10亿参数和30亿参数的轻量级纯文本模型 官方数据显示，与同等规模的“中小型”大模型相比，Llama 3.2 11B和90B表现出了超越闭源模型的性能。

今天凌晨，大新闻不断。一边是 OpenAI 的高层又又又动荡了，另一边被誉为「真・Open AI」的 Meta 对 Llama 模型来了一波大更新：不仅推出了支持图像 ...

在Scaling-up中，近几年的benchmarks越来越偏向于加入更多困难的例子，或者给予所谓“权威”来源更多权重，研究员也因此更倾向于优化模型在困难任务上的表现，导致在难度一致性上慢性恶化。

00 后国人一作登上 Nature，这篇大模型论文引起热议。简单来说，论文发现：更大且更遵循指令的大模型也变得更不可靠了，某些情况下 GPT-4 在回答可靠性上还不如 GPT-3 。 与早期模型相比，有更多算力和人类反馈加持的最新模型，在回答可靠性上实际愈加恶化了。

A new molecule, based on llama antibodies, could counteract opioid effects by preventing drugs from binding to brain ...

Meta’s multilingual Llama family of models has reached version 3.2, with the bump from 3.1 signifying that several Llama models are now multimodal. Llama 3.2 11B — a compact model — and 90B ...

Meta’s Llama 3.2 has been developed to redefined how large language models (LLMs) interact with visual data. By introducing a groundbreaking architecture that seamlessly integrates image ...

Learn More Today at its annual Meta Connect developer conference, Meta launched Llama Stack distributions, a comprehensive suite of tools designed to simplify AI deployment across a wide range of ...

Earlier this week Meta unveiled Llama 3.2, a major advancement in artificial intelligence (AI) designed for edge devices. This release brings enhanced performance and introduces models capable of ...

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn ...