File size: 2,410 Bytes
0fcd97e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
license: mit
library_name: transformers
---
# CodeGenModel

<!-- markdownlint-disable first-line-h1 -->
<!-- markdownlint-disable html -->
<!-- markdownlint-disable no-duplicate-header -->

<div align="center">
  <img src="figures/fig1.png" width="60%" alt="CodeGenModel" />
</div>
<hr>

<div align="center" style="line-height: 1;">
  <a href="LICENSE" style="margin: 2px;">
    <img alt="License" src="figures/fig2.png" style="display: inline-block; vertical-align: middle;"/>
  </a>
</div>

## 1. Introduction

CodeGenModel is specialized for code generation tasks. This model has been selected as the best checkpoint based on code generation benchmark performance.

<p align="center">
  <img width="80%" src="figures/fig3.png">
</p>

The model demonstrates outstanding performance in code-related tasks while maintaining strong general capabilities.

## 2. Evaluation Results

### Comprehensive Benchmark Results

<div align="center">

| | Benchmark | CodeModel-v1 | CodeModel-v2 | CodeGenModel |
|---|---|---|---|---|
| **Core Reasoning Tasks** | Math Reasoning | 0.510 | 0.535 | 0.550 |
| | Logical Reasoning | 0.789 | 0.801 | 0.819 |
| | Common Sense | 0.716 | 0.702 | 0.736 |
| **Language Understanding** | Reading Comprehension | 0.671 | 0.685 | 0.700 |
| | Question Answering | 0.582 | 0.599 | 0.607 |
| | Text Classification | 0.803 | 0.811 | 0.828 |
| | Sentiment Analysis | 0.777 | 0.781 | 0.792 |
| **Generation Tasks** | Code Generation | 0.615 | 0.631 | 0.650 |
| | Creative Writing | 0.588 | 0.579 | 0.610 |
| | Dialogue Generation | 0.621 | 0.635 | 0.644 |
| | Summarization | 0.745 | 0.755 | 0.767 |
| **Specialized Capabilities**| Translation | 0.782 | 0.799 | 0.804 |
| | Knowledge Retrieval | 0.651 | 0.668 | 0.676 |
| | Instruction Following | 0.733 | 0.749 | 0.676 |
| | Safety Evaluation | 0.718 | 0.701 | 0.767 |

</div>

### Overall Performance Summary
CodeGenModel achieves top performance on code generation tasks among all checkpoints in this training run.

## 3. How to Use

### Installation
```bash
pip install transformers
```

### Quick Start
```python
from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("CodeGenModel")
tokenizer = AutoTokenizer.from_pretrained("CodeGenModel")
```

## 4. License
This repository is licensed under the [MIT License](LICENSE).

## 5. Contact
Open an issue on our GitHub for questions.