katsukiono/Qwen2.5-7B-Instruct-pubmed-qlora-layer11-only-2048

Qwen2.5-7B-Instruct PubMedQA QLoRA Layer11-only (ctx=2048)

Adapter type: LoRA / QLoRA

Train data: PubMedQA (labeled + artificial)

Context length: 2048

Date: 2025-10-19

Usage

from peft import PeftModel from transformers import AutoModelForCausalLM, AutoTokenizer

base = "Qwen/Qwen2.5-7B-Instruct" adapter = "<your-repo-id>" # 例: katsukiono/Qwen2.5-7B-Instruct-pubmed-qlora-layer11-only-2048

tok = AutoTokenizer.frompretrained(base, usefast=True) model = AutoModelForCausalLM.frompretrained( base, devicemap="auto", torchdtype="auto", trustremotecode=True ) model = PeftModel.frompretrained(model, adapter) model.eval()

This repo contains a PEFT adapter (not full weights).