Training in progress, step 813, 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 50503544
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:955a7736768d247a560a9140fb276728dbf3ce8a247f711ae94ba90d6474d8a5
|
| 3 |
size 50503544
|
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 sha256:6c82abff08445366f36490751ea0383991f39d4a02ce9e61d511211b0344d4d7
|
| 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 sha256:14d88616e2b1a8f39174a04caaef9eb4ea5a1a57d4aa5d8c386b7412fc210885
|
| 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 sha256:9d410a86c567093ae7e3350b0cb4f76b6d6af0c68f2cfaab1e7764a570593615
|
| 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:d0c0b78924a966d0ff851c4df819a0fe6307f3bd376b9608c06fab58c12d7e42
|
| 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": 204,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -4323,6 +4323,1413 @@
|
|
| 4323 |
"eval_samples_per_second": 10.58,
|
| 4324 |
"eval_steps_per_second": 5.298,
|
| 4325 |
"step": 612
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4326 |
}
|
| 4327 |
],
|
| 4328 |
"logging_steps": 1,
|
|
@@ -4337,12 +5744,12 @@
|
|
| 4337 |
"should_evaluate": false,
|
| 4338 |
"should_log": false,
|
| 4339 |
"should_save": true,
|
| 4340 |
-
"should_training_stop":
|
| 4341 |
},
|
| 4342 |
"attributes": {}
|
| 4343 |
}
|
| 4344 |
},
|
| 4345 |
-
"total_flos": 1.
|
| 4346 |
"train_batch_size": 1,
|
| 4347 |
"trial_name": null,
|
| 4348 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9993853718500307,
|
| 5 |
"eval_steps": 204,
|
| 6 |
+
"global_step": 813,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 4323 |
"eval_samples_per_second": 10.58,
|
| 4324 |
"eval_steps_per_second": 5.298,
|
| 4325 |
"step": 612
|
| 4326 |
+
},
|
| 4327 |
+
{
|
| 4328 |
+
"epoch": 0.7535341118623233,
|
| 4329 |
+
"grad_norm": 0.45276275277137756,
|
| 4330 |
+
"learning_rate": 1.6016523480379382e-05,
|
| 4331 |
+
"loss": 0.8856,
|
| 4332 |
+
"step": 613
|
| 4333 |
+
},
|
| 4334 |
+
{
|
| 4335 |
+
"epoch": 0.7547633681622619,
|
| 4336 |
+
"grad_norm": 0.6476858854293823,
|
| 4337 |
+
"learning_rate": 1.5865801734305668e-05,
|
| 4338 |
+
"loss": 0.8316,
|
| 4339 |
+
"step": 614
|
| 4340 |
+
},
|
| 4341 |
+
{
|
| 4342 |
+
"epoch": 0.7559926244622004,
|
| 4343 |
+
"grad_norm": 0.6815754771232605,
|
| 4344 |
+
"learning_rate": 1.5715658669950634e-05,
|
| 4345 |
+
"loss": 0.6487,
|
| 4346 |
+
"step": 615
|
| 4347 |
+
},
|
| 4348 |
+
{
|
| 4349 |
+
"epoch": 0.7572218807621389,
|
| 4350 |
+
"grad_norm": 0.7536165118217468,
|
| 4351 |
+
"learning_rate": 1.5566096832710154e-05,
|
| 4352 |
+
"loss": 0.9835,
|
| 4353 |
+
"step": 616
|
| 4354 |
+
},
|
| 4355 |
+
{
|
| 4356 |
+
"epoch": 0.7584511370620775,
|
| 4357 |
+
"grad_norm": 1.0182470083236694,
|
| 4358 |
+
"learning_rate": 1.541711875812641e-05,
|
| 4359 |
+
"loss": 1.172,
|
| 4360 |
+
"step": 617
|
| 4361 |
+
},
|
| 4362 |
+
{
|
| 4363 |
+
"epoch": 0.759680393362016,
|
| 4364 |
+
"grad_norm": 2.240806818008423,
|
| 4365 |
+
"learning_rate": 1.5268726971845037e-05,
|
| 4366 |
+
"loss": 1.3291,
|
| 4367 |
+
"step": 618
|
| 4368 |
+
},
|
| 4369 |
+
{
|
| 4370 |
+
"epoch": 0.7609096496619545,
|
| 4371 |
+
"grad_norm": 3.0141806602478027,
|
| 4372 |
+
"learning_rate": 1.5120923989572244e-05,
|
| 4373 |
+
"loss": 0.8101,
|
| 4374 |
+
"step": 619
|
| 4375 |
+
},
|
| 4376 |
+
{
|
| 4377 |
+
"epoch": 0.762138905961893,
|
| 4378 |
+
"grad_norm": 3.7101502418518066,
|
| 4379 |
+
"learning_rate": 1.4973712317032135e-05,
|
| 4380 |
+
"loss": 1.0581,
|
| 4381 |
+
"step": 620
|
| 4382 |
+
},
|
| 4383 |
+
{
|
| 4384 |
+
"epoch": 0.7633681622618316,
|
| 4385 |
+
"grad_norm": 3.892627000808716,
|
| 4386 |
+
"learning_rate": 1.482709444992425e-05,
|
| 4387 |
+
"loss": 1.007,
|
| 4388 |
+
"step": 621
|
| 4389 |
+
},
|
| 4390 |
+
{
|
| 4391 |
+
"epoch": 0.7645974185617701,
|
| 4392 |
+
"grad_norm": 3.3281090259552,
|
| 4393 |
+
"learning_rate": 1.4681072873881312e-05,
|
| 4394 |
+
"loss": 0.7403,
|
| 4395 |
+
"step": 622
|
| 4396 |
+
},
|
| 4397 |
+
{
|
| 4398 |
+
"epoch": 0.7658266748617086,
|
| 4399 |
+
"grad_norm": 3.755115509033203,
|
| 4400 |
+
"learning_rate": 1.4535650064427003e-05,
|
| 4401 |
+
"loss": 0.9939,
|
| 4402 |
+
"step": 623
|
| 4403 |
+
},
|
| 4404 |
+
{
|
| 4405 |
+
"epoch": 0.7670559311616472,
|
| 4406 |
+
"grad_norm": 2.45129132270813,
|
| 4407 |
+
"learning_rate": 1.439082848693406e-05,
|
| 4408 |
+
"loss": 0.6902,
|
| 4409 |
+
"step": 624
|
| 4410 |
+
},
|
| 4411 |
+
{
|
| 4412 |
+
"epoch": 0.7682851874615857,
|
| 4413 |
+
"grad_norm": 3.0480661392211914,
|
| 4414 |
+
"learning_rate": 1.4246610596582444e-05,
|
| 4415 |
+
"loss": 0.6467,
|
| 4416 |
+
"step": 625
|
| 4417 |
+
},
|
| 4418 |
+
{
|
| 4419 |
+
"epoch": 0.7695144437615242,
|
| 4420 |
+
"grad_norm": 0.2060498744249344,
|
| 4421 |
+
"learning_rate": 1.41029988383177e-05,
|
| 4422 |
+
"loss": 0.9195,
|
| 4423 |
+
"step": 626
|
| 4424 |
+
},
|
| 4425 |
+
{
|
| 4426 |
+
"epoch": 0.7707437000614629,
|
| 4427 |
+
"grad_norm": 0.21947383880615234,
|
| 4428 |
+
"learning_rate": 1.3959995646809549e-05,
|
| 4429 |
+
"loss": 1.0644,
|
| 4430 |
+
"step": 627
|
| 4431 |
+
},
|
| 4432 |
+
{
|
| 4433 |
+
"epoch": 0.7719729563614014,
|
| 4434 |
+
"grad_norm": 0.26282572746276855,
|
| 4435 |
+
"learning_rate": 1.381760344641061e-05,
|
| 4436 |
+
"loss": 1.3907,
|
| 4437 |
+
"step": 628
|
| 4438 |
+
},
|
| 4439 |
+
{
|
| 4440 |
+
"epoch": 0.7732022126613399,
|
| 4441 |
+
"grad_norm": 0.2563353180885315,
|
| 4442 |
+
"learning_rate": 1.3675824651115276e-05,
|
| 4443 |
+
"loss": 1.2281,
|
| 4444 |
+
"step": 629
|
| 4445 |
+
},
|
| 4446 |
+
{
|
| 4447 |
+
"epoch": 0.7744314689612785,
|
| 4448 |
+
"grad_norm": 0.3003043532371521,
|
| 4449 |
+
"learning_rate": 1.3534661664518817e-05,
|
| 4450 |
+
"loss": 1.2247,
|
| 4451 |
+
"step": 630
|
| 4452 |
+
},
|
| 4453 |
+
{
|
| 4454 |
+
"epoch": 0.775660725261217,
|
| 4455 |
+
"grad_norm": 0.30270689725875854,
|
| 4456 |
+
"learning_rate": 1.339411687977657e-05,
|
| 4457 |
+
"loss": 1.3467,
|
| 4458 |
+
"step": 631
|
| 4459 |
+
},
|
| 4460 |
+
{
|
| 4461 |
+
"epoch": 0.7768899815611555,
|
| 4462 |
+
"grad_norm": 0.3199015259742737,
|
| 4463 |
+
"learning_rate": 1.325419267956346e-05,
|
| 4464 |
+
"loss": 1.1064,
|
| 4465 |
+
"step": 632
|
| 4466 |
+
},
|
| 4467 |
+
{
|
| 4468 |
+
"epoch": 0.778119237861094,
|
| 4469 |
+
"grad_norm": 0.33701401948928833,
|
| 4470 |
+
"learning_rate": 1.3114891436033522e-05,
|
| 4471 |
+
"loss": 1.1541,
|
| 4472 |
+
"step": 633
|
| 4473 |
+
},
|
| 4474 |
+
{
|
| 4475 |
+
"epoch": 0.7793484941610326,
|
| 4476 |
+
"grad_norm": 0.3596295714378357,
|
| 4477 |
+
"learning_rate": 1.2976215510779755e-05,
|
| 4478 |
+
"loss": 1.0702,
|
| 4479 |
+
"step": 634
|
| 4480 |
+
},
|
| 4481 |
+
{
|
| 4482 |
+
"epoch": 0.7805777504609711,
|
| 4483 |
+
"grad_norm": 0.4676484167575836,
|
| 4484 |
+
"learning_rate": 1.2838167254794004e-05,
|
| 4485 |
+
"loss": 0.9612,
|
| 4486 |
+
"step": 635
|
| 4487 |
+
},
|
| 4488 |
+
{
|
| 4489 |
+
"epoch": 0.7818070067609096,
|
| 4490 |
+
"grad_norm": 0.43727314472198486,
|
| 4491 |
+
"learning_rate": 1.2700749008427205e-05,
|
| 4492 |
+
"loss": 1.171,
|
| 4493 |
+
"step": 636
|
| 4494 |
+
},
|
| 4495 |
+
{
|
| 4496 |
+
"epoch": 0.7830362630608482,
|
| 4497 |
+
"grad_norm": 0.6096376180648804,
|
| 4498 |
+
"learning_rate": 1.2563963101349619e-05,
|
| 4499 |
+
"loss": 1.2183,
|
| 4500 |
+
"step": 637
|
| 4501 |
+
},
|
| 4502 |
+
{
|
| 4503 |
+
"epoch": 0.7842655193607867,
|
| 4504 |
+
"grad_norm": 0.7051495909690857,
|
| 4505 |
+
"learning_rate": 1.2427811852511395e-05,
|
| 4506 |
+
"loss": 0.7788,
|
| 4507 |
+
"step": 638
|
| 4508 |
+
},
|
| 4509 |
+
{
|
| 4510 |
+
"epoch": 0.7854947756607252,
|
| 4511 |
+
"grad_norm": 0.6735429167747498,
|
| 4512 |
+
"learning_rate": 1.2292297570103229e-05,
|
| 4513 |
+
"loss": 0.7001,
|
| 4514 |
+
"step": 639
|
| 4515 |
+
},
|
| 4516 |
+
{
|
| 4517 |
+
"epoch": 0.7867240319606638,
|
| 4518 |
+
"grad_norm": 0.740139901638031,
|
| 4519 |
+
"learning_rate": 1.2157422551517228e-05,
|
| 4520 |
+
"loss": 0.8659,
|
| 4521 |
+
"step": 640
|
| 4522 |
+
},
|
| 4523 |
+
{
|
| 4524 |
+
"epoch": 0.7879532882606023,
|
| 4525 |
+
"grad_norm": 0.8181758522987366,
|
| 4526 |
+
"learning_rate": 1.202318908330795e-05,
|
| 4527 |
+
"loss": 1.1304,
|
| 4528 |
+
"step": 641
|
| 4529 |
+
},
|
| 4530 |
+
{
|
| 4531 |
+
"epoch": 0.7891825445605408,
|
| 4532 |
+
"grad_norm": 1.1049834489822388,
|
| 4533 |
+
"learning_rate": 1.188959944115372e-05,
|
| 4534 |
+
"loss": 1.0549,
|
| 4535 |
+
"step": 642
|
| 4536 |
+
},
|
| 4537 |
+
{
|
| 4538 |
+
"epoch": 0.7904118008604795,
|
| 4539 |
+
"grad_norm": 2.664985418319702,
|
| 4540 |
+
"learning_rate": 1.1756655889817953e-05,
|
| 4541 |
+
"loss": 1.2242,
|
| 4542 |
+
"step": 643
|
| 4543 |
+
},
|
| 4544 |
+
{
|
| 4545 |
+
"epoch": 0.791641057160418,
|
| 4546 |
+
"grad_norm": 3.3801634311676025,
|
| 4547 |
+
"learning_rate": 1.1624360683110819e-05,
|
| 4548 |
+
"loss": 0.5464,
|
| 4549 |
+
"step": 644
|
| 4550 |
+
},
|
| 4551 |
+
{
|
| 4552 |
+
"epoch": 0.7928703134603565,
|
| 4553 |
+
"grad_norm": 3.139307975769043,
|
| 4554 |
+
"learning_rate": 1.1492716063850973e-05,
|
| 4555 |
+
"loss": 0.9791,
|
| 4556 |
+
"step": 645
|
| 4557 |
+
},
|
| 4558 |
+
{
|
| 4559 |
+
"epoch": 0.794099569760295,
|
| 4560 |
+
"grad_norm": 3.501133918762207,
|
| 4561 |
+
"learning_rate": 1.1361724263827633e-05,
|
| 4562 |
+
"loss": 1.0201,
|
| 4563 |
+
"step": 646
|
| 4564 |
+
},
|
| 4565 |
+
{
|
| 4566 |
+
"epoch": 0.7953288260602336,
|
| 4567 |
+
"grad_norm": 2.9995079040527344,
|
| 4568 |
+
"learning_rate": 1.123138750376262e-05,
|
| 4569 |
+
"loss": 1.0347,
|
| 4570 |
+
"step": 647
|
| 4571 |
+
},
|
| 4572 |
+
{
|
| 4573 |
+
"epoch": 0.7965580823601721,
|
| 4574 |
+
"grad_norm": 3.3414182662963867,
|
| 4575 |
+
"learning_rate": 1.1101707993272825e-05,
|
| 4576 |
+
"loss": 0.912,
|
| 4577 |
+
"step": 648
|
| 4578 |
+
},
|
| 4579 |
+
{
|
| 4580 |
+
"epoch": 0.7977873386601106,
|
| 4581 |
+
"grad_norm": 2.3033857345581055,
|
| 4582 |
+
"learning_rate": 1.097268793083266e-05,
|
| 4583 |
+
"loss": 0.551,
|
| 4584 |
+
"step": 649
|
| 4585 |
+
},
|
| 4586 |
+
{
|
| 4587 |
+
"epoch": 0.7990165949600492,
|
| 4588 |
+
"grad_norm": 2.474606990814209,
|
| 4589 |
+
"learning_rate": 1.084432950373685e-05,
|
| 4590 |
+
"loss": 0.4882,
|
| 4591 |
+
"step": 650
|
| 4592 |
+
},
|
| 4593 |
+
{
|
| 4594 |
+
"epoch": 0.8002458512599877,
|
| 4595 |
+
"grad_norm": 0.21653778851032257,
|
| 4596 |
+
"learning_rate": 1.071663488806331e-05,
|
| 4597 |
+
"loss": 1.1304,
|
| 4598 |
+
"step": 651
|
| 4599 |
+
},
|
| 4600 |
+
{
|
| 4601 |
+
"epoch": 0.8014751075599262,
|
| 4602 |
+
"grad_norm": 0.19837996363639832,
|
| 4603 |
+
"learning_rate": 1.0589606248636292e-05,
|
| 4604 |
+
"loss": 0.8167,
|
| 4605 |
+
"step": 652
|
| 4606 |
+
},
|
| 4607 |
+
{
|
| 4608 |
+
"epoch": 0.8027043638598648,
|
| 4609 |
+
"grad_norm": 0.2383168786764145,
|
| 4610 |
+
"learning_rate": 1.0463245738989636e-05,
|
| 4611 |
+
"loss": 1.0729,
|
| 4612 |
+
"step": 653
|
| 4613 |
+
},
|
| 4614 |
+
{
|
| 4615 |
+
"epoch": 0.8039336201598033,
|
| 4616 |
+
"grad_norm": 0.2854044437408447,
|
| 4617 |
+
"learning_rate": 1.0337555501330281e-05,
|
| 4618 |
+
"loss": 1.1909,
|
| 4619 |
+
"step": 654
|
| 4620 |
+
},
|
| 4621 |
+
{
|
| 4622 |
+
"epoch": 0.8051628764597418,
|
| 4623 |
+
"grad_norm": 0.2671623229980469,
|
| 4624 |
+
"learning_rate": 1.0212537666501976e-05,
|
| 4625 |
+
"loss": 1.2883,
|
| 4626 |
+
"step": 655
|
| 4627 |
+
},
|
| 4628 |
+
{
|
| 4629 |
+
"epoch": 0.8063921327596804,
|
| 4630 |
+
"grad_norm": 0.2792122960090637,
|
| 4631 |
+
"learning_rate": 1.0088194353949137e-05,
|
| 4632 |
+
"loss": 1.0991,
|
| 4633 |
+
"step": 656
|
| 4634 |
+
},
|
| 4635 |
+
{
|
| 4636 |
+
"epoch": 0.8076213890596189,
|
| 4637 |
+
"grad_norm": 0.2957600951194763,
|
| 4638 |
+
"learning_rate": 9.96452767168089e-06,
|
| 4639 |
+
"loss": 1.0984,
|
| 4640 |
+
"step": 657
|
| 4641 |
+
},
|
| 4642 |
+
{
|
| 4643 |
+
"epoch": 0.8088506453595574,
|
| 4644 |
+
"grad_norm": 0.31414932012557983,
|
| 4645 |
+
"learning_rate": 9.841539716235387e-06,
|
| 4646 |
+
"loss": 1.0569,
|
| 4647 |
+
"step": 658
|
| 4648 |
+
},
|
| 4649 |
+
{
|
| 4650 |
+
"epoch": 0.8100799016594961,
|
| 4651 |
+
"grad_norm": 0.3483067452907562,
|
| 4652 |
+
"learning_rate": 9.719232572644187e-06,
|
| 4653 |
+
"loss": 1.0767,
|
| 4654 |
+
"step": 659
|
| 4655 |
+
},
|
| 4656 |
+
{
|
| 4657 |
+
"epoch": 0.8113091579594346,
|
| 4658 |
+
"grad_norm": 0.3564923107624054,
|
| 4659 |
+
"learning_rate": 9.597608314396978e-06,
|
| 4660 |
+
"loss": 1.2318,
|
| 4661 |
+
"step": 660
|
| 4662 |
+
},
|
| 4663 |
+
{
|
| 4664 |
+
"epoch": 0.8125384142593731,
|
| 4665 |
+
"grad_norm": 0.40873420238494873,
|
| 4666 |
+
"learning_rate": 9.476669003406403e-06,
|
| 4667 |
+
"loss": 1.2112,
|
| 4668 |
+
"step": 661
|
| 4669 |
+
},
|
| 4670 |
+
{
|
| 4671 |
+
"epoch": 0.8137676705593117,
|
| 4672 |
+
"grad_norm": 0.4222586452960968,
|
| 4673 |
+
"learning_rate": 9.356416689973108e-06,
|
| 4674 |
+
"loss": 1.066,
|
| 4675 |
+
"step": 662
|
| 4676 |
+
},
|
| 4677 |
+
{
|
| 4678 |
+
"epoch": 0.8149969268592502,
|
| 4679 |
+
"grad_norm": 0.5865674614906311,
|
| 4680 |
+
"learning_rate": 9.236853412750935e-06,
|
| 4681 |
+
"loss": 1.0605,
|
| 4682 |
+
"step": 663
|
| 4683 |
+
},
|
| 4684 |
+
{
|
| 4685 |
+
"epoch": 0.8162261831591887,
|
| 4686 |
+
"grad_norm": 0.7420896291732788,
|
| 4687 |
+
"learning_rate": 9.11798119871245e-06,
|
| 4688 |
+
"loss": 0.7511,
|
| 4689 |
+
"step": 664
|
| 4690 |
+
},
|
| 4691 |
+
{
|
| 4692 |
+
"epoch": 0.8174554394591272,
|
| 4693 |
+
"grad_norm": 0.6792820692062378,
|
| 4694 |
+
"learning_rate": 8.99980206311452e-06,
|
| 4695 |
+
"loss": 0.9264,
|
| 4696 |
+
"step": 665
|
| 4697 |
+
},
|
| 4698 |
+
{
|
| 4699 |
+
"epoch": 0.8186846957590658,
|
| 4700 |
+
"grad_norm": 0.6937956213951111,
|
| 4701 |
+
"learning_rate": 8.882318009464125e-06,
|
| 4702 |
+
"loss": 0.6352,
|
| 4703 |
+
"step": 666
|
| 4704 |
+
},
|
| 4705 |
+
{
|
| 4706 |
+
"epoch": 0.8199139520590043,
|
| 4707 |
+
"grad_norm": 0.7615432739257812,
|
| 4708 |
+
"learning_rate": 8.765531029484476e-06,
|
| 4709 |
+
"loss": 0.9749,
|
| 4710 |
+
"step": 667
|
| 4711 |
+
},
|
| 4712 |
+
{
|
| 4713 |
+
"epoch": 0.8211432083589428,
|
| 4714 |
+
"grad_norm": 1.0325685739517212,
|
| 4715 |
+
"learning_rate": 8.64944310308114e-06,
|
| 4716 |
+
"loss": 1.2544,
|
| 4717 |
+
"step": 668
|
| 4718 |
+
},
|
| 4719 |
+
{
|
| 4720 |
+
"epoch": 0.8223724646588814,
|
| 4721 |
+
"grad_norm": 3.312368631362915,
|
| 4722 |
+
"learning_rate": 8.534056198308582e-06,
|
| 4723 |
+
"loss": 1.1743,
|
| 4724 |
+
"step": 669
|
| 4725 |
+
},
|
| 4726 |
+
{
|
| 4727 |
+
"epoch": 0.8236017209588199,
|
| 4728 |
+
"grad_norm": 2.5028300285339355,
|
| 4729 |
+
"learning_rate": 8.419372271336745e-06,
|
| 4730 |
+
"loss": 0.6261,
|
| 4731 |
+
"step": 670
|
| 4732 |
+
},
|
| 4733 |
+
{
|
| 4734 |
+
"epoch": 0.8248309772587584,
|
| 4735 |
+
"grad_norm": 2.5545358657836914,
|
| 4736 |
+
"learning_rate": 8.305393266417887e-06,
|
| 4737 |
+
"loss": 0.6315,
|
| 4738 |
+
"step": 671
|
| 4739 |
+
},
|
| 4740 |
+
{
|
| 4741 |
+
"epoch": 0.826060233558697,
|
| 4742 |
+
"grad_norm": 2.4186129570007324,
|
| 4743 |
+
"learning_rate": 8.192121115853602e-06,
|
| 4744 |
+
"loss": 0.6056,
|
| 4745 |
+
"step": 672
|
| 4746 |
+
},
|
| 4747 |
+
{
|
| 4748 |
+
"epoch": 0.8272894898586355,
|
| 4749 |
+
"grad_norm": 2.6672956943511963,
|
| 4750 |
+
"learning_rate": 8.079557739962128e-06,
|
| 4751 |
+
"loss": 0.644,
|
| 4752 |
+
"step": 673
|
| 4753 |
+
},
|
| 4754 |
+
{
|
| 4755 |
+
"epoch": 0.828518746158574,
|
| 4756 |
+
"grad_norm": 3.801270008087158,
|
| 4757 |
+
"learning_rate": 7.967705047045715e-06,
|
| 4758 |
+
"loss": 1.0116,
|
| 4759 |
+
"step": 674
|
| 4760 |
+
},
|
| 4761 |
+
{
|
| 4762 |
+
"epoch": 0.8297480024585125,
|
| 4763 |
+
"grad_norm": 2.186417579650879,
|
| 4764 |
+
"learning_rate": 7.856564933358324e-06,
|
| 4765 |
+
"loss": 0.4153,
|
| 4766 |
+
"step": 675
|
| 4767 |
+
},
|
| 4768 |
+
{
|
| 4769 |
+
"epoch": 0.8309772587584512,
|
| 4770 |
+
"grad_norm": 0.2179417610168457,
|
| 4771 |
+
"learning_rate": 7.746139283073473e-06,
|
| 4772 |
+
"loss": 1.0057,
|
| 4773 |
+
"step": 676
|
| 4774 |
+
},
|
| 4775 |
+
{
|
| 4776 |
+
"epoch": 0.8322065150583897,
|
| 4777 |
+
"grad_norm": 0.2457554191350937,
|
| 4778 |
+
"learning_rate": 7.636429968252257e-06,
|
| 4779 |
+
"loss": 1.2948,
|
| 4780 |
+
"step": 677
|
| 4781 |
+
},
|
| 4782 |
+
{
|
| 4783 |
+
"epoch": 0.8334357713583282,
|
| 4784 |
+
"grad_norm": 0.2506256401538849,
|
| 4785 |
+
"learning_rate": 7.527438848811652e-06,
|
| 4786 |
+
"loss": 1.1521,
|
| 4787 |
+
"step": 678
|
| 4788 |
+
},
|
| 4789 |
+
{
|
| 4790 |
+
"epoch": 0.8346650276582668,
|
| 4791 |
+
"grad_norm": 0.25727221369743347,
|
| 4792 |
+
"learning_rate": 7.4191677724929906e-06,
|
| 4793 |
+
"loss": 1.1952,
|
| 4794 |
+
"step": 679
|
| 4795 |
+
},
|
| 4796 |
+
{
|
| 4797 |
+
"epoch": 0.8358942839582053,
|
| 4798 |
+
"grad_norm": 0.265831857919693,
|
| 4799 |
+
"learning_rate": 7.31161857483057e-06,
|
| 4800 |
+
"loss": 1.1952,
|
| 4801 |
+
"step": 680
|
| 4802 |
+
},
|
| 4803 |
+
{
|
| 4804 |
+
"epoch": 0.8371235402581438,
|
| 4805 |
+
"grad_norm": 0.27371159195899963,
|
| 4806 |
+
"learning_rate": 7.204793079120636e-06,
|
| 4807 |
+
"loss": 1.1563,
|
| 4808 |
+
"step": 681
|
| 4809 |
+
},
|
| 4810 |
+
{
|
| 4811 |
+
"epoch": 0.8383527965580824,
|
| 4812 |
+
"grad_norm": 0.30959704518318176,
|
| 4813 |
+
"learning_rate": 7.0986930963903575e-06,
|
| 4814 |
+
"loss": 1.122,
|
| 4815 |
+
"step": 682
|
| 4816 |
+
},
|
| 4817 |
+
{
|
| 4818 |
+
"epoch": 0.8395820528580209,
|
| 4819 |
+
"grad_norm": 0.335997074842453,
|
| 4820 |
+
"learning_rate": 6.993320425367222e-06,
|
| 4821 |
+
"loss": 1.0265,
|
| 4822 |
+
"step": 683
|
| 4823 |
+
},
|
| 4824 |
+
{
|
| 4825 |
+
"epoch": 0.8408113091579594,
|
| 4826 |
+
"grad_norm": 0.3063446879386902,
|
| 4827 |
+
"learning_rate": 6.8886768524485e-06,
|
| 4828 |
+
"loss": 0.8463,
|
| 4829 |
+
"step": 684
|
| 4830 |
+
},
|
| 4831 |
+
{
|
| 4832 |
+
"epoch": 0.842040565457898,
|
| 4833 |
+
"grad_norm": 0.34628668427467346,
|
| 4834 |
+
"learning_rate": 6.7847641516709635e-06,
|
| 4835 |
+
"loss": 0.9597,
|
| 4836 |
+
"step": 685
|
| 4837 |
+
},
|
| 4838 |
+
{
|
| 4839 |
+
"epoch": 0.8432698217578365,
|
| 4840 |
+
"grad_norm": 0.38880714774131775,
|
| 4841 |
+
"learning_rate": 6.681584084680787e-06,
|
| 4842 |
+
"loss": 0.9935,
|
| 4843 |
+
"step": 686
|
| 4844 |
+
},
|
| 4845 |
+
{
|
| 4846 |
+
"epoch": 0.844499078057775,
|
| 4847 |
+
"grad_norm": 0.45289453864097595,
|
| 4848 |
+
"learning_rate": 6.579138400703716e-06,
|
| 4849 |
+
"loss": 0.911,
|
| 4850 |
+
"step": 687
|
| 4851 |
+
},
|
| 4852 |
+
{
|
| 4853 |
+
"epoch": 0.8457283343577136,
|
| 4854 |
+
"grad_norm": 0.697723388671875,
|
| 4855 |
+
"learning_rate": 6.4774288365154035e-06,
|
| 4856 |
+
"loss": 0.9489,
|
| 4857 |
+
"step": 688
|
| 4858 |
+
},
|
| 4859 |
+
{
|
| 4860 |
+
"epoch": 0.8469575906576521,
|
| 4861 |
+
"grad_norm": 0.7846165895462036,
|
| 4862 |
+
"learning_rate": 6.376457116411971e-06,
|
| 4863 |
+
"loss": 0.9332,
|
| 4864 |
+
"step": 689
|
| 4865 |
+
},
|
| 4866 |
+
{
|
| 4867 |
+
"epoch": 0.8481868469575906,
|
| 4868 |
+
"grad_norm": 0.6980641484260559,
|
| 4869 |
+
"learning_rate": 6.2762249521807645e-06,
|
| 4870 |
+
"loss": 0.8282,
|
| 4871 |
+
"step": 690
|
| 4872 |
+
},
|
| 4873 |
+
{
|
| 4874 |
+
"epoch": 0.8494161032575291,
|
| 4875 |
+
"grad_norm": 0.6430233716964722,
|
| 4876 |
+
"learning_rate": 6.17673404307132e-06,
|
| 4877 |
+
"loss": 0.7686,
|
| 4878 |
+
"step": 691
|
| 4879 |
+
},
|
| 4880 |
+
{
|
| 4881 |
+
"epoch": 0.8506453595574678,
|
| 4882 |
+
"grad_norm": 0.7281519174575806,
|
| 4883 |
+
"learning_rate": 6.077986075766612e-06,
|
| 4884 |
+
"loss": 0.9116,
|
| 4885 |
+
"step": 692
|
| 4886 |
+
},
|
| 4887 |
+
{
|
| 4888 |
+
"epoch": 0.8518746158574063,
|
| 4889 |
+
"grad_norm": 1.0536761283874512,
|
| 4890 |
+
"learning_rate": 5.979982724354366e-06,
|
| 4891 |
+
"loss": 1.1009,
|
| 4892 |
+
"step": 693
|
| 4893 |
+
},
|
| 4894 |
+
{
|
| 4895 |
+
"epoch": 0.8531038721573448,
|
| 4896 |
+
"grad_norm": 1.9653775691986084,
|
| 4897 |
+
"learning_rate": 5.882725650298787e-06,
|
| 4898 |
+
"loss": 1.1204,
|
| 4899 |
+
"step": 694
|
| 4900 |
+
},
|
| 4901 |
+
{
|
| 4902 |
+
"epoch": 0.8543331284572834,
|
| 4903 |
+
"grad_norm": 3.2125988006591797,
|
| 4904 |
+
"learning_rate": 5.7862165024123175e-06,
|
| 4905 |
+
"loss": 1.3958,
|
| 4906 |
+
"step": 695
|
| 4907 |
+
},
|
| 4908 |
+
{
|
| 4909 |
+
"epoch": 0.8555623847572219,
|
| 4910 |
+
"grad_norm": 3.528750419616699,
|
| 4911 |
+
"learning_rate": 5.690456916827691e-06,
|
| 4912 |
+
"loss": 0.9974,
|
| 4913 |
+
"step": 696
|
| 4914 |
+
},
|
| 4915 |
+
{
|
| 4916 |
+
"epoch": 0.8567916410571604,
|
| 4917 |
+
"grad_norm": 3.617098808288574,
|
| 4918 |
+
"learning_rate": 5.5954485169702306e-06,
|
| 4919 |
+
"loss": 1.1263,
|
| 4920 |
+
"step": 697
|
| 4921 |
+
},
|
| 4922 |
+
{
|
| 4923 |
+
"epoch": 0.858020897357099,
|
| 4924 |
+
"grad_norm": 2.575523614883423,
|
| 4925 |
+
"learning_rate": 5.501192913530301e-06,
|
| 4926 |
+
"loss": 0.6504,
|
| 4927 |
+
"step": 698
|
| 4928 |
+
},
|
| 4929 |
+
{
|
| 4930 |
+
"epoch": 0.8592501536570375,
|
| 4931 |
+
"grad_norm": 3.0249316692352295,
|
| 4932 |
+
"learning_rate": 5.407691704435991e-06,
|
| 4933 |
+
"loss": 0.6346,
|
| 4934 |
+
"step": 699
|
| 4935 |
+
},
|
| 4936 |
+
{
|
| 4937 |
+
"epoch": 0.860479409956976,
|
| 4938 |
+
"grad_norm": 2.3461408615112305,
|
| 4939 |
+
"learning_rate": 5.314946474826066e-06,
|
| 4940 |
+
"loss": 0.4619,
|
| 4941 |
+
"step": 700
|
| 4942 |
+
},
|
| 4943 |
+
{
|
| 4944 |
+
"epoch": 0.8617086662569146,
|
| 4945 |
+
"grad_norm": 0.2000708281993866,
|
| 4946 |
+
"learning_rate": 5.222958797023036e-06,
|
| 4947 |
+
"loss": 0.9165,
|
| 4948 |
+
"step": 701
|
| 4949 |
+
},
|
| 4950 |
+
{
|
| 4951 |
+
"epoch": 0.8629379225568531,
|
| 4952 |
+
"grad_norm": 0.23434923589229584,
|
| 4953 |
+
"learning_rate": 5.13173023050656e-06,
|
| 4954 |
+
"loss": 1.1903,
|
| 4955 |
+
"step": 702
|
| 4956 |
+
},
|
| 4957 |
+
{
|
| 4958 |
+
"epoch": 0.8641671788567916,
|
| 4959 |
+
"grad_norm": 0.25243085622787476,
|
| 4960 |
+
"learning_rate": 5.041262321886958e-06,
|
| 4961 |
+
"loss": 1.2185,
|
| 4962 |
+
"step": 703
|
| 4963 |
+
},
|
| 4964 |
+
{
|
| 4965 |
+
"epoch": 0.8653964351567301,
|
| 4966 |
+
"grad_norm": 0.25832071900367737,
|
| 4967 |
+
"learning_rate": 4.951556604879048e-06,
|
| 4968 |
+
"loss": 1.2881,
|
| 4969 |
+
"step": 704
|
| 4970 |
+
},
|
| 4971 |
+
{
|
| 4972 |
+
"epoch": 0.8666256914566687,
|
| 4973 |
+
"grad_norm": 0.25066548585891724,
|
| 4974 |
+
"learning_rate": 4.862614600276061e-06,
|
| 4975 |
+
"loss": 1.1272,
|
| 4976 |
+
"step": 705
|
| 4977 |
+
},
|
| 4978 |
+
{
|
| 4979 |
+
"epoch": 0.8678549477566072,
|
| 4980 |
+
"grad_norm": 0.27538540959358215,
|
| 4981 |
+
"learning_rate": 4.774437815923938e-06,
|
| 4982 |
+
"loss": 1.003,
|
| 4983 |
+
"step": 706
|
| 4984 |
+
},
|
| 4985 |
+
{
|
| 4986 |
+
"epoch": 0.8690842040565457,
|
| 4987 |
+
"grad_norm": 0.27678731083869934,
|
| 4988 |
+
"learning_rate": 4.687027746695727e-06,
|
| 4989 |
+
"loss": 1.0579,
|
| 4990 |
+
"step": 707
|
| 4991 |
+
},
|
| 4992 |
+
{
|
| 4993 |
+
"epoch": 0.8703134603564844,
|
| 4994 |
+
"grad_norm": 0.3305651843547821,
|
| 4995 |
+
"learning_rate": 4.600385874466256e-06,
|
| 4996 |
+
"loss": 1.0734,
|
| 4997 |
+
"step": 708
|
| 4998 |
+
},
|
| 4999 |
+
{
|
| 5000 |
+
"epoch": 0.8715427166564229,
|
| 5001 |
+
"grad_norm": 0.3679453134536743,
|
| 5002 |
+
"learning_rate": 4.514513668087011e-06,
|
| 5003 |
+
"loss": 1.1202,
|
| 5004 |
+
"step": 709
|
| 5005 |
+
},
|
| 5006 |
+
{
|
| 5007 |
+
"epoch": 0.8727719729563614,
|
| 5008 |
+
"grad_norm": 0.32898518443107605,
|
| 5009 |
+
"learning_rate": 4.429412583361209e-06,
|
| 5010 |
+
"loss": 1.1176,
|
| 5011 |
+
"step": 710
|
| 5012 |
+
},
|
| 5013 |
+
{
|
| 5014 |
+
"epoch": 0.8740012292563,
|
| 5015 |
+
"grad_norm": 0.37726449966430664,
|
| 5016 |
+
"learning_rate": 4.34508406301915e-06,
|
| 5017 |
+
"loss": 1.1331,
|
| 5018 |
+
"step": 711
|
| 5019 |
+
},
|
| 5020 |
+
{
|
| 5021 |
+
"epoch": 0.8752304855562385,
|
| 5022 |
+
"grad_norm": 0.44210806488990784,
|
| 5023 |
+
"learning_rate": 4.261529536693737e-06,
|
| 5024 |
+
"loss": 0.9606,
|
| 5025 |
+
"step": 712
|
| 5026 |
+
},
|
| 5027 |
+
{
|
| 5028 |
+
"epoch": 0.876459741856177,
|
| 5029 |
+
"grad_norm": 0.5424479246139526,
|
| 5030 |
+
"learning_rate": 4.178750420896255e-06,
|
| 5031 |
+
"loss": 0.9004,
|
| 5032 |
+
"step": 713
|
| 5033 |
+
},
|
| 5034 |
+
{
|
| 5035 |
+
"epoch": 0.8776889981561156,
|
| 5036 |
+
"grad_norm": 0.7757157683372498,
|
| 5037 |
+
"learning_rate": 4.0967481189923384e-06,
|
| 5038 |
+
"loss": 0.8007,
|
| 5039 |
+
"step": 714
|
| 5040 |
+
},
|
| 5041 |
+
{
|
| 5042 |
+
"epoch": 0.8789182544560541,
|
| 5043 |
+
"grad_norm": 0.6526496410369873,
|
| 5044 |
+
"learning_rate": 4.015524021178196e-06,
|
| 5045 |
+
"loss": 0.7106,
|
| 5046 |
+
"step": 715
|
| 5047 |
+
},
|
| 5048 |
+
{
|
| 5049 |
+
"epoch": 0.8801475107559926,
|
| 5050 |
+
"grad_norm": 0.6553987264633179,
|
| 5051 |
+
"learning_rate": 3.935079504457034e-06,
|
| 5052 |
+
"loss": 0.668,
|
| 5053 |
+
"step": 716
|
| 5054 |
+
},
|
| 5055 |
+
{
|
| 5056 |
+
"epoch": 0.8813767670559312,
|
| 5057 |
+
"grad_norm": 0.795111358165741,
|
| 5058 |
+
"learning_rate": 3.8554159326157304e-06,
|
| 5059 |
+
"loss": 1.0273,
|
| 5060 |
+
"step": 717
|
| 5061 |
+
},
|
| 5062 |
+
{
|
| 5063 |
+
"epoch": 0.8826060233558697,
|
| 5064 |
+
"grad_norm": 1.3710993528366089,
|
| 5065 |
+
"learning_rate": 3.7765346562016744e-06,
|
| 5066 |
+
"loss": 1.0677,
|
| 5067 |
+
"step": 718
|
| 5068 |
+
},
|
| 5069 |
+
{
|
| 5070 |
+
"epoch": 0.8838352796558082,
|
| 5071 |
+
"grad_norm": 2.689974308013916,
|
| 5072 |
+
"learning_rate": 3.6984370124999058e-06,
|
| 5073 |
+
"loss": 1.3152,
|
| 5074 |
+
"step": 719
|
| 5075 |
+
},
|
| 5076 |
+
{
|
| 5077 |
+
"epoch": 0.8850645359557467,
|
| 5078 |
+
"grad_norm": 3.0526108741760254,
|
| 5079 |
+
"learning_rate": 3.621124325510422e-06,
|
| 5080 |
+
"loss": 0.7108,
|
| 5081 |
+
"step": 720
|
| 5082 |
+
},
|
| 5083 |
+
{
|
| 5084 |
+
"epoch": 0.8862937922556853,
|
| 5085 |
+
"grad_norm": 2.8882553577423096,
|
| 5086 |
+
"learning_rate": 3.5445979059257505e-06,
|
| 5087 |
+
"loss": 0.7952,
|
| 5088 |
+
"step": 721
|
| 5089 |
+
},
|
| 5090 |
+
{
|
| 5091 |
+
"epoch": 0.8875230485556238,
|
| 5092 |
+
"grad_norm": 2.545741319656372,
|
| 5093 |
+
"learning_rate": 3.4688590511087304e-06,
|
| 5094 |
+
"loss": 0.5186,
|
| 5095 |
+
"step": 722
|
| 5096 |
+
},
|
| 5097 |
+
{
|
| 5098 |
+
"epoch": 0.8887523048555623,
|
| 5099 |
+
"grad_norm": 3.898621082305908,
|
| 5100 |
+
"learning_rate": 3.3939090450704925e-06,
|
| 5101 |
+
"loss": 0.921,
|
| 5102 |
+
"step": 723
|
| 5103 |
+
},
|
| 5104 |
+
{
|
| 5105 |
+
"epoch": 0.8899815611555009,
|
| 5106 |
+
"grad_norm": 2.621055841445923,
|
| 5107 |
+
"learning_rate": 3.3197491584487093e-06,
|
| 5108 |
+
"loss": 0.7413,
|
| 5109 |
+
"step": 724
|
| 5110 |
+
},
|
| 5111 |
+
{
|
| 5112 |
+
"epoch": 0.8912108174554395,
|
| 5113 |
+
"grad_norm": 2.9799768924713135,
|
| 5114 |
+
"learning_rate": 3.246380648486058e-06,
|
| 5115 |
+
"loss": 0.5268,
|
| 5116 |
+
"step": 725
|
| 5117 |
+
},
|
| 5118 |
+
{
|
| 5119 |
+
"epoch": 0.892440073755378,
|
| 5120 |
+
"grad_norm": 0.2122289389371872,
|
| 5121 |
+
"learning_rate": 3.1738047590088803e-06,
|
| 5122 |
+
"loss": 1.0725,
|
| 5123 |
+
"step": 726
|
| 5124 |
+
},
|
| 5125 |
+
{
|
| 5126 |
+
"epoch": 0.8936693300553166,
|
| 5127 |
+
"grad_norm": 0.2591450810432434,
|
| 5128 |
+
"learning_rate": 3.10202272040615e-06,
|
| 5129 |
+
"loss": 1.4448,
|
| 5130 |
+
"step": 727
|
| 5131 |
+
},
|
| 5132 |
+
{
|
| 5133 |
+
"epoch": 0.8948985863552551,
|
| 5134 |
+
"grad_norm": 0.22906170785427094,
|
| 5135 |
+
"learning_rate": 3.0310357496085405e-06,
|
| 5136 |
+
"loss": 1.0041,
|
| 5137 |
+
"step": 728
|
| 5138 |
+
},
|
| 5139 |
+
{
|
| 5140 |
+
"epoch": 0.8961278426551936,
|
| 5141 |
+
"grad_norm": 0.2589828073978424,
|
| 5142 |
+
"learning_rate": 2.9608450500678565e-06,
|
| 5143 |
+
"loss": 1.2252,
|
| 5144 |
+
"step": 729
|
| 5145 |
+
},
|
| 5146 |
+
{
|
| 5147 |
+
"epoch": 0.8973570989551322,
|
| 5148 |
+
"grad_norm": 0.27138885855674744,
|
| 5149 |
+
"learning_rate": 2.8914518117366006e-06,
|
| 5150 |
+
"loss": 1.2409,
|
| 5151 |
+
"step": 730
|
| 5152 |
+
},
|
| 5153 |
+
{
|
| 5154 |
+
"epoch": 0.8985863552550707,
|
| 5155 |
+
"grad_norm": 0.2877761423587799,
|
| 5156 |
+
"learning_rate": 2.8228572110478133e-06,
|
| 5157 |
+
"loss": 1.1005,
|
| 5158 |
+
"step": 731
|
| 5159 |
+
},
|
| 5160 |
+
{
|
| 5161 |
+
"epoch": 0.8998156115550092,
|
| 5162 |
+
"grad_norm": 0.3168583810329437,
|
| 5163 |
+
"learning_rate": 2.755062410895104e-06,
|
| 5164 |
+
"loss": 0.9683,
|
| 5165 |
+
"step": 732
|
| 5166 |
+
},
|
| 5167 |
+
{
|
| 5168 |
+
"epoch": 0.9010448678549478,
|
| 5169 |
+
"grad_norm": 0.3234322667121887,
|
| 5170 |
+
"learning_rate": 2.6880685606129664e-06,
|
| 5171 |
+
"loss": 0.979,
|
| 5172 |
+
"step": 733
|
| 5173 |
+
},
|
| 5174 |
+
{
|
| 5175 |
+
"epoch": 0.9022741241548863,
|
| 5176 |
+
"grad_norm": 0.33790323138237,
|
| 5177 |
+
"learning_rate": 2.62187679595729e-06,
|
| 5178 |
+
"loss": 1.1439,
|
| 5179 |
+
"step": 734
|
| 5180 |
+
},
|
| 5181 |
+
{
|
| 5182 |
+
"epoch": 0.9035033804548248,
|
| 5183 |
+
"grad_norm": 0.34143367409706116,
|
| 5184 |
+
"learning_rate": 2.55648823908608e-06,
|
| 5185 |
+
"loss": 1.1412,
|
| 5186 |
+
"step": 735
|
| 5187 |
+
},
|
| 5188 |
+
{
|
| 5189 |
+
"epoch": 0.9047326367547633,
|
| 5190 |
+
"grad_norm": 0.41084036231040955,
|
| 5191 |
+
"learning_rate": 2.4919039985404626e-06,
|
| 5192 |
+
"loss": 1.1742,
|
| 5193 |
+
"step": 736
|
| 5194 |
+
},
|
| 5195 |
+
{
|
| 5196 |
+
"epoch": 0.9059618930547019,
|
| 5197 |
+
"grad_norm": 0.5593938231468201,
|
| 5198 |
+
"learning_rate": 2.428125169225881e-06,
|
| 5199 |
+
"loss": 1.0794,
|
| 5200 |
+
"step": 737
|
| 5201 |
+
},
|
| 5202 |
+
{
|
| 5203 |
+
"epoch": 0.9071911493546404,
|
| 5204 |
+
"grad_norm": 0.7228390574455261,
|
| 5205 |
+
"learning_rate": 2.36515283239353e-06,
|
| 5206 |
+
"loss": 0.7684,
|
| 5207 |
+
"step": 738
|
| 5208 |
+
},
|
| 5209 |
+
{
|
| 5210 |
+
"epoch": 0.9084204056545789,
|
| 5211 |
+
"grad_norm": 0.6799226999282837,
|
| 5212 |
+
"learning_rate": 2.3029880556220074e-06,
|
| 5213 |
+
"loss": 1.0537,
|
| 5214 |
+
"step": 739
|
| 5215 |
+
},
|
| 5216 |
+
{
|
| 5217 |
+
"epoch": 0.9096496619545175,
|
| 5218 |
+
"grad_norm": 0.6442016959190369,
|
| 5219 |
+
"learning_rate": 2.241631892799262e-06,
|
| 5220 |
+
"loss": 0.498,
|
| 5221 |
+
"step": 740
|
| 5222 |
+
},
|
| 5223 |
+
{
|
| 5224 |
+
"epoch": 0.9108789182544561,
|
| 5225 |
+
"grad_norm": 0.8138104677200317,
|
| 5226 |
+
"learning_rate": 2.181085384104703e-06,
|
| 5227 |
+
"loss": 0.9415,
|
| 5228 |
+
"step": 741
|
| 5229 |
+
},
|
| 5230 |
+
{
|
| 5231 |
+
"epoch": 0.9121081745543946,
|
| 5232 |
+
"grad_norm": 0.864895224571228,
|
| 5233 |
+
"learning_rate": 2.121349555991525e-06,
|
| 5234 |
+
"loss": 1.0131,
|
| 5235 |
+
"step": 742
|
| 5236 |
+
},
|
| 5237 |
+
{
|
| 5238 |
+
"epoch": 0.9133374308543332,
|
| 5239 |
+
"grad_norm": 1.3241431713104248,
|
| 5240 |
+
"learning_rate": 2.0624254211693894e-06,
|
| 5241 |
+
"loss": 1.0531,
|
| 5242 |
+
"step": 743
|
| 5243 |
+
},
|
| 5244 |
+
{
|
| 5245 |
+
"epoch": 0.9145666871542717,
|
| 5246 |
+
"grad_norm": 2.810883045196533,
|
| 5247 |
+
"learning_rate": 2.004313978587186e-06,
|
| 5248 |
+
"loss": 0.8746,
|
| 5249 |
+
"step": 744
|
| 5250 |
+
},
|
| 5251 |
+
{
|
| 5252 |
+
"epoch": 0.9157959434542102,
|
| 5253 |
+
"grad_norm": 3.3806657791137695,
|
| 5254 |
+
"learning_rate": 1.9470162134161143e-06,
|
| 5255 |
+
"loss": 0.804,
|
| 5256 |
+
"step": 745
|
| 5257 |
+
},
|
| 5258 |
+
{
|
| 5259 |
+
"epoch": 0.9170251997541488,
|
| 5260 |
+
"grad_norm": 3.9051411151885986,
|
| 5261 |
+
"learning_rate": 1.8905330970330259e-06,
|
| 5262 |
+
"loss": 1.1265,
|
| 5263 |
+
"step": 746
|
| 5264 |
+
},
|
| 5265 |
+
{
|
| 5266 |
+
"epoch": 0.9182544560540873,
|
| 5267 |
+
"grad_norm": 2.50091290473938,
|
| 5268 |
+
"learning_rate": 1.83486558700387e-06,
|
| 5269 |
+
"loss": 0.8042,
|
| 5270 |
+
"step": 747
|
| 5271 |
+
},
|
| 5272 |
+
{
|
| 5273 |
+
"epoch": 0.9194837123540258,
|
| 5274 |
+
"grad_norm": 2.444403648376465,
|
| 5275 |
+
"learning_rate": 1.78001462706755e-06,
|
| 5276 |
+
"loss": 0.7185,
|
| 5277 |
+
"step": 748
|
| 5278 |
+
},
|
| 5279 |
+
{
|
| 5280 |
+
"epoch": 0.9207129686539643,
|
| 5281 |
+
"grad_norm": 2.565976619720459,
|
| 5282 |
+
"learning_rate": 1.7259811471198706e-06,
|
| 5283 |
+
"loss": 0.5854,
|
| 5284 |
+
"step": 749
|
| 5285 |
+
},
|
| 5286 |
+
{
|
| 5287 |
+
"epoch": 0.9219422249539029,
|
| 5288 |
+
"grad_norm": 3.5841028690338135,
|
| 5289 |
+
"learning_rate": 1.6727660631977893e-06,
|
| 5290 |
+
"loss": 1.1298,
|
| 5291 |
+
"step": 750
|
| 5292 |
+
},
|
| 5293 |
+
{
|
| 5294 |
+
"epoch": 0.9231714812538414,
|
| 5295 |
+
"grad_norm": 0.18373464047908783,
|
| 5296 |
+
"learning_rate": 1.620370277463884e-06,
|
| 5297 |
+
"loss": 0.798,
|
| 5298 |
+
"step": 751
|
| 5299 |
+
},
|
| 5300 |
+
{
|
| 5301 |
+
"epoch": 0.9244007375537799,
|
| 5302 |
+
"grad_norm": 0.23574425280094147,
|
| 5303 |
+
"learning_rate": 1.5687946781910378e-06,
|
| 5304 |
+
"loss": 1.1075,
|
| 5305 |
+
"step": 752
|
| 5306 |
+
},
|
| 5307 |
+
{
|
| 5308 |
+
"epoch": 0.9256299938537185,
|
| 5309 |
+
"grad_norm": 0.25111323595046997,
|
| 5310 |
+
"learning_rate": 1.5180401397474343e-06,
|
| 5311 |
+
"loss": 1.2677,
|
| 5312 |
+
"step": 753
|
| 5313 |
+
},
|
| 5314 |
+
{
|
| 5315 |
+
"epoch": 0.926859250153657,
|
| 5316 |
+
"grad_norm": 0.23998814821243286,
|
| 5317 |
+
"learning_rate": 1.4681075225816854e-06,
|
| 5318 |
+
"loss": 1.0985,
|
| 5319 |
+
"step": 754
|
| 5320 |
+
},
|
| 5321 |
+
{
|
| 5322 |
+
"epoch": 0.9280885064535955,
|
| 5323 |
+
"grad_norm": 0.2664302587509155,
|
| 5324 |
+
"learning_rate": 1.4189976732082666e-06,
|
| 5325 |
+
"loss": 1.0119,
|
| 5326 |
+
"step": 755
|
| 5327 |
+
},
|
| 5328 |
+
{
|
| 5329 |
+
"epoch": 0.9293177627535341,
|
| 5330 |
+
"grad_norm": 0.28511667251586914,
|
| 5331 |
+
"learning_rate": 1.3707114241931328e-06,
|
| 5332 |
+
"loss": 1.2079,
|
| 5333 |
+
"step": 756
|
| 5334 |
+
},
|
| 5335 |
+
{
|
| 5336 |
+
"epoch": 0.9305470190534727,
|
| 5337 |
+
"grad_norm": 0.278188556432724,
|
| 5338 |
+
"learning_rate": 1.3232495941396639e-06,
|
| 5339 |
+
"loss": 1.1176,
|
| 5340 |
+
"step": 757
|
| 5341 |
+
},
|
| 5342 |
+
{
|
| 5343 |
+
"epoch": 0.9317762753534112,
|
| 5344 |
+
"grad_norm": 0.3081459105014801,
|
| 5345 |
+
"learning_rate": 1.2766129876747413e-06,
|
| 5346 |
+
"loss": 0.9467,
|
| 5347 |
+
"step": 758
|
| 5348 |
+
},
|
| 5349 |
+
{
|
| 5350 |
+
"epoch": 0.9330055316533498,
|
| 5351 |
+
"grad_norm": 0.34005647897720337,
|
| 5352 |
+
"learning_rate": 1.2308023954351043e-06,
|
| 5353 |
+
"loss": 1.0071,
|
| 5354 |
+
"step": 759
|
| 5355 |
+
},
|
| 5356 |
+
{
|
| 5357 |
+
"epoch": 0.9342347879532883,
|
| 5358 |
+
"grad_norm": 0.3434988856315613,
|
| 5359 |
+
"learning_rate": 1.1858185940539779e-06,
|
| 5360 |
+
"loss": 1.1407,
|
| 5361 |
+
"step": 760
|
| 5362 |
+
},
|
| 5363 |
+
{
|
| 5364 |
+
"epoch": 0.9354640442532268,
|
| 5365 |
+
"grad_norm": 0.38441202044487,
|
| 5366 |
+
"learning_rate": 1.1416623461478704e-06,
|
| 5367 |
+
"loss": 1.316,
|
| 5368 |
+
"step": 761
|
| 5369 |
+
},
|
| 5370 |
+
{
|
| 5371 |
+
"epoch": 0.9366933005531654,
|
| 5372 |
+
"grad_norm": 0.42989110946655273,
|
| 5373 |
+
"learning_rate": 1.0983344003036912e-06,
|
| 5374 |
+
"loss": 1.2446,
|
| 5375 |
+
"step": 762
|
| 5376 |
+
},
|
| 5377 |
+
{
|
| 5378 |
+
"epoch": 0.9379225568531039,
|
| 5379 |
+
"grad_norm": 0.4977233111858368,
|
| 5380 |
+
"learning_rate": 1.055835491066004e-06,
|
| 5381 |
+
"loss": 0.7431,
|
| 5382 |
+
"step": 763
|
| 5383 |
+
},
|
| 5384 |
+
{
|
| 5385 |
+
"epoch": 0.9391518131530424,
|
| 5386 |
+
"grad_norm": 0.6383669376373291,
|
| 5387 |
+
"learning_rate": 1.014166338924627e-06,
|
| 5388 |
+
"loss": 0.5211,
|
| 5389 |
+
"step": 764
|
| 5390 |
+
},
|
| 5391 |
+
{
|
| 5392 |
+
"epoch": 0.9403810694529809,
|
| 5393 |
+
"grad_norm": 0.7088152766227722,
|
| 5394 |
+
"learning_rate": 9.733276503023692e-07,
|
| 5395 |
+
"loss": 0.537,
|
| 5396 |
+
"step": 765
|
| 5397 |
+
},
|
| 5398 |
+
{
|
| 5399 |
+
"epoch": 0.9416103257529195,
|
| 5400 |
+
"grad_norm": 0.6624244451522827,
|
| 5401 |
+
"learning_rate": 9.33320117543085e-07,
|
| 5402 |
+
"loss": 0.8375,
|
| 5403 |
+
"step": 766
|
| 5404 |
+
},
|
| 5405 |
+
{
|
| 5406 |
+
"epoch": 0.942839582052858,
|
| 5407 |
+
"grad_norm": 0.7807971239089966,
|
| 5408 |
+
"learning_rate": 8.941444188999393e-07,
|
| 5409 |
+
"loss": 0.7689,
|
| 5410 |
+
"step": 767
|
| 5411 |
+
},
|
| 5412 |
+
{
|
| 5413 |
+
"epoch": 0.9440688383527965,
|
| 5414 |
+
"grad_norm": 1.2817997932434082,
|
| 5415 |
+
"learning_rate": 8.558012185238939e-07,
|
| 5416 |
+
"loss": 1.1247,
|
| 5417 |
+
"step": 768
|
| 5418 |
+
},
|
| 5419 |
+
{
|
| 5420 |
+
"epoch": 0.9452980946527351,
|
| 5421 |
+
"grad_norm": 3.324708938598633,
|
| 5422 |
+
"learning_rate": 8.182911664524562e-07,
|
| 5423 |
+
"loss": 1.0981,
|
| 5424 |
+
"step": 769
|
| 5425 |
+
},
|
| 5426 |
+
{
|
| 5427 |
+
"epoch": 0.9465273509526736,
|
| 5428 |
+
"grad_norm": 3.574130058288574,
|
| 5429 |
+
"learning_rate": 7.816148985986483e-07,
|
| 5430 |
+
"loss": 0.6504,
|
| 5431 |
+
"step": 770
|
| 5432 |
+
},
|
| 5433 |
+
{
|
| 5434 |
+
"epoch": 0.9477566072526121,
|
| 5435 |
+
"grad_norm": 3.618056297302246,
|
| 5436 |
+
"learning_rate": 7.457730367402549e-07,
|
| 5437 |
+
"loss": 0.6916,
|
| 5438 |
+
"step": 771
|
| 5439 |
+
},
|
| 5440 |
+
{
|
| 5441 |
+
"epoch": 0.9489858635525507,
|
| 5442 |
+
"grad_norm": 2.9001710414886475,
|
| 5443 |
+
"learning_rate": 7.107661885092321e-07,
|
| 5444 |
+
"loss": 0.8261,
|
| 5445 |
+
"step": 772
|
| 5446 |
+
},
|
| 5447 |
+
{
|
| 5448 |
+
"epoch": 0.9502151198524893,
|
| 5449 |
+
"grad_norm": 2.4319024085998535,
|
| 5450 |
+
"learning_rate": 6.765949473814648e-07,
|
| 5451 |
+
"loss": 0.9126,
|
| 5452 |
+
"step": 773
|
| 5453 |
+
},
|
| 5454 |
+
{
|
| 5455 |
+
"epoch": 0.9514443761524278,
|
| 5456 |
+
"grad_norm": 3.393937110900879,
|
| 5457 |
+
"learning_rate": 6.432598926666589e-07,
|
| 5458 |
+
"loss": 0.8197,
|
| 5459 |
+
"step": 774
|
| 5460 |
+
},
|
| 5461 |
+
{
|
| 5462 |
+
"epoch": 0.9526736324523664,
|
| 5463 |
+
"grad_norm": 3.406257390975952,
|
| 5464 |
+
"learning_rate": 6.107615894985375e-07,
|
| 5465 |
+
"loss": 0.9015,
|
| 5466 |
+
"step": 775
|
| 5467 |
+
},
|
| 5468 |
+
{
|
| 5469 |
+
"epoch": 0.9539028887523049,
|
| 5470 |
+
"grad_norm": 0.21785248816013336,
|
| 5471 |
+
"learning_rate": 5.791005888252765e-07,
|
| 5472 |
+
"loss": 0.9666,
|
| 5473 |
+
"step": 776
|
| 5474 |
+
},
|
| 5475 |
+
{
|
| 5476 |
+
"epoch": 0.9551321450522434,
|
| 5477 |
+
"grad_norm": 0.22744600474834442,
|
| 5478 |
+
"learning_rate": 5.482774274001401e-07,
|
| 5479 |
+
"loss": 1.1927,
|
| 5480 |
+
"step": 777
|
| 5481 |
+
},
|
| 5482 |
+
{
|
| 5483 |
+
"epoch": 0.956361401352182,
|
| 5484 |
+
"grad_norm": 0.26558175683021545,
|
| 5485 |
+
"learning_rate": 5.18292627772382e-07,
|
| 5486 |
+
"loss": 1.3098,
|
| 5487 |
+
"step": 778
|
| 5488 |
+
},
|
| 5489 |
+
{
|
| 5490 |
+
"epoch": 0.9575906576521205,
|
| 5491 |
+
"grad_norm": 0.25804319977760315,
|
| 5492 |
+
"learning_rate": 4.891466982783977e-07,
|
| 5493 |
+
"loss": 1.2578,
|
| 5494 |
+
"step": 779
|
| 5495 |
+
},
|
| 5496 |
+
{
|
| 5497 |
+
"epoch": 0.958819913952059,
|
| 5498 |
+
"grad_norm": 0.2447444498538971,
|
| 5499 |
+
"learning_rate": 4.60840133033108e-07,
|
| 5500 |
+
"loss": 0.9595,
|
| 5501 |
+
"step": 780
|
| 5502 |
+
},
|
| 5503 |
+
{
|
| 5504 |
+
"epoch": 0.9600491702519975,
|
| 5505 |
+
"grad_norm": 0.2831968069076538,
|
| 5506 |
+
"learning_rate": 4.3337341192157265e-07,
|
| 5507 |
+
"loss": 1.0676,
|
| 5508 |
+
"step": 781
|
| 5509 |
+
},
|
| 5510 |
+
{
|
| 5511 |
+
"epoch": 0.9612784265519361,
|
| 5512 |
+
"grad_norm": 0.2845923602581024,
|
| 5513 |
+
"learning_rate": 4.067470005908625e-07,
|
| 5514 |
+
"loss": 0.9645,
|
| 5515 |
+
"step": 782
|
| 5516 |
+
},
|
| 5517 |
+
{
|
| 5518 |
+
"epoch": 0.9625076828518746,
|
| 5519 |
+
"grad_norm": 0.3282490372657776,
|
| 5520 |
+
"learning_rate": 3.809613504421661e-07,
|
| 5521 |
+
"loss": 0.8996,
|
| 5522 |
+
"step": 783
|
| 5523 |
+
},
|
| 5524 |
+
{
|
| 5525 |
+
"epoch": 0.9637369391518131,
|
| 5526 |
+
"grad_norm": 0.3805837631225586,
|
| 5527 |
+
"learning_rate": 3.5601689862311826e-07,
|
| 5528 |
+
"loss": 0.9756,
|
| 5529 |
+
"step": 784
|
| 5530 |
+
},
|
| 5531 |
+
{
|
| 5532 |
+
"epoch": 0.9649661954517517,
|
| 5533 |
+
"grad_norm": 0.3583146631717682,
|
| 5534 |
+
"learning_rate": 3.3191406802041693e-07,
|
| 5535 |
+
"loss": 1.0574,
|
| 5536 |
+
"step": 785
|
| 5537 |
+
},
|
| 5538 |
+
{
|
| 5539 |
+
"epoch": 0.9661954517516902,
|
| 5540 |
+
"grad_norm": 0.4697728753089905,
|
| 5541 |
+
"learning_rate": 3.0865326725263435e-07,
|
| 5542 |
+
"loss": 1.4301,
|
| 5543 |
+
"step": 786
|
| 5544 |
+
},
|
| 5545 |
+
{
|
| 5546 |
+
"epoch": 0.9674247080516287,
|
| 5547 |
+
"grad_norm": 0.6786931157112122,
|
| 5548 |
+
"learning_rate": 2.8623489066329503e-07,
|
| 5549 |
+
"loss": 1.0623,
|
| 5550 |
+
"step": 787
|
| 5551 |
+
},
|
| 5552 |
+
{
|
| 5553 |
+
"epoch": 0.9686539643515673,
|
| 5554 |
+
"grad_norm": 0.6815045475959778,
|
| 5555 |
+
"learning_rate": 2.646593183142088e-07,
|
| 5556 |
+
"loss": 0.7782,
|
| 5557 |
+
"step": 788
|
| 5558 |
+
},
|
| 5559 |
+
{
|
| 5560 |
+
"epoch": 0.9698832206515058,
|
| 5561 |
+
"grad_norm": 0.7787659764289856,
|
| 5562 |
+
"learning_rate": 2.4392691597898143e-07,
|
| 5563 |
+
"loss": 0.8097,
|
| 5564 |
+
"step": 789
|
| 5565 |
+
},
|
| 5566 |
+
{
|
| 5567 |
+
"epoch": 0.9711124769514444,
|
| 5568 |
+
"grad_norm": 0.7595354318618774,
|
| 5569 |
+
"learning_rate": 2.2403803513686428e-07,
|
| 5570 |
+
"loss": 0.9502,
|
| 5571 |
+
"step": 790
|
| 5572 |
+
},
|
| 5573 |
+
{
|
| 5574 |
+
"epoch": 0.972341733251383,
|
| 5575 |
+
"grad_norm": 0.8005998730659485,
|
| 5576 |
+
"learning_rate": 2.0499301296676432e-07,
|
| 5577 |
+
"loss": 0.7545,
|
| 5578 |
+
"step": 791
|
| 5579 |
+
},
|
| 5580 |
+
{
|
| 5581 |
+
"epoch": 0.9735709895513215,
|
| 5582 |
+
"grad_norm": 1.0289498567581177,
|
| 5583 |
+
"learning_rate": 1.8679217234154334e-07,
|
| 5584 |
+
"loss": 1.2025,
|
| 5585 |
+
"step": 792
|
| 5586 |
+
},
|
| 5587 |
+
{
|
| 5588 |
+
"epoch": 0.97480024585126,
|
| 5589 |
+
"grad_norm": 1.7563585042953491,
|
| 5590 |
+
"learning_rate": 1.6943582182253336e-07,
|
| 5591 |
+
"loss": 1.1175,
|
| 5592 |
+
"step": 793
|
| 5593 |
+
},
|
| 5594 |
+
{
|
| 5595 |
+
"epoch": 0.9760295021511985,
|
| 5596 |
+
"grad_norm": 3.998744487762451,
|
| 5597 |
+
"learning_rate": 1.5292425565430757e-07,
|
| 5598 |
+
"loss": 1.2125,
|
| 5599 |
+
"step": 794
|
| 5600 |
+
},
|
| 5601 |
+
{
|
| 5602 |
+
"epoch": 0.9772587584511371,
|
| 5603 |
+
"grad_norm": 2.2866718769073486,
|
| 5604 |
+
"learning_rate": 1.372577537597064e-07,
|
| 5605 |
+
"loss": 0.3866,
|
| 5606 |
+
"step": 795
|
| 5607 |
+
},
|
| 5608 |
+
{
|
| 5609 |
+
"epoch": 0.9784880147510756,
|
| 5610 |
+
"grad_norm": 3.594325542449951,
|
| 5611 |
+
"learning_rate": 1.224365817350692e-07,
|
| 5612 |
+
"loss": 0.8022,
|
| 5613 |
+
"step": 796
|
| 5614 |
+
},
|
| 5615 |
+
{
|
| 5616 |
+
"epoch": 0.9797172710510141,
|
| 5617 |
+
"grad_norm": 2.57466197013855,
|
| 5618 |
+
"learning_rate": 1.0846099084574346e-07,
|
| 5619 |
+
"loss": 0.8229,
|
| 5620 |
+
"step": 797
|
| 5621 |
+
},
|
| 5622 |
+
{
|
| 5623 |
+
"epoch": 0.9809465273509527,
|
| 5624 |
+
"grad_norm": 2.2920215129852295,
|
| 5625 |
+
"learning_rate": 9.533121802183797e-08,
|
| 5626 |
+
"loss": 0.509,
|
| 5627 |
+
"step": 798
|
| 5628 |
+
},
|
| 5629 |
+
{
|
| 5630 |
+
"epoch": 0.9821757836508912,
|
| 5631 |
+
"grad_norm": 2.825840711593628,
|
| 5632 |
+
"learning_rate": 8.304748585417078e-08,
|
| 5633 |
+
"loss": 0.9473,
|
| 5634 |
+
"step": 799
|
| 5635 |
+
},
|
| 5636 |
+
{
|
| 5637 |
+
"epoch": 0.9834050399508297,
|
| 5638 |
+
"grad_norm": 3.6677334308624268,
|
| 5639 |
+
"learning_rate": 7.161000259053308e-08,
|
| 5640 |
+
"loss": 0.7938,
|
| 5641 |
+
"step": 800
|
| 5642 |
+
},
|
| 5643 |
+
{
|
| 5644 |
+
"epoch": 0.9846342962507683,
|
| 5645 |
+
"grad_norm": 0.2495596706867218,
|
| 5646 |
+
"learning_rate": 6.10189621321422e-08,
|
| 5647 |
+
"loss": 1.2466,
|
| 5648 |
+
"step": 801
|
| 5649 |
+
},
|
| 5650 |
+
{
|
| 5651 |
+
"epoch": 0.9858635525507068,
|
| 5652 |
+
"grad_norm": 0.2586044371128082,
|
| 5653 |
+
"learning_rate": 5.127454403034415e-08,
|
| 5654 |
+
"loss": 1.3163,
|
| 5655 |
+
"step": 802
|
| 5656 |
+
},
|
| 5657 |
+
{
|
| 5658 |
+
"epoch": 0.9870928088506453,
|
| 5659 |
+
"grad_norm": 0.2908371388912201,
|
| 5660 |
+
"learning_rate": 4.2376913483599404e-08,
|
| 5661 |
+
"loss": 1.2477,
|
| 5662 |
+
"step": 803
|
| 5663 |
+
},
|
| 5664 |
+
{
|
| 5665 |
+
"epoch": 0.9883220651505839,
|
| 5666 |
+
"grad_norm": 0.3172759413719177,
|
| 5667 |
+
"learning_rate": 3.4326221334640695e-08,
|
| 5668 |
+
"loss": 1.1196,
|
| 5669 |
+
"step": 804
|
| 5670 |
+
},
|
| 5671 |
+
{
|
| 5672 |
+
"epoch": 0.9895513214505224,
|
| 5673 |
+
"grad_norm": 0.33819660544395447,
|
| 5674 |
+
"learning_rate": 2.712260406795286e-08,
|
| 5675 |
+
"loss": 1.0077,
|
| 5676 |
+
"step": 805
|
| 5677 |
+
},
|
| 5678 |
+
{
|
| 5679 |
+
"epoch": 0.990780577750461,
|
| 5680 |
+
"grad_norm": 0.353371262550354,
|
| 5681 |
+
"learning_rate": 2.076618380744133e-08,
|
| 5682 |
+
"loss": 1.0194,
|
| 5683 |
+
"step": 806
|
| 5684 |
+
},
|
| 5685 |
+
{
|
| 5686 |
+
"epoch": 0.9920098340503996,
|
| 5687 |
+
"grad_norm": 0.4687137007713318,
|
| 5688 |
+
"learning_rate": 1.525706831437268e-08,
|
| 5689 |
+
"loss": 1.199,
|
| 5690 |
+
"step": 807
|
| 5691 |
+
},
|
| 5692 |
+
{
|
| 5693 |
+
"epoch": 0.9932390903503381,
|
| 5694 |
+
"grad_norm": 0.6552625298500061,
|
| 5695 |
+
"learning_rate": 1.0595350985526109e-08,
|
| 5696 |
+
"loss": 0.5601,
|
| 5697 |
+
"step": 808
|
| 5698 |
+
},
|
| 5699 |
+
{
|
| 5700 |
+
"epoch": 0.9944683466502766,
|
| 5701 |
+
"grad_norm": 0.6662526726722717,
|
| 5702 |
+
"learning_rate": 6.781110851633576e-09,
|
| 5703 |
+
"loss": 0.8824,
|
| 5704 |
+
"step": 809
|
| 5705 |
+
},
|
| 5706 |
+
{
|
| 5707 |
+
"epoch": 0.9956976029502151,
|
| 5708 |
+
"grad_norm": 1.0151597261428833,
|
| 5709 |
+
"learning_rate": 3.814412576025328e-09,
|
| 5710 |
+
"loss": 1.1593,
|
| 5711 |
+
"step": 810
|
| 5712 |
+
},
|
| 5713 |
+
{
|
| 5714 |
+
"epoch": 0.9969268592501537,
|
| 5715 |
+
"grad_norm": 2.71447491645813,
|
| 5716 |
+
"learning_rate": 1.6953064535474295e-09,
|
| 5717 |
+
"loss": 0.7048,
|
| 5718 |
+
"step": 811
|
| 5719 |
+
},
|
| 5720 |
+
{
|
| 5721 |
+
"epoch": 0.9981561155500922,
|
| 5722 |
+
"grad_norm": 3.1842808723449707,
|
| 5723 |
+
"learning_rate": 4.238284096902412e-10,
|
| 5724 |
+
"loss": 0.8335,
|
| 5725 |
+
"step": 812
|
| 5726 |
+
},
|
| 5727 |
+
{
|
| 5728 |
+
"epoch": 0.9993853718500307,
|
| 5729 |
+
"grad_norm": 2.6523120403289795,
|
| 5730 |
+
"learning_rate": 0.0,
|
| 5731 |
+
"loss": 0.6896,
|
| 5732 |
+
"step": 813
|
| 5733 |
}
|
| 5734 |
],
|
| 5735 |
"logging_steps": 1,
|
|
|
|
| 5744 |
"should_evaluate": false,
|
| 5745 |
"should_log": false,
|
| 5746 |
"should_save": true,
|
| 5747 |
+
"should_training_stop": true
|
| 5748 |
},
|
| 5749 |
"attributes": {}
|
| 5750 |
}
|
| 5751 |
},
|
| 5752 |
+
"total_flos": 1.6756836906985062e+17,
|
| 5753 |
"train_batch_size": 1,
|
| 5754 |
"trial_name": null,
|
| 5755 |
"trial_params": null
|