benchmark/embedding/benchmark.py [261:274]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
        total_time = time.time() - start_time

        # Filter out failed requests
        valid_results = [
            (elapsed, tokens) for elapsed, tokens in results if elapsed > 0
        ]
        if not valid_results:
            return 0, 0, 0

        # Calculate performance metrics
        total_elapsed = sum(elapsed for elapsed, _ in valid_results)
        avg_latency = total_elapsed / len(valid_results)
        total_tokens = sum(tokens for _, tokens in valid_results)
        throughput = total_tokens / total_time
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



benchmark/embedding/benchmark_http.py [173:186]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
        total_time = time.time() - start_time

        # Filter out failed requests and sum up tokens
        valid_results = [
            (elapsed, tokens) for elapsed, tokens in results if elapsed > 0
        ]
        if not valid_results:
            return 0, 0, 0

        # Calculate average latency and throughput
        total_elapsed = sum(elapsed for elapsed, _ in valid_results)
        avg_latency = total_elapsed / len(valid_results)
        total_tokens = sum(tokens for _, tokens in valid_results)
        throughput = total_tokens / total_time
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -