Fine-tune base models with LoRA on ePub-derived book text to emulate author voices.
npx skills add https://github.com/anionzo/skill --skill book-sft-pipeline-anionzo
Please help me install this Agent Skill. Skill: book-sft-pipeline Source: https://github.com/anionzo/skill/tree/main/skills/book-sft-pipeline Command: npx skills add https://github.com/anionzo/skill --skill book-sft-pipeline-anionzo
Book data is often scattered, long-form, and difficult to structure for training. This Skill consolidates extraction, segmentation, and author-style fine-tuning into a repeatable pipeline.
Fine-tune a base model on a public-domain book to imitate its author's voice.
High-intent search queries and answers about installing and using this skill.
To create an SFT dataset from ePub books, extract text, segment long-form content, and apply diverse prompt templates. This pipeline outputs training JSONL datasets and validation scaffolds structured for author-voice fine-tuning.
Yes, you can train a base model with a LoRA adapter to capture distinctive cadence and diction. This approach applies author-voice style-transfer to emulate the specific writing style of a book's author.
The best way to structure long-form literature for model training is using a segmentation pipeline. This process breaks down scattered book data into tailored segments, enabling diverse instruction templates for fine-tuning.
You need a base model with a LoRA adapter to perform book style-transfer. Careful data curation and diverse prompt templates are also required to output training data, adapters, and evaluation artifacts.
Fine-tuning on book data requires careful data curation because book data is scattered and long-form. Proper curation and diverse prompt templates ensure the model accurately captures the author's distinctive voice rather than memorizing text.
phonowellConvert books into SFT-ready JSONL datasets and train LoRA models.
niznik-devOrchestrate SLURM fine-tuning jobs with torchtune and evaluate with inspect-ai.
muratcankoylanConvert ePub books into JSONL SFT datasets and train LoRA style models.
hoangnb24Automate ePub-to-JSONL dataset creation and LoRA fine-tuning.
