Uploaded checkpoint-30000
Browse files- model.safetensors +1 -1
- optimizer.pt +1 -1
- rng_state.pth +2 -2
- scheduler.pt +1 -1
- trainer_state.json +3511 -3
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2836579040
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2b2ff417a72c490b70cdd9645bc0fad3d74d5768ff33fe870472376d130cca26
|
| 3 |
size 2836579040
|
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5673376169
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d4feefb8dc1e69e287b0c1c32430c100d0d01e66f215074bcf262dcede980934
|
| 3 |
size 5673376169
|
rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8f6fab5b98c9117992015ae4aa1f5dc6bad25ab862e59672747f31c88ae9efe6
|
| 3 |
+
size 14244
|
scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7def393ab2a8c7ec0497fd0cb247710e6273a07ccc7d3c4a47916ec7e85eb0c0
|
| 3 |
size 1064
|
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 5000,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -17547,6 +17547,3514 @@
|
|
| 17547 |
"eval_samples_per_second": 15.379,
|
| 17548 |
"eval_steps_per_second": 15.379,
|
| 17549 |
"step": 25000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 17550 |
}
|
| 17551 |
],
|
| 17552 |
"logging_steps": 10,
|
|
@@ -17554,7 +21062,7 @@
|
|
| 17554 |
"num_input_tokens_seen": 0,
|
| 17555 |
"num_train_epochs": 1,
|
| 17556 |
"save_steps": 5000,
|
| 17557 |
-
"total_flos": 4.
|
| 17558 |
"train_batch_size": 1,
|
| 17559 |
"trial_name": null,
|
| 17560 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.75,
|
| 5 |
"eval_steps": 5000,
|
| 6 |
+
"global_step": 30000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 17547 |
"eval_samples_per_second": 15.379,
|
| 17548 |
"eval_steps_per_second": 15.379,
|
| 17549 |
"step": 25000
|
| 17550 |
+
},
|
| 17551 |
+
{
|
| 17552 |
+
"epoch": 0.63,
|
| 17553 |
+
"grad_norm": 65.5,
|
| 17554 |
+
"learning_rate": 1.6915254237288136e-07,
|
| 17555 |
+
"loss": 1.1602,
|
| 17556 |
+
"step": 25010
|
| 17557 |
+
},
|
| 17558 |
+
{
|
| 17559 |
+
"epoch": 0.63,
|
| 17560 |
+
"grad_norm": 63.5,
|
| 17561 |
+
"learning_rate": 1.688135593220339e-07,
|
| 17562 |
+
"loss": 1.1774,
|
| 17563 |
+
"step": 25020
|
| 17564 |
+
},
|
| 17565 |
+
{
|
| 17566 |
+
"epoch": 0.63,
|
| 17567 |
+
"grad_norm": 64.5,
|
| 17568 |
+
"learning_rate": 1.6847457627118644e-07,
|
| 17569 |
+
"loss": 1.1692,
|
| 17570 |
+
"step": 25030
|
| 17571 |
+
},
|
| 17572 |
+
{
|
| 17573 |
+
"epoch": 0.63,
|
| 17574 |
+
"grad_norm": 67.5,
|
| 17575 |
+
"learning_rate": 1.68135593220339e-07,
|
| 17576 |
+
"loss": 1.1752,
|
| 17577 |
+
"step": 25040
|
| 17578 |
+
},
|
| 17579 |
+
{
|
| 17580 |
+
"epoch": 0.63,
|
| 17581 |
+
"grad_norm": 65.5,
|
| 17582 |
+
"learning_rate": 1.677966101694915e-07,
|
| 17583 |
+
"loss": 1.1779,
|
| 17584 |
+
"step": 25050
|
| 17585 |
+
},
|
| 17586 |
+
{
|
| 17587 |
+
"epoch": 0.63,
|
| 17588 |
+
"grad_norm": 71.0,
|
| 17589 |
+
"learning_rate": 1.6745762711864405e-07,
|
| 17590 |
+
"loss": 1.1757,
|
| 17591 |
+
"step": 25060
|
| 17592 |
+
},
|
| 17593 |
+
{
|
| 17594 |
+
"epoch": 0.63,
|
| 17595 |
+
"grad_norm": 66.5,
|
| 17596 |
+
"learning_rate": 1.671186440677966e-07,
|
| 17597 |
+
"loss": 1.16,
|
| 17598 |
+
"step": 25070
|
| 17599 |
+
},
|
| 17600 |
+
{
|
| 17601 |
+
"epoch": 0.63,
|
| 17602 |
+
"grad_norm": 63.25,
|
| 17603 |
+
"learning_rate": 1.6677966101694916e-07,
|
| 17604 |
+
"loss": 1.1812,
|
| 17605 |
+
"step": 25080
|
| 17606 |
+
},
|
| 17607 |
+
{
|
| 17608 |
+
"epoch": 0.63,
|
| 17609 |
+
"grad_norm": 66.0,
|
| 17610 |
+
"learning_rate": 1.664406779661017e-07,
|
| 17611 |
+
"loss": 1.1798,
|
| 17612 |
+
"step": 25090
|
| 17613 |
+
},
|
| 17614 |
+
{
|
| 17615 |
+
"epoch": 0.63,
|
| 17616 |
+
"grad_norm": 64.0,
|
| 17617 |
+
"learning_rate": 1.6610169491525421e-07,
|
| 17618 |
+
"loss": 1.1654,
|
| 17619 |
+
"step": 25100
|
| 17620 |
+
},
|
| 17621 |
+
{
|
| 17622 |
+
"epoch": 0.63,
|
| 17623 |
+
"grad_norm": 64.5,
|
| 17624 |
+
"learning_rate": 1.6576271186440677e-07,
|
| 17625 |
+
"loss": 1.1492,
|
| 17626 |
+
"step": 25110
|
| 17627 |
+
},
|
| 17628 |
+
{
|
| 17629 |
+
"epoch": 0.63,
|
| 17630 |
+
"grad_norm": 68.5,
|
| 17631 |
+
"learning_rate": 1.6542372881355932e-07,
|
| 17632 |
+
"loss": 1.1509,
|
| 17633 |
+
"step": 25120
|
| 17634 |
+
},
|
| 17635 |
+
{
|
| 17636 |
+
"epoch": 0.63,
|
| 17637 |
+
"grad_norm": 66.5,
|
| 17638 |
+
"learning_rate": 1.6508474576271188e-07,
|
| 17639 |
+
"loss": 1.151,
|
| 17640 |
+
"step": 25130
|
| 17641 |
+
},
|
| 17642 |
+
{
|
| 17643 |
+
"epoch": 0.63,
|
| 17644 |
+
"grad_norm": 63.75,
|
| 17645 |
+
"learning_rate": 1.647457627118644e-07,
|
| 17646 |
+
"loss": 1.1538,
|
| 17647 |
+
"step": 25140
|
| 17648 |
+
},
|
| 17649 |
+
{
|
| 17650 |
+
"epoch": 0.63,
|
| 17651 |
+
"grad_norm": 63.75,
|
| 17652 |
+
"learning_rate": 1.6440677966101693e-07,
|
| 17653 |
+
"loss": 1.1404,
|
| 17654 |
+
"step": 25150
|
| 17655 |
+
},
|
| 17656 |
+
{
|
| 17657 |
+
"epoch": 0.63,
|
| 17658 |
+
"grad_norm": 65.5,
|
| 17659 |
+
"learning_rate": 1.6406779661016949e-07,
|
| 17660 |
+
"loss": 1.1591,
|
| 17661 |
+
"step": 25160
|
| 17662 |
+
},
|
| 17663 |
+
{
|
| 17664 |
+
"epoch": 0.63,
|
| 17665 |
+
"grad_norm": 63.25,
|
| 17666 |
+
"learning_rate": 1.6372881355932201e-07,
|
| 17667 |
+
"loss": 1.161,
|
| 17668 |
+
"step": 25170
|
| 17669 |
+
},
|
| 17670 |
+
{
|
| 17671 |
+
"epoch": 0.63,
|
| 17672 |
+
"grad_norm": 65.5,
|
| 17673 |
+
"learning_rate": 1.6338983050847457e-07,
|
| 17674 |
+
"loss": 1.1404,
|
| 17675 |
+
"step": 25180
|
| 17676 |
+
},
|
| 17677 |
+
{
|
| 17678 |
+
"epoch": 0.63,
|
| 17679 |
+
"grad_norm": 64.0,
|
| 17680 |
+
"learning_rate": 1.6305084745762712e-07,
|
| 17681 |
+
"loss": 1.1778,
|
| 17682 |
+
"step": 25190
|
| 17683 |
+
},
|
| 17684 |
+
{
|
| 17685 |
+
"epoch": 0.63,
|
| 17686 |
+
"grad_norm": 63.75,
|
| 17687 |
+
"learning_rate": 1.6271186440677965e-07,
|
| 17688 |
+
"loss": 1.1592,
|
| 17689 |
+
"step": 25200
|
| 17690 |
+
},
|
| 17691 |
+
{
|
| 17692 |
+
"epoch": 0.63,
|
| 17693 |
+
"grad_norm": 63.0,
|
| 17694 |
+
"learning_rate": 1.6237288135593218e-07,
|
| 17695 |
+
"loss": 1.1485,
|
| 17696 |
+
"step": 25210
|
| 17697 |
+
},
|
| 17698 |
+
{
|
| 17699 |
+
"epoch": 0.63,
|
| 17700 |
+
"grad_norm": 65.5,
|
| 17701 |
+
"learning_rate": 1.6203389830508473e-07,
|
| 17702 |
+
"loss": 1.1548,
|
| 17703 |
+
"step": 25220
|
| 17704 |
+
},
|
| 17705 |
+
{
|
| 17706 |
+
"epoch": 0.63,
|
| 17707 |
+
"grad_norm": 70.0,
|
| 17708 |
+
"learning_rate": 1.6169491525423729e-07,
|
| 17709 |
+
"loss": 1.1533,
|
| 17710 |
+
"step": 25230
|
| 17711 |
+
},
|
| 17712 |
+
{
|
| 17713 |
+
"epoch": 0.63,
|
| 17714 |
+
"grad_norm": 66.0,
|
| 17715 |
+
"learning_rate": 1.6135593220338984e-07,
|
| 17716 |
+
"loss": 1.1705,
|
| 17717 |
+
"step": 25240
|
| 17718 |
+
},
|
| 17719 |
+
{
|
| 17720 |
+
"epoch": 0.63,
|
| 17721 |
+
"grad_norm": 66.5,
|
| 17722 |
+
"learning_rate": 1.6101694915254234e-07,
|
| 17723 |
+
"loss": 1.1547,
|
| 17724 |
+
"step": 25250
|
| 17725 |
+
},
|
| 17726 |
+
{
|
| 17727 |
+
"epoch": 0.63,
|
| 17728 |
+
"grad_norm": 67.5,
|
| 17729 |
+
"learning_rate": 1.606779661016949e-07,
|
| 17730 |
+
"loss": 1.1725,
|
| 17731 |
+
"step": 25260
|
| 17732 |
+
},
|
| 17733 |
+
{
|
| 17734 |
+
"epoch": 0.63,
|
| 17735 |
+
"grad_norm": 67.0,
|
| 17736 |
+
"learning_rate": 1.6033898305084745e-07,
|
| 17737 |
+
"loss": 1.17,
|
| 17738 |
+
"step": 25270
|
| 17739 |
+
},
|
| 17740 |
+
{
|
| 17741 |
+
"epoch": 0.63,
|
| 17742 |
+
"grad_norm": 65.5,
|
| 17743 |
+
"learning_rate": 1.6e-07,
|
| 17744 |
+
"loss": 1.1567,
|
| 17745 |
+
"step": 25280
|
| 17746 |
+
},
|
| 17747 |
+
{
|
| 17748 |
+
"epoch": 0.63,
|
| 17749 |
+
"grad_norm": 65.0,
|
| 17750 |
+
"learning_rate": 1.5966101694915253e-07,
|
| 17751 |
+
"loss": 1.1564,
|
| 17752 |
+
"step": 25290
|
| 17753 |
+
},
|
| 17754 |
+
{
|
| 17755 |
+
"epoch": 0.63,
|
| 17756 |
+
"grad_norm": 65.5,
|
| 17757 |
+
"learning_rate": 1.5932203389830506e-07,
|
| 17758 |
+
"loss": 1.1429,
|
| 17759 |
+
"step": 25300
|
| 17760 |
+
},
|
| 17761 |
+
{
|
| 17762 |
+
"epoch": 0.63,
|
| 17763 |
+
"grad_norm": 65.0,
|
| 17764 |
+
"learning_rate": 1.589830508474576e-07,
|
| 17765 |
+
"loss": 1.1923,
|
| 17766 |
+
"step": 25310
|
| 17767 |
+
},
|
| 17768 |
+
{
|
| 17769 |
+
"epoch": 0.63,
|
| 17770 |
+
"grad_norm": 66.0,
|
| 17771 |
+
"learning_rate": 1.5864406779661017e-07,
|
| 17772 |
+
"loss": 1.1442,
|
| 17773 |
+
"step": 25320
|
| 17774 |
+
},
|
| 17775 |
+
{
|
| 17776 |
+
"epoch": 0.63,
|
| 17777 |
+
"grad_norm": 65.0,
|
| 17778 |
+
"learning_rate": 1.583050847457627e-07,
|
| 17779 |
+
"loss": 1.1426,
|
| 17780 |
+
"step": 25330
|
| 17781 |
+
},
|
| 17782 |
+
{
|
| 17783 |
+
"epoch": 0.63,
|
| 17784 |
+
"grad_norm": 63.25,
|
| 17785 |
+
"learning_rate": 1.5796610169491525e-07,
|
| 17786 |
+
"loss": 1.1524,
|
| 17787 |
+
"step": 25340
|
| 17788 |
+
},
|
| 17789 |
+
{
|
| 17790 |
+
"epoch": 0.63,
|
| 17791 |
+
"grad_norm": 62.75,
|
| 17792 |
+
"learning_rate": 1.576271186440678e-07,
|
| 17793 |
+
"loss": 1.1665,
|
| 17794 |
+
"step": 25350
|
| 17795 |
+
},
|
| 17796 |
+
{
|
| 17797 |
+
"epoch": 0.63,
|
| 17798 |
+
"grad_norm": 67.0,
|
| 17799 |
+
"learning_rate": 1.5728813559322033e-07,
|
| 17800 |
+
"loss": 1.1656,
|
| 17801 |
+
"step": 25360
|
| 17802 |
+
},
|
| 17803 |
+
{
|
| 17804 |
+
"epoch": 0.63,
|
| 17805 |
+
"grad_norm": 65.0,
|
| 17806 |
+
"learning_rate": 1.5694915254237286e-07,
|
| 17807 |
+
"loss": 1.1632,
|
| 17808 |
+
"step": 25370
|
| 17809 |
+
},
|
| 17810 |
+
{
|
| 17811 |
+
"epoch": 0.63,
|
| 17812 |
+
"grad_norm": 65.5,
|
| 17813 |
+
"learning_rate": 1.566101694915254e-07,
|
| 17814 |
+
"loss": 1.1635,
|
| 17815 |
+
"step": 25380
|
| 17816 |
+
},
|
| 17817 |
+
{
|
| 17818 |
+
"epoch": 0.63,
|
| 17819 |
+
"grad_norm": 67.5,
|
| 17820 |
+
"learning_rate": 1.5627118644067797e-07,
|
| 17821 |
+
"loss": 1.1564,
|
| 17822 |
+
"step": 25390
|
| 17823 |
+
},
|
| 17824 |
+
{
|
| 17825 |
+
"epoch": 0.64,
|
| 17826 |
+
"grad_norm": 66.0,
|
| 17827 |
+
"learning_rate": 1.5593220338983052e-07,
|
| 17828 |
+
"loss": 1.1746,
|
| 17829 |
+
"step": 25400
|
| 17830 |
+
},
|
| 17831 |
+
{
|
| 17832 |
+
"epoch": 0.64,
|
| 17833 |
+
"grad_norm": 64.5,
|
| 17834 |
+
"learning_rate": 1.5559322033898302e-07,
|
| 17835 |
+
"loss": 1.1615,
|
| 17836 |
+
"step": 25410
|
| 17837 |
+
},
|
| 17838 |
+
{
|
| 17839 |
+
"epoch": 0.64,
|
| 17840 |
+
"grad_norm": 64.0,
|
| 17841 |
+
"learning_rate": 1.5525423728813558e-07,
|
| 17842 |
+
"loss": 1.1514,
|
| 17843 |
+
"step": 25420
|
| 17844 |
+
},
|
| 17845 |
+
{
|
| 17846 |
+
"epoch": 0.64,
|
| 17847 |
+
"grad_norm": 67.0,
|
| 17848 |
+
"learning_rate": 1.5491525423728813e-07,
|
| 17849 |
+
"loss": 1.1442,
|
| 17850 |
+
"step": 25430
|
| 17851 |
+
},
|
| 17852 |
+
{
|
| 17853 |
+
"epoch": 0.64,
|
| 17854 |
+
"grad_norm": 64.5,
|
| 17855 |
+
"learning_rate": 1.5457627118644068e-07,
|
| 17856 |
+
"loss": 1.1624,
|
| 17857 |
+
"step": 25440
|
| 17858 |
+
},
|
| 17859 |
+
{
|
| 17860 |
+
"epoch": 0.64,
|
| 17861 |
+
"grad_norm": 65.5,
|
| 17862 |
+
"learning_rate": 1.542372881355932e-07,
|
| 17863 |
+
"loss": 1.1658,
|
| 17864 |
+
"step": 25450
|
| 17865 |
+
},
|
| 17866 |
+
{
|
| 17867 |
+
"epoch": 0.64,
|
| 17868 |
+
"grad_norm": 65.0,
|
| 17869 |
+
"learning_rate": 1.5389830508474574e-07,
|
| 17870 |
+
"loss": 1.1648,
|
| 17871 |
+
"step": 25460
|
| 17872 |
+
},
|
| 17873 |
+
{
|
| 17874 |
+
"epoch": 0.64,
|
| 17875 |
+
"grad_norm": 66.0,
|
| 17876 |
+
"learning_rate": 1.535593220338983e-07,
|
| 17877 |
+
"loss": 1.163,
|
| 17878 |
+
"step": 25470
|
| 17879 |
+
},
|
| 17880 |
+
{
|
| 17881 |
+
"epoch": 0.64,
|
| 17882 |
+
"grad_norm": 66.5,
|
| 17883 |
+
"learning_rate": 1.5322033898305085e-07,
|
| 17884 |
+
"loss": 1.1667,
|
| 17885 |
+
"step": 25480
|
| 17886 |
+
},
|
| 17887 |
+
{
|
| 17888 |
+
"epoch": 0.64,
|
| 17889 |
+
"grad_norm": 65.0,
|
| 17890 |
+
"learning_rate": 1.5288135593220338e-07,
|
| 17891 |
+
"loss": 1.1479,
|
| 17892 |
+
"step": 25490
|
| 17893 |
+
},
|
| 17894 |
+
{
|
| 17895 |
+
"epoch": 0.64,
|
| 17896 |
+
"grad_norm": 63.25,
|
| 17897 |
+
"learning_rate": 1.5254237288135593e-07,
|
| 17898 |
+
"loss": 1.1403,
|
| 17899 |
+
"step": 25500
|
| 17900 |
+
},
|
| 17901 |
+
{
|
| 17902 |
+
"epoch": 0.64,
|
| 17903 |
+
"grad_norm": 64.5,
|
| 17904 |
+
"learning_rate": 1.5220338983050846e-07,
|
| 17905 |
+
"loss": 1.1393,
|
| 17906 |
+
"step": 25510
|
| 17907 |
+
},
|
| 17908 |
+
{
|
| 17909 |
+
"epoch": 0.64,
|
| 17910 |
+
"grad_norm": 63.25,
|
| 17911 |
+
"learning_rate": 1.51864406779661e-07,
|
| 17912 |
+
"loss": 1.1531,
|
| 17913 |
+
"step": 25520
|
| 17914 |
+
},
|
| 17915 |
+
{
|
| 17916 |
+
"epoch": 0.64,
|
| 17917 |
+
"grad_norm": 68.0,
|
| 17918 |
+
"learning_rate": 1.5152542372881354e-07,
|
| 17919 |
+
"loss": 1.1799,
|
| 17920 |
+
"step": 25530
|
| 17921 |
+
},
|
| 17922 |
+
{
|
| 17923 |
+
"epoch": 0.64,
|
| 17924 |
+
"grad_norm": 64.0,
|
| 17925 |
+
"learning_rate": 1.511864406779661e-07,
|
| 17926 |
+
"loss": 1.17,
|
| 17927 |
+
"step": 25540
|
| 17928 |
+
},
|
| 17929 |
+
{
|
| 17930 |
+
"epoch": 0.64,
|
| 17931 |
+
"grad_norm": 67.5,
|
| 17932 |
+
"learning_rate": 1.5084745762711865e-07,
|
| 17933 |
+
"loss": 1.1592,
|
| 17934 |
+
"step": 25550
|
| 17935 |
+
},
|
| 17936 |
+
{
|
| 17937 |
+
"epoch": 0.64,
|
| 17938 |
+
"grad_norm": 65.5,
|
| 17939 |
+
"learning_rate": 1.5050847457627118e-07,
|
| 17940 |
+
"loss": 1.168,
|
| 17941 |
+
"step": 25560
|
| 17942 |
+
},
|
| 17943 |
+
{
|
| 17944 |
+
"epoch": 0.64,
|
| 17945 |
+
"grad_norm": 68.5,
|
| 17946 |
+
"learning_rate": 1.501694915254237e-07,
|
| 17947 |
+
"loss": 1.1706,
|
| 17948 |
+
"step": 25570
|
| 17949 |
+
},
|
| 17950 |
+
{
|
| 17951 |
+
"epoch": 0.64,
|
| 17952 |
+
"grad_norm": 67.0,
|
| 17953 |
+
"learning_rate": 1.4983050847457626e-07,
|
| 17954 |
+
"loss": 1.1594,
|
| 17955 |
+
"step": 25580
|
| 17956 |
+
},
|
| 17957 |
+
{
|
| 17958 |
+
"epoch": 0.64,
|
| 17959 |
+
"grad_norm": 63.0,
|
| 17960 |
+
"learning_rate": 1.494915254237288e-07,
|
| 17961 |
+
"loss": 1.1591,
|
| 17962 |
+
"step": 25590
|
| 17963 |
+
},
|
| 17964 |
+
{
|
| 17965 |
+
"epoch": 0.64,
|
| 17966 |
+
"grad_norm": 65.0,
|
| 17967 |
+
"learning_rate": 1.4915254237288137e-07,
|
| 17968 |
+
"loss": 1.158,
|
| 17969 |
+
"step": 25600
|
| 17970 |
+
},
|
| 17971 |
+
{
|
| 17972 |
+
"epoch": 0.64,
|
| 17973 |
+
"grad_norm": 65.0,
|
| 17974 |
+
"learning_rate": 1.488135593220339e-07,
|
| 17975 |
+
"loss": 1.164,
|
| 17976 |
+
"step": 25610
|
| 17977 |
+
},
|
| 17978 |
+
{
|
| 17979 |
+
"epoch": 0.64,
|
| 17980 |
+
"grad_norm": 65.5,
|
| 17981 |
+
"learning_rate": 1.4847457627118642e-07,
|
| 17982 |
+
"loss": 1.1634,
|
| 17983 |
+
"step": 25620
|
| 17984 |
+
},
|
| 17985 |
+
{
|
| 17986 |
+
"epoch": 0.64,
|
| 17987 |
+
"grad_norm": 64.5,
|
| 17988 |
+
"learning_rate": 1.4813559322033897e-07,
|
| 17989 |
+
"loss": 1.1551,
|
| 17990 |
+
"step": 25630
|
| 17991 |
+
},
|
| 17992 |
+
{
|
| 17993 |
+
"epoch": 0.64,
|
| 17994 |
+
"grad_norm": 65.5,
|
| 17995 |
+
"learning_rate": 1.4779661016949153e-07,
|
| 17996 |
+
"loss": 1.1518,
|
| 17997 |
+
"step": 25640
|
| 17998 |
+
},
|
| 17999 |
+
{
|
| 18000 |
+
"epoch": 0.64,
|
| 18001 |
+
"grad_norm": 66.0,
|
| 18002 |
+
"learning_rate": 1.4745762711864406e-07,
|
| 18003 |
+
"loss": 1.1638,
|
| 18004 |
+
"step": 25650
|
| 18005 |
+
},
|
| 18006 |
+
{
|
| 18007 |
+
"epoch": 0.64,
|
| 18008 |
+
"grad_norm": 67.5,
|
| 18009 |
+
"learning_rate": 1.471186440677966e-07,
|
| 18010 |
+
"loss": 1.176,
|
| 18011 |
+
"step": 25660
|
| 18012 |
+
},
|
| 18013 |
+
{
|
| 18014 |
+
"epoch": 0.64,
|
| 18015 |
+
"grad_norm": 67.0,
|
| 18016 |
+
"learning_rate": 1.4677966101694914e-07,
|
| 18017 |
+
"loss": 1.1748,
|
| 18018 |
+
"step": 25670
|
| 18019 |
+
},
|
| 18020 |
+
{
|
| 18021 |
+
"epoch": 0.64,
|
| 18022 |
+
"grad_norm": 68.0,
|
| 18023 |
+
"learning_rate": 1.464406779661017e-07,
|
| 18024 |
+
"loss": 1.1563,
|
| 18025 |
+
"step": 25680
|
| 18026 |
+
},
|
| 18027 |
+
{
|
| 18028 |
+
"epoch": 0.64,
|
| 18029 |
+
"grad_norm": 66.5,
|
| 18030 |
+
"learning_rate": 1.4610169491525422e-07,
|
| 18031 |
+
"loss": 1.1587,
|
| 18032 |
+
"step": 25690
|
| 18033 |
+
},
|
| 18034 |
+
{
|
| 18035 |
+
"epoch": 0.64,
|
| 18036 |
+
"grad_norm": 69.0,
|
| 18037 |
+
"learning_rate": 1.4576271186440677e-07,
|
| 18038 |
+
"loss": 1.1564,
|
| 18039 |
+
"step": 25700
|
| 18040 |
+
},
|
| 18041 |
+
{
|
| 18042 |
+
"epoch": 0.64,
|
| 18043 |
+
"grad_norm": 68.0,
|
| 18044 |
+
"learning_rate": 1.4542372881355933e-07,
|
| 18045 |
+
"loss": 1.1485,
|
| 18046 |
+
"step": 25710
|
| 18047 |
+
},
|
| 18048 |
+
{
|
| 18049 |
+
"epoch": 0.64,
|
| 18050 |
+
"grad_norm": 67.5,
|
| 18051 |
+
"learning_rate": 1.4508474576271186e-07,
|
| 18052 |
+
"loss": 1.1504,
|
| 18053 |
+
"step": 25720
|
| 18054 |
+
},
|
| 18055 |
+
{
|
| 18056 |
+
"epoch": 0.64,
|
| 18057 |
+
"grad_norm": 64.0,
|
| 18058 |
+
"learning_rate": 1.4474576271186438e-07,
|
| 18059 |
+
"loss": 1.1607,
|
| 18060 |
+
"step": 25730
|
| 18061 |
+
},
|
| 18062 |
+
{
|
| 18063 |
+
"epoch": 0.64,
|
| 18064 |
+
"grad_norm": 66.0,
|
| 18065 |
+
"learning_rate": 1.4440677966101694e-07,
|
| 18066 |
+
"loss": 1.1671,
|
| 18067 |
+
"step": 25740
|
| 18068 |
+
},
|
| 18069 |
+
{
|
| 18070 |
+
"epoch": 0.64,
|
| 18071 |
+
"grad_norm": 65.0,
|
| 18072 |
+
"learning_rate": 1.440677966101695e-07,
|
| 18073 |
+
"loss": 1.16,
|
| 18074 |
+
"step": 25750
|
| 18075 |
+
},
|
| 18076 |
+
{
|
| 18077 |
+
"epoch": 0.64,
|
| 18078 |
+
"grad_norm": 66.0,
|
| 18079 |
+
"learning_rate": 1.4372881355932205e-07,
|
| 18080 |
+
"loss": 1.1692,
|
| 18081 |
+
"step": 25760
|
| 18082 |
+
},
|
| 18083 |
+
{
|
| 18084 |
+
"epoch": 0.64,
|
| 18085 |
+
"grad_norm": 68.0,
|
| 18086 |
+
"learning_rate": 1.4338983050847455e-07,
|
| 18087 |
+
"loss": 1.1652,
|
| 18088 |
+
"step": 25770
|
| 18089 |
+
},
|
| 18090 |
+
{
|
| 18091 |
+
"epoch": 0.64,
|
| 18092 |
+
"grad_norm": 66.0,
|
| 18093 |
+
"learning_rate": 1.430508474576271e-07,
|
| 18094 |
+
"loss": 1.1545,
|
| 18095 |
+
"step": 25780
|
| 18096 |
+
},
|
| 18097 |
+
{
|
| 18098 |
+
"epoch": 0.64,
|
| 18099 |
+
"grad_norm": 64.5,
|
| 18100 |
+
"learning_rate": 1.4271186440677966e-07,
|
| 18101 |
+
"loss": 1.1608,
|
| 18102 |
+
"step": 25790
|
| 18103 |
+
},
|
| 18104 |
+
{
|
| 18105 |
+
"epoch": 0.65,
|
| 18106 |
+
"grad_norm": 64.0,
|
| 18107 |
+
"learning_rate": 1.423728813559322e-07,
|
| 18108 |
+
"loss": 1.151,
|
| 18109 |
+
"step": 25800
|
| 18110 |
+
},
|
| 18111 |
+
{
|
| 18112 |
+
"epoch": 0.65,
|
| 18113 |
+
"grad_norm": 65.0,
|
| 18114 |
+
"learning_rate": 1.4203389830508474e-07,
|
| 18115 |
+
"loss": 1.1878,
|
| 18116 |
+
"step": 25810
|
| 18117 |
+
},
|
| 18118 |
+
{
|
| 18119 |
+
"epoch": 0.65,
|
| 18120 |
+
"grad_norm": 67.0,
|
| 18121 |
+
"learning_rate": 1.4169491525423727e-07,
|
| 18122 |
+
"loss": 1.169,
|
| 18123 |
+
"step": 25820
|
| 18124 |
+
},
|
| 18125 |
+
{
|
| 18126 |
+
"epoch": 0.65,
|
| 18127 |
+
"grad_norm": 66.5,
|
| 18128 |
+
"learning_rate": 1.4135593220338982e-07,
|
| 18129 |
+
"loss": 1.1569,
|
| 18130 |
+
"step": 25830
|
| 18131 |
+
},
|
| 18132 |
+
{
|
| 18133 |
+
"epoch": 0.65,
|
| 18134 |
+
"grad_norm": 70.0,
|
| 18135 |
+
"learning_rate": 1.4101694915254237e-07,
|
| 18136 |
+
"loss": 1.1676,
|
| 18137 |
+
"step": 25840
|
| 18138 |
+
},
|
| 18139 |
+
{
|
| 18140 |
+
"epoch": 0.65,
|
| 18141 |
+
"grad_norm": 64.5,
|
| 18142 |
+
"learning_rate": 1.406779661016949e-07,
|
| 18143 |
+
"loss": 1.1609,
|
| 18144 |
+
"step": 25850
|
| 18145 |
+
},
|
| 18146 |
+
{
|
| 18147 |
+
"epoch": 0.65,
|
| 18148 |
+
"grad_norm": 65.0,
|
| 18149 |
+
"learning_rate": 1.4033898305084746e-07,
|
| 18150 |
+
"loss": 1.1647,
|
| 18151 |
+
"step": 25860
|
| 18152 |
+
},
|
| 18153 |
+
{
|
| 18154 |
+
"epoch": 0.65,
|
| 18155 |
+
"grad_norm": 64.5,
|
| 18156 |
+
"learning_rate": 1.4e-07,
|
| 18157 |
+
"loss": 1.1587,
|
| 18158 |
+
"step": 25870
|
| 18159 |
+
},
|
| 18160 |
+
{
|
| 18161 |
+
"epoch": 0.65,
|
| 18162 |
+
"grad_norm": 63.5,
|
| 18163 |
+
"learning_rate": 1.3966101694915254e-07,
|
| 18164 |
+
"loss": 1.1649,
|
| 18165 |
+
"step": 25880
|
| 18166 |
+
},
|
| 18167 |
+
{
|
| 18168 |
+
"epoch": 0.65,
|
| 18169 |
+
"grad_norm": 64.0,
|
| 18170 |
+
"learning_rate": 1.3932203389830506e-07,
|
| 18171 |
+
"loss": 1.1553,
|
| 18172 |
+
"step": 25890
|
| 18173 |
+
},
|
| 18174 |
+
{
|
| 18175 |
+
"epoch": 0.65,
|
| 18176 |
+
"grad_norm": 64.5,
|
| 18177 |
+
"learning_rate": 1.3898305084745762e-07,
|
| 18178 |
+
"loss": 1.1602,
|
| 18179 |
+
"step": 25900
|
| 18180 |
+
},
|
| 18181 |
+
{
|
| 18182 |
+
"epoch": 0.65,
|
| 18183 |
+
"grad_norm": 67.5,
|
| 18184 |
+
"learning_rate": 1.3864406779661017e-07,
|
| 18185 |
+
"loss": 1.1675,
|
| 18186 |
+
"step": 25910
|
| 18187 |
+
},
|
| 18188 |
+
{
|
| 18189 |
+
"epoch": 0.65,
|
| 18190 |
+
"grad_norm": 66.0,
|
| 18191 |
+
"learning_rate": 1.3830508474576273e-07,
|
| 18192 |
+
"loss": 1.1869,
|
| 18193 |
+
"step": 25920
|
| 18194 |
+
},
|
| 18195 |
+
{
|
| 18196 |
+
"epoch": 0.65,
|
| 18197 |
+
"grad_norm": 66.0,
|
| 18198 |
+
"learning_rate": 1.3796610169491523e-07,
|
| 18199 |
+
"loss": 1.1519,
|
| 18200 |
+
"step": 25930
|
| 18201 |
+
},
|
| 18202 |
+
{
|
| 18203 |
+
"epoch": 0.65,
|
| 18204 |
+
"grad_norm": 64.0,
|
| 18205 |
+
"learning_rate": 1.3762711864406778e-07,
|
| 18206 |
+
"loss": 1.1623,
|
| 18207 |
+
"step": 25940
|
| 18208 |
+
},
|
| 18209 |
+
{
|
| 18210 |
+
"epoch": 0.65,
|
| 18211 |
+
"grad_norm": 63.25,
|
| 18212 |
+
"learning_rate": 1.3728813559322034e-07,
|
| 18213 |
+
"loss": 1.1409,
|
| 18214 |
+
"step": 25950
|
| 18215 |
+
},
|
| 18216 |
+
{
|
| 18217 |
+
"epoch": 0.65,
|
| 18218 |
+
"grad_norm": 63.5,
|
| 18219 |
+
"learning_rate": 1.369491525423729e-07,
|
| 18220 |
+
"loss": 1.1447,
|
| 18221 |
+
"step": 25960
|
| 18222 |
+
},
|
| 18223 |
+
{
|
| 18224 |
+
"epoch": 0.65,
|
| 18225 |
+
"grad_norm": 64.5,
|
| 18226 |
+
"learning_rate": 1.3661016949152542e-07,
|
| 18227 |
+
"loss": 1.1648,
|
| 18228 |
+
"step": 25970
|
| 18229 |
+
},
|
| 18230 |
+
{
|
| 18231 |
+
"epoch": 0.65,
|
| 18232 |
+
"grad_norm": 66.5,
|
| 18233 |
+
"learning_rate": 1.3627118644067795e-07,
|
| 18234 |
+
"loss": 1.1488,
|
| 18235 |
+
"step": 25980
|
| 18236 |
+
},
|
| 18237 |
+
{
|
| 18238 |
+
"epoch": 0.65,
|
| 18239 |
+
"grad_norm": 64.0,
|
| 18240 |
+
"learning_rate": 1.359322033898305e-07,
|
| 18241 |
+
"loss": 1.1572,
|
| 18242 |
+
"step": 25990
|
| 18243 |
+
},
|
| 18244 |
+
{
|
| 18245 |
+
"epoch": 0.65,
|
| 18246 |
+
"grad_norm": 64.5,
|
| 18247 |
+
"learning_rate": 1.3559322033898305e-07,
|
| 18248 |
+
"loss": 1.1588,
|
| 18249 |
+
"step": 26000
|
| 18250 |
+
},
|
| 18251 |
+
{
|
| 18252 |
+
"epoch": 0.65,
|
| 18253 |
+
"grad_norm": 64.0,
|
| 18254 |
+
"learning_rate": 1.3525423728813558e-07,
|
| 18255 |
+
"loss": 1.1545,
|
| 18256 |
+
"step": 26010
|
| 18257 |
+
},
|
| 18258 |
+
{
|
| 18259 |
+
"epoch": 0.65,
|
| 18260 |
+
"grad_norm": 66.5,
|
| 18261 |
+
"learning_rate": 1.3491525423728814e-07,
|
| 18262 |
+
"loss": 1.152,
|
| 18263 |
+
"step": 26020
|
| 18264 |
+
},
|
| 18265 |
+
{
|
| 18266 |
+
"epoch": 0.65,
|
| 18267 |
+
"grad_norm": 68.5,
|
| 18268 |
+
"learning_rate": 1.3457627118644066e-07,
|
| 18269 |
+
"loss": 1.1726,
|
| 18270 |
+
"step": 26030
|
| 18271 |
+
},
|
| 18272 |
+
{
|
| 18273 |
+
"epoch": 0.65,
|
| 18274 |
+
"grad_norm": 65.5,
|
| 18275 |
+
"learning_rate": 1.3423728813559322e-07,
|
| 18276 |
+
"loss": 1.1567,
|
| 18277 |
+
"step": 26040
|
| 18278 |
+
},
|
| 18279 |
+
{
|
| 18280 |
+
"epoch": 0.65,
|
| 18281 |
+
"grad_norm": 65.0,
|
| 18282 |
+
"learning_rate": 1.3389830508474575e-07,
|
| 18283 |
+
"loss": 1.1695,
|
| 18284 |
+
"step": 26050
|
| 18285 |
+
},
|
| 18286 |
+
{
|
| 18287 |
+
"epoch": 0.65,
|
| 18288 |
+
"grad_norm": 64.0,
|
| 18289 |
+
"learning_rate": 1.335593220338983e-07,
|
| 18290 |
+
"loss": 1.1547,
|
| 18291 |
+
"step": 26060
|
| 18292 |
+
},
|
| 18293 |
+
{
|
| 18294 |
+
"epoch": 0.65,
|
| 18295 |
+
"grad_norm": 66.5,
|
| 18296 |
+
"learning_rate": 1.3322033898305085e-07,
|
| 18297 |
+
"loss": 1.1592,
|
| 18298 |
+
"step": 26070
|
| 18299 |
+
},
|
| 18300 |
+
{
|
| 18301 |
+
"epoch": 0.65,
|
| 18302 |
+
"grad_norm": 67.5,
|
| 18303 |
+
"learning_rate": 1.3288135593220338e-07,
|
| 18304 |
+
"loss": 1.1573,
|
| 18305 |
+
"step": 26080
|
| 18306 |
+
},
|
| 18307 |
+
{
|
| 18308 |
+
"epoch": 0.65,
|
| 18309 |
+
"grad_norm": 67.0,
|
| 18310 |
+
"learning_rate": 1.325423728813559e-07,
|
| 18311 |
+
"loss": 1.1546,
|
| 18312 |
+
"step": 26090
|
| 18313 |
+
},
|
| 18314 |
+
{
|
| 18315 |
+
"epoch": 0.65,
|
| 18316 |
+
"grad_norm": 67.5,
|
| 18317 |
+
"learning_rate": 1.3220338983050846e-07,
|
| 18318 |
+
"loss": 1.1682,
|
| 18319 |
+
"step": 26100
|
| 18320 |
+
},
|
| 18321 |
+
{
|
| 18322 |
+
"epoch": 0.65,
|
| 18323 |
+
"grad_norm": 67.0,
|
| 18324 |
+
"learning_rate": 1.3186440677966102e-07,
|
| 18325 |
+
"loss": 1.1654,
|
| 18326 |
+
"step": 26110
|
| 18327 |
+
},
|
| 18328 |
+
{
|
| 18329 |
+
"epoch": 0.65,
|
| 18330 |
+
"grad_norm": 69.5,
|
| 18331 |
+
"learning_rate": 1.3152542372881357e-07,
|
| 18332 |
+
"loss": 1.1477,
|
| 18333 |
+
"step": 26120
|
| 18334 |
+
},
|
| 18335 |
+
{
|
| 18336 |
+
"epoch": 0.65,
|
| 18337 |
+
"grad_norm": 65.0,
|
| 18338 |
+
"learning_rate": 1.311864406779661e-07,
|
| 18339 |
+
"loss": 1.1548,
|
| 18340 |
+
"step": 26130
|
| 18341 |
+
},
|
| 18342 |
+
{
|
| 18343 |
+
"epoch": 0.65,
|
| 18344 |
+
"grad_norm": 64.5,
|
| 18345 |
+
"learning_rate": 1.3084745762711863e-07,
|
| 18346 |
+
"loss": 1.1542,
|
| 18347 |
+
"step": 26140
|
| 18348 |
+
},
|
| 18349 |
+
{
|
| 18350 |
+
"epoch": 0.65,
|
| 18351 |
+
"grad_norm": 65.5,
|
| 18352 |
+
"learning_rate": 1.3050847457627118e-07,
|
| 18353 |
+
"loss": 1.1593,
|
| 18354 |
+
"step": 26150
|
| 18355 |
+
},
|
| 18356 |
+
{
|
| 18357 |
+
"epoch": 0.65,
|
| 18358 |
+
"grad_norm": 65.0,
|
| 18359 |
+
"learning_rate": 1.3016949152542374e-07,
|
| 18360 |
+
"loss": 1.1389,
|
| 18361 |
+
"step": 26160
|
| 18362 |
+
},
|
| 18363 |
+
{
|
| 18364 |
+
"epoch": 0.65,
|
| 18365 |
+
"grad_norm": 65.0,
|
| 18366 |
+
"learning_rate": 1.2983050847457626e-07,
|
| 18367 |
+
"loss": 1.1708,
|
| 18368 |
+
"step": 26170
|
| 18369 |
+
},
|
| 18370 |
+
{
|
| 18371 |
+
"epoch": 0.65,
|
| 18372 |
+
"grad_norm": 63.0,
|
| 18373 |
+
"learning_rate": 1.2949152542372882e-07,
|
| 18374 |
+
"loss": 1.1391,
|
| 18375 |
+
"step": 26180
|
| 18376 |
+
},
|
| 18377 |
+
{
|
| 18378 |
+
"epoch": 0.65,
|
| 18379 |
+
"grad_norm": 65.5,
|
| 18380 |
+
"learning_rate": 1.2915254237288135e-07,
|
| 18381 |
+
"loss": 1.1426,
|
| 18382 |
+
"step": 26190
|
| 18383 |
+
},
|
| 18384 |
+
{
|
| 18385 |
+
"epoch": 0.66,
|
| 18386 |
+
"grad_norm": 65.5,
|
| 18387 |
+
"learning_rate": 1.288135593220339e-07,
|
| 18388 |
+
"loss": 1.156,
|
| 18389 |
+
"step": 26200
|
| 18390 |
+
},
|
| 18391 |
+
{
|
| 18392 |
+
"epoch": 0.66,
|
| 18393 |
+
"grad_norm": 65.5,
|
| 18394 |
+
"learning_rate": 1.2847457627118643e-07,
|
| 18395 |
+
"loss": 1.1458,
|
| 18396 |
+
"step": 26210
|
| 18397 |
+
},
|
| 18398 |
+
{
|
| 18399 |
+
"epoch": 0.66,
|
| 18400 |
+
"grad_norm": 64.0,
|
| 18401 |
+
"learning_rate": 1.2813559322033898e-07,
|
| 18402 |
+
"loss": 1.1718,
|
| 18403 |
+
"step": 26220
|
| 18404 |
+
},
|
| 18405 |
+
{
|
| 18406 |
+
"epoch": 0.66,
|
| 18407 |
+
"grad_norm": 70.5,
|
| 18408 |
+
"learning_rate": 1.2779661016949154e-07,
|
| 18409 |
+
"loss": 1.1524,
|
| 18410 |
+
"step": 26230
|
| 18411 |
+
},
|
| 18412 |
+
{
|
| 18413 |
+
"epoch": 0.66,
|
| 18414 |
+
"grad_norm": 65.0,
|
| 18415 |
+
"learning_rate": 1.2745762711864406e-07,
|
| 18416 |
+
"loss": 1.1687,
|
| 18417 |
+
"step": 26240
|
| 18418 |
+
},
|
| 18419 |
+
{
|
| 18420 |
+
"epoch": 0.66,
|
| 18421 |
+
"grad_norm": 68.0,
|
| 18422 |
+
"learning_rate": 1.271186440677966e-07,
|
| 18423 |
+
"loss": 1.186,
|
| 18424 |
+
"step": 26250
|
| 18425 |
+
},
|
| 18426 |
+
{
|
| 18427 |
+
"epoch": 0.66,
|
| 18428 |
+
"grad_norm": 66.0,
|
| 18429 |
+
"learning_rate": 1.2677966101694914e-07,
|
| 18430 |
+
"loss": 1.16,
|
| 18431 |
+
"step": 26260
|
| 18432 |
+
},
|
| 18433 |
+
{
|
| 18434 |
+
"epoch": 0.66,
|
| 18435 |
+
"grad_norm": 64.0,
|
| 18436 |
+
"learning_rate": 1.264406779661017e-07,
|
| 18437 |
+
"loss": 1.1376,
|
| 18438 |
+
"step": 26270
|
| 18439 |
+
},
|
| 18440 |
+
{
|
| 18441 |
+
"epoch": 0.66,
|
| 18442 |
+
"grad_norm": 68.5,
|
| 18443 |
+
"learning_rate": 1.2610169491525425e-07,
|
| 18444 |
+
"loss": 1.1513,
|
| 18445 |
+
"step": 26280
|
| 18446 |
+
},
|
| 18447 |
+
{
|
| 18448 |
+
"epoch": 0.66,
|
| 18449 |
+
"grad_norm": 65.0,
|
| 18450 |
+
"learning_rate": 1.2576271186440675e-07,
|
| 18451 |
+
"loss": 1.1554,
|
| 18452 |
+
"step": 26290
|
| 18453 |
+
},
|
| 18454 |
+
{
|
| 18455 |
+
"epoch": 0.66,
|
| 18456 |
+
"grad_norm": 65.0,
|
| 18457 |
+
"learning_rate": 1.254237288135593e-07,
|
| 18458 |
+
"loss": 1.1686,
|
| 18459 |
+
"step": 26300
|
| 18460 |
+
},
|
| 18461 |
+
{
|
| 18462 |
+
"epoch": 0.66,
|
| 18463 |
+
"grad_norm": 65.0,
|
| 18464 |
+
"learning_rate": 1.2508474576271186e-07,
|
| 18465 |
+
"loss": 1.1488,
|
| 18466 |
+
"step": 26310
|
| 18467 |
+
},
|
| 18468 |
+
{
|
| 18469 |
+
"epoch": 0.66,
|
| 18470 |
+
"grad_norm": 69.5,
|
| 18471 |
+
"learning_rate": 1.2474576271186442e-07,
|
| 18472 |
+
"loss": 1.1584,
|
| 18473 |
+
"step": 26320
|
| 18474 |
+
},
|
| 18475 |
+
{
|
| 18476 |
+
"epoch": 0.66,
|
| 18477 |
+
"grad_norm": 65.0,
|
| 18478 |
+
"learning_rate": 1.2440677966101694e-07,
|
| 18479 |
+
"loss": 1.1547,
|
| 18480 |
+
"step": 26330
|
| 18481 |
+
},
|
| 18482 |
+
{
|
| 18483 |
+
"epoch": 0.66,
|
| 18484 |
+
"grad_norm": 65.5,
|
| 18485 |
+
"learning_rate": 1.240677966101695e-07,
|
| 18486 |
+
"loss": 1.1438,
|
| 18487 |
+
"step": 26340
|
| 18488 |
+
},
|
| 18489 |
+
{
|
| 18490 |
+
"epoch": 0.66,
|
| 18491 |
+
"grad_norm": 63.5,
|
| 18492 |
+
"learning_rate": 1.2372881355932203e-07,
|
| 18493 |
+
"loss": 1.158,
|
| 18494 |
+
"step": 26350
|
| 18495 |
+
},
|
| 18496 |
+
{
|
| 18497 |
+
"epoch": 0.66,
|
| 18498 |
+
"grad_norm": 65.0,
|
| 18499 |
+
"learning_rate": 1.2338983050847458e-07,
|
| 18500 |
+
"loss": 1.1535,
|
| 18501 |
+
"step": 26360
|
| 18502 |
+
},
|
| 18503 |
+
{
|
| 18504 |
+
"epoch": 0.66,
|
| 18505 |
+
"grad_norm": 66.5,
|
| 18506 |
+
"learning_rate": 1.230508474576271e-07,
|
| 18507 |
+
"loss": 1.1466,
|
| 18508 |
+
"step": 26370
|
| 18509 |
+
},
|
| 18510 |
+
{
|
| 18511 |
+
"epoch": 0.66,
|
| 18512 |
+
"grad_norm": 66.5,
|
| 18513 |
+
"learning_rate": 1.2271186440677966e-07,
|
| 18514 |
+
"loss": 1.1663,
|
| 18515 |
+
"step": 26380
|
| 18516 |
+
},
|
| 18517 |
+
{
|
| 18518 |
+
"epoch": 0.66,
|
| 18519 |
+
"grad_norm": 66.0,
|
| 18520 |
+
"learning_rate": 1.223728813559322e-07,
|
| 18521 |
+
"loss": 1.1487,
|
| 18522 |
+
"step": 26390
|
| 18523 |
+
},
|
| 18524 |
+
{
|
| 18525 |
+
"epoch": 0.66,
|
| 18526 |
+
"grad_norm": 66.5,
|
| 18527 |
+
"learning_rate": 1.2203389830508474e-07,
|
| 18528 |
+
"loss": 1.1752,
|
| 18529 |
+
"step": 26400
|
| 18530 |
+
},
|
| 18531 |
+
{
|
| 18532 |
+
"epoch": 0.66,
|
| 18533 |
+
"grad_norm": 64.5,
|
| 18534 |
+
"learning_rate": 1.2169491525423727e-07,
|
| 18535 |
+
"loss": 1.1582,
|
| 18536 |
+
"step": 26410
|
| 18537 |
+
},
|
| 18538 |
+
{
|
| 18539 |
+
"epoch": 0.66,
|
| 18540 |
+
"grad_norm": 65.5,
|
| 18541 |
+
"learning_rate": 1.2135593220338983e-07,
|
| 18542 |
+
"loss": 1.1549,
|
| 18543 |
+
"step": 26420
|
| 18544 |
+
},
|
| 18545 |
+
{
|
| 18546 |
+
"epoch": 0.66,
|
| 18547 |
+
"grad_norm": 64.0,
|
| 18548 |
+
"learning_rate": 1.2101694915254235e-07,
|
| 18549 |
+
"loss": 1.1754,
|
| 18550 |
+
"step": 26430
|
| 18551 |
+
},
|
| 18552 |
+
{
|
| 18553 |
+
"epoch": 0.66,
|
| 18554 |
+
"grad_norm": 66.0,
|
| 18555 |
+
"learning_rate": 1.206779661016949e-07,
|
| 18556 |
+
"loss": 1.1594,
|
| 18557 |
+
"step": 26440
|
| 18558 |
+
},
|
| 18559 |
+
{
|
| 18560 |
+
"epoch": 0.66,
|
| 18561 |
+
"grad_norm": 66.0,
|
| 18562 |
+
"learning_rate": 1.2033898305084746e-07,
|
| 18563 |
+
"loss": 1.1614,
|
| 18564 |
+
"step": 26450
|
| 18565 |
+
},
|
| 18566 |
+
{
|
| 18567 |
+
"epoch": 0.66,
|
| 18568 |
+
"grad_norm": 63.25,
|
| 18569 |
+
"learning_rate": 1.2e-07,
|
| 18570 |
+
"loss": 1.171,
|
| 18571 |
+
"step": 26460
|
| 18572 |
+
},
|
| 18573 |
+
{
|
| 18574 |
+
"epoch": 0.66,
|
| 18575 |
+
"grad_norm": 69.5,
|
| 18576 |
+
"learning_rate": 1.1966101694915254e-07,
|
| 18577 |
+
"loss": 1.1617,
|
| 18578 |
+
"step": 26470
|
| 18579 |
+
},
|
| 18580 |
+
{
|
| 18581 |
+
"epoch": 0.66,
|
| 18582 |
+
"grad_norm": 65.5,
|
| 18583 |
+
"learning_rate": 1.193220338983051e-07,
|
| 18584 |
+
"loss": 1.1552,
|
| 18585 |
+
"step": 26480
|
| 18586 |
+
},
|
| 18587 |
+
{
|
| 18588 |
+
"epoch": 0.66,
|
| 18589 |
+
"grad_norm": 64.5,
|
| 18590 |
+
"learning_rate": 1.1898305084745763e-07,
|
| 18591 |
+
"loss": 1.1751,
|
| 18592 |
+
"step": 26490
|
| 18593 |
+
},
|
| 18594 |
+
{
|
| 18595 |
+
"epoch": 0.66,
|
| 18596 |
+
"grad_norm": 65.5,
|
| 18597 |
+
"learning_rate": 1.1864406779661017e-07,
|
| 18598 |
+
"loss": 1.1537,
|
| 18599 |
+
"step": 26500
|
| 18600 |
+
},
|
| 18601 |
+
{
|
| 18602 |
+
"epoch": 0.66,
|
| 18603 |
+
"grad_norm": 65.0,
|
| 18604 |
+
"learning_rate": 1.1830508474576271e-07,
|
| 18605 |
+
"loss": 1.1505,
|
| 18606 |
+
"step": 26510
|
| 18607 |
+
},
|
| 18608 |
+
{
|
| 18609 |
+
"epoch": 0.66,
|
| 18610 |
+
"grad_norm": 64.5,
|
| 18611 |
+
"learning_rate": 1.1796610169491525e-07,
|
| 18612 |
+
"loss": 1.1589,
|
| 18613 |
+
"step": 26520
|
| 18614 |
+
},
|
| 18615 |
+
{
|
| 18616 |
+
"epoch": 0.66,
|
| 18617 |
+
"grad_norm": 65.0,
|
| 18618 |
+
"learning_rate": 1.176271186440678e-07,
|
| 18619 |
+
"loss": 1.1575,
|
| 18620 |
+
"step": 26530
|
| 18621 |
+
},
|
| 18622 |
+
{
|
| 18623 |
+
"epoch": 0.66,
|
| 18624 |
+
"grad_norm": 67.0,
|
| 18625 |
+
"learning_rate": 1.1728813559322033e-07,
|
| 18626 |
+
"loss": 1.1615,
|
| 18627 |
+
"step": 26540
|
| 18628 |
+
},
|
| 18629 |
+
{
|
| 18630 |
+
"epoch": 0.66,
|
| 18631 |
+
"grad_norm": 64.0,
|
| 18632 |
+
"learning_rate": 1.1694915254237288e-07,
|
| 18633 |
+
"loss": 1.1684,
|
| 18634 |
+
"step": 26550
|
| 18635 |
+
},
|
| 18636 |
+
{
|
| 18637 |
+
"epoch": 0.66,
|
| 18638 |
+
"grad_norm": 64.0,
|
| 18639 |
+
"learning_rate": 1.1661016949152541e-07,
|
| 18640 |
+
"loss": 1.1511,
|
| 18641 |
+
"step": 26560
|
| 18642 |
+
},
|
| 18643 |
+
{
|
| 18644 |
+
"epoch": 0.66,
|
| 18645 |
+
"grad_norm": 73.0,
|
| 18646 |
+
"learning_rate": 1.1627118644067797e-07,
|
| 18647 |
+
"loss": 1.1818,
|
| 18648 |
+
"step": 26570
|
| 18649 |
+
},
|
| 18650 |
+
{
|
| 18651 |
+
"epoch": 0.66,
|
| 18652 |
+
"grad_norm": 66.5,
|
| 18653 |
+
"learning_rate": 1.1593220338983051e-07,
|
| 18654 |
+
"loss": 1.1594,
|
| 18655 |
+
"step": 26580
|
| 18656 |
+
},
|
| 18657 |
+
{
|
| 18658 |
+
"epoch": 0.66,
|
| 18659 |
+
"grad_norm": 66.0,
|
| 18660 |
+
"learning_rate": 1.1559322033898305e-07,
|
| 18661 |
+
"loss": 1.1641,
|
| 18662 |
+
"step": 26590
|
| 18663 |
+
},
|
| 18664 |
+
{
|
| 18665 |
+
"epoch": 0.67,
|
| 18666 |
+
"grad_norm": 71.0,
|
| 18667 |
+
"learning_rate": 1.1525423728813559e-07,
|
| 18668 |
+
"loss": 1.1727,
|
| 18669 |
+
"step": 26600
|
| 18670 |
+
},
|
| 18671 |
+
{
|
| 18672 |
+
"epoch": 0.67,
|
| 18673 |
+
"grad_norm": 67.0,
|
| 18674 |
+
"learning_rate": 1.1491525423728814e-07,
|
| 18675 |
+
"loss": 1.1762,
|
| 18676 |
+
"step": 26610
|
| 18677 |
+
},
|
| 18678 |
+
{
|
| 18679 |
+
"epoch": 0.67,
|
| 18680 |
+
"grad_norm": 64.5,
|
| 18681 |
+
"learning_rate": 1.1457627118644067e-07,
|
| 18682 |
+
"loss": 1.1572,
|
| 18683 |
+
"step": 26620
|
| 18684 |
+
},
|
| 18685 |
+
{
|
| 18686 |
+
"epoch": 0.67,
|
| 18687 |
+
"grad_norm": 65.0,
|
| 18688 |
+
"learning_rate": 1.1423728813559322e-07,
|
| 18689 |
+
"loss": 1.1373,
|
| 18690 |
+
"step": 26630
|
| 18691 |
+
},
|
| 18692 |
+
{
|
| 18693 |
+
"epoch": 0.67,
|
| 18694 |
+
"grad_norm": 67.0,
|
| 18695 |
+
"learning_rate": 1.1389830508474575e-07,
|
| 18696 |
+
"loss": 1.1662,
|
| 18697 |
+
"step": 26640
|
| 18698 |
+
},
|
| 18699 |
+
{
|
| 18700 |
+
"epoch": 0.67,
|
| 18701 |
+
"grad_norm": 64.5,
|
| 18702 |
+
"learning_rate": 1.135593220338983e-07,
|
| 18703 |
+
"loss": 1.1405,
|
| 18704 |
+
"step": 26650
|
| 18705 |
+
},
|
| 18706 |
+
{
|
| 18707 |
+
"epoch": 0.67,
|
| 18708 |
+
"grad_norm": 66.5,
|
| 18709 |
+
"learning_rate": 1.1322033898305085e-07,
|
| 18710 |
+
"loss": 1.1456,
|
| 18711 |
+
"step": 26660
|
| 18712 |
+
},
|
| 18713 |
+
{
|
| 18714 |
+
"epoch": 0.67,
|
| 18715 |
+
"grad_norm": 64.5,
|
| 18716 |
+
"learning_rate": 1.1288135593220339e-07,
|
| 18717 |
+
"loss": 1.1323,
|
| 18718 |
+
"step": 26670
|
| 18719 |
+
},
|
| 18720 |
+
{
|
| 18721 |
+
"epoch": 0.67,
|
| 18722 |
+
"grad_norm": 69.0,
|
| 18723 |
+
"learning_rate": 1.1254237288135593e-07,
|
| 18724 |
+
"loss": 1.1493,
|
| 18725 |
+
"step": 26680
|
| 18726 |
+
},
|
| 18727 |
+
{
|
| 18728 |
+
"epoch": 0.67,
|
| 18729 |
+
"grad_norm": 63.25,
|
| 18730 |
+
"learning_rate": 1.1220338983050846e-07,
|
| 18731 |
+
"loss": 1.1563,
|
| 18732 |
+
"step": 26690
|
| 18733 |
+
},
|
| 18734 |
+
{
|
| 18735 |
+
"epoch": 0.67,
|
| 18736 |
+
"grad_norm": 66.0,
|
| 18737 |
+
"learning_rate": 1.1186440677966101e-07,
|
| 18738 |
+
"loss": 1.166,
|
| 18739 |
+
"step": 26700
|
| 18740 |
+
},
|
| 18741 |
+
{
|
| 18742 |
+
"epoch": 0.67,
|
| 18743 |
+
"grad_norm": 66.5,
|
| 18744 |
+
"learning_rate": 1.1152542372881356e-07,
|
| 18745 |
+
"loss": 1.1597,
|
| 18746 |
+
"step": 26710
|
| 18747 |
+
},
|
| 18748 |
+
{
|
| 18749 |
+
"epoch": 0.67,
|
| 18750 |
+
"grad_norm": 66.5,
|
| 18751 |
+
"learning_rate": 1.1118644067796609e-07,
|
| 18752 |
+
"loss": 1.1509,
|
| 18753 |
+
"step": 26720
|
| 18754 |
+
},
|
| 18755 |
+
{
|
| 18756 |
+
"epoch": 0.67,
|
| 18757 |
+
"grad_norm": 67.0,
|
| 18758 |
+
"learning_rate": 1.1084745762711865e-07,
|
| 18759 |
+
"loss": 1.1594,
|
| 18760 |
+
"step": 26730
|
| 18761 |
+
},
|
| 18762 |
+
{
|
| 18763 |
+
"epoch": 0.67,
|
| 18764 |
+
"grad_norm": 68.5,
|
| 18765 |
+
"learning_rate": 1.1050847457627119e-07,
|
| 18766 |
+
"loss": 1.2039,
|
| 18767 |
+
"step": 26740
|
| 18768 |
+
},
|
| 18769 |
+
{
|
| 18770 |
+
"epoch": 0.67,
|
| 18771 |
+
"grad_norm": 65.0,
|
| 18772 |
+
"learning_rate": 1.1016949152542372e-07,
|
| 18773 |
+
"loss": 1.1834,
|
| 18774 |
+
"step": 26750
|
| 18775 |
+
},
|
| 18776 |
+
{
|
| 18777 |
+
"epoch": 0.67,
|
| 18778 |
+
"grad_norm": 71.0,
|
| 18779 |
+
"learning_rate": 1.0983050847457627e-07,
|
| 18780 |
+
"loss": 1.1813,
|
| 18781 |
+
"step": 26760
|
| 18782 |
+
},
|
| 18783 |
+
{
|
| 18784 |
+
"epoch": 0.67,
|
| 18785 |
+
"grad_norm": 65.0,
|
| 18786 |
+
"learning_rate": 1.094915254237288e-07,
|
| 18787 |
+
"loss": 1.1655,
|
| 18788 |
+
"step": 26770
|
| 18789 |
+
},
|
| 18790 |
+
{
|
| 18791 |
+
"epoch": 0.67,
|
| 18792 |
+
"grad_norm": 67.0,
|
| 18793 |
+
"learning_rate": 1.0915254237288135e-07,
|
| 18794 |
+
"loss": 1.1869,
|
| 18795 |
+
"step": 26780
|
| 18796 |
+
},
|
| 18797 |
+
{
|
| 18798 |
+
"epoch": 0.67,
|
| 18799 |
+
"grad_norm": 65.5,
|
| 18800 |
+
"learning_rate": 1.0881355932203389e-07,
|
| 18801 |
+
"loss": 1.1891,
|
| 18802 |
+
"step": 26790
|
| 18803 |
+
},
|
| 18804 |
+
{
|
| 18805 |
+
"epoch": 0.67,
|
| 18806 |
+
"grad_norm": 64.5,
|
| 18807 |
+
"learning_rate": 1.0847457627118643e-07,
|
| 18808 |
+
"loss": 1.1724,
|
| 18809 |
+
"step": 26800
|
| 18810 |
+
},
|
| 18811 |
+
{
|
| 18812 |
+
"epoch": 0.67,
|
| 18813 |
+
"grad_norm": 71.0,
|
| 18814 |
+
"learning_rate": 1.0813559322033897e-07,
|
| 18815 |
+
"loss": 1.1609,
|
| 18816 |
+
"step": 26810
|
| 18817 |
+
},
|
| 18818 |
+
{
|
| 18819 |
+
"epoch": 0.67,
|
| 18820 |
+
"grad_norm": 70.5,
|
| 18821 |
+
"learning_rate": 1.0779661016949151e-07,
|
| 18822 |
+
"loss": 1.1671,
|
| 18823 |
+
"step": 26820
|
| 18824 |
+
},
|
| 18825 |
+
{
|
| 18826 |
+
"epoch": 0.67,
|
| 18827 |
+
"grad_norm": 66.5,
|
| 18828 |
+
"learning_rate": 1.0745762711864406e-07,
|
| 18829 |
+
"loss": 1.1725,
|
| 18830 |
+
"step": 26830
|
| 18831 |
+
},
|
| 18832 |
+
{
|
| 18833 |
+
"epoch": 0.67,
|
| 18834 |
+
"grad_norm": 65.0,
|
| 18835 |
+
"learning_rate": 1.0711864406779661e-07,
|
| 18836 |
+
"loss": 1.1763,
|
| 18837 |
+
"step": 26840
|
| 18838 |
+
},
|
| 18839 |
+
{
|
| 18840 |
+
"epoch": 0.67,
|
| 18841 |
+
"grad_norm": 69.5,
|
| 18842 |
+
"learning_rate": 1.0677966101694914e-07,
|
| 18843 |
+
"loss": 1.1771,
|
| 18844 |
+
"step": 26850
|
| 18845 |
+
},
|
| 18846 |
+
{
|
| 18847 |
+
"epoch": 0.67,
|
| 18848 |
+
"grad_norm": 65.0,
|
| 18849 |
+
"learning_rate": 1.0644067796610169e-07,
|
| 18850 |
+
"loss": 1.1843,
|
| 18851 |
+
"step": 26860
|
| 18852 |
+
},
|
| 18853 |
+
{
|
| 18854 |
+
"epoch": 0.67,
|
| 18855 |
+
"grad_norm": 72.5,
|
| 18856 |
+
"learning_rate": 1.0610169491525423e-07,
|
| 18857 |
+
"loss": 1.1825,
|
| 18858 |
+
"step": 26870
|
| 18859 |
+
},
|
| 18860 |
+
{
|
| 18861 |
+
"epoch": 0.67,
|
| 18862 |
+
"grad_norm": 64.5,
|
| 18863 |
+
"learning_rate": 1.0576271186440677e-07,
|
| 18864 |
+
"loss": 1.1688,
|
| 18865 |
+
"step": 26880
|
| 18866 |
+
},
|
| 18867 |
+
{
|
| 18868 |
+
"epoch": 0.67,
|
| 18869 |
+
"grad_norm": 63.75,
|
| 18870 |
+
"learning_rate": 1.0542372881355931e-07,
|
| 18871 |
+
"loss": 1.1708,
|
| 18872 |
+
"step": 26890
|
| 18873 |
+
},
|
| 18874 |
+
{
|
| 18875 |
+
"epoch": 0.67,
|
| 18876 |
+
"grad_norm": 66.5,
|
| 18877 |
+
"learning_rate": 1.0508474576271186e-07,
|
| 18878 |
+
"loss": 1.1755,
|
| 18879 |
+
"step": 26900
|
| 18880 |
+
},
|
| 18881 |
+
{
|
| 18882 |
+
"epoch": 0.67,
|
| 18883 |
+
"grad_norm": 66.5,
|
| 18884 |
+
"learning_rate": 1.047457627118644e-07,
|
| 18885 |
+
"loss": 1.1559,
|
| 18886 |
+
"step": 26910
|
| 18887 |
+
},
|
| 18888 |
+
{
|
| 18889 |
+
"epoch": 0.67,
|
| 18890 |
+
"grad_norm": 65.0,
|
| 18891 |
+
"learning_rate": 1.0440677966101695e-07,
|
| 18892 |
+
"loss": 1.1474,
|
| 18893 |
+
"step": 26920
|
| 18894 |
+
},
|
| 18895 |
+
{
|
| 18896 |
+
"epoch": 0.67,
|
| 18897 |
+
"grad_norm": 67.5,
|
| 18898 |
+
"learning_rate": 1.0406779661016948e-07,
|
| 18899 |
+
"loss": 1.1922,
|
| 18900 |
+
"step": 26930
|
| 18901 |
+
},
|
| 18902 |
+
{
|
| 18903 |
+
"epoch": 0.67,
|
| 18904 |
+
"grad_norm": 69.0,
|
| 18905 |
+
"learning_rate": 1.0372881355932203e-07,
|
| 18906 |
+
"loss": 1.1789,
|
| 18907 |
+
"step": 26940
|
| 18908 |
+
},
|
| 18909 |
+
{
|
| 18910 |
+
"epoch": 0.67,
|
| 18911 |
+
"grad_norm": 67.5,
|
| 18912 |
+
"learning_rate": 1.0338983050847456e-07,
|
| 18913 |
+
"loss": 1.1707,
|
| 18914 |
+
"step": 26950
|
| 18915 |
+
},
|
| 18916 |
+
{
|
| 18917 |
+
"epoch": 0.67,
|
| 18918 |
+
"grad_norm": 67.5,
|
| 18919 |
+
"learning_rate": 1.0305084745762711e-07,
|
| 18920 |
+
"loss": 1.1641,
|
| 18921 |
+
"step": 26960
|
| 18922 |
+
},
|
| 18923 |
+
{
|
| 18924 |
+
"epoch": 0.67,
|
| 18925 |
+
"grad_norm": 67.5,
|
| 18926 |
+
"learning_rate": 1.0271186440677965e-07,
|
| 18927 |
+
"loss": 1.179,
|
| 18928 |
+
"step": 26970
|
| 18929 |
+
},
|
| 18930 |
+
{
|
| 18931 |
+
"epoch": 0.67,
|
| 18932 |
+
"grad_norm": 63.75,
|
| 18933 |
+
"learning_rate": 1.023728813559322e-07,
|
| 18934 |
+
"loss": 1.1731,
|
| 18935 |
+
"step": 26980
|
| 18936 |
+
},
|
| 18937 |
+
{
|
| 18938 |
+
"epoch": 0.67,
|
| 18939 |
+
"grad_norm": 69.5,
|
| 18940 |
+
"learning_rate": 1.0203389830508474e-07,
|
| 18941 |
+
"loss": 1.1675,
|
| 18942 |
+
"step": 26990
|
| 18943 |
+
},
|
| 18944 |
+
{
|
| 18945 |
+
"epoch": 0.68,
|
| 18946 |
+
"grad_norm": 65.5,
|
| 18947 |
+
"learning_rate": 1.0169491525423729e-07,
|
| 18948 |
+
"loss": 1.176,
|
| 18949 |
+
"step": 27000
|
| 18950 |
+
},
|
| 18951 |
+
{
|
| 18952 |
+
"epoch": 0.68,
|
| 18953 |
+
"grad_norm": 66.0,
|
| 18954 |
+
"learning_rate": 1.0135593220338982e-07,
|
| 18955 |
+
"loss": 1.1738,
|
| 18956 |
+
"step": 27010
|
| 18957 |
+
},
|
| 18958 |
+
{
|
| 18959 |
+
"epoch": 0.68,
|
| 18960 |
+
"grad_norm": 67.5,
|
| 18961 |
+
"learning_rate": 1.0101694915254237e-07,
|
| 18962 |
+
"loss": 1.1805,
|
| 18963 |
+
"step": 27020
|
| 18964 |
+
},
|
| 18965 |
+
{
|
| 18966 |
+
"epoch": 0.68,
|
| 18967 |
+
"grad_norm": 67.0,
|
| 18968 |
+
"learning_rate": 1.006779661016949e-07,
|
| 18969 |
+
"loss": 1.184,
|
| 18970 |
+
"step": 27030
|
| 18971 |
+
},
|
| 18972 |
+
{
|
| 18973 |
+
"epoch": 0.68,
|
| 18974 |
+
"grad_norm": 65.0,
|
| 18975 |
+
"learning_rate": 1.0033898305084745e-07,
|
| 18976 |
+
"loss": 1.174,
|
| 18977 |
+
"step": 27040
|
| 18978 |
+
},
|
| 18979 |
+
{
|
| 18980 |
+
"epoch": 0.68,
|
| 18981 |
+
"grad_norm": 68.0,
|
| 18982 |
+
"learning_rate": 1e-07,
|
| 18983 |
+
"loss": 1.1811,
|
| 18984 |
+
"step": 27050
|
| 18985 |
+
},
|
| 18986 |
+
{
|
| 18987 |
+
"epoch": 0.68,
|
| 18988 |
+
"grad_norm": 68.0,
|
| 18989 |
+
"learning_rate": 9.966101694915254e-08,
|
| 18990 |
+
"loss": 1.1762,
|
| 18991 |
+
"step": 27060
|
| 18992 |
+
},
|
| 18993 |
+
{
|
| 18994 |
+
"epoch": 0.68,
|
| 18995 |
+
"grad_norm": 66.5,
|
| 18996 |
+
"learning_rate": 9.932203389830508e-08,
|
| 18997 |
+
"loss": 1.1601,
|
| 18998 |
+
"step": 27070
|
| 18999 |
+
},
|
| 19000 |
+
{
|
| 19001 |
+
"epoch": 0.68,
|
| 19002 |
+
"grad_norm": 64.5,
|
| 19003 |
+
"learning_rate": 9.898305084745762e-08,
|
| 19004 |
+
"loss": 1.1618,
|
| 19005 |
+
"step": 27080
|
| 19006 |
+
},
|
| 19007 |
+
{
|
| 19008 |
+
"epoch": 0.68,
|
| 19009 |
+
"grad_norm": 66.5,
|
| 19010 |
+
"learning_rate": 9.864406779661016e-08,
|
| 19011 |
+
"loss": 1.1715,
|
| 19012 |
+
"step": 27090
|
| 19013 |
+
},
|
| 19014 |
+
{
|
| 19015 |
+
"epoch": 0.68,
|
| 19016 |
+
"grad_norm": 66.0,
|
| 19017 |
+
"learning_rate": 9.830508474576271e-08,
|
| 19018 |
+
"loss": 1.1831,
|
| 19019 |
+
"step": 27100
|
| 19020 |
+
},
|
| 19021 |
+
{
|
| 19022 |
+
"epoch": 0.68,
|
| 19023 |
+
"grad_norm": 64.0,
|
| 19024 |
+
"learning_rate": 9.796610169491524e-08,
|
| 19025 |
+
"loss": 1.1904,
|
| 19026 |
+
"step": 27110
|
| 19027 |
+
},
|
| 19028 |
+
{
|
| 19029 |
+
"epoch": 0.68,
|
| 19030 |
+
"grad_norm": 65.0,
|
| 19031 |
+
"learning_rate": 9.76271186440678e-08,
|
| 19032 |
+
"loss": 1.1563,
|
| 19033 |
+
"step": 27120
|
| 19034 |
+
},
|
| 19035 |
+
{
|
| 19036 |
+
"epoch": 0.68,
|
| 19037 |
+
"grad_norm": 66.0,
|
| 19038 |
+
"learning_rate": 9.728813559322034e-08,
|
| 19039 |
+
"loss": 1.1901,
|
| 19040 |
+
"step": 27130
|
| 19041 |
+
},
|
| 19042 |
+
{
|
| 19043 |
+
"epoch": 0.68,
|
| 19044 |
+
"grad_norm": 64.0,
|
| 19045 |
+
"learning_rate": 9.694915254237288e-08,
|
| 19046 |
+
"loss": 1.1703,
|
| 19047 |
+
"step": 27140
|
| 19048 |
+
},
|
| 19049 |
+
{
|
| 19050 |
+
"epoch": 0.68,
|
| 19051 |
+
"grad_norm": 67.5,
|
| 19052 |
+
"learning_rate": 9.661016949152542e-08,
|
| 19053 |
+
"loss": 1.1721,
|
| 19054 |
+
"step": 27150
|
| 19055 |
+
},
|
| 19056 |
+
{
|
| 19057 |
+
"epoch": 0.68,
|
| 19058 |
+
"grad_norm": 65.0,
|
| 19059 |
+
"learning_rate": 9.627118644067796e-08,
|
| 19060 |
+
"loss": 1.1758,
|
| 19061 |
+
"step": 27160
|
| 19062 |
+
},
|
| 19063 |
+
{
|
| 19064 |
+
"epoch": 0.68,
|
| 19065 |
+
"grad_norm": 65.5,
|
| 19066 |
+
"learning_rate": 9.59322033898305e-08,
|
| 19067 |
+
"loss": 1.1511,
|
| 19068 |
+
"step": 27170
|
| 19069 |
+
},
|
| 19070 |
+
{
|
| 19071 |
+
"epoch": 0.68,
|
| 19072 |
+
"grad_norm": 66.0,
|
| 19073 |
+
"learning_rate": 9.559322033898305e-08,
|
| 19074 |
+
"loss": 1.1678,
|
| 19075 |
+
"step": 27180
|
| 19076 |
+
},
|
| 19077 |
+
{
|
| 19078 |
+
"epoch": 0.68,
|
| 19079 |
+
"grad_norm": 67.5,
|
| 19080 |
+
"learning_rate": 9.525423728813558e-08,
|
| 19081 |
+
"loss": 1.1904,
|
| 19082 |
+
"step": 27190
|
| 19083 |
+
},
|
| 19084 |
+
{
|
| 19085 |
+
"epoch": 0.68,
|
| 19086 |
+
"grad_norm": 64.5,
|
| 19087 |
+
"learning_rate": 9.491525423728814e-08,
|
| 19088 |
+
"loss": 1.171,
|
| 19089 |
+
"step": 27200
|
| 19090 |
+
},
|
| 19091 |
+
{
|
| 19092 |
+
"epoch": 0.68,
|
| 19093 |
+
"grad_norm": 69.0,
|
| 19094 |
+
"learning_rate": 9.457627118644066e-08,
|
| 19095 |
+
"loss": 1.1611,
|
| 19096 |
+
"step": 27210
|
| 19097 |
+
},
|
| 19098 |
+
{
|
| 19099 |
+
"epoch": 0.68,
|
| 19100 |
+
"grad_norm": 69.0,
|
| 19101 |
+
"learning_rate": 9.423728813559322e-08,
|
| 19102 |
+
"loss": 1.1792,
|
| 19103 |
+
"step": 27220
|
| 19104 |
+
},
|
| 19105 |
+
{
|
| 19106 |
+
"epoch": 0.68,
|
| 19107 |
+
"grad_norm": 68.0,
|
| 19108 |
+
"learning_rate": 9.389830508474576e-08,
|
| 19109 |
+
"loss": 1.193,
|
| 19110 |
+
"step": 27230
|
| 19111 |
+
},
|
| 19112 |
+
{
|
| 19113 |
+
"epoch": 0.68,
|
| 19114 |
+
"grad_norm": 66.0,
|
| 19115 |
+
"learning_rate": 9.35593220338983e-08,
|
| 19116 |
+
"loss": 1.2138,
|
| 19117 |
+
"step": 27240
|
| 19118 |
+
},
|
| 19119 |
+
{
|
| 19120 |
+
"epoch": 0.68,
|
| 19121 |
+
"grad_norm": 67.0,
|
| 19122 |
+
"learning_rate": 9.322033898305084e-08,
|
| 19123 |
+
"loss": 1.1808,
|
| 19124 |
+
"step": 27250
|
| 19125 |
+
},
|
| 19126 |
+
{
|
| 19127 |
+
"epoch": 0.68,
|
| 19128 |
+
"grad_norm": 69.5,
|
| 19129 |
+
"learning_rate": 9.28813559322034e-08,
|
| 19130 |
+
"loss": 1.1754,
|
| 19131 |
+
"step": 27260
|
| 19132 |
+
},
|
| 19133 |
+
{
|
| 19134 |
+
"epoch": 0.68,
|
| 19135 |
+
"grad_norm": 64.0,
|
| 19136 |
+
"learning_rate": 9.254237288135592e-08,
|
| 19137 |
+
"loss": 1.1714,
|
| 19138 |
+
"step": 27270
|
| 19139 |
+
},
|
| 19140 |
+
{
|
| 19141 |
+
"epoch": 0.68,
|
| 19142 |
+
"grad_norm": 67.5,
|
| 19143 |
+
"learning_rate": 9.220338983050848e-08,
|
| 19144 |
+
"loss": 1.1801,
|
| 19145 |
+
"step": 27280
|
| 19146 |
+
},
|
| 19147 |
+
{
|
| 19148 |
+
"epoch": 0.68,
|
| 19149 |
+
"grad_norm": 66.0,
|
| 19150 |
+
"learning_rate": 9.1864406779661e-08,
|
| 19151 |
+
"loss": 1.1587,
|
| 19152 |
+
"step": 27290
|
| 19153 |
+
},
|
| 19154 |
+
{
|
| 19155 |
+
"epoch": 0.68,
|
| 19156 |
+
"grad_norm": 66.5,
|
| 19157 |
+
"learning_rate": 9.152542372881356e-08,
|
| 19158 |
+
"loss": 1.1749,
|
| 19159 |
+
"step": 27300
|
| 19160 |
+
},
|
| 19161 |
+
{
|
| 19162 |
+
"epoch": 0.68,
|
| 19163 |
+
"grad_norm": 66.5,
|
| 19164 |
+
"learning_rate": 9.11864406779661e-08,
|
| 19165 |
+
"loss": 1.1671,
|
| 19166 |
+
"step": 27310
|
| 19167 |
+
},
|
| 19168 |
+
{
|
| 19169 |
+
"epoch": 0.68,
|
| 19170 |
+
"grad_norm": 67.0,
|
| 19171 |
+
"learning_rate": 9.084745762711864e-08,
|
| 19172 |
+
"loss": 1.1646,
|
| 19173 |
+
"step": 27320
|
| 19174 |
+
},
|
| 19175 |
+
{
|
| 19176 |
+
"epoch": 0.68,
|
| 19177 |
+
"grad_norm": 64.5,
|
| 19178 |
+
"learning_rate": 9.050847457627118e-08,
|
| 19179 |
+
"loss": 1.1718,
|
| 19180 |
+
"step": 27330
|
| 19181 |
+
},
|
| 19182 |
+
{
|
| 19183 |
+
"epoch": 0.68,
|
| 19184 |
+
"grad_norm": 67.5,
|
| 19185 |
+
"learning_rate": 9.016949152542372e-08,
|
| 19186 |
+
"loss": 1.1836,
|
| 19187 |
+
"step": 27340
|
| 19188 |
+
},
|
| 19189 |
+
{
|
| 19190 |
+
"epoch": 0.68,
|
| 19191 |
+
"grad_norm": 66.5,
|
| 19192 |
+
"learning_rate": 8.983050847457626e-08,
|
| 19193 |
+
"loss": 1.1577,
|
| 19194 |
+
"step": 27350
|
| 19195 |
+
},
|
| 19196 |
+
{
|
| 19197 |
+
"epoch": 0.68,
|
| 19198 |
+
"grad_norm": 64.0,
|
| 19199 |
+
"learning_rate": 8.949152542372882e-08,
|
| 19200 |
+
"loss": 1.1663,
|
| 19201 |
+
"step": 27360
|
| 19202 |
+
},
|
| 19203 |
+
{
|
| 19204 |
+
"epoch": 0.68,
|
| 19205 |
+
"grad_norm": 65.5,
|
| 19206 |
+
"learning_rate": 8.915254237288134e-08,
|
| 19207 |
+
"loss": 1.1814,
|
| 19208 |
+
"step": 27370
|
| 19209 |
+
},
|
| 19210 |
+
{
|
| 19211 |
+
"epoch": 0.68,
|
| 19212 |
+
"grad_norm": 64.0,
|
| 19213 |
+
"learning_rate": 8.88135593220339e-08,
|
| 19214 |
+
"loss": 1.1874,
|
| 19215 |
+
"step": 27380
|
| 19216 |
+
},
|
| 19217 |
+
{
|
| 19218 |
+
"epoch": 0.68,
|
| 19219 |
+
"grad_norm": 65.5,
|
| 19220 |
+
"learning_rate": 8.847457627118644e-08,
|
| 19221 |
+
"loss": 1.1757,
|
| 19222 |
+
"step": 27390
|
| 19223 |
+
},
|
| 19224 |
+
{
|
| 19225 |
+
"epoch": 0.69,
|
| 19226 |
+
"grad_norm": 64.5,
|
| 19227 |
+
"learning_rate": 8.813559322033898e-08,
|
| 19228 |
+
"loss": 1.1597,
|
| 19229 |
+
"step": 27400
|
| 19230 |
+
},
|
| 19231 |
+
{
|
| 19232 |
+
"epoch": 0.69,
|
| 19233 |
+
"grad_norm": 67.5,
|
| 19234 |
+
"learning_rate": 8.779661016949152e-08,
|
| 19235 |
+
"loss": 1.1702,
|
| 19236 |
+
"step": 27410
|
| 19237 |
+
},
|
| 19238 |
+
{
|
| 19239 |
+
"epoch": 0.69,
|
| 19240 |
+
"grad_norm": 66.0,
|
| 19241 |
+
"learning_rate": 8.745762711864406e-08,
|
| 19242 |
+
"loss": 1.1581,
|
| 19243 |
+
"step": 27420
|
| 19244 |
+
},
|
| 19245 |
+
{
|
| 19246 |
+
"epoch": 0.69,
|
| 19247 |
+
"grad_norm": 65.0,
|
| 19248 |
+
"learning_rate": 8.71186440677966e-08,
|
| 19249 |
+
"loss": 1.1751,
|
| 19250 |
+
"step": 27430
|
| 19251 |
+
},
|
| 19252 |
+
{
|
| 19253 |
+
"epoch": 0.69,
|
| 19254 |
+
"grad_norm": 63.25,
|
| 19255 |
+
"learning_rate": 8.677966101694916e-08,
|
| 19256 |
+
"loss": 1.1562,
|
| 19257 |
+
"step": 27440
|
| 19258 |
+
},
|
| 19259 |
+
{
|
| 19260 |
+
"epoch": 0.69,
|
| 19261 |
+
"grad_norm": 65.0,
|
| 19262 |
+
"learning_rate": 8.644067796610168e-08,
|
| 19263 |
+
"loss": 1.1559,
|
| 19264 |
+
"step": 27450
|
| 19265 |
+
},
|
| 19266 |
+
{
|
| 19267 |
+
"epoch": 0.69,
|
| 19268 |
+
"grad_norm": 65.5,
|
| 19269 |
+
"learning_rate": 8.610169491525424e-08,
|
| 19270 |
+
"loss": 1.192,
|
| 19271 |
+
"step": 27460
|
| 19272 |
+
},
|
| 19273 |
+
{
|
| 19274 |
+
"epoch": 0.69,
|
| 19275 |
+
"grad_norm": 66.0,
|
| 19276 |
+
"learning_rate": 8.576271186440677e-08,
|
| 19277 |
+
"loss": 1.1701,
|
| 19278 |
+
"step": 27470
|
| 19279 |
+
},
|
| 19280 |
+
{
|
| 19281 |
+
"epoch": 0.69,
|
| 19282 |
+
"grad_norm": 64.0,
|
| 19283 |
+
"learning_rate": 8.542372881355932e-08,
|
| 19284 |
+
"loss": 1.1595,
|
| 19285 |
+
"step": 27480
|
| 19286 |
+
},
|
| 19287 |
+
{
|
| 19288 |
+
"epoch": 0.69,
|
| 19289 |
+
"grad_norm": 65.5,
|
| 19290 |
+
"learning_rate": 8.508474576271186e-08,
|
| 19291 |
+
"loss": 1.1636,
|
| 19292 |
+
"step": 27490
|
| 19293 |
+
},
|
| 19294 |
+
{
|
| 19295 |
+
"epoch": 0.69,
|
| 19296 |
+
"grad_norm": 65.0,
|
| 19297 |
+
"learning_rate": 8.47457627118644e-08,
|
| 19298 |
+
"loss": 1.1743,
|
| 19299 |
+
"step": 27500
|
| 19300 |
+
},
|
| 19301 |
+
{
|
| 19302 |
+
"epoch": 0.69,
|
| 19303 |
+
"grad_norm": 67.5,
|
| 19304 |
+
"learning_rate": 8.440677966101694e-08,
|
| 19305 |
+
"loss": 1.1932,
|
| 19306 |
+
"step": 27510
|
| 19307 |
+
},
|
| 19308 |
+
{
|
| 19309 |
+
"epoch": 0.69,
|
| 19310 |
+
"grad_norm": 63.75,
|
| 19311 |
+
"learning_rate": 8.40677966101695e-08,
|
| 19312 |
+
"loss": 1.1806,
|
| 19313 |
+
"step": 27520
|
| 19314 |
+
},
|
| 19315 |
+
{
|
| 19316 |
+
"epoch": 0.69,
|
| 19317 |
+
"grad_norm": 64.5,
|
| 19318 |
+
"learning_rate": 8.372881355932203e-08,
|
| 19319 |
+
"loss": 1.1705,
|
| 19320 |
+
"step": 27530
|
| 19321 |
+
},
|
| 19322 |
+
{
|
| 19323 |
+
"epoch": 0.69,
|
| 19324 |
+
"grad_norm": 67.0,
|
| 19325 |
+
"learning_rate": 8.338983050847458e-08,
|
| 19326 |
+
"loss": 1.1849,
|
| 19327 |
+
"step": 27540
|
| 19328 |
+
},
|
| 19329 |
+
{
|
| 19330 |
+
"epoch": 0.69,
|
| 19331 |
+
"grad_norm": 68.0,
|
| 19332 |
+
"learning_rate": 8.305084745762711e-08,
|
| 19333 |
+
"loss": 1.1698,
|
| 19334 |
+
"step": 27550
|
| 19335 |
+
},
|
| 19336 |
+
{
|
| 19337 |
+
"epoch": 0.69,
|
| 19338 |
+
"grad_norm": 69.0,
|
| 19339 |
+
"learning_rate": 8.271186440677966e-08,
|
| 19340 |
+
"loss": 1.1745,
|
| 19341 |
+
"step": 27560
|
| 19342 |
+
},
|
| 19343 |
+
{
|
| 19344 |
+
"epoch": 0.69,
|
| 19345 |
+
"grad_norm": 64.5,
|
| 19346 |
+
"learning_rate": 8.23728813559322e-08,
|
| 19347 |
+
"loss": 1.1622,
|
| 19348 |
+
"step": 27570
|
| 19349 |
+
},
|
| 19350 |
+
{
|
| 19351 |
+
"epoch": 0.69,
|
| 19352 |
+
"grad_norm": 68.0,
|
| 19353 |
+
"learning_rate": 8.203389830508474e-08,
|
| 19354 |
+
"loss": 1.1893,
|
| 19355 |
+
"step": 27580
|
| 19356 |
+
},
|
| 19357 |
+
{
|
| 19358 |
+
"epoch": 0.69,
|
| 19359 |
+
"grad_norm": 65.5,
|
| 19360 |
+
"learning_rate": 8.169491525423728e-08,
|
| 19361 |
+
"loss": 1.18,
|
| 19362 |
+
"step": 27590
|
| 19363 |
+
},
|
| 19364 |
+
{
|
| 19365 |
+
"epoch": 0.69,
|
| 19366 |
+
"grad_norm": 66.5,
|
| 19367 |
+
"learning_rate": 8.135593220338982e-08,
|
| 19368 |
+
"loss": 1.165,
|
| 19369 |
+
"step": 27600
|
| 19370 |
+
},
|
| 19371 |
+
{
|
| 19372 |
+
"epoch": 0.69,
|
| 19373 |
+
"grad_norm": 64.5,
|
| 19374 |
+
"learning_rate": 8.101694915254237e-08,
|
| 19375 |
+
"loss": 1.2114,
|
| 19376 |
+
"step": 27610
|
| 19377 |
+
},
|
| 19378 |
+
{
|
| 19379 |
+
"epoch": 0.69,
|
| 19380 |
+
"grad_norm": 64.5,
|
| 19381 |
+
"learning_rate": 8.067796610169492e-08,
|
| 19382 |
+
"loss": 1.1631,
|
| 19383 |
+
"step": 27620
|
| 19384 |
+
},
|
| 19385 |
+
{
|
| 19386 |
+
"epoch": 0.69,
|
| 19387 |
+
"grad_norm": 72.0,
|
| 19388 |
+
"learning_rate": 8.033898305084745e-08,
|
| 19389 |
+
"loss": 1.1922,
|
| 19390 |
+
"step": 27630
|
| 19391 |
+
},
|
| 19392 |
+
{
|
| 19393 |
+
"epoch": 0.69,
|
| 19394 |
+
"grad_norm": 66.0,
|
| 19395 |
+
"learning_rate": 8e-08,
|
| 19396 |
+
"loss": 1.1569,
|
| 19397 |
+
"step": 27640
|
| 19398 |
+
},
|
| 19399 |
+
{
|
| 19400 |
+
"epoch": 0.69,
|
| 19401 |
+
"grad_norm": 64.5,
|
| 19402 |
+
"learning_rate": 7.966101694915253e-08,
|
| 19403 |
+
"loss": 1.1786,
|
| 19404 |
+
"step": 27650
|
| 19405 |
+
},
|
| 19406 |
+
{
|
| 19407 |
+
"epoch": 0.69,
|
| 19408 |
+
"grad_norm": 67.0,
|
| 19409 |
+
"learning_rate": 7.932203389830508e-08,
|
| 19410 |
+
"loss": 1.1804,
|
| 19411 |
+
"step": 27660
|
| 19412 |
+
},
|
| 19413 |
+
{
|
| 19414 |
+
"epoch": 0.69,
|
| 19415 |
+
"grad_norm": 66.5,
|
| 19416 |
+
"learning_rate": 7.898305084745762e-08,
|
| 19417 |
+
"loss": 1.1891,
|
| 19418 |
+
"step": 27670
|
| 19419 |
+
},
|
| 19420 |
+
{
|
| 19421 |
+
"epoch": 0.69,
|
| 19422 |
+
"grad_norm": 66.0,
|
| 19423 |
+
"learning_rate": 7.864406779661017e-08,
|
| 19424 |
+
"loss": 1.1856,
|
| 19425 |
+
"step": 27680
|
| 19426 |
+
},
|
| 19427 |
+
{
|
| 19428 |
+
"epoch": 0.69,
|
| 19429 |
+
"grad_norm": 67.5,
|
| 19430 |
+
"learning_rate": 7.83050847457627e-08,
|
| 19431 |
+
"loss": 1.1739,
|
| 19432 |
+
"step": 27690
|
| 19433 |
+
},
|
| 19434 |
+
{
|
| 19435 |
+
"epoch": 0.69,
|
| 19436 |
+
"grad_norm": 68.0,
|
| 19437 |
+
"learning_rate": 7.796610169491526e-08,
|
| 19438 |
+
"loss": 1.1885,
|
| 19439 |
+
"step": 27700
|
| 19440 |
+
},
|
| 19441 |
+
{
|
| 19442 |
+
"epoch": 0.69,
|
| 19443 |
+
"grad_norm": 67.0,
|
| 19444 |
+
"learning_rate": 7.762711864406779e-08,
|
| 19445 |
+
"loss": 1.1659,
|
| 19446 |
+
"step": 27710
|
| 19447 |
+
},
|
| 19448 |
+
{
|
| 19449 |
+
"epoch": 0.69,
|
| 19450 |
+
"grad_norm": 64.0,
|
| 19451 |
+
"learning_rate": 7.728813559322034e-08,
|
| 19452 |
+
"loss": 1.1677,
|
| 19453 |
+
"step": 27720
|
| 19454 |
+
},
|
| 19455 |
+
{
|
| 19456 |
+
"epoch": 0.69,
|
| 19457 |
+
"grad_norm": 66.0,
|
| 19458 |
+
"learning_rate": 7.694915254237287e-08,
|
| 19459 |
+
"loss": 1.2044,
|
| 19460 |
+
"step": 27730
|
| 19461 |
+
},
|
| 19462 |
+
{
|
| 19463 |
+
"epoch": 0.69,
|
| 19464 |
+
"grad_norm": 65.5,
|
| 19465 |
+
"learning_rate": 7.661016949152542e-08,
|
| 19466 |
+
"loss": 1.16,
|
| 19467 |
+
"step": 27740
|
| 19468 |
+
},
|
| 19469 |
+
{
|
| 19470 |
+
"epoch": 0.69,
|
| 19471 |
+
"grad_norm": 66.5,
|
| 19472 |
+
"learning_rate": 7.627118644067796e-08,
|
| 19473 |
+
"loss": 1.1727,
|
| 19474 |
+
"step": 27750
|
| 19475 |
+
},
|
| 19476 |
+
{
|
| 19477 |
+
"epoch": 0.69,
|
| 19478 |
+
"grad_norm": 66.0,
|
| 19479 |
+
"learning_rate": 7.59322033898305e-08,
|
| 19480 |
+
"loss": 1.1746,
|
| 19481 |
+
"step": 27760
|
| 19482 |
+
},
|
| 19483 |
+
{
|
| 19484 |
+
"epoch": 0.69,
|
| 19485 |
+
"grad_norm": 66.0,
|
| 19486 |
+
"learning_rate": 7.559322033898305e-08,
|
| 19487 |
+
"loss": 1.1771,
|
| 19488 |
+
"step": 27770
|
| 19489 |
+
},
|
| 19490 |
+
{
|
| 19491 |
+
"epoch": 0.69,
|
| 19492 |
+
"grad_norm": 65.0,
|
| 19493 |
+
"learning_rate": 7.525423728813559e-08,
|
| 19494 |
+
"loss": 1.1711,
|
| 19495 |
+
"step": 27780
|
| 19496 |
+
},
|
| 19497 |
+
{
|
| 19498 |
+
"epoch": 0.69,
|
| 19499 |
+
"grad_norm": 64.0,
|
| 19500 |
+
"learning_rate": 7.491525423728813e-08,
|
| 19501 |
+
"loss": 1.1563,
|
| 19502 |
+
"step": 27790
|
| 19503 |
+
},
|
| 19504 |
+
{
|
| 19505 |
+
"epoch": 0.69,
|
| 19506 |
+
"grad_norm": 69.0,
|
| 19507 |
+
"learning_rate": 7.457627118644068e-08,
|
| 19508 |
+
"loss": 1.1984,
|
| 19509 |
+
"step": 27800
|
| 19510 |
+
},
|
| 19511 |
+
{
|
| 19512 |
+
"epoch": 0.7,
|
| 19513 |
+
"grad_norm": 67.0,
|
| 19514 |
+
"learning_rate": 7.423728813559321e-08,
|
| 19515 |
+
"loss": 1.181,
|
| 19516 |
+
"step": 27810
|
| 19517 |
+
},
|
| 19518 |
+
{
|
| 19519 |
+
"epoch": 0.7,
|
| 19520 |
+
"grad_norm": 71.0,
|
| 19521 |
+
"learning_rate": 7.389830508474576e-08,
|
| 19522 |
+
"loss": 1.1896,
|
| 19523 |
+
"step": 27820
|
| 19524 |
+
},
|
| 19525 |
+
{
|
| 19526 |
+
"epoch": 0.7,
|
| 19527 |
+
"grad_norm": 64.0,
|
| 19528 |
+
"learning_rate": 7.35593220338983e-08,
|
| 19529 |
+
"loss": 1.1725,
|
| 19530 |
+
"step": 27830
|
| 19531 |
+
},
|
| 19532 |
+
{
|
| 19533 |
+
"epoch": 0.7,
|
| 19534 |
+
"grad_norm": 70.5,
|
| 19535 |
+
"learning_rate": 7.322033898305085e-08,
|
| 19536 |
+
"loss": 1.1948,
|
| 19537 |
+
"step": 27840
|
| 19538 |
+
},
|
| 19539 |
+
{
|
| 19540 |
+
"epoch": 0.7,
|
| 19541 |
+
"grad_norm": 67.0,
|
| 19542 |
+
"learning_rate": 7.288135593220339e-08,
|
| 19543 |
+
"loss": 1.1976,
|
| 19544 |
+
"step": 27850
|
| 19545 |
+
},
|
| 19546 |
+
{
|
| 19547 |
+
"epoch": 0.7,
|
| 19548 |
+
"grad_norm": 65.0,
|
| 19549 |
+
"learning_rate": 7.254237288135593e-08,
|
| 19550 |
+
"loss": 1.1856,
|
| 19551 |
+
"step": 27860
|
| 19552 |
+
},
|
| 19553 |
+
{
|
| 19554 |
+
"epoch": 0.7,
|
| 19555 |
+
"grad_norm": 65.5,
|
| 19556 |
+
"learning_rate": 7.220338983050847e-08,
|
| 19557 |
+
"loss": 1.1882,
|
| 19558 |
+
"step": 27870
|
| 19559 |
+
},
|
| 19560 |
+
{
|
| 19561 |
+
"epoch": 0.7,
|
| 19562 |
+
"grad_norm": 66.0,
|
| 19563 |
+
"learning_rate": 7.186440677966102e-08,
|
| 19564 |
+
"loss": 1.1767,
|
| 19565 |
+
"step": 27880
|
| 19566 |
+
},
|
| 19567 |
+
{
|
| 19568 |
+
"epoch": 0.7,
|
| 19569 |
+
"grad_norm": 66.0,
|
| 19570 |
+
"learning_rate": 7.152542372881355e-08,
|
| 19571 |
+
"loss": 1.1746,
|
| 19572 |
+
"step": 27890
|
| 19573 |
+
},
|
| 19574 |
+
{
|
| 19575 |
+
"epoch": 0.7,
|
| 19576 |
+
"grad_norm": 66.5,
|
| 19577 |
+
"learning_rate": 7.11864406779661e-08,
|
| 19578 |
+
"loss": 1.1752,
|
| 19579 |
+
"step": 27900
|
| 19580 |
+
},
|
| 19581 |
+
{
|
| 19582 |
+
"epoch": 0.7,
|
| 19583 |
+
"grad_norm": 64.0,
|
| 19584 |
+
"learning_rate": 7.084745762711863e-08,
|
| 19585 |
+
"loss": 1.1606,
|
| 19586 |
+
"step": 27910
|
| 19587 |
+
},
|
| 19588 |
+
{
|
| 19589 |
+
"epoch": 0.7,
|
| 19590 |
+
"grad_norm": 71.0,
|
| 19591 |
+
"learning_rate": 7.050847457627119e-08,
|
| 19592 |
+
"loss": 1.1768,
|
| 19593 |
+
"step": 27920
|
| 19594 |
+
},
|
| 19595 |
+
{
|
| 19596 |
+
"epoch": 0.7,
|
| 19597 |
+
"grad_norm": 66.0,
|
| 19598 |
+
"learning_rate": 7.016949152542373e-08,
|
| 19599 |
+
"loss": 1.1674,
|
| 19600 |
+
"step": 27930
|
| 19601 |
+
},
|
| 19602 |
+
{
|
| 19603 |
+
"epoch": 0.7,
|
| 19604 |
+
"grad_norm": 66.5,
|
| 19605 |
+
"learning_rate": 6.983050847457627e-08,
|
| 19606 |
+
"loss": 1.1703,
|
| 19607 |
+
"step": 27940
|
| 19608 |
+
},
|
| 19609 |
+
{
|
| 19610 |
+
"epoch": 0.7,
|
| 19611 |
+
"grad_norm": 69.0,
|
| 19612 |
+
"learning_rate": 6.949152542372881e-08,
|
| 19613 |
+
"loss": 1.169,
|
| 19614 |
+
"step": 27950
|
| 19615 |
+
},
|
| 19616 |
+
{
|
| 19617 |
+
"epoch": 0.7,
|
| 19618 |
+
"grad_norm": 66.5,
|
| 19619 |
+
"learning_rate": 6.915254237288136e-08,
|
| 19620 |
+
"loss": 1.1728,
|
| 19621 |
+
"step": 27960
|
| 19622 |
+
},
|
| 19623 |
+
{
|
| 19624 |
+
"epoch": 0.7,
|
| 19625 |
+
"grad_norm": 72.0,
|
| 19626 |
+
"learning_rate": 6.881355932203389e-08,
|
| 19627 |
+
"loss": 1.1645,
|
| 19628 |
+
"step": 27970
|
| 19629 |
+
},
|
| 19630 |
+
{
|
| 19631 |
+
"epoch": 0.7,
|
| 19632 |
+
"grad_norm": 66.5,
|
| 19633 |
+
"learning_rate": 6.847457627118645e-08,
|
| 19634 |
+
"loss": 1.1994,
|
| 19635 |
+
"step": 27980
|
| 19636 |
+
},
|
| 19637 |
+
{
|
| 19638 |
+
"epoch": 0.7,
|
| 19639 |
+
"grad_norm": 67.0,
|
| 19640 |
+
"learning_rate": 6.813559322033897e-08,
|
| 19641 |
+
"loss": 1.1788,
|
| 19642 |
+
"step": 27990
|
| 19643 |
+
},
|
| 19644 |
+
{
|
| 19645 |
+
"epoch": 0.7,
|
| 19646 |
+
"grad_norm": 66.5,
|
| 19647 |
+
"learning_rate": 6.779661016949153e-08,
|
| 19648 |
+
"loss": 1.1951,
|
| 19649 |
+
"step": 28000
|
| 19650 |
+
},
|
| 19651 |
+
{
|
| 19652 |
+
"epoch": 0.7,
|
| 19653 |
+
"grad_norm": 67.5,
|
| 19654 |
+
"learning_rate": 6.745762711864407e-08,
|
| 19655 |
+
"loss": 1.1631,
|
| 19656 |
+
"step": 28010
|
| 19657 |
+
},
|
| 19658 |
+
{
|
| 19659 |
+
"epoch": 0.7,
|
| 19660 |
+
"grad_norm": 63.75,
|
| 19661 |
+
"learning_rate": 6.711864406779661e-08,
|
| 19662 |
+
"loss": 1.1755,
|
| 19663 |
+
"step": 28020
|
| 19664 |
+
},
|
| 19665 |
+
{
|
| 19666 |
+
"epoch": 0.7,
|
| 19667 |
+
"grad_norm": 64.5,
|
| 19668 |
+
"learning_rate": 6.677966101694915e-08,
|
| 19669 |
+
"loss": 1.1843,
|
| 19670 |
+
"step": 28030
|
| 19671 |
+
},
|
| 19672 |
+
{
|
| 19673 |
+
"epoch": 0.7,
|
| 19674 |
+
"grad_norm": 67.0,
|
| 19675 |
+
"learning_rate": 6.644067796610169e-08,
|
| 19676 |
+
"loss": 1.1787,
|
| 19677 |
+
"step": 28040
|
| 19678 |
+
},
|
| 19679 |
+
{
|
| 19680 |
+
"epoch": 0.7,
|
| 19681 |
+
"grad_norm": 69.5,
|
| 19682 |
+
"learning_rate": 6.610169491525423e-08,
|
| 19683 |
+
"loss": 1.1746,
|
| 19684 |
+
"step": 28050
|
| 19685 |
+
},
|
| 19686 |
+
{
|
| 19687 |
+
"epoch": 0.7,
|
| 19688 |
+
"grad_norm": 68.0,
|
| 19689 |
+
"learning_rate": 6.576271186440679e-08,
|
| 19690 |
+
"loss": 1.1861,
|
| 19691 |
+
"step": 28060
|
| 19692 |
+
},
|
| 19693 |
+
{
|
| 19694 |
+
"epoch": 0.7,
|
| 19695 |
+
"grad_norm": 66.0,
|
| 19696 |
+
"learning_rate": 6.542372881355931e-08,
|
| 19697 |
+
"loss": 1.1767,
|
| 19698 |
+
"step": 28070
|
| 19699 |
+
},
|
| 19700 |
+
{
|
| 19701 |
+
"epoch": 0.7,
|
| 19702 |
+
"grad_norm": 67.5,
|
| 19703 |
+
"learning_rate": 6.508474576271187e-08,
|
| 19704 |
+
"loss": 1.1804,
|
| 19705 |
+
"step": 28080
|
| 19706 |
+
},
|
| 19707 |
+
{
|
| 19708 |
+
"epoch": 0.7,
|
| 19709 |
+
"grad_norm": 64.5,
|
| 19710 |
+
"learning_rate": 6.474576271186441e-08,
|
| 19711 |
+
"loss": 1.1867,
|
| 19712 |
+
"step": 28090
|
| 19713 |
+
},
|
| 19714 |
+
{
|
| 19715 |
+
"epoch": 0.7,
|
| 19716 |
+
"grad_norm": 68.0,
|
| 19717 |
+
"learning_rate": 6.440677966101695e-08,
|
| 19718 |
+
"loss": 1.1565,
|
| 19719 |
+
"step": 28100
|
| 19720 |
+
},
|
| 19721 |
+
{
|
| 19722 |
+
"epoch": 0.7,
|
| 19723 |
+
"grad_norm": 68.0,
|
| 19724 |
+
"learning_rate": 6.406779661016949e-08,
|
| 19725 |
+
"loss": 1.1659,
|
| 19726 |
+
"step": 28110
|
| 19727 |
+
},
|
| 19728 |
+
{
|
| 19729 |
+
"epoch": 0.7,
|
| 19730 |
+
"grad_norm": 66.0,
|
| 19731 |
+
"learning_rate": 6.372881355932203e-08,
|
| 19732 |
+
"loss": 1.1844,
|
| 19733 |
+
"step": 28120
|
| 19734 |
+
},
|
| 19735 |
+
{
|
| 19736 |
+
"epoch": 0.7,
|
| 19737 |
+
"grad_norm": 66.0,
|
| 19738 |
+
"learning_rate": 6.338983050847457e-08,
|
| 19739 |
+
"loss": 1.1578,
|
| 19740 |
+
"step": 28130
|
| 19741 |
+
},
|
| 19742 |
+
{
|
| 19743 |
+
"epoch": 0.7,
|
| 19744 |
+
"grad_norm": 66.5,
|
| 19745 |
+
"learning_rate": 6.305084745762713e-08,
|
| 19746 |
+
"loss": 1.1916,
|
| 19747 |
+
"step": 28140
|
| 19748 |
+
},
|
| 19749 |
+
{
|
| 19750 |
+
"epoch": 0.7,
|
| 19751 |
+
"grad_norm": 68.5,
|
| 19752 |
+
"learning_rate": 6.271186440677965e-08,
|
| 19753 |
+
"loss": 1.1999,
|
| 19754 |
+
"step": 28150
|
| 19755 |
+
},
|
| 19756 |
+
{
|
| 19757 |
+
"epoch": 0.7,
|
| 19758 |
+
"grad_norm": 68.0,
|
| 19759 |
+
"learning_rate": 6.237288135593221e-08,
|
| 19760 |
+
"loss": 1.1608,
|
| 19761 |
+
"step": 28160
|
| 19762 |
+
},
|
| 19763 |
+
{
|
| 19764 |
+
"epoch": 0.7,
|
| 19765 |
+
"grad_norm": 65.0,
|
| 19766 |
+
"learning_rate": 6.203389830508475e-08,
|
| 19767 |
+
"loss": 1.1643,
|
| 19768 |
+
"step": 28170
|
| 19769 |
+
},
|
| 19770 |
+
{
|
| 19771 |
+
"epoch": 0.7,
|
| 19772 |
+
"grad_norm": 68.0,
|
| 19773 |
+
"learning_rate": 6.169491525423729e-08,
|
| 19774 |
+
"loss": 1.1638,
|
| 19775 |
+
"step": 28180
|
| 19776 |
+
},
|
| 19777 |
+
{
|
| 19778 |
+
"epoch": 0.7,
|
| 19779 |
+
"grad_norm": 68.0,
|
| 19780 |
+
"learning_rate": 6.135593220338983e-08,
|
| 19781 |
+
"loss": 1.191,
|
| 19782 |
+
"step": 28190
|
| 19783 |
+
},
|
| 19784 |
+
{
|
| 19785 |
+
"epoch": 0.7,
|
| 19786 |
+
"grad_norm": 66.5,
|
| 19787 |
+
"learning_rate": 6.101694915254237e-08,
|
| 19788 |
+
"loss": 1.162,
|
| 19789 |
+
"step": 28200
|
| 19790 |
+
},
|
| 19791 |
+
{
|
| 19792 |
+
"epoch": 0.71,
|
| 19793 |
+
"grad_norm": 67.0,
|
| 19794 |
+
"learning_rate": 6.067796610169491e-08,
|
| 19795 |
+
"loss": 1.1963,
|
| 19796 |
+
"step": 28210
|
| 19797 |
+
},
|
| 19798 |
+
{
|
| 19799 |
+
"epoch": 0.71,
|
| 19800 |
+
"grad_norm": 65.5,
|
| 19801 |
+
"learning_rate": 6.033898305084745e-08,
|
| 19802 |
+
"loss": 1.1595,
|
| 19803 |
+
"step": 28220
|
| 19804 |
+
},
|
| 19805 |
+
{
|
| 19806 |
+
"epoch": 0.71,
|
| 19807 |
+
"grad_norm": 66.0,
|
| 19808 |
+
"learning_rate": 6e-08,
|
| 19809 |
+
"loss": 1.1813,
|
| 19810 |
+
"step": 28230
|
| 19811 |
+
},
|
| 19812 |
+
{
|
| 19813 |
+
"epoch": 0.71,
|
| 19814 |
+
"grad_norm": 64.5,
|
| 19815 |
+
"learning_rate": 5.966101694915255e-08,
|
| 19816 |
+
"loss": 1.1716,
|
| 19817 |
+
"step": 28240
|
| 19818 |
+
},
|
| 19819 |
+
{
|
| 19820 |
+
"epoch": 0.71,
|
| 19821 |
+
"grad_norm": 67.5,
|
| 19822 |
+
"learning_rate": 5.932203389830508e-08,
|
| 19823 |
+
"loss": 1.1809,
|
| 19824 |
+
"step": 28250
|
| 19825 |
+
},
|
| 19826 |
+
{
|
| 19827 |
+
"epoch": 0.71,
|
| 19828 |
+
"grad_norm": 64.0,
|
| 19829 |
+
"learning_rate": 5.8983050847457624e-08,
|
| 19830 |
+
"loss": 1.176,
|
| 19831 |
+
"step": 28260
|
| 19832 |
+
},
|
| 19833 |
+
{
|
| 19834 |
+
"epoch": 0.71,
|
| 19835 |
+
"grad_norm": 67.0,
|
| 19836 |
+
"learning_rate": 5.8644067796610165e-08,
|
| 19837 |
+
"loss": 1.1815,
|
| 19838 |
+
"step": 28270
|
| 19839 |
+
},
|
| 19840 |
+
{
|
| 19841 |
+
"epoch": 0.71,
|
| 19842 |
+
"grad_norm": 63.75,
|
| 19843 |
+
"learning_rate": 5.8305084745762706e-08,
|
| 19844 |
+
"loss": 1.1776,
|
| 19845 |
+
"step": 28280
|
| 19846 |
+
},
|
| 19847 |
+
{
|
| 19848 |
+
"epoch": 0.71,
|
| 19849 |
+
"grad_norm": 65.0,
|
| 19850 |
+
"learning_rate": 5.7966101694915253e-08,
|
| 19851 |
+
"loss": 1.1663,
|
| 19852 |
+
"step": 28290
|
| 19853 |
+
},
|
| 19854 |
+
{
|
| 19855 |
+
"epoch": 0.71,
|
| 19856 |
+
"grad_norm": 67.0,
|
| 19857 |
+
"learning_rate": 5.7627118644067794e-08,
|
| 19858 |
+
"loss": 1.1622,
|
| 19859 |
+
"step": 28300
|
| 19860 |
+
},
|
| 19861 |
+
{
|
| 19862 |
+
"epoch": 0.71,
|
| 19863 |
+
"grad_norm": 65.0,
|
| 19864 |
+
"learning_rate": 5.7288135593220335e-08,
|
| 19865 |
+
"loss": 1.1581,
|
| 19866 |
+
"step": 28310
|
| 19867 |
+
},
|
| 19868 |
+
{
|
| 19869 |
+
"epoch": 0.71,
|
| 19870 |
+
"grad_norm": 65.0,
|
| 19871 |
+
"learning_rate": 5.6949152542372876e-08,
|
| 19872 |
+
"loss": 1.16,
|
| 19873 |
+
"step": 28320
|
| 19874 |
+
},
|
| 19875 |
+
{
|
| 19876 |
+
"epoch": 0.71,
|
| 19877 |
+
"grad_norm": 66.5,
|
| 19878 |
+
"learning_rate": 5.6610169491525424e-08,
|
| 19879 |
+
"loss": 1.1714,
|
| 19880 |
+
"step": 28330
|
| 19881 |
+
},
|
| 19882 |
+
{
|
| 19883 |
+
"epoch": 0.71,
|
| 19884 |
+
"grad_norm": 65.5,
|
| 19885 |
+
"learning_rate": 5.6271186440677964e-08,
|
| 19886 |
+
"loss": 1.1625,
|
| 19887 |
+
"step": 28340
|
| 19888 |
+
},
|
| 19889 |
+
{
|
| 19890 |
+
"epoch": 0.71,
|
| 19891 |
+
"grad_norm": 66.0,
|
| 19892 |
+
"learning_rate": 5.5932203389830505e-08,
|
| 19893 |
+
"loss": 1.1681,
|
| 19894 |
+
"step": 28350
|
| 19895 |
+
},
|
| 19896 |
+
{
|
| 19897 |
+
"epoch": 0.71,
|
| 19898 |
+
"grad_norm": 66.0,
|
| 19899 |
+
"learning_rate": 5.5593220338983046e-08,
|
| 19900 |
+
"loss": 1.168,
|
| 19901 |
+
"step": 28360
|
| 19902 |
+
},
|
| 19903 |
+
{
|
| 19904 |
+
"epoch": 0.71,
|
| 19905 |
+
"grad_norm": 66.5,
|
| 19906 |
+
"learning_rate": 5.5254237288135594e-08,
|
| 19907 |
+
"loss": 1.1839,
|
| 19908 |
+
"step": 28370
|
| 19909 |
+
},
|
| 19910 |
+
{
|
| 19911 |
+
"epoch": 0.71,
|
| 19912 |
+
"grad_norm": 67.0,
|
| 19913 |
+
"learning_rate": 5.4915254237288135e-08,
|
| 19914 |
+
"loss": 1.1759,
|
| 19915 |
+
"step": 28380
|
| 19916 |
+
},
|
| 19917 |
+
{
|
| 19918 |
+
"epoch": 0.71,
|
| 19919 |
+
"grad_norm": 70.0,
|
| 19920 |
+
"learning_rate": 5.4576271186440676e-08,
|
| 19921 |
+
"loss": 1.1736,
|
| 19922 |
+
"step": 28390
|
| 19923 |
+
},
|
| 19924 |
+
{
|
| 19925 |
+
"epoch": 0.71,
|
| 19926 |
+
"grad_norm": 67.0,
|
| 19927 |
+
"learning_rate": 5.4237288135593217e-08,
|
| 19928 |
+
"loss": 1.1766,
|
| 19929 |
+
"step": 28400
|
| 19930 |
+
},
|
| 19931 |
+
{
|
| 19932 |
+
"epoch": 0.71,
|
| 19933 |
+
"grad_norm": 66.0,
|
| 19934 |
+
"learning_rate": 5.389830508474576e-08,
|
| 19935 |
+
"loss": 1.1752,
|
| 19936 |
+
"step": 28410
|
| 19937 |
+
},
|
| 19938 |
+
{
|
| 19939 |
+
"epoch": 0.71,
|
| 19940 |
+
"grad_norm": 63.25,
|
| 19941 |
+
"learning_rate": 5.3559322033898305e-08,
|
| 19942 |
+
"loss": 1.1758,
|
| 19943 |
+
"step": 28420
|
| 19944 |
+
},
|
| 19945 |
+
{
|
| 19946 |
+
"epoch": 0.71,
|
| 19947 |
+
"grad_norm": 68.0,
|
| 19948 |
+
"learning_rate": 5.3220338983050846e-08,
|
| 19949 |
+
"loss": 1.1935,
|
| 19950 |
+
"step": 28430
|
| 19951 |
+
},
|
| 19952 |
+
{
|
| 19953 |
+
"epoch": 0.71,
|
| 19954 |
+
"grad_norm": 68.0,
|
| 19955 |
+
"learning_rate": 5.288135593220339e-08,
|
| 19956 |
+
"loss": 1.1833,
|
| 19957 |
+
"step": 28440
|
| 19958 |
+
},
|
| 19959 |
+
{
|
| 19960 |
+
"epoch": 0.71,
|
| 19961 |
+
"grad_norm": 69.5,
|
| 19962 |
+
"learning_rate": 5.254237288135593e-08,
|
| 19963 |
+
"loss": 1.1751,
|
| 19964 |
+
"step": 28450
|
| 19965 |
+
},
|
| 19966 |
+
{
|
| 19967 |
+
"epoch": 0.71,
|
| 19968 |
+
"grad_norm": 67.5,
|
| 19969 |
+
"learning_rate": 5.2203389830508475e-08,
|
| 19970 |
+
"loss": 1.1774,
|
| 19971 |
+
"step": 28460
|
| 19972 |
+
},
|
| 19973 |
+
{
|
| 19974 |
+
"epoch": 0.71,
|
| 19975 |
+
"grad_norm": 68.5,
|
| 19976 |
+
"learning_rate": 5.1864406779661016e-08,
|
| 19977 |
+
"loss": 1.1812,
|
| 19978 |
+
"step": 28470
|
| 19979 |
+
},
|
| 19980 |
+
{
|
| 19981 |
+
"epoch": 0.71,
|
| 19982 |
+
"grad_norm": 64.5,
|
| 19983 |
+
"learning_rate": 5.152542372881356e-08,
|
| 19984 |
+
"loss": 1.1931,
|
| 19985 |
+
"step": 28480
|
| 19986 |
+
},
|
| 19987 |
+
{
|
| 19988 |
+
"epoch": 0.71,
|
| 19989 |
+
"grad_norm": 68.0,
|
| 19990 |
+
"learning_rate": 5.11864406779661e-08,
|
| 19991 |
+
"loss": 1.2,
|
| 19992 |
+
"step": 28490
|
| 19993 |
+
},
|
| 19994 |
+
{
|
| 19995 |
+
"epoch": 0.71,
|
| 19996 |
+
"grad_norm": 65.5,
|
| 19997 |
+
"learning_rate": 5.0847457627118645e-08,
|
| 19998 |
+
"loss": 1.1767,
|
| 19999 |
+
"step": 28500
|
| 20000 |
+
},
|
| 20001 |
+
{
|
| 20002 |
+
"epoch": 0.71,
|
| 20003 |
+
"grad_norm": 68.0,
|
| 20004 |
+
"learning_rate": 5.0508474576271186e-08,
|
| 20005 |
+
"loss": 1.1891,
|
| 20006 |
+
"step": 28510
|
| 20007 |
+
},
|
| 20008 |
+
{
|
| 20009 |
+
"epoch": 0.71,
|
| 20010 |
+
"grad_norm": 68.0,
|
| 20011 |
+
"learning_rate": 5.016949152542373e-08,
|
| 20012 |
+
"loss": 1.1878,
|
| 20013 |
+
"step": 28520
|
| 20014 |
+
},
|
| 20015 |
+
{
|
| 20016 |
+
"epoch": 0.71,
|
| 20017 |
+
"grad_norm": 66.0,
|
| 20018 |
+
"learning_rate": 4.983050847457627e-08,
|
| 20019 |
+
"loss": 1.1729,
|
| 20020 |
+
"step": 28530
|
| 20021 |
+
},
|
| 20022 |
+
{
|
| 20023 |
+
"epoch": 0.71,
|
| 20024 |
+
"grad_norm": 67.0,
|
| 20025 |
+
"learning_rate": 4.949152542372881e-08,
|
| 20026 |
+
"loss": 1.189,
|
| 20027 |
+
"step": 28540
|
| 20028 |
+
},
|
| 20029 |
+
{
|
| 20030 |
+
"epoch": 0.71,
|
| 20031 |
+
"grad_norm": 66.0,
|
| 20032 |
+
"learning_rate": 4.9152542372881357e-08,
|
| 20033 |
+
"loss": 1.1898,
|
| 20034 |
+
"step": 28550
|
| 20035 |
+
},
|
| 20036 |
+
{
|
| 20037 |
+
"epoch": 0.71,
|
| 20038 |
+
"grad_norm": 68.5,
|
| 20039 |
+
"learning_rate": 4.88135593220339e-08,
|
| 20040 |
+
"loss": 1.1752,
|
| 20041 |
+
"step": 28560
|
| 20042 |
+
},
|
| 20043 |
+
{
|
| 20044 |
+
"epoch": 0.71,
|
| 20045 |
+
"grad_norm": 63.75,
|
| 20046 |
+
"learning_rate": 4.847457627118644e-08,
|
| 20047 |
+
"loss": 1.1693,
|
| 20048 |
+
"step": 28570
|
| 20049 |
+
},
|
| 20050 |
+
{
|
| 20051 |
+
"epoch": 0.71,
|
| 20052 |
+
"grad_norm": 66.5,
|
| 20053 |
+
"learning_rate": 4.813559322033898e-08,
|
| 20054 |
+
"loss": 1.1751,
|
| 20055 |
+
"step": 28580
|
| 20056 |
+
},
|
| 20057 |
+
{
|
| 20058 |
+
"epoch": 0.71,
|
| 20059 |
+
"grad_norm": 65.0,
|
| 20060 |
+
"learning_rate": 4.779661016949153e-08,
|
| 20061 |
+
"loss": 1.18,
|
| 20062 |
+
"step": 28590
|
| 20063 |
+
},
|
| 20064 |
+
{
|
| 20065 |
+
"epoch": 0.71,
|
| 20066 |
+
"grad_norm": 67.5,
|
| 20067 |
+
"learning_rate": 4.745762711864407e-08,
|
| 20068 |
+
"loss": 1.168,
|
| 20069 |
+
"step": 28600
|
| 20070 |
+
},
|
| 20071 |
+
{
|
| 20072 |
+
"epoch": 0.72,
|
| 20073 |
+
"grad_norm": 67.0,
|
| 20074 |
+
"learning_rate": 4.711864406779661e-08,
|
| 20075 |
+
"loss": 1.154,
|
| 20076 |
+
"step": 28610
|
| 20077 |
+
},
|
| 20078 |
+
{
|
| 20079 |
+
"epoch": 0.72,
|
| 20080 |
+
"grad_norm": 64.0,
|
| 20081 |
+
"learning_rate": 4.677966101694915e-08,
|
| 20082 |
+
"loss": 1.1494,
|
| 20083 |
+
"step": 28620
|
| 20084 |
+
},
|
| 20085 |
+
{
|
| 20086 |
+
"epoch": 0.72,
|
| 20087 |
+
"grad_norm": 71.0,
|
| 20088 |
+
"learning_rate": 4.64406779661017e-08,
|
| 20089 |
+
"loss": 1.1787,
|
| 20090 |
+
"step": 28630
|
| 20091 |
+
},
|
| 20092 |
+
{
|
| 20093 |
+
"epoch": 0.72,
|
| 20094 |
+
"grad_norm": 65.5,
|
| 20095 |
+
"learning_rate": 4.610169491525424e-08,
|
| 20096 |
+
"loss": 1.1676,
|
| 20097 |
+
"step": 28640
|
| 20098 |
+
},
|
| 20099 |
+
{
|
| 20100 |
+
"epoch": 0.72,
|
| 20101 |
+
"grad_norm": 66.5,
|
| 20102 |
+
"learning_rate": 4.576271186440678e-08,
|
| 20103 |
+
"loss": 1.1857,
|
| 20104 |
+
"step": 28650
|
| 20105 |
+
},
|
| 20106 |
+
{
|
| 20107 |
+
"epoch": 0.72,
|
| 20108 |
+
"grad_norm": 68.5,
|
| 20109 |
+
"learning_rate": 4.542372881355932e-08,
|
| 20110 |
+
"loss": 1.1773,
|
| 20111 |
+
"step": 28660
|
| 20112 |
+
},
|
| 20113 |
+
{
|
| 20114 |
+
"epoch": 0.72,
|
| 20115 |
+
"grad_norm": 68.0,
|
| 20116 |
+
"learning_rate": 4.508474576271186e-08,
|
| 20117 |
+
"loss": 1.194,
|
| 20118 |
+
"step": 28670
|
| 20119 |
+
},
|
| 20120 |
+
{
|
| 20121 |
+
"epoch": 0.72,
|
| 20122 |
+
"grad_norm": 65.0,
|
| 20123 |
+
"learning_rate": 4.474576271186441e-08,
|
| 20124 |
+
"loss": 1.1772,
|
| 20125 |
+
"step": 28680
|
| 20126 |
+
},
|
| 20127 |
+
{
|
| 20128 |
+
"epoch": 0.72,
|
| 20129 |
+
"grad_norm": 66.5,
|
| 20130 |
+
"learning_rate": 4.440677966101695e-08,
|
| 20131 |
+
"loss": 1.1646,
|
| 20132 |
+
"step": 28690
|
| 20133 |
+
},
|
| 20134 |
+
{
|
| 20135 |
+
"epoch": 0.72,
|
| 20136 |
+
"grad_norm": 63.5,
|
| 20137 |
+
"learning_rate": 4.406779661016949e-08,
|
| 20138 |
+
"loss": 1.18,
|
| 20139 |
+
"step": 28700
|
| 20140 |
+
},
|
| 20141 |
+
{
|
| 20142 |
+
"epoch": 0.72,
|
| 20143 |
+
"grad_norm": 68.5,
|
| 20144 |
+
"learning_rate": 4.372881355932203e-08,
|
| 20145 |
+
"loss": 1.1891,
|
| 20146 |
+
"step": 28710
|
| 20147 |
+
},
|
| 20148 |
+
{
|
| 20149 |
+
"epoch": 0.72,
|
| 20150 |
+
"grad_norm": 66.5,
|
| 20151 |
+
"learning_rate": 4.338983050847458e-08,
|
| 20152 |
+
"loss": 1.1779,
|
| 20153 |
+
"step": 28720
|
| 20154 |
+
},
|
| 20155 |
+
{
|
| 20156 |
+
"epoch": 0.72,
|
| 20157 |
+
"grad_norm": 64.5,
|
| 20158 |
+
"learning_rate": 4.305084745762712e-08,
|
| 20159 |
+
"loss": 1.1591,
|
| 20160 |
+
"step": 28730
|
| 20161 |
+
},
|
| 20162 |
+
{
|
| 20163 |
+
"epoch": 0.72,
|
| 20164 |
+
"grad_norm": 66.5,
|
| 20165 |
+
"learning_rate": 4.271186440677966e-08,
|
| 20166 |
+
"loss": 1.2068,
|
| 20167 |
+
"step": 28740
|
| 20168 |
+
},
|
| 20169 |
+
{
|
| 20170 |
+
"epoch": 0.72,
|
| 20171 |
+
"grad_norm": 67.5,
|
| 20172 |
+
"learning_rate": 4.23728813559322e-08,
|
| 20173 |
+
"loss": 1.2019,
|
| 20174 |
+
"step": 28750
|
| 20175 |
+
},
|
| 20176 |
+
{
|
| 20177 |
+
"epoch": 0.72,
|
| 20178 |
+
"grad_norm": 67.0,
|
| 20179 |
+
"learning_rate": 4.203389830508475e-08,
|
| 20180 |
+
"loss": 1.1812,
|
| 20181 |
+
"step": 28760
|
| 20182 |
+
},
|
| 20183 |
+
{
|
| 20184 |
+
"epoch": 0.72,
|
| 20185 |
+
"grad_norm": 68.5,
|
| 20186 |
+
"learning_rate": 4.169491525423729e-08,
|
| 20187 |
+
"loss": 1.1768,
|
| 20188 |
+
"step": 28770
|
| 20189 |
+
},
|
| 20190 |
+
{
|
| 20191 |
+
"epoch": 0.72,
|
| 20192 |
+
"grad_norm": 65.5,
|
| 20193 |
+
"learning_rate": 4.135593220338983e-08,
|
| 20194 |
+
"loss": 1.1845,
|
| 20195 |
+
"step": 28780
|
| 20196 |
+
},
|
| 20197 |
+
{
|
| 20198 |
+
"epoch": 0.72,
|
| 20199 |
+
"grad_norm": 66.5,
|
| 20200 |
+
"learning_rate": 4.101694915254237e-08,
|
| 20201 |
+
"loss": 1.2024,
|
| 20202 |
+
"step": 28790
|
| 20203 |
+
},
|
| 20204 |
+
{
|
| 20205 |
+
"epoch": 0.72,
|
| 20206 |
+
"grad_norm": 66.0,
|
| 20207 |
+
"learning_rate": 4.067796610169491e-08,
|
| 20208 |
+
"loss": 1.1843,
|
| 20209 |
+
"step": 28800
|
| 20210 |
+
},
|
| 20211 |
+
{
|
| 20212 |
+
"epoch": 0.72,
|
| 20213 |
+
"grad_norm": 66.0,
|
| 20214 |
+
"learning_rate": 4.033898305084746e-08,
|
| 20215 |
+
"loss": 1.1749,
|
| 20216 |
+
"step": 28810
|
| 20217 |
+
},
|
| 20218 |
+
{
|
| 20219 |
+
"epoch": 0.72,
|
| 20220 |
+
"grad_norm": 66.5,
|
| 20221 |
+
"learning_rate": 4e-08,
|
| 20222 |
+
"loss": 1.169,
|
| 20223 |
+
"step": 28820
|
| 20224 |
+
},
|
| 20225 |
+
{
|
| 20226 |
+
"epoch": 0.72,
|
| 20227 |
+
"grad_norm": 64.0,
|
| 20228 |
+
"learning_rate": 3.966101694915254e-08,
|
| 20229 |
+
"loss": 1.1715,
|
| 20230 |
+
"step": 28830
|
| 20231 |
+
},
|
| 20232 |
+
{
|
| 20233 |
+
"epoch": 0.72,
|
| 20234 |
+
"grad_norm": 64.5,
|
| 20235 |
+
"learning_rate": 3.932203389830508e-08,
|
| 20236 |
+
"loss": 1.1653,
|
| 20237 |
+
"step": 28840
|
| 20238 |
+
},
|
| 20239 |
+
{
|
| 20240 |
+
"epoch": 0.72,
|
| 20241 |
+
"grad_norm": 67.5,
|
| 20242 |
+
"learning_rate": 3.898305084745763e-08,
|
| 20243 |
+
"loss": 1.1682,
|
| 20244 |
+
"step": 28850
|
| 20245 |
+
},
|
| 20246 |
+
{
|
| 20247 |
+
"epoch": 0.72,
|
| 20248 |
+
"grad_norm": 66.5,
|
| 20249 |
+
"learning_rate": 3.864406779661017e-08,
|
| 20250 |
+
"loss": 1.183,
|
| 20251 |
+
"step": 28860
|
| 20252 |
+
},
|
| 20253 |
+
{
|
| 20254 |
+
"epoch": 0.72,
|
| 20255 |
+
"grad_norm": 68.5,
|
| 20256 |
+
"learning_rate": 3.830508474576271e-08,
|
| 20257 |
+
"loss": 1.1728,
|
| 20258 |
+
"step": 28870
|
| 20259 |
+
},
|
| 20260 |
+
{
|
| 20261 |
+
"epoch": 0.72,
|
| 20262 |
+
"grad_norm": 65.0,
|
| 20263 |
+
"learning_rate": 3.796610169491525e-08,
|
| 20264 |
+
"loss": 1.1596,
|
| 20265 |
+
"step": 28880
|
| 20266 |
+
},
|
| 20267 |
+
{
|
| 20268 |
+
"epoch": 0.72,
|
| 20269 |
+
"grad_norm": 65.5,
|
| 20270 |
+
"learning_rate": 3.7627118644067794e-08,
|
| 20271 |
+
"loss": 1.1782,
|
| 20272 |
+
"step": 28890
|
| 20273 |
+
},
|
| 20274 |
+
{
|
| 20275 |
+
"epoch": 0.72,
|
| 20276 |
+
"grad_norm": 67.0,
|
| 20277 |
+
"learning_rate": 3.728813559322034e-08,
|
| 20278 |
+
"loss": 1.1856,
|
| 20279 |
+
"step": 28900
|
| 20280 |
+
},
|
| 20281 |
+
{
|
| 20282 |
+
"epoch": 0.72,
|
| 20283 |
+
"grad_norm": 66.5,
|
| 20284 |
+
"learning_rate": 3.694915254237288e-08,
|
| 20285 |
+
"loss": 1.1533,
|
| 20286 |
+
"step": 28910
|
| 20287 |
+
},
|
| 20288 |
+
{
|
| 20289 |
+
"epoch": 0.72,
|
| 20290 |
+
"grad_norm": 65.0,
|
| 20291 |
+
"learning_rate": 3.661016949152542e-08,
|
| 20292 |
+
"loss": 1.1762,
|
| 20293 |
+
"step": 28920
|
| 20294 |
+
},
|
| 20295 |
+
{
|
| 20296 |
+
"epoch": 0.72,
|
| 20297 |
+
"grad_norm": 67.0,
|
| 20298 |
+
"learning_rate": 3.6271186440677964e-08,
|
| 20299 |
+
"loss": 1.1801,
|
| 20300 |
+
"step": 28930
|
| 20301 |
+
},
|
| 20302 |
+
{
|
| 20303 |
+
"epoch": 0.72,
|
| 20304 |
+
"grad_norm": 65.5,
|
| 20305 |
+
"learning_rate": 3.593220338983051e-08,
|
| 20306 |
+
"loss": 1.18,
|
| 20307 |
+
"step": 28940
|
| 20308 |
+
},
|
| 20309 |
+
{
|
| 20310 |
+
"epoch": 0.72,
|
| 20311 |
+
"grad_norm": 67.5,
|
| 20312 |
+
"learning_rate": 3.559322033898305e-08,
|
| 20313 |
+
"loss": 1.1894,
|
| 20314 |
+
"step": 28950
|
| 20315 |
+
},
|
| 20316 |
+
{
|
| 20317 |
+
"epoch": 0.72,
|
| 20318 |
+
"grad_norm": 67.0,
|
| 20319 |
+
"learning_rate": 3.5254237288135593e-08,
|
| 20320 |
+
"loss": 1.1711,
|
| 20321 |
+
"step": 28960
|
| 20322 |
+
},
|
| 20323 |
+
{
|
| 20324 |
+
"epoch": 0.72,
|
| 20325 |
+
"grad_norm": 66.0,
|
| 20326 |
+
"learning_rate": 3.4915254237288134e-08,
|
| 20327 |
+
"loss": 1.1621,
|
| 20328 |
+
"step": 28970
|
| 20329 |
+
},
|
| 20330 |
+
{
|
| 20331 |
+
"epoch": 0.72,
|
| 20332 |
+
"grad_norm": 65.5,
|
| 20333 |
+
"learning_rate": 3.457627118644068e-08,
|
| 20334 |
+
"loss": 1.1823,
|
| 20335 |
+
"step": 28980
|
| 20336 |
+
},
|
| 20337 |
+
{
|
| 20338 |
+
"epoch": 0.72,
|
| 20339 |
+
"grad_norm": 64.5,
|
| 20340 |
+
"learning_rate": 3.423728813559322e-08,
|
| 20341 |
+
"loss": 1.2095,
|
| 20342 |
+
"step": 28990
|
| 20343 |
+
},
|
| 20344 |
+
{
|
| 20345 |
+
"epoch": 0.72,
|
| 20346 |
+
"grad_norm": 67.0,
|
| 20347 |
+
"learning_rate": 3.3898305084745764e-08,
|
| 20348 |
+
"loss": 1.1891,
|
| 20349 |
+
"step": 29000
|
| 20350 |
+
},
|
| 20351 |
+
{
|
| 20352 |
+
"epoch": 0.73,
|
| 20353 |
+
"grad_norm": 67.0,
|
| 20354 |
+
"learning_rate": 3.3559322033898305e-08,
|
| 20355 |
+
"loss": 1.1973,
|
| 20356 |
+
"step": 29010
|
| 20357 |
+
},
|
| 20358 |
+
{
|
| 20359 |
+
"epoch": 0.73,
|
| 20360 |
+
"grad_norm": 67.0,
|
| 20361 |
+
"learning_rate": 3.3220338983050845e-08,
|
| 20362 |
+
"loss": 1.1556,
|
| 20363 |
+
"step": 29020
|
| 20364 |
+
},
|
| 20365 |
+
{
|
| 20366 |
+
"epoch": 0.73,
|
| 20367 |
+
"grad_norm": 66.0,
|
| 20368 |
+
"learning_rate": 3.288135593220339e-08,
|
| 20369 |
+
"loss": 1.1879,
|
| 20370 |
+
"step": 29030
|
| 20371 |
+
},
|
| 20372 |
+
{
|
| 20373 |
+
"epoch": 0.73,
|
| 20374 |
+
"grad_norm": 67.5,
|
| 20375 |
+
"learning_rate": 3.2542372881355934e-08,
|
| 20376 |
+
"loss": 1.1944,
|
| 20377 |
+
"step": 29040
|
| 20378 |
+
},
|
| 20379 |
+
{
|
| 20380 |
+
"epoch": 0.73,
|
| 20381 |
+
"grad_norm": 68.0,
|
| 20382 |
+
"learning_rate": 3.2203389830508475e-08,
|
| 20383 |
+
"loss": 1.1761,
|
| 20384 |
+
"step": 29050
|
| 20385 |
+
},
|
| 20386 |
+
{
|
| 20387 |
+
"epoch": 0.73,
|
| 20388 |
+
"grad_norm": 66.0,
|
| 20389 |
+
"learning_rate": 3.1864406779661016e-08,
|
| 20390 |
+
"loss": 1.1798,
|
| 20391 |
+
"step": 29060
|
| 20392 |
+
},
|
| 20393 |
+
{
|
| 20394 |
+
"epoch": 0.73,
|
| 20395 |
+
"grad_norm": 66.0,
|
| 20396 |
+
"learning_rate": 3.152542372881356e-08,
|
| 20397 |
+
"loss": 1.1858,
|
| 20398 |
+
"step": 29070
|
| 20399 |
+
},
|
| 20400 |
+
{
|
| 20401 |
+
"epoch": 0.73,
|
| 20402 |
+
"grad_norm": 66.5,
|
| 20403 |
+
"learning_rate": 3.1186440677966104e-08,
|
| 20404 |
+
"loss": 1.1662,
|
| 20405 |
+
"step": 29080
|
| 20406 |
+
},
|
| 20407 |
+
{
|
| 20408 |
+
"epoch": 0.73,
|
| 20409 |
+
"grad_norm": 65.0,
|
| 20410 |
+
"learning_rate": 3.0847457627118645e-08,
|
| 20411 |
+
"loss": 1.1988,
|
| 20412 |
+
"step": 29090
|
| 20413 |
+
},
|
| 20414 |
+
{
|
| 20415 |
+
"epoch": 0.73,
|
| 20416 |
+
"grad_norm": 66.0,
|
| 20417 |
+
"learning_rate": 3.0508474576271186e-08,
|
| 20418 |
+
"loss": 1.166,
|
| 20419 |
+
"step": 29100
|
| 20420 |
+
},
|
| 20421 |
+
{
|
| 20422 |
+
"epoch": 0.73,
|
| 20423 |
+
"grad_norm": 66.5,
|
| 20424 |
+
"learning_rate": 3.016949152542373e-08,
|
| 20425 |
+
"loss": 1.1616,
|
| 20426 |
+
"step": 29110
|
| 20427 |
+
},
|
| 20428 |
+
{
|
| 20429 |
+
"epoch": 0.73,
|
| 20430 |
+
"grad_norm": 65.5,
|
| 20431 |
+
"learning_rate": 2.9830508474576274e-08,
|
| 20432 |
+
"loss": 1.1717,
|
| 20433 |
+
"step": 29120
|
| 20434 |
+
},
|
| 20435 |
+
{
|
| 20436 |
+
"epoch": 0.73,
|
| 20437 |
+
"grad_norm": 66.5,
|
| 20438 |
+
"learning_rate": 2.9491525423728812e-08,
|
| 20439 |
+
"loss": 1.1695,
|
| 20440 |
+
"step": 29130
|
| 20441 |
+
},
|
| 20442 |
+
{
|
| 20443 |
+
"epoch": 0.73,
|
| 20444 |
+
"grad_norm": 66.0,
|
| 20445 |
+
"learning_rate": 2.9152542372881353e-08,
|
| 20446 |
+
"loss": 1.146,
|
| 20447 |
+
"step": 29140
|
| 20448 |
+
},
|
| 20449 |
+
{
|
| 20450 |
+
"epoch": 0.73,
|
| 20451 |
+
"grad_norm": 67.5,
|
| 20452 |
+
"learning_rate": 2.8813559322033897e-08,
|
| 20453 |
+
"loss": 1.1694,
|
| 20454 |
+
"step": 29150
|
| 20455 |
+
},
|
| 20456 |
+
{
|
| 20457 |
+
"epoch": 0.73,
|
| 20458 |
+
"grad_norm": 64.5,
|
| 20459 |
+
"learning_rate": 2.8474576271186438e-08,
|
| 20460 |
+
"loss": 1.179,
|
| 20461 |
+
"step": 29160
|
| 20462 |
+
},
|
| 20463 |
+
{
|
| 20464 |
+
"epoch": 0.73,
|
| 20465 |
+
"grad_norm": 67.0,
|
| 20466 |
+
"learning_rate": 2.8135593220338982e-08,
|
| 20467 |
+
"loss": 1.1701,
|
| 20468 |
+
"step": 29170
|
| 20469 |
+
},
|
| 20470 |
+
{
|
| 20471 |
+
"epoch": 0.73,
|
| 20472 |
+
"grad_norm": 66.0,
|
| 20473 |
+
"learning_rate": 2.7796610169491523e-08,
|
| 20474 |
+
"loss": 1.1695,
|
| 20475 |
+
"step": 29180
|
| 20476 |
+
},
|
| 20477 |
+
{
|
| 20478 |
+
"epoch": 0.73,
|
| 20479 |
+
"grad_norm": 67.0,
|
| 20480 |
+
"learning_rate": 2.7457627118644067e-08,
|
| 20481 |
+
"loss": 1.1776,
|
| 20482 |
+
"step": 29190
|
| 20483 |
+
},
|
| 20484 |
+
{
|
| 20485 |
+
"epoch": 0.73,
|
| 20486 |
+
"grad_norm": 66.0,
|
| 20487 |
+
"learning_rate": 2.7118644067796608e-08,
|
| 20488 |
+
"loss": 1.1782,
|
| 20489 |
+
"step": 29200
|
| 20490 |
+
},
|
| 20491 |
+
{
|
| 20492 |
+
"epoch": 0.73,
|
| 20493 |
+
"grad_norm": 65.0,
|
| 20494 |
+
"learning_rate": 2.6779661016949152e-08,
|
| 20495 |
+
"loss": 1.1798,
|
| 20496 |
+
"step": 29210
|
| 20497 |
+
},
|
| 20498 |
+
{
|
| 20499 |
+
"epoch": 0.73,
|
| 20500 |
+
"grad_norm": 64.0,
|
| 20501 |
+
"learning_rate": 2.6440677966101693e-08,
|
| 20502 |
+
"loss": 1.1631,
|
| 20503 |
+
"step": 29220
|
| 20504 |
+
},
|
| 20505 |
+
{
|
| 20506 |
+
"epoch": 0.73,
|
| 20507 |
+
"grad_norm": 64.0,
|
| 20508 |
+
"learning_rate": 2.6101694915254238e-08,
|
| 20509 |
+
"loss": 1.1623,
|
| 20510 |
+
"step": 29230
|
| 20511 |
+
},
|
| 20512 |
+
{
|
| 20513 |
+
"epoch": 0.73,
|
| 20514 |
+
"grad_norm": 68.5,
|
| 20515 |
+
"learning_rate": 2.576271186440678e-08,
|
| 20516 |
+
"loss": 1.2007,
|
| 20517 |
+
"step": 29240
|
| 20518 |
+
},
|
| 20519 |
+
{
|
| 20520 |
+
"epoch": 0.73,
|
| 20521 |
+
"grad_norm": 64.5,
|
| 20522 |
+
"learning_rate": 2.5423728813559323e-08,
|
| 20523 |
+
"loss": 1.1659,
|
| 20524 |
+
"step": 29250
|
| 20525 |
+
},
|
| 20526 |
+
{
|
| 20527 |
+
"epoch": 0.73,
|
| 20528 |
+
"grad_norm": 67.5,
|
| 20529 |
+
"learning_rate": 2.5084745762711864e-08,
|
| 20530 |
+
"loss": 1.1919,
|
| 20531 |
+
"step": 29260
|
| 20532 |
+
},
|
| 20533 |
+
{
|
| 20534 |
+
"epoch": 0.73,
|
| 20535 |
+
"grad_norm": 65.5,
|
| 20536 |
+
"learning_rate": 2.4745762711864405e-08,
|
| 20537 |
+
"loss": 1.1692,
|
| 20538 |
+
"step": 29270
|
| 20539 |
+
},
|
| 20540 |
+
{
|
| 20541 |
+
"epoch": 0.73,
|
| 20542 |
+
"grad_norm": 65.5,
|
| 20543 |
+
"learning_rate": 2.440677966101695e-08,
|
| 20544 |
+
"loss": 1.2005,
|
| 20545 |
+
"step": 29280
|
| 20546 |
+
},
|
| 20547 |
+
{
|
| 20548 |
+
"epoch": 0.73,
|
| 20549 |
+
"grad_norm": 67.5,
|
| 20550 |
+
"learning_rate": 2.406779661016949e-08,
|
| 20551 |
+
"loss": 1.1626,
|
| 20552 |
+
"step": 29290
|
| 20553 |
+
},
|
| 20554 |
+
{
|
| 20555 |
+
"epoch": 0.73,
|
| 20556 |
+
"grad_norm": 64.5,
|
| 20557 |
+
"learning_rate": 2.3728813559322034e-08,
|
| 20558 |
+
"loss": 1.1873,
|
| 20559 |
+
"step": 29300
|
| 20560 |
+
},
|
| 20561 |
+
{
|
| 20562 |
+
"epoch": 0.73,
|
| 20563 |
+
"grad_norm": 70.0,
|
| 20564 |
+
"learning_rate": 2.3389830508474575e-08,
|
| 20565 |
+
"loss": 1.1779,
|
| 20566 |
+
"step": 29310
|
| 20567 |
+
},
|
| 20568 |
+
{
|
| 20569 |
+
"epoch": 0.73,
|
| 20570 |
+
"grad_norm": 67.0,
|
| 20571 |
+
"learning_rate": 2.305084745762712e-08,
|
| 20572 |
+
"loss": 1.1553,
|
| 20573 |
+
"step": 29320
|
| 20574 |
+
},
|
| 20575 |
+
{
|
| 20576 |
+
"epoch": 0.73,
|
| 20577 |
+
"grad_norm": 64.5,
|
| 20578 |
+
"learning_rate": 2.271186440677966e-08,
|
| 20579 |
+
"loss": 1.1923,
|
| 20580 |
+
"step": 29330
|
| 20581 |
+
},
|
| 20582 |
+
{
|
| 20583 |
+
"epoch": 0.73,
|
| 20584 |
+
"grad_norm": 65.5,
|
| 20585 |
+
"learning_rate": 2.2372881355932204e-08,
|
| 20586 |
+
"loss": 1.1708,
|
| 20587 |
+
"step": 29340
|
| 20588 |
+
},
|
| 20589 |
+
{
|
| 20590 |
+
"epoch": 0.73,
|
| 20591 |
+
"grad_norm": 69.0,
|
| 20592 |
+
"learning_rate": 2.2033898305084745e-08,
|
| 20593 |
+
"loss": 1.1977,
|
| 20594 |
+
"step": 29350
|
| 20595 |
+
},
|
| 20596 |
+
{
|
| 20597 |
+
"epoch": 0.73,
|
| 20598 |
+
"grad_norm": 64.5,
|
| 20599 |
+
"learning_rate": 2.169491525423729e-08,
|
| 20600 |
+
"loss": 1.1713,
|
| 20601 |
+
"step": 29360
|
| 20602 |
+
},
|
| 20603 |
+
{
|
| 20604 |
+
"epoch": 0.73,
|
| 20605 |
+
"grad_norm": 66.0,
|
| 20606 |
+
"learning_rate": 2.135593220338983e-08,
|
| 20607 |
+
"loss": 1.1634,
|
| 20608 |
+
"step": 29370
|
| 20609 |
+
},
|
| 20610 |
+
{
|
| 20611 |
+
"epoch": 0.73,
|
| 20612 |
+
"grad_norm": 70.0,
|
| 20613 |
+
"learning_rate": 2.1016949152542374e-08,
|
| 20614 |
+
"loss": 1.1726,
|
| 20615 |
+
"step": 29380
|
| 20616 |
+
},
|
| 20617 |
+
{
|
| 20618 |
+
"epoch": 0.73,
|
| 20619 |
+
"grad_norm": 65.5,
|
| 20620 |
+
"learning_rate": 2.0677966101694915e-08,
|
| 20621 |
+
"loss": 1.189,
|
| 20622 |
+
"step": 29390
|
| 20623 |
+
},
|
| 20624 |
+
{
|
| 20625 |
+
"epoch": 0.73,
|
| 20626 |
+
"grad_norm": 65.5,
|
| 20627 |
+
"learning_rate": 2.0338983050847456e-08,
|
| 20628 |
+
"loss": 1.1715,
|
| 20629 |
+
"step": 29400
|
| 20630 |
+
},
|
| 20631 |
+
{
|
| 20632 |
+
"epoch": 0.74,
|
| 20633 |
+
"grad_norm": 65.0,
|
| 20634 |
+
"learning_rate": 2e-08,
|
| 20635 |
+
"loss": 1.1636,
|
| 20636 |
+
"step": 29410
|
| 20637 |
+
},
|
| 20638 |
+
{
|
| 20639 |
+
"epoch": 0.74,
|
| 20640 |
+
"grad_norm": 65.5,
|
| 20641 |
+
"learning_rate": 1.966101694915254e-08,
|
| 20642 |
+
"loss": 1.1809,
|
| 20643 |
+
"step": 29420
|
| 20644 |
+
},
|
| 20645 |
+
{
|
| 20646 |
+
"epoch": 0.74,
|
| 20647 |
+
"grad_norm": 66.0,
|
| 20648 |
+
"learning_rate": 1.9322033898305086e-08,
|
| 20649 |
+
"loss": 1.1693,
|
| 20650 |
+
"step": 29430
|
| 20651 |
+
},
|
| 20652 |
+
{
|
| 20653 |
+
"epoch": 0.74,
|
| 20654 |
+
"grad_norm": 71.5,
|
| 20655 |
+
"learning_rate": 1.8983050847457626e-08,
|
| 20656 |
+
"loss": 1.1905,
|
| 20657 |
+
"step": 29440
|
| 20658 |
+
},
|
| 20659 |
+
{
|
| 20660 |
+
"epoch": 0.74,
|
| 20661 |
+
"grad_norm": 64.5,
|
| 20662 |
+
"learning_rate": 1.864406779661017e-08,
|
| 20663 |
+
"loss": 1.1722,
|
| 20664 |
+
"step": 29450
|
| 20665 |
+
},
|
| 20666 |
+
{
|
| 20667 |
+
"epoch": 0.74,
|
| 20668 |
+
"grad_norm": 67.5,
|
| 20669 |
+
"learning_rate": 1.830508474576271e-08,
|
| 20670 |
+
"loss": 1.1826,
|
| 20671 |
+
"step": 29460
|
| 20672 |
+
},
|
| 20673 |
+
{
|
| 20674 |
+
"epoch": 0.74,
|
| 20675 |
+
"grad_norm": 67.0,
|
| 20676 |
+
"learning_rate": 1.7966101694915256e-08,
|
| 20677 |
+
"loss": 1.1745,
|
| 20678 |
+
"step": 29470
|
| 20679 |
+
},
|
| 20680 |
+
{
|
| 20681 |
+
"epoch": 0.74,
|
| 20682 |
+
"grad_norm": 64.5,
|
| 20683 |
+
"learning_rate": 1.7627118644067797e-08,
|
| 20684 |
+
"loss": 1.1653,
|
| 20685 |
+
"step": 29480
|
| 20686 |
+
},
|
| 20687 |
+
{
|
| 20688 |
+
"epoch": 0.74,
|
| 20689 |
+
"grad_norm": 67.5,
|
| 20690 |
+
"learning_rate": 1.728813559322034e-08,
|
| 20691 |
+
"loss": 1.1924,
|
| 20692 |
+
"step": 29490
|
| 20693 |
+
},
|
| 20694 |
+
{
|
| 20695 |
+
"epoch": 0.74,
|
| 20696 |
+
"grad_norm": 66.5,
|
| 20697 |
+
"learning_rate": 1.6949152542372882e-08,
|
| 20698 |
+
"loss": 1.1706,
|
| 20699 |
+
"step": 29500
|
| 20700 |
+
},
|
| 20701 |
+
{
|
| 20702 |
+
"epoch": 0.74,
|
| 20703 |
+
"grad_norm": 69.0,
|
| 20704 |
+
"learning_rate": 1.6610169491525423e-08,
|
| 20705 |
+
"loss": 1.1748,
|
| 20706 |
+
"step": 29510
|
| 20707 |
+
},
|
| 20708 |
+
{
|
| 20709 |
+
"epoch": 0.74,
|
| 20710 |
+
"grad_norm": 66.0,
|
| 20711 |
+
"learning_rate": 1.6271186440677967e-08,
|
| 20712 |
+
"loss": 1.1892,
|
| 20713 |
+
"step": 29520
|
| 20714 |
+
},
|
| 20715 |
+
{
|
| 20716 |
+
"epoch": 0.74,
|
| 20717 |
+
"grad_norm": 67.0,
|
| 20718 |
+
"learning_rate": 1.5932203389830508e-08,
|
| 20719 |
+
"loss": 1.1983,
|
| 20720 |
+
"step": 29530
|
| 20721 |
+
},
|
| 20722 |
+
{
|
| 20723 |
+
"epoch": 0.74,
|
| 20724 |
+
"grad_norm": 65.5,
|
| 20725 |
+
"learning_rate": 1.5593220338983052e-08,
|
| 20726 |
+
"loss": 1.1682,
|
| 20727 |
+
"step": 29540
|
| 20728 |
+
},
|
| 20729 |
+
{
|
| 20730 |
+
"epoch": 0.74,
|
| 20731 |
+
"grad_norm": 66.0,
|
| 20732 |
+
"learning_rate": 1.5254237288135593e-08,
|
| 20733 |
+
"loss": 1.1612,
|
| 20734 |
+
"step": 29550
|
| 20735 |
+
},
|
| 20736 |
+
{
|
| 20737 |
+
"epoch": 0.74,
|
| 20738 |
+
"grad_norm": 64.5,
|
| 20739 |
+
"learning_rate": 1.4915254237288137e-08,
|
| 20740 |
+
"loss": 1.1761,
|
| 20741 |
+
"step": 29560
|
| 20742 |
+
},
|
| 20743 |
+
{
|
| 20744 |
+
"epoch": 0.74,
|
| 20745 |
+
"grad_norm": 68.5,
|
| 20746 |
+
"learning_rate": 1.4576271186440676e-08,
|
| 20747 |
+
"loss": 1.1729,
|
| 20748 |
+
"step": 29570
|
| 20749 |
+
},
|
| 20750 |
+
{
|
| 20751 |
+
"epoch": 0.74,
|
| 20752 |
+
"grad_norm": 71.5,
|
| 20753 |
+
"learning_rate": 1.4237288135593219e-08,
|
| 20754 |
+
"loss": 1.1787,
|
| 20755 |
+
"step": 29580
|
| 20756 |
+
},
|
| 20757 |
+
{
|
| 20758 |
+
"epoch": 0.74,
|
| 20759 |
+
"grad_norm": 66.0,
|
| 20760 |
+
"learning_rate": 1.3898305084745762e-08,
|
| 20761 |
+
"loss": 1.1603,
|
| 20762 |
+
"step": 29590
|
| 20763 |
+
},
|
| 20764 |
+
{
|
| 20765 |
+
"epoch": 0.74,
|
| 20766 |
+
"grad_norm": 68.0,
|
| 20767 |
+
"learning_rate": 1.3559322033898304e-08,
|
| 20768 |
+
"loss": 1.1768,
|
| 20769 |
+
"step": 29600
|
| 20770 |
+
},
|
| 20771 |
+
{
|
| 20772 |
+
"epoch": 0.74,
|
| 20773 |
+
"grad_norm": 64.0,
|
| 20774 |
+
"learning_rate": 1.3220338983050847e-08,
|
| 20775 |
+
"loss": 1.1934,
|
| 20776 |
+
"step": 29610
|
| 20777 |
+
},
|
| 20778 |
+
{
|
| 20779 |
+
"epoch": 0.74,
|
| 20780 |
+
"grad_norm": 66.0,
|
| 20781 |
+
"learning_rate": 1.288135593220339e-08,
|
| 20782 |
+
"loss": 1.1794,
|
| 20783 |
+
"step": 29620
|
| 20784 |
+
},
|
| 20785 |
+
{
|
| 20786 |
+
"epoch": 0.74,
|
| 20787 |
+
"grad_norm": 63.75,
|
| 20788 |
+
"learning_rate": 1.2542372881355932e-08,
|
| 20789 |
+
"loss": 1.1709,
|
| 20790 |
+
"step": 29630
|
| 20791 |
+
},
|
| 20792 |
+
{
|
| 20793 |
+
"epoch": 0.74,
|
| 20794 |
+
"grad_norm": 67.0,
|
| 20795 |
+
"learning_rate": 1.2203389830508474e-08,
|
| 20796 |
+
"loss": 1.1842,
|
| 20797 |
+
"step": 29640
|
| 20798 |
+
},
|
| 20799 |
+
{
|
| 20800 |
+
"epoch": 0.74,
|
| 20801 |
+
"grad_norm": 70.0,
|
| 20802 |
+
"learning_rate": 1.1864406779661017e-08,
|
| 20803 |
+
"loss": 1.2028,
|
| 20804 |
+
"step": 29650
|
| 20805 |
+
},
|
| 20806 |
+
{
|
| 20807 |
+
"epoch": 0.74,
|
| 20808 |
+
"grad_norm": 64.5,
|
| 20809 |
+
"learning_rate": 1.152542372881356e-08,
|
| 20810 |
+
"loss": 1.1769,
|
| 20811 |
+
"step": 29660
|
| 20812 |
+
},
|
| 20813 |
+
{
|
| 20814 |
+
"epoch": 0.74,
|
| 20815 |
+
"grad_norm": 69.0,
|
| 20816 |
+
"learning_rate": 1.1186440677966102e-08,
|
| 20817 |
+
"loss": 1.1695,
|
| 20818 |
+
"step": 29670
|
| 20819 |
+
},
|
| 20820 |
+
{
|
| 20821 |
+
"epoch": 0.74,
|
| 20822 |
+
"grad_norm": 64.5,
|
| 20823 |
+
"learning_rate": 1.0847457627118645e-08,
|
| 20824 |
+
"loss": 1.1755,
|
| 20825 |
+
"step": 29680
|
| 20826 |
+
},
|
| 20827 |
+
{
|
| 20828 |
+
"epoch": 0.74,
|
| 20829 |
+
"grad_norm": 66.5,
|
| 20830 |
+
"learning_rate": 1.0508474576271187e-08,
|
| 20831 |
+
"loss": 1.1646,
|
| 20832 |
+
"step": 29690
|
| 20833 |
+
},
|
| 20834 |
+
{
|
| 20835 |
+
"epoch": 0.74,
|
| 20836 |
+
"grad_norm": 66.0,
|
| 20837 |
+
"learning_rate": 1.0169491525423728e-08,
|
| 20838 |
+
"loss": 1.1852,
|
| 20839 |
+
"step": 29700
|
| 20840 |
+
},
|
| 20841 |
+
{
|
| 20842 |
+
"epoch": 0.74,
|
| 20843 |
+
"grad_norm": 65.5,
|
| 20844 |
+
"learning_rate": 9.83050847457627e-09,
|
| 20845 |
+
"loss": 1.1742,
|
| 20846 |
+
"step": 29710
|
| 20847 |
+
},
|
| 20848 |
+
{
|
| 20849 |
+
"epoch": 0.74,
|
| 20850 |
+
"grad_norm": 67.0,
|
| 20851 |
+
"learning_rate": 9.491525423728813e-09,
|
| 20852 |
+
"loss": 1.1842,
|
| 20853 |
+
"step": 29720
|
| 20854 |
+
},
|
| 20855 |
+
{
|
| 20856 |
+
"epoch": 0.74,
|
| 20857 |
+
"grad_norm": 67.0,
|
| 20858 |
+
"learning_rate": 9.152542372881356e-09,
|
| 20859 |
+
"loss": 1.1906,
|
| 20860 |
+
"step": 29730
|
| 20861 |
+
},
|
| 20862 |
+
{
|
| 20863 |
+
"epoch": 0.74,
|
| 20864 |
+
"grad_norm": 65.5,
|
| 20865 |
+
"learning_rate": 8.813559322033898e-09,
|
| 20866 |
+
"loss": 1.1782,
|
| 20867 |
+
"step": 29740
|
| 20868 |
+
},
|
| 20869 |
+
{
|
| 20870 |
+
"epoch": 0.74,
|
| 20871 |
+
"grad_norm": 67.0,
|
| 20872 |
+
"learning_rate": 8.474576271186441e-09,
|
| 20873 |
+
"loss": 1.1841,
|
| 20874 |
+
"step": 29750
|
| 20875 |
+
},
|
| 20876 |
+
{
|
| 20877 |
+
"epoch": 0.74,
|
| 20878 |
+
"grad_norm": 64.0,
|
| 20879 |
+
"learning_rate": 8.135593220338983e-09,
|
| 20880 |
+
"loss": 1.1766,
|
| 20881 |
+
"step": 29760
|
| 20882 |
+
},
|
| 20883 |
+
{
|
| 20884 |
+
"epoch": 0.74,
|
| 20885 |
+
"grad_norm": 63.75,
|
| 20886 |
+
"learning_rate": 7.796610169491526e-09,
|
| 20887 |
+
"loss": 1.1705,
|
| 20888 |
+
"step": 29770
|
| 20889 |
+
},
|
| 20890 |
+
{
|
| 20891 |
+
"epoch": 0.74,
|
| 20892 |
+
"grad_norm": 66.5,
|
| 20893 |
+
"learning_rate": 7.457627118644069e-09,
|
| 20894 |
+
"loss": 1.1767,
|
| 20895 |
+
"step": 29780
|
| 20896 |
+
},
|
| 20897 |
+
{
|
| 20898 |
+
"epoch": 0.74,
|
| 20899 |
+
"grad_norm": 66.5,
|
| 20900 |
+
"learning_rate": 7.1186440677966095e-09,
|
| 20901 |
+
"loss": 1.1694,
|
| 20902 |
+
"step": 29790
|
| 20903 |
+
},
|
| 20904 |
+
{
|
| 20905 |
+
"epoch": 0.74,
|
| 20906 |
+
"grad_norm": 64.5,
|
| 20907 |
+
"learning_rate": 6.779661016949152e-09,
|
| 20908 |
+
"loss": 1.1741,
|
| 20909 |
+
"step": 29800
|
| 20910 |
+
},
|
| 20911 |
+
{
|
| 20912 |
+
"epoch": 0.75,
|
| 20913 |
+
"grad_norm": 72.0,
|
| 20914 |
+
"learning_rate": 6.440677966101695e-09,
|
| 20915 |
+
"loss": 1.1896,
|
| 20916 |
+
"step": 29810
|
| 20917 |
+
},
|
| 20918 |
+
{
|
| 20919 |
+
"epoch": 0.75,
|
| 20920 |
+
"grad_norm": 67.0,
|
| 20921 |
+
"learning_rate": 6.101694915254237e-09,
|
| 20922 |
+
"loss": 1.1807,
|
| 20923 |
+
"step": 29820
|
| 20924 |
+
},
|
| 20925 |
+
{
|
| 20926 |
+
"epoch": 0.75,
|
| 20927 |
+
"grad_norm": 65.0,
|
| 20928 |
+
"learning_rate": 5.76271186440678e-09,
|
| 20929 |
+
"loss": 1.1864,
|
| 20930 |
+
"step": 29830
|
| 20931 |
+
},
|
| 20932 |
+
{
|
| 20933 |
+
"epoch": 0.75,
|
| 20934 |
+
"grad_norm": 67.0,
|
| 20935 |
+
"learning_rate": 5.423728813559322e-09,
|
| 20936 |
+
"loss": 1.1829,
|
| 20937 |
+
"step": 29840
|
| 20938 |
+
},
|
| 20939 |
+
{
|
| 20940 |
+
"epoch": 0.75,
|
| 20941 |
+
"grad_norm": 64.5,
|
| 20942 |
+
"learning_rate": 5.084745762711864e-09,
|
| 20943 |
+
"loss": 1.1958,
|
| 20944 |
+
"step": 29850
|
| 20945 |
+
},
|
| 20946 |
+
{
|
| 20947 |
+
"epoch": 0.75,
|
| 20948 |
+
"grad_norm": 66.5,
|
| 20949 |
+
"learning_rate": 4.745762711864407e-09,
|
| 20950 |
+
"loss": 1.1793,
|
| 20951 |
+
"step": 29860
|
| 20952 |
+
},
|
| 20953 |
+
{
|
| 20954 |
+
"epoch": 0.75,
|
| 20955 |
+
"grad_norm": 64.5,
|
| 20956 |
+
"learning_rate": 4.406779661016949e-09,
|
| 20957 |
+
"loss": 1.1734,
|
| 20958 |
+
"step": 29870
|
| 20959 |
+
},
|
| 20960 |
+
{
|
| 20961 |
+
"epoch": 0.75,
|
| 20962 |
+
"grad_norm": 65.0,
|
| 20963 |
+
"learning_rate": 4.067796610169492e-09,
|
| 20964 |
+
"loss": 1.184,
|
| 20965 |
+
"step": 29880
|
| 20966 |
+
},
|
| 20967 |
+
{
|
| 20968 |
+
"epoch": 0.75,
|
| 20969 |
+
"grad_norm": 66.0,
|
| 20970 |
+
"learning_rate": 3.728813559322034e-09,
|
| 20971 |
+
"loss": 1.1666,
|
| 20972 |
+
"step": 29890
|
| 20973 |
+
},
|
| 20974 |
+
{
|
| 20975 |
+
"epoch": 0.75,
|
| 20976 |
+
"grad_norm": 69.5,
|
| 20977 |
+
"learning_rate": 3.389830508474576e-09,
|
| 20978 |
+
"loss": 1.1738,
|
| 20979 |
+
"step": 29900
|
| 20980 |
+
},
|
| 20981 |
+
{
|
| 20982 |
+
"epoch": 0.75,
|
| 20983 |
+
"grad_norm": 71.5,
|
| 20984 |
+
"learning_rate": 3.0508474576271186e-09,
|
| 20985 |
+
"loss": 1.1815,
|
| 20986 |
+
"step": 29910
|
| 20987 |
+
},
|
| 20988 |
+
{
|
| 20989 |
+
"epoch": 0.75,
|
| 20990 |
+
"grad_norm": 68.5,
|
| 20991 |
+
"learning_rate": 2.711864406779661e-09,
|
| 20992 |
+
"loss": 1.1795,
|
| 20993 |
+
"step": 29920
|
| 20994 |
+
},
|
| 20995 |
+
{
|
| 20996 |
+
"epoch": 0.75,
|
| 20997 |
+
"grad_norm": 66.5,
|
| 20998 |
+
"learning_rate": 2.3728813559322033e-09,
|
| 20999 |
+
"loss": 1.1775,
|
| 21000 |
+
"step": 29930
|
| 21001 |
+
},
|
| 21002 |
+
{
|
| 21003 |
+
"epoch": 0.75,
|
| 21004 |
+
"grad_norm": 65.0,
|
| 21005 |
+
"learning_rate": 2.033898305084746e-09,
|
| 21006 |
+
"loss": 1.1742,
|
| 21007 |
+
"step": 29940
|
| 21008 |
+
},
|
| 21009 |
+
{
|
| 21010 |
+
"epoch": 0.75,
|
| 21011 |
+
"grad_norm": 68.0,
|
| 21012 |
+
"learning_rate": 1.694915254237288e-09,
|
| 21013 |
+
"loss": 1.1615,
|
| 21014 |
+
"step": 29950
|
| 21015 |
+
},
|
| 21016 |
+
{
|
| 21017 |
+
"epoch": 0.75,
|
| 21018 |
+
"grad_norm": 64.5,
|
| 21019 |
+
"learning_rate": 1.3559322033898306e-09,
|
| 21020 |
+
"loss": 1.1785,
|
| 21021 |
+
"step": 29960
|
| 21022 |
+
},
|
| 21023 |
+
{
|
| 21024 |
+
"epoch": 0.75,
|
| 21025 |
+
"grad_norm": 65.0,
|
| 21026 |
+
"learning_rate": 1.016949152542373e-09,
|
| 21027 |
+
"loss": 1.1834,
|
| 21028 |
+
"step": 29970
|
| 21029 |
+
},
|
| 21030 |
+
{
|
| 21031 |
+
"epoch": 0.75,
|
| 21032 |
+
"grad_norm": 65.0,
|
| 21033 |
+
"learning_rate": 6.779661016949153e-10,
|
| 21034 |
+
"loss": 1.1747,
|
| 21035 |
+
"step": 29980
|
| 21036 |
+
},
|
| 21037 |
+
{
|
| 21038 |
+
"epoch": 0.75,
|
| 21039 |
+
"grad_norm": 68.5,
|
| 21040 |
+
"learning_rate": 3.3898305084745764e-10,
|
| 21041 |
+
"loss": 1.1884,
|
| 21042 |
+
"step": 29990
|
| 21043 |
+
},
|
| 21044 |
+
{
|
| 21045 |
+
"epoch": 0.75,
|
| 21046 |
+
"grad_norm": 66.5,
|
| 21047 |
+
"learning_rate": 0.0,
|
| 21048 |
+
"loss": 1.2117,
|
| 21049 |
+
"step": 30000
|
| 21050 |
+
},
|
| 21051 |
+
{
|
| 21052 |
+
"epoch": 0.75,
|
| 21053 |
+
"eval_loss": 1.169899582862854,
|
| 21054 |
+
"eval_runtime": 65.2594,
|
| 21055 |
+
"eval_samples_per_second": 15.323,
|
| 21056 |
+
"eval_steps_per_second": 15.323,
|
| 21057 |
+
"step": 30000
|
| 21058 |
}
|
| 21059 |
],
|
| 21060 |
"logging_steps": 10,
|
|
|
|
| 21062 |
"num_input_tokens_seen": 0,
|
| 21063 |
"num_train_epochs": 1,
|
| 21064 |
"save_steps": 5000,
|
| 21065 |
+
"total_flos": 4.841766125568e+17,
|
| 21066 |
"train_batch_size": 1,
|
| 21067 |
"trial_name": null,
|
| 21068 |
"trial_params": null
|