python 33 lines · 1 tab

A clean PyTorch training loop with validation and checkpoints

Dr. Elena Vasquez Apr 2026

1 tab

import torch

best_val_loss = float('inf')

for epoch in range(1, num_epochs + 1):
    model.train()
    train_loss = 0.0
    for batch in train_loader:
        inputs, targets = [item.to(device) for item in batch]
        optimizer.zero_grad(set_to_none=True)
        outputs = model(inputs)
        loss = criterion(outputs, targets)
        loss.backward()
        torch.nn.utils.clip_grad_norm_(model.parameters(), max_norm=1.0)
        optimizer.step()
        train_loss += loss.item() * inputs.size(0)

    model.eval()
    val_loss = 0.0
    with torch.no_grad():
        for batch in valid_loader:
            inputs, targets = [item.to(device) for item in batch]
            outputs = model(inputs)
            loss = criterion(outputs, targets)
            val_loss += loss.item() * inputs.size(0)

    val_loss /= len(valid_loader.dataset)
    if val_loss < best_val_loss:
        best_val_loss = val_loss
        torch.save({'model_state': model.state_dict()}, 'best_model.pt')

    scheduler.step(val_loss)
    print(f'epoch={epoch} val_loss={val_loss:.4f}')

1 file · python Explain with highlit

The training loop is where research code either becomes maintainable or turns into a mess. I keep it explicit: train phase, validation phase, scheduler step, metric tracking, and checkpoint saving. That structure pays off immediately when experiments fail halfway through or need to be resumed on another machine.

Related snips

ruby

class SignupForm
  include ActiveModel::Model
  include ActiveModel::Attributes

  attribute :account_name, :string
  attribute :email, :string

Shallow Controller, Deep Params: Form Object Pattern

rails activemodel form-object

by codesnips 3 tabs

html

<!DOCTYPE html>
<html lang="en">
<head>
  <meta charset="UTF-8">
  <title>Form Validation Example</title>
  <style>

HTML forms with validation and accessibility

html forms validation

by Alex Chang 1 tab

typescript

import axios from 'axios';

export type NormalizedErrors = {
  fields: Record<string, string>;
  formLevel: string | null;
};

Frontend: normalize and display server validation errors

ux typescript react

by codesnips 3 tabs

php

<?php

namespace App\Http\Requests;

use Illuminate\Foundation\Http\FormRequest;
use Illuminate\Validation\Rule;

Laravel form requests for validation

laravel validation form-requests

by Carlos Mendez 2 tabs

python

import torch.nn as nn

class SmallCNN(nn.Module):
    def __init__(self, num_classes: int) -> None:
        super().__init__()
        self.features = nn.Sequential(

Convolutional neural networks for image classification in PyTorch

pytorch cnn computer-vision

by Dr. Elena Vasquez 1 tab

ruby

class ReindexCheckpoint < ApplicationRecord
  enum status: { idle: 0, running: 1, done: 2, failed: 3 }

  validates :index_name, presence: true, uniqueness: true

  def self.for(index_name)

Safer Background Reindex: slice batches + checkpoints

rails reliability elasticsearch

by codesnips 4 tabs

Share this code

Here's the card — post it anywhere.

A clean PyTorch training loop with validation and checkpoints — share card

Link copied