gemma-medical-qa-lora

Gemma Medical QA LoRA (ChatDoctor-HealthCareMagic)

This repository contains a LoRA adapter fine-tuned on a medical Q&A dataset to adapt google/gemma-2-2b-it for healthcare-style question answering.
It is not a full model checkpoint; you must load the base model and then attach this LoRA adapter.

Model Details

Model Description

• Model type: LoRA adapter for a decoder-only causal language model (Gemma-2B-IT)
• Base model: google/gemma-2-2b-it (gated on Hugging Face; requires access + authentication)
• Task: Medical question answering / health advice style responses (English)
• Training method: Parameter-efficient fine-tuning (PEFT) with QLoRA (4-bit quantization + LoRA)
• Language(s): English
• License: Same usage constraints as the base model and the dataset; check both before deployment

Model Sources

• Base model: google/gemma-2-2b-it
• Dataset: lavita/ChatDoctor-HealthCareMagic-100k

Intended Use

Direct Use

• Educational / research use for demonstrating QLoRA + LoRA fine-tuning on a medical QA dataset
• Running inference for medical-QA style prompts with safety disclaimers

Out-of-Scope Use

• Not for real-world medical diagnosis, triage, or treatment decisions
• Not a substitute for licensed clinical advice
• High-risk usage (children, pregnancy, emergencies, prescription decisions) should be avoided without professional oversight

Bias, Risks, and Limitations

• Medical datasets may contain incorrect, outdated, or unsafe advice.
• The model may hallucinate facts, miss contraindications, or fail to identify emergencies.
• Outputs can reflect dataset biases and may under-serve certain populations or rare conditions.
• This adapter was trained for demonstration purposes and may not be clinically validated.

Recommendations

• Always include a disclaimer that outputs are informational and not medical advice.
• Encourage users to consult a clinician for diagnosis/treatment.
• Do not use as a standalone triage tool; consider adding rule-based safety filters for red-flag symptoms.
• Evaluate on curated clinical benchmarks before any serious use.

How to Get Started

1) Install dependencies

pip install -U transformers accelerate bitsandbytes peft

Load Base Model + Attach LoRA Adapter & Run Inference

Important
google/gemma-2-2b-it is a gated model on Hugging Face.
You must have access to the model and be authenticated (e.g. huggingface-cli login) before running the code below.

---

1) Load the Base Model and Attach the LoRA Adapter

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig
from peft import PeftModel

BASE_MODEL = "google/gemma-2-2b-it"
ADAPTER_REPO = "KULLANICI_ADIN/gemma-medical-qa-lora"  # change to your Hugging Face repo id

# 4-bit quantization config (QLoRA-style loading)
bnb_config = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_quant_type="nf4",
    bnb_4bit_compute_dtype=torch.float16,
    bnb_4bit_use_double_quant=True,
)

# Load tokenizer and base model
tokenizer = AutoTokenizer.from_pretrained(BASE_MODEL, token=True)
base_model = AutoModelForCausalLM.from_pretrained(
    BASE_MODEL,
    device_map="auto",
    quantization_config=bnb_config,
    token=True,
)

# Attach LoRA adapter
model = PeftModel.from_pretrained(base_model, ADAPTER_REPO)
model.eval()

How to Get Started with the Model

Use the code below to get started with the model.

[More Information Needed]

Training Details

Training Data

<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->

[More Information Needed]

Training Procedure

<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->

Preprocessing [optional]

[More Information Needed]

Training Hyperparameters

• Training regime: [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->

Speeds, Sizes, Times [optional]

<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->

[More Information Needed]

Evaluation

<!-- This section describes the evaluation protocols and provides the results. -->

Testing Data, Factors & Metrics

Testing Data

<!-- This should link to a Dataset Card if possible. -->

[More Information Needed]

Factors

<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->

[More Information Needed]

Metrics

<!-- These are the evaluation metrics being used, ideally with a description of why. -->

[More Information Needed]

Results

[More Information Needed]

Summary

Model Examination [optional]

<!-- Relevant interpretability work for the model goes here -->

[More Information Needed]

Environmental Impact

<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->

Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019).

• Hardware Type: [More Information Needed]
• Hours used: [More Information Needed]
• Cloud Provider: [More Information Needed]
• Compute Region: [More Information Needed]
• Carbon Emitted: [More Information Needed]

Technical Specifications [optional]

Model Architecture and Objective

[More Information Needed]

Compute Infrastructure

[More Information Needed]

Hardware

[More Information Needed]

Software

[More Information Needed]

Citation [optional]

<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->

BibTeX:

[More Information Needed]

APA:

[More Information Needed]

Glossary [optional]

<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->

[More Information Needed]

More Information [optional]

[More Information Needed]

Model Card Authors [optional]

[More Information Needed]

Model Card Contact

[More Information Needed]

Framework versions

• PEFT 0.18.0