Training in progress, step 8000, 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 sha256:5f8a3cc4fd3d109560b7a96fdbe13f14a6e27436a0d8c767e80e006e65b5b757
|
| 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 sha256:86699da39a3fab83cd8c5fdd36440d7f77f6445f7569eff311e9b0123ff619ce
|
| 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 sha256:5dafd50c49589ddc23b254f5413b0d2faf890d41ca48d64ead8a7084105e608a
|
| 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:fc42f55e2d0842057dc8ff559f0ec21b6e555cf7df0d60cb6c4bece9d16e8ae6
|
| 3 |
size 1000
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 1.5398365259170532,
|
| 3 |
"best_model_checkpoint": "model/chessformer-3/checkpoint-7000",
|
| 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,
|
|
@@ -24619,6 +24619,3522 @@
|
|
| 24619 |
"eval_samples_per_second": 534.768,
|
| 24620 |
"eval_steps_per_second": 66.865,
|
| 24621 |
"step": 7000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 24622 |
}
|
| 24623 |
],
|
| 24624 |
"logging_steps": 2,
|
|
@@ -24638,7 +28154,7 @@
|
|
| 24638 |
"attributes": {}
|
| 24639 |
}
|
| 24640 |
},
|
| 24641 |
-
"total_flos": 1.
|
| 24642 |
"train_batch_size": 768,
|
| 24643 |
"trial_name": null,
|
| 24644 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 1.5398365259170532,
|
| 3 |
"best_model_checkpoint": "model/chessformer-3/checkpoint-7000",
|
| 4 |
+
"epoch": 0.35569783468943134,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 8000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 24619 |
"eval_samples_per_second": 534.768,
|
| 24620 |
"eval_steps_per_second": 66.865,
|
| 24621 |
"step": 7000
|
| 24622 |
+
},
|
| 24623 |
+
{
|
| 24624 |
+
"epoch": 0.31132452981192477,
|
| 24625 |
+
"grad_norm": 0.08610174804925919,
|
| 24626 |
+
"learning_rate": 0.0009888111241591819,
|
| 24627 |
+
"loss": 1.5637,
|
| 24628 |
+
"step": 7002
|
| 24629 |
+
},
|
| 24630 |
+
{
|
| 24631 |
+
"epoch": 0.31141345427059713,
|
| 24632 |
+
"grad_norm": 0.08628767728805542,
|
| 24633 |
+
"learning_rate": 0.0009888036941112907,
|
| 24634 |
+
"loss": 1.5625,
|
| 24635 |
+
"step": 7004
|
| 24636 |
+
},
|
| 24637 |
+
{
|
| 24638 |
+
"epoch": 0.3115023787292695,
|
| 24639 |
+
"grad_norm": 0.08559640496969223,
|
| 24640 |
+
"learning_rate": 0.0009887962616251654,
|
| 24641 |
+
"loss": 1.5633,
|
| 24642 |
+
"step": 7006
|
| 24643 |
+
},
|
| 24644 |
+
{
|
| 24645 |
+
"epoch": 0.31159130318794187,
|
| 24646 |
+
"grad_norm": 0.08582499623298645,
|
| 24647 |
+
"learning_rate": 0.0009887888267008432,
|
| 24648 |
+
"loss": 1.5718,
|
| 24649 |
+
"step": 7008
|
| 24650 |
+
},
|
| 24651 |
+
{
|
| 24652 |
+
"epoch": 0.3116802276466142,
|
| 24653 |
+
"grad_norm": 0.08453387767076492,
|
| 24654 |
+
"learning_rate": 0.0009887813893383612,
|
| 24655 |
+
"loss": 1.5644,
|
| 24656 |
+
"step": 7010
|
| 24657 |
+
},
|
| 24658 |
+
{
|
| 24659 |
+
"epoch": 0.31176915210528655,
|
| 24660 |
+
"grad_norm": 0.0880834087729454,
|
| 24661 |
+
"learning_rate": 0.0009887739495377564,
|
| 24662 |
+
"loss": 1.5675,
|
| 24663 |
+
"step": 7012
|
| 24664 |
+
},
|
| 24665 |
+
{
|
| 24666 |
+
"epoch": 0.3118580765639589,
|
| 24667 |
+
"grad_norm": 0.08896942436695099,
|
| 24668 |
+
"learning_rate": 0.0009887665072990661,
|
| 24669 |
+
"loss": 1.5601,
|
| 24670 |
+
"step": 7014
|
| 24671 |
+
},
|
| 24672 |
+
{
|
| 24673 |
+
"epoch": 0.3119470010226313,
|
| 24674 |
+
"grad_norm": 0.08615525811910629,
|
| 24675 |
+
"learning_rate": 0.0009887590626223273,
|
| 24676 |
+
"loss": 1.5647,
|
| 24677 |
+
"step": 7016
|
| 24678 |
+
},
|
| 24679 |
+
{
|
| 24680 |
+
"epoch": 0.31203592548130366,
|
| 24681 |
+
"grad_norm": 0.08899851888418198,
|
| 24682 |
+
"learning_rate": 0.0009887516155075772,
|
| 24683 |
+
"loss": 1.5663,
|
| 24684 |
+
"step": 7018
|
| 24685 |
+
},
|
| 24686 |
+
{
|
| 24687 |
+
"epoch": 0.31212484993997597,
|
| 24688 |
+
"grad_norm": 0.08589426428079605,
|
| 24689 |
+
"learning_rate": 0.000988744165954853,
|
| 24690 |
+
"loss": 1.5629,
|
| 24691 |
+
"step": 7020
|
| 24692 |
+
},
|
| 24693 |
+
{
|
| 24694 |
+
"epoch": 0.31221377439864834,
|
| 24695 |
+
"grad_norm": 0.0895439013838768,
|
| 24696 |
+
"learning_rate": 0.0009887367139641916,
|
| 24697 |
+
"loss": 1.5655,
|
| 24698 |
+
"step": 7022
|
| 24699 |
+
},
|
| 24700 |
+
{
|
| 24701 |
+
"epoch": 0.3123026988573207,
|
| 24702 |
+
"grad_norm": 0.08594179153442383,
|
| 24703 |
+
"learning_rate": 0.0009887292595356304,
|
| 24704 |
+
"loss": 1.5574,
|
| 24705 |
+
"step": 7024
|
| 24706 |
+
},
|
| 24707 |
+
{
|
| 24708 |
+
"epoch": 0.31239162331599307,
|
| 24709 |
+
"grad_norm": 0.08816447108983994,
|
| 24710 |
+
"learning_rate": 0.0009887218026692066,
|
| 24711 |
+
"loss": 1.5686,
|
| 24712 |
+
"step": 7026
|
| 24713 |
+
},
|
| 24714 |
+
{
|
| 24715 |
+
"epoch": 0.31248054777466544,
|
| 24716 |
+
"grad_norm": 0.08560863137245178,
|
| 24717 |
+
"learning_rate": 0.0009887143433649573,
|
| 24718 |
+
"loss": 1.5662,
|
| 24719 |
+
"step": 7028
|
| 24720 |
+
},
|
| 24721 |
+
{
|
| 24722 |
+
"epoch": 0.31256947223333775,
|
| 24723 |
+
"grad_norm": 0.0848335549235344,
|
| 24724 |
+
"learning_rate": 0.0009887068816229196,
|
| 24725 |
+
"loss": 1.5647,
|
| 24726 |
+
"step": 7030
|
| 24727 |
+
},
|
| 24728 |
+
{
|
| 24729 |
+
"epoch": 0.3126583966920101,
|
| 24730 |
+
"grad_norm": 0.08963989466428757,
|
| 24731 |
+
"learning_rate": 0.000988699417443131,
|
| 24732 |
+
"loss": 1.5681,
|
| 24733 |
+
"step": 7032
|
| 24734 |
+
},
|
| 24735 |
+
{
|
| 24736 |
+
"epoch": 0.3127473211506825,
|
| 24737 |
+
"grad_norm": 0.08423878997564316,
|
| 24738 |
+
"learning_rate": 0.0009886919508256286,
|
| 24739 |
+
"loss": 1.5634,
|
| 24740 |
+
"step": 7034
|
| 24741 |
+
},
|
| 24742 |
+
{
|
| 24743 |
+
"epoch": 0.31283624560935486,
|
| 24744 |
+
"grad_norm": 0.08075986057519913,
|
| 24745 |
+
"learning_rate": 0.0009886844817704496,
|
| 24746 |
+
"loss": 1.5645,
|
| 24747 |
+
"step": 7036
|
| 24748 |
+
},
|
| 24749 |
+
{
|
| 24750 |
+
"epoch": 0.3129251700680272,
|
| 24751 |
+
"grad_norm": 0.07971926033496857,
|
| 24752 |
+
"learning_rate": 0.0009886770102776313,
|
| 24753 |
+
"loss": 1.5615,
|
| 24754 |
+
"step": 7038
|
| 24755 |
+
},
|
| 24756 |
+
{
|
| 24757 |
+
"epoch": 0.3130140945266996,
|
| 24758 |
+
"grad_norm": 0.08668291568756104,
|
| 24759 |
+
"learning_rate": 0.0009886695363472107,
|
| 24760 |
+
"loss": 1.5657,
|
| 24761 |
+
"step": 7040
|
| 24762 |
+
},
|
| 24763 |
+
{
|
| 24764 |
+
"epoch": 0.3131030189853719,
|
| 24765 |
+
"grad_norm": 0.08376525342464447,
|
| 24766 |
+
"learning_rate": 0.0009886620599792258,
|
| 24767 |
+
"loss": 1.5634,
|
| 24768 |
+
"step": 7042
|
| 24769 |
+
},
|
| 24770 |
+
{
|
| 24771 |
+
"epoch": 0.3131919434440443,
|
| 24772 |
+
"grad_norm": 0.0836341604590416,
|
| 24773 |
+
"learning_rate": 0.0009886545811737132,
|
| 24774 |
+
"loss": 1.5597,
|
| 24775 |
+
"step": 7044
|
| 24776 |
+
},
|
| 24777 |
+
{
|
| 24778 |
+
"epoch": 0.31328086790271664,
|
| 24779 |
+
"grad_norm": 0.08614721894264221,
|
| 24780 |
+
"learning_rate": 0.0009886470999307102,
|
| 24781 |
+
"loss": 1.5616,
|
| 24782 |
+
"step": 7046
|
| 24783 |
+
},
|
| 24784 |
+
{
|
| 24785 |
+
"epoch": 0.313369792361389,
|
| 24786 |
+
"grad_norm": 0.08370938897132874,
|
| 24787 |
+
"learning_rate": 0.0009886396162502547,
|
| 24788 |
+
"loss": 1.5695,
|
| 24789 |
+
"step": 7048
|
| 24790 |
+
},
|
| 24791 |
+
{
|
| 24792 |
+
"epoch": 0.3134587168200614,
|
| 24793 |
+
"grad_norm": 0.08297300338745117,
|
| 24794 |
+
"learning_rate": 0.0009886321301323835,
|
| 24795 |
+
"loss": 1.5669,
|
| 24796 |
+
"step": 7050
|
| 24797 |
+
},
|
| 24798 |
+
{
|
| 24799 |
+
"epoch": 0.3135476412787337,
|
| 24800 |
+
"grad_norm": 0.07920370995998383,
|
| 24801 |
+
"learning_rate": 0.0009886246415771343,
|
| 24802 |
+
"loss": 1.5671,
|
| 24803 |
+
"step": 7052
|
| 24804 |
+
},
|
| 24805 |
+
{
|
| 24806 |
+
"epoch": 0.31363656573740606,
|
| 24807 |
+
"grad_norm": 0.08777054399251938,
|
| 24808 |
+
"learning_rate": 0.000988617150584544,
|
| 24809 |
+
"loss": 1.5684,
|
| 24810 |
+
"step": 7054
|
| 24811 |
+
},
|
| 24812 |
+
{
|
| 24813 |
+
"epoch": 0.3137254901960784,
|
| 24814 |
+
"grad_norm": 0.08385943621397018,
|
| 24815 |
+
"learning_rate": 0.0009886096571546505,
|
| 24816 |
+
"loss": 1.5594,
|
| 24817 |
+
"step": 7056
|
| 24818 |
+
},
|
| 24819 |
+
{
|
| 24820 |
+
"epoch": 0.3138144146547508,
|
| 24821 |
+
"grad_norm": 0.08605042845010757,
|
| 24822 |
+
"learning_rate": 0.0009886021612874908,
|
| 24823 |
+
"loss": 1.5674,
|
| 24824 |
+
"step": 7058
|
| 24825 |
+
},
|
| 24826 |
+
{
|
| 24827 |
+
"epoch": 0.31390333911342316,
|
| 24828 |
+
"grad_norm": 0.08706225454807281,
|
| 24829 |
+
"learning_rate": 0.0009885946629831024,
|
| 24830 |
+
"loss": 1.5579,
|
| 24831 |
+
"step": 7060
|
| 24832 |
+
},
|
| 24833 |
+
{
|
| 24834 |
+
"epoch": 0.31399226357209553,
|
| 24835 |
+
"grad_norm": 0.08403518795967102,
|
| 24836 |
+
"learning_rate": 0.000988587162241523,
|
| 24837 |
+
"loss": 1.5624,
|
| 24838 |
+
"step": 7062
|
| 24839 |
+
},
|
| 24840 |
+
{
|
| 24841 |
+
"epoch": 0.31408118803076784,
|
| 24842 |
+
"grad_norm": 0.08342042565345764,
|
| 24843 |
+
"learning_rate": 0.0009885796590627894,
|
| 24844 |
+
"loss": 1.5651,
|
| 24845 |
+
"step": 7064
|
| 24846 |
+
},
|
| 24847 |
+
{
|
| 24848 |
+
"epoch": 0.3141701124894402,
|
| 24849 |
+
"grad_norm": 0.08213895559310913,
|
| 24850 |
+
"learning_rate": 0.0009885721534469395,
|
| 24851 |
+
"loss": 1.5653,
|
| 24852 |
+
"step": 7066
|
| 24853 |
+
},
|
| 24854 |
+
{
|
| 24855 |
+
"epoch": 0.3142590369481126,
|
| 24856 |
+
"grad_norm": 0.08500027656555176,
|
| 24857 |
+
"learning_rate": 0.0009885646453940105,
|
| 24858 |
+
"loss": 1.5693,
|
| 24859 |
+
"step": 7068
|
| 24860 |
+
},
|
| 24861 |
+
{
|
| 24862 |
+
"epoch": 0.31434796140678495,
|
| 24863 |
+
"grad_norm": 0.08024073392152786,
|
| 24864 |
+
"learning_rate": 0.0009885571349040401,
|
| 24865 |
+
"loss": 1.565,
|
| 24866 |
+
"step": 7070
|
| 24867 |
+
},
|
| 24868 |
+
{
|
| 24869 |
+
"epoch": 0.3144368858654573,
|
| 24870 |
+
"grad_norm": 0.08580642938613892,
|
| 24871 |
+
"learning_rate": 0.0009885496219770656,
|
| 24872 |
+
"loss": 1.5634,
|
| 24873 |
+
"step": 7072
|
| 24874 |
+
},
|
| 24875 |
+
{
|
| 24876 |
+
"epoch": 0.31452581032412963,
|
| 24877 |
+
"grad_norm": 0.08305441588163376,
|
| 24878 |
+
"learning_rate": 0.0009885421066131244,
|
| 24879 |
+
"loss": 1.5678,
|
| 24880 |
+
"step": 7074
|
| 24881 |
+
},
|
| 24882 |
+
{
|
| 24883 |
+
"epoch": 0.314614734782802,
|
| 24884 |
+
"grad_norm": 0.0872965082526207,
|
| 24885 |
+
"learning_rate": 0.000988534588812254,
|
| 24886 |
+
"loss": 1.5619,
|
| 24887 |
+
"step": 7076
|
| 24888 |
+
},
|
| 24889 |
+
{
|
| 24890 |
+
"epoch": 0.31470365924147436,
|
| 24891 |
+
"grad_norm": 0.08917077630758286,
|
| 24892 |
+
"learning_rate": 0.0009885270685744921,
|
| 24893 |
+
"loss": 1.5585,
|
| 24894 |
+
"step": 7078
|
| 24895 |
+
},
|
| 24896 |
+
{
|
| 24897 |
+
"epoch": 0.31479258370014673,
|
| 24898 |
+
"grad_norm": 0.08678824454545975,
|
| 24899 |
+
"learning_rate": 0.000988519545899876,
|
| 24900 |
+
"loss": 1.5599,
|
| 24901 |
+
"step": 7080
|
| 24902 |
+
},
|
| 24903 |
+
{
|
| 24904 |
+
"epoch": 0.3148815081588191,
|
| 24905 |
+
"grad_norm": 0.08195692300796509,
|
| 24906 |
+
"learning_rate": 0.0009885120207884434,
|
| 24907 |
+
"loss": 1.565,
|
| 24908 |
+
"step": 7082
|
| 24909 |
+
},
|
| 24910 |
+
{
|
| 24911 |
+
"epoch": 0.3149704326174914,
|
| 24912 |
+
"grad_norm": 0.08080308139324188,
|
| 24913 |
+
"learning_rate": 0.000988504493240232,
|
| 24914 |
+
"loss": 1.5663,
|
| 24915 |
+
"step": 7084
|
| 24916 |
+
},
|
| 24917 |
+
{
|
| 24918 |
+
"epoch": 0.3150593570761638,
|
| 24919 |
+
"grad_norm": 0.08524754643440247,
|
| 24920 |
+
"learning_rate": 0.0009884969632552787,
|
| 24921 |
+
"loss": 1.5669,
|
| 24922 |
+
"step": 7086
|
| 24923 |
+
},
|
| 24924 |
+
{
|
| 24925 |
+
"epoch": 0.31514828153483615,
|
| 24926 |
+
"grad_norm": 0.08281772583723068,
|
| 24927 |
+
"learning_rate": 0.0009884894308336215,
|
| 24928 |
+
"loss": 1.5625,
|
| 24929 |
+
"step": 7088
|
| 24930 |
+
},
|
| 24931 |
+
{
|
| 24932 |
+
"epoch": 0.3152372059935085,
|
| 24933 |
+
"grad_norm": 0.08267832547426224,
|
| 24934 |
+
"learning_rate": 0.0009884818959752981,
|
| 24935 |
+
"loss": 1.5649,
|
| 24936 |
+
"step": 7090
|
| 24937 |
+
},
|
| 24938 |
+
{
|
| 24939 |
+
"epoch": 0.3153261304521809,
|
| 24940 |
+
"grad_norm": 0.08353657275438309,
|
| 24941 |
+
"learning_rate": 0.0009884743586803457,
|
| 24942 |
+
"loss": 1.5639,
|
| 24943 |
+
"step": 7092
|
| 24944 |
+
},
|
| 24945 |
+
{
|
| 24946 |
+
"epoch": 0.31541505491085325,
|
| 24947 |
+
"grad_norm": 0.08342524617910385,
|
| 24948 |
+
"learning_rate": 0.0009884668189488023,
|
| 24949 |
+
"loss": 1.5646,
|
| 24950 |
+
"step": 7094
|
| 24951 |
+
},
|
| 24952 |
+
{
|
| 24953 |
+
"epoch": 0.31550397936952557,
|
| 24954 |
+
"grad_norm": 0.08173046261072159,
|
| 24955 |
+
"learning_rate": 0.0009884592767807053,
|
| 24956 |
+
"loss": 1.5661,
|
| 24957 |
+
"step": 7096
|
| 24958 |
+
},
|
| 24959 |
+
{
|
| 24960 |
+
"epoch": 0.31559290382819793,
|
| 24961 |
+
"grad_norm": 0.08680406212806702,
|
| 24962 |
+
"learning_rate": 0.0009884517321760923,
|
| 24963 |
+
"loss": 1.5619,
|
| 24964 |
+
"step": 7098
|
| 24965 |
+
},
|
| 24966 |
+
{
|
| 24967 |
+
"epoch": 0.3156818282868703,
|
| 24968 |
+
"grad_norm": 0.08208227157592773,
|
| 24969 |
+
"learning_rate": 0.000988444185135001,
|
| 24970 |
+
"loss": 1.5639,
|
| 24971 |
+
"step": 7100
|
| 24972 |
+
},
|
| 24973 |
+
{
|
| 24974 |
+
"epoch": 0.31577075274554267,
|
| 24975 |
+
"grad_norm": 0.0871683731675148,
|
| 24976 |
+
"learning_rate": 0.0009884366356574692,
|
| 24977 |
+
"loss": 1.5657,
|
| 24978 |
+
"step": 7102
|
| 24979 |
+
},
|
| 24980 |
+
{
|
| 24981 |
+
"epoch": 0.31585967720421504,
|
| 24982 |
+
"grad_norm": 0.08410225808620453,
|
| 24983 |
+
"learning_rate": 0.0009884290837435341,
|
| 24984 |
+
"loss": 1.5616,
|
| 24985 |
+
"step": 7104
|
| 24986 |
+
},
|
| 24987 |
+
{
|
| 24988 |
+
"epoch": 0.31594860166288735,
|
| 24989 |
+
"grad_norm": 0.08295659720897675,
|
| 24990 |
+
"learning_rate": 0.0009884215293932339,
|
| 24991 |
+
"loss": 1.5653,
|
| 24992 |
+
"step": 7106
|
| 24993 |
+
},
|
| 24994 |
+
{
|
| 24995 |
+
"epoch": 0.3160375261215597,
|
| 24996 |
+
"grad_norm": 0.08682281523942947,
|
| 24997 |
+
"learning_rate": 0.000988413972606606,
|
| 24998 |
+
"loss": 1.5636,
|
| 24999 |
+
"step": 7108
|
| 25000 |
+
},
|
| 25001 |
+
{
|
| 25002 |
+
"epoch": 0.3161264505802321,
|
| 25003 |
+
"grad_norm": 0.08395165205001831,
|
| 25004 |
+
"learning_rate": 0.000988406413383688,
|
| 25005 |
+
"loss": 1.5628,
|
| 25006 |
+
"step": 7110
|
| 25007 |
+
},
|
| 25008 |
+
{
|
| 25009 |
+
"epoch": 0.31621537503890446,
|
| 25010 |
+
"grad_norm": 0.08330830186605453,
|
| 25011 |
+
"learning_rate": 0.0009883988517245177,
|
| 25012 |
+
"loss": 1.5585,
|
| 25013 |
+
"step": 7112
|
| 25014 |
+
},
|
| 25015 |
+
{
|
| 25016 |
+
"epoch": 0.3163042994975768,
|
| 25017 |
+
"grad_norm": 0.08235388249158859,
|
| 25018 |
+
"learning_rate": 0.0009883912876291329,
|
| 25019 |
+
"loss": 1.5712,
|
| 25020 |
+
"step": 7114
|
| 25021 |
+
},
|
| 25022 |
+
{
|
| 25023 |
+
"epoch": 0.3163932239562492,
|
| 25024 |
+
"grad_norm": 0.08317794650793076,
|
| 25025 |
+
"learning_rate": 0.0009883837210975714,
|
| 25026 |
+
"loss": 1.5658,
|
| 25027 |
+
"step": 7116
|
| 25028 |
+
},
|
| 25029 |
+
{
|
| 25030 |
+
"epoch": 0.3164821484149215,
|
| 25031 |
+
"grad_norm": 0.08337473124265671,
|
| 25032 |
+
"learning_rate": 0.0009883761521298708,
|
| 25033 |
+
"loss": 1.5615,
|
| 25034 |
+
"step": 7118
|
| 25035 |
+
},
|
| 25036 |
+
{
|
| 25037 |
+
"epoch": 0.3165710728735939,
|
| 25038 |
+
"grad_norm": 0.08434969931840897,
|
| 25039 |
+
"learning_rate": 0.000988368580726069,
|
| 25040 |
+
"loss": 1.5617,
|
| 25041 |
+
"step": 7120
|
| 25042 |
+
},
|
| 25043 |
+
{
|
| 25044 |
+
"epoch": 0.31665999733226624,
|
| 25045 |
+
"grad_norm": 0.08485856652259827,
|
| 25046 |
+
"learning_rate": 0.0009883610068862035,
|
| 25047 |
+
"loss": 1.5657,
|
| 25048 |
+
"step": 7122
|
| 25049 |
+
},
|
| 25050 |
+
{
|
| 25051 |
+
"epoch": 0.3167489217909386,
|
| 25052 |
+
"grad_norm": 0.08327708393335342,
|
| 25053 |
+
"learning_rate": 0.0009883534306103121,
|
| 25054 |
+
"loss": 1.5656,
|
| 25055 |
+
"step": 7124
|
| 25056 |
+
},
|
| 25057 |
+
{
|
| 25058 |
+
"epoch": 0.316837846249611,
|
| 25059 |
+
"grad_norm": 0.08332667499780655,
|
| 25060 |
+
"learning_rate": 0.000988345851898433,
|
| 25061 |
+
"loss": 1.5645,
|
| 25062 |
+
"step": 7126
|
| 25063 |
+
},
|
| 25064 |
+
{
|
| 25065 |
+
"epoch": 0.3169267707082833,
|
| 25066 |
+
"grad_norm": 0.08230622857809067,
|
| 25067 |
+
"learning_rate": 0.0009883382707506036,
|
| 25068 |
+
"loss": 1.5623,
|
| 25069 |
+
"step": 7128
|
| 25070 |
+
},
|
| 25071 |
+
{
|
| 25072 |
+
"epoch": 0.31701569516695566,
|
| 25073 |
+
"grad_norm": 0.08090732246637344,
|
| 25074 |
+
"learning_rate": 0.0009883306871668618,
|
| 25075 |
+
"loss": 1.5695,
|
| 25076 |
+
"step": 7130
|
| 25077 |
+
},
|
| 25078 |
+
{
|
| 25079 |
+
"epoch": 0.317104619625628,
|
| 25080 |
+
"grad_norm": 0.08214423060417175,
|
| 25081 |
+
"learning_rate": 0.0009883231011472455,
|
| 25082 |
+
"loss": 1.5633,
|
| 25083 |
+
"step": 7132
|
| 25084 |
+
},
|
| 25085 |
+
{
|
| 25086 |
+
"epoch": 0.3171935440843004,
|
| 25087 |
+
"grad_norm": 0.08126901835203171,
|
| 25088 |
+
"learning_rate": 0.0009883155126917925,
|
| 25089 |
+
"loss": 1.5634,
|
| 25090 |
+
"step": 7134
|
| 25091 |
+
},
|
| 25092 |
+
{
|
| 25093 |
+
"epoch": 0.31728246854297276,
|
| 25094 |
+
"grad_norm": 0.08316788822412491,
|
| 25095 |
+
"learning_rate": 0.0009883079218005407,
|
| 25096 |
+
"loss": 1.5626,
|
| 25097 |
+
"step": 7136
|
| 25098 |
+
},
|
| 25099 |
+
{
|
| 25100 |
+
"epoch": 0.31737139300164513,
|
| 25101 |
+
"grad_norm": 0.0886952131986618,
|
| 25102 |
+
"learning_rate": 0.000988300328473528,
|
| 25103 |
+
"loss": 1.5624,
|
| 25104 |
+
"step": 7138
|
| 25105 |
+
},
|
| 25106 |
+
{
|
| 25107 |
+
"epoch": 0.31746031746031744,
|
| 25108 |
+
"grad_norm": 0.0825994685292244,
|
| 25109 |
+
"learning_rate": 0.0009882927327107922,
|
| 25110 |
+
"loss": 1.5671,
|
| 25111 |
+
"step": 7140
|
| 25112 |
+
},
|
| 25113 |
+
{
|
| 25114 |
+
"epoch": 0.3175492419189898,
|
| 25115 |
+
"grad_norm": 0.09127391129732132,
|
| 25116 |
+
"learning_rate": 0.000988285134512371,
|
| 25117 |
+
"loss": 1.5592,
|
| 25118 |
+
"step": 7142
|
| 25119 |
+
},
|
| 25120 |
+
{
|
| 25121 |
+
"epoch": 0.3176381663776622,
|
| 25122 |
+
"grad_norm": 0.08798851072788239,
|
| 25123 |
+
"learning_rate": 0.0009882775338783028,
|
| 25124 |
+
"loss": 1.5618,
|
| 25125 |
+
"step": 7144
|
| 25126 |
+
},
|
| 25127 |
+
{
|
| 25128 |
+
"epoch": 0.31772709083633455,
|
| 25129 |
+
"grad_norm": 0.09108838438987732,
|
| 25130 |
+
"learning_rate": 0.000988269930808625,
|
| 25131 |
+
"loss": 1.5644,
|
| 25132 |
+
"step": 7146
|
| 25133 |
+
},
|
| 25134 |
+
{
|
| 25135 |
+
"epoch": 0.3178160152950069,
|
| 25136 |
+
"grad_norm": 0.08250788599252701,
|
| 25137 |
+
"learning_rate": 0.0009882623253033758,
|
| 25138 |
+
"loss": 1.5662,
|
| 25139 |
+
"step": 7148
|
| 25140 |
+
},
|
| 25141 |
+
{
|
| 25142 |
+
"epoch": 0.3179049397536792,
|
| 25143 |
+
"grad_norm": 0.08027481287717819,
|
| 25144 |
+
"learning_rate": 0.000988254717362593,
|
| 25145 |
+
"loss": 1.5601,
|
| 25146 |
+
"step": 7150
|
| 25147 |
+
},
|
| 25148 |
+
{
|
| 25149 |
+
"epoch": 0.3179938642123516,
|
| 25150 |
+
"grad_norm": 0.08304357528686523,
|
| 25151 |
+
"learning_rate": 0.0009882471069863146,
|
| 25152 |
+
"loss": 1.5578,
|
| 25153 |
+
"step": 7152
|
| 25154 |
+
},
|
| 25155 |
+
{
|
| 25156 |
+
"epoch": 0.31808278867102396,
|
| 25157 |
+
"grad_norm": 0.08454793691635132,
|
| 25158 |
+
"learning_rate": 0.0009882394941745788,
|
| 25159 |
+
"loss": 1.5567,
|
| 25160 |
+
"step": 7154
|
| 25161 |
+
},
|
| 25162 |
+
{
|
| 25163 |
+
"epoch": 0.31817171312969633,
|
| 25164 |
+
"grad_norm": 0.08314481377601624,
|
| 25165 |
+
"learning_rate": 0.0009882318789274233,
|
| 25166 |
+
"loss": 1.5692,
|
| 25167 |
+
"step": 7156
|
| 25168 |
+
},
|
| 25169 |
+
{
|
| 25170 |
+
"epoch": 0.3182606375883687,
|
| 25171 |
+
"grad_norm": 0.08372924476861954,
|
| 25172 |
+
"learning_rate": 0.000988224261244886,
|
| 25173 |
+
"loss": 1.5597,
|
| 25174 |
+
"step": 7158
|
| 25175 |
+
},
|
| 25176 |
+
{
|
| 25177 |
+
"epoch": 0.318349562047041,
|
| 25178 |
+
"grad_norm": 0.08202824741601944,
|
| 25179 |
+
"learning_rate": 0.000988216641127005,
|
| 25180 |
+
"loss": 1.5669,
|
| 25181 |
+
"step": 7160
|
| 25182 |
+
},
|
| 25183 |
+
{
|
| 25184 |
+
"epoch": 0.3184384865057134,
|
| 25185 |
+
"grad_norm": 0.08950933814048767,
|
| 25186 |
+
"learning_rate": 0.0009882090185738186,
|
| 25187 |
+
"loss": 1.5682,
|
| 25188 |
+
"step": 7162
|
| 25189 |
+
},
|
| 25190 |
+
{
|
| 25191 |
+
"epoch": 0.31852741096438575,
|
| 25192 |
+
"grad_norm": 0.08493774384260178,
|
| 25193 |
+
"learning_rate": 0.0009882013935853644,
|
| 25194 |
+
"loss": 1.5685,
|
| 25195 |
+
"step": 7164
|
| 25196 |
+
},
|
| 25197 |
+
{
|
| 25198 |
+
"epoch": 0.3186163354230581,
|
| 25199 |
+
"grad_norm": 0.08362516015768051,
|
| 25200 |
+
"learning_rate": 0.0009881937661616806,
|
| 25201 |
+
"loss": 1.5685,
|
| 25202 |
+
"step": 7166
|
| 25203 |
+
},
|
| 25204 |
+
{
|
| 25205 |
+
"epoch": 0.3187052598817305,
|
| 25206 |
+
"grad_norm": 0.08802633732557297,
|
| 25207 |
+
"learning_rate": 0.0009881861363028053,
|
| 25208 |
+
"loss": 1.5684,
|
| 25209 |
+
"step": 7168
|
| 25210 |
+
},
|
| 25211 |
+
{
|
| 25212 |
+
"epoch": 0.31879418434040285,
|
| 25213 |
+
"grad_norm": 0.08594125509262085,
|
| 25214 |
+
"learning_rate": 0.0009881785040087765,
|
| 25215 |
+
"loss": 1.5581,
|
| 25216 |
+
"step": 7170
|
| 25217 |
+
},
|
| 25218 |
+
{
|
| 25219 |
+
"epoch": 0.31888310879907517,
|
| 25220 |
+
"grad_norm": 0.08420670032501221,
|
| 25221 |
+
"learning_rate": 0.0009881708692796325,
|
| 25222 |
+
"loss": 1.5626,
|
| 25223 |
+
"step": 7172
|
| 25224 |
+
},
|
| 25225 |
+
{
|
| 25226 |
+
"epoch": 0.31897203325774753,
|
| 25227 |
+
"grad_norm": 0.08396308124065399,
|
| 25228 |
+
"learning_rate": 0.000988163232115411,
|
| 25229 |
+
"loss": 1.5624,
|
| 25230 |
+
"step": 7174
|
| 25231 |
+
},
|
| 25232 |
+
{
|
| 25233 |
+
"epoch": 0.3190609577164199,
|
| 25234 |
+
"grad_norm": 0.08624763786792755,
|
| 25235 |
+
"learning_rate": 0.00098815559251615,
|
| 25236 |
+
"loss": 1.5667,
|
| 25237 |
+
"step": 7176
|
| 25238 |
+
},
|
| 25239 |
+
{
|
| 25240 |
+
"epoch": 0.31914988217509227,
|
| 25241 |
+
"grad_norm": 0.0823800191283226,
|
| 25242 |
+
"learning_rate": 0.0009881479504818883,
|
| 25243 |
+
"loss": 1.5654,
|
| 25244 |
+
"step": 7178
|
| 25245 |
+
},
|
| 25246 |
+
{
|
| 25247 |
+
"epoch": 0.31923880663376464,
|
| 25248 |
+
"grad_norm": 0.08368480205535889,
|
| 25249 |
+
"learning_rate": 0.0009881403060126635,
|
| 25250 |
+
"loss": 1.5673,
|
| 25251 |
+
"step": 7180
|
| 25252 |
+
},
|
| 25253 |
+
{
|
| 25254 |
+
"epoch": 0.31932773109243695,
|
| 25255 |
+
"grad_norm": 0.08373897522687912,
|
| 25256 |
+
"learning_rate": 0.0009881326591085135,
|
| 25257 |
+
"loss": 1.565,
|
| 25258 |
+
"step": 7182
|
| 25259 |
+
},
|
| 25260 |
+
{
|
| 25261 |
+
"epoch": 0.3194166555511093,
|
| 25262 |
+
"grad_norm": 0.08619046956300735,
|
| 25263 |
+
"learning_rate": 0.0009881250097694772,
|
| 25264 |
+
"loss": 1.5597,
|
| 25265 |
+
"step": 7184
|
| 25266 |
+
},
|
| 25267 |
+
{
|
| 25268 |
+
"epoch": 0.3195055800097817,
|
| 25269 |
+
"grad_norm": 0.08758384734392166,
|
| 25270 |
+
"learning_rate": 0.000988117357995592,
|
| 25271 |
+
"loss": 1.5618,
|
| 25272 |
+
"step": 7186
|
| 25273 |
+
},
|
| 25274 |
+
{
|
| 25275 |
+
"epoch": 0.31959450446845405,
|
| 25276 |
+
"grad_norm": 0.08243545889854431,
|
| 25277 |
+
"learning_rate": 0.0009881097037868966,
|
| 25278 |
+
"loss": 1.5549,
|
| 25279 |
+
"step": 7188
|
| 25280 |
+
},
|
| 25281 |
+
{
|
| 25282 |
+
"epoch": 0.3196834289271264,
|
| 25283 |
+
"grad_norm": 0.07898540049791336,
|
| 25284 |
+
"learning_rate": 0.000988102047143429,
|
| 25285 |
+
"loss": 1.5597,
|
| 25286 |
+
"step": 7190
|
| 25287 |
+
},
|
| 25288 |
+
{
|
| 25289 |
+
"epoch": 0.3197723533857988,
|
| 25290 |
+
"grad_norm": 0.08437183499336243,
|
| 25291 |
+
"learning_rate": 0.0009880943880652274,
|
| 25292 |
+
"loss": 1.5637,
|
| 25293 |
+
"step": 7192
|
| 25294 |
+
},
|
| 25295 |
+
{
|
| 25296 |
+
"epoch": 0.3198612778444711,
|
| 25297 |
+
"grad_norm": 0.08325666189193726,
|
| 25298 |
+
"learning_rate": 0.0009880867265523296,
|
| 25299 |
+
"loss": 1.5613,
|
| 25300 |
+
"step": 7194
|
| 25301 |
+
},
|
| 25302 |
+
{
|
| 25303 |
+
"epoch": 0.31995020230314347,
|
| 25304 |
+
"grad_norm": 0.08307388424873352,
|
| 25305 |
+
"learning_rate": 0.0009880790626047747,
|
| 25306 |
+
"loss": 1.5627,
|
| 25307 |
+
"step": 7196
|
| 25308 |
+
},
|
| 25309 |
+
{
|
| 25310 |
+
"epoch": 0.32003912676181584,
|
| 25311 |
+
"grad_norm": 0.08156055212020874,
|
| 25312 |
+
"learning_rate": 0.0009880713962226,
|
| 25313 |
+
"loss": 1.5666,
|
| 25314 |
+
"step": 7198
|
| 25315 |
+
},
|
| 25316 |
+
{
|
| 25317 |
+
"epoch": 0.3201280512204882,
|
| 25318 |
+
"grad_norm": 0.08196540176868439,
|
| 25319 |
+
"learning_rate": 0.0009880637274058443,
|
| 25320 |
+
"loss": 1.5627,
|
| 25321 |
+
"step": 7200
|
| 25322 |
+
},
|
| 25323 |
+
{
|
| 25324 |
+
"epoch": 0.3202169756791606,
|
| 25325 |
+
"grad_norm": 0.08538369089365005,
|
| 25326 |
+
"learning_rate": 0.0009880560561545458,
|
| 25327 |
+
"loss": 1.5644,
|
| 25328 |
+
"step": 7202
|
| 25329 |
+
},
|
| 25330 |
+
{
|
| 25331 |
+
"epoch": 0.3203059001378329,
|
| 25332 |
+
"grad_norm": 0.08306334167718887,
|
| 25333 |
+
"learning_rate": 0.0009880483824687427,
|
| 25334 |
+
"loss": 1.5582,
|
| 25335 |
+
"step": 7204
|
| 25336 |
+
},
|
| 25337 |
+
{
|
| 25338 |
+
"epoch": 0.32039482459650526,
|
| 25339 |
+
"grad_norm": 0.08345794677734375,
|
| 25340 |
+
"learning_rate": 0.0009880407063484731,
|
| 25341 |
+
"loss": 1.5538,
|
| 25342 |
+
"step": 7206
|
| 25343 |
+
},
|
| 25344 |
+
{
|
| 25345 |
+
"epoch": 0.3204837490551776,
|
| 25346 |
+
"grad_norm": 0.08639951795339584,
|
| 25347 |
+
"learning_rate": 0.0009880330277937757,
|
| 25348 |
+
"loss": 1.5601,
|
| 25349 |
+
"step": 7208
|
| 25350 |
+
},
|
| 25351 |
+
{
|
| 25352 |
+
"epoch": 0.32057267351385,
|
| 25353 |
+
"grad_norm": 0.08315160125494003,
|
| 25354 |
+
"learning_rate": 0.0009880253468046883,
|
| 25355 |
+
"loss": 1.5591,
|
| 25356 |
+
"step": 7210
|
| 25357 |
+
},
|
| 25358 |
+
{
|
| 25359 |
+
"epoch": 0.32066159797252236,
|
| 25360 |
+
"grad_norm": 0.0834464579820633,
|
| 25361 |
+
"learning_rate": 0.0009880176633812493,
|
| 25362 |
+
"loss": 1.5663,
|
| 25363 |
+
"step": 7212
|
| 25364 |
+
},
|
| 25365 |
+
{
|
| 25366 |
+
"epoch": 0.3207505224311947,
|
| 25367 |
+
"grad_norm": 0.08205483853816986,
|
| 25368 |
+
"learning_rate": 0.0009880099775234977,
|
| 25369 |
+
"loss": 1.5662,
|
| 25370 |
+
"step": 7214
|
| 25371 |
+
},
|
| 25372 |
+
{
|
| 25373 |
+
"epoch": 0.32083944688986704,
|
| 25374 |
+
"grad_norm": 0.08744161576032639,
|
| 25375 |
+
"learning_rate": 0.000988002289231471,
|
| 25376 |
+
"loss": 1.5548,
|
| 25377 |
+
"step": 7216
|
| 25378 |
+
},
|
| 25379 |
+
{
|
| 25380 |
+
"epoch": 0.3209283713485394,
|
| 25381 |
+
"grad_norm": 0.08676169067621231,
|
| 25382 |
+
"learning_rate": 0.000987994598505208,
|
| 25383 |
+
"loss": 1.5595,
|
| 25384 |
+
"step": 7218
|
| 25385 |
+
},
|
| 25386 |
+
{
|
| 25387 |
+
"epoch": 0.3210172958072118,
|
| 25388 |
+
"grad_norm": 0.08752325177192688,
|
| 25389 |
+
"learning_rate": 0.0009879869053447468,
|
| 25390 |
+
"loss": 1.5624,
|
| 25391 |
+
"step": 7220
|
| 25392 |
+
},
|
| 25393 |
+
{
|
| 25394 |
+
"epoch": 0.32110622026588415,
|
| 25395 |
+
"grad_norm": 0.08283118903636932,
|
| 25396 |
+
"learning_rate": 0.0009879792097501258,
|
| 25397 |
+
"loss": 1.5675,
|
| 25398 |
+
"step": 7222
|
| 25399 |
+
},
|
| 25400 |
+
{
|
| 25401 |
+
"epoch": 0.3211951447245565,
|
| 25402 |
+
"grad_norm": 0.08117381483316422,
|
| 25403 |
+
"learning_rate": 0.0009879715117213838,
|
| 25404 |
+
"loss": 1.5604,
|
| 25405 |
+
"step": 7224
|
| 25406 |
+
},
|
| 25407 |
+
{
|
| 25408 |
+
"epoch": 0.3212840691832288,
|
| 25409 |
+
"grad_norm": 0.08399328589439392,
|
| 25410 |
+
"learning_rate": 0.0009879638112585587,
|
| 25411 |
+
"loss": 1.5532,
|
| 25412 |
+
"step": 7226
|
| 25413 |
+
},
|
| 25414 |
+
{
|
| 25415 |
+
"epoch": 0.3213729936419012,
|
| 25416 |
+
"grad_norm": 0.0844864770770073,
|
| 25417 |
+
"learning_rate": 0.0009879561083616893,
|
| 25418 |
+
"loss": 1.5662,
|
| 25419 |
+
"step": 7228
|
| 25420 |
+
},
|
| 25421 |
+
{
|
| 25422 |
+
"epoch": 0.32146191810057356,
|
| 25423 |
+
"grad_norm": 0.08030157536268234,
|
| 25424 |
+
"learning_rate": 0.0009879484030308136,
|
| 25425 |
+
"loss": 1.5637,
|
| 25426 |
+
"step": 7230
|
| 25427 |
+
},
|
| 25428 |
+
{
|
| 25429 |
+
"epoch": 0.32155084255924593,
|
| 25430 |
+
"grad_norm": 0.08214158564805984,
|
| 25431 |
+
"learning_rate": 0.0009879406952659705,
|
| 25432 |
+
"loss": 1.5642,
|
| 25433 |
+
"step": 7232
|
| 25434 |
+
},
|
| 25435 |
+
{
|
| 25436 |
+
"epoch": 0.3216397670179183,
|
| 25437 |
+
"grad_norm": 0.08249089121818542,
|
| 25438 |
+
"learning_rate": 0.0009879329850671981,
|
| 25439 |
+
"loss": 1.5614,
|
| 25440 |
+
"step": 7234
|
| 25441 |
+
},
|
| 25442 |
+
{
|
| 25443 |
+
"epoch": 0.3217286914765906,
|
| 25444 |
+
"grad_norm": 0.0858563482761383,
|
| 25445 |
+
"learning_rate": 0.000987925272434535,
|
| 25446 |
+
"loss": 1.5587,
|
| 25447 |
+
"step": 7236
|
| 25448 |
+
},
|
| 25449 |
+
{
|
| 25450 |
+
"epoch": 0.321817615935263,
|
| 25451 |
+
"grad_norm": 0.08245126157999039,
|
| 25452 |
+
"learning_rate": 0.0009879175573680196,
|
| 25453 |
+
"loss": 1.5641,
|
| 25454 |
+
"step": 7238
|
| 25455 |
+
},
|
| 25456 |
+
{
|
| 25457 |
+
"epoch": 0.32190654039393535,
|
| 25458 |
+
"grad_norm": 0.0817423164844513,
|
| 25459 |
+
"learning_rate": 0.0009879098398676905,
|
| 25460 |
+
"loss": 1.5613,
|
| 25461 |
+
"step": 7240
|
| 25462 |
+
},
|
| 25463 |
+
{
|
| 25464 |
+
"epoch": 0.3219954648526077,
|
| 25465 |
+
"grad_norm": 0.08229317516088486,
|
| 25466 |
+
"learning_rate": 0.0009879021199335862,
|
| 25467 |
+
"loss": 1.558,
|
| 25468 |
+
"step": 7242
|
| 25469 |
+
},
|
| 25470 |
+
{
|
| 25471 |
+
"epoch": 0.3220843893112801,
|
| 25472 |
+
"grad_norm": 0.0869976207613945,
|
| 25473 |
+
"learning_rate": 0.000987894397565745,
|
| 25474 |
+
"loss": 1.556,
|
| 25475 |
+
"step": 7244
|
| 25476 |
+
},
|
| 25477 |
+
{
|
| 25478 |
+
"epoch": 0.32217331376995245,
|
| 25479 |
+
"grad_norm": 0.08708877116441727,
|
| 25480 |
+
"learning_rate": 0.0009878866727642056,
|
| 25481 |
+
"loss": 1.5642,
|
| 25482 |
+
"step": 7246
|
| 25483 |
+
},
|
| 25484 |
+
{
|
| 25485 |
+
"epoch": 0.32226223822862476,
|
| 25486 |
+
"grad_norm": 0.08197760581970215,
|
| 25487 |
+
"learning_rate": 0.0009878789455290064,
|
| 25488 |
+
"loss": 1.5577,
|
| 25489 |
+
"step": 7248
|
| 25490 |
+
},
|
| 25491 |
+
{
|
| 25492 |
+
"epoch": 0.32235116268729713,
|
| 25493 |
+
"grad_norm": 0.08504566550254822,
|
| 25494 |
+
"learning_rate": 0.0009878712158601863,
|
| 25495 |
+
"loss": 1.5633,
|
| 25496 |
+
"step": 7250
|
| 25497 |
+
},
|
| 25498 |
+
{
|
| 25499 |
+
"epoch": 0.3224400871459695,
|
| 25500 |
+
"grad_norm": 0.0837409570813179,
|
| 25501 |
+
"learning_rate": 0.0009878634837577832,
|
| 25502 |
+
"loss": 1.5607,
|
| 25503 |
+
"step": 7252
|
| 25504 |
+
},
|
| 25505 |
+
{
|
| 25506 |
+
"epoch": 0.32252901160464187,
|
| 25507 |
+
"grad_norm": 0.08278852701187134,
|
| 25508 |
+
"learning_rate": 0.0009878557492218364,
|
| 25509 |
+
"loss": 1.5618,
|
| 25510 |
+
"step": 7254
|
| 25511 |
+
},
|
| 25512 |
+
{
|
| 25513 |
+
"epoch": 0.32261793606331424,
|
| 25514 |
+
"grad_norm": 0.07882179319858551,
|
| 25515 |
+
"learning_rate": 0.0009878480122523842,
|
| 25516 |
+
"loss": 1.5579,
|
| 25517 |
+
"step": 7256
|
| 25518 |
+
},
|
| 25519 |
+
{
|
| 25520 |
+
"epoch": 0.32270686052198655,
|
| 25521 |
+
"grad_norm": 0.08126863837242126,
|
| 25522 |
+
"learning_rate": 0.0009878402728494648,
|
| 25523 |
+
"loss": 1.5578,
|
| 25524 |
+
"step": 7258
|
| 25525 |
+
},
|
| 25526 |
+
{
|
| 25527 |
+
"epoch": 0.3227957849806589,
|
| 25528 |
+
"grad_norm": 0.08077924698591232,
|
| 25529 |
+
"learning_rate": 0.0009878325310131173,
|
| 25530 |
+
"loss": 1.5598,
|
| 25531 |
+
"step": 7260
|
| 25532 |
+
},
|
| 25533 |
+
{
|
| 25534 |
+
"epoch": 0.3228847094393313,
|
| 25535 |
+
"grad_norm": 0.08375802636146545,
|
| 25536 |
+
"learning_rate": 0.0009878247867433803,
|
| 25537 |
+
"loss": 1.5595,
|
| 25538 |
+
"step": 7262
|
| 25539 |
+
},
|
| 25540 |
+
{
|
| 25541 |
+
"epoch": 0.32297363389800365,
|
| 25542 |
+
"grad_norm": 0.08182679116725922,
|
| 25543 |
+
"learning_rate": 0.000987817040040292,
|
| 25544 |
+
"loss": 1.5558,
|
| 25545 |
+
"step": 7264
|
| 25546 |
+
},
|
| 25547 |
+
{
|
| 25548 |
+
"epoch": 0.323062558356676,
|
| 25549 |
+
"grad_norm": 0.08209866285324097,
|
| 25550 |
+
"learning_rate": 0.0009878092909038916,
|
| 25551 |
+
"loss": 1.5582,
|
| 25552 |
+
"step": 7266
|
| 25553 |
+
},
|
| 25554 |
+
{
|
| 25555 |
+
"epoch": 0.32315148281534833,
|
| 25556 |
+
"grad_norm": 0.08449520170688629,
|
| 25557 |
+
"learning_rate": 0.0009878015393342172,
|
| 25558 |
+
"loss": 1.56,
|
| 25559 |
+
"step": 7268
|
| 25560 |
+
},
|
| 25561 |
+
{
|
| 25562 |
+
"epoch": 0.3232404072740207,
|
| 25563 |
+
"grad_norm": 0.07824712991714478,
|
| 25564 |
+
"learning_rate": 0.000987793785331308,
|
| 25565 |
+
"loss": 1.5512,
|
| 25566 |
+
"step": 7270
|
| 25567 |
+
},
|
| 25568 |
+
{
|
| 25569 |
+
"epoch": 0.32332933173269307,
|
| 25570 |
+
"grad_norm": 0.0843387246131897,
|
| 25571 |
+
"learning_rate": 0.0009877860288952022,
|
| 25572 |
+
"loss": 1.5599,
|
| 25573 |
+
"step": 7272
|
| 25574 |
+
},
|
| 25575 |
+
{
|
| 25576 |
+
"epoch": 0.32341825619136544,
|
| 25577 |
+
"grad_norm": 0.08296310156583786,
|
| 25578 |
+
"learning_rate": 0.000987778270025939,
|
| 25579 |
+
"loss": 1.5652,
|
| 25580 |
+
"step": 7274
|
| 25581 |
+
},
|
| 25582 |
+
{
|
| 25583 |
+
"epoch": 0.3235071806500378,
|
| 25584 |
+
"grad_norm": 0.08399280905723572,
|
| 25585 |
+
"learning_rate": 0.0009877705087235566,
|
| 25586 |
+
"loss": 1.5624,
|
| 25587 |
+
"step": 7276
|
| 25588 |
+
},
|
| 25589 |
+
{
|
| 25590 |
+
"epoch": 0.3235961051087102,
|
| 25591 |
+
"grad_norm": 0.08151458948850632,
|
| 25592 |
+
"learning_rate": 0.000987762744988094,
|
| 25593 |
+
"loss": 1.5661,
|
| 25594 |
+
"step": 7278
|
| 25595 |
+
},
|
| 25596 |
+
{
|
| 25597 |
+
"epoch": 0.3236850295673825,
|
| 25598 |
+
"grad_norm": 0.08649469912052155,
|
| 25599 |
+
"learning_rate": 0.00098775497881959,
|
| 25600 |
+
"loss": 1.5541,
|
| 25601 |
+
"step": 7280
|
| 25602 |
+
},
|
| 25603 |
+
{
|
| 25604 |
+
"epoch": 0.32377395402605486,
|
| 25605 |
+
"grad_norm": 0.0848914235830307,
|
| 25606 |
+
"learning_rate": 0.0009877472102180831,
|
| 25607 |
+
"loss": 1.5622,
|
| 25608 |
+
"step": 7282
|
| 25609 |
+
},
|
| 25610 |
+
{
|
| 25611 |
+
"epoch": 0.3238628784847272,
|
| 25612 |
+
"grad_norm": 0.08599822223186493,
|
| 25613 |
+
"learning_rate": 0.0009877394391836123,
|
| 25614 |
+
"loss": 1.5586,
|
| 25615 |
+
"step": 7284
|
| 25616 |
+
},
|
| 25617 |
+
{
|
| 25618 |
+
"epoch": 0.3239518029433996,
|
| 25619 |
+
"grad_norm": 0.09074999392032623,
|
| 25620 |
+
"learning_rate": 0.000987731665716216,
|
| 25621 |
+
"loss": 1.5584,
|
| 25622 |
+
"step": 7286
|
| 25623 |
+
},
|
| 25624 |
+
{
|
| 25625 |
+
"epoch": 0.32404072740207196,
|
| 25626 |
+
"grad_norm": 0.08198599517345428,
|
| 25627 |
+
"learning_rate": 0.0009877238898159332,
|
| 25628 |
+
"loss": 1.5623,
|
| 25629 |
+
"step": 7288
|
| 25630 |
+
},
|
| 25631 |
+
{
|
| 25632 |
+
"epoch": 0.3241296518607443,
|
| 25633 |
+
"grad_norm": 0.08142637461423874,
|
| 25634 |
+
"learning_rate": 0.0009877161114828026,
|
| 25635 |
+
"loss": 1.5591,
|
| 25636 |
+
"step": 7290
|
| 25637 |
+
},
|
| 25638 |
+
{
|
| 25639 |
+
"epoch": 0.32421857631941664,
|
| 25640 |
+
"grad_norm": 0.08540283888578415,
|
| 25641 |
+
"learning_rate": 0.0009877083307168633,
|
| 25642 |
+
"loss": 1.562,
|
| 25643 |
+
"step": 7292
|
| 25644 |
+
},
|
| 25645 |
+
{
|
| 25646 |
+
"epoch": 0.324307500778089,
|
| 25647 |
+
"grad_norm": 0.08285167068243027,
|
| 25648 |
+
"learning_rate": 0.0009877005475181539,
|
| 25649 |
+
"loss": 1.5581,
|
| 25650 |
+
"step": 7294
|
| 25651 |
+
},
|
| 25652 |
+
{
|
| 25653 |
+
"epoch": 0.3243964252367614,
|
| 25654 |
+
"grad_norm": 0.08604173362255096,
|
| 25655 |
+
"learning_rate": 0.000987692761886713,
|
| 25656 |
+
"loss": 1.5646,
|
| 25657 |
+
"step": 7296
|
| 25658 |
+
},
|
| 25659 |
+
{
|
| 25660 |
+
"epoch": 0.32448534969543374,
|
| 25661 |
+
"grad_norm": 0.08079639822244644,
|
| 25662 |
+
"learning_rate": 0.0009876849738225798,
|
| 25663 |
+
"loss": 1.5611,
|
| 25664 |
+
"step": 7298
|
| 25665 |
+
},
|
| 25666 |
+
{
|
| 25667 |
+
"epoch": 0.3245742741541061,
|
| 25668 |
+
"grad_norm": 0.08307557553052902,
|
| 25669 |
+
"learning_rate": 0.000987677183325793,
|
| 25670 |
+
"loss": 1.5594,
|
| 25671 |
+
"step": 7300
|
| 25672 |
+
},
|
| 25673 |
+
{
|
| 25674 |
+
"epoch": 0.3246631986127784,
|
| 25675 |
+
"grad_norm": 0.08180294185876846,
|
| 25676 |
+
"learning_rate": 0.0009876693903963913,
|
| 25677 |
+
"loss": 1.5631,
|
| 25678 |
+
"step": 7302
|
| 25679 |
+
},
|
| 25680 |
+
{
|
| 25681 |
+
"epoch": 0.3247521230714508,
|
| 25682 |
+
"grad_norm": 0.08266652375459671,
|
| 25683 |
+
"learning_rate": 0.000987661595034414,
|
| 25684 |
+
"loss": 1.559,
|
| 25685 |
+
"step": 7304
|
| 25686 |
+
},
|
| 25687 |
+
{
|
| 25688 |
+
"epoch": 0.32484104753012316,
|
| 25689 |
+
"grad_norm": 0.07898559421300888,
|
| 25690 |
+
"learning_rate": 0.0009876537972398995,
|
| 25691 |
+
"loss": 1.5604,
|
| 25692 |
+
"step": 7306
|
| 25693 |
+
},
|
| 25694 |
+
{
|
| 25695 |
+
"epoch": 0.32492997198879553,
|
| 25696 |
+
"grad_norm": 0.0870475322008133,
|
| 25697 |
+
"learning_rate": 0.000987645997012887,
|
| 25698 |
+
"loss": 1.5617,
|
| 25699 |
+
"step": 7308
|
| 25700 |
+
},
|
| 25701 |
+
{
|
| 25702 |
+
"epoch": 0.3250188964474679,
|
| 25703 |
+
"grad_norm": 0.08088637888431549,
|
| 25704 |
+
"learning_rate": 0.0009876381943534154,
|
| 25705 |
+
"loss": 1.5604,
|
| 25706 |
+
"step": 7310
|
| 25707 |
+
},
|
| 25708 |
+
{
|
| 25709 |
+
"epoch": 0.3251078209061402,
|
| 25710 |
+
"grad_norm": 0.08533639460802078,
|
| 25711 |
+
"learning_rate": 0.0009876303892615234,
|
| 25712 |
+
"loss": 1.5555,
|
| 25713 |
+
"step": 7312
|
| 25714 |
+
},
|
| 25715 |
+
{
|
| 25716 |
+
"epoch": 0.3251967453648126,
|
| 25717 |
+
"grad_norm": 0.08240604400634766,
|
| 25718 |
+
"learning_rate": 0.0009876225817372502,
|
| 25719 |
+
"loss": 1.5609,
|
| 25720 |
+
"step": 7314
|
| 25721 |
+
},
|
| 25722 |
+
{
|
| 25723 |
+
"epoch": 0.32528566982348495,
|
| 25724 |
+
"grad_norm": 0.08196206390857697,
|
| 25725 |
+
"learning_rate": 0.0009876147717806346,
|
| 25726 |
+
"loss": 1.5579,
|
| 25727 |
+
"step": 7316
|
| 25728 |
+
},
|
| 25729 |
+
{
|
| 25730 |
+
"epoch": 0.3253745942821573,
|
| 25731 |
+
"grad_norm": 0.08041080087423325,
|
| 25732 |
+
"learning_rate": 0.0009876069593917154,
|
| 25733 |
+
"loss": 1.5621,
|
| 25734 |
+
"step": 7318
|
| 25735 |
+
},
|
| 25736 |
+
{
|
| 25737 |
+
"epoch": 0.3254635187408297,
|
| 25738 |
+
"grad_norm": 0.07944405823945999,
|
| 25739 |
+
"learning_rate": 0.000987599144570532,
|
| 25740 |
+
"loss": 1.5583,
|
| 25741 |
+
"step": 7320
|
| 25742 |
+
},
|
| 25743 |
+
{
|
| 25744 |
+
"epoch": 0.32555244319950205,
|
| 25745 |
+
"grad_norm": 0.0833417996764183,
|
| 25746 |
+
"learning_rate": 0.000987591327317123,
|
| 25747 |
+
"loss": 1.5588,
|
| 25748 |
+
"step": 7322
|
| 25749 |
+
},
|
| 25750 |
+
{
|
| 25751 |
+
"epoch": 0.32564136765817436,
|
| 25752 |
+
"grad_norm": 0.08267171680927277,
|
| 25753 |
+
"learning_rate": 0.0009875835076315273,
|
| 25754 |
+
"loss": 1.5576,
|
| 25755 |
+
"step": 7324
|
| 25756 |
+
},
|
| 25757 |
+
{
|
| 25758 |
+
"epoch": 0.32573029211684673,
|
| 25759 |
+
"grad_norm": 0.08149762451648712,
|
| 25760 |
+
"learning_rate": 0.0009875756855137845,
|
| 25761 |
+
"loss": 1.5535,
|
| 25762 |
+
"step": 7326
|
| 25763 |
+
},
|
| 25764 |
+
{
|
| 25765 |
+
"epoch": 0.3258192165755191,
|
| 25766 |
+
"grad_norm": 0.0837164968252182,
|
| 25767 |
+
"learning_rate": 0.0009875678609639331,
|
| 25768 |
+
"loss": 1.5554,
|
| 25769 |
+
"step": 7328
|
| 25770 |
+
},
|
| 25771 |
+
{
|
| 25772 |
+
"epoch": 0.32590814103419147,
|
| 25773 |
+
"grad_norm": 0.08050676435232162,
|
| 25774 |
+
"learning_rate": 0.0009875600339820123,
|
| 25775 |
+
"loss": 1.561,
|
| 25776 |
+
"step": 7330
|
| 25777 |
+
},
|
| 25778 |
+
{
|
| 25779 |
+
"epoch": 0.32599706549286384,
|
| 25780 |
+
"grad_norm": 0.08078650385141373,
|
| 25781 |
+
"learning_rate": 0.000987552204568061,
|
| 25782 |
+
"loss": 1.5543,
|
| 25783 |
+
"step": 7332
|
| 25784 |
+
},
|
| 25785 |
+
{
|
| 25786 |
+
"epoch": 0.32608598995153615,
|
| 25787 |
+
"grad_norm": 0.08506888151168823,
|
| 25788 |
+
"learning_rate": 0.0009875443727221185,
|
| 25789 |
+
"loss": 1.5606,
|
| 25790 |
+
"step": 7334
|
| 25791 |
+
},
|
| 25792 |
+
{
|
| 25793 |
+
"epoch": 0.3261749144102085,
|
| 25794 |
+
"grad_norm": 0.08433322608470917,
|
| 25795 |
+
"learning_rate": 0.0009875365384442237,
|
| 25796 |
+
"loss": 1.5563,
|
| 25797 |
+
"step": 7336
|
| 25798 |
+
},
|
| 25799 |
+
{
|
| 25800 |
+
"epoch": 0.3262638388688809,
|
| 25801 |
+
"grad_norm": 0.08262262493371964,
|
| 25802 |
+
"learning_rate": 0.000987528701734416,
|
| 25803 |
+
"loss": 1.565,
|
| 25804 |
+
"step": 7338
|
| 25805 |
+
},
|
| 25806 |
+
{
|
| 25807 |
+
"epoch": 0.32635276332755325,
|
| 25808 |
+
"grad_norm": 0.08206149190664291,
|
| 25809 |
+
"learning_rate": 0.0009875208625927338,
|
| 25810 |
+
"loss": 1.5562,
|
| 25811 |
+
"step": 7340
|
| 25812 |
+
},
|
| 25813 |
+
{
|
| 25814 |
+
"epoch": 0.3264416877862256,
|
| 25815 |
+
"grad_norm": 0.08632387220859528,
|
| 25816 |
+
"learning_rate": 0.0009875130210192167,
|
| 25817 |
+
"loss": 1.5566,
|
| 25818 |
+
"step": 7342
|
| 25819 |
+
},
|
| 25820 |
+
{
|
| 25821 |
+
"epoch": 0.32653061224489793,
|
| 25822 |
+
"grad_norm": 0.08040108531713486,
|
| 25823 |
+
"learning_rate": 0.0009875051770139038,
|
| 25824 |
+
"loss": 1.5591,
|
| 25825 |
+
"step": 7344
|
| 25826 |
+
},
|
| 25827 |
+
{
|
| 25828 |
+
"epoch": 0.3266195367035703,
|
| 25829 |
+
"grad_norm": 0.08796191215515137,
|
| 25830 |
+
"learning_rate": 0.0009874973305768343,
|
| 25831 |
+
"loss": 1.5635,
|
| 25832 |
+
"step": 7346
|
| 25833 |
+
},
|
| 25834 |
+
{
|
| 25835 |
+
"epoch": 0.32670846116224267,
|
| 25836 |
+
"grad_norm": 0.08850188553333282,
|
| 25837 |
+
"learning_rate": 0.000987489481708047,
|
| 25838 |
+
"loss": 1.5547,
|
| 25839 |
+
"step": 7348
|
| 25840 |
+
},
|
| 25841 |
+
{
|
| 25842 |
+
"epoch": 0.32679738562091504,
|
| 25843 |
+
"grad_norm": 0.08623901754617691,
|
| 25844 |
+
"learning_rate": 0.000987481630407581,
|
| 25845 |
+
"loss": 1.5583,
|
| 25846 |
+
"step": 7350
|
| 25847 |
+
},
|
| 25848 |
+
{
|
| 25849 |
+
"epoch": 0.3268863100795874,
|
| 25850 |
+
"grad_norm": 0.08186235278844833,
|
| 25851 |
+
"learning_rate": 0.0009874737766754763,
|
| 25852 |
+
"loss": 1.5568,
|
| 25853 |
+
"step": 7352
|
| 25854 |
+
},
|
| 25855 |
+
{
|
| 25856 |
+
"epoch": 0.3269752345382598,
|
| 25857 |
+
"grad_norm": 0.08264327794313431,
|
| 25858 |
+
"learning_rate": 0.000987465920511771,
|
| 25859 |
+
"loss": 1.5589,
|
| 25860 |
+
"step": 7354
|
| 25861 |
+
},
|
| 25862 |
+
{
|
| 25863 |
+
"epoch": 0.3270641589969321,
|
| 25864 |
+
"grad_norm": 0.08318621665239334,
|
| 25865 |
+
"learning_rate": 0.000987458061916505,
|
| 25866 |
+
"loss": 1.5578,
|
| 25867 |
+
"step": 7356
|
| 25868 |
+
},
|
| 25869 |
+
{
|
| 25870 |
+
"epoch": 0.32715308345560445,
|
| 25871 |
+
"grad_norm": 0.08507546037435532,
|
| 25872 |
+
"learning_rate": 0.0009874502008897174,
|
| 25873 |
+
"loss": 1.5628,
|
| 25874 |
+
"step": 7358
|
| 25875 |
+
},
|
| 25876 |
+
{
|
| 25877 |
+
"epoch": 0.3272420079142768,
|
| 25878 |
+
"grad_norm": 0.08034215867519379,
|
| 25879 |
+
"learning_rate": 0.000987442337431447,
|
| 25880 |
+
"loss": 1.5597,
|
| 25881 |
+
"step": 7360
|
| 25882 |
+
},
|
| 25883 |
+
{
|
| 25884 |
+
"epoch": 0.3273309323729492,
|
| 25885 |
+
"grad_norm": 0.08105006068944931,
|
| 25886 |
+
"learning_rate": 0.0009874344715417333,
|
| 25887 |
+
"loss": 1.5562,
|
| 25888 |
+
"step": 7362
|
| 25889 |
+
},
|
| 25890 |
+
{
|
| 25891 |
+
"epoch": 0.32741985683162156,
|
| 25892 |
+
"grad_norm": 0.07907504588365555,
|
| 25893 |
+
"learning_rate": 0.0009874266032206156,
|
| 25894 |
+
"loss": 1.5649,
|
| 25895 |
+
"step": 7364
|
| 25896 |
+
},
|
| 25897 |
+
{
|
| 25898 |
+
"epoch": 0.32750878129029387,
|
| 25899 |
+
"grad_norm": 0.08351552486419678,
|
| 25900 |
+
"learning_rate": 0.0009874187324681331,
|
| 25901 |
+
"loss": 1.563,
|
| 25902 |
+
"step": 7366
|
| 25903 |
+
},
|
| 25904 |
+
{
|
| 25905 |
+
"epoch": 0.32759770574896624,
|
| 25906 |
+
"grad_norm": 0.07940589636564255,
|
| 25907 |
+
"learning_rate": 0.0009874108592843253,
|
| 25908 |
+
"loss": 1.5597,
|
| 25909 |
+
"step": 7368
|
| 25910 |
+
},
|
| 25911 |
+
{
|
| 25912 |
+
"epoch": 0.3276866302076386,
|
| 25913 |
+
"grad_norm": 0.0827498808503151,
|
| 25914 |
+
"learning_rate": 0.000987402983669231,
|
| 25915 |
+
"loss": 1.5544,
|
| 25916 |
+
"step": 7370
|
| 25917 |
+
},
|
| 25918 |
+
{
|
| 25919 |
+
"epoch": 0.327775554666311,
|
| 25920 |
+
"grad_norm": 0.08331042528152466,
|
| 25921 |
+
"learning_rate": 0.0009873951056228898,
|
| 25922 |
+
"loss": 1.56,
|
| 25923 |
+
"step": 7372
|
| 25924 |
+
},
|
| 25925 |
+
{
|
| 25926 |
+
"epoch": 0.32786447912498334,
|
| 25927 |
+
"grad_norm": 0.08180607855319977,
|
| 25928 |
+
"learning_rate": 0.000987387225145341,
|
| 25929 |
+
"loss": 1.5565,
|
| 25930 |
+
"step": 7374
|
| 25931 |
+
},
|
| 25932 |
+
{
|
| 25933 |
+
"epoch": 0.3279534035836557,
|
| 25934 |
+
"grad_norm": 0.08340569585561752,
|
| 25935 |
+
"learning_rate": 0.0009873793422366234,
|
| 25936 |
+
"loss": 1.5608,
|
| 25937 |
+
"step": 7376
|
| 25938 |
+
},
|
| 25939 |
+
{
|
| 25940 |
+
"epoch": 0.328042328042328,
|
| 25941 |
+
"grad_norm": 0.08175882697105408,
|
| 25942 |
+
"learning_rate": 0.000987371456896777,
|
| 25943 |
+
"loss": 1.5607,
|
| 25944 |
+
"step": 7378
|
| 25945 |
+
},
|
| 25946 |
+
{
|
| 25947 |
+
"epoch": 0.3281312525010004,
|
| 25948 |
+
"grad_norm": 0.0852338895201683,
|
| 25949 |
+
"learning_rate": 0.000987363569125841,
|
| 25950 |
+
"loss": 1.5584,
|
| 25951 |
+
"step": 7380
|
| 25952 |
+
},
|
| 25953 |
+
{
|
| 25954 |
+
"epoch": 0.32822017695967276,
|
| 25955 |
+
"grad_norm": 0.0846090242266655,
|
| 25956 |
+
"learning_rate": 0.0009873556789238545,
|
| 25957 |
+
"loss": 1.5586,
|
| 25958 |
+
"step": 7382
|
| 25959 |
+
},
|
| 25960 |
+
{
|
| 25961 |
+
"epoch": 0.32830910141834513,
|
| 25962 |
+
"grad_norm": 0.07967184484004974,
|
| 25963 |
+
"learning_rate": 0.000987347786290857,
|
| 25964 |
+
"loss": 1.5576,
|
| 25965 |
+
"step": 7384
|
| 25966 |
+
},
|
| 25967 |
+
{
|
| 25968 |
+
"epoch": 0.3283980258770175,
|
| 25969 |
+
"grad_norm": 0.08349616080522537,
|
| 25970 |
+
"learning_rate": 0.000987339891226888,
|
| 25971 |
+
"loss": 1.5628,
|
| 25972 |
+
"step": 7386
|
| 25973 |
+
},
|
| 25974 |
+
{
|
| 25975 |
+
"epoch": 0.3284869503356898,
|
| 25976 |
+
"grad_norm": 0.08276060223579407,
|
| 25977 |
+
"learning_rate": 0.0009873319937319864,
|
| 25978 |
+
"loss": 1.5597,
|
| 25979 |
+
"step": 7388
|
| 25980 |
+
},
|
| 25981 |
+
{
|
| 25982 |
+
"epoch": 0.3285758747943622,
|
| 25983 |
+
"grad_norm": 0.08043144643306732,
|
| 25984 |
+
"learning_rate": 0.0009873240938061921,
|
| 25985 |
+
"loss": 1.5537,
|
| 25986 |
+
"step": 7390
|
| 25987 |
+
},
|
| 25988 |
+
{
|
| 25989 |
+
"epoch": 0.32866479925303455,
|
| 25990 |
+
"grad_norm": 0.08394316583871841,
|
| 25991 |
+
"learning_rate": 0.0009873161914495444,
|
| 25992 |
+
"loss": 1.5573,
|
| 25993 |
+
"step": 7392
|
| 25994 |
+
},
|
| 25995 |
+
{
|
| 25996 |
+
"epoch": 0.3287537237117069,
|
| 25997 |
+
"grad_norm": 0.08426385372877121,
|
| 25998 |
+
"learning_rate": 0.0009873082866620824,
|
| 25999 |
+
"loss": 1.558,
|
| 26000 |
+
"step": 7394
|
| 26001 |
+
},
|
| 26002 |
+
{
|
| 26003 |
+
"epoch": 0.3288426481703793,
|
| 26004 |
+
"grad_norm": 0.08154802769422531,
|
| 26005 |
+
"learning_rate": 0.000987300379443846,
|
| 26006 |
+
"loss": 1.5572,
|
| 26007 |
+
"step": 7396
|
| 26008 |
+
},
|
| 26009 |
+
{
|
| 26010 |
+
"epoch": 0.3289315726290516,
|
| 26011 |
+
"grad_norm": 0.08731735497713089,
|
| 26012 |
+
"learning_rate": 0.0009872924697948745,
|
| 26013 |
+
"loss": 1.5567,
|
| 26014 |
+
"step": 7398
|
| 26015 |
+
},
|
| 26016 |
+
{
|
| 26017 |
+
"epoch": 0.32902049708772396,
|
| 26018 |
+
"grad_norm": 0.08999709039926529,
|
| 26019 |
+
"learning_rate": 0.000987284557715207,
|
| 26020 |
+
"loss": 1.556,
|
| 26021 |
+
"step": 7400
|
| 26022 |
+
},
|
| 26023 |
+
{
|
| 26024 |
+
"epoch": 0.32910942154639633,
|
| 26025 |
+
"grad_norm": 0.08462511003017426,
|
| 26026 |
+
"learning_rate": 0.0009872766432048835,
|
| 26027 |
+
"loss": 1.5585,
|
| 26028 |
+
"step": 7402
|
| 26029 |
+
},
|
| 26030 |
+
{
|
| 26031 |
+
"epoch": 0.3291983460050687,
|
| 26032 |
+
"grad_norm": 0.08666058629751205,
|
| 26033 |
+
"learning_rate": 0.000987268726263943,
|
| 26034 |
+
"loss": 1.5567,
|
| 26035 |
+
"step": 7404
|
| 26036 |
+
},
|
| 26037 |
+
{
|
| 26038 |
+
"epoch": 0.32928727046374107,
|
| 26039 |
+
"grad_norm": 0.08502695709466934,
|
| 26040 |
+
"learning_rate": 0.0009872608068924253,
|
| 26041 |
+
"loss": 1.5566,
|
| 26042 |
+
"step": 7406
|
| 26043 |
+
},
|
| 26044 |
+
{
|
| 26045 |
+
"epoch": 0.32937619492241343,
|
| 26046 |
+
"grad_norm": 0.08488810807466507,
|
| 26047 |
+
"learning_rate": 0.0009872528850903698,
|
| 26048 |
+
"loss": 1.5529,
|
| 26049 |
+
"step": 7408
|
| 26050 |
+
},
|
| 26051 |
+
{
|
| 26052 |
+
"epoch": 0.32946511938108575,
|
| 26053 |
+
"grad_norm": 0.08286993950605392,
|
| 26054 |
+
"learning_rate": 0.000987244960857816,
|
| 26055 |
+
"loss": 1.5639,
|
| 26056 |
+
"step": 7410
|
| 26057 |
+
},
|
| 26058 |
+
{
|
| 26059 |
+
"epoch": 0.3295540438397581,
|
| 26060 |
+
"grad_norm": 0.0867101177573204,
|
| 26061 |
+
"learning_rate": 0.0009872370341948036,
|
| 26062 |
+
"loss": 1.559,
|
| 26063 |
+
"step": 7412
|
| 26064 |
+
},
|
| 26065 |
+
{
|
| 26066 |
+
"epoch": 0.3296429682984305,
|
| 26067 |
+
"grad_norm": 0.0828615054488182,
|
| 26068 |
+
"learning_rate": 0.0009872291051013719,
|
| 26069 |
+
"loss": 1.5587,
|
| 26070 |
+
"step": 7414
|
| 26071 |
+
},
|
| 26072 |
+
{
|
| 26073 |
+
"epoch": 0.32973189275710285,
|
| 26074 |
+
"grad_norm": 0.08587847650051117,
|
| 26075 |
+
"learning_rate": 0.0009872211735775605,
|
| 26076 |
+
"loss": 1.5599,
|
| 26077 |
+
"step": 7416
|
| 26078 |
+
},
|
| 26079 |
+
{
|
| 26080 |
+
"epoch": 0.3298208172157752,
|
| 26081 |
+
"grad_norm": 0.08348828554153442,
|
| 26082 |
+
"learning_rate": 0.000987213239623409,
|
| 26083 |
+
"loss": 1.5556,
|
| 26084 |
+
"step": 7418
|
| 26085 |
+
},
|
| 26086 |
+
{
|
| 26087 |
+
"epoch": 0.32990974167444753,
|
| 26088 |
+
"grad_norm": 0.08008415997028351,
|
| 26089 |
+
"learning_rate": 0.0009872053032389573,
|
| 26090 |
+
"loss": 1.5642,
|
| 26091 |
+
"step": 7420
|
| 26092 |
+
},
|
| 26093 |
+
{
|
| 26094 |
+
"epoch": 0.3299986661331199,
|
| 26095 |
+
"grad_norm": 0.0780973806977272,
|
| 26096 |
+
"learning_rate": 0.0009871973644242442,
|
| 26097 |
+
"loss": 1.5563,
|
| 26098 |
+
"step": 7422
|
| 26099 |
+
},
|
| 26100 |
+
{
|
| 26101 |
+
"epoch": 0.33008759059179227,
|
| 26102 |
+
"grad_norm": 0.08216983824968338,
|
| 26103 |
+
"learning_rate": 0.00098718942317931,
|
| 26104 |
+
"loss": 1.5608,
|
| 26105 |
+
"step": 7424
|
| 26106 |
+
},
|
| 26107 |
+
{
|
| 26108 |
+
"epoch": 0.33017651505046464,
|
| 26109 |
+
"grad_norm": 0.08794374763965607,
|
| 26110 |
+
"learning_rate": 0.0009871814795041941,
|
| 26111 |
+
"loss": 1.5622,
|
| 26112 |
+
"step": 7426
|
| 26113 |
+
},
|
| 26114 |
+
{
|
| 26115 |
+
"epoch": 0.330265439509137,
|
| 26116 |
+
"grad_norm": 0.09090631455183029,
|
| 26117 |
+
"learning_rate": 0.0009871735333989362,
|
| 26118 |
+
"loss": 1.5562,
|
| 26119 |
+
"step": 7428
|
| 26120 |
+
},
|
| 26121 |
+
{
|
| 26122 |
+
"epoch": 0.3303543639678094,
|
| 26123 |
+
"grad_norm": 0.08172343671321869,
|
| 26124 |
+
"learning_rate": 0.0009871655848635757,
|
| 26125 |
+
"loss": 1.5569,
|
| 26126 |
+
"step": 7430
|
| 26127 |
+
},
|
| 26128 |
+
{
|
| 26129 |
+
"epoch": 0.3304432884264817,
|
| 26130 |
+
"grad_norm": 0.08407451957464218,
|
| 26131 |
+
"learning_rate": 0.0009871576338981525,
|
| 26132 |
+
"loss": 1.5586,
|
| 26133 |
+
"step": 7432
|
| 26134 |
+
},
|
| 26135 |
+
{
|
| 26136 |
+
"epoch": 0.33053221288515405,
|
| 26137 |
+
"grad_norm": 0.08738183975219727,
|
| 26138 |
+
"learning_rate": 0.000987149680502706,
|
| 26139 |
+
"loss": 1.5631,
|
| 26140 |
+
"step": 7434
|
| 26141 |
+
},
|
| 26142 |
+
{
|
| 26143 |
+
"epoch": 0.3306211373438264,
|
| 26144 |
+
"grad_norm": 0.09222719818353653,
|
| 26145 |
+
"learning_rate": 0.000987141724677276,
|
| 26146 |
+
"loss": 1.5581,
|
| 26147 |
+
"step": 7436
|
| 26148 |
+
},
|
| 26149 |
+
{
|
| 26150 |
+
"epoch": 0.3307100618024988,
|
| 26151 |
+
"grad_norm": 0.08091361075639725,
|
| 26152 |
+
"learning_rate": 0.0009871337664219024,
|
| 26153 |
+
"loss": 1.5561,
|
| 26154 |
+
"step": 7438
|
| 26155 |
+
},
|
| 26156 |
+
{
|
| 26157 |
+
"epoch": 0.33079898626117116,
|
| 26158 |
+
"grad_norm": 0.09107191115617752,
|
| 26159 |
+
"learning_rate": 0.0009871258057366247,
|
| 26160 |
+
"loss": 1.5559,
|
| 26161 |
+
"step": 7440
|
| 26162 |
+
},
|
| 26163 |
+
{
|
| 26164 |
+
"epoch": 0.33088791071984347,
|
| 26165 |
+
"grad_norm": 0.08714121580123901,
|
| 26166 |
+
"learning_rate": 0.0009871178426214826,
|
| 26167 |
+
"loss": 1.5662,
|
| 26168 |
+
"step": 7442
|
| 26169 |
+
},
|
| 26170 |
+
{
|
| 26171 |
+
"epoch": 0.33097683517851584,
|
| 26172 |
+
"grad_norm": 0.0870809406042099,
|
| 26173 |
+
"learning_rate": 0.000987109877076516,
|
| 26174 |
+
"loss": 1.5584,
|
| 26175 |
+
"step": 7444
|
| 26176 |
+
},
|
| 26177 |
+
{
|
| 26178 |
+
"epoch": 0.3310657596371882,
|
| 26179 |
+
"grad_norm": 0.0869300365447998,
|
| 26180 |
+
"learning_rate": 0.0009871019091017641,
|
| 26181 |
+
"loss": 1.5596,
|
| 26182 |
+
"step": 7446
|
| 26183 |
+
},
|
| 26184 |
+
{
|
| 26185 |
+
"epoch": 0.3311546840958606,
|
| 26186 |
+
"grad_norm": 0.08757723122835159,
|
| 26187 |
+
"learning_rate": 0.0009870939386972674,
|
| 26188 |
+
"loss": 1.558,
|
| 26189 |
+
"step": 7448
|
| 26190 |
+
},
|
| 26191 |
+
{
|
| 26192 |
+
"epoch": 0.33124360855453294,
|
| 26193 |
+
"grad_norm": 0.08233436197042465,
|
| 26194 |
+
"learning_rate": 0.000987085965863065,
|
| 26195 |
+
"loss": 1.5573,
|
| 26196 |
+
"step": 7450
|
| 26197 |
+
},
|
| 26198 |
+
{
|
| 26199 |
+
"epoch": 0.33133253301320525,
|
| 26200 |
+
"grad_norm": 0.08261076360940933,
|
| 26201 |
+
"learning_rate": 0.0009870779905991971,
|
| 26202 |
+
"loss": 1.5639,
|
| 26203 |
+
"step": 7452
|
| 26204 |
+
},
|
| 26205 |
+
{
|
| 26206 |
+
"epoch": 0.3314214574718776,
|
| 26207 |
+
"grad_norm": 0.08498179167509079,
|
| 26208 |
+
"learning_rate": 0.0009870700129057032,
|
| 26209 |
+
"loss": 1.5584,
|
| 26210 |
+
"step": 7454
|
| 26211 |
+
},
|
| 26212 |
+
{
|
| 26213 |
+
"epoch": 0.33151038193055,
|
| 26214 |
+
"grad_norm": 0.08367268741130829,
|
| 26215 |
+
"learning_rate": 0.0009870620327826233,
|
| 26216 |
+
"loss": 1.5633,
|
| 26217 |
+
"step": 7456
|
| 26218 |
+
},
|
| 26219 |
+
{
|
| 26220 |
+
"epoch": 0.33159930638922236,
|
| 26221 |
+
"grad_norm": 0.08087131381034851,
|
| 26222 |
+
"learning_rate": 0.0009870540502299973,
|
| 26223 |
+
"loss": 1.5592,
|
| 26224 |
+
"step": 7458
|
| 26225 |
+
},
|
| 26226 |
+
{
|
| 26227 |
+
"epoch": 0.3316882308478947,
|
| 26228 |
+
"grad_norm": 0.08614420145750046,
|
| 26229 |
+
"learning_rate": 0.0009870460652478645,
|
| 26230 |
+
"loss": 1.5625,
|
| 26231 |
+
"step": 7460
|
| 26232 |
+
},
|
| 26233 |
+
{
|
| 26234 |
+
"epoch": 0.3317771553065671,
|
| 26235 |
+
"grad_norm": 0.08858582377433777,
|
| 26236 |
+
"learning_rate": 0.0009870380778362654,
|
| 26237 |
+
"loss": 1.5612,
|
| 26238 |
+
"step": 7462
|
| 26239 |
+
},
|
| 26240 |
+
{
|
| 26241 |
+
"epoch": 0.3318660797652394,
|
| 26242 |
+
"grad_norm": 0.08249551802873611,
|
| 26243 |
+
"learning_rate": 0.0009870300879952394,
|
| 26244 |
+
"loss": 1.5645,
|
| 26245 |
+
"step": 7464
|
| 26246 |
+
},
|
| 26247 |
+
{
|
| 26248 |
+
"epoch": 0.3319550042239118,
|
| 26249 |
+
"grad_norm": 0.08597932755947113,
|
| 26250 |
+
"learning_rate": 0.0009870220957248264,
|
| 26251 |
+
"loss": 1.5571,
|
| 26252 |
+
"step": 7466
|
| 26253 |
+
},
|
| 26254 |
+
{
|
| 26255 |
+
"epoch": 0.33204392868258414,
|
| 26256 |
+
"grad_norm": 0.08219864219427109,
|
| 26257 |
+
"learning_rate": 0.0009870141010250666,
|
| 26258 |
+
"loss": 1.5608,
|
| 26259 |
+
"step": 7468
|
| 26260 |
+
},
|
| 26261 |
+
{
|
| 26262 |
+
"epoch": 0.3321328531412565,
|
| 26263 |
+
"grad_norm": 0.08492123335599899,
|
| 26264 |
+
"learning_rate": 0.0009870061038959994,
|
| 26265 |
+
"loss": 1.5632,
|
| 26266 |
+
"step": 7470
|
| 26267 |
+
},
|
| 26268 |
+
{
|
| 26269 |
+
"epoch": 0.3322217775999289,
|
| 26270 |
+
"grad_norm": 0.08149886131286621,
|
| 26271 |
+
"learning_rate": 0.0009869981043376648,
|
| 26272 |
+
"loss": 1.5541,
|
| 26273 |
+
"step": 7472
|
| 26274 |
+
},
|
| 26275 |
+
{
|
| 26276 |
+
"epoch": 0.3323107020586012,
|
| 26277 |
+
"grad_norm": 0.08603468537330627,
|
| 26278 |
+
"learning_rate": 0.000986990102350103,
|
| 26279 |
+
"loss": 1.5528,
|
| 26280 |
+
"step": 7474
|
| 26281 |
+
},
|
| 26282 |
+
{
|
| 26283 |
+
"epoch": 0.33239962651727356,
|
| 26284 |
+
"grad_norm": 0.0839407667517662,
|
| 26285 |
+
"learning_rate": 0.0009869820979333539,
|
| 26286 |
+
"loss": 1.5573,
|
| 26287 |
+
"step": 7476
|
| 26288 |
+
},
|
| 26289 |
+
{
|
| 26290 |
+
"epoch": 0.33248855097594593,
|
| 26291 |
+
"grad_norm": 0.07948873937129974,
|
| 26292 |
+
"learning_rate": 0.0009869740910874569,
|
| 26293 |
+
"loss": 1.5646,
|
| 26294 |
+
"step": 7478
|
| 26295 |
+
},
|
| 26296 |
+
{
|
| 26297 |
+
"epoch": 0.3325774754346183,
|
| 26298 |
+
"grad_norm": 0.08331003040075302,
|
| 26299 |
+
"learning_rate": 0.0009869660818124526,
|
| 26300 |
+
"loss": 1.5538,
|
| 26301 |
+
"step": 7480
|
| 26302 |
+
},
|
| 26303 |
+
{
|
| 26304 |
+
"epoch": 0.33266639989329067,
|
| 26305 |
+
"grad_norm": 0.08100567013025284,
|
| 26306 |
+
"learning_rate": 0.0009869580701083805,
|
| 26307 |
+
"loss": 1.5628,
|
| 26308 |
+
"step": 7482
|
| 26309 |
+
},
|
| 26310 |
+
{
|
| 26311 |
+
"epoch": 0.33275532435196303,
|
| 26312 |
+
"grad_norm": 0.08268828690052032,
|
| 26313 |
+
"learning_rate": 0.0009869500559752807,
|
| 26314 |
+
"loss": 1.5591,
|
| 26315 |
+
"step": 7484
|
| 26316 |
+
},
|
| 26317 |
+
{
|
| 26318 |
+
"epoch": 0.33284424881063535,
|
| 26319 |
+
"grad_norm": 0.08128658682107925,
|
| 26320 |
+
"learning_rate": 0.0009869420394131932,
|
| 26321 |
+
"loss": 1.559,
|
| 26322 |
+
"step": 7486
|
| 26323 |
+
},
|
| 26324 |
+
{
|
| 26325 |
+
"epoch": 0.3329331732693077,
|
| 26326 |
+
"grad_norm": 0.07909125089645386,
|
| 26327 |
+
"learning_rate": 0.0009869340204221582,
|
| 26328 |
+
"loss": 1.5581,
|
| 26329 |
+
"step": 7488
|
| 26330 |
+
},
|
| 26331 |
+
{
|
| 26332 |
+
"epoch": 0.3330220977279801,
|
| 26333 |
+
"grad_norm": 0.08491683751344681,
|
| 26334 |
+
"learning_rate": 0.0009869259990022152,
|
| 26335 |
+
"loss": 1.5523,
|
| 26336 |
+
"step": 7490
|
| 26337 |
+
},
|
| 26338 |
+
{
|
| 26339 |
+
"epoch": 0.33311102218665245,
|
| 26340 |
+
"grad_norm": 0.08070467412471771,
|
| 26341 |
+
"learning_rate": 0.0009869179751534046,
|
| 26342 |
+
"loss": 1.5558,
|
| 26343 |
+
"step": 7492
|
| 26344 |
+
},
|
| 26345 |
+
{
|
| 26346 |
+
"epoch": 0.3331999466453248,
|
| 26347 |
+
"grad_norm": 0.08046550303697586,
|
| 26348 |
+
"learning_rate": 0.0009869099488757662,
|
| 26349 |
+
"loss": 1.5511,
|
| 26350 |
+
"step": 7494
|
| 26351 |
+
},
|
| 26352 |
+
{
|
| 26353 |
+
"epoch": 0.33328887110399713,
|
| 26354 |
+
"grad_norm": 0.08419077098369598,
|
| 26355 |
+
"learning_rate": 0.0009869019201693403,
|
| 26356 |
+
"loss": 1.5632,
|
| 26357 |
+
"step": 7496
|
| 26358 |
+
},
|
| 26359 |
+
{
|
| 26360 |
+
"epoch": 0.3333777955626695,
|
| 26361 |
+
"grad_norm": 0.07918644696474075,
|
| 26362 |
+
"learning_rate": 0.0009868938890341668,
|
| 26363 |
+
"loss": 1.5552,
|
| 26364 |
+
"step": 7498
|
| 26365 |
+
},
|
| 26366 |
+
{
|
| 26367 |
+
"epoch": 0.33346672002134187,
|
| 26368 |
+
"grad_norm": 0.08392981439828873,
|
| 26369 |
+
"learning_rate": 0.0009868858554702856,
|
| 26370 |
+
"loss": 1.5588,
|
| 26371 |
+
"step": 7500
|
| 26372 |
+
},
|
| 26373 |
+
{
|
| 26374 |
+
"epoch": 0.33346672002134187,
|
| 26375 |
+
"eval_loss": 1.5330219268798828,
|
| 26376 |
+
"eval_runtime": 12.4092,
|
| 26377 |
+
"eval_samples_per_second": 556.847,
|
| 26378 |
+
"eval_steps_per_second": 69.626,
|
| 26379 |
+
"step": 7500
|
| 26380 |
+
},
|
| 26381 |
+
{
|
| 26382 |
+
"epoch": 0.33355564448001424,
|
| 26383 |
+
"grad_norm": 0.09008117020130157,
|
| 26384 |
+
"learning_rate": 0.0009868778194777371,
|
| 26385 |
+
"loss": 1.5609,
|
| 26386 |
+
"step": 7502
|
| 26387 |
+
},
|
| 26388 |
+
{
|
| 26389 |
+
"epoch": 0.3336445689386866,
|
| 26390 |
+
"grad_norm": 0.08291777223348618,
|
| 26391 |
+
"learning_rate": 0.0009868697810565613,
|
| 26392 |
+
"loss": 1.5649,
|
| 26393 |
+
"step": 7504
|
| 26394 |
+
},
|
| 26395 |
+
{
|
| 26396 |
+
"epoch": 0.33373349339735897,
|
| 26397 |
+
"grad_norm": 0.08472921699285507,
|
| 26398 |
+
"learning_rate": 0.0009868617402067979,
|
| 26399 |
+
"loss": 1.5604,
|
| 26400 |
+
"step": 7506
|
| 26401 |
+
},
|
| 26402 |
+
{
|
| 26403 |
+
"epoch": 0.3338224178560313,
|
| 26404 |
+
"grad_norm": 0.08425983041524887,
|
| 26405 |
+
"learning_rate": 0.0009868536969284876,
|
| 26406 |
+
"loss": 1.5625,
|
| 26407 |
+
"step": 7508
|
| 26408 |
+
},
|
| 26409 |
+
{
|
| 26410 |
+
"epoch": 0.33391134231470365,
|
| 26411 |
+
"grad_norm": 0.08368197828531265,
|
| 26412 |
+
"learning_rate": 0.0009868456512216702,
|
| 26413 |
+
"loss": 1.5517,
|
| 26414 |
+
"step": 7510
|
| 26415 |
+
},
|
| 26416 |
+
{
|
| 26417 |
+
"epoch": 0.334000266773376,
|
| 26418 |
+
"grad_norm": 0.07804959267377853,
|
| 26419 |
+
"learning_rate": 0.0009868376030863857,
|
| 26420 |
+
"loss": 1.5606,
|
| 26421 |
+
"step": 7512
|
| 26422 |
+
},
|
| 26423 |
+
{
|
| 26424 |
+
"epoch": 0.3340891912320484,
|
| 26425 |
+
"grad_norm": 0.08549796044826508,
|
| 26426 |
+
"learning_rate": 0.0009868295525226746,
|
| 26427 |
+
"loss": 1.5547,
|
| 26428 |
+
"step": 7514
|
| 26429 |
+
},
|
| 26430 |
+
{
|
| 26431 |
+
"epoch": 0.33417811569072076,
|
| 26432 |
+
"grad_norm": 0.08111416548490524,
|
| 26433 |
+
"learning_rate": 0.0009868214995305768,
|
| 26434 |
+
"loss": 1.5567,
|
| 26435 |
+
"step": 7516
|
| 26436 |
+
},
|
| 26437 |
+
{
|
| 26438 |
+
"epoch": 0.33426704014939307,
|
| 26439 |
+
"grad_norm": 0.07972585409879684,
|
| 26440 |
+
"learning_rate": 0.0009868134441101326,
|
| 26441 |
+
"loss": 1.5515,
|
| 26442 |
+
"step": 7518
|
| 26443 |
+
},
|
| 26444 |
+
{
|
| 26445 |
+
"epoch": 0.33435596460806544,
|
| 26446 |
+
"grad_norm": 0.0834115669131279,
|
| 26447 |
+
"learning_rate": 0.0009868053862613821,
|
| 26448 |
+
"loss": 1.5563,
|
| 26449 |
+
"step": 7520
|
| 26450 |
+
},
|
| 26451 |
+
{
|
| 26452 |
+
"epoch": 0.3344448890667378,
|
| 26453 |
+
"grad_norm": 0.08134639263153076,
|
| 26454 |
+
"learning_rate": 0.0009867973259843656,
|
| 26455 |
+
"loss": 1.5539,
|
| 26456 |
+
"step": 7522
|
| 26457 |
+
},
|
| 26458 |
+
{
|
| 26459 |
+
"epoch": 0.3345338135254102,
|
| 26460 |
+
"grad_norm": 0.07953674346208572,
|
| 26461 |
+
"learning_rate": 0.0009867892632791232,
|
| 26462 |
+
"loss": 1.5566,
|
| 26463 |
+
"step": 7524
|
| 26464 |
+
},
|
| 26465 |
+
{
|
| 26466 |
+
"epoch": 0.33462273798408254,
|
| 26467 |
+
"grad_norm": 0.07886262983083725,
|
| 26468 |
+
"learning_rate": 0.000986781198145695,
|
| 26469 |
+
"loss": 1.5549,
|
| 26470 |
+
"step": 7526
|
| 26471 |
+
},
|
| 26472 |
+
{
|
| 26473 |
+
"epoch": 0.33471166244275485,
|
| 26474 |
+
"grad_norm": 0.08340603113174438,
|
| 26475 |
+
"learning_rate": 0.0009867731305841217,
|
| 26476 |
+
"loss": 1.5609,
|
| 26477 |
+
"step": 7528
|
| 26478 |
+
},
|
| 26479 |
+
{
|
| 26480 |
+
"epoch": 0.3348005869014272,
|
| 26481 |
+
"grad_norm": 0.08290226757526398,
|
| 26482 |
+
"learning_rate": 0.000986765060594443,
|
| 26483 |
+
"loss": 1.5561,
|
| 26484 |
+
"step": 7530
|
| 26485 |
+
},
|
| 26486 |
+
{
|
| 26487 |
+
"epoch": 0.3348895113600996,
|
| 26488 |
+
"grad_norm": 0.08421459048986435,
|
| 26489 |
+
"learning_rate": 0.0009867569881766994,
|
| 26490 |
+
"loss": 1.5549,
|
| 26491 |
+
"step": 7532
|
| 26492 |
+
},
|
| 26493 |
+
{
|
| 26494 |
+
"epoch": 0.33497843581877196,
|
| 26495 |
+
"grad_norm": 0.08067211508750916,
|
| 26496 |
+
"learning_rate": 0.0009867489133309312,
|
| 26497 |
+
"loss": 1.5613,
|
| 26498 |
+
"step": 7534
|
| 26499 |
+
},
|
| 26500 |
+
{
|
| 26501 |
+
"epoch": 0.3350673602774443,
|
| 26502 |
+
"grad_norm": 0.08332887291908264,
|
| 26503 |
+
"learning_rate": 0.0009867408360571786,
|
| 26504 |
+
"loss": 1.5533,
|
| 26505 |
+
"step": 7536
|
| 26506 |
+
},
|
| 26507 |
+
{
|
| 26508 |
+
"epoch": 0.3351562847361167,
|
| 26509 |
+
"grad_norm": 0.08278394490480423,
|
| 26510 |
+
"learning_rate": 0.0009867327563554819,
|
| 26511 |
+
"loss": 1.5563,
|
| 26512 |
+
"step": 7538
|
| 26513 |
+
},
|
| 26514 |
+
{
|
| 26515 |
+
"epoch": 0.335245209194789,
|
| 26516 |
+
"grad_norm": 0.08533370494842529,
|
| 26517 |
+
"learning_rate": 0.0009867246742258813,
|
| 26518 |
+
"loss": 1.558,
|
| 26519 |
+
"step": 7540
|
| 26520 |
+
},
|
| 26521 |
+
{
|
| 26522 |
+
"epoch": 0.3353341336534614,
|
| 26523 |
+
"grad_norm": 0.08526033163070679,
|
| 26524 |
+
"learning_rate": 0.0009867165896684173,
|
| 26525 |
+
"loss": 1.5586,
|
| 26526 |
+
"step": 7542
|
| 26527 |
+
},
|
| 26528 |
+
{
|
| 26529 |
+
"epoch": 0.33542305811213374,
|
| 26530 |
+
"grad_norm": 0.0807008370757103,
|
| 26531 |
+
"learning_rate": 0.0009867085026831304,
|
| 26532 |
+
"loss": 1.563,
|
| 26533 |
+
"step": 7544
|
| 26534 |
+
},
|
| 26535 |
+
{
|
| 26536 |
+
"epoch": 0.3355119825708061,
|
| 26537 |
+
"grad_norm": 0.08265620470046997,
|
| 26538 |
+
"learning_rate": 0.0009867004132700606,
|
| 26539 |
+
"loss": 1.5492,
|
| 26540 |
+
"step": 7546
|
| 26541 |
+
},
|
| 26542 |
+
{
|
| 26543 |
+
"epoch": 0.3356009070294785,
|
| 26544 |
+
"grad_norm": 0.08214646577835083,
|
| 26545 |
+
"learning_rate": 0.0009866923214292482,
|
| 26546 |
+
"loss": 1.5571,
|
| 26547 |
+
"step": 7548
|
| 26548 |
+
},
|
| 26549 |
+
{
|
| 26550 |
+
"epoch": 0.3356898314881508,
|
| 26551 |
+
"grad_norm": 0.0798763632774353,
|
| 26552 |
+
"learning_rate": 0.0009866842271607336,
|
| 26553 |
+
"loss": 1.5533,
|
| 26554 |
+
"step": 7550
|
| 26555 |
+
},
|
| 26556 |
+
{
|
| 26557 |
+
"epoch": 0.33577875594682316,
|
| 26558 |
+
"grad_norm": 0.07971591502428055,
|
| 26559 |
+
"learning_rate": 0.0009866761304645575,
|
| 26560 |
+
"loss": 1.5582,
|
| 26561 |
+
"step": 7552
|
| 26562 |
+
},
|
| 26563 |
+
{
|
| 26564 |
+
"epoch": 0.33586768040549553,
|
| 26565 |
+
"grad_norm": 0.08346015959978104,
|
| 26566 |
+
"learning_rate": 0.00098666803134076,
|
| 26567 |
+
"loss": 1.5546,
|
| 26568 |
+
"step": 7554
|
| 26569 |
+
},
|
| 26570 |
+
{
|
| 26571 |
+
"epoch": 0.3359566048641679,
|
| 26572 |
+
"grad_norm": 0.08312725275754929,
|
| 26573 |
+
"learning_rate": 0.0009866599297893817,
|
| 26574 |
+
"loss": 1.5637,
|
| 26575 |
+
"step": 7556
|
| 26576 |
+
},
|
| 26577 |
+
{
|
| 26578 |
+
"epoch": 0.33604552932284026,
|
| 26579 |
+
"grad_norm": 0.07956881821155548,
|
| 26580 |
+
"learning_rate": 0.0009866518258104628,
|
| 26581 |
+
"loss": 1.5593,
|
| 26582 |
+
"step": 7558
|
| 26583 |
+
},
|
| 26584 |
+
{
|
| 26585 |
+
"epoch": 0.33613445378151263,
|
| 26586 |
+
"grad_norm": 0.08073802292346954,
|
| 26587 |
+
"learning_rate": 0.0009866437194040439,
|
| 26588 |
+
"loss": 1.5635,
|
| 26589 |
+
"step": 7560
|
| 26590 |
+
},
|
| 26591 |
+
{
|
| 26592 |
+
"epoch": 0.33622337824018494,
|
| 26593 |
+
"grad_norm": 0.08115088939666748,
|
| 26594 |
+
"learning_rate": 0.0009866356105701652,
|
| 26595 |
+
"loss": 1.5571,
|
| 26596 |
+
"step": 7562
|
| 26597 |
+
},
|
| 26598 |
+
{
|
| 26599 |
+
"epoch": 0.3363123026988573,
|
| 26600 |
+
"grad_norm": 0.08693643659353256,
|
| 26601 |
+
"learning_rate": 0.0009866274993088674,
|
| 26602 |
+
"loss": 1.5552,
|
| 26603 |
+
"step": 7564
|
| 26604 |
+
},
|
| 26605 |
+
{
|
| 26606 |
+
"epoch": 0.3364012271575297,
|
| 26607 |
+
"grad_norm": 0.0820782408118248,
|
| 26608 |
+
"learning_rate": 0.0009866193856201907,
|
| 26609 |
+
"loss": 1.5554,
|
| 26610 |
+
"step": 7566
|
| 26611 |
+
},
|
| 26612 |
+
{
|
| 26613 |
+
"epoch": 0.33649015161620205,
|
| 26614 |
+
"grad_norm": 0.08132241666316986,
|
| 26615 |
+
"learning_rate": 0.000986611269504176,
|
| 26616 |
+
"loss": 1.5558,
|
| 26617 |
+
"step": 7568
|
| 26618 |
+
},
|
| 26619 |
+
{
|
| 26620 |
+
"epoch": 0.3365790760748744,
|
| 26621 |
+
"grad_norm": 0.08526284992694855,
|
| 26622 |
+
"learning_rate": 0.0009866031509608633,
|
| 26623 |
+
"loss": 1.5552,
|
| 26624 |
+
"step": 7570
|
| 26625 |
+
},
|
| 26626 |
+
{
|
| 26627 |
+
"epoch": 0.33666800053354673,
|
| 26628 |
+
"grad_norm": 0.08132081478834152,
|
| 26629 |
+
"learning_rate": 0.0009865950299902935,
|
| 26630 |
+
"loss": 1.5619,
|
| 26631 |
+
"step": 7572
|
| 26632 |
+
},
|
| 26633 |
+
{
|
| 26634 |
+
"epoch": 0.3367569249922191,
|
| 26635 |
+
"grad_norm": 0.08150200545787811,
|
| 26636 |
+
"learning_rate": 0.0009865869065925068,
|
| 26637 |
+
"loss": 1.5563,
|
| 26638 |
+
"step": 7574
|
| 26639 |
+
},
|
| 26640 |
+
{
|
| 26641 |
+
"epoch": 0.33684584945089147,
|
| 26642 |
+
"grad_norm": 0.0802740678191185,
|
| 26643 |
+
"learning_rate": 0.0009865787807675438,
|
| 26644 |
+
"loss": 1.5508,
|
| 26645 |
+
"step": 7576
|
| 26646 |
+
},
|
| 26647 |
+
{
|
| 26648 |
+
"epoch": 0.33693477390956383,
|
| 26649 |
+
"grad_norm": 0.08373408019542694,
|
| 26650 |
+
"learning_rate": 0.0009865706525154453,
|
| 26651 |
+
"loss": 1.5592,
|
| 26652 |
+
"step": 7578
|
| 26653 |
+
},
|
| 26654 |
+
{
|
| 26655 |
+
"epoch": 0.3370236983682362,
|
| 26656 |
+
"grad_norm": 0.08197604864835739,
|
| 26657 |
+
"learning_rate": 0.0009865625218362514,
|
| 26658 |
+
"loss": 1.5523,
|
| 26659 |
+
"step": 7580
|
| 26660 |
+
},
|
| 26661 |
+
{
|
| 26662 |
+
"epoch": 0.3371126228269085,
|
| 26663 |
+
"grad_norm": 0.08425270020961761,
|
| 26664 |
+
"learning_rate": 0.0009865543887300028,
|
| 26665 |
+
"loss": 1.557,
|
| 26666 |
+
"step": 7582
|
| 26667 |
+
},
|
| 26668 |
+
{
|
| 26669 |
+
"epoch": 0.3372015472855809,
|
| 26670 |
+
"grad_norm": 0.08232254534959793,
|
| 26671 |
+
"learning_rate": 0.00098654625319674,
|
| 26672 |
+
"loss": 1.5567,
|
| 26673 |
+
"step": 7584
|
| 26674 |
+
},
|
| 26675 |
+
{
|
| 26676 |
+
"epoch": 0.33729047174425325,
|
| 26677 |
+
"grad_norm": 0.08390845358371735,
|
| 26678 |
+
"learning_rate": 0.0009865381152365043,
|
| 26679 |
+
"loss": 1.558,
|
| 26680 |
+
"step": 7586
|
| 26681 |
+
},
|
| 26682 |
+
{
|
| 26683 |
+
"epoch": 0.3373793962029256,
|
| 26684 |
+
"grad_norm": 0.08453506231307983,
|
| 26685 |
+
"learning_rate": 0.0009865299748493352,
|
| 26686 |
+
"loss": 1.5561,
|
| 26687 |
+
"step": 7588
|
| 26688 |
+
},
|
| 26689 |
+
{
|
| 26690 |
+
"epoch": 0.337468320661598,
|
| 26691 |
+
"grad_norm": 0.08450014144182205,
|
| 26692 |
+
"learning_rate": 0.0009865218320352742,
|
| 26693 |
+
"loss": 1.5515,
|
| 26694 |
+
"step": 7590
|
| 26695 |
+
},
|
| 26696 |
+
{
|
| 26697 |
+
"epoch": 0.33755724512027035,
|
| 26698 |
+
"grad_norm": 0.083348348736763,
|
| 26699 |
+
"learning_rate": 0.0009865136867943615,
|
| 26700 |
+
"loss": 1.5583,
|
| 26701 |
+
"step": 7592
|
| 26702 |
+
},
|
| 26703 |
+
{
|
| 26704 |
+
"epoch": 0.33764616957894267,
|
| 26705 |
+
"grad_norm": 0.0832839161157608,
|
| 26706 |
+
"learning_rate": 0.0009865055391266374,
|
| 26707 |
+
"loss": 1.5618,
|
| 26708 |
+
"step": 7594
|
| 26709 |
+
},
|
| 26710 |
+
{
|
| 26711 |
+
"epoch": 0.33773509403761504,
|
| 26712 |
+
"grad_norm": 0.07885057479143143,
|
| 26713 |
+
"learning_rate": 0.0009864973890321431,
|
| 26714 |
+
"loss": 1.5528,
|
| 26715 |
+
"step": 7596
|
| 26716 |
+
},
|
| 26717 |
+
{
|
| 26718 |
+
"epoch": 0.3378240184962874,
|
| 26719 |
+
"grad_norm": 0.0822066068649292,
|
| 26720 |
+
"learning_rate": 0.0009864892365109193,
|
| 26721 |
+
"loss": 1.5585,
|
| 26722 |
+
"step": 7598
|
| 26723 |
+
},
|
| 26724 |
+
{
|
| 26725 |
+
"epoch": 0.33791294295495977,
|
| 26726 |
+
"grad_norm": 0.08152958005666733,
|
| 26727 |
+
"learning_rate": 0.0009864810815630061,
|
| 26728 |
+
"loss": 1.559,
|
| 26729 |
+
"step": 7600
|
| 26730 |
+
},
|
| 26731 |
+
{
|
| 26732 |
+
"epoch": 0.33800186741363214,
|
| 26733 |
+
"grad_norm": 0.08215297013521194,
|
| 26734 |
+
"learning_rate": 0.0009864729241884447,
|
| 26735 |
+
"loss": 1.5569,
|
| 26736 |
+
"step": 7602
|
| 26737 |
+
},
|
| 26738 |
+
{
|
| 26739 |
+
"epoch": 0.33809079187230445,
|
| 26740 |
+
"grad_norm": 0.08144047111272812,
|
| 26741 |
+
"learning_rate": 0.0009864647643872757,
|
| 26742 |
+
"loss": 1.558,
|
| 26743 |
+
"step": 7604
|
| 26744 |
+
},
|
| 26745 |
+
{
|
| 26746 |
+
"epoch": 0.3381797163309768,
|
| 26747 |
+
"grad_norm": 0.08268721401691437,
|
| 26748 |
+
"learning_rate": 0.0009864566021595396,
|
| 26749 |
+
"loss": 1.5643,
|
| 26750 |
+
"step": 7606
|
| 26751 |
+
},
|
| 26752 |
+
{
|
| 26753 |
+
"epoch": 0.3382686407896492,
|
| 26754 |
+
"grad_norm": 0.08536458760499954,
|
| 26755 |
+
"learning_rate": 0.000986448437505277,
|
| 26756 |
+
"loss": 1.5528,
|
| 26757 |
+
"step": 7608
|
| 26758 |
+
},
|
| 26759 |
+
{
|
| 26760 |
+
"epoch": 0.33835756524832156,
|
| 26761 |
+
"grad_norm": 0.07940022647380829,
|
| 26762 |
+
"learning_rate": 0.0009864402704245292,
|
| 26763 |
+
"loss": 1.5514,
|
| 26764 |
+
"step": 7610
|
| 26765 |
+
},
|
| 26766 |
+
{
|
| 26767 |
+
"epoch": 0.3384464897069939,
|
| 26768 |
+
"grad_norm": 0.0803755670785904,
|
| 26769 |
+
"learning_rate": 0.0009864321009173365,
|
| 26770 |
+
"loss": 1.5582,
|
| 26771 |
+
"step": 7612
|
| 26772 |
+
},
|
| 26773 |
+
{
|
| 26774 |
+
"epoch": 0.3385354141656663,
|
| 26775 |
+
"grad_norm": 0.08060748130083084,
|
| 26776 |
+
"learning_rate": 0.0009864239289837395,
|
| 26777 |
+
"loss": 1.559,
|
| 26778 |
+
"step": 7614
|
| 26779 |
+
},
|
| 26780 |
+
{
|
| 26781 |
+
"epoch": 0.3386243386243386,
|
| 26782 |
+
"grad_norm": 0.07905225455760956,
|
| 26783 |
+
"learning_rate": 0.0009864157546237795,
|
| 26784 |
+
"loss": 1.5599,
|
| 26785 |
+
"step": 7616
|
| 26786 |
+
},
|
| 26787 |
+
{
|
| 26788 |
+
"epoch": 0.338713263083011,
|
| 26789 |
+
"grad_norm": 0.08046825975179672,
|
| 26790 |
+
"learning_rate": 0.0009864075778374967,
|
| 26791 |
+
"loss": 1.5617,
|
| 26792 |
+
"step": 7618
|
| 26793 |
+
},
|
| 26794 |
+
{
|
| 26795 |
+
"epoch": 0.33880218754168334,
|
| 26796 |
+
"grad_norm": 0.08096635341644287,
|
| 26797 |
+
"learning_rate": 0.0009863993986249325,
|
| 26798 |
+
"loss": 1.5552,
|
| 26799 |
+
"step": 7620
|
| 26800 |
+
},
|
| 26801 |
+
{
|
| 26802 |
+
"epoch": 0.3388911120003557,
|
| 26803 |
+
"grad_norm": 0.08172665536403656,
|
| 26804 |
+
"learning_rate": 0.000986391216986127,
|
| 26805 |
+
"loss": 1.5593,
|
| 26806 |
+
"step": 7622
|
| 26807 |
+
},
|
| 26808 |
+
{
|
| 26809 |
+
"epoch": 0.3389800364590281,
|
| 26810 |
+
"grad_norm": 0.0854748860001564,
|
| 26811 |
+
"learning_rate": 0.0009863830329211217,
|
| 26812 |
+
"loss": 1.5591,
|
| 26813 |
+
"step": 7624
|
| 26814 |
+
},
|
| 26815 |
+
{
|
| 26816 |
+
"epoch": 0.3390689609177004,
|
| 26817 |
+
"grad_norm": 0.0805579125881195,
|
| 26818 |
+
"learning_rate": 0.0009863748464299569,
|
| 26819 |
+
"loss": 1.5578,
|
| 26820 |
+
"step": 7626
|
| 26821 |
+
},
|
| 26822 |
+
{
|
| 26823 |
+
"epoch": 0.33915788537637276,
|
| 26824 |
+
"grad_norm": 0.07932641357183456,
|
| 26825 |
+
"learning_rate": 0.0009863666575126737,
|
| 26826 |
+
"loss": 1.5491,
|
| 26827 |
+
"step": 7628
|
| 26828 |
+
},
|
| 26829 |
+
{
|
| 26830 |
+
"epoch": 0.3392468098350451,
|
| 26831 |
+
"grad_norm": 0.0846443846821785,
|
| 26832 |
+
"learning_rate": 0.000986358466169313,
|
| 26833 |
+
"loss": 1.5591,
|
| 26834 |
+
"step": 7630
|
| 26835 |
+
},
|
| 26836 |
+
{
|
| 26837 |
+
"epoch": 0.3393357342937175,
|
| 26838 |
+
"grad_norm": 0.08422157168388367,
|
| 26839 |
+
"learning_rate": 0.0009863502723999153,
|
| 26840 |
+
"loss": 1.5564,
|
| 26841 |
+
"step": 7632
|
| 26842 |
+
},
|
| 26843 |
+
{
|
| 26844 |
+
"epoch": 0.33942465875238986,
|
| 26845 |
+
"grad_norm": 0.08966944366693497,
|
| 26846 |
+
"learning_rate": 0.0009863420762045217,
|
| 26847 |
+
"loss": 1.5575,
|
| 26848 |
+
"step": 7634
|
| 26849 |
+
},
|
| 26850 |
+
{
|
| 26851 |
+
"epoch": 0.3395135832110622,
|
| 26852 |
+
"grad_norm": 0.08580625057220459,
|
| 26853 |
+
"learning_rate": 0.000986333877583173,
|
| 26854 |
+
"loss": 1.5568,
|
| 26855 |
+
"step": 7636
|
| 26856 |
+
},
|
| 26857 |
+
{
|
| 26858 |
+
"epoch": 0.33960250766973454,
|
| 26859 |
+
"grad_norm": 0.08479603379964828,
|
| 26860 |
+
"learning_rate": 0.0009863256765359108,
|
| 26861 |
+
"loss": 1.5607,
|
| 26862 |
+
"step": 7638
|
| 26863 |
+
},
|
| 26864 |
+
{
|
| 26865 |
+
"epoch": 0.3396914321284069,
|
| 26866 |
+
"grad_norm": 0.0841260626912117,
|
| 26867 |
+
"learning_rate": 0.000986317473062775,
|
| 26868 |
+
"loss": 1.5547,
|
| 26869 |
+
"step": 7640
|
| 26870 |
+
},
|
| 26871 |
+
{
|
| 26872 |
+
"epoch": 0.3397803565870793,
|
| 26873 |
+
"grad_norm": 0.08255398273468018,
|
| 26874 |
+
"learning_rate": 0.0009863092671638068,
|
| 26875 |
+
"loss": 1.5613,
|
| 26876 |
+
"step": 7642
|
| 26877 |
+
},
|
| 26878 |
+
{
|
| 26879 |
+
"epoch": 0.33986928104575165,
|
| 26880 |
+
"grad_norm": 0.0812351405620575,
|
| 26881 |
+
"learning_rate": 0.0009863010588390473,
|
| 26882 |
+
"loss": 1.5575,
|
| 26883 |
+
"step": 7644
|
| 26884 |
+
},
|
| 26885 |
+
{
|
| 26886 |
+
"epoch": 0.339958205504424,
|
| 26887 |
+
"grad_norm": 0.0809001475572586,
|
| 26888 |
+
"learning_rate": 0.0009862928480885375,
|
| 26889 |
+
"loss": 1.5592,
|
| 26890 |
+
"step": 7646
|
| 26891 |
+
},
|
| 26892 |
+
{
|
| 26893 |
+
"epoch": 0.34004712996309633,
|
| 26894 |
+
"grad_norm": 0.0820452943444252,
|
| 26895 |
+
"learning_rate": 0.0009862846349123183,
|
| 26896 |
+
"loss": 1.5551,
|
| 26897 |
+
"step": 7648
|
| 26898 |
+
},
|
| 26899 |
+
{
|
| 26900 |
+
"epoch": 0.3401360544217687,
|
| 26901 |
+
"grad_norm": 0.08290841430425644,
|
| 26902 |
+
"learning_rate": 0.0009862764193104303,
|
| 26903 |
+
"loss": 1.5607,
|
| 26904 |
+
"step": 7650
|
| 26905 |
+
},
|
| 26906 |
+
{
|
| 26907 |
+
"epoch": 0.34022497888044106,
|
| 26908 |
+
"grad_norm": 0.08015654981136322,
|
| 26909 |
+
"learning_rate": 0.0009862682012829152,
|
| 26910 |
+
"loss": 1.5555,
|
| 26911 |
+
"step": 7652
|
| 26912 |
+
},
|
| 26913 |
+
{
|
| 26914 |
+
"epoch": 0.34031390333911343,
|
| 26915 |
+
"grad_norm": 0.08518238365650177,
|
| 26916 |
+
"learning_rate": 0.0009862599808298134,
|
| 26917 |
+
"loss": 1.5524,
|
| 26918 |
+
"step": 7654
|
| 26919 |
+
},
|
| 26920 |
+
{
|
| 26921 |
+
"epoch": 0.3404028277977858,
|
| 26922 |
+
"grad_norm": 0.08613689988851547,
|
| 26923 |
+
"learning_rate": 0.000986251757951166,
|
| 26924 |
+
"loss": 1.556,
|
| 26925 |
+
"step": 7656
|
| 26926 |
+
},
|
| 26927 |
+
{
|
| 26928 |
+
"epoch": 0.3404917522564581,
|
| 26929 |
+
"grad_norm": 0.08661675453186035,
|
| 26930 |
+
"learning_rate": 0.0009862435326470143,
|
| 26931 |
+
"loss": 1.5564,
|
| 26932 |
+
"step": 7658
|
| 26933 |
+
},
|
| 26934 |
+
{
|
| 26935 |
+
"epoch": 0.3405806767151305,
|
| 26936 |
+
"grad_norm": 0.07869388163089752,
|
| 26937 |
+
"learning_rate": 0.000986235304917399,
|
| 26938 |
+
"loss": 1.5577,
|
| 26939 |
+
"step": 7660
|
| 26940 |
+
},
|
| 26941 |
+
{
|
| 26942 |
+
"epoch": 0.34066960117380285,
|
| 26943 |
+
"grad_norm": 0.08099169284105301,
|
| 26944 |
+
"learning_rate": 0.0009862270747623616,
|
| 26945 |
+
"loss": 1.5534,
|
| 26946 |
+
"step": 7662
|
| 26947 |
+
},
|
| 26948 |
+
{
|
| 26949 |
+
"epoch": 0.3407585256324752,
|
| 26950 |
+
"grad_norm": 0.08418874442577362,
|
| 26951 |
+
"learning_rate": 0.0009862188421819425,
|
| 26952 |
+
"loss": 1.5575,
|
| 26953 |
+
"step": 7664
|
| 26954 |
+
},
|
| 26955 |
+
{
|
| 26956 |
+
"epoch": 0.3408474500911476,
|
| 26957 |
+
"grad_norm": 0.08251459896564484,
|
| 26958 |
+
"learning_rate": 0.0009862106071761834,
|
| 26959 |
+
"loss": 1.5578,
|
| 26960 |
+
"step": 7666
|
| 26961 |
+
},
|
| 26962 |
+
{
|
| 26963 |
+
"epoch": 0.34093637454981995,
|
| 26964 |
+
"grad_norm": 0.08254079520702362,
|
| 26965 |
+
"learning_rate": 0.0009862023697451248,
|
| 26966 |
+
"loss": 1.5507,
|
| 26967 |
+
"step": 7668
|
| 26968 |
+
},
|
| 26969 |
+
{
|
| 26970 |
+
"epoch": 0.34102529900849227,
|
| 26971 |
+
"grad_norm": 0.08215059340000153,
|
| 26972 |
+
"learning_rate": 0.0009861941298888082,
|
| 26973 |
+
"loss": 1.5562,
|
| 26974 |
+
"step": 7670
|
| 26975 |
+
},
|
| 26976 |
+
{
|
| 26977 |
+
"epoch": 0.34111422346716463,
|
| 26978 |
+
"grad_norm": 0.08373280614614487,
|
| 26979 |
+
"learning_rate": 0.0009861858876072747,
|
| 26980 |
+
"loss": 1.5594,
|
| 26981 |
+
"step": 7672
|
| 26982 |
+
},
|
| 26983 |
+
{
|
| 26984 |
+
"epoch": 0.341203147925837,
|
| 26985 |
+
"grad_norm": 0.07917515188455582,
|
| 26986 |
+
"learning_rate": 0.000986177642900565,
|
| 26987 |
+
"loss": 1.5542,
|
| 26988 |
+
"step": 7674
|
| 26989 |
+
},
|
| 26990 |
+
{
|
| 26991 |
+
"epoch": 0.34129207238450937,
|
| 26992 |
+
"grad_norm": 0.08267196267843246,
|
| 26993 |
+
"learning_rate": 0.0009861693957687208,
|
| 26994 |
+
"loss": 1.5612,
|
| 26995 |
+
"step": 7676
|
| 26996 |
+
},
|
| 26997 |
+
{
|
| 26998 |
+
"epoch": 0.34138099684318174,
|
| 26999 |
+
"grad_norm": 0.08618869632482529,
|
| 27000 |
+
"learning_rate": 0.0009861611462117829,
|
| 27001 |
+
"loss": 1.5604,
|
| 27002 |
+
"step": 7678
|
| 27003 |
+
},
|
| 27004 |
+
{
|
| 27005 |
+
"epoch": 0.34146992130185405,
|
| 27006 |
+
"grad_norm": 0.08248917013406754,
|
| 27007 |
+
"learning_rate": 0.0009861528942297923,
|
| 27008 |
+
"loss": 1.5624,
|
| 27009 |
+
"step": 7680
|
| 27010 |
+
},
|
| 27011 |
+
{
|
| 27012 |
+
"epoch": 0.3415588457605264,
|
| 27013 |
+
"grad_norm": 0.08377496898174286,
|
| 27014 |
+
"learning_rate": 0.0009861446398227904,
|
| 27015 |
+
"loss": 1.5614,
|
| 27016 |
+
"step": 7682
|
| 27017 |
+
},
|
| 27018 |
+
{
|
| 27019 |
+
"epoch": 0.3416477702191988,
|
| 27020 |
+
"grad_norm": 0.08078130334615707,
|
| 27021 |
+
"learning_rate": 0.0009861363829908186,
|
| 27022 |
+
"loss": 1.5528,
|
| 27023 |
+
"step": 7684
|
| 27024 |
+
},
|
| 27025 |
+
{
|
| 27026 |
+
"epoch": 0.34173669467787116,
|
| 27027 |
+
"grad_norm": 0.07990232855081558,
|
| 27028 |
+
"learning_rate": 0.0009861281237339176,
|
| 27029 |
+
"loss": 1.5554,
|
| 27030 |
+
"step": 7686
|
| 27031 |
+
},
|
| 27032 |
+
{
|
| 27033 |
+
"epoch": 0.3418256191365435,
|
| 27034 |
+
"grad_norm": 0.07614683359861374,
|
| 27035 |
+
"learning_rate": 0.0009861198620521288,
|
| 27036 |
+
"loss": 1.5562,
|
| 27037 |
+
"step": 7688
|
| 27038 |
+
},
|
| 27039 |
+
{
|
| 27040 |
+
"epoch": 0.3419145435952159,
|
| 27041 |
+
"grad_norm": 0.08139005303382874,
|
| 27042 |
+
"learning_rate": 0.0009861115979454935,
|
| 27043 |
+
"loss": 1.5544,
|
| 27044 |
+
"step": 7690
|
| 27045 |
+
},
|
| 27046 |
+
{
|
| 27047 |
+
"epoch": 0.3420034680538882,
|
| 27048 |
+
"grad_norm": 0.08052778244018555,
|
| 27049 |
+
"learning_rate": 0.000986103331414053,
|
| 27050 |
+
"loss": 1.5592,
|
| 27051 |
+
"step": 7692
|
| 27052 |
+
},
|
| 27053 |
+
{
|
| 27054 |
+
"epoch": 0.3420923925125606,
|
| 27055 |
+
"grad_norm": 0.08632178604602814,
|
| 27056 |
+
"learning_rate": 0.0009860950624578481,
|
| 27057 |
+
"loss": 1.5574,
|
| 27058 |
+
"step": 7694
|
| 27059 |
+
},
|
| 27060 |
+
{
|
| 27061 |
+
"epoch": 0.34218131697123294,
|
| 27062 |
+
"grad_norm": 0.0841991975903511,
|
| 27063 |
+
"learning_rate": 0.0009860867910769207,
|
| 27064 |
+
"loss": 1.5576,
|
| 27065 |
+
"step": 7696
|
| 27066 |
+
},
|
| 27067 |
+
{
|
| 27068 |
+
"epoch": 0.3422702414299053,
|
| 27069 |
+
"grad_norm": 0.08402567356824875,
|
| 27070 |
+
"learning_rate": 0.0009860785172713115,
|
| 27071 |
+
"loss": 1.5525,
|
| 27072 |
+
"step": 7698
|
| 27073 |
+
},
|
| 27074 |
+
{
|
| 27075 |
+
"epoch": 0.3423591658885777,
|
| 27076 |
+
"grad_norm": 0.08191527426242828,
|
| 27077 |
+
"learning_rate": 0.000986070241041062,
|
| 27078 |
+
"loss": 1.5613,
|
| 27079 |
+
"step": 7700
|
| 27080 |
+
},
|
| 27081 |
+
{
|
| 27082 |
+
"epoch": 0.34244809034725,
|
| 27083 |
+
"grad_norm": 0.08203962445259094,
|
| 27084 |
+
"learning_rate": 0.0009860619623862136,
|
| 27085 |
+
"loss": 1.5565,
|
| 27086 |
+
"step": 7702
|
| 27087 |
+
},
|
| 27088 |
+
{
|
| 27089 |
+
"epoch": 0.34253701480592236,
|
| 27090 |
+
"grad_norm": 0.07740230858325958,
|
| 27091 |
+
"learning_rate": 0.0009860536813068072,
|
| 27092 |
+
"loss": 1.5574,
|
| 27093 |
+
"step": 7704
|
| 27094 |
+
},
|
| 27095 |
+
{
|
| 27096 |
+
"epoch": 0.3426259392645947,
|
| 27097 |
+
"grad_norm": 0.07994545996189117,
|
| 27098 |
+
"learning_rate": 0.0009860453978028844,
|
| 27099 |
+
"loss": 1.5573,
|
| 27100 |
+
"step": 7706
|
| 27101 |
+
},
|
| 27102 |
+
{
|
| 27103 |
+
"epoch": 0.3427148637232671,
|
| 27104 |
+
"grad_norm": 0.08103641122579575,
|
| 27105 |
+
"learning_rate": 0.0009860371118744866,
|
| 27106 |
+
"loss": 1.5566,
|
| 27107 |
+
"step": 7708
|
| 27108 |
+
},
|
| 27109 |
+
{
|
| 27110 |
+
"epoch": 0.34280378818193946,
|
| 27111 |
+
"grad_norm": 0.08051252365112305,
|
| 27112 |
+
"learning_rate": 0.0009860288235216549,
|
| 27113 |
+
"loss": 1.5603,
|
| 27114 |
+
"step": 7710
|
| 27115 |
+
},
|
| 27116 |
+
{
|
| 27117 |
+
"epoch": 0.3428927126406118,
|
| 27118 |
+
"grad_norm": 0.08033636957406998,
|
| 27119 |
+
"learning_rate": 0.0009860205327444308,
|
| 27120 |
+
"loss": 1.5545,
|
| 27121 |
+
"step": 7712
|
| 27122 |
+
},
|
| 27123 |
+
{
|
| 27124 |
+
"epoch": 0.34298163709928414,
|
| 27125 |
+
"grad_norm": 0.08459188044071198,
|
| 27126 |
+
"learning_rate": 0.0009860122395428555,
|
| 27127 |
+
"loss": 1.5595,
|
| 27128 |
+
"step": 7714
|
| 27129 |
+
},
|
| 27130 |
+
{
|
| 27131 |
+
"epoch": 0.3430705615579565,
|
| 27132 |
+
"grad_norm": 0.08079506456851959,
|
| 27133 |
+
"learning_rate": 0.0009860039439169705,
|
| 27134 |
+
"loss": 1.5565,
|
| 27135 |
+
"step": 7716
|
| 27136 |
+
},
|
| 27137 |
+
{
|
| 27138 |
+
"epoch": 0.3431594860166289,
|
| 27139 |
+
"grad_norm": 0.08086486160755157,
|
| 27140 |
+
"learning_rate": 0.0009859956458668174,
|
| 27141 |
+
"loss": 1.5585,
|
| 27142 |
+
"step": 7718
|
| 27143 |
+
},
|
| 27144 |
+
{
|
| 27145 |
+
"epoch": 0.34324841047530125,
|
| 27146 |
+
"grad_norm": 0.0799533799290657,
|
| 27147 |
+
"learning_rate": 0.000985987345392437,
|
| 27148 |
+
"loss": 1.5492,
|
| 27149 |
+
"step": 7720
|
| 27150 |
+
},
|
| 27151 |
+
{
|
| 27152 |
+
"epoch": 0.3433373349339736,
|
| 27153 |
+
"grad_norm": 0.07945334911346436,
|
| 27154 |
+
"learning_rate": 0.0009859790424938714,
|
| 27155 |
+
"loss": 1.5486,
|
| 27156 |
+
"step": 7722
|
| 27157 |
+
},
|
| 27158 |
+
{
|
| 27159 |
+
"epoch": 0.3434262593926459,
|
| 27160 |
+
"grad_norm": 0.07963025569915771,
|
| 27161 |
+
"learning_rate": 0.0009859707371711614,
|
| 27162 |
+
"loss": 1.5606,
|
| 27163 |
+
"step": 7724
|
| 27164 |
+
},
|
| 27165 |
+
{
|
| 27166 |
+
"epoch": 0.3435151838513183,
|
| 27167 |
+
"grad_norm": 0.08368024230003357,
|
| 27168 |
+
"learning_rate": 0.0009859624294243487,
|
| 27169 |
+
"loss": 1.5599,
|
| 27170 |
+
"step": 7726
|
| 27171 |
+
},
|
| 27172 |
+
{
|
| 27173 |
+
"epoch": 0.34360410830999066,
|
| 27174 |
+
"grad_norm": 0.07972510904073715,
|
| 27175 |
+
"learning_rate": 0.000985954119253475,
|
| 27176 |
+
"loss": 1.5461,
|
| 27177 |
+
"step": 7728
|
| 27178 |
+
},
|
| 27179 |
+
{
|
| 27180 |
+
"epoch": 0.34369303276866303,
|
| 27181 |
+
"grad_norm": 0.08038385957479477,
|
| 27182 |
+
"learning_rate": 0.000985945806658581,
|
| 27183 |
+
"loss": 1.5568,
|
| 27184 |
+
"step": 7730
|
| 27185 |
+
},
|
| 27186 |
+
{
|
| 27187 |
+
"epoch": 0.3437819572273354,
|
| 27188 |
+
"grad_norm": 0.08230438828468323,
|
| 27189 |
+
"learning_rate": 0.0009859374916397093,
|
| 27190 |
+
"loss": 1.5553,
|
| 27191 |
+
"step": 7732
|
| 27192 |
+
},
|
| 27193 |
+
{
|
| 27194 |
+
"epoch": 0.3438708816860077,
|
| 27195 |
+
"grad_norm": 0.08172792196273804,
|
| 27196 |
+
"learning_rate": 0.0009859291741969004,
|
| 27197 |
+
"loss": 1.5544,
|
| 27198 |
+
"step": 7734
|
| 27199 |
+
},
|
| 27200 |
+
{
|
| 27201 |
+
"epoch": 0.3439598061446801,
|
| 27202 |
+
"grad_norm": 0.08643286675214767,
|
| 27203 |
+
"learning_rate": 0.0009859208543301962,
|
| 27204 |
+
"loss": 1.5551,
|
| 27205 |
+
"step": 7736
|
| 27206 |
+
},
|
| 27207 |
+
{
|
| 27208 |
+
"epoch": 0.34404873060335245,
|
| 27209 |
+
"grad_norm": 0.08269612491130829,
|
| 27210 |
+
"learning_rate": 0.0009859125320396379,
|
| 27211 |
+
"loss": 1.5538,
|
| 27212 |
+
"step": 7738
|
| 27213 |
+
},
|
| 27214 |
+
{
|
| 27215 |
+
"epoch": 0.3441376550620248,
|
| 27216 |
+
"grad_norm": 0.08148770779371262,
|
| 27217 |
+
"learning_rate": 0.0009859042073252673,
|
| 27218 |
+
"loss": 1.5511,
|
| 27219 |
+
"step": 7740
|
| 27220 |
+
},
|
| 27221 |
+
{
|
| 27222 |
+
"epoch": 0.3442265795206972,
|
| 27223 |
+
"grad_norm": 0.0830964595079422,
|
| 27224 |
+
"learning_rate": 0.0009858958801871257,
|
| 27225 |
+
"loss": 1.559,
|
| 27226 |
+
"step": 7742
|
| 27227 |
+
},
|
| 27228 |
+
{
|
| 27229 |
+
"epoch": 0.34431550397936955,
|
| 27230 |
+
"grad_norm": 0.08528967946767807,
|
| 27231 |
+
"learning_rate": 0.0009858875506252552,
|
| 27232 |
+
"loss": 1.5578,
|
| 27233 |
+
"step": 7744
|
| 27234 |
+
},
|
| 27235 |
+
{
|
| 27236 |
+
"epoch": 0.34440442843804187,
|
| 27237 |
+
"grad_norm": 0.08497735857963562,
|
| 27238 |
+
"learning_rate": 0.0009858792186396967,
|
| 27239 |
+
"loss": 1.5565,
|
| 27240 |
+
"step": 7746
|
| 27241 |
+
},
|
| 27242 |
+
{
|
| 27243 |
+
"epoch": 0.34449335289671423,
|
| 27244 |
+
"grad_norm": 0.07968219369649887,
|
| 27245 |
+
"learning_rate": 0.0009858708842304922,
|
| 27246 |
+
"loss": 1.5606,
|
| 27247 |
+
"step": 7748
|
| 27248 |
+
},
|
| 27249 |
+
{
|
| 27250 |
+
"epoch": 0.3445822773553866,
|
| 27251 |
+
"grad_norm": 0.08317109197378159,
|
| 27252 |
+
"learning_rate": 0.0009858625473976828,
|
| 27253 |
+
"loss": 1.5534,
|
| 27254 |
+
"step": 7750
|
| 27255 |
+
},
|
| 27256 |
+
{
|
| 27257 |
+
"epoch": 0.34467120181405897,
|
| 27258 |
+
"grad_norm": 0.07973136752843857,
|
| 27259 |
+
"learning_rate": 0.0009858542081413106,
|
| 27260 |
+
"loss": 1.5536,
|
| 27261 |
+
"step": 7752
|
| 27262 |
+
},
|
| 27263 |
+
{
|
| 27264 |
+
"epoch": 0.34476012627273134,
|
| 27265 |
+
"grad_norm": 0.07634979486465454,
|
| 27266 |
+
"learning_rate": 0.0009858458664614169,
|
| 27267 |
+
"loss": 1.5492,
|
| 27268 |
+
"step": 7754
|
| 27269 |
+
},
|
| 27270 |
+
{
|
| 27271 |
+
"epoch": 0.34484905073140365,
|
| 27272 |
+
"grad_norm": 0.07837677747011185,
|
| 27273 |
+
"learning_rate": 0.0009858375223580434,
|
| 27274 |
+
"loss": 1.5548,
|
| 27275 |
+
"step": 7756
|
| 27276 |
+
},
|
| 27277 |
+
{
|
| 27278 |
+
"epoch": 0.344937975190076,
|
| 27279 |
+
"grad_norm": 0.07845926284790039,
|
| 27280 |
+
"learning_rate": 0.0009858291758312317,
|
| 27281 |
+
"loss": 1.5537,
|
| 27282 |
+
"step": 7758
|
| 27283 |
+
},
|
| 27284 |
+
{
|
| 27285 |
+
"epoch": 0.3450268996487484,
|
| 27286 |
+
"grad_norm": 0.08067924529314041,
|
| 27287 |
+
"learning_rate": 0.0009858208268810232,
|
| 27288 |
+
"loss": 1.5546,
|
| 27289 |
+
"step": 7760
|
| 27290 |
+
},
|
| 27291 |
+
{
|
| 27292 |
+
"epoch": 0.34511582410742075,
|
| 27293 |
+
"grad_norm": 0.07944106310606003,
|
| 27294 |
+
"learning_rate": 0.00098581247550746,
|
| 27295 |
+
"loss": 1.5523,
|
| 27296 |
+
"step": 7762
|
| 27297 |
+
},
|
| 27298 |
+
{
|
| 27299 |
+
"epoch": 0.3452047485660931,
|
| 27300 |
+
"grad_norm": 0.08254885673522949,
|
| 27301 |
+
"learning_rate": 0.0009858041217105835,
|
| 27302 |
+
"loss": 1.5549,
|
| 27303 |
+
"step": 7764
|
| 27304 |
+
},
|
| 27305 |
+
{
|
| 27306 |
+
"epoch": 0.34529367302476544,
|
| 27307 |
+
"grad_norm": 0.08257626742124557,
|
| 27308 |
+
"learning_rate": 0.0009857957654904352,
|
| 27309 |
+
"loss": 1.5568,
|
| 27310 |
+
"step": 7766
|
| 27311 |
+
},
|
| 27312 |
+
{
|
| 27313 |
+
"epoch": 0.3453825974834378,
|
| 27314 |
+
"grad_norm": 0.07757585495710373,
|
| 27315 |
+
"learning_rate": 0.0009857874068470575,
|
| 27316 |
+
"loss": 1.5542,
|
| 27317 |
+
"step": 7768
|
| 27318 |
+
},
|
| 27319 |
+
{
|
| 27320 |
+
"epoch": 0.34547152194211017,
|
| 27321 |
+
"grad_norm": 0.07840435206890106,
|
| 27322 |
+
"learning_rate": 0.000985779045780491,
|
| 27323 |
+
"loss": 1.5541,
|
| 27324 |
+
"step": 7770
|
| 27325 |
+
},
|
| 27326 |
+
{
|
| 27327 |
+
"epoch": 0.34556044640078254,
|
| 27328 |
+
"grad_norm": 0.08048916608095169,
|
| 27329 |
+
"learning_rate": 0.0009857706822907783,
|
| 27330 |
+
"loss": 1.5509,
|
| 27331 |
+
"step": 7772
|
| 27332 |
+
},
|
| 27333 |
+
{
|
| 27334 |
+
"epoch": 0.3456493708594549,
|
| 27335 |
+
"grad_norm": 0.08014550805091858,
|
| 27336 |
+
"learning_rate": 0.0009857623163779606,
|
| 27337 |
+
"loss": 1.5515,
|
| 27338 |
+
"step": 7774
|
| 27339 |
+
},
|
| 27340 |
+
{
|
| 27341 |
+
"epoch": 0.3457382953181273,
|
| 27342 |
+
"grad_norm": 0.08095970749855042,
|
| 27343 |
+
"learning_rate": 0.00098575394804208,
|
| 27344 |
+
"loss": 1.5545,
|
| 27345 |
+
"step": 7776
|
| 27346 |
+
},
|
| 27347 |
+
{
|
| 27348 |
+
"epoch": 0.3458272197767996,
|
| 27349 |
+
"grad_norm": 0.0833306685090065,
|
| 27350 |
+
"learning_rate": 0.000985745577283178,
|
| 27351 |
+
"loss": 1.5512,
|
| 27352 |
+
"step": 7778
|
| 27353 |
+
},
|
| 27354 |
+
{
|
| 27355 |
+
"epoch": 0.34591614423547196,
|
| 27356 |
+
"grad_norm": 0.08204789459705353,
|
| 27357 |
+
"learning_rate": 0.0009857372041012966,
|
| 27358 |
+
"loss": 1.5512,
|
| 27359 |
+
"step": 7780
|
| 27360 |
+
},
|
| 27361 |
+
{
|
| 27362 |
+
"epoch": 0.3460050686941443,
|
| 27363 |
+
"grad_norm": 0.08149214088916779,
|
| 27364 |
+
"learning_rate": 0.000985728828496477,
|
| 27365 |
+
"loss": 1.5549,
|
| 27366 |
+
"step": 7782
|
| 27367 |
+
},
|
| 27368 |
+
{
|
| 27369 |
+
"epoch": 0.3460939931528167,
|
| 27370 |
+
"grad_norm": 0.08480332046747208,
|
| 27371 |
+
"learning_rate": 0.0009857204504687617,
|
| 27372 |
+
"loss": 1.5529,
|
| 27373 |
+
"step": 7784
|
| 27374 |
+
},
|
| 27375 |
+
{
|
| 27376 |
+
"epoch": 0.34618291761148906,
|
| 27377 |
+
"grad_norm": 0.0808916836977005,
|
| 27378 |
+
"learning_rate": 0.0009857120700181921,
|
| 27379 |
+
"loss": 1.5539,
|
| 27380 |
+
"step": 7786
|
| 27381 |
+
},
|
| 27382 |
+
{
|
| 27383 |
+
"epoch": 0.3462718420701614,
|
| 27384 |
+
"grad_norm": 0.08434953540563583,
|
| 27385 |
+
"learning_rate": 0.0009857036871448101,
|
| 27386 |
+
"loss": 1.5492,
|
| 27387 |
+
"step": 7788
|
| 27388 |
+
},
|
| 27389 |
+
{
|
| 27390 |
+
"epoch": 0.34636076652883374,
|
| 27391 |
+
"grad_norm": 0.077853262424469,
|
| 27392 |
+
"learning_rate": 0.0009856953018486576,
|
| 27393 |
+
"loss": 1.5522,
|
| 27394 |
+
"step": 7790
|
| 27395 |
+
},
|
| 27396 |
+
{
|
| 27397 |
+
"epoch": 0.3464496909875061,
|
| 27398 |
+
"grad_norm": 0.08755781501531601,
|
| 27399 |
+
"learning_rate": 0.0009856869141297762,
|
| 27400 |
+
"loss": 1.5561,
|
| 27401 |
+
"step": 7792
|
| 27402 |
+
},
|
| 27403 |
+
{
|
| 27404 |
+
"epoch": 0.3465386154461785,
|
| 27405 |
+
"grad_norm": 0.0840546116232872,
|
| 27406 |
+
"learning_rate": 0.0009856785239882079,
|
| 27407 |
+
"loss": 1.5522,
|
| 27408 |
+
"step": 7794
|
| 27409 |
+
},
|
| 27410 |
+
{
|
| 27411 |
+
"epoch": 0.34662753990485085,
|
| 27412 |
+
"grad_norm": 0.08388349413871765,
|
| 27413 |
+
"learning_rate": 0.0009856701314239942,
|
| 27414 |
+
"loss": 1.5519,
|
| 27415 |
+
"step": 7796
|
| 27416 |
+
},
|
| 27417 |
+
{
|
| 27418 |
+
"epoch": 0.3467164643635232,
|
| 27419 |
+
"grad_norm": 0.07955602556467056,
|
| 27420 |
+
"learning_rate": 0.0009856617364371776,
|
| 27421 |
+
"loss": 1.5584,
|
| 27422 |
+
"step": 7798
|
| 27423 |
+
},
|
| 27424 |
+
{
|
| 27425 |
+
"epoch": 0.3468053888221955,
|
| 27426 |
+
"grad_norm": 0.08307410031557083,
|
| 27427 |
+
"learning_rate": 0.0009856533390277995,
|
| 27428 |
+
"loss": 1.5527,
|
| 27429 |
+
"step": 7800
|
| 27430 |
+
},
|
| 27431 |
+
{
|
| 27432 |
+
"epoch": 0.3468943132808679,
|
| 27433 |
+
"grad_norm": 0.08166033774614334,
|
| 27434 |
+
"learning_rate": 0.000985644939195902,
|
| 27435 |
+
"loss": 1.5575,
|
| 27436 |
+
"step": 7802
|
| 27437 |
+
},
|
| 27438 |
+
{
|
| 27439 |
+
"epoch": 0.34698323773954026,
|
| 27440 |
+
"grad_norm": 0.08017229288816452,
|
| 27441 |
+
"learning_rate": 0.0009856365369415269,
|
| 27442 |
+
"loss": 1.5536,
|
| 27443 |
+
"step": 7804
|
| 27444 |
+
},
|
| 27445 |
+
{
|
| 27446 |
+
"epoch": 0.34707216219821263,
|
| 27447 |
+
"grad_norm": 0.08459261059761047,
|
| 27448 |
+
"learning_rate": 0.000985628132264716,
|
| 27449 |
+
"loss": 1.5624,
|
| 27450 |
+
"step": 7806
|
| 27451 |
+
},
|
| 27452 |
+
{
|
| 27453 |
+
"epoch": 0.347161086656885,
|
| 27454 |
+
"grad_norm": 0.07887223362922668,
|
| 27455 |
+
"learning_rate": 0.0009856197251655117,
|
| 27456 |
+
"loss": 1.5504,
|
| 27457 |
+
"step": 7808
|
| 27458 |
+
},
|
| 27459 |
+
{
|
| 27460 |
+
"epoch": 0.3472500111155573,
|
| 27461 |
+
"grad_norm": 0.08019403368234634,
|
| 27462 |
+
"learning_rate": 0.0009856113156439554,
|
| 27463 |
+
"loss": 1.5524,
|
| 27464 |
+
"step": 7810
|
| 27465 |
+
},
|
| 27466 |
+
{
|
| 27467 |
+
"epoch": 0.3473389355742297,
|
| 27468 |
+
"grad_norm": 0.0818062275648117,
|
| 27469 |
+
"learning_rate": 0.0009856029037000893,
|
| 27470 |
+
"loss": 1.5556,
|
| 27471 |
+
"step": 7812
|
| 27472 |
+
},
|
| 27473 |
+
{
|
| 27474 |
+
"epoch": 0.34742786003290205,
|
| 27475 |
+
"grad_norm": 0.07944361120462418,
|
| 27476 |
+
"learning_rate": 0.0009855944893339553,
|
| 27477 |
+
"loss": 1.5543,
|
| 27478 |
+
"step": 7814
|
| 27479 |
+
},
|
| 27480 |
+
{
|
| 27481 |
+
"epoch": 0.3475167844915744,
|
| 27482 |
+
"grad_norm": 0.07994072139263153,
|
| 27483 |
+
"learning_rate": 0.0009855860725455953,
|
| 27484 |
+
"loss": 1.5454,
|
| 27485 |
+
"step": 7816
|
| 27486 |
+
},
|
| 27487 |
+
{
|
| 27488 |
+
"epoch": 0.3476057089502468,
|
| 27489 |
+
"grad_norm": 0.07912899553775787,
|
| 27490 |
+
"learning_rate": 0.0009855776533350513,
|
| 27491 |
+
"loss": 1.5526,
|
| 27492 |
+
"step": 7818
|
| 27493 |
+
},
|
| 27494 |
+
{
|
| 27495 |
+
"epoch": 0.3476946334089191,
|
| 27496 |
+
"grad_norm": 0.08150925487279892,
|
| 27497 |
+
"learning_rate": 0.0009855692317023654,
|
| 27498 |
+
"loss": 1.5507,
|
| 27499 |
+
"step": 7820
|
| 27500 |
+
},
|
| 27501 |
+
{
|
| 27502 |
+
"epoch": 0.34778355786759146,
|
| 27503 |
+
"grad_norm": 0.0804237350821495,
|
| 27504 |
+
"learning_rate": 0.0009855608076475798,
|
| 27505 |
+
"loss": 1.5562,
|
| 27506 |
+
"step": 7822
|
| 27507 |
+
},
|
| 27508 |
+
{
|
| 27509 |
+
"epoch": 0.34787248232626383,
|
| 27510 |
+
"grad_norm": 0.07889696210622787,
|
| 27511 |
+
"learning_rate": 0.000985552381170736,
|
| 27512 |
+
"loss": 1.5572,
|
| 27513 |
+
"step": 7824
|
| 27514 |
+
},
|
| 27515 |
+
{
|
| 27516 |
+
"epoch": 0.3479614067849362,
|
| 27517 |
+
"grad_norm": 0.07692207396030426,
|
| 27518 |
+
"learning_rate": 0.0009855439522718766,
|
| 27519 |
+
"loss": 1.5534,
|
| 27520 |
+
"step": 7826
|
| 27521 |
+
},
|
| 27522 |
+
{
|
| 27523 |
+
"epoch": 0.34805033124360857,
|
| 27524 |
+
"grad_norm": 0.07672964036464691,
|
| 27525 |
+
"learning_rate": 0.0009855355209510432,
|
| 27526 |
+
"loss": 1.553,
|
| 27527 |
+
"step": 7828
|
| 27528 |
+
},
|
| 27529 |
+
{
|
| 27530 |
+
"epoch": 0.34813925570228094,
|
| 27531 |
+
"grad_norm": 0.07981939613819122,
|
| 27532 |
+
"learning_rate": 0.000985527087208278,
|
| 27533 |
+
"loss": 1.5549,
|
| 27534 |
+
"step": 7830
|
| 27535 |
+
},
|
| 27536 |
+
{
|
| 27537 |
+
"epoch": 0.34822818016095325,
|
| 27538 |
+
"grad_norm": 0.07698217779397964,
|
| 27539 |
+
"learning_rate": 0.0009855186510436232,
|
| 27540 |
+
"loss": 1.5567,
|
| 27541 |
+
"step": 7832
|
| 27542 |
+
},
|
| 27543 |
+
{
|
| 27544 |
+
"epoch": 0.3483171046196256,
|
| 27545 |
+
"grad_norm": 0.07928860932588577,
|
| 27546 |
+
"learning_rate": 0.000985510212457121,
|
| 27547 |
+
"loss": 1.5513,
|
| 27548 |
+
"step": 7834
|
| 27549 |
+
},
|
| 27550 |
+
{
|
| 27551 |
+
"epoch": 0.348406029078298,
|
| 27552 |
+
"grad_norm": 0.08210493624210358,
|
| 27553 |
+
"learning_rate": 0.0009855017714488128,
|
| 27554 |
+
"loss": 1.5503,
|
| 27555 |
+
"step": 7836
|
| 27556 |
+
},
|
| 27557 |
+
{
|
| 27558 |
+
"epoch": 0.34849495353697035,
|
| 27559 |
+
"grad_norm": 0.08063147217035294,
|
| 27560 |
+
"learning_rate": 0.0009854933280187416,
|
| 27561 |
+
"loss": 1.5549,
|
| 27562 |
+
"step": 7838
|
| 27563 |
+
},
|
| 27564 |
+
{
|
| 27565 |
+
"epoch": 0.3485838779956427,
|
| 27566 |
+
"grad_norm": 0.07979092001914978,
|
| 27567 |
+
"learning_rate": 0.0009854848821669487,
|
| 27568 |
+
"loss": 1.5482,
|
| 27569 |
+
"step": 7840
|
| 27570 |
+
},
|
| 27571 |
+
{
|
| 27572 |
+
"epoch": 0.34867280245431503,
|
| 27573 |
+
"grad_norm": 0.0807296484708786,
|
| 27574 |
+
"learning_rate": 0.0009854764338934768,
|
| 27575 |
+
"loss": 1.5546,
|
| 27576 |
+
"step": 7842
|
| 27577 |
+
},
|
| 27578 |
+
{
|
| 27579 |
+
"epoch": 0.3487617269129874,
|
| 27580 |
+
"grad_norm": 0.082137331366539,
|
| 27581 |
+
"learning_rate": 0.0009854679831983678,
|
| 27582 |
+
"loss": 1.5572,
|
| 27583 |
+
"step": 7844
|
| 27584 |
+
},
|
| 27585 |
+
{
|
| 27586 |
+
"epoch": 0.34885065137165977,
|
| 27587 |
+
"grad_norm": 0.0817376971244812,
|
| 27588 |
+
"learning_rate": 0.000985459530081664,
|
| 27589 |
+
"loss": 1.5512,
|
| 27590 |
+
"step": 7846
|
| 27591 |
+
},
|
| 27592 |
+
{
|
| 27593 |
+
"epoch": 0.34893957583033214,
|
| 27594 |
+
"grad_norm": 0.08156748861074448,
|
| 27595 |
+
"learning_rate": 0.0009854510745434075,
|
| 27596 |
+
"loss": 1.5563,
|
| 27597 |
+
"step": 7848
|
| 27598 |
+
},
|
| 27599 |
+
{
|
| 27600 |
+
"epoch": 0.3490285002890045,
|
| 27601 |
+
"grad_norm": 0.08468926697969437,
|
| 27602 |
+
"learning_rate": 0.0009854426165836403,
|
| 27603 |
+
"loss": 1.5537,
|
| 27604 |
+
"step": 7850
|
| 27605 |
+
},
|
| 27606 |
+
{
|
| 27607 |
+
"epoch": 0.3491174247476769,
|
| 27608 |
+
"grad_norm": 0.0817328691482544,
|
| 27609 |
+
"learning_rate": 0.000985434156202405,
|
| 27610 |
+
"loss": 1.559,
|
| 27611 |
+
"step": 7852
|
| 27612 |
+
},
|
| 27613 |
+
{
|
| 27614 |
+
"epoch": 0.3492063492063492,
|
| 27615 |
+
"grad_norm": 0.07793274521827698,
|
| 27616 |
+
"learning_rate": 0.0009854256933997435,
|
| 27617 |
+
"loss": 1.5527,
|
| 27618 |
+
"step": 7854
|
| 27619 |
+
},
|
| 27620 |
+
{
|
| 27621 |
+
"epoch": 0.34929527366502156,
|
| 27622 |
+
"grad_norm": 0.07918530702590942,
|
| 27623 |
+
"learning_rate": 0.0009854172281756979,
|
| 27624 |
+
"loss": 1.554,
|
| 27625 |
+
"step": 7856
|
| 27626 |
+
},
|
| 27627 |
+
{
|
| 27628 |
+
"epoch": 0.3493841981236939,
|
| 27629 |
+
"grad_norm": 0.08301304280757904,
|
| 27630 |
+
"learning_rate": 0.0009854087605303105,
|
| 27631 |
+
"loss": 1.554,
|
| 27632 |
+
"step": 7858
|
| 27633 |
+
},
|
| 27634 |
+
{
|
| 27635 |
+
"epoch": 0.3494731225823663,
|
| 27636 |
+
"grad_norm": 0.07774464040994644,
|
| 27637 |
+
"learning_rate": 0.0009854002904636238,
|
| 27638 |
+
"loss": 1.5517,
|
| 27639 |
+
"step": 7860
|
| 27640 |
+
},
|
| 27641 |
+
{
|
| 27642 |
+
"epoch": 0.34956204704103866,
|
| 27643 |
+
"grad_norm": 0.08034802228212357,
|
| 27644 |
+
"learning_rate": 0.00098539181797568,
|
| 27645 |
+
"loss": 1.5544,
|
| 27646 |
+
"step": 7862
|
| 27647 |
+
},
|
| 27648 |
+
{
|
| 27649 |
+
"epoch": 0.34965097149971097,
|
| 27650 |
+
"grad_norm": 0.07757385820150375,
|
| 27651 |
+
"learning_rate": 0.0009853833430665212,
|
| 27652 |
+
"loss": 1.5548,
|
| 27653 |
+
"step": 7864
|
| 27654 |
+
},
|
| 27655 |
+
{
|
| 27656 |
+
"epoch": 0.34973989595838334,
|
| 27657 |
+
"grad_norm": 0.08288813382387161,
|
| 27658 |
+
"learning_rate": 0.0009853748657361896,
|
| 27659 |
+
"loss": 1.5527,
|
| 27660 |
+
"step": 7866
|
| 27661 |
+
},
|
| 27662 |
+
{
|
| 27663 |
+
"epoch": 0.3498288204170557,
|
| 27664 |
+
"grad_norm": 0.08583445101976395,
|
| 27665 |
+
"learning_rate": 0.0009853663859847276,
|
| 27666 |
+
"loss": 1.5544,
|
| 27667 |
+
"step": 7868
|
| 27668 |
+
},
|
| 27669 |
+
{
|
| 27670 |
+
"epoch": 0.3499177448757281,
|
| 27671 |
+
"grad_norm": 0.07990463078022003,
|
| 27672 |
+
"learning_rate": 0.0009853579038121775,
|
| 27673 |
+
"loss": 1.5537,
|
| 27674 |
+
"step": 7870
|
| 27675 |
+
},
|
| 27676 |
+
{
|
| 27677 |
+
"epoch": 0.35000666933440044,
|
| 27678 |
+
"grad_norm": 0.08092455565929413,
|
| 27679 |
+
"learning_rate": 0.0009853494192185817,
|
| 27680 |
+
"loss": 1.5494,
|
| 27681 |
+
"step": 7872
|
| 27682 |
+
},
|
| 27683 |
+
{
|
| 27684 |
+
"epoch": 0.3500955937930728,
|
| 27685 |
+
"grad_norm": 0.07774668186903,
|
| 27686 |
+
"learning_rate": 0.0009853409322039823,
|
| 27687 |
+
"loss": 1.5512,
|
| 27688 |
+
"step": 7874
|
| 27689 |
+
},
|
| 27690 |
+
{
|
| 27691 |
+
"epoch": 0.3501845182517451,
|
| 27692 |
+
"grad_norm": 0.08283083885908127,
|
| 27693 |
+
"learning_rate": 0.000985332442768422,
|
| 27694 |
+
"loss": 1.5538,
|
| 27695 |
+
"step": 7876
|
| 27696 |
+
},
|
| 27697 |
+
{
|
| 27698 |
+
"epoch": 0.3502734427104175,
|
| 27699 |
+
"grad_norm": 0.08365800231695175,
|
| 27700 |
+
"learning_rate": 0.0009853239509119427,
|
| 27701 |
+
"loss": 1.5525,
|
| 27702 |
+
"step": 7878
|
| 27703 |
+
},
|
| 27704 |
+
{
|
| 27705 |
+
"epoch": 0.35036236716908986,
|
| 27706 |
+
"grad_norm": 0.077533058822155,
|
| 27707 |
+
"learning_rate": 0.000985315456634587,
|
| 27708 |
+
"loss": 1.5596,
|
| 27709 |
+
"step": 7880
|
| 27710 |
+
},
|
| 27711 |
+
{
|
| 27712 |
+
"epoch": 0.35045129162776223,
|
| 27713 |
+
"grad_norm": 0.08021321892738342,
|
| 27714 |
+
"learning_rate": 0.0009853069599363973,
|
| 27715 |
+
"loss": 1.5468,
|
| 27716 |
+
"step": 7882
|
| 27717 |
+
},
|
| 27718 |
+
{
|
| 27719 |
+
"epoch": 0.3505402160864346,
|
| 27720 |
+
"grad_norm": 0.11996711045503616,
|
| 27721 |
+
"learning_rate": 0.0009852984608174158,
|
| 27722 |
+
"loss": 1.5533,
|
| 27723 |
+
"step": 7884
|
| 27724 |
+
},
|
| 27725 |
+
{
|
| 27726 |
+
"epoch": 0.3506291405451069,
|
| 27727 |
+
"grad_norm": 0.3232218623161316,
|
| 27728 |
+
"learning_rate": 0.0009852899592776852,
|
| 27729 |
+
"loss": 1.6783,
|
| 27730 |
+
"step": 7886
|
| 27731 |
+
},
|
| 27732 |
+
{
|
| 27733 |
+
"epoch": 0.3507180650037793,
|
| 27734 |
+
"grad_norm": 4.166322231292725,
|
| 27735 |
+
"learning_rate": 0.0009852814553172476,
|
| 27736 |
+
"loss": 2.4033,
|
| 27737 |
+
"step": 7888
|
| 27738 |
+
},
|
| 27739 |
+
{
|
| 27740 |
+
"epoch": 0.35080698946245165,
|
| 27741 |
+
"grad_norm": 105.5245132446289,
|
| 27742 |
+
"learning_rate": 0.0009852729489361457,
|
| 27743 |
+
"loss": 3.5532,
|
| 27744 |
+
"step": 7890
|
| 27745 |
+
},
|
| 27746 |
+
{
|
| 27747 |
+
"epoch": 0.350895913921124,
|
| 27748 |
+
"grad_norm": 5.243870258331299,
|
| 27749 |
+
"learning_rate": 0.0009852644401344218,
|
| 27750 |
+
"loss": 7.1315,
|
| 27751 |
+
"step": 7892
|
| 27752 |
+
},
|
| 27753 |
+
{
|
| 27754 |
+
"epoch": 0.3509848383797964,
|
| 27755 |
+
"grad_norm": 6.443427085876465,
|
| 27756 |
+
"learning_rate": 0.0009852559289121182,
|
| 27757 |
+
"loss": 7.392,
|
| 27758 |
+
"step": 7894
|
| 27759 |
+
},
|
| 27760 |
+
{
|
| 27761 |
+
"epoch": 0.3510737628384687,
|
| 27762 |
+
"grad_norm": 11.5606050491333,
|
| 27763 |
+
"learning_rate": 0.0009852474152692774,
|
| 27764 |
+
"loss": 7.5255,
|
| 27765 |
+
"step": 7896
|
| 27766 |
+
},
|
| 27767 |
+
{
|
| 27768 |
+
"epoch": 0.35116268729714106,
|
| 27769 |
+
"grad_norm": 2.048408031463623,
|
| 27770 |
+
"learning_rate": 0.000985238899205942,
|
| 27771 |
+
"loss": 7.2354,
|
| 27772 |
+
"step": 7898
|
| 27773 |
+
},
|
| 27774 |
+
{
|
| 27775 |
+
"epoch": 0.35125161175581343,
|
| 27776 |
+
"grad_norm": 0.9983798265457153,
|
| 27777 |
+
"learning_rate": 0.0009852303807221545,
|
| 27778 |
+
"loss": 6.5717,
|
| 27779 |
+
"step": 7900
|
| 27780 |
+
},
|
| 27781 |
+
{
|
| 27782 |
+
"epoch": 0.3513405362144858,
|
| 27783 |
+
"grad_norm": 6.880504608154297,
|
| 27784 |
+
"learning_rate": 0.0009852218598179572,
|
| 27785 |
+
"loss": 6.7333,
|
| 27786 |
+
"step": 7902
|
| 27787 |
+
},
|
| 27788 |
+
{
|
| 27789 |
+
"epoch": 0.35142946067315817,
|
| 27790 |
+
"grad_norm": 0.685417890548706,
|
| 27791 |
+
"learning_rate": 0.0009852133364933928,
|
| 27792 |
+
"loss": 6.3766,
|
| 27793 |
+
"step": 7904
|
| 27794 |
+
},
|
| 27795 |
+
{
|
| 27796 |
+
"epoch": 0.35151838513183054,
|
| 27797 |
+
"grad_norm": 1.0315533876419067,
|
| 27798 |
+
"learning_rate": 0.0009852048107485039,
|
| 27799 |
+
"loss": 6.1477,
|
| 27800 |
+
"step": 7906
|
| 27801 |
+
},
|
| 27802 |
+
{
|
| 27803 |
+
"epoch": 0.35160730959050285,
|
| 27804 |
+
"grad_norm": 0.4206119179725647,
|
| 27805 |
+
"learning_rate": 0.0009851962825833324,
|
| 27806 |
+
"loss": 6.0279,
|
| 27807 |
+
"step": 7908
|
| 27808 |
+
},
|
| 27809 |
+
{
|
| 27810 |
+
"epoch": 0.3516962340491752,
|
| 27811 |
+
"grad_norm": 0.34929385781288147,
|
| 27812 |
+
"learning_rate": 0.0009851877519979218,
|
| 27813 |
+
"loss": 5.9475,
|
| 27814 |
+
"step": 7910
|
| 27815 |
+
},
|
| 27816 |
+
{
|
| 27817 |
+
"epoch": 0.3517851585078476,
|
| 27818 |
+
"grad_norm": 0.33057817816734314,
|
| 27819 |
+
"learning_rate": 0.000985179218992314,
|
| 27820 |
+
"loss": 5.8579,
|
| 27821 |
+
"step": 7912
|
| 27822 |
+
},
|
| 27823 |
+
{
|
| 27824 |
+
"epoch": 0.35187408296651995,
|
| 27825 |
+
"grad_norm": 0.2894933819770813,
|
| 27826 |
+
"learning_rate": 0.0009851706835665515,
|
| 27827 |
+
"loss": 5.8023,
|
| 27828 |
+
"step": 7914
|
| 27829 |
+
},
|
| 27830 |
+
{
|
| 27831 |
+
"epoch": 0.3519630074251923,
|
| 27832 |
+
"grad_norm": 0.21461133658885956,
|
| 27833 |
+
"learning_rate": 0.0009851621457206774,
|
| 27834 |
+
"loss": 5.7603,
|
| 27835 |
+
"step": 7916
|
| 27836 |
+
},
|
| 27837 |
+
{
|
| 27838 |
+
"epoch": 0.35205193188386463,
|
| 27839 |
+
"grad_norm": 0.21909675002098083,
|
| 27840 |
+
"learning_rate": 0.0009851536054547338,
|
| 27841 |
+
"loss": 5.7207,
|
| 27842 |
+
"step": 7918
|
| 27843 |
+
},
|
| 27844 |
+
{
|
| 27845 |
+
"epoch": 0.352140856342537,
|
| 27846 |
+
"grad_norm": 0.3142562508583069,
|
| 27847 |
+
"learning_rate": 0.0009851450627687635,
|
| 27848 |
+
"loss": 5.6821,
|
| 27849 |
+
"step": 7920
|
| 27850 |
+
},
|
| 27851 |
+
{
|
| 27852 |
+
"epoch": 0.35222978080120937,
|
| 27853 |
+
"grad_norm": 1.3621059656143188,
|
| 27854 |
+
"learning_rate": 0.000985136517662809,
|
| 27855 |
+
"loss": 5.6821,
|
| 27856 |
+
"step": 7922
|
| 27857 |
+
},
|
| 27858 |
+
{
|
| 27859 |
+
"epoch": 0.35231870525988174,
|
| 27860 |
+
"grad_norm": 0.7668919563293457,
|
| 27861 |
+
"learning_rate": 0.0009851279701369134,
|
| 27862 |
+
"loss": 5.6241,
|
| 27863 |
+
"step": 7924
|
| 27864 |
+
},
|
| 27865 |
+
{
|
| 27866 |
+
"epoch": 0.3524076297185541,
|
| 27867 |
+
"grad_norm": 0.3618534207344055,
|
| 27868 |
+
"learning_rate": 0.0009851194201911187,
|
| 27869 |
+
"loss": 5.5482,
|
| 27870 |
+
"step": 7926
|
| 27871 |
+
},
|
| 27872 |
+
{
|
| 27873 |
+
"epoch": 0.3524965541772265,
|
| 27874 |
+
"grad_norm": 0.2870240807533264,
|
| 27875 |
+
"learning_rate": 0.0009851108678254677,
|
| 27876 |
+
"loss": 5.5032,
|
| 27877 |
+
"step": 7928
|
| 27878 |
+
},
|
| 27879 |
+
{
|
| 27880 |
+
"epoch": 0.3525854786358988,
|
| 27881 |
+
"grad_norm": 0.485879510641098,
|
| 27882 |
+
"learning_rate": 0.0009851023130400035,
|
| 27883 |
+
"loss": 5.4567,
|
| 27884 |
+
"step": 7930
|
| 27885 |
+
},
|
| 27886 |
+
{
|
| 27887 |
+
"epoch": 0.35267440309457115,
|
| 27888 |
+
"grad_norm": 0.2687022387981415,
|
| 27889 |
+
"learning_rate": 0.000985093755834768,
|
| 27890 |
+
"loss": 5.3988,
|
| 27891 |
+
"step": 7932
|
| 27892 |
+
},
|
| 27893 |
+
{
|
| 27894 |
+
"epoch": 0.3527633275532435,
|
| 27895 |
+
"grad_norm": 0.3306155204772949,
|
| 27896 |
+
"learning_rate": 0.0009850851962098046,
|
| 27897 |
+
"loss": 5.3285,
|
| 27898 |
+
"step": 7934
|
| 27899 |
+
},
|
| 27900 |
+
{
|
| 27901 |
+
"epoch": 0.3528522520119159,
|
| 27902 |
+
"grad_norm": 0.7935633063316345,
|
| 27903 |
+
"learning_rate": 0.0009850766341651556,
|
| 27904 |
+
"loss": 5.2442,
|
| 27905 |
+
"step": 7936
|
| 27906 |
+
},
|
| 27907 |
+
{
|
| 27908 |
+
"epoch": 0.35294117647058826,
|
| 27909 |
+
"grad_norm": 1.1152472496032715,
|
| 27910 |
+
"learning_rate": 0.000985068069700864,
|
| 27911 |
+
"loss": 5.2263,
|
| 27912 |
+
"step": 7938
|
| 27913 |
+
},
|
| 27914 |
+
{
|
| 27915 |
+
"epoch": 0.35303010092926057,
|
| 27916 |
+
"grad_norm": 1.460869550704956,
|
| 27917 |
+
"learning_rate": 0.0009850595028169722,
|
| 27918 |
+
"loss": 5.2365,
|
| 27919 |
+
"step": 7940
|
| 27920 |
+
},
|
| 27921 |
+
{
|
| 27922 |
+
"epoch": 0.35311902538793294,
|
| 27923 |
+
"grad_norm": 1.1066856384277344,
|
| 27924 |
+
"learning_rate": 0.0009850509335135231,
|
| 27925 |
+
"loss": 5.1598,
|
| 27926 |
+
"step": 7942
|
| 27927 |
+
},
|
| 27928 |
+
{
|
| 27929 |
+
"epoch": 0.3532079498466053,
|
| 27930 |
+
"grad_norm": 0.8395753502845764,
|
| 27931 |
+
"learning_rate": 0.0009850423617905595,
|
| 27932 |
+
"loss": 5.0372,
|
| 27933 |
+
"step": 7944
|
| 27934 |
+
},
|
| 27935 |
+
{
|
| 27936 |
+
"epoch": 0.3532968743052777,
|
| 27937 |
+
"grad_norm": 0.550400972366333,
|
| 27938 |
+
"learning_rate": 0.0009850337876481242,
|
| 27939 |
+
"loss": 4.9278,
|
| 27940 |
+
"step": 7946
|
| 27941 |
+
},
|
| 27942 |
+
{
|
| 27943 |
+
"epoch": 0.35338579876395004,
|
| 27944 |
+
"grad_norm": 0.7241742014884949,
|
| 27945 |
+
"learning_rate": 0.0009850252110862596,
|
| 27946 |
+
"loss": 4.8471,
|
| 27947 |
+
"step": 7948
|
| 27948 |
+
},
|
| 27949 |
+
{
|
| 27950 |
+
"epoch": 0.35347472322262236,
|
| 27951 |
+
"grad_norm": 1.1229575872421265,
|
| 27952 |
+
"learning_rate": 0.000985016632105009,
|
| 27953 |
+
"loss": 4.7441,
|
| 27954 |
+
"step": 7950
|
| 27955 |
+
},
|
| 27956 |
+
{
|
| 27957 |
+
"epoch": 0.3535636476812947,
|
| 27958 |
+
"grad_norm": 0.5880857110023499,
|
| 27959 |
+
"learning_rate": 0.0009850080507044147,
|
| 27960 |
+
"loss": 4.6106,
|
| 27961 |
+
"step": 7952
|
| 27962 |
+
},
|
| 27963 |
+
{
|
| 27964 |
+
"epoch": 0.3536525721399671,
|
| 27965 |
+
"grad_norm": 0.6695611476898193,
|
| 27966 |
+
"learning_rate": 0.0009849994668845196,
|
| 27967 |
+
"loss": 4.4359,
|
| 27968 |
+
"step": 7954
|
| 27969 |
+
},
|
| 27970 |
+
{
|
| 27971 |
+
"epoch": 0.35374149659863946,
|
| 27972 |
+
"grad_norm": 0.4951477646827698,
|
| 27973 |
+
"learning_rate": 0.000984990880645367,
|
| 27974 |
+
"loss": 4.3121,
|
| 27975 |
+
"step": 7956
|
| 27976 |
+
},
|
| 27977 |
+
{
|
| 27978 |
+
"epoch": 0.35383042105731183,
|
| 27979 |
+
"grad_norm": 1.49424147605896,
|
| 27980 |
+
"learning_rate": 0.0009849822919869994,
|
| 27981 |
+
"loss": 4.3655,
|
| 27982 |
+
"step": 7958
|
| 27983 |
+
},
|
| 27984 |
+
{
|
| 27985 |
+
"epoch": 0.3539193455159842,
|
| 27986 |
+
"grad_norm": 0.6731409430503845,
|
| 27987 |
+
"learning_rate": 0.0009849737009094595,
|
| 27988 |
+
"loss": 4.1954,
|
| 27989 |
+
"step": 7960
|
| 27990 |
+
},
|
| 27991 |
+
{
|
| 27992 |
+
"epoch": 0.3540082699746565,
|
| 27993 |
+
"grad_norm": 0.657111644744873,
|
| 27994 |
+
"learning_rate": 0.00098496510741279,
|
| 27995 |
+
"loss": 3.9788,
|
| 27996 |
+
"step": 7962
|
| 27997 |
+
},
|
| 27998 |
+
{
|
| 27999 |
+
"epoch": 0.3540971944333289,
|
| 28000 |
+
"grad_norm": 1.2278788089752197,
|
| 28001 |
+
"learning_rate": 0.0009849565114970346,
|
| 28002 |
+
"loss": 3.8825,
|
| 28003 |
+
"step": 7964
|
| 28004 |
+
},
|
| 28005 |
+
{
|
| 28006 |
+
"epoch": 0.35418611889200124,
|
| 28007 |
+
"grad_norm": 1.1738439798355103,
|
| 28008 |
+
"learning_rate": 0.0009849479131622352,
|
| 28009 |
+
"loss": 3.7363,
|
| 28010 |
+
"step": 7966
|
| 28011 |
+
},
|
| 28012 |
+
{
|
| 28013 |
+
"epoch": 0.3542750433506736,
|
| 28014 |
+
"grad_norm": 0.9582828283309937,
|
| 28015 |
+
"learning_rate": 0.000984939312408435,
|
| 28016 |
+
"loss": 3.5467,
|
| 28017 |
+
"step": 7968
|
| 28018 |
+
},
|
| 28019 |
+
{
|
| 28020 |
+
"epoch": 0.354363967809346,
|
| 28021 |
+
"grad_norm": 0.9599164724349976,
|
| 28022 |
+
"learning_rate": 0.0009849307092356773,
|
| 28023 |
+
"loss": 3.3789,
|
| 28024 |
+
"step": 7970
|
| 28025 |
+
},
|
| 28026 |
+
{
|
| 28027 |
+
"epoch": 0.3544528922680183,
|
| 28028 |
+
"grad_norm": 0.8798990249633789,
|
| 28029 |
+
"learning_rate": 0.0009849221036440049,
|
| 28030 |
+
"loss": 3.201,
|
| 28031 |
+
"step": 7972
|
| 28032 |
+
},
|
| 28033 |
+
{
|
| 28034 |
+
"epoch": 0.35454181672669066,
|
| 28035 |
+
"grad_norm": 0.837184488773346,
|
| 28036 |
+
"learning_rate": 0.0009849134956334603,
|
| 28037 |
+
"loss": 3.0279,
|
| 28038 |
+
"step": 7974
|
| 28039 |
+
},
|
| 28040 |
+
{
|
| 28041 |
+
"epoch": 0.35463074118536303,
|
| 28042 |
+
"grad_norm": 0.869229793548584,
|
| 28043 |
+
"learning_rate": 0.0009849048852040867,
|
| 28044 |
+
"loss": 2.9193,
|
| 28045 |
+
"step": 7976
|
| 28046 |
+
},
|
| 28047 |
+
{
|
| 28048 |
+
"epoch": 0.3547196656440354,
|
| 28049 |
+
"grad_norm": 1.0861220359802246,
|
| 28050 |
+
"learning_rate": 0.000984896272355927,
|
| 28051 |
+
"loss": 2.7289,
|
| 28052 |
+
"step": 7978
|
| 28053 |
+
},
|
| 28054 |
+
{
|
| 28055 |
+
"epoch": 0.35480859010270777,
|
| 28056 |
+
"grad_norm": 0.8610190749168396,
|
| 28057 |
+
"learning_rate": 0.000984887657089024,
|
| 28058 |
+
"loss": 2.5584,
|
| 28059 |
+
"step": 7980
|
| 28060 |
+
},
|
| 28061 |
+
{
|
| 28062 |
+
"epoch": 0.35489751456138013,
|
| 28063 |
+
"grad_norm": 0.6836897134780884,
|
| 28064 |
+
"learning_rate": 0.0009848790394034214,
|
| 28065 |
+
"loss": 2.4055,
|
| 28066 |
+
"step": 7982
|
| 28067 |
+
},
|
| 28068 |
+
{
|
| 28069 |
+
"epoch": 0.35498643902005245,
|
| 28070 |
+
"grad_norm": 0.7401456832885742,
|
| 28071 |
+
"learning_rate": 0.0009848704192991613,
|
| 28072 |
+
"loss": 2.2528,
|
| 28073 |
+
"step": 7984
|
| 28074 |
+
},
|
| 28075 |
+
{
|
| 28076 |
+
"epoch": 0.3550753634787248,
|
| 28077 |
+
"grad_norm": 0.9028481245040894,
|
| 28078 |
+
"learning_rate": 0.0009848617967762872,
|
| 28079 |
+
"loss": 2.3112,
|
| 28080 |
+
"step": 7986
|
| 28081 |
+
},
|
| 28082 |
+
{
|
| 28083 |
+
"epoch": 0.3551642879373972,
|
| 28084 |
+
"grad_norm": 0.3118695318698883,
|
| 28085 |
+
"learning_rate": 0.000984853171834842,
|
| 28086 |
+
"loss": 2.1296,
|
| 28087 |
+
"step": 7988
|
| 28088 |
+
},
|
| 28089 |
+
{
|
| 28090 |
+
"epoch": 0.35525321239606955,
|
| 28091 |
+
"grad_norm": 0.30334171652793884,
|
| 28092 |
+
"learning_rate": 0.0009848445444748685,
|
| 28093 |
+
"loss": 2.0048,
|
| 28094 |
+
"step": 7990
|
| 28095 |
+
},
|
| 28096 |
+
{
|
| 28097 |
+
"epoch": 0.3553421368547419,
|
| 28098 |
+
"grad_norm": 0.19824688136577606,
|
| 28099 |
+
"learning_rate": 0.0009848359146964103,
|
| 28100 |
+
"loss": 1.9316,
|
| 28101 |
+
"step": 7992
|
| 28102 |
+
},
|
| 28103 |
+
{
|
| 28104 |
+
"epoch": 0.35543106131341423,
|
| 28105 |
+
"grad_norm": 0.2128942757844925,
|
| 28106 |
+
"learning_rate": 0.0009848272824995096,
|
| 28107 |
+
"loss": 1.8792,
|
| 28108 |
+
"step": 7994
|
| 28109 |
+
},
|
| 28110 |
+
{
|
| 28111 |
+
"epoch": 0.3555199857720866,
|
| 28112 |
+
"grad_norm": 0.15974895656108856,
|
| 28113 |
+
"learning_rate": 0.0009848186478842104,
|
| 28114 |
+
"loss": 1.8306,
|
| 28115 |
+
"step": 7996
|
| 28116 |
+
},
|
| 28117 |
+
{
|
| 28118 |
+
"epoch": 0.35560891023075897,
|
| 28119 |
+
"grad_norm": 0.13069583475589752,
|
| 28120 |
+
"learning_rate": 0.000984810010850555,
|
| 28121 |
+
"loss": 1.7945,
|
| 28122 |
+
"step": 7998
|
| 28123 |
+
},
|
| 28124 |
+
{
|
| 28125 |
+
"epoch": 0.35569783468943134,
|
| 28126 |
+
"grad_norm": 0.13678595423698425,
|
| 28127 |
+
"learning_rate": 0.000984801371398587,
|
| 28128 |
+
"loss": 1.7696,
|
| 28129 |
+
"step": 8000
|
| 28130 |
+
},
|
| 28131 |
+
{
|
| 28132 |
+
"epoch": 0.35569783468943134,
|
| 28133 |
+
"eval_loss": 1.6807708740234375,
|
| 28134 |
+
"eval_runtime": 12.3592,
|
| 28135 |
+
"eval_samples_per_second": 559.097,
|
| 28136 |
+
"eval_steps_per_second": 69.907,
|
| 28137 |
+
"step": 8000
|
| 28138 |
}
|
| 28139 |
],
|
| 28140 |
"logging_steps": 2,
|
|
|
|
| 28154 |
"attributes": {}
|
| 28155 |
}
|
| 28156 |
},
|
| 28157 |
+
"total_flos": 1.711784985624576e+19,
|
| 28158 |
"train_batch_size": 768,
|
| 28159 |
"trial_name": null,
|
| 28160 |
"trial_params": null
|