Many top language models now err on the side of caution, refusing harmless prompts that merely sound risky – an ‘over-refusal' behavior that affects their usefulness in real-world scenarios. A new ...

The latest addition is the Phi-4 Reasoning — a 14 billion-parameter model built by applying a supervised fine-tuning (SFT) algorithm to the Phi-4 base model. The researchers also derived the Phi-4 ...

Tencent Holdings' stockpiles of AI chips should protect it from U.S. restrictions, its president said on Wednesday, after it ...

Oklo stock soared Wednesday after the nuclear-power startup reported a better-than-expected first-quarter loss and reiterated ...