Training in progress, step 1011, 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 159967880
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b662b0fc7aac68525d0f688f390c8ad9870852fbf2dd6cb97fbfc1ece7c5d598
|
| 3 |
size 159967880
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 320194002
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b7f474172a83fcd8820c4f98277a6048b8181d77934e81481cd77060814392c6
|
| 3 |
size 320194002
|
last-checkpoint/rng_state_0.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14512
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:58265f7d6bfa2f240115921535bdab2d9168829e78e7fcd7bbfd4c601b33fc1e
|
| 3 |
size 14512
|
last-checkpoint/rng_state_1.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14512
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7d3366f225c3d2f610762730db695898cfa0429711eba5664413a852bccb0150
|
| 3 |
size 14512
|
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 sha256:7d2ae9412584d60dfb3a14dc71755f8ba52a5c928738cf6019a6761ddc1e1565
|
| 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": 337,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -4749,6 +4749,2373 @@
|
|
| 4749 |
"eval_samples_per_second": 5.949,
|
| 4750 |
"eval_steps_per_second": 2.975,
|
| 4751 |
"step": 674
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4752 |
}
|
| 4753 |
],
|
| 4754 |
"logging_steps": 1,
|
|
@@ -4768,7 +7135,7 @@
|
|
| 4768 |
"attributes": {}
|
| 4769 |
}
|
| 4770 |
},
|
| 4771 |
-
"total_flos":
|
| 4772 |
"train_batch_size": 1,
|
| 4773 |
"trial_name": null,
|
| 4774 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.7508354994430003,
|
| 5 |
"eval_steps": 337,
|
| 6 |
+
"global_step": 1011,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 4749 |
"eval_samples_per_second": 5.949,
|
| 4750 |
"eval_steps_per_second": 2.975,
|
| 4751 |
"step": 674
|
| 4752 |
+
},
|
| 4753 |
+
{
|
| 4754 |
+
"epoch": 0.5012996658002228,
|
| 4755 |
+
"grad_norm": 0.0007262330618686974,
|
| 4756 |
+
"learning_rate": 5.2786234659122726e-05,
|
| 4757 |
+
"loss": 0.0001,
|
| 4758 |
+
"step": 675
|
| 4759 |
+
},
|
| 4760 |
+
{
|
| 4761 |
+
"epoch": 0.5020423319717787,
|
| 4762 |
+
"grad_norm": 0.007732240483164787,
|
| 4763 |
+
"learning_rate": 5.266521150051492e-05,
|
| 4764 |
+
"loss": 0.0003,
|
| 4765 |
+
"step": 676
|
| 4766 |
+
},
|
| 4767 |
+
{
|
| 4768 |
+
"epoch": 0.5027849981433345,
|
| 4769 |
+
"grad_norm": 0.0026611273642629385,
|
| 4770 |
+
"learning_rate": 5.254417268084514e-05,
|
| 4771 |
+
"loss": 0.0001,
|
| 4772 |
+
"step": 677
|
| 4773 |
+
},
|
| 4774 |
+
{
|
| 4775 |
+
"epoch": 0.5035276643148905,
|
| 4776 |
+
"grad_norm": 0.00148585916031152,
|
| 4777 |
+
"learning_rate": 5.242311891135016e-05,
|
| 4778 |
+
"loss": 0.0002,
|
| 4779 |
+
"step": 678
|
| 4780 |
+
},
|
| 4781 |
+
{
|
| 4782 |
+
"epoch": 0.5042703304864463,
|
| 4783 |
+
"grad_norm": 0.0006069787195883691,
|
| 4784 |
+
"learning_rate": 5.2302050903354606e-05,
|
| 4785 |
+
"loss": 0.0001,
|
| 4786 |
+
"step": 679
|
| 4787 |
+
},
|
| 4788 |
+
{
|
| 4789 |
+
"epoch": 0.5050129966580023,
|
| 4790 |
+
"grad_norm": 0.002375812502577901,
|
| 4791 |
+
"learning_rate": 5.218096936826681e-05,
|
| 4792 |
+
"loss": 0.0002,
|
| 4793 |
+
"step": 680
|
| 4794 |
+
},
|
| 4795 |
+
{
|
| 4796 |
+
"epoch": 0.5057556628295581,
|
| 4797 |
+
"grad_norm": 0.003138911910355091,
|
| 4798 |
+
"learning_rate": 5.205987501757452e-05,
|
| 4799 |
+
"loss": 0.0002,
|
| 4800 |
+
"step": 681
|
| 4801 |
+
},
|
| 4802 |
+
{
|
| 4803 |
+
"epoch": 0.506498329001114,
|
| 4804 |
+
"grad_norm": 0.001015645801089704,
|
| 4805 |
+
"learning_rate": 5.193876856284085e-05,
|
| 4806 |
+
"loss": 0.0001,
|
| 4807 |
+
"step": 682
|
| 4808 |
+
},
|
| 4809 |
+
{
|
| 4810 |
+
"epoch": 0.5072409951726699,
|
| 4811 |
+
"grad_norm": 0.005205164197832346,
|
| 4812 |
+
"learning_rate": 5.1817650715699994e-05,
|
| 4813 |
+
"loss": 0.0002,
|
| 4814 |
+
"step": 683
|
| 4815 |
+
},
|
| 4816 |
+
{
|
| 4817 |
+
"epoch": 0.5079836613442258,
|
| 4818 |
+
"grad_norm": 0.0015316219069063663,
|
| 4819 |
+
"learning_rate": 5.1696522187853126e-05,
|
| 4820 |
+
"loss": 0.0002,
|
| 4821 |
+
"step": 684
|
| 4822 |
+
},
|
| 4823 |
+
{
|
| 4824 |
+
"epoch": 0.5087263275157816,
|
| 4825 |
+
"grad_norm": 0.0026633795350790024,
|
| 4826 |
+
"learning_rate": 5.157538369106414e-05,
|
| 4827 |
+
"loss": 0.0002,
|
| 4828 |
+
"step": 685
|
| 4829 |
+
},
|
| 4830 |
+
{
|
| 4831 |
+
"epoch": 0.5094689936873376,
|
| 4832 |
+
"grad_norm": 0.0016374588012695312,
|
| 4833 |
+
"learning_rate": 5.145423593715557e-05,
|
| 4834 |
+
"loss": 0.0001,
|
| 4835 |
+
"step": 686
|
| 4836 |
+
},
|
| 4837 |
+
{
|
| 4838 |
+
"epoch": 0.5102116598588934,
|
| 4839 |
+
"grad_norm": 0.002312192926183343,
|
| 4840 |
+
"learning_rate": 5.133307963800429e-05,
|
| 4841 |
+
"loss": 0.0002,
|
| 4842 |
+
"step": 687
|
| 4843 |
+
},
|
| 4844 |
+
{
|
| 4845 |
+
"epoch": 0.5109543260304493,
|
| 4846 |
+
"grad_norm": 0.004305702168494463,
|
| 4847 |
+
"learning_rate": 5.121191550553741e-05,
|
| 4848 |
+
"loss": 0.0002,
|
| 4849 |
+
"step": 688
|
| 4850 |
+
},
|
| 4851 |
+
{
|
| 4852 |
+
"epoch": 0.5116969922020052,
|
| 4853 |
+
"grad_norm": 0.005227786023169756,
|
| 4854 |
+
"learning_rate": 5.1090744251728064e-05,
|
| 4855 |
+
"loss": 0.0002,
|
| 4856 |
+
"step": 689
|
| 4857 |
+
},
|
| 4858 |
+
{
|
| 4859 |
+
"epoch": 0.5124396583735611,
|
| 4860 |
+
"grad_norm": 0.0004952704184688628,
|
| 4861 |
+
"learning_rate": 5.096956658859122e-05,
|
| 4862 |
+
"loss": 0.0001,
|
| 4863 |
+
"step": 690
|
| 4864 |
+
},
|
| 4865 |
+
{
|
| 4866 |
+
"epoch": 0.5131823245451169,
|
| 4867 |
+
"grad_norm": 0.0006610776763409376,
|
| 4868 |
+
"learning_rate": 5.084838322817954e-05,
|
| 4869 |
+
"loss": 0.0001,
|
| 4870 |
+
"step": 691
|
| 4871 |
+
},
|
| 4872 |
+
{
|
| 4873 |
+
"epoch": 0.5139249907166729,
|
| 4874 |
+
"grad_norm": 0.002141596982255578,
|
| 4875 |
+
"learning_rate": 5.072719488257914e-05,
|
| 4876 |
+
"loss": 0.0002,
|
| 4877 |
+
"step": 692
|
| 4878 |
+
},
|
| 4879 |
+
{
|
| 4880 |
+
"epoch": 0.5146676568882287,
|
| 4881 |
+
"grad_norm": 0.0008098665275610983,
|
| 4882 |
+
"learning_rate": 5.060600226390543e-05,
|
| 4883 |
+
"loss": 0.0001,
|
| 4884 |
+
"step": 693
|
| 4885 |
+
},
|
| 4886 |
+
{
|
| 4887 |
+
"epoch": 0.5154103230597846,
|
| 4888 |
+
"grad_norm": 0.0018381529953330755,
|
| 4889 |
+
"learning_rate": 5.048480608429893e-05,
|
| 4890 |
+
"loss": 0.0001,
|
| 4891 |
+
"step": 694
|
| 4892 |
+
},
|
| 4893 |
+
{
|
| 4894 |
+
"epoch": 0.5161529892313406,
|
| 4895 |
+
"grad_norm": 0.0011757278116419911,
|
| 4896 |
+
"learning_rate": 5.0363607055921094e-05,
|
| 4897 |
+
"loss": 0.0002,
|
| 4898 |
+
"step": 695
|
| 4899 |
+
},
|
| 4900 |
+
{
|
| 4901 |
+
"epoch": 0.5168956554028964,
|
| 4902 |
+
"grad_norm": 0.00043161376379430294,
|
| 4903 |
+
"learning_rate": 5.0242405890950093e-05,
|
| 4904 |
+
"loss": 0.0001,
|
| 4905 |
+
"step": 696
|
| 4906 |
+
},
|
| 4907 |
+
{
|
| 4908 |
+
"epoch": 0.5176383215744523,
|
| 4909 |
+
"grad_norm": 0.0008273344719782472,
|
| 4910 |
+
"learning_rate": 5.01212033015767e-05,
|
| 4911 |
+
"loss": 0.0001,
|
| 4912 |
+
"step": 697
|
| 4913 |
+
},
|
| 4914 |
+
{
|
| 4915 |
+
"epoch": 0.5183809877460082,
|
| 4916 |
+
"grad_norm": 0.0035381957422941923,
|
| 4917 |
+
"learning_rate": 5e-05,
|
| 4918 |
+
"loss": 0.0002,
|
| 4919 |
+
"step": 698
|
| 4920 |
+
},
|
| 4921 |
+
{
|
| 4922 |
+
"epoch": 0.5191236539175641,
|
| 4923 |
+
"grad_norm": 0.0015232819132506847,
|
| 4924 |
+
"learning_rate": 4.9878796698423324e-05,
|
| 4925 |
+
"loss": 0.0001,
|
| 4926 |
+
"step": 699
|
| 4927 |
+
},
|
| 4928 |
+
{
|
| 4929 |
+
"epoch": 0.5198663200891199,
|
| 4930 |
+
"grad_norm": 0.0007589098531752825,
|
| 4931 |
+
"learning_rate": 4.975759410904992e-05,
|
| 4932 |
+
"loss": 0.0001,
|
| 4933 |
+
"step": 700
|
| 4934 |
+
},
|
| 4935 |
+
{
|
| 4936 |
+
"epoch": 0.5206089862606759,
|
| 4937 |
+
"grad_norm": 0.0017526590963825583,
|
| 4938 |
+
"learning_rate": 4.963639294407893e-05,
|
| 4939 |
+
"loss": 0.0001,
|
| 4940 |
+
"step": 701
|
| 4941 |
+
},
|
| 4942 |
+
{
|
| 4943 |
+
"epoch": 0.5213516524322317,
|
| 4944 |
+
"grad_norm": 0.0024000522680580616,
|
| 4945 |
+
"learning_rate": 4.951519391570108e-05,
|
| 4946 |
+
"loss": 0.0001,
|
| 4947 |
+
"step": 702
|
| 4948 |
+
},
|
| 4949 |
+
{
|
| 4950 |
+
"epoch": 0.5220943186037876,
|
| 4951 |
+
"grad_norm": 0.027467429637908936,
|
| 4952 |
+
"learning_rate": 4.9393997736094596e-05,
|
| 4953 |
+
"loss": 0.0002,
|
| 4954 |
+
"step": 703
|
| 4955 |
+
},
|
| 4956 |
+
{
|
| 4957 |
+
"epoch": 0.5228369847753435,
|
| 4958 |
+
"grad_norm": 0.000491057347971946,
|
| 4959 |
+
"learning_rate": 4.9272805117420865e-05,
|
| 4960 |
+
"loss": 0.0001,
|
| 4961 |
+
"step": 704
|
| 4962 |
+
},
|
| 4963 |
+
{
|
| 4964 |
+
"epoch": 0.5235796509468994,
|
| 4965 |
+
"grad_norm": 0.0012586608063429594,
|
| 4966 |
+
"learning_rate": 4.915161677182048e-05,
|
| 4967 |
+
"loss": 0.0001,
|
| 4968 |
+
"step": 705
|
| 4969 |
+
},
|
| 4970 |
+
{
|
| 4971 |
+
"epoch": 0.5243223171184552,
|
| 4972 |
+
"grad_norm": 0.0025400435552001,
|
| 4973 |
+
"learning_rate": 4.903043341140879e-05,
|
| 4974 |
+
"loss": 0.0001,
|
| 4975 |
+
"step": 706
|
| 4976 |
+
},
|
| 4977 |
+
{
|
| 4978 |
+
"epoch": 0.5250649832900112,
|
| 4979 |
+
"grad_norm": 0.0027627914678305387,
|
| 4980 |
+
"learning_rate": 4.890925574827195e-05,
|
| 4981 |
+
"loss": 0.0002,
|
| 4982 |
+
"step": 707
|
| 4983 |
+
},
|
| 4984 |
+
{
|
| 4985 |
+
"epoch": 0.525807649461567,
|
| 4986 |
+
"grad_norm": 0.0010417604353278875,
|
| 4987 |
+
"learning_rate": 4.878808449446259e-05,
|
| 4988 |
+
"loss": 0.0001,
|
| 4989 |
+
"step": 708
|
| 4990 |
+
},
|
| 4991 |
+
{
|
| 4992 |
+
"epoch": 0.5265503156331229,
|
| 4993 |
+
"grad_norm": 0.0006638254853896797,
|
| 4994 |
+
"learning_rate": 4.866692036199572e-05,
|
| 4995 |
+
"loss": 0.0001,
|
| 4996 |
+
"step": 709
|
| 4997 |
+
},
|
| 4998 |
+
{
|
| 4999 |
+
"epoch": 0.5272929818046788,
|
| 5000 |
+
"grad_norm": 0.0013286214089021087,
|
| 5001 |
+
"learning_rate": 4.854576406284443e-05,
|
| 5002 |
+
"loss": 0.0001,
|
| 5003 |
+
"step": 710
|
| 5004 |
+
},
|
| 5005 |
+
{
|
| 5006 |
+
"epoch": 0.5280356479762347,
|
| 5007 |
+
"grad_norm": 0.004056660924106836,
|
| 5008 |
+
"learning_rate": 4.8424616308935875e-05,
|
| 5009 |
+
"loss": 0.0001,
|
| 5010 |
+
"step": 711
|
| 5011 |
+
},
|
| 5012 |
+
{
|
| 5013 |
+
"epoch": 0.5287783141477905,
|
| 5014 |
+
"grad_norm": 0.0007322711171582341,
|
| 5015 |
+
"learning_rate": 4.830347781214689e-05,
|
| 5016 |
+
"loss": 0.0001,
|
| 5017 |
+
"step": 712
|
| 5018 |
+
},
|
| 5019 |
+
{
|
| 5020 |
+
"epoch": 0.5295209803193465,
|
| 5021 |
+
"grad_norm": 0.004066810943186283,
|
| 5022 |
+
"learning_rate": 4.8182349284300024e-05,
|
| 5023 |
+
"loss": 0.0002,
|
| 5024 |
+
"step": 713
|
| 5025 |
+
},
|
| 5026 |
+
{
|
| 5027 |
+
"epoch": 0.5302636464909023,
|
| 5028 |
+
"grad_norm": 0.0007455990416929126,
|
| 5029 |
+
"learning_rate": 4.806123143715916e-05,
|
| 5030 |
+
"loss": 0.0001,
|
| 5031 |
+
"step": 714
|
| 5032 |
+
},
|
| 5033 |
+
{
|
| 5034 |
+
"epoch": 0.5310063126624582,
|
| 5035 |
+
"grad_norm": 0.0005849769804626703,
|
| 5036 |
+
"learning_rate": 4.7940124982425497e-05,
|
| 5037 |
+
"loss": 0.0001,
|
| 5038 |
+
"step": 715
|
| 5039 |
+
},
|
| 5040 |
+
{
|
| 5041 |
+
"epoch": 0.5317489788340141,
|
| 5042 |
+
"grad_norm": 0.0005887853913009167,
|
| 5043 |
+
"learning_rate": 4.781903063173321e-05,
|
| 5044 |
+
"loss": 0.0001,
|
| 5045 |
+
"step": 716
|
| 5046 |
+
},
|
| 5047 |
+
{
|
| 5048 |
+
"epoch": 0.53249164500557,
|
| 5049 |
+
"grad_norm": 0.0007402475457638502,
|
| 5050 |
+
"learning_rate": 4.76979490966454e-05,
|
| 5051 |
+
"loss": 0.0001,
|
| 5052 |
+
"step": 717
|
| 5053 |
+
},
|
| 5054 |
+
{
|
| 5055 |
+
"epoch": 0.5332343111771258,
|
| 5056 |
+
"grad_norm": 0.0029941131360828876,
|
| 5057 |
+
"learning_rate": 4.757688108864986e-05,
|
| 5058 |
+
"loss": 0.0002,
|
| 5059 |
+
"step": 718
|
| 5060 |
+
},
|
| 5061 |
+
{
|
| 5062 |
+
"epoch": 0.5339769773486818,
|
| 5063 |
+
"grad_norm": 0.0025099741760641336,
|
| 5064 |
+
"learning_rate": 4.7455827319154873e-05,
|
| 5065 |
+
"loss": 0.0002,
|
| 5066 |
+
"step": 719
|
| 5067 |
+
},
|
| 5068 |
+
{
|
| 5069 |
+
"epoch": 0.5347196435202377,
|
| 5070 |
+
"grad_norm": 0.0005227324436418712,
|
| 5071 |
+
"learning_rate": 4.73347884994851e-05,
|
| 5072 |
+
"loss": 0.0001,
|
| 5073 |
+
"step": 720
|
| 5074 |
+
},
|
| 5075 |
+
{
|
| 5076 |
+
"epoch": 0.5354623096917935,
|
| 5077 |
+
"grad_norm": 0.0016509697306901217,
|
| 5078 |
+
"learning_rate": 4.7213765340877286e-05,
|
| 5079 |
+
"loss": 0.0001,
|
| 5080 |
+
"step": 721
|
| 5081 |
+
},
|
| 5082 |
+
{
|
| 5083 |
+
"epoch": 0.5362049758633495,
|
| 5084 |
+
"grad_norm": 0.0011602011509239674,
|
| 5085 |
+
"learning_rate": 4.709275855447621e-05,
|
| 5086 |
+
"loss": 0.0001,
|
| 5087 |
+
"step": 722
|
| 5088 |
+
},
|
| 5089 |
+
{
|
| 5090 |
+
"epoch": 0.5369476420349053,
|
| 5091 |
+
"grad_norm": 0.0008615905535407364,
|
| 5092 |
+
"learning_rate": 4.697176885133042e-05,
|
| 5093 |
+
"loss": 0.0001,
|
| 5094 |
+
"step": 723
|
| 5095 |
+
},
|
| 5096 |
+
{
|
| 5097 |
+
"epoch": 0.5376903082064612,
|
| 5098 |
+
"grad_norm": 0.005168153438717127,
|
| 5099 |
+
"learning_rate": 4.6850796942388106e-05,
|
| 5100 |
+
"loss": 0.0002,
|
| 5101 |
+
"step": 724
|
| 5102 |
+
},
|
| 5103 |
+
{
|
| 5104 |
+
"epoch": 0.5384329743780171,
|
| 5105 |
+
"grad_norm": 0.0014478538651019335,
|
| 5106 |
+
"learning_rate": 4.6729843538492847e-05,
|
| 5107 |
+
"loss": 0.0001,
|
| 5108 |
+
"step": 725
|
| 5109 |
+
},
|
| 5110 |
+
{
|
| 5111 |
+
"epoch": 0.539175640549573,
|
| 5112 |
+
"grad_norm": 0.8397618532180786,
|
| 5113 |
+
"learning_rate": 4.660890935037954e-05,
|
| 5114 |
+
"loss": 0.013,
|
| 5115 |
+
"step": 726
|
| 5116 |
+
},
|
| 5117 |
+
{
|
| 5118 |
+
"epoch": 0.5399183067211288,
|
| 5119 |
+
"grad_norm": 0.0026152590289711952,
|
| 5120 |
+
"learning_rate": 4.648799508867012e-05,
|
| 5121 |
+
"loss": 0.0002,
|
| 5122 |
+
"step": 727
|
| 5123 |
+
},
|
| 5124 |
+
{
|
| 5125 |
+
"epoch": 0.5406609728926848,
|
| 5126 |
+
"grad_norm": 0.007058785296976566,
|
| 5127 |
+
"learning_rate": 4.636710146386947e-05,
|
| 5128 |
+
"loss": 0.0002,
|
| 5129 |
+
"step": 728
|
| 5130 |
+
},
|
| 5131 |
+
{
|
| 5132 |
+
"epoch": 0.5414036390642406,
|
| 5133 |
+
"grad_norm": 0.00043141836067661643,
|
| 5134 |
+
"learning_rate": 4.6246229186361184e-05,
|
| 5135 |
+
"loss": 0.0001,
|
| 5136 |
+
"step": 729
|
| 5137 |
+
},
|
| 5138 |
+
{
|
| 5139 |
+
"epoch": 0.5421463052357965,
|
| 5140 |
+
"grad_norm": 0.0015156837180256844,
|
| 5141 |
+
"learning_rate": 4.612537896640346e-05,
|
| 5142 |
+
"loss": 0.0002,
|
| 5143 |
+
"step": 730
|
| 5144 |
+
},
|
| 5145 |
+
{
|
| 5146 |
+
"epoch": 0.5428889714073524,
|
| 5147 |
+
"grad_norm": 0.0005109444609843194,
|
| 5148 |
+
"learning_rate": 4.600455151412482e-05,
|
| 5149 |
+
"loss": 0.0001,
|
| 5150 |
+
"step": 731
|
| 5151 |
+
},
|
| 5152 |
+
{
|
| 5153 |
+
"epoch": 0.5436316375789083,
|
| 5154 |
+
"grad_norm": 0.0052424585446715355,
|
| 5155 |
+
"learning_rate": 4.588374753952001e-05,
|
| 5156 |
+
"loss": 0.0003,
|
| 5157 |
+
"step": 732
|
| 5158 |
+
},
|
| 5159 |
+
{
|
| 5160 |
+
"epoch": 0.5443743037504641,
|
| 5161 |
+
"grad_norm": 0.0008050160249695182,
|
| 5162 |
+
"learning_rate": 4.576296775244588e-05,
|
| 5163 |
+
"loss": 0.0001,
|
| 5164 |
+
"step": 733
|
| 5165 |
+
},
|
| 5166 |
+
{
|
| 5167 |
+
"epoch": 0.5451169699220201,
|
| 5168 |
+
"grad_norm": 0.0020469501614570618,
|
| 5169 |
+
"learning_rate": 4.564221286261709e-05,
|
| 5170 |
+
"loss": 0.0002,
|
| 5171 |
+
"step": 734
|
| 5172 |
+
},
|
| 5173 |
+
{
|
| 5174 |
+
"epoch": 0.5458596360935759,
|
| 5175 |
+
"grad_norm": 0.000345387845300138,
|
| 5176 |
+
"learning_rate": 4.552148357960205e-05,
|
| 5177 |
+
"loss": 0.0001,
|
| 5178 |
+
"step": 735
|
| 5179 |
+
},
|
| 5180 |
+
{
|
| 5181 |
+
"epoch": 0.5466023022651318,
|
| 5182 |
+
"grad_norm": 0.002280315151438117,
|
| 5183 |
+
"learning_rate": 4.5400780612818626e-05,
|
| 5184 |
+
"loss": 0.0002,
|
| 5185 |
+
"step": 736
|
| 5186 |
+
},
|
| 5187 |
+
{
|
| 5188 |
+
"epoch": 0.5473449684366877,
|
| 5189 |
+
"grad_norm": 0.0018161894986405969,
|
| 5190 |
+
"learning_rate": 4.528010467153016e-05,
|
| 5191 |
+
"loss": 0.0002,
|
| 5192 |
+
"step": 737
|
| 5193 |
+
},
|
| 5194 |
+
{
|
| 5195 |
+
"epoch": 0.5480876346082436,
|
| 5196 |
+
"grad_norm": 0.0008220197050832212,
|
| 5197 |
+
"learning_rate": 4.515945646484105e-05,
|
| 5198 |
+
"loss": 0.0001,
|
| 5199 |
+
"step": 738
|
| 5200 |
+
},
|
| 5201 |
+
{
|
| 5202 |
+
"epoch": 0.5488303007797994,
|
| 5203 |
+
"grad_norm": 0.000890806782990694,
|
| 5204 |
+
"learning_rate": 4.503883670169285e-05,
|
| 5205 |
+
"loss": 0.0001,
|
| 5206 |
+
"step": 739
|
| 5207 |
+
},
|
| 5208 |
+
{
|
| 5209 |
+
"epoch": 0.5495729669513554,
|
| 5210 |
+
"grad_norm": 0.0015704066026955843,
|
| 5211 |
+
"learning_rate": 4.491824609085991e-05,
|
| 5212 |
+
"loss": 0.0002,
|
| 5213 |
+
"step": 740
|
| 5214 |
+
},
|
| 5215 |
+
{
|
| 5216 |
+
"epoch": 0.5503156331229112,
|
| 5217 |
+
"grad_norm": 0.003599775955080986,
|
| 5218 |
+
"learning_rate": 4.4797685340945295e-05,
|
| 5219 |
+
"loss": 0.0002,
|
| 5220 |
+
"step": 741
|
| 5221 |
+
},
|
| 5222 |
+
{
|
| 5223 |
+
"epoch": 0.5510582992944671,
|
| 5224 |
+
"grad_norm": 0.0044060661457479,
|
| 5225 |
+
"learning_rate": 4.467715516037659e-05,
|
| 5226 |
+
"loss": 0.0002,
|
| 5227 |
+
"step": 742
|
| 5228 |
+
},
|
| 5229 |
+
{
|
| 5230 |
+
"epoch": 0.551800965466023,
|
| 5231 |
+
"grad_norm": 0.0007429459365084767,
|
| 5232 |
+
"learning_rate": 4.4556656257401786e-05,
|
| 5233 |
+
"loss": 0.0001,
|
| 5234 |
+
"step": 743
|
| 5235 |
+
},
|
| 5236 |
+
{
|
| 5237 |
+
"epoch": 0.5525436316375789,
|
| 5238 |
+
"grad_norm": 0.0007204132853075862,
|
| 5239 |
+
"learning_rate": 4.443618934008502e-05,
|
| 5240 |
+
"loss": 0.0001,
|
| 5241 |
+
"step": 744
|
| 5242 |
+
},
|
| 5243 |
+
{
|
| 5244 |
+
"epoch": 0.5532862978091347,
|
| 5245 |
+
"grad_norm": 0.05012976750731468,
|
| 5246 |
+
"learning_rate": 4.4315755116302554e-05,
|
| 5247 |
+
"loss": 0.0007,
|
| 5248 |
+
"step": 745
|
| 5249 |
+
},
|
| 5250 |
+
{
|
| 5251 |
+
"epoch": 0.5540289639806907,
|
| 5252 |
+
"grad_norm": 0.0010998403886333108,
|
| 5253 |
+
"learning_rate": 4.4195354293738484e-05,
|
| 5254 |
+
"loss": 0.0001,
|
| 5255 |
+
"step": 746
|
| 5256 |
+
},
|
| 5257 |
+
{
|
| 5258 |
+
"epoch": 0.5547716301522466,
|
| 5259 |
+
"grad_norm": 0.0022817838471382856,
|
| 5260 |
+
"learning_rate": 4.4074987579880685e-05,
|
| 5261 |
+
"loss": 0.0002,
|
| 5262 |
+
"step": 747
|
| 5263 |
+
},
|
| 5264 |
+
{
|
| 5265 |
+
"epoch": 0.5555142963238024,
|
| 5266 |
+
"grad_norm": 0.009643745608627796,
|
| 5267 |
+
"learning_rate": 4.395465568201655e-05,
|
| 5268 |
+
"loss": 0.0005,
|
| 5269 |
+
"step": 748
|
| 5270 |
+
},
|
| 5271 |
+
{
|
| 5272 |
+
"epoch": 0.5562569624953584,
|
| 5273 |
+
"grad_norm": 0.0010438463650643826,
|
| 5274 |
+
"learning_rate": 4.38343593072289e-05,
|
| 5275 |
+
"loss": 0.0001,
|
| 5276 |
+
"step": 749
|
| 5277 |
+
},
|
| 5278 |
+
{
|
| 5279 |
+
"epoch": 0.5569996286669142,
|
| 5280 |
+
"grad_norm": 0.0021892802324146032,
|
| 5281 |
+
"learning_rate": 4.3714099162391875e-05,
|
| 5282 |
+
"loss": 0.0001,
|
| 5283 |
+
"step": 750
|
| 5284 |
+
},
|
| 5285 |
+
{
|
| 5286 |
+
"epoch": 0.5577422948384702,
|
| 5287 |
+
"grad_norm": 0.04245134815573692,
|
| 5288 |
+
"learning_rate": 4.359387595416665e-05,
|
| 5289 |
+
"loss": 0.001,
|
| 5290 |
+
"step": 751
|
| 5291 |
+
},
|
| 5292 |
+
{
|
| 5293 |
+
"epoch": 0.558484961010026,
|
| 5294 |
+
"grad_norm": 0.001306456862948835,
|
| 5295 |
+
"learning_rate": 4.347369038899744e-05,
|
| 5296 |
+
"loss": 0.0001,
|
| 5297 |
+
"step": 752
|
| 5298 |
+
},
|
| 5299 |
+
{
|
| 5300 |
+
"epoch": 0.5592276271815819,
|
| 5301 |
+
"grad_norm": 0.0017448252765461802,
|
| 5302 |
+
"learning_rate": 4.335354317310715e-05,
|
| 5303 |
+
"loss": 0.0001,
|
| 5304 |
+
"step": 753
|
| 5305 |
+
},
|
| 5306 |
+
{
|
| 5307 |
+
"epoch": 0.5599702933531377,
|
| 5308 |
+
"grad_norm": 0.014149321243166924,
|
| 5309 |
+
"learning_rate": 4.323343501249346e-05,
|
| 5310 |
+
"loss": 0.0002,
|
| 5311 |
+
"step": 754
|
| 5312 |
+
},
|
| 5313 |
+
{
|
| 5314 |
+
"epoch": 0.5607129595246937,
|
| 5315 |
+
"grad_norm": 0.0010777488350868225,
|
| 5316 |
+
"learning_rate": 4.3113366612924465e-05,
|
| 5317 |
+
"loss": 0.0001,
|
| 5318 |
+
"step": 755
|
| 5319 |
+
},
|
| 5320 |
+
{
|
| 5321 |
+
"epoch": 0.5614556256962495,
|
| 5322 |
+
"grad_norm": 0.0005878026131540537,
|
| 5323 |
+
"learning_rate": 4.299333867993469e-05,
|
| 5324 |
+
"loss": 0.0001,
|
| 5325 |
+
"step": 756
|
| 5326 |
+
},
|
| 5327 |
+
{
|
| 5328 |
+
"epoch": 0.5621982918678055,
|
| 5329 |
+
"grad_norm": 0.004690335597842932,
|
| 5330 |
+
"learning_rate": 4.287335191882083e-05,
|
| 5331 |
+
"loss": 0.0003,
|
| 5332 |
+
"step": 757
|
| 5333 |
+
},
|
| 5334 |
+
{
|
| 5335 |
+
"epoch": 0.5629409580393613,
|
| 5336 |
+
"grad_norm": 0.002779589965939522,
|
| 5337 |
+
"learning_rate": 4.275340703463767e-05,
|
| 5338 |
+
"loss": 0.0002,
|
| 5339 |
+
"step": 758
|
| 5340 |
+
},
|
| 5341 |
+
{
|
| 5342 |
+
"epoch": 0.5636836242109172,
|
| 5343 |
+
"grad_norm": 0.01008251216262579,
|
| 5344 |
+
"learning_rate": 4.2633504732193896e-05,
|
| 5345 |
+
"loss": 0.0003,
|
| 5346 |
+
"step": 759
|
| 5347 |
+
},
|
| 5348 |
+
{
|
| 5349 |
+
"epoch": 0.564426290382473,
|
| 5350 |
+
"grad_norm": 0.0012684455141425133,
|
| 5351 |
+
"learning_rate": 4.2513645716048025e-05,
|
| 5352 |
+
"loss": 0.0002,
|
| 5353 |
+
"step": 760
|
| 5354 |
+
},
|
| 5355 |
+
{
|
| 5356 |
+
"epoch": 0.565168956554029,
|
| 5357 |
+
"grad_norm": 0.0006100367172621191,
|
| 5358 |
+
"learning_rate": 4.239383069050417e-05,
|
| 5359 |
+
"loss": 0.0001,
|
| 5360 |
+
"step": 761
|
| 5361 |
+
},
|
| 5362 |
+
{
|
| 5363 |
+
"epoch": 0.5659116227255848,
|
| 5364 |
+
"grad_norm": 0.0015486053889617324,
|
| 5365 |
+
"learning_rate": 4.227406035960798e-05,
|
| 5366 |
+
"loss": 0.0001,
|
| 5367 |
+
"step": 762
|
| 5368 |
+
},
|
| 5369 |
+
{
|
| 5370 |
+
"epoch": 0.5666542888971408,
|
| 5371 |
+
"grad_norm": 0.0052903578616678715,
|
| 5372 |
+
"learning_rate": 4.215433542714248e-05,
|
| 5373 |
+
"loss": 0.0003,
|
| 5374 |
+
"step": 763
|
| 5375 |
+
},
|
| 5376 |
+
{
|
| 5377 |
+
"epoch": 0.5673969550686966,
|
| 5378 |
+
"grad_norm": 0.005356297362595797,
|
| 5379 |
+
"learning_rate": 4.203465659662388e-05,
|
| 5380 |
+
"loss": 0.0002,
|
| 5381 |
+
"step": 764
|
| 5382 |
+
},
|
| 5383 |
+
{
|
| 5384 |
+
"epoch": 0.5681396212402525,
|
| 5385 |
+
"grad_norm": 0.003509636502712965,
|
| 5386 |
+
"learning_rate": 4.191502457129756e-05,
|
| 5387 |
+
"loss": 0.0001,
|
| 5388 |
+
"step": 765
|
| 5389 |
+
},
|
| 5390 |
+
{
|
| 5391 |
+
"epoch": 0.5688822874118084,
|
| 5392 |
+
"grad_norm": 0.013686946593225002,
|
| 5393 |
+
"learning_rate": 4.17954400541338e-05,
|
| 5394 |
+
"loss": 0.0004,
|
| 5395 |
+
"step": 766
|
| 5396 |
+
},
|
| 5397 |
+
{
|
| 5398 |
+
"epoch": 0.5696249535833643,
|
| 5399 |
+
"grad_norm": 0.0035359973553568125,
|
| 5400 |
+
"learning_rate": 4.16759037478238e-05,
|
| 5401 |
+
"loss": 0.0002,
|
| 5402 |
+
"step": 767
|
| 5403 |
+
},
|
| 5404 |
+
{
|
| 5405 |
+
"epoch": 0.5703676197549201,
|
| 5406 |
+
"grad_norm": 0.0006897081038914621,
|
| 5407 |
+
"learning_rate": 4.1556416354775364e-05,
|
| 5408 |
+
"loss": 0.0001,
|
| 5409 |
+
"step": 768
|
| 5410 |
+
},
|
| 5411 |
+
{
|
| 5412 |
+
"epoch": 0.5711102859264761,
|
| 5413 |
+
"grad_norm": 0.000840733468066901,
|
| 5414 |
+
"learning_rate": 4.1436978577108975e-05,
|
| 5415 |
+
"loss": 0.0001,
|
| 5416 |
+
"step": 769
|
| 5417 |
+
},
|
| 5418 |
+
{
|
| 5419 |
+
"epoch": 0.5718529520980319,
|
| 5420 |
+
"grad_norm": 0.001890829997137189,
|
| 5421 |
+
"learning_rate": 4.131759111665349e-05,
|
| 5422 |
+
"loss": 0.0001,
|
| 5423 |
+
"step": 770
|
| 5424 |
+
},
|
| 5425 |
+
{
|
| 5426 |
+
"epoch": 0.5725956182695878,
|
| 5427 |
+
"grad_norm": 0.006861279718577862,
|
| 5428 |
+
"learning_rate": 4.1198254674942154e-05,
|
| 5429 |
+
"loss": 0.0003,
|
| 5430 |
+
"step": 771
|
| 5431 |
+
},
|
| 5432 |
+
{
|
| 5433 |
+
"epoch": 0.5733382844411437,
|
| 5434 |
+
"grad_norm": 0.020367255434393883,
|
| 5435 |
+
"learning_rate": 4.107896995320839e-05,
|
| 5436 |
+
"loss": 0.0001,
|
| 5437 |
+
"step": 772
|
| 5438 |
+
},
|
| 5439 |
+
{
|
| 5440 |
+
"epoch": 0.5740809506126996,
|
| 5441 |
+
"grad_norm": 0.0010303305462002754,
|
| 5442 |
+
"learning_rate": 4.095973765238175e-05,
|
| 5443 |
+
"loss": 0.0001,
|
| 5444 |
+
"step": 773
|
| 5445 |
+
},
|
| 5446 |
+
{
|
| 5447 |
+
"epoch": 0.5748236167842555,
|
| 5448 |
+
"grad_norm": 0.0006230950239114463,
|
| 5449 |
+
"learning_rate": 4.0840558473083664e-05,
|
| 5450 |
+
"loss": 0.0001,
|
| 5451 |
+
"step": 774
|
| 5452 |
+
},
|
| 5453 |
+
{
|
| 5454 |
+
"epoch": 0.5755662829558114,
|
| 5455 |
+
"grad_norm": 0.06197100505232811,
|
| 5456 |
+
"learning_rate": 4.0721433115623544e-05,
|
| 5457 |
+
"loss": 0.0016,
|
| 5458 |
+
"step": 775
|
| 5459 |
+
},
|
| 5460 |
+
{
|
| 5461 |
+
"epoch": 0.5763089491273673,
|
| 5462 |
+
"grad_norm": 0.015006426721811295,
|
| 5463 |
+
"learning_rate": 4.060236227999441e-05,
|
| 5464 |
+
"loss": 0.0003,
|
| 5465 |
+
"step": 776
|
| 5466 |
+
},
|
| 5467 |
+
{
|
| 5468 |
+
"epoch": 0.5770516152989231,
|
| 5469 |
+
"grad_norm": 0.0008427058928646147,
|
| 5470 |
+
"learning_rate": 4.0483346665869e-05,
|
| 5471 |
+
"loss": 0.0001,
|
| 5472 |
+
"step": 777
|
| 5473 |
+
},
|
| 5474 |
+
{
|
| 5475 |
+
"epoch": 0.5777942814704791,
|
| 5476 |
+
"grad_norm": 0.0018404822330921888,
|
| 5477 |
+
"learning_rate": 4.036438697259551e-05,
|
| 5478 |
+
"loss": 0.0002,
|
| 5479 |
+
"step": 778
|
| 5480 |
+
},
|
| 5481 |
+
{
|
| 5482 |
+
"epoch": 0.5785369476420349,
|
| 5483 |
+
"grad_norm": 0.0025395648553967476,
|
| 5484 |
+
"learning_rate": 4.0245483899193595e-05,
|
| 5485 |
+
"loss": 0.0001,
|
| 5486 |
+
"step": 779
|
| 5487 |
+
},
|
| 5488 |
+
{
|
| 5489 |
+
"epoch": 0.5792796138135908,
|
| 5490 |
+
"grad_norm": 0.0011295550502836704,
|
| 5491 |
+
"learning_rate": 4.012663814435014e-05,
|
| 5492 |
+
"loss": 0.0001,
|
| 5493 |
+
"step": 780
|
| 5494 |
+
},
|
| 5495 |
+
{
|
| 5496 |
+
"epoch": 0.5800222799851467,
|
| 5497 |
+
"grad_norm": 0.0014332979917526245,
|
| 5498 |
+
"learning_rate": 4.0007850406415234e-05,
|
| 5499 |
+
"loss": 0.0002,
|
| 5500 |
+
"step": 781
|
| 5501 |
+
},
|
| 5502 |
+
{
|
| 5503 |
+
"epoch": 0.5807649461567026,
|
| 5504 |
+
"grad_norm": 0.0006570751429535449,
|
| 5505 |
+
"learning_rate": 3.9889121383398115e-05,
|
| 5506 |
+
"loss": 0.0001,
|
| 5507 |
+
"step": 782
|
| 5508 |
+
},
|
| 5509 |
+
{
|
| 5510 |
+
"epoch": 0.5815076123282584,
|
| 5511 |
+
"grad_norm": 0.0014288886450231075,
|
| 5512 |
+
"learning_rate": 3.977045177296291e-05,
|
| 5513 |
+
"loss": 0.0002,
|
| 5514 |
+
"step": 783
|
| 5515 |
+
},
|
| 5516 |
+
{
|
| 5517 |
+
"epoch": 0.5822502784998144,
|
| 5518 |
+
"grad_norm": 0.000917198951356113,
|
| 5519 |
+
"learning_rate": 3.965184227242474e-05,
|
| 5520 |
+
"loss": 0.0001,
|
| 5521 |
+
"step": 784
|
| 5522 |
+
},
|
| 5523 |
+
{
|
| 5524 |
+
"epoch": 0.5829929446713702,
|
| 5525 |
+
"grad_norm": 0.002546856412664056,
|
| 5526 |
+
"learning_rate": 3.95332935787454e-05,
|
| 5527 |
+
"loss": 0.0001,
|
| 5528 |
+
"step": 785
|
| 5529 |
+
},
|
| 5530 |
+
{
|
| 5531 |
+
"epoch": 0.5837356108429261,
|
| 5532 |
+
"grad_norm": 0.0009835307719185948,
|
| 5533 |
+
"learning_rate": 3.941480638852948e-05,
|
| 5534 |
+
"loss": 0.0001,
|
| 5535 |
+
"step": 786
|
| 5536 |
+
},
|
| 5537 |
+
{
|
| 5538 |
+
"epoch": 0.584478277014482,
|
| 5539 |
+
"grad_norm": 0.10560861229896545,
|
| 5540 |
+
"learning_rate": 3.929638139802007e-05,
|
| 5541 |
+
"loss": 0.0006,
|
| 5542 |
+
"step": 787
|
| 5543 |
+
},
|
| 5544 |
+
{
|
| 5545 |
+
"epoch": 0.5852209431860379,
|
| 5546 |
+
"grad_norm": 0.0006187596009112895,
|
| 5547 |
+
"learning_rate": 3.917801930309486e-05,
|
| 5548 |
+
"loss": 0.0001,
|
| 5549 |
+
"step": 788
|
| 5550 |
+
},
|
| 5551 |
+
{
|
| 5552 |
+
"epoch": 0.5859636093575937,
|
| 5553 |
+
"grad_norm": 0.027732783928513527,
|
| 5554 |
+
"learning_rate": 3.90597207992619e-05,
|
| 5555 |
+
"loss": 0.0005,
|
| 5556 |
+
"step": 789
|
| 5557 |
+
},
|
| 5558 |
+
{
|
| 5559 |
+
"epoch": 0.5867062755291497,
|
| 5560 |
+
"grad_norm": 0.003759243292734027,
|
| 5561 |
+
"learning_rate": 3.8941486581655615e-05,
|
| 5562 |
+
"loss": 0.0002,
|
| 5563 |
+
"step": 790
|
| 5564 |
+
},
|
| 5565 |
+
{
|
| 5566 |
+
"epoch": 0.5874489417007055,
|
| 5567 |
+
"grad_norm": 0.002819777699187398,
|
| 5568 |
+
"learning_rate": 3.882331734503263e-05,
|
| 5569 |
+
"loss": 0.0001,
|
| 5570 |
+
"step": 791
|
| 5571 |
+
},
|
| 5572 |
+
{
|
| 5573 |
+
"epoch": 0.5881916078722614,
|
| 5574 |
+
"grad_norm": 0.0007966597331687808,
|
| 5575 |
+
"learning_rate": 3.870521378376777e-05,
|
| 5576 |
+
"loss": 0.0001,
|
| 5577 |
+
"step": 792
|
| 5578 |
+
},
|
| 5579 |
+
{
|
| 5580 |
+
"epoch": 0.5889342740438173,
|
| 5581 |
+
"grad_norm": 0.0005825799889862537,
|
| 5582 |
+
"learning_rate": 3.8587176591849916e-05,
|
| 5583 |
+
"loss": 0.0001,
|
| 5584 |
+
"step": 793
|
| 5585 |
+
},
|
| 5586 |
+
{
|
| 5587 |
+
"epoch": 0.5896769402153732,
|
| 5588 |
+
"grad_norm": 0.002244536532089114,
|
| 5589 |
+
"learning_rate": 3.846920646287799e-05,
|
| 5590 |
+
"loss": 0.0001,
|
| 5591 |
+
"step": 794
|
| 5592 |
+
},
|
| 5593 |
+
{
|
| 5594 |
+
"epoch": 0.590419606386929,
|
| 5595 |
+
"grad_norm": 0.002780564595013857,
|
| 5596 |
+
"learning_rate": 3.8351304090056825e-05,
|
| 5597 |
+
"loss": 0.0002,
|
| 5598 |
+
"step": 795
|
| 5599 |
+
},
|
| 5600 |
+
{
|
| 5601 |
+
"epoch": 0.591162272558485,
|
| 5602 |
+
"grad_norm": 0.0009447059710510075,
|
| 5603 |
+
"learning_rate": 3.8233470166193126e-05,
|
| 5604 |
+
"loss": 0.0001,
|
| 5605 |
+
"step": 796
|
| 5606 |
+
},
|
| 5607 |
+
{
|
| 5608 |
+
"epoch": 0.5919049387300408,
|
| 5609 |
+
"grad_norm": 0.001110229641199112,
|
| 5610 |
+
"learning_rate": 3.8115705383691355e-05,
|
| 5611 |
+
"loss": 0.0001,
|
| 5612 |
+
"step": 797
|
| 5613 |
+
},
|
| 5614 |
+
{
|
| 5615 |
+
"epoch": 0.5926476049015967,
|
| 5616 |
+
"grad_norm": 0.0013889704132452607,
|
| 5617 |
+
"learning_rate": 3.799801043454971e-05,
|
| 5618 |
+
"loss": 0.0001,
|
| 5619 |
+
"step": 798
|
| 5620 |
+
},
|
| 5621 |
+
{
|
| 5622 |
+
"epoch": 0.5933902710731526,
|
| 5623 |
+
"grad_norm": 0.0008447144064120948,
|
| 5624 |
+
"learning_rate": 3.7880386010356054e-05,
|
| 5625 |
+
"loss": 0.0001,
|
| 5626 |
+
"step": 799
|
| 5627 |
+
},
|
| 5628 |
+
{
|
| 5629 |
+
"epoch": 0.5941329372447085,
|
| 5630 |
+
"grad_norm": 0.003610602580010891,
|
| 5631 |
+
"learning_rate": 3.776283280228381e-05,
|
| 5632 |
+
"loss": 0.0001,
|
| 5633 |
+
"step": 800
|
| 5634 |
+
},
|
| 5635 |
+
{
|
| 5636 |
+
"epoch": 0.5948756034162644,
|
| 5637 |
+
"grad_norm": 0.005582943558692932,
|
| 5638 |
+
"learning_rate": 3.764535150108798e-05,
|
| 5639 |
+
"loss": 0.0003,
|
| 5640 |
+
"step": 801
|
| 5641 |
+
},
|
| 5642 |
+
{
|
| 5643 |
+
"epoch": 0.5956182695878203,
|
| 5644 |
+
"grad_norm": 0.003268187865614891,
|
| 5645 |
+
"learning_rate": 3.752794279710094e-05,
|
| 5646 |
+
"loss": 0.0002,
|
| 5647 |
+
"step": 802
|
| 5648 |
+
},
|
| 5649 |
+
{
|
| 5650 |
+
"epoch": 0.5963609357593762,
|
| 5651 |
+
"grad_norm": 0.0018831202760338783,
|
| 5652 |
+
"learning_rate": 3.741060738022858e-05,
|
| 5653 |
+
"loss": 0.0002,
|
| 5654 |
+
"step": 803
|
| 5655 |
+
},
|
| 5656 |
+
{
|
| 5657 |
+
"epoch": 0.597103601930932,
|
| 5658 |
+
"grad_norm": 0.0007239828119054437,
|
| 5659 |
+
"learning_rate": 3.7293345939946065e-05,
|
| 5660 |
+
"loss": 0.0001,
|
| 5661 |
+
"step": 804
|
| 5662 |
+
},
|
| 5663 |
+
{
|
| 5664 |
+
"epoch": 0.597846268102488,
|
| 5665 |
+
"grad_norm": 0.21732911467552185,
|
| 5666 |
+
"learning_rate": 3.717615916529395e-05,
|
| 5667 |
+
"loss": 0.0017,
|
| 5668 |
+
"step": 805
|
| 5669 |
+
},
|
| 5670 |
+
{
|
| 5671 |
+
"epoch": 0.5985889342740438,
|
| 5672 |
+
"grad_norm": 0.00127478688955307,
|
| 5673 |
+
"learning_rate": 3.705904774487396e-05,
|
| 5674 |
+
"loss": 0.0002,
|
| 5675 |
+
"step": 806
|
| 5676 |
+
},
|
| 5677 |
+
{
|
| 5678 |
+
"epoch": 0.5993316004455997,
|
| 5679 |
+
"grad_norm": 0.0045968894846737385,
|
| 5680 |
+
"learning_rate": 3.6942012366845096e-05,
|
| 5681 |
+
"loss": 0.0001,
|
| 5682 |
+
"step": 807
|
| 5683 |
+
},
|
| 5684 |
+
{
|
| 5685 |
+
"epoch": 0.6000742666171556,
|
| 5686 |
+
"grad_norm": 0.001871331944130361,
|
| 5687 |
+
"learning_rate": 3.6825053718919464e-05,
|
| 5688 |
+
"loss": 0.0002,
|
| 5689 |
+
"step": 808
|
| 5690 |
+
},
|
| 5691 |
+
{
|
| 5692 |
+
"epoch": 0.6008169327887115,
|
| 5693 |
+
"grad_norm": 0.0010791551321744919,
|
| 5694 |
+
"learning_rate": 3.670817248835836e-05,
|
| 5695 |
+
"loss": 0.0001,
|
| 5696 |
+
"step": 809
|
| 5697 |
+
},
|
| 5698 |
+
{
|
| 5699 |
+
"epoch": 0.6015595989602673,
|
| 5700 |
+
"grad_norm": 0.0006027501658536494,
|
| 5701 |
+
"learning_rate": 3.6591369361968124e-05,
|
| 5702 |
+
"loss": 0.0001,
|
| 5703 |
+
"step": 810
|
| 5704 |
+
},
|
| 5705 |
+
{
|
| 5706 |
+
"epoch": 0.6023022651318233,
|
| 5707 |
+
"grad_norm": 0.0032688211649656296,
|
| 5708 |
+
"learning_rate": 3.6474645026096185e-05,
|
| 5709 |
+
"loss": 0.0002,
|
| 5710 |
+
"step": 811
|
| 5711 |
+
},
|
| 5712 |
+
{
|
| 5713 |
+
"epoch": 0.6030449313033791,
|
| 5714 |
+
"grad_norm": 0.00113378232344985,
|
| 5715 |
+
"learning_rate": 3.6358000166626966e-05,
|
| 5716 |
+
"loss": 0.0001,
|
| 5717 |
+
"step": 812
|
| 5718 |
+
},
|
| 5719 |
+
{
|
| 5720 |
+
"epoch": 0.603787597474935,
|
| 5721 |
+
"grad_norm": 0.0009587003150954843,
|
| 5722 |
+
"learning_rate": 3.624143546897784e-05,
|
| 5723 |
+
"loss": 0.0001,
|
| 5724 |
+
"step": 813
|
| 5725 |
+
},
|
| 5726 |
+
{
|
| 5727 |
+
"epoch": 0.6045302636464909,
|
| 5728 |
+
"grad_norm": 0.0017409041756764054,
|
| 5729 |
+
"learning_rate": 3.612495161809522e-05,
|
| 5730 |
+
"loss": 0.0001,
|
| 5731 |
+
"step": 814
|
| 5732 |
+
},
|
| 5733 |
+
{
|
| 5734 |
+
"epoch": 0.6052729298180468,
|
| 5735 |
+
"grad_norm": 0.0004892799770459533,
|
| 5736 |
+
"learning_rate": 3.60085492984504e-05,
|
| 5737 |
+
"loss": 0.0001,
|
| 5738 |
+
"step": 815
|
| 5739 |
+
},
|
| 5740 |
+
{
|
| 5741 |
+
"epoch": 0.6060155959896026,
|
| 5742 |
+
"grad_norm": 0.0016567412531003356,
|
| 5743 |
+
"learning_rate": 3.589222919403562e-05,
|
| 5744 |
+
"loss": 0.0002,
|
| 5745 |
+
"step": 816
|
| 5746 |
+
},
|
| 5747 |
+
{
|
| 5748 |
+
"epoch": 0.6067582621611586,
|
| 5749 |
+
"grad_norm": 0.0016312827356159687,
|
| 5750 |
+
"learning_rate": 3.577599198835998e-05,
|
| 5751 |
+
"loss": 0.0001,
|
| 5752 |
+
"step": 817
|
| 5753 |
+
},
|
| 5754 |
+
{
|
| 5755 |
+
"epoch": 0.6075009283327144,
|
| 5756 |
+
"grad_norm": 0.0007810606039129198,
|
| 5757 |
+
"learning_rate": 3.5659838364445505e-05,
|
| 5758 |
+
"loss": 0.0001,
|
| 5759 |
+
"step": 818
|
| 5760 |
+
},
|
| 5761 |
+
{
|
| 5762 |
+
"epoch": 0.6082435945042703,
|
| 5763 |
+
"grad_norm": 0.0013547971611842513,
|
| 5764 |
+
"learning_rate": 3.554376900482301e-05,
|
| 5765 |
+
"loss": 0.0001,
|
| 5766 |
+
"step": 819
|
| 5767 |
+
},
|
| 5768 |
+
{
|
| 5769 |
+
"epoch": 0.6089862606758262,
|
| 5770 |
+
"grad_norm": 0.0006009297212585807,
|
| 5771 |
+
"learning_rate": 3.542778459152826e-05,
|
| 5772 |
+
"loss": 0.0001,
|
| 5773 |
+
"step": 820
|
| 5774 |
+
},
|
| 5775 |
+
{
|
| 5776 |
+
"epoch": 0.6097289268473821,
|
| 5777 |
+
"grad_norm": 0.0010081218788400292,
|
| 5778 |
+
"learning_rate": 3.531188580609778e-05,
|
| 5779 |
+
"loss": 0.0001,
|
| 5780 |
+
"step": 821
|
| 5781 |
+
},
|
| 5782 |
+
{
|
| 5783 |
+
"epoch": 0.6104715930189379,
|
| 5784 |
+
"grad_norm": 0.00042164497426711023,
|
| 5785 |
+
"learning_rate": 3.5196073329565015e-05,
|
| 5786 |
+
"loss": 0.0001,
|
| 5787 |
+
"step": 822
|
| 5788 |
+
},
|
| 5789 |
+
{
|
| 5790 |
+
"epoch": 0.6112142591904939,
|
| 5791 |
+
"grad_norm": 0.0007461085333488882,
|
| 5792 |
+
"learning_rate": 3.508034784245616e-05,
|
| 5793 |
+
"loss": 0.0001,
|
| 5794 |
+
"step": 823
|
| 5795 |
+
},
|
| 5796 |
+
{
|
| 5797 |
+
"epoch": 0.6119569253620497,
|
| 5798 |
+
"grad_norm": 1.3910738229751587,
|
| 5799 |
+
"learning_rate": 3.4964710024786354e-05,
|
| 5800 |
+
"loss": 0.0984,
|
| 5801 |
+
"step": 824
|
| 5802 |
+
},
|
| 5803 |
+
{
|
| 5804 |
+
"epoch": 0.6126995915336056,
|
| 5805 |
+
"grad_norm": 0.0007660177652724087,
|
| 5806 |
+
"learning_rate": 3.484916055605548e-05,
|
| 5807 |
+
"loss": 0.0001,
|
| 5808 |
+
"step": 825
|
| 5809 |
+
},
|
| 5810 |
+
{
|
| 5811 |
+
"epoch": 0.6134422577051615,
|
| 5812 |
+
"grad_norm": 0.07438488304615021,
|
| 5813 |
+
"learning_rate": 3.473370011524435e-05,
|
| 5814 |
+
"loss": 0.0012,
|
| 5815 |
+
"step": 826
|
| 5816 |
+
},
|
| 5817 |
+
{
|
| 5818 |
+
"epoch": 0.6141849238767174,
|
| 5819 |
+
"grad_norm": 0.0015772825572639704,
|
| 5820 |
+
"learning_rate": 3.461832938081059e-05,
|
| 5821 |
+
"loss": 0.0002,
|
| 5822 |
+
"step": 827
|
| 5823 |
+
},
|
| 5824 |
+
{
|
| 5825 |
+
"epoch": 0.6149275900482734,
|
| 5826 |
+
"grad_norm": 0.0016182976542040706,
|
| 5827 |
+
"learning_rate": 3.450304903068475e-05,
|
| 5828 |
+
"loss": 0.0001,
|
| 5829 |
+
"step": 828
|
| 5830 |
+
},
|
| 5831 |
+
{
|
| 5832 |
+
"epoch": 0.6156702562198292,
|
| 5833 |
+
"grad_norm": 0.0015479021240025759,
|
| 5834 |
+
"learning_rate": 3.4387859742266225e-05,
|
| 5835 |
+
"loss": 0.0001,
|
| 5836 |
+
"step": 829
|
| 5837 |
+
},
|
| 5838 |
+
{
|
| 5839 |
+
"epoch": 0.6164129223913851,
|
| 5840 |
+
"grad_norm": 0.00273375422693789,
|
| 5841 |
+
"learning_rate": 3.427276219241933e-05,
|
| 5842 |
+
"loss": 0.0002,
|
| 5843 |
+
"step": 830
|
| 5844 |
+
},
|
| 5845 |
+
{
|
| 5846 |
+
"epoch": 0.617155588562941,
|
| 5847 |
+
"grad_norm": 0.001004122314043343,
|
| 5848 |
+
"learning_rate": 3.415775705746934e-05,
|
| 5849 |
+
"loss": 0.0001,
|
| 5850 |
+
"step": 831
|
| 5851 |
+
},
|
| 5852 |
+
{
|
| 5853 |
+
"epoch": 0.6178982547344969,
|
| 5854 |
+
"grad_norm": 0.008125795051455498,
|
| 5855 |
+
"learning_rate": 3.4042845013198456e-05,
|
| 5856 |
+
"loss": 0.0002,
|
| 5857 |
+
"step": 832
|
| 5858 |
+
},
|
| 5859 |
+
{
|
| 5860 |
+
"epoch": 0.6186409209060527,
|
| 5861 |
+
"grad_norm": 0.13623641431331635,
|
| 5862 |
+
"learning_rate": 3.392802673484193e-05,
|
| 5863 |
+
"loss": 0.0018,
|
| 5864 |
+
"step": 833
|
| 5865 |
+
},
|
| 5866 |
+
{
|
| 5867 |
+
"epoch": 0.6193835870776087,
|
| 5868 |
+
"grad_norm": 0.000562251138035208,
|
| 5869 |
+
"learning_rate": 3.381330289708396e-05,
|
| 5870 |
+
"loss": 0.0001,
|
| 5871 |
+
"step": 834
|
| 5872 |
+
},
|
| 5873 |
+
{
|
| 5874 |
+
"epoch": 0.6201262532491645,
|
| 5875 |
+
"grad_norm": 0.0013340302975848317,
|
| 5876 |
+
"learning_rate": 3.3698674174053844e-05,
|
| 5877 |
+
"loss": 0.0001,
|
| 5878 |
+
"step": 835
|
| 5879 |
+
},
|
| 5880 |
+
{
|
| 5881 |
+
"epoch": 0.6208689194207204,
|
| 5882 |
+
"grad_norm": 0.0081217335537076,
|
| 5883 |
+
"learning_rate": 3.358414123932195e-05,
|
| 5884 |
+
"loss": 0.0005,
|
| 5885 |
+
"step": 836
|
| 5886 |
+
},
|
| 5887 |
+
{
|
| 5888 |
+
"epoch": 0.6216115855922763,
|
| 5889 |
+
"grad_norm": 0.059895727783441544,
|
| 5890 |
+
"learning_rate": 3.346970476589583e-05,
|
| 5891 |
+
"loss": 0.0021,
|
| 5892 |
+
"step": 837
|
| 5893 |
+
},
|
| 5894 |
+
{
|
| 5895 |
+
"epoch": 0.6223542517638322,
|
| 5896 |
+
"grad_norm": 0.0007285134051926434,
|
| 5897 |
+
"learning_rate": 3.3355365426216164e-05,
|
| 5898 |
+
"loss": 0.0001,
|
| 5899 |
+
"step": 838
|
| 5900 |
+
},
|
| 5901 |
+
{
|
| 5902 |
+
"epoch": 0.623096917935388,
|
| 5903 |
+
"grad_norm": 0.0223313607275486,
|
| 5904 |
+
"learning_rate": 3.324112389215293e-05,
|
| 5905 |
+
"loss": 0.0005,
|
| 5906 |
+
"step": 839
|
| 5907 |
+
},
|
| 5908 |
+
{
|
| 5909 |
+
"epoch": 0.623839584106944,
|
| 5910 |
+
"grad_norm": 0.0015575829893350601,
|
| 5911 |
+
"learning_rate": 3.31269808350013e-05,
|
| 5912 |
+
"loss": 0.0001,
|
| 5913 |
+
"step": 840
|
| 5914 |
+
},
|
| 5915 |
+
{
|
| 5916 |
+
"epoch": 0.6245822502784998,
|
| 5917 |
+
"grad_norm": 0.001839678268879652,
|
| 5918 |
+
"learning_rate": 3.3012936925477857e-05,
|
| 5919 |
+
"loss": 0.0001,
|
| 5920 |
+
"step": 841
|
| 5921 |
+
},
|
| 5922 |
+
{
|
| 5923 |
+
"epoch": 0.6253249164500557,
|
| 5924 |
+
"grad_norm": 0.0008310491102747619,
|
| 5925 |
+
"learning_rate": 3.289899283371657e-05,
|
| 5926 |
+
"loss": 0.0001,
|
| 5927 |
+
"step": 842
|
| 5928 |
+
},
|
| 5929 |
+
{
|
| 5930 |
+
"epoch": 0.6260675826216116,
|
| 5931 |
+
"grad_norm": 0.0006663409876637161,
|
| 5932 |
+
"learning_rate": 3.278514922926484e-05,
|
| 5933 |
+
"loss": 0.0001,
|
| 5934 |
+
"step": 843
|
| 5935 |
+
},
|
| 5936 |
+
{
|
| 5937 |
+
"epoch": 0.6268102487931675,
|
| 5938 |
+
"grad_norm": 0.0015589938266202807,
|
| 5939 |
+
"learning_rate": 3.267140678107964e-05,
|
| 5940 |
+
"loss": 0.0001,
|
| 5941 |
+
"step": 844
|
| 5942 |
+
},
|
| 5943 |
+
{
|
| 5944 |
+
"epoch": 0.6275529149647233,
|
| 5945 |
+
"grad_norm": 0.003392728278413415,
|
| 5946 |
+
"learning_rate": 3.2557766157523465e-05,
|
| 5947 |
+
"loss": 0.0001,
|
| 5948 |
+
"step": 845
|
| 5949 |
+
},
|
| 5950 |
+
{
|
| 5951 |
+
"epoch": 0.6282955811362793,
|
| 5952 |
+
"grad_norm": 0.0010547033743932843,
|
| 5953 |
+
"learning_rate": 3.244422802636057e-05,
|
| 5954 |
+
"loss": 0.0001,
|
| 5955 |
+
"step": 846
|
| 5956 |
+
},
|
| 5957 |
+
{
|
| 5958 |
+
"epoch": 0.6290382473078351,
|
| 5959 |
+
"grad_norm": 0.0024194575380533934,
|
| 5960 |
+
"learning_rate": 3.2330793054752873e-05,
|
| 5961 |
+
"loss": 0.0002,
|
| 5962 |
+
"step": 847
|
| 5963 |
+
},
|
| 5964 |
+
{
|
| 5965 |
+
"epoch": 0.629780913479391,
|
| 5966 |
+
"grad_norm": 0.0006333817727863789,
|
| 5967 |
+
"learning_rate": 3.2217461909256185e-05,
|
| 5968 |
+
"loss": 0.0001,
|
| 5969 |
+
"step": 848
|
| 5970 |
+
},
|
| 5971 |
+
{
|
| 5972 |
+
"epoch": 0.6305235796509469,
|
| 5973 |
+
"grad_norm": 0.0008936111116781831,
|
| 5974 |
+
"learning_rate": 3.210423525581615e-05,
|
| 5975 |
+
"loss": 0.0001,
|
| 5976 |
+
"step": 849
|
| 5977 |
+
},
|
| 5978 |
+
{
|
| 5979 |
+
"epoch": 0.6312662458225028,
|
| 5980 |
+
"grad_norm": 0.0010497324401512742,
|
| 5981 |
+
"learning_rate": 3.199111375976449e-05,
|
| 5982 |
+
"loss": 0.0001,
|
| 5983 |
+
"step": 850
|
| 5984 |
+
},
|
| 5985 |
+
{
|
| 5986 |
+
"epoch": 0.6320089119940586,
|
| 5987 |
+
"grad_norm": 0.010097804479300976,
|
| 5988 |
+
"learning_rate": 3.1878098085814924e-05,
|
| 5989 |
+
"loss": 0.0005,
|
| 5990 |
+
"step": 851
|
| 5991 |
+
},
|
| 5992 |
+
{
|
| 5993 |
+
"epoch": 0.6327515781656146,
|
| 5994 |
+
"grad_norm": 0.003449138719588518,
|
| 5995 |
+
"learning_rate": 3.176518889805941e-05,
|
| 5996 |
+
"loss": 0.0001,
|
| 5997 |
+
"step": 852
|
| 5998 |
+
},
|
| 5999 |
+
{
|
| 6000 |
+
"epoch": 0.6334942443371704,
|
| 6001 |
+
"grad_norm": 0.020226208493113518,
|
| 6002 |
+
"learning_rate": 3.1652386859964156e-05,
|
| 6003 |
+
"loss": 0.0004,
|
| 6004 |
+
"step": 853
|
| 6005 |
+
},
|
| 6006 |
+
{
|
| 6007 |
+
"epoch": 0.6342369105087263,
|
| 6008 |
+
"grad_norm": 0.0008942090207710862,
|
| 6009 |
+
"learning_rate": 3.153969263436579e-05,
|
| 6010 |
+
"loss": 0.0001,
|
| 6011 |
+
"step": 854
|
| 6012 |
+
},
|
| 6013 |
+
{
|
| 6014 |
+
"epoch": 0.6349795766802823,
|
| 6015 |
+
"grad_norm": 0.005887329112738371,
|
| 6016 |
+
"learning_rate": 3.142710688346733e-05,
|
| 6017 |
+
"loss": 0.0002,
|
| 6018 |
+
"step": 855
|
| 6019 |
+
},
|
| 6020 |
+
{
|
| 6021 |
+
"epoch": 0.6357222428518381,
|
| 6022 |
+
"grad_norm": 0.0013140842784196138,
|
| 6023 |
+
"learning_rate": 3.131463026883449e-05,
|
| 6024 |
+
"loss": 0.0001,
|
| 6025 |
+
"step": 856
|
| 6026 |
+
},
|
| 6027 |
+
{
|
| 6028 |
+
"epoch": 0.636464909023394,
|
| 6029 |
+
"grad_norm": 0.001480376347899437,
|
| 6030 |
+
"learning_rate": 3.12022634513916e-05,
|
| 6031 |
+
"loss": 0.0002,
|
| 6032 |
+
"step": 857
|
| 6033 |
+
},
|
| 6034 |
+
{
|
| 6035 |
+
"epoch": 0.6372075751949499,
|
| 6036 |
+
"grad_norm": 0.0014736526645720005,
|
| 6037 |
+
"learning_rate": 3.109000709141788e-05,
|
| 6038 |
+
"loss": 0.0001,
|
| 6039 |
+
"step": 858
|
| 6040 |
+
},
|
| 6041 |
+
{
|
| 6042 |
+
"epoch": 0.6379502413665058,
|
| 6043 |
+
"grad_norm": 0.0005073303473182023,
|
| 6044 |
+
"learning_rate": 3.0977861848543444e-05,
|
| 6045 |
+
"loss": 0.0001,
|
| 6046 |
+
"step": 859
|
| 6047 |
+
},
|
| 6048 |
+
{
|
| 6049 |
+
"epoch": 0.6386929075380616,
|
| 6050 |
+
"grad_norm": 0.0009759328095242381,
|
| 6051 |
+
"learning_rate": 3.086582838174551e-05,
|
| 6052 |
+
"loss": 0.0001,
|
| 6053 |
+
"step": 860
|
| 6054 |
+
},
|
| 6055 |
+
{
|
| 6056 |
+
"epoch": 0.6394355737096176,
|
| 6057 |
+
"grad_norm": 0.0012429956113919616,
|
| 6058 |
+
"learning_rate": 3.0753907349344464e-05,
|
| 6059 |
+
"loss": 0.0001,
|
| 6060 |
+
"step": 861
|
| 6061 |
+
},
|
| 6062 |
+
{
|
| 6063 |
+
"epoch": 0.6401782398811734,
|
| 6064 |
+
"grad_norm": 0.003437102073803544,
|
| 6065 |
+
"learning_rate": 3.064209940899998e-05,
|
| 6066 |
+
"loss": 0.0002,
|
| 6067 |
+
"step": 862
|
| 6068 |
+
},
|
| 6069 |
+
{
|
| 6070 |
+
"epoch": 0.6409209060527293,
|
| 6071 |
+
"grad_norm": 0.0044972775503993034,
|
| 6072 |
+
"learning_rate": 3.0530405217707256e-05,
|
| 6073 |
+
"loss": 0.0003,
|
| 6074 |
+
"step": 863
|
| 6075 |
+
},
|
| 6076 |
+
{
|
| 6077 |
+
"epoch": 0.6416635722242852,
|
| 6078 |
+
"grad_norm": 0.0003766078152693808,
|
| 6079 |
+
"learning_rate": 3.0418825431793052e-05,
|
| 6080 |
+
"loss": 0.0001,
|
| 6081 |
+
"step": 864
|
| 6082 |
+
},
|
| 6083 |
+
{
|
| 6084 |
+
"epoch": 0.6424062383958411,
|
| 6085 |
+
"grad_norm": 0.000993401394225657,
|
| 6086 |
+
"learning_rate": 3.0307360706911896e-05,
|
| 6087 |
+
"loss": 0.0001,
|
| 6088 |
+
"step": 865
|
| 6089 |
+
},
|
| 6090 |
+
{
|
| 6091 |
+
"epoch": 0.6431489045673969,
|
| 6092 |
+
"grad_norm": 0.0007390630780719221,
|
| 6093 |
+
"learning_rate": 3.019601169804216e-05,
|
| 6094 |
+
"loss": 0.0001,
|
| 6095 |
+
"step": 866
|
| 6096 |
+
},
|
| 6097 |
+
{
|
| 6098 |
+
"epoch": 0.6438915707389529,
|
| 6099 |
+
"grad_norm": 0.0011397154303267598,
|
| 6100 |
+
"learning_rate": 3.0084779059482322e-05,
|
| 6101 |
+
"loss": 0.0001,
|
| 6102 |
+
"step": 867
|
| 6103 |
+
},
|
| 6104 |
+
{
|
| 6105 |
+
"epoch": 0.6446342369105087,
|
| 6106 |
+
"grad_norm": 0.004833963233977556,
|
| 6107 |
+
"learning_rate": 2.997366344484698e-05,
|
| 6108 |
+
"loss": 0.0002,
|
| 6109 |
+
"step": 868
|
| 6110 |
+
},
|
| 6111 |
+
{
|
| 6112 |
+
"epoch": 0.6453769030820646,
|
| 6113 |
+
"grad_norm": 0.007242708466947079,
|
| 6114 |
+
"learning_rate": 2.9862665507063147e-05,
|
| 6115 |
+
"loss": 0.0002,
|
| 6116 |
+
"step": 869
|
| 6117 |
+
},
|
| 6118 |
+
{
|
| 6119 |
+
"epoch": 0.6461195692536205,
|
| 6120 |
+
"grad_norm": 0.0021677876356989145,
|
| 6121 |
+
"learning_rate": 2.975178589836632e-05,
|
| 6122 |
+
"loss": 0.0001,
|
| 6123 |
+
"step": 870
|
| 6124 |
+
},
|
| 6125 |
+
{
|
| 6126 |
+
"epoch": 0.6468622354251764,
|
| 6127 |
+
"grad_norm": 0.0013036590535193682,
|
| 6128 |
+
"learning_rate": 2.9641025270296722e-05,
|
| 6129 |
+
"loss": 0.0001,
|
| 6130 |
+
"step": 871
|
| 6131 |
+
},
|
| 6132 |
+
{
|
| 6133 |
+
"epoch": 0.6476049015967322,
|
| 6134 |
+
"grad_norm": 0.0008346426184289157,
|
| 6135 |
+
"learning_rate": 2.953038427369537e-05,
|
| 6136 |
+
"loss": 0.0001,
|
| 6137 |
+
"step": 872
|
| 6138 |
+
},
|
| 6139 |
+
{
|
| 6140 |
+
"epoch": 0.6483475677682882,
|
| 6141 |
+
"grad_norm": 0.0006587895331904292,
|
| 6142 |
+
"learning_rate": 2.9419863558700396e-05,
|
| 6143 |
+
"loss": 0.0001,
|
| 6144 |
+
"step": 873
|
| 6145 |
+
},
|
| 6146 |
+
{
|
| 6147 |
+
"epoch": 0.649090233939844,
|
| 6148 |
+
"grad_norm": 0.0005403941031545401,
|
| 6149 |
+
"learning_rate": 2.9309463774743046e-05,
|
| 6150 |
+
"loss": 0.0001,
|
| 6151 |
+
"step": 874
|
| 6152 |
+
},
|
| 6153 |
+
{
|
| 6154 |
+
"epoch": 0.6498329001113999,
|
| 6155 |
+
"grad_norm": 0.001244431477971375,
|
| 6156 |
+
"learning_rate": 2.919918557054405e-05,
|
| 6157 |
+
"loss": 0.0001,
|
| 6158 |
+
"step": 875
|
| 6159 |
+
},
|
| 6160 |
+
{
|
| 6161 |
+
"epoch": 0.6505755662829558,
|
| 6162 |
+
"grad_norm": 0.004742435645312071,
|
| 6163 |
+
"learning_rate": 2.9089029594109683e-05,
|
| 6164 |
+
"loss": 0.0002,
|
| 6165 |
+
"step": 876
|
| 6166 |
+
},
|
| 6167 |
+
{
|
| 6168 |
+
"epoch": 0.6513182324545117,
|
| 6169 |
+
"grad_norm": 0.0021011109929531813,
|
| 6170 |
+
"learning_rate": 2.8978996492727994e-05,
|
| 6171 |
+
"loss": 0.0002,
|
| 6172 |
+
"step": 877
|
| 6173 |
+
},
|
| 6174 |
+
{
|
| 6175 |
+
"epoch": 0.6520608986260675,
|
| 6176 |
+
"grad_norm": 0.001175252953544259,
|
| 6177 |
+
"learning_rate": 2.886908691296504e-05,
|
| 6178 |
+
"loss": 0.0001,
|
| 6179 |
+
"step": 878
|
| 6180 |
+
},
|
| 6181 |
+
{
|
| 6182 |
+
"epoch": 0.6528035647976235,
|
| 6183 |
+
"grad_norm": 0.0017059053061529994,
|
| 6184 |
+
"learning_rate": 2.8759301500660972e-05,
|
| 6185 |
+
"loss": 0.0001,
|
| 6186 |
+
"step": 879
|
| 6187 |
+
},
|
| 6188 |
+
{
|
| 6189 |
+
"epoch": 0.6535462309691794,
|
| 6190 |
+
"grad_norm": 0.0007660607225261629,
|
| 6191 |
+
"learning_rate": 2.864964090092644e-05,
|
| 6192 |
+
"loss": 0.0001,
|
| 6193 |
+
"step": 880
|
| 6194 |
+
},
|
| 6195 |
+
{
|
| 6196 |
+
"epoch": 0.6542888971407352,
|
| 6197 |
+
"grad_norm": 0.001221244689077139,
|
| 6198 |
+
"learning_rate": 2.854010575813856e-05,
|
| 6199 |
+
"loss": 0.0001,
|
| 6200 |
+
"step": 881
|
| 6201 |
+
},
|
| 6202 |
+
{
|
| 6203 |
+
"epoch": 0.6550315633122912,
|
| 6204 |
+
"grad_norm": 0.0004929836140945554,
|
| 6205 |
+
"learning_rate": 2.8430696715937337e-05,
|
| 6206 |
+
"loss": 0.0001,
|
| 6207 |
+
"step": 882
|
| 6208 |
+
},
|
| 6209 |
+
{
|
| 6210 |
+
"epoch": 0.655774229483847,
|
| 6211 |
+
"grad_norm": 0.0009183657239191234,
|
| 6212 |
+
"learning_rate": 2.8321414417221746e-05,
|
| 6213 |
+
"loss": 0.0001,
|
| 6214 |
+
"step": 883
|
| 6215 |
+
},
|
| 6216 |
+
{
|
| 6217 |
+
"epoch": 0.6565168956554029,
|
| 6218 |
+
"grad_norm": 0.0003201389918103814,
|
| 6219 |
+
"learning_rate": 2.8212259504146043e-05,
|
| 6220 |
+
"loss": 0.0001,
|
| 6221 |
+
"step": 884
|
| 6222 |
+
},
|
| 6223 |
+
{
|
| 6224 |
+
"epoch": 0.6572595618269588,
|
| 6225 |
+
"grad_norm": 0.0023057435173541307,
|
| 6226 |
+
"learning_rate": 2.8103232618115903e-05,
|
| 6227 |
+
"loss": 0.0001,
|
| 6228 |
+
"step": 885
|
| 6229 |
+
},
|
| 6230 |
+
{
|
| 6231 |
+
"epoch": 0.6580022279985147,
|
| 6232 |
+
"grad_norm": 0.0004039146879222244,
|
| 6233 |
+
"learning_rate": 2.7994334399784772e-05,
|
| 6234 |
+
"loss": 0.0001,
|
| 6235 |
+
"step": 886
|
| 6236 |
+
},
|
| 6237 |
+
{
|
| 6238 |
+
"epoch": 0.6587448941700705,
|
| 6239 |
+
"grad_norm": 0.001328380312770605,
|
| 6240 |
+
"learning_rate": 2.7885565489049946e-05,
|
| 6241 |
+
"loss": 0.0001,
|
| 6242 |
+
"step": 887
|
| 6243 |
+
},
|
| 6244 |
+
{
|
| 6245 |
+
"epoch": 0.6594875603416265,
|
| 6246 |
+
"grad_norm": 0.0009456725674681365,
|
| 6247 |
+
"learning_rate": 2.777692652504895e-05,
|
| 6248 |
+
"loss": 0.0001,
|
| 6249 |
+
"step": 888
|
| 6250 |
+
},
|
| 6251 |
+
{
|
| 6252 |
+
"epoch": 0.6602302265131823,
|
| 6253 |
+
"grad_norm": 0.0006427007610909641,
|
| 6254 |
+
"learning_rate": 2.766841814615573e-05,
|
| 6255 |
+
"loss": 0.0001,
|
| 6256 |
+
"step": 889
|
| 6257 |
+
},
|
| 6258 |
+
{
|
| 6259 |
+
"epoch": 0.6609728926847382,
|
| 6260 |
+
"grad_norm": 0.0006173664005473256,
|
| 6261 |
+
"learning_rate": 2.7560040989976892e-05,
|
| 6262 |
+
"loss": 0.0001,
|
| 6263 |
+
"step": 890
|
| 6264 |
+
},
|
| 6265 |
+
{
|
| 6266 |
+
"epoch": 0.6617155588562941,
|
| 6267 |
+
"grad_norm": 0.0005695072468370199,
|
| 6268 |
+
"learning_rate": 2.7451795693347936e-05,
|
| 6269 |
+
"loss": 0.0001,
|
| 6270 |
+
"step": 891
|
| 6271 |
+
},
|
| 6272 |
+
{
|
| 6273 |
+
"epoch": 0.66245822502785,
|
| 6274 |
+
"grad_norm": 0.000993382534943521,
|
| 6275 |
+
"learning_rate": 2.734368289232959e-05,
|
| 6276 |
+
"loss": 0.0001,
|
| 6277 |
+
"step": 892
|
| 6278 |
+
},
|
| 6279 |
+
{
|
| 6280 |
+
"epoch": 0.6632008911994058,
|
| 6281 |
+
"grad_norm": 0.00402169581502676,
|
| 6282 |
+
"learning_rate": 2.723570322220399e-05,
|
| 6283 |
+
"loss": 0.0001,
|
| 6284 |
+
"step": 893
|
| 6285 |
+
},
|
| 6286 |
+
{
|
| 6287 |
+
"epoch": 0.6639435573709618,
|
| 6288 |
+
"grad_norm": 0.00042257923632860184,
|
| 6289 |
+
"learning_rate": 2.7127857317470968e-05,
|
| 6290 |
+
"loss": 0.0001,
|
| 6291 |
+
"step": 894
|
| 6292 |
+
},
|
| 6293 |
+
{
|
| 6294 |
+
"epoch": 0.6646862235425176,
|
| 6295 |
+
"grad_norm": 0.0011959762778133154,
|
| 6296 |
+
"learning_rate": 2.702014581184439e-05,
|
| 6297 |
+
"loss": 0.0001,
|
| 6298 |
+
"step": 895
|
| 6299 |
+
},
|
| 6300 |
+
{
|
| 6301 |
+
"epoch": 0.6654288897140735,
|
| 6302 |
+
"grad_norm": 0.0022542597725987434,
|
| 6303 |
+
"learning_rate": 2.6912569338248315e-05,
|
| 6304 |
+
"loss": 0.0001,
|
| 6305 |
+
"step": 896
|
| 6306 |
+
},
|
| 6307 |
+
{
|
| 6308 |
+
"epoch": 0.6661715558856294,
|
| 6309 |
+
"grad_norm": 0.0031469352543354034,
|
| 6310 |
+
"learning_rate": 2.6805128528813384e-05,
|
| 6311 |
+
"loss": 0.0002,
|
| 6312 |
+
"step": 897
|
| 6313 |
+
},
|
| 6314 |
+
{
|
| 6315 |
+
"epoch": 0.6669142220571853,
|
| 6316 |
+
"grad_norm": 0.0017010547453537583,
|
| 6317 |
+
"learning_rate": 2.6697824014873075e-05,
|
| 6318 |
+
"loss": 0.0002,
|
| 6319 |
+
"step": 898
|
| 6320 |
+
},
|
| 6321 |
+
{
|
| 6322 |
+
"epoch": 0.6676568882287411,
|
| 6323 |
+
"grad_norm": 0.003644139738753438,
|
| 6324 |
+
"learning_rate": 2.6590656426959902e-05,
|
| 6325 |
+
"loss": 0.0002,
|
| 6326 |
+
"step": 899
|
| 6327 |
+
},
|
| 6328 |
+
{
|
| 6329 |
+
"epoch": 0.6683995544002971,
|
| 6330 |
+
"grad_norm": 0.0016295432578772306,
|
| 6331 |
+
"learning_rate": 2.6483626394801873e-05,
|
| 6332 |
+
"loss": 0.0002,
|
| 6333 |
+
"step": 900
|
| 6334 |
+
},
|
| 6335 |
+
{
|
| 6336 |
+
"epoch": 0.6691422205718529,
|
| 6337 |
+
"grad_norm": 0.012327441945672035,
|
| 6338 |
+
"learning_rate": 2.637673454731868e-05,
|
| 6339 |
+
"loss": 0.0003,
|
| 6340 |
+
"step": 901
|
| 6341 |
+
},
|
| 6342 |
+
{
|
| 6343 |
+
"epoch": 0.6698848867434088,
|
| 6344 |
+
"grad_norm": 0.2909511923789978,
|
| 6345 |
+
"learning_rate": 2.626998151261798e-05,
|
| 6346 |
+
"loss": 0.004,
|
| 6347 |
+
"step": 902
|
| 6348 |
+
},
|
| 6349 |
+
{
|
| 6350 |
+
"epoch": 0.6706275529149647,
|
| 6351 |
+
"grad_norm": 0.001305720885284245,
|
| 6352 |
+
"learning_rate": 2.6163367917991822e-05,
|
| 6353 |
+
"loss": 0.0001,
|
| 6354 |
+
"step": 903
|
| 6355 |
+
},
|
| 6356 |
+
{
|
| 6357 |
+
"epoch": 0.6713702190865206,
|
| 6358 |
+
"grad_norm": 0.0014534946531057358,
|
| 6359 |
+
"learning_rate": 2.605689438991282e-05,
|
| 6360 |
+
"loss": 0.0001,
|
| 6361 |
+
"step": 904
|
| 6362 |
+
},
|
| 6363 |
+
{
|
| 6364 |
+
"epoch": 0.6721128852580764,
|
| 6365 |
+
"grad_norm": 0.003072174033150077,
|
| 6366 |
+
"learning_rate": 2.595056155403063e-05,
|
| 6367 |
+
"loss": 0.0002,
|
| 6368 |
+
"step": 905
|
| 6369 |
+
},
|
| 6370 |
+
{
|
| 6371 |
+
"epoch": 0.6728555514296324,
|
| 6372 |
+
"grad_norm": 0.004174454137682915,
|
| 6373 |
+
"learning_rate": 2.5844370035168073e-05,
|
| 6374 |
+
"loss": 0.0002,
|
| 6375 |
+
"step": 906
|
| 6376 |
+
},
|
| 6377 |
+
{
|
| 6378 |
+
"epoch": 0.6735982176011883,
|
| 6379 |
+
"grad_norm": 0.001340761547908187,
|
| 6380 |
+
"learning_rate": 2.5738320457317698e-05,
|
| 6381 |
+
"loss": 0.0001,
|
| 6382 |
+
"step": 907
|
| 6383 |
+
},
|
| 6384 |
+
{
|
| 6385 |
+
"epoch": 0.6743408837727441,
|
| 6386 |
+
"grad_norm": 0.5977171063423157,
|
| 6387 |
+
"learning_rate": 2.5632413443637888e-05,
|
| 6388 |
+
"loss": 0.0386,
|
| 6389 |
+
"step": 908
|
| 6390 |
+
},
|
| 6391 |
+
{
|
| 6392 |
+
"epoch": 0.6750835499443001,
|
| 6393 |
+
"grad_norm": 0.000558204366825521,
|
| 6394 |
+
"learning_rate": 2.552664961644936e-05,
|
| 6395 |
+
"loss": 0.0001,
|
| 6396 |
+
"step": 909
|
| 6397 |
+
},
|
| 6398 |
+
{
|
| 6399 |
+
"epoch": 0.6758262161158559,
|
| 6400 |
+
"grad_norm": 0.0011802364606410265,
|
| 6401 |
+
"learning_rate": 2.5421029597231478e-05,
|
| 6402 |
+
"loss": 0.0001,
|
| 6403 |
+
"step": 910
|
| 6404 |
+
},
|
| 6405 |
+
{
|
| 6406 |
+
"epoch": 0.6765688822874119,
|
| 6407 |
+
"grad_norm": 0.000828778138384223,
|
| 6408 |
+
"learning_rate": 2.5315554006618485e-05,
|
| 6409 |
+
"loss": 0.0001,
|
| 6410 |
+
"step": 911
|
| 6411 |
+
},
|
| 6412 |
+
{
|
| 6413 |
+
"epoch": 0.6773115484589677,
|
| 6414 |
+
"grad_norm": 0.0003953818522859365,
|
| 6415 |
+
"learning_rate": 2.5210223464396053e-05,
|
| 6416 |
+
"loss": 0.0001,
|
| 6417 |
+
"step": 912
|
| 6418 |
+
},
|
| 6419 |
+
{
|
| 6420 |
+
"epoch": 0.6780542146305236,
|
| 6421 |
+
"grad_norm": 0.0005724510992877185,
|
| 6422 |
+
"learning_rate": 2.5105038589497433e-05,
|
| 6423 |
+
"loss": 0.0001,
|
| 6424 |
+
"step": 913
|
| 6425 |
+
},
|
| 6426 |
+
{
|
| 6427 |
+
"epoch": 0.6787968808020794,
|
| 6428 |
+
"grad_norm": 0.0006651621079072356,
|
| 6429 |
+
"learning_rate": 2.500000000000001e-05,
|
| 6430 |
+
"loss": 0.0001,
|
| 6431 |
+
"step": 914
|
| 6432 |
+
},
|
| 6433 |
+
{
|
| 6434 |
+
"epoch": 0.6795395469736354,
|
| 6435 |
+
"grad_norm": 0.0022599331568926573,
|
| 6436 |
+
"learning_rate": 2.4895108313121508e-05,
|
| 6437 |
+
"loss": 0.0002,
|
| 6438 |
+
"step": 915
|
| 6439 |
+
},
|
| 6440 |
+
{
|
| 6441 |
+
"epoch": 0.6802822131451912,
|
| 6442 |
+
"grad_norm": 0.0014916729414835572,
|
| 6443 |
+
"learning_rate": 2.4790364145216495e-05,
|
| 6444 |
+
"loss": 0.0001,
|
| 6445 |
+
"step": 916
|
| 6446 |
+
},
|
| 6447 |
+
{
|
| 6448 |
+
"epoch": 0.6810248793167472,
|
| 6449 |
+
"grad_norm": 0.004133202601224184,
|
| 6450 |
+
"learning_rate": 2.4685768111772644e-05,
|
| 6451 |
+
"loss": 0.0002,
|
| 6452 |
+
"step": 917
|
| 6453 |
+
},
|
| 6454 |
+
{
|
| 6455 |
+
"epoch": 0.681767545488303,
|
| 6456 |
+
"grad_norm": 0.06581906974315643,
|
| 6457 |
+
"learning_rate": 2.458132082740724e-05,
|
| 6458 |
+
"loss": 0.0024,
|
| 6459 |
+
"step": 918
|
| 6460 |
+
},
|
| 6461 |
+
{
|
| 6462 |
+
"epoch": 0.6825102116598589,
|
| 6463 |
+
"grad_norm": 0.0005587110063061118,
|
| 6464 |
+
"learning_rate": 2.4477022905863427e-05,
|
| 6465 |
+
"loss": 0.0001,
|
| 6466 |
+
"step": 919
|
| 6467 |
+
},
|
| 6468 |
+
{
|
| 6469 |
+
"epoch": 0.6832528778314148,
|
| 6470 |
+
"grad_norm": 0.0006018027779646218,
|
| 6471 |
+
"learning_rate": 2.4372874960006743e-05,
|
| 6472 |
+
"loss": 0.0001,
|
| 6473 |
+
"step": 920
|
| 6474 |
+
},
|
| 6475 |
+
{
|
| 6476 |
+
"epoch": 0.6839955440029707,
|
| 6477 |
+
"grad_norm": 0.0009498025756329298,
|
| 6478 |
+
"learning_rate": 2.4268877601821423e-05,
|
| 6479 |
+
"loss": 0.0001,
|
| 6480 |
+
"step": 921
|
| 6481 |
+
},
|
| 6482 |
+
{
|
| 6483 |
+
"epoch": 0.6847382101745265,
|
| 6484 |
+
"grad_norm": 0.0018847642932087183,
|
| 6485 |
+
"learning_rate": 2.4165031442406855e-05,
|
| 6486 |
+
"loss": 0.0001,
|
| 6487 |
+
"step": 922
|
| 6488 |
+
},
|
| 6489 |
+
{
|
| 6490 |
+
"epoch": 0.6854808763460825,
|
| 6491 |
+
"grad_norm": 0.0015400544507429004,
|
| 6492 |
+
"learning_rate": 2.4061337091973918e-05,
|
| 6493 |
+
"loss": 0.0001,
|
| 6494 |
+
"step": 923
|
| 6495 |
+
},
|
| 6496 |
+
{
|
| 6497 |
+
"epoch": 0.6862235425176383,
|
| 6498 |
+
"grad_norm": 0.0018971724202856421,
|
| 6499 |
+
"learning_rate": 2.3957795159841517e-05,
|
| 6500 |
+
"loss": 0.0001,
|
| 6501 |
+
"step": 924
|
| 6502 |
+
},
|
| 6503 |
+
{
|
| 6504 |
+
"epoch": 0.6869662086891942,
|
| 6505 |
+
"grad_norm": 0.004078839905560017,
|
| 6506 |
+
"learning_rate": 2.385440625443287e-05,
|
| 6507 |
+
"loss": 0.0002,
|
| 6508 |
+
"step": 925
|
| 6509 |
+
},
|
| 6510 |
+
{
|
| 6511 |
+
"epoch": 0.68770887486075,
|
| 6512 |
+
"grad_norm": 0.0012903795577585697,
|
| 6513 |
+
"learning_rate": 2.3751170983272e-05,
|
| 6514 |
+
"loss": 0.0002,
|
| 6515 |
+
"step": 926
|
| 6516 |
+
},
|
| 6517 |
+
{
|
| 6518 |
+
"epoch": 0.688451541032306,
|
| 6519 |
+
"grad_norm": 0.0016058724140748382,
|
| 6520 |
+
"learning_rate": 2.3648089952980202e-05,
|
| 6521 |
+
"loss": 0.0001,
|
| 6522 |
+
"step": 927
|
| 6523 |
+
},
|
| 6524 |
+
{
|
| 6525 |
+
"epoch": 0.6891942072038618,
|
| 6526 |
+
"grad_norm": 0.002747948979958892,
|
| 6527 |
+
"learning_rate": 2.3545163769272383e-05,
|
| 6528 |
+
"loss": 0.0001,
|
| 6529 |
+
"step": 928
|
| 6530 |
+
},
|
| 6531 |
+
{
|
| 6532 |
+
"epoch": 0.6899368733754178,
|
| 6533 |
+
"grad_norm": 0.0010071592405438423,
|
| 6534 |
+
"learning_rate": 2.3442393036953613e-05,
|
| 6535 |
+
"loss": 0.0001,
|
| 6536 |
+
"step": 929
|
| 6537 |
+
},
|
| 6538 |
+
{
|
| 6539 |
+
"epoch": 0.6906795395469736,
|
| 6540 |
+
"grad_norm": 0.003560745855793357,
|
| 6541 |
+
"learning_rate": 2.333977835991545e-05,
|
| 6542 |
+
"loss": 0.0001,
|
| 6543 |
+
"step": 930
|
| 6544 |
+
},
|
| 6545 |
+
{
|
| 6546 |
+
"epoch": 0.6914222057185295,
|
| 6547 |
+
"grad_norm": 0.0034868186339735985,
|
| 6548 |
+
"learning_rate": 2.3237320341132497e-05,
|
| 6549 |
+
"loss": 0.0002,
|
| 6550 |
+
"step": 931
|
| 6551 |
+
},
|
| 6552 |
+
{
|
| 6553 |
+
"epoch": 0.6921648718900854,
|
| 6554 |
+
"grad_norm": 0.0018978634616360068,
|
| 6555 |
+
"learning_rate": 2.3135019582658802e-05,
|
| 6556 |
+
"loss": 0.0002,
|
| 6557 |
+
"step": 932
|
| 6558 |
+
},
|
| 6559 |
+
{
|
| 6560 |
+
"epoch": 0.6929075380616413,
|
| 6561 |
+
"grad_norm": 0.000817135616671294,
|
| 6562 |
+
"learning_rate": 2.3032876685624367e-05,
|
| 6563 |
+
"loss": 0.0001,
|
| 6564 |
+
"step": 933
|
| 6565 |
+
},
|
| 6566 |
+
{
|
| 6567 |
+
"epoch": 0.6936502042331972,
|
| 6568 |
+
"grad_norm": 0.002365557011216879,
|
| 6569 |
+
"learning_rate": 2.2930892250231516e-05,
|
| 6570 |
+
"loss": 0.0001,
|
| 6571 |
+
"step": 934
|
| 6572 |
+
},
|
| 6573 |
+
{
|
| 6574 |
+
"epoch": 0.6943928704047531,
|
| 6575 |
+
"grad_norm": 0.0016304274322465062,
|
| 6576 |
+
"learning_rate": 2.282906687575151e-05,
|
| 6577 |
+
"loss": 0.0001,
|
| 6578 |
+
"step": 935
|
| 6579 |
+
},
|
| 6580 |
+
{
|
| 6581 |
+
"epoch": 0.695135536576309,
|
| 6582 |
+
"grad_norm": 0.004060409963130951,
|
| 6583 |
+
"learning_rate": 2.2727401160520876e-05,
|
| 6584 |
+
"loss": 0.0002,
|
| 6585 |
+
"step": 936
|
| 6586 |
+
},
|
| 6587 |
+
{
|
| 6588 |
+
"epoch": 0.6958782027478648,
|
| 6589 |
+
"grad_norm": 0.0014790298882871866,
|
| 6590 |
+
"learning_rate": 2.262589570193805e-05,
|
| 6591 |
+
"loss": 0.0001,
|
| 6592 |
+
"step": 937
|
| 6593 |
+
},
|
| 6594 |
+
{
|
| 6595 |
+
"epoch": 0.6966208689194208,
|
| 6596 |
+
"grad_norm": 0.0010480373166501522,
|
| 6597 |
+
"learning_rate": 2.25245510964597e-05,
|
| 6598 |
+
"loss": 0.0001,
|
| 6599 |
+
"step": 938
|
| 6600 |
+
},
|
| 6601 |
+
{
|
| 6602 |
+
"epoch": 0.6973635350909766,
|
| 6603 |
+
"grad_norm": 0.0022476809099316597,
|
| 6604 |
+
"learning_rate": 2.2423367939597385e-05,
|
| 6605 |
+
"loss": 0.0002,
|
| 6606 |
+
"step": 939
|
| 6607 |
+
},
|
| 6608 |
+
{
|
| 6609 |
+
"epoch": 0.6981062012625325,
|
| 6610 |
+
"grad_norm": 0.0014067289885133505,
|
| 6611 |
+
"learning_rate": 2.2322346825913886e-05,
|
| 6612 |
+
"loss": 0.0001,
|
| 6613 |
+
"step": 940
|
| 6614 |
+
},
|
| 6615 |
+
{
|
| 6616 |
+
"epoch": 0.6988488674340884,
|
| 6617 |
+
"grad_norm": 0.002156575210392475,
|
| 6618 |
+
"learning_rate": 2.2221488349019903e-05,
|
| 6619 |
+
"loss": 0.0001,
|
| 6620 |
+
"step": 941
|
| 6621 |
+
},
|
| 6622 |
+
{
|
| 6623 |
+
"epoch": 0.6995915336056443,
|
| 6624 |
+
"grad_norm": 0.0006094371201470494,
|
| 6625 |
+
"learning_rate": 2.2120793101570364e-05,
|
| 6626 |
+
"loss": 0.0001,
|
| 6627 |
+
"step": 942
|
| 6628 |
+
},
|
| 6629 |
+
{
|
| 6630 |
+
"epoch": 0.7003341997772001,
|
| 6631 |
+
"grad_norm": 0.0005834063631482422,
|
| 6632 |
+
"learning_rate": 2.202026167526111e-05,
|
| 6633 |
+
"loss": 0.0001,
|
| 6634 |
+
"step": 943
|
| 6635 |
+
},
|
| 6636 |
+
{
|
| 6637 |
+
"epoch": 0.7010768659487561,
|
| 6638 |
+
"grad_norm": 0.00110190873965621,
|
| 6639 |
+
"learning_rate": 2.1919894660825365e-05,
|
| 6640 |
+
"loss": 0.0001,
|
| 6641 |
+
"step": 944
|
| 6642 |
+
},
|
| 6643 |
+
{
|
| 6644 |
+
"epoch": 0.7018195321203119,
|
| 6645 |
+
"grad_norm": 0.06204232573509216,
|
| 6646 |
+
"learning_rate": 2.1819692648030165e-05,
|
| 6647 |
+
"loss": 0.0018,
|
| 6648 |
+
"step": 945
|
| 6649 |
+
},
|
| 6650 |
+
{
|
| 6651 |
+
"epoch": 0.7025621982918678,
|
| 6652 |
+
"grad_norm": 0.002225680975243449,
|
| 6653 |
+
"learning_rate": 2.171965622567308e-05,
|
| 6654 |
+
"loss": 0.0001,
|
| 6655 |
+
"step": 946
|
| 6656 |
+
},
|
| 6657 |
+
{
|
| 6658 |
+
"epoch": 0.7033048644634237,
|
| 6659 |
+
"grad_norm": 0.005498223472386599,
|
| 6660 |
+
"learning_rate": 2.1619785981578573e-05,
|
| 6661 |
+
"loss": 0.0002,
|
| 6662 |
+
"step": 947
|
| 6663 |
+
},
|
| 6664 |
+
{
|
| 6665 |
+
"epoch": 0.7040475306349796,
|
| 6666 |
+
"grad_norm": 0.2603996992111206,
|
| 6667 |
+
"learning_rate": 2.1520082502594685e-05,
|
| 6668 |
+
"loss": 0.0036,
|
| 6669 |
+
"step": 948
|
| 6670 |
+
},
|
| 6671 |
+
{
|
| 6672 |
+
"epoch": 0.7047901968065354,
|
| 6673 |
+
"grad_norm": 0.0005693411221727729,
|
| 6674 |
+
"learning_rate": 2.1420546374589484e-05,
|
| 6675 |
+
"loss": 0.0001,
|
| 6676 |
+
"step": 949
|
| 6677 |
+
},
|
| 6678 |
+
{
|
| 6679 |
+
"epoch": 0.7055328629780914,
|
| 6680 |
+
"grad_norm": 0.0005677467561326921,
|
| 6681 |
+
"learning_rate": 2.132117818244771e-05,
|
| 6682 |
+
"loss": 0.0001,
|
| 6683 |
+
"step": 950
|
| 6684 |
+
},
|
| 6685 |
+
{
|
| 6686 |
+
"epoch": 0.7062755291496472,
|
| 6687 |
+
"grad_norm": 0.019135845825076103,
|
| 6688 |
+
"learning_rate": 2.1221978510067236e-05,
|
| 6689 |
+
"loss": 0.0006,
|
| 6690 |
+
"step": 951
|
| 6691 |
+
},
|
| 6692 |
+
{
|
| 6693 |
+
"epoch": 0.7070181953212031,
|
| 6694 |
+
"grad_norm": 0.3902997672557831,
|
| 6695 |
+
"learning_rate": 2.1122947940355747e-05,
|
| 6696 |
+
"loss": 0.0052,
|
| 6697 |
+
"step": 952
|
| 6698 |
+
},
|
| 6699 |
+
{
|
| 6700 |
+
"epoch": 0.707760861492759,
|
| 6701 |
+
"grad_norm": 0.0012116495054215193,
|
| 6702 |
+
"learning_rate": 2.1024087055227248e-05,
|
| 6703 |
+
"loss": 0.0001,
|
| 6704 |
+
"step": 953
|
| 6705 |
+
},
|
| 6706 |
+
{
|
| 6707 |
+
"epoch": 0.7085035276643149,
|
| 6708 |
+
"grad_norm": 0.3175068795681,
|
| 6709 |
+
"learning_rate": 2.0925396435598664e-05,
|
| 6710 |
+
"loss": 0.0045,
|
| 6711 |
+
"step": 954
|
| 6712 |
+
},
|
| 6713 |
+
{
|
| 6714 |
+
"epoch": 0.7092461938358707,
|
| 6715 |
+
"grad_norm": 0.015600265935063362,
|
| 6716 |
+
"learning_rate": 2.0826876661386395e-05,
|
| 6717 |
+
"loss": 0.0002,
|
| 6718 |
+
"step": 955
|
| 6719 |
+
},
|
| 6720 |
+
{
|
| 6721 |
+
"epoch": 0.7099888600074267,
|
| 6722 |
+
"grad_norm": 0.0010375302517786622,
|
| 6723 |
+
"learning_rate": 2.0728528311502976e-05,
|
| 6724 |
+
"loss": 0.0001,
|
| 6725 |
+
"step": 956
|
| 6726 |
+
},
|
| 6727 |
+
{
|
| 6728 |
+
"epoch": 0.7107315261789825,
|
| 6729 |
+
"grad_norm": 0.0014974784571677446,
|
| 6730 |
+
"learning_rate": 2.0630351963853587e-05,
|
| 6731 |
+
"loss": 0.0001,
|
| 6732 |
+
"step": 957
|
| 6733 |
+
},
|
| 6734 |
+
{
|
| 6735 |
+
"epoch": 0.7114741923505384,
|
| 6736 |
+
"grad_norm": 0.0019292763900011778,
|
| 6737 |
+
"learning_rate": 2.053234819533276e-05,
|
| 6738 |
+
"loss": 0.0001,
|
| 6739 |
+
"step": 958
|
| 6740 |
+
},
|
| 6741 |
+
{
|
| 6742 |
+
"epoch": 0.7122168585220943,
|
| 6743 |
+
"grad_norm": 0.0008647122303955257,
|
| 6744 |
+
"learning_rate": 2.0434517581820896e-05,
|
| 6745 |
+
"loss": 0.0001,
|
| 6746 |
+
"step": 959
|
| 6747 |
+
},
|
| 6748 |
+
{
|
| 6749 |
+
"epoch": 0.7129595246936502,
|
| 6750 |
+
"grad_norm": 0.0007554054609499872,
|
| 6751 |
+
"learning_rate": 2.0336860698180905e-05,
|
| 6752 |
+
"loss": 0.0001,
|
| 6753 |
+
"step": 960
|
| 6754 |
+
},
|
| 6755 |
+
{
|
| 6756 |
+
"epoch": 0.7137021908652061,
|
| 6757 |
+
"grad_norm": 0.01203538291156292,
|
| 6758 |
+
"learning_rate": 2.0239378118254905e-05,
|
| 6759 |
+
"loss": 0.0004,
|
| 6760 |
+
"step": 961
|
| 6761 |
+
},
|
| 6762 |
+
{
|
| 6763 |
+
"epoch": 0.714444857036762,
|
| 6764 |
+
"grad_norm": 0.005261108744889498,
|
| 6765 |
+
"learning_rate": 2.0142070414860704e-05,
|
| 6766 |
+
"loss": 0.0003,
|
| 6767 |
+
"step": 962
|
| 6768 |
+
},
|
| 6769 |
+
{
|
| 6770 |
+
"epoch": 0.7151875232083179,
|
| 6771 |
+
"grad_norm": 0.006285266485065222,
|
| 6772 |
+
"learning_rate": 2.0044938159788584e-05,
|
| 6773 |
+
"loss": 0.0002,
|
| 6774 |
+
"step": 963
|
| 6775 |
+
},
|
| 6776 |
+
{
|
| 6777 |
+
"epoch": 0.7159301893798737,
|
| 6778 |
+
"grad_norm": 0.0008547162869945168,
|
| 6779 |
+
"learning_rate": 1.9947981923797853e-05,
|
| 6780 |
+
"loss": 0.0001,
|
| 6781 |
+
"step": 964
|
| 6782 |
+
},
|
| 6783 |
+
{
|
| 6784 |
+
"epoch": 0.7166728555514297,
|
| 6785 |
+
"grad_norm": 0.00040477412403561175,
|
| 6786 |
+
"learning_rate": 1.9851202276613523e-05,
|
| 6787 |
+
"loss": 0.0001,
|
| 6788 |
+
"step": 965
|
| 6789 |
+
},
|
| 6790 |
+
{
|
| 6791 |
+
"epoch": 0.7174155217229855,
|
| 6792 |
+
"grad_norm": 0.0028770139906555414,
|
| 6793 |
+
"learning_rate": 1.975459978692291e-05,
|
| 6794 |
+
"loss": 0.0003,
|
| 6795 |
+
"step": 966
|
| 6796 |
+
},
|
| 6797 |
+
{
|
| 6798 |
+
"epoch": 0.7181581878945414,
|
| 6799 |
+
"grad_norm": 0.001716628554277122,
|
| 6800 |
+
"learning_rate": 1.96581750223724e-05,
|
| 6801 |
+
"loss": 0.0002,
|
| 6802 |
+
"step": 967
|
| 6803 |
+
},
|
| 6804 |
+
{
|
| 6805 |
+
"epoch": 0.7189008540660973,
|
| 6806 |
+
"grad_norm": 0.0005640503368340433,
|
| 6807 |
+
"learning_rate": 1.9561928549563968e-05,
|
| 6808 |
+
"loss": 0.0001,
|
| 6809 |
+
"step": 968
|
| 6810 |
+
},
|
| 6811 |
+
{
|
| 6812 |
+
"epoch": 0.7196435202376532,
|
| 6813 |
+
"grad_norm": 0.001024580909870565,
|
| 6814 |
+
"learning_rate": 1.9465860934052006e-05,
|
| 6815 |
+
"loss": 0.0001,
|
| 6816 |
+
"step": 969
|
| 6817 |
+
},
|
| 6818 |
+
{
|
| 6819 |
+
"epoch": 0.720386186409209,
|
| 6820 |
+
"grad_norm": 0.0007752843666821718,
|
| 6821 |
+
"learning_rate": 1.936997274033986e-05,
|
| 6822 |
+
"loss": 0.0001,
|
| 6823 |
+
"step": 970
|
| 6824 |
+
},
|
| 6825 |
+
{
|
| 6826 |
+
"epoch": 0.721128852580765,
|
| 6827 |
+
"grad_norm": 0.0008537242538295686,
|
| 6828 |
+
"learning_rate": 1.927426453187663e-05,
|
| 6829 |
+
"loss": 0.0001,
|
| 6830 |
+
"step": 971
|
| 6831 |
+
},
|
| 6832 |
+
{
|
| 6833 |
+
"epoch": 0.7218715187523208,
|
| 6834 |
+
"grad_norm": 0.0009459343855269253,
|
| 6835 |
+
"learning_rate": 1.9178736871053736e-05,
|
| 6836 |
+
"loss": 0.0001,
|
| 6837 |
+
"step": 972
|
| 6838 |
+
},
|
| 6839 |
+
{
|
| 6840 |
+
"epoch": 0.7226141849238767,
|
| 6841 |
+
"grad_norm": 0.0008696299046278,
|
| 6842 |
+
"learning_rate": 1.9083390319201764e-05,
|
| 6843 |
+
"loss": 0.0001,
|
| 6844 |
+
"step": 973
|
| 6845 |
+
},
|
| 6846 |
+
{
|
| 6847 |
+
"epoch": 0.7233568510954326,
|
| 6848 |
+
"grad_norm": 0.0020470714662224054,
|
| 6849 |
+
"learning_rate": 1.8988225436587002e-05,
|
| 6850 |
+
"loss": 0.0002,
|
| 6851 |
+
"step": 974
|
| 6852 |
+
},
|
| 6853 |
+
{
|
| 6854 |
+
"epoch": 0.7240995172669885,
|
| 6855 |
+
"grad_norm": 0.0032244205940514803,
|
| 6856 |
+
"learning_rate": 1.889324278240829e-05,
|
| 6857 |
+
"loss": 0.0002,
|
| 6858 |
+
"step": 975
|
| 6859 |
+
},
|
| 6860 |
+
{
|
| 6861 |
+
"epoch": 0.7248421834385443,
|
| 6862 |
+
"grad_norm": 0.007482845336198807,
|
| 6863 |
+
"learning_rate": 1.8798442914793663e-05,
|
| 6864 |
+
"loss": 0.0002,
|
| 6865 |
+
"step": 976
|
| 6866 |
+
},
|
| 6867 |
+
{
|
| 6868 |
+
"epoch": 0.7255848496101003,
|
| 6869 |
+
"grad_norm": 0.25979605317115784,
|
| 6870 |
+
"learning_rate": 1.8703826390797048e-05,
|
| 6871 |
+
"loss": 0.0077,
|
| 6872 |
+
"step": 977
|
| 6873 |
+
},
|
| 6874 |
+
{
|
| 6875 |
+
"epoch": 0.7263275157816561,
|
| 6876 |
+
"grad_norm": 0.0036151015665382147,
|
| 6877 |
+
"learning_rate": 1.8609393766395085e-05,
|
| 6878 |
+
"loss": 0.0002,
|
| 6879 |
+
"step": 978
|
| 6880 |
+
},
|
| 6881 |
+
{
|
| 6882 |
+
"epoch": 0.727070181953212,
|
| 6883 |
+
"grad_norm": 1.0644394159317017,
|
| 6884 |
+
"learning_rate": 1.8515145596483723e-05,
|
| 6885 |
+
"loss": 0.0138,
|
| 6886 |
+
"step": 979
|
| 6887 |
+
},
|
| 6888 |
+
{
|
| 6889 |
+
"epoch": 0.7278128481247679,
|
| 6890 |
+
"grad_norm": 0.0007425799267366529,
|
| 6891 |
+
"learning_rate": 1.842108243487513e-05,
|
| 6892 |
+
"loss": 0.0001,
|
| 6893 |
+
"step": 980
|
| 6894 |
+
},
|
| 6895 |
+
{
|
| 6896 |
+
"epoch": 0.7285555142963238,
|
| 6897 |
+
"grad_norm": 0.0005094231455586851,
|
| 6898 |
+
"learning_rate": 1.832720483429426e-05,
|
| 6899 |
+
"loss": 0.0001,
|
| 6900 |
+
"step": 981
|
| 6901 |
+
},
|
| 6902 |
+
{
|
| 6903 |
+
"epoch": 0.7292981804678796,
|
| 6904 |
+
"grad_norm": 0.0007252280483953655,
|
| 6905 |
+
"learning_rate": 1.8233513346375756e-05,
|
| 6906 |
+
"loss": 0.0001,
|
| 6907 |
+
"step": 982
|
| 6908 |
+
},
|
| 6909 |
+
{
|
| 6910 |
+
"epoch": 0.7300408466394356,
|
| 6911 |
+
"grad_norm": 0.0005672296392731369,
|
| 6912 |
+
"learning_rate": 1.814000852166059e-05,
|
| 6913 |
+
"loss": 0.0001,
|
| 6914 |
+
"step": 983
|
| 6915 |
+
},
|
| 6916 |
+
{
|
| 6917 |
+
"epoch": 0.7307835128109914,
|
| 6918 |
+
"grad_norm": 0.9842846989631653,
|
| 6919 |
+
"learning_rate": 1.8046690909592917e-05,
|
| 6920 |
+
"loss": 0.1766,
|
| 6921 |
+
"step": 984
|
| 6922 |
+
},
|
| 6923 |
+
{
|
| 6924 |
+
"epoch": 0.7315261789825473,
|
| 6925 |
+
"grad_norm": 0.000559376843739301,
|
| 6926 |
+
"learning_rate": 1.7953561058516828e-05,
|
| 6927 |
+
"loss": 0.0001,
|
| 6928 |
+
"step": 985
|
| 6929 |
+
},
|
| 6930 |
+
{
|
| 6931 |
+
"epoch": 0.7322688451541032,
|
| 6932 |
+
"grad_norm": 0.0007592244073748589,
|
| 6933 |
+
"learning_rate": 1.7860619515673033e-05,
|
| 6934 |
+
"loss": 0.0001,
|
| 6935 |
+
"step": 986
|
| 6936 |
+
},
|
| 6937 |
+
{
|
| 6938 |
+
"epoch": 0.7330115113256591,
|
| 6939 |
+
"grad_norm": 0.0009743353584781289,
|
| 6940 |
+
"learning_rate": 1.7767866827195805e-05,
|
| 6941 |
+
"loss": 0.0001,
|
| 6942 |
+
"step": 987
|
| 6943 |
+
},
|
| 6944 |
+
{
|
| 6945 |
+
"epoch": 0.733754177497215,
|
| 6946 |
+
"grad_norm": 0.0004740802978631109,
|
| 6947 |
+
"learning_rate": 1.767530353810968e-05,
|
| 6948 |
+
"loss": 0.0001,
|
| 6949 |
+
"step": 988
|
| 6950 |
+
},
|
| 6951 |
+
{
|
| 6952 |
+
"epoch": 0.7344968436687709,
|
| 6953 |
+
"grad_norm": 0.001709680538624525,
|
| 6954 |
+
"learning_rate": 1.7582930192326186e-05,
|
| 6955 |
+
"loss": 0.0001,
|
| 6956 |
+
"step": 989
|
| 6957 |
+
},
|
| 6958 |
+
{
|
| 6959 |
+
"epoch": 0.7352395098403268,
|
| 6960 |
+
"grad_norm": 0.004645473323762417,
|
| 6961 |
+
"learning_rate": 1.7490747332640833e-05,
|
| 6962 |
+
"loss": 0.0002,
|
| 6963 |
+
"step": 990
|
| 6964 |
+
},
|
| 6965 |
+
{
|
| 6966 |
+
"epoch": 0.7359821760118826,
|
| 6967 |
+
"grad_norm": 0.002532870974391699,
|
| 6968 |
+
"learning_rate": 1.7398755500729734e-05,
|
| 6969 |
+
"loss": 0.0001,
|
| 6970 |
+
"step": 991
|
| 6971 |
+
},
|
| 6972 |
+
{
|
| 6973 |
+
"epoch": 0.7367248421834386,
|
| 6974 |
+
"grad_norm": 0.0024906517937779427,
|
| 6975 |
+
"learning_rate": 1.730695523714652e-05,
|
| 6976 |
+
"loss": 0.0001,
|
| 6977 |
+
"step": 992
|
| 6978 |
+
},
|
| 6979 |
+
{
|
| 6980 |
+
"epoch": 0.7374675083549944,
|
| 6981 |
+
"grad_norm": 0.006111454218626022,
|
| 6982 |
+
"learning_rate": 1.7215347081319188e-05,
|
| 6983 |
+
"loss": 0.0003,
|
| 6984 |
+
"step": 993
|
| 6985 |
+
},
|
| 6986 |
+
{
|
| 6987 |
+
"epoch": 0.7382101745265504,
|
| 6988 |
+
"grad_norm": 0.004187372513115406,
|
| 6989 |
+
"learning_rate": 1.7123931571546827e-05,
|
| 6990 |
+
"loss": 0.0002,
|
| 6991 |
+
"step": 994
|
| 6992 |
+
},
|
| 6993 |
+
{
|
| 6994 |
+
"epoch": 0.7389528406981062,
|
| 6995 |
+
"grad_norm": 0.006407550070434809,
|
| 6996 |
+
"learning_rate": 1.703270924499656e-05,
|
| 6997 |
+
"loss": 0.0004,
|
| 6998 |
+
"step": 995
|
| 6999 |
+
},
|
| 7000 |
+
{
|
| 7001 |
+
"epoch": 0.7396955068696621,
|
| 7002 |
+
"grad_norm": 0.01187540590763092,
|
| 7003 |
+
"learning_rate": 1.6941680637700334e-05,
|
| 7004 |
+
"loss": 0.0005,
|
| 7005 |
+
"step": 996
|
| 7006 |
+
},
|
| 7007 |
+
{
|
| 7008 |
+
"epoch": 0.740438173041218,
|
| 7009 |
+
"grad_norm": 0.008408035151660442,
|
| 7010 |
+
"learning_rate": 1.68508462845518e-05,
|
| 7011 |
+
"loss": 0.0003,
|
| 7012 |
+
"step": 997
|
| 7013 |
+
},
|
| 7014 |
+
{
|
| 7015 |
+
"epoch": 0.7411808392127739,
|
| 7016 |
+
"grad_norm": 0.00872902199625969,
|
| 7017 |
+
"learning_rate": 1.6760206719303105e-05,
|
| 7018 |
+
"loss": 0.0003,
|
| 7019 |
+
"step": 998
|
| 7020 |
+
},
|
| 7021 |
+
{
|
| 7022 |
+
"epoch": 0.7419235053843297,
|
| 7023 |
+
"grad_norm": 0.013007326982915401,
|
| 7024 |
+
"learning_rate": 1.6669762474561857e-05,
|
| 7025 |
+
"loss": 0.0005,
|
| 7026 |
+
"step": 999
|
| 7027 |
+
},
|
| 7028 |
+
{
|
| 7029 |
+
"epoch": 0.7426661715558857,
|
| 7030 |
+
"grad_norm": 0.044673044234514236,
|
| 7031 |
+
"learning_rate": 1.657951408178788e-05,
|
| 7032 |
+
"loss": 0.0013,
|
| 7033 |
+
"step": 1000
|
| 7034 |
+
},
|
| 7035 |
+
{
|
| 7036 |
+
"epoch": 0.7434088377274415,
|
| 7037 |
+
"grad_norm": 0.017271386459469795,
|
| 7038 |
+
"learning_rate": 1.6489462071290214e-05,
|
| 7039 |
+
"loss": 0.0006,
|
| 7040 |
+
"step": 1001
|
| 7041 |
+
},
|
| 7042 |
+
{
|
| 7043 |
+
"epoch": 0.7441515038989974,
|
| 7044 |
+
"grad_norm": 0.03663730248808861,
|
| 7045 |
+
"learning_rate": 1.639960697222388e-05,
|
| 7046 |
+
"loss": 0.001,
|
| 7047 |
+
"step": 1002
|
| 7048 |
+
},
|
| 7049 |
+
{
|
| 7050 |
+
"epoch": 0.7448941700705533,
|
| 7051 |
+
"grad_norm": 0.01725134626030922,
|
| 7052 |
+
"learning_rate": 1.630994931258688e-05,
|
| 7053 |
+
"loss": 0.0005,
|
| 7054 |
+
"step": 1003
|
| 7055 |
+
},
|
| 7056 |
+
{
|
| 7057 |
+
"epoch": 0.7456368362421092,
|
| 7058 |
+
"grad_norm": 0.0050296480767428875,
|
| 7059 |
+
"learning_rate": 1.622048961921699e-05,
|
| 7060 |
+
"loss": 0.0003,
|
| 7061 |
+
"step": 1004
|
| 7062 |
+
},
|
| 7063 |
+
{
|
| 7064 |
+
"epoch": 0.746379502413665,
|
| 7065 |
+
"grad_norm": 0.013799471780657768,
|
| 7066 |
+
"learning_rate": 1.6131228417788765e-05,
|
| 7067 |
+
"loss": 0.0004,
|
| 7068 |
+
"step": 1005
|
| 7069 |
+
},
|
| 7070 |
+
{
|
| 7071 |
+
"epoch": 0.747122168585221,
|
| 7072 |
+
"grad_norm": 0.005248996429145336,
|
| 7073 |
+
"learning_rate": 1.6042166232810347e-05,
|
| 7074 |
+
"loss": 0.0003,
|
| 7075 |
+
"step": 1006
|
| 7076 |
+
},
|
| 7077 |
+
{
|
| 7078 |
+
"epoch": 0.7478648347567768,
|
| 7079 |
+
"grad_norm": 0.0058412919752299786,
|
| 7080 |
+
"learning_rate": 1.595330358762047e-05,
|
| 7081 |
+
"loss": 0.0005,
|
| 7082 |
+
"step": 1007
|
| 7083 |
+
},
|
| 7084 |
+
{
|
| 7085 |
+
"epoch": 0.7486075009283327,
|
| 7086 |
+
"grad_norm": 0.006610256619751453,
|
| 7087 |
+
"learning_rate": 1.5864641004385384e-05,
|
| 7088 |
+
"loss": 0.0003,
|
| 7089 |
+
"step": 1008
|
| 7090 |
+
},
|
| 7091 |
+
{
|
| 7092 |
+
"epoch": 0.7493501670998886,
|
| 7093 |
+
"grad_norm": 0.004714768845587969,
|
| 7094 |
+
"learning_rate": 1.577617900409566e-05,
|
| 7095 |
+
"loss": 0.0002,
|
| 7096 |
+
"step": 1009
|
| 7097 |
+
},
|
| 7098 |
+
{
|
| 7099 |
+
"epoch": 0.7500928332714445,
|
| 7100 |
+
"grad_norm": 0.005967407487332821,
|
| 7101 |
+
"learning_rate": 1.5687918106563326e-05,
|
| 7102 |
+
"loss": 0.0003,
|
| 7103 |
+
"step": 1010
|
| 7104 |
+
},
|
| 7105 |
+
{
|
| 7106 |
+
"epoch": 0.7508354994430003,
|
| 7107 |
+
"grad_norm": 0.004380329977720976,
|
| 7108 |
+
"learning_rate": 1.5599858830418628e-05,
|
| 7109 |
+
"loss": 0.0003,
|
| 7110 |
+
"step": 1011
|
| 7111 |
+
},
|
| 7112 |
+
{
|
| 7113 |
+
"epoch": 0.7508354994430003,
|
| 7114 |
+
"eval_loss": 9.535554272588342e-05,
|
| 7115 |
+
"eval_runtime": 187.7576,
|
| 7116 |
+
"eval_samples_per_second": 6.04,
|
| 7117 |
+
"eval_steps_per_second": 3.02,
|
| 7118 |
+
"step": 1011
|
| 7119 |
}
|
| 7120 |
],
|
| 7121 |
"logging_steps": 1,
|
|
|
|
| 7135 |
"attributes": {}
|
| 7136 |
}
|
| 7137 |
},
|
| 7138 |
+
"total_flos": 1.321305510883885e+18,
|
| 7139 |
"train_batch_size": 1,
|
| 7140 |
"trial_name": null,
|
| 7141 |
"trial_params": null
|