Overview

Hi. This model is fine-tuned on Qwen 2.5 3B, optimized (somewhat) to generate better science fiction novels in Chinese. It is intended for research and educational use, particularly for generating stylistic, narrative-rich content reminiscent of Chinese sci-fi literature.

Base model: Qwen 2.5 3B
Fine-tuning method: QLoRA
Training purpose: To enhance narrative generation in the style of modern Chinese science fiction

Dataset is not released due to copyright concerns.

This is my first attempt at training and publishing a language model. I was a high school senior still learning about LLMs. The model should have limitations or unexpected behavior, and feedback is very welcome. ☺️☺️☺️

(Edited March 5, 2026)

license: cc-by-nc-4.0