Training in progress, step 999, 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 48679352
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9cca9f27bcd2b1b3f2781ff3348caff7dfdc338608e16d5c986672579a69edd5
|
3 |
size 48679352
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 25152884
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:654f5ad232053c56b03c0dff973796ed6f3efd97c27655f3b61e2051fcd9b288
|
3 |
size 25152884
|
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:a3e495d9430e77a9a8235a79020af44fedd824541bf26d3b88fae05d6a5811e3
|
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 sha256:83c8739da06509b17311b3ff617932a8ea714dea69a235792df21c2f31217d11
|
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": 0.
|
5 |
"eval_steps": 500,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -4669,6 +4669,2337 @@
|
|
4669 |
"learning_rate": 5.0177825624378114e-05,
|
4670 |
"loss": 0.9829,
|
4671 |
"step": 666
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4672 |
}
|
4673 |
],
|
4674 |
"logging_steps": 1,
|
@@ -4688,7 +7019,7 @@
|
|
4688 |
"attributes": {}
|
4689 |
}
|
4690 |
},
|
4691 |
-
"total_flos":
|
4692 |
"train_batch_size": 4,
|
4693 |
"trial_name": null,
|
4694 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.006640057693394173,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 999,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
4669 |
"learning_rate": 5.0177825624378114e-05,
|
4670 |
"loss": 0.9829,
|
4671 |
"step": 666
|
4672 |
+
},
|
4673 |
+
{
|
4674 |
+
"epoch": 0.004433351833327241,
|
4675 |
+
"grad_norm": 0.8198716044425964,
|
4676 |
+
"learning_rate": 5.0059275319202084e-05,
|
4677 |
+
"loss": 0.7453,
|
4678 |
+
"step": 667
|
4679 |
+
},
|
4680 |
+
{
|
4681 |
+
"epoch": 0.004439998537725033,
|
4682 |
+
"grad_norm": 0.6317304372787476,
|
4683 |
+
"learning_rate": 4.9940724680797914e-05,
|
4684 |
+
"loss": 0.8972,
|
4685 |
+
"step": 668
|
4686 |
+
},
|
4687 |
+
{
|
4688 |
+
"epoch": 0.004446645242122824,
|
4689 |
+
"grad_norm": 0.7842250466346741,
|
4690 |
+
"learning_rate": 4.98221743756219e-05,
|
4691 |
+
"loss": 0.7495,
|
4692 |
+
"step": 669
|
4693 |
+
},
|
4694 |
+
{
|
4695 |
+
"epoch": 0.004453291946520616,
|
4696 |
+
"grad_norm": 0.7934935092926025,
|
4697 |
+
"learning_rate": 4.970362507012848e-05,
|
4698 |
+
"loss": 0.8944,
|
4699 |
+
"step": 670
|
4700 |
+
},
|
4701 |
+
{
|
4702 |
+
"epoch": 0.004459938650918409,
|
4703 |
+
"grad_norm": 0.858690619468689,
|
4704 |
+
"learning_rate": 4.958507743076645e-05,
|
4705 |
+
"loss": 0.8814,
|
4706 |
+
"step": 671
|
4707 |
+
},
|
4708 |
+
{
|
4709 |
+
"epoch": 0.0044665853553162,
|
4710 |
+
"grad_norm": 0.7755974531173706,
|
4711 |
+
"learning_rate": 4.946653212397527e-05,
|
4712 |
+
"loss": 0.8873,
|
4713 |
+
"step": 672
|
4714 |
+
},
|
4715 |
+
{
|
4716 |
+
"epoch": 0.004473232059713992,
|
4717 |
+
"grad_norm": 0.995033860206604,
|
4718 |
+
"learning_rate": 4.934798981618127e-05,
|
4719 |
+
"loss": 0.7748,
|
4720 |
+
"step": 673
|
4721 |
+
},
|
4722 |
+
{
|
4723 |
+
"epoch": 0.004479878764111784,
|
4724 |
+
"grad_norm": 1.0964568853378296,
|
4725 |
+
"learning_rate": 4.922945117379394e-05,
|
4726 |
+
"loss": 0.629,
|
4727 |
+
"step": 674
|
4728 |
+
},
|
4729 |
+
{
|
4730 |
+
"epoch": 0.0044865254685095765,
|
4731 |
+
"grad_norm": 0.901854932308197,
|
4732 |
+
"learning_rate": 4.911091686320213e-05,
|
4733 |
+
"loss": 0.6407,
|
4734 |
+
"step": 675
|
4735 |
+
},
|
4736 |
+
{
|
4737 |
+
"epoch": 0.004493172172907368,
|
4738 |
+
"grad_norm": 0.9707088470458984,
|
4739 |
+
"learning_rate": 4.8992387550770373e-05,
|
4740 |
+
"loss": 0.8739,
|
4741 |
+
"step": 676
|
4742 |
+
},
|
4743 |
+
{
|
4744 |
+
"epoch": 0.00449981887730516,
|
4745 |
+
"grad_norm": 0.9366260766983032,
|
4746 |
+
"learning_rate": 4.887386390283508e-05,
|
4747 |
+
"loss": 0.7199,
|
4748 |
+
"step": 677
|
4749 |
+
},
|
4750 |
+
{
|
4751 |
+
"epoch": 0.004506465581702952,
|
4752 |
+
"grad_norm": 0.7978320717811584,
|
4753 |
+
"learning_rate": 4.875534658570081e-05,
|
4754 |
+
"loss": 0.735,
|
4755 |
+
"step": 678
|
4756 |
+
},
|
4757 |
+
{
|
4758 |
+
"epoch": 0.004513112286100744,
|
4759 |
+
"grad_norm": 0.917753279209137,
|
4760 |
+
"learning_rate": 4.863683626563657e-05,
|
4761 |
+
"loss": 0.6231,
|
4762 |
+
"step": 679
|
4763 |
+
},
|
4764 |
+
{
|
4765 |
+
"epoch": 0.004519758990498536,
|
4766 |
+
"grad_norm": 0.9080862402915955,
|
4767 |
+
"learning_rate": 4.851833360887201e-05,
|
4768 |
+
"loss": 0.9354,
|
4769 |
+
"step": 680
|
4770 |
+
},
|
4771 |
+
{
|
4772 |
+
"epoch": 0.004526405694896328,
|
4773 |
+
"grad_norm": 0.7667110562324524,
|
4774 |
+
"learning_rate": 4.839983928159371e-05,
|
4775 |
+
"loss": 0.7752,
|
4776 |
+
"step": 681
|
4777 |
+
},
|
4778 |
+
{
|
4779 |
+
"epoch": 0.00453305239929412,
|
4780 |
+
"grad_norm": 1.119476318359375,
|
4781 |
+
"learning_rate": 4.828135394994137e-05,
|
4782 |
+
"loss": 0.9401,
|
4783 |
+
"step": 682
|
4784 |
+
},
|
4785 |
+
{
|
4786 |
+
"epoch": 0.004539699103691912,
|
4787 |
+
"grad_norm": 1.1330437660217285,
|
4788 |
+
"learning_rate": 4.816287828000418e-05,
|
4789 |
+
"loss": 0.8267,
|
4790 |
+
"step": 683
|
4791 |
+
},
|
4792 |
+
{
|
4793 |
+
"epoch": 0.004546345808089704,
|
4794 |
+
"grad_norm": 0.904588520526886,
|
4795 |
+
"learning_rate": 4.804441293781702e-05,
|
4796 |
+
"loss": 0.8909,
|
4797 |
+
"step": 684
|
4798 |
+
},
|
4799 |
+
{
|
4800 |
+
"epoch": 0.004552992512487496,
|
4801 |
+
"grad_norm": 1.2153210639953613,
|
4802 |
+
"learning_rate": 4.7925958589356675e-05,
|
4803 |
+
"loss": 0.9012,
|
4804 |
+
"step": 685
|
4805 |
+
},
|
4806 |
+
{
|
4807 |
+
"epoch": 0.004559639216885288,
|
4808 |
+
"grad_norm": 1.0748893022537231,
|
4809 |
+
"learning_rate": 4.780751590053813e-05,
|
4810 |
+
"loss": 0.6757,
|
4811 |
+
"step": 686
|
4812 |
+
},
|
4813 |
+
{
|
4814 |
+
"epoch": 0.00456628592128308,
|
4815 |
+
"grad_norm": 1.0900492668151855,
|
4816 |
+
"learning_rate": 4.768908553721085e-05,
|
4817 |
+
"loss": 0.7261,
|
4818 |
+
"step": 687
|
4819 |
+
},
|
4820 |
+
{
|
4821 |
+
"epoch": 0.004572932625680872,
|
4822 |
+
"grad_norm": 1.0078903436660767,
|
4823 |
+
"learning_rate": 4.757066816515494e-05,
|
4824 |
+
"loss": 0.8643,
|
4825 |
+
"step": 688
|
4826 |
+
},
|
4827 |
+
{
|
4828 |
+
"epoch": 0.0045795793300786635,
|
4829 |
+
"grad_norm": 1.585972547531128,
|
4830 |
+
"learning_rate": 4.74522644500776e-05,
|
4831 |
+
"loss": 0.8907,
|
4832 |
+
"step": 689
|
4833 |
+
},
|
4834 |
+
{
|
4835 |
+
"epoch": 0.004586226034476456,
|
4836 |
+
"grad_norm": 1.3707166910171509,
|
4837 |
+
"learning_rate": 4.7333875057609126e-05,
|
4838 |
+
"loss": 0.8671,
|
4839 |
+
"step": 690
|
4840 |
+
},
|
4841 |
+
{
|
4842 |
+
"epoch": 0.004592872738874248,
|
4843 |
+
"grad_norm": 1.412148118019104,
|
4844 |
+
"learning_rate": 4.721550065329938e-05,
|
4845 |
+
"loss": 0.9209,
|
4846 |
+
"step": 691
|
4847 |
+
},
|
4848 |
+
{
|
4849 |
+
"epoch": 0.0045995194432720396,
|
4850 |
+
"grad_norm": 1.1696654558181763,
|
4851 |
+
"learning_rate": 4.70971419026139e-05,
|
4852 |
+
"loss": 0.6559,
|
4853 |
+
"step": 692
|
4854 |
+
},
|
4855 |
+
{
|
4856 |
+
"epoch": 0.004606166147669831,
|
4857 |
+
"grad_norm": 1.614428997039795,
|
4858 |
+
"learning_rate": 4.697879947093027e-05,
|
4859 |
+
"loss": 0.9432,
|
4860 |
+
"step": 693
|
4861 |
+
},
|
4862 |
+
{
|
4863 |
+
"epoch": 0.004612812852067624,
|
4864 |
+
"grad_norm": 1.4621384143829346,
|
4865 |
+
"learning_rate": 4.6860474023534335e-05,
|
4866 |
+
"loss": 0.6001,
|
4867 |
+
"step": 694
|
4868 |
+
},
|
4869 |
+
{
|
4870 |
+
"epoch": 0.004619459556465416,
|
4871 |
+
"grad_norm": 1.82318913936615,
|
4872 |
+
"learning_rate": 4.674216622561644e-05,
|
4873 |
+
"loss": 0.9496,
|
4874 |
+
"step": 695
|
4875 |
+
},
|
4876 |
+
{
|
4877 |
+
"epoch": 0.004626106260863207,
|
4878 |
+
"grad_norm": 1.641218662261963,
|
4879 |
+
"learning_rate": 4.662387674226771e-05,
|
4880 |
+
"loss": 0.8027,
|
4881 |
+
"step": 696
|
4882 |
+
},
|
4883 |
+
{
|
4884 |
+
"epoch": 0.004632752965260999,
|
4885 |
+
"grad_norm": 3.2597484588623047,
|
4886 |
+
"learning_rate": 4.6505606238476326e-05,
|
4887 |
+
"loss": 1.388,
|
4888 |
+
"step": 697
|
4889 |
+
},
|
4890 |
+
{
|
4891 |
+
"epoch": 0.004639399669658792,
|
4892 |
+
"grad_norm": 2.241243839263916,
|
4893 |
+
"learning_rate": 4.6387355379123734e-05,
|
4894 |
+
"loss": 1.2085,
|
4895 |
+
"step": 698
|
4896 |
+
},
|
4897 |
+
{
|
4898 |
+
"epoch": 0.0046460463740565835,
|
4899 |
+
"grad_norm": 2.4520256519317627,
|
4900 |
+
"learning_rate": 4.6269124828981014e-05,
|
4901 |
+
"loss": 0.5715,
|
4902 |
+
"step": 699
|
4903 |
+
},
|
4904 |
+
{
|
4905 |
+
"epoch": 0.004652693078454375,
|
4906 |
+
"grad_norm": 2.561048984527588,
|
4907 |
+
"learning_rate": 4.615091525270501e-05,
|
4908 |
+
"loss": 0.8268,
|
4909 |
+
"step": 700
|
4910 |
+
},
|
4911 |
+
{
|
4912 |
+
"epoch": 0.004659339782852167,
|
4913 |
+
"grad_norm": 0.3658062219619751,
|
4914 |
+
"learning_rate": 4.603272731483467e-05,
|
4915 |
+
"loss": 1.0607,
|
4916 |
+
"step": 701
|
4917 |
+
},
|
4918 |
+
{
|
4919 |
+
"epoch": 0.00466598648724996,
|
4920 |
+
"grad_norm": 0.5691250562667847,
|
4921 |
+
"learning_rate": 4.5914561679787345e-05,
|
4922 |
+
"loss": 1.1228,
|
4923 |
+
"step": 702
|
4924 |
+
},
|
4925 |
+
{
|
4926 |
+
"epoch": 0.004672633191647751,
|
4927 |
+
"grad_norm": 0.4382160007953644,
|
4928 |
+
"learning_rate": 4.579641901185491e-05,
|
4929 |
+
"loss": 0.8865,
|
4930 |
+
"step": 703
|
4931 |
+
},
|
4932 |
+
{
|
4933 |
+
"epoch": 0.004679279896045543,
|
4934 |
+
"grad_norm": 0.6933655142784119,
|
4935 |
+
"learning_rate": 4.567829997520023e-05,
|
4936 |
+
"loss": 0.7971,
|
4937 |
+
"step": 704
|
4938 |
+
},
|
4939 |
+
{
|
4940 |
+
"epoch": 0.004685926600443335,
|
4941 |
+
"grad_norm": 0.6793907880783081,
|
4942 |
+
"learning_rate": 4.5560205233853266e-05,
|
4943 |
+
"loss": 0.9222,
|
4944 |
+
"step": 705
|
4945 |
+
},
|
4946 |
+
{
|
4947 |
+
"epoch": 0.004692573304841127,
|
4948 |
+
"grad_norm": 0.842363178730011,
|
4949 |
+
"learning_rate": 4.544213545170741e-05,
|
4950 |
+
"loss": 0.8264,
|
4951 |
+
"step": 706
|
4952 |
+
},
|
4953 |
+
{
|
4954 |
+
"epoch": 0.004699220009238919,
|
4955 |
+
"grad_norm": 0.6511426568031311,
|
4956 |
+
"learning_rate": 4.5324091292515724e-05,
|
4957 |
+
"loss": 1.1472,
|
4958 |
+
"step": 707
|
4959 |
+
},
|
4960 |
+
{
|
4961 |
+
"epoch": 0.004705866713636711,
|
4962 |
+
"grad_norm": 0.5798910856246948,
|
4963 |
+
"learning_rate": 4.520607341988724e-05,
|
4964 |
+
"loss": 0.8783,
|
4965 |
+
"step": 708
|
4966 |
+
},
|
4967 |
+
{
|
4968 |
+
"epoch": 0.004712513418034503,
|
4969 |
+
"grad_norm": 0.5842803120613098,
|
4970 |
+
"learning_rate": 4.5088082497283235e-05,
|
4971 |
+
"loss": 1.0191,
|
4972 |
+
"step": 709
|
4973 |
+
},
|
4974 |
+
{
|
4975 |
+
"epoch": 0.004719160122432295,
|
4976 |
+
"grad_norm": 0.5944225788116455,
|
4977 |
+
"learning_rate": 4.497011918801347e-05,
|
4978 |
+
"loss": 0.8818,
|
4979 |
+
"step": 710
|
4980 |
+
},
|
4981 |
+
{
|
4982 |
+
"epoch": 0.004725806826830087,
|
4983 |
+
"grad_norm": 0.6699796319007874,
|
4984 |
+
"learning_rate": 4.485218415523242e-05,
|
4985 |
+
"loss": 0.8876,
|
4986 |
+
"step": 711
|
4987 |
+
},
|
4988 |
+
{
|
4989 |
+
"epoch": 0.004732453531227879,
|
4990 |
+
"grad_norm": 0.6302338242530823,
|
4991 |
+
"learning_rate": 4.473427806193567e-05,
|
4992 |
+
"loss": 0.8118,
|
4993 |
+
"step": 712
|
4994 |
+
},
|
4995 |
+
{
|
4996 |
+
"epoch": 0.004739100235625671,
|
4997 |
+
"grad_norm": 0.5817951560020447,
|
4998 |
+
"learning_rate": 4.4616401570956075e-05,
|
4999 |
+
"loss": 0.8042,
|
5000 |
+
"step": 713
|
5001 |
+
},
|
5002 |
+
{
|
5003 |
+
"epoch": 0.004745746940023463,
|
5004 |
+
"grad_norm": 0.8302817940711975,
|
5005 |
+
"learning_rate": 4.4498555344960115e-05,
|
5006 |
+
"loss": 0.5791,
|
5007 |
+
"step": 714
|
5008 |
+
},
|
5009 |
+
{
|
5010 |
+
"epoch": 0.004752393644421255,
|
5011 |
+
"grad_norm": 0.6934455633163452,
|
5012 |
+
"learning_rate": 4.4380740046444066e-05,
|
5013 |
+
"loss": 0.8314,
|
5014 |
+
"step": 715
|
5015 |
+
},
|
5016 |
+
{
|
5017 |
+
"epoch": 0.0047590403488190465,
|
5018 |
+
"grad_norm": 0.8211054801940918,
|
5019 |
+
"learning_rate": 4.4262956337730385e-05,
|
5020 |
+
"loss": 0.8583,
|
5021 |
+
"step": 716
|
5022 |
+
},
|
5023 |
+
{
|
5024 |
+
"epoch": 0.004765687053216839,
|
5025 |
+
"grad_norm": 0.8365226984024048,
|
5026 |
+
"learning_rate": 4.4145204880963945e-05,
|
5027 |
+
"loss": 0.7647,
|
5028 |
+
"step": 717
|
5029 |
+
},
|
5030 |
+
{
|
5031 |
+
"epoch": 0.004772333757614631,
|
5032 |
+
"grad_norm": 0.7704525589942932,
|
5033 |
+
"learning_rate": 4.402748633810826e-05,
|
5034 |
+
"loss": 0.8246,
|
5035 |
+
"step": 718
|
5036 |
+
},
|
5037 |
+
{
|
5038 |
+
"epoch": 0.004778980462012423,
|
5039 |
+
"grad_norm": 0.9756360054016113,
|
5040 |
+
"learning_rate": 4.390980137094187e-05,
|
5041 |
+
"loss": 0.7665,
|
5042 |
+
"step": 719
|
5043 |
+
},
|
5044 |
+
{
|
5045 |
+
"epoch": 0.004785627166410214,
|
5046 |
+
"grad_norm": 0.8577277660369873,
|
5047 |
+
"learning_rate": 4.379215064105454e-05,
|
5048 |
+
"loss": 0.7484,
|
5049 |
+
"step": 720
|
5050 |
+
},
|
5051 |
+
{
|
5052 |
+
"epoch": 0.004792273870808007,
|
5053 |
+
"grad_norm": 0.7188462615013123,
|
5054 |
+
"learning_rate": 4.367453480984356e-05,
|
5055 |
+
"loss": 0.8432,
|
5056 |
+
"step": 721
|
5057 |
+
},
|
5058 |
+
{
|
5059 |
+
"epoch": 0.004798920575205799,
|
5060 |
+
"grad_norm": 0.82792067527771,
|
5061 |
+
"learning_rate": 4.355695453851e-05,
|
5062 |
+
"loss": 0.9031,
|
5063 |
+
"step": 722
|
5064 |
+
},
|
5065 |
+
{
|
5066 |
+
"epoch": 0.0048055672796035905,
|
5067 |
+
"grad_norm": 0.7311772704124451,
|
5068 |
+
"learning_rate": 4.3439410488055075e-05,
|
5069 |
+
"loss": 0.7162,
|
5070 |
+
"step": 723
|
5071 |
+
},
|
5072 |
+
{
|
5073 |
+
"epoch": 0.004812213984001382,
|
5074 |
+
"grad_norm": 0.7862158417701721,
|
5075 |
+
"learning_rate": 4.332190331927636e-05,
|
5076 |
+
"loss": 0.9094,
|
5077 |
+
"step": 724
|
5078 |
+
},
|
5079 |
+
{
|
5080 |
+
"epoch": 0.004818860688399175,
|
5081 |
+
"grad_norm": 0.8980942964553833,
|
5082 |
+
"learning_rate": 4.3204433692764096e-05,
|
5083 |
+
"loss": 0.9385,
|
5084 |
+
"step": 725
|
5085 |
+
},
|
5086 |
+
{
|
5087 |
+
"epoch": 0.0048255073927969665,
|
5088 |
+
"grad_norm": 0.8081440925598145,
|
5089 |
+
"learning_rate": 4.3087002268897434e-05,
|
5090 |
+
"loss": 0.81,
|
5091 |
+
"step": 726
|
5092 |
+
},
|
5093 |
+
{
|
5094 |
+
"epoch": 0.004832154097194758,
|
5095 |
+
"grad_norm": 1.0013748407363892,
|
5096 |
+
"learning_rate": 4.296960970784082e-05,
|
5097 |
+
"loss": 0.7476,
|
5098 |
+
"step": 727
|
5099 |
+
},
|
5100 |
+
{
|
5101 |
+
"epoch": 0.00483880080159255,
|
5102 |
+
"grad_norm": 0.9266440272331238,
|
5103 |
+
"learning_rate": 4.285225666954016e-05,
|
5104 |
+
"loss": 0.8057,
|
5105 |
+
"step": 728
|
5106 |
+
},
|
5107 |
+
{
|
5108 |
+
"epoch": 0.004845447505990343,
|
5109 |
+
"grad_norm": 0.9051468968391418,
|
5110 |
+
"learning_rate": 4.273494381371926e-05,
|
5111 |
+
"loss": 0.939,
|
5112 |
+
"step": 729
|
5113 |
+
},
|
5114 |
+
{
|
5115 |
+
"epoch": 0.004852094210388134,
|
5116 |
+
"grad_norm": 0.8008094429969788,
|
5117 |
+
"learning_rate": 4.2617671799875944e-05,
|
5118 |
+
"loss": 0.8005,
|
5119 |
+
"step": 730
|
5120 |
+
},
|
5121 |
+
{
|
5122 |
+
"epoch": 0.004858740914785926,
|
5123 |
+
"grad_norm": 0.9573123455047607,
|
5124 |
+
"learning_rate": 4.2500441287278505e-05,
|
5125 |
+
"loss": 0.7531,
|
5126 |
+
"step": 731
|
5127 |
+
},
|
5128 |
+
{
|
5129 |
+
"epoch": 0.004865387619183718,
|
5130 |
+
"grad_norm": 1.2024757862091064,
|
5131 |
+
"learning_rate": 4.238325293496186e-05,
|
5132 |
+
"loss": 0.6332,
|
5133 |
+
"step": 732
|
5134 |
+
},
|
5135 |
+
{
|
5136 |
+
"epoch": 0.0048720343235815105,
|
5137 |
+
"grad_norm": 1.0710391998291016,
|
5138 |
+
"learning_rate": 4.226610740172396e-05,
|
5139 |
+
"loss": 0.9639,
|
5140 |
+
"step": 733
|
5141 |
+
},
|
5142 |
+
{
|
5143 |
+
"epoch": 0.004878681027979302,
|
5144 |
+
"grad_norm": 1.2133551836013794,
|
5145 |
+
"learning_rate": 4.214900534612205e-05,
|
5146 |
+
"loss": 0.5005,
|
5147 |
+
"step": 734
|
5148 |
+
},
|
5149 |
+
{
|
5150 |
+
"epoch": 0.004885327732377094,
|
5151 |
+
"grad_norm": 1.379006266593933,
|
5152 |
+
"learning_rate": 4.203194742646893e-05,
|
5153 |
+
"loss": 0.8146,
|
5154 |
+
"step": 735
|
5155 |
+
},
|
5156 |
+
{
|
5157 |
+
"epoch": 0.004891974436774886,
|
5158 |
+
"grad_norm": 1.496587872505188,
|
5159 |
+
"learning_rate": 4.191493430082929e-05,
|
5160 |
+
"loss": 0.7647,
|
5161 |
+
"step": 736
|
5162 |
+
},
|
5163 |
+
{
|
5164 |
+
"epoch": 0.004898621141172678,
|
5165 |
+
"grad_norm": 1.0434892177581787,
|
5166 |
+
"learning_rate": 4.179796662701597e-05,
|
5167 |
+
"loss": 0.8637,
|
5168 |
+
"step": 737
|
5169 |
+
},
|
5170 |
+
{
|
5171 |
+
"epoch": 0.00490526784557047,
|
5172 |
+
"grad_norm": 1.0825583934783936,
|
5173 |
+
"learning_rate": 4.168104506258636e-05,
|
5174 |
+
"loss": 0.6391,
|
5175 |
+
"step": 738
|
5176 |
+
},
|
5177 |
+
{
|
5178 |
+
"epoch": 0.004911914549968262,
|
5179 |
+
"grad_norm": 0.9926093220710754,
|
5180 |
+
"learning_rate": 4.1564170264838595e-05,
|
5181 |
+
"loss": 0.8085,
|
5182 |
+
"step": 739
|
5183 |
+
},
|
5184 |
+
{
|
5185 |
+
"epoch": 0.004918561254366054,
|
5186 |
+
"grad_norm": 1.1647216081619263,
|
5187 |
+
"learning_rate": 4.144734289080791e-05,
|
5188 |
+
"loss": 0.6383,
|
5189 |
+
"step": 740
|
5190 |
+
},
|
5191 |
+
{
|
5192 |
+
"epoch": 0.004925207958763846,
|
5193 |
+
"grad_norm": 1.0999730825424194,
|
5194 |
+
"learning_rate": 4.133056359726289e-05,
|
5195 |
+
"loss": 0.7778,
|
5196 |
+
"step": 741
|
5197 |
+
},
|
5198 |
+
{
|
5199 |
+
"epoch": 0.004931854663161638,
|
5200 |
+
"grad_norm": 1.5499298572540283,
|
5201 |
+
"learning_rate": 4.121383304070191e-05,
|
5202 |
+
"loss": 0.7831,
|
5203 |
+
"step": 742
|
5204 |
+
},
|
5205 |
+
{
|
5206 |
+
"epoch": 0.00493850136755943,
|
5207 |
+
"grad_norm": 1.425614595413208,
|
5208 |
+
"learning_rate": 4.1097151877349316e-05,
|
5209 |
+
"loss": 0.8449,
|
5210 |
+
"step": 743
|
5211 |
+
},
|
5212 |
+
{
|
5213 |
+
"epoch": 0.004945148071957222,
|
5214 |
+
"grad_norm": 1.5261919498443604,
|
5215 |
+
"learning_rate": 4.0980520763151796e-05,
|
5216 |
+
"loss": 0.7073,
|
5217 |
+
"step": 744
|
5218 |
+
},
|
5219 |
+
{
|
5220 |
+
"epoch": 0.004951794776355014,
|
5221 |
+
"grad_norm": 1.5911787748336792,
|
5222 |
+
"learning_rate": 4.086394035377463e-05,
|
5223 |
+
"loss": 0.6253,
|
5224 |
+
"step": 745
|
5225 |
+
},
|
5226 |
+
{
|
5227 |
+
"epoch": 0.004958441480752806,
|
5228 |
+
"grad_norm": 1.6266251802444458,
|
5229 |
+
"learning_rate": 4.074741130459813e-05,
|
5230 |
+
"loss": 0.8432,
|
5231 |
+
"step": 746
|
5232 |
+
},
|
5233 |
+
{
|
5234 |
+
"epoch": 0.004965088185150597,
|
5235 |
+
"grad_norm": 1.2186232805252075,
|
5236 |
+
"learning_rate": 4.063093427071376e-05,
|
5237 |
+
"loss": 0.6454,
|
5238 |
+
"step": 747
|
5239 |
+
},
|
5240 |
+
{
|
5241 |
+
"epoch": 0.00497173488954839,
|
5242 |
+
"grad_norm": 1.5160404443740845,
|
5243 |
+
"learning_rate": 4.051450990692073e-05,
|
5244 |
+
"loss": 0.805,
|
5245 |
+
"step": 748
|
5246 |
+
},
|
5247 |
+
{
|
5248 |
+
"epoch": 0.004978381593946182,
|
5249 |
+
"grad_norm": 2.0402166843414307,
|
5250 |
+
"learning_rate": 4.039813886772201e-05,
|
5251 |
+
"loss": 0.8304,
|
5252 |
+
"step": 749
|
5253 |
+
},
|
5254 |
+
{
|
5255 |
+
"epoch": 0.0049850282983439735,
|
5256 |
+
"grad_norm": 2.949624538421631,
|
5257 |
+
"learning_rate": 4.028182180732088e-05,
|
5258 |
+
"loss": 0.6917,
|
5259 |
+
"step": 750
|
5260 |
+
},
|
5261 |
+
{
|
5262 |
+
"epoch": 0.004991675002741765,
|
5263 |
+
"grad_norm": 0.35819950699806213,
|
5264 |
+
"learning_rate": 4.016555937961711e-05,
|
5265 |
+
"loss": 0.8131,
|
5266 |
+
"step": 751
|
5267 |
+
},
|
5268 |
+
{
|
5269 |
+
"epoch": 0.004998321707139558,
|
5270 |
+
"grad_norm": 0.43927833437919617,
|
5271 |
+
"learning_rate": 4.0049352238203376e-05,
|
5272 |
+
"loss": 0.857,
|
5273 |
+
"step": 752
|
5274 |
+
},
|
5275 |
+
{
|
5276 |
+
"epoch": 0.00500496841153735,
|
5277 |
+
"grad_norm": 0.4738449156284332,
|
5278 |
+
"learning_rate": 3.9933201036361554e-05,
|
5279 |
+
"loss": 0.8975,
|
5280 |
+
"step": 753
|
5281 |
+
},
|
5282 |
+
{
|
5283 |
+
"epoch": 0.005011615115935141,
|
5284 |
+
"grad_norm": 1.1976778507232666,
|
5285 |
+
"learning_rate": 3.981710642705906e-05,
|
5286 |
+
"loss": 0.7099,
|
5287 |
+
"step": 754
|
5288 |
+
},
|
5289 |
+
{
|
5290 |
+
"epoch": 0.005018261820332933,
|
5291 |
+
"grad_norm": 0.5854752063751221,
|
5292 |
+
"learning_rate": 3.970106906294509e-05,
|
5293 |
+
"loss": 0.9164,
|
5294 |
+
"step": 755
|
5295 |
+
},
|
5296 |
+
{
|
5297 |
+
"epoch": 0.005024908524730726,
|
5298 |
+
"grad_norm": 0.5547404289245605,
|
5299 |
+
"learning_rate": 3.9585089596347094e-05,
|
5300 |
+
"loss": 0.8185,
|
5301 |
+
"step": 756
|
5302 |
+
},
|
5303 |
+
{
|
5304 |
+
"epoch": 0.005031555229128517,
|
5305 |
+
"grad_norm": 0.7054506540298462,
|
5306 |
+
"learning_rate": 3.946916867926702e-05,
|
5307 |
+
"loss": 0.9904,
|
5308 |
+
"step": 757
|
5309 |
+
},
|
5310 |
+
{
|
5311 |
+
"epoch": 0.005038201933526309,
|
5312 |
+
"grad_norm": 0.5721423029899597,
|
5313 |
+
"learning_rate": 3.935330696337768e-05,
|
5314 |
+
"loss": 0.8113,
|
5315 |
+
"step": 758
|
5316 |
+
},
|
5317 |
+
{
|
5318 |
+
"epoch": 0.005044848637924101,
|
5319 |
+
"grad_norm": 0.7183482050895691,
|
5320 |
+
"learning_rate": 3.923750510001903e-05,
|
5321 |
+
"loss": 1.132,
|
5322 |
+
"step": 759
|
5323 |
+
},
|
5324 |
+
{
|
5325 |
+
"epoch": 0.0050514953423218935,
|
5326 |
+
"grad_norm": 0.6827608346939087,
|
5327 |
+
"learning_rate": 3.9121763740194616e-05,
|
5328 |
+
"loss": 0.8093,
|
5329 |
+
"step": 760
|
5330 |
+
},
|
5331 |
+
{
|
5332 |
+
"epoch": 0.005058142046719685,
|
5333 |
+
"grad_norm": 0.6405389308929443,
|
5334 |
+
"learning_rate": 3.900608353456782e-05,
|
5335 |
+
"loss": 0.7248,
|
5336 |
+
"step": 761
|
5337 |
+
},
|
5338 |
+
{
|
5339 |
+
"epoch": 0.005064788751117477,
|
5340 |
+
"grad_norm": 0.9901874661445618,
|
5341 |
+
"learning_rate": 3.889046513345818e-05,
|
5342 |
+
"loss": 0.6536,
|
5343 |
+
"step": 762
|
5344 |
+
},
|
5345 |
+
{
|
5346 |
+
"epoch": 0.005071435455515269,
|
5347 |
+
"grad_norm": 0.7273656725883484,
|
5348 |
+
"learning_rate": 3.87749091868379e-05,
|
5349 |
+
"loss": 0.7783,
|
5350 |
+
"step": 763
|
5351 |
+
},
|
5352 |
+
{
|
5353 |
+
"epoch": 0.005078082159913061,
|
5354 |
+
"grad_norm": 0.7552725672721863,
|
5355 |
+
"learning_rate": 3.8659416344328e-05,
|
5356 |
+
"loss": 0.8353,
|
5357 |
+
"step": 764
|
5358 |
+
},
|
5359 |
+
{
|
5360 |
+
"epoch": 0.005084728864310853,
|
5361 |
+
"grad_norm": 0.9276883602142334,
|
5362 |
+
"learning_rate": 3.854398725519479e-05,
|
5363 |
+
"loss": 0.804,
|
5364 |
+
"step": 765
|
5365 |
+
},
|
5366 |
+
{
|
5367 |
+
"epoch": 0.005091375568708645,
|
5368 |
+
"grad_norm": 0.6831343173980713,
|
5369 |
+
"learning_rate": 3.842862256834613e-05,
|
5370 |
+
"loss": 0.8294,
|
5371 |
+
"step": 766
|
5372 |
+
},
|
5373 |
+
{
|
5374 |
+
"epoch": 0.005098022273106437,
|
5375 |
+
"grad_norm": 0.8653516173362732,
|
5376 |
+
"learning_rate": 3.831332293232787e-05,
|
5377 |
+
"loss": 0.7905,
|
5378 |
+
"step": 767
|
5379 |
+
},
|
5380 |
+
{
|
5381 |
+
"epoch": 0.005104668977504229,
|
5382 |
+
"grad_norm": 0.788194477558136,
|
5383 |
+
"learning_rate": 3.819808899532015e-05,
|
5384 |
+
"loss": 0.896,
|
5385 |
+
"step": 768
|
5386 |
+
},
|
5387 |
+
{
|
5388 |
+
"epoch": 0.005111315681902021,
|
5389 |
+
"grad_norm": 0.8511560559272766,
|
5390 |
+
"learning_rate": 3.808292140513379e-05,
|
5391 |
+
"loss": 0.7863,
|
5392 |
+
"step": 769
|
5393 |
+
},
|
5394 |
+
{
|
5395 |
+
"epoch": 0.005117962386299813,
|
5396 |
+
"grad_norm": 0.868413507938385,
|
5397 |
+
"learning_rate": 3.796782080920659e-05,
|
5398 |
+
"loss": 0.8043,
|
5399 |
+
"step": 770
|
5400 |
+
},
|
5401 |
+
{
|
5402 |
+
"epoch": 0.005124609090697605,
|
5403 |
+
"grad_norm": 0.8420473337173462,
|
5404 |
+
"learning_rate": 3.7852787854599746e-05,
|
5405 |
+
"loss": 0.9915,
|
5406 |
+
"step": 771
|
5407 |
+
},
|
5408 |
+
{
|
5409 |
+
"epoch": 0.005131255795095397,
|
5410 |
+
"grad_norm": 0.8845599293708801,
|
5411 |
+
"learning_rate": 3.77378231879942e-05,
|
5412 |
+
"loss": 0.7352,
|
5413 |
+
"step": 772
|
5414 |
+
},
|
5415 |
+
{
|
5416 |
+
"epoch": 0.005137902499493189,
|
5417 |
+
"grad_norm": 0.820316731929779,
|
5418 |
+
"learning_rate": 3.762292745568702e-05,
|
5419 |
+
"loss": 0.7933,
|
5420 |
+
"step": 773
|
5421 |
+
},
|
5422 |
+
{
|
5423 |
+
"epoch": 0.0051445492038909805,
|
5424 |
+
"grad_norm": 1.0592056512832642,
|
5425 |
+
"learning_rate": 3.750810130358769e-05,
|
5426 |
+
"loss": 0.8795,
|
5427 |
+
"step": 774
|
5428 |
+
},
|
5429 |
+
{
|
5430 |
+
"epoch": 0.005151195908288773,
|
5431 |
+
"grad_norm": 0.7946357131004333,
|
5432 |
+
"learning_rate": 3.7393345377214586e-05,
|
5433 |
+
"loss": 0.6885,
|
5434 |
+
"step": 775
|
5435 |
+
},
|
5436 |
+
{
|
5437 |
+
"epoch": 0.005157842612686565,
|
5438 |
+
"grad_norm": 0.8453227877616882,
|
5439 |
+
"learning_rate": 3.727866032169127e-05,
|
5440 |
+
"loss": 0.7128,
|
5441 |
+
"step": 776
|
5442 |
+
},
|
5443 |
+
{
|
5444 |
+
"epoch": 0.005164489317084357,
|
5445 |
+
"grad_norm": 1.225663185119629,
|
5446 |
+
"learning_rate": 3.716404678174288e-05,
|
5447 |
+
"loss": 0.9135,
|
5448 |
+
"step": 777
|
5449 |
+
},
|
5450 |
+
{
|
5451 |
+
"epoch": 0.005171136021482148,
|
5452 |
+
"grad_norm": 0.8996406197547913,
|
5453 |
+
"learning_rate": 3.704950540169256e-05,
|
5454 |
+
"loss": 0.8883,
|
5455 |
+
"step": 778
|
5456 |
+
},
|
5457 |
+
{
|
5458 |
+
"epoch": 0.005177782725879941,
|
5459 |
+
"grad_norm": 0.9237585067749023,
|
5460 |
+
"learning_rate": 3.693503682545775e-05,
|
5461 |
+
"loss": 0.8482,
|
5462 |
+
"step": 779
|
5463 |
+
},
|
5464 |
+
{
|
5465 |
+
"epoch": 0.005184429430277733,
|
5466 |
+
"grad_norm": 1.2108267545700073,
|
5467 |
+
"learning_rate": 3.682064169654663e-05,
|
5468 |
+
"loss": 0.8083,
|
5469 |
+
"step": 780
|
5470 |
+
},
|
5471 |
+
{
|
5472 |
+
"epoch": 0.005191076134675524,
|
5473 |
+
"grad_norm": 0.9889872670173645,
|
5474 |
+
"learning_rate": 3.670632065805445e-05,
|
5475 |
+
"loss": 0.9583,
|
5476 |
+
"step": 781
|
5477 |
+
},
|
5478 |
+
{
|
5479 |
+
"epoch": 0.005197722839073316,
|
5480 |
+
"grad_norm": 0.9854968190193176,
|
5481 |
+
"learning_rate": 3.659207435265998e-05,
|
5482 |
+
"loss": 0.5858,
|
5483 |
+
"step": 782
|
5484 |
+
},
|
5485 |
+
{
|
5486 |
+
"epoch": 0.005204369543471109,
|
5487 |
+
"grad_norm": 0.9853740930557251,
|
5488 |
+
"learning_rate": 3.6477903422621865e-05,
|
5489 |
+
"loss": 0.7543,
|
5490 |
+
"step": 783
|
5491 |
+
},
|
5492 |
+
{
|
5493 |
+
"epoch": 0.0052110162478689005,
|
5494 |
+
"grad_norm": 1.0569560527801514,
|
5495 |
+
"learning_rate": 3.6363808509774996e-05,
|
5496 |
+
"loss": 0.9121,
|
5497 |
+
"step": 784
|
5498 |
+
},
|
5499 |
+
{
|
5500 |
+
"epoch": 0.005217662952266692,
|
5501 |
+
"grad_norm": 1.209022879600525,
|
5502 |
+
"learning_rate": 3.6249790255526915e-05,
|
5503 |
+
"loss": 0.9692,
|
5504 |
+
"step": 785
|
5505 |
+
},
|
5506 |
+
{
|
5507 |
+
"epoch": 0.005224309656664484,
|
5508 |
+
"grad_norm": 0.822544515132904,
|
5509 |
+
"learning_rate": 3.6135849300854225e-05,
|
5510 |
+
"loss": 0.5065,
|
5511 |
+
"step": 786
|
5512 |
+
},
|
5513 |
+
{
|
5514 |
+
"epoch": 0.005230956361062277,
|
5515 |
+
"grad_norm": 1.006661057472229,
|
5516 |
+
"learning_rate": 3.602198628629893e-05,
|
5517 |
+
"loss": 0.7771,
|
5518 |
+
"step": 787
|
5519 |
+
},
|
5520 |
+
{
|
5521 |
+
"epoch": 0.005237603065460068,
|
5522 |
+
"grad_norm": 1.003117322921753,
|
5523 |
+
"learning_rate": 3.590820185196497e-05,
|
5524 |
+
"loss": 0.7955,
|
5525 |
+
"step": 788
|
5526 |
+
},
|
5527 |
+
{
|
5528 |
+
"epoch": 0.00524424976985786,
|
5529 |
+
"grad_norm": 1.7347395420074463,
|
5530 |
+
"learning_rate": 3.579449663751444e-05,
|
5531 |
+
"loss": 0.7385,
|
5532 |
+
"step": 789
|
5533 |
+
},
|
5534 |
+
{
|
5535 |
+
"epoch": 0.005250896474255652,
|
5536 |
+
"grad_norm": 1.0848127603530884,
|
5537 |
+
"learning_rate": 3.5680871282164144e-05,
|
5538 |
+
"loss": 0.5635,
|
5539 |
+
"step": 790
|
5540 |
+
},
|
5541 |
+
{
|
5542 |
+
"epoch": 0.005257543178653444,
|
5543 |
+
"grad_norm": 1.8770695924758911,
|
5544 |
+
"learning_rate": 3.556732642468189e-05,
|
5545 |
+
"loss": 0.9407,
|
5546 |
+
"step": 791
|
5547 |
+
},
|
5548 |
+
{
|
5549 |
+
"epoch": 0.005264189883051236,
|
5550 |
+
"grad_norm": 1.2975075244903564,
|
5551 |
+
"learning_rate": 3.545386270338299e-05,
|
5552 |
+
"loss": 0.8492,
|
5553 |
+
"step": 792
|
5554 |
+
},
|
5555 |
+
{
|
5556 |
+
"epoch": 0.005270836587449028,
|
5557 |
+
"grad_norm": 1.0128837823867798,
|
5558 |
+
"learning_rate": 3.5340480756126624e-05,
|
5559 |
+
"loss": 0.6228,
|
5560 |
+
"step": 793
|
5561 |
+
},
|
5562 |
+
{
|
5563 |
+
"epoch": 0.00527748329184682,
|
5564 |
+
"grad_norm": 1.1561847925186157,
|
5565 |
+
"learning_rate": 3.522718122031229e-05,
|
5566 |
+
"loss": 0.6094,
|
5567 |
+
"step": 794
|
5568 |
+
},
|
5569 |
+
{
|
5570 |
+
"epoch": 0.005284129996244612,
|
5571 |
+
"grad_norm": 1.5360347032546997,
|
5572 |
+
"learning_rate": 3.5113964732876106e-05,
|
5573 |
+
"loss": 0.865,
|
5574 |
+
"step": 795
|
5575 |
+
},
|
5576 |
+
{
|
5577 |
+
"epoch": 0.005290776700642404,
|
5578 |
+
"grad_norm": 1.4779386520385742,
|
5579 |
+
"learning_rate": 3.500083193028741e-05,
|
5580 |
+
"loss": 0.6847,
|
5581 |
+
"step": 796
|
5582 |
+
},
|
5583 |
+
{
|
5584 |
+
"epoch": 0.005297423405040196,
|
5585 |
+
"grad_norm": 1.67745041847229,
|
5586 |
+
"learning_rate": 3.488778344854504e-05,
|
5587 |
+
"loss": 0.6983,
|
5588 |
+
"step": 797
|
5589 |
+
},
|
5590 |
+
{
|
5591 |
+
"epoch": 0.005304070109437988,
|
5592 |
+
"grad_norm": 1.8071433305740356,
|
5593 |
+
"learning_rate": 3.477481992317384e-05,
|
5594 |
+
"loss": 0.7865,
|
5595 |
+
"step": 798
|
5596 |
+
},
|
5597 |
+
{
|
5598 |
+
"epoch": 0.00531071681383578,
|
5599 |
+
"grad_norm": 2.242330312728882,
|
5600 |
+
"learning_rate": 3.466194198922102e-05,
|
5601 |
+
"loss": 0.9295,
|
5602 |
+
"step": 799
|
5603 |
+
},
|
5604 |
+
{
|
5605 |
+
"epoch": 0.005317363518233572,
|
5606 |
+
"grad_norm": 3.3643980026245117,
|
5607 |
+
"learning_rate": 3.4549150281252636e-05,
|
5608 |
+
"loss": 0.7402,
|
5609 |
+
"step": 800
|
5610 |
+
},
|
5611 |
+
{
|
5612 |
+
"epoch": 0.0053240102226313635,
|
5613 |
+
"grad_norm": 0.3782943785190582,
|
5614 |
+
"learning_rate": 3.4436445433350015e-05,
|
5615 |
+
"loss": 0.9188,
|
5616 |
+
"step": 801
|
5617 |
+
},
|
5618 |
+
{
|
5619 |
+
"epoch": 0.005330656927029156,
|
5620 |
+
"grad_norm": 0.539129376411438,
|
5621 |
+
"learning_rate": 3.432382807910615e-05,
|
5622 |
+
"loss": 0.8105,
|
5623 |
+
"step": 802
|
5624 |
+
},
|
5625 |
+
{
|
5626 |
+
"epoch": 0.005337303631426948,
|
5627 |
+
"grad_norm": 0.8282445669174194,
|
5628 |
+
"learning_rate": 3.421129885162223e-05,
|
5629 |
+
"loss": 0.7792,
|
5630 |
+
"step": 803
|
5631 |
+
},
|
5632 |
+
{
|
5633 |
+
"epoch": 0.00534395033582474,
|
5634 |
+
"grad_norm": 0.5992372035980225,
|
5635 |
+
"learning_rate": 3.409885838350397e-05,
|
5636 |
+
"loss": 0.6944,
|
5637 |
+
"step": 804
|
5638 |
+
},
|
5639 |
+
{
|
5640 |
+
"epoch": 0.005350597040222531,
|
5641 |
+
"grad_norm": 0.5927280187606812,
|
5642 |
+
"learning_rate": 3.3986507306858125e-05,
|
5643 |
+
"loss": 0.8305,
|
5644 |
+
"step": 805
|
5645 |
+
},
|
5646 |
+
{
|
5647 |
+
"epoch": 0.005357243744620324,
|
5648 |
+
"grad_norm": 0.6903488636016846,
|
5649 |
+
"learning_rate": 3.387424625328892e-05,
|
5650 |
+
"loss": 0.8415,
|
5651 |
+
"step": 806
|
5652 |
+
},
|
5653 |
+
{
|
5654 |
+
"epoch": 0.005363890449018116,
|
5655 |
+
"grad_norm": 0.5096772313117981,
|
5656 |
+
"learning_rate": 3.376207585389447e-05,
|
5657 |
+
"loss": 0.8794,
|
5658 |
+
"step": 807
|
5659 |
+
},
|
5660 |
+
{
|
5661 |
+
"epoch": 0.0053705371534159075,
|
5662 |
+
"grad_norm": 0.6946663856506348,
|
5663 |
+
"learning_rate": 3.3649996739263326e-05,
|
5664 |
+
"loss": 1.0822,
|
5665 |
+
"step": 808
|
5666 |
+
},
|
5667 |
+
{
|
5668 |
+
"epoch": 0.005377183857813699,
|
5669 |
+
"grad_norm": 0.6458315253257751,
|
5670 |
+
"learning_rate": 3.353800953947079e-05,
|
5671 |
+
"loss": 0.8732,
|
5672 |
+
"step": 809
|
5673 |
+
},
|
5674 |
+
{
|
5675 |
+
"epoch": 0.005383830562211492,
|
5676 |
+
"grad_norm": 0.5500379204750061,
|
5677 |
+
"learning_rate": 3.342611488407549e-05,
|
5678 |
+
"loss": 0.9068,
|
5679 |
+
"step": 810
|
5680 |
+
},
|
5681 |
+
{
|
5682 |
+
"epoch": 0.0053904772666092836,
|
5683 |
+
"grad_norm": 0.8139527440071106,
|
5684 |
+
"learning_rate": 3.331431340211576e-05,
|
5685 |
+
"loss": 0.9336,
|
5686 |
+
"step": 811
|
5687 |
+
},
|
5688 |
+
{
|
5689 |
+
"epoch": 0.005397123971007075,
|
5690 |
+
"grad_norm": 0.6902498602867126,
|
5691 |
+
"learning_rate": 3.3202605722106186e-05,
|
5692 |
+
"loss": 0.8824,
|
5693 |
+
"step": 812
|
5694 |
+
},
|
5695 |
+
{
|
5696 |
+
"epoch": 0.005403770675404867,
|
5697 |
+
"grad_norm": 0.6358487606048584,
|
5698 |
+
"learning_rate": 3.309099247203401e-05,
|
5699 |
+
"loss": 0.9743,
|
5700 |
+
"step": 813
|
5701 |
+
},
|
5702 |
+
{
|
5703 |
+
"epoch": 0.00541041737980266,
|
5704 |
+
"grad_norm": 0.8419442176818848,
|
5705 |
+
"learning_rate": 3.297947427935559e-05,
|
5706 |
+
"loss": 0.8796,
|
5707 |
+
"step": 814
|
5708 |
+
},
|
5709 |
+
{
|
5710 |
+
"epoch": 0.005417064084200451,
|
5711 |
+
"grad_norm": 0.8624058365821838,
|
5712 |
+
"learning_rate": 3.286805177099293e-05,
|
5713 |
+
"loss": 0.8735,
|
5714 |
+
"step": 815
|
5715 |
+
},
|
5716 |
+
{
|
5717 |
+
"epoch": 0.005423710788598243,
|
5718 |
+
"grad_norm": 1.113830327987671,
|
5719 |
+
"learning_rate": 3.275672557333015e-05,
|
5720 |
+
"loss": 0.8796,
|
5721 |
+
"step": 816
|
5722 |
+
},
|
5723 |
+
{
|
5724 |
+
"epoch": 0.005430357492996035,
|
5725 |
+
"grad_norm": 0.7120754718780518,
|
5726 |
+
"learning_rate": 3.264549631220985e-05,
|
5727 |
+
"loss": 0.8,
|
5728 |
+
"step": 817
|
5729 |
+
},
|
5730 |
+
{
|
5731 |
+
"epoch": 0.0054370041973938275,
|
5732 |
+
"grad_norm": 0.8254979252815247,
|
5733 |
+
"learning_rate": 3.253436461292978e-05,
|
5734 |
+
"loss": 0.702,
|
5735 |
+
"step": 818
|
5736 |
+
},
|
5737 |
+
{
|
5738 |
+
"epoch": 0.005443650901791619,
|
5739 |
+
"grad_norm": 0.7647942900657654,
|
5740 |
+
"learning_rate": 3.2423331100239177e-05,
|
5741 |
+
"loss": 0.8197,
|
5742 |
+
"step": 819
|
5743 |
+
},
|
5744 |
+
{
|
5745 |
+
"epoch": 0.005450297606189411,
|
5746 |
+
"grad_norm": 0.6956480145454407,
|
5747 |
+
"learning_rate": 3.231239639833531e-05,
|
5748 |
+
"loss": 0.8276,
|
5749 |
+
"step": 820
|
5750 |
+
},
|
5751 |
+
{
|
5752 |
+
"epoch": 0.005456944310587203,
|
5753 |
+
"grad_norm": 0.9934280514717102,
|
5754 |
+
"learning_rate": 3.2201561130859957e-05,
|
5755 |
+
"loss": 0.8602,
|
5756 |
+
"step": 821
|
5757 |
+
},
|
5758 |
+
{
|
5759 |
+
"epoch": 0.005463591014984995,
|
5760 |
+
"grad_norm": 0.9009010791778564,
|
5761 |
+
"learning_rate": 3.209082592089591e-05,
|
5762 |
+
"loss": 0.8127,
|
5763 |
+
"step": 822
|
5764 |
+
},
|
5765 |
+
{
|
5766 |
+
"epoch": 0.005470237719382787,
|
5767 |
+
"grad_norm": 1.0514200925827026,
|
5768 |
+
"learning_rate": 3.1980191390963484e-05,
|
5769 |
+
"loss": 0.9751,
|
5770 |
+
"step": 823
|
5771 |
+
},
|
5772 |
+
{
|
5773 |
+
"epoch": 0.005476884423780579,
|
5774 |
+
"grad_norm": 1.3160456418991089,
|
5775 |
+
"learning_rate": 3.1869658163017e-05,
|
5776 |
+
"loss": 0.8358,
|
5777 |
+
"step": 824
|
5778 |
+
},
|
5779 |
+
{
|
5780 |
+
"epoch": 0.005483531128178371,
|
5781 |
+
"grad_norm": 1.054132342338562,
|
5782 |
+
"learning_rate": 3.175922685844125e-05,
|
5783 |
+
"loss": 0.6201,
|
5784 |
+
"step": 825
|
5785 |
+
},
|
5786 |
+
{
|
5787 |
+
"epoch": 0.005490177832576163,
|
5788 |
+
"grad_norm": 0.984657883644104,
|
5789 |
+
"learning_rate": 3.164889809804808e-05,
|
5790 |
+
"loss": 1.0336,
|
5791 |
+
"step": 826
|
5792 |
+
},
|
5793 |
+
{
|
5794 |
+
"epoch": 0.005496824536973955,
|
5795 |
+
"grad_norm": 1.1126635074615479,
|
5796 |
+
"learning_rate": 3.1538672502072817e-05,
|
5797 |
+
"loss": 0.8523,
|
5798 |
+
"step": 827
|
5799 |
+
},
|
5800 |
+
{
|
5801 |
+
"epoch": 0.005503471241371747,
|
5802 |
+
"grad_norm": 0.7525351047515869,
|
5803 |
+
"learning_rate": 3.142855069017093e-05,
|
5804 |
+
"loss": 0.568,
|
5805 |
+
"step": 828
|
5806 |
+
},
|
5807 |
+
{
|
5808 |
+
"epoch": 0.005510117945769539,
|
5809 |
+
"grad_norm": 0.9859227538108826,
|
5810 |
+
"learning_rate": 3.131853328141434e-05,
|
5811 |
+
"loss": 0.7202,
|
5812 |
+
"step": 829
|
5813 |
+
},
|
5814 |
+
{
|
5815 |
+
"epoch": 0.005516764650167331,
|
5816 |
+
"grad_norm": 0.9517953991889954,
|
5817 |
+
"learning_rate": 3.12086208942881e-05,
|
5818 |
+
"loss": 0.7246,
|
5819 |
+
"step": 830
|
5820 |
+
},
|
5821 |
+
{
|
5822 |
+
"epoch": 0.005523411354565123,
|
5823 |
+
"grad_norm": 0.9761250019073486,
|
5824 |
+
"learning_rate": 3.1098814146686834e-05,
|
5825 |
+
"loss": 1.0411,
|
5826 |
+
"step": 831
|
5827 |
+
},
|
5828 |
+
{
|
5829 |
+
"epoch": 0.0055300580589629144,
|
5830 |
+
"grad_norm": 1.040299415588379,
|
5831 |
+
"learning_rate": 3.098911365591129e-05,
|
5832 |
+
"loss": 0.8767,
|
5833 |
+
"step": 832
|
5834 |
+
},
|
5835 |
+
{
|
5836 |
+
"epoch": 0.005536704763360707,
|
5837 |
+
"grad_norm": 0.9226474165916443,
|
5838 |
+
"learning_rate": 3.087952003866492e-05,
|
5839 |
+
"loss": 0.6856,
|
5840 |
+
"step": 833
|
5841 |
+
},
|
5842 |
+
{
|
5843 |
+
"epoch": 0.005543351467758499,
|
5844 |
+
"grad_norm": 0.9735934138298035,
|
5845 |
+
"learning_rate": 3.077003391105033e-05,
|
5846 |
+
"loss": 0.4014,
|
5847 |
+
"step": 834
|
5848 |
+
},
|
5849 |
+
{
|
5850 |
+
"epoch": 0.0055499981721562905,
|
5851 |
+
"grad_norm": 1.027713418006897,
|
5852 |
+
"learning_rate": 3.0660655888565825e-05,
|
5853 |
+
"loss": 0.8747,
|
5854 |
+
"step": 835
|
5855 |
+
},
|
5856 |
+
{
|
5857 |
+
"epoch": 0.005556644876554082,
|
5858 |
+
"grad_norm": 1.1405994892120361,
|
5859 |
+
"learning_rate": 3.055138658610202e-05,
|
5860 |
+
"loss": 0.6844,
|
5861 |
+
"step": 836
|
5862 |
+
},
|
5863 |
+
{
|
5864 |
+
"epoch": 0.005563291580951875,
|
5865 |
+
"grad_norm": 1.0301321744918823,
|
5866 |
+
"learning_rate": 3.0442226617938307e-05,
|
5867 |
+
"loss": 0.6939,
|
5868 |
+
"step": 837
|
5869 |
+
},
|
5870 |
+
{
|
5871 |
+
"epoch": 0.005569938285349667,
|
5872 |
+
"grad_norm": 1.1772515773773193,
|
5873 |
+
"learning_rate": 3.0333176597739476e-05,
|
5874 |
+
"loss": 0.4741,
|
5875 |
+
"step": 838
|
5876 |
+
},
|
5877 |
+
{
|
5878 |
+
"epoch": 0.005576584989747458,
|
5879 |
+
"grad_norm": 1.322670340538025,
|
5880 |
+
"learning_rate": 3.022423713855218e-05,
|
5881 |
+
"loss": 0.6345,
|
5882 |
+
"step": 839
|
5883 |
+
},
|
5884 |
+
{
|
5885 |
+
"epoch": 0.00558323169414525,
|
5886 |
+
"grad_norm": 1.449883222579956,
|
5887 |
+
"learning_rate": 3.0115408852801535e-05,
|
5888 |
+
"loss": 1.018,
|
5889 |
+
"step": 840
|
5890 |
+
},
|
5891 |
+
{
|
5892 |
+
"epoch": 0.005589878398543043,
|
5893 |
+
"grad_norm": 1.06947660446167,
|
5894 |
+
"learning_rate": 3.00066923522877e-05,
|
5895 |
+
"loss": 0.6099,
|
5896 |
+
"step": 841
|
5897 |
+
},
|
5898 |
+
{
|
5899 |
+
"epoch": 0.0055965251029408344,
|
5900 |
+
"grad_norm": 1.6659729480743408,
|
5901 |
+
"learning_rate": 2.9898088248182366e-05,
|
5902 |
+
"loss": 0.6934,
|
5903 |
+
"step": 842
|
5904 |
+
},
|
5905 |
+
{
|
5906 |
+
"epoch": 0.005603171807338626,
|
5907 |
+
"grad_norm": 1.4211032390594482,
|
5908 |
+
"learning_rate": 2.9789597151025438e-05,
|
5909 |
+
"loss": 0.596,
|
5910 |
+
"step": 843
|
5911 |
+
},
|
5912 |
+
{
|
5913 |
+
"epoch": 0.005609818511736418,
|
5914 |
+
"grad_norm": 1.7341675758361816,
|
5915 |
+
"learning_rate": 2.968121967072145e-05,
|
5916 |
+
"loss": 0.9223,
|
5917 |
+
"step": 844
|
5918 |
+
},
|
5919 |
+
{
|
5920 |
+
"epoch": 0.0056164652161342105,
|
5921 |
+
"grad_norm": 1.3527710437774658,
|
5922 |
+
"learning_rate": 2.9572956416536267e-05,
|
5923 |
+
"loss": 0.4806,
|
5924 |
+
"step": 845
|
5925 |
+
},
|
5926 |
+
{
|
5927 |
+
"epoch": 0.005623111920532002,
|
5928 |
+
"grad_norm": 1.7114077806472778,
|
5929 |
+
"learning_rate": 2.946480799709358e-05,
|
5930 |
+
"loss": 0.7749,
|
5931 |
+
"step": 846
|
5932 |
+
},
|
5933 |
+
{
|
5934 |
+
"epoch": 0.005629758624929794,
|
5935 |
+
"grad_norm": 2.36133074760437,
|
5936 |
+
"learning_rate": 2.935677502037152e-05,
|
5937 |
+
"loss": 0.9725,
|
5938 |
+
"step": 847
|
5939 |
+
},
|
5940 |
+
{
|
5941 |
+
"epoch": 0.005636405329327586,
|
5942 |
+
"grad_norm": 1.736711025238037,
|
5943 |
+
"learning_rate": 2.924885809369926e-05,
|
5944 |
+
"loss": 0.9024,
|
5945 |
+
"step": 848
|
5946 |
+
},
|
5947 |
+
{
|
5948 |
+
"epoch": 0.005643052033725378,
|
5949 |
+
"grad_norm": 2.3337650299072266,
|
5950 |
+
"learning_rate": 2.9141057823753548e-05,
|
5951 |
+
"loss": 0.6143,
|
5952 |
+
"step": 849
|
5953 |
+
},
|
5954 |
+
{
|
5955 |
+
"epoch": 0.00564969873812317,
|
5956 |
+
"grad_norm": 4.121450424194336,
|
5957 |
+
"learning_rate": 2.9033374816555338e-05,
|
5958 |
+
"loss": 1.1379,
|
5959 |
+
"step": 850
|
5960 |
+
},
|
5961 |
+
{
|
5962 |
+
"epoch": 0.005656345442520962,
|
5963 |
+
"grad_norm": 0.3683931827545166,
|
5964 |
+
"learning_rate": 2.892580967746633e-05,
|
5965 |
+
"loss": 0.9842,
|
5966 |
+
"step": 851
|
5967 |
+
},
|
5968 |
+
{
|
5969 |
+
"epoch": 0.005662992146918754,
|
5970 |
+
"grad_norm": 0.49938151240348816,
|
5971 |
+
"learning_rate": 2.8818363011185647e-05,
|
5972 |
+
"loss": 0.9464,
|
5973 |
+
"step": 852
|
5974 |
+
},
|
5975 |
+
{
|
5976 |
+
"epoch": 0.005669638851316546,
|
5977 |
+
"grad_norm": 0.6275522708892822,
|
5978 |
+
"learning_rate": 2.8711035421746367e-05,
|
5979 |
+
"loss": 0.7492,
|
5980 |
+
"step": 853
|
5981 |
+
},
|
5982 |
+
{
|
5983 |
+
"epoch": 0.005676285555714338,
|
5984 |
+
"grad_norm": 0.5167999267578125,
|
5985 |
+
"learning_rate": 2.86038275125122e-05,
|
5986 |
+
"loss": 0.6106,
|
5987 |
+
"step": 854
|
5988 |
+
},
|
5989 |
+
{
|
5990 |
+
"epoch": 0.00568293226011213,
|
5991 |
+
"grad_norm": 0.7422475814819336,
|
5992 |
+
"learning_rate": 2.8496739886173995e-05,
|
5993 |
+
"loss": 0.713,
|
5994 |
+
"step": 855
|
5995 |
+
},
|
5996 |
+
{
|
5997 |
+
"epoch": 0.005689578964509922,
|
5998 |
+
"grad_norm": 0.6441875100135803,
|
5999 |
+
"learning_rate": 2.83897731447464e-05,
|
6000 |
+
"loss": 0.6015,
|
6001 |
+
"step": 856
|
6002 |
+
},
|
6003 |
+
{
|
6004 |
+
"epoch": 0.005696225668907714,
|
6005 |
+
"grad_norm": 0.7126911878585815,
|
6006 |
+
"learning_rate": 2.8282927889564542e-05,
|
6007 |
+
"loss": 0.7766,
|
6008 |
+
"step": 857
|
6009 |
+
},
|
6010 |
+
{
|
6011 |
+
"epoch": 0.005702872373305506,
|
6012 |
+
"grad_norm": 0.7499037981033325,
|
6013 |
+
"learning_rate": 2.8176204721280563e-05,
|
6014 |
+
"loss": 0.6112,
|
6015 |
+
"step": 858
|
6016 |
+
},
|
6017 |
+
{
|
6018 |
+
"epoch": 0.0057095190777032975,
|
6019 |
+
"grad_norm": 0.5948365330696106,
|
6020 |
+
"learning_rate": 2.806960423986026e-05,
|
6021 |
+
"loss": 0.8759,
|
6022 |
+
"step": 859
|
6023 |
+
},
|
6024 |
+
{
|
6025 |
+
"epoch": 0.00571616578210109,
|
6026 |
+
"grad_norm": 0.6240098476409912,
|
6027 |
+
"learning_rate": 2.7963127044579697e-05,
|
6028 |
+
"loss": 0.9862,
|
6029 |
+
"step": 860
|
6030 |
+
},
|
6031 |
+
{
|
6032 |
+
"epoch": 0.005722812486498882,
|
6033 |
+
"grad_norm": 0.7282832264900208,
|
6034 |
+
"learning_rate": 2.7856773734021913e-05,
|
6035 |
+
"loss": 0.6017,
|
6036 |
+
"step": 861
|
6037 |
+
},
|
6038 |
+
{
|
6039 |
+
"epoch": 0.005729459190896674,
|
6040 |
+
"grad_norm": 0.6713678240776062,
|
6041 |
+
"learning_rate": 2.7750544906073463e-05,
|
6042 |
+
"loss": 0.8692,
|
6043 |
+
"step": 862
|
6044 |
+
},
|
6045 |
+
{
|
6046 |
+
"epoch": 0.005736105895294465,
|
6047 |
+
"grad_norm": 0.6113742589950562,
|
6048 |
+
"learning_rate": 2.764444115792113e-05,
|
6049 |
+
"loss": 0.9001,
|
6050 |
+
"step": 863
|
6051 |
+
},
|
6052 |
+
{
|
6053 |
+
"epoch": 0.005742752599692258,
|
6054 |
+
"grad_norm": 0.6918427348136902,
|
6055 |
+
"learning_rate": 2.7538463086048504e-05,
|
6056 |
+
"loss": 0.8363,
|
6057 |
+
"step": 864
|
6058 |
+
},
|
6059 |
+
{
|
6060 |
+
"epoch": 0.00574939930409005,
|
6061 |
+
"grad_norm": 0.9629685282707214,
|
6062 |
+
"learning_rate": 2.743261128623269e-05,
|
6063 |
+
"loss": 0.9178,
|
6064 |
+
"step": 865
|
6065 |
+
},
|
6066 |
+
{
|
6067 |
+
"epoch": 0.005756046008487841,
|
6068 |
+
"grad_norm": 0.7158761024475098,
|
6069 |
+
"learning_rate": 2.7326886353540916e-05,
|
6070 |
+
"loss": 0.7963,
|
6071 |
+
"step": 866
|
6072 |
+
},
|
6073 |
+
{
|
6074 |
+
"epoch": 0.005762692712885633,
|
6075 |
+
"grad_norm": 0.6633831858634949,
|
6076 |
+
"learning_rate": 2.7221288882327167e-05,
|
6077 |
+
"loss": 0.7058,
|
6078 |
+
"step": 867
|
6079 |
+
},
|
6080 |
+
{
|
6081 |
+
"epoch": 0.005769339417283426,
|
6082 |
+
"grad_norm": 0.7890012264251709,
|
6083 |
+
"learning_rate": 2.711581946622893e-05,
|
6084 |
+
"loss": 0.7596,
|
6085 |
+
"step": 868
|
6086 |
+
},
|
6087 |
+
{
|
6088 |
+
"epoch": 0.0057759861216812175,
|
6089 |
+
"grad_norm": 0.743654727935791,
|
6090 |
+
"learning_rate": 2.701047869816381e-05,
|
6091 |
+
"loss": 0.9293,
|
6092 |
+
"step": 869
|
6093 |
+
},
|
6094 |
+
{
|
6095 |
+
"epoch": 0.005782632826079009,
|
6096 |
+
"grad_norm": 0.9647628664970398,
|
6097 |
+
"learning_rate": 2.6905267170326143e-05,
|
6098 |
+
"loss": 0.7642,
|
6099 |
+
"step": 870
|
6100 |
+
},
|
6101 |
+
{
|
6102 |
+
"epoch": 0.005789279530476801,
|
6103 |
+
"grad_norm": 0.7934918403625488,
|
6104 |
+
"learning_rate": 2.6800185474183725e-05,
|
6105 |
+
"loss": 0.9243,
|
6106 |
+
"step": 871
|
6107 |
+
},
|
6108 |
+
{
|
6109 |
+
"epoch": 0.005795926234874594,
|
6110 |
+
"grad_norm": 0.8000086545944214,
|
6111 |
+
"learning_rate": 2.6695234200474495e-05,
|
6112 |
+
"loss": 0.9556,
|
6113 |
+
"step": 872
|
6114 |
+
},
|
6115 |
+
{
|
6116 |
+
"epoch": 0.005802572939272385,
|
6117 |
+
"grad_norm": 0.774133563041687,
|
6118 |
+
"learning_rate": 2.6590413939203228e-05,
|
6119 |
+
"loss": 0.7859,
|
6120 |
+
"step": 873
|
6121 |
+
},
|
6122 |
+
{
|
6123 |
+
"epoch": 0.005809219643670177,
|
6124 |
+
"grad_norm": 0.994034469127655,
|
6125 |
+
"learning_rate": 2.6485725279638113e-05,
|
6126 |
+
"loss": 0.8363,
|
6127 |
+
"step": 874
|
6128 |
+
},
|
6129 |
+
{
|
6130 |
+
"epoch": 0.005815866348067969,
|
6131 |
+
"grad_norm": 0.8750971555709839,
|
6132 |
+
"learning_rate": 2.6381168810307533e-05,
|
6133 |
+
"loss": 0.92,
|
6134 |
+
"step": 875
|
6135 |
+
},
|
6136 |
+
{
|
6137 |
+
"epoch": 0.005822513052465761,
|
6138 |
+
"grad_norm": 1.1995213031768799,
|
6139 |
+
"learning_rate": 2.627674511899678e-05,
|
6140 |
+
"loss": 0.7646,
|
6141 |
+
"step": 876
|
6142 |
+
},
|
6143 |
+
{
|
6144 |
+
"epoch": 0.005829159756863553,
|
6145 |
+
"grad_norm": 0.8142304420471191,
|
6146 |
+
"learning_rate": 2.6172454792744706e-05,
|
6147 |
+
"loss": 0.867,
|
6148 |
+
"step": 877
|
6149 |
+
},
|
6150 |
+
{
|
6151 |
+
"epoch": 0.005835806461261345,
|
6152 |
+
"grad_norm": 0.9330615401268005,
|
6153 |
+
"learning_rate": 2.6068298417840364e-05,
|
6154 |
+
"loss": 0.8755,
|
6155 |
+
"step": 878
|
6156 |
+
},
|
6157 |
+
{
|
6158 |
+
"epoch": 0.005842453165659137,
|
6159 |
+
"grad_norm": 1.0684067010879517,
|
6160 |
+
"learning_rate": 2.596427657981979e-05,
|
6161 |
+
"loss": 0.8391,
|
6162 |
+
"step": 879
|
6163 |
+
},
|
6164 |
+
{
|
6165 |
+
"epoch": 0.005849099870056929,
|
6166 |
+
"grad_norm": 0.7989206910133362,
|
6167 |
+
"learning_rate": 2.5860389863462765e-05,
|
6168 |
+
"loss": 0.7431,
|
6169 |
+
"step": 880
|
6170 |
+
},
|
6171 |
+
{
|
6172 |
+
"epoch": 0.005855746574454721,
|
6173 |
+
"grad_norm": 0.8344511389732361,
|
6174 |
+
"learning_rate": 2.575663885278935e-05,
|
6175 |
+
"loss": 0.725,
|
6176 |
+
"step": 881
|
6177 |
+
},
|
6178 |
+
{
|
6179 |
+
"epoch": 0.005862393278852513,
|
6180 |
+
"grad_norm": 1.2674167156219482,
|
6181 |
+
"learning_rate": 2.5653024131056802e-05,
|
6182 |
+
"loss": 0.9218,
|
6183 |
+
"step": 882
|
6184 |
+
},
|
6185 |
+
{
|
6186 |
+
"epoch": 0.005869039983250305,
|
6187 |
+
"grad_norm": 1.206439733505249,
|
6188 |
+
"learning_rate": 2.5549546280756137e-05,
|
6189 |
+
"loss": 0.7342,
|
6190 |
+
"step": 883
|
6191 |
+
},
|
6192 |
+
{
|
6193 |
+
"epoch": 0.005875686687648097,
|
6194 |
+
"grad_norm": 1.1278153657913208,
|
6195 |
+
"learning_rate": 2.544620588360897e-05,
|
6196 |
+
"loss": 0.8005,
|
6197 |
+
"step": 884
|
6198 |
+
},
|
6199 |
+
{
|
6200 |
+
"epoch": 0.005882333392045889,
|
6201 |
+
"grad_norm": 1.1391501426696777,
|
6202 |
+
"learning_rate": 2.5343003520564158e-05,
|
6203 |
+
"loss": 0.8529,
|
6204 |
+
"step": 885
|
6205 |
+
},
|
6206 |
+
{
|
6207 |
+
"epoch": 0.0058889800964436806,
|
6208 |
+
"grad_norm": 0.9958661794662476,
|
6209 |
+
"learning_rate": 2.5239939771794563e-05,
|
6210 |
+
"loss": 0.7329,
|
6211 |
+
"step": 886
|
6212 |
+
},
|
6213 |
+
{
|
6214 |
+
"epoch": 0.005895626800841473,
|
6215 |
+
"grad_norm": 1.1319859027862549,
|
6216 |
+
"learning_rate": 2.5137015216693848e-05,
|
6217 |
+
"loss": 0.7973,
|
6218 |
+
"step": 887
|
6219 |
+
},
|
6220 |
+
{
|
6221 |
+
"epoch": 0.005902273505239265,
|
6222 |
+
"grad_norm": 1.1556273698806763,
|
6223 |
+
"learning_rate": 2.5034230433873153e-05,
|
6224 |
+
"loss": 0.8337,
|
6225 |
+
"step": 888
|
6226 |
+
},
|
6227 |
+
{
|
6228 |
+
"epoch": 0.005908920209637057,
|
6229 |
+
"grad_norm": 1.0099126100540161,
|
6230 |
+
"learning_rate": 2.4931586001157835e-05,
|
6231 |
+
"loss": 0.7135,
|
6232 |
+
"step": 889
|
6233 |
+
},
|
6234 |
+
{
|
6235 |
+
"epoch": 0.005915566914034848,
|
6236 |
+
"grad_norm": 1.1238211393356323,
|
6237 |
+
"learning_rate": 2.4829082495584242e-05,
|
6238 |
+
"loss": 0.5285,
|
6239 |
+
"step": 890
|
6240 |
+
},
|
6241 |
+
{
|
6242 |
+
"epoch": 0.005922213618432641,
|
6243 |
+
"grad_norm": 1.2216733694076538,
|
6244 |
+
"learning_rate": 2.4726720493396512e-05,
|
6245 |
+
"loss": 0.7336,
|
6246 |
+
"step": 891
|
6247 |
+
},
|
6248 |
+
{
|
6249 |
+
"epoch": 0.005928860322830433,
|
6250 |
+
"grad_norm": 1.766696572303772,
|
6251 |
+
"learning_rate": 2.4624500570043278e-05,
|
6252 |
+
"loss": 0.8474,
|
6253 |
+
"step": 892
|
6254 |
+
},
|
6255 |
+
{
|
6256 |
+
"epoch": 0.0059355070272282245,
|
6257 |
+
"grad_norm": 1.1542131900787354,
|
6258 |
+
"learning_rate": 2.4522423300174437e-05,
|
6259 |
+
"loss": 0.7923,
|
6260 |
+
"step": 893
|
6261 |
+
},
|
6262 |
+
{
|
6263 |
+
"epoch": 0.005942153731626016,
|
6264 |
+
"grad_norm": 1.332521915435791,
|
6265 |
+
"learning_rate": 2.4420489257637906e-05,
|
6266 |
+
"loss": 0.7754,
|
6267 |
+
"step": 894
|
6268 |
+
},
|
6269 |
+
{
|
6270 |
+
"epoch": 0.005948800436023809,
|
6271 |
+
"grad_norm": 1.6716405153274536,
|
6272 |
+
"learning_rate": 2.4318699015476493e-05,
|
6273 |
+
"loss": 0.7301,
|
6274 |
+
"step": 895
|
6275 |
+
},
|
6276 |
+
{
|
6277 |
+
"epoch": 0.0059554471404216006,
|
6278 |
+
"grad_norm": 1.6850249767303467,
|
6279 |
+
"learning_rate": 2.421705314592452e-05,
|
6280 |
+
"loss": 0.7965,
|
6281 |
+
"step": 896
|
6282 |
+
},
|
6283 |
+
{
|
6284 |
+
"epoch": 0.005962093844819392,
|
6285 |
+
"grad_norm": 1.4353415966033936,
|
6286 |
+
"learning_rate": 2.4115552220404765e-05,
|
6287 |
+
"loss": 0.6851,
|
6288 |
+
"step": 897
|
6289 |
+
},
|
6290 |
+
{
|
6291 |
+
"epoch": 0.005968740549217184,
|
6292 |
+
"grad_norm": 2.0517685413360596,
|
6293 |
+
"learning_rate": 2.4014196809525102e-05,
|
6294 |
+
"loss": 1.0993,
|
6295 |
+
"step": 898
|
6296 |
+
},
|
6297 |
+
{
|
6298 |
+
"epoch": 0.005975387253614977,
|
6299 |
+
"grad_norm": 2.061101198196411,
|
6300 |
+
"learning_rate": 2.391298748307544e-05,
|
6301 |
+
"loss": 0.6542,
|
6302 |
+
"step": 899
|
6303 |
+
},
|
6304 |
+
{
|
6305 |
+
"epoch": 0.005982033958012768,
|
6306 |
+
"grad_norm": 2.91339373588562,
|
6307 |
+
"learning_rate": 2.3811924810024384e-05,
|
6308 |
+
"loss": 0.7534,
|
6309 |
+
"step": 900
|
6310 |
+
},
|
6311 |
+
{
|
6312 |
+
"epoch": 0.00598868066241056,
|
6313 |
+
"grad_norm": 0.3465811312198639,
|
6314 |
+
"learning_rate": 2.3711009358516102e-05,
|
6315 |
+
"loss": 0.9504,
|
6316 |
+
"step": 901
|
6317 |
+
},
|
6318 |
+
{
|
6319 |
+
"epoch": 0.005995327366808352,
|
6320 |
+
"grad_norm": 0.5000481605529785,
|
6321 |
+
"learning_rate": 2.3610241695867163e-05,
|
6322 |
+
"loss": 0.8834,
|
6323 |
+
"step": 902
|
6324 |
+
},
|
6325 |
+
{
|
6326 |
+
"epoch": 0.0060019740712061445,
|
6327 |
+
"grad_norm": 0.871720552444458,
|
6328 |
+
"learning_rate": 2.3509622388563313e-05,
|
6329 |
+
"loss": 0.6535,
|
6330 |
+
"step": 903
|
6331 |
+
},
|
6332 |
+
{
|
6333 |
+
"epoch": 0.006008620775603936,
|
6334 |
+
"grad_norm": 0.9901016354560852,
|
6335 |
+
"learning_rate": 2.3409152002256247e-05,
|
6336 |
+
"loss": 1.0592,
|
6337 |
+
"step": 904
|
6338 |
+
},
|
6339 |
+
{
|
6340 |
+
"epoch": 0.006015267480001728,
|
6341 |
+
"grad_norm": 0.6650891304016113,
|
6342 |
+
"learning_rate": 2.3308831101760486e-05,
|
6343 |
+
"loss": 0.9605,
|
6344 |
+
"step": 905
|
6345 |
+
},
|
6346 |
+
{
|
6347 |
+
"epoch": 0.00602191418439952,
|
6348 |
+
"grad_norm": 0.6247522830963135,
|
6349 |
+
"learning_rate": 2.3208660251050158e-05,
|
6350 |
+
"loss": 0.8668,
|
6351 |
+
"step": 906
|
6352 |
+
},
|
6353 |
+
{
|
6354 |
+
"epoch": 0.006028560888797312,
|
6355 |
+
"grad_norm": 0.8756898641586304,
|
6356 |
+
"learning_rate": 2.3108640013255968e-05,
|
6357 |
+
"loss": 0.9702,
|
6358 |
+
"step": 907
|
6359 |
+
},
|
6360 |
+
{
|
6361 |
+
"epoch": 0.006035207593195104,
|
6362 |
+
"grad_norm": 0.618049144744873,
|
6363 |
+
"learning_rate": 2.30087709506618e-05,
|
6364 |
+
"loss": 0.7552,
|
6365 |
+
"step": 908
|
6366 |
+
},
|
6367 |
+
{
|
6368 |
+
"epoch": 0.006041854297592896,
|
6369 |
+
"grad_norm": 0.8775025010108948,
|
6370 |
+
"learning_rate": 2.2909053624701715e-05,
|
6371 |
+
"loss": 0.7956,
|
6372 |
+
"step": 909
|
6373 |
+
},
|
6374 |
+
{
|
6375 |
+
"epoch": 0.006048501001990688,
|
6376 |
+
"grad_norm": 0.6528658866882324,
|
6377 |
+
"learning_rate": 2.2809488595956745e-05,
|
6378 |
+
"loss": 0.7783,
|
6379 |
+
"step": 910
|
6380 |
+
},
|
6381 |
+
{
|
6382 |
+
"epoch": 0.00605514770638848,
|
6383 |
+
"grad_norm": 0.5745322704315186,
|
6384 |
+
"learning_rate": 2.2710076424151788e-05,
|
6385 |
+
"loss": 0.9813,
|
6386 |
+
"step": 911
|
6387 |
+
},
|
6388 |
+
{
|
6389 |
+
"epoch": 0.006061794410786272,
|
6390 |
+
"grad_norm": 0.8731324076652527,
|
6391 |
+
"learning_rate": 2.2610817668152413e-05,
|
6392 |
+
"loss": 0.8946,
|
6393 |
+
"step": 912
|
6394 |
+
},
|
6395 |
+
{
|
6396 |
+
"epoch": 0.006068441115184064,
|
6397 |
+
"grad_norm": 0.7682494521141052,
|
6398 |
+
"learning_rate": 2.251171288596171e-05,
|
6399 |
+
"loss": 0.9534,
|
6400 |
+
"step": 913
|
6401 |
+
},
|
6402 |
+
{
|
6403 |
+
"epoch": 0.006075087819581856,
|
6404 |
+
"grad_norm": 0.7490772008895874,
|
6405 |
+
"learning_rate": 2.241276263471716e-05,
|
6406 |
+
"loss": 0.8044,
|
6407 |
+
"step": 914
|
6408 |
+
},
|
6409 |
+
{
|
6410 |
+
"epoch": 0.006081734523979648,
|
6411 |
+
"grad_norm": 0.7320804595947266,
|
6412 |
+
"learning_rate": 2.2313967470687593e-05,
|
6413 |
+
"loss": 0.7225,
|
6414 |
+
"step": 915
|
6415 |
+
},
|
6416 |
+
{
|
6417 |
+
"epoch": 0.00608838122837744,
|
6418 |
+
"grad_norm": 0.8209467530250549,
|
6419 |
+
"learning_rate": 2.2215327949269886e-05,
|
6420 |
+
"loss": 0.773,
|
6421 |
+
"step": 916
|
6422 |
+
},
|
6423 |
+
{
|
6424 |
+
"epoch": 0.0060950279327752314,
|
6425 |
+
"grad_norm": 0.7817612290382385,
|
6426 |
+
"learning_rate": 2.2116844624986037e-05,
|
6427 |
+
"loss": 0.8568,
|
6428 |
+
"step": 917
|
6429 |
+
},
|
6430 |
+
{
|
6431 |
+
"epoch": 0.006101674637173024,
|
6432 |
+
"grad_norm": 0.7729702591896057,
|
6433 |
+
"learning_rate": 2.201851805147986e-05,
|
6434 |
+
"loss": 0.9524,
|
6435 |
+
"step": 918
|
6436 |
+
},
|
6437 |
+
{
|
6438 |
+
"epoch": 0.006108321341570816,
|
6439 |
+
"grad_norm": 0.8721408247947693,
|
6440 |
+
"learning_rate": 2.192034878151404e-05,
|
6441 |
+
"loss": 0.9534,
|
6442 |
+
"step": 919
|
6443 |
+
},
|
6444 |
+
{
|
6445 |
+
"epoch": 0.0061149680459686075,
|
6446 |
+
"grad_norm": 0.7396866083145142,
|
6447 |
+
"learning_rate": 2.1822337366966898e-05,
|
6448 |
+
"loss": 0.8427,
|
6449 |
+
"step": 920
|
6450 |
+
},
|
6451 |
+
{
|
6452 |
+
"epoch": 0.006121614750366399,
|
6453 |
+
"grad_norm": 0.8488925695419312,
|
6454 |
+
"learning_rate": 2.1724484358829338e-05,
|
6455 |
+
"loss": 0.8558,
|
6456 |
+
"step": 921
|
6457 |
+
},
|
6458 |
+
{
|
6459 |
+
"epoch": 0.006128261454764192,
|
6460 |
+
"grad_norm": 0.8868795037269592,
|
6461 |
+
"learning_rate": 2.1626790307201778e-05,
|
6462 |
+
"loss": 0.6768,
|
6463 |
+
"step": 922
|
6464 |
+
},
|
6465 |
+
{
|
6466 |
+
"epoch": 0.006134908159161984,
|
6467 |
+
"grad_norm": 0.9627169966697693,
|
6468 |
+
"learning_rate": 2.1529255761291052e-05,
|
6469 |
+
"loss": 0.6102,
|
6470 |
+
"step": 923
|
6471 |
+
},
|
6472 |
+
{
|
6473 |
+
"epoch": 0.006141554863559775,
|
6474 |
+
"grad_norm": 0.9614560604095459,
|
6475 |
+
"learning_rate": 2.1431881269407233e-05,
|
6476 |
+
"loss": 0.7419,
|
6477 |
+
"step": 924
|
6478 |
+
},
|
6479 |
+
{
|
6480 |
+
"epoch": 0.006148201567957567,
|
6481 |
+
"grad_norm": 0.9442191123962402,
|
6482 |
+
"learning_rate": 2.1334667378960644e-05,
|
6483 |
+
"loss": 0.9619,
|
6484 |
+
"step": 925
|
6485 |
+
},
|
6486 |
+
{
|
6487 |
+
"epoch": 0.00615484827235536,
|
6488 |
+
"grad_norm": 0.8390161991119385,
|
6489 |
+
"learning_rate": 2.123761463645878e-05,
|
6490 |
+
"loss": 0.8088,
|
6491 |
+
"step": 926
|
6492 |
+
},
|
6493 |
+
{
|
6494 |
+
"epoch": 0.0061614949767531514,
|
6495 |
+
"grad_norm": 0.9244080781936646,
|
6496 |
+
"learning_rate": 2.114072358750321e-05,
|
6497 |
+
"loss": 0.875,
|
6498 |
+
"step": 927
|
6499 |
+
},
|
6500 |
+
{
|
6501 |
+
"epoch": 0.006168141681150943,
|
6502 |
+
"grad_norm": 0.9828105568885803,
|
6503 |
+
"learning_rate": 2.1043994776786473e-05,
|
6504 |
+
"loss": 0.7579,
|
6505 |
+
"step": 928
|
6506 |
+
},
|
6507 |
+
{
|
6508 |
+
"epoch": 0.006174788385548735,
|
6509 |
+
"grad_norm": 0.8843576908111572,
|
6510 |
+
"learning_rate": 2.0947428748089066e-05,
|
6511 |
+
"loss": 0.6549,
|
6512 |
+
"step": 929
|
6513 |
+
},
|
6514 |
+
{
|
6515 |
+
"epoch": 0.0061814350899465275,
|
6516 |
+
"grad_norm": 1.1693994998931885,
|
6517 |
+
"learning_rate": 2.0851026044276406e-05,
|
6518 |
+
"loss": 0.9206,
|
6519 |
+
"step": 930
|
6520 |
+
},
|
6521 |
+
{
|
6522 |
+
"epoch": 0.006188081794344319,
|
6523 |
+
"grad_norm": 1.1045140027999878,
|
6524 |
+
"learning_rate": 2.0754787207295696e-05,
|
6525 |
+
"loss": 0.8276,
|
6526 |
+
"step": 931
|
6527 |
+
},
|
6528 |
+
{
|
6529 |
+
"epoch": 0.006194728498742111,
|
6530 |
+
"grad_norm": 0.9659402966499329,
|
6531 |
+
"learning_rate": 2.0658712778172982e-05,
|
6532 |
+
"loss": 0.8977,
|
6533 |
+
"step": 932
|
6534 |
+
},
|
6535 |
+
{
|
6536 |
+
"epoch": 0.006201375203139903,
|
6537 |
+
"grad_norm": 0.9211580157279968,
|
6538 |
+
"learning_rate": 2.0562803297009996e-05,
|
6539 |
+
"loss": 0.7761,
|
6540 |
+
"step": 933
|
6541 |
+
},
|
6542 |
+
{
|
6543 |
+
"epoch": 0.006208021907537695,
|
6544 |
+
"grad_norm": 1.2015657424926758,
|
6545 |
+
"learning_rate": 2.0467059302981263e-05,
|
6546 |
+
"loss": 0.8824,
|
6547 |
+
"step": 934
|
6548 |
+
},
|
6549 |
+
{
|
6550 |
+
"epoch": 0.006214668611935487,
|
6551 |
+
"grad_norm": 1.3020975589752197,
|
6552 |
+
"learning_rate": 2.0371481334330912e-05,
|
6553 |
+
"loss": 0.9115,
|
6554 |
+
"step": 935
|
6555 |
+
},
|
6556 |
+
{
|
6557 |
+
"epoch": 0.006221315316333279,
|
6558 |
+
"grad_norm": 0.8667546510696411,
|
6559 |
+
"learning_rate": 2.0276069928369746e-05,
|
6560 |
+
"loss": 0.5838,
|
6561 |
+
"step": 936
|
6562 |
+
},
|
6563 |
+
{
|
6564 |
+
"epoch": 0.006227962020731071,
|
6565 |
+
"grad_norm": 1.300489068031311,
|
6566 |
+
"learning_rate": 2.0180825621472227e-05,
|
6567 |
+
"loss": 0.6866,
|
6568 |
+
"step": 937
|
6569 |
+
},
|
6570 |
+
{
|
6571 |
+
"epoch": 0.006234608725128863,
|
6572 |
+
"grad_norm": 1.0658702850341797,
|
6573 |
+
"learning_rate": 2.0085748949073446e-05,
|
6574 |
+
"loss": 0.795,
|
6575 |
+
"step": 938
|
6576 |
+
},
|
6577 |
+
{
|
6578 |
+
"epoch": 0.006241255429526655,
|
6579 |
+
"grad_norm": 1.0691542625427246,
|
6580 |
+
"learning_rate": 1.9990840445666054e-05,
|
6581 |
+
"loss": 0.88,
|
6582 |
+
"step": 939
|
6583 |
+
},
|
6584 |
+
{
|
6585 |
+
"epoch": 0.006247902133924447,
|
6586 |
+
"grad_norm": 1.6725701093673706,
|
6587 |
+
"learning_rate": 1.9896100644797317e-05,
|
6588 |
+
"loss": 1.0337,
|
6589 |
+
"step": 940
|
6590 |
+
},
|
6591 |
+
{
|
6592 |
+
"epoch": 0.006254548838322239,
|
6593 |
+
"grad_norm": 1.0296176671981812,
|
6594 |
+
"learning_rate": 1.9801530079066137e-05,
|
6595 |
+
"loss": 0.4251,
|
6596 |
+
"step": 941
|
6597 |
+
},
|
6598 |
+
{
|
6599 |
+
"epoch": 0.006261195542720031,
|
6600 |
+
"grad_norm": 1.954781174659729,
|
6601 |
+
"learning_rate": 1.9707129280120036e-05,
|
6602 |
+
"loss": 0.7169,
|
6603 |
+
"step": 942
|
6604 |
+
},
|
6605 |
+
{
|
6606 |
+
"epoch": 0.006267842247117823,
|
6607 |
+
"grad_norm": 1.4644161462783813,
|
6608 |
+
"learning_rate": 1.96128987786521e-05,
|
6609 |
+
"loss": 0.49,
|
6610 |
+
"step": 943
|
6611 |
+
},
|
6612 |
+
{
|
6613 |
+
"epoch": 0.0062744889515156145,
|
6614 |
+
"grad_norm": 1.8161664009094238,
|
6615 |
+
"learning_rate": 1.9518839104398075e-05,
|
6616 |
+
"loss": 0.9943,
|
6617 |
+
"step": 944
|
6618 |
+
},
|
6619 |
+
{
|
6620 |
+
"epoch": 0.006281135655913407,
|
6621 |
+
"grad_norm": 1.5679110288619995,
|
6622 |
+
"learning_rate": 1.942495078613341e-05,
|
6623 |
+
"loss": 0.8934,
|
6624 |
+
"step": 945
|
6625 |
+
},
|
6626 |
+
{
|
6627 |
+
"epoch": 0.006287782360311199,
|
6628 |
+
"grad_norm": 1.837007999420166,
|
6629 |
+
"learning_rate": 1.933123435167018e-05,
|
6630 |
+
"loss": 0.8163,
|
6631 |
+
"step": 946
|
6632 |
+
},
|
6633 |
+
{
|
6634 |
+
"epoch": 0.006294429064708991,
|
6635 |
+
"grad_norm": 1.9493184089660645,
|
6636 |
+
"learning_rate": 1.9237690327854236e-05,
|
6637 |
+
"loss": 0.9895,
|
6638 |
+
"step": 947
|
6639 |
+
},
|
6640 |
+
{
|
6641 |
+
"epoch": 0.006301075769106782,
|
6642 |
+
"grad_norm": 1.4512784481048584,
|
6643 |
+
"learning_rate": 1.9144319240562125e-05,
|
6644 |
+
"loss": 0.6742,
|
6645 |
+
"step": 948
|
6646 |
+
},
|
6647 |
+
{
|
6648 |
+
"epoch": 0.006307722473504575,
|
6649 |
+
"grad_norm": 1.9372994899749756,
|
6650 |
+
"learning_rate": 1.9051121614698276e-05,
|
6651 |
+
"loss": 0.6534,
|
6652 |
+
"step": 949
|
6653 |
+
},
|
6654 |
+
{
|
6655 |
+
"epoch": 0.006314369177902367,
|
6656 |
+
"grad_norm": 4.261740207672119,
|
6657 |
+
"learning_rate": 1.8958097974191907e-05,
|
6658 |
+
"loss": 0.7541,
|
6659 |
+
"step": 950
|
6660 |
+
},
|
6661 |
+
{
|
6662 |
+
"epoch": 0.006321015882300158,
|
6663 |
+
"grad_norm": 0.39013972878456116,
|
6664 |
+
"learning_rate": 1.8865248841994143e-05,
|
6665 |
+
"loss": 0.7921,
|
6666 |
+
"step": 951
|
6667 |
+
},
|
6668 |
+
{
|
6669 |
+
"epoch": 0.00632766258669795,
|
6670 |
+
"grad_norm": 0.5134124755859375,
|
6671 |
+
"learning_rate": 1.877257474007511e-05,
|
6672 |
+
"loss": 0.961,
|
6673 |
+
"step": 952
|
6674 |
+
},
|
6675 |
+
{
|
6676 |
+
"epoch": 0.006334309291095743,
|
6677 |
+
"grad_norm": 0.9366634488105774,
|
6678 |
+
"learning_rate": 1.8680076189420964e-05,
|
6679 |
+
"loss": 0.8541,
|
6680 |
+
"step": 953
|
6681 |
+
},
|
6682 |
+
{
|
6683 |
+
"epoch": 0.0063409559954935345,
|
6684 |
+
"grad_norm": 0.532407820224762,
|
6685 |
+
"learning_rate": 1.858775371003094e-05,
|
6686 |
+
"loss": 1.0254,
|
6687 |
+
"step": 954
|
6688 |
+
},
|
6689 |
+
{
|
6690 |
+
"epoch": 0.006347602699891326,
|
6691 |
+
"grad_norm": 0.7640518546104431,
|
6692 |
+
"learning_rate": 1.849560782091445e-05,
|
6693 |
+
"loss": 0.9868,
|
6694 |
+
"step": 955
|
6695 |
+
},
|
6696 |
+
{
|
6697 |
+
"epoch": 0.006354249404289118,
|
6698 |
+
"grad_norm": 0.49766799807548523,
|
6699 |
+
"learning_rate": 1.840363904008819e-05,
|
6700 |
+
"loss": 0.8374,
|
6701 |
+
"step": 956
|
6702 |
+
},
|
6703 |
+
{
|
6704 |
+
"epoch": 0.006360896108686911,
|
6705 |
+
"grad_norm": 0.7397210597991943,
|
6706 |
+
"learning_rate": 1.831184788457321e-05,
|
6707 |
+
"loss": 0.8034,
|
6708 |
+
"step": 957
|
6709 |
+
},
|
6710 |
+
{
|
6711 |
+
"epoch": 0.006367542813084702,
|
6712 |
+
"grad_norm": 0.5797079801559448,
|
6713 |
+
"learning_rate": 1.8220234870391983e-05,
|
6714 |
+
"loss": 1.059,
|
6715 |
+
"step": 958
|
6716 |
+
},
|
6717 |
+
{
|
6718 |
+
"epoch": 0.006374189517482494,
|
6719 |
+
"grad_norm": 0.634558379650116,
|
6720 |
+
"learning_rate": 1.8128800512565513e-05,
|
6721 |
+
"loss": 0.8126,
|
6722 |
+
"step": 959
|
6723 |
+
},
|
6724 |
+
{
|
6725 |
+
"epoch": 0.006380836221880286,
|
6726 |
+
"grad_norm": 0.604607343673706,
|
6727 |
+
"learning_rate": 1.8037545325110504e-05,
|
6728 |
+
"loss": 0.9754,
|
6729 |
+
"step": 960
|
6730 |
+
},
|
6731 |
+
{
|
6732 |
+
"epoch": 0.006387482926278078,
|
6733 |
+
"grad_norm": 0.6816580295562744,
|
6734 |
+
"learning_rate": 1.7946469821036354e-05,
|
6735 |
+
"loss": 0.7716,
|
6736 |
+
"step": 961
|
6737 |
+
},
|
6738 |
+
{
|
6739 |
+
"epoch": 0.00639412963067587,
|
6740 |
+
"grad_norm": 0.6408112049102783,
|
6741 |
+
"learning_rate": 1.7855574512342397e-05,
|
6742 |
+
"loss": 0.6089,
|
6743 |
+
"step": 962
|
6744 |
+
},
|
6745 |
+
{
|
6746 |
+
"epoch": 0.006400776335073662,
|
6747 |
+
"grad_norm": 0.609083890914917,
|
6748 |
+
"learning_rate": 1.7764859910014893e-05,
|
6749 |
+
"loss": 0.9005,
|
6750 |
+
"step": 963
|
6751 |
+
},
|
6752 |
+
{
|
6753 |
+
"epoch": 0.006407423039471454,
|
6754 |
+
"grad_norm": 1.0609853267669678,
|
6755 |
+
"learning_rate": 1.7674326524024298e-05,
|
6756 |
+
"loss": 0.764,
|
6757 |
+
"step": 964
|
6758 |
+
},
|
6759 |
+
{
|
6760 |
+
"epoch": 0.006414069743869246,
|
6761 |
+
"grad_norm": 0.5911771059036255,
|
6762 |
+
"learning_rate": 1.7583974863322274e-05,
|
6763 |
+
"loss": 1.0236,
|
6764 |
+
"step": 965
|
6765 |
+
},
|
6766 |
+
{
|
6767 |
+
"epoch": 0.006420716448267038,
|
6768 |
+
"grad_norm": 0.6928749084472656,
|
6769 |
+
"learning_rate": 1.7493805435838874e-05,
|
6770 |
+
"loss": 0.7597,
|
6771 |
+
"step": 966
|
6772 |
+
},
|
6773 |
+
{
|
6774 |
+
"epoch": 0.00642736315266483,
|
6775 |
+
"grad_norm": 0.8709490299224854,
|
6776 |
+
"learning_rate": 1.7403818748479715e-05,
|
6777 |
+
"loss": 0.6517,
|
6778 |
+
"step": 967
|
6779 |
+
},
|
6780 |
+
{
|
6781 |
+
"epoch": 0.006434009857062622,
|
6782 |
+
"grad_norm": 0.6218889355659485,
|
6783 |
+
"learning_rate": 1.7314015307123133e-05,
|
6784 |
+
"loss": 0.9283,
|
6785 |
+
"step": 968
|
6786 |
+
},
|
6787 |
+
{
|
6788 |
+
"epoch": 0.006440656561460414,
|
6789 |
+
"grad_norm": 0.7528246641159058,
|
6790 |
+
"learning_rate": 1.7224395616617235e-05,
|
6791 |
+
"loss": 0.7588,
|
6792 |
+
"step": 969
|
6793 |
+
},
|
6794 |
+
{
|
6795 |
+
"epoch": 0.006447303265858206,
|
6796 |
+
"grad_norm": 0.7978760004043579,
|
6797 |
+
"learning_rate": 1.713496018077717e-05,
|
6798 |
+
"loss": 0.6017,
|
6799 |
+
"step": 970
|
6800 |
+
},
|
6801 |
+
{
|
6802 |
+
"epoch": 0.0064539499702559976,
|
6803 |
+
"grad_norm": 0.8703947067260742,
|
6804 |
+
"learning_rate": 1.7045709502382285e-05,
|
6805 |
+
"loss": 0.8674,
|
6806 |
+
"step": 971
|
6807 |
+
},
|
6808 |
+
{
|
6809 |
+
"epoch": 0.00646059667465379,
|
6810 |
+
"grad_norm": 0.8785549402236938,
|
6811 |
+
"learning_rate": 1.6956644083173262e-05,
|
6812 |
+
"loss": 0.6892,
|
6813 |
+
"step": 972
|
6814 |
+
},
|
6815 |
+
{
|
6816 |
+
"epoch": 0.006467243379051582,
|
6817 |
+
"grad_norm": 0.9214485883712769,
|
6818 |
+
"learning_rate": 1.6867764423849296e-05,
|
6819 |
+
"loss": 0.6507,
|
6820 |
+
"step": 973
|
6821 |
+
},
|
6822 |
+
{
|
6823 |
+
"epoch": 0.006473890083449374,
|
6824 |
+
"grad_norm": 0.7563501000404358,
|
6825 |
+
"learning_rate": 1.677907102406529e-05,
|
6826 |
+
"loss": 0.7163,
|
6827 |
+
"step": 974
|
6828 |
+
},
|
6829 |
+
{
|
6830 |
+
"epoch": 0.006480536787847165,
|
6831 |
+
"grad_norm": 0.9271652102470398,
|
6832 |
+
"learning_rate": 1.66905643824291e-05,
|
6833 |
+
"loss": 0.8399,
|
6834 |
+
"step": 975
|
6835 |
+
},
|
6836 |
+
{
|
6837 |
+
"epoch": 0.006487183492244958,
|
6838 |
+
"grad_norm": 0.906379759311676,
|
6839 |
+
"learning_rate": 1.660224499649861e-05,
|
6840 |
+
"loss": 0.6255,
|
6841 |
+
"step": 976
|
6842 |
+
},
|
6843 |
+
{
|
6844 |
+
"epoch": 0.00649383019664275,
|
6845 |
+
"grad_norm": 0.7881162762641907,
|
6846 |
+
"learning_rate": 1.6514113362779092e-05,
|
6847 |
+
"loss": 0.7797,
|
6848 |
+
"step": 977
|
6849 |
+
},
|
6850 |
+
{
|
6851 |
+
"epoch": 0.0065004769010405415,
|
6852 |
+
"grad_norm": 0.9204055666923523,
|
6853 |
+
"learning_rate": 1.6426169976720245e-05,
|
6854 |
+
"loss": 0.8043,
|
6855 |
+
"step": 978
|
6856 |
+
},
|
6857 |
+
{
|
6858 |
+
"epoch": 0.006507123605438333,
|
6859 |
+
"grad_norm": 0.9636942148208618,
|
6860 |
+
"learning_rate": 1.6338415332713576e-05,
|
6861 |
+
"loss": 0.8895,
|
6862 |
+
"step": 979
|
6863 |
+
},
|
6864 |
+
{
|
6865 |
+
"epoch": 0.006513770309836126,
|
6866 |
+
"grad_norm": 0.8163102269172668,
|
6867 |
+
"learning_rate": 1.6250849924089484e-05,
|
6868 |
+
"loss": 0.9835,
|
6869 |
+
"step": 980
|
6870 |
+
},
|
6871 |
+
{
|
6872 |
+
"epoch": 0.006520417014233918,
|
6873 |
+
"grad_norm": 1.0948930978775024,
|
6874 |
+
"learning_rate": 1.616347424311455e-05,
|
6875 |
+
"loss": 0.58,
|
6876 |
+
"step": 981
|
6877 |
+
},
|
6878 |
+
{
|
6879 |
+
"epoch": 0.006527063718631709,
|
6880 |
+
"grad_norm": 1.122653603553772,
|
6881 |
+
"learning_rate": 1.607628878098879e-05,
|
6882 |
+
"loss": 0.8378,
|
6883 |
+
"step": 982
|
6884 |
+
},
|
6885 |
+
{
|
6886 |
+
"epoch": 0.006533710423029501,
|
6887 |
+
"grad_norm": 0.9559472799301147,
|
6888 |
+
"learning_rate": 1.5989294027842867e-05,
|
6889 |
+
"loss": 0.6431,
|
6890 |
+
"step": 983
|
6891 |
+
},
|
6892 |
+
{
|
6893 |
+
"epoch": 0.006540357127427294,
|
6894 |
+
"grad_norm": 1.3170665502548218,
|
6895 |
+
"learning_rate": 1.59024904727353e-05,
|
6896 |
+
"loss": 0.4975,
|
6897 |
+
"step": 984
|
6898 |
+
},
|
6899 |
+
{
|
6900 |
+
"epoch": 0.006547003831825085,
|
6901 |
+
"grad_norm": 1.3573777675628662,
|
6902 |
+
"learning_rate": 1.581587860364977e-05,
|
6903 |
+
"loss": 0.7293,
|
6904 |
+
"step": 985
|
6905 |
+
},
|
6906 |
+
{
|
6907 |
+
"epoch": 0.006553650536222877,
|
6908 |
+
"grad_norm": 0.9671066403388977,
|
6909 |
+
"learning_rate": 1.5729458907492327e-05,
|
6910 |
+
"loss": 0.9002,
|
6911 |
+
"step": 986
|
6912 |
+
},
|
6913 |
+
{
|
6914 |
+
"epoch": 0.006560297240620669,
|
6915 |
+
"grad_norm": 1.218948245048523,
|
6916 |
+
"learning_rate": 1.5643231870088772e-05,
|
6917 |
+
"loss": 0.8514,
|
6918 |
+
"step": 987
|
6919 |
+
},
|
6920 |
+
{
|
6921 |
+
"epoch": 0.0065669439450184615,
|
6922 |
+
"grad_norm": 1.1144795417785645,
|
6923 |
+
"learning_rate": 1.5557197976181744e-05,
|
6924 |
+
"loss": 0.8713,
|
6925 |
+
"step": 988
|
6926 |
+
},
|
6927 |
+
{
|
6928 |
+
"epoch": 0.006573590649416253,
|
6929 |
+
"grad_norm": 1.354432463645935,
|
6930 |
+
"learning_rate": 1.5471357709428093e-05,
|
6931 |
+
"loss": 0.9549,
|
6932 |
+
"step": 989
|
6933 |
+
},
|
6934 |
+
{
|
6935 |
+
"epoch": 0.006580237353814045,
|
6936 |
+
"grad_norm": 1.4001054763793945,
|
6937 |
+
"learning_rate": 1.5385711552396227e-05,
|
6938 |
+
"loss": 1.1086,
|
6939 |
+
"step": 990
|
6940 |
+
},
|
6941 |
+
{
|
6942 |
+
"epoch": 0.006586884058211837,
|
6943 |
+
"grad_norm": 1.5661379098892212,
|
6944 |
+
"learning_rate": 1.5300259986563232e-05,
|
6945 |
+
"loss": 0.8944,
|
6946 |
+
"step": 991
|
6947 |
+
},
|
6948 |
+
{
|
6949 |
+
"epoch": 0.006593530762609629,
|
6950 |
+
"grad_norm": 1.2172316312789917,
|
6951 |
+
"learning_rate": 1.5215003492312351e-05,
|
6952 |
+
"loss": 1.0221,
|
6953 |
+
"step": 992
|
6954 |
+
},
|
6955 |
+
{
|
6956 |
+
"epoch": 0.006600177467007421,
|
6957 |
+
"grad_norm": 1.4772396087646484,
|
6958 |
+
"learning_rate": 1.5129942548930115e-05,
|
6959 |
+
"loss": 0.4565,
|
6960 |
+
"step": 993
|
6961 |
+
},
|
6962 |
+
{
|
6963 |
+
"epoch": 0.006606824171405213,
|
6964 |
+
"grad_norm": 1.6136627197265625,
|
6965 |
+
"learning_rate": 1.5045077634603787e-05,
|
6966 |
+
"loss": 0.9532,
|
6967 |
+
"step": 994
|
6968 |
+
},
|
6969 |
+
{
|
6970 |
+
"epoch": 0.006613470875803005,
|
6971 |
+
"grad_norm": 1.4357903003692627,
|
6972 |
+
"learning_rate": 1.4960409226418576e-05,
|
6973 |
+
"loss": 0.5671,
|
6974 |
+
"step": 995
|
6975 |
+
},
|
6976 |
+
{
|
6977 |
+
"epoch": 0.006620117580200797,
|
6978 |
+
"grad_norm": 1.6323310136795044,
|
6979 |
+
"learning_rate": 1.4875937800354988e-05,
|
6980 |
+
"loss": 0.8964,
|
6981 |
+
"step": 996
|
6982 |
+
},
|
6983 |
+
{
|
6984 |
+
"epoch": 0.006626764284598589,
|
6985 |
+
"grad_norm": 1.6246354579925537,
|
6986 |
+
"learning_rate": 1.4791663831286184e-05,
|
6987 |
+
"loss": 0.9035,
|
6988 |
+
"step": 997
|
6989 |
+
},
|
6990 |
+
{
|
6991 |
+
"epoch": 0.006633410988996381,
|
6992 |
+
"grad_norm": 2.0185391902923584,
|
6993 |
+
"learning_rate": 1.470758779297528e-05,
|
6994 |
+
"loss": 0.9129,
|
6995 |
+
"step": 998
|
6996 |
+
},
|
6997 |
+
{
|
6998 |
+
"epoch": 0.006640057693394173,
|
6999 |
+
"grad_norm": 1.7859017848968506,
|
7000 |
+
"learning_rate": 1.4623710158072663e-05,
|
7001 |
+
"loss": 0.772,
|
7002 |
+
"step": 999
|
7003 |
}
|
7004 |
],
|
7005 |
"logging_steps": 1,
|
|
|
7019 |
"attributes": {}
|
7020 |
}
|
7021 |
},
|
7022 |
+
"total_flos": 2.7808050135957504e+17,
|
7023 |
"train_batch_size": 4,
|
7024 |
"trial_name": null,
|
7025 |
"trial_params": null
|