Wing Lian f34eef546a update doc and use P2P=LOC for brittle grpo test (#2649) ...

* update doc and skip brittle grpo test

* fix the path to run the multigpu tests

* increase timeout, use LOC instead of NVL

* typo

* use hf cache from s3 backed cloudfront

* mark grpo as flaky test dues to vllm start

2025-05-12 14:17:25 -04:00

..

__init__.py

Various fixes for CI, save_only_model for RL, prevent packing multiprocessing deadlocks (#2661)

2025-05-12 10:51:18 -04:00

cicd.sh

Various fixes for CI, save_only_model for RL, prevent packing multiprocessing deadlocks (#2661)

2025-05-12 10:51:18 -04:00

cleanup.py

Various fixes for CI, save_only_model for RL, prevent packing multiprocessing deadlocks (#2661)

2025-05-12 10:51:18 -04:00

cleanup.sh

Various fixes for CI, save_only_model for RL, prevent packing multiprocessing deadlocks (#2661)

2025-05-12 10:51:18 -04:00

Dockerfile.jinja

Sequence parallelism (#2412)

2025-03-21 12:43:55 -04:00

e2e_tests.py

update doc and use P2P=LOC for brittle grpo test (#2649)

2025-05-12 14:17:25 -04:00

multigpu.py

Codecov fixes / improvements (#2549)

2025-04-23 10:33:30 -04:00

multigpu.sh

update trl to 0.17.0 (#2560)

2025-04-27 19:19:53 -04:00

single_gpu.py

Various fixes for CI, save_only_model for RL, prevent packing multiprocessing deadlocks (#2661)

2025-05-12 10:51:18 -04:00