flexium 0.2.0a2

Flexium.AI - Flexible Resource Allocation for GPU training with live migration

Flexium

Zero-downtime GPU migration for PyTorch training
Move running training jobs between GPUs without stopping them.

Quick Start • Features • How It Works • Sign Up Free

---

What is Flexium?

Flexium enables live GPU migration for PyTorch training. Move your training between GPUs without stopping, checkpointing, or losing progress.

import flexium.auto

with flexium.auto.run():
    # Your existing training code - unchanged
    for epoch in range(100):
        for batch in dataloader:
            loss = model(batch).sum()
            loss.backward()
            optimizer.step()
    # Training transparently migrates between GPUs when needed

Zero code changes. Zero downtime. Zero memory residue.

Features

• Live Migration - Move training between GPUs on the same server without stopping
• Zero Code Changes - Just wrap your training loop
• Zero Memory Residue - Source GPU is completely freed
• PyTorch & Lightning - Works with PyTorch and PyTorch Lightning
• Web Dashboard - Visual GPU management and monitoring
• Pause/Resume - Temporarily free GPU, resume on any available GPU

Quick Start

1. Install

pip install flexium

2. Get Your API Key

Sign up for free at https://flexium.ai

3. Add Two Lines to Your Training

import flexium.auto

with flexium.auto.run():
    # Your existing training code - no changes needed
    model = MyModel().cuda()
    for epoch in range(100):
        train_one_epoch(model)

That's it! Your training is now migratable.

How It Works

┌─────────────────────────────────────────────────────────────┐
│                    YOUR TRAINING CODE                        │
│                                                              │
│   with flexium.auto.run():                                  │
│       model.train()   ◄─── Training runs normally           │
└─────────────────────────────────────────────────────────────┘
                              │
                              │ Heartbeat
                              ▼
┌─────────────────────────────────────────────────────────────┐
│                   FLEXIUM SERVER                             │
│                                                              │
│   Monitors your GPUs and coordinates migrations             │
│   - Web dashboard for visual management                     │
│   - One-click migration between GPUs                        │
│   - Pause/resume training jobs                              │
└─────────────────────────────────────────────────────────────┘

When migration is triggered (via dashboard or CLI), Flexium:

1. Captures complete GPU state at driver level (driver 580+)
2. Frees source GPU completely (0 MB residue)
3. Restores on target GPU seamlessly

Your training code never knows it moved.

Use Cases

GPU Sharing

# Free up GPUs for teammates without killing your job
with flexium.auto.run():
    train_model()  # Can be migrated via dashboard anytime

Resource Optimization

# Move jobs between GPUs based on memory requirements
with flexium.auto.run():
    train_model()  # Migrate to GPU with more VRAM when needed

Pause for Priority Jobs

# Pause your job to free GPU, resume later
with flexium.auto.run():
    train_model()  # Pause via dashboard, resume on any GPU

Requirements

• Python 3.8+
• PyTorch 2.0+
• NVIDIA GPU with driver 580+
• CUDA 12.4+
• Linux x86_64

Documentation

• Getting Started
• API Reference
• Examples
• Troubleshooting

Support

• Docs: docs.flexium.ai
• Discord: Join our community
• Issues: GitHub Issues
• Email: [e-mail]

License

Apache 2.0 - See LICENSE for details.

---

Built with ❤️ by the Flexium team
flexium.ai