Input: "I want to undo my last commit but keep the files staged."
- Output: git reset --soft HEAD~1
Input: "Create a new branch called 'feature-login' and switch to it."
- Output: git checkout -b feature-login

🔧 Training Details

Model Type: Causal Language Model (Fine-tuned Llama-3)
Base Model: unsloth/llama-3-8b-bnb-4bit
Developed by: daimon-ml
Language: English (en)
License: Apache-2.0

This model was fine-tuned using Unsloth, allowing for faster training and efficient 4-bit quantization. It leverages the Llama-3 architecture to ensure high reasoning capabilities even for edge-case Git scenarios.

📦 How to Use

You can load this model using the transformers library or via unsloth:

python

from transformers import AutoTokenizer, AutoModelForCausalLM

model_name = "daimon-ml/gitfluent-llama3-8b"

tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name)

prompt = "How do I stash my changes?"
inputs = tokenizer(prompt, return_tensors="pt")

# Generate output
outputs = model.generate(**inputs, max_new_tokens=50)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))