File size: 3,253 Bytes
b750673
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d64006e
 
 
 
 
 
 
 
 
 
 
b750673
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d64006e
 
 
 
b750673
 
 
 
 
 
 
 
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
---
license: mit
base_model: naver-clova-ix/donut-base
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: donut-base-sroie-metrics-combined-new
  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. -->

# donut-base-sroie-metrics-combined-new

This model is a fine-tuned version of [naver-clova-ix/donut-base](https://huggingface.co/naver-clova-ix/donut-base) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.3400
- Bleu score: 0.0856
- Precisions: [0.8478260869565217, 0.8017817371937639, 0.7755102040816326, 0.755223880597015]
- Brevity penalty: 0.1078
- Length ratio: 0.3099
- Translation length: 506
- Reference length: 1633
- Cer: 0.7597
- Wer: 0.8305
- Cer Hugging Face: 0.7664
- Wer Hugging Face: 0.8347

## 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: 1
- eval_batch_size: 1
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 2
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 4
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step | Validation Loss | Bleu score | Precisions                                                                       | Brevity penalty | Length ratio | Translation length | Reference length | Cer    | Wer    | Cer Hugging Face | Wer Hugging Face |
|:-------------:|:-----:|:----:|:---------------:|:----------:|:--------------------------------------------------------------------------------:|:---------------:|:------------:|:------------------:|:----------------:|:------:|:------:|:----------------:|:----------------:|
| 0.9692        | 1.0   | 253  | 0.4901          | 0.0746     | [0.8011928429423459, 0.726457399103139, 0.6760925449871465, 0.6295180722891566]  | 0.1058          | 0.3080       | 503                | 1633             | 0.7672 | 0.8440 | 0.7741           | 0.8478           |
| 0.437         | 2.0   | 506  | 0.3906          | 0.0824     | [0.8382642998027613, 0.7755555555555556, 0.7353689567430025, 0.6964285714285714] | 0.1085          | 0.3105       | 507                | 1633             | 0.7611 | 0.8328 | 0.7675           | 0.8367           |
| 0.2997        | 3.0   | 759  | 0.3565          | 0.0858     | [0.828125, 0.778021978021978, 0.7462311557788944, 0.718475073313783]             | 0.1120          | 0.3135       | 512                | 1633             | 0.7640 | 0.8363 | 0.7703           | 0.8397           |
| 0.2168        | 4.0   | 1012 | 0.3400          | 0.0856     | [0.8478260869565217, 0.8017817371937639, 0.7755102040816326, 0.755223880597015]  | 0.1078          | 0.3099       | 506                | 1633             | 0.7597 | 0.8305 | 0.7664           | 0.8347           |


### Framework versions

- Transformers 4.41.0.dev0
- Pytorch 2.1.0
- Datasets 2.19.0
- Tokenizers 0.19.1