GitMylo/nsfwcaption-qwen3-vl-8b-v2-safetensors
GitMylo/nsfwcaption-qwen3-vl-8b-v2-safetensors is an 8 billion parameter Qwen3 VL model developed by GitMylo, specifically fine-tuned for NSFW image captioning. This model is designed to better understand and describe NSFW concepts in images, providing uncensored captions while remaining censored for general instruction tasks. Its primary strength lies in generating detailed descriptions for explicit visual content.
Loading preview...
Model Overview
GitMylo/nsfwcaption-qwen3-vl-8b-v2-safetensors is an 8 billion parameter vision-language model based on the Qwen3 architecture, developed by GitMylo. This version is a fine-tuned iteration specifically optimized for generating captions for NSFW (Not Safe For Work) images. The model aims to provide more accurate and detailed descriptions of explicit visual content.
Key Capabilities
- NSFW Image Captioning: The model has been fine-tuned to better understand and describe NSFW concepts within images.
- Uncensored Caption Generation: It generates uncensored captions for explicit content, distinguishing itself from models that might refuse or censor such descriptions.
Limitations and Considerations
- Instruction Following: While uncensored for captioning, the model remains censored for general instruction-following tasks and will refuse NSFW instructions outside of image captioning.
- Captioning Accuracy: Users should be aware that the model might occasionally misidentify characters or provide incorrect captions. Providing additional context to the model is recommended for improved accuracy.
- Character Mix-ups: There is a known issue where the model might mix up characters or provide inaccurate details; prompting to exclude character names can mitigate this.