File size: 2,574 Bytes
a2d70b7
 
 
 
 
 
ef86d3c
 
a2d70b7
 
 
 
 
 
 
ef86d3c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
a2d70b7
 
 
 
 
 
 
ef86d3c
a2d70b7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
95
96
97
---
library_name: transformers
license: mit
base_model: microsoft/mdeberta-v3-base
tags:
- generated_from_trainer
datasets:
- google/xtreme
metrics:
- precision
- recall
- f1
- accuracy
model-index:
- name: mdeberta-v3-base-panx-wikiann-en
  results:
  - task:
      name: Token Classification
      type: token-classification
    dataset:
      name: google/xtreme PAN-X.en
      type: google/xtreme
      args: PAN-X.en
    metrics:
    - name: Precision
      type: precision
      value: 0.8285338502007477
    - name: Recall
      type: recall
      value: 0.8461049059804892
    - name: F1
      type: f1
      value: 0.8372271964185787
    - name: Accuracy
      type: accuracy
      value: 0.9318317274262442
---

<!-- 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. -->

# mdeberta-v3-base-panx-wikiann-en

This model is a fine-tuned version of [microsoft/mdeberta-v3-base](https://huggingface.co/microsoft/mdeberta-v3-base) on the google/xtreme PAN-X.en dataset.
It achieves the following results on the evaluation set:
- Loss: 0.2520
- Precision: 0.8285
- Recall: 0.8461
- F1: 0.8372
- Accuracy: 0.9318

## 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: 4
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 8
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5.0
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1     | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:---------:|:------:|:------:|:--------:|
| 0.4565        | 1.0   | 625  | 0.2651          | 0.7942    | 0.8198 | 0.8068 | 0.9215   |
| 0.2612        | 2.0   | 1250 | 0.2490          | 0.8043    | 0.8285 | 0.8162 | 0.9257   |
| 0.2184        | 3.0   | 1875 | 0.2471          | 0.8175    | 0.8353 | 0.8263 | 0.9294   |
| 0.1636        | 4.0   | 2500 | 0.2493          | 0.8195    | 0.8434 | 0.8313 | 0.9308   |
| 0.1408        | 5.0   | 3125 | 0.2520          | 0.8285    | 0.8461 | 0.8372 | 0.9318   |


### Framework versions

- Transformers 4.44.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.1
- Tokenizers 0.19.1