File size: 2,827 Bytes
da788d6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
license: apache-2.0
base_model: distilbert-base-multilingual-cased
tags:
- generated_from_trainer
metrics:
- accuracy
- precision
- recall
- f1
model-index:
- name: spa-eng-pos-tagging-v2
  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. -->

# spa-eng-pos-tagging-v2

This model is a fine-tuned version of [distilbert-base-multilingual-cased](https://huggingface.co/distilbert-base-multilingual-cased) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.4432
- Accuracy: 0.8418
- Precision: 0.8395
- Recall: 0.7600
- F1: 0.7676
- Hamming Loss: 0.1582

## 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: 1e-05
- train_batch_size: 16
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 12

### Training results

| Training Loss | Epoch | Step  | Validation Loss | Accuracy | Precision | Recall | F1     | Hamming Loss |
|:-------------:|:-----:|:-----:|:---------------:|:--------:|:---------:|:------:|:------:|:------------:|
| 1.4285        | 1.0   | 1744  | 1.2584          | 0.5671   | 0.6506    | 0.4372 | 0.4716 | 0.4329       |
| 1.1788        | 2.0   | 3488  | 1.0023          | 0.6388   | 0.6753    | 0.5323 | 0.5578 | 0.3612       |
| 0.9144        | 3.0   | 5232  | 0.7885          | 0.7093   | 0.7259    | 0.6091 | 0.6281 | 0.2907       |
| 0.78          | 4.0   | 6976  | 0.6970          | 0.7439   | 0.7517    | 0.6527 | 0.6673 | 0.2561       |
| 0.6565        | 5.0   | 8720  | 0.6072          | 0.7765   | 0.7792    | 0.6838 | 0.6952 | 0.2235       |
| 0.5845        | 6.0   | 10464 | 0.5438          | 0.7995   | 0.7974    | 0.7125 | 0.7221 | 0.2005       |
| 0.5158        | 7.0   | 12208 | 0.5127          | 0.8132   | 0.8180    | 0.7250 | 0.7362 | 0.1868       |
| 0.4697        | 8.0   | 13952 | 0.4939          | 0.8186   | 0.8188    | 0.7345 | 0.7438 | 0.1814       |
| 0.4251        | 9.0   | 15696 | 0.4712          | 0.8334   | 0.8349    | 0.7502 | 0.7591 | 0.1666       |
| 0.4039        | 10.0  | 17440 | 0.4564          | 0.8381   | 0.8382    | 0.7538 | 0.7629 | 0.1619       |
| 0.3826        | 11.0  | 19184 | 0.4479          | 0.8397   | 0.8399    | 0.7565 | 0.7656 | 0.1603       |
| 0.3691        | 12.0  | 20928 | 0.4432          | 0.8418   | 0.8395    | 0.7600 | 0.7676 | 0.1582       |


### Framework versions

- Transformers 4.32.0
- Pytorch 2.0.1+cu118
- Tokenizers 0.13.3