---
base_model: Qwen/Qwen2.5-7B-Instruct
library_name: peft
pipeline_tag: text-generation
tags:
- qwen2.5
- peft
- lora
- frequent-itemset-extraction
---

# Qwen2.5-7B SFT Frequent Itemset Extractor

This repository stores the exact PEFT LoRA SFT adapter used for the thesis
evaluation. It is intended to be loaded on top of `Qwen/Qwen2.5-7B-Instruct` with PEFT.

- Adapter checkpoint SHA-256: `f1dfb89164225c3f1f107a486271acab6c2d076e791241360bdc0d6466e416e2`
- Expected loader: `Qwen2ForCausalLM.from_pretrained('Qwen/Qwen2.5-7B-Instruct', quantization_config=NF4, attn_implementation="sdpa")` plus `PeftModel.from_pretrained(...)`
- Expected eval profile: `primary_v3` two-phase decoding