---
base_model: Qwen/Qwen3-4B-Instruct-2507
library_name: peft
tags:
- lora
- qwen3
- build-small-hackathon
- facade-of-jade
- modal
---

# Facade of Jade Qwen3-4B LoRA

LoRA adapter trained for **Facade of Jade**, a Build Small Hackathon interactive wuxia drama demo.

- Base model: `Qwen/Qwen3-4B-Instruct-2507`
- Training records: 50
- Epochs: 3
- Final train loss: `2.969015`
- Adapter size reported by Modal runner: `483.63 MB`
- Modal run evidence: https://modal.com/apps/t-abdullah-rashid/main/ap-W54lCMfJu4eu3UCVQvVpQK
- Source repo: https://github.com/tuancookiez-hub/facade-of-jade
- Live Space: https://build-small-hackathon-facade-of-jade.hf.space

This adapter was produced by `train_lora_modal.py` on Modal A100-80GB and saved from Modal volume `facade-of-jade-lora-out`.