chartreuse-verte/prose-rewriter-4b-v1.4

TEXT GENERATIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:4BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Aug 31, 2026License:agpl-3.0Architecture:Transformer Open Weights Featherless Exclusive Cold

The chartreuse-verte/prose-rewriter-4b-v1.4 is a 4 billion parameter language model based on Qwen3-4B-Base, specifically designed for paragraph-level prose rewriting. It transforms LLM-generated text into more human-like prose while preserving original semantics. This model excels at increasing sentence length variety and reducing AI-generated linguistic patterns, making it ideal for refining fictional content.

Loading preview...

Overview

chartreuse-verte/prose-rewriter-4b-v1.4 is a specialized 4 billion parameter model built on Qwen/Qwen3-4B-Base with a merged LoRA adapter. Its primary function is to rewrite AI-generated prose at the paragraph level, making it sound more human while retaining the original meaning. This version significantly improves upon its predecessor, v1.3, by editing more frequently and introducing greater sentence length variety.

Key Capabilities and Features

  • Prose Rewriting: Transforms LLM-written text into more natural, human-like prose.
  • Semantic Preservation: Maintains the core meaning of the input paragraph during rewriting.
  • Enhanced Variety: Doubles sentence-length variety compared to v1.3, actively breaking and joining sentences.
  • Reduced AI Patterns: Significantly lowers the presence of 'banned constructions' and 'slop lexicon density' often found in AI-generated text.
  • Contextual Modes: Supports match, inflate, and compress modes to guide the rewriting process based on desired output length.
  • Optimized for Fictional Prose: Trained on diverse human prose datasets including writing prompts, fanfiction, and roleplay forum content.
  • Untied lm_head: Features an untied output head for improved performance and specific architectural integrity.

When to Use This Model

  • Refining LLM Outputs: Ideal for post-processing text generated by other large language models to improve its naturalness and readability.
  • Creative Writing: Useful for authors or content creators looking to humanize AI-assisted fictional prose.
  • Roleplay Content: Specifically trained with a roleplay-forum register, making it suitable for roleplay scenarios.
  • Paragraph-Level Edits: Best suited for inputs between 15 and 70 words; longer inputs should be split.

Limitations

  • Not an Instruct Model: Designed for a single, specific task with a fixed prompt format.
  • Fictional Prose Only: Not intended for technical documentation or non-narrative text.
  • English Only: Supports narrative English prose (third and first person fiction, dialogue).
  • AI Detector Vulnerability: Will not bypass AI detection tools as it preserves core word choices and structures.

Finetunes
1 models