Quantum physicists have shrunk and “de-censored” DeepSeek R1

Quantum physicists at Multiverse Computing have developed a smaller, "de-censored" version of the AI reasoning model DeepSeek R1, eliminating Chinese censorship from the original model. By using tensor networks borrowed from quantum physics, they were able to significantly reduce the model's size while maintaining its performance. The modified model, DeepSeek R1 Slim, was tested on politically sensitive questions and produced uncensored, factual responses comparable to Western models. This work is part of a broader effort to compress and manipulate existing AI models, aiming to make them more efficient and cost-effective.