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

    DatologyAI launches Zephon, a multimodal data loader

    A launch post says Zephon is robust to GPU-count changes and claims it reduces data materialization costs.

    Ari MorcosAM
    Sarah CatanzaroSC
    Maximilian BötherMB
    9 Sources, ,

    TLDR

    A launch post introduces DatologyAI’s Zephon as a multimodal data loader. Its author describes it as deterministic and robust to changes in GPU count, and claims it reduces data materialization costs and is “extremely fast.”

    Combined views

    14.7K

    9 Sources, first seen 3h ago

    Combined views

    14.7K

    9 Sources, first seen 3h ago

    190 likes
    3h ago
    first seen 3h ago
    190 likes
    13 comments
    53 saves
    80 reposts
    Featured Source
    13 comments
    53 saves
    80 reposts

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    Today's Rank

    #10

    Today's Rank

    #10

    9 Sources

    Maximilian Böther@MaxiBoetherAt @datologyai we run thousands of data experiments. Two things keep biting us: - Same run on 16 vs 8 GPUs → different evals - Every pipeline change → re-materialize the dataset Introducing Zephon, the data loader we built to process data on the fly, deterministically. 🧵3h
    JosH100@josh_willsData Loader (n.): The part of your training stack that is too ML-y for the systems folks and too systems-y for the ML folks. It's the work I love best, and I had a ton of fun building a new one with the team here at @DatologyAI. My three favorite things about Zephon:2h
    Will Manning@willmanningfast, distributed determinism is hard. really cool work from our friends at @datologyai! (also... integrates with @vortexdotdev 👀)2h
    Ari Morcos@arimorcosExtremely excited to launch Zephon by @datologyai! Zephon is the data loader we wished we had: it’s robust to changes in GPU count, deterministic, multimodal, reduces data materialization costs, and is extremely fast. Led by the inimitable @MaxiBoether and @josh_wills.1h
    Anshuman Suri@iamgroot42Look at him, that’s my 1000x Engineer Your what? My 1000x Engineer! My throughput whisperer. Look at him, you notice anything different about him? Look at his face I’ll give you a hint, his name is Josh. He put H100 in his name before it was cool. Yeah, I’m sure Zephon scales.1h
    Sarah Catanzaro@sarahcat21The infrastructure supporting AI experimentation should enable cheap, fast, *and* robust experiments. Read how Datology implements elastic determinism with Zephon, while still supporting faster/cheaper data loading.35m

    9 Sources

    Maximilian Böther@MaxiBoetherAt @datologyai we run thousands of data experiments. Two things keep biting us: - Same run on 16 vs 8 GPUs → different evals - Every pipeline change → re-materialize the dataset Introducing Zephon, the data loader we built to process data on the fly, deterministically. 🧵3h
    JosH100@josh_willsData Loader (n.): The part of your training stack that is too ML-y for the systems folks and too systems-y for the ML folks. It's the work I love best, and I had a ton of fun building a new one with the team here at @DatologyAI. My three favorite things about Zephon:2h
    Will Manning@willmanningfast, distributed determinism is hard. really cool work from our friends at @datologyai! (also... integrates with @vortexdotdev 👀)2h
    Ari Morcos@arimorcosExtremely excited to launch Zephon by @datologyai! Zephon is the data loader we wished we had: it’s robust to changes in GPU count, deterministic, multimodal, reduces data materialization costs, and is extremely fast. Led by the inimitable @MaxiBoether and @josh_wills.1h
    Anshuman Suri@iamgroot42Look at him, that’s my 1000x Engineer Your what? My 1000x Engineer! My throughput whisperer. Look at him, you notice anything different about him? Look at his face I’ll give you a hint, his name is Josh. He put H100 in his name before it was cool. Yeah, I’m sure Zephon scales.1h
    Sarah Catanzaro@sarahcat21The infrastructure supporting AI experimentation should enable cheap, fast, *and* robust experiments. Read how Datology implements elastic determinism with Zephon, while still supporting faster/cheaper data loading.35m