File size: 2,779 Bytes
9588452
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
license: apache-2.0
base_model: DevAibest/opus-mt-en-fr-finetuned-en-to-fr
tags:
- generated_from_trainer
datasets:
- opus_books
metrics:
- bleu
model-index:
- name: opus-mt-finetuned-en-to-fr
  results:
  - task:
      name: Sequence-to-sequence Language Modeling
      type: text2text-generation
    dataset:
      name: opus_books
      type: opus_books
      config: en-fr
      split: train
      args: en-fr
    metrics:
    - name: Bleu
      type: bleu
      value: 26.9265
---

<!-- 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. -->

# opus-mt-finetuned-en-to-fr

This model is a fine-tuned version of [DevAibest/opus-mt-en-fr-finetuned-en-to-fr](https://huggingface.co/DevAibest/opus-mt-en-fr-finetuned-en-to-fr) on the opus_books dataset.
It achieves the following results on the evaluation set:
- Loss: 1.6538
- Bleu: 26.9265
- Gen Len: 34.161

## 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: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 15

### Training results

| Training Loss | Epoch | Step | Validation Loss | Bleu    | Gen Len |
|:-------------:|:-----:|:----:|:---------------:|:-------:|:-------:|
| No log        | 1.0   | 398  | 1.6540          | 26.2729 | 34.3078 |
| 1.764         | 2.0   | 796  | 1.6364          | 26.484  | 34.4081 |
| 1.6142        | 3.0   | 1194 | 1.6313          | 26.533  | 34.273  |
| 1.5079        | 4.0   | 1592 | 1.6291          | 26.6765 | 34.1672 |
| 1.5079        | 5.0   | 1990 | 1.6323          | 26.8527 | 34.0352 |
| 1.4218        | 6.0   | 2388 | 1.6314          | 26.8336 | 34.1142 |
| 1.3423        | 7.0   | 2786 | 1.6316          | 26.8037 | 34.2929 |
| 1.2812        | 8.0   | 3184 | 1.6361          | 26.9189 | 34.2471 |
| 1.2421        | 9.0   | 3582 | 1.6369          | 26.9398 | 34.2674 |
| 1.2421        | 10.0  | 3980 | 1.6413          | 26.9029 | 34.2462 |
| 1.1991        | 11.0  | 4378 | 1.6445          | 26.9221 | 34.1811 |
| 1.1562        | 12.0  | 4776 | 1.6495          | 26.8817 | 34.249  |
| 1.1392        | 13.0  | 5174 | 1.6517          | 26.8928 | 34.1777 |
| 1.1194        | 14.0  | 5572 | 1.6531          | 26.8547 | 34.2342 |
| 1.1194        | 15.0  | 5970 | 1.6538          | 26.9265 | 34.161  |


### Framework versions

- Transformers 4.33.0
- Pytorch 2.0.0
- Datasets 2.1.0
- Tokenizers 0.13.3