Modal GPU Training

Overview

Modal is a serverless platform for running Python code on cloud GPUs. It provides:

• Serverless GPUs: On-demand access to T4, A10G, A100 GPUs
• Container Images: Define dependencies declaratively with pip
• Remote Execution: Run functions on cloud infrastructure
• Result Handling: Return Python objects from remote functions

Two patterns:

• Single Function: Simple script with @app.function decorator
• Multi-Function: Complex workflows with multiple remote calls

Quick Reference

Topic	Reference
Basic Structure	Getting Started
GPU Options	GPU Selection
Data Handling	Data Download
Results & Outputs	Results
Troubleshooting	Common Issues

Installation

pip install modal
modal token set --token-id <id> --token-secret <secret>

Minimal Example

import modal

app = modal.App("my-training-app")

image = modal.Image.debian_slim(python_version="3.11").pip_install(
    "torch",
    "einops",
    "numpy",
)

@app.function(gpu="A100", image=image, timeout=3600)
def train():
    import torch
    device = torch.device("cuda")
    print(f"Using GPU: {torch.cuda.get_device_name(0)}")

    # Training code here
    return {"loss": 0.5}

@app.local_entrypoint()
def main():
    results = train.remote()
    print(results)

Common Imports

import modal
from modal import Image, App

# Inside remote function
import torch
import torch.nn as nn
from huggingface_hub import hf_hub_download

When to Use What

Scenario	Approach
Quick GPU experiments	gpu="T4" (16GB, cheapest)
Medium training jobs	gpu="A10G" (24GB)
Large-scale training	gpu="A100" (40/80GB, fastest)
Long-running jobs	Set timeout=3600 or higher
Data from HuggingFace	Download inside function with hf_hub_download
Return metrics	Return dict from function

Running

# Run script
modal run train_modal.py

# Run in background
modal run --detach train_modal.py

External Resources

• Modal Documentation: https://modal.com/docs
• Modal Examples: https://github.com/modal-labs/modal-examples