axonml-tui 0.2.8

Terminal User Interface for Axonml ML Framework
Documentation

Overview

axonml-tui is a comprehensive terminal user interface for the AxonML machine learning framework. It provides real-time visualization of neural network architectures, dataset statistics, training progress, loss curves, and a file browser for navigating models and datasets - all from the comfort of your terminal.

Features

  • Model Architecture Visualization - View layer structures, parameter counts, shapes, and trainable status with interactive navigation
  • Dataset Explorer - Analyze class distributions, feature statistics, data splits, and preview dataset metadata
  • Real-Time Training Monitor - Track epochs, batches, loss/accuracy metrics with sparkline trends and ETA estimation
  • Interactive Graphs - Visualize loss curves, accuracy charts, and learning rate schedules with switchable chart types
  • File Browser - Navigate directories, preview model and dataset files, with support for multiple formats
  • Help System - Comprehensive keyboard shortcut reference organized by category
  • Themeable UI - NexusForge-inspired color scheme with teal, terracotta, and cream accents

Modules

Module Description
app Main application state, tab management, and navigation logic
event Keyboard input handling and event routing to views
theme NexusForge color palette and style presets (teal, terracotta, cream)
ui Core rendering logic for header, content, footer, and help overlay
views::model Neural network architecture display with layer details
views::data Dataset statistics, class distribution, and feature tables
views::training Real-time training progress with gauges and sparklines
views::graphs Loss, accuracy, and learning rate chart rendering
views::files Tree-based file browser with preview pane
views::help Keyboard shortcut reference organized by category

Usage

Add axonml-tui to your Cargo.toml:

[dependencies]
axonml-tui = "0.1.0"

Basic Example

use axonml_tui::run;

fn main() -> std::io::Result<()> {
    // Launch the TUI with no pre-loaded files
    run(None, None)
}

Loading a Model on Startup

use axonml_tui::run;
use std::path::PathBuf;

fn main() -> std::io::Result<()> {
    let model_path = PathBuf::from("models/mnist_classifier.axonml");
    run(Some(model_path), None)
}

Loading Both Model and Dataset

use axonml_tui::run;
use std::path::PathBuf;

fn main() -> std::io::Result<()> {
    let model_path = PathBuf::from("models/resnet.axonml");
    let data_path = PathBuf::from("datasets/cifar10");
    run(Some(model_path), Some(data_path))
}

Keyboard Shortcuts

Global

Key Action
Tab / Shift+Tab Switch between views
1 - 5 Jump to specific view (Model, Data, Training, Graphs, Files)
? Toggle help overlay
q Quit application

Navigation

Key Action
j / Down Move selection down
k / Up Move selection up
h / Left Collapse / Previous panel
l / Right Expand / Next panel
Enter Select / Open

Model View

Key Action
d Toggle detailed layer view
Enter View layer details

Training View

Key Action
p Pause / Resume training
r Refresh training data

Graphs View

Key Action
< / > Switch chart type (Loss / Accuracy / LR)
z Toggle zoom mode

Files View

Key Action
Enter Open file / Toggle directory
Backspace / u Go to parent directory
~ Go to home directory

Supported File Formats

Category Extensions
Models .axonml, .onnx, .pt, .pth, .safetensors, .h5, .keras
Datasets .npz, .npy, .csv, .parquet, .arrow, .tfrecord
Config .toml, .yaml, .yml, .json

Tests

Run the test suite:

cargo test -p axonml-tui

Run with verbose output:

cargo test -p axonml-tui -- --nocapture

Dependencies

  • ratatui - Terminal UI framework
  • crossterm - Cross-platform terminal manipulation
  • tokio - Async runtime
  • serde / serde_json - Serialization
  • chrono - Time handling
  • dirs - File system utilities

License

Licensed under either of:

at your option.