Fable 5 Has a Built-In Anti-Distillation Mechanism — and Its False Positive Rate Is Absurdly High
量子位 QbitAI3 days agoIncidentAnthropic's Fable 5 is reported to include a built-in anti-distillation mechanism that intentionally lowers output quality when it suspects its responses are being used to train competing models. While the intent is to protect proprietary intelligence, the false positive rate is described as unreasonably high. This means ordinary developers and researchers may routinely receive degraded answers without knowing why.
我們讓 Claude 撰寫 CUDA 核心並教導開源模型！Hugging Face 發表 Upskill 專案★ 80
Hugging Face Blog137 days agoRelease### Background and Challenge: Why Is CUDA Programming So Hard for AI? CUDA (Compute Unified Device Architecture) is a parallel computing platform and…
投資於效能：利用 LLM 洞察微調小型模型 — CFM 案例研究★ 75
Hugging Face Blog558 days agoBusinessThis case study from Hugging Face details how quantitative asset management firm Capital Fund Management (CFM) has optimized its investment and research…
Google 推出全新開源大語言模型 Gemma 2：9B 與 27B 雙版本，效能超越同級模型★ 85
Hugging Face Blog717 days agoReleaseGoogle has officially launched the next generation of its open-source large language model, Gemma 2, with an initial release in two sizes — 9B (9 billion…
Hugging Face 開源 SD-Small 與 SD-Tiny 知識蒸餾程式碼與模型權重★ 70
Hugging Face Blog1,048 days agoReleaseWhile Stable Diffusion (SD) 1.5 has demonstrated powerful image generation capabilities, its 860 million parameter count still presents challenges for edge…