QLoRA (rank 16) adapter for microsoft/Phi-3.5-mini-instruct, trained via Unsloth for domain adaptation — next-token/causal-LM objective on lipid biochemistry and Raman/IR spectroscopy abstracts, not an instruction-tuned task.

Part of LipidOS, a citation-grounded RAG system for the Raman/lipid literature where every generated claim must cite retrieved evidence it actually came from.

107,665 PubMed abstracts (two MeSH-term pools: bulk lipid biochemistry/lipidomics/membrane biophysics, and Raman/IR/vibrational spectroscopy vocabulary), trained on 64,000 of them (60.6% of one epoch, step-capped to a fixed compute budget).

Held-out perplexity on the same 2,000-example set, base vs. fine-tuned:

Citation-grounding was separately verified to survive the fine-tune: 0/3 hallucinated evidence citations across real domain questions plus a deliberate out-of-domain probe (details in the main repo's docs/solutions.md).

This model is designed to answer only through a grounded-retrieval + citation-checking pipeline — it should not be used for open-ended generation without retrieved evidence attached. See scripts/build_evidence.py and scripts/generate_finetuned_answer.py for the reference pipeline.

A standalone merged version (no peft required) is available at srikarjy025/lipidos-phi3-domain-adapt-merged.

This llama model was trained 2x faster with Unsloth.