch": 0.031041567109009,
|
| 3339 |
+
"grad_norm": 2.088942289352417,
|
| 3340 |
+
"learning_rate": 0.0001556665446131007,
|
| 3341 |
+
"loss": 2.9455,
|
| 3342 |
+
"step": 475
|
| 3343 |
+
},
|
| 3344 |
+
{
|
| 3345 |
+
"epoch": 0.031106917776606912,
|
| 3346 |
+
"grad_norm": 2.370609998703003,
|
| 3347 |
+
"learning_rate": 0.00015549126419388536,
|
| 3348 |
+
"loss": 3.0924,
|
| 3349 |
+
"step": 476
|
| 3350 |
+
},
|
| 3351 |
+
{
|
| 3352 |
+
"epoch": 0.031172268444204825,
|
| 3353 |
+
"grad_norm": 2.1024417877197266,
|
| 3354 |
+
"learning_rate": 0.0001553157370846009,
|
| 3355 |
+
"loss": 2.857,
|
| 3356 |
+
"step": 477
|
| 3357 |
+
},
|
| 3358 |
+
{
|
| 3359 |
+
"epoch": 0.03123761911180274,
|
| 3360 |
+
"grad_norm": 2.367189407348633,
|
| 3361 |
+
"learning_rate": 0.00015513996406556465,
|
| 3362 |
+
"loss": 2.9313,
|
| 3363 |
+
"step": 478
|
| 3364 |
+
},
|
| 3365 |
+
{
|
| 3366 |
+
"epoch": 0.031302969779400655,
|
| 3367 |
+
"grad_norm": 2.203998327255249,
|
| 3368 |
+
"learning_rate": 0.00015496394591818716,
|
| 3369 |
+
"loss": 3.0986,
|
| 3370 |
+
"step": 479
|
| 3371 |
+
},
|
| 3372 |
+
{
|
| 3373 |
+
"epoch": 0.03136832044699857,
|
| 3374 |
+
"grad_norm": 2.1466145515441895,
|
| 3375 |
+
"learning_rate": 0.0001547876834249687,
|
| 3376 |
+
"loss": 2.7535,
|
| 3377 |
+
"step": 480
|
| 3378 |
+
},
|
| 3379 |
+
{
|
| 3380 |
+
"epoch": 0.03143367111459648,
|
| 3381 |
+
"grad_norm": 2.2550671100616455,
|
| 3382 |
+
"learning_rate": 0.00015461117736949577,
|
| 3383 |
+
"loss": 3.0209,
|
| 3384 |
+
"step": 481
|
| 3385 |
+
},
|
| 3386 |
+
{
|
| 3387 |
+
"epoch": 0.031499021782194395,
|
| 3388 |
+
"grad_norm": 2.7715814113616943,
|
| 3389 |
+
"learning_rate": 0.00015443442853643762,
|
| 3390 |
+
"loss": 3.1244,
|
| 3391 |
+
"step": 482
|
| 3392 |
+
},
|
| 3393 |
+
{
|
| 3394 |
+
"epoch": 0.03156437244979231,
|
| 3395 |
+
"grad_norm": 2.6216328144073486,
|
| 3396 |
+
"learning_rate": 0.00015425743771154294,
|
| 3397 |
+
"loss": 3.0967,
|
| 3398 |
+
"step": 483
|
| 3399 |
+
},
|
| 3400 |
+
{
|
| 3401 |
+
"epoch": 0.03162972311739022,
|
| 3402 |
+
"grad_norm": 2.9239511489868164,
|
| 3403 |
+
"learning_rate": 0.00015408020568163602,
|
| 3404 |
+
"loss": 2.9292,
|
| 3405 |
+
"step": 484
|
| 3406 |
+
},
|
| 3407 |
+
{
|
| 3408 |
+
"epoch": 0.031695073784988134,
|
| 3409 |
+
"grad_norm": 2.6827921867370605,
|
| 3410 |
+
"learning_rate": 0.00015390273323461352,
|
| 3411 |
+
"loss": 3.0856,
|
| 3412 |
+
"step": 485
|
| 3413 |
+
},
|
| 3414 |
+
{
|
| 3415 |
+
"epoch": 0.03176042445258605,
|
| 3416 |
+
"grad_norm": 2.7837822437286377,
|
| 3417 |
+
"learning_rate": 0.0001537250211594409,
|
| 3418 |
+
"loss": 2.9937,
|
| 3419 |
+
"step": 486
|
| 3420 |
+
},
|
| 3421 |
+
{
|
| 3422 |
+
"epoch": 0.03182577512018396,
|
| 3423 |
+
"grad_norm": 2.8517332077026367,
|
| 3424 |
+
"learning_rate": 0.0001535470702461489,
|
| 3425 |
+
"loss": 3.18,
|
| 3426 |
+
"step": 487
|
| 3427 |
+
},
|
| 3428 |
+
{
|
| 3429 |
+
"epoch": 0.031891125787781874,
|
| 3430 |
+
"grad_norm": 3.131091594696045,
|
| 3431 |
+
"learning_rate": 0.00015336888128583,
|
| 3432 |
+
"loss": 3.0751,
|
| 3433 |
+
"step": 488
|
| 3434 |
+
},
|
| 3435 |
+
{
|
| 3436 |
+
"epoch": 0.03195647645537979,
|
| 3437 |
+
"grad_norm": 3.4372997283935547,
|
| 3438 |
+
"learning_rate": 0.000153190455070635,
|
| 3439 |
+
"loss": 3.1115,
|
| 3440 |
+
"step": 489
|
| 3441 |
+
},
|
| 3442 |
+
{
|
| 3443 |
+
"epoch": 0.0320218271229777,
|
| 3444 |
+
"grad_norm": 3.2346112728118896,
|
| 3445 |
+
"learning_rate": 0.00015301179239376938,
|
| 3446 |
+
"loss": 3.1384,
|
| 3447 |
+
"step": 490
|
| 3448 |
+
},
|
| 3449 |
+
{
|
| 3450 |
+
"epoch": 0.03208717779057562,
|
| 3451 |
+
"grad_norm": 3.102898120880127,
|
| 3452 |
+
"learning_rate": 0.00015283289404948976,
|
| 3453 |
+
"loss": 3.0725,
|
| 3454 |
+
"step": 491
|
| 3455 |
+
},
|
| 3456 |
+
{
|
| 3457 |
+
"epoch": 0.032152528458173533,
|
| 3458 |
+
"grad_norm": 3.6215391159057617,
|
| 3459 |
+
"learning_rate": 0.0001526537608331006,
|
| 3460 |
+
"loss": 2.9572,
|
| 3461 |
+
"step": 492
|
| 3462 |
+
},
|
| 3463 |
+
{
|
| 3464 |
+
"epoch": 0.03221787912577145,
|
| 3465 |
+
"grad_norm": 3.090087652206421,
|
| 3466 |
+
"learning_rate": 0.00015247439354095041,
|
| 3467 |
+
"loss": 2.9867,
|
| 3468 |
+
"step": 493
|
| 3469 |
+
},
|
| 3470 |
+
{
|
| 3471 |
+
"epoch": 0.03228322979336936,
|
| 3472 |
+
"grad_norm": 3.4262967109680176,
|
| 3473 |
+
"learning_rate": 0.00015229479297042823,
|
| 3474 |
+
"loss": 2.8331,
|
| 3475 |
+
"step": 494
|
| 3476 |
+
},
|
| 3477 |
+
{
|
| 3478 |
+
"epoch": 0.03234858046096727,
|
| 3479 |
+
"grad_norm": 3.6537046432495117,
|
| 3480 |
+
"learning_rate": 0.00015211495991996027,
|
| 3481 |
+
"loss": 3.0767,
|
| 3482 |
+
"step": 495
|
| 3483 |
+
},
|
| 3484 |
+
{
|
| 3485 |
+
"epoch": 0.032413931128565186,
|
| 3486 |
+
"grad_norm": 3.5232338905334473,
|
| 3487 |
+
"learning_rate": 0.0001519348951890062,
|
| 3488 |
+
"loss": 2.8771,
|
| 3489 |
+
"step": 496
|
| 3490 |
+
},
|
| 3491 |
+
{
|
| 3492 |
+
"epoch": 0.0324792817961631,
|
| 3493 |
+
"grad_norm": 4.434952259063721,
|
| 3494 |
+
"learning_rate": 0.0001517545995780556,
|
| 3495 |
+
"loss": 2.8009,
|
| 3496 |
+
"step": 497
|
| 3497 |
+
},
|
| 3498 |
+
{
|
| 3499 |
+
"epoch": 0.03254463246376101,
|
| 3500 |
+
"grad_norm": 6.826545238494873,
|
| 3501 |
+
"learning_rate": 0.00015157407388862452,
|
| 3502 |
+
"loss": 2.8267,
|
| 3503 |
+
"step": 498
|
| 3504 |
+
},
|
| 3505 |
+
{
|
| 3506 |
+
"epoch": 0.032609983131358926,
|
| 3507 |
+
"grad_norm": 4.22343111038208,
|
| 3508 |
+
"learning_rate": 0.00015139331892325179,
|
| 3509 |
+
"loss": 2.3913,
|
| 3510 |
+
"step": 499
|
| 3511 |
+
},
|
| 3512 |
+
{
|
| 3513 |
+
"epoch": 0.03267533379895684,
|
| 3514 |
+
"grad_norm": 6.091843128204346,
|
| 3515 |
+
"learning_rate": 0.0001512123354854955,
|
| 3516 |
+
"loss": 2.8988,
|
| 3517 |
+
"step": 500
|
| 3518 |
+
},
|
| 3519 |
+
{
|
| 3520 |
+
"epoch": 0.03274068446655475,
|
| 3521 |
+
"grad_norm": 11.71518611907959,
|
| 3522 |
+
"learning_rate": 0.0001510311243799295,
|
| 3523 |
+
"loss": 3.6692,
|
| 3524 |
+
"step": 501
|
| 3525 |
+
},
|
| 3526 |
+
{
|
| 3527 |
+
"epoch": 0.032806035134152665,
|
| 3528 |
+
"grad_norm": 7.051513671875,
|
| 3529 |
+
"learning_rate": 0.00015084968641213958,
|
| 3530 |
+
"loss": 3.5298,
|
| 3531 |
+
"step": 502
|
| 3532 |
+
},
|
| 3533 |
+
{
|
| 3534 |
+
"epoch": 0.03287138580175058,
|
| 3535 |
+
"grad_norm": 4.335301399230957,
|
| 3536 |
+
"learning_rate": 0.00015066802238872023,
|
| 3537 |
+
"loss": 3.234,
|
| 3538 |
+
"step": 503
|
| 3539 |
+
},
|
| 3540 |
+
{
|
| 3541 |
+
"epoch": 0.03293673646934849,
|
| 3542 |
+
"grad_norm": 3.7799558639526367,
|
| 3543 |
+
"learning_rate": 0.0001504861331172709,
|
| 3544 |
+
"loss": 3.0844,
|
| 3545 |
+
"step": 504
|
| 3546 |
+
},
|
| 3547 |
+
{
|
| 3548 |
+
"epoch": 0.033002087136946405,
|
| 3549 |
+
"grad_norm": 2.185335397720337,
|
| 3550 |
+
"learning_rate": 0.0001503040194063922,
|
| 3551 |
+
"loss": 3.1615,
|
| 3552 |
+
"step": 505
|
| 3553 |
+
},
|
| 3554 |
+
{
|
| 3555 |
+
"epoch": 0.033067437804544325,
|
| 3556 |
+
"grad_norm": 1.9385499954223633,
|
| 3557 |
+
"learning_rate": 0.00015012168206568268,
|
| 3558 |
+
"loss": 3.1164,
|
| 3559 |
+
"step": 506
|
| 3560 |
+
},
|
| 3561 |
+
{
|
| 3562 |
+
"epoch": 0.03313278847214224,
|
| 3563 |
+
"grad_norm": 1.9973067045211792,
|
| 3564 |
+
"learning_rate": 0.00014993912190573505,
|
| 3565 |
+
"loss": 3.0556,
|
| 3566 |
+
"step": 507
|
| 3567 |
+
},
|
| 3568 |
+
{
|
| 3569 |
+
"epoch": 0.03319813913974015,
|
| 3570 |
+
"grad_norm": 1.8199883699417114,
|
| 3571 |
+
"learning_rate": 0.00014975633973813242,
|
| 3572 |
+
"loss": 2.9567,
|
| 3573 |
+
"step": 508
|
| 3574 |
+
},
|
| 3575 |
+
{
|
| 3576 |
+
"epoch": 0.033263489807338065,
|
| 3577 |
+
"grad_norm": 1.5722929239273071,
|
| 3578 |
+
"learning_rate": 0.00014957333637544503,
|
| 3579 |
+
"loss": 2.9897,
|
| 3580 |
+
"step": 509
|
| 3581 |
+
},
|
| 3582 |
+
{
|
| 3583 |
+
"epoch": 0.03332884047493598,
|
| 3584 |
+
"grad_norm": 1.9580084085464478,
|
| 3585 |
+
"learning_rate": 0.00014939011263122634,
|
| 3586 |
+
"loss": 3.0393,
|
| 3587 |
+
"step": 510
|
| 3588 |
+
},
|
| 3589 |
+
{
|
| 3590 |
+
"epoch": 0.03339419114253389,
|
| 3591 |
+
"grad_norm": 1.7566349506378174,
|
| 3592 |
+
"learning_rate": 0.0001492066693200096,
|
| 3593 |
+
"loss": 3.1124,
|
| 3594 |
+
"step": 511
|
| 3595 |
+
},
|
| 3596 |
+
{
|
| 3597 |
+
"epoch": 0.033459541810131804,
|
| 3598 |
+
"grad_norm": 1.995408058166504,
|
| 3599 |
+
"learning_rate": 0.00014902300725730413,
|
| 3600 |
+
"loss": 3.0233,
|
| 3601 |
+
"step": 512
|
| 3602 |
+
},
|
| 3603 |
+
{
|
| 3604 |
+
"epoch": 0.03352489247772972,
|
| 3605 |
+
"grad_norm": 1.8557460308074951,
|
| 3606 |
+
"learning_rate": 0.00014883912725959167,
|
| 3607 |
+
"loss": 3.0566,
|
| 3608 |
+
"step": 513
|
| 3609 |
+
},
|
| 3610 |
+
{
|
| 3611 |
+
"epoch": 0.03359024314532763,
|
| 3612 |
+
"grad_norm": 1.7373842000961304,
|
| 3613 |
+
"learning_rate": 0.00014865503014432292,
|
| 3614 |
+
"loss": 2.9989,
|
| 3615 |
+
"step": 514
|
| 3616 |
+
},
|
| 3617 |
+
{
|
| 3618 |
+
"epoch": 0.033655593812925544,
|
| 3619 |
+
"grad_norm": 2.1281046867370605,
|
| 3620 |
+
"learning_rate": 0.00014847071672991367,
|
| 3621 |
+
"loss": 3.1251,
|
| 3622 |
+
"step": 515
|
| 3623 |
+
},
|
| 3624 |
+
{
|
| 3625 |
+
"epoch": 0.03372094448052346,
|
| 3626 |
+
"grad_norm": 1.8922642469406128,
|
| 3627 |
+
"learning_rate": 0.0001482861878357414,
|
| 3628 |
+
"loss": 3.0562,
|
| 3629 |
+
"step": 516
|
| 3630 |
+
},
|
| 3631 |
+
{
|
| 3632 |
+
"epoch": 0.03378629514812137,
|
| 3633 |
+
"grad_norm": 2.2692742347717285,
|
| 3634 |
+
"learning_rate": 0.00014810144428214144,
|
| 3635 |
+
"loss": 3.1552,
|
| 3636 |
+
"step": 517
|
| 3637 |
+
},
|
| 3638 |
+
{
|
| 3639 |
+
"epoch": 0.03385164581571928,
|
| 3640 |
+
"grad_norm": 1.7412115335464478,
|
| 3641 |
+
"learning_rate": 0.0001479164868904034,
|
| 3642 |
+
"loss": 2.8901,
|
| 3643 |
+
"step": 518
|
| 3644 |
+
},
|
| 3645 |
+
{
|
| 3646 |
+
"epoch": 0.0339169964833172,
|
| 3647 |
+
"grad_norm": 2.91524338722229,
|
| 3648 |
+
"learning_rate": 0.00014773131648276758,
|
| 3649 |
+
"loss": 3.2572,
|
| 3650 |
+
"step": 519
|
| 3651 |
+
},
|
| 3652 |
+
{
|
| 3653 |
+
"epoch": 0.03398234715091512,
|
| 3654 |
+
"grad_norm": 2.537008047103882,
|
| 3655 |
+
"learning_rate": 0.00014754593388242117,
|
| 3656 |
+
"loss": 3.1831,
|
| 3657 |
+
"step": 520
|
| 3658 |
+
},
|
| 3659 |
+
{
|
| 3660 |
+
"epoch": 0.03404769781851303,
|
| 3661 |
+
"grad_norm": 2.0767822265625,
|
| 3662 |
+
"learning_rate": 0.0001473603399134948,
|
| 3663 |
+
"loss": 2.9038,
|
| 3664 |
+
"step": 521
|
| 3665 |
+
},
|
| 3666 |
+
{
|
| 3667 |
+
"epoch": 0.03411304848611094,
|
| 3668 |
+
"grad_norm": 2.3586618900299072,
|
| 3669 |
+
"learning_rate": 0.0001471745354010586,
|
| 3670 |
+
"loss": 3.1186,
|
| 3671 |
+
"step": 522
|
| 3672 |
+
},
|
| 3673 |
+
{
|
| 3674 |
+
"epoch": 0.034178399153708856,
|
| 3675 |
+
"grad_norm": 2.0097970962524414,
|
| 3676 |
+
"learning_rate": 0.00014698852117111884,
|
| 3677 |
+
"loss": 3.0701,
|
| 3678 |
+
"step": 523
|
| 3679 |
+
},
|
| 3680 |
+
{
|
| 3681 |
+
"epoch": 0.03424374982130677,
|
| 3682 |
+
"grad_norm": 2.0309152603149414,
|
| 3683 |
+
"learning_rate": 0.000146802298050614,
|
| 3684 |
+
"loss": 3.044,
|
| 3685 |
+
"step": 524
|
| 3686 |
+
},
|
| 3687 |
+
{
|
| 3688 |
+
"epoch": 0.03430910048890468,
|
| 3689 |
+
"grad_norm": 2.0452139377593994,
|
| 3690 |
+
"learning_rate": 0.0001466158668674112,
|
| 3691 |
+
"loss": 2.9056,
|
| 3692 |
+
"step": 525
|
| 3693 |
+
},
|
| 3694 |
+
{
|
| 3695 |
+
"epoch": 0.034374451156502596,
|
| 3696 |
+
"grad_norm": 2.242048740386963,
|
| 3697 |
+
"learning_rate": 0.00014642922845030257,
|
| 3698 |
+
"loss": 3.3416,
|
| 3699 |
+
"step": 526
|
| 3700 |
+
},
|
| 3701 |
+
{
|
| 3702 |
+
"epoch": 0.03443980182410051,
|
| 3703 |
+
"grad_norm": 2.0582847595214844,
|
| 3704 |
+
"learning_rate": 0.0001462423836290015,
|
| 3705 |
+
"loss": 2.9922,
|
| 3706 |
+
"step": 527
|
| 3707 |
+
},
|
| 3708 |
+
{
|
| 3709 |
+
"epoch": 0.03450515249169842,
|
| 3710 |
+
"grad_norm": 1.8688050508499146,
|
| 3711 |
+
"learning_rate": 0.00014605533323413887,
|
| 3712 |
+
"loss": 2.9809,
|
| 3713 |
+
"step": 528
|
| 3714 |
+
},
|
| 3715 |
+
{
|
| 3716 |
+
"epoch": 0.034570503159296335,
|
| 3717 |
+
"grad_norm": 2.1673102378845215,
|
| 3718 |
+
"learning_rate": 0.00014586807809725962,
|
| 3719 |
+
"loss": 2.9412,
|
| 3720 |
+
"step": 529
|
| 3721 |
+
},
|
| 3722 |
+
{
|
| 3723 |
+
"epoch": 0.03463585382689425,
|
| 3724 |
+
"grad_norm": 2.2136361598968506,
|
| 3725 |
+
"learning_rate": 0.00014568061905081875,
|
| 3726 |
+
"loss": 2.9397,
|
| 3727 |
+
"step": 530
|
| 3728 |
+
},
|
| 3729 |
+
{
|
| 3730 |
+
"epoch": 0.03470120449449216,
|
| 3731 |
+
"grad_norm": 2.265554428100586,
|
| 3732 |
+
"learning_rate": 0.00014549295692817778,
|
| 3733 |
+
"loss": 3.094,
|
| 3734 |
+
"step": 531
|
| 3735 |
+
},
|
| 3736 |
+
{
|
| 3737 |
+
"epoch": 0.034766555162090075,
|
| 3738 |
+
"grad_norm": 2.590942621231079,
|
| 3739 |
+
"learning_rate": 0.00014530509256360102,
|
| 3740 |
+
"loss": 3.0893,
|
| 3741 |
+
"step": 532
|
| 3742 |
+
},
|
| 3743 |
+
{
|
| 3744 |
+
"epoch": 0.03483190582968799,
|
| 3745 |
+
"grad_norm": 2.161839485168457,
|
| 3746 |
+
"learning_rate": 0.00014511702679225193,
|
| 3747 |
+
"loss": 2.8916,
|
| 3748 |
+
"step": 533
|
| 3749 |
+
},
|
| 3750 |
+
{
|
| 3751 |
+
"epoch": 0.03489725649728591,
|
| 3752 |
+
"grad_norm": 2.8425076007843018,
|
| 3753 |
+
"learning_rate": 0.0001449287604501893,
|
| 3754 |
+
"loss": 3.0664,
|
| 3755 |
+
"step": 534
|
| 3756 |
+
},
|
| 3757 |
+
{
|
| 3758 |
+
"epoch": 0.03496260716488382,
|
| 3759 |
+
"grad_norm": 2.508582592010498,
|
| 3760 |
+
"learning_rate": 0.00014474029437436348,
|
| 3761 |
+
"loss": 3.0965,
|
| 3762 |
+
"step": 535
|
| 3763 |
+
},
|
| 3764 |
+
{
|
| 3765 |
+
"epoch": 0.035027957832481735,
|
| 3766 |
+
"grad_norm": 2.5080628395080566,
|
| 3767 |
+
"learning_rate": 0.00014455162940261285,
|
| 3768 |
+
"loss": 2.913,
|
| 3769 |
+
"step": 536
|
| 3770 |
+
},
|
| 3771 |
+
{
|
| 3772 |
+
"epoch": 0.03509330850007965,
|
| 3773 |
+
"grad_norm": 2.7472963333129883,
|
| 3774 |
+
"learning_rate": 0.0001443627663736599,
|
| 3775 |
+
"loss": 3.034,
|
| 3776 |
+
"step": 537
|
| 3777 |
+
},
|
| 3778 |
+
{
|
| 3779 |
+
"epoch": 0.03515865916767756,
|
| 3780 |
+
"grad_norm": 3.1874053478240967,
|
| 3781 |
+
"learning_rate": 0.00014417370612710778,
|
| 3782 |
+
"loss": 3.1924,
|
| 3783 |
+
"step": 538
|
| 3784 |
+
},
|
| 3785 |
+
{
|
| 3786 |
+
"epoch": 0.035224009835275474,
|
| 3787 |
+
"grad_norm": 3.3476574420928955,
|
| 3788 |
+
"learning_rate": 0.00014398444950343623,
|
| 3789 |
+
"loss": 2.8451,
|
| 3790 |
+
"step": 539
|
| 3791 |
+
},
|
| 3792 |
+
{
|
| 3793 |
+
"epoch": 0.03528936050287339,
|
| 3794 |
+
"grad_norm": 3.191981315612793,
|
| 3795 |
+
"learning_rate": 0.00014379499734399798,
|
| 3796 |
+
"loss": 3.0234,
|
| 3797 |
+
"step": 540
|
| 3798 |
+
},
|
| 3799 |
+
{
|
| 3800 |
+
"epoch": 0.0353547111704713,
|
| 3801 |
+
"grad_norm": 3.4992892742156982,
|
| 3802 |
+
"learning_rate": 0.0001436053504910151,
|
| 3803 |
+
"loss": 2.6897,
|
| 3804 |
+
"step": 541
|
| 3805 |
+
},
|
| 3806 |
+
{
|
| 3807 |
+
"epoch": 0.035420061838069214,
|
| 3808 |
+
"grad_norm": 3.7841637134552,
|
| 3809 |
+
"learning_rate": 0.0001434155097875752,
|
| 3810 |
+
"loss": 3.2661,
|
| 3811 |
+
"step": 542
|
| 3812 |
+
},
|
| 3813 |
+
{
|
| 3814 |
+
"epoch": 0.03548541250566713,
|
| 3815 |
+
"grad_norm": 4.506886005401611,
|
| 3816 |
+
"learning_rate": 0.00014322547607762762,
|
| 3817 |
+
"loss": 2.9592,
|
| 3818 |
+
"step": 543
|
| 3819 |
+
},
|
| 3820 |
+
{
|
| 3821 |
+
"epoch": 0.03555076317326504,
|
| 3822 |
+
"grad_norm": 4.825909614562988,
|
| 3823 |
+
"learning_rate": 0.0001430352502059797,
|
| 3824 |
+
"loss": 2.7436,
|
| 3825 |
+
"step": 544
|
| 3826 |
+
},
|
| 3827 |
+
{
|
| 3828 |
+
"epoch": 0.035616113840862954,
|
| 3829 |
+
"grad_norm": 3.0600483417510986,
|
| 3830 |
+
"learning_rate": 0.0001428448330182931,
|
| 3831 |
+
"loss": 2.778,
|
| 3832 |
+
"step": 545
|
| 3833 |
+
},
|
| 3834 |
+
{
|
| 3835 |
+
"epoch": 0.03568146450846087,
|
| 3836 |
+
"grad_norm": 9.061102867126465,
|
| 3837 |
+
"learning_rate": 0.00014265422536107993,
|
| 3838 |
+
"loss": 3.1647,
|
| 3839 |
+
"step": 546
|
| 3840 |
+
},
|
| 3841 |
+
{
|
| 3842 |
+
"epoch": 0.03574681517605878,
|
| 3843 |
+
"grad_norm": 7.136674880981445,
|
| 3844 |
+
"learning_rate": 0.00014246342808169914,
|
| 3845 |
+
"loss": 2.5426,
|
| 3846 |
+
"step": 547
|
| 3847 |
+
},
|
| 3848 |
+
{
|
| 3849 |
+
"epoch": 0.0358121658436567,
|
| 3850 |
+
"grad_norm": 5.525428295135498,
|
| 3851 |
+
"learning_rate": 0.00014227244202835257,
|
| 3852 |
+
"loss": 2.6582,
|
| 3853 |
+
"step": 548
|
| 3854 |
+
},
|
| 3855 |
+
{
|
| 3856 |
+
"epoch": 0.03587751651125461,
|
| 3857 |
+
"grad_norm": 5.315718650817871,
|
| 3858 |
+
"learning_rate": 0.0001420812680500813,
|
| 3859 |
+
"loss": 2.8048,
|
| 3860 |
+
"step": 549
|
| 3861 |
+
},
|
| 3862 |
+
{
|
| 3863 |
+
"epoch": 0.035942867178852526,
|
| 3864 |
+
"grad_norm": 8.281085014343262,
|
| 3865 |
+
"learning_rate": 0.00014188990699676184,
|
| 3866 |
+
"loss": 2.8449,
|
| 3867 |
+
"step": 550
|
| 3868 |
+
},
|
| 3869 |
+
{
|
| 3870 |
+
"epoch": 0.03600821784645044,
|
| 3871 |
+
"grad_norm": 8.362018585205078,
|
| 3872 |
+
"learning_rate": 0.00014169835971910238,
|
| 3873 |
+
"loss": 3.1894,
|
| 3874 |
+
"step": 551
|
| 3875 |
+
},
|
| 3876 |
+
{
|
| 3877 |
+
"epoch": 0.03607356851404835,
|
| 3878 |
+
"grad_norm": 11.041610717773438,
|
| 3879 |
+
"learning_rate": 0.0001415066270686389,
|
| 3880 |
+
"loss": 3.4059,
|
| 3881 |
+
"step": 552
|
| 3882 |
+
},
|
| 3883 |
+
{
|
| 3884 |
+
"epoch": 0.036138919181646266,
|
| 3885 |
+
"grad_norm": 3.9808542728424072,
|
| 3886 |
+
"learning_rate": 0.00014131470989773158,
|
| 3887 |
+
"loss": 3.2764,
|
| 3888 |
+
"step": 553
|
| 3889 |
+
},
|
| 3890 |
+
{
|
| 3891 |
+
"epoch": 0.03620426984924418,
|
| 3892 |
+
"grad_norm": 13.12489128112793,
|
| 3893 |
+
"learning_rate": 0.0001411226090595608,
|
| 3894 |
+
"loss": 3.2788,
|
| 3895 |
+
"step": 554
|
| 3896 |
+
},
|
| 3897 |
+
{
|
| 3898 |
+
"epoch": 0.03626962051684209,
|
| 3899 |
+
"grad_norm": 8.727335929870605,
|
| 3900 |
+
"learning_rate": 0.00014093032540812348,
|
| 3901 |
+
"loss": 3.2385,
|
| 3902 |
+
"step": 555
|
| 3903 |
+
},
|
| 3904 |
+
{
|
| 3905 |
+
"epoch": 0.036334971184440006,
|
| 3906 |
+
"grad_norm": 3.04437255859375,
|
| 3907 |
+
"learning_rate": 0.0001407378597982293,
|
| 3908 |
+
"loss": 3.2088,
|
| 3909 |
+
"step": 556
|
| 3910 |
+
},
|
| 3911 |
+
{
|
| 3912 |
+
"epoch": 0.03640032185203792,
|
| 3913 |
+
"grad_norm": 3.878338575363159,
|
| 3914 |
+
"learning_rate": 0.00014054521308549673,
|
| 3915 |
+
"loss": 3.1957,
|
| 3916 |
+
"step": 557
|
| 3917 |
+
},
|
| 3918 |
+
{
|
| 3919 |
+
"epoch": 0.03646567251963583,
|
| 3920 |
+
"grad_norm": 2.566128969192505,
|
| 3921 |
+
"learning_rate": 0.0001403523861263495,
|
| 3922 |
+
"loss": 3.0187,
|
| 3923 |
+
"step": 558
|
| 3924 |
+
},
|
| 3925 |
+
{
|
| 3926 |
+
"epoch": 0.036531023187233745,
|
| 3927 |
+
"grad_norm": 2.8429207801818848,
|
| 3928 |
+
"learning_rate": 0.00014015937977801256,
|
| 3929 |
+
"loss": 3.2802,
|
| 3930 |
+
"step": 559
|
| 3931 |
+
},
|
| 3932 |
+
{
|
| 3933 |
+
"epoch": 0.03659637385483166,
|
| 3934 |
+
"grad_norm": 2.0213189125061035,
|
| 3935 |
+
"learning_rate": 0.00013996619489850822,
|
| 3936 |
+
"loss": 3.1732,
|
| 3937 |
+
"step": 560
|
| 3938 |
+
},
|
| 3939 |
+
{
|
| 3940 |
+
"epoch": 0.03666172452242957,
|
| 3941 |
+
"grad_norm": 2.244999408721924,
|
| 3942 |
+
"learning_rate": 0.00013977283234665273,
|
| 3943 |
+
"loss": 3.1004,
|
| 3944 |
+
"step": 561
|
| 3945 |
+
},
|
| 3946 |
+
{
|
| 3947 |
+
"epoch": 0.036727075190027485,
|
| 3948 |
+
"grad_norm": 1.8090487718582153,
|
| 3949 |
+
"learning_rate": 0.00013957929298205195,
|
| 3950 |
+
"loss": 3.0267,
|
| 3951 |
+
"step": 562
|
| 3952 |
+
},
|
| 3953 |
+
{
|
| 3954 |
+
"epoch": 0.036792425857625405,
|
| 3955 |
+
"grad_norm": 1.7424795627593994,
|
| 3956 |
+
"learning_rate": 0.00013938557766509792,
|
| 3957 |
+
"loss": 2.8685,
|
| 3958 |
+
"step": 563
|
| 3959 |
+
},
|
| 3960 |
+
{
|
| 3961 |
+
"epoch": 0.03685777652522332,
|
| 3962 |
+
"grad_norm": 2.0039384365081787,
|
| 3963 |
+
"learning_rate": 0.0001391916872569648,
|
| 3964 |
+
"loss": 2.9659,
|
| 3965 |
+
"step": 564
|
| 3966 |
+
},
|
| 3967 |
+
{
|
| 3968 |
+
"epoch": 0.03692312719282123,
|
| 3969 |
+
"grad_norm": 1.816052794456482,
|
| 3970 |
+
"learning_rate": 0.00013899762261960518,
|
| 3971 |
+
"loss": 3.0037,
|
| 3972 |
+
"step": 565
|
| 3973 |
+
},
|
| 3974 |
+
{
|
| 3975 |
+
"epoch": 0.036988477860419144,
|
| 3976 |
+
"grad_norm": 1.7479966878890991,
|
| 3977 |
+
"learning_rate": 0.0001388033846157462,
|
| 3978 |
+
"loss": 2.8999,
|
| 3979 |
+
"step": 566
|
| 3980 |
+
},
|
| 3981 |
+
{
|
| 3982 |
+
"epoch": 0.03705382852801706,
|
| 3983 |
+
"grad_norm": 1.7347345352172852,
|
| 3984 |
+
"learning_rate": 0.0001386089741088857,
|
| 3985 |
+
"loss": 2.9378,
|
| 3986 |
+
"step": 567
|
| 3987 |
+
},
|
| 3988 |
+
{
|
| 3989 |
+
"epoch": 0.03711917919561497,
|
| 3990 |
+
"grad_norm": 1.8642627000808716,
|
| 3991 |
+
"learning_rate": 0.00013841439196328836,
|
| 3992 |
+
"loss": 3.0348,
|
| 3993 |
+
"step": 568
|
| 3994 |
+
},
|
| 3995 |
+
{
|
| 3996 |
+
"epoch": 0.037184529863212884,
|
| 3997 |
+
"grad_norm": 1.9142097234725952,
|
| 3998 |
+
"learning_rate": 0.00013821963904398193,
|
| 3999 |
+
"loss": 3.1449,
|
| 4000 |
+
"step": 569
|
| 4001 |
+
},
|
| 4002 |
+
{
|
| 4003 |
+
"epoch": 0.0372498805308108,
|
| 4004 |
+
"grad_norm": 1.9393805265426636,
|
| 4005 |
+
"learning_rate": 0.00013802471621675338,
|
| 4006 |
+
"loss": 3.0023,
|
| 4007 |
+
"step": 570
|
| 4008 |
+
},
|
| 4009 |
+
{
|
| 4010 |
+
"epoch": 0.03731523119840871,
|
| 4011 |
+
"grad_norm": 1.9792324304580688,
|
| 4012 |
+
"learning_rate": 0.00013782962434814492,
|
| 4013 |
+
"loss": 3.0849,
|
| 4014 |
+
"step": 571
|
| 4015 |
+
},
|
| 4016 |
+
{
|
| 4017 |
+
"epoch": 0.037380581866006624,
|
| 4018 |
+
"grad_norm": 2.1985344886779785,
|
| 4019 |
+
"learning_rate": 0.00013763436430545034,
|
| 4020 |
+
"loss": 2.9901,
|
| 4021 |
+
"step": 572
|
| 4022 |
+
},
|
| 4023 |
+
{
|
| 4024 |
+
"epoch": 0.03744593253360454,
|
| 4025 |
+
"grad_norm": 1.9945766925811768,
|
| 4026 |
+
"learning_rate": 0.00013743893695671096,
|
| 4027 |
+
"loss": 2.9633,
|
| 4028 |
+
"step": 573
|
| 4029 |
+
},
|
| 4030 |
+
{
|
| 4031 |
+
"epoch": 0.03751128320120245,
|
| 4032 |
+
"grad_norm": 1.930016279220581,
|
| 4033 |
+
"learning_rate": 0.00013724334317071198,
|
| 4034 |
+
"loss": 3.007,
|
| 4035 |
+
"step": 574
|
| 4036 |
+
},
|
| 4037 |
+
{
|
| 4038 |
+
"epoch": 0.03757663386880036,
|
| 4039 |
+
"grad_norm": 2.070556402206421,
|
| 4040 |
+
"learning_rate": 0.00013704758381697844,
|
| 4041 |
+
"loss": 3.1439,
|
| 4042 |
+
"step": 575
|
| 4043 |
+
},
|
| 4044 |
+
{
|
| 4045 |
+
"epoch": 0.037641984536398276,
|
| 4046 |
+
"grad_norm": 2.0270776748657227,
|
| 4047 |
+
"learning_rate": 0.00013685165976577146,
|
| 4048 |
+
"loss": 3.0146,
|
| 4049 |
+
"step": 576
|
| 4050 |
+
},
|
| 4051 |
+
{
|
| 4052 |
+
"epoch": 0.037707335203996196,
|
| 4053 |
+
"grad_norm": 2.047429323196411,
|
| 4054 |
+
"learning_rate": 0.0001366555718880843,
|
| 4055 |
+
"loss": 2.8983,
|
| 4056 |
+
"step": 577
|
| 4057 |
+
},
|
| 4058 |
+
{
|
| 4059 |
+
"epoch": 0.03777268587159411,
|
| 4060 |
+
"grad_norm": 2.2146401405334473,
|
| 4061 |
+
"learning_rate": 0.00013645932105563844,
|
| 4062 |
+
"loss": 2.9096,
|
| 4063 |
+
"step": 578
|
| 4064 |
+
},
|
| 4065 |
+
{
|
| 4066 |
+
"epoch": 0.03783803653919202,
|
| 4067 |
+
"grad_norm": 2.0988502502441406,
|
| 4068 |
+
"learning_rate": 0.00013626290814088005,
|
| 4069 |
+
"loss": 2.9559,
|
| 4070 |
+
"step": 579
|
| 4071 |
+
},
|
| 4072 |
+
{
|
| 4073 |
+
"epoch": 0.037903387206789936,
|
| 4074 |
+
"grad_norm": 2.1318140029907227,
|
| 4075 |
+
"learning_rate": 0.00013606633401697557,
|
| 4076 |
+
"loss": 2.8678,
|
| 4077 |
+
"step": 580
|
| 4078 |
+
},
|
| 4079 |
+
{
|
| 4080 |
+
"epoch": 0.03796873787438785,
|
| 4081 |
+
"grad_norm": 2.47351336479187,
|
| 4082 |
+
"learning_rate": 0.00013586959955780824,
|
| 4083 |
+
"loss": 3.1913,
|
| 4084 |
+
"step": 581
|
| 4085 |
+
},
|
| 4086 |
+
{
|
| 4087 |
+
"epoch": 0.03803408854198576,
|
| 4088 |
+
"grad_norm": 2.630089044570923,
|
| 4089 |
+
"learning_rate": 0.00013567270563797398,
|
| 4090 |
+
"loss": 3.4317,
|
| 4091 |
+
"step": 582
|
| 4092 |
+
},
|
| 4093 |
+
{
|
| 4094 |
+
"epoch": 0.038099439209583676,
|
| 4095 |
+
"grad_norm": 2.557398557662964,
|
| 4096 |
+
"learning_rate": 0.00013547565313277776,
|
| 4097 |
+
"loss": 2.9004,
|
| 4098 |
+
"step": 583
|
| 4099 |
+
},
|
| 4100 |
+
{
|
| 4101 |
+
"epoch": 0.03816478987718159,
|
| 4102 |
+
"grad_norm": 2.377253293991089,
|
| 4103 |
+
"learning_rate": 0.00013527844291822948,
|
| 4104 |
+
"loss": 2.8715,
|
| 4105 |
+
"step": 584
|
| 4106 |
+
},
|
| 4107 |
+
{
|
| 4108 |
+
"epoch": 0.0382301405447795,
|
| 4109 |
+
"grad_norm": 2.532958984375,
|
| 4110 |
+
"learning_rate": 0.0001350810758710401,
|
| 4111 |
+
"loss": 3.0972,
|
| 4112 |
+
"step": 585
|
| 4113 |
+
},
|
| 4114 |
+
{
|
| 4115 |
+
"epoch": 0.038295491212377415,
|
| 4116 |
+
"grad_norm": 2.6120903491973877,
|
| 4117 |
+
"learning_rate": 0.00013488355286861783,
|
| 4118 |
+
"loss": 2.9928,
|
| 4119 |
+
"step": 586
|
| 4120 |
+
},
|
| 4121 |
+
{
|
| 4122 |
+
"epoch": 0.03836084187997533,
|
| 4123 |
+
"grad_norm": 2.865553617477417,
|
| 4124 |
+
"learning_rate": 0.0001346858747890642,
|
| 4125 |
+
"loss": 2.9036,
|
| 4126 |
+
"step": 587
|
| 4127 |
+
},
|
| 4128 |
+
{
|
| 4129 |
+
"epoch": 0.03842619254757324,
|
| 4130 |
+
"grad_norm": 2.734539270401001,
|
| 4131 |
+
"learning_rate": 0.00013448804251117003,
|
| 4132 |
+
"loss": 3.1063,
|
| 4133 |
+
"step": 588
|
| 4134 |
+
},
|
| 4135 |
+
{
|
| 4136 |
+
"epoch": 0.038491543215171155,
|
| 4137 |
+
"grad_norm": 3.559826374053955,
|
| 4138 |
+
"learning_rate": 0.0001342900569144119,
|
| 4139 |
+
"loss": 3.0002,
|
| 4140 |
+
"step": 589
|
| 4141 |
+
},
|
| 4142 |
+
{
|
| 4143 |
+
"epoch": 0.03855689388276907,
|
| 4144 |
+
"grad_norm": 3.4512901306152344,
|
| 4145 |
+
"learning_rate": 0.0001340919188789477,
|
| 4146 |
+
"loss": 3.0635,
|
| 4147 |
+
"step": 590
|
| 4148 |
+
},
|
| 4149 |
+
{
|
| 4150 |
+
"epoch": 0.03862224455036699,
|
| 4151 |
+
"grad_norm": 4.525444030761719,
|
| 4152 |
+
"learning_rate": 0.00013389362928561317,
|
| 4153 |
+
"loss": 2.904,
|
| 4154 |
+
"step": 591
|
| 4155 |
+
},
|
| 4156 |
+
{
|
| 4157 |
+
"epoch": 0.0386875952179649,
|
| 4158 |
+
"grad_norm": 2.985776424407959,
|
| 4159 |
+
"learning_rate": 0.00013369518901591772,
|
| 4160 |
+
"loss": 2.8948,
|
| 4161 |
+
"step": 592
|
| 4162 |
+
},
|
| 4163 |
+
{
|
| 4164 |
+
"epoch": 0.038752945885562814,
|
| 4165 |
+
"grad_norm": 3.0237293243408203,
|
| 4166 |
+
"learning_rate": 0.00013349659895204067,
|
| 4167 |
+
"loss": 2.6273,
|
| 4168 |
+
"step": 593
|
| 4169 |
+
},
|
| 4170 |
+
{
|
| 4171 |
+
"epoch": 0.03881829655316073,
|
| 4172 |
+
"grad_norm": 3.293497323989868,
|
| 4173 |
+
"learning_rate": 0.0001332978599768272,
|
| 4174 |
+
"loss": 2.5571,
|
| 4175 |
+
"step": 594
|
| 4176 |
+
},
|
| 4177 |
+
{
|
| 4178 |
+
"epoch": 0.03888364722075864,
|
| 4179 |
+
"grad_norm": 4.987462997436523,
|
| 4180 |
+
"learning_rate": 0.00013309897297378455,
|
| 4181 |
+
"loss": 3.0358,
|
| 4182 |
+
"step": 595
|
| 4183 |
+
},
|
| 4184 |
+
{
|
| 4185 |
+
"epoch": 0.038948997888356554,
|
| 4186 |
+
"grad_norm": 4.07577657699585,
|
| 4187 |
+
"learning_rate": 0.00013289993882707797,
|
| 4188 |
+
"loss": 2.6679,
|
| 4189 |
+
"step": 596
|
| 4190 |
+
},
|
| 4191 |
+
{
|
| 4192 |
+
"epoch": 0.03901434855595447,
|
| 4193 |
+
"grad_norm": 4.8090901374816895,
|
| 4194 |
+
"learning_rate": 0.00013270075842152678,
|
| 4195 |
+
"loss": 2.8967,
|
| 4196 |
+
"step": 597
|
| 4197 |
+
},
|
| 4198 |
+
{
|
| 4199 |
+
"epoch": 0.03907969922355238,
|
| 4200 |
+
"grad_norm": 6.000687599182129,
|
| 4201 |
+
"learning_rate": 0.00013250143264260074,
|
| 4202 |
+
"loss": 2.8839,
|
| 4203 |
+
"step": 598
|
| 4204 |
+
},
|
| 4205 |
+
{
|
| 4206 |
+
"epoch": 0.039145049891150294,
|
| 4207 |
+
"grad_norm": 9.676702499389648,
|
| 4208 |
+
"learning_rate": 0.0001323019623764156,
|
| 4209 |
+
"loss": 3.0498,
|
| 4210 |
+
"step": 599
|
| 4211 |
+
},
|
| 4212 |
+
{
|
| 4213 |
+
"epoch": 0.03921040055874821,
|
| 4214 |
+
"grad_norm": 15.539387702941895,
|
| 4215 |
+
"learning_rate": 0.00013210234850972964,
|
| 4216 |
+
"loss": 2.8019,
|
| 4217 |
+
"step": 600
|
| 4218 |
+
},
|
| 4219 |
+
{
|
| 4220 |
+
"epoch": 0.03927575122634612,
|
| 4221 |
+
"grad_norm": 32.17361068725586,
|
| 4222 |
+
"learning_rate": 0.0001319025919299394,
|
| 4223 |
+
"loss": 3.9383,
|
| 4224 |
+
"step": 601
|
| 4225 |
+
},
|
| 4226 |
+
{
|
| 4227 |
+
"epoch": 0.03934110189394403,
|
| 4228 |
+
"grad_norm": 17.129863739013672,
|
| 4229 |
+
"learning_rate": 0.00013170269352507597,
|
| 4230 |
+
"loss": 3.4502,
|
| 4231 |
+
"step": 602
|
| 4232 |
+
},
|
| 4233 |
+
{
|
| 4234 |
+
"epoch": 0.039406452561541946,
|
| 4235 |
+
"grad_norm": 5.781350135803223,
|
| 4236 |
+
"learning_rate": 0.0001315026541838008,
|
| 4237 |
+
"loss": 3.1775,
|
| 4238 |
+
"step": 603
|
| 4239 |
+
},
|
| 4240 |
+
{
|
| 4241 |
+
"epoch": 0.03947180322913986,
|
| 4242 |
+
"grad_norm": 20.643762588500977,
|
| 4243 |
+
"learning_rate": 0.00013130247479540202,
|
| 4244 |
+
"loss": 3.394,
|
| 4245 |
+
"step": 604
|
| 4246 |
+
},
|
| 4247 |
+
{
|
| 4248 |
+
"epoch": 0.03953715389673778,
|
| 4249 |
+
"grad_norm": 15.9223051071167,
|
| 4250 |
+
"learning_rate": 0.00013110215624979025,
|
| 4251 |
+
"loss": 3.3603,
|
| 4252 |
+
"step": 605
|
| 4253 |
+
},
|
| 4254 |
+
{
|
| 4255 |
+
"epoch": 0.03960250456433569,
|
| 4256 |
+
"grad_norm": 3.7204947471618652,
|
| 4257 |
+
"learning_rate": 0.00013090169943749476,
|
| 4258 |
+
"loss": 3.1395,
|
| 4259 |
+
"step": 606
|
| 4260 |
+
},
|
| 4261 |
+
{
|
| 4262 |
+
"epoch": 0.039667855231933606,
|
| 4263 |
+
"grad_norm": 2.6271538734436035,
|
| 4264 |
+
"learning_rate": 0.00013070110524965954,
|
| 4265 |
+
"loss": 3.0419,
|
| 4266 |
+
"step": 607
|
| 4267 |
+
},
|
| 4268 |
+
{
|
| 4269 |
+
"epoch": 0.03973320589953152,
|
| 4270 |
+
"grad_norm": 2.098278284072876,
|
| 4271 |
+
"learning_rate": 0.00013050037457803924,
|
| 4272 |
+
"loss": 3.0074,
|
| 4273 |
+
"step": 608
|
| 4274 |
+
},
|
| 4275 |
+
{
|
| 4276 |
+
"epoch": 0.03979855656712943,
|
| 4277 |
+
"grad_norm": 2.050483226776123,
|
| 4278 |
+
"learning_rate": 0.0001302995083149953,
|
| 4279 |
+
"loss": 3.1576,
|
| 4280 |
+
"step": 609
|
| 4281 |
+
},
|
| 4282 |
+
{
|
| 4283 |
+
"epoch": 0.039863907234727346,
|
| 4284 |
+
"grad_norm": 3.23539662361145,
|
| 4285 |
+
"learning_rate": 0.0001300985073534919,
|
| 4286 |
+
"loss": 3.1858,
|
| 4287 |
+
"step": 610
|
| 4288 |
+
},
|
| 4289 |
+
{
|
| 4290 |
+
"epoch": 0.03992925790232526,
|
| 4291 |
+
"grad_norm": 2.2334954738616943,
|
| 4292 |
+
"learning_rate": 0.00012989737258709203,
|
| 4293 |
+
"loss": 3.1913,
|
| 4294 |
+
"step": 611
|
| 4295 |
+
},
|
| 4296 |
+
{
|
| 4297 |
+
"epoch": 0.03999460856992317,
|
| 4298 |
+
"grad_norm": 1.9432390928268433,
|
| 4299 |
+
"learning_rate": 0.00012969610490995358,
|
| 4300 |
+
"loss": 3.0337,
|
| 4301 |
+
"step": 612
|
| 4302 |
+
},
|
| 4303 |
+
{
|
| 4304 |
+
"epoch": 0.040059959237521085,
|
| 4305 |
+
"grad_norm": 1.6395848989486694,
|
| 4306 |
+
"learning_rate": 0.00012949470521682528,
|
| 4307 |
+
"loss": 2.802,
|
| 4308 |
+
"step": 613
|
| 4309 |
+
},
|
| 4310 |
+
{
|
| 4311 |
+
"epoch": 0.040125309905119,
|
| 4312 |
+
"grad_norm": 1.7931872606277466,
|
| 4313 |
+
"learning_rate": 0.0001292931744030427,
|
| 4314 |
+
"loss": 3.0174,
|
| 4315 |
+
"step": 614
|
| 4316 |
+
},
|
| 4317 |
+
{
|
| 4318 |
+
"epoch": 0.04019066057271691,
|
| 4319 |
+
"grad_norm": 1.8623560667037964,
|
| 4320 |
+
"learning_rate": 0.0001290915133645243,
|
| 4321 |
+
"loss": 3.0511,
|
| 4322 |
+
"step": 615
|
| 4323 |
+
},
|
| 4324 |
+
{
|
| 4325 |
+
"epoch": 0.040256011240314825,
|
| 4326 |
+
"grad_norm": 1.8954159021377563,
|
| 4327 |
+
"learning_rate": 0.00012888972299776754,
|
| 4328 |
+
"loss": 3.0324,
|
| 4329 |
+
"step": 616
|
| 4330 |
+
},
|
| 4331 |
+
{
|
| 4332 |
+
"epoch": 0.04032136190791274,
|
| 4333 |
+
"grad_norm": 1.8624866008758545,
|
| 4334 |
+
"learning_rate": 0.00012868780419984482,
|
| 4335 |
+
"loss": 2.8377,
|
| 4336 |
+
"step": 617
|
| 4337 |
+
},
|
| 4338 |
+
{
|
| 4339 |
+
"epoch": 0.04038671257551065,
|
| 4340 |
+
"grad_norm": 1.9576878547668457,
|
| 4341 |
+
"learning_rate": 0.00012848575786839943,
|
| 4342 |
+
"loss": 3.0353,
|
| 4343 |
+
"step": 618
|
| 4344 |
+
},
|
| 4345 |
+
{
|
| 4346 |
+
"epoch": 0.040452063243108564,
|
| 4347 |
+
"grad_norm": 2.503295660018921,
|
| 4348 |
+
"learning_rate": 0.0001282835849016416,
|
| 4349 |
+
"loss": 3.1967,
|
| 4350 |
+
"step": 619
|
| 4351 |
+
},
|
| 4352 |
+
{
|
| 4353 |
+
"epoch": 0.040517413910706485,
|
| 4354 |
+
"grad_norm": 2.1141245365142822,
|
| 4355 |
+
"learning_rate": 0.00012808128619834461,
|
| 4356 |
+
"loss": 3.1058,
|
| 4357 |
+
"step": 620
|
| 4358 |
+
},
|
| 4359 |
+
{
|
| 4360 |
+
"epoch": 0.0405827645783044,
|
| 4361 |
+
"grad_norm": 1.9845447540283203,
|
| 4362 |
+
"learning_rate": 0.0001278788626578407,
|
| 4363 |
+
"loss": 2.9753,
|
| 4364 |
+
"step": 621
|
| 4365 |
+
},
|
| 4366 |
+
{
|
| 4367 |
+
"epoch": 0.04064811524590231,
|
| 4368 |
+
"grad_norm": 2.3292746543884277,
|
| 4369 |
+
"learning_rate": 0.00012767631518001698,
|
| 4370 |
+
"loss": 3.3823,
|
| 4371 |
+
"step": 622
|
| 4372 |
+
},
|
| 4373 |
+
{
|
| 4374 |
+
"epoch": 0.040713465913500224,
|
| 4375 |
+
"grad_norm": 1.9060697555541992,
|
| 4376 |
+
"learning_rate": 0.00012747364466531163,
|
| 4377 |
+
"loss": 2.8823,
|
| 4378 |
+
"step": 623
|
| 4379 |
+
},
|
| 4380 |
+
{
|
| 4381 |
+
"epoch": 0.04077881658109814,
|
| 4382 |
+
"grad_norm": 2.004152774810791,
|
| 4383 |
+
"learning_rate": 0.00012727085201470973,
|
| 4384 |
+
"loss": 3.0116,
|
| 4385 |
+
"step": 624
|
| 4386 |
+
},
|
| 4387 |
+
{
|
| 4388 |
+
"epoch": 0.04084416724869605,
|
| 4389 |
+
"grad_norm": 2.2095632553100586,
|
| 4390 |
+
"learning_rate": 0.00012706793812973941,
|
| 4391 |
+
"loss": 3.1009,
|
| 4392 |
+
"step": 625
|
| 4393 |
+
},
|
| 4394 |
+
{
|
| 4395 |
+
"epoch": 0.040909517916293964,
|
| 4396 |
+
"grad_norm": 2.1803863048553467,
|
| 4397 |
+
"learning_rate": 0.0001268649039124677,
|
| 4398 |
+
"loss": 2.8572,
|
| 4399 |
+
"step": 626
|
| 4400 |
+
},
|
| 4401 |
+
{
|
| 4402 |
+
"epoch": 0.04097486858389188,
|
| 4403 |
+
"grad_norm": 2.2665598392486572,
|
| 4404 |
+
"learning_rate": 0.00012666175026549662,
|
| 4405 |
+
"loss": 3.0518,
|
| 4406 |
+
"step": 627
|
| 4407 |
+
},
|
| 4408 |
+
{
|
| 4409 |
+
"epoch": 0.04104021925148979,
|
| 4410 |
+
"grad_norm": 2.5089712142944336,
|
| 4411 |
+
"learning_rate": 0.000126458478091959,
|
| 4412 |
+
"loss": 3.1027,
|
| 4413 |
+
"step": 628
|
| 4414 |
+
},
|
| 4415 |
+
{
|
| 4416 |
+
"epoch": 0.0411055699190877,
|
| 4417 |
+
"grad_norm": 2.315622329711914,
|
| 4418 |
+
"learning_rate": 0.00012625508829551473,
|
| 4419 |
+
"loss": 3.0324,
|
| 4420 |
+
"step": 629
|
| 4421 |
+
},
|
| 4422 |
+
{
|
| 4423 |
+
"epoch": 0.041170920586685616,
|
| 4424 |
+
"grad_norm": 2.4968509674072266,
|
| 4425 |
+
"learning_rate": 0.00012605158178034654,
|
| 4426 |
+
"loss": 2.9036,
|
| 4427 |
+
"step": 630
|
| 4428 |
+
},
|
| 4429 |
+
{
|
| 4430 |
+
"epoch": 0.04123627125428353,
|
| 4431 |
+
"grad_norm": 2.250495433807373,
|
| 4432 |
+
"learning_rate": 0.00012584795945115603,
|
| 4433 |
+
"loss": 2.816,
|
| 4434 |
+
"step": 631
|
| 4435 |
+
},
|
| 4436 |
+
{
|
| 4437 |
+
"epoch": 0.04130162192188144,
|
| 4438 |
+
"grad_norm": 2.2009661197662354,
|
| 4439 |
+
"learning_rate": 0.0001256442222131597,
|
| 4440 |
+
"loss": 2.8616,
|
| 4441 |
+
"step": 632
|
| 4442 |
+
},
|
| 4443 |
+
{
|
| 4444 |
+
"epoch": 0.041366972589479356,
|
| 4445 |
+
"grad_norm": 2.7355098724365234,
|
| 4446 |
+
"learning_rate": 0.0001254403709720848,
|
| 4447 |
+
"loss": 3.047,
|
| 4448 |
+
"step": 633
|
| 4449 |
+
},
|
| 4450 |
+
{
|
| 4451 |
+
"epoch": 0.041432323257077276,
|
| 4452 |
+
"grad_norm": 3.254558563232422,
|
| 4453 |
+
"learning_rate": 0.0001252364066341655,
|
| 4454 |
+
"loss": 3.284,
|
| 4455 |
+
"step": 634
|
| 4456 |
+
},
|
| 4457 |
+
{
|
| 4458 |
+
"epoch": 0.04149767392467519,
|
| 4459 |
+
"grad_norm": 3.5515785217285156,
|
| 4460 |
+
"learning_rate": 0.00012503233010613865,
|
| 4461 |
+
"loss": 3.206,
|
| 4462 |
+
"step": 635
|
| 4463 |
+
},
|
| 4464 |
+
{
|
| 4465 |
+
"epoch": 0.0415630245922731,
|
| 4466 |
+
"grad_norm": 2.704427480697632,
|
| 4467 |
+
"learning_rate": 0.00012482814229523997,
|
| 4468 |
+
"loss": 3.2234,
|
| 4469 |
+
"step": 636
|
| 4470 |
+
},
|
| 4471 |
+
{
|
| 4472 |
+
"epoch": 0.041628375259871016,
|
| 4473 |
+
"grad_norm": 3.272678852081299,
|
| 4474 |
+
"learning_rate": 0.00012462384410919975,
|
| 4475 |
+
"loss": 3.1285,
|
| 4476 |
+
"step": 637
|
| 4477 |
+
},
|
| 4478 |
+
{
|
| 4479 |
+
"epoch": 0.04169372592746893,
|
| 4480 |
+
"grad_norm": 3.133451461791992,
|
| 4481 |
+
"learning_rate": 0.00012441943645623903,
|
| 4482 |
+
"loss": 2.9501,
|
| 4483 |
+
"step": 638
|
| 4484 |
+
},
|
| 4485 |
+
{
|
| 4486 |
+
"epoch": 0.04175907659506684,
|
| 4487 |
+
"grad_norm": 3.1467037200927734,
|
| 4488 |
+
"learning_rate": 0.00012421492024506555,
|
| 4489 |
+
"loss": 3.1572,
|
| 4490 |
+
"step": 639
|
| 4491 |
+
},
|
| 4492 |
+
{
|
| 4493 |
+
"epoch": 0.041824427262664755,
|
| 4494 |
+
"grad_norm": 3.4635956287384033,
|
| 4495 |
+
"learning_rate": 0.00012401029638486953,
|
| 4496 |
+
"loss": 2.9697,
|
| 4497 |
+
"step": 640
|
| 4498 |
+
},
|
| 4499 |
+
{
|
| 4500 |
+
"epoch": 0.04188977793026267,
|
| 4501 |
+
"grad_norm": 4.2859320640563965,
|
| 4502 |
+
"learning_rate": 0.0001238055657853198,
|
| 4503 |
+
"loss": 3.2693,
|
| 4504 |
+
"step": 641
|
| 4505 |
+
},
|
| 4506 |
+
{
|
| 4507 |
+
"epoch": 0.04195512859786058,
|
| 4508 |
+
"grad_norm": 3.362356662750244,
|
| 4509 |
+
"learning_rate": 0.00012360072935655982,
|
| 4510 |
+
"loss": 2.9823,
|
| 4511 |
+
"step": 642
|
| 4512 |
+
},
|
| 4513 |
+
{
|
| 4514 |
+
"epoch": 0.042020479265458495,
|
| 4515 |
+
"grad_norm": 4.812988758087158,
|
| 4516 |
+
"learning_rate": 0.00012339578800920332,
|
| 4517 |
+
"loss": 2.679,
|
| 4518 |
+
"step": 643
|
| 4519 |
+
},
|
| 4520 |
+
{
|
| 4521 |
+
"epoch": 0.04208582993305641,
|
| 4522 |
+
"grad_norm": 3.6487369537353516,
|
| 4523 |
+
"learning_rate": 0.00012319074265433063,
|
| 4524 |
+
"loss": 2.6908,
|
| 4525 |
+
"step": 644
|
| 4526 |
+
},
|
| 4527 |
+
{
|
| 4528 |
+
"epoch": 0.04215118060065432,
|
| 4529 |
+
"grad_norm": 4.461695194244385,
|
| 4530 |
+
"learning_rate": 0.00012298559420348437,
|
| 4531 |
+
"loss": 3.1041,
|
| 4532 |
+
"step": 645
|
| 4533 |
+
},
|
| 4534 |
+
{
|
| 4535 |
+
"epoch": 0.042216531268252235,
|
| 4536 |
+
"grad_norm": 3.465256929397583,
|
| 4537 |
+
"learning_rate": 0.00012278034356866545,
|
| 4538 |
+
"loss": 2.6494,
|
| 4539 |
+
"step": 646
|
| 4540 |
+
},
|
| 4541 |
+
{
|
| 4542 |
+
"epoch": 0.04228188193585015,
|
| 4543 |
+
"grad_norm": 5.122613906860352,
|
| 4544 |
+
"learning_rate": 0.00012257499166232907,
|
| 4545 |
+
"loss": 3.0645,
|
| 4546 |
+
"step": 647
|
| 4547 |
+
},
|
| 4548 |
+
{
|
| 4549 |
+
"epoch": 0.04234723260344807,
|
| 4550 |
+
"grad_norm": 4.229803085327148,
|
| 4551 |
+
"learning_rate": 0.0001223695393973807,
|
| 4552 |
+
"loss": 2.926,
|
| 4553 |
+
"step": 648
|
| 4554 |
+
},
|
| 4555 |
+
{
|
| 4556 |
+
"epoch": 0.04241258327104598,
|
| 4557 |
+
"grad_norm": 5.684296607971191,
|
| 4558 |
+
"learning_rate": 0.0001221639876871719,
|
| 4559 |
+
"loss": 3.0356,
|
| 4560 |
+
"step": 649
|
| 4561 |
+
},
|
| 4562 |
+
{
|
| 4563 |
+
"epoch": 0.042477933938643894,
|
| 4564 |
+
"grad_norm": 5.228229999542236,
|
| 4565 |
+
"learning_rate": 0.0001219583374454963,
|
| 4566 |
+
"loss": 2.5974,
|
| 4567 |
+
"step": 650
|
| 4568 |
+
},
|
| 4569 |
+
{
|
| 4570 |
+
"epoch": 0.04254328460624181,
|
| 4571 |
+
"grad_norm": 16.97333526611328,
|
| 4572 |
+
"learning_rate": 0.00012175258958658564,
|
| 4573 |
+
"loss": 3.5355,
|
| 4574 |
+
"step": 651
|
| 4575 |
+
},
|
| 4576 |
+
{
|
| 4577 |
+
"epoch": 0.04260863527383972,
|
| 4578 |
+
"grad_norm": 6.152676105499268,
|
| 4579 |
+
"learning_rate": 0.00012154674502510555,
|
| 4580 |
+
"loss": 3.3318,
|
| 4581 |
+
"step": 652
|
| 4582 |
+
},
|
| 4583 |
+
{
|
| 4584 |
+
"epoch": 0.042673985941437634,
|
| 4585 |
+
"grad_norm": 12.877490043640137,
|
| 4586 |
+
"learning_rate": 0.00012134080467615159,
|
| 4587 |
+
"loss": 3.3578,
|
| 4588 |
+
"step": 653
|
| 4589 |
+
},
|
| 4590 |
+
{
|
| 4591 |
+
"epoch": 0.04273933660903555,
|
| 4592 |
+
"grad_norm": 15.447893142700195,
|
| 4593 |
+
"learning_rate": 0.00012113476945524513,
|
| 4594 |
+
"loss": 3.5556,
|
| 4595 |
+
"step": 654
|
| 4596 |
+
},
|
| 4597 |
+
{
|
| 4598 |
+
"epoch": 0.04280468727663346,
|
| 4599 |
+
"grad_norm": 7.900871276855469,
|
| 4600 |
+
"learning_rate": 0.00012092864027832933,
|
| 4601 |
+
"loss": 3.2906,
|
| 4602 |
+
"step": 655
|
| 4603 |
+
},
|
| 4604 |
+
{
|
| 4605 |
+
"epoch": 0.04287003794423137,
|
| 4606 |
+
"grad_norm": 3.3897945880889893,
|
| 4607 |
+
"learning_rate": 0.000120722418061765,
|
| 4608 |
+
"loss": 2.9694,
|
| 4609 |
+
"step": 656
|
| 4610 |
+
},
|
| 4611 |
+
{
|
| 4612 |
+
"epoch": 0.04293538861182929,
|
| 4613 |
+
"grad_norm": 3.3307113647460938,
|
| 4614 |
+
"learning_rate": 0.0001205161037223266,
|
| 4615 |
+
"loss": 3.2379,
|
| 4616 |
+
"step": 657
|
| 4617 |
+
},
|
| 4618 |
+
{
|
| 4619 |
+
"epoch": 0.0430007392794272,
|
| 4620 |
+
"grad_norm": 2.4997718334198,
|
| 4621 |
+
"learning_rate": 0.00012030969817719808,
|
| 4622 |
+
"loss": 3.101,
|
| 4623 |
+
"step": 658
|
| 4624 |
+
},
|
| 4625 |
+
{
|
| 4626 |
+
"epoch": 0.04306608994702511,
|
| 4627 |
+
"grad_norm": 2.665391445159912,
|
| 4628 |
+
"learning_rate": 0.00012010320234396894,
|
| 4629 |
+
"loss": 3.2717,
|
| 4630 |
+
"step": 659
|
| 4631 |
+
},
|
| 4632 |
+
{
|
| 4633 |
+
"epoch": 0.043131440614623026,
|
| 4634 |
+
"grad_norm": 3.035580635070801,
|
| 4635 |
+
"learning_rate": 0.00011989661714062999,
|
| 4636 |
+
"loss": 3.1335,
|
| 4637 |
+
"step": 660
|
| 4638 |
+
},
|
| 4639 |
+
{
|
| 4640 |
+
"epoch": 0.04319679128222094,
|
| 4641 |
+
"grad_norm": 1.964550495147705,
|
| 4642 |
+
"learning_rate": 0.0001196899434855693,
|
| 4643 |
+
"loss": 2.9821,
|
| 4644 |
+
"step": 661
|
| 4645 |
+
},
|
| 4646 |
+
{
|
| 4647 |
+
"epoch": 0.04326214194981886,
|
| 4648 |
+
"grad_norm": 2.2508432865142822,
|
| 4649 |
+
"learning_rate": 0.00011948318229756827,
|
| 4650 |
+
"loss": 2.8584,
|
| 4651 |
+
"step": 662
|
| 4652 |
+
},
|
| 4653 |
+
{
|
| 4654 |
+
"epoch": 0.04332749261741677,
|
| 4655 |
+
"grad_norm": 1.820594310760498,
|
| 4656 |
+
"learning_rate": 0.00011927633449579735,
|
| 4657 |
+
"loss": 2.8469,
|
| 4658 |
+
"step": 663
|
| 4659 |
+
},
|
| 4660 |
+
{
|
| 4661 |
+
"epoch": 0.043392843285014686,
|
| 4662 |
+
"grad_norm": 2.0927326679229736,
|
| 4663 |
+
"learning_rate": 0.0001190694009998121,
|
| 4664 |
+
"loss": 2.9665,
|
| 4665 |
+
"step": 664
|
| 4666 |
+
},
|
| 4667 |
+
{
|
| 4668 |
+
"epoch": 0.0434581939526126,
|
| 4669 |
+
"grad_norm": 1.9903268814086914,
|
| 4670 |
+
"learning_rate": 0.00011886238272954897,
|
| 4671 |
+
"loss": 3.0826,
|
| 4672 |
+
"step": 665
|
| 4673 |
+
},
|
| 4674 |
+
{
|
| 4675 |
+
"epoch": 0.04352354462021051,
|
| 4676 |
+
"grad_norm": 2.0736825466156006,
|
| 4677 |
+
"learning_rate": 0.00011865528060532127,
|
| 4678 |
+
"loss": 3.2923,
|
| 4679 |
+
"step": 666
|
| 4680 |
+
},
|
| 4681 |
+
{
|
| 4682 |
+
"epoch": 0.043588895287808425,
|
| 4683 |
+
"grad_norm": 1.9887921810150146,
|
| 4684 |
+
"learning_rate": 0.0001184480955478152,
|
| 4685 |
+
"loss": 3.1273,
|
| 4686 |
+
"step": 667
|
| 4687 |
+
},
|
| 4688 |
+
{
|
| 4689 |
+
"epoch": 0.04365424595540634,
|
| 4690 |
+
"grad_norm": 1.898668646812439,
|
| 4691 |
+
"learning_rate": 0.00011824082847808558,
|
| 4692 |
+
"loss": 3.0536,
|
| 4693 |
+
"step": 668
|
| 4694 |
+
},
|
| 4695 |
+
{
|
| 4696 |
+
"epoch": 0.04371959662300425,
|
| 4697 |
+
"grad_norm": 2.2333860397338867,
|
| 4698 |
+
"learning_rate": 0.00011803348031755179,
|
| 4699 |
+
"loss": 2.7913,
|
| 4700 |
+
"step": 669
|
| 4701 |
+
},
|
| 4702 |
+
{
|
| 4703 |
+
"epoch": 0.043784947290602165,
|
| 4704 |
+
"grad_norm": 2.0350661277770996,
|
| 4705 |
+
"learning_rate": 0.0001178260519879937,
|
| 4706 |
+
"loss": 2.8106,
|
| 4707 |
+
"step": 670
|
| 4708 |
+
},
|
| 4709 |
+
{
|
| 4710 |
+
"epoch": 0.04385029795820008,
|
| 4711 |
+
"grad_norm": 2.0229201316833496,
|
| 4712 |
+
"learning_rate": 0.00011761854441154767,
|
| 4713 |
+
"loss": 2.9744,
|
| 4714 |
+
"step": 671
|
| 4715 |
+
},
|
| 4716 |
+
{
|
| 4717 |
+
"epoch": 0.04391564862579799,
|
| 4718 |
+
"grad_norm": 2.2588329315185547,
|
| 4719 |
+
"learning_rate": 0.00011741095851070228,
|
| 4720 |
+
"loss": 3.1313,
|
| 4721 |
+
"step": 672
|
| 4722 |
+
},
|
| 4723 |
+
{
|
| 4724 |
+
"epoch": 0.043980999293395905,
|
| 4725 |
+
"grad_norm": 2.1912214756011963,
|
| 4726 |
+
"learning_rate": 0.00011720329520829429,
|
| 4727 |
+
"loss": 2.7155,
|
| 4728 |
+
"step": 673
|
| 4729 |
+
},
|
| 4730 |
+
{
|
| 4731 |
+
"epoch": 0.04404634996099382,
|
| 4732 |
+
"grad_norm": 2.086419105529785,
|
| 4733 |
+
"learning_rate": 0.0001169955554275046,
|
| 4734 |
+
"loss": 2.9657,
|
| 4735 |
+
"step": 674
|
| 4736 |
+
},
|
| 4737 |
+
{
|
| 4738 |
+
"epoch": 0.04411170062859173,
|
| 4739 |
+
"grad_norm": 2.2901723384857178,
|
| 4740 |
+
"learning_rate": 0.0001167877400918541,
|
| 4741 |
+
"loss": 3.1845,
|
| 4742 |
+
"step": 675
|
| 4743 |
+
},
|
| 4744 |
+
{
|
| 4745 |
+
"epoch": 0.04417705129618965,
|
| 4746 |
+
"grad_norm": 2.07425594329834,
|
| 4747 |
+
"learning_rate": 0.00011657985012519952,
|
| 4748 |
+
"loss": 2.9775,
|
| 4749 |
+
"step": 676
|
| 4750 |
+
},
|
| 4751 |
+
{
|
| 4752 |
+
"epoch": 0.044242401963787564,
|
| 4753 |
+
"grad_norm": 2.1768293380737305,
|
| 4754 |
+
"learning_rate": 0.00011637188645172944,
|
| 4755 |
+
"loss": 3.2035,
|
| 4756 |
+
"step": 677
|
| 4757 |
+
},
|
| 4758 |
+
{
|
| 4759 |
+
"epoch": 0.04430775263138548,
|
| 4760 |
+
"grad_norm": 2.202507734298706,
|
| 4761 |
+
"learning_rate": 0.00011616384999596006,
|
| 4762 |
+
"loss": 3.0454,
|
| 4763 |
+
"step": 678
|
| 4764 |
+
},
|
| 4765 |
+
{
|
| 4766 |
+
"epoch": 0.04437310329898339,
|
| 4767 |
+
"grad_norm": 2.209707736968994,
|
| 4768 |
+
"learning_rate": 0.00011595574168273111,
|
| 4769 |
+
"loss": 3.1953,
|
| 4770 |
+
"step": 679
|
| 4771 |
+
},
|
| 4772 |
+
{
|
| 4773 |
+
"epoch": 0.044438453966581304,
|
| 4774 |
+
"grad_norm": 2.144770383834839,
|
| 4775 |
+
"learning_rate": 0.0001157475624372018,
|
| 4776 |
+
"loss": 2.9541,
|
| 4777 |
+
"step": 680
|
| 4778 |
+
},
|
| 4779 |
+
{
|
| 4780 |
+
"epoch": 0.04450380463417922,
|
| 4781 |
+
"grad_norm": 2.4799458980560303,
|
| 4782 |
+
"learning_rate": 0.0001155393131848467,
|
| 4783 |
+
"loss": 3.2559,
|
| 4784 |
+
"step": 681
|
| 4785 |
+
},
|
| 4786 |
+
{
|
| 4787 |
+
"epoch": 0.04456915530177713,
|
| 4788 |
+
"grad_norm": 2.293464183807373,
|
| 4789 |
+
"learning_rate": 0.00011533099485145155,
|
| 4790 |
+
"loss": 3.1855,
|
| 4791 |
+
"step": 682
|
| 4792 |
+
},
|
| 4793 |
+
{
|
| 4794 |
+
"epoch": 0.04463450596937504,
|
| 4795 |
+
"grad_norm": 2.608764886856079,
|
| 4796 |
+
"learning_rate": 0.00011512260836310924,
|
| 4797 |
+
"loss": 2.9465,
|
| 4798 |
+
"step": 683
|
| 4799 |
+
},
|
| 4800 |
+
{
|
| 4801 |
+
"epoch": 0.04469985663697296,
|
| 4802 |
+
"grad_norm": 2.3550198078155518,
|
| 4803 |
+
"learning_rate": 0.00011491415464621562,
|
| 4804 |
+
"loss": 2.914,
|
| 4805 |
+
"step": 684
|
| 4806 |
+
},
|
| 4807 |
+
{
|
| 4808 |
+
"epoch": 0.04476520730457087,
|
| 4809 |
+
"grad_norm": 2.5664079189300537,
|
| 4810 |
+
"learning_rate": 0.00011470563462746541,
|
| 4811 |
+
"loss": 2.8957,
|
| 4812 |
+
"step": 685
|
| 4813 |
+
},
|
| 4814 |
+
{
|
| 4815 |
+
"epoch": 0.04483055797216878,
|
| 4816 |
+
"grad_norm": 2.570154905319214,
|
| 4817 |
+
"learning_rate": 0.00011449704923384812,
|
| 4818 |
+
"loss": 3.1221,
|
| 4819 |
+
"step": 686
|
| 4820 |
+
},
|
| 4821 |
+
{
|
| 4822 |
+
"epoch": 0.044895908639766696,
|
| 4823 |
+
"grad_norm": 6.49520206451416,
|
| 4824 |
+
"learning_rate": 0.00011428839939264382,
|
| 4825 |
+
"loss": 3.0106,
|
| 4826 |
+
"step": 687
|
| 4827 |
+
},
|
| 4828 |
+
{
|
| 4829 |
+
"epoch": 0.04496125930736461,
|
| 4830 |
+
"grad_norm": 3.2794463634490967,
|
| 4831 |
+
"learning_rate": 0.0001140796860314191,
|
| 4832 |
+
"loss": 3.0885,
|
| 4833 |
+
"step": 688
|
| 4834 |
+
},
|
| 4835 |
+
{
|
| 4836 |
+
"epoch": 0.04502660997496252,
|
| 4837 |
+
"grad_norm": 4.257885932922363,
|
| 4838 |
+
"learning_rate": 0.00011387091007802297,
|
| 4839 |
+
"loss": 3.0201,
|
| 4840 |
+
"step": 689
|
| 4841 |
+
},
|
| 4842 |
+
{
|
| 4843 |
+
"epoch": 0.045091960642560436,
|
| 4844 |
+
"grad_norm": 4.146145820617676,
|
| 4845 |
+
"learning_rate": 0.0001136620724605827,
|
| 4846 |
+
"loss": 3.2533,
|
| 4847 |
+
"step": 690
|
| 4848 |
+
},
|
| 4849 |
+
{
|
| 4850 |
+
"epoch": 0.045157311310158356,
|
| 4851 |
+
"grad_norm": 3.103212356567383,
|
| 4852 |
+
"learning_rate": 0.00011345317410749964,
|
| 4853 |
+
"loss": 2.7706,
|
| 4854 |
+
"step": 691
|
| 4855 |
+
},
|
| 4856 |
+
{
|
| 4857 |
+
"epoch": 0.04522266197775627,
|
| 4858 |
+
"grad_norm": 3.2333109378814697,
|
| 4859 |
+
"learning_rate": 0.00011324421594744516,
|
| 4860 |
+
"loss": 2.8265,
|
| 4861 |
+
"step": 692
|
| 4862 |
+
},
|
| 4863 |
+
{
|
| 4864 |
+
"epoch": 0.04528801264535418,
|
| 4865 |
+
"grad_norm": 5.440866947174072,
|
| 4866 |
+
"learning_rate": 0.00011303519890935656,
|
| 4867 |
+
"loss": 2.9213,
|
| 4868 |
+
"step": 693
|
| 4869 |
+
},
|
| 4870 |
+
{
|
| 4871 |
+
"epoch": 0.045353363312952095,
|
| 4872 |
+
"grad_norm": 4.879655838012695,
|
| 4873 |
+
"learning_rate": 0.00011282612392243286,
|
| 4874 |
+
"loss": 2.8601,
|
| 4875 |
+
"step": 694
|
| 4876 |
+
},
|
| 4877 |
+
{
|
| 4878 |
+
"epoch": 0.04541871398055001,
|
| 4879 |
+
"grad_norm": 3.385190486907959,
|
| 4880 |
+
"learning_rate": 0.00011261699191613066,
|
| 4881 |
+
"loss": 2.8724,
|
| 4882 |
+
"step": 695
|
| 4883 |
+
},
|
| 4884 |
+
{
|
| 4885 |
+
"epoch": 0.04548406464814792,
|
| 4886 |
+
"grad_norm": 5.638006687164307,
|
| 4887 |
+
"learning_rate": 0.00011240780382016005,
|
| 4888 |
+
"loss": 2.6965,
|
| 4889 |
+
"step": 696
|
| 4890 |
+
},
|
| 4891 |
+
{
|
| 4892 |
+
"epoch": 0.045549415315745835,
|
| 4893 |
+
"grad_norm": 6.110171318054199,
|
| 4894 |
+
"learning_rate": 0.00011219856056448051,
|
| 4895 |
+
"loss": 3.0768,
|
| 4896 |
+
"step": 697
|
| 4897 |
+
},
|
| 4898 |
+
{
|
| 4899 |
+
"epoch": 0.04561476598334375,
|
| 4900 |
+
"grad_norm": 3.941072463989258,
|
| 4901 |
+
"learning_rate": 0.00011198926307929664,
|
| 4902 |
+
"loss": 2.6718,
|
| 4903 |
+
"step": 698
|
| 4904 |
+
},
|
| 4905 |
+
{
|
| 4906 |
+
"epoch": 0.04568011665094166,
|
| 4907 |
+
"grad_norm": 3.8805644512176514,
|
| 4908 |
+
"learning_rate": 0.00011177991229505431,
|
| 4909 |
+
"loss": 2.6361,
|
| 4910 |
+
"step": 699
|
| 4911 |
+
},
|
| 4912 |
+
{
|
| 4913 |
+
"epoch": 0.045745467318539575,
|
| 4914 |
+
"grad_norm": 5.5727972984313965,
|
| 4915 |
+
"learning_rate": 0.00011157050914243614,
|
| 4916 |
+
"loss": 2.6005,
|
| 4917 |
+
"step": 700
|
| 4918 |
+
},
|
| 4919 |
+
{
|
| 4920 |
+
"epoch": 0.04581081798613749,
|
| 4921 |
+
"grad_norm": 14.964508056640625,
|
| 4922 |
+
"learning_rate": 0.00011136105455235766,
|
| 4923 |
+
"loss": 3.436,
|
| 4924 |
+
"step": 701
|
| 4925 |
+
},
|
| 4926 |
+
{
|
| 4927 |
+
"epoch": 0.0458761686537354,
|
| 4928 |
+
"grad_norm": 8.832298278808594,
|
| 4929 |
+
"learning_rate": 0.00011115154945596305,
|
| 4930 |
+
"loss": 3.2624,
|
| 4931 |
+
"step": 702
|
| 4932 |
+
},
|
| 4933 |
+
{
|
| 4934 |
+
"epoch": 0.045941519321333314,
|
| 4935 |
+
"grad_norm": 8.178778648376465,
|
| 4936 |
+
"learning_rate": 0.00011094199478462095,
|
| 4937 |
+
"loss": 3.0941,
|
| 4938 |
+
"step": 703
|
| 4939 |
+
},
|
| 4940 |
+
{
|
| 4941 |
+
"epoch": 0.04600686998893123,
|
| 4942 |
+
"grad_norm": 6.393551826477051,
|
| 4943 |
+
"learning_rate": 0.00011073239146992054,
|
| 4944 |
+
"loss": 3.2571,
|
| 4945 |
+
"step": 704
|
| 4946 |
+
},
|
| 4947 |
+
{
|
| 4948 |
+
"epoch": 0.04607222065652915,
|
| 4949 |
+
"grad_norm": 3.2871720790863037,
|
| 4950 |
+
"learning_rate": 0.00011052274044366711,
|
| 4951 |
+
"loss": 3.2479,
|
| 4952 |
+
"step": 705
|
| 4953 |
+
},
|
| 4954 |
+
{
|
| 4955 |
+
"epoch": 0.04613757132412706,
|
| 4956 |
+
"grad_norm": 4.299026966094971,
|
| 4957 |
+
"learning_rate": 0.00011031304263787812,
|
| 4958 |
+
"loss": 3.0419,
|
| 4959 |
+
"step": 706
|
| 4960 |
+
},
|
| 4961 |
+
{
|
| 4962 |
+
"epoch": 0.046202921991724974,
|
| 4963 |
+
"grad_norm": 2.4954004287719727,
|
| 4964 |
+
"learning_rate": 0.00011010329898477891,
|
| 4965 |
+
"loss": 3.1365,
|
| 4966 |
+
"step": 707
|
| 4967 |
+
},
|
| 4968 |
+
{
|
| 4969 |
+
"epoch": 0.04626827265932289,
|
| 4970 |
+
"grad_norm": 2.39926815032959,
|
| 4971 |
+
"learning_rate": 0.0001098935104167988,
|
| 4972 |
+
"loss": 3.1031,
|
| 4973 |
+
"step": 708
|
| 4974 |
+
},
|
| 4975 |
+
{
|
| 4976 |
+
"epoch": 0.0463336233269208,
|
| 4977 |
+
"grad_norm": 1.965259313583374,
|
| 4978 |
+
"learning_rate": 0.00010968367786656663,
|
| 4979 |
+
"loss": 3.0447,
|
| 4980 |
+
"step": 709
|
| 4981 |
+
},
|
| 4982 |
+
{
|
| 4983 |
+
"epoch": 0.046398973994518714,
|
| 4984 |
+
"grad_norm": 2.1634223461151123,
|
| 4985 |
+
"learning_rate": 0.00010947380226690684,
|
| 4986 |
+
"loss": 3.0273,
|
| 4987 |
+
"step": 710
|
| 4988 |
+
},
|
| 4989 |
+
{
|
| 4990 |
+
"epoch": 0.04646432466211663,
|
| 4991 |
+
"grad_norm": 1.858903408050537,
|
| 4992 |
+
"learning_rate": 0.00010926388455083522,
|
| 4993 |
+
"loss": 3.0245,
|
| 4994 |
+
"step": 711
|
| 4995 |
+
},
|
| 4996 |
+
{
|
| 4997 |
+
"epoch": 0.04652967532971454,
|
| 4998 |
+
"grad_norm": 2.939605474472046,
|
| 4999 |
+
"learning_rate": 0.00010905392565155477,
|
| 5000 |
+
"loss": 3.1052,
|
| 5001 |
+
"step": 712
|
| 5002 |
+
},
|
| 5003 |
+
{
|
| 5004 |
+
"epoch": 0.04659502599731245,
|
| 5005 |
+
"grad_norm": 1.939249873161316,
|
| 5006 |
+
"learning_rate": 0.00010884392650245165,
|
| 5007 |
+
"loss": 2.9436,
|
| 5008 |
+
"step": 713
|
| 5009 |
+
},
|
| 5010 |
+
{
|
| 5011 |
+
"epoch": 0.046660376664910366,
|
| 5012 |
+
"grad_norm": 1.8927584886550903,
|
| 5013 |
+
"learning_rate": 0.00010863388803709089,
|
| 5014 |
+
"loss": 3.0052,
|
| 5015 |
+
"step": 714
|
| 5016 |
+
},
|
| 5017 |
+
{
|
| 5018 |
+
"epoch": 0.04672572733250828,
|
| 5019 |
+
"grad_norm": 2.38651704788208,
|
| 5020 |
+
"learning_rate": 0.00010842381118921232,
|
| 5021 |
+
"loss": 2.8575,
|
| 5022 |
+
"step": 715
|
| 5023 |
+
},
|
| 5024 |
+
{
|
| 5025 |
+
"epoch": 0.04679107800010619,
|
| 5026 |
+
"grad_norm": 1.9476335048675537,
|
| 5027 |
+
"learning_rate": 0.00010821369689272638,
|
| 5028 |
+
"loss": 2.8868,
|
| 5029 |
+
"step": 716
|
| 5030 |
+
},
|
| 5031 |
+
{
|
| 5032 |
+
"epoch": 0.046856428667704106,
|
| 5033 |
+
"grad_norm": 2.5600156784057617,
|
| 5034 |
+
"learning_rate": 0.00010800354608171003,
|
| 5035 |
+
"loss": 3.0852,
|
| 5036 |
+
"step": 717
|
| 5037 |
+
},
|
| 5038 |
+
{
|
| 5039 |
+
"epoch": 0.04692177933530202,
|
| 5040 |
+
"grad_norm": 1.9691524505615234,
|
| 5041 |
+
"learning_rate": 0.00010779335969040252,
|
| 5042 |
+
"loss": 3.1693,
|
| 5043 |
+
"step": 718
|
| 5044 |
+
},
|
| 5045 |
+
{
|
| 5046 |
+
"epoch": 0.04698713000289994,
|
| 5047 |
+
"grad_norm": 2.0690503120422363,
|
| 5048 |
+
"learning_rate": 0.00010758313865320134,
|
| 5049 |
+
"loss": 3.0951,
|
| 5050 |
+
"step": 719
|
| 5051 |
+
},
|
| 5052 |
+
{
|
| 5053 |
+
"epoch": 0.04705248067049785,
|
| 5054 |
+
"grad_norm": 1.8928483724594116,
|
| 5055 |
+
"learning_rate": 0.00010737288390465792,
|
| 5056 |
+
"loss": 2.949,
|
| 5057 |
+
"step": 720
|
| 5058 |
+
},
|
| 5059 |
+
{
|
| 5060 |
+
"epoch": 0.047117831338095766,
|
| 5061 |
+
"grad_norm": 1.9570032358169556,
|
| 5062 |
+
"learning_rate": 0.00010716259637947357,
|
| 5063 |
+
"loss": 3.0925,
|
| 5064 |
+
"step": 721
|
| 5065 |
+
},
|
| 5066 |
+
{
|
| 5067 |
+
"epoch": 0.04718318200569368,
|
| 5068 |
+
"grad_norm": 2.15844988822937,
|
| 5069 |
+
"learning_rate": 0.00010695227701249537,
|
| 5070 |
+
"loss": 3.1285,
|
| 5071 |
+
"step": 722
|
| 5072 |
+
},
|
| 5073 |
+
{
|
| 5074 |
+
"epoch": 0.04724853267329159,
|
| 5075 |
+
"grad_norm": 2.1201603412628174,
|
| 5076 |
+
"learning_rate": 0.00010674192673871191,
|
| 5077 |
+
"loss": 2.9518,
|
| 5078 |
+
"step": 723
|
| 5079 |
+
},
|
| 5080 |
+
{
|
| 5081 |
+
"epoch": 0.047313883340889505,
|
| 5082 |
+
"grad_norm": 2.0877747535705566,
|
| 5083 |
+
"learning_rate": 0.00010653154649324917,
|
| 5084 |
+
"loss": 2.8219,
|
| 5085 |
+
"step": 724
|
| 5086 |
+
},
|
| 5087 |
+
{
|
| 5088 |
+
"epoch": 0.04737923400848742,
|
| 5089 |
+
"grad_norm": 2.1260170936584473,
|
| 5090 |
+
"learning_rate": 0.00010632113721136636,
|
| 5091 |
+
"loss": 2.989,
|
| 5092 |
+
"step": 725
|
| 5093 |
+
},
|
| 5094 |
+
{
|
| 5095 |
+
"epoch": 0.04744458467608533,
|
| 5096 |
+
"grad_norm": 2.3242621421813965,
|
| 5097 |
+
"learning_rate": 0.00010611069982845183,
|
| 5098 |
+
"loss": 3.1765,
|
| 5099 |
+
"step": 726
|
| 5100 |
+
},
|
| 5101 |
+
{
|
| 5102 |
+
"epoch": 0.047509935343683245,
|
| 5103 |
+
"grad_norm": 2.1452298164367676,
|
| 5104 |
+
"learning_rate": 0.00010590023528001884,
|
| 5105 |
+
"loss": 2.9238,
|
| 5106 |
+
"step": 727
|
| 5107 |
+
},
|
| 5108 |
+
{
|
| 5109 |
+
"epoch": 0.04757528601128116,
|
| 5110 |
+
"grad_norm": 2.3453991413116455,
|
| 5111 |
+
"learning_rate": 0.00010568974450170139,
|
| 5112 |
+
"loss": 2.9934,
|
| 5113 |
+
"step": 728
|
| 5114 |
+
},
|
| 5115 |
+
{
|
| 5116 |
+
"epoch": 0.04764063667887907,
|
| 5117 |
+
"grad_norm": 2.5484986305236816,
|
| 5118 |
+
"learning_rate": 0.00010547922842925008,
|
| 5119 |
+
"loss": 2.8265,
|
| 5120 |
+
"step": 729
|
| 5121 |
+
},
|
| 5122 |
+
{
|
| 5123 |
+
"epoch": 0.047705987346476984,
|
| 5124 |
+
"grad_norm": 2.7285537719726562,
|
| 5125 |
+
"learning_rate": 0.00010526868799852796,
|
| 5126 |
+
"loss": 3.4688,
|
| 5127 |
+
"step": 730
|
| 5128 |
+
},
|
| 5129 |
+
{
|
| 5130 |
+
"epoch": 0.0477713380140749,
|
| 5131 |
+
"grad_norm": 2.3990156650543213,
|
| 5132 |
+
"learning_rate": 0.0001050581241455064,
|
| 5133 |
+
"loss": 3.0999,
|
| 5134 |
+
"step": 731
|
| 5135 |
+
},
|
| 5136 |
+
{
|
| 5137 |
+
"epoch": 0.04783668868167281,
|
| 5138 |
+
"grad_norm": 3.584718704223633,
|
| 5139 |
+
"learning_rate": 0.00010484753780626089,
|
| 5140 |
+
"loss": 2.789,
|
| 5141 |
+
"step": 732
|
| 5142 |
+
},
|
| 5143 |
+
{
|
| 5144 |
+
"epoch": 0.04790203934927073,
|
| 5145 |
+
"grad_norm": 2.8623907566070557,
|
| 5146 |
+
"learning_rate": 0.00010463692991696685,
|
| 5147 |
+
"loss": 3.1882,
|
| 5148 |
+
"step": 733
|
| 5149 |
+
},
|
| 5150 |
+
{
|
| 5151 |
+
"epoch": 0.047967390016868644,
|
| 5152 |
+
"grad_norm": 4.1186909675598145,
|
| 5153 |
+
"learning_rate": 0.00010442630141389549,
|
| 5154 |
+
"loss": 2.9204,
|
| 5155 |
+
"step": 734
|
| 5156 |
+
},
|
| 5157 |
+
{
|
| 5158 |
+
"epoch": 0.04803274068446656,
|
| 5159 |
+
"grad_norm": 3.359563112258911,
|
| 5160 |
+
"learning_rate": 0.00010421565323340971,
|
| 5161 |
+
"loss": 3.1909,
|
| 5162 |
+
"step": 735
|
| 5163 |
+
},
|
| 5164 |
+
{
|
| 5165 |
+
"epoch": 0.04809809135206447,
|
| 5166 |
+
"grad_norm": 2.727571725845337,
|
| 5167 |
+
"learning_rate": 0.00010400498631195992,
|
| 5168 |
+
"loss": 3.3027,
|
| 5169 |
+
"step": 736
|
| 5170 |
+
},
|
| 5171 |
+
{
|
| 5172 |
+
"epoch": 0.048163442019662384,
|
| 5173 |
+
"grad_norm": 3.851203680038452,
|
| 5174 |
+
"learning_rate": 0.00010379430158607975,
|
| 5175 |
+
"loss": 3.3401,
|
| 5176 |
+
"step": 737
|
| 5177 |
+
},
|
| 5178 |
+
{
|
| 5179 |
+
"epoch": 0.0482287926872603,
|
| 5180 |
+
"grad_norm": 2.8851330280303955,
|
| 5181 |
+
"learning_rate": 0.000103583599992382,
|
| 5182 |
+
"loss": 3.091,
|
| 5183 |
+
"step": 738
|
| 5184 |
+
},
|
| 5185 |
+
{
|
| 5186 |
+
"epoch": 0.04829414335485821,
|
| 5187 |
+
"grad_norm": 2.6836740970611572,
|
| 5188 |
+
"learning_rate": 0.0001033728824675545,
|
| 5189 |
+
"loss": 2.9343,
|
| 5190 |
+
"step": 739
|
| 5191 |
+
},
|
| 5192 |
+
{
|
| 5193 |
+
"epoch": 0.04835949402245612,
|
| 5194 |
+
"grad_norm": 3.1017136573791504,
|
| 5195 |
+
"learning_rate": 0.0001031621499483559,
|
| 5196 |
+
"loss": 2.9053,
|
| 5197 |
+
"step": 740
|
| 5198 |
+
},
|
| 5199 |
+
{
|
| 5200 |
+
"epoch": 0.048424844690054036,
|
| 5201 |
+
"grad_norm": 3.8753886222839355,
|
| 5202 |
+
"learning_rate": 0.00010295140337161146,
|
| 5203 |
+
"loss": 2.8989,
|
| 5204 |
+
"step": 741
|
| 5205 |
+
},
|
| 5206 |
+
{
|
| 5207 |
+
"epoch": 0.04849019535765195,
|
| 5208 |
+
"grad_norm": 3.4842278957366943,
|
| 5209 |
+
"learning_rate": 0.00010274064367420897,
|
| 5210 |
+
"loss": 2.7641,
|
| 5211 |
+
"step": 742
|
| 5212 |
+
},
|
| 5213 |
+
{
|
| 5214 |
+
"epoch": 0.04855554602524986,
|
| 5215 |
+
"grad_norm": 3.0435166358947754,
|
| 5216 |
+
"learning_rate": 0.00010252987179309459,
|
| 5217 |
+
"loss": 3.0448,
|
| 5218 |
+
"step": 743
|
| 5219 |
+
},
|
| 5220 |
+
{
|
| 5221 |
+
"epoch": 0.048620896692847776,
|
| 5222 |
+
"grad_norm": 3.1794517040252686,
|
| 5223 |
+
"learning_rate": 0.00010231908866526851,
|
| 5224 |
+
"loss": 2.8337,
|
| 5225 |
+
"step": 744
|
| 5226 |
+
},
|
| 5227 |
+
{
|
| 5228 |
+
"epoch": 0.04868624736044569,
|
| 5229 |
+
"grad_norm": 7.32763147354126,
|
| 5230 |
+
"learning_rate": 0.00010210829522778111,
|
| 5231 |
+
"loss": 3.0432,
|
| 5232 |
+
"step": 745
|
| 5233 |
+
},
|
| 5234 |
+
{
|
| 5235 |
+
"epoch": 0.0487515980280436,
|
| 5236 |
+
"grad_norm": 3.655820846557617,
|
| 5237 |
+
"learning_rate": 0.00010189749241772844,
|
| 5238 |
+
"loss": 2.8711,
|
| 5239 |
+
"step": 746
|
| 5240 |
+
},
|
| 5241 |
+
{
|
| 5242 |
+
"epoch": 0.048816948695641516,
|
| 5243 |
+
"grad_norm": 3.610841751098633,
|
| 5244 |
+
"learning_rate": 0.00010168668117224825,
|
| 5245 |
+
"loss": 2.6454,
|
| 5246 |
+
"step": 747
|
| 5247 |
+
},
|
| 5248 |
+
{
|
| 5249 |
+
"epoch": 0.048882299363239436,
|
| 5250 |
+
"grad_norm": 4.61665153503418,
|
| 5251 |
+
"learning_rate": 0.00010147586242851585,
|
| 5252 |
+
"loss": 2.8728,
|
| 5253 |
+
"step": 748
|
| 5254 |
+
},
|
| 5255 |
+
{
|
| 5256 |
+
"epoch": 0.04894765003083735,
|
| 5257 |
+
"grad_norm": 6.172382354736328,
|
| 5258 |
+
"learning_rate": 0.00010126503712373982,
|
| 5259 |
+
"loss": 2.882,
|
| 5260 |
+
"step": 749
|
| 5261 |
+
},
|
| 5262 |
+
{
|
| 5263 |
+
"epoch": 0.04901300069843526,
|
| 5264 |
+
"grad_norm": 12.269244194030762,
|
| 5265 |
+
"learning_rate": 0.00010105420619515798,
|
| 5266 |
+
"loss": 3.1222,
|
| 5267 |
+
"step": 750
|
| 5268 |
+
},
|
| 5269 |
+
{
|
| 5270 |
+
"epoch": 0.04901300069843526,
|
| 5271 |
+
"eval_loss": 4.020401954650879,
|
| 5272 |
+
"eval_runtime": 1547.5208,
|
| 5273 |
+
"eval_samples_per_second": 16.654,
|
| 5274 |
+
"eval_steps_per_second": 8.327,
|
| 5275 |
+
"step": 750
|
| 5276 |
}
|
| 5277 |
],
|
| 5278 |
"logging_steps": 1,
|
|
|
|
| 5292 |
"attributes": {}
|
| 5293 |
}
|
| 5294 |
},
|
| 5295 |
+
"total_flos": 1.109738548887552e+18,
|
| 5296 |
"train_batch_size": 2,
|
| 5297 |
"trial_name": null,
|
| 5298 |
"trial_params": null
|