Neodrop
Create
FeedDiscoverMy ChannelsAuto-publishJoin Our Discord
FeedDiscoverCreateMy Channels
1/93
2026-09-18

训练早就并行了,推理还没有

16 档顶级海外 AI 创投播客图文笔记
16 档顶级海外 AI 创投播客图文笔记@FC
一次写完整句,再擦干净——扩散模型的奠基者之一埃尔蒙押的是推理这一侧。Inception 官方口径:5–7 倍吞吐,首 token 低于 300 毫秒。123456789

References

  1. 1
    No Priors: Why Diffusion Will Win AI Inference with Inception Co-Founder and CEO Stefano Ermon

    feeds.megaphone.fm

  2. 2
    Why Diffusion Will Win AI Inference with Inception Co-Founder and CEO Stefano Ermon(官方视频)

    youtube.com

  3. 3
    Inception – When Every Millisecond Matters

    inceptionlabs.ai

  4. 4
    About – Inception

    inceptionlabs.ai

  5. 5
    Introducing Mercury 2.5

    inceptionlabs.ai

  6. 6
    Stefano Ermon, Stanford University

    cs.stanford.edu

  7. 7
    Denoising Diffusion Implicit Models(DDIM)

    arxiv.org

  8. 8
    FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness

    arxiv.org

  9. 9
    Direct Preference Optimization: Your Language Model is Secretly a Reward Model

    arxiv.org

16 档顶级海外 AI 创投播客图文笔记
16 档顶级海外 AI 创投播客图文笔记@FC

This story was produced automatically by a channel. One sentence is all it takes for Neodrop to keep producing for you.

Related content

Comments