---
tags:
- text-generation
- transformers
- peft
- qlora
- bitsandbytes
- mistral
- mistral-7b
- fine-tune
license: apache-2.0
---

# my-qlora-mistral7b-instruct

This is a **QLoRA fine-tuned** version of the [mistralai/Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) model.
It was fine-tuned using **Low-Rank Adaptation (LoRA)** in 4-bit precision for efficiency on consumer GPUs.

## 🚀 Model Details
- **Base model**: mistralai/Mistral-7B-Instruct-v0.2
- **Fine-tuning method**: QLoRA with PEFT
- **Quantization**: 4-bit (bitsandbytes)
- **Task**: Instruction following / conversational AI
- **Dataset**: Custom instruction-response pairs
- **Training environment**: Google Colab Pro (T4 / A100 GPU)

## 📦 How to Use
```python
# First, make sure you have the necessary libraries installed:
# pip install transformers peft bitsandbytes accelerate

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline, BitsAndBytesConfig
from peft import PeftModel
from accelerate import infer_auto_device_map, dispatch_model

fine_tuned_model_id = "Falah/my-qlora-mistral7b-instruct"
base_model_id = "mistralai/Mistral-7B-Instruct-v0.2"

print("Loading tokenizer...")
tokenizer = AutoTokenizer.from_pretrained(fine_tuned_model_id)

print("Loading base model with quantization...")
bnb_config = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_use_double_quant=True,
    bnb_4bit_quant_type="nf4",
    bnb_4bit_compute_dtype=torch.float16
)
base_model = AutoModelForCausalLM.from_pretrained(
    base_model_id,
    quantization_config=bnb_config,
    device_map=None, # Load to CPU initially
    torch_dtype=torch.float16,
    trust_remote_code=True,
)

print("Loading PEFT adapter onto the base model...")
model = PeftModel.from_pretrained(base_model, fine_tuned_model_id)

print("Dispatching model to devices...")
device_map = infer_auto_device_map(model, dtype=torch.float16)
model = dispatch_model(model, device_map=device_map)

# Ensure the model is in evaluation mode
model.eval()

print("Creating text generation pipeline...")
generator = pipeline(
    "text-generation",
    model=model,
    tokenizer=tokenizer,
    torch_dtype=torch.float16,
    device_map="auto",
)

# Define a sample user prompt
user_prompt = "Write a short story about a robot learning to love."

# Format the prompt
formatted_prompt = f"[INST] {user_prompt} [/INST]"

# Generate text
outputs = generator(
    formatted_prompt,
    max_new_tokens=200,
    num_return_sequences=1,
    do_sample=True,
    temperature=0.7,
    top_k=50,
    top_p=0.95,
)

# Print the generated text
for i, output in enumerate(outputs):
    print(f"Generated Output {i+1}:\n{output['generated_text']}")
```