llm2025_main_rev2_sft02

This is a LoRA adapter fine-tuned on structured output tasks (REV2 SFT02). It focuses on extracting pure Output only, strictly filtering spam/repetition.

Base Model

Qwen/Qwen3-4B-Instruct-2507

Training Data (Mixed)

• 75%: daichira/structured-hard-sft-4k (High Quality)
• 25%: u-10bei/structured_data_with_cot_dataset_512_v4 (Filtered Output-only)

Method

• Enhanced Output-only extraction
• Filters: Length Limit, Keyword Spam, Repetition Check
• Unsloth optimized training