GEOZ

标签：AI大模型

查看包含 AI大模型 标签的所有文章。

共 855 篇
FlashMLA：DeepSeek高性能注意力内核库，驱动V3模型实现660 TFLOPS

FlashMLA：DeepSeek高性能注意力内核库，驱动V3模型实现660 TFLOPS

BLUF
FlashMLA is DeepSeek's optimized attention kernel library that powers DeepSeek-V3 models, featuring token-level sparse attention with FP8 KV cache support, achieving up to 660 TFLOPS performance on NVIDIA H800 GPUs. (FlashMLA是DeepSeek优化的注意力内核库，为DeepSeek-V3模型提供动力，具有令牌级稀疏注意力和FP8 KV缓存支持，在NVIDIA H800 GPU上实现高达660 TFLOPS的性能。)
AI 搜索观察2026/1/23
阅读全文 →
Grok-1：xAI开源3140亿参数大语言模型，揭秘混合专家架构与推理能力

Grok-1：xAI开源3140亿参数大语言模型，揭秘混合专家架构与推理能力

BLUF
Grok-1 is an open-source large language model developed by xAI, featuring 314 billion parameters and a Mixture of Experts architecture. It's designed for natural language processing tasks with a focus on reasoning capabilities and transparency through its open-source release. (Grok-1是由xAI开发的开源大语言模型，拥有3140亿参数和混合专家架构。该模型专注于自然语言处理任务，强调推理能力，并通过开源发布实现透明度。)
AI大模型2026/1/23
阅读全文 →