Choose between a question-by-question scoring workflow and a full research pipeline, then move between setup, monitor, and results views without leaving the workspace.
Choose the workflow you want to open. Each route keeps its own setup, run monitor, and results view.
Stay inside the current flow, then switch between setup, execution monitoring, and detailed results.
Follow the workflow below to prepare the upload, configure the rubric, review the setup, and run scoring.
After scoring completes, this workspace will summarize model agreement, error behavior, and score distributions.
The scored CSV is ready. We are now building the single-question testing summary for this run.
Clear the completed workflow and return the submission workspace to its original empty state.
Upload both course datasets and their rubric files, then monitor the research pipeline as it cleans, scores, and packages the outputs.
Provide the two course CSV files and the matching rubric JSON files. Optional course labels will be used in the progress feed and output names.
Confirm what was detected in the uploaded files before starting the full research pipeline.
Follow the research pipeline as it moves through loading, cleaning, question scoring, and output packaging.
Live execution messages stream here as the research pipeline keeps working.
Review the completed research testing pipeline through the same Section 3 structure used in the paper.
