peterbuitho/VietPoet-Qwen3.5-9B
VietPoet-Qwen3.5-9B by peterbuitho is a 9 billion parameter Qwen3.5-based language model specifically fine-tuned for generating Vietnamese lục bát poems. It excels at adhering to the formal rules of lục bát poetry, achieving high rule scores when used with a line-by-line sampler and rule checker. This model is optimized for creative writing tasks focused on a specific poetic form, offering a specialized tool for Vietnamese poetry generation.
Loading preview...
Overview
peterbuitho/VietPoet-Qwen3.5-9B is a specialized 9 billion parameter language model built upon the Qwen3.5 architecture. It has been fine-tuned using QLoRA on a dataset of 8,000 Vietnamese poems to generate lục bát poetry, a traditional Vietnamese poetic form.
Key Capabilities
- Lục Bát Poetry Generation: The model is specifically trained to produce Vietnamese lục bát poems, focusing on their unique structure and rules.
- High Formal Accuracy: When integrated with a line-by-line sampler and rule checker (from github.com/peterbuitho/ThoLucBat), the model achieves a mean rule score of 0.994 and generates 98% fully valid poems in terms of form.
- Qwen Chat Format: It utilizes the Qwen chat format for prompting, expecting specific system and user turns to guide poem generation.
Training Details
The model was fine-tuned using QLoRA (4-bit base) with specific parameters including r=16, alpha=16, lr 2e-4, and 2 epochs. The training data was sourced from the phamson02/vietnamese-poetry-corpus, filtered to ensure compliance with lục bát rules.
Good For
- Automated Vietnamese Poetry Creation: Ideal for applications requiring the generation of formally correct Vietnamese lục bát poems.
- Integration with Rule Checkers: Best utilized in conjunction with external rule-checking and sampling mechanisms to maximize formal accuracy.
- Creative Writing Tools: Suitable for developers building tools for creative writing, particularly in the context of Vietnamese literature.