AI BriefingDeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic: 9B-sized DeepSeek-V4-Pro, Uncensored Version
xtools-at/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic
About the project
A fine-tuned model based on Jackrong's DeepSeek-V4-Pro-Qwen3.5-9B. It maintains the structure and functionality of the original model while being produced as 'Uncensored' and 'Abliterated' variants with response restrictions removed. It is distributed under the Apache 2.0 license, allowing free commercial and personal use.
With a scale of 9.4 billion parameters, it is optimized for text generation and reasoning tasks. It supports problem-solving in mathematics and STEM fields, as well as tool-use and function-calling capabilities. It features a multimodal (image-text-to-text) structure that processes inputs combining images and text.
It supports six languages, including English, Chinese, Korean, Japanese, Spanish, and Russian. It is compatible with the Transformers library and provides optimized inference performance through Unsloth. It has been downloaded 772 times and has 222 likes on Hugging Face.
It is suitable for developers who want more free responses by bypassing the safety measures or content filtering of existing models. It can be utilized when running an LLM in a local environment or on a self-hosted server, particularly when a model without restrictions on specific topics or formats is required.
xtools-at/DeepSeek-V4-Pro-Qwen3.5-9B-UNCENSORED-heretic
The original page has no description.
text-generation
This introduction was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report errors, attribution issues, or removal requests via Contact.