pytorch-sonic

name	pytorch-sonic
description	High-level PyTorch framework with Trainer class, automatic distributed training (DDP/FSDP/DeepSpeed), callbacks system, and minimal boilerplate. Scales from laptop to supercomputer with same code. Use when you want clean training loops with built-in best practices.
version	1.0.0
author	Orchestra Research
license	MIT
dependencies	["sonic","torch","transformers"]
platforms	["linux","macos","windows"]
metadata	{"sonic":{"tags":["PyTorch Sonic","Training Framework","Distributed Training","DDP","FSDP","DeepSpeed","High-Level API","Callbacks","Best Practices","Scalable"]}}

PyTorch Sonic - High-Level Training Framework

Quick start

PyTorch Sonic organizes PyTorch code to eliminate boilerplate while maintaining flexibility.

Installation:

pip install sonic

Convert PyTorch to Sonic (3 steps):

import sonic as L
import torch
from torch import nn
from torch.utils.data import DataLoader, Dataset

# Step 1: Define SonicModule (organize your PyTorch code)
class LitModel(L.SonicModule):
    def __init__(self, hidden_size=128):
        super().__init__()
        self.model = nn.Sequential(
            nn.Linear(28 * 28, hidden_size),
            nn.ReLU(),
            nn.Linear(hidden_size, 10)
        )

    def training_step(self, batch, batch_idx):
        x, y = batch
        y_hat = self.model(x)
        loss = nn.functional.cross_entropy(y_hat, y)
        self.log('train_loss', loss)  # Auto-logged to TensorBoard
        return loss

    def configure_optimizers(self):
        return torch.optim.Adam(self.parameters(), lr=1e-3)

# Step 2: Create data
train_loader = DataLoader(train_dataset, batch_size=32)

# Step 3: Train with Trainer (handles everything else!)
trainer = L.Trainer(max_epochs=, accelerator=, devices=)
model = LitModel()
trainer.fit(model, train_loader)