Training in progress, step 2868, 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 2145944
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a96170c4566c1e7efa993538ecb2474b7056c80c01787e9d312ce31d75111369
|
| 3 |
size 2145944
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4310020
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:635f9699d60977f7f49fef0fac4d8dd2e9540f390cee9cca0ae8d57bdec47c46
|
| 3 |
size 4310020
|
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:a9966c454e255711c675f388ff89a77f36c50534180a271dc2d40e7f43b870bf
|
| 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 sha256:237b866b1fce3ebf6f30679e5de802610141e1fcea25e04865362188175cac93
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -18410,6 +18410,1679 @@
|
|
| 18410 |
"learning_rate": 2.2784275177278934e-06,
|
| 18411 |
"loss": 9.3997,
|
| 18412 |
"step": 2629
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18413 |
}
|
| 18414 |
],
|
| 18415 |
"logging_steps": 1,
|
|
@@ -18429,7 +20102,7 @@
|
|
| 18429 |
"attributes": {}
|
| 18430 |
}
|
| 18431 |
},
|
| 18432 |
-
"total_flos":
|
| 18433 |
"train_batch_size": 4,
|
| 18434 |
"trial_name": null,
|
| 18435 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9893066574680924,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 2868,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 18410 |
"learning_rate": 2.2784275177278934e-06,
|
| 18411 |
"loss": 9.3997,
|
| 18412 |
"step": 2629
|
| 18413 |
+
},
|
| 18414 |
+
{
|
| 18415 |
+
"epoch": 0.9072093825457054,
|
| 18416 |
+
"grad_norm": 1.3021107912063599,
|
| 18417 |
+
"learning_rate": 2.2617097018509613e-06,
|
| 18418 |
+
"loss": 9.5147,
|
| 18419 |
+
"step": 2630
|
| 18420 |
+
},
|
| 18421 |
+
{
|
| 18422 |
+
"epoch": 0.9075543290789928,
|
| 18423 |
+
"grad_norm": 1.280267357826233,
|
| 18424 |
+
"learning_rate": 2.2450520256057038e-06,
|
| 18425 |
+
"loss": 9.5056,
|
| 18426 |
+
"step": 2631
|
| 18427 |
+
},
|
| 18428 |
+
{
|
| 18429 |
+
"epoch": 0.9078992756122801,
|
| 18430 |
+
"grad_norm": 1.3371871709823608,
|
| 18431 |
+
"learning_rate": 2.22845450997709e-06,
|
| 18432 |
+
"loss": 9.521,
|
| 18433 |
+
"step": 2632
|
| 18434 |
+
},
|
| 18435 |
+
{
|
| 18436 |
+
"epoch": 0.9082442221455674,
|
| 18437 |
+
"grad_norm": 1.3242119550704956,
|
| 18438 |
+
"learning_rate": 2.2119171758743117e-06,
|
| 18439 |
+
"loss": 9.4992,
|
| 18440 |
+
"step": 2633
|
| 18441 |
+
},
|
| 18442 |
+
{
|
| 18443 |
+
"epoch": 0.9085891686788548,
|
| 18444 |
+
"grad_norm": 1.357699990272522,
|
| 18445 |
+
"learning_rate": 2.19544004413072e-06,
|
| 18446 |
+
"loss": 9.4417,
|
| 18447 |
+
"step": 2634
|
| 18448 |
+
},
|
| 18449 |
+
{
|
| 18450 |
+
"epoch": 0.9089341152121421,
|
| 18451 |
+
"grad_norm": 1.3349027633666992,
|
| 18452 |
+
"learning_rate": 2.1790231355038495e-06,
|
| 18453 |
+
"loss": 9.462,
|
| 18454 |
+
"step": 2635
|
| 18455 |
+
},
|
| 18456 |
+
{
|
| 18457 |
+
"epoch": 0.9092790617454295,
|
| 18458 |
+
"grad_norm": 1.4189931154251099,
|
| 18459 |
+
"learning_rate": 2.162666470675334e-06,
|
| 18460 |
+
"loss": 9.4083,
|
| 18461 |
+
"step": 2636
|
| 18462 |
+
},
|
| 18463 |
+
{
|
| 18464 |
+
"epoch": 0.9096240082787168,
|
| 18465 |
+
"grad_norm": 1.5067139863967896,
|
| 18466 |
+
"learning_rate": 2.146370070250958e-06,
|
| 18467 |
+
"loss": 9.4424,
|
| 18468 |
+
"step": 2637
|
| 18469 |
+
},
|
| 18470 |
+
{
|
| 18471 |
+
"epoch": 0.9099689548120041,
|
| 18472 |
+
"grad_norm": 1.5032390356063843,
|
| 18473 |
+
"learning_rate": 2.130133954760538e-06,
|
| 18474 |
+
"loss": 9.4019,
|
| 18475 |
+
"step": 2638
|
| 18476 |
+
},
|
| 18477 |
+
{
|
| 18478 |
+
"epoch": 0.9103139013452914,
|
| 18479 |
+
"grad_norm": 1.3871209621429443,
|
| 18480 |
+
"learning_rate": 2.1139581446580017e-06,
|
| 18481 |
+
"loss": 9.5032,
|
| 18482 |
+
"step": 2639
|
| 18483 |
+
},
|
| 18484 |
+
{
|
| 18485 |
+
"epoch": 0.9106588478785789,
|
| 18486 |
+
"grad_norm": 1.5128953456878662,
|
| 18487 |
+
"learning_rate": 2.097842660321242e-06,
|
| 18488 |
+
"loss": 9.3989,
|
| 18489 |
+
"step": 2640
|
| 18490 |
+
},
|
| 18491 |
+
{
|
| 18492 |
+
"epoch": 0.9110037944118662,
|
| 18493 |
+
"grad_norm": 1.3251484632492065,
|
| 18494 |
+
"learning_rate": 2.081787522052203e-06,
|
| 18495 |
+
"loss": 9.5056,
|
| 18496 |
+
"step": 2641
|
| 18497 |
+
},
|
| 18498 |
+
{
|
| 18499 |
+
"epoch": 0.9113487409451535,
|
| 18500 |
+
"grad_norm": 1.3176518678665161,
|
| 18501 |
+
"learning_rate": 2.0657927500767894e-06,
|
| 18502 |
+
"loss": 9.4636,
|
| 18503 |
+
"step": 2642
|
| 18504 |
+
},
|
| 18505 |
+
{
|
| 18506 |
+
"epoch": 0.9116936874784408,
|
| 18507 |
+
"grad_norm": 1.5699663162231445,
|
| 18508 |
+
"learning_rate": 2.0498583645448487e-06,
|
| 18509 |
+
"loss": 9.4639,
|
| 18510 |
+
"step": 2643
|
| 18511 |
+
},
|
| 18512 |
+
{
|
| 18513 |
+
"epoch": 0.9120386340117281,
|
| 18514 |
+
"grad_norm": 1.4477311372756958,
|
| 18515 |
+
"learning_rate": 2.0339843855301744e-06,
|
| 18516 |
+
"loss": 9.4008,
|
| 18517 |
+
"step": 2644
|
| 18518 |
+
},
|
| 18519 |
+
{
|
| 18520 |
+
"epoch": 0.9123835805450156,
|
| 18521 |
+
"grad_norm": 1.6083184480667114,
|
| 18522 |
+
"learning_rate": 2.018170833030436e-06,
|
| 18523 |
+
"loss": 9.4396,
|
| 18524 |
+
"step": 2645
|
| 18525 |
+
},
|
| 18526 |
+
{
|
| 18527 |
+
"epoch": 0.9127285270783029,
|
| 18528 |
+
"grad_norm": 1.6080248355865479,
|
| 18529 |
+
"learning_rate": 2.0024177269672094e-06,
|
| 18530 |
+
"loss": 9.4432,
|
| 18531 |
+
"step": 2646
|
| 18532 |
+
},
|
| 18533 |
+
{
|
| 18534 |
+
"epoch": 0.9130734736115902,
|
| 18535 |
+
"grad_norm": 1.4727885723114014,
|
| 18536 |
+
"learning_rate": 1.986725087185898e-06,
|
| 18537 |
+
"loss": 9.4662,
|
| 18538 |
+
"step": 2647
|
| 18539 |
+
},
|
| 18540 |
+
{
|
| 18541 |
+
"epoch": 0.9134184201448775,
|
| 18542 |
+
"grad_norm": 1.6938202381134033,
|
| 18543 |
+
"learning_rate": 1.9710929334557484e-06,
|
| 18544 |
+
"loss": 9.332,
|
| 18545 |
+
"step": 2648
|
| 18546 |
+
},
|
| 18547 |
+
{
|
| 18548 |
+
"epoch": 0.9137633666781649,
|
| 18549 |
+
"grad_norm": 1.6540645360946655,
|
| 18550 |
+
"learning_rate": 1.9555212854697803e-06,
|
| 18551 |
+
"loss": 9.3841,
|
| 18552 |
+
"step": 2649
|
| 18553 |
+
},
|
| 18554 |
+
{
|
| 18555 |
+
"epoch": 0.9141083132114523,
|
| 18556 |
+
"grad_norm": 1.7224019765853882,
|
| 18557 |
+
"learning_rate": 1.940010162844824e-06,
|
| 18558 |
+
"loss": 9.3694,
|
| 18559 |
+
"step": 2650
|
| 18560 |
+
},
|
| 18561 |
+
{
|
| 18562 |
+
"epoch": 0.9144532597447396,
|
| 18563 |
+
"grad_norm": 0.7851664423942566,
|
| 18564 |
+
"learning_rate": 1.9245595851214328e-06,
|
| 18565 |
+
"loss": 9.6529,
|
| 18566 |
+
"step": 2651
|
| 18567 |
+
},
|
| 18568 |
+
{
|
| 18569 |
+
"epoch": 0.9147982062780269,
|
| 18570 |
+
"grad_norm": 0.9760227203369141,
|
| 18571 |
+
"learning_rate": 1.909169571763908e-06,
|
| 18572 |
+
"loss": 9.5619,
|
| 18573 |
+
"step": 2652
|
| 18574 |
+
},
|
| 18575 |
+
{
|
| 18576 |
+
"epoch": 0.9151431528113142,
|
| 18577 |
+
"grad_norm": 1.0906972885131836,
|
| 18578 |
+
"learning_rate": 1.8938401421602359e-06,
|
| 18579 |
+
"loss": 9.4804,
|
| 18580 |
+
"step": 2653
|
| 18581 |
+
},
|
| 18582 |
+
{
|
| 18583 |
+
"epoch": 0.9154880993446016,
|
| 18584 |
+
"grad_norm": 1.1999458074569702,
|
| 18585 |
+
"learning_rate": 1.8785713156221018e-06,
|
| 18586 |
+
"loss": 9.527,
|
| 18587 |
+
"step": 2654
|
| 18588 |
+
},
|
| 18589 |
+
{
|
| 18590 |
+
"epoch": 0.9158330458778889,
|
| 18591 |
+
"grad_norm": 1.1160448789596558,
|
| 18592 |
+
"learning_rate": 1.863363111384836e-06,
|
| 18593 |
+
"loss": 9.5069,
|
| 18594 |
+
"step": 2655
|
| 18595 |
+
},
|
| 18596 |
+
{
|
| 18597 |
+
"epoch": 0.9161779924111763,
|
| 18598 |
+
"grad_norm": 1.0647850036621094,
|
| 18599 |
+
"learning_rate": 1.8482155486073739e-06,
|
| 18600 |
+
"loss": 9.5693,
|
| 18601 |
+
"step": 2656
|
| 18602 |
+
},
|
| 18603 |
+
{
|
| 18604 |
+
"epoch": 0.9165229389444636,
|
| 18605 |
+
"grad_norm": 1.2221078872680664,
|
| 18606 |
+
"learning_rate": 1.8331286463722951e-06,
|
| 18607 |
+
"loss": 9.4765,
|
| 18608 |
+
"step": 2657
|
| 18609 |
+
},
|
| 18610 |
+
{
|
| 18611 |
+
"epoch": 0.916867885477751,
|
| 18612 |
+
"grad_norm": 1.1700892448425293,
|
| 18613 |
+
"learning_rate": 1.8181024236857246e-06,
|
| 18614 |
+
"loss": 9.5458,
|
| 18615 |
+
"step": 2658
|
| 18616 |
+
},
|
| 18617 |
+
{
|
| 18618 |
+
"epoch": 0.9172128320110383,
|
| 18619 |
+
"grad_norm": 1.2416577339172363,
|
| 18620 |
+
"learning_rate": 1.8031368994773756e-06,
|
| 18621 |
+
"loss": 9.5379,
|
| 18622 |
+
"step": 2659
|
| 18623 |
+
},
|
| 18624 |
+
{
|
| 18625 |
+
"epoch": 0.9175577785443256,
|
| 18626 |
+
"grad_norm": 1.1829198598861694,
|
| 18627 |
+
"learning_rate": 1.788232092600478e-06,
|
| 18628 |
+
"loss": 9.5421,
|
| 18629 |
+
"step": 2660
|
| 18630 |
+
},
|
| 18631 |
+
{
|
| 18632 |
+
"epoch": 0.917902725077613,
|
| 18633 |
+
"grad_norm": 1.1688567399978638,
|
| 18634 |
+
"learning_rate": 1.7733880218317788e-06,
|
| 18635 |
+
"loss": 9.4909,
|
| 18636 |
+
"step": 2661
|
| 18637 |
+
},
|
| 18638 |
+
{
|
| 18639 |
+
"epoch": 0.9182476716109003,
|
| 18640 |
+
"grad_norm": 1.1145144701004028,
|
| 18641 |
+
"learning_rate": 1.7586047058714972e-06,
|
| 18642 |
+
"loss": 9.5701,
|
| 18643 |
+
"step": 2662
|
| 18644 |
+
},
|
| 18645 |
+
{
|
| 18646 |
+
"epoch": 0.9185926181441877,
|
| 18647 |
+
"grad_norm": 1.1775919198989868,
|
| 18648 |
+
"learning_rate": 1.74388216334333e-06,
|
| 18649 |
+
"loss": 9.5376,
|
| 18650 |
+
"step": 2663
|
| 18651 |
+
},
|
| 18652 |
+
{
|
| 18653 |
+
"epoch": 0.918937564677475,
|
| 18654 |
+
"grad_norm": 1.2331610918045044,
|
| 18655 |
+
"learning_rate": 1.7292204127944134e-06,
|
| 18656 |
+
"loss": 9.4995,
|
| 18657 |
+
"step": 2664
|
| 18658 |
+
},
|
| 18659 |
+
{
|
| 18660 |
+
"epoch": 0.9192825112107623,
|
| 18661 |
+
"grad_norm": 1.2776384353637695,
|
| 18662 |
+
"learning_rate": 1.714619472695278e-06,
|
| 18663 |
+
"loss": 9.5589,
|
| 18664 |
+
"step": 2665
|
| 18665 |
+
},
|
| 18666 |
+
{
|
| 18667 |
+
"epoch": 0.9196274577440496,
|
| 18668 |
+
"grad_norm": 1.1205222606658936,
|
| 18669 |
+
"learning_rate": 1.7000793614398714e-06,
|
| 18670 |
+
"loss": 9.586,
|
| 18671 |
+
"step": 2666
|
| 18672 |
+
},
|
| 18673 |
+
{
|
| 18674 |
+
"epoch": 0.9199724042773371,
|
| 18675 |
+
"grad_norm": 1.172733187675476,
|
| 18676 |
+
"learning_rate": 1.6856000973455022e-06,
|
| 18677 |
+
"loss": 9.5626,
|
| 18678 |
+
"step": 2667
|
| 18679 |
+
},
|
| 18680 |
+
{
|
| 18681 |
+
"epoch": 0.9203173508106244,
|
| 18682 |
+
"grad_norm": 1.1725226640701294,
|
| 18683 |
+
"learning_rate": 1.6711816986528238e-06,
|
| 18684 |
+
"loss": 9.5409,
|
| 18685 |
+
"step": 2668
|
| 18686 |
+
},
|
| 18687 |
+
{
|
| 18688 |
+
"epoch": 0.9206622973439117,
|
| 18689 |
+
"grad_norm": 1.3246572017669678,
|
| 18690 |
+
"learning_rate": 1.6568241835258068e-06,
|
| 18691 |
+
"loss": 9.5028,
|
| 18692 |
+
"step": 2669
|
| 18693 |
+
},
|
| 18694 |
+
{
|
| 18695 |
+
"epoch": 0.921007243877199,
|
| 18696 |
+
"grad_norm": 1.4405431747436523,
|
| 18697 |
+
"learning_rate": 1.6425275700517385e-06,
|
| 18698 |
+
"loss": 9.4395,
|
| 18699 |
+
"step": 2670
|
| 18700 |
+
},
|
| 18701 |
+
{
|
| 18702 |
+
"epoch": 0.9213521904104863,
|
| 18703 |
+
"grad_norm": 1.2822954654693604,
|
| 18704 |
+
"learning_rate": 1.6282918762411614e-06,
|
| 18705 |
+
"loss": 9.5234,
|
| 18706 |
+
"step": 2671
|
| 18707 |
+
},
|
| 18708 |
+
{
|
| 18709 |
+
"epoch": 0.9216971369437738,
|
| 18710 |
+
"grad_norm": 1.3265615701675415,
|
| 18711 |
+
"learning_rate": 1.614117120027886e-06,
|
| 18712 |
+
"loss": 9.4913,
|
| 18713 |
+
"step": 2672
|
| 18714 |
+
},
|
| 18715 |
+
{
|
| 18716 |
+
"epoch": 0.9220420834770611,
|
| 18717 |
+
"grad_norm": 1.2790005207061768,
|
| 18718 |
+
"learning_rate": 1.6000033192689611e-06,
|
| 18719 |
+
"loss": 9.4907,
|
| 18720 |
+
"step": 2673
|
| 18721 |
+
},
|
| 18722 |
+
{
|
| 18723 |
+
"epoch": 0.9223870300103484,
|
| 18724 |
+
"grad_norm": 1.3700817823410034,
|
| 18725 |
+
"learning_rate": 1.5859504917446366e-06,
|
| 18726 |
+
"loss": 9.48,
|
| 18727 |
+
"step": 2674
|
| 18728 |
+
},
|
| 18729 |
+
{
|
| 18730 |
+
"epoch": 0.9227319765436357,
|
| 18731 |
+
"grad_norm": 1.3636354207992554,
|
| 18732 |
+
"learning_rate": 1.5719586551583454e-06,
|
| 18733 |
+
"loss": 9.4867,
|
| 18734 |
+
"step": 2675
|
| 18735 |
+
},
|
| 18736 |
+
{
|
| 18737 |
+
"epoch": 0.9230769230769231,
|
| 18738 |
+
"grad_norm": 1.2950481176376343,
|
| 18739 |
+
"learning_rate": 1.5580278271366878e-06,
|
| 18740 |
+
"loss": 9.4936,
|
| 18741 |
+
"step": 2676
|
| 18742 |
+
},
|
| 18743 |
+
{
|
| 18744 |
+
"epoch": 0.9234218696102104,
|
| 18745 |
+
"grad_norm": 1.3804364204406738,
|
| 18746 |
+
"learning_rate": 1.5441580252294253e-06,
|
| 18747 |
+
"loss": 9.4613,
|
| 18748 |
+
"step": 2677
|
| 18749 |
+
},
|
| 18750 |
+
{
|
| 18751 |
+
"epoch": 0.9237668161434978,
|
| 18752 |
+
"grad_norm": 1.369942307472229,
|
| 18753 |
+
"learning_rate": 1.5303492669094089e-06,
|
| 18754 |
+
"loss": 9.5052,
|
| 18755 |
+
"step": 2678
|
| 18756 |
+
},
|
| 18757 |
+
{
|
| 18758 |
+
"epoch": 0.9241117626767851,
|
| 18759 |
+
"grad_norm": 1.2750253677368164,
|
| 18760 |
+
"learning_rate": 1.5166015695726122e-06,
|
| 18761 |
+
"loss": 9.4562,
|
| 18762 |
+
"step": 2679
|
| 18763 |
+
},
|
| 18764 |
+
{
|
| 18765 |
+
"epoch": 0.9244567092100724,
|
| 18766 |
+
"grad_norm": 1.348370909690857,
|
| 18767 |
+
"learning_rate": 1.5029149505380646e-06,
|
| 18768 |
+
"loss": 9.4602,
|
| 18769 |
+
"step": 2680
|
| 18770 |
+
},
|
| 18771 |
+
{
|
| 18772 |
+
"epoch": 0.9248016557433598,
|
| 18773 |
+
"grad_norm": 1.2539187669754028,
|
| 18774 |
+
"learning_rate": 1.4892894270478853e-06,
|
| 18775 |
+
"loss": 9.525,
|
| 18776 |
+
"step": 2681
|
| 18777 |
+
},
|
| 18778 |
+
{
|
| 18779 |
+
"epoch": 0.9251466022766471,
|
| 18780 |
+
"grad_norm": 1.474637508392334,
|
| 18781 |
+
"learning_rate": 1.4757250162671822e-06,
|
| 18782 |
+
"loss": 9.449,
|
| 18783 |
+
"step": 2682
|
| 18784 |
+
},
|
| 18785 |
+
{
|
| 18786 |
+
"epoch": 0.9254915488099345,
|
| 18787 |
+
"grad_norm": 1.484521508216858,
|
| 18788 |
+
"learning_rate": 1.4622217352841138e-06,
|
| 18789 |
+
"loss": 9.3875,
|
| 18790 |
+
"step": 2683
|
| 18791 |
+
},
|
| 18792 |
+
{
|
| 18793 |
+
"epoch": 0.9258364953432218,
|
| 18794 |
+
"grad_norm": 1.2464600801467896,
|
| 18795 |
+
"learning_rate": 1.448779601109801e-06,
|
| 18796 |
+
"loss": 9.4846,
|
| 18797 |
+
"step": 2684
|
| 18798 |
+
},
|
| 18799 |
+
{
|
| 18800 |
+
"epoch": 0.9261814418765092,
|
| 18801 |
+
"grad_norm": 1.404441237449646,
|
| 18802 |
+
"learning_rate": 1.4353986306783418e-06,
|
| 18803 |
+
"loss": 9.516,
|
| 18804 |
+
"step": 2685
|
| 18805 |
+
},
|
| 18806 |
+
{
|
| 18807 |
+
"epoch": 0.9265263884097965,
|
| 18808 |
+
"grad_norm": 1.3945449590682983,
|
| 18809 |
+
"learning_rate": 1.4220788408468021e-06,
|
| 18810 |
+
"loss": 9.4727,
|
| 18811 |
+
"step": 2686
|
| 18812 |
+
},
|
| 18813 |
+
{
|
| 18814 |
+
"epoch": 0.9268713349430838,
|
| 18815 |
+
"grad_norm": 1.3472886085510254,
|
| 18816 |
+
"learning_rate": 1.4088202483951374e-06,
|
| 18817 |
+
"loss": 9.4367,
|
| 18818 |
+
"step": 2687
|
| 18819 |
+
},
|
| 18820 |
+
{
|
| 18821 |
+
"epoch": 0.9272162814763711,
|
| 18822 |
+
"grad_norm": 1.377700686454773,
|
| 18823 |
+
"learning_rate": 1.3956228700262252e-06,
|
| 18824 |
+
"loss": 9.5139,
|
| 18825 |
+
"step": 2688
|
| 18826 |
+
},
|
| 18827 |
+
{
|
| 18828 |
+
"epoch": 0.9275612280096585,
|
| 18829 |
+
"grad_norm": 1.389609456062317,
|
| 18830 |
+
"learning_rate": 1.3824867223658388e-06,
|
| 18831 |
+
"loss": 9.4322,
|
| 18832 |
+
"step": 2689
|
| 18833 |
+
},
|
| 18834 |
+
{
|
| 18835 |
+
"epoch": 0.9279061745429459,
|
| 18836 |
+
"grad_norm": 1.4232558012008667,
|
| 18837 |
+
"learning_rate": 1.3694118219626074e-06,
|
| 18838 |
+
"loss": 9.5109,
|
| 18839 |
+
"step": 2690
|
| 18840 |
+
},
|
| 18841 |
+
{
|
| 18842 |
+
"epoch": 0.9282511210762332,
|
| 18843 |
+
"grad_norm": 1.3743571043014526,
|
| 18844 |
+
"learning_rate": 1.3563981852879827e-06,
|
| 18845 |
+
"loss": 9.4921,
|
| 18846 |
+
"step": 2691
|
| 18847 |
+
},
|
| 18848 |
+
{
|
| 18849 |
+
"epoch": 0.9285960676095205,
|
| 18850 |
+
"grad_norm": 1.4875822067260742,
|
| 18851 |
+
"learning_rate": 1.3434458287362672e-06,
|
| 18852 |
+
"loss": 9.463,
|
| 18853 |
+
"step": 2692
|
| 18854 |
+
},
|
| 18855 |
+
{
|
| 18856 |
+
"epoch": 0.9289410141428078,
|
| 18857 |
+
"grad_norm": 1.3445839881896973,
|
| 18858 |
+
"learning_rate": 1.3305547686245422e-06,
|
| 18859 |
+
"loss": 9.4212,
|
| 18860 |
+
"step": 2693
|
| 18861 |
+
},
|
| 18862 |
+
{
|
| 18863 |
+
"epoch": 0.9292859606760953,
|
| 18864 |
+
"grad_norm": 1.6044926643371582,
|
| 18865 |
+
"learning_rate": 1.3177250211926728e-06,
|
| 18866 |
+
"loss": 9.4379,
|
| 18867 |
+
"step": 2694
|
| 18868 |
+
},
|
| 18869 |
+
{
|
| 18870 |
+
"epoch": 0.9296309072093826,
|
| 18871 |
+
"grad_norm": 1.5688241720199585,
|
| 18872 |
+
"learning_rate": 1.3049566026033023e-06,
|
| 18873 |
+
"loss": 9.4778,
|
| 18874 |
+
"step": 2695
|
| 18875 |
+
},
|
| 18876 |
+
{
|
| 18877 |
+
"epoch": 0.9299758537426699,
|
| 18878 |
+
"grad_norm": 1.6058199405670166,
|
| 18879 |
+
"learning_rate": 1.2922495289417913e-06,
|
| 18880 |
+
"loss": 9.421,
|
| 18881 |
+
"step": 2696
|
| 18882 |
+
},
|
| 18883 |
+
{
|
| 18884 |
+
"epoch": 0.9303208002759572,
|
| 18885 |
+
"grad_norm": 1.693510890007019,
|
| 18886 |
+
"learning_rate": 1.2796038162162239e-06,
|
| 18887 |
+
"loss": 9.354,
|
| 18888 |
+
"step": 2697
|
| 18889 |
+
},
|
| 18890 |
+
{
|
| 18891 |
+
"epoch": 0.9306657468092445,
|
| 18892 |
+
"grad_norm": 1.5772984027862549,
|
| 18893 |
+
"learning_rate": 1.2670194803573954e-06,
|
| 18894 |
+
"loss": 9.3735,
|
| 18895 |
+
"step": 2698
|
| 18896 |
+
},
|
| 18897 |
+
{
|
| 18898 |
+
"epoch": 0.931010693342532,
|
| 18899 |
+
"grad_norm": 1.614292860031128,
|
| 18900 |
+
"learning_rate": 1.2544965372187635e-06,
|
| 18901 |
+
"loss": 9.3702,
|
| 18902 |
+
"step": 2699
|
| 18903 |
+
},
|
| 18904 |
+
{
|
| 18905 |
+
"epoch": 0.9313556398758193,
|
| 18906 |
+
"grad_norm": 1.9061168432235718,
|
| 18907 |
+
"learning_rate": 1.2420350025764528e-06,
|
| 18908 |
+
"loss": 9.2773,
|
| 18909 |
+
"step": 2700
|
| 18910 |
+
},
|
| 18911 |
+
{
|
| 18912 |
+
"epoch": 0.9317005864091066,
|
| 18913 |
+
"grad_norm": 0.8620436787605286,
|
| 18914 |
+
"learning_rate": 1.2296348921292333e-06,
|
| 18915 |
+
"loss": 9.6332,
|
| 18916 |
+
"step": 2701
|
| 18917 |
+
},
|
| 18918 |
+
{
|
| 18919 |
+
"epoch": 0.9320455329423939,
|
| 18920 |
+
"grad_norm": 0.8850879073143005,
|
| 18921 |
+
"learning_rate": 1.2172962214984763e-06,
|
| 18922 |
+
"loss": 9.6099,
|
| 18923 |
+
"step": 2702
|
| 18924 |
+
},
|
| 18925 |
+
{
|
| 18926 |
+
"epoch": 0.9323904794756813,
|
| 18927 |
+
"grad_norm": 1.0009554624557495,
|
| 18928 |
+
"learning_rate": 1.2050190062281752e-06,
|
| 18929 |
+
"loss": 9.5471,
|
| 18930 |
+
"step": 2703
|
| 18931 |
+
},
|
| 18932 |
+
{
|
| 18933 |
+
"epoch": 0.9327354260089686,
|
| 18934 |
+
"grad_norm": 1.0762373208999634,
|
| 18935 |
+
"learning_rate": 1.1928032617848805e-06,
|
| 18936 |
+
"loss": 9.5482,
|
| 18937 |
+
"step": 2704
|
| 18938 |
+
},
|
| 18939 |
+
{
|
| 18940 |
+
"epoch": 0.933080372542256,
|
| 18941 |
+
"grad_norm": 1.0936650037765503,
|
| 18942 |
+
"learning_rate": 1.1806490035577267e-06,
|
| 18943 |
+
"loss": 9.5399,
|
| 18944 |
+
"step": 2705
|
| 18945 |
+
},
|
| 18946 |
+
{
|
| 18947 |
+
"epoch": 0.9334253190755433,
|
| 18948 |
+
"grad_norm": 1.043621301651001,
|
| 18949 |
+
"learning_rate": 1.16855624685836e-06,
|
| 18950 |
+
"loss": 9.5344,
|
| 18951 |
+
"step": 2706
|
| 18952 |
+
},
|
| 18953 |
+
{
|
| 18954 |
+
"epoch": 0.9337702656088306,
|
| 18955 |
+
"grad_norm": 1.2308648824691772,
|
| 18956 |
+
"learning_rate": 1.1565250069209776e-06,
|
| 18957 |
+
"loss": 9.5064,
|
| 18958 |
+
"step": 2707
|
| 18959 |
+
},
|
| 18960 |
+
{
|
| 18961 |
+
"epoch": 0.934115212142118,
|
| 18962 |
+
"grad_norm": 1.162726640701294,
|
| 18963 |
+
"learning_rate": 1.1445552989022668e-06,
|
| 18964 |
+
"loss": 9.5275,
|
| 18965 |
+
"step": 2708
|
| 18966 |
+
},
|
| 18967 |
+
{
|
| 18968 |
+
"epoch": 0.9344601586754053,
|
| 18969 |
+
"grad_norm": 1.1094176769256592,
|
| 18970 |
+
"learning_rate": 1.132647137881393e-06,
|
| 18971 |
+
"loss": 9.5538,
|
| 18972 |
+
"step": 2709
|
| 18973 |
+
},
|
| 18974 |
+
{
|
| 18975 |
+
"epoch": 0.9348051052086926,
|
| 18976 |
+
"grad_norm": 1.255376935005188,
|
| 18977 |
+
"learning_rate": 1.120800538859995e-06,
|
| 18978 |
+
"loss": 9.463,
|
| 18979 |
+
"step": 2710
|
| 18980 |
+
},
|
| 18981 |
+
{
|
| 18982 |
+
"epoch": 0.93515005174198,
|
| 18983 |
+
"grad_norm": 1.1070146560668945,
|
| 18984 |
+
"learning_rate": 1.1090155167621518e-06,
|
| 18985 |
+
"loss": 9.5253,
|
| 18986 |
+
"step": 2711
|
| 18987 |
+
},
|
| 18988 |
+
{
|
| 18989 |
+
"epoch": 0.9354949982752674,
|
| 18990 |
+
"grad_norm": 1.3558119535446167,
|
| 18991 |
+
"learning_rate": 1.0972920864343705e-06,
|
| 18992 |
+
"loss": 9.4959,
|
| 18993 |
+
"step": 2712
|
| 18994 |
+
},
|
| 18995 |
+
{
|
| 18996 |
+
"epoch": 0.9358399448085547,
|
| 18997 |
+
"grad_norm": 1.2252329587936401,
|
| 18998 |
+
"learning_rate": 1.085630262645565e-06,
|
| 18999 |
+
"loss": 9.5362,
|
| 19000 |
+
"step": 2713
|
| 19001 |
+
},
|
| 19002 |
+
{
|
| 19003 |
+
"epoch": 0.936184891341842,
|
| 19004 |
+
"grad_norm": 1.1986970901489258,
|
| 19005 |
+
"learning_rate": 1.07403006008705e-06,
|
| 19006 |
+
"loss": 9.4938,
|
| 19007 |
+
"step": 2714
|
| 19008 |
+
},
|
| 19009 |
+
{
|
| 19010 |
+
"epoch": 0.9365298378751293,
|
| 19011 |
+
"grad_norm": 1.1853275299072266,
|
| 19012 |
+
"learning_rate": 1.062491493372486e-06,
|
| 19013 |
+
"loss": 9.5321,
|
| 19014 |
+
"step": 2715
|
| 19015 |
+
},
|
| 19016 |
+
{
|
| 19017 |
+
"epoch": 0.9368747844084166,
|
| 19018 |
+
"grad_norm": 1.169337511062622,
|
| 19019 |
+
"learning_rate": 1.0510145770379177e-06,
|
| 19020 |
+
"loss": 9.5066,
|
| 19021 |
+
"step": 2716
|
| 19022 |
+
},
|
| 19023 |
+
{
|
| 19024 |
+
"epoch": 0.9372197309417041,
|
| 19025 |
+
"grad_norm": 1.2148215770721436,
|
| 19026 |
+
"learning_rate": 1.0395993255416957e-06,
|
| 19027 |
+
"loss": 9.5333,
|
| 19028 |
+
"step": 2717
|
| 19029 |
+
},
|
| 19030 |
+
{
|
| 19031 |
+
"epoch": 0.9375646774749914,
|
| 19032 |
+
"grad_norm": 1.3632956743240356,
|
| 19033 |
+
"learning_rate": 1.0282457532645119e-06,
|
| 19034 |
+
"loss": 9.4924,
|
| 19035 |
+
"step": 2718
|
| 19036 |
+
},
|
| 19037 |
+
{
|
| 19038 |
+
"epoch": 0.9379096240082787,
|
| 19039 |
+
"grad_norm": 1.1406636238098145,
|
| 19040 |
+
"learning_rate": 1.0169538745093242e-06,
|
| 19041 |
+
"loss": 9.5274,
|
| 19042 |
+
"step": 2719
|
| 19043 |
+
},
|
| 19044 |
+
{
|
| 19045 |
+
"epoch": 0.938254570541566,
|
| 19046 |
+
"grad_norm": 1.1914732456207275,
|
| 19047 |
+
"learning_rate": 1.0057237035014044e-06,
|
| 19048 |
+
"loss": 9.4695,
|
| 19049 |
+
"step": 2720
|
| 19050 |
+
},
|
| 19051 |
+
{
|
| 19052 |
+
"epoch": 0.9385995170748535,
|
| 19053 |
+
"grad_norm": 1.4516353607177734,
|
| 19054 |
+
"learning_rate": 9.945552543882685e-07,
|
| 19055 |
+
"loss": 9.4253,
|
| 19056 |
+
"step": 2721
|
| 19057 |
+
},
|
| 19058 |
+
{
|
| 19059 |
+
"epoch": 0.9389444636081408,
|
| 19060 |
+
"grad_norm": 1.3653576374053955,
|
| 19061 |
+
"learning_rate": 9.834485412396677e-07,
|
| 19062 |
+
"loss": 9.4781,
|
| 19063 |
+
"step": 2722
|
| 19064 |
+
},
|
| 19065 |
+
{
|
| 19066 |
+
"epoch": 0.9392894101414281,
|
| 19067 |
+
"grad_norm": 1.1982650756835938,
|
| 19068 |
+
"learning_rate": 9.724035780476092e-07,
|
| 19069 |
+
"loss": 9.599,
|
| 19070 |
+
"step": 2723
|
| 19071 |
+
},
|
| 19072 |
+
{
|
| 19073 |
+
"epoch": 0.9396343566747154,
|
| 19074 |
+
"grad_norm": 1.244680643081665,
|
| 19075 |
+
"learning_rate": 9.6142037872628e-07,
|
| 19076 |
+
"loss": 9.5415,
|
| 19077 |
+
"step": 2724
|
| 19078 |
+
},
|
| 19079 |
+
{
|
| 19080 |
+
"epoch": 0.9399793032080027,
|
| 19081 |
+
"grad_norm": 1.288313388824463,
|
| 19082 |
+
"learning_rate": 9.504989571120726e-07,
|
| 19083 |
+
"loss": 9.5385,
|
| 19084 |
+
"step": 2725
|
| 19085 |
+
},
|
| 19086 |
+
{
|
| 19087 |
+
"epoch": 0.9403242497412901,
|
| 19088 |
+
"grad_norm": 1.185012698173523,
|
| 19089 |
+
"learning_rate": 9.396393269635484e-07,
|
| 19090 |
+
"loss": 9.5646,
|
| 19091 |
+
"step": 2726
|
| 19092 |
+
},
|
| 19093 |
+
{
|
| 19094 |
+
"epoch": 0.9406691962745775,
|
| 19095 |
+
"grad_norm": 1.2785496711730957,
|
| 19096 |
+
"learning_rate": 9.28841501961425e-07,
|
| 19097 |
+
"loss": 9.496,
|
| 19098 |
+
"step": 2727
|
| 19099 |
+
},
|
| 19100 |
+
{
|
| 19101 |
+
"epoch": 0.9410141428078648,
|
| 19102 |
+
"grad_norm": 1.2989834547042847,
|
| 19103 |
+
"learning_rate": 9.1810549570856e-07,
|
| 19104 |
+
"loss": 9.5301,
|
| 19105 |
+
"step": 2728
|
| 19106 |
+
},
|
| 19107 |
+
{
|
| 19108 |
+
"epoch": 0.9413590893411521,
|
| 19109 |
+
"grad_norm": 1.2136149406433105,
|
| 19110 |
+
"learning_rate": 9.074313217299457e-07,
|
| 19111 |
+
"loss": 9.5197,
|
| 19112 |
+
"step": 2729
|
| 19113 |
+
},
|
| 19114 |
+
{
|
| 19115 |
+
"epoch": 0.9417040358744395,
|
| 19116 |
+
"grad_norm": 1.4056475162506104,
|
| 19117 |
+
"learning_rate": 8.968189934726534e-07,
|
| 19118 |
+
"loss": 9.5294,
|
| 19119 |
+
"step": 2730
|
| 19120 |
+
},
|
| 19121 |
+
{
|
| 19122 |
+
"epoch": 0.9420489824077268,
|
| 19123 |
+
"grad_norm": 1.2778812646865845,
|
| 19124 |
+
"learning_rate": 8.862685243058666e-07,
|
| 19125 |
+
"loss": 9.455,
|
| 19126 |
+
"step": 2731
|
| 19127 |
+
},
|
| 19128 |
+
{
|
| 19129 |
+
"epoch": 0.9423939289410141,
|
| 19130 |
+
"grad_norm": 1.3463318347930908,
|
| 19131 |
+
"learning_rate": 8.757799275208311e-07,
|
| 19132 |
+
"loss": 9.5346,
|
| 19133 |
+
"step": 2732
|
| 19134 |
+
},
|
| 19135 |
+
{
|
| 19136 |
+
"epoch": 0.9427388754743015,
|
| 19137 |
+
"grad_norm": 1.2876229286193848,
|
| 19138 |
+
"learning_rate": 8.653532163308387e-07,
|
| 19139 |
+
"loss": 9.5036,
|
| 19140 |
+
"step": 2733
|
| 19141 |
+
},
|
| 19142 |
+
{
|
| 19143 |
+
"epoch": 0.9430838220075888,
|
| 19144 |
+
"grad_norm": 1.3723164796829224,
|
| 19145 |
+
"learning_rate": 8.549884038712375e-07,
|
| 19146 |
+
"loss": 9.5081,
|
| 19147 |
+
"step": 2734
|
| 19148 |
+
},
|
| 19149 |
+
{
|
| 19150 |
+
"epoch": 0.9434287685408762,
|
| 19151 |
+
"grad_norm": 1.4172887802124023,
|
| 19152 |
+
"learning_rate": 8.446855031993717e-07,
|
| 19153 |
+
"loss": 9.4852,
|
| 19154 |
+
"step": 2735
|
| 19155 |
+
},
|
| 19156 |
+
{
|
| 19157 |
+
"epoch": 0.9437737150741635,
|
| 19158 |
+
"grad_norm": 1.4235061407089233,
|
| 19159 |
+
"learning_rate": 8.344445272946199e-07,
|
| 19160 |
+
"loss": 9.486,
|
| 19161 |
+
"step": 2736
|
| 19162 |
+
},
|
| 19163 |
+
{
|
| 19164 |
+
"epoch": 0.9441186616074508,
|
| 19165 |
+
"grad_norm": 1.449661374092102,
|
| 19166 |
+
"learning_rate": 8.24265489058329e-07,
|
| 19167 |
+
"loss": 9.4349,
|
| 19168 |
+
"step": 2737
|
| 19169 |
+
},
|
| 19170 |
+
{
|
| 19171 |
+
"epoch": 0.9444636081407382,
|
| 19172 |
+
"grad_norm": 1.4169197082519531,
|
| 19173 |
+
"learning_rate": 8.1414840131383e-07,
|
| 19174 |
+
"loss": 9.4753,
|
| 19175 |
+
"step": 2738
|
| 19176 |
+
},
|
| 19177 |
+
{
|
| 19178 |
+
"epoch": 0.9448085546740256,
|
| 19179 |
+
"grad_norm": 1.4273600578308105,
|
| 19180 |
+
"learning_rate": 8.040932768063947e-07,
|
| 19181 |
+
"loss": 9.4446,
|
| 19182 |
+
"step": 2739
|
| 19183 |
+
},
|
| 19184 |
+
{
|
| 19185 |
+
"epoch": 0.9451535012073129,
|
| 19186 |
+
"grad_norm": 1.4585797786712646,
|
| 19187 |
+
"learning_rate": 7.941001282032512e-07,
|
| 19188 |
+
"loss": 9.4683,
|
| 19189 |
+
"step": 2740
|
| 19190 |
+
},
|
| 19191 |
+
{
|
| 19192 |
+
"epoch": 0.9454984477406002,
|
| 19193 |
+
"grad_norm": 1.433382272720337,
|
| 19194 |
+
"learning_rate": 7.841689680935349e-07,
|
| 19195 |
+
"loss": 9.4656,
|
| 19196 |
+
"step": 2741
|
| 19197 |
+
},
|
| 19198 |
+
{
|
| 19199 |
+
"epoch": 0.9458433942738875,
|
| 19200 |
+
"grad_norm": 1.298423409461975,
|
| 19201 |
+
"learning_rate": 7.742998089883102e-07,
|
| 19202 |
+
"loss": 9.4483,
|
| 19203 |
+
"step": 2742
|
| 19204 |
+
},
|
| 19205 |
+
{
|
| 19206 |
+
"epoch": 0.9461883408071748,
|
| 19207 |
+
"grad_norm": 1.4828376770019531,
|
| 19208 |
+
"learning_rate": 7.644926633205208e-07,
|
| 19209 |
+
"loss": 9.4068,
|
| 19210 |
+
"step": 2743
|
| 19211 |
+
},
|
| 19212 |
+
{
|
| 19213 |
+
"epoch": 0.9465332873404623,
|
| 19214 |
+
"grad_norm": 1.4661272764205933,
|
| 19215 |
+
"learning_rate": 7.547475434449835e-07,
|
| 19216 |
+
"loss": 9.4218,
|
| 19217 |
+
"step": 2744
|
| 19218 |
+
},
|
| 19219 |
+
{
|
| 19220 |
+
"epoch": 0.9468782338737496,
|
| 19221 |
+
"grad_norm": 1.4254372119903564,
|
| 19222 |
+
"learning_rate": 7.450644616383951e-07,
|
| 19223 |
+
"loss": 9.4599,
|
| 19224 |
+
"step": 2745
|
| 19225 |
+
},
|
| 19226 |
+
{
|
| 19227 |
+
"epoch": 0.9472231804070369,
|
| 19228 |
+
"grad_norm": 1.6096229553222656,
|
| 19229 |
+
"learning_rate": 7.354434300992752e-07,
|
| 19230 |
+
"loss": 9.4529,
|
| 19231 |
+
"step": 2746
|
| 19232 |
+
},
|
| 19233 |
+
{
|
| 19234 |
+
"epoch": 0.9475681269403242,
|
| 19235 |
+
"grad_norm": 1.6057524681091309,
|
| 19236 |
+
"learning_rate": 7.258844609479953e-07,
|
| 19237 |
+
"loss": 9.3971,
|
| 19238 |
+
"step": 2747
|
| 19239 |
+
},
|
| 19240 |
+
{
|
| 19241 |
+
"epoch": 0.9479130734736116,
|
| 19242 |
+
"grad_norm": 1.5125082731246948,
|
| 19243 |
+
"learning_rate": 7.163875662267117e-07,
|
| 19244 |
+
"loss": 9.417,
|
| 19245 |
+
"step": 2748
|
| 19246 |
+
},
|
| 19247 |
+
{
|
| 19248 |
+
"epoch": 0.948258020006899,
|
| 19249 |
+
"grad_norm": 1.7600480318069458,
|
| 19250 |
+
"learning_rate": 7.069527578994151e-07,
|
| 19251 |
+
"loss": 9.2774,
|
| 19252 |
+
"step": 2749
|
| 19253 |
+
},
|
| 19254 |
+
{
|
| 19255 |
+
"epoch": 0.9486029665401863,
|
| 19256 |
+
"grad_norm": 1.6905750036239624,
|
| 19257 |
+
"learning_rate": 6.975800478518646e-07,
|
| 19258 |
+
"loss": 9.3521,
|
| 19259 |
+
"step": 2750
|
| 19260 |
+
},
|
| 19261 |
+
{
|
| 19262 |
+
"epoch": 0.9489479130734736,
|
| 19263 |
+
"grad_norm": 0.99737948179245,
|
| 19264 |
+
"learning_rate": 6.88269447891593e-07,
|
| 19265 |
+
"loss": 9.533,
|
| 19266 |
+
"step": 2751
|
| 19267 |
+
},
|
| 19268 |
+
{
|
| 19269 |
+
"epoch": 0.9492928596067609,
|
| 19270 |
+
"grad_norm": 0.9953950643539429,
|
| 19271 |
+
"learning_rate": 6.790209697478789e-07,
|
| 19272 |
+
"loss": 9.5993,
|
| 19273 |
+
"step": 2752
|
| 19274 |
+
},
|
| 19275 |
+
{
|
| 19276 |
+
"epoch": 0.9496378061400483,
|
| 19277 |
+
"grad_norm": 1.0281240940093994,
|
| 19278 |
+
"learning_rate": 6.698346250717524e-07,
|
| 19279 |
+
"loss": 9.5267,
|
| 19280 |
+
"step": 2753
|
| 19281 |
+
},
|
| 19282 |
+
{
|
| 19283 |
+
"epoch": 0.9499827526733357,
|
| 19284 |
+
"grad_norm": 1.020925760269165,
|
| 19285 |
+
"learning_rate": 6.607104254359675e-07,
|
| 19286 |
+
"loss": 9.5643,
|
| 19287 |
+
"step": 2754
|
| 19288 |
+
},
|
| 19289 |
+
{
|
| 19290 |
+
"epoch": 0.950327699206623,
|
| 19291 |
+
"grad_norm": 1.0711194276809692,
|
| 19292 |
+
"learning_rate": 6.516483823349795e-07,
|
| 19293 |
+
"loss": 9.5639,
|
| 19294 |
+
"step": 2755
|
| 19295 |
+
},
|
| 19296 |
+
{
|
| 19297 |
+
"epoch": 0.9506726457399103,
|
| 19298 |
+
"grad_norm": 1.0857421159744263,
|
| 19299 |
+
"learning_rate": 6.426485071849564e-07,
|
| 19300 |
+
"loss": 9.4995,
|
| 19301 |
+
"step": 2756
|
| 19302 |
+
},
|
| 19303 |
+
{
|
| 19304 |
+
"epoch": 0.9510175922731977,
|
| 19305 |
+
"grad_norm": 1.262204885482788,
|
| 19306 |
+
"learning_rate": 6.337108113237344e-07,
|
| 19307 |
+
"loss": 9.4864,
|
| 19308 |
+
"step": 2757
|
| 19309 |
+
},
|
| 19310 |
+
{
|
| 19311 |
+
"epoch": 0.951362538806485,
|
| 19312 |
+
"grad_norm": 1.0766891241073608,
|
| 19313 |
+
"learning_rate": 6.248353060108292e-07,
|
| 19314 |
+
"loss": 9.5252,
|
| 19315 |
+
"step": 2758
|
| 19316 |
+
},
|
| 19317 |
+
{
|
| 19318 |
+
"epoch": 0.9517074853397723,
|
| 19319 |
+
"grad_norm": 1.153833031654358,
|
| 19320 |
+
"learning_rate": 6.160220024273966e-07,
|
| 19321 |
+
"loss": 9.5178,
|
| 19322 |
+
"step": 2759
|
| 19323 |
+
},
|
| 19324 |
+
{
|
| 19325 |
+
"epoch": 0.9520524318730597,
|
| 19326 |
+
"grad_norm": 1.1806145906448364,
|
| 19327 |
+
"learning_rate": 6.072709116762442e-07,
|
| 19328 |
+
"loss": 9.538,
|
| 19329 |
+
"step": 2760
|
| 19330 |
+
},
|
| 19331 |
+
{
|
| 19332 |
+
"epoch": 0.952397378406347,
|
| 19333 |
+
"grad_norm": 1.146377682685852,
|
| 19334 |
+
"learning_rate": 5.98582044781798e-07,
|
| 19335 |
+
"loss": 9.534,
|
| 19336 |
+
"step": 2761
|
| 19337 |
+
},
|
| 19338 |
+
{
|
| 19339 |
+
"epoch": 0.9527423249396344,
|
| 19340 |
+
"grad_norm": 1.1945569515228271,
|
| 19341 |
+
"learning_rate": 5.899554126901075e-07,
|
| 19342 |
+
"loss": 9.5208,
|
| 19343 |
+
"step": 2762
|
| 19344 |
+
},
|
| 19345 |
+
{
|
| 19346 |
+
"epoch": 0.9530872714729217,
|
| 19347 |
+
"grad_norm": 1.286086916923523,
|
| 19348 |
+
"learning_rate": 5.813910262687905e-07,
|
| 19349 |
+
"loss": 9.4976,
|
| 19350 |
+
"step": 2763
|
| 19351 |
+
},
|
| 19352 |
+
{
|
| 19353 |
+
"epoch": 0.953432218006209,
|
| 19354 |
+
"grad_norm": 1.2750720977783203,
|
| 19355 |
+
"learning_rate": 5.728888963070945e-07,
|
| 19356 |
+
"loss": 9.5186,
|
| 19357 |
+
"step": 2764
|
| 19358 |
+
},
|
| 19359 |
+
{
|
| 19360 |
+
"epoch": 0.9537771645394963,
|
| 19361 |
+
"grad_norm": 1.2140814065933228,
|
| 19362 |
+
"learning_rate": 5.644490335157959e-07,
|
| 19363 |
+
"loss": 9.4896,
|
| 19364 |
+
"step": 2765
|
| 19365 |
+
},
|
| 19366 |
+
{
|
| 19367 |
+
"epoch": 0.9541221110727838,
|
| 19368 |
+
"grad_norm": 1.1700353622436523,
|
| 19369 |
+
"learning_rate": 5.560714485272512e-07,
|
| 19370 |
+
"loss": 9.6212,
|
| 19371 |
+
"step": 2766
|
| 19372 |
+
},
|
| 19373 |
+
{
|
| 19374 |
+
"epoch": 0.9544670576060711,
|
| 19375 |
+
"grad_norm": 1.3271950483322144,
|
| 19376 |
+
"learning_rate": 5.477561518953566e-07,
|
| 19377 |
+
"loss": 9.4967,
|
| 19378 |
+
"step": 2767
|
| 19379 |
+
},
|
| 19380 |
+
{
|
| 19381 |
+
"epoch": 0.9548120041393584,
|
| 19382 |
+
"grad_norm": 1.1459087133407593,
|
| 19383 |
+
"learning_rate": 5.395031540955275e-07,
|
| 19384 |
+
"loss": 9.5539,
|
| 19385 |
+
"step": 2768
|
| 19386 |
+
},
|
| 19387 |
+
{
|
| 19388 |
+
"epoch": 0.9551569506726457,
|
| 19389 |
+
"grad_norm": 1.122616171836853,
|
| 19390 |
+
"learning_rate": 5.313124655247192e-07,
|
| 19391 |
+
"loss": 9.5894,
|
| 19392 |
+
"step": 2769
|
| 19393 |
+
},
|
| 19394 |
+
{
|
| 19395 |
+
"epoch": 0.955501897205933,
|
| 19396 |
+
"grad_norm": 1.3789931535720825,
|
| 19397 |
+
"learning_rate": 5.231840965013668e-07,
|
| 19398 |
+
"loss": 9.4521,
|
| 19399 |
+
"step": 2770
|
| 19400 |
+
},
|
| 19401 |
+
{
|
| 19402 |
+
"epoch": 0.9558468437392205,
|
| 19403 |
+
"grad_norm": 1.2291425466537476,
|
| 19404 |
+
"learning_rate": 5.151180572654235e-07,
|
| 19405 |
+
"loss": 9.5169,
|
| 19406 |
+
"step": 2771
|
| 19407 |
+
},
|
| 19408 |
+
{
|
| 19409 |
+
"epoch": 0.9561917902725078,
|
| 19410 |
+
"grad_norm": 1.1856980323791504,
|
| 19411 |
+
"learning_rate": 5.071143579782889e-07,
|
| 19412 |
+
"loss": 9.4942,
|
| 19413 |
+
"step": 2772
|
| 19414 |
+
},
|
| 19415 |
+
{
|
| 19416 |
+
"epoch": 0.9565367368057951,
|
| 19417 |
+
"grad_norm": 1.2218326330184937,
|
| 19418 |
+
"learning_rate": 4.99173008722853e-07,
|
| 19419 |
+
"loss": 9.55,
|
| 19420 |
+
"step": 2773
|
| 19421 |
+
},
|
| 19422 |
+
{
|
| 19423 |
+
"epoch": 0.9568816833390824,
|
| 19424 |
+
"grad_norm": 1.2583118677139282,
|
| 19425 |
+
"learning_rate": 4.91294019503441e-07,
|
| 19426 |
+
"loss": 9.538,
|
| 19427 |
+
"step": 2774
|
| 19428 |
+
},
|
| 19429 |
+
{
|
| 19430 |
+
"epoch": 0.9572266298723697,
|
| 19431 |
+
"grad_norm": 1.3726310729980469,
|
| 19432 |
+
"learning_rate": 4.834774002458409e-07,
|
| 19433 |
+
"loss": 9.4831,
|
| 19434 |
+
"step": 2775
|
| 19435 |
+
},
|
| 19436 |
+
{
|
| 19437 |
+
"epoch": 0.9575715764056572,
|
| 19438 |
+
"grad_norm": 1.3480925559997559,
|
| 19439 |
+
"learning_rate": 4.757231607972534e-07,
|
| 19440 |
+
"loss": 9.4674,
|
| 19441 |
+
"step": 2776
|
| 19442 |
+
},
|
| 19443 |
+
{
|
| 19444 |
+
"epoch": 0.9579165229389445,
|
| 19445 |
+
"grad_norm": 1.2078763246536255,
|
| 19446 |
+
"learning_rate": 4.680313109262813e-07,
|
| 19447 |
+
"loss": 9.4641,
|
| 19448 |
+
"step": 2777
|
| 19449 |
+
},
|
| 19450 |
+
{
|
| 19451 |
+
"epoch": 0.9582614694722318,
|
| 19452 |
+
"grad_norm": 1.2556449174880981,
|
| 19453 |
+
"learning_rate": 4.6040186032296206e-07,
|
| 19454 |
+
"loss": 9.4801,
|
| 19455 |
+
"step": 2778
|
| 19456 |
+
},
|
| 19457 |
+
{
|
| 19458 |
+
"epoch": 0.9586064160055191,
|
| 19459 |
+
"grad_norm": 1.2832894325256348,
|
| 19460 |
+
"learning_rate": 4.5283481859869635e-07,
|
| 19461 |
+
"loss": 9.5116,
|
| 19462 |
+
"step": 2779
|
| 19463 |
+
},
|
| 19464 |
+
{
|
| 19465 |
+
"epoch": 0.9589513625388065,
|
| 19466 |
+
"grad_norm": 1.3231313228607178,
|
| 19467 |
+
"learning_rate": 4.4533019528628093e-07,
|
| 19468 |
+
"loss": 9.4913,
|
| 19469 |
+
"step": 2780
|
| 19470 |
+
},
|
| 19471 |
+
{
|
| 19472 |
+
"epoch": 0.9592963090720938,
|
| 19473 |
+
"grad_norm": 1.2611724138259888,
|
| 19474 |
+
"learning_rate": 4.3788799983986997e-07,
|
| 19475 |
+
"loss": 9.521,
|
| 19476 |
+
"step": 2781
|
| 19477 |
+
},
|
| 19478 |
+
{
|
| 19479 |
+
"epoch": 0.9596412556053812,
|
| 19480 |
+
"grad_norm": 1.2805002927780151,
|
| 19481 |
+
"learning_rate": 4.305082416349804e-07,
|
| 19482 |
+
"loss": 9.502,
|
| 19483 |
+
"step": 2782
|
| 19484 |
+
},
|
| 19485 |
+
{
|
| 19486 |
+
"epoch": 0.9599862021386685,
|
| 19487 |
+
"grad_norm": 1.3639451265335083,
|
| 19488 |
+
"learning_rate": 4.231909299684533e-07,
|
| 19489 |
+
"loss": 9.4736,
|
| 19490 |
+
"step": 2783
|
| 19491 |
+
},
|
| 19492 |
+
{
|
| 19493 |
+
"epoch": 0.9603311486719558,
|
| 19494 |
+
"grad_norm": 1.4248766899108887,
|
| 19495 |
+
"learning_rate": 4.159360740584817e-07,
|
| 19496 |
+
"loss": 9.5154,
|
| 19497 |
+
"step": 2784
|
| 19498 |
+
},
|
| 19499 |
+
{
|
| 19500 |
+
"epoch": 0.9606760952052432,
|
| 19501 |
+
"grad_norm": 1.5050346851348877,
|
| 19502 |
+
"learning_rate": 4.0874368304457676e-07,
|
| 19503 |
+
"loss": 9.4806,
|
| 19504 |
+
"step": 2785
|
| 19505 |
+
},
|
| 19506 |
+
{
|
| 19507 |
+
"epoch": 0.9610210417385305,
|
| 19508 |
+
"grad_norm": 1.28929603099823,
|
| 19509 |
+
"learning_rate": 4.016137659875463e-07,
|
| 19510 |
+
"loss": 9.5005,
|
| 19511 |
+
"step": 2786
|
| 19512 |
+
},
|
| 19513 |
+
{
|
| 19514 |
+
"epoch": 0.9613659882718179,
|
| 19515 |
+
"grad_norm": 1.1755974292755127,
|
| 19516 |
+
"learning_rate": 3.945463318695053e-07,
|
| 19517 |
+
"loss": 9.4845,
|
| 19518 |
+
"step": 2787
|
| 19519 |
+
},
|
| 19520 |
+
{
|
| 19521 |
+
"epoch": 0.9617109348051052,
|
| 19522 |
+
"grad_norm": 1.4888191223144531,
|
| 19523 |
+
"learning_rate": 3.8754138959383733e-07,
|
| 19524 |
+
"loss": 9.4333,
|
| 19525 |
+
"step": 2788
|
| 19526 |
+
},
|
| 19527 |
+
{
|
| 19528 |
+
"epoch": 0.9620558813383926,
|
| 19529 |
+
"grad_norm": 1.3838120698928833,
|
| 19530 |
+
"learning_rate": 3.805989479852279e-07,
|
| 19531 |
+
"loss": 9.4622,
|
| 19532 |
+
"step": 2789
|
| 19533 |
+
},
|
| 19534 |
+
{
|
| 19535 |
+
"epoch": 0.9624008278716799,
|
| 19536 |
+
"grad_norm": 1.3965057134628296,
|
| 19537 |
+
"learning_rate": 3.7371901578959756e-07,
|
| 19538 |
+
"loss": 9.4135,
|
| 19539 |
+
"step": 2790
|
| 19540 |
+
},
|
| 19541 |
+
{
|
| 19542 |
+
"epoch": 0.9627457744049672,
|
| 19543 |
+
"grad_norm": 1.3354548215866089,
|
| 19544 |
+
"learning_rate": 3.6690160167413554e-07,
|
| 19545 |
+
"loss": 9.4748,
|
| 19546 |
+
"step": 2791
|
| 19547 |
+
},
|
| 19548 |
+
{
|
| 19549 |
+
"epoch": 0.9630907209382545,
|
| 19550 |
+
"grad_norm": 1.4212675094604492,
|
| 19551 |
+
"learning_rate": 3.6014671422727185e-07,
|
| 19552 |
+
"loss": 9.4848,
|
| 19553 |
+
"step": 2792
|
| 19554 |
+
},
|
| 19555 |
+
{
|
| 19556 |
+
"epoch": 0.9634356674715419,
|
| 19557 |
+
"grad_norm": 1.4828577041625977,
|
| 19558 |
+
"learning_rate": 3.5345436195866053e-07,
|
| 19559 |
+
"loss": 9.4245,
|
| 19560 |
+
"step": 2793
|
| 19561 |
+
},
|
| 19562 |
+
{
|
| 19563 |
+
"epoch": 0.9637806140048293,
|
| 19564 |
+
"grad_norm": 1.4509541988372803,
|
| 19565 |
+
"learning_rate": 3.468245532991743e-07,
|
| 19566 |
+
"loss": 9.4091,
|
| 19567 |
+
"step": 2794
|
| 19568 |
+
},
|
| 19569 |
+
{
|
| 19570 |
+
"epoch": 0.9641255605381166,
|
| 19571 |
+
"grad_norm": 1.5361698865890503,
|
| 19572 |
+
"learning_rate": 3.4025729660089877e-07,
|
| 19573 |
+
"loss": 9.4698,
|
| 19574 |
+
"step": 2795
|
| 19575 |
+
},
|
| 19576 |
+
{
|
| 19577 |
+
"epoch": 0.9644705070714039,
|
| 19578 |
+
"grad_norm": 1.5707025527954102,
|
| 19579 |
+
"learning_rate": 3.3375260013711604e-07,
|
| 19580 |
+
"loss": 9.4206,
|
| 19581 |
+
"step": 2796
|
| 19582 |
+
},
|
| 19583 |
+
{
|
| 19584 |
+
"epoch": 0.9648154536046912,
|
| 19585 |
+
"grad_norm": 1.47184157371521,
|
| 19586 |
+
"learning_rate": 3.273104721023046e-07,
|
| 19587 |
+
"loss": 9.3935,
|
| 19588 |
+
"step": 2797
|
| 19589 |
+
},
|
| 19590 |
+
{
|
| 19591 |
+
"epoch": 0.9651604001379787,
|
| 19592 |
+
"grad_norm": 1.5686029195785522,
|
| 19593 |
+
"learning_rate": 3.209309206121058e-07,
|
| 19594 |
+
"loss": 9.3636,
|
| 19595 |
+
"step": 2798
|
| 19596 |
+
},
|
| 19597 |
+
{
|
| 19598 |
+
"epoch": 0.965505346671266,
|
| 19599 |
+
"grad_norm": 1.5438534021377563,
|
| 19600 |
+
"learning_rate": 3.1461395370334104e-07,
|
| 19601 |
+
"loss": 9.3861,
|
| 19602 |
+
"step": 2799
|
| 19603 |
+
},
|
| 19604 |
+
{
|
| 19605 |
+
"epoch": 0.9658502932045533,
|
| 19606 |
+
"grad_norm": 1.6539479494094849,
|
| 19607 |
+
"learning_rate": 3.0835957933397773e-07,
|
| 19608 |
+
"loss": 9.398,
|
| 19609 |
+
"step": 2800
|
| 19610 |
+
},
|
| 19611 |
+
{
|
| 19612 |
+
"epoch": 0.9661952397378406,
|
| 19613 |
+
"grad_norm": 0.774260401725769,
|
| 19614 |
+
"learning_rate": 3.0216780538314116e-07,
|
| 19615 |
+
"loss": 9.6004,
|
| 19616 |
+
"step": 2801
|
| 19617 |
+
},
|
| 19618 |
+
{
|
| 19619 |
+
"epoch": 0.9665401862711279,
|
| 19620 |
+
"grad_norm": 0.9886148571968079,
|
| 19621 |
+
"learning_rate": 2.960386396510972e-07,
|
| 19622 |
+
"loss": 9.5417,
|
| 19623 |
+
"step": 2802
|
| 19624 |
+
},
|
| 19625 |
+
{
|
| 19626 |
+
"epoch": 0.9668851328044153,
|
| 19627 |
+
"grad_norm": 1.0178803205490112,
|
| 19628 |
+
"learning_rate": 2.8997208985921953e-07,
|
| 19629 |
+
"loss": 9.5666,
|
| 19630 |
+
"step": 2803
|
| 19631 |
+
},
|
| 19632 |
+
{
|
| 19633 |
+
"epoch": 0.9672300793377027,
|
| 19634 |
+
"grad_norm": 1.0676816701889038,
|
| 19635 |
+
"learning_rate": 2.8396816365001687e-07,
|
| 19636 |
+
"loss": 9.499,
|
| 19637 |
+
"step": 2804
|
| 19638 |
+
},
|
| 19639 |
+
{
|
| 19640 |
+
"epoch": 0.96757502587099,
|
| 19641 |
+
"grad_norm": 1.0291966199874878,
|
| 19642 |
+
"learning_rate": 2.7802686858710016e-07,
|
| 19643 |
+
"loss": 9.5364,
|
| 19644 |
+
"step": 2805
|
| 19645 |
+
},
|
| 19646 |
+
{
|
| 19647 |
+
"epoch": 0.9679199724042773,
|
| 19648 |
+
"grad_norm": 1.100340723991394,
|
| 19649 |
+
"learning_rate": 2.7214821215518214e-07,
|
| 19650 |
+
"loss": 9.5105,
|
| 19651 |
+
"step": 2806
|
| 19652 |
+
},
|
| 19653 |
+
{
|
| 19654 |
+
"epoch": 0.9682649189375647,
|
| 19655 |
+
"grad_norm": 1.1685614585876465,
|
| 19656 |
+
"learning_rate": 2.6633220176006667e-07,
|
| 19657 |
+
"loss": 9.5038,
|
| 19658 |
+
"step": 2807
|
| 19659 |
+
},
|
| 19660 |
+
{
|
| 19661 |
+
"epoch": 0.968609865470852,
|
| 19662 |
+
"grad_norm": 1.1775288581848145,
|
| 19663 |
+
"learning_rate": 2.6057884472862617e-07,
|
| 19664 |
+
"loss": 9.5383,
|
| 19665 |
+
"step": 2808
|
| 19666 |
+
},
|
| 19667 |
+
{
|
| 19668 |
+
"epoch": 0.9689548120041394,
|
| 19669 |
+
"grad_norm": 1.2399282455444336,
|
| 19670 |
+
"learning_rate": 2.548881483088128e-07,
|
| 19671 |
+
"loss": 9.4643,
|
| 19672 |
+
"step": 2809
|
| 19673 |
+
},
|
| 19674 |
+
{
|
| 19675 |
+
"epoch": 0.9692997585374267,
|
| 19676 |
+
"grad_norm": 1.2756638526916504,
|
| 19677 |
+
"learning_rate": 2.49260119669642e-07,
|
| 19678 |
+
"loss": 9.4916,
|
| 19679 |
+
"step": 2810
|
| 19680 |
+
},
|
| 19681 |
+
{
|
| 19682 |
+
"epoch": 0.969644705070714,
|
| 19683 |
+
"grad_norm": 1.3724730014801025,
|
| 19684 |
+
"learning_rate": 2.4369476590118123e-07,
|
| 19685 |
+
"loss": 9.4872,
|
| 19686 |
+
"step": 2811
|
| 19687 |
+
},
|
| 19688 |
+
{
|
| 19689 |
+
"epoch": 0.9699896516040014,
|
| 19690 |
+
"grad_norm": 1.1623852252960205,
|
| 19691 |
+
"learning_rate": 2.381920940145277e-07,
|
| 19692 |
+
"loss": 9.5659,
|
| 19693 |
+
"step": 2812
|
| 19694 |
+
},
|
| 19695 |
+
{
|
| 19696 |
+
"epoch": 0.9703345981372887,
|
| 19697 |
+
"grad_norm": 1.3232759237289429,
|
| 19698 |
+
"learning_rate": 2.3275211094183623e-07,
|
| 19699 |
+
"loss": 9.4726,
|
| 19700 |
+
"step": 2813
|
| 19701 |
+
},
|
| 19702 |
+
{
|
| 19703 |
+
"epoch": 0.970679544670576,
|
| 19704 |
+
"grad_norm": 1.105460286140442,
|
| 19705 |
+
"learning_rate": 2.2737482353626937e-07,
|
| 19706 |
+
"loss": 9.5364,
|
| 19707 |
+
"step": 2814
|
| 19708 |
+
},
|
| 19709 |
+
{
|
| 19710 |
+
"epoch": 0.9710244912038634,
|
| 19711 |
+
"grad_norm": 1.130436897277832,
|
| 19712 |
+
"learning_rate": 2.2206023857201385e-07,
|
| 19713 |
+
"loss": 9.5565,
|
| 19714 |
+
"step": 2815
|
| 19715 |
+
},
|
| 19716 |
+
{
|
| 19717 |
+
"epoch": 0.9713694377371508,
|
| 19718 |
+
"grad_norm": 1.2147849798202515,
|
| 19719 |
+
"learning_rate": 2.1680836274426962e-07,
|
| 19720 |
+
"loss": 9.5194,
|
| 19721 |
+
"step": 2816
|
| 19722 |
+
},
|
| 19723 |
+
{
|
| 19724 |
+
"epoch": 0.9717143842704381,
|
| 19725 |
+
"grad_norm": 1.2784674167633057,
|
| 19726 |
+
"learning_rate": 2.1161920266922763e-07,
|
| 19727 |
+
"loss": 9.4549,
|
| 19728 |
+
"step": 2817
|
| 19729 |
+
},
|
| 19730 |
+
{
|
| 19731 |
+
"epoch": 0.9720593308037254,
|
| 19732 |
+
"grad_norm": 1.3010627031326294,
|
| 19733 |
+
"learning_rate": 2.0649276488408086e-07,
|
| 19734 |
+
"loss": 9.4602,
|
| 19735 |
+
"step": 2818
|
| 19736 |
+
},
|
| 19737 |
+
{
|
| 19738 |
+
"epoch": 0.9724042773370127,
|
| 19739 |
+
"grad_norm": 1.2807101011276245,
|
| 19740 |
+
"learning_rate": 2.014290558469911e-07,
|
| 19741 |
+
"loss": 9.5456,
|
| 19742 |
+
"step": 2819
|
| 19743 |
+
},
|
| 19744 |
+
{
|
| 19745 |
+
"epoch": 0.9727492238703,
|
| 19746 |
+
"grad_norm": 1.301635503768921,
|
| 19747 |
+
"learning_rate": 1.964280819371167e-07,
|
| 19748 |
+
"loss": 9.5044,
|
| 19749 |
+
"step": 2820
|
| 19750 |
+
},
|
| 19751 |
+
{
|
| 19752 |
+
"epoch": 0.9730941704035875,
|
| 19753 |
+
"grad_norm": 1.1405683755874634,
|
| 19754 |
+
"learning_rate": 1.914898494545736e-07,
|
| 19755 |
+
"loss": 9.5604,
|
| 19756 |
+
"step": 2821
|
| 19757 |
+
},
|
| 19758 |
+
{
|
| 19759 |
+
"epoch": 0.9734391169368748,
|
| 19760 |
+
"grad_norm": 1.261732816696167,
|
| 19761 |
+
"learning_rate": 1.8661436462042437e-07,
|
| 19762 |
+
"loss": 9.4851,
|
| 19763 |
+
"step": 2822
|
| 19764 |
+
},
|
| 19765 |
+
{
|
| 19766 |
+
"epoch": 0.9737840634701621,
|
| 19767 |
+
"grad_norm": 1.2781423330307007,
|
| 19768 |
+
"learning_rate": 1.8180163357671143e-07,
|
| 19769 |
+
"loss": 9.5201,
|
| 19770 |
+
"step": 2823
|
| 19771 |
+
},
|
| 19772 |
+
{
|
| 19773 |
+
"epoch": 0.9741290100034494,
|
| 19774 |
+
"grad_norm": 1.2555526494979858,
|
| 19775 |
+
"learning_rate": 1.7705166238639047e-07,
|
| 19776 |
+
"loss": 9.4532,
|
| 19777 |
+
"step": 2824
|
| 19778 |
+
},
|
| 19779 |
+
{
|
| 19780 |
+
"epoch": 0.9744739565367369,
|
| 19781 |
+
"grad_norm": 1.2804704904556274,
|
| 19782 |
+
"learning_rate": 1.7236445703338044e-07,
|
| 19783 |
+
"loss": 9.5293,
|
| 19784 |
+
"step": 2825
|
| 19785 |
+
},
|
| 19786 |
+
{
|
| 19787 |
+
"epoch": 0.9748189030700242,
|
| 19788 |
+
"grad_norm": 1.3036941289901733,
|
| 19789 |
+
"learning_rate": 1.677400234225135e-07,
|
| 19790 |
+
"loss": 9.536,
|
| 19791 |
+
"step": 2826
|
| 19792 |
+
},
|
| 19793 |
+
{
|
| 19794 |
+
"epoch": 0.9751638496033115,
|
| 19795 |
+
"grad_norm": 1.2297347784042358,
|
| 19796 |
+
"learning_rate": 1.6317836737955172e-07,
|
| 19797 |
+
"loss": 9.496,
|
| 19798 |
+
"step": 2827
|
| 19799 |
+
},
|
| 19800 |
+
{
|
| 19801 |
+
"epoch": 0.9755087961365988,
|
| 19802 |
+
"grad_norm": 1.2707699537277222,
|
| 19803 |
+
"learning_rate": 1.586794946511594e-07,
|
| 19804 |
+
"loss": 9.4632,
|
| 19805 |
+
"step": 2828
|
| 19806 |
+
},
|
| 19807 |
+
{
|
| 19808 |
+
"epoch": 0.9758537426698861,
|
| 19809 |
+
"grad_norm": 1.3694789409637451,
|
| 19810 |
+
"learning_rate": 1.542434109049251e-07,
|
| 19811 |
+
"loss": 9.4874,
|
| 19812 |
+
"step": 2829
|
| 19813 |
+
},
|
| 19814 |
+
{
|
| 19815 |
+
"epoch": 0.9761986892031735,
|
| 19816 |
+
"grad_norm": 1.3027503490447998,
|
| 19817 |
+
"learning_rate": 1.4987012172932302e-07,
|
| 19818 |
+
"loss": 9.4417,
|
| 19819 |
+
"step": 2830
|
| 19820 |
+
},
|
| 19821 |
+
{
|
| 19822 |
+
"epoch": 0.9765436357364609,
|
| 19823 |
+
"grad_norm": 1.300711750984192,
|
| 19824 |
+
"learning_rate": 1.4555963263372385e-07,
|
| 19825 |
+
"loss": 9.5223,
|
| 19826 |
+
"step": 2831
|
| 19827 |
+
},
|
| 19828 |
+
{
|
| 19829 |
+
"epoch": 0.9768885822697482,
|
| 19830 |
+
"grad_norm": 1.29239821434021,
|
| 19831 |
+
"learning_rate": 1.413119490483894e-07,
|
| 19832 |
+
"loss": 9.4713,
|
| 19833 |
+
"step": 2832
|
| 19834 |
+
},
|
| 19835 |
+
{
|
| 19836 |
+
"epoch": 0.9772335288030355,
|
| 19837 |
+
"grad_norm": 1.3773186206817627,
|
| 19838 |
+
"learning_rate": 1.3712707632445032e-07,
|
| 19839 |
+
"loss": 9.4819,
|
| 19840 |
+
"step": 2833
|
| 19841 |
+
},
|
| 19842 |
+
{
|
| 19843 |
+
"epoch": 0.9775784753363229,
|
| 19844 |
+
"grad_norm": 1.4483076333999634,
|
| 19845 |
+
"learning_rate": 1.3300501973392277e-07,
|
| 19846 |
+
"loss": 9.4047,
|
| 19847 |
+
"step": 2834
|
| 19848 |
+
},
|
| 19849 |
+
{
|
| 19850 |
+
"epoch": 0.9779234218696102,
|
| 19851 |
+
"grad_norm": 1.4170414209365845,
|
| 19852 |
+
"learning_rate": 1.2894578446968065e-07,
|
| 19853 |
+
"loss": 9.5096,
|
| 19854 |
+
"step": 2835
|
| 19855 |
+
},
|
| 19856 |
+
{
|
| 19857 |
+
"epoch": 0.9782683684028975,
|
| 19858 |
+
"grad_norm": 1.5087085962295532,
|
| 19859 |
+
"learning_rate": 1.2494937564545562e-07,
|
| 19860 |
+
"loss": 9.4874,
|
| 19861 |
+
"step": 2836
|
| 19862 |
+
},
|
| 19863 |
+
{
|
| 19864 |
+
"epoch": 0.9786133149361849,
|
| 19865 |
+
"grad_norm": 1.5026702880859375,
|
| 19866 |
+
"learning_rate": 1.2101579829583154e-07,
|
| 19867 |
+
"loss": 9.4319,
|
| 19868 |
+
"step": 2837
|
| 19869 |
+
},
|
| 19870 |
+
{
|
| 19871 |
+
"epoch": 0.9789582614694722,
|
| 19872 |
+
"grad_norm": 1.3315801620483398,
|
| 19873 |
+
"learning_rate": 1.1714505737625004e-07,
|
| 19874 |
+
"loss": 9.4616,
|
| 19875 |
+
"step": 2838
|
| 19876 |
+
},
|
| 19877 |
+
{
|
| 19878 |
+
"epoch": 0.9793032080027596,
|
| 19879 |
+
"grad_norm": 1.5373655557632446,
|
| 19880 |
+
"learning_rate": 1.133371577629716e-07,
|
| 19881 |
+
"loss": 9.4924,
|
| 19882 |
+
"step": 2839
|
| 19883 |
+
},
|
| 19884 |
+
{
|
| 19885 |
+
"epoch": 0.9796481545360469,
|
| 19886 |
+
"grad_norm": 1.4027314186096191,
|
| 19887 |
+
"learning_rate": 1.095921042531145e-07,
|
| 19888 |
+
"loss": 9.3959,
|
| 19889 |
+
"step": 2840
|
| 19890 |
+
},
|
| 19891 |
+
{
|
| 19892 |
+
"epoch": 0.9799931010693342,
|
| 19893 |
+
"grad_norm": 1.4827989339828491,
|
| 19894 |
+
"learning_rate": 1.0590990156461034e-07,
|
| 19895 |
+
"loss": 9.4222,
|
| 19896 |
+
"step": 2841
|
| 19897 |
+
},
|
| 19898 |
+
{
|
| 19899 |
+
"epoch": 0.9803380476026216,
|
| 19900 |
+
"grad_norm": 1.365110158920288,
|
| 19901 |
+
"learning_rate": 1.022905543362096e-07,
|
| 19902 |
+
"loss": 9.4457,
|
| 19903 |
+
"step": 2842
|
| 19904 |
+
},
|
| 19905 |
+
{
|
| 19906 |
+
"epoch": 0.980682994135909,
|
| 19907 |
+
"grad_norm": 1.4800422191619873,
|
| 19908 |
+
"learning_rate": 9.873406712749279e-08,
|
| 19909 |
+
"loss": 9.4845,
|
| 19910 |
+
"step": 2843
|
| 19911 |
+
},
|
| 19912 |
+
{
|
| 19913 |
+
"epoch": 0.9810279406691963,
|
| 19914 |
+
"grad_norm": 1.482553243637085,
|
| 19915 |
+
"learning_rate": 9.524044441883706e-08,
|
| 19916 |
+
"loss": 9.4586,
|
| 19917 |
+
"step": 2844
|
| 19918 |
+
},
|
| 19919 |
+
{
|
| 19920 |
+
"epoch": 0.9813728872024836,
|
| 19921 |
+
"grad_norm": 1.5155751705169678,
|
| 19922 |
+
"learning_rate": 9.180969061143852e-08,
|
| 19923 |
+
"loss": 9.4402,
|
| 19924 |
+
"step": 2845
|
| 19925 |
+
},
|
| 19926 |
+
{
|
| 19927 |
+
"epoch": 0.9817178337357709,
|
| 19928 |
+
"grad_norm": 1.5355010032653809,
|
| 19929 |
+
"learning_rate": 8.844181002727325e-08,
|
| 19930 |
+
"loss": 9.4685,
|
| 19931 |
+
"step": 2846
|
| 19932 |
+
},
|
| 19933 |
+
{
|
| 19934 |
+
"epoch": 0.9820627802690582,
|
| 19935 |
+
"grad_norm": 1.5655204057693481,
|
| 19936 |
+
"learning_rate": 8.513680690913073e-08,
|
| 19937 |
+
"loss": 9.4197,
|
| 19938 |
+
"step": 2847
|
| 19939 |
+
},
|
| 19940 |
+
{
|
| 19941 |
+
"epoch": 0.9824077268023457,
|
| 19942 |
+
"grad_norm": 1.5435409545898438,
|
| 19943 |
+
"learning_rate": 8.189468542057488e-08,
|
| 19944 |
+
"loss": 9.4517,
|
| 19945 |
+
"step": 2848
|
| 19946 |
+
},
|
| 19947 |
+
{
|
| 19948 |
+
"epoch": 0.982752673335633,
|
| 19949 |
+
"grad_norm": 1.6485852003097534,
|
| 19950 |
+
"learning_rate": 7.871544964596633e-08,
|
| 19951 |
+
"loss": 9.4181,
|
| 19952 |
+
"step": 2849
|
| 19953 |
+
},
|
| 19954 |
+
{
|
| 19955 |
+
"epoch": 0.9830976198689203,
|
| 19956 |
+
"grad_norm": 1.6750664710998535,
|
| 19957 |
+
"learning_rate": 7.559910359042355e-08,
|
| 19958 |
+
"loss": 9.361,
|
| 19959 |
+
"step": 2850
|
| 19960 |
+
},
|
| 19961 |
+
{
|
| 19962 |
+
"epoch": 0.9834425664022076,
|
| 19963 |
+
"grad_norm": 0.9129282832145691,
|
| 19964 |
+
"learning_rate": 7.254565117985613e-08,
|
| 19965 |
+
"loss": 9.5804,
|
| 19966 |
+
"step": 2851
|
| 19967 |
+
},
|
| 19968 |
+
{
|
| 19969 |
+
"epoch": 0.983787512935495,
|
| 19970 |
+
"grad_norm": 1.0424609184265137,
|
| 19971 |
+
"learning_rate": 6.955509626093703e-08,
|
| 19972 |
+
"loss": 9.5371,
|
| 19973 |
+
"step": 2852
|
| 19974 |
+
},
|
| 19975 |
+
{
|
| 19976 |
+
"epoch": 0.9841324594687824,
|
| 19977 |
+
"grad_norm": 0.9472970366477966,
|
| 19978 |
+
"learning_rate": 6.662744260109155e-08,
|
| 19979 |
+
"loss": 9.6361,
|
| 19980 |
+
"step": 2853
|
| 19981 |
+
},
|
| 19982 |
+
{
|
| 19983 |
+
"epoch": 0.9844774060020697,
|
| 19984 |
+
"grad_norm": 1.1693047285079956,
|
| 19985 |
+
"learning_rate": 6.376269388852496e-08,
|
| 19986 |
+
"loss": 9.4819,
|
| 19987 |
+
"step": 2854
|
| 19988 |
+
},
|
| 19989 |
+
{
|
| 19990 |
+
"epoch": 0.984822352535357,
|
| 19991 |
+
"grad_norm": 0.99554044008255,
|
| 19992 |
+
"learning_rate": 6.096085373217264e-08,
|
| 19993 |
+
"loss": 9.5079,
|
| 19994 |
+
"step": 2855
|
| 19995 |
+
},
|
| 19996 |
+
{
|
| 19997 |
+
"epoch": 0.9851672990686443,
|
| 19998 |
+
"grad_norm": 0.9872909188270569,
|
| 19999 |
+
"learning_rate": 5.822192566173334e-08,
|
| 20000 |
+
"loss": 9.5455,
|
| 20001 |
+
"step": 2856
|
| 20002 |
+
},
|
| 20003 |
+
{
|
| 20004 |
+
"epoch": 0.9855122456019317,
|
| 20005 |
+
"grad_norm": 1.08110773563385,
|
| 20006 |
+
"learning_rate": 5.554591312765811e-08,
|
| 20007 |
+
"loss": 9.5288,
|
| 20008 |
+
"step": 2857
|
| 20009 |
+
},
|
| 20010 |
+
{
|
| 20011 |
+
"epoch": 0.985857192135219,
|
| 20012 |
+
"grad_norm": 1.1271926164627075,
|
| 20013 |
+
"learning_rate": 5.2932819501111395e-08,
|
| 20014 |
+
"loss": 9.5112,
|
| 20015 |
+
"step": 2858
|
| 20016 |
+
},
|
| 20017 |
+
{
|
| 20018 |
+
"epoch": 0.9862021386685064,
|
| 20019 |
+
"grad_norm": 1.2304562330245972,
|
| 20020 |
+
"learning_rate": 5.038264807402105e-08,
|
| 20021 |
+
"loss": 9.5027,
|
| 20022 |
+
"step": 2859
|
| 20023 |
+
},
|
| 20024 |
+
{
|
| 20025 |
+
"epoch": 0.9865470852017937,
|
| 20026 |
+
"grad_norm": 1.0815626382827759,
|
| 20027 |
+
"learning_rate": 4.789540205902832e-08,
|
| 20028 |
+
"loss": 9.5418,
|
| 20029 |
+
"step": 2860
|
| 20030 |
+
},
|
| 20031 |
+
{
|
| 20032 |
+
"epoch": 0.9868920317350811,
|
| 20033 |
+
"grad_norm": 1.279129147529602,
|
| 20034 |
+
"learning_rate": 4.547108458951566e-08,
|
| 20035 |
+
"loss": 9.5028,
|
| 20036 |
+
"step": 2861
|
| 20037 |
+
},
|
| 20038 |
+
{
|
| 20039 |
+
"epoch": 0.9872369782683684,
|
| 20040 |
+
"grad_norm": 1.1527676582336426,
|
| 20041 |
+
"learning_rate": 4.310969871958448e-08,
|
| 20042 |
+
"loss": 9.541,
|
| 20043 |
+
"step": 2862
|
| 20044 |
+
},
|
| 20045 |
+
{
|
| 20046 |
+
"epoch": 0.9875819248016557,
|
| 20047 |
+
"grad_norm": 1.260764241218567,
|
| 20048 |
+
"learning_rate": 4.0811247424049625e-08,
|
| 20049 |
+
"loss": 9.4835,
|
| 20050 |
+
"step": 2863
|
| 20051 |
+
},
|
| 20052 |
+
{
|
| 20053 |
+
"epoch": 0.9879268713349431,
|
| 20054 |
+
"grad_norm": 1.153011679649353,
|
| 20055 |
+
"learning_rate": 3.857573359845601e-08,
|
| 20056 |
+
"loss": 9.5011,
|
| 20057 |
+
"step": 2864
|
| 20058 |
+
},
|
| 20059 |
+
{
|
| 20060 |
+
"epoch": 0.9882718178682304,
|
| 20061 |
+
"grad_norm": 1.124871015548706,
|
| 20062 |
+
"learning_rate": 3.6403160059050865e-08,
|
| 20063 |
+
"loss": 9.5657,
|
| 20064 |
+
"step": 2865
|
| 20065 |
+
},
|
| 20066 |
+
{
|
| 20067 |
+
"epoch": 0.9886167644015178,
|
| 20068 |
+
"grad_norm": 1.1694023609161377,
|
| 20069 |
+
"learning_rate": 3.4293529542800406e-08,
|
| 20070 |
+
"loss": 9.4759,
|
| 20071 |
+
"step": 2866
|
| 20072 |
+
},
|
| 20073 |
+
{
|
| 20074 |
+
"epoch": 0.9889617109348051,
|
| 20075 |
+
"grad_norm": 1.1712218523025513,
|
| 20076 |
+
"learning_rate": 3.224684470735651e-08,
|
| 20077 |
+
"loss": 9.5351,
|
| 20078 |
+
"step": 2867
|
| 20079 |
+
},
|
| 20080 |
+
{
|
| 20081 |
+
"epoch": 0.9893066574680924,
|
| 20082 |
+
"grad_norm": 1.2313039302825928,
|
| 20083 |
+
"learning_rate": 3.0263108131095566e-08,
|
| 20084 |
+
"loss": 9.5316,
|
| 20085 |
+
"step": 2868
|
| 20086 |
}
|
| 20087 |
],
|
| 20088 |
"logging_steps": 1,
|
|
|
|
| 20102 |
"attributes": {}
|
| 20103 |
}
|
| 20104 |
},
|
| 20105 |
+
"total_flos": 595349320237056.0,
|
| 20106 |
"train_batch_size": 4,
|
| 20107 |
"trial_name": null,
|
| 20108 |
"trial_params": null
|