Skip to content
@AI-Hypercomputer

AI Hypercomputer

Reference implementations, benchmarks, recipes, and all things Google Cloud AI Hypercomputer

Popular repositories Loading

  1. maxtext maxtext Public

    A simple, performant, and scalable Jax LLM!

    Python 2.5k 645

  2. JetStream JetStream Public

    JetStream is a throughput and memory optimized engine for LLM inference on XLA devices, starting with TPUs (and GPUs in future -- PRs welcome).

    Python 473 70

  3. maxdiffusion maxdiffusion Public

    Python 395 94

  4. RecML RecML Public

    Python 230 23

  5. xpk xpk Public

    xpk (Accelerated Processing Kit, pronounced x-p-k,) is a software tool to help Cloud developers to orchestrate training jobs on accelerators such as TPUs and GPUs on GKE.

    Python 196 96

  6. gpu-recipes gpu-recipes Public

    Recipes for reproducing training and serving benchmarks for large machine learning models using GPUs on Google Cloud.

    Python 151 94

Repositories

Showing 10 of 23 repositories
  • pathways-utils Public

    Package of Pathways-on-Cloud utilities

    AI-Hypercomputer/pathways-utils's past year of commit activity
    Python 39 Apache-2.0 14 0 28 Updated Nov 12, 2026
  • maxtext Public

    A simple, performant, and scalable Jax LLM!

    AI-Hypercomputer/maxtext's past year of commit activity
    Python 2,460 Apache-2.0 645 67 (1 issue needs help) 477 Updated Nov 12, 2026
  • maxdiffusion Public
    AI-Hypercomputer/maxdiffusion's past year of commit activity
    Python 395 Apache-2.0 94 11 22 Updated Nov 12, 2026
  • RecML Public
    AI-Hypercomputer/RecML's past year of commit activity
    Python 230 Apache-2.0 23 3 99 Updated Nov 12, 2026
  • AI-Hypercomputer/ml-goodput-measurement's past year of commit activity
    Python 43 Apache-2.0 10 2 3 Updated Nov 12, 2026
  • AI-Hypercomputer/accelerator-agents's past year of commit activity
    Python 103 Apache-2.0 24 1 28 Updated Nov 11, 2026
  • accelerator-microbenchmarks Public

    TPUMS (TPU Microbenchmark Suite) is a modular, high-fidelity benchmarking and profiling framework designed to evaluate the compute, memory, and interconnect performance of Cloud TPUs using JAX.

    AI-Hypercomputer/accelerator-microbenchmarks's past year of commit activity
    Python 40 Apache-2.0 44 4 8 Updated Nov 11, 2026
  • tpu-codesign Public

    Hardware-grounded model architecture design principles, MoE co-design patterns, low-precision numerics, and parallelism guides for training and serving on Google Cloud TPUs.

    AI-Hypercomputer/tpu-codesign's past year of commit activity
    0 Apache-2.0 0 0 0 Updated Nov 12, 2026
  • gpu-recipes Public

    Recipes for reproducing training and serving benchmarks for large machine learning models using GPUs on Google Cloud.

    AI-Hypercomputer/gpu-recipes's past year of commit activity
    Python 151 Apache-2.0 94 8 34 Updated Nov 10, 2026
  • AI-Hypercomputer/google-cloud-mldiagnostics's past year of commit activity
    Python 14 Apache-2.0 4 5 3 Updated Nov 5, 2026