Jump to content

Connect Leeroopedia MCP: Equip your AI agents to search best practices, build plans, verify code, diagnose failures, and look up hyperparameter defaults.

Implementation:Marker Inc Korea AutoRAG Dashboard Run

From Leeroopedia
Knowledge Sources
Domains Visualization, RAG Pipeline Optimization
Last Updated 2026-02-12 00:00 GMT

Overview

Concrete tool for launching an interactive web dashboard to visualize optimization trial results, provided by the AutoRAG framework.

Description

The run function in autorag.dashboard assembles and serves an interactive Panel-based web application for exploring the results of an AutoRAG optimization trial. It reads the trial's summary.csv and per-node result files, constructs a tabbed interface with a trial summary view, individual node detail views, and a YAML configuration reference, and launches a local web server.

The trial summary tab renders a markdown document showing each node's best module name, its parameter configuration, and its metric values. Each node tab displays an interactive Tabulator table of all candidate modules with their scores and a detail button that loads the full result DataFrame, along with strip plots and box plots (rendered via Matplotlib and Seaborn) showing metric distributions across candidates. The YAML tab presents the exact configuration file used for the trial, formatted as a code block for easy reference.

Usage

Import and call the run function (or invoke it via the AutoRAG CLI) after an optimization trial has completed. Point it at the trial directory (e.g., my_project/0/) to launch the dashboard. The dashboard opens in the default web browser and remains active until the process is terminated.

Code Reference

Source Location

  • Repository: AutoRAG
  • File: autorag/dashboard.py (lines 176-199)

Signature

def run(trial_dir: str, port: int = 7690):

Import

from autorag.dashboard import run as dashboard_run

I/O Contract

Inputs

Name Type Required Description
trial_dir str yes Path to the completed trial directory. Must contain a summary.csv file, at least one node line subdirectory with node results, and a config.yaml file.
port int no TCP port on which the Panel web server will listen. Default is 7690.

Outputs

Name Type Description
dashboard Interactive Panel web application A local web server is started and the dashboard is opened in the default web browser. The function blocks until the server is stopped. No return value.

Usage Examples

Basic Usage

from autorag.dashboard import run as dashboard_run

# Launch the dashboard for a completed trial
dashboard_run(trial_dir="my_autorag_project/0")
# Opens http://localhost:7690 in the default browser

Custom Port

from autorag.dashboard import run as dashboard_run

# Launch on a custom port (useful when running multiple dashboards)
dashboard_run(trial_dir="my_autorag_project/0", port=8080)
# Opens http://localhost:8080 in the default browser

CLI Usage

# From the command line (equivalent to the Python API):
# autorag dashboard --trial_dir my_autorag_project/0 --port 7690

Related Pages

Implements Principle

Requires Environment

Page Connections

Double-click a node to navigate. Hold to expand connections.
Principle
Implementation
Heuristic
Environment