matthewagi/essay-author-rewrite-qwen3-0.6b-pg-jasper-step100
The matthewagi/essay-author-rewrite-qwen3-0.6b-pg-jasper-step100 is a 0.8 billion parameter Qwen3-based model, fine-tuned for 100 steps using Prime-RL. It specializes in rewriting essays in the style of Paul Graham, leveraging a Jasper PG author reward and original drift gate. This model is optimized for stylistic text transformation, specifically adapting existing text to a target author's writing style.
Loading preview...
Model Overview
This model, matthewagi/essay-author-rewrite-qwen3-0.6b-pg-jasper-step100, is a specialized Qwen3-based language model with 0.8 billion parameters. It has undergone 100 steps of fine-tuning using the Prime-RL framework, specifically designed for essay rewriting. The training incorporates a Jasper PG author reward and an original drift gate to guide the stylistic transformation.
Key Capabilities
- Authorial Style Transfer: Rewrites input text to emulate the writing style of a target author, specifically Paul Graham.
- Stylistic Consistency: Utilizes a reward model to maintain stylistic coherence during the rewriting process.
- Efficient Fine-tuning: Represents a checkpoint from a targeted Prime-RL run, demonstrating focused adaptation.
Training Details
The model was trained with specific parameters to achieve its stylistic rewriting capability:
- Target Author: Paul Graham
- Source Truncation: Processes the first 1024 model-tokenized source tokens.
- Max Completion Tokens: Generates completions up to 1400 tokens.
Use Cases
This model is particularly suited for applications requiring:
- Content Rephrasing: Adapting existing essays or articles to a distinct authorial voice.
- Creative Writing Assistance: Generating text that aligns with a specific stylistic persona.
- Educational Tools: Demonstrating or practicing different writing styles.