N
Nyxis
Prompts / Eval

DatasetTable

Stacked table for an eval dataset — input / expected / actual / score with bucket filters.

Preview

Dataset

6 / 6 rows

Installation

pnpm add nyxis-ui

Usage

import { DatasetTable, type EvalRow } from 'nyxis-ui';

const rows: EvalRow[] = [
  {
    id: '1',
    input: 'PR #42 — refactor: split chat barrel',
    expected: '- Split chat barrel\n- Per-component subpaths',
    actual: '- Split chat barrel\n- Per-component subpaths',
    score: 0.94,
    status: 'pass',
    latencyMs: 420,
    costUsd: 0.0023,
  },
  // ...
];

<DatasetTable rows={rows} />;

Anatomy

  • Header: title + filtered/total row count + bucket filter pills (low / mid / high with counts).
  • Search input: matches input, expected, actual, and notes.
  • Each row shows input / expected / actual columns truncated to 2 lines.
  • Click a row to expand the full text plus latency, cost, and notes.
  • Score bucket thresholds default to [0.5, 0.8] and are configurable via scoreBuckets.