fermion-research 0.1.8
Five-value sub-2-bit LLMs: chat with a ~2 GB 8B container at native-runtime speed, load it as a Transformers model, or serve it on an OpenAI-compatible endpoint
Five-value sub-2-bit LLMs: chat with a ~2 GB 8B container at native-runtime speed, load it as a Transformers model, or serve it on an OpenAI-compatible endpoint

Tom Lee, Fundstrat Global Advisors' head of research, calls the market's AI capex fear a bullish sign. He does not read it as a warning of an approaching top. Steve Eisman, known for shorting the 2008 housing bubble, warned this week that markets could fall s…

中国企業のMoonshot AIが自社製AIモデル「Kimi K3」をオープンモデルとして公開しました。Releasing the model weights and technical report of Kimi K3.Kimi K3 is our most capable model: a 2.8T MoE model with native visual understanding and a 1M-token context window.New model architecture: 2.5x the…

AI's success in solving complex math problems may accelerate advancements in computational research, reshaping future problem-solving approaches. The post AI cracks second FrontierMath benchmark problem on absolute Galois groups, signaling a shift for computa…

Your thymus begins shrinking in your 20s, and it doesn't stop. This small gland tucked behind your breastbone slowly transforms from active immune tissue into fat, and for years, scientists treated that decline as a harmless quirk of aging. New research sugge…