This repository provides the necessary instructions to reproduce a specific workload on Google Cloud TPUs. The focus is on reliably achieving a performance metric (e.g. throughput) that demonstrates the combined hardware and software stack on TPUs.
./training: instructions to reproduce the training performance of popular LLMs, diffusion, and other models with PyTorch and JAX.
./inference: instructions to reproduce inference performance.
./microbenchmarks: instructions for low-level TPU benchmarks such as matrix multiplication performance and memory bandwidth.
Note: This is not an officially supported Google product. This project is not eligible for the Google Open Source Software Vulnerability Rewards Program.
AboutNo description, website, or topics provided.
Resources License Contributing Stars Watchers ForksYou can’t perform that action at this time.
RetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.4