89 lines
2.3 KiB
Markdown
89 lines
2.3 KiB
Markdown
---
|
|
pipeline_tag: text-generation
|
|
---
|
|
|
|
## 🧠 Fine-tuned CodeLlama on LeetCode Problems
|
|
|
|
**This model is a fine-tuned version of [`codellama/CodeLlama-7b-Instruct-hf`](https://huggingface.co/codellama/CodeLlama-7b-Instruct-hf) on the [`greengerong/leetcode`](https://huggingface.co/datasets/greengerong/leetcode) dataset. It has been instruction-tuned to generate Python solutions from LeetCode-style problem descriptions.**
|
|
---
|
|
|
|
## 📦 Model Formats Available
|
|
|
|
- **Transformers-compatible (`.safetensors`)** — for use via 🤗 Transformers.
|
|
- **GGUF (`.gguf`)** — for use via [llama.cpp](https://github.com/ggerganov/llama.cpp), including `llama-server`, `llama-cpp-python`, and other compatible tools.
|
|
|
|
---
|
|
|
|
## 🔗 Example Usage (Transformers)
|
|
|
|
```python
|
|
from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
|
|
|
|
model_id = "harshism1/codellama-leetcode-finetuned"
|
|
tokenizer = AutoTokenizer.from_pretrained(model_id)
|
|
model = AutoModelForCausalLM.from_pretrained(model_id)
|
|
|
|
pipe = pipeline("text-generation", model=model, tokenizer=tokenizer)
|
|
|
|
prompt = """You are an AI assistant. Solve the following problem:
|
|
|
|
Given an array of integers, return indices of the two numbers such that they add up to a specific target.
|
|
|
|
## Solution
|
|
"""
|
|
|
|
result = pipe(prompt, max_new_tokens=256, do_sample=True, temperature=0.7)
|
|
print(result[0]["generated_text"])
|
|
```
|
|
|
|
|
|
|
|
## ⚙️ Usage with `llama.cpp`
|
|
|
|
You can run the model using tools in the [`llama.cpp`](https://github.com/ggerganov/llama.cpp) ecosystem. Make sure you have the `.gguf` version of the model (e.g., `codellama-leetcode.gguf`).
|
|
|
|
### 🐍 Using `llama-cpp-python`
|
|
|
|
Install:
|
|
|
|
```bash
|
|
pip install llama-cpp-python
|
|
```
|
|
Then use:
|
|
|
|
```
|
|
from llama_cpp import Llama
|
|
|
|
llm = Llama(
|
|
model_path="codellama-leetcode.gguf",
|
|
n_ctx=4096,
|
|
n_gpu_layers=99 # adjust based on your GPU
|
|
)
|
|
|
|
prompt = """### Problem
|
|
Given an array of integers, return indices of the two numbers such that they add up to a specific target.
|
|
|
|
## Solution
|
|
"""
|
|
|
|
output = llm(prompt, max_tokens=256)
|
|
print(output["choices"][0]["text"])
|
|
```
|
|
|
|
|
|
### 🖥️ Using llama-server
|
|
|
|
Start the server:
|
|
|
|
```
|
|
llama-server --model codellama-leetcode.gguf --port 8000 --n_gpu_layers 99
|
|
```
|
|
|
|
Then send a request:
|
|
|
|
```
|
|
curl http://localhost:8000/completion -d '{
|
|
"prompt": "### Problem\nGiven an array of integers...\n\n## Solution\n",
|
|
"n_predict": 256
|
|
}'
|
|
``` |