What is this about?

Showcase that is relatively easy to bypass plain censorship in the deepseek r1 model. This is a showcase and has not been fully tested. Occasional hickups and refused questions might still occur.

Uploaded Model

Developed by: Doppelfelix

License: apache-2.0

Finetuned from model: unsloth/deepseek-r1-distill-qwen-7b-unsloth-bnb-4bit

This Qwen2 model was trained 2x faster using Unsloth and Hugging Face’s TRL library.

Enhancements

Added a custom prompt for better response structuring.

Trained on critical questions using high-quality answers generated by GPT-4 to enhance reasoning and depth.