Model save
Browse files- README.md +83 -0
- adapter_model.safetensors +1 -1
- all_results.json +9 -0
- runs/Jun05_05-37-11_7bdd13775218/events.out.tfevents.1717565872.7bdd13775218.92549.0 +2 -2
- train_results.json +9 -0
- trainer_state.json +2341 -0
README.md
ADDED
@@ -0,0 +1,83 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: gemma
|
3 |
+
library_name: peft
|
4 |
+
tags:
|
5 |
+
- trl
|
6 |
+
- sft
|
7 |
+
- generated_from_trainer
|
8 |
+
base_model: google/gemma-2b
|
9 |
+
datasets:
|
10 |
+
- generator
|
11 |
+
model-index:
|
12 |
+
- name: gemma2b-summarize-gemini1_5flash-64k
|
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 |
+
# gemma2b-summarize-gemini1_5flash-64k
|
20 |
+
|
21 |
+
This model is a fine-tuned version of [google/gemma-2b](https://huggingface.co/google/gemma-2b) on the generator dataset.
|
22 |
+
It achieves the following results on the evaluation set:
|
23 |
+
- Loss: 2.5425
|
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: 8
|
45 |
+
- seed: 42
|
46 |
+
- distributed_type: multi-GPU
|
47 |
+
- num_devices: 4
|
48 |
+
- gradient_accumulation_steps: 2
|
49 |
+
- total_train_batch_size: 64
|
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: 15
|
55 |
+
|
56 |
+
### Training results
|
57 |
+
|
58 |
+
| Training Loss | Epoch | Step | Validation Loss |
|
59 |
+
|:-------------:|:-------:|:----:|:---------------:|
|
60 |
+
| 1.2724 | 0.9952 | 104 | 2.5531 |
|
61 |
+
| 1.0792 | 2.0 | 209 | 2.4752 |
|
62 |
+
| 1.0312 | 2.9952 | 313 | 2.4632 |
|
63 |
+
| 0.9957 | 4.0 | 418 | 2.4705 |
|
64 |
+
| 0.9775 | 4.9952 | 522 | 2.4764 |
|
65 |
+
| 0.9584 | 6.0 | 627 | 2.4870 |
|
66 |
+
| 0.9368 | 6.9952 | 731 | 2.4975 |
|
67 |
+
| 0.93 | 8.0 | 836 | 2.5067 |
|
68 |
+
| 0.9195 | 8.9952 | 940 | 2.5168 |
|
69 |
+
| 0.912 | 10.0 | 1045 | 2.5271 |
|
70 |
+
| 0.9003 | 10.9952 | 1149 | 2.5356 |
|
71 |
+
| 0.9032 | 12.0 | 1254 | 2.5401 |
|
72 |
+
| 0.9006 | 12.9952 | 1358 | 2.5426 |
|
73 |
+
| 0.9007 | 14.0 | 1463 | 2.5433 |
|
74 |
+
| 0.896 | 14.9282 | 1560 | 2.5425 |
|
75 |
+
|
76 |
+
|
77 |
+
### Framework versions
|
78 |
+
|
79 |
+
- PEFT 0.11.1
|
80 |
+
- Transformers 4.40.1
|
81 |
+
- Pytorch 2.2.0+cu121
|
82 |
+
- Datasets 2.19.2
|
83 |
+
- 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 19644912
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:65c2b237a9d6e4f42820f11fb6a46e7fa64092ce0eaf8bfe6216180cfe7ee76f
|
3 |
size 19644912
|
all_results.json
ADDED
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 14.92822966507177,
|
3 |
+
"total_flos": 1.2217365722824704e+18,
|
4 |
+
"train_loss": 1.0227742999027938,
|
5 |
+
"train_runtime": 5355.4835,
|
6 |
+
"train_samples": 63353,
|
7 |
+
"train_samples_per_second": 18.668,
|
8 |
+
"train_steps_per_second": 0.291
|
9 |
+
}
|
runs/Jun05_05-37-11_7bdd13775218/events.out.tfevents.1717565872.7bdd13775218.92549.0
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:fa71059a8babe57c7d18e52f24f32852969c97b76db65c2239c7ff868373bc12
|
3 |
+
size 75804
|
train_results.json
ADDED
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 14.92822966507177,
|
3 |
+
"total_flos": 1.2217365722824704e+18,
|
4 |
+
"train_loss": 1.0227742999027938,
|
5 |
+
"train_runtime": 5355.4835,
|
6 |
+
"train_samples": 63353,
|
7 |
+
"train_samples_per_second": 18.668,
|
8 |
+
"train_steps_per_second": 0.291
|
9 |
+
}
|
trainer_state.json
ADDED
@@ -0,0 +1,2341 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 14.92822966507177,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 1560,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.009569377990430622,
|
13 |
+
"grad_norm": 2.171875,
|
14 |
+
"learning_rate": 1.282051282051282e-06,
|
15 |
+
"loss": 3.0241,
|
16 |
+
"step": 1
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.04784688995215311,
|
20 |
+
"grad_norm": 2.40625,
|
21 |
+
"learning_rate": 6.41025641025641e-06,
|
22 |
+
"loss": 3.0093,
|
23 |
+
"step": 5
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 0.09569377990430622,
|
27 |
+
"grad_norm": 2.09375,
|
28 |
+
"learning_rate": 1.282051282051282e-05,
|
29 |
+
"loss": 3.0072,
|
30 |
+
"step": 10
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 0.14354066985645933,
|
34 |
+
"grad_norm": 1.984375,
|
35 |
+
"learning_rate": 1.923076923076923e-05,
|
36 |
+
"loss": 2.9982,
|
37 |
+
"step": 15
|
38 |
+
},
|
39 |
+
{
|
40 |
+
"epoch": 0.19138755980861244,
|
41 |
+
"grad_norm": 2.015625,
|
42 |
+
"learning_rate": 2.564102564102564e-05,
|
43 |
+
"loss": 2.9122,
|
44 |
+
"step": 20
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.23923444976076555,
|
48 |
+
"grad_norm": 2.53125,
|
49 |
+
"learning_rate": 3.205128205128206e-05,
|
50 |
+
"loss": 2.785,
|
51 |
+
"step": 25
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.28708133971291866,
|
55 |
+
"grad_norm": 1.875,
|
56 |
+
"learning_rate": 3.846153846153846e-05,
|
57 |
+
"loss": 2.6063,
|
58 |
+
"step": 30
|
59 |
+
},
|
60 |
+
{
|
61 |
+
"epoch": 0.3349282296650718,
|
62 |
+
"grad_norm": 2.65625,
|
63 |
+
"learning_rate": 4.4871794871794874e-05,
|
64 |
+
"loss": 2.4641,
|
65 |
+
"step": 35
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"epoch": 0.3827751196172249,
|
69 |
+
"grad_norm": 1.890625,
|
70 |
+
"learning_rate": 5.128205128205128e-05,
|
71 |
+
"loss": 2.3442,
|
72 |
+
"step": 40
|
73 |
+
},
|
74 |
+
{
|
75 |
+
"epoch": 0.430622009569378,
|
76 |
+
"grad_norm": 73.5,
|
77 |
+
"learning_rate": 5.769230769230769e-05,
|
78 |
+
"loss": 2.2165,
|
79 |
+
"step": 45
|
80 |
+
},
|
81 |
+
{
|
82 |
+
"epoch": 0.4784688995215311,
|
83 |
+
"grad_norm": 1.1171875,
|
84 |
+
"learning_rate": 6.410256410256412e-05,
|
85 |
+
"loss": 2.0922,
|
86 |
+
"step": 50
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.5263157894736842,
|
90 |
+
"grad_norm": 0.87890625,
|
91 |
+
"learning_rate": 7.051282051282052e-05,
|
92 |
+
"loss": 1.9553,
|
93 |
+
"step": 55
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.5741626794258373,
|
97 |
+
"grad_norm": 0.89453125,
|
98 |
+
"learning_rate": 7.692307692307693e-05,
|
99 |
+
"loss": 1.8161,
|
100 |
+
"step": 60
|
101 |
+
},
|
102 |
+
{
|
103 |
+
"epoch": 0.6220095693779905,
|
104 |
+
"grad_norm": 0.65234375,
|
105 |
+
"learning_rate": 8.333333333333334e-05,
|
106 |
+
"loss": 1.6939,
|
107 |
+
"step": 65
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"epoch": 0.6698564593301436,
|
111 |
+
"grad_norm": 0.52734375,
|
112 |
+
"learning_rate": 8.974358974358975e-05,
|
113 |
+
"loss": 1.6032,
|
114 |
+
"step": 70
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 0.7177033492822966,
|
118 |
+
"grad_norm": 0.546875,
|
119 |
+
"learning_rate": 9.615384615384617e-05,
|
120 |
+
"loss": 1.5079,
|
121 |
+
"step": 75
|
122 |
+
},
|
123 |
+
{
|
124 |
+
"epoch": 0.7655502392344498,
|
125 |
+
"grad_norm": 0.498046875,
|
126 |
+
"learning_rate": 0.00010256410256410256,
|
127 |
+
"loss": 1.4389,
|
128 |
+
"step": 80
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.8133971291866029,
|
132 |
+
"grad_norm": 0.67578125,
|
133 |
+
"learning_rate": 0.00010897435897435896,
|
134 |
+
"loss": 1.3952,
|
135 |
+
"step": 85
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 0.861244019138756,
|
139 |
+
"grad_norm": 0.302734375,
|
140 |
+
"learning_rate": 0.00011538461538461538,
|
141 |
+
"loss": 1.3286,
|
142 |
+
"step": 90
|
143 |
+
},
|
144 |
+
{
|
145 |
+
"epoch": 0.9090909090909091,
|
146 |
+
"grad_norm": 0.55859375,
|
147 |
+
"learning_rate": 0.00012179487179487179,
|
148 |
+
"loss": 1.3036,
|
149 |
+
"step": 95
|
150 |
+
},
|
151 |
+
{
|
152 |
+
"epoch": 0.9569377990430622,
|
153 |
+
"grad_norm": 0.515625,
|
154 |
+
"learning_rate": 0.00012820512820512823,
|
155 |
+
"loss": 1.2724,
|
156 |
+
"step": 100
|
157 |
+
},
|
158 |
+
{
|
159 |
+
"epoch": 0.9952153110047847,
|
160 |
+
"eval_loss": 2.5530645847320557,
|
161 |
+
"eval_runtime": 0.6584,
|
162 |
+
"eval_samples_per_second": 15.189,
|
163 |
+
"eval_steps_per_second": 1.519,
|
164 |
+
"step": 104
|
165 |
+
},
|
166 |
+
{
|
167 |
+
"epoch": 1.0047846889952152,
|
168 |
+
"grad_norm": 0.31640625,
|
169 |
+
"learning_rate": 0.00013461538461538464,
|
170 |
+
"loss": 1.2461,
|
171 |
+
"step": 105
|
172 |
+
},
|
173 |
+
{
|
174 |
+
"epoch": 1.0526315789473684,
|
175 |
+
"grad_norm": 0.486328125,
|
176 |
+
"learning_rate": 0.00014102564102564104,
|
177 |
+
"loss": 1.2256,
|
178 |
+
"step": 110
|
179 |
+
},
|
180 |
+
{
|
181 |
+
"epoch": 1.1004784688995215,
|
182 |
+
"grad_norm": 0.255859375,
|
183 |
+
"learning_rate": 0.00014743589743589745,
|
184 |
+
"loss": 1.2074,
|
185 |
+
"step": 115
|
186 |
+
},
|
187 |
+
{
|
188 |
+
"epoch": 1.1483253588516746,
|
189 |
+
"grad_norm": 0.396484375,
|
190 |
+
"learning_rate": 0.00015384615384615385,
|
191 |
+
"loss": 1.1911,
|
192 |
+
"step": 120
|
193 |
+
},
|
194 |
+
{
|
195 |
+
"epoch": 1.1961722488038278,
|
196 |
+
"grad_norm": 1.0390625,
|
197 |
+
"learning_rate": 0.00016025641025641028,
|
198 |
+
"loss": 1.1684,
|
199 |
+
"step": 125
|
200 |
+
},
|
201 |
+
{
|
202 |
+
"epoch": 1.244019138755981,
|
203 |
+
"grad_norm": 0.453125,
|
204 |
+
"learning_rate": 0.0001666666666666667,
|
205 |
+
"loss": 1.1506,
|
206 |
+
"step": 130
|
207 |
+
},
|
208 |
+
{
|
209 |
+
"epoch": 1.291866028708134,
|
210 |
+
"grad_norm": 0.56640625,
|
211 |
+
"learning_rate": 0.0001730769230769231,
|
212 |
+
"loss": 1.1572,
|
213 |
+
"step": 135
|
214 |
+
},
|
215 |
+
{
|
216 |
+
"epoch": 1.339712918660287,
|
217 |
+
"grad_norm": 0.75,
|
218 |
+
"learning_rate": 0.0001794871794871795,
|
219 |
+
"loss": 1.136,
|
220 |
+
"step": 140
|
221 |
+
},
|
222 |
+
{
|
223 |
+
"epoch": 1.38755980861244,
|
224 |
+
"grad_norm": 0.7890625,
|
225 |
+
"learning_rate": 0.0001858974358974359,
|
226 |
+
"loss": 1.138,
|
227 |
+
"step": 145
|
228 |
+
},
|
229 |
+
{
|
230 |
+
"epoch": 1.4354066985645932,
|
231 |
+
"grad_norm": 0.2890625,
|
232 |
+
"learning_rate": 0.00019230769230769233,
|
233 |
+
"loss": 1.1339,
|
234 |
+
"step": 150
|
235 |
+
},
|
236 |
+
{
|
237 |
+
"epoch": 1.4832535885167464,
|
238 |
+
"grad_norm": 0.71875,
|
239 |
+
"learning_rate": 0.00019871794871794874,
|
240 |
+
"loss": 1.1258,
|
241 |
+
"step": 155
|
242 |
+
},
|
243 |
+
{
|
244 |
+
"epoch": 1.5311004784688995,
|
245 |
+
"grad_norm": 1.078125,
|
246 |
+
"learning_rate": 0.00019999599453798524,
|
247 |
+
"loss": 1.118,
|
248 |
+
"step": 160
|
249 |
+
},
|
250 |
+
{
|
251 |
+
"epoch": 1.5789473684210527,
|
252 |
+
"grad_norm": 1.1015625,
|
253 |
+
"learning_rate": 0.00019997972289848503,
|
254 |
+
"loss": 1.1211,
|
255 |
+
"step": 165
|
256 |
+
},
|
257 |
+
{
|
258 |
+
"epoch": 1.6267942583732058,
|
259 |
+
"grad_norm": 0.291015625,
|
260 |
+
"learning_rate": 0.0001999509367752813,
|
261 |
+
"loss": 1.109,
|
262 |
+
"step": 170
|
263 |
+
},
|
264 |
+
{
|
265 |
+
"epoch": 1.674641148325359,
|
266 |
+
"grad_norm": 0.27734375,
|
267 |
+
"learning_rate": 0.00019990963977153936,
|
268 |
+
"loss": 1.1016,
|
269 |
+
"step": 175
|
270 |
+
},
|
271 |
+
{
|
272 |
+
"epoch": 1.722488038277512,
|
273 |
+
"grad_norm": 0.37890625,
|
274 |
+
"learning_rate": 0.00019985583705641418,
|
275 |
+
"loss": 1.1047,
|
276 |
+
"step": 180
|
277 |
+
},
|
278 |
+
{
|
279 |
+
"epoch": 1.7703349282296652,
|
280 |
+
"grad_norm": 0.310546875,
|
281 |
+
"learning_rate": 0.00019978953536440336,
|
282 |
+
"loss": 1.0877,
|
283 |
+
"step": 185
|
284 |
+
},
|
285 |
+
{
|
286 |
+
"epoch": 1.8181818181818183,
|
287 |
+
"grad_norm": 0.267578125,
|
288 |
+
"learning_rate": 0.0001997107429945041,
|
289 |
+
"loss": 1.0849,
|
290 |
+
"step": 190
|
291 |
+
},
|
292 |
+
{
|
293 |
+
"epoch": 1.8660287081339713,
|
294 |
+
"grad_norm": 0.306640625,
|
295 |
+
"learning_rate": 0.00019961946980917456,
|
296 |
+
"loss": 1.0945,
|
297 |
+
"step": 195
|
298 |
+
},
|
299 |
+
{
|
300 |
+
"epoch": 1.9138755980861244,
|
301 |
+
"grad_norm": 0.376953125,
|
302 |
+
"learning_rate": 0.0001995157272330992,
|
303 |
+
"loss": 1.0796,
|
304 |
+
"step": 200
|
305 |
+
},
|
306 |
+
{
|
307 |
+
"epoch": 1.9617224880382775,
|
308 |
+
"grad_norm": 0.412109375,
|
309 |
+
"learning_rate": 0.00019939952825175888,
|
310 |
+
"loss": 1.0792,
|
311 |
+
"step": 205
|
312 |
+
},
|
313 |
+
{
|
314 |
+
"epoch": 2.0,
|
315 |
+
"eval_loss": 2.475158214569092,
|
316 |
+
"eval_runtime": 0.5364,
|
317 |
+
"eval_samples_per_second": 18.643,
|
318 |
+
"eval_steps_per_second": 1.864,
|
319 |
+
"step": 209
|
320 |
+
},
|
321 |
+
{
|
322 |
+
"epoch": 2.0095693779904304,
|
323 |
+
"grad_norm": 0.337890625,
|
324 |
+
"learning_rate": 0.0001992708874098054,
|
325 |
+
"loss": 1.0691,
|
326 |
+
"step": 210
|
327 |
+
},
|
328 |
+
{
|
329 |
+
"epoch": 2.0574162679425836,
|
330 |
+
"grad_norm": 0.326171875,
|
331 |
+
"learning_rate": 0.00019912982080924103,
|
332 |
+
"loss": 1.0586,
|
333 |
+
"step": 215
|
334 |
+
},
|
335 |
+
{
|
336 |
+
"epoch": 2.1052631578947367,
|
337 |
+
"grad_norm": 0.31640625,
|
338 |
+
"learning_rate": 0.00019897634610740287,
|
339 |
+
"loss": 1.0492,
|
340 |
+
"step": 220
|
341 |
+
},
|
342 |
+
{
|
343 |
+
"epoch": 2.15311004784689,
|
344 |
+
"grad_norm": 0.33203125,
|
345 |
+
"learning_rate": 0.0001988104825147528,
|
346 |
+
"loss": 1.0467,
|
347 |
+
"step": 225
|
348 |
+
},
|
349 |
+
{
|
350 |
+
"epoch": 2.200956937799043,
|
351 |
+
"grad_norm": 0.80078125,
|
352 |
+
"learning_rate": 0.00019863225079247285,
|
353 |
+
"loss": 1.0492,
|
354 |
+
"step": 230
|
355 |
+
},
|
356 |
+
{
|
357 |
+
"epoch": 2.248803827751196,
|
358 |
+
"grad_norm": 0.37890625,
|
359 |
+
"learning_rate": 0.00019844167324986657,
|
360 |
+
"loss": 1.0444,
|
361 |
+
"step": 235
|
362 |
+
},
|
363 |
+
{
|
364 |
+
"epoch": 2.2966507177033493,
|
365 |
+
"grad_norm": 0.396484375,
|
366 |
+
"learning_rate": 0.00019823877374156647,
|
367 |
+
"loss": 1.049,
|
368 |
+
"step": 240
|
369 |
+
},
|
370 |
+
{
|
371 |
+
"epoch": 2.3444976076555024,
|
372 |
+
"grad_norm": 0.353515625,
|
373 |
+
"learning_rate": 0.00019802357766454827,
|
374 |
+
"loss": 1.047,
|
375 |
+
"step": 245
|
376 |
+
},
|
377 |
+
{
|
378 |
+
"epoch": 2.3923444976076556,
|
379 |
+
"grad_norm": 0.57421875,
|
380 |
+
"learning_rate": 0.00019779611195495177,
|
381 |
+
"loss": 1.044,
|
382 |
+
"step": 250
|
383 |
+
},
|
384 |
+
{
|
385 |
+
"epoch": 2.4401913875598087,
|
386 |
+
"grad_norm": 0.341796875,
|
387 |
+
"learning_rate": 0.00019755640508470942,
|
388 |
+
"loss": 1.0424,
|
389 |
+
"step": 255
|
390 |
+
},
|
391 |
+
{
|
392 |
+
"epoch": 2.488038277511962,
|
393 |
+
"grad_norm": 0.3125,
|
394 |
+
"learning_rate": 0.00019730448705798239,
|
395 |
+
"loss": 1.0362,
|
396 |
+
"step": 260
|
397 |
+
},
|
398 |
+
{
|
399 |
+
"epoch": 2.535885167464115,
|
400 |
+
"grad_norm": 0.423828125,
|
401 |
+
"learning_rate": 0.00019704038940740505,
|
402 |
+
"loss": 1.031,
|
403 |
+
"step": 265
|
404 |
+
},
|
405 |
+
{
|
406 |
+
"epoch": 2.583732057416268,
|
407 |
+
"grad_norm": 0.57421875,
|
408 |
+
"learning_rate": 0.00019676414519013781,
|
409 |
+
"loss": 1.0416,
|
410 |
+
"step": 270
|
411 |
+
},
|
412 |
+
{
|
413 |
+
"epoch": 2.6315789473684212,
|
414 |
+
"grad_norm": 1.0859375,
|
415 |
+
"learning_rate": 0.0001964757889837296,
|
416 |
+
"loss": 1.0326,
|
417 |
+
"step": 275
|
418 |
+
},
|
419 |
+
{
|
420 |
+
"epoch": 2.679425837320574,
|
421 |
+
"grad_norm": 0.408203125,
|
422 |
+
"learning_rate": 0.0001961753568817896,
|
423 |
+
"loss": 1.0317,
|
424 |
+
"step": 280
|
425 |
+
},
|
426 |
+
{
|
427 |
+
"epoch": 2.7272727272727275,
|
428 |
+
"grad_norm": 0.3828125,
|
429 |
+
"learning_rate": 0.00019586288648946947,
|
430 |
+
"loss": 1.0449,
|
431 |
+
"step": 285
|
432 |
+
},
|
433 |
+
{
|
434 |
+
"epoch": 2.77511961722488,
|
435 |
+
"grad_norm": 0.396484375,
|
436 |
+
"learning_rate": 0.0001955384169187563,
|
437 |
+
"loss": 1.0297,
|
438 |
+
"step": 290
|
439 |
+
},
|
440 |
+
{
|
441 |
+
"epoch": 2.8229665071770333,
|
442 |
+
"grad_norm": 0.55078125,
|
443 |
+
"learning_rate": 0.00019520198878357703,
|
444 |
+
"loss": 1.0319,
|
445 |
+
"step": 295
|
446 |
+
},
|
447 |
+
{
|
448 |
+
"epoch": 2.8708133971291865,
|
449 |
+
"grad_norm": 0.5625,
|
450 |
+
"learning_rate": 0.00019485364419471454,
|
451 |
+
"loss": 1.031,
|
452 |
+
"step": 300
|
453 |
+
},
|
454 |
+
{
|
455 |
+
"epoch": 2.9186602870813396,
|
456 |
+
"grad_norm": 0.3359375,
|
457 |
+
"learning_rate": 0.00019449342675453707,
|
458 |
+
"loss": 1.0256,
|
459 |
+
"step": 305
|
460 |
+
},
|
461 |
+
{
|
462 |
+
"epoch": 2.9665071770334928,
|
463 |
+
"grad_norm": 0.3359375,
|
464 |
+
"learning_rate": 0.00019412138155154002,
|
465 |
+
"loss": 1.0312,
|
466 |
+
"step": 310
|
467 |
+
},
|
468 |
+
{
|
469 |
+
"epoch": 2.9952153110047846,
|
470 |
+
"eval_loss": 2.463235855102539,
|
471 |
+
"eval_runtime": 0.6563,
|
472 |
+
"eval_samples_per_second": 15.237,
|
473 |
+
"eval_steps_per_second": 1.524,
|
474 |
+
"step": 313
|
475 |
+
},
|
476 |
+
{
|
477 |
+
"epoch": 3.014354066985646,
|
478 |
+
"grad_norm": 0.37890625,
|
479 |
+
"learning_rate": 0.00019373755515470254,
|
480 |
+
"loss": 1.0296,
|
481 |
+
"step": 315
|
482 |
+
},
|
483 |
+
{
|
484 |
+
"epoch": 3.062200956937799,
|
485 |
+
"grad_norm": 0.330078125,
|
486 |
+
"learning_rate": 0.0001933419956076584,
|
487 |
+
"loss": 1.0058,
|
488 |
+
"step": 320
|
489 |
+
},
|
490 |
+
{
|
491 |
+
"epoch": 3.110047846889952,
|
492 |
+
"grad_norm": 0.34375,
|
493 |
+
"learning_rate": 0.00019293475242268223,
|
494 |
+
"loss": 1.0065,
|
495 |
+
"step": 325
|
496 |
+
},
|
497 |
+
{
|
498 |
+
"epoch": 3.1578947368421053,
|
499 |
+
"grad_norm": 0.40625,
|
500 |
+
"learning_rate": 0.00019251587657449236,
|
501 |
+
"loss": 1.0095,
|
502 |
+
"step": 330
|
503 |
+
},
|
504 |
+
{
|
505 |
+
"epoch": 3.2057416267942584,
|
506 |
+
"grad_norm": 0.8203125,
|
507 |
+
"learning_rate": 0.0001920854204938699,
|
508 |
+
"loss": 1.0087,
|
509 |
+
"step": 335
|
510 |
+
},
|
511 |
+
{
|
512 |
+
"epoch": 3.2535885167464116,
|
513 |
+
"grad_norm": 0.353515625,
|
514 |
+
"learning_rate": 0.00019164343806109632,
|
515 |
+
"loss": 1.0066,
|
516 |
+
"step": 340
|
517 |
+
},
|
518 |
+
{
|
519 |
+
"epoch": 3.3014354066985647,
|
520 |
+
"grad_norm": 0.486328125,
|
521 |
+
"learning_rate": 0.00019118998459920902,
|
522 |
+
"loss": 1.002,
|
523 |
+
"step": 345
|
524 |
+
},
|
525 |
+
{
|
526 |
+
"epoch": 3.349282296650718,
|
527 |
+
"grad_norm": 0.6484375,
|
528 |
+
"learning_rate": 0.00019072511686707663,
|
529 |
+
"loss": 1.0099,
|
530 |
+
"step": 350
|
531 |
+
},
|
532 |
+
{
|
533 |
+
"epoch": 3.397129186602871,
|
534 |
+
"grad_norm": 0.3671875,
|
535 |
+
"learning_rate": 0.00019024889305229456,
|
536 |
+
"loss": 0.9971,
|
537 |
+
"step": 355
|
538 |
+
},
|
539 |
+
{
|
540 |
+
"epoch": 3.444976076555024,
|
541 |
+
"grad_norm": 0.306640625,
|
542 |
+
"learning_rate": 0.0001897613727639014,
|
543 |
+
"loss": 0.9993,
|
544 |
+
"step": 360
|
545 |
+
},
|
546 |
+
{
|
547 |
+
"epoch": 3.492822966507177,
|
548 |
+
"grad_norm": 0.318359375,
|
549 |
+
"learning_rate": 0.00018926261702491797,
|
550 |
+
"loss": 1.0025,
|
551 |
+
"step": 365
|
552 |
+
},
|
553 |
+
{
|
554 |
+
"epoch": 3.5406698564593304,
|
555 |
+
"grad_norm": 0.46875,
|
556 |
+
"learning_rate": 0.00018875268826470872,
|
557 |
+
"loss": 0.9953,
|
558 |
+
"step": 370
|
559 |
+
},
|
560 |
+
{
|
561 |
+
"epoch": 3.588516746411483,
|
562 |
+
"grad_norm": 0.470703125,
|
563 |
+
"learning_rate": 0.0001882316503111678,
|
564 |
+
"loss": 0.9988,
|
565 |
+
"step": 375
|
566 |
+
},
|
567 |
+
{
|
568 |
+
"epoch": 3.6363636363636362,
|
569 |
+
"grad_norm": 0.71484375,
|
570 |
+
"learning_rate": 0.00018769956838272936,
|
571 |
+
"loss": 1.0103,
|
572 |
+
"step": 380
|
573 |
+
},
|
574 |
+
{
|
575 |
+
"epoch": 3.6842105263157894,
|
576 |
+
"grad_norm": 0.6171875,
|
577 |
+
"learning_rate": 0.00018715650908020427,
|
578 |
+
"loss": 1.0031,
|
579 |
+
"step": 385
|
580 |
+
},
|
581 |
+
{
|
582 |
+
"epoch": 3.7320574162679425,
|
583 |
+
"grad_norm": 0.330078125,
|
584 |
+
"learning_rate": 0.00018660254037844388,
|
585 |
+
"loss": 0.9989,
|
586 |
+
"step": 390
|
587 |
+
},
|
588 |
+
{
|
589 |
+
"epoch": 3.7799043062200957,
|
590 |
+
"grad_norm": 0.396484375,
|
591 |
+
"learning_rate": 0.00018603773161783124,
|
592 |
+
"loss": 0.9975,
|
593 |
+
"step": 395
|
594 |
+
},
|
595 |
+
{
|
596 |
+
"epoch": 3.827751196172249,
|
597 |
+
"grad_norm": 0.482421875,
|
598 |
+
"learning_rate": 0.00018546215349560203,
|
599 |
+
"loss": 0.9895,
|
600 |
+
"step": 400
|
601 |
+
},
|
602 |
+
{
|
603 |
+
"epoch": 3.875598086124402,
|
604 |
+
"grad_norm": 0.56640625,
|
605 |
+
"learning_rate": 0.00018487587805699526,
|
606 |
+
"loss": 0.9941,
|
607 |
+
"step": 405
|
608 |
+
},
|
609 |
+
{
|
610 |
+
"epoch": 3.923444976076555,
|
611 |
+
"grad_norm": 0.65625,
|
612 |
+
"learning_rate": 0.00018427897868623534,
|
613 |
+
"loss": 0.9996,
|
614 |
+
"step": 410
|
615 |
+
},
|
616 |
+
{
|
617 |
+
"epoch": 3.971291866028708,
|
618 |
+
"grad_norm": 0.8046875,
|
619 |
+
"learning_rate": 0.00018367153009734655,
|
620 |
+
"loss": 0.9957,
|
621 |
+
"step": 415
|
622 |
+
},
|
623 |
+
{
|
624 |
+
"epoch": 4.0,
|
625 |
+
"eval_loss": 2.470475912094116,
|
626 |
+
"eval_runtime": 0.5362,
|
627 |
+
"eval_samples_per_second": 18.649,
|
628 |
+
"eval_steps_per_second": 1.865,
|
629 |
+
"step": 418
|
630 |
+
},
|
631 |
+
{
|
632 |
+
"epoch": 4.019138755980861,
|
633 |
+
"grad_norm": 0.59375,
|
634 |
+
"learning_rate": 0.00018305360832480117,
|
635 |
+
"loss": 0.9875,
|
636 |
+
"step": 420
|
637 |
+
},
|
638 |
+
{
|
639 |
+
"epoch": 4.0669856459330145,
|
640 |
+
"grad_norm": 0.70703125,
|
641 |
+
"learning_rate": 0.00018242529071400214,
|
642 |
+
"loss": 0.9719,
|
643 |
+
"step": 425
|
644 |
+
},
|
645 |
+
{
|
646 |
+
"epoch": 4.114832535885167,
|
647 |
+
"grad_norm": 0.408203125,
|
648 |
+
"learning_rate": 0.00018178665591160172,
|
649 |
+
"loss": 0.9655,
|
650 |
+
"step": 430
|
651 |
+
},
|
652 |
+
{
|
653 |
+
"epoch": 4.162679425837321,
|
654 |
+
"grad_norm": 0.427734375,
|
655 |
+
"learning_rate": 0.00018113778385565733,
|
656 |
+
"loss": 0.9659,
|
657 |
+
"step": 435
|
658 |
+
},
|
659 |
+
{
|
660 |
+
"epoch": 4.2105263157894735,
|
661 |
+
"grad_norm": 0.625,
|
662 |
+
"learning_rate": 0.00018047875576562557,
|
663 |
+
"loss": 0.9758,
|
664 |
+
"step": 440
|
665 |
+
},
|
666 |
+
{
|
667 |
+
"epoch": 4.258373205741627,
|
668 |
+
"grad_norm": 0.55859375,
|
669 |
+
"learning_rate": 0.0001798096541321961,
|
670 |
+
"loss": 0.983,
|
671 |
+
"step": 445
|
672 |
+
},
|
673 |
+
{
|
674 |
+
"epoch": 4.30622009569378,
|
675 |
+
"grad_norm": 0.390625,
|
676 |
+
"learning_rate": 0.0001791305627069662,
|
677 |
+
"loss": 0.979,
|
678 |
+
"step": 450
|
679 |
+
},
|
680 |
+
{
|
681 |
+
"epoch": 4.354066985645933,
|
682 |
+
"grad_norm": 0.35546875,
|
683 |
+
"learning_rate": 0.00017844156649195759,
|
684 |
+
"loss": 0.9764,
|
685 |
+
"step": 455
|
686 |
+
},
|
687 |
+
{
|
688 |
+
"epoch": 4.401913875598086,
|
689 |
+
"grad_norm": 0.4453125,
|
690 |
+
"learning_rate": 0.0001777427517289766,
|
691 |
+
"loss": 0.9775,
|
692 |
+
"step": 460
|
693 |
+
},
|
694 |
+
{
|
695 |
+
"epoch": 4.44976076555024,
|
696 |
+
"grad_norm": 0.80078125,
|
697 |
+
"learning_rate": 0.00017703420588881946,
|
698 |
+
"loss": 0.9746,
|
699 |
+
"step": 465
|
700 |
+
},
|
701 |
+
{
|
702 |
+
"epoch": 4.497607655502392,
|
703 |
+
"grad_norm": 0.373046875,
|
704 |
+
"learning_rate": 0.00017631601766032336,
|
705 |
+
"loss": 0.972,
|
706 |
+
"step": 470
|
707 |
+
},
|
708 |
+
{
|
709 |
+
"epoch": 4.545454545454545,
|
710 |
+
"grad_norm": 0.44140625,
|
711 |
+
"learning_rate": 0.00017558827693926534,
|
712 |
+
"loss": 0.9814,
|
713 |
+
"step": 475
|
714 |
+
},
|
715 |
+
{
|
716 |
+
"epoch": 4.5933014354066986,
|
717 |
+
"grad_norm": 0.32421875,
|
718 |
+
"learning_rate": 0.00017485107481711012,
|
719 |
+
"loss": 0.9813,
|
720 |
+
"step": 480
|
721 |
+
},
|
722 |
+
{
|
723 |
+
"epoch": 4.641148325358852,
|
724 |
+
"grad_norm": 0.408203125,
|
725 |
+
"learning_rate": 0.00017410450356960795,
|
726 |
+
"loss": 0.9811,
|
727 |
+
"step": 485
|
728 |
+
},
|
729 |
+
{
|
730 |
+
"epoch": 4.688995215311005,
|
731 |
+
"grad_norm": 0.3515625,
|
732 |
+
"learning_rate": 0.0001733486566452446,
|
733 |
+
"loss": 0.9705,
|
734 |
+
"step": 490
|
735 |
+
},
|
736 |
+
{
|
737 |
+
"epoch": 4.7368421052631575,
|
738 |
+
"grad_norm": 0.353515625,
|
739 |
+
"learning_rate": 0.00017258362865354426,
|
740 |
+
"loss": 0.985,
|
741 |
+
"step": 495
|
742 |
+
},
|
743 |
+
{
|
744 |
+
"epoch": 4.784688995215311,
|
745 |
+
"grad_norm": 0.380859375,
|
746 |
+
"learning_rate": 0.0001718095153532274,
|
747 |
+
"loss": 0.9817,
|
748 |
+
"step": 500
|
749 |
+
},
|
750 |
+
{
|
751 |
+
"epoch": 4.832535885167464,
|
752 |
+
"grad_norm": 0.474609375,
|
753 |
+
"learning_rate": 0.00017102641364022457,
|
754 |
+
"loss": 0.9686,
|
755 |
+
"step": 505
|
756 |
+
},
|
757 |
+
{
|
758 |
+
"epoch": 4.880382775119617,
|
759 |
+
"grad_norm": 0.384765625,
|
760 |
+
"learning_rate": 0.00017023442153554777,
|
761 |
+
"loss": 0.9838,
|
762 |
+
"step": 510
|
763 |
+
},
|
764 |
+
{
|
765 |
+
"epoch": 4.92822966507177,
|
766 |
+
"grad_norm": 0.53515625,
|
767 |
+
"learning_rate": 0.00016943363817302135,
|
768 |
+
"loss": 0.9774,
|
769 |
+
"step": 515
|
770 |
+
},
|
771 |
+
{
|
772 |
+
"epoch": 4.976076555023924,
|
773 |
+
"grad_norm": 0.50390625,
|
774 |
+
"learning_rate": 0.0001686241637868734,
|
775 |
+
"loss": 0.9775,
|
776 |
+
"step": 520
|
777 |
+
},
|
778 |
+
{
|
779 |
+
"epoch": 4.9952153110047846,
|
780 |
+
"eval_loss": 2.476422071456909,
|
781 |
+
"eval_runtime": 0.6478,
|
782 |
+
"eval_samples_per_second": 15.436,
|
783 |
+
"eval_steps_per_second": 1.544,
|
784 |
+
"step": 522
|
785 |
+
},
|
786 |
+
{
|
787 |
+
"epoch": 5.023923444976076,
|
788 |
+
"grad_norm": 0.625,
|
789 |
+
"learning_rate": 0.0001678060996991891,
|
790 |
+
"loss": 0.9684,
|
791 |
+
"step": 525
|
792 |
+
},
|
793 |
+
{
|
794 |
+
"epoch": 5.07177033492823,
|
795 |
+
"grad_norm": 0.33203125,
|
796 |
+
"learning_rate": 0.00016697954830722868,
|
797 |
+
"loss": 0.9614,
|
798 |
+
"step": 530
|
799 |
+
},
|
800 |
+
{
|
801 |
+
"epoch": 5.119617224880383,
|
802 |
+
"grad_norm": 0.376953125,
|
803 |
+
"learning_rate": 0.00016614461307061,
|
804 |
+
"loss": 0.963,
|
805 |
+
"step": 535
|
806 |
+
},
|
807 |
+
{
|
808 |
+
"epoch": 5.167464114832536,
|
809 |
+
"grad_norm": 0.439453125,
|
810 |
+
"learning_rate": 0.0001653013984983585,
|
811 |
+
"loss": 0.9558,
|
812 |
+
"step": 540
|
813 |
+
},
|
814 |
+
{
|
815 |
+
"epoch": 5.215311004784689,
|
816 |
+
"grad_norm": 0.375,
|
817 |
+
"learning_rate": 0.00016445001013582608,
|
818 |
+
"loss": 0.9533,
|
819 |
+
"step": 545
|
820 |
+
},
|
821 |
+
{
|
822 |
+
"epoch": 5.2631578947368425,
|
823 |
+
"grad_norm": 0.40234375,
|
824 |
+
"learning_rate": 0.0001635905545514795,
|
825 |
+
"loss": 0.9664,
|
826 |
+
"step": 550
|
827 |
+
},
|
828 |
+
{
|
829 |
+
"epoch": 5.311004784688995,
|
830 |
+
"grad_norm": 0.380859375,
|
831 |
+
"learning_rate": 0.00016272313932356162,
|
832 |
+
"loss": 0.9552,
|
833 |
+
"step": 555
|
834 |
+
},
|
835 |
+
{
|
836 |
+
"epoch": 5.358851674641148,
|
837 |
+
"grad_norm": 0.416015625,
|
838 |
+
"learning_rate": 0.0001618478730266255,
|
839 |
+
"loss": 0.9505,
|
840 |
+
"step": 560
|
841 |
+
},
|
842 |
+
{
|
843 |
+
"epoch": 5.4066985645933014,
|
844 |
+
"grad_norm": 0.39453125,
|
845 |
+
"learning_rate": 0.00016096486521794434,
|
846 |
+
"loss": 0.964,
|
847 |
+
"step": 565
|
848 |
+
},
|
849 |
+
{
|
850 |
+
"epoch": 5.454545454545454,
|
851 |
+
"grad_norm": 0.416015625,
|
852 |
+
"learning_rate": 0.0001600742264237979,
|
853 |
+
"loss": 0.9558,
|
854 |
+
"step": 570
|
855 |
+
},
|
856 |
+
{
|
857 |
+
"epoch": 5.502392344497608,
|
858 |
+
"grad_norm": 0.5625,
|
859 |
+
"learning_rate": 0.0001591760681256382,
|
860 |
+
"loss": 0.9573,
|
861 |
+
"step": 575
|
862 |
+
},
|
863 |
+
{
|
864 |
+
"epoch": 5.55023923444976,
|
865 |
+
"grad_norm": 0.3515625,
|
866 |
+
"learning_rate": 0.00015827050274613513,
|
867 |
+
"loss": 0.9587,
|
868 |
+
"step": 580
|
869 |
+
},
|
870 |
+
{
|
871 |
+
"epoch": 5.598086124401914,
|
872 |
+
"grad_norm": 0.421875,
|
873 |
+
"learning_rate": 0.0001573576436351046,
|
874 |
+
"loss": 0.9638,
|
875 |
+
"step": 585
|
876 |
+
},
|
877 |
+
{
|
878 |
+
"epoch": 5.645933014354067,
|
879 |
+
"grad_norm": 0.4140625,
|
880 |
+
"learning_rate": 0.0001564376050553205,
|
881 |
+
"loss": 0.9513,
|
882 |
+
"step": 590
|
883 |
+
},
|
884 |
+
{
|
885 |
+
"epoch": 5.69377990430622,
|
886 |
+
"grad_norm": 0.62109375,
|
887 |
+
"learning_rate": 0.0001555105021682123,
|
888 |
+
"loss": 0.9627,
|
889 |
+
"step": 595
|
890 |
+
},
|
891 |
+
{
|
892 |
+
"epoch": 5.741626794258373,
|
893 |
+
"grad_norm": 0.62890625,
|
894 |
+
"learning_rate": 0.00015457645101945046,
|
895 |
+
"loss": 0.9546,
|
896 |
+
"step": 600
|
897 |
+
},
|
898 |
+
{
|
899 |
+
"epoch": 5.7894736842105265,
|
900 |
+
"grad_norm": 0.361328125,
|
901 |
+
"learning_rate": 0.00015363556852442085,
|
902 |
+
"loss": 0.9481,
|
903 |
+
"step": 605
|
904 |
+
},
|
905 |
+
{
|
906 |
+
"epoch": 5.837320574162679,
|
907 |
+
"grad_norm": 0.462890625,
|
908 |
+
"learning_rate": 0.00015268797245359035,
|
909 |
+
"loss": 0.9629,
|
910 |
+
"step": 610
|
911 |
+
},
|
912 |
+
{
|
913 |
+
"epoch": 5.885167464114833,
|
914 |
+
"grad_norm": 0.33984375,
|
915 |
+
"learning_rate": 0.00015173378141776568,
|
916 |
+
"loss": 0.9519,
|
917 |
+
"step": 615
|
918 |
+
},
|
919 |
+
{
|
920 |
+
"epoch": 5.9330143540669855,
|
921 |
+
"grad_norm": 0.423828125,
|
922 |
+
"learning_rate": 0.0001507731148532468,
|
923 |
+
"loss": 0.9584,
|
924 |
+
"step": 620
|
925 |
+
},
|
926 |
+
{
|
927 |
+
"epoch": 5.980861244019139,
|
928 |
+
"grad_norm": 0.3828125,
|
929 |
+
"learning_rate": 0.00014980609300687683,
|
930 |
+
"loss": 0.9584,
|
931 |
+
"step": 625
|
932 |
+
},
|
933 |
+
{
|
934 |
+
"epoch": 6.0,
|
935 |
+
"eval_loss": 2.487025737762451,
|
936 |
+
"eval_runtime": 0.5347,
|
937 |
+
"eval_samples_per_second": 18.7,
|
938 |
+
"eval_steps_per_second": 1.87,
|
939 |
+
"step": 627
|
940 |
+
},
|
941 |
+
{
|
942 |
+
"epoch": 6.028708133971292,
|
943 |
+
"grad_norm": 0.361328125,
|
944 |
+
"learning_rate": 0.00014883283692099112,
|
945 |
+
"loss": 0.9435,
|
946 |
+
"step": 630
|
947 |
+
},
|
948 |
+
{
|
949 |
+
"epoch": 6.076555023923445,
|
950 |
+
"grad_norm": 0.388671875,
|
951 |
+
"learning_rate": 0.000147853468418266,
|
952 |
+
"loss": 0.9462,
|
953 |
+
"step": 635
|
954 |
+
},
|
955 |
+
{
|
956 |
+
"epoch": 6.124401913875598,
|
957 |
+
"grad_norm": 0.443359375,
|
958 |
+
"learning_rate": 0.00014686811008647038,
|
959 |
+
"loss": 0.9372,
|
960 |
+
"step": 640
|
961 |
+
},
|
962 |
+
{
|
963 |
+
"epoch": 6.172248803827751,
|
964 |
+
"grad_norm": 0.37109375,
|
965 |
+
"learning_rate": 0.00014587688526312143,
|
966 |
+
"loss": 0.9405,
|
967 |
+
"step": 645
|
968 |
+
},
|
969 |
+
{
|
970 |
+
"epoch": 6.220095693779904,
|
971 |
+
"grad_norm": 0.474609375,
|
972 |
+
"learning_rate": 0.00014487991802004623,
|
973 |
+
"loss": 0.942,
|
974 |
+
"step": 650
|
975 |
+
},
|
976 |
+
{
|
977 |
+
"epoch": 6.267942583732057,
|
978 |
+
"grad_norm": 0.3984375,
|
979 |
+
"learning_rate": 0.00014387733314785193,
|
980 |
+
"loss": 0.9495,
|
981 |
+
"step": 655
|
982 |
+
},
|
983 |
+
{
|
984 |
+
"epoch": 6.315789473684211,
|
985 |
+
"grad_norm": 0.38671875,
|
986 |
+
"learning_rate": 0.00014286925614030542,
|
987 |
+
"loss": 0.9415,
|
988 |
+
"step": 660
|
989 |
+
},
|
990 |
+
{
|
991 |
+
"epoch": 6.363636363636363,
|
992 |
+
"grad_norm": 0.390625,
|
993 |
+
"learning_rate": 0.00014185581317862546,
|
994 |
+
"loss": 0.94,
|
995 |
+
"step": 665
|
996 |
+
},
|
997 |
+
{
|
998 |
+
"epoch": 6.411483253588517,
|
999 |
+
"grad_norm": 0.48046875,
|
1000 |
+
"learning_rate": 0.00014083713111568842,
|
1001 |
+
"loss": 0.9344,
|
1002 |
+
"step": 670
|
1003 |
+
},
|
1004 |
+
{
|
1005 |
+
"epoch": 6.45933014354067,
|
1006 |
+
"grad_norm": 0.439453125,
|
1007 |
+
"learning_rate": 0.0001398133374601501,
|
1008 |
+
"loss": 0.9438,
|
1009 |
+
"step": 675
|
1010 |
+
},
|
1011 |
+
{
|
1012 |
+
"epoch": 6.507177033492823,
|
1013 |
+
"grad_norm": 0.333984375,
|
1014 |
+
"learning_rate": 0.0001387845603604855,
|
1015 |
+
"loss": 0.941,
|
1016 |
+
"step": 680
|
1017 |
+
},
|
1018 |
+
{
|
1019 |
+
"epoch": 6.555023923444976,
|
1020 |
+
"grad_norm": 0.5078125,
|
1021 |
+
"learning_rate": 0.00013775092858894837,
|
1022 |
+
"loss": 0.9433,
|
1023 |
+
"step": 685
|
1024 |
+
},
|
1025 |
+
{
|
1026 |
+
"epoch": 6.6028708133971294,
|
1027 |
+
"grad_norm": 0.3515625,
|
1028 |
+
"learning_rate": 0.00013671257152545277,
|
1029 |
+
"loss": 0.9433,
|
1030 |
+
"step": 690
|
1031 |
+
},
|
1032 |
+
{
|
1033 |
+
"epoch": 6.650717703349282,
|
1034 |
+
"grad_norm": 0.451171875,
|
1035 |
+
"learning_rate": 0.00013566961914137867,
|
1036 |
+
"loss": 0.9435,
|
1037 |
+
"step": 695
|
1038 |
+
},
|
1039 |
+
{
|
1040 |
+
"epoch": 6.698564593301436,
|
1041 |
+
"grad_norm": 0.41015625,
|
1042 |
+
"learning_rate": 0.00013462220198330328,
|
1043 |
+
"loss": 0.9431,
|
1044 |
+
"step": 700
|
1045 |
+
},
|
1046 |
+
{
|
1047 |
+
"epoch": 6.746411483253588,
|
1048 |
+
"grad_norm": 0.44140625,
|
1049 |
+
"learning_rate": 0.0001335704511566605,
|
1050 |
+
"loss": 0.9409,
|
1051 |
+
"step": 705
|
1052 |
+
},
|
1053 |
+
{
|
1054 |
+
"epoch": 6.794258373205742,
|
1055 |
+
"grad_norm": 0.546875,
|
1056 |
+
"learning_rate": 0.0001325144983093305,
|
1057 |
+
"loss": 0.9437,
|
1058 |
+
"step": 710
|
1059 |
+
},
|
1060 |
+
{
|
1061 |
+
"epoch": 6.842105263157895,
|
1062 |
+
"grad_norm": 0.412109375,
|
1063 |
+
"learning_rate": 0.00013145447561516138,
|
1064 |
+
"loss": 0.9491,
|
1065 |
+
"step": 715
|
1066 |
+
},
|
1067 |
+
{
|
1068 |
+
"epoch": 6.889952153110048,
|
1069 |
+
"grad_norm": 0.38671875,
|
1070 |
+
"learning_rate": 0.0001303905157574247,
|
1071 |
+
"loss": 0.9445,
|
1072 |
+
"step": 720
|
1073 |
+
},
|
1074 |
+
{
|
1075 |
+
"epoch": 6.937799043062201,
|
1076 |
+
"grad_norm": 0.380859375,
|
1077 |
+
"learning_rate": 0.00012932275191220776,
|
1078 |
+
"loss": 0.9315,
|
1079 |
+
"step": 725
|
1080 |
+
},
|
1081 |
+
{
|
1082 |
+
"epoch": 6.985645933014354,
|
1083 |
+
"grad_norm": 0.36328125,
|
1084 |
+
"learning_rate": 0.0001282513177317437,
|
1085 |
+
"loss": 0.9368,
|
1086 |
+
"step": 730
|
1087 |
+
},
|
1088 |
+
{
|
1089 |
+
"epoch": 6.9952153110047846,
|
1090 |
+
"eval_loss": 2.497544765472412,
|
1091 |
+
"eval_runtime": 0.6485,
|
1092 |
+
"eval_samples_per_second": 15.419,
|
1093 |
+
"eval_steps_per_second": 1.542,
|
1094 |
+
"step": 731
|
1095 |
+
},
|
1096 |
+
{
|
1097 |
+
"epoch": 7.033492822966507,
|
1098 |
+
"grad_norm": 0.69921875,
|
1099 |
+
"learning_rate": 0.00012717634732768243,
|
1100 |
+
"loss": 0.9238,
|
1101 |
+
"step": 735
|
1102 |
+
},
|
1103 |
+
{
|
1104 |
+
"epoch": 7.08133971291866,
|
1105 |
+
"grad_norm": 0.39453125,
|
1106 |
+
"learning_rate": 0.00012609797525430373,
|
1107 |
+
"loss": 0.9235,
|
1108 |
+
"step": 740
|
1109 |
+
},
|
1110 |
+
{
|
1111 |
+
"epoch": 7.1291866028708135,
|
1112 |
+
"grad_norm": 0.3984375,
|
1113 |
+
"learning_rate": 0.00012501633649167495,
|
1114 |
+
"loss": 0.9148,
|
1115 |
+
"step": 745
|
1116 |
+
},
|
1117 |
+
{
|
1118 |
+
"epoch": 7.177033492822966,
|
1119 |
+
"grad_norm": 0.455078125,
|
1120 |
+
"learning_rate": 0.0001239315664287558,
|
1121 |
+
"loss": 0.9182,
|
1122 |
+
"step": 750
|
1123 |
+
},
|
1124 |
+
{
|
1125 |
+
"epoch": 7.22488038277512,
|
1126 |
+
"grad_norm": 0.435546875,
|
1127 |
+
"learning_rate": 0.00012284380084645139,
|
1128 |
+
"loss": 0.9237,
|
1129 |
+
"step": 755
|
1130 |
+
},
|
1131 |
+
{
|
1132 |
+
"epoch": 7.2727272727272725,
|
1133 |
+
"grad_norm": 0.443359375,
|
1134 |
+
"learning_rate": 0.00012175317590061674,
|
1135 |
+
"loss": 0.93,
|
1136 |
+
"step": 760
|
1137 |
+
},
|
1138 |
+
{
|
1139 |
+
"epoch": 7.320574162679426,
|
1140 |
+
"grad_norm": 0.75,
|
1141 |
+
"learning_rate": 0.00012065982810501404,
|
1142 |
+
"loss": 0.9268,
|
1143 |
+
"step": 765
|
1144 |
+
},
|
1145 |
+
{
|
1146 |
+
"epoch": 7.368421052631579,
|
1147 |
+
"grad_norm": 0.48828125,
|
1148 |
+
"learning_rate": 0.00011956389431422507,
|
1149 |
+
"loss": 0.9214,
|
1150 |
+
"step": 770
|
1151 |
+
},
|
1152 |
+
{
|
1153 |
+
"epoch": 7.416267942583732,
|
1154 |
+
"grad_norm": 0.6015625,
|
1155 |
+
"learning_rate": 0.00011846551170652127,
|
1156 |
+
"loss": 0.931,
|
1157 |
+
"step": 775
|
1158 |
+
},
|
1159 |
+
{
|
1160 |
+
"epoch": 7.464114832535885,
|
1161 |
+
"grad_norm": 0.453125,
|
1162 |
+
"learning_rate": 0.00011736481776669306,
|
1163 |
+
"loss": 0.9316,
|
1164 |
+
"step": 780
|
1165 |
+
},
|
1166 |
+
{
|
1167 |
+
"epoch": 7.511961722488039,
|
1168 |
+
"grad_norm": 0.369140625,
|
1169 |
+
"learning_rate": 0.0001162619502688407,
|
1170 |
+
"loss": 0.9272,
|
1171 |
+
"step": 785
|
1172 |
+
},
|
1173 |
+
{
|
1174 |
+
"epoch": 7.559808612440191,
|
1175 |
+
"grad_norm": 0.435546875,
|
1176 |
+
"learning_rate": 0.00011515704725912926,
|
1177 |
+
"loss": 0.9219,
|
1178 |
+
"step": 790
|
1179 |
+
},
|
1180 |
+
{
|
1181 |
+
"epoch": 7.607655502392344,
|
1182 |
+
"grad_norm": 0.42578125,
|
1183 |
+
"learning_rate": 0.00011405024703850929,
|
1184 |
+
"loss": 0.9363,
|
1185 |
+
"step": 795
|
1186 |
+
},
|
1187 |
+
{
|
1188 |
+
"epoch": 7.655502392344498,
|
1189 |
+
"grad_norm": 0.5390625,
|
1190 |
+
"learning_rate": 0.00011294168814540553,
|
1191 |
+
"loss": 0.9388,
|
1192 |
+
"step": 800
|
1193 |
+
},
|
1194 |
+
{
|
1195 |
+
"epoch": 7.703349282296651,
|
1196 |
+
"grad_norm": 0.431640625,
|
1197 |
+
"learning_rate": 0.00011183150933837632,
|
1198 |
+
"loss": 0.9284,
|
1199 |
+
"step": 805
|
1200 |
+
},
|
1201 |
+
{
|
1202 |
+
"epoch": 7.751196172248804,
|
1203 |
+
"grad_norm": 0.353515625,
|
1204 |
+
"learning_rate": 0.00011071984957874479,
|
1205 |
+
"loss": 0.9222,
|
1206 |
+
"step": 810
|
1207 |
+
},
|
1208 |
+
{
|
1209 |
+
"epoch": 7.7990430622009566,
|
1210 |
+
"grad_norm": 0.42578125,
|
1211 |
+
"learning_rate": 0.00010960684801320536,
|
1212 |
+
"loss": 0.9335,
|
1213 |
+
"step": 815
|
1214 |
+
},
|
1215 |
+
{
|
1216 |
+
"epoch": 7.84688995215311,
|
1217 |
+
"grad_norm": 0.6171875,
|
1218 |
+
"learning_rate": 0.00010849264395640649,
|
1219 |
+
"loss": 0.9382,
|
1220 |
+
"step": 820
|
1221 |
+
},
|
1222 |
+
{
|
1223 |
+
"epoch": 7.894736842105263,
|
1224 |
+
"grad_norm": 0.66015625,
|
1225 |
+
"learning_rate": 0.00010737737687351284,
|
1226 |
+
"loss": 0.9414,
|
1227 |
+
"step": 825
|
1228 |
+
},
|
1229 |
+
{
|
1230 |
+
"epoch": 7.942583732057416,
|
1231 |
+
"grad_norm": 0.5234375,
|
1232 |
+
"learning_rate": 0.0001062611863627482,
|
1233 |
+
"loss": 0.9374,
|
1234 |
+
"step": 830
|
1235 |
+
},
|
1236 |
+
{
|
1237 |
+
"epoch": 7.990430622009569,
|
1238 |
+
"grad_norm": 0.33984375,
|
1239 |
+
"learning_rate": 0.00010514421213792205,
|
1240 |
+
"loss": 0.93,
|
1241 |
+
"step": 835
|
1242 |
+
},
|
1243 |
+
{
|
1244 |
+
"epoch": 8.0,
|
1245 |
+
"eval_loss": 2.506725311279297,
|
1246 |
+
"eval_runtime": 0.5363,
|
1247 |
+
"eval_samples_per_second": 18.647,
|
1248 |
+
"eval_steps_per_second": 1.865,
|
1249 |
+
"step": 836
|
1250 |
+
},
|
1251 |
+
{
|
1252 |
+
"epoch": 8.038277511961722,
|
1253 |
+
"grad_norm": 0.390625,
|
1254 |
+
"learning_rate": 0.00010402659401094152,
|
1255 |
+
"loss": 0.9129,
|
1256 |
+
"step": 840
|
1257 |
+
},
|
1258 |
+
{
|
1259 |
+
"epoch": 8.086124401913876,
|
1260 |
+
"grad_norm": 0.35546875,
|
1261 |
+
"learning_rate": 0.00010290847187431113,
|
1262 |
+
"loss": 0.9107,
|
1263 |
+
"step": 845
|
1264 |
+
},
|
1265 |
+
{
|
1266 |
+
"epoch": 8.133971291866029,
|
1267 |
+
"grad_norm": 0.36328125,
|
1268 |
+
"learning_rate": 0.00010178998568362243,
|
1269 |
+
"loss": 0.9226,
|
1270 |
+
"step": 850
|
1271 |
+
},
|
1272 |
+
{
|
1273 |
+
"epoch": 8.181818181818182,
|
1274 |
+
"grad_norm": 0.345703125,
|
1275 |
+
"learning_rate": 0.00010067127544003563,
|
1276 |
+
"loss": 0.9184,
|
1277 |
+
"step": 855
|
1278 |
+
},
|
1279 |
+
{
|
1280 |
+
"epoch": 8.229665071770334,
|
1281 |
+
"grad_norm": 0.35546875,
|
1282 |
+
"learning_rate": 9.955248117275566e-05,
|
1283 |
+
"loss": 0.915,
|
1284 |
+
"step": 860
|
1285 |
+
},
|
1286 |
+
{
|
1287 |
+
"epoch": 8.277511961722489,
|
1288 |
+
"grad_norm": 0.396484375,
|
1289 |
+
"learning_rate": 9.843374292150488e-05,
|
1290 |
+
"loss": 0.9234,
|
1291 |
+
"step": 865
|
1292 |
+
},
|
1293 |
+
{
|
1294 |
+
"epoch": 8.325358851674642,
|
1295 |
+
"grad_norm": 0.3671875,
|
1296 |
+
"learning_rate": 9.73152007189939e-05,
|
1297 |
+
"loss": 0.9169,
|
1298 |
+
"step": 870
|
1299 |
+
},
|
1300 |
+
{
|
1301 |
+
"epoch": 8.373205741626794,
|
1302 |
+
"grad_norm": 0.373046875,
|
1303 |
+
"learning_rate": 9.619699457339405e-05,
|
1304 |
+
"loss": 0.9131,
|
1305 |
+
"step": 875
|
1306 |
+
},
|
1307 |
+
{
|
1308 |
+
"epoch": 8.421052631578947,
|
1309 |
+
"grad_norm": 0.3671875,
|
1310 |
+
"learning_rate": 9.507926445081219e-05,
|
1311 |
+
"loss": 0.9189,
|
1312 |
+
"step": 880
|
1313 |
+
},
|
1314 |
+
{
|
1315 |
+
"epoch": 8.4688995215311,
|
1316 |
+
"grad_norm": 0.359375,
|
1317 |
+
"learning_rate": 9.396215025777139e-05,
|
1318 |
+
"loss": 0.9125,
|
1319 |
+
"step": 885
|
1320 |
+
},
|
1321 |
+
{
|
1322 |
+
"epoch": 8.516746411483254,
|
1323 |
+
"grad_norm": 0.341796875,
|
1324 |
+
"learning_rate": 9.284579182369867e-05,
|
1325 |
+
"loss": 0.9089,
|
1326 |
+
"step": 890
|
1327 |
+
},
|
1328 |
+
{
|
1329 |
+
"epoch": 8.564593301435407,
|
1330 |
+
"grad_norm": 0.4453125,
|
1331 |
+
"learning_rate": 9.173032888342244e-05,
|
1332 |
+
"loss": 0.9221,
|
1333 |
+
"step": 895
|
1334 |
+
},
|
1335 |
+
{
|
1336 |
+
"epoch": 8.61244019138756,
|
1337 |
+
"grad_norm": 0.5546875,
|
1338 |
+
"learning_rate": 9.061590105968208e-05,
|
1339 |
+
"loss": 0.9155,
|
1340 |
+
"step": 900
|
1341 |
+
},
|
1342 |
+
{
|
1343 |
+
"epoch": 8.660287081339712,
|
1344 |
+
"grad_norm": 0.55859375,
|
1345 |
+
"learning_rate": 8.950264784565112e-05,
|
1346 |
+
"loss": 0.9286,
|
1347 |
+
"step": 905
|
1348 |
+
},
|
1349 |
+
{
|
1350 |
+
"epoch": 8.708133971291867,
|
1351 |
+
"grad_norm": 0.36328125,
|
1352 |
+
"learning_rate": 8.839070858747697e-05,
|
1353 |
+
"loss": 0.9199,
|
1354 |
+
"step": 910
|
1355 |
+
},
|
1356 |
+
{
|
1357 |
+
"epoch": 8.75598086124402,
|
1358 |
+
"grad_norm": 0.365234375,
|
1359 |
+
"learning_rate": 8.728022246683894e-05,
|
1360 |
+
"loss": 0.908,
|
1361 |
+
"step": 915
|
1362 |
+
},
|
1363 |
+
{
|
1364 |
+
"epoch": 8.803827751196172,
|
1365 |
+
"grad_norm": 0.345703125,
|
1366 |
+
"learning_rate": 8.617132848352671e-05,
|
1367 |
+
"loss": 0.9169,
|
1368 |
+
"step": 920
|
1369 |
+
},
|
1370 |
+
{
|
1371 |
+
"epoch": 8.851674641148325,
|
1372 |
+
"grad_norm": 0.390625,
|
1373 |
+
"learning_rate": 8.506416543804182e-05,
|
1374 |
+
"loss": 0.9238,
|
1375 |
+
"step": 925
|
1376 |
+
},
|
1377 |
+
{
|
1378 |
+
"epoch": 8.89952153110048,
|
1379 |
+
"grad_norm": 0.5703125,
|
1380 |
+
"learning_rate": 8.395887191422397e-05,
|
1381 |
+
"loss": 0.9209,
|
1382 |
+
"step": 930
|
1383 |
+
},
|
1384 |
+
{
|
1385 |
+
"epoch": 8.947368421052632,
|
1386 |
+
"grad_norm": 0.46875,
|
1387 |
+
"learning_rate": 8.285558626190447e-05,
|
1388 |
+
"loss": 0.9189,
|
1389 |
+
"step": 935
|
1390 |
+
},
|
1391 |
+
{
|
1392 |
+
"epoch": 8.995215311004785,
|
1393 |
+
"grad_norm": 0.337890625,
|
1394 |
+
"learning_rate": 8.175444657958876e-05,
|
1395 |
+
"loss": 0.9195,
|
1396 |
+
"step": 940
|
1397 |
+
},
|
1398 |
+
{
|
1399 |
+
"epoch": 8.995215311004785,
|
1400 |
+
"eval_loss": 2.5168216228485107,
|
1401 |
+
"eval_runtime": 0.6511,
|
1402 |
+
"eval_samples_per_second": 15.36,
|
1403 |
+
"eval_steps_per_second": 1.536,
|
1404 |
+
"step": 940
|
1405 |
+
},
|
1406 |
+
{
|
1407 |
+
"epoch": 9.043062200956937,
|
1408 |
+
"grad_norm": 0.40625,
|
1409 |
+
"learning_rate": 8.065559069717088e-05,
|
1410 |
+
"loss": 0.9021,
|
1411 |
+
"step": 945
|
1412 |
+
},
|
1413 |
+
{
|
1414 |
+
"epoch": 9.090909090909092,
|
1415 |
+
"grad_norm": 0.41015625,
|
1416 |
+
"learning_rate": 7.955915615868111e-05,
|
1417 |
+
"loss": 0.9103,
|
1418 |
+
"step": 950
|
1419 |
+
},
|
1420 |
+
{
|
1421 |
+
"epoch": 9.138755980861244,
|
1422 |
+
"grad_norm": 0.392578125,
|
1423 |
+
"learning_rate": 7.846528020506957e-05,
|
1424 |
+
"loss": 0.8983,
|
1425 |
+
"step": 955
|
1426 |
+
},
|
1427 |
+
{
|
1428 |
+
"epoch": 9.186602870813397,
|
1429 |
+
"grad_norm": 0.36328125,
|
1430 |
+
"learning_rate": 7.73740997570278e-05,
|
1431 |
+
"loss": 0.9037,
|
1432 |
+
"step": 960
|
1433 |
+
},
|
1434 |
+
{
|
1435 |
+
"epoch": 9.23444976076555,
|
1436 |
+
"grad_norm": 0.353515625,
|
1437 |
+
"learning_rate": 7.628575139785024e-05,
|
1438 |
+
"loss": 0.9036,
|
1439 |
+
"step": 965
|
1440 |
+
},
|
1441 |
+
{
|
1442 |
+
"epoch": 9.282296650717702,
|
1443 |
+
"grad_norm": 0.388671875,
|
1444 |
+
"learning_rate": 7.520037135633816e-05,
|
1445 |
+
"loss": 0.9051,
|
1446 |
+
"step": 970
|
1447 |
+
},
|
1448 |
+
{
|
1449 |
+
"epoch": 9.330143540669857,
|
1450 |
+
"grad_norm": 0.439453125,
|
1451 |
+
"learning_rate": 7.411809548974792e-05,
|
1452 |
+
"loss": 0.9186,
|
1453 |
+
"step": 975
|
1454 |
+
},
|
1455 |
+
{
|
1456 |
+
"epoch": 9.37799043062201,
|
1457 |
+
"grad_norm": 0.365234375,
|
1458 |
+
"learning_rate": 7.303905926678564e-05,
|
1459 |
+
"loss": 0.9095,
|
1460 |
+
"step": 980
|
1461 |
+
},
|
1462 |
+
{
|
1463 |
+
"epoch": 9.425837320574162,
|
1464 |
+
"grad_norm": 0.375,
|
1465 |
+
"learning_rate": 7.196339775065042e-05,
|
1466 |
+
"loss": 0.9069,
|
1467 |
+
"step": 985
|
1468 |
+
},
|
1469 |
+
{
|
1470 |
+
"epoch": 9.473684210526315,
|
1471 |
+
"grad_norm": 0.376953125,
|
1472 |
+
"learning_rate": 7.089124558212871e-05,
|
1473 |
+
"loss": 0.908,
|
1474 |
+
"step": 990
|
1475 |
+
},
|
1476 |
+
{
|
1477 |
+
"epoch": 9.52153110047847,
|
1478 |
+
"grad_norm": 0.38671875,
|
1479 |
+
"learning_rate": 6.982273696274106e-05,
|
1480 |
+
"loss": 0.9121,
|
1481 |
+
"step": 995
|
1482 |
+
},
|
1483 |
+
{
|
1484 |
+
"epoch": 9.569377990430622,
|
1485 |
+
"grad_norm": 0.41796875,
|
1486 |
+
"learning_rate": 6.875800563794425e-05,
|
1487 |
+
"loss": 0.9097,
|
1488 |
+
"step": 1000
|
1489 |
+
},
|
1490 |
+
{
|
1491 |
+
"epoch": 9.617224880382775,
|
1492 |
+
"grad_norm": 0.361328125,
|
1493 |
+
"learning_rate": 6.769718488039023e-05,
|
1494 |
+
"loss": 0.9101,
|
1495 |
+
"step": 1005
|
1496 |
+
},
|
1497 |
+
{
|
1498 |
+
"epoch": 9.665071770334928,
|
1499 |
+
"grad_norm": 0.3515625,
|
1500 |
+
"learning_rate": 6.664040747324437e-05,
|
1501 |
+
"loss": 0.9117,
|
1502 |
+
"step": 1010
|
1503 |
+
},
|
1504 |
+
{
|
1505 |
+
"epoch": 9.712918660287082,
|
1506 |
+
"grad_norm": 0.375,
|
1507 |
+
"learning_rate": 6.558780569356507e-05,
|
1508 |
+
"loss": 0.9106,
|
1509 |
+
"step": 1015
|
1510 |
+
},
|
1511 |
+
{
|
1512 |
+
"epoch": 9.760765550239235,
|
1513 |
+
"grad_norm": 0.376953125,
|
1514 |
+
"learning_rate": 6.453951129574644e-05,
|
1515 |
+
"loss": 0.9037,
|
1516 |
+
"step": 1020
|
1517 |
+
},
|
1518 |
+
{
|
1519 |
+
"epoch": 9.808612440191387,
|
1520 |
+
"grad_norm": 0.40234375,
|
1521 |
+
"learning_rate": 6.349565549502676e-05,
|
1522 |
+
"loss": 0.9033,
|
1523 |
+
"step": 1025
|
1524 |
+
},
|
1525 |
+
{
|
1526 |
+
"epoch": 9.85645933014354,
|
1527 |
+
"grad_norm": 0.3515625,
|
1528 |
+
"learning_rate": 6.245636895106402e-05,
|
1529 |
+
"loss": 0.9156,
|
1530 |
+
"step": 1030
|
1531 |
+
},
|
1532 |
+
{
|
1533 |
+
"epoch": 9.904306220095695,
|
1534 |
+
"grad_norm": 0.357421875,
|
1535 |
+
"learning_rate": 6.142178175158149e-05,
|
1536 |
+
"loss": 0.9082,
|
1537 |
+
"step": 1035
|
1538 |
+
},
|
1539 |
+
{
|
1540 |
+
"epoch": 9.952153110047847,
|
1541 |
+
"grad_norm": 0.451171875,
|
1542 |
+
"learning_rate": 6.039202339608432e-05,
|
1543 |
+
"loss": 0.9147,
|
1544 |
+
"step": 1040
|
1545 |
+
},
|
1546 |
+
{
|
1547 |
+
"epoch": 10.0,
|
1548 |
+
"grad_norm": 0.50390625,
|
1549 |
+
"learning_rate": 5.9367222779650334e-05,
|
1550 |
+
"loss": 0.912,
|
1551 |
+
"step": 1045
|
1552 |
+
},
|
1553 |
+
{
|
1554 |
+
"epoch": 10.0,
|
1555 |
+
"eval_loss": 2.5271153450012207,
|
1556 |
+
"eval_runtime": 0.5317,
|
1557 |
+
"eval_samples_per_second": 18.809,
|
1558 |
+
"eval_steps_per_second": 1.881,
|
1559 |
+
"step": 1045
|
1560 |
+
},
|
1561 |
+
{
|
1562 |
+
"epoch": 10.047846889952153,
|
1563 |
+
"grad_norm": 0.439453125,
|
1564 |
+
"learning_rate": 5.834750817679606e-05,
|
1565 |
+
"loss": 0.9084,
|
1566 |
+
"step": 1050
|
1567 |
+
},
|
1568 |
+
{
|
1569 |
+
"epoch": 10.095693779904305,
|
1570 |
+
"grad_norm": 0.3515625,
|
1571 |
+
"learning_rate": 5.733300722542045e-05,
|
1572 |
+
"loss": 0.897,
|
1573 |
+
"step": 1055
|
1574 |
+
},
|
1575 |
+
{
|
1576 |
+
"epoch": 10.14354066985646,
|
1577 |
+
"grad_norm": 0.376953125,
|
1578 |
+
"learning_rate": 5.6323846910828735e-05,
|
1579 |
+
"loss": 0.8998,
|
1580 |
+
"step": 1060
|
1581 |
+
},
|
1582 |
+
{
|
1583 |
+
"epoch": 10.191387559808613,
|
1584 |
+
"grad_norm": 0.349609375,
|
1585 |
+
"learning_rate": 5.5320153549837415e-05,
|
1586 |
+
"loss": 0.9026,
|
1587 |
+
"step": 1065
|
1588 |
+
},
|
1589 |
+
{
|
1590 |
+
"epoch": 10.239234449760765,
|
1591 |
+
"grad_norm": 0.3515625,
|
1592 |
+
"learning_rate": 5.432205277496327e-05,
|
1593 |
+
"loss": 0.8996,
|
1594 |
+
"step": 1070
|
1595 |
+
},
|
1596 |
+
{
|
1597 |
+
"epoch": 10.287081339712918,
|
1598 |
+
"grad_norm": 0.359375,
|
1599 |
+
"learning_rate": 5.33296695186977e-05,
|
1600 |
+
"loss": 0.9,
|
1601 |
+
"step": 1075
|
1602 |
+
},
|
1603 |
+
{
|
1604 |
+
"epoch": 10.334928229665072,
|
1605 |
+
"grad_norm": 0.41015625,
|
1606 |
+
"learning_rate": 5.234312799786921e-05,
|
1607 |
+
"loss": 0.8988,
|
1608 |
+
"step": 1080
|
1609 |
+
},
|
1610 |
+
{
|
1611 |
+
"epoch": 10.382775119617225,
|
1612 |
+
"grad_norm": 0.41015625,
|
1613 |
+
"learning_rate": 5.1362551698094964e-05,
|
1614 |
+
"loss": 0.8969,
|
1615 |
+
"step": 1085
|
1616 |
+
},
|
1617 |
+
{
|
1618 |
+
"epoch": 10.430622009569378,
|
1619 |
+
"grad_norm": 0.35546875,
|
1620 |
+
"learning_rate": 5.0388063358324134e-05,
|
1621 |
+
"loss": 0.8954,
|
1622 |
+
"step": 1090
|
1623 |
+
},
|
1624 |
+
{
|
1625 |
+
"epoch": 10.47846889952153,
|
1626 |
+
"grad_norm": 0.3515625,
|
1627 |
+
"learning_rate": 4.9419784955474524e-05,
|
1628 |
+
"loss": 0.8991,
|
1629 |
+
"step": 1095
|
1630 |
+
},
|
1631 |
+
{
|
1632 |
+
"epoch": 10.526315789473685,
|
1633 |
+
"grad_norm": 0.359375,
|
1634 |
+
"learning_rate": 4.845783768916482e-05,
|
1635 |
+
"loss": 0.9074,
|
1636 |
+
"step": 1100
|
1637 |
+
},
|
1638 |
+
{
|
1639 |
+
"epoch": 10.574162679425838,
|
1640 |
+
"grad_norm": 0.353515625,
|
1641 |
+
"learning_rate": 4.7502341966544e-05,
|
1642 |
+
"loss": 0.9012,
|
1643 |
+
"step": 1105
|
1644 |
+
},
|
1645 |
+
{
|
1646 |
+
"epoch": 10.62200956937799,
|
1647 |
+
"grad_norm": 0.34765625,
|
1648 |
+
"learning_rate": 4.6553417387219886e-05,
|
1649 |
+
"loss": 0.8993,
|
1650 |
+
"step": 1110
|
1651 |
+
},
|
1652 |
+
{
|
1653 |
+
"epoch": 10.669856459330143,
|
1654 |
+
"grad_norm": 0.333984375,
|
1655 |
+
"learning_rate": 4.5611182728288895e-05,
|
1656 |
+
"loss": 0.9073,
|
1657 |
+
"step": 1115
|
1658 |
+
},
|
1659 |
+
{
|
1660 |
+
"epoch": 10.717703349282296,
|
1661 |
+
"grad_norm": 0.341796875,
|
1662 |
+
"learning_rate": 4.467575592946864e-05,
|
1663 |
+
"loss": 0.9073,
|
1664 |
+
"step": 1120
|
1665 |
+
},
|
1666 |
+
{
|
1667 |
+
"epoch": 10.76555023923445,
|
1668 |
+
"grad_norm": 0.349609375,
|
1669 |
+
"learning_rate": 4.374725407833532e-05,
|
1670 |
+
"loss": 0.902,
|
1671 |
+
"step": 1125
|
1672 |
+
},
|
1673 |
+
{
|
1674 |
+
"epoch": 10.813397129186603,
|
1675 |
+
"grad_norm": 0.380859375,
|
1676 |
+
"learning_rate": 4.282579339566802e-05,
|
1677 |
+
"loss": 0.8993,
|
1678 |
+
"step": 1130
|
1679 |
+
},
|
1680 |
+
{
|
1681 |
+
"epoch": 10.861244019138756,
|
1682 |
+
"grad_norm": 0.37109375,
|
1683 |
+
"learning_rate": 4.1911489220901236e-05,
|
1684 |
+
"loss": 0.9002,
|
1685 |
+
"step": 1135
|
1686 |
+
},
|
1687 |
+
{
|
1688 |
+
"epoch": 10.909090909090908,
|
1689 |
+
"grad_norm": 0.3515625,
|
1690 |
+
"learning_rate": 4.100445599768774e-05,
|
1691 |
+
"loss": 0.9115,
|
1692 |
+
"step": 1140
|
1693 |
+
},
|
1694 |
+
{
|
1695 |
+
"epoch": 10.956937799043063,
|
1696 |
+
"grad_norm": 0.34375,
|
1697 |
+
"learning_rate": 4.0104807259573716e-05,
|
1698 |
+
"loss": 0.9003,
|
1699 |
+
"step": 1145
|
1700 |
+
},
|
1701 |
+
{
|
1702 |
+
"epoch": 10.995215311004785,
|
1703 |
+
"eval_loss": 2.535585403442383,
|
1704 |
+
"eval_runtime": 0.6791,
|
1705 |
+
"eval_samples_per_second": 14.726,
|
1706 |
+
"eval_steps_per_second": 1.473,
|
1707 |
+
"step": 1149
|
1708 |
+
},
|
1709 |
+
{
|
1710 |
+
"epoch": 11.004784688995215,
|
1711 |
+
"grad_norm": 0.345703125,
|
1712 |
+
"learning_rate": 3.9212655615787804e-05,
|
1713 |
+
"loss": 0.9032,
|
1714 |
+
"step": 1150
|
1715 |
+
},
|
1716 |
+
{
|
1717 |
+
"epoch": 11.052631578947368,
|
1718 |
+
"grad_norm": 0.34765625,
|
1719 |
+
"learning_rate": 3.832811273714569e-05,
|
1720 |
+
"loss": 0.902,
|
1721 |
+
"step": 1155
|
1722 |
+
},
|
1723 |
+
{
|
1724 |
+
"epoch": 11.10047846889952,
|
1725 |
+
"grad_norm": 0.36328125,
|
1726 |
+
"learning_rate": 3.745128934207225e-05,
|
1727 |
+
"loss": 0.8894,
|
1728 |
+
"step": 1160
|
1729 |
+
},
|
1730 |
+
{
|
1731 |
+
"epoch": 11.148325358851675,
|
1732 |
+
"grad_norm": 0.388671875,
|
1733 |
+
"learning_rate": 3.6582295182742964e-05,
|
1734 |
+
"loss": 0.8935,
|
1735 |
+
"step": 1165
|
1736 |
+
},
|
1737 |
+
{
|
1738 |
+
"epoch": 11.196172248803828,
|
1739 |
+
"grad_norm": 0.365234375,
|
1740 |
+
"learning_rate": 3.5721239031346066e-05,
|
1741 |
+
"loss": 0.8997,
|
1742 |
+
"step": 1170
|
1743 |
+
},
|
1744 |
+
{
|
1745 |
+
"epoch": 11.24401913875598,
|
1746 |
+
"grad_norm": 0.3359375,
|
1747 |
+
"learning_rate": 3.4868228666467704e-05,
|
1748 |
+
"loss": 0.8972,
|
1749 |
+
"step": 1175
|
1750 |
+
},
|
1751 |
+
{
|
1752 |
+
"epoch": 11.291866028708133,
|
1753 |
+
"grad_norm": 0.349609375,
|
1754 |
+
"learning_rate": 3.402337085960119e-05,
|
1755 |
+
"loss": 0.9028,
|
1756 |
+
"step": 1180
|
1757 |
+
},
|
1758 |
+
{
|
1759 |
+
"epoch": 11.339712918660288,
|
1760 |
+
"grad_norm": 0.341796875,
|
1761 |
+
"learning_rate": 3.318677136178228e-05,
|
1762 |
+
"loss": 0.8943,
|
1763 |
+
"step": 1185
|
1764 |
+
},
|
1765 |
+
{
|
1766 |
+
"epoch": 11.38755980861244,
|
1767 |
+
"grad_norm": 0.345703125,
|
1768 |
+
"learning_rate": 3.235853489035241e-05,
|
1769 |
+
"loss": 0.8851,
|
1770 |
+
"step": 1190
|
1771 |
+
},
|
1772 |
+
{
|
1773 |
+
"epoch": 11.435406698564593,
|
1774 |
+
"grad_norm": 0.384765625,
|
1775 |
+
"learning_rate": 3.153876511585122e-05,
|
1776 |
+
"loss": 0.8917,
|
1777 |
+
"step": 1195
|
1778 |
+
},
|
1779 |
+
{
|
1780 |
+
"epoch": 11.483253588516746,
|
1781 |
+
"grad_norm": 0.392578125,
|
1782 |
+
"learning_rate": 3.072756464904006e-05,
|
1783 |
+
"loss": 0.8974,
|
1784 |
+
"step": 1200
|
1785 |
+
},
|
1786 |
+
{
|
1787 |
+
"epoch": 11.5311004784689,
|
1788 |
+
"grad_norm": 0.337890625,
|
1789 |
+
"learning_rate": 2.9925035028058134e-05,
|
1790 |
+
"loss": 0.8951,
|
1791 |
+
"step": 1205
|
1792 |
+
},
|
1793 |
+
{
|
1794 |
+
"epoch": 11.578947368421053,
|
1795 |
+
"grad_norm": 0.330078125,
|
1796 |
+
"learning_rate": 2.9131276705713006e-05,
|
1797 |
+
"loss": 0.893,
|
1798 |
+
"step": 1210
|
1799 |
+
},
|
1800 |
+
{
|
1801 |
+
"epoch": 11.626794258373206,
|
1802 |
+
"grad_norm": 0.345703125,
|
1803 |
+
"learning_rate": 2.8346389036906828e-05,
|
1804 |
+
"loss": 0.899,
|
1805 |
+
"step": 1215
|
1806 |
+
},
|
1807 |
+
{
|
1808 |
+
"epoch": 11.674641148325358,
|
1809 |
+
"grad_norm": 0.337890625,
|
1810 |
+
"learning_rate": 2.7570470266200176e-05,
|
1811 |
+
"loss": 0.9037,
|
1812 |
+
"step": 1220
|
1813 |
+
},
|
1814 |
+
{
|
1815 |
+
"epoch": 11.722488038277511,
|
1816 |
+
"grad_norm": 0.33984375,
|
1817 |
+
"learning_rate": 2.68036175155147e-05,
|
1818 |
+
"loss": 0.8917,
|
1819 |
+
"step": 1225
|
1820 |
+
},
|
1821 |
+
{
|
1822 |
+
"epoch": 11.770334928229666,
|
1823 |
+
"grad_norm": 0.37890625,
|
1824 |
+
"learning_rate": 2.6045926771976303e-05,
|
1825 |
+
"loss": 0.9061,
|
1826 |
+
"step": 1230
|
1827 |
+
},
|
1828 |
+
{
|
1829 |
+
"epoch": 11.818181818181818,
|
1830 |
+
"grad_norm": 0.3671875,
|
1831 |
+
"learning_rate": 2.529749287590042e-05,
|
1832 |
+
"loss": 0.9028,
|
1833 |
+
"step": 1235
|
1834 |
+
},
|
1835 |
+
{
|
1836 |
+
"epoch": 11.866028708133971,
|
1837 |
+
"grad_norm": 0.337890625,
|
1838 |
+
"learning_rate": 2.4558409508920986e-05,
|
1839 |
+
"loss": 0.8946,
|
1840 |
+
"step": 1240
|
1841 |
+
},
|
1842 |
+
{
|
1843 |
+
"epoch": 11.913875598086124,
|
1844 |
+
"grad_norm": 0.349609375,
|
1845 |
+
"learning_rate": 2.382876918226409e-05,
|
1846 |
+
"loss": 0.8938,
|
1847 |
+
"step": 1245
|
1848 |
+
},
|
1849 |
+
{
|
1850 |
+
"epoch": 11.961722488038278,
|
1851 |
+
"grad_norm": 0.36328125,
|
1852 |
+
"learning_rate": 2.3108663225168435e-05,
|
1853 |
+
"loss": 0.9032,
|
1854 |
+
"step": 1250
|
1855 |
+
},
|
1856 |
+
{
|
1857 |
+
"epoch": 12.0,
|
1858 |
+
"eval_loss": 2.5401320457458496,
|
1859 |
+
"eval_runtime": 0.536,
|
1860 |
+
"eval_samples_per_second": 18.658,
|
1861 |
+
"eval_steps_per_second": 1.866,
|
1862 |
+
"step": 1254
|
1863 |
+
},
|
1864 |
+
{
|
1865 |
+
"epoch": 12.009569377990431,
|
1866 |
+
"grad_norm": 0.33984375,
|
1867 |
+
"learning_rate": 2.239818177345364e-05,
|
1868 |
+
"loss": 0.9023,
|
1869 |
+
"step": 1255
|
1870 |
+
},
|
1871 |
+
{
|
1872 |
+
"epoch": 12.057416267942584,
|
1873 |
+
"grad_norm": 0.337890625,
|
1874 |
+
"learning_rate": 2.1697413758237784e-05,
|
1875 |
+
"loss": 0.8902,
|
1876 |
+
"step": 1260
|
1877 |
+
},
|
1878 |
+
{
|
1879 |
+
"epoch": 12.105263157894736,
|
1880 |
+
"grad_norm": 0.359375,
|
1881 |
+
"learning_rate": 2.1006446894806065e-05,
|
1882 |
+
"loss": 0.8958,
|
1883 |
+
"step": 1265
|
1884 |
+
},
|
1885 |
+
{
|
1886 |
+
"epoch": 12.15311004784689,
|
1887 |
+
"grad_norm": 0.337890625,
|
1888 |
+
"learning_rate": 2.032536767163141e-05,
|
1889 |
+
"loss": 0.8949,
|
1890 |
+
"step": 1270
|
1891 |
+
},
|
1892 |
+
{
|
1893 |
+
"epoch": 12.200956937799043,
|
1894 |
+
"grad_norm": 0.3671875,
|
1895 |
+
"learning_rate": 1.965426133954854e-05,
|
1896 |
+
"loss": 0.9023,
|
1897 |
+
"step": 1275
|
1898 |
+
},
|
1899 |
+
{
|
1900 |
+
"epoch": 12.248803827751196,
|
1901 |
+
"grad_norm": 0.333984375,
|
1902 |
+
"learning_rate": 1.8993211901083353e-05,
|
1903 |
+
"loss": 0.895,
|
1904 |
+
"step": 1280
|
1905 |
+
},
|
1906 |
+
{
|
1907 |
+
"epoch": 12.296650717703349,
|
1908 |
+
"grad_norm": 0.328125,
|
1909 |
+
"learning_rate": 1.8342302099938057e-05,
|
1910 |
+
"loss": 0.8925,
|
1911 |
+
"step": 1285
|
1912 |
+
},
|
1913 |
+
{
|
1914 |
+
"epoch": 12.344497607655502,
|
1915 |
+
"grad_norm": 0.328125,
|
1916 |
+
"learning_rate": 1.7701613410634365e-05,
|
1917 |
+
"loss": 0.8986,
|
1918 |
+
"step": 1290
|
1919 |
+
},
|
1920 |
+
{
|
1921 |
+
"epoch": 12.392344497607656,
|
1922 |
+
"grad_norm": 0.33203125,
|
1923 |
+
"learning_rate": 1.7071226028315113e-05,
|
1924 |
+
"loss": 0.8922,
|
1925 |
+
"step": 1295
|
1926 |
+
},
|
1927 |
+
{
|
1928 |
+
"epoch": 12.440191387559809,
|
1929 |
+
"grad_norm": 0.35546875,
|
1930 |
+
"learning_rate": 1.6451218858706374e-05,
|
1931 |
+
"loss": 0.8878,
|
1932 |
+
"step": 1300
|
1933 |
+
},
|
1934 |
+
{
|
1935 |
+
"epoch": 12.488038277511961,
|
1936 |
+
"grad_norm": 0.337890625,
|
1937 |
+
"learning_rate": 1.584166950824061e-05,
|
1938 |
+
"loss": 0.8992,
|
1939 |
+
"step": 1305
|
1940 |
+
},
|
1941 |
+
{
|
1942 |
+
"epoch": 12.535885167464114,
|
1943 |
+
"grad_norm": 0.361328125,
|
1944 |
+
"learning_rate": 1.5242654274342894e-05,
|
1945 |
+
"loss": 0.8879,
|
1946 |
+
"step": 1310
|
1947 |
+
},
|
1948 |
+
{
|
1949 |
+
"epoch": 12.583732057416269,
|
1950 |
+
"grad_norm": 0.353515625,
|
1951 |
+
"learning_rate": 1.4654248135880621e-05,
|
1952 |
+
"loss": 0.8942,
|
1953 |
+
"step": 1315
|
1954 |
+
},
|
1955 |
+
{
|
1956 |
+
"epoch": 12.631578947368421,
|
1957 |
+
"grad_norm": 0.353515625,
|
1958 |
+
"learning_rate": 1.4076524743778319e-05,
|
1959 |
+
"loss": 0.8957,
|
1960 |
+
"step": 1320
|
1961 |
+
},
|
1962 |
+
{
|
1963 |
+
"epoch": 12.679425837320574,
|
1964 |
+
"grad_norm": 0.33203125,
|
1965 |
+
"learning_rate": 1.350955641179893e-05,
|
1966 |
+
"loss": 0.8981,
|
1967 |
+
"step": 1325
|
1968 |
+
},
|
1969 |
+
{
|
1970 |
+
"epoch": 12.727272727272727,
|
1971 |
+
"grad_norm": 0.353515625,
|
1972 |
+
"learning_rate": 1.295341410749208e-05,
|
1973 |
+
"loss": 0.8952,
|
1974 |
+
"step": 1330
|
1975 |
+
},
|
1976 |
+
{
|
1977 |
+
"epoch": 12.775119617224881,
|
1978 |
+
"grad_norm": 0.34375,
|
1979 |
+
"learning_rate": 1.2408167443311214e-05,
|
1980 |
+
"loss": 0.8945,
|
1981 |
+
"step": 1335
|
1982 |
+
},
|
1983 |
+
{
|
1984 |
+
"epoch": 12.822966507177034,
|
1985 |
+
"grad_norm": 0.359375,
|
1986 |
+
"learning_rate": 1.1873884667900125e-05,
|
1987 |
+
"loss": 0.8851,
|
1988 |
+
"step": 1340
|
1989 |
+
},
|
1990 |
+
{
|
1991 |
+
"epoch": 12.870813397129186,
|
1992 |
+
"grad_norm": 0.34375,
|
1993 |
+
"learning_rate": 1.1350632657550253e-05,
|
1994 |
+
"loss": 0.8922,
|
1995 |
+
"step": 1345
|
1996 |
+
},
|
1997 |
+
{
|
1998 |
+
"epoch": 12.91866028708134,
|
1999 |
+
"grad_norm": 0.3515625,
|
2000 |
+
"learning_rate": 1.083847690782972e-05,
|
2001 |
+
"loss": 0.8926,
|
2002 |
+
"step": 1350
|
2003 |
+
},
|
2004 |
+
{
|
2005 |
+
"epoch": 12.966507177033494,
|
2006 |
+
"grad_norm": 0.349609375,
|
2007 |
+
"learning_rate": 1.0337481525385362e-05,
|
2008 |
+
"loss": 0.9006,
|
2009 |
+
"step": 1355
|
2010 |
+
},
|
2011 |
+
{
|
2012 |
+
"epoch": 12.995215311004785,
|
2013 |
+
"eval_loss": 2.5425758361816406,
|
2014 |
+
"eval_runtime": 0.6353,
|
2015 |
+
"eval_samples_per_second": 15.741,
|
2016 |
+
"eval_steps_per_second": 1.574,
|
2017 |
+
"step": 1358
|
2018 |
+
},
|
2019 |
+
{
|
2020 |
+
"epoch": 13.014354066985646,
|
2021 |
+
"grad_norm": 0.345703125,
|
2022 |
+
"learning_rate": 9.8477092199184e-06,
|
2023 |
+
"loss": 0.9002,
|
2024 |
+
"step": 1360
|
2025 |
+
},
|
2026 |
+
{
|
2027 |
+
"epoch": 13.062200956937799,
|
2028 |
+
"grad_norm": 0.34375,
|
2029 |
+
"learning_rate": 9.369221296335006e-06,
|
2030 |
+
"loss": 0.9007,
|
2031 |
+
"step": 1365
|
2032 |
+
},
|
2033 |
+
{
|
2034 |
+
"epoch": 13.110047846889952,
|
2035 |
+
"grad_norm": 0.34765625,
|
2036 |
+
"learning_rate": 8.902077647072881e-06,
|
2037 |
+
"loss": 0.8882,
|
2038 |
+
"step": 1370
|
2039 |
+
},
|
2040 |
+
{
|
2041 |
+
"epoch": 13.157894736842104,
|
2042 |
+
"grad_norm": 0.34375,
|
2043 |
+
"learning_rate": 8.446336744604378e-06,
|
2044 |
+
"loss": 0.888,
|
2045 |
+
"step": 1375
|
2046 |
+
},
|
2047 |
+
{
|
2048 |
+
"epoch": 13.205741626794259,
|
2049 |
+
"grad_norm": 0.33203125,
|
2050 |
+
"learning_rate": 8.002055634117578e-06,
|
2051 |
+
"loss": 0.8869,
|
2052 |
+
"step": 1380
|
2053 |
+
},
|
2054 |
+
{
|
2055 |
+
"epoch": 13.253588516746412,
|
2056 |
+
"grad_norm": 0.34375,
|
2057 |
+
"learning_rate": 7.569289926375933e-06,
|
2058 |
+
"loss": 0.8931,
|
2059 |
+
"step": 1385
|
2060 |
+
},
|
2061 |
+
{
|
2062 |
+
"epoch": 13.301435406698564,
|
2063 |
+
"grad_norm": 0.326171875,
|
2064 |
+
"learning_rate": 7.148093790757371e-06,
|
2065 |
+
"loss": 0.8958,
|
2066 |
+
"step": 1390
|
2067 |
+
},
|
2068 |
+
{
|
2069 |
+
"epoch": 13.349282296650717,
|
2070 |
+
"grad_norm": 0.330078125,
|
2071 |
+
"learning_rate": 6.738519948473976e-06,
|
2072 |
+
"loss": 0.8914,
|
2073 |
+
"step": 1395
|
2074 |
+
},
|
2075 |
+
{
|
2076 |
+
"epoch": 13.397129186602871,
|
2077 |
+
"grad_norm": 0.349609375,
|
2078 |
+
"learning_rate": 6.3406196659728465e-06,
|
2079 |
+
"loss": 0.8975,
|
2080 |
+
"step": 1400
|
2081 |
+
},
|
2082 |
+
{
|
2083 |
+
"epoch": 13.444976076555024,
|
2084 |
+
"grad_norm": 0.330078125,
|
2085 |
+
"learning_rate": 5.954442748519073e-06,
|
2086 |
+
"loss": 0.8908,
|
2087 |
+
"step": 1405
|
2088 |
+
},
|
2089 |
+
{
|
2090 |
+
"epoch": 13.492822966507177,
|
2091 |
+
"grad_norm": 0.345703125,
|
2092 |
+
"learning_rate": 5.580037533961546e-06,
|
2093 |
+
"loss": 0.8938,
|
2094 |
+
"step": 1410
|
2095 |
+
},
|
2096 |
+
{
|
2097 |
+
"epoch": 13.54066985645933,
|
2098 |
+
"grad_norm": 0.349609375,
|
2099 |
+
"learning_rate": 5.217450886682584e-06,
|
2100 |
+
"loss": 0.8958,
|
2101 |
+
"step": 1415
|
2102 |
+
},
|
2103 |
+
{
|
2104 |
+
"epoch": 13.588516746411484,
|
2105 |
+
"grad_norm": 0.3359375,
|
2106 |
+
"learning_rate": 4.866728191731829e-06,
|
2107 |
+
"loss": 0.8905,
|
2108 |
+
"step": 1420
|
2109 |
+
},
|
2110 |
+
{
|
2111 |
+
"epoch": 13.636363636363637,
|
2112 |
+
"grad_norm": 0.330078125,
|
2113 |
+
"learning_rate": 4.527913349145441e-06,
|
2114 |
+
"loss": 0.8995,
|
2115 |
+
"step": 1425
|
2116 |
+
},
|
2117 |
+
{
|
2118 |
+
"epoch": 13.68421052631579,
|
2119 |
+
"grad_norm": 0.3671875,
|
2120 |
+
"learning_rate": 4.20104876845111e-06,
|
2121 |
+
"loss": 0.895,
|
2122 |
+
"step": 1430
|
2123 |
+
},
|
2124 |
+
{
|
2125 |
+
"epoch": 13.732057416267942,
|
2126 |
+
"grad_norm": 0.33203125,
|
2127 |
+
"learning_rate": 3.886175363359646e-06,
|
2128 |
+
"loss": 0.8964,
|
2129 |
+
"step": 1435
|
2130 |
+
},
|
2131 |
+
{
|
2132 |
+
"epoch": 13.779904306220097,
|
2133 |
+
"grad_norm": 0.34375,
|
2134 |
+
"learning_rate": 3.5833325466437694e-06,
|
2135 |
+
"loss": 0.891,
|
2136 |
+
"step": 1440
|
2137 |
+
},
|
2138 |
+
{
|
2139 |
+
"epoch": 13.82775119617225,
|
2140 |
+
"grad_norm": 0.341796875,
|
2141 |
+
"learning_rate": 3.2925582252048338e-06,
|
2142 |
+
"loss": 0.8903,
|
2143 |
+
"step": 1445
|
2144 |
+
},
|
2145 |
+
{
|
2146 |
+
"epoch": 13.875598086124402,
|
2147 |
+
"grad_norm": 0.3359375,
|
2148 |
+
"learning_rate": 3.013888795328057e-06,
|
2149 |
+
"loss": 0.8926,
|
2150 |
+
"step": 1450
|
2151 |
+
},
|
2152 |
+
{
|
2153 |
+
"epoch": 13.923444976076555,
|
2154 |
+
"grad_norm": 0.330078125,
|
2155 |
+
"learning_rate": 2.7473591381266708e-06,
|
2156 |
+
"loss": 0.8921,
|
2157 |
+
"step": 1455
|
2158 |
+
},
|
2159 |
+
{
|
2160 |
+
"epoch": 13.971291866028707,
|
2161 |
+
"grad_norm": 0.357421875,
|
2162 |
+
"learning_rate": 2.4930026151759766e-06,
|
2163 |
+
"loss": 0.9007,
|
2164 |
+
"step": 1460
|
2165 |
+
},
|
2166 |
+
{
|
2167 |
+
"epoch": 14.0,
|
2168 |
+
"eval_loss": 2.5432519912719727,
|
2169 |
+
"eval_runtime": 0.5348,
|
2170 |
+
"eval_samples_per_second": 18.7,
|
2171 |
+
"eval_steps_per_second": 1.87,
|
2172 |
+
"step": 1463
|
2173 |
+
},
|
2174 |
+
{
|
2175 |
+
"epoch": 14.019138755980862,
|
2176 |
+
"grad_norm": 0.3359375,
|
2177 |
+
"learning_rate": 2.250851064337367e-06,
|
2178 |
+
"loss": 0.8857,
|
2179 |
+
"step": 1465
|
2180 |
+
},
|
2181 |
+
{
|
2182 |
+
"epoch": 14.066985645933014,
|
2183 |
+
"grad_norm": 0.330078125,
|
2184 |
+
"learning_rate": 2.0209347957732328e-06,
|
2185 |
+
"loss": 0.897,
|
2186 |
+
"step": 1470
|
2187 |
+
},
|
2188 |
+
{
|
2189 |
+
"epoch": 14.114832535885167,
|
2190 |
+
"grad_norm": 0.337890625,
|
2191 |
+
"learning_rate": 1.8032825881530213e-06,
|
2192 |
+
"loss": 0.8863,
|
2193 |
+
"step": 1475
|
2194 |
+
},
|
2195 |
+
{
|
2196 |
+
"epoch": 14.16267942583732,
|
2197 |
+
"grad_norm": 0.3359375,
|
2198 |
+
"learning_rate": 1.5979216850509848e-06,
|
2199 |
+
"loss": 0.8968,
|
2200 |
+
"step": 1480
|
2201 |
+
},
|
2202 |
+
{
|
2203 |
+
"epoch": 14.210526315789474,
|
2204 |
+
"grad_norm": 0.345703125,
|
2205 |
+
"learning_rate": 1.404877791536141e-06,
|
2206 |
+
"loss": 0.8934,
|
2207 |
+
"step": 1485
|
2208 |
+
},
|
2209 |
+
{
|
2210 |
+
"epoch": 14.258373205741627,
|
2211 |
+
"grad_norm": 0.3671875,
|
2212 |
+
"learning_rate": 1.2241750709546917e-06,
|
2213 |
+
"loss": 0.8868,
|
2214 |
+
"step": 1490
|
2215 |
+
},
|
2216 |
+
{
|
2217 |
+
"epoch": 14.30622009569378,
|
2218 |
+
"grad_norm": 0.341796875,
|
2219 |
+
"learning_rate": 1.055836141905553e-06,
|
2220 |
+
"loss": 0.8992,
|
2221 |
+
"step": 1495
|
2222 |
+
},
|
2223 |
+
{
|
2224 |
+
"epoch": 14.354066985645932,
|
2225 |
+
"grad_norm": 0.337890625,
|
2226 |
+
"learning_rate": 8.998820754091531e-07,
|
2227 |
+
"loss": 0.8934,
|
2228 |
+
"step": 1500
|
2229 |
+
},
|
2230 |
+
{
|
2231 |
+
"epoch": 14.401913875598087,
|
2232 |
+
"grad_norm": 0.3515625,
|
2233 |
+
"learning_rate": 7.563323922699983e-07,
|
2234 |
+
"loss": 0.8897,
|
2235 |
+
"step": 1505
|
2236 |
+
},
|
2237 |
+
{
|
2238 |
+
"epoch": 14.44976076555024,
|
2239 |
+
"grad_norm": 0.328125,
|
2240 |
+
"learning_rate": 6.25205060633205e-07,
|
2241 |
+
"loss": 0.8929,
|
2242 |
+
"step": 1510
|
2243 |
+
},
|
2244 |
+
{
|
2245 |
+
"epoch": 14.497607655502392,
|
2246 |
+
"grad_norm": 0.3359375,
|
2247 |
+
"learning_rate": 5.065164937354428e-07,
|
2248 |
+
"loss": 0.9045,
|
2249 |
+
"step": 1515
|
2250 |
+
},
|
2251 |
+
{
|
2252 |
+
"epoch": 14.545454545454545,
|
2253 |
+
"grad_norm": 0.328125,
|
2254 |
+
"learning_rate": 4.0028154785050063e-07,
|
2255 |
+
"loss": 0.8877,
|
2256 |
+
"step": 1520
|
2257 |
+
},
|
2258 |
+
{
|
2259 |
+
"epoch": 14.593301435406698,
|
2260 |
+
"grad_norm": 0.3515625,
|
2261 |
+
"learning_rate": 3.065135204296965e-07,
|
2262 |
+
"loss": 0.8995,
|
2263 |
+
"step": 1525
|
2264 |
+
},
|
2265 |
+
{
|
2266 |
+
"epoch": 14.641148325358852,
|
2267 |
+
"grad_norm": 0.3359375,
|
2268 |
+
"learning_rate": 2.2522414843748618e-07,
|
2269 |
+
"loss": 0.8919,
|
2270 |
+
"step": 1530
|
2271 |
+
},
|
2272 |
+
{
|
2273 |
+
"epoch": 14.688995215311005,
|
2274 |
+
"grad_norm": 0.353515625,
|
2275 |
+
"learning_rate": 1.5642360688225e-07,
|
2276 |
+
"loss": 0.8905,
|
2277 |
+
"step": 1535
|
2278 |
+
},
|
2279 |
+
{
|
2280 |
+
"epoch": 14.736842105263158,
|
2281 |
+
"grad_norm": 0.33203125,
|
2282 |
+
"learning_rate": 1.0012050754277802e-07,
|
2283 |
+
"loss": 0.8943,
|
2284 |
+
"step": 1540
|
2285 |
+
},
|
2286 |
+
{
|
2287 |
+
"epoch": 14.78468899521531,
|
2288 |
+
"grad_norm": 0.34375,
|
2289 |
+
"learning_rate": 5.632189789027687e-08,
|
2290 |
+
"loss": 0.8943,
|
2291 |
+
"step": 1545
|
2292 |
+
},
|
2293 |
+
{
|
2294 |
+
"epoch": 14.832535885167465,
|
2295 |
+
"grad_norm": 0.330078125,
|
2296 |
+
"learning_rate": 2.5033260206275277e-08,
|
2297 |
+
"loss": 0.8969,
|
2298 |
+
"step": 1550
|
2299 |
+
},
|
2300 |
+
{
|
2301 |
+
"epoch": 14.880382775119617,
|
2302 |
+
"grad_norm": 0.326171875,
|
2303 |
+
"learning_rate": 6.25851089636198e-09,
|
2304 |
+
"loss": 0.8944,
|
2305 |
+
"step": 1555
|
2306 |
+
},
|
2307 |
+
{
|
2308 |
+
"epoch": 14.92822966507177,
|
2309 |
+
"grad_norm": 0.330078125,
|
2310 |
+
"learning_rate": 0.0,
|
2311 |
+
"loss": 0.896,
|
2312 |
+
"step": 1560
|
2313 |
+
},
|
2314 |
+
{
|
2315 |
+
"epoch": 14.92822966507177,
|
2316 |
+
"eval_loss": 2.542541980743408,
|
2317 |
+
"eval_runtime": 0.5324,
|
2318 |
+
"eval_samples_per_second": 18.782,
|
2319 |
+
"eval_steps_per_second": 1.878,
|
2320 |
+
"step": 1560
|
2321 |
+
},
|
2322 |
+
{
|
2323 |
+
"epoch": 14.92822966507177,
|
2324 |
+
"step": 1560,
|
2325 |
+
"total_flos": 1.2217365722824704e+18,
|
2326 |
+
"train_loss": 1.0227742999027938,
|
2327 |
+
"train_runtime": 5355.4835,
|
2328 |
+
"train_samples_per_second": 18.668,
|
2329 |
+
"train_steps_per_second": 0.291
|
2330 |
+
}
|
2331 |
+
],
|
2332 |
+
"logging_steps": 5,
|
2333 |
+
"max_steps": 1560,
|
2334 |
+
"num_input_tokens_seen": 0,
|
2335 |
+
"num_train_epochs": 15,
|
2336 |
+
"save_steps": 100,
|
2337 |
+
"total_flos": 1.2217365722824704e+18,
|
2338 |
+
"train_batch_size": 8,
|
2339 |
+
"trial_name": null,
|
2340 |
+
"trial_params": null
|
2341 |
+
}
|