Skip to content

Conversation

terrykong
Copy link
Contributor

@terrykong terrykong commented Aug 17, 2025

the 16k and 24k tests run in the CI with 8k and 16k staged checkpoints, but external users can still run these with the base checkpoint since the metric check always makes sure it doesn't regress against the base checkpoint

tools/launch tests/test_suites/llm/grpo-deepscaler-1.5b-8K.sh

EXTRA_ENV="NRL_DEEPSCALER_8K_CKPT=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B" \
  tools/launch tests/test_suites/llm/grpo-deepscaler-1.5b-16K.sh

EXTRA_ENV="NRL_DEEPSCALER_16K_CKPT=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B" \
  tools/launch tests/test_suites/llm/grpo-deepscaler-1.5b-24K.sh

Signed-off-by: Terry Kong <terryk@nvidia.com>

updated recipes + hacks

Signed-off-by: Terry Kong <terryk@nvidia.com>

better for memory

Signed-off-by: Terry Kong <terryk@nvidia.com>

adjust step time

Signed-off-by: Terry Kong <terryk@nvidia.com>

fix steps

Signed-off-by: Terry Kong <terryk@nvidia.com>

disable eval this round

Signed-off-by: Terry Kong <terryk@nvidia.com>

comment

Signed-off-by: Terry Kong <terryk@nvidia.com>

lint

Signed-off-by: Terry Kong <terryk@nvidia.com>

revert

Signed-off-by: Terry Kong <terryk@nvidia.com>

fix some config issues uncovered by placing configs in the correct dir

Signed-off-by: Terry Kong <terryk@nvidia.com>

chore: enable accuracy test for deepscaler + enable 16k and 24k tests

Signed-off-by: Terry Kong <terryk@nvidia.com>

get recipes right

Signed-off-by: Terry Kong <terryk@nvidia.com>
Signed-off-by: Terry Kong <terryk@nvidia.com>
@github-actions github-actions bot added the documentation Improvements or additions to documentation label Aug 17, 2025
Signed-off-by: Terry Kong <terryk@nvidia.com>
@terrykong terrykong changed the title tests: enable 8k/16k/24k deepscaler nightly tests test: enable 8k/16k/24k deepscaler nightly tests Aug 17, 2025
Signed-off-by: Terry Kong <terryk@nvidia.com>
Signed-off-by: Terry Kong <terryk@nvidia.com>
Signed-off-by: Terry Kong <terryk@nvidia.com>
Copy link
Contributor

@yuki-97 yuki-97 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

thanks! LGTM except the below two questions.

Signed-off-by: Terry Kong <terryk@nvidia.com>
Signed-off-by: Terry Kong <terryk@nvidia.com>
@terrykong terrykong linked an issue Aug 18, 2025 that may be closed by this pull request
@terrykong terrykong enabled auto-merge August 19, 2025 16:20
@terrykong terrykong added this pull request to the merge queue Aug 19, 2025
Merged via the queue into main with commit d149a62 Aug 19, 2025
19 checks passed
@terrykong terrykong deleted the tk/remaining-deepscaler branch August 19, 2025 20:35
jveronvialard pushed a commit that referenced this pull request Aug 27, 2025
Signed-off-by: Terry Kong <terryk@nvidia.com>
Signed-off-by: Julien Veron Vialard <jveronvialar@nvidia.com>
soodoshll pushed a commit to soodoshll/RL that referenced this pull request Sep 4, 2025
Signed-off-by: Terry Kong <terryk@nvidia.com>
Signed-off-by: Qidong Su <qidongs@nvidia.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
documentation Improvements or additions to documentation
Projects
None yet
Development

Successfully merging this pull request may close these issues.

Enable Deepscaler Aimee check in nightly tests
2 participants