Training in progress, step 801, checkpoint
Browse files
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 101184122
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c51a4ea12d5db42e949ec42e890dcb5292b3e9117b9b9eedcae3ea31bc13a1e5
|
| 3 |
size 101184122
|
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:cad93a7e3572e6b82d445f60e2ebd8578a7fe66abb7b9d5234247eef2b871e82
|
| 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:f067d6ff1a98806fe29d9018f9017970da944606a8b344add0e88cb1ed62ea89
|
| 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 sha256:39b4ce5437221792fe7f1d6337ae622063ecf52bab27cd4a3727195d1826d53a
|
| 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": 201,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -4260,6 +4260,1392 @@
|
|
| 4260 |
"eval_samples_per_second": 22.249,
|
| 4261 |
"eval_steps_per_second": 5.566,
|
| 4262 |
"step": 603
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4263 |
}
|
| 4264 |
],
|
| 4265 |
"logging_steps": 1,
|
|
@@ -4274,12 +5660,12 @@
|
|
| 4274 |
"should_evaluate": false,
|
| 4275 |
"should_log": false,
|
| 4276 |
"should_save": true,
|
| 4277 |
-
"should_training_stop":
|
| 4278 |
},
|
| 4279 |
"attributes": {}
|
| 4280 |
}
|
| 4281 |
},
|
| 4282 |
-
"total_flos":
|
| 4283 |
"train_batch_size": 2,
|
| 4284 |
"trial_name": null,
|
| 4285 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9990645463049579,
|
| 5 |
"eval_steps": 201,
|
| 6 |
+
"global_step": 801,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 4260 |
"eval_samples_per_second": 22.249,
|
| 4261 |
"eval_steps_per_second": 5.566,
|
| 4262 |
"step": 603
|
| 4263 |
+
},
|
| 4264 |
+
{
|
| 4265 |
+
"epoch": 0.7533520424072342,
|
| 4266 |
+
"grad_norm": NaN,
|
| 4267 |
+
"learning_rate": 3.207769905442359e-05,
|
| 4268 |
+
"loss": 0.0,
|
| 4269 |
+
"step": 604
|
| 4270 |
+
},
|
| 4271 |
+
{
|
| 4272 |
+
"epoch": 0.7545993140006236,
|
| 4273 |
+
"grad_norm": NaN,
|
| 4274 |
+
"learning_rate": 3.177127490813706e-05,
|
| 4275 |
+
"loss": 0.0,
|
| 4276 |
+
"step": 605
|
| 4277 |
+
},
|
| 4278 |
+
{
|
| 4279 |
+
"epoch": 0.7558465855940131,
|
| 4280 |
+
"grad_norm": NaN,
|
| 4281 |
+
"learning_rate": 3.1466044712824805e-05,
|
| 4282 |
+
"loss": 0.0,
|
| 4283 |
+
"step": 606
|
| 4284 |
+
},
|
| 4285 |
+
{
|
| 4286 |
+
"epoch": 0.7570938571874025,
|
| 4287 |
+
"grad_norm": NaN,
|
| 4288 |
+
"learning_rate": 3.1162013809783955e-05,
|
| 4289 |
+
"loss": 0.0,
|
| 4290 |
+
"step": 607
|
| 4291 |
+
},
|
| 4292 |
+
{
|
| 4293 |
+
"epoch": 0.758341128780792,
|
| 4294 |
+
"grad_norm": NaN,
|
| 4295 |
+
"learning_rate": 3.0859187519324806e-05,
|
| 4296 |
+
"loss": 0.0,
|
| 4297 |
+
"step": 608
|
| 4298 |
+
},
|
| 4299 |
+
{
|
| 4300 |
+
"epoch": 0.7595884003741815,
|
| 4301 |
+
"grad_norm": NaN,
|
| 4302 |
+
"learning_rate": 3.055757114067794e-05,
|
| 4303 |
+
"loss": 0.0,
|
| 4304 |
+
"step": 609
|
| 4305 |
+
},
|
| 4306 |
+
{
|
| 4307 |
+
"epoch": 0.760835671967571,
|
| 4308 |
+
"grad_norm": NaN,
|
| 4309 |
+
"learning_rate": 3.025716995190141e-05,
|
| 4310 |
+
"loss": 0.0,
|
| 4311 |
+
"step": 610
|
| 4312 |
+
},
|
| 4313 |
+
{
|
| 4314 |
+
"epoch": 0.7620829435609604,
|
| 4315 |
+
"grad_norm": NaN,
|
| 4316 |
+
"learning_rate": 2.995798920978825e-05,
|
| 4317 |
+
"loss": 0.0,
|
| 4318 |
+
"step": 611
|
| 4319 |
+
},
|
| 4320 |
+
{
|
| 4321 |
+
"epoch": 0.7633302151543498,
|
| 4322 |
+
"grad_norm": NaN,
|
| 4323 |
+
"learning_rate": 2.966003414977475e-05,
|
| 4324 |
+
"loss": 0.0,
|
| 4325 |
+
"step": 612
|
| 4326 |
+
},
|
| 4327 |
+
{
|
| 4328 |
+
"epoch": 0.7645774867477393,
|
| 4329 |
+
"grad_norm": NaN,
|
| 4330 |
+
"learning_rate": 2.9363309985848585e-05,
|
| 4331 |
+
"loss": 0.0,
|
| 4332 |
+
"step": 613
|
| 4333 |
+
},
|
| 4334 |
+
{
|
| 4335 |
+
"epoch": 0.7658247583411287,
|
| 4336 |
+
"grad_norm": NaN,
|
| 4337 |
+
"learning_rate": 2.9067821910457704e-05,
|
| 4338 |
+
"loss": 0.0,
|
| 4339 |
+
"step": 614
|
| 4340 |
+
},
|
| 4341 |
+
{
|
| 4342 |
+
"epoch": 0.7670720299345183,
|
| 4343 |
+
"grad_norm": NaN,
|
| 4344 |
+
"learning_rate": 2.877357509441947e-05,
|
| 4345 |
+
"loss": 0.0,
|
| 4346 |
+
"step": 615
|
| 4347 |
+
},
|
| 4348 |
+
{
|
| 4349 |
+
"epoch": 0.7683193015279077,
|
| 4350 |
+
"grad_norm": NaN,
|
| 4351 |
+
"learning_rate": 2.8480574686830142e-05,
|
| 4352 |
+
"loss": 0.0,
|
| 4353 |
+
"step": 616
|
| 4354 |
+
},
|
| 4355 |
+
{
|
| 4356 |
+
"epoch": 0.7695665731212972,
|
| 4357 |
+
"grad_norm": NaN,
|
| 4358 |
+
"learning_rate": 2.8188825814974795e-05,
|
| 4359 |
+
"loss": 0.0,
|
| 4360 |
+
"step": 617
|
| 4361 |
+
},
|
| 4362 |
+
{
|
| 4363 |
+
"epoch": 0.7708138447146866,
|
| 4364 |
+
"grad_norm": NaN,
|
| 4365 |
+
"learning_rate": 2.7898333584237534e-05,
|
| 4366 |
+
"loss": 0.0,
|
| 4367 |
+
"step": 618
|
| 4368 |
+
},
|
| 4369 |
+
{
|
| 4370 |
+
"epoch": 0.7720611163080761,
|
| 4371 |
+
"grad_norm": NaN,
|
| 4372 |
+
"learning_rate": 2.7609103078012166e-05,
|
| 4373 |
+
"loss": 0.0,
|
| 4374 |
+
"step": 619
|
| 4375 |
+
},
|
| 4376 |
+
{
|
| 4377 |
+
"epoch": 0.7733083879014655,
|
| 4378 |
+
"grad_norm": NaN,
|
| 4379 |
+
"learning_rate": 2.7321139357613412e-05,
|
| 4380 |
+
"loss": 0.0,
|
| 4381 |
+
"step": 620
|
| 4382 |
+
},
|
| 4383 |
+
{
|
| 4384 |
+
"epoch": 0.774555659494855,
|
| 4385 |
+
"grad_norm": NaN,
|
| 4386 |
+
"learning_rate": 2.703444746218802e-05,
|
| 4387 |
+
"loss": 0.0,
|
| 4388 |
+
"step": 621
|
| 4389 |
+
},
|
| 4390 |
+
{
|
| 4391 |
+
"epoch": 0.7758029310882445,
|
| 4392 |
+
"grad_norm": NaN,
|
| 4393 |
+
"learning_rate": 2.6749032408626907e-05,
|
| 4394 |
+
"loss": 0.0,
|
| 4395 |
+
"step": 622
|
| 4396 |
+
},
|
| 4397 |
+
{
|
| 4398 |
+
"epoch": 0.7770502026816339,
|
| 4399 |
+
"grad_norm": NaN,
|
| 4400 |
+
"learning_rate": 2.646489919147721e-05,
|
| 4401 |
+
"loss": 0.0,
|
| 4402 |
+
"step": 623
|
| 4403 |
+
},
|
| 4404 |
+
{
|
| 4405 |
+
"epoch": 0.7782974742750234,
|
| 4406 |
+
"grad_norm": NaN,
|
| 4407 |
+
"learning_rate": 2.6182052782854806e-05,
|
| 4408 |
+
"loss": 0.0,
|
| 4409 |
+
"step": 624
|
| 4410 |
+
},
|
| 4411 |
+
{
|
| 4412 |
+
"epoch": 0.7795447458684128,
|
| 4413 |
+
"grad_norm": NaN,
|
| 4414 |
+
"learning_rate": 2.59004981323575e-05,
|
| 4415 |
+
"loss": 0.0,
|
| 4416 |
+
"step": 625
|
| 4417 |
+
},
|
| 4418 |
+
{
|
| 4419 |
+
"epoch": 0.7807920174618023,
|
| 4420 |
+
"grad_norm": NaN,
|
| 4421 |
+
"learning_rate": 2.5620240166978226e-05,
|
| 4422 |
+
"loss": 0.0,
|
| 4423 |
+
"step": 626
|
| 4424 |
+
},
|
| 4425 |
+
{
|
| 4426 |
+
"epoch": 0.7820392890551918,
|
| 4427 |
+
"grad_norm": NaN,
|
| 4428 |
+
"learning_rate": 2.5341283791018988e-05,
|
| 4429 |
+
"loss": 0.0,
|
| 4430 |
+
"step": 627
|
| 4431 |
+
},
|
| 4432 |
+
{
|
| 4433 |
+
"epoch": 0.7832865606485813,
|
| 4434 |
+
"grad_norm": NaN,
|
| 4435 |
+
"learning_rate": 2.5063633886004935e-05,
|
| 4436 |
+
"loss": 0.0,
|
| 4437 |
+
"step": 628
|
| 4438 |
+
},
|
| 4439 |
+
{
|
| 4440 |
+
"epoch": 0.7845338322419707,
|
| 4441 |
+
"grad_norm": NaN,
|
| 4442 |
+
"learning_rate": 2.4787295310598913e-05,
|
| 4443 |
+
"loss": 0.0,
|
| 4444 |
+
"step": 629
|
| 4445 |
+
},
|
| 4446 |
+
{
|
| 4447 |
+
"epoch": 0.7857811038353602,
|
| 4448 |
+
"grad_norm": NaN,
|
| 4449 |
+
"learning_rate": 2.45122729005166e-05,
|
| 4450 |
+
"loss": 0.0,
|
| 4451 |
+
"step": 630
|
| 4452 |
+
},
|
| 4453 |
+
{
|
| 4454 |
+
"epoch": 0.7870283754287496,
|
| 4455 |
+
"grad_norm": NaN,
|
| 4456 |
+
"learning_rate": 2.423857146844164e-05,
|
| 4457 |
+
"loss": 0.0,
|
| 4458 |
+
"step": 631
|
| 4459 |
+
},
|
| 4460 |
+
{
|
| 4461 |
+
"epoch": 0.788275647022139,
|
| 4462 |
+
"grad_norm": NaN,
|
| 4463 |
+
"learning_rate": 2.3966195803941715e-05,
|
| 4464 |
+
"loss": 0.0,
|
| 4465 |
+
"step": 632
|
| 4466 |
+
},
|
| 4467 |
+
{
|
| 4468 |
+
"epoch": 0.7895229186155285,
|
| 4469 |
+
"grad_norm": NaN,
|
| 4470 |
+
"learning_rate": 2.3695150673384437e-05,
|
| 4471 |
+
"loss": 0.0,
|
| 4472 |
+
"step": 633
|
| 4473 |
+
},
|
| 4474 |
+
{
|
| 4475 |
+
"epoch": 0.790770190208918,
|
| 4476 |
+
"grad_norm": NaN,
|
| 4477 |
+
"learning_rate": 2.3425440819854185e-05,
|
| 4478 |
+
"loss": 0.0,
|
| 4479 |
+
"step": 634
|
| 4480 |
+
},
|
| 4481 |
+
{
|
| 4482 |
+
"epoch": 0.7920174618023075,
|
| 4483 |
+
"grad_norm": NaN,
|
| 4484 |
+
"learning_rate": 2.3157070963068984e-05,
|
| 4485 |
+
"loss": 0.0,
|
| 4486 |
+
"step": 635
|
| 4487 |
+
},
|
| 4488 |
+
{
|
| 4489 |
+
"epoch": 0.7932647333956969,
|
| 4490 |
+
"grad_norm": NaN,
|
| 4491 |
+
"learning_rate": 2.2890045799297876e-05,
|
| 4492 |
+
"loss": 0.0,
|
| 4493 |
+
"step": 636
|
| 4494 |
+
},
|
| 4495 |
+
{
|
| 4496 |
+
"epoch": 0.7945120049890864,
|
| 4497 |
+
"grad_norm": NaN,
|
| 4498 |
+
"learning_rate": 2.2624370001278838e-05,
|
| 4499 |
+
"loss": 0.0,
|
| 4500 |
+
"step": 637
|
| 4501 |
+
},
|
| 4502 |
+
{
|
| 4503 |
+
"epoch": 0.7957592765824758,
|
| 4504 |
+
"grad_norm": NaN,
|
| 4505 |
+
"learning_rate": 2.2360048218136985e-05,
|
| 4506 |
+
"loss": 0.0,
|
| 4507 |
+
"step": 638
|
| 4508 |
+
},
|
| 4509 |
+
{
|
| 4510 |
+
"epoch": 0.7970065481758652,
|
| 4511 |
+
"grad_norm": NaN,
|
| 4512 |
+
"learning_rate": 2.2097085075303138e-05,
|
| 4513 |
+
"loss": 0.0,
|
| 4514 |
+
"step": 639
|
| 4515 |
+
},
|
| 4516 |
+
{
|
| 4517 |
+
"epoch": 0.7982538197692548,
|
| 4518 |
+
"grad_norm": NaN,
|
| 4519 |
+
"learning_rate": 2.1835485174433002e-05,
|
| 4520 |
+
"loss": 0.0,
|
| 4521 |
+
"step": 640
|
| 4522 |
+
},
|
| 4523 |
+
{
|
| 4524 |
+
"epoch": 0.7995010913626442,
|
| 4525 |
+
"grad_norm": NaN,
|
| 4526 |
+
"learning_rate": 2.1575253093326586e-05,
|
| 4527 |
+
"loss": 0.0,
|
| 4528 |
+
"step": 641
|
| 4529 |
+
},
|
| 4530 |
+
{
|
| 4531 |
+
"epoch": 0.8007483629560337,
|
| 4532 |
+
"grad_norm": NaN,
|
| 4533 |
+
"learning_rate": 2.131639338584801e-05,
|
| 4534 |
+
"loss": 0.0,
|
| 4535 |
+
"step": 642
|
| 4536 |
+
},
|
| 4537 |
+
{
|
| 4538 |
+
"epoch": 0.8019956345494231,
|
| 4539 |
+
"grad_norm": NaN,
|
| 4540 |
+
"learning_rate": 2.1058910581846013e-05,
|
| 4541 |
+
"loss": 0.0,
|
| 4542 |
+
"step": 643
|
| 4543 |
+
},
|
| 4544 |
+
{
|
| 4545 |
+
"epoch": 0.8032429061428126,
|
| 4546 |
+
"grad_norm": NaN,
|
| 4547 |
+
"learning_rate": 2.0802809187074434e-05,
|
| 4548 |
+
"loss": 0.0,
|
| 4549 |
+
"step": 644
|
| 4550 |
+
},
|
| 4551 |
+
{
|
| 4552 |
+
"epoch": 0.804490177736202,
|
| 4553 |
+
"grad_norm": NaN,
|
| 4554 |
+
"learning_rate": 2.05480936831136e-05,
|
| 4555 |
+
"loss": 0.0,
|
| 4556 |
+
"step": 645
|
| 4557 |
+
},
|
| 4558 |
+
{
|
| 4559 |
+
"epoch": 0.8057374493295916,
|
| 4560 |
+
"grad_norm": NaN,
|
| 4561 |
+
"learning_rate": 2.0294768527291796e-05,
|
| 4562 |
+
"loss": 0.0,
|
| 4563 |
+
"step": 646
|
| 4564 |
+
},
|
| 4565 |
+
{
|
| 4566 |
+
"epoch": 0.806984720922981,
|
| 4567 |
+
"grad_norm": NaN,
|
| 4568 |
+
"learning_rate": 2.004283815260717e-05,
|
| 4569 |
+
"loss": 0.0,
|
| 4570 |
+
"step": 647
|
| 4571 |
+
},
|
| 4572 |
+
{
|
| 4573 |
+
"epoch": 0.8082319925163705,
|
| 4574 |
+
"grad_norm": NaN,
|
| 4575 |
+
"learning_rate": 1.9792306967650398e-05,
|
| 4576 |
+
"loss": 0.0,
|
| 4577 |
+
"step": 648
|
| 4578 |
+
},
|
| 4579 |
+
{
|
| 4580 |
+
"epoch": 0.8094792641097599,
|
| 4581 |
+
"grad_norm": NaN,
|
| 4582 |
+
"learning_rate": 1.9543179356527252e-05,
|
| 4583 |
+
"loss": 0.0,
|
| 4584 |
+
"step": 649
|
| 4585 |
+
},
|
| 4586 |
+
{
|
| 4587 |
+
"epoch": 0.8107265357031493,
|
| 4588 |
+
"grad_norm": NaN,
|
| 4589 |
+
"learning_rate": 1.9295459678782168e-05,
|
| 4590 |
+
"loss": 0.0,
|
| 4591 |
+
"step": 650
|
| 4592 |
+
},
|
| 4593 |
+
{
|
| 4594 |
+
"epoch": 0.8119738072965388,
|
| 4595 |
+
"grad_norm": NaN,
|
| 4596 |
+
"learning_rate": 1.904915226932169e-05,
|
| 4597 |
+
"loss": 0.0,
|
| 4598 |
+
"step": 651
|
| 4599 |
+
},
|
| 4600 |
+
{
|
| 4601 |
+
"epoch": 0.8132210788899282,
|
| 4602 |
+
"grad_norm": NaN,
|
| 4603 |
+
"learning_rate": 1.88042614383388e-05,
|
| 4604 |
+
"loss": 0.0,
|
| 4605 |
+
"step": 652
|
| 4606 |
+
},
|
| 4607 |
+
{
|
| 4608 |
+
"epoch": 0.8144683504833178,
|
| 4609 |
+
"grad_norm": NaN,
|
| 4610 |
+
"learning_rate": 1.856079147123746e-05,
|
| 4611 |
+
"loss": 0.0,
|
| 4612 |
+
"step": 653
|
| 4613 |
+
},
|
| 4614 |
+
{
|
| 4615 |
+
"epoch": 0.8157156220767072,
|
| 4616 |
+
"grad_norm": NaN,
|
| 4617 |
+
"learning_rate": 1.8318746628557526e-05,
|
| 4618 |
+
"loss": 0.0,
|
| 4619 |
+
"step": 654
|
| 4620 |
+
},
|
| 4621 |
+
{
|
| 4622 |
+
"epoch": 0.8169628936700967,
|
| 4623 |
+
"grad_norm": NaN,
|
| 4624 |
+
"learning_rate": 1.8078131145900267e-05,
|
| 4625 |
+
"loss": 0.0,
|
| 4626 |
+
"step": 655
|
| 4627 |
+
},
|
| 4628 |
+
{
|
| 4629 |
+
"epoch": 0.8182101652634861,
|
| 4630 |
+
"grad_norm": NaN,
|
| 4631 |
+
"learning_rate": 1.7838949233854284e-05,
|
| 4632 |
+
"loss": 0.0,
|
| 4633 |
+
"step": 656
|
| 4634 |
+
},
|
| 4635 |
+
{
|
| 4636 |
+
"epoch": 0.8194574368568756,
|
| 4637 |
+
"grad_norm": NaN,
|
| 4638 |
+
"learning_rate": 1.760120507792169e-05,
|
| 4639 |
+
"loss": 0.0,
|
| 4640 |
+
"step": 657
|
| 4641 |
+
},
|
| 4642 |
+
{
|
| 4643 |
+
"epoch": 0.820704708450265,
|
| 4644 |
+
"grad_norm": NaN,
|
| 4645 |
+
"learning_rate": 1.7364902838445106e-05,
|
| 4646 |
+
"loss": 0.0,
|
| 4647 |
+
"step": 658
|
| 4648 |
+
},
|
| 4649 |
+
{
|
| 4650 |
+
"epoch": 0.8219519800436546,
|
| 4651 |
+
"grad_norm": NaN,
|
| 4652 |
+
"learning_rate": 1.713004665053457e-05,
|
| 4653 |
+
"loss": 0.0,
|
| 4654 |
+
"step": 659
|
| 4655 |
+
},
|
| 4656 |
+
{
|
| 4657 |
+
"epoch": 0.823199251637044,
|
| 4658 |
+
"grad_norm": NaN,
|
| 4659 |
+
"learning_rate": 1.6896640623995318e-05,
|
| 4660 |
+
"loss": 0.0,
|
| 4661 |
+
"step": 660
|
| 4662 |
+
},
|
| 4663 |
+
{
|
| 4664 |
+
"epoch": 0.8244465232304334,
|
| 4665 |
+
"grad_norm": NaN,
|
| 4666 |
+
"learning_rate": 1.666468884325596e-05,
|
| 4667 |
+
"loss": 0.0,
|
| 4668 |
+
"step": 661
|
| 4669 |
+
},
|
| 4670 |
+
{
|
| 4671 |
+
"epoch": 0.8256937948238229,
|
| 4672 |
+
"grad_norm": NaN,
|
| 4673 |
+
"learning_rate": 1.6434195367296802e-05,
|
| 4674 |
+
"loss": 0.0,
|
| 4675 |
+
"step": 662
|
| 4676 |
+
},
|
| 4677 |
+
{
|
| 4678 |
+
"epoch": 0.8269410664172123,
|
| 4679 |
+
"grad_norm": NaN,
|
| 4680 |
+
"learning_rate": 1.6205164229578994e-05,
|
| 4681 |
+
"loss": 0.0,
|
| 4682 |
+
"step": 663
|
| 4683 |
+
},
|
| 4684 |
+
{
|
| 4685 |
+
"epoch": 0.8281883380106018,
|
| 4686 |
+
"grad_norm": NaN,
|
| 4687 |
+
"learning_rate": 1.5977599437973867e-05,
|
| 4688 |
+
"loss": 0.0,
|
| 4689 |
+
"step": 664
|
| 4690 |
+
},
|
| 4691 |
+
{
|
| 4692 |
+
"epoch": 0.8294356096039913,
|
| 4693 |
+
"grad_norm": NaN,
|
| 4694 |
+
"learning_rate": 1.5751504974692733e-05,
|
| 4695 |
+
"loss": 0.0,
|
| 4696 |
+
"step": 665
|
| 4697 |
+
},
|
| 4698 |
+
{
|
| 4699 |
+
"epoch": 0.8306828811973808,
|
| 4700 |
+
"grad_norm": NaN,
|
| 4701 |
+
"learning_rate": 1.55268847962174e-05,
|
| 4702 |
+
"loss": 0.0,
|
| 4703 |
+
"step": 666
|
| 4704 |
+
},
|
| 4705 |
+
{
|
| 4706 |
+
"epoch": 0.8319301527907702,
|
| 4707 |
+
"grad_norm": NaN,
|
| 4708 |
+
"learning_rate": 1.5303742833230673e-05,
|
| 4709 |
+
"loss": 0.0,
|
| 4710 |
+
"step": 667
|
| 4711 |
+
},
|
| 4712 |
+
{
|
| 4713 |
+
"epoch": 0.8331774243841596,
|
| 4714 |
+
"grad_norm": NaN,
|
| 4715 |
+
"learning_rate": 1.5082082990547796e-05,
|
| 4716 |
+
"loss": 0.0,
|
| 4717 |
+
"step": 668
|
| 4718 |
+
},
|
| 4719 |
+
{
|
| 4720 |
+
"epoch": 0.8344246959775491,
|
| 4721 |
+
"grad_norm": NaN,
|
| 4722 |
+
"learning_rate": 1.4861909147048025e-05,
|
| 4723 |
+
"loss": 0.0,
|
| 4724 |
+
"step": 669
|
| 4725 |
+
},
|
| 4726 |
+
{
|
| 4727 |
+
"epoch": 0.8356719675709385,
|
| 4728 |
+
"grad_norm": NaN,
|
| 4729 |
+
"learning_rate": 1.464322515560671e-05,
|
| 4730 |
+
"loss": 0.0,
|
| 4731 |
+
"step": 670
|
| 4732 |
+
},
|
| 4733 |
+
{
|
| 4734 |
+
"epoch": 0.836919239164328,
|
| 4735 |
+
"grad_norm": NaN,
|
| 4736 |
+
"learning_rate": 1.4426034843027969e-05,
|
| 4737 |
+
"loss": 0.0,
|
| 4738 |
+
"step": 671
|
| 4739 |
+
},
|
| 4740 |
+
{
|
| 4741 |
+
"epoch": 0.8381665107577175,
|
| 4742 |
+
"grad_norm": NaN,
|
| 4743 |
+
"learning_rate": 1.4210342009977628e-05,
|
| 4744 |
+
"loss": 0.0,
|
| 4745 |
+
"step": 672
|
| 4746 |
+
},
|
| 4747 |
+
{
|
| 4748 |
+
"epoch": 0.839413782351107,
|
| 4749 |
+
"grad_norm": NaN,
|
| 4750 |
+
"learning_rate": 1.3996150430916799e-05,
|
| 4751 |
+
"loss": 0.0,
|
| 4752 |
+
"step": 673
|
| 4753 |
+
},
|
| 4754 |
+
{
|
| 4755 |
+
"epoch": 0.8406610539444964,
|
| 4756 |
+
"grad_norm": NaN,
|
| 4757 |
+
"learning_rate": 1.378346385403575e-05,
|
| 4758 |
+
"loss": 0.0,
|
| 4759 |
+
"step": 674
|
| 4760 |
+
},
|
| 4761 |
+
{
|
| 4762 |
+
"epoch": 0.8419083255378859,
|
| 4763 |
+
"grad_norm": NaN,
|
| 4764 |
+
"learning_rate": 1.357228600118836e-05,
|
| 4765 |
+
"loss": 0.0,
|
| 4766 |
+
"step": 675
|
| 4767 |
+
},
|
| 4768 |
+
{
|
| 4769 |
+
"epoch": 0.8431555971312753,
|
| 4770 |
+
"grad_norm": NaN,
|
| 4771 |
+
"learning_rate": 1.3362620567827033e-05,
|
| 4772 |
+
"loss": 0.0,
|
| 4773 |
+
"step": 676
|
| 4774 |
+
},
|
| 4775 |
+
{
|
| 4776 |
+
"epoch": 0.8444028687246647,
|
| 4777 |
+
"grad_norm": NaN,
|
| 4778 |
+
"learning_rate": 1.3154471222937903e-05,
|
| 4779 |
+
"loss": 0.0,
|
| 4780 |
+
"step": 677
|
| 4781 |
+
},
|
| 4782 |
+
{
|
| 4783 |
+
"epoch": 0.8456501403180543,
|
| 4784 |
+
"grad_norm": NaN,
|
| 4785 |
+
"learning_rate": 1.2947841608976718e-05,
|
| 4786 |
+
"loss": 0.0,
|
| 4787 |
+
"step": 678
|
| 4788 |
+
},
|
| 4789 |
+
{
|
| 4790 |
+
"epoch": 0.8468974119114437,
|
| 4791 |
+
"grad_norm": NaN,
|
| 4792 |
+
"learning_rate": 1.2742735341805145e-05,
|
| 4793 |
+
"loss": 0.0,
|
| 4794 |
+
"step": 679
|
| 4795 |
+
},
|
| 4796 |
+
{
|
| 4797 |
+
"epoch": 0.8481446835048332,
|
| 4798 |
+
"grad_norm": NaN,
|
| 4799 |
+
"learning_rate": 1.253915601062734e-05,
|
| 4800 |
+
"loss": 0.0,
|
| 4801 |
+
"step": 680
|
| 4802 |
+
},
|
| 4803 |
+
{
|
| 4804 |
+
"epoch": 0.8493919550982226,
|
| 4805 |
+
"grad_norm": NaN,
|
| 4806 |
+
"learning_rate": 1.2337107177927365e-05,
|
| 4807 |
+
"loss": 0.0,
|
| 4808 |
+
"step": 681
|
| 4809 |
+
},
|
| 4810 |
+
{
|
| 4811 |
+
"epoch": 0.8506392266916121,
|
| 4812 |
+
"grad_norm": NaN,
|
| 4813 |
+
"learning_rate": 1.213659237940662e-05,
|
| 4814 |
+
"loss": 0.0,
|
| 4815 |
+
"step": 682
|
| 4816 |
+
},
|
| 4817 |
+
{
|
| 4818 |
+
"epoch": 0.8518864982850015,
|
| 4819 |
+
"grad_norm": NaN,
|
| 4820 |
+
"learning_rate": 1.1937615123922052e-05,
|
| 4821 |
+
"loss": 0.0,
|
| 4822 |
+
"step": 683
|
| 4823 |
+
},
|
| 4824 |
+
{
|
| 4825 |
+
"epoch": 0.8531337698783911,
|
| 4826 |
+
"grad_norm": NaN,
|
| 4827 |
+
"learning_rate": 1.174017889342489e-05,
|
| 4828 |
+
"loss": 0.0,
|
| 4829 |
+
"step": 684
|
| 4830 |
+
},
|
| 4831 |
+
{
|
| 4832 |
+
"epoch": 0.8543810414717805,
|
| 4833 |
+
"grad_norm": NaN,
|
| 4834 |
+
"learning_rate": 1.1544287142899446e-05,
|
| 4835 |
+
"loss": 0.0,
|
| 4836 |
+
"step": 685
|
| 4837 |
+
},
|
| 4838 |
+
{
|
| 4839 |
+
"epoch": 0.85562831306517,
|
| 4840 |
+
"grad_norm": NaN,
|
| 4841 |
+
"learning_rate": 1.1349943300302913e-05,
|
| 4842 |
+
"loss": 0.0,
|
| 4843 |
+
"step": 686
|
| 4844 |
+
},
|
| 4845 |
+
{
|
| 4846 |
+
"epoch": 0.8568755846585594,
|
| 4847 |
+
"grad_norm": NaN,
|
| 4848 |
+
"learning_rate": 1.1157150766505253e-05,
|
| 4849 |
+
"loss": 0.0,
|
| 4850 |
+
"step": 687
|
| 4851 |
+
},
|
| 4852 |
+
{
|
| 4853 |
+
"epoch": 0.8581228562519488,
|
| 4854 |
+
"grad_norm": NaN,
|
| 4855 |
+
"learning_rate": 1.0965912915229625e-05,
|
| 4856 |
+
"loss": 0.0,
|
| 4857 |
+
"step": 688
|
| 4858 |
+
},
|
| 4859 |
+
{
|
| 4860 |
+
"epoch": 0.8593701278453383,
|
| 4861 |
+
"grad_norm": NaN,
|
| 4862 |
+
"learning_rate": 1.0776233092993527e-05,
|
| 4863 |
+
"loss": 0.0,
|
| 4864 |
+
"step": 689
|
| 4865 |
+
},
|
| 4866 |
+
{
|
| 4867 |
+
"epoch": 0.8606173994387278,
|
| 4868 |
+
"grad_norm": NaN,
|
| 4869 |
+
"learning_rate": 1.0588114619050028e-05,
|
| 4870 |
+
"loss": 0.0,
|
| 4871 |
+
"step": 690
|
| 4872 |
+
},
|
| 4873 |
+
{
|
| 4874 |
+
"epoch": 0.8618646710321173,
|
| 4875 |
+
"grad_norm": NaN,
|
| 4876 |
+
"learning_rate": 1.040156078532989e-05,
|
| 4877 |
+
"loss": 0.0,
|
| 4878 |
+
"step": 691
|
| 4879 |
+
},
|
| 4880 |
+
{
|
| 4881 |
+
"epoch": 0.8631119426255067,
|
| 4882 |
+
"grad_norm": NaN,
|
| 4883 |
+
"learning_rate": 1.0216574856383742e-05,
|
| 4884 |
+
"loss": 0.0,
|
| 4885 |
+
"step": 692
|
| 4886 |
+
},
|
| 4887 |
+
{
|
| 4888 |
+
"epoch": 0.8643592142188962,
|
| 4889 |
+
"grad_norm": NaN,
|
| 4890 |
+
"learning_rate": 1.0033160069325166e-05,
|
| 4891 |
+
"loss": 0.0,
|
| 4892 |
+
"step": 693
|
| 4893 |
+
},
|
| 4894 |
+
{
|
| 4895 |
+
"epoch": 0.8656064858122856,
|
| 4896 |
+
"grad_norm": NaN,
|
| 4897 |
+
"learning_rate": 9.851319633773926e-06,
|
| 4898 |
+
"loss": 0.0,
|
| 4899 |
+
"step": 694
|
| 4900 |
+
},
|
| 4901 |
+
{
|
| 4902 |
+
"epoch": 0.866853757405675,
|
| 4903 |
+
"grad_norm": NaN,
|
| 4904 |
+
"learning_rate": 9.671056731799777e-06,
|
| 4905 |
+
"loss": 0.0,
|
| 4906 |
+
"step": 695
|
| 4907 |
+
},
|
| 4908 |
+
{
|
| 4909 |
+
"epoch": 0.8681010289990645,
|
| 4910 |
+
"grad_norm": NaN,
|
| 4911 |
+
"learning_rate": 9.49237451786692e-06,
|
| 4912 |
+
"loss": 0.0,
|
| 4913 |
+
"step": 696
|
| 4914 |
+
},
|
| 4915 |
+
{
|
| 4916 |
+
"epoch": 0.869348300592454,
|
| 4917 |
+
"grad_norm": NaN,
|
| 4918 |
+
"learning_rate": 9.315276118778627e-06,
|
| 4919 |
+
"loss": 0.0,
|
| 4920 |
+
"step": 697
|
| 4921 |
+
},
|
| 4922 |
+
{
|
| 4923 |
+
"epoch": 0.8705955721858435,
|
| 4924 |
+
"grad_norm": NaN,
|
| 4925 |
+
"learning_rate": 9.139764633622617e-06,
|
| 4926 |
+
"loss": 0.0,
|
| 4927 |
+
"step": 698
|
| 4928 |
+
},
|
| 4929 |
+
{
|
| 4930 |
+
"epoch": 0.8718428437792329,
|
| 4931 |
+
"grad_norm": NaN,
|
| 4932 |
+
"learning_rate": 8.965843133716933e-06,
|
| 4933 |
+
"loss": 0.0,
|
| 4934 |
+
"step": 699
|
| 4935 |
+
},
|
| 4936 |
+
{
|
| 4937 |
+
"epoch": 0.8730901153726224,
|
| 4938 |
+
"grad_norm": NaN,
|
| 4939 |
+
"learning_rate": 8.793514662555946e-06,
|
| 4940 |
+
"loss": 0.0,
|
| 4941 |
+
"step": 700
|
| 4942 |
+
},
|
| 4943 |
+
{
|
| 4944 |
+
"epoch": 0.8743373869660118,
|
| 4945 |
+
"grad_norm": NaN,
|
| 4946 |
+
"learning_rate": 8.622782235757276e-06,
|
| 4947 |
+
"loss": 0.0,
|
| 4948 |
+
"step": 701
|
| 4949 |
+
},
|
| 4950 |
+
{
|
| 4951 |
+
"epoch": 0.8755846585594013,
|
| 4952 |
+
"grad_norm": NaN,
|
| 4953 |
+
"learning_rate": 8.453648841009021e-06,
|
| 4954 |
+
"loss": 0.0,
|
| 4955 |
+
"step": 702
|
| 4956 |
+
},
|
| 4957 |
+
{
|
| 4958 |
+
"epoch": 0.8768319301527908,
|
| 4959 |
+
"grad_norm": NaN,
|
| 4960 |
+
"learning_rate": 8.286117438017337e-06,
|
| 4961 |
+
"loss": 0.0,
|
| 4962 |
+
"step": 703
|
| 4963 |
+
},
|
| 4964 |
+
{
|
| 4965 |
+
"epoch": 0.8780792017461803,
|
| 4966 |
+
"grad_norm": NaN,
|
| 4967 |
+
"learning_rate": 8.120190958454843e-06,
|
| 4968 |
+
"loss": 0.0,
|
| 4969 |
+
"step": 704
|
| 4970 |
+
},
|
| 4971 |
+
{
|
| 4972 |
+
"epoch": 0.8793264733395697,
|
| 4973 |
+
"grad_norm": NaN,
|
| 4974 |
+
"learning_rate": 7.955872305909152e-06,
|
| 4975 |
+
"loss": 0.0,
|
| 4976 |
+
"step": 705
|
| 4977 |
+
},
|
| 4978 |
+
{
|
| 4979 |
+
"epoch": 0.8805737449329591,
|
| 4980 |
+
"grad_norm": NaN,
|
| 4981 |
+
"learning_rate": 7.793164355832127e-06,
|
| 4982 |
+
"loss": 0.0,
|
| 4983 |
+
"step": 706
|
| 4984 |
+
},
|
| 4985 |
+
{
|
| 4986 |
+
"epoch": 0.8818210165263486,
|
| 4987 |
+
"grad_norm": NaN,
|
| 4988 |
+
"learning_rate": 7.632069955489585e-06,
|
| 4989 |
+
"loss": 0.0,
|
| 4990 |
+
"step": 707
|
| 4991 |
+
},
|
| 4992 |
+
{
|
| 4993 |
+
"epoch": 0.883068288119738,
|
| 4994 |
+
"grad_norm": NaN,
|
| 4995 |
+
"learning_rate": 7.472591923911398e-06,
|
| 4996 |
+
"loss": 0.0,
|
| 4997 |
+
"step": 708
|
| 4998 |
+
},
|
| 4999 |
+
{
|
| 5000 |
+
"epoch": 0.8843155597131276,
|
| 5001 |
+
"grad_norm": NaN,
|
| 5002 |
+
"learning_rate": 7.314733051842282e-06,
|
| 5003 |
+
"loss": 0.0,
|
| 5004 |
+
"step": 709
|
| 5005 |
+
},
|
| 5006 |
+
{
|
| 5007 |
+
"epoch": 0.885562831306517,
|
| 5008 |
+
"grad_norm": NaN,
|
| 5009 |
+
"learning_rate": 7.158496101692802e-06,
|
| 5010 |
+
"loss": 0.0,
|
| 5011 |
+
"step": 710
|
| 5012 |
+
},
|
| 5013 |
+
{
|
| 5014 |
+
"epoch": 0.8868101028999065,
|
| 5015 |
+
"grad_norm": NaN,
|
| 5016 |
+
"learning_rate": 7.003883807491185e-06,
|
| 5017 |
+
"loss": 0.0,
|
| 5018 |
+
"step": 711
|
| 5019 |
+
},
|
| 5020 |
+
{
|
| 5021 |
+
"epoch": 0.8880573744932959,
|
| 5022 |
+
"grad_norm": NaN,
|
| 5023 |
+
"learning_rate": 6.85089887483541e-06,
|
| 5024 |
+
"loss": 0.0,
|
| 5025 |
+
"step": 712
|
| 5026 |
+
},
|
| 5027 |
+
{
|
| 5028 |
+
"epoch": 0.8893046460866854,
|
| 5029 |
+
"grad_norm": NaN,
|
| 5030 |
+
"learning_rate": 6.699543980845801e-06,
|
| 5031 |
+
"loss": 0.0,
|
| 5032 |
+
"step": 713
|
| 5033 |
+
},
|
| 5034 |
+
{
|
| 5035 |
+
"epoch": 0.8905519176800748,
|
| 5036 |
+
"grad_norm": NaN,
|
| 5037 |
+
"learning_rate": 6.549821774118325e-06,
|
| 5038 |
+
"loss": 0.0,
|
| 5039 |
+
"step": 714
|
| 5040 |
+
},
|
| 5041 |
+
{
|
| 5042 |
+
"epoch": 0.8917991892734642,
|
| 5043 |
+
"grad_norm": NaN,
|
| 5044 |
+
"learning_rate": 6.401734874678089e-06,
|
| 5045 |
+
"loss": 0.0,
|
| 5046 |
+
"step": 715
|
| 5047 |
+
},
|
| 5048 |
+
{
|
| 5049 |
+
"epoch": 0.8930464608668538,
|
| 5050 |
+
"grad_norm": NaN,
|
| 5051 |
+
"learning_rate": 6.255285873933569e-06,
|
| 5052 |
+
"loss": 0.0,
|
| 5053 |
+
"step": 716
|
| 5054 |
+
},
|
| 5055 |
+
{
|
| 5056 |
+
"epoch": 0.8942937324602432,
|
| 5057 |
+
"grad_norm": NaN,
|
| 5058 |
+
"learning_rate": 6.110477334631326e-06,
|
| 5059 |
+
"loss": 0.0,
|
| 5060 |
+
"step": 717
|
| 5061 |
+
},
|
| 5062 |
+
{
|
| 5063 |
+
"epoch": 0.8955410040536327,
|
| 5064 |
+
"grad_norm": NaN,
|
| 5065 |
+
"learning_rate": 5.967311790811014e-06,
|
| 5066 |
+
"loss": 0.0,
|
| 5067 |
+
"step": 718
|
| 5068 |
+
},
|
| 5069 |
+
{
|
| 5070 |
+
"epoch": 0.8967882756470221,
|
| 5071 |
+
"grad_norm": NaN,
|
| 5072 |
+
"learning_rate": 5.825791747761123e-06,
|
| 5073 |
+
"loss": 0.0,
|
| 5074 |
+
"step": 719
|
| 5075 |
+
},
|
| 5076 |
+
{
|
| 5077 |
+
"epoch": 0.8980355472404116,
|
| 5078 |
+
"grad_norm": NaN,
|
| 5079 |
+
"learning_rate": 5.685919681975149e-06,
|
| 5080 |
+
"loss": 0.0,
|
| 5081 |
+
"step": 720
|
| 5082 |
+
},
|
| 5083 |
+
{
|
| 5084 |
+
"epoch": 0.899282818833801,
|
| 5085 |
+
"grad_norm": NaN,
|
| 5086 |
+
"learning_rate": 5.547698041108229e-06,
|
| 5087 |
+
"loss": 0.0,
|
| 5088 |
+
"step": 721
|
| 5089 |
+
},
|
| 5090 |
+
{
|
| 5091 |
+
"epoch": 0.9005300904271906,
|
| 5092 |
+
"grad_norm": NaN,
|
| 5093 |
+
"learning_rate": 5.4111292439342986e-06,
|
| 5094 |
+
"loss": 0.0,
|
| 5095 |
+
"step": 722
|
| 5096 |
+
},
|
| 5097 |
+
{
|
| 5098 |
+
"epoch": 0.90177736202058,
|
| 5099 |
+
"grad_norm": NaN,
|
| 5100 |
+
"learning_rate": 5.276215680303831e-06,
|
| 5101 |
+
"loss": 0.0,
|
| 5102 |
+
"step": 723
|
| 5103 |
+
},
|
| 5104 |
+
{
|
| 5105 |
+
"epoch": 0.9030246336139695,
|
| 5106 |
+
"grad_norm": NaN,
|
| 5107 |
+
"learning_rate": 5.14295971110188e-06,
|
| 5108 |
+
"loss": 0.0,
|
| 5109 |
+
"step": 724
|
| 5110 |
+
},
|
| 5111 |
+
{
|
| 5112 |
+
"epoch": 0.9042719052073589,
|
| 5113 |
+
"grad_norm": NaN,
|
| 5114 |
+
"learning_rate": 5.011363668206948e-06,
|
| 5115 |
+
"loss": 0.0,
|
| 5116 |
+
"step": 725
|
| 5117 |
+
},
|
| 5118 |
+
{
|
| 5119 |
+
"epoch": 0.9055191768007483,
|
| 5120 |
+
"grad_norm": NaN,
|
| 5121 |
+
"learning_rate": 4.881429854450004e-06,
|
| 5122 |
+
"loss": 0.0,
|
| 5123 |
+
"step": 726
|
| 5124 |
+
},
|
| 5125 |
+
{
|
| 5126 |
+
"epoch": 0.9067664483941378,
|
| 5127 |
+
"grad_norm": NaN,
|
| 5128 |
+
"learning_rate": 4.753160543574331e-06,
|
| 5129 |
+
"loss": 0.0,
|
| 5130 |
+
"step": 727
|
| 5131 |
+
},
|
| 5132 |
+
{
|
| 5133 |
+
"epoch": 0.9080137199875273,
|
| 5134 |
+
"grad_norm": NaN,
|
| 5135 |
+
"learning_rate": 4.626557980195623e-06,
|
| 5136 |
+
"loss": 0.0,
|
| 5137 |
+
"step": 728
|
| 5138 |
+
},
|
| 5139 |
+
{
|
| 5140 |
+
"epoch": 0.9092609915809168,
|
| 5141 |
+
"grad_norm": NaN,
|
| 5142 |
+
"learning_rate": 4.501624379762803e-06,
|
| 5143 |
+
"loss": 0.0,
|
| 5144 |
+
"step": 729
|
| 5145 |
+
},
|
| 5146 |
+
{
|
| 5147 |
+
"epoch": 0.9105082631743062,
|
| 5148 |
+
"grad_norm": NaN,
|
| 5149 |
+
"learning_rate": 4.3783619285191705e-06,
|
| 5150 |
+
"loss": 0.0,
|
| 5151 |
+
"step": 730
|
| 5152 |
+
},
|
| 5153 |
+
{
|
| 5154 |
+
"epoch": 0.9117555347676957,
|
| 5155 |
+
"grad_norm": NaN,
|
| 5156 |
+
"learning_rate": 4.2567727834641915e-06,
|
| 5157 |
+
"loss": 0.0,
|
| 5158 |
+
"step": 731
|
| 5159 |
+
},
|
| 5160 |
+
{
|
| 5161 |
+
"epoch": 0.9130028063610851,
|
| 5162 |
+
"grad_norm": NaN,
|
| 5163 |
+
"learning_rate": 4.136859072315758e-06,
|
| 5164 |
+
"loss": 0.0,
|
| 5165 |
+
"step": 732
|
| 5166 |
+
},
|
| 5167 |
+
{
|
| 5168 |
+
"epoch": 0.9142500779544745,
|
| 5169 |
+
"grad_norm": NaN,
|
| 5170 |
+
"learning_rate": 4.018622893472912e-06,
|
| 5171 |
+
"loss": 0.0,
|
| 5172 |
+
"step": 733
|
| 5173 |
+
},
|
| 5174 |
+
{
|
| 5175 |
+
"epoch": 0.915497349547864,
|
| 5176 |
+
"grad_norm": NaN,
|
| 5177 |
+
"learning_rate": 3.902066315979158e-06,
|
| 5178 |
+
"loss": 0.0,
|
| 5179 |
+
"step": 734
|
| 5180 |
+
},
|
| 5181 |
+
{
|
| 5182 |
+
"epoch": 0.9167446211412535,
|
| 5183 |
+
"grad_norm": NaN,
|
| 5184 |
+
"learning_rate": 3.787191379486288e-06,
|
| 5185 |
+
"loss": 0.0,
|
| 5186 |
+
"step": 735
|
| 5187 |
+
},
|
| 5188 |
+
{
|
| 5189 |
+
"epoch": 0.917991892734643,
|
| 5190 |
+
"grad_norm": NaN,
|
| 5191 |
+
"learning_rate": 3.674000094218577e-06,
|
| 5192 |
+
"loss": 0.0,
|
| 5193 |
+
"step": 736
|
| 5194 |
+
},
|
| 5195 |
+
{
|
| 5196 |
+
"epoch": 0.9192391643280324,
|
| 5197 |
+
"grad_norm": NaN,
|
| 5198 |
+
"learning_rate": 3.562494440937769e-06,
|
| 5199 |
+
"loss": 0.0,
|
| 5200 |
+
"step": 737
|
| 5201 |
+
},
|
| 5202 |
+
{
|
| 5203 |
+
"epoch": 0.9204864359214219,
|
| 5204 |
+
"grad_norm": NaN,
|
| 5205 |
+
"learning_rate": 3.4526763709082476e-06,
|
| 5206 |
+
"loss": 0.0,
|
| 5207 |
+
"step": 738
|
| 5208 |
+
},
|
| 5209 |
+
{
|
| 5210 |
+
"epoch": 0.9217337075148113,
|
| 5211 |
+
"grad_norm": NaN,
|
| 5212 |
+
"learning_rate": 3.344547805862985e-06,
|
| 5213 |
+
"loss": 0.0,
|
| 5214 |
+
"step": 739
|
| 5215 |
+
},
|
| 5216 |
+
{
|
| 5217 |
+
"epoch": 0.9229809791082008,
|
| 5218 |
+
"grad_norm": NaN,
|
| 5219 |
+
"learning_rate": 3.2381106379699488e-06,
|
| 5220 |
+
"loss": 0.0,
|
| 5221 |
+
"step": 740
|
| 5222 |
+
},
|
| 5223 |
+
{
|
| 5224 |
+
"epoch": 0.9242282507015903,
|
| 5225 |
+
"grad_norm": NaN,
|
| 5226 |
+
"learning_rate": 3.1333667297989035e-06,
|
| 5227 |
+
"loss": 0.0,
|
| 5228 |
+
"step": 741
|
| 5229 |
+
},
|
| 5230 |
+
{
|
| 5231 |
+
"epoch": 0.9254755222949798,
|
| 5232 |
+
"grad_norm": NaN,
|
| 5233 |
+
"learning_rate": 3.030317914288816e-06,
|
| 5234 |
+
"loss": 0.0,
|
| 5235 |
+
"step": 742
|
| 5236 |
+
},
|
| 5237 |
+
{
|
| 5238 |
+
"epoch": 0.9267227938883692,
|
| 5239 |
+
"grad_norm": NaN,
|
| 5240 |
+
"learning_rate": 2.928965994715882e-06,
|
| 5241 |
+
"loss": 0.0,
|
| 5242 |
+
"step": 743
|
| 5243 |
+
},
|
| 5244 |
+
{
|
| 5245 |
+
"epoch": 0.9279700654817586,
|
| 5246 |
+
"grad_norm": NaN,
|
| 5247 |
+
"learning_rate": 2.8293127446618383e-06,
|
| 5248 |
+
"loss": 0.0,
|
| 5249 |
+
"step": 744
|
| 5250 |
+
},
|
| 5251 |
+
{
|
| 5252 |
+
"epoch": 0.9292173370751481,
|
| 5253 |
+
"grad_norm": NaN,
|
| 5254 |
+
"learning_rate": 2.7313599079830666e-06,
|
| 5255 |
+
"loss": 0.0,
|
| 5256 |
+
"step": 745
|
| 5257 |
+
},
|
| 5258 |
+
{
|
| 5259 |
+
"epoch": 0.9304646086685375,
|
| 5260 |
+
"grad_norm": NaN,
|
| 5261 |
+
"learning_rate": 2.63510919877995e-06,
|
| 5262 |
+
"loss": 0.0,
|
| 5263 |
+
"step": 746
|
| 5264 |
+
},
|
| 5265 |
+
{
|
| 5266 |
+
"epoch": 0.9317118802619271,
|
| 5267 |
+
"grad_norm": NaN,
|
| 5268 |
+
"learning_rate": 2.540562301366922e-06,
|
| 5269 |
+
"loss": 0.0,
|
| 5270 |
+
"step": 747
|
| 5271 |
+
},
|
| 5272 |
+
{
|
| 5273 |
+
"epoch": 0.9329591518553165,
|
| 5274 |
+
"grad_norm": NaN,
|
| 5275 |
+
"learning_rate": 2.447720870243064e-06,
|
| 5276 |
+
"loss": 0.0,
|
| 5277 |
+
"step": 748
|
| 5278 |
+
},
|
| 5279 |
+
{
|
| 5280 |
+
"epoch": 0.934206423448706,
|
| 5281 |
+
"grad_norm": NaN,
|
| 5282 |
+
"learning_rate": 2.3565865300630206e-06,
|
| 5283 |
+
"loss": 0.0,
|
| 5284 |
+
"step": 749
|
| 5285 |
+
},
|
| 5286 |
+
{
|
| 5287 |
+
"epoch": 0.9354536950420954,
|
| 5288 |
+
"grad_norm": NaN,
|
| 5289 |
+
"learning_rate": 2.267160875608687e-06,
|
| 5290 |
+
"loss": 0.0,
|
| 5291 |
+
"step": 750
|
| 5292 |
+
},
|
| 5293 |
+
{
|
| 5294 |
+
"epoch": 0.9367009666354849,
|
| 5295 |
+
"grad_norm": NaN,
|
| 5296 |
+
"learning_rate": 2.179445471761221e-06,
|
| 5297 |
+
"loss": 0.0,
|
| 5298 |
+
"step": 751
|
| 5299 |
+
},
|
| 5300 |
+
{
|
| 5301 |
+
"epoch": 0.9379482382288743,
|
| 5302 |
+
"grad_norm": NaN,
|
| 5303 |
+
"learning_rate": 2.0934418534737098e-06,
|
| 5304 |
+
"loss": 0.0,
|
| 5305 |
+
"step": 752
|
| 5306 |
+
},
|
| 5307 |
+
{
|
| 5308 |
+
"epoch": 0.9391955098222639,
|
| 5309 |
+
"grad_norm": NaN,
|
| 5310 |
+
"learning_rate": 2.0091515257442904e-06,
|
| 5311 |
+
"loss": 0.0,
|
| 5312 |
+
"step": 753
|
| 5313 |
+
},
|
| 5314 |
+
{
|
| 5315 |
+
"epoch": 0.9404427814156533,
|
| 5316 |
+
"grad_norm": NaN,
|
| 5317 |
+
"learning_rate": 1.926575963589805e-06,
|
| 5318 |
+
"loss": 0.0,
|
| 5319 |
+
"step": 754
|
| 5320 |
+
},
|
| 5321 |
+
{
|
| 5322 |
+
"epoch": 0.9416900530090427,
|
| 5323 |
+
"grad_norm": NaN,
|
| 5324 |
+
"learning_rate": 1.8457166120199987e-06,
|
| 5325 |
+
"loss": 0.0,
|
| 5326 |
+
"step": 755
|
| 5327 |
+
},
|
| 5328 |
+
{
|
| 5329 |
+
"epoch": 0.9429373246024322,
|
| 5330 |
+
"grad_norm": NaN,
|
| 5331 |
+
"learning_rate": 1.7665748860122512e-06,
|
| 5332 |
+
"loss": 0.0,
|
| 5333 |
+
"step": 756
|
| 5334 |
+
},
|
| 5335 |
+
{
|
| 5336 |
+
"epoch": 0.9441845961958216,
|
| 5337 |
+
"grad_norm": NaN,
|
| 5338 |
+
"learning_rate": 1.689152170486752e-06,
|
| 5339 |
+
"loss": 0.0,
|
| 5340 |
+
"step": 757
|
| 5341 |
+
},
|
| 5342 |
+
{
|
| 5343 |
+
"epoch": 0.9454318677892111,
|
| 5344 |
+
"grad_norm": NaN,
|
| 5345 |
+
"learning_rate": 1.6134498202823645e-06,
|
| 5346 |
+
"loss": 0.0,
|
| 5347 |
+
"step": 758
|
| 5348 |
+
},
|
| 5349 |
+
{
|
| 5350 |
+
"epoch": 0.9466791393826005,
|
| 5351 |
+
"grad_norm": NaN,
|
| 5352 |
+
"learning_rate": 1.5394691601328338e-06,
|
| 5353 |
+
"loss": 0.0,
|
| 5354 |
+
"step": 759
|
| 5355 |
+
},
|
| 5356 |
+
{
|
| 5357 |
+
"epoch": 0.9479264109759901,
|
| 5358 |
+
"grad_norm": NaN,
|
| 5359 |
+
"learning_rate": 1.467211484643627e-06,
|
| 5360 |
+
"loss": 0.0,
|
| 5361 |
+
"step": 760
|
| 5362 |
+
},
|
| 5363 |
+
{
|
| 5364 |
+
"epoch": 0.9491736825693795,
|
| 5365 |
+
"grad_norm": NaN,
|
| 5366 |
+
"learning_rate": 1.3966780582693185e-06,
|
| 5367 |
+
"loss": 0.0,
|
| 5368 |
+
"step": 761
|
| 5369 |
+
},
|
| 5370 |
+
{
|
| 5371 |
+
"epoch": 0.950420954162769,
|
| 5372 |
+
"grad_norm": NaN,
|
| 5373 |
+
"learning_rate": 1.3278701152913742e-06,
|
| 5374 |
+
"loss": 0.0,
|
| 5375 |
+
"step": 762
|
| 5376 |
+
},
|
| 5377 |
+
{
|
| 5378 |
+
"epoch": 0.9516682257561584,
|
| 5379 |
+
"grad_norm": NaN,
|
| 5380 |
+
"learning_rate": 1.2607888597966688e-06,
|
| 5381 |
+
"loss": 0.0,
|
| 5382 |
+
"step": 763
|
| 5383 |
+
},
|
| 5384 |
+
{
|
| 5385 |
+
"epoch": 0.9529154973495478,
|
| 5386 |
+
"grad_norm": NaN,
|
| 5387 |
+
"learning_rate": 1.195435465656325e-06,
|
| 5388 |
+
"loss": 0.0,
|
| 5389 |
+
"step": 764
|
| 5390 |
+
},
|
| 5391 |
+
{
|
| 5392 |
+
"epoch": 0.9541627689429373,
|
| 5393 |
+
"grad_norm": NaN,
|
| 5394 |
+
"learning_rate": 1.131811076505196e-06,
|
| 5395 |
+
"loss": 0.0,
|
| 5396 |
+
"step": 765
|
| 5397 |
+
},
|
| 5398 |
+
{
|
| 5399 |
+
"epoch": 0.9554100405363268,
|
| 5400 |
+
"grad_norm": NaN,
|
| 5401 |
+
"learning_rate": 1.0699168057218823e-06,
|
| 5402 |
+
"loss": 0.0,
|
| 5403 |
+
"step": 766
|
| 5404 |
+
},
|
| 5405 |
+
{
|
| 5406 |
+
"epoch": 0.9566573121297163,
|
| 5407 |
+
"grad_norm": NaN,
|
| 5408 |
+
"learning_rate": 1.0097537364091914e-06,
|
| 5409 |
+
"loss": 0.0,
|
| 5410 |
+
"step": 767
|
| 5411 |
+
},
|
| 5412 |
+
{
|
| 5413 |
+
"epoch": 0.9579045837231057,
|
| 5414 |
+
"grad_norm": NaN,
|
| 5415 |
+
"learning_rate": 9.513229213752417e-07,
|
| 5416 |
+
"loss": 0.0,
|
| 5417 |
+
"step": 768
|
| 5418 |
+
},
|
| 5419 |
+
{
|
| 5420 |
+
"epoch": 0.9591518553164952,
|
| 5421 |
+
"grad_norm": NaN,
|
| 5422 |
+
"learning_rate": 8.946253831150109e-07,
|
| 5423 |
+
"loss": 0.0,
|
| 5424 |
+
"step": 769
|
| 5425 |
+
},
|
| 5426 |
+
{
|
| 5427 |
+
"epoch": 0.9603991269098846,
|
| 5428 |
+
"grad_norm": NaN,
|
| 5429 |
+
"learning_rate": 8.396621137924388e-07,
|
| 5430 |
+
"loss": 0.0,
|
| 5431 |
+
"step": 770
|
| 5432 |
+
},
|
| 5433 |
+
{
|
| 5434 |
+
"epoch": 0.961646398503274,
|
| 5435 |
+
"grad_norm": NaN,
|
| 5436 |
+
"learning_rate": 7.864340752230859e-07,
|
| 5437 |
+
"loss": 0.0,
|
| 5438 |
+
"step": 771
|
| 5439 |
+
},
|
| 5440 |
+
{
|
| 5441 |
+
"epoch": 0.9628936700966636,
|
| 5442 |
+
"grad_norm": NaN,
|
| 5443 |
+
"learning_rate": 7.349421988572691e-07,
|
| 5444 |
+
"loss": 0.0,
|
| 5445 |
+
"step": 772
|
| 5446 |
+
},
|
| 5447 |
+
{
|
| 5448 |
+
"epoch": 0.964140941690053,
|
| 5449 |
+
"grad_norm": NaN,
|
| 5450 |
+
"learning_rate": 6.851873857638192e-07,
|
| 5451 |
+
"loss": 0.0,
|
| 5452 |
+
"step": 773
|
| 5453 |
+
},
|
| 5454 |
+
{
|
| 5455 |
+
"epoch": 0.9653882132834425,
|
| 5456 |
+
"grad_norm": NaN,
|
| 5457 |
+
"learning_rate": 6.371705066142264e-07,
|
| 5458 |
+
"loss": 0.0,
|
| 5459 |
+
"step": 774
|
| 5460 |
+
},
|
| 5461 |
+
{
|
| 5462 |
+
"epoch": 0.9666354848768319,
|
| 5463 |
+
"grad_norm": NaN,
|
| 5464 |
+
"learning_rate": 5.908924016674977e-07,
|
| 5465 |
+
"loss": 0.0,
|
| 5466 |
+
"step": 775
|
| 5467 |
+
},
|
| 5468 |
+
{
|
| 5469 |
+
"epoch": 0.9678827564702214,
|
| 5470 |
+
"grad_norm": NaN,
|
| 5471 |
+
"learning_rate": 5.463538807553903e-07,
|
| 5472 |
+
"loss": 0.0,
|
| 5473 |
+
"step": 776
|
| 5474 |
+
},
|
| 5475 |
+
{
|
| 5476 |
+
"epoch": 0.9691300280636108,
|
| 5477 |
+
"grad_norm": NaN,
|
| 5478 |
+
"learning_rate": 5.035557232682564e-07,
|
| 5479 |
+
"loss": 0.0,
|
| 5480 |
+
"step": 777
|
| 5481 |
+
},
|
| 5482 |
+
{
|
| 5483 |
+
"epoch": 0.9703772996570003,
|
| 5484 |
+
"grad_norm": NaN,
|
| 5485 |
+
"learning_rate": 4.624986781414098e-07,
|
| 5486 |
+
"loss": 0.0,
|
| 5487 |
+
"step": 778
|
| 5488 |
+
},
|
| 5489 |
+
{
|
| 5490 |
+
"epoch": 0.9716245712503898,
|
| 5491 |
+
"grad_norm": NaN,
|
| 5492 |
+
"learning_rate": 4.231834638420362e-07,
|
| 5493 |
+
"loss": 0.0,
|
| 5494 |
+
"step": 779
|
| 5495 |
+
},
|
| 5496 |
+
{
|
| 5497 |
+
"epoch": 0.9728718428437793,
|
| 5498 |
+
"grad_norm": NaN,
|
| 5499 |
+
"learning_rate": 3.8561076835657017e-07,
|
| 5500 |
+
"loss": 0.0,
|
| 5501 |
+
"step": 780
|
| 5502 |
+
},
|
| 5503 |
+
{
|
| 5504 |
+
"epoch": 0.9741191144371687,
|
| 5505 |
+
"grad_norm": NaN,
|
| 5506 |
+
"learning_rate": 3.4978124917871556e-07,
|
| 5507 |
+
"loss": 0.0,
|
| 5508 |
+
"step": 781
|
| 5509 |
+
},
|
| 5510 |
+
{
|
| 5511 |
+
"epoch": 0.9753663860305581,
|
| 5512 |
+
"grad_norm": NaN,
|
| 5513 |
+
"learning_rate": 3.1569553329788836e-07,
|
| 5514 |
+
"loss": 0.0,
|
| 5515 |
+
"step": 782
|
| 5516 |
+
},
|
| 5517 |
+
{
|
| 5518 |
+
"epoch": 0.9766136576239476,
|
| 5519 |
+
"grad_norm": NaN,
|
| 5520 |
+
"learning_rate": 2.8335421718829193e-07,
|
| 5521 |
+
"loss": 0.0,
|
| 5522 |
+
"step": 783
|
| 5523 |
+
},
|
| 5524 |
+
{
|
| 5525 |
+
"epoch": 0.977860929217337,
|
| 5526 |
+
"grad_norm": NaN,
|
| 5527 |
+
"learning_rate": 2.527578667984365e-07,
|
| 5528 |
+
"loss": 0.0,
|
| 5529 |
+
"step": 784
|
| 5530 |
+
},
|
| 5531 |
+
{
|
| 5532 |
+
"epoch": 0.9791082008107266,
|
| 5533 |
+
"grad_norm": NaN,
|
| 5534 |
+
"learning_rate": 2.239070175412694e-07,
|
| 5535 |
+
"loss": 0.0,
|
| 5536 |
+
"step": 785
|
| 5537 |
+
},
|
| 5538 |
+
{
|
| 5539 |
+
"epoch": 0.980355472404116,
|
| 5540 |
+
"grad_norm": NaN,
|
| 5541 |
+
"learning_rate": 1.9680217428479364e-07,
|
| 5542 |
+
"loss": 0.0,
|
| 5543 |
+
"step": 786
|
| 5544 |
+
},
|
| 5545 |
+
{
|
| 5546 |
+
"epoch": 0.9816027439975055,
|
| 5547 |
+
"grad_norm": NaN,
|
| 5548 |
+
"learning_rate": 1.714438113431971e-07,
|
| 5549 |
+
"loss": 0.0,
|
| 5550 |
+
"step": 787
|
| 5551 |
+
},
|
| 5552 |
+
{
|
| 5553 |
+
"epoch": 0.9828500155908949,
|
| 5554 |
+
"grad_norm": NaN,
|
| 5555 |
+
"learning_rate": 1.4783237246862592e-07,
|
| 5556 |
+
"loss": 0.0,
|
| 5557 |
+
"step": 788
|
| 5558 |
+
},
|
| 5559 |
+
{
|
| 5560 |
+
"epoch": 0.9840972871842844,
|
| 5561 |
+
"grad_norm": NaN,
|
| 5562 |
+
"learning_rate": 1.259682708433574e-07,
|
| 5563 |
+
"loss": 0.0,
|
| 5564 |
+
"step": 789
|
| 5565 |
+
},
|
| 5566 |
+
{
|
| 5567 |
+
"epoch": 0.9853445587776738,
|
| 5568 |
+
"grad_norm": NaN,
|
| 5569 |
+
"learning_rate": 1.0585188907260569e-07,
|
| 5570 |
+
"loss": 0.0,
|
| 5571 |
+
"step": 790
|
| 5572 |
+
},
|
| 5573 |
+
{
|
| 5574 |
+
"epoch": 0.9865918303710633,
|
| 5575 |
+
"grad_norm": NaN,
|
| 5576 |
+
"learning_rate": 8.748357917780503e-08,
|
| 5577 |
+
"loss": 0.0,
|
| 5578 |
+
"step": 791
|
| 5579 |
+
},
|
| 5580 |
+
{
|
| 5581 |
+
"epoch": 0.9878391019644528,
|
| 5582 |
+
"grad_norm": NaN,
|
| 5583 |
+
"learning_rate": 7.086366259044796e-08,
|
| 5584 |
+
"loss": 0.0,
|
| 5585 |
+
"step": 792
|
| 5586 |
+
},
|
| 5587 |
+
{
|
| 5588 |
+
"epoch": 0.9890863735578422,
|
| 5589 |
+
"grad_norm": NaN,
|
| 5590 |
+
"learning_rate": 5.59924301464898e-08,
|
| 5591 |
+
"loss": 0.0,
|
| 5592 |
+
"step": 793
|
| 5593 |
+
},
|
| 5594 |
+
{
|
| 5595 |
+
"epoch": 0.9903336451512317,
|
| 5596 |
+
"grad_norm": NaN,
|
| 5597 |
+
"learning_rate": 4.287014208120832e-08,
|
| 5598 |
+
"loss": 0.0,
|
| 5599 |
+
"step": 794
|
| 5600 |
+
},
|
| 5601 |
+
{
|
| 5602 |
+
"epoch": 0.9915809167446211,
|
| 5603 |
+
"grad_norm": NaN,
|
| 5604 |
+
"learning_rate": 3.149702802470733e-08,
|
| 5605 |
+
"loss": 0.0,
|
| 5606 |
+
"step": 795
|
| 5607 |
+
},
|
| 5608 |
+
{
|
| 5609 |
+
"epoch": 0.9928281883380106,
|
| 5610 |
+
"grad_norm": NaN,
|
| 5611 |
+
"learning_rate": 2.1873286997875498e-08,
|
| 5612 |
+
"loss": 0.0,
|
| 5613 |
+
"step": 796
|
| 5614 |
+
},
|
| 5615 |
+
{
|
| 5616 |
+
"epoch": 0.9940754599314,
|
| 5617 |
+
"grad_norm": NaN,
|
| 5618 |
+
"learning_rate": 1.3999087408866906e-08,
|
| 5619 |
+
"loss": 0.0,
|
| 5620 |
+
"step": 797
|
| 5621 |
+
},
|
| 5622 |
+
{
|
| 5623 |
+
"epoch": 0.9953227315247896,
|
| 5624 |
+
"grad_norm": NaN,
|
| 5625 |
+
"learning_rate": 7.874567050214499e-09,
|
| 5626 |
+
"loss": 0.0,
|
| 5627 |
+
"step": 798
|
| 5628 |
+
},
|
| 5629 |
+
{
|
| 5630 |
+
"epoch": 0.996570003118179,
|
| 5631 |
+
"grad_norm": NaN,
|
| 5632 |
+
"learning_rate": 3.4998330963764705e-09,
|
| 5633 |
+
"loss": 0.0,
|
| 5634 |
+
"step": 799
|
| 5635 |
+
},
|
| 5636 |
+
{
|
| 5637 |
+
"epoch": 0.9978172747115684,
|
| 5638 |
+
"grad_norm": NaN,
|
| 5639 |
+
"learning_rate": 8.749621018822041e-10,
|
| 5640 |
+
"loss": 0.0,
|
| 5641 |
+
"step": 800
|
| 5642 |
+
},
|
| 5643 |
+
{
|
| 5644 |
+
"epoch": 0.9990645463049579,
|
| 5645 |
+
"grad_norm": NaN,
|
| 5646 |
+
"learning_rate": 0.0,
|
| 5647 |
+
"loss": 0.0,
|
| 5648 |
+
"step": 801
|
| 5649 |
}
|
| 5650 |
],
|
| 5651 |
"logging_steps": 1,
|
|
|
|
| 5660 |
"should_evaluate": false,
|
| 5661 |
"should_log": false,
|
| 5662 |
"should_save": true,
|
| 5663 |
+
"should_training_stop": true
|
| 5664 |
},
|
| 5665 |
"attributes": {}
|
| 5666 |
}
|
| 5667 |
},
|
| 5668 |
+
"total_flos": 3.299641599763415e+17,
|
| 5669 |
"train_batch_size": 2,
|
| 5670 |
"trial_name": null,
|
| 5671 |
"trial_params": null
|