Training in progress, step 1044, 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 17640136
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:43388660cb71ed584203e4789b8a35953c132f30781a796aba86ab8edaa79b8b
|
| 3 |
size 17640136
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 9569204
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8461677d2dfb1d6f198f8a7f7b8b3358af860be5dc0f2884a089ab89119e52dc
|
| 3 |
size 9569204
|
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:ddaaf88fa715a3952eaf5f63ebfe03e656e5c9c40261b0a09b2c5467725e8930
|
| 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:c729447b6b732ff2668c562e8a044d8d5a17738e8f84aa51416aac42a89dc821
|
| 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": 348,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -4903,6 +4903,2450 @@
|
|
| 4903 |
"eval_samples_per_second": 48.265,
|
| 4904 |
"eval_steps_per_second": 24.132,
|
| 4905 |
"step": 696
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4906 |
}
|
| 4907 |
],
|
| 4908 |
"logging_steps": 1,
|
|
@@ -4922,7 +7366,7 @@
|
|
| 4922 |
"attributes": {}
|
| 4923 |
}
|
| 4924 |
},
|
| 4925 |
-
"total_flos": 1.
|
| 4926 |
"train_batch_size": 2,
|
| 4927 |
"trial_name": null,
|
| 4928 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.5940256045519203,
|
| 5 |
"eval_steps": 348,
|
| 6 |
+
"global_step": 1044,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 4903 |
"eval_samples_per_second": 48.265,
|
| 4904 |
"eval_steps_per_second": 24.132,
|
| 4905 |
"step": 696
|
| 4906 |
+
},
|
| 4907 |
+
{
|
| 4908 |
+
"epoch": 0.3965860597439545,
|
| 4909 |
+
"grad_norm": 0.5474138855934143,
|
| 4910 |
+
"learning_rate": 0.0001009092762264271,
|
| 4911 |
+
"loss": 1.4683,
|
| 4912 |
+
"step": 697
|
| 4913 |
+
},
|
| 4914 |
+
{
|
| 4915 |
+
"epoch": 0.39715504978662874,
|
| 4916 |
+
"grad_norm": 0.6160016059875488,
|
| 4917 |
+
"learning_rate": 0.00010068196128122975,
|
| 4918 |
+
"loss": 1.6705,
|
| 4919 |
+
"step": 698
|
| 4920 |
+
},
|
| 4921 |
+
{
|
| 4922 |
+
"epoch": 0.397724039829303,
|
| 4923 |
+
"grad_norm": 0.5745415687561035,
|
| 4924 |
+
"learning_rate": 0.00010045464281197327,
|
| 4925 |
+
"loss": 1.5104,
|
| 4926 |
+
"step": 699
|
| 4927 |
+
},
|
| 4928 |
+
{
|
| 4929 |
+
"epoch": 0.39829302987197723,
|
| 4930 |
+
"grad_norm": 0.5802525281906128,
|
| 4931 |
+
"learning_rate": 0.0001002273219933339,
|
| 4932 |
+
"loss": 1.4029,
|
| 4933 |
+
"step": 700
|
| 4934 |
+
},
|
| 4935 |
+
{
|
| 4936 |
+
"epoch": 0.3988620199146515,
|
| 4937 |
+
"grad_norm": 0.5592519044876099,
|
| 4938 |
+
"learning_rate": 0.0001,
|
| 4939 |
+
"loss": 1.6325,
|
| 4940 |
+
"step": 701
|
| 4941 |
+
},
|
| 4942 |
+
{
|
| 4943 |
+
"epoch": 0.3994310099573257,
|
| 4944 |
+
"grad_norm": 0.6051873564720154,
|
| 4945 |
+
"learning_rate": 9.977267800666613e-05,
|
| 4946 |
+
"loss": 1.688,
|
| 4947 |
+
"step": 702
|
| 4948 |
+
},
|
| 4949 |
+
{
|
| 4950 |
+
"epoch": 0.4,
|
| 4951 |
+
"grad_norm": 0.5836036205291748,
|
| 4952 |
+
"learning_rate": 9.954535718802675e-05,
|
| 4953 |
+
"loss": 1.3107,
|
| 4954 |
+
"step": 703
|
| 4955 |
+
},
|
| 4956 |
+
{
|
| 4957 |
+
"epoch": 0.40056899004267427,
|
| 4958 |
+
"grad_norm": 0.5733322501182556,
|
| 4959 |
+
"learning_rate": 9.931803871877028e-05,
|
| 4960 |
+
"loss": 1.7469,
|
| 4961 |
+
"step": 704
|
| 4962 |
+
},
|
| 4963 |
+
{
|
| 4964 |
+
"epoch": 0.4011379800853485,
|
| 4965 |
+
"grad_norm": 0.5718969106674194,
|
| 4966 |
+
"learning_rate": 9.909072377357294e-05,
|
| 4967 |
+
"loss": 1.3822,
|
| 4968 |
+
"step": 705
|
| 4969 |
+
},
|
| 4970 |
+
{
|
| 4971 |
+
"epoch": 0.40170697012802276,
|
| 4972 |
+
"grad_norm": 0.5877561569213867,
|
| 4973 |
+
"learning_rate": 9.88634135270928e-05,
|
| 4974 |
+
"loss": 1.6344,
|
| 4975 |
+
"step": 706
|
| 4976 |
+
},
|
| 4977 |
+
{
|
| 4978 |
+
"epoch": 0.402275960170697,
|
| 4979 |
+
"grad_norm": 0.5636436939239502,
|
| 4980 |
+
"learning_rate": 9.863610915396365e-05,
|
| 4981 |
+
"loss": 1.5552,
|
| 4982 |
+
"step": 707
|
| 4983 |
+
},
|
| 4984 |
+
{
|
| 4985 |
+
"epoch": 0.40284495021337124,
|
| 4986 |
+
"grad_norm": 0.5809296369552612,
|
| 4987 |
+
"learning_rate": 9.840881182878895e-05,
|
| 4988 |
+
"loss": 1.3633,
|
| 4989 |
+
"step": 708
|
| 4990 |
+
},
|
| 4991 |
+
{
|
| 4992 |
+
"epoch": 0.40341394025604554,
|
| 4993 |
+
"grad_norm": 0.5500168204307556,
|
| 4994 |
+
"learning_rate": 9.81815227261357e-05,
|
| 4995 |
+
"loss": 1.4063,
|
| 4996 |
+
"step": 709
|
| 4997 |
+
},
|
| 4998 |
+
{
|
| 4999 |
+
"epoch": 0.4039829302987198,
|
| 5000 |
+
"grad_norm": 0.5806904435157776,
|
| 5001 |
+
"learning_rate": 9.795424302052836e-05,
|
| 5002 |
+
"loss": 1.5629,
|
| 5003 |
+
"step": 710
|
| 5004 |
+
},
|
| 5005 |
+
{
|
| 5006 |
+
"epoch": 0.40455192034139403,
|
| 5007 |
+
"grad_norm": 0.5868257880210876,
|
| 5008 |
+
"learning_rate": 9.77269738864429e-05,
|
| 5009 |
+
"loss": 1.3655,
|
| 5010 |
+
"step": 711
|
| 5011 |
+
},
|
| 5012 |
+
{
|
| 5013 |
+
"epoch": 0.4051209103840683,
|
| 5014 |
+
"grad_norm": 0.5417432188987732,
|
| 5015 |
+
"learning_rate": 9.749971649830071e-05,
|
| 5016 |
+
"loss": 1.4914,
|
| 5017 |
+
"step": 712
|
| 5018 |
+
},
|
| 5019 |
+
{
|
| 5020 |
+
"epoch": 0.4056899004267425,
|
| 5021 |
+
"grad_norm": 0.6012546420097351,
|
| 5022 |
+
"learning_rate": 9.727247203046234e-05,
|
| 5023 |
+
"loss": 1.5365,
|
| 5024 |
+
"step": 713
|
| 5025 |
+
},
|
| 5026 |
+
{
|
| 5027 |
+
"epoch": 0.40625889046941677,
|
| 5028 |
+
"grad_norm": 0.5691578388214111,
|
| 5029 |
+
"learning_rate": 9.704524165722174e-05,
|
| 5030 |
+
"loss": 1.5959,
|
| 5031 |
+
"step": 714
|
| 5032 |
+
},
|
| 5033 |
+
{
|
| 5034 |
+
"epoch": 0.406827880512091,
|
| 5035 |
+
"grad_norm": 0.5487850904464722,
|
| 5036 |
+
"learning_rate": 9.681802655279986e-05,
|
| 5037 |
+
"loss": 1.4469,
|
| 5038 |
+
"step": 715
|
| 5039 |
+
},
|
| 5040 |
+
{
|
| 5041 |
+
"epoch": 0.4073968705547653,
|
| 5042 |
+
"grad_norm": 0.6280918121337891,
|
| 5043 |
+
"learning_rate": 9.659082789133884e-05,
|
| 5044 |
+
"loss": 1.338,
|
| 5045 |
+
"step": 716
|
| 5046 |
+
},
|
| 5047 |
+
{
|
| 5048 |
+
"epoch": 0.40796586059743956,
|
| 5049 |
+
"grad_norm": 0.5909377932548523,
|
| 5050 |
+
"learning_rate": 9.63636468468959e-05,
|
| 5051 |
+
"loss": 1.6272,
|
| 5052 |
+
"step": 717
|
| 5053 |
+
},
|
| 5054 |
+
{
|
| 5055 |
+
"epoch": 0.4085348506401138,
|
| 5056 |
+
"grad_norm": 0.6044595837593079,
|
| 5057 |
+
"learning_rate": 9.613648459343708e-05,
|
| 5058 |
+
"loss": 1.5717,
|
| 5059 |
+
"step": 718
|
| 5060 |
+
},
|
| 5061 |
+
{
|
| 5062 |
+
"epoch": 0.40910384068278804,
|
| 5063 |
+
"grad_norm": 0.5833640098571777,
|
| 5064 |
+
"learning_rate": 9.590934230483149e-05,
|
| 5065 |
+
"loss": 1.4213,
|
| 5066 |
+
"step": 719
|
| 5067 |
+
},
|
| 5068 |
+
{
|
| 5069 |
+
"epoch": 0.4096728307254623,
|
| 5070 |
+
"grad_norm": 0.6057854890823364,
|
| 5071 |
+
"learning_rate": 9.568222115484488e-05,
|
| 5072 |
+
"loss": 1.4861,
|
| 5073 |
+
"step": 720
|
| 5074 |
+
},
|
| 5075 |
+
{
|
| 5076 |
+
"epoch": 0.41024182076813653,
|
| 5077 |
+
"grad_norm": 0.5813032984733582,
|
| 5078 |
+
"learning_rate": 9.54551223171339e-05,
|
| 5079 |
+
"loss": 1.5329,
|
| 5080 |
+
"step": 721
|
| 5081 |
+
},
|
| 5082 |
+
{
|
| 5083 |
+
"epoch": 0.41081081081081083,
|
| 5084 |
+
"grad_norm": 0.5498741865158081,
|
| 5085 |
+
"learning_rate": 9.522804696523991e-05,
|
| 5086 |
+
"loss": 1.4457,
|
| 5087 |
+
"step": 722
|
| 5088 |
+
},
|
| 5089 |
+
{
|
| 5090 |
+
"epoch": 0.4113798008534851,
|
| 5091 |
+
"grad_norm": 0.5357645750045776,
|
| 5092 |
+
"learning_rate": 9.500099627258282e-05,
|
| 5093 |
+
"loss": 1.2792,
|
| 5094 |
+
"step": 723
|
| 5095 |
+
},
|
| 5096 |
+
{
|
| 5097 |
+
"epoch": 0.4119487908961593,
|
| 5098 |
+
"grad_norm": 0.5478993654251099,
|
| 5099 |
+
"learning_rate": 9.477397141245519e-05,
|
| 5100 |
+
"loss": 1.5071,
|
| 5101 |
+
"step": 724
|
| 5102 |
+
},
|
| 5103 |
+
{
|
| 5104 |
+
"epoch": 0.41251778093883357,
|
| 5105 |
+
"grad_norm": 0.5776642560958862,
|
| 5106 |
+
"learning_rate": 9.454697355801598e-05,
|
| 5107 |
+
"loss": 1.3664,
|
| 5108 |
+
"step": 725
|
| 5109 |
+
},
|
| 5110 |
+
{
|
| 5111 |
+
"epoch": 0.4130867709815078,
|
| 5112 |
+
"grad_norm": 0.6283994913101196,
|
| 5113 |
+
"learning_rate": 9.432000388228473e-05,
|
| 5114 |
+
"loss": 1.3994,
|
| 5115 |
+
"step": 726
|
| 5116 |
+
},
|
| 5117 |
+
{
|
| 5118 |
+
"epoch": 0.41365576102418206,
|
| 5119 |
+
"grad_norm": 0.6153956651687622,
|
| 5120 |
+
"learning_rate": 9.409306355813529e-05,
|
| 5121 |
+
"loss": 1.2524,
|
| 5122 |
+
"step": 727
|
| 5123 |
+
},
|
| 5124 |
+
{
|
| 5125 |
+
"epoch": 0.41422475106685636,
|
| 5126 |
+
"grad_norm": 0.5952728986740112,
|
| 5127 |
+
"learning_rate": 9.386615375828984e-05,
|
| 5128 |
+
"loss": 1.5941,
|
| 5129 |
+
"step": 728
|
| 5130 |
+
},
|
| 5131 |
+
{
|
| 5132 |
+
"epoch": 0.4147937411095306,
|
| 5133 |
+
"grad_norm": 0.5799689292907715,
|
| 5134 |
+
"learning_rate": 9.36392756553129e-05,
|
| 5135 |
+
"loss": 1.3113,
|
| 5136 |
+
"step": 729
|
| 5137 |
+
},
|
| 5138 |
+
{
|
| 5139 |
+
"epoch": 0.41536273115220484,
|
| 5140 |
+
"grad_norm": 0.5933107733726501,
|
| 5141 |
+
"learning_rate": 9.341243042160503e-05,
|
| 5142 |
+
"loss": 1.6378,
|
| 5143 |
+
"step": 730
|
| 5144 |
+
},
|
| 5145 |
+
{
|
| 5146 |
+
"epoch": 0.4159317211948791,
|
| 5147 |
+
"grad_norm": 0.5808780789375305,
|
| 5148 |
+
"learning_rate": 9.318561922939711e-05,
|
| 5149 |
+
"loss": 1.663,
|
| 5150 |
+
"step": 731
|
| 5151 |
+
},
|
| 5152 |
+
{
|
| 5153 |
+
"epoch": 0.41650071123755333,
|
| 5154 |
+
"grad_norm": 0.5661304593086243,
|
| 5155 |
+
"learning_rate": 9.295884325074398e-05,
|
| 5156 |
+
"loss": 1.4145,
|
| 5157 |
+
"step": 732
|
| 5158 |
+
},
|
| 5159 |
+
{
|
| 5160 |
+
"epoch": 0.4170697012802276,
|
| 5161 |
+
"grad_norm": 0.577038049697876,
|
| 5162 |
+
"learning_rate": 9.273210365751862e-05,
|
| 5163 |
+
"loss": 1.4288,
|
| 5164 |
+
"step": 733
|
| 5165 |
+
},
|
| 5166 |
+
{
|
| 5167 |
+
"epoch": 0.4176386913229018,
|
| 5168 |
+
"grad_norm": 0.5904839038848877,
|
| 5169 |
+
"learning_rate": 9.250540162140597e-05,
|
| 5170 |
+
"loss": 1.5257,
|
| 5171 |
+
"step": 734
|
| 5172 |
+
},
|
| 5173 |
+
{
|
| 5174 |
+
"epoch": 0.4182076813655761,
|
| 5175 |
+
"grad_norm": 0.5645294785499573,
|
| 5176 |
+
"learning_rate": 9.227873831389677e-05,
|
| 5177 |
+
"loss": 1.4073,
|
| 5178 |
+
"step": 735
|
| 5179 |
+
},
|
| 5180 |
+
{
|
| 5181 |
+
"epoch": 0.41877667140825037,
|
| 5182 |
+
"grad_norm": 0.5541549921035767,
|
| 5183 |
+
"learning_rate": 9.205211490628173e-05,
|
| 5184 |
+
"loss": 1.3965,
|
| 5185 |
+
"step": 736
|
| 5186 |
+
},
|
| 5187 |
+
{
|
| 5188 |
+
"epoch": 0.4193456614509246,
|
| 5189 |
+
"grad_norm": 0.6137387752532959,
|
| 5190 |
+
"learning_rate": 9.18255325696454e-05,
|
| 5191 |
+
"loss": 1.3849,
|
| 5192 |
+
"step": 737
|
| 5193 |
+
},
|
| 5194 |
+
{
|
| 5195 |
+
"epoch": 0.41991465149359886,
|
| 5196 |
+
"grad_norm": 0.588316798210144,
|
| 5197 |
+
"learning_rate": 9.159899247486004e-05,
|
| 5198 |
+
"loss": 1.4989,
|
| 5199 |
+
"step": 738
|
| 5200 |
+
},
|
| 5201 |
+
{
|
| 5202 |
+
"epoch": 0.4204836415362731,
|
| 5203 |
+
"grad_norm": 0.567848265171051,
|
| 5204 |
+
"learning_rate": 9.13724957925797e-05,
|
| 5205 |
+
"loss": 1.4989,
|
| 5206 |
+
"step": 739
|
| 5207 |
+
},
|
| 5208 |
+
{
|
| 5209 |
+
"epoch": 0.42105263157894735,
|
| 5210 |
+
"grad_norm": 0.554695188999176,
|
| 5211 |
+
"learning_rate": 9.114604369323395e-05,
|
| 5212 |
+
"loss": 1.4509,
|
| 5213 |
+
"step": 740
|
| 5214 |
+
},
|
| 5215 |
+
{
|
| 5216 |
+
"epoch": 0.42162162162162165,
|
| 5217 |
+
"grad_norm": 0.5507339835166931,
|
| 5218 |
+
"learning_rate": 9.091963734702208e-05,
|
| 5219 |
+
"loss": 1.364,
|
| 5220 |
+
"step": 741
|
| 5221 |
+
},
|
| 5222 |
+
{
|
| 5223 |
+
"epoch": 0.4221906116642959,
|
| 5224 |
+
"grad_norm": 0.569786012172699,
|
| 5225 |
+
"learning_rate": 9.069327792390695e-05,
|
| 5226 |
+
"loss": 1.5775,
|
| 5227 |
+
"step": 742
|
| 5228 |
+
},
|
| 5229 |
+
{
|
| 5230 |
+
"epoch": 0.42275960170697013,
|
| 5231 |
+
"grad_norm": 0.563234806060791,
|
| 5232 |
+
"learning_rate": 9.046696659360894e-05,
|
| 5233 |
+
"loss": 1.4557,
|
| 5234 |
+
"step": 743
|
| 5235 |
+
},
|
| 5236 |
+
{
|
| 5237 |
+
"epoch": 0.4233285917496444,
|
| 5238 |
+
"grad_norm": 0.5537723302841187,
|
| 5239 |
+
"learning_rate": 9.024070452559986e-05,
|
| 5240 |
+
"loss": 1.443,
|
| 5241 |
+
"step": 744
|
| 5242 |
+
},
|
| 5243 |
+
{
|
| 5244 |
+
"epoch": 0.4238975817923186,
|
| 5245 |
+
"grad_norm": 0.5786699056625366,
|
| 5246 |
+
"learning_rate": 9.001449288909702e-05,
|
| 5247 |
+
"loss": 1.2683,
|
| 5248 |
+
"step": 745
|
| 5249 |
+
},
|
| 5250 |
+
{
|
| 5251 |
+
"epoch": 0.42446657183499287,
|
| 5252 |
+
"grad_norm": 0.55182945728302,
|
| 5253 |
+
"learning_rate": 8.978833285305705e-05,
|
| 5254 |
+
"loss": 1.4565,
|
| 5255 |
+
"step": 746
|
| 5256 |
+
},
|
| 5257 |
+
{
|
| 5258 |
+
"epoch": 0.42503556187766717,
|
| 5259 |
+
"grad_norm": 0.5818150043487549,
|
| 5260 |
+
"learning_rate": 8.956222558616998e-05,
|
| 5261 |
+
"loss": 1.6502,
|
| 5262 |
+
"step": 747
|
| 5263 |
+
},
|
| 5264 |
+
{
|
| 5265 |
+
"epoch": 0.4256045519203414,
|
| 5266 |
+
"grad_norm": 0.6044638752937317,
|
| 5267 |
+
"learning_rate": 8.933617225685319e-05,
|
| 5268 |
+
"loss": 1.4631,
|
| 5269 |
+
"step": 748
|
| 5270 |
+
},
|
| 5271 |
+
{
|
| 5272 |
+
"epoch": 0.42617354196301566,
|
| 5273 |
+
"grad_norm": 0.568188488483429,
|
| 5274 |
+
"learning_rate": 8.91101740332453e-05,
|
| 5275 |
+
"loss": 1.5997,
|
| 5276 |
+
"step": 749
|
| 5277 |
+
},
|
| 5278 |
+
{
|
| 5279 |
+
"epoch": 0.4267425320056899,
|
| 5280 |
+
"grad_norm": 0.5530648231506348,
|
| 5281 |
+
"learning_rate": 8.888423208320008e-05,
|
| 5282 |
+
"loss": 1.174,
|
| 5283 |
+
"step": 750
|
| 5284 |
+
},
|
| 5285 |
+
{
|
| 5286 |
+
"epoch": 0.42731152204836415,
|
| 5287 |
+
"grad_norm": 0.5782289505004883,
|
| 5288 |
+
"learning_rate": 8.865834757428064e-05,
|
| 5289 |
+
"loss": 1.5198,
|
| 5290 |
+
"step": 751
|
| 5291 |
+
},
|
| 5292 |
+
{
|
| 5293 |
+
"epoch": 0.4278805120910384,
|
| 5294 |
+
"grad_norm": 0.5685307383537292,
|
| 5295 |
+
"learning_rate": 8.843252167375322e-05,
|
| 5296 |
+
"loss": 1.5545,
|
| 5297 |
+
"step": 752
|
| 5298 |
+
},
|
| 5299 |
+
{
|
| 5300 |
+
"epoch": 0.42844950213371263,
|
| 5301 |
+
"grad_norm": 0.5832937359809875,
|
| 5302 |
+
"learning_rate": 8.820675554858115e-05,
|
| 5303 |
+
"loss": 1.5776,
|
| 5304 |
+
"step": 753
|
| 5305 |
+
},
|
| 5306 |
+
{
|
| 5307 |
+
"epoch": 0.42901849217638693,
|
| 5308 |
+
"grad_norm": 0.6279184818267822,
|
| 5309 |
+
"learning_rate": 8.7981050365419e-05,
|
| 5310 |
+
"loss": 1.5975,
|
| 5311 |
+
"step": 754
|
| 5312 |
+
},
|
| 5313 |
+
{
|
| 5314 |
+
"epoch": 0.4295874822190612,
|
| 5315 |
+
"grad_norm": 0.5440697073936462,
|
| 5316 |
+
"learning_rate": 8.775540729060618e-05,
|
| 5317 |
+
"loss": 1.3772,
|
| 5318 |
+
"step": 755
|
| 5319 |
+
},
|
| 5320 |
+
{
|
| 5321 |
+
"epoch": 0.4301564722617354,
|
| 5322 |
+
"grad_norm": 0.6341460347175598,
|
| 5323 |
+
"learning_rate": 8.752982749016139e-05,
|
| 5324 |
+
"loss": 1.573,
|
| 5325 |
+
"step": 756
|
| 5326 |
+
},
|
| 5327 |
+
{
|
| 5328 |
+
"epoch": 0.43072546230440967,
|
| 5329 |
+
"grad_norm": 0.5840321779251099,
|
| 5330 |
+
"learning_rate": 8.730431212977625e-05,
|
| 5331 |
+
"loss": 1.5051,
|
| 5332 |
+
"step": 757
|
| 5333 |
+
},
|
| 5334 |
+
{
|
| 5335 |
+
"epoch": 0.4312944523470839,
|
| 5336 |
+
"grad_norm": 0.5965592265129089,
|
| 5337 |
+
"learning_rate": 8.70788623748094e-05,
|
| 5338 |
+
"loss": 1.5323,
|
| 5339 |
+
"step": 758
|
| 5340 |
+
},
|
| 5341 |
+
{
|
| 5342 |
+
"epoch": 0.43186344238975816,
|
| 5343 |
+
"grad_norm": 0.5905702710151672,
|
| 5344 |
+
"learning_rate": 8.68534793902805e-05,
|
| 5345 |
+
"loss": 1.4051,
|
| 5346 |
+
"step": 759
|
| 5347 |
+
},
|
| 5348 |
+
{
|
| 5349 |
+
"epoch": 0.43243243243243246,
|
| 5350 |
+
"grad_norm": 0.5640906691551208,
|
| 5351 |
+
"learning_rate": 8.662816434086404e-05,
|
| 5352 |
+
"loss": 1.6614,
|
| 5353 |
+
"step": 760
|
| 5354 |
+
},
|
| 5355 |
+
{
|
| 5356 |
+
"epoch": 0.4330014224751067,
|
| 5357 |
+
"grad_norm": 0.5574825406074524,
|
| 5358 |
+
"learning_rate": 8.64029183908836e-05,
|
| 5359 |
+
"loss": 1.3464,
|
| 5360 |
+
"step": 761
|
| 5361 |
+
},
|
| 5362 |
+
{
|
| 5363 |
+
"epoch": 0.43357041251778095,
|
| 5364 |
+
"grad_norm": 0.5866842865943909,
|
| 5365 |
+
"learning_rate": 8.617774270430566e-05,
|
| 5366 |
+
"loss": 1.4531,
|
| 5367 |
+
"step": 762
|
| 5368 |
+
},
|
| 5369 |
+
{
|
| 5370 |
+
"epoch": 0.4341394025604552,
|
| 5371 |
+
"grad_norm": 0.6260978579521179,
|
| 5372 |
+
"learning_rate": 8.595263844473353e-05,
|
| 5373 |
+
"loss": 1.4005,
|
| 5374 |
+
"step": 763
|
| 5375 |
+
},
|
| 5376 |
+
{
|
| 5377 |
+
"epoch": 0.43470839260312943,
|
| 5378 |
+
"grad_norm": 0.5732872486114502,
|
| 5379 |
+
"learning_rate": 8.572760677540154e-05,
|
| 5380 |
+
"loss": 1.366,
|
| 5381 |
+
"step": 764
|
| 5382 |
+
},
|
| 5383 |
+
{
|
| 5384 |
+
"epoch": 0.4352773826458037,
|
| 5385 |
+
"grad_norm": 0.5682139992713928,
|
| 5386 |
+
"learning_rate": 8.550264885916877e-05,
|
| 5387 |
+
"loss": 1.359,
|
| 5388 |
+
"step": 765
|
| 5389 |
+
},
|
| 5390 |
+
{
|
| 5391 |
+
"epoch": 0.435846372688478,
|
| 5392 |
+
"grad_norm": 0.5898922085762024,
|
| 5393 |
+
"learning_rate": 8.527776585851328e-05,
|
| 5394 |
+
"loss": 1.5197,
|
| 5395 |
+
"step": 766
|
| 5396 |
+
},
|
| 5397 |
+
{
|
| 5398 |
+
"epoch": 0.4364153627311522,
|
| 5399 |
+
"grad_norm": 0.5902604460716248,
|
| 5400 |
+
"learning_rate": 8.505295893552594e-05,
|
| 5401 |
+
"loss": 1.4844,
|
| 5402 |
+
"step": 767
|
| 5403 |
+
},
|
| 5404 |
+
{
|
| 5405 |
+
"epoch": 0.43698435277382647,
|
| 5406 |
+
"grad_norm": 0.6057772040367126,
|
| 5407 |
+
"learning_rate": 8.482822925190452e-05,
|
| 5408 |
+
"loss": 1.5739,
|
| 5409 |
+
"step": 768
|
| 5410 |
+
},
|
| 5411 |
+
{
|
| 5412 |
+
"epoch": 0.4375533428165007,
|
| 5413 |
+
"grad_norm": 0.5546793341636658,
|
| 5414 |
+
"learning_rate": 8.460357796894773e-05,
|
| 5415 |
+
"loss": 1.4748,
|
| 5416 |
+
"step": 769
|
| 5417 |
+
},
|
| 5418 |
+
{
|
| 5419 |
+
"epoch": 0.43812233285917496,
|
| 5420 |
+
"grad_norm": 0.5493602156639099,
|
| 5421 |
+
"learning_rate": 8.437900624754895e-05,
|
| 5422 |
+
"loss": 1.3922,
|
| 5423 |
+
"step": 770
|
| 5424 |
+
},
|
| 5425 |
+
{
|
| 5426 |
+
"epoch": 0.4386913229018492,
|
| 5427 |
+
"grad_norm": 0.5499581098556519,
|
| 5428 |
+
"learning_rate": 8.415451524819058e-05,
|
| 5429 |
+
"loss": 1.574,
|
| 5430 |
+
"step": 771
|
| 5431 |
+
},
|
| 5432 |
+
{
|
| 5433 |
+
"epoch": 0.43926031294452345,
|
| 5434 |
+
"grad_norm": 0.5515440702438354,
|
| 5435 |
+
"learning_rate": 8.393010613093781e-05,
|
| 5436 |
+
"loss": 1.3672,
|
| 5437 |
+
"step": 772
|
| 5438 |
+
},
|
| 5439 |
+
{
|
| 5440 |
+
"epoch": 0.43982930298719775,
|
| 5441 |
+
"grad_norm": 0.5613058805465698,
|
| 5442 |
+
"learning_rate": 8.370578005543278e-05,
|
| 5443 |
+
"loss": 1.3815,
|
| 5444 |
+
"step": 773
|
| 5445 |
+
},
|
| 5446 |
+
{
|
| 5447 |
+
"epoch": 0.440398293029872,
|
| 5448 |
+
"grad_norm": 0.5643707513809204,
|
| 5449 |
+
"learning_rate": 8.348153818088844e-05,
|
| 5450 |
+
"loss": 1.5947,
|
| 5451 |
+
"step": 774
|
| 5452 |
+
},
|
| 5453 |
+
{
|
| 5454 |
+
"epoch": 0.44096728307254623,
|
| 5455 |
+
"grad_norm": 0.6310828924179077,
|
| 5456 |
+
"learning_rate": 8.325738166608263e-05,
|
| 5457 |
+
"loss": 1.5413,
|
| 5458 |
+
"step": 775
|
| 5459 |
+
},
|
| 5460 |
+
{
|
| 5461 |
+
"epoch": 0.4415362731152205,
|
| 5462 |
+
"grad_norm": 0.6655511856079102,
|
| 5463 |
+
"learning_rate": 8.303331166935209e-05,
|
| 5464 |
+
"loss": 1.5198,
|
| 5465 |
+
"step": 776
|
| 5466 |
+
},
|
| 5467 |
+
{
|
| 5468 |
+
"epoch": 0.4421052631578947,
|
| 5469 |
+
"grad_norm": 0.5539633631706238,
|
| 5470 |
+
"learning_rate": 8.280932934858652e-05,
|
| 5471 |
+
"loss": 1.4308,
|
| 5472 |
+
"step": 777
|
| 5473 |
+
},
|
| 5474 |
+
{
|
| 5475 |
+
"epoch": 0.44267425320056897,
|
| 5476 |
+
"grad_norm": 0.5974248647689819,
|
| 5477 |
+
"learning_rate": 8.25854358612225e-05,
|
| 5478 |
+
"loss": 1.537,
|
| 5479 |
+
"step": 778
|
| 5480 |
+
},
|
| 5481 |
+
{
|
| 5482 |
+
"epoch": 0.44324324324324327,
|
| 5483 |
+
"grad_norm": 0.5987525582313538,
|
| 5484 |
+
"learning_rate": 8.236163236423767e-05,
|
| 5485 |
+
"loss": 1.5318,
|
| 5486 |
+
"step": 779
|
| 5487 |
+
},
|
| 5488 |
+
{
|
| 5489 |
+
"epoch": 0.4438122332859175,
|
| 5490 |
+
"grad_norm": 0.5623188018798828,
|
| 5491 |
+
"learning_rate": 8.213792001414445e-05,
|
| 5492 |
+
"loss": 1.6016,
|
| 5493 |
+
"step": 780
|
| 5494 |
+
},
|
| 5495 |
+
{
|
| 5496 |
+
"epoch": 0.44438122332859176,
|
| 5497 |
+
"grad_norm": 0.5642153024673462,
|
| 5498 |
+
"learning_rate": 8.191429996698436e-05,
|
| 5499 |
+
"loss": 1.4452,
|
| 5500 |
+
"step": 781
|
| 5501 |
+
},
|
| 5502 |
+
{
|
| 5503 |
+
"epoch": 0.444950213371266,
|
| 5504 |
+
"grad_norm": 0.6042040586471558,
|
| 5505 |
+
"learning_rate": 8.1690773378322e-05,
|
| 5506 |
+
"loss": 1.6325,
|
| 5507 |
+
"step": 782
|
| 5508 |
+
},
|
| 5509 |
+
{
|
| 5510 |
+
"epoch": 0.44551920341394025,
|
| 5511 |
+
"grad_norm": 0.5777531862258911,
|
| 5512 |
+
"learning_rate": 8.146734140323896e-05,
|
| 5513 |
+
"loss": 1.6388,
|
| 5514 |
+
"step": 783
|
| 5515 |
+
},
|
| 5516 |
+
{
|
| 5517 |
+
"epoch": 0.4460881934566145,
|
| 5518 |
+
"grad_norm": 0.5600481629371643,
|
| 5519 |
+
"learning_rate": 8.124400519632788e-05,
|
| 5520 |
+
"loss": 1.5077,
|
| 5521 |
+
"step": 784
|
| 5522 |
+
},
|
| 5523 |
+
{
|
| 5524 |
+
"epoch": 0.4466571834992888,
|
| 5525 |
+
"grad_norm": 0.5644223690032959,
|
| 5526 |
+
"learning_rate": 8.102076591168655e-05,
|
| 5527 |
+
"loss": 1.4056,
|
| 5528 |
+
"step": 785
|
| 5529 |
+
},
|
| 5530 |
+
{
|
| 5531 |
+
"epoch": 0.44722617354196303,
|
| 5532 |
+
"grad_norm": 0.6023853421211243,
|
| 5533 |
+
"learning_rate": 8.079762470291191e-05,
|
| 5534 |
+
"loss": 1.713,
|
| 5535 |
+
"step": 786
|
| 5536 |
+
},
|
| 5537 |
+
{
|
| 5538 |
+
"epoch": 0.4477951635846373,
|
| 5539 |
+
"grad_norm": 0.5626102685928345,
|
| 5540 |
+
"learning_rate": 8.05745827230941e-05,
|
| 5541 |
+
"loss": 1.5125,
|
| 5542 |
+
"step": 787
|
| 5543 |
+
},
|
| 5544 |
+
{
|
| 5545 |
+
"epoch": 0.4483641536273115,
|
| 5546 |
+
"grad_norm": 0.5824998617172241,
|
| 5547 |
+
"learning_rate": 8.035164112481048e-05,
|
| 5548 |
+
"loss": 1.4695,
|
| 5549 |
+
"step": 788
|
| 5550 |
+
},
|
| 5551 |
+
{
|
| 5552 |
+
"epoch": 0.44893314366998577,
|
| 5553 |
+
"grad_norm": 0.5714951157569885,
|
| 5554 |
+
"learning_rate": 8.01288010601197e-05,
|
| 5555 |
+
"loss": 1.4452,
|
| 5556 |
+
"step": 789
|
| 5557 |
+
},
|
| 5558 |
+
{
|
| 5559 |
+
"epoch": 0.44950213371266,
|
| 5560 |
+
"grad_norm": 0.5934897065162659,
|
| 5561 |
+
"learning_rate": 7.990606368055564e-05,
|
| 5562 |
+
"loss": 1.5389,
|
| 5563 |
+
"step": 790
|
| 5564 |
+
},
|
| 5565 |
+
{
|
| 5566 |
+
"epoch": 0.45007112375533426,
|
| 5567 |
+
"grad_norm": 0.5794687867164612,
|
| 5568 |
+
"learning_rate": 7.968343013712167e-05,
|
| 5569 |
+
"loss": 1.4127,
|
| 5570 |
+
"step": 791
|
| 5571 |
+
},
|
| 5572 |
+
{
|
| 5573 |
+
"epoch": 0.45064011379800856,
|
| 5574 |
+
"grad_norm": 0.5628656148910522,
|
| 5575 |
+
"learning_rate": 7.946090158028455e-05,
|
| 5576 |
+
"loss": 1.4798,
|
| 5577 |
+
"step": 792
|
| 5578 |
+
},
|
| 5579 |
+
{
|
| 5580 |
+
"epoch": 0.4512091038406828,
|
| 5581 |
+
"grad_norm": 0.5794563293457031,
|
| 5582 |
+
"learning_rate": 7.923847915996851e-05,
|
| 5583 |
+
"loss": 1.5584,
|
| 5584 |
+
"step": 793
|
| 5585 |
+
},
|
| 5586 |
+
{
|
| 5587 |
+
"epoch": 0.45177809388335705,
|
| 5588 |
+
"grad_norm": 0.5685121417045593,
|
| 5589 |
+
"learning_rate": 7.901616402554933e-05,
|
| 5590 |
+
"loss": 1.51,
|
| 5591 |
+
"step": 794
|
| 5592 |
+
},
|
| 5593 |
+
{
|
| 5594 |
+
"epoch": 0.4523470839260313,
|
| 5595 |
+
"grad_norm": 0.568209171295166,
|
| 5596 |
+
"learning_rate": 7.87939573258483e-05,
|
| 5597 |
+
"loss": 1.5588,
|
| 5598 |
+
"step": 795
|
| 5599 |
+
},
|
| 5600 |
+
{
|
| 5601 |
+
"epoch": 0.45291607396870553,
|
| 5602 |
+
"grad_norm": 0.5683977603912354,
|
| 5603 |
+
"learning_rate": 7.857186020912647e-05,
|
| 5604 |
+
"loss": 1.4482,
|
| 5605 |
+
"step": 796
|
| 5606 |
+
},
|
| 5607 |
+
{
|
| 5608 |
+
"epoch": 0.4534850640113798,
|
| 5609 |
+
"grad_norm": 0.5802903771400452,
|
| 5610 |
+
"learning_rate": 7.834987382307861e-05,
|
| 5611 |
+
"loss": 1.5827,
|
| 5612 |
+
"step": 797
|
| 5613 |
+
},
|
| 5614 |
+
{
|
| 5615 |
+
"epoch": 0.4540540540540541,
|
| 5616 |
+
"grad_norm": 0.5780924558639526,
|
| 5617 |
+
"learning_rate": 7.812799931482721e-05,
|
| 5618 |
+
"loss": 1.4595,
|
| 5619 |
+
"step": 798
|
| 5620 |
+
},
|
| 5621 |
+
{
|
| 5622 |
+
"epoch": 0.4546230440967283,
|
| 5623 |
+
"grad_norm": 0.5929847359657288,
|
| 5624 |
+
"learning_rate": 7.790623783091677e-05,
|
| 5625 |
+
"loss": 1.5512,
|
| 5626 |
+
"step": 799
|
| 5627 |
+
},
|
| 5628 |
+
{
|
| 5629 |
+
"epoch": 0.45519203413940257,
|
| 5630 |
+
"grad_norm": 0.5519236326217651,
|
| 5631 |
+
"learning_rate": 7.768459051730752e-05,
|
| 5632 |
+
"loss": 1.4239,
|
| 5633 |
+
"step": 800
|
| 5634 |
+
},
|
| 5635 |
+
{
|
| 5636 |
+
"epoch": 0.4557610241820768,
|
| 5637 |
+
"grad_norm": 0.5426004528999329,
|
| 5638 |
+
"learning_rate": 7.74630585193699e-05,
|
| 5639 |
+
"loss": 1.3005,
|
| 5640 |
+
"step": 801
|
| 5641 |
+
},
|
| 5642 |
+
{
|
| 5643 |
+
"epoch": 0.45633001422475106,
|
| 5644 |
+
"grad_norm": 0.6065943241119385,
|
| 5645 |
+
"learning_rate": 7.724164298187838e-05,
|
| 5646 |
+
"loss": 1.3966,
|
| 5647 |
+
"step": 802
|
| 5648 |
+
},
|
| 5649 |
+
{
|
| 5650 |
+
"epoch": 0.4568990042674253,
|
| 5651 |
+
"grad_norm": 0.5971605777740479,
|
| 5652 |
+
"learning_rate": 7.70203450490056e-05,
|
| 5653 |
+
"loss": 1.5944,
|
| 5654 |
+
"step": 803
|
| 5655 |
+
},
|
| 5656 |
+
{
|
| 5657 |
+
"epoch": 0.45746799431009955,
|
| 5658 |
+
"grad_norm": 0.5548596978187561,
|
| 5659 |
+
"learning_rate": 7.679916586431654e-05,
|
| 5660 |
+
"loss": 1.4323,
|
| 5661 |
+
"step": 804
|
| 5662 |
+
},
|
| 5663 |
+
{
|
| 5664 |
+
"epoch": 0.45803698435277385,
|
| 5665 |
+
"grad_norm": 0.5478107929229736,
|
| 5666 |
+
"learning_rate": 7.657810657076243e-05,
|
| 5667 |
+
"loss": 1.3819,
|
| 5668 |
+
"step": 805
|
| 5669 |
+
},
|
| 5670 |
+
{
|
| 5671 |
+
"epoch": 0.4586059743954481,
|
| 5672 |
+
"grad_norm": 0.5837447047233582,
|
| 5673 |
+
"learning_rate": 7.635716831067505e-05,
|
| 5674 |
+
"loss": 1.3941,
|
| 5675 |
+
"step": 806
|
| 5676 |
+
},
|
| 5677 |
+
{
|
| 5678 |
+
"epoch": 0.45917496443812233,
|
| 5679 |
+
"grad_norm": 0.5920546650886536,
|
| 5680 |
+
"learning_rate": 7.613635222576072e-05,
|
| 5681 |
+
"loss": 1.5395,
|
| 5682 |
+
"step": 807
|
| 5683 |
+
},
|
| 5684 |
+
{
|
| 5685 |
+
"epoch": 0.4597439544807966,
|
| 5686 |
+
"grad_norm": 0.6047683358192444,
|
| 5687 |
+
"learning_rate": 7.59156594570944e-05,
|
| 5688 |
+
"loss": 1.4169,
|
| 5689 |
+
"step": 808
|
| 5690 |
+
},
|
| 5691 |
+
{
|
| 5692 |
+
"epoch": 0.4603129445234708,
|
| 5693 |
+
"grad_norm": 0.5774646401405334,
|
| 5694 |
+
"learning_rate": 7.569509114511386e-05,
|
| 5695 |
+
"loss": 1.5108,
|
| 5696 |
+
"step": 809
|
| 5697 |
+
},
|
| 5698 |
+
{
|
| 5699 |
+
"epoch": 0.46088193456614507,
|
| 5700 |
+
"grad_norm": 0.5855366587638855,
|
| 5701 |
+
"learning_rate": 7.547464842961362e-05,
|
| 5702 |
+
"loss": 1.6545,
|
| 5703 |
+
"step": 810
|
| 5704 |
+
},
|
| 5705 |
+
{
|
| 5706 |
+
"epoch": 0.46145092460881937,
|
| 5707 |
+
"grad_norm": 0.5752539038658142,
|
| 5708 |
+
"learning_rate": 7.52543324497393e-05,
|
| 5709 |
+
"loss": 1.6431,
|
| 5710 |
+
"step": 811
|
| 5711 |
+
},
|
| 5712 |
+
{
|
| 5713 |
+
"epoch": 0.4620199146514936,
|
| 5714 |
+
"grad_norm": 0.5689989328384399,
|
| 5715 |
+
"learning_rate": 7.503414434398151e-05,
|
| 5716 |
+
"loss": 1.2883,
|
| 5717 |
+
"step": 812
|
| 5718 |
+
},
|
| 5719 |
+
{
|
| 5720 |
+
"epoch": 0.46258890469416786,
|
| 5721 |
+
"grad_norm": 0.6341901421546936,
|
| 5722 |
+
"learning_rate": 7.481408525017013e-05,
|
| 5723 |
+
"loss": 1.4223,
|
| 5724 |
+
"step": 813
|
| 5725 |
+
},
|
| 5726 |
+
{
|
| 5727 |
+
"epoch": 0.4631578947368421,
|
| 5728 |
+
"grad_norm": 0.6005092263221741,
|
| 5729 |
+
"learning_rate": 7.459415630546842e-05,
|
| 5730 |
+
"loss": 1.5522,
|
| 5731 |
+
"step": 814
|
| 5732 |
+
},
|
| 5733 |
+
{
|
| 5734 |
+
"epoch": 0.46372688477951635,
|
| 5735 |
+
"grad_norm": 0.6249240636825562,
|
| 5736 |
+
"learning_rate": 7.437435864636691e-05,
|
| 5737 |
+
"loss": 1.5459,
|
| 5738 |
+
"step": 815
|
| 5739 |
+
},
|
| 5740 |
+
{
|
| 5741 |
+
"epoch": 0.4642958748221906,
|
| 5742 |
+
"grad_norm": 0.5745651125907898,
|
| 5743 |
+
"learning_rate": 7.415469340867787e-05,
|
| 5744 |
+
"loss": 1.6287,
|
| 5745 |
+
"step": 816
|
| 5746 |
+
},
|
| 5747 |
+
{
|
| 5748 |
+
"epoch": 0.4648648648648649,
|
| 5749 |
+
"grad_norm": 0.5915263891220093,
|
| 5750 |
+
"learning_rate": 7.393516172752919e-05,
|
| 5751 |
+
"loss": 1.4738,
|
| 5752 |
+
"step": 817
|
| 5753 |
+
},
|
| 5754 |
+
{
|
| 5755 |
+
"epoch": 0.46543385490753914,
|
| 5756 |
+
"grad_norm": 0.5895527601242065,
|
| 5757 |
+
"learning_rate": 7.371576473735867e-05,
|
| 5758 |
+
"loss": 1.6939,
|
| 5759 |
+
"step": 818
|
| 5760 |
+
},
|
| 5761 |
+
{
|
| 5762 |
+
"epoch": 0.4660028449502134,
|
| 5763 |
+
"grad_norm": 0.5770692825317383,
|
| 5764 |
+
"learning_rate": 7.349650357190807e-05,
|
| 5765 |
+
"loss": 1.4264,
|
| 5766 |
+
"step": 819
|
| 5767 |
+
},
|
| 5768 |
+
{
|
| 5769 |
+
"epoch": 0.4665718349928876,
|
| 5770 |
+
"grad_norm": 0.6085241436958313,
|
| 5771 |
+
"learning_rate": 7.327737936421721e-05,
|
| 5772 |
+
"loss": 1.5019,
|
| 5773 |
+
"step": 820
|
| 5774 |
+
},
|
| 5775 |
+
{
|
| 5776 |
+
"epoch": 0.46714082503556187,
|
| 5777 |
+
"grad_norm": 0.5652032494544983,
|
| 5778 |
+
"learning_rate": 7.305839324661823e-05,
|
| 5779 |
+
"loss": 1.3324,
|
| 5780 |
+
"step": 821
|
| 5781 |
+
},
|
| 5782 |
+
{
|
| 5783 |
+
"epoch": 0.4677098150782361,
|
| 5784 |
+
"grad_norm": 0.5609267950057983,
|
| 5785 |
+
"learning_rate": 7.283954635072968e-05,
|
| 5786 |
+
"loss": 1.3902,
|
| 5787 |
+
"step": 822
|
| 5788 |
+
},
|
| 5789 |
+
{
|
| 5790 |
+
"epoch": 0.46827880512091036,
|
| 5791 |
+
"grad_norm": 0.5592348575592041,
|
| 5792 |
+
"learning_rate": 7.262083980745069e-05,
|
| 5793 |
+
"loss": 1.4362,
|
| 5794 |
+
"step": 823
|
| 5795 |
+
},
|
| 5796 |
+
{
|
| 5797 |
+
"epoch": 0.46884779516358466,
|
| 5798 |
+
"grad_norm": 0.5790618658065796,
|
| 5799 |
+
"learning_rate": 7.240227474695509e-05,
|
| 5800 |
+
"loss": 1.4753,
|
| 5801 |
+
"step": 824
|
| 5802 |
+
},
|
| 5803 |
+
{
|
| 5804 |
+
"epoch": 0.4694167852062589,
|
| 5805 |
+
"grad_norm": 0.5804809927940369,
|
| 5806 |
+
"learning_rate": 7.218385229868559e-05,
|
| 5807 |
+
"loss": 1.2719,
|
| 5808 |
+
"step": 825
|
| 5809 |
+
},
|
| 5810 |
+
{
|
| 5811 |
+
"epoch": 0.46998577524893315,
|
| 5812 |
+
"grad_norm": 0.5487887859344482,
|
| 5813 |
+
"learning_rate": 7.196557359134794e-05,
|
| 5814 |
+
"loss": 1.3212,
|
| 5815 |
+
"step": 826
|
| 5816 |
+
},
|
| 5817 |
+
{
|
| 5818 |
+
"epoch": 0.4705547652916074,
|
| 5819 |
+
"grad_norm": 0.5842025876045227,
|
| 5820 |
+
"learning_rate": 7.174743975290513e-05,
|
| 5821 |
+
"loss": 1.5622,
|
| 5822 |
+
"step": 827
|
| 5823 |
+
},
|
| 5824 |
+
{
|
| 5825 |
+
"epoch": 0.47112375533428164,
|
| 5826 |
+
"grad_norm": 0.580644428730011,
|
| 5827 |
+
"learning_rate": 7.152945191057154e-05,
|
| 5828 |
+
"loss": 1.4567,
|
| 5829 |
+
"step": 828
|
| 5830 |
+
},
|
| 5831 |
+
{
|
| 5832 |
+
"epoch": 0.4716927453769559,
|
| 5833 |
+
"grad_norm": 0.5735095739364624,
|
| 5834 |
+
"learning_rate": 7.131161119080712e-05,
|
| 5835 |
+
"loss": 1.4547,
|
| 5836 |
+
"step": 829
|
| 5837 |
+
},
|
| 5838 |
+
{
|
| 5839 |
+
"epoch": 0.4722617354196302,
|
| 5840 |
+
"grad_norm": 0.5592243671417236,
|
| 5841 |
+
"learning_rate": 7.109391871931142e-05,
|
| 5842 |
+
"loss": 1.3144,
|
| 5843 |
+
"step": 830
|
| 5844 |
+
},
|
| 5845 |
+
{
|
| 5846 |
+
"epoch": 0.4728307254623044,
|
| 5847 |
+
"grad_norm": 0.581495463848114,
|
| 5848 |
+
"learning_rate": 7.087637562101813e-05,
|
| 5849 |
+
"loss": 1.5145,
|
| 5850 |
+
"step": 831
|
| 5851 |
+
},
|
| 5852 |
+
{
|
| 5853 |
+
"epoch": 0.47339971550497867,
|
| 5854 |
+
"grad_norm": 0.5653107762336731,
|
| 5855 |
+
"learning_rate": 7.065898302008886e-05,
|
| 5856 |
+
"loss": 1.388,
|
| 5857 |
+
"step": 832
|
| 5858 |
+
},
|
| 5859 |
+
{
|
| 5860 |
+
"epoch": 0.4739687055476529,
|
| 5861 |
+
"grad_norm": 0.5776169300079346,
|
| 5862 |
+
"learning_rate": 7.04417420399077e-05,
|
| 5863 |
+
"loss": 1.5059,
|
| 5864 |
+
"step": 833
|
| 5865 |
+
},
|
| 5866 |
+
{
|
| 5867 |
+
"epoch": 0.47453769559032716,
|
| 5868 |
+
"grad_norm": 0.556419312953949,
|
| 5869 |
+
"learning_rate": 7.02246538030751e-05,
|
| 5870 |
+
"loss": 1.3933,
|
| 5871 |
+
"step": 834
|
| 5872 |
+
},
|
| 5873 |
+
{
|
| 5874 |
+
"epoch": 0.4751066856330014,
|
| 5875 |
+
"grad_norm": 0.5605750679969788,
|
| 5876 |
+
"learning_rate": 7.000771943140218e-05,
|
| 5877 |
+
"loss": 1.4677,
|
| 5878 |
+
"step": 835
|
| 5879 |
+
},
|
| 5880 |
+
{
|
| 5881 |
+
"epoch": 0.4756756756756757,
|
| 5882 |
+
"grad_norm": 0.5609278678894043,
|
| 5883 |
+
"learning_rate": 6.979094004590507e-05,
|
| 5884 |
+
"loss": 1.4526,
|
| 5885 |
+
"step": 836
|
| 5886 |
+
},
|
| 5887 |
+
{
|
| 5888 |
+
"epoch": 0.47624466571834995,
|
| 5889 |
+
"grad_norm": 0.5990177392959595,
|
| 5890 |
+
"learning_rate": 6.957431676679896e-05,
|
| 5891 |
+
"loss": 1.6215,
|
| 5892 |
+
"step": 837
|
| 5893 |
+
},
|
| 5894 |
+
{
|
| 5895 |
+
"epoch": 0.4768136557610242,
|
| 5896 |
+
"grad_norm": 0.5737520456314087,
|
| 5897 |
+
"learning_rate": 6.935785071349228e-05,
|
| 5898 |
+
"loss": 1.3985,
|
| 5899 |
+
"step": 838
|
| 5900 |
+
},
|
| 5901 |
+
{
|
| 5902 |
+
"epoch": 0.47738264580369844,
|
| 5903 |
+
"grad_norm": 0.5521170496940613,
|
| 5904 |
+
"learning_rate": 6.914154300458115e-05,
|
| 5905 |
+
"loss": 1.6527,
|
| 5906 |
+
"step": 839
|
| 5907 |
+
},
|
| 5908 |
+
{
|
| 5909 |
+
"epoch": 0.4779516358463727,
|
| 5910 |
+
"grad_norm": 0.5809024572372437,
|
| 5911 |
+
"learning_rate": 6.892539475784326e-05,
|
| 5912 |
+
"loss": 1.5697,
|
| 5913 |
+
"step": 840
|
| 5914 |
+
},
|
| 5915 |
+
{
|
| 5916 |
+
"epoch": 0.4785206258890469,
|
| 5917 |
+
"grad_norm": 0.6158897876739502,
|
| 5918 |
+
"learning_rate": 6.870940709023237e-05,
|
| 5919 |
+
"loss": 1.48,
|
| 5920 |
+
"step": 841
|
| 5921 |
+
},
|
| 5922 |
+
{
|
| 5923 |
+
"epoch": 0.47908961593172117,
|
| 5924 |
+
"grad_norm": 0.5950735807418823,
|
| 5925 |
+
"learning_rate": 6.849358111787246e-05,
|
| 5926 |
+
"loss": 1.3335,
|
| 5927 |
+
"step": 842
|
| 5928 |
+
},
|
| 5929 |
+
{
|
| 5930 |
+
"epoch": 0.47965860597439547,
|
| 5931 |
+
"grad_norm": 0.5788929462432861,
|
| 5932 |
+
"learning_rate": 6.82779179560519e-05,
|
| 5933 |
+
"loss": 1.4746,
|
| 5934 |
+
"step": 843
|
| 5935 |
+
},
|
| 5936 |
+
{
|
| 5937 |
+
"epoch": 0.4802275960170697,
|
| 5938 |
+
"grad_norm": 0.6169467568397522,
|
| 5939 |
+
"learning_rate": 6.806241871921777e-05,
|
| 5940 |
+
"loss": 1.2997,
|
| 5941 |
+
"step": 844
|
| 5942 |
+
},
|
| 5943 |
+
{
|
| 5944 |
+
"epoch": 0.48079658605974396,
|
| 5945 |
+
"grad_norm": 0.5850261449813843,
|
| 5946 |
+
"learning_rate": 6.784708452096998e-05,
|
| 5947 |
+
"loss": 1.2293,
|
| 5948 |
+
"step": 845
|
| 5949 |
+
},
|
| 5950 |
+
{
|
| 5951 |
+
"epoch": 0.4813655761024182,
|
| 5952 |
+
"grad_norm": 0.5514947772026062,
|
| 5953 |
+
"learning_rate": 6.763191647405568e-05,
|
| 5954 |
+
"loss": 1.3825,
|
| 5955 |
+
"step": 846
|
| 5956 |
+
},
|
| 5957 |
+
{
|
| 5958 |
+
"epoch": 0.48193456614509245,
|
| 5959 |
+
"grad_norm": 0.5753430128097534,
|
| 5960 |
+
"learning_rate": 6.741691569036338e-05,
|
| 5961 |
+
"loss": 1.5195,
|
| 5962 |
+
"step": 847
|
| 5963 |
+
},
|
| 5964 |
+
{
|
| 5965 |
+
"epoch": 0.4825035561877667,
|
| 5966 |
+
"grad_norm": 0.5876197814941406,
|
| 5967 |
+
"learning_rate": 6.720208328091732e-05,
|
| 5968 |
+
"loss": 1.4453,
|
| 5969 |
+
"step": 848
|
| 5970 |
+
},
|
| 5971 |
+
{
|
| 5972 |
+
"epoch": 0.483072546230441,
|
| 5973 |
+
"grad_norm": 0.5744032859802246,
|
| 5974 |
+
"learning_rate": 6.69874203558716e-05,
|
| 5975 |
+
"loss": 1.4914,
|
| 5976 |
+
"step": 849
|
| 5977 |
+
},
|
| 5978 |
+
{
|
| 5979 |
+
"epoch": 0.48364153627311524,
|
| 5980 |
+
"grad_norm": 0.5800637006759644,
|
| 5981 |
+
"learning_rate": 6.677292802450447e-05,
|
| 5982 |
+
"loss": 1.4932,
|
| 5983 |
+
"step": 850
|
| 5984 |
+
},
|
| 5985 |
+
{
|
| 5986 |
+
"epoch": 0.4842105263157895,
|
| 5987 |
+
"grad_norm": 0.5554024577140808,
|
| 5988 |
+
"learning_rate": 6.655860739521271e-05,
|
| 5989 |
+
"loss": 1.1795,
|
| 5990 |
+
"step": 851
|
| 5991 |
+
},
|
| 5992 |
+
{
|
| 5993 |
+
"epoch": 0.4847795163584637,
|
| 5994 |
+
"grad_norm": 0.5711913704872131,
|
| 5995 |
+
"learning_rate": 6.634445957550577e-05,
|
| 5996 |
+
"loss": 1.486,
|
| 5997 |
+
"step": 852
|
| 5998 |
+
},
|
| 5999 |
+
{
|
| 6000 |
+
"epoch": 0.48534850640113797,
|
| 6001 |
+
"grad_norm": 0.5684107542037964,
|
| 6002 |
+
"learning_rate": 6.613048567200013e-05,
|
| 6003 |
+
"loss": 1.3984,
|
| 6004 |
+
"step": 853
|
| 6005 |
+
},
|
| 6006 |
+
{
|
| 6007 |
+
"epoch": 0.4859174964438122,
|
| 6008 |
+
"grad_norm": 0.5672001242637634,
|
| 6009 |
+
"learning_rate": 6.591668679041359e-05,
|
| 6010 |
+
"loss": 1.4811,
|
| 6011 |
+
"step": 854
|
| 6012 |
+
},
|
| 6013 |
+
{
|
| 6014 |
+
"epoch": 0.4864864864864865,
|
| 6015 |
+
"grad_norm": 0.5804989337921143,
|
| 6016 |
+
"learning_rate": 6.570306403555937e-05,
|
| 6017 |
+
"loss": 1.3624,
|
| 6018 |
+
"step": 855
|
| 6019 |
+
},
|
| 6020 |
+
{
|
| 6021 |
+
"epoch": 0.48705547652916076,
|
| 6022 |
+
"grad_norm": 0.6067745089530945,
|
| 6023 |
+
"learning_rate": 6.548961851134072e-05,
|
| 6024 |
+
"loss": 1.4192,
|
| 6025 |
+
"step": 856
|
| 6026 |
+
},
|
| 6027 |
+
{
|
| 6028 |
+
"epoch": 0.487624466571835,
|
| 6029 |
+
"grad_norm": 0.576329231262207,
|
| 6030 |
+
"learning_rate": 6.527635132074493e-05,
|
| 6031 |
+
"loss": 1.6314,
|
| 6032 |
+
"step": 857
|
| 6033 |
+
},
|
| 6034 |
+
{
|
| 6035 |
+
"epoch": 0.48819345661450925,
|
| 6036 |
+
"grad_norm": 0.5863393545150757,
|
| 6037 |
+
"learning_rate": 6.506326356583781e-05,
|
| 6038 |
+
"loss": 1.5669,
|
| 6039 |
+
"step": 858
|
| 6040 |
+
},
|
| 6041 |
+
{
|
| 6042 |
+
"epoch": 0.4887624466571835,
|
| 6043 |
+
"grad_norm": 0.6074771285057068,
|
| 6044 |
+
"learning_rate": 6.485035634775796e-05,
|
| 6045 |
+
"loss": 1.3334,
|
| 6046 |
+
"step": 859
|
| 6047 |
+
},
|
| 6048 |
+
{
|
| 6049 |
+
"epoch": 0.48933143669985774,
|
| 6050 |
+
"grad_norm": 0.5837851166725159,
|
| 6051 |
+
"learning_rate": 6.463763076671091e-05,
|
| 6052 |
+
"loss": 1.607,
|
| 6053 |
+
"step": 860
|
| 6054 |
+
},
|
| 6055 |
+
{
|
| 6056 |
+
"epoch": 0.489900426742532,
|
| 6057 |
+
"grad_norm": 0.5989742875099182,
|
| 6058 |
+
"learning_rate": 6.442508792196369e-05,
|
| 6059 |
+
"loss": 1.4518,
|
| 6060 |
+
"step": 861
|
| 6061 |
+
},
|
| 6062 |
+
{
|
| 6063 |
+
"epoch": 0.4904694167852063,
|
| 6064 |
+
"grad_norm": 0.5692201852798462,
|
| 6065 |
+
"learning_rate": 6.4212728911839e-05,
|
| 6066 |
+
"loss": 1.3878,
|
| 6067 |
+
"step": 862
|
| 6068 |
+
},
|
| 6069 |
+
{
|
| 6070 |
+
"epoch": 0.4910384068278805,
|
| 6071 |
+
"grad_norm": 0.6134719252586365,
|
| 6072 |
+
"learning_rate": 6.400055483370957e-05,
|
| 6073 |
+
"loss": 1.5154,
|
| 6074 |
+
"step": 863
|
| 6075 |
+
},
|
| 6076 |
+
{
|
| 6077 |
+
"epoch": 0.49160739687055477,
|
| 6078 |
+
"grad_norm": 0.5494038462638855,
|
| 6079 |
+
"learning_rate": 6.378856678399255e-05,
|
| 6080 |
+
"loss": 1.2968,
|
| 6081 |
+
"step": 864
|
| 6082 |
+
},
|
| 6083 |
+
{
|
| 6084 |
+
"epoch": 0.492176386913229,
|
| 6085 |
+
"grad_norm": 0.5780492424964905,
|
| 6086 |
+
"learning_rate": 6.357676585814366e-05,
|
| 6087 |
+
"loss": 1.5766,
|
| 6088 |
+
"step": 865
|
| 6089 |
+
},
|
| 6090 |
+
{
|
| 6091 |
+
"epoch": 0.49274537695590326,
|
| 6092 |
+
"grad_norm": 0.5398704409599304,
|
| 6093 |
+
"learning_rate": 6.336515315065168e-05,
|
| 6094 |
+
"loss": 1.4446,
|
| 6095 |
+
"step": 866
|
| 6096 |
+
},
|
| 6097 |
+
{
|
| 6098 |
+
"epoch": 0.4933143669985775,
|
| 6099 |
+
"grad_norm": 0.5509852170944214,
|
| 6100 |
+
"learning_rate": 6.315372975503285e-05,
|
| 6101 |
+
"loss": 1.4465,
|
| 6102 |
+
"step": 867
|
| 6103 |
+
},
|
| 6104 |
+
{
|
| 6105 |
+
"epoch": 0.4938833570412518,
|
| 6106 |
+
"grad_norm": 0.6671035885810852,
|
| 6107 |
+
"learning_rate": 6.294249676382508e-05,
|
| 6108 |
+
"loss": 1.706,
|
| 6109 |
+
"step": 868
|
| 6110 |
+
},
|
| 6111 |
+
{
|
| 6112 |
+
"epoch": 0.49445234708392605,
|
| 6113 |
+
"grad_norm": 0.579408586025238,
|
| 6114 |
+
"learning_rate": 6.273145526858236e-05,
|
| 6115 |
+
"loss": 1.5695,
|
| 6116 |
+
"step": 869
|
| 6117 |
+
},
|
| 6118 |
+
{
|
| 6119 |
+
"epoch": 0.4950213371266003,
|
| 6120 |
+
"grad_norm": 0.571058988571167,
|
| 6121 |
+
"learning_rate": 6.252060635986911e-05,
|
| 6122 |
+
"loss": 1.3541,
|
| 6123 |
+
"step": 870
|
| 6124 |
+
},
|
| 6125 |
+
{
|
| 6126 |
+
"epoch": 0.49559032716927454,
|
| 6127 |
+
"grad_norm": 0.5792422890663147,
|
| 6128 |
+
"learning_rate": 6.230995112725454e-05,
|
| 6129 |
+
"loss": 1.4329,
|
| 6130 |
+
"step": 871
|
| 6131 |
+
},
|
| 6132 |
+
{
|
| 6133 |
+
"epoch": 0.4961593172119488,
|
| 6134 |
+
"grad_norm": 0.5893927216529846,
|
| 6135 |
+
"learning_rate": 6.209949065930706e-05,
|
| 6136 |
+
"loss": 1.4674,
|
| 6137 |
+
"step": 872
|
| 6138 |
+
},
|
| 6139 |
+
{
|
| 6140 |
+
"epoch": 0.496728307254623,
|
| 6141 |
+
"grad_norm": 0.5954142212867737,
|
| 6142 |
+
"learning_rate": 6.188922604358865e-05,
|
| 6143 |
+
"loss": 1.4462,
|
| 6144 |
+
"step": 873
|
| 6145 |
+
},
|
| 6146 |
+
{
|
| 6147 |
+
"epoch": 0.4972972972972973,
|
| 6148 |
+
"grad_norm": 0.6741952896118164,
|
| 6149 |
+
"learning_rate": 6.16791583666492e-05,
|
| 6150 |
+
"loss": 1.6458,
|
| 6151 |
+
"step": 874
|
| 6152 |
+
},
|
| 6153 |
+
{
|
| 6154 |
+
"epoch": 0.49786628733997157,
|
| 6155 |
+
"grad_norm": 0.6125763654708862,
|
| 6156 |
+
"learning_rate": 6.146928871402081e-05,
|
| 6157 |
+
"loss": 1.5387,
|
| 6158 |
+
"step": 875
|
| 6159 |
+
},
|
| 6160 |
+
{
|
| 6161 |
+
"epoch": 0.4984352773826458,
|
| 6162 |
+
"grad_norm": 0.5839952230453491,
|
| 6163 |
+
"learning_rate": 6.12596181702124e-05,
|
| 6164 |
+
"loss": 1.6821,
|
| 6165 |
+
"step": 876
|
| 6166 |
+
},
|
| 6167 |
+
{
|
| 6168 |
+
"epoch": 0.49900426742532006,
|
| 6169 |
+
"grad_norm": 0.5859706401824951,
|
| 6170 |
+
"learning_rate": 6.1050147818704e-05,
|
| 6171 |
+
"loss": 1.4713,
|
| 6172 |
+
"step": 877
|
| 6173 |
+
},
|
| 6174 |
+
{
|
| 6175 |
+
"epoch": 0.4995732574679943,
|
| 6176 |
+
"grad_norm": 0.5910811424255371,
|
| 6177 |
+
"learning_rate": 6.0840878741941057e-05,
|
| 6178 |
+
"loss": 1.59,
|
| 6179 |
+
"step": 878
|
| 6180 |
+
},
|
| 6181 |
+
{
|
| 6182 |
+
"epoch": 0.5001422475106686,
|
| 6183 |
+
"grad_norm": 0.6297405958175659,
|
| 6184 |
+
"learning_rate": 6.063181202132901e-05,
|
| 6185 |
+
"loss": 1.5881,
|
| 6186 |
+
"step": 879
|
| 6187 |
+
},
|
| 6188 |
+
{
|
| 6189 |
+
"epoch": 0.5007112375533428,
|
| 6190 |
+
"grad_norm": 0.5714183449745178,
|
| 6191 |
+
"learning_rate": 6.0422948737227504e-05,
|
| 6192 |
+
"loss": 1.5894,
|
| 6193 |
+
"step": 880
|
| 6194 |
+
},
|
| 6195 |
+
{
|
| 6196 |
+
"epoch": 0.5012802275960171,
|
| 6197 |
+
"grad_norm": 0.5969492197036743,
|
| 6198 |
+
"learning_rate": 6.0214289968945004e-05,
|
| 6199 |
+
"loss": 1.6697,
|
| 6200 |
+
"step": 881
|
| 6201 |
+
},
|
| 6202 |
+
{
|
| 6203 |
+
"epoch": 0.5018492176386913,
|
| 6204 |
+
"grad_norm": 0.5817530155181885,
|
| 6205 |
+
"learning_rate": 6.000583679473315e-05,
|
| 6206 |
+
"loss": 1.5806,
|
| 6207 |
+
"step": 882
|
| 6208 |
+
},
|
| 6209 |
+
{
|
| 6210 |
+
"epoch": 0.5024182076813656,
|
| 6211 |
+
"grad_norm": 0.5869944095611572,
|
| 6212 |
+
"learning_rate": 5.979759029178107e-05,
|
| 6213 |
+
"loss": 1.4565,
|
| 6214 |
+
"step": 883
|
| 6215 |
+
},
|
| 6216 |
+
{
|
| 6217 |
+
"epoch": 0.5029871977240399,
|
| 6218 |
+
"grad_norm": 0.5745888948440552,
|
| 6219 |
+
"learning_rate": 5.958955153621004e-05,
|
| 6220 |
+
"loss": 1.5645,
|
| 6221 |
+
"step": 884
|
| 6222 |
+
},
|
| 6223 |
+
{
|
| 6224 |
+
"epoch": 0.5035561877667141,
|
| 6225 |
+
"grad_norm": 0.549628734588623,
|
| 6226 |
+
"learning_rate": 5.938172160306765e-05,
|
| 6227 |
+
"loss": 1.5017,
|
| 6228 |
+
"step": 885
|
| 6229 |
+
},
|
| 6230 |
+
{
|
| 6231 |
+
"epoch": 0.5041251778093884,
|
| 6232 |
+
"grad_norm": 0.5471094250679016,
|
| 6233 |
+
"learning_rate": 5.9174101566322504e-05,
|
| 6234 |
+
"loss": 1.2781,
|
| 6235 |
+
"step": 886
|
| 6236 |
+
},
|
| 6237 |
+
{
|
| 6238 |
+
"epoch": 0.5046941678520626,
|
| 6239 |
+
"grad_norm": 0.5772054195404053,
|
| 6240 |
+
"learning_rate": 5.896669249885851e-05,
|
| 6241 |
+
"loss": 1.386,
|
| 6242 |
+
"step": 887
|
| 6243 |
+
},
|
| 6244 |
+
{
|
| 6245 |
+
"epoch": 0.5052631578947369,
|
| 6246 |
+
"grad_norm": 0.6143761873245239,
|
| 6247 |
+
"learning_rate": 5.875949547246939e-05,
|
| 6248 |
+
"loss": 1.5432,
|
| 6249 |
+
"step": 888
|
| 6250 |
+
},
|
| 6251 |
+
{
|
| 6252 |
+
"epoch": 0.505832147937411,
|
| 6253 |
+
"grad_norm": 0.5768917202949524,
|
| 6254 |
+
"learning_rate": 5.8552511557853204e-05,
|
| 6255 |
+
"loss": 1.6945,
|
| 6256 |
+
"step": 889
|
| 6257 |
+
},
|
| 6258 |
+
{
|
| 6259 |
+
"epoch": 0.5064011379800853,
|
| 6260 |
+
"grad_norm": 0.5644556283950806,
|
| 6261 |
+
"learning_rate": 5.8345741824606617e-05,
|
| 6262 |
+
"loss": 1.5163,
|
| 6263 |
+
"step": 890
|
| 6264 |
+
},
|
| 6265 |
+
{
|
| 6266 |
+
"epoch": 0.5069701280227596,
|
| 6267 |
+
"grad_norm": 0.6083329916000366,
|
| 6268 |
+
"learning_rate": 5.813918734121955e-05,
|
| 6269 |
+
"loss": 1.7979,
|
| 6270 |
+
"step": 891
|
| 6271 |
+
},
|
| 6272 |
+
{
|
| 6273 |
+
"epoch": 0.5075391180654338,
|
| 6274 |
+
"grad_norm": 0.5543102025985718,
|
| 6275 |
+
"learning_rate": 5.7932849175069705e-05,
|
| 6276 |
+
"loss": 1.5558,
|
| 6277 |
+
"step": 892
|
| 6278 |
+
},
|
| 6279 |
+
{
|
| 6280 |
+
"epoch": 0.5081081081081081,
|
| 6281 |
+
"grad_norm": 0.6090741753578186,
|
| 6282 |
+
"learning_rate": 5.7726728392416874e-05,
|
| 6283 |
+
"loss": 1.6233,
|
| 6284 |
+
"step": 893
|
| 6285 |
+
},
|
| 6286 |
+
{
|
| 6287 |
+
"epoch": 0.5086770981507823,
|
| 6288 |
+
"grad_norm": 0.556496798992157,
|
| 6289 |
+
"learning_rate": 5.7520826058397525e-05,
|
| 6290 |
+
"loss": 1.5755,
|
| 6291 |
+
"step": 894
|
| 6292 |
+
},
|
| 6293 |
+
{
|
| 6294 |
+
"epoch": 0.5092460881934566,
|
| 6295 |
+
"grad_norm": 0.6258504986763,
|
| 6296 |
+
"learning_rate": 5.731514323701927e-05,
|
| 6297 |
+
"loss": 1.6054,
|
| 6298 |
+
"step": 895
|
| 6299 |
+
},
|
| 6300 |
+
{
|
| 6301 |
+
"epoch": 0.5098150782361308,
|
| 6302 |
+
"grad_norm": 0.6283307671546936,
|
| 6303 |
+
"learning_rate": 5.7109680991155364e-05,
|
| 6304 |
+
"loss": 1.8276,
|
| 6305 |
+
"step": 896
|
| 6306 |
+
},
|
| 6307 |
+
{
|
| 6308 |
+
"epoch": 0.5103840682788051,
|
| 6309 |
+
"grad_norm": 0.5817832946777344,
|
| 6310 |
+
"learning_rate": 5.690444038253935e-05,
|
| 6311 |
+
"loss": 1.6388,
|
| 6312 |
+
"step": 897
|
| 6313 |
+
},
|
| 6314 |
+
{
|
| 6315 |
+
"epoch": 0.5109530583214794,
|
| 6316 |
+
"grad_norm": 0.5892955660820007,
|
| 6317 |
+
"learning_rate": 5.669942247175933e-05,
|
| 6318 |
+
"loss": 1.2641,
|
| 6319 |
+
"step": 898
|
| 6320 |
+
},
|
| 6321 |
+
{
|
| 6322 |
+
"epoch": 0.5115220483641536,
|
| 6323 |
+
"grad_norm": 0.5834968686103821,
|
| 6324 |
+
"learning_rate": 5.649462831825265e-05,
|
| 6325 |
+
"loss": 1.4207,
|
| 6326 |
+
"step": 899
|
| 6327 |
+
},
|
| 6328 |
+
{
|
| 6329 |
+
"epoch": 0.5120910384068279,
|
| 6330 |
+
"grad_norm": 0.5753495693206787,
|
| 6331 |
+
"learning_rate": 5.629005898030035e-05,
|
| 6332 |
+
"loss": 1.4724,
|
| 6333 |
+
"step": 900
|
| 6334 |
+
},
|
| 6335 |
+
{
|
| 6336 |
+
"epoch": 0.5126600284495021,
|
| 6337 |
+
"grad_norm": 0.6050419211387634,
|
| 6338 |
+
"learning_rate": 5.608571551502175e-05,
|
| 6339 |
+
"loss": 1.7189,
|
| 6340 |
+
"step": 901
|
| 6341 |
+
},
|
| 6342 |
+
{
|
| 6343 |
+
"epoch": 0.5132290184921764,
|
| 6344 |
+
"grad_norm": 0.5946124196052551,
|
| 6345 |
+
"learning_rate": 5.588159897836902e-05,
|
| 6346 |
+
"loss": 1.3803,
|
| 6347 |
+
"step": 902
|
| 6348 |
+
},
|
| 6349 |
+
{
|
| 6350 |
+
"epoch": 0.5137980085348507,
|
| 6351 |
+
"grad_norm": 0.5731397867202759,
|
| 6352 |
+
"learning_rate": 5.56777104251216e-05,
|
| 6353 |
+
"loss": 1.7426,
|
| 6354 |
+
"step": 903
|
| 6355 |
+
},
|
| 6356 |
+
{
|
| 6357 |
+
"epoch": 0.5143669985775249,
|
| 6358 |
+
"grad_norm": 0.5813397169113159,
|
| 6359 |
+
"learning_rate": 5.5474050908880814e-05,
|
| 6360 |
+
"loss": 1.4898,
|
| 6361 |
+
"step": 904
|
| 6362 |
+
},
|
| 6363 |
+
{
|
| 6364 |
+
"epoch": 0.5149359886201992,
|
| 6365 |
+
"grad_norm": 0.5610973834991455,
|
| 6366 |
+
"learning_rate": 5.5270621482064465e-05,
|
| 6367 |
+
"loss": 1.4937,
|
| 6368 |
+
"step": 905
|
| 6369 |
+
},
|
| 6370 |
+
{
|
| 6371 |
+
"epoch": 0.5155049786628734,
|
| 6372 |
+
"grad_norm": 0.5550079941749573,
|
| 6373 |
+
"learning_rate": 5.50674231959013e-05,
|
| 6374 |
+
"loss": 1.3543,
|
| 6375 |
+
"step": 906
|
| 6376 |
+
},
|
| 6377 |
+
{
|
| 6378 |
+
"epoch": 0.5160739687055477,
|
| 6379 |
+
"grad_norm": 0.596593976020813,
|
| 6380 |
+
"learning_rate": 5.4864457100425783e-05,
|
| 6381 |
+
"loss": 1.5856,
|
| 6382 |
+
"step": 907
|
| 6383 |
+
},
|
| 6384 |
+
{
|
| 6385 |
+
"epoch": 0.5166429587482219,
|
| 6386 |
+
"grad_norm": 0.6018926501274109,
|
| 6387 |
+
"learning_rate": 5.4661724244472355e-05,
|
| 6388 |
+
"loss": 1.5092,
|
| 6389 |
+
"step": 908
|
| 6390 |
+
},
|
| 6391 |
+
{
|
| 6392 |
+
"epoch": 0.5172119487908962,
|
| 6393 |
+
"grad_norm": 0.6650524735450745,
|
| 6394 |
+
"learning_rate": 5.4459225675670264e-05,
|
| 6395 |
+
"loss": 1.7059,
|
| 6396 |
+
"step": 909
|
| 6397 |
+
},
|
| 6398 |
+
{
|
| 6399 |
+
"epoch": 0.5177809388335705,
|
| 6400 |
+
"grad_norm": 0.5858013033866882,
|
| 6401 |
+
"learning_rate": 5.425696244043807e-05,
|
| 6402 |
+
"loss": 1.4591,
|
| 6403 |
+
"step": 910
|
| 6404 |
+
},
|
| 6405 |
+
{
|
| 6406 |
+
"epoch": 0.5183499288762446,
|
| 6407 |
+
"grad_norm": 0.555473268032074,
|
| 6408 |
+
"learning_rate": 5.405493558397824e-05,
|
| 6409 |
+
"loss": 1.401,
|
| 6410 |
+
"step": 911
|
| 6411 |
+
},
|
| 6412 |
+
{
|
| 6413 |
+
"epoch": 0.518918918918919,
|
| 6414 |
+
"grad_norm": 0.6246885061264038,
|
| 6415 |
+
"learning_rate": 5.385314615027168e-05,
|
| 6416 |
+
"loss": 1.4415,
|
| 6417 |
+
"step": 912
|
| 6418 |
+
},
|
| 6419 |
+
{
|
| 6420 |
+
"epoch": 0.5194879089615931,
|
| 6421 |
+
"grad_norm": 0.608062207698822,
|
| 6422 |
+
"learning_rate": 5.365159518207252e-05,
|
| 6423 |
+
"loss": 1.4239,
|
| 6424 |
+
"step": 913
|
| 6425 |
+
},
|
| 6426 |
+
{
|
| 6427 |
+
"epoch": 0.5200568990042674,
|
| 6428 |
+
"grad_norm": 0.5979565382003784,
|
| 6429 |
+
"learning_rate": 5.345028372090256e-05,
|
| 6430 |
+
"loss": 1.4656,
|
| 6431 |
+
"step": 914
|
| 6432 |
+
},
|
| 6433 |
+
{
|
| 6434 |
+
"epoch": 0.5206258890469416,
|
| 6435 |
+
"grad_norm": 0.6553084254264832,
|
| 6436 |
+
"learning_rate": 5.324921280704589e-05,
|
| 6437 |
+
"loss": 1.4609,
|
| 6438 |
+
"step": 915
|
| 6439 |
+
},
|
| 6440 |
+
{
|
| 6441 |
+
"epoch": 0.5211948790896159,
|
| 6442 |
+
"grad_norm": 0.5839146971702576,
|
| 6443 |
+
"learning_rate": 5.304838347954363e-05,
|
| 6444 |
+
"loss": 1.5546,
|
| 6445 |
+
"step": 916
|
| 6446 |
+
},
|
| 6447 |
+
{
|
| 6448 |
+
"epoch": 0.5217638691322902,
|
| 6449 |
+
"grad_norm": 0.5618466734886169,
|
| 6450 |
+
"learning_rate": 5.284779677618841e-05,
|
| 6451 |
+
"loss": 1.4078,
|
| 6452 |
+
"step": 917
|
| 6453 |
+
},
|
| 6454 |
+
{
|
| 6455 |
+
"epoch": 0.5223328591749644,
|
| 6456 |
+
"grad_norm": 0.6020224690437317,
|
| 6457 |
+
"learning_rate": 5.264745373351923e-05,
|
| 6458 |
+
"loss": 1.568,
|
| 6459 |
+
"step": 918
|
| 6460 |
+
},
|
| 6461 |
+
{
|
| 6462 |
+
"epoch": 0.5229018492176387,
|
| 6463 |
+
"grad_norm": 0.6049513220787048,
|
| 6464 |
+
"learning_rate": 5.244735538681584e-05,
|
| 6465 |
+
"loss": 1.3196,
|
| 6466 |
+
"step": 919
|
| 6467 |
+
},
|
| 6468 |
+
{
|
| 6469 |
+
"epoch": 0.5234708392603129,
|
| 6470 |
+
"grad_norm": 0.5781171917915344,
|
| 6471 |
+
"learning_rate": 5.224750277009358e-05,
|
| 6472 |
+
"loss": 1.5366,
|
| 6473 |
+
"step": 920
|
| 6474 |
+
},
|
| 6475 |
+
{
|
| 6476 |
+
"epoch": 0.5240398293029872,
|
| 6477 |
+
"grad_norm": 0.6478269696235657,
|
| 6478 |
+
"learning_rate": 5.204789691609793e-05,
|
| 6479 |
+
"loss": 1.5281,
|
| 6480 |
+
"step": 921
|
| 6481 |
+
},
|
| 6482 |
+
{
|
| 6483 |
+
"epoch": 0.5246088193456615,
|
| 6484 |
+
"grad_norm": 0.598915696144104,
|
| 6485 |
+
"learning_rate": 5.184853885629921e-05,
|
| 6486 |
+
"loss": 1.5734,
|
| 6487 |
+
"step": 922
|
| 6488 |
+
},
|
| 6489 |
+
{
|
| 6490 |
+
"epoch": 0.5251778093883357,
|
| 6491 |
+
"grad_norm": 0.589694619178772,
|
| 6492 |
+
"learning_rate": 5.1649429620887334e-05,
|
| 6493 |
+
"loss": 1.4307,
|
| 6494 |
+
"step": 923
|
| 6495 |
+
},
|
| 6496 |
+
{
|
| 6497 |
+
"epoch": 0.52574679943101,
|
| 6498 |
+
"grad_norm": 0.5483283996582031,
|
| 6499 |
+
"learning_rate": 5.145057023876634e-05,
|
| 6500 |
+
"loss": 1.4334,
|
| 6501 |
+
"step": 924
|
| 6502 |
+
},
|
| 6503 |
+
{
|
| 6504 |
+
"epoch": 0.5263157894736842,
|
| 6505 |
+
"grad_norm": 0.5908382534980774,
|
| 6506 |
+
"learning_rate": 5.125196173754914e-05,
|
| 6507 |
+
"loss": 1.588,
|
| 6508 |
+
"step": 925
|
| 6509 |
+
},
|
| 6510 |
+
{
|
| 6511 |
+
"epoch": 0.5268847795163585,
|
| 6512 |
+
"grad_norm": 0.5898739695549011,
|
| 6513 |
+
"learning_rate": 5.105360514355222e-05,
|
| 6514 |
+
"loss": 1.5685,
|
| 6515 |
+
"step": 926
|
| 6516 |
+
},
|
| 6517 |
+
{
|
| 6518 |
+
"epoch": 0.5274537695590327,
|
| 6519 |
+
"grad_norm": 0.6079673171043396,
|
| 6520 |
+
"learning_rate": 5.0855501481790305e-05,
|
| 6521 |
+
"loss": 1.4421,
|
| 6522 |
+
"step": 927
|
| 6523 |
+
},
|
| 6524 |
+
{
|
| 6525 |
+
"epoch": 0.528022759601707,
|
| 6526 |
+
"grad_norm": 0.5824552178382874,
|
| 6527 |
+
"learning_rate": 5.0657651775971146e-05,
|
| 6528 |
+
"loss": 1.3472,
|
| 6529 |
+
"step": 928
|
| 6530 |
+
},
|
| 6531 |
+
{
|
| 6532 |
+
"epoch": 0.5285917496443813,
|
| 6533 |
+
"grad_norm": 0.5997583866119385,
|
| 6534 |
+
"learning_rate": 5.046005704849015e-05,
|
| 6535 |
+
"loss": 1.6292,
|
| 6536 |
+
"step": 929
|
| 6537 |
+
},
|
| 6538 |
+
{
|
| 6539 |
+
"epoch": 0.5291607396870555,
|
| 6540 |
+
"grad_norm": 0.5740709900856018,
|
| 6541 |
+
"learning_rate": 5.026271832042506e-05,
|
| 6542 |
+
"loss": 1.4085,
|
| 6543 |
+
"step": 930
|
| 6544 |
+
},
|
| 6545 |
+
{
|
| 6546 |
+
"epoch": 0.5297297297297298,
|
| 6547 |
+
"grad_norm": 0.5683955550193787,
|
| 6548 |
+
"learning_rate": 5.0065636611530767e-05,
|
| 6549 |
+
"loss": 1.4722,
|
| 6550 |
+
"step": 931
|
| 6551 |
+
},
|
| 6552 |
+
{
|
| 6553 |
+
"epoch": 0.530298719772404,
|
| 6554 |
+
"grad_norm": 0.5909097790718079,
|
| 6555 |
+
"learning_rate": 4.986881294023397e-05,
|
| 6556 |
+
"loss": 1.5688,
|
| 6557 |
+
"step": 932
|
| 6558 |
+
},
|
| 6559 |
+
{
|
| 6560 |
+
"epoch": 0.5308677098150782,
|
| 6561 |
+
"grad_norm": 0.5723986029624939,
|
| 6562 |
+
"learning_rate": 4.967224832362807e-05,
|
| 6563 |
+
"loss": 1.718,
|
| 6564 |
+
"step": 933
|
| 6565 |
+
},
|
| 6566 |
+
{
|
| 6567 |
+
"epoch": 0.5314366998577524,
|
| 6568 |
+
"grad_norm": 0.6397773623466492,
|
| 6569 |
+
"learning_rate": 4.947594377746769e-05,
|
| 6570 |
+
"loss": 1.5896,
|
| 6571 |
+
"step": 934
|
| 6572 |
+
},
|
| 6573 |
+
{
|
| 6574 |
+
"epoch": 0.5320056899004267,
|
| 6575 |
+
"grad_norm": 0.6130902171134949,
|
| 6576 |
+
"learning_rate": 4.9279900316163466e-05,
|
| 6577 |
+
"loss": 1.5974,
|
| 6578 |
+
"step": 935
|
| 6579 |
+
},
|
| 6580 |
+
{
|
| 6581 |
+
"epoch": 0.532574679943101,
|
| 6582 |
+
"grad_norm": 0.5888193845748901,
|
| 6583 |
+
"learning_rate": 4.908411895277704e-05,
|
| 6584 |
+
"loss": 1.569,
|
| 6585 |
+
"step": 936
|
| 6586 |
+
},
|
| 6587 |
+
{
|
| 6588 |
+
"epoch": 0.5331436699857752,
|
| 6589 |
+
"grad_norm": 0.5966805219650269,
|
| 6590 |
+
"learning_rate": 4.8888600699015496e-05,
|
| 6591 |
+
"loss": 1.4014,
|
| 6592 |
+
"step": 937
|
| 6593 |
+
},
|
| 6594 |
+
{
|
| 6595 |
+
"epoch": 0.5337126600284495,
|
| 6596 |
+
"grad_norm": 0.6131336092948914,
|
| 6597 |
+
"learning_rate": 4.869334656522644e-05,
|
| 6598 |
+
"loss": 1.5619,
|
| 6599 |
+
"step": 938
|
| 6600 |
+
},
|
| 6601 |
+
{
|
| 6602 |
+
"epoch": 0.5342816500711237,
|
| 6603 |
+
"grad_norm": 0.5846887826919556,
|
| 6604 |
+
"learning_rate": 4.849835756039254e-05,
|
| 6605 |
+
"loss": 1.5674,
|
| 6606 |
+
"step": 939
|
| 6607 |
+
},
|
| 6608 |
+
{
|
| 6609 |
+
"epoch": 0.534850640113798,
|
| 6610 |
+
"grad_norm": 0.5879199504852295,
|
| 6611 |
+
"learning_rate": 4.830363469212631e-05,
|
| 6612 |
+
"loss": 1.6148,
|
| 6613 |
+
"step": 940
|
| 6614 |
+
},
|
| 6615 |
+
{
|
| 6616 |
+
"epoch": 0.5354196301564723,
|
| 6617 |
+
"grad_norm": 0.6081675887107849,
|
| 6618 |
+
"learning_rate": 4.8109178966665194e-05,
|
| 6619 |
+
"loss": 1.5329,
|
| 6620 |
+
"step": 941
|
| 6621 |
+
},
|
| 6622 |
+
{
|
| 6623 |
+
"epoch": 0.5359886201991465,
|
| 6624 |
+
"grad_norm": 0.5982802510261536,
|
| 6625 |
+
"learning_rate": 4.791499138886603e-05,
|
| 6626 |
+
"loss": 1.5198,
|
| 6627 |
+
"step": 942
|
| 6628 |
+
},
|
| 6629 |
+
{
|
| 6630 |
+
"epoch": 0.5365576102418208,
|
| 6631 |
+
"grad_norm": 0.5899128913879395,
|
| 6632 |
+
"learning_rate": 4.7721072962199975e-05,
|
| 6633 |
+
"loss": 1.331,
|
| 6634 |
+
"step": 943
|
| 6635 |
+
},
|
| 6636 |
+
{
|
| 6637 |
+
"epoch": 0.537126600284495,
|
| 6638 |
+
"grad_norm": 0.6289139986038208,
|
| 6639 |
+
"learning_rate": 4.7527424688747535e-05,
|
| 6640 |
+
"loss": 1.3543,
|
| 6641 |
+
"step": 944
|
| 6642 |
+
},
|
| 6643 |
+
{
|
| 6644 |
+
"epoch": 0.5376955903271693,
|
| 6645 |
+
"grad_norm": 0.5747124552726746,
|
| 6646 |
+
"learning_rate": 4.733404756919287e-05,
|
| 6647 |
+
"loss": 1.2679,
|
| 6648 |
+
"step": 945
|
| 6649 |
+
},
|
| 6650 |
+
{
|
| 6651 |
+
"epoch": 0.5382645803698435,
|
| 6652 |
+
"grad_norm": 0.5888437032699585,
|
| 6653 |
+
"learning_rate": 4.7140942602819236e-05,
|
| 6654 |
+
"loss": 1.3506,
|
| 6655 |
+
"step": 946
|
| 6656 |
+
},
|
| 6657 |
+
{
|
| 6658 |
+
"epoch": 0.5388335704125178,
|
| 6659 |
+
"grad_norm": 0.6044580936431885,
|
| 6660 |
+
"learning_rate": 4.694811078750338e-05,
|
| 6661 |
+
"loss": 1.5955,
|
| 6662 |
+
"step": 947
|
| 6663 |
+
},
|
| 6664 |
+
{
|
| 6665 |
+
"epoch": 0.5394025604551921,
|
| 6666 |
+
"grad_norm": 0.6149877905845642,
|
| 6667 |
+
"learning_rate": 4.6755553119710524e-05,
|
| 6668 |
+
"loss": 1.5836,
|
| 6669 |
+
"step": 948
|
| 6670 |
+
},
|
| 6671 |
+
{
|
| 6672 |
+
"epoch": 0.5399715504978663,
|
| 6673 |
+
"grad_norm": 0.6135841012001038,
|
| 6674 |
+
"learning_rate": 4.656327059448937e-05,
|
| 6675 |
+
"loss": 1.4659,
|
| 6676 |
+
"step": 949
|
| 6677 |
+
},
|
| 6678 |
+
{
|
| 6679 |
+
"epoch": 0.5405405405405406,
|
| 6680 |
+
"grad_norm": 0.5868760943412781,
|
| 6681 |
+
"learning_rate": 4.637126420546653e-05,
|
| 6682 |
+
"loss": 1.3821,
|
| 6683 |
+
"step": 950
|
| 6684 |
+
},
|
| 6685 |
+
{
|
| 6686 |
+
"epoch": 0.5411095305832148,
|
| 6687 |
+
"grad_norm": 0.6109480261802673,
|
| 6688 |
+
"learning_rate": 4.6179534944842e-05,
|
| 6689 |
+
"loss": 1.5173,
|
| 6690 |
+
"step": 951
|
| 6691 |
+
},
|
| 6692 |
+
{
|
| 6693 |
+
"epoch": 0.5416785206258891,
|
| 6694 |
+
"grad_norm": 0.6133657693862915,
|
| 6695 |
+
"learning_rate": 4.5988083803383464e-05,
|
| 6696 |
+
"loss": 1.6325,
|
| 6697 |
+
"step": 952
|
| 6698 |
+
},
|
| 6699 |
+
{
|
| 6700 |
+
"epoch": 0.5422475106685632,
|
| 6701 |
+
"grad_norm": 0.593211829662323,
|
| 6702 |
+
"learning_rate": 4.57969117704215e-05,
|
| 6703 |
+
"loss": 1.361,
|
| 6704 |
+
"step": 953
|
| 6705 |
+
},
|
| 6706 |
+
{
|
| 6707 |
+
"epoch": 0.5428165007112375,
|
| 6708 |
+
"grad_norm": 0.5881854891777039,
|
| 6709 |
+
"learning_rate": 4.560601983384447e-05,
|
| 6710 |
+
"loss": 1.3796,
|
| 6711 |
+
"step": 954
|
| 6712 |
+
},
|
| 6713 |
+
{
|
| 6714 |
+
"epoch": 0.5433854907539118,
|
| 6715 |
+
"grad_norm": 0.65924471616745,
|
| 6716 |
+
"learning_rate": 4.5415408980093096e-05,
|
| 6717 |
+
"loss": 1.5899,
|
| 6718 |
+
"step": 955
|
| 6719 |
+
},
|
| 6720 |
+
{
|
| 6721 |
+
"epoch": 0.543954480796586,
|
| 6722 |
+
"grad_norm": 0.6201925277709961,
|
| 6723 |
+
"learning_rate": 4.522508019415587e-05,
|
| 6724 |
+
"loss": 1.536,
|
| 6725 |
+
"step": 956
|
| 6726 |
+
},
|
| 6727 |
+
{
|
| 6728 |
+
"epoch": 0.5445234708392603,
|
| 6729 |
+
"grad_norm": 0.5619149208068848,
|
| 6730 |
+
"learning_rate": 4.50350344595635e-05,
|
| 6731 |
+
"loss": 1.3624,
|
| 6732 |
+
"step": 957
|
| 6733 |
+
},
|
| 6734 |
+
{
|
| 6735 |
+
"epoch": 0.5450924608819345,
|
| 6736 |
+
"grad_norm": 0.5680489540100098,
|
| 6737 |
+
"learning_rate": 4.484527275838404e-05,
|
| 6738 |
+
"loss": 1.4247,
|
| 6739 |
+
"step": 958
|
| 6740 |
+
},
|
| 6741 |
+
{
|
| 6742 |
+
"epoch": 0.5456614509246088,
|
| 6743 |
+
"grad_norm": 0.5449238419532776,
|
| 6744 |
+
"learning_rate": 4.4655796071217937e-05,
|
| 6745 |
+
"loss": 1.3423,
|
| 6746 |
+
"step": 959
|
| 6747 |
+
},
|
| 6748 |
+
{
|
| 6749 |
+
"epoch": 0.5462304409672831,
|
| 6750 |
+
"grad_norm": 0.6032193899154663,
|
| 6751 |
+
"learning_rate": 4.446660537719256e-05,
|
| 6752 |
+
"loss": 1.6294,
|
| 6753 |
+
"step": 960
|
| 6754 |
+
},
|
| 6755 |
+
{
|
| 6756 |
+
"epoch": 0.5467994310099573,
|
| 6757 |
+
"grad_norm": 0.5516905784606934,
|
| 6758 |
+
"learning_rate": 4.427770165395766e-05,
|
| 6759 |
+
"loss": 1.3738,
|
| 6760 |
+
"step": 961
|
| 6761 |
+
},
|
| 6762 |
+
{
|
| 6763 |
+
"epoch": 0.5473684210526316,
|
| 6764 |
+
"grad_norm": 0.6235291361808777,
|
| 6765 |
+
"learning_rate": 4.4089085877679904e-05,
|
| 6766 |
+
"loss": 1.4602,
|
| 6767 |
+
"step": 962
|
| 6768 |
+
},
|
| 6769 |
+
{
|
| 6770 |
+
"epoch": 0.5479374110953058,
|
| 6771 |
+
"grad_norm": 0.6051345467567444,
|
| 6772 |
+
"learning_rate": 4.3900759023037974e-05,
|
| 6773 |
+
"loss": 1.3761,
|
| 6774 |
+
"step": 963
|
| 6775 |
+
},
|
| 6776 |
+
{
|
| 6777 |
+
"epoch": 0.5485064011379801,
|
| 6778 |
+
"grad_norm": 0.5858922600746155,
|
| 6779 |
+
"learning_rate": 4.3712722063217693e-05,
|
| 6780 |
+
"loss": 1.5158,
|
| 6781 |
+
"step": 964
|
| 6782 |
+
},
|
| 6783 |
+
{
|
| 6784 |
+
"epoch": 0.5490753911806543,
|
| 6785 |
+
"grad_norm": 0.5914279818534851,
|
| 6786 |
+
"learning_rate": 4.3524975969906636e-05,
|
| 6787 |
+
"loss": 1.3333,
|
| 6788 |
+
"step": 965
|
| 6789 |
+
},
|
| 6790 |
+
{
|
| 6791 |
+
"epoch": 0.5496443812233286,
|
| 6792 |
+
"grad_norm": 0.5849418044090271,
|
| 6793 |
+
"learning_rate": 4.3337521713289407e-05,
|
| 6794 |
+
"loss": 1.5459,
|
| 6795 |
+
"step": 966
|
| 6796 |
+
},
|
| 6797 |
+
{
|
| 6798 |
+
"epoch": 0.5502133712660029,
|
| 6799 |
+
"grad_norm": 0.5740037560462952,
|
| 6800 |
+
"learning_rate": 4.315036026204262e-05,
|
| 6801 |
+
"loss": 1.3858,
|
| 6802 |
+
"step": 967
|
| 6803 |
+
},
|
| 6804 |
+
{
|
| 6805 |
+
"epoch": 0.5507823613086771,
|
| 6806 |
+
"grad_norm": 0.5611101984977722,
|
| 6807 |
+
"learning_rate": 4.296349258332967e-05,
|
| 6808 |
+
"loss": 1.3895,
|
| 6809 |
+
"step": 968
|
| 6810 |
+
},
|
| 6811 |
+
{
|
| 6812 |
+
"epoch": 0.5513513513513514,
|
| 6813 |
+
"grad_norm": 0.585473895072937,
|
| 6814 |
+
"learning_rate": 4.277691964279594e-05,
|
| 6815 |
+
"loss": 1.2682,
|
| 6816 |
+
"step": 969
|
| 6817 |
+
},
|
| 6818 |
+
{
|
| 6819 |
+
"epoch": 0.5519203413940256,
|
| 6820 |
+
"grad_norm": 0.6113364100456238,
|
| 6821 |
+
"learning_rate": 4.259064240456374e-05,
|
| 6822 |
+
"loss": 1.4292,
|
| 6823 |
+
"step": 970
|
| 6824 |
+
},
|
| 6825 |
+
{
|
| 6826 |
+
"epoch": 0.5524893314366999,
|
| 6827 |
+
"grad_norm": 0.6335917115211487,
|
| 6828 |
+
"learning_rate": 4.2404661831227276e-05,
|
| 6829 |
+
"loss": 1.4529,
|
| 6830 |
+
"step": 971
|
| 6831 |
+
},
|
| 6832 |
+
{
|
| 6833 |
+
"epoch": 0.5530583214793741,
|
| 6834 |
+
"grad_norm": 0.574226975440979,
|
| 6835 |
+
"learning_rate": 4.2218978883847835e-05,
|
| 6836 |
+
"loss": 1.5254,
|
| 6837 |
+
"step": 972
|
| 6838 |
+
},
|
| 6839 |
+
{
|
| 6840 |
+
"epoch": 0.5536273115220484,
|
| 6841 |
+
"grad_norm": 0.5865671038627625,
|
| 6842 |
+
"learning_rate": 4.203359452194863e-05,
|
| 6843 |
+
"loss": 1.5265,
|
| 6844 |
+
"step": 973
|
| 6845 |
+
},
|
| 6846 |
+
{
|
| 6847 |
+
"epoch": 0.5541963015647227,
|
| 6848 |
+
"grad_norm": 0.5852011442184448,
|
| 6849 |
+
"learning_rate": 4.184850970350992e-05,
|
| 6850 |
+
"loss": 1.5834,
|
| 6851 |
+
"step": 974
|
| 6852 |
+
},
|
| 6853 |
+
{
|
| 6854 |
+
"epoch": 0.5547652916073968,
|
| 6855 |
+
"grad_norm": 0.6045235395431519,
|
| 6856 |
+
"learning_rate": 4.166372538496408e-05,
|
| 6857 |
+
"loss": 1.3905,
|
| 6858 |
+
"step": 975
|
| 6859 |
+
},
|
| 6860 |
+
{
|
| 6861 |
+
"epoch": 0.5553342816500711,
|
| 6862 |
+
"grad_norm": 0.558691143989563,
|
| 6863 |
+
"learning_rate": 4.147924252119063e-05,
|
| 6864 |
+
"loss": 1.5088,
|
| 6865 |
+
"step": 976
|
| 6866 |
+
},
|
| 6867 |
+
{
|
| 6868 |
+
"epoch": 0.5559032716927453,
|
| 6869 |
+
"grad_norm": 0.5623577237129211,
|
| 6870 |
+
"learning_rate": 4.129506206551138e-05,
|
| 6871 |
+
"loss": 1.3502,
|
| 6872 |
+
"step": 977
|
| 6873 |
+
},
|
| 6874 |
+
{
|
| 6875 |
+
"epoch": 0.5564722617354196,
|
| 6876 |
+
"grad_norm": 0.5946846604347229,
|
| 6877 |
+
"learning_rate": 4.1111184969685354e-05,
|
| 6878 |
+
"loss": 1.3884,
|
| 6879 |
+
"step": 978
|
| 6880 |
+
},
|
| 6881 |
+
{
|
| 6882 |
+
"epoch": 0.5570412517780939,
|
| 6883 |
+
"grad_norm": 0.5882412195205688,
|
| 6884 |
+
"learning_rate": 4.0927612183903976e-05,
|
| 6885 |
+
"loss": 1.542,
|
| 6886 |
+
"step": 979
|
| 6887 |
+
},
|
| 6888 |
+
{
|
| 6889 |
+
"epoch": 0.5576102418207681,
|
| 6890 |
+
"grad_norm": 0.577912449836731,
|
| 6891 |
+
"learning_rate": 4.0744344656786124e-05,
|
| 6892 |
+
"loss": 1.324,
|
| 6893 |
+
"step": 980
|
| 6894 |
+
},
|
| 6895 |
+
{
|
| 6896 |
+
"epoch": 0.5581792318634424,
|
| 6897 |
+
"grad_norm": 0.5644152164459229,
|
| 6898 |
+
"learning_rate": 4.056138333537326e-05,
|
| 6899 |
+
"loss": 1.2746,
|
| 6900 |
+
"step": 981
|
| 6901 |
+
},
|
| 6902 |
+
{
|
| 6903 |
+
"epoch": 0.5587482219061166,
|
| 6904 |
+
"grad_norm": 0.6058292984962463,
|
| 6905 |
+
"learning_rate": 4.037872916512455e-05,
|
| 6906 |
+
"loss": 1.5404,
|
| 6907 |
+
"step": 982
|
| 6908 |
+
},
|
| 6909 |
+
{
|
| 6910 |
+
"epoch": 0.5593172119487909,
|
| 6911 |
+
"grad_norm": 0.6061570644378662,
|
| 6912 |
+
"learning_rate": 4.019638308991189e-05,
|
| 6913 |
+
"loss": 1.3896,
|
| 6914 |
+
"step": 983
|
| 6915 |
+
},
|
| 6916 |
+
{
|
| 6917 |
+
"epoch": 0.5598862019914651,
|
| 6918 |
+
"grad_norm": 0.6102644205093384,
|
| 6919 |
+
"learning_rate": 4.0014346052015114e-05,
|
| 6920 |
+
"loss": 1.5365,
|
| 6921 |
+
"step": 984
|
| 6922 |
+
},
|
| 6923 |
+
{
|
| 6924 |
+
"epoch": 0.5604551920341394,
|
| 6925 |
+
"grad_norm": 0.5747568011283875,
|
| 6926 |
+
"learning_rate": 3.983261899211708e-05,
|
| 6927 |
+
"loss": 1.4337,
|
| 6928 |
+
"step": 985
|
| 6929 |
+
},
|
| 6930 |
+
{
|
| 6931 |
+
"epoch": 0.5610241820768137,
|
| 6932 |
+
"grad_norm": 0.5756990909576416,
|
| 6933 |
+
"learning_rate": 3.965120284929878e-05,
|
| 6934 |
+
"loss": 1.4752,
|
| 6935 |
+
"step": 986
|
| 6936 |
+
},
|
| 6937 |
+
{
|
| 6938 |
+
"epoch": 0.5615931721194879,
|
| 6939 |
+
"grad_norm": 0.570568323135376,
|
| 6940 |
+
"learning_rate": 3.947009856103465e-05,
|
| 6941 |
+
"loss": 1.4064,
|
| 6942 |
+
"step": 987
|
| 6943 |
+
},
|
| 6944 |
+
{
|
| 6945 |
+
"epoch": 0.5621621621621622,
|
| 6946 |
+
"grad_norm": 0.6102871298789978,
|
| 6947 |
+
"learning_rate": 3.928930706318752e-05,
|
| 6948 |
+
"loss": 1.5697,
|
| 6949 |
+
"step": 988
|
| 6950 |
+
},
|
| 6951 |
+
{
|
| 6952 |
+
"epoch": 0.5627311522048364,
|
| 6953 |
+
"grad_norm": 0.555619478225708,
|
| 6954 |
+
"learning_rate": 3.910882929000387e-05,
|
| 6955 |
+
"loss": 1.2905,
|
| 6956 |
+
"step": 989
|
| 6957 |
+
},
|
| 6958 |
+
{
|
| 6959 |
+
"epoch": 0.5633001422475107,
|
| 6960 |
+
"grad_norm": 0.6053213477134705,
|
| 6961 |
+
"learning_rate": 3.892866617410901e-05,
|
| 6962 |
+
"loss": 1.4823,
|
| 6963 |
+
"step": 990
|
| 6964 |
+
},
|
| 6965 |
+
{
|
| 6966 |
+
"epoch": 0.5638691322901849,
|
| 6967 |
+
"grad_norm": 0.5635027289390564,
|
| 6968 |
+
"learning_rate": 3.874881864650224e-05,
|
| 6969 |
+
"loss": 1.2325,
|
| 6970 |
+
"step": 991
|
| 6971 |
+
},
|
| 6972 |
+
{
|
| 6973 |
+
"epoch": 0.5644381223328592,
|
| 6974 |
+
"grad_norm": 0.6095726490020752,
|
| 6975 |
+
"learning_rate": 3.8569287636552024e-05,
|
| 6976 |
+
"loss": 1.5359,
|
| 6977 |
+
"step": 992
|
| 6978 |
+
},
|
| 6979 |
+
{
|
| 6980 |
+
"epoch": 0.5650071123755335,
|
| 6981 |
+
"grad_norm": 0.5644766092300415,
|
| 6982 |
+
"learning_rate": 3.839007407199129e-05,
|
| 6983 |
+
"loss": 1.277,
|
| 6984 |
+
"step": 993
|
| 6985 |
+
},
|
| 6986 |
+
{
|
| 6987 |
+
"epoch": 0.5655761024182077,
|
| 6988 |
+
"grad_norm": 0.5609472393989563,
|
| 6989 |
+
"learning_rate": 3.821117887891249e-05,
|
| 6990 |
+
"loss": 1.2394,
|
| 6991 |
+
"step": 994
|
| 6992 |
+
},
|
| 6993 |
+
{
|
| 6994 |
+
"epoch": 0.566145092460882,
|
| 6995 |
+
"grad_norm": 0.6164161562919617,
|
| 6996 |
+
"learning_rate": 3.803260298176288e-05,
|
| 6997 |
+
"loss": 1.5458,
|
| 6998 |
+
"step": 995
|
| 6999 |
+
},
|
| 7000 |
+
{
|
| 7001 |
+
"epoch": 0.5667140825035561,
|
| 7002 |
+
"grad_norm": 0.6040405631065369,
|
| 7003 |
+
"learning_rate": 3.7854347303339754e-05,
|
| 7004 |
+
"loss": 1.2356,
|
| 7005 |
+
"step": 996
|
| 7006 |
+
},
|
| 7007 |
+
{
|
| 7008 |
+
"epoch": 0.5672830725462304,
|
| 7009 |
+
"grad_norm": 0.6196702718734741,
|
| 7010 |
+
"learning_rate": 3.767641276478563e-05,
|
| 7011 |
+
"loss": 1.5923,
|
| 7012 |
+
"step": 997
|
| 7013 |
+
},
|
| 7014 |
+
{
|
| 7015 |
+
"epoch": 0.5678520625889047,
|
| 7016 |
+
"grad_norm": 0.5526005029678345,
|
| 7017 |
+
"learning_rate": 3.749880028558364e-05,
|
| 7018 |
+
"loss": 1.5057,
|
| 7019 |
+
"step": 998
|
| 7020 |
+
},
|
| 7021 |
+
{
|
| 7022 |
+
"epoch": 0.5684210526315789,
|
| 7023 |
+
"grad_norm": 0.5806797742843628,
|
| 7024 |
+
"learning_rate": 3.732151078355253e-05,
|
| 7025 |
+
"loss": 1.5355,
|
| 7026 |
+
"step": 999
|
| 7027 |
+
},
|
| 7028 |
+
{
|
| 7029 |
+
"epoch": 0.5689900426742532,
|
| 7030 |
+
"grad_norm": 0.5680354237556458,
|
| 7031 |
+
"learning_rate": 3.7144545174842115e-05,
|
| 7032 |
+
"loss": 1.4381,
|
| 7033 |
+
"step": 1000
|
| 7034 |
+
},
|
| 7035 |
+
{
|
| 7036 |
+
"epoch": 0.5695590327169274,
|
| 7037 |
+
"grad_norm": 0.5921180248260498,
|
| 7038 |
+
"learning_rate": 3.6967904373928475e-05,
|
| 7039 |
+
"loss": 1.3444,
|
| 7040 |
+
"step": 1001
|
| 7041 |
+
},
|
| 7042 |
+
{
|
| 7043 |
+
"epoch": 0.5701280227596017,
|
| 7044 |
+
"grad_norm": 0.5849342942237854,
|
| 7045 |
+
"learning_rate": 3.6791589293609184e-05,
|
| 7046 |
+
"loss": 1.3836,
|
| 7047 |
+
"step": 1002
|
| 7048 |
+
},
|
| 7049 |
+
{
|
| 7050 |
+
"epoch": 0.5706970128022759,
|
| 7051 |
+
"grad_norm": 0.5548643469810486,
|
| 7052 |
+
"learning_rate": 3.661560084499874e-05,
|
| 7053 |
+
"loss": 1.4809,
|
| 7054 |
+
"step": 1003
|
| 7055 |
+
},
|
| 7056 |
+
{
|
| 7057 |
+
"epoch": 0.5712660028449502,
|
| 7058 |
+
"grad_norm": 0.5976467132568359,
|
| 7059 |
+
"learning_rate": 3.64399399375237e-05,
|
| 7060 |
+
"loss": 1.4543,
|
| 7061 |
+
"step": 1004
|
| 7062 |
+
},
|
| 7063 |
+
{
|
| 7064 |
+
"epoch": 0.5718349928876245,
|
| 7065 |
+
"grad_norm": 0.588699996471405,
|
| 7066 |
+
"learning_rate": 3.6264607478918037e-05,
|
| 7067 |
+
"loss": 1.4448,
|
| 7068 |
+
"step": 1005
|
| 7069 |
+
},
|
| 7070 |
+
{
|
| 7071 |
+
"epoch": 0.5724039829302987,
|
| 7072 |
+
"grad_norm": 0.5786314606666565,
|
| 7073 |
+
"learning_rate": 3.608960437521844e-05,
|
| 7074 |
+
"loss": 1.769,
|
| 7075 |
+
"step": 1006
|
| 7076 |
+
},
|
| 7077 |
+
{
|
| 7078 |
+
"epoch": 0.572972972972973,
|
| 7079 |
+
"grad_norm": 0.6124690771102905,
|
| 7080 |
+
"learning_rate": 3.591493153075966e-05,
|
| 7081 |
+
"loss": 1.6527,
|
| 7082 |
+
"step": 1007
|
| 7083 |
+
},
|
| 7084 |
+
{
|
| 7085 |
+
"epoch": 0.5735419630156472,
|
| 7086 |
+
"grad_norm": 0.5587359070777893,
|
| 7087 |
+
"learning_rate": 3.5740589848169894e-05,
|
| 7088 |
+
"loss": 1.2819,
|
| 7089 |
+
"step": 1008
|
| 7090 |
+
},
|
| 7091 |
+
{
|
| 7092 |
+
"epoch": 0.5741109530583215,
|
| 7093 |
+
"grad_norm": 0.6170410513877869,
|
| 7094 |
+
"learning_rate": 3.556658022836594e-05,
|
| 7095 |
+
"loss": 1.5858,
|
| 7096 |
+
"step": 1009
|
| 7097 |
+
},
|
| 7098 |
+
{
|
| 7099 |
+
"epoch": 0.5746799431009957,
|
| 7100 |
+
"grad_norm": 0.5927881002426147,
|
| 7101 |
+
"learning_rate": 3.5392903570548694e-05,
|
| 7102 |
+
"loss": 1.6321,
|
| 7103 |
+
"step": 1010
|
| 7104 |
+
},
|
| 7105 |
+
{
|
| 7106 |
+
"epoch": 0.57524893314367,
|
| 7107 |
+
"grad_norm": 0.5902583599090576,
|
| 7108 |
+
"learning_rate": 3.521956077219847e-05,
|
| 7109 |
+
"loss": 1.5162,
|
| 7110 |
+
"step": 1011
|
| 7111 |
+
},
|
| 7112 |
+
{
|
| 7113 |
+
"epoch": 0.5758179231863443,
|
| 7114 |
+
"grad_norm": 0.6113704442977905,
|
| 7115 |
+
"learning_rate": 3.504655272907028e-05,
|
| 7116 |
+
"loss": 1.6929,
|
| 7117 |
+
"step": 1012
|
| 7118 |
+
},
|
| 7119 |
+
{
|
| 7120 |
+
"epoch": 0.5763869132290185,
|
| 7121 |
+
"grad_norm": 0.5586623549461365,
|
| 7122 |
+
"learning_rate": 3.4873880335189427e-05,
|
| 7123 |
+
"loss": 1.3555,
|
| 7124 |
+
"step": 1013
|
| 7125 |
+
},
|
| 7126 |
+
{
|
| 7127 |
+
"epoch": 0.5769559032716928,
|
| 7128 |
+
"grad_norm": 0.5992634296417236,
|
| 7129 |
+
"learning_rate": 3.470154448284659e-05,
|
| 7130 |
+
"loss": 1.6901,
|
| 7131 |
+
"step": 1014
|
| 7132 |
+
},
|
| 7133 |
+
{
|
| 7134 |
+
"epoch": 0.577524893314367,
|
| 7135 |
+
"grad_norm": 0.5722742676734924,
|
| 7136 |
+
"learning_rate": 3.452954606259343e-05,
|
| 7137 |
+
"loss": 1.386,
|
| 7138 |
+
"step": 1015
|
| 7139 |
+
},
|
| 7140 |
+
{
|
| 7141 |
+
"epoch": 0.5780938833570413,
|
| 7142 |
+
"grad_norm": 0.6090911030769348,
|
| 7143 |
+
"learning_rate": 3.435788596323789e-05,
|
| 7144 |
+
"loss": 1.528,
|
| 7145 |
+
"step": 1016
|
| 7146 |
+
},
|
| 7147 |
+
{
|
| 7148 |
+
"epoch": 0.5786628733997156,
|
| 7149 |
+
"grad_norm": 0.5943465828895569,
|
| 7150 |
+
"learning_rate": 3.41865650718396e-05,
|
| 7151 |
+
"loss": 1.4567,
|
| 7152 |
+
"step": 1017
|
| 7153 |
+
},
|
| 7154 |
+
{
|
| 7155 |
+
"epoch": 0.5792318634423897,
|
| 7156 |
+
"grad_norm": 0.5948119163513184,
|
| 7157 |
+
"learning_rate": 3.4015584273705425e-05,
|
| 7158 |
+
"loss": 1.4926,
|
| 7159 |
+
"step": 1018
|
| 7160 |
+
},
|
| 7161 |
+
{
|
| 7162 |
+
"epoch": 0.579800853485064,
|
| 7163 |
+
"grad_norm": 0.6115890741348267,
|
| 7164 |
+
"learning_rate": 3.384494445238471e-05,
|
| 7165 |
+
"loss": 1.4113,
|
| 7166 |
+
"step": 1019
|
| 7167 |
+
},
|
| 7168 |
+
{
|
| 7169 |
+
"epoch": 0.5803698435277382,
|
| 7170 |
+
"grad_norm": 0.5682458281517029,
|
| 7171 |
+
"learning_rate": 3.367464648966471e-05,
|
| 7172 |
+
"loss": 1.514,
|
| 7173 |
+
"step": 1020
|
| 7174 |
+
},
|
| 7175 |
+
{
|
| 7176 |
+
"epoch": 0.5809388335704125,
|
| 7177 |
+
"grad_norm": 0.5994877219200134,
|
| 7178 |
+
"learning_rate": 3.350469126556627e-05,
|
| 7179 |
+
"loss": 1.495,
|
| 7180 |
+
"step": 1021
|
| 7181 |
+
},
|
| 7182 |
+
{
|
| 7183 |
+
"epoch": 0.5815078236130867,
|
| 7184 |
+
"grad_norm": 0.5887535810470581,
|
| 7185 |
+
"learning_rate": 3.333507965833905e-05,
|
| 7186 |
+
"loss": 1.6428,
|
| 7187 |
+
"step": 1022
|
| 7188 |
+
},
|
| 7189 |
+
{
|
| 7190 |
+
"epoch": 0.582076813655761,
|
| 7191 |
+
"grad_norm": 0.5758301615715027,
|
| 7192 |
+
"learning_rate": 3.316581254445701e-05,
|
| 7193 |
+
"loss": 1.4076,
|
| 7194 |
+
"step": 1023
|
| 7195 |
+
},
|
| 7196 |
+
{
|
| 7197 |
+
"epoch": 0.5826458036984353,
|
| 7198 |
+
"grad_norm": 0.6117954850196838,
|
| 7199 |
+
"learning_rate": 3.299689079861408e-05,
|
| 7200 |
+
"loss": 1.4471,
|
| 7201 |
+
"step": 1024
|
| 7202 |
+
},
|
| 7203 |
+
{
|
| 7204 |
+
"epoch": 0.5832147937411095,
|
| 7205 |
+
"grad_norm": 0.6079879999160767,
|
| 7206 |
+
"learning_rate": 3.2828315293719245e-05,
|
| 7207 |
+
"loss": 1.485,
|
| 7208 |
+
"step": 1025
|
| 7209 |
+
},
|
| 7210 |
+
{
|
| 7211 |
+
"epoch": 0.5837837837837838,
|
| 7212 |
+
"grad_norm": 0.5936009287834167,
|
| 7213 |
+
"learning_rate": 3.266008690089253e-05,
|
| 7214 |
+
"loss": 1.6109,
|
| 7215 |
+
"step": 1026
|
| 7216 |
+
},
|
| 7217 |
+
{
|
| 7218 |
+
"epoch": 0.584352773826458,
|
| 7219 |
+
"grad_norm": 0.5736754536628723,
|
| 7220 |
+
"learning_rate": 3.24922064894601e-05,
|
| 7221 |
+
"loss": 1.4451,
|
| 7222 |
+
"step": 1027
|
| 7223 |
+
},
|
| 7224 |
+
{
|
| 7225 |
+
"epoch": 0.5849217638691323,
|
| 7226 |
+
"grad_norm": 0.5830667018890381,
|
| 7227 |
+
"learning_rate": 3.23246749269499e-05,
|
| 7228 |
+
"loss": 1.499,
|
| 7229 |
+
"step": 1028
|
| 7230 |
+
},
|
| 7231 |
+
{
|
| 7232 |
+
"epoch": 0.5854907539118065,
|
| 7233 |
+
"grad_norm": 0.5929978489875793,
|
| 7234 |
+
"learning_rate": 3.2157493079087343e-05,
|
| 7235 |
+
"loss": 1.5964,
|
| 7236 |
+
"step": 1029
|
| 7237 |
+
},
|
| 7238 |
+
{
|
| 7239 |
+
"epoch": 0.5860597439544808,
|
| 7240 |
+
"grad_norm": 0.5748528242111206,
|
| 7241 |
+
"learning_rate": 3.1990661809790445e-05,
|
| 7242 |
+
"loss": 1.3425,
|
| 7243 |
+
"step": 1030
|
| 7244 |
+
},
|
| 7245 |
+
{
|
| 7246 |
+
"epoch": 0.5866287339971551,
|
| 7247 |
+
"grad_norm": 0.6261157393455505,
|
| 7248 |
+
"learning_rate": 3.18241819811658e-05,
|
| 7249 |
+
"loss": 1.4458,
|
| 7250 |
+
"step": 1031
|
| 7251 |
+
},
|
| 7252 |
+
{
|
| 7253 |
+
"epoch": 0.5871977240398293,
|
| 7254 |
+
"grad_norm": 0.5736514925956726,
|
| 7255 |
+
"learning_rate": 3.165805445350383e-05,
|
| 7256 |
+
"loss": 1.3948,
|
| 7257 |
+
"step": 1032
|
| 7258 |
+
},
|
| 7259 |
+
{
|
| 7260 |
+
"epoch": 0.5877667140825036,
|
| 7261 |
+
"grad_norm": 0.6165857911109924,
|
| 7262 |
+
"learning_rate": 3.149228008527437e-05,
|
| 7263 |
+
"loss": 1.6043,
|
| 7264 |
+
"step": 1033
|
| 7265 |
+
},
|
| 7266 |
+
{
|
| 7267 |
+
"epoch": 0.5883357041251778,
|
| 7268 |
+
"grad_norm": 0.6109797954559326,
|
| 7269 |
+
"learning_rate": 3.132685973312251e-05,
|
| 7270 |
+
"loss": 1.5376,
|
| 7271 |
+
"step": 1034
|
| 7272 |
+
},
|
| 7273 |
+
{
|
| 7274 |
+
"epoch": 0.5889046941678521,
|
| 7275 |
+
"grad_norm": 0.5716987252235413,
|
| 7276 |
+
"learning_rate": 3.116179425186361e-05,
|
| 7277 |
+
"loss": 1.3554,
|
| 7278 |
+
"step": 1035
|
| 7279 |
+
},
|
| 7280 |
+
{
|
| 7281 |
+
"epoch": 0.5894736842105263,
|
| 7282 |
+
"grad_norm": 0.6563665866851807,
|
| 7283 |
+
"learning_rate": 3.099708449447956e-05,
|
| 7284 |
+
"loss": 1.4934,
|
| 7285 |
+
"step": 1036
|
| 7286 |
+
},
|
| 7287 |
+
{
|
| 7288 |
+
"epoch": 0.5900426742532006,
|
| 7289 |
+
"grad_norm": 0.6072697043418884,
|
| 7290 |
+
"learning_rate": 3.083273131211382e-05,
|
| 7291 |
+
"loss": 1.3181,
|
| 7292 |
+
"step": 1037
|
| 7293 |
+
},
|
| 7294 |
+
{
|
| 7295 |
+
"epoch": 0.5906116642958749,
|
| 7296 |
+
"grad_norm": 0.5769975781440735,
|
| 7297 |
+
"learning_rate": 3.066873555406727e-05,
|
| 7298 |
+
"loss": 1.5376,
|
| 7299 |
+
"step": 1038
|
| 7300 |
+
},
|
| 7301 |
+
{
|
| 7302 |
+
"epoch": 0.591180654338549,
|
| 7303 |
+
"grad_norm": 0.58552485704422,
|
| 7304 |
+
"learning_rate": 3.0505098067793937e-05,
|
| 7305 |
+
"loss": 1.3483,
|
| 7306 |
+
"step": 1039
|
| 7307 |
+
},
|
| 7308 |
+
{
|
| 7309 |
+
"epoch": 0.5917496443812233,
|
| 7310 |
+
"grad_norm": 0.6377474069595337,
|
| 7311 |
+
"learning_rate": 3.0341819698896202e-05,
|
| 7312 |
+
"loss": 1.6044,
|
| 7313 |
+
"step": 1040
|
| 7314 |
+
},
|
| 7315 |
+
{
|
| 7316 |
+
"epoch": 0.5923186344238975,
|
| 7317 |
+
"grad_norm": 0.5746393203735352,
|
| 7318 |
+
"learning_rate": 3.017890129112094e-05,
|
| 7319 |
+
"loss": 1.5081,
|
| 7320 |
+
"step": 1041
|
| 7321 |
+
},
|
| 7322 |
+
{
|
| 7323 |
+
"epoch": 0.5928876244665718,
|
| 7324 |
+
"grad_norm": 0.5879509449005127,
|
| 7325 |
+
"learning_rate": 3.0016343686354775e-05,
|
| 7326 |
+
"loss": 1.7884,
|
| 7327 |
+
"step": 1042
|
| 7328 |
+
},
|
| 7329 |
+
{
|
| 7330 |
+
"epoch": 0.5934566145092461,
|
| 7331 |
+
"grad_norm": 0.5871498584747314,
|
| 7332 |
+
"learning_rate": 2.9854147724619886e-05,
|
| 7333 |
+
"loss": 1.4425,
|
| 7334 |
+
"step": 1043
|
| 7335 |
+
},
|
| 7336 |
+
{
|
| 7337 |
+
"epoch": 0.5940256045519203,
|
| 7338 |
+
"grad_norm": 0.6417199373245239,
|
| 7339 |
+
"learning_rate": 2.9692314244069764e-05,
|
| 7340 |
+
"loss": 1.4729,
|
| 7341 |
+
"step": 1044
|
| 7342 |
+
},
|
| 7343 |
+
{
|
| 7344 |
+
"epoch": 0.5940256045519203,
|
| 7345 |
+
"eval_loss": 1.4645270109176636,
|
| 7346 |
+
"eval_runtime": 16.2716,
|
| 7347 |
+
"eval_samples_per_second": 45.478,
|
| 7348 |
+
"eval_steps_per_second": 22.739,
|
| 7349 |
+
"step": 1044
|
| 7350 |
}
|
| 7351 |
],
|
| 7352 |
"logging_steps": 1,
|
|
|
|
| 7366 |
"attributes": {}
|
| 7367 |
}
|
| 7368 |
},
|
| 7369 |
+
"total_flos": 1.856000524497715e+16,
|
| 7370 |
"train_batch_size": 2,
|
| 7371 |
"trial_name": null,
|
| 7372 |
"trial_params": null
|