axonml-cli

Overview

axonml-cli is the official command-line interface for the AxonML machine learning framework. It provides a comprehensive toolkit for managing ML projects, training models, evaluating performance, and deploying to production.

The CLI supports the full ML workflow from project initialization to model serving, with integrations for Kaggle, Weights & Biases, and a built-in model hub.

Features

• Project Management - Create and initialize AxonML projects with customizable templates and configurations.

• Training & Evaluation - Train models from configuration files, resume from checkpoints, and evaluate with comprehensive metrics.

• Model Operations - Convert between formats (ONNX, SafeTensors, Ferrite), quantize models (Q4, Q8, F16), and inspect architectures.

• Data Management - Upload, analyze, validate, and preview datasets with automatic type detection and statistics.

• Deployment - Export models for production, start inference servers, and manage dashboard services.

• Integrations - Kaggle dataset downloads, W&B experiment tracking, and pretrained model hub access.

• Terminal UI - Interactive TUI for exploring models and datasets with real-time visualization.

• GPU Support - Detect, benchmark, and manage GPU devices for accelerated training.

Modules

Installation

Install from crates.io:

cargo install axonml-cli

Or build from source:

cargo build --release -p axonml-cli

Command Reference

Usage

Project Commands

# Create a new AxonML project
axonml new my-project --template default

# Initialize AxonML in an existing directory
axonml init --name my-project

# Generate a Rust training project scaffold
axonml scaffold generate my-trainer --template training --wandb

Training Commands

# Train a model from configuration
axonml train --config config.toml --epochs 50 --device cuda:0

# Resume training from a checkpoint
axonml resume checkpoint.pt --epochs 20 --lr 0.0001

# Evaluate model performance
axonml eval model.pt data/ --metrics accuracy,f1,precision,recall

# Make predictions
axonml predict model.pt input.json --format json --top-k 5

Model Commands

# Inspect model architecture
axonml inspect model.pt --detailed --show-params 5

# Convert model formats
axonml convert model.pt model.onnx --to onnx --optimize

# Export for deployment
axonml export model.pt ./deploy --format onnx --target cuda --quantize

# Quantize model
axonml quant convert model.pt --target Q8_0 --output model_q8.axon

# Generate evaluation report
axonml report model.pt --data test/ --format html --confusion-matrix

Data Commands

# Analyze a dataset
axonml data analyze ./data --detailed --recommend

# Upload and configure dataset
axonml data upload ./images --data-type image --task classification

# Validate dataset structure
axonml data validate ./data --check-balance --check-missing

# Preview dataset samples
axonml data preview ./data --num-samples 10 --random

Workspace Commands

# Load model into workspace
axonml load model model.pt --name my-model

# Load dataset into workspace
axonml load data ./dataset --data-type tabular

# Analyze loaded model
axonml analyze model --detailed --output report.json

# Generate comprehensive report
axonml analyze report --format html --visualize

Benchmarking

# Benchmark model performance
axonml bench model model.pt --iterations 100 --device cuda:0

# Benchmark at different batch sizes
axonml bench inference model.pt --batch-sizes 1,2,4,8,16,32

# Compare multiple models
axonml bench compare "model1.pt,model2.pt,model3.pt" --iterations 50

# Benchmark hardware capabilities
axonml bench hardware --iterations 10

GPU Management

# List available GPUs
axonml gpu list

# Show detailed GPU information
axonml gpu info

# Select GPU for training
axonml gpu select cuda:0 --persistent

# Benchmark GPU performance
axonml gpu bench --all --iterations 10

Hub & Kaggle Integration

# List pretrained models
axonml hub list

# Download pretrained weights
axonml hub download resnet50 --force

# Configure Kaggle credentials
axonml kaggle login --username USER --key API_KEY

# Search Kaggle datasets
axonml kaggle search "image classification" --limit 20

# Download Kaggle dataset
axonml kaggle download username/dataset-name --output ./data

Dashboard & Server

# Start dashboard and API server
axonml start --port 3000 --dashboard-port 8080

# Start only the API server
axonml start --server --foreground

# Check service status
axonml status --detailed

# View logs
axonml logs --follow --lines 100

# Stop services
axonml stop

Additional Commands

# Create model/dataset bundle
axonml zip create --output bundle.axonzip --model model.pt --data ./data

# Extract bundle
axonml zip extract bundle.axonzip --output ./extracted

# Rename model
axonml rename model model.pt new_model.pt

# Launch terminal UI
axonml tui --model model.pt --data ./dataset

Configuration

The CLI uses axonml.toml for project configuration:

[project]
name = "my-ml-project"
version = "0.1.0"
description = "My machine learning project"

[training]
epochs = 50
batch_size = 32
learning_rate = 0.001
device = "cuda:0"
checkpoint_frequency = 1
output_dir = "./output"
num_workers = 4

[training.optimizer]
name = "adam"
weight_decay = 0.0001
beta1 = 0.9
beta2 = 0.999

[training.scheduler]
name = "cosine"
t_max = 50
eta_min = 0.00001
warmup_epochs = 5

[model]
architecture = "resnet18"
num_classes = 10
dropout = 0.1

[data]
train_path = "./data/train"
val_path = "./data/val"
val_split = 0.1
augmentation = true
shuffle = true
normalize = true

Global Options

# Enable verbose output
axonml --verbose <command>

# Suppress all output except errors
axonml --quiet <command>

Tests

Run the test suite:

cargo test -p axonml-cli

Run integration tests:

cargo test -p axonml-cli --test cli_integration_test

License

Licensed under either of:

• MIT License
• Apache License, Version 2.0

at your option.