Instructions to use MatteoKhan/CodeLlama-7B-Merged-Python with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use MatteoKhan/CodeLlama-7B-Merged-Python with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="MatteoKhan/CodeLlama-7B-Merged-Python")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("MatteoKhan/CodeLlama-7B-Merged-Python")
model = AutoModelForCausalLM.from_pretrained("MatteoKhan/CodeLlama-7B-Merged-Python")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use MatteoKhan/CodeLlama-7B-Merged-Python with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "MatteoKhan/CodeLlama-7B-Merged-Python"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "MatteoKhan/CodeLlama-7B-Merged-Python",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/MatteoKhan/CodeLlama-7B-Merged-Python

SGLang

How to use MatteoKhan/CodeLlama-7B-Merged-Python with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "MatteoKhan/CodeLlama-7B-Merged-Python" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "MatteoKhan/CodeLlama-7B-Merged-Python",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "MatteoKhan/CodeLlama-7B-Merged-Python" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "MatteoKhan/CodeLlama-7B-Merged-Python",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use MatteoKhan/CodeLlama-7B-Merged-Python with Docker Model Runner:
```
docker model run hf.co/MatteoKhan/CodeLlama-7B-Merged-Python
```

🚀 CodeLlama-Hybrid-7B: Optimized for Code Generation

📌 Overview

CodeLlama-Hybrid-7B is an experimental hybrid language model that merges the capabilities of two CodeLlama variants. Built using MergeKit, this model is optimized for programming-related tasks, balancing efficiency and performance in code generation and understanding.

🔗 Created by: Matteo Khan
🎓 Affiliation: Apprentice at TW3 Partners (Generative AI Research)
📍 License: MIT

🔗 Connect with me on LinkedIn
🔍 Model on Hugging Face

🧠 Model Details

Model Type: Hybrid Language Model (Merged for Code Generation)
Parent Models:
- CodeLlama-7B
- CodeLlama-7B-Python
Merging Technique: Linear Merge (MergeKit)
Tokenizer Source: codellama/CodeLlama-7b-hf

🎯 Intended Use

This model is designed for code-related tasks and experimentation in hybrid model optimization. Possible applications include:

✅ Code Generation
✅ Code Completion & Assistance
✅ Code Understanding & Refactoring
✅ Exploration of Model Merging Effects on Programming Tasks

⚠️ Limitations & Considerations

While CodeLlama-Hybrid-7B provides enhanced code generation capabilities, it inherits some limitations from its parent models:

❌ May produce incorrect or insecure code
⚠️ Can generate biased, offensive, or inappropriate content
🔄 Merging may introduce unpredictable behaviors
📉 Performance may vary depending on the programming language and context

🔬 Merging Process & Configuration

This is not a newly trained model, but rather a merge of existing models using the following configuration:

merge_method: linear
dtype: float16
allow_crimes: true
models:
  - model: "codellama/CodeLlama-7b-hf"
    parameters:
      t: 1.0
      weight: 0.5
  - model: "codellama/CodeLlama-7b-Python-hf"
    parameters:
      t: 1.0
      weight: 0.5
parameters:
  normalize: true
  int8_mask: false
  ignore_mismatched_sizes: true
layers:
  - pattern: "model.*"
tokenizer_source: "codellama/CodeLlama-7b-hf"

📊 No formal evaluation has been conducted yet. Users are encouraged to benchmark and share feedback!

🌍 Environmental Impact

By utilizing model merging instead of training from scratch, CodeLlama-Hybrid-7B significantly reduces computational and environmental costs.

🚀 How to Use

from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "MatteoKhan/CodeLlama-Hybrid-7B"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name)

# Example usage
prompt = "Write a Python function to calculate Fibonacci numbers."
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_length=200)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(response)

📩 Feedback & Contact: Reach out via Hugging Face.

🎉 Happy Coding! 🚀

Downloads last month: 13

Safetensors

Model size

7B params

Tensor type

F32

Model tree for MatteoKhan/CodeLlama-7B-Merged-Python

Base model

codellama/CodeLlama-7b-Python-hf

Finetuned

(8)

this model