PDP Qwen3-8B RQ3 SFT LoRA Adapter
Model Summary
This repository contains an RQ3 supervised fine-tuning LoRA adapter for prosecution decision prediction (PDP) experiments based on Qwen/Qwen3-8B.
- Repository:
Julian2002/PDP-Qwen3-8B-SFT-LoRA - Source directory:
RQ3_SFT_Qwen3-8B_train_seed42_original - Upload mode:
adapter_only - Dataset:
PDP-Bench / pdp2k_rq3_sft
Task
Given suspect information, procedural information, and factual information, the model is trained to generate structured prosecutorial reasoning and a final decision in the PDP setting.
The target decision space contains four labels:
起诉相对不起诉法定不起诉存疑不起诉
Training Output Format
The supervised target follows the native Qwen3-style format used in this project:
<think>
...
</think>
【适用法条】
刑法：...
刑事诉讼法：...
刑事诉讼规则：...
【审查分析】
...
【最终结论】
决定：...
Loading
This repository stores a LoRA adapter, not a merged standalone base model. Load it together with Qwen/Qwen3-8B, or merge it into the base model before standalone inference or vLLM evaluation.
Notes
This model card was generated automatically by scripts/upload_RQ3_sft_checkpoint_to_hf.sh.
- Downloads last month
- 84