task_charts__dumbbell__gap_rank_row_label

Contract

1. Domain: charts
2. Scene id: dumbbell
3. Source implementation domain/group: charts/dumbbell
4. Query ids: largest_gap_rank_row_label, smallest_gap_rank_row_label
5. Semantic query details are recorded in query_id and trace params.

Implementation

1. Registered class: trace_tasks.tasks.charts.dumbbell.gap_rank_row_label.ChartsDumbbellGapRankRowLabelTask
2. Prompt lookup domain/group: charts/dumbbell
3. Generation is deterministic from instance_seed, explicit params, prompt bundle, renderer config, and code versions.
4. Answers and annotation are produced from the same metadata execution trace.

Annotation Contract

1. Answer schema: string_label.
2. Annotation schema: segment.
3. Annotation is one segment: two [x, y] pixel points formatted [[x0, y0], [x1, y1]], connecting the two colored dot centers for the selected dumbbell row.
4. Renderer context such as legends, axes, decorative labels, titles, and distractor text is metadata unless the task explicitly asks for it as annotation.

Program Contract

Program: arg_extreme(row_label, abs(value(series_a,row_label)-value(series_b,row_label)), direction={largest,smallest}, rank={1,2}); output=string_label; annotation=segment(answer_row_dot_centers); scene=dumbbell; scope=gap_rank_row_label

Candidate set: the visible paired endpoint markers, connectors, and category labels inside the gap_rank_row_label objective scope. Operands: prompt-bound labels, categories, series names, thresholds, intervals, references, and encoded chart values, plus the active query id's comparator, direction, target role, or extremum focus when present. Operation: evaluate arg_extreme over the candidate set using the filters, comparisons, aggregations, rankings, projections, or counterfactual edits named in the program expression; generation enforces a unique final answer. Output binding: answer is the string_label value bound by string_label. Annotation witnesses: segment witnesses bound by segment(answer_row_dot_centers). Annotation is one segment: two [x, y] pixel points formatted [[x0, y0], [x1, y1]], connecting the two colored dot centers for the selected dumbbell row. Renderer context such as legends, axes, decorative labels, titles, and distractor text is metadata unless the task explicitly asks for it as annotation. Query ids: largest_gap_rank_row_label, smallest_gap_rank_row_label.

Reasoning Operations

Families: ranking, formula_evaluation

Query Details

Query id	Program signature	Answer schema	Annotation schema
largest_gap_rank_row_label	selection.gap_rank_row_label	string_label	segment
smallest_gap_rank_row_label	selection.gap_rank_row_label	string_label	segment

Rank position is task-internal metadata and is sampled only from 1 and 2, yielding largest/second-largest or smallest/second-smallest prompts.