Model save
Browse files- README.md +78 -0
- adapter_model.safetensors +1 -1
- all_results.json +9 -0
- runs/Aug10_05-21-27_e8f6cb577ded/events.out.tfevents.1723267852.e8f6cb577ded.180766.0 +2 -2
- train_results.json +9 -0
- trainer_state.json +2173 -0
README.md
ADDED
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
base_model: mistralai/Mistral-7B-v0.3
|
3 |
+
datasets:
|
4 |
+
- generator
|
5 |
+
library_name: peft
|
6 |
+
license: apache-2.0
|
7 |
+
tags:
|
8 |
+
- trl
|
9 |
+
- sft
|
10 |
+
- generated_from_trainer
|
11 |
+
model-index:
|
12 |
+
- name: mistral_7b_0_3-closedqa-gpt4o-100k
|
13 |
+
results: []
|
14 |
+
---
|
15 |
+
|
16 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
17 |
+
should probably proofread and complete it, then remove this comment. -->
|
18 |
+
|
19 |
+
# mistral_7b_0_3-closedqa-gpt4o-100k
|
20 |
+
|
21 |
+
This model is a fine-tuned version of [mistralai/Mistral-7B-v0.3](https://huggingface.co/mistralai/Mistral-7B-v0.3) on the generator dataset.
|
22 |
+
It achieves the following results on the evaluation set:
|
23 |
+
- Loss: 1.7580
|
24 |
+
|
25 |
+
## Model description
|
26 |
+
|
27 |
+
More information needed
|
28 |
+
|
29 |
+
## Intended uses & limitations
|
30 |
+
|
31 |
+
More information needed
|
32 |
+
|
33 |
+
## Training and evaluation data
|
34 |
+
|
35 |
+
More information needed
|
36 |
+
|
37 |
+
## Training procedure
|
38 |
+
|
39 |
+
### Training hyperparameters
|
40 |
+
|
41 |
+
The following hyperparameters were used during training:
|
42 |
+
- learning_rate: 0.0002
|
43 |
+
- train_batch_size: 8
|
44 |
+
- eval_batch_size: 4
|
45 |
+
- seed: 42
|
46 |
+
- distributed_type: multi-GPU
|
47 |
+
- num_devices: 8
|
48 |
+
- gradient_accumulation_steps: 2
|
49 |
+
- total_train_batch_size: 128
|
50 |
+
- total_eval_batch_size: 32
|
51 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
52 |
+
- lr_scheduler_type: cosine
|
53 |
+
- lr_scheduler_warmup_ratio: 0.1
|
54 |
+
- num_epochs: 10
|
55 |
+
|
56 |
+
### Training results
|
57 |
+
|
58 |
+
| Training Loss | Epoch | Step | Validation Loss |
|
59 |
+
|:-------------:|:-----:|:----:|:---------------:|
|
60 |
+
| 0.7127 | 1.0 | 146 | 1.7010 |
|
61 |
+
| 0.6581 | 2.0 | 292 | 1.6853 |
|
62 |
+
| 0.6323 | 3.0 | 438 | 1.6891 |
|
63 |
+
| 0.6067 | 4.0 | 584 | 1.7035 |
|
64 |
+
| 0.591 | 5.0 | 730 | 1.7226 |
|
65 |
+
| 0.5752 | 6.0 | 876 | 1.7259 |
|
66 |
+
| 0.5603 | 7.0 | 1022 | 1.7422 |
|
67 |
+
| 0.5632 | 8.0 | 1168 | 1.7483 |
|
68 |
+
| 0.5497 | 9.0 | 1314 | 1.7570 |
|
69 |
+
| 0.5353 | 10.0 | 1460 | 1.7580 |
|
70 |
+
|
71 |
+
|
72 |
+
### Framework versions
|
73 |
+
|
74 |
+
- PEFT 0.12.0
|
75 |
+
- Transformers 4.44.0
|
76 |
+
- Pytorch 2.2.0+cu121
|
77 |
+
- Datasets 2.20.0
|
78 |
+
- Tokenizers 0.19.1
|
adapter_model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 591447088
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0afc889d9040d98f57ac27fff09868e2858d5fa4eba9848582bf6db9f2736cf3
|
3 |
size 591447088
|
all_results.json
ADDED
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 10.0,
|
3 |
+
"total_flos": 8.183666472539652e+18,
|
4 |
+
"train_loss": 0.6181400966970888,
|
5 |
+
"train_runtime": 11966.6323,
|
6 |
+
"train_samples": 111440,
|
7 |
+
"train_samples_per_second": 15.569,
|
8 |
+
"train_steps_per_second": 0.122
|
9 |
+
}
|
runs/Aug10_05-21-27_e8f6cb577ded/events.out.tfevents.1723267852.e8f6cb577ded.180766.0
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9004177d0ebe8f4443b9605f1586b0e90bd13a5381e3635e094419e787107289
|
3 |
+
size 70731
|
train_results.json
ADDED
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 10.0,
|
3 |
+
"total_flos": 8.183666472539652e+18,
|
4 |
+
"train_loss": 0.6181400966970888,
|
5 |
+
"train_runtime": 11966.6323,
|
6 |
+
"train_samples": 111440,
|
7 |
+
"train_samples_per_second": 15.569,
|
8 |
+
"train_steps_per_second": 0.122
|
9 |
+
}
|
trainer_state.json
ADDED
@@ -0,0 +1,2173 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 10.0,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 1460,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.00684931506849315,
|
13 |
+
"grad_norm": 8.315443992614746,
|
14 |
+
"learning_rate": 1.3698630136986302e-06,
|
15 |
+
"loss": 1.7877,
|
16 |
+
"step": 1
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.03424657534246575,
|
20 |
+
"grad_norm": 7.738222599029541,
|
21 |
+
"learning_rate": 6.849315068493151e-06,
|
22 |
+
"loss": 1.7409,
|
23 |
+
"step": 5
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 0.0684931506849315,
|
27 |
+
"grad_norm": 7.3871989250183105,
|
28 |
+
"learning_rate": 1.3698630136986302e-05,
|
29 |
+
"loss": 1.6148,
|
30 |
+
"step": 10
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 0.10273972602739725,
|
34 |
+
"grad_norm": 4.8537421226501465,
|
35 |
+
"learning_rate": 2.0547945205479453e-05,
|
36 |
+
"loss": 1.3419,
|
37 |
+
"step": 15
|
38 |
+
},
|
39 |
+
{
|
40 |
+
"epoch": 0.136986301369863,
|
41 |
+
"grad_norm": 2.9443578720092773,
|
42 |
+
"learning_rate": 2.7397260273972603e-05,
|
43 |
+
"loss": 1.1536,
|
44 |
+
"step": 20
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.17123287671232876,
|
48 |
+
"grad_norm": 1.285714864730835,
|
49 |
+
"learning_rate": 3.424657534246575e-05,
|
50 |
+
"loss": 1.0747,
|
51 |
+
"step": 25
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.2054794520547945,
|
55 |
+
"grad_norm": 0.9238005876541138,
|
56 |
+
"learning_rate": 4.1095890410958905e-05,
|
57 |
+
"loss": 0.9979,
|
58 |
+
"step": 30
|
59 |
+
},
|
60 |
+
{
|
61 |
+
"epoch": 0.23972602739726026,
|
62 |
+
"grad_norm": 0.6104592680931091,
|
63 |
+
"learning_rate": 4.794520547945205e-05,
|
64 |
+
"loss": 0.9327,
|
65 |
+
"step": 35
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"epoch": 0.273972602739726,
|
69 |
+
"grad_norm": 0.6212075352668762,
|
70 |
+
"learning_rate": 5.479452054794521e-05,
|
71 |
+
"loss": 0.8795,
|
72 |
+
"step": 40
|
73 |
+
},
|
74 |
+
{
|
75 |
+
"epoch": 0.3082191780821918,
|
76 |
+
"grad_norm": 0.4866364002227783,
|
77 |
+
"learning_rate": 6.164383561643835e-05,
|
78 |
+
"loss": 0.859,
|
79 |
+
"step": 45
|
80 |
+
},
|
81 |
+
{
|
82 |
+
"epoch": 0.3424657534246575,
|
83 |
+
"grad_norm": 0.33247315883636475,
|
84 |
+
"learning_rate": 6.84931506849315e-05,
|
85 |
+
"loss": 0.831,
|
86 |
+
"step": 50
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.3767123287671233,
|
90 |
+
"grad_norm": 0.33287501335144043,
|
91 |
+
"learning_rate": 7.534246575342466e-05,
|
92 |
+
"loss": 0.8186,
|
93 |
+
"step": 55
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.410958904109589,
|
97 |
+
"grad_norm": 0.40740513801574707,
|
98 |
+
"learning_rate": 8.219178082191781e-05,
|
99 |
+
"loss": 0.8011,
|
100 |
+
"step": 60
|
101 |
+
},
|
102 |
+
{
|
103 |
+
"epoch": 0.4452054794520548,
|
104 |
+
"grad_norm": 0.766364574432373,
|
105 |
+
"learning_rate": 8.904109589041096e-05,
|
106 |
+
"loss": 0.7926,
|
107 |
+
"step": 65
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"epoch": 0.4794520547945205,
|
111 |
+
"grad_norm": 0.7360146045684814,
|
112 |
+
"learning_rate": 9.58904109589041e-05,
|
113 |
+
"loss": 0.7767,
|
114 |
+
"step": 70
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 0.5136986301369864,
|
118 |
+
"grad_norm": 0.3070668578147888,
|
119 |
+
"learning_rate": 0.00010273972602739728,
|
120 |
+
"loss": 0.7761,
|
121 |
+
"step": 75
|
122 |
+
},
|
123 |
+
{
|
124 |
+
"epoch": 0.547945205479452,
|
125 |
+
"grad_norm": 0.7730902433395386,
|
126 |
+
"learning_rate": 0.00010958904109589041,
|
127 |
+
"loss": 0.778,
|
128 |
+
"step": 80
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.5821917808219178,
|
132 |
+
"grad_norm": 0.4104633033275604,
|
133 |
+
"learning_rate": 0.00011643835616438356,
|
134 |
+
"loss": 0.7555,
|
135 |
+
"step": 85
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 0.6164383561643836,
|
139 |
+
"grad_norm": 0.2929731011390686,
|
140 |
+
"learning_rate": 0.0001232876712328767,
|
141 |
+
"loss": 0.7522,
|
142 |
+
"step": 90
|
143 |
+
},
|
144 |
+
{
|
145 |
+
"epoch": 0.6506849315068494,
|
146 |
+
"grad_norm": 0.33198031783103943,
|
147 |
+
"learning_rate": 0.00013013698630136988,
|
148 |
+
"loss": 0.7533,
|
149 |
+
"step": 95
|
150 |
+
},
|
151 |
+
{
|
152 |
+
"epoch": 0.684931506849315,
|
153 |
+
"grad_norm": 0.35038191080093384,
|
154 |
+
"learning_rate": 0.000136986301369863,
|
155 |
+
"loss": 0.7433,
|
156 |
+
"step": 100
|
157 |
+
},
|
158 |
+
{
|
159 |
+
"epoch": 0.7191780821917808,
|
160 |
+
"grad_norm": 0.380381315946579,
|
161 |
+
"learning_rate": 0.00014383561643835618,
|
162 |
+
"loss": 0.7446,
|
163 |
+
"step": 105
|
164 |
+
},
|
165 |
+
{
|
166 |
+
"epoch": 0.7534246575342466,
|
167 |
+
"grad_norm": 0.3344345986843109,
|
168 |
+
"learning_rate": 0.00015068493150684933,
|
169 |
+
"loss": 0.7325,
|
170 |
+
"step": 110
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"epoch": 0.7876712328767124,
|
174 |
+
"grad_norm": 0.5609772205352783,
|
175 |
+
"learning_rate": 0.00015753424657534247,
|
176 |
+
"loss": 0.7339,
|
177 |
+
"step": 115
|
178 |
+
},
|
179 |
+
{
|
180 |
+
"epoch": 0.821917808219178,
|
181 |
+
"grad_norm": 0.5873807668685913,
|
182 |
+
"learning_rate": 0.00016438356164383562,
|
183 |
+
"loss": 0.7253,
|
184 |
+
"step": 120
|
185 |
+
},
|
186 |
+
{
|
187 |
+
"epoch": 0.8561643835616438,
|
188 |
+
"grad_norm": 0.5051682591438293,
|
189 |
+
"learning_rate": 0.00017123287671232877,
|
190 |
+
"loss": 0.7272,
|
191 |
+
"step": 125
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"epoch": 0.8904109589041096,
|
195 |
+
"grad_norm": 0.3545282483100891,
|
196 |
+
"learning_rate": 0.00017808219178082192,
|
197 |
+
"loss": 0.7126,
|
198 |
+
"step": 130
|
199 |
+
},
|
200 |
+
{
|
201 |
+
"epoch": 0.9246575342465754,
|
202 |
+
"grad_norm": 0.40065136551856995,
|
203 |
+
"learning_rate": 0.0001849315068493151,
|
204 |
+
"loss": 0.7165,
|
205 |
+
"step": 135
|
206 |
+
},
|
207 |
+
{
|
208 |
+
"epoch": 0.958904109589041,
|
209 |
+
"grad_norm": 0.6324636936187744,
|
210 |
+
"learning_rate": 0.0001917808219178082,
|
211 |
+
"loss": 0.7096,
|
212 |
+
"step": 140
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 0.9931506849315068,
|
216 |
+
"grad_norm": 0.8474305868148804,
|
217 |
+
"learning_rate": 0.00019863013698630139,
|
218 |
+
"loss": 0.7127,
|
219 |
+
"step": 145
|
220 |
+
},
|
221 |
+
{
|
222 |
+
"epoch": 1.0,
|
223 |
+
"eval_loss": 1.7009907960891724,
|
224 |
+
"eval_runtime": 0.6831,
|
225 |
+
"eval_samples_per_second": 8.783,
|
226 |
+
"eval_steps_per_second": 1.464,
|
227 |
+
"step": 146
|
228 |
+
},
|
229 |
+
{
|
230 |
+
"epoch": 1.0273972602739727,
|
231 |
+
"grad_norm": 0.34465500712394714,
|
232 |
+
"learning_rate": 0.00019999542705801296,
|
233 |
+
"loss": 0.7106,
|
234 |
+
"step": 150
|
235 |
+
},
|
236 |
+
{
|
237 |
+
"epoch": 1.0616438356164384,
|
238 |
+
"grad_norm": 0.32777366042137146,
|
239 |
+
"learning_rate": 0.00019997685019798912,
|
240 |
+
"loss": 0.6947,
|
241 |
+
"step": 155
|
242 |
+
},
|
243 |
+
{
|
244 |
+
"epoch": 1.095890410958904,
|
245 |
+
"grad_norm": 0.34322458505630493,
|
246 |
+
"learning_rate": 0.00019994398626371643,
|
247 |
+
"loss": 0.6952,
|
248 |
+
"step": 160
|
249 |
+
},
|
250 |
+
{
|
251 |
+
"epoch": 1.13013698630137,
|
252 |
+
"grad_norm": 0.2865244150161743,
|
253 |
+
"learning_rate": 0.00019989683995157677,
|
254 |
+
"loss": 0.6965,
|
255 |
+
"step": 165
|
256 |
+
},
|
257 |
+
{
|
258 |
+
"epoch": 1.1643835616438356,
|
259 |
+
"grad_norm": 0.2771136462688446,
|
260 |
+
"learning_rate": 0.0001998354179989585,
|
261 |
+
"loss": 0.6933,
|
262 |
+
"step": 170
|
263 |
+
},
|
264 |
+
{
|
265 |
+
"epoch": 1.1986301369863013,
|
266 |
+
"grad_norm": 0.3072766661643982,
|
267 |
+
"learning_rate": 0.00019975972918329356,
|
268 |
+
"loss": 0.6938,
|
269 |
+
"step": 175
|
270 |
+
},
|
271 |
+
{
|
272 |
+
"epoch": 1.2328767123287672,
|
273 |
+
"grad_norm": 0.2912730872631073,
|
274 |
+
"learning_rate": 0.00019966978432080316,
|
275 |
+
"loss": 0.6913,
|
276 |
+
"step": 180
|
277 |
+
},
|
278 |
+
{
|
279 |
+
"epoch": 1.2671232876712328,
|
280 |
+
"grad_norm": 0.33161672949790955,
|
281 |
+
"learning_rate": 0.00019956559626495212,
|
282 |
+
"loss": 0.6856,
|
283 |
+
"step": 185
|
284 |
+
},
|
285 |
+
{
|
286 |
+
"epoch": 1.3013698630136985,
|
287 |
+
"grad_norm": 0.40179818868637085,
|
288 |
+
"learning_rate": 0.00019944717990461207,
|
289 |
+
"loss": 0.6863,
|
290 |
+
"step": 190
|
291 |
+
},
|
292 |
+
{
|
293 |
+
"epoch": 1.3356164383561644,
|
294 |
+
"grad_norm": 0.32012802362442017,
|
295 |
+
"learning_rate": 0.00019931455216193382,
|
296 |
+
"loss": 0.6828,
|
297 |
+
"step": 195
|
298 |
+
},
|
299 |
+
{
|
300 |
+
"epoch": 1.36986301369863,
|
301 |
+
"grad_norm": 0.9038059115409851,
|
302 |
+
"learning_rate": 0.000199167731989929,
|
303 |
+
"loss": 0.6874,
|
304 |
+
"step": 200
|
305 |
+
},
|
306 |
+
{
|
307 |
+
"epoch": 1.404109589041096,
|
308 |
+
"grad_norm": 0.8754851222038269,
|
309 |
+
"learning_rate": 0.00019900674036976173,
|
310 |
+
"loss": 0.6794,
|
311 |
+
"step": 205
|
312 |
+
},
|
313 |
+
{
|
314 |
+
"epoch": 1.4383561643835616,
|
315 |
+
"grad_norm": 0.39323532581329346,
|
316 |
+
"learning_rate": 0.00019883160030775016,
|
317 |
+
"loss": 0.6799,
|
318 |
+
"step": 210
|
319 |
+
},
|
320 |
+
{
|
321 |
+
"epoch": 1.4726027397260273,
|
322 |
+
"grad_norm": 0.44316402077674866,
|
323 |
+
"learning_rate": 0.00019864233683207906,
|
324 |
+
"loss": 0.6801,
|
325 |
+
"step": 215
|
326 |
+
},
|
327 |
+
{
|
328 |
+
"epoch": 1.5068493150684932,
|
329 |
+
"grad_norm": 0.3385327160358429,
|
330 |
+
"learning_rate": 0.00019843897698922284,
|
331 |
+
"loss": 0.6759,
|
332 |
+
"step": 220
|
333 |
+
},
|
334 |
+
{
|
335 |
+
"epoch": 1.541095890410959,
|
336 |
+
"grad_norm": 0.26804545521736145,
|
337 |
+
"learning_rate": 0.00019822154984008088,
|
338 |
+
"loss": 0.6704,
|
339 |
+
"step": 225
|
340 |
+
},
|
341 |
+
{
|
342 |
+
"epoch": 1.5753424657534247,
|
343 |
+
"grad_norm": 0.26624366641044617,
|
344 |
+
"learning_rate": 0.0001979900864558242,
|
345 |
+
"loss": 0.6747,
|
346 |
+
"step": 230
|
347 |
+
},
|
348 |
+
{
|
349 |
+
"epoch": 1.6095890410958904,
|
350 |
+
"grad_norm": 0.2834051549434662,
|
351 |
+
"learning_rate": 0.00019774461991345577,
|
352 |
+
"loss": 0.6674,
|
353 |
+
"step": 235
|
354 |
+
},
|
355 |
+
{
|
356 |
+
"epoch": 1.643835616438356,
|
357 |
+
"grad_norm": 0.36631157994270325,
|
358 |
+
"learning_rate": 0.00019748518529108316,
|
359 |
+
"loss": 0.6702,
|
360 |
+
"step": 240
|
361 |
+
},
|
362 |
+
{
|
363 |
+
"epoch": 1.678082191780822,
|
364 |
+
"grad_norm": 0.2652304768562317,
|
365 |
+
"learning_rate": 0.00019721181966290613,
|
366 |
+
"loss": 0.6704,
|
367 |
+
"step": 245
|
368 |
+
},
|
369 |
+
{
|
370 |
+
"epoch": 1.7123287671232876,
|
371 |
+
"grad_norm": 0.46855708956718445,
|
372 |
+
"learning_rate": 0.00019692456209391846,
|
373 |
+
"loss": 0.6782,
|
374 |
+
"step": 250
|
375 |
+
},
|
376 |
+
{
|
377 |
+
"epoch": 1.7465753424657535,
|
378 |
+
"grad_norm": 0.27823150157928467,
|
379 |
+
"learning_rate": 0.0001966234536343253,
|
380 |
+
"loss": 0.665,
|
381 |
+
"step": 255
|
382 |
+
},
|
383 |
+
{
|
384 |
+
"epoch": 1.7808219178082192,
|
385 |
+
"grad_norm": 0.32231074571609497,
|
386 |
+
"learning_rate": 0.00019630853731367713,
|
387 |
+
"loss": 0.6651,
|
388 |
+
"step": 260
|
389 |
+
},
|
390 |
+
{
|
391 |
+
"epoch": 1.8150684931506849,
|
392 |
+
"grad_norm": 0.24578699469566345,
|
393 |
+
"learning_rate": 0.00019597985813472052,
|
394 |
+
"loss": 0.6665,
|
395 |
+
"step": 265
|
396 |
+
},
|
397 |
+
{
|
398 |
+
"epoch": 1.8493150684931505,
|
399 |
+
"grad_norm": 0.31806036829948425,
|
400 |
+
"learning_rate": 0.0001956374630669672,
|
401 |
+
"loss": 0.6672,
|
402 |
+
"step": 270
|
403 |
+
},
|
404 |
+
{
|
405 |
+
"epoch": 1.8835616438356164,
|
406 |
+
"grad_norm": 0.31710395216941833,
|
407 |
+
"learning_rate": 0.00019528140103998177,
|
408 |
+
"loss": 0.6609,
|
409 |
+
"step": 275
|
410 |
+
},
|
411 |
+
{
|
412 |
+
"epoch": 1.9178082191780823,
|
413 |
+
"grad_norm": 0.28619030117988586,
|
414 |
+
"learning_rate": 0.00019491172293638968,
|
415 |
+
"loss": 0.6631,
|
416 |
+
"step": 280
|
417 |
+
},
|
418 |
+
{
|
419 |
+
"epoch": 1.952054794520548,
|
420 |
+
"grad_norm": 0.3997630774974823,
|
421 |
+
"learning_rate": 0.0001945284815846057,
|
422 |
+
"loss": 0.6575,
|
423 |
+
"step": 285
|
424 |
+
},
|
425 |
+
{
|
426 |
+
"epoch": 1.9863013698630136,
|
427 |
+
"grad_norm": 0.30261415243148804,
|
428 |
+
"learning_rate": 0.00019413173175128473,
|
429 |
+
"loss": 0.6581,
|
430 |
+
"step": 290
|
431 |
+
},
|
432 |
+
{
|
433 |
+
"epoch": 2.0,
|
434 |
+
"eval_loss": 1.6853089332580566,
|
435 |
+
"eval_runtime": 0.6813,
|
436 |
+
"eval_samples_per_second": 8.806,
|
437 |
+
"eval_steps_per_second": 1.468,
|
438 |
+
"step": 292
|
439 |
+
},
|
440 |
+
{
|
441 |
+
"epoch": 2.0205479452054793,
|
442 |
+
"grad_norm": 0.2559691071510315,
|
443 |
+
"learning_rate": 0.00019372153013349523,
|
444 |
+
"loss": 0.6464,
|
445 |
+
"step": 295
|
446 |
+
},
|
447 |
+
{
|
448 |
+
"epoch": 2.0547945205479454,
|
449 |
+
"grad_norm": 0.25488296151161194,
|
450 |
+
"learning_rate": 0.00019329793535061723,
|
451 |
+
"loss": 0.6446,
|
452 |
+
"step": 300
|
453 |
+
},
|
454 |
+
{
|
455 |
+
"epoch": 2.089041095890411,
|
456 |
+
"grad_norm": 0.27149608731269836,
|
457 |
+
"learning_rate": 0.0001928610079359652,
|
458 |
+
"loss": 0.6391,
|
459 |
+
"step": 305
|
460 |
+
},
|
461 |
+
{
|
462 |
+
"epoch": 2.1232876712328768,
|
463 |
+
"grad_norm": 0.27013513445854187,
|
464 |
+
"learning_rate": 0.00019241081032813772,
|
465 |
+
"loss": 0.6451,
|
466 |
+
"step": 310
|
467 |
+
},
|
468 |
+
{
|
469 |
+
"epoch": 2.1575342465753424,
|
470 |
+
"grad_norm": 0.287147581577301,
|
471 |
+
"learning_rate": 0.00019194740686209464,
|
472 |
+
"loss": 0.642,
|
473 |
+
"step": 315
|
474 |
+
},
|
475 |
+
{
|
476 |
+
"epoch": 2.191780821917808,
|
477 |
+
"grad_norm": 0.41998761892318726,
|
478 |
+
"learning_rate": 0.0001914708637599636,
|
479 |
+
"loss": 0.6395,
|
480 |
+
"step": 320
|
481 |
+
},
|
482 |
+
{
|
483 |
+
"epoch": 2.2260273972602738,
|
484 |
+
"grad_norm": 0.2858888506889343,
|
485 |
+
"learning_rate": 0.00019098124912157632,
|
486 |
+
"loss": 0.6389,
|
487 |
+
"step": 325
|
488 |
+
},
|
489 |
+
{
|
490 |
+
"epoch": 2.26027397260274,
|
491 |
+
"grad_norm": 0.34377697110176086,
|
492 |
+
"learning_rate": 0.00019047863291473717,
|
493 |
+
"loss": 0.6422,
|
494 |
+
"step": 330
|
495 |
+
},
|
496 |
+
{
|
497 |
+
"epoch": 2.2945205479452055,
|
498 |
+
"grad_norm": 0.2444482445716858,
|
499 |
+
"learning_rate": 0.00018996308696522433,
|
500 |
+
"loss": 0.6355,
|
501 |
+
"step": 335
|
502 |
+
},
|
503 |
+
{
|
504 |
+
"epoch": 2.328767123287671,
|
505 |
+
"grad_norm": 0.308186411857605,
|
506 |
+
"learning_rate": 0.0001894346849465257,
|
507 |
+
"loss": 0.6403,
|
508 |
+
"step": 340
|
509 |
+
},
|
510 |
+
{
|
511 |
+
"epoch": 2.363013698630137,
|
512 |
+
"grad_norm": 0.25512081384658813,
|
513 |
+
"learning_rate": 0.00018889350236931055,
|
514 |
+
"loss": 0.6417,
|
515 |
+
"step": 345
|
516 |
+
},
|
517 |
+
{
|
518 |
+
"epoch": 2.3972602739726026,
|
519 |
+
"grad_norm": 0.3035546839237213,
|
520 |
+
"learning_rate": 0.00018833961657063885,
|
521 |
+
"loss": 0.6366,
|
522 |
+
"step": 350
|
523 |
+
},
|
524 |
+
{
|
525 |
+
"epoch": 2.4315068493150687,
|
526 |
+
"grad_norm": 0.31536582112312317,
|
527 |
+
"learning_rate": 0.0001877731067029096,
|
528 |
+
"loss": 0.6376,
|
529 |
+
"step": 355
|
530 |
+
},
|
531 |
+
{
|
532 |
+
"epoch": 2.4657534246575343,
|
533 |
+
"grad_norm": 0.45909973978996277,
|
534 |
+
"learning_rate": 0.00018719405372254948,
|
535 |
+
"loss": 0.635,
|
536 |
+
"step": 360
|
537 |
+
},
|
538 |
+
{
|
539 |
+
"epoch": 2.5,
|
540 |
+
"grad_norm": 0.39519354701042175,
|
541 |
+
"learning_rate": 0.00018660254037844388,
|
542 |
+
"loss": 0.6384,
|
543 |
+
"step": 365
|
544 |
+
},
|
545 |
+
{
|
546 |
+
"epoch": 2.5342465753424657,
|
547 |
+
"grad_norm": 0.29281747341156006,
|
548 |
+
"learning_rate": 0.00018599865120011192,
|
549 |
+
"loss": 0.6316,
|
550 |
+
"step": 370
|
551 |
+
},
|
552 |
+
{
|
553 |
+
"epoch": 2.5684931506849313,
|
554 |
+
"grad_norm": 0.24112151563167572,
|
555 |
+
"learning_rate": 0.00018538247248562674,
|
556 |
+
"loss": 0.6363,
|
557 |
+
"step": 375
|
558 |
+
},
|
559 |
+
{
|
560 |
+
"epoch": 2.602739726027397,
|
561 |
+
"grad_norm": 0.2908940017223358,
|
562 |
+
"learning_rate": 0.00018475409228928312,
|
563 |
+
"loss": 0.6346,
|
564 |
+
"step": 380
|
565 |
+
},
|
566 |
+
{
|
567 |
+
"epoch": 2.636986301369863,
|
568 |
+
"grad_norm": 0.30711865425109863,
|
569 |
+
"learning_rate": 0.0001841136004090144,
|
570 |
+
"loss": 0.6453,
|
571 |
+
"step": 385
|
572 |
+
},
|
573 |
+
{
|
574 |
+
"epoch": 2.671232876712329,
|
575 |
+
"grad_norm": 0.24229027330875397,
|
576 |
+
"learning_rate": 0.00018346108837355972,
|
577 |
+
"loss": 0.6373,
|
578 |
+
"step": 390
|
579 |
+
},
|
580 |
+
{
|
581 |
+
"epoch": 2.7054794520547945,
|
582 |
+
"grad_norm": 0.28023508191108704,
|
583 |
+
"learning_rate": 0.00018279664942938447,
|
584 |
+
"loss": 0.6366,
|
585 |
+
"step": 395
|
586 |
+
},
|
587 |
+
{
|
588 |
+
"epoch": 2.73972602739726,
|
589 |
+
"grad_norm": 0.36763280630111694,
|
590 |
+
"learning_rate": 0.00018212037852735486,
|
591 |
+
"loss": 0.6461,
|
592 |
+
"step": 400
|
593 |
+
},
|
594 |
+
{
|
595 |
+
"epoch": 2.7739726027397262,
|
596 |
+
"grad_norm": 0.3579126000404358,
|
597 |
+
"learning_rate": 0.0001814323723091692,
|
598 |
+
"loss": 0.6433,
|
599 |
+
"step": 405
|
600 |
+
},
|
601 |
+
{
|
602 |
+
"epoch": 2.808219178082192,
|
603 |
+
"grad_norm": 0.30816444754600525,
|
604 |
+
"learning_rate": 0.00018073272909354727,
|
605 |
+
"loss": 0.6319,
|
606 |
+
"step": 410
|
607 |
+
},
|
608 |
+
{
|
609 |
+
"epoch": 2.8424657534246576,
|
610 |
+
"grad_norm": 0.4557536542415619,
|
611 |
+
"learning_rate": 0.00018002154886218033,
|
612 |
+
"loss": 0.6384,
|
613 |
+
"step": 415
|
614 |
+
},
|
615 |
+
{
|
616 |
+
"epoch": 2.8767123287671232,
|
617 |
+
"grad_norm": 0.4224196672439575,
|
618 |
+
"learning_rate": 0.00017929893324544332,
|
619 |
+
"loss": 0.6355,
|
620 |
+
"step": 420
|
621 |
+
},
|
622 |
+
{
|
623 |
+
"epoch": 2.910958904109589,
|
624 |
+
"grad_norm": 0.285664826631546,
|
625 |
+
"learning_rate": 0.00017856498550787144,
|
626 |
+
"loss": 0.6408,
|
627 |
+
"step": 425
|
628 |
+
},
|
629 |
+
{
|
630 |
+
"epoch": 2.9452054794520546,
|
631 |
+
"grad_norm": 0.29482561349868774,
|
632 |
+
"learning_rate": 0.00017781981053340337,
|
633 |
+
"loss": 0.6351,
|
634 |
+
"step": 430
|
635 |
+
},
|
636 |
+
{
|
637 |
+
"epoch": 2.9794520547945207,
|
638 |
+
"grad_norm": 0.31438663601875305,
|
639 |
+
"learning_rate": 0.00017706351481039284,
|
640 |
+
"loss": 0.6323,
|
641 |
+
"step": 435
|
642 |
+
},
|
643 |
+
{
|
644 |
+
"epoch": 3.0,
|
645 |
+
"eval_loss": 1.6891189813613892,
|
646 |
+
"eval_runtime": 0.6806,
|
647 |
+
"eval_samples_per_second": 8.815,
|
648 |
+
"eval_steps_per_second": 1.469,
|
649 |
+
"step": 438
|
650 |
+
},
|
651 |
+
{
|
652 |
+
"epoch": 3.0136986301369864,
|
653 |
+
"grad_norm": 0.235789492726326,
|
654 |
+
"learning_rate": 0.00017629620641639103,
|
655 |
+
"loss": 0.6314,
|
656 |
+
"step": 440
|
657 |
+
},
|
658 |
+
{
|
659 |
+
"epoch": 3.047945205479452,
|
660 |
+
"grad_norm": 0.2618357241153717,
|
661 |
+
"learning_rate": 0.00017551799500270198,
|
662 |
+
"loss": 0.6064,
|
663 |
+
"step": 445
|
664 |
+
},
|
665 |
+
{
|
666 |
+
"epoch": 3.0821917808219177,
|
667 |
+
"grad_norm": 0.24724790453910828,
|
668 |
+
"learning_rate": 0.00017472899177871297,
|
669 |
+
"loss": 0.6096,
|
670 |
+
"step": 450
|
671 |
+
},
|
672 |
+
{
|
673 |
+
"epoch": 3.1164383561643834,
|
674 |
+
"grad_norm": 0.25246715545654297,
|
675 |
+
"learning_rate": 0.00017392930949600217,
|
676 |
+
"loss": 0.6088,
|
677 |
+
"step": 455
|
678 |
+
},
|
679 |
+
{
|
680 |
+
"epoch": 3.1506849315068495,
|
681 |
+
"grad_norm": 0.24573026597499847,
|
682 |
+
"learning_rate": 0.00017311906243222614,
|
683 |
+
"loss": 0.6184,
|
684 |
+
"step": 460
|
685 |
+
},
|
686 |
+
{
|
687 |
+
"epoch": 3.184931506849315,
|
688 |
+
"grad_norm": 0.2606354057788849,
|
689 |
+
"learning_rate": 0.00017229836637478902,
|
690 |
+
"loss": 0.6113,
|
691 |
+
"step": 465
|
692 |
+
},
|
693 |
+
{
|
694 |
+
"epoch": 3.219178082191781,
|
695 |
+
"grad_norm": 0.3043690025806427,
|
696 |
+
"learning_rate": 0.00017146733860429612,
|
697 |
+
"loss": 0.6092,
|
698 |
+
"step": 470
|
699 |
+
},
|
700 |
+
{
|
701 |
+
"epoch": 3.2534246575342465,
|
702 |
+
"grad_norm": 0.26179400086402893,
|
703 |
+
"learning_rate": 0.00017062609787779403,
|
704 |
+
"loss": 0.6173,
|
705 |
+
"step": 475
|
706 |
+
},
|
707 |
+
{
|
708 |
+
"epoch": 3.287671232876712,
|
709 |
+
"grad_norm": 0.2980283498764038,
|
710 |
+
"learning_rate": 0.00016977476441179992,
|
711 |
+
"loss": 0.6173,
|
712 |
+
"step": 480
|
713 |
+
},
|
714 |
+
{
|
715 |
+
"epoch": 3.3219178082191783,
|
716 |
+
"grad_norm": 0.2769377529621124,
|
717 |
+
"learning_rate": 0.0001689134598651219,
|
718 |
+
"loss": 0.6181,
|
719 |
+
"step": 485
|
720 |
+
},
|
721 |
+
{
|
722 |
+
"epoch": 3.356164383561644,
|
723 |
+
"grad_norm": 0.31351810693740845,
|
724 |
+
"learning_rate": 0.0001680423073214737,
|
725 |
+
"loss": 0.6223,
|
726 |
+
"step": 490
|
727 |
+
},
|
728 |
+
{
|
729 |
+
"epoch": 3.3904109589041096,
|
730 |
+
"grad_norm": 0.2710462510585785,
|
731 |
+
"learning_rate": 0.00016716143127188548,
|
732 |
+
"loss": 0.6171,
|
733 |
+
"step": 495
|
734 |
+
},
|
735 |
+
{
|
736 |
+
"epoch": 3.4246575342465753,
|
737 |
+
"grad_norm": 0.2538503110408783,
|
738 |
+
"learning_rate": 0.00016627095759691362,
|
739 |
+
"loss": 0.6132,
|
740 |
+
"step": 500
|
741 |
+
},
|
742 |
+
{
|
743 |
+
"epoch": 3.458904109589041,
|
744 |
+
"grad_norm": 0.2977152168750763,
|
745 |
+
"learning_rate": 0.0001653710135486518,
|
746 |
+
"loss": 0.613,
|
747 |
+
"step": 505
|
748 |
+
},
|
749 |
+
{
|
750 |
+
"epoch": 3.493150684931507,
|
751 |
+
"grad_norm": 0.2564401626586914,
|
752 |
+
"learning_rate": 0.00016446172773254629,
|
753 |
+
"loss": 0.6124,
|
754 |
+
"step": 510
|
755 |
+
},
|
756 |
+
{
|
757 |
+
"epoch": 3.5273972602739727,
|
758 |
+
"grad_norm": 0.2717874050140381,
|
759 |
+
"learning_rate": 0.00016354323008901776,
|
760 |
+
"loss": 0.6104,
|
761 |
+
"step": 515
|
762 |
+
},
|
763 |
+
{
|
764 |
+
"epoch": 3.5616438356164384,
|
765 |
+
"grad_norm": 0.3490482270717621,
|
766 |
+
"learning_rate": 0.0001626156518748922,
|
767 |
+
"loss": 0.625,
|
768 |
+
"step": 520
|
769 |
+
},
|
770 |
+
{
|
771 |
+
"epoch": 3.595890410958904,
|
772 |
+
"grad_norm": 0.2741421163082123,
|
773 |
+
"learning_rate": 0.00016167912564464383,
|
774 |
+
"loss": 0.6094,
|
775 |
+
"step": 525
|
776 |
+
},
|
777 |
+
{
|
778 |
+
"epoch": 3.6301369863013697,
|
779 |
+
"grad_norm": 0.269045352935791,
|
780 |
+
"learning_rate": 0.0001607337852314527,
|
781 |
+
"loss": 0.6053,
|
782 |
+
"step": 530
|
783 |
+
},
|
784 |
+
{
|
785 |
+
"epoch": 3.6643835616438354,
|
786 |
+
"grad_norm": 0.30331557989120483,
|
787 |
+
"learning_rate": 0.0001597797657280792,
|
788 |
+
"loss": 0.6173,
|
789 |
+
"step": 535
|
790 |
+
},
|
791 |
+
{
|
792 |
+
"epoch": 3.6986301369863015,
|
793 |
+
"grad_norm": 0.27356013655662537,
|
794 |
+
"learning_rate": 0.00015881720346755905,
|
795 |
+
"loss": 0.6095,
|
796 |
+
"step": 540
|
797 |
+
},
|
798 |
+
{
|
799 |
+
"epoch": 3.732876712328767,
|
800 |
+
"grad_norm": 0.2557332515716553,
|
801 |
+
"learning_rate": 0.00015784623600372042,
|
802 |
+
"loss": 0.6128,
|
803 |
+
"step": 545
|
804 |
+
},
|
805 |
+
{
|
806 |
+
"epoch": 3.767123287671233,
|
807 |
+
"grad_norm": 0.26268497109413147,
|
808 |
+
"learning_rate": 0.00015686700209152738,
|
809 |
+
"loss": 0.614,
|
810 |
+
"step": 550
|
811 |
+
},
|
812 |
+
{
|
813 |
+
"epoch": 3.8013698630136985,
|
814 |
+
"grad_norm": 0.240408256649971,
|
815 |
+
"learning_rate": 0.00015587964166725095,
|
816 |
+
"loss": 0.62,
|
817 |
+
"step": 555
|
818 |
+
},
|
819 |
+
{
|
820 |
+
"epoch": 3.8356164383561646,
|
821 |
+
"grad_norm": 0.2302975207567215,
|
822 |
+
"learning_rate": 0.00015488429582847192,
|
823 |
+
"loss": 0.6076,
|
824 |
+
"step": 560
|
825 |
+
},
|
826 |
+
{
|
827 |
+
"epoch": 3.8698630136986303,
|
828 |
+
"grad_norm": 0.24237115681171417,
|
829 |
+
"learning_rate": 0.00015388110681391725,
|
830 |
+
"loss": 0.6193,
|
831 |
+
"step": 565
|
832 |
+
},
|
833 |
+
{
|
834 |
+
"epoch": 3.904109589041096,
|
835 |
+
"grad_norm": 0.23819388449192047,
|
836 |
+
"learning_rate": 0.0001528702179831338,
|
837 |
+
"loss": 0.6047,
|
838 |
+
"step": 570
|
839 |
+
},
|
840 |
+
{
|
841 |
+
"epoch": 3.9383561643835616,
|
842 |
+
"grad_norm": 0.2723640203475952,
|
843 |
+
"learning_rate": 0.00015185177379600152,
|
844 |
+
"loss": 0.611,
|
845 |
+
"step": 575
|
846 |
+
},
|
847 |
+
{
|
848 |
+
"epoch": 3.9726027397260273,
|
849 |
+
"grad_norm": 0.24371425807476044,
|
850 |
+
"learning_rate": 0.00015082591979208976,
|
851 |
+
"loss": 0.6067,
|
852 |
+
"step": 580
|
853 |
+
},
|
854 |
+
{
|
855 |
+
"epoch": 4.0,
|
856 |
+
"eval_loss": 1.7035120725631714,
|
857 |
+
"eval_runtime": 0.6811,
|
858 |
+
"eval_samples_per_second": 8.81,
|
859 |
+
"eval_steps_per_second": 1.468,
|
860 |
+
"step": 584
|
861 |
+
},
|
862 |
+
{
|
863 |
+
"epoch": 4.006849315068493,
|
864 |
+
"grad_norm": 0.28401726484298706,
|
865 |
+
"learning_rate": 0.000149792802569859,
|
866 |
+
"loss": 0.6113,
|
867 |
+
"step": 585
|
868 |
+
},
|
869 |
+
{
|
870 |
+
"epoch": 4.041095890410959,
|
871 |
+
"grad_norm": 0.2556915581226349,
|
872 |
+
"learning_rate": 0.00014875256976571135,
|
873 |
+
"loss": 0.5917,
|
874 |
+
"step": 590
|
875 |
+
},
|
876 |
+
{
|
877 |
+
"epoch": 4.075342465753424,
|
878 |
+
"grad_norm": 0.2777804434299469,
|
879 |
+
"learning_rate": 0.0001477053700328929,
|
880 |
+
"loss": 0.591,
|
881 |
+
"step": 595
|
882 |
+
},
|
883 |
+
{
|
884 |
+
"epoch": 4.109589041095891,
|
885 |
+
"grad_norm": 0.24701425433158875,
|
886 |
+
"learning_rate": 0.00014665135302025035,
|
887 |
+
"loss": 0.59,
|
888 |
+
"step": 600
|
889 |
+
},
|
890 |
+
{
|
891 |
+
"epoch": 4.1438356164383565,
|
892 |
+
"grad_norm": 0.24922576546669006,
|
893 |
+
"learning_rate": 0.00014559066935084588,
|
894 |
+
"loss": 0.5878,
|
895 |
+
"step": 605
|
896 |
+
},
|
897 |
+
{
|
898 |
+
"epoch": 4.178082191780822,
|
899 |
+
"grad_norm": 0.2835874557495117,
|
900 |
+
"learning_rate": 0.00014452347060043237,
|
901 |
+
"loss": 0.5827,
|
902 |
+
"step": 610
|
903 |
+
},
|
904 |
+
{
|
905 |
+
"epoch": 4.212328767123288,
|
906 |
+
"grad_norm": 0.2816501557826996,
|
907 |
+
"learning_rate": 0.00014344990927579268,
|
908 |
+
"loss": 0.5902,
|
909 |
+
"step": 615
|
910 |
+
},
|
911 |
+
{
|
912 |
+
"epoch": 4.2465753424657535,
|
913 |
+
"grad_norm": 0.27082914113998413,
|
914 |
+
"learning_rate": 0.0001423701387929459,
|
915 |
+
"loss": 0.5943,
|
916 |
+
"step": 620
|
917 |
+
},
|
918 |
+
{
|
919 |
+
"epoch": 4.280821917808219,
|
920 |
+
"grad_norm": 0.26985329389572144,
|
921 |
+
"learning_rate": 0.0001412843134552235,
|
922 |
+
"loss": 0.601,
|
923 |
+
"step": 625
|
924 |
+
},
|
925 |
+
{
|
926 |
+
"epoch": 4.315068493150685,
|
927 |
+
"grad_norm": 0.2813054621219635,
|
928 |
+
"learning_rate": 0.00014019258843121893,
|
929 |
+
"loss": 0.5896,
|
930 |
+
"step": 630
|
931 |
+
},
|
932 |
+
{
|
933 |
+
"epoch": 4.3493150684931505,
|
934 |
+
"grad_norm": 0.24405089020729065,
|
935 |
+
"learning_rate": 0.0001390951197326134,
|
936 |
+
"loss": 0.591,
|
937 |
+
"step": 635
|
938 |
+
},
|
939 |
+
{
|
940 |
+
"epoch": 4.383561643835616,
|
941 |
+
"grad_norm": 0.2506767511367798,
|
942 |
+
"learning_rate": 0.00013799206419188103,
|
943 |
+
"loss": 0.5957,
|
944 |
+
"step": 640
|
945 |
+
},
|
946 |
+
{
|
947 |
+
"epoch": 4.417808219178082,
|
948 |
+
"grad_norm": 0.28518110513687134,
|
949 |
+
"learning_rate": 0.00013688357943987732,
|
950 |
+
"loss": 0.5847,
|
951 |
+
"step": 645
|
952 |
+
},
|
953 |
+
{
|
954 |
+
"epoch": 4.4520547945205475,
|
955 |
+
"grad_norm": 0.25432756543159485,
|
956 |
+
"learning_rate": 0.0001357698238833126,
|
957 |
+
"loss": 0.5985,
|
958 |
+
"step": 650
|
959 |
+
},
|
960 |
+
{
|
961 |
+
"epoch": 4.486301369863014,
|
962 |
+
"grad_norm": 0.24484550952911377,
|
963 |
+
"learning_rate": 0.0001346509566821153,
|
964 |
+
"loss": 0.5928,
|
965 |
+
"step": 655
|
966 |
+
},
|
967 |
+
{
|
968 |
+
"epoch": 4.52054794520548,
|
969 |
+
"grad_norm": 0.2853868305683136,
|
970 |
+
"learning_rate": 0.00013352713772668765,
|
971 |
+
"loss": 0.5968,
|
972 |
+
"step": 660
|
973 |
+
},
|
974 |
+
{
|
975 |
+
"epoch": 4.554794520547945,
|
976 |
+
"grad_norm": 0.25753411650657654,
|
977 |
+
"learning_rate": 0.00013239852761505626,
|
978 |
+
"loss": 0.5945,
|
979 |
+
"step": 665
|
980 |
+
},
|
981 |
+
{
|
982 |
+
"epoch": 4.589041095890411,
|
983 |
+
"grad_norm": 0.2740824818611145,
|
984 |
+
"learning_rate": 0.00013126528762992247,
|
985 |
+
"loss": 0.5944,
|
986 |
+
"step": 670
|
987 |
+
},
|
988 |
+
{
|
989 |
+
"epoch": 4.623287671232877,
|
990 |
+
"grad_norm": 0.24608436226844788,
|
991 |
+
"learning_rate": 0.00013012757971561415,
|
992 |
+
"loss": 0.596,
|
993 |
+
"step": 675
|
994 |
+
},
|
995 |
+
{
|
996 |
+
"epoch": 4.657534246575342,
|
997 |
+
"grad_norm": 0.24703024327754974,
|
998 |
+
"learning_rate": 0.00012898556645494325,
|
999 |
+
"loss": 0.6013,
|
1000 |
+
"step": 680
|
1001 |
+
},
|
1002 |
+
{
|
1003 |
+
"epoch": 4.691780821917808,
|
1004 |
+
"grad_norm": 0.26772698760032654,
|
1005 |
+
"learning_rate": 0.0001278394110459724,
|
1006 |
+
"loss": 0.5945,
|
1007 |
+
"step": 685
|
1008 |
+
},
|
1009 |
+
{
|
1010 |
+
"epoch": 4.726027397260274,
|
1011 |
+
"grad_norm": 0.24686093628406525,
|
1012 |
+
"learning_rate": 0.0001266892772786929,
|
1013 |
+
"loss": 0.5904,
|
1014 |
+
"step": 690
|
1015 |
+
},
|
1016 |
+
{
|
1017 |
+
"epoch": 4.760273972602739,
|
1018 |
+
"grad_norm": 0.27200302481651306,
|
1019 |
+
"learning_rate": 0.0001255353295116187,
|
1020 |
+
"loss": 0.5915,
|
1021 |
+
"step": 695
|
1022 |
+
},
|
1023 |
+
{
|
1024 |
+
"epoch": 4.794520547945205,
|
1025 |
+
"grad_norm": 0.2462131232023239,
|
1026 |
+
"learning_rate": 0.00012437773264829897,
|
1027 |
+
"loss": 0.5896,
|
1028 |
+
"step": 700
|
1029 |
+
},
|
1030 |
+
{
|
1031 |
+
"epoch": 4.828767123287671,
|
1032 |
+
"grad_norm": 0.25482824444770813,
|
1033 |
+
"learning_rate": 0.00012321665211375256,
|
1034 |
+
"loss": 0.5921,
|
1035 |
+
"step": 705
|
1036 |
+
},
|
1037 |
+
{
|
1038 |
+
"epoch": 4.863013698630137,
|
1039 |
+
"grad_norm": 0.2556120455265045,
|
1040 |
+
"learning_rate": 0.00012205225383082843,
|
1041 |
+
"loss": 0.5944,
|
1042 |
+
"step": 710
|
1043 |
+
},
|
1044 |
+
{
|
1045 |
+
"epoch": 4.897260273972603,
|
1046 |
+
"grad_norm": 0.2709575593471527,
|
1047 |
+
"learning_rate": 0.00012088470419649432,
|
1048 |
+
"loss": 0.5975,
|
1049 |
+
"step": 715
|
1050 |
+
},
|
1051 |
+
{
|
1052 |
+
"epoch": 4.931506849315069,
|
1053 |
+
"grad_norm": 0.29646211862564087,
|
1054 |
+
"learning_rate": 0.00011971417005805818,
|
1055 |
+
"loss": 0.6007,
|
1056 |
+
"step": 720
|
1057 |
+
},
|
1058 |
+
{
|
1059 |
+
"epoch": 4.965753424657534,
|
1060 |
+
"grad_norm": 0.2482611984014511,
|
1061 |
+
"learning_rate": 0.0001185408186893251,
|
1062 |
+
"loss": 0.5927,
|
1063 |
+
"step": 725
|
1064 |
+
},
|
1065 |
+
{
|
1066 |
+
"epoch": 5.0,
|
1067 |
+
"grad_norm": 0.24462343752384186,
|
1068 |
+
"learning_rate": 0.00011736481776669306,
|
1069 |
+
"loss": 0.591,
|
1070 |
+
"step": 730
|
1071 |
+
},
|
1072 |
+
{
|
1073 |
+
"epoch": 5.0,
|
1074 |
+
"eval_loss": 1.7225574254989624,
|
1075 |
+
"eval_runtime": 0.6804,
|
1076 |
+
"eval_samples_per_second": 8.818,
|
1077 |
+
"eval_steps_per_second": 1.47,
|
1078 |
+
"step": 730
|
1079 |
+
},
|
1080 |
+
{
|
1081 |
+
"epoch": 5.034246575342466,
|
1082 |
+
"grad_norm": 0.25259608030319214,
|
1083 |
+
"learning_rate": 0.00011618633534519141,
|
1084 |
+
"loss": 0.5762,
|
1085 |
+
"step": 735
|
1086 |
+
},
|
1087 |
+
{
|
1088 |
+
"epoch": 5.068493150684931,
|
1089 |
+
"grad_norm": 0.24844078719615936,
|
1090 |
+
"learning_rate": 0.00011500553983446527,
|
1091 |
+
"loss": 0.5762,
|
1092 |
+
"step": 740
|
1093 |
+
},
|
1094 |
+
{
|
1095 |
+
"epoch": 5.102739726027397,
|
1096 |
+
"grad_norm": 0.2850824296474457,
|
1097 |
+
"learning_rate": 0.00011382259997470899,
|
1098 |
+
"loss": 0.5744,
|
1099 |
+
"step": 745
|
1100 |
+
},
|
1101 |
+
{
|
1102 |
+
"epoch": 5.136986301369863,
|
1103 |
+
"grad_norm": 0.27126604318618774,
|
1104 |
+
"learning_rate": 0.00011263768481255264,
|
1105 |
+
"loss": 0.5713,
|
1106 |
+
"step": 750
|
1107 |
+
},
|
1108 |
+
{
|
1109 |
+
"epoch": 5.171232876712328,
|
1110 |
+
"grad_norm": 0.25230973958969116,
|
1111 |
+
"learning_rate": 0.00011145096367690444,
|
1112 |
+
"loss": 0.5755,
|
1113 |
+
"step": 755
|
1114 |
+
},
|
1115 |
+
{
|
1116 |
+
"epoch": 5.205479452054795,
|
1117 |
+
"grad_norm": 0.24937325716018677,
|
1118 |
+
"learning_rate": 0.00011026260615475333,
|
1119 |
+
"loss": 0.5752,
|
1120 |
+
"step": 760
|
1121 |
+
},
|
1122 |
+
{
|
1123 |
+
"epoch": 5.239726027397261,
|
1124 |
+
"grad_norm": 0.26235413551330566,
|
1125 |
+
"learning_rate": 0.00010907278206693395,
|
1126 |
+
"loss": 0.5793,
|
1127 |
+
"step": 765
|
1128 |
+
},
|
1129 |
+
{
|
1130 |
+
"epoch": 5.273972602739726,
|
1131 |
+
"grad_norm": 0.2632244825363159,
|
1132 |
+
"learning_rate": 0.00010788166144385888,
|
1133 |
+
"loss": 0.576,
|
1134 |
+
"step": 770
|
1135 |
+
},
|
1136 |
+
{
|
1137 |
+
"epoch": 5.308219178082192,
|
1138 |
+
"grad_norm": 0.2701359689235687,
|
1139 |
+
"learning_rate": 0.00010668941450122055,
|
1140 |
+
"loss": 0.5812,
|
1141 |
+
"step": 775
|
1142 |
+
},
|
1143 |
+
{
|
1144 |
+
"epoch": 5.342465753424658,
|
1145 |
+
"grad_norm": 0.2537064552307129,
|
1146 |
+
"learning_rate": 0.0001054962116156667,
|
1147 |
+
"loss": 0.5754,
|
1148 |
+
"step": 780
|
1149 |
+
},
|
1150 |
+
{
|
1151 |
+
"epoch": 5.376712328767123,
|
1152 |
+
"grad_norm": 0.2470828741788864,
|
1153 |
+
"learning_rate": 0.00010430222330045304,
|
1154 |
+
"loss": 0.5707,
|
1155 |
+
"step": 785
|
1156 |
+
},
|
1157 |
+
{
|
1158 |
+
"epoch": 5.410958904109589,
|
1159 |
+
"grad_norm": 0.2509852945804596,
|
1160 |
+
"learning_rate": 0.0001031076201810762,
|
1161 |
+
"loss": 0.5762,
|
1162 |
+
"step": 790
|
1163 |
+
},
|
1164 |
+
{
|
1165 |
+
"epoch": 5.445205479452055,
|
1166 |
+
"grad_norm": 0.2670098543167114,
|
1167 |
+
"learning_rate": 0.00010191257297089052,
|
1168 |
+
"loss": 0.5763,
|
1169 |
+
"step": 795
|
1170 |
+
},
|
1171 |
+
{
|
1172 |
+
"epoch": 5.47945205479452,
|
1173 |
+
"grad_norm": 0.25170961022377014,
|
1174 |
+
"learning_rate": 0.00010071725244671282,
|
1175 |
+
"loss": 0.5704,
|
1176 |
+
"step": 800
|
1177 |
+
},
|
1178 |
+
{
|
1179 |
+
"epoch": 5.513698630136986,
|
1180 |
+
"grad_norm": 0.2560670077800751,
|
1181 |
+
"learning_rate": 9.952182942441733e-05,
|
1182 |
+
"loss": 0.5796,
|
1183 |
+
"step": 805
|
1184 |
+
},
|
1185 |
+
{
|
1186 |
+
"epoch": 5.5479452054794525,
|
1187 |
+
"grad_norm": 0.25187042355537415,
|
1188 |
+
"learning_rate": 9.83264747345259e-05,
|
1189 |
+
"loss": 0.5722,
|
1190 |
+
"step": 810
|
1191 |
+
},
|
1192 |
+
{
|
1193 |
+
"epoch": 5.582191780821918,
|
1194 |
+
"grad_norm": 0.2625672519207001,
|
1195 |
+
"learning_rate": 9.713135919779515e-05,
|
1196 |
+
"loss": 0.5762,
|
1197 |
+
"step": 815
|
1198 |
+
},
|
1199 |
+
{
|
1200 |
+
"epoch": 5.616438356164384,
|
1201 |
+
"grad_norm": 0.25329843163490295,
|
1202 |
+
"learning_rate": 9.593665360080599e-05,
|
1203 |
+
"loss": 0.5744,
|
1204 |
+
"step": 820
|
1205 |
+
},
|
1206 |
+
{
|
1207 |
+
"epoch": 5.6506849315068495,
|
1208 |
+
"grad_norm": 0.25838255882263184,
|
1209 |
+
"learning_rate": 9.474252867155732e-05,
|
1210 |
+
"loss": 0.5759,
|
1211 |
+
"step": 825
|
1212 |
+
},
|
1213 |
+
{
|
1214 |
+
"epoch": 5.684931506849315,
|
1215 |
+
"grad_norm": 0.28136056661605835,
|
1216 |
+
"learning_rate": 9.354915505506839e-05,
|
1217 |
+
"loss": 0.5768,
|
1218 |
+
"step": 830
|
1219 |
+
},
|
1220 |
+
{
|
1221 |
+
"epoch": 5.719178082191781,
|
1222 |
+
"grad_norm": 0.24633042514324188,
|
1223 |
+
"learning_rate": 9.235670328899293e-05,
|
1224 |
+
"loss": 0.5794,
|
1225 |
+
"step": 835
|
1226 |
+
},
|
1227 |
+
{
|
1228 |
+
"epoch": 5.7534246575342465,
|
1229 |
+
"grad_norm": 0.2591402232646942,
|
1230 |
+
"learning_rate": 9.116534377924883e-05,
|
1231 |
+
"loss": 0.578,
|
1232 |
+
"step": 840
|
1233 |
+
},
|
1234 |
+
{
|
1235 |
+
"epoch": 5.787671232876712,
|
1236 |
+
"grad_norm": 0.27811384201049805,
|
1237 |
+
"learning_rate": 8.997524677566627e-05,
|
1238 |
+
"loss": 0.5818,
|
1239 |
+
"step": 845
|
1240 |
+
},
|
1241 |
+
{
|
1242 |
+
"epoch": 5.821917808219178,
|
1243 |
+
"grad_norm": 0.2641887962818146,
|
1244 |
+
"learning_rate": 8.878658234765858e-05,
|
1245 |
+
"loss": 0.5756,
|
1246 |
+
"step": 850
|
1247 |
+
},
|
1248 |
+
{
|
1249 |
+
"epoch": 5.8561643835616435,
|
1250 |
+
"grad_norm": 0.26102420687675476,
|
1251 |
+
"learning_rate": 8.759952035991844e-05,
|
1252 |
+
"loss": 0.5817,
|
1253 |
+
"step": 855
|
1254 |
+
},
|
1255 |
+
{
|
1256 |
+
"epoch": 5.890410958904109,
|
1257 |
+
"grad_norm": 0.25208574533462524,
|
1258 |
+
"learning_rate": 8.641423044814374e-05,
|
1259 |
+
"loss": 0.5792,
|
1260 |
+
"step": 860
|
1261 |
+
},
|
1262 |
+
{
|
1263 |
+
"epoch": 5.924657534246576,
|
1264 |
+
"grad_norm": 0.2486419528722763,
|
1265 |
+
"learning_rate": 8.5230881994796e-05,
|
1266 |
+
"loss": 0.5743,
|
1267 |
+
"step": 865
|
1268 |
+
},
|
1269 |
+
{
|
1270 |
+
"epoch": 5.958904109589041,
|
1271 |
+
"grad_norm": 0.2543000280857086,
|
1272 |
+
"learning_rate": 8.404964410489485e-05,
|
1273 |
+
"loss": 0.575,
|
1274 |
+
"step": 870
|
1275 |
+
},
|
1276 |
+
{
|
1277 |
+
"epoch": 5.993150684931507,
|
1278 |
+
"grad_norm": 0.2586722671985626,
|
1279 |
+
"learning_rate": 8.287068558185225e-05,
|
1280 |
+
"loss": 0.5752,
|
1281 |
+
"step": 875
|
1282 |
+
},
|
1283 |
+
{
|
1284 |
+
"epoch": 6.0,
|
1285 |
+
"eval_loss": 1.725889801979065,
|
1286 |
+
"eval_runtime": 0.6823,
|
1287 |
+
"eval_samples_per_second": 8.794,
|
1288 |
+
"eval_steps_per_second": 1.466,
|
1289 |
+
"step": 876
|
1290 |
+
},
|
1291 |
+
{
|
1292 |
+
"epoch": 6.027397260273973,
|
1293 |
+
"grad_norm": 0.2684203088283539,
|
1294 |
+
"learning_rate": 8.169417490335007e-05,
|
1295 |
+
"loss": 0.5647,
|
1296 |
+
"step": 880
|
1297 |
+
},
|
1298 |
+
{
|
1299 |
+
"epoch": 6.061643835616438,
|
1300 |
+
"grad_norm": 0.2663749158382416,
|
1301 |
+
"learning_rate": 8.052028019726371e-05,
|
1302 |
+
"loss": 0.5576,
|
1303 |
+
"step": 885
|
1304 |
+
},
|
1305 |
+
{
|
1306 |
+
"epoch": 6.095890410958904,
|
1307 |
+
"grad_norm": 0.2585398256778717,
|
1308 |
+
"learning_rate": 7.934916921763628e-05,
|
1309 |
+
"loss": 0.5512,
|
1310 |
+
"step": 890
|
1311 |
+
},
|
1312 |
+
{
|
1313 |
+
"epoch": 6.13013698630137,
|
1314 |
+
"grad_norm": 0.2593580484390259,
|
1315 |
+
"learning_rate": 7.818100932070546e-05,
|
1316 |
+
"loss": 0.5513,
|
1317 |
+
"step": 895
|
1318 |
+
},
|
1319 |
+
{
|
1320 |
+
"epoch": 6.164383561643835,
|
1321 |
+
"grad_norm": 0.2640758454799652,
|
1322 |
+
"learning_rate": 7.701596744098818e-05,
|
1323 |
+
"loss": 0.561,
|
1324 |
+
"step": 900
|
1325 |
+
},
|
1326 |
+
{
|
1327 |
+
"epoch": 6.198630136986301,
|
1328 |
+
"grad_norm": 0.29837489128112793,
|
1329 |
+
"learning_rate": 7.585421006742463e-05,
|
1330 |
+
"loss": 0.5576,
|
1331 |
+
"step": 905
|
1332 |
+
},
|
1333 |
+
{
|
1334 |
+
"epoch": 6.232876712328767,
|
1335 |
+
"grad_norm": 0.2711084485054016,
|
1336 |
+
"learning_rate": 7.469590321958662e-05,
|
1337 |
+
"loss": 0.5632,
|
1338 |
+
"step": 910
|
1339 |
+
},
|
1340 |
+
{
|
1341 |
+
"epoch": 6.267123287671233,
|
1342 |
+
"grad_norm": 0.27940109372138977,
|
1343 |
+
"learning_rate": 7.354121242395254e-05,
|
1344 |
+
"loss": 0.5597,
|
1345 |
+
"step": 915
|
1346 |
+
},
|
1347 |
+
{
|
1348 |
+
"epoch": 6.301369863013699,
|
1349 |
+
"grad_norm": 0.2598007321357727,
|
1350 |
+
"learning_rate": 7.239030269025311e-05,
|
1351 |
+
"loss": 0.5634,
|
1352 |
+
"step": 920
|
1353 |
+
},
|
1354 |
+
{
|
1355 |
+
"epoch": 6.335616438356165,
|
1356 |
+
"grad_norm": 0.27142032980918884,
|
1357 |
+
"learning_rate": 7.124333848789091e-05,
|
1358 |
+
"loss": 0.5684,
|
1359 |
+
"step": 925
|
1360 |
+
},
|
1361 |
+
{
|
1362 |
+
"epoch": 6.36986301369863,
|
1363 |
+
"grad_norm": 0.2744521200656891,
|
1364 |
+
"learning_rate": 7.010048372243698e-05,
|
1365 |
+
"loss": 0.566,
|
1366 |
+
"step": 930
|
1367 |
+
},
|
1368 |
+
{
|
1369 |
+
"epoch": 6.404109589041096,
|
1370 |
+
"grad_norm": 0.2650752067565918,
|
1371 |
+
"learning_rate": 6.8961901712208e-05,
|
1372 |
+
"loss": 0.5633,
|
1373 |
+
"step": 935
|
1374 |
+
},
|
1375 |
+
{
|
1376 |
+
"epoch": 6.438356164383562,
|
1377 |
+
"grad_norm": 0.2859334349632263,
|
1378 |
+
"learning_rate": 6.782775516492771e-05,
|
1379 |
+
"loss": 0.5622,
|
1380 |
+
"step": 940
|
1381 |
+
},
|
1382 |
+
{
|
1383 |
+
"epoch": 6.472602739726027,
|
1384 |
+
"grad_norm": 0.2793692946434021,
|
1385 |
+
"learning_rate": 6.669820615447522e-05,
|
1386 |
+
"loss": 0.5618,
|
1387 |
+
"step": 945
|
1388 |
+
},
|
1389 |
+
{
|
1390 |
+
"epoch": 6.506849315068493,
|
1391 |
+
"grad_norm": 0.2735882103443146,
|
1392 |
+
"learning_rate": 6.5573416097724e-05,
|
1393 |
+
"loss": 0.5674,
|
1394 |
+
"step": 950
|
1395 |
+
},
|
1396 |
+
{
|
1397 |
+
"epoch": 6.541095890410959,
|
1398 |
+
"grad_norm": 0.26911288499832153,
|
1399 |
+
"learning_rate": 6.445354573147484e-05,
|
1400 |
+
"loss": 0.5641,
|
1401 |
+
"step": 955
|
1402 |
+
},
|
1403 |
+
{
|
1404 |
+
"epoch": 6.575342465753424,
|
1405 |
+
"grad_norm": 0.2591806948184967,
|
1406 |
+
"learning_rate": 6.333875508948593e-05,
|
1407 |
+
"loss": 0.5689,
|
1408 |
+
"step": 960
|
1409 |
+
},
|
1410 |
+
{
|
1411 |
+
"epoch": 6.609589041095891,
|
1412 |
+
"grad_norm": 0.25813576579093933,
|
1413 |
+
"learning_rate": 6.22292034796035e-05,
|
1414 |
+
"loss": 0.5609,
|
1415 |
+
"step": 965
|
1416 |
+
},
|
1417 |
+
{
|
1418 |
+
"epoch": 6.6438356164383565,
|
1419 |
+
"grad_norm": 0.26337507367134094,
|
1420 |
+
"learning_rate": 6.112504946099604e-05,
|
1421 |
+
"loss": 0.5577,
|
1422 |
+
"step": 970
|
1423 |
+
},
|
1424 |
+
{
|
1425 |
+
"epoch": 6.678082191780822,
|
1426 |
+
"grad_norm": 0.2651839852333069,
|
1427 |
+
"learning_rate": 6.0026450821495536e-05,
|
1428 |
+
"loss": 0.5593,
|
1429 |
+
"step": 975
|
1430 |
+
},
|
1431 |
+
{
|
1432 |
+
"epoch": 6.712328767123288,
|
1433 |
+
"grad_norm": 0.274582177400589,
|
1434 |
+
"learning_rate": 5.8933564555049105e-05,
|
1435 |
+
"loss": 0.5597,
|
1436 |
+
"step": 980
|
1437 |
+
},
|
1438 |
+
{
|
1439 |
+
"epoch": 6.7465753424657535,
|
1440 |
+
"grad_norm": 0.26564931869506836,
|
1441 |
+
"learning_rate": 5.784654683928391e-05,
|
1442 |
+
"loss": 0.5645,
|
1443 |
+
"step": 985
|
1444 |
+
},
|
1445 |
+
{
|
1446 |
+
"epoch": 6.780821917808219,
|
1447 |
+
"grad_norm": 0.2597637176513672,
|
1448 |
+
"learning_rate": 5.6765553013188766e-05,
|
1449 |
+
"loss": 0.5592,
|
1450 |
+
"step": 990
|
1451 |
+
},
|
1452 |
+
{
|
1453 |
+
"epoch": 6.815068493150685,
|
1454 |
+
"grad_norm": 0.2663344144821167,
|
1455 |
+
"learning_rate": 5.5690737554915604e-05,
|
1456 |
+
"loss": 0.5657,
|
1457 |
+
"step": 995
|
1458 |
+
},
|
1459 |
+
{
|
1460 |
+
"epoch": 6.8493150684931505,
|
1461 |
+
"grad_norm": 0.2605944275856018,
|
1462 |
+
"learning_rate": 5.462225405970401e-05,
|
1463 |
+
"loss": 0.5604,
|
1464 |
+
"step": 1000
|
1465 |
+
},
|
1466 |
+
{
|
1467 |
+
"epoch": 6.883561643835616,
|
1468 |
+
"grad_norm": 0.2578333914279938,
|
1469 |
+
"learning_rate": 5.3560255217931785e-05,
|
1470 |
+
"loss": 0.5713,
|
1471 |
+
"step": 1005
|
1472 |
+
},
|
1473 |
+
{
|
1474 |
+
"epoch": 6.917808219178082,
|
1475 |
+
"grad_norm": 0.2837528586387634,
|
1476 |
+
"learning_rate": 5.2504892793295e-05,
|
1477 |
+
"loss": 0.5609,
|
1478 |
+
"step": 1010
|
1479 |
+
},
|
1480 |
+
{
|
1481 |
+
"epoch": 6.9520547945205475,
|
1482 |
+
"grad_norm": 0.26841509342193604,
|
1483 |
+
"learning_rate": 5.145631760112022e-05,
|
1484 |
+
"loss": 0.5696,
|
1485 |
+
"step": 1015
|
1486 |
+
},
|
1487 |
+
{
|
1488 |
+
"epoch": 6.986301369863014,
|
1489 |
+
"grad_norm": 0.27431559562683105,
|
1490 |
+
"learning_rate": 5.041467948681269e-05,
|
1491 |
+
"loss": 0.5603,
|
1492 |
+
"step": 1020
|
1493 |
+
},
|
1494 |
+
{
|
1495 |
+
"epoch": 7.0,
|
1496 |
+
"eval_loss": 1.7421818971633911,
|
1497 |
+
"eval_runtime": 0.6796,
|
1498 |
+
"eval_samples_per_second": 8.829,
|
1499 |
+
"eval_steps_per_second": 1.471,
|
1500 |
+
"step": 1022
|
1501 |
+
},
|
1502 |
+
{
|
1503 |
+
"epoch": 7.02054794520548,
|
1504 |
+
"grad_norm": 0.2642446458339691,
|
1505 |
+
"learning_rate": 4.9380127304442634e-05,
|
1506 |
+
"loss": 0.5559,
|
1507 |
+
"step": 1025
|
1508 |
+
},
|
1509 |
+
{
|
1510 |
+
"epoch": 7.054794520547945,
|
1511 |
+
"grad_norm": 0.25995686650276184,
|
1512 |
+
"learning_rate": 4.835280889547351e-05,
|
1513 |
+
"loss": 0.5473,
|
1514 |
+
"step": 1030
|
1515 |
+
},
|
1516 |
+
{
|
1517 |
+
"epoch": 7.089041095890411,
|
1518 |
+
"grad_norm": 0.26262959837913513,
|
1519 |
+
"learning_rate": 4.733287106763481e-05,
|
1520 |
+
"loss": 0.5516,
|
1521 |
+
"step": 1035
|
1522 |
+
},
|
1523 |
+
{
|
1524 |
+
"epoch": 7.123287671232877,
|
1525 |
+
"grad_norm": 0.2740737199783325,
|
1526 |
+
"learning_rate": 4.6320459573942856e-05,
|
1527 |
+
"loss": 0.5395,
|
1528 |
+
"step": 1040
|
1529 |
+
},
|
1530 |
+
{
|
1531 |
+
"epoch": 7.157534246575342,
|
1532 |
+
"grad_norm": 0.2638036012649536,
|
1533 |
+
"learning_rate": 4.531571909187197e-05,
|
1534 |
+
"loss": 0.5507,
|
1535 |
+
"step": 1045
|
1536 |
+
},
|
1537 |
+
{
|
1538 |
+
"epoch": 7.191780821917808,
|
1539 |
+
"grad_norm": 0.2760368585586548,
|
1540 |
+
"learning_rate": 4.431879320267972e-05,
|
1541 |
+
"loss": 0.5516,
|
1542 |
+
"step": 1050
|
1543 |
+
},
|
1544 |
+
{
|
1545 |
+
"epoch": 7.226027397260274,
|
1546 |
+
"grad_norm": 0.26356157660484314,
|
1547 |
+
"learning_rate": 4.332982437088825e-05,
|
1548 |
+
"loss": 0.5528,
|
1549 |
+
"step": 1055
|
1550 |
+
},
|
1551 |
+
{
|
1552 |
+
"epoch": 7.260273972602739,
|
1553 |
+
"grad_norm": 0.27334779500961304,
|
1554 |
+
"learning_rate": 4.2348953923925916e-05,
|
1555 |
+
"loss": 0.547,
|
1556 |
+
"step": 1060
|
1557 |
+
},
|
1558 |
+
{
|
1559 |
+
"epoch": 7.294520547945205,
|
1560 |
+
"grad_norm": 0.2645105719566345,
|
1561 |
+
"learning_rate": 4.137632203193086e-05,
|
1562 |
+
"loss": 0.5449,
|
1563 |
+
"step": 1065
|
1564 |
+
},
|
1565 |
+
{
|
1566 |
+
"epoch": 7.328767123287671,
|
1567 |
+
"grad_norm": 0.2697451114654541,
|
1568 |
+
"learning_rate": 4.041206768772022e-05,
|
1569 |
+
"loss": 0.5484,
|
1570 |
+
"step": 1070
|
1571 |
+
},
|
1572 |
+
{
|
1573 |
+
"epoch": 7.363013698630137,
|
1574 |
+
"grad_norm": 0.2794542610645294,
|
1575 |
+
"learning_rate": 3.9456328686927525e-05,
|
1576 |
+
"loss": 0.5527,
|
1577 |
+
"step": 1075
|
1578 |
+
},
|
1579 |
+
{
|
1580 |
+
"epoch": 7.397260273972603,
|
1581 |
+
"grad_norm": 0.27488037943840027,
|
1582 |
+
"learning_rate": 3.850924160831115e-05,
|
1583 |
+
"loss": 0.5473,
|
1584 |
+
"step": 1080
|
1585 |
+
},
|
1586 |
+
{
|
1587 |
+
"epoch": 7.431506849315069,
|
1588 |
+
"grad_norm": 0.2687772810459137,
|
1589 |
+
"learning_rate": 3.757094179423672e-05,
|
1590 |
+
"loss": 0.5505,
|
1591 |
+
"step": 1085
|
1592 |
+
},
|
1593 |
+
{
|
1594 |
+
"epoch": 7.465753424657534,
|
1595 |
+
"grad_norm": 0.2724910080432892,
|
1596 |
+
"learning_rate": 3.6641563331336125e-05,
|
1597 |
+
"loss": 0.5615,
|
1598 |
+
"step": 1090
|
1599 |
+
},
|
1600 |
+
{
|
1601 |
+
"epoch": 7.5,
|
1602 |
+
"grad_norm": 0.26319825649261475,
|
1603 |
+
"learning_rate": 3.5721239031346066e-05,
|
1604 |
+
"loss": 0.5571,
|
1605 |
+
"step": 1095
|
1606 |
+
},
|
1607 |
+
{
|
1608 |
+
"epoch": 7.534246575342466,
|
1609 |
+
"grad_norm": 0.270577609539032,
|
1610 |
+
"learning_rate": 3.4810100412128747e-05,
|
1611 |
+
"loss": 0.5428,
|
1612 |
+
"step": 1100
|
1613 |
+
},
|
1614 |
+
{
|
1615 |
+
"epoch": 7.568493150684931,
|
1616 |
+
"grad_norm": 0.2661505341529846,
|
1617 |
+
"learning_rate": 3.3908277678877445e-05,
|
1618 |
+
"loss": 0.5515,
|
1619 |
+
"step": 1105
|
1620 |
+
},
|
1621 |
+
{
|
1622 |
+
"epoch": 7.602739726027397,
|
1623 |
+
"grad_norm": 0.2684604823589325,
|
1624 |
+
"learning_rate": 3.3015899705509734e-05,
|
1625 |
+
"loss": 0.5513,
|
1626 |
+
"step": 1110
|
1627 |
+
},
|
1628 |
+
{
|
1629 |
+
"epoch": 7.636986301369863,
|
1630 |
+
"grad_norm": 0.2670838534832001,
|
1631 |
+
"learning_rate": 3.21330940162508e-05,
|
1632 |
+
"loss": 0.5553,
|
1633 |
+
"step": 1115
|
1634 |
+
},
|
1635 |
+
{
|
1636 |
+
"epoch": 7.671232876712329,
|
1637 |
+
"grad_norm": 0.26764294505119324,
|
1638 |
+
"learning_rate": 3.125998676740987e-05,
|
1639 |
+
"loss": 0.5553,
|
1640 |
+
"step": 1120
|
1641 |
+
},
|
1642 |
+
{
|
1643 |
+
"epoch": 7.705479452054795,
|
1644 |
+
"grad_norm": 0.2651881277561188,
|
1645 |
+
"learning_rate": 3.0396702729352023e-05,
|
1646 |
+
"loss": 0.5502,
|
1647 |
+
"step": 1125
|
1648 |
+
},
|
1649 |
+
{
|
1650 |
+
"epoch": 7.739726027397261,
|
1651 |
+
"grad_norm": 0.27107974886894226,
|
1652 |
+
"learning_rate": 2.9543365268667867e-05,
|
1653 |
+
"loss": 0.5522,
|
1654 |
+
"step": 1130
|
1655 |
+
},
|
1656 |
+
{
|
1657 |
+
"epoch": 7.773972602739726,
|
1658 |
+
"grad_norm": 0.2676118314266205,
|
1659 |
+
"learning_rate": 2.8700096330544012e-05,
|
1660 |
+
"loss": 0.5506,
|
1661 |
+
"step": 1135
|
1662 |
+
},
|
1663 |
+
{
|
1664 |
+
"epoch": 7.808219178082192,
|
1665 |
+
"grad_norm": 0.2681746184825897,
|
1666 |
+
"learning_rate": 2.7867016421336776e-05,
|
1667 |
+
"loss": 0.5473,
|
1668 |
+
"step": 1140
|
1669 |
+
},
|
1670 |
+
{
|
1671 |
+
"epoch": 7.842465753424658,
|
1672 |
+
"grad_norm": 0.27224835753440857,
|
1673 |
+
"learning_rate": 2.7044244591351232e-05,
|
1674 |
+
"loss": 0.5493,
|
1675 |
+
"step": 1145
|
1676 |
+
},
|
1677 |
+
{
|
1678 |
+
"epoch": 7.876712328767123,
|
1679 |
+
"grad_norm": 0.2668222486972809,
|
1680 |
+
"learning_rate": 2.6231898417828603e-05,
|
1681 |
+
"loss": 0.55,
|
1682 |
+
"step": 1150
|
1683 |
+
},
|
1684 |
+
{
|
1685 |
+
"epoch": 7.910958904109589,
|
1686 |
+
"grad_norm": 0.2646133005619049,
|
1687 |
+
"learning_rate": 2.5430093988143778e-05,
|
1688 |
+
"loss": 0.5532,
|
1689 |
+
"step": 1155
|
1690 |
+
},
|
1691 |
+
{
|
1692 |
+
"epoch": 7.945205479452055,
|
1693 |
+
"grad_norm": 0.26807960867881775,
|
1694 |
+
"learning_rate": 2.4638945883216235e-05,
|
1695 |
+
"loss": 0.555,
|
1696 |
+
"step": 1160
|
1697 |
+
},
|
1698 |
+
{
|
1699 |
+
"epoch": 7.97945205479452,
|
1700 |
+
"grad_norm": 0.2676006853580475,
|
1701 |
+
"learning_rate": 2.385856716113587e-05,
|
1702 |
+
"loss": 0.5632,
|
1703 |
+
"step": 1165
|
1704 |
+
},
|
1705 |
+
{
|
1706 |
+
"epoch": 8.0,
|
1707 |
+
"eval_loss": 1.7482503652572632,
|
1708 |
+
"eval_runtime": 0.6829,
|
1709 |
+
"eval_samples_per_second": 8.786,
|
1710 |
+
"eval_steps_per_second": 1.464,
|
1711 |
+
"step": 1168
|
1712 |
+
},
|
1713 |
+
{
|
1714 |
+
"epoch": 8.013698630136986,
|
1715 |
+
"grad_norm": 0.2698109447956085,
|
1716 |
+
"learning_rate": 2.3089069341006565e-05,
|
1717 |
+
"loss": 0.5524,
|
1718 |
+
"step": 1170
|
1719 |
+
},
|
1720 |
+
{
|
1721 |
+
"epoch": 8.047945205479452,
|
1722 |
+
"grad_norm": 0.26906144618988037,
|
1723 |
+
"learning_rate": 2.2330562387009745e-05,
|
1724 |
+
"loss": 0.5412,
|
1725 |
+
"step": 1175
|
1726 |
+
},
|
1727 |
+
{
|
1728 |
+
"epoch": 8.082191780821917,
|
1729 |
+
"grad_norm": 0.2829205095767975,
|
1730 |
+
"learning_rate": 2.1583154692689976e-05,
|
1731 |
+
"loss": 0.5406,
|
1732 |
+
"step": 1180
|
1733 |
+
},
|
1734 |
+
{
|
1735 |
+
"epoch": 8.116438356164384,
|
1736 |
+
"grad_norm": 0.27515366673469543,
|
1737 |
+
"learning_rate": 2.08469530654652e-05,
|
1738 |
+
"loss": 0.5447,
|
1739 |
+
"step": 1185
|
1740 |
+
},
|
1741 |
+
{
|
1742 |
+
"epoch": 8.150684931506849,
|
1743 |
+
"grad_norm": 0.26556894183158875,
|
1744 |
+
"learning_rate": 2.0122062711363532e-05,
|
1745 |
+
"loss": 0.5426,
|
1746 |
+
"step": 1190
|
1747 |
+
},
|
1748 |
+
{
|
1749 |
+
"epoch": 8.184931506849315,
|
1750 |
+
"grad_norm": 0.2701828181743622,
|
1751 |
+
"learning_rate": 1.9408587219988805e-05,
|
1752 |
+
"loss": 0.5375,
|
1753 |
+
"step": 1195
|
1754 |
+
},
|
1755 |
+
{
|
1756 |
+
"epoch": 8.219178082191782,
|
1757 |
+
"grad_norm": 0.2658767104148865,
|
1758 |
+
"learning_rate": 1.8706628549717452e-05,
|
1759 |
+
"loss": 0.5402,
|
1760 |
+
"step": 1200
|
1761 |
+
},
|
1762 |
+
{
|
1763 |
+
"epoch": 8.253424657534246,
|
1764 |
+
"grad_norm": 0.28211283683776855,
|
1765 |
+
"learning_rate": 1.8016287013128018e-05,
|
1766 |
+
"loss": 0.5416,
|
1767 |
+
"step": 1205
|
1768 |
+
},
|
1769 |
+
{
|
1770 |
+
"epoch": 8.287671232876713,
|
1771 |
+
"grad_norm": 0.281766802072525,
|
1772 |
+
"learning_rate": 1.7337661262666294e-05,
|
1773 |
+
"loss": 0.546,
|
1774 |
+
"step": 1210
|
1775 |
+
},
|
1776 |
+
{
|
1777 |
+
"epoch": 8.321917808219178,
|
1778 |
+
"grad_norm": 0.2836877405643463,
|
1779 |
+
"learning_rate": 1.6670848276547334e-05,
|
1780 |
+
"loss": 0.5415,
|
1781 |
+
"step": 1215
|
1782 |
+
},
|
1783 |
+
{
|
1784 |
+
"epoch": 8.356164383561644,
|
1785 |
+
"grad_norm": 0.2742989659309387,
|
1786 |
+
"learning_rate": 1.601594334489702e-05,
|
1787 |
+
"loss": 0.5483,
|
1788 |
+
"step": 1220
|
1789 |
+
},
|
1790 |
+
{
|
1791 |
+
"epoch": 8.39041095890411,
|
1792 |
+
"grad_norm": 0.2683957517147064,
|
1793 |
+
"learning_rate": 1.5373040056134814e-05,
|
1794 |
+
"loss": 0.5433,
|
1795 |
+
"step": 1225
|
1796 |
+
},
|
1797 |
+
{
|
1798 |
+
"epoch": 8.424657534246576,
|
1799 |
+
"grad_norm": 0.2692102789878845,
|
1800 |
+
"learning_rate": 1.474223028359939e-05,
|
1801 |
+
"loss": 0.5465,
|
1802 |
+
"step": 1230
|
1803 |
+
},
|
1804 |
+
{
|
1805 |
+
"epoch": 8.45890410958904,
|
1806 |
+
"grad_norm": 0.2703090310096741,
|
1807 |
+
"learning_rate": 1.4123604172419713e-05,
|
1808 |
+
"loss": 0.5405,
|
1809 |
+
"step": 1235
|
1810 |
+
},
|
1811 |
+
{
|
1812 |
+
"epoch": 8.493150684931507,
|
1813 |
+
"grad_norm": 0.27296507358551025,
|
1814 |
+
"learning_rate": 1.3517250126632986e-05,
|
1815 |
+
"loss": 0.5424,
|
1816 |
+
"step": 1240
|
1817 |
+
},
|
1818 |
+
{
|
1819 |
+
"epoch": 8.527397260273972,
|
1820 |
+
"grad_norm": 0.26967960596084595,
|
1821 |
+
"learning_rate": 1.292325479655131e-05,
|
1822 |
+
"loss": 0.5428,
|
1823 |
+
"step": 1245
|
1824 |
+
},
|
1825 |
+
{
|
1826 |
+
"epoch": 8.561643835616438,
|
1827 |
+
"grad_norm": 0.27562007308006287,
|
1828 |
+
"learning_rate": 1.2341703066379074e-05,
|
1829 |
+
"loss": 0.546,
|
1830 |
+
"step": 1250
|
1831 |
+
},
|
1832 |
+
{
|
1833 |
+
"epoch": 8.595890410958905,
|
1834 |
+
"grad_norm": 0.2698553204536438,
|
1835 |
+
"learning_rate": 1.1772678042082607e-05,
|
1836 |
+
"loss": 0.5463,
|
1837 |
+
"step": 1255
|
1838 |
+
},
|
1839 |
+
{
|
1840 |
+
"epoch": 8.63013698630137,
|
1841 |
+
"grad_norm": 0.27123743295669556,
|
1842 |
+
"learning_rate": 1.1216261039514087e-05,
|
1843 |
+
"loss": 0.5423,
|
1844 |
+
"step": 1260
|
1845 |
+
},
|
1846 |
+
{
|
1847 |
+
"epoch": 8.664383561643836,
|
1848 |
+
"grad_norm": 0.27804234623908997,
|
1849 |
+
"learning_rate": 1.0672531572791178e-05,
|
1850 |
+
"loss": 0.5481,
|
1851 |
+
"step": 1265
|
1852 |
+
},
|
1853 |
+
{
|
1854 |
+
"epoch": 8.698630136986301,
|
1855 |
+
"grad_norm": 0.26643478870391846,
|
1856 |
+
"learning_rate": 1.0141567342934132e-05,
|
1857 |
+
"loss": 0.5415,
|
1858 |
+
"step": 1270
|
1859 |
+
},
|
1860 |
+
{
|
1861 |
+
"epoch": 8.732876712328768,
|
1862 |
+
"grad_norm": 0.2655141353607178,
|
1863 |
+
"learning_rate": 9.623444226762035e-06,
|
1864 |
+
"loss": 0.5439,
|
1865 |
+
"step": 1275
|
1866 |
+
},
|
1867 |
+
{
|
1868 |
+
"epoch": 8.767123287671232,
|
1869 |
+
"grad_norm": 0.26757049560546875,
|
1870 |
+
"learning_rate": 9.118236266049707e-06,
|
1871 |
+
"loss": 0.5495,
|
1872 |
+
"step": 1280
|
1873 |
+
},
|
1874 |
+
{
|
1875 |
+
"epoch": 8.801369863013699,
|
1876 |
+
"grad_norm": 0.2669236958026886,
|
1877 |
+
"learning_rate": 8.626015656946895e-06,
|
1878 |
+
"loss": 0.5473,
|
1879 |
+
"step": 1285
|
1880 |
+
},
|
1881 |
+
{
|
1882 |
+
"epoch": 8.835616438356164,
|
1883 |
+
"grad_norm": 0.2733878791332245,
|
1884 |
+
"learning_rate": 8.146852739661105e-06,
|
1885 |
+
"loss": 0.5449,
|
1886 |
+
"step": 1290
|
1887 |
+
},
|
1888 |
+
{
|
1889 |
+
"epoch": 8.86986301369863,
|
1890 |
+
"grad_norm": 0.27196043729782104,
|
1891 |
+
"learning_rate": 7.6808159884057e-06,
|
1892 |
+
"loss": 0.5451,
|
1893 |
+
"step": 1295
|
1894 |
+
},
|
1895 |
+
{
|
1896 |
+
"epoch": 8.904109589041095,
|
1897 |
+
"grad_norm": 0.27573108673095703,
|
1898 |
+
"learning_rate": 7.2279720016148244e-06,
|
1899 |
+
"loss": 0.5422,
|
1900 |
+
"step": 1300
|
1901 |
+
},
|
1902 |
+
{
|
1903 |
+
"epoch": 8.938356164383562,
|
1904 |
+
"grad_norm": 0.2723158299922943,
|
1905 |
+
"learning_rate": 6.788385492426053e-06,
|
1906 |
+
"loss": 0.548,
|
1907 |
+
"step": 1305
|
1908 |
+
},
|
1909 |
+
{
|
1910 |
+
"epoch": 8.972602739726028,
|
1911 |
+
"grad_norm": 0.2763724625110626,
|
1912 |
+
"learning_rate": 6.36211927943271e-06,
|
1913 |
+
"loss": 0.5497,
|
1914 |
+
"step": 1310
|
1915 |
+
},
|
1916 |
+
{
|
1917 |
+
"epoch": 9.0,
|
1918 |
+
"eval_loss": 1.7569636106491089,
|
1919 |
+
"eval_runtime": 0.6815,
|
1920 |
+
"eval_samples_per_second": 8.804,
|
1921 |
+
"eval_steps_per_second": 1.467,
|
1922 |
+
"step": 1314
|
1923 |
+
},
|
1924 |
+
{
|
1925 |
+
"epoch": 9.006849315068493,
|
1926 |
+
"grad_norm": 0.26691538095474243,
|
1927 |
+
"learning_rate": 5.949234277706861e-06,
|
1928 |
+
"loss": 0.5373,
|
1929 |
+
"step": 1315
|
1930 |
+
},
|
1931 |
+
{
|
1932 |
+
"epoch": 9.04109589041096,
|
1933 |
+
"grad_norm": 0.26734665036201477,
|
1934 |
+
"learning_rate": 5.549789490094304e-06,
|
1935 |
+
"loss": 0.5391,
|
1936 |
+
"step": 1320
|
1937 |
+
},
|
1938 |
+
{
|
1939 |
+
"epoch": 9.075342465753424,
|
1940 |
+
"grad_norm": 0.2708768844604492,
|
1941 |
+
"learning_rate": 5.163841998782837e-06,
|
1942 |
+
"loss": 0.5492,
|
1943 |
+
"step": 1325
|
1944 |
+
},
|
1945 |
+
{
|
1946 |
+
"epoch": 9.10958904109589,
|
1947 |
+
"grad_norm": 0.2653951942920685,
|
1948 |
+
"learning_rate": 4.79144695714504e-06,
|
1949 |
+
"loss": 0.5358,
|
1950 |
+
"step": 1330
|
1951 |
+
},
|
1952 |
+
{
|
1953 |
+
"epoch": 9.143835616438356,
|
1954 |
+
"grad_norm": 0.27374231815338135,
|
1955 |
+
"learning_rate": 4.432657581856525e-06,
|
1956 |
+
"loss": 0.5426,
|
1957 |
+
"step": 1335
|
1958 |
+
},
|
1959 |
+
{
|
1960 |
+
"epoch": 9.178082191780822,
|
1961 |
+
"grad_norm": 0.2698308229446411,
|
1962 |
+
"learning_rate": 4.087525145291204e-06,
|
1963 |
+
"loss": 0.54,
|
1964 |
+
"step": 1340
|
1965 |
+
},
|
1966 |
+
{
|
1967 |
+
"epoch": 9.212328767123287,
|
1968 |
+
"grad_norm": 0.2661896049976349,
|
1969 |
+
"learning_rate": 3.7560989681941992e-06,
|
1970 |
+
"loss": 0.5396,
|
1971 |
+
"step": 1345
|
1972 |
+
},
|
1973 |
+
{
|
1974 |
+
"epoch": 9.246575342465754,
|
1975 |
+
"grad_norm": 0.26976555585861206,
|
1976 |
+
"learning_rate": 3.4384264126337328e-06,
|
1977 |
+
"loss": 0.5421,
|
1978 |
+
"step": 1350
|
1979 |
+
},
|
1980 |
+
{
|
1981 |
+
"epoch": 9.280821917808218,
|
1982 |
+
"grad_norm": 0.26978161931037903,
|
1983 |
+
"learning_rate": 3.1345528752329212e-06,
|
1984 |
+
"loss": 0.5419,
|
1985 |
+
"step": 1355
|
1986 |
+
},
|
1987 |
+
{
|
1988 |
+
"epoch": 9.315068493150685,
|
1989 |
+
"grad_norm": 0.2713252902030945,
|
1990 |
+
"learning_rate": 2.8445217806824077e-06,
|
1991 |
+
"loss": 0.5367,
|
1992 |
+
"step": 1360
|
1993 |
+
},
|
1994 |
+
{
|
1995 |
+
"epoch": 9.349315068493151,
|
1996 |
+
"grad_norm": 0.27367061376571655,
|
1997 |
+
"learning_rate": 2.5683745755348044e-06,
|
1998 |
+
"loss": 0.5384,
|
1999 |
+
"step": 1365
|
2000 |
+
},
|
2001 |
+
{
|
2002 |
+
"epoch": 9.383561643835616,
|
2003 |
+
"grad_norm": 0.2732658386230469,
|
2004 |
+
"learning_rate": 2.30615072228183e-06,
|
2005 |
+
"loss": 0.5452,
|
2006 |
+
"step": 1370
|
2007 |
+
},
|
2008 |
+
{
|
2009 |
+
"epoch": 9.417808219178083,
|
2010 |
+
"grad_norm": 0.2684633731842041,
|
2011 |
+
"learning_rate": 2.057887693714988e-06,
|
2012 |
+
"loss": 0.5354,
|
2013 |
+
"step": 1375
|
2014 |
+
},
|
2015 |
+
{
|
2016 |
+
"epoch": 9.452054794520548,
|
2017 |
+
"grad_norm": 0.2733153700828552,
|
2018 |
+
"learning_rate": 1.8236209675705274e-06,
|
2019 |
+
"loss": 0.5406,
|
2020 |
+
"step": 1380
|
2021 |
+
},
|
2022 |
+
{
|
2023 |
+
"epoch": 9.486301369863014,
|
2024 |
+
"grad_norm": 0.265996515750885,
|
2025 |
+
"learning_rate": 1.6033840214595308e-06,
|
2026 |
+
"loss": 0.5394,
|
2027 |
+
"step": 1385
|
2028 |
+
},
|
2029 |
+
{
|
2030 |
+
"epoch": 9.520547945205479,
|
2031 |
+
"grad_norm": 0.26887837052345276,
|
2032 |
+
"learning_rate": 1.397208328083921e-06,
|
2033 |
+
"loss": 0.5407,
|
2034 |
+
"step": 1390
|
2035 |
+
},
|
2036 |
+
{
|
2037 |
+
"epoch": 9.554794520547945,
|
2038 |
+
"grad_norm": 0.26908043026924133,
|
2039 |
+
"learning_rate": 1.205123350738746e-06,
|
2040 |
+
"loss": 0.5366,
|
2041 |
+
"step": 1395
|
2042 |
+
},
|
2043 |
+
{
|
2044 |
+
"epoch": 9.58904109589041,
|
2045 |
+
"grad_norm": 0.2689751088619232,
|
2046 |
+
"learning_rate": 1.0271565391018922e-06,
|
2047 |
+
"loss": 0.5381,
|
2048 |
+
"step": 1400
|
2049 |
+
},
|
2050 |
+
{
|
2051 |
+
"epoch": 9.623287671232877,
|
2052 |
+
"grad_norm": 0.2709926664829254,
|
2053 |
+
"learning_rate": 8.633333253113995e-07,
|
2054 |
+
"loss": 0.5383,
|
2055 |
+
"step": 1405
|
2056 |
+
},
|
2057 |
+
{
|
2058 |
+
"epoch": 9.657534246575342,
|
2059 |
+
"grad_norm": 0.26522693037986755,
|
2060 |
+
"learning_rate": 7.136771203310245e-07,
|
2061 |
+
"loss": 0.5417,
|
2062 |
+
"step": 1410
|
2063 |
+
},
|
2064 |
+
{
|
2065 |
+
"epoch": 9.691780821917808,
|
2066 |
+
"grad_norm": 0.2700962722301483,
|
2067 |
+
"learning_rate": 5.782093106048159e-07,
|
2068 |
+
"loss": 0.5441,
|
2069 |
+
"step": 1415
|
2070 |
+
},
|
2071 |
+
{
|
2072 |
+
"epoch": 9.726027397260275,
|
2073 |
+
"grad_norm": 0.2700207531452179,
|
2074 |
+
"learning_rate": 4.569492550008603e-07,
|
2075 |
+
"loss": 0.5373,
|
2076 |
+
"step": 1420
|
2077 |
+
},
|
2078 |
+
{
|
2079 |
+
"epoch": 9.76027397260274,
|
2080 |
+
"grad_norm": 0.26531684398651123,
|
2081 |
+
"learning_rate": 3.49914282044872e-07,
|
2082 |
+
"loss": 0.5395,
|
2083 |
+
"step": 1425
|
2084 |
+
},
|
2085 |
+
{
|
2086 |
+
"epoch": 9.794520547945206,
|
2087 |
+
"grad_norm": 0.2695051431655884,
|
2088 |
+
"learning_rate": 2.5711968744382974e-07,
|
2089 |
+
"loss": 0.536,
|
2090 |
+
"step": 1430
|
2091 |
+
},
|
2092 |
+
{
|
2093 |
+
"epoch": 9.82876712328767,
|
2094 |
+
"grad_norm": 0.2686808407306671,
|
2095 |
+
"learning_rate": 1.7857873190019192e-07,
|
2096 |
+
"loss": 0.544,
|
2097 |
+
"step": 1435
|
2098 |
+
},
|
2099 |
+
{
|
2100 |
+
"epoch": 9.863013698630137,
|
2101 |
+
"grad_norm": 0.27493226528167725,
|
2102 |
+
"learning_rate": 1.143026392168789e-07,
|
2103 |
+
"loss": 0.5456,
|
2104 |
+
"step": 1440
|
2105 |
+
},
|
2106 |
+
{
|
2107 |
+
"epoch": 9.897260273972602,
|
2108 |
+
"grad_norm": 0.27204445004463196,
|
2109 |
+
"learning_rate": 6.430059469334504e-08,
|
2110 |
+
"loss": 0.5423,
|
2111 |
+
"step": 1445
|
2112 |
+
},
|
2113 |
+
{
|
2114 |
+
"epoch": 9.931506849315069,
|
2115 |
+
"grad_norm": 0.27562659978866577,
|
2116 |
+
"learning_rate": 2.8579743813006432e-08,
|
2117 |
+
"loss": 0.5435,
|
2118 |
+
"step": 1450
|
2119 |
+
},
|
2120 |
+
{
|
2121 |
+
"epoch": 9.965753424657533,
|
2122 |
+
"grad_norm": 0.27623289823532104,
|
2123 |
+
"learning_rate": 7.145191222035497e-09,
|
2124 |
+
"loss": 0.5455,
|
2125 |
+
"step": 1455
|
2126 |
+
},
|
2127 |
+
{
|
2128 |
+
"epoch": 10.0,
|
2129 |
+
"grad_norm": 0.26718658208847046,
|
2130 |
+
"learning_rate": 0.0,
|
2131 |
+
"loss": 0.5353,
|
2132 |
+
"step": 1460
|
2133 |
+
},
|
2134 |
+
{
|
2135 |
+
"epoch": 10.0,
|
2136 |
+
"eval_loss": 1.7579952478408813,
|
2137 |
+
"eval_runtime": 0.6696,
|
2138 |
+
"eval_samples_per_second": 8.961,
|
2139 |
+
"eval_steps_per_second": 1.493,
|
2140 |
+
"step": 1460
|
2141 |
+
},
|
2142 |
+
{
|
2143 |
+
"epoch": 10.0,
|
2144 |
+
"step": 1460,
|
2145 |
+
"total_flos": 8.183666472539652e+18,
|
2146 |
+
"train_loss": 0.6181400966970888,
|
2147 |
+
"train_runtime": 11966.6323,
|
2148 |
+
"train_samples_per_second": 15.569,
|
2149 |
+
"train_steps_per_second": 0.122
|
2150 |
+
}
|
2151 |
+
],
|
2152 |
+
"logging_steps": 5,
|
2153 |
+
"max_steps": 1460,
|
2154 |
+
"num_input_tokens_seen": 0,
|
2155 |
+
"num_train_epochs": 10,
|
2156 |
+
"save_steps": 100,
|
2157 |
+
"stateful_callbacks": {
|
2158 |
+
"TrainerControl": {
|
2159 |
+
"args": {
|
2160 |
+
"should_epoch_stop": false,
|
2161 |
+
"should_evaluate": false,
|
2162 |
+
"should_log": false,
|
2163 |
+
"should_save": true,
|
2164 |
+
"should_training_stop": true
|
2165 |
+
},
|
2166 |
+
"attributes": {}
|
2167 |
+
}
|
2168 |
+
},
|
2169 |
+
"total_flos": 8.183666472539652e+18,
|
2170 |
+
"train_batch_size": 8,
|
2171 |
+
"trial_name": null,
|
2172 |
+
"trial_params": null
|
2173 |
+
}
|