Fanucci
commited on
Training in progress, step 1000, 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 671149168
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d14fa1c6cd0a8c1098c2de693eacd3e9fd9509d70fdc5893f7c7bc6fa28a5515
|
| 3 |
size 671149168
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1342555602
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d145b6c083f2e7222313302dd9978b6bc171ff03d6651b2280de2698f84eda21
|
| 3 |
size 1342555602
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5f5a61412c4341dc129410787ef47b348a882a43e4433910d7fa909787ab3112
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e56f8bb31e77104b8ecd639f65cf9bb955635271d71868e7ef99f855e32f8bbf
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 3.1351470947265625,
|
| 3 |
"best_model_checkpoint": "miner_id_24/checkpoint-600",
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 200,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -5647,6 +5647,1414 @@
|
|
| 5647 |
"eval_samples_per_second": 5.635,
|
| 5648 |
"eval_steps_per_second": 1.409,
|
| 5649 |
"step": 800
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5650 |
}
|
| 5651 |
],
|
| 5652 |
"logging_steps": 1,
|
|
@@ -5661,7 +7069,7 @@
|
|
| 5661 |
"early_stopping_threshold": 0.0
|
| 5662 |
},
|
| 5663 |
"attributes": {
|
| 5664 |
-
"early_stopping_patience_counter":
|
| 5665 |
}
|
| 5666 |
},
|
| 5667 |
"TrainerControl": {
|
|
@@ -5675,7 +7083,7 @@
|
|
| 5675 |
"attributes": {}
|
| 5676 |
}
|
| 5677 |
},
|
| 5678 |
-
"total_flos":
|
| 5679 |
"train_batch_size": 4,
|
| 5680 |
"trial_name": null,
|
| 5681 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 3.1351470947265625,
|
| 3 |
"best_model_checkpoint": "miner_id_24/checkpoint-600",
|
| 4 |
+
"epoch": 0.3199488081906895,
|
| 5 |
"eval_steps": 200,
|
| 6 |
+
"global_step": 1000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 5647 |
"eval_samples_per_second": 5.635,
|
| 5648 |
"eval_steps_per_second": 1.409,
|
| 5649 |
"step": 800
|
| 5650 |
+
},
|
| 5651 |
+
{
|
| 5652 |
+
"epoch": 0.2562789953607423,
|
| 5653 |
+
"grad_norm": 472340.75,
|
| 5654 |
+
"learning_rate": 0.0003403490006745532,
|
| 5655 |
+
"loss": 18.063,
|
| 5656 |
+
"step": 801
|
| 5657 |
+
},
|
| 5658 |
+
{
|
| 5659 |
+
"epoch": 0.256598944168933,
|
| 5660 |
+
"grad_norm": 32471.482421875,
|
| 5661 |
+
"learning_rate": 0.00034020654447263155,
|
| 5662 |
+
"loss": 26.347,
|
| 5663 |
+
"step": 802
|
| 5664 |
+
},
|
| 5665 |
+
{
|
| 5666 |
+
"epoch": 0.25691889297712367,
|
| 5667 |
+
"grad_norm": 151727.5,
|
| 5668 |
+
"learning_rate": 0.00034006395012645086,
|
| 5669 |
+
"loss": 24.9178,
|
| 5670 |
+
"step": 803
|
| 5671 |
+
},
|
| 5672 |
+
{
|
| 5673 |
+
"epoch": 0.25723884178531436,
|
| 5674 |
+
"grad_norm": 43865.3671875,
|
| 5675 |
+
"learning_rate": 0.00033992121778729826,
|
| 5676 |
+
"loss": 24.8413,
|
| 5677 |
+
"step": 804
|
| 5678 |
+
},
|
| 5679 |
+
{
|
| 5680 |
+
"epoch": 0.25755879059350506,
|
| 5681 |
+
"grad_norm": 47757.33984375,
|
| 5682 |
+
"learning_rate": 0.00033977834760660745,
|
| 5683 |
+
"loss": 25.444,
|
| 5684 |
+
"step": 805
|
| 5685 |
+
},
|
| 5686 |
+
{
|
| 5687 |
+
"epoch": 0.25787873940169576,
|
| 5688 |
+
"grad_norm": 14128.7724609375,
|
| 5689 |
+
"learning_rate": 0.00033963533973595825,
|
| 5690 |
+
"loss": 25.3313,
|
| 5691 |
+
"step": 806
|
| 5692 |
+
},
|
| 5693 |
+
{
|
| 5694 |
+
"epoch": 0.2581986882098864,
|
| 5695 |
+
"grad_norm": 1130996.625,
|
| 5696 |
+
"learning_rate": 0.00033949219432707667,
|
| 5697 |
+
"loss": 26.6114,
|
| 5698 |
+
"step": 807
|
| 5699 |
+
},
|
| 5700 |
+
{
|
| 5701 |
+
"epoch": 0.2585186370180771,
|
| 5702 |
+
"grad_norm": 4760.6435546875,
|
| 5703 |
+
"learning_rate": 0.0003393489115318347,
|
| 5704 |
+
"loss": 25.624,
|
| 5705 |
+
"step": 808
|
| 5706 |
+
},
|
| 5707 |
+
{
|
| 5708 |
+
"epoch": 0.2588385858262678,
|
| 5709 |
+
"grad_norm": 365082.65625,
|
| 5710 |
+
"learning_rate": 0.00033920549150225,
|
| 5711 |
+
"loss": 26.4853,
|
| 5712 |
+
"step": 809
|
| 5713 |
+
},
|
| 5714 |
+
{
|
| 5715 |
+
"epoch": 0.2591585346344585,
|
| 5716 |
+
"grad_norm": 972.4833374023438,
|
| 5717 |
+
"learning_rate": 0.0003390619343904857,
|
| 5718 |
+
"loss": 26.3925,
|
| 5719 |
+
"step": 810
|
| 5720 |
+
},
|
| 5721 |
+
{
|
| 5722 |
+
"epoch": 0.2594784834426492,
|
| 5723 |
+
"grad_norm": 56389.19140625,
|
| 5724 |
+
"learning_rate": 0.0003389182403488507,
|
| 5725 |
+
"loss": 25.6556,
|
| 5726 |
+
"step": 811
|
| 5727 |
+
},
|
| 5728 |
+
{
|
| 5729 |
+
"epoch": 0.2597984322508399,
|
| 5730 |
+
"grad_norm": 21225.556640625,
|
| 5731 |
+
"learning_rate": 0.0003387744095297988,
|
| 5732 |
+
"loss": 24.6976,
|
| 5733 |
+
"step": 812
|
| 5734 |
+
},
|
| 5735 |
+
{
|
| 5736 |
+
"epoch": 0.26011838105903057,
|
| 5737 |
+
"grad_norm": 1668.451416015625,
|
| 5738 |
+
"learning_rate": 0.00033863044208592925,
|
| 5739 |
+
"loss": 22.8956,
|
| 5740 |
+
"step": 813
|
| 5741 |
+
},
|
| 5742 |
+
{
|
| 5743 |
+
"epoch": 0.26043832986722126,
|
| 5744 |
+
"grad_norm": 19362.66015625,
|
| 5745 |
+
"learning_rate": 0.00033848633816998603,
|
| 5746 |
+
"loss": 24.3001,
|
| 5747 |
+
"step": 814
|
| 5748 |
+
},
|
| 5749 |
+
{
|
| 5750 |
+
"epoch": 0.26075827867541196,
|
| 5751 |
+
"grad_norm": 32938.8125,
|
| 5752 |
+
"learning_rate": 0.000338342097934858,
|
| 5753 |
+
"loss": 23.744,
|
| 5754 |
+
"step": 815
|
| 5755 |
+
},
|
| 5756 |
+
{
|
| 5757 |
+
"epoch": 0.2610782274836026,
|
| 5758 |
+
"grad_norm": 2615.0556640625,
|
| 5759 |
+
"learning_rate": 0.00033819772153357875,
|
| 5760 |
+
"loss": 23.1489,
|
| 5761 |
+
"step": 816
|
| 5762 |
+
},
|
| 5763 |
+
{
|
| 5764 |
+
"epoch": 0.2613981762917933,
|
| 5765 |
+
"grad_norm": 2801.425048828125,
|
| 5766 |
+
"learning_rate": 0.00033805320911932597,
|
| 5767 |
+
"loss": 21.2644,
|
| 5768 |
+
"step": 817
|
| 5769 |
+
},
|
| 5770 |
+
{
|
| 5771 |
+
"epoch": 0.261718125099984,
|
| 5772 |
+
"grad_norm": 26318.5703125,
|
| 5773 |
+
"learning_rate": 0.00033790856084542223,
|
| 5774 |
+
"loss": 20.3282,
|
| 5775 |
+
"step": 818
|
| 5776 |
+
},
|
| 5777 |
+
{
|
| 5778 |
+
"epoch": 0.2620380739081747,
|
| 5779 |
+
"grad_norm": 16279.3955078125,
|
| 5780 |
+
"learning_rate": 0.0003377637768653337,
|
| 5781 |
+
"loss": 17.2759,
|
| 5782 |
+
"step": 819
|
| 5783 |
+
},
|
| 5784 |
+
{
|
| 5785 |
+
"epoch": 0.2623580227163654,
|
| 5786 |
+
"grad_norm": 13162.205078125,
|
| 5787 |
+
"learning_rate": 0.0003376188573326708,
|
| 5788 |
+
"loss": 16.7004,
|
| 5789 |
+
"step": 820
|
| 5790 |
+
},
|
| 5791 |
+
{
|
| 5792 |
+
"epoch": 0.2626779715245561,
|
| 5793 |
+
"grad_norm": 42018.765625,
|
| 5794 |
+
"learning_rate": 0.00033747380240118773,
|
| 5795 |
+
"loss": 16.1286,
|
| 5796 |
+
"step": 821
|
| 5797 |
+
},
|
| 5798 |
+
{
|
| 5799 |
+
"epoch": 0.26299792033274677,
|
| 5800 |
+
"grad_norm": 2374.1533203125,
|
| 5801 |
+
"learning_rate": 0.0003373286122247823,
|
| 5802 |
+
"loss": 17.0077,
|
| 5803 |
+
"step": 822
|
| 5804 |
+
},
|
| 5805 |
+
{
|
| 5806 |
+
"epoch": 0.26331786914093747,
|
| 5807 |
+
"grad_norm": 2869.451904296875,
|
| 5808 |
+
"learning_rate": 0.000337183286957496,
|
| 5809 |
+
"loss": 16.8362,
|
| 5810 |
+
"step": 823
|
| 5811 |
+
},
|
| 5812 |
+
{
|
| 5813 |
+
"epoch": 0.26363781794912816,
|
| 5814 |
+
"grad_norm": 169.99826049804688,
|
| 5815 |
+
"learning_rate": 0.0003370378267535132,
|
| 5816 |
+
"loss": 16.7969,
|
| 5817 |
+
"step": 824
|
| 5818 |
+
},
|
| 5819 |
+
{
|
| 5820 |
+
"epoch": 0.2639577667573188,
|
| 5821 |
+
"grad_norm": 118.742919921875,
|
| 5822 |
+
"learning_rate": 0.00033689223176716187,
|
| 5823 |
+
"loss": 13.8806,
|
| 5824 |
+
"step": 825
|
| 5825 |
+
},
|
| 5826 |
+
{
|
| 5827 |
+
"epoch": 0.2642777155655095,
|
| 5828 |
+
"grad_norm": 21638.884765625,
|
| 5829 |
+
"learning_rate": 0.00033674650215291296,
|
| 5830 |
+
"loss": 12.1067,
|
| 5831 |
+
"step": 826
|
| 5832 |
+
},
|
| 5833 |
+
{
|
| 5834 |
+
"epoch": 0.2645976643737002,
|
| 5835 |
+
"grad_norm": 2294.127197265625,
|
| 5836 |
+
"learning_rate": 0.00033660063806537993,
|
| 5837 |
+
"loss": 12.1123,
|
| 5838 |
+
"step": 827
|
| 5839 |
+
},
|
| 5840 |
+
{
|
| 5841 |
+
"epoch": 0.2649176131818909,
|
| 5842 |
+
"grad_norm": 34425.94921875,
|
| 5843 |
+
"learning_rate": 0.0003364546396593192,
|
| 5844 |
+
"loss": 13.1807,
|
| 5845 |
+
"step": 828
|
| 5846 |
+
},
|
| 5847 |
+
{
|
| 5848 |
+
"epoch": 0.2652375619900816,
|
| 5849 |
+
"grad_norm": 600.3480224609375,
|
| 5850 |
+
"learning_rate": 0.00033630850708962957,
|
| 5851 |
+
"loss": 12.5817,
|
| 5852 |
+
"step": 829
|
| 5853 |
+
},
|
| 5854 |
+
{
|
| 5855 |
+
"epoch": 0.2655575107982723,
|
| 5856 |
+
"grad_norm": 82.4616928100586,
|
| 5857 |
+
"learning_rate": 0.00033616224051135225,
|
| 5858 |
+
"loss": 12.0756,
|
| 5859 |
+
"step": 830
|
| 5860 |
+
},
|
| 5861 |
+
{
|
| 5862 |
+
"epoch": 0.265877459606463,
|
| 5863 |
+
"grad_norm": 69.83650207519531,
|
| 5864 |
+
"learning_rate": 0.0003360158400796705,
|
| 5865 |
+
"loss": 10.1748,
|
| 5866 |
+
"step": 831
|
| 5867 |
+
},
|
| 5868 |
+
{
|
| 5869 |
+
"epoch": 0.26619740841465367,
|
| 5870 |
+
"grad_norm": 150.00209045410156,
|
| 5871 |
+
"learning_rate": 0.0003358693059499098,
|
| 5872 |
+
"loss": 8.8591,
|
| 5873 |
+
"step": 832
|
| 5874 |
+
},
|
| 5875 |
+
{
|
| 5876 |
+
"epoch": 0.26651735722284436,
|
| 5877 |
+
"grad_norm": 30079.291015625,
|
| 5878 |
+
"learning_rate": 0.0003357226382775373,
|
| 5879 |
+
"loss": 8.8002,
|
| 5880 |
+
"step": 833
|
| 5881 |
+
},
|
| 5882 |
+
{
|
| 5883 |
+
"epoch": 0.26683730603103506,
|
| 5884 |
+
"grad_norm": 18880.88671875,
|
| 5885 |
+
"learning_rate": 0.00033557583721816194,
|
| 5886 |
+
"loss": 10.3358,
|
| 5887 |
+
"step": 834
|
| 5888 |
+
},
|
| 5889 |
+
{
|
| 5890 |
+
"epoch": 0.2671572548392257,
|
| 5891 |
+
"grad_norm": 7514.67333984375,
|
| 5892 |
+
"learning_rate": 0.00033542890292753404,
|
| 5893 |
+
"loss": 12.7088,
|
| 5894 |
+
"step": 835
|
| 5895 |
+
},
|
| 5896 |
+
{
|
| 5897 |
+
"epoch": 0.2674772036474164,
|
| 5898 |
+
"grad_norm": 165107.78125,
|
| 5899 |
+
"learning_rate": 0.0003352818355615455,
|
| 5900 |
+
"loss": 15.0574,
|
| 5901 |
+
"step": 836
|
| 5902 |
+
},
|
| 5903 |
+
{
|
| 5904 |
+
"epoch": 0.2677971524556071,
|
| 5905 |
+
"grad_norm": 10771.0146484375,
|
| 5906 |
+
"learning_rate": 0.0003351346352762292,
|
| 5907 |
+
"loss": 16.6562,
|
| 5908 |
+
"step": 837
|
| 5909 |
+
},
|
| 5910 |
+
{
|
| 5911 |
+
"epoch": 0.2681171012637978,
|
| 5912 |
+
"grad_norm": 32579.517578125,
|
| 5913 |
+
"learning_rate": 0.00033498730222775917,
|
| 5914 |
+
"loss": 16.5778,
|
| 5915 |
+
"step": 838
|
| 5916 |
+
},
|
| 5917 |
+
{
|
| 5918 |
+
"epoch": 0.2684370500719885,
|
| 5919 |
+
"grad_norm": 737471.125,
|
| 5920 |
+
"learning_rate": 0.00033483983657245017,
|
| 5921 |
+
"loss": 17.4677,
|
| 5922 |
+
"step": 839
|
| 5923 |
+
},
|
| 5924 |
+
{
|
| 5925 |
+
"epoch": 0.2687569988801792,
|
| 5926 |
+
"grad_norm": 1129485.5,
|
| 5927 |
+
"learning_rate": 0.00033469223846675783,
|
| 5928 |
+
"loss": 16.0327,
|
| 5929 |
+
"step": 840
|
| 5930 |
+
},
|
| 5931 |
+
{
|
| 5932 |
+
"epoch": 0.26907694768836987,
|
| 5933 |
+
"grad_norm": 21750792.0,
|
| 5934 |
+
"learning_rate": 0.0003345445080672781,
|
| 5935 |
+
"loss": 16.6764,
|
| 5936 |
+
"step": 841
|
| 5937 |
+
},
|
| 5938 |
+
{
|
| 5939 |
+
"epoch": 0.26939689649656057,
|
| 5940 |
+
"grad_norm": 626546.5625,
|
| 5941 |
+
"learning_rate": 0.0003343966455307474,
|
| 5942 |
+
"loss": 18.6074,
|
| 5943 |
+
"step": 842
|
| 5944 |
+
},
|
| 5945 |
+
{
|
| 5946 |
+
"epoch": 0.26971684530475126,
|
| 5947 |
+
"grad_norm": 1119687.125,
|
| 5948 |
+
"learning_rate": 0.0003342486510140424,
|
| 5949 |
+
"loss": 19.7998,
|
| 5950 |
+
"step": 843
|
| 5951 |
+
},
|
| 5952 |
+
{
|
| 5953 |
+
"epoch": 0.2700367941129419,
|
| 5954 |
+
"grad_norm": 4357446.0,
|
| 5955 |
+
"learning_rate": 0.00033410052467417985,
|
| 5956 |
+
"loss": 17.5433,
|
| 5957 |
+
"step": 844
|
| 5958 |
+
},
|
| 5959 |
+
{
|
| 5960 |
+
"epoch": 0.2703567429211326,
|
| 5961 |
+
"grad_norm": 617141.25,
|
| 5962 |
+
"learning_rate": 0.0003339522666683159,
|
| 5963 |
+
"loss": 19.1042,
|
| 5964 |
+
"step": 845
|
| 5965 |
+
},
|
| 5966 |
+
{
|
| 5967 |
+
"epoch": 0.2706766917293233,
|
| 5968 |
+
"grad_norm": 355878.9375,
|
| 5969 |
+
"learning_rate": 0.000333803877153747,
|
| 5970 |
+
"loss": 20.8072,
|
| 5971 |
+
"step": 846
|
| 5972 |
+
},
|
| 5973 |
+
{
|
| 5974 |
+
"epoch": 0.270996640537514,
|
| 5975 |
+
"grad_norm": 624676.4375,
|
| 5976 |
+
"learning_rate": 0.0003336553562879088,
|
| 5977 |
+
"loss": 14.6778,
|
| 5978 |
+
"step": 847
|
| 5979 |
+
},
|
| 5980 |
+
{
|
| 5981 |
+
"epoch": 0.2713165893457047,
|
| 5982 |
+
"grad_norm": 330477.0625,
|
| 5983 |
+
"learning_rate": 0.00033350670422837637,
|
| 5984 |
+
"loss": 19.176,
|
| 5985 |
+
"step": 848
|
| 5986 |
+
},
|
| 5987 |
+
{
|
| 5988 |
+
"epoch": 0.2716365381538954,
|
| 5989 |
+
"grad_norm": 941151.25,
|
| 5990 |
+
"learning_rate": 0.0003333579211328639,
|
| 5991 |
+
"loss": 21.5224,
|
| 5992 |
+
"step": 849
|
| 5993 |
+
},
|
| 5994 |
+
{
|
| 5995 |
+
"epoch": 0.2719564869620861,
|
| 5996 |
+
"grad_norm": 3218129.5,
|
| 5997 |
+
"learning_rate": 0.00033320900715922477,
|
| 5998 |
+
"loss": 23.2502,
|
| 5999 |
+
"step": 850
|
| 6000 |
+
},
|
| 6001 |
+
{
|
| 6002 |
+
"epoch": 0.27227643577027677,
|
| 6003 |
+
"grad_norm": 1361448.25,
|
| 6004 |
+
"learning_rate": 0.00033305996246545106,
|
| 6005 |
+
"loss": 23.624,
|
| 6006 |
+
"step": 851
|
| 6007 |
+
},
|
| 6008 |
+
{
|
| 6009 |
+
"epoch": 0.27259638457846747,
|
| 6010 |
+
"grad_norm": 282881.1875,
|
| 6011 |
+
"learning_rate": 0.00033291078720967365,
|
| 6012 |
+
"loss": 24.8605,
|
| 6013 |
+
"step": 852
|
| 6014 |
+
},
|
| 6015 |
+
{
|
| 6016 |
+
"epoch": 0.27291633338665816,
|
| 6017 |
+
"grad_norm": 652137.875,
|
| 6018 |
+
"learning_rate": 0.0003327614815501618,
|
| 6019 |
+
"loss": 25.1488,
|
| 6020 |
+
"step": 853
|
| 6021 |
+
},
|
| 6022 |
+
{
|
| 6023 |
+
"epoch": 0.2732362821948488,
|
| 6024 |
+
"grad_norm": 11083729.0,
|
| 6025 |
+
"learning_rate": 0.0003326120456453232,
|
| 6026 |
+
"loss": 26.0858,
|
| 6027 |
+
"step": 854
|
| 6028 |
+
},
|
| 6029 |
+
{
|
| 6030 |
+
"epoch": 0.2735562310030395,
|
| 6031 |
+
"grad_norm": 3168683.75,
|
| 6032 |
+
"learning_rate": 0.0003324624796537039,
|
| 6033 |
+
"loss": 26.6461,
|
| 6034 |
+
"step": 855
|
| 6035 |
+
},
|
| 6036 |
+
{
|
| 6037 |
+
"epoch": 0.2738761798112302,
|
| 6038 |
+
"grad_norm": 3187546.0,
|
| 6039 |
+
"learning_rate": 0.00033231278373398773,
|
| 6040 |
+
"loss": 25.3838,
|
| 6041 |
+
"step": 856
|
| 6042 |
+
},
|
| 6043 |
+
{
|
| 6044 |
+
"epoch": 0.2741961286194209,
|
| 6045 |
+
"grad_norm": 396312.03125,
|
| 6046 |
+
"learning_rate": 0.00033216295804499636,
|
| 6047 |
+
"loss": 25.4869,
|
| 6048 |
+
"step": 857
|
| 6049 |
+
},
|
| 6050 |
+
{
|
| 6051 |
+
"epoch": 0.2745160774276116,
|
| 6052 |
+
"grad_norm": 1157155.25,
|
| 6053 |
+
"learning_rate": 0.00033201300274568934,
|
| 6054 |
+
"loss": 26.773,
|
| 6055 |
+
"step": 858
|
| 6056 |
+
},
|
| 6057 |
+
{
|
| 6058 |
+
"epoch": 0.2748360262358023,
|
| 6059 |
+
"grad_norm": 1152320.75,
|
| 6060 |
+
"learning_rate": 0.00033186291799516354,
|
| 6061 |
+
"loss": 26.3161,
|
| 6062 |
+
"step": 859
|
| 6063 |
+
},
|
| 6064 |
+
{
|
| 6065 |
+
"epoch": 0.275155975043993,
|
| 6066 |
+
"grad_norm": 3598519.75,
|
| 6067 |
+
"learning_rate": 0.00033171270395265334,
|
| 6068 |
+
"loss": 24.7172,
|
| 6069 |
+
"step": 860
|
| 6070 |
+
},
|
| 6071 |
+
{
|
| 6072 |
+
"epoch": 0.27547592385218367,
|
| 6073 |
+
"grad_norm": 4380961.0,
|
| 6074 |
+
"learning_rate": 0.00033156236077753006,
|
| 6075 |
+
"loss": 25.4326,
|
| 6076 |
+
"step": 861
|
| 6077 |
+
},
|
| 6078 |
+
{
|
| 6079 |
+
"epoch": 0.27579587266037436,
|
| 6080 |
+
"grad_norm": 219964.46875,
|
| 6081 |
+
"learning_rate": 0.00033141188862930235,
|
| 6082 |
+
"loss": 25.278,
|
| 6083 |
+
"step": 862
|
| 6084 |
+
},
|
| 6085 |
+
{
|
| 6086 |
+
"epoch": 0.276115821468565,
|
| 6087 |
+
"grad_norm": 9706468.0,
|
| 6088 |
+
"learning_rate": 0.0003312612876676153,
|
| 6089 |
+
"loss": 24.5966,
|
| 6090 |
+
"step": 863
|
| 6091 |
+
},
|
| 6092 |
+
{
|
| 6093 |
+
"epoch": 0.2764357702767557,
|
| 6094 |
+
"grad_norm": 398041.8125,
|
| 6095 |
+
"learning_rate": 0.0003311105580522511,
|
| 6096 |
+
"loss": 23.8814,
|
| 6097 |
+
"step": 864
|
| 6098 |
+
},
|
| 6099 |
+
{
|
| 6100 |
+
"epoch": 0.2767557190849464,
|
| 6101 |
+
"grad_norm": 520553.59375,
|
| 6102 |
+
"learning_rate": 0.00033095969994312806,
|
| 6103 |
+
"loss": 22.963,
|
| 6104 |
+
"step": 865
|
| 6105 |
+
},
|
| 6106 |
+
{
|
| 6107 |
+
"epoch": 0.2770756678931371,
|
| 6108 |
+
"grad_norm": 210379.578125,
|
| 6109 |
+
"learning_rate": 0.00033080871350030103,
|
| 6110 |
+
"loss": 25.1315,
|
| 6111 |
+
"step": 866
|
| 6112 |
+
},
|
| 6113 |
+
{
|
| 6114 |
+
"epoch": 0.2773956167013278,
|
| 6115 |
+
"grad_norm": 1593170.0,
|
| 6116 |
+
"learning_rate": 0.00033065759888396097,
|
| 6117 |
+
"loss": 25.1688,
|
| 6118 |
+
"step": 867
|
| 6119 |
+
},
|
| 6120 |
+
{
|
| 6121 |
+
"epoch": 0.2777155655095185,
|
| 6122 |
+
"grad_norm": 714547.0625,
|
| 6123 |
+
"learning_rate": 0.0003305063562544348,
|
| 6124 |
+
"loss": 24.1757,
|
| 6125 |
+
"step": 868
|
| 6126 |
+
},
|
| 6127 |
+
{
|
| 6128 |
+
"epoch": 0.2780355143177092,
|
| 6129 |
+
"grad_norm": 1169557.25,
|
| 6130 |
+
"learning_rate": 0.00033035498577218533,
|
| 6131 |
+
"loss": 24.4053,
|
| 6132 |
+
"step": 869
|
| 6133 |
+
},
|
| 6134 |
+
{
|
| 6135 |
+
"epoch": 0.27835546312589987,
|
| 6136 |
+
"grad_norm": 957691.8125,
|
| 6137 |
+
"learning_rate": 0.0003302034875978109,
|
| 6138 |
+
"loss": 22.7094,
|
| 6139 |
+
"step": 870
|
| 6140 |
+
},
|
| 6141 |
+
{
|
| 6142 |
+
"epoch": 0.27867541193409057,
|
| 6143 |
+
"grad_norm": 261482.71875,
|
| 6144 |
+
"learning_rate": 0.0003300518618920454,
|
| 6145 |
+
"loss": 22.1931,
|
| 6146 |
+
"step": 871
|
| 6147 |
+
},
|
| 6148 |
+
{
|
| 6149 |
+
"epoch": 0.27899536074228126,
|
| 6150 |
+
"grad_norm": 193108.375,
|
| 6151 |
+
"learning_rate": 0.00032990010881575805,
|
| 6152 |
+
"loss": 23.5923,
|
| 6153 |
+
"step": 872
|
| 6154 |
+
},
|
| 6155 |
+
{
|
| 6156 |
+
"epoch": 0.2793153095504719,
|
| 6157 |
+
"grad_norm": 75381.0078125,
|
| 6158 |
+
"learning_rate": 0.00032974822852995314,
|
| 6159 |
+
"loss": 22.1936,
|
| 6160 |
+
"step": 873
|
| 6161 |
+
},
|
| 6162 |
+
{
|
| 6163 |
+
"epoch": 0.2796352583586626,
|
| 6164 |
+
"grad_norm": 101304.7265625,
|
| 6165 |
+
"learning_rate": 0.00032959622119576996,
|
| 6166 |
+
"loss": 23.972,
|
| 6167 |
+
"step": 874
|
| 6168 |
+
},
|
| 6169 |
+
{
|
| 6170 |
+
"epoch": 0.2799552071668533,
|
| 6171 |
+
"grad_norm": 797569.8125,
|
| 6172 |
+
"learning_rate": 0.00032944408697448255,
|
| 6173 |
+
"loss": 23.1834,
|
| 6174 |
+
"step": 875
|
| 6175 |
+
},
|
| 6176 |
+
{
|
| 6177 |
+
"epoch": 0.280275155975044,
|
| 6178 |
+
"grad_norm": 54563.6953125,
|
| 6179 |
+
"learning_rate": 0.0003292918260274997,
|
| 6180 |
+
"loss": 19.8762,
|
| 6181 |
+
"step": 876
|
| 6182 |
+
},
|
| 6183 |
+
{
|
| 6184 |
+
"epoch": 0.2805951047832347,
|
| 6185 |
+
"grad_norm": 765206.625,
|
| 6186 |
+
"learning_rate": 0.0003291394385163645,
|
| 6187 |
+
"loss": 17.3695,
|
| 6188 |
+
"step": 877
|
| 6189 |
+
},
|
| 6190 |
+
{
|
| 6191 |
+
"epoch": 0.2809150535914254,
|
| 6192 |
+
"grad_norm": 145896.859375,
|
| 6193 |
+
"learning_rate": 0.00032898692460275446,
|
| 6194 |
+
"loss": 14.1639,
|
| 6195 |
+
"step": 878
|
| 6196 |
+
},
|
| 6197 |
+
{
|
| 6198 |
+
"epoch": 0.2812350023996161,
|
| 6199 |
+
"grad_norm": 78758.6640625,
|
| 6200 |
+
"learning_rate": 0.000328834284448481,
|
| 6201 |
+
"loss": 18.8681,
|
| 6202 |
+
"step": 879
|
| 6203 |
+
},
|
| 6204 |
+
{
|
| 6205 |
+
"epoch": 0.28155495120780677,
|
| 6206 |
+
"grad_norm": 878149.4375,
|
| 6207 |
+
"learning_rate": 0.00032868151821548966,
|
| 6208 |
+
"loss": 25.91,
|
| 6209 |
+
"step": 880
|
| 6210 |
+
},
|
| 6211 |
+
{
|
| 6212 |
+
"epoch": 0.28187490001599746,
|
| 6213 |
+
"grad_norm": 2420195.5,
|
| 6214 |
+
"learning_rate": 0.00032852862606585966,
|
| 6215 |
+
"loss": 30.7789,
|
| 6216 |
+
"step": 881
|
| 6217 |
+
},
|
| 6218 |
+
{
|
| 6219 |
+
"epoch": 0.2821948488241881,
|
| 6220 |
+
"grad_norm": 305033.84375,
|
| 6221 |
+
"learning_rate": 0.0003283756081618039,
|
| 6222 |
+
"loss": 32.8377,
|
| 6223 |
+
"step": 882
|
| 6224 |
+
},
|
| 6225 |
+
{
|
| 6226 |
+
"epoch": 0.2825147976323788,
|
| 6227 |
+
"grad_norm": 422428.90625,
|
| 6228 |
+
"learning_rate": 0.00032822246466566855,
|
| 6229 |
+
"loss": 32.3465,
|
| 6230 |
+
"step": 883
|
| 6231 |
+
},
|
| 6232 |
+
{
|
| 6233 |
+
"epoch": 0.2828347464405695,
|
| 6234 |
+
"grad_norm": 78632.2578125,
|
| 6235 |
+
"learning_rate": 0.0003280691957399332,
|
| 6236 |
+
"loss": 32.9845,
|
| 6237 |
+
"step": 884
|
| 6238 |
+
},
|
| 6239 |
+
{
|
| 6240 |
+
"epoch": 0.2831546952487602,
|
| 6241 |
+
"grad_norm": 286373.71875,
|
| 6242 |
+
"learning_rate": 0.00032791580154721035,
|
| 6243 |
+
"loss": 34.3491,
|
| 6244 |
+
"step": 885
|
| 6245 |
+
},
|
| 6246 |
+
{
|
| 6247 |
+
"epoch": 0.2834746440569509,
|
| 6248 |
+
"grad_norm": 12756404.0,
|
| 6249 |
+
"learning_rate": 0.0003277622822502455,
|
| 6250 |
+
"loss": 32.8115,
|
| 6251 |
+
"step": 886
|
| 6252 |
+
},
|
| 6253 |
+
{
|
| 6254 |
+
"epoch": 0.2837945928651416,
|
| 6255 |
+
"grad_norm": 259764.84375,
|
| 6256 |
+
"learning_rate": 0.00032760863801191696,
|
| 6257 |
+
"loss": 33.0063,
|
| 6258 |
+
"step": 887
|
| 6259 |
+
},
|
| 6260 |
+
{
|
| 6261 |
+
"epoch": 0.2841145416733323,
|
| 6262 |
+
"grad_norm": 1352647.25,
|
| 6263 |
+
"learning_rate": 0.0003274548689952354,
|
| 6264 |
+
"loss": 33.3023,
|
| 6265 |
+
"step": 888
|
| 6266 |
+
},
|
| 6267 |
+
{
|
| 6268 |
+
"epoch": 0.284434490481523,
|
| 6269 |
+
"grad_norm": 2583883.25,
|
| 6270 |
+
"learning_rate": 0.000327300975363344,
|
| 6271 |
+
"loss": 34.0707,
|
| 6272 |
+
"step": 889
|
| 6273 |
+
},
|
| 6274 |
+
{
|
| 6275 |
+
"epoch": 0.28475443928971367,
|
| 6276 |
+
"grad_norm": 768095.625,
|
| 6277 |
+
"learning_rate": 0.0003271469572795181,
|
| 6278 |
+
"loss": 33.3534,
|
| 6279 |
+
"step": 890
|
| 6280 |
+
},
|
| 6281 |
+
{
|
| 6282 |
+
"epoch": 0.2850743880979043,
|
| 6283 |
+
"grad_norm": 38086.27734375,
|
| 6284 |
+
"learning_rate": 0.0003269928149071652,
|
| 6285 |
+
"loss": 35.1093,
|
| 6286 |
+
"step": 891
|
| 6287 |
+
},
|
| 6288 |
+
{
|
| 6289 |
+
"epoch": 0.285394336906095,
|
| 6290 |
+
"grad_norm": 11101828.0,
|
| 6291 |
+
"learning_rate": 0.00032683854840982453,
|
| 6292 |
+
"loss": 36.15,
|
| 6293 |
+
"step": 892
|
| 6294 |
+
},
|
| 6295 |
+
{
|
| 6296 |
+
"epoch": 0.2857142857142857,
|
| 6297 |
+
"grad_norm": 936936.6875,
|
| 6298 |
+
"learning_rate": 0.000326684157951167,
|
| 6299 |
+
"loss": 35.3095,
|
| 6300 |
+
"step": 893
|
| 6301 |
+
},
|
| 6302 |
+
{
|
| 6303 |
+
"epoch": 0.2860342345224764,
|
| 6304 |
+
"grad_norm": 1324180.875,
|
| 6305 |
+
"learning_rate": 0.0003265296436949952,
|
| 6306 |
+
"loss": 34.8645,
|
| 6307 |
+
"step": 894
|
| 6308 |
+
},
|
| 6309 |
+
{
|
| 6310 |
+
"epoch": 0.2863541833306671,
|
| 6311 |
+
"grad_norm": 108139.625,
|
| 6312 |
+
"learning_rate": 0.00032637500580524285,
|
| 6313 |
+
"loss": 34.3561,
|
| 6314 |
+
"step": 895
|
| 6315 |
+
},
|
| 6316 |
+
{
|
| 6317 |
+
"epoch": 0.2866741321388578,
|
| 6318 |
+
"grad_norm": 200987.484375,
|
| 6319 |
+
"learning_rate": 0.00032622024444597497,
|
| 6320 |
+
"loss": 36.2547,
|
| 6321 |
+
"step": 896
|
| 6322 |
+
},
|
| 6323 |
+
{
|
| 6324 |
+
"epoch": 0.2869940809470485,
|
| 6325 |
+
"grad_norm": 147518.8125,
|
| 6326 |
+
"learning_rate": 0.00032606535978138763,
|
| 6327 |
+
"loss": 34.7164,
|
| 6328 |
+
"step": 897
|
| 6329 |
+
},
|
| 6330 |
+
{
|
| 6331 |
+
"epoch": 0.2873140297552392,
|
| 6332 |
+
"grad_norm": 1045666.5625,
|
| 6333 |
+
"learning_rate": 0.0003259103519758076,
|
| 6334 |
+
"loss": 34.6775,
|
| 6335 |
+
"step": 898
|
| 6336 |
+
},
|
| 6337 |
+
{
|
| 6338 |
+
"epoch": 0.28763397856342987,
|
| 6339 |
+
"grad_norm": 728341.4375,
|
| 6340 |
+
"learning_rate": 0.0003257552211936924,
|
| 6341 |
+
"loss": 34.3695,
|
| 6342 |
+
"step": 899
|
| 6343 |
+
},
|
| 6344 |
+
{
|
| 6345 |
+
"epoch": 0.28795392737162057,
|
| 6346 |
+
"grad_norm": 47716.6015625,
|
| 6347 |
+
"learning_rate": 0.00032559996759962985,
|
| 6348 |
+
"loss": 34.6398,
|
| 6349 |
+
"step": 900
|
| 6350 |
+
},
|
| 6351 |
+
{
|
| 6352 |
+
"epoch": 0.2882738761798112,
|
| 6353 |
+
"grad_norm": 2483745.0,
|
| 6354 |
+
"learning_rate": 0.0003254445913583383,
|
| 6355 |
+
"loss": 34.867,
|
| 6356 |
+
"step": 901
|
| 6357 |
+
},
|
| 6358 |
+
{
|
| 6359 |
+
"epoch": 0.2885938249880019,
|
| 6360 |
+
"grad_norm": 4208087.5,
|
| 6361 |
+
"learning_rate": 0.00032528909263466614,
|
| 6362 |
+
"loss": 33.2978,
|
| 6363 |
+
"step": 902
|
| 6364 |
+
},
|
| 6365 |
+
{
|
| 6366 |
+
"epoch": 0.2889137737961926,
|
| 6367 |
+
"grad_norm": 668148.4375,
|
| 6368 |
+
"learning_rate": 0.0003251334715935915,
|
| 6369 |
+
"loss": 32.0358,
|
| 6370 |
+
"step": 903
|
| 6371 |
+
},
|
| 6372 |
+
{
|
| 6373 |
+
"epoch": 0.2892337226043833,
|
| 6374 |
+
"grad_norm": 382070.53125,
|
| 6375 |
+
"learning_rate": 0.0003249777284002227,
|
| 6376 |
+
"loss": 34.4869,
|
| 6377 |
+
"step": 904
|
| 6378 |
+
},
|
| 6379 |
+
{
|
| 6380 |
+
"epoch": 0.289553671412574,
|
| 6381 |
+
"grad_norm": 2095629.125,
|
| 6382 |
+
"learning_rate": 0.00032482186321979717,
|
| 6383 |
+
"loss": 32.9963,
|
| 6384 |
+
"step": 905
|
| 6385 |
+
},
|
| 6386 |
+
{
|
| 6387 |
+
"epoch": 0.2898736202207647,
|
| 6388 |
+
"grad_norm": 721848.8125,
|
| 6389 |
+
"learning_rate": 0.00032466587621768226,
|
| 6390 |
+
"loss": 36.1113,
|
| 6391 |
+
"step": 906
|
| 6392 |
+
},
|
| 6393 |
+
{
|
| 6394 |
+
"epoch": 0.2901935690289554,
|
| 6395 |
+
"grad_norm": 826675.1875,
|
| 6396 |
+
"learning_rate": 0.0003245097675593743,
|
| 6397 |
+
"loss": 33.8295,
|
| 6398 |
+
"step": 907
|
| 6399 |
+
},
|
| 6400 |
+
{
|
| 6401 |
+
"epoch": 0.2905135178371461,
|
| 6402 |
+
"grad_norm": 1047299.1875,
|
| 6403 |
+
"learning_rate": 0.0003243535374104985,
|
| 6404 |
+
"loss": 33.8889,
|
| 6405 |
+
"step": 908
|
| 6406 |
+
},
|
| 6407 |
+
{
|
| 6408 |
+
"epoch": 0.29083346664533677,
|
| 6409 |
+
"grad_norm": 83939.1484375,
|
| 6410 |
+
"learning_rate": 0.00032419718593680943,
|
| 6411 |
+
"loss": 34.5913,
|
| 6412 |
+
"step": 909
|
| 6413 |
+
},
|
| 6414 |
+
{
|
| 6415 |
+
"epoch": 0.2911534154535274,
|
| 6416 |
+
"grad_norm": 69454.4609375,
|
| 6417 |
+
"learning_rate": 0.00032404071330419,
|
| 6418 |
+
"loss": 34.2654,
|
| 6419 |
+
"step": 910
|
| 6420 |
+
},
|
| 6421 |
+
{
|
| 6422 |
+
"epoch": 0.2914733642617181,
|
| 6423 |
+
"grad_norm": 22721304.0,
|
| 6424 |
+
"learning_rate": 0.00032388411967865184,
|
| 6425 |
+
"loss": 33.8081,
|
| 6426 |
+
"step": 911
|
| 6427 |
+
},
|
| 6428 |
+
{
|
| 6429 |
+
"epoch": 0.2917933130699088,
|
| 6430 |
+
"grad_norm": 357308.65625,
|
| 6431 |
+
"learning_rate": 0.0003237274052263349,
|
| 6432 |
+
"loss": 32.9188,
|
| 6433 |
+
"step": 912
|
| 6434 |
+
},
|
| 6435 |
+
{
|
| 6436 |
+
"epoch": 0.2921132618780995,
|
| 6437 |
+
"grad_norm": 278778.5,
|
| 6438 |
+
"learning_rate": 0.0003235705701135074,
|
| 6439 |
+
"loss": 32.0302,
|
| 6440 |
+
"step": 913
|
| 6441 |
+
},
|
| 6442 |
+
{
|
| 6443 |
+
"epoch": 0.2924332106862902,
|
| 6444 |
+
"grad_norm": 4960189.5,
|
| 6445 |
+
"learning_rate": 0.0003234136145065655,
|
| 6446 |
+
"loss": 32.5315,
|
| 6447 |
+
"step": 914
|
| 6448 |
+
},
|
| 6449 |
+
{
|
| 6450 |
+
"epoch": 0.2927531594944809,
|
| 6451 |
+
"grad_norm": 55529.8125,
|
| 6452 |
+
"learning_rate": 0.0003232565385720331,
|
| 6453 |
+
"loss": 32.6047,
|
| 6454 |
+
"step": 915
|
| 6455 |
+
},
|
| 6456 |
+
{
|
| 6457 |
+
"epoch": 0.2930731083026716,
|
| 6458 |
+
"grad_norm": 1051795.0,
|
| 6459 |
+
"learning_rate": 0.000323099342476562,
|
| 6460 |
+
"loss": 32.6275,
|
| 6461 |
+
"step": 916
|
| 6462 |
+
},
|
| 6463 |
+
{
|
| 6464 |
+
"epoch": 0.2933930571108623,
|
| 6465 |
+
"grad_norm": 1545867.0,
|
| 6466 |
+
"learning_rate": 0.0003229420263869313,
|
| 6467 |
+
"loss": 34.8654,
|
| 6468 |
+
"step": 917
|
| 6469 |
+
},
|
| 6470 |
+
{
|
| 6471 |
+
"epoch": 0.29371300591905297,
|
| 6472 |
+
"grad_norm": 58457.1015625,
|
| 6473 |
+
"learning_rate": 0.0003227845904700475,
|
| 6474 |
+
"loss": 30.3776,
|
| 6475 |
+
"step": 918
|
| 6476 |
+
},
|
| 6477 |
+
{
|
| 6478 |
+
"epoch": 0.29403295472724367,
|
| 6479 |
+
"grad_norm": 1061561.125,
|
| 6480 |
+
"learning_rate": 0.00032262703489294414,
|
| 6481 |
+
"loss": 31.8167,
|
| 6482 |
+
"step": 919
|
| 6483 |
+
},
|
| 6484 |
+
{
|
| 6485 |
+
"epoch": 0.2943529035354343,
|
| 6486 |
+
"grad_norm": 1284448.125,
|
| 6487 |
+
"learning_rate": 0.00032246935982278185,
|
| 6488 |
+
"loss": 31.0466,
|
| 6489 |
+
"step": 920
|
| 6490 |
+
},
|
| 6491 |
+
{
|
| 6492 |
+
"epoch": 0.294672852343625,
|
| 6493 |
+
"grad_norm": 319359.875,
|
| 6494 |
+
"learning_rate": 0.0003223115654268479,
|
| 6495 |
+
"loss": 29.9253,
|
| 6496 |
+
"step": 921
|
| 6497 |
+
},
|
| 6498 |
+
{
|
| 6499 |
+
"epoch": 0.2949928011518157,
|
| 6500 |
+
"grad_norm": 1560787.0,
|
| 6501 |
+
"learning_rate": 0.0003221536518725563,
|
| 6502 |
+
"loss": 30.4316,
|
| 6503 |
+
"step": 922
|
| 6504 |
+
},
|
| 6505 |
+
{
|
| 6506 |
+
"epoch": 0.2953127499600064,
|
| 6507 |
+
"grad_norm": 2599579.75,
|
| 6508 |
+
"learning_rate": 0.00032199561932744734,
|
| 6509 |
+
"loss": 29.6555,
|
| 6510 |
+
"step": 923
|
| 6511 |
+
},
|
| 6512 |
+
{
|
| 6513 |
+
"epoch": 0.2956326987681971,
|
| 6514 |
+
"grad_norm": 76841.046875,
|
| 6515 |
+
"learning_rate": 0.0003218374679591877,
|
| 6516 |
+
"loss": 29.959,
|
| 6517 |
+
"step": 924
|
| 6518 |
+
},
|
| 6519 |
+
{
|
| 6520 |
+
"epoch": 0.2959526475763878,
|
| 6521 |
+
"grad_norm": 34526.2109375,
|
| 6522 |
+
"learning_rate": 0.0003216791979355699,
|
| 6523 |
+
"loss": 29.0153,
|
| 6524 |
+
"step": 925
|
| 6525 |
+
},
|
| 6526 |
+
{
|
| 6527 |
+
"epoch": 0.2962725963845785,
|
| 6528 |
+
"grad_norm": 8784803.0,
|
| 6529 |
+
"learning_rate": 0.0003215208094245127,
|
| 6530 |
+
"loss": 30.9955,
|
| 6531 |
+
"step": 926
|
| 6532 |
+
},
|
| 6533 |
+
{
|
| 6534 |
+
"epoch": 0.2965925451927692,
|
| 6535 |
+
"grad_norm": 39772.76171875,
|
| 6536 |
+
"learning_rate": 0.00032136230259406016,
|
| 6537 |
+
"loss": 31.03,
|
| 6538 |
+
"step": 927
|
| 6539 |
+
},
|
| 6540 |
+
{
|
| 6541 |
+
"epoch": 0.29691249400095987,
|
| 6542 |
+
"grad_norm": 44792.16796875,
|
| 6543 |
+
"learning_rate": 0.00032120367761238227,
|
| 6544 |
+
"loss": 31.4258,
|
| 6545 |
+
"step": 928
|
| 6546 |
+
},
|
| 6547 |
+
{
|
| 6548 |
+
"epoch": 0.2972324428091505,
|
| 6549 |
+
"grad_norm": 399403.78125,
|
| 6550 |
+
"learning_rate": 0.00032104493464777404,
|
| 6551 |
+
"loss": 33.6536,
|
| 6552 |
+
"step": 929
|
| 6553 |
+
},
|
| 6554 |
+
{
|
| 6555 |
+
"epoch": 0.2975523916173412,
|
| 6556 |
+
"grad_norm": 869917.6875,
|
| 6557 |
+
"learning_rate": 0.0003208860738686558,
|
| 6558 |
+
"loss": 32.0902,
|
| 6559 |
+
"step": 930
|
| 6560 |
+
},
|
| 6561 |
+
{
|
| 6562 |
+
"epoch": 0.2978723404255319,
|
| 6563 |
+
"grad_norm": 78028.140625,
|
| 6564 |
+
"learning_rate": 0.00032072709544357306,
|
| 6565 |
+
"loss": 32.4948,
|
| 6566 |
+
"step": 931
|
| 6567 |
+
},
|
| 6568 |
+
{
|
| 6569 |
+
"epoch": 0.2981922892337226,
|
| 6570 |
+
"grad_norm": 3609.693603515625,
|
| 6571 |
+
"learning_rate": 0.0003205679995411958,
|
| 6572 |
+
"loss": 32.1421,
|
| 6573 |
+
"step": 932
|
| 6574 |
+
},
|
| 6575 |
+
{
|
| 6576 |
+
"epoch": 0.2985122380419133,
|
| 6577 |
+
"grad_norm": 36646.99609375,
|
| 6578 |
+
"learning_rate": 0.00032040878633031884,
|
| 6579 |
+
"loss": 25.8291,
|
| 6580 |
+
"step": 933
|
| 6581 |
+
},
|
| 6582 |
+
{
|
| 6583 |
+
"epoch": 0.298832186850104,
|
| 6584 |
+
"grad_norm": 6927.86328125,
|
| 6585 |
+
"learning_rate": 0.0003202494559798614,
|
| 6586 |
+
"loss": 26.697,
|
| 6587 |
+
"step": 934
|
| 6588 |
+
},
|
| 6589 |
+
{
|
| 6590 |
+
"epoch": 0.2991521356582947,
|
| 6591 |
+
"grad_norm": 745786.875,
|
| 6592 |
+
"learning_rate": 0.0003200900086588671,
|
| 6593 |
+
"loss": 22.2229,
|
| 6594 |
+
"step": 935
|
| 6595 |
+
},
|
| 6596 |
+
{
|
| 6597 |
+
"epoch": 0.2994720844664854,
|
| 6598 |
+
"grad_norm": 208755.265625,
|
| 6599 |
+
"learning_rate": 0.0003199304445365035,
|
| 6600 |
+
"loss": 18.1356,
|
| 6601 |
+
"step": 936
|
| 6602 |
+
},
|
| 6603 |
+
{
|
| 6604 |
+
"epoch": 0.2997920332746761,
|
| 6605 |
+
"grad_norm": 7917.80859375,
|
| 6606 |
+
"learning_rate": 0.0003197707637820621,
|
| 6607 |
+
"loss": 16.6657,
|
| 6608 |
+
"step": 937
|
| 6609 |
+
},
|
| 6610 |
+
{
|
| 6611 |
+
"epoch": 0.30011198208286677,
|
| 6612 |
+
"grad_norm": 498.3856506347656,
|
| 6613 |
+
"learning_rate": 0.00031961096656495824,
|
| 6614 |
+
"loss": 15.2576,
|
| 6615 |
+
"step": 938
|
| 6616 |
+
},
|
| 6617 |
+
{
|
| 6618 |
+
"epoch": 0.3004319308910574,
|
| 6619 |
+
"grad_norm": 245.81466674804688,
|
| 6620 |
+
"learning_rate": 0.00031945105305473075,
|
| 6621 |
+
"loss": 10.8893,
|
| 6622 |
+
"step": 939
|
| 6623 |
+
},
|
| 6624 |
+
{
|
| 6625 |
+
"epoch": 0.3007518796992481,
|
| 6626 |
+
"grad_norm": 946.4498291015625,
|
| 6627 |
+
"learning_rate": 0.000319291023421042,
|
| 6628 |
+
"loss": 7.6083,
|
| 6629 |
+
"step": 940
|
| 6630 |
+
},
|
| 6631 |
+
{
|
| 6632 |
+
"epoch": 0.3010718285074388,
|
| 6633 |
+
"grad_norm": 136.80340576171875,
|
| 6634 |
+
"learning_rate": 0.0003191308778336772,
|
| 6635 |
+
"loss": 7.4698,
|
| 6636 |
+
"step": 941
|
| 6637 |
+
},
|
| 6638 |
+
{
|
| 6639 |
+
"epoch": 0.3013917773156295,
|
| 6640 |
+
"grad_norm": 423.9597473144531,
|
| 6641 |
+
"learning_rate": 0.000318970616462545,
|
| 6642 |
+
"loss": 8.2354,
|
| 6643 |
+
"step": 942
|
| 6644 |
+
},
|
| 6645 |
+
{
|
| 6646 |
+
"epoch": 0.3017117261238202,
|
| 6647 |
+
"grad_norm": 12609.2431640625,
|
| 6648 |
+
"learning_rate": 0.00031881023947767656,
|
| 6649 |
+
"loss": 8.5875,
|
| 6650 |
+
"step": 943
|
| 6651 |
+
},
|
| 6652 |
+
{
|
| 6653 |
+
"epoch": 0.3020316749320109,
|
| 6654 |
+
"grad_norm": 63.26814651489258,
|
| 6655 |
+
"learning_rate": 0.000318649747049226,
|
| 6656 |
+
"loss": 9.1278,
|
| 6657 |
+
"step": 944
|
| 6658 |
+
},
|
| 6659 |
+
{
|
| 6660 |
+
"epoch": 0.3023516237402016,
|
| 6661 |
+
"grad_norm": 1825.032470703125,
|
| 6662 |
+
"learning_rate": 0.0003184891393474696,
|
| 6663 |
+
"loss": 9.1317,
|
| 6664 |
+
"step": 945
|
| 6665 |
+
},
|
| 6666 |
+
{
|
| 6667 |
+
"epoch": 0.3026715725483923,
|
| 6668 |
+
"grad_norm": 76.4158706665039,
|
| 6669 |
+
"learning_rate": 0.0003183284165428061,
|
| 6670 |
+
"loss": 8.2171,
|
| 6671 |
+
"step": 946
|
| 6672 |
+
},
|
| 6673 |
+
{
|
| 6674 |
+
"epoch": 0.30299152135658297,
|
| 6675 |
+
"grad_norm": 14688.9970703125,
|
| 6676 |
+
"learning_rate": 0.00031816757880575664,
|
| 6677 |
+
"loss": 7.872,
|
| 6678 |
+
"step": 947
|
| 6679 |
+
},
|
| 6680 |
+
{
|
| 6681 |
+
"epoch": 0.3033114701647736,
|
| 6682 |
+
"grad_norm": 21862.908203125,
|
| 6683 |
+
"learning_rate": 0.0003180066263069637,
|
| 6684 |
+
"loss": 7.8133,
|
| 6685 |
+
"step": 948
|
| 6686 |
+
},
|
| 6687 |
+
{
|
| 6688 |
+
"epoch": 0.3036314189729643,
|
| 6689 |
+
"grad_norm": 132.2907257080078,
|
| 6690 |
+
"learning_rate": 0.000317845559217192,
|
| 6691 |
+
"loss": 7.9487,
|
| 6692 |
+
"step": 949
|
| 6693 |
+
},
|
| 6694 |
+
{
|
| 6695 |
+
"epoch": 0.303951367781155,
|
| 6696 |
+
"grad_norm": 125.90247344970703,
|
| 6697 |
+
"learning_rate": 0.00031768437770732775,
|
| 6698 |
+
"loss": 7.259,
|
| 6699 |
+
"step": 950
|
| 6700 |
+
},
|
| 6701 |
+
{
|
| 6702 |
+
"epoch": 0.3042713165893457,
|
| 6703 |
+
"grad_norm": 71.32536315917969,
|
| 6704 |
+
"learning_rate": 0.0003175230819483784,
|
| 6705 |
+
"loss": 6.9182,
|
| 6706 |
+
"step": 951
|
| 6707 |
+
},
|
| 6708 |
+
{
|
| 6709 |
+
"epoch": 0.3045912653975364,
|
| 6710 |
+
"grad_norm": 315.1258239746094,
|
| 6711 |
+
"learning_rate": 0.0003173616721114726,
|
| 6712 |
+
"loss": 6.9307,
|
| 6713 |
+
"step": 952
|
| 6714 |
+
},
|
| 6715 |
+
{
|
| 6716 |
+
"epoch": 0.3049112142057271,
|
| 6717 |
+
"grad_norm": 13234.1328125,
|
| 6718 |
+
"learning_rate": 0.0003172001483678603,
|
| 6719 |
+
"loss": 6.8386,
|
| 6720 |
+
"step": 953
|
| 6721 |
+
},
|
| 6722 |
+
{
|
| 6723 |
+
"epoch": 0.3052311630139178,
|
| 6724 |
+
"grad_norm": 37406.86328125,
|
| 6725 |
+
"learning_rate": 0.00031703851088891197,
|
| 6726 |
+
"loss": 7.2485,
|
| 6727 |
+
"step": 954
|
| 6728 |
+
},
|
| 6729 |
+
{
|
| 6730 |
+
"epoch": 0.3055511118221085,
|
| 6731 |
+
"grad_norm": 352988.6875,
|
| 6732 |
+
"learning_rate": 0.0003168767598461191,
|
| 6733 |
+
"loss": 6.8438,
|
| 6734 |
+
"step": 955
|
| 6735 |
+
},
|
| 6736 |
+
{
|
| 6737 |
+
"epoch": 0.3058710606302992,
|
| 6738 |
+
"grad_norm": 262.562744140625,
|
| 6739 |
+
"learning_rate": 0.00031671489541109326,
|
| 6740 |
+
"loss": 8.3405,
|
| 6741 |
+
"step": 956
|
| 6742 |
+
},
|
| 6743 |
+
{
|
| 6744 |
+
"epoch": 0.3061910094384898,
|
| 6745 |
+
"grad_norm": 1149.9610595703125,
|
| 6746 |
+
"learning_rate": 0.0003165529177555668,
|
| 6747 |
+
"loss": 8.3142,
|
| 6748 |
+
"step": 957
|
| 6749 |
+
},
|
| 6750 |
+
{
|
| 6751 |
+
"epoch": 0.3065109582466805,
|
| 6752 |
+
"grad_norm": 331.04718017578125,
|
| 6753 |
+
"learning_rate": 0.0003163908270513917,
|
| 6754 |
+
"loss": 8.1344,
|
| 6755 |
+
"step": 958
|
| 6756 |
+
},
|
| 6757 |
+
{
|
| 6758 |
+
"epoch": 0.3068309070548712,
|
| 6759 |
+
"grad_norm": 112.65670013427734,
|
| 6760 |
+
"learning_rate": 0.0003162286234705403,
|
| 6761 |
+
"loss": 7.2487,
|
| 6762 |
+
"step": 959
|
| 6763 |
+
},
|
| 6764 |
+
{
|
| 6765 |
+
"epoch": 0.3071508558630619,
|
| 6766 |
+
"grad_norm": 115.52394104003906,
|
| 6767 |
+
"learning_rate": 0.00031606630718510433,
|
| 6768 |
+
"loss": 7.4428,
|
| 6769 |
+
"step": 960
|
| 6770 |
+
},
|
| 6771 |
+
{
|
| 6772 |
+
"epoch": 0.3074708046712526,
|
| 6773 |
+
"grad_norm": 409.6588134765625,
|
| 6774 |
+
"learning_rate": 0.00031590387836729543,
|
| 6775 |
+
"loss": 6.6182,
|
| 6776 |
+
"step": 961
|
| 6777 |
+
},
|
| 6778 |
+
{
|
| 6779 |
+
"epoch": 0.3077907534794433,
|
| 6780 |
+
"grad_norm": 140.50241088867188,
|
| 6781 |
+
"learning_rate": 0.0003157413371894444,
|
| 6782 |
+
"loss": 7.4124,
|
| 6783 |
+
"step": 962
|
| 6784 |
+
},
|
| 6785 |
+
{
|
| 6786 |
+
"epoch": 0.308110702287634,
|
| 6787 |
+
"grad_norm": 126.857666015625,
|
| 6788 |
+
"learning_rate": 0.0003155786838240014,
|
| 6789 |
+
"loss": 6.8525,
|
| 6790 |
+
"step": 963
|
| 6791 |
+
},
|
| 6792 |
+
{
|
| 6793 |
+
"epoch": 0.3084306510958247,
|
| 6794 |
+
"grad_norm": 4386.052734375,
|
| 6795 |
+
"learning_rate": 0.0003154159184435355,
|
| 6796 |
+
"loss": 7.0282,
|
| 6797 |
+
"step": 964
|
| 6798 |
+
},
|
| 6799 |
+
{
|
| 6800 |
+
"epoch": 0.3087505999040154,
|
| 6801 |
+
"grad_norm": 79.79589080810547,
|
| 6802 |
+
"learning_rate": 0.00031525304122073476,
|
| 6803 |
+
"loss": 7.3552,
|
| 6804 |
+
"step": 965
|
| 6805 |
+
},
|
| 6806 |
+
{
|
| 6807 |
+
"epoch": 0.3090705487122061,
|
| 6808 |
+
"grad_norm": 207.43475341796875,
|
| 6809 |
+
"learning_rate": 0.0003150900523284056,
|
| 6810 |
+
"loss": 6.9154,
|
| 6811 |
+
"step": 966
|
| 6812 |
+
},
|
| 6813 |
+
{
|
| 6814 |
+
"epoch": 0.3093904975203967,
|
| 6815 |
+
"grad_norm": 44.77931594848633,
|
| 6816 |
+
"learning_rate": 0.00031492695193947325,
|
| 6817 |
+
"loss": 6.9198,
|
| 6818 |
+
"step": 967
|
| 6819 |
+
},
|
| 6820 |
+
{
|
| 6821 |
+
"epoch": 0.3097104463285874,
|
| 6822 |
+
"grad_norm": 103.20629119873047,
|
| 6823 |
+
"learning_rate": 0.0003147637402269812,
|
| 6824 |
+
"loss": 7.0039,
|
| 6825 |
+
"step": 968
|
| 6826 |
+
},
|
| 6827 |
+
{
|
| 6828 |
+
"epoch": 0.3100303951367781,
|
| 6829 |
+
"grad_norm": 31.550006866455078,
|
| 6830 |
+
"learning_rate": 0.0003146004173640908,
|
| 6831 |
+
"loss": 6.7761,
|
| 6832 |
+
"step": 969
|
| 6833 |
+
},
|
| 6834 |
+
{
|
| 6835 |
+
"epoch": 0.3103503439449688,
|
| 6836 |
+
"grad_norm": 14.499809265136719,
|
| 6837 |
+
"learning_rate": 0.0003144369835240815,
|
| 6838 |
+
"loss": 6.6403,
|
| 6839 |
+
"step": 970
|
| 6840 |
+
},
|
| 6841 |
+
{
|
| 6842 |
+
"epoch": 0.3106702927531595,
|
| 6843 |
+
"grad_norm": 47.8282470703125,
|
| 6844 |
+
"learning_rate": 0.00031427343888035047,
|
| 6845 |
+
"loss": 6.626,
|
| 6846 |
+
"step": 971
|
| 6847 |
+
},
|
| 6848 |
+
{
|
| 6849 |
+
"epoch": 0.3109902415613502,
|
| 6850 |
+
"grad_norm": 15.349066734313965,
|
| 6851 |
+
"learning_rate": 0.00031410978360641253,
|
| 6852 |
+
"loss": 7.1983,
|
| 6853 |
+
"step": 972
|
| 6854 |
+
},
|
| 6855 |
+
{
|
| 6856 |
+
"epoch": 0.3113101903695409,
|
| 6857 |
+
"grad_norm": 12.059926986694336,
|
| 6858 |
+
"learning_rate": 0.0003139460178758997,
|
| 6859 |
+
"loss": 6.6756,
|
| 6860 |
+
"step": 973
|
| 6861 |
+
},
|
| 6862 |
+
{
|
| 6863 |
+
"epoch": 0.3116301391777316,
|
| 6864 |
+
"grad_norm": 42.02625274658203,
|
| 6865 |
+
"learning_rate": 0.0003137821418625613,
|
| 6866 |
+
"loss": 6.7595,
|
| 6867 |
+
"step": 974
|
| 6868 |
+
},
|
| 6869 |
+
{
|
| 6870 |
+
"epoch": 0.3119500879859223,
|
| 6871 |
+
"grad_norm": 18.728851318359375,
|
| 6872 |
+
"learning_rate": 0.0003136181557402637,
|
| 6873 |
+
"loss": 6.7694,
|
| 6874 |
+
"step": 975
|
| 6875 |
+
},
|
| 6876 |
+
{
|
| 6877 |
+
"epoch": 0.3122700367941129,
|
| 6878 |
+
"grad_norm": 17.670698165893555,
|
| 6879 |
+
"learning_rate": 0.00031345405968298995,
|
| 6880 |
+
"loss": 6.9929,
|
| 6881 |
+
"step": 976
|
| 6882 |
+
},
|
| 6883 |
+
{
|
| 6884 |
+
"epoch": 0.3125899856023036,
|
| 6885 |
+
"grad_norm": 12.548014640808105,
|
| 6886 |
+
"learning_rate": 0.00031328985386483974,
|
| 6887 |
+
"loss": 6.7901,
|
| 6888 |
+
"step": 977
|
| 6889 |
+
},
|
| 6890 |
+
{
|
| 6891 |
+
"epoch": 0.3129099344104943,
|
| 6892 |
+
"grad_norm": 15.494758605957031,
|
| 6893 |
+
"learning_rate": 0.0003131255384600294,
|
| 6894 |
+
"loss": 6.6275,
|
| 6895 |
+
"step": 978
|
| 6896 |
+
},
|
| 6897 |
+
{
|
| 6898 |
+
"epoch": 0.313229883218685,
|
| 6899 |
+
"grad_norm": 22.92428207397461,
|
| 6900 |
+
"learning_rate": 0.00031296111364289143,
|
| 6901 |
+
"loss": 6.7544,
|
| 6902 |
+
"step": 979
|
| 6903 |
+
},
|
| 6904 |
+
{
|
| 6905 |
+
"epoch": 0.3135498320268757,
|
| 6906 |
+
"grad_norm": 8.640791893005371,
|
| 6907 |
+
"learning_rate": 0.00031279657958787434,
|
| 6908 |
+
"loss": 6.6713,
|
| 6909 |
+
"step": 980
|
| 6910 |
+
},
|
| 6911 |
+
{
|
| 6912 |
+
"epoch": 0.3138697808350664,
|
| 6913 |
+
"grad_norm": 11.602739334106445,
|
| 6914 |
+
"learning_rate": 0.00031263193646954255,
|
| 6915 |
+
"loss": 6.8735,
|
| 6916 |
+
"step": 981
|
| 6917 |
+
},
|
| 6918 |
+
{
|
| 6919 |
+
"epoch": 0.3141897296432571,
|
| 6920 |
+
"grad_norm": 7.338200569152832,
|
| 6921 |
+
"learning_rate": 0.0003124671844625763,
|
| 6922 |
+
"loss": 6.8,
|
| 6923 |
+
"step": 982
|
| 6924 |
+
},
|
| 6925 |
+
{
|
| 6926 |
+
"epoch": 0.3145096784514478,
|
| 6927 |
+
"grad_norm": 8.646294593811035,
|
| 6928 |
+
"learning_rate": 0.00031230232374177127,
|
| 6929 |
+
"loss": 6.4102,
|
| 6930 |
+
"step": 983
|
| 6931 |
+
},
|
| 6932 |
+
{
|
| 6933 |
+
"epoch": 0.3148296272596385,
|
| 6934 |
+
"grad_norm": 8.402074813842773,
|
| 6935 |
+
"learning_rate": 0.0003121373544820385,
|
| 6936 |
+
"loss": 6.7865,
|
| 6937 |
+
"step": 984
|
| 6938 |
+
},
|
| 6939 |
+
{
|
| 6940 |
+
"epoch": 0.3151495760678292,
|
| 6941 |
+
"grad_norm": 9.351253509521484,
|
| 6942 |
+
"learning_rate": 0.00031197227685840414,
|
| 6943 |
+
"loss": 6.6593,
|
| 6944 |
+
"step": 985
|
| 6945 |
+
},
|
| 6946 |
+
{
|
| 6947 |
+
"epoch": 0.3154695248760198,
|
| 6948 |
+
"grad_norm": 13.015589714050293,
|
| 6949 |
+
"learning_rate": 0.0003118070910460094,
|
| 6950 |
+
"loss": 6.7232,
|
| 6951 |
+
"step": 986
|
| 6952 |
+
},
|
| 6953 |
+
{
|
| 6954 |
+
"epoch": 0.3157894736842105,
|
| 6955 |
+
"grad_norm": 10.829553604125977,
|
| 6956 |
+
"learning_rate": 0.0003116417972201102,
|
| 6957 |
+
"loss": 6.6565,
|
| 6958 |
+
"step": 987
|
| 6959 |
+
},
|
| 6960 |
+
{
|
| 6961 |
+
"epoch": 0.3161094224924012,
|
| 6962 |
+
"grad_norm": 8.133262634277344,
|
| 6963 |
+
"learning_rate": 0.00031147639555607713,
|
| 6964 |
+
"loss": 6.8315,
|
| 6965 |
+
"step": 988
|
| 6966 |
+
},
|
| 6967 |
+
{
|
| 6968 |
+
"epoch": 0.3164293713005919,
|
| 6969 |
+
"grad_norm": 5.855452537536621,
|
| 6970 |
+
"learning_rate": 0.0003113108862293952,
|
| 6971 |
+
"loss": 6.7788,
|
| 6972 |
+
"step": 989
|
| 6973 |
+
},
|
| 6974 |
+
{
|
| 6975 |
+
"epoch": 0.3167493201087826,
|
| 6976 |
+
"grad_norm": 8.390470504760742,
|
| 6977 |
+
"learning_rate": 0.0003111452694156634,
|
| 6978 |
+
"loss": 6.6702,
|
| 6979 |
+
"step": 990
|
| 6980 |
+
},
|
| 6981 |
+
{
|
| 6982 |
+
"epoch": 0.3170692689169733,
|
| 6983 |
+
"grad_norm": 7.9397969245910645,
|
| 6984 |
+
"learning_rate": 0.0003109795452905952,
|
| 6985 |
+
"loss": 6.8752,
|
| 6986 |
+
"step": 991
|
| 6987 |
+
},
|
| 6988 |
+
{
|
| 6989 |
+
"epoch": 0.317389217725164,
|
| 6990 |
+
"grad_norm": 9.200600624084473,
|
| 6991 |
+
"learning_rate": 0.0003108137140300175,
|
| 6992 |
+
"loss": 6.7201,
|
| 6993 |
+
"step": 992
|
| 6994 |
+
},
|
| 6995 |
+
{
|
| 6996 |
+
"epoch": 0.3177091665333547,
|
| 6997 |
+
"grad_norm": 6.002403259277344,
|
| 6998 |
+
"learning_rate": 0.0003106477758098712,
|
| 6999 |
+
"loss": 6.5127,
|
| 7000 |
+
"step": 993
|
| 7001 |
+
},
|
| 7002 |
+
{
|
| 7003 |
+
"epoch": 0.3180291153415454,
|
| 7004 |
+
"grad_norm": 7.685515880584717,
|
| 7005 |
+
"learning_rate": 0.00031048173080621034,
|
| 7006 |
+
"loss": 6.6819,
|
| 7007 |
+
"step": 994
|
| 7008 |
+
},
|
| 7009 |
+
{
|
| 7010 |
+
"epoch": 0.318349064149736,
|
| 7011 |
+
"grad_norm": 6.026432037353516,
|
| 7012 |
+
"learning_rate": 0.0003103155791952026,
|
| 7013 |
+
"loss": 6.6265,
|
| 7014 |
+
"step": 995
|
| 7015 |
+
},
|
| 7016 |
+
{
|
| 7017 |
+
"epoch": 0.3186690129579267,
|
| 7018 |
+
"grad_norm": 7.417787551879883,
|
| 7019 |
+
"learning_rate": 0.0003101493211531285,
|
| 7020 |
+
"loss": 6.6986,
|
| 7021 |
+
"step": 996
|
| 7022 |
+
},
|
| 7023 |
+
{
|
| 7024 |
+
"epoch": 0.3189889617661174,
|
| 7025 |
+
"grad_norm": 12.300549507141113,
|
| 7026 |
+
"learning_rate": 0.00030998295685638166,
|
| 7027 |
+
"loss": 6.87,
|
| 7028 |
+
"step": 997
|
| 7029 |
+
},
|
| 7030 |
+
{
|
| 7031 |
+
"epoch": 0.3193089105743081,
|
| 7032 |
+
"grad_norm": 7.922842025756836,
|
| 7033 |
+
"learning_rate": 0.00030981648648146837,
|
| 7034 |
+
"loss": 6.8495,
|
| 7035 |
+
"step": 998
|
| 7036 |
+
},
|
| 7037 |
+
{
|
| 7038 |
+
"epoch": 0.3196288593824988,
|
| 7039 |
+
"grad_norm": 18.15106201171875,
|
| 7040 |
+
"learning_rate": 0.00030964991020500745,
|
| 7041 |
+
"loss": 6.6748,
|
| 7042 |
+
"step": 999
|
| 7043 |
+
},
|
| 7044 |
+
{
|
| 7045 |
+
"epoch": 0.3199488081906895,
|
| 7046 |
+
"grad_norm": 7.818212985992432,
|
| 7047 |
+
"learning_rate": 0.0003094832282037299,
|
| 7048 |
+
"loss": 6.6174,
|
| 7049 |
+
"step": 1000
|
| 7050 |
+
},
|
| 7051 |
+
{
|
| 7052 |
+
"epoch": 0.3199488081906895,
|
| 7053 |
+
"eval_loss": 3.358996868133545,
|
| 7054 |
+
"eval_runtime": 233.9815,
|
| 7055 |
+
"eval_samples_per_second": 5.624,
|
| 7056 |
+
"eval_steps_per_second": 1.406,
|
| 7057 |
+
"step": 1000
|
| 7058 |
}
|
| 7059 |
],
|
| 7060 |
"logging_steps": 1,
|
|
|
|
| 7069 |
"early_stopping_threshold": 0.0
|
| 7070 |
},
|
| 7071 |
"attributes": {
|
| 7072 |
+
"early_stopping_patience_counter": 2
|
| 7073 |
}
|
| 7074 |
},
|
| 7075 |
"TrainerControl": {
|
|
|
|
| 7083 |
"attributes": {}
|
| 7084 |
}
|
| 7085 |
},
|
| 7086 |
+
"total_flos": 9.201317197790577e+17,
|
| 7087 |
"train_batch_size": 4,
|
| 7088 |
"trial_name": null,
|
| 7089 |
"trial_params": null
|