File size: 881 Bytes
2caf5a6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
license: mit
datasets:
- vishnun/CodevsNL
language:
- en
metrics:
- accuracy
library_name: transformers
pipeline_tag: text-classification
tags:
- code
- nli
---

## PreFace
Code vs Natural language classification using bert-small from prajwall, below are the metrics achieved

## Training Metrics

| Epoch | Training Loss | Validation Loss | Accuracy |
|-------|---------------|-----------------|----------|
| 1     | 0.022500      | 0.012705        | 0.997203 |
| 2     | 0.008700      | 0.013107        | 0.996880 |
| 3     | 0.002700      | 0.014081        | 0.997633 |
| 4     | 0.001800      | 0.010666        | 0.997526 |
| 5     | 0.000900      | 0.010800        | 0.998063 |

## More

- Github repo for installable python package: https://github.com/Vishnunkumar
- Space on the extraction of code blocks from screenshots: https://huggingface.co/spaces/vishnun/SnapCode