Model Card for Gemma 4 E2B IT — Primary Healthcare Fine-tune
Fine-tuned version of unsloth/gemma-4-E2B-it on a large medical QA dataset for primary healthcare, developed as part of the CSCI 566 (Deep Learning) course project at USC.
Model Details
Model Description
This model is a LoRA fine-tune of Gemma 4 E2B IT, trained on ~146K medical multiple-choice QA examples to improve performance on clinical reasoning tasks relevant to primary healthcare. Both LoRA adapter weights and quantized GGUF files are provided.
Model type: Causal Language Model (LoRA fine-tune)
Answer medical multiple-choice questions and provide evidence-based clinical reasoning for primary healthcare queries.
Downstream Use
Can be further fine-tuned or used as a backbone for medical decision-support tools, clinical education assistants, or healthcare chatbots.
Out-of-Scope Use
This model is not intended for direct clinical diagnosis or as a replacement for qualified medical professionals. It should not be used for emergency medical advice or high-stakes clinical decisions without human oversight.
Bias, Risks, and Limitations
The model may produce plausible-sounding but incorrect medical information.
Performance may degrade on rare conditions, non-English queries, or highly specialized subspecialties.
Recommendations
Always consult a qualified healthcare professional for personal medical advice. Outputs from this model should be treated as informational only and verified against authoritative clinical sources.
~146,354 medical multiple-choice QA examples in ShareGPT chat format, covering a broad range of primary care topics including internal medicine, pharmacology, and clinical reasoning. A validation split of ~5,548 examples was held out during training.
The system prompt used during training:
"You are a knowledgeable medical AI assistant. Provide accurate, evidence-based medical information. Always recommend consulting a qualified healthcare professional for personal medical advice."
Training Procedure
Training Hyperparameters
Training regime: bf16 mixed precision
Fine-tuning method: LoRA via Unsloth + TRL SFTTrainer
Held-out validation split (~5,548 examples) from the same medical QA distribution as training data, plus standard medical benchmarks (MedQA, MedMCQA).
Metrics
Accuracy on multiple-choice medical QA benchmarks
Results
Benchmark
Base Model
Fine-tuned
CareQA
53.98%
36.61%
HEADQA
58.64%
40.59%
MedQA
47.13%
32.29%
MMLU (Medical)
62.26%
44.34%
MMLU-Pro (Medical)
32.76%
16.01%
PubMedQA
48.30%
34.10%
Summary
Fine-tuning Gemma 4 E2B IT on the primary healthcare dataset did not improve benchmark accuracy — performance declined across all benchmarks compared to the base model. This suggests the base model's general medical knowledge was partially overridden by the narrow training distribution, likely due to the small model capacity (2B parameters) and single epoch of training.
Technical Specifications
Model Architecture and Objective
Decoder-only transformer (Gemma 4 E2B IT) with LoRA adapters applied to attention projection layers. Trained with next-token prediction (SFT) on chat-formatted medical QA data.
