test(train): bump RLVR seq-length integ timeout to 3h - #1
Closed
guanweim wants to merge 1 commit into
Closed
Conversation
GRPO RLVR on a 9B model (qwen3-5-9b) at 8K sequence length is far more compute-heavy than the 1B llama-based RLVR tests that share the 1h timeout. The job submits and runs correctly but does not reach Completed within 3600s, so the test asserted 'InProgress' == 'Completed' and failed. Raise max_wait_time to 10800 (3h), matching the budget already used by the Nova RLVR test in this file for a large model.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Raise
max_wait_timefrom 3600 (1h) to 10800 (3h) intest_rlvr_trainer_lora_with_sequence_length.Why
GRPO RLVR on a 9B model (
huggingface-vlm-qwen3-5-9b) at 8K sequence length is far heavier than the 1B llama-based RLVR tests in this file that share the 1h timeout. In CI the job submitted and ran correctly (recipe resolved, reward fn verified 3/3, training job created) but had not reachedCompletedwithin 3600s, so the test failed onassert 'InProgress' == 'Completed'— a timeout, not a real training failure.3h matches the budget already used by
test_rlvr_trainer_nova_workflowfor a large model in the same file.Note
Stacked on aws#5965 (which introduces this test); base is
context_length.