新浪网29 天
为超越 OpenAI GPT-4，Meta 不惜使用争议数据训练 Llama 3
这些文件显示，Meta 在开发其最新 AI 模型 Llama 3 的过程中，高管和研究人员将超越 OpenAI 的 GPT-4 视为核心目标，并在内部讨论中表现出极强的竞争 ...
TweakTown6 天
Meta accused of downloading torrents of 81.7TB of pirated books to train its Llama AI models
TL;DR: Meta is accused of illegally torrenting 81.7TB of pirated books to train its Llama AI models, according to a lawsuit in California. Authors allege Meta used sources like Z-Library and ...
Digit13 天
DeepSeek vs Meta: 5 Things Mark Zuckerberg Teased About Llama 4 and the Future of Open ...
With Meta’s Llama series already making waves in the AI community, the forthcoming Llama 4 promises to push boundaries even further. Zuckerberg outlined ambitious plans for this next-generation ...
Insider Monkey8 天
Meta Platforms, Inc. (META) Lags in AI Race as DeepSeek Surpasses Llama Model
The Piper Sandler note has revealed that DeepSeek is ahead of Meta’s latest Llama model, but lags behind OpenAI and Anthropic by about six months. The firm further indicates that Meta has ...