jialinyyzz/humanizer

VISIONPricing:Input $0.1 / Cached $0.08 / Output $0.2Concurrent Unit Cost:1Model Size:7.9BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 11, 2026License:apache-2.0Architecture:Transformer0.9K Open Weights Featherless Exclusive Cold

The jialinyyzz/humanizer is a 12 billion parameter model, fine-tuned from google/gemma-4-12B, designed to rewrite AI-generated text into more human-like prose. It maintains factual accuracy, preserving numbers, units, dates, names, and quotes, and supports both English and Chinese. This model excels at making AI drafts (emails, essays, reports) read naturally, with a focus on local, offline operation and robust quantization techniques for efficient deployment.

Loading preview...

Overview

The jialinyyzz/humanizer is a 12 billion parameter model, fine-tuned from Google's Gemma-4-12B, specifically engineered to transform AI-written drafts into text that reads as if a human wrote it. It operates locally and offline, ensuring privacy and efficiency. The model is designed to meticulously preserve all factual details such as numbers, units, dates, names, and quotes, while significantly improving the naturalness and flow of the language in both English and Chinese.

Key Capabilities

  • Human-like Rewriting: Rewrites AI-generated text (emails, essays, reports, forum posts) to sound natural and human, not robotic.
  • Fact Fidelity: Trained to maintain every number, unit, date, name, and quote from the original draft without alteration.
  • Multilingual Support: Processes and rewrites drafts in both English and Chinese.
  • Local Operation: Designed to run entirely offline, with dedicated applications for Mac (Apple silicon) and Windows, or via llama.cpp and other runtimes.
  • Advanced Quantization: Utilizes sophisticated quantization-aware training and distillation techniques to provide smaller GGUF files (down to 3.9 GB) with minimal quality loss, enabling deployment on devices with as little as 8 GB of memory.
  • High AI-Detector Evasion: Achieves 95% human judgment by Originality.ai at its strictest setting, significantly outperforming previous versions and public baselines.

Good For

  • Content Creators & Writers: Perfect for refining AI-generated content to remove robotic phrasing and enhance readability.
  • Professionals: Ideal for humanizing emails, reports, and other business communications drafted by AI.
  • Students & Academics: Useful for polishing essays and papers to ensure a natural, human tone while preserving factual integrity.
  • Developers & Integrators: Provides flexible deployment options via llama.cpp, MLX, transformers, vLLM, and Ollama, with detailed instructions and prompt formats for seamless integration into applications and agents.
  • Users with Limited Resources: Optimized quantized versions allow for efficient local execution on consumer hardware, including machines with 8 GB RAM.

Finetunes
2 models