File size: 2,632 Bytes
197ce61
 
4f084da
 
 
 
 
 
 
 
197ce61
4f084da
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
license: mit
base_model: urduhack/roberta-urdu-small
tags:
- generated_from_trainer
datasets:
- wikiann
model-index:
- name: UrduNER
  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. -->

# UrduNER

This model is a fine-tuned version of [urduhack/roberta-urdu-small](https://huggingface.co/urduhack/roberta-urdu-small) on the wikiann dataset.
It achieves the following results on the evaluation set:
- Loss: 0.1163
- Overall Precision: 0.9540
- Overall Recall: 0.9553
- Overall F1: 0.9546
- Overall Accuracy: 0.9836
- Loc F1: 0.9643
- Org F1: 0.9448
- Per F1: 0.9491

## 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: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 7

### Training results

| Training Loss | Epoch | Step | Validation Loss | Overall Precision | Overall Recall | Overall F1 | Overall Accuracy | Loc F1 | Org F1 | Per F1 |
|:-------------:|:-----:|:----:|:---------------:|:-----------------:|:--------------:|:----------:|:----------------:|:------:|:------:|:------:|
| 0.248         | 1.0   | 1250 | 0.0920          | 0.8906            | 0.8991         | 0.8948     | 0.9687           | 0.9086 | 0.8686 | 0.8995 |
| 0.1169        | 2.0   | 2500 | 0.0761          | 0.9302            | 0.9390         | 0.9346     | 0.9791           | 0.9501 | 0.9045 | 0.9400 |
| 0.07          | 3.0   | 3750 | 0.0831          | 0.9394            | 0.9451         | 0.9422     | 0.9805           | 0.9505 | 0.9348 | 0.9361 |
| 0.029         | 4.0   | 5000 | 0.1102          | 0.9311            | 0.9431         | 0.9371     | 0.9784           | 0.9469 | 0.9305 | 0.9279 |
| 0.0134        | 5.0   | 6250 | 0.1225          | 0.9442            | 0.9519         | 0.9480     | 0.9820           | 0.9593 | 0.9438 | 0.9337 |
| 0.0107        | 6.0   | 7500 | 0.1087          | 0.9515            | 0.9566         | 0.9541     | 0.9837           | 0.9660 | 0.9423 | 0.9466 |
| 0.005         | 7.0   | 8750 | 0.1163          | 0.9540            | 0.9553         | 0.9546     | 0.9836           | 0.9643 | 0.9448 | 0.9491 |


### Framework versions

- Transformers 4.32.0
- Pytorch 2.0.1+cu118
- Datasets 2.14.4
- Tokenizers 0.13.3