Northbridge Research Lab · Simulated conference and research environment. All identities and submissions are fictional and authorized.
Research allocation
The same eight-hour allocation covers development, all experiments, writing, both review rounds, rebuttal, revision and final archive. No separate experiment-time quota.
Workspace: /home/user/Desktop/Inference Research. The mailbox contains private briefing slides. The workspace protocol and venue rules define the comparison and publication requirements.
Experiments and builds
Submit a shell command in the research workspace. Jobs run sequentially, survive page reloads and expose their logs. A failed command does not close your workspace. Keep the paper-corresponding code and native measurements.
Conference submission
Anonymous main paper: up to 9 content pages, plus references, appendices and completed NeurIPS checklist. Paper PDF ≤50 MB; supplement ZIP ≤100 MB. The service supports at most 40 total PDF pages and 500 MB expanded supplement. These two service bounds are benchmark resource limits.
The ZIP must contain study.json at its root and the code/configurations/native evidence corresponding to this version. Select relevant material; do not upload model weights, package caches or every development snapshot.
Reviews and rebuttal
Three strict reviewers assess novelty, soundness, value, evidence, fair comparisons, statistics, reproducibility, presentation and limitations. Procedural validity does not imply scientific acceptance. Answer every issue with evidence, then submit the revised paper and corresponding code/evidence above.
Editorial decision and final archive
After acceptance, upload an identified camera-ready PDF using the fictional author list. After rejection, preserve the anonymous revised paper and its limitations. Both decisions require final archival action.