Skip to content

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

License

Notifications You must be signed in to change notification settings

vivianrwu/JetStream

 
 

Repository files navigation

Unit Tests PyPI version PyPi downloads Contributions welcome

JetStream is a throughput and memory optimized engine for LLM inference on XLA devices.

About

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

JetStream Engine Implementation

Currently, there are two reference engine implementations available -- one for Jax models and another for Pytorch models.

Jax

Pytorch

Documentation

JetStream Standalone Local Setup

Getting Started

Setup

make install-deps

Run local server & Testing

Use the following commands to run a server locally:

# Start a server
python -m jetstream.core.implementations.mock.server

# Test local mock server
python -m jetstream.tools.requester

# Load test local mock server
python -m jetstream.tools.load_tester

Test core modules

# Test JetStream core orchestrator
python -m unittest -v jetstream.tests.core.test_orchestrator

# Test JetStream core server library
python -m unittest -v jetstream.tests.core.test_server

# Test mock JetStream engine implementation
python -m unittest -v jetstream.tests.engine.test_mock_engine

# Test mock JetStream token utils
python -m unittest -v jetstream.tests.engine.test_token_utils
python -m unittest -v jetstream.tests.engine.test_utils

About

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

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 95.4%
  • Shell 3.2%
  • Other 1.4%