Recent literature and method comparison
Keep search queries, source locations and disagreements; state which full texts were actually read.
Computer science & AI
Compare methods, review code, test small CPU prototypes and examine evaluation reproducibility.
Start with a focused task
Algorithm and evaluation design
State the question, assumptions, design and comparison conditions; flag choices needing researcher review.
Dataset, code and leakage checks
Check variables, units, missing values, duplicates and provenance; record changes without silent imputation or deletion.
CPU prototypes and metric calculation
Check material and method suitability first; execute required calculations and deliver code, results and limitations.
Baselines, ablations and reproduction checks
Use independent checks, alternative explanations or applicable sensitivity checks; report failure conditions and unresolved issues.
Experiment figures and error analysis
Create figures and tables from checked results with units, captions and sources; retain reusable code.
Algorithm papers and technical reports
Draft from actual materials and results, check numbers and citations, and identify drafts and unknowns.
Code review, revision and presentations
Deliver issue lists, point-by-point responses or presentation material; distinguish corrections, missing evidence and unresolved issues.
Current scope
arXiv peer-review status is unverified. Reading papers or code does not run training or reproduce experiments. Key conclusions need human review.
Research examples
Browse research outputs and find a starting point for your task.