Training in progress, step 9000, checkpoint
Browse files
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1856040378
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f3168371b4d34bcda066e6f74a82bfec8b457ef9d00edf98c5480d2293d8f7e3
|
| 3 |
size 1856040378
|
last-checkpoint/pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 928000378
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0fd0ab7b3c558e973628879af87172e59cc108ce97049f2ac5663634d4e2a338
|
| 3 |
size 928000378
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a3816ae240b6071ab7e92dda2e725d12e18bf87285ee93c87739a1a68be57ea7
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1000
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bff2269a00d40d32d61117bbc42ac1ebc05e60fa114002712151b20a202ac481
|
| 3 |
size 1000
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 1.
|
| 3 |
-
"best_model_checkpoint": "model/chessformer-3/checkpoint-
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -28135,6 +28135,3522 @@
|
|
| 28135 |
"eval_samples_per_second": 559.097,
|
| 28136 |
"eval_steps_per_second": 69.907,
|
| 28137 |
"step": 8000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 28138 |
}
|
| 28139 |
],
|
| 28140 |
"logging_steps": 2,
|
|
@@ -28154,7 +31670,7 @@
|
|
| 28154 |
"attributes": {}
|
| 28155 |
}
|
| 28156 |
},
|
| 28157 |
-
"total_flos": 1.
|
| 28158 |
"train_batch_size": 768,
|
| 28159 |
"trial_name": null,
|
| 28160 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 1.51895010471344,
|
| 3 |
+
"best_model_checkpoint": "model/chessformer-3/checkpoint-9000",
|
| 4 |
+
"epoch": 0.4001600640256102,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 9000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 28135 |
"eval_samples_per_second": 559.097,
|
| 28136 |
"eval_steps_per_second": 69.907,
|
| 28137 |
"step": 8000
|
| 28138 |
+
},
|
| 28139 |
+
{
|
| 28140 |
+
"epoch": 0.3557867591481037,
|
| 28141 |
+
"grad_norm": 0.12552303075790405,
|
| 28142 |
+
"learning_rate": 0.0009847927295283491,
|
| 28143 |
+
"loss": 1.7432,
|
| 28144 |
+
"step": 8002
|
| 28145 |
+
},
|
| 28146 |
+
{
|
| 28147 |
+
"epoch": 0.355875683606776,
|
| 28148 |
+
"grad_norm": 0.12398079037666321,
|
| 28149 |
+
"learning_rate": 0.0009847840852398846,
|
| 28150 |
+
"loss": 1.7169,
|
| 28151 |
+
"step": 8004
|
| 28152 |
+
},
|
| 28153 |
+
{
|
| 28154 |
+
"epoch": 0.3559646080654484,
|
| 28155 |
+
"grad_norm": 0.11141873896121979,
|
| 28156 |
+
"learning_rate": 0.0009847754385332369,
|
| 28157 |
+
"loss": 1.6993,
|
| 28158 |
+
"step": 8006
|
| 28159 |
+
},
|
| 28160 |
+
{
|
| 28161 |
+
"epoch": 0.35605353252412075,
|
| 28162 |
+
"grad_norm": 0.1098860651254654,
|
| 28163 |
+
"learning_rate": 0.0009847667894084485,
|
| 28164 |
+
"loss": 1.6952,
|
| 28165 |
+
"step": 8008
|
| 28166 |
+
},
|
| 28167 |
+
{
|
| 28168 |
+
"epoch": 0.3561424569827931,
|
| 28169 |
+
"grad_norm": 0.10826461762189865,
|
| 28170 |
+
"learning_rate": 0.000984758137865563,
|
| 28171 |
+
"loss": 1.6817,
|
| 28172 |
+
"step": 8010
|
| 28173 |
+
},
|
| 28174 |
+
{
|
| 28175 |
+
"epoch": 0.3562313814414655,
|
| 28176 |
+
"grad_norm": 0.10351142287254333,
|
| 28177 |
+
"learning_rate": 0.0009847494839046234,
|
| 28178 |
+
"loss": 1.6737,
|
| 28179 |
+
"step": 8012
|
| 28180 |
+
},
|
| 28181 |
+
{
|
| 28182 |
+
"epoch": 0.35632030590013786,
|
| 28183 |
+
"grad_norm": 0.09795521944761276,
|
| 28184 |
+
"learning_rate": 0.000984740827525673,
|
| 28185 |
+
"loss": 1.6679,
|
| 28186 |
+
"step": 8014
|
| 28187 |
+
},
|
| 28188 |
+
{
|
| 28189 |
+
"epoch": 0.35640923035881017,
|
| 28190 |
+
"grad_norm": 0.0955873504281044,
|
| 28191 |
+
"learning_rate": 0.000984732168728755,
|
| 28192 |
+
"loss": 1.6526,
|
| 28193 |
+
"step": 8016
|
| 28194 |
+
},
|
| 28195 |
+
{
|
| 28196 |
+
"epoch": 0.35649815481748254,
|
| 28197 |
+
"grad_norm": 0.09694499522447586,
|
| 28198 |
+
"learning_rate": 0.0009847235075139123,
|
| 28199 |
+
"loss": 1.6522,
|
| 28200 |
+
"step": 8018
|
| 28201 |
+
},
|
| 28202 |
+
{
|
| 28203 |
+
"epoch": 0.3565870792761549,
|
| 28204 |
+
"grad_norm": 0.09016109257936478,
|
| 28205 |
+
"learning_rate": 0.0009847148438811883,
|
| 28206 |
+
"loss": 1.6439,
|
| 28207 |
+
"step": 8020
|
| 28208 |
+
},
|
| 28209 |
+
{
|
| 28210 |
+
"epoch": 0.3566760037348273,
|
| 28211 |
+
"grad_norm": 0.08890122175216675,
|
| 28212 |
+
"learning_rate": 0.0009847061778306262,
|
| 28213 |
+
"loss": 1.6345,
|
| 28214 |
+
"step": 8022
|
| 28215 |
+
},
|
| 28216 |
+
{
|
| 28217 |
+
"epoch": 0.35676492819349964,
|
| 28218 |
+
"grad_norm": 0.09018497169017792,
|
| 28219 |
+
"learning_rate": 0.0009846975093622692,
|
| 28220 |
+
"loss": 1.6403,
|
| 28221 |
+
"step": 8024
|
| 28222 |
+
},
|
| 28223 |
+
{
|
| 28224 |
+
"epoch": 0.35685385265217195,
|
| 28225 |
+
"grad_norm": 0.0861065536737442,
|
| 28226 |
+
"learning_rate": 0.0009846888384761607,
|
| 28227 |
+
"loss": 1.6304,
|
| 28228 |
+
"step": 8026
|
| 28229 |
+
},
|
| 28230 |
+
{
|
| 28231 |
+
"epoch": 0.3569427771108443,
|
| 28232 |
+
"grad_norm": 0.08618265390396118,
|
| 28233 |
+
"learning_rate": 0.0009846801651723436,
|
| 28234 |
+
"loss": 1.6258,
|
| 28235 |
+
"step": 8028
|
| 28236 |
+
},
|
| 28237 |
+
{
|
| 28238 |
+
"epoch": 0.3570317015695167,
|
| 28239 |
+
"grad_norm": 0.0936831384897232,
|
| 28240 |
+
"learning_rate": 0.0009846714894508617,
|
| 28241 |
+
"loss": 1.6244,
|
| 28242 |
+
"step": 8030
|
| 28243 |
+
},
|
| 28244 |
+
{
|
| 28245 |
+
"epoch": 0.35712062602818906,
|
| 28246 |
+
"grad_norm": 0.08654095232486725,
|
| 28247 |
+
"learning_rate": 0.0009846628113117577,
|
| 28248 |
+
"loss": 1.6206,
|
| 28249 |
+
"step": 8032
|
| 28250 |
+
},
|
| 28251 |
+
{
|
| 28252 |
+
"epoch": 0.3572095504868614,
|
| 28253 |
+
"grad_norm": 0.09224793314933777,
|
| 28254 |
+
"learning_rate": 0.0009846541307550753,
|
| 28255 |
+
"loss": 1.6189,
|
| 28256 |
+
"step": 8034
|
| 28257 |
+
},
|
| 28258 |
+
{
|
| 28259 |
+
"epoch": 0.3572984749455338,
|
| 28260 |
+
"grad_norm": 0.08937020599842072,
|
| 28261 |
+
"learning_rate": 0.0009846454477808575,
|
| 28262 |
+
"loss": 1.6198,
|
| 28263 |
+
"step": 8036
|
| 28264 |
+
},
|
| 28265 |
+
{
|
| 28266 |
+
"epoch": 0.3573873994042061,
|
| 28267 |
+
"grad_norm": 0.09124995023012161,
|
| 28268 |
+
"learning_rate": 0.0009846367623891478,
|
| 28269 |
+
"loss": 1.6108,
|
| 28270 |
+
"step": 8038
|
| 28271 |
+
},
|
| 28272 |
+
{
|
| 28273 |
+
"epoch": 0.3574763238628785,
|
| 28274 |
+
"grad_norm": 0.08880987763404846,
|
| 28275 |
+
"learning_rate": 0.0009846280745799895,
|
| 28276 |
+
"loss": 1.6126,
|
| 28277 |
+
"step": 8040
|
| 28278 |
+
},
|
| 28279 |
+
{
|
| 28280 |
+
"epoch": 0.35756524832155084,
|
| 28281 |
+
"grad_norm": 0.0875345841050148,
|
| 28282 |
+
"learning_rate": 0.000984619384353426,
|
| 28283 |
+
"loss": 1.6134,
|
| 28284 |
+
"step": 8042
|
| 28285 |
+
},
|
| 28286 |
+
{
|
| 28287 |
+
"epoch": 0.3576541727802232,
|
| 28288 |
+
"grad_norm": 0.08219029754400253,
|
| 28289 |
+
"learning_rate": 0.0009846106917095006,
|
| 28290 |
+
"loss": 1.6066,
|
| 28291 |
+
"step": 8044
|
| 28292 |
+
},
|
| 28293 |
+
{
|
| 28294 |
+
"epoch": 0.3577430972388956,
|
| 28295 |
+
"grad_norm": 0.08467015624046326,
|
| 28296 |
+
"learning_rate": 0.0009846019966482564,
|
| 28297 |
+
"loss": 1.6128,
|
| 28298 |
+
"step": 8046
|
| 28299 |
+
},
|
| 28300 |
+
{
|
| 28301 |
+
"epoch": 0.3578320216975679,
|
| 28302 |
+
"grad_norm": 0.08793745934963226,
|
| 28303 |
+
"learning_rate": 0.0009845932991697373,
|
| 28304 |
+
"loss": 1.6093,
|
| 28305 |
+
"step": 8048
|
| 28306 |
+
},
|
| 28307 |
+
{
|
| 28308 |
+
"epoch": 0.35792094615624026,
|
| 28309 |
+
"grad_norm": 0.08105975389480591,
|
| 28310 |
+
"learning_rate": 0.000984584599273986,
|
| 28311 |
+
"loss": 1.602,
|
| 28312 |
+
"step": 8050
|
| 28313 |
+
},
|
| 28314 |
+
{
|
| 28315 |
+
"epoch": 0.35800987061491263,
|
| 28316 |
+
"grad_norm": 0.08754699677228928,
|
| 28317 |
+
"learning_rate": 0.0009845758969610466,
|
| 28318 |
+
"loss": 1.596,
|
| 28319 |
+
"step": 8052
|
| 28320 |
+
},
|
| 28321 |
+
{
|
| 28322 |
+
"epoch": 0.358098795073585,
|
| 28323 |
+
"grad_norm": 0.08423270285129547,
|
| 28324 |
+
"learning_rate": 0.000984567192230962,
|
| 28325 |
+
"loss": 1.5994,
|
| 28326 |
+
"step": 8054
|
| 28327 |
+
},
|
| 28328 |
+
{
|
| 28329 |
+
"epoch": 0.35818771953225736,
|
| 28330 |
+
"grad_norm": 0.08522899448871613,
|
| 28331 |
+
"learning_rate": 0.000984558485083776,
|
| 28332 |
+
"loss": 1.6003,
|
| 28333 |
+
"step": 8056
|
| 28334 |
+
},
|
| 28335 |
+
{
|
| 28336 |
+
"epoch": 0.35827664399092973,
|
| 28337 |
+
"grad_norm": 0.08278319239616394,
|
| 28338 |
+
"learning_rate": 0.0009845497755195318,
|
| 28339 |
+
"loss": 1.6026,
|
| 28340 |
+
"step": 8058
|
| 28341 |
+
},
|
| 28342 |
+
{
|
| 28343 |
+
"epoch": 0.35836556844960205,
|
| 28344 |
+
"grad_norm": 0.08185625076293945,
|
| 28345 |
+
"learning_rate": 0.0009845410635382728,
|
| 28346 |
+
"loss": 1.5918,
|
| 28347 |
+
"step": 8060
|
| 28348 |
+
},
|
| 28349 |
+
{
|
| 28350 |
+
"epoch": 0.3584544929082744,
|
| 28351 |
+
"grad_norm": 0.08351411670446396,
|
| 28352 |
+
"learning_rate": 0.0009845323491400426,
|
| 28353 |
+
"loss": 1.6026,
|
| 28354 |
+
"step": 8062
|
| 28355 |
+
},
|
| 28356 |
+
{
|
| 28357 |
+
"epoch": 0.3585434173669468,
|
| 28358 |
+
"grad_norm": 0.08093686401844025,
|
| 28359 |
+
"learning_rate": 0.0009845236323248845,
|
| 28360 |
+
"loss": 1.6011,
|
| 28361 |
+
"step": 8064
|
| 28362 |
+
},
|
| 28363 |
+
{
|
| 28364 |
+
"epoch": 0.35863234182561915,
|
| 28365 |
+
"grad_norm": 0.08070489019155502,
|
| 28366 |
+
"learning_rate": 0.0009845149130928421,
|
| 28367 |
+
"loss": 1.5963,
|
| 28368 |
+
"step": 8066
|
| 28369 |
+
},
|
| 28370 |
+
{
|
| 28371 |
+
"epoch": 0.3587212662842915,
|
| 28372 |
+
"grad_norm": 0.08184243738651276,
|
| 28373 |
+
"learning_rate": 0.0009845061914439592,
|
| 28374 |
+
"loss": 1.5887,
|
| 28375 |
+
"step": 8068
|
| 28376 |
+
},
|
| 28377 |
+
{
|
| 28378 |
+
"epoch": 0.35881019074296383,
|
| 28379 |
+
"grad_norm": 0.08118681609630585,
|
| 28380 |
+
"learning_rate": 0.0009844974673782786,
|
| 28381 |
+
"loss": 1.5864,
|
| 28382 |
+
"step": 8070
|
| 28383 |
+
},
|
| 28384 |
+
{
|
| 28385 |
+
"epoch": 0.3588991152016362,
|
| 28386 |
+
"grad_norm": 0.08214083313941956,
|
| 28387 |
+
"learning_rate": 0.0009844887408958444,
|
| 28388 |
+
"loss": 1.5933,
|
| 28389 |
+
"step": 8072
|
| 28390 |
+
},
|
| 28391 |
+
{
|
| 28392 |
+
"epoch": 0.35898803966030857,
|
| 28393 |
+
"grad_norm": 0.07901991903781891,
|
| 28394 |
+
"learning_rate": 0.0009844800119967,
|
| 28395 |
+
"loss": 1.5918,
|
| 28396 |
+
"step": 8074
|
| 28397 |
+
},
|
| 28398 |
+
{
|
| 28399 |
+
"epoch": 0.35907696411898093,
|
| 28400 |
+
"grad_norm": 0.0798892080783844,
|
| 28401 |
+
"learning_rate": 0.000984471280680889,
|
| 28402 |
+
"loss": 1.5866,
|
| 28403 |
+
"step": 8076
|
| 28404 |
+
},
|
| 28405 |
+
{
|
| 28406 |
+
"epoch": 0.3591658885776533,
|
| 28407 |
+
"grad_norm": 0.08176968991756439,
|
| 28408 |
+
"learning_rate": 0.0009844625469484546,
|
| 28409 |
+
"loss": 1.5841,
|
| 28410 |
+
"step": 8078
|
| 28411 |
+
},
|
| 28412 |
+
{
|
| 28413 |
+
"epoch": 0.3592548130363256,
|
| 28414 |
+
"grad_norm": 0.0794360563158989,
|
| 28415 |
+
"learning_rate": 0.0009844538107994405,
|
| 28416 |
+
"loss": 1.589,
|
| 28417 |
+
"step": 8080
|
| 28418 |
+
},
|
| 28419 |
+
{
|
| 28420 |
+
"epoch": 0.359343737494998,
|
| 28421 |
+
"grad_norm": 0.08003082126379013,
|
| 28422 |
+
"learning_rate": 0.0009844450722338906,
|
| 28423 |
+
"loss": 1.5884,
|
| 28424 |
+
"step": 8082
|
| 28425 |
+
},
|
| 28426 |
+
{
|
| 28427 |
+
"epoch": 0.35943266195367035,
|
| 28428 |
+
"grad_norm": 0.08419719338417053,
|
| 28429 |
+
"learning_rate": 0.000984436331251848,
|
| 28430 |
+
"loss": 1.5887,
|
| 28431 |
+
"step": 8084
|
| 28432 |
+
},
|
| 28433 |
+
{
|
| 28434 |
+
"epoch": 0.3595215864123427,
|
| 28435 |
+
"grad_norm": 0.07977907359600067,
|
| 28436 |
+
"learning_rate": 0.0009844275878533569,
|
| 28437 |
+
"loss": 1.5834,
|
| 28438 |
+
"step": 8086
|
| 28439 |
+
},
|
| 28440 |
+
{
|
| 28441 |
+
"epoch": 0.3596105108710151,
|
| 28442 |
+
"grad_norm": 0.08345074206590652,
|
| 28443 |
+
"learning_rate": 0.0009844188420384604,
|
| 28444 |
+
"loss": 1.584,
|
| 28445 |
+
"step": 8088
|
| 28446 |
+
},
|
| 28447 |
+
{
|
| 28448 |
+
"epoch": 0.35969943532968746,
|
| 28449 |
+
"grad_norm": 0.08102136850357056,
|
| 28450 |
+
"learning_rate": 0.0009844100938072023,
|
| 28451 |
+
"loss": 1.5883,
|
| 28452 |
+
"step": 8090
|
| 28453 |
+
},
|
| 28454 |
+
{
|
| 28455 |
+
"epoch": 0.35978835978835977,
|
| 28456 |
+
"grad_norm": 0.08167117834091187,
|
| 28457 |
+
"learning_rate": 0.0009844013431596263,
|
| 28458 |
+
"loss": 1.5822,
|
| 28459 |
+
"step": 8092
|
| 28460 |
+
},
|
| 28461 |
+
{
|
| 28462 |
+
"epoch": 0.35987728424703214,
|
| 28463 |
+
"grad_norm": 0.0804857537150383,
|
| 28464 |
+
"learning_rate": 0.000984392590095776,
|
| 28465 |
+
"loss": 1.5835,
|
| 28466 |
+
"step": 8094
|
| 28467 |
+
},
|
| 28468 |
+
{
|
| 28469 |
+
"epoch": 0.3599662087057045,
|
| 28470 |
+
"grad_norm": 0.08060716092586517,
|
| 28471 |
+
"learning_rate": 0.000984383834615695,
|
| 28472 |
+
"loss": 1.5853,
|
| 28473 |
+
"step": 8096
|
| 28474 |
+
},
|
| 28475 |
+
{
|
| 28476 |
+
"epoch": 0.3600551331643769,
|
| 28477 |
+
"grad_norm": 0.08844839036464691,
|
| 28478 |
+
"learning_rate": 0.000984375076719427,
|
| 28479 |
+
"loss": 1.582,
|
| 28480 |
+
"step": 8098
|
| 28481 |
+
},
|
| 28482 |
+
{
|
| 28483 |
+
"epoch": 0.36014405762304924,
|
| 28484 |
+
"grad_norm": 0.07595451176166534,
|
| 28485 |
+
"learning_rate": 0.0009843663164070158,
|
| 28486 |
+
"loss": 1.5848,
|
| 28487 |
+
"step": 8100
|
| 28488 |
+
},
|
| 28489 |
+
{
|
| 28490 |
+
"epoch": 0.36023298208172155,
|
| 28491 |
+
"grad_norm": 0.08353257924318314,
|
| 28492 |
+
"learning_rate": 0.000984357553678505,
|
| 28493 |
+
"loss": 1.5796,
|
| 28494 |
+
"step": 8102
|
| 28495 |
+
},
|
| 28496 |
+
{
|
| 28497 |
+
"epoch": 0.3603219065403939,
|
| 28498 |
+
"grad_norm": 0.08055184781551361,
|
| 28499 |
+
"learning_rate": 0.0009843487885339384,
|
| 28500 |
+
"loss": 1.5814,
|
| 28501 |
+
"step": 8104
|
| 28502 |
+
},
|
| 28503 |
+
{
|
| 28504 |
+
"epoch": 0.3604108309990663,
|
| 28505 |
+
"grad_norm": 0.07868727296590805,
|
| 28506 |
+
"learning_rate": 0.0009843400209733593,
|
| 28507 |
+
"loss": 1.5861,
|
| 28508 |
+
"step": 8106
|
| 28509 |
+
},
|
| 28510 |
+
{
|
| 28511 |
+
"epoch": 0.36049975545773866,
|
| 28512 |
+
"grad_norm": 0.08025962114334106,
|
| 28513 |
+
"learning_rate": 0.000984331250996812,
|
| 28514 |
+
"loss": 1.5845,
|
| 28515 |
+
"step": 8108
|
| 28516 |
+
},
|
| 28517 |
+
{
|
| 28518 |
+
"epoch": 0.360588679916411,
|
| 28519 |
+
"grad_norm": 0.07819036394357681,
|
| 28520 |
+
"learning_rate": 0.0009843224786043401,
|
| 28521 |
+
"loss": 1.5807,
|
| 28522 |
+
"step": 8110
|
| 28523 |
+
},
|
| 28524 |
+
{
|
| 28525 |
+
"epoch": 0.3606776043750834,
|
| 28526 |
+
"grad_norm": 0.07715727388858795,
|
| 28527 |
+
"learning_rate": 0.0009843137037959873,
|
| 28528 |
+
"loss": 1.5842,
|
| 28529 |
+
"step": 8112
|
| 28530 |
+
},
|
| 28531 |
+
{
|
| 28532 |
+
"epoch": 0.3607665288337557,
|
| 28533 |
+
"grad_norm": 0.08035559952259064,
|
| 28534 |
+
"learning_rate": 0.0009843049265717975,
|
| 28535 |
+
"loss": 1.5746,
|
| 28536 |
+
"step": 8114
|
| 28537 |
+
},
|
| 28538 |
+
{
|
| 28539 |
+
"epoch": 0.3608554532924281,
|
| 28540 |
+
"grad_norm": 0.08042401075363159,
|
| 28541 |
+
"learning_rate": 0.000984296146931814,
|
| 28542 |
+
"loss": 1.5779,
|
| 28543 |
+
"step": 8116
|
| 28544 |
+
},
|
| 28545 |
+
{
|
| 28546 |
+
"epoch": 0.36094437775110044,
|
| 28547 |
+
"grad_norm": 0.0793127790093422,
|
| 28548 |
+
"learning_rate": 0.000984287364876081,
|
| 28549 |
+
"loss": 1.5806,
|
| 28550 |
+
"step": 8118
|
| 28551 |
+
},
|
| 28552 |
+
{
|
| 28553 |
+
"epoch": 0.3610333022097728,
|
| 28554 |
+
"grad_norm": 0.07924724370241165,
|
| 28555 |
+
"learning_rate": 0.0009842785804046424,
|
| 28556 |
+
"loss": 1.5812,
|
| 28557 |
+
"step": 8120
|
| 28558 |
+
},
|
| 28559 |
+
{
|
| 28560 |
+
"epoch": 0.3611222266684452,
|
| 28561 |
+
"grad_norm": 0.08291906118392944,
|
| 28562 |
+
"learning_rate": 0.0009842697935175417,
|
| 28563 |
+
"loss": 1.5734,
|
| 28564 |
+
"step": 8122
|
| 28565 |
+
},
|
| 28566 |
+
{
|
| 28567 |
+
"epoch": 0.3612111511271175,
|
| 28568 |
+
"grad_norm": 0.08052452653646469,
|
| 28569 |
+
"learning_rate": 0.0009842610042148229,
|
| 28570 |
+
"loss": 1.575,
|
| 28571 |
+
"step": 8124
|
| 28572 |
+
},
|
| 28573 |
+
{
|
| 28574 |
+
"epoch": 0.36130007558578986,
|
| 28575 |
+
"grad_norm": 0.07844741642475128,
|
| 28576 |
+
"learning_rate": 0.0009842522124965298,
|
| 28577 |
+
"loss": 1.5802,
|
| 28578 |
+
"step": 8126
|
| 28579 |
+
},
|
| 28580 |
+
{
|
| 28581 |
+
"epoch": 0.3613890000444622,
|
| 28582 |
+
"grad_norm": 0.08405445516109467,
|
| 28583 |
+
"learning_rate": 0.0009842434183627063,
|
| 28584 |
+
"loss": 1.571,
|
| 28585 |
+
"step": 8128
|
| 28586 |
+
},
|
| 28587 |
+
{
|
| 28588 |
+
"epoch": 0.3614779245031346,
|
| 28589 |
+
"grad_norm": 0.07999279350042343,
|
| 28590 |
+
"learning_rate": 0.0009842346218133964,
|
| 28591 |
+
"loss": 1.5703,
|
| 28592 |
+
"step": 8130
|
| 28593 |
+
},
|
| 28594 |
+
{
|
| 28595 |
+
"epoch": 0.36156684896180696,
|
| 28596 |
+
"grad_norm": 0.08422758430242538,
|
| 28597 |
+
"learning_rate": 0.0009842258228486436,
|
| 28598 |
+
"loss": 1.5774,
|
| 28599 |
+
"step": 8132
|
| 28600 |
+
},
|
| 28601 |
+
{
|
| 28602 |
+
"epoch": 0.3616557734204793,
|
| 28603 |
+
"grad_norm": 0.07723760604858398,
|
| 28604 |
+
"learning_rate": 0.0009842170214684923,
|
| 28605 |
+
"loss": 1.5756,
|
| 28606 |
+
"step": 8134
|
| 28607 |
+
},
|
| 28608 |
+
{
|
| 28609 |
+
"epoch": 0.36174469787915164,
|
| 28610 |
+
"grad_norm": 0.07770794630050659,
|
| 28611 |
+
"learning_rate": 0.0009842082176729858,
|
| 28612 |
+
"loss": 1.5674,
|
| 28613 |
+
"step": 8136
|
| 28614 |
+
},
|
| 28615 |
+
{
|
| 28616 |
+
"epoch": 0.361833622337824,
|
| 28617 |
+
"grad_norm": 0.07809870690107346,
|
| 28618 |
+
"learning_rate": 0.0009841994114621686,
|
| 28619 |
+
"loss": 1.5741,
|
| 28620 |
+
"step": 8138
|
| 28621 |
+
},
|
| 28622 |
+
{
|
| 28623 |
+
"epoch": 0.3619225467964964,
|
| 28624 |
+
"grad_norm": 0.08160432428121567,
|
| 28625 |
+
"learning_rate": 0.0009841906028360842,
|
| 28626 |
+
"loss": 1.5726,
|
| 28627 |
+
"step": 8140
|
| 28628 |
+
},
|
| 28629 |
+
{
|
| 28630 |
+
"epoch": 0.36201147125516875,
|
| 28631 |
+
"grad_norm": 0.08070631325244904,
|
| 28632 |
+
"learning_rate": 0.0009841817917947767,
|
| 28633 |
+
"loss": 1.5736,
|
| 28634 |
+
"step": 8142
|
| 28635 |
+
},
|
| 28636 |
+
{
|
| 28637 |
+
"epoch": 0.3621003957138411,
|
| 28638 |
+
"grad_norm": 0.07993360608816147,
|
| 28639 |
+
"learning_rate": 0.0009841729783382903,
|
| 28640 |
+
"loss": 1.5757,
|
| 28641 |
+
"step": 8144
|
| 28642 |
+
},
|
| 28643 |
+
{
|
| 28644 |
+
"epoch": 0.36218932017251343,
|
| 28645 |
+
"grad_norm": 0.07653278112411499,
|
| 28646 |
+
"learning_rate": 0.0009841641624666684,
|
| 28647 |
+
"loss": 1.5737,
|
| 28648 |
+
"step": 8146
|
| 28649 |
+
},
|
| 28650 |
+
{
|
| 28651 |
+
"epoch": 0.3622782446311858,
|
| 28652 |
+
"grad_norm": 0.07840976864099503,
|
| 28653 |
+
"learning_rate": 0.0009841553441799554,
|
| 28654 |
+
"loss": 1.5702,
|
| 28655 |
+
"step": 8148
|
| 28656 |
+
},
|
| 28657 |
+
{
|
| 28658 |
+
"epoch": 0.36236716908985817,
|
| 28659 |
+
"grad_norm": 0.07674574106931686,
|
| 28660 |
+
"learning_rate": 0.0009841465234781952,
|
| 28661 |
+
"loss": 1.5761,
|
| 28662 |
+
"step": 8150
|
| 28663 |
+
},
|
| 28664 |
+
{
|
| 28665 |
+
"epoch": 0.36245609354853053,
|
| 28666 |
+
"grad_norm": 0.08122869580984116,
|
| 28667 |
+
"learning_rate": 0.0009841377003614318,
|
| 28668 |
+
"loss": 1.5696,
|
| 28669 |
+
"step": 8152
|
| 28670 |
+
},
|
| 28671 |
+
{
|
| 28672 |
+
"epoch": 0.3625450180072029,
|
| 28673 |
+
"grad_norm": 0.07849126309156418,
|
| 28674 |
+
"learning_rate": 0.0009841288748297092,
|
| 28675 |
+
"loss": 1.5692,
|
| 28676 |
+
"step": 8154
|
| 28677 |
+
},
|
| 28678 |
+
{
|
| 28679 |
+
"epoch": 0.3626339424658752,
|
| 28680 |
+
"grad_norm": 0.0775504857301712,
|
| 28681 |
+
"learning_rate": 0.0009841200468830714,
|
| 28682 |
+
"loss": 1.5748,
|
| 28683 |
+
"step": 8156
|
| 28684 |
+
},
|
| 28685 |
+
{
|
| 28686 |
+
"epoch": 0.3627228669245476,
|
| 28687 |
+
"grad_norm": 0.07808183133602142,
|
| 28688 |
+
"learning_rate": 0.0009841112165215624,
|
| 28689 |
+
"loss": 1.5702,
|
| 28690 |
+
"step": 8158
|
| 28691 |
+
},
|
| 28692 |
+
{
|
| 28693 |
+
"epoch": 0.36281179138321995,
|
| 28694 |
+
"grad_norm": 0.0813973993062973,
|
| 28695 |
+
"learning_rate": 0.0009841023837452263,
|
| 28696 |
+
"loss": 1.5721,
|
| 28697 |
+
"step": 8160
|
| 28698 |
+
},
|
| 28699 |
+
{
|
| 28700 |
+
"epoch": 0.3629007158418923,
|
| 28701 |
+
"grad_norm": 0.07763282209634781,
|
| 28702 |
+
"learning_rate": 0.0009840935485541072,
|
| 28703 |
+
"loss": 1.5666,
|
| 28704 |
+
"step": 8162
|
| 28705 |
+
},
|
| 28706 |
+
{
|
| 28707 |
+
"epoch": 0.3629896403005647,
|
| 28708 |
+
"grad_norm": 0.07707773894071579,
|
| 28709 |
+
"learning_rate": 0.000984084710948249,
|
| 28710 |
+
"loss": 1.5699,
|
| 28711 |
+
"step": 8164
|
| 28712 |
+
},
|
| 28713 |
+
{
|
| 28714 |
+
"epoch": 0.36307856475923705,
|
| 28715 |
+
"grad_norm": 0.07617053389549255,
|
| 28716 |
+
"learning_rate": 0.0009840758709276961,
|
| 28717 |
+
"loss": 1.5731,
|
| 28718 |
+
"step": 8166
|
| 28719 |
+
},
|
| 28720 |
+
{
|
| 28721 |
+
"epoch": 0.36316748921790937,
|
| 28722 |
+
"grad_norm": 0.07609335333108902,
|
| 28723 |
+
"learning_rate": 0.0009840670284924921,
|
| 28724 |
+
"loss": 1.5734,
|
| 28725 |
+
"step": 8168
|
| 28726 |
+
},
|
| 28727 |
+
{
|
| 28728 |
+
"epoch": 0.36325641367658174,
|
| 28729 |
+
"grad_norm": 0.07503112405538559,
|
| 28730 |
+
"learning_rate": 0.0009840581836426817,
|
| 28731 |
+
"loss": 1.5689,
|
| 28732 |
+
"step": 8170
|
| 28733 |
+
},
|
| 28734 |
+
{
|
| 28735 |
+
"epoch": 0.3633453381352541,
|
| 28736 |
+
"grad_norm": 0.07643923163414001,
|
| 28737 |
+
"learning_rate": 0.0009840493363783085,
|
| 28738 |
+
"loss": 1.5717,
|
| 28739 |
+
"step": 8172
|
| 28740 |
+
},
|
| 28741 |
+
{
|
| 28742 |
+
"epoch": 0.36343426259392647,
|
| 28743 |
+
"grad_norm": 0.07574044167995453,
|
| 28744 |
+
"learning_rate": 0.0009840404866994171,
|
| 28745 |
+
"loss": 1.569,
|
| 28746 |
+
"step": 8174
|
| 28747 |
+
},
|
| 28748 |
+
{
|
| 28749 |
+
"epoch": 0.36352318705259884,
|
| 28750 |
+
"grad_norm": 0.07417095452547073,
|
| 28751 |
+
"learning_rate": 0.0009840316346060513,
|
| 28752 |
+
"loss": 1.569,
|
| 28753 |
+
"step": 8176
|
| 28754 |
+
},
|
| 28755 |
+
{
|
| 28756 |
+
"epoch": 0.36361211151127115,
|
| 28757 |
+
"grad_norm": 0.07547099888324738,
|
| 28758 |
+
"learning_rate": 0.0009840227800982553,
|
| 28759 |
+
"loss": 1.5751,
|
| 28760 |
+
"step": 8178
|
| 28761 |
+
},
|
| 28762 |
+
{
|
| 28763 |
+
"epoch": 0.3637010359699435,
|
| 28764 |
+
"grad_norm": 0.07645308971405029,
|
| 28765 |
+
"learning_rate": 0.0009840139231760735,
|
| 28766 |
+
"loss": 1.5648,
|
| 28767 |
+
"step": 8180
|
| 28768 |
+
},
|
| 28769 |
+
{
|
| 28770 |
+
"epoch": 0.3637899604286159,
|
| 28771 |
+
"grad_norm": 0.07473346590995789,
|
| 28772 |
+
"learning_rate": 0.0009840050638395498,
|
| 28773 |
+
"loss": 1.5661,
|
| 28774 |
+
"step": 8182
|
| 28775 |
+
},
|
| 28776 |
+
{
|
| 28777 |
+
"epoch": 0.36387888488728826,
|
| 28778 |
+
"grad_norm": 0.07807636260986328,
|
| 28779 |
+
"learning_rate": 0.0009839962020887286,
|
| 28780 |
+
"loss": 1.5724,
|
| 28781 |
+
"step": 8184
|
| 28782 |
+
},
|
| 28783 |
+
{
|
| 28784 |
+
"epoch": 0.3639678093459606,
|
| 28785 |
+
"grad_norm": 0.07300268113613129,
|
| 28786 |
+
"learning_rate": 0.0009839873379236537,
|
| 28787 |
+
"loss": 1.571,
|
| 28788 |
+
"step": 8186
|
| 28789 |
+
},
|
| 28790 |
+
{
|
| 28791 |
+
"epoch": 0.36405673380463294,
|
| 28792 |
+
"grad_norm": 0.07703215628862381,
|
| 28793 |
+
"learning_rate": 0.00098397847134437,
|
| 28794 |
+
"loss": 1.5722,
|
| 28795 |
+
"step": 8188
|
| 28796 |
+
},
|
| 28797 |
+
{
|
| 28798 |
+
"epoch": 0.3641456582633053,
|
| 28799 |
+
"grad_norm": 0.07669734954833984,
|
| 28800 |
+
"learning_rate": 0.000983969602350921,
|
| 28801 |
+
"loss": 1.5676,
|
| 28802 |
+
"step": 8190
|
| 28803 |
+
},
|
| 28804 |
+
{
|
| 28805 |
+
"epoch": 0.3642345827219777,
|
| 28806 |
+
"grad_norm": 0.07603123784065247,
|
| 28807 |
+
"learning_rate": 0.0009839607309433515,
|
| 28808 |
+
"loss": 1.5688,
|
| 28809 |
+
"step": 8192
|
| 28810 |
+
},
|
| 28811 |
+
{
|
| 28812 |
+
"epoch": 0.36432350718065004,
|
| 28813 |
+
"grad_norm": 0.07759499549865723,
|
| 28814 |
+
"learning_rate": 0.0009839518571217055,
|
| 28815 |
+
"loss": 1.5643,
|
| 28816 |
+
"step": 8194
|
| 28817 |
+
},
|
| 28818 |
+
{
|
| 28819 |
+
"epoch": 0.3644124316393224,
|
| 28820 |
+
"grad_norm": 0.07855424284934998,
|
| 28821 |
+
"learning_rate": 0.0009839429808860274,
|
| 28822 |
+
"loss": 1.5641,
|
| 28823 |
+
"step": 8196
|
| 28824 |
+
},
|
| 28825 |
+
{
|
| 28826 |
+
"epoch": 0.3645013560979948,
|
| 28827 |
+
"grad_norm": 0.07806675881147385,
|
| 28828 |
+
"learning_rate": 0.0009839341022363612,
|
| 28829 |
+
"loss": 1.572,
|
| 28830 |
+
"step": 8198
|
| 28831 |
+
},
|
| 28832 |
+
{
|
| 28833 |
+
"epoch": 0.3645902805566671,
|
| 28834 |
+
"grad_norm": 0.08267048001289368,
|
| 28835 |
+
"learning_rate": 0.0009839252211727514,
|
| 28836 |
+
"loss": 1.5711,
|
| 28837 |
+
"step": 8200
|
| 28838 |
+
},
|
| 28839 |
+
{
|
| 28840 |
+
"epoch": 0.36467920501533946,
|
| 28841 |
+
"grad_norm": 0.07525893300771713,
|
| 28842 |
+
"learning_rate": 0.0009839163376952425,
|
| 28843 |
+
"loss": 1.5687,
|
| 28844 |
+
"step": 8202
|
| 28845 |
+
},
|
| 28846 |
+
{
|
| 28847 |
+
"epoch": 0.3647681294740118,
|
| 28848 |
+
"grad_norm": 0.07745590060949326,
|
| 28849 |
+
"learning_rate": 0.0009839074518038784,
|
| 28850 |
+
"loss": 1.5629,
|
| 28851 |
+
"step": 8204
|
| 28852 |
+
},
|
| 28853 |
+
{
|
| 28854 |
+
"epoch": 0.3648570539326842,
|
| 28855 |
+
"grad_norm": 0.07969274371862411,
|
| 28856 |
+
"learning_rate": 0.0009838985634987036,
|
| 28857 |
+
"loss": 1.5702,
|
| 28858 |
+
"step": 8206
|
| 28859 |
+
},
|
| 28860 |
+
{
|
| 28861 |
+
"epoch": 0.36494597839135656,
|
| 28862 |
+
"grad_norm": 0.0816626101732254,
|
| 28863 |
+
"learning_rate": 0.0009838896727797624,
|
| 28864 |
+
"loss": 1.5599,
|
| 28865 |
+
"step": 8208
|
| 28866 |
+
},
|
| 28867 |
+
{
|
| 28868 |
+
"epoch": 0.3650349028500289,
|
| 28869 |
+
"grad_norm": 0.07653602957725525,
|
| 28870 |
+
"learning_rate": 0.0009838807796470992,
|
| 28871 |
+
"loss": 1.5618,
|
| 28872 |
+
"step": 8210
|
| 28873 |
+
},
|
| 28874 |
+
{
|
| 28875 |
+
"epoch": 0.36512382730870124,
|
| 28876 |
+
"grad_norm": 0.08516646176576614,
|
| 28877 |
+
"learning_rate": 0.0009838718841007586,
|
| 28878 |
+
"loss": 1.5642,
|
| 28879 |
+
"step": 8212
|
| 28880 |
+
},
|
| 28881 |
+
{
|
| 28882 |
+
"epoch": 0.3652127517673736,
|
| 28883 |
+
"grad_norm": 0.07569476962089539,
|
| 28884 |
+
"learning_rate": 0.0009838629861407843,
|
| 28885 |
+
"loss": 1.5597,
|
| 28886 |
+
"step": 8214
|
| 28887 |
+
},
|
| 28888 |
+
{
|
| 28889 |
+
"epoch": 0.365301676226046,
|
| 28890 |
+
"grad_norm": 0.07699557393789291,
|
| 28891 |
+
"learning_rate": 0.0009838540857672214,
|
| 28892 |
+
"loss": 1.5649,
|
| 28893 |
+
"step": 8216
|
| 28894 |
+
},
|
| 28895 |
+
{
|
| 28896 |
+
"epoch": 0.36539060068471835,
|
| 28897 |
+
"grad_norm": 0.07943487912416458,
|
| 28898 |
+
"learning_rate": 0.000983845182980114,
|
| 28899 |
+
"loss": 1.5606,
|
| 28900 |
+
"step": 8218
|
| 28901 |
+
},
|
| 28902 |
+
{
|
| 28903 |
+
"epoch": 0.3654795251433907,
|
| 28904 |
+
"grad_norm": 0.07496616989374161,
|
| 28905 |
+
"learning_rate": 0.0009838362777795065,
|
| 28906 |
+
"loss": 1.5649,
|
| 28907 |
+
"step": 8220
|
| 28908 |
+
},
|
| 28909 |
+
{
|
| 28910 |
+
"epoch": 0.36556844960206303,
|
| 28911 |
+
"grad_norm": 0.07536393404006958,
|
| 28912 |
+
"learning_rate": 0.0009838273701654433,
|
| 28913 |
+
"loss": 1.5696,
|
| 28914 |
+
"step": 8222
|
| 28915 |
+
},
|
| 28916 |
+
{
|
| 28917 |
+
"epoch": 0.3656573740607354,
|
| 28918 |
+
"grad_norm": 0.07574569433927536,
|
| 28919 |
+
"learning_rate": 0.0009838184601379688,
|
| 28920 |
+
"loss": 1.5623,
|
| 28921 |
+
"step": 8224
|
| 28922 |
+
},
|
| 28923 |
+
{
|
| 28924 |
+
"epoch": 0.36574629851940776,
|
| 28925 |
+
"grad_norm": 0.07682286202907562,
|
| 28926 |
+
"learning_rate": 0.0009838095476971274,
|
| 28927 |
+
"loss": 1.5676,
|
| 28928 |
+
"step": 8226
|
| 28929 |
+
},
|
| 28930 |
+
{
|
| 28931 |
+
"epoch": 0.36583522297808013,
|
| 28932 |
+
"grad_norm": 0.08097031712532043,
|
| 28933 |
+
"learning_rate": 0.000983800632842964,
|
| 28934 |
+
"loss": 1.5641,
|
| 28935 |
+
"step": 8228
|
| 28936 |
+
},
|
| 28937 |
+
{
|
| 28938 |
+
"epoch": 0.3659241474367525,
|
| 28939 |
+
"grad_norm": 0.07592246681451797,
|
| 28940 |
+
"learning_rate": 0.0009837917155755226,
|
| 28941 |
+
"loss": 1.5562,
|
| 28942 |
+
"step": 8230
|
| 28943 |
+
},
|
| 28944 |
+
{
|
| 28945 |
+
"epoch": 0.3660130718954248,
|
| 28946 |
+
"grad_norm": 0.07768161594867706,
|
| 28947 |
+
"learning_rate": 0.0009837827958948477,
|
| 28948 |
+
"loss": 1.5638,
|
| 28949 |
+
"step": 8232
|
| 28950 |
+
},
|
| 28951 |
+
{
|
| 28952 |
+
"epoch": 0.3661019963540972,
|
| 28953 |
+
"grad_norm": 0.07560420036315918,
|
| 28954 |
+
"learning_rate": 0.0009837738738009841,
|
| 28955 |
+
"loss": 1.5625,
|
| 28956 |
+
"step": 8234
|
| 28957 |
+
},
|
| 28958 |
+
{
|
| 28959 |
+
"epoch": 0.36619092081276955,
|
| 28960 |
+
"grad_norm": 0.07644996047019958,
|
| 28961 |
+
"learning_rate": 0.000983764949293976,
|
| 28962 |
+
"loss": 1.5602,
|
| 28963 |
+
"step": 8236
|
| 28964 |
+
},
|
| 28965 |
+
{
|
| 28966 |
+
"epoch": 0.3662798452714419,
|
| 28967 |
+
"grad_norm": 0.07451751828193665,
|
| 28968 |
+
"learning_rate": 0.0009837560223738679,
|
| 28969 |
+
"loss": 1.5608,
|
| 28970 |
+
"step": 8238
|
| 28971 |
+
},
|
| 28972 |
+
{
|
| 28973 |
+
"epoch": 0.3663687697301143,
|
| 28974 |
+
"grad_norm": 0.0761575847864151,
|
| 28975 |
+
"learning_rate": 0.0009837470930407046,
|
| 28976 |
+
"loss": 1.5542,
|
| 28977 |
+
"step": 8240
|
| 28978 |
+
},
|
| 28979 |
+
{
|
| 28980 |
+
"epoch": 0.36645769418878665,
|
| 28981 |
+
"grad_norm": 0.0770341232419014,
|
| 28982 |
+
"learning_rate": 0.0009837381612945305,
|
| 28983 |
+
"loss": 1.5584,
|
| 28984 |
+
"step": 8242
|
| 28985 |
+
},
|
| 28986 |
+
{
|
| 28987 |
+
"epoch": 0.36654661864745897,
|
| 28988 |
+
"grad_norm": 0.07663694024085999,
|
| 28989 |
+
"learning_rate": 0.0009837292271353902,
|
| 28990 |
+
"loss": 1.5646,
|
| 28991 |
+
"step": 8244
|
| 28992 |
+
},
|
| 28993 |
+
{
|
| 28994 |
+
"epoch": 0.36663554310613133,
|
| 28995 |
+
"grad_norm": 0.0755605474114418,
|
| 28996 |
+
"learning_rate": 0.0009837202905633282,
|
| 28997 |
+
"loss": 1.5621,
|
| 28998 |
+
"step": 8246
|
| 28999 |
+
},
|
| 29000 |
+
{
|
| 29001 |
+
"epoch": 0.3667244675648037,
|
| 29002 |
+
"grad_norm": 0.07471001893281937,
|
| 29003 |
+
"learning_rate": 0.0009837113515783892,
|
| 29004 |
+
"loss": 1.5641,
|
| 29005 |
+
"step": 8248
|
| 29006 |
+
},
|
| 29007 |
+
{
|
| 29008 |
+
"epoch": 0.36681339202347607,
|
| 29009 |
+
"grad_norm": 0.07854770869016647,
|
| 29010 |
+
"learning_rate": 0.0009837024101806173,
|
| 29011 |
+
"loss": 1.5628,
|
| 29012 |
+
"step": 8250
|
| 29013 |
+
},
|
| 29014 |
+
{
|
| 29015 |
+
"epoch": 0.36690231648214844,
|
| 29016 |
+
"grad_norm": 0.07697580009698868,
|
| 29017 |
+
"learning_rate": 0.000983693466370058,
|
| 29018 |
+
"loss": 1.561,
|
| 29019 |
+
"step": 8252
|
| 29020 |
+
},
|
| 29021 |
+
{
|
| 29022 |
+
"epoch": 0.36699124094082075,
|
| 29023 |
+
"grad_norm": 0.07174623012542725,
|
| 29024 |
+
"learning_rate": 0.000983684520146755,
|
| 29025 |
+
"loss": 1.5571,
|
| 29026 |
+
"step": 8254
|
| 29027 |
+
},
|
| 29028 |
+
{
|
| 29029 |
+
"epoch": 0.3670801653994931,
|
| 29030 |
+
"grad_norm": 0.0765925943851471,
|
| 29031 |
+
"learning_rate": 0.0009836755715107535,
|
| 29032 |
+
"loss": 1.5634,
|
| 29033 |
+
"step": 8256
|
| 29034 |
+
},
|
| 29035 |
+
{
|
| 29036 |
+
"epoch": 0.3671690898581655,
|
| 29037 |
+
"grad_norm": 0.07561610639095306,
|
| 29038 |
+
"learning_rate": 0.000983666620462098,
|
| 29039 |
+
"loss": 1.5588,
|
| 29040 |
+
"step": 8258
|
| 29041 |
+
},
|
| 29042 |
+
{
|
| 29043 |
+
"epoch": 0.36725801431683786,
|
| 29044 |
+
"grad_norm": 0.07434527575969696,
|
| 29045 |
+
"learning_rate": 0.000983657667000833,
|
| 29046 |
+
"loss": 1.5617,
|
| 29047 |
+
"step": 8260
|
| 29048 |
+
},
|
| 29049 |
+
{
|
| 29050 |
+
"epoch": 0.3673469387755102,
|
| 29051 |
+
"grad_norm": 0.07405678927898407,
|
| 29052 |
+
"learning_rate": 0.0009836487111270034,
|
| 29053 |
+
"loss": 1.5646,
|
| 29054 |
+
"step": 8262
|
| 29055 |
+
},
|
| 29056 |
+
{
|
| 29057 |
+
"epoch": 0.36743586323418254,
|
| 29058 |
+
"grad_norm": 0.07639020681381226,
|
| 29059 |
+
"learning_rate": 0.0009836397528406534,
|
| 29060 |
+
"loss": 1.5586,
|
| 29061 |
+
"step": 8264
|
| 29062 |
+
},
|
| 29063 |
+
{
|
| 29064 |
+
"epoch": 0.3675247876928549,
|
| 29065 |
+
"grad_norm": 0.07487154006958008,
|
| 29066 |
+
"learning_rate": 0.0009836307921418284,
|
| 29067 |
+
"loss": 1.5547,
|
| 29068 |
+
"step": 8266
|
| 29069 |
+
},
|
| 29070 |
+
{
|
| 29071 |
+
"epoch": 0.36761371215152727,
|
| 29072 |
+
"grad_norm": 0.07496780157089233,
|
| 29073 |
+
"learning_rate": 0.0009836218290305728,
|
| 29074 |
+
"loss": 1.5611,
|
| 29075 |
+
"step": 8268
|
| 29076 |
+
},
|
| 29077 |
+
{
|
| 29078 |
+
"epoch": 0.36770263661019964,
|
| 29079 |
+
"grad_norm": 0.07398813217878342,
|
| 29080 |
+
"learning_rate": 0.000983612863506931,
|
| 29081 |
+
"loss": 1.558,
|
| 29082 |
+
"step": 8270
|
| 29083 |
+
},
|
| 29084 |
+
{
|
| 29085 |
+
"epoch": 0.367791561068872,
|
| 29086 |
+
"grad_norm": 0.07573290914297104,
|
| 29087 |
+
"learning_rate": 0.000983603895570948,
|
| 29088 |
+
"loss": 1.5595,
|
| 29089 |
+
"step": 8272
|
| 29090 |
+
},
|
| 29091 |
+
{
|
| 29092 |
+
"epoch": 0.3678804855275444,
|
| 29093 |
+
"grad_norm": 0.07165679335594177,
|
| 29094 |
+
"learning_rate": 0.0009835949252226686,
|
| 29095 |
+
"loss": 1.5649,
|
| 29096 |
+
"step": 8274
|
| 29097 |
+
},
|
| 29098 |
+
{
|
| 29099 |
+
"epoch": 0.3679694099862167,
|
| 29100 |
+
"grad_norm": 0.07558693736791611,
|
| 29101 |
+
"learning_rate": 0.0009835859524621373,
|
| 29102 |
+
"loss": 1.5568,
|
| 29103 |
+
"step": 8276
|
| 29104 |
+
},
|
| 29105 |
+
{
|
| 29106 |
+
"epoch": 0.36805833444488906,
|
| 29107 |
+
"grad_norm": 0.0720970556139946,
|
| 29108 |
+
"learning_rate": 0.000983576977289399,
|
| 29109 |
+
"loss": 1.5583,
|
| 29110 |
+
"step": 8278
|
| 29111 |
+
},
|
| 29112 |
+
{
|
| 29113 |
+
"epoch": 0.3681472589035614,
|
| 29114 |
+
"grad_norm": 0.07370097190141678,
|
| 29115 |
+
"learning_rate": 0.0009835679997044985,
|
| 29116 |
+
"loss": 1.5612,
|
| 29117 |
+
"step": 8280
|
| 29118 |
+
},
|
| 29119 |
+
{
|
| 29120 |
+
"epoch": 0.3682361833622338,
|
| 29121 |
+
"grad_norm": 0.07401144504547119,
|
| 29122 |
+
"learning_rate": 0.0009835590197074806,
|
| 29123 |
+
"loss": 1.5596,
|
| 29124 |
+
"step": 8282
|
| 29125 |
+
},
|
| 29126 |
+
{
|
| 29127 |
+
"epoch": 0.36832510782090616,
|
| 29128 |
+
"grad_norm": 0.07251115888357162,
|
| 29129 |
+
"learning_rate": 0.0009835500372983902,
|
| 29130 |
+
"loss": 1.5569,
|
| 29131 |
+
"step": 8284
|
| 29132 |
+
},
|
| 29133 |
+
{
|
| 29134 |
+
"epoch": 0.3684140322795785,
|
| 29135 |
+
"grad_norm": 0.07301607728004456,
|
| 29136 |
+
"learning_rate": 0.0009835410524772718,
|
| 29137 |
+
"loss": 1.5591,
|
| 29138 |
+
"step": 8286
|
| 29139 |
+
},
|
| 29140 |
+
{
|
| 29141 |
+
"epoch": 0.36850295673825084,
|
| 29142 |
+
"grad_norm": 0.07856374233961105,
|
| 29143 |
+
"learning_rate": 0.0009835320652441702,
|
| 29144 |
+
"loss": 1.5658,
|
| 29145 |
+
"step": 8288
|
| 29146 |
+
},
|
| 29147 |
+
{
|
| 29148 |
+
"epoch": 0.3685918811969232,
|
| 29149 |
+
"grad_norm": 0.07428453117609024,
|
| 29150 |
+
"learning_rate": 0.0009835230755991305,
|
| 29151 |
+
"loss": 1.5631,
|
| 29152 |
+
"step": 8290
|
| 29153 |
+
},
|
| 29154 |
+
{
|
| 29155 |
+
"epoch": 0.3686808056555956,
|
| 29156 |
+
"grad_norm": 0.07632127404212952,
|
| 29157 |
+
"learning_rate": 0.0009835140835421977,
|
| 29158 |
+
"loss": 1.5659,
|
| 29159 |
+
"step": 8292
|
| 29160 |
+
},
|
| 29161 |
+
{
|
| 29162 |
+
"epoch": 0.36876973011426795,
|
| 29163 |
+
"grad_norm": 0.07355497777462006,
|
| 29164 |
+
"learning_rate": 0.0009835050890734161,
|
| 29165 |
+
"loss": 1.5582,
|
| 29166 |
+
"step": 8294
|
| 29167 |
+
},
|
| 29168 |
+
{
|
| 29169 |
+
"epoch": 0.3688586545729403,
|
| 29170 |
+
"grad_norm": 0.07494645565748215,
|
| 29171 |
+
"learning_rate": 0.0009834960921928308,
|
| 29172 |
+
"loss": 1.5584,
|
| 29173 |
+
"step": 8296
|
| 29174 |
+
},
|
| 29175 |
+
{
|
| 29176 |
+
"epoch": 0.3689475790316126,
|
| 29177 |
+
"grad_norm": 0.07737577706575394,
|
| 29178 |
+
"learning_rate": 0.000983487092900487,
|
| 29179 |
+
"loss": 1.5562,
|
| 29180 |
+
"step": 8298
|
| 29181 |
+
},
|
| 29182 |
+
{
|
| 29183 |
+
"epoch": 0.369036503490285,
|
| 29184 |
+
"grad_norm": 0.07639828324317932,
|
| 29185 |
+
"learning_rate": 0.000983478091196429,
|
| 29186 |
+
"loss": 1.5646,
|
| 29187 |
+
"step": 8300
|
| 29188 |
+
},
|
| 29189 |
+
{
|
| 29190 |
+
"epoch": 0.36912542794895736,
|
| 29191 |
+
"grad_norm": 0.07329989224672318,
|
| 29192 |
+
"learning_rate": 0.0009834690870807023,
|
| 29193 |
+
"loss": 1.5579,
|
| 29194 |
+
"step": 8302
|
| 29195 |
+
},
|
| 29196 |
+
{
|
| 29197 |
+
"epoch": 0.36921435240762973,
|
| 29198 |
+
"grad_norm": 0.07461026310920715,
|
| 29199 |
+
"learning_rate": 0.0009834600805533516,
|
| 29200 |
+
"loss": 1.5522,
|
| 29201 |
+
"step": 8304
|
| 29202 |
+
},
|
| 29203 |
+
{
|
| 29204 |
+
"epoch": 0.3693032768663021,
|
| 29205 |
+
"grad_norm": 0.07623513787984848,
|
| 29206 |
+
"learning_rate": 0.0009834510716144214,
|
| 29207 |
+
"loss": 1.5643,
|
| 29208 |
+
"step": 8306
|
| 29209 |
+
},
|
| 29210 |
+
{
|
| 29211 |
+
"epoch": 0.3693922013249744,
|
| 29212 |
+
"grad_norm": 0.07590184360742569,
|
| 29213 |
+
"learning_rate": 0.0009834420602639574,
|
| 29214 |
+
"loss": 1.5595,
|
| 29215 |
+
"step": 8308
|
| 29216 |
+
},
|
| 29217 |
+
{
|
| 29218 |
+
"epoch": 0.3694811257836468,
|
| 29219 |
+
"grad_norm": 0.07395032793283463,
|
| 29220 |
+
"learning_rate": 0.0009834330465020038,
|
| 29221 |
+
"loss": 1.5598,
|
| 29222 |
+
"step": 8310
|
| 29223 |
+
},
|
| 29224 |
+
{
|
| 29225 |
+
"epoch": 0.36957005024231915,
|
| 29226 |
+
"grad_norm": 0.07140004634857178,
|
| 29227 |
+
"learning_rate": 0.0009834240303286063,
|
| 29228 |
+
"loss": 1.5628,
|
| 29229 |
+
"step": 8312
|
| 29230 |
+
},
|
| 29231 |
+
{
|
| 29232 |
+
"epoch": 0.3696589747009915,
|
| 29233 |
+
"grad_norm": 0.0748644694685936,
|
| 29234 |
+
"learning_rate": 0.0009834150117438091,
|
| 29235 |
+
"loss": 1.5601,
|
| 29236 |
+
"step": 8314
|
| 29237 |
+
},
|
| 29238 |
+
{
|
| 29239 |
+
"epoch": 0.3697478991596639,
|
| 29240 |
+
"grad_norm": 0.07483812421560287,
|
| 29241 |
+
"learning_rate": 0.0009834059907476579,
|
| 29242 |
+
"loss": 1.5533,
|
| 29243 |
+
"step": 8316
|
| 29244 |
+
},
|
| 29245 |
+
{
|
| 29246 |
+
"epoch": 0.3698368236183362,
|
| 29247 |
+
"grad_norm": 0.07460799068212509,
|
| 29248 |
+
"learning_rate": 0.0009833969673401972,
|
| 29249 |
+
"loss": 1.5583,
|
| 29250 |
+
"step": 8318
|
| 29251 |
+
},
|
| 29252 |
+
{
|
| 29253 |
+
"epoch": 0.36992574807700856,
|
| 29254 |
+
"grad_norm": 0.0767635703086853,
|
| 29255 |
+
"learning_rate": 0.0009833879415214722,
|
| 29256 |
+
"loss": 1.5656,
|
| 29257 |
+
"step": 8320
|
| 29258 |
+
},
|
| 29259 |
+
{
|
| 29260 |
+
"epoch": 0.37001467253568093,
|
| 29261 |
+
"grad_norm": 0.07315431535243988,
|
| 29262 |
+
"learning_rate": 0.0009833789132915282,
|
| 29263 |
+
"loss": 1.5562,
|
| 29264 |
+
"step": 8322
|
| 29265 |
+
},
|
| 29266 |
+
{
|
| 29267 |
+
"epoch": 0.3701035969943533,
|
| 29268 |
+
"grad_norm": 0.07698900997638702,
|
| 29269 |
+
"learning_rate": 0.0009833698826504096,
|
| 29270 |
+
"loss": 1.5598,
|
| 29271 |
+
"step": 8324
|
| 29272 |
+
},
|
| 29273 |
+
{
|
| 29274 |
+
"epoch": 0.37019252145302567,
|
| 29275 |
+
"grad_norm": 0.07581165432929993,
|
| 29276 |
+
"learning_rate": 0.0009833608495981618,
|
| 29277 |
+
"loss": 1.5574,
|
| 29278 |
+
"step": 8326
|
| 29279 |
+
},
|
| 29280 |
+
{
|
| 29281 |
+
"epoch": 0.37028144591169804,
|
| 29282 |
+
"grad_norm": 0.07427161186933517,
|
| 29283 |
+
"learning_rate": 0.0009833518141348298,
|
| 29284 |
+
"loss": 1.5555,
|
| 29285 |
+
"step": 8328
|
| 29286 |
+
},
|
| 29287 |
+
{
|
| 29288 |
+
"epoch": 0.37037037037037035,
|
| 29289 |
+
"grad_norm": 0.07862924039363861,
|
| 29290 |
+
"learning_rate": 0.000983342776260459,
|
| 29291 |
+
"loss": 1.5601,
|
| 29292 |
+
"step": 8330
|
| 29293 |
+
},
|
| 29294 |
+
{
|
| 29295 |
+
"epoch": 0.3704592948290427,
|
| 29296 |
+
"grad_norm": 0.07485569268465042,
|
| 29297 |
+
"learning_rate": 0.000983333735975094,
|
| 29298 |
+
"loss": 1.5526,
|
| 29299 |
+
"step": 8332
|
| 29300 |
+
},
|
| 29301 |
+
{
|
| 29302 |
+
"epoch": 0.3705482192877151,
|
| 29303 |
+
"grad_norm": 0.07545112073421478,
|
| 29304 |
+
"learning_rate": 0.0009833246932787801,
|
| 29305 |
+
"loss": 1.5548,
|
| 29306 |
+
"step": 8334
|
| 29307 |
+
},
|
| 29308 |
+
{
|
| 29309 |
+
"epoch": 0.37063714374638745,
|
| 29310 |
+
"grad_norm": 0.07797663658857346,
|
| 29311 |
+
"learning_rate": 0.0009833156481715624,
|
| 29312 |
+
"loss": 1.5603,
|
| 29313 |
+
"step": 8336
|
| 29314 |
+
},
|
| 29315 |
+
{
|
| 29316 |
+
"epoch": 0.3707260682050598,
|
| 29317 |
+
"grad_norm": 0.07360795885324478,
|
| 29318 |
+
"learning_rate": 0.0009833066006534861,
|
| 29319 |
+
"loss": 1.5535,
|
| 29320 |
+
"step": 8338
|
| 29321 |
+
},
|
| 29322 |
+
{
|
| 29323 |
+
"epoch": 0.37081499266373213,
|
| 29324 |
+
"grad_norm": 0.07508852332830429,
|
| 29325 |
+
"learning_rate": 0.0009832975507245963,
|
| 29326 |
+
"loss": 1.562,
|
| 29327 |
+
"step": 8340
|
| 29328 |
+
},
|
| 29329 |
+
{
|
| 29330 |
+
"epoch": 0.3709039171224045,
|
| 29331 |
+
"grad_norm": 0.07691474258899689,
|
| 29332 |
+
"learning_rate": 0.000983288498384938,
|
| 29333 |
+
"loss": 1.5559,
|
| 29334 |
+
"step": 8342
|
| 29335 |
+
},
|
| 29336 |
+
{
|
| 29337 |
+
"epoch": 0.37099284158107687,
|
| 29338 |
+
"grad_norm": 0.07170414924621582,
|
| 29339 |
+
"learning_rate": 0.0009832794436345565,
|
| 29340 |
+
"loss": 1.5595,
|
| 29341 |
+
"step": 8344
|
| 29342 |
+
},
|
| 29343 |
+
{
|
| 29344 |
+
"epoch": 0.37108176603974924,
|
| 29345 |
+
"grad_norm": 0.07313521206378937,
|
| 29346 |
+
"learning_rate": 0.000983270386473497,
|
| 29347 |
+
"loss": 1.5568,
|
| 29348 |
+
"step": 8346
|
| 29349 |
+
},
|
| 29350 |
+
{
|
| 29351 |
+
"epoch": 0.3711706904984216,
|
| 29352 |
+
"grad_norm": 0.07443317025899887,
|
| 29353 |
+
"learning_rate": 0.0009832613269018043,
|
| 29354 |
+
"loss": 1.5551,
|
| 29355 |
+
"step": 8348
|
| 29356 |
+
},
|
| 29357 |
+
{
|
| 29358 |
+
"epoch": 0.371259614957094,
|
| 29359 |
+
"grad_norm": 0.0727194994688034,
|
| 29360 |
+
"learning_rate": 0.000983252264919524,
|
| 29361 |
+
"loss": 1.5558,
|
| 29362 |
+
"step": 8350
|
| 29363 |
+
},
|
| 29364 |
+
{
|
| 29365 |
+
"epoch": 0.3713485394157663,
|
| 29366 |
+
"grad_norm": 0.07491002231836319,
|
| 29367 |
+
"learning_rate": 0.0009832432005267013,
|
| 29368 |
+
"loss": 1.5553,
|
| 29369 |
+
"step": 8352
|
| 29370 |
+
},
|
| 29371 |
+
{
|
| 29372 |
+
"epoch": 0.37143746387443866,
|
| 29373 |
+
"grad_norm": 0.07130132615566254,
|
| 29374 |
+
"learning_rate": 0.0009832341337233813,
|
| 29375 |
+
"loss": 1.5515,
|
| 29376 |
+
"step": 8354
|
| 29377 |
+
},
|
| 29378 |
+
{
|
| 29379 |
+
"epoch": 0.371526388333111,
|
| 29380 |
+
"grad_norm": 0.0775056853890419,
|
| 29381 |
+
"learning_rate": 0.0009832250645096091,
|
| 29382 |
+
"loss": 1.5612,
|
| 29383 |
+
"step": 8356
|
| 29384 |
+
},
|
| 29385 |
+
{
|
| 29386 |
+
"epoch": 0.3716153127917834,
|
| 29387 |
+
"grad_norm": 0.0759596973657608,
|
| 29388 |
+
"learning_rate": 0.00098321599288543,
|
| 29389 |
+
"loss": 1.554,
|
| 29390 |
+
"step": 8358
|
| 29391 |
+
},
|
| 29392 |
+
{
|
| 29393 |
+
"epoch": 0.37170423725045576,
|
| 29394 |
+
"grad_norm": 0.0767047107219696,
|
| 29395 |
+
"learning_rate": 0.0009832069188508896,
|
| 29396 |
+
"loss": 1.5613,
|
| 29397 |
+
"step": 8360
|
| 29398 |
+
},
|
| 29399 |
+
{
|
| 29400 |
+
"epoch": 0.3717931617091281,
|
| 29401 |
+
"grad_norm": 0.07511959224939346,
|
| 29402 |
+
"learning_rate": 0.0009831978424060328,
|
| 29403 |
+
"loss": 1.5553,
|
| 29404 |
+
"step": 8362
|
| 29405 |
+
},
|
| 29406 |
+
{
|
| 29407 |
+
"epoch": 0.37188208616780044,
|
| 29408 |
+
"grad_norm": 0.07674729824066162,
|
| 29409 |
+
"learning_rate": 0.0009831887635509049,
|
| 29410 |
+
"loss": 1.5505,
|
| 29411 |
+
"step": 8364
|
| 29412 |
+
},
|
| 29413 |
+
{
|
| 29414 |
+
"epoch": 0.3719710106264728,
|
| 29415 |
+
"grad_norm": 0.07360707968473434,
|
| 29416 |
+
"learning_rate": 0.0009831796822855512,
|
| 29417 |
+
"loss": 1.556,
|
| 29418 |
+
"step": 8366
|
| 29419 |
+
},
|
| 29420 |
+
{
|
| 29421 |
+
"epoch": 0.3720599350851452,
|
| 29422 |
+
"grad_norm": 0.0746261477470398,
|
| 29423 |
+
"learning_rate": 0.000983170598610017,
|
| 29424 |
+
"loss": 1.5553,
|
| 29425 |
+
"step": 8368
|
| 29426 |
+
},
|
| 29427 |
+
{
|
| 29428 |
+
"epoch": 0.37214885954381755,
|
| 29429 |
+
"grad_norm": 0.07465831935405731,
|
| 29430 |
+
"learning_rate": 0.0009831615125243478,
|
| 29431 |
+
"loss": 1.55,
|
| 29432 |
+
"step": 8370
|
| 29433 |
+
},
|
| 29434 |
+
{
|
| 29435 |
+
"epoch": 0.37223778400248986,
|
| 29436 |
+
"grad_norm": 0.07530109584331512,
|
| 29437 |
+
"learning_rate": 0.0009831524240285887,
|
| 29438 |
+
"loss": 1.5556,
|
| 29439 |
+
"step": 8372
|
| 29440 |
+
},
|
| 29441 |
+
{
|
| 29442 |
+
"epoch": 0.3723267084611622,
|
| 29443 |
+
"grad_norm": 0.07392653822898865,
|
| 29444 |
+
"learning_rate": 0.0009831433331227851,
|
| 29445 |
+
"loss": 1.5582,
|
| 29446 |
+
"step": 8374
|
| 29447 |
+
},
|
| 29448 |
+
{
|
| 29449 |
+
"epoch": 0.3724156329198346,
|
| 29450 |
+
"grad_norm": 0.0778929591178894,
|
| 29451 |
+
"learning_rate": 0.0009831342398069823,
|
| 29452 |
+
"loss": 1.5604,
|
| 29453 |
+
"step": 8376
|
| 29454 |
+
},
|
| 29455 |
+
{
|
| 29456 |
+
"epoch": 0.37250455737850696,
|
| 29457 |
+
"grad_norm": 0.07217849791049957,
|
| 29458 |
+
"learning_rate": 0.000983125144081226,
|
| 29459 |
+
"loss": 1.5528,
|
| 29460 |
+
"step": 8378
|
| 29461 |
+
},
|
| 29462 |
+
{
|
| 29463 |
+
"epoch": 0.37259348183717933,
|
| 29464 |
+
"grad_norm": 0.07852315157651901,
|
| 29465 |
+
"learning_rate": 0.000983116045945561,
|
| 29466 |
+
"loss": 1.56,
|
| 29467 |
+
"step": 8380
|
| 29468 |
+
},
|
| 29469 |
+
{
|
| 29470 |
+
"epoch": 0.3726824062958517,
|
| 29471 |
+
"grad_norm": 0.07284464687108994,
|
| 29472 |
+
"learning_rate": 0.000983106945400033,
|
| 29473 |
+
"loss": 1.5594,
|
| 29474 |
+
"step": 8382
|
| 29475 |
+
},
|
| 29476 |
+
{
|
| 29477 |
+
"epoch": 0.372771330754524,
|
| 29478 |
+
"grad_norm": 0.07296974211931229,
|
| 29479 |
+
"learning_rate": 0.0009830978424446874,
|
| 29480 |
+
"loss": 1.5558,
|
| 29481 |
+
"step": 8384
|
| 29482 |
+
},
|
| 29483 |
+
{
|
| 29484 |
+
"epoch": 0.3728602552131964,
|
| 29485 |
+
"grad_norm": 0.07269974797964096,
|
| 29486 |
+
"learning_rate": 0.0009830887370795695,
|
| 29487 |
+
"loss": 1.5563,
|
| 29488 |
+
"step": 8386
|
| 29489 |
+
},
|
| 29490 |
+
{
|
| 29491 |
+
"epoch": 0.37294917967186875,
|
| 29492 |
+
"grad_norm": 0.07310895621776581,
|
| 29493 |
+
"learning_rate": 0.0009830796293047249,
|
| 29494 |
+
"loss": 1.559,
|
| 29495 |
+
"step": 8388
|
| 29496 |
+
},
|
| 29497 |
+
{
|
| 29498 |
+
"epoch": 0.3730381041305411,
|
| 29499 |
+
"grad_norm": 0.07368705421686172,
|
| 29500 |
+
"learning_rate": 0.000983070519120199,
|
| 29501 |
+
"loss": 1.5569,
|
| 29502 |
+
"step": 8390
|
| 29503 |
+
},
|
| 29504 |
+
{
|
| 29505 |
+
"epoch": 0.3731270285892135,
|
| 29506 |
+
"grad_norm": 0.07440350204706192,
|
| 29507 |
+
"learning_rate": 0.000983061406526037,
|
| 29508 |
+
"loss": 1.5542,
|
| 29509 |
+
"step": 8392
|
| 29510 |
+
},
|
| 29511 |
+
{
|
| 29512 |
+
"epoch": 0.3732159530478858,
|
| 29513 |
+
"grad_norm": 0.07380738109350204,
|
| 29514 |
+
"learning_rate": 0.0009830522915222845,
|
| 29515 |
+
"loss": 1.5558,
|
| 29516 |
+
"step": 8394
|
| 29517 |
+
},
|
| 29518 |
+
{
|
| 29519 |
+
"epoch": 0.37330487750655816,
|
| 29520 |
+
"grad_norm": 0.0736425444483757,
|
| 29521 |
+
"learning_rate": 0.000983043174108987,
|
| 29522 |
+
"loss": 1.5493,
|
| 29523 |
+
"step": 8396
|
| 29524 |
+
},
|
| 29525 |
+
{
|
| 29526 |
+
"epoch": 0.37339380196523053,
|
| 29527 |
+
"grad_norm": 0.07320426404476166,
|
| 29528 |
+
"learning_rate": 0.0009830340542861902,
|
| 29529 |
+
"loss": 1.5516,
|
| 29530 |
+
"step": 8398
|
| 29531 |
+
},
|
| 29532 |
+
{
|
| 29533 |
+
"epoch": 0.3734827264239029,
|
| 29534 |
+
"grad_norm": 0.07398249953985214,
|
| 29535 |
+
"learning_rate": 0.0009830249320539391,
|
| 29536 |
+
"loss": 1.5563,
|
| 29537 |
+
"step": 8400
|
| 29538 |
+
},
|
| 29539 |
+
{
|
| 29540 |
+
"epoch": 0.37357165088257527,
|
| 29541 |
+
"grad_norm": 0.0731641873717308,
|
| 29542 |
+
"learning_rate": 0.0009830158074122795,
|
| 29543 |
+
"loss": 1.5547,
|
| 29544 |
+
"step": 8402
|
| 29545 |
+
},
|
| 29546 |
+
{
|
| 29547 |
+
"epoch": 0.37366057534124764,
|
| 29548 |
+
"grad_norm": 0.07476949691772461,
|
| 29549 |
+
"learning_rate": 0.000983006680361257,
|
| 29550 |
+
"loss": 1.5563,
|
| 29551 |
+
"step": 8404
|
| 29552 |
+
},
|
| 29553 |
+
{
|
| 29554 |
+
"epoch": 0.37374949979991995,
|
| 29555 |
+
"grad_norm": 0.07096338272094727,
|
| 29556 |
+
"learning_rate": 0.000982997550900917,
|
| 29557 |
+
"loss": 1.5579,
|
| 29558 |
+
"step": 8406
|
| 29559 |
+
},
|
| 29560 |
+
{
|
| 29561 |
+
"epoch": 0.3738384242585923,
|
| 29562 |
+
"grad_norm": 0.0731477290391922,
|
| 29563 |
+
"learning_rate": 0.0009829884190313048,
|
| 29564 |
+
"loss": 1.5571,
|
| 29565 |
+
"step": 8408
|
| 29566 |
+
},
|
| 29567 |
+
{
|
| 29568 |
+
"epoch": 0.3739273487172647,
|
| 29569 |
+
"grad_norm": 0.07443249970674515,
|
| 29570 |
+
"learning_rate": 0.0009829792847524663,
|
| 29571 |
+
"loss": 1.5524,
|
| 29572 |
+
"step": 8410
|
| 29573 |
+
},
|
| 29574 |
+
{
|
| 29575 |
+
"epoch": 0.37401627317593705,
|
| 29576 |
+
"grad_norm": 0.07522628456354141,
|
| 29577 |
+
"learning_rate": 0.000982970148064447,
|
| 29578 |
+
"loss": 1.5515,
|
| 29579 |
+
"step": 8412
|
| 29580 |
+
},
|
| 29581 |
+
{
|
| 29582 |
+
"epoch": 0.3741051976346094,
|
| 29583 |
+
"grad_norm": 0.08042244613170624,
|
| 29584 |
+
"learning_rate": 0.0009829610089672925,
|
| 29585 |
+
"loss": 1.552,
|
| 29586 |
+
"step": 8414
|
| 29587 |
+
},
|
| 29588 |
+
{
|
| 29589 |
+
"epoch": 0.37419412209328173,
|
| 29590 |
+
"grad_norm": 0.0764075294137001,
|
| 29591 |
+
"learning_rate": 0.0009829518674610482,
|
| 29592 |
+
"loss": 1.5509,
|
| 29593 |
+
"step": 8416
|
| 29594 |
+
},
|
| 29595 |
+
{
|
| 29596 |
+
"epoch": 0.3742830465519541,
|
| 29597 |
+
"grad_norm": 0.07679541409015656,
|
| 29598 |
+
"learning_rate": 0.0009829427235457599,
|
| 29599 |
+
"loss": 1.5558,
|
| 29600 |
+
"step": 8418
|
| 29601 |
+
},
|
| 29602 |
+
{
|
| 29603 |
+
"epoch": 0.37437197101062647,
|
| 29604 |
+
"grad_norm": 0.07601416856050491,
|
| 29605 |
+
"learning_rate": 0.0009829335772214732,
|
| 29606 |
+
"loss": 1.5551,
|
| 29607 |
+
"step": 8420
|
| 29608 |
+
},
|
| 29609 |
+
{
|
| 29610 |
+
"epoch": 0.37446089546929884,
|
| 29611 |
+
"grad_norm": 0.07198473066091537,
|
| 29612 |
+
"learning_rate": 0.0009829244284882336,
|
| 29613 |
+
"loss": 1.5515,
|
| 29614 |
+
"step": 8422
|
| 29615 |
+
},
|
| 29616 |
+
{
|
| 29617 |
+
"epoch": 0.3745498199279712,
|
| 29618 |
+
"grad_norm": 0.0754043385386467,
|
| 29619 |
+
"learning_rate": 0.0009829152773460869,
|
| 29620 |
+
"loss": 1.5522,
|
| 29621 |
+
"step": 8424
|
| 29622 |
+
},
|
| 29623 |
+
{
|
| 29624 |
+
"epoch": 0.3746387443866436,
|
| 29625 |
+
"grad_norm": 0.07835698872804642,
|
| 29626 |
+
"learning_rate": 0.0009829061237950783,
|
| 29627 |
+
"loss": 1.5558,
|
| 29628 |
+
"step": 8426
|
| 29629 |
+
},
|
| 29630 |
+
{
|
| 29631 |
+
"epoch": 0.3747276688453159,
|
| 29632 |
+
"grad_norm": 0.07373791188001633,
|
| 29633 |
+
"learning_rate": 0.000982896967835254,
|
| 29634 |
+
"loss": 1.5485,
|
| 29635 |
+
"step": 8428
|
| 29636 |
+
},
|
| 29637 |
+
{
|
| 29638 |
+
"epoch": 0.37481659330398825,
|
| 29639 |
+
"grad_norm": 0.07449112087488174,
|
| 29640 |
+
"learning_rate": 0.0009828878094666595,
|
| 29641 |
+
"loss": 1.5524,
|
| 29642 |
+
"step": 8430
|
| 29643 |
+
},
|
| 29644 |
+
{
|
| 29645 |
+
"epoch": 0.3749055177626606,
|
| 29646 |
+
"grad_norm": 0.07234968990087509,
|
| 29647 |
+
"learning_rate": 0.0009828786486893405,
|
| 29648 |
+
"loss": 1.5579,
|
| 29649 |
+
"step": 8432
|
| 29650 |
+
},
|
| 29651 |
+
{
|
| 29652 |
+
"epoch": 0.374994442221333,
|
| 29653 |
+
"grad_norm": 0.07672835886478424,
|
| 29654 |
+
"learning_rate": 0.0009828694855033425,
|
| 29655 |
+
"loss": 1.5522,
|
| 29656 |
+
"step": 8434
|
| 29657 |
+
},
|
| 29658 |
+
{
|
| 29659 |
+
"epoch": 0.37508336668000536,
|
| 29660 |
+
"grad_norm": 0.07505165040493011,
|
| 29661 |
+
"learning_rate": 0.0009828603199087114,
|
| 29662 |
+
"loss": 1.5515,
|
| 29663 |
+
"step": 8436
|
| 29664 |
+
},
|
| 29665 |
+
{
|
| 29666 |
+
"epoch": 0.37517229113867767,
|
| 29667 |
+
"grad_norm": 0.07347539067268372,
|
| 29668 |
+
"learning_rate": 0.0009828511519054928,
|
| 29669 |
+
"loss": 1.552,
|
| 29670 |
+
"step": 8438
|
| 29671 |
+
},
|
| 29672 |
+
{
|
| 29673 |
+
"epoch": 0.37526121559735004,
|
| 29674 |
+
"grad_norm": 0.07479852437973022,
|
| 29675 |
+
"learning_rate": 0.0009828419814937326,
|
| 29676 |
+
"loss": 1.5485,
|
| 29677 |
+
"step": 8440
|
| 29678 |
+
},
|
| 29679 |
+
{
|
| 29680 |
+
"epoch": 0.3753501400560224,
|
| 29681 |
+
"grad_norm": 0.0721561387181282,
|
| 29682 |
+
"learning_rate": 0.0009828328086734766,
|
| 29683 |
+
"loss": 1.5542,
|
| 29684 |
+
"step": 8442
|
| 29685 |
+
},
|
| 29686 |
+
{
|
| 29687 |
+
"epoch": 0.3754390645146948,
|
| 29688 |
+
"grad_norm": 0.07447963207960129,
|
| 29689 |
+
"learning_rate": 0.0009828236334447702,
|
| 29690 |
+
"loss": 1.5502,
|
| 29691 |
+
"step": 8444
|
| 29692 |
+
},
|
| 29693 |
+
{
|
| 29694 |
+
"epoch": 0.37552798897336714,
|
| 29695 |
+
"grad_norm": 0.07521109282970428,
|
| 29696 |
+
"learning_rate": 0.0009828144558076595,
|
| 29697 |
+
"loss": 1.5478,
|
| 29698 |
+
"step": 8446
|
| 29699 |
+
},
|
| 29700 |
+
{
|
| 29701 |
+
"epoch": 0.37561691343203946,
|
| 29702 |
+
"grad_norm": 0.08029814064502716,
|
| 29703 |
+
"learning_rate": 0.00098280527576219,
|
| 29704 |
+
"loss": 1.5552,
|
| 29705 |
+
"step": 8448
|
| 29706 |
+
},
|
| 29707 |
+
{
|
| 29708 |
+
"epoch": 0.3757058378907118,
|
| 29709 |
+
"grad_norm": 0.07279892265796661,
|
| 29710 |
+
"learning_rate": 0.0009827960933084076,
|
| 29711 |
+
"loss": 1.549,
|
| 29712 |
+
"step": 8450
|
| 29713 |
+
},
|
| 29714 |
+
{
|
| 29715 |
+
"epoch": 0.3757947623493842,
|
| 29716 |
+
"grad_norm": 0.07226986438035965,
|
| 29717 |
+
"learning_rate": 0.0009827869084463583,
|
| 29718 |
+
"loss": 1.5512,
|
| 29719 |
+
"step": 8452
|
| 29720 |
+
},
|
| 29721 |
+
{
|
| 29722 |
+
"epoch": 0.37588368680805656,
|
| 29723 |
+
"grad_norm": 0.07428599148988724,
|
| 29724 |
+
"learning_rate": 0.0009827777211760878,
|
| 29725 |
+
"loss": 1.5563,
|
| 29726 |
+
"step": 8454
|
| 29727 |
+
},
|
| 29728 |
+
{
|
| 29729 |
+
"epoch": 0.37597261126672893,
|
| 29730 |
+
"grad_norm": 0.07335691154003143,
|
| 29731 |
+
"learning_rate": 0.0009827685314976419,
|
| 29732 |
+
"loss": 1.5522,
|
| 29733 |
+
"step": 8456
|
| 29734 |
+
},
|
| 29735 |
+
{
|
| 29736 |
+
"epoch": 0.3760615357254013,
|
| 29737 |
+
"grad_norm": 0.0702577456831932,
|
| 29738 |
+
"learning_rate": 0.000982759339411066,
|
| 29739 |
+
"loss": 1.5493,
|
| 29740 |
+
"step": 8458
|
| 29741 |
+
},
|
| 29742 |
+
{
|
| 29743 |
+
"epoch": 0.3761504601840736,
|
| 29744 |
+
"grad_norm": 0.07333244383335114,
|
| 29745 |
+
"learning_rate": 0.000982750144916407,
|
| 29746 |
+
"loss": 1.5525,
|
| 29747 |
+
"step": 8460
|
| 29748 |
+
},
|
| 29749 |
+
{
|
| 29750 |
+
"epoch": 0.376239384642746,
|
| 29751 |
+
"grad_norm": 0.07639877498149872,
|
| 29752 |
+
"learning_rate": 0.0009827409480137098,
|
| 29753 |
+
"loss": 1.5517,
|
| 29754 |
+
"step": 8462
|
| 29755 |
+
},
|
| 29756 |
+
{
|
| 29757 |
+
"epoch": 0.37632830910141835,
|
| 29758 |
+
"grad_norm": 0.07263181358575821,
|
| 29759 |
+
"learning_rate": 0.0009827317487030207,
|
| 29760 |
+
"loss": 1.549,
|
| 29761 |
+
"step": 8464
|
| 29762 |
+
},
|
| 29763 |
+
{
|
| 29764 |
+
"epoch": 0.3764172335600907,
|
| 29765 |
+
"grad_norm": 0.07310689985752106,
|
| 29766 |
+
"learning_rate": 0.0009827225469843854,
|
| 29767 |
+
"loss": 1.5462,
|
| 29768 |
+
"step": 8466
|
| 29769 |
+
},
|
| 29770 |
+
{
|
| 29771 |
+
"epoch": 0.3765061580187631,
|
| 29772 |
+
"grad_norm": 0.07368572801351547,
|
| 29773 |
+
"learning_rate": 0.00098271334285785,
|
| 29774 |
+
"loss": 1.5532,
|
| 29775 |
+
"step": 8468
|
| 29776 |
+
},
|
| 29777 |
+
{
|
| 29778 |
+
"epoch": 0.3765950824774354,
|
| 29779 |
+
"grad_norm": 0.07141252607107162,
|
| 29780 |
+
"learning_rate": 0.0009827041363234604,
|
| 29781 |
+
"loss": 1.5565,
|
| 29782 |
+
"step": 8470
|
| 29783 |
+
},
|
| 29784 |
+
{
|
| 29785 |
+
"epoch": 0.37668400693610776,
|
| 29786 |
+
"grad_norm": 0.07508658617734909,
|
| 29787 |
+
"learning_rate": 0.0009826949273812622,
|
| 29788 |
+
"loss": 1.5514,
|
| 29789 |
+
"step": 8472
|
| 29790 |
+
},
|
| 29791 |
+
{
|
| 29792 |
+
"epoch": 0.37677293139478013,
|
| 29793 |
+
"grad_norm": 0.07459234446287155,
|
| 29794 |
+
"learning_rate": 0.0009826857160313018,
|
| 29795 |
+
"loss": 1.5596,
|
| 29796 |
+
"step": 8474
|
| 29797 |
+
},
|
| 29798 |
+
{
|
| 29799 |
+
"epoch": 0.3768618558534525,
|
| 29800 |
+
"grad_norm": 0.07409926503896713,
|
| 29801 |
+
"learning_rate": 0.0009826765022736249,
|
| 29802 |
+
"loss": 1.5514,
|
| 29803 |
+
"step": 8476
|
| 29804 |
+
},
|
| 29805 |
+
{
|
| 29806 |
+
"epoch": 0.37695078031212487,
|
| 29807 |
+
"grad_norm": 0.07309357076883316,
|
| 29808 |
+
"learning_rate": 0.0009826672861082775,
|
| 29809 |
+
"loss": 1.5485,
|
| 29810 |
+
"step": 8478
|
| 29811 |
+
},
|
| 29812 |
+
{
|
| 29813 |
+
"epoch": 0.37703970477079723,
|
| 29814 |
+
"grad_norm": 0.07403724640607834,
|
| 29815 |
+
"learning_rate": 0.0009826580675353054,
|
| 29816 |
+
"loss": 1.5558,
|
| 29817 |
+
"step": 8480
|
| 29818 |
+
},
|
| 29819 |
+
{
|
| 29820 |
+
"epoch": 0.37712862922946955,
|
| 29821 |
+
"grad_norm": 0.07526498287916183,
|
| 29822 |
+
"learning_rate": 0.000982648846554755,
|
| 29823 |
+
"loss": 1.5511,
|
| 29824 |
+
"step": 8482
|
| 29825 |
+
},
|
| 29826 |
+
{
|
| 29827 |
+
"epoch": 0.3772175536881419,
|
| 29828 |
+
"grad_norm": 0.07380390167236328,
|
| 29829 |
+
"learning_rate": 0.0009826396231666717,
|
| 29830 |
+
"loss": 1.5511,
|
| 29831 |
+
"step": 8484
|
| 29832 |
+
},
|
| 29833 |
+
{
|
| 29834 |
+
"epoch": 0.3773064781468143,
|
| 29835 |
+
"grad_norm": 0.07699756324291229,
|
| 29836 |
+
"learning_rate": 0.0009826303973711021,
|
| 29837 |
+
"loss": 1.5507,
|
| 29838 |
+
"step": 8486
|
| 29839 |
+
},
|
| 29840 |
+
{
|
| 29841 |
+
"epoch": 0.37739540260548665,
|
| 29842 |
+
"grad_norm": 0.07383593916893005,
|
| 29843 |
+
"learning_rate": 0.000982621169168092,
|
| 29844 |
+
"loss": 1.5523,
|
| 29845 |
+
"step": 8488
|
| 29846 |
+
},
|
| 29847 |
+
{
|
| 29848 |
+
"epoch": 0.377484327064159,
|
| 29849 |
+
"grad_norm": 0.07610490173101425,
|
| 29850 |
+
"learning_rate": 0.0009826119385576873,
|
| 29851 |
+
"loss": 1.5494,
|
| 29852 |
+
"step": 8490
|
| 29853 |
+
},
|
| 29854 |
+
{
|
| 29855 |
+
"epoch": 0.37757325152283133,
|
| 29856 |
+
"grad_norm": 0.07537056505680084,
|
| 29857 |
+
"learning_rate": 0.000982602705539934,
|
| 29858 |
+
"loss": 1.5563,
|
| 29859 |
+
"step": 8492
|
| 29860 |
+
},
|
| 29861 |
+
{
|
| 29862 |
+
"epoch": 0.3776621759815037,
|
| 29863 |
+
"grad_norm": 0.07339107990264893,
|
| 29864 |
+
"learning_rate": 0.0009825934701148784,
|
| 29865 |
+
"loss": 1.5511,
|
| 29866 |
+
"step": 8494
|
| 29867 |
+
},
|
| 29868 |
+
{
|
| 29869 |
+
"epoch": 0.37775110044017607,
|
| 29870 |
+
"grad_norm": 0.0728975385427475,
|
| 29871 |
+
"learning_rate": 0.0009825842322825668,
|
| 29872 |
+
"loss": 1.5491,
|
| 29873 |
+
"step": 8496
|
| 29874 |
+
},
|
| 29875 |
+
{
|
| 29876 |
+
"epoch": 0.37784002489884844,
|
| 29877 |
+
"grad_norm": 0.0715377926826477,
|
| 29878 |
+
"learning_rate": 0.0009825749920430444,
|
| 29879 |
+
"loss": 1.5493,
|
| 29880 |
+
"step": 8498
|
| 29881 |
+
},
|
| 29882 |
+
{
|
| 29883 |
+
"epoch": 0.3779289493575208,
|
| 29884 |
+
"grad_norm": 0.07521523535251617,
|
| 29885 |
+
"learning_rate": 0.000982565749396358,
|
| 29886 |
+
"loss": 1.5483,
|
| 29887 |
+
"step": 8500
|
| 29888 |
+
},
|
| 29889 |
+
{
|
| 29890 |
+
"epoch": 0.3779289493575208,
|
| 29891 |
+
"eval_loss": 1.525511622428894,
|
| 29892 |
+
"eval_runtime": 12.3728,
|
| 29893 |
+
"eval_samples_per_second": 558.484,
|
| 29894 |
+
"eval_steps_per_second": 69.831,
|
| 29895 |
+
"step": 8500
|
| 29896 |
+
},
|
| 29897 |
+
{
|
| 29898 |
+
"epoch": 0.3780178738161931,
|
| 29899 |
+
"grad_norm": 0.07613959908485413,
|
| 29900 |
+
"learning_rate": 0.0009825565043425537,
|
| 29901 |
+
"loss": 1.5473,
|
| 29902 |
+
"step": 8502
|
| 29903 |
+
},
|
| 29904 |
+
{
|
| 29905 |
+
"epoch": 0.3781067982748655,
|
| 29906 |
+
"grad_norm": 0.07284019142389297,
|
| 29907 |
+
"learning_rate": 0.0009825472568816775,
|
| 29908 |
+
"loss": 1.5514,
|
| 29909 |
+
"step": 8504
|
| 29910 |
+
},
|
| 29911 |
+
{
|
| 29912 |
+
"epoch": 0.37819572273353785,
|
| 29913 |
+
"grad_norm": 0.07196670770645142,
|
| 29914 |
+
"learning_rate": 0.0009825380070137752,
|
| 29915 |
+
"loss": 1.557,
|
| 29916 |
+
"step": 8506
|
| 29917 |
+
},
|
| 29918 |
+
{
|
| 29919 |
+
"epoch": 0.3782846471922102,
|
| 29920 |
+
"grad_norm": 0.07932493090629578,
|
| 29921 |
+
"learning_rate": 0.0009825287547388934,
|
| 29922 |
+
"loss": 1.5504,
|
| 29923 |
+
"step": 8508
|
| 29924 |
+
},
|
| 29925 |
+
{
|
| 29926 |
+
"epoch": 0.3783735716508826,
|
| 29927 |
+
"grad_norm": 0.07043527066707611,
|
| 29928 |
+
"learning_rate": 0.000982519500057078,
|
| 29929 |
+
"loss": 1.5411,
|
| 29930 |
+
"step": 8510
|
| 29931 |
+
},
|
| 29932 |
+
{
|
| 29933 |
+
"epoch": 0.37846249610955496,
|
| 29934 |
+
"grad_norm": 0.07196062058210373,
|
| 29935 |
+
"learning_rate": 0.0009825102429683755,
|
| 29936 |
+
"loss": 1.5507,
|
| 29937 |
+
"step": 8512
|
| 29938 |
+
},
|
| 29939 |
+
{
|
| 29940 |
+
"epoch": 0.37855142056822727,
|
| 29941 |
+
"grad_norm": 0.07716414332389832,
|
| 29942 |
+
"learning_rate": 0.0009825009834728315,
|
| 29943 |
+
"loss": 1.5461,
|
| 29944 |
+
"step": 8514
|
| 29945 |
+
},
|
| 29946 |
+
{
|
| 29947 |
+
"epoch": 0.37864034502689964,
|
| 29948 |
+
"grad_norm": 0.06990046054124832,
|
| 29949 |
+
"learning_rate": 0.0009824917215704926,
|
| 29950 |
+
"loss": 1.5522,
|
| 29951 |
+
"step": 8516
|
| 29952 |
+
},
|
| 29953 |
+
{
|
| 29954 |
+
"epoch": 0.378729269485572,
|
| 29955 |
+
"grad_norm": 0.07274769991636276,
|
| 29956 |
+
"learning_rate": 0.000982482457261405,
|
| 29957 |
+
"loss": 1.5507,
|
| 29958 |
+
"step": 8518
|
| 29959 |
+
},
|
| 29960 |
+
{
|
| 29961 |
+
"epoch": 0.3788181939442444,
|
| 29962 |
+
"grad_norm": 0.07060352712869644,
|
| 29963 |
+
"learning_rate": 0.0009824731905456149,
|
| 29964 |
+
"loss": 1.5491,
|
| 29965 |
+
"step": 8520
|
| 29966 |
+
},
|
| 29967 |
+
{
|
| 29968 |
+
"epoch": 0.37890711840291674,
|
| 29969 |
+
"grad_norm": 0.07348603010177612,
|
| 29970 |
+
"learning_rate": 0.000982463921423168,
|
| 29971 |
+
"loss": 1.5473,
|
| 29972 |
+
"step": 8522
|
| 29973 |
+
},
|
| 29974 |
+
{
|
| 29975 |
+
"epoch": 0.37899604286158906,
|
| 29976 |
+
"grad_norm": 0.07370968163013458,
|
| 29977 |
+
"learning_rate": 0.0009824546498941114,
|
| 29978 |
+
"loss": 1.5564,
|
| 29979 |
+
"step": 8524
|
| 29980 |
+
},
|
| 29981 |
+
{
|
| 29982 |
+
"epoch": 0.3790849673202614,
|
| 29983 |
+
"grad_norm": 0.07592252641916275,
|
| 29984 |
+
"learning_rate": 0.0009824453759584907,
|
| 29985 |
+
"loss": 1.5535,
|
| 29986 |
+
"step": 8526
|
| 29987 |
+
},
|
| 29988 |
+
{
|
| 29989 |
+
"epoch": 0.3791738917789338,
|
| 29990 |
+
"grad_norm": 0.07057470828294754,
|
| 29991 |
+
"learning_rate": 0.0009824360996163524,
|
| 29992 |
+
"loss": 1.5567,
|
| 29993 |
+
"step": 8528
|
| 29994 |
+
},
|
| 29995 |
+
{
|
| 29996 |
+
"epoch": 0.37926281623760616,
|
| 29997 |
+
"grad_norm": 0.07211954891681671,
|
| 29998 |
+
"learning_rate": 0.000982426820867743,
|
| 29999 |
+
"loss": 1.5548,
|
| 30000 |
+
"step": 8530
|
| 30001 |
+
},
|
| 30002 |
+
{
|
| 30003 |
+
"epoch": 0.3793517406962785,
|
| 30004 |
+
"grad_norm": 0.06937380880117416,
|
| 30005 |
+
"learning_rate": 0.0009824175397127083,
|
| 30006 |
+
"loss": 1.5457,
|
| 30007 |
+
"step": 8532
|
| 30008 |
+
},
|
| 30009 |
+
{
|
| 30010 |
+
"epoch": 0.3794406651549509,
|
| 30011 |
+
"grad_norm": 0.07182077318429947,
|
| 30012 |
+
"learning_rate": 0.0009824082561512948,
|
| 30013 |
+
"loss": 1.5589,
|
| 30014 |
+
"step": 8534
|
| 30015 |
+
},
|
| 30016 |
+
{
|
| 30017 |
+
"epoch": 0.3795295896136232,
|
| 30018 |
+
"grad_norm": 0.07062894105911255,
|
| 30019 |
+
"learning_rate": 0.000982398970183549,
|
| 30020 |
+
"loss": 1.5499,
|
| 30021 |
+
"step": 8536
|
| 30022 |
+
},
|
| 30023 |
+
{
|
| 30024 |
+
"epoch": 0.3796185140722956,
|
| 30025 |
+
"grad_norm": 0.07044274359941483,
|
| 30026 |
+
"learning_rate": 0.000982389681809517,
|
| 30027 |
+
"loss": 1.5555,
|
| 30028 |
+
"step": 8538
|
| 30029 |
+
},
|
| 30030 |
+
{
|
| 30031 |
+
"epoch": 0.37970743853096794,
|
| 30032 |
+
"grad_norm": 0.0717586874961853,
|
| 30033 |
+
"learning_rate": 0.0009823803910292452,
|
| 30034 |
+
"loss": 1.5452,
|
| 30035 |
+
"step": 8540
|
| 30036 |
+
},
|
| 30037 |
+
{
|
| 30038 |
+
"epoch": 0.3797963629896403,
|
| 30039 |
+
"grad_norm": 0.07334475219249725,
|
| 30040 |
+
"learning_rate": 0.0009823710978427798,
|
| 30041 |
+
"loss": 1.5576,
|
| 30042 |
+
"step": 8542
|
| 30043 |
+
},
|
| 30044 |
+
{
|
| 30045 |
+
"epoch": 0.3798852874483127,
|
| 30046 |
+
"grad_norm": 0.07173279672861099,
|
| 30047 |
+
"learning_rate": 0.0009823618022501675,
|
| 30048 |
+
"loss": 1.5451,
|
| 30049 |
+
"step": 8544
|
| 30050 |
+
},
|
| 30051 |
+
{
|
| 30052 |
+
"epoch": 0.379974211906985,
|
| 30053 |
+
"grad_norm": 0.07294796407222748,
|
| 30054 |
+
"learning_rate": 0.0009823525042514543,
|
| 30055 |
+
"loss": 1.5547,
|
| 30056 |
+
"step": 8546
|
| 30057 |
+
},
|
| 30058 |
+
{
|
| 30059 |
+
"epoch": 0.38006313636565736,
|
| 30060 |
+
"grad_norm": 0.0728803277015686,
|
| 30061 |
+
"learning_rate": 0.0009823432038466868,
|
| 30062 |
+
"loss": 1.5461,
|
| 30063 |
+
"step": 8548
|
| 30064 |
+
},
|
| 30065 |
+
{
|
| 30066 |
+
"epoch": 0.38015206082432973,
|
| 30067 |
+
"grad_norm": 0.070876345038414,
|
| 30068 |
+
"learning_rate": 0.0009823339010359114,
|
| 30069 |
+
"loss": 1.5525,
|
| 30070 |
+
"step": 8550
|
| 30071 |
+
},
|
| 30072 |
+
{
|
| 30073 |
+
"epoch": 0.3802409852830021,
|
| 30074 |
+
"grad_norm": 0.07284954190254211,
|
| 30075 |
+
"learning_rate": 0.0009823245958191745,
|
| 30076 |
+
"loss": 1.5486,
|
| 30077 |
+
"step": 8552
|
| 30078 |
+
},
|
| 30079 |
+
{
|
| 30080 |
+
"epoch": 0.38032990974167447,
|
| 30081 |
+
"grad_norm": 0.0750528946518898,
|
| 30082 |
+
"learning_rate": 0.0009823152881965223,
|
| 30083 |
+
"loss": 1.5507,
|
| 30084 |
+
"step": 8554
|
| 30085 |
+
},
|
| 30086 |
+
{
|
| 30087 |
+
"epoch": 0.3804188342003468,
|
| 30088 |
+
"grad_norm": 0.07205671817064285,
|
| 30089 |
+
"learning_rate": 0.0009823059781680014,
|
| 30090 |
+
"loss": 1.5513,
|
| 30091 |
+
"step": 8556
|
| 30092 |
+
},
|
| 30093 |
+
{
|
| 30094 |
+
"epoch": 0.38050775865901915,
|
| 30095 |
+
"grad_norm": 0.07429669797420502,
|
| 30096 |
+
"learning_rate": 0.0009822966657336582,
|
| 30097 |
+
"loss": 1.5503,
|
| 30098 |
+
"step": 8558
|
| 30099 |
+
},
|
| 30100 |
+
{
|
| 30101 |
+
"epoch": 0.3805966831176915,
|
| 30102 |
+
"grad_norm": 0.0736270323395729,
|
| 30103 |
+
"learning_rate": 0.0009822873508935394,
|
| 30104 |
+
"loss": 1.5513,
|
| 30105 |
+
"step": 8560
|
| 30106 |
+
},
|
| 30107 |
+
{
|
| 30108 |
+
"epoch": 0.3806856075763639,
|
| 30109 |
+
"grad_norm": 0.0691053494811058,
|
| 30110 |
+
"learning_rate": 0.0009822780336476908,
|
| 30111 |
+
"loss": 1.5482,
|
| 30112 |
+
"step": 8562
|
| 30113 |
+
},
|
| 30114 |
+
{
|
| 30115 |
+
"epoch": 0.38077453203503625,
|
| 30116 |
+
"grad_norm": 0.0720055103302002,
|
| 30117 |
+
"learning_rate": 0.0009822687139961596,
|
| 30118 |
+
"loss": 1.5469,
|
| 30119 |
+
"step": 8564
|
| 30120 |
+
},
|
| 30121 |
+
{
|
| 30122 |
+
"epoch": 0.3808634564937086,
|
| 30123 |
+
"grad_norm": 0.07141195237636566,
|
| 30124 |
+
"learning_rate": 0.0009822593919389922,
|
| 30125 |
+
"loss": 1.5539,
|
| 30126 |
+
"step": 8566
|
| 30127 |
+
},
|
| 30128 |
+
{
|
| 30129 |
+
"epoch": 0.38095238095238093,
|
| 30130 |
+
"grad_norm": 0.07182452082633972,
|
| 30131 |
+
"learning_rate": 0.0009822500674762345,
|
| 30132 |
+
"loss": 1.5476,
|
| 30133 |
+
"step": 8568
|
| 30134 |
+
},
|
| 30135 |
+
{
|
| 30136 |
+
"epoch": 0.3810413054110533,
|
| 30137 |
+
"grad_norm": 0.07438525557518005,
|
| 30138 |
+
"learning_rate": 0.0009822407406079336,
|
| 30139 |
+
"loss": 1.5486,
|
| 30140 |
+
"step": 8570
|
| 30141 |
+
},
|
| 30142 |
+
{
|
| 30143 |
+
"epoch": 0.38113022986972567,
|
| 30144 |
+
"grad_norm": 0.07334297895431519,
|
| 30145 |
+
"learning_rate": 0.0009822314113341359,
|
| 30146 |
+
"loss": 1.549,
|
| 30147 |
+
"step": 8572
|
| 30148 |
+
},
|
| 30149 |
+
{
|
| 30150 |
+
"epoch": 0.38121915432839804,
|
| 30151 |
+
"grad_norm": 0.07467876374721527,
|
| 30152 |
+
"learning_rate": 0.0009822220796548878,
|
| 30153 |
+
"loss": 1.5524,
|
| 30154 |
+
"step": 8574
|
| 30155 |
+
},
|
| 30156 |
+
{
|
| 30157 |
+
"epoch": 0.3813080787870704,
|
| 30158 |
+
"grad_norm": 0.0723448172211647,
|
| 30159 |
+
"learning_rate": 0.0009822127455702359,
|
| 30160 |
+
"loss": 1.5459,
|
| 30161 |
+
"step": 8576
|
| 30162 |
+
},
|
| 30163 |
+
{
|
| 30164 |
+
"epoch": 0.3813970032457427,
|
| 30165 |
+
"grad_norm": 0.07537683844566345,
|
| 30166 |
+
"learning_rate": 0.0009822034090802268,
|
| 30167 |
+
"loss": 1.5465,
|
| 30168 |
+
"step": 8578
|
| 30169 |
+
},
|
| 30170 |
+
{
|
| 30171 |
+
"epoch": 0.3814859277044151,
|
| 30172 |
+
"grad_norm": 0.07435756176710129,
|
| 30173 |
+
"learning_rate": 0.0009821940701849072,
|
| 30174 |
+
"loss": 1.5497,
|
| 30175 |
+
"step": 8580
|
| 30176 |
+
},
|
| 30177 |
+
{
|
| 30178 |
+
"epoch": 0.38157485216308745,
|
| 30179 |
+
"grad_norm": 0.07319629937410355,
|
| 30180 |
+
"learning_rate": 0.0009821847288843232,
|
| 30181 |
+
"loss": 1.5454,
|
| 30182 |
+
"step": 8582
|
| 30183 |
+
},
|
| 30184 |
+
{
|
| 30185 |
+
"epoch": 0.3816637766217598,
|
| 30186 |
+
"grad_norm": 0.07365921884775162,
|
| 30187 |
+
"learning_rate": 0.000982175385178522,
|
| 30188 |
+
"loss": 1.5457,
|
| 30189 |
+
"step": 8584
|
| 30190 |
+
},
|
| 30191 |
+
{
|
| 30192 |
+
"epoch": 0.3817527010804322,
|
| 30193 |
+
"grad_norm": 0.06942031532526016,
|
| 30194 |
+
"learning_rate": 0.00098216603906755,
|
| 30195 |
+
"loss": 1.5484,
|
| 30196 |
+
"step": 8586
|
| 30197 |
+
},
|
| 30198 |
+
{
|
| 30199 |
+
"epoch": 0.38184162553910456,
|
| 30200 |
+
"grad_norm": 0.07433439046144485,
|
| 30201 |
+
"learning_rate": 0.0009821566905514536,
|
| 30202 |
+
"loss": 1.5524,
|
| 30203 |
+
"step": 8588
|
| 30204 |
+
},
|
| 30205 |
+
{
|
| 30206 |
+
"epoch": 0.38193054999777687,
|
| 30207 |
+
"grad_norm": 0.07098887860774994,
|
| 30208 |
+
"learning_rate": 0.0009821473396302797,
|
| 30209 |
+
"loss": 1.5419,
|
| 30210 |
+
"step": 8590
|
| 30211 |
+
},
|
| 30212 |
+
{
|
| 30213 |
+
"epoch": 0.38201947445644924,
|
| 30214 |
+
"grad_norm": 0.07336100190877914,
|
| 30215 |
+
"learning_rate": 0.0009821379863040748,
|
| 30216 |
+
"loss": 1.5489,
|
| 30217 |
+
"step": 8592
|
| 30218 |
+
},
|
| 30219 |
+
{
|
| 30220 |
+
"epoch": 0.3821083989151216,
|
| 30221 |
+
"grad_norm": 0.07159991562366486,
|
| 30222 |
+
"learning_rate": 0.0009821286305728854,
|
| 30223 |
+
"loss": 1.5495,
|
| 30224 |
+
"step": 8594
|
| 30225 |
+
},
|
| 30226 |
+
{
|
| 30227 |
+
"epoch": 0.382197323373794,
|
| 30228 |
+
"grad_norm": 0.07283421605825424,
|
| 30229 |
+
"learning_rate": 0.0009821192724367585,
|
| 30230 |
+
"loss": 1.5474,
|
| 30231 |
+
"step": 8596
|
| 30232 |
+
},
|
| 30233 |
+
{
|
| 30234 |
+
"epoch": 0.38228624783246634,
|
| 30235 |
+
"grad_norm": 0.07494281232357025,
|
| 30236 |
+
"learning_rate": 0.0009821099118957407,
|
| 30237 |
+
"loss": 1.5481,
|
| 30238 |
+
"step": 8598
|
| 30239 |
+
},
|
| 30240 |
+
{
|
| 30241 |
+
"epoch": 0.38237517229113865,
|
| 30242 |
+
"grad_norm": 0.07011394947767258,
|
| 30243 |
+
"learning_rate": 0.0009821005489498787,
|
| 30244 |
+
"loss": 1.544,
|
| 30245 |
+
"step": 8600
|
| 30246 |
+
},
|
| 30247 |
+
{
|
| 30248 |
+
"epoch": 0.382464096749811,
|
| 30249 |
+
"grad_norm": 0.07218565046787262,
|
| 30250 |
+
"learning_rate": 0.000982091183599219,
|
| 30251 |
+
"loss": 1.5442,
|
| 30252 |
+
"step": 8602
|
| 30253 |
+
},
|
| 30254 |
+
{
|
| 30255 |
+
"epoch": 0.3825530212084834,
|
| 30256 |
+
"grad_norm": 0.07564288377761841,
|
| 30257 |
+
"learning_rate": 0.0009820818158438084,
|
| 30258 |
+
"loss": 1.5573,
|
| 30259 |
+
"step": 8604
|
| 30260 |
+
},
|
| 30261 |
+
{
|
| 30262 |
+
"epoch": 0.38264194566715576,
|
| 30263 |
+
"grad_norm": 0.07179772108793259,
|
| 30264 |
+
"learning_rate": 0.0009820724456836938,
|
| 30265 |
+
"loss": 1.5556,
|
| 30266 |
+
"step": 8606
|
| 30267 |
+
},
|
| 30268 |
+
{
|
| 30269 |
+
"epoch": 0.3827308701258281,
|
| 30270 |
+
"grad_norm": 0.07555480301380157,
|
| 30271 |
+
"learning_rate": 0.0009820630731189218,
|
| 30272 |
+
"loss": 1.5486,
|
| 30273 |
+
"step": 8608
|
| 30274 |
+
},
|
| 30275 |
+
{
|
| 30276 |
+
"epoch": 0.3828197945845005,
|
| 30277 |
+
"grad_norm": 0.07220586389303207,
|
| 30278 |
+
"learning_rate": 0.0009820536981495391,
|
| 30279 |
+
"loss": 1.5496,
|
| 30280 |
+
"step": 8610
|
| 30281 |
+
},
|
| 30282 |
+
{
|
| 30283 |
+
"epoch": 0.3829087190431728,
|
| 30284 |
+
"grad_norm": 0.07378540188074112,
|
| 30285 |
+
"learning_rate": 0.0009820443207755928,
|
| 30286 |
+
"loss": 1.5498,
|
| 30287 |
+
"step": 8612
|
| 30288 |
+
},
|
| 30289 |
+
{
|
| 30290 |
+
"epoch": 0.3829976435018452,
|
| 30291 |
+
"grad_norm": 0.07111547887325287,
|
| 30292 |
+
"learning_rate": 0.000982034940997129,
|
| 30293 |
+
"loss": 1.5494,
|
| 30294 |
+
"step": 8614
|
| 30295 |
+
},
|
| 30296 |
+
{
|
| 30297 |
+
"epoch": 0.38308656796051754,
|
| 30298 |
+
"grad_norm": 0.07289877533912659,
|
| 30299 |
+
"learning_rate": 0.000982025558814195,
|
| 30300 |
+
"loss": 1.5496,
|
| 30301 |
+
"step": 8616
|
| 30302 |
+
},
|
| 30303 |
+
{
|
| 30304 |
+
"epoch": 0.3831754924191899,
|
| 30305 |
+
"grad_norm": 0.07126215100288391,
|
| 30306 |
+
"learning_rate": 0.0009820161742268379,
|
| 30307 |
+
"loss": 1.5444,
|
| 30308 |
+
"step": 8618
|
| 30309 |
+
},
|
| 30310 |
+
{
|
| 30311 |
+
"epoch": 0.3832644168778623,
|
| 30312 |
+
"grad_norm": 0.07101909816265106,
|
| 30313 |
+
"learning_rate": 0.0009820067872351038,
|
| 30314 |
+
"loss": 1.5456,
|
| 30315 |
+
"step": 8620
|
| 30316 |
+
},
|
| 30317 |
+
{
|
| 30318 |
+
"epoch": 0.3833533413365346,
|
| 30319 |
+
"grad_norm": 0.0723530650138855,
|
| 30320 |
+
"learning_rate": 0.0009819973978390398,
|
| 30321 |
+
"loss": 1.5452,
|
| 30322 |
+
"step": 8622
|
| 30323 |
+
},
|
| 30324 |
+
{
|
| 30325 |
+
"epoch": 0.38344226579520696,
|
| 30326 |
+
"grad_norm": 0.07396049052476883,
|
| 30327 |
+
"learning_rate": 0.0009819880060386928,
|
| 30328 |
+
"loss": 1.5498,
|
| 30329 |
+
"step": 8624
|
| 30330 |
+
},
|
| 30331 |
+
{
|
| 30332 |
+
"epoch": 0.38353119025387933,
|
| 30333 |
+
"grad_norm": 0.07335137575864792,
|
| 30334 |
+
"learning_rate": 0.0009819786118341097,
|
| 30335 |
+
"loss": 1.554,
|
| 30336 |
+
"step": 8626
|
| 30337 |
+
},
|
| 30338 |
+
{
|
| 30339 |
+
"epoch": 0.3836201147125517,
|
| 30340 |
+
"grad_norm": 0.07116411626338959,
|
| 30341 |
+
"learning_rate": 0.0009819692152253372,
|
| 30342 |
+
"loss": 1.5461,
|
| 30343 |
+
"step": 8628
|
| 30344 |
+
},
|
| 30345 |
+
{
|
| 30346 |
+
"epoch": 0.38370903917122406,
|
| 30347 |
+
"grad_norm": 0.07290813326835632,
|
| 30348 |
+
"learning_rate": 0.0009819598162124224,
|
| 30349 |
+
"loss": 1.5431,
|
| 30350 |
+
"step": 8630
|
| 30351 |
+
},
|
| 30352 |
+
{
|
| 30353 |
+
"epoch": 0.3837979636298964,
|
| 30354 |
+
"grad_norm": 0.07139547169208527,
|
| 30355 |
+
"learning_rate": 0.0009819504147954119,
|
| 30356 |
+
"loss": 1.5439,
|
| 30357 |
+
"step": 8632
|
| 30358 |
+
},
|
| 30359 |
+
{
|
| 30360 |
+
"epoch": 0.38388688808856875,
|
| 30361 |
+
"grad_norm": 0.06995617598295212,
|
| 30362 |
+
"learning_rate": 0.0009819410109743528,
|
| 30363 |
+
"loss": 1.5455,
|
| 30364 |
+
"step": 8634
|
| 30365 |
+
},
|
| 30366 |
+
{
|
| 30367 |
+
"epoch": 0.3839758125472411,
|
| 30368 |
+
"grad_norm": 0.07103978097438812,
|
| 30369 |
+
"learning_rate": 0.0009819316047492918,
|
| 30370 |
+
"loss": 1.5439,
|
| 30371 |
+
"step": 8636
|
| 30372 |
+
},
|
| 30373 |
+
{
|
| 30374 |
+
"epoch": 0.3840647370059135,
|
| 30375 |
+
"grad_norm": 0.07140542566776276,
|
| 30376 |
+
"learning_rate": 0.000981922196120276,
|
| 30377 |
+
"loss": 1.5404,
|
| 30378 |
+
"step": 8638
|
| 30379 |
+
},
|
| 30380 |
+
{
|
| 30381 |
+
"epoch": 0.38415366146458585,
|
| 30382 |
+
"grad_norm": 0.0695871040225029,
|
| 30383 |
+
"learning_rate": 0.0009819127850873525,
|
| 30384 |
+
"loss": 1.5443,
|
| 30385 |
+
"step": 8640
|
| 30386 |
+
},
|
| 30387 |
+
{
|
| 30388 |
+
"epoch": 0.3842425859232582,
|
| 30389 |
+
"grad_norm": 0.07270847260951996,
|
| 30390 |
+
"learning_rate": 0.0009819033716505679,
|
| 30391 |
+
"loss": 1.5423,
|
| 30392 |
+
"step": 8642
|
| 30393 |
+
},
|
| 30394 |
+
{
|
| 30395 |
+
"epoch": 0.38433151038193053,
|
| 30396 |
+
"grad_norm": 0.07431674748659134,
|
| 30397 |
+
"learning_rate": 0.000981893955809969,
|
| 30398 |
+
"loss": 1.5434,
|
| 30399 |
+
"step": 8644
|
| 30400 |
+
},
|
| 30401 |
+
{
|
| 30402 |
+
"epoch": 0.3844204348406029,
|
| 30403 |
+
"grad_norm": 0.07491683959960938,
|
| 30404 |
+
"learning_rate": 0.0009818845375656035,
|
| 30405 |
+
"loss": 1.5443,
|
| 30406 |
+
"step": 8646
|
| 30407 |
+
},
|
| 30408 |
+
{
|
| 30409 |
+
"epoch": 0.38450935929927527,
|
| 30410 |
+
"grad_norm": 0.0724981278181076,
|
| 30411 |
+
"learning_rate": 0.0009818751169175177,
|
| 30412 |
+
"loss": 1.5455,
|
| 30413 |
+
"step": 8648
|
| 30414 |
+
},
|
| 30415 |
+
{
|
| 30416 |
+
"epoch": 0.38459828375794763,
|
| 30417 |
+
"grad_norm": 0.07452180981636047,
|
| 30418 |
+
"learning_rate": 0.0009818656938657589,
|
| 30419 |
+
"loss": 1.55,
|
| 30420 |
+
"step": 8650
|
| 30421 |
+
},
|
| 30422 |
+
{
|
| 30423 |
+
"epoch": 0.38468720821662,
|
| 30424 |
+
"grad_norm": 0.07273737341165543,
|
| 30425 |
+
"learning_rate": 0.000981856268410374,
|
| 30426 |
+
"loss": 1.5422,
|
| 30427 |
+
"step": 8652
|
| 30428 |
+
},
|
| 30429 |
+
{
|
| 30430 |
+
"epoch": 0.3847761326752923,
|
| 30431 |
+
"grad_norm": 0.07237744331359863,
|
| 30432 |
+
"learning_rate": 0.0009818468405514101,
|
| 30433 |
+
"loss": 1.5468,
|
| 30434 |
+
"step": 8654
|
| 30435 |
+
},
|
| 30436 |
+
{
|
| 30437 |
+
"epoch": 0.3848650571339647,
|
| 30438 |
+
"grad_norm": 0.07318416237831116,
|
| 30439 |
+
"learning_rate": 0.0009818374102889141,
|
| 30440 |
+
"loss": 1.5456,
|
| 30441 |
+
"step": 8656
|
| 30442 |
+
},
|
| 30443 |
+
{
|
| 30444 |
+
"epoch": 0.38495398159263705,
|
| 30445 |
+
"grad_norm": 0.0707939937710762,
|
| 30446 |
+
"learning_rate": 0.0009818279776229333,
|
| 30447 |
+
"loss": 1.5468,
|
| 30448 |
+
"step": 8658
|
| 30449 |
+
},
|
| 30450 |
+
{
|
| 30451 |
+
"epoch": 0.3850429060513094,
|
| 30452 |
+
"grad_norm": 0.07714451104402542,
|
| 30453 |
+
"learning_rate": 0.0009818185425535145,
|
| 30454 |
+
"loss": 1.55,
|
| 30455 |
+
"step": 8660
|
| 30456 |
+
},
|
| 30457 |
+
{
|
| 30458 |
+
"epoch": 0.3851318305099818,
|
| 30459 |
+
"grad_norm": 0.07308610528707504,
|
| 30460 |
+
"learning_rate": 0.0009818091050807047,
|
| 30461 |
+
"loss": 1.5422,
|
| 30462 |
+
"step": 8662
|
| 30463 |
+
},
|
| 30464 |
+
{
|
| 30465 |
+
"epoch": 0.38522075496865416,
|
| 30466 |
+
"grad_norm": 0.07573612779378891,
|
| 30467 |
+
"learning_rate": 0.0009817996652045512,
|
| 30468 |
+
"loss": 1.5528,
|
| 30469 |
+
"step": 8664
|
| 30470 |
+
},
|
| 30471 |
+
{
|
| 30472 |
+
"epoch": 0.38530967942732647,
|
| 30473 |
+
"grad_norm": 0.07352116703987122,
|
| 30474 |
+
"learning_rate": 0.0009817902229251009,
|
| 30475 |
+
"loss": 1.5453,
|
| 30476 |
+
"step": 8666
|
| 30477 |
+
},
|
| 30478 |
+
{
|
| 30479 |
+
"epoch": 0.38539860388599884,
|
| 30480 |
+
"grad_norm": 0.0721651017665863,
|
| 30481 |
+
"learning_rate": 0.000981780778242401,
|
| 30482 |
+
"loss": 1.5484,
|
| 30483 |
+
"step": 8668
|
| 30484 |
+
},
|
| 30485 |
+
{
|
| 30486 |
+
"epoch": 0.3854875283446712,
|
| 30487 |
+
"grad_norm": 0.07210839539766312,
|
| 30488 |
+
"learning_rate": 0.0009817713311564987,
|
| 30489 |
+
"loss": 1.5465,
|
| 30490 |
+
"step": 8670
|
| 30491 |
+
},
|
| 30492 |
+
{
|
| 30493 |
+
"epoch": 0.3855764528033436,
|
| 30494 |
+
"grad_norm": 0.07044050097465515,
|
| 30495 |
+
"learning_rate": 0.000981761881667441,
|
| 30496 |
+
"loss": 1.5464,
|
| 30497 |
+
"step": 8672
|
| 30498 |
+
},
|
| 30499 |
+
{
|
| 30500 |
+
"epoch": 0.38566537726201594,
|
| 30501 |
+
"grad_norm": 0.07243029773235321,
|
| 30502 |
+
"learning_rate": 0.000981752429775275,
|
| 30503 |
+
"loss": 1.5452,
|
| 30504 |
+
"step": 8674
|
| 30505 |
+
},
|
| 30506 |
+
{
|
| 30507 |
+
"epoch": 0.38575430172068825,
|
| 30508 |
+
"grad_norm": 0.07179372757673264,
|
| 30509 |
+
"learning_rate": 0.000981742975480048,
|
| 30510 |
+
"loss": 1.5457,
|
| 30511 |
+
"step": 8676
|
| 30512 |
+
},
|
| 30513 |
+
{
|
| 30514 |
+
"epoch": 0.3858432261793606,
|
| 30515 |
+
"grad_norm": 0.07337214797735214,
|
| 30516 |
+
"learning_rate": 0.000981733518781807,
|
| 30517 |
+
"loss": 1.5459,
|
| 30518 |
+
"step": 8678
|
| 30519 |
+
},
|
| 30520 |
+
{
|
| 30521 |
+
"epoch": 0.385932150638033,
|
| 30522 |
+
"grad_norm": 0.0745837390422821,
|
| 30523 |
+
"learning_rate": 0.0009817240596805994,
|
| 30524 |
+
"loss": 1.5494,
|
| 30525 |
+
"step": 8680
|
| 30526 |
+
},
|
| 30527 |
+
{
|
| 30528 |
+
"epoch": 0.38602107509670536,
|
| 30529 |
+
"grad_norm": 0.072391077876091,
|
| 30530 |
+
"learning_rate": 0.0009817145981764722,
|
| 30531 |
+
"loss": 1.5427,
|
| 30532 |
+
"step": 8682
|
| 30533 |
+
},
|
| 30534 |
+
{
|
| 30535 |
+
"epoch": 0.3861099995553777,
|
| 30536 |
+
"grad_norm": 0.07101153582334518,
|
| 30537 |
+
"learning_rate": 0.0009817051342694725,
|
| 30538 |
+
"loss": 1.5478,
|
| 30539 |
+
"step": 8684
|
| 30540 |
+
},
|
| 30541 |
+
{
|
| 30542 |
+
"epoch": 0.38619892401405004,
|
| 30543 |
+
"grad_norm": 0.07185351103544235,
|
| 30544 |
+
"learning_rate": 0.0009816956679596475,
|
| 30545 |
+
"loss": 1.5428,
|
| 30546 |
+
"step": 8686
|
| 30547 |
+
},
|
| 30548 |
+
{
|
| 30549 |
+
"epoch": 0.3862878484727224,
|
| 30550 |
+
"grad_norm": 0.07106965035200119,
|
| 30551 |
+
"learning_rate": 0.000981686199247045,
|
| 30552 |
+
"loss": 1.5515,
|
| 30553 |
+
"step": 8688
|
| 30554 |
+
},
|
| 30555 |
+
{
|
| 30556 |
+
"epoch": 0.3863767729313948,
|
| 30557 |
+
"grad_norm": 0.07175727933645248,
|
| 30558 |
+
"learning_rate": 0.0009816767281317113,
|
| 30559 |
+
"loss": 1.5422,
|
| 30560 |
+
"step": 8690
|
| 30561 |
+
},
|
| 30562 |
+
{
|
| 30563 |
+
"epoch": 0.38646569739006714,
|
| 30564 |
+
"grad_norm": 0.0721508041024208,
|
| 30565 |
+
"learning_rate": 0.0009816672546136944,
|
| 30566 |
+
"loss": 1.5444,
|
| 30567 |
+
"step": 8692
|
| 30568 |
+
},
|
| 30569 |
+
{
|
| 30570 |
+
"epoch": 0.3865546218487395,
|
| 30571 |
+
"grad_norm": 0.07354867458343506,
|
| 30572 |
+
"learning_rate": 0.0009816577786930414,
|
| 30573 |
+
"loss": 1.5476,
|
| 30574 |
+
"step": 8694
|
| 30575 |
+
},
|
| 30576 |
+
{
|
| 30577 |
+
"epoch": 0.3866435463074119,
|
| 30578 |
+
"grad_norm": 0.07081717252731323,
|
| 30579 |
+
"learning_rate": 0.000981648300369799,
|
| 30580 |
+
"loss": 1.5489,
|
| 30581 |
+
"step": 8696
|
| 30582 |
+
},
|
| 30583 |
+
{
|
| 30584 |
+
"epoch": 0.3867324707660842,
|
| 30585 |
+
"grad_norm": 0.07528056204319,
|
| 30586 |
+
"learning_rate": 0.0009816388196440154,
|
| 30587 |
+
"loss": 1.5507,
|
| 30588 |
+
"step": 8698
|
| 30589 |
+
},
|
| 30590 |
+
{
|
| 30591 |
+
"epoch": 0.38682139522475656,
|
| 30592 |
+
"grad_norm": 0.07244568318128586,
|
| 30593 |
+
"learning_rate": 0.000981629336515737,
|
| 30594 |
+
"loss": 1.5481,
|
| 30595 |
+
"step": 8700
|
| 30596 |
+
},
|
| 30597 |
+
{
|
| 30598 |
+
"epoch": 0.3869103196834289,
|
| 30599 |
+
"grad_norm": 0.07286922633647919,
|
| 30600 |
+
"learning_rate": 0.0009816198509850118,
|
| 30601 |
+
"loss": 1.5445,
|
| 30602 |
+
"step": 8702
|
| 30603 |
+
},
|
| 30604 |
+
{
|
| 30605 |
+
"epoch": 0.3869992441421013,
|
| 30606 |
+
"grad_norm": 0.06972247362136841,
|
| 30607 |
+
"learning_rate": 0.000981610363051887,
|
| 30608 |
+
"loss": 1.5486,
|
| 30609 |
+
"step": 8704
|
| 30610 |
+
},
|
| 30611 |
+
{
|
| 30612 |
+
"epoch": 0.38708816860077366,
|
| 30613 |
+
"grad_norm": 0.07090491056442261,
|
| 30614 |
+
"learning_rate": 0.0009816008727164093,
|
| 30615 |
+
"loss": 1.5433,
|
| 30616 |
+
"step": 8706
|
| 30617 |
+
},
|
| 30618 |
+
{
|
| 30619 |
+
"epoch": 0.387177093059446,
|
| 30620 |
+
"grad_norm": 0.0725286453962326,
|
| 30621 |
+
"learning_rate": 0.0009815913799786269,
|
| 30622 |
+
"loss": 1.5452,
|
| 30623 |
+
"step": 8708
|
| 30624 |
+
},
|
| 30625 |
+
{
|
| 30626 |
+
"epoch": 0.38726601751811834,
|
| 30627 |
+
"grad_norm": 0.07091175764799118,
|
| 30628 |
+
"learning_rate": 0.0009815818848385865,
|
| 30629 |
+
"loss": 1.5429,
|
| 30630 |
+
"step": 8710
|
| 30631 |
+
},
|
| 30632 |
+
{
|
| 30633 |
+
"epoch": 0.3873549419767907,
|
| 30634 |
+
"grad_norm": 0.07452746480703354,
|
| 30635 |
+
"learning_rate": 0.0009815723872963358,
|
| 30636 |
+
"loss": 1.5374,
|
| 30637 |
+
"step": 8712
|
| 30638 |
+
},
|
| 30639 |
+
{
|
| 30640 |
+
"epoch": 0.3874438664354631,
|
| 30641 |
+
"grad_norm": 0.07199020683765411,
|
| 30642 |
+
"learning_rate": 0.0009815628873519223,
|
| 30643 |
+
"loss": 1.5389,
|
| 30644 |
+
"step": 8714
|
| 30645 |
+
},
|
| 30646 |
+
{
|
| 30647 |
+
"epoch": 0.38753279089413545,
|
| 30648 |
+
"grad_norm": 0.07434893399477005,
|
| 30649 |
+
"learning_rate": 0.0009815533850053928,
|
| 30650 |
+
"loss": 1.5465,
|
| 30651 |
+
"step": 8716
|
| 30652 |
+
},
|
| 30653 |
+
{
|
| 30654 |
+
"epoch": 0.3876217153528078,
|
| 30655 |
+
"grad_norm": 0.0721900537610054,
|
| 30656 |
+
"learning_rate": 0.0009815438802567955,
|
| 30657 |
+
"loss": 1.5464,
|
| 30658 |
+
"step": 8718
|
| 30659 |
+
},
|
| 30660 |
+
{
|
| 30661 |
+
"epoch": 0.38771063981148013,
|
| 30662 |
+
"grad_norm": 0.0768774077296257,
|
| 30663 |
+
"learning_rate": 0.000981534373106177,
|
| 30664 |
+
"loss": 1.5477,
|
| 30665 |
+
"step": 8720
|
| 30666 |
+
},
|
| 30667 |
+
{
|
| 30668 |
+
"epoch": 0.3877995642701525,
|
| 30669 |
+
"grad_norm": 0.07451849430799484,
|
| 30670 |
+
"learning_rate": 0.0009815248635535853,
|
| 30671 |
+
"loss": 1.5439,
|
| 30672 |
+
"step": 8722
|
| 30673 |
+
},
|
| 30674 |
+
{
|
| 30675 |
+
"epoch": 0.38788848872882487,
|
| 30676 |
+
"grad_norm": 0.0778982937335968,
|
| 30677 |
+
"learning_rate": 0.0009815153515990679,
|
| 30678 |
+
"loss": 1.5425,
|
| 30679 |
+
"step": 8724
|
| 30680 |
+
},
|
| 30681 |
+
{
|
| 30682 |
+
"epoch": 0.38797741318749723,
|
| 30683 |
+
"grad_norm": 0.07124931365251541,
|
| 30684 |
+
"learning_rate": 0.0009815058372426716,
|
| 30685 |
+
"loss": 1.5497,
|
| 30686 |
+
"step": 8726
|
| 30687 |
+
},
|
| 30688 |
+
{
|
| 30689 |
+
"epoch": 0.3880663376461696,
|
| 30690 |
+
"grad_norm": 0.0718124508857727,
|
| 30691 |
+
"learning_rate": 0.0009814963204844444,
|
| 30692 |
+
"loss": 1.5436,
|
| 30693 |
+
"step": 8728
|
| 30694 |
+
},
|
| 30695 |
+
{
|
| 30696 |
+
"epoch": 0.3881552621048419,
|
| 30697 |
+
"grad_norm": 0.07384263724088669,
|
| 30698 |
+
"learning_rate": 0.0009814868013244336,
|
| 30699 |
+
"loss": 1.5444,
|
| 30700 |
+
"step": 8730
|
| 30701 |
+
},
|
| 30702 |
+
{
|
| 30703 |
+
"epoch": 0.3882441865635143,
|
| 30704 |
+
"grad_norm": 0.07269671559333801,
|
| 30705 |
+
"learning_rate": 0.0009814772797626867,
|
| 30706 |
+
"loss": 1.5482,
|
| 30707 |
+
"step": 8732
|
| 30708 |
+
},
|
| 30709 |
+
{
|
| 30710 |
+
"epoch": 0.38833311102218665,
|
| 30711 |
+
"grad_norm": 0.0724627822637558,
|
| 30712 |
+
"learning_rate": 0.0009814677557992513,
|
| 30713 |
+
"loss": 1.5495,
|
| 30714 |
+
"step": 8734
|
| 30715 |
+
},
|
| 30716 |
+
{
|
| 30717 |
+
"epoch": 0.388422035480859,
|
| 30718 |
+
"grad_norm": 0.07544320821762085,
|
| 30719 |
+
"learning_rate": 0.0009814582294341747,
|
| 30720 |
+
"loss": 1.5459,
|
| 30721 |
+
"step": 8736
|
| 30722 |
+
},
|
| 30723 |
+
{
|
| 30724 |
+
"epoch": 0.3885109599395314,
|
| 30725 |
+
"grad_norm": 0.0732409730553627,
|
| 30726 |
+
"learning_rate": 0.0009814487006675047,
|
| 30727 |
+
"loss": 1.543,
|
| 30728 |
+
"step": 8738
|
| 30729 |
+
},
|
| 30730 |
+
{
|
| 30731 |
+
"epoch": 0.3885998843982037,
|
| 30732 |
+
"grad_norm": 0.071528859436512,
|
| 30733 |
+
"learning_rate": 0.0009814391694992885,
|
| 30734 |
+
"loss": 1.5489,
|
| 30735 |
+
"step": 8740
|
| 30736 |
+
},
|
| 30737 |
+
{
|
| 30738 |
+
"epoch": 0.38868880885687607,
|
| 30739 |
+
"grad_norm": 0.07471916079521179,
|
| 30740 |
+
"learning_rate": 0.000981429635929574,
|
| 30741 |
+
"loss": 1.5427,
|
| 30742 |
+
"step": 8742
|
| 30743 |
+
},
|
| 30744 |
+
{
|
| 30745 |
+
"epoch": 0.38877773331554843,
|
| 30746 |
+
"grad_norm": 0.07275721430778503,
|
| 30747 |
+
"learning_rate": 0.0009814200999584085,
|
| 30748 |
+
"loss": 1.5449,
|
| 30749 |
+
"step": 8744
|
| 30750 |
+
},
|
| 30751 |
+
{
|
| 30752 |
+
"epoch": 0.3888666577742208,
|
| 30753 |
+
"grad_norm": 0.07258014380931854,
|
| 30754 |
+
"learning_rate": 0.0009814105615858395,
|
| 30755 |
+
"loss": 1.543,
|
| 30756 |
+
"step": 8746
|
| 30757 |
+
},
|
| 30758 |
+
{
|
| 30759 |
+
"epoch": 0.38895558223289317,
|
| 30760 |
+
"grad_norm": 0.07144364714622498,
|
| 30761 |
+
"learning_rate": 0.0009814010208119147,
|
| 30762 |
+
"loss": 1.5451,
|
| 30763 |
+
"step": 8748
|
| 30764 |
+
},
|
| 30765 |
+
{
|
| 30766 |
+
"epoch": 0.38904450669156554,
|
| 30767 |
+
"grad_norm": 0.07227019965648651,
|
| 30768 |
+
"learning_rate": 0.0009813914776366816,
|
| 30769 |
+
"loss": 1.5423,
|
| 30770 |
+
"step": 8750
|
| 30771 |
+
},
|
| 30772 |
+
{
|
| 30773 |
+
"epoch": 0.38913343115023785,
|
| 30774 |
+
"grad_norm": 0.0736045092344284,
|
| 30775 |
+
"learning_rate": 0.0009813819320601883,
|
| 30776 |
+
"loss": 1.5435,
|
| 30777 |
+
"step": 8752
|
| 30778 |
+
},
|
| 30779 |
+
{
|
| 30780 |
+
"epoch": 0.3892223556089102,
|
| 30781 |
+
"grad_norm": 0.07181594520807266,
|
| 30782 |
+
"learning_rate": 0.0009813723840824819,
|
| 30783 |
+
"loss": 1.5413,
|
| 30784 |
+
"step": 8754
|
| 30785 |
+
},
|
| 30786 |
+
{
|
| 30787 |
+
"epoch": 0.3893112800675826,
|
| 30788 |
+
"grad_norm": 0.07333064079284668,
|
| 30789 |
+
"learning_rate": 0.0009813628337036098,
|
| 30790 |
+
"loss": 1.5514,
|
| 30791 |
+
"step": 8756
|
| 30792 |
+
},
|
| 30793 |
+
{
|
| 30794 |
+
"epoch": 0.38940020452625496,
|
| 30795 |
+
"grad_norm": 0.07221698760986328,
|
| 30796 |
+
"learning_rate": 0.0009813532809236202,
|
| 30797 |
+
"loss": 1.5435,
|
| 30798 |
+
"step": 8758
|
| 30799 |
+
},
|
| 30800 |
+
{
|
| 30801 |
+
"epoch": 0.3894891289849273,
|
| 30802 |
+
"grad_norm": 0.07169820368289948,
|
| 30803 |
+
"learning_rate": 0.0009813437257425606,
|
| 30804 |
+
"loss": 1.5474,
|
| 30805 |
+
"step": 8760
|
| 30806 |
+
},
|
| 30807 |
+
{
|
| 30808 |
+
"epoch": 0.38957805344359964,
|
| 30809 |
+
"grad_norm": 0.07131782174110413,
|
| 30810 |
+
"learning_rate": 0.0009813341681604785,
|
| 30811 |
+
"loss": 1.5431,
|
| 30812 |
+
"step": 8762
|
| 30813 |
+
},
|
| 30814 |
+
{
|
| 30815 |
+
"epoch": 0.389666977902272,
|
| 30816 |
+
"grad_norm": 0.07381466031074524,
|
| 30817 |
+
"learning_rate": 0.0009813246081774218,
|
| 30818 |
+
"loss": 1.551,
|
| 30819 |
+
"step": 8764
|
| 30820 |
+
},
|
| 30821 |
+
{
|
| 30822 |
+
"epoch": 0.3897559023609444,
|
| 30823 |
+
"grad_norm": 0.07252717018127441,
|
| 30824 |
+
"learning_rate": 0.000981315045793438,
|
| 30825 |
+
"loss": 1.542,
|
| 30826 |
+
"step": 8766
|
| 30827 |
+
},
|
| 30828 |
+
{
|
| 30829 |
+
"epoch": 0.38984482681961674,
|
| 30830 |
+
"grad_norm": 0.07131364196538925,
|
| 30831 |
+
"learning_rate": 0.0009813054810085748,
|
| 30832 |
+
"loss": 1.5416,
|
| 30833 |
+
"step": 8768
|
| 30834 |
+
},
|
| 30835 |
+
{
|
| 30836 |
+
"epoch": 0.3899337512782891,
|
| 30837 |
+
"grad_norm": 0.07210860401391983,
|
| 30838 |
+
"learning_rate": 0.00098129591382288,
|
| 30839 |
+
"loss": 1.5414,
|
| 30840 |
+
"step": 8770
|
| 30841 |
+
},
|
| 30842 |
+
{
|
| 30843 |
+
"epoch": 0.3900226757369615,
|
| 30844 |
+
"grad_norm": 0.07035349309444427,
|
| 30845 |
+
"learning_rate": 0.0009812863442364014,
|
| 30846 |
+
"loss": 1.5424,
|
| 30847 |
+
"step": 8772
|
| 30848 |
+
},
|
| 30849 |
+
{
|
| 30850 |
+
"epoch": 0.3901116001956338,
|
| 30851 |
+
"grad_norm": 0.07026588171720505,
|
| 30852 |
+
"learning_rate": 0.0009812767722491864,
|
| 30853 |
+
"loss": 1.5484,
|
| 30854 |
+
"step": 8774
|
| 30855 |
+
},
|
| 30856 |
+
{
|
| 30857 |
+
"epoch": 0.39020052465430616,
|
| 30858 |
+
"grad_norm": 0.07101976126432419,
|
| 30859 |
+
"learning_rate": 0.000981267197861283,
|
| 30860 |
+
"loss": 1.5419,
|
| 30861 |
+
"step": 8776
|
| 30862 |
+
},
|
| 30863 |
+
{
|
| 30864 |
+
"epoch": 0.3902894491129785,
|
| 30865 |
+
"grad_norm": 0.07162394374608994,
|
| 30866 |
+
"learning_rate": 0.0009812576210727392,
|
| 30867 |
+
"loss": 1.5482,
|
| 30868 |
+
"step": 8778
|
| 30869 |
+
},
|
| 30870 |
+
{
|
| 30871 |
+
"epoch": 0.3903783735716509,
|
| 30872 |
+
"grad_norm": 0.07118743658065796,
|
| 30873 |
+
"learning_rate": 0.0009812480418836024,
|
| 30874 |
+
"loss": 1.5401,
|
| 30875 |
+
"step": 8780
|
| 30876 |
+
},
|
| 30877 |
+
{
|
| 30878 |
+
"epoch": 0.39046729803032326,
|
| 30879 |
+
"grad_norm": 0.0704973116517067,
|
| 30880 |
+
"learning_rate": 0.0009812384602939203,
|
| 30881 |
+
"loss": 1.5413,
|
| 30882 |
+
"step": 8782
|
| 30883 |
+
},
|
| 30884 |
+
{
|
| 30885 |
+
"epoch": 0.3905562224889956,
|
| 30886 |
+
"grad_norm": 0.07269991189241409,
|
| 30887 |
+
"learning_rate": 0.000981228876303741,
|
| 30888 |
+
"loss": 1.5438,
|
| 30889 |
+
"step": 8784
|
| 30890 |
+
},
|
| 30891 |
+
{
|
| 30892 |
+
"epoch": 0.39064514694766794,
|
| 30893 |
+
"grad_norm": 0.07165592908859253,
|
| 30894 |
+
"learning_rate": 0.0009812192899131122,
|
| 30895 |
+
"loss": 1.548,
|
| 30896 |
+
"step": 8786
|
| 30897 |
+
},
|
| 30898 |
+
{
|
| 30899 |
+
"epoch": 0.3907340714063403,
|
| 30900 |
+
"grad_norm": 0.07121030986309052,
|
| 30901 |
+
"learning_rate": 0.0009812097011220816,
|
| 30902 |
+
"loss": 1.5459,
|
| 30903 |
+
"step": 8788
|
| 30904 |
+
},
|
| 30905 |
+
{
|
| 30906 |
+
"epoch": 0.3908229958650127,
|
| 30907 |
+
"grad_norm": 0.07146286964416504,
|
| 30908 |
+
"learning_rate": 0.0009812001099306974,
|
| 30909 |
+
"loss": 1.539,
|
| 30910 |
+
"step": 8790
|
| 30911 |
+
},
|
| 30912 |
+
{
|
| 30913 |
+
"epoch": 0.39091192032368505,
|
| 30914 |
+
"grad_norm": 0.07126247137784958,
|
| 30915 |
+
"learning_rate": 0.0009811905163390068,
|
| 30916 |
+
"loss": 1.5442,
|
| 30917 |
+
"step": 8792
|
| 30918 |
+
},
|
| 30919 |
+
{
|
| 30920 |
+
"epoch": 0.3910008447823574,
|
| 30921 |
+
"grad_norm": 0.06753192096948624,
|
| 30922 |
+
"learning_rate": 0.0009811809203470582,
|
| 30923 |
+
"loss": 1.546,
|
| 30924 |
+
"step": 8794
|
| 30925 |
+
},
|
| 30926 |
+
{
|
| 30927 |
+
"epoch": 0.3910897692410297,
|
| 30928 |
+
"grad_norm": 0.0699109360575676,
|
| 30929 |
+
"learning_rate": 0.0009811713219548992,
|
| 30930 |
+
"loss": 1.542,
|
| 30931 |
+
"step": 8796
|
| 30932 |
+
},
|
| 30933 |
+
{
|
| 30934 |
+
"epoch": 0.3911786936997021,
|
| 30935 |
+
"grad_norm": 0.0715608224272728,
|
| 30936 |
+
"learning_rate": 0.0009811617211625781,
|
| 30937 |
+
"loss": 1.5429,
|
| 30938 |
+
"step": 8798
|
| 30939 |
+
},
|
| 30940 |
+
{
|
| 30941 |
+
"epoch": 0.39126761815837446,
|
| 30942 |
+
"grad_norm": 0.07075057178735733,
|
| 30943 |
+
"learning_rate": 0.0009811521179701422,
|
| 30944 |
+
"loss": 1.5433,
|
| 30945 |
+
"step": 8800
|
| 30946 |
+
},
|
| 30947 |
+
{
|
| 30948 |
+
"epoch": 0.39135654261704683,
|
| 30949 |
+
"grad_norm": 0.07018981128931046,
|
| 30950 |
+
"learning_rate": 0.0009811425123776398,
|
| 30951 |
+
"loss": 1.5432,
|
| 30952 |
+
"step": 8802
|
| 30953 |
+
},
|
| 30954 |
+
{
|
| 30955 |
+
"epoch": 0.3914454670757192,
|
| 30956 |
+
"grad_norm": 0.07095321267843246,
|
| 30957 |
+
"learning_rate": 0.0009811329043851185,
|
| 30958 |
+
"loss": 1.5474,
|
| 30959 |
+
"step": 8804
|
| 30960 |
+
},
|
| 30961 |
+
{
|
| 30962 |
+
"epoch": 0.3915343915343915,
|
| 30963 |
+
"grad_norm": 0.07263996452093124,
|
| 30964 |
+
"learning_rate": 0.0009811232939926267,
|
| 30965 |
+
"loss": 1.5394,
|
| 30966 |
+
"step": 8806
|
| 30967 |
+
},
|
| 30968 |
+
{
|
| 30969 |
+
"epoch": 0.3916233159930639,
|
| 30970 |
+
"grad_norm": 0.07189413160085678,
|
| 30971 |
+
"learning_rate": 0.0009811136812002117,
|
| 30972 |
+
"loss": 1.5441,
|
| 30973 |
+
"step": 8808
|
| 30974 |
+
},
|
| 30975 |
+
{
|
| 30976 |
+
"epoch": 0.39171224045173625,
|
| 30977 |
+
"grad_norm": 0.0719655454158783,
|
| 30978 |
+
"learning_rate": 0.000981104066007922,
|
| 30979 |
+
"loss": 1.545,
|
| 30980 |
+
"step": 8810
|
| 30981 |
+
},
|
| 30982 |
+
{
|
| 30983 |
+
"epoch": 0.3918011649104086,
|
| 30984 |
+
"grad_norm": 0.0707225650548935,
|
| 30985 |
+
"learning_rate": 0.0009810944484158052,
|
| 30986 |
+
"loss": 1.5435,
|
| 30987 |
+
"step": 8812
|
| 30988 |
+
},
|
| 30989 |
+
{
|
| 30990 |
+
"epoch": 0.391890089369081,
|
| 30991 |
+
"grad_norm": 0.07110429555177689,
|
| 30992 |
+
"learning_rate": 0.0009810848284239097,
|
| 30993 |
+
"loss": 1.5402,
|
| 30994 |
+
"step": 8814
|
| 30995 |
+
},
|
| 30996 |
+
{
|
| 30997 |
+
"epoch": 0.3919790138277533,
|
| 30998 |
+
"grad_norm": 0.07303628325462341,
|
| 30999 |
+
"learning_rate": 0.000981075206032283,
|
| 31000 |
+
"loss": 1.5408,
|
| 31001 |
+
"step": 8816
|
| 31002 |
+
},
|
| 31003 |
+
{
|
| 31004 |
+
"epoch": 0.39206793828642567,
|
| 31005 |
+
"grad_norm": 0.07266764342784882,
|
| 31006 |
+
"learning_rate": 0.0009810655812409735,
|
| 31007 |
+
"loss": 1.5399,
|
| 31008 |
+
"step": 8818
|
| 31009 |
+
},
|
| 31010 |
+
{
|
| 31011 |
+
"epoch": 0.39215686274509803,
|
| 31012 |
+
"grad_norm": 0.07289233058691025,
|
| 31013 |
+
"learning_rate": 0.0009810559540500289,
|
| 31014 |
+
"loss": 1.5437,
|
| 31015 |
+
"step": 8820
|
| 31016 |
+
},
|
| 31017 |
+
{
|
| 31018 |
+
"epoch": 0.3922457872037704,
|
| 31019 |
+
"grad_norm": 0.07148449867963791,
|
| 31020 |
+
"learning_rate": 0.0009810463244594974,
|
| 31021 |
+
"loss": 1.5445,
|
| 31022 |
+
"step": 8822
|
| 31023 |
+
},
|
| 31024 |
+
{
|
| 31025 |
+
"epoch": 0.39233471166244277,
|
| 31026 |
+
"grad_norm": 0.0700794905424118,
|
| 31027 |
+
"learning_rate": 0.0009810366924694269,
|
| 31028 |
+
"loss": 1.5361,
|
| 31029 |
+
"step": 8824
|
| 31030 |
+
},
|
| 31031 |
+
{
|
| 31032 |
+
"epoch": 0.39242363612111514,
|
| 31033 |
+
"grad_norm": 0.07210569828748703,
|
| 31034 |
+
"learning_rate": 0.0009810270580798656,
|
| 31035 |
+
"loss": 1.5488,
|
| 31036 |
+
"step": 8826
|
| 31037 |
+
},
|
| 31038 |
+
{
|
| 31039 |
+
"epoch": 0.39251256057978745,
|
| 31040 |
+
"grad_norm": 0.07074844092130661,
|
| 31041 |
+
"learning_rate": 0.0009810174212908614,
|
| 31042 |
+
"loss": 1.5454,
|
| 31043 |
+
"step": 8828
|
| 31044 |
+
},
|
| 31045 |
+
{
|
| 31046 |
+
"epoch": 0.3926014850384598,
|
| 31047 |
+
"grad_norm": 0.07008279860019684,
|
| 31048 |
+
"learning_rate": 0.0009810077821024625,
|
| 31049 |
+
"loss": 1.5463,
|
| 31050 |
+
"step": 8830
|
| 31051 |
+
},
|
| 31052 |
+
{
|
| 31053 |
+
"epoch": 0.3926904094971322,
|
| 31054 |
+
"grad_norm": 0.07124333083629608,
|
| 31055 |
+
"learning_rate": 0.0009809981405147171,
|
| 31056 |
+
"loss": 1.5415,
|
| 31057 |
+
"step": 8832
|
| 31058 |
+
},
|
| 31059 |
+
{
|
| 31060 |
+
"epoch": 0.39277933395580455,
|
| 31061 |
+
"grad_norm": 0.06977679580450058,
|
| 31062 |
+
"learning_rate": 0.000980988496527673,
|
| 31063 |
+
"loss": 1.5459,
|
| 31064 |
+
"step": 8834
|
| 31065 |
+
},
|
| 31066 |
+
{
|
| 31067 |
+
"epoch": 0.3928682584144769,
|
| 31068 |
+
"grad_norm": 0.0721663162112236,
|
| 31069 |
+
"learning_rate": 0.0009809788501413783,
|
| 31070 |
+
"loss": 1.5433,
|
| 31071 |
+
"step": 8836
|
| 31072 |
+
},
|
| 31073 |
+
{
|
| 31074 |
+
"epoch": 0.39295718287314924,
|
| 31075 |
+
"grad_norm": 0.07385817915201187,
|
| 31076 |
+
"learning_rate": 0.0009809692013558816,
|
| 31077 |
+
"loss": 1.545,
|
| 31078 |
+
"step": 8838
|
| 31079 |
+
},
|
| 31080 |
+
{
|
| 31081 |
+
"epoch": 0.3930461073318216,
|
| 31082 |
+
"grad_norm": 0.07491130381822586,
|
| 31083 |
+
"learning_rate": 0.0009809595501712302,
|
| 31084 |
+
"loss": 1.5419,
|
| 31085 |
+
"step": 8840
|
| 31086 |
+
},
|
| 31087 |
+
{
|
| 31088 |
+
"epoch": 0.39313503179049397,
|
| 31089 |
+
"grad_norm": 0.07294177263975143,
|
| 31090 |
+
"learning_rate": 0.0009809498965874732,
|
| 31091 |
+
"loss": 1.5461,
|
| 31092 |
+
"step": 8842
|
| 31093 |
+
},
|
| 31094 |
+
{
|
| 31095 |
+
"epoch": 0.39322395624916634,
|
| 31096 |
+
"grad_norm": 0.07299856841564178,
|
| 31097 |
+
"learning_rate": 0.000980940240604658,
|
| 31098 |
+
"loss": 1.5455,
|
| 31099 |
+
"step": 8844
|
| 31100 |
+
},
|
| 31101 |
+
{
|
| 31102 |
+
"epoch": 0.3933128807078387,
|
| 31103 |
+
"grad_norm": 0.07228664308786392,
|
| 31104 |
+
"learning_rate": 0.0009809305822228332,
|
| 31105 |
+
"loss": 1.5422,
|
| 31106 |
+
"step": 8846
|
| 31107 |
+
},
|
| 31108 |
+
{
|
| 31109 |
+
"epoch": 0.3934018051665111,
|
| 31110 |
+
"grad_norm": 0.07752872258424759,
|
| 31111 |
+
"learning_rate": 0.0009809209214420467,
|
| 31112 |
+
"loss": 1.5404,
|
| 31113 |
+
"step": 8848
|
| 31114 |
+
},
|
| 31115 |
+
{
|
| 31116 |
+
"epoch": 0.3934907296251834,
|
| 31117 |
+
"grad_norm": 0.07412002980709076,
|
| 31118 |
+
"learning_rate": 0.000980911258262347,
|
| 31119 |
+
"loss": 1.5446,
|
| 31120 |
+
"step": 8850
|
| 31121 |
+
},
|
| 31122 |
+
{
|
| 31123 |
+
"epoch": 0.39357965408385576,
|
| 31124 |
+
"grad_norm": 0.06998312473297119,
|
| 31125 |
+
"learning_rate": 0.0009809015926837818,
|
| 31126 |
+
"loss": 1.5454,
|
| 31127 |
+
"step": 8852
|
| 31128 |
+
},
|
| 31129 |
+
{
|
| 31130 |
+
"epoch": 0.3936685785425281,
|
| 31131 |
+
"grad_norm": 0.0724993646144867,
|
| 31132 |
+
"learning_rate": 0.0009808919247064,
|
| 31133 |
+
"loss": 1.5369,
|
| 31134 |
+
"step": 8854
|
| 31135 |
+
},
|
| 31136 |
+
{
|
| 31137 |
+
"epoch": 0.3937575030012005,
|
| 31138 |
+
"grad_norm": 0.0712260976433754,
|
| 31139 |
+
"learning_rate": 0.0009808822543302493,
|
| 31140 |
+
"loss": 1.544,
|
| 31141 |
+
"step": 8856
|
| 31142 |
+
},
|
| 31143 |
+
{
|
| 31144 |
+
"epoch": 0.39384642745987286,
|
| 31145 |
+
"grad_norm": 0.07351220399141312,
|
| 31146 |
+
"learning_rate": 0.0009808725815553781,
|
| 31147 |
+
"loss": 1.5481,
|
| 31148 |
+
"step": 8858
|
| 31149 |
+
},
|
| 31150 |
+
{
|
| 31151 |
+
"epoch": 0.3939353519185452,
|
| 31152 |
+
"grad_norm": 0.07386364787817001,
|
| 31153 |
+
"learning_rate": 0.0009808629063818347,
|
| 31154 |
+
"loss": 1.5355,
|
| 31155 |
+
"step": 8860
|
| 31156 |
+
},
|
| 31157 |
+
{
|
| 31158 |
+
"epoch": 0.39402427637721754,
|
| 31159 |
+
"grad_norm": 0.0700744241476059,
|
| 31160 |
+
"learning_rate": 0.000980853228809667,
|
| 31161 |
+
"loss": 1.5344,
|
| 31162 |
+
"step": 8862
|
| 31163 |
+
},
|
| 31164 |
+
{
|
| 31165 |
+
"epoch": 0.3941132008358899,
|
| 31166 |
+
"grad_norm": 0.06920643895864487,
|
| 31167 |
+
"learning_rate": 0.0009808435488389239,
|
| 31168 |
+
"loss": 1.5445,
|
| 31169 |
+
"step": 8864
|
| 31170 |
+
},
|
| 31171 |
+
{
|
| 31172 |
+
"epoch": 0.3942021252945623,
|
| 31173 |
+
"grad_norm": 0.07501189410686493,
|
| 31174 |
+
"learning_rate": 0.0009808338664696532,
|
| 31175 |
+
"loss": 1.545,
|
| 31176 |
+
"step": 8866
|
| 31177 |
+
},
|
| 31178 |
+
{
|
| 31179 |
+
"epoch": 0.39429104975323465,
|
| 31180 |
+
"grad_norm": 0.07265200465917587,
|
| 31181 |
+
"learning_rate": 0.0009808241817019035,
|
| 31182 |
+
"loss": 1.5478,
|
| 31183 |
+
"step": 8868
|
| 31184 |
+
},
|
| 31185 |
+
{
|
| 31186 |
+
"epoch": 0.39437997421190696,
|
| 31187 |
+
"grad_norm": 0.07012397050857544,
|
| 31188 |
+
"learning_rate": 0.000980814494535723,
|
| 31189 |
+
"loss": 1.5415,
|
| 31190 |
+
"step": 8870
|
| 31191 |
+
},
|
| 31192 |
+
{
|
| 31193 |
+
"epoch": 0.3944688986705793,
|
| 31194 |
+
"grad_norm": 0.07346207648515701,
|
| 31195 |
+
"learning_rate": 0.0009808048049711597,
|
| 31196 |
+
"loss": 1.5456,
|
| 31197 |
+
"step": 8872
|
| 31198 |
+
},
|
| 31199 |
+
{
|
| 31200 |
+
"epoch": 0.3945578231292517,
|
| 31201 |
+
"grad_norm": 0.06875230371952057,
|
| 31202 |
+
"learning_rate": 0.0009807951130082625,
|
| 31203 |
+
"loss": 1.5495,
|
| 31204 |
+
"step": 8874
|
| 31205 |
+
},
|
| 31206 |
+
{
|
| 31207 |
+
"epoch": 0.39464674758792406,
|
| 31208 |
+
"grad_norm": 0.07028649747371674,
|
| 31209 |
+
"learning_rate": 0.0009807854186470793,
|
| 31210 |
+
"loss": 1.5382,
|
| 31211 |
+
"step": 8876
|
| 31212 |
+
},
|
| 31213 |
+
{
|
| 31214 |
+
"epoch": 0.39473567204659643,
|
| 31215 |
+
"grad_norm": 0.07126179337501526,
|
| 31216 |
+
"learning_rate": 0.0009807757218876587,
|
| 31217 |
+
"loss": 1.5412,
|
| 31218 |
+
"step": 8878
|
| 31219 |
+
},
|
| 31220 |
+
{
|
| 31221 |
+
"epoch": 0.3948245965052688,
|
| 31222 |
+
"grad_norm": 0.07412000000476837,
|
| 31223 |
+
"learning_rate": 0.000980766022730049,
|
| 31224 |
+
"loss": 1.5441,
|
| 31225 |
+
"step": 8880
|
| 31226 |
+
},
|
| 31227 |
+
{
|
| 31228 |
+
"epoch": 0.3949135209639411,
|
| 31229 |
+
"grad_norm": 0.06931138783693314,
|
| 31230 |
+
"learning_rate": 0.0009807563211742987,
|
| 31231 |
+
"loss": 1.5408,
|
| 31232 |
+
"step": 8882
|
| 31233 |
+
},
|
| 31234 |
+
{
|
| 31235 |
+
"epoch": 0.3950024454226135,
|
| 31236 |
+
"grad_norm": 0.06960853189229965,
|
| 31237 |
+
"learning_rate": 0.0009807466172204557,
|
| 31238 |
+
"loss": 1.5467,
|
| 31239 |
+
"step": 8884
|
| 31240 |
+
},
|
| 31241 |
+
{
|
| 31242 |
+
"epoch": 0.39509136988128585,
|
| 31243 |
+
"grad_norm": 0.07029549777507782,
|
| 31244 |
+
"learning_rate": 0.000980736910868569,
|
| 31245 |
+
"loss": 1.5485,
|
| 31246 |
+
"step": 8886
|
| 31247 |
+
},
|
| 31248 |
+
{
|
| 31249 |
+
"epoch": 0.3951802943399582,
|
| 31250 |
+
"grad_norm": 0.070205919444561,
|
| 31251 |
+
"learning_rate": 0.0009807272021186868,
|
| 31252 |
+
"loss": 1.544,
|
| 31253 |
+
"step": 8888
|
| 31254 |
+
},
|
| 31255 |
+
{
|
| 31256 |
+
"epoch": 0.3952692187986306,
|
| 31257 |
+
"grad_norm": 0.07060658931732178,
|
| 31258 |
+
"learning_rate": 0.0009807174909708576,
|
| 31259 |
+
"loss": 1.542,
|
| 31260 |
+
"step": 8890
|
| 31261 |
+
},
|
| 31262 |
+
{
|
| 31263 |
+
"epoch": 0.3953581432573029,
|
| 31264 |
+
"grad_norm": 0.070548877120018,
|
| 31265 |
+
"learning_rate": 0.0009807077774251296,
|
| 31266 |
+
"loss": 1.5451,
|
| 31267 |
+
"step": 8892
|
| 31268 |
+
},
|
| 31269 |
+
{
|
| 31270 |
+
"epoch": 0.39544706771597526,
|
| 31271 |
+
"grad_norm": 0.06973063200712204,
|
| 31272 |
+
"learning_rate": 0.0009806980614815515,
|
| 31273 |
+
"loss": 1.5422,
|
| 31274 |
+
"step": 8894
|
| 31275 |
+
},
|
| 31276 |
+
{
|
| 31277 |
+
"epoch": 0.39553599217464763,
|
| 31278 |
+
"grad_norm": 0.07283268868923187,
|
| 31279 |
+
"learning_rate": 0.0009806883431401718,
|
| 31280 |
+
"loss": 1.5433,
|
| 31281 |
+
"step": 8896
|
| 31282 |
+
},
|
| 31283 |
+
{
|
| 31284 |
+
"epoch": 0.39562491663332,
|
| 31285 |
+
"grad_norm": 0.07797042280435562,
|
| 31286 |
+
"learning_rate": 0.0009806786224010387,
|
| 31287 |
+
"loss": 1.549,
|
| 31288 |
+
"step": 8898
|
| 31289 |
+
},
|
| 31290 |
+
{
|
| 31291 |
+
"epoch": 0.39571384109199237,
|
| 31292 |
+
"grad_norm": 0.07383356988430023,
|
| 31293 |
+
"learning_rate": 0.000980668899264201,
|
| 31294 |
+
"loss": 1.5419,
|
| 31295 |
+
"step": 8900
|
| 31296 |
+
},
|
| 31297 |
+
{
|
| 31298 |
+
"epoch": 0.39580276555066474,
|
| 31299 |
+
"grad_norm": 0.08115527033805847,
|
| 31300 |
+
"learning_rate": 0.0009806591737297069,
|
| 31301 |
+
"loss": 1.5418,
|
| 31302 |
+
"step": 8902
|
| 31303 |
+
},
|
| 31304 |
+
{
|
| 31305 |
+
"epoch": 0.39589169000933705,
|
| 31306 |
+
"grad_norm": 0.0753684863448143,
|
| 31307 |
+
"learning_rate": 0.0009806494457976049,
|
| 31308 |
+
"loss": 1.5422,
|
| 31309 |
+
"step": 8904
|
| 31310 |
+
},
|
| 31311 |
+
{
|
| 31312 |
+
"epoch": 0.3959806144680094,
|
| 31313 |
+
"grad_norm": 0.07579310983419418,
|
| 31314 |
+
"learning_rate": 0.0009806397154679437,
|
| 31315 |
+
"loss": 1.5431,
|
| 31316 |
+
"step": 8906
|
| 31317 |
+
},
|
| 31318 |
+
{
|
| 31319 |
+
"epoch": 0.3960695389266818,
|
| 31320 |
+
"grad_norm": 0.0755925178527832,
|
| 31321 |
+
"learning_rate": 0.0009806299827407722,
|
| 31322 |
+
"loss": 1.5424,
|
| 31323 |
+
"step": 8908
|
| 31324 |
+
},
|
| 31325 |
+
{
|
| 31326 |
+
"epoch": 0.39615846338535415,
|
| 31327 |
+
"grad_norm": 0.07460242509841919,
|
| 31328 |
+
"learning_rate": 0.0009806202476161383,
|
| 31329 |
+
"loss": 1.5464,
|
| 31330 |
+
"step": 8910
|
| 31331 |
+
},
|
| 31332 |
+
{
|
| 31333 |
+
"epoch": 0.3962473878440265,
|
| 31334 |
+
"grad_norm": 0.07479722797870636,
|
| 31335 |
+
"learning_rate": 0.000980610510094091,
|
| 31336 |
+
"loss": 1.5461,
|
| 31337 |
+
"step": 8912
|
| 31338 |
+
},
|
| 31339 |
+
{
|
| 31340 |
+
"epoch": 0.39633631230269883,
|
| 31341 |
+
"grad_norm": 0.0737161934375763,
|
| 31342 |
+
"learning_rate": 0.0009806007701746785,
|
| 31343 |
+
"loss": 1.5415,
|
| 31344 |
+
"step": 8914
|
| 31345 |
+
},
|
| 31346 |
+
{
|
| 31347 |
+
"epoch": 0.3964252367613712,
|
| 31348 |
+
"grad_norm": 0.07179709523916245,
|
| 31349 |
+
"learning_rate": 0.0009805910278579495,
|
| 31350 |
+
"loss": 1.5388,
|
| 31351 |
+
"step": 8916
|
| 31352 |
+
},
|
| 31353 |
+
{
|
| 31354 |
+
"epoch": 0.39651416122004357,
|
| 31355 |
+
"grad_norm": 0.07436472177505493,
|
| 31356 |
+
"learning_rate": 0.0009805812831439528,
|
| 31357 |
+
"loss": 1.5408,
|
| 31358 |
+
"step": 8918
|
| 31359 |
+
},
|
| 31360 |
+
{
|
| 31361 |
+
"epoch": 0.39660308567871594,
|
| 31362 |
+
"grad_norm": 0.07385652512311935,
|
| 31363 |
+
"learning_rate": 0.000980571536032737,
|
| 31364 |
+
"loss": 1.5375,
|
| 31365 |
+
"step": 8920
|
| 31366 |
+
},
|
| 31367 |
+
{
|
| 31368 |
+
"epoch": 0.3966920101373883,
|
| 31369 |
+
"grad_norm": 0.07151640951633453,
|
| 31370 |
+
"learning_rate": 0.0009805617865243504,
|
| 31371 |
+
"loss": 1.5407,
|
| 31372 |
+
"step": 8922
|
| 31373 |
+
},
|
| 31374 |
+
{
|
| 31375 |
+
"epoch": 0.3967809345960606,
|
| 31376 |
+
"grad_norm": 0.07092677056789398,
|
| 31377 |
+
"learning_rate": 0.000980552034618842,
|
| 31378 |
+
"loss": 1.5415,
|
| 31379 |
+
"step": 8924
|
| 31380 |
+
},
|
| 31381 |
+
{
|
| 31382 |
+
"epoch": 0.396869859054733,
|
| 31383 |
+
"grad_norm": 0.0716884657740593,
|
| 31384 |
+
"learning_rate": 0.0009805422803162603,
|
| 31385 |
+
"loss": 1.5341,
|
| 31386 |
+
"step": 8926
|
| 31387 |
+
},
|
| 31388 |
+
{
|
| 31389 |
+
"epoch": 0.39695878351340536,
|
| 31390 |
+
"grad_norm": 0.07016546279191971,
|
| 31391 |
+
"learning_rate": 0.0009805325236166538,
|
| 31392 |
+
"loss": 1.5385,
|
| 31393 |
+
"step": 8928
|
| 31394 |
+
},
|
| 31395 |
+
{
|
| 31396 |
+
"epoch": 0.3970477079720777,
|
| 31397 |
+
"grad_norm": 0.07406767457723618,
|
| 31398 |
+
"learning_rate": 0.0009805227645200713,
|
| 31399 |
+
"loss": 1.5411,
|
| 31400 |
+
"step": 8930
|
| 31401 |
+
},
|
| 31402 |
+
{
|
| 31403 |
+
"epoch": 0.3971366324307501,
|
| 31404 |
+
"grad_norm": 0.0695648342370987,
|
| 31405 |
+
"learning_rate": 0.0009805130030265616,
|
| 31406 |
+
"loss": 1.5376,
|
| 31407 |
+
"step": 8932
|
| 31408 |
+
},
|
| 31409 |
+
{
|
| 31410 |
+
"epoch": 0.39722555688942246,
|
| 31411 |
+
"grad_norm": 0.07370419800281525,
|
| 31412 |
+
"learning_rate": 0.0009805032391361733,
|
| 31413 |
+
"loss": 1.5464,
|
| 31414 |
+
"step": 8934
|
| 31415 |
+
},
|
| 31416 |
+
{
|
| 31417 |
+
"epoch": 0.3973144813480948,
|
| 31418 |
+
"grad_norm": 0.07572710514068604,
|
| 31419 |
+
"learning_rate": 0.0009804934728489551,
|
| 31420 |
+
"loss": 1.5397,
|
| 31421 |
+
"step": 8936
|
| 31422 |
+
},
|
| 31423 |
+
{
|
| 31424 |
+
"epoch": 0.39740340580676714,
|
| 31425 |
+
"grad_norm": 0.07438923418521881,
|
| 31426 |
+
"learning_rate": 0.0009804837041649556,
|
| 31427 |
+
"loss": 1.5395,
|
| 31428 |
+
"step": 8938
|
| 31429 |
+
},
|
| 31430 |
+
{
|
| 31431 |
+
"epoch": 0.3974923302654395,
|
| 31432 |
+
"grad_norm": 0.07321964204311371,
|
| 31433 |
+
"learning_rate": 0.0009804739330842235,
|
| 31434 |
+
"loss": 1.5405,
|
| 31435 |
+
"step": 8940
|
| 31436 |
+
},
|
| 31437 |
+
{
|
| 31438 |
+
"epoch": 0.3975812547241119,
|
| 31439 |
+
"grad_norm": 0.07280636578798294,
|
| 31440 |
+
"learning_rate": 0.000980464159606808,
|
| 31441 |
+
"loss": 1.5442,
|
| 31442 |
+
"step": 8942
|
| 31443 |
+
},
|
| 31444 |
+
{
|
| 31445 |
+
"epoch": 0.39767017918278424,
|
| 31446 |
+
"grad_norm": 0.07473626732826233,
|
| 31447 |
+
"learning_rate": 0.0009804543837327573,
|
| 31448 |
+
"loss": 1.5427,
|
| 31449 |
+
"step": 8944
|
| 31450 |
+
},
|
| 31451 |
+
{
|
| 31452 |
+
"epoch": 0.39775910364145656,
|
| 31453 |
+
"grad_norm": 0.06971806287765503,
|
| 31454 |
+
"learning_rate": 0.0009804446054621203,
|
| 31455 |
+
"loss": 1.5431,
|
| 31456 |
+
"step": 8946
|
| 31457 |
+
},
|
| 31458 |
+
{
|
| 31459 |
+
"epoch": 0.3978480281001289,
|
| 31460 |
+
"grad_norm": 0.07413081079721451,
|
| 31461 |
+
"learning_rate": 0.0009804348247949462,
|
| 31462 |
+
"loss": 1.5433,
|
| 31463 |
+
"step": 8948
|
| 31464 |
+
},
|
| 31465 |
+
{
|
| 31466 |
+
"epoch": 0.3979369525588013,
|
| 31467 |
+
"grad_norm": 0.07259391993284225,
|
| 31468 |
+
"learning_rate": 0.0009804250417312832,
|
| 31469 |
+
"loss": 1.5446,
|
| 31470 |
+
"step": 8950
|
| 31471 |
+
},
|
| 31472 |
+
{
|
| 31473 |
+
"epoch": 0.39802587701747366,
|
| 31474 |
+
"grad_norm": 0.0727652907371521,
|
| 31475 |
+
"learning_rate": 0.0009804152562711804,
|
| 31476 |
+
"loss": 1.5387,
|
| 31477 |
+
"step": 8952
|
| 31478 |
+
},
|
| 31479 |
+
{
|
| 31480 |
+
"epoch": 0.39811480147614603,
|
| 31481 |
+
"grad_norm": 0.07275709509849548,
|
| 31482 |
+
"learning_rate": 0.0009804054684146865,
|
| 31483 |
+
"loss": 1.5379,
|
| 31484 |
+
"step": 8954
|
| 31485 |
+
},
|
| 31486 |
+
{
|
| 31487 |
+
"epoch": 0.3982037259348184,
|
| 31488 |
+
"grad_norm": 0.07379709184169769,
|
| 31489 |
+
"learning_rate": 0.0009803956781618505,
|
| 31490 |
+
"loss": 1.5357,
|
| 31491 |
+
"step": 8956
|
| 31492 |
+
},
|
| 31493 |
+
{
|
| 31494 |
+
"epoch": 0.3982926503934907,
|
| 31495 |
+
"grad_norm": 0.06894747912883759,
|
| 31496 |
+
"learning_rate": 0.000980385885512721,
|
| 31497 |
+
"loss": 1.544,
|
| 31498 |
+
"step": 8958
|
| 31499 |
+
},
|
| 31500 |
+
{
|
| 31501 |
+
"epoch": 0.3983815748521631,
|
| 31502 |
+
"grad_norm": 0.06997144967317581,
|
| 31503 |
+
"learning_rate": 0.000980376090467347,
|
| 31504 |
+
"loss": 1.5365,
|
| 31505 |
+
"step": 8960
|
| 31506 |
+
},
|
| 31507 |
+
{
|
| 31508 |
+
"epoch": 0.39847049931083545,
|
| 31509 |
+
"grad_norm": 0.07101274281740189,
|
| 31510 |
+
"learning_rate": 0.0009803662930257773,
|
| 31511 |
+
"loss": 1.5355,
|
| 31512 |
+
"step": 8962
|
| 31513 |
+
},
|
| 31514 |
+
{
|
| 31515 |
+
"epoch": 0.3985594237695078,
|
| 31516 |
+
"grad_norm": 0.07142171263694763,
|
| 31517 |
+
"learning_rate": 0.0009803564931880606,
|
| 31518 |
+
"loss": 1.5441,
|
| 31519 |
+
"step": 8964
|
| 31520 |
+
},
|
| 31521 |
+
{
|
| 31522 |
+
"epoch": 0.3986483482281802,
|
| 31523 |
+
"grad_norm": 0.06851017475128174,
|
| 31524 |
+
"learning_rate": 0.0009803466909542463,
|
| 31525 |
+
"loss": 1.5381,
|
| 31526 |
+
"step": 8966
|
| 31527 |
+
},
|
| 31528 |
+
{
|
| 31529 |
+
"epoch": 0.3987372726868525,
|
| 31530 |
+
"grad_norm": 0.06920833140611649,
|
| 31531 |
+
"learning_rate": 0.0009803368863243829,
|
| 31532 |
+
"loss": 1.5397,
|
| 31533 |
+
"step": 8968
|
| 31534 |
+
},
|
| 31535 |
+
{
|
| 31536 |
+
"epoch": 0.39882619714552486,
|
| 31537 |
+
"grad_norm": 0.07191641628742218,
|
| 31538 |
+
"learning_rate": 0.0009803270792985194,
|
| 31539 |
+
"loss": 1.541,
|
| 31540 |
+
"step": 8970
|
| 31541 |
+
},
|
| 31542 |
+
{
|
| 31543 |
+
"epoch": 0.39891512160419723,
|
| 31544 |
+
"grad_norm": 0.07115970551967621,
|
| 31545 |
+
"learning_rate": 0.0009803172698767046,
|
| 31546 |
+
"loss": 1.5363,
|
| 31547 |
+
"step": 8972
|
| 31548 |
+
},
|
| 31549 |
+
{
|
| 31550 |
+
"epoch": 0.3990040460628696,
|
| 31551 |
+
"grad_norm": 0.06930098682641983,
|
| 31552 |
+
"learning_rate": 0.0009803074580589874,
|
| 31553 |
+
"loss": 1.5372,
|
| 31554 |
+
"step": 8974
|
| 31555 |
+
},
|
| 31556 |
+
{
|
| 31557 |
+
"epoch": 0.39909297052154197,
|
| 31558 |
+
"grad_norm": 0.06866835802793503,
|
| 31559 |
+
"learning_rate": 0.0009802976438454173,
|
| 31560 |
+
"loss": 1.5429,
|
| 31561 |
+
"step": 8976
|
| 31562 |
+
},
|
| 31563 |
+
{
|
| 31564 |
+
"epoch": 0.39918189498021434,
|
| 31565 |
+
"grad_norm": 0.06998659670352936,
|
| 31566 |
+
"learning_rate": 0.0009802878272360423,
|
| 31567 |
+
"loss": 1.537,
|
| 31568 |
+
"step": 8978
|
| 31569 |
+
},
|
| 31570 |
+
{
|
| 31571 |
+
"epoch": 0.39927081943888665,
|
| 31572 |
+
"grad_norm": 0.07126189023256302,
|
| 31573 |
+
"learning_rate": 0.0009802780082309123,
|
| 31574 |
+
"loss": 1.5386,
|
| 31575 |
+
"step": 8980
|
| 31576 |
+
},
|
| 31577 |
+
{
|
| 31578 |
+
"epoch": 0.399359743897559,
|
| 31579 |
+
"grad_norm": 0.0688776820898056,
|
| 31580 |
+
"learning_rate": 0.0009802681868300758,
|
| 31581 |
+
"loss": 1.5404,
|
| 31582 |
+
"step": 8982
|
| 31583 |
+
},
|
| 31584 |
+
{
|
| 31585 |
+
"epoch": 0.3994486683562314,
|
| 31586 |
+
"grad_norm": 0.07415543496608734,
|
| 31587 |
+
"learning_rate": 0.0009802583630335818,
|
| 31588 |
+
"loss": 1.5461,
|
| 31589 |
+
"step": 8984
|
| 31590 |
+
},
|
| 31591 |
+
{
|
| 31592 |
+
"epoch": 0.39953759281490375,
|
| 31593 |
+
"grad_norm": 0.06990732252597809,
|
| 31594 |
+
"learning_rate": 0.0009802485368414792,
|
| 31595 |
+
"loss": 1.54,
|
| 31596 |
+
"step": 8986
|
| 31597 |
+
},
|
| 31598 |
+
{
|
| 31599 |
+
"epoch": 0.3996265172735761,
|
| 31600 |
+
"grad_norm": 0.06985145807266235,
|
| 31601 |
+
"learning_rate": 0.0009802387082538174,
|
| 31602 |
+
"loss": 1.5392,
|
| 31603 |
+
"step": 8988
|
| 31604 |
+
},
|
| 31605 |
+
{
|
| 31606 |
+
"epoch": 0.39971544173224843,
|
| 31607 |
+
"grad_norm": 0.07134158909320831,
|
| 31608 |
+
"learning_rate": 0.000980228877270645,
|
| 31609 |
+
"loss": 1.5432,
|
| 31610 |
+
"step": 8990
|
| 31611 |
+
},
|
| 31612 |
+
{
|
| 31613 |
+
"epoch": 0.3998043661909208,
|
| 31614 |
+
"grad_norm": 0.07347475737333298,
|
| 31615 |
+
"learning_rate": 0.0009802190438920116,
|
| 31616 |
+
"loss": 1.5434,
|
| 31617 |
+
"step": 8992
|
| 31618 |
+
},
|
| 31619 |
+
{
|
| 31620 |
+
"epoch": 0.39989329064959317,
|
| 31621 |
+
"grad_norm": 0.07028983533382416,
|
| 31622 |
+
"learning_rate": 0.0009802092081179657,
|
| 31623 |
+
"loss": 1.5385,
|
| 31624 |
+
"step": 8994
|
| 31625 |
+
},
|
| 31626 |
+
{
|
| 31627 |
+
"epoch": 0.39998221510826554,
|
| 31628 |
+
"grad_norm": 0.07134208083152771,
|
| 31629 |
+
"learning_rate": 0.0009801993699485566,
|
| 31630 |
+
"loss": 1.5446,
|
| 31631 |
+
"step": 8996
|
| 31632 |
+
},
|
| 31633 |
+
{
|
| 31634 |
+
"epoch": 0.4000711395669379,
|
| 31635 |
+
"grad_norm": 0.07063104212284088,
|
| 31636 |
+
"learning_rate": 0.000980189529383833,
|
| 31637 |
+
"loss": 1.5452,
|
| 31638 |
+
"step": 8998
|
| 31639 |
+
},
|
| 31640 |
+
{
|
| 31641 |
+
"epoch": 0.4001600640256102,
|
| 31642 |
+
"grad_norm": 0.06996823847293854,
|
| 31643 |
+
"learning_rate": 0.0009801796864238447,
|
| 31644 |
+
"loss": 1.5371,
|
| 31645 |
+
"step": 9000
|
| 31646 |
+
},
|
| 31647 |
+
{
|
| 31648 |
+
"epoch": 0.4001600640256102,
|
| 31649 |
+
"eval_loss": 1.51895010471344,
|
| 31650 |
+
"eval_runtime": 12.3629,
|
| 31651 |
+
"eval_samples_per_second": 558.929,
|
| 31652 |
+
"eval_steps_per_second": 69.886,
|
| 31653 |
+
"step": 9000
|
| 31654 |
}
|
| 31655 |
],
|
| 31656 |
"logging_steps": 2,
|
|
|
|
| 31670 |
"attributes": {}
|
| 31671 |
}
|
| 31672 |
},
|
| 31673 |
+
"total_flos": 1.925758108827648e+19,
|
| 31674 |
"train_batch_size": 768,
|
| 31675 |
"trial_name": null,
|
| 31676 |
"trial_params": null
|