Training in progress, step 22000, 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 sha256:4531ecca714f467930734a738ae96aa6d8c7dde0ae2f856e01bfd7c2718c30e2
|
| 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 sha256:5d73be928861bba842ddafd6c57448be532897ea8cdcffe6ba29b9511de401e3
|
| 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 sha256:206e7b0fb5a1dee1afc9d0786fbf979a9b9a48953b1e20b8dd66c2319f838ad0
|
| 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 sha256:3e2e1da63578bda22917ec19daaed13e356a28b0f07b3611af0830560ee695f8
|
| 3 |
size 1000
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 1.
|
| 3 |
-
"best_model_checkpoint": "model/chessformer-3/checkpoint-
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -73843,6 +73843,3522 @@
|
|
| 73843 |
"eval_samples_per_second": 556.179,
|
| 73844 |
"eval_steps_per_second": 69.543,
|
| 73845 |
"step": 21000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 73846 |
}
|
| 73847 |
],
|
| 73848 |
"logging_steps": 2,
|
|
@@ -73862,7 +77378,7 @@
|
|
| 73862 |
"attributes": {}
|
| 73863 |
}
|
| 73864 |
},
|
| 73865 |
-
"total_flos": 4.
|
| 73866 |
"train_batch_size": 768,
|
| 73867 |
"trial_name": null,
|
| 73868 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 1.4320533275604248,
|
| 3 |
+
"best_model_checkpoint": "model/chessformer-3/checkpoint-22000",
|
| 4 |
+
"epoch": 0.9781690453959362,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 22000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 73843 |
"eval_samples_per_second": 556.179,
|
| 73844 |
"eval_steps_per_second": 69.543,
|
| 73845 |
"step": 21000
|
| 73846 |
+
},
|
| 73847 |
+
{
|
| 73848 |
+
"epoch": 0.9337957405184296,
|
| 73849 |
+
"grad_norm": 0.05756537243723869,
|
| 73850 |
+
"learning_rate": 1.1783100692341508e-05,
|
| 73851 |
+
"loss": 1.4479,
|
| 73852 |
+
"step": 21002
|
| 73853 |
+
},
|
| 73854 |
+
{
|
| 73855 |
+
"epoch": 0.933884664977102,
|
| 73856 |
+
"grad_norm": 0.0578317753970623,
|
| 73857 |
+
"learning_rate": 1.17515715586613e-05,
|
| 73858 |
+
"loss": 1.4463,
|
| 73859 |
+
"step": 21004
|
| 73860 |
+
},
|
| 73861 |
+
{
|
| 73862 |
+
"epoch": 0.9339735894357744,
|
| 73863 |
+
"grad_norm": 0.056406132876873016,
|
| 73864 |
+
"learning_rate": 1.172008416260839e-05,
|
| 73865 |
+
"loss": 1.4472,
|
| 73866 |
+
"step": 21006
|
| 73867 |
+
},
|
| 73868 |
+
{
|
| 73869 |
+
"epoch": 0.9340625138944467,
|
| 73870 |
+
"grad_norm": 0.0582900270819664,
|
| 73871 |
+
"learning_rate": 1.1688638506874405e-05,
|
| 73872 |
+
"loss": 1.452,
|
| 73873 |
+
"step": 21008
|
| 73874 |
+
},
|
| 73875 |
+
{
|
| 73876 |
+
"epoch": 0.934151438353119,
|
| 73877 |
+
"grad_norm": 0.05787406116724014,
|
| 73878 |
+
"learning_rate": 1.1657234594147636e-05,
|
| 73879 |
+
"loss": 1.4507,
|
| 73880 |
+
"step": 21010
|
| 73881 |
+
},
|
| 73882 |
+
{
|
| 73883 |
+
"epoch": 0.9342403628117913,
|
| 73884 |
+
"grad_norm": 0.05701528117060661,
|
| 73885 |
+
"learning_rate": 1.1625872427112439e-05,
|
| 73886 |
+
"loss": 1.4506,
|
| 73887 |
+
"step": 21012
|
| 73888 |
+
},
|
| 73889 |
+
{
|
| 73890 |
+
"epoch": 0.9343292872704637,
|
| 73891 |
+
"grad_norm": 0.05795227363705635,
|
| 73892 |
+
"learning_rate": 1.1594552008449888e-05,
|
| 73893 |
+
"loss": 1.4454,
|
| 73894 |
+
"step": 21014
|
| 73895 |
+
},
|
| 73896 |
+
{
|
| 73897 |
+
"epoch": 0.9344182117291361,
|
| 73898 |
+
"grad_norm": 0.05739319324493408,
|
| 73899 |
+
"learning_rate": 1.1563273340837289e-05,
|
| 73900 |
+
"loss": 1.451,
|
| 73901 |
+
"step": 21016
|
| 73902 |
+
},
|
| 73903 |
+
{
|
| 73904 |
+
"epoch": 0.9345071361878085,
|
| 73905 |
+
"grad_norm": 0.05824971944093704,
|
| 73906 |
+
"learning_rate": 1.1532036426948667e-05,
|
| 73907 |
+
"loss": 1.4479,
|
| 73908 |
+
"step": 21018
|
| 73909 |
+
},
|
| 73910 |
+
{
|
| 73911 |
+
"epoch": 0.9345960606464808,
|
| 73912 |
+
"grad_norm": 0.05758555978536606,
|
| 73913 |
+
"learning_rate": 1.1500841269454166e-05,
|
| 73914 |
+
"loss": 1.4407,
|
| 73915 |
+
"step": 21020
|
| 73916 |
+
},
|
| 73917 |
+
{
|
| 73918 |
+
"epoch": 0.9346849851051532,
|
| 73919 |
+
"grad_norm": 0.05768615007400513,
|
| 73920 |
+
"learning_rate": 1.1469687871020428e-05,
|
| 73921 |
+
"loss": 1.4472,
|
| 73922 |
+
"step": 21022
|
| 73923 |
+
},
|
| 73924 |
+
{
|
| 73925 |
+
"epoch": 0.9347739095638256,
|
| 73926 |
+
"grad_norm": 0.057626668363809586,
|
| 73927 |
+
"learning_rate": 1.143857623431066e-05,
|
| 73928 |
+
"loss": 1.4431,
|
| 73929 |
+
"step": 21024
|
| 73930 |
+
},
|
| 73931 |
+
{
|
| 73932 |
+
"epoch": 0.9348628340224979,
|
| 73933 |
+
"grad_norm": 0.057193391025066376,
|
| 73934 |
+
"learning_rate": 1.1407506361984509e-05,
|
| 73935 |
+
"loss": 1.4451,
|
| 73936 |
+
"step": 21026
|
| 73937 |
+
},
|
| 73938 |
+
{
|
| 73939 |
+
"epoch": 0.9349517584811703,
|
| 73940 |
+
"grad_norm": 0.05753851309418678,
|
| 73941 |
+
"learning_rate": 1.1376478256697797e-05,
|
| 73942 |
+
"loss": 1.4413,
|
| 73943 |
+
"step": 21028
|
| 73944 |
+
},
|
| 73945 |
+
{
|
| 73946 |
+
"epoch": 0.9350406829398425,
|
| 73947 |
+
"grad_norm": 0.05880241096019745,
|
| 73948 |
+
"learning_rate": 1.1345491921102957e-05,
|
| 73949 |
+
"loss": 1.4487,
|
| 73950 |
+
"step": 21030
|
| 73951 |
+
},
|
| 73952 |
+
{
|
| 73953 |
+
"epoch": 0.9351296073985149,
|
| 73954 |
+
"grad_norm": 0.057462744414806366,
|
| 73955 |
+
"learning_rate": 1.1314547357849037e-05,
|
| 73956 |
+
"loss": 1.4432,
|
| 73957 |
+
"step": 21032
|
| 73958 |
+
},
|
| 73959 |
+
{
|
| 73960 |
+
"epoch": 0.9352185318571873,
|
| 73961 |
+
"grad_norm": 0.05817759782075882,
|
| 73962 |
+
"learning_rate": 1.1283644569581087e-05,
|
| 73963 |
+
"loss": 1.4497,
|
| 73964 |
+
"step": 21034
|
| 73965 |
+
},
|
| 73966 |
+
{
|
| 73967 |
+
"epoch": 0.9353074563158597,
|
| 73968 |
+
"grad_norm": 0.058256879448890686,
|
| 73969 |
+
"learning_rate": 1.1252783558940882e-05,
|
| 73970 |
+
"loss": 1.4495,
|
| 73971 |
+
"step": 21036
|
| 73972 |
+
},
|
| 73973 |
+
{
|
| 73974 |
+
"epoch": 0.935396380774532,
|
| 73975 |
+
"grad_norm": 0.05921512097120285,
|
| 73976 |
+
"learning_rate": 1.1221964328566647e-05,
|
| 73977 |
+
"loss": 1.444,
|
| 73978 |
+
"step": 21038
|
| 73979 |
+
},
|
| 73980 |
+
{
|
| 73981 |
+
"epoch": 0.9354853052332044,
|
| 73982 |
+
"grad_norm": 0.05717106908559799,
|
| 73983 |
+
"learning_rate": 1.119118688109283e-05,
|
| 73984 |
+
"loss": 1.4431,
|
| 73985 |
+
"step": 21040
|
| 73986 |
+
},
|
| 73987 |
+
{
|
| 73988 |
+
"epoch": 0.9355742296918768,
|
| 73989 |
+
"grad_norm": 0.058333754539489746,
|
| 73990 |
+
"learning_rate": 1.1160451219150548e-05,
|
| 73991 |
+
"loss": 1.4481,
|
| 73992 |
+
"step": 21042
|
| 73993 |
+
},
|
| 73994 |
+
{
|
| 73995 |
+
"epoch": 0.9356631541505491,
|
| 73996 |
+
"grad_norm": 0.05773552879691124,
|
| 73997 |
+
"learning_rate": 1.112975734536703e-05,
|
| 73998 |
+
"loss": 1.4508,
|
| 73999 |
+
"step": 21044
|
| 74000 |
+
},
|
| 74001 |
+
{
|
| 74002 |
+
"epoch": 0.9357520786092215,
|
| 74003 |
+
"grad_norm": 0.058473654091358185,
|
| 74004 |
+
"learning_rate": 1.1099105262366294e-05,
|
| 74005 |
+
"loss": 1.452,
|
| 74006 |
+
"step": 21046
|
| 74007 |
+
},
|
| 74008 |
+
{
|
| 74009 |
+
"epoch": 0.9358410030678939,
|
| 74010 |
+
"grad_norm": 0.057561490684747696,
|
| 74011 |
+
"learning_rate": 1.1068494972768572e-05,
|
| 74012 |
+
"loss": 1.4452,
|
| 74013 |
+
"step": 21048
|
| 74014 |
+
},
|
| 74015 |
+
{
|
| 74016 |
+
"epoch": 0.9359299275265662,
|
| 74017 |
+
"grad_norm": 0.05912725254893303,
|
| 74018 |
+
"learning_rate": 1.1037926479190497e-05,
|
| 74019 |
+
"loss": 1.4493,
|
| 74020 |
+
"step": 21050
|
| 74021 |
+
},
|
| 74022 |
+
{
|
| 74023 |
+
"epoch": 0.9360188519852385,
|
| 74024 |
+
"grad_norm": 0.05738065019249916,
|
| 74025 |
+
"learning_rate": 1.1007399784245308e-05,
|
| 74026 |
+
"loss": 1.4438,
|
| 74027 |
+
"step": 21052
|
| 74028 |
+
},
|
| 74029 |
+
{
|
| 74030 |
+
"epoch": 0.9361077764439109,
|
| 74031 |
+
"grad_norm": 0.05802926421165466,
|
| 74032 |
+
"learning_rate": 1.0976914890542422e-05,
|
| 74033 |
+
"loss": 1.4481,
|
| 74034 |
+
"step": 21054
|
| 74035 |
+
},
|
| 74036 |
+
{
|
| 74037 |
+
"epoch": 0.9361967009025832,
|
| 74038 |
+
"grad_norm": 0.05792355164885521,
|
| 74039 |
+
"learning_rate": 1.0946471800688029e-05,
|
| 74040 |
+
"loss": 1.4466,
|
| 74041 |
+
"step": 21056
|
| 74042 |
+
},
|
| 74043 |
+
{
|
| 74044 |
+
"epoch": 0.9362856253612556,
|
| 74045 |
+
"grad_norm": 0.059775542467832565,
|
| 74046 |
+
"learning_rate": 1.0916070517284327e-05,
|
| 74047 |
+
"loss": 1.4437,
|
| 74048 |
+
"step": 21058
|
| 74049 |
+
},
|
| 74050 |
+
{
|
| 74051 |
+
"epoch": 0.936374549819928,
|
| 74052 |
+
"grad_norm": 0.058133359998464584,
|
| 74053 |
+
"learning_rate": 1.0885711042930235e-05,
|
| 74054 |
+
"loss": 1.4453,
|
| 74055 |
+
"step": 21060
|
| 74056 |
+
},
|
| 74057 |
+
{
|
| 74058 |
+
"epoch": 0.9364634742786003,
|
| 74059 |
+
"grad_norm": 0.05763556808233261,
|
| 74060 |
+
"learning_rate": 1.0855393380221068e-05,
|
| 74061 |
+
"loss": 1.4554,
|
| 74062 |
+
"step": 21062
|
| 74063 |
+
},
|
| 74064 |
+
{
|
| 74065 |
+
"epoch": 0.9365523987372727,
|
| 74066 |
+
"grad_norm": 0.058549582958221436,
|
| 74067 |
+
"learning_rate": 1.0825117531748474e-05,
|
| 74068 |
+
"loss": 1.447,
|
| 74069 |
+
"step": 21064
|
| 74070 |
+
},
|
| 74071 |
+
{
|
| 74072 |
+
"epoch": 0.9366413231959451,
|
| 74073 |
+
"grad_norm": 0.057294197380542755,
|
| 74074 |
+
"learning_rate": 1.0794883500100494e-05,
|
| 74075 |
+
"loss": 1.4436,
|
| 74076 |
+
"step": 21066
|
| 74077 |
+
},
|
| 74078 |
+
{
|
| 74079 |
+
"epoch": 0.9367302476546174,
|
| 74080 |
+
"grad_norm": 0.05782110616564751,
|
| 74081 |
+
"learning_rate": 1.0764691287861783e-05,
|
| 74082 |
+
"loss": 1.4539,
|
| 74083 |
+
"step": 21068
|
| 74084 |
+
},
|
| 74085 |
+
{
|
| 74086 |
+
"epoch": 0.9368191721132898,
|
| 74087 |
+
"grad_norm": 0.05737898498773575,
|
| 74088 |
+
"learning_rate": 1.0734540897613221e-05,
|
| 74089 |
+
"loss": 1.4546,
|
| 74090 |
+
"step": 21070
|
| 74091 |
+
},
|
| 74092 |
+
{
|
| 74093 |
+
"epoch": 0.9369080965719622,
|
| 74094 |
+
"grad_norm": 0.05745162069797516,
|
| 74095 |
+
"learning_rate": 1.0704432331932245e-05,
|
| 74096 |
+
"loss": 1.4459,
|
| 74097 |
+
"step": 21072
|
| 74098 |
+
},
|
| 74099 |
+
{
|
| 74100 |
+
"epoch": 0.9369970210306344,
|
| 74101 |
+
"grad_norm": 0.05851050093770027,
|
| 74102 |
+
"learning_rate": 1.0674365593392633e-05,
|
| 74103 |
+
"loss": 1.4427,
|
| 74104 |
+
"step": 21074
|
| 74105 |
+
},
|
| 74106 |
+
{
|
| 74107 |
+
"epoch": 0.9370859454893068,
|
| 74108 |
+
"grad_norm": 0.058115605264902115,
|
| 74109 |
+
"learning_rate": 1.0644340684564657e-05,
|
| 74110 |
+
"loss": 1.442,
|
| 74111 |
+
"step": 21076
|
| 74112 |
+
},
|
| 74113 |
+
{
|
| 74114 |
+
"epoch": 0.9371748699479792,
|
| 74115 |
+
"grad_norm": 0.05693921446800232,
|
| 74116 |
+
"learning_rate": 1.061435760801499e-05,
|
| 74117 |
+
"loss": 1.4476,
|
| 74118 |
+
"step": 21078
|
| 74119 |
+
},
|
| 74120 |
+
{
|
| 74121 |
+
"epoch": 0.9372637944066515,
|
| 74122 |
+
"grad_norm": 0.05739055946469307,
|
| 74123 |
+
"learning_rate": 1.0584416366306581e-05,
|
| 74124 |
+
"loss": 1.4492,
|
| 74125 |
+
"step": 21080
|
| 74126 |
+
},
|
| 74127 |
+
{
|
| 74128 |
+
"epoch": 0.9373527188653239,
|
| 74129 |
+
"grad_norm": 0.05716922506690025,
|
| 74130 |
+
"learning_rate": 1.0554516961999105e-05,
|
| 74131 |
+
"loss": 1.4438,
|
| 74132 |
+
"step": 21082
|
| 74133 |
+
},
|
| 74134 |
+
{
|
| 74135 |
+
"epoch": 0.9374416433239963,
|
| 74136 |
+
"grad_norm": 0.05674619972705841,
|
| 74137 |
+
"learning_rate": 1.0524659397648462e-05,
|
| 74138 |
+
"loss": 1.4482,
|
| 74139 |
+
"step": 21084
|
| 74140 |
+
},
|
| 74141 |
+
{
|
| 74142 |
+
"epoch": 0.9375305677826686,
|
| 74143 |
+
"grad_norm": 0.058578867465257645,
|
| 74144 |
+
"learning_rate": 1.0494843675806942e-05,
|
| 74145 |
+
"loss": 1.4419,
|
| 74146 |
+
"step": 21086
|
| 74147 |
+
},
|
| 74148 |
+
{
|
| 74149 |
+
"epoch": 0.937619492241341,
|
| 74150 |
+
"grad_norm": 0.05818881839513779,
|
| 74151 |
+
"learning_rate": 1.0465069799023397e-05,
|
| 74152 |
+
"loss": 1.4496,
|
| 74153 |
+
"step": 21088
|
| 74154 |
+
},
|
| 74155 |
+
{
|
| 74156 |
+
"epoch": 0.9377084167000134,
|
| 74157 |
+
"grad_norm": 0.058019597083330154,
|
| 74158 |
+
"learning_rate": 1.0435337769843012e-05,
|
| 74159 |
+
"loss": 1.4423,
|
| 74160 |
+
"step": 21090
|
| 74161 |
+
},
|
| 74162 |
+
{
|
| 74163 |
+
"epoch": 0.9377973411586857,
|
| 74164 |
+
"grad_norm": 0.05781329795718193,
|
| 74165 |
+
"learning_rate": 1.0405647590807366e-05,
|
| 74166 |
+
"loss": 1.4467,
|
| 74167 |
+
"step": 21092
|
| 74168 |
+
},
|
| 74169 |
+
{
|
| 74170 |
+
"epoch": 0.9378862656173581,
|
| 74171 |
+
"grad_norm": 0.05810556560754776,
|
| 74172 |
+
"learning_rate": 1.0375999264454483e-05,
|
| 74173 |
+
"loss": 1.4453,
|
| 74174 |
+
"step": 21094
|
| 74175 |
+
},
|
| 74176 |
+
{
|
| 74177 |
+
"epoch": 0.9379751900760304,
|
| 74178 |
+
"grad_norm": 0.05800723284482956,
|
| 74179 |
+
"learning_rate": 1.0346392793319005e-05,
|
| 74180 |
+
"loss": 1.442,
|
| 74181 |
+
"step": 21096
|
| 74182 |
+
},
|
| 74183 |
+
{
|
| 74184 |
+
"epoch": 0.9380641145347027,
|
| 74185 |
+
"grad_norm": 0.0583004355430603,
|
| 74186 |
+
"learning_rate": 1.031682817993157e-05,
|
| 74187 |
+
"loss": 1.447,
|
| 74188 |
+
"step": 21098
|
| 74189 |
+
},
|
| 74190 |
+
{
|
| 74191 |
+
"epoch": 0.9381530389933751,
|
| 74192 |
+
"grad_norm": 0.05934945121407509,
|
| 74193 |
+
"learning_rate": 1.0287305426819771e-05,
|
| 74194 |
+
"loss": 1.4457,
|
| 74195 |
+
"step": 21100
|
| 74196 |
+
},
|
| 74197 |
+
{
|
| 74198 |
+
"epoch": 0.9382419634520475,
|
| 74199 |
+
"grad_norm": 0.05703364312648773,
|
| 74200 |
+
"learning_rate": 1.0257824536507088e-05,
|
| 74201 |
+
"loss": 1.4435,
|
| 74202 |
+
"step": 21102
|
| 74203 |
+
},
|
| 74204 |
+
{
|
| 74205 |
+
"epoch": 0.9383308879107198,
|
| 74206 |
+
"grad_norm": 0.057916540652513504,
|
| 74207 |
+
"learning_rate": 1.0228385511513782e-05,
|
| 74208 |
+
"loss": 1.4447,
|
| 74209 |
+
"step": 21104
|
| 74210 |
+
},
|
| 74211 |
+
{
|
| 74212 |
+
"epoch": 0.9384198123693922,
|
| 74213 |
+
"grad_norm": 0.05742233991622925,
|
| 74214 |
+
"learning_rate": 1.0198988354356509e-05,
|
| 74215 |
+
"loss": 1.442,
|
| 74216 |
+
"step": 21106
|
| 74217 |
+
},
|
| 74218 |
+
{
|
| 74219 |
+
"epoch": 0.9385087368280646,
|
| 74220 |
+
"grad_norm": 0.05896897241473198,
|
| 74221 |
+
"learning_rate": 1.0169633067548145e-05,
|
| 74222 |
+
"loss": 1.4476,
|
| 74223 |
+
"step": 21108
|
| 74224 |
+
},
|
| 74225 |
+
{
|
| 74226 |
+
"epoch": 0.9385976612867369,
|
| 74227 |
+
"grad_norm": 0.05869178846478462,
|
| 74228 |
+
"learning_rate": 1.0140319653598185e-05,
|
| 74229 |
+
"loss": 1.4472,
|
| 74230 |
+
"step": 21110
|
| 74231 |
+
},
|
| 74232 |
+
{
|
| 74233 |
+
"epoch": 0.9386865857454093,
|
| 74234 |
+
"grad_norm": 0.05712299793958664,
|
| 74235 |
+
"learning_rate": 1.0111048115012456e-05,
|
| 74236 |
+
"loss": 1.4457,
|
| 74237 |
+
"step": 21112
|
| 74238 |
+
},
|
| 74239 |
+
{
|
| 74240 |
+
"epoch": 0.9387755102040817,
|
| 74241 |
+
"grad_norm": 0.05764850974082947,
|
| 74242 |
+
"learning_rate": 1.0081818454293124e-05,
|
| 74243 |
+
"loss": 1.4435,
|
| 74244 |
+
"step": 21114
|
| 74245 |
+
},
|
| 74246 |
+
{
|
| 74247 |
+
"epoch": 0.938864434662754,
|
| 74248 |
+
"grad_norm": 0.05720280855894089,
|
| 74249 |
+
"learning_rate": 1.0052630673939023e-05,
|
| 74250 |
+
"loss": 1.4491,
|
| 74251 |
+
"step": 21116
|
| 74252 |
+
},
|
| 74253 |
+
{
|
| 74254 |
+
"epoch": 0.9389533591214263,
|
| 74255 |
+
"grad_norm": 0.058444440364837646,
|
| 74256 |
+
"learning_rate": 1.0023484776445158e-05,
|
| 74257 |
+
"loss": 1.4496,
|
| 74258 |
+
"step": 21118
|
| 74259 |
+
},
|
| 74260 |
+
{
|
| 74261 |
+
"epoch": 0.9390422835800987,
|
| 74262 |
+
"grad_norm": 0.0576569139957428,
|
| 74263 |
+
"learning_rate": 9.994380764303091e-06,
|
| 74264 |
+
"loss": 1.4523,
|
| 74265 |
+
"step": 21120
|
| 74266 |
+
},
|
| 74267 |
+
{
|
| 74268 |
+
"epoch": 0.939131208038771,
|
| 74269 |
+
"grad_norm": 0.05701880529522896,
|
| 74270 |
+
"learning_rate": 9.96531864000072e-06,
|
| 74271 |
+
"loss": 1.4496,
|
| 74272 |
+
"step": 21122
|
| 74273 |
+
},
|
| 74274 |
+
{
|
| 74275 |
+
"epoch": 0.9392201324974434,
|
| 74276 |
+
"grad_norm": 0.05704936385154724,
|
| 74277 |
+
"learning_rate": 9.936298406022393e-06,
|
| 74278 |
+
"loss": 1.4445,
|
| 74279 |
+
"step": 21124
|
| 74280 |
+
},
|
| 74281 |
+
{
|
| 74282 |
+
"epoch": 0.9393090569561158,
|
| 74283 |
+
"grad_norm": 0.05782592296600342,
|
| 74284 |
+
"learning_rate": 9.907320064848956e-06,
|
| 74285 |
+
"loss": 1.4462,
|
| 74286 |
+
"step": 21126
|
| 74287 |
+
},
|
| 74288 |
+
{
|
| 74289 |
+
"epoch": 0.9393979814147881,
|
| 74290 |
+
"grad_norm": 0.05813451111316681,
|
| 74291 |
+
"learning_rate": 9.878383618957599e-06,
|
| 74292 |
+
"loss": 1.4475,
|
| 74293 |
+
"step": 21128
|
| 74294 |
+
},
|
| 74295 |
+
{
|
| 74296 |
+
"epoch": 0.9394869058734605,
|
| 74297 |
+
"grad_norm": 0.05756333842873573,
|
| 74298 |
+
"learning_rate": 9.849489070821893e-06,
|
| 74299 |
+
"loss": 1.4443,
|
| 74300 |
+
"step": 21130
|
| 74301 |
+
},
|
| 74302 |
+
{
|
| 74303 |
+
"epoch": 0.9395758303321329,
|
| 74304 |
+
"grad_norm": 0.05767510086297989,
|
| 74305 |
+
"learning_rate": 9.820636422911811e-06,
|
| 74306 |
+
"loss": 1.4527,
|
| 74307 |
+
"step": 21132
|
| 74308 |
+
},
|
| 74309 |
+
{
|
| 74310 |
+
"epoch": 0.9396647547908052,
|
| 74311 |
+
"grad_norm": 0.057892054319381714,
|
| 74312 |
+
"learning_rate": 9.791825677693988e-06,
|
| 74313 |
+
"loss": 1.4458,
|
| 74314 |
+
"step": 21134
|
| 74315 |
+
},
|
| 74316 |
+
{
|
| 74317 |
+
"epoch": 0.9397536792494776,
|
| 74318 |
+
"grad_norm": 0.057954948395490646,
|
| 74319 |
+
"learning_rate": 9.763056837631123e-06,
|
| 74320 |
+
"loss": 1.4412,
|
| 74321 |
+
"step": 21136
|
| 74322 |
+
},
|
| 74323 |
+
{
|
| 74324 |
+
"epoch": 0.93984260370815,
|
| 74325 |
+
"grad_norm": 0.05756092816591263,
|
| 74326 |
+
"learning_rate": 9.734329905182527e-06,
|
| 74327 |
+
"loss": 1.4492,
|
| 74328 |
+
"step": 21138
|
| 74329 |
+
},
|
| 74330 |
+
{
|
| 74331 |
+
"epoch": 0.9399315281668222,
|
| 74332 |
+
"grad_norm": 0.05734538286924362,
|
| 74333 |
+
"learning_rate": 9.705644882803954e-06,
|
| 74334 |
+
"loss": 1.4451,
|
| 74335 |
+
"step": 21140
|
| 74336 |
+
},
|
| 74337 |
+
{
|
| 74338 |
+
"epoch": 0.9400204526254946,
|
| 74339 |
+
"grad_norm": 0.057080693542957306,
|
| 74340 |
+
"learning_rate": 9.677001772947614e-06,
|
| 74341 |
+
"loss": 1.4529,
|
| 74342 |
+
"step": 21142
|
| 74343 |
+
},
|
| 74344 |
+
{
|
| 74345 |
+
"epoch": 0.940109377084167,
|
| 74346 |
+
"grad_norm": 0.057391393929719925,
|
| 74347 |
+
"learning_rate": 9.648400578061822e-06,
|
| 74348 |
+
"loss": 1.4459,
|
| 74349 |
+
"step": 21144
|
| 74350 |
+
},
|
| 74351 |
+
{
|
| 74352 |
+
"epoch": 0.9401983015428393,
|
| 74353 |
+
"grad_norm": 0.0573163703083992,
|
| 74354 |
+
"learning_rate": 9.619841300591736e-06,
|
| 74355 |
+
"loss": 1.4446,
|
| 74356 |
+
"step": 21146
|
| 74357 |
+
},
|
| 74358 |
+
{
|
| 74359 |
+
"epoch": 0.9402872260015117,
|
| 74360 |
+
"grad_norm": 0.058304496109485626,
|
| 74361 |
+
"learning_rate": 9.591323942978624e-06,
|
| 74362 |
+
"loss": 1.4483,
|
| 74363 |
+
"step": 21148
|
| 74364 |
+
},
|
| 74365 |
+
{
|
| 74366 |
+
"epoch": 0.9403761504601841,
|
| 74367 |
+
"grad_norm": 0.059254664927721024,
|
| 74368 |
+
"learning_rate": 9.562848507660316e-06,
|
| 74369 |
+
"loss": 1.4455,
|
| 74370 |
+
"step": 21150
|
| 74371 |
+
},
|
| 74372 |
+
{
|
| 74373 |
+
"epoch": 0.9404650749188564,
|
| 74374 |
+
"grad_norm": 0.05732493847608566,
|
| 74375 |
+
"learning_rate": 9.534414997070973e-06,
|
| 74376 |
+
"loss": 1.4498,
|
| 74377 |
+
"step": 21152
|
| 74378 |
+
},
|
| 74379 |
+
{
|
| 74380 |
+
"epoch": 0.9405539993775288,
|
| 74381 |
+
"grad_norm": 0.05808182805776596,
|
| 74382 |
+
"learning_rate": 9.506023413641263e-06,
|
| 74383 |
+
"loss": 1.4464,
|
| 74384 |
+
"step": 21154
|
| 74385 |
+
},
|
| 74386 |
+
{
|
| 74387 |
+
"epoch": 0.9406429238362012,
|
| 74388 |
+
"grad_norm": 0.05795228108763695,
|
| 74389 |
+
"learning_rate": 9.477673759798188e-06,
|
| 74390 |
+
"loss": 1.444,
|
| 74391 |
+
"step": 21156
|
| 74392 |
+
},
|
| 74393 |
+
{
|
| 74394 |
+
"epoch": 0.9407318482948736,
|
| 74395 |
+
"grad_norm": 0.0579368956387043,
|
| 74396 |
+
"learning_rate": 9.449366037965313e-06,
|
| 74397 |
+
"loss": 1.4475,
|
| 74398 |
+
"step": 21158
|
| 74399 |
+
},
|
| 74400 |
+
{
|
| 74401 |
+
"epoch": 0.9408207727535458,
|
| 74402 |
+
"grad_norm": 0.05821996554732323,
|
| 74403 |
+
"learning_rate": 9.421100250562309e-06,
|
| 74404 |
+
"loss": 1.4519,
|
| 74405 |
+
"step": 21160
|
| 74406 |
+
},
|
| 74407 |
+
{
|
| 74408 |
+
"epoch": 0.9409096972122182,
|
| 74409 |
+
"grad_norm": 0.058779843151569366,
|
| 74410 |
+
"learning_rate": 9.392876400005579e-06,
|
| 74411 |
+
"loss": 1.4436,
|
| 74412 |
+
"step": 21162
|
| 74413 |
+
},
|
| 74414 |
+
{
|
| 74415 |
+
"epoch": 0.9409986216708905,
|
| 74416 |
+
"grad_norm": 0.05720444768667221,
|
| 74417 |
+
"learning_rate": 9.364694488707858e-06,
|
| 74418 |
+
"loss": 1.4545,
|
| 74419 |
+
"step": 21164
|
| 74420 |
+
},
|
| 74421 |
+
{
|
| 74422 |
+
"epoch": 0.9410875461295629,
|
| 74423 |
+
"grad_norm": 0.057297226041555405,
|
| 74424 |
+
"learning_rate": 9.336554519078221e-06,
|
| 74425 |
+
"loss": 1.4425,
|
| 74426 |
+
"step": 21166
|
| 74427 |
+
},
|
| 74428 |
+
{
|
| 74429 |
+
"epoch": 0.9411764705882353,
|
| 74430 |
+
"grad_norm": 0.057402316480875015,
|
| 74431 |
+
"learning_rate": 9.308456493522133e-06,
|
| 74432 |
+
"loss": 1.4458,
|
| 74433 |
+
"step": 21168
|
| 74434 |
+
},
|
| 74435 |
+
{
|
| 74436 |
+
"epoch": 0.9412653950469076,
|
| 74437 |
+
"grad_norm": 0.05686284974217415,
|
| 74438 |
+
"learning_rate": 9.280400414441613e-06,
|
| 74439 |
+
"loss": 1.4437,
|
| 74440 |
+
"step": 21170
|
| 74441 |
+
},
|
| 74442 |
+
{
|
| 74443 |
+
"epoch": 0.94135431950558,
|
| 74444 |
+
"grad_norm": 0.05828374624252319,
|
| 74445 |
+
"learning_rate": 9.252386284235026e-06,
|
| 74446 |
+
"loss": 1.4463,
|
| 74447 |
+
"step": 21172
|
| 74448 |
+
},
|
| 74449 |
+
{
|
| 74450 |
+
"epoch": 0.9414432439642524,
|
| 74451 |
+
"grad_norm": 0.05804366618394852,
|
| 74452 |
+
"learning_rate": 9.224414105297064e-06,
|
| 74453 |
+
"loss": 1.4455,
|
| 74454 |
+
"step": 21174
|
| 74455 |
+
},
|
| 74456 |
+
{
|
| 74457 |
+
"epoch": 0.9415321684229248,
|
| 74458 |
+
"grad_norm": 0.058524325489997864,
|
| 74459 |
+
"learning_rate": 9.19648388001898e-06,
|
| 74460 |
+
"loss": 1.4406,
|
| 74461 |
+
"step": 21176
|
| 74462 |
+
},
|
| 74463 |
+
{
|
| 74464 |
+
"epoch": 0.9416210928815971,
|
| 74465 |
+
"grad_norm": 0.05712493881583214,
|
| 74466 |
+
"learning_rate": 9.168595610788365e-06,
|
| 74467 |
+
"loss": 1.445,
|
| 74468 |
+
"step": 21178
|
| 74469 |
+
},
|
| 74470 |
+
{
|
| 74471 |
+
"epoch": 0.9417100173402695,
|
| 74472 |
+
"grad_norm": 0.05792125314474106,
|
| 74473 |
+
"learning_rate": 9.140749299989205e-06,
|
| 74474 |
+
"loss": 1.4381,
|
| 74475 |
+
"step": 21180
|
| 74476 |
+
},
|
| 74477 |
+
{
|
| 74478 |
+
"epoch": 0.9417989417989417,
|
| 74479 |
+
"grad_norm": 0.05810127779841423,
|
| 74480 |
+
"learning_rate": 9.112944950001978e-06,
|
| 74481 |
+
"loss": 1.4472,
|
| 74482 |
+
"step": 21182
|
| 74483 |
+
},
|
| 74484 |
+
{
|
| 74485 |
+
"epoch": 0.9418878662576141,
|
| 74486 |
+
"grad_norm": 0.05826450139284134,
|
| 74487 |
+
"learning_rate": 9.085182563203453e-06,
|
| 74488 |
+
"loss": 1.4468,
|
| 74489 |
+
"step": 21184
|
| 74490 |
+
},
|
| 74491 |
+
{
|
| 74492 |
+
"epoch": 0.9419767907162865,
|
| 74493 |
+
"grad_norm": 0.0577828474342823,
|
| 74494 |
+
"learning_rate": 9.057462141966898e-06,
|
| 74495 |
+
"loss": 1.4437,
|
| 74496 |
+
"step": 21186
|
| 74497 |
+
},
|
| 74498 |
+
{
|
| 74499 |
+
"epoch": 0.9420657151749589,
|
| 74500 |
+
"grad_norm": 0.05752667412161827,
|
| 74501 |
+
"learning_rate": 9.02978368866203e-06,
|
| 74502 |
+
"loss": 1.4479,
|
| 74503 |
+
"step": 21188
|
| 74504 |
+
},
|
| 74505 |
+
{
|
| 74506 |
+
"epoch": 0.9421546396336312,
|
| 74507 |
+
"grad_norm": 0.05732182040810585,
|
| 74508 |
+
"learning_rate": 9.002147205654843e-06,
|
| 74509 |
+
"loss": 1.4476,
|
| 74510 |
+
"step": 21190
|
| 74511 |
+
},
|
| 74512 |
+
{
|
| 74513 |
+
"epoch": 0.9422435640923036,
|
| 74514 |
+
"grad_norm": 0.057963550090789795,
|
| 74515 |
+
"learning_rate": 8.974552695307948e-06,
|
| 74516 |
+
"loss": 1.4481,
|
| 74517 |
+
"step": 21192
|
| 74518 |
+
},
|
| 74519 |
+
{
|
| 74520 |
+
"epoch": 0.942332488550976,
|
| 74521 |
+
"grad_norm": 0.05983150750398636,
|
| 74522 |
+
"learning_rate": 8.947000159980067e-06,
|
| 74523 |
+
"loss": 1.4485,
|
| 74524 |
+
"step": 21194
|
| 74525 |
+
},
|
| 74526 |
+
{
|
| 74527 |
+
"epoch": 0.9424214130096483,
|
| 74528 |
+
"grad_norm": 0.05860733240842819,
|
| 74529 |
+
"learning_rate": 8.919489602026653e-06,
|
| 74530 |
+
"loss": 1.4393,
|
| 74531 |
+
"step": 21196
|
| 74532 |
+
},
|
| 74533 |
+
{
|
| 74534 |
+
"epoch": 0.9425103374683207,
|
| 74535 |
+
"grad_norm": 0.05784778296947479,
|
| 74536 |
+
"learning_rate": 8.892021023799435e-06,
|
| 74537 |
+
"loss": 1.4464,
|
| 74538 |
+
"step": 21198
|
| 74539 |
+
},
|
| 74540 |
+
{
|
| 74541 |
+
"epoch": 0.9425992619269931,
|
| 74542 |
+
"grad_norm": 0.05762090906500816,
|
| 74543 |
+
"learning_rate": 8.864594427646478e-06,
|
| 74544 |
+
"loss": 1.4431,
|
| 74545 |
+
"step": 21200
|
| 74546 |
+
},
|
| 74547 |
+
{
|
| 74548 |
+
"epoch": 0.9426881863856654,
|
| 74549 |
+
"grad_norm": 0.0579993836581707,
|
| 74550 |
+
"learning_rate": 8.837209815912295e-06,
|
| 74551 |
+
"loss": 1.4431,
|
| 74552 |
+
"step": 21202
|
| 74553 |
+
},
|
| 74554 |
+
{
|
| 74555 |
+
"epoch": 0.9427771108443377,
|
| 74556 |
+
"grad_norm": 0.0585738830268383,
|
| 74557 |
+
"learning_rate": 8.809867190938014e-06,
|
| 74558 |
+
"loss": 1.4496,
|
| 74559 |
+
"step": 21204
|
| 74560 |
+
},
|
| 74561 |
+
{
|
| 74562 |
+
"epoch": 0.94286603530301,
|
| 74563 |
+
"grad_norm": 0.05800570920109749,
|
| 74564 |
+
"learning_rate": 8.782566555060822e-06,
|
| 74565 |
+
"loss": 1.4428,
|
| 74566 |
+
"step": 21206
|
| 74567 |
+
},
|
| 74568 |
+
{
|
| 74569 |
+
"epoch": 0.9429549597616824,
|
| 74570 |
+
"grad_norm": 0.058509331196546555,
|
| 74571 |
+
"learning_rate": 8.755307910614574e-06,
|
| 74572 |
+
"loss": 1.4455,
|
| 74573 |
+
"step": 21208
|
| 74574 |
+
},
|
| 74575 |
+
{
|
| 74576 |
+
"epoch": 0.9430438842203548,
|
| 74577 |
+
"grad_norm": 0.05728688836097717,
|
| 74578 |
+
"learning_rate": 8.728091259929404e-06,
|
| 74579 |
+
"loss": 1.4459,
|
| 74580 |
+
"step": 21210
|
| 74581 |
+
},
|
| 74582 |
+
{
|
| 74583 |
+
"epoch": 0.9431328086790272,
|
| 74584 |
+
"grad_norm": 0.057143598794937134,
|
| 74585 |
+
"learning_rate": 8.70091660533201e-06,
|
| 74586 |
+
"loss": 1.4493,
|
| 74587 |
+
"step": 21212
|
| 74588 |
+
},
|
| 74589 |
+
{
|
| 74590 |
+
"epoch": 0.9432217331376995,
|
| 74591 |
+
"grad_norm": 0.05791422724723816,
|
| 74592 |
+
"learning_rate": 8.673783949145364e-06,
|
| 74593 |
+
"loss": 1.4482,
|
| 74594 |
+
"step": 21214
|
| 74595 |
+
},
|
| 74596 |
+
{
|
| 74597 |
+
"epoch": 0.9433106575963719,
|
| 74598 |
+
"grad_norm": 0.05727590247988701,
|
| 74599 |
+
"learning_rate": 8.646693293688834e-06,
|
| 74600 |
+
"loss": 1.4514,
|
| 74601 |
+
"step": 21216
|
| 74602 |
+
},
|
| 74603 |
+
{
|
| 74604 |
+
"epoch": 0.9433995820550443,
|
| 74605 |
+
"grad_norm": 0.05757603794336319,
|
| 74606 |
+
"learning_rate": 8.61964464127829e-06,
|
| 74607 |
+
"loss": 1.4526,
|
| 74608 |
+
"step": 21218
|
| 74609 |
+
},
|
| 74610 |
+
{
|
| 74611 |
+
"epoch": 0.9434885065137166,
|
| 74612 |
+
"grad_norm": 0.057514291256666183,
|
| 74613 |
+
"learning_rate": 8.592637994225994e-06,
|
| 74614 |
+
"loss": 1.4421,
|
| 74615 |
+
"step": 21220
|
| 74616 |
+
},
|
| 74617 |
+
{
|
| 74618 |
+
"epoch": 0.943577430972389,
|
| 74619 |
+
"grad_norm": 0.05774332210421562,
|
| 74620 |
+
"learning_rate": 8.565673354840543e-06,
|
| 74621 |
+
"loss": 1.4437,
|
| 74622 |
+
"step": 21222
|
| 74623 |
+
},
|
| 74624 |
+
{
|
| 74625 |
+
"epoch": 0.9436663554310614,
|
| 74626 |
+
"grad_norm": 0.05754677578806877,
|
| 74627 |
+
"learning_rate": 8.53875072542698e-06,
|
| 74628 |
+
"loss": 1.4513,
|
| 74629 |
+
"step": 21224
|
| 74630 |
+
},
|
| 74631 |
+
{
|
| 74632 |
+
"epoch": 0.9437552798897336,
|
| 74633 |
+
"grad_norm": 0.05838412791490555,
|
| 74634 |
+
"learning_rate": 8.511870108286857e-06,
|
| 74635 |
+
"loss": 1.449,
|
| 74636 |
+
"step": 21226
|
| 74637 |
+
},
|
| 74638 |
+
{
|
| 74639 |
+
"epoch": 0.943844204348406,
|
| 74640 |
+
"grad_norm": 0.05649451166391373,
|
| 74641 |
+
"learning_rate": 8.485031505718e-06,
|
| 74642 |
+
"loss": 1.4425,
|
| 74643 |
+
"step": 21228
|
| 74644 |
+
},
|
| 74645 |
+
{
|
| 74646 |
+
"epoch": 0.9439331288070784,
|
| 74647 |
+
"grad_norm": 0.05882545933127403,
|
| 74648 |
+
"learning_rate": 8.458234920014684e-06,
|
| 74649 |
+
"loss": 1.4436,
|
| 74650 |
+
"step": 21230
|
| 74651 |
+
},
|
| 74652 |
+
{
|
| 74653 |
+
"epoch": 0.9440220532657507,
|
| 74654 |
+
"grad_norm": 0.05716662108898163,
|
| 74655 |
+
"learning_rate": 8.431480353467524e-06,
|
| 74656 |
+
"loss": 1.4402,
|
| 74657 |
+
"step": 21232
|
| 74658 |
+
},
|
| 74659 |
+
{
|
| 74660 |
+
"epoch": 0.9441109777244231,
|
| 74661 |
+
"grad_norm": 0.05844126269221306,
|
| 74662 |
+
"learning_rate": 8.404767808363744e-06,
|
| 74663 |
+
"loss": 1.4457,
|
| 74664 |
+
"step": 21234
|
| 74665 |
+
},
|
| 74666 |
+
{
|
| 74667 |
+
"epoch": 0.9441999021830955,
|
| 74668 |
+
"grad_norm": 0.05845796316862106,
|
| 74669 |
+
"learning_rate": 8.378097286986852e-06,
|
| 74670 |
+
"loss": 1.4508,
|
| 74671 |
+
"step": 21236
|
| 74672 |
+
},
|
| 74673 |
+
{
|
| 74674 |
+
"epoch": 0.9442888266417678,
|
| 74675 |
+
"grad_norm": 0.059185273945331573,
|
| 74676 |
+
"learning_rate": 8.351468791616634e-06,
|
| 74677 |
+
"loss": 1.4452,
|
| 74678 |
+
"step": 21238
|
| 74679 |
+
},
|
| 74680 |
+
{
|
| 74681 |
+
"epoch": 0.9443777511004402,
|
| 74682 |
+
"grad_norm": 0.05794282630085945,
|
| 74683 |
+
"learning_rate": 8.324882324529548e-06,
|
| 74684 |
+
"loss": 1.4503,
|
| 74685 |
+
"step": 21240
|
| 74686 |
+
},
|
| 74687 |
+
{
|
| 74688 |
+
"epoch": 0.9444666755591126,
|
| 74689 |
+
"grad_norm": 0.05867988243699074,
|
| 74690 |
+
"learning_rate": 8.29833788799822e-06,
|
| 74691 |
+
"loss": 1.4503,
|
| 74692 |
+
"step": 21242
|
| 74693 |
+
},
|
| 74694 |
+
{
|
| 74695 |
+
"epoch": 0.9445556000177849,
|
| 74696 |
+
"grad_norm": 0.05754483863711357,
|
| 74697 |
+
"learning_rate": 8.271835484291835e-06,
|
| 74698 |
+
"loss": 1.4445,
|
| 74699 |
+
"step": 21244
|
| 74700 |
+
},
|
| 74701 |
+
{
|
| 74702 |
+
"epoch": 0.9446445244764573,
|
| 74703 |
+
"grad_norm": 0.05763980373740196,
|
| 74704 |
+
"learning_rate": 8.245375115675912e-06,
|
| 74705 |
+
"loss": 1.4464,
|
| 74706 |
+
"step": 21246
|
| 74707 |
+
},
|
| 74708 |
+
{
|
| 74709 |
+
"epoch": 0.9447334489351296,
|
| 74710 |
+
"grad_norm": 0.05797286331653595,
|
| 74711 |
+
"learning_rate": 8.218956784412479e-06,
|
| 74712 |
+
"loss": 1.448,
|
| 74713 |
+
"step": 21248
|
| 74714 |
+
},
|
| 74715 |
+
{
|
| 74716 |
+
"epoch": 0.9448223733938019,
|
| 74717 |
+
"grad_norm": 0.05742442235350609,
|
| 74718 |
+
"learning_rate": 8.192580492759783e-06,
|
| 74719 |
+
"loss": 1.4453,
|
| 74720 |
+
"step": 21250
|
| 74721 |
+
},
|
| 74722 |
+
{
|
| 74723 |
+
"epoch": 0.9449112978524743,
|
| 74724 |
+
"grad_norm": 0.05741652473807335,
|
| 74725 |
+
"learning_rate": 8.16624624297263e-06,
|
| 74726 |
+
"loss": 1.4499,
|
| 74727 |
+
"step": 21252
|
| 74728 |
+
},
|
| 74729 |
+
{
|
| 74730 |
+
"epoch": 0.9450002223111467,
|
| 74731 |
+
"grad_norm": 0.058388851583004,
|
| 74732 |
+
"learning_rate": 8.139954037302222e-06,
|
| 74733 |
+
"loss": 1.4424,
|
| 74734 |
+
"step": 21254
|
| 74735 |
+
},
|
| 74736 |
+
{
|
| 74737 |
+
"epoch": 0.945089146769819,
|
| 74738 |
+
"grad_norm": 0.058313120156526566,
|
| 74739 |
+
"learning_rate": 8.113703877996092e-06,
|
| 74740 |
+
"loss": 1.4426,
|
| 74741 |
+
"step": 21256
|
| 74742 |
+
},
|
| 74743 |
+
{
|
| 74744 |
+
"epoch": 0.9451780712284914,
|
| 74745 |
+
"grad_norm": 0.05811810865998268,
|
| 74746 |
+
"learning_rate": 8.087495767298226e-06,
|
| 74747 |
+
"loss": 1.441,
|
| 74748 |
+
"step": 21258
|
| 74749 |
+
},
|
| 74750 |
+
{
|
| 74751 |
+
"epoch": 0.9452669956871638,
|
| 74752 |
+
"grad_norm": 0.057785093784332275,
|
| 74753 |
+
"learning_rate": 8.061329707448939e-06,
|
| 74754 |
+
"loss": 1.4496,
|
| 74755 |
+
"step": 21260
|
| 74756 |
+
},
|
| 74757 |
+
{
|
| 74758 |
+
"epoch": 0.9453559201458361,
|
| 74759 |
+
"grad_norm": 0.05778678134083748,
|
| 74760 |
+
"learning_rate": 8.035205700685167e-06,
|
| 74761 |
+
"loss": 1.4433,
|
| 74762 |
+
"step": 21262
|
| 74763 |
+
},
|
| 74764 |
+
{
|
| 74765 |
+
"epoch": 0.9454448446045085,
|
| 74766 |
+
"grad_norm": 0.057463571429252625,
|
| 74767 |
+
"learning_rate": 8.00912374924001e-06,
|
| 74768 |
+
"loss": 1.4445,
|
| 74769 |
+
"step": 21264
|
| 74770 |
+
},
|
| 74771 |
+
{
|
| 74772 |
+
"epoch": 0.9455337690631809,
|
| 74773 |
+
"grad_norm": 0.058068107813596725,
|
| 74774 |
+
"learning_rate": 7.983083855343132e-06,
|
| 74775 |
+
"loss": 1.4452,
|
| 74776 |
+
"step": 21266
|
| 74777 |
+
},
|
| 74778 |
+
{
|
| 74779 |
+
"epoch": 0.9456226935218531,
|
| 74780 |
+
"grad_norm": 0.05680304020643234,
|
| 74781 |
+
"learning_rate": 7.957086021220417e-06,
|
| 74782 |
+
"loss": 1.4465,
|
| 74783 |
+
"step": 21268
|
| 74784 |
+
},
|
| 74785 |
+
{
|
| 74786 |
+
"epoch": 0.9457116179805255,
|
| 74787 |
+
"grad_norm": 0.057669758796691895,
|
| 74788 |
+
"learning_rate": 7.931130249094365e-06,
|
| 74789 |
+
"loss": 1.4493,
|
| 74790 |
+
"step": 21270
|
| 74791 |
+
},
|
| 74792 |
+
{
|
| 74793 |
+
"epoch": 0.9458005424391979,
|
| 74794 |
+
"grad_norm": 0.056808967143297195,
|
| 74795 |
+
"learning_rate": 7.905216541183869e-06,
|
| 74796 |
+
"loss": 1.4448,
|
| 74797 |
+
"step": 21272
|
| 74798 |
+
},
|
| 74799 |
+
{
|
| 74800 |
+
"epoch": 0.9458894668978702,
|
| 74801 |
+
"grad_norm": 0.05728383734822273,
|
| 74802 |
+
"learning_rate": 7.879344899703932e-06,
|
| 74803 |
+
"loss": 1.4457,
|
| 74804 |
+
"step": 21274
|
| 74805 |
+
},
|
| 74806 |
+
{
|
| 74807 |
+
"epoch": 0.9459783913565426,
|
| 74808 |
+
"grad_norm": 0.05765995383262634,
|
| 74809 |
+
"learning_rate": 7.853515326866345e-06,
|
| 74810 |
+
"loss": 1.4424,
|
| 74811 |
+
"step": 21276
|
| 74812 |
+
},
|
| 74813 |
+
{
|
| 74814 |
+
"epoch": 0.946067315815215,
|
| 74815 |
+
"grad_norm": 0.0581035315990448,
|
| 74816 |
+
"learning_rate": 7.827727824879116e-06,
|
| 74817 |
+
"loss": 1.443,
|
| 74818 |
+
"step": 21278
|
| 74819 |
+
},
|
| 74820 |
+
{
|
| 74821 |
+
"epoch": 0.9461562402738873,
|
| 74822 |
+
"grad_norm": 0.05767706036567688,
|
| 74823 |
+
"learning_rate": 7.801982395946649e-06,
|
| 74824 |
+
"loss": 1.4423,
|
| 74825 |
+
"step": 21280
|
| 74826 |
+
},
|
| 74827 |
+
{
|
| 74828 |
+
"epoch": 0.9462451647325597,
|
| 74829 |
+
"grad_norm": 0.057066384702920914,
|
| 74830 |
+
"learning_rate": 7.776279042269685e-06,
|
| 74831 |
+
"loss": 1.4482,
|
| 74832 |
+
"step": 21282
|
| 74833 |
+
},
|
| 74834 |
+
{
|
| 74835 |
+
"epoch": 0.9463340891912321,
|
| 74836 |
+
"grad_norm": 0.058075472712516785,
|
| 74837 |
+
"learning_rate": 7.750617766045688e-06,
|
| 74838 |
+
"loss": 1.4452,
|
| 74839 |
+
"step": 21284
|
| 74840 |
+
},
|
| 74841 |
+
{
|
| 74842 |
+
"epoch": 0.9464230136499044,
|
| 74843 |
+
"grad_norm": 0.05739322304725647,
|
| 74844 |
+
"learning_rate": 7.724998569468066e-06,
|
| 74845 |
+
"loss": 1.4477,
|
| 74846 |
+
"step": 21286
|
| 74847 |
+
},
|
| 74848 |
+
{
|
| 74849 |
+
"epoch": 0.9465119381085768,
|
| 74850 |
+
"grad_norm": 0.05727776512503624,
|
| 74851 |
+
"learning_rate": 7.699421454726963e-06,
|
| 74852 |
+
"loss": 1.447,
|
| 74853 |
+
"step": 21288
|
| 74854 |
+
},
|
| 74855 |
+
{
|
| 74856 |
+
"epoch": 0.9466008625672491,
|
| 74857 |
+
"grad_norm": 0.057564686983823776,
|
| 74858 |
+
"learning_rate": 7.67388642400879e-06,
|
| 74859 |
+
"loss": 1.4455,
|
| 74860 |
+
"step": 21290
|
| 74861 |
+
},
|
| 74862 |
+
{
|
| 74863 |
+
"epoch": 0.9466897870259214,
|
| 74864 |
+
"grad_norm": 0.05791555717587471,
|
| 74865 |
+
"learning_rate": 7.648393479496419e-06,
|
| 74866 |
+
"loss": 1.4429,
|
| 74867 |
+
"step": 21292
|
| 74868 |
+
},
|
| 74869 |
+
{
|
| 74870 |
+
"epoch": 0.9467787114845938,
|
| 74871 |
+
"grad_norm": 0.057518370449543,
|
| 74872 |
+
"learning_rate": 7.622942623369156e-06,
|
| 74873 |
+
"loss": 1.4435,
|
| 74874 |
+
"step": 21294
|
| 74875 |
+
},
|
| 74876 |
+
{
|
| 74877 |
+
"epoch": 0.9468676359432662,
|
| 74878 |
+
"grad_norm": 0.05689391866326332,
|
| 74879 |
+
"learning_rate": 7.597533857802541e-06,
|
| 74880 |
+
"loss": 1.4461,
|
| 74881 |
+
"step": 21296
|
| 74882 |
+
},
|
| 74883 |
+
{
|
| 74884 |
+
"epoch": 0.9469565604019385,
|
| 74885 |
+
"grad_norm": 0.058041494339704514,
|
| 74886 |
+
"learning_rate": 7.57216718496867e-06,
|
| 74887 |
+
"loss": 1.4456,
|
| 74888 |
+
"step": 21298
|
| 74889 |
+
},
|
| 74890 |
+
{
|
| 74891 |
+
"epoch": 0.9470454848606109,
|
| 74892 |
+
"grad_norm": 0.057841625064611435,
|
| 74893 |
+
"learning_rate": 7.5468426070360865e-06,
|
| 74894 |
+
"loss": 1.4501,
|
| 74895 |
+
"step": 21300
|
| 74896 |
+
},
|
| 74897 |
+
{
|
| 74898 |
+
"epoch": 0.9471344093192833,
|
| 74899 |
+
"grad_norm": 0.057104241102933884,
|
| 74900 |
+
"learning_rate": 7.521560126169502e-06,
|
| 74901 |
+
"loss": 1.4433,
|
| 74902 |
+
"step": 21302
|
| 74903 |
+
},
|
| 74904 |
+
{
|
| 74905 |
+
"epoch": 0.9472233337779556,
|
| 74906 |
+
"grad_norm": 0.058944664895534515,
|
| 74907 |
+
"learning_rate": 7.4963197445302445e-06,
|
| 74908 |
+
"loss": 1.4469,
|
| 74909 |
+
"step": 21304
|
| 74910 |
+
},
|
| 74911 |
+
{
|
| 74912 |
+
"epoch": 0.947312258236628,
|
| 74913 |
+
"grad_norm": 0.05764506757259369,
|
| 74914 |
+
"learning_rate": 7.471121464275976e-06,
|
| 74915 |
+
"loss": 1.4485,
|
| 74916 |
+
"step": 21306
|
| 74917 |
+
},
|
| 74918 |
+
{
|
| 74919 |
+
"epoch": 0.9474011826953004,
|
| 74920 |
+
"grad_norm": 0.05761958658695221,
|
| 74921 |
+
"learning_rate": 7.445965287560752e-06,
|
| 74922 |
+
"loss": 1.4491,
|
| 74923 |
+
"step": 21308
|
| 74924 |
+
},
|
| 74925 |
+
{
|
| 74926 |
+
"epoch": 0.9474901071539727,
|
| 74927 |
+
"grad_norm": 0.0575152151286602,
|
| 74928 |
+
"learning_rate": 7.420851216535019e-06,
|
| 74929 |
+
"loss": 1.4428,
|
| 74930 |
+
"step": 21310
|
| 74931 |
+
},
|
| 74932 |
+
{
|
| 74933 |
+
"epoch": 0.947579031612645,
|
| 74934 |
+
"grad_norm": 0.058228038251399994,
|
| 74935 |
+
"learning_rate": 7.3957792533456695e-06,
|
| 74936 |
+
"loss": 1.4463,
|
| 74937 |
+
"step": 21312
|
| 74938 |
+
},
|
| 74939 |
+
{
|
| 74940 |
+
"epoch": 0.9476679560713174,
|
| 74941 |
+
"grad_norm": 0.057448986917734146,
|
| 74942 |
+
"learning_rate": 7.3707494001358785e-06,
|
| 74943 |
+
"loss": 1.4482,
|
| 74944 |
+
"step": 21314
|
| 74945 |
+
},
|
| 74946 |
+
{
|
| 74947 |
+
"epoch": 0.9477568805299897,
|
| 74948 |
+
"grad_norm": 0.05910767614841461,
|
| 74949 |
+
"learning_rate": 7.3457616590454916e-06,
|
| 74950 |
+
"loss": 1.445,
|
| 74951 |
+
"step": 21316
|
| 74952 |
+
},
|
| 74953 |
+
{
|
| 74954 |
+
"epoch": 0.9478458049886621,
|
| 74955 |
+
"grad_norm": 0.057866428047418594,
|
| 74956 |
+
"learning_rate": 7.320816032210353e-06,
|
| 74957 |
+
"loss": 1.4464,
|
| 74958 |
+
"step": 21318
|
| 74959 |
+
},
|
| 74960 |
+
{
|
| 74961 |
+
"epoch": 0.9479347294473345,
|
| 74962 |
+
"grad_norm": 0.05850965157151222,
|
| 74963 |
+
"learning_rate": 7.295912521763038e-06,
|
| 74964 |
+
"loss": 1.4437,
|
| 74965 |
+
"step": 21320
|
| 74966 |
+
},
|
| 74967 |
+
{
|
| 74968 |
+
"epoch": 0.9480236539060068,
|
| 74969 |
+
"grad_norm": 0.057498205453157425,
|
| 74970 |
+
"learning_rate": 7.271051129832451e-06,
|
| 74971 |
+
"loss": 1.449,
|
| 74972 |
+
"step": 21322
|
| 74973 |
+
},
|
| 74974 |
+
{
|
| 74975 |
+
"epoch": 0.9481125783646792,
|
| 74976 |
+
"grad_norm": 0.057792823761701584,
|
| 74977 |
+
"learning_rate": 7.246231858543784e-06,
|
| 74978 |
+
"loss": 1.4393,
|
| 74979 |
+
"step": 21324
|
| 74980 |
+
},
|
| 74981 |
+
{
|
| 74982 |
+
"epoch": 0.9482015028233516,
|
| 74983 |
+
"grad_norm": 0.058098308742046356,
|
| 74984 |
+
"learning_rate": 7.221454710018671e-06,
|
| 74985 |
+
"loss": 1.4419,
|
| 74986 |
+
"step": 21326
|
| 74987 |
+
},
|
| 74988 |
+
{
|
| 74989 |
+
"epoch": 0.948290427282024,
|
| 74990 |
+
"grad_norm": 0.058741819113492966,
|
| 74991 |
+
"learning_rate": 7.196719686375308e-06,
|
| 74992 |
+
"loss": 1.4463,
|
| 74993 |
+
"step": 21328
|
| 74994 |
+
},
|
| 74995 |
+
{
|
| 74996 |
+
"epoch": 0.9483793517406963,
|
| 74997 |
+
"grad_norm": 0.0579305998980999,
|
| 74998 |
+
"learning_rate": 7.1720267897279475e-06,
|
| 74999 |
+
"loss": 1.4438,
|
| 75000 |
+
"step": 21330
|
| 75001 |
+
},
|
| 75002 |
+
{
|
| 75003 |
+
"epoch": 0.9484682761993687,
|
| 75004 |
+
"grad_norm": 0.05758350342512131,
|
| 75005 |
+
"learning_rate": 7.147376022187624e-06,
|
| 75006 |
+
"loss": 1.4539,
|
| 75007 |
+
"step": 21332
|
| 75008 |
+
},
|
| 75009 |
+
{
|
| 75010 |
+
"epoch": 0.9485572006580409,
|
| 75011 |
+
"grad_norm": 0.05874762684106827,
|
| 75012 |
+
"learning_rate": 7.122767385861484e-06,
|
| 75013 |
+
"loss": 1.4462,
|
| 75014 |
+
"step": 21334
|
| 75015 |
+
},
|
| 75016 |
+
{
|
| 75017 |
+
"epoch": 0.9486461251167133,
|
| 75018 |
+
"grad_norm": 0.05720106512308121,
|
| 75019 |
+
"learning_rate": 7.09820088285329e-06,
|
| 75020 |
+
"loss": 1.4421,
|
| 75021 |
+
"step": 21336
|
| 75022 |
+
},
|
| 75023 |
+
{
|
| 75024 |
+
"epoch": 0.9487350495753857,
|
| 75025 |
+
"grad_norm": 0.057676248252391815,
|
| 75026 |
+
"learning_rate": 7.073676515263028e-06,
|
| 75027 |
+
"loss": 1.4495,
|
| 75028 |
+
"step": 21338
|
| 75029 |
+
},
|
| 75030 |
+
{
|
| 75031 |
+
"epoch": 0.948823974034058,
|
| 75032 |
+
"grad_norm": 0.05915232002735138,
|
| 75033 |
+
"learning_rate": 7.049194285187077e-06,
|
| 75034 |
+
"loss": 1.4458,
|
| 75035 |
+
"step": 21340
|
| 75036 |
+
},
|
| 75037 |
+
{
|
| 75038 |
+
"epoch": 0.9489128984927304,
|
| 75039 |
+
"grad_norm": 0.05813376605510712,
|
| 75040 |
+
"learning_rate": 7.02475419471843e-06,
|
| 75041 |
+
"loss": 1.4398,
|
| 75042 |
+
"step": 21342
|
| 75043 |
+
},
|
| 75044 |
+
{
|
| 75045 |
+
"epoch": 0.9490018229514028,
|
| 75046 |
+
"grad_norm": 0.057365693151950836,
|
| 75047 |
+
"learning_rate": 7.000356245946249e-06,
|
| 75048 |
+
"loss": 1.4464,
|
| 75049 |
+
"step": 21344
|
| 75050 |
+
},
|
| 75051 |
+
{
|
| 75052 |
+
"epoch": 0.9490907474100752,
|
| 75053 |
+
"grad_norm": 0.05754502862691879,
|
| 75054 |
+
"learning_rate": 6.976000440956198e-06,
|
| 75055 |
+
"loss": 1.4493,
|
| 75056 |
+
"step": 21346
|
| 75057 |
+
},
|
| 75058 |
+
{
|
| 75059 |
+
"epoch": 0.9491796718687475,
|
| 75060 |
+
"grad_norm": 0.058054231107234955,
|
| 75061 |
+
"learning_rate": 6.9516867818302796e-06,
|
| 75062 |
+
"loss": 1.4469,
|
| 75063 |
+
"step": 21348
|
| 75064 |
+
},
|
| 75065 |
+
{
|
| 75066 |
+
"epoch": 0.9492685963274199,
|
| 75067 |
+
"grad_norm": 0.05699177458882332,
|
| 75068 |
+
"learning_rate": 6.927415270647053e-06,
|
| 75069 |
+
"loss": 1.4509,
|
| 75070 |
+
"step": 21350
|
| 75071 |
+
},
|
| 75072 |
+
{
|
| 75073 |
+
"epoch": 0.9493575207860923,
|
| 75074 |
+
"grad_norm": 0.058153219521045685,
|
| 75075 |
+
"learning_rate": 6.903185909481191e-06,
|
| 75076 |
+
"loss": 1.4432,
|
| 75077 |
+
"step": 21352
|
| 75078 |
+
},
|
| 75079 |
+
{
|
| 75080 |
+
"epoch": 0.9494464452447646,
|
| 75081 |
+
"grad_norm": 0.05815225467085838,
|
| 75082 |
+
"learning_rate": 6.878998700403982e-06,
|
| 75083 |
+
"loss": 1.439,
|
| 75084 |
+
"step": 21354
|
| 75085 |
+
},
|
| 75086 |
+
{
|
| 75087 |
+
"epoch": 0.9495353697034369,
|
| 75088 |
+
"grad_norm": 0.05740995332598686,
|
| 75089 |
+
"learning_rate": 6.854853645483106e-06,
|
| 75090 |
+
"loss": 1.4472,
|
| 75091 |
+
"step": 21356
|
| 75092 |
+
},
|
| 75093 |
+
{
|
| 75094 |
+
"epoch": 0.9496242941621093,
|
| 75095 |
+
"grad_norm": 0.05770250782370567,
|
| 75096 |
+
"learning_rate": 6.830750746782633e-06,
|
| 75097 |
+
"loss": 1.4437,
|
| 75098 |
+
"step": 21358
|
| 75099 |
+
},
|
| 75100 |
+
{
|
| 75101 |
+
"epoch": 0.9497132186207816,
|
| 75102 |
+
"grad_norm": 0.05776236578822136,
|
| 75103 |
+
"learning_rate": 6.806690006362859e-06,
|
| 75104 |
+
"loss": 1.4447,
|
| 75105 |
+
"step": 21360
|
| 75106 |
+
},
|
| 75107 |
+
{
|
| 75108 |
+
"epoch": 0.949802143079454,
|
| 75109 |
+
"grad_norm": 0.057613179087638855,
|
| 75110 |
+
"learning_rate": 6.7826714262806396e-06,
|
| 75111 |
+
"loss": 1.4448,
|
| 75112 |
+
"step": 21362
|
| 75113 |
+
},
|
| 75114 |
+
{
|
| 75115 |
+
"epoch": 0.9498910675381264,
|
| 75116 |
+
"grad_norm": 0.0582854337990284,
|
| 75117 |
+
"learning_rate": 6.758695008589221e-06,
|
| 75118 |
+
"loss": 1.4439,
|
| 75119 |
+
"step": 21364
|
| 75120 |
+
},
|
| 75121 |
+
{
|
| 75122 |
+
"epoch": 0.9499799919967987,
|
| 75123 |
+
"grad_norm": 0.057772018015384674,
|
| 75124 |
+
"learning_rate": 6.734760755338243e-06,
|
| 75125 |
+
"loss": 1.4434,
|
| 75126 |
+
"step": 21366
|
| 75127 |
+
},
|
| 75128 |
+
{
|
| 75129 |
+
"epoch": 0.9500689164554711,
|
| 75130 |
+
"grad_norm": 0.057995546609163284,
|
| 75131 |
+
"learning_rate": 6.7108686685735665e-06,
|
| 75132 |
+
"loss": 1.444,
|
| 75133 |
+
"step": 21368
|
| 75134 |
+
},
|
| 75135 |
+
{
|
| 75136 |
+
"epoch": 0.9501578409141435,
|
| 75137 |
+
"grad_norm": 0.05806703120470047,
|
| 75138 |
+
"learning_rate": 6.687018750337726e-06,
|
| 75139 |
+
"loss": 1.4412,
|
| 75140 |
+
"step": 21370
|
| 75141 |
+
},
|
| 75142 |
+
{
|
| 75143 |
+
"epoch": 0.9502467653728158,
|
| 75144 |
+
"grad_norm": 0.056987687945365906,
|
| 75145 |
+
"learning_rate": 6.663211002669534e-06,
|
| 75146 |
+
"loss": 1.4426,
|
| 75147 |
+
"step": 21372
|
| 75148 |
+
},
|
| 75149 |
+
{
|
| 75150 |
+
"epoch": 0.9503356898314882,
|
| 75151 |
+
"grad_norm": 0.05830015987157822,
|
| 75152 |
+
"learning_rate": 6.639445427604085e-06,
|
| 75153 |
+
"loss": 1.4377,
|
| 75154 |
+
"step": 21374
|
| 75155 |
+
},
|
| 75156 |
+
{
|
| 75157 |
+
"epoch": 0.9504246142901606,
|
| 75158 |
+
"grad_norm": 0.057094842195510864,
|
| 75159 |
+
"learning_rate": 6.615722027173032e-06,
|
| 75160 |
+
"loss": 1.4469,
|
| 75161 |
+
"step": 21376
|
| 75162 |
+
},
|
| 75163 |
+
{
|
| 75164 |
+
"epoch": 0.9505135387488328,
|
| 75165 |
+
"grad_norm": 0.057993218302726746,
|
| 75166 |
+
"learning_rate": 6.592040803404309e-06,
|
| 75167 |
+
"loss": 1.4415,
|
| 75168 |
+
"step": 21378
|
| 75169 |
+
},
|
| 75170 |
+
{
|
| 75171 |
+
"epoch": 0.9506024632075052,
|
| 75172 |
+
"grad_norm": 0.05756436288356781,
|
| 75173 |
+
"learning_rate": 6.5684017583223506e-06,
|
| 75174 |
+
"loss": 1.4454,
|
| 75175 |
+
"step": 21380
|
| 75176 |
+
},
|
| 75177 |
+
{
|
| 75178 |
+
"epoch": 0.9506913876661776,
|
| 75179 |
+
"grad_norm": 0.05718113109469414,
|
| 75180 |
+
"learning_rate": 6.544804893947876e-06,
|
| 75181 |
+
"loss": 1.4446,
|
| 75182 |
+
"step": 21382
|
| 75183 |
+
},
|
| 75184 |
+
{
|
| 75185 |
+
"epoch": 0.9507803121248499,
|
| 75186 |
+
"grad_norm": 0.05671248957514763,
|
| 75187 |
+
"learning_rate": 6.521250212298046e-06,
|
| 75188 |
+
"loss": 1.4501,
|
| 75189 |
+
"step": 21384
|
| 75190 |
+
},
|
| 75191 |
+
{
|
| 75192 |
+
"epoch": 0.9508692365835223,
|
| 75193 |
+
"grad_norm": 0.05659397318959236,
|
| 75194 |
+
"learning_rate": 6.497737715386476e-06,
|
| 75195 |
+
"loss": 1.4458,
|
| 75196 |
+
"step": 21386
|
| 75197 |
+
},
|
| 75198 |
+
{
|
| 75199 |
+
"epoch": 0.9509581610421947,
|
| 75200 |
+
"grad_norm": 0.05791900306940079,
|
| 75201 |
+
"learning_rate": 6.474267405223e-06,
|
| 75202 |
+
"loss": 1.4472,
|
| 75203 |
+
"step": 21388
|
| 75204 |
+
},
|
| 75205 |
+
{
|
| 75206 |
+
"epoch": 0.951047085500867,
|
| 75207 |
+
"grad_norm": 0.05804454907774925,
|
| 75208 |
+
"learning_rate": 6.450839283814125e-06,
|
| 75209 |
+
"loss": 1.4465,
|
| 75210 |
+
"step": 21390
|
| 75211 |
+
},
|
| 75212 |
+
{
|
| 75213 |
+
"epoch": 0.9511360099595394,
|
| 75214 |
+
"grad_norm": 0.05747028812766075,
|
| 75215 |
+
"learning_rate": 6.4274533531624715e-06,
|
| 75216 |
+
"loss": 1.4468,
|
| 75217 |
+
"step": 21392
|
| 75218 |
+
},
|
| 75219 |
+
{
|
| 75220 |
+
"epoch": 0.9512249344182118,
|
| 75221 |
+
"grad_norm": 0.058244530111551285,
|
| 75222 |
+
"learning_rate": 6.404109615267218e-06,
|
| 75223 |
+
"loss": 1.4506,
|
| 75224 |
+
"step": 21394
|
| 75225 |
+
},
|
| 75226 |
+
{
|
| 75227 |
+
"epoch": 0.9513138588768841,
|
| 75228 |
+
"grad_norm": 0.057229701429605484,
|
| 75229 |
+
"learning_rate": 6.380808072123934e-06,
|
| 75230 |
+
"loss": 1.4488,
|
| 75231 |
+
"step": 21396
|
| 75232 |
+
},
|
| 75233 |
+
{
|
| 75234 |
+
"epoch": 0.9514027833355564,
|
| 75235 |
+
"grad_norm": 0.05814434215426445,
|
| 75236 |
+
"learning_rate": 6.357548725724416e-06,
|
| 75237 |
+
"loss": 1.4449,
|
| 75238 |
+
"step": 21398
|
| 75239 |
+
},
|
| 75240 |
+
{
|
| 75241 |
+
"epoch": 0.9514917077942288,
|
| 75242 |
+
"grad_norm": 0.057511039078235626,
|
| 75243 |
+
"learning_rate": 6.334331578057018e-06,
|
| 75244 |
+
"loss": 1.4408,
|
| 75245 |
+
"step": 21400
|
| 75246 |
+
},
|
| 75247 |
+
{
|
| 75248 |
+
"epoch": 0.9515806322529011,
|
| 75249 |
+
"grad_norm": 0.05873566493391991,
|
| 75250 |
+
"learning_rate": 6.31115663110654e-06,
|
| 75251 |
+
"loss": 1.449,
|
| 75252 |
+
"step": 21402
|
| 75253 |
+
},
|
| 75254 |
+
{
|
| 75255 |
+
"epoch": 0.9516695567115735,
|
| 75256 |
+
"grad_norm": 0.05769232288002968,
|
| 75257 |
+
"learning_rate": 6.288023886854011e-06,
|
| 75258 |
+
"loss": 1.4451,
|
| 75259 |
+
"step": 21404
|
| 75260 |
+
},
|
| 75261 |
+
{
|
| 75262 |
+
"epoch": 0.9517584811702459,
|
| 75263 |
+
"grad_norm": 0.05827988311648369,
|
| 75264 |
+
"learning_rate": 6.264933347276847e-06,
|
| 75265 |
+
"loss": 1.4442,
|
| 75266 |
+
"step": 21406
|
| 75267 |
+
},
|
| 75268 |
+
{
|
| 75269 |
+
"epoch": 0.9518474056289182,
|
| 75270 |
+
"grad_norm": 0.05725441128015518,
|
| 75271 |
+
"learning_rate": 6.2418850143490805e-06,
|
| 75272 |
+
"loss": 1.4455,
|
| 75273 |
+
"step": 21408
|
| 75274 |
+
},
|
| 75275 |
+
{
|
| 75276 |
+
"epoch": 0.9519363300875906,
|
| 75277 |
+
"grad_norm": 0.05901114642620087,
|
| 75278 |
+
"learning_rate": 6.218878890040858e-06,
|
| 75279 |
+
"loss": 1.4456,
|
| 75280 |
+
"step": 21410
|
| 75281 |
+
},
|
| 75282 |
+
{
|
| 75283 |
+
"epoch": 0.952025254546263,
|
| 75284 |
+
"grad_norm": 0.05790838971734047,
|
| 75285 |
+
"learning_rate": 6.195914976318884e-06,
|
| 75286 |
+
"loss": 1.4424,
|
| 75287 |
+
"step": 21412
|
| 75288 |
+
},
|
| 75289 |
+
{
|
| 75290 |
+
"epoch": 0.9521141790049353,
|
| 75291 |
+
"grad_norm": 0.057518161833286285,
|
| 75292 |
+
"learning_rate": 6.17299327514631e-06,
|
| 75293 |
+
"loss": 1.45,
|
| 75294 |
+
"step": 21414
|
| 75295 |
+
},
|
| 75296 |
+
{
|
| 75297 |
+
"epoch": 0.9522031034636077,
|
| 75298 |
+
"grad_norm": 0.05830460786819458,
|
| 75299 |
+
"learning_rate": 6.150113788482403e-06,
|
| 75300 |
+
"loss": 1.451,
|
| 75301 |
+
"step": 21416
|
| 75302 |
+
},
|
| 75303 |
+
{
|
| 75304 |
+
"epoch": 0.9522920279222801,
|
| 75305 |
+
"grad_norm": 0.057020124047994614,
|
| 75306 |
+
"learning_rate": 6.127276518283153e-06,
|
| 75307 |
+
"loss": 1.4477,
|
| 75308 |
+
"step": 21418
|
| 75309 |
+
},
|
| 75310 |
+
{
|
| 75311 |
+
"epoch": 0.9523809523809523,
|
| 75312 |
+
"grad_norm": 0.05747748538851738,
|
| 75313 |
+
"learning_rate": 6.104481466500667e-06,
|
| 75314 |
+
"loss": 1.4502,
|
| 75315 |
+
"step": 21420
|
| 75316 |
+
},
|
| 75317 |
+
{
|
| 75318 |
+
"epoch": 0.9524698768396247,
|
| 75319 |
+
"grad_norm": 0.05771753937005997,
|
| 75320 |
+
"learning_rate": 6.081728635083661e-06,
|
| 75321 |
+
"loss": 1.4461,
|
| 75322 |
+
"step": 21422
|
| 75323 |
+
},
|
| 75324 |
+
{
|
| 75325 |
+
"epoch": 0.9525588012982971,
|
| 75326 |
+
"grad_norm": 0.057178083807229996,
|
| 75327 |
+
"learning_rate": 6.059018025977137e-06,
|
| 75328 |
+
"loss": 1.4498,
|
| 75329 |
+
"step": 21424
|
| 75330 |
+
},
|
| 75331 |
+
{
|
| 75332 |
+
"epoch": 0.9526477257569694,
|
| 75333 |
+
"grad_norm": 0.05818738043308258,
|
| 75334 |
+
"learning_rate": 6.036349641122429e-06,
|
| 75335 |
+
"loss": 1.4535,
|
| 75336 |
+
"step": 21426
|
| 75337 |
+
},
|
| 75338 |
+
{
|
| 75339 |
+
"epoch": 0.9527366502156418,
|
| 75340 |
+
"grad_norm": 0.057868827134370804,
|
| 75341 |
+
"learning_rate": 6.013723482457434e-06,
|
| 75342 |
+
"loss": 1.4438,
|
| 75343 |
+
"step": 21428
|
| 75344 |
+
},
|
| 75345 |
+
{
|
| 75346 |
+
"epoch": 0.9528255746743142,
|
| 75347 |
+
"grad_norm": 0.05839633569121361,
|
| 75348 |
+
"learning_rate": 5.9911395519162695e-06,
|
| 75349 |
+
"loss": 1.4444,
|
| 75350 |
+
"step": 21430
|
| 75351 |
+
},
|
| 75352 |
+
{
|
| 75353 |
+
"epoch": 0.9529144991329865,
|
| 75354 |
+
"grad_norm": 0.05750703439116478,
|
| 75355 |
+
"learning_rate": 5.968597851429502e-06,
|
| 75356 |
+
"loss": 1.4501,
|
| 75357 |
+
"step": 21432
|
| 75358 |
+
},
|
| 75359 |
+
{
|
| 75360 |
+
"epoch": 0.9530034235916589,
|
| 75361 |
+
"grad_norm": 0.05813831835985184,
|
| 75362 |
+
"learning_rate": 5.946098382924148e-06,
|
| 75363 |
+
"loss": 1.4435,
|
| 75364 |
+
"step": 21434
|
| 75365 |
+
},
|
| 75366 |
+
{
|
| 75367 |
+
"epoch": 0.9530923480503313,
|
| 75368 |
+
"grad_norm": 0.05906336382031441,
|
| 75369 |
+
"learning_rate": 5.9236411483235e-06,
|
| 75370 |
+
"loss": 1.4442,
|
| 75371 |
+
"step": 21436
|
| 75372 |
+
},
|
| 75373 |
+
{
|
| 75374 |
+
"epoch": 0.9531812725090036,
|
| 75375 |
+
"grad_norm": 0.05761095881462097,
|
| 75376 |
+
"learning_rate": 5.901226149547356e-06,
|
| 75377 |
+
"loss": 1.4428,
|
| 75378 |
+
"step": 21438
|
| 75379 |
+
},
|
| 75380 |
+
{
|
| 75381 |
+
"epoch": 0.953270196967676,
|
| 75382 |
+
"grad_norm": 0.058021754026412964,
|
| 75383 |
+
"learning_rate": 5.8788533885117956e-06,
|
| 75384 |
+
"loss": 1.4458,
|
| 75385 |
+
"step": 21440
|
| 75386 |
+
},
|
| 75387 |
+
{
|
| 75388 |
+
"epoch": 0.9533591214263483,
|
| 75389 |
+
"grad_norm": 0.057688046246767044,
|
| 75390 |
+
"learning_rate": 5.856522867129343e-06,
|
| 75391 |
+
"loss": 1.4438,
|
| 75392 |
+
"step": 21442
|
| 75393 |
+
},
|
| 75394 |
+
{
|
| 75395 |
+
"epoch": 0.9534480458850206,
|
| 75396 |
+
"grad_norm": 0.05762839689850807,
|
| 75397 |
+
"learning_rate": 5.834234587309028e-06,
|
| 75398 |
+
"loss": 1.4459,
|
| 75399 |
+
"step": 21444
|
| 75400 |
+
},
|
| 75401 |
+
{
|
| 75402 |
+
"epoch": 0.953536970343693,
|
| 75403 |
+
"grad_norm": 0.05795801803469658,
|
| 75404 |
+
"learning_rate": 5.811988550955993e-06,
|
| 75405 |
+
"loss": 1.4487,
|
| 75406 |
+
"step": 21446
|
| 75407 |
+
},
|
| 75408 |
+
{
|
| 75409 |
+
"epoch": 0.9536258948023654,
|
| 75410 |
+
"grad_norm": 0.0580856017768383,
|
| 75411 |
+
"learning_rate": 5.789784759971994e-06,
|
| 75412 |
+
"loss": 1.4503,
|
| 75413 |
+
"step": 21448
|
| 75414 |
+
},
|
| 75415 |
+
{
|
| 75416 |
+
"epoch": 0.9537148192610377,
|
| 75417 |
+
"grad_norm": 0.05720821022987366,
|
| 75418 |
+
"learning_rate": 5.767623216255125e-06,
|
| 75419 |
+
"loss": 1.4477,
|
| 75420 |
+
"step": 21450
|
| 75421 |
+
},
|
| 75422 |
+
{
|
| 75423 |
+
"epoch": 0.9538037437197101,
|
| 75424 |
+
"grad_norm": 0.057731300592422485,
|
| 75425 |
+
"learning_rate": 5.745503921699868e-06,
|
| 75426 |
+
"loss": 1.4478,
|
| 75427 |
+
"step": 21452
|
| 75428 |
+
},
|
| 75429 |
+
{
|
| 75430 |
+
"epoch": 0.9538926681783825,
|
| 75431 |
+
"grad_norm": 0.05763096734881401,
|
| 75432 |
+
"learning_rate": 5.7234268781969915e-06,
|
| 75433 |
+
"loss": 1.4447,
|
| 75434 |
+
"step": 21454
|
| 75435 |
+
},
|
| 75436 |
+
{
|
| 75437 |
+
"epoch": 0.9539815926370548,
|
| 75438 |
+
"grad_norm": 0.057338956743478775,
|
| 75439 |
+
"learning_rate": 5.701392087633761e-06,
|
| 75440 |
+
"loss": 1.445,
|
| 75441 |
+
"step": 21456
|
| 75442 |
+
},
|
| 75443 |
+
{
|
| 75444 |
+
"epoch": 0.9540705170957272,
|
| 75445 |
+
"grad_norm": 0.057891324162483215,
|
| 75446 |
+
"learning_rate": 5.679399551893893e-06,
|
| 75447 |
+
"loss": 1.4418,
|
| 75448 |
+
"step": 21458
|
| 75449 |
+
},
|
| 75450 |
+
{
|
| 75451 |
+
"epoch": 0.9541594415543996,
|
| 75452 |
+
"grad_norm": 0.05737922713160515,
|
| 75453 |
+
"learning_rate": 5.657449272857385e-06,
|
| 75454 |
+
"loss": 1.4444,
|
| 75455 |
+
"step": 21460
|
| 75456 |
+
},
|
| 75457 |
+
{
|
| 75458 |
+
"epoch": 0.954248366013072,
|
| 75459 |
+
"grad_norm": 0.05823858082294464,
|
| 75460 |
+
"learning_rate": 5.635541252400511e-06,
|
| 75461 |
+
"loss": 1.4367,
|
| 75462 |
+
"step": 21462
|
| 75463 |
+
},
|
| 75464 |
+
{
|
| 75465 |
+
"epoch": 0.9543372904717442,
|
| 75466 |
+
"grad_norm": 0.05750085040926933,
|
| 75467 |
+
"learning_rate": 5.61367549239622e-06,
|
| 75468 |
+
"loss": 1.4491,
|
| 75469 |
+
"step": 21464
|
| 75470 |
+
},
|
| 75471 |
+
{
|
| 75472 |
+
"epoch": 0.9544262149304166,
|
| 75473 |
+
"grad_norm": 0.05752246081829071,
|
| 75474 |
+
"learning_rate": 5.591851994713681e-06,
|
| 75475 |
+
"loss": 1.4429,
|
| 75476 |
+
"step": 21466
|
| 75477 |
+
},
|
| 75478 |
+
{
|
| 75479 |
+
"epoch": 0.9545151393890889,
|
| 75480 |
+
"grad_norm": 0.057769421488046646,
|
| 75481 |
+
"learning_rate": 5.5700707612184044e-06,
|
| 75482 |
+
"loss": 1.4498,
|
| 75483 |
+
"step": 21468
|
| 75484 |
+
},
|
| 75485 |
+
{
|
| 75486 |
+
"epoch": 0.9546040638477613,
|
| 75487 |
+
"grad_norm": 0.057797472923994064,
|
| 75488 |
+
"learning_rate": 5.548331793772288e-06,
|
| 75489 |
+
"loss": 1.4463,
|
| 75490 |
+
"step": 21470
|
| 75491 |
+
},
|
| 75492 |
+
{
|
| 75493 |
+
"epoch": 0.9546929883064337,
|
| 75494 |
+
"grad_norm": 0.057377446442842484,
|
| 75495 |
+
"learning_rate": 5.52663509423379e-06,
|
| 75496 |
+
"loss": 1.4473,
|
| 75497 |
+
"step": 21472
|
| 75498 |
+
},
|
| 75499 |
+
{
|
| 75500 |
+
"epoch": 0.954781912765106,
|
| 75501 |
+
"grad_norm": 0.05764816701412201,
|
| 75502 |
+
"learning_rate": 5.504980664457593e-06,
|
| 75503 |
+
"loss": 1.4454,
|
| 75504 |
+
"step": 21474
|
| 75505 |
+
},
|
| 75506 |
+
{
|
| 75507 |
+
"epoch": 0.9548708372237784,
|
| 75508 |
+
"grad_norm": 0.05828756093978882,
|
| 75509 |
+
"learning_rate": 5.483368506294828e-06,
|
| 75510 |
+
"loss": 1.4446,
|
| 75511 |
+
"step": 21476
|
| 75512 |
+
},
|
| 75513 |
+
{
|
| 75514 |
+
"epoch": 0.9549597616824508,
|
| 75515 |
+
"grad_norm": 0.058148372918367386,
|
| 75516 |
+
"learning_rate": 5.461798621593017e-06,
|
| 75517 |
+
"loss": 1.441,
|
| 75518 |
+
"step": 21478
|
| 75519 |
+
},
|
| 75520 |
+
{
|
| 75521 |
+
"epoch": 0.9550486861411231,
|
| 75522 |
+
"grad_norm": 0.05808459222316742,
|
| 75523 |
+
"learning_rate": 5.440271012195963e-06,
|
| 75524 |
+
"loss": 1.4473,
|
| 75525 |
+
"step": 21480
|
| 75526 |
+
},
|
| 75527 |
+
{
|
| 75528 |
+
"epoch": 0.9551376105997955,
|
| 75529 |
+
"grad_norm": 0.05690108239650726,
|
| 75530 |
+
"learning_rate": 5.4187856799440275e-06,
|
| 75531 |
+
"loss": 1.4451,
|
| 75532 |
+
"step": 21482
|
| 75533 |
+
},
|
| 75534 |
+
{
|
| 75535 |
+
"epoch": 0.9552265350584679,
|
| 75536 |
+
"grad_norm": 0.05813979357481003,
|
| 75537 |
+
"learning_rate": 5.3973426266737955e-06,
|
| 75538 |
+
"loss": 1.4465,
|
| 75539 |
+
"step": 21484
|
| 75540 |
+
},
|
| 75541 |
+
{
|
| 75542 |
+
"epoch": 0.9553154595171401,
|
| 75543 |
+
"grad_norm": 0.057522471994161606,
|
| 75544 |
+
"learning_rate": 5.3759418542184134e-06,
|
| 75545 |
+
"loss": 1.4487,
|
| 75546 |
+
"step": 21486
|
| 75547 |
+
},
|
| 75548 |
+
{
|
| 75549 |
+
"epoch": 0.9554043839758125,
|
| 75550 |
+
"grad_norm": 0.05860830470919609,
|
| 75551 |
+
"learning_rate": 5.354583364407251e-06,
|
| 75552 |
+
"loss": 1.4444,
|
| 75553 |
+
"step": 21488
|
| 75554 |
+
},
|
| 75555 |
+
{
|
| 75556 |
+
"epoch": 0.9554933084344849,
|
| 75557 |
+
"grad_norm": 0.05663066357374191,
|
| 75558 |
+
"learning_rate": 5.333267159066124e-06,
|
| 75559 |
+
"loss": 1.444,
|
| 75560 |
+
"step": 21490
|
| 75561 |
+
},
|
| 75562 |
+
{
|
| 75563 |
+
"epoch": 0.9555822328931572,
|
| 75564 |
+
"grad_norm": 0.058597080409526825,
|
| 75565 |
+
"learning_rate": 5.311993240017243e-06,
|
| 75566 |
+
"loss": 1.4472,
|
| 75567 |
+
"step": 21492
|
| 75568 |
+
},
|
| 75569 |
+
{
|
| 75570 |
+
"epoch": 0.9556711573518296,
|
| 75571 |
+
"grad_norm": 0.057328660041093826,
|
| 75572 |
+
"learning_rate": 5.290761609079209e-06,
|
| 75573 |
+
"loss": 1.447,
|
| 75574 |
+
"step": 21494
|
| 75575 |
+
},
|
| 75576 |
+
{
|
| 75577 |
+
"epoch": 0.955760081810502,
|
| 75578 |
+
"grad_norm": 0.0580228753387928,
|
| 75579 |
+
"learning_rate": 5.269572268066958e-06,
|
| 75580 |
+
"loss": 1.4468,
|
| 75581 |
+
"step": 21496
|
| 75582 |
+
},
|
| 75583 |
+
{
|
| 75584 |
+
"epoch": 0.9558490062691744,
|
| 75585 |
+
"grad_norm": 0.05755842849612236,
|
| 75586 |
+
"learning_rate": 5.248425218791874e-06,
|
| 75587 |
+
"loss": 1.4492,
|
| 75588 |
+
"step": 21498
|
| 75589 |
+
},
|
| 75590 |
+
{
|
| 75591 |
+
"epoch": 0.9559379307278467,
|
| 75592 |
+
"grad_norm": 0.05705071613192558,
|
| 75593 |
+
"learning_rate": 5.227320463061791e-06,
|
| 75594 |
+
"loss": 1.4469,
|
| 75595 |
+
"step": 21500
|
| 75596 |
+
},
|
| 75597 |
+
{
|
| 75598 |
+
"epoch": 0.9559379307278467,
|
| 75599 |
+
"eval_loss": 1.432187557220459,
|
| 75600 |
+
"eval_runtime": 12.4558,
|
| 75601 |
+
"eval_samples_per_second": 554.762,
|
| 75602 |
+
"eval_steps_per_second": 69.365,
|
| 75603 |
+
"step": 21500
|
| 75604 |
+
},
|
| 75605 |
+
{
|
| 75606 |
+
"epoch": 0.9560268551865191,
|
| 75607 |
+
"grad_norm": 0.057521793991327286,
|
| 75608 |
+
"learning_rate": 5.206258002680653e-06,
|
| 75609 |
+
"loss": 1.4425,
|
| 75610 |
+
"step": 21502
|
| 75611 |
+
},
|
| 75612 |
+
{
|
| 75613 |
+
"epoch": 0.9561157796451915,
|
| 75614 |
+
"grad_norm": 0.05715570226311684,
|
| 75615 |
+
"learning_rate": 5.18523783944913e-06,
|
| 75616 |
+
"loss": 1.4504,
|
| 75617 |
+
"step": 21504
|
| 75618 |
+
},
|
| 75619 |
+
{
|
| 75620 |
+
"epoch": 0.9562047041038638,
|
| 75621 |
+
"grad_norm": 0.05779632553458214,
|
| 75622 |
+
"learning_rate": 5.164259975164009e-06,
|
| 75623 |
+
"loss": 1.4429,
|
| 75624 |
+
"step": 21506
|
| 75625 |
+
},
|
| 75626 |
+
{
|
| 75627 |
+
"epoch": 0.9562936285625361,
|
| 75628 |
+
"grad_norm": 0.059060875326395035,
|
| 75629 |
+
"learning_rate": 5.143324411618577e-06,
|
| 75630 |
+
"loss": 1.4463,
|
| 75631 |
+
"step": 21508
|
| 75632 |
+
},
|
| 75633 |
+
{
|
| 75634 |
+
"epoch": 0.9563825530212084,
|
| 75635 |
+
"grad_norm": 0.05711085721850395,
|
| 75636 |
+
"learning_rate": 5.122431150602624e-06,
|
| 75637 |
+
"loss": 1.4455,
|
| 75638 |
+
"step": 21510
|
| 75639 |
+
},
|
| 75640 |
+
{
|
| 75641 |
+
"epoch": 0.9564714774798808,
|
| 75642 |
+
"grad_norm": 0.05797237157821655,
|
| 75643 |
+
"learning_rate": 5.101580193902055e-06,
|
| 75644 |
+
"loss": 1.4448,
|
| 75645 |
+
"step": 21512
|
| 75646 |
+
},
|
| 75647 |
+
{
|
| 75648 |
+
"epoch": 0.9565604019385532,
|
| 75649 |
+
"grad_norm": 0.057503245770931244,
|
| 75650 |
+
"learning_rate": 5.080771543299389e-06,
|
| 75651 |
+
"loss": 1.4502,
|
| 75652 |
+
"step": 21514
|
| 75653 |
+
},
|
| 75654 |
+
{
|
| 75655 |
+
"epoch": 0.9566493263972256,
|
| 75656 |
+
"grad_norm": 0.05919162929058075,
|
| 75657 |
+
"learning_rate": 5.060005200573425e-06,
|
| 75658 |
+
"loss": 1.4475,
|
| 75659 |
+
"step": 21516
|
| 75660 |
+
},
|
| 75661 |
+
{
|
| 75662 |
+
"epoch": 0.9567382508558979,
|
| 75663 |
+
"grad_norm": 0.05886862426996231,
|
| 75664 |
+
"learning_rate": 5.039281167499299e-06,
|
| 75665 |
+
"loss": 1.4526,
|
| 75666 |
+
"step": 21518
|
| 75667 |
+
},
|
| 75668 |
+
{
|
| 75669 |
+
"epoch": 0.9568271753145703,
|
| 75670 |
+
"grad_norm": 0.058441340923309326,
|
| 75671 |
+
"learning_rate": 5.018599445848648e-06,
|
| 75672 |
+
"loss": 1.4441,
|
| 75673 |
+
"step": 21520
|
| 75674 |
+
},
|
| 75675 |
+
{
|
| 75676 |
+
"epoch": 0.9569160997732427,
|
| 75677 |
+
"grad_norm": 0.05824016034603119,
|
| 75678 |
+
"learning_rate": 4.997960037389449e-06,
|
| 75679 |
+
"loss": 1.4449,
|
| 75680 |
+
"step": 21522
|
| 75681 |
+
},
|
| 75682 |
+
{
|
| 75683 |
+
"epoch": 0.957005024231915,
|
| 75684 |
+
"grad_norm": 0.057525213807821274,
|
| 75685 |
+
"learning_rate": 4.97736294388601e-06,
|
| 75686 |
+
"loss": 1.4477,
|
| 75687 |
+
"step": 21524
|
| 75688 |
+
},
|
| 75689 |
+
{
|
| 75690 |
+
"epoch": 0.9570939486905874,
|
| 75691 |
+
"grad_norm": 0.058176856487989426,
|
| 75692 |
+
"learning_rate": 4.956808167099092e-06,
|
| 75693 |
+
"loss": 1.4471,
|
| 75694 |
+
"step": 21526
|
| 75695 |
+
},
|
| 75696 |
+
{
|
| 75697 |
+
"epoch": 0.9571828731492596,
|
| 75698 |
+
"grad_norm": 0.057681165635585785,
|
| 75699 |
+
"learning_rate": 4.936295708785732e-06,
|
| 75700 |
+
"loss": 1.4497,
|
| 75701 |
+
"step": 21528
|
| 75702 |
+
},
|
| 75703 |
+
{
|
| 75704 |
+
"epoch": 0.957271797607932,
|
| 75705 |
+
"grad_norm": 0.05783999338746071,
|
| 75706 |
+
"learning_rate": 4.915825570699584e-06,
|
| 75707 |
+
"loss": 1.4408,
|
| 75708 |
+
"step": 21530
|
| 75709 |
+
},
|
| 75710 |
+
{
|
| 75711 |
+
"epoch": 0.9573607220666044,
|
| 75712 |
+
"grad_norm": 0.05777215585112572,
|
| 75713 |
+
"learning_rate": 4.895397754590414e-06,
|
| 75714 |
+
"loss": 1.4461,
|
| 75715 |
+
"step": 21532
|
| 75716 |
+
},
|
| 75717 |
+
{
|
| 75718 |
+
"epoch": 0.9574496465252768,
|
| 75719 |
+
"grad_norm": 0.058080073446035385,
|
| 75720 |
+
"learning_rate": 4.875012262204492e-06,
|
| 75721 |
+
"loss": 1.451,
|
| 75722 |
+
"step": 21534
|
| 75723 |
+
},
|
| 75724 |
+
{
|
| 75725 |
+
"epoch": 0.9575385709839491,
|
| 75726 |
+
"grad_norm": 0.05700400471687317,
|
| 75727 |
+
"learning_rate": 4.854669095284425e-06,
|
| 75728 |
+
"loss": 1.4511,
|
| 75729 |
+
"step": 21536
|
| 75730 |
+
},
|
| 75731 |
+
{
|
| 75732 |
+
"epoch": 0.9576274954426215,
|
| 75733 |
+
"grad_norm": 0.05668550357222557,
|
| 75734 |
+
"learning_rate": 4.834368255569322e-06,
|
| 75735 |
+
"loss": 1.4435,
|
| 75736 |
+
"step": 21538
|
| 75737 |
+
},
|
| 75738 |
+
{
|
| 75739 |
+
"epoch": 0.9577164199012939,
|
| 75740 |
+
"grad_norm": 0.05787520855665207,
|
| 75741 |
+
"learning_rate": 4.81410974479457e-06,
|
| 75742 |
+
"loss": 1.4419,
|
| 75743 |
+
"step": 21540
|
| 75744 |
+
},
|
| 75745 |
+
{
|
| 75746 |
+
"epoch": 0.9578053443599662,
|
| 75747 |
+
"grad_norm": 0.05773087963461876,
|
| 75748 |
+
"learning_rate": 4.793893564691843e-06,
|
| 75749 |
+
"loss": 1.4458,
|
| 75750 |
+
"step": 21542
|
| 75751 |
+
},
|
| 75752 |
+
{
|
| 75753 |
+
"epoch": 0.9578942688186386,
|
| 75754 |
+
"grad_norm": 0.057870473712682724,
|
| 75755 |
+
"learning_rate": 4.773719716989477e-06,
|
| 75756 |
+
"loss": 1.4454,
|
| 75757 |
+
"step": 21544
|
| 75758 |
+
},
|
| 75759 |
+
{
|
| 75760 |
+
"epoch": 0.957983193277311,
|
| 75761 |
+
"grad_norm": 0.05800451338291168,
|
| 75762 |
+
"learning_rate": 4.753588203411929e-06,
|
| 75763 |
+
"loss": 1.4423,
|
| 75764 |
+
"step": 21546
|
| 75765 |
+
},
|
| 75766 |
+
{
|
| 75767 |
+
"epoch": 0.9580721177359833,
|
| 75768 |
+
"grad_norm": 0.05686897784471512,
|
| 75769 |
+
"learning_rate": 4.7334990256800995e-06,
|
| 75770 |
+
"loss": 1.4462,
|
| 75771 |
+
"step": 21548
|
| 75772 |
+
},
|
| 75773 |
+
{
|
| 75774 |
+
"epoch": 0.9581610421946556,
|
| 75775 |
+
"grad_norm": 0.056949593126773834,
|
| 75776 |
+
"learning_rate": 4.7134521855113355e-06,
|
| 75777 |
+
"loss": 1.4418,
|
| 75778 |
+
"step": 21550
|
| 75779 |
+
},
|
| 75780 |
+
{
|
| 75781 |
+
"epoch": 0.958249966653328,
|
| 75782 |
+
"grad_norm": 0.05743638426065445,
|
| 75783 |
+
"learning_rate": 4.693447684619379e-06,
|
| 75784 |
+
"loss": 1.4471,
|
| 75785 |
+
"step": 21552
|
| 75786 |
+
},
|
| 75787 |
+
{
|
| 75788 |
+
"epoch": 0.9583388911120003,
|
| 75789 |
+
"grad_norm": 0.05902915075421333,
|
| 75790 |
+
"learning_rate": 4.673485524714305e-06,
|
| 75791 |
+
"loss": 1.4437,
|
| 75792 |
+
"step": 21554
|
| 75793 |
+
},
|
| 75794 |
+
{
|
| 75795 |
+
"epoch": 0.9584278155706727,
|
| 75796 |
+
"grad_norm": 0.058458685874938965,
|
| 75797 |
+
"learning_rate": 4.653565707502416e-06,
|
| 75798 |
+
"loss": 1.4422,
|
| 75799 |
+
"step": 21556
|
| 75800 |
+
},
|
| 75801 |
+
{
|
| 75802 |
+
"epoch": 0.9585167400293451,
|
| 75803 |
+
"grad_norm": 0.05795848369598389,
|
| 75804 |
+
"learning_rate": 4.633688234686739e-06,
|
| 75805 |
+
"loss": 1.4426,
|
| 75806 |
+
"step": 21558
|
| 75807 |
+
},
|
| 75808 |
+
{
|
| 75809 |
+
"epoch": 0.9586056644880174,
|
| 75810 |
+
"grad_norm": 0.058213118463754654,
|
| 75811 |
+
"learning_rate": 4.613853107966359e-06,
|
| 75812 |
+
"loss": 1.4477,
|
| 75813 |
+
"step": 21560
|
| 75814 |
+
},
|
| 75815 |
+
{
|
| 75816 |
+
"epoch": 0.9586945889466898,
|
| 75817 |
+
"grad_norm": 0.05738683417439461,
|
| 75818 |
+
"learning_rate": 4.59406032903692e-06,
|
| 75819 |
+
"loss": 1.4455,
|
| 75820 |
+
"step": 21562
|
| 75821 |
+
},
|
| 75822 |
+
{
|
| 75823 |
+
"epoch": 0.9587835134053622,
|
| 75824 |
+
"grad_norm": 0.057969436049461365,
|
| 75825 |
+
"learning_rate": 4.5743098995903455e-06,
|
| 75826 |
+
"loss": 1.4482,
|
| 75827 |
+
"step": 21564
|
| 75828 |
+
},
|
| 75829 |
+
{
|
| 75830 |
+
"epoch": 0.9588724378640345,
|
| 75831 |
+
"grad_norm": 0.05753438174724579,
|
| 75832 |
+
"learning_rate": 4.554601821315063e-06,
|
| 75833 |
+
"loss": 1.4418,
|
| 75834 |
+
"step": 21566
|
| 75835 |
+
},
|
| 75836 |
+
{
|
| 75837 |
+
"epoch": 0.9589613623227069,
|
| 75838 |
+
"grad_norm": 0.056868597865104675,
|
| 75839 |
+
"learning_rate": 4.534936095895781e-06,
|
| 75840 |
+
"loss": 1.449,
|
| 75841 |
+
"step": 21568
|
| 75842 |
+
},
|
| 75843 |
+
{
|
| 75844 |
+
"epoch": 0.9590502867813793,
|
| 75845 |
+
"grad_norm": 0.057507824152708054,
|
| 75846 |
+
"learning_rate": 4.515312725013598e-06,
|
| 75847 |
+
"loss": 1.4416,
|
| 75848 |
+
"step": 21570
|
| 75849 |
+
},
|
| 75850 |
+
{
|
| 75851 |
+
"epoch": 0.9591392112400515,
|
| 75852 |
+
"grad_norm": 0.05689457431435585,
|
| 75853 |
+
"learning_rate": 4.495731710346007e-06,
|
| 75854 |
+
"loss": 1.4416,
|
| 75855 |
+
"step": 21572
|
| 75856 |
+
},
|
| 75857 |
+
{
|
| 75858 |
+
"epoch": 0.9592281356987239,
|
| 75859 |
+
"grad_norm": 0.0581645630300045,
|
| 75860 |
+
"learning_rate": 4.47619305356689e-06,
|
| 75861 |
+
"loss": 1.446,
|
| 75862 |
+
"step": 21574
|
| 75863 |
+
},
|
| 75864 |
+
{
|
| 75865 |
+
"epoch": 0.9593170601573963,
|
| 75866 |
+
"grad_norm": 0.058234803378582,
|
| 75867 |
+
"learning_rate": 4.4566967563465234e-06,
|
| 75868 |
+
"loss": 1.4436,
|
| 75869 |
+
"step": 21576
|
| 75870 |
+
},
|
| 75871 |
+
{
|
| 75872 |
+
"epoch": 0.9594059846160686,
|
| 75873 |
+
"grad_norm": 0.05846071243286133,
|
| 75874 |
+
"learning_rate": 4.4372428203514615e-06,
|
| 75875 |
+
"loss": 1.4399,
|
| 75876 |
+
"step": 21578
|
| 75877 |
+
},
|
| 75878 |
+
{
|
| 75879 |
+
"epoch": 0.959494909074741,
|
| 75880 |
+
"grad_norm": 0.057688742876052856,
|
| 75881 |
+
"learning_rate": 4.417831247244819e-06,
|
| 75882 |
+
"loss": 1.4489,
|
| 75883 |
+
"step": 21580
|
| 75884 |
+
},
|
| 75885 |
+
{
|
| 75886 |
+
"epoch": 0.9595838335334134,
|
| 75887 |
+
"grad_norm": 0.058130089193582535,
|
| 75888 |
+
"learning_rate": 4.398462038685824e-06,
|
| 75889 |
+
"loss": 1.4435,
|
| 75890 |
+
"step": 21582
|
| 75891 |
+
},
|
| 75892 |
+
{
|
| 75893 |
+
"epoch": 0.9596727579920857,
|
| 75894 |
+
"grad_norm": 0.05822238698601723,
|
| 75895 |
+
"learning_rate": 4.3791351963304304e-06,
|
| 75896 |
+
"loss": 1.4503,
|
| 75897 |
+
"step": 21584
|
| 75898 |
+
},
|
| 75899 |
+
{
|
| 75900 |
+
"epoch": 0.9597616824507581,
|
| 75901 |
+
"grad_norm": 0.05791193246841431,
|
| 75902 |
+
"learning_rate": 4.3598507218306495e-06,
|
| 75903 |
+
"loss": 1.4482,
|
| 75904 |
+
"step": 21586
|
| 75905 |
+
},
|
| 75906 |
+
{
|
| 75907 |
+
"epoch": 0.9598506069094305,
|
| 75908 |
+
"grad_norm": 0.05749112367630005,
|
| 75909 |
+
"learning_rate": 4.340608616835051e-06,
|
| 75910 |
+
"loss": 1.4458,
|
| 75911 |
+
"step": 21588
|
| 75912 |
+
},
|
| 75913 |
+
{
|
| 75914 |
+
"epoch": 0.9599395313681028,
|
| 75915 |
+
"grad_norm": 0.0575961209833622,
|
| 75916 |
+
"learning_rate": 4.321408882988542e-06,
|
| 75917 |
+
"loss": 1.4476,
|
| 75918 |
+
"step": 21590
|
| 75919 |
+
},
|
| 75920 |
+
{
|
| 75921 |
+
"epoch": 0.9600284558267752,
|
| 75922 |
+
"grad_norm": 0.05781961977481842,
|
| 75923 |
+
"learning_rate": 4.302251521932366e-06,
|
| 75924 |
+
"loss": 1.4448,
|
| 75925 |
+
"step": 21592
|
| 75926 |
+
},
|
| 75927 |
+
{
|
| 75928 |
+
"epoch": 0.9601173802854475,
|
| 75929 |
+
"grad_norm": 0.05863596498966217,
|
| 75930 |
+
"learning_rate": 4.283136535304155e-06,
|
| 75931 |
+
"loss": 1.4491,
|
| 75932 |
+
"step": 21594
|
| 75933 |
+
},
|
| 75934 |
+
{
|
| 75935 |
+
"epoch": 0.9602063047441198,
|
| 75936 |
+
"grad_norm": 0.057458702474832535,
|
| 75937 |
+
"learning_rate": 4.264063924738104e-06,
|
| 75938 |
+
"loss": 1.4479,
|
| 75939 |
+
"step": 21596
|
| 75940 |
+
},
|
| 75941 |
+
{
|
| 75942 |
+
"epoch": 0.9602952292027922,
|
| 75943 |
+
"grad_norm": 0.057083528488874435,
|
| 75944 |
+
"learning_rate": 4.2450336918644085e-06,
|
| 75945 |
+
"loss": 1.4482,
|
| 75946 |
+
"step": 21598
|
| 75947 |
+
},
|
| 75948 |
+
{
|
| 75949 |
+
"epoch": 0.9603841536614646,
|
| 75950 |
+
"grad_norm": 0.05720750615000725,
|
| 75951 |
+
"learning_rate": 4.226045838309989e-06,
|
| 75952 |
+
"loss": 1.4473,
|
| 75953 |
+
"step": 21600
|
| 75954 |
+
},
|
| 75955 |
+
{
|
| 75956 |
+
"epoch": 0.9604730781201369,
|
| 75957 |
+
"grad_norm": 0.05795248597860336,
|
| 75958 |
+
"learning_rate": 4.207100365697936e-06,
|
| 75959 |
+
"loss": 1.4519,
|
| 75960 |
+
"step": 21602
|
| 75961 |
+
},
|
| 75962 |
+
{
|
| 75963 |
+
"epoch": 0.9605620025788093,
|
| 75964 |
+
"grad_norm": 0.057493992149829865,
|
| 75965 |
+
"learning_rate": 4.188197275647898e-06,
|
| 75966 |
+
"loss": 1.4457,
|
| 75967 |
+
"step": 21604
|
| 75968 |
+
},
|
| 75969 |
+
{
|
| 75970 |
+
"epoch": 0.9606509270374817,
|
| 75971 |
+
"grad_norm": 0.0582110695540905,
|
| 75972 |
+
"learning_rate": 4.169336569775695e-06,
|
| 75973 |
+
"loss": 1.4448,
|
| 75974 |
+
"step": 21606
|
| 75975 |
+
},
|
| 75976 |
+
{
|
| 75977 |
+
"epoch": 0.960739851496154,
|
| 75978 |
+
"grad_norm": 0.05797712132334709,
|
| 75979 |
+
"learning_rate": 4.150518249693647e-06,
|
| 75980 |
+
"loss": 1.4529,
|
| 75981 |
+
"step": 21608
|
| 75982 |
+
},
|
| 75983 |
+
{
|
| 75984 |
+
"epoch": 0.9608287759548264,
|
| 75985 |
+
"grad_norm": 0.0583646297454834,
|
| 75986 |
+
"learning_rate": 4.1317423170104675e-06,
|
| 75987 |
+
"loss": 1.4422,
|
| 75988 |
+
"step": 21610
|
| 75989 |
+
},
|
| 75990 |
+
{
|
| 75991 |
+
"epoch": 0.9609177004134988,
|
| 75992 |
+
"grad_norm": 0.05701034516096115,
|
| 75993 |
+
"learning_rate": 4.113008773331151e-06,
|
| 75994 |
+
"loss": 1.4434,
|
| 75995 |
+
"step": 21612
|
| 75996 |
+
},
|
| 75997 |
+
{
|
| 75998 |
+
"epoch": 0.9610066248721711,
|
| 75999 |
+
"grad_norm": 0.05788358673453331,
|
| 76000 |
+
"learning_rate": 4.094317620257138e-06,
|
| 76001 |
+
"loss": 1.4412,
|
| 76002 |
+
"step": 21614
|
| 76003 |
+
},
|
| 76004 |
+
{
|
| 76005 |
+
"epoch": 0.9610955493308434,
|
| 76006 |
+
"grad_norm": 0.05787377804517746,
|
| 76007 |
+
"learning_rate": 4.075668859386261e-06,
|
| 76008 |
+
"loss": 1.4417,
|
| 76009 |
+
"step": 21616
|
| 76010 |
+
},
|
| 76011 |
+
{
|
| 76012 |
+
"epoch": 0.9611844737895158,
|
| 76013 |
+
"grad_norm": 0.05681522190570831,
|
| 76014 |
+
"learning_rate": 4.057062492312691e-06,
|
| 76015 |
+
"loss": 1.4483,
|
| 76016 |
+
"step": 21618
|
| 76017 |
+
},
|
| 76018 |
+
{
|
| 76019 |
+
"epoch": 0.9612733982481881,
|
| 76020 |
+
"grad_norm": 0.05852191522717476,
|
| 76021 |
+
"learning_rate": 4.038498520626932e-06,
|
| 76022 |
+
"loss": 1.4491,
|
| 76023 |
+
"step": 21620
|
| 76024 |
+
},
|
| 76025 |
+
{
|
| 76026 |
+
"epoch": 0.9613623227068605,
|
| 76027 |
+
"grad_norm": 0.05827983468770981,
|
| 76028 |
+
"learning_rate": 4.0199769459159376e-06,
|
| 76029 |
+
"loss": 1.4491,
|
| 76030 |
+
"step": 21622
|
| 76031 |
+
},
|
| 76032 |
+
{
|
| 76033 |
+
"epoch": 0.9614512471655329,
|
| 76034 |
+
"grad_norm": 0.05857324227690697,
|
| 76035 |
+
"learning_rate": 4.001497769763107e-06,
|
| 76036 |
+
"loss": 1.4505,
|
| 76037 |
+
"step": 21624
|
| 76038 |
+
},
|
| 76039 |
+
{
|
| 76040 |
+
"epoch": 0.9615401716242052,
|
| 76041 |
+
"grad_norm": 0.058631282299757004,
|
| 76042 |
+
"learning_rate": 3.983060993747956e-06,
|
| 76043 |
+
"loss": 1.4475,
|
| 76044 |
+
"step": 21626
|
| 76045 |
+
},
|
| 76046 |
+
{
|
| 76047 |
+
"epoch": 0.9616290960828776,
|
| 76048 |
+
"grad_norm": 0.056906893849372864,
|
| 76049 |
+
"learning_rate": 3.964666619446666e-06,
|
| 76050 |
+
"loss": 1.4458,
|
| 76051 |
+
"step": 21628
|
| 76052 |
+
},
|
| 76053 |
+
{
|
| 76054 |
+
"epoch": 0.96171802054155,
|
| 76055 |
+
"grad_norm": 0.057671476155519485,
|
| 76056 |
+
"learning_rate": 3.946314648431648e-06,
|
| 76057 |
+
"loss": 1.4407,
|
| 76058 |
+
"step": 21630
|
| 76059 |
+
},
|
| 76060 |
+
{
|
| 76061 |
+
"epoch": 0.9618069450002223,
|
| 76062 |
+
"grad_norm": 0.057756319642066956,
|
| 76063 |
+
"learning_rate": 3.928005082271646e-06,
|
| 76064 |
+
"loss": 1.4427,
|
| 76065 |
+
"step": 21632
|
| 76066 |
+
},
|
| 76067 |
+
{
|
| 76068 |
+
"epoch": 0.9618958694588947,
|
| 76069 |
+
"grad_norm": 0.05779024213552475,
|
| 76070 |
+
"learning_rate": 3.909737922531909e-06,
|
| 76071 |
+
"loss": 1.4454,
|
| 76072 |
+
"step": 21634
|
| 76073 |
+
},
|
| 76074 |
+
{
|
| 76075 |
+
"epoch": 0.961984793917567,
|
| 76076 |
+
"grad_norm": 0.05701223015785217,
|
| 76077 |
+
"learning_rate": 3.891513170773964e-06,
|
| 76078 |
+
"loss": 1.4496,
|
| 76079 |
+
"step": 21636
|
| 76080 |
+
},
|
| 76081 |
+
{
|
| 76082 |
+
"epoch": 0.9620737183762393,
|
| 76083 |
+
"grad_norm": 0.05769617483019829,
|
| 76084 |
+
"learning_rate": 3.873330828555788e-06,
|
| 76085 |
+
"loss": 1.443,
|
| 76086 |
+
"step": 21638
|
| 76087 |
+
},
|
| 76088 |
+
{
|
| 76089 |
+
"epoch": 0.9621626428349117,
|
| 76090 |
+
"grad_norm": 0.05806345120072365,
|
| 76091 |
+
"learning_rate": 3.855190897431637e-06,
|
| 76092 |
+
"loss": 1.4428,
|
| 76093 |
+
"step": 21640
|
| 76094 |
+
},
|
| 76095 |
+
{
|
| 76096 |
+
"epoch": 0.9622515672935841,
|
| 76097 |
+
"grad_norm": 0.0573996864259243,
|
| 76098 |
+
"learning_rate": 3.837093378952217e-06,
|
| 76099 |
+
"loss": 1.4492,
|
| 76100 |
+
"step": 21642
|
| 76101 |
+
},
|
| 76102 |
+
{
|
| 76103 |
+
"epoch": 0.9623404917522564,
|
| 76104 |
+
"grad_norm": 0.05671732872724533,
|
| 76105 |
+
"learning_rate": 3.81903827466451e-06,
|
| 76106 |
+
"loss": 1.4445,
|
| 76107 |
+
"step": 21644
|
| 76108 |
+
},
|
| 76109 |
+
{
|
| 76110 |
+
"epoch": 0.9624294162109288,
|
| 76111 |
+
"grad_norm": 0.057457756251096725,
|
| 76112 |
+
"learning_rate": 3.801025586112061e-06,
|
| 76113 |
+
"loss": 1.4476,
|
| 76114 |
+
"step": 21646
|
| 76115 |
+
},
|
| 76116 |
+
{
|
| 76117 |
+
"epoch": 0.9625183406696012,
|
| 76118 |
+
"grad_norm": 0.05706573650240898,
|
| 76119 |
+
"learning_rate": 3.783055314834638e-06,
|
| 76120 |
+
"loss": 1.4462,
|
| 76121 |
+
"step": 21648
|
| 76122 |
+
},
|
| 76123 |
+
{
|
| 76124 |
+
"epoch": 0.9626072651282735,
|
| 76125 |
+
"grad_norm": 0.05761359632015228,
|
| 76126 |
+
"learning_rate": 3.765127462368345e-06,
|
| 76127 |
+
"loss": 1.4483,
|
| 76128 |
+
"step": 21650
|
| 76129 |
+
},
|
| 76130 |
+
{
|
| 76131 |
+
"epoch": 0.9626961895869459,
|
| 76132 |
+
"grad_norm": 0.057819437235593796,
|
| 76133 |
+
"learning_rate": 3.7472420302458455e-06,
|
| 76134 |
+
"loss": 1.4464,
|
| 76135 |
+
"step": 21652
|
| 76136 |
+
},
|
| 76137 |
+
{
|
| 76138 |
+
"epoch": 0.9627851140456183,
|
| 76139 |
+
"grad_norm": 0.05833171680569649,
|
| 76140 |
+
"learning_rate": 3.729399019996027e-06,
|
| 76141 |
+
"loss": 1.4449,
|
| 76142 |
+
"step": 21654
|
| 76143 |
+
},
|
| 76144 |
+
{
|
| 76145 |
+
"epoch": 0.9628740385042907,
|
| 76146 |
+
"grad_norm": 0.057252366095781326,
|
| 76147 |
+
"learning_rate": 3.71159843314417e-06,
|
| 76148 |
+
"loss": 1.4462,
|
| 76149 |
+
"step": 21656
|
| 76150 |
+
},
|
| 76151 |
+
{
|
| 76152 |
+
"epoch": 0.9629629629629629,
|
| 76153 |
+
"grad_norm": 0.058833926916122437,
|
| 76154 |
+
"learning_rate": 3.69384027121189e-06,
|
| 76155 |
+
"loss": 1.4441,
|
| 76156 |
+
"step": 21658
|
| 76157 |
+
},
|
| 76158 |
+
{
|
| 76159 |
+
"epoch": 0.9630518874216353,
|
| 76160 |
+
"grad_norm": 0.05726613849401474,
|
| 76161 |
+
"learning_rate": 3.676124535717307e-06,
|
| 76162 |
+
"loss": 1.4473,
|
| 76163 |
+
"step": 21660
|
| 76164 |
+
},
|
| 76165 |
+
{
|
| 76166 |
+
"epoch": 0.9631408118803076,
|
| 76167 |
+
"grad_norm": 0.05816185846924782,
|
| 76168 |
+
"learning_rate": 3.658451228174875e-06,
|
| 76169 |
+
"loss": 1.4407,
|
| 76170 |
+
"step": 21662
|
| 76171 |
+
},
|
| 76172 |
+
{
|
| 76173 |
+
"epoch": 0.96322973633898,
|
| 76174 |
+
"grad_norm": 0.05848447605967522,
|
| 76175 |
+
"learning_rate": 3.640820350095331e-06,
|
| 76176 |
+
"loss": 1.4431,
|
| 76177 |
+
"step": 21664
|
| 76178 |
+
},
|
| 76179 |
+
{
|
| 76180 |
+
"epoch": 0.9633186607976524,
|
| 76181 |
+
"grad_norm": 0.05794060230255127,
|
| 76182 |
+
"learning_rate": 3.6232319029858017e-06,
|
| 76183 |
+
"loss": 1.4511,
|
| 76184 |
+
"step": 21666
|
| 76185 |
+
},
|
| 76186 |
+
{
|
| 76187 |
+
"epoch": 0.9634075852563247,
|
| 76188 |
+
"grad_norm": 0.05679847672581673,
|
| 76189 |
+
"learning_rate": 3.6056858883499187e-06,
|
| 76190 |
+
"loss": 1.4469,
|
| 76191 |
+
"step": 21668
|
| 76192 |
+
},
|
| 76193 |
+
{
|
| 76194 |
+
"epoch": 0.9634965097149971,
|
| 76195 |
+
"grad_norm": 0.05763591080904007,
|
| 76196 |
+
"learning_rate": 3.588182307687482e-06,
|
| 76197 |
+
"loss": 1.4494,
|
| 76198 |
+
"step": 21670
|
| 76199 |
+
},
|
| 76200 |
+
{
|
| 76201 |
+
"epoch": 0.9635854341736695,
|
| 76202 |
+
"grad_norm": 0.05757317319512367,
|
| 76203 |
+
"learning_rate": 3.5707211624949055e-06,
|
| 76204 |
+
"loss": 1.4486,
|
| 76205 |
+
"step": 21672
|
| 76206 |
+
},
|
| 76207 |
+
{
|
| 76208 |
+
"epoch": 0.9636743586323419,
|
| 76209 |
+
"grad_norm": 0.057650867849588394,
|
| 76210 |
+
"learning_rate": 3.5533024542647174e-06,
|
| 76211 |
+
"loss": 1.4514,
|
| 76212 |
+
"step": 21674
|
| 76213 |
+
},
|
| 76214 |
+
{
|
| 76215 |
+
"epoch": 0.9637632830910142,
|
| 76216 |
+
"grad_norm": 0.057070985436439514,
|
| 76217 |
+
"learning_rate": 3.53592618448606e-06,
|
| 76218 |
+
"loss": 1.4482,
|
| 76219 |
+
"step": 21676
|
| 76220 |
+
},
|
| 76221 |
+
{
|
| 76222 |
+
"epoch": 0.9638522075496866,
|
| 76223 |
+
"grad_norm": 0.05852733552455902,
|
| 76224 |
+
"learning_rate": 3.5185923546442457e-06,
|
| 76225 |
+
"loss": 1.448,
|
| 76226 |
+
"step": 21678
|
| 76227 |
+
},
|
| 76228 |
+
{
|
| 76229 |
+
"epoch": 0.9639411320083588,
|
| 76230 |
+
"grad_norm": 0.058048900216817856,
|
| 76231 |
+
"learning_rate": 3.501300966221088e-06,
|
| 76232 |
+
"loss": 1.4493,
|
| 76233 |
+
"step": 21680
|
| 76234 |
+
},
|
| 76235 |
+
{
|
| 76236 |
+
"epoch": 0.9640300564670312,
|
| 76237 |
+
"grad_norm": 0.05750441551208496,
|
| 76238 |
+
"learning_rate": 3.4840520206947392e-06,
|
| 76239 |
+
"loss": 1.4466,
|
| 76240 |
+
"step": 21682
|
| 76241 |
+
},
|
| 76242 |
+
{
|
| 76243 |
+
"epoch": 0.9641189809257036,
|
| 76244 |
+
"grad_norm": 0.05819224938750267,
|
| 76245 |
+
"learning_rate": 3.4668455195396854e-06,
|
| 76246 |
+
"loss": 1.4456,
|
| 76247 |
+
"step": 21684
|
| 76248 |
+
},
|
| 76249 |
+
{
|
| 76250 |
+
"epoch": 0.964207905384376,
|
| 76251 |
+
"grad_norm": 0.057015545666217804,
|
| 76252 |
+
"learning_rate": 3.449681464226806e-06,
|
| 76253 |
+
"loss": 1.4428,
|
| 76254 |
+
"step": 21686
|
| 76255 |
+
},
|
| 76256 |
+
{
|
| 76257 |
+
"epoch": 0.9642968298430483,
|
| 76258 |
+
"grad_norm": 0.056825630366802216,
|
| 76259 |
+
"learning_rate": 3.4325598562234274e-06,
|
| 76260 |
+
"loss": 1.447,
|
| 76261 |
+
"step": 21688
|
| 76262 |
+
},
|
| 76263 |
+
{
|
| 76264 |
+
"epoch": 0.9643857543017207,
|
| 76265 |
+
"grad_norm": 0.05699056014418602,
|
| 76266 |
+
"learning_rate": 3.415480696993101e-06,
|
| 76267 |
+
"loss": 1.4432,
|
| 76268 |
+
"step": 21690
|
| 76269 |
+
},
|
| 76270 |
+
{
|
| 76271 |
+
"epoch": 0.9644746787603931,
|
| 76272 |
+
"grad_norm": 0.057919371873140335,
|
| 76273 |
+
"learning_rate": 3.3984439879958807e-06,
|
| 76274 |
+
"loss": 1.4484,
|
| 76275 |
+
"step": 21692
|
| 76276 |
+
},
|
| 76277 |
+
{
|
| 76278 |
+
"epoch": 0.9645636032190654,
|
| 76279 |
+
"grad_norm": 0.05731486156582832,
|
| 76280 |
+
"learning_rate": 3.381449730688102e-06,
|
| 76281 |
+
"loss": 1.4495,
|
| 76282 |
+
"step": 21694
|
| 76283 |
+
},
|
| 76284 |
+
{
|
| 76285 |
+
"epoch": 0.9646525276777378,
|
| 76286 |
+
"grad_norm": 0.05739450827240944,
|
| 76287 |
+
"learning_rate": 3.3644979265225474e-06,
|
| 76288 |
+
"loss": 1.449,
|
| 76289 |
+
"step": 21696
|
| 76290 |
+
},
|
| 76291 |
+
{
|
| 76292 |
+
"epoch": 0.9647414521364102,
|
| 76293 |
+
"grad_norm": 0.058767009526491165,
|
| 76294 |
+
"learning_rate": 3.3475885769482796e-06,
|
| 76295 |
+
"loss": 1.4444,
|
| 76296 |
+
"step": 21698
|
| 76297 |
+
},
|
| 76298 |
+
{
|
| 76299 |
+
"epoch": 0.9648303765950825,
|
| 76300 |
+
"grad_norm": 0.05814867466688156,
|
| 76301 |
+
"learning_rate": 3.330721683410809e-06,
|
| 76302 |
+
"loss": 1.4522,
|
| 76303 |
+
"step": 21700
|
| 76304 |
+
},
|
| 76305 |
+
{
|
| 76306 |
+
"epoch": 0.9649193010537548,
|
| 76307 |
+
"grad_norm": 0.05803316831588745,
|
| 76308 |
+
"learning_rate": 3.313897247352038e-06,
|
| 76309 |
+
"loss": 1.4435,
|
| 76310 |
+
"step": 21702
|
| 76311 |
+
},
|
| 76312 |
+
{
|
| 76313 |
+
"epoch": 0.9650082255124272,
|
| 76314 |
+
"grad_norm": 0.05720685422420502,
|
| 76315 |
+
"learning_rate": 3.297115270210149e-06,
|
| 76316 |
+
"loss": 1.4459,
|
| 76317 |
+
"step": 21704
|
| 76318 |
+
},
|
| 76319 |
+
{
|
| 76320 |
+
"epoch": 0.9650971499710995,
|
| 76321 |
+
"grad_norm": 0.05773816257715225,
|
| 76322 |
+
"learning_rate": 3.2803757534197175e-06,
|
| 76323 |
+
"loss": 1.4459,
|
| 76324 |
+
"step": 21706
|
| 76325 |
+
},
|
| 76326 |
+
{
|
| 76327 |
+
"epoch": 0.9651860744297719,
|
| 76328 |
+
"grad_norm": 0.05913296714425087,
|
| 76329 |
+
"learning_rate": 3.263678698411765e-06,
|
| 76330 |
+
"loss": 1.4439,
|
| 76331 |
+
"step": 21708
|
| 76332 |
+
},
|
| 76333 |
+
{
|
| 76334 |
+
"epoch": 0.9652749988884443,
|
| 76335 |
+
"grad_norm": 0.057658907026052475,
|
| 76336 |
+
"learning_rate": 3.2470241066136495e-06,
|
| 76337 |
+
"loss": 1.4482,
|
| 76338 |
+
"step": 21710
|
| 76339 |
+
},
|
| 76340 |
+
{
|
| 76341 |
+
"epoch": 0.9653639233471166,
|
| 76342 |
+
"grad_norm": 0.05754270777106285,
|
| 76343 |
+
"learning_rate": 3.230411979448955e-06,
|
| 76344 |
+
"loss": 1.4479,
|
| 76345 |
+
"step": 21712
|
| 76346 |
+
},
|
| 76347 |
+
{
|
| 76348 |
+
"epoch": 0.965452847805789,
|
| 76349 |
+
"grad_norm": 0.05791141465306282,
|
| 76350 |
+
"learning_rate": 3.2138423183378785e-06,
|
| 76351 |
+
"loss": 1.4476,
|
| 76352 |
+
"step": 21714
|
| 76353 |
+
},
|
| 76354 |
+
{
|
| 76355 |
+
"epoch": 0.9655417722644614,
|
| 76356 |
+
"grad_norm": 0.057228460907936096,
|
| 76357 |
+
"learning_rate": 3.1973151246967868e-06,
|
| 76358 |
+
"loss": 1.4455,
|
| 76359 |
+
"step": 21716
|
| 76360 |
+
},
|
| 76361 |
+
{
|
| 76362 |
+
"epoch": 0.9656306967231337,
|
| 76363 |
+
"grad_norm": 0.05856222286820412,
|
| 76364 |
+
"learning_rate": 3.18083039993855e-06,
|
| 76365 |
+
"loss": 1.4421,
|
| 76366 |
+
"step": 21718
|
| 76367 |
+
},
|
| 76368 |
+
{
|
| 76369 |
+
"epoch": 0.9657196211818061,
|
| 76370 |
+
"grad_norm": 0.05858870595693588,
|
| 76371 |
+
"learning_rate": 3.1643881454723744e-06,
|
| 76372 |
+
"loss": 1.4433,
|
| 76373 |
+
"step": 21720
|
| 76374 |
+
},
|
| 76375 |
+
{
|
| 76376 |
+
"epoch": 0.9658085456404785,
|
| 76377 |
+
"grad_norm": 0.057810429483652115,
|
| 76378 |
+
"learning_rate": 3.1479883627037464e-06,
|
| 76379 |
+
"loss": 1.4414,
|
| 76380 |
+
"step": 21722
|
| 76381 |
+
},
|
| 76382 |
+
{
|
| 76383 |
+
"epoch": 0.9658974700991507,
|
| 76384 |
+
"grad_norm": 0.05786995589733124,
|
| 76385 |
+
"learning_rate": 3.131631053034656e-06,
|
| 76386 |
+
"loss": 1.4479,
|
| 76387 |
+
"step": 21724
|
| 76388 |
+
},
|
| 76389 |
+
{
|
| 76390 |
+
"epoch": 0.9659863945578231,
|
| 76391 |
+
"grad_norm": 0.05765226110816002,
|
| 76392 |
+
"learning_rate": 3.115316217863373e-06,
|
| 76393 |
+
"loss": 1.444,
|
| 76394 |
+
"step": 21726
|
| 76395 |
+
},
|
| 76396 |
+
{
|
| 76397 |
+
"epoch": 0.9660753190164955,
|
| 76398 |
+
"grad_norm": 0.05825477093458176,
|
| 76399 |
+
"learning_rate": 3.0990438585845048e-06,
|
| 76400 |
+
"loss": 1.4505,
|
| 76401 |
+
"step": 21728
|
| 76402 |
+
},
|
| 76403 |
+
{
|
| 76404 |
+
"epoch": 0.9661642434751678,
|
| 76405 |
+
"grad_norm": 0.057248134166002274,
|
| 76406 |
+
"learning_rate": 3.082813976589216e-06,
|
| 76407 |
+
"loss": 1.4484,
|
| 76408 |
+
"step": 21730
|
| 76409 |
+
},
|
| 76410 |
+
{
|
| 76411 |
+
"epoch": 0.9662531679338402,
|
| 76412 |
+
"grad_norm": 0.057661522179841995,
|
| 76413 |
+
"learning_rate": 3.0666265732647856e-06,
|
| 76414 |
+
"loss": 1.4497,
|
| 76415 |
+
"step": 21732
|
| 76416 |
+
},
|
| 76417 |
+
{
|
| 76418 |
+
"epoch": 0.9663420923925126,
|
| 76419 |
+
"grad_norm": 0.05723179504275322,
|
| 76420 |
+
"learning_rate": 3.0504816499950516e-06,
|
| 76421 |
+
"loss": 1.4442,
|
| 76422 |
+
"step": 21734
|
| 76423 |
+
},
|
| 76424 |
+
{
|
| 76425 |
+
"epoch": 0.9664310168511849,
|
| 76426 |
+
"grad_norm": 0.05735337361693382,
|
| 76427 |
+
"learning_rate": 3.034379208160076e-06,
|
| 76428 |
+
"loss": 1.4482,
|
| 76429 |
+
"step": 21736
|
| 76430 |
+
},
|
| 76431 |
+
{
|
| 76432 |
+
"epoch": 0.9665199413098573,
|
| 76433 |
+
"grad_norm": 0.05809729918837547,
|
| 76434 |
+
"learning_rate": 3.018319249136481e-06,
|
| 76435 |
+
"loss": 1.4436,
|
| 76436 |
+
"step": 21738
|
| 76437 |
+
},
|
| 76438 |
+
{
|
| 76439 |
+
"epoch": 0.9666088657685297,
|
| 76440 |
+
"grad_norm": 0.05790387839078903,
|
| 76441 |
+
"learning_rate": 3.002301774297056e-06,
|
| 76442 |
+
"loss": 1.4404,
|
| 76443 |
+
"step": 21740
|
| 76444 |
+
},
|
| 76445 |
+
{
|
| 76446 |
+
"epoch": 0.966697790227202,
|
| 76447 |
+
"grad_norm": 0.05688699334859848,
|
| 76448 |
+
"learning_rate": 2.9863267850110955e-06,
|
| 76449 |
+
"loss": 1.449,
|
| 76450 |
+
"step": 21742
|
| 76451 |
+
},
|
| 76452 |
+
{
|
| 76453 |
+
"epoch": 0.9667867146858744,
|
| 76454 |
+
"grad_norm": 0.056998882442712784,
|
| 76455 |
+
"learning_rate": 2.9703942826441734e-06,
|
| 76456 |
+
"loss": 1.4472,
|
| 76457 |
+
"step": 21744
|
| 76458 |
+
},
|
| 76459 |
+
{
|
| 76460 |
+
"epoch": 0.9668756391445467,
|
| 76461 |
+
"grad_norm": 0.058484215289354324,
|
| 76462 |
+
"learning_rate": 2.9545042685583112e-06,
|
| 76463 |
+
"loss": 1.4486,
|
| 76464 |
+
"step": 21746
|
| 76465 |
+
},
|
| 76466 |
+
{
|
| 76467 |
+
"epoch": 0.966964563603219,
|
| 76468 |
+
"grad_norm": 0.05785754323005676,
|
| 76469 |
+
"learning_rate": 2.938656744111812e-06,
|
| 76470 |
+
"loss": 1.4462,
|
| 76471 |
+
"step": 21748
|
| 76472 |
+
},
|
| 76473 |
+
{
|
| 76474 |
+
"epoch": 0.9670534880618914,
|
| 76475 |
+
"grad_norm": 0.05740215256810188,
|
| 76476 |
+
"learning_rate": 2.9228517106594244e-06,
|
| 76477 |
+
"loss": 1.4409,
|
| 76478 |
+
"step": 21750
|
| 76479 |
+
},
|
| 76480 |
+
{
|
| 76481 |
+
"epoch": 0.9671424125205638,
|
| 76482 |
+
"grad_norm": 0.057988543063402176,
|
| 76483 |
+
"learning_rate": 2.9070891695521796e-06,
|
| 76484 |
+
"loss": 1.4468,
|
| 76485 |
+
"step": 21752
|
| 76486 |
+
},
|
| 76487 |
+
{
|
| 76488 |
+
"epoch": 0.9672313369792361,
|
| 76489 |
+
"grad_norm": 0.05775817483663559,
|
| 76490 |
+
"learning_rate": 2.8913691221376102e-06,
|
| 76491 |
+
"loss": 1.4534,
|
| 76492 |
+
"step": 21754
|
| 76493 |
+
},
|
| 76494 |
+
{
|
| 76495 |
+
"epoch": 0.9673202614379085,
|
| 76496 |
+
"grad_norm": 0.05813591554760933,
|
| 76497 |
+
"learning_rate": 2.8756915697594756e-06,
|
| 76498 |
+
"loss": 1.4463,
|
| 76499 |
+
"step": 21756
|
| 76500 |
+
},
|
| 76501 |
+
{
|
| 76502 |
+
"epoch": 0.9674091858965809,
|
| 76503 |
+
"grad_norm": 0.05779729038476944,
|
| 76504 |
+
"learning_rate": 2.86005651375798e-06,
|
| 76505 |
+
"loss": 1.4475,
|
| 76506 |
+
"step": 21758
|
| 76507 |
+
},
|
| 76508 |
+
{
|
| 76509 |
+
"epoch": 0.9674981103552532,
|
| 76510 |
+
"grad_norm": 0.05689067393541336,
|
| 76511 |
+
"learning_rate": 2.844463955469723e-06,
|
| 76512 |
+
"loss": 1.4441,
|
| 76513 |
+
"step": 21760
|
| 76514 |
+
},
|
| 76515 |
+
{
|
| 76516 |
+
"epoch": 0.9675870348139256,
|
| 76517 |
+
"grad_norm": 0.05785556882619858,
|
| 76518 |
+
"learning_rate": 2.828913896227525e-06,
|
| 76519 |
+
"loss": 1.446,
|
| 76520 |
+
"step": 21762
|
| 76521 |
+
},
|
| 76522 |
+
{
|
| 76523 |
+
"epoch": 0.967675959272598,
|
| 76524 |
+
"grad_norm": 0.0574394054710865,
|
| 76525 |
+
"learning_rate": 2.8134063373607687e-06,
|
| 76526 |
+
"loss": 1.4418,
|
| 76527 |
+
"step": 21764
|
| 76528 |
+
},
|
| 76529 |
+
{
|
| 76530 |
+
"epoch": 0.9677648837312702,
|
| 76531 |
+
"grad_norm": 0.05662890523672104,
|
| 76532 |
+
"learning_rate": 2.79794128019506e-06,
|
| 76533 |
+
"loss": 1.4432,
|
| 76534 |
+
"step": 21766
|
| 76535 |
+
},
|
| 76536 |
+
{
|
| 76537 |
+
"epoch": 0.9678538081899426,
|
| 76538 |
+
"grad_norm": 0.0582660436630249,
|
| 76539 |
+
"learning_rate": 2.7825187260523966e-06,
|
| 76540 |
+
"loss": 1.4476,
|
| 76541 |
+
"step": 21768
|
| 76542 |
+
},
|
| 76543 |
+
{
|
| 76544 |
+
"epoch": 0.967942732648615,
|
| 76545 |
+
"grad_norm": 0.05858684331178665,
|
| 76546 |
+
"learning_rate": 2.767138676251224e-06,
|
| 76547 |
+
"loss": 1.4451,
|
| 76548 |
+
"step": 21770
|
| 76549 |
+
},
|
| 76550 |
+
{
|
| 76551 |
+
"epoch": 0.9680316571072873,
|
| 76552 |
+
"grad_norm": 0.056393761187791824,
|
| 76553 |
+
"learning_rate": 2.7518011321062687e-06,
|
| 76554 |
+
"loss": 1.444,
|
| 76555 |
+
"step": 21772
|
| 76556 |
+
},
|
| 76557 |
+
{
|
| 76558 |
+
"epoch": 0.9681205815659597,
|
| 76559 |
+
"grad_norm": 0.058356158435344696,
|
| 76560 |
+
"learning_rate": 2.736506094928648e-06,
|
| 76561 |
+
"loss": 1.4424,
|
| 76562 |
+
"step": 21774
|
| 76563 |
+
},
|
| 76564 |
+
{
|
| 76565 |
+
"epoch": 0.9682095060246321,
|
| 76566 |
+
"grad_norm": 0.05726497247815132,
|
| 76567 |
+
"learning_rate": 2.7212535660258718e-06,
|
| 76568 |
+
"loss": 1.4445,
|
| 76569 |
+
"step": 21776
|
| 76570 |
+
},
|
| 76571 |
+
{
|
| 76572 |
+
"epoch": 0.9682984304833044,
|
| 76573 |
+
"grad_norm": 0.057576339691877365,
|
| 76574 |
+
"learning_rate": 2.7060435467017865e-06,
|
| 76575 |
+
"loss": 1.4466,
|
| 76576 |
+
"step": 21778
|
| 76577 |
+
},
|
| 76578 |
+
{
|
| 76579 |
+
"epoch": 0.9683873549419768,
|
| 76580 |
+
"grad_norm": 0.05778598040342331,
|
| 76581 |
+
"learning_rate": 2.6908760382565735e-06,
|
| 76582 |
+
"loss": 1.448,
|
| 76583 |
+
"step": 21780
|
| 76584 |
+
},
|
| 76585 |
+
{
|
| 76586 |
+
"epoch": 0.9684762794006492,
|
| 76587 |
+
"grad_norm": 0.056812673807144165,
|
| 76588 |
+
"learning_rate": 2.6757510419868624e-06,
|
| 76589 |
+
"loss": 1.4487,
|
| 76590 |
+
"step": 21782
|
| 76591 |
+
},
|
| 76592 |
+
{
|
| 76593 |
+
"epoch": 0.9685652038593215,
|
| 76594 |
+
"grad_norm": 0.05817960575222969,
|
| 76595 |
+
"learning_rate": 2.660668559185564e-06,
|
| 76596 |
+
"loss": 1.4498,
|
| 76597 |
+
"step": 21784
|
| 76598 |
+
},
|
| 76599 |
+
{
|
| 76600 |
+
"epoch": 0.9686541283179939,
|
| 76601 |
+
"grad_norm": 0.05673222243785858,
|
| 76602 |
+
"learning_rate": 2.6456285911420353e-06,
|
| 76603 |
+
"loss": 1.4489,
|
| 76604 |
+
"step": 21786
|
| 76605 |
+
},
|
| 76606 |
+
{
|
| 76607 |
+
"epoch": 0.9687430527766662,
|
| 76608 |
+
"grad_norm": 0.058771610260009766,
|
| 76609 |
+
"learning_rate": 2.63063113914197e-06,
|
| 76610 |
+
"loss": 1.4485,
|
| 76611 |
+
"step": 21788
|
| 76612 |
+
},
|
| 76613 |
+
{
|
| 76614 |
+
"epoch": 0.9688319772353385,
|
| 76615 |
+
"grad_norm": 0.05711560696363449,
|
| 76616 |
+
"learning_rate": 2.615676204467343e-06,
|
| 76617 |
+
"loss": 1.4452,
|
| 76618 |
+
"step": 21790
|
| 76619 |
+
},
|
| 76620 |
+
{
|
| 76621 |
+
"epoch": 0.9689209016940109,
|
| 76622 |
+
"grad_norm": 0.057385995984077454,
|
| 76623 |
+
"learning_rate": 2.6007637883966872e-06,
|
| 76624 |
+
"loss": 1.4476,
|
| 76625 |
+
"step": 21792
|
| 76626 |
+
},
|
| 76627 |
+
{
|
| 76628 |
+
"epoch": 0.9690098261526833,
|
| 76629 |
+
"grad_norm": 0.05783890560269356,
|
| 76630 |
+
"learning_rate": 2.5858938922046495e-06,
|
| 76631 |
+
"loss": 1.4458,
|
| 76632 |
+
"step": 21794
|
| 76633 |
+
},
|
| 76634 |
+
{
|
| 76635 |
+
"epoch": 0.9690987506113556,
|
| 76636 |
+
"grad_norm": 0.05787234753370285,
|
| 76637 |
+
"learning_rate": 2.571066517162435e-06,
|
| 76638 |
+
"loss": 1.4456,
|
| 76639 |
+
"step": 21796
|
| 76640 |
+
},
|
| 76641 |
+
{
|
| 76642 |
+
"epoch": 0.969187675070028,
|
| 76643 |
+
"grad_norm": 0.05799291655421257,
|
| 76644 |
+
"learning_rate": 2.556281664537585e-06,
|
| 76645 |
+
"loss": 1.4469,
|
| 76646 |
+
"step": 21798
|
| 76647 |
+
},
|
| 76648 |
+
{
|
| 76649 |
+
"epoch": 0.9692765995287004,
|
| 76650 |
+
"grad_norm": 0.05841020122170448,
|
| 76651 |
+
"learning_rate": 2.541539335593923e-06,
|
| 76652 |
+
"loss": 1.4422,
|
| 76653 |
+
"step": 21800
|
| 76654 |
+
},
|
| 76655 |
+
{
|
| 76656 |
+
"epoch": 0.9693655239873727,
|
| 76657 |
+
"grad_norm": 0.057031866163015366,
|
| 76658 |
+
"learning_rate": 2.526839531591718e-06,
|
| 76659 |
+
"loss": 1.4509,
|
| 76660 |
+
"step": 21802
|
| 76661 |
+
},
|
| 76662 |
+
{
|
| 76663 |
+
"epoch": 0.9694544484460451,
|
| 76664 |
+
"grad_norm": 0.05724103003740311,
|
| 76665 |
+
"learning_rate": 2.5121822537875206e-06,
|
| 76666 |
+
"loss": 1.4438,
|
| 76667 |
+
"step": 21804
|
| 76668 |
+
},
|
| 76669 |
+
{
|
| 76670 |
+
"epoch": 0.9695433729047175,
|
| 76671 |
+
"grad_norm": 0.057726915925741196,
|
| 76672 |
+
"learning_rate": 2.4975675034343835e-06,
|
| 76673 |
+
"loss": 1.4444,
|
| 76674 |
+
"step": 21806
|
| 76675 |
+
},
|
| 76676 |
+
{
|
| 76677 |
+
"epoch": 0.9696322973633899,
|
| 76678 |
+
"grad_norm": 0.058166489005088806,
|
| 76679 |
+
"learning_rate": 2.482995281781586e-06,
|
| 76680 |
+
"loss": 1.4455,
|
| 76681 |
+
"step": 21808
|
| 76682 |
+
},
|
| 76683 |
+
{
|
| 76684 |
+
"epoch": 0.9697212218220621,
|
| 76685 |
+
"grad_norm": 0.058684688061475754,
|
| 76686 |
+
"learning_rate": 2.4684655900748532e-06,
|
| 76687 |
+
"loss": 1.4486,
|
| 76688 |
+
"step": 21810
|
| 76689 |
+
},
|
| 76690 |
+
{
|
| 76691 |
+
"epoch": 0.9698101462807345,
|
| 76692 |
+
"grad_norm": 0.05730533227324486,
|
| 76693 |
+
"learning_rate": 2.453978429556247e-06,
|
| 76694 |
+
"loss": 1.4456,
|
| 76695 |
+
"step": 21812
|
| 76696 |
+
},
|
| 76697 |
+
{
|
| 76698 |
+
"epoch": 0.9698990707394068,
|
| 76699 |
+
"grad_norm": 0.057590242475271225,
|
| 76700 |
+
"learning_rate": 2.439533801464111e-06,
|
| 76701 |
+
"loss": 1.4445,
|
| 76702 |
+
"step": 21814
|
| 76703 |
+
},
|
| 76704 |
+
{
|
| 76705 |
+
"epoch": 0.9699879951980792,
|
| 76706 |
+
"grad_norm": 0.057827651500701904,
|
| 76707 |
+
"learning_rate": 2.425131707033346e-06,
|
| 76708 |
+
"loss": 1.4461,
|
| 76709 |
+
"step": 21816
|
| 76710 |
+
},
|
| 76711 |
+
{
|
| 76712 |
+
"epoch": 0.9700769196567516,
|
| 76713 |
+
"grad_norm": 0.05834771320223808,
|
| 76714 |
+
"learning_rate": 2.4107721474950774e-06,
|
| 76715 |
+
"loss": 1.4415,
|
| 76716 |
+
"step": 21818
|
| 76717 |
+
},
|
| 76718 |
+
{
|
| 76719 |
+
"epoch": 0.970165844115424,
|
| 76720 |
+
"grad_norm": 0.057678621262311935,
|
| 76721 |
+
"learning_rate": 2.3964551240767684e-06,
|
| 76722 |
+
"loss": 1.4477,
|
| 76723 |
+
"step": 21820
|
| 76724 |
+
},
|
| 76725 |
+
{
|
| 76726 |
+
"epoch": 0.9702547685740963,
|
| 76727 |
+
"grad_norm": 0.05750925838947296,
|
| 76728 |
+
"learning_rate": 2.382180638002385e-06,
|
| 76729 |
+
"loss": 1.4441,
|
| 76730 |
+
"step": 21822
|
| 76731 |
+
},
|
| 76732 |
+
{
|
| 76733 |
+
"epoch": 0.9703436930327687,
|
| 76734 |
+
"grad_norm": 0.056883055716753006,
|
| 76735 |
+
"learning_rate": 2.3679486904921164e-06,
|
| 76736 |
+
"loss": 1.4452,
|
| 76737 |
+
"step": 21824
|
| 76738 |
+
},
|
| 76739 |
+
{
|
| 76740 |
+
"epoch": 0.970432617491441,
|
| 76741 |
+
"grad_norm": 0.057405877858400345,
|
| 76742 |
+
"learning_rate": 2.353759282762602e-06,
|
| 76743 |
+
"loss": 1.4487,
|
| 76744 |
+
"step": 21826
|
| 76745 |
+
},
|
| 76746 |
+
{
|
| 76747 |
+
"epoch": 0.9705215419501134,
|
| 76748 |
+
"grad_norm": 0.056650012731552124,
|
| 76749 |
+
"learning_rate": 2.339612416026815e-06,
|
| 76750 |
+
"loss": 1.4426,
|
| 76751 |
+
"step": 21828
|
| 76752 |
+
},
|
| 76753 |
+
{
|
| 76754 |
+
"epoch": 0.9706104664087858,
|
| 76755 |
+
"grad_norm": 0.05729759484529495,
|
| 76756 |
+
"learning_rate": 2.3255080914940106e-06,
|
| 76757 |
+
"loss": 1.4483,
|
| 76758 |
+
"step": 21830
|
| 76759 |
+
},
|
| 76760 |
+
{
|
| 76761 |
+
"epoch": 0.970699390867458,
|
| 76762 |
+
"grad_norm": 0.05801773816347122,
|
| 76763 |
+
"learning_rate": 2.3114463103700023e-06,
|
| 76764 |
+
"loss": 1.4429,
|
| 76765 |
+
"step": 21832
|
| 76766 |
+
},
|
| 76767 |
+
{
|
| 76768 |
+
"epoch": 0.9707883153261304,
|
| 76769 |
+
"grad_norm": 0.057637542486190796,
|
| 76770 |
+
"learning_rate": 2.2974270738567726e-06,
|
| 76771 |
+
"loss": 1.4442,
|
| 76772 |
+
"step": 21834
|
| 76773 |
+
},
|
| 76774 |
+
{
|
| 76775 |
+
"epoch": 0.9708772397848028,
|
| 76776 |
+
"grad_norm": 0.058352191001176834,
|
| 76777 |
+
"learning_rate": 2.2834503831528076e-06,
|
| 76778 |
+
"loss": 1.4448,
|
| 76779 |
+
"step": 21836
|
| 76780 |
+
},
|
| 76781 |
+
{
|
| 76782 |
+
"epoch": 0.9709661642434751,
|
| 76783 |
+
"grad_norm": 0.057397257536649704,
|
| 76784 |
+
"learning_rate": 2.2695162394528735e-06,
|
| 76785 |
+
"loss": 1.4414,
|
| 76786 |
+
"step": 21838
|
| 76787 |
+
},
|
| 76788 |
+
{
|
| 76789 |
+
"epoch": 0.9710550887021475,
|
| 76790 |
+
"grad_norm": 0.05693025141954422,
|
| 76791 |
+
"learning_rate": 2.2556246439481286e-06,
|
| 76792 |
+
"loss": 1.4481,
|
| 76793 |
+
"step": 21840
|
| 76794 |
+
},
|
| 76795 |
+
{
|
| 76796 |
+
"epoch": 0.9711440131608199,
|
| 76797 |
+
"grad_norm": 0.05720305070281029,
|
| 76798 |
+
"learning_rate": 2.2417755978260123e-06,
|
| 76799 |
+
"loss": 1.452,
|
| 76800 |
+
"step": 21842
|
| 76801 |
+
},
|
| 76802 |
+
{
|
| 76803 |
+
"epoch": 0.9712329376194923,
|
| 76804 |
+
"grad_norm": 0.057424396276474,
|
| 76805 |
+
"learning_rate": 2.2279691022705216e-06,
|
| 76806 |
+
"loss": 1.4438,
|
| 76807 |
+
"step": 21844
|
| 76808 |
+
},
|
| 76809 |
+
{
|
| 76810 |
+
"epoch": 0.9713218620781646,
|
| 76811 |
+
"grad_norm": 0.057667020708322525,
|
| 76812 |
+
"learning_rate": 2.2142051584618237e-06,
|
| 76813 |
+
"loss": 1.4478,
|
| 76814 |
+
"step": 21846
|
| 76815 |
+
},
|
| 76816 |
+
{
|
| 76817 |
+
"epoch": 0.971410786536837,
|
| 76818 |
+
"grad_norm": 0.058313172310590744,
|
| 76819 |
+
"learning_rate": 2.200483767576589e-06,
|
| 76820 |
+
"loss": 1.4476,
|
| 76821 |
+
"step": 21848
|
| 76822 |
+
},
|
| 76823 |
+
{
|
| 76824 |
+
"epoch": 0.9714997109955094,
|
| 76825 |
+
"grad_norm": 0.057696230709552765,
|
| 76826 |
+
"learning_rate": 2.1868049307877113e-06,
|
| 76827 |
+
"loss": 1.4433,
|
| 76828 |
+
"step": 21850
|
| 76829 |
+
},
|
| 76830 |
+
{
|
| 76831 |
+
"epoch": 0.9715886354541817,
|
| 76832 |
+
"grad_norm": 0.057298045605421066,
|
| 76833 |
+
"learning_rate": 2.1731686492644787e-06,
|
| 76834 |
+
"loss": 1.447,
|
| 76835 |
+
"step": 21852
|
| 76836 |
+
},
|
| 76837 |
+
{
|
| 76838 |
+
"epoch": 0.971677559912854,
|
| 76839 |
+
"grad_norm": 0.05838226154446602,
|
| 76840 |
+
"learning_rate": 2.1595749241726805e-06,
|
| 76841 |
+
"loss": 1.4426,
|
| 76842 |
+
"step": 21854
|
| 76843 |
+
},
|
| 76844 |
+
{
|
| 76845 |
+
"epoch": 0.9717664843715264,
|
| 76846 |
+
"grad_norm": 0.057391196489334106,
|
| 76847 |
+
"learning_rate": 2.146023756674331e-06,
|
| 76848 |
+
"loss": 1.4436,
|
| 76849 |
+
"step": 21856
|
| 76850 |
+
},
|
| 76851 |
+
{
|
| 76852 |
+
"epoch": 0.9718554088301987,
|
| 76853 |
+
"grad_norm": 0.05735941603779793,
|
| 76854 |
+
"learning_rate": 2.132515147927838e-06,
|
| 76855 |
+
"loss": 1.4449,
|
| 76856 |
+
"step": 21858
|
| 76857 |
+
},
|
| 76858 |
+
{
|
| 76859 |
+
"epoch": 0.9719443332888711,
|
| 76860 |
+
"grad_norm": 0.05849364399909973,
|
| 76861 |
+
"learning_rate": 2.1190490990879994e-06,
|
| 76862 |
+
"loss": 1.4509,
|
| 76863 |
+
"step": 21860
|
| 76864 |
+
},
|
| 76865 |
+
{
|
| 76866 |
+
"epoch": 0.9720332577475435,
|
| 76867 |
+
"grad_norm": 0.05746670067310333,
|
| 76868 |
+
"learning_rate": 2.10562561130595e-06,
|
| 76869 |
+
"loss": 1.4453,
|
| 76870 |
+
"step": 21862
|
| 76871 |
+
},
|
| 76872 |
+
{
|
| 76873 |
+
"epoch": 0.9721221822062158,
|
| 76874 |
+
"grad_norm": 0.05694587528705597,
|
| 76875 |
+
"learning_rate": 2.0922446857291055e-06,
|
| 76876 |
+
"loss": 1.4459,
|
| 76877 |
+
"step": 21864
|
| 76878 |
+
},
|
| 76879 |
+
{
|
| 76880 |
+
"epoch": 0.9722111066648882,
|
| 76881 |
+
"grad_norm": 0.05823846161365509,
|
| 76882 |
+
"learning_rate": 2.0789063235014396e-06,
|
| 76883 |
+
"loss": 1.4386,
|
| 76884 |
+
"step": 21866
|
| 76885 |
+
},
|
| 76886 |
+
{
|
| 76887 |
+
"epoch": 0.9723000311235606,
|
| 76888 |
+
"grad_norm": 0.05770362168550491,
|
| 76889 |
+
"learning_rate": 2.065610525763095e-06,
|
| 76890 |
+
"loss": 1.4468,
|
| 76891 |
+
"step": 21868
|
| 76892 |
+
},
|
| 76893 |
+
{
|
| 76894 |
+
"epoch": 0.9723889555822329,
|
| 76895 |
+
"grad_norm": 0.05771014466881752,
|
| 76896 |
+
"learning_rate": 2.052357293650775e-06,
|
| 76897 |
+
"loss": 1.4437,
|
| 76898 |
+
"step": 21870
|
| 76899 |
+
},
|
| 76900 |
+
{
|
| 76901 |
+
"epoch": 0.9724778800409053,
|
| 76902 |
+
"grad_norm": 0.058271683752536774,
|
| 76903 |
+
"learning_rate": 2.039146628297295e-06,
|
| 76904 |
+
"loss": 1.446,
|
| 76905 |
+
"step": 21872
|
| 76906 |
+
},
|
| 76907 |
+
{
|
| 76908 |
+
"epoch": 0.9725668044995777,
|
| 76909 |
+
"grad_norm": 0.057467687875032425,
|
| 76910 |
+
"learning_rate": 2.0259785308319176e-06,
|
| 76911 |
+
"loss": 1.4474,
|
| 76912 |
+
"step": 21874
|
| 76913 |
+
},
|
| 76914 |
+
{
|
| 76915 |
+
"epoch": 0.9726557289582499,
|
| 76916 |
+
"grad_norm": 0.058236099779605865,
|
| 76917 |
+
"learning_rate": 2.012853002380466e-06,
|
| 76918 |
+
"loss": 1.4466,
|
| 76919 |
+
"step": 21876
|
| 76920 |
+
},
|
| 76921 |
+
{
|
| 76922 |
+
"epoch": 0.9727446534169223,
|
| 76923 |
+
"grad_norm": 0.057283271104097366,
|
| 76924 |
+
"learning_rate": 1.9997700440649302e-06,
|
| 76925 |
+
"loss": 1.4443,
|
| 76926 |
+
"step": 21878
|
| 76927 |
+
},
|
| 76928 |
+
{
|
| 76929 |
+
"epoch": 0.9728335778755947,
|
| 76930 |
+
"grad_norm": 0.057595402002334595,
|
| 76931 |
+
"learning_rate": 1.9867296570036387e-06,
|
| 76932 |
+
"loss": 1.4457,
|
| 76933 |
+
"step": 21880
|
| 76934 |
+
},
|
| 76935 |
+
{
|
| 76936 |
+
"epoch": 0.972922502334267,
|
| 76937 |
+
"grad_norm": 0.05784250423312187,
|
| 76938 |
+
"learning_rate": 1.973731842311366e-06,
|
| 76939 |
+
"loss": 1.4421,
|
| 76940 |
+
"step": 21882
|
| 76941 |
+
},
|
| 76942 |
+
{
|
| 76943 |
+
"epoch": 0.9730114267929394,
|
| 76944 |
+
"grad_norm": 0.057085875421762466,
|
| 76945 |
+
"learning_rate": 1.9607766010991677e-06,
|
| 76946 |
+
"loss": 1.4472,
|
| 76947 |
+
"step": 21884
|
| 76948 |
+
},
|
| 76949 |
+
{
|
| 76950 |
+
"epoch": 0.9731003512516118,
|
| 76951 |
+
"grad_norm": 0.05725502967834473,
|
| 76952 |
+
"learning_rate": 1.9478639344746585e-06,
|
| 76953 |
+
"loss": 1.4512,
|
| 76954 |
+
"step": 21886
|
| 76955 |
+
},
|
| 76956 |
+
{
|
| 76957 |
+
"epoch": 0.9731892757102841,
|
| 76958 |
+
"grad_norm": 0.057032741606235504,
|
| 76959 |
+
"learning_rate": 1.934993843541566e-06,
|
| 76960 |
+
"loss": 1.4412,
|
| 76961 |
+
"step": 21888
|
| 76962 |
+
},
|
| 76963 |
+
{
|
| 76964 |
+
"epoch": 0.9732782001689565,
|
| 76965 |
+
"grad_norm": 0.056812822818756104,
|
| 76966 |
+
"learning_rate": 1.9221663294000657e-06,
|
| 76967 |
+
"loss": 1.4454,
|
| 76968 |
+
"step": 21890
|
| 76969 |
+
},
|
| 76970 |
+
{
|
| 76971 |
+
"epoch": 0.9733671246276289,
|
| 76972 |
+
"grad_norm": 0.057856421917676926,
|
| 76973 |
+
"learning_rate": 1.9093813931467807e-06,
|
| 76974 |
+
"loss": 1.4446,
|
| 76975 |
+
"step": 21892
|
| 76976 |
+
},
|
| 76977 |
+
{
|
| 76978 |
+
"epoch": 0.9734560490863012,
|
| 76979 |
+
"grad_norm": 0.05846314877271652,
|
| 76980 |
+
"learning_rate": 1.8966390358746145e-06,
|
| 76981 |
+
"loss": 1.4441,
|
| 76982 |
+
"step": 21894
|
| 76983 |
+
},
|
| 76984 |
+
{
|
| 76985 |
+
"epoch": 0.9735449735449735,
|
| 76986 |
+
"grad_norm": 0.05669744685292244,
|
| 76987 |
+
"learning_rate": 1.883939258672751e-06,
|
| 76988 |
+
"loss": 1.4504,
|
| 76989 |
+
"step": 21896
|
| 76990 |
+
},
|
| 76991 |
+
{
|
| 76992 |
+
"epoch": 0.9736338980036459,
|
| 76993 |
+
"grad_norm": 0.05733288824558258,
|
| 76994 |
+
"learning_rate": 1.8712820626268778e-06,
|
| 76995 |
+
"loss": 1.453,
|
| 76996 |
+
"step": 21898
|
| 76997 |
+
},
|
| 76998 |
+
{
|
| 76999 |
+
"epoch": 0.9737228224623182,
|
| 77000 |
+
"grad_norm": 0.05750066787004471,
|
| 77001 |
+
"learning_rate": 1.8586674488190736e-06,
|
| 77002 |
+
"loss": 1.4435,
|
| 77003 |
+
"step": 21900
|
| 77004 |
+
},
|
| 77005 |
+
{
|
| 77006 |
+
"epoch": 0.9738117469209906,
|
| 77007 |
+
"grad_norm": 0.05767687410116196,
|
| 77008 |
+
"learning_rate": 1.8460954183275314e-06,
|
| 77009 |
+
"loss": 1.4426,
|
| 77010 |
+
"step": 21902
|
| 77011 |
+
},
|
| 77012 |
+
{
|
| 77013 |
+
"epoch": 0.973900671379663,
|
| 77014 |
+
"grad_norm": 0.05756646394729614,
|
| 77015 |
+
"learning_rate": 1.8335659722271136e-06,
|
| 77016 |
+
"loss": 1.4368,
|
| 77017 |
+
"step": 21904
|
| 77018 |
+
},
|
| 77019 |
+
{
|
| 77020 |
+
"epoch": 0.9739895958383353,
|
| 77021 |
+
"grad_norm": 0.057719554752111435,
|
| 77022 |
+
"learning_rate": 1.8210791115887971e-06,
|
| 77023 |
+
"loss": 1.4422,
|
| 77024 |
+
"step": 21906
|
| 77025 |
+
},
|
| 77026 |
+
{
|
| 77027 |
+
"epoch": 0.9740785202970077,
|
| 77028 |
+
"grad_norm": 0.05758875980973244,
|
| 77029 |
+
"learning_rate": 1.8086348374800055e-06,
|
| 77030 |
+
"loss": 1.4432,
|
| 77031 |
+
"step": 21908
|
| 77032 |
+
},
|
| 77033 |
+
{
|
| 77034 |
+
"epoch": 0.9741674447556801,
|
| 77035 |
+
"grad_norm": 0.0580807626247406,
|
| 77036 |
+
"learning_rate": 1.7962331509646103e-06,
|
| 77037 |
+
"loss": 1.4417,
|
| 77038 |
+
"step": 21910
|
| 77039 |
+
},
|
| 77040 |
+
{
|
| 77041 |
+
"epoch": 0.9742563692143524,
|
| 77042 |
+
"grad_norm": 0.057357918471097946,
|
| 77043 |
+
"learning_rate": 1.7838740531027076e-06,
|
| 77044 |
+
"loss": 1.4442,
|
| 77045 |
+
"step": 21912
|
| 77046 |
+
},
|
| 77047 |
+
{
|
| 77048 |
+
"epoch": 0.9743452936730248,
|
| 77049 |
+
"grad_norm": 0.05980123206973076,
|
| 77050 |
+
"learning_rate": 1.7715575449508413e-06,
|
| 77051 |
+
"loss": 1.4484,
|
| 77052 |
+
"step": 21914
|
| 77053 |
+
},
|
| 77054 |
+
{
|
| 77055 |
+
"epoch": 0.9744342181316972,
|
| 77056 |
+
"grad_norm": 0.05732397362589836,
|
| 77057 |
+
"learning_rate": 1.7592836275618361e-06,
|
| 77058 |
+
"loss": 1.4435,
|
| 77059 |
+
"step": 21916
|
| 77060 |
+
},
|
| 77061 |
+
{
|
| 77062 |
+
"epoch": 0.9745231425903694,
|
| 77063 |
+
"grad_norm": 0.05790344625711441,
|
| 77064 |
+
"learning_rate": 1.7470523019849084e-06,
|
| 77065 |
+
"loss": 1.446,
|
| 77066 |
+
"step": 21918
|
| 77067 |
+
},
|
| 77068 |
+
{
|
| 77069 |
+
"epoch": 0.9746120670490418,
|
| 77070 |
+
"grad_norm": 0.0583200603723526,
|
| 77071 |
+
"learning_rate": 1.734863569265721e-06,
|
| 77072 |
+
"loss": 1.4472,
|
| 77073 |
+
"step": 21920
|
| 77074 |
+
},
|
| 77075 |
+
{
|
| 77076 |
+
"epoch": 0.9747009915077142,
|
| 77077 |
+
"grad_norm": 0.057706188410520554,
|
| 77078 |
+
"learning_rate": 1.7227174304461635e-06,
|
| 77079 |
+
"loss": 1.4404,
|
| 77080 |
+
"step": 21922
|
| 77081 |
+
},
|
| 77082 |
+
{
|
| 77083 |
+
"epoch": 0.9747899159663865,
|
| 77084 |
+
"grad_norm": 0.05923660844564438,
|
| 77085 |
+
"learning_rate": 1.7106138865645716e-06,
|
| 77086 |
+
"loss": 1.4471,
|
| 77087 |
+
"step": 21924
|
| 77088 |
+
},
|
| 77089 |
+
{
|
| 77090 |
+
"epoch": 0.9748788404250589,
|
| 77091 |
+
"grad_norm": 0.057506807148456573,
|
| 77092 |
+
"learning_rate": 1.698552938655562e-06,
|
| 77093 |
+
"loss": 1.4454,
|
| 77094 |
+
"step": 21926
|
| 77095 |
+
},
|
| 77096 |
+
{
|
| 77097 |
+
"epoch": 0.9749677648837313,
|
| 77098 |
+
"grad_norm": 0.057615265250205994,
|
| 77099 |
+
"learning_rate": 1.6865345877502546e-06,
|
| 77100 |
+
"loss": 1.4465,
|
| 77101 |
+
"step": 21928
|
| 77102 |
+
},
|
| 77103 |
+
{
|
| 77104 |
+
"epoch": 0.9750566893424036,
|
| 77105 |
+
"grad_norm": 0.05670270696282387,
|
| 77106 |
+
"learning_rate": 1.6745588348758833e-06,
|
| 77107 |
+
"loss": 1.4418,
|
| 77108 |
+
"step": 21930
|
| 77109 |
+
},
|
| 77110 |
+
{
|
| 77111 |
+
"epoch": 0.975145613801076,
|
| 77112 |
+
"grad_norm": 0.057712372392416,
|
| 77113 |
+
"learning_rate": 1.662625681056351e-06,
|
| 77114 |
+
"loss": 1.4462,
|
| 77115 |
+
"step": 21932
|
| 77116 |
+
},
|
| 77117 |
+
{
|
| 77118 |
+
"epoch": 0.9752345382597484,
|
| 77119 |
+
"grad_norm": 0.05813715234398842,
|
| 77120 |
+
"learning_rate": 1.65073512731162e-06,
|
| 77121 |
+
"loss": 1.446,
|
| 77122 |
+
"step": 21934
|
| 77123 |
+
},
|
| 77124 |
+
{
|
| 77125 |
+
"epoch": 0.9753234627184207,
|
| 77126 |
+
"grad_norm": 0.05662925913929939,
|
| 77127 |
+
"learning_rate": 1.6388871746582102e-06,
|
| 77128 |
+
"loss": 1.4411,
|
| 77129 |
+
"step": 21936
|
| 77130 |
+
},
|
| 77131 |
+
{
|
| 77132 |
+
"epoch": 0.9754123871770931,
|
| 77133 |
+
"grad_norm": 0.056773193180561066,
|
| 77134 |
+
"learning_rate": 1.627081824108978e-06,
|
| 77135 |
+
"loss": 1.4469,
|
| 77136 |
+
"step": 21938
|
| 77137 |
+
},
|
| 77138 |
+
{
|
| 77139 |
+
"epoch": 0.9755013116357654,
|
| 77140 |
+
"grad_norm": 0.057524967938661575,
|
| 77141 |
+
"learning_rate": 1.6153190766730053e-06,
|
| 77142 |
+
"loss": 1.4463,
|
| 77143 |
+
"step": 21940
|
| 77144 |
+
},
|
| 77145 |
+
{
|
| 77146 |
+
"epoch": 0.9755902360944377,
|
| 77147 |
+
"grad_norm": 0.05839113891124725,
|
| 77148 |
+
"learning_rate": 1.6035989333558765e-06,
|
| 77149 |
+
"loss": 1.4537,
|
| 77150 |
+
"step": 21942
|
| 77151 |
+
},
|
| 77152 |
+
{
|
| 77153 |
+
"epoch": 0.9756791605531101,
|
| 77154 |
+
"grad_norm": 0.058153193444013596,
|
| 77155 |
+
"learning_rate": 1.5919213951594569e-06,
|
| 77156 |
+
"loss": 1.4476,
|
| 77157 |
+
"step": 21944
|
| 77158 |
+
},
|
| 77159 |
+
{
|
| 77160 |
+
"epoch": 0.9757680850117825,
|
| 77161 |
+
"grad_norm": 0.05714964494109154,
|
| 77162 |
+
"learning_rate": 1.5802864630820591e-06,
|
| 77163 |
+
"loss": 1.4424,
|
| 77164 |
+
"step": 21946
|
| 77165 |
+
},
|
| 77166 |
+
{
|
| 77167 |
+
"epoch": 0.9758570094704548,
|
| 77168 |
+
"grad_norm": 0.0575794093310833,
|
| 77169 |
+
"learning_rate": 1.5686941381182206e-06,
|
| 77170 |
+
"loss": 1.4466,
|
| 77171 |
+
"step": 21948
|
| 77172 |
+
},
|
| 77173 |
+
{
|
| 77174 |
+
"epoch": 0.9759459339291272,
|
| 77175 |
+
"grad_norm": 0.057474132627248764,
|
| 77176 |
+
"learning_rate": 1.5571444212588715e-06,
|
| 77177 |
+
"loss": 1.4455,
|
| 77178 |
+
"step": 21950
|
| 77179 |
+
},
|
| 77180 |
+
{
|
| 77181 |
+
"epoch": 0.9760348583877996,
|
| 77182 |
+
"grad_norm": 0.05650651827454567,
|
| 77183 |
+
"learning_rate": 1.5456373134914437e-06,
|
| 77184 |
+
"loss": 1.449,
|
| 77185 |
+
"step": 21952
|
| 77186 |
+
},
|
| 77187 |
+
{
|
| 77188 |
+
"epoch": 0.9761237828464719,
|
| 77189 |
+
"grad_norm": 0.05746529996395111,
|
| 77190 |
+
"learning_rate": 1.534172815799484e-06,
|
| 77191 |
+
"loss": 1.4398,
|
| 77192 |
+
"step": 21954
|
| 77193 |
+
},
|
| 77194 |
+
{
|
| 77195 |
+
"epoch": 0.9762127073051443,
|
| 77196 |
+
"grad_norm": 0.05808810144662857,
|
| 77197 |
+
"learning_rate": 1.522750929163208e-06,
|
| 77198 |
+
"loss": 1.439,
|
| 77199 |
+
"step": 21956
|
| 77200 |
+
},
|
| 77201 |
+
{
|
| 77202 |
+
"epoch": 0.9763016317638167,
|
| 77203 |
+
"grad_norm": 0.05777526646852493,
|
| 77204 |
+
"learning_rate": 1.5113716545588352e-06,
|
| 77205 |
+
"loss": 1.4447,
|
| 77206 |
+
"step": 21958
|
| 77207 |
+
},
|
| 77208 |
+
{
|
| 77209 |
+
"epoch": 0.976390556222489,
|
| 77210 |
+
"grad_norm": 0.057585395872592926,
|
| 77211 |
+
"learning_rate": 1.5000349929591982e-06,
|
| 77212 |
+
"loss": 1.4507,
|
| 77213 |
+
"step": 21960
|
| 77214 |
+
},
|
| 77215 |
+
{
|
| 77216 |
+
"epoch": 0.9764794806811613,
|
| 77217 |
+
"grad_norm": 0.05766819417476654,
|
| 77218 |
+
"learning_rate": 1.4887409453333555e-06,
|
| 77219 |
+
"loss": 1.4515,
|
| 77220 |
+
"step": 21962
|
| 77221 |
+
},
|
| 77222 |
+
{
|
| 77223 |
+
"epoch": 0.9765684051398337,
|
| 77224 |
+
"grad_norm": 0.05719029903411865,
|
| 77225 |
+
"learning_rate": 1.4774895126468125e-06,
|
| 77226 |
+
"loss": 1.4468,
|
| 77227 |
+
"step": 21964
|
| 77228 |
+
},
|
| 77229 |
+
{
|
| 77230 |
+
"epoch": 0.976657329598506,
|
| 77231 |
+
"grad_norm": 0.05869666114449501,
|
| 77232 |
+
"learning_rate": 1.4662806958614105e-06,
|
| 77233 |
+
"loss": 1.4479,
|
| 77234 |
+
"step": 21966
|
| 77235 |
+
},
|
| 77236 |
+
{
|
| 77237 |
+
"epoch": 0.9767462540571784,
|
| 77238 |
+
"grad_norm": 0.057645734399557114,
|
| 77239 |
+
"learning_rate": 1.4551144959352724e-06,
|
| 77240 |
+
"loss": 1.4539,
|
| 77241 |
+
"step": 21968
|
| 77242 |
+
},
|
| 77243 |
+
{
|
| 77244 |
+
"epoch": 0.9768351785158508,
|
| 77245 |
+
"grad_norm": 0.058074142783880234,
|
| 77246 |
+
"learning_rate": 1.443990913822968e-06,
|
| 77247 |
+
"loss": 1.4441,
|
| 77248 |
+
"step": 21970
|
| 77249 |
+
},
|
| 77250 |
+
{
|
| 77251 |
+
"epoch": 0.9769241029745231,
|
| 77252 |
+
"grad_norm": 0.057652901858091354,
|
| 77253 |
+
"learning_rate": 1.432909950475403e-06,
|
| 77254 |
+
"loss": 1.4502,
|
| 77255 |
+
"step": 21972
|
| 77256 |
+
},
|
| 77257 |
+
{
|
| 77258 |
+
"epoch": 0.9770130274331955,
|
| 77259 |
+
"grad_norm": 0.056621208786964417,
|
| 77260 |
+
"learning_rate": 1.4218716068398196e-06,
|
| 77261 |
+
"loss": 1.4418,
|
| 77262 |
+
"step": 21974
|
| 77263 |
+
},
|
| 77264 |
+
{
|
| 77265 |
+
"epoch": 0.9771019518918679,
|
| 77266 |
+
"grad_norm": 0.05749661475419998,
|
| 77267 |
+
"learning_rate": 1.4108758838597969e-06,
|
| 77268 |
+
"loss": 1.4433,
|
| 77269 |
+
"step": 21976
|
| 77270 |
+
},
|
| 77271 |
+
{
|
| 77272 |
+
"epoch": 0.9771908763505402,
|
| 77273 |
+
"grad_norm": 0.05827660858631134,
|
| 77274 |
+
"learning_rate": 1.3999227824753047e-06,
|
| 77275 |
+
"loss": 1.4444,
|
| 77276 |
+
"step": 21978
|
| 77277 |
+
},
|
| 77278 |
+
{
|
| 77279 |
+
"epoch": 0.9772798008092126,
|
| 77280 |
+
"grad_norm": 0.057929717004299164,
|
| 77281 |
+
"learning_rate": 1.3890123036227054e-06,
|
| 77282 |
+
"loss": 1.4509,
|
| 77283 |
+
"step": 21980
|
| 77284 |
+
},
|
| 77285 |
+
{
|
| 77286 |
+
"epoch": 0.977368725267885,
|
| 77287 |
+
"grad_norm": 0.058121420443058014,
|
| 77288 |
+
"learning_rate": 1.3781444482345863e-06,
|
| 77289 |
+
"loss": 1.4477,
|
| 77290 |
+
"step": 21982
|
| 77291 |
+
},
|
| 77292 |
+
{
|
| 77293 |
+
"epoch": 0.9774576497265572,
|
| 77294 |
+
"grad_norm": 0.057763341814279556,
|
| 77295 |
+
"learning_rate": 1.3673192172400927e-06,
|
| 77296 |
+
"loss": 1.4427,
|
| 77297 |
+
"step": 21984
|
| 77298 |
+
},
|
| 77299 |
+
{
|
| 77300 |
+
"epoch": 0.9775465741852296,
|
| 77301 |
+
"grad_norm": 0.05874239280819893,
|
| 77302 |
+
"learning_rate": 1.3565366115645405e-06,
|
| 77303 |
+
"loss": 1.4478,
|
| 77304 |
+
"step": 21986
|
| 77305 |
+
},
|
| 77306 |
+
{
|
| 77307 |
+
"epoch": 0.977635498643902,
|
| 77308 |
+
"grad_norm": 0.05708703398704529,
|
| 77309 |
+
"learning_rate": 1.3457966321296921e-06,
|
| 77310 |
+
"loss": 1.4451,
|
| 77311 |
+
"step": 21988
|
| 77312 |
+
},
|
| 77313 |
+
{
|
| 77314 |
+
"epoch": 0.9777244231025743,
|
| 77315 |
+
"grad_norm": 0.05787283182144165,
|
| 77316 |
+
"learning_rate": 1.335099279853591e-06,
|
| 77317 |
+
"loss": 1.4457,
|
| 77318 |
+
"step": 21990
|
| 77319 |
+
},
|
| 77320 |
+
{
|
| 77321 |
+
"epoch": 0.9778133475612467,
|
| 77322 |
+
"grad_norm": 0.05789671465754509,
|
| 77323 |
+
"learning_rate": 1.3244445556507834e-06,
|
| 77324 |
+
"loss": 1.445,
|
| 77325 |
+
"step": 21992
|
| 77326 |
+
},
|
| 77327 |
+
{
|
| 77328 |
+
"epoch": 0.9779022720199191,
|
| 77329 |
+
"grad_norm": 0.0576728992164135,
|
| 77330 |
+
"learning_rate": 1.3138324604320961e-06,
|
| 77331 |
+
"loss": 1.4467,
|
| 77332 |
+
"step": 21994
|
| 77333 |
+
},
|
| 77334 |
+
{
|
| 77335 |
+
"epoch": 0.9779911964785915,
|
| 77336 |
+
"grad_norm": 0.05753236636519432,
|
| 77337 |
+
"learning_rate": 1.3032629951045817e-06,
|
| 77338 |
+
"loss": 1.4464,
|
| 77339 |
+
"step": 21996
|
| 77340 |
+
},
|
| 77341 |
+
{
|
| 77342 |
+
"epoch": 0.9780801209372638,
|
| 77343 |
+
"grad_norm": 0.05750956013798714,
|
| 77344 |
+
"learning_rate": 1.2927361605718502e-06,
|
| 77345 |
+
"loss": 1.4436,
|
| 77346 |
+
"step": 21998
|
| 77347 |
+
},
|
| 77348 |
+
{
|
| 77349 |
+
"epoch": 0.9781690453959362,
|
| 77350 |
+
"grad_norm": 0.05757934972643852,
|
| 77351 |
+
"learning_rate": 1.2822519577337932e-06,
|
| 77352 |
+
"loss": 1.4457,
|
| 77353 |
+
"step": 22000
|
| 77354 |
+
},
|
| 77355 |
+
{
|
| 77356 |
+
"epoch": 0.9781690453959362,
|
| 77357 |
+
"eval_loss": 1.4320533275604248,
|
| 77358 |
+
"eval_runtime": 12.4698,
|
| 77359 |
+
"eval_samples_per_second": 554.141,
|
| 77360 |
+
"eval_steps_per_second": 69.288,
|
| 77361 |
+
"step": 22000
|
| 77362 |
}
|
| 77363 |
],
|
| 77364 |
"logging_steps": 2,
|
|
|
|
| 77378 |
"attributes": {}
|
| 77379 |
}
|
| 77380 |
},
|
| 77381 |
+
"total_flos": 4.707408710467584e+19,
|
| 77382 |
"train_batch_size": 768,
|
| 77383 |
"trial_name": null,
|
| 77384 |
"trial_params": null
|