@wavequery/conductor/dist/esm/evaluation/eval-reporter.js

Modular LLM orchestration framework

import { Logger, LogLevel } from "@/utils/logger";
export class EvalReporter {
    constructor() {
        this.logger = new Logger({
            level: LogLevel.DEBUG,
            prefix: "EvalReporter",
        });
    }
    generateReport(results, options = {
        format: "markdown",
        includeMetadata: true,
        includeTimestamp: true,
        includeSummary: true,
        includeRecommendations: true,
    }) {
        try {
            const sections = [
                options.includeSummary && this.generateSummary(results),
                this.generateDetailed(results, options),
                options.includeRecommendations && this.generateRecommendations(results),
            ].filter(Boolean);
            return sections.join("\n\n");
        }
        catch (error) {
            this.logger.error("Report generation failed:", error);
            throw error;
        }
    }
    async generateAsyncReport(results, options) {
        return new Promise((resolve, reject) => {
            try {
                const report = this.generateReport(results, options);
                resolve(report);
            }
            catch (error) {
                reject(error);
            }
        });
    }
    generateSummary(results) {
        const avgScore = results.reduce((sum, r) => sum + r.score, 0) / results.length;
        const timestamp = new Date().toISOString();
        const successfulMetrics = results.filter((r) => r.score >= 0.7).length;
        return `
        # Evaluation Summary
        - Overall Score: ${(avgScore * 100).toFixed(2)}%
        - Total Metrics: ${results.length}
        - Successful Metrics: ${successfulMetrics}
        - Failed Metrics: ${results.length - successfulMetrics}
        - Timestamp: ${timestamp}
    `.trim();
    }
    generateDetailed(results, options) {
        return `
          # Detailed Results

          ${results
            .map((r) => `
          ## ${r.metricName}
          - Score: ${(r.score * 100).toFixed(2)}%
          ${options.includeMetadata ? this.formatMetadata(r.metadata) : ""}
          ${options.includeTimestamp ? `- Timestamp: ${r.timestamp.toISOString()}` : ""}
          ${r.evaluator ? `- Evaluator: ${r.evaluator}` : ""}
          `)
            .join("\n")}
          `.trim();
    }
    formatMetadata(metadata) {
        return Object.entries(metadata)
            .map(([key, value]) => `- ${key}: ${JSON.stringify(value)}`)
            .join("\n");
    }
    generateRecommendations(results) {
        const lowScores = results.filter((r) => r.score < 0.7);
        if (lowScores.length === 0) {
            return "# Recommendations\nAll metrics are performing well! No immediate improvements needed.";
        }
        return `
          # Recommendations

          ${lowScores
            .map((r) => `
          - Improve ${r.metricName}
            - Current score: ${(r.score * 100).toFixed(2)}%
            - Action needed: ${this.getRecommendation(r)}
          `)
            .join("\n")}
          `.trim();
    }
    getRecommendation(result) {
        switch (result.metricName) {
            case "responseTime":
                return "Consider optimizing performance or increasing compute resources";
            case "tokenUsage":
                return "Review prompt engineering to reduce token consumption";
            case "accuracy":
                return "Review model configuration and input preprocessing";
            default:
                return "Review implementation and configuration";
        }
    }
    exportToJSON(results) {
        return JSON.stringify({
            summary: {
                averageScore: results.reduce((sum, r) => sum + r.score, 0) / results.length,
                totalMetrics: results.length,
                timestamp: new Date().toISOString(),
            },
            results: results.map((r) => ({
                ...r,
                timestamp: r.timestamp.toISOString(),
            })),
        }, null, 2);
    }
}
//# sourceMappingURL=eval-reporter.js.map