AI 導讀
DeepInfer 是 DeepSeek 釋出的全新推理框架，已在 V4.1‑Flash 上執行，提供高吞吐量與低延遲。
- 高效能：在 H800 上達 580 TFLOPS 注意力運算，DeepGEMM 1350+ FP8 TFLOPS 矩陣乘法。
推薦理由
DeepSeek 釋出高效推理框架 DeepInfer，助使用者以更低成本、低延遲部署大模型，降低 AI 社群算力與能源門檻。
T
Thom Wolf (@Thom_Wolf)@Thom_Wolf
·2 小時前
DeepSeek's open-source streak is getting absurd: - FlashMLA: 580 TFLOPS attention on H800 - DeepEP: MoE comms - DeepGEMM: 1350+ FP8 TFLOPS matmul - 3FS: they built an entire file system - and today: the full inference engine behind V4.1-Flash, Apache 2.0 Mad respect
🚀 Introducing DeepInfer, our in-house inference framework, now fully open source! Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more
推文來源 · X (Twitter)在 X 參與討論
來源： Thom Wolf (@Thom_Wolf) (x.com)
在 X 查看推文