Training in progress, step 2600, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 201361312
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2a4bcf338cd5bcfdfca322442e0437d0756fba523cd3c6e969d1ab63c60649d6
|
| 3 |
size 201361312
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 102537812
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9d2c3f69d1b1d65ff82b5361a177d936c38774c67f7e7c9e80fbae95019ee61a
|
| 3 |
size 102537812
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3f17f98bb0e89af780ffffb5792ad56e5907144164a75d1f2d972da513c20656
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2bee69a6ac763ec64bcb9204c4b05b515ace2ec25288895042da824c3899c85a
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 1.602339506149292,
|
| 3 |
"best_model_checkpoint": "miner_id_24/checkpoint-2200",
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 200,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -16911,6 +16911,1414 @@
|
|
| 16911 |
"eval_samples_per_second": 8.736,
|
| 16912 |
"eval_steps_per_second": 4.373,
|
| 16913 |
"step": 2400
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 16914 |
}
|
| 16915 |
],
|
| 16916 |
"logging_steps": 1,
|
|
@@ -16925,7 +18333,7 @@
|
|
| 16925 |
"early_stopping_threshold": 0.0
|
| 16926 |
},
|
| 16927 |
"attributes": {
|
| 16928 |
-
"early_stopping_patience_counter":
|
| 16929 |
}
|
| 16930 |
},
|
| 16931 |
"TrainerControl": {
|
|
@@ -16939,7 +18347,7 @@
|
|
| 16939 |
"attributes": {}
|
| 16940 |
}
|
| 16941 |
},
|
| 16942 |
-
"total_flos": 2.
|
| 16943 |
"train_batch_size": 2,
|
| 16944 |
"trial_name": null,
|
| 16945 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 1.602339506149292,
|
| 3 |
"best_model_checkpoint": "miner_id_24/checkpoint-2200",
|
| 4 |
+
"epoch": 0.12408872343725764,
|
| 5 |
"eval_steps": 200,
|
| 6 |
+
"global_step": 2600,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 16911 |
"eval_samples_per_second": 8.736,
|
| 16912 |
"eval_steps_per_second": 4.373,
|
| 16913 |
"step": 2400
|
| 16914 |
+
},
|
| 16915 |
+
{
|
| 16916 |
+
"epoch": 0.11459116345109831,
|
| 16917 |
+
"grad_norm": 6.055756092071533,
|
| 16918 |
+
"learning_rate": 0.00019929798684826006,
|
| 16919 |
+
"loss": 7.0865,
|
| 16920 |
+
"step": 2401
|
| 16921 |
+
},
|
| 16922 |
+
{
|
| 16923 |
+
"epoch": 0.11463888988318956,
|
| 16924 |
+
"grad_norm": 7.1241044998168945,
|
| 16925 |
+
"learning_rate": 0.00019929739525163437,
|
| 16926 |
+
"loss": 6.9193,
|
| 16927 |
+
"step": 2402
|
| 16928 |
+
},
|
| 16929 |
+
{
|
| 16930 |
+
"epoch": 0.1146866163152808,
|
| 16931 |
+
"grad_norm": 5.732908725738525,
|
| 16932 |
+
"learning_rate": 0.00019929680340671896,
|
| 16933 |
+
"loss": 6.8239,
|
| 16934 |
+
"step": 2403
|
| 16935 |
+
},
|
| 16936 |
+
{
|
| 16937 |
+
"epoch": 0.11473434274737207,
|
| 16938 |
+
"grad_norm": 6.3990278244018555,
|
| 16939 |
+
"learning_rate": 0.00019929621131351525,
|
| 16940 |
+
"loss": 7.5857,
|
| 16941 |
+
"step": 2404
|
| 16942 |
+
},
|
| 16943 |
+
{
|
| 16944 |
+
"epoch": 0.11478206917946332,
|
| 16945 |
+
"grad_norm": 6.51461124420166,
|
| 16946 |
+
"learning_rate": 0.00019929561897202478,
|
| 16947 |
+
"loss": 7.5126,
|
| 16948 |
+
"step": 2405
|
| 16949 |
+
},
|
| 16950 |
+
{
|
| 16951 |
+
"epoch": 0.11482979561155457,
|
| 16952 |
+
"grad_norm": 4.333196640014648,
|
| 16953 |
+
"learning_rate": 0.000199295026382249,
|
| 16954 |
+
"loss": 5.407,
|
| 16955 |
+
"step": 2406
|
| 16956 |
+
},
|
| 16957 |
+
{
|
| 16958 |
+
"epoch": 0.11487752204364582,
|
| 16959 |
+
"grad_norm": 8.812541007995605,
|
| 16960 |
+
"learning_rate": 0.0001992944335441894,
|
| 16961 |
+
"loss": 7.8813,
|
| 16962 |
+
"step": 2407
|
| 16963 |
+
},
|
| 16964 |
+
{
|
| 16965 |
+
"epoch": 0.11492524847573707,
|
| 16966 |
+
"grad_norm": 6.706300735473633,
|
| 16967 |
+
"learning_rate": 0.00019929384045784746,
|
| 16968 |
+
"loss": 7.3434,
|
| 16969 |
+
"step": 2408
|
| 16970 |
+
},
|
| 16971 |
+
{
|
| 16972 |
+
"epoch": 0.11497297490782833,
|
| 16973 |
+
"grad_norm": 5.9792375564575195,
|
| 16974 |
+
"learning_rate": 0.00019929324712322465,
|
| 16975 |
+
"loss": 6.588,
|
| 16976 |
+
"step": 2409
|
| 16977 |
+
},
|
| 16978 |
+
{
|
| 16979 |
+
"epoch": 0.11502070133991958,
|
| 16980 |
+
"grad_norm": 5.693054676055908,
|
| 16981 |
+
"learning_rate": 0.00019929265354032247,
|
| 16982 |
+
"loss": 6.9219,
|
| 16983 |
+
"step": 2410
|
| 16984 |
+
},
|
| 16985 |
+
{
|
| 16986 |
+
"epoch": 0.11506842777201083,
|
| 16987 |
+
"grad_norm": 6.320680141448975,
|
| 16988 |
+
"learning_rate": 0.00019929205970914242,
|
| 16989 |
+
"loss": 6.0497,
|
| 16990 |
+
"step": 2411
|
| 16991 |
+
},
|
| 16992 |
+
{
|
| 16993 |
+
"epoch": 0.11511615420410208,
|
| 16994 |
+
"grad_norm": 7.449224948883057,
|
| 16995 |
+
"learning_rate": 0.00019929146562968594,
|
| 16996 |
+
"loss": 6.0061,
|
| 16997 |
+
"step": 2412
|
| 16998 |
+
},
|
| 16999 |
+
{
|
| 17000 |
+
"epoch": 0.11516388063619334,
|
| 17001 |
+
"grad_norm": 5.945936679840088,
|
| 17002 |
+
"learning_rate": 0.00019929087130195457,
|
| 17003 |
+
"loss": 6.7282,
|
| 17004 |
+
"step": 2413
|
| 17005 |
+
},
|
| 17006 |
+
{
|
| 17007 |
+
"epoch": 0.1152116070682846,
|
| 17008 |
+
"grad_norm": 5.9501752853393555,
|
| 17009 |
+
"learning_rate": 0.00019929027672594977,
|
| 17010 |
+
"loss": 6.9589,
|
| 17011 |
+
"step": 2414
|
| 17012 |
+
},
|
| 17013 |
+
{
|
| 17014 |
+
"epoch": 0.11525933350037584,
|
| 17015 |
+
"grad_norm": 5.984325408935547,
|
| 17016 |
+
"learning_rate": 0.00019928968190167302,
|
| 17017 |
+
"loss": 6.0555,
|
| 17018 |
+
"step": 2415
|
| 17019 |
+
},
|
| 17020 |
+
{
|
| 17021 |
+
"epoch": 0.1153070599324671,
|
| 17022 |
+
"grad_norm": 5.344329833984375,
|
| 17023 |
+
"learning_rate": 0.00019928908682912582,
|
| 17024 |
+
"loss": 6.0593,
|
| 17025 |
+
"step": 2416
|
| 17026 |
+
},
|
| 17027 |
+
{
|
| 17028 |
+
"epoch": 0.11535478636455836,
|
| 17029 |
+
"grad_norm": 7.7345709800720215,
|
| 17030 |
+
"learning_rate": 0.0001992884915083096,
|
| 17031 |
+
"loss": 7.2564,
|
| 17032 |
+
"step": 2417
|
| 17033 |
+
},
|
| 17034 |
+
{
|
| 17035 |
+
"epoch": 0.1154025127966496,
|
| 17036 |
+
"grad_norm": 6.209690570831299,
|
| 17037 |
+
"learning_rate": 0.00019928789593922597,
|
| 17038 |
+
"loss": 6.8254,
|
| 17039 |
+
"step": 2418
|
| 17040 |
+
},
|
| 17041 |
+
{
|
| 17042 |
+
"epoch": 0.11545023922874086,
|
| 17043 |
+
"grad_norm": 7.814291477203369,
|
| 17044 |
+
"learning_rate": 0.00019928730012187628,
|
| 17045 |
+
"loss": 6.2167,
|
| 17046 |
+
"step": 2419
|
| 17047 |
+
},
|
| 17048 |
+
{
|
| 17049 |
+
"epoch": 0.1154979656608321,
|
| 17050 |
+
"grad_norm": 7.086947917938232,
|
| 17051 |
+
"learning_rate": 0.00019928670405626212,
|
| 17052 |
+
"loss": 6.982,
|
| 17053 |
+
"step": 2420
|
| 17054 |
+
},
|
| 17055 |
+
{
|
| 17056 |
+
"epoch": 0.11554569209292337,
|
| 17057 |
+
"grad_norm": 5.73114538192749,
|
| 17058 |
+
"learning_rate": 0.00019928610774238496,
|
| 17059 |
+
"loss": 6.7713,
|
| 17060 |
+
"step": 2421
|
| 17061 |
+
},
|
| 17062 |
+
{
|
| 17063 |
+
"epoch": 0.11559341852501462,
|
| 17064 |
+
"grad_norm": 6.70130729675293,
|
| 17065 |
+
"learning_rate": 0.00019928551118024627,
|
| 17066 |
+
"loss": 6.346,
|
| 17067 |
+
"step": 2422
|
| 17068 |
+
},
|
| 17069 |
+
{
|
| 17070 |
+
"epoch": 0.11564114495710587,
|
| 17071 |
+
"grad_norm": 6.0923662185668945,
|
| 17072 |
+
"learning_rate": 0.00019928491436984756,
|
| 17073 |
+
"loss": 7.3247,
|
| 17074 |
+
"step": 2423
|
| 17075 |
+
},
|
| 17076 |
+
{
|
| 17077 |
+
"epoch": 0.11568887138919712,
|
| 17078 |
+
"grad_norm": 5.1853861808776855,
|
| 17079 |
+
"learning_rate": 0.0001992843173111903,
|
| 17080 |
+
"loss": 5.7999,
|
| 17081 |
+
"step": 2424
|
| 17082 |
+
},
|
| 17083 |
+
{
|
| 17084 |
+
"epoch": 0.11573659782128838,
|
| 17085 |
+
"grad_norm": 6.79941463470459,
|
| 17086 |
+
"learning_rate": 0.000199283720004276,
|
| 17087 |
+
"loss": 5.7225,
|
| 17088 |
+
"step": 2425
|
| 17089 |
+
},
|
| 17090 |
+
{
|
| 17091 |
+
"epoch": 0.11578432425337963,
|
| 17092 |
+
"grad_norm": 6.806683540344238,
|
| 17093 |
+
"learning_rate": 0.00019928312244910616,
|
| 17094 |
+
"loss": 8.1717,
|
| 17095 |
+
"step": 2426
|
| 17096 |
+
},
|
| 17097 |
+
{
|
| 17098 |
+
"epoch": 0.11583205068547088,
|
| 17099 |
+
"grad_norm": 6.361385822296143,
|
| 17100 |
+
"learning_rate": 0.00019928252464568224,
|
| 17101 |
+
"loss": 7.508,
|
| 17102 |
+
"step": 2427
|
| 17103 |
+
},
|
| 17104 |
+
{
|
| 17105 |
+
"epoch": 0.11587977711756213,
|
| 17106 |
+
"grad_norm": 7.511900424957275,
|
| 17107 |
+
"learning_rate": 0.0001992819265940058,
|
| 17108 |
+
"loss": 7.6279,
|
| 17109 |
+
"step": 2428
|
| 17110 |
+
},
|
| 17111 |
+
{
|
| 17112 |
+
"epoch": 0.1159275035496534,
|
| 17113 |
+
"grad_norm": 6.408151149749756,
|
| 17114 |
+
"learning_rate": 0.00019928132829407825,
|
| 17115 |
+
"loss": 6.464,
|
| 17116 |
+
"step": 2429
|
| 17117 |
+
},
|
| 17118 |
+
{
|
| 17119 |
+
"epoch": 0.11597522998174464,
|
| 17120 |
+
"grad_norm": 6.161413192749023,
|
| 17121 |
+
"learning_rate": 0.00019928072974590116,
|
| 17122 |
+
"loss": 7.1698,
|
| 17123 |
+
"step": 2430
|
| 17124 |
+
},
|
| 17125 |
+
{
|
| 17126 |
+
"epoch": 0.11602295641383589,
|
| 17127 |
+
"grad_norm": 6.958288669586182,
|
| 17128 |
+
"learning_rate": 0.00019928013094947598,
|
| 17129 |
+
"loss": 8.821,
|
| 17130 |
+
"step": 2431
|
| 17131 |
+
},
|
| 17132 |
+
{
|
| 17133 |
+
"epoch": 0.11607068284592714,
|
| 17134 |
+
"grad_norm": 4.9188127517700195,
|
| 17135 |
+
"learning_rate": 0.00019927953190480423,
|
| 17136 |
+
"loss": 5.397,
|
| 17137 |
+
"step": 2432
|
| 17138 |
+
},
|
| 17139 |
+
{
|
| 17140 |
+
"epoch": 0.11611840927801839,
|
| 17141 |
+
"grad_norm": 5.682304382324219,
|
| 17142 |
+
"learning_rate": 0.0001992789326118874,
|
| 17143 |
+
"loss": 7.3442,
|
| 17144 |
+
"step": 2433
|
| 17145 |
+
},
|
| 17146 |
+
{
|
| 17147 |
+
"epoch": 0.11616613571010966,
|
| 17148 |
+
"grad_norm": 5.912992000579834,
|
| 17149 |
+
"learning_rate": 0.00019927833307072698,
|
| 17150 |
+
"loss": 6.825,
|
| 17151 |
+
"step": 2434
|
| 17152 |
+
},
|
| 17153 |
+
{
|
| 17154 |
+
"epoch": 0.1162138621422009,
|
| 17155 |
+
"grad_norm": 5.315921783447266,
|
| 17156 |
+
"learning_rate": 0.00019927773328132448,
|
| 17157 |
+
"loss": 4.3194,
|
| 17158 |
+
"step": 2435
|
| 17159 |
+
},
|
| 17160 |
+
{
|
| 17161 |
+
"epoch": 0.11626158857429215,
|
| 17162 |
+
"grad_norm": 6.76012659072876,
|
| 17163 |
+
"learning_rate": 0.00019927713324368145,
|
| 17164 |
+
"loss": 6.5639,
|
| 17165 |
+
"step": 2436
|
| 17166 |
+
},
|
| 17167 |
+
{
|
| 17168 |
+
"epoch": 0.1163093150063834,
|
| 17169 |
+
"grad_norm": 6.277029037475586,
|
| 17170 |
+
"learning_rate": 0.00019927653295779928,
|
| 17171 |
+
"loss": 6.0674,
|
| 17172 |
+
"step": 2437
|
| 17173 |
+
},
|
| 17174 |
+
{
|
| 17175 |
+
"epoch": 0.11635704143847467,
|
| 17176 |
+
"grad_norm": 5.538710117340088,
|
| 17177 |
+
"learning_rate": 0.00019927593242367956,
|
| 17178 |
+
"loss": 7.0892,
|
| 17179 |
+
"step": 2438
|
| 17180 |
+
},
|
| 17181 |
+
{
|
| 17182 |
+
"epoch": 0.11640476787056592,
|
| 17183 |
+
"grad_norm": 6.0107245445251465,
|
| 17184 |
+
"learning_rate": 0.00019927533164132376,
|
| 17185 |
+
"loss": 6.2243,
|
| 17186 |
+
"step": 2439
|
| 17187 |
+
},
|
| 17188 |
+
{
|
| 17189 |
+
"epoch": 0.11645249430265717,
|
| 17190 |
+
"grad_norm": 8.592703819274902,
|
| 17191 |
+
"learning_rate": 0.0001992747306107334,
|
| 17192 |
+
"loss": 5.2615,
|
| 17193 |
+
"step": 2440
|
| 17194 |
+
},
|
| 17195 |
+
{
|
| 17196 |
+
"epoch": 0.11650022073474842,
|
| 17197 |
+
"grad_norm": 5.739934921264648,
|
| 17198 |
+
"learning_rate": 0.00019927412933190992,
|
| 17199 |
+
"loss": 5.8984,
|
| 17200 |
+
"step": 2441
|
| 17201 |
+
},
|
| 17202 |
+
{
|
| 17203 |
+
"epoch": 0.11654794716683968,
|
| 17204 |
+
"grad_norm": 5.398654460906982,
|
| 17205 |
+
"learning_rate": 0.0001992735278048549,
|
| 17206 |
+
"loss": 5.6214,
|
| 17207 |
+
"step": 2442
|
| 17208 |
+
},
|
| 17209 |
+
{
|
| 17210 |
+
"epoch": 0.11659567359893093,
|
| 17211 |
+
"grad_norm": 6.197778224945068,
|
| 17212 |
+
"learning_rate": 0.00019927292602956982,
|
| 17213 |
+
"loss": 6.3904,
|
| 17214 |
+
"step": 2443
|
| 17215 |
+
},
|
| 17216 |
+
{
|
| 17217 |
+
"epoch": 0.11664340003102218,
|
| 17218 |
+
"grad_norm": 5.15541410446167,
|
| 17219 |
+
"learning_rate": 0.00019927232400605616,
|
| 17220 |
+
"loss": 5.1736,
|
| 17221 |
+
"step": 2444
|
| 17222 |
+
},
|
| 17223 |
+
{
|
| 17224 |
+
"epoch": 0.11669112646311343,
|
| 17225 |
+
"grad_norm": 7.815628528594971,
|
| 17226 |
+
"learning_rate": 0.00019927172173431546,
|
| 17227 |
+
"loss": 7.705,
|
| 17228 |
+
"step": 2445
|
| 17229 |
+
},
|
| 17230 |
+
{
|
| 17231 |
+
"epoch": 0.11673885289520469,
|
| 17232 |
+
"grad_norm": 6.1312642097473145,
|
| 17233 |
+
"learning_rate": 0.00019927111921434922,
|
| 17234 |
+
"loss": 6.9547,
|
| 17235 |
+
"step": 2446
|
| 17236 |
+
},
|
| 17237 |
+
{
|
| 17238 |
+
"epoch": 0.11678657932729594,
|
| 17239 |
+
"grad_norm": 5.695800304412842,
|
| 17240 |
+
"learning_rate": 0.00019927051644615893,
|
| 17241 |
+
"loss": 5.9635,
|
| 17242 |
+
"step": 2447
|
| 17243 |
+
},
|
| 17244 |
+
{
|
| 17245 |
+
"epoch": 0.11683430575938719,
|
| 17246 |
+
"grad_norm": 7.247018814086914,
|
| 17247 |
+
"learning_rate": 0.00019926991342974609,
|
| 17248 |
+
"loss": 6.8249,
|
| 17249 |
+
"step": 2448
|
| 17250 |
+
},
|
| 17251 |
+
{
|
| 17252 |
+
"epoch": 0.11688203219147844,
|
| 17253 |
+
"grad_norm": 6.522912979125977,
|
| 17254 |
+
"learning_rate": 0.00019926931016511223,
|
| 17255 |
+
"loss": 6.7136,
|
| 17256 |
+
"step": 2449
|
| 17257 |
+
},
|
| 17258 |
+
{
|
| 17259 |
+
"epoch": 0.1169297586235697,
|
| 17260 |
+
"grad_norm": 5.864326477050781,
|
| 17261 |
+
"learning_rate": 0.00019926870665225885,
|
| 17262 |
+
"loss": 6.6803,
|
| 17263 |
+
"step": 2450
|
| 17264 |
+
},
|
| 17265 |
+
{
|
| 17266 |
+
"epoch": 0.11697748505566095,
|
| 17267 |
+
"grad_norm": 6.620709419250488,
|
| 17268 |
+
"learning_rate": 0.0001992681028911875,
|
| 17269 |
+
"loss": 7.1,
|
| 17270 |
+
"step": 2451
|
| 17271 |
+
},
|
| 17272 |
+
{
|
| 17273 |
+
"epoch": 0.1170252114877522,
|
| 17274 |
+
"grad_norm": 6.445756435394287,
|
| 17275 |
+
"learning_rate": 0.0001992674988818996,
|
| 17276 |
+
"loss": 6.0606,
|
| 17277 |
+
"step": 2452
|
| 17278 |
+
},
|
| 17279 |
+
{
|
| 17280 |
+
"epoch": 0.11707293791984345,
|
| 17281 |
+
"grad_norm": 5.338548183441162,
|
| 17282 |
+
"learning_rate": 0.00019926689462439673,
|
| 17283 |
+
"loss": 4.951,
|
| 17284 |
+
"step": 2453
|
| 17285 |
+
},
|
| 17286 |
+
{
|
| 17287 |
+
"epoch": 0.11712066435193472,
|
| 17288 |
+
"grad_norm": 5.401038646697998,
|
| 17289 |
+
"learning_rate": 0.00019926629011868035,
|
| 17290 |
+
"loss": 5.7122,
|
| 17291 |
+
"step": 2454
|
| 17292 |
+
},
|
| 17293 |
+
{
|
| 17294 |
+
"epoch": 0.11716839078402597,
|
| 17295 |
+
"grad_norm": 7.513417720794678,
|
| 17296 |
+
"learning_rate": 0.00019926568536475205,
|
| 17297 |
+
"loss": 7.0187,
|
| 17298 |
+
"step": 2455
|
| 17299 |
+
},
|
| 17300 |
+
{
|
| 17301 |
+
"epoch": 0.11721611721611722,
|
| 17302 |
+
"grad_norm": 7.435956954956055,
|
| 17303 |
+
"learning_rate": 0.00019926508036261328,
|
| 17304 |
+
"loss": 6.212,
|
| 17305 |
+
"step": 2456
|
| 17306 |
+
},
|
| 17307 |
+
{
|
| 17308 |
+
"epoch": 0.11726384364820847,
|
| 17309 |
+
"grad_norm": 6.338344097137451,
|
| 17310 |
+
"learning_rate": 0.00019926447511226555,
|
| 17311 |
+
"loss": 7.8148,
|
| 17312 |
+
"step": 2457
|
| 17313 |
+
},
|
| 17314 |
+
{
|
| 17315 |
+
"epoch": 0.11731157008029972,
|
| 17316 |
+
"grad_norm": 6.532636642456055,
|
| 17317 |
+
"learning_rate": 0.00019926386961371042,
|
| 17318 |
+
"loss": 7.2181,
|
| 17319 |
+
"step": 2458
|
| 17320 |
+
},
|
| 17321 |
+
{
|
| 17322 |
+
"epoch": 0.11735929651239098,
|
| 17323 |
+
"grad_norm": 5.906259536743164,
|
| 17324 |
+
"learning_rate": 0.00019926326386694935,
|
| 17325 |
+
"loss": 7.3303,
|
| 17326 |
+
"step": 2459
|
| 17327 |
+
},
|
| 17328 |
+
{
|
| 17329 |
+
"epoch": 0.11740702294448223,
|
| 17330 |
+
"grad_norm": 5.578773498535156,
|
| 17331 |
+
"learning_rate": 0.00019926265787198388,
|
| 17332 |
+
"loss": 5.9464,
|
| 17333 |
+
"step": 2460
|
| 17334 |
+
},
|
| 17335 |
+
{
|
| 17336 |
+
"epoch": 0.11745474937657348,
|
| 17337 |
+
"grad_norm": 6.0162272453308105,
|
| 17338 |
+
"learning_rate": 0.00019926205162881555,
|
| 17339 |
+
"loss": 6.7526,
|
| 17340 |
+
"step": 2461
|
| 17341 |
+
},
|
| 17342 |
+
{
|
| 17343 |
+
"epoch": 0.11750247580866473,
|
| 17344 |
+
"grad_norm": 4.859315395355225,
|
| 17345 |
+
"learning_rate": 0.00019926144513744584,
|
| 17346 |
+
"loss": 5.4383,
|
| 17347 |
+
"step": 2462
|
| 17348 |
+
},
|
| 17349 |
+
{
|
| 17350 |
+
"epoch": 0.11755020224075599,
|
| 17351 |
+
"grad_norm": 5.332747936248779,
|
| 17352 |
+
"learning_rate": 0.00019926083839787628,
|
| 17353 |
+
"loss": 6.8147,
|
| 17354 |
+
"step": 2463
|
| 17355 |
+
},
|
| 17356 |
+
{
|
| 17357 |
+
"epoch": 0.11759792867284724,
|
| 17358 |
+
"grad_norm": 6.401904106140137,
|
| 17359 |
+
"learning_rate": 0.00019926023141010837,
|
| 17360 |
+
"loss": 5.5658,
|
| 17361 |
+
"step": 2464
|
| 17362 |
+
},
|
| 17363 |
+
{
|
| 17364 |
+
"epoch": 0.11764565510493849,
|
| 17365 |
+
"grad_norm": 8.178959846496582,
|
| 17366 |
+
"learning_rate": 0.00019925962417414366,
|
| 17367 |
+
"loss": 5.4159,
|
| 17368 |
+
"step": 2465
|
| 17369 |
+
},
|
| 17370 |
+
{
|
| 17371 |
+
"epoch": 0.11769338153702974,
|
| 17372 |
+
"grad_norm": 4.893832683563232,
|
| 17373 |
+
"learning_rate": 0.00019925901668998362,
|
| 17374 |
+
"loss": 6.2554,
|
| 17375 |
+
"step": 2466
|
| 17376 |
+
},
|
| 17377 |
+
{
|
| 17378 |
+
"epoch": 0.117741107969121,
|
| 17379 |
+
"grad_norm": 6.215792655944824,
|
| 17380 |
+
"learning_rate": 0.00019925840895762984,
|
| 17381 |
+
"loss": 5.6131,
|
| 17382 |
+
"step": 2467
|
| 17383 |
+
},
|
| 17384 |
+
{
|
| 17385 |
+
"epoch": 0.11778883440121225,
|
| 17386 |
+
"grad_norm": 103.55281829833984,
|
| 17387 |
+
"learning_rate": 0.00019925780097708378,
|
| 17388 |
+
"loss": 10.3719,
|
| 17389 |
+
"step": 2468
|
| 17390 |
+
},
|
| 17391 |
+
{
|
| 17392 |
+
"epoch": 0.1178365608333035,
|
| 17393 |
+
"grad_norm": 6.744437217712402,
|
| 17394 |
+
"learning_rate": 0.00019925719274834695,
|
| 17395 |
+
"loss": 4.8652,
|
| 17396 |
+
"step": 2469
|
| 17397 |
+
},
|
| 17398 |
+
{
|
| 17399 |
+
"epoch": 0.11788428726539475,
|
| 17400 |
+
"grad_norm": 6.799299240112305,
|
| 17401 |
+
"learning_rate": 0.00019925658427142096,
|
| 17402 |
+
"loss": 6.5246,
|
| 17403 |
+
"step": 2470
|
| 17404 |
+
},
|
| 17405 |
+
{
|
| 17406 |
+
"epoch": 0.11793201369748602,
|
| 17407 |
+
"grad_norm": 6.361636161804199,
|
| 17408 |
+
"learning_rate": 0.00019925597554630722,
|
| 17409 |
+
"loss": 7.6175,
|
| 17410 |
+
"step": 2471
|
| 17411 |
+
},
|
| 17412 |
+
{
|
| 17413 |
+
"epoch": 0.11797974012957727,
|
| 17414 |
+
"grad_norm": 4.605724334716797,
|
| 17415 |
+
"learning_rate": 0.00019925536657300734,
|
| 17416 |
+
"loss": 5.1498,
|
| 17417 |
+
"step": 2472
|
| 17418 |
+
},
|
| 17419 |
+
{
|
| 17420 |
+
"epoch": 0.11802746656166851,
|
| 17421 |
+
"grad_norm": 5.402807235717773,
|
| 17422 |
+
"learning_rate": 0.0001992547573515228,
|
| 17423 |
+
"loss": 6.0529,
|
| 17424 |
+
"step": 2473
|
| 17425 |
+
},
|
| 17426 |
+
{
|
| 17427 |
+
"epoch": 0.11807519299375976,
|
| 17428 |
+
"grad_norm": 8.144805908203125,
|
| 17429 |
+
"learning_rate": 0.00019925414788185512,
|
| 17430 |
+
"loss": 8.934,
|
| 17431 |
+
"step": 2474
|
| 17432 |
+
},
|
| 17433 |
+
{
|
| 17434 |
+
"epoch": 0.11812291942585103,
|
| 17435 |
+
"grad_norm": 10.172551155090332,
|
| 17436 |
+
"learning_rate": 0.00019925353816400583,
|
| 17437 |
+
"loss": 7.2408,
|
| 17438 |
+
"step": 2475
|
| 17439 |
+
},
|
| 17440 |
+
{
|
| 17441 |
+
"epoch": 0.11817064585794228,
|
| 17442 |
+
"grad_norm": 9.197928428649902,
|
| 17443 |
+
"learning_rate": 0.00019925292819797648,
|
| 17444 |
+
"loss": 6.8291,
|
| 17445 |
+
"step": 2476
|
| 17446 |
+
},
|
| 17447 |
+
{
|
| 17448 |
+
"epoch": 0.11821837229003353,
|
| 17449 |
+
"grad_norm": 606.15283203125,
|
| 17450 |
+
"learning_rate": 0.00019925231798376854,
|
| 17451 |
+
"loss": 8.0119,
|
| 17452 |
+
"step": 2477
|
| 17453 |
+
},
|
| 17454 |
+
{
|
| 17455 |
+
"epoch": 0.11826609872212478,
|
| 17456 |
+
"grad_norm": 16.890789031982422,
|
| 17457 |
+
"learning_rate": 0.0001992517075213836,
|
| 17458 |
+
"loss": 7.1259,
|
| 17459 |
+
"step": 2478
|
| 17460 |
+
},
|
| 17461 |
+
{
|
| 17462 |
+
"epoch": 0.11831382515421604,
|
| 17463 |
+
"grad_norm": 44.77732467651367,
|
| 17464 |
+
"learning_rate": 0.00019925109681082315,
|
| 17465 |
+
"loss": 7.1127,
|
| 17466 |
+
"step": 2479
|
| 17467 |
+
},
|
| 17468 |
+
{
|
| 17469 |
+
"epoch": 0.11836155158630729,
|
| 17470 |
+
"grad_norm": 278.022705078125,
|
| 17471 |
+
"learning_rate": 0.00019925048585208872,
|
| 17472 |
+
"loss": 6.3405,
|
| 17473 |
+
"step": 2480
|
| 17474 |
+
},
|
| 17475 |
+
{
|
| 17476 |
+
"epoch": 0.11840927801839854,
|
| 17477 |
+
"grad_norm": 9.286020278930664,
|
| 17478 |
+
"learning_rate": 0.0001992498746451818,
|
| 17479 |
+
"loss": 8.3945,
|
| 17480 |
+
"step": 2481
|
| 17481 |
+
},
|
| 17482 |
+
{
|
| 17483 |
+
"epoch": 0.11845700445048979,
|
| 17484 |
+
"grad_norm": 4.785390853881836,
|
| 17485 |
+
"learning_rate": 0.00019924926319010404,
|
| 17486 |
+
"loss": 5.2656,
|
| 17487 |
+
"step": 2482
|
| 17488 |
+
},
|
| 17489 |
+
{
|
| 17490 |
+
"epoch": 0.11850473088258104,
|
| 17491 |
+
"grad_norm": 5.759562969207764,
|
| 17492 |
+
"learning_rate": 0.00019924865148685683,
|
| 17493 |
+
"loss": 6.6575,
|
| 17494 |
+
"step": 2483
|
| 17495 |
+
},
|
| 17496 |
+
{
|
| 17497 |
+
"epoch": 0.1185524573146723,
|
| 17498 |
+
"grad_norm": 5.401790142059326,
|
| 17499 |
+
"learning_rate": 0.00019924803953544177,
|
| 17500 |
+
"loss": 6.0652,
|
| 17501 |
+
"step": 2484
|
| 17502 |
+
},
|
| 17503 |
+
{
|
| 17504 |
+
"epoch": 0.11860018374676355,
|
| 17505 |
+
"grad_norm": 6.856231212615967,
|
| 17506 |
+
"learning_rate": 0.00019924742733586038,
|
| 17507 |
+
"loss": 7.3388,
|
| 17508 |
+
"step": 2485
|
| 17509 |
+
},
|
| 17510 |
+
{
|
| 17511 |
+
"epoch": 0.1186479101788548,
|
| 17512 |
+
"grad_norm": 4.967160701751709,
|
| 17513 |
+
"learning_rate": 0.0001992468148881142,
|
| 17514 |
+
"loss": 6.0109,
|
| 17515 |
+
"step": 2486
|
| 17516 |
+
},
|
| 17517 |
+
{
|
| 17518 |
+
"epoch": 0.11869563661094605,
|
| 17519 |
+
"grad_norm": 7.75101900100708,
|
| 17520 |
+
"learning_rate": 0.00019924620219220472,
|
| 17521 |
+
"loss": 6.6262,
|
| 17522 |
+
"step": 2487
|
| 17523 |
+
},
|
| 17524 |
+
{
|
| 17525 |
+
"epoch": 0.11874336304303731,
|
| 17526 |
+
"grad_norm": 8.207610130310059,
|
| 17527 |
+
"learning_rate": 0.00019924558924813352,
|
| 17528 |
+
"loss": 7.5386,
|
| 17529 |
+
"step": 2488
|
| 17530 |
+
},
|
| 17531 |
+
{
|
| 17532 |
+
"epoch": 0.11879108947512856,
|
| 17533 |
+
"grad_norm": 6.139005661010742,
|
| 17534 |
+
"learning_rate": 0.00019924497605590212,
|
| 17535 |
+
"loss": 6.0931,
|
| 17536 |
+
"step": 2489
|
| 17537 |
+
},
|
| 17538 |
+
{
|
| 17539 |
+
"epoch": 0.11883881590721981,
|
| 17540 |
+
"grad_norm": 5.760986328125,
|
| 17541 |
+
"learning_rate": 0.00019924436261551206,
|
| 17542 |
+
"loss": 5.4283,
|
| 17543 |
+
"step": 2490
|
| 17544 |
+
},
|
| 17545 |
+
{
|
| 17546 |
+
"epoch": 0.11888654233931106,
|
| 17547 |
+
"grad_norm": 6.050114631652832,
|
| 17548 |
+
"learning_rate": 0.00019924374892696484,
|
| 17549 |
+
"loss": 6.4558,
|
| 17550 |
+
"step": 2491
|
| 17551 |
+
},
|
| 17552 |
+
{
|
| 17553 |
+
"epoch": 0.11893426877140233,
|
| 17554 |
+
"grad_norm": 6.406091690063477,
|
| 17555 |
+
"learning_rate": 0.00019924313499026201,
|
| 17556 |
+
"loss": 6.0646,
|
| 17557 |
+
"step": 2492
|
| 17558 |
+
},
|
| 17559 |
+
{
|
| 17560 |
+
"epoch": 0.11898199520349358,
|
| 17561 |
+
"grad_norm": 5.947465419769287,
|
| 17562 |
+
"learning_rate": 0.00019924252080540512,
|
| 17563 |
+
"loss": 7.7813,
|
| 17564 |
+
"step": 2493
|
| 17565 |
+
},
|
| 17566 |
+
{
|
| 17567 |
+
"epoch": 0.11902972163558483,
|
| 17568 |
+
"grad_norm": 6.351478576660156,
|
| 17569 |
+
"learning_rate": 0.00019924190637239571,
|
| 17570 |
+
"loss": 6.2155,
|
| 17571 |
+
"step": 2494
|
| 17572 |
+
},
|
| 17573 |
+
{
|
| 17574 |
+
"epoch": 0.11907744806767608,
|
| 17575 |
+
"grad_norm": 5.51107931137085,
|
| 17576 |
+
"learning_rate": 0.0001992412916912353,
|
| 17577 |
+
"loss": 5.8992,
|
| 17578 |
+
"step": 2495
|
| 17579 |
+
},
|
| 17580 |
+
{
|
| 17581 |
+
"epoch": 0.11912517449976734,
|
| 17582 |
+
"grad_norm": 5.350305080413818,
|
| 17583 |
+
"learning_rate": 0.00019924067676192544,
|
| 17584 |
+
"loss": 5.2331,
|
| 17585 |
+
"step": 2496
|
| 17586 |
+
},
|
| 17587 |
+
{
|
| 17588 |
+
"epoch": 0.11917290093185859,
|
| 17589 |
+
"grad_norm": 5.645650386810303,
|
| 17590 |
+
"learning_rate": 0.0001992400615844676,
|
| 17591 |
+
"loss": 5.9623,
|
| 17592 |
+
"step": 2497
|
| 17593 |
+
},
|
| 17594 |
+
{
|
| 17595 |
+
"epoch": 0.11922062736394984,
|
| 17596 |
+
"grad_norm": 5.326278209686279,
|
| 17597 |
+
"learning_rate": 0.00019923944615886344,
|
| 17598 |
+
"loss": 5.8638,
|
| 17599 |
+
"step": 2498
|
| 17600 |
+
},
|
| 17601 |
+
{
|
| 17602 |
+
"epoch": 0.11926835379604109,
|
| 17603 |
+
"grad_norm": 5.611948490142822,
|
| 17604 |
+
"learning_rate": 0.00019923883048511443,
|
| 17605 |
+
"loss": 6.6996,
|
| 17606 |
+
"step": 2499
|
| 17607 |
+
},
|
| 17608 |
+
{
|
| 17609 |
+
"epoch": 0.11931608022813235,
|
| 17610 |
+
"grad_norm": 5.947208404541016,
|
| 17611 |
+
"learning_rate": 0.00019923821456322208,
|
| 17612 |
+
"loss": 5.2053,
|
| 17613 |
+
"step": 2500
|
| 17614 |
+
},
|
| 17615 |
+
{
|
| 17616 |
+
"epoch": 0.1193638066602236,
|
| 17617 |
+
"grad_norm": 6.137307167053223,
|
| 17618 |
+
"learning_rate": 0.000199237598393188,
|
| 17619 |
+
"loss": 6.1276,
|
| 17620 |
+
"step": 2501
|
| 17621 |
+
},
|
| 17622 |
+
{
|
| 17623 |
+
"epoch": 0.11941153309231485,
|
| 17624 |
+
"grad_norm": 7.9021525382995605,
|
| 17625 |
+
"learning_rate": 0.0001992369819750137,
|
| 17626 |
+
"loss": 8.74,
|
| 17627 |
+
"step": 2502
|
| 17628 |
+
},
|
| 17629 |
+
{
|
| 17630 |
+
"epoch": 0.1194592595244061,
|
| 17631 |
+
"grad_norm": 8.133113861083984,
|
| 17632 |
+
"learning_rate": 0.00019923636530870068,
|
| 17633 |
+
"loss": 8.2945,
|
| 17634 |
+
"step": 2503
|
| 17635 |
+
},
|
| 17636 |
+
{
|
| 17637 |
+
"epoch": 0.11950698595649736,
|
| 17638 |
+
"grad_norm": 4.672719478607178,
|
| 17639 |
+
"learning_rate": 0.00019923574839425054,
|
| 17640 |
+
"loss": 5.661,
|
| 17641 |
+
"step": 2504
|
| 17642 |
+
},
|
| 17643 |
+
{
|
| 17644 |
+
"epoch": 0.11955471238858861,
|
| 17645 |
+
"grad_norm": 5.1231513023376465,
|
| 17646 |
+
"learning_rate": 0.0001992351312316648,
|
| 17647 |
+
"loss": 5.0149,
|
| 17648 |
+
"step": 2505
|
| 17649 |
+
},
|
| 17650 |
+
{
|
| 17651 |
+
"epoch": 0.11960243882067986,
|
| 17652 |
+
"grad_norm": 6.5395121574401855,
|
| 17653 |
+
"learning_rate": 0.00019923451382094498,
|
| 17654 |
+
"loss": 6.6384,
|
| 17655 |
+
"step": 2506
|
| 17656 |
+
},
|
| 17657 |
+
{
|
| 17658 |
+
"epoch": 0.11965016525277111,
|
| 17659 |
+
"grad_norm": 6.963812828063965,
|
| 17660 |
+
"learning_rate": 0.00019923389616209266,
|
| 17661 |
+
"loss": 7.6418,
|
| 17662 |
+
"step": 2507
|
| 17663 |
+
},
|
| 17664 |
+
{
|
| 17665 |
+
"epoch": 0.11969789168486236,
|
| 17666 |
+
"grad_norm": 5.314905643463135,
|
| 17667 |
+
"learning_rate": 0.0001992332782551094,
|
| 17668 |
+
"loss": 5.357,
|
| 17669 |
+
"step": 2508
|
| 17670 |
+
},
|
| 17671 |
+
{
|
| 17672 |
+
"epoch": 0.11974561811695363,
|
| 17673 |
+
"grad_norm": 6.953101634979248,
|
| 17674 |
+
"learning_rate": 0.0001992326600999967,
|
| 17675 |
+
"loss": 7.4247,
|
| 17676 |
+
"step": 2509
|
| 17677 |
+
},
|
| 17678 |
+
{
|
| 17679 |
+
"epoch": 0.11979334454904488,
|
| 17680 |
+
"grad_norm": 5.65818977355957,
|
| 17681 |
+
"learning_rate": 0.0001992320416967561,
|
| 17682 |
+
"loss": 6.9697,
|
| 17683 |
+
"step": 2510
|
| 17684 |
+
},
|
| 17685 |
+
{
|
| 17686 |
+
"epoch": 0.11984107098113612,
|
| 17687 |
+
"grad_norm": 5.882226467132568,
|
| 17688 |
+
"learning_rate": 0.0001992314230453892,
|
| 17689 |
+
"loss": 6.6894,
|
| 17690 |
+
"step": 2511
|
| 17691 |
+
},
|
| 17692 |
+
{
|
| 17693 |
+
"epoch": 0.11988879741322737,
|
| 17694 |
+
"grad_norm": 7.456608295440674,
|
| 17695 |
+
"learning_rate": 0.00019923080414589752,
|
| 17696 |
+
"loss": 6.5442,
|
| 17697 |
+
"step": 2512
|
| 17698 |
+
},
|
| 17699 |
+
{
|
| 17700 |
+
"epoch": 0.11993652384531864,
|
| 17701 |
+
"grad_norm": 5.068438529968262,
|
| 17702 |
+
"learning_rate": 0.0001992301849982826,
|
| 17703 |
+
"loss": 5.6924,
|
| 17704 |
+
"step": 2513
|
| 17705 |
+
},
|
| 17706 |
+
{
|
| 17707 |
+
"epoch": 0.11998425027740989,
|
| 17708 |
+
"grad_norm": 5.8276519775390625,
|
| 17709 |
+
"learning_rate": 0.00019922956560254599,
|
| 17710 |
+
"loss": 4.5949,
|
| 17711 |
+
"step": 2514
|
| 17712 |
+
},
|
| 17713 |
+
{
|
| 17714 |
+
"epoch": 0.12003197670950114,
|
| 17715 |
+
"grad_norm": 7.3889617919921875,
|
| 17716 |
+
"learning_rate": 0.00019922894595868923,
|
| 17717 |
+
"loss": 7.027,
|
| 17718 |
+
"step": 2515
|
| 17719 |
+
},
|
| 17720 |
+
{
|
| 17721 |
+
"epoch": 0.12007970314159239,
|
| 17722 |
+
"grad_norm": 5.222685813903809,
|
| 17723 |
+
"learning_rate": 0.00019922832606671386,
|
| 17724 |
+
"loss": 5.7551,
|
| 17725 |
+
"step": 2516
|
| 17726 |
+
},
|
| 17727 |
+
{
|
| 17728 |
+
"epoch": 0.12012742957368365,
|
| 17729 |
+
"grad_norm": 5.6962127685546875,
|
| 17730 |
+
"learning_rate": 0.0001992277059266215,
|
| 17731 |
+
"loss": 5.0557,
|
| 17732 |
+
"step": 2517
|
| 17733 |
+
},
|
| 17734 |
+
{
|
| 17735 |
+
"epoch": 0.1201751560057749,
|
| 17736 |
+
"grad_norm": 5.449974060058594,
|
| 17737 |
+
"learning_rate": 0.00019922708553841363,
|
| 17738 |
+
"loss": 6.3609,
|
| 17739 |
+
"step": 2518
|
| 17740 |
+
},
|
| 17741 |
+
{
|
| 17742 |
+
"epoch": 0.12022288243786615,
|
| 17743 |
+
"grad_norm": 5.272806167602539,
|
| 17744 |
+
"learning_rate": 0.00019922646490209183,
|
| 17745 |
+
"loss": 5.3785,
|
| 17746 |
+
"step": 2519
|
| 17747 |
+
},
|
| 17748 |
+
{
|
| 17749 |
+
"epoch": 0.1202706088699574,
|
| 17750 |
+
"grad_norm": 6.4713921546936035,
|
| 17751 |
+
"learning_rate": 0.00019922584401765763,
|
| 17752 |
+
"loss": 7.2317,
|
| 17753 |
+
"step": 2520
|
| 17754 |
+
},
|
| 17755 |
+
{
|
| 17756 |
+
"epoch": 0.12031833530204866,
|
| 17757 |
+
"grad_norm": 5.631217002868652,
|
| 17758 |
+
"learning_rate": 0.0001992252228851126,
|
| 17759 |
+
"loss": 5.7852,
|
| 17760 |
+
"step": 2521
|
| 17761 |
+
},
|
| 17762 |
+
{
|
| 17763 |
+
"epoch": 0.12036606173413991,
|
| 17764 |
+
"grad_norm": 6.146111488342285,
|
| 17765 |
+
"learning_rate": 0.0001992246015044583,
|
| 17766 |
+
"loss": 5.7136,
|
| 17767 |
+
"step": 2522
|
| 17768 |
+
},
|
| 17769 |
+
{
|
| 17770 |
+
"epoch": 0.12041378816623116,
|
| 17771 |
+
"grad_norm": 7.096044063568115,
|
| 17772 |
+
"learning_rate": 0.00019922397987569626,
|
| 17773 |
+
"loss": 7.2231,
|
| 17774 |
+
"step": 2523
|
| 17775 |
+
},
|
| 17776 |
+
{
|
| 17777 |
+
"epoch": 0.12046151459832241,
|
| 17778 |
+
"grad_norm": 7.921234607696533,
|
| 17779 |
+
"learning_rate": 0.00019922335799882806,
|
| 17780 |
+
"loss": 5.5,
|
| 17781 |
+
"step": 2524
|
| 17782 |
+
},
|
| 17783 |
+
{
|
| 17784 |
+
"epoch": 0.12050924103041367,
|
| 17785 |
+
"grad_norm": 5.085870265960693,
|
| 17786 |
+
"learning_rate": 0.00019922273587385524,
|
| 17787 |
+
"loss": 4.867,
|
| 17788 |
+
"step": 2525
|
| 17789 |
+
},
|
| 17790 |
+
{
|
| 17791 |
+
"epoch": 0.12055696746250492,
|
| 17792 |
+
"grad_norm": 6.939750671386719,
|
| 17793 |
+
"learning_rate": 0.00019922211350077934,
|
| 17794 |
+
"loss": 6.7561,
|
| 17795 |
+
"step": 2526
|
| 17796 |
+
},
|
| 17797 |
+
{
|
| 17798 |
+
"epoch": 0.12060469389459617,
|
| 17799 |
+
"grad_norm": 7.504082679748535,
|
| 17800 |
+
"learning_rate": 0.00019922149087960196,
|
| 17801 |
+
"loss": 7.1749,
|
| 17802 |
+
"step": 2527
|
| 17803 |
+
},
|
| 17804 |
+
{
|
| 17805 |
+
"epoch": 0.12065242032668742,
|
| 17806 |
+
"grad_norm": 6.798505783081055,
|
| 17807 |
+
"learning_rate": 0.00019922086801032462,
|
| 17808 |
+
"loss": 6.8441,
|
| 17809 |
+
"step": 2528
|
| 17810 |
+
},
|
| 17811 |
+
{
|
| 17812 |
+
"epoch": 0.12070014675877869,
|
| 17813 |
+
"grad_norm": 6.547543048858643,
|
| 17814 |
+
"learning_rate": 0.00019922024489294892,
|
| 17815 |
+
"loss": 4.8769,
|
| 17816 |
+
"step": 2529
|
| 17817 |
+
},
|
| 17818 |
+
{
|
| 17819 |
+
"epoch": 0.12074787319086994,
|
| 17820 |
+
"grad_norm": 8.873817443847656,
|
| 17821 |
+
"learning_rate": 0.00019921962152747634,
|
| 17822 |
+
"loss": 9.6741,
|
| 17823 |
+
"step": 2530
|
| 17824 |
+
},
|
| 17825 |
+
{
|
| 17826 |
+
"epoch": 0.12079559962296119,
|
| 17827 |
+
"grad_norm": 6.475502014160156,
|
| 17828 |
+
"learning_rate": 0.0001992189979139085,
|
| 17829 |
+
"loss": 6.5672,
|
| 17830 |
+
"step": 2531
|
| 17831 |
+
},
|
| 17832 |
+
{
|
| 17833 |
+
"epoch": 0.12084332605505244,
|
| 17834 |
+
"grad_norm": 4.202572345733643,
|
| 17835 |
+
"learning_rate": 0.00019921837405224694,
|
| 17836 |
+
"loss": 3.9452,
|
| 17837 |
+
"step": 2532
|
| 17838 |
+
},
|
| 17839 |
+
{
|
| 17840 |
+
"epoch": 0.1208910524871437,
|
| 17841 |
+
"grad_norm": 6.775956153869629,
|
| 17842 |
+
"learning_rate": 0.00019921774994249324,
|
| 17843 |
+
"loss": 4.9947,
|
| 17844 |
+
"step": 2533
|
| 17845 |
+
},
|
| 17846 |
+
{
|
| 17847 |
+
"epoch": 0.12093877891923495,
|
| 17848 |
+
"grad_norm": 7.0207695960998535,
|
| 17849 |
+
"learning_rate": 0.00019921712558464895,
|
| 17850 |
+
"loss": 7.7427,
|
| 17851 |
+
"step": 2534
|
| 17852 |
+
},
|
| 17853 |
+
{
|
| 17854 |
+
"epoch": 0.1209865053513262,
|
| 17855 |
+
"grad_norm": 6.175810813903809,
|
| 17856 |
+
"learning_rate": 0.00019921650097871562,
|
| 17857 |
+
"loss": 6.728,
|
| 17858 |
+
"step": 2535
|
| 17859 |
+
},
|
| 17860 |
+
{
|
| 17861 |
+
"epoch": 0.12103423178341745,
|
| 17862 |
+
"grad_norm": 6.027743816375732,
|
| 17863 |
+
"learning_rate": 0.0001992158761246948,
|
| 17864 |
+
"loss": 6.0176,
|
| 17865 |
+
"step": 2536
|
| 17866 |
+
},
|
| 17867 |
+
{
|
| 17868 |
+
"epoch": 0.1210819582155087,
|
| 17869 |
+
"grad_norm": 7.800528049468994,
|
| 17870 |
+
"learning_rate": 0.00019921525102258811,
|
| 17871 |
+
"loss": 7.3638,
|
| 17872 |
+
"step": 2537
|
| 17873 |
+
},
|
| 17874 |
+
{
|
| 17875 |
+
"epoch": 0.12112968464759996,
|
| 17876 |
+
"grad_norm": 6.8329267501831055,
|
| 17877 |
+
"learning_rate": 0.00019921462567239705,
|
| 17878 |
+
"loss": 7.3964,
|
| 17879 |
+
"step": 2538
|
| 17880 |
+
},
|
| 17881 |
+
{
|
| 17882 |
+
"epoch": 0.12117741107969121,
|
| 17883 |
+
"grad_norm": 6.654507637023926,
|
| 17884 |
+
"learning_rate": 0.0001992140000741232,
|
| 17885 |
+
"loss": 7.6877,
|
| 17886 |
+
"step": 2539
|
| 17887 |
+
},
|
| 17888 |
+
{
|
| 17889 |
+
"epoch": 0.12122513751178246,
|
| 17890 |
+
"grad_norm": 6.265644073486328,
|
| 17891 |
+
"learning_rate": 0.00019921337422776816,
|
| 17892 |
+
"loss": 6.2082,
|
| 17893 |
+
"step": 2540
|
| 17894 |
+
},
|
| 17895 |
+
{
|
| 17896 |
+
"epoch": 0.12127286394387371,
|
| 17897 |
+
"grad_norm": 5.6409196853637695,
|
| 17898 |
+
"learning_rate": 0.00019921274813333346,
|
| 17899 |
+
"loss": 6.288,
|
| 17900 |
+
"step": 2541
|
| 17901 |
+
},
|
| 17902 |
+
{
|
| 17903 |
+
"epoch": 0.12132059037596497,
|
| 17904 |
+
"grad_norm": 6.226003170013428,
|
| 17905 |
+
"learning_rate": 0.00019921212179082064,
|
| 17906 |
+
"loss": 6.6538,
|
| 17907 |
+
"step": 2542
|
| 17908 |
+
},
|
| 17909 |
+
{
|
| 17910 |
+
"epoch": 0.12136831680805622,
|
| 17911 |
+
"grad_norm": 4.8436384201049805,
|
| 17912 |
+
"learning_rate": 0.00019921149520023135,
|
| 17913 |
+
"loss": 4.5971,
|
| 17914 |
+
"step": 2543
|
| 17915 |
+
},
|
| 17916 |
+
{
|
| 17917 |
+
"epoch": 0.12141604324014747,
|
| 17918 |
+
"grad_norm": 6.016689777374268,
|
| 17919 |
+
"learning_rate": 0.00019921086836156707,
|
| 17920 |
+
"loss": 6.8544,
|
| 17921 |
+
"step": 2544
|
| 17922 |
+
},
|
| 17923 |
+
{
|
| 17924 |
+
"epoch": 0.12146376967223872,
|
| 17925 |
+
"grad_norm": 4.865900039672852,
|
| 17926 |
+
"learning_rate": 0.0001992102412748294,
|
| 17927 |
+
"loss": 4.7976,
|
| 17928 |
+
"step": 2545
|
| 17929 |
+
},
|
| 17930 |
+
{
|
| 17931 |
+
"epoch": 0.12151149610432999,
|
| 17932 |
+
"grad_norm": 4.419651508331299,
|
| 17933 |
+
"learning_rate": 0.0001992096139400199,
|
| 17934 |
+
"loss": 5.0167,
|
| 17935 |
+
"step": 2546
|
| 17936 |
+
},
|
| 17937 |
+
{
|
| 17938 |
+
"epoch": 0.12155922253642124,
|
| 17939 |
+
"grad_norm": 6.08146858215332,
|
| 17940 |
+
"learning_rate": 0.0001992089863571402,
|
| 17941 |
+
"loss": 6.2101,
|
| 17942 |
+
"step": 2547
|
| 17943 |
+
},
|
| 17944 |
+
{
|
| 17945 |
+
"epoch": 0.12160694896851248,
|
| 17946 |
+
"grad_norm": 6.761473178863525,
|
| 17947 |
+
"learning_rate": 0.00019920835852619176,
|
| 17948 |
+
"loss": 6.9868,
|
| 17949 |
+
"step": 2548
|
| 17950 |
+
},
|
| 17951 |
+
{
|
| 17952 |
+
"epoch": 0.12165467540060373,
|
| 17953 |
+
"grad_norm": 4.87887716293335,
|
| 17954 |
+
"learning_rate": 0.00019920773044717626,
|
| 17955 |
+
"loss": 5.0713,
|
| 17956 |
+
"step": 2549
|
| 17957 |
+
},
|
| 17958 |
+
{
|
| 17959 |
+
"epoch": 0.121702401832695,
|
| 17960 |
+
"grad_norm": 5.633859157562256,
|
| 17961 |
+
"learning_rate": 0.0001992071021200952,
|
| 17962 |
+
"loss": 6.1883,
|
| 17963 |
+
"step": 2550
|
| 17964 |
+
},
|
| 17965 |
+
{
|
| 17966 |
+
"epoch": 0.12175012826478625,
|
| 17967 |
+
"grad_norm": 6.28723669052124,
|
| 17968 |
+
"learning_rate": 0.0001992064735449502,
|
| 17969 |
+
"loss": 5.2651,
|
| 17970 |
+
"step": 2551
|
| 17971 |
+
},
|
| 17972 |
+
{
|
| 17973 |
+
"epoch": 0.1217978546968775,
|
| 17974 |
+
"grad_norm": 5.057219982147217,
|
| 17975 |
+
"learning_rate": 0.00019920584472174274,
|
| 17976 |
+
"loss": 4.7898,
|
| 17977 |
+
"step": 2552
|
| 17978 |
+
},
|
| 17979 |
+
{
|
| 17980 |
+
"epoch": 0.12184558112896875,
|
| 17981 |
+
"grad_norm": 7.6184821128845215,
|
| 17982 |
+
"learning_rate": 0.0001992052156504745,
|
| 17983 |
+
"loss": 7.9772,
|
| 17984 |
+
"step": 2553
|
| 17985 |
+
},
|
| 17986 |
+
{
|
| 17987 |
+
"epoch": 0.12189330756106001,
|
| 17988 |
+
"grad_norm": 7.5784173011779785,
|
| 17989 |
+
"learning_rate": 0.000199204586331147,
|
| 17990 |
+
"loss": 7.3766,
|
| 17991 |
+
"step": 2554
|
| 17992 |
+
},
|
| 17993 |
+
{
|
| 17994 |
+
"epoch": 0.12194103399315126,
|
| 17995 |
+
"grad_norm": 6.459564208984375,
|
| 17996 |
+
"learning_rate": 0.00019920395676376181,
|
| 17997 |
+
"loss": 5.3843,
|
| 17998 |
+
"step": 2555
|
| 17999 |
+
},
|
| 18000 |
+
{
|
| 18001 |
+
"epoch": 0.12198876042524251,
|
| 18002 |
+
"grad_norm": 8.033954620361328,
|
| 18003 |
+
"learning_rate": 0.00019920332694832048,
|
| 18004 |
+
"loss": 8.4444,
|
| 18005 |
+
"step": 2556
|
| 18006 |
+
},
|
| 18007 |
+
{
|
| 18008 |
+
"epoch": 0.12203648685733376,
|
| 18009 |
+
"grad_norm": 7.824962615966797,
|
| 18010 |
+
"learning_rate": 0.00019920269688482466,
|
| 18011 |
+
"loss": 7.2764,
|
| 18012 |
+
"step": 2557
|
| 18013 |
+
},
|
| 18014 |
+
{
|
| 18015 |
+
"epoch": 0.12208421328942502,
|
| 18016 |
+
"grad_norm": 7.024128437042236,
|
| 18017 |
+
"learning_rate": 0.00019920206657327588,
|
| 18018 |
+
"loss": 6.7798,
|
| 18019 |
+
"step": 2558
|
| 18020 |
+
},
|
| 18021 |
+
{
|
| 18022 |
+
"epoch": 0.12213193972151627,
|
| 18023 |
+
"grad_norm": 7.976163864135742,
|
| 18024 |
+
"learning_rate": 0.00019920143601367575,
|
| 18025 |
+
"loss": 6.9575,
|
| 18026 |
+
"step": 2559
|
| 18027 |
+
},
|
| 18028 |
+
{
|
| 18029 |
+
"epoch": 0.12217966615360752,
|
| 18030 |
+
"grad_norm": 6.063900470733643,
|
| 18031 |
+
"learning_rate": 0.0001992008052060258,
|
| 18032 |
+
"loss": 5.2735,
|
| 18033 |
+
"step": 2560
|
| 18034 |
+
},
|
| 18035 |
+
{
|
| 18036 |
+
"epoch": 0.12222739258569877,
|
| 18037 |
+
"grad_norm": 10.203490257263184,
|
| 18038 |
+
"learning_rate": 0.0001992001741503276,
|
| 18039 |
+
"loss": 6.9637,
|
| 18040 |
+
"step": 2561
|
| 18041 |
+
},
|
| 18042 |
+
{
|
| 18043 |
+
"epoch": 0.12227511901779002,
|
| 18044 |
+
"grad_norm": 6.814461708068848,
|
| 18045 |
+
"learning_rate": 0.0001991995428465828,
|
| 18046 |
+
"loss": 7.4683,
|
| 18047 |
+
"step": 2562
|
| 18048 |
+
},
|
| 18049 |
+
{
|
| 18050 |
+
"epoch": 0.12232284544988128,
|
| 18051 |
+
"grad_norm": 5.394496917724609,
|
| 18052 |
+
"learning_rate": 0.00019919891129479292,
|
| 18053 |
+
"loss": 5.9349,
|
| 18054 |
+
"step": 2563
|
| 18055 |
+
},
|
| 18056 |
+
{
|
| 18057 |
+
"epoch": 0.12237057188197253,
|
| 18058 |
+
"grad_norm": 7.58441162109375,
|
| 18059 |
+
"learning_rate": 0.00019919827949495952,
|
| 18060 |
+
"loss": 6.5405,
|
| 18061 |
+
"step": 2564
|
| 18062 |
+
},
|
| 18063 |
+
{
|
| 18064 |
+
"epoch": 0.12241829831406378,
|
| 18065 |
+
"grad_norm": 5.798213958740234,
|
| 18066 |
+
"learning_rate": 0.00019919764744708422,
|
| 18067 |
+
"loss": 6.6679,
|
| 18068 |
+
"step": 2565
|
| 18069 |
+
},
|
| 18070 |
+
{
|
| 18071 |
+
"epoch": 0.12246602474615503,
|
| 18072 |
+
"grad_norm": 5.101596355438232,
|
| 18073 |
+
"learning_rate": 0.0001991970151511686,
|
| 18074 |
+
"loss": 5.3157,
|
| 18075 |
+
"step": 2566
|
| 18076 |
+
},
|
| 18077 |
+
{
|
| 18078 |
+
"epoch": 0.1225137511782463,
|
| 18079 |
+
"grad_norm": 4.84085750579834,
|
| 18080 |
+
"learning_rate": 0.00019919638260721423,
|
| 18081 |
+
"loss": 5.0614,
|
| 18082 |
+
"step": 2567
|
| 18083 |
+
},
|
| 18084 |
+
{
|
| 18085 |
+
"epoch": 0.12256147761033755,
|
| 18086 |
+
"grad_norm": 6.472250461578369,
|
| 18087 |
+
"learning_rate": 0.00019919574981522268,
|
| 18088 |
+
"loss": 6.5636,
|
| 18089 |
+
"step": 2568
|
| 18090 |
+
},
|
| 18091 |
+
{
|
| 18092 |
+
"epoch": 0.1226092040424288,
|
| 18093 |
+
"grad_norm": 5.555691719055176,
|
| 18094 |
+
"learning_rate": 0.00019919511677519557,
|
| 18095 |
+
"loss": 5.9893,
|
| 18096 |
+
"step": 2569
|
| 18097 |
+
},
|
| 18098 |
+
{
|
| 18099 |
+
"epoch": 0.12265693047452005,
|
| 18100 |
+
"grad_norm": 6.802445411682129,
|
| 18101 |
+
"learning_rate": 0.00019919448348713445,
|
| 18102 |
+
"loss": 7.2901,
|
| 18103 |
+
"step": 2570
|
| 18104 |
+
},
|
| 18105 |
+
{
|
| 18106 |
+
"epoch": 0.12270465690661131,
|
| 18107 |
+
"grad_norm": 5.74739408493042,
|
| 18108 |
+
"learning_rate": 0.00019919384995104093,
|
| 18109 |
+
"loss": 5.9203,
|
| 18110 |
+
"step": 2571
|
| 18111 |
+
},
|
| 18112 |
+
{
|
| 18113 |
+
"epoch": 0.12275238333870256,
|
| 18114 |
+
"grad_norm": 6.161860942840576,
|
| 18115 |
+
"learning_rate": 0.00019919321616691655,
|
| 18116 |
+
"loss": 7.1177,
|
| 18117 |
+
"step": 2572
|
| 18118 |
+
},
|
| 18119 |
+
{
|
| 18120 |
+
"epoch": 0.12280010977079381,
|
| 18121 |
+
"grad_norm": 5.15494441986084,
|
| 18122 |
+
"learning_rate": 0.00019919258213476292,
|
| 18123 |
+
"loss": 5.502,
|
| 18124 |
+
"step": 2573
|
| 18125 |
+
},
|
| 18126 |
+
{
|
| 18127 |
+
"epoch": 0.12284783620288506,
|
| 18128 |
+
"grad_norm": 6.421548366546631,
|
| 18129 |
+
"learning_rate": 0.00019919194785458167,
|
| 18130 |
+
"loss": 5.7436,
|
| 18131 |
+
"step": 2574
|
| 18132 |
+
},
|
| 18133 |
+
{
|
| 18134 |
+
"epoch": 0.12289556263497632,
|
| 18135 |
+
"grad_norm": 6.586711883544922,
|
| 18136 |
+
"learning_rate": 0.0001991913133263743,
|
| 18137 |
+
"loss": 6.9608,
|
| 18138 |
+
"step": 2575
|
| 18139 |
+
},
|
| 18140 |
+
{
|
| 18141 |
+
"epoch": 0.12294328906706757,
|
| 18142 |
+
"grad_norm": 5.92076301574707,
|
| 18143 |
+
"learning_rate": 0.00019919067855014247,
|
| 18144 |
+
"loss": 6.092,
|
| 18145 |
+
"step": 2576
|
| 18146 |
+
},
|
| 18147 |
+
{
|
| 18148 |
+
"epoch": 0.12299101549915882,
|
| 18149 |
+
"grad_norm": 6.7879862785339355,
|
| 18150 |
+
"learning_rate": 0.00019919004352588767,
|
| 18151 |
+
"loss": 6.7476,
|
| 18152 |
+
"step": 2577
|
| 18153 |
+
},
|
| 18154 |
+
{
|
| 18155 |
+
"epoch": 0.12303874193125007,
|
| 18156 |
+
"grad_norm": 8.76463508605957,
|
| 18157 |
+
"learning_rate": 0.00019918940825361164,
|
| 18158 |
+
"loss": 8.9568,
|
| 18159 |
+
"step": 2578
|
| 18160 |
+
},
|
| 18161 |
+
{
|
| 18162 |
+
"epoch": 0.12308646836334133,
|
| 18163 |
+
"grad_norm": 4.8239665031433105,
|
| 18164 |
+
"learning_rate": 0.00019918877273331583,
|
| 18165 |
+
"loss": 5.2598,
|
| 18166 |
+
"step": 2579
|
| 18167 |
+
},
|
| 18168 |
+
{
|
| 18169 |
+
"epoch": 0.12313419479543258,
|
| 18170 |
+
"grad_norm": 6.723465919494629,
|
| 18171 |
+
"learning_rate": 0.00019918813696500189,
|
| 18172 |
+
"loss": 6.6971,
|
| 18173 |
+
"step": 2580
|
| 18174 |
+
},
|
| 18175 |
+
{
|
| 18176 |
+
"epoch": 0.12318192122752383,
|
| 18177 |
+
"grad_norm": 6.38710355758667,
|
| 18178 |
+
"learning_rate": 0.00019918750094867144,
|
| 18179 |
+
"loss": 5.7313,
|
| 18180 |
+
"step": 2581
|
| 18181 |
+
},
|
| 18182 |
+
{
|
| 18183 |
+
"epoch": 0.12322964765961508,
|
| 18184 |
+
"grad_norm": 6.35195779800415,
|
| 18185 |
+
"learning_rate": 0.000199186864684326,
|
| 18186 |
+
"loss": 7.8379,
|
| 18187 |
+
"step": 2582
|
| 18188 |
+
},
|
| 18189 |
+
{
|
| 18190 |
+
"epoch": 0.12327737409170635,
|
| 18191 |
+
"grad_norm": 6.694102764129639,
|
| 18192 |
+
"learning_rate": 0.0001991862281719672,
|
| 18193 |
+
"loss": 6.2678,
|
| 18194 |
+
"step": 2583
|
| 18195 |
+
},
|
| 18196 |
+
{
|
| 18197 |
+
"epoch": 0.1233251005237976,
|
| 18198 |
+
"grad_norm": 6.242228984832764,
|
| 18199 |
+
"learning_rate": 0.00019918559141159664,
|
| 18200 |
+
"loss": 5.813,
|
| 18201 |
+
"step": 2584
|
| 18202 |
+
},
|
| 18203 |
+
{
|
| 18204 |
+
"epoch": 0.12337282695588885,
|
| 18205 |
+
"grad_norm": 7.03825569152832,
|
| 18206 |
+
"learning_rate": 0.00019918495440321586,
|
| 18207 |
+
"loss": 6.2827,
|
| 18208 |
+
"step": 2585
|
| 18209 |
+
},
|
| 18210 |
+
{
|
| 18211 |
+
"epoch": 0.1234205533879801,
|
| 18212 |
+
"grad_norm": 4.9181389808654785,
|
| 18213 |
+
"learning_rate": 0.0001991843171468265,
|
| 18214 |
+
"loss": 4.9544,
|
| 18215 |
+
"step": 2586
|
| 18216 |
+
},
|
| 18217 |
+
{
|
| 18218 |
+
"epoch": 0.12346827982007134,
|
| 18219 |
+
"grad_norm": 7.315206050872803,
|
| 18220 |
+
"learning_rate": 0.00019918367964243014,
|
| 18221 |
+
"loss": 8.3353,
|
| 18222 |
+
"step": 2587
|
| 18223 |
+
},
|
| 18224 |
+
{
|
| 18225 |
+
"epoch": 0.12351600625216261,
|
| 18226 |
+
"grad_norm": 6.019472122192383,
|
| 18227 |
+
"learning_rate": 0.0001991830418900284,
|
| 18228 |
+
"loss": 6.734,
|
| 18229 |
+
"step": 2588
|
| 18230 |
+
},
|
| 18231 |
+
{
|
| 18232 |
+
"epoch": 0.12356373268425386,
|
| 18233 |
+
"grad_norm": 5.389978885650635,
|
| 18234 |
+
"learning_rate": 0.00019918240388962284,
|
| 18235 |
+
"loss": 6.5434,
|
| 18236 |
+
"step": 2589
|
| 18237 |
+
},
|
| 18238 |
+
{
|
| 18239 |
+
"epoch": 0.12361145911634511,
|
| 18240 |
+
"grad_norm": 6.509700298309326,
|
| 18241 |
+
"learning_rate": 0.00019918176564121508,
|
| 18242 |
+
"loss": 7.3647,
|
| 18243 |
+
"step": 2590
|
| 18244 |
+
},
|
| 18245 |
+
{
|
| 18246 |
+
"epoch": 0.12365918554843636,
|
| 18247 |
+
"grad_norm": 8.090117454528809,
|
| 18248 |
+
"learning_rate": 0.0001991811271448067,
|
| 18249 |
+
"loss": 7.2696,
|
| 18250 |
+
"step": 2591
|
| 18251 |
+
},
|
| 18252 |
+
{
|
| 18253 |
+
"epoch": 0.12370691198052762,
|
| 18254 |
+
"grad_norm": 5.416383266448975,
|
| 18255 |
+
"learning_rate": 0.00019918048840039928,
|
| 18256 |
+
"loss": 6.4276,
|
| 18257 |
+
"step": 2592
|
| 18258 |
+
},
|
| 18259 |
+
{
|
| 18260 |
+
"epoch": 0.12375463841261887,
|
| 18261 |
+
"grad_norm": 6.610803127288818,
|
| 18262 |
+
"learning_rate": 0.00019917984940799445,
|
| 18263 |
+
"loss": 5.4854,
|
| 18264 |
+
"step": 2593
|
| 18265 |
+
},
|
| 18266 |
+
{
|
| 18267 |
+
"epoch": 0.12380236484471012,
|
| 18268 |
+
"grad_norm": 4.918051242828369,
|
| 18269 |
+
"learning_rate": 0.0001991792101675938,
|
| 18270 |
+
"loss": 4.2269,
|
| 18271 |
+
"step": 2594
|
| 18272 |
+
},
|
| 18273 |
+
{
|
| 18274 |
+
"epoch": 0.12385009127680137,
|
| 18275 |
+
"grad_norm": 6.454699993133545,
|
| 18276 |
+
"learning_rate": 0.0001991785706791989,
|
| 18277 |
+
"loss": 6.3255,
|
| 18278 |
+
"step": 2595
|
| 18279 |
+
},
|
| 18280 |
+
{
|
| 18281 |
+
"epoch": 0.12389781770889263,
|
| 18282 |
+
"grad_norm": 10.754586219787598,
|
| 18283 |
+
"learning_rate": 0.0001991779309428114,
|
| 18284 |
+
"loss": 10.1455,
|
| 18285 |
+
"step": 2596
|
| 18286 |
+
},
|
| 18287 |
+
{
|
| 18288 |
+
"epoch": 0.12394554414098388,
|
| 18289 |
+
"grad_norm": 7.033049583435059,
|
| 18290 |
+
"learning_rate": 0.00019917729095843286,
|
| 18291 |
+
"loss": 6.9458,
|
| 18292 |
+
"step": 2597
|
| 18293 |
+
},
|
| 18294 |
+
{
|
| 18295 |
+
"epoch": 0.12399327057307513,
|
| 18296 |
+
"grad_norm": 5.376648426055908,
|
| 18297 |
+
"learning_rate": 0.0001991766507260649,
|
| 18298 |
+
"loss": 5.361,
|
| 18299 |
+
"step": 2598
|
| 18300 |
+
},
|
| 18301 |
+
{
|
| 18302 |
+
"epoch": 0.12404099700516638,
|
| 18303 |
+
"grad_norm": 7.248912811279297,
|
| 18304 |
+
"learning_rate": 0.0001991760102457091,
|
| 18305 |
+
"loss": 7.2075,
|
| 18306 |
+
"step": 2599
|
| 18307 |
+
},
|
| 18308 |
+
{
|
| 18309 |
+
"epoch": 0.12408872343725764,
|
| 18310 |
+
"grad_norm": 5.227941989898682,
|
| 18311 |
+
"learning_rate": 0.0001991753695173671,
|
| 18312 |
+
"loss": 5.2552,
|
| 18313 |
+
"step": 2600
|
| 18314 |
+
},
|
| 18315 |
+
{
|
| 18316 |
+
"epoch": 0.12408872343725764,
|
| 18317 |
+
"eval_loss": 1.616467833518982,
|
| 18318 |
+
"eval_runtime": 96.4878,
|
| 18319 |
+
"eval_samples_per_second": 8.737,
|
| 18320 |
+
"eval_steps_per_second": 4.374,
|
| 18321 |
+
"step": 2600
|
| 18322 |
}
|
| 18323 |
],
|
| 18324 |
"logging_steps": 1,
|
|
|
|
| 18333 |
"early_stopping_threshold": 0.0
|
| 18334 |
},
|
| 18335 |
"attributes": {
|
| 18336 |
+
"early_stopping_patience_counter": 2
|
| 18337 |
}
|
| 18338 |
},
|
| 18339 |
"TrainerControl": {
|
|
|
|
| 18347 |
"attributes": {}
|
| 18348 |
}
|
| 18349 |
},
|
| 18350 |
+
"total_flos": 2.410799318433792e+17,
|
| 18351 |
"train_batch_size": 2,
|
| 18352 |
"trial_name": null,
|
| 18353 |
"trial_params": null
|