• Home
  • Technology
  • Gaming
  • Entertainment
  • World & Business
  • Science
  • Sports
  • AI
HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
    • HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
    • Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI
    AI

    Account Views NVIDIA Criticism as Disappointment Like Europe

    X reply from AI observer draws parallel between SA comments and personal feelings toward Europe.

    T(
    SE
    XL
    4 Sources, 28d ago, first seen 28d ago

    TLDR

    An anonymous account focused on AI, effective accelerationism, and open-source model releases replied to discussion of NVIDIA criticism. The post states that SA's comments read as disappointment in the company for not performing better, much like the author's view of Europe. It questions whether that comparison fully explains the tone yet concludes the remarks still seem to extend into personal and specific territory beyond the analogy. The reply appears among visible responses on X addressing the original NVIDIA remarks. The account is known for fine-tuning and sharing models.

    Combined views

    126.3K

    4 Sources, first seen 28d ago

    Combined views

    126.3K

    4 Sources, first seen 28d ago

    502 likes
    502 likes
    49 comments
    108 saves
    54 reposts

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    49 comments
    108 saves
    54 reposts
    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    4 Sources

    @SemiAnalysis_NVIDIA Research 🚀 has produced some great research, like LatentMoE (used in Kimi K3) and GatedDeltaNets (used in Qwen). But for e2e frontier training, NVIDIA's bureaucratic culture has produced embarrassing models like Nemotron3 Ultra. Despite NVIDIA Research having amazing talent, Nemotron3 Ultra, with 550B total params (55B active), is getting mogged by all the Chinese models, including even Qwen3.8 27B parameters, which has ~20x fewer parameters.
    @teortaxesTexCan anyone explain to me why SA is going after Nemotron 3 Ultra with such dogged vindictiveness? Nobody cares about Nemotron3. Nobody thinks Nvidia has a frontier LLM team. Architecture R&D output tends to be anticorrelated with model quality, if anything.
    @xlr8harder@teortaxesTex I always read SA's NVIDIA criticism as a bit like how I feel about Europe, sort of disappointed in them for not being better, so calling them out on it? I don't know though, it does still seem to go beyond that and get quite personal and specific.
    @wzenusRT @SemiAnalysis_: NVIDIA Research 🚀 has produced some great research, like LatentMoE (used in Kimi K3) and GatedDeltaNets (used in Qwen).…

    4 Sources

    @SemiAnalysis_NVIDIA Research 🚀 has produced some great research, like LatentMoE (used in Kimi K3) and GatedDeltaNets (used in Qwen). But for e2e frontier training, NVIDIA's bureaucratic culture has produced embarrassing models like Nemotron3 Ultra. Despite NVIDIA Research having amazing talent, Nemotron3 Ultra, with 550B total params (55B active), is getting mogged by all the Chinese models, including even Qwen3.8 27B parameters, which has ~20x fewer parameters.
    @teortaxesTexCan anyone explain to me why SA is going after Nemotron 3 Ultra with such dogged vindictiveness? Nobody cares about Nemotron3. Nobody thinks Nvidia has a frontier LLM team. Architecture R&D output tends to be anticorrelated with model quality, if anything.
    @xlr8harder@teortaxesTex I always read SA's NVIDIA criticism as a bit like how I feel about Europe, sort of disappointed in them for not being better, so calling them out on it? I don't know though, it does still seem to go beyond that and get quite personal and specific.
    @wzenusRT @SemiAnalysis_: NVIDIA Research 🚀 has produced some great research, like LatentMoE (used in Kimi K3) and GatedDeltaNets (used in Qwen).…