Fault Tolerant Llama training

(pytorch.org)

52 points | by Mougatine 4 days ago ago

8 comments

  • d4l3k 12 hours ago ago

    Hey, nice to see this here!

    I'm the primary author so happy to answer any questions you might have!

  • zxexz 6 hours ago ago

    This is awesome, can’t wait to try out these techniques. At least a week a year of my time for the past few years has gone towards recovering from a fault crashing a training run. Sometimes environment related, sometimes shared storage, sometimes just because a slightly faulty IB cable.

  • bjt12345 10 hours ago ago

    This is severely underrated work, why aren't there more mid sized companies helping this? Ultra Ethernet just got released.

    • foobiekr 6 hours ago ago

      Ultra Ethernet will do almost nothing. It’s a rubber stamped version of Broadcom’s design and Marcel/Cisco/etc will just add it to their asics. Remains to be seen if SpecrumX will or Connectix. If not, none of it matters.

      These chips are $30m-$100m projects a pop. After the embarrassingly brutal failure of Barefoot nobody is going to do ASICs.

  • anonymousDan 3 hours ago ago

    What kind of failures are you typically concerned with here?

  • timzaman 12 hours ago ago

    300 L40s? What's this, 1998?

    • d4l3k 10 hours ago ago

      Hey Tim, how's it going?

      Interested in lending PyTorch some compute? :)

      torchft can handle much larger scales but for public multi-day demonstration run this is what we had available. Point of this blog was to demonstrate correctness of the quorum algorithm and recovery with a stock PyTorch stack and not so much peak flops.

      Stay tuned though -- planning on doing some much larger demos on B200s!

    • kcorbitt 11 hours ago ago

      I was curious about this so I had o3 do a bit of research. Turns out 300 L40s have more compute than any supercomputer before 2013 (and arguably before 2016, depending on how you count reduced-precision FLOPs).

      https://chatgpt.com/share/685dea79-26ec-8002-bd62-7ed83aedf4...