Instructions to use nttx/b7d1970d-8f9a-4cae-8180-b2e8c45a7da2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use nttx/b7d1970d-8f9a-4cae-8180-b2e8c45a7da2 with PEFT:
from peft import PeftModel from transformers import AutoModelForCausalLM base_model = AutoModelForCausalLM.from_pretrained("katuni4ka/tiny-random-dbrx") model = PeftModel.from_pretrained(base_model, "nttx/b7d1970d-8f9a-4cae-8180-b2e8c45a7da2") - Notebooks
- Google Colab
- Kaggle
Training in progress, step 1200, 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 5752
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:46d727ef0af70eeabeb4773ef88c608d767a507ccc740f851c6336c45687168f
|
| 3 |
size 5752
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15814
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d01f324d9ce4269badbd34877b6daf5599db653feeb0bcf4a87c03590a2cab60
|
| 3 |
size 15814
|
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:e00cb835fc366aa9f5489a07d0e068f8672dc83206d40c9854181d9d7a9a805c
|
| 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 sha256:8b7384accc06cf321f008aa8494a569142e2fcc7cdc888583cfeddfa9f6e36fe
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 11.5,
|
| 3 |
"best_model_checkpoint": "miner_id_24/checkpoint-300",
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 300,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -6339,6 +6339,2114 @@
|
|
| 6339 |
"eval_samples_per_second": 149.243,
|
| 6340 |
"eval_steps_per_second": 74.621,
|
| 6341 |
"step": 900
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6342 |
}
|
| 6343 |
],
|
| 6344 |
"logging_steps": 1,
|
|
@@ -6353,7 +8461,7 @@
|
|
| 6353 |
"early_stopping_threshold": 0.0
|
| 6354 |
},
|
| 6355 |
"attributes": {
|
| 6356 |
-
"early_stopping_patience_counter":
|
| 6357 |
}
|
| 6358 |
},
|
| 6359 |
"TrainerControl": {
|
|
@@ -6362,12 +8470,12 @@
|
|
| 6362 |
"should_evaluate": false,
|
| 6363 |
"should_log": false,
|
| 6364 |
"should_save": true,
|
| 6365 |
-
"should_training_stop":
|
| 6366 |
},
|
| 6367 |
"attributes": {}
|
| 6368 |
}
|
| 6369 |
},
|
| 6370 |
-
"total_flos":
|
| 6371 |
"train_batch_size": 2,
|
| 6372 |
"trial_name": null,
|
| 6373 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 11.5,
|
| 3 |
"best_model_checkpoint": "miner_id_24/checkpoint-300",
|
| 4 |
+
"epoch": 0.16387286197125397,
|
| 5 |
"eval_steps": 300,
|
| 6 |
+
"global_step": 1200,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 6339 |
"eval_samples_per_second": 149.243,
|
| 6340 |
"eval_steps_per_second": 74.621,
|
| 6341 |
"step": 900
|
| 6342 |
+
},
|
| 6343 |
+
{
|
| 6344 |
+
"epoch": 0.12304120719674985,
|
| 6345 |
+
"grad_norm": 0.0009241271764039993,
|
| 6346 |
+
"learning_rate": 0.00015928737427682032,
|
| 6347 |
+
"loss": 46.0,
|
| 6348 |
+
"step": 901
|
| 6349 |
+
},
|
| 6350 |
+
{
|
| 6351 |
+
"epoch": 0.12317776791505923,
|
| 6352 |
+
"grad_norm": 0.0003521353646647185,
|
| 6353 |
+
"learning_rate": 0.00015920272918543257,
|
| 6354 |
+
"loss": 46.0,
|
| 6355 |
+
"step": 902
|
| 6356 |
+
},
|
| 6357 |
+
{
|
| 6358 |
+
"epoch": 0.12331432863336861,
|
| 6359 |
+
"grad_norm": 0.0004476907488424331,
|
| 6360 |
+
"learning_rate": 0.00015911801873599933,
|
| 6361 |
+
"loss": 46.0,
|
| 6362 |
+
"step": 903
|
| 6363 |
+
},
|
| 6364 |
+
{
|
| 6365 |
+
"epoch": 0.12345088935167799,
|
| 6366 |
+
"grad_norm": 0.0008625802001915872,
|
| 6367 |
+
"learning_rate": 0.00015903324302203836,
|
| 6368 |
+
"loss": 46.0,
|
| 6369 |
+
"step": 904
|
| 6370 |
+
},
|
| 6371 |
+
{
|
| 6372 |
+
"epoch": 0.12358745006998736,
|
| 6373 |
+
"grad_norm": 0.000347345310728997,
|
| 6374 |
+
"learning_rate": 0.00015894840213713952,
|
| 6375 |
+
"loss": 46.0,
|
| 6376 |
+
"step": 905
|
| 6377 |
+
},
|
| 6378 |
+
{
|
| 6379 |
+
"epoch": 0.12372401078829674,
|
| 6380 |
+
"grad_norm": 0.00042961168219335377,
|
| 6381 |
+
"learning_rate": 0.0001588634961749646,
|
| 6382 |
+
"loss": 46.0,
|
| 6383 |
+
"step": 906
|
| 6384 |
+
},
|
| 6385 |
+
{
|
| 6386 |
+
"epoch": 0.12386057150660612,
|
| 6387 |
+
"grad_norm": 0.0005479567334987223,
|
| 6388 |
+
"learning_rate": 0.00015877852522924732,
|
| 6389 |
+
"loss": 46.0,
|
| 6390 |
+
"step": 907
|
| 6391 |
+
},
|
| 6392 |
+
{
|
| 6393 |
+
"epoch": 0.1239971322249155,
|
| 6394 |
+
"grad_norm": 0.000546832219697535,
|
| 6395 |
+
"learning_rate": 0.00015869348939379302,
|
| 6396 |
+
"loss": 46.0,
|
| 6397 |
+
"step": 908
|
| 6398 |
+
},
|
| 6399 |
+
{
|
| 6400 |
+
"epoch": 0.12413369294322488,
|
| 6401 |
+
"grad_norm": 0.0005068538011983037,
|
| 6402 |
+
"learning_rate": 0.0001586083887624787,
|
| 6403 |
+
"loss": 46.0,
|
| 6404 |
+
"step": 909
|
| 6405 |
+
},
|
| 6406 |
+
{
|
| 6407 |
+
"epoch": 0.12427025366153426,
|
| 6408 |
+
"grad_norm": 0.0005834728945046663,
|
| 6409 |
+
"learning_rate": 0.00015852322342925295,
|
| 6410 |
+
"loss": 46.0,
|
| 6411 |
+
"step": 910
|
| 6412 |
+
},
|
| 6413 |
+
{
|
| 6414 |
+
"epoch": 0.12440681437984363,
|
| 6415 |
+
"grad_norm": 0.0004933670861646533,
|
| 6416 |
+
"learning_rate": 0.00015843799348813574,
|
| 6417 |
+
"loss": 46.0,
|
| 6418 |
+
"step": 911
|
| 6419 |
+
},
|
| 6420 |
+
{
|
| 6421 |
+
"epoch": 0.12454337509815301,
|
| 6422 |
+
"grad_norm": 0.0011595729738473892,
|
| 6423 |
+
"learning_rate": 0.0001583526990332184,
|
| 6424 |
+
"loss": 46.0,
|
| 6425 |
+
"step": 912
|
| 6426 |
+
},
|
| 6427 |
+
{
|
| 6428 |
+
"epoch": 0.12467993581646239,
|
| 6429 |
+
"grad_norm": 0.00029905018163844943,
|
| 6430 |
+
"learning_rate": 0.00015826734015866344,
|
| 6431 |
+
"loss": 46.0,
|
| 6432 |
+
"step": 913
|
| 6433 |
+
},
|
| 6434 |
+
{
|
| 6435 |
+
"epoch": 0.12481649653477177,
|
| 6436 |
+
"grad_norm": 0.000434244517236948,
|
| 6437 |
+
"learning_rate": 0.00015818191695870452,
|
| 6438 |
+
"loss": 46.0,
|
| 6439 |
+
"step": 914
|
| 6440 |
+
},
|
| 6441 |
+
{
|
| 6442 |
+
"epoch": 0.12495305725308115,
|
| 6443 |
+
"grad_norm": 0.0006040096050128341,
|
| 6444 |
+
"learning_rate": 0.00015809642952764632,
|
| 6445 |
+
"loss": 46.0,
|
| 6446 |
+
"step": 915
|
| 6447 |
+
},
|
| 6448 |
+
{
|
| 6449 |
+
"epoch": 0.12508961797139054,
|
| 6450 |
+
"grad_norm": 0.0007016469608061016,
|
| 6451 |
+
"learning_rate": 0.00015801087795986438,
|
| 6452 |
+
"loss": 46.0,
|
| 6453 |
+
"step": 916
|
| 6454 |
+
},
|
| 6455 |
+
{
|
| 6456 |
+
"epoch": 0.1252261786896999,
|
| 6457 |
+
"grad_norm": 0.0003513791016303003,
|
| 6458 |
+
"learning_rate": 0.0001579252623498051,
|
| 6459 |
+
"loss": 46.0,
|
| 6460 |
+
"step": 917
|
| 6461 |
+
},
|
| 6462 |
+
{
|
| 6463 |
+
"epoch": 0.1253627394080093,
|
| 6464 |
+
"grad_norm": 0.0003278540389146656,
|
| 6465 |
+
"learning_rate": 0.0001578395827919855,
|
| 6466 |
+
"loss": 46.0,
|
| 6467 |
+
"step": 918
|
| 6468 |
+
},
|
| 6469 |
+
{
|
| 6470 |
+
"epoch": 0.12549930012631866,
|
| 6471 |
+
"grad_norm": 0.000646027154289186,
|
| 6472 |
+
"learning_rate": 0.00015775383938099332,
|
| 6473 |
+
"loss": 46.0,
|
| 6474 |
+
"step": 919
|
| 6475 |
+
},
|
| 6476 |
+
{
|
| 6477 |
+
"epoch": 0.12563586084462805,
|
| 6478 |
+
"grad_norm": 0.0005723676295019686,
|
| 6479 |
+
"learning_rate": 0.00015766803221148673,
|
| 6480 |
+
"loss": 46.0,
|
| 6481 |
+
"step": 920
|
| 6482 |
+
},
|
| 6483 |
+
{
|
| 6484 |
+
"epoch": 0.12577242156293741,
|
| 6485 |
+
"grad_norm": 0.0015359300887212157,
|
| 6486 |
+
"learning_rate": 0.00015758216137819422,
|
| 6487 |
+
"loss": 46.0,
|
| 6488 |
+
"step": 921
|
| 6489 |
+
},
|
| 6490 |
+
{
|
| 6491 |
+
"epoch": 0.1259089822812468,
|
| 6492 |
+
"grad_norm": 0.00043970157275907695,
|
| 6493 |
+
"learning_rate": 0.0001574962269759147,
|
| 6494 |
+
"loss": 46.0,
|
| 6495 |
+
"step": 922
|
| 6496 |
+
},
|
| 6497 |
+
{
|
| 6498 |
+
"epoch": 0.12604554299955617,
|
| 6499 |
+
"grad_norm": 0.00028622214449569583,
|
| 6500 |
+
"learning_rate": 0.00015741022909951716,
|
| 6501 |
+
"loss": 46.0,
|
| 6502 |
+
"step": 923
|
| 6503 |
+
},
|
| 6504 |
+
{
|
| 6505 |
+
"epoch": 0.12618210371786556,
|
| 6506 |
+
"grad_norm": 0.00038283158210106194,
|
| 6507 |
+
"learning_rate": 0.00015732416784394065,
|
| 6508 |
+
"loss": 46.0,
|
| 6509 |
+
"step": 924
|
| 6510 |
+
},
|
| 6511 |
+
{
|
| 6512 |
+
"epoch": 0.12631866443617493,
|
| 6513 |
+
"grad_norm": 0.00039500248385593295,
|
| 6514 |
+
"learning_rate": 0.00015723804330419422,
|
| 6515 |
+
"loss": 46.0,
|
| 6516 |
+
"step": 925
|
| 6517 |
+
},
|
| 6518 |
+
{
|
| 6519 |
+
"epoch": 0.12645522515448432,
|
| 6520 |
+
"grad_norm": 0.0004930765135213733,
|
| 6521 |
+
"learning_rate": 0.00015715185557535689,
|
| 6522 |
+
"loss": 46.0,
|
| 6523 |
+
"step": 926
|
| 6524 |
+
},
|
| 6525 |
+
{
|
| 6526 |
+
"epoch": 0.12659178587279368,
|
| 6527 |
+
"grad_norm": 0.0005405626725405455,
|
| 6528 |
+
"learning_rate": 0.00015706560475257727,
|
| 6529 |
+
"loss": 46.0,
|
| 6530 |
+
"step": 927
|
| 6531 |
+
},
|
| 6532 |
+
{
|
| 6533 |
+
"epoch": 0.12672834659110307,
|
| 6534 |
+
"grad_norm": 0.00043443331378512084,
|
| 6535 |
+
"learning_rate": 0.00015697929093107365,
|
| 6536 |
+
"loss": 46.0,
|
| 6537 |
+
"step": 928
|
| 6538 |
+
},
|
| 6539 |
+
{
|
| 6540 |
+
"epoch": 0.12686490730941244,
|
| 6541 |
+
"grad_norm": 0.0005110527272336185,
|
| 6542 |
+
"learning_rate": 0.000156892914206134,
|
| 6543 |
+
"loss": 46.0,
|
| 6544 |
+
"step": 929
|
| 6545 |
+
},
|
| 6546 |
+
{
|
| 6547 |
+
"epoch": 0.12700146802772183,
|
| 6548 |
+
"grad_norm": 0.00031187915010377765,
|
| 6549 |
+
"learning_rate": 0.00015680647467311557,
|
| 6550 |
+
"loss": 46.0,
|
| 6551 |
+
"step": 930
|
| 6552 |
+
},
|
| 6553 |
+
{
|
| 6554 |
+
"epoch": 0.1271380287460312,
|
| 6555 |
+
"grad_norm": 0.0001981136156246066,
|
| 6556 |
+
"learning_rate": 0.00015671997242744511,
|
| 6557 |
+
"loss": 46.0,
|
| 6558 |
+
"step": 931
|
| 6559 |
+
},
|
| 6560 |
+
{
|
| 6561 |
+
"epoch": 0.1272745894643406,
|
| 6562 |
+
"grad_norm": 0.0026429896242916584,
|
| 6563 |
+
"learning_rate": 0.00015663340756461844,
|
| 6564 |
+
"loss": 46.0,
|
| 6565 |
+
"step": 932
|
| 6566 |
+
},
|
| 6567 |
+
{
|
| 6568 |
+
"epoch": 0.12741115018264995,
|
| 6569 |
+
"grad_norm": 0.0008942155982367694,
|
| 6570 |
+
"learning_rate": 0.0001565467801802006,
|
| 6571 |
+
"loss": 46.0,
|
| 6572 |
+
"step": 933
|
| 6573 |
+
},
|
| 6574 |
+
{
|
| 6575 |
+
"epoch": 0.12754771090095934,
|
| 6576 |
+
"grad_norm": 0.0006840588175691664,
|
| 6577 |
+
"learning_rate": 0.00015646009036982567,
|
| 6578 |
+
"loss": 46.0,
|
| 6579 |
+
"step": 934
|
| 6580 |
+
},
|
| 6581 |
+
{
|
| 6582 |
+
"epoch": 0.1276842716192687,
|
| 6583 |
+
"grad_norm": 0.0009611474233679473,
|
| 6584 |
+
"learning_rate": 0.00015637333822919656,
|
| 6585 |
+
"loss": 46.0,
|
| 6586 |
+
"step": 935
|
| 6587 |
+
},
|
| 6588 |
+
{
|
| 6589 |
+
"epoch": 0.1278208323375781,
|
| 6590 |
+
"grad_norm": 0.000530791119672358,
|
| 6591 |
+
"learning_rate": 0.00015628652385408508,
|
| 6592 |
+
"loss": 46.0,
|
| 6593 |
+
"step": 936
|
| 6594 |
+
},
|
| 6595 |
+
{
|
| 6596 |
+
"epoch": 0.12795739305588746,
|
| 6597 |
+
"grad_norm": 0.00041494445758871734,
|
| 6598 |
+
"learning_rate": 0.00015619964734033172,
|
| 6599 |
+
"loss": 46.0,
|
| 6600 |
+
"step": 937
|
| 6601 |
+
},
|
| 6602 |
+
{
|
| 6603 |
+
"epoch": 0.12809395377419686,
|
| 6604 |
+
"grad_norm": 0.0008938443497754633,
|
| 6605 |
+
"learning_rate": 0.00015611270878384552,
|
| 6606 |
+
"loss": 46.0,
|
| 6607 |
+
"step": 938
|
| 6608 |
+
},
|
| 6609 |
+
{
|
| 6610 |
+
"epoch": 0.12823051449250622,
|
| 6611 |
+
"grad_norm": 0.000739588460419327,
|
| 6612 |
+
"learning_rate": 0.00015602570828060407,
|
| 6613 |
+
"loss": 46.0,
|
| 6614 |
+
"step": 939
|
| 6615 |
+
},
|
| 6616 |
+
{
|
| 6617 |
+
"epoch": 0.1283670752108156,
|
| 6618 |
+
"grad_norm": 0.0025961471255868673,
|
| 6619 |
+
"learning_rate": 0.00015593864592665333,
|
| 6620 |
+
"loss": 46.0,
|
| 6621 |
+
"step": 940
|
| 6622 |
+
},
|
| 6623 |
+
{
|
| 6624 |
+
"epoch": 0.12850363592912498,
|
| 6625 |
+
"grad_norm": 0.00049801473505795,
|
| 6626 |
+
"learning_rate": 0.00015585152181810753,
|
| 6627 |
+
"loss": 46.0,
|
| 6628 |
+
"step": 941
|
| 6629 |
+
},
|
| 6630 |
+
{
|
| 6631 |
+
"epoch": 0.12864019664743437,
|
| 6632 |
+
"grad_norm": 0.000621246756054461,
|
| 6633 |
+
"learning_rate": 0.00015576433605114912,
|
| 6634 |
+
"loss": 46.0,
|
| 6635 |
+
"step": 942
|
| 6636 |
+
},
|
| 6637 |
+
{
|
| 6638 |
+
"epoch": 0.12877675736574373,
|
| 6639 |
+
"grad_norm": 0.0007328742649406195,
|
| 6640 |
+
"learning_rate": 0.00015567708872202854,
|
| 6641 |
+
"loss": 46.0,
|
| 6642 |
+
"step": 943
|
| 6643 |
+
},
|
| 6644 |
+
{
|
| 6645 |
+
"epoch": 0.12891331808405312,
|
| 6646 |
+
"grad_norm": 0.005625641904771328,
|
| 6647 |
+
"learning_rate": 0.00015558977992706426,
|
| 6648 |
+
"loss": 46.0,
|
| 6649 |
+
"step": 944
|
| 6650 |
+
},
|
| 6651 |
+
{
|
| 6652 |
+
"epoch": 0.1290498788023625,
|
| 6653 |
+
"grad_norm": 0.0005238280282355845,
|
| 6654 |
+
"learning_rate": 0.00015550240976264253,
|
| 6655 |
+
"loss": 46.0,
|
| 6656 |
+
"step": 945
|
| 6657 |
+
},
|
| 6658 |
+
{
|
| 6659 |
+
"epoch": 0.12918643952067188,
|
| 6660 |
+
"grad_norm": 0.001820914214476943,
|
| 6661 |
+
"learning_rate": 0.0001554149783252175,
|
| 6662 |
+
"loss": 46.0,
|
| 6663 |
+
"step": 946
|
| 6664 |
+
},
|
| 6665 |
+
{
|
| 6666 |
+
"epoch": 0.12932300023898124,
|
| 6667 |
+
"grad_norm": 0.0006223174277693033,
|
| 6668 |
+
"learning_rate": 0.0001553274857113108,
|
| 6669 |
+
"loss": 46.0,
|
| 6670 |
+
"step": 947
|
| 6671 |
+
},
|
| 6672 |
+
{
|
| 6673 |
+
"epoch": 0.12945956095729064,
|
| 6674 |
+
"grad_norm": 0.0015029089991003275,
|
| 6675 |
+
"learning_rate": 0.00015523993201751167,
|
| 6676 |
+
"loss": 46.0,
|
| 6677 |
+
"step": 948
|
| 6678 |
+
},
|
| 6679 |
+
{
|
| 6680 |
+
"epoch": 0.1295961216756,
|
| 6681 |
+
"grad_norm": 0.0012488181237131357,
|
| 6682 |
+
"learning_rate": 0.00015515231734047677,
|
| 6683 |
+
"loss": 46.0,
|
| 6684 |
+
"step": 949
|
| 6685 |
+
},
|
| 6686 |
+
{
|
| 6687 |
+
"epoch": 0.1297326823939094,
|
| 6688 |
+
"grad_norm": 0.0007277615368366241,
|
| 6689 |
+
"learning_rate": 0.0001550646417769301,
|
| 6690 |
+
"loss": 46.0,
|
| 6691 |
+
"step": 950
|
| 6692 |
+
},
|
| 6693 |
+
{
|
| 6694 |
+
"epoch": 0.12986924311221876,
|
| 6695 |
+
"grad_norm": 0.0024758039508014917,
|
| 6696 |
+
"learning_rate": 0.0001549769054236629,
|
| 6697 |
+
"loss": 46.0,
|
| 6698 |
+
"step": 951
|
| 6699 |
+
},
|
| 6700 |
+
{
|
| 6701 |
+
"epoch": 0.13000580383052815,
|
| 6702 |
+
"grad_norm": 0.00043562057544477284,
|
| 6703 |
+
"learning_rate": 0.00015488910837753342,
|
| 6704 |
+
"loss": 46.0,
|
| 6705 |
+
"step": 952
|
| 6706 |
+
},
|
| 6707 |
+
{
|
| 6708 |
+
"epoch": 0.1301423645488375,
|
| 6709 |
+
"grad_norm": 0.0006010486977174878,
|
| 6710 |
+
"learning_rate": 0.00015480125073546704,
|
| 6711 |
+
"loss": 46.0,
|
| 6712 |
+
"step": 953
|
| 6713 |
+
},
|
| 6714 |
+
{
|
| 6715 |
+
"epoch": 0.1302789252671469,
|
| 6716 |
+
"grad_norm": 0.00039981160080060363,
|
| 6717 |
+
"learning_rate": 0.0001547133325944559,
|
| 6718 |
+
"loss": 46.0,
|
| 6719 |
+
"step": 954
|
| 6720 |
+
},
|
| 6721 |
+
{
|
| 6722 |
+
"epoch": 0.13041548598545627,
|
| 6723 |
+
"grad_norm": 0.00056524045066908,
|
| 6724 |
+
"learning_rate": 0.00015462535405155902,
|
| 6725 |
+
"loss": 46.0,
|
| 6726 |
+
"step": 955
|
| 6727 |
+
},
|
| 6728 |
+
{
|
| 6729 |
+
"epoch": 0.13055204670376566,
|
| 6730 |
+
"grad_norm": 0.00019664541468955576,
|
| 6731 |
+
"learning_rate": 0.00015453731520390215,
|
| 6732 |
+
"loss": 46.0,
|
| 6733 |
+
"step": 956
|
| 6734 |
+
},
|
| 6735 |
+
{
|
| 6736 |
+
"epoch": 0.13068860742207505,
|
| 6737 |
+
"grad_norm": 0.000747493002563715,
|
| 6738 |
+
"learning_rate": 0.0001544492161486775,
|
| 6739 |
+
"loss": 46.0,
|
| 6740 |
+
"step": 957
|
| 6741 |
+
},
|
| 6742 |
+
{
|
| 6743 |
+
"epoch": 0.13082516814038442,
|
| 6744 |
+
"grad_norm": 0.00021618347091134638,
|
| 6745 |
+
"learning_rate": 0.00015436105698314384,
|
| 6746 |
+
"loss": 46.0,
|
| 6747 |
+
"step": 958
|
| 6748 |
+
},
|
| 6749 |
+
{
|
| 6750 |
+
"epoch": 0.1309617288586938,
|
| 6751 |
+
"grad_norm": 0.0005658991285599768,
|
| 6752 |
+
"learning_rate": 0.0001542728378046262,
|
| 6753 |
+
"loss": 46.0,
|
| 6754 |
+
"step": 959
|
| 6755 |
+
},
|
| 6756 |
+
{
|
| 6757 |
+
"epoch": 0.13109828957700317,
|
| 6758 |
+
"grad_norm": 0.0010320099536329508,
|
| 6759 |
+
"learning_rate": 0.00015418455871051592,
|
| 6760 |
+
"loss": 46.0,
|
| 6761 |
+
"step": 960
|
| 6762 |
+
},
|
| 6763 |
+
{
|
| 6764 |
+
"epoch": 0.13123485029531257,
|
| 6765 |
+
"grad_norm": 0.00037752182106487453,
|
| 6766 |
+
"learning_rate": 0.00015409621979827048,
|
| 6767 |
+
"loss": 46.0,
|
| 6768 |
+
"step": 961
|
| 6769 |
+
},
|
| 6770 |
+
{
|
| 6771 |
+
"epoch": 0.13137141101362193,
|
| 6772 |
+
"grad_norm": 0.0003615982714109123,
|
| 6773 |
+
"learning_rate": 0.0001540078211654135,
|
| 6774 |
+
"loss": 46.0,
|
| 6775 |
+
"step": 962
|
| 6776 |
+
},
|
| 6777 |
+
{
|
| 6778 |
+
"epoch": 0.13150797173193132,
|
| 6779 |
+
"grad_norm": 0.00041104850242845714,
|
| 6780 |
+
"learning_rate": 0.0001539193629095343,
|
| 6781 |
+
"loss": 46.0,
|
| 6782 |
+
"step": 963
|
| 6783 |
+
},
|
| 6784 |
+
{
|
| 6785 |
+
"epoch": 0.1316445324502407,
|
| 6786 |
+
"grad_norm": 0.0005330987041816115,
|
| 6787 |
+
"learning_rate": 0.00015383084512828824,
|
| 6788 |
+
"loss": 46.0,
|
| 6789 |
+
"step": 964
|
| 6790 |
+
},
|
| 6791 |
+
{
|
| 6792 |
+
"epoch": 0.13178109316855008,
|
| 6793 |
+
"grad_norm": 0.0003800500126089901,
|
| 6794 |
+
"learning_rate": 0.00015374226791939628,
|
| 6795 |
+
"loss": 46.0,
|
| 6796 |
+
"step": 965
|
| 6797 |
+
},
|
| 6798 |
+
{
|
| 6799 |
+
"epoch": 0.13191765388685944,
|
| 6800 |
+
"grad_norm": 0.0006545698852278292,
|
| 6801 |
+
"learning_rate": 0.000153653631380645,
|
| 6802 |
+
"loss": 46.0,
|
| 6803 |
+
"step": 966
|
| 6804 |
+
},
|
| 6805 |
+
{
|
| 6806 |
+
"epoch": 0.13205421460516883,
|
| 6807 |
+
"grad_norm": 0.0020852810703217983,
|
| 6808 |
+
"learning_rate": 0.0001535649356098865,
|
| 6809 |
+
"loss": 46.0,
|
| 6810 |
+
"step": 967
|
| 6811 |
+
},
|
| 6812 |
+
{
|
| 6813 |
+
"epoch": 0.1321907753234782,
|
| 6814 |
+
"grad_norm": 0.00031325622694566846,
|
| 6815 |
+
"learning_rate": 0.00015347618070503827,
|
| 6816 |
+
"loss": 46.0,
|
| 6817 |
+
"step": 968
|
| 6818 |
+
},
|
| 6819 |
+
{
|
| 6820 |
+
"epoch": 0.1323273360417876,
|
| 6821 |
+
"grad_norm": 0.00019234443607274443,
|
| 6822 |
+
"learning_rate": 0.0001533873667640831,
|
| 6823 |
+
"loss": 46.0,
|
| 6824 |
+
"step": 969
|
| 6825 |
+
},
|
| 6826 |
+
{
|
| 6827 |
+
"epoch": 0.13246389676009696,
|
| 6828 |
+
"grad_norm": 0.00038069483707658947,
|
| 6829 |
+
"learning_rate": 0.00015329849388506886,
|
| 6830 |
+
"loss": 46.0,
|
| 6831 |
+
"step": 970
|
| 6832 |
+
},
|
| 6833 |
+
{
|
| 6834 |
+
"epoch": 0.13260045747840635,
|
| 6835 |
+
"grad_norm": 0.00045511999633163214,
|
| 6836 |
+
"learning_rate": 0.00015320956216610866,
|
| 6837 |
+
"loss": 46.0,
|
| 6838 |
+
"step": 971
|
| 6839 |
+
},
|
| 6840 |
+
{
|
| 6841 |
+
"epoch": 0.1327370181967157,
|
| 6842 |
+
"grad_norm": 0.0005886334110982716,
|
| 6843 |
+
"learning_rate": 0.00015312057170538035,
|
| 6844 |
+
"loss": 46.0,
|
| 6845 |
+
"step": 972
|
| 6846 |
+
},
|
| 6847 |
+
{
|
| 6848 |
+
"epoch": 0.1328735789150251,
|
| 6849 |
+
"grad_norm": 0.0009905985789373517,
|
| 6850 |
+
"learning_rate": 0.00015303152260112682,
|
| 6851 |
+
"loss": 46.0,
|
| 6852 |
+
"step": 973
|
| 6853 |
+
},
|
| 6854 |
+
{
|
| 6855 |
+
"epoch": 0.13301013963333447,
|
| 6856 |
+
"grad_norm": 0.0010108448332175612,
|
| 6857 |
+
"learning_rate": 0.00015294241495165557,
|
| 6858 |
+
"loss": 46.0,
|
| 6859 |
+
"step": 974
|
| 6860 |
+
},
|
| 6861 |
+
{
|
| 6862 |
+
"epoch": 0.13314670035164386,
|
| 6863 |
+
"grad_norm": 0.0005389642901718616,
|
| 6864 |
+
"learning_rate": 0.00015285324885533884,
|
| 6865 |
+
"loss": 46.0,
|
| 6866 |
+
"step": 975
|
| 6867 |
+
},
|
| 6868 |
+
{
|
| 6869 |
+
"epoch": 0.13328326106995322,
|
| 6870 |
+
"grad_norm": 0.000410493987146765,
|
| 6871 |
+
"learning_rate": 0.0001527640244106133,
|
| 6872 |
+
"loss": 46.0,
|
| 6873 |
+
"step": 976
|
| 6874 |
+
},
|
| 6875 |
+
{
|
| 6876 |
+
"epoch": 0.13341982178826262,
|
| 6877 |
+
"grad_norm": 0.001756677869707346,
|
| 6878 |
+
"learning_rate": 0.00015267474171598005,
|
| 6879 |
+
"loss": 46.0,
|
| 6880 |
+
"step": 977
|
| 6881 |
+
},
|
| 6882 |
+
{
|
| 6883 |
+
"epoch": 0.13355638250657198,
|
| 6884 |
+
"grad_norm": 0.0003254815237596631,
|
| 6885 |
+
"learning_rate": 0.0001525854008700046,
|
| 6886 |
+
"loss": 46.0,
|
| 6887 |
+
"step": 978
|
| 6888 |
+
},
|
| 6889 |
+
{
|
| 6890 |
+
"epoch": 0.13369294322488137,
|
| 6891 |
+
"grad_norm": 0.0003471802920103073,
|
| 6892 |
+
"learning_rate": 0.00015249600197131651,
|
| 6893 |
+
"loss": 46.0,
|
| 6894 |
+
"step": 979
|
| 6895 |
+
},
|
| 6896 |
+
{
|
| 6897 |
+
"epoch": 0.13382950394319074,
|
| 6898 |
+
"grad_norm": 0.0015930512454360723,
|
| 6899 |
+
"learning_rate": 0.0001524065451186095,
|
| 6900 |
+
"loss": 46.0,
|
| 6901 |
+
"step": 980
|
| 6902 |
+
},
|
| 6903 |
+
{
|
| 6904 |
+
"epoch": 0.13396606466150013,
|
| 6905 |
+
"grad_norm": 0.00028746266616508365,
|
| 6906 |
+
"learning_rate": 0.0001523170304106413,
|
| 6907 |
+
"loss": 46.0,
|
| 6908 |
+
"step": 981
|
| 6909 |
+
},
|
| 6910 |
+
{
|
| 6911 |
+
"epoch": 0.1341026253798095,
|
| 6912 |
+
"grad_norm": 0.0005899532698094845,
|
| 6913 |
+
"learning_rate": 0.0001522274579462334,
|
| 6914 |
+
"loss": 46.0,
|
| 6915 |
+
"step": 982
|
| 6916 |
+
},
|
| 6917 |
+
{
|
| 6918 |
+
"epoch": 0.13423918609811888,
|
| 6919 |
+
"grad_norm": 0.0005316737224347889,
|
| 6920 |
+
"learning_rate": 0.00015213782782427123,
|
| 6921 |
+
"loss": 46.0,
|
| 6922 |
+
"step": 983
|
| 6923 |
+
},
|
| 6924 |
+
{
|
| 6925 |
+
"epoch": 0.13437574681642825,
|
| 6926 |
+
"grad_norm": 0.00045934764784760773,
|
| 6927 |
+
"learning_rate": 0.00015204814014370372,
|
| 6928 |
+
"loss": 46.0,
|
| 6929 |
+
"step": 984
|
| 6930 |
+
},
|
| 6931 |
+
{
|
| 6932 |
+
"epoch": 0.13451230753473764,
|
| 6933 |
+
"grad_norm": 0.001285345759242773,
|
| 6934 |
+
"learning_rate": 0.00015195839500354335,
|
| 6935 |
+
"loss": 46.0,
|
| 6936 |
+
"step": 985
|
| 6937 |
+
},
|
| 6938 |
+
{
|
| 6939 |
+
"epoch": 0.134648868253047,
|
| 6940 |
+
"grad_norm": 0.0005873920163139701,
|
| 6941 |
+
"learning_rate": 0.00015186859250286615,
|
| 6942 |
+
"loss": 46.0,
|
| 6943 |
+
"step": 986
|
| 6944 |
+
},
|
| 6945 |
+
{
|
| 6946 |
+
"epoch": 0.1347854289713564,
|
| 6947 |
+
"grad_norm": 0.0011875568889081478,
|
| 6948 |
+
"learning_rate": 0.00015177873274081137,
|
| 6949 |
+
"loss": 46.0,
|
| 6950 |
+
"step": 987
|
| 6951 |
+
},
|
| 6952 |
+
{
|
| 6953 |
+
"epoch": 0.13492198968966576,
|
| 6954 |
+
"grad_norm": 0.001056193490512669,
|
| 6955 |
+
"learning_rate": 0.00015168881581658147,
|
| 6956 |
+
"loss": 46.0,
|
| 6957 |
+
"step": 988
|
| 6958 |
+
},
|
| 6959 |
+
{
|
| 6960 |
+
"epoch": 0.13505855040797515,
|
| 6961 |
+
"grad_norm": 0.0009340514661744237,
|
| 6962 |
+
"learning_rate": 0.00015159884182944211,
|
| 6963 |
+
"loss": 46.0,
|
| 6964 |
+
"step": 989
|
| 6965 |
+
},
|
| 6966 |
+
{
|
| 6967 |
+
"epoch": 0.13519511112628452,
|
| 6968 |
+
"grad_norm": 0.0011555576929822564,
|
| 6969 |
+
"learning_rate": 0.00015150881087872185,
|
| 6970 |
+
"loss": 46.0,
|
| 6971 |
+
"step": 990
|
| 6972 |
+
},
|
| 6973 |
+
{
|
| 6974 |
+
"epoch": 0.1353316718445939,
|
| 6975 |
+
"grad_norm": 0.001084683695808053,
|
| 6976 |
+
"learning_rate": 0.00015141872306381215,
|
| 6977 |
+
"loss": 46.0,
|
| 6978 |
+
"step": 991
|
| 6979 |
+
},
|
| 6980 |
+
{
|
| 6981 |
+
"epoch": 0.13546823256290327,
|
| 6982 |
+
"grad_norm": 0.000519106222782284,
|
| 6983 |
+
"learning_rate": 0.00015132857848416733,
|
| 6984 |
+
"loss": 46.0,
|
| 6985 |
+
"step": 992
|
| 6986 |
+
},
|
| 6987 |
+
{
|
| 6988 |
+
"epoch": 0.13560479328121267,
|
| 6989 |
+
"grad_norm": 0.001280359923839569,
|
| 6990 |
+
"learning_rate": 0.00015123837723930424,
|
| 6991 |
+
"loss": 46.0,
|
| 6992 |
+
"step": 993
|
| 6993 |
+
},
|
| 6994 |
+
{
|
| 6995 |
+
"epoch": 0.13574135399952203,
|
| 6996 |
+
"grad_norm": 0.0007028987165540457,
|
| 6997 |
+
"learning_rate": 0.00015114811942880242,
|
| 6998 |
+
"loss": 46.0,
|
| 6999 |
+
"step": 994
|
| 7000 |
+
},
|
| 7001 |
+
{
|
| 7002 |
+
"epoch": 0.13587791471783142,
|
| 7003 |
+
"grad_norm": 0.0005714019644074142,
|
| 7004 |
+
"learning_rate": 0.00015105780515230376,
|
| 7005 |
+
"loss": 46.0,
|
| 7006 |
+
"step": 995
|
| 7007 |
+
},
|
| 7008 |
+
{
|
| 7009 |
+
"epoch": 0.13601447543614079,
|
| 7010 |
+
"grad_norm": 0.0009716853965073824,
|
| 7011 |
+
"learning_rate": 0.00015096743450951258,
|
| 7012 |
+
"loss": 46.0,
|
| 7013 |
+
"step": 996
|
| 7014 |
+
},
|
| 7015 |
+
{
|
| 7016 |
+
"epoch": 0.13615103615445018,
|
| 7017 |
+
"grad_norm": 0.0006497084395959973,
|
| 7018 |
+
"learning_rate": 0.00015087700760019532,
|
| 7019 |
+
"loss": 46.0,
|
| 7020 |
+
"step": 997
|
| 7021 |
+
},
|
| 7022 |
+
{
|
| 7023 |
+
"epoch": 0.13628759687275954,
|
| 7024 |
+
"grad_norm": 0.0013521420769393444,
|
| 7025 |
+
"learning_rate": 0.00015078652452418063,
|
| 7026 |
+
"loss": 46.0,
|
| 7027 |
+
"step": 998
|
| 7028 |
+
},
|
| 7029 |
+
{
|
| 7030 |
+
"epoch": 0.13642415759106893,
|
| 7031 |
+
"grad_norm": 0.0008505037403665483,
|
| 7032 |
+
"learning_rate": 0.00015069598538135906,
|
| 7033 |
+
"loss": 46.0,
|
| 7034 |
+
"step": 999
|
| 7035 |
+
},
|
| 7036 |
+
{
|
| 7037 |
+
"epoch": 0.1365607183093783,
|
| 7038 |
+
"grad_norm": 0.0011916455114260316,
|
| 7039 |
+
"learning_rate": 0.00015060539027168316,
|
| 7040 |
+
"loss": 46.0,
|
| 7041 |
+
"step": 1000
|
| 7042 |
+
},
|
| 7043 |
+
{
|
| 7044 |
+
"epoch": 0.1366972790276877,
|
| 7045 |
+
"grad_norm": 0.0005008620209991932,
|
| 7046 |
+
"learning_rate": 0.00015051473929516722,
|
| 7047 |
+
"loss": 46.0,
|
| 7048 |
+
"step": 1001
|
| 7049 |
+
},
|
| 7050 |
+
{
|
| 7051 |
+
"epoch": 0.13683383974599705,
|
| 7052 |
+
"grad_norm": 0.0006068138754926622,
|
| 7053 |
+
"learning_rate": 0.00015042403255188723,
|
| 7054 |
+
"loss": 46.0,
|
| 7055 |
+
"step": 1002
|
| 7056 |
+
},
|
| 7057 |
+
{
|
| 7058 |
+
"epoch": 0.13697040046430645,
|
| 7059 |
+
"grad_norm": 0.0003954765561502427,
|
| 7060 |
+
"learning_rate": 0.00015033327014198075,
|
| 7061 |
+
"loss": 46.0,
|
| 7062 |
+
"step": 1003
|
| 7063 |
+
},
|
| 7064 |
+
{
|
| 7065 |
+
"epoch": 0.1371069611826158,
|
| 7066 |
+
"grad_norm": 0.0005776135949417949,
|
| 7067 |
+
"learning_rate": 0.00015024245216564667,
|
| 7068 |
+
"loss": 46.0,
|
| 7069 |
+
"step": 1004
|
| 7070 |
+
},
|
| 7071 |
+
{
|
| 7072 |
+
"epoch": 0.1372435219009252,
|
| 7073 |
+
"grad_norm": 0.0008297267486341298,
|
| 7074 |
+
"learning_rate": 0.00015015157872314542,
|
| 7075 |
+
"loss": 46.0,
|
| 7076 |
+
"step": 1005
|
| 7077 |
+
},
|
| 7078 |
+
{
|
| 7079 |
+
"epoch": 0.13738008261923457,
|
| 7080 |
+
"grad_norm": 0.000555426231585443,
|
| 7081 |
+
"learning_rate": 0.00015006064991479853,
|
| 7082 |
+
"loss": 46.0,
|
| 7083 |
+
"step": 1006
|
| 7084 |
+
},
|
| 7085 |
+
{
|
| 7086 |
+
"epoch": 0.13751664333754396,
|
| 7087 |
+
"grad_norm": 0.0002457842347212136,
|
| 7088 |
+
"learning_rate": 0.0001499696658409887,
|
| 7089 |
+
"loss": 46.0,
|
| 7090 |
+
"step": 1007
|
| 7091 |
+
},
|
| 7092 |
+
{
|
| 7093 |
+
"epoch": 0.13765320405585332,
|
| 7094 |
+
"grad_norm": 0.0005408066790550947,
|
| 7095 |
+
"learning_rate": 0.00014987862660215966,
|
| 7096 |
+
"loss": 46.0,
|
| 7097 |
+
"step": 1008
|
| 7098 |
+
},
|
| 7099 |
+
{
|
| 7100 |
+
"epoch": 0.13778976477416272,
|
| 7101 |
+
"grad_norm": 0.0010539703071117401,
|
| 7102 |
+
"learning_rate": 0.00014978753229881594,
|
| 7103 |
+
"loss": 46.0,
|
| 7104 |
+
"step": 1009
|
| 7105 |
+
},
|
| 7106 |
+
{
|
| 7107 |
+
"epoch": 0.13792632549247208,
|
| 7108 |
+
"grad_norm": 0.00045947683975100517,
|
| 7109 |
+
"learning_rate": 0.00014969638303152295,
|
| 7110 |
+
"loss": 46.0,
|
| 7111 |
+
"step": 1010
|
| 7112 |
+
},
|
| 7113 |
+
{
|
| 7114 |
+
"epoch": 0.13806288621078147,
|
| 7115 |
+
"grad_norm": 0.0004952670424245298,
|
| 7116 |
+
"learning_rate": 0.0001496051789009068,
|
| 7117 |
+
"loss": 46.0,
|
| 7118 |
+
"step": 1011
|
| 7119 |
+
},
|
| 7120 |
+
{
|
| 7121 |
+
"epoch": 0.13819944692909084,
|
| 7122 |
+
"grad_norm": 0.0006827415782026947,
|
| 7123 |
+
"learning_rate": 0.00014951392000765411,
|
| 7124 |
+
"loss": 46.0,
|
| 7125 |
+
"step": 1012
|
| 7126 |
+
},
|
| 7127 |
+
{
|
| 7128 |
+
"epoch": 0.13833600764740023,
|
| 7129 |
+
"grad_norm": 0.000486463715787977,
|
| 7130 |
+
"learning_rate": 0.000149422606452512,
|
| 7131 |
+
"loss": 46.0,
|
| 7132 |
+
"step": 1013
|
| 7133 |
+
},
|
| 7134 |
+
{
|
| 7135 |
+
"epoch": 0.1384725683657096,
|
| 7136 |
+
"grad_norm": 0.0002832711033988744,
|
| 7137 |
+
"learning_rate": 0.00014933123833628785,
|
| 7138 |
+
"loss": 46.0,
|
| 7139 |
+
"step": 1014
|
| 7140 |
+
},
|
| 7141 |
+
{
|
| 7142 |
+
"epoch": 0.13860912908401898,
|
| 7143 |
+
"grad_norm": 0.0004940008511766791,
|
| 7144 |
+
"learning_rate": 0.00014923981575984936,
|
| 7145 |
+
"loss": 46.0,
|
| 7146 |
+
"step": 1015
|
| 7147 |
+
},
|
| 7148 |
+
{
|
| 7149 |
+
"epoch": 0.13874568980232835,
|
| 7150 |
+
"grad_norm": 0.0008655837154947221,
|
| 7151 |
+
"learning_rate": 0.00014914833882412435,
|
| 7152 |
+
"loss": 46.0,
|
| 7153 |
+
"step": 1016
|
| 7154 |
+
},
|
| 7155 |
+
{
|
| 7156 |
+
"epoch": 0.13888225052063774,
|
| 7157 |
+
"grad_norm": 0.0007375786663033068,
|
| 7158 |
+
"learning_rate": 0.00014905680763010058,
|
| 7159 |
+
"loss": 46.0,
|
| 7160 |
+
"step": 1017
|
| 7161 |
+
},
|
| 7162 |
+
{
|
| 7163 |
+
"epoch": 0.1390188112389471,
|
| 7164 |
+
"grad_norm": 0.0006843106239102781,
|
| 7165 |
+
"learning_rate": 0.00014896522227882578,
|
| 7166 |
+
"loss": 46.0,
|
| 7167 |
+
"step": 1018
|
| 7168 |
+
},
|
| 7169 |
+
{
|
| 7170 |
+
"epoch": 0.1391553719572565,
|
| 7171 |
+
"grad_norm": 0.001130104181356728,
|
| 7172 |
+
"learning_rate": 0.00014887358287140744,
|
| 7173 |
+
"loss": 46.0,
|
| 7174 |
+
"step": 1019
|
| 7175 |
+
},
|
| 7176 |
+
{
|
| 7177 |
+
"epoch": 0.13929193267556586,
|
| 7178 |
+
"grad_norm": 0.0008909485186450183,
|
| 7179 |
+
"learning_rate": 0.00014878188950901276,
|
| 7180 |
+
"loss": 46.0,
|
| 7181 |
+
"step": 1020
|
| 7182 |
+
},
|
| 7183 |
+
{
|
| 7184 |
+
"epoch": 0.13942849339387525,
|
| 7185 |
+
"grad_norm": 0.005642786156386137,
|
| 7186 |
+
"learning_rate": 0.0001486901422928684,
|
| 7187 |
+
"loss": 46.0,
|
| 7188 |
+
"step": 1021
|
| 7189 |
+
},
|
| 7190 |
+
{
|
| 7191 |
+
"epoch": 0.13956505411218462,
|
| 7192 |
+
"grad_norm": 0.0005347135593183339,
|
| 7193 |
+
"learning_rate": 0.0001485983413242606,
|
| 7194 |
+
"loss": 46.0,
|
| 7195 |
+
"step": 1022
|
| 7196 |
+
},
|
| 7197 |
+
{
|
| 7198 |
+
"epoch": 0.139701614830494,
|
| 7199 |
+
"grad_norm": 0.000356485164957121,
|
| 7200 |
+
"learning_rate": 0.00014850648670453493,
|
| 7201 |
+
"loss": 46.0,
|
| 7202 |
+
"step": 1023
|
| 7203 |
+
},
|
| 7204 |
+
{
|
| 7205 |
+
"epoch": 0.13983817554880337,
|
| 7206 |
+
"grad_norm": 0.0006373700452968478,
|
| 7207 |
+
"learning_rate": 0.00014841457853509606,
|
| 7208 |
+
"loss": 46.0,
|
| 7209 |
+
"step": 1024
|
| 7210 |
+
},
|
| 7211 |
+
{
|
| 7212 |
+
"epoch": 0.13997473626711276,
|
| 7213 |
+
"grad_norm": 0.00020091190526727587,
|
| 7214 |
+
"learning_rate": 0.0001483226169174079,
|
| 7215 |
+
"loss": 46.0,
|
| 7216 |
+
"step": 1025
|
| 7217 |
+
},
|
| 7218 |
+
{
|
| 7219 |
+
"epoch": 0.14011129698542216,
|
| 7220 |
+
"grad_norm": 0.00042303564259782434,
|
| 7221 |
+
"learning_rate": 0.00014823060195299337,
|
| 7222 |
+
"loss": 46.0,
|
| 7223 |
+
"step": 1026
|
| 7224 |
+
},
|
| 7225 |
+
{
|
| 7226 |
+
"epoch": 0.14024785770373152,
|
| 7227 |
+
"grad_norm": 0.0004543966497294605,
|
| 7228 |
+
"learning_rate": 0.00014813853374343419,
|
| 7229 |
+
"loss": 46.0,
|
| 7230 |
+
"step": 1027
|
| 7231 |
+
},
|
| 7232 |
+
{
|
| 7233 |
+
"epoch": 0.1403844184220409,
|
| 7234 |
+
"grad_norm": 0.00042528280755504966,
|
| 7235 |
+
"learning_rate": 0.00014804641239037097,
|
| 7236 |
+
"loss": 46.0,
|
| 7237 |
+
"step": 1028
|
| 7238 |
+
},
|
| 7239 |
+
{
|
| 7240 |
+
"epoch": 0.14052097914035028,
|
| 7241 |
+
"grad_norm": 0.0005864663980901241,
|
| 7242 |
+
"learning_rate": 0.00014795423799550284,
|
| 7243 |
+
"loss": 46.0,
|
| 7244 |
+
"step": 1029
|
| 7245 |
+
},
|
| 7246 |
+
{
|
| 7247 |
+
"epoch": 0.14065753985865967,
|
| 7248 |
+
"grad_norm": 0.0005385400145314634,
|
| 7249 |
+
"learning_rate": 0.00014786201066058766,
|
| 7250 |
+
"loss": 46.0,
|
| 7251 |
+
"step": 1030
|
| 7252 |
+
},
|
| 7253 |
+
{
|
| 7254 |
+
"epoch": 0.14079410057696903,
|
| 7255 |
+
"grad_norm": 0.0011653905967250466,
|
| 7256 |
+
"learning_rate": 0.00014776973048744165,
|
| 7257 |
+
"loss": 46.0,
|
| 7258 |
+
"step": 1031
|
| 7259 |
+
},
|
| 7260 |
+
{
|
| 7261 |
+
"epoch": 0.14093066129527843,
|
| 7262 |
+
"grad_norm": 0.0004561395035125315,
|
| 7263 |
+
"learning_rate": 0.0001476773975779393,
|
| 7264 |
+
"loss": 46.0,
|
| 7265 |
+
"step": 1032
|
| 7266 |
+
},
|
| 7267 |
+
{
|
| 7268 |
+
"epoch": 0.1410672220135878,
|
| 7269 |
+
"grad_norm": 0.000643297506030649,
|
| 7270 |
+
"learning_rate": 0.00014758501203401348,
|
| 7271 |
+
"loss": 46.0,
|
| 7272 |
+
"step": 1033
|
| 7273 |
+
},
|
| 7274 |
+
{
|
| 7275 |
+
"epoch": 0.14120378273189718,
|
| 7276 |
+
"grad_norm": 0.0006550021353177726,
|
| 7277 |
+
"learning_rate": 0.00014749257395765502,
|
| 7278 |
+
"loss": 46.0,
|
| 7279 |
+
"step": 1034
|
| 7280 |
+
},
|
| 7281 |
+
{
|
| 7282 |
+
"epoch": 0.14134034345020655,
|
| 7283 |
+
"grad_norm": 0.0007623559795320034,
|
| 7284 |
+
"learning_rate": 0.0001474000834509128,
|
| 7285 |
+
"loss": 46.0,
|
| 7286 |
+
"step": 1035
|
| 7287 |
+
},
|
| 7288 |
+
{
|
| 7289 |
+
"epoch": 0.14147690416851594,
|
| 7290 |
+
"grad_norm": 0.000623580242972821,
|
| 7291 |
+
"learning_rate": 0.00014730754061589355,
|
| 7292 |
+
"loss": 46.0,
|
| 7293 |
+
"step": 1036
|
| 7294 |
+
},
|
| 7295 |
+
{
|
| 7296 |
+
"epoch": 0.1416134648868253,
|
| 7297 |
+
"grad_norm": 0.0010572531027719378,
|
| 7298 |
+
"learning_rate": 0.00014721494555476188,
|
| 7299 |
+
"loss": 46.0,
|
| 7300 |
+
"step": 1037
|
| 7301 |
+
},
|
| 7302 |
+
{
|
| 7303 |
+
"epoch": 0.1417500256051347,
|
| 7304 |
+
"grad_norm": 0.0011201991001144052,
|
| 7305 |
+
"learning_rate": 0.00014712229836973988,
|
| 7306 |
+
"loss": 46.0,
|
| 7307 |
+
"step": 1038
|
| 7308 |
+
},
|
| 7309 |
+
{
|
| 7310 |
+
"epoch": 0.14188658632344406,
|
| 7311 |
+
"grad_norm": 0.001960804220288992,
|
| 7312 |
+
"learning_rate": 0.00014702959916310736,
|
| 7313 |
+
"loss": 46.0,
|
| 7314 |
+
"step": 1039
|
| 7315 |
+
},
|
| 7316 |
+
{
|
| 7317 |
+
"epoch": 0.14202314704175345,
|
| 7318 |
+
"grad_norm": 0.0050703976303339005,
|
| 7319 |
+
"learning_rate": 0.00014693684803720138,
|
| 7320 |
+
"loss": 46.0,
|
| 7321 |
+
"step": 1040
|
| 7322 |
+
},
|
| 7323 |
+
{
|
| 7324 |
+
"epoch": 0.14215970776006281,
|
| 7325 |
+
"grad_norm": 0.00040476518915966153,
|
| 7326 |
+
"learning_rate": 0.0001468440450944165,
|
| 7327 |
+
"loss": 46.0,
|
| 7328 |
+
"step": 1041
|
| 7329 |
+
},
|
| 7330 |
+
{
|
| 7331 |
+
"epoch": 0.1422962684783722,
|
| 7332 |
+
"grad_norm": 0.0007858510361984372,
|
| 7333 |
+
"learning_rate": 0.00014675119043720437,
|
| 7334 |
+
"loss": 46.0,
|
| 7335 |
+
"step": 1042
|
| 7336 |
+
},
|
| 7337 |
+
{
|
| 7338 |
+
"epoch": 0.14243282919668157,
|
| 7339 |
+
"grad_norm": 0.0007758094579912722,
|
| 7340 |
+
"learning_rate": 0.0001466582841680737,
|
| 7341 |
+
"loss": 46.0,
|
| 7342 |
+
"step": 1043
|
| 7343 |
+
},
|
| 7344 |
+
{
|
| 7345 |
+
"epoch": 0.14256938991499096,
|
| 7346 |
+
"grad_norm": 0.0008653284166939557,
|
| 7347 |
+
"learning_rate": 0.00014656532638959035,
|
| 7348 |
+
"loss": 46.0,
|
| 7349 |
+
"step": 1044
|
| 7350 |
+
},
|
| 7351 |
+
{
|
| 7352 |
+
"epoch": 0.14270595063330033,
|
| 7353 |
+
"grad_norm": 0.0004421341873239726,
|
| 7354 |
+
"learning_rate": 0.00014647231720437686,
|
| 7355 |
+
"loss": 46.0,
|
| 7356 |
+
"step": 1045
|
| 7357 |
+
},
|
| 7358 |
+
{
|
| 7359 |
+
"epoch": 0.14284251135160972,
|
| 7360 |
+
"grad_norm": 0.0008486118167638779,
|
| 7361 |
+
"learning_rate": 0.0001463792567151126,
|
| 7362 |
+
"loss": 46.0,
|
| 7363 |
+
"step": 1046
|
| 7364 |
+
},
|
| 7365 |
+
{
|
| 7366 |
+
"epoch": 0.14297907206991908,
|
| 7367 |
+
"grad_norm": 0.000525649928022176,
|
| 7368 |
+
"learning_rate": 0.0001462861450245336,
|
| 7369 |
+
"loss": 46.0,
|
| 7370 |
+
"step": 1047
|
| 7371 |
+
},
|
| 7372 |
+
{
|
| 7373 |
+
"epoch": 0.14311563278822848,
|
| 7374 |
+
"grad_norm": 0.0017683632904663682,
|
| 7375 |
+
"learning_rate": 0.00014619298223543235,
|
| 7376 |
+
"loss": 46.0,
|
| 7377 |
+
"step": 1048
|
| 7378 |
+
},
|
| 7379 |
+
{
|
| 7380 |
+
"epoch": 0.14325219350653784,
|
| 7381 |
+
"grad_norm": 0.0012217290932312608,
|
| 7382 |
+
"learning_rate": 0.00014609976845065783,
|
| 7383 |
+
"loss": 46.0,
|
| 7384 |
+
"step": 1049
|
| 7385 |
+
},
|
| 7386 |
+
{
|
| 7387 |
+
"epoch": 0.14338875422484723,
|
| 7388 |
+
"grad_norm": 0.0011137262918055058,
|
| 7389 |
+
"learning_rate": 0.00014600650377311522,
|
| 7390 |
+
"loss": 46.0,
|
| 7391 |
+
"step": 1050
|
| 7392 |
+
},
|
| 7393 |
+
{
|
| 7394 |
+
"epoch": 0.1435253149431566,
|
| 7395 |
+
"grad_norm": 0.0003707687428686768,
|
| 7396 |
+
"learning_rate": 0.00014591318830576598,
|
| 7397 |
+
"loss": 46.0,
|
| 7398 |
+
"step": 1051
|
| 7399 |
+
},
|
| 7400 |
+
{
|
| 7401 |
+
"epoch": 0.143661875661466,
|
| 7402 |
+
"grad_norm": 0.000414914742577821,
|
| 7403 |
+
"learning_rate": 0.0001458198221516276,
|
| 7404 |
+
"loss": 46.0,
|
| 7405 |
+
"step": 1052
|
| 7406 |
+
},
|
| 7407 |
+
{
|
| 7408 |
+
"epoch": 0.14379843637977535,
|
| 7409 |
+
"grad_norm": 0.0008973225485533476,
|
| 7410 |
+
"learning_rate": 0.0001457264054137735,
|
| 7411 |
+
"loss": 46.0,
|
| 7412 |
+
"step": 1053
|
| 7413 |
+
},
|
| 7414 |
+
{
|
| 7415 |
+
"epoch": 0.14393499709808474,
|
| 7416 |
+
"grad_norm": 0.00040008637006394565,
|
| 7417 |
+
"learning_rate": 0.000145632938195333,
|
| 7418 |
+
"loss": 46.0,
|
| 7419 |
+
"step": 1054
|
| 7420 |
+
},
|
| 7421 |
+
{
|
| 7422 |
+
"epoch": 0.1440715578163941,
|
| 7423 |
+
"grad_norm": 0.0005400644731707871,
|
| 7424 |
+
"learning_rate": 0.0001455394205994911,
|
| 7425 |
+
"loss": 46.0,
|
| 7426 |
+
"step": 1055
|
| 7427 |
+
},
|
| 7428 |
+
{
|
| 7429 |
+
"epoch": 0.1442081185347035,
|
| 7430 |
+
"grad_norm": 0.00028061779448762536,
|
| 7431 |
+
"learning_rate": 0.00014544585272948843,
|
| 7432 |
+
"loss": 46.0,
|
| 7433 |
+
"step": 1056
|
| 7434 |
+
},
|
| 7435 |
+
{
|
| 7436 |
+
"epoch": 0.14434467925301286,
|
| 7437 |
+
"grad_norm": 0.0007635858491994441,
|
| 7438 |
+
"learning_rate": 0.00014535223468862114,
|
| 7439 |
+
"loss": 46.0,
|
| 7440 |
+
"step": 1057
|
| 7441 |
+
},
|
| 7442 |
+
{
|
| 7443 |
+
"epoch": 0.14448123997132226,
|
| 7444 |
+
"grad_norm": 0.0004812279948964715,
|
| 7445 |
+
"learning_rate": 0.00014525856658024076,
|
| 7446 |
+
"loss": 46.0,
|
| 7447 |
+
"step": 1058
|
| 7448 |
+
},
|
| 7449 |
+
{
|
| 7450 |
+
"epoch": 0.14461780068963162,
|
| 7451 |
+
"grad_norm": 0.0004690811620093882,
|
| 7452 |
+
"learning_rate": 0.00014516484850775406,
|
| 7453 |
+
"loss": 46.0,
|
| 7454 |
+
"step": 1059
|
| 7455 |
+
},
|
| 7456 |
+
{
|
| 7457 |
+
"epoch": 0.144754361407941,
|
| 7458 |
+
"grad_norm": 0.0008514091605320573,
|
| 7459 |
+
"learning_rate": 0.00014507108057462296,
|
| 7460 |
+
"loss": 46.0,
|
| 7461 |
+
"step": 1060
|
| 7462 |
+
},
|
| 7463 |
+
{
|
| 7464 |
+
"epoch": 0.14489092212625038,
|
| 7465 |
+
"grad_norm": 0.00031304339063353837,
|
| 7466 |
+
"learning_rate": 0.00014497726288436458,
|
| 7467 |
+
"loss": 46.0,
|
| 7468 |
+
"step": 1061
|
| 7469 |
+
},
|
| 7470 |
+
{
|
| 7471 |
+
"epoch": 0.14502748284455977,
|
| 7472 |
+
"grad_norm": 0.00015942241589073092,
|
| 7473 |
+
"learning_rate": 0.00014488339554055073,
|
| 7474 |
+
"loss": 46.0,
|
| 7475 |
+
"step": 1062
|
| 7476 |
+
},
|
| 7477 |
+
{
|
| 7478 |
+
"epoch": 0.14516404356286913,
|
| 7479 |
+
"grad_norm": 0.0012250308645889163,
|
| 7480 |
+
"learning_rate": 0.0001447894786468082,
|
| 7481 |
+
"loss": 46.0,
|
| 7482 |
+
"step": 1063
|
| 7483 |
+
},
|
| 7484 |
+
{
|
| 7485 |
+
"epoch": 0.14530060428117852,
|
| 7486 |
+
"grad_norm": 0.00036729895509779453,
|
| 7487 |
+
"learning_rate": 0.00014469551230681844,
|
| 7488 |
+
"loss": 46.0,
|
| 7489 |
+
"step": 1064
|
| 7490 |
+
},
|
| 7491 |
+
{
|
| 7492 |
+
"epoch": 0.1454371649994879,
|
| 7493 |
+
"grad_norm": 0.000257661915384233,
|
| 7494 |
+
"learning_rate": 0.00014460149662431747,
|
| 7495 |
+
"loss": 46.0,
|
| 7496 |
+
"step": 1065
|
| 7497 |
+
},
|
| 7498 |
+
{
|
| 7499 |
+
"epoch": 0.14557372571779728,
|
| 7500 |
+
"grad_norm": 0.000458430964499712,
|
| 7501 |
+
"learning_rate": 0.00014450743170309584,
|
| 7502 |
+
"loss": 46.0,
|
| 7503 |
+
"step": 1066
|
| 7504 |
+
},
|
| 7505 |
+
{
|
| 7506 |
+
"epoch": 0.14571028643610665,
|
| 7507 |
+
"grad_norm": 0.0004429294203873724,
|
| 7508 |
+
"learning_rate": 0.00014441331764699836,
|
| 7509 |
+
"loss": 46.0,
|
| 7510 |
+
"step": 1067
|
| 7511 |
+
},
|
| 7512 |
+
{
|
| 7513 |
+
"epoch": 0.14584684715441604,
|
| 7514 |
+
"grad_norm": 0.0005923425196669996,
|
| 7515 |
+
"learning_rate": 0.00014431915455992414,
|
| 7516 |
+
"loss": 46.0,
|
| 7517 |
+
"step": 1068
|
| 7518 |
+
},
|
| 7519 |
+
{
|
| 7520 |
+
"epoch": 0.1459834078727254,
|
| 7521 |
+
"grad_norm": 0.0003316183283459395,
|
| 7522 |
+
"learning_rate": 0.00014422494254582647,
|
| 7523 |
+
"loss": 46.0,
|
| 7524 |
+
"step": 1069
|
| 7525 |
+
},
|
| 7526 |
+
{
|
| 7527 |
+
"epoch": 0.1461199685910348,
|
| 7528 |
+
"grad_norm": 0.0004024969239253551,
|
| 7529 |
+
"learning_rate": 0.0001441306817087125,
|
| 7530 |
+
"loss": 46.0,
|
| 7531 |
+
"step": 1070
|
| 7532 |
+
},
|
| 7533 |
+
{
|
| 7534 |
+
"epoch": 0.14625652930934416,
|
| 7535 |
+
"grad_norm": 0.00029238680144771934,
|
| 7536 |
+
"learning_rate": 0.00014403637215264353,
|
| 7537 |
+
"loss": 46.0,
|
| 7538 |
+
"step": 1071
|
| 7539 |
+
},
|
| 7540 |
+
{
|
| 7541 |
+
"epoch": 0.14639309002765355,
|
| 7542 |
+
"grad_norm": 0.00044409476686269045,
|
| 7543 |
+
"learning_rate": 0.00014394201398173437,
|
| 7544 |
+
"loss": 46.0,
|
| 7545 |
+
"step": 1072
|
| 7546 |
+
},
|
| 7547 |
+
{
|
| 7548 |
+
"epoch": 0.1465296507459629,
|
| 7549 |
+
"grad_norm": 0.000767083081882447,
|
| 7550 |
+
"learning_rate": 0.00014384760730015364,
|
| 7551 |
+
"loss": 46.0,
|
| 7552 |
+
"step": 1073
|
| 7553 |
+
},
|
| 7554 |
+
{
|
| 7555 |
+
"epoch": 0.1466662114642723,
|
| 7556 |
+
"grad_norm": 0.00029706236091442406,
|
| 7557 |
+
"learning_rate": 0.00014375315221212357,
|
| 7558 |
+
"loss": 46.0,
|
| 7559 |
+
"step": 1074
|
| 7560 |
+
},
|
| 7561 |
+
{
|
| 7562 |
+
"epoch": 0.14680277218258167,
|
| 7563 |
+
"grad_norm": 0.0004188843595329672,
|
| 7564 |
+
"learning_rate": 0.00014365864882191968,
|
| 7565 |
+
"loss": 46.0,
|
| 7566 |
+
"step": 1075
|
| 7567 |
+
},
|
| 7568 |
+
{
|
| 7569 |
+
"epoch": 0.14693933290089106,
|
| 7570 |
+
"grad_norm": 0.0005889105377718806,
|
| 7571 |
+
"learning_rate": 0.0001435640972338709,
|
| 7572 |
+
"loss": 46.0,
|
| 7573 |
+
"step": 1076
|
| 7574 |
+
},
|
| 7575 |
+
{
|
| 7576 |
+
"epoch": 0.14707589361920043,
|
| 7577 |
+
"grad_norm": 0.00072791165439412,
|
| 7578 |
+
"learning_rate": 0.00014346949755235944,
|
| 7579 |
+
"loss": 46.0,
|
| 7580 |
+
"step": 1077
|
| 7581 |
+
},
|
| 7582 |
+
{
|
| 7583 |
+
"epoch": 0.14721245433750982,
|
| 7584 |
+
"grad_norm": 0.0006283425609581172,
|
| 7585 |
+
"learning_rate": 0.00014337484988182042,
|
| 7586 |
+
"loss": 46.0,
|
| 7587 |
+
"step": 1078
|
| 7588 |
+
},
|
| 7589 |
+
{
|
| 7590 |
+
"epoch": 0.14734901505581918,
|
| 7591 |
+
"grad_norm": 0.000486049015307799,
|
| 7592 |
+
"learning_rate": 0.00014328015432674214,
|
| 7593 |
+
"loss": 46.0,
|
| 7594 |
+
"step": 1079
|
| 7595 |
+
},
|
| 7596 |
+
{
|
| 7597 |
+
"epoch": 0.14748557577412857,
|
| 7598 |
+
"grad_norm": 0.0007576828938908875,
|
| 7599 |
+
"learning_rate": 0.00014318541099166555,
|
| 7600 |
+
"loss": 46.0,
|
| 7601 |
+
"step": 1080
|
| 7602 |
+
},
|
| 7603 |
+
{
|
| 7604 |
+
"epoch": 0.14762213649243794,
|
| 7605 |
+
"grad_norm": 0.0010862492490559816,
|
| 7606 |
+
"learning_rate": 0.00014309061998118454,
|
| 7607 |
+
"loss": 46.0,
|
| 7608 |
+
"step": 1081
|
| 7609 |
+
},
|
| 7610 |
+
{
|
| 7611 |
+
"epoch": 0.14775869721074733,
|
| 7612 |
+
"grad_norm": 0.0005167250637896359,
|
| 7613 |
+
"learning_rate": 0.00014299578139994557,
|
| 7614 |
+
"loss": 46.0,
|
| 7615 |
+
"step": 1082
|
| 7616 |
+
},
|
| 7617 |
+
{
|
| 7618 |
+
"epoch": 0.1478952579290567,
|
| 7619 |
+
"grad_norm": 0.002369736786931753,
|
| 7620 |
+
"learning_rate": 0.00014290089535264755,
|
| 7621 |
+
"loss": 46.0,
|
| 7622 |
+
"step": 1083
|
| 7623 |
+
},
|
| 7624 |
+
{
|
| 7625 |
+
"epoch": 0.1480318186473661,
|
| 7626 |
+
"grad_norm": 0.0003688117431011051,
|
| 7627 |
+
"learning_rate": 0.0001428059619440419,
|
| 7628 |
+
"loss": 46.0,
|
| 7629 |
+
"step": 1084
|
| 7630 |
+
},
|
| 7631 |
+
{
|
| 7632 |
+
"epoch": 0.14816837936567545,
|
| 7633 |
+
"grad_norm": 0.0003458712890278548,
|
| 7634 |
+
"learning_rate": 0.00014271098127893218,
|
| 7635 |
+
"loss": 46.0,
|
| 7636 |
+
"step": 1085
|
| 7637 |
+
},
|
| 7638 |
+
{
|
| 7639 |
+
"epoch": 0.14830494008398484,
|
| 7640 |
+
"grad_norm": 0.0013860361650586128,
|
| 7641 |
+
"learning_rate": 0.0001426159534621743,
|
| 7642 |
+
"loss": 46.0,
|
| 7643 |
+
"step": 1086
|
| 7644 |
+
},
|
| 7645 |
+
{
|
| 7646 |
+
"epoch": 0.1484415008022942,
|
| 7647 |
+
"grad_norm": 0.001035936875268817,
|
| 7648 |
+
"learning_rate": 0.00014252087859867608,
|
| 7649 |
+
"loss": 46.0,
|
| 7650 |
+
"step": 1087
|
| 7651 |
+
},
|
| 7652 |
+
{
|
| 7653 |
+
"epoch": 0.1485780615206036,
|
| 7654 |
+
"grad_norm": 0.00632870476692915,
|
| 7655 |
+
"learning_rate": 0.00014242575679339738,
|
| 7656 |
+
"loss": 46.0,
|
| 7657 |
+
"step": 1088
|
| 7658 |
+
},
|
| 7659 |
+
{
|
| 7660 |
+
"epoch": 0.14871462223891296,
|
| 7661 |
+
"grad_norm": 0.0006592991994693875,
|
| 7662 |
+
"learning_rate": 0.00014233058815134978,
|
| 7663 |
+
"loss": 46.0,
|
| 7664 |
+
"step": 1089
|
| 7665 |
+
},
|
| 7666 |
+
{
|
| 7667 |
+
"epoch": 0.14885118295722236,
|
| 7668 |
+
"grad_norm": 0.0028478745371103287,
|
| 7669 |
+
"learning_rate": 0.00014223537277759666,
|
| 7670 |
+
"loss": 46.0,
|
| 7671 |
+
"step": 1090
|
| 7672 |
+
},
|
| 7673 |
+
{
|
| 7674 |
+
"epoch": 0.14898774367553172,
|
| 7675 |
+
"grad_norm": 0.0005855032941326499,
|
| 7676 |
+
"learning_rate": 0.00014214011077725292,
|
| 7677 |
+
"loss": 46.0,
|
| 7678 |
+
"step": 1091
|
| 7679 |
+
},
|
| 7680 |
+
{
|
| 7681 |
+
"epoch": 0.1491243043938411,
|
| 7682 |
+
"grad_norm": 0.0023942850530147552,
|
| 7683 |
+
"learning_rate": 0.00014204480225548494,
|
| 7684 |
+
"loss": 46.0,
|
| 7685 |
+
"step": 1092
|
| 7686 |
+
},
|
| 7687 |
+
{
|
| 7688 |
+
"epoch": 0.1492608651121505,
|
| 7689 |
+
"grad_norm": 0.0013539772480726242,
|
| 7690 |
+
"learning_rate": 0.00014194944731751058,
|
| 7691 |
+
"loss": 46.0,
|
| 7692 |
+
"step": 1093
|
| 7693 |
+
},
|
| 7694 |
+
{
|
| 7695 |
+
"epoch": 0.14939742583045987,
|
| 7696 |
+
"grad_norm": 0.0009589577093720436,
|
| 7697 |
+
"learning_rate": 0.00014185404606859877,
|
| 7698 |
+
"loss": 46.0,
|
| 7699 |
+
"step": 1094
|
| 7700 |
+
},
|
| 7701 |
+
{
|
| 7702 |
+
"epoch": 0.14953398654876926,
|
| 7703 |
+
"grad_norm": 0.0003180347557645291,
|
| 7704 |
+
"learning_rate": 0.00014175859861406966,
|
| 7705 |
+
"loss": 46.0,
|
| 7706 |
+
"step": 1095
|
| 7707 |
+
},
|
| 7708 |
+
{
|
| 7709 |
+
"epoch": 0.14967054726707862,
|
| 7710 |
+
"grad_norm": 0.0014342650538310409,
|
| 7711 |
+
"learning_rate": 0.00014166310505929434,
|
| 7712 |
+
"loss": 46.0,
|
| 7713 |
+
"step": 1096
|
| 7714 |
+
},
|
| 7715 |
+
{
|
| 7716 |
+
"epoch": 0.14980710798538802,
|
| 7717 |
+
"grad_norm": 0.0006643772940151393,
|
| 7718 |
+
"learning_rate": 0.00014156756550969492,
|
| 7719 |
+
"loss": 46.0,
|
| 7720 |
+
"step": 1097
|
| 7721 |
+
},
|
| 7722 |
+
{
|
| 7723 |
+
"epoch": 0.14994366870369738,
|
| 7724 |
+
"grad_norm": 0.001409722724929452,
|
| 7725 |
+
"learning_rate": 0.00014147198007074415,
|
| 7726 |
+
"loss": 46.0,
|
| 7727 |
+
"step": 1098
|
| 7728 |
+
},
|
| 7729 |
+
{
|
| 7730 |
+
"epoch": 0.15008022942200677,
|
| 7731 |
+
"grad_norm": 0.0008714981959201396,
|
| 7732 |
+
"learning_rate": 0.00014137634884796557,
|
| 7733 |
+
"loss": 46.0,
|
| 7734 |
+
"step": 1099
|
| 7735 |
+
},
|
| 7736 |
+
{
|
| 7737 |
+
"epoch": 0.15021679014031614,
|
| 7738 |
+
"grad_norm": 0.00160513399168849,
|
| 7739 |
+
"learning_rate": 0.00014128067194693316,
|
| 7740 |
+
"loss": 46.0,
|
| 7741 |
+
"step": 1100
|
| 7742 |
+
},
|
| 7743 |
+
{
|
| 7744 |
+
"epoch": 0.15035335085862553,
|
| 7745 |
+
"grad_norm": 0.0004049557028338313,
|
| 7746 |
+
"learning_rate": 0.0001411849494732713,
|
| 7747 |
+
"loss": 46.0,
|
| 7748 |
+
"step": 1101
|
| 7749 |
+
},
|
| 7750 |
+
{
|
| 7751 |
+
"epoch": 0.1504899115769349,
|
| 7752 |
+
"grad_norm": 0.0005186764756217599,
|
| 7753 |
+
"learning_rate": 0.00014108918153265485,
|
| 7754 |
+
"loss": 46.0,
|
| 7755 |
+
"step": 1102
|
| 7756 |
+
},
|
| 7757 |
+
{
|
| 7758 |
+
"epoch": 0.15062647229524428,
|
| 7759 |
+
"grad_norm": 0.002497048582881689,
|
| 7760 |
+
"learning_rate": 0.00014099336823080865,
|
| 7761 |
+
"loss": 46.0,
|
| 7762 |
+
"step": 1103
|
| 7763 |
+
},
|
| 7764 |
+
{
|
| 7765 |
+
"epoch": 0.15076303301355365,
|
| 7766 |
+
"grad_norm": 0.00020665116608142853,
|
| 7767 |
+
"learning_rate": 0.00014089750967350781,
|
| 7768 |
+
"loss": 46.0,
|
| 7769 |
+
"step": 1104
|
| 7770 |
+
},
|
| 7771 |
+
{
|
| 7772 |
+
"epoch": 0.15089959373186304,
|
| 7773 |
+
"grad_norm": 0.0005294004804454744,
|
| 7774 |
+
"learning_rate": 0.0001408016059665773,
|
| 7775 |
+
"loss": 46.0,
|
| 7776 |
+
"step": 1105
|
| 7777 |
+
},
|
| 7778 |
+
{
|
| 7779 |
+
"epoch": 0.1510361544501724,
|
| 7780 |
+
"grad_norm": 0.00033093662932515144,
|
| 7781 |
+
"learning_rate": 0.00014070565721589195,
|
| 7782 |
+
"loss": 46.0,
|
| 7783 |
+
"step": 1106
|
| 7784 |
+
},
|
| 7785 |
+
{
|
| 7786 |
+
"epoch": 0.1511727151684818,
|
| 7787 |
+
"grad_norm": 0.0003122551424894482,
|
| 7788 |
+
"learning_rate": 0.00014060966352737628,
|
| 7789 |
+
"loss": 46.0,
|
| 7790 |
+
"step": 1107
|
| 7791 |
+
},
|
| 7792 |
+
{
|
| 7793 |
+
"epoch": 0.15130927588679116,
|
| 7794 |
+
"grad_norm": 0.0006798842805437744,
|
| 7795 |
+
"learning_rate": 0.00014051362500700447,
|
| 7796 |
+
"loss": 46.0,
|
| 7797 |
+
"step": 1108
|
| 7798 |
+
},
|
| 7799 |
+
{
|
| 7800 |
+
"epoch": 0.15144583660510055,
|
| 7801 |
+
"grad_norm": 0.0005911933840252459,
|
| 7802 |
+
"learning_rate": 0.00014041754176080017,
|
| 7803 |
+
"loss": 46.0,
|
| 7804 |
+
"step": 1109
|
| 7805 |
+
},
|
| 7806 |
+
{
|
| 7807 |
+
"epoch": 0.15158239732340992,
|
| 7808 |
+
"grad_norm": 0.0010168857406824827,
|
| 7809 |
+
"learning_rate": 0.00014032141389483648,
|
| 7810 |
+
"loss": 46.0,
|
| 7811 |
+
"step": 1110
|
| 7812 |
+
},
|
| 7813 |
+
{
|
| 7814 |
+
"epoch": 0.1517189580417193,
|
| 7815 |
+
"grad_norm": 0.00033409081515856087,
|
| 7816 |
+
"learning_rate": 0.00014022524151523563,
|
| 7817 |
+
"loss": 46.0,
|
| 7818 |
+
"step": 1111
|
| 7819 |
+
},
|
| 7820 |
+
{
|
| 7821 |
+
"epoch": 0.15185551876002867,
|
| 7822 |
+
"grad_norm": 0.0002926045854110271,
|
| 7823 |
+
"learning_rate": 0.00014012902472816907,
|
| 7824 |
+
"loss": 46.0,
|
| 7825 |
+
"step": 1112
|
| 7826 |
+
},
|
| 7827 |
+
{
|
| 7828 |
+
"epoch": 0.15199207947833807,
|
| 7829 |
+
"grad_norm": 0.0013335029361769557,
|
| 7830 |
+
"learning_rate": 0.00014003276363985727,
|
| 7831 |
+
"loss": 46.0,
|
| 7832 |
+
"step": 1113
|
| 7833 |
+
},
|
| 7834 |
+
{
|
| 7835 |
+
"epoch": 0.15212864019664743,
|
| 7836 |
+
"grad_norm": 0.000854438403621316,
|
| 7837 |
+
"learning_rate": 0.00013993645835656953,
|
| 7838 |
+
"loss": 46.0,
|
| 7839 |
+
"step": 1114
|
| 7840 |
+
},
|
| 7841 |
+
{
|
| 7842 |
+
"epoch": 0.15226520091495682,
|
| 7843 |
+
"grad_norm": 0.0009656418114900589,
|
| 7844 |
+
"learning_rate": 0.00013984010898462416,
|
| 7845 |
+
"loss": 46.0,
|
| 7846 |
+
"step": 1115
|
| 7847 |
+
},
|
| 7848 |
+
{
|
| 7849 |
+
"epoch": 0.1524017616332662,
|
| 7850 |
+
"grad_norm": 0.00042812732863239944,
|
| 7851 |
+
"learning_rate": 0.00013974371563038785,
|
| 7852 |
+
"loss": 46.0,
|
| 7853 |
+
"step": 1116
|
| 7854 |
+
},
|
| 7855 |
+
{
|
| 7856 |
+
"epoch": 0.15253832235157558,
|
| 7857 |
+
"grad_norm": 0.0006366227171383798,
|
| 7858 |
+
"learning_rate": 0.00013964727840027604,
|
| 7859 |
+
"loss": 46.0,
|
| 7860 |
+
"step": 1117
|
| 7861 |
+
},
|
| 7862 |
+
{
|
| 7863 |
+
"epoch": 0.15267488306988494,
|
| 7864 |
+
"grad_norm": 0.0011450116289779544,
|
| 7865 |
+
"learning_rate": 0.00013955079740075256,
|
| 7866 |
+
"loss": 46.0,
|
| 7867 |
+
"step": 1118
|
| 7868 |
+
},
|
| 7869 |
+
{
|
| 7870 |
+
"epoch": 0.15281144378819433,
|
| 7871 |
+
"grad_norm": 0.00023903950932435691,
|
| 7872 |
+
"learning_rate": 0.00013945427273832954,
|
| 7873 |
+
"loss": 46.0,
|
| 7874 |
+
"step": 1119
|
| 7875 |
+
},
|
| 7876 |
+
{
|
| 7877 |
+
"epoch": 0.1529480045065037,
|
| 7878 |
+
"grad_norm": 0.0011047361185774207,
|
| 7879 |
+
"learning_rate": 0.0001393577045195673,
|
| 7880 |
+
"loss": 46.0,
|
| 7881 |
+
"step": 1120
|
| 7882 |
+
},
|
| 7883 |
+
{
|
| 7884 |
+
"epoch": 0.1530845652248131,
|
| 7885 |
+
"grad_norm": 0.0004785344353877008,
|
| 7886 |
+
"learning_rate": 0.0001392610928510743,
|
| 7887 |
+
"loss": 46.0,
|
| 7888 |
+
"step": 1121
|
| 7889 |
+
},
|
| 7890 |
+
{
|
| 7891 |
+
"epoch": 0.15322112594312245,
|
| 7892 |
+
"grad_norm": 0.0008018110529519618,
|
| 7893 |
+
"learning_rate": 0.00013916443783950694,
|
| 7894 |
+
"loss": 46.0,
|
| 7895 |
+
"step": 1122
|
| 7896 |
+
},
|
| 7897 |
+
{
|
| 7898 |
+
"epoch": 0.15335768666143185,
|
| 7899 |
+
"grad_norm": 0.0012497154530137777,
|
| 7900 |
+
"learning_rate": 0.00013906773959156948,
|
| 7901 |
+
"loss": 46.0,
|
| 7902 |
+
"step": 1123
|
| 7903 |
+
},
|
| 7904 |
+
{
|
| 7905 |
+
"epoch": 0.1534942473797412,
|
| 7906 |
+
"grad_norm": 0.00042997964192181826,
|
| 7907 |
+
"learning_rate": 0.00013897099821401384,
|
| 7908 |
+
"loss": 46.0,
|
| 7909 |
+
"step": 1124
|
| 7910 |
+
},
|
| 7911 |
+
{
|
| 7912 |
+
"epoch": 0.1536308080980506,
|
| 7913 |
+
"grad_norm": 0.0003085716161876917,
|
| 7914 |
+
"learning_rate": 0.00013887421381363968,
|
| 7915 |
+
"loss": 46.0,
|
| 7916 |
+
"step": 1125
|
| 7917 |
+
},
|
| 7918 |
+
{
|
| 7919 |
+
"epoch": 0.15376736881635997,
|
| 7920 |
+
"grad_norm": 0.000396199116948992,
|
| 7921 |
+
"learning_rate": 0.00013877738649729405,
|
| 7922 |
+
"loss": 46.0,
|
| 7923 |
+
"step": 1126
|
| 7924 |
+
},
|
| 7925 |
+
{
|
| 7926 |
+
"epoch": 0.15390392953466936,
|
| 7927 |
+
"grad_norm": 0.0006908946088515222,
|
| 7928 |
+
"learning_rate": 0.00013868051637187144,
|
| 7929 |
+
"loss": 46.0,
|
| 7930 |
+
"step": 1127
|
| 7931 |
+
},
|
| 7932 |
+
{
|
| 7933 |
+
"epoch": 0.15404049025297872,
|
| 7934 |
+
"grad_norm": 0.0003736602666322142,
|
| 7935 |
+
"learning_rate": 0.00013858360354431355,
|
| 7936 |
+
"loss": 46.0,
|
| 7937 |
+
"step": 1128
|
| 7938 |
+
},
|
| 7939 |
+
{
|
| 7940 |
+
"epoch": 0.15417705097128812,
|
| 7941 |
+
"grad_norm": 0.0006938680890016258,
|
| 7942 |
+
"learning_rate": 0.00013848664812160925,
|
| 7943 |
+
"loss": 46.0,
|
| 7944 |
+
"step": 1129
|
| 7945 |
+
},
|
| 7946 |
+
{
|
| 7947 |
+
"epoch": 0.15431361168959748,
|
| 7948 |
+
"grad_norm": 0.0005900769028812647,
|
| 7949 |
+
"learning_rate": 0.00013838965021079446,
|
| 7950 |
+
"loss": 46.0,
|
| 7951 |
+
"step": 1130
|
| 7952 |
+
},
|
| 7953 |
+
{
|
| 7954 |
+
"epoch": 0.15445017240790687,
|
| 7955 |
+
"grad_norm": 0.002096734009683132,
|
| 7956 |
+
"learning_rate": 0.00013829260991895197,
|
| 7957 |
+
"loss": 46.0,
|
| 7958 |
+
"step": 1131
|
| 7959 |
+
},
|
| 7960 |
+
{
|
| 7961 |
+
"epoch": 0.15458673312621624,
|
| 7962 |
+
"grad_norm": 0.0011866495478898287,
|
| 7963 |
+
"learning_rate": 0.00013819552735321134,
|
| 7964 |
+
"loss": 46.0,
|
| 7965 |
+
"step": 1132
|
| 7966 |
+
},
|
| 7967 |
+
{
|
| 7968 |
+
"epoch": 0.15472329384452563,
|
| 7969 |
+
"grad_norm": 0.00037106405943632126,
|
| 7970 |
+
"learning_rate": 0.00013809840262074885,
|
| 7971 |
+
"loss": 46.0,
|
| 7972 |
+
"step": 1133
|
| 7973 |
+
},
|
| 7974 |
+
{
|
| 7975 |
+
"epoch": 0.154859854562835,
|
| 7976 |
+
"grad_norm": 0.0006705286214128137,
|
| 7977 |
+
"learning_rate": 0.0001380012358287873,
|
| 7978 |
+
"loss": 46.0,
|
| 7979 |
+
"step": 1134
|
| 7980 |
+
},
|
| 7981 |
+
{
|
| 7982 |
+
"epoch": 0.15499641528114438,
|
| 7983 |
+
"grad_norm": 0.00040015694685280323,
|
| 7984 |
+
"learning_rate": 0.0001379040270845959,
|
| 7985 |
+
"loss": 46.0,
|
| 7986 |
+
"step": 1135
|
| 7987 |
+
},
|
| 7988 |
+
{
|
| 7989 |
+
"epoch": 0.15513297599945375,
|
| 7990 |
+
"grad_norm": 0.0005712392157875001,
|
| 7991 |
+
"learning_rate": 0.00013780677649549025,
|
| 7992 |
+
"loss": 46.0,
|
| 7993 |
+
"step": 1136
|
| 7994 |
+
},
|
| 7995 |
+
{
|
| 7996 |
+
"epoch": 0.15526953671776314,
|
| 7997 |
+
"grad_norm": 0.00047067005652934313,
|
| 7998 |
+
"learning_rate": 0.00013770948416883205,
|
| 7999 |
+
"loss": 46.0,
|
| 8000 |
+
"step": 1137
|
| 8001 |
+
},
|
| 8002 |
+
{
|
| 8003 |
+
"epoch": 0.1554060974360725,
|
| 8004 |
+
"grad_norm": 0.0008378413622267544,
|
| 8005 |
+
"learning_rate": 0.00013761215021202916,
|
| 8006 |
+
"loss": 46.0,
|
| 8007 |
+
"step": 1138
|
| 8008 |
+
},
|
| 8009 |
+
{
|
| 8010 |
+
"epoch": 0.1555426581543819,
|
| 8011 |
+
"grad_norm": 0.0007883374928496778,
|
| 8012 |
+
"learning_rate": 0.00013751477473253533,
|
| 8013 |
+
"loss": 46.0,
|
| 8014 |
+
"step": 1139
|
| 8015 |
+
},
|
| 8016 |
+
{
|
| 8017 |
+
"epoch": 0.15567921887269126,
|
| 8018 |
+
"grad_norm": 0.0012233637971803546,
|
| 8019 |
+
"learning_rate": 0.0001374173578378502,
|
| 8020 |
+
"loss": 46.0,
|
| 8021 |
+
"step": 1140
|
| 8022 |
+
},
|
| 8023 |
+
{
|
| 8024 |
+
"epoch": 0.15581577959100065,
|
| 8025 |
+
"grad_norm": 0.004130385350435972,
|
| 8026 |
+
"learning_rate": 0.00013731989963551913,
|
| 8027 |
+
"loss": 46.0,
|
| 8028 |
+
"step": 1141
|
| 8029 |
+
},
|
| 8030 |
+
{
|
| 8031 |
+
"epoch": 0.15595234030931002,
|
| 8032 |
+
"grad_norm": 0.0029437027405947447,
|
| 8033 |
+
"learning_rate": 0.00013722240023313306,
|
| 8034 |
+
"loss": 46.0,
|
| 8035 |
+
"step": 1142
|
| 8036 |
+
},
|
| 8037 |
+
{
|
| 8038 |
+
"epoch": 0.1560889010276194,
|
| 8039 |
+
"grad_norm": 0.0003660391375888139,
|
| 8040 |
+
"learning_rate": 0.00013712485973832838,
|
| 8041 |
+
"loss": 46.0,
|
| 8042 |
+
"step": 1143
|
| 8043 |
+
},
|
| 8044 |
+
{
|
| 8045 |
+
"epoch": 0.15622546174592877,
|
| 8046 |
+
"grad_norm": 0.0020895125344395638,
|
| 8047 |
+
"learning_rate": 0.00013702727825878693,
|
| 8048 |
+
"loss": 46.0,
|
| 8049 |
+
"step": 1144
|
| 8050 |
+
},
|
| 8051 |
+
{
|
| 8052 |
+
"epoch": 0.15636202246423817,
|
| 8053 |
+
"grad_norm": 0.0016986053669825196,
|
| 8054 |
+
"learning_rate": 0.00013692965590223573,
|
| 8055 |
+
"loss": 46.0,
|
| 8056 |
+
"step": 1145
|
| 8057 |
+
},
|
| 8058 |
+
{
|
| 8059 |
+
"epoch": 0.15649858318254753,
|
| 8060 |
+
"grad_norm": 0.0005671007093042135,
|
| 8061 |
+
"learning_rate": 0.00013683199277644693,
|
| 8062 |
+
"loss": 46.0,
|
| 8063 |
+
"step": 1146
|
| 8064 |
+
},
|
| 8065 |
+
{
|
| 8066 |
+
"epoch": 0.15663514390085692,
|
| 8067 |
+
"grad_norm": 0.0009818869875743985,
|
| 8068 |
+
"learning_rate": 0.00013673428898923774,
|
| 8069 |
+
"loss": 46.0,
|
| 8070 |
+
"step": 1147
|
| 8071 |
+
},
|
| 8072 |
+
{
|
| 8073 |
+
"epoch": 0.15677170461916629,
|
| 8074 |
+
"grad_norm": 0.0004315339319873601,
|
| 8075 |
+
"learning_rate": 0.00013663654464847022,
|
| 8076 |
+
"loss": 46.0,
|
| 8077 |
+
"step": 1148
|
| 8078 |
+
},
|
| 8079 |
+
{
|
| 8080 |
+
"epoch": 0.15690826533747568,
|
| 8081 |
+
"grad_norm": 0.0006189457490108907,
|
| 8082 |
+
"learning_rate": 0.0001365387598620512,
|
| 8083 |
+
"loss": 46.0,
|
| 8084 |
+
"step": 1149
|
| 8085 |
+
},
|
| 8086 |
+
{
|
| 8087 |
+
"epoch": 0.15704482605578504,
|
| 8088 |
+
"grad_norm": 0.0007538103964179754,
|
| 8089 |
+
"learning_rate": 0.00013644093473793215,
|
| 8090 |
+
"loss": 46.0,
|
| 8091 |
+
"step": 1150
|
| 8092 |
+
},
|
| 8093 |
+
{
|
| 8094 |
+
"epoch": 0.15718138677409443,
|
| 8095 |
+
"grad_norm": 0.0005718530155718327,
|
| 8096 |
+
"learning_rate": 0.00013634306938410911,
|
| 8097 |
+
"loss": 46.0,
|
| 8098 |
+
"step": 1151
|
| 8099 |
+
},
|
| 8100 |
+
{
|
| 8101 |
+
"epoch": 0.1573179474924038,
|
| 8102 |
+
"grad_norm": 0.0006374249351210892,
|
| 8103 |
+
"learning_rate": 0.00013624516390862244,
|
| 8104 |
+
"loss": 46.0,
|
| 8105 |
+
"step": 1152
|
| 8106 |
+
},
|
| 8107 |
+
{
|
| 8108 |
+
"epoch": 0.1574545082107132,
|
| 8109 |
+
"grad_norm": 0.0007302735466510057,
|
| 8110 |
+
"learning_rate": 0.00013614721841955692,
|
| 8111 |
+
"loss": 46.0,
|
| 8112 |
+
"step": 1153
|
| 8113 |
+
},
|
| 8114 |
+
{
|
| 8115 |
+
"epoch": 0.15759106892902255,
|
| 8116 |
+
"grad_norm": 0.0005098398542031646,
|
| 8117 |
+
"learning_rate": 0.00013604923302504147,
|
| 8118 |
+
"loss": 46.0,
|
| 8119 |
+
"step": 1154
|
| 8120 |
+
},
|
| 8121 |
+
{
|
| 8122 |
+
"epoch": 0.15772762964733195,
|
| 8123 |
+
"grad_norm": 0.0005700102774426341,
|
| 8124 |
+
"learning_rate": 0.00013595120783324902,
|
| 8125 |
+
"loss": 46.0,
|
| 8126 |
+
"step": 1155
|
| 8127 |
+
},
|
| 8128 |
+
{
|
| 8129 |
+
"epoch": 0.1578641903656413,
|
| 8130 |
+
"grad_norm": 0.00040146647370420396,
|
| 8131 |
+
"learning_rate": 0.00013585314295239644,
|
| 8132 |
+
"loss": 46.0,
|
| 8133 |
+
"step": 1156
|
| 8134 |
+
},
|
| 8135 |
+
{
|
| 8136 |
+
"epoch": 0.1580007510839507,
|
| 8137 |
+
"grad_norm": 0.0008773392182774842,
|
| 8138 |
+
"learning_rate": 0.00013575503849074444,
|
| 8139 |
+
"loss": 46.0,
|
| 8140 |
+
"step": 1157
|
| 8141 |
+
},
|
| 8142 |
+
{
|
| 8143 |
+
"epoch": 0.15813731180226007,
|
| 8144 |
+
"grad_norm": 0.0003678193024825305,
|
| 8145 |
+
"learning_rate": 0.0001356568945565974,
|
| 8146 |
+
"loss": 46.0,
|
| 8147 |
+
"step": 1158
|
| 8148 |
+
},
|
| 8149 |
+
{
|
| 8150 |
+
"epoch": 0.15827387252056946,
|
| 8151 |
+
"grad_norm": 0.0013553998433053493,
|
| 8152 |
+
"learning_rate": 0.0001355587112583033,
|
| 8153 |
+
"loss": 46.0,
|
| 8154 |
+
"step": 1159
|
| 8155 |
+
},
|
| 8156 |
+
{
|
| 8157 |
+
"epoch": 0.15841043323887882,
|
| 8158 |
+
"grad_norm": 0.0009887435007840395,
|
| 8159 |
+
"learning_rate": 0.00013546048870425356,
|
| 8160 |
+
"loss": 46.0,
|
| 8161 |
+
"step": 1160
|
| 8162 |
+
},
|
| 8163 |
+
{
|
| 8164 |
+
"epoch": 0.15854699395718821,
|
| 8165 |
+
"grad_norm": 0.0019033915596082807,
|
| 8166 |
+
"learning_rate": 0.00013536222700288303,
|
| 8167 |
+
"loss": 46.0,
|
| 8168 |
+
"step": 1161
|
| 8169 |
+
},
|
| 8170 |
+
{
|
| 8171 |
+
"epoch": 0.1586835546754976,
|
| 8172 |
+
"grad_norm": 0.0006128349923528731,
|
| 8173 |
+
"learning_rate": 0.00013526392626266956,
|
| 8174 |
+
"loss": 46.0,
|
| 8175 |
+
"step": 1162
|
| 8176 |
+
},
|
| 8177 |
+
{
|
| 8178 |
+
"epoch": 0.15882011539380697,
|
| 8179 |
+
"grad_norm": 0.0005847832653671503,
|
| 8180 |
+
"learning_rate": 0.00013516558659213432,
|
| 8181 |
+
"loss": 46.0,
|
| 8182 |
+
"step": 1163
|
| 8183 |
+
},
|
| 8184 |
+
{
|
| 8185 |
+
"epoch": 0.15895667611211636,
|
| 8186 |
+
"grad_norm": 0.0005258521996438503,
|
| 8187 |
+
"learning_rate": 0.00013506720809984137,
|
| 8188 |
+
"loss": 46.0,
|
| 8189 |
+
"step": 1164
|
| 8190 |
+
},
|
| 8191 |
+
{
|
| 8192 |
+
"epoch": 0.15909323683042573,
|
| 8193 |
+
"grad_norm": 0.00040099999750964344,
|
| 8194 |
+
"learning_rate": 0.0001349687908943976,
|
| 8195 |
+
"loss": 46.0,
|
| 8196 |
+
"step": 1165
|
| 8197 |
+
},
|
| 8198 |
+
{
|
| 8199 |
+
"epoch": 0.15922979754873512,
|
| 8200 |
+
"grad_norm": 0.0007036002352833748,
|
| 8201 |
+
"learning_rate": 0.0001348703350844527,
|
| 8202 |
+
"loss": 46.0,
|
| 8203 |
+
"step": 1166
|
| 8204 |
+
},
|
| 8205 |
+
{
|
| 8206 |
+
"epoch": 0.15936635826704448,
|
| 8207 |
+
"grad_norm": 0.00037679230445064604,
|
| 8208 |
+
"learning_rate": 0.00013477184077869892,
|
| 8209 |
+
"loss": 46.0,
|
| 8210 |
+
"step": 1167
|
| 8211 |
+
},
|
| 8212 |
+
{
|
| 8213 |
+
"epoch": 0.15950291898535388,
|
| 8214 |
+
"grad_norm": 0.0005223838961683214,
|
| 8215 |
+
"learning_rate": 0.000134673308085871,
|
| 8216 |
+
"loss": 46.0,
|
| 8217 |
+
"step": 1168
|
| 8218 |
+
},
|
| 8219 |
+
{
|
| 8220 |
+
"epoch": 0.15963947970366324,
|
| 8221 |
+
"grad_norm": 0.0008007617434486747,
|
| 8222 |
+
"learning_rate": 0.0001345747371147461,
|
| 8223 |
+
"loss": 46.0,
|
| 8224 |
+
"step": 1169
|
| 8225 |
+
},
|
| 8226 |
+
{
|
| 8227 |
+
"epoch": 0.15977604042197263,
|
| 8228 |
+
"grad_norm": 0.000517090258654207,
|
| 8229 |
+
"learning_rate": 0.0001344761279741437,
|
| 8230 |
+
"loss": 46.0,
|
| 8231 |
+
"step": 1170
|
| 8232 |
+
},
|
| 8233 |
+
{
|
| 8234 |
+
"epoch": 0.159912601140282,
|
| 8235 |
+
"grad_norm": 0.0008526128367520869,
|
| 8236 |
+
"learning_rate": 0.0001343774807729253,
|
| 8237 |
+
"loss": 46.0,
|
| 8238 |
+
"step": 1171
|
| 8239 |
+
},
|
| 8240 |
+
{
|
| 8241 |
+
"epoch": 0.1600491618585914,
|
| 8242 |
+
"grad_norm": 0.0011307375971227884,
|
| 8243 |
+
"learning_rate": 0.0001342787956199945,
|
| 8244 |
+
"loss": 46.0,
|
| 8245 |
+
"step": 1172
|
| 8246 |
+
},
|
| 8247 |
+
{
|
| 8248 |
+
"epoch": 0.16018572257690075,
|
| 8249 |
+
"grad_norm": 0.00033380292006768286,
|
| 8250 |
+
"learning_rate": 0.00013418007262429668,
|
| 8251 |
+
"loss": 46.0,
|
| 8252 |
+
"step": 1173
|
| 8253 |
+
},
|
| 8254 |
+
{
|
| 8255 |
+
"epoch": 0.16032228329521014,
|
| 8256 |
+
"grad_norm": 0.000278162129689008,
|
| 8257 |
+
"learning_rate": 0.00013408131189481911,
|
| 8258 |
+
"loss": 46.0,
|
| 8259 |
+
"step": 1174
|
| 8260 |
+
},
|
| 8261 |
+
{
|
| 8262 |
+
"epoch": 0.1604588440135195,
|
| 8263 |
+
"grad_norm": 0.00258398219011724,
|
| 8264 |
+
"learning_rate": 0.00013398251354059077,
|
| 8265 |
+
"loss": 46.0,
|
| 8266 |
+
"step": 1175
|
| 8267 |
+
},
|
| 8268 |
+
{
|
| 8269 |
+
"epoch": 0.1605954047318289,
|
| 8270 |
+
"grad_norm": 0.0006168753025121987,
|
| 8271 |
+
"learning_rate": 0.000133883677670682,
|
| 8272 |
+
"loss": 46.0,
|
| 8273 |
+
"step": 1176
|
| 8274 |
+
},
|
| 8275 |
+
{
|
| 8276 |
+
"epoch": 0.16073196545013826,
|
| 8277 |
+
"grad_norm": 0.00029901755624450743,
|
| 8278 |
+
"learning_rate": 0.0001337848043942047,
|
| 8279 |
+
"loss": 46.0,
|
| 8280 |
+
"step": 1177
|
| 8281 |
+
},
|
| 8282 |
+
{
|
| 8283 |
+
"epoch": 0.16086852616844766,
|
| 8284 |
+
"grad_norm": 0.000997790601104498,
|
| 8285 |
+
"learning_rate": 0.00013368589382031196,
|
| 8286 |
+
"loss": 46.0,
|
| 8287 |
+
"step": 1178
|
| 8288 |
+
},
|
| 8289 |
+
{
|
| 8290 |
+
"epoch": 0.16100508688675702,
|
| 8291 |
+
"grad_norm": 0.0006558905588462949,
|
| 8292 |
+
"learning_rate": 0.00013358694605819814,
|
| 8293 |
+
"loss": 46.0,
|
| 8294 |
+
"step": 1179
|
| 8295 |
+
},
|
| 8296 |
+
{
|
| 8297 |
+
"epoch": 0.1611416476050664,
|
| 8298 |
+
"grad_norm": 0.0006217307527549565,
|
| 8299 |
+
"learning_rate": 0.00013348796121709862,
|
| 8300 |
+
"loss": 46.0,
|
| 8301 |
+
"step": 1180
|
| 8302 |
+
},
|
| 8303 |
+
{
|
| 8304 |
+
"epoch": 0.16127820832337578,
|
| 8305 |
+
"grad_norm": 0.0004898210754618049,
|
| 8306 |
+
"learning_rate": 0.00013338893940628973,
|
| 8307 |
+
"loss": 46.0,
|
| 8308 |
+
"step": 1181
|
| 8309 |
+
},
|
| 8310 |
+
{
|
| 8311 |
+
"epoch": 0.16141476904168517,
|
| 8312 |
+
"grad_norm": 0.0009382545249536633,
|
| 8313 |
+
"learning_rate": 0.00013328988073508852,
|
| 8314 |
+
"loss": 46.0,
|
| 8315 |
+
"step": 1182
|
| 8316 |
+
},
|
| 8317 |
+
{
|
| 8318 |
+
"epoch": 0.16155132975999453,
|
| 8319 |
+
"grad_norm": 0.0006358098471537232,
|
| 8320 |
+
"learning_rate": 0.00013319078531285285,
|
| 8321 |
+
"loss": 46.0,
|
| 8322 |
+
"step": 1183
|
| 8323 |
+
},
|
| 8324 |
+
{
|
| 8325 |
+
"epoch": 0.16168789047830393,
|
| 8326 |
+
"grad_norm": 0.0012355463113635778,
|
| 8327 |
+
"learning_rate": 0.00013309165324898112,
|
| 8328 |
+
"loss": 46.0,
|
| 8329 |
+
"step": 1184
|
| 8330 |
+
},
|
| 8331 |
+
{
|
| 8332 |
+
"epoch": 0.1618244511966133,
|
| 8333 |
+
"grad_norm": 0.0005907994927838445,
|
| 8334 |
+
"learning_rate": 0.00013299248465291214,
|
| 8335 |
+
"loss": 46.0,
|
| 8336 |
+
"step": 1185
|
| 8337 |
+
},
|
| 8338 |
+
{
|
| 8339 |
+
"epoch": 0.16196101191492268,
|
| 8340 |
+
"grad_norm": 0.002715210895985365,
|
| 8341 |
+
"learning_rate": 0.00013289327963412513,
|
| 8342 |
+
"loss": 46.0,
|
| 8343 |
+
"step": 1186
|
| 8344 |
+
},
|
| 8345 |
+
{
|
| 8346 |
+
"epoch": 0.16209757263323205,
|
| 8347 |
+
"grad_norm": 0.001038241432979703,
|
| 8348 |
+
"learning_rate": 0.00013279403830213942,
|
| 8349 |
+
"loss": 46.0,
|
| 8350 |
+
"step": 1187
|
| 8351 |
+
},
|
| 8352 |
+
{
|
| 8353 |
+
"epoch": 0.16223413335154144,
|
| 8354 |
+
"grad_norm": 0.0015468295896425843,
|
| 8355 |
+
"learning_rate": 0.00013269476076651447,
|
| 8356 |
+
"loss": 46.0,
|
| 8357 |
+
"step": 1188
|
| 8358 |
+
},
|
| 8359 |
+
{
|
| 8360 |
+
"epoch": 0.1623706940698508,
|
| 8361 |
+
"grad_norm": 0.0017287740483880043,
|
| 8362 |
+
"learning_rate": 0.00013259544713684974,
|
| 8363 |
+
"loss": 46.0,
|
| 8364 |
+
"step": 1189
|
| 8365 |
+
},
|
| 8366 |
+
{
|
| 8367 |
+
"epoch": 0.1625072547881602,
|
| 8368 |
+
"grad_norm": 0.0018615883309394121,
|
| 8369 |
+
"learning_rate": 0.00013249609752278454,
|
| 8370 |
+
"loss": 46.0,
|
| 8371 |
+
"step": 1190
|
| 8372 |
+
},
|
| 8373 |
+
{
|
| 8374 |
+
"epoch": 0.16264381550646956,
|
| 8375 |
+
"grad_norm": 0.00047380104660987854,
|
| 8376 |
+
"learning_rate": 0.0001323967120339978,
|
| 8377 |
+
"loss": 46.0,
|
| 8378 |
+
"step": 1191
|
| 8379 |
+
},
|
| 8380 |
+
{
|
| 8381 |
+
"epoch": 0.16278037622477895,
|
| 8382 |
+
"grad_norm": 0.001483297673985362,
|
| 8383 |
+
"learning_rate": 0.00013229729078020823,
|
| 8384 |
+
"loss": 46.0,
|
| 8385 |
+
"step": 1192
|
| 8386 |
+
},
|
| 8387 |
+
{
|
| 8388 |
+
"epoch": 0.16291693694308831,
|
| 8389 |
+
"grad_norm": 0.0004559273656923324,
|
| 8390 |
+
"learning_rate": 0.00013219783387117385,
|
| 8391 |
+
"loss": 46.0,
|
| 8392 |
+
"step": 1193
|
| 8393 |
+
},
|
| 8394 |
+
{
|
| 8395 |
+
"epoch": 0.1630534976613977,
|
| 8396 |
+
"grad_norm": 0.0008615512633696198,
|
| 8397 |
+
"learning_rate": 0.00013209834141669213,
|
| 8398 |
+
"loss": 46.0,
|
| 8399 |
+
"step": 1194
|
| 8400 |
+
},
|
| 8401 |
+
{
|
| 8402 |
+
"epoch": 0.16319005837970707,
|
| 8403 |
+
"grad_norm": 0.0005908702732995152,
|
| 8404 |
+
"learning_rate": 0.0001319988135265998,
|
| 8405 |
+
"loss": 46.0,
|
| 8406 |
+
"step": 1195
|
| 8407 |
+
},
|
| 8408 |
+
{
|
| 8409 |
+
"epoch": 0.16332661909801646,
|
| 8410 |
+
"grad_norm": 0.0008730573463253677,
|
| 8411 |
+
"learning_rate": 0.00013189925031077267,
|
| 8412 |
+
"loss": 46.0,
|
| 8413 |
+
"step": 1196
|
| 8414 |
+
},
|
| 8415 |
+
{
|
| 8416 |
+
"epoch": 0.16346317981632583,
|
| 8417 |
+
"grad_norm": 0.0012465447653084993,
|
| 8418 |
+
"learning_rate": 0.00013179965187912554,
|
| 8419 |
+
"loss": 46.0,
|
| 8420 |
+
"step": 1197
|
| 8421 |
+
},
|
| 8422 |
+
{
|
| 8423 |
+
"epoch": 0.16359974053463522,
|
| 8424 |
+
"grad_norm": 0.0013986461563035846,
|
| 8425 |
+
"learning_rate": 0.00013170001834161209,
|
| 8426 |
+
"loss": 46.0,
|
| 8427 |
+
"step": 1198
|
| 8428 |
+
},
|
| 8429 |
+
{
|
| 8430 |
+
"epoch": 0.16373630125294458,
|
| 8431 |
+
"grad_norm": 0.0011632711393758655,
|
| 8432 |
+
"learning_rate": 0.0001316003498082248,
|
| 8433 |
+
"loss": 46.0,
|
| 8434 |
+
"step": 1199
|
| 8435 |
+
},
|
| 8436 |
+
{
|
| 8437 |
+
"epoch": 0.16387286197125397,
|
| 8438 |
+
"grad_norm": 0.001200903090648353,
|
| 8439 |
+
"learning_rate": 0.0001315006463889948,
|
| 8440 |
+
"loss": 46.0,
|
| 8441 |
+
"step": 1200
|
| 8442 |
+
},
|
| 8443 |
+
{
|
| 8444 |
+
"epoch": 0.16387286197125397,
|
| 8445 |
+
"eval_loss": 11.5,
|
| 8446 |
+
"eval_runtime": 20.5706,
|
| 8447 |
+
"eval_samples_per_second": 149.923,
|
| 8448 |
+
"eval_steps_per_second": 74.961,
|
| 8449 |
+
"step": 1200
|
| 8450 |
}
|
| 8451 |
],
|
| 8452 |
"logging_steps": 1,
|
|
|
|
| 8461 |
"early_stopping_threshold": 0.0
|
| 8462 |
},
|
| 8463 |
"attributes": {
|
| 8464 |
+
"early_stopping_patience_counter": 3
|
| 8465 |
}
|
| 8466 |
},
|
| 8467 |
"TrainerControl": {
|
|
|
|
| 8470 |
"should_evaluate": false,
|
| 8471 |
"should_log": false,
|
| 8472 |
"should_save": true,
|
| 8473 |
+
"should_training_stop": true
|
| 8474 |
},
|
| 8475 |
"attributes": {}
|
| 8476 |
}
|
| 8477 |
},
|
| 8478 |
+
"total_flos": 42923841650688.0,
|
| 8479 |
"train_batch_size": 2,
|
| 8480 |
"trial_name": null,
|
| 8481 |
"trial_params": null
|