File size: 2,209 Bytes
9d1ffbc
aff08a4
 
9d1ffbc
 
 
aff08a4
9d1ffbc
 
 
aff08a4
9d1ffbc
 
aff08a4
 
 
 
 
 
14a4094
aff08a4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
14a4094
 
 
 
 
 
 
 
 
 
 
aff08a4
 
 
 
 
 
 
 
 
 
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
---
title: CodeLlama Code Generator
emoji: 🦙
colorFrom: blue
colorTo: green
sdk: gradio
sdk_version: 4.19.2
app_file: app.py
pinned: false
license: bigscience-openrail-m
short_description: Interactive CodeLlama code generation demo
---

# CodeLlama Code Generator

This is an interactive demo of the CodeLlama-7b model for generating code completions. The application provides a simple interface where you can enter a code prompt and get AI-generated code completions.

## Features

- Interactive code generation with CodeLlama-7b model
- Adjustable parameters (temperature, max length, etc.)
- Example prompts to get started quickly
- Real-time generation with timing information

## How to Use

1. Enter a code prompt in the input box (e.g., a function signature or class definition)
2. Adjust the generation parameters if needed:
   - **Max Length**: Controls the maximum length of the generated text
   - **Temperature**: Controls randomness (lower = more deterministic)
   - **Top-p**: Controls diversity via nucleus sampling
   - **Top-k**: Controls diversity via top-k sampling
3. Click "Generate Code" to get your completion
4. Try different prompts and parameters to see how they affect the output

## Examples

The demo includes several example prompts to help you get started:

- Function to implement exponential backoff for network pings
- Fibonacci sequence implementation
- Binary search tree class
- Asynchronous data fetching function

## Authentication Requirements

**Important**: CodeLlama is a gated model that requires authentication to access. To use this demo:

1. You must accept the model's license at [meta-llama/CodeLlama-7b-hf](https://huggingface.co/meta-llama/CodeLlama-7b-hf)
2. You need to set your Hugging Face token in the Space's settings:
   - Go to Settings > Repository Secrets > Add
   - Create a secret named `HF_TOKEN` with your Hugging Face token as the value

Without proper authentication, the demo will show a limited interface with instructions.

## Technical Details

This demo uses:
- CodeLlama-7b model from Meta
- Hugging Face Transformers library
- Gradio for the web interface

## License

This demo is provided under the BigScience OpenRAIL-M license.