File size: 3,048 Bytes
5f3511d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
---
license: gemma
base_model: google/gemma-2b
tags:
- generated_from_trainer
model-index:
- name: G0515HMA22H
  results: []
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# G0515HMA22H

This model is a fine-tuned version of [google/gemma-2b](https://huggingface.co/google/gemma-2b) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.1322

## Model description

More information needed

## Intended uses & limitations

More information needed

## Training and evaluation data

More information needed

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 0.0003
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 16
- total_train_batch_size: 128
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine_with_restarts
- lr_scheduler_warmup_steps: 80
- num_epochs: 3
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 3.1841        | 0.09  | 10   | 2.8751          |
| 2.5107        | 0.18  | 20   | 1.9774          |
| 1.4768        | 0.27  | 30   | 0.8765          |
| 0.4892        | 0.36  | 40   | 0.2067          |
| 0.1718        | 0.45  | 50   | 0.1561          |
| 0.1522        | 0.54  | 60   | 0.1523          |
| 0.1497        | 0.63  | 70   | 0.1483          |
| 0.15          | 0.73  | 80   | 0.1496          |
| 0.1434        | 0.82  | 90   | 0.1502          |
| 0.1464        | 0.91  | 100  | 0.1485          |
| 0.1496        | 1.0   | 110  | 0.1545          |
| 0.1448        | 1.09  | 120  | 0.1500          |
| 0.1452        | 1.18  | 130  | 0.1482          |
| 0.1457        | 1.27  | 140  | 0.1470          |
| 0.1489        | 1.36  | 150  | 0.1466          |
| 0.1425        | 1.45  | 160  | 0.1505          |
| 0.1444        | 1.54  | 170  | 0.1477          |
| 0.1463        | 1.63  | 180  | 0.1462          |
| 0.1459        | 1.72  | 190  | 0.1485          |
| 0.1459        | 1.81  | 200  | 0.1476          |
| 0.1475        | 1.9   | 210  | 0.1470          |
| 0.1469        | 1.99  | 220  | 0.1461          |
| 0.1429        | 2.08  | 230  | 0.1453          |
| 0.1386        | 2.18  | 240  | 0.1431          |
| 0.14          | 2.27  | 250  | 0.1416          |
| 0.1406        | 2.36  | 260  | 0.1402          |
| 0.1379        | 2.45  | 270  | 0.1397          |
| 0.1349        | 2.54  | 280  | 0.1366          |
| 0.1358        | 2.63  | 290  | 0.1348          |
| 0.1319        | 2.72  | 300  | 0.1333          |
| 0.1337        | 2.81  | 310  | 0.1322          |
| 0.1322        | 2.9   | 320  | 0.1322          |
| 0.1335        | 2.99  | 330  | 0.1322          |


### Framework versions

- Transformers 4.36.0.dev0
- Pytorch 2.1.2+cu121
- Datasets 2.14.6
- Tokenizers 0.14.0