Training in progress, step 1200, 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 335604696
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:43dc3e67c4dd1820bbd2270c7e0b12353f9c7369b0bcf3fad771b83ac8447e3d
|
| 3 |
size 335604696
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 170920532
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a98ec5d1c2471dadc6ffbe5e4f0ddc0eccb1bf4306b2e42107164b9ba171922c
|
| 3 |
size 170920532
|
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 sha256:140ab48bfe96a1d0da1e6c5c3b63214bcd35f3a91e2df3df83ed97a96aa7a467
|
| 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 sha256:040ca33075c084cc6ee3696af40933b3a026e71b68847c6f3c2869db1e7ef945
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.9684450030326843,
|
| 3 |
"best_model_checkpoint": "miner_id_24/checkpoint-600",
|
| 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,
|
|
@@ -7055,6 +7055,1414 @@
|
|
| 7055 |
"eval_samples_per_second": 6.453,
|
| 7056 |
"eval_steps_per_second": 3.239,
|
| 7057 |
"step": 1000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7058 |
}
|
| 7059 |
],
|
| 7060 |
"logging_steps": 1,
|
|
@@ -7069,7 +8477,7 @@
|
|
| 7069 |
"early_stopping_threshold": 0.0
|
| 7070 |
},
|
| 7071 |
"attributes": {
|
| 7072 |
-
"early_stopping_patience_counter":
|
| 7073 |
}
|
| 7074 |
},
|
| 7075 |
"TrainerControl": {
|
|
@@ -7078,12 +8486,12 @@
|
|
| 7078 |
"should_evaluate": false,
|
| 7079 |
"should_log": false,
|
| 7080 |
"should_save": true,
|
| 7081 |
-
"should_training_stop":
|
| 7082 |
},
|
| 7083 |
"attributes": {}
|
| 7084 |
}
|
| 7085 |
},
|
| 7086 |
-
"total_flos": 1.
|
| 7087 |
"train_batch_size": 2,
|
| 7088 |
"trial_name": null,
|
| 7089 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.9684450030326843,
|
| 3 |
"best_model_checkpoint": "miner_id_24/checkpoint-600",
|
| 4 |
+
"epoch": 0.18482864844050828,
|
| 5 |
"eval_steps": 200,
|
| 6 |
+
"global_step": 1200,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 7055 |
"eval_samples_per_second": 6.453,
|
| 7056 |
"eval_steps_per_second": 3.239,
|
| 7057 |
"step": 1000
|
| 7058 |
+
},
|
| 7059 |
+
{
|
| 7060 |
+
"epoch": 0.15417789757412398,
|
| 7061 |
+
"grad_norm": 13.766051292419434,
|
| 7062 |
+
"learning_rate": 0.00019877211733515348,
|
| 7063 |
+
"loss": 4.6898,
|
| 7064 |
+
"step": 1001
|
| 7065 |
+
},
|
| 7066 |
+
{
|
| 7067 |
+
"epoch": 0.1543319214478244,
|
| 7068 |
+
"grad_norm": 11.941194534301758,
|
| 7069 |
+
"learning_rate": 0.00019876959212519163,
|
| 7070 |
+
"loss": 4.276,
|
| 7071 |
+
"step": 1002
|
| 7072 |
+
},
|
| 7073 |
+
{
|
| 7074 |
+
"epoch": 0.15448594532152485,
|
| 7075 |
+
"grad_norm": 14.202241897583008,
|
| 7076 |
+
"learning_rate": 0.00019876706433735134,
|
| 7077 |
+
"loss": 3.6016,
|
| 7078 |
+
"step": 1003
|
| 7079 |
+
},
|
| 7080 |
+
{
|
| 7081 |
+
"epoch": 0.15463996919522527,
|
| 7082 |
+
"grad_norm": 15.519389152526855,
|
| 7083 |
+
"learning_rate": 0.0001987645339716985,
|
| 7084 |
+
"loss": 4.9497,
|
| 7085 |
+
"step": 1004
|
| 7086 |
+
},
|
| 7087 |
+
{
|
| 7088 |
+
"epoch": 0.1547939930689257,
|
| 7089 |
+
"grad_norm": 16.54033660888672,
|
| 7090 |
+
"learning_rate": 0.0001987620010282992,
|
| 7091 |
+
"loss": 3.3404,
|
| 7092 |
+
"step": 1005
|
| 7093 |
+
},
|
| 7094 |
+
{
|
| 7095 |
+
"epoch": 0.1549480169426261,
|
| 7096 |
+
"grad_norm": 10.095898628234863,
|
| 7097 |
+
"learning_rate": 0.00019875946550721954,
|
| 7098 |
+
"loss": 3.3351,
|
| 7099 |
+
"step": 1006
|
| 7100 |
+
},
|
| 7101 |
+
{
|
| 7102 |
+
"epoch": 0.15510204081632653,
|
| 7103 |
+
"grad_norm": 14.892594337463379,
|
| 7104 |
+
"learning_rate": 0.0001987569274085257,
|
| 7105 |
+
"loss": 3.2248,
|
| 7106 |
+
"step": 1007
|
| 7107 |
+
},
|
| 7108 |
+
{
|
| 7109 |
+
"epoch": 0.15525606469002695,
|
| 7110 |
+
"grad_norm": 19.26637840270996,
|
| 7111 |
+
"learning_rate": 0.00019875438673228393,
|
| 7112 |
+
"loss": 4.6162,
|
| 7113 |
+
"step": 1008
|
| 7114 |
+
},
|
| 7115 |
+
{
|
| 7116 |
+
"epoch": 0.15541008856372737,
|
| 7117 |
+
"grad_norm": 13.137590408325195,
|
| 7118 |
+
"learning_rate": 0.00019875184347856054,
|
| 7119 |
+
"loss": 3.9372,
|
| 7120 |
+
"step": 1009
|
| 7121 |
+
},
|
| 7122 |
+
{
|
| 7123 |
+
"epoch": 0.1555641124374278,
|
| 7124 |
+
"grad_norm": 14.01908016204834,
|
| 7125 |
+
"learning_rate": 0.0001987492976474219,
|
| 7126 |
+
"loss": 4.2331,
|
| 7127 |
+
"step": 1010
|
| 7128 |
+
},
|
| 7129 |
+
{
|
| 7130 |
+
"epoch": 0.1557181363111282,
|
| 7131 |
+
"grad_norm": 12.255888938903809,
|
| 7132 |
+
"learning_rate": 0.00019874674923893443,
|
| 7133 |
+
"loss": 4.0807,
|
| 7134 |
+
"step": 1011
|
| 7135 |
+
},
|
| 7136 |
+
{
|
| 7137 |
+
"epoch": 0.15587216018482866,
|
| 7138 |
+
"grad_norm": 13.065942764282227,
|
| 7139 |
+
"learning_rate": 0.0001987441982531647,
|
| 7140 |
+
"loss": 4.3647,
|
| 7141 |
+
"step": 1012
|
| 7142 |
+
},
|
| 7143 |
+
{
|
| 7144 |
+
"epoch": 0.15602618405852908,
|
| 7145 |
+
"grad_norm": 9.445647239685059,
|
| 7146 |
+
"learning_rate": 0.00019874164469017925,
|
| 7147 |
+
"loss": 3.397,
|
| 7148 |
+
"step": 1013
|
| 7149 |
+
},
|
| 7150 |
+
{
|
| 7151 |
+
"epoch": 0.1561802079322295,
|
| 7152 |
+
"grad_norm": 17.212614059448242,
|
| 7153 |
+
"learning_rate": 0.00019873908855004476,
|
| 7154 |
+
"loss": 4.5294,
|
| 7155 |
+
"step": 1014
|
| 7156 |
+
},
|
| 7157 |
+
{
|
| 7158 |
+
"epoch": 0.15633423180592992,
|
| 7159 |
+
"grad_norm": 12.967076301574707,
|
| 7160 |
+
"learning_rate": 0.00019873652983282792,
|
| 7161 |
+
"loss": 4.6614,
|
| 7162 |
+
"step": 1015
|
| 7163 |
+
},
|
| 7164 |
+
{
|
| 7165 |
+
"epoch": 0.15648825567963034,
|
| 7166 |
+
"grad_norm": 12.160512924194336,
|
| 7167 |
+
"learning_rate": 0.00019873396853859553,
|
| 7168 |
+
"loss": 3.7298,
|
| 7169 |
+
"step": 1016
|
| 7170 |
+
},
|
| 7171 |
+
{
|
| 7172 |
+
"epoch": 0.15664227955333077,
|
| 7173 |
+
"grad_norm": 10.568840980529785,
|
| 7174 |
+
"learning_rate": 0.00019873140466741438,
|
| 7175 |
+
"loss": 4.1098,
|
| 7176 |
+
"step": 1017
|
| 7177 |
+
},
|
| 7178 |
+
{
|
| 7179 |
+
"epoch": 0.1567963034270312,
|
| 7180 |
+
"grad_norm": 10.045523643493652,
|
| 7181 |
+
"learning_rate": 0.00019872883821935147,
|
| 7182 |
+
"loss": 3.9284,
|
| 7183 |
+
"step": 1018
|
| 7184 |
+
},
|
| 7185 |
+
{
|
| 7186 |
+
"epoch": 0.1569503273007316,
|
| 7187 |
+
"grad_norm": 12.165972709655762,
|
| 7188 |
+
"learning_rate": 0.00019872626919447374,
|
| 7189 |
+
"loss": 3.5241,
|
| 7190 |
+
"step": 1019
|
| 7191 |
+
},
|
| 7192 |
+
{
|
| 7193 |
+
"epoch": 0.15710435117443203,
|
| 7194 |
+
"grad_norm": 13.945481300354004,
|
| 7195 |
+
"learning_rate": 0.00019872369759284827,
|
| 7196 |
+
"loss": 4.4213,
|
| 7197 |
+
"step": 1020
|
| 7198 |
+
},
|
| 7199 |
+
{
|
| 7200 |
+
"epoch": 0.15725837504813245,
|
| 7201 |
+
"grad_norm": 14.0652437210083,
|
| 7202 |
+
"learning_rate": 0.00019872112341454213,
|
| 7203 |
+
"loss": 4.4031,
|
| 7204 |
+
"step": 1021
|
| 7205 |
+
},
|
| 7206 |
+
{
|
| 7207 |
+
"epoch": 0.1574123989218329,
|
| 7208 |
+
"grad_norm": 11.660938262939453,
|
| 7209 |
+
"learning_rate": 0.00019871854665962256,
|
| 7210 |
+
"loss": 3.1114,
|
| 7211 |
+
"step": 1022
|
| 7212 |
+
},
|
| 7213 |
+
{
|
| 7214 |
+
"epoch": 0.15756642279553332,
|
| 7215 |
+
"grad_norm": 11.772308349609375,
|
| 7216 |
+
"learning_rate": 0.00019871596732815675,
|
| 7217 |
+
"loss": 4.7416,
|
| 7218 |
+
"step": 1023
|
| 7219 |
+
},
|
| 7220 |
+
{
|
| 7221 |
+
"epoch": 0.15772044666923374,
|
| 7222 |
+
"grad_norm": 10.968940734863281,
|
| 7223 |
+
"learning_rate": 0.0001987133854202121,
|
| 7224 |
+
"loss": 2.9122,
|
| 7225 |
+
"step": 1024
|
| 7226 |
+
},
|
| 7227 |
+
{
|
| 7228 |
+
"epoch": 0.15787447054293416,
|
| 7229 |
+
"grad_norm": 11.12600326538086,
|
| 7230 |
+
"learning_rate": 0.00019871080093585594,
|
| 7231 |
+
"loss": 3.5455,
|
| 7232 |
+
"step": 1025
|
| 7233 |
+
},
|
| 7234 |
+
{
|
| 7235 |
+
"epoch": 0.15802849441663458,
|
| 7236 |
+
"grad_norm": 9.834585189819336,
|
| 7237 |
+
"learning_rate": 0.00019870821387515573,
|
| 7238 |
+
"loss": 3.1168,
|
| 7239 |
+
"step": 1026
|
| 7240 |
+
},
|
| 7241 |
+
{
|
| 7242 |
+
"epoch": 0.158182518290335,
|
| 7243 |
+
"grad_norm": 15.116072654724121,
|
| 7244 |
+
"learning_rate": 0.000198705624238179,
|
| 7245 |
+
"loss": 4.0013,
|
| 7246 |
+
"step": 1027
|
| 7247 |
+
},
|
| 7248 |
+
{
|
| 7249 |
+
"epoch": 0.15833654216403542,
|
| 7250 |
+
"grad_norm": 10.250100135803223,
|
| 7251 |
+
"learning_rate": 0.00019870303202499336,
|
| 7252 |
+
"loss": 3.0386,
|
| 7253 |
+
"step": 1028
|
| 7254 |
+
},
|
| 7255 |
+
{
|
| 7256 |
+
"epoch": 0.15849056603773584,
|
| 7257 |
+
"grad_norm": 13.875730514526367,
|
| 7258 |
+
"learning_rate": 0.00019870043723566646,
|
| 7259 |
+
"loss": 4.0619,
|
| 7260 |
+
"step": 1029
|
| 7261 |
+
},
|
| 7262 |
+
{
|
| 7263 |
+
"epoch": 0.15864458991143626,
|
| 7264 |
+
"grad_norm": 10.340214729309082,
|
| 7265 |
+
"learning_rate": 0.00019869783987026596,
|
| 7266 |
+
"loss": 3.6729,
|
| 7267 |
+
"step": 1030
|
| 7268 |
+
},
|
| 7269 |
+
{
|
| 7270 |
+
"epoch": 0.15879861378513668,
|
| 7271 |
+
"grad_norm": 11.000602722167969,
|
| 7272 |
+
"learning_rate": 0.00019869523992885975,
|
| 7273 |
+
"loss": 3.91,
|
| 7274 |
+
"step": 1031
|
| 7275 |
+
},
|
| 7276 |
+
{
|
| 7277 |
+
"epoch": 0.15895263765883713,
|
| 7278 |
+
"grad_norm": 11.994555473327637,
|
| 7279 |
+
"learning_rate": 0.0001986926374115156,
|
| 7280 |
+
"loss": 4.1724,
|
| 7281 |
+
"step": 1032
|
| 7282 |
+
},
|
| 7283 |
+
{
|
| 7284 |
+
"epoch": 0.15910666153253755,
|
| 7285 |
+
"grad_norm": 12.995914459228516,
|
| 7286 |
+
"learning_rate": 0.0001986900323183015,
|
| 7287 |
+
"loss": 3.8724,
|
| 7288 |
+
"step": 1033
|
| 7289 |
+
},
|
| 7290 |
+
{
|
| 7291 |
+
"epoch": 0.15926068540623797,
|
| 7292 |
+
"grad_norm": 10.791321754455566,
|
| 7293 |
+
"learning_rate": 0.00019868742464928542,
|
| 7294 |
+
"loss": 3.7237,
|
| 7295 |
+
"step": 1034
|
| 7296 |
+
},
|
| 7297 |
+
{
|
| 7298 |
+
"epoch": 0.1594147092799384,
|
| 7299 |
+
"grad_norm": 9.404729843139648,
|
| 7300 |
+
"learning_rate": 0.00019868481440453542,
|
| 7301 |
+
"loss": 4.1626,
|
| 7302 |
+
"step": 1035
|
| 7303 |
+
},
|
| 7304 |
+
{
|
| 7305 |
+
"epoch": 0.15956873315363881,
|
| 7306 |
+
"grad_norm": 13.564080238342285,
|
| 7307 |
+
"learning_rate": 0.00019868220158411964,
|
| 7308 |
+
"loss": 5.0002,
|
| 7309 |
+
"step": 1036
|
| 7310 |
+
},
|
| 7311 |
+
{
|
| 7312 |
+
"epoch": 0.15972275702733923,
|
| 7313 |
+
"grad_norm": 11.500726699829102,
|
| 7314 |
+
"learning_rate": 0.00019867958618810625,
|
| 7315 |
+
"loss": 4.0815,
|
| 7316 |
+
"step": 1037
|
| 7317 |
+
},
|
| 7318 |
+
{
|
| 7319 |
+
"epoch": 0.15987678090103966,
|
| 7320 |
+
"grad_norm": 12.543424606323242,
|
| 7321 |
+
"learning_rate": 0.0001986769682165635,
|
| 7322 |
+
"loss": 3.9962,
|
| 7323 |
+
"step": 1038
|
| 7324 |
+
},
|
| 7325 |
+
{
|
| 7326 |
+
"epoch": 0.16003080477474008,
|
| 7327 |
+
"grad_norm": 14.21129035949707,
|
| 7328 |
+
"learning_rate": 0.00019867434766955976,
|
| 7329 |
+
"loss": 4.1891,
|
| 7330 |
+
"step": 1039
|
| 7331 |
+
},
|
| 7332 |
+
{
|
| 7333 |
+
"epoch": 0.1601848286484405,
|
| 7334 |
+
"grad_norm": 12.011770248413086,
|
| 7335 |
+
"learning_rate": 0.00019867172454716343,
|
| 7336 |
+
"loss": 3.8437,
|
| 7337 |
+
"step": 1040
|
| 7338 |
+
},
|
| 7339 |
+
{
|
| 7340 |
+
"epoch": 0.16033885252214095,
|
| 7341 |
+
"grad_norm": 9.131732940673828,
|
| 7342 |
+
"learning_rate": 0.0001986690988494429,
|
| 7343 |
+
"loss": 2.8849,
|
| 7344 |
+
"step": 1041
|
| 7345 |
+
},
|
| 7346 |
+
{
|
| 7347 |
+
"epoch": 0.16049287639584137,
|
| 7348 |
+
"grad_norm": 11.264022827148438,
|
| 7349 |
+
"learning_rate": 0.0001986664705764668,
|
| 7350 |
+
"loss": 3.5034,
|
| 7351 |
+
"step": 1042
|
| 7352 |
+
},
|
| 7353 |
+
{
|
| 7354 |
+
"epoch": 0.1606469002695418,
|
| 7355 |
+
"grad_norm": 10.755845069885254,
|
| 7356 |
+
"learning_rate": 0.00019866383972830365,
|
| 7357 |
+
"loss": 3.9478,
|
| 7358 |
+
"step": 1043
|
| 7359 |
+
},
|
| 7360 |
+
{
|
| 7361 |
+
"epoch": 0.1608009241432422,
|
| 7362 |
+
"grad_norm": 11.155594825744629,
|
| 7363 |
+
"learning_rate": 0.00019866120630502215,
|
| 7364 |
+
"loss": 3.8415,
|
| 7365 |
+
"step": 1044
|
| 7366 |
+
},
|
| 7367 |
+
{
|
| 7368 |
+
"epoch": 0.16095494801694263,
|
| 7369 |
+
"grad_norm": 11.492998123168945,
|
| 7370 |
+
"learning_rate": 0.00019865857030669103,
|
| 7371 |
+
"loss": 3.6514,
|
| 7372 |
+
"step": 1045
|
| 7373 |
+
},
|
| 7374 |
+
{
|
| 7375 |
+
"epoch": 0.16110897189064305,
|
| 7376 |
+
"grad_norm": 12.405851364135742,
|
| 7377 |
+
"learning_rate": 0.00019865593173337912,
|
| 7378 |
+
"loss": 4.4151,
|
| 7379 |
+
"step": 1046
|
| 7380 |
+
},
|
| 7381 |
+
{
|
| 7382 |
+
"epoch": 0.16126299576434347,
|
| 7383 |
+
"grad_norm": 14.843172073364258,
|
| 7384 |
+
"learning_rate": 0.0001986532905851552,
|
| 7385 |
+
"loss": 4.0748,
|
| 7386 |
+
"step": 1047
|
| 7387 |
+
},
|
| 7388 |
+
{
|
| 7389 |
+
"epoch": 0.1614170196380439,
|
| 7390 |
+
"grad_norm": 13.064985275268555,
|
| 7391 |
+
"learning_rate": 0.00019865064686208826,
|
| 7392 |
+
"loss": 3.959,
|
| 7393 |
+
"step": 1048
|
| 7394 |
+
},
|
| 7395 |
+
{
|
| 7396 |
+
"epoch": 0.1615710435117443,
|
| 7397 |
+
"grad_norm": 10.112977027893066,
|
| 7398 |
+
"learning_rate": 0.0001986480005642473,
|
| 7399 |
+
"loss": 4.4101,
|
| 7400 |
+
"step": 1049
|
| 7401 |
+
},
|
| 7402 |
+
{
|
| 7403 |
+
"epoch": 0.16172506738544473,
|
| 7404 |
+
"grad_norm": 13.397727012634277,
|
| 7405 |
+
"learning_rate": 0.00019864535169170143,
|
| 7406 |
+
"loss": 4.5429,
|
| 7407 |
+
"step": 1050
|
| 7408 |
+
},
|
| 7409 |
+
{
|
| 7410 |
+
"epoch": 0.16187909125914518,
|
| 7411 |
+
"grad_norm": 11.528743743896484,
|
| 7412 |
+
"learning_rate": 0.00019864270024451966,
|
| 7413 |
+
"loss": 4.748,
|
| 7414 |
+
"step": 1051
|
| 7415 |
+
},
|
| 7416 |
+
{
|
| 7417 |
+
"epoch": 0.1620331151328456,
|
| 7418 |
+
"grad_norm": 11.48531436920166,
|
| 7419 |
+
"learning_rate": 0.00019864004622277133,
|
| 7420 |
+
"loss": 4.2897,
|
| 7421 |
+
"step": 1052
|
| 7422 |
+
},
|
| 7423 |
+
{
|
| 7424 |
+
"epoch": 0.16218713900654602,
|
| 7425 |
+
"grad_norm": 9.885492324829102,
|
| 7426 |
+
"learning_rate": 0.00019863738962652564,
|
| 7427 |
+
"loss": 3.9293,
|
| 7428 |
+
"step": 1053
|
| 7429 |
+
},
|
| 7430 |
+
{
|
| 7431 |
+
"epoch": 0.16234116288024644,
|
| 7432 |
+
"grad_norm": 11.141953468322754,
|
| 7433 |
+
"learning_rate": 0.00019863473045585194,
|
| 7434 |
+
"loss": 3.8289,
|
| 7435 |
+
"step": 1054
|
| 7436 |
+
},
|
| 7437 |
+
{
|
| 7438 |
+
"epoch": 0.16249518675394686,
|
| 7439 |
+
"grad_norm": 12.676016807556152,
|
| 7440 |
+
"learning_rate": 0.00019863206871081963,
|
| 7441 |
+
"loss": 4.2164,
|
| 7442 |
+
"step": 1055
|
| 7443 |
+
},
|
| 7444 |
+
{
|
| 7445 |
+
"epoch": 0.16264921062764728,
|
| 7446 |
+
"grad_norm": 11.620012283325195,
|
| 7447 |
+
"learning_rate": 0.0001986294043914982,
|
| 7448 |
+
"loss": 3.8046,
|
| 7449 |
+
"step": 1056
|
| 7450 |
+
},
|
| 7451 |
+
{
|
| 7452 |
+
"epoch": 0.1628032345013477,
|
| 7453 |
+
"grad_norm": 15.559549331665039,
|
| 7454 |
+
"learning_rate": 0.00019862673749795716,
|
| 7455 |
+
"loss": 3.3443,
|
| 7456 |
+
"step": 1057
|
| 7457 |
+
},
|
| 7458 |
+
{
|
| 7459 |
+
"epoch": 0.16295725837504813,
|
| 7460 |
+
"grad_norm": 10.871837615966797,
|
| 7461 |
+
"learning_rate": 0.00019862406803026613,
|
| 7462 |
+
"loss": 4.1935,
|
| 7463 |
+
"step": 1058
|
| 7464 |
+
},
|
| 7465 |
+
{
|
| 7466 |
+
"epoch": 0.16311128224874855,
|
| 7467 |
+
"grad_norm": 11.921292304992676,
|
| 7468 |
+
"learning_rate": 0.0001986213959884948,
|
| 7469 |
+
"loss": 4.2124,
|
| 7470 |
+
"step": 1059
|
| 7471 |
+
},
|
| 7472 |
+
{
|
| 7473 |
+
"epoch": 0.16326530612244897,
|
| 7474 |
+
"grad_norm": 15.105125427246094,
|
| 7475 |
+
"learning_rate": 0.00019861872137271285,
|
| 7476 |
+
"loss": 4.2305,
|
| 7477 |
+
"step": 1060
|
| 7478 |
+
},
|
| 7479 |
+
{
|
| 7480 |
+
"epoch": 0.16341932999614942,
|
| 7481 |
+
"grad_norm": 17.650894165039062,
|
| 7482 |
+
"learning_rate": 0.00019861604418299015,
|
| 7483 |
+
"loss": 3.8902,
|
| 7484 |
+
"step": 1061
|
| 7485 |
+
},
|
| 7486 |
+
{
|
| 7487 |
+
"epoch": 0.16357335386984984,
|
| 7488 |
+
"grad_norm": 10.674752235412598,
|
| 7489 |
+
"learning_rate": 0.00019861336441939656,
|
| 7490 |
+
"loss": 4.66,
|
| 7491 |
+
"step": 1062
|
| 7492 |
+
},
|
| 7493 |
+
{
|
| 7494 |
+
"epoch": 0.16372737774355026,
|
| 7495 |
+
"grad_norm": 12.202083587646484,
|
| 7496 |
+
"learning_rate": 0.000198610682082002,
|
| 7497 |
+
"loss": 4.3085,
|
| 7498 |
+
"step": 1063
|
| 7499 |
+
},
|
| 7500 |
+
{
|
| 7501 |
+
"epoch": 0.16388140161725068,
|
| 7502 |
+
"grad_norm": 12.168536186218262,
|
| 7503 |
+
"learning_rate": 0.00019860799717087653,
|
| 7504 |
+
"loss": 3.9641,
|
| 7505 |
+
"step": 1064
|
| 7506 |
+
},
|
| 7507 |
+
{
|
| 7508 |
+
"epoch": 0.1640354254909511,
|
| 7509 |
+
"grad_norm": 13.570276260375977,
|
| 7510 |
+
"learning_rate": 0.00019860530968609017,
|
| 7511 |
+
"loss": 4.1331,
|
| 7512 |
+
"step": 1065
|
| 7513 |
+
},
|
| 7514 |
+
{
|
| 7515 |
+
"epoch": 0.16418944936465152,
|
| 7516 |
+
"grad_norm": 14.154974937438965,
|
| 7517 |
+
"learning_rate": 0.00019860261962771305,
|
| 7518 |
+
"loss": 4.6501,
|
| 7519 |
+
"step": 1066
|
| 7520 |
+
},
|
| 7521 |
+
{
|
| 7522 |
+
"epoch": 0.16434347323835194,
|
| 7523 |
+
"grad_norm": 10.965476989746094,
|
| 7524 |
+
"learning_rate": 0.00019859992699581546,
|
| 7525 |
+
"loss": 4.6674,
|
| 7526 |
+
"step": 1067
|
| 7527 |
+
},
|
| 7528 |
+
{
|
| 7529 |
+
"epoch": 0.16449749711205236,
|
| 7530 |
+
"grad_norm": 10.643817901611328,
|
| 7531 |
+
"learning_rate": 0.0001985972317904676,
|
| 7532 |
+
"loss": 4.053,
|
| 7533 |
+
"step": 1068
|
| 7534 |
+
},
|
| 7535 |
+
{
|
| 7536 |
+
"epoch": 0.16465152098575278,
|
| 7537 |
+
"grad_norm": 13.31205940246582,
|
| 7538 |
+
"learning_rate": 0.00019859453401173987,
|
| 7539 |
+
"loss": 3.4928,
|
| 7540 |
+
"step": 1069
|
| 7541 |
+
},
|
| 7542 |
+
{
|
| 7543 |
+
"epoch": 0.16480554485945323,
|
| 7544 |
+
"grad_norm": 12.753203392028809,
|
| 7545 |
+
"learning_rate": 0.00019859183365970265,
|
| 7546 |
+
"loss": 4.4367,
|
| 7547 |
+
"step": 1070
|
| 7548 |
+
},
|
| 7549 |
+
{
|
| 7550 |
+
"epoch": 0.16495956873315365,
|
| 7551 |
+
"grad_norm": 12.968934059143066,
|
| 7552 |
+
"learning_rate": 0.00019858913073442645,
|
| 7553 |
+
"loss": 3.9816,
|
| 7554 |
+
"step": 1071
|
| 7555 |
+
},
|
| 7556 |
+
{
|
| 7557 |
+
"epoch": 0.16511359260685407,
|
| 7558 |
+
"grad_norm": 10.421770095825195,
|
| 7559 |
+
"learning_rate": 0.00019858642523598173,
|
| 7560 |
+
"loss": 3.779,
|
| 7561 |
+
"step": 1072
|
| 7562 |
+
},
|
| 7563 |
+
{
|
| 7564 |
+
"epoch": 0.1652676164805545,
|
| 7565 |
+
"grad_norm": 12.015799522399902,
|
| 7566 |
+
"learning_rate": 0.00019858371716443922,
|
| 7567 |
+
"loss": 2.9187,
|
| 7568 |
+
"step": 1073
|
| 7569 |
+
},
|
| 7570 |
+
{
|
| 7571 |
+
"epoch": 0.1654216403542549,
|
| 7572 |
+
"grad_norm": 12.27148723602295,
|
| 7573 |
+
"learning_rate": 0.00019858100651986953,
|
| 7574 |
+
"loss": 4.164,
|
| 7575 |
+
"step": 1074
|
| 7576 |
+
},
|
| 7577 |
+
{
|
| 7578 |
+
"epoch": 0.16557566422795533,
|
| 7579 |
+
"grad_norm": 12.459403038024902,
|
| 7580 |
+
"learning_rate": 0.0001985782933023434,
|
| 7581 |
+
"loss": 3.5718,
|
| 7582 |
+
"step": 1075
|
| 7583 |
+
},
|
| 7584 |
+
{
|
| 7585 |
+
"epoch": 0.16572968810165575,
|
| 7586 |
+
"grad_norm": 14.004292488098145,
|
| 7587 |
+
"learning_rate": 0.0001985755775119317,
|
| 7588 |
+
"loss": 4.3172,
|
| 7589 |
+
"step": 1076
|
| 7590 |
+
},
|
| 7591 |
+
{
|
| 7592 |
+
"epoch": 0.16588371197535617,
|
| 7593 |
+
"grad_norm": 9.558106422424316,
|
| 7594 |
+
"learning_rate": 0.0001985728591487053,
|
| 7595 |
+
"loss": 3.3111,
|
| 7596 |
+
"step": 1077
|
| 7597 |
+
},
|
| 7598 |
+
{
|
| 7599 |
+
"epoch": 0.1660377358490566,
|
| 7600 |
+
"grad_norm": 12.947151184082031,
|
| 7601 |
+
"learning_rate": 0.00019857013821273508,
|
| 7602 |
+
"loss": 3.8423,
|
| 7603 |
+
"step": 1078
|
| 7604 |
+
},
|
| 7605 |
+
{
|
| 7606 |
+
"epoch": 0.16619175972275702,
|
| 7607 |
+
"grad_norm": 10.47056770324707,
|
| 7608 |
+
"learning_rate": 0.00019856741470409214,
|
| 7609 |
+
"loss": 3.513,
|
| 7610 |
+
"step": 1079
|
| 7611 |
+
},
|
| 7612 |
+
{
|
| 7613 |
+
"epoch": 0.16634578359645746,
|
| 7614 |
+
"grad_norm": 12.660852432250977,
|
| 7615 |
+
"learning_rate": 0.00019856468862284752,
|
| 7616 |
+
"loss": 3.8525,
|
| 7617 |
+
"step": 1080
|
| 7618 |
+
},
|
| 7619 |
+
{
|
| 7620 |
+
"epoch": 0.16649980747015788,
|
| 7621 |
+
"grad_norm": 13.666932106018066,
|
| 7622 |
+
"learning_rate": 0.00019856195996907237,
|
| 7623 |
+
"loss": 4.3106,
|
| 7624 |
+
"step": 1081
|
| 7625 |
+
},
|
| 7626 |
+
{
|
| 7627 |
+
"epoch": 0.1666538313438583,
|
| 7628 |
+
"grad_norm": 14.849276542663574,
|
| 7629 |
+
"learning_rate": 0.00019855922874283794,
|
| 7630 |
+
"loss": 3.5716,
|
| 7631 |
+
"step": 1082
|
| 7632 |
+
},
|
| 7633 |
+
{
|
| 7634 |
+
"epoch": 0.16680785521755873,
|
| 7635 |
+
"grad_norm": 10.565242767333984,
|
| 7636 |
+
"learning_rate": 0.00019855649494421548,
|
| 7637 |
+
"loss": 3.5383,
|
| 7638 |
+
"step": 1083
|
| 7639 |
+
},
|
| 7640 |
+
{
|
| 7641 |
+
"epoch": 0.16696187909125915,
|
| 7642 |
+
"grad_norm": 12.083426475524902,
|
| 7643 |
+
"learning_rate": 0.00019855375857327637,
|
| 7644 |
+
"loss": 4.0519,
|
| 7645 |
+
"step": 1084
|
| 7646 |
+
},
|
| 7647 |
+
{
|
| 7648 |
+
"epoch": 0.16711590296495957,
|
| 7649 |
+
"grad_norm": 8.973551750183105,
|
| 7650 |
+
"learning_rate": 0.00019855101963009198,
|
| 7651 |
+
"loss": 3.2653,
|
| 7652 |
+
"step": 1085
|
| 7653 |
+
},
|
| 7654 |
+
{
|
| 7655 |
+
"epoch": 0.16726992683866,
|
| 7656 |
+
"grad_norm": 12.055974960327148,
|
| 7657 |
+
"learning_rate": 0.00019854827811473382,
|
| 7658 |
+
"loss": 4.0457,
|
| 7659 |
+
"step": 1086
|
| 7660 |
+
},
|
| 7661 |
+
{
|
| 7662 |
+
"epoch": 0.1674239507123604,
|
| 7663 |
+
"grad_norm": 10.313488960266113,
|
| 7664 |
+
"learning_rate": 0.0001985455340272735,
|
| 7665 |
+
"loss": 3.2663,
|
| 7666 |
+
"step": 1087
|
| 7667 |
+
},
|
| 7668 |
+
{
|
| 7669 |
+
"epoch": 0.16757797458606083,
|
| 7670 |
+
"grad_norm": 11.116025924682617,
|
| 7671 |
+
"learning_rate": 0.00019854278736778258,
|
| 7672 |
+
"loss": 3.4076,
|
| 7673 |
+
"step": 1088
|
| 7674 |
+
},
|
| 7675 |
+
{
|
| 7676 |
+
"epoch": 0.16773199845976125,
|
| 7677 |
+
"grad_norm": 13.926247596740723,
|
| 7678 |
+
"learning_rate": 0.00019854003813633275,
|
| 7679 |
+
"loss": 3.4324,
|
| 7680 |
+
"step": 1089
|
| 7681 |
+
},
|
| 7682 |
+
{
|
| 7683 |
+
"epoch": 0.1678860223334617,
|
| 7684 |
+
"grad_norm": 11.820225715637207,
|
| 7685 |
+
"learning_rate": 0.00019853728633299577,
|
| 7686 |
+
"loss": 3.3995,
|
| 7687 |
+
"step": 1090
|
| 7688 |
+
},
|
| 7689 |
+
{
|
| 7690 |
+
"epoch": 0.16804004620716212,
|
| 7691 |
+
"grad_norm": 10.893301010131836,
|
| 7692 |
+
"learning_rate": 0.00019853453195784346,
|
| 7693 |
+
"loss": 2.9227,
|
| 7694 |
+
"step": 1091
|
| 7695 |
+
},
|
| 7696 |
+
{
|
| 7697 |
+
"epoch": 0.16819407008086254,
|
| 7698 |
+
"grad_norm": 20.633623123168945,
|
| 7699 |
+
"learning_rate": 0.00019853177501094775,
|
| 7700 |
+
"loss": 3.9035,
|
| 7701 |
+
"step": 1092
|
| 7702 |
+
},
|
| 7703 |
+
{
|
| 7704 |
+
"epoch": 0.16834809395456296,
|
| 7705 |
+
"grad_norm": 13.161808013916016,
|
| 7706 |
+
"learning_rate": 0.00019852901549238054,
|
| 7707 |
+
"loss": 4.2875,
|
| 7708 |
+
"step": 1093
|
| 7709 |
+
},
|
| 7710 |
+
{
|
| 7711 |
+
"epoch": 0.16850211782826338,
|
| 7712 |
+
"grad_norm": 13.590312004089355,
|
| 7713 |
+
"learning_rate": 0.00019852625340221387,
|
| 7714 |
+
"loss": 3.5538,
|
| 7715 |
+
"step": 1094
|
| 7716 |
+
},
|
| 7717 |
+
{
|
| 7718 |
+
"epoch": 0.1686561417019638,
|
| 7719 |
+
"grad_norm": 14.807574272155762,
|
| 7720 |
+
"learning_rate": 0.00019852348874051985,
|
| 7721 |
+
"loss": 4.1179,
|
| 7722 |
+
"step": 1095
|
| 7723 |
+
},
|
| 7724 |
+
{
|
| 7725 |
+
"epoch": 0.16881016557566422,
|
| 7726 |
+
"grad_norm": 11.80431079864502,
|
| 7727 |
+
"learning_rate": 0.00019852072150737063,
|
| 7728 |
+
"loss": 4.2079,
|
| 7729 |
+
"step": 1096
|
| 7730 |
+
},
|
| 7731 |
+
{
|
| 7732 |
+
"epoch": 0.16896418944936464,
|
| 7733 |
+
"grad_norm": 13.151602745056152,
|
| 7734 |
+
"learning_rate": 0.00019851795170283841,
|
| 7735 |
+
"loss": 3.6325,
|
| 7736 |
+
"step": 1097
|
| 7737 |
+
},
|
| 7738 |
+
{
|
| 7739 |
+
"epoch": 0.16911821332306506,
|
| 7740 |
+
"grad_norm": 12.30171012878418,
|
| 7741 |
+
"learning_rate": 0.0001985151793269955,
|
| 7742 |
+
"loss": 3.3386,
|
| 7743 |
+
"step": 1098
|
| 7744 |
+
},
|
| 7745 |
+
{
|
| 7746 |
+
"epoch": 0.16927223719676548,
|
| 7747 |
+
"grad_norm": 12.336485862731934,
|
| 7748 |
+
"learning_rate": 0.00019851240437991426,
|
| 7749 |
+
"loss": 3.0809,
|
| 7750 |
+
"step": 1099
|
| 7751 |
+
},
|
| 7752 |
+
{
|
| 7753 |
+
"epoch": 0.16942626107046593,
|
| 7754 |
+
"grad_norm": 11.41134262084961,
|
| 7755 |
+
"learning_rate": 0.00019850962686166713,
|
| 7756 |
+
"loss": 3.272,
|
| 7757 |
+
"step": 1100
|
| 7758 |
+
},
|
| 7759 |
+
{
|
| 7760 |
+
"epoch": 0.16958028494416635,
|
| 7761 |
+
"grad_norm": 10.431635856628418,
|
| 7762 |
+
"learning_rate": 0.0001985068467723266,
|
| 7763 |
+
"loss": 3.7402,
|
| 7764 |
+
"step": 1101
|
| 7765 |
+
},
|
| 7766 |
+
{
|
| 7767 |
+
"epoch": 0.16973430881786677,
|
| 7768 |
+
"grad_norm": 13.894745826721191,
|
| 7769 |
+
"learning_rate": 0.00019850406411196518,
|
| 7770 |
+
"loss": 4.3794,
|
| 7771 |
+
"step": 1102
|
| 7772 |
+
},
|
| 7773 |
+
{
|
| 7774 |
+
"epoch": 0.1698883326915672,
|
| 7775 |
+
"grad_norm": 12.396195411682129,
|
| 7776 |
+
"learning_rate": 0.00019850127888065554,
|
| 7777 |
+
"loss": 4.2671,
|
| 7778 |
+
"step": 1103
|
| 7779 |
+
},
|
| 7780 |
+
{
|
| 7781 |
+
"epoch": 0.17004235656526762,
|
| 7782 |
+
"grad_norm": 11.540474891662598,
|
| 7783 |
+
"learning_rate": 0.0001984984910784704,
|
| 7784 |
+
"loss": 3.9258,
|
| 7785 |
+
"step": 1104
|
| 7786 |
+
},
|
| 7787 |
+
{
|
| 7788 |
+
"epoch": 0.17019638043896804,
|
| 7789 |
+
"grad_norm": 9.679957389831543,
|
| 7790 |
+
"learning_rate": 0.00019849570070548244,
|
| 7791 |
+
"loss": 3.8347,
|
| 7792 |
+
"step": 1105
|
| 7793 |
+
},
|
| 7794 |
+
{
|
| 7795 |
+
"epoch": 0.17035040431266846,
|
| 7796 |
+
"grad_norm": 13.799325942993164,
|
| 7797 |
+
"learning_rate": 0.00019849290776176458,
|
| 7798 |
+
"loss": 5.0299,
|
| 7799 |
+
"step": 1106
|
| 7800 |
+
},
|
| 7801 |
+
{
|
| 7802 |
+
"epoch": 0.17050442818636888,
|
| 7803 |
+
"grad_norm": 11.714168548583984,
|
| 7804 |
+
"learning_rate": 0.00019849011224738967,
|
| 7805 |
+
"loss": 3.7388,
|
| 7806 |
+
"step": 1107
|
| 7807 |
+
},
|
| 7808 |
+
{
|
| 7809 |
+
"epoch": 0.1706584520600693,
|
| 7810 |
+
"grad_norm": 13.339337348937988,
|
| 7811 |
+
"learning_rate": 0.00019848731416243067,
|
| 7812 |
+
"loss": 3.9534,
|
| 7813 |
+
"step": 1108
|
| 7814 |
+
},
|
| 7815 |
+
{
|
| 7816 |
+
"epoch": 0.17081247593376975,
|
| 7817 |
+
"grad_norm": 15.734074592590332,
|
| 7818 |
+
"learning_rate": 0.00019848451350696063,
|
| 7819 |
+
"loss": 4.3108,
|
| 7820 |
+
"step": 1109
|
| 7821 |
+
},
|
| 7822 |
+
{
|
| 7823 |
+
"epoch": 0.17096649980747017,
|
| 7824 |
+
"grad_norm": 12.882152557373047,
|
| 7825 |
+
"learning_rate": 0.00019848171028105261,
|
| 7826 |
+
"loss": 3.7053,
|
| 7827 |
+
"step": 1110
|
| 7828 |
+
},
|
| 7829 |
+
{
|
| 7830 |
+
"epoch": 0.1711205236811706,
|
| 7831 |
+
"grad_norm": 11.955374717712402,
|
| 7832 |
+
"learning_rate": 0.0001984789044847798,
|
| 7833 |
+
"loss": 4.1183,
|
| 7834 |
+
"step": 1111
|
| 7835 |
+
},
|
| 7836 |
+
{
|
| 7837 |
+
"epoch": 0.171274547554871,
|
| 7838 |
+
"grad_norm": 12.456283569335938,
|
| 7839 |
+
"learning_rate": 0.00019847609611821544,
|
| 7840 |
+
"loss": 3.0288,
|
| 7841 |
+
"step": 1112
|
| 7842 |
+
},
|
| 7843 |
+
{
|
| 7844 |
+
"epoch": 0.17142857142857143,
|
| 7845 |
+
"grad_norm": 10.662981033325195,
|
| 7846 |
+
"learning_rate": 0.0001984732851814328,
|
| 7847 |
+
"loss": 4.7529,
|
| 7848 |
+
"step": 1113
|
| 7849 |
+
},
|
| 7850 |
+
{
|
| 7851 |
+
"epoch": 0.17158259530227185,
|
| 7852 |
+
"grad_norm": 12.153194427490234,
|
| 7853 |
+
"learning_rate": 0.0001984704716745053,
|
| 7854 |
+
"loss": 3.9459,
|
| 7855 |
+
"step": 1114
|
| 7856 |
+
},
|
| 7857 |
+
{
|
| 7858 |
+
"epoch": 0.17173661917597227,
|
| 7859 |
+
"grad_norm": 16.472930908203125,
|
| 7860 |
+
"learning_rate": 0.00019846765559750632,
|
| 7861 |
+
"loss": 3.1947,
|
| 7862 |
+
"step": 1115
|
| 7863 |
+
},
|
| 7864 |
+
{
|
| 7865 |
+
"epoch": 0.1718906430496727,
|
| 7866 |
+
"grad_norm": 11.6087646484375,
|
| 7867 |
+
"learning_rate": 0.00019846483695050936,
|
| 7868 |
+
"loss": 3.9445,
|
| 7869 |
+
"step": 1116
|
| 7870 |
+
},
|
| 7871 |
+
{
|
| 7872 |
+
"epoch": 0.1720446669233731,
|
| 7873 |
+
"grad_norm": 11.422113418579102,
|
| 7874 |
+
"learning_rate": 0.00019846201573358801,
|
| 7875 |
+
"loss": 3.2062,
|
| 7876 |
+
"step": 1117
|
| 7877 |
+
},
|
| 7878 |
+
{
|
| 7879 |
+
"epoch": 0.17219869079707353,
|
| 7880 |
+
"grad_norm": 16.624980926513672,
|
| 7881 |
+
"learning_rate": 0.0001984591919468159,
|
| 7882 |
+
"loss": 3.8645,
|
| 7883 |
+
"step": 1118
|
| 7884 |
+
},
|
| 7885 |
+
{
|
| 7886 |
+
"epoch": 0.17235271467077398,
|
| 7887 |
+
"grad_norm": 11.967522621154785,
|
| 7888 |
+
"learning_rate": 0.0001984563655902667,
|
| 7889 |
+
"loss": 3.8582,
|
| 7890 |
+
"step": 1119
|
| 7891 |
+
},
|
| 7892 |
+
{
|
| 7893 |
+
"epoch": 0.1725067385444744,
|
| 7894 |
+
"grad_norm": 13.318442344665527,
|
| 7895 |
+
"learning_rate": 0.00019845353666401422,
|
| 7896 |
+
"loss": 4.3134,
|
| 7897 |
+
"step": 1120
|
| 7898 |
+
},
|
| 7899 |
+
{
|
| 7900 |
+
"epoch": 0.17266076241817482,
|
| 7901 |
+
"grad_norm": 8.190085411071777,
|
| 7902 |
+
"learning_rate": 0.00019845070516813227,
|
| 7903 |
+
"loss": 3.484,
|
| 7904 |
+
"step": 1121
|
| 7905 |
+
},
|
| 7906 |
+
{
|
| 7907 |
+
"epoch": 0.17281478629187524,
|
| 7908 |
+
"grad_norm": 17.924562454223633,
|
| 7909 |
+
"learning_rate": 0.00019844787110269478,
|
| 7910 |
+
"loss": 3.7665,
|
| 7911 |
+
"step": 1122
|
| 7912 |
+
},
|
| 7913 |
+
{
|
| 7914 |
+
"epoch": 0.17296881016557566,
|
| 7915 |
+
"grad_norm": 11.064155578613281,
|
| 7916 |
+
"learning_rate": 0.0001984450344677757,
|
| 7917 |
+
"loss": 3.0238,
|
| 7918 |
+
"step": 1123
|
| 7919 |
+
},
|
| 7920 |
+
{
|
| 7921 |
+
"epoch": 0.17312283403927609,
|
| 7922 |
+
"grad_norm": 16.522754669189453,
|
| 7923 |
+
"learning_rate": 0.00019844219526344903,
|
| 7924 |
+
"loss": 3.9628,
|
| 7925 |
+
"step": 1124
|
| 7926 |
+
},
|
| 7927 |
+
{
|
| 7928 |
+
"epoch": 0.1732768579129765,
|
| 7929 |
+
"grad_norm": 10.773133277893066,
|
| 7930 |
+
"learning_rate": 0.00019843935348978892,
|
| 7931 |
+
"loss": 4.2028,
|
| 7932 |
+
"step": 1125
|
| 7933 |
+
},
|
| 7934 |
+
{
|
| 7935 |
+
"epoch": 0.17343088178667693,
|
| 7936 |
+
"grad_norm": 12.943816184997559,
|
| 7937 |
+
"learning_rate": 0.00019843650914686954,
|
| 7938 |
+
"loss": 4.1469,
|
| 7939 |
+
"step": 1126
|
| 7940 |
+
},
|
| 7941 |
+
{
|
| 7942 |
+
"epoch": 0.17358490566037735,
|
| 7943 |
+
"grad_norm": 9.31995677947998,
|
| 7944 |
+
"learning_rate": 0.0001984336622347651,
|
| 7945 |
+
"loss": 2.9169,
|
| 7946 |
+
"step": 1127
|
| 7947 |
+
},
|
| 7948 |
+
{
|
| 7949 |
+
"epoch": 0.17373892953407777,
|
| 7950 |
+
"grad_norm": 12.664985656738281,
|
| 7951 |
+
"learning_rate": 0.0001984308127535499,
|
| 7952 |
+
"loss": 3.5709,
|
| 7953 |
+
"step": 1128
|
| 7954 |
+
},
|
| 7955 |
+
{
|
| 7956 |
+
"epoch": 0.17389295340777822,
|
| 7957 |
+
"grad_norm": 11.307038307189941,
|
| 7958 |
+
"learning_rate": 0.00019842796070329837,
|
| 7959 |
+
"loss": 5.146,
|
| 7960 |
+
"step": 1129
|
| 7961 |
+
},
|
| 7962 |
+
{
|
| 7963 |
+
"epoch": 0.17404697728147864,
|
| 7964 |
+
"grad_norm": 8.186049461364746,
|
| 7965 |
+
"learning_rate": 0.0001984251060840849,
|
| 7966 |
+
"loss": 3.5036,
|
| 7967 |
+
"step": 1130
|
| 7968 |
+
},
|
| 7969 |
+
{
|
| 7970 |
+
"epoch": 0.17420100115517906,
|
| 7971 |
+
"grad_norm": 11.325016975402832,
|
| 7972 |
+
"learning_rate": 0.00019842224889598397,
|
| 7973 |
+
"loss": 3.5391,
|
| 7974 |
+
"step": 1131
|
| 7975 |
+
},
|
| 7976 |
+
{
|
| 7977 |
+
"epoch": 0.17435502502887948,
|
| 7978 |
+
"grad_norm": 13.908296585083008,
|
| 7979 |
+
"learning_rate": 0.00019841938913907022,
|
| 7980 |
+
"loss": 3.5762,
|
| 7981 |
+
"step": 1132
|
| 7982 |
+
},
|
| 7983 |
+
{
|
| 7984 |
+
"epoch": 0.1745090489025799,
|
| 7985 |
+
"grad_norm": 9.789925575256348,
|
| 7986 |
+
"learning_rate": 0.00019841652681341825,
|
| 7987 |
+
"loss": 3.9647,
|
| 7988 |
+
"step": 1133
|
| 7989 |
+
},
|
| 7990 |
+
{
|
| 7991 |
+
"epoch": 0.17466307277628032,
|
| 7992 |
+
"grad_norm": 15.169290542602539,
|
| 7993 |
+
"learning_rate": 0.00019841366191910277,
|
| 7994 |
+
"loss": 4.6646,
|
| 7995 |
+
"step": 1134
|
| 7996 |
+
},
|
| 7997 |
+
{
|
| 7998 |
+
"epoch": 0.17481709664998074,
|
| 7999 |
+
"grad_norm": 11.169201850891113,
|
| 8000 |
+
"learning_rate": 0.00019841079445619855,
|
| 8001 |
+
"loss": 3.1769,
|
| 8002 |
+
"step": 1135
|
| 8003 |
+
},
|
| 8004 |
+
{
|
| 8005 |
+
"epoch": 0.17497112052368116,
|
| 8006 |
+
"grad_norm": 10.863900184631348,
|
| 8007 |
+
"learning_rate": 0.00019840792442478043,
|
| 8008 |
+
"loss": 3.8673,
|
| 8009 |
+
"step": 1136
|
| 8010 |
+
},
|
| 8011 |
+
{
|
| 8012 |
+
"epoch": 0.17512514439738158,
|
| 8013 |
+
"grad_norm": 11.557327270507812,
|
| 8014 |
+
"learning_rate": 0.0001984050518249233,
|
| 8015 |
+
"loss": 4.0111,
|
| 8016 |
+
"step": 1137
|
| 8017 |
+
},
|
| 8018 |
+
{
|
| 8019 |
+
"epoch": 0.17527916827108203,
|
| 8020 |
+
"grad_norm": 10.25669002532959,
|
| 8021 |
+
"learning_rate": 0.0001984021766567022,
|
| 8022 |
+
"loss": 4.0542,
|
| 8023 |
+
"step": 1138
|
| 8024 |
+
},
|
| 8025 |
+
{
|
| 8026 |
+
"epoch": 0.17543319214478245,
|
| 8027 |
+
"grad_norm": 12.508610725402832,
|
| 8028 |
+
"learning_rate": 0.0001983992989201921,
|
| 8029 |
+
"loss": 3.0514,
|
| 8030 |
+
"step": 1139
|
| 8031 |
+
},
|
| 8032 |
+
{
|
| 8033 |
+
"epoch": 0.17558721601848287,
|
| 8034 |
+
"grad_norm": 13.459620475769043,
|
| 8035 |
+
"learning_rate": 0.00019839641861546815,
|
| 8036 |
+
"loss": 4.1518,
|
| 8037 |
+
"step": 1140
|
| 8038 |
+
},
|
| 8039 |
+
{
|
| 8040 |
+
"epoch": 0.1757412398921833,
|
| 8041 |
+
"grad_norm": 15.084485054016113,
|
| 8042 |
+
"learning_rate": 0.00019839353574260552,
|
| 8043 |
+
"loss": 4.3405,
|
| 8044 |
+
"step": 1141
|
| 8045 |
+
},
|
| 8046 |
+
{
|
| 8047 |
+
"epoch": 0.1758952637658837,
|
| 8048 |
+
"grad_norm": 11.985688209533691,
|
| 8049 |
+
"learning_rate": 0.00019839065030167945,
|
| 8050 |
+
"loss": 4.4696,
|
| 8051 |
+
"step": 1142
|
| 8052 |
+
},
|
| 8053 |
+
{
|
| 8054 |
+
"epoch": 0.17604928763958413,
|
| 8055 |
+
"grad_norm": 13.175622940063477,
|
| 8056 |
+
"learning_rate": 0.00019838776229276524,
|
| 8057 |
+
"loss": 3.7165,
|
| 8058 |
+
"step": 1143
|
| 8059 |
+
},
|
| 8060 |
+
{
|
| 8061 |
+
"epoch": 0.17620331151328456,
|
| 8062 |
+
"grad_norm": 12.478694915771484,
|
| 8063 |
+
"learning_rate": 0.00019838487171593827,
|
| 8064 |
+
"loss": 3.9842,
|
| 8065 |
+
"step": 1144
|
| 8066 |
+
},
|
| 8067 |
+
{
|
| 8068 |
+
"epoch": 0.17635733538698498,
|
| 8069 |
+
"grad_norm": 11.12195110321045,
|
| 8070 |
+
"learning_rate": 0.00019838197857127398,
|
| 8071 |
+
"loss": 3.8946,
|
| 8072 |
+
"step": 1145
|
| 8073 |
+
},
|
| 8074 |
+
{
|
| 8075 |
+
"epoch": 0.1765113592606854,
|
| 8076 |
+
"grad_norm": 11.980460166931152,
|
| 8077 |
+
"learning_rate": 0.0001983790828588479,
|
| 8078 |
+
"loss": 4.5101,
|
| 8079 |
+
"step": 1146
|
| 8080 |
+
},
|
| 8081 |
+
{
|
| 8082 |
+
"epoch": 0.17666538313438582,
|
| 8083 |
+
"grad_norm": 11.580781936645508,
|
| 8084 |
+
"learning_rate": 0.0001983761845787356,
|
| 8085 |
+
"loss": 3.2412,
|
| 8086 |
+
"step": 1147
|
| 8087 |
+
},
|
| 8088 |
+
{
|
| 8089 |
+
"epoch": 0.17681940700808627,
|
| 8090 |
+
"grad_norm": 11.876729965209961,
|
| 8091 |
+
"learning_rate": 0.0001983732837310127,
|
| 8092 |
+
"loss": 4.362,
|
| 8093 |
+
"step": 1148
|
| 8094 |
+
},
|
| 8095 |
+
{
|
| 8096 |
+
"epoch": 0.1769734308817867,
|
| 8097 |
+
"grad_norm": 9.922520637512207,
|
| 8098 |
+
"learning_rate": 0.00019837038031575495,
|
| 8099 |
+
"loss": 3.7501,
|
| 8100 |
+
"step": 1149
|
| 8101 |
+
},
|
| 8102 |
+
{
|
| 8103 |
+
"epoch": 0.1771274547554871,
|
| 8104 |
+
"grad_norm": 10.539559364318848,
|
| 8105 |
+
"learning_rate": 0.00019836747433303812,
|
| 8106 |
+
"loss": 2.8852,
|
| 8107 |
+
"step": 1150
|
| 8108 |
+
},
|
| 8109 |
+
{
|
| 8110 |
+
"epoch": 0.17728147862918753,
|
| 8111 |
+
"grad_norm": 12.098252296447754,
|
| 8112 |
+
"learning_rate": 0.000198364565782938,
|
| 8113 |
+
"loss": 3.8561,
|
| 8114 |
+
"step": 1151
|
| 8115 |
+
},
|
| 8116 |
+
{
|
| 8117 |
+
"epoch": 0.17743550250288795,
|
| 8118 |
+
"grad_norm": 17.50993537902832,
|
| 8119 |
+
"learning_rate": 0.0001983616546655306,
|
| 8120 |
+
"loss": 3.773,
|
| 8121 |
+
"step": 1152
|
| 8122 |
+
},
|
| 8123 |
+
{
|
| 8124 |
+
"epoch": 0.17758952637658837,
|
| 8125 |
+
"grad_norm": 10.848986625671387,
|
| 8126 |
+
"learning_rate": 0.0001983587409808918,
|
| 8127 |
+
"loss": 4.1584,
|
| 8128 |
+
"step": 1153
|
| 8129 |
+
},
|
| 8130 |
+
{
|
| 8131 |
+
"epoch": 0.1777435502502888,
|
| 8132 |
+
"grad_norm": 13.124414443969727,
|
| 8133 |
+
"learning_rate": 0.00019835582472909775,
|
| 8134 |
+
"loss": 4.1641,
|
| 8135 |
+
"step": 1154
|
| 8136 |
+
},
|
| 8137 |
+
{
|
| 8138 |
+
"epoch": 0.1778975741239892,
|
| 8139 |
+
"grad_norm": 13.227132797241211,
|
| 8140 |
+
"learning_rate": 0.0001983529059102245,
|
| 8141 |
+
"loss": 3.6555,
|
| 8142 |
+
"step": 1155
|
| 8143 |
+
},
|
| 8144 |
+
{
|
| 8145 |
+
"epoch": 0.17805159799768963,
|
| 8146 |
+
"grad_norm": 11.310081481933594,
|
| 8147 |
+
"learning_rate": 0.0001983499845243482,
|
| 8148 |
+
"loss": 3.4316,
|
| 8149 |
+
"step": 1156
|
| 8150 |
+
},
|
| 8151 |
+
{
|
| 8152 |
+
"epoch": 0.17820562187139005,
|
| 8153 |
+
"grad_norm": 9.132098197937012,
|
| 8154 |
+
"learning_rate": 0.00019834706057154518,
|
| 8155 |
+
"loss": 3.1699,
|
| 8156 |
+
"step": 1157
|
| 8157 |
+
},
|
| 8158 |
+
{
|
| 8159 |
+
"epoch": 0.1783596457450905,
|
| 8160 |
+
"grad_norm": 12.124452590942383,
|
| 8161 |
+
"learning_rate": 0.00019834413405189172,
|
| 8162 |
+
"loss": 4.3226,
|
| 8163 |
+
"step": 1158
|
| 8164 |
+
},
|
| 8165 |
+
{
|
| 8166 |
+
"epoch": 0.17851366961879092,
|
| 8167 |
+
"grad_norm": 14.201220512390137,
|
| 8168 |
+
"learning_rate": 0.0001983412049654642,
|
| 8169 |
+
"loss": 4.4813,
|
| 8170 |
+
"step": 1159
|
| 8171 |
+
},
|
| 8172 |
+
{
|
| 8173 |
+
"epoch": 0.17866769349249134,
|
| 8174 |
+
"grad_norm": 13.82433032989502,
|
| 8175 |
+
"learning_rate": 0.00019833827331233904,
|
| 8176 |
+
"loss": 3.5625,
|
| 8177 |
+
"step": 1160
|
| 8178 |
+
},
|
| 8179 |
+
{
|
| 8180 |
+
"epoch": 0.17882171736619176,
|
| 8181 |
+
"grad_norm": 10.633478164672852,
|
| 8182 |
+
"learning_rate": 0.0001983353390925928,
|
| 8183 |
+
"loss": 3.9063,
|
| 8184 |
+
"step": 1161
|
| 8185 |
+
},
|
| 8186 |
+
{
|
| 8187 |
+
"epoch": 0.17897574123989218,
|
| 8188 |
+
"grad_norm": 12.822907447814941,
|
| 8189 |
+
"learning_rate": 0.000198332402306302,
|
| 8190 |
+
"loss": 3.7799,
|
| 8191 |
+
"step": 1162
|
| 8192 |
+
},
|
| 8193 |
+
{
|
| 8194 |
+
"epoch": 0.1791297651135926,
|
| 8195 |
+
"grad_norm": 13.253988265991211,
|
| 8196 |
+
"learning_rate": 0.00019832946295354338,
|
| 8197 |
+
"loss": 4.2672,
|
| 8198 |
+
"step": 1163
|
| 8199 |
+
},
|
| 8200 |
+
{
|
| 8201 |
+
"epoch": 0.17928378898729302,
|
| 8202 |
+
"grad_norm": 10.326937675476074,
|
| 8203 |
+
"learning_rate": 0.0001983265210343936,
|
| 8204 |
+
"loss": 3.556,
|
| 8205 |
+
"step": 1164
|
| 8206 |
+
},
|
| 8207 |
+
{
|
| 8208 |
+
"epoch": 0.17943781286099345,
|
| 8209 |
+
"grad_norm": 14.797069549560547,
|
| 8210 |
+
"learning_rate": 0.00019832357654892943,
|
| 8211 |
+
"loss": 3.9052,
|
| 8212 |
+
"step": 1165
|
| 8213 |
+
},
|
| 8214 |
+
{
|
| 8215 |
+
"epoch": 0.17959183673469387,
|
| 8216 |
+
"grad_norm": 9.644969940185547,
|
| 8217 |
+
"learning_rate": 0.0001983206294972278,
|
| 8218 |
+
"loss": 4.4052,
|
| 8219 |
+
"step": 1166
|
| 8220 |
+
},
|
| 8221 |
+
{
|
| 8222 |
+
"epoch": 0.17974586060839431,
|
| 8223 |
+
"grad_norm": 11.694543838500977,
|
| 8224 |
+
"learning_rate": 0.00019831767987936553,
|
| 8225 |
+
"loss": 4.3418,
|
| 8226 |
+
"step": 1167
|
| 8227 |
+
},
|
| 8228 |
+
{
|
| 8229 |
+
"epoch": 0.17989988448209474,
|
| 8230 |
+
"grad_norm": 12.655614852905273,
|
| 8231 |
+
"learning_rate": 0.00019831472769541965,
|
| 8232 |
+
"loss": 4.1458,
|
| 8233 |
+
"step": 1168
|
| 8234 |
+
},
|
| 8235 |
+
{
|
| 8236 |
+
"epoch": 0.18005390835579516,
|
| 8237 |
+
"grad_norm": 12.079744338989258,
|
| 8238 |
+
"learning_rate": 0.00019831177294546724,
|
| 8239 |
+
"loss": 4.4483,
|
| 8240 |
+
"step": 1169
|
| 8241 |
+
},
|
| 8242 |
+
{
|
| 8243 |
+
"epoch": 0.18020793222949558,
|
| 8244 |
+
"grad_norm": 12.279813766479492,
|
| 8245 |
+
"learning_rate": 0.00019830881562958537,
|
| 8246 |
+
"loss": 3.5655,
|
| 8247 |
+
"step": 1170
|
| 8248 |
+
},
|
| 8249 |
+
{
|
| 8250 |
+
"epoch": 0.180361956103196,
|
| 8251 |
+
"grad_norm": 13.201979637145996,
|
| 8252 |
+
"learning_rate": 0.00019830585574785124,
|
| 8253 |
+
"loss": 3.9816,
|
| 8254 |
+
"step": 1171
|
| 8255 |
+
},
|
| 8256 |
+
{
|
| 8257 |
+
"epoch": 0.18051597997689642,
|
| 8258 |
+
"grad_norm": 10.753480911254883,
|
| 8259 |
+
"learning_rate": 0.00019830289330034212,
|
| 8260 |
+
"loss": 3.5746,
|
| 8261 |
+
"step": 1172
|
| 8262 |
+
},
|
| 8263 |
+
{
|
| 8264 |
+
"epoch": 0.18067000385059684,
|
| 8265 |
+
"grad_norm": 10.541348457336426,
|
| 8266 |
+
"learning_rate": 0.00019829992828713532,
|
| 8267 |
+
"loss": 3.9633,
|
| 8268 |
+
"step": 1173
|
| 8269 |
+
},
|
| 8270 |
+
{
|
| 8271 |
+
"epoch": 0.18082402772429726,
|
| 8272 |
+
"grad_norm": 10.566235542297363,
|
| 8273 |
+
"learning_rate": 0.0001982969607083082,
|
| 8274 |
+
"loss": 3.4668,
|
| 8275 |
+
"step": 1174
|
| 8276 |
+
},
|
| 8277 |
+
{
|
| 8278 |
+
"epoch": 0.18097805159799768,
|
| 8279 |
+
"grad_norm": 12.87186050415039,
|
| 8280 |
+
"learning_rate": 0.00019829399056393821,
|
| 8281 |
+
"loss": 4.3093,
|
| 8282 |
+
"step": 1175
|
| 8283 |
+
},
|
| 8284 |
+
{
|
| 8285 |
+
"epoch": 0.1811320754716981,
|
| 8286 |
+
"grad_norm": 12.723098754882812,
|
| 8287 |
+
"learning_rate": 0.00019829101785410295,
|
| 8288 |
+
"loss": 4.4819,
|
| 8289 |
+
"step": 1176
|
| 8290 |
+
},
|
| 8291 |
+
{
|
| 8292 |
+
"epoch": 0.18128609934539855,
|
| 8293 |
+
"grad_norm": 11.136910438537598,
|
| 8294 |
+
"learning_rate": 0.00019828804257887995,
|
| 8295 |
+
"loss": 4.1809,
|
| 8296 |
+
"step": 1177
|
| 8297 |
+
},
|
| 8298 |
+
{
|
| 8299 |
+
"epoch": 0.18144012321909897,
|
| 8300 |
+
"grad_norm": 14.080367088317871,
|
| 8301 |
+
"learning_rate": 0.00019828506473834687,
|
| 8302 |
+
"loss": 4.3867,
|
| 8303 |
+
"step": 1178
|
| 8304 |
+
},
|
| 8305 |
+
{
|
| 8306 |
+
"epoch": 0.1815941470927994,
|
| 8307 |
+
"grad_norm": 11.699310302734375,
|
| 8308 |
+
"learning_rate": 0.00019828208433258143,
|
| 8309 |
+
"loss": 4.1318,
|
| 8310 |
+
"step": 1179
|
| 8311 |
+
},
|
| 8312 |
+
{
|
| 8313 |
+
"epoch": 0.1817481709664998,
|
| 8314 |
+
"grad_norm": 11.105619430541992,
|
| 8315 |
+
"learning_rate": 0.00019827910136166138,
|
| 8316 |
+
"loss": 4.3103,
|
| 8317 |
+
"step": 1180
|
| 8318 |
+
},
|
| 8319 |
+
{
|
| 8320 |
+
"epoch": 0.18190219484020023,
|
| 8321 |
+
"grad_norm": 9.191035270690918,
|
| 8322 |
+
"learning_rate": 0.00019827611582566468,
|
| 8323 |
+
"loss": 3.6743,
|
| 8324 |
+
"step": 1181
|
| 8325 |
+
},
|
| 8326 |
+
{
|
| 8327 |
+
"epoch": 0.18205621871390065,
|
| 8328 |
+
"grad_norm": 11.332408905029297,
|
| 8329 |
+
"learning_rate": 0.00019827312772466914,
|
| 8330 |
+
"loss": 4.1401,
|
| 8331 |
+
"step": 1182
|
| 8332 |
+
},
|
| 8333 |
+
{
|
| 8334 |
+
"epoch": 0.18221024258760107,
|
| 8335 |
+
"grad_norm": 11.520444869995117,
|
| 8336 |
+
"learning_rate": 0.0001982701370587528,
|
| 8337 |
+
"loss": 4.4887,
|
| 8338 |
+
"step": 1183
|
| 8339 |
+
},
|
| 8340 |
+
{
|
| 8341 |
+
"epoch": 0.1823642664613015,
|
| 8342 |
+
"grad_norm": 11.725013732910156,
|
| 8343 |
+
"learning_rate": 0.00019826714382799375,
|
| 8344 |
+
"loss": 3.4896,
|
| 8345 |
+
"step": 1184
|
| 8346 |
+
},
|
| 8347 |
+
{
|
| 8348 |
+
"epoch": 0.18251829033500191,
|
| 8349 |
+
"grad_norm": 17.180004119873047,
|
| 8350 |
+
"learning_rate": 0.00019826414803247001,
|
| 8351 |
+
"loss": 3.5693,
|
| 8352 |
+
"step": 1185
|
| 8353 |
+
},
|
| 8354 |
+
{
|
| 8355 |
+
"epoch": 0.18267231420870234,
|
| 8356 |
+
"grad_norm": 11.406875610351562,
|
| 8357 |
+
"learning_rate": 0.00019826114967225992,
|
| 8358 |
+
"loss": 3.2492,
|
| 8359 |
+
"step": 1186
|
| 8360 |
+
},
|
| 8361 |
+
{
|
| 8362 |
+
"epoch": 0.18282633808240278,
|
| 8363 |
+
"grad_norm": 10.967105865478516,
|
| 8364 |
+
"learning_rate": 0.00019825814874744157,
|
| 8365 |
+
"loss": 4.1061,
|
| 8366 |
+
"step": 1187
|
| 8367 |
+
},
|
| 8368 |
+
{
|
| 8369 |
+
"epoch": 0.1829803619561032,
|
| 8370 |
+
"grad_norm": 14.942412376403809,
|
| 8371 |
+
"learning_rate": 0.00019825514525809341,
|
| 8372 |
+
"loss": 4.4003,
|
| 8373 |
+
"step": 1188
|
| 8374 |
+
},
|
| 8375 |
+
{
|
| 8376 |
+
"epoch": 0.18313438582980363,
|
| 8377 |
+
"grad_norm": 11.987462997436523,
|
| 8378 |
+
"learning_rate": 0.0001982521392042938,
|
| 8379 |
+
"loss": 4.2834,
|
| 8380 |
+
"step": 1189
|
| 8381 |
+
},
|
| 8382 |
+
{
|
| 8383 |
+
"epoch": 0.18328840970350405,
|
| 8384 |
+
"grad_norm": 12.808433532714844,
|
| 8385 |
+
"learning_rate": 0.00019824913058612116,
|
| 8386 |
+
"loss": 3.9433,
|
| 8387 |
+
"step": 1190
|
| 8388 |
+
},
|
| 8389 |
+
{
|
| 8390 |
+
"epoch": 0.18344243357720447,
|
| 8391 |
+
"grad_norm": 12.015241622924805,
|
| 8392 |
+
"learning_rate": 0.00019824611940365402,
|
| 8393 |
+
"loss": 3.7709,
|
| 8394 |
+
"step": 1191
|
| 8395 |
+
},
|
| 8396 |
+
{
|
| 8397 |
+
"epoch": 0.1835964574509049,
|
| 8398 |
+
"grad_norm": 11.753364562988281,
|
| 8399 |
+
"learning_rate": 0.000198243105656971,
|
| 8400 |
+
"loss": 4.1389,
|
| 8401 |
+
"step": 1192
|
| 8402 |
+
},
|
| 8403 |
+
{
|
| 8404 |
+
"epoch": 0.1837504813246053,
|
| 8405 |
+
"grad_norm": 9.650666236877441,
|
| 8406 |
+
"learning_rate": 0.00019824008934615076,
|
| 8407 |
+
"loss": 4.0265,
|
| 8408 |
+
"step": 1193
|
| 8409 |
+
},
|
| 8410 |
+
{
|
| 8411 |
+
"epoch": 0.18390450519830573,
|
| 8412 |
+
"grad_norm": 12.346750259399414,
|
| 8413 |
+
"learning_rate": 0.00019823707047127205,
|
| 8414 |
+
"loss": 4.0431,
|
| 8415 |
+
"step": 1194
|
| 8416 |
+
},
|
| 8417 |
+
{
|
| 8418 |
+
"epoch": 0.18405852907200615,
|
| 8419 |
+
"grad_norm": 16.240116119384766,
|
| 8420 |
+
"learning_rate": 0.00019823404903241359,
|
| 8421 |
+
"loss": 3.9108,
|
| 8422 |
+
"step": 1195
|
| 8423 |
+
},
|
| 8424 |
+
{
|
| 8425 |
+
"epoch": 0.1842125529457066,
|
| 8426 |
+
"grad_norm": 14.475204467773438,
|
| 8427 |
+
"learning_rate": 0.00019823102502965426,
|
| 8428 |
+
"loss": 3.6084,
|
| 8429 |
+
"step": 1196
|
| 8430 |
+
},
|
| 8431 |
+
{
|
| 8432 |
+
"epoch": 0.18436657681940702,
|
| 8433 |
+
"grad_norm": 12.212177276611328,
|
| 8434 |
+
"learning_rate": 0.00019822799846307306,
|
| 8435 |
+
"loss": 3.5312,
|
| 8436 |
+
"step": 1197
|
| 8437 |
+
},
|
| 8438 |
+
{
|
| 8439 |
+
"epoch": 0.18452060069310744,
|
| 8440 |
+
"grad_norm": 15.028864860534668,
|
| 8441 |
+
"learning_rate": 0.00019822496933274888,
|
| 8442 |
+
"loss": 4.6225,
|
| 8443 |
+
"step": 1198
|
| 8444 |
+
},
|
| 8445 |
+
{
|
| 8446 |
+
"epoch": 0.18467462456680786,
|
| 8447 |
+
"grad_norm": 10.804827690124512,
|
| 8448 |
+
"learning_rate": 0.00019822193763876085,
|
| 8449 |
+
"loss": 2.7079,
|
| 8450 |
+
"step": 1199
|
| 8451 |
+
},
|
| 8452 |
+
{
|
| 8453 |
+
"epoch": 0.18482864844050828,
|
| 8454 |
+
"grad_norm": 12.293673515319824,
|
| 8455 |
+
"learning_rate": 0.00019821890338118806,
|
| 8456 |
+
"loss": 3.888,
|
| 8457 |
+
"step": 1200
|
| 8458 |
+
},
|
| 8459 |
+
{
|
| 8460 |
+
"epoch": 0.18482864844050828,
|
| 8461 |
+
"eval_loss": 0.9737382531166077,
|
| 8462 |
+
"eval_runtime": 40.4706,
|
| 8463 |
+
"eval_samples_per_second": 6.449,
|
| 8464 |
+
"eval_steps_per_second": 3.237,
|
| 8465 |
+
"step": 1200
|
| 8466 |
}
|
| 8467 |
],
|
| 8468 |
"logging_steps": 1,
|
|
|
|
| 8477 |
"early_stopping_threshold": 0.0
|
| 8478 |
},
|
| 8479 |
"attributes": {
|
| 8480 |
+
"early_stopping_patience_counter": 3
|
| 8481 |
}
|
| 8482 |
},
|
| 8483 |
"TrainerControl": {
|
|
|
|
| 8486 |
"should_evaluate": false,
|
| 8487 |
"should_log": false,
|
| 8488 |
"should_save": true,
|
| 8489 |
+
"should_training_stop": true
|
| 8490 |
},
|
| 8491 |
"attributes": {}
|
| 8492 |
}
|
| 8493 |
},
|
| 8494 |
+
"total_flos": 1.745589977040814e+17,
|
| 8495 |
"train_batch_size": 2,
|
| 8496 |
"trial_name": null,
|
| 8497 |
"trial_params": null
|