Training in progress, step 900, 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 72396376
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d2e77beca278bbed323654e4748505a06737dd198d15b01bf8ffe802826ecb88
|
| 3 |
size 72396376
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 144903627
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2afa46872c49537f9910ac9b4c25cee71441514db37d8a18567d2067699cb072
|
| 3 |
size 144903627
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e39111815cccf95fa47da6bf39c2ea3d1d65fbed26ef6c0116b3e3704b3c778f
|
| 3 |
+
size 14581
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1337
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4aea7e1e23a759eea4f6e68fc962f7860f8560d63692b96bf20701ff4436e21f
|
| 3 |
size 1337
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,10 +1,10 @@
|
|
| 1 |
{
|
| 2 |
-
"best_global_step":
|
| 3 |
-
"best_metric": 1.
|
| 4 |
-
"best_model_checkpoint":
|
| 5 |
-
"epoch": 0.
|
| 6 |
"eval_steps": 75,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -3588,6 +3588,2902 @@
|
|
| 3588 |
"learning_rate": 0.0,
|
| 3589 |
"loss": 1.5317,
|
| 3590 |
"step": 500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3591 |
}
|
| 3592 |
],
|
| 3593 |
"logging_steps": 1,
|
|
@@ -3602,12 +6498,12 @@
|
|
| 3602 |
"should_evaluate": false,
|
| 3603 |
"should_log": false,
|
| 3604 |
"should_save": true,
|
| 3605 |
-
"should_training_stop":
|
| 3606 |
},
|
| 3607 |
"attributes": {}
|
| 3608 |
}
|
| 3609 |
},
|
| 3610 |
-
"total_flos":
|
| 3611 |
"train_batch_size": 4,
|
| 3612 |
"trial_name": null,
|
| 3613 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_global_step": 900,
|
| 3 |
+
"best_metric": 1.46342933177948,
|
| 4 |
+
"best_model_checkpoint": "core/outputs/663d8eb4-5a22-49f9-99b2-65b09be5f21c/checkpoint-900",
|
| 5 |
+
"epoch": 0.01725096317877748,
|
| 6 |
"eval_steps": 75,
|
| 7 |
+
"global_step": 900,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 3588 |
"learning_rate": 0.0,
|
| 3589 |
"loss": 1.5317,
|
| 3590 |
"step": 500
|
| 3591 |
+
},
|
| 3592 |
+
{
|
| 3593 |
+
"epoch": 0.009603036169519465,
|
| 3594 |
+
"grad_norm": 0.09532026946544647,
|
| 3595 |
+
"learning_rate": 1.6336019480728855e-06,
|
| 3596 |
+
"loss": 1.5502,
|
| 3597 |
+
"step": 501
|
| 3598 |
+
},
|
| 3599 |
+
{
|
| 3600 |
+
"epoch": 0.009622203906384774,
|
| 3601 |
+
"grad_norm": 0.10625909268856049,
|
| 3602 |
+
"learning_rate": 3.267203896145771e-06,
|
| 3603 |
+
"loss": 1.5264,
|
| 3604 |
+
"step": 502
|
| 3605 |
+
},
|
| 3606 |
+
{
|
| 3607 |
+
"epoch": 0.009641371643250081,
|
| 3608 |
+
"grad_norm": 0.1184196025133133,
|
| 3609 |
+
"learning_rate": 4.900805844218657e-06,
|
| 3610 |
+
"loss": 1.5642,
|
| 3611 |
+
"step": 503
|
| 3612 |
+
},
|
| 3613 |
+
{
|
| 3614 |
+
"epoch": 0.00966053938011539,
|
| 3615 |
+
"grad_norm": 0.08634921908378601,
|
| 3616 |
+
"learning_rate": 6.534407792291542e-06,
|
| 3617 |
+
"loss": 1.1418,
|
| 3618 |
+
"step": 504
|
| 3619 |
+
},
|
| 3620 |
+
{
|
| 3621 |
+
"epoch": 0.009679707116980699,
|
| 3622 |
+
"grad_norm": 0.08168788999319077,
|
| 3623 |
+
"learning_rate": 8.168009740364426e-06,
|
| 3624 |
+
"loss": 1.3338,
|
| 3625 |
+
"step": 505
|
| 3626 |
+
},
|
| 3627 |
+
{
|
| 3628 |
+
"epoch": 0.009698874853846006,
|
| 3629 |
+
"grad_norm": 0.1057819277048111,
|
| 3630 |
+
"learning_rate": 9.801611688437313e-06,
|
| 3631 |
+
"loss": 1.2433,
|
| 3632 |
+
"step": 506
|
| 3633 |
+
},
|
| 3634 |
+
{
|
| 3635 |
+
"epoch": 0.009718042590711315,
|
| 3636 |
+
"grad_norm": 0.09045206010341644,
|
| 3637 |
+
"learning_rate": 1.1435213636510199e-05,
|
| 3638 |
+
"loss": 1.815,
|
| 3639 |
+
"step": 507
|
| 3640 |
+
},
|
| 3641 |
+
{
|
| 3642 |
+
"epoch": 0.009737210327576623,
|
| 3643 |
+
"grad_norm": 0.08739877492189407,
|
| 3644 |
+
"learning_rate": 1.3068815584583084e-05,
|
| 3645 |
+
"loss": 1.3414,
|
| 3646 |
+
"step": 508
|
| 3647 |
+
},
|
| 3648 |
+
{
|
| 3649 |
+
"epoch": 0.00975637806444193,
|
| 3650 |
+
"grad_norm": 0.11166685819625854,
|
| 3651 |
+
"learning_rate": 1.4702417532655969e-05,
|
| 3652 |
+
"loss": 1.6414,
|
| 3653 |
+
"step": 509
|
| 3654 |
+
},
|
| 3655 |
+
{
|
| 3656 |
+
"epoch": 0.00977554580130724,
|
| 3657 |
+
"grad_norm": 0.10074058920145035,
|
| 3658 |
+
"learning_rate": 1.6336019480728853e-05,
|
| 3659 |
+
"loss": 1.6147,
|
| 3660 |
+
"step": 510
|
| 3661 |
+
},
|
| 3662 |
+
{
|
| 3663 |
+
"epoch": 0.009794713538172548,
|
| 3664 |
+
"grad_norm": 0.14303070306777954,
|
| 3665 |
+
"learning_rate": 1.7969621428801738e-05,
|
| 3666 |
+
"loss": 1.4928,
|
| 3667 |
+
"step": 511
|
| 3668 |
+
},
|
| 3669 |
+
{
|
| 3670 |
+
"epoch": 0.009813881275037857,
|
| 3671 |
+
"grad_norm": 0.10371147096157074,
|
| 3672 |
+
"learning_rate": 1.9603223376874627e-05,
|
| 3673 |
+
"loss": 1.3396,
|
| 3674 |
+
"step": 512
|
| 3675 |
+
},
|
| 3676 |
+
{
|
| 3677 |
+
"epoch": 0.009833049011903164,
|
| 3678 |
+
"grad_norm": 0.11297748982906342,
|
| 3679 |
+
"learning_rate": 2.1236825324947512e-05,
|
| 3680 |
+
"loss": 1.3404,
|
| 3681 |
+
"step": 513
|
| 3682 |
+
},
|
| 3683 |
+
{
|
| 3684 |
+
"epoch": 0.009852216748768473,
|
| 3685 |
+
"grad_norm": 0.10189443081617355,
|
| 3686 |
+
"learning_rate": 2.2870427273020397e-05,
|
| 3687 |
+
"loss": 1.6061,
|
| 3688 |
+
"step": 514
|
| 3689 |
+
},
|
| 3690 |
+
{
|
| 3691 |
+
"epoch": 0.009871384485633782,
|
| 3692 |
+
"grad_norm": 0.1111370325088501,
|
| 3693 |
+
"learning_rate": 2.4504029221093282e-05,
|
| 3694 |
+
"loss": 1.1975,
|
| 3695 |
+
"step": 515
|
| 3696 |
+
},
|
| 3697 |
+
{
|
| 3698 |
+
"epoch": 0.009890552222499089,
|
| 3699 |
+
"grad_norm": 0.12068242579698563,
|
| 3700 |
+
"learning_rate": 2.6137631169166168e-05,
|
| 3701 |
+
"loss": 1.723,
|
| 3702 |
+
"step": 516
|
| 3703 |
+
},
|
| 3704 |
+
{
|
| 3705 |
+
"epoch": 0.009909719959364398,
|
| 3706 |
+
"grad_norm": 0.11446703225374222,
|
| 3707 |
+
"learning_rate": 2.7771233117239053e-05,
|
| 3708 |
+
"loss": 1.8779,
|
| 3709 |
+
"step": 517
|
| 3710 |
+
},
|
| 3711 |
+
{
|
| 3712 |
+
"epoch": 0.009928887696229707,
|
| 3713 |
+
"grad_norm": 0.09416982531547546,
|
| 3714 |
+
"learning_rate": 2.9404835065311938e-05,
|
| 3715 |
+
"loss": 1.7864,
|
| 3716 |
+
"step": 518
|
| 3717 |
+
},
|
| 3718 |
+
{
|
| 3719 |
+
"epoch": 0.009948055433095014,
|
| 3720 |
+
"grad_norm": 0.11554675549268723,
|
| 3721 |
+
"learning_rate": 3.103843701338482e-05,
|
| 3722 |
+
"loss": 1.7024,
|
| 3723 |
+
"step": 519
|
| 3724 |
+
},
|
| 3725 |
+
{
|
| 3726 |
+
"epoch": 0.009967223169960323,
|
| 3727 |
+
"grad_norm": 0.10419266670942307,
|
| 3728 |
+
"learning_rate": 3.2672038961457705e-05,
|
| 3729 |
+
"loss": 1.5895,
|
| 3730 |
+
"step": 520
|
| 3731 |
+
},
|
| 3732 |
+
{
|
| 3733 |
+
"epoch": 0.009986390906825631,
|
| 3734 |
+
"grad_norm": 0.10057644546031952,
|
| 3735 |
+
"learning_rate": 3.430564090953059e-05,
|
| 3736 |
+
"loss": 1.5948,
|
| 3737 |
+
"step": 521
|
| 3738 |
+
},
|
| 3739 |
+
{
|
| 3740 |
+
"epoch": 0.010005558643690939,
|
| 3741 |
+
"grad_norm": 0.10852479189634323,
|
| 3742 |
+
"learning_rate": 3.5939242857603476e-05,
|
| 3743 |
+
"loss": 1.617,
|
| 3744 |
+
"step": 522
|
| 3745 |
+
},
|
| 3746 |
+
{
|
| 3747 |
+
"epoch": 0.010024726380556247,
|
| 3748 |
+
"grad_norm": 0.18556846678256989,
|
| 3749 |
+
"learning_rate": 3.757284480567637e-05,
|
| 3750 |
+
"loss": 1.4999,
|
| 3751 |
+
"step": 523
|
| 3752 |
+
},
|
| 3753 |
+
{
|
| 3754 |
+
"epoch": 0.010043894117421556,
|
| 3755 |
+
"grad_norm": 0.08714650571346283,
|
| 3756 |
+
"learning_rate": 3.920644675374925e-05,
|
| 3757 |
+
"loss": 1.4746,
|
| 3758 |
+
"step": 524
|
| 3759 |
+
},
|
| 3760 |
+
{
|
| 3761 |
+
"epoch": 0.010063061854286865,
|
| 3762 |
+
"grad_norm": 0.15116101503372192,
|
| 3763 |
+
"learning_rate": 4.084004870182214e-05,
|
| 3764 |
+
"loss": 1.5842,
|
| 3765 |
+
"step": 525
|
| 3766 |
+
},
|
| 3767 |
+
{
|
| 3768 |
+
"epoch": 0.010063061854286865,
|
| 3769 |
+
"eval_loss": 1.499491572380066,
|
| 3770 |
+
"eval_runtime": 2505.8714,
|
| 3771 |
+
"eval_samples_per_second": 37.012,
|
| 3772 |
+
"eval_steps_per_second": 18.506,
|
| 3773 |
+
"step": 525
|
| 3774 |
+
},
|
| 3775 |
+
{
|
| 3776 |
+
"epoch": 0.010063061854286865,
|
| 3777 |
+
"eval_loss": 1.5245558023452759,
|
| 3778 |
+
"eval_runtime": 2507.2653,
|
| 3779 |
+
"eval_samples_per_second": 36.992,
|
| 3780 |
+
"eval_steps_per_second": 18.496,
|
| 3781 |
+
"step": 525
|
| 3782 |
+
},
|
| 3783 |
+
{
|
| 3784 |
+
"epoch": 0.010082229591152172,
|
| 3785 |
+
"grad_norm": 0.09549321979284286,
|
| 3786 |
+
"learning_rate": 4.2473650649895024e-05,
|
| 3787 |
+
"loss": 2.0039,
|
| 3788 |
+
"step": 526
|
| 3789 |
+
},
|
| 3790 |
+
{
|
| 3791 |
+
"epoch": 0.010101397328017481,
|
| 3792 |
+
"grad_norm": 0.10219141095876694,
|
| 3793 |
+
"learning_rate": 4.410725259796791e-05,
|
| 3794 |
+
"loss": 1.3952,
|
| 3795 |
+
"step": 527
|
| 3796 |
+
},
|
| 3797 |
+
{
|
| 3798 |
+
"epoch": 0.01012056506488279,
|
| 3799 |
+
"grad_norm": 0.08407086879014969,
|
| 3800 |
+
"learning_rate": 4.5740854546040794e-05,
|
| 3801 |
+
"loss": 1.022,
|
| 3802 |
+
"step": 528
|
| 3803 |
+
},
|
| 3804 |
+
{
|
| 3805 |
+
"epoch": 0.010139732801748097,
|
| 3806 |
+
"grad_norm": 0.08318034559488297,
|
| 3807 |
+
"learning_rate": 4.737445649411368e-05,
|
| 3808 |
+
"loss": 1.1778,
|
| 3809 |
+
"step": 529
|
| 3810 |
+
},
|
| 3811 |
+
{
|
| 3812 |
+
"epoch": 0.010158900538613406,
|
| 3813 |
+
"grad_norm": 0.08296982944011688,
|
| 3814 |
+
"learning_rate": 4.9008058442186565e-05,
|
| 3815 |
+
"loss": 1.3715,
|
| 3816 |
+
"step": 530
|
| 3817 |
+
},
|
| 3818 |
+
{
|
| 3819 |
+
"epoch": 0.010178068275478715,
|
| 3820 |
+
"grad_norm": 0.09953635931015015,
|
| 3821 |
+
"learning_rate": 5.064166039025946e-05,
|
| 3822 |
+
"loss": 1.5036,
|
| 3823 |
+
"step": 531
|
| 3824 |
+
},
|
| 3825 |
+
{
|
| 3826 |
+
"epoch": 0.010197236012344022,
|
| 3827 |
+
"grad_norm": 0.0981762632727623,
|
| 3828 |
+
"learning_rate": 5.2275262338332335e-05,
|
| 3829 |
+
"loss": 1.5204,
|
| 3830 |
+
"step": 532
|
| 3831 |
+
},
|
| 3832 |
+
{
|
| 3833 |
+
"epoch": 0.01021640374920933,
|
| 3834 |
+
"grad_norm": 0.14627604186534882,
|
| 3835 |
+
"learning_rate": 5.390886428640523e-05,
|
| 3836 |
+
"loss": 1.5033,
|
| 3837 |
+
"step": 533
|
| 3838 |
+
},
|
| 3839 |
+
{
|
| 3840 |
+
"epoch": 0.01023557148607464,
|
| 3841 |
+
"grad_norm": 0.10711963474750519,
|
| 3842 |
+
"learning_rate": 5.5542466234478106e-05,
|
| 3843 |
+
"loss": 1.7255,
|
| 3844 |
+
"step": 534
|
| 3845 |
+
},
|
| 3846 |
+
{
|
| 3847 |
+
"epoch": 0.010254739222939948,
|
| 3848 |
+
"grad_norm": 0.0885913074016571,
|
| 3849 |
+
"learning_rate": 5.7176068182551e-05,
|
| 3850 |
+
"loss": 1.2385,
|
| 3851 |
+
"step": 535
|
| 3852 |
+
},
|
| 3853 |
+
{
|
| 3854 |
+
"epoch": 0.010273906959805255,
|
| 3855 |
+
"grad_norm": 0.08486426621675491,
|
| 3856 |
+
"learning_rate": 5.8809670130623876e-05,
|
| 3857 |
+
"loss": 1.5619,
|
| 3858 |
+
"step": 536
|
| 3859 |
+
},
|
| 3860 |
+
{
|
| 3861 |
+
"epoch": 0.010293074696670564,
|
| 3862 |
+
"grad_norm": 0.0885050967335701,
|
| 3863 |
+
"learning_rate": 6.044327207869677e-05,
|
| 3864 |
+
"loss": 1.1614,
|
| 3865 |
+
"step": 537
|
| 3866 |
+
},
|
| 3867 |
+
{
|
| 3868 |
+
"epoch": 0.010312242433535873,
|
| 3869 |
+
"grad_norm": 0.12057748436927795,
|
| 3870 |
+
"learning_rate": 6.207687402676964e-05,
|
| 3871 |
+
"loss": 1.5633,
|
| 3872 |
+
"step": 538
|
| 3873 |
+
},
|
| 3874 |
+
{
|
| 3875 |
+
"epoch": 0.01033141017040118,
|
| 3876 |
+
"grad_norm": 0.09845733642578125,
|
| 3877 |
+
"learning_rate": 6.371047597484253e-05,
|
| 3878 |
+
"loss": 1.7861,
|
| 3879 |
+
"step": 539
|
| 3880 |
+
},
|
| 3881 |
+
{
|
| 3882 |
+
"epoch": 0.010350577907266489,
|
| 3883 |
+
"grad_norm": 0.10790391266345978,
|
| 3884 |
+
"learning_rate": 6.534407792291541e-05,
|
| 3885 |
+
"loss": 1.3479,
|
| 3886 |
+
"step": 540
|
| 3887 |
+
},
|
| 3888 |
+
{
|
| 3889 |
+
"epoch": 0.010369745644131798,
|
| 3890 |
+
"grad_norm": 0.11300645023584366,
|
| 3891 |
+
"learning_rate": 6.69776798709883e-05,
|
| 3892 |
+
"loss": 1.7842,
|
| 3893 |
+
"step": 541
|
| 3894 |
+
},
|
| 3895 |
+
{
|
| 3896 |
+
"epoch": 0.010388913380997105,
|
| 3897 |
+
"grad_norm": 0.10352309048175812,
|
| 3898 |
+
"learning_rate": 6.861128181906118e-05,
|
| 3899 |
+
"loss": 1.4253,
|
| 3900 |
+
"step": 542
|
| 3901 |
+
},
|
| 3902 |
+
{
|
| 3903 |
+
"epoch": 0.010408081117862414,
|
| 3904 |
+
"grad_norm": 0.11383506655693054,
|
| 3905 |
+
"learning_rate": 7.024488376713407e-05,
|
| 3906 |
+
"loss": 1.1164,
|
| 3907 |
+
"step": 543
|
| 3908 |
+
},
|
| 3909 |
+
{
|
| 3910 |
+
"epoch": 0.010427248854727723,
|
| 3911 |
+
"grad_norm": 0.14779718220233917,
|
| 3912 |
+
"learning_rate": 7.187848571520695e-05,
|
| 3913 |
+
"loss": 1.4464,
|
| 3914 |
+
"step": 544
|
| 3915 |
+
},
|
| 3916 |
+
{
|
| 3917 |
+
"epoch": 0.01044641659159303,
|
| 3918 |
+
"grad_norm": 0.09846003353595734,
|
| 3919 |
+
"learning_rate": 7.351208766327984e-05,
|
| 3920 |
+
"loss": 1.1064,
|
| 3921 |
+
"step": 545
|
| 3922 |
+
},
|
| 3923 |
+
{
|
| 3924 |
+
"epoch": 0.010465584328458339,
|
| 3925 |
+
"grad_norm": 0.10775422304868698,
|
| 3926 |
+
"learning_rate": 7.514568961135274e-05,
|
| 3927 |
+
"loss": 1.2104,
|
| 3928 |
+
"step": 546
|
| 3929 |
+
},
|
| 3930 |
+
{
|
| 3931 |
+
"epoch": 0.010484752065323647,
|
| 3932 |
+
"grad_norm": 0.10013894736766815,
|
| 3933 |
+
"learning_rate": 7.677929155942561e-05,
|
| 3934 |
+
"loss": 1.4019,
|
| 3935 |
+
"step": 547
|
| 3936 |
+
},
|
| 3937 |
+
{
|
| 3938 |
+
"epoch": 0.010503919802188956,
|
| 3939 |
+
"grad_norm": 0.10368606448173523,
|
| 3940 |
+
"learning_rate": 7.84128935074985e-05,
|
| 3941 |
+
"loss": 1.2399,
|
| 3942 |
+
"step": 548
|
| 3943 |
+
},
|
| 3944 |
+
{
|
| 3945 |
+
"epoch": 0.010523087539054263,
|
| 3946 |
+
"grad_norm": 0.0994802713394165,
|
| 3947 |
+
"learning_rate": 8.004649545557138e-05,
|
| 3948 |
+
"loss": 1.43,
|
| 3949 |
+
"step": 549
|
| 3950 |
+
},
|
| 3951 |
+
{
|
| 3952 |
+
"epoch": 0.010542255275919572,
|
| 3953 |
+
"grad_norm": 0.10836998373270035,
|
| 3954 |
+
"learning_rate": 8.168009740364428e-05,
|
| 3955 |
+
"loss": 1.0461,
|
| 3956 |
+
"step": 550
|
| 3957 |
+
},
|
| 3958 |
+
{
|
| 3959 |
+
"epoch": 0.010561423012784881,
|
| 3960 |
+
"grad_norm": 0.12247351557016373,
|
| 3961 |
+
"learning_rate": 8.331369935171716e-05,
|
| 3962 |
+
"loss": 1.3185,
|
| 3963 |
+
"step": 551
|
| 3964 |
+
},
|
| 3965 |
+
{
|
| 3966 |
+
"epoch": 0.010580590749650188,
|
| 3967 |
+
"grad_norm": 0.12315574288368225,
|
| 3968 |
+
"learning_rate": 8.494730129979005e-05,
|
| 3969 |
+
"loss": 1.5978,
|
| 3970 |
+
"step": 552
|
| 3971 |
+
},
|
| 3972 |
+
{
|
| 3973 |
+
"epoch": 0.010599758486515497,
|
| 3974 |
+
"grad_norm": 0.09850146621465683,
|
| 3975 |
+
"learning_rate": 8.658090324786293e-05,
|
| 3976 |
+
"loss": 1.613,
|
| 3977 |
+
"step": 553
|
| 3978 |
+
},
|
| 3979 |
+
{
|
| 3980 |
+
"epoch": 0.010618926223380806,
|
| 3981 |
+
"grad_norm": 0.08749821037054062,
|
| 3982 |
+
"learning_rate": 8.821450519593582e-05,
|
| 3983 |
+
"loss": 1.4552,
|
| 3984 |
+
"step": 554
|
| 3985 |
+
},
|
| 3986 |
+
{
|
| 3987 |
+
"epoch": 0.010638093960246113,
|
| 3988 |
+
"grad_norm": 0.09449183195829391,
|
| 3989 |
+
"learning_rate": 8.98481071440087e-05,
|
| 3990 |
+
"loss": 1.231,
|
| 3991 |
+
"step": 555
|
| 3992 |
+
},
|
| 3993 |
+
{
|
| 3994 |
+
"epoch": 0.010657261697111422,
|
| 3995 |
+
"grad_norm": 0.09725350141525269,
|
| 3996 |
+
"learning_rate": 9.148170909208159e-05,
|
| 3997 |
+
"loss": 1.6669,
|
| 3998 |
+
"step": 556
|
| 3999 |
+
},
|
| 4000 |
+
{
|
| 4001 |
+
"epoch": 0.01067642943397673,
|
| 4002 |
+
"grad_norm": 0.09609991312026978,
|
| 4003 |
+
"learning_rate": 9.311531104015447e-05,
|
| 4004 |
+
"loss": 1.1775,
|
| 4005 |
+
"step": 557
|
| 4006 |
+
},
|
| 4007 |
+
{
|
| 4008 |
+
"epoch": 0.010695597170842038,
|
| 4009 |
+
"grad_norm": 0.13277539610862732,
|
| 4010 |
+
"learning_rate": 9.474891298822736e-05,
|
| 4011 |
+
"loss": 1.2718,
|
| 4012 |
+
"step": 558
|
| 4013 |
+
},
|
| 4014 |
+
{
|
| 4015 |
+
"epoch": 0.010714764907707347,
|
| 4016 |
+
"grad_norm": 0.09620299935340881,
|
| 4017 |
+
"learning_rate": 9.638251493630024e-05,
|
| 4018 |
+
"loss": 1.4025,
|
| 4019 |
+
"step": 559
|
| 4020 |
+
},
|
| 4021 |
+
{
|
| 4022 |
+
"epoch": 0.010733932644572656,
|
| 4023 |
+
"grad_norm": 0.10179946571588516,
|
| 4024 |
+
"learning_rate": 9.801611688437313e-05,
|
| 4025 |
+
"loss": 1.1225,
|
| 4026 |
+
"step": 560
|
| 4027 |
+
},
|
| 4028 |
+
{
|
| 4029 |
+
"epoch": 0.010753100381437964,
|
| 4030 |
+
"grad_norm": 0.12255925685167313,
|
| 4031 |
+
"learning_rate": 9.801529597559957e-05,
|
| 4032 |
+
"loss": 1.2726,
|
| 4033 |
+
"step": 561
|
| 4034 |
+
},
|
| 4035 |
+
{
|
| 4036 |
+
"epoch": 0.010772268118303271,
|
| 4037 |
+
"grad_norm": 0.12259944528341293,
|
| 4038 |
+
"learning_rate": 9.801283327706357e-05,
|
| 4039 |
+
"loss": 1.372,
|
| 4040 |
+
"step": 562
|
| 4041 |
+
},
|
| 4042 |
+
{
|
| 4043 |
+
"epoch": 0.01079143585516858,
|
| 4044 |
+
"grad_norm": 0.10003650188446045,
|
| 4045 |
+
"learning_rate": 9.800872887211833e-05,
|
| 4046 |
+
"loss": 1.6154,
|
| 4047 |
+
"step": 563
|
| 4048 |
+
},
|
| 4049 |
+
{
|
| 4050 |
+
"epoch": 0.01081060359203389,
|
| 4051 |
+
"grad_norm": 0.11308196187019348,
|
| 4052 |
+
"learning_rate": 9.800298289968272e-05,
|
| 4053 |
+
"loss": 1.49,
|
| 4054 |
+
"step": 564
|
| 4055 |
+
},
|
| 4056 |
+
{
|
| 4057 |
+
"epoch": 0.010829771328899196,
|
| 4058 |
+
"grad_norm": 0.11465234309434891,
|
| 4059 |
+
"learning_rate": 9.799559555423655e-05,
|
| 4060 |
+
"loss": 1.076,
|
| 4061 |
+
"step": 565
|
| 4062 |
+
},
|
| 4063 |
+
{
|
| 4064 |
+
"epoch": 0.010848939065764505,
|
| 4065 |
+
"grad_norm": 0.10720562189817429,
|
| 4066 |
+
"learning_rate": 9.798656708581404e-05,
|
| 4067 |
+
"loss": 1.5567,
|
| 4068 |
+
"step": 566
|
| 4069 |
+
},
|
| 4070 |
+
{
|
| 4071 |
+
"epoch": 0.010868106802629814,
|
| 4072 |
+
"grad_norm": 0.14240263402462006,
|
| 4073 |
+
"learning_rate": 9.797589779999525e-05,
|
| 4074 |
+
"loss": 1.2253,
|
| 4075 |
+
"step": 567
|
| 4076 |
+
},
|
| 4077 |
+
{
|
| 4078 |
+
"epoch": 0.010887274539495121,
|
| 4079 |
+
"grad_norm": 0.11476835608482361,
|
| 4080 |
+
"learning_rate": 9.79635880578959e-05,
|
| 4081 |
+
"loss": 1.6748,
|
| 4082 |
+
"step": 568
|
| 4083 |
+
},
|
| 4084 |
+
{
|
| 4085 |
+
"epoch": 0.01090644227636043,
|
| 4086 |
+
"grad_norm": 0.09139933437108994,
|
| 4087 |
+
"learning_rate": 9.7949638276155e-05,
|
| 4088 |
+
"loss": 1.8657,
|
| 4089 |
+
"step": 569
|
| 4090 |
+
},
|
| 4091 |
+
{
|
| 4092 |
+
"epoch": 0.010925610013225739,
|
| 4093 |
+
"grad_norm": 0.16544809937477112,
|
| 4094 |
+
"learning_rate": 9.793404892692088e-05,
|
| 4095 |
+
"loss": 1.7588,
|
| 4096 |
+
"step": 570
|
| 4097 |
+
},
|
| 4098 |
+
{
|
| 4099 |
+
"epoch": 0.010944777750091048,
|
| 4100 |
+
"grad_norm": 0.12964791059494019,
|
| 4101 |
+
"learning_rate": 9.791682053783504e-05,
|
| 4102 |
+
"loss": 1.7762,
|
| 4103 |
+
"step": 571
|
| 4104 |
+
},
|
| 4105 |
+
{
|
| 4106 |
+
"epoch": 0.010963945486956355,
|
| 4107 |
+
"grad_norm": 0.09350966662168503,
|
| 4108 |
+
"learning_rate": 9.789795369201445e-05,
|
| 4109 |
+
"loss": 1.3184,
|
| 4110 |
+
"step": 572
|
| 4111 |
+
},
|
| 4112 |
+
{
|
| 4113 |
+
"epoch": 0.010983113223821664,
|
| 4114 |
+
"grad_norm": 0.12951631844043732,
|
| 4115 |
+
"learning_rate": 9.787744902803176e-05,
|
| 4116 |
+
"loss": 1.0515,
|
| 4117 |
+
"step": 573
|
| 4118 |
+
},
|
| 4119 |
+
{
|
| 4120 |
+
"epoch": 0.011002280960686972,
|
| 4121 |
+
"grad_norm": 0.16301530599594116,
|
| 4122 |
+
"learning_rate": 9.785530723989363e-05,
|
| 4123 |
+
"loss": 1.5014,
|
| 4124 |
+
"step": 574
|
| 4125 |
+
},
|
| 4126 |
+
{
|
| 4127 |
+
"epoch": 0.01102144869755228,
|
| 4128 |
+
"grad_norm": 0.12670457363128662,
|
| 4129 |
+
"learning_rate": 9.783152907701736e-05,
|
| 4130 |
+
"loss": 1.717,
|
| 4131 |
+
"step": 575
|
| 4132 |
+
},
|
| 4133 |
+
{
|
| 4134 |
+
"epoch": 0.011040616434417588,
|
| 4135 |
+
"grad_norm": 0.0955829843878746,
|
| 4136 |
+
"learning_rate": 9.780611534420538e-05,
|
| 4137 |
+
"loss": 1.8241,
|
| 4138 |
+
"step": 576
|
| 4139 |
+
},
|
| 4140 |
+
{
|
| 4141 |
+
"epoch": 0.011059784171282897,
|
| 4142 |
+
"grad_norm": 0.09355156123638153,
|
| 4143 |
+
"learning_rate": 9.777906690161814e-05,
|
| 4144 |
+
"loss": 1.4054,
|
| 4145 |
+
"step": 577
|
| 4146 |
+
},
|
| 4147 |
+
{
|
| 4148 |
+
"epoch": 0.011078951908148204,
|
| 4149 |
+
"grad_norm": 0.11253128945827484,
|
| 4150 |
+
"learning_rate": 9.775038466474491e-05,
|
| 4151 |
+
"loss": 1.8637,
|
| 4152 |
+
"step": 578
|
| 4153 |
+
},
|
| 4154 |
+
{
|
| 4155 |
+
"epoch": 0.011098119645013513,
|
| 4156 |
+
"grad_norm": 0.09193957597017288,
|
| 4157 |
+
"learning_rate": 9.772006960437286e-05,
|
| 4158 |
+
"loss": 1.5736,
|
| 4159 |
+
"step": 579
|
| 4160 |
+
},
|
| 4161 |
+
{
|
| 4162 |
+
"epoch": 0.011117287381878822,
|
| 4163 |
+
"grad_norm": 0.13004140555858612,
|
| 4164 |
+
"learning_rate": 9.768812274655415e-05,
|
| 4165 |
+
"loss": 1.5249,
|
| 4166 |
+
"step": 580
|
| 4167 |
+
},
|
| 4168 |
+
{
|
| 4169 |
+
"epoch": 0.011136455118744129,
|
| 4170 |
+
"grad_norm": 0.1326868236064911,
|
| 4171 |
+
"learning_rate": 9.765454517257119e-05,
|
| 4172 |
+
"loss": 1.4847,
|
| 4173 |
+
"step": 581
|
| 4174 |
+
},
|
| 4175 |
+
{
|
| 4176 |
+
"epoch": 0.011155622855609438,
|
| 4177 |
+
"grad_norm": 0.13382641971111298,
|
| 4178 |
+
"learning_rate": 9.76193380189001e-05,
|
| 4179 |
+
"loss": 1.2997,
|
| 4180 |
+
"step": 582
|
| 4181 |
+
},
|
| 4182 |
+
{
|
| 4183 |
+
"epoch": 0.011174790592474747,
|
| 4184 |
+
"grad_norm": 0.0922945886850357,
|
| 4185 |
+
"learning_rate": 9.758250247717226e-05,
|
| 4186 |
+
"loss": 1.4587,
|
| 4187 |
+
"step": 583
|
| 4188 |
+
},
|
| 4189 |
+
{
|
| 4190 |
+
"epoch": 0.011193958329340056,
|
| 4191 |
+
"grad_norm": 0.11700108647346497,
|
| 4192 |
+
"learning_rate": 9.754403979413388e-05,
|
| 4193 |
+
"loss": 1.5027,
|
| 4194 |
+
"step": 584
|
| 4195 |
+
},
|
| 4196 |
+
{
|
| 4197 |
+
"epoch": 0.011213126066205363,
|
| 4198 |
+
"grad_norm": 0.10824666172266006,
|
| 4199 |
+
"learning_rate": 9.750395127160385e-05,
|
| 4200 |
+
"loss": 1.4262,
|
| 4201 |
+
"step": 585
|
| 4202 |
+
},
|
| 4203 |
+
{
|
| 4204 |
+
"epoch": 0.011232293803070672,
|
| 4205 |
+
"grad_norm": 0.11390513181686401,
|
| 4206 |
+
"learning_rate": 9.746223826642974e-05,
|
| 4207 |
+
"loss": 1.6256,
|
| 4208 |
+
"step": 586
|
| 4209 |
+
},
|
| 4210 |
+
{
|
| 4211 |
+
"epoch": 0.01125146153993598,
|
| 4212 |
+
"grad_norm": 0.15232987701892853,
|
| 4213 |
+
"learning_rate": 9.741890219044177e-05,
|
| 4214 |
+
"loss": 1.5036,
|
| 4215 |
+
"step": 587
|
| 4216 |
+
},
|
| 4217 |
+
{
|
| 4218 |
+
"epoch": 0.011270629276801288,
|
| 4219 |
+
"grad_norm": 0.11328087747097015,
|
| 4220 |
+
"learning_rate": 9.737394451040511e-05,
|
| 4221 |
+
"loss": 1.3001,
|
| 4222 |
+
"step": 588
|
| 4223 |
+
},
|
| 4224 |
+
{
|
| 4225 |
+
"epoch": 0.011289797013666596,
|
| 4226 |
+
"grad_norm": 0.10069140046834946,
|
| 4227 |
+
"learning_rate": 9.732736674797018e-05,
|
| 4228 |
+
"loss": 1.5349,
|
| 4229 |
+
"step": 589
|
| 4230 |
+
},
|
| 4231 |
+
{
|
| 4232 |
+
"epoch": 0.011308964750531905,
|
| 4233 |
+
"grad_norm": 0.08682256937026978,
|
| 4234 |
+
"learning_rate": 9.727917047962118e-05,
|
| 4235 |
+
"loss": 1.2649,
|
| 4236 |
+
"step": 590
|
| 4237 |
+
},
|
| 4238 |
+
{
|
| 4239 |
+
"epoch": 0.011328132487397212,
|
| 4240 |
+
"grad_norm": 0.10409464687108994,
|
| 4241 |
+
"learning_rate": 9.722935733662273e-05,
|
| 4242 |
+
"loss": 1.3541,
|
| 4243 |
+
"step": 591
|
| 4244 |
+
},
|
| 4245 |
+
{
|
| 4246 |
+
"epoch": 0.011347300224262521,
|
| 4247 |
+
"grad_norm": 0.10544151812791824,
|
| 4248 |
+
"learning_rate": 9.717792900496464e-05,
|
| 4249 |
+
"loss": 1.1391,
|
| 4250 |
+
"step": 592
|
| 4251 |
+
},
|
| 4252 |
+
{
|
| 4253 |
+
"epoch": 0.01136646796112783,
|
| 4254 |
+
"grad_norm": 0.12220872938632965,
|
| 4255 |
+
"learning_rate": 9.712488722530483e-05,
|
| 4256 |
+
"loss": 1.3058,
|
| 4257 |
+
"step": 593
|
| 4258 |
+
},
|
| 4259 |
+
{
|
| 4260 |
+
"epoch": 0.011385635697993137,
|
| 4261 |
+
"grad_norm": 0.12459158897399902,
|
| 4262 |
+
"learning_rate": 9.70702337929105e-05,
|
| 4263 |
+
"loss": 1.7773,
|
| 4264 |
+
"step": 594
|
| 4265 |
+
},
|
| 4266 |
+
{
|
| 4267 |
+
"epoch": 0.011404803434858446,
|
| 4268 |
+
"grad_norm": 0.1203218549489975,
|
| 4269 |
+
"learning_rate": 9.701397055759726e-05,
|
| 4270 |
+
"loss": 1.3244,
|
| 4271 |
+
"step": 595
|
| 4272 |
+
},
|
| 4273 |
+
{
|
| 4274 |
+
"epoch": 0.011423971171723755,
|
| 4275 |
+
"grad_norm": 0.08355801552534103,
|
| 4276 |
+
"learning_rate": 9.695609942366658e-05,
|
| 4277 |
+
"loss": 1.2935,
|
| 4278 |
+
"step": 596
|
| 4279 |
+
},
|
| 4280 |
+
{
|
| 4281 |
+
"epoch": 0.011443138908589064,
|
| 4282 |
+
"grad_norm": 0.11499644815921783,
|
| 4283 |
+
"learning_rate": 9.689662234984131e-05,
|
| 4284 |
+
"loss": 1.3095,
|
| 4285 |
+
"step": 597
|
| 4286 |
+
},
|
| 4287 |
+
{
|
| 4288 |
+
"epoch": 0.01146230664545437,
|
| 4289 |
+
"grad_norm": 0.09302666038274765,
|
| 4290 |
+
"learning_rate": 9.683554134919944e-05,
|
| 4291 |
+
"loss": 1.2796,
|
| 4292 |
+
"step": 598
|
| 4293 |
+
},
|
| 4294 |
+
{
|
| 4295 |
+
"epoch": 0.01148147438231968,
|
| 4296 |
+
"grad_norm": 0.11288128793239594,
|
| 4297 |
+
"learning_rate": 9.677285848910587e-05,
|
| 4298 |
+
"loss": 1.4899,
|
| 4299 |
+
"step": 599
|
| 4300 |
+
},
|
| 4301 |
+
{
|
| 4302 |
+
"epoch": 0.011500642119184988,
|
| 4303 |
+
"grad_norm": 0.10734209418296814,
|
| 4304 |
+
"learning_rate": 9.670857589114252e-05,
|
| 4305 |
+
"loss": 1.6273,
|
| 4306 |
+
"step": 600
|
| 4307 |
+
},
|
| 4308 |
+
{
|
| 4309 |
+
"epoch": 0.011500642119184988,
|
| 4310 |
+
"eval_loss": 1.491734266281128,
|
| 4311 |
+
"eval_runtime": 2535.026,
|
| 4312 |
+
"eval_samples_per_second": 36.587,
|
| 4313 |
+
"eval_steps_per_second": 18.293,
|
| 4314 |
+
"step": 600
|
| 4315 |
+
},
|
| 4316 |
+
{
|
| 4317 |
+
"epoch": 0.011500642119184988,
|
| 4318 |
+
"eval_loss": 1.5211025476455688,
|
| 4319 |
+
"eval_runtime": 2506.1864,
|
| 4320 |
+
"eval_samples_per_second": 37.008,
|
| 4321 |
+
"eval_steps_per_second": 18.504,
|
| 4322 |
+
"step": 600
|
| 4323 |
+
},
|
| 4324 |
+
{
|
| 4325 |
+
"epoch": 0.011519809856050296,
|
| 4326 |
+
"grad_norm": 0.10411401838064194,
|
| 4327 |
+
"learning_rate": 9.664269573103652e-05,
|
| 4328 |
+
"loss": 1.6011,
|
| 4329 |
+
"step": 601
|
| 4330 |
+
},
|
| 4331 |
+
{
|
| 4332 |
+
"epoch": 0.011538977592915604,
|
| 4333 |
+
"grad_norm": 0.10562720894813538,
|
| 4334 |
+
"learning_rate": 9.657522023858646e-05,
|
| 4335 |
+
"loss": 1.6277,
|
| 4336 |
+
"step": 602
|
| 4337 |
+
},
|
| 4338 |
+
{
|
| 4339 |
+
"epoch": 0.011558145329780913,
|
| 4340 |
+
"grad_norm": 0.09079369902610779,
|
| 4341 |
+
"learning_rate": 9.65061516975871e-05,
|
| 4342 |
+
"loss": 1.3665,
|
| 4343 |
+
"step": 603
|
| 4344 |
+
},
|
| 4345 |
+
{
|
| 4346 |
+
"epoch": 0.01157731306664622,
|
| 4347 |
+
"grad_norm": 0.11086009442806244,
|
| 4348 |
+
"learning_rate": 9.643549244575191e-05,
|
| 4349 |
+
"loss": 1.7168,
|
| 4350 |
+
"step": 604
|
| 4351 |
+
},
|
| 4352 |
+
{
|
| 4353 |
+
"epoch": 0.01159648080351153,
|
| 4354 |
+
"grad_norm": 0.127419114112854,
|
| 4355 |
+
"learning_rate": 9.636324487463404e-05,
|
| 4356 |
+
"loss": 1.4687,
|
| 4357 |
+
"step": 605
|
| 4358 |
+
},
|
| 4359 |
+
{
|
| 4360 |
+
"epoch": 0.011615648540376838,
|
| 4361 |
+
"grad_norm": 0.09277164936065674,
|
| 4362 |
+
"learning_rate": 9.628941142954533e-05,
|
| 4363 |
+
"loss": 1.27,
|
| 4364 |
+
"step": 606
|
| 4365 |
+
},
|
| 4366 |
+
{
|
| 4367 |
+
"epoch": 0.011634816277242147,
|
| 4368 |
+
"grad_norm": 0.12536202371120453,
|
| 4369 |
+
"learning_rate": 9.621399460947357e-05,
|
| 4370 |
+
"loss": 1.4982,
|
| 4371 |
+
"step": 607
|
| 4372 |
+
},
|
| 4373 |
+
{
|
| 4374 |
+
"epoch": 0.011653984014107454,
|
| 4375 |
+
"grad_norm": 0.10327380150556564,
|
| 4376 |
+
"learning_rate": 9.61369969669979e-05,
|
| 4377 |
+
"loss": 1.1826,
|
| 4378 |
+
"step": 608
|
| 4379 |
+
},
|
| 4380 |
+
{
|
| 4381 |
+
"epoch": 0.011673151750972763,
|
| 4382 |
+
"grad_norm": 0.1040838211774826,
|
| 4383 |
+
"learning_rate": 9.605842110820246e-05,
|
| 4384 |
+
"loss": 1.4192,
|
| 4385 |
+
"step": 609
|
| 4386 |
+
},
|
| 4387 |
+
{
|
| 4388 |
+
"epoch": 0.011692319487838072,
|
| 4389 |
+
"grad_norm": 0.129541277885437,
|
| 4390 |
+
"learning_rate": 9.597826969258811e-05,
|
| 4391 |
+
"loss": 1.6659,
|
| 4392 |
+
"step": 610
|
| 4393 |
+
},
|
| 4394 |
+
{
|
| 4395 |
+
"epoch": 0.011711487224703379,
|
| 4396 |
+
"grad_norm": 0.10935177654027939,
|
| 4397 |
+
"learning_rate": 9.589654543298245e-05,
|
| 4398 |
+
"loss": 1.5881,
|
| 4399 |
+
"step": 611
|
| 4400 |
+
},
|
| 4401 |
+
{
|
| 4402 |
+
"epoch": 0.011730654961568688,
|
| 4403 |
+
"grad_norm": 0.09962373971939087,
|
| 4404 |
+
"learning_rate": 9.581325109544805e-05,
|
| 4405 |
+
"loss": 1.3581,
|
| 4406 |
+
"step": 612
|
| 4407 |
+
},
|
| 4408 |
+
{
|
| 4409 |
+
"epoch": 0.011749822698433996,
|
| 4410 |
+
"grad_norm": 0.13899528980255127,
|
| 4411 |
+
"learning_rate": 9.572838949918872e-05,
|
| 4412 |
+
"loss": 1.5192,
|
| 4413 |
+
"step": 613
|
| 4414 |
+
},
|
| 4415 |
+
{
|
| 4416 |
+
"epoch": 0.011768990435299304,
|
| 4417 |
+
"grad_norm": 0.11136510223150253,
|
| 4418 |
+
"learning_rate": 9.564196351645423e-05,
|
| 4419 |
+
"loss": 1.3238,
|
| 4420 |
+
"step": 614
|
| 4421 |
+
},
|
| 4422 |
+
{
|
| 4423 |
+
"epoch": 0.011788158172164612,
|
| 4424 |
+
"grad_norm": 0.11565709859132767,
|
| 4425 |
+
"learning_rate": 9.555397607244296e-05,
|
| 4426 |
+
"loss": 1.6724,
|
| 4427 |
+
"step": 615
|
| 4428 |
+
},
|
| 4429 |
+
{
|
| 4430 |
+
"epoch": 0.011807325909029921,
|
| 4431 |
+
"grad_norm": 0.09210363775491714,
|
| 4432 |
+
"learning_rate": 9.546443014520298e-05,
|
| 4433 |
+
"loss": 1.2795,
|
| 4434 |
+
"step": 616
|
| 4435 |
+
},
|
| 4436 |
+
{
|
| 4437 |
+
"epoch": 0.011826493645895228,
|
| 4438 |
+
"grad_norm": 0.10333288460969925,
|
| 4439 |
+
"learning_rate": 9.537332876553123e-05,
|
| 4440 |
+
"loss": 1.4106,
|
| 4441 |
+
"step": 617
|
| 4442 |
+
},
|
| 4443 |
+
{
|
| 4444 |
+
"epoch": 0.011845661382760537,
|
| 4445 |
+
"grad_norm": 0.13616147637367249,
|
| 4446 |
+
"learning_rate": 9.528067501687089e-05,
|
| 4447 |
+
"loss": 1.3582,
|
| 4448 |
+
"step": 618
|
| 4449 |
+
},
|
| 4450 |
+
{
|
| 4451 |
+
"epoch": 0.011864829119625846,
|
| 4452 |
+
"grad_norm": 0.10424721240997314,
|
| 4453 |
+
"learning_rate": 9.518647203520714e-05,
|
| 4454 |
+
"loss": 1.4855,
|
| 4455 |
+
"step": 619
|
| 4456 |
+
},
|
| 4457 |
+
{
|
| 4458 |
+
"epoch": 0.011883996856491155,
|
| 4459 |
+
"grad_norm": 0.11424586921930313,
|
| 4460 |
+
"learning_rate": 9.509072300896087e-05,
|
| 4461 |
+
"loss": 1.1964,
|
| 4462 |
+
"step": 620
|
| 4463 |
+
},
|
| 4464 |
+
{
|
| 4465 |
+
"epoch": 0.011903164593356462,
|
| 4466 |
+
"grad_norm": 0.16829033195972443,
|
| 4467 |
+
"learning_rate": 9.499343117888088e-05,
|
| 4468 |
+
"loss": 1.3607,
|
| 4469 |
+
"step": 621
|
| 4470 |
+
},
|
| 4471 |
+
{
|
| 4472 |
+
"epoch": 0.01192233233022177,
|
| 4473 |
+
"grad_norm": 0.11840435117483139,
|
| 4474 |
+
"learning_rate": 9.489459983793413e-05,
|
| 4475 |
+
"loss": 1.5524,
|
| 4476 |
+
"step": 622
|
| 4477 |
+
},
|
| 4478 |
+
{
|
| 4479 |
+
"epoch": 0.01194150006708708,
|
| 4480 |
+
"grad_norm": 0.13831564784049988,
|
| 4481 |
+
"learning_rate": 9.479423233119428e-05,
|
| 4482 |
+
"loss": 1.5572,
|
| 4483 |
+
"step": 623
|
| 4484 |
+
},
|
| 4485 |
+
{
|
| 4486 |
+
"epoch": 0.011960667803952387,
|
| 4487 |
+
"grad_norm": 0.12457295507192612,
|
| 4488 |
+
"learning_rate": 9.469233205572857e-05,
|
| 4489 |
+
"loss": 1.5482,
|
| 4490 |
+
"step": 624
|
| 4491 |
+
},
|
| 4492 |
+
{
|
| 4493 |
+
"epoch": 0.011979835540817696,
|
| 4494 |
+
"grad_norm": 0.14689622819423676,
|
| 4495 |
+
"learning_rate": 9.458890246048266e-05,
|
| 4496 |
+
"loss": 1.2769,
|
| 4497 |
+
"step": 625
|
| 4498 |
+
},
|
| 4499 |
+
{
|
| 4500 |
+
"epoch": 0.011999003277683004,
|
| 4501 |
+
"grad_norm": 0.08986423164606094,
|
| 4502 |
+
"learning_rate": 9.44839470461641e-05,
|
| 4503 |
+
"loss": 1.0757,
|
| 4504 |
+
"step": 626
|
| 4505 |
+
},
|
| 4506 |
+
{
|
| 4507 |
+
"epoch": 0.012018171014548312,
|
| 4508 |
+
"grad_norm": 0.1021321564912796,
|
| 4509 |
+
"learning_rate": 9.437746936512368e-05,
|
| 4510 |
+
"loss": 1.5207,
|
| 4511 |
+
"step": 627
|
| 4512 |
+
},
|
| 4513 |
+
{
|
| 4514 |
+
"epoch": 0.01203733875141362,
|
| 4515 |
+
"grad_norm": 0.14323778450489044,
|
| 4516 |
+
"learning_rate": 9.426947302123529e-05,
|
| 4517 |
+
"loss": 1.5927,
|
| 4518 |
+
"step": 628
|
| 4519 |
+
},
|
| 4520 |
+
{
|
| 4521 |
+
"epoch": 0.01205650648827893,
|
| 4522 |
+
"grad_norm": 0.11163625121116638,
|
| 4523 |
+
"learning_rate": 9.415996166977393e-05,
|
| 4524 |
+
"loss": 1.2186,
|
| 4525 |
+
"step": 629
|
| 4526 |
+
},
|
| 4527 |
+
{
|
| 4528 |
+
"epoch": 0.012075674225144236,
|
| 4529 |
+
"grad_norm": 0.11363939940929413,
|
| 4530 |
+
"learning_rate": 9.404893901729192e-05,
|
| 4531 |
+
"loss": 1.2431,
|
| 4532 |
+
"step": 630
|
| 4533 |
+
},
|
| 4534 |
+
{
|
| 4535 |
+
"epoch": 0.012094841962009545,
|
| 4536 |
+
"grad_norm": 0.124871626496315,
|
| 4537 |
+
"learning_rate": 9.393640882149359e-05,
|
| 4538 |
+
"loss": 1.3805,
|
| 4539 |
+
"step": 631
|
| 4540 |
+
},
|
| 4541 |
+
{
|
| 4542 |
+
"epoch": 0.012114009698874854,
|
| 4543 |
+
"grad_norm": 0.12126212567090988,
|
| 4544 |
+
"learning_rate": 9.382237489110791e-05,
|
| 4545 |
+
"loss": 1.253,
|
| 4546 |
+
"step": 632
|
| 4547 |
+
},
|
| 4548 |
+
{
|
| 4549 |
+
"epoch": 0.012133177435740163,
|
| 4550 |
+
"grad_norm": 0.1411461979150772,
|
| 4551 |
+
"learning_rate": 9.370684108575978e-05,
|
| 4552 |
+
"loss": 1.5281,
|
| 4553 |
+
"step": 633
|
| 4554 |
+
},
|
| 4555 |
+
{
|
| 4556 |
+
"epoch": 0.01215234517260547,
|
| 4557 |
+
"grad_norm": 0.09744168072938919,
|
| 4558 |
+
"learning_rate": 9.358981131583925e-05,
|
| 4559 |
+
"loss": 1.3321,
|
| 4560 |
+
"step": 634
|
| 4561 |
+
},
|
| 4562 |
+
{
|
| 4563 |
+
"epoch": 0.012171512909470779,
|
| 4564 |
+
"grad_norm": 0.13379117846488953,
|
| 4565 |
+
"learning_rate": 9.347128954236917e-05,
|
| 4566 |
+
"loss": 1.7078,
|
| 4567 |
+
"step": 635
|
| 4568 |
+
},
|
| 4569 |
+
{
|
| 4570 |
+
"epoch": 0.012190680646336088,
|
| 4571 |
+
"grad_norm": 0.18476705253124237,
|
| 4572 |
+
"learning_rate": 9.335127977687121e-05,
|
| 4573 |
+
"loss": 1.3149,
|
| 4574 |
+
"step": 636
|
| 4575 |
+
},
|
| 4576 |
+
{
|
| 4577 |
+
"epoch": 0.012209848383201395,
|
| 4578 |
+
"grad_norm": 0.13794240355491638,
|
| 4579 |
+
"learning_rate": 9.322978608123005e-05,
|
| 4580 |
+
"loss": 1.2862,
|
| 4581 |
+
"step": 637
|
| 4582 |
+
},
|
| 4583 |
+
{
|
| 4584 |
+
"epoch": 0.012229016120066704,
|
| 4585 |
+
"grad_norm": 0.1311839371919632,
|
| 4586 |
+
"learning_rate": 9.31068125675559e-05,
|
| 4587 |
+
"loss": 1.7977,
|
| 4588 |
+
"step": 638
|
| 4589 |
+
},
|
| 4590 |
+
{
|
| 4591 |
+
"epoch": 0.012248183856932012,
|
| 4592 |
+
"grad_norm": 0.10724909603595734,
|
| 4593 |
+
"learning_rate": 9.298236339804526e-05,
|
| 4594 |
+
"loss": 1.1626,
|
| 4595 |
+
"step": 639
|
| 4596 |
+
},
|
| 4597 |
+
{
|
| 4598 |
+
"epoch": 0.01226735159379732,
|
| 4599 |
+
"grad_norm": 0.09627658873796463,
|
| 4600 |
+
"learning_rate": 9.285644278484013e-05,
|
| 4601 |
+
"loss": 1.5093,
|
| 4602 |
+
"step": 640
|
| 4603 |
+
},
|
| 4604 |
+
{
|
| 4605 |
+
"epoch": 0.012286519330662628,
|
| 4606 |
+
"grad_norm": 0.13738052546977997,
|
| 4607 |
+
"learning_rate": 9.272905498988542e-05,
|
| 4608 |
+
"loss": 1.1338,
|
| 4609 |
+
"step": 641
|
| 4610 |
+
},
|
| 4611 |
+
{
|
| 4612 |
+
"epoch": 0.012305687067527937,
|
| 4613 |
+
"grad_norm": 0.11568105965852737,
|
| 4614 |
+
"learning_rate": 9.26002043247847e-05,
|
| 4615 |
+
"loss": 1.2151,
|
| 4616 |
+
"step": 642
|
| 4617 |
+
},
|
| 4618 |
+
{
|
| 4619 |
+
"epoch": 0.012324854804393246,
|
| 4620 |
+
"grad_norm": 0.11026597023010254,
|
| 4621 |
+
"learning_rate": 9.246989515065422e-05,
|
| 4622 |
+
"loss": 1.5823,
|
| 4623 |
+
"step": 643
|
| 4624 |
+
},
|
| 4625 |
+
{
|
| 4626 |
+
"epoch": 0.012344022541258553,
|
| 4627 |
+
"grad_norm": 0.10940872132778168,
|
| 4628 |
+
"learning_rate": 9.233813187797535e-05,
|
| 4629 |
+
"loss": 1.3081,
|
| 4630 |
+
"step": 644
|
| 4631 |
+
},
|
| 4632 |
+
{
|
| 4633 |
+
"epoch": 0.012363190278123862,
|
| 4634 |
+
"grad_norm": 0.12110668420791626,
|
| 4635 |
+
"learning_rate": 9.220491896644531e-05,
|
| 4636 |
+
"loss": 1.1608,
|
| 4637 |
+
"step": 645
|
| 4638 |
+
},
|
| 4639 |
+
{
|
| 4640 |
+
"epoch": 0.012382358014989171,
|
| 4641 |
+
"grad_norm": 0.1306304782629013,
|
| 4642 |
+
"learning_rate": 9.207026092482622e-05,
|
| 4643 |
+
"loss": 1.5766,
|
| 4644 |
+
"step": 646
|
| 4645 |
+
},
|
| 4646 |
+
{
|
| 4647 |
+
"epoch": 0.012401525751854478,
|
| 4648 |
+
"grad_norm": 0.11147546768188477,
|
| 4649 |
+
"learning_rate": 9.193416231079247e-05,
|
| 4650 |
+
"loss": 1.462,
|
| 4651 |
+
"step": 647
|
| 4652 |
+
},
|
| 4653 |
+
{
|
| 4654 |
+
"epoch": 0.012420693488719787,
|
| 4655 |
+
"grad_norm": 0.1513271927833557,
|
| 4656 |
+
"learning_rate": 9.179662773077646e-05,
|
| 4657 |
+
"loss": 1.2902,
|
| 4658 |
+
"step": 648
|
| 4659 |
+
},
|
| 4660 |
+
{
|
| 4661 |
+
"epoch": 0.012439861225585096,
|
| 4662 |
+
"grad_norm": 0.12458869814872742,
|
| 4663 |
+
"learning_rate": 9.165766183981272e-05,
|
| 4664 |
+
"loss": 1.6558,
|
| 4665 |
+
"step": 649
|
| 4666 |
+
},
|
| 4667 |
+
{
|
| 4668 |
+
"epoch": 0.012459028962450403,
|
| 4669 |
+
"grad_norm": 0.12529686093330383,
|
| 4670 |
+
"learning_rate": 9.151726934138035e-05,
|
| 4671 |
+
"loss": 1.519,
|
| 4672 |
+
"step": 650
|
| 4673 |
+
},
|
| 4674 |
+
{
|
| 4675 |
+
"epoch": 0.012478196699315712,
|
| 4676 |
+
"grad_norm": 0.1223490983247757,
|
| 4677 |
+
"learning_rate": 9.137545498724373e-05,
|
| 4678 |
+
"loss": 1.4077,
|
| 4679 |
+
"step": 651
|
| 4680 |
+
},
|
| 4681 |
+
{
|
| 4682 |
+
"epoch": 0.01249736443618102,
|
| 4683 |
+
"grad_norm": 0.15136773884296417,
|
| 4684 |
+
"learning_rate": 9.123222357729193e-05,
|
| 4685 |
+
"loss": 1.2649,
|
| 4686 |
+
"step": 652
|
| 4687 |
+
},
|
| 4688 |
+
{
|
| 4689 |
+
"epoch": 0.012516532173046328,
|
| 4690 |
+
"grad_norm": 0.08339743316173553,
|
| 4691 |
+
"learning_rate": 9.108757995937594e-05,
|
| 4692 |
+
"loss": 1.0947,
|
| 4693 |
+
"step": 653
|
| 4694 |
+
},
|
| 4695 |
+
{
|
| 4696 |
+
"epoch": 0.012535699909911636,
|
| 4697 |
+
"grad_norm": 0.12441645562648773,
|
| 4698 |
+
"learning_rate": 9.094152902914481e-05,
|
| 4699 |
+
"loss": 1.2436,
|
| 4700 |
+
"step": 654
|
| 4701 |
+
},
|
| 4702 |
+
{
|
| 4703 |
+
"epoch": 0.012554867646776945,
|
| 4704 |
+
"grad_norm": 0.12047876417636871,
|
| 4705 |
+
"learning_rate": 9.07940757298799e-05,
|
| 4706 |
+
"loss": 1.3707,
|
| 4707 |
+
"step": 655
|
| 4708 |
+
},
|
| 4709 |
+
{
|
| 4710 |
+
"epoch": 0.012574035383642254,
|
| 4711 |
+
"grad_norm": 0.11989863961935043,
|
| 4712 |
+
"learning_rate": 9.064522505232752e-05,
|
| 4713 |
+
"loss": 1.4951,
|
| 4714 |
+
"step": 656
|
| 4715 |
+
},
|
| 4716 |
+
{
|
| 4717 |
+
"epoch": 0.012593203120507561,
|
| 4718 |
+
"grad_norm": 0.09021518379449844,
|
| 4719 |
+
"learning_rate": 9.049498203453006e-05,
|
| 4720 |
+
"loss": 1.6233,
|
| 4721 |
+
"step": 657
|
| 4722 |
+
},
|
| 4723 |
+
{
|
| 4724 |
+
"epoch": 0.01261237085737287,
|
| 4725 |
+
"grad_norm": 0.17760628461837769,
|
| 4726 |
+
"learning_rate": 9.034335176165543e-05,
|
| 4727 |
+
"loss": 1.4375,
|
| 4728 |
+
"step": 658
|
| 4729 |
+
},
|
| 4730 |
+
{
|
| 4731 |
+
"epoch": 0.012631538594238179,
|
| 4732 |
+
"grad_norm": 0.11390668898820877,
|
| 4733 |
+
"learning_rate": 9.0190339365825e-05,
|
| 4734 |
+
"loss": 1.7167,
|
| 4735 |
+
"step": 659
|
| 4736 |
+
},
|
| 4737 |
+
{
|
| 4738 |
+
"epoch": 0.012650706331103486,
|
| 4739 |
+
"grad_norm": 0.13511599600315094,
|
| 4740 |
+
"learning_rate": 9.003595002593983e-05,
|
| 4741 |
+
"loss": 1.5552,
|
| 4742 |
+
"step": 660
|
| 4743 |
+
},
|
| 4744 |
+
{
|
| 4745 |
+
"epoch": 0.012669874067968795,
|
| 4746 |
+
"grad_norm": 0.12574054300785065,
|
| 4747 |
+
"learning_rate": 8.988018896750548e-05,
|
| 4748 |
+
"loss": 1.6231,
|
| 4749 |
+
"step": 661
|
| 4750 |
+
},
|
| 4751 |
+
{
|
| 4752 |
+
"epoch": 0.012689041804834104,
|
| 4753 |
+
"grad_norm": 0.10890061408281326,
|
| 4754 |
+
"learning_rate": 8.9723061462455e-05,
|
| 4755 |
+
"loss": 1.06,
|
| 4756 |
+
"step": 662
|
| 4757 |
+
},
|
| 4758 |
+
{
|
| 4759 |
+
"epoch": 0.01270820954169941,
|
| 4760 |
+
"grad_norm": 0.10770639032125473,
|
| 4761 |
+
"learning_rate": 8.95645728289707e-05,
|
| 4762 |
+
"loss": 1.5683,
|
| 4763 |
+
"step": 663
|
| 4764 |
+
},
|
| 4765 |
+
{
|
| 4766 |
+
"epoch": 0.01272737727856472,
|
| 4767 |
+
"grad_norm": 0.11729026585817337,
|
| 4768 |
+
"learning_rate": 8.94047284313039e-05,
|
| 4769 |
+
"loss": 1.3508,
|
| 4770 |
+
"step": 664
|
| 4771 |
+
},
|
| 4772 |
+
{
|
| 4773 |
+
"epoch": 0.012746545015430029,
|
| 4774 |
+
"grad_norm": 0.1106027290225029,
|
| 4775 |
+
"learning_rate": 8.924353367959359e-05,
|
| 4776 |
+
"loss": 1.3145,
|
| 4777 |
+
"step": 665
|
| 4778 |
+
},
|
| 4779 |
+
{
|
| 4780 |
+
"epoch": 0.012765712752295336,
|
| 4781 |
+
"grad_norm": 0.1163991391658783,
|
| 4782 |
+
"learning_rate": 8.908099402968317e-05,
|
| 4783 |
+
"loss": 1.2162,
|
| 4784 |
+
"step": 666
|
| 4785 |
+
},
|
| 4786 |
+
{
|
| 4787 |
+
"epoch": 0.012784880489160644,
|
| 4788 |
+
"grad_norm": 0.11139512062072754,
|
| 4789 |
+
"learning_rate": 8.891711498293591e-05,
|
| 4790 |
+
"loss": 1.6249,
|
| 4791 |
+
"step": 667
|
| 4792 |
+
},
|
| 4793 |
+
{
|
| 4794 |
+
"epoch": 0.012804048226025953,
|
| 4795 |
+
"grad_norm": 0.1063053160905838,
|
| 4796 |
+
"learning_rate": 8.875190208604863e-05,
|
| 4797 |
+
"loss": 1.7123,
|
| 4798 |
+
"step": 668
|
| 4799 |
+
},
|
| 4800 |
+
{
|
| 4801 |
+
"epoch": 0.012823215962891262,
|
| 4802 |
+
"grad_norm": 0.0996391698718071,
|
| 4803 |
+
"learning_rate": 8.858536093086406e-05,
|
| 4804 |
+
"loss": 1.4169,
|
| 4805 |
+
"step": 669
|
| 4806 |
+
},
|
| 4807 |
+
{
|
| 4808 |
+
"epoch": 0.01284238369975657,
|
| 4809 |
+
"grad_norm": 0.15188683569431305,
|
| 4810 |
+
"learning_rate": 8.841749715418151e-05,
|
| 4811 |
+
"loss": 1.3692,
|
| 4812 |
+
"step": 670
|
| 4813 |
+
},
|
| 4814 |
+
{
|
| 4815 |
+
"epoch": 0.012861551436621878,
|
| 4816 |
+
"grad_norm": 0.11893344670534134,
|
| 4817 |
+
"learning_rate": 8.824831643756612e-05,
|
| 4818 |
+
"loss": 1.263,
|
| 4819 |
+
"step": 671
|
| 4820 |
+
},
|
| 4821 |
+
{
|
| 4822 |
+
"epoch": 0.012880719173487187,
|
| 4823 |
+
"grad_norm": 0.11241691559553146,
|
| 4824 |
+
"learning_rate": 8.807782450715656e-05,
|
| 4825 |
+
"loss": 1.3337,
|
| 4826 |
+
"step": 672
|
| 4827 |
+
},
|
| 4828 |
+
{
|
| 4829 |
+
"epoch": 0.012899886910352494,
|
| 4830 |
+
"grad_norm": 0.12966130673885345,
|
| 4831 |
+
"learning_rate": 8.790602713347123e-05,
|
| 4832 |
+
"loss": 1.2688,
|
| 4833 |
+
"step": 673
|
| 4834 |
+
},
|
| 4835 |
+
{
|
| 4836 |
+
"epoch": 0.012919054647217803,
|
| 4837 |
+
"grad_norm": 0.12770654261112213,
|
| 4838 |
+
"learning_rate": 8.77329301312129e-05,
|
| 4839 |
+
"loss": 1.1834,
|
| 4840 |
+
"step": 674
|
| 4841 |
+
},
|
| 4842 |
+
{
|
| 4843 |
+
"epoch": 0.012938222384083112,
|
| 4844 |
+
"grad_norm": 0.14243365824222565,
|
| 4845 |
+
"learning_rate": 8.755853935907198e-05,
|
| 4846 |
+
"loss": 1.4371,
|
| 4847 |
+
"step": 675
|
| 4848 |
+
},
|
| 4849 |
+
{
|
| 4850 |
+
"epoch": 0.012938222384083112,
|
| 4851 |
+
"eval_loss": 1.4821537733078003,
|
| 4852 |
+
"eval_runtime": 2514.7611,
|
| 4853 |
+
"eval_samples_per_second": 36.881,
|
| 4854 |
+
"eval_steps_per_second": 18.441,
|
| 4855 |
+
"step": 675
|
| 4856 |
+
},
|
| 4857 |
+
{
|
| 4858 |
+
"epoch": 0.012938222384083112,
|
| 4859 |
+
"eval_loss": 1.5171287059783936,
|
| 4860 |
+
"eval_runtime": 2514.0016,
|
| 4861 |
+
"eval_samples_per_second": 36.893,
|
| 4862 |
+
"eval_steps_per_second": 18.446,
|
| 4863 |
+
"step": 675
|
| 4864 |
+
},
|
| 4865 |
+
{
|
| 4866 |
+
"epoch": 0.012957390120948419,
|
| 4867 |
+
"grad_norm": 0.09096322953701019,
|
| 4868 |
+
"learning_rate": 8.738286071952813e-05,
|
| 4869 |
+
"loss": 1.412,
|
| 4870 |
+
"step": 676
|
| 4871 |
+
},
|
| 4872 |
+
{
|
| 4873 |
+
"epoch": 0.012976557857813728,
|
| 4874 |
+
"grad_norm": 0.14551272988319397,
|
| 4875 |
+
"learning_rate": 8.720590015865057e-05,
|
| 4876 |
+
"loss": 1.2728,
|
| 4877 |
+
"step": 677
|
| 4878 |
+
},
|
| 4879 |
+
{
|
| 4880 |
+
"epoch": 0.012995725594679037,
|
| 4881 |
+
"grad_norm": 0.11868377774953842,
|
| 4882 |
+
"learning_rate": 8.702766366589681e-05,
|
| 4883 |
+
"loss": 1.6341,
|
| 4884 |
+
"step": 678
|
| 4885 |
+
},
|
| 4886 |
+
{
|
| 4887 |
+
"epoch": 0.013014893331544345,
|
| 4888 |
+
"grad_norm": 0.1134573444724083,
|
| 4889 |
+
"learning_rate": 8.684815727390988e-05,
|
| 4890 |
+
"loss": 1.6379,
|
| 4891 |
+
"step": 679
|
| 4892 |
+
},
|
| 4893 |
+
{
|
| 4894 |
+
"epoch": 0.013034061068409653,
|
| 4895 |
+
"grad_norm": 0.12391621619462967,
|
| 4896 |
+
"learning_rate": 8.666738705831421e-05,
|
| 4897 |
+
"loss": 1.3885,
|
| 4898 |
+
"step": 680
|
| 4899 |
+
},
|
| 4900 |
+
{
|
| 4901 |
+
"epoch": 0.013053228805274961,
|
| 4902 |
+
"grad_norm": 0.1294897496700287,
|
| 4903 |
+
"learning_rate": 8.648535913750994e-05,
|
| 4904 |
+
"loss": 1.7702,
|
| 4905 |
+
"step": 681
|
| 4906 |
+
},
|
| 4907 |
+
{
|
| 4908 |
+
"epoch": 0.01307239654214027,
|
| 4909 |
+
"grad_norm": 0.13557648658752441,
|
| 4910 |
+
"learning_rate": 8.630207967246585e-05,
|
| 4911 |
+
"loss": 1.5324,
|
| 4912 |
+
"step": 682
|
| 4913 |
+
},
|
| 4914 |
+
{
|
| 4915 |
+
"epoch": 0.013091564279005577,
|
| 4916 |
+
"grad_norm": 0.09349766373634338,
|
| 4917 |
+
"learning_rate": 8.611755486651088e-05,
|
| 4918 |
+
"loss": 1.6017,
|
| 4919 |
+
"step": 683
|
| 4920 |
+
},
|
| 4921 |
+
{
|
| 4922 |
+
"epoch": 0.013110732015870886,
|
| 4923 |
+
"grad_norm": 0.18742144107818604,
|
| 4924 |
+
"learning_rate": 8.593179096512411e-05,
|
| 4925 |
+
"loss": 1.4779,
|
| 4926 |
+
"step": 684
|
| 4927 |
+
},
|
| 4928 |
+
{
|
| 4929 |
+
"epoch": 0.013129899752736195,
|
| 4930 |
+
"grad_norm": 0.1266089230775833,
|
| 4931 |
+
"learning_rate": 8.574479425572338e-05,
|
| 4932 |
+
"loss": 1.425,
|
| 4933 |
+
"step": 685
|
| 4934 |
+
},
|
| 4935 |
+
{
|
| 4936 |
+
"epoch": 0.013149067489601502,
|
| 4937 |
+
"grad_norm": 0.12737752497196198,
|
| 4938 |
+
"learning_rate": 8.555657106745254e-05,
|
| 4939 |
+
"loss": 1.2625,
|
| 4940 |
+
"step": 686
|
| 4941 |
+
},
|
| 4942 |
+
{
|
| 4943 |
+
"epoch": 0.013168235226466811,
|
| 4944 |
+
"grad_norm": 0.16241122782230377,
|
| 4945 |
+
"learning_rate": 8.536712777096716e-05,
|
| 4946 |
+
"loss": 1.5357,
|
| 4947 |
+
"step": 687
|
| 4948 |
+
},
|
| 4949 |
+
{
|
| 4950 |
+
"epoch": 0.01318740296333212,
|
| 4951 |
+
"grad_norm": 0.14777013659477234,
|
| 4952 |
+
"learning_rate": 8.517647077821897e-05,
|
| 4953 |
+
"loss": 1.4624,
|
| 4954 |
+
"step": 688
|
| 4955 |
+
},
|
| 4956 |
+
{
|
| 4957 |
+
"epoch": 0.013206570700197427,
|
| 4958 |
+
"grad_norm": 0.14423668384552002,
|
| 4959 |
+
"learning_rate": 8.498460654223876e-05,
|
| 4960 |
+
"loss": 1.627,
|
| 4961 |
+
"step": 689
|
| 4962 |
+
},
|
| 4963 |
+
{
|
| 4964 |
+
"epoch": 0.013225738437062736,
|
| 4965 |
+
"grad_norm": 0.162040114402771,
|
| 4966 |
+
"learning_rate": 8.479154155691808e-05,
|
| 4967 |
+
"loss": 1.2931,
|
| 4968 |
+
"step": 690
|
| 4969 |
+
},
|
| 4970 |
+
{
|
| 4971 |
+
"epoch": 0.013244906173928045,
|
| 4972 |
+
"grad_norm": 0.12327907979488373,
|
| 4973 |
+
"learning_rate": 8.459728235678932e-05,
|
| 4974 |
+
"loss": 1.7891,
|
| 4975 |
+
"step": 691
|
| 4976 |
+
},
|
| 4977 |
+
{
|
| 4978 |
+
"epoch": 0.013264073910793353,
|
| 4979 |
+
"grad_norm": 0.11705397814512253,
|
| 4980 |
+
"learning_rate": 8.440183551680466e-05,
|
| 4981 |
+
"loss": 1.8122,
|
| 4982 |
+
"step": 692
|
| 4983 |
+
},
|
| 4984 |
+
{
|
| 4985 |
+
"epoch": 0.01328324164765866,
|
| 4986 |
+
"grad_norm": 0.10630516707897186,
|
| 4987 |
+
"learning_rate": 8.420520765211347e-05,
|
| 4988 |
+
"loss": 1.1619,
|
| 4989 |
+
"step": 693
|
| 4990 |
+
},
|
| 4991 |
+
{
|
| 4992 |
+
"epoch": 0.01330240938452397,
|
| 4993 |
+
"grad_norm": 0.15495729446411133,
|
| 4994 |
+
"learning_rate": 8.400740541783835e-05,
|
| 4995 |
+
"loss": 1.4287,
|
| 4996 |
+
"step": 694
|
| 4997 |
+
},
|
| 4998 |
+
{
|
| 4999 |
+
"epoch": 0.013321577121389278,
|
| 5000 |
+
"grad_norm": 0.14559215307235718,
|
| 5001 |
+
"learning_rate": 8.380843550885003e-05,
|
| 5002 |
+
"loss": 1.0827,
|
| 5003 |
+
"step": 695
|
| 5004 |
+
},
|
| 5005 |
+
{
|
| 5006 |
+
"epoch": 0.013340744858254585,
|
| 5007 |
+
"grad_norm": 0.14757879078388214,
|
| 5008 |
+
"learning_rate": 8.360830465954068e-05,
|
| 5009 |
+
"loss": 1.3853,
|
| 5010 |
+
"step": 696
|
| 5011 |
+
},
|
| 5012 |
+
{
|
| 5013 |
+
"epoch": 0.013359912595119894,
|
| 5014 |
+
"grad_norm": 0.11220613121986389,
|
| 5015 |
+
"learning_rate": 8.340701964359591e-05,
|
| 5016 |
+
"loss": 1.2396,
|
| 5017 |
+
"step": 697
|
| 5018 |
+
},
|
| 5019 |
+
{
|
| 5020 |
+
"epoch": 0.013379080331985203,
|
| 5021 |
+
"grad_norm": 0.1057950034737587,
|
| 5022 |
+
"learning_rate": 8.32045872737657e-05,
|
| 5023 |
+
"loss": 1.2326,
|
| 5024 |
+
"step": 698
|
| 5025 |
+
},
|
| 5026 |
+
{
|
| 5027 |
+
"epoch": 0.01339824806885051,
|
| 5028 |
+
"grad_norm": 0.10915949940681458,
|
| 5029 |
+
"learning_rate": 8.300101440163359e-05,
|
| 5030 |
+
"loss": 1.1097,
|
| 5031 |
+
"step": 699
|
| 5032 |
+
},
|
| 5033 |
+
{
|
| 5034 |
+
"epoch": 0.013417415805715819,
|
| 5035 |
+
"grad_norm": 0.11548012495040894,
|
| 5036 |
+
"learning_rate": 8.279630791738493e-05,
|
| 5037 |
+
"loss": 1.6202,
|
| 5038 |
+
"step": 700
|
| 5039 |
+
},
|
| 5040 |
+
{
|
| 5041 |
+
"epoch": 0.013436583542581128,
|
| 5042 |
+
"grad_norm": 0.19324012100696564,
|
| 5043 |
+
"learning_rate": 8.259047474957364e-05,
|
| 5044 |
+
"loss": 1.486,
|
| 5045 |
+
"step": 701
|
| 5046 |
+
},
|
| 5047 |
+
{
|
| 5048 |
+
"epoch": 0.013455751279446435,
|
| 5049 |
+
"grad_norm": 0.12308330833911896,
|
| 5050 |
+
"learning_rate": 8.238352186488769e-05,
|
| 5051 |
+
"loss": 1.4034,
|
| 5052 |
+
"step": 702
|
| 5053 |
+
},
|
| 5054 |
+
{
|
| 5055 |
+
"epoch": 0.013474919016311744,
|
| 5056 |
+
"grad_norm": 0.1256161630153656,
|
| 5057 |
+
"learning_rate": 8.217545626791327e-05,
|
| 5058 |
+
"loss": 1.7307,
|
| 5059 |
+
"step": 703
|
| 5060 |
+
},
|
| 5061 |
+
{
|
| 5062 |
+
"epoch": 0.013494086753177053,
|
| 5063 |
+
"grad_norm": 0.12383873015642166,
|
| 5064 |
+
"learning_rate": 8.19662850008978e-05,
|
| 5065 |
+
"loss": 1.4614,
|
| 5066 |
+
"step": 704
|
| 5067 |
+
},
|
| 5068 |
+
{
|
| 5069 |
+
"epoch": 0.013513254490042361,
|
| 5070 |
+
"grad_norm": 0.11147335171699524,
|
| 5071 |
+
"learning_rate": 8.175601514351145e-05,
|
| 5072 |
+
"loss": 1.2766,
|
| 5073 |
+
"step": 705
|
| 5074 |
+
},
|
| 5075 |
+
{
|
| 5076 |
+
"epoch": 0.013532422226907669,
|
| 5077 |
+
"grad_norm": 0.12207506597042084,
|
| 5078 |
+
"learning_rate": 8.154465381260771e-05,
|
| 5079 |
+
"loss": 1.3754,
|
| 5080 |
+
"step": 706
|
| 5081 |
+
},
|
| 5082 |
+
{
|
| 5083 |
+
"epoch": 0.013551589963772977,
|
| 5084 |
+
"grad_norm": 0.09853308647871017,
|
| 5085 |
+
"learning_rate": 8.133220816198227e-05,
|
| 5086 |
+
"loss": 1.2895,
|
| 5087 |
+
"step": 707
|
| 5088 |
+
},
|
| 5089 |
+
{
|
| 5090 |
+
"epoch": 0.013570757700638286,
|
| 5091 |
+
"grad_norm": 0.16966822743415833,
|
| 5092 |
+
"learning_rate": 8.111868538213106e-05,
|
| 5093 |
+
"loss": 1.1311,
|
| 5094 |
+
"step": 708
|
| 5095 |
+
},
|
| 5096 |
+
{
|
| 5097 |
+
"epoch": 0.013589925437503593,
|
| 5098 |
+
"grad_norm": 0.19110709428787231,
|
| 5099 |
+
"learning_rate": 8.090409270000685e-05,
|
| 5100 |
+
"loss": 1.9365,
|
| 5101 |
+
"step": 709
|
| 5102 |
+
},
|
| 5103 |
+
{
|
| 5104 |
+
"epoch": 0.013609093174368902,
|
| 5105 |
+
"grad_norm": 0.09257730841636658,
|
| 5106 |
+
"learning_rate": 8.068843737877462e-05,
|
| 5107 |
+
"loss": 1.5642,
|
| 5108 |
+
"step": 710
|
| 5109 |
+
},
|
| 5110 |
+
{
|
| 5111 |
+
"epoch": 0.013628260911234211,
|
| 5112 |
+
"grad_norm": 0.103863924741745,
|
| 5113 |
+
"learning_rate": 8.047172671756572e-05,
|
| 5114 |
+
"loss": 1.0652,
|
| 5115 |
+
"step": 711
|
| 5116 |
+
},
|
| 5117 |
+
{
|
| 5118 |
+
"epoch": 0.013647428648099518,
|
| 5119 |
+
"grad_norm": 0.117330402135849,
|
| 5120 |
+
"learning_rate": 8.025396805123084e-05,
|
| 5121 |
+
"loss": 1.3352,
|
| 5122 |
+
"step": 712
|
| 5123 |
+
},
|
| 5124 |
+
{
|
| 5125 |
+
"epoch": 0.013666596384964827,
|
| 5126 |
+
"grad_norm": 0.13342571258544922,
|
| 5127 |
+
"learning_rate": 8.003516875009178e-05,
|
| 5128 |
+
"loss": 1.5733,
|
| 5129 |
+
"step": 713
|
| 5130 |
+
},
|
| 5131 |
+
{
|
| 5132 |
+
"epoch": 0.013685764121830136,
|
| 5133 |
+
"grad_norm": 0.13681145012378693,
|
| 5134 |
+
"learning_rate": 7.98153362196919e-05,
|
| 5135 |
+
"loss": 1.3734,
|
| 5136 |
+
"step": 714
|
| 5137 |
+
},
|
| 5138 |
+
{
|
| 5139 |
+
"epoch": 0.013704931858695445,
|
| 5140 |
+
"grad_norm": 0.1097632646560669,
|
| 5141 |
+
"learning_rate": 7.959447790054562e-05,
|
| 5142 |
+
"loss": 1.5992,
|
| 5143 |
+
"step": 715
|
| 5144 |
+
},
|
| 5145 |
+
{
|
| 5146 |
+
"epoch": 0.013724099595560752,
|
| 5147 |
+
"grad_norm": 0.15187889337539673,
|
| 5148 |
+
"learning_rate": 7.937260126788639e-05,
|
| 5149 |
+
"loss": 1.3056,
|
| 5150 |
+
"step": 716
|
| 5151 |
+
},
|
| 5152 |
+
{
|
| 5153 |
+
"epoch": 0.01374326733242606,
|
| 5154 |
+
"grad_norm": 0.13137780129909515,
|
| 5155 |
+
"learning_rate": 7.91497138314139e-05,
|
| 5156 |
+
"loss": 1.8176,
|
| 5157 |
+
"step": 717
|
| 5158 |
+
},
|
| 5159 |
+
{
|
| 5160 |
+
"epoch": 0.01376243506929137,
|
| 5161 |
+
"grad_norm": 0.09313193708658218,
|
| 5162 |
+
"learning_rate": 7.89258231350397e-05,
|
| 5163 |
+
"loss": 1.3236,
|
| 5164 |
+
"step": 718
|
| 5165 |
+
},
|
| 5166 |
+
{
|
| 5167 |
+
"epoch": 0.013781602806156677,
|
| 5168 |
+
"grad_norm": 0.14475251734256744,
|
| 5169 |
+
"learning_rate": 7.8700936756632e-05,
|
| 5170 |
+
"loss": 1.7993,
|
| 5171 |
+
"step": 719
|
| 5172 |
+
},
|
| 5173 |
+
{
|
| 5174 |
+
"epoch": 0.013800770543021985,
|
| 5175 |
+
"grad_norm": 0.14910048246383667,
|
| 5176 |
+
"learning_rate": 7.847506230775914e-05,
|
| 5177 |
+
"loss": 1.402,
|
| 5178 |
+
"step": 720
|
| 5179 |
+
},
|
| 5180 |
+
{
|
| 5181 |
+
"epoch": 0.013819938279887294,
|
| 5182 |
+
"grad_norm": 0.11825945228338242,
|
| 5183 |
+
"learning_rate": 7.824820743343202e-05,
|
| 5184 |
+
"loss": 1.266,
|
| 5185 |
+
"step": 721
|
| 5186 |
+
},
|
| 5187 |
+
{
|
| 5188 |
+
"epoch": 0.013839106016752601,
|
| 5189 |
+
"grad_norm": 0.10240902751684189,
|
| 5190 |
+
"learning_rate": 7.802037981184523e-05,
|
| 5191 |
+
"loss": 1.3654,
|
| 5192 |
+
"step": 722
|
| 5193 |
+
},
|
| 5194 |
+
{
|
| 5195 |
+
"epoch": 0.01385827375361791,
|
| 5196 |
+
"grad_norm": 0.14030222594738007,
|
| 5197 |
+
"learning_rate": 7.779158715411727e-05,
|
| 5198 |
+
"loss": 1.2747,
|
| 5199 |
+
"step": 723
|
| 5200 |
+
},
|
| 5201 |
+
{
|
| 5202 |
+
"epoch": 0.013877441490483219,
|
| 5203 |
+
"grad_norm": 0.1306154876947403,
|
| 5204 |
+
"learning_rate": 7.756183720402953e-05,
|
| 5205 |
+
"loss": 1.6009,
|
| 5206 |
+
"step": 724
|
| 5207 |
+
},
|
| 5208 |
+
{
|
| 5209 |
+
"epoch": 0.013896609227348526,
|
| 5210 |
+
"grad_norm": 0.1406806856393814,
|
| 5211 |
+
"learning_rate": 7.733113773776421e-05,
|
| 5212 |
+
"loss": 1.3401,
|
| 5213 |
+
"step": 725
|
| 5214 |
+
},
|
| 5215 |
+
{
|
| 5216 |
+
"epoch": 0.013915776964213835,
|
| 5217 |
+
"grad_norm": 0.16746580600738525,
|
| 5218 |
+
"learning_rate": 7.709949656364105e-05,
|
| 5219 |
+
"loss": 1.7965,
|
| 5220 |
+
"step": 726
|
| 5221 |
+
},
|
| 5222 |
+
{
|
| 5223 |
+
"epoch": 0.013934944701079144,
|
| 5224 |
+
"grad_norm": 0.11773744970560074,
|
| 5225 |
+
"learning_rate": 7.686692152185316e-05,
|
| 5226 |
+
"loss": 1.488,
|
| 5227 |
+
"step": 727
|
| 5228 |
+
},
|
| 5229 |
+
{
|
| 5230 |
+
"epoch": 0.013954112437944453,
|
| 5231 |
+
"grad_norm": 0.1136757880449295,
|
| 5232 |
+
"learning_rate": 7.663342048420154e-05,
|
| 5233 |
+
"loss": 1.6238,
|
| 5234 |
+
"step": 728
|
| 5235 |
+
},
|
| 5236 |
+
{
|
| 5237 |
+
"epoch": 0.01397328017480976,
|
| 5238 |
+
"grad_norm": 0.10973573476076126,
|
| 5239 |
+
"learning_rate": 7.639900135382878e-05,
|
| 5240 |
+
"loss": 1.4109,
|
| 5241 |
+
"step": 729
|
| 5242 |
+
},
|
| 5243 |
+
{
|
| 5244 |
+
"epoch": 0.013992447911675069,
|
| 5245 |
+
"grad_norm": 0.11506867408752441,
|
| 5246 |
+
"learning_rate": 7.616367206495144e-05,
|
| 5247 |
+
"loss": 1.4051,
|
| 5248 |
+
"step": 730
|
| 5249 |
+
},
|
| 5250 |
+
{
|
| 5251 |
+
"epoch": 0.014011615648540377,
|
| 5252 |
+
"grad_norm": 0.16806761920452118,
|
| 5253 |
+
"learning_rate": 7.59274405825916e-05,
|
| 5254 |
+
"loss": 1.4472,
|
| 5255 |
+
"step": 731
|
| 5256 |
+
},
|
| 5257 |
+
{
|
| 5258 |
+
"epoch": 0.014030783385405685,
|
| 5259 |
+
"grad_norm": 0.0913873165845871,
|
| 5260 |
+
"learning_rate": 7.569031490230721e-05,
|
| 5261 |
+
"loss": 1.2027,
|
| 5262 |
+
"step": 732
|
| 5263 |
+
},
|
| 5264 |
+
{
|
| 5265 |
+
"epoch": 0.014049951122270993,
|
| 5266 |
+
"grad_norm": 0.12903502583503723,
|
| 5267 |
+
"learning_rate": 7.545230304992151e-05,
|
| 5268 |
+
"loss": 1.3709,
|
| 5269 |
+
"step": 733
|
| 5270 |
+
},
|
| 5271 |
+
{
|
| 5272 |
+
"epoch": 0.014069118859136302,
|
| 5273 |
+
"grad_norm": 0.13405589759349823,
|
| 5274 |
+
"learning_rate": 7.521341308125137e-05,
|
| 5275 |
+
"loss": 1.4219,
|
| 5276 |
+
"step": 734
|
| 5277 |
+
},
|
| 5278 |
+
{
|
| 5279 |
+
"epoch": 0.01408828659600161,
|
| 5280 |
+
"grad_norm": 0.09643625468015671,
|
| 5281 |
+
"learning_rate": 7.497365308183465e-05,
|
| 5282 |
+
"loss": 1.2863,
|
| 5283 |
+
"step": 735
|
| 5284 |
+
},
|
| 5285 |
+
{
|
| 5286 |
+
"epoch": 0.014107454332866918,
|
| 5287 |
+
"grad_norm": 0.11383571475744247,
|
| 5288 |
+
"learning_rate": 7.47330311666565e-05,
|
| 5289 |
+
"loss": 1.3964,
|
| 5290 |
+
"step": 736
|
| 5291 |
+
},
|
| 5292 |
+
{
|
| 5293 |
+
"epoch": 0.014126622069732227,
|
| 5294 |
+
"grad_norm": 0.10297911614179611,
|
| 5295 |
+
"learning_rate": 7.449155547987473e-05,
|
| 5296 |
+
"loss": 1.8177,
|
| 5297 |
+
"step": 737
|
| 5298 |
+
},
|
| 5299 |
+
{
|
| 5300 |
+
"epoch": 0.014145789806597534,
|
| 5301 |
+
"grad_norm": 0.12988147139549255,
|
| 5302 |
+
"learning_rate": 7.424923419454413e-05,
|
| 5303 |
+
"loss": 1.6083,
|
| 5304 |
+
"step": 738
|
| 5305 |
+
},
|
| 5306 |
+
{
|
| 5307 |
+
"epoch": 0.014164957543462843,
|
| 5308 |
+
"grad_norm": 0.10675617307424545,
|
| 5309 |
+
"learning_rate": 7.400607551233985e-05,
|
| 5310 |
+
"loss": 1.3063,
|
| 5311 |
+
"step": 739
|
| 5312 |
+
},
|
| 5313 |
+
{
|
| 5314 |
+
"epoch": 0.014184125280328152,
|
| 5315 |
+
"grad_norm": 0.1297062337398529,
|
| 5316 |
+
"learning_rate": 7.376208766327985e-05,
|
| 5317 |
+
"loss": 1.2958,
|
| 5318 |
+
"step": 740
|
| 5319 |
+
},
|
| 5320 |
+
{
|
| 5321 |
+
"epoch": 0.01420329301719346,
|
| 5322 |
+
"grad_norm": 0.12037523835897446,
|
| 5323 |
+
"learning_rate": 7.351727890544627e-05,
|
| 5324 |
+
"loss": 1.0633,
|
| 5325 |
+
"step": 741
|
| 5326 |
+
},
|
| 5327 |
+
{
|
| 5328 |
+
"epoch": 0.014222460754058768,
|
| 5329 |
+
"grad_norm": 0.1311262995004654,
|
| 5330 |
+
"learning_rate": 7.3271657524706e-05,
|
| 5331 |
+
"loss": 1.6545,
|
| 5332 |
+
"step": 742
|
| 5333 |
+
},
|
| 5334 |
+
{
|
| 5335 |
+
"epoch": 0.014241628490924077,
|
| 5336 |
+
"grad_norm": 0.14785340428352356,
|
| 5337 |
+
"learning_rate": 7.302523183443017e-05,
|
| 5338 |
+
"loss": 1.624,
|
| 5339 |
+
"step": 743
|
| 5340 |
+
},
|
| 5341 |
+
{
|
| 5342 |
+
"epoch": 0.014260796227789386,
|
| 5343 |
+
"grad_norm": 0.11753561347723007,
|
| 5344 |
+
"learning_rate": 7.277801017521278e-05,
|
| 5345 |
+
"loss": 1.2393,
|
| 5346 |
+
"step": 744
|
| 5347 |
+
},
|
| 5348 |
+
{
|
| 5349 |
+
"epoch": 0.014279963964654693,
|
| 5350 |
+
"grad_norm": 0.1260678768157959,
|
| 5351 |
+
"learning_rate": 7.253000091458846e-05,
|
| 5352 |
+
"loss": 1.3294,
|
| 5353 |
+
"step": 745
|
| 5354 |
+
},
|
| 5355 |
+
{
|
| 5356 |
+
"epoch": 0.014299131701520001,
|
| 5357 |
+
"grad_norm": 0.12359932065010071,
|
| 5358 |
+
"learning_rate": 7.22812124467492e-05,
|
| 5359 |
+
"loss": 1.799,
|
| 5360 |
+
"step": 746
|
| 5361 |
+
},
|
| 5362 |
+
{
|
| 5363 |
+
"epoch": 0.01431829943838531,
|
| 5364 |
+
"grad_norm": 0.14787715673446655,
|
| 5365 |
+
"learning_rate": 7.20316531922603e-05,
|
| 5366 |
+
"loss": 1.7779,
|
| 5367 |
+
"step": 747
|
| 5368 |
+
},
|
| 5369 |
+
{
|
| 5370 |
+
"epoch": 0.014337467175250617,
|
| 5371 |
+
"grad_norm": 0.1288924664258957,
|
| 5372 |
+
"learning_rate": 7.178133159777526e-05,
|
| 5373 |
+
"loss": 1.6288,
|
| 5374 |
+
"step": 748
|
| 5375 |
+
},
|
| 5376 |
+
{
|
| 5377 |
+
"epoch": 0.014356634912115926,
|
| 5378 |
+
"grad_norm": 0.1537061333656311,
|
| 5379 |
+
"learning_rate": 7.153025613574998e-05,
|
| 5380 |
+
"loss": 1.2679,
|
| 5381 |
+
"step": 749
|
| 5382 |
+
},
|
| 5383 |
+
{
|
| 5384 |
+
"epoch": 0.014375802648981235,
|
| 5385 |
+
"grad_norm": 0.14872634410858154,
|
| 5386 |
+
"learning_rate": 7.127843530415601e-05,
|
| 5387 |
+
"loss": 1.3101,
|
| 5388 |
+
"step": 750
|
| 5389 |
+
},
|
| 5390 |
+
{
|
| 5391 |
+
"epoch": 0.014375802648981235,
|
| 5392 |
+
"eval_loss": 1.4733232259750366,
|
| 5393 |
+
"eval_runtime": 2557.9542,
|
| 5394 |
+
"eval_samples_per_second": 36.259,
|
| 5395 |
+
"eval_steps_per_second": 18.129,
|
| 5396 |
+
"step": 750
|
| 5397 |
+
},
|
| 5398 |
+
{
|
| 5399 |
+
"epoch": 0.014375802648981235,
|
| 5400 |
+
"eval_loss": 1.5129666328430176,
|
| 5401 |
+
"eval_runtime": 2524.1983,
|
| 5402 |
+
"eval_samples_per_second": 36.744,
|
| 5403 |
+
"eval_steps_per_second": 18.372,
|
| 5404 |
+
"step": 750
|
| 5405 |
+
},
|
| 5406 |
+
{
|
| 5407 |
+
"epoch": 0.014394970385846544,
|
| 5408 |
+
"grad_norm": 0.14070995151996613,
|
| 5409 |
+
"learning_rate": 7.102587762619283e-05,
|
| 5410 |
+
"loss": 1.5705,
|
| 5411 |
+
"step": 751
|
| 5412 |
+
},
|
| 5413 |
+
{
|
| 5414 |
+
"epoch": 0.014414138122711851,
|
| 5415 |
+
"grad_norm": 0.1251671314239502,
|
| 5416 |
+
"learning_rate": 7.077259164999948e-05,
|
| 5417 |
+
"loss": 1.6186,
|
| 5418 |
+
"step": 752
|
| 5419 |
+
},
|
| 5420 |
+
{
|
| 5421 |
+
"epoch": 0.01443330585957716,
|
| 5422 |
+
"grad_norm": 0.16820262372493744,
|
| 5423 |
+
"learning_rate": 7.051858594836516e-05,
|
| 5424 |
+
"loss": 1.7081,
|
| 5425 |
+
"step": 753
|
| 5426 |
+
},
|
| 5427 |
+
{
|
| 5428 |
+
"epoch": 0.014452473596442469,
|
| 5429 |
+
"grad_norm": 0.12928839027881622,
|
| 5430 |
+
"learning_rate": 7.026386911843915e-05,
|
| 5431 |
+
"loss": 1.565,
|
| 5432 |
+
"step": 754
|
| 5433 |
+
},
|
| 5434 |
+
{
|
| 5435 |
+
"epoch": 0.014471641333307776,
|
| 5436 |
+
"grad_norm": 0.13493424654006958,
|
| 5437 |
+
"learning_rate": 7.000844978143971e-05,
|
| 5438 |
+
"loss": 1.6153,
|
| 5439 |
+
"step": 755
|
| 5440 |
+
},
|
| 5441 |
+
{
|
| 5442 |
+
"epoch": 0.014490809070173085,
|
| 5443 |
+
"grad_norm": 0.12037918716669083,
|
| 5444 |
+
"learning_rate": 6.975233658236241e-05,
|
| 5445 |
+
"loss": 1.7057,
|
| 5446 |
+
"step": 756
|
| 5447 |
+
},
|
| 5448 |
+
{
|
| 5449 |
+
"epoch": 0.014509976807038394,
|
| 5450 |
+
"grad_norm": 0.14160001277923584,
|
| 5451 |
+
"learning_rate": 6.949553818968743e-05,
|
| 5452 |
+
"loss": 1.3393,
|
| 5453 |
+
"step": 757
|
| 5454 |
+
},
|
| 5455 |
+
{
|
| 5456 |
+
"epoch": 0.0145291445439037,
|
| 5457 |
+
"grad_norm": 0.12361916899681091,
|
| 5458 |
+
"learning_rate": 6.923806329508627e-05,
|
| 5459 |
+
"loss": 1.6205,
|
| 5460 |
+
"step": 758
|
| 5461 |
+
},
|
| 5462 |
+
{
|
| 5463 |
+
"epoch": 0.01454831228076901,
|
| 5464 |
+
"grad_norm": 0.11223111301660538,
|
| 5465 |
+
"learning_rate": 6.897992061312742e-05,
|
| 5466 |
+
"loss": 1.8051,
|
| 5467 |
+
"step": 759
|
| 5468 |
+
},
|
| 5469 |
+
{
|
| 5470 |
+
"epoch": 0.014567480017634318,
|
| 5471 |
+
"grad_norm": 0.10139868408441544,
|
| 5472 |
+
"learning_rate": 6.872111888098156e-05,
|
| 5473 |
+
"loss": 1.1499,
|
| 5474 |
+
"step": 760
|
| 5475 |
+
},
|
| 5476 |
+
{
|
| 5477 |
+
"epoch": 0.014586647754499625,
|
| 5478 |
+
"grad_norm": 0.14185746014118195,
|
| 5479 |
+
"learning_rate": 6.846166685812576e-05,
|
| 5480 |
+
"loss": 1.6056,
|
| 5481 |
+
"step": 761
|
| 5482 |
+
},
|
| 5483 |
+
{
|
| 5484 |
+
"epoch": 0.014605815491364934,
|
| 5485 |
+
"grad_norm": 0.12362541258335114,
|
| 5486 |
+
"learning_rate": 6.8201573326047e-05,
|
| 5487 |
+
"loss": 1.0905,
|
| 5488 |
+
"step": 762
|
| 5489 |
+
},
|
| 5490 |
+
{
|
| 5491 |
+
"epoch": 0.014624983228230243,
|
| 5492 |
+
"grad_norm": 0.14459389448165894,
|
| 5493 |
+
"learning_rate": 6.794084708794496e-05,
|
| 5494 |
+
"loss": 1.4114,
|
| 5495 |
+
"step": 763
|
| 5496 |
+
},
|
| 5497 |
+
{
|
| 5498 |
+
"epoch": 0.014644150965095552,
|
| 5499 |
+
"grad_norm": 0.13407780230045319,
|
| 5500 |
+
"learning_rate": 6.767949696843406e-05,
|
| 5501 |
+
"loss": 1.4494,
|
| 5502 |
+
"step": 764
|
| 5503 |
+
},
|
| 5504 |
+
{
|
| 5505 |
+
"epoch": 0.014663318701960859,
|
| 5506 |
+
"grad_norm": 0.10757341235876083,
|
| 5507 |
+
"learning_rate": 6.74175318132448e-05,
|
| 5508 |
+
"loss": 1.6365,
|
| 5509 |
+
"step": 765
|
| 5510 |
+
},
|
| 5511 |
+
{
|
| 5512 |
+
"epoch": 0.014682486438826168,
|
| 5513 |
+
"grad_norm": 0.11026965826749802,
|
| 5514 |
+
"learning_rate": 6.71549604889244e-05,
|
| 5515 |
+
"loss": 1.435,
|
| 5516 |
+
"step": 766
|
| 5517 |
+
},
|
| 5518 |
+
{
|
| 5519 |
+
"epoch": 0.014701654175691477,
|
| 5520 |
+
"grad_norm": 0.12049650400876999,
|
| 5521 |
+
"learning_rate": 6.689179188253655e-05,
|
| 5522 |
+
"loss": 1.4351,
|
| 5523 |
+
"step": 767
|
| 5524 |
+
},
|
| 5525 |
+
{
|
| 5526 |
+
"epoch": 0.014720821912556784,
|
| 5527 |
+
"grad_norm": 0.10785556584596634,
|
| 5528 |
+
"learning_rate": 6.662803490136081e-05,
|
| 5529 |
+
"loss": 1.496,
|
| 5530 |
+
"step": 768
|
| 5531 |
+
},
|
| 5532 |
+
{
|
| 5533 |
+
"epoch": 0.014739989649422093,
|
| 5534 |
+
"grad_norm": 0.09526213258504868,
|
| 5535 |
+
"learning_rate": 6.636369847259102e-05,
|
| 5536 |
+
"loss": 1.3857,
|
| 5537 |
+
"step": 769
|
| 5538 |
+
},
|
| 5539 |
+
{
|
| 5540 |
+
"epoch": 0.014759157386287402,
|
| 5541 |
+
"grad_norm": 0.12694235146045685,
|
| 5542 |
+
"learning_rate": 6.609879154303313e-05,
|
| 5543 |
+
"loss": 1.5483,
|
| 5544 |
+
"step": 770
|
| 5545 |
+
},
|
| 5546 |
+
{
|
| 5547 |
+
"epoch": 0.014778325123152709,
|
| 5548 |
+
"grad_norm": 0.1330501288175583,
|
| 5549 |
+
"learning_rate": 6.583332307880244e-05,
|
| 5550 |
+
"loss": 1.4761,
|
| 5551 |
+
"step": 771
|
| 5552 |
+
},
|
| 5553 |
+
{
|
| 5554 |
+
"epoch": 0.014797492860018018,
|
| 5555 |
+
"grad_norm": 0.21376430988311768,
|
| 5556 |
+
"learning_rate": 6.556730206502014e-05,
|
| 5557 |
+
"loss": 1.491,
|
| 5558 |
+
"step": 772
|
| 5559 |
+
},
|
| 5560 |
+
{
|
| 5561 |
+
"epoch": 0.014816660596883326,
|
| 5562 |
+
"grad_norm": 0.14251284301280975,
|
| 5563 |
+
"learning_rate": 6.530073750550914e-05,
|
| 5564 |
+
"loss": 1.3677,
|
| 5565 |
+
"step": 773
|
| 5566 |
+
},
|
| 5567 |
+
{
|
| 5568 |
+
"epoch": 0.014835828333748633,
|
| 5569 |
+
"grad_norm": 0.14213839173316956,
|
| 5570 |
+
"learning_rate": 6.503363842248935e-05,
|
| 5571 |
+
"loss": 1.6587,
|
| 5572 |
+
"step": 774
|
| 5573 |
+
},
|
| 5574 |
+
{
|
| 5575 |
+
"epoch": 0.014854996070613942,
|
| 5576 |
+
"grad_norm": 0.13434764742851257,
|
| 5577 |
+
"learning_rate": 6.476601385627237e-05,
|
| 5578 |
+
"loss": 1.4309,
|
| 5579 |
+
"step": 775
|
| 5580 |
+
},
|
| 5581 |
+
{
|
| 5582 |
+
"epoch": 0.014874163807479251,
|
| 5583 |
+
"grad_norm": 0.12511897087097168,
|
| 5584 |
+
"learning_rate": 6.449787286495535e-05,
|
| 5585 |
+
"loss": 1.4375,
|
| 5586 |
+
"step": 776
|
| 5587 |
+
},
|
| 5588 |
+
{
|
| 5589 |
+
"epoch": 0.01489333154434456,
|
| 5590 |
+
"grad_norm": 0.12347516417503357,
|
| 5591 |
+
"learning_rate": 6.422922452411462e-05,
|
| 5592 |
+
"loss": 1.2932,
|
| 5593 |
+
"step": 777
|
| 5594 |
+
},
|
| 5595 |
+
{
|
| 5596 |
+
"epoch": 0.014912499281209867,
|
| 5597 |
+
"grad_norm": 0.0852687656879425,
|
| 5598 |
+
"learning_rate": 6.396007792649831e-05,
|
| 5599 |
+
"loss": 1.3767,
|
| 5600 |
+
"step": 778
|
| 5601 |
+
},
|
| 5602 |
+
{
|
| 5603 |
+
"epoch": 0.014931667018075176,
|
| 5604 |
+
"grad_norm": 0.14227139949798584,
|
| 5605 |
+
"learning_rate": 6.36904421817188e-05,
|
| 5606 |
+
"loss": 1.4536,
|
| 5607 |
+
"step": 779
|
| 5608 |
+
},
|
| 5609 |
+
{
|
| 5610 |
+
"epoch": 0.014950834754940485,
|
| 5611 |
+
"grad_norm": 0.13570551574230194,
|
| 5612 |
+
"learning_rate": 6.342032641594417e-05,
|
| 5613 |
+
"loss": 1.7646,
|
| 5614 |
+
"step": 780
|
| 5615 |
+
},
|
| 5616 |
+
{
|
| 5617 |
+
"epoch": 0.014970002491805792,
|
| 5618 |
+
"grad_norm": 0.14132581651210785,
|
| 5619 |
+
"learning_rate": 6.31497397715895e-05,
|
| 5620 |
+
"loss": 1.4213,
|
| 5621 |
+
"step": 781
|
| 5622 |
+
},
|
| 5623 |
+
{
|
| 5624 |
+
"epoch": 0.0149891702286711,
|
| 5625 |
+
"grad_norm": 0.13665030896663666,
|
| 5626 |
+
"learning_rate": 6.287869140700733e-05,
|
| 5627 |
+
"loss": 1.0851,
|
| 5628 |
+
"step": 782
|
| 5629 |
+
},
|
| 5630 |
+
{
|
| 5631 |
+
"epoch": 0.01500833796553641,
|
| 5632 |
+
"grad_norm": 0.1134626641869545,
|
| 5633 |
+
"learning_rate": 6.260719049617774e-05,
|
| 5634 |
+
"loss": 1.1517,
|
| 5635 |
+
"step": 783
|
| 5636 |
+
},
|
| 5637 |
+
{
|
| 5638 |
+
"epoch": 0.015027505702401717,
|
| 5639 |
+
"grad_norm": 0.1048184409737587,
|
| 5640 |
+
"learning_rate": 6.233524622839781e-05,
|
| 5641 |
+
"loss": 1.2068,
|
| 5642 |
+
"step": 784
|
| 5643 |
+
},
|
| 5644 |
+
{
|
| 5645 |
+
"epoch": 0.015046673439267026,
|
| 5646 |
+
"grad_norm": 0.10722368210554123,
|
| 5647 |
+
"learning_rate": 6.206286780797056e-05,
|
| 5648 |
+
"loss": 1.5125,
|
| 5649 |
+
"step": 785
|
| 5650 |
+
},
|
| 5651 |
+
{
|
| 5652 |
+
"epoch": 0.015065841176132334,
|
| 5653 |
+
"grad_norm": 0.1135949194431305,
|
| 5654 |
+
"learning_rate": 6.179006445389355e-05,
|
| 5655 |
+
"loss": 1.3569,
|
| 5656 |
+
"step": 786
|
| 5657 |
+
},
|
| 5658 |
+
{
|
| 5659 |
+
"epoch": 0.015085008912997643,
|
| 5660 |
+
"grad_norm": 0.14543752372264862,
|
| 5661 |
+
"learning_rate": 6.151684539954667e-05,
|
| 5662 |
+
"loss": 1.4925,
|
| 5663 |
+
"step": 787
|
| 5664 |
+
},
|
| 5665 |
+
{
|
| 5666 |
+
"epoch": 0.01510417664986295,
|
| 5667 |
+
"grad_norm": 0.18014617264270782,
|
| 5668 |
+
"learning_rate": 6.12432198923798e-05,
|
| 5669 |
+
"loss": 1.7051,
|
| 5670 |
+
"step": 788
|
| 5671 |
+
},
|
| 5672 |
+
{
|
| 5673 |
+
"epoch": 0.01512334438672826,
|
| 5674 |
+
"grad_norm": 0.14088796079158783,
|
| 5675 |
+
"learning_rate": 6.096919719359972e-05,
|
| 5676 |
+
"loss": 1.7405,
|
| 5677 |
+
"step": 789
|
| 5678 |
+
},
|
| 5679 |
+
{
|
| 5680 |
+
"epoch": 0.015142512123593568,
|
| 5681 |
+
"grad_norm": 0.10348264127969742,
|
| 5682 |
+
"learning_rate": 6.069478657785659e-05,
|
| 5683 |
+
"loss": 1.6423,
|
| 5684 |
+
"step": 790
|
| 5685 |
+
},
|
| 5686 |
+
{
|
| 5687 |
+
"epoch": 0.015161679860458875,
|
| 5688 |
+
"grad_norm": 0.20880816876888275,
|
| 5689 |
+
"learning_rate": 6.0419997332930257e-05,
|
| 5690 |
+
"loss": 1.375,
|
| 5691 |
+
"step": 791
|
| 5692 |
+
},
|
| 5693 |
+
{
|
| 5694 |
+
"epoch": 0.015180847597324184,
|
| 5695 |
+
"grad_norm": 0.14411678910255432,
|
| 5696 |
+
"learning_rate": 6.014483875941564e-05,
|
| 5697 |
+
"loss": 1.1963,
|
| 5698 |
+
"step": 792
|
| 5699 |
+
},
|
| 5700 |
+
{
|
| 5701 |
+
"epoch": 0.015200015334189493,
|
| 5702 |
+
"grad_norm": 0.11277009546756744,
|
| 5703 |
+
"learning_rate": 5.986932017040813e-05,
|
| 5704 |
+
"loss": 1.5241,
|
| 5705 |
+
"step": 793
|
| 5706 |
+
},
|
| 5707 |
+
{
|
| 5708 |
+
"epoch": 0.0152191830710548,
|
| 5709 |
+
"grad_norm": 0.11883819848299026,
|
| 5710 |
+
"learning_rate": 5.959345089118825e-05,
|
| 5711 |
+
"loss": 1.035,
|
| 5712 |
+
"step": 794
|
| 5713 |
+
},
|
| 5714 |
+
{
|
| 5715 |
+
"epoch": 0.015238350807920109,
|
| 5716 |
+
"grad_norm": 0.13488759100437164,
|
| 5717 |
+
"learning_rate": 5.9317240258906155e-05,
|
| 5718 |
+
"loss": 1.4593,
|
| 5719 |
+
"step": 795
|
| 5720 |
+
},
|
| 5721 |
+
{
|
| 5722 |
+
"epoch": 0.015257518544785418,
|
| 5723 |
+
"grad_norm": 0.15183618664741516,
|
| 5724 |
+
"learning_rate": 5.904069762226545e-05,
|
| 5725 |
+
"loss": 1.2563,
|
| 5726 |
+
"step": 796
|
| 5727 |
+
},
|
| 5728 |
+
{
|
| 5729 |
+
"epoch": 0.015276686281650725,
|
| 5730 |
+
"grad_norm": 0.14672145247459412,
|
| 5731 |
+
"learning_rate": 5.876383234120691e-05,
|
| 5732 |
+
"loss": 1.2782,
|
| 5733 |
+
"step": 797
|
| 5734 |
+
},
|
| 5735 |
+
{
|
| 5736 |
+
"epoch": 0.015295854018516034,
|
| 5737 |
+
"grad_norm": 0.1341315656900406,
|
| 5738 |
+
"learning_rate": 5.84866537865916e-05,
|
| 5739 |
+
"loss": 1.4439,
|
| 5740 |
+
"step": 798
|
| 5741 |
+
},
|
| 5742 |
+
{
|
| 5743 |
+
"epoch": 0.015315021755381342,
|
| 5744 |
+
"grad_norm": 0.11749543994665146,
|
| 5745 |
+
"learning_rate": 5.820917133988376e-05,
|
| 5746 |
+
"loss": 1.244,
|
| 5747 |
+
"step": 799
|
| 5748 |
+
},
|
| 5749 |
+
{
|
| 5750 |
+
"epoch": 0.015334189492246651,
|
| 5751 |
+
"grad_norm": 0.12214719504117966,
|
| 5752 |
+
"learning_rate": 5.793139439283322e-05,
|
| 5753 |
+
"loss": 1.3007,
|
| 5754 |
+
"step": 800
|
| 5755 |
+
},
|
| 5756 |
+
{
|
| 5757 |
+
"epoch": 0.015353357229111958,
|
| 5758 |
+
"grad_norm": 0.13016489148139954,
|
| 5759 |
+
"learning_rate": 5.765333234715756e-05,
|
| 5760 |
+
"loss": 1.469,
|
| 5761 |
+
"step": 801
|
| 5762 |
+
},
|
| 5763 |
+
{
|
| 5764 |
+
"epoch": 0.015372524965977267,
|
| 5765 |
+
"grad_norm": 0.16239777207374573,
|
| 5766 |
+
"learning_rate": 5.737499461422387e-05,
|
| 5767 |
+
"loss": 1.2896,
|
| 5768 |
+
"step": 802
|
| 5769 |
+
},
|
| 5770 |
+
{
|
| 5771 |
+
"epoch": 0.015391692702842576,
|
| 5772 |
+
"grad_norm": 0.13130241632461548,
|
| 5773 |
+
"learning_rate": 5.7096390614730276e-05,
|
| 5774 |
+
"loss": 1.5951,
|
| 5775 |
+
"step": 803
|
| 5776 |
+
},
|
| 5777 |
+
{
|
| 5778 |
+
"epoch": 0.015410860439707883,
|
| 5779 |
+
"grad_norm": 0.14913310110569,
|
| 5780 |
+
"learning_rate": 5.6817529778387e-05,
|
| 5781 |
+
"loss": 1.3024,
|
| 5782 |
+
"step": 804
|
| 5783 |
+
},
|
| 5784 |
+
{
|
| 5785 |
+
"epoch": 0.015430028176573192,
|
| 5786 |
+
"grad_norm": 0.12097875028848648,
|
| 5787 |
+
"learning_rate": 5.653842154359724e-05,
|
| 5788 |
+
"loss": 1.1658,
|
| 5789 |
+
"step": 805
|
| 5790 |
+
},
|
| 5791 |
+
{
|
| 5792 |
+
"epoch": 0.0154491959134385,
|
| 5793 |
+
"grad_norm": 0.18160229921340942,
|
| 5794 |
+
"learning_rate": 5.625907535713775e-05,
|
| 5795 |
+
"loss": 1.3462,
|
| 5796 |
+
"step": 806
|
| 5797 |
+
},
|
| 5798 |
+
{
|
| 5799 |
+
"epoch": 0.015468363650303808,
|
| 5800 |
+
"grad_norm": 0.1442558765411377,
|
| 5801 |
+
"learning_rate": 5.5979500673839004e-05,
|
| 5802 |
+
"loss": 1.3229,
|
| 5803 |
+
"step": 807
|
| 5804 |
+
},
|
| 5805 |
+
{
|
| 5806 |
+
"epoch": 0.015487531387169117,
|
| 5807 |
+
"grad_norm": 0.1372595578432083,
|
| 5808 |
+
"learning_rate": 5.5699706956265294e-05,
|
| 5809 |
+
"loss": 1.5033,
|
| 5810 |
+
"step": 808
|
| 5811 |
+
},
|
| 5812 |
+
{
|
| 5813 |
+
"epoch": 0.015506699124034426,
|
| 5814 |
+
"grad_norm": 0.13811466097831726,
|
| 5815 |
+
"learning_rate": 5.541970367439436e-05,
|
| 5816 |
+
"loss": 1.2323,
|
| 5817 |
+
"step": 809
|
| 5818 |
+
},
|
| 5819 |
+
{
|
| 5820 |
+
"epoch": 0.015525866860899733,
|
| 5821 |
+
"grad_norm": 0.09932932257652283,
|
| 5822 |
+
"learning_rate": 5.513950030529698e-05,
|
| 5823 |
+
"loss": 1.4305,
|
| 5824 |
+
"step": 810
|
| 5825 |
+
},
|
| 5826 |
+
{
|
| 5827 |
+
"epoch": 0.015545034597765042,
|
| 5828 |
+
"grad_norm": 0.1343100517988205,
|
| 5829 |
+
"learning_rate": 5.485910633281609e-05,
|
| 5830 |
+
"loss": 1.3607,
|
| 5831 |
+
"step": 811
|
| 5832 |
+
},
|
| 5833 |
+
{
|
| 5834 |
+
"epoch": 0.01556420233463035,
|
| 5835 |
+
"grad_norm": 0.1169649213552475,
|
| 5836 |
+
"learning_rate": 5.457853124724585e-05,
|
| 5837 |
+
"loss": 1.5473,
|
| 5838 |
+
"step": 812
|
| 5839 |
+
},
|
| 5840 |
+
{
|
| 5841 |
+
"epoch": 0.01558337007149566,
|
| 5842 |
+
"grad_norm": 0.10100696980953217,
|
| 5843 |
+
"learning_rate": 5.4297784545010445e-05,
|
| 5844 |
+
"loss": 1.2088,
|
| 5845 |
+
"step": 813
|
| 5846 |
+
},
|
| 5847 |
+
{
|
| 5848 |
+
"epoch": 0.015602537808360966,
|
| 5849 |
+
"grad_norm": 0.15452361106872559,
|
| 5850 |
+
"learning_rate": 5.4016875728342645e-05,
|
| 5851 |
+
"loss": 1.193,
|
| 5852 |
+
"step": 814
|
| 5853 |
+
},
|
| 5854 |
+
{
|
| 5855 |
+
"epoch": 0.015621705545226275,
|
| 5856 |
+
"grad_norm": 0.12977555394172668,
|
| 5857 |
+
"learning_rate": 5.373581430496215e-05,
|
| 5858 |
+
"loss": 1.5655,
|
| 5859 |
+
"step": 815
|
| 5860 |
+
},
|
| 5861 |
+
{
|
| 5862 |
+
"epoch": 0.015640873282091582,
|
| 5863 |
+
"grad_norm": 0.14951717853546143,
|
| 5864 |
+
"learning_rate": 5.3454609787753874e-05,
|
| 5865 |
+
"loss": 1.2403,
|
| 5866 |
+
"step": 816
|
| 5867 |
+
},
|
| 5868 |
+
{
|
| 5869 |
+
"epoch": 0.01566004101895689,
|
| 5870 |
+
"grad_norm": 0.15313448011875153,
|
| 5871 |
+
"learning_rate": 5.3173271694445934e-05,
|
| 5872 |
+
"loss": 1.3189,
|
| 5873 |
+
"step": 817
|
| 5874 |
+
},
|
| 5875 |
+
{
|
| 5876 |
+
"epoch": 0.0156792087558222,
|
| 5877 |
+
"grad_norm": 0.15912052989006042,
|
| 5878 |
+
"learning_rate": 5.289180954728744e-05,
|
| 5879 |
+
"loss": 1.7737,
|
| 5880 |
+
"step": 818
|
| 5881 |
+
},
|
| 5882 |
+
{
|
| 5883 |
+
"epoch": 0.01569837649268751,
|
| 5884 |
+
"grad_norm": 0.11720007658004761,
|
| 5885 |
+
"learning_rate": 5.2610232872726344e-05,
|
| 5886 |
+
"loss": 1.5078,
|
| 5887 |
+
"step": 819
|
| 5888 |
+
},
|
| 5889 |
+
{
|
| 5890 |
+
"epoch": 0.015717544229552818,
|
| 5891 |
+
"grad_norm": 0.1603204905986786,
|
| 5892 |
+
"learning_rate": 5.232855120108686e-05,
|
| 5893 |
+
"loss": 1.5153,
|
| 5894 |
+
"step": 820
|
| 5895 |
+
},
|
| 5896 |
+
{
|
| 5897 |
+
"epoch": 0.015736711966418127,
|
| 5898 |
+
"grad_norm": 0.1426832377910614,
|
| 5899 |
+
"learning_rate": 5.204677406624701e-05,
|
| 5900 |
+
"loss": 1.3705,
|
| 5901 |
+
"step": 821
|
| 5902 |
+
},
|
| 5903 |
+
{
|
| 5904 |
+
"epoch": 0.015755879703283432,
|
| 5905 |
+
"grad_norm": 0.18909388780593872,
|
| 5906 |
+
"learning_rate": 5.176491100531584e-05,
|
| 5907 |
+
"loss": 1.5634,
|
| 5908 |
+
"step": 822
|
| 5909 |
+
},
|
| 5910 |
+
{
|
| 5911 |
+
"epoch": 0.01577504744014874,
|
| 5912 |
+
"grad_norm": 0.11742043495178223,
|
| 5913 |
+
"learning_rate": 5.148297155831072e-05,
|
| 5914 |
+
"loss": 1.7318,
|
| 5915 |
+
"step": 823
|
| 5916 |
+
},
|
| 5917 |
+
{
|
| 5918 |
+
"epoch": 0.01579421517701405,
|
| 5919 |
+
"grad_norm": 0.1602964997291565,
|
| 5920 |
+
"learning_rate": 5.1200965267834384e-05,
|
| 5921 |
+
"loss": 1.2726,
|
| 5922 |
+
"step": 824
|
| 5923 |
+
},
|
| 5924 |
+
{
|
| 5925 |
+
"epoch": 0.01581338291387936,
|
| 5926 |
+
"grad_norm": 0.12343866378068924,
|
| 5927 |
+
"learning_rate": 5.091890167875197e-05,
|
| 5928 |
+
"loss": 1.344,
|
| 5929 |
+
"step": 825
|
| 5930 |
+
},
|
| 5931 |
+
{
|
| 5932 |
+
"epoch": 0.01581338291387936,
|
| 5933 |
+
"eval_loss": 1.467738151550293,
|
| 5934 |
+
"eval_runtime": 2532.0079,
|
| 5935 |
+
"eval_samples_per_second": 36.63,
|
| 5936 |
+
"eval_steps_per_second": 18.315,
|
| 5937 |
+
"step": 825
|
| 5938 |
+
},
|
| 5939 |
+
{
|
| 5940 |
+
"epoch": 0.01581338291387936,
|
| 5941 |
+
"eval_loss": 1.5087077617645264,
|
| 5942 |
+
"eval_runtime": 2531.6284,
|
| 5943 |
+
"eval_samples_per_second": 36.636,
|
| 5944 |
+
"eval_steps_per_second": 18.318,
|
| 5945 |
+
"step": 825
|
| 5946 |
+
},
|
| 5947 |
+
{
|
| 5948 |
+
"epoch": 0.015832550650744667,
|
| 5949 |
+
"grad_norm": 0.1388055682182312,
|
| 5950 |
+
"learning_rate": 5.0636790337867974e-05,
|
| 5951 |
+
"loss": 1.8056,
|
| 5952 |
+
"step": 826
|
| 5953 |
+
},
|
| 5954 |
+
{
|
| 5955 |
+
"epoch": 0.015851718387609976,
|
| 5956 |
+
"grad_norm": 0.14199969172477722,
|
| 5957 |
+
"learning_rate": 5.035464079360311e-05,
|
| 5958 |
+
"loss": 1.4404,
|
| 5959 |
+
"step": 827
|
| 5960 |
+
},
|
| 5961 |
+
{
|
| 5962 |
+
"epoch": 0.01587088612447528,
|
| 5963 |
+
"grad_norm": 0.12653231620788574,
|
| 5964 |
+
"learning_rate": 5.007246259567111e-05,
|
| 5965 |
+
"loss": 1.5827,
|
| 5966 |
+
"step": 828
|
| 5967 |
+
},
|
| 5968 |
+
{
|
| 5969 |
+
"epoch": 0.01589005386134059,
|
| 5970 |
+
"grad_norm": 0.12284441292285919,
|
| 5971 |
+
"learning_rate": 4.979026529475555e-05,
|
| 5972 |
+
"loss": 1.1822,
|
| 5973 |
+
"step": 829
|
| 5974 |
+
},
|
| 5975 |
+
{
|
| 5976 |
+
"epoch": 0.0159092215982059,
|
| 5977 |
+
"grad_norm": 0.13183721899986267,
|
| 5978 |
+
"learning_rate": 4.9508058442186563e-05,
|
| 5979 |
+
"loss": 1.6911,
|
| 5980 |
+
"step": 830
|
| 5981 |
+
},
|
| 5982 |
+
{
|
| 5983 |
+
"epoch": 0.015928389335071208,
|
| 5984 |
+
"grad_norm": 0.17767372727394104,
|
| 5985 |
+
"learning_rate": 4.922585158961757e-05,
|
| 5986 |
+
"loss": 1.2546,
|
| 5987 |
+
"step": 831
|
| 5988 |
+
},
|
| 5989 |
+
{
|
| 5990 |
+
"epoch": 0.015947557071936517,
|
| 5991 |
+
"grad_norm": 0.11520732939243317,
|
| 5992 |
+
"learning_rate": 4.894365428870203e-05,
|
| 5993 |
+
"loss": 1.3013,
|
| 5994 |
+
"step": 832
|
| 5995 |
+
},
|
| 5996 |
+
{
|
| 5997 |
+
"epoch": 0.015966724808801826,
|
| 5998 |
+
"grad_norm": 0.1350078135728836,
|
| 5999 |
+
"learning_rate": 4.866147609077002e-05,
|
| 6000 |
+
"loss": 1.8165,
|
| 6001 |
+
"step": 833
|
| 6002 |
+
},
|
| 6003 |
+
{
|
| 6004 |
+
"epoch": 0.015985892545667135,
|
| 6005 |
+
"grad_norm": 0.12163300812244415,
|
| 6006 |
+
"learning_rate": 4.837932654650515e-05,
|
| 6007 |
+
"loss": 1.5469,
|
| 6008 |
+
"step": 834
|
| 6009 |
+
},
|
| 6010 |
+
{
|
| 6011 |
+
"epoch": 0.01600506028253244,
|
| 6012 |
+
"grad_norm": 0.13528017699718475,
|
| 6013 |
+
"learning_rate": 4.8097215205621156e-05,
|
| 6014 |
+
"loss": 1.0834,
|
| 6015 |
+
"step": 835
|
| 6016 |
+
},
|
| 6017 |
+
{
|
| 6018 |
+
"epoch": 0.01602422801939775,
|
| 6019 |
+
"grad_norm": 0.1524435132741928,
|
| 6020 |
+
"learning_rate": 4.7815151616538764e-05,
|
| 6021 |
+
"loss": 1.3997,
|
| 6022 |
+
"step": 836
|
| 6023 |
+
},
|
| 6024 |
+
{
|
| 6025 |
+
"epoch": 0.016043395756263058,
|
| 6026 |
+
"grad_norm": 0.10823335498571396,
|
| 6027 |
+
"learning_rate": 4.7533145326062416e-05,
|
| 6028 |
+
"loss": 1.592,
|
| 6029 |
+
"step": 837
|
| 6030 |
+
},
|
| 6031 |
+
{
|
| 6032 |
+
"epoch": 0.016062563493128366,
|
| 6033 |
+
"grad_norm": 0.09441324323415756,
|
| 6034 |
+
"learning_rate": 4.725120587905729e-05,
|
| 6035 |
+
"loss": 1.3389,
|
| 6036 |
+
"step": 838
|
| 6037 |
+
},
|
| 6038 |
+
{
|
| 6039 |
+
"epoch": 0.016081731229993675,
|
| 6040 |
+
"grad_norm": 0.12699145078659058,
|
| 6041 |
+
"learning_rate": 4.6969342818126124e-05,
|
| 6042 |
+
"loss": 1.5813,
|
| 6043 |
+
"step": 839
|
| 6044 |
+
},
|
| 6045 |
+
{
|
| 6046 |
+
"epoch": 0.016100898966858984,
|
| 6047 |
+
"grad_norm": 0.16241148114204407,
|
| 6048 |
+
"learning_rate": 4.668756568328627e-05,
|
| 6049 |
+
"loss": 1.5295,
|
| 6050 |
+
"step": 840
|
| 6051 |
+
},
|
| 6052 |
+
{
|
| 6053 |
+
"epoch": 0.016120066703724293,
|
| 6054 |
+
"grad_norm": 0.14051668345928192,
|
| 6055 |
+
"learning_rate": 4.640588401164678e-05,
|
| 6056 |
+
"loss": 1.6318,
|
| 6057 |
+
"step": 841
|
| 6058 |
+
},
|
| 6059 |
+
{
|
| 6060 |
+
"epoch": 0.0161392344405896,
|
| 6061 |
+
"grad_norm": 0.10982175916433334,
|
| 6062 |
+
"learning_rate": 4.612430733708568e-05,
|
| 6063 |
+
"loss": 1.1175,
|
| 6064 |
+
"step": 842
|
| 6065 |
+
},
|
| 6066 |
+
{
|
| 6067 |
+
"epoch": 0.016158402177454907,
|
| 6068 |
+
"grad_norm": 0.11348648369312286,
|
| 6069 |
+
"learning_rate": 4.5842845189927214e-05,
|
| 6070 |
+
"loss": 1.4904,
|
| 6071 |
+
"step": 843
|
| 6072 |
+
},
|
| 6073 |
+
{
|
| 6074 |
+
"epoch": 0.016177569914320216,
|
| 6075 |
+
"grad_norm": 0.1736602485179901,
|
| 6076 |
+
"learning_rate": 4.556150709661925e-05,
|
| 6077 |
+
"loss": 1.5621,
|
| 6078 |
+
"step": 844
|
| 6079 |
+
},
|
| 6080 |
+
{
|
| 6081 |
+
"epoch": 0.016196737651185525,
|
| 6082 |
+
"grad_norm": 0.153645321726799,
|
| 6083 |
+
"learning_rate": 4.5280302579410976e-05,
|
| 6084 |
+
"loss": 1.6168,
|
| 6085 |
+
"step": 845
|
| 6086 |
+
},
|
| 6087 |
+
{
|
| 6088 |
+
"epoch": 0.016215905388050834,
|
| 6089 |
+
"grad_norm": 0.12852360308170319,
|
| 6090 |
+
"learning_rate": 4.4999241156030496e-05,
|
| 6091 |
+
"loss": 1.0716,
|
| 6092 |
+
"step": 846
|
| 6093 |
+
},
|
| 6094 |
+
{
|
| 6095 |
+
"epoch": 0.016235073124916143,
|
| 6096 |
+
"grad_norm": 0.10657412558794022,
|
| 6097 |
+
"learning_rate": 4.471833233936268e-05,
|
| 6098 |
+
"loss": 1.876,
|
| 6099 |
+
"step": 847
|
| 6100 |
+
},
|
| 6101 |
+
{
|
| 6102 |
+
"epoch": 0.016254240861781448,
|
| 6103 |
+
"grad_norm": 0.15090149641036987,
|
| 6104 |
+
"learning_rate": 4.443758563712728e-05,
|
| 6105 |
+
"loss": 1.3017,
|
| 6106 |
+
"step": 848
|
| 6107 |
+
},
|
| 6108 |
+
{
|
| 6109 |
+
"epoch": 0.016273408598646757,
|
| 6110 |
+
"grad_norm": 0.18283763527870178,
|
| 6111 |
+
"learning_rate": 4.415701055155704e-05,
|
| 6112 |
+
"loss": 1.8782,
|
| 6113 |
+
"step": 849
|
| 6114 |
+
},
|
| 6115 |
+
{
|
| 6116 |
+
"epoch": 0.016292576335512066,
|
| 6117 |
+
"grad_norm": 0.12587769329547882,
|
| 6118 |
+
"learning_rate": 4.3876616579076144e-05,
|
| 6119 |
+
"loss": 1.647,
|
| 6120 |
+
"step": 850
|
| 6121 |
+
},
|
| 6122 |
+
{
|
| 6123 |
+
"epoch": 0.016311744072377374,
|
| 6124 |
+
"grad_norm": 0.19330890476703644,
|
| 6125 |
+
"learning_rate": 4.3596413209978774e-05,
|
| 6126 |
+
"loss": 1.8159,
|
| 6127 |
+
"step": 851
|
| 6128 |
+
},
|
| 6129 |
+
{
|
| 6130 |
+
"epoch": 0.016330911809242683,
|
| 6131 |
+
"grad_norm": 0.16183598339557648,
|
| 6132 |
+
"learning_rate": 4.331640992810784e-05,
|
| 6133 |
+
"loss": 1.2802,
|
| 6134 |
+
"step": 852
|
| 6135 |
+
},
|
| 6136 |
+
{
|
| 6137 |
+
"epoch": 0.016350079546107992,
|
| 6138 |
+
"grad_norm": 0.1407604068517685,
|
| 6139 |
+
"learning_rate": 4.303661621053412e-05,
|
| 6140 |
+
"loss": 1.1006,
|
| 6141 |
+
"step": 853
|
| 6142 |
+
},
|
| 6143 |
+
{
|
| 6144 |
+
"epoch": 0.0163692472829733,
|
| 6145 |
+
"grad_norm": 0.15612633526325226,
|
| 6146 |
+
"learning_rate": 4.275704152723539e-05,
|
| 6147 |
+
"loss": 1.4314,
|
| 6148 |
+
"step": 854
|
| 6149 |
+
},
|
| 6150 |
+
{
|
| 6151 |
+
"epoch": 0.016388415019838606,
|
| 6152 |
+
"grad_norm": 0.1078529953956604,
|
| 6153 |
+
"learning_rate": 4.2477695340775884e-05,
|
| 6154 |
+
"loss": 1.8136,
|
| 6155 |
+
"step": 855
|
| 6156 |
+
},
|
| 6157 |
+
{
|
| 6158 |
+
"epoch": 0.016407582756703915,
|
| 6159 |
+
"grad_norm": 0.1912553310394287,
|
| 6160 |
+
"learning_rate": 4.2198587105986126e-05,
|
| 6161 |
+
"loss": 1.303,
|
| 6162 |
+
"step": 856
|
| 6163 |
+
},
|
| 6164 |
+
{
|
| 6165 |
+
"epoch": 0.016426750493569224,
|
| 6166 |
+
"grad_norm": 0.10706888884305954,
|
| 6167 |
+
"learning_rate": 4.1919726269642844e-05,
|
| 6168 |
+
"loss": 1.119,
|
| 6169 |
+
"step": 857
|
| 6170 |
+
},
|
| 6171 |
+
{
|
| 6172 |
+
"epoch": 0.016445918230434533,
|
| 6173 |
+
"grad_norm": 0.13584940135478973,
|
| 6174 |
+
"learning_rate": 4.164112227014926e-05,
|
| 6175 |
+
"loss": 1.1873,
|
| 6176 |
+
"step": 858
|
| 6177 |
+
},
|
| 6178 |
+
{
|
| 6179 |
+
"epoch": 0.016465085967299842,
|
| 6180 |
+
"grad_norm": 0.12513437867164612,
|
| 6181 |
+
"learning_rate": 4.136278453721558e-05,
|
| 6182 |
+
"loss": 1.1956,
|
| 6183 |
+
"step": 859
|
| 6184 |
+
},
|
| 6185 |
+
{
|
| 6186 |
+
"epoch": 0.01648425370416515,
|
| 6187 |
+
"grad_norm": 0.14521580934524536,
|
| 6188 |
+
"learning_rate": 4.108472249153991e-05,
|
| 6189 |
+
"loss": 1.3369,
|
| 6190 |
+
"step": 860
|
| 6191 |
+
},
|
| 6192 |
+
{
|
| 6193 |
+
"epoch": 0.016503421441030456,
|
| 6194 |
+
"grad_norm": 0.12698303163051605,
|
| 6195 |
+
"learning_rate": 4.080694554448936e-05,
|
| 6196 |
+
"loss": 1.6408,
|
| 6197 |
+
"step": 861
|
| 6198 |
+
},
|
| 6199 |
+
{
|
| 6200 |
+
"epoch": 0.016522589177895765,
|
| 6201 |
+
"grad_norm": 0.10518490523099899,
|
| 6202 |
+
"learning_rate": 4.052946309778153e-05,
|
| 6203 |
+
"loss": 1.6527,
|
| 6204 |
+
"step": 862
|
| 6205 |
+
},
|
| 6206 |
+
{
|
| 6207 |
+
"epoch": 0.016541756914761074,
|
| 6208 |
+
"grad_norm": 0.11144135892391205,
|
| 6209 |
+
"learning_rate": 4.025228454316622e-05,
|
| 6210 |
+
"loss": 1.451,
|
| 6211 |
+
"step": 863
|
| 6212 |
+
},
|
| 6213 |
+
{
|
| 6214 |
+
"epoch": 0.016560924651626382,
|
| 6215 |
+
"grad_norm": 0.1137765496969223,
|
| 6216 |
+
"learning_rate": 3.997541926210768e-05,
|
| 6217 |
+
"loss": 1.1639,
|
| 6218 |
+
"step": 864
|
| 6219 |
+
},
|
| 6220 |
+
{
|
| 6221 |
+
"epoch": 0.01658009238849169,
|
| 6222 |
+
"grad_norm": 0.11371007561683655,
|
| 6223 |
+
"learning_rate": 3.9698876625466985e-05,
|
| 6224 |
+
"loss": 1.5035,
|
| 6225 |
+
"step": 865
|
| 6226 |
+
},
|
| 6227 |
+
{
|
| 6228 |
+
"epoch": 0.016599260125357,
|
| 6229 |
+
"grad_norm": 0.17619284987449646,
|
| 6230 |
+
"learning_rate": 3.942266599318488e-05,
|
| 6231 |
+
"loss": 1.4435,
|
| 6232 |
+
"step": 866
|
| 6233 |
+
},
|
| 6234 |
+
{
|
| 6235 |
+
"epoch": 0.01661842786222231,
|
| 6236 |
+
"grad_norm": 0.1566302627325058,
|
| 6237 |
+
"learning_rate": 3.9146796713965005e-05,
|
| 6238 |
+
"loss": 1.4801,
|
| 6239 |
+
"step": 867
|
| 6240 |
+
},
|
| 6241 |
+
{
|
| 6242 |
+
"epoch": 0.016637595599087614,
|
| 6243 |
+
"grad_norm": 0.17392432689666748,
|
| 6244 |
+
"learning_rate": 3.887127812495748e-05,
|
| 6245 |
+
"loss": 1.2481,
|
| 6246 |
+
"step": 868
|
| 6247 |
+
},
|
| 6248 |
+
{
|
| 6249 |
+
"epoch": 0.016656763335952923,
|
| 6250 |
+
"grad_norm": 0.1386740505695343,
|
| 6251 |
+
"learning_rate": 3.859611955144288e-05,
|
| 6252 |
+
"loss": 1.4188,
|
| 6253 |
+
"step": 869
|
| 6254 |
+
},
|
| 6255 |
+
{
|
| 6256 |
+
"epoch": 0.016675931072818232,
|
| 6257 |
+
"grad_norm": 0.1529177874326706,
|
| 6258 |
+
"learning_rate": 3.8321330306516535e-05,
|
| 6259 |
+
"loss": 1.7615,
|
| 6260 |
+
"step": 870
|
| 6261 |
+
},
|
| 6262 |
+
{
|
| 6263 |
+
"epoch": 0.01669509880968354,
|
| 6264 |
+
"grad_norm": 0.13424637913703918,
|
| 6265 |
+
"learning_rate": 3.804691969077342e-05,
|
| 6266 |
+
"loss": 1.4151,
|
| 6267 |
+
"step": 871
|
| 6268 |
+
},
|
| 6269 |
+
{
|
| 6270 |
+
"epoch": 0.01671426654654885,
|
| 6271 |
+
"grad_norm": 0.11477918922901154,
|
| 6272 |
+
"learning_rate": 3.7772896991993334e-05,
|
| 6273 |
+
"loss": 1.3577,
|
| 6274 |
+
"step": 872
|
| 6275 |
+
},
|
| 6276 |
+
{
|
| 6277 |
+
"epoch": 0.01673343428341416,
|
| 6278 |
+
"grad_norm": 0.17588411271572113,
|
| 6279 |
+
"learning_rate": 3.749927148482647e-05,
|
| 6280 |
+
"loss": 1.551,
|
| 6281 |
+
"step": 873
|
| 6282 |
+
},
|
| 6283 |
+
{
|
| 6284 |
+
"epoch": 0.016752602020279464,
|
| 6285 |
+
"grad_norm": 0.1304953545331955,
|
| 6286 |
+
"learning_rate": 3.7226052430479594e-05,
|
| 6287 |
+
"loss": 1.2935,
|
| 6288 |
+
"step": 874
|
| 6289 |
+
},
|
| 6290 |
+
{
|
| 6291 |
+
"epoch": 0.016771769757144773,
|
| 6292 |
+
"grad_norm": 0.10543167591094971,
|
| 6293 |
+
"learning_rate": 3.695324907640256e-05,
|
| 6294 |
+
"loss": 1.2343,
|
| 6295 |
+
"step": 875
|
| 6296 |
+
},
|
| 6297 |
+
{
|
| 6298 |
+
"epoch": 0.01679093749401008,
|
| 6299 |
+
"grad_norm": 0.12932707369327545,
|
| 6300 |
+
"learning_rate": 3.668087065597534e-05,
|
| 6301 |
+
"loss": 1.8215,
|
| 6302 |
+
"step": 876
|
| 6303 |
+
},
|
| 6304 |
+
{
|
| 6305 |
+
"epoch": 0.01681010523087539,
|
| 6306 |
+
"grad_norm": 0.1890428364276886,
|
| 6307 |
+
"learning_rate": 3.640892638819539e-05,
|
| 6308 |
+
"loss": 1.9125,
|
| 6309 |
+
"step": 877
|
| 6310 |
+
},
|
| 6311 |
+
{
|
| 6312 |
+
"epoch": 0.0168292729677407,
|
| 6313 |
+
"grad_norm": 0.14469286799430847,
|
| 6314 |
+
"learning_rate": 3.61374254773658e-05,
|
| 6315 |
+
"loss": 1.521,
|
| 6316 |
+
"step": 878
|
| 6317 |
+
},
|
| 6318 |
+
{
|
| 6319 |
+
"epoch": 0.016848440704606008,
|
| 6320 |
+
"grad_norm": 0.12144976854324341,
|
| 6321 |
+
"learning_rate": 3.586637711278363e-05,
|
| 6322 |
+
"loss": 1.2526,
|
| 6323 |
+
"step": 879
|
| 6324 |
+
},
|
| 6325 |
+
{
|
| 6326 |
+
"epoch": 0.016867608441471317,
|
| 6327 |
+
"grad_norm": 0.2042318731546402,
|
| 6328 |
+
"learning_rate": 3.5595790468428974e-05,
|
| 6329 |
+
"loss": 1.351,
|
| 6330 |
+
"step": 880
|
| 6331 |
+
},
|
| 6332 |
+
{
|
| 6333 |
+
"epoch": 0.016886776178336622,
|
| 6334 |
+
"grad_norm": 0.15050017833709717,
|
| 6335 |
+
"learning_rate": 3.532567470265434e-05,
|
| 6336 |
+
"loss": 1.2813,
|
| 6337 |
+
"step": 881
|
| 6338 |
+
},
|
| 6339 |
+
{
|
| 6340 |
+
"epoch": 0.01690594391520193,
|
| 6341 |
+
"grad_norm": 0.1257251501083374,
|
| 6342 |
+
"learning_rate": 3.5056038957874806e-05,
|
| 6343 |
+
"loss": 1.4732,
|
| 6344 |
+
"step": 882
|
| 6345 |
+
},
|
| 6346 |
+
{
|
| 6347 |
+
"epoch": 0.01692511165206724,
|
| 6348 |
+
"grad_norm": 0.16458237171173096,
|
| 6349 |
+
"learning_rate": 3.478689236025853e-05,
|
| 6350 |
+
"loss": 1.169,
|
| 6351 |
+
"step": 883
|
| 6352 |
+
},
|
| 6353 |
+
{
|
| 6354 |
+
"epoch": 0.01694427938893255,
|
| 6355 |
+
"grad_norm": 0.1077469140291214,
|
| 6356 |
+
"learning_rate": 3.451824401941778e-05,
|
| 6357 |
+
"loss": 1.4696,
|
| 6358 |
+
"step": 884
|
| 6359 |
+
},
|
| 6360 |
+
{
|
| 6361 |
+
"epoch": 0.016963447125797858,
|
| 6362 |
+
"grad_norm": 0.18079222738742828,
|
| 6363 |
+
"learning_rate": 3.425010302810077e-05,
|
| 6364 |
+
"loss": 1.1604,
|
| 6365 |
+
"step": 885
|
| 6366 |
+
},
|
| 6367 |
+
{
|
| 6368 |
+
"epoch": 0.016982614862663167,
|
| 6369 |
+
"grad_norm": 0.1286560446023941,
|
| 6370 |
+
"learning_rate": 3.398247846188376e-05,
|
| 6371 |
+
"loss": 1.4951,
|
| 6372 |
+
"step": 886
|
| 6373 |
+
},
|
| 6374 |
+
{
|
| 6375 |
+
"epoch": 0.017001782599528472,
|
| 6376 |
+
"grad_norm": 0.13332948088645935,
|
| 6377 |
+
"learning_rate": 3.3715379378863995e-05,
|
| 6378 |
+
"loss": 1.18,
|
| 6379 |
+
"step": 887
|
| 6380 |
+
},
|
| 6381 |
+
{
|
| 6382 |
+
"epoch": 0.01702095033639378,
|
| 6383 |
+
"grad_norm": 0.1720653772354126,
|
| 6384 |
+
"learning_rate": 3.344881481935299e-05,
|
| 6385 |
+
"loss": 1.606,
|
| 6386 |
+
"step": 888
|
| 6387 |
+
},
|
| 6388 |
+
{
|
| 6389 |
+
"epoch": 0.01704011807325909,
|
| 6390 |
+
"grad_norm": 0.17775127291679382,
|
| 6391 |
+
"learning_rate": 3.3182793805570675e-05,
|
| 6392 |
+
"loss": 2.1922,
|
| 6393 |
+
"step": 889
|
| 6394 |
+
},
|
| 6395 |
+
{
|
| 6396 |
+
"epoch": 0.0170592858101244,
|
| 6397 |
+
"grad_norm": 0.1610146164894104,
|
| 6398 |
+
"learning_rate": 3.291732534134e-05,
|
| 6399 |
+
"loss": 1.4212,
|
| 6400 |
+
"step": 890
|
| 6401 |
+
},
|
| 6402 |
+
{
|
| 6403 |
+
"epoch": 0.017078453546989707,
|
| 6404 |
+
"grad_norm": 3.0537126064300537,
|
| 6405 |
+
"learning_rate": 3.26524184117821e-05,
|
| 6406 |
+
"loss": 2.2348,
|
| 6407 |
+
"step": 891
|
| 6408 |
+
},
|
| 6409 |
+
{
|
| 6410 |
+
"epoch": 0.017097621283855016,
|
| 6411 |
+
"grad_norm": 0.1203082874417305,
|
| 6412 |
+
"learning_rate": 3.2388081983012314e-05,
|
| 6413 |
+
"loss": 1.2908,
|
| 6414 |
+
"step": 892
|
| 6415 |
+
},
|
| 6416 |
+
{
|
| 6417 |
+
"epoch": 0.017116789020720325,
|
| 6418 |
+
"grad_norm": 0.13465985655784607,
|
| 6419 |
+
"learning_rate": 3.212432500183657e-05,
|
| 6420 |
+
"loss": 1.128,
|
| 6421 |
+
"step": 893
|
| 6422 |
+
},
|
| 6423 |
+
{
|
| 6424 |
+
"epoch": 0.01713595675758563,
|
| 6425 |
+
"grad_norm": 0.1260255128145218,
|
| 6426 |
+
"learning_rate": 3.1861156395448755e-05,
|
| 6427 |
+
"loss": 1.7501,
|
| 6428 |
+
"step": 894
|
| 6429 |
+
},
|
| 6430 |
+
{
|
| 6431 |
+
"epoch": 0.01715512449445094,
|
| 6432 |
+
"grad_norm": 0.19205020368099213,
|
| 6433 |
+
"learning_rate": 3.159858507112833e-05,
|
| 6434 |
+
"loss": 1.4086,
|
| 6435 |
+
"step": 895
|
| 6436 |
+
},
|
| 6437 |
+
{
|
| 6438 |
+
"epoch": 0.017174292231316248,
|
| 6439 |
+
"grad_norm": 0.11290293186903,
|
| 6440 |
+
"learning_rate": 3.1336619915939084e-05,
|
| 6441 |
+
"loss": 1.4479,
|
| 6442 |
+
"step": 896
|
| 6443 |
+
},
|
| 6444 |
+
{
|
| 6445 |
+
"epoch": 0.017193459968181557,
|
| 6446 |
+
"grad_norm": 0.12862160801887512,
|
| 6447 |
+
"learning_rate": 3.107526979642819e-05,
|
| 6448 |
+
"loss": 1.3166,
|
| 6449 |
+
"step": 897
|
| 6450 |
+
},
|
| 6451 |
+
{
|
| 6452 |
+
"epoch": 0.017212627705046866,
|
| 6453 |
+
"grad_norm": 0.12488026171922684,
|
| 6454 |
+
"learning_rate": 3.0814543558326143e-05,
|
| 6455 |
+
"loss": 1.2115,
|
| 6456 |
+
"step": 898
|
| 6457 |
+
},
|
| 6458 |
+
{
|
| 6459 |
+
"epoch": 0.017231795441912175,
|
| 6460 |
+
"grad_norm": 0.15966151654720306,
|
| 6461 |
+
"learning_rate": 3.055445002624737e-05,
|
| 6462 |
+
"loss": 1.5335,
|
| 6463 |
+
"step": 899
|
| 6464 |
+
},
|
| 6465 |
+
{
|
| 6466 |
+
"epoch": 0.01725096317877748,
|
| 6467 |
+
"grad_norm": 0.12605780363082886,
|
| 6468 |
+
"learning_rate": 3.0294998003391566e-05,
|
| 6469 |
+
"loss": 1.5669,
|
| 6470 |
+
"step": 900
|
| 6471 |
+
},
|
| 6472 |
+
{
|
| 6473 |
+
"epoch": 0.01725096317877748,
|
| 6474 |
+
"eval_loss": 1.46342933177948,
|
| 6475 |
+
"eval_runtime": 2529.5354,
|
| 6476 |
+
"eval_samples_per_second": 36.666,
|
| 6477 |
+
"eval_steps_per_second": 18.333,
|
| 6478 |
+
"step": 900
|
| 6479 |
+
},
|
| 6480 |
+
{
|
| 6481 |
+
"epoch": 0.01725096317877748,
|
| 6482 |
+
"eval_loss": 1.5044798851013184,
|
| 6483 |
+
"eval_runtime": 2568.8275,
|
| 6484 |
+
"eval_samples_per_second": 36.105,
|
| 6485 |
+
"eval_steps_per_second": 18.053,
|
| 6486 |
+
"step": 900
|
| 6487 |
}
|
| 6488 |
],
|
| 6489 |
"logging_steps": 1,
|
|
|
|
| 6498 |
"should_evaluate": false,
|
| 6499 |
"should_log": false,
|
| 6500 |
"should_save": true,
|
| 6501 |
+
"should_training_stop": true
|
| 6502 |
},
|
| 6503 |
"attributes": {}
|
| 6504 |
}
|
| 6505 |
},
|
| 6506 |
+
"total_flos": 1.1870280980727398e+17,
|
| 6507 |
"train_batch_size": 4,
|
| 6508 |
"trial_name": null,
|
| 6509 |
"trial_params": null
|