REMORSEHow can we help?
Find CSV requirements and report explanations below, or contact us in English or Korean.
Report an issue
Bugs and feature requests for Remorsearch. Include the steps, the expected result and the actual result.
GitHub Issues →Report a security problem
Do not open a public issue. Email us with the subject line “Security”. We will confirm receipt.
[email protected]Frequently asked questions
What is Sublate Evaluate?
A free browser tool that compares paired baseline and candidate measurements by instance ID. It uses your objective, runtime and validity fields; it does not run a solver.
What does a CSV need?
Use the downloadable template. Every row needs a unique instance ID, development/holdout/stress split, baseline and candidate objectives, non-negative runtimes in milliseconds and true/false validity flags. Files are limited to 1 MiB and 2,000 data rows.
Why is a comparison blocked?
Reported invalid baseline or candidate results, runtime violations, or regressions beyond your selected allowance block the comparison. The report identifies the affected instances.
Why is a comparison inconclusive?
A missing holdout or stress split leaves the evidence incomplete. Add those measurements to review the candidate against all required splits.
Does my evaluation file leave the browser?
CSV contents stay in memory in this browser. Evaluate uses no account, server upload, browser storage or external model call. Reset clears the loaded data.
How do I request a paid review?
Prepare a local brief at /pilot/ or email [email protected]. A feasibility review starts at KRW 800,000 for up to 6 delivery hours. A one-solver/problem pilot starts at KRW 3,500,000 total including the review, up to 40 delivery hours. Scope, taxes, compute budget and schedule are confirmed in a written quote. Improvement is not guaranteed.
Can I run a local solver project?
Download the Sublate pilot kit from /pilot/. Connect a local Python adapter, baseline, scorer and instances, then use doctor to review missing setup and permissions. The macOS runner uses the Python standard library. Execution and the optional paid Anthropic proposal connector require separate local approvals. No live paid API call has been verified.
What is Remorsearch UX QA?
An open-source MIT Agent Skill for UX research and QA in your own agent host. See its product page and GitHub repository for installation and current requirements.
When can I use Maccept?
Maccept is in development. Its public release and intended end-to-end Siri interaction are not yet verified.
Do you support Korean?
Support is available in English and Korean.