Skip to content

Add Cua-S1 multimodal stage profiling and RTX 4090 results - #15

Closed
Levius-Fubuki wants to merge 13 commits into
ThinkFlowLab:mainfrom
Levius-Fubuki:codex/cua-multimodal-profile
Closed

Levius-Fubuki wants to merge 13 commits into
ThinkFlowLab:mainfrom
Levius-Fubuki:codex/cua-multimodal-profile

Conversation

@Levius-Fubuki

@Levius-Fubuki Levius-Fubuki commented Sep 28, 2026 •

Copy link
Copy Markdown
Collaborator

Closed after runtime-only cleanup

This PR contributed only profiling/benchmark tooling and documentation. Those files have been removed from the submitted scope; its current runtime tree is identical to #12, so it no longer has an independent implementation change.

The branch and commit history are retained. Original tools, tests and documentation remain available at 138c4d11fe943e9fb4b559f6a1bbe12a27b98698. No runtime code was changed by this cleanup.

@hsliuustc0106 hsliuustc0106 left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

Reviewed original snapshot 4c605e3 and the follow-up diff through 04acf99b35d263abf8ed6aefe744a472fc83c7b9. No new actionable findings.

The follow-up fixes the inherited extreme-aspect-ratio validation and lock-cleanup test race from #12. Reviewed the protocol/test changes and documentation updates. Protocol and HTTP tests on this head: 61 passed in 14.82s. No accelerator execution was performed.

Validation and scope of the original snapshot review:

No new actionable findings in profiling delta. Instrumentation is scoped and restored; baseline timing precedes profiling. CPU profiling tests: 25 passed. GPU results not rerun.

Remove archived experiment outputs and assistant planning notes. Keep runtime code, reproducible tools, and model verification metadata unchanged. Link historical reports to an immutable commit and ignore future local artifacts. Remove only tests tied to the deleted historical evidence bundles.
Remove archived experiment outputs and assistant planning notes. Keep runtime code, reproducible tools, and model verification metadata unchanged. Link historical reports to an immutable commit and ignore future local artifacts. Remove only tests tied to the deleted historical evidence bundles.
Remove supplemental tests, recipe tools, documentation, CI, and configuration changes from the PR diff. Runtime Python source and inline third-party notice are unchanged. Validation uses the pre-cleanup test/tool snapshot outside the checkout.
Remove supplemental tests, recipe tools, documentation, CI, and configuration changes from the PR diff. Runtime Python source and inline third-party notice are unchanged. Validation uses the pre-cleanup test/tool snapshot outside the checkout.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants