Wing Lian 598c965043 use train_loss for sp test

2026-03-22 12:00:55 -04:00

..

patched

use train_loss for sp test

2026-03-22 12:00:55 -04:00

solo

fix flaky tests; should be using train loss from final step rather than final avg train loss

2026-03-22 10:38:46 -04:00

__init__.py

Attempt to run multigpu in PR CI for now to ensure it works (#1815) [skip ci]

2024-08-09 11:50:13 -04:00

test_dist_muon_fsdp2.py

fix flaky tests; should be using train loss from final step rather than final avg train loss

2026-03-22 10:38:46 -04:00

test_eval.py

Add ruff, remove black, isort, flake8, pylint (#3092)

2025-08-23 23:37:33 -04:00

test_fp8_fsdp2.py

fix flaky tests; should be using train loss from final step rather than final avg train loss

2026-03-22 10:38:46 -04:00

test_fsdp1.py

fix flaky tests; should be using train loss from final step rather than final avg train loss

2026-03-22 10:38:46 -04:00

test_fsdp2.py

fix flaky tests; should be using train loss from final step rather than final avg train loss

2026-03-22 10:38:46 -04:00

test_gemma3.py

fix flaky tests; should be using train loss from final step rather than final avg train loss

2026-03-22 10:38:46 -04:00

test_llama.py

fix flaky tests; should be using train loss from final step rather than final avg train loss

2026-03-22 10:38:46 -04:00

test_locking.py

Data loader refactor (#2707)

2025-06-10 19:53:07 -04:00

test_ray.py

fix flaky tests; should be using train loss from final step rather than final avg train loss

2026-03-22 10:38:46 -04:00

test_tp.py

fix flaky tests; should be using train loss from final step rather than final avg train loss

2026-03-22 10:38:46 -04:00