Training in progress, step 1346, 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 sha256:2e11b51dcd9f84eb65ffeaa57bcb3520f67defcc54ae73d71fdb62f8baecbf3d
|
| 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:e3830f2d2282c8027eb8004ed649397a6a316e6d178d4d93c50015ffd9ca3b30
|
| 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 sha256:9156fcf9a0a22eda1751efc419df7fe24ed5bdc3a98306d3676f8d312987cdf9
|
| 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:de9fde295b905b7e74df6039ca6d815366f74d4a8f805150f15fb508d5f722b0
|
| 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 sha256:7fa527cc920b9cc0cdeb63b9a663aaf204f9bd87acb41a3f6cfff6215c103244
|
| 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,
|
|
@@ -7116,6 +7116,2351 @@
|
|
| 7116 |
"eval_samples_per_second": 6.04,
|
| 7117 |
"eval_steps_per_second": 3.02,
|
| 7118 |
"step": 1011
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7119 |
}
|
| 7120 |
],
|
| 7121 |
"logging_steps": 1,
|
|
@@ -7130,12 +9475,12 @@
|
|
| 7130 |
"should_evaluate": false,
|
| 7131 |
"should_log": false,
|
| 7132 |
"should_save": true,
|
| 7133 |
-
"should_training_stop":
|
| 7134 |
},
|
| 7135 |
"attributes": {}
|
| 7136 |
}
|
| 7137 |
},
|
| 7138 |
-
"total_flos": 1.
|
| 7139 |
"train_batch_size": 1,
|
| 7140 |
"trial_name": null,
|
| 7141 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.999628666914222,
|
| 5 |
"eval_steps": 337,
|
| 6 |
+
"global_step": 1346,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 7116 |
"eval_samples_per_second": 6.04,
|
| 7117 |
"eval_steps_per_second": 3.02,
|
| 7118 |
"step": 1011
|
| 7119 |
+
},
|
| 7120 |
+
{
|
| 7121 |
+
"epoch": 0.7515781656145563,
|
| 7122 |
+
"grad_norm": 0.010244094766676426,
|
| 7123 |
+
"learning_rate": 1.5512001693107124e-05,
|
| 7124 |
+
"loss": 0.0003,
|
| 7125 |
+
"step": 1012
|
| 7126 |
+
},
|
| 7127 |
+
{
|
| 7128 |
+
"epoch": 0.7523208317861121,
|
| 7129 |
+
"grad_norm": 0.003399641951546073,
|
| 7130 |
+
"learning_rate": 1.5424347210886538e-05,
|
| 7131 |
+
"loss": 0.0002,
|
| 7132 |
+
"step": 1013
|
| 7133 |
+
},
|
| 7134 |
+
{
|
| 7135 |
+
"epoch": 0.753063497957668,
|
| 7136 |
+
"grad_norm": 0.004665118642151356,
|
| 7137 |
+
"learning_rate": 1.53368958988238e-05,
|
| 7138 |
+
"loss": 0.0003,
|
| 7139 |
+
"step": 1014
|
| 7140 |
+
},
|
| 7141 |
+
{
|
| 7142 |
+
"epoch": 0.753806164129224,
|
| 7143 |
+
"grad_norm": 0.007897808216512203,
|
| 7144 |
+
"learning_rate": 1.5249648270791966e-05,
|
| 7145 |
+
"loss": 0.0004,
|
| 7146 |
+
"step": 1015
|
| 7147 |
+
},
|
| 7148 |
+
{
|
| 7149 |
+
"epoch": 0.7545488303007798,
|
| 7150 |
+
"grad_norm": 0.005702964495867491,
|
| 7151 |
+
"learning_rate": 1.5162604839467265e-05,
|
| 7152 |
+
"loss": 0.0003,
|
| 7153 |
+
"step": 1016
|
| 7154 |
+
},
|
| 7155 |
+
{
|
| 7156 |
+
"epoch": 0.7552914964723357,
|
| 7157 |
+
"grad_norm": 0.1535305380821228,
|
| 7158 |
+
"learning_rate": 1.5075766116325985e-05,
|
| 7159 |
+
"loss": 0.0013,
|
| 7160 |
+
"step": 1017
|
| 7161 |
+
},
|
| 7162 |
+
{
|
| 7163 |
+
"epoch": 0.7560341626438916,
|
| 7164 |
+
"grad_norm": 0.0065938313491642475,
|
| 7165 |
+
"learning_rate": 1.4989132611641576e-05,
|
| 7166 |
+
"loss": 0.0003,
|
| 7167 |
+
"step": 1018
|
| 7168 |
+
},
|
| 7169 |
+
{
|
| 7170 |
+
"epoch": 0.7567768288154475,
|
| 7171 |
+
"grad_norm": 0.011125356890261173,
|
| 7172 |
+
"learning_rate": 1.4902704834481584e-05,
|
| 7173 |
+
"loss": 0.0004,
|
| 7174 |
+
"step": 1019
|
| 7175 |
+
},
|
| 7176 |
+
{
|
| 7177 |
+
"epoch": 0.7575194949870033,
|
| 7178 |
+
"grad_norm": 0.010162554681301117,
|
| 7179 |
+
"learning_rate": 1.4816483292704708e-05,
|
| 7180 |
+
"loss": 0.0004,
|
| 7181 |
+
"step": 1020
|
| 7182 |
+
},
|
| 7183 |
+
{
|
| 7184 |
+
"epoch": 0.7582621611585593,
|
| 7185 |
+
"grad_norm": 0.012406081892549992,
|
| 7186 |
+
"learning_rate": 1.4730468492957722e-05,
|
| 7187 |
+
"loss": 0.0004,
|
| 7188 |
+
"step": 1021
|
| 7189 |
+
},
|
| 7190 |
+
{
|
| 7191 |
+
"epoch": 0.7590048273301151,
|
| 7192 |
+
"grad_norm": 0.010129008442163467,
|
| 7193 |
+
"learning_rate": 1.4644660940672627e-05,
|
| 7194 |
+
"loss": 0.0004,
|
| 7195 |
+
"step": 1022
|
| 7196 |
+
},
|
| 7197 |
+
{
|
| 7198 |
+
"epoch": 0.759747493501671,
|
| 7199 |
+
"grad_norm": 0.018571769818663597,
|
| 7200 |
+
"learning_rate": 1.4559061140063578e-05,
|
| 7201 |
+
"loss": 0.0006,
|
| 7202 |
+
"step": 1023
|
| 7203 |
+
},
|
| 7204 |
+
{
|
| 7205 |
+
"epoch": 0.7604901596732269,
|
| 7206 |
+
"grad_norm": 0.01117369718849659,
|
| 7207 |
+
"learning_rate": 1.4473669594123956e-05,
|
| 7208 |
+
"loss": 0.0004,
|
| 7209 |
+
"step": 1024
|
| 7210 |
+
},
|
| 7211 |
+
{
|
| 7212 |
+
"epoch": 0.7612328258447828,
|
| 7213 |
+
"grad_norm": 0.011828822083771229,
|
| 7214 |
+
"learning_rate": 1.4388486804623464e-05,
|
| 7215 |
+
"loss": 0.0005,
|
| 7216 |
+
"step": 1025
|
| 7217 |
+
},
|
| 7218 |
+
{
|
| 7219 |
+
"epoch": 0.7619754920163386,
|
| 7220 |
+
"grad_norm": 0.006083609536290169,
|
| 7221 |
+
"learning_rate": 1.4303513272105057e-05,
|
| 7222 |
+
"loss": 0.0003,
|
| 7223 |
+
"step": 1026
|
| 7224 |
+
},
|
| 7225 |
+
{
|
| 7226 |
+
"epoch": 0.7627181581878946,
|
| 7227 |
+
"grad_norm": 0.0038173352368175983,
|
| 7228 |
+
"learning_rate": 1.4218749495882133e-05,
|
| 7229 |
+
"loss": 0.0002,
|
| 7230 |
+
"step": 1027
|
| 7231 |
+
},
|
| 7232 |
+
{
|
| 7233 |
+
"epoch": 0.7634608243594504,
|
| 7234 |
+
"grad_norm": 0.003378799417987466,
|
| 7235 |
+
"learning_rate": 1.4134195974035525e-05,
|
| 7236 |
+
"loss": 0.0002,
|
| 7237 |
+
"step": 1028
|
| 7238 |
+
},
|
| 7239 |
+
{
|
| 7240 |
+
"epoch": 0.7642034905310063,
|
| 7241 |
+
"grad_norm": 0.0023517953231930733,
|
| 7242 |
+
"learning_rate": 1.4049853203410596e-05,
|
| 7243 |
+
"loss": 0.0002,
|
| 7244 |
+
"step": 1029
|
| 7245 |
+
},
|
| 7246 |
+
{
|
| 7247 |
+
"epoch": 0.7649461567025622,
|
| 7248 |
+
"grad_norm": 0.005170909222215414,
|
| 7249 |
+
"learning_rate": 1.396572167961427e-05,
|
| 7250 |
+
"loss": 0.0002,
|
| 7251 |
+
"step": 1030
|
| 7252 |
+
},
|
| 7253 |
+
{
|
| 7254 |
+
"epoch": 0.7656888228741181,
|
| 7255 |
+
"grad_norm": 0.003187798196449876,
|
| 7256 |
+
"learning_rate": 1.3881801897012225e-05,
|
| 7257 |
+
"loss": 0.0002,
|
| 7258 |
+
"step": 1031
|
| 7259 |
+
},
|
| 7260 |
+
{
|
| 7261 |
+
"epoch": 0.7664314890456739,
|
| 7262 |
+
"grad_norm": 0.0076558575965464115,
|
| 7263 |
+
"learning_rate": 1.3798094348725871e-05,
|
| 7264 |
+
"loss": 0.0002,
|
| 7265 |
+
"step": 1032
|
| 7266 |
+
},
|
| 7267 |
+
{
|
| 7268 |
+
"epoch": 0.7671741552172299,
|
| 7269 |
+
"grad_norm": 0.002008821815252304,
|
| 7270 |
+
"learning_rate": 1.3714599526629551e-05,
|
| 7271 |
+
"loss": 0.0002,
|
| 7272 |
+
"step": 1033
|
| 7273 |
+
},
|
| 7274 |
+
{
|
| 7275 |
+
"epoch": 0.7679168213887857,
|
| 7276 |
+
"grad_norm": 0.011446031741797924,
|
| 7277 |
+
"learning_rate": 1.3631317921347563e-05,
|
| 7278 |
+
"loss": 0.0003,
|
| 7279 |
+
"step": 1034
|
| 7280 |
+
},
|
| 7281 |
+
{
|
| 7282 |
+
"epoch": 0.7686594875603416,
|
| 7283 |
+
"grad_norm": 0.0025135111063718796,
|
| 7284 |
+
"learning_rate": 1.3548250022251375e-05,
|
| 7285 |
+
"loss": 0.0002,
|
| 7286 |
+
"step": 1035
|
| 7287 |
+
},
|
| 7288 |
+
{
|
| 7289 |
+
"epoch": 0.7694021537318975,
|
| 7290 |
+
"grad_norm": 0.0011204791953787208,
|
| 7291 |
+
"learning_rate": 1.346539631745663e-05,
|
| 7292 |
+
"loss": 0.0001,
|
| 7293 |
+
"step": 1036
|
| 7294 |
+
},
|
| 7295 |
+
{
|
| 7296 |
+
"epoch": 0.7701448199034534,
|
| 7297 |
+
"grad_norm": 0.002433129120618105,
|
| 7298 |
+
"learning_rate": 1.3382757293820409e-05,
|
| 7299 |
+
"loss": 0.0002,
|
| 7300 |
+
"step": 1037
|
| 7301 |
+
},
|
| 7302 |
+
{
|
| 7303 |
+
"epoch": 0.7708874860750092,
|
| 7304 |
+
"grad_norm": 0.004581818822771311,
|
| 7305 |
+
"learning_rate": 1.330033343693824e-05,
|
| 7306 |
+
"loss": 0.0003,
|
| 7307 |
+
"step": 1038
|
| 7308 |
+
},
|
| 7309 |
+
{
|
| 7310 |
+
"epoch": 0.7716301522465652,
|
| 7311 |
+
"grad_norm": 0.001932491664774716,
|
| 7312 |
+
"learning_rate": 1.3218125231141354e-05,
|
| 7313 |
+
"loss": 0.0002,
|
| 7314 |
+
"step": 1039
|
| 7315 |
+
},
|
| 7316 |
+
{
|
| 7317 |
+
"epoch": 0.7723728184181211,
|
| 7318 |
+
"grad_norm": 0.001757967984303832,
|
| 7319 |
+
"learning_rate": 1.3136133159493802e-05,
|
| 7320 |
+
"loss": 0.0002,
|
| 7321 |
+
"step": 1040
|
| 7322 |
+
},
|
| 7323 |
+
{
|
| 7324 |
+
"epoch": 0.7731154845896769,
|
| 7325 |
+
"grad_norm": 0.0024122344329953194,
|
| 7326 |
+
"learning_rate": 1.3054357703789554e-05,
|
| 7327 |
+
"loss": 0.0001,
|
| 7328 |
+
"step": 1041
|
| 7329 |
+
},
|
| 7330 |
+
{
|
| 7331 |
+
"epoch": 0.7738581507612329,
|
| 7332 |
+
"grad_norm": 0.003989276476204395,
|
| 7333 |
+
"learning_rate": 1.297279934454978e-05,
|
| 7334 |
+
"loss": 0.0003,
|
| 7335 |
+
"step": 1042
|
| 7336 |
+
},
|
| 7337 |
+
{
|
| 7338 |
+
"epoch": 0.7746008169327887,
|
| 7339 |
+
"grad_norm": 0.10799296200275421,
|
| 7340 |
+
"learning_rate": 1.2891458561019915e-05,
|
| 7341 |
+
"loss": 0.0017,
|
| 7342 |
+
"step": 1043
|
| 7343 |
+
},
|
| 7344 |
+
{
|
| 7345 |
+
"epoch": 0.7753434831043446,
|
| 7346 |
+
"grad_norm": 0.005773243959993124,
|
| 7347 |
+
"learning_rate": 1.2810335831166954e-05,
|
| 7348 |
+
"loss": 0.0002,
|
| 7349 |
+
"step": 1044
|
| 7350 |
+
},
|
| 7351 |
+
{
|
| 7352 |
+
"epoch": 0.7760861492759005,
|
| 7353 |
+
"grad_norm": 0.003262764774262905,
|
| 7354 |
+
"learning_rate": 1.2729431631676525e-05,
|
| 7355 |
+
"loss": 0.0002,
|
| 7356 |
+
"step": 1045
|
| 7357 |
+
},
|
| 7358 |
+
{
|
| 7359 |
+
"epoch": 0.7768288154474564,
|
| 7360 |
+
"grad_norm": 0.0027453494258224964,
|
| 7361 |
+
"learning_rate": 1.264874643795021e-05,
|
| 7362 |
+
"loss": 0.0002,
|
| 7363 |
+
"step": 1046
|
| 7364 |
+
},
|
| 7365 |
+
{
|
| 7366 |
+
"epoch": 0.7775714816190122,
|
| 7367 |
+
"grad_norm": 0.030763821676373482,
|
| 7368 |
+
"learning_rate": 1.2568280724102632e-05,
|
| 7369 |
+
"loss": 0.001,
|
| 7370 |
+
"step": 1047
|
| 7371 |
+
},
|
| 7372 |
+
{
|
| 7373 |
+
"epoch": 0.7783141477905682,
|
| 7374 |
+
"grad_norm": 0.004409801680594683,
|
| 7375 |
+
"learning_rate": 1.2488034962958789e-05,
|
| 7376 |
+
"loss": 0.0002,
|
| 7377 |
+
"step": 1048
|
| 7378 |
+
},
|
| 7379 |
+
{
|
| 7380 |
+
"epoch": 0.779056813962124,
|
| 7381 |
+
"grad_norm": 0.004938923753798008,
|
| 7382 |
+
"learning_rate": 1.2408009626051137e-05,
|
| 7383 |
+
"loss": 0.0003,
|
| 7384 |
+
"step": 1049
|
| 7385 |
+
},
|
| 7386 |
+
{
|
| 7387 |
+
"epoch": 0.77979948013368,
|
| 7388 |
+
"grad_norm": 0.007012166548520327,
|
| 7389 |
+
"learning_rate": 1.2328205183616965e-05,
|
| 7390 |
+
"loss": 0.0003,
|
| 7391 |
+
"step": 1050
|
| 7392 |
+
},
|
| 7393 |
+
{
|
| 7394 |
+
"epoch": 0.7805421463052358,
|
| 7395 |
+
"grad_norm": 0.004935544449836016,
|
| 7396 |
+
"learning_rate": 1.2248622104595532e-05,
|
| 7397 |
+
"loss": 0.0002,
|
| 7398 |
+
"step": 1051
|
| 7399 |
+
},
|
| 7400 |
+
{
|
| 7401 |
+
"epoch": 0.7812848124767917,
|
| 7402 |
+
"grad_norm": 0.002522197552025318,
|
| 7403 |
+
"learning_rate": 1.216926085662536e-05,
|
| 7404 |
+
"loss": 0.0002,
|
| 7405 |
+
"step": 1052
|
| 7406 |
+
},
|
| 7407 |
+
{
|
| 7408 |
+
"epoch": 0.7820274786483475,
|
| 7409 |
+
"grad_norm": 0.001708640600554645,
|
| 7410 |
+
"learning_rate": 1.209012190604143e-05,
|
| 7411 |
+
"loss": 0.0002,
|
| 7412 |
+
"step": 1053
|
| 7413 |
+
},
|
| 7414 |
+
{
|
| 7415 |
+
"epoch": 0.7827701448199035,
|
| 7416 |
+
"grad_norm": 0.001268695225007832,
|
| 7417 |
+
"learning_rate": 1.2011205717872537e-05,
|
| 7418 |
+
"loss": 0.0001,
|
| 7419 |
+
"step": 1054
|
| 7420 |
+
},
|
| 7421 |
+
{
|
| 7422 |
+
"epoch": 0.7835128109914593,
|
| 7423 |
+
"grad_norm": 0.0013489356497302651,
|
| 7424 |
+
"learning_rate": 1.1932512755838449e-05,
|
| 7425 |
+
"loss": 0.0001,
|
| 7426 |
+
"step": 1055
|
| 7427 |
+
},
|
| 7428 |
+
{
|
| 7429 |
+
"epoch": 0.7842554771630152,
|
| 7430 |
+
"grad_norm": 0.002102524507790804,
|
| 7431 |
+
"learning_rate": 1.1854043482347248e-05,
|
| 7432 |
+
"loss": 0.0001,
|
| 7433 |
+
"step": 1056
|
| 7434 |
+
},
|
| 7435 |
+
{
|
| 7436 |
+
"epoch": 0.7849981433345711,
|
| 7437 |
+
"grad_norm": 0.0030939369462430477,
|
| 7438 |
+
"learning_rate": 1.1775798358492629e-05,
|
| 7439 |
+
"loss": 0.0002,
|
| 7440 |
+
"step": 1057
|
| 7441 |
+
},
|
| 7442 |
+
{
|
| 7443 |
+
"epoch": 0.785740809506127,
|
| 7444 |
+
"grad_norm": 0.001534225302748382,
|
| 7445 |
+
"learning_rate": 1.1697777844051105e-05,
|
| 7446 |
+
"loss": 0.0001,
|
| 7447 |
+
"step": 1058
|
| 7448 |
+
},
|
| 7449 |
+
{
|
| 7450 |
+
"epoch": 0.7864834756776828,
|
| 7451 |
+
"grad_norm": 0.008333349600434303,
|
| 7452 |
+
"learning_rate": 1.1619982397479412e-05,
|
| 7453 |
+
"loss": 0.0002,
|
| 7454 |
+
"step": 1059
|
| 7455 |
+
},
|
| 7456 |
+
{
|
| 7457 |
+
"epoch": 0.7872261418492388,
|
| 7458 |
+
"grad_norm": 0.2474621683359146,
|
| 7459 |
+
"learning_rate": 1.1542412475911758e-05,
|
| 7460 |
+
"loss": 0.0036,
|
| 7461 |
+
"step": 1060
|
| 7462 |
+
},
|
| 7463 |
+
{
|
| 7464 |
+
"epoch": 0.7879688080207946,
|
| 7465 |
+
"grad_norm": 0.0020573630463331938,
|
| 7466 |
+
"learning_rate": 1.1465068535157097e-05,
|
| 7467 |
+
"loss": 0.0002,
|
| 7468 |
+
"step": 1061
|
| 7469 |
+
},
|
| 7470 |
+
{
|
| 7471 |
+
"epoch": 0.7887114741923505,
|
| 7472 |
+
"grad_norm": 0.009289558976888657,
|
| 7473 |
+
"learning_rate": 1.1387951029696542e-05,
|
| 7474 |
+
"loss": 0.0003,
|
| 7475 |
+
"step": 1062
|
| 7476 |
+
},
|
| 7477 |
+
{
|
| 7478 |
+
"epoch": 0.7894541403639064,
|
| 7479 |
+
"grad_norm": 0.0011494681239128113,
|
| 7480 |
+
"learning_rate": 1.1311060412680647e-05,
|
| 7481 |
+
"loss": 0.0001,
|
| 7482 |
+
"step": 1063
|
| 7483 |
+
},
|
| 7484 |
+
{
|
| 7485 |
+
"epoch": 0.7901968065354623,
|
| 7486 |
+
"grad_norm": 0.002744733588770032,
|
| 7487 |
+
"learning_rate": 1.1234397135926705e-05,
|
| 7488 |
+
"loss": 0.0002,
|
| 7489 |
+
"step": 1064
|
| 7490 |
+
},
|
| 7491 |
+
{
|
| 7492 |
+
"epoch": 0.7909394727070181,
|
| 7493 |
+
"grad_norm": 0.0026154378429055214,
|
| 7494 |
+
"learning_rate": 1.1157961649916194e-05,
|
| 7495 |
+
"loss": 0.0001,
|
| 7496 |
+
"step": 1065
|
| 7497 |
+
},
|
| 7498 |
+
{
|
| 7499 |
+
"epoch": 0.7916821388785741,
|
| 7500 |
+
"grad_norm": 0.002585667185485363,
|
| 7501 |
+
"learning_rate": 1.1081754403791999e-05,
|
| 7502 |
+
"loss": 0.0002,
|
| 7503 |
+
"step": 1066
|
| 7504 |
+
},
|
| 7505 |
+
{
|
| 7506 |
+
"epoch": 0.79242480505013,
|
| 7507 |
+
"grad_norm": 0.0019874395802617073,
|
| 7508 |
+
"learning_rate": 1.100577584535592e-05,
|
| 7509 |
+
"loss": 0.0002,
|
| 7510 |
+
"step": 1067
|
| 7511 |
+
},
|
| 7512 |
+
{
|
| 7513 |
+
"epoch": 0.7931674712216858,
|
| 7514 |
+
"grad_norm": 0.0020060345996171236,
|
| 7515 |
+
"learning_rate": 1.0930026421065887e-05,
|
| 7516 |
+
"loss": 0.0002,
|
| 7517 |
+
"step": 1068
|
| 7518 |
+
},
|
| 7519 |
+
{
|
| 7520 |
+
"epoch": 0.7939101373932418,
|
| 7521 |
+
"grad_norm": 0.0012965823989361525,
|
| 7522 |
+
"learning_rate": 1.0854506576033491e-05,
|
| 7523 |
+
"loss": 0.0001,
|
| 7524 |
+
"step": 1069
|
| 7525 |
+
},
|
| 7526 |
+
{
|
| 7527 |
+
"epoch": 0.7946528035647976,
|
| 7528 |
+
"grad_norm": 0.0032621456775814295,
|
| 7529 |
+
"learning_rate": 1.0779216754021215e-05,
|
| 7530 |
+
"loss": 0.0002,
|
| 7531 |
+
"step": 1070
|
| 7532 |
+
},
|
| 7533 |
+
{
|
| 7534 |
+
"epoch": 0.7953954697363536,
|
| 7535 |
+
"grad_norm": 0.002112373011186719,
|
| 7536 |
+
"learning_rate": 1.0704157397439962e-05,
|
| 7537 |
+
"loss": 0.0001,
|
| 7538 |
+
"step": 1071
|
| 7539 |
+
},
|
| 7540 |
+
{
|
| 7541 |
+
"epoch": 0.7961381359079094,
|
| 7542 |
+
"grad_norm": 0.00291498564183712,
|
| 7543 |
+
"learning_rate": 1.062932894734639e-05,
|
| 7544 |
+
"loss": 0.0002,
|
| 7545 |
+
"step": 1072
|
| 7546 |
+
},
|
| 7547 |
+
{
|
| 7548 |
+
"epoch": 0.7968808020794653,
|
| 7549 |
+
"grad_norm": 0.005537567660212517,
|
| 7550 |
+
"learning_rate": 1.0554731843440275e-05,
|
| 7551 |
+
"loss": 0.0002,
|
| 7552 |
+
"step": 1073
|
| 7553 |
+
},
|
| 7554 |
+
{
|
| 7555 |
+
"epoch": 0.7976234682510212,
|
| 7556 |
+
"grad_norm": 0.002763421507552266,
|
| 7557 |
+
"learning_rate": 1.0480366524062042e-05,
|
| 7558 |
+
"loss": 0.0002,
|
| 7559 |
+
"step": 1074
|
| 7560 |
+
},
|
| 7561 |
+
{
|
| 7562 |
+
"epoch": 0.7983661344225771,
|
| 7563 |
+
"grad_norm": 0.007374095730483532,
|
| 7564 |
+
"learning_rate": 1.0406233426190076e-05,
|
| 7565 |
+
"loss": 0.0003,
|
| 7566 |
+
"step": 1075
|
| 7567 |
+
},
|
| 7568 |
+
{
|
| 7569 |
+
"epoch": 0.7991088005941329,
|
| 7570 |
+
"grad_norm": 0.009125984273850918,
|
| 7571 |
+
"learning_rate": 1.0332332985438248e-05,
|
| 7572 |
+
"loss": 0.0002,
|
| 7573 |
+
"step": 1076
|
| 7574 |
+
},
|
| 7575 |
+
{
|
| 7576 |
+
"epoch": 0.7998514667656889,
|
| 7577 |
+
"grad_norm": 0.004513624124228954,
|
| 7578 |
+
"learning_rate": 1.0258665636053272e-05,
|
| 7579 |
+
"loss": 0.0003,
|
| 7580 |
+
"step": 1077
|
| 7581 |
+
},
|
| 7582 |
+
{
|
| 7583 |
+
"epoch": 0.8005941329372447,
|
| 7584 |
+
"grad_norm": 0.002031454350799322,
|
| 7585 |
+
"learning_rate": 1.0185231810912222e-05,
|
| 7586 |
+
"loss": 0.0002,
|
| 7587 |
+
"step": 1078
|
| 7588 |
+
},
|
| 7589 |
+
{
|
| 7590 |
+
"epoch": 0.8013367991088006,
|
| 7591 |
+
"grad_norm": 0.0018436467507854104,
|
| 7592 |
+
"learning_rate": 1.0112031941519934e-05,
|
| 7593 |
+
"loss": 0.0002,
|
| 7594 |
+
"step": 1079
|
| 7595 |
+
},
|
| 7596 |
+
{
|
| 7597 |
+
"epoch": 0.8020794652803565,
|
| 7598 |
+
"grad_norm": 0.02794628217816353,
|
| 7599 |
+
"learning_rate": 1.003906645800653e-05,
|
| 7600 |
+
"loss": 0.0003,
|
| 7601 |
+
"step": 1080
|
| 7602 |
+
},
|
| 7603 |
+
{
|
| 7604 |
+
"epoch": 0.8028221314519124,
|
| 7605 |
+
"grad_norm": 0.0030993130058050156,
|
| 7606 |
+
"learning_rate": 9.966335789124809e-06,
|
| 7607 |
+
"loss": 0.0002,
|
| 7608 |
+
"step": 1081
|
| 7609 |
+
},
|
| 7610 |
+
{
|
| 7611 |
+
"epoch": 0.8035647976234682,
|
| 7612 |
+
"grad_norm": 0.01633264310657978,
|
| 7613 |
+
"learning_rate": 9.893840362247809e-06,
|
| 7614 |
+
"loss": 0.0004,
|
| 7615 |
+
"step": 1082
|
| 7616 |
+
},
|
| 7617 |
+
{
|
| 7618 |
+
"epoch": 0.8043074637950242,
|
| 7619 |
+
"grad_norm": 0.002037476282566786,
|
| 7620 |
+
"learning_rate": 9.821580603366259e-06,
|
| 7621 |
+
"loss": 0.0001,
|
| 7622 |
+
"step": 1083
|
| 7623 |
+
},
|
| 7624 |
+
{
|
| 7625 |
+
"epoch": 0.80505012996658,
|
| 7626 |
+
"grad_norm": 0.0021626856178045273,
|
| 7627 |
+
"learning_rate": 9.749556937086085e-06,
|
| 7628 |
+
"loss": 0.0001,
|
| 7629 |
+
"step": 1084
|
| 7630 |
+
},
|
| 7631 |
+
{
|
| 7632 |
+
"epoch": 0.8057927961381359,
|
| 7633 |
+
"grad_norm": 0.0009408566984347999,
|
| 7634 |
+
"learning_rate": 9.677769786625867e-06,
|
| 7635 |
+
"loss": 0.0001,
|
| 7636 |
+
"step": 1085
|
| 7637 |
+
},
|
| 7638 |
+
{
|
| 7639 |
+
"epoch": 0.8065354623096918,
|
| 7640 |
+
"grad_norm": 0.0013037261087447405,
|
| 7641 |
+
"learning_rate": 9.606219573814445e-06,
|
| 7642 |
+
"loss": 0.0001,
|
| 7643 |
+
"step": 1086
|
| 7644 |
+
},
|
| 7645 |
+
{
|
| 7646 |
+
"epoch": 0.8072781284812477,
|
| 7647 |
+
"grad_norm": 0.0014423932880163193,
|
| 7648 |
+
"learning_rate": 9.534906719088355e-06,
|
| 7649 |
+
"loss": 0.0002,
|
| 7650 |
+
"step": 1087
|
| 7651 |
+
},
|
| 7652 |
+
{
|
| 7653 |
+
"epoch": 0.8080207946528035,
|
| 7654 |
+
"grad_norm": 0.0028046793304383755,
|
| 7655 |
+
"learning_rate": 9.46383164148939e-06,
|
| 7656 |
+
"loss": 0.0002,
|
| 7657 |
+
"step": 1088
|
| 7658 |
+
},
|
| 7659 |
+
{
|
| 7660 |
+
"epoch": 0.8087634608243595,
|
| 7661 |
+
"grad_norm": 0.0011104003060609102,
|
| 7662 |
+
"learning_rate": 9.392994758662178e-06,
|
| 7663 |
+
"loss": 0.0001,
|
| 7664 |
+
"step": 1089
|
| 7665 |
+
},
|
| 7666 |
+
{
|
| 7667 |
+
"epoch": 0.8095061269959153,
|
| 7668 |
+
"grad_norm": 0.0008305896772071719,
|
| 7669 |
+
"learning_rate": 9.322396486851626e-06,
|
| 7670 |
+
"loss": 0.0001,
|
| 7671 |
+
"step": 1090
|
| 7672 |
+
},
|
| 7673 |
+
{
|
| 7674 |
+
"epoch": 0.8102487931674712,
|
| 7675 |
+
"grad_norm": 0.049922533333301544,
|
| 7676 |
+
"learning_rate": 9.252037240900619e-06,
|
| 7677 |
+
"loss": 0.0017,
|
| 7678 |
+
"step": 1091
|
| 7679 |
+
},
|
| 7680 |
+
{
|
| 7681 |
+
"epoch": 0.810991459339027,
|
| 7682 |
+
"grad_norm": 0.0017735987203195691,
|
| 7683 |
+
"learning_rate": 9.181917434247416e-06,
|
| 7684 |
+
"loss": 0.0002,
|
| 7685 |
+
"step": 1092
|
| 7686 |
+
},
|
| 7687 |
+
{
|
| 7688 |
+
"epoch": 0.811734125510583,
|
| 7689 |
+
"grad_norm": 0.0012938451254740357,
|
| 7690 |
+
"learning_rate": 9.112037478923363e-06,
|
| 7691 |
+
"loss": 0.0002,
|
| 7692 |
+
"step": 1093
|
| 7693 |
+
},
|
| 7694 |
+
{
|
| 7695 |
+
"epoch": 0.8124767916821389,
|
| 7696 |
+
"grad_norm": 0.0025100386701524258,
|
| 7697 |
+
"learning_rate": 9.042397785550405e-06,
|
| 7698 |
+
"loss": 0.0002,
|
| 7699 |
+
"step": 1094
|
| 7700 |
+
},
|
| 7701 |
+
{
|
| 7702 |
+
"epoch": 0.8132194578536948,
|
| 7703 |
+
"grad_norm": 2.1535964012145996,
|
| 7704 |
+
"learning_rate": 8.972998763338685e-06,
|
| 7705 |
+
"loss": 0.291,
|
| 7706 |
+
"step": 1095
|
| 7707 |
+
},
|
| 7708 |
+
{
|
| 7709 |
+
"epoch": 0.8139621240252507,
|
| 7710 |
+
"grad_norm": 0.02024056576192379,
|
| 7711 |
+
"learning_rate": 8.903840820084096e-06,
|
| 7712 |
+
"loss": 0.0004,
|
| 7713 |
+
"step": 1096
|
| 7714 |
+
},
|
| 7715 |
+
{
|
| 7716 |
+
"epoch": 0.8147047901968065,
|
| 7717 |
+
"grad_norm": 0.002208460820838809,
|
| 7718 |
+
"learning_rate": 8.834924362165992e-06,
|
| 7719 |
+
"loss": 0.0002,
|
| 7720 |
+
"step": 1097
|
| 7721 |
+
},
|
| 7722 |
+
{
|
| 7723 |
+
"epoch": 0.8154474563683625,
|
| 7724 |
+
"grad_norm": 0.0023833250161260366,
|
| 7725 |
+
"learning_rate": 8.766249794544662e-06,
|
| 7726 |
+
"loss": 0.0002,
|
| 7727 |
+
"step": 1098
|
| 7728 |
+
},
|
| 7729 |
+
{
|
| 7730 |
+
"epoch": 0.8161901225399183,
|
| 7731 |
+
"grad_norm": 0.003369487589225173,
|
| 7732 |
+
"learning_rate": 8.697817520759093e-06,
|
| 7733 |
+
"loss": 0.0002,
|
| 7734 |
+
"step": 1099
|
| 7735 |
+
},
|
| 7736 |
+
{
|
| 7737 |
+
"epoch": 0.8169327887114742,
|
| 7738 |
+
"grad_norm": 0.003595678135752678,
|
| 7739 |
+
"learning_rate": 8.629627942924473e-06,
|
| 7740 |
+
"loss": 0.0002,
|
| 7741 |
+
"step": 1100
|
| 7742 |
+
},
|
| 7743 |
+
{
|
| 7744 |
+
"epoch": 0.8176754548830301,
|
| 7745 |
+
"grad_norm": 0.011070462875068188,
|
| 7746 |
+
"learning_rate": 8.561681461729926e-06,
|
| 7747 |
+
"loss": 0.0003,
|
| 7748 |
+
"step": 1101
|
| 7749 |
+
},
|
| 7750 |
+
{
|
| 7751 |
+
"epoch": 0.818418121054586,
|
| 7752 |
+
"grad_norm": 0.000978004070930183,
|
| 7753 |
+
"learning_rate": 8.49397847643606e-06,
|
| 7754 |
+
"loss": 0.0001,
|
| 7755 |
+
"step": 1102
|
| 7756 |
+
},
|
| 7757 |
+
{
|
| 7758 |
+
"epoch": 0.8191607872261418,
|
| 7759 |
+
"grad_norm": 0.0015130506362766027,
|
| 7760 |
+
"learning_rate": 8.426519384872733e-06,
|
| 7761 |
+
"loss": 0.0001,
|
| 7762 |
+
"step": 1103
|
| 7763 |
+
},
|
| 7764 |
+
{
|
| 7765 |
+
"epoch": 0.8199034533976978,
|
| 7766 |
+
"grad_norm": 0.0014117389218881726,
|
| 7767 |
+
"learning_rate": 8.359304583436644e-06,
|
| 7768 |
+
"loss": 0.0002,
|
| 7769 |
+
"step": 1104
|
| 7770 |
+
},
|
| 7771 |
+
{
|
| 7772 |
+
"epoch": 0.8206461195692536,
|
| 7773 |
+
"grad_norm": 0.0007138338405638933,
|
| 7774 |
+
"learning_rate": 8.29233446708898e-06,
|
| 7775 |
+
"loss": 0.0001,
|
| 7776 |
+
"step": 1105
|
| 7777 |
+
},
|
| 7778 |
+
{
|
| 7779 |
+
"epoch": 0.8213887857408095,
|
| 7780 |
+
"grad_norm": 0.0014467401197180152,
|
| 7781 |
+
"learning_rate": 8.225609429353187e-06,
|
| 7782 |
+
"loss": 0.0001,
|
| 7783 |
+
"step": 1106
|
| 7784 |
+
},
|
| 7785 |
+
{
|
| 7786 |
+
"epoch": 0.8221314519123654,
|
| 7787 |
+
"grad_norm": 0.0005772042204625905,
|
| 7788 |
+
"learning_rate": 8.159129862312548e-06,
|
| 7789 |
+
"loss": 0.0001,
|
| 7790 |
+
"step": 1107
|
| 7791 |
+
},
|
| 7792 |
+
{
|
| 7793 |
+
"epoch": 0.8228741180839213,
|
| 7794 |
+
"grad_norm": 0.0007005248917266726,
|
| 7795 |
+
"learning_rate": 8.092896156607987e-06,
|
| 7796 |
+
"loss": 0.0001,
|
| 7797 |
+
"step": 1108
|
| 7798 |
+
},
|
| 7799 |
+
{
|
| 7800 |
+
"epoch": 0.8236167842554771,
|
| 7801 |
+
"grad_norm": 0.0023481727112084627,
|
| 7802 |
+
"learning_rate": 8.026908701435681e-06,
|
| 7803 |
+
"loss": 0.0001,
|
| 7804 |
+
"step": 1109
|
| 7805 |
+
},
|
| 7806 |
+
{
|
| 7807 |
+
"epoch": 0.8243594504270331,
|
| 7808 |
+
"grad_norm": 0.0042024110443890095,
|
| 7809 |
+
"learning_rate": 7.96116788454485e-06,
|
| 7810 |
+
"loss": 0.0002,
|
| 7811 |
+
"step": 1110
|
| 7812 |
+
},
|
| 7813 |
+
{
|
| 7814 |
+
"epoch": 0.8251021165985889,
|
| 7815 |
+
"grad_norm": 0.001470068353228271,
|
| 7816 |
+
"learning_rate": 7.895674092235395e-06,
|
| 7817 |
+
"loss": 0.0001,
|
| 7818 |
+
"step": 1111
|
| 7819 |
+
},
|
| 7820 |
+
{
|
| 7821 |
+
"epoch": 0.8258447827701448,
|
| 7822 |
+
"grad_norm": 0.0019429409876465797,
|
| 7823 |
+
"learning_rate": 7.830427709355725e-06,
|
| 7824 |
+
"loss": 0.0002,
|
| 7825 |
+
"step": 1112
|
| 7826 |
+
},
|
| 7827 |
+
{
|
| 7828 |
+
"epoch": 0.8265874489417007,
|
| 7829 |
+
"grad_norm": 0.006123952101916075,
|
| 7830 |
+
"learning_rate": 7.765429119300394e-06,
|
| 7831 |
+
"loss": 0.0003,
|
| 7832 |
+
"step": 1113
|
| 7833 |
+
},
|
| 7834 |
+
{
|
| 7835 |
+
"epoch": 0.8273301151132566,
|
| 7836 |
+
"grad_norm": 0.0007870638510212302,
|
| 7837 |
+
"learning_rate": 7.700678704007947e-06,
|
| 7838 |
+
"loss": 0.0001,
|
| 7839 |
+
"step": 1114
|
| 7840 |
+
},
|
| 7841 |
+
{
|
| 7842 |
+
"epoch": 0.8280727812848124,
|
| 7843 |
+
"grad_norm": 0.0008543855510652065,
|
| 7844 |
+
"learning_rate": 7.636176843958598e-06,
|
| 7845 |
+
"loss": 0.0001,
|
| 7846 |
+
"step": 1115
|
| 7847 |
+
},
|
| 7848 |
+
{
|
| 7849 |
+
"epoch": 0.8288154474563684,
|
| 7850 |
+
"grad_norm": 0.010817396454513073,
|
| 7851 |
+
"learning_rate": 7.571923918172053e-06,
|
| 7852 |
+
"loss": 0.0002,
|
| 7853 |
+
"step": 1116
|
| 7854 |
+
},
|
| 7855 |
+
{
|
| 7856 |
+
"epoch": 0.8295581136279242,
|
| 7857 |
+
"grad_norm": 0.0015762196853756905,
|
| 7858 |
+
"learning_rate": 7.507920304205202e-06,
|
| 7859 |
+
"loss": 0.0002,
|
| 7860 |
+
"step": 1117
|
| 7861 |
+
},
|
| 7862 |
+
{
|
| 7863 |
+
"epoch": 0.8303007797994801,
|
| 7864 |
+
"grad_norm": 0.0009204100351780653,
|
| 7865 |
+
"learning_rate": 7.444166378150013e-06,
|
| 7866 |
+
"loss": 0.0001,
|
| 7867 |
+
"step": 1118
|
| 7868 |
+
},
|
| 7869 |
+
{
|
| 7870 |
+
"epoch": 0.831043445971036,
|
| 7871 |
+
"grad_norm": 0.0015323157422244549,
|
| 7872 |
+
"learning_rate": 7.3806625146312205e-06,
|
| 7873 |
+
"loss": 0.0002,
|
| 7874 |
+
"step": 1119
|
| 7875 |
+
},
|
| 7876 |
+
{
|
| 7877 |
+
"epoch": 0.8317861121425919,
|
| 7878 |
+
"grad_norm": 0.001088933553546667,
|
| 7879 |
+
"learning_rate": 7.317409086804189e-06,
|
| 7880 |
+
"loss": 0.0001,
|
| 7881 |
+
"step": 1120
|
| 7882 |
+
},
|
| 7883 |
+
{
|
| 7884 |
+
"epoch": 0.8325287783141478,
|
| 7885 |
+
"grad_norm": 0.002421481069177389,
|
| 7886 |
+
"learning_rate": 7.2544064663526815e-06,
|
| 7887 |
+
"loss": 0.0002,
|
| 7888 |
+
"step": 1121
|
| 7889 |
+
},
|
| 7890 |
+
{
|
| 7891 |
+
"epoch": 0.8332714444857037,
|
| 7892 |
+
"grad_norm": 0.0009983620839193463,
|
| 7893 |
+
"learning_rate": 7.191655023486682e-06,
|
| 7894 |
+
"loss": 0.0001,
|
| 7895 |
+
"step": 1122
|
| 7896 |
+
},
|
| 7897 |
+
{
|
| 7898 |
+
"epoch": 0.8340141106572596,
|
| 7899 |
+
"grad_norm": 0.003855592804029584,
|
| 7900 |
+
"learning_rate": 7.129155126940268e-06,
|
| 7901 |
+
"loss": 0.0002,
|
| 7902 |
+
"step": 1123
|
| 7903 |
+
},
|
| 7904 |
+
{
|
| 7905 |
+
"epoch": 0.8347567768288154,
|
| 7906 |
+
"grad_norm": 0.003245392581447959,
|
| 7907 |
+
"learning_rate": 7.066907143969353e-06,
|
| 7908 |
+
"loss": 0.0002,
|
| 7909 |
+
"step": 1124
|
| 7910 |
+
},
|
| 7911 |
+
{
|
| 7912 |
+
"epoch": 0.8354994430003714,
|
| 7913 |
+
"grad_norm": 0.005832053255289793,
|
| 7914 |
+
"learning_rate": 7.004911440349615e-06,
|
| 7915 |
+
"loss": 0.0003,
|
| 7916 |
+
"step": 1125
|
| 7917 |
+
},
|
| 7918 |
+
{
|
| 7919 |
+
"epoch": 0.8362421091719272,
|
| 7920 |
+
"grad_norm": 0.0075843255035579205,
|
| 7921 |
+
"learning_rate": 6.94316838037431e-06,
|
| 7922 |
+
"loss": 0.0002,
|
| 7923 |
+
"step": 1126
|
| 7924 |
+
},
|
| 7925 |
+
{
|
| 7926 |
+
"epoch": 0.8369847753434831,
|
| 7927 |
+
"grad_norm": 0.019108982756733894,
|
| 7928 |
+
"learning_rate": 6.881678326852136e-06,
|
| 7929 |
+
"loss": 0.0005,
|
| 7930 |
+
"step": 1127
|
| 7931 |
+
},
|
| 7932 |
+
{
|
| 7933 |
+
"epoch": 0.837727441515039,
|
| 7934 |
+
"grad_norm": 0.0029295405838638544,
|
| 7935 |
+
"learning_rate": 6.820441641105063e-06,
|
| 7936 |
+
"loss": 0.0003,
|
| 7937 |
+
"step": 1128
|
| 7938 |
+
},
|
| 7939 |
+
{
|
| 7940 |
+
"epoch": 0.8384701076865949,
|
| 7941 |
+
"grad_norm": 0.0009995042346417904,
|
| 7942 |
+
"learning_rate": 6.759458682966296e-06,
|
| 7943 |
+
"loss": 0.0001,
|
| 7944 |
+
"step": 1129
|
| 7945 |
+
},
|
| 7946 |
+
{
|
| 7947 |
+
"epoch": 0.8392127738581507,
|
| 7948 |
+
"grad_norm": 0.0011027619475498796,
|
| 7949 |
+
"learning_rate": 6.698729810778065e-06,
|
| 7950 |
+
"loss": 0.0002,
|
| 7951 |
+
"step": 1130
|
| 7952 |
+
},
|
| 7953 |
+
{
|
| 7954 |
+
"epoch": 0.8399554400297067,
|
| 7955 |
+
"grad_norm": 0.0015378465177491307,
|
| 7956 |
+
"learning_rate": 6.6382553813896075e-06,
|
| 7957 |
+
"loss": 0.0001,
|
| 7958 |
+
"step": 1131
|
| 7959 |
+
},
|
| 7960 |
+
{
|
| 7961 |
+
"epoch": 0.8406981062012625,
|
| 7962 |
+
"grad_norm": 0.0011789267882704735,
|
| 7963 |
+
"learning_rate": 6.578035750154976e-06,
|
| 7964 |
+
"loss": 0.0001,
|
| 7965 |
+
"step": 1132
|
| 7966 |
+
},
|
| 7967 |
+
{
|
| 7968 |
+
"epoch": 0.8414407723728184,
|
| 7969 |
+
"grad_norm": 0.003139357315376401,
|
| 7970 |
+
"learning_rate": 6.518071270931059e-06,
|
| 7971 |
+
"loss": 0.0002,
|
| 7972 |
+
"step": 1133
|
| 7973 |
+
},
|
| 7974 |
+
{
|
| 7975 |
+
"epoch": 0.8421834385443743,
|
| 7976 |
+
"grad_norm": 0.004962114151567221,
|
| 7977 |
+
"learning_rate": 6.458362296075399e-06,
|
| 7978 |
+
"loss": 0.0003,
|
| 7979 |
+
"step": 1134
|
| 7980 |
+
},
|
| 7981 |
+
{
|
| 7982 |
+
"epoch": 0.8429261047159302,
|
| 7983 |
+
"grad_norm": 0.0013825440546497703,
|
| 7984 |
+
"learning_rate": 6.398909176444207e-06,
|
| 7985 |
+
"loss": 0.0001,
|
| 7986 |
+
"step": 1135
|
| 7987 |
+
},
|
| 7988 |
+
{
|
| 7989 |
+
"epoch": 0.843668770887486,
|
| 7990 |
+
"grad_norm": 0.0008226705831475556,
|
| 7991 |
+
"learning_rate": 6.339712261390213e-06,
|
| 7992 |
+
"loss": 0.0001,
|
| 7993 |
+
"step": 1136
|
| 7994 |
+
},
|
| 7995 |
+
{
|
| 7996 |
+
"epoch": 0.844411437059042,
|
| 7997 |
+
"grad_norm": 0.00134433398488909,
|
| 7998 |
+
"learning_rate": 6.280771898760718e-06,
|
| 7999 |
+
"loss": 0.0001,
|
| 8000 |
+
"step": 1137
|
| 8001 |
+
},
|
| 8002 |
+
{
|
| 8003 |
+
"epoch": 0.8451541032305978,
|
| 8004 |
+
"grad_norm": 0.0013929035048931837,
|
| 8005 |
+
"learning_rate": 6.222088434895462e-06,
|
| 8006 |
+
"loss": 0.0001,
|
| 8007 |
+
"step": 1138
|
| 8008 |
+
},
|
| 8009 |
+
{
|
| 8010 |
+
"epoch": 0.8458967694021537,
|
| 8011 |
+
"grad_norm": 0.00138406315818429,
|
| 8012 |
+
"learning_rate": 6.163662214624616e-06,
|
| 8013 |
+
"loss": 0.0002,
|
| 8014 |
+
"step": 1139
|
| 8015 |
+
},
|
| 8016 |
+
{
|
| 8017 |
+
"epoch": 0.8466394355737096,
|
| 8018 |
+
"grad_norm": 0.0019744529854506254,
|
| 8019 |
+
"learning_rate": 6.1054935812667855e-06,
|
| 8020 |
+
"loss": 0.0002,
|
| 8021 |
+
"step": 1140
|
| 8022 |
+
},
|
| 8023 |
+
{
|
| 8024 |
+
"epoch": 0.8473821017452655,
|
| 8025 |
+
"grad_norm": 0.0007763861794956028,
|
| 8026 |
+
"learning_rate": 6.047582876626934e-06,
|
| 8027 |
+
"loss": 0.0001,
|
| 8028 |
+
"step": 1141
|
| 8029 |
+
},
|
| 8030 |
+
{
|
| 8031 |
+
"epoch": 0.8481247679168213,
|
| 8032 |
+
"grad_norm": 0.0008731714333407581,
|
| 8033 |
+
"learning_rate": 5.989930440994451e-06,
|
| 8034 |
+
"loss": 0.0001,
|
| 8035 |
+
"step": 1142
|
| 8036 |
+
},
|
| 8037 |
+
{
|
| 8038 |
+
"epoch": 0.8488674340883773,
|
| 8039 |
+
"grad_norm": 0.0010884518269449472,
|
| 8040 |
+
"learning_rate": 5.932536613141065e-06,
|
| 8041 |
+
"loss": 0.0001,
|
| 8042 |
+
"step": 1143
|
| 8043 |
+
},
|
| 8044 |
+
{
|
| 8045 |
+
"epoch": 0.8496101002599331,
|
| 8046 |
+
"grad_norm": 0.004011042881757021,
|
| 8047 |
+
"learning_rate": 5.875401730318947e-06,
|
| 8048 |
+
"loss": 0.0002,
|
| 8049 |
+
"step": 1144
|
| 8050 |
+
},
|
| 8051 |
+
{
|
| 8052 |
+
"epoch": 0.850352766431489,
|
| 8053 |
+
"grad_norm": 0.0016273773508146405,
|
| 8054 |
+
"learning_rate": 5.818526128258622e-06,
|
| 8055 |
+
"loss": 0.0002,
|
| 8056 |
+
"step": 1145
|
| 8057 |
+
},
|
| 8058 |
+
{
|
| 8059 |
+
"epoch": 0.8510954326030449,
|
| 8060 |
+
"grad_norm": 0.001861872267909348,
|
| 8061 |
+
"learning_rate": 5.7619101411671095e-06,
|
| 8062 |
+
"loss": 0.0001,
|
| 8063 |
+
"step": 1146
|
| 8064 |
+
},
|
| 8065 |
+
{
|
| 8066 |
+
"epoch": 0.8518380987746008,
|
| 8067 |
+
"grad_norm": 0.0037269063759595156,
|
| 8068 |
+
"learning_rate": 5.7055541017258855e-06,
|
| 8069 |
+
"loss": 0.0003,
|
| 8070 |
+
"step": 1147
|
| 8071 |
+
},
|
| 8072 |
+
{
|
| 8073 |
+
"epoch": 0.8525807649461568,
|
| 8074 |
+
"grad_norm": 0.0021908970084041357,
|
| 8075 |
+
"learning_rate": 5.649458341088915e-06,
|
| 8076 |
+
"loss": 0.0002,
|
| 8077 |
+
"step": 1148
|
| 8078 |
+
},
|
| 8079 |
+
{
|
| 8080 |
+
"epoch": 0.8533234311177126,
|
| 8081 |
+
"grad_norm": 0.002067559864372015,
|
| 8082 |
+
"learning_rate": 5.593623188880781e-06,
|
| 8083 |
+
"loss": 0.0002,
|
| 8084 |
+
"step": 1149
|
| 8085 |
+
},
|
| 8086 |
+
{
|
| 8087 |
+
"epoch": 0.8540660972892685,
|
| 8088 |
+
"grad_norm": 0.004860437475144863,
|
| 8089 |
+
"learning_rate": 5.538048973194698e-06,
|
| 8090 |
+
"loss": 0.0002,
|
| 8091 |
+
"step": 1150
|
| 8092 |
+
},
|
| 8093 |
+
{
|
| 8094 |
+
"epoch": 0.8548087634608244,
|
| 8095 |
+
"grad_norm": 0.00816015899181366,
|
| 8096 |
+
"learning_rate": 5.482736020590551e-06,
|
| 8097 |
+
"loss": 0.0002,
|
| 8098 |
+
"step": 1151
|
| 8099 |
+
},
|
| 8100 |
+
{
|
| 8101 |
+
"epoch": 0.8555514296323803,
|
| 8102 |
+
"grad_norm": 0.23404280841350555,
|
| 8103 |
+
"learning_rate": 5.427684656093074e-06,
|
| 8104 |
+
"loss": 0.0006,
|
| 8105 |
+
"step": 1152
|
| 8106 |
+
},
|
| 8107 |
+
{
|
| 8108 |
+
"epoch": 0.8562940958039361,
|
| 8109 |
+
"grad_norm": 0.0022261200938373804,
|
| 8110 |
+
"learning_rate": 5.372895203189837e-06,
|
| 8111 |
+
"loss": 0.0002,
|
| 8112 |
+
"step": 1153
|
| 8113 |
+
},
|
| 8114 |
+
{
|
| 8115 |
+
"epoch": 0.8570367619754921,
|
| 8116 |
+
"grad_norm": 0.001256439252756536,
|
| 8117 |
+
"learning_rate": 5.318367983829392e-06,
|
| 8118 |
+
"loss": 0.0001,
|
| 8119 |
+
"step": 1154
|
| 8120 |
+
},
|
| 8121 |
+
{
|
| 8122 |
+
"epoch": 0.8577794281470479,
|
| 8123 |
+
"grad_norm": 0.0017608028138056397,
|
| 8124 |
+
"learning_rate": 5.2641033184194165e-06,
|
| 8125 |
+
"loss": 0.0002,
|
| 8126 |
+
"step": 1155
|
| 8127 |
+
},
|
| 8128 |
+
{
|
| 8129 |
+
"epoch": 0.8585220943186038,
|
| 8130 |
+
"grad_norm": 0.0018713950412347913,
|
| 8131 |
+
"learning_rate": 5.210101525824746e-06,
|
| 8132 |
+
"loss": 0.0001,
|
| 8133 |
+
"step": 1156
|
| 8134 |
+
},
|
| 8135 |
+
{
|
| 8136 |
+
"epoch": 0.8592647604901597,
|
| 8137 |
+
"grad_norm": 0.0007540153455920517,
|
| 8138 |
+
"learning_rate": 5.156362923365588e-06,
|
| 8139 |
+
"loss": 0.0001,
|
| 8140 |
+
"step": 1157
|
| 8141 |
+
},
|
| 8142 |
+
{
|
| 8143 |
+
"epoch": 0.8600074266617156,
|
| 8144 |
+
"grad_norm": 0.0012859681155532598,
|
| 8145 |
+
"learning_rate": 5.102887826815589e-06,
|
| 8146 |
+
"loss": 0.0001,
|
| 8147 |
+
"step": 1158
|
| 8148 |
+
},
|
| 8149 |
+
{
|
| 8150 |
+
"epoch": 0.8607500928332714,
|
| 8151 |
+
"grad_norm": 0.0008075116202235222,
|
| 8152 |
+
"learning_rate": 5.049676550400034e-06,
|
| 8153 |
+
"loss": 0.0001,
|
| 8154 |
+
"step": 1159
|
| 8155 |
+
},
|
| 8156 |
+
{
|
| 8157 |
+
"epoch": 0.8614927590048274,
|
| 8158 |
+
"grad_norm": 0.0007994707557372749,
|
| 8159 |
+
"learning_rate": 4.996729406793943e-06,
|
| 8160 |
+
"loss": 0.0001,
|
| 8161 |
+
"step": 1160
|
| 8162 |
+
},
|
| 8163 |
+
{
|
| 8164 |
+
"epoch": 0.8622354251763832,
|
| 8165 |
+
"grad_norm": 0.004248554352670908,
|
| 8166 |
+
"learning_rate": 4.944046707120287e-06,
|
| 8167 |
+
"loss": 0.0002,
|
| 8168 |
+
"step": 1161
|
| 8169 |
+
},
|
| 8170 |
+
{
|
| 8171 |
+
"epoch": 0.8629780913479391,
|
| 8172 |
+
"grad_norm": 0.0005359674105420709,
|
| 8173 |
+
"learning_rate": 4.891628760948114e-06,
|
| 8174 |
+
"loss": 0.0001,
|
| 8175 |
+
"step": 1162
|
| 8176 |
+
},
|
| 8177 |
+
{
|
| 8178 |
+
"epoch": 0.863720757519495,
|
| 8179 |
+
"grad_norm": 0.0008803227683529258,
|
| 8180 |
+
"learning_rate": 4.839475876290789e-06,
|
| 8181 |
+
"loss": 0.0001,
|
| 8182 |
+
"step": 1163
|
| 8183 |
+
},
|
| 8184 |
+
{
|
| 8185 |
+
"epoch": 0.8644634236910509,
|
| 8186 |
+
"grad_norm": 0.0011372326407581568,
|
| 8187 |
+
"learning_rate": 4.787588359604106e-06,
|
| 8188 |
+
"loss": 0.0002,
|
| 8189 |
+
"step": 1164
|
| 8190 |
+
},
|
| 8191 |
+
{
|
| 8192 |
+
"epoch": 0.8652060898626067,
|
| 8193 |
+
"grad_norm": 0.004412113688886166,
|
| 8194 |
+
"learning_rate": 4.735966515784562e-06,
|
| 8195 |
+
"loss": 0.0002,
|
| 8196 |
+
"step": 1165
|
| 8197 |
+
},
|
| 8198 |
+
{
|
| 8199 |
+
"epoch": 0.8659487560341627,
|
| 8200 |
+
"grad_norm": 0.0014664451591670513,
|
| 8201 |
+
"learning_rate": 4.684610648167503e-06,
|
| 8202 |
+
"loss": 0.0002,
|
| 8203 |
+
"step": 1166
|
| 8204 |
+
},
|
| 8205 |
+
{
|
| 8206 |
+
"epoch": 0.8666914222057185,
|
| 8207 |
+
"grad_norm": 0.00369854923337698,
|
| 8208 |
+
"learning_rate": 4.633521058525408e-06,
|
| 8209 |
+
"loss": 0.0002,
|
| 8210 |
+
"step": 1167
|
| 8211 |
+
},
|
| 8212 |
+
{
|
| 8213 |
+
"epoch": 0.8674340883772744,
|
| 8214 |
+
"grad_norm": 0.0015638495096936822,
|
| 8215 |
+
"learning_rate": 4.582698047066036e-06,
|
| 8216 |
+
"loss": 0.0001,
|
| 8217 |
+
"step": 1168
|
| 8218 |
+
},
|
| 8219 |
+
{
|
| 8220 |
+
"epoch": 0.8681767545488303,
|
| 8221 |
+
"grad_norm": 0.026589730754494667,
|
| 8222 |
+
"learning_rate": 4.532141912430743e-06,
|
| 8223 |
+
"loss": 0.0005,
|
| 8224 |
+
"step": 1169
|
| 8225 |
+
},
|
| 8226 |
+
{
|
| 8227 |
+
"epoch": 0.8689194207203862,
|
| 8228 |
+
"grad_norm": 0.0014144079759716988,
|
| 8229 |
+
"learning_rate": 4.4818529516926726e-06,
|
| 8230 |
+
"loss": 0.0001,
|
| 8231 |
+
"step": 1170
|
| 8232 |
+
},
|
| 8233 |
+
{
|
| 8234 |
+
"epoch": 0.869662086891942,
|
| 8235 |
+
"grad_norm": 0.0023864826653152704,
|
| 8236 |
+
"learning_rate": 4.4318314603550074e-06,
|
| 8237 |
+
"loss": 0.0002,
|
| 8238 |
+
"step": 1171
|
| 8239 |
+
},
|
| 8240 |
+
{
|
| 8241 |
+
"epoch": 0.870404753063498,
|
| 8242 |
+
"grad_norm": 0.00443720119073987,
|
| 8243 |
+
"learning_rate": 4.382077732349299e-06,
|
| 8244 |
+
"loss": 0.0002,
|
| 8245 |
+
"step": 1172
|
| 8246 |
+
},
|
| 8247 |
+
{
|
| 8248 |
+
"epoch": 0.8711474192350538,
|
| 8249 |
+
"grad_norm": 0.0019203081028535962,
|
| 8250 |
+
"learning_rate": 4.332592060033636e-06,
|
| 8251 |
+
"loss": 0.0001,
|
| 8252 |
+
"step": 1173
|
| 8253 |
+
},
|
| 8254 |
+
{
|
| 8255 |
+
"epoch": 0.8718900854066097,
|
| 8256 |
+
"grad_norm": 0.0016560673248022795,
|
| 8257 |
+
"learning_rate": 4.283374734191037e-06,
|
| 8258 |
+
"loss": 0.0002,
|
| 8259 |
+
"step": 1174
|
| 8260 |
+
},
|
| 8261 |
+
{
|
| 8262 |
+
"epoch": 0.8726327515781657,
|
| 8263 |
+
"grad_norm": 0.0018693654565140605,
|
| 8264 |
+
"learning_rate": 4.234426044027645e-06,
|
| 8265 |
+
"loss": 0.0001,
|
| 8266 |
+
"step": 1175
|
| 8267 |
+
},
|
| 8268 |
+
{
|
| 8269 |
+
"epoch": 0.8733754177497215,
|
| 8270 |
+
"grad_norm": 0.041886404156684875,
|
| 8271 |
+
"learning_rate": 4.18574627717111e-06,
|
| 8272 |
+
"loss": 0.0008,
|
| 8273 |
+
"step": 1176
|
| 8274 |
+
},
|
| 8275 |
+
{
|
| 8276 |
+
"epoch": 0.8741180839212774,
|
| 8277 |
+
"grad_norm": 0.0033249545376747847,
|
| 8278 |
+
"learning_rate": 4.137335719668822e-06,
|
| 8279 |
+
"loss": 0.0001,
|
| 8280 |
+
"step": 1177
|
| 8281 |
+
},
|
| 8282 |
+
{
|
| 8283 |
+
"epoch": 0.8748607500928333,
|
| 8284 |
+
"grad_norm": 0.0029907547868788242,
|
| 8285 |
+
"learning_rate": 4.089194655986306e-06,
|
| 8286 |
+
"loss": 0.0002,
|
| 8287 |
+
"step": 1178
|
| 8288 |
+
},
|
| 8289 |
+
{
|
| 8290 |
+
"epoch": 0.8756034162643892,
|
| 8291 |
+
"grad_norm": 0.0019523641094565392,
|
| 8292 |
+
"learning_rate": 4.041323369005479e-06,
|
| 8293 |
+
"loss": 0.0002,
|
| 8294 |
+
"step": 1179
|
| 8295 |
+
},
|
| 8296 |
+
{
|
| 8297 |
+
"epoch": 0.876346082435945,
|
| 8298 |
+
"grad_norm": 0.0027865059673786163,
|
| 8299 |
+
"learning_rate": 3.9937221400230484e-06,
|
| 8300 |
+
"loss": 0.0001,
|
| 8301 |
+
"step": 1180
|
| 8302 |
+
},
|
| 8303 |
+
{
|
| 8304 |
+
"epoch": 0.877088748607501,
|
| 8305 |
+
"grad_norm": 0.0016474600415676832,
|
| 8306 |
+
"learning_rate": 3.946391248748821e-06,
|
| 8307 |
+
"loss": 0.0002,
|
| 8308 |
+
"step": 1181
|
| 8309 |
+
},
|
| 8310 |
+
{
|
| 8311 |
+
"epoch": 0.8778314147790568,
|
| 8312 |
+
"grad_norm": 0.0016686657909303904,
|
| 8313 |
+
"learning_rate": 3.899330973304083e-06,
|
| 8314 |
+
"loss": 0.0001,
|
| 8315 |
+
"step": 1182
|
| 8316 |
+
},
|
| 8317 |
+
{
|
| 8318 |
+
"epoch": 0.8785740809506127,
|
| 8319 |
+
"grad_norm": 0.03477117419242859,
|
| 8320 |
+
"learning_rate": 3.8525415902199154e-06,
|
| 8321 |
+
"loss": 0.0013,
|
| 8322 |
+
"step": 1183
|
| 8323 |
+
},
|
| 8324 |
+
{
|
| 8325 |
+
"epoch": 0.8793167471221686,
|
| 8326 |
+
"grad_norm": 0.000568350194953382,
|
| 8327 |
+
"learning_rate": 3.8060233744356633e-06,
|
| 8328 |
+
"loss": 0.0001,
|
| 8329 |
+
"step": 1184
|
| 8330 |
+
},
|
| 8331 |
+
{
|
| 8332 |
+
"epoch": 0.8800594132937245,
|
| 8333 |
+
"grad_norm": 0.0022050561383366585,
|
| 8334 |
+
"learning_rate": 3.7597765992972267e-06,
|
| 8335 |
+
"loss": 0.0002,
|
| 8336 |
+
"step": 1185
|
| 8337 |
+
},
|
| 8338 |
+
{
|
| 8339 |
+
"epoch": 0.8808020794652803,
|
| 8340 |
+
"grad_norm": 0.0006907075876370072,
|
| 8341 |
+
"learning_rate": 3.7138015365554833e-06,
|
| 8342 |
+
"loss": 0.0001,
|
| 8343 |
+
"step": 1186
|
| 8344 |
+
},
|
| 8345 |
+
{
|
| 8346 |
+
"epoch": 0.8815447456368363,
|
| 8347 |
+
"grad_norm": 0.0008904426940716803,
|
| 8348 |
+
"learning_rate": 3.6680984563647434e-06,
|
| 8349 |
+
"loss": 0.0001,
|
| 8350 |
+
"step": 1187
|
| 8351 |
+
},
|
| 8352 |
+
{
|
| 8353 |
+
"epoch": 0.8822874118083921,
|
| 8354 |
+
"grad_norm": 0.0031244743149727583,
|
| 8355 |
+
"learning_rate": 3.6226676272810735e-06,
|
| 8356 |
+
"loss": 0.0002,
|
| 8357 |
+
"step": 1188
|
| 8358 |
+
},
|
| 8359 |
+
{
|
| 8360 |
+
"epoch": 0.883030077979948,
|
| 8361 |
+
"grad_norm": 0.0016301183495670557,
|
| 8362 |
+
"learning_rate": 3.5775093162607963e-06,
|
| 8363 |
+
"loss": 0.0001,
|
| 8364 |
+
"step": 1189
|
| 8365 |
+
},
|
| 8366 |
+
{
|
| 8367 |
+
"epoch": 0.8837727441515039,
|
| 8368 |
+
"grad_norm": 0.006416819058358669,
|
| 8369 |
+
"learning_rate": 3.5326237886588732e-06,
|
| 8370 |
+
"loss": 0.0003,
|
| 8371 |
+
"step": 1190
|
| 8372 |
+
},
|
| 8373 |
+
{
|
| 8374 |
+
"epoch": 0.8845154103230598,
|
| 8375 |
+
"grad_norm": 0.0009310771129094064,
|
| 8376 |
+
"learning_rate": 3.488011308227379e-06,
|
| 8377 |
+
"loss": 0.0001,
|
| 8378 |
+
"step": 1191
|
| 8379 |
+
},
|
| 8380 |
+
{
|
| 8381 |
+
"epoch": 0.8852580764946156,
|
| 8382 |
+
"grad_norm": 0.00095487164799124,
|
| 8383 |
+
"learning_rate": 3.443672137113901e-06,
|
| 8384 |
+
"loss": 0.0001,
|
| 8385 |
+
"step": 1192
|
| 8386 |
+
},
|
| 8387 |
+
{
|
| 8388 |
+
"epoch": 0.8860007426661716,
|
| 8389 |
+
"grad_norm": 0.0011116062523797154,
|
| 8390 |
+
"learning_rate": 3.3996065358600782e-06,
|
| 8391 |
+
"loss": 0.0001,
|
| 8392 |
+
"step": 1193
|
| 8393 |
+
},
|
| 8394 |
+
{
|
| 8395 |
+
"epoch": 0.8867434088377274,
|
| 8396 |
+
"grad_norm": 0.0024930539075285196,
|
| 8397 |
+
"learning_rate": 3.3558147633999728e-06,
|
| 8398 |
+
"loss": 0.0002,
|
| 8399 |
+
"step": 1194
|
| 8400 |
+
},
|
| 8401 |
+
{
|
| 8402 |
+
"epoch": 0.8874860750092833,
|
| 8403 |
+
"grad_norm": 0.0017195651307702065,
|
| 8404 |
+
"learning_rate": 3.3122970770586514e-06,
|
| 8405 |
+
"loss": 0.0001,
|
| 8406 |
+
"step": 1195
|
| 8407 |
+
},
|
| 8408 |
+
{
|
| 8409 |
+
"epoch": 0.8882287411808392,
|
| 8410 |
+
"grad_norm": 0.0015204905066639185,
|
| 8411 |
+
"learning_rate": 3.269053732550581e-06,
|
| 8412 |
+
"loss": 0.0002,
|
| 8413 |
+
"step": 1196
|
| 8414 |
+
},
|
| 8415 |
+
{
|
| 8416 |
+
"epoch": 0.8889714073523951,
|
| 8417 |
+
"grad_norm": 0.005054555833339691,
|
| 8418 |
+
"learning_rate": 3.226084983978195e-06,
|
| 8419 |
+
"loss": 0.0003,
|
| 8420 |
+
"step": 1197
|
| 8421 |
+
},
|
| 8422 |
+
{
|
| 8423 |
+
"epoch": 0.8897140735239509,
|
| 8424 |
+
"grad_norm": 0.0018111229874193668,
|
| 8425 |
+
"learning_rate": 3.183391083830345e-06,
|
| 8426 |
+
"loss": 0.0001,
|
| 8427 |
+
"step": 1198
|
| 8428 |
+
},
|
| 8429 |
+
{
|
| 8430 |
+
"epoch": 0.8904567396955069,
|
| 8431 |
+
"grad_norm": 0.026347478851675987,
|
| 8432 |
+
"learning_rate": 3.1409722829808863e-06,
|
| 8433 |
+
"loss": 0.0005,
|
| 8434 |
+
"step": 1199
|
| 8435 |
+
},
|
| 8436 |
+
{
|
| 8437 |
+
"epoch": 0.8911994058670628,
|
| 8438 |
+
"grad_norm": 0.002853583311662078,
|
| 8439 |
+
"learning_rate": 3.0988288306871115e-06,
|
| 8440 |
+
"loss": 0.0002,
|
| 8441 |
+
"step": 1200
|
| 8442 |
+
},
|
| 8443 |
+
{
|
| 8444 |
+
"epoch": 0.8919420720386186,
|
| 8445 |
+
"grad_norm": 0.0017346449894830585,
|
| 8446 |
+
"learning_rate": 3.0569609745883743e-06,
|
| 8447 |
+
"loss": 0.0001,
|
| 8448 |
+
"step": 1201
|
| 8449 |
+
},
|
| 8450 |
+
{
|
| 8451 |
+
"epoch": 0.8926847382101746,
|
| 8452 |
+
"grad_norm": 0.31705546379089355,
|
| 8453 |
+
"learning_rate": 3.0153689607045845e-06,
|
| 8454 |
+
"loss": 0.0023,
|
| 8455 |
+
"step": 1202
|
| 8456 |
+
},
|
| 8457 |
+
{
|
| 8458 |
+
"epoch": 0.8934274043817304,
|
| 8459 |
+
"grad_norm": 0.2009972184896469,
|
| 8460 |
+
"learning_rate": 2.9740530334347594e-06,
|
| 8461 |
+
"loss": 0.0012,
|
| 8462 |
+
"step": 1203
|
| 8463 |
+
},
|
| 8464 |
+
{
|
| 8465 |
+
"epoch": 0.8941700705532863,
|
| 8466 |
+
"grad_norm": 0.009426708333194256,
|
| 8467 |
+
"learning_rate": 2.933013435555637e-06,
|
| 8468 |
+
"loss": 0.0003,
|
| 8469 |
+
"step": 1204
|
| 8470 |
+
},
|
| 8471 |
+
{
|
| 8472 |
+
"epoch": 0.8949127367248422,
|
| 8473 |
+
"grad_norm": 0.0008448066073469818,
|
| 8474 |
+
"learning_rate": 2.8922504082201764e-06,
|
| 8475 |
+
"loss": 0.0001,
|
| 8476 |
+
"step": 1205
|
| 8477 |
+
},
|
| 8478 |
+
{
|
| 8479 |
+
"epoch": 0.8956554028963981,
|
| 8480 |
+
"grad_norm": 0.0007227634778246284,
|
| 8481 |
+
"learning_rate": 2.8517641909562077e-06,
|
| 8482 |
+
"loss": 0.0001,
|
| 8483 |
+
"step": 1206
|
| 8484 |
+
},
|
| 8485 |
+
{
|
| 8486 |
+
"epoch": 0.8963980690679539,
|
| 8487 |
+
"grad_norm": 0.007234448567032814,
|
| 8488 |
+
"learning_rate": 2.811555021664969e-06,
|
| 8489 |
+
"loss": 0.0003,
|
| 8490 |
+
"step": 1207
|
| 8491 |
+
},
|
| 8492 |
+
{
|
| 8493 |
+
"epoch": 0.8971407352395099,
|
| 8494 |
+
"grad_norm": 0.003920075949281454,
|
| 8495 |
+
"learning_rate": 2.7716231366197663e-06,
|
| 8496 |
+
"loss": 0.0002,
|
| 8497 |
+
"step": 1208
|
| 8498 |
+
},
|
| 8499 |
+
{
|
| 8500 |
+
"epoch": 0.8978834014110657,
|
| 8501 |
+
"grad_norm": 0.0014649624936282635,
|
| 8502 |
+
"learning_rate": 2.7319687704645326e-06,
|
| 8503 |
+
"loss": 0.0001,
|
| 8504 |
+
"step": 1209
|
| 8505 |
+
},
|
| 8506 |
+
{
|
| 8507 |
+
"epoch": 0.8986260675826216,
|
| 8508 |
+
"grad_norm": 0.0005429857992567122,
|
| 8509 |
+
"learning_rate": 2.692592156212487e-06,
|
| 8510 |
+
"loss": 0.0001,
|
| 8511 |
+
"step": 1210
|
| 8512 |
+
},
|
| 8513 |
+
{
|
| 8514 |
+
"epoch": 0.8993687337541775,
|
| 8515 |
+
"grad_norm": 0.004393384791910648,
|
| 8516 |
+
"learning_rate": 2.653493525244721e-06,
|
| 8517 |
+
"loss": 0.0002,
|
| 8518 |
+
"step": 1211
|
| 8519 |
+
},
|
| 8520 |
+
{
|
| 8521 |
+
"epoch": 0.9001113999257334,
|
| 8522 |
+
"grad_norm": 0.0011112246429547668,
|
| 8523 |
+
"learning_rate": 2.614673107308896e-06,
|
| 8524 |
+
"loss": 0.0001,
|
| 8525 |
+
"step": 1212
|
| 8526 |
+
},
|
| 8527 |
+
{
|
| 8528 |
+
"epoch": 0.9008540660972892,
|
| 8529 |
+
"grad_norm": 0.0010931448778137565,
|
| 8530 |
+
"learning_rate": 2.5761311305178614e-06,
|
| 8531 |
+
"loss": 0.0001,
|
| 8532 |
+
"step": 1213
|
| 8533 |
+
},
|
| 8534 |
+
{
|
| 8535 |
+
"epoch": 0.9015967322688452,
|
| 8536 |
+
"grad_norm": 0.0010206797160208225,
|
| 8537 |
+
"learning_rate": 2.5378678213483054e-06,
|
| 8538 |
+
"loss": 0.0001,
|
| 8539 |
+
"step": 1214
|
| 8540 |
+
},
|
| 8541 |
+
{
|
| 8542 |
+
"epoch": 0.902339398440401,
|
| 8543 |
+
"grad_norm": 0.0024138889275491238,
|
| 8544 |
+
"learning_rate": 2.499883404639436e-06,
|
| 8545 |
+
"loss": 0.0002,
|
| 8546 |
+
"step": 1215
|
| 8547 |
+
},
|
| 8548 |
+
{
|
| 8549 |
+
"epoch": 0.903082064611957,
|
| 8550 |
+
"grad_norm": 0.0011341659119352698,
|
| 8551 |
+
"learning_rate": 2.462178103591678e-06,
|
| 8552 |
+
"loss": 0.0001,
|
| 8553 |
+
"step": 1216
|
| 8554 |
+
},
|
| 8555 |
+
{
|
| 8556 |
+
"epoch": 0.9038247307835128,
|
| 8557 |
+
"grad_norm": 0.0011666857171803713,
|
| 8558 |
+
"learning_rate": 2.424752139765318e-06,
|
| 8559 |
+
"loss": 0.0001,
|
| 8560 |
+
"step": 1217
|
| 8561 |
+
},
|
| 8562 |
+
{
|
| 8563 |
+
"epoch": 0.9045673969550687,
|
| 8564 |
+
"grad_norm": 0.0019645995926111937,
|
| 8565 |
+
"learning_rate": 2.3876057330792346e-06,
|
| 8566 |
+
"loss": 0.0002,
|
| 8567 |
+
"step": 1218
|
| 8568 |
+
},
|
| 8569 |
+
{
|
| 8570 |
+
"epoch": 0.9053100631266245,
|
| 8571 |
+
"grad_norm": 0.0026836565230041742,
|
| 8572 |
+
"learning_rate": 2.350739101809624e-06,
|
| 8573 |
+
"loss": 0.0003,
|
| 8574 |
+
"step": 1219
|
| 8575 |
+
},
|
| 8576 |
+
{
|
| 8577 |
+
"epoch": 0.9060527292981805,
|
| 8578 |
+
"grad_norm": 0.0015338718658313155,
|
| 8579 |
+
"learning_rate": 2.314152462588659e-06,
|
| 8580 |
+
"loss": 0.0001,
|
| 8581 |
+
"step": 1220
|
| 8582 |
+
},
|
| 8583 |
+
{
|
| 8584 |
+
"epoch": 0.9067953954697363,
|
| 8585 |
+
"grad_norm": 0.007026324979960918,
|
| 8586 |
+
"learning_rate": 2.2778460304032634e-06,
|
| 8587 |
+
"loss": 0.0003,
|
| 8588 |
+
"step": 1221
|
| 8589 |
+
},
|
| 8590 |
+
{
|
| 8591 |
+
"epoch": 0.9075380616412922,
|
| 8592 |
+
"grad_norm": 0.0013827058719471097,
|
| 8593 |
+
"learning_rate": 2.2418200185938487e-06,
|
| 8594 |
+
"loss": 0.0002,
|
| 8595 |
+
"step": 1222
|
| 8596 |
+
},
|
| 8597 |
+
{
|
| 8598 |
+
"epoch": 0.9082807278128481,
|
| 8599 |
+
"grad_norm": 0.0026905061677098274,
|
| 8600 |
+
"learning_rate": 2.206074638853023e-06,
|
| 8601 |
+
"loss": 0.0002,
|
| 8602 |
+
"step": 1223
|
| 8603 |
+
},
|
| 8604 |
+
{
|
| 8605 |
+
"epoch": 0.909023393984404,
|
| 8606 |
+
"grad_norm": 0.001551009714603424,
|
| 8607 |
+
"learning_rate": 2.170610101224385e-06,
|
| 8608 |
+
"loss": 0.0001,
|
| 8609 |
+
"step": 1224
|
| 8610 |
+
},
|
| 8611 |
+
{
|
| 8612 |
+
"epoch": 0.9097660601559598,
|
| 8613 |
+
"grad_norm": 0.004318626597523689,
|
| 8614 |
+
"learning_rate": 2.1354266141012824e-06,
|
| 8615 |
+
"loss": 0.0002,
|
| 8616 |
+
"step": 1225
|
| 8617 |
+
},
|
| 8618 |
+
{
|
| 8619 |
+
"epoch": 0.9105087263275158,
|
| 8620 |
+
"grad_norm": 0.002215152606368065,
|
| 8621 |
+
"learning_rate": 2.100524384225555e-06,
|
| 8622 |
+
"loss": 0.0001,
|
| 8623 |
+
"step": 1226
|
| 8624 |
+
},
|
| 8625 |
+
{
|
| 8626 |
+
"epoch": 0.9112513924990717,
|
| 8627 |
+
"grad_norm": 0.011130552738904953,
|
| 8628 |
+
"learning_rate": 2.0659036166863766e-06,
|
| 8629 |
+
"loss": 0.0004,
|
| 8630 |
+
"step": 1227
|
| 8631 |
+
},
|
| 8632 |
+
{
|
| 8633 |
+
"epoch": 0.9119940586706275,
|
| 8634 |
+
"grad_norm": 0.0008930191397666931,
|
| 8635 |
+
"learning_rate": 2.0315645149189935e-06,
|
| 8636 |
+
"loss": 0.0001,
|
| 8637 |
+
"step": 1228
|
| 8638 |
+
},
|
| 8639 |
+
{
|
| 8640 |
+
"epoch": 0.9127367248421835,
|
| 8641 |
+
"grad_norm": 0.0026403944939374924,
|
| 8642 |
+
"learning_rate": 1.99750728070357e-06,
|
| 8643 |
+
"loss": 0.0002,
|
| 8644 |
+
"step": 1229
|
| 8645 |
+
},
|
| 8646 |
+
{
|
| 8647 |
+
"epoch": 0.9134793910137393,
|
| 8648 |
+
"grad_norm": 0.0009700857335701585,
|
| 8649 |
+
"learning_rate": 1.9637321141639743e-06,
|
| 8650 |
+
"loss": 0.0001,
|
| 8651 |
+
"step": 1230
|
| 8652 |
+
},
|
| 8653 |
+
{
|
| 8654 |
+
"epoch": 0.9142220571852953,
|
| 8655 |
+
"grad_norm": 0.045662786811590195,
|
| 8656 |
+
"learning_rate": 1.930239213766638e-06,
|
| 8657 |
+
"loss": 0.0009,
|
| 8658 |
+
"step": 1231
|
| 8659 |
+
},
|
| 8660 |
+
{
|
| 8661 |
+
"epoch": 0.9149647233568511,
|
| 8662 |
+
"grad_norm": 0.02511802315711975,
|
| 8663 |
+
"learning_rate": 1.8970287763193429e-06,
|
| 8664 |
+
"loss": 0.0007,
|
| 8665 |
+
"step": 1232
|
| 8666 |
+
},
|
| 8667 |
+
{
|
| 8668 |
+
"epoch": 0.915707389528407,
|
| 8669 |
+
"grad_norm": 0.001158213010057807,
|
| 8670 |
+
"learning_rate": 1.8641009969700917e-06,
|
| 8671 |
+
"loss": 0.0001,
|
| 8672 |
+
"step": 1233
|
| 8673 |
+
},
|
| 8674 |
+
{
|
| 8675 |
+
"epoch": 0.9164500556999629,
|
| 8676 |
+
"grad_norm": 0.00307093164883554,
|
| 8677 |
+
"learning_rate": 1.8314560692059835e-06,
|
| 8678 |
+
"loss": 0.0002,
|
| 8679 |
+
"step": 1234
|
| 8680 |
+
},
|
| 8681 |
+
{
|
| 8682 |
+
"epoch": 0.9171927218715188,
|
| 8683 |
+
"grad_norm": 0.0006863306043669581,
|
| 8684 |
+
"learning_rate": 1.799094184852018e-06,
|
| 8685 |
+
"loss": 0.0001,
|
| 8686 |
+
"step": 1235
|
| 8687 |
+
},
|
| 8688 |
+
{
|
| 8689 |
+
"epoch": 0.9179353880430746,
|
| 8690 |
+
"grad_norm": 0.001991459634155035,
|
| 8691 |
+
"learning_rate": 1.7670155340700323e-06,
|
| 8692 |
+
"loss": 0.0002,
|
| 8693 |
+
"step": 1236
|
| 8694 |
+
},
|
| 8695 |
+
{
|
| 8696 |
+
"epoch": 0.9186780542146306,
|
| 8697 |
+
"grad_norm": 0.0033494269009679556,
|
| 8698 |
+
"learning_rate": 1.7352203053575157e-06,
|
| 8699 |
+
"loss": 0.0003,
|
| 8700 |
+
"step": 1237
|
| 8701 |
+
},
|
| 8702 |
+
{
|
| 8703 |
+
"epoch": 0.9194207203861864,
|
| 8704 |
+
"grad_norm": 0.0011704204371199012,
|
| 8705 |
+
"learning_rate": 1.70370868554659e-06,
|
| 8706 |
+
"loss": 0.0001,
|
| 8707 |
+
"step": 1238
|
| 8708 |
+
},
|
| 8709 |
+
{
|
| 8710 |
+
"epoch": 0.9201633865577423,
|
| 8711 |
+
"grad_norm": 0.000724752142559737,
|
| 8712 |
+
"learning_rate": 1.6724808598028108e-06,
|
| 8713 |
+
"loss": 0.0001,
|
| 8714 |
+
"step": 1239
|
| 8715 |
+
},
|
| 8716 |
+
{
|
| 8717 |
+
"epoch": 0.9209060527292982,
|
| 8718 |
+
"grad_norm": 0.002084789564833045,
|
| 8719 |
+
"learning_rate": 1.6415370116241724e-06,
|
| 8720 |
+
"loss": 0.0001,
|
| 8721 |
+
"step": 1240
|
| 8722 |
+
},
|
| 8723 |
+
{
|
| 8724 |
+
"epoch": 0.9216487189008541,
|
| 8725 |
+
"grad_norm": 0.0011653484543785453,
|
| 8726 |
+
"learning_rate": 1.6108773228399543e-06,
|
| 8727 |
+
"loss": 0.0001,
|
| 8728 |
+
"step": 1241
|
| 8729 |
+
},
|
| 8730 |
+
{
|
| 8731 |
+
"epoch": 0.9223913850724099,
|
| 8732 |
+
"grad_norm": 0.001394697930663824,
|
| 8733 |
+
"learning_rate": 1.5805019736097104e-06,
|
| 8734 |
+
"loss": 0.0002,
|
| 8735 |
+
"step": 1242
|
| 8736 |
+
},
|
| 8737 |
+
{
|
| 8738 |
+
"epoch": 0.9231340512439659,
|
| 8739 |
+
"grad_norm": 0.0010144009720534086,
|
| 8740 |
+
"learning_rate": 1.5504111424221646e-06,
|
| 8741 |
+
"loss": 0.0001,
|
| 8742 |
+
"step": 1243
|
| 8743 |
+
},
|
| 8744 |
+
{
|
| 8745 |
+
"epoch": 0.9238767174155217,
|
| 8746 |
+
"grad_norm": 0.012748640961945057,
|
| 8747 |
+
"learning_rate": 1.5206050060942e-06,
|
| 8748 |
+
"loss": 0.0001,
|
| 8749 |
+
"step": 1244
|
| 8750 |
+
},
|
| 8751 |
+
{
|
| 8752 |
+
"epoch": 0.9246193835870776,
|
| 8753 |
+
"grad_norm": 0.0012531749671325088,
|
| 8754 |
+
"learning_rate": 1.4910837397697886e-06,
|
| 8755 |
+
"loss": 0.0001,
|
| 8756 |
+
"step": 1245
|
| 8757 |
+
},
|
| 8758 |
+
{
|
| 8759 |
+
"epoch": 0.9253620497586335,
|
| 8760 |
+
"grad_norm": 0.0027611353434622288,
|
| 8761 |
+
"learning_rate": 1.4618475169190017e-06,
|
| 8762 |
+
"loss": 0.0002,
|
| 8763 |
+
"step": 1246
|
| 8764 |
+
},
|
| 8765 |
+
{
|
| 8766 |
+
"epoch": 0.9261047159301894,
|
| 8767 |
+
"grad_norm": 0.0028265246655792,
|
| 8768 |
+
"learning_rate": 1.4328965093369283e-06,
|
| 8769 |
+
"loss": 0.0001,
|
| 8770 |
+
"step": 1247
|
| 8771 |
+
},
|
| 8772 |
+
{
|
| 8773 |
+
"epoch": 0.9268473821017452,
|
| 8774 |
+
"grad_norm": 0.0014653587713837624,
|
| 8775 |
+
"learning_rate": 1.4042308871427324e-06,
|
| 8776 |
+
"loss": 0.0001,
|
| 8777 |
+
"step": 1248
|
| 8778 |
+
},
|
| 8779 |
+
{
|
| 8780 |
+
"epoch": 0.9275900482733012,
|
| 8781 |
+
"grad_norm": 0.0027727442793548107,
|
| 8782 |
+
"learning_rate": 1.3758508187786068e-06,
|
| 8783 |
+
"loss": 0.0002,
|
| 8784 |
+
"step": 1249
|
| 8785 |
+
},
|
| 8786 |
+
{
|
| 8787 |
+
"epoch": 0.928332714444857,
|
| 8788 |
+
"grad_norm": 0.003438390791416168,
|
| 8789 |
+
"learning_rate": 1.3477564710088098e-06,
|
| 8790 |
+
"loss": 0.0002,
|
| 8791 |
+
"step": 1250
|
| 8792 |
+
},
|
| 8793 |
+
{
|
| 8794 |
+
"epoch": 0.9290753806164129,
|
| 8795 |
+
"grad_norm": 0.0032505146227777004,
|
| 8796 |
+
"learning_rate": 1.3199480089186756e-06,
|
| 8797 |
+
"loss": 0.0002,
|
| 8798 |
+
"step": 1251
|
| 8799 |
+
},
|
| 8800 |
+
{
|
| 8801 |
+
"epoch": 0.9298180467879688,
|
| 8802 |
+
"grad_norm": 0.005036122631281614,
|
| 8803 |
+
"learning_rate": 1.2924255959136266e-06,
|
| 8804 |
+
"loss": 0.0002,
|
| 8805 |
+
"step": 1252
|
| 8806 |
+
},
|
| 8807 |
+
{
|
| 8808 |
+
"epoch": 0.9305607129595247,
|
| 8809 |
+
"grad_norm": 0.011345870792865753,
|
| 8810 |
+
"learning_rate": 1.2651893937182635e-06,
|
| 8811 |
+
"loss": 0.0003,
|
| 8812 |
+
"step": 1253
|
| 8813 |
+
},
|
| 8814 |
+
{
|
| 8815 |
+
"epoch": 0.9313033791310806,
|
| 8816 |
+
"grad_norm": 0.0007504152599722147,
|
| 8817 |
+
"learning_rate": 1.2382395623753485e-06,
|
| 8818 |
+
"loss": 0.0001,
|
| 8819 |
+
"step": 1254
|
| 8820 |
+
},
|
| 8821 |
+
{
|
| 8822 |
+
"epoch": 0.9320460453026365,
|
| 8823 |
+
"grad_norm": 0.0009204870439134538,
|
| 8824 |
+
"learning_rate": 1.2115762602449177e-06,
|
| 8825 |
+
"loss": 0.0001,
|
| 8826 |
+
"step": 1255
|
| 8827 |
+
},
|
| 8828 |
+
{
|
| 8829 |
+
"epoch": 0.9327887114741924,
|
| 8830 |
+
"grad_norm": 0.021037423983216286,
|
| 8831 |
+
"learning_rate": 1.1851996440033319e-06,
|
| 8832 |
+
"loss": 0.0007,
|
| 8833 |
+
"step": 1256
|
| 8834 |
+
},
|
| 8835 |
+
{
|
| 8836 |
+
"epoch": 0.9335313776457482,
|
| 8837 |
+
"grad_norm": 0.05482451990246773,
|
| 8838 |
+
"learning_rate": 1.1591098686423496e-06,
|
| 8839 |
+
"loss": 0.0006,
|
| 8840 |
+
"step": 1257
|
| 8841 |
+
},
|
| 8842 |
+
{
|
| 8843 |
+
"epoch": 0.9342740438173042,
|
| 8844 |
+
"grad_norm": 0.0006609958363696933,
|
| 8845 |
+
"learning_rate": 1.1333070874682216e-06,
|
| 8846 |
+
"loss": 0.0001,
|
| 8847 |
+
"step": 1258
|
| 8848 |
+
},
|
| 8849 |
+
{
|
| 8850 |
+
"epoch": 0.93501670998886,
|
| 8851 |
+
"grad_norm": 0.0011097380192950368,
|
| 8852 |
+
"learning_rate": 1.1077914521007981e-06,
|
| 8853 |
+
"loss": 0.0002,
|
| 8854 |
+
"step": 1259
|
| 8855 |
+
},
|
| 8856 |
+
{
|
| 8857 |
+
"epoch": 0.9357593761604159,
|
| 8858 |
+
"grad_norm": 0.0019438870949670672,
|
| 8859 |
+
"learning_rate": 1.0825631124726232e-06,
|
| 8860 |
+
"loss": 0.0001,
|
| 8861 |
+
"step": 1260
|
| 8862 |
+
},
|
| 8863 |
+
{
|
| 8864 |
+
"epoch": 0.9365020423319718,
|
| 8865 |
+
"grad_norm": 0.0007544804248027503,
|
| 8866 |
+
"learning_rate": 1.0576222168280636e-06,
|
| 8867 |
+
"loss": 0.0001,
|
| 8868 |
+
"step": 1261
|
| 8869 |
+
},
|
| 8870 |
+
{
|
| 8871 |
+
"epoch": 0.9372447085035277,
|
| 8872 |
+
"grad_norm": 0.007262110244482756,
|
| 8873 |
+
"learning_rate": 1.0329689117224262e-06,
|
| 8874 |
+
"loss": 0.0004,
|
| 8875 |
+
"step": 1262
|
| 8876 |
+
},
|
| 8877 |
+
{
|
| 8878 |
+
"epoch": 0.9379873746750835,
|
| 8879 |
+
"grad_norm": 0.0008670915849506855,
|
| 8880 |
+
"learning_rate": 1.008603342021125e-06,
|
| 8881 |
+
"loss": 0.0001,
|
| 8882 |
+
"step": 1263
|
| 8883 |
+
},
|
| 8884 |
+
{
|
| 8885 |
+
"epoch": 0.9387300408466395,
|
| 8886 |
+
"grad_norm": 0.0009119977476075292,
|
| 8887 |
+
"learning_rate": 9.845256508987989e-07,
|
| 8888 |
+
"loss": 0.0001,
|
| 8889 |
+
"step": 1264
|
| 8890 |
+
},
|
| 8891 |
+
{
|
| 8892 |
+
"epoch": 0.9394727070181953,
|
| 8893 |
+
"grad_norm": 0.00047243438893929124,
|
| 8894 |
+
"learning_rate": 9.607359798384785e-07,
|
| 8895 |
+
"loss": 0.0001,
|
| 8896 |
+
"step": 1265
|
| 8897 |
+
},
|
| 8898 |
+
{
|
| 8899 |
+
"epoch": 0.9402153731897512,
|
| 8900 |
+
"grad_norm": 0.0028327691834419966,
|
| 8901 |
+
"learning_rate": 9.372344686307655e-07,
|
| 8902 |
+
"loss": 0.0002,
|
| 8903 |
+
"step": 1266
|
| 8904 |
+
},
|
| 8905 |
+
{
|
| 8906 |
+
"epoch": 0.9409580393613071,
|
| 8907 |
+
"grad_norm": 0.044161271303892136,
|
| 8908 |
+
"learning_rate": 9.140212553730043e-07,
|
| 8909 |
+
"loss": 0.0016,
|
| 8910 |
+
"step": 1267
|
| 8911 |
+
},
|
| 8912 |
+
{
|
| 8913 |
+
"epoch": 0.941700705532863,
|
| 8914 |
+
"grad_norm": 0.3684042692184448,
|
| 8915 |
+
"learning_rate": 8.910964764684671e-07,
|
| 8916 |
+
"loss": 0.0089,
|
| 8917 |
+
"step": 1268
|
| 8918 |
+
},
|
| 8919 |
+
{
|
| 8920 |
+
"epoch": 0.9424433717044188,
|
| 8921 |
+
"grad_norm": 0.002094242488965392,
|
| 8922 |
+
"learning_rate": 8.684602666255481e-07,
|
| 8923 |
+
"loss": 0.0001,
|
| 8924 |
+
"step": 1269
|
| 8925 |
+
},
|
| 8926 |
+
{
|
| 8927 |
+
"epoch": 0.9431860378759748,
|
| 8928 |
+
"grad_norm": 0.0017989203333854675,
|
| 8929 |
+
"learning_rate": 8.46112758857004e-07,
|
| 8930 |
+
"loss": 0.0002,
|
| 8931 |
+
"step": 1270
|
| 8932 |
+
},
|
| 8933 |
+
{
|
| 8934 |
+
"epoch": 0.9439287040475306,
|
| 8935 |
+
"grad_norm": 0.0011526981834322214,
|
| 8936 |
+
"learning_rate": 8.240540844791145e-07,
|
| 8937 |
+
"loss": 0.0001,
|
| 8938 |
+
"step": 1271
|
| 8939 |
+
},
|
| 8940 |
+
{
|
| 8941 |
+
"epoch": 0.9446713702190865,
|
| 8942 |
+
"grad_norm": 0.0008104105945676565,
|
| 8943 |
+
"learning_rate": 8.022843731109674e-07,
|
| 8944 |
+
"loss": 0.0001,
|
| 8945 |
+
"step": 1272
|
| 8946 |
+
},
|
| 8947 |
+
{
|
| 8948 |
+
"epoch": 0.9454140363906424,
|
| 8949 |
+
"grad_norm": 0.0018936687847599387,
|
| 8950 |
+
"learning_rate": 7.8080375267367e-07,
|
| 8951 |
+
"loss": 0.0002,
|
| 8952 |
+
"step": 1273
|
| 8953 |
+
},
|
| 8954 |
+
{
|
| 8955 |
+
"epoch": 0.9461567025621983,
|
| 8956 |
+
"grad_norm": 0.0015682062366977334,
|
| 8957 |
+
"learning_rate": 7.596123493895991e-07,
|
| 8958 |
+
"loss": 0.0001,
|
| 8959 |
+
"step": 1274
|
| 8960 |
+
},
|
| 8961 |
+
{
|
| 8962 |
+
"epoch": 0.9468993687337541,
|
| 8963 |
+
"grad_norm": 0.0024748966097831726,
|
| 8964 |
+
"learning_rate": 7.387102877816521e-07,
|
| 8965 |
+
"loss": 0.0002,
|
| 8966 |
+
"step": 1275
|
| 8967 |
+
},
|
| 8968 |
+
{
|
| 8969 |
+
"epoch": 0.9476420349053101,
|
| 8970 |
+
"grad_norm": 0.0018025932367891073,
|
| 8971 |
+
"learning_rate": 7.180976906725423e-07,
|
| 8972 |
+
"loss": 0.0001,
|
| 8973 |
+
"step": 1276
|
| 8974 |
+
},
|
| 8975 |
+
{
|
| 8976 |
+
"epoch": 0.9483847010768659,
|
| 8977 |
+
"grad_norm": 0.0015493928221985698,
|
| 8978 |
+
"learning_rate": 6.977746791840434e-07,
|
| 8979 |
+
"loss": 0.0002,
|
| 8980 |
+
"step": 1277
|
| 8981 |
+
},
|
| 8982 |
+
{
|
| 8983 |
+
"epoch": 0.9491273672484218,
|
| 8984 |
+
"grad_norm": 0.004556954838335514,
|
| 8985 |
+
"learning_rate": 6.777413727363069e-07,
|
| 8986 |
+
"loss": 0.0002,
|
| 8987 |
+
"step": 1278
|
| 8988 |
+
},
|
| 8989 |
+
{
|
| 8990 |
+
"epoch": 0.9498700334199777,
|
| 8991 |
+
"grad_norm": 0.018575742840766907,
|
| 8992 |
+
"learning_rate": 6.579978890471294e-07,
|
| 8993 |
+
"loss": 0.0003,
|
| 8994 |
+
"step": 1279
|
| 8995 |
+
},
|
| 8996 |
+
{
|
| 8997 |
+
"epoch": 0.9506126995915336,
|
| 8998 |
+
"grad_norm": 0.002247869735583663,
|
| 8999 |
+
"learning_rate": 6.385443441312978e-07,
|
| 9000 |
+
"loss": 0.0002,
|
| 9001 |
+
"step": 1280
|
| 9002 |
+
},
|
| 9003 |
+
{
|
| 9004 |
+
"epoch": 0.9513553657630895,
|
| 9005 |
+
"grad_norm": 0.0014160757418721914,
|
| 9006 |
+
"learning_rate": 6.193808522998723e-07,
|
| 9007 |
+
"loss": 0.0001,
|
| 9008 |
+
"step": 1281
|
| 9009 |
+
},
|
| 9010 |
+
{
|
| 9011 |
+
"epoch": 0.9520980319346454,
|
| 9012 |
+
"grad_norm": 0.0015209565171971917,
|
| 9013 |
+
"learning_rate": 6.005075261595494e-07,
|
| 9014 |
+
"loss": 0.0002,
|
| 9015 |
+
"step": 1282
|
| 9016 |
+
},
|
| 9017 |
+
{
|
| 9018 |
+
"epoch": 0.9528406981062013,
|
| 9019 |
+
"grad_norm": 0.0009286428103223443,
|
| 9020 |
+
"learning_rate": 5.81924476611967e-07,
|
| 9021 |
+
"loss": 0.0001,
|
| 9022 |
+
"step": 1283
|
| 9023 |
+
},
|
| 9024 |
+
{
|
| 9025 |
+
"epoch": 0.9535833642777571,
|
| 9026 |
+
"grad_norm": 0.013380018062889576,
|
| 9027 |
+
"learning_rate": 5.63631812853066e-07,
|
| 9028 |
+
"loss": 0.0002,
|
| 9029 |
+
"step": 1284
|
| 9030 |
+
},
|
| 9031 |
+
{
|
| 9032 |
+
"epoch": 0.9543260304493131,
|
| 9033 |
+
"grad_norm": 0.00046919341548345983,
|
| 9034 |
+
"learning_rate": 5.456296423724527e-07,
|
| 9035 |
+
"loss": 0.0001,
|
| 9036 |
+
"step": 1285
|
| 9037 |
+
},
|
| 9038 |
+
{
|
| 9039 |
+
"epoch": 0.9550686966208689,
|
| 9040 |
+
"grad_norm": 0.001202006940729916,
|
| 9041 |
+
"learning_rate": 5.279180709527765e-07,
|
| 9042 |
+
"loss": 0.0001,
|
| 9043 |
+
"step": 1286
|
| 9044 |
+
},
|
| 9045 |
+
{
|
| 9046 |
+
"epoch": 0.9558113627924248,
|
| 9047 |
+
"grad_norm": 0.0014856454217806458,
|
| 9048 |
+
"learning_rate": 5.104972026690802e-07,
|
| 9049 |
+
"loss": 0.0001,
|
| 9050 |
+
"step": 1287
|
| 9051 |
+
},
|
| 9052 |
+
{
|
| 9053 |
+
"epoch": 0.9565540289639807,
|
| 9054 |
+
"grad_norm": 0.0008986307075247169,
|
| 9055 |
+
"learning_rate": 4.933671398882067e-07,
|
| 9056 |
+
"loss": 0.0001,
|
| 9057 |
+
"step": 1288
|
| 9058 |
+
},
|
| 9059 |
+
{
|
| 9060 |
+
"epoch": 0.9572966951355366,
|
| 9061 |
+
"grad_norm": 0.0012058410793542862,
|
| 9062 |
+
"learning_rate": 4.765279832682101e-07,
|
| 9063 |
+
"loss": 0.0001,
|
| 9064 |
+
"step": 1289
|
| 9065 |
+
},
|
| 9066 |
+
{
|
| 9067 |
+
"epoch": 0.9580393613070924,
|
| 9068 |
+
"grad_norm": 0.003975100349634886,
|
| 9069 |
+
"learning_rate": 4.5997983175773417e-07,
|
| 9070 |
+
"loss": 0.0002,
|
| 9071 |
+
"step": 1290
|
| 9072 |
+
},
|
| 9073 |
+
{
|
| 9074 |
+
"epoch": 0.9587820274786484,
|
| 9075 |
+
"grad_norm": 0.003919736947864294,
|
| 9076 |
+
"learning_rate": 4.4372278259545155e-07,
|
| 9077 |
+
"loss": 0.0003,
|
| 9078 |
+
"step": 1291
|
| 9079 |
+
},
|
| 9080 |
+
{
|
| 9081 |
+
"epoch": 0.9595246936502042,
|
| 9082 |
+
"grad_norm": 0.019034747034311295,
|
| 9083 |
+
"learning_rate": 4.277569313094809e-07,
|
| 9084 |
+
"loss": 0.0006,
|
| 9085 |
+
"step": 1292
|
| 9086 |
+
},
|
| 9087 |
+
{
|
| 9088 |
+
"epoch": 0.9602673598217601,
|
| 9089 |
+
"grad_norm": 0.004752116743475199,
|
| 9090 |
+
"learning_rate": 4.120823717168432e-07,
|
| 9091 |
+
"loss": 0.0002,
|
| 9092 |
+
"step": 1293
|
| 9093 |
+
},
|
| 9094 |
+
{
|
| 9095 |
+
"epoch": 0.961010025993316,
|
| 9096 |
+
"grad_norm": 0.0008414981421083212,
|
| 9097 |
+
"learning_rate": 3.9669919592288384e-07,
|
| 9098 |
+
"loss": 0.0001,
|
| 9099 |
+
"step": 1294
|
| 9100 |
+
},
|
| 9101 |
+
{
|
| 9102 |
+
"epoch": 0.9617526921648719,
|
| 9103 |
+
"grad_norm": 0.0018463104497641325,
|
| 9104 |
+
"learning_rate": 3.8160749432076815e-07,
|
| 9105 |
+
"loss": 0.0002,
|
| 9106 |
+
"step": 1295
|
| 9107 |
+
},
|
| 9108 |
+
{
|
| 9109 |
+
"epoch": 0.9624953583364277,
|
| 9110 |
+
"grad_norm": 0.0009904575999826193,
|
| 9111 |
+
"learning_rate": 3.6680735559090906e-07,
|
| 9112 |
+
"loss": 0.0001,
|
| 9113 |
+
"step": 1296
|
| 9114 |
+
},
|
| 9115 |
+
{
|
| 9116 |
+
"epoch": 0.9632380245079837,
|
| 9117 |
+
"grad_norm": 0.001538589014671743,
|
| 9118 |
+
"learning_rate": 3.5229886670046783e-07,
|
| 9119 |
+
"loss": 0.0002,
|
| 9120 |
+
"step": 1297
|
| 9121 |
+
},
|
| 9122 |
+
{
|
| 9123 |
+
"epoch": 0.9639806906795395,
|
| 9124 |
+
"grad_norm": 0.0009757834486663342,
|
| 9125 |
+
"learning_rate": 3.380821129028489e-07,
|
| 9126 |
+
"loss": 0.0001,
|
| 9127 |
+
"step": 1298
|
| 9128 |
+
},
|
| 9129 |
+
{
|
| 9130 |
+
"epoch": 0.9647233568510954,
|
| 9131 |
+
"grad_norm": 0.00236718263477087,
|
| 9132 |
+
"learning_rate": 3.241571777371888e-07,
|
| 9133 |
+
"loss": 0.0002,
|
| 9134 |
+
"step": 1299
|
| 9135 |
+
},
|
| 9136 |
+
{
|
| 9137 |
+
"epoch": 0.9654660230226513,
|
| 9138 |
+
"grad_norm": 0.00395069969817996,
|
| 9139 |
+
"learning_rate": 3.1052414302786823e-07,
|
| 9140 |
+
"loss": 0.0002,
|
| 9141 |
+
"step": 1300
|
| 9142 |
+
},
|
| 9143 |
+
{
|
| 9144 |
+
"epoch": 0.9662086891942072,
|
| 9145 |
+
"grad_norm": 0.006602929905056953,
|
| 9146 |
+
"learning_rate": 2.971830888840177e-07,
|
| 9147 |
+
"loss": 0.0003,
|
| 9148 |
+
"step": 1301
|
| 9149 |
+
},
|
| 9150 |
+
{
|
| 9151 |
+
"epoch": 0.966951355365763,
|
| 9152 |
+
"grad_norm": 0.0023442876990884542,
|
| 9153 |
+
"learning_rate": 2.8413409369907886e-07,
|
| 9154 |
+
"loss": 0.0002,
|
| 9155 |
+
"step": 1302
|
| 9156 |
+
},
|
| 9157 |
+
{
|
| 9158 |
+
"epoch": 0.967694021537319,
|
| 9159 |
+
"grad_norm": 0.012941988185048103,
|
| 9160 |
+
"learning_rate": 2.7137723415030516e-07,
|
| 9161 |
+
"loss": 0.0003,
|
| 9162 |
+
"step": 1303
|
| 9163 |
+
},
|
| 9164 |
+
{
|
| 9165 |
+
"epoch": 0.9684366877088748,
|
| 9166 |
+
"grad_norm": 0.005490366835147142,
|
| 9167 |
+
"learning_rate": 2.589125851983509e-07,
|
| 9168 |
+
"loss": 0.0002,
|
| 9169 |
+
"step": 1304
|
| 9170 |
+
},
|
| 9171 |
+
{
|
| 9172 |
+
"epoch": 0.9691793538804307,
|
| 9173 |
+
"grad_norm": 0.0007061643409542739,
|
| 9174 |
+
"learning_rate": 2.46740220086783e-07,
|
| 9175 |
+
"loss": 0.0001,
|
| 9176 |
+
"step": 1305
|
| 9177 |
+
},
|
| 9178 |
+
{
|
| 9179 |
+
"epoch": 0.9699220200519866,
|
| 9180 |
+
"grad_norm": 0.0017634638352319598,
|
| 9181 |
+
"learning_rate": 2.3486021034170857e-07,
|
| 9182 |
+
"loss": 0.0002,
|
| 9183 |
+
"step": 1306
|
| 9184 |
+
},
|
| 9185 |
+
{
|
| 9186 |
+
"epoch": 0.9706646862235425,
|
| 9187 |
+
"grad_norm": 0.0028723692521452904,
|
| 9188 |
+
"learning_rate": 2.232726257713036e-07,
|
| 9189 |
+
"loss": 0.0002,
|
| 9190 |
+
"step": 1307
|
| 9191 |
+
},
|
| 9192 |
+
{
|
| 9193 |
+
"epoch": 0.9714073523950985,
|
| 9194 |
+
"grad_norm": 0.0016753628151491284,
|
| 9195 |
+
"learning_rate": 2.1197753446542957e-07,
|
| 9196 |
+
"loss": 0.0001,
|
| 9197 |
+
"step": 1308
|
| 9198 |
+
},
|
| 9199 |
+
{
|
| 9200 |
+
"epoch": 0.9721500185666543,
|
| 9201 |
+
"grad_norm": 0.02156355232000351,
|
| 9202 |
+
"learning_rate": 2.009750027952284e-07,
|
| 9203 |
+
"loss": 0.0007,
|
| 9204 |
+
"step": 1309
|
| 9205 |
+
},
|
| 9206 |
+
{
|
| 9207 |
+
"epoch": 0.9728926847382102,
|
| 9208 |
+
"grad_norm": 0.0010258769616484642,
|
| 9209 |
+
"learning_rate": 1.9026509541272275e-07,
|
| 9210 |
+
"loss": 0.0001,
|
| 9211 |
+
"step": 1310
|
| 9212 |
+
},
|
| 9213 |
+
{
|
| 9214 |
+
"epoch": 0.973635350909766,
|
| 9215 |
+
"grad_norm": 0.001431930111721158,
|
| 9216 |
+
"learning_rate": 1.798478752504551e-07,
|
| 9217 |
+
"loss": 0.0001,
|
| 9218 |
+
"step": 1311
|
| 9219 |
+
},
|
| 9220 |
+
{
|
| 9221 |
+
"epoch": 0.974378017081322,
|
| 9222 |
+
"grad_norm": 0.7691323161125183,
|
| 9223 |
+
"learning_rate": 1.6972340352110483e-07,
|
| 9224 |
+
"loss": 0.0318,
|
| 9225 |
+
"step": 1312
|
| 9226 |
+
},
|
| 9227 |
+
{
|
| 9228 |
+
"epoch": 0.9751206832528778,
|
| 9229 |
+
"grad_norm": 0.005196771118789911,
|
| 9230 |
+
"learning_rate": 1.598917397171218e-07,
|
| 9231 |
+
"loss": 0.0003,
|
| 9232 |
+
"step": 1313
|
| 9233 |
+
},
|
| 9234 |
+
{
|
| 9235 |
+
"epoch": 0.9758633494244338,
|
| 9236 |
+
"grad_norm": 0.0011321887141093612,
|
| 9237 |
+
"learning_rate": 1.503529416103988e-07,
|
| 9238 |
+
"loss": 0.0002,
|
| 9239 |
+
"step": 1314
|
| 9240 |
+
},
|
| 9241 |
+
{
|
| 9242 |
+
"epoch": 0.9766060155959896,
|
| 9243 |
+
"grad_norm": 0.0005147146875970066,
|
| 9244 |
+
"learning_rate": 1.4110706525190531e-07,
|
| 9245 |
+
"loss": 0.0001,
|
| 9246 |
+
"step": 1315
|
| 9247 |
+
},
|
| 9248 |
+
{
|
| 9249 |
+
"epoch": 0.9773486817675455,
|
| 9250 |
+
"grad_norm": 0.008908793330192566,
|
| 9251 |
+
"learning_rate": 1.3215416497138754e-07,
|
| 9252 |
+
"loss": 0.0003,
|
| 9253 |
+
"step": 1316
|
| 9254 |
+
},
|
| 9255 |
+
{
|
| 9256 |
+
"epoch": 0.9780913479391014,
|
| 9257 |
+
"grad_norm": 0.002536405110731721,
|
| 9258 |
+
"learning_rate": 1.234942933770189e-07,
|
| 9259 |
+
"loss": 0.0002,
|
| 9260 |
+
"step": 1317
|
| 9261 |
+
},
|
| 9262 |
+
{
|
| 9263 |
+
"epoch": 0.9788340141106573,
|
| 9264 |
+
"grad_norm": 0.0017400149954482913,
|
| 9265 |
+
"learning_rate": 1.1512750135511674e-07,
|
| 9266 |
+
"loss": 0.0001,
|
| 9267 |
+
"step": 1318
|
| 9268 |
+
},
|
| 9269 |
+
{
|
| 9270 |
+
"epoch": 0.9795766802822131,
|
| 9271 |
+
"grad_norm": 0.0011383086675778031,
|
| 9272 |
+
"learning_rate": 1.0705383806982606e-07,
|
| 9273 |
+
"loss": 0.0001,
|
| 9274 |
+
"step": 1319
|
| 9275 |
+
},
|
| 9276 |
+
{
|
| 9277 |
+
"epoch": 0.9803193464537691,
|
| 9278 |
+
"grad_norm": 0.0018386361189186573,
|
| 9279 |
+
"learning_rate": 9.927335096283629e-08,
|
| 9280 |
+
"loss": 0.0002,
|
| 9281 |
+
"step": 1320
|
| 9282 |
+
},
|
| 9283 |
+
{
|
| 9284 |
+
"epoch": 0.9810620126253249,
|
| 9285 |
+
"grad_norm": 0.0018964792834594846,
|
| 9286 |
+
"learning_rate": 9.178608575310388e-08,
|
| 9287 |
+
"loss": 0.0001,
|
| 9288 |
+
"step": 1321
|
| 9289 |
+
},
|
| 9290 |
+
{
|
| 9291 |
+
"epoch": 0.9818046787968808,
|
| 9292 |
+
"grad_norm": 0.001377547043375671,
|
| 9293 |
+
"learning_rate": 8.459208643659122e-08,
|
| 9294 |
+
"loss": 0.0001,
|
| 9295 |
+
"step": 1322
|
| 9296 |
+
},
|
| 9297 |
+
{
|
| 9298 |
+
"epoch": 0.9825473449684367,
|
| 9299 |
+
"grad_norm": 0.0015313441399484873,
|
| 9300 |
+
"learning_rate": 7.769139528598368e-08,
|
| 9301 |
+
"loss": 0.0001,
|
| 9302 |
+
"step": 1323
|
| 9303 |
+
},
|
| 9304 |
+
{
|
| 9305 |
+
"epoch": 0.9832900111399926,
|
| 9306 |
+
"grad_norm": 0.00252739991992712,
|
| 9307 |
+
"learning_rate": 7.108405285046749e-08,
|
| 9308 |
+
"loss": 0.0002,
|
| 9309 |
+
"step": 1324
|
| 9310 |
+
},
|
| 9311 |
+
{
|
| 9312 |
+
"epoch": 0.9840326773115484,
|
| 9313 |
+
"grad_norm": 0.0026342032942920923,
|
| 9314 |
+
"learning_rate": 6.477009795547995e-08,
|
| 9315 |
+
"loss": 0.0002,
|
| 9316 |
+
"step": 1325
|
| 9317 |
+
},
|
| 9318 |
+
{
|
| 9319 |
+
"epoch": 0.9847753434831044,
|
| 9320 |
+
"grad_norm": 0.0036653559654951096,
|
| 9321 |
+
"learning_rate": 5.8749567702481856e-08,
|
| 9322 |
+
"loss": 0.0002,
|
| 9323 |
+
"step": 1326
|
| 9324 |
+
},
|
| 9325 |
+
{
|
| 9326 |
+
"epoch": 0.9855180096546602,
|
| 9327 |
+
"grad_norm": 0.031000560149550438,
|
| 9328 |
+
"learning_rate": 5.3022497468735446e-08,
|
| 9329 |
+
"loss": 0.0012,
|
| 9330 |
+
"step": 1327
|
| 9331 |
+
},
|
| 9332 |
+
{
|
| 9333 |
+
"epoch": 0.9862606758262161,
|
| 9334 |
+
"grad_norm": 0.01462267991155386,
|
| 9335 |
+
"learning_rate": 4.7588920907110094e-08,
|
| 9336 |
+
"loss": 0.0005,
|
| 9337 |
+
"step": 1328
|
| 9338 |
+
},
|
| 9339 |
+
{
|
| 9340 |
+
"epoch": 0.987003341997772,
|
| 9341 |
+
"grad_norm": 0.0009502037428319454,
|
| 9342 |
+
"learning_rate": 4.244886994587138e-08,
|
| 9343 |
+
"loss": 0.0001,
|
| 9344 |
+
"step": 1329
|
| 9345 |
+
},
|
| 9346 |
+
{
|
| 9347 |
+
"epoch": 0.9877460081693279,
|
| 9348 |
+
"grad_norm": 0.0015973957488313317,
|
| 9349 |
+
"learning_rate": 3.760237478849793e-08,
|
| 9350 |
+
"loss": 0.0001,
|
| 9351 |
+
"step": 1330
|
| 9352 |
+
},
|
| 9353 |
+
{
|
| 9354 |
+
"epoch": 0.9884886743408837,
|
| 9355 |
+
"grad_norm": 0.0013975318288430572,
|
| 9356 |
+
"learning_rate": 3.3049463913498166e-08,
|
| 9357 |
+
"loss": 0.0001,
|
| 9358 |
+
"step": 1331
|
| 9359 |
+
},
|
| 9360 |
+
{
|
| 9361 |
+
"epoch": 0.9892313405124397,
|
| 9362 |
+
"grad_norm": 0.06534797698259354,
|
| 9363 |
+
"learning_rate": 2.8790164074254943e-08,
|
| 9364 |
+
"loss": 0.0017,
|
| 9365 |
+
"step": 1332
|
| 9366 |
+
},
|
| 9367 |
+
{
|
| 9368 |
+
"epoch": 0.9899740066839955,
|
| 9369 |
+
"grad_norm": 0.000878207094501704,
|
| 9370 |
+
"learning_rate": 2.4824500298858967e-08,
|
| 9371 |
+
"loss": 0.0001,
|
| 9372 |
+
"step": 1333
|
| 9373 |
+
},
|
| 9374 |
+
{
|
| 9375 |
+
"epoch": 0.9907166728555514,
|
| 9376 |
+
"grad_norm": 0.0011413278989493847,
|
| 9377 |
+
"learning_rate": 2.1152495889970035e-08,
|
| 9378 |
+
"loss": 0.0001,
|
| 9379 |
+
"step": 1334
|
| 9380 |
+
},
|
| 9381 |
+
{
|
| 9382 |
+
"epoch": 0.9914593390271074,
|
| 9383 |
+
"grad_norm": 0.0010686512105166912,
|
| 9384 |
+
"learning_rate": 1.777417242467272e-08,
|
| 9385 |
+
"loss": 0.0001,
|
| 9386 |
+
"step": 1335
|
| 9387 |
+
},
|
| 9388 |
+
{
|
| 9389 |
+
"epoch": 0.9922020051986632,
|
| 9390 |
+
"grad_norm": 0.001142362249083817,
|
| 9391 |
+
"learning_rate": 1.4689549754337562e-08,
|
| 9392 |
+
"loss": 0.0001,
|
| 9393 |
+
"step": 1336
|
| 9394 |
+
},
|
| 9395 |
+
{
|
| 9396 |
+
"epoch": 0.9929446713702191,
|
| 9397 |
+
"grad_norm": 0.0008279301691800356,
|
| 9398 |
+
"learning_rate": 1.189864600454338e-08,
|
| 9399 |
+
"loss": 0.0001,
|
| 9400 |
+
"step": 1337
|
| 9401 |
+
},
|
| 9402 |
+
{
|
| 9403 |
+
"epoch": 0.993687337541775,
|
| 9404 |
+
"grad_norm": 0.0007734330138191581,
|
| 9405 |
+
"learning_rate": 9.401477574932926e-09,
|
| 9406 |
+
"loss": 0.0001,
|
| 9407 |
+
"step": 1338
|
| 9408 |
+
},
|
| 9409 |
+
{
|
| 9410 |
+
"epoch": 0.9944300037133309,
|
| 9411 |
+
"grad_norm": 0.004689542576670647,
|
| 9412 |
+
"learning_rate": 7.198059139118529e-09,
|
| 9413 |
+
"loss": 0.0002,
|
| 9414 |
+
"step": 1339
|
| 9415 |
+
},
|
| 9416 |
+
{
|
| 9417 |
+
"epoch": 0.9951726698848867,
|
| 9418 |
+
"grad_norm": 0.0011075050570070744,
|
| 9419 |
+
"learning_rate": 5.2884036446265714e-09,
|
| 9420 |
+
"loss": 0.0001,
|
| 9421 |
+
"step": 1340
|
| 9422 |
+
},
|
| 9423 |
+
{
|
| 9424 |
+
"epoch": 0.9959153360564427,
|
| 9425 |
+
"grad_norm": 0.0022147931158542633,
|
| 9426 |
+
"learning_rate": 3.672522312786475e-09,
|
| 9427 |
+
"loss": 0.0002,
|
| 9428 |
+
"step": 1341
|
| 9429 |
+
},
|
| 9430 |
+
{
|
| 9431 |
+
"epoch": 0.9966580022279985,
|
| 9432 |
+
"grad_norm": 0.0008529072511009872,
|
| 9433 |
+
"learning_rate": 2.3504246386918393e-09,
|
| 9434 |
+
"loss": 0.0001,
|
| 9435 |
+
"step": 1342
|
| 9436 |
+
},
|
| 9437 |
+
{
|
| 9438 |
+
"epoch": 0.9974006683995544,
|
| 9439 |
+
"grad_norm": 0.0028021601028740406,
|
| 9440 |
+
"learning_rate": 1.3221183911282798e-09,
|
| 9441 |
+
"loss": 0.0002,
|
| 9442 |
+
"step": 1343
|
| 9443 |
+
},
|
| 9444 |
+
{
|
| 9445 |
+
"epoch": 0.9981433345711103,
|
| 9446 |
+
"grad_norm": 0.0028829630464315414,
|
| 9447 |
+
"learning_rate": 5.876096125234653e-10,
|
| 9448 |
+
"loss": 0.0001,
|
| 9449 |
+
"step": 1344
|
| 9450 |
+
},
|
| 9451 |
+
{
|
| 9452 |
+
"epoch": 0.9988860007426662,
|
| 9453 |
+
"grad_norm": 0.002654945245012641,
|
| 9454 |
+
"learning_rate": 1.4690261893601786e-10,
|
| 9455 |
+
"loss": 0.0002,
|
| 9456 |
+
"step": 1345
|
| 9457 |
+
},
|
| 9458 |
+
{
|
| 9459 |
+
"epoch": 0.999628666914222,
|
| 9460 |
+
"grad_norm": 0.0032625223975628614,
|
| 9461 |
+
"learning_rate": 0.0,
|
| 9462 |
+
"loss": 0.0001,
|
| 9463 |
+
"step": 1346
|
| 9464 |
}
|
| 9465 |
],
|
| 9466 |
"logging_steps": 1,
|
|
|
|
| 9475 |
"should_evaluate": false,
|
| 9476 |
"should_log": false,
|
| 9477 |
"should_save": true,
|
| 9478 |
+
"should_training_stop": true
|
| 9479 |
},
|
| 9480 |
"attributes": {}
|
| 9481 |
}
|
| 9482 |
},
|
| 9483 |
+
"total_flos": 1.7591268226010972e+18,
|
| 9484 |
"train_batch_size": 1,
|
| 9485 |
"trial_name": null,
|
| 9486 |
"trial_params": null
|