Hacker GitHub

← All repos

AI-Hypercomputer/maxtext

View on GitHub ↗ ★ 2,455 Python

A simple, performant, and scalable Jax LLM!

9 mentions on Hacker News by 8 people: 2 stories, 7 comments · first 2023-04-20 · latest 2025-05-15

Mentions per year 9 mentions · busiest year 2024 (6)
20082026
AllStoriesComments NewestTopOldest
  1. …but rather this repo contains various attention implementations in MaxText/layers/attentions.py [0] https://github.com/AI-Hypercomputer/maxtext
    comment by yarri · 2025-05-15 · in AlphaEvolve: A Gemini-powered coding agent for designing advanced algorithms
  2. …weights and pretrain on additional data. It looks like the instructions for doing that are here: https://github.com/AI-Hypercomputer/maxtext/blob/main/gettin...
    comment by lern_too_spel · 2024-12-12 · in Trillium TPU Is GA
  3. Did you consider using https://github.com/AI-Hypercomputer/maxtext ? It has a Jax llama implementation, and gets decent MFU on TPU and GPU (I've only tried it on NVidia GPU, not AMD).
    comment by logicchains · 2024-09-24 · in We fine-tuned Llama 405B on AMD GPUs
  4. …going into detail: https://www.youtube.com/watch?v=5QsM1K9ahtw https://github.com/google/JetStream https://github.com/google/maxtext
    comment by kiratp · 2024-06-14 · in Cost of self hosting Llama-3 8B-Instruct
  5. This is absolutely the case, TPUs scale very well: https://github.com/google/maxtext .
    comment by logicchains · 2024-06-13 · in How Meta trains large language models at scale
  6. …too, but they've made great progress on caching JIT compiled functions to reduce latency.) [1]: https://github.com/google/maxtext?tab=readme-ov-file#ahead-o...
    comment by ubj · 2024-04-24 · in Maxtext: A simple, performant and scalable Jax LLM
  7. ▲ 115
    Maxtext: A simple, performant and scalable Jax LLM
    story by zerojames · 2024-04-24 · 7 comments
  8. …and it runs on pods last time I tried) and see how it performs in practice. EDIT: reading over https://github.com/google/maxtext , it looks pretty delightful. I was in the TPU scene back in 2020, and there was no way to do ahead of time…
    comment by sillysaurusx · 2023-11-11 · in Google Cloud TPU Multislice Training
  9. ▲ 2
    Google’s maxtext – A simple, performant and scalable Jax LLM
    story by tim_sw · 2023-04-20 · 2 comments