quantification by phantom

According to the news of "brokerage China" on the evening of 25th, the relevant personage of Mirage Quantitative responded to the news of "DeepSeek's new-generation AI model R2 was released earlier", saying: the official news shall prevail. On July 17, 2023, Mirage Quantitative established DeepSeek AI Company, and then released the DeepSeek-R1 model in January this year. Previously, Reuters yesterday cited three people familiar with the matter as saying that DeepSeek is accelerating the release of its R2 model, originally planned for May, but is now working to bring it forward. The sources said...

DeepSeek, an AI company under Phantom Quantitative, officially released the DeepSeek-R1 model on January 20 and open-sourced the model weights at the same time. According to the official introduction, DeepSeek-R1 uses reinforcement learning technology on a large scale in the post-training phase, which greatly improves the model's reasoning ability with only a very small amount of labeled data. In math, code, natural language reasoning and other tasks, the performance is comparable to the official version of OpenAI o1. DeepSeek claims that the DeepSeek-R1 distillation mini-model outperforms Ope...