---
license: mit
tags:
- medical
- dental
- lora
- qwen
- instruction-tuning
- unsloth
- adapter
- peft
- transformers
---

# 🦷 doctor-dental-implant-LoRA-Qwen2.5-7B-Instruct

This is a **LoRA adapter** fine-tuned with [Unsloth](https://github.com/unslothai/unsloth) on a domain-specific dataset that combines:

- **Realistic doctor–patient conversations**  
- **Dental implant Q&A** extracted from Straumann® technical manuals

> 🔬 Designed to make Qwen2.5-7B-Instruct capable of answering both general health questions and dental-specific scenarios.

---

## 🧠 Base Model

- **Base**: [`Qwen/Qwen2.5-VL-7B-Instruct`](https://huggingface.co/Qwen/Qwen2.5-VL-7B-Instruct)
- **Adapter**: LoRA (PEFT-based)

This repo contains only the **LoRA adapter weights**, not the full model.

---

## 🗂 Files

| File                    | Purpose                            |
|-------------------------|-------------------------------------|
| `adapter_model.safetensors` | LoRA weight file (for PEFT loading) |
| `adapter_config.json`   | LoRA hyperparameter configuration |
| `tokenizer.json`, `vocab.json`, `merges.txt` | Tokenizer (shared with base model) |

---

## 📦 How to Use

```python
from transformers import AutoTokenizer, AutoModelForCausalLM
from peft import PeftModel

# Load base model
base = AutoModelForCausalLM.from_pretrained("Qwen/Qwen2.5-VL-7B-Instruct", trust_remote_code=True)
tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2.5-VL-7B-Instruct", trust_remote_code=True)

# Load adapter
model = PeftModel.from_pretrained(base, "BirdieByte1024/doctor-dental-implant-LoRA-Qwen2.5-7B-Instruct")