michsethowusu/opani-coder_1b-merged-16bit

Opani Coder 1B (16-bit)

A fine-tuned Llama 3.2 1B model for coding assistance in Twi (Akan language), helping Twi speakers learn programming in their native language.

Model Details

• Base Model: Llama 3.2 1B
• Precision: 16-bit full model
• Language: Twi with code examples
• Fine-tuning: Full model fine-tuning
• Developed by: michsethowusu

Installation

pip install torch transformers

Usage

You can test the model using this HF Space - https://huggingface.co/spaces/michsethowusu/Opani-Coder-DEMO

from transformers import AutoTokenizer, AutoModelForCausalLM, TextStreamer
import torch

# Load model
model_id = "michsethowusu/opani-coder_1b-merged-16bit"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.float16,
    device_map="auto",
)

# Prepare message
messages = [
    {"role": "user", "content": "Kyerɛkyerɛ nea for loop yɛ"}
]

text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True,
)

# Generate response
streamer = TextStreamer(tokenizer, skip_prompt=True)
_ = model.generate(
    **tokenizer(text, return_tensors="pt").to(model.device),
    max_new_tokens=512,
    temperature=0.7,
    top_p=0.9,
    streamer=streamer,
)

Use Cases

• Explaining programming concepts in Twi
• Code generation with Twi commentary
• Debugging assistance in Twi
• Translating coding tutorials to Twi

Limitations

• Optimized for Twi language programming interactions
• May mix English technical terms
• Verify generated code before production use

Citation

@misc{opani-coder_1b_2024,
  author = {michsethowusu},
  title = {Opani Coder 1B: Fine-tuned Llama 3.2 1B for Twi Coding Assistance},
  year = {2024},
  publisher = {HuggingFace},
  url = {https://huggingface.co/michsethowusu/opani-coder_1b-merged-16bit}
}

Contact

• HuggingFace: @michsethowusu