Training in progress, step 1176, 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 36981072
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e82ddad86ac334ffad594328e555eeb2f2611b427d2fb920cf5ac34d9e794974
|
| 3 |
size 36981072
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 19859524
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:95b95ad303046b43abd66b860a905811dbad5ed4b33a144c4296caa2e42f9437
|
| 3 |
size 19859524
|
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 sha256:4dc7943cea8ad7d13591d68000c924b350a90128d02ac6e3163ee8e139add986
|
| 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:f75e367a8f7de08bfa82633d08c68150dc46bb7628dc56e91078e717a9e30144
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -6181,6 +6181,2072 @@
|
|
| 6181 |
"learning_rate": 1.4763424932128388e-05,
|
| 6182 |
"loss": 0.6212,
|
| 6183 |
"step": 882
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6184 |
}
|
| 6185 |
],
|
| 6186 |
"logging_steps": 1,
|
|
@@ -6195,12 +8261,12 @@
|
|
| 6195 |
"should_evaluate": false,
|
| 6196 |
"should_log": false,
|
| 6197 |
"should_save": true,
|
| 6198 |
-
"should_training_stop":
|
| 6199 |
},
|
| 6200 |
"attributes": {}
|
| 6201 |
}
|
| 6202 |
},
|
| 6203 |
-
"total_flos": 1.
|
| 6204 |
"train_batch_size": 4,
|
| 6205 |
"trial_name": null,
|
| 6206 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0004252604720392,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 1176,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 6181 |
"learning_rate": 1.4763424932128388e-05,
|
| 6182 |
"loss": 0.6212,
|
| 6183 |
"step": 882
|
| 6184 |
+
},
|
| 6185 |
+
{
|
| 6186 |
+
"epoch": 0.7510099936210929,
|
| 6187 |
+
"grad_norm": 0.32501527667045593,
|
| 6188 |
+
"learning_rate": 1.4668381972599649e-05,
|
| 6189 |
+
"loss": 0.6416,
|
| 6190 |
+
"step": 883
|
| 6191 |
+
},
|
| 6192 |
+
{
|
| 6193 |
+
"epoch": 0.7518605145651711,
|
| 6194 |
+
"grad_norm": 0.31923580169677734,
|
| 6195 |
+
"learning_rate": 1.4573593314746293e-05,
|
| 6196 |
+
"loss": 0.6278,
|
| 6197 |
+
"step": 884
|
| 6198 |
+
},
|
| 6199 |
+
{
|
| 6200 |
+
"epoch": 0.7527110355092494,
|
| 6201 |
+
"grad_norm": 0.3235868811607361,
|
| 6202 |
+
"learning_rate": 1.4479059640815962e-05,
|
| 6203 |
+
"loss": 0.6789,
|
| 6204 |
+
"step": 885
|
| 6205 |
+
},
|
| 6206 |
+
{
|
| 6207 |
+
"epoch": 0.7535615564533277,
|
| 6208 |
+
"grad_norm": 0.3378104269504547,
|
| 6209 |
+
"learning_rate": 1.4384781631221112e-05,
|
| 6210 |
+
"loss": 0.6761,
|
| 6211 |
+
"step": 886
|
| 6212 |
+
},
|
| 6213 |
+
{
|
| 6214 |
+
"epoch": 0.7544120773974059,
|
| 6215 |
+
"grad_norm": 0.3208213448524475,
|
| 6216 |
+
"learning_rate": 1.4290759964533962e-05,
|
| 6217 |
+
"loss": 0.6583,
|
| 6218 |
+
"step": 887
|
| 6219 |
+
},
|
| 6220 |
+
{
|
| 6221 |
+
"epoch": 0.7552625983414841,
|
| 6222 |
+
"grad_norm": 0.3205471336841583,
|
| 6223 |
+
"learning_rate": 1.4196995317481748e-05,
|
| 6224 |
+
"loss": 0.6265,
|
| 6225 |
+
"step": 888
|
| 6226 |
+
},
|
| 6227 |
+
{
|
| 6228 |
+
"epoch": 0.7561131192855624,
|
| 6229 |
+
"grad_norm": 0.3536736071109772,
|
| 6230 |
+
"learning_rate": 1.4103488364941776e-05,
|
| 6231 |
+
"loss": 0.6643,
|
| 6232 |
+
"step": 889
|
| 6233 |
+
},
|
| 6234 |
+
{
|
| 6235 |
+
"epoch": 0.7569636402296407,
|
| 6236 |
+
"grad_norm": 0.33717280626296997,
|
| 6237 |
+
"learning_rate": 1.4010239779936535e-05,
|
| 6238 |
+
"loss": 0.6037,
|
| 6239 |
+
"step": 890
|
| 6240 |
+
},
|
| 6241 |
+
{
|
| 6242 |
+
"epoch": 0.7578141611737189,
|
| 6243 |
+
"grad_norm": 0.3463943302631378,
|
| 6244 |
+
"learning_rate": 1.391725023362897e-05,
|
| 6245 |
+
"loss": 0.6694,
|
| 6246 |
+
"step": 891
|
| 6247 |
+
},
|
| 6248 |
+
{
|
| 6249 |
+
"epoch": 0.7586646821177971,
|
| 6250 |
+
"grad_norm": 0.3252984881401062,
|
| 6251 |
+
"learning_rate": 1.3824520395317498e-05,
|
| 6252 |
+
"loss": 0.6439,
|
| 6253 |
+
"step": 892
|
| 6254 |
+
},
|
| 6255 |
+
{
|
| 6256 |
+
"epoch": 0.7595152030618754,
|
| 6257 |
+
"grad_norm": 0.3410145044326782,
|
| 6258 |
+
"learning_rate": 1.3732050932431344e-05,
|
| 6259 |
+
"loss": 0.6065,
|
| 6260 |
+
"step": 893
|
| 6261 |
+
},
|
| 6262 |
+
{
|
| 6263 |
+
"epoch": 0.7603657240059536,
|
| 6264 |
+
"grad_norm": 0.3252216577529907,
|
| 6265 |
+
"learning_rate": 1.3639842510525603e-05,
|
| 6266 |
+
"loss": 0.6637,
|
| 6267 |
+
"step": 894
|
| 6268 |
+
},
|
| 6269 |
+
{
|
| 6270 |
+
"epoch": 0.7612162449500319,
|
| 6271 |
+
"grad_norm": 0.32330164313316345,
|
| 6272 |
+
"learning_rate": 1.3547895793276561e-05,
|
| 6273 |
+
"loss": 0.6251,
|
| 6274 |
+
"step": 895
|
| 6275 |
+
},
|
| 6276 |
+
{
|
| 6277 |
+
"epoch": 0.7620667658941102,
|
| 6278 |
+
"grad_norm": 0.3421441614627838,
|
| 6279 |
+
"learning_rate": 1.3456211442476813e-05,
|
| 6280 |
+
"loss": 0.6146,
|
| 6281 |
+
"step": 896
|
| 6282 |
+
},
|
| 6283 |
+
{
|
| 6284 |
+
"epoch": 0.7629172868381884,
|
| 6285 |
+
"grad_norm": 0.3628736734390259,
|
| 6286 |
+
"learning_rate": 1.3364790118030618e-05,
|
| 6287 |
+
"loss": 0.6169,
|
| 6288 |
+
"step": 897
|
| 6289 |
+
},
|
| 6290 |
+
{
|
| 6291 |
+
"epoch": 0.7637678077822666,
|
| 6292 |
+
"grad_norm": 0.3343038856983185,
|
| 6293 |
+
"learning_rate": 1.3273632477949017e-05,
|
| 6294 |
+
"loss": 0.6101,
|
| 6295 |
+
"step": 898
|
| 6296 |
+
},
|
| 6297 |
+
{
|
| 6298 |
+
"epoch": 0.7646183287263448,
|
| 6299 |
+
"grad_norm": 0.4127245247364044,
|
| 6300 |
+
"learning_rate": 1.3182739178345233e-05,
|
| 6301 |
+
"loss": 0.6604,
|
| 6302 |
+
"step": 899
|
| 6303 |
+
},
|
| 6304 |
+
{
|
| 6305 |
+
"epoch": 0.7654688496704232,
|
| 6306 |
+
"grad_norm": 0.6632758975028992,
|
| 6307 |
+
"learning_rate": 1.309211087342982e-05,
|
| 6308 |
+
"loss": 0.5966,
|
| 6309 |
+
"step": 900
|
| 6310 |
+
},
|
| 6311 |
+
{
|
| 6312 |
+
"epoch": 0.7663193706145014,
|
| 6313 |
+
"grad_norm": 0.24257014691829681,
|
| 6314 |
+
"learning_rate": 1.3001748215506065e-05,
|
| 6315 |
+
"loss": 0.6016,
|
| 6316 |
+
"step": 901
|
| 6317 |
+
},
|
| 6318 |
+
{
|
| 6319 |
+
"epoch": 0.7671698915585796,
|
| 6320 |
+
"grad_norm": 0.2611919939517975,
|
| 6321 |
+
"learning_rate": 1.2911651854965178e-05,
|
| 6322 |
+
"loss": 0.628,
|
| 6323 |
+
"step": 902
|
| 6324 |
+
},
|
| 6325 |
+
{
|
| 6326 |
+
"epoch": 0.7680204125026578,
|
| 6327 |
+
"grad_norm": 0.2762000858783722,
|
| 6328 |
+
"learning_rate": 1.2821822440281783e-05,
|
| 6329 |
+
"loss": 0.6044,
|
| 6330 |
+
"step": 903
|
| 6331 |
+
},
|
| 6332 |
+
{
|
| 6333 |
+
"epoch": 0.7688709334467361,
|
| 6334 |
+
"grad_norm": 0.2796517312526703,
|
| 6335 |
+
"learning_rate": 1.273226061800904e-05,
|
| 6336 |
+
"loss": 0.6305,
|
| 6337 |
+
"step": 904
|
| 6338 |
+
},
|
| 6339 |
+
{
|
| 6340 |
+
"epoch": 0.7697214543908144,
|
| 6341 |
+
"grad_norm": 0.2846098840236664,
|
| 6342 |
+
"learning_rate": 1.2642967032774128e-05,
|
| 6343 |
+
"loss": 0.6548,
|
| 6344 |
+
"step": 905
|
| 6345 |
+
},
|
| 6346 |
+
{
|
| 6347 |
+
"epoch": 0.7705719753348926,
|
| 6348 |
+
"grad_norm": 0.28865575790405273,
|
| 6349 |
+
"learning_rate": 1.2553942327273605e-05,
|
| 6350 |
+
"loss": 0.6322,
|
| 6351 |
+
"step": 906
|
| 6352 |
+
},
|
| 6353 |
+
{
|
| 6354 |
+
"epoch": 0.7714224962789709,
|
| 6355 |
+
"grad_norm": 0.28172197937965393,
|
| 6356 |
+
"learning_rate": 1.2465187142268685e-05,
|
| 6357 |
+
"loss": 0.5971,
|
| 6358 |
+
"step": 907
|
| 6359 |
+
},
|
| 6360 |
+
{
|
| 6361 |
+
"epoch": 0.7722730172230491,
|
| 6362 |
+
"grad_norm": 0.2969709634780884,
|
| 6363 |
+
"learning_rate": 1.2376702116580763e-05,
|
| 6364 |
+
"loss": 0.6288,
|
| 6365 |
+
"step": 908
|
| 6366 |
+
},
|
| 6367 |
+
{
|
| 6368 |
+
"epoch": 0.7731235381671273,
|
| 6369 |
+
"grad_norm": 0.276475191116333,
|
| 6370 |
+
"learning_rate": 1.2288487887086681e-05,
|
| 6371 |
+
"loss": 0.6318,
|
| 6372 |
+
"step": 909
|
| 6373 |
+
},
|
| 6374 |
+
{
|
| 6375 |
+
"epoch": 0.7739740591112056,
|
| 6376 |
+
"grad_norm": 0.29293322563171387,
|
| 6377 |
+
"learning_rate": 1.2200545088714254e-05,
|
| 6378 |
+
"loss": 0.6459,
|
| 6379 |
+
"step": 910
|
| 6380 |
+
},
|
| 6381 |
+
{
|
| 6382 |
+
"epoch": 0.7748245800552839,
|
| 6383 |
+
"grad_norm": 0.291227251291275,
|
| 6384 |
+
"learning_rate": 1.2112874354437614e-05,
|
| 6385 |
+
"loss": 0.6064,
|
| 6386 |
+
"step": 911
|
| 6387 |
+
},
|
| 6388 |
+
{
|
| 6389 |
+
"epoch": 0.7756751009993621,
|
| 6390 |
+
"grad_norm": 0.31140944361686707,
|
| 6391 |
+
"learning_rate": 1.2025476315272744e-05,
|
| 6392 |
+
"loss": 0.6764,
|
| 6393 |
+
"step": 912
|
| 6394 |
+
},
|
| 6395 |
+
{
|
| 6396 |
+
"epoch": 0.7765256219434403,
|
| 6397 |
+
"grad_norm": 0.29568618535995483,
|
| 6398 |
+
"learning_rate": 1.1938351600272817e-05,
|
| 6399 |
+
"loss": 0.6456,
|
| 6400 |
+
"step": 913
|
| 6401 |
+
},
|
| 6402 |
+
{
|
| 6403 |
+
"epoch": 0.7773761428875186,
|
| 6404 |
+
"grad_norm": 0.3133503496646881,
|
| 6405 |
+
"learning_rate": 1.1851500836523816e-05,
|
| 6406 |
+
"loss": 0.6271,
|
| 6407 |
+
"step": 914
|
| 6408 |
+
},
|
| 6409 |
+
{
|
| 6410 |
+
"epoch": 0.7782266638315969,
|
| 6411 |
+
"grad_norm": 0.30188944935798645,
|
| 6412 |
+
"learning_rate": 1.1764924649139858e-05,
|
| 6413 |
+
"loss": 0.6469,
|
| 6414 |
+
"step": 915
|
| 6415 |
+
},
|
| 6416 |
+
{
|
| 6417 |
+
"epoch": 0.7790771847756751,
|
| 6418 |
+
"grad_norm": 0.30510157346725464,
|
| 6419 |
+
"learning_rate": 1.1678623661258864e-05,
|
| 6420 |
+
"loss": 0.6443,
|
| 6421 |
+
"step": 916
|
| 6422 |
+
},
|
| 6423 |
+
{
|
| 6424 |
+
"epoch": 0.7799277057197533,
|
| 6425 |
+
"grad_norm": 0.301517516374588,
|
| 6426 |
+
"learning_rate": 1.1592598494037898e-05,
|
| 6427 |
+
"loss": 0.626,
|
| 6428 |
+
"step": 917
|
| 6429 |
+
},
|
| 6430 |
+
{
|
| 6431 |
+
"epoch": 0.7807782266638316,
|
| 6432 |
+
"grad_norm": 0.31465768814086914,
|
| 6433 |
+
"learning_rate": 1.1506849766648886e-05,
|
| 6434 |
+
"loss": 0.6503,
|
| 6435 |
+
"step": 918
|
| 6436 |
+
},
|
| 6437 |
+
{
|
| 6438 |
+
"epoch": 0.7816287476079098,
|
| 6439 |
+
"grad_norm": 0.3222067952156067,
|
| 6440 |
+
"learning_rate": 1.1421378096273972e-05,
|
| 6441 |
+
"loss": 0.6937,
|
| 6442 |
+
"step": 919
|
| 6443 |
+
},
|
| 6444 |
+
{
|
| 6445 |
+
"epoch": 0.7824792685519881,
|
| 6446 |
+
"grad_norm": 0.30539044737815857,
|
| 6447 |
+
"learning_rate": 1.1336184098101221e-05,
|
| 6448 |
+
"loss": 0.6604,
|
| 6449 |
+
"step": 920
|
| 6450 |
+
},
|
| 6451 |
+
{
|
| 6452 |
+
"epoch": 0.7833297894960664,
|
| 6453 |
+
"grad_norm": 0.3119628429412842,
|
| 6454 |
+
"learning_rate": 1.1251268385320091e-05,
|
| 6455 |
+
"loss": 0.6394,
|
| 6456 |
+
"step": 921
|
| 6457 |
+
},
|
| 6458 |
+
{
|
| 6459 |
+
"epoch": 0.7841803104401446,
|
| 6460 |
+
"grad_norm": 0.306523859500885,
|
| 6461 |
+
"learning_rate": 1.1166631569117115e-05,
|
| 6462 |
+
"loss": 0.637,
|
| 6463 |
+
"step": 922
|
| 6464 |
+
},
|
| 6465 |
+
{
|
| 6466 |
+
"epoch": 0.7850308313842228,
|
| 6467 |
+
"grad_norm": 0.3049357533454895,
|
| 6468 |
+
"learning_rate": 1.1082274258671376e-05,
|
| 6469 |
+
"loss": 0.6671,
|
| 6470 |
+
"step": 923
|
| 6471 |
+
},
|
| 6472 |
+
{
|
| 6473 |
+
"epoch": 0.785881352328301,
|
| 6474 |
+
"grad_norm": 0.3031104803085327,
|
| 6475 |
+
"learning_rate": 1.0998197061150277e-05,
|
| 6476 |
+
"loss": 0.6483,
|
| 6477 |
+
"step": 924
|
| 6478 |
+
},
|
| 6479 |
+
{
|
| 6480 |
+
"epoch": 0.7867318732723794,
|
| 6481 |
+
"grad_norm": 0.3212115466594696,
|
| 6482 |
+
"learning_rate": 1.0914400581705037e-05,
|
| 6483 |
+
"loss": 0.6402,
|
| 6484 |
+
"step": 925
|
| 6485 |
+
},
|
| 6486 |
+
{
|
| 6487 |
+
"epoch": 0.7875823942164576,
|
| 6488 |
+
"grad_norm": 0.31629765033721924,
|
| 6489 |
+
"learning_rate": 1.083088542346638e-05,
|
| 6490 |
+
"loss": 0.653,
|
| 6491 |
+
"step": 926
|
| 6492 |
+
},
|
| 6493 |
+
{
|
| 6494 |
+
"epoch": 0.7884329151605358,
|
| 6495 |
+
"grad_norm": 0.3214990198612213,
|
| 6496 |
+
"learning_rate": 1.0747652187540253e-05,
|
| 6497 |
+
"loss": 0.6705,
|
| 6498 |
+
"step": 927
|
| 6499 |
+
},
|
| 6500 |
+
{
|
| 6501 |
+
"epoch": 0.789283436104614,
|
| 6502 |
+
"grad_norm": 0.3067534565925598,
|
| 6503 |
+
"learning_rate": 1.0664701473003397e-05,
|
| 6504 |
+
"loss": 0.653,
|
| 6505 |
+
"step": 928
|
| 6506 |
+
},
|
| 6507 |
+
{
|
| 6508 |
+
"epoch": 0.7901339570486923,
|
| 6509 |
+
"grad_norm": 0.32290220260620117,
|
| 6510 |
+
"learning_rate": 1.0582033876899127e-05,
|
| 6511 |
+
"loss": 0.6478,
|
| 6512 |
+
"step": 929
|
| 6513 |
+
},
|
| 6514 |
+
{
|
| 6515 |
+
"epoch": 0.7909844779927706,
|
| 6516 |
+
"grad_norm": 0.3029451072216034,
|
| 6517 |
+
"learning_rate": 1.0499649994232958e-05,
|
| 6518 |
+
"loss": 0.6453,
|
| 6519 |
+
"step": 930
|
| 6520 |
+
},
|
| 6521 |
+
{
|
| 6522 |
+
"epoch": 0.7918349989368488,
|
| 6523 |
+
"grad_norm": 0.3304339051246643,
|
| 6524 |
+
"learning_rate": 1.041755041796839e-05,
|
| 6525 |
+
"loss": 0.7113,
|
| 6526 |
+
"step": 931
|
| 6527 |
+
},
|
| 6528 |
+
{
|
| 6529 |
+
"epoch": 0.7926855198809271,
|
| 6530 |
+
"grad_norm": 0.32017937302589417,
|
| 6531 |
+
"learning_rate": 1.0335735739022568e-05,
|
| 6532 |
+
"loss": 0.6078,
|
| 6533 |
+
"step": 932
|
| 6534 |
+
},
|
| 6535 |
+
{
|
| 6536 |
+
"epoch": 0.7935360408250053,
|
| 6537 |
+
"grad_norm": 0.3028521239757538,
|
| 6538 |
+
"learning_rate": 1.0254206546262141e-05,
|
| 6539 |
+
"loss": 0.6831,
|
| 6540 |
+
"step": 933
|
| 6541 |
+
},
|
| 6542 |
+
{
|
| 6543 |
+
"epoch": 0.7943865617690835,
|
| 6544 |
+
"grad_norm": 0.3357234001159668,
|
| 6545 |
+
"learning_rate": 1.0172963426498866e-05,
|
| 6546 |
+
"loss": 0.6582,
|
| 6547 |
+
"step": 934
|
| 6548 |
+
},
|
| 6549 |
+
{
|
| 6550 |
+
"epoch": 0.7952370827131618,
|
| 6551 |
+
"grad_norm": 0.3475031554698944,
|
| 6552 |
+
"learning_rate": 1.009200696448554e-05,
|
| 6553 |
+
"loss": 0.6487,
|
| 6554 |
+
"step": 935
|
| 6555 |
+
},
|
| 6556 |
+
{
|
| 6557 |
+
"epoch": 0.7960876036572401,
|
| 6558 |
+
"grad_norm": 0.3296676278114319,
|
| 6559 |
+
"learning_rate": 1.0011337742911674e-05,
|
| 6560 |
+
"loss": 0.6466,
|
| 6561 |
+
"step": 936
|
| 6562 |
+
},
|
| 6563 |
+
{
|
| 6564 |
+
"epoch": 0.7969381246013183,
|
| 6565 |
+
"grad_norm": 0.3364199101924896,
|
| 6566 |
+
"learning_rate": 9.930956342399384e-06,
|
| 6567 |
+
"loss": 0.6709,
|
| 6568 |
+
"step": 937
|
| 6569 |
+
},
|
| 6570 |
+
{
|
| 6571 |
+
"epoch": 0.7977886455453965,
|
| 6572 |
+
"grad_norm": 0.3327937424182892,
|
| 6573 |
+
"learning_rate": 9.850863341499134e-06,
|
| 6574 |
+
"loss": 0.6581,
|
| 6575 |
+
"step": 938
|
| 6576 |
+
},
|
| 6577 |
+
{
|
| 6578 |
+
"epoch": 0.7986391664894747,
|
| 6579 |
+
"grad_norm": 0.33358824253082275,
|
| 6580 |
+
"learning_rate": 9.771059316685666e-06,
|
| 6581 |
+
"loss": 0.6086,
|
| 6582 |
+
"step": 939
|
| 6583 |
+
},
|
| 6584 |
+
{
|
| 6585 |
+
"epoch": 0.7994896874335531,
|
| 6586 |
+
"grad_norm": 0.33920568227767944,
|
| 6587 |
+
"learning_rate": 9.691544842353728e-06,
|
| 6588 |
+
"loss": 0.6441,
|
| 6589 |
+
"step": 940
|
| 6590 |
+
},
|
| 6591 |
+
{
|
| 6592 |
+
"epoch": 0.8003402083776313,
|
| 6593 |
+
"grad_norm": 0.3285887837409973,
|
| 6594 |
+
"learning_rate": 9.612320490814093e-06,
|
| 6595 |
+
"loss": 0.6728,
|
| 6596 |
+
"step": 941
|
| 6597 |
+
},
|
| 6598 |
+
{
|
| 6599 |
+
"epoch": 0.8011907293217095,
|
| 6600 |
+
"grad_norm": 0.3406343162059784,
|
| 6601 |
+
"learning_rate": 9.533386832289266e-06,
|
| 6602 |
+
"loss": 0.6358,
|
| 6603 |
+
"step": 942
|
| 6604 |
+
},
|
| 6605 |
+
{
|
| 6606 |
+
"epoch": 0.8020412502657878,
|
| 6607 |
+
"grad_norm": 0.33286091685295105,
|
| 6608 |
+
"learning_rate": 9.454744434909551e-06,
|
| 6609 |
+
"loss": 0.6777,
|
| 6610 |
+
"step": 943
|
| 6611 |
+
},
|
| 6612 |
+
{
|
| 6613 |
+
"epoch": 0.802891771209866,
|
| 6614 |
+
"grad_norm": 0.3307068943977356,
|
| 6615 |
+
"learning_rate": 9.37639386470882e-06,
|
| 6616 |
+
"loss": 0.6318,
|
| 6617 |
+
"step": 944
|
| 6618 |
+
},
|
| 6619 |
+
{
|
| 6620 |
+
"epoch": 0.8037422921539443,
|
| 6621 |
+
"grad_norm": 0.34263697266578674,
|
| 6622 |
+
"learning_rate": 9.298335685620546e-06,
|
| 6623 |
+
"loss": 0.607,
|
| 6624 |
+
"step": 945
|
| 6625 |
+
},
|
| 6626 |
+
{
|
| 6627 |
+
"epoch": 0.8045928130980226,
|
| 6628 |
+
"grad_norm": 0.3440447151660919,
|
| 6629 |
+
"learning_rate": 9.22057045947366e-06,
|
| 6630 |
+
"loss": 0.6929,
|
| 6631 |
+
"step": 946
|
| 6632 |
+
},
|
| 6633 |
+
{
|
| 6634 |
+
"epoch": 0.8054433340421008,
|
| 6635 |
+
"grad_norm": 0.384171724319458,
|
| 6636 |
+
"learning_rate": 9.143098745988532e-06,
|
| 6637 |
+
"loss": 0.6745,
|
| 6638 |
+
"step": 947
|
| 6639 |
+
},
|
| 6640 |
+
{
|
| 6641 |
+
"epoch": 0.806293854986179,
|
| 6642 |
+
"grad_norm": 0.34123048186302185,
|
| 6643 |
+
"learning_rate": 9.065921102773045e-06,
|
| 6644 |
+
"loss": 0.6201,
|
| 6645 |
+
"step": 948
|
| 6646 |
+
},
|
| 6647 |
+
{
|
| 6648 |
+
"epoch": 0.8071443759302572,
|
| 6649 |
+
"grad_norm": 0.40928083658218384,
|
| 6650 |
+
"learning_rate": 8.98903808531838e-06,
|
| 6651 |
+
"loss": 0.6848,
|
| 6652 |
+
"step": 949
|
| 6653 |
+
},
|
| 6654 |
+
{
|
| 6655 |
+
"epoch": 0.8079948968743356,
|
| 6656 |
+
"grad_norm": 0.619849681854248,
|
| 6657 |
+
"learning_rate": 8.912450246995214e-06,
|
| 6658 |
+
"loss": 0.6359,
|
| 6659 |
+
"step": 950
|
| 6660 |
+
},
|
| 6661 |
+
{
|
| 6662 |
+
"epoch": 0.8088454178184138,
|
| 6663 |
+
"grad_norm": 0.24706372618675232,
|
| 6664 |
+
"learning_rate": 8.836158139049583e-06,
|
| 6665 |
+
"loss": 0.5911,
|
| 6666 |
+
"step": 951
|
| 6667 |
+
},
|
| 6668 |
+
{
|
| 6669 |
+
"epoch": 0.809695938762492,
|
| 6670 |
+
"grad_norm": 0.2719581425189972,
|
| 6671 |
+
"learning_rate": 8.760162310599046e-06,
|
| 6672 |
+
"loss": 0.5919,
|
| 6673 |
+
"step": 952
|
| 6674 |
+
},
|
| 6675 |
+
{
|
| 6676 |
+
"epoch": 0.8105464597065702,
|
| 6677 |
+
"grad_norm": 0.27172034978866577,
|
| 6678 |
+
"learning_rate": 8.684463308628616e-06,
|
| 6679 |
+
"loss": 0.6026,
|
| 6680 |
+
"step": 953
|
| 6681 |
+
},
|
| 6682 |
+
{
|
| 6683 |
+
"epoch": 0.8113969806506485,
|
| 6684 |
+
"grad_norm": 0.29025357961654663,
|
| 6685 |
+
"learning_rate": 8.60906167798694e-06,
|
| 6686 |
+
"loss": 0.6429,
|
| 6687 |
+
"step": 954
|
| 6688 |
+
},
|
| 6689 |
+
{
|
| 6690 |
+
"epoch": 0.8122475015947268,
|
| 6691 |
+
"grad_norm": 0.2830718159675598,
|
| 6692 |
+
"learning_rate": 8.533957961382238e-06,
|
| 6693 |
+
"loss": 0.6626,
|
| 6694 |
+
"step": 955
|
| 6695 |
+
},
|
| 6696 |
+
{
|
| 6697 |
+
"epoch": 0.813098022538805,
|
| 6698 |
+
"grad_norm": 0.3042718768119812,
|
| 6699 |
+
"learning_rate": 8.45915269937856e-06,
|
| 6700 |
+
"loss": 0.6581,
|
| 6701 |
+
"step": 956
|
| 6702 |
+
},
|
| 6703 |
+
{
|
| 6704 |
+
"epoch": 0.8139485434828833,
|
| 6705 |
+
"grad_norm": 0.2902892529964447,
|
| 6706 |
+
"learning_rate": 8.384646430391718e-06,
|
| 6707 |
+
"loss": 0.6694,
|
| 6708 |
+
"step": 957
|
| 6709 |
+
},
|
| 6710 |
+
{
|
| 6711 |
+
"epoch": 0.8147990644269615,
|
| 6712 |
+
"grad_norm": 0.29462185502052307,
|
| 6713 |
+
"learning_rate": 8.31043969068559e-06,
|
| 6714 |
+
"loss": 0.6201,
|
| 6715 |
+
"step": 958
|
| 6716 |
+
},
|
| 6717 |
+
{
|
| 6718 |
+
"epoch": 0.8156495853710397,
|
| 6719 |
+
"grad_norm": 0.3070443868637085,
|
| 6720 |
+
"learning_rate": 8.236533014368098e-06,
|
| 6721 |
+
"loss": 0.6926,
|
| 6722 |
+
"step": 959
|
| 6723 |
+
},
|
| 6724 |
+
{
|
| 6725 |
+
"epoch": 0.816500106315118,
|
| 6726 |
+
"grad_norm": 0.2813088893890381,
|
| 6727 |
+
"learning_rate": 8.1629269333875e-06,
|
| 6728 |
+
"loss": 0.608,
|
| 6729 |
+
"step": 960
|
| 6730 |
+
},
|
| 6731 |
+
{
|
| 6732 |
+
"epoch": 0.8173506272591963,
|
| 6733 |
+
"grad_norm": 0.28812092542648315,
|
| 6734 |
+
"learning_rate": 8.08962197752844e-06,
|
| 6735 |
+
"loss": 0.5971,
|
| 6736 |
+
"step": 961
|
| 6737 |
+
},
|
| 6738 |
+
{
|
| 6739 |
+
"epoch": 0.8182011482032745,
|
| 6740 |
+
"grad_norm": 0.289363831281662,
|
| 6741 |
+
"learning_rate": 8.016618674408232e-06,
|
| 6742 |
+
"loss": 0.6114,
|
| 6743 |
+
"step": 962
|
| 6744 |
+
},
|
| 6745 |
+
{
|
| 6746 |
+
"epoch": 0.8190516691473527,
|
| 6747 |
+
"grad_norm": 0.31117239594459534,
|
| 6748 |
+
"learning_rate": 7.943917549473034e-06,
|
| 6749 |
+
"loss": 0.6626,
|
| 6750 |
+
"step": 963
|
| 6751 |
+
},
|
| 6752 |
+
{
|
| 6753 |
+
"epoch": 0.819902190091431,
|
| 6754 |
+
"grad_norm": 0.3049095869064331,
|
| 6755 |
+
"learning_rate": 7.871519125994014e-06,
|
| 6756 |
+
"loss": 0.6402,
|
| 6757 |
+
"step": 964
|
| 6758 |
+
},
|
| 6759 |
+
{
|
| 6760 |
+
"epoch": 0.8207527110355093,
|
| 6761 |
+
"grad_norm": 0.31192365288734436,
|
| 6762 |
+
"learning_rate": 7.79942392506367e-06,
|
| 6763 |
+
"loss": 0.6394,
|
| 6764 |
+
"step": 965
|
| 6765 |
+
},
|
| 6766 |
+
{
|
| 6767 |
+
"epoch": 0.8216032319795875,
|
| 6768 |
+
"grad_norm": 0.3218645751476288,
|
| 6769 |
+
"learning_rate": 7.727632465591995e-06,
|
| 6770 |
+
"loss": 0.6883,
|
| 6771 |
+
"step": 966
|
| 6772 |
+
},
|
| 6773 |
+
{
|
| 6774 |
+
"epoch": 0.8224537529236657,
|
| 6775 |
+
"grad_norm": 0.31128594279289246,
|
| 6776 |
+
"learning_rate": 7.656145264302822e-06,
|
| 6777 |
+
"loss": 0.6311,
|
| 6778 |
+
"step": 967
|
| 6779 |
+
},
|
| 6780 |
+
{
|
| 6781 |
+
"epoch": 0.823304273867744,
|
| 6782 |
+
"grad_norm": 0.3170647919178009,
|
| 6783 |
+
"learning_rate": 7.584962835730025e-06,
|
| 6784 |
+
"loss": 0.634,
|
| 6785 |
+
"step": 968
|
| 6786 |
+
},
|
| 6787 |
+
{
|
| 6788 |
+
"epoch": 0.8241547948118222,
|
| 6789 |
+
"grad_norm": 0.3140490651130676,
|
| 6790 |
+
"learning_rate": 7.514085692213907e-06,
|
| 6791 |
+
"loss": 0.6386,
|
| 6792 |
+
"step": 969
|
| 6793 |
+
},
|
| 6794 |
+
{
|
| 6795 |
+
"epoch": 0.8250053157559005,
|
| 6796 |
+
"grad_norm": 0.31740131974220276,
|
| 6797 |
+
"learning_rate": 7.443514343897395e-06,
|
| 6798 |
+
"loss": 0.6564,
|
| 6799 |
+
"step": 970
|
| 6800 |
+
},
|
| 6801 |
+
{
|
| 6802 |
+
"epoch": 0.8258558366999788,
|
| 6803 |
+
"grad_norm": 0.31138667464256287,
|
| 6804 |
+
"learning_rate": 7.373249298722506e-06,
|
| 6805 |
+
"loss": 0.6541,
|
| 6806 |
+
"step": 971
|
| 6807 |
+
},
|
| 6808 |
+
{
|
| 6809 |
+
"epoch": 0.826706357644057,
|
| 6810 |
+
"grad_norm": 0.31116360425949097,
|
| 6811 |
+
"learning_rate": 7.303291062426559e-06,
|
| 6812 |
+
"loss": 0.611,
|
| 6813 |
+
"step": 972
|
| 6814 |
+
},
|
| 6815 |
+
{
|
| 6816 |
+
"epoch": 0.8275568785881352,
|
| 6817 |
+
"grad_norm": 0.3222910165786743,
|
| 6818 |
+
"learning_rate": 7.23364013853865e-06,
|
| 6819 |
+
"loss": 0.6609,
|
| 6820 |
+
"step": 973
|
| 6821 |
+
},
|
| 6822 |
+
{
|
| 6823 |
+
"epoch": 0.8284073995322134,
|
| 6824 |
+
"grad_norm": 0.33877184987068176,
|
| 6825 |
+
"learning_rate": 7.164297028375911e-06,
|
| 6826 |
+
"loss": 0.6892,
|
| 6827 |
+
"step": 974
|
| 6828 |
+
},
|
| 6829 |
+
{
|
| 6830 |
+
"epoch": 0.8292579204762918,
|
| 6831 |
+
"grad_norm": 0.326835960149765,
|
| 6832 |
+
"learning_rate": 7.095262231040029e-06,
|
| 6833 |
+
"loss": 0.5931,
|
| 6834 |
+
"step": 975
|
| 6835 |
+
},
|
| 6836 |
+
{
|
| 6837 |
+
"epoch": 0.83010844142037,
|
| 6838 |
+
"grad_norm": 0.31615501642227173,
|
| 6839 |
+
"learning_rate": 7.02653624341354e-06,
|
| 6840 |
+
"loss": 0.5906,
|
| 6841 |
+
"step": 976
|
| 6842 |
+
},
|
| 6843 |
+
{
|
| 6844 |
+
"epoch": 0.8309589623644482,
|
| 6845 |
+
"grad_norm": 0.3255496323108673,
|
| 6846 |
+
"learning_rate": 6.9581195601563175e-06,
|
| 6847 |
+
"loss": 0.6243,
|
| 6848 |
+
"step": 977
|
| 6849 |
+
},
|
| 6850 |
+
{
|
| 6851 |
+
"epoch": 0.8318094833085264,
|
| 6852 |
+
"grad_norm": 0.3078942894935608,
|
| 6853 |
+
"learning_rate": 6.890012673702028e-06,
|
| 6854 |
+
"loss": 0.6353,
|
| 6855 |
+
"step": 978
|
| 6856 |
+
},
|
| 6857 |
+
{
|
| 6858 |
+
"epoch": 0.8326600042526047,
|
| 6859 |
+
"grad_norm": 0.3214848041534424,
|
| 6860 |
+
"learning_rate": 6.8222160742544895e-06,
|
| 6861 |
+
"loss": 0.6501,
|
| 6862 |
+
"step": 979
|
| 6863 |
+
},
|
| 6864 |
+
{
|
| 6865 |
+
"epoch": 0.833510525196683,
|
| 6866 |
+
"grad_norm": 0.3080413341522217,
|
| 6867 |
+
"learning_rate": 6.754730249784269e-06,
|
| 6868 |
+
"loss": 0.6749,
|
| 6869 |
+
"step": 980
|
| 6870 |
+
},
|
| 6871 |
+
{
|
| 6872 |
+
"epoch": 0.8343610461407612,
|
| 6873 |
+
"grad_norm": 0.3155469298362732,
|
| 6874 |
+
"learning_rate": 6.687555686025071e-06,
|
| 6875 |
+
"loss": 0.6513,
|
| 6876 |
+
"step": 981
|
| 6877 |
+
},
|
| 6878 |
+
{
|
| 6879 |
+
"epoch": 0.8352115670848395,
|
| 6880 |
+
"grad_norm": 0.3147723972797394,
|
| 6881 |
+
"learning_rate": 6.6206928664703076e-06,
|
| 6882 |
+
"loss": 0.6342,
|
| 6883 |
+
"step": 982
|
| 6884 |
+
},
|
| 6885 |
+
{
|
| 6886 |
+
"epoch": 0.8360620880289177,
|
| 6887 |
+
"grad_norm": 0.3308815062046051,
|
| 6888 |
+
"learning_rate": 6.554142272369562e-06,
|
| 6889 |
+
"loss": 0.6499,
|
| 6890 |
+
"step": 983
|
| 6891 |
+
},
|
| 6892 |
+
{
|
| 6893 |
+
"epoch": 0.8369126089729959,
|
| 6894 |
+
"grad_norm": 0.33099743723869324,
|
| 6895 |
+
"learning_rate": 6.487904382725191e-06,
|
| 6896 |
+
"loss": 0.6842,
|
| 6897 |
+
"step": 984
|
| 6898 |
+
},
|
| 6899 |
+
{
|
| 6900 |
+
"epoch": 0.8377631299170742,
|
| 6901 |
+
"grad_norm": 0.3191700875759125,
|
| 6902 |
+
"learning_rate": 6.421979674288797e-06,
|
| 6903 |
+
"loss": 0.6083,
|
| 6904 |
+
"step": 985
|
| 6905 |
+
},
|
| 6906 |
+
{
|
| 6907 |
+
"epoch": 0.8386136508611525,
|
| 6908 |
+
"grad_norm": 0.3238374888896942,
|
| 6909 |
+
"learning_rate": 6.356368621557884e-06,
|
| 6910 |
+
"loss": 0.649,
|
| 6911 |
+
"step": 986
|
| 6912 |
+
},
|
| 6913 |
+
{
|
| 6914 |
+
"epoch": 0.8394641718052307,
|
| 6915 |
+
"grad_norm": 0.33162933588027954,
|
| 6916 |
+
"learning_rate": 6.291071696772361e-06,
|
| 6917 |
+
"loss": 0.6485,
|
| 6918 |
+
"step": 987
|
| 6919 |
+
},
|
| 6920 |
+
{
|
| 6921 |
+
"epoch": 0.8403146927493089,
|
| 6922 |
+
"grad_norm": 0.3484635055065155,
|
| 6923 |
+
"learning_rate": 6.226089369911209e-06,
|
| 6924 |
+
"loss": 0.6364,
|
| 6925 |
+
"step": 988
|
| 6926 |
+
},
|
| 6927 |
+
{
|
| 6928 |
+
"epoch": 0.8411652136933871,
|
| 6929 |
+
"grad_norm": 0.31507810950279236,
|
| 6930 |
+
"learning_rate": 6.161422108689035e-06,
|
| 6931 |
+
"loss": 0.6333,
|
| 6932 |
+
"step": 989
|
| 6933 |
+
},
|
| 6934 |
+
{
|
| 6935 |
+
"epoch": 0.8420157346374655,
|
| 6936 |
+
"grad_norm": 0.34592968225479126,
|
| 6937 |
+
"learning_rate": 6.0970703785527885e-06,
|
| 6938 |
+
"loss": 0.6297,
|
| 6939 |
+
"step": 990
|
| 6940 |
+
},
|
| 6941 |
+
{
|
| 6942 |
+
"epoch": 0.8428662555815437,
|
| 6943 |
+
"grad_norm": 0.34727707505226135,
|
| 6944 |
+
"learning_rate": 6.033034642678298e-06,
|
| 6945 |
+
"loss": 0.643,
|
| 6946 |
+
"step": 991
|
| 6947 |
+
},
|
| 6948 |
+
{
|
| 6949 |
+
"epoch": 0.8437167765256219,
|
| 6950 |
+
"grad_norm": 0.3230702579021454,
|
| 6951 |
+
"learning_rate": 5.969315361967087e-06,
|
| 6952 |
+
"loss": 0.6469,
|
| 6953 |
+
"step": 992
|
| 6954 |
+
},
|
| 6955 |
+
{
|
| 6956 |
+
"epoch": 0.8445672974697002,
|
| 6957 |
+
"grad_norm": 0.326414555311203,
|
| 6958 |
+
"learning_rate": 5.905912995042895e-06,
|
| 6959 |
+
"loss": 0.632,
|
| 6960 |
+
"step": 993
|
| 6961 |
+
},
|
| 6962 |
+
{
|
| 6963 |
+
"epoch": 0.8454178184137784,
|
| 6964 |
+
"grad_norm": 0.35592177510261536,
|
| 6965 |
+
"learning_rate": 5.842827998248496e-06,
|
| 6966 |
+
"loss": 0.6419,
|
| 6967 |
+
"step": 994
|
| 6968 |
+
},
|
| 6969 |
+
{
|
| 6970 |
+
"epoch": 0.8462683393578567,
|
| 6971 |
+
"grad_norm": 0.340466171503067,
|
| 6972 |
+
"learning_rate": 5.780060825642358e-06,
|
| 6973 |
+
"loss": 0.6777,
|
| 6974 |
+
"step": 995
|
| 6975 |
+
},
|
| 6976 |
+
{
|
| 6977 |
+
"epoch": 0.847118860301935,
|
| 6978 |
+
"grad_norm": 0.3180592954158783,
|
| 6979 |
+
"learning_rate": 5.71761192899537e-06,
|
| 6980 |
+
"loss": 0.6048,
|
| 6981 |
+
"step": 996
|
| 6982 |
+
},
|
| 6983 |
+
{
|
| 6984 |
+
"epoch": 0.8479693812460132,
|
| 6985 |
+
"grad_norm": 0.3615611493587494,
|
| 6986 |
+
"learning_rate": 5.655481757787651e-06,
|
| 6987 |
+
"loss": 0.6771,
|
| 6988 |
+
"step": 997
|
| 6989 |
+
},
|
| 6990 |
+
{
|
| 6991 |
+
"epoch": 0.8488199021900914,
|
| 6992 |
+
"grad_norm": 0.3619909882545471,
|
| 6993 |
+
"learning_rate": 5.5936707592052274e-06,
|
| 6994 |
+
"loss": 0.6663,
|
| 6995 |
+
"step": 998
|
| 6996 |
+
},
|
| 6997 |
+
{
|
| 6998 |
+
"epoch": 0.8496704231341696,
|
| 6999 |
+
"grad_norm": 0.3606683909893036,
|
| 7000 |
+
"learning_rate": 5.532179378136904e-06,
|
| 7001 |
+
"loss": 0.6028,
|
| 7002 |
+
"step": 999
|
| 7003 |
+
},
|
| 7004 |
+
{
|
| 7005 |
+
"epoch": 0.850520944078248,
|
| 7006 |
+
"grad_norm": 0.6209678649902344,
|
| 7007 |
+
"learning_rate": 5.471008057170973e-06,
|
| 7008 |
+
"loss": 0.544,
|
| 7009 |
+
"step": 1000
|
| 7010 |
+
},
|
| 7011 |
+
{
|
| 7012 |
+
"epoch": 0.8513714650223262,
|
| 7013 |
+
"grad_norm": 0.23854555189609528,
|
| 7014 |
+
"learning_rate": 5.410157236592111e-06,
|
| 7015 |
+
"loss": 0.562,
|
| 7016 |
+
"step": 1001
|
| 7017 |
+
},
|
| 7018 |
+
{
|
| 7019 |
+
"epoch": 0.8522219859664044,
|
| 7020 |
+
"grad_norm": 0.2729296088218689,
|
| 7021 |
+
"learning_rate": 5.3496273543781396e-06,
|
| 7022 |
+
"loss": 0.6074,
|
| 7023 |
+
"step": 1002
|
| 7024 |
+
},
|
| 7025 |
+
{
|
| 7026 |
+
"epoch": 0.8530725069104826,
|
| 7027 |
+
"grad_norm": 0.27674445509910583,
|
| 7028 |
+
"learning_rate": 5.289418846196931e-06,
|
| 7029 |
+
"loss": 0.6595,
|
| 7030 |
+
"step": 1003
|
| 7031 |
+
},
|
| 7032 |
+
{
|
| 7033 |
+
"epoch": 0.8539230278545609,
|
| 7034 |
+
"grad_norm": 0.2779724597930908,
|
| 7035 |
+
"learning_rate": 5.229532145403226e-06,
|
| 7036 |
+
"loss": 0.6383,
|
| 7037 |
+
"step": 1004
|
| 7038 |
+
},
|
| 7039 |
+
{
|
| 7040 |
+
"epoch": 0.8547735487986392,
|
| 7041 |
+
"grad_norm": 0.2887398600578308,
|
| 7042 |
+
"learning_rate": 5.1699676830355595e-06,
|
| 7043 |
+
"loss": 0.6396,
|
| 7044 |
+
"step": 1005
|
| 7045 |
+
},
|
| 7046 |
+
{
|
| 7047 |
+
"epoch": 0.8556240697427174,
|
| 7048 |
+
"grad_norm": 0.2952885925769806,
|
| 7049 |
+
"learning_rate": 5.110725887813089e-06,
|
| 7050 |
+
"loss": 0.7069,
|
| 7051 |
+
"step": 1006
|
| 7052 |
+
},
|
| 7053 |
+
{
|
| 7054 |
+
"epoch": 0.8564745906867957,
|
| 7055 |
+
"grad_norm": 0.2868542969226837,
|
| 7056 |
+
"learning_rate": 5.051807186132623e-06,
|
| 7057 |
+
"loss": 0.6296,
|
| 7058 |
+
"step": 1007
|
| 7059 |
+
},
|
| 7060 |
+
{
|
| 7061 |
+
"epoch": 0.8573251116308739,
|
| 7062 |
+
"grad_norm": 0.28352710604667664,
|
| 7063 |
+
"learning_rate": 4.993212002065411e-06,
|
| 7064 |
+
"loss": 0.625,
|
| 7065 |
+
"step": 1008
|
| 7066 |
+
},
|
| 7067 |
+
{
|
| 7068 |
+
"epoch": 0.8581756325749521,
|
| 7069 |
+
"grad_norm": 0.30322712659835815,
|
| 7070 |
+
"learning_rate": 4.934940757354223e-06,
|
| 7071 |
+
"loss": 0.6734,
|
| 7072 |
+
"step": 1009
|
| 7073 |
+
},
|
| 7074 |
+
{
|
| 7075 |
+
"epoch": 0.8590261535190304,
|
| 7076 |
+
"grad_norm": 0.3017674386501312,
|
| 7077 |
+
"learning_rate": 4.8769938714102015e-06,
|
| 7078 |
+
"loss": 0.6175,
|
| 7079 |
+
"step": 1010
|
| 7080 |
+
},
|
| 7081 |
+
{
|
| 7082 |
+
"epoch": 0.8598766744631087,
|
| 7083 |
+
"grad_norm": 0.2911052405834198,
|
| 7084 |
+
"learning_rate": 4.819371761309937e-06,
|
| 7085 |
+
"loss": 0.6666,
|
| 7086 |
+
"step": 1011
|
| 7087 |
+
},
|
| 7088 |
+
{
|
| 7089 |
+
"epoch": 0.8607271954071869,
|
| 7090 |
+
"grad_norm": 0.2917759418487549,
|
| 7091 |
+
"learning_rate": 4.762074841792391e-06,
|
| 7092 |
+
"loss": 0.6376,
|
| 7093 |
+
"step": 1012
|
| 7094 |
+
},
|
| 7095 |
+
{
|
| 7096 |
+
"epoch": 0.8615777163512651,
|
| 7097 |
+
"grad_norm": 0.305550217628479,
|
| 7098 |
+
"learning_rate": 4.705103525255983e-06,
|
| 7099 |
+
"loss": 0.6405,
|
| 7100 |
+
"step": 1013
|
| 7101 |
+
},
|
| 7102 |
+
{
|
| 7103 |
+
"epoch": 0.8624282372953433,
|
| 7104 |
+
"grad_norm": 0.28329166769981384,
|
| 7105 |
+
"learning_rate": 4.648458221755531e-06,
|
| 7106 |
+
"loss": 0.5857,
|
| 7107 |
+
"step": 1014
|
| 7108 |
+
},
|
| 7109 |
+
{
|
| 7110 |
+
"epoch": 0.8632787582394217,
|
| 7111 |
+
"grad_norm": 0.3194465935230255,
|
| 7112 |
+
"learning_rate": 4.592139338999413e-06,
|
| 7113 |
+
"loss": 0.6446,
|
| 7114 |
+
"step": 1015
|
| 7115 |
+
},
|
| 7116 |
+
{
|
| 7117 |
+
"epoch": 0.8641292791834999,
|
| 7118 |
+
"grad_norm": 0.3141101002693176,
|
| 7119 |
+
"learning_rate": 4.53614728234652e-06,
|
| 7120 |
+
"loss": 0.6597,
|
| 7121 |
+
"step": 1016
|
| 7122 |
+
},
|
| 7123 |
+
{
|
| 7124 |
+
"epoch": 0.8649798001275781,
|
| 7125 |
+
"grad_norm": 0.2985946238040924,
|
| 7126 |
+
"learning_rate": 4.480482454803409e-06,
|
| 7127 |
+
"loss": 0.6309,
|
| 7128 |
+
"step": 1017
|
| 7129 |
+
},
|
| 7130 |
+
{
|
| 7131 |
+
"epoch": 0.8658303210716564,
|
| 7132 |
+
"grad_norm": 0.3038308918476105,
|
| 7133 |
+
"learning_rate": 4.425145257021407e-06,
|
| 7134 |
+
"loss": 0.6476,
|
| 7135 |
+
"step": 1018
|
| 7136 |
+
},
|
| 7137 |
+
{
|
| 7138 |
+
"epoch": 0.8666808420157346,
|
| 7139 |
+
"grad_norm": 0.3121652603149414,
|
| 7140 |
+
"learning_rate": 4.370136087293658e-06,
|
| 7141 |
+
"loss": 0.6613,
|
| 7142 |
+
"step": 1019
|
| 7143 |
+
},
|
| 7144 |
+
{
|
| 7145 |
+
"epoch": 0.8675313629598129,
|
| 7146 |
+
"grad_norm": 0.31806936860084534,
|
| 7147 |
+
"learning_rate": 4.3154553415523435e-06,
|
| 7148 |
+
"loss": 0.669,
|
| 7149 |
+
"step": 1020
|
| 7150 |
+
},
|
| 7151 |
+
{
|
| 7152 |
+
"epoch": 0.8683818839038911,
|
| 7153 |
+
"grad_norm": 0.3062193989753723,
|
| 7154 |
+
"learning_rate": 4.261103413365758e-06,
|
| 7155 |
+
"loss": 0.6823,
|
| 7156 |
+
"step": 1021
|
| 7157 |
+
},
|
| 7158 |
+
{
|
| 7159 |
+
"epoch": 0.8692324048479694,
|
| 7160 |
+
"grad_norm": 0.2995758056640625,
|
| 7161 |
+
"learning_rate": 4.207080693935556e-06,
|
| 7162 |
+
"loss": 0.5908,
|
| 7163 |
+
"step": 1022
|
| 7164 |
+
},
|
| 7165 |
+
{
|
| 7166 |
+
"epoch": 0.8700829257920476,
|
| 7167 |
+
"grad_norm": 0.31240278482437134,
|
| 7168 |
+
"learning_rate": 4.153387572093836e-06,
|
| 7169 |
+
"loss": 0.6536,
|
| 7170 |
+
"step": 1023
|
| 7171 |
+
},
|
| 7172 |
+
{
|
| 7173 |
+
"epoch": 0.8709334467361258,
|
| 7174 |
+
"grad_norm": 0.29285261034965515,
|
| 7175 |
+
"learning_rate": 4.100024434300437e-06,
|
| 7176 |
+
"loss": 0.5865,
|
| 7177 |
+
"step": 1024
|
| 7178 |
+
},
|
| 7179 |
+
{
|
| 7180 |
+
"epoch": 0.8717839676802042,
|
| 7181 |
+
"grad_norm": 0.30669569969177246,
|
| 7182 |
+
"learning_rate": 4.046991664640081e-06,
|
| 7183 |
+
"loss": 0.657,
|
| 7184 |
+
"step": 1025
|
| 7185 |
+
},
|
| 7186 |
+
{
|
| 7187 |
+
"epoch": 0.8726344886242824,
|
| 7188 |
+
"grad_norm": 0.3080507218837738,
|
| 7189 |
+
"learning_rate": 3.994289644819682e-06,
|
| 7190 |
+
"loss": 0.6843,
|
| 7191 |
+
"step": 1026
|
| 7192 |
+
},
|
| 7193 |
+
{
|
| 7194 |
+
"epoch": 0.8734850095683606,
|
| 7195 |
+
"grad_norm": 0.3329055905342102,
|
| 7196 |
+
"learning_rate": 3.941918754165519e-06,
|
| 7197 |
+
"loss": 0.6768,
|
| 7198 |
+
"step": 1027
|
| 7199 |
+
},
|
| 7200 |
+
{
|
| 7201 |
+
"epoch": 0.8743355305124388,
|
| 7202 |
+
"grad_norm": 0.3102298080921173,
|
| 7203 |
+
"learning_rate": 3.889879369620591e-06,
|
| 7204 |
+
"loss": 0.6481,
|
| 7205 |
+
"step": 1028
|
| 7206 |
+
},
|
| 7207 |
+
{
|
| 7208 |
+
"epoch": 0.8751860514565171,
|
| 7209 |
+
"grad_norm": 0.3252778947353363,
|
| 7210 |
+
"learning_rate": 3.838171865741813e-06,
|
| 7211 |
+
"loss": 0.6475,
|
| 7212 |
+
"step": 1029
|
| 7213 |
+
},
|
| 7214 |
+
{
|
| 7215 |
+
"epoch": 0.8760365724005954,
|
| 7216 |
+
"grad_norm": 0.3287936747074127,
|
| 7217 |
+
"learning_rate": 3.786796614697413e-06,
|
| 7218 |
+
"loss": 0.6508,
|
| 7219 |
+
"step": 1030
|
| 7220 |
+
},
|
| 7221 |
+
{
|
| 7222 |
+
"epoch": 0.8768870933446736,
|
| 7223 |
+
"grad_norm": 0.3356083631515503,
|
| 7224 |
+
"learning_rate": 3.735753986264162e-06,
|
| 7225 |
+
"loss": 0.6309,
|
| 7226 |
+
"step": 1031
|
| 7227 |
+
},
|
| 7228 |
+
{
|
| 7229 |
+
"epoch": 0.8777376142887519,
|
| 7230 |
+
"grad_norm": 0.32422903180122375,
|
| 7231 |
+
"learning_rate": 3.6850443478247953e-06,
|
| 7232 |
+
"loss": 0.6436,
|
| 7233 |
+
"step": 1032
|
| 7234 |
+
},
|
| 7235 |
+
{
|
| 7236 |
+
"epoch": 0.8785881352328301,
|
| 7237 |
+
"grad_norm": 0.3286559581756592,
|
| 7238 |
+
"learning_rate": 3.634668064365304e-06,
|
| 7239 |
+
"loss": 0.6278,
|
| 7240 |
+
"step": 1033
|
| 7241 |
+
},
|
| 7242 |
+
{
|
| 7243 |
+
"epoch": 0.8794386561769083,
|
| 7244 |
+
"grad_norm": 0.3390061557292938,
|
| 7245 |
+
"learning_rate": 3.5846254984723605e-06,
|
| 7246 |
+
"loss": 0.7267,
|
| 7247 |
+
"step": 1034
|
| 7248 |
+
},
|
| 7249 |
+
{
|
| 7250 |
+
"epoch": 0.8802891771209866,
|
| 7251 |
+
"grad_norm": 0.29817309975624084,
|
| 7252 |
+
"learning_rate": 3.5349170103306515e-06,
|
| 7253 |
+
"loss": 0.5605,
|
| 7254 |
+
"step": 1035
|
| 7255 |
+
},
|
| 7256 |
+
{
|
| 7257 |
+
"epoch": 0.8811396980650649,
|
| 7258 |
+
"grad_norm": 0.334392786026001,
|
| 7259 |
+
"learning_rate": 3.485542957720339e-06,
|
| 7260 |
+
"loss": 0.6439,
|
| 7261 |
+
"step": 1036
|
| 7262 |
+
},
|
| 7263 |
+
{
|
| 7264 |
+
"epoch": 0.8819902190091431,
|
| 7265 |
+
"grad_norm": 0.3293307423591614,
|
| 7266 |
+
"learning_rate": 3.4365036960144647e-06,
|
| 7267 |
+
"loss": 0.5955,
|
| 7268 |
+
"step": 1037
|
| 7269 |
+
},
|
| 7270 |
+
{
|
| 7271 |
+
"epoch": 0.8828407399532213,
|
| 7272 |
+
"grad_norm": 0.33913591504096985,
|
| 7273 |
+
"learning_rate": 3.3877995781763704e-06,
|
| 7274 |
+
"loss": 0.6884,
|
| 7275 |
+
"step": 1038
|
| 7276 |
+
},
|
| 7277 |
+
{
|
| 7278 |
+
"epoch": 0.8836912608972995,
|
| 7279 |
+
"grad_norm": 0.34623920917510986,
|
| 7280 |
+
"learning_rate": 3.3394309547572055e-06,
|
| 7281 |
+
"loss": 0.6698,
|
| 7282 |
+
"step": 1039
|
| 7283 |
+
},
|
| 7284 |
+
{
|
| 7285 |
+
"epoch": 0.8845417818413779,
|
| 7286 |
+
"grad_norm": 0.3281223773956299,
|
| 7287 |
+
"learning_rate": 3.29139817389334e-06,
|
| 7288 |
+
"loss": 0.6039,
|
| 7289 |
+
"step": 1040
|
| 7290 |
+
},
|
| 7291 |
+
{
|
| 7292 |
+
"epoch": 0.8853923027854561,
|
| 7293 |
+
"grad_norm": 0.33946382999420166,
|
| 7294 |
+
"learning_rate": 3.243701581303937e-06,
|
| 7295 |
+
"loss": 0.6989,
|
| 7296 |
+
"step": 1041
|
| 7297 |
+
},
|
| 7298 |
+
{
|
| 7299 |
+
"epoch": 0.8862428237295343,
|
| 7300 |
+
"grad_norm": 0.3334444463253021,
|
| 7301 |
+
"learning_rate": 3.1963415202883817e-06,
|
| 7302 |
+
"loss": 0.6411,
|
| 7303 |
+
"step": 1042
|
| 7304 |
+
},
|
| 7305 |
+
{
|
| 7306 |
+
"epoch": 0.8870933446736126,
|
| 7307 |
+
"grad_norm": 0.33876511454582214,
|
| 7308 |
+
"learning_rate": 3.1493183317238885e-06,
|
| 7309 |
+
"loss": 0.6504,
|
| 7310 |
+
"step": 1043
|
| 7311 |
+
},
|
| 7312 |
+
{
|
| 7313 |
+
"epoch": 0.8879438656176908,
|
| 7314 |
+
"grad_norm": 0.35493630170822144,
|
| 7315 |
+
"learning_rate": 3.1026323540629676e-06,
|
| 7316 |
+
"loss": 0.6613,
|
| 7317 |
+
"step": 1044
|
| 7318 |
+
},
|
| 7319 |
+
{
|
| 7320 |
+
"epoch": 0.8887943865617691,
|
| 7321 |
+
"grad_norm": 0.34328144788742065,
|
| 7322 |
+
"learning_rate": 3.056283923331077e-06,
|
| 7323 |
+
"loss": 0.6736,
|
| 7324 |
+
"step": 1045
|
| 7325 |
+
},
|
| 7326 |
+
{
|
| 7327 |
+
"epoch": 0.8896449075058473,
|
| 7328 |
+
"grad_norm": 0.37152862548828125,
|
| 7329 |
+
"learning_rate": 3.0102733731241116e-06,
|
| 7330 |
+
"loss": 0.6827,
|
| 7331 |
+
"step": 1046
|
| 7332 |
+
},
|
| 7333 |
+
{
|
| 7334 |
+
"epoch": 0.8904954284499256,
|
| 7335 |
+
"grad_norm": 0.359235554933548,
|
| 7336 |
+
"learning_rate": 2.9646010346061013e-06,
|
| 7337 |
+
"loss": 0.5788,
|
| 7338 |
+
"step": 1047
|
| 7339 |
+
},
|
| 7340 |
+
{
|
| 7341 |
+
"epoch": 0.8913459493940038,
|
| 7342 |
+
"grad_norm": 0.3687358498573303,
|
| 7343 |
+
"learning_rate": 2.919267236506723e-06,
|
| 7344 |
+
"loss": 0.6111,
|
| 7345 |
+
"step": 1048
|
| 7346 |
+
},
|
| 7347 |
+
{
|
| 7348 |
+
"epoch": 0.892196470338082,
|
| 7349 |
+
"grad_norm": 0.3757440745830536,
|
| 7350 |
+
"learning_rate": 2.8742723051190246e-06,
|
| 7351 |
+
"loss": 0.6099,
|
| 7352 |
+
"step": 1049
|
| 7353 |
+
},
|
| 7354 |
+
{
|
| 7355 |
+
"epoch": 0.8930469912821604,
|
| 7356 |
+
"grad_norm": 0.5482544898986816,
|
| 7357 |
+
"learning_rate": 2.8296165642970108e-06,
|
| 7358 |
+
"loss": 0.5869,
|
| 7359 |
+
"step": 1050
|
| 7360 |
+
},
|
| 7361 |
+
{
|
| 7362 |
+
"epoch": 0.8938975122262386,
|
| 7363 |
+
"grad_norm": 0.2548101544380188,
|
| 7364 |
+
"learning_rate": 2.7853003354533556e-06,
|
| 7365 |
+
"loss": 0.6081,
|
| 7366 |
+
"step": 1051
|
| 7367 |
+
},
|
| 7368 |
+
{
|
| 7369 |
+
"epoch": 0.8947480331703168,
|
| 7370 |
+
"grad_norm": 0.28014668822288513,
|
| 7371 |
+
"learning_rate": 2.7413239375570832e-06,
|
| 7372 |
+
"loss": 0.5798,
|
| 7373 |
+
"step": 1052
|
| 7374 |
+
},
|
| 7375 |
+
{
|
| 7376 |
+
"epoch": 0.895598554114395,
|
| 7377 |
+
"grad_norm": 0.2735235095024109,
|
| 7378 |
+
"learning_rate": 2.697687687131234e-06,
|
| 7379 |
+
"loss": 0.6342,
|
| 7380 |
+
"step": 1053
|
| 7381 |
+
},
|
| 7382 |
+
{
|
| 7383 |
+
"epoch": 0.8964490750584733,
|
| 7384 |
+
"grad_norm": 0.280996173620224,
|
| 7385 |
+
"learning_rate": 2.65439189825063e-06,
|
| 7386 |
+
"loss": 0.656,
|
| 7387 |
+
"step": 1054
|
| 7388 |
+
},
|
| 7389 |
+
{
|
| 7390 |
+
"epoch": 0.8972995960025516,
|
| 7391 |
+
"grad_norm": 0.28959381580352783,
|
| 7392 |
+
"learning_rate": 2.6114368825395975e-06,
|
| 7393 |
+
"loss": 0.6531,
|
| 7394 |
+
"step": 1055
|
| 7395 |
+
},
|
| 7396 |
+
{
|
| 7397 |
+
"epoch": 0.8981501169466298,
|
| 7398 |
+
"grad_norm": 0.2938001751899719,
|
| 7399 |
+
"learning_rate": 2.5688229491697357e-06,
|
| 7400 |
+
"loss": 0.6901,
|
| 7401 |
+
"step": 1056
|
| 7402 |
+
},
|
| 7403 |
+
{
|
| 7404 |
+
"epoch": 0.899000637890708,
|
| 7405 |
+
"grad_norm": 0.2926790714263916,
|
| 7406 |
+
"learning_rate": 2.5265504048576473e-06,
|
| 7407 |
+
"loss": 0.6705,
|
| 7408 |
+
"step": 1057
|
| 7409 |
+
},
|
| 7410 |
+
{
|
| 7411 |
+
"epoch": 0.8998511588347863,
|
| 7412 |
+
"grad_norm": 0.3077324628829956,
|
| 7413 |
+
"learning_rate": 2.48461955386281e-06,
|
| 7414 |
+
"loss": 0.6255,
|
| 7415 |
+
"step": 1058
|
| 7416 |
+
},
|
| 7417 |
+
{
|
| 7418 |
+
"epoch": 0.9007016797788645,
|
| 7419 |
+
"grad_norm": 0.3055940866470337,
|
| 7420 |
+
"learning_rate": 2.4430306979853047e-06,
|
| 7421 |
+
"loss": 0.643,
|
| 7422 |
+
"step": 1059
|
| 7423 |
+
},
|
| 7424 |
+
{
|
| 7425 |
+
"epoch": 0.9015522007229428,
|
| 7426 |
+
"grad_norm": 0.3011234998703003,
|
| 7427 |
+
"learning_rate": 2.4017841365637005e-06,
|
| 7428 |
+
"loss": 0.6488,
|
| 7429 |
+
"step": 1060
|
| 7430 |
+
},
|
| 7431 |
+
{
|
| 7432 |
+
"epoch": 0.9024027216670211,
|
| 7433 |
+
"grad_norm": 0.2801249027252197,
|
| 7434 |
+
"learning_rate": 2.360880166472862e-06,
|
| 7435 |
+
"loss": 0.6099,
|
| 7436 |
+
"step": 1061
|
| 7437 |
+
},
|
| 7438 |
+
{
|
| 7439 |
+
"epoch": 0.9032532426110993,
|
| 7440 |
+
"grad_norm": 0.28682729601860046,
|
| 7441 |
+
"learning_rate": 2.3203190821218525e-06,
|
| 7442 |
+
"loss": 0.6104,
|
| 7443 |
+
"step": 1062
|
| 7444 |
+
},
|
| 7445 |
+
{
|
| 7446 |
+
"epoch": 0.9041037635551775,
|
| 7447 |
+
"grad_norm": 0.30458736419677734,
|
| 7448 |
+
"learning_rate": 2.280101175451771e-06,
|
| 7449 |
+
"loss": 0.6631,
|
| 7450 |
+
"step": 1063
|
| 7451 |
+
},
|
| 7452 |
+
{
|
| 7453 |
+
"epoch": 0.9049542844992557,
|
| 7454 |
+
"grad_norm": 0.3022899627685547,
|
| 7455 |
+
"learning_rate": 2.2402267359336872e-06,
|
| 7456 |
+
"loss": 0.5977,
|
| 7457 |
+
"step": 1064
|
| 7458 |
+
},
|
| 7459 |
+
{
|
| 7460 |
+
"epoch": 0.9058048054433341,
|
| 7461 |
+
"grad_norm": 0.3096393346786499,
|
| 7462 |
+
"learning_rate": 2.200696050566525e-06,
|
| 7463 |
+
"loss": 0.6303,
|
| 7464 |
+
"step": 1065
|
| 7465 |
+
},
|
| 7466 |
+
{
|
| 7467 |
+
"epoch": 0.9066553263874123,
|
| 7468 |
+
"grad_norm": 0.3267126977443695,
|
| 7469 |
+
"learning_rate": 2.1615094038750396e-06,
|
| 7470 |
+
"loss": 0.6633,
|
| 7471 |
+
"step": 1066
|
| 7472 |
+
},
|
| 7473 |
+
{
|
| 7474 |
+
"epoch": 0.9075058473314905,
|
| 7475 |
+
"grad_norm": 0.31724464893341064,
|
| 7476 |
+
"learning_rate": 2.122667077907731e-06,
|
| 7477 |
+
"loss": 0.7064,
|
| 7478 |
+
"step": 1067
|
| 7479 |
+
},
|
| 7480 |
+
{
|
| 7481 |
+
"epoch": 0.9083563682755688,
|
| 7482 |
+
"grad_norm": 0.31232479214668274,
|
| 7483 |
+
"learning_rate": 2.0841693522348218e-06,
|
| 7484 |
+
"loss": 0.6281,
|
| 7485 |
+
"step": 1068
|
| 7486 |
+
},
|
| 7487 |
+
{
|
| 7488 |
+
"epoch": 0.909206889219647,
|
| 7489 |
+
"grad_norm": 0.3088105320930481,
|
| 7490 |
+
"learning_rate": 2.0460165039462686e-06,
|
| 7491 |
+
"loss": 0.6261,
|
| 7492 |
+
"step": 1069
|
| 7493 |
+
},
|
| 7494 |
+
{
|
| 7495 |
+
"epoch": 0.9100574101637253,
|
| 7496 |
+
"grad_norm": 0.32464417815208435,
|
| 7497 |
+
"learning_rate": 2.008208807649725e-06,
|
| 7498 |
+
"loss": 0.6309,
|
| 7499 |
+
"step": 1070
|
| 7500 |
+
},
|
| 7501 |
+
{
|
| 7502 |
+
"epoch": 0.9109079311078035,
|
| 7503 |
+
"grad_norm": 0.31593137979507446,
|
| 7504 |
+
"learning_rate": 1.9707465354686205e-06,
|
| 7505 |
+
"loss": 0.6712,
|
| 7506 |
+
"step": 1071
|
| 7507 |
+
},
|
| 7508 |
+
{
|
| 7509 |
+
"epoch": 0.9117584520518818,
|
| 7510 |
+
"grad_norm": 0.31863364577293396,
|
| 7511 |
+
"learning_rate": 1.9336299570401393e-06,
|
| 7512 |
+
"loss": 0.6461,
|
| 7513 |
+
"step": 1072
|
| 7514 |
+
},
|
| 7515 |
+
{
|
| 7516 |
+
"epoch": 0.91260897299596,
|
| 7517 |
+
"grad_norm": 0.31302958726882935,
|
| 7518 |
+
"learning_rate": 1.8968593395133405e-06,
|
| 7519 |
+
"loss": 0.6243,
|
| 7520 |
+
"step": 1073
|
| 7521 |
+
},
|
| 7522 |
+
{
|
| 7523 |
+
"epoch": 0.9134594939400382,
|
| 7524 |
+
"grad_norm": 0.3122914433479309,
|
| 7525 |
+
"learning_rate": 1.8604349475471905e-06,
|
| 7526 |
+
"loss": 0.6707,
|
| 7527 |
+
"step": 1074
|
| 7528 |
+
},
|
| 7529 |
+
{
|
| 7530 |
+
"epoch": 0.9143100148841166,
|
| 7531 |
+
"grad_norm": 0.329847514629364,
|
| 7532 |
+
"learning_rate": 1.8243570433086831e-06,
|
| 7533 |
+
"loss": 0.7134,
|
| 7534 |
+
"step": 1075
|
| 7535 |
+
},
|
| 7536 |
+
{
|
| 7537 |
+
"epoch": 0.9151605358281948,
|
| 7538 |
+
"grad_norm": 0.3198718726634979,
|
| 7539 |
+
"learning_rate": 1.7886258864709239e-06,
|
| 7540 |
+
"loss": 0.6754,
|
| 7541 |
+
"step": 1076
|
| 7542 |
+
},
|
| 7543 |
+
{
|
| 7544 |
+
"epoch": 0.916011056772273,
|
| 7545 |
+
"grad_norm": 0.33064860105514526,
|
| 7546 |
+
"learning_rate": 1.753241734211314e-06,
|
| 7547 |
+
"loss": 0.6347,
|
| 7548 |
+
"step": 1077
|
| 7549 |
+
},
|
| 7550 |
+
{
|
| 7551 |
+
"epoch": 0.9168615777163512,
|
| 7552 |
+
"grad_norm": 0.322177529335022,
|
| 7553 |
+
"learning_rate": 1.7182048412096307e-06,
|
| 7554 |
+
"loss": 0.6491,
|
| 7555 |
+
"step": 1078
|
| 7556 |
+
},
|
| 7557 |
+
{
|
| 7558 |
+
"epoch": 0.9177120986604295,
|
| 7559 |
+
"grad_norm": 0.3152576982975006,
|
| 7560 |
+
"learning_rate": 1.6835154596462621e-06,
|
| 7561 |
+
"loss": 0.6448,
|
| 7562 |
+
"step": 1079
|
| 7563 |
+
},
|
| 7564 |
+
{
|
| 7565 |
+
"epoch": 0.9185626196045078,
|
| 7566 |
+
"grad_norm": 0.3428654074668884,
|
| 7567 |
+
"learning_rate": 1.6491738392003187e-06,
|
| 7568 |
+
"loss": 0.6187,
|
| 7569 |
+
"step": 1080
|
| 7570 |
+
},
|
| 7571 |
+
{
|
| 7572 |
+
"epoch": 0.919413140548586,
|
| 7573 |
+
"grad_norm": 0.32502835988998413,
|
| 7574 |
+
"learning_rate": 1.6151802270479355e-06,
|
| 7575 |
+
"loss": 0.6161,
|
| 7576 |
+
"step": 1081
|
| 7577 |
+
},
|
| 7578 |
+
{
|
| 7579 |
+
"epoch": 0.9202636614926643,
|
| 7580 |
+
"grad_norm": 0.3320343792438507,
|
| 7581 |
+
"learning_rate": 1.5815348678603792e-06,
|
| 7582 |
+
"loss": 0.667,
|
| 7583 |
+
"step": 1082
|
| 7584 |
+
},
|
| 7585 |
+
{
|
| 7586 |
+
"epoch": 0.9211141824367425,
|
| 7587 |
+
"grad_norm": 0.3216160535812378,
|
| 7588 |
+
"learning_rate": 1.548238003802377e-06,
|
| 7589 |
+
"loss": 0.6874,
|
| 7590 |
+
"step": 1083
|
| 7591 |
+
},
|
| 7592 |
+
{
|
| 7593 |
+
"epoch": 0.9219647033808207,
|
| 7594 |
+
"grad_norm": 0.325003445148468,
|
| 7595 |
+
"learning_rate": 1.5152898745303123e-06,
|
| 7596 |
+
"loss": 0.6607,
|
| 7597 |
+
"step": 1084
|
| 7598 |
+
},
|
| 7599 |
+
{
|
| 7600 |
+
"epoch": 0.922815224324899,
|
| 7601 |
+
"grad_norm": 0.3372570872306824,
|
| 7602 |
+
"learning_rate": 1.4826907171905546e-06,
|
| 7603 |
+
"loss": 0.6605,
|
| 7604 |
+
"step": 1085
|
| 7605 |
+
},
|
| 7606 |
+
{
|
| 7607 |
+
"epoch": 0.9236657452689773,
|
| 7608 |
+
"grad_norm": 0.3317967653274536,
|
| 7609 |
+
"learning_rate": 1.45044076641771e-06,
|
| 7610 |
+
"loss": 0.6208,
|
| 7611 |
+
"step": 1086
|
| 7612 |
+
},
|
| 7613 |
+
{
|
| 7614 |
+
"epoch": 0.9245162662130555,
|
| 7615 |
+
"grad_norm": 0.33367931842803955,
|
| 7616 |
+
"learning_rate": 1.4185402543329396e-06,
|
| 7617 |
+
"loss": 0.6562,
|
| 7618 |
+
"step": 1087
|
| 7619 |
+
},
|
| 7620 |
+
{
|
| 7621 |
+
"epoch": 0.9253667871571337,
|
| 7622 |
+
"grad_norm": 0.31913578510284424,
|
| 7623 |
+
"learning_rate": 1.3869894105423109e-06,
|
| 7624 |
+
"loss": 0.6753,
|
| 7625 |
+
"step": 1088
|
| 7626 |
+
},
|
| 7627 |
+
{
|
| 7628 |
+
"epoch": 0.926217308101212,
|
| 7629 |
+
"grad_norm": 0.3329998552799225,
|
| 7630 |
+
"learning_rate": 1.3557884621351213e-06,
|
| 7631 |
+
"loss": 0.6742,
|
| 7632 |
+
"step": 1089
|
| 7633 |
+
},
|
| 7634 |
+
{
|
| 7635 |
+
"epoch": 0.9270678290452903,
|
| 7636 |
+
"grad_norm": 0.3438052833080292,
|
| 7637 |
+
"learning_rate": 1.3249376336822883e-06,
|
| 7638 |
+
"loss": 0.6314,
|
| 7639 |
+
"step": 1090
|
| 7640 |
+
},
|
| 7641 |
+
{
|
| 7642 |
+
"epoch": 0.9279183499893685,
|
| 7643 |
+
"grad_norm": 0.3445602357387543,
|
| 7644 |
+
"learning_rate": 1.2944371472346838e-06,
|
| 7645 |
+
"loss": 0.6325,
|
| 7646 |
+
"step": 1091
|
| 7647 |
+
},
|
| 7648 |
+
{
|
| 7649 |
+
"epoch": 0.9287688709334467,
|
| 7650 |
+
"grad_norm": 0.3479834794998169,
|
| 7651 |
+
"learning_rate": 1.264287222321614e-06,
|
| 7652 |
+
"loss": 0.6958,
|
| 7653 |
+
"step": 1092
|
| 7654 |
+
},
|
| 7655 |
+
{
|
| 7656 |
+
"epoch": 0.929619391877525,
|
| 7657 |
+
"grad_norm": 0.34561192989349365,
|
| 7658 |
+
"learning_rate": 1.2344880759491585e-06,
|
| 7659 |
+
"loss": 0.6485,
|
| 7660 |
+
"step": 1093
|
| 7661 |
+
},
|
| 7662 |
+
{
|
| 7663 |
+
"epoch": 0.9304699128216032,
|
| 7664 |
+
"grad_norm": 0.32325616478919983,
|
| 7665 |
+
"learning_rate": 1.205039922598672e-06,
|
| 7666 |
+
"loss": 0.6593,
|
| 7667 |
+
"step": 1094
|
| 7668 |
+
},
|
| 7669 |
+
{
|
| 7670 |
+
"epoch": 0.9313204337656815,
|
| 7671 |
+
"grad_norm": 0.37606281042099,
|
| 7672 |
+
"learning_rate": 1.1759429742251749e-06,
|
| 7673 |
+
"loss": 0.6598,
|
| 7674 |
+
"step": 1095
|
| 7675 |
+
},
|
| 7676 |
+
{
|
| 7677 |
+
"epoch": 0.9321709547097597,
|
| 7678 |
+
"grad_norm": 0.33404406905174255,
|
| 7679 |
+
"learning_rate": 1.147197440255926e-06,
|
| 7680 |
+
"loss": 0.642,
|
| 7681 |
+
"step": 1096
|
| 7682 |
+
},
|
| 7683 |
+
{
|
| 7684 |
+
"epoch": 0.933021475653838,
|
| 7685 |
+
"grad_norm": 0.3421410322189331,
|
| 7686 |
+
"learning_rate": 1.1188035275887965e-06,
|
| 7687 |
+
"loss": 0.6384,
|
| 7688 |
+
"step": 1097
|
| 7689 |
+
},
|
| 7690 |
+
{
|
| 7691 |
+
"epoch": 0.9338719965979162,
|
| 7692 |
+
"grad_norm": 0.35476282238960266,
|
| 7693 |
+
"learning_rate": 1.0907614405908816e-06,
|
| 7694 |
+
"loss": 0.6603,
|
| 7695 |
+
"step": 1098
|
| 7696 |
+
},
|
| 7697 |
+
{
|
| 7698 |
+
"epoch": 0.9347225175419944,
|
| 7699 |
+
"grad_norm": 0.37484419345855713,
|
| 7700 |
+
"learning_rate": 1.063071381096964e-06,
|
| 7701 |
+
"loss": 0.5992,
|
| 7702 |
+
"step": 1099
|
| 7703 |
+
},
|
| 7704 |
+
{
|
| 7705 |
+
"epoch": 0.9355730384860728,
|
| 7706 |
+
"grad_norm": 0.5027537941932678,
|
| 7707 |
+
"learning_rate": 1.035733548408091e-06,
|
| 7708 |
+
"loss": 0.6615,
|
| 7709 |
+
"step": 1100
|
| 7710 |
+
},
|
| 7711 |
+
{
|
| 7712 |
+
"epoch": 0.936423559430151,
|
| 7713 |
+
"grad_norm": 0.254710853099823,
|
| 7714 |
+
"learning_rate": 1.0087481392901331e-06,
|
| 7715 |
+
"loss": 0.6408,
|
| 7716 |
+
"step": 1101
|
| 7717 |
+
},
|
| 7718 |
+
{
|
| 7719 |
+
"epoch": 0.9372740803742292,
|
| 7720 |
+
"grad_norm": 0.27126064896583557,
|
| 7721 |
+
"learning_rate": 9.821153479723788e-07,
|
| 7722 |
+
"loss": 0.606,
|
| 7723 |
+
"step": 1102
|
| 7724 |
+
},
|
| 7725 |
+
{
|
| 7726 |
+
"epoch": 0.9381246013183074,
|
| 7727 |
+
"grad_norm": 0.2831709086894989,
|
| 7728 |
+
"learning_rate": 9.558353661461072e-07,
|
| 7729 |
+
"loss": 0.634,
|
| 7730 |
+
"step": 1103
|
| 7731 |
+
},
|
| 7732 |
+
{
|
| 7733 |
+
"epoch": 0.9389751222623857,
|
| 7734 |
+
"grad_norm": 0.30248603224754333,
|
| 7735 |
+
"learning_rate": 9.299083829632516e-07,
|
| 7736 |
+
"loss": 0.7414,
|
| 7737 |
+
"step": 1104
|
| 7738 |
+
},
|
| 7739 |
+
{
|
| 7740 |
+
"epoch": 0.939825643206464,
|
| 7741 |
+
"grad_norm": 0.2874879539012909,
|
| 7742 |
+
"learning_rate": 9.043345850349939e-07,
|
| 7743 |
+
"loss": 0.6089,
|
| 7744 |
+
"step": 1105
|
| 7745 |
+
},
|
| 7746 |
+
{
|
| 7747 |
+
"epoch": 0.9406761641505422,
|
| 7748 |
+
"grad_norm": 0.28575336933135986,
|
| 7749 |
+
"learning_rate": 8.791141564304495e-07,
|
| 7750 |
+
"loss": 0.6352,
|
| 7751 |
+
"step": 1106
|
| 7752 |
+
},
|
| 7753 |
+
{
|
| 7754 |
+
"epoch": 0.9415266850946205,
|
| 7755 |
+
"grad_norm": 0.29282501339912415,
|
| 7756 |
+
"learning_rate": 8.542472786753408e-07,
|
| 7757 |
+
"loss": 0.651,
|
| 7758 |
+
"step": 1107
|
| 7759 |
+
},
|
| 7760 |
+
{
|
| 7761 |
+
"epoch": 0.9423772060386987,
|
| 7762 |
+
"grad_norm": 0.2857988476753235,
|
| 7763 |
+
"learning_rate": 8.297341307506645e-07,
|
| 7764 |
+
"loss": 0.612,
|
| 7765 |
+
"step": 1108
|
| 7766 |
+
},
|
| 7767 |
+
{
|
| 7768 |
+
"epoch": 0.9432277269827769,
|
| 7769 |
+
"grad_norm": 0.28160417079925537,
|
| 7770 |
+
"learning_rate": 8.055748890914483e-07,
|
| 7771 |
+
"loss": 0.6208,
|
| 7772 |
+
"step": 1109
|
| 7773 |
+
},
|
| 7774 |
+
{
|
| 7775 |
+
"epoch": 0.9440782479268552,
|
| 7776 |
+
"grad_norm": 0.27699196338653564,
|
| 7777 |
+
"learning_rate": 7.817697275854296e-07,
|
| 7778 |
+
"loss": 0.5592,
|
| 7779 |
+
"step": 1110
|
| 7780 |
+
},
|
| 7781 |
+
{
|
| 7782 |
+
"epoch": 0.9449287688709335,
|
| 7783 |
+
"grad_norm": 0.3014472424983978,
|
| 7784 |
+
"learning_rate": 7.583188175718625e-07,
|
| 7785 |
+
"loss": 0.6837,
|
| 7786 |
+
"step": 1111
|
| 7787 |
+
},
|
| 7788 |
+
{
|
| 7789 |
+
"epoch": 0.9457792898150117,
|
| 7790 |
+
"grad_norm": 0.31287431716918945,
|
| 7791 |
+
"learning_rate": 7.352223278402292e-07,
|
| 7792 |
+
"loss": 0.6498,
|
| 7793 |
+
"step": 1112
|
| 7794 |
+
},
|
| 7795 |
+
{
|
| 7796 |
+
"epoch": 0.9466298107590899,
|
| 7797 |
+
"grad_norm": 0.30627578496932983,
|
| 7798 |
+
"learning_rate": 7.124804246290473e-07,
|
| 7799 |
+
"loss": 0.6542,
|
| 7800 |
+
"step": 1113
|
| 7801 |
+
},
|
| 7802 |
+
{
|
| 7803 |
+
"epoch": 0.9474803317031681,
|
| 7804 |
+
"grad_norm": 0.29300999641418457,
|
| 7805 |
+
"learning_rate": 6.900932716246866e-07,
|
| 7806 |
+
"loss": 0.6397,
|
| 7807 |
+
"step": 1114
|
| 7808 |
+
},
|
| 7809 |
+
{
|
| 7810 |
+
"epoch": 0.9483308526472465,
|
| 7811 |
+
"grad_norm": 0.29144996404647827,
|
| 7812 |
+
"learning_rate": 6.680610299601708e-07,
|
| 7813 |
+
"loss": 0.6867,
|
| 7814 |
+
"step": 1115
|
| 7815 |
+
},
|
| 7816 |
+
{
|
| 7817 |
+
"epoch": 0.9491813735913247,
|
| 7818 |
+
"grad_norm": 0.3168846666812897,
|
| 7819 |
+
"learning_rate": 6.46383858214028e-07,
|
| 7820 |
+
"loss": 0.6927,
|
| 7821 |
+
"step": 1116
|
| 7822 |
+
},
|
| 7823 |
+
{
|
| 7824 |
+
"epoch": 0.9500318945354029,
|
| 7825 |
+
"grad_norm": 0.31854063272476196,
|
| 7826 |
+
"learning_rate": 6.250619124091473e-07,
|
| 7827 |
+
"loss": 0.6628,
|
| 7828 |
+
"step": 1117
|
| 7829 |
+
},
|
| 7830 |
+
{
|
| 7831 |
+
"epoch": 0.9508824154794812,
|
| 7832 |
+
"grad_norm": 0.30669549107551575,
|
| 7833 |
+
"learning_rate": 6.040953460116516e-07,
|
| 7834 |
+
"loss": 0.6212,
|
| 7835 |
+
"step": 1118
|
| 7836 |
+
},
|
| 7837 |
+
{
|
| 7838 |
+
"epoch": 0.9517329364235594,
|
| 7839 |
+
"grad_norm": 0.308658242225647,
|
| 7840 |
+
"learning_rate": 5.834843099298048e-07,
|
| 7841 |
+
"loss": 0.6451,
|
| 7842 |
+
"step": 1119
|
| 7843 |
+
},
|
| 7844 |
+
{
|
| 7845 |
+
"epoch": 0.9525834573676377,
|
| 7846 |
+
"grad_norm": 0.3079119324684143,
|
| 7847 |
+
"learning_rate": 5.632289525129065e-07,
|
| 7848 |
+
"loss": 0.652,
|
| 7849 |
+
"step": 1120
|
| 7850 |
+
},
|
| 7851 |
+
{
|
| 7852 |
+
"epoch": 0.953433978311716,
|
| 7853 |
+
"grad_norm": 0.2991427779197693,
|
| 7854 |
+
"learning_rate": 5.433294195502481e-07,
|
| 7855 |
+
"loss": 0.6479,
|
| 7856 |
+
"step": 1121
|
| 7857 |
+
},
|
| 7858 |
+
{
|
| 7859 |
+
"epoch": 0.9542844992557942,
|
| 7860 |
+
"grad_norm": 0.31203827261924744,
|
| 7861 |
+
"learning_rate": 5.237858542700425e-07,
|
| 7862 |
+
"loss": 0.6435,
|
| 7863 |
+
"step": 1122
|
| 7864 |
+
},
|
| 7865 |
+
{
|
| 7866 |
+
"epoch": 0.9551350201998724,
|
| 7867 |
+
"grad_norm": 0.3181011378765106,
|
| 7868 |
+
"learning_rate": 5.045983973383961e-07,
|
| 7869 |
+
"loss": 0.6504,
|
| 7870 |
+
"step": 1123
|
| 7871 |
+
},
|
| 7872 |
+
{
|
| 7873 |
+
"epoch": 0.9559855411439506,
|
| 7874 |
+
"grad_norm": 0.3141401708126068,
|
| 7875 |
+
"learning_rate": 4.857671868583158e-07,
|
| 7876 |
+
"loss": 0.6495,
|
| 7877 |
+
"step": 1124
|
| 7878 |
+
},
|
| 7879 |
+
{
|
| 7880 |
+
"epoch": 0.956836062088029,
|
| 7881 |
+
"grad_norm": 0.3229545056819916,
|
| 7882 |
+
"learning_rate": 4.6729235836869277e-07,
|
| 7883 |
+
"loss": 0.64,
|
| 7884 |
+
"step": 1125
|
| 7885 |
+
},
|
| 7886 |
+
{
|
| 7887 |
+
"epoch": 0.9576865830321072,
|
| 7888 |
+
"grad_norm": 0.3335203230381012,
|
| 7889 |
+
"learning_rate": 4.4917404484334237e-07,
|
| 7890 |
+
"loss": 0.6703,
|
| 7891 |
+
"step": 1126
|
| 7892 |
+
},
|
| 7893 |
+
{
|
| 7894 |
+
"epoch": 0.9585371039761854,
|
| 7895 |
+
"grad_norm": 0.311173677444458,
|
| 7896 |
+
"learning_rate": 4.314123766900213e-07,
|
| 7897 |
+
"loss": 0.6414,
|
| 7898 |
+
"step": 1127
|
| 7899 |
+
},
|
| 7900 |
+
{
|
| 7901 |
+
"epoch": 0.9593876249202636,
|
| 7902 |
+
"grad_norm": 0.3213408887386322,
|
| 7903 |
+
"learning_rate": 4.1400748174952877e-07,
|
| 7904 |
+
"loss": 0.6878,
|
| 7905 |
+
"step": 1128
|
| 7906 |
+
},
|
| 7907 |
+
{
|
| 7908 |
+
"epoch": 0.9602381458643419,
|
| 7909 |
+
"grad_norm": 0.32059377431869507,
|
| 7910 |
+
"learning_rate": 3.969594852947511e-07,
|
| 7911 |
+
"loss": 0.6144,
|
| 7912 |
+
"step": 1129
|
| 7913 |
+
},
|
| 7914 |
+
{
|
| 7915 |
+
"epoch": 0.9610886668084202,
|
| 7916 |
+
"grad_norm": 0.32189103960990906,
|
| 7917 |
+
"learning_rate": 3.802685100297798e-07,
|
| 7918 |
+
"loss": 0.6547,
|
| 7919 |
+
"step": 1130
|
| 7920 |
+
},
|
| 7921 |
+
{
|
| 7922 |
+
"epoch": 0.9619391877524984,
|
| 7923 |
+
"grad_norm": 0.3400660455226898,
|
| 7924 |
+
"learning_rate": 3.639346760890283e-07,
|
| 7925 |
+
"loss": 0.6598,
|
| 7926 |
+
"step": 1131
|
| 7927 |
+
},
|
| 7928 |
+
{
|
| 7929 |
+
"epoch": 0.9627897086965767,
|
| 7930 |
+
"grad_norm": 0.3212508261203766,
|
| 7931 |
+
"learning_rate": 3.4795810103634417e-07,
|
| 7932 |
+
"loss": 0.6392,
|
| 7933 |
+
"step": 1132
|
| 7934 |
+
},
|
| 7935 |
+
{
|
| 7936 |
+
"epoch": 0.9636402296406549,
|
| 7937 |
+
"grad_norm": 0.31696587800979614,
|
| 7938 |
+
"learning_rate": 3.3233889986419855e-07,
|
| 7939 |
+
"loss": 0.6333,
|
| 7940 |
+
"step": 1133
|
| 7941 |
+
},
|
| 7942 |
+
{
|
| 7943 |
+
"epoch": 0.9644907505847331,
|
| 7944 |
+
"grad_norm": 0.3382854461669922,
|
| 7945 |
+
"learning_rate": 3.170771849928367e-07,
|
| 7946 |
+
"loss": 0.6498,
|
| 7947 |
+
"step": 1134
|
| 7948 |
+
},
|
| 7949 |
+
{
|
| 7950 |
+
"epoch": 0.9653412715288114,
|
| 7951 |
+
"grad_norm": 0.3380298316478729,
|
| 7952 |
+
"learning_rate": 3.021730662694511e-07,
|
| 7953 |
+
"loss": 0.6686,
|
| 7954 |
+
"step": 1135
|
| 7955 |
+
},
|
| 7956 |
+
{
|
| 7957 |
+
"epoch": 0.9661917924728897,
|
| 7958 |
+
"grad_norm": 0.3369327783584595,
|
| 7959 |
+
"learning_rate": 2.8762665096744854e-07,
|
| 7960 |
+
"loss": 0.643,
|
| 7961 |
+
"step": 1136
|
| 7962 |
+
},
|
| 7963 |
+
{
|
| 7964 |
+
"epoch": 0.9670423134169679,
|
| 7965 |
+
"grad_norm": 0.3265814483165741,
|
| 7966 |
+
"learning_rate": 2.734380437856121e-07,
|
| 7967 |
+
"loss": 0.6238,
|
| 7968 |
+
"step": 1137
|
| 7969 |
+
},
|
| 7970 |
+
{
|
| 7971 |
+
"epoch": 0.9678928343610461,
|
| 7972 |
+
"grad_norm": 0.33164817094802856,
|
| 7973 |
+
"learning_rate": 2.5960734684739584e-07,
|
| 7974 |
+
"loss": 0.6256,
|
| 7975 |
+
"step": 1138
|
| 7976 |
+
},
|
| 7977 |
+
{
|
| 7978 |
+
"epoch": 0.9687433553051243,
|
| 7979 |
+
"grad_norm": 0.3356388807296753,
|
| 7980 |
+
"learning_rate": 2.4613465970015347e-07,
|
| 7981 |
+
"loss": 0.6595,
|
| 7982 |
+
"step": 1139
|
| 7983 |
+
},
|
| 7984 |
+
{
|
| 7985 |
+
"epoch": 0.9695938762492027,
|
| 7986 |
+
"grad_norm": 0.34756919741630554,
|
| 7987 |
+
"learning_rate": 2.3302007931444992e-07,
|
| 7988 |
+
"loss": 0.6228,
|
| 7989 |
+
"step": 1140
|
| 7990 |
+
},
|
| 7991 |
+
{
|
| 7992 |
+
"epoch": 0.9704443971932809,
|
| 7993 |
+
"grad_norm": 0.32481443881988525,
|
| 7994 |
+
"learning_rate": 2.202637000833563e-07,
|
| 7995 |
+
"loss": 0.6224,
|
| 7996 |
+
"step": 1141
|
| 7997 |
+
},
|
| 7998 |
+
{
|
| 7999 |
+
"epoch": 0.9712949181373591,
|
| 8000 |
+
"grad_norm": 0.3473244607448578,
|
| 8001 |
+
"learning_rate": 2.0786561382176163e-07,
|
| 8002 |
+
"loss": 0.6656,
|
| 8003 |
+
"step": 1142
|
| 8004 |
+
},
|
| 8005 |
+
{
|
| 8006 |
+
"epoch": 0.9721454390814374,
|
| 8007 |
+
"grad_norm": 0.36506223678588867,
|
| 8008 |
+
"learning_rate": 1.9582590976571224e-07,
|
| 8009 |
+
"loss": 0.6685,
|
| 8010 |
+
"step": 1143
|
| 8011 |
+
},
|
| 8012 |
+
{
|
| 8013 |
+
"epoch": 0.9729959600255156,
|
| 8014 |
+
"grad_norm": 0.36091652512550354,
|
| 8015 |
+
"learning_rate": 1.841446745717901e-07,
|
| 8016 |
+
"loss": 0.6762,
|
| 8017 |
+
"step": 1144
|
| 8018 |
+
},
|
| 8019 |
+
{
|
| 8020 |
+
"epoch": 0.9738464809695939,
|
| 8021 |
+
"grad_norm": 0.3407830595970154,
|
| 8022 |
+
"learning_rate": 1.7282199231645758e-07,
|
| 8023 |
+
"loss": 0.6402,
|
| 8024 |
+
"step": 1145
|
| 8025 |
+
},
|
| 8026 |
+
{
|
| 8027 |
+
"epoch": 0.9746970019136721,
|
| 8028 |
+
"grad_norm": 0.3659832775592804,
|
| 8029 |
+
"learning_rate": 1.618579444954693e-07,
|
| 8030 |
+
"loss": 0.6457,
|
| 8031 |
+
"step": 1146
|
| 8032 |
+
},
|
| 8033 |
+
{
|
| 8034 |
+
"epoch": 0.9755475228577504,
|
| 8035 |
+
"grad_norm": 0.3384697139263153,
|
| 8036 |
+
"learning_rate": 1.5125261002330026e-07,
|
| 8037 |
+
"loss": 0.6015,
|
| 8038 |
+
"step": 1147
|
| 8039 |
+
},
|
| 8040 |
+
{
|
| 8041 |
+
"epoch": 0.9763980438018286,
|
| 8042 |
+
"grad_norm": 0.3543066680431366,
|
| 8043 |
+
"learning_rate": 1.4100606523255177e-07,
|
| 8044 |
+
"loss": 0.6212,
|
| 8045 |
+
"step": 1148
|
| 8046 |
+
},
|
| 8047 |
+
{
|
| 8048 |
+
"epoch": 0.9772485647459068,
|
| 8049 |
+
"grad_norm": 0.3659920394420624,
|
| 8050 |
+
"learning_rate": 1.3111838387340758e-07,
|
| 8051 |
+
"loss": 0.6027,
|
| 8052 |
+
"step": 1149
|
| 8053 |
+
},
|
| 8054 |
+
{
|
| 8055 |
+
"epoch": 0.9780990856899852,
|
| 8056 |
+
"grad_norm": 0.47490641474723816,
|
| 8057 |
+
"learning_rate": 1.21589637113112e-07,
|
| 8058 |
+
"loss": 0.5485,
|
| 8059 |
+
"step": 1150
|
| 8060 |
+
},
|
| 8061 |
+
{
|
| 8062 |
+
"epoch": 0.9789496066340634,
|
| 8063 |
+
"grad_norm": 0.26568692922592163,
|
| 8064 |
+
"learning_rate": 1.1241989353545923e-07,
|
| 8065 |
+
"loss": 0.6297,
|
| 8066 |
+
"step": 1151
|
| 8067 |
+
},
|
| 8068 |
+
{
|
| 8069 |
+
"epoch": 0.9798001275781416,
|
| 8070 |
+
"grad_norm": 0.2851766347885132,
|
| 8071 |
+
"learning_rate": 1.036092191402882e-07,
|
| 8072 |
+
"loss": 0.6266,
|
| 8073 |
+
"step": 1152
|
| 8074 |
+
},
|
| 8075 |
+
{
|
| 8076 |
+
"epoch": 0.9806506485222198,
|
| 8077 |
+
"grad_norm": 0.2778666913509369,
|
| 8078 |
+
"learning_rate": 9.515767734302184e-08,
|
| 8079 |
+
"loss": 0.621,
|
| 8080 |
+
"step": 1153
|
| 8081 |
+
},
|
| 8082 |
+
{
|
| 8083 |
+
"epoch": 0.9815011694662981,
|
| 8084 |
+
"grad_norm": 0.2930733859539032,
|
| 8085 |
+
"learning_rate": 8.706532897417851e-08,
|
| 8086 |
+
"loss": 0.6171,
|
| 8087 |
+
"step": 1154
|
| 8088 |
+
},
|
| 8089 |
+
{
|
| 8090 |
+
"epoch": 0.9823516904103764,
|
| 8091 |
+
"grad_norm": 0.2878072261810303,
|
| 8092 |
+
"learning_rate": 7.933223227898911e-08,
|
| 8093 |
+
"loss": 0.6623,
|
| 8094 |
+
"step": 1155
|
| 8095 |
+
},
|
| 8096 |
+
{
|
| 8097 |
+
"epoch": 0.9832022113544546,
|
| 8098 |
+
"grad_norm": 0.2896723747253418,
|
| 8099 |
+
"learning_rate": 7.195844291693065e-08,
|
| 8100 |
+
"loss": 0.6461,
|
| 8101 |
+
"step": 1156
|
| 8102 |
+
},
|
| 8103 |
+
{
|
| 8104 |
+
"epoch": 0.9840527322985329,
|
| 8105 |
+
"grad_norm": 0.3108197748661041,
|
| 8106 |
+
"learning_rate": 6.494401396134331e-08,
|
| 8107 |
+
"loss": 0.6485,
|
| 8108 |
+
"step": 1157
|
| 8109 |
+
},
|
| 8110 |
+
{
|
| 8111 |
+
"epoch": 0.9849032532426111,
|
| 8112 |
+
"grad_norm": 0.3045186698436737,
|
| 8113 |
+
"learning_rate": 5.82889958990418e-08,
|
| 8114 |
+
"loss": 0.6488,
|
| 8115 |
+
"step": 1158
|
| 8116 |
+
},
|
| 8117 |
+
{
|
| 8118 |
+
"epoch": 0.9857537741866893,
|
| 8119 |
+
"grad_norm": 0.32476016879081726,
|
| 8120 |
+
"learning_rate": 5.1993436629965694e-08,
|
| 8121 |
+
"loss": 0.6823,
|
| 8122 |
+
"step": 1159
|
| 8123 |
+
},
|
| 8124 |
+
{
|
| 8125 |
+
"epoch": 0.9866042951307676,
|
| 8126 |
+
"grad_norm": 0.31088775396347046,
|
| 8127 |
+
"learning_rate": 4.605738146682414e-08,
|
| 8128 |
+
"loss": 0.6667,
|
| 8129 |
+
"step": 1160
|
| 8130 |
+
},
|
| 8131 |
+
{
|
| 8132 |
+
"epoch": 0.9874548160748459,
|
| 8133 |
+
"grad_norm": 0.30500373244285583,
|
| 8134 |
+
"learning_rate": 4.048087313476279e-08,
|
| 8135 |
+
"loss": 0.6505,
|
| 8136 |
+
"step": 1161
|
| 8137 |
+
},
|
| 8138 |
+
{
|
| 8139 |
+
"epoch": 0.9883053370189241,
|
| 8140 |
+
"grad_norm": 0.3153252899646759,
|
| 8141 |
+
"learning_rate": 3.526395177108066e-08,
|
| 8142 |
+
"loss": 0.6084,
|
| 8143 |
+
"step": 1162
|
| 8144 |
+
},
|
| 8145 |
+
{
|
| 8146 |
+
"epoch": 0.9891558579630023,
|
| 8147 |
+
"grad_norm": 0.32147717475891113,
|
| 8148 |
+
"learning_rate": 3.040665492491379e-08,
|
| 8149 |
+
"loss": 0.6466,
|
| 8150 |
+
"step": 1163
|
| 8151 |
+
},
|
| 8152 |
+
{
|
| 8153 |
+
"epoch": 0.9900063789070805,
|
| 8154 |
+
"grad_norm": 0.3214706778526306,
|
| 8155 |
+
"learning_rate": 2.590901755697983e-08,
|
| 8156 |
+
"loss": 0.6246,
|
| 8157 |
+
"step": 1164
|
| 8158 |
+
},
|
| 8159 |
+
{
|
| 8160 |
+
"epoch": 0.9908568998511589,
|
| 8161 |
+
"grad_norm": 0.3158264458179474,
|
| 8162 |
+
"learning_rate": 2.1771072039322714e-08,
|
| 8163 |
+
"loss": 0.6216,
|
| 8164 |
+
"step": 1165
|
| 8165 |
+
},
|
| 8166 |
+
{
|
| 8167 |
+
"epoch": 0.9917074207952371,
|
| 8168 |
+
"grad_norm": 0.32929104566574097,
|
| 8169 |
+
"learning_rate": 1.7992848155079512e-08,
|
| 8170 |
+
"loss": 0.644,
|
| 8171 |
+
"step": 1166
|
| 8172 |
+
},
|
| 8173 |
+
{
|
| 8174 |
+
"epoch": 0.9925579417393153,
|
| 8175 |
+
"grad_norm": 0.3237769305706024,
|
| 8176 |
+
"learning_rate": 1.4574373098275029e-08,
|
| 8177 |
+
"loss": 0.6019,
|
| 8178 |
+
"step": 1167
|
| 8179 |
+
},
|
| 8180 |
+
{
|
| 8181 |
+
"epoch": 0.9934084626833936,
|
| 8182 |
+
"grad_norm": 0.3426662087440491,
|
| 8183 |
+
"learning_rate": 1.1515671473599776e-08,
|
| 8184 |
+
"loss": 0.6455,
|
| 8185 |
+
"step": 1168
|
| 8186 |
+
},
|
| 8187 |
+
{
|
| 8188 |
+
"epoch": 0.9942589836274718,
|
| 8189 |
+
"grad_norm": 0.3314763903617859,
|
| 8190 |
+
"learning_rate": 8.816765296271179e-09,
|
| 8191 |
+
"loss": 0.6551,
|
| 8192 |
+
"step": 1169
|
| 8193 |
+
},
|
| 8194 |
+
{
|
| 8195 |
+
"epoch": 0.9951095045715501,
|
| 8196 |
+
"grad_norm": 0.33676841855049133,
|
| 8197 |
+
"learning_rate": 6.477673991850397e-09,
|
| 8198 |
+
"loss": 0.6562,
|
| 8199 |
+
"step": 1170
|
| 8200 |
+
},
|
| 8201 |
+
{
|
| 8202 |
+
"epoch": 0.9959600255156283,
|
| 8203 |
+
"grad_norm": 0.34772470593452454,
|
| 8204 |
+
"learning_rate": 4.498414396092443e-09,
|
| 8205 |
+
"loss": 0.6711,
|
| 8206 |
+
"step": 1171
|
| 8207 |
+
},
|
| 8208 |
+
{
|
| 8209 |
+
"epoch": 0.9968105464597066,
|
| 8210 |
+
"grad_norm": 0.3249342143535614,
|
| 8211 |
+
"learning_rate": 2.879000754851813e-09,
|
| 8212 |
+
"loss": 0.5821,
|
| 8213 |
+
"step": 1172
|
| 8214 |
+
},
|
| 8215 |
+
{
|
| 8216 |
+
"epoch": 0.9976610674037848,
|
| 8217 |
+
"grad_norm": 0.3405115008354187,
|
| 8218 |
+
"learning_rate": 1.6194447239714638e-09,
|
| 8219 |
+
"loss": 0.6668,
|
| 8220 |
+
"step": 1173
|
| 8221 |
+
},
|
| 8222 |
+
{
|
| 8223 |
+
"epoch": 0.998511588347863,
|
| 8224 |
+
"grad_norm": 0.35243573784828186,
|
| 8225 |
+
"learning_rate": 7.197553691939973e-10,
|
| 8226 |
+
"loss": 0.6637,
|
| 8227 |
+
"step": 1174
|
| 8228 |
+
},
|
| 8229 |
+
{
|
| 8230 |
+
"epoch": 0.9993621092919414,
|
| 8231 |
+
"grad_norm": 0.3634124994277954,
|
| 8232 |
+
"learning_rate": 1.7993916607839113e-10,
|
| 8233 |
+
"loss": 0.5855,
|
| 8234 |
+
"step": 1175
|
| 8235 |
+
},
|
| 8236 |
+
{
|
| 8237 |
+
"epoch": 0.9993621092919414,
|
| 8238 |
+
"eval_loss": 0.6406263709068298,
|
| 8239 |
+
"eval_runtime": 23.3139,
|
| 8240 |
+
"eval_samples_per_second": 42.464,
|
| 8241 |
+
"eval_steps_per_second": 10.637,
|
| 8242 |
+
"step": 1175
|
| 8243 |
+
},
|
| 8244 |
+
{
|
| 8245 |
+
"epoch": 1.0004252604720392,
|
| 8246 |
+
"grad_norm": 0.9211934804916382,
|
| 8247 |
+
"learning_rate": 0.0,
|
| 8248 |
+
"loss": 0.9193,
|
| 8249 |
+
"step": 1176
|
| 8250 |
}
|
| 8251 |
],
|
| 8252 |
"logging_steps": 1,
|
|
|
|
| 8261 |
"should_evaluate": false,
|
| 8262 |
"should_log": false,
|
| 8263 |
"should_save": true,
|
| 8264 |
+
"should_training_stop": true
|
| 8265 |
},
|
| 8266 |
"attributes": {}
|
| 8267 |
}
|
| 8268 |
},
|
| 8269 |
+
"total_flos": 1.5271204993420493e+17,
|
| 8270 |
"train_batch_size": 4,
|
| 8271 |
"trial_name": null,
|
| 8272 |
"trial_params": null
|