DEDUPED Hot AI repos. Real use cases. What people build. Get the Monday 5

How to train and validate a YOLO model on your dataset with Ultralytics

Fine-tune YOLO26n on a dataset config, then check its metrics on the validation set.

Short answer: Install with pip install ultralytics; then in Python load the model with YOLO("yolo26n.pt"); call model.train with data set to your dataset yaml (the README uses coco8.yaml), epochs and imgsz. All steps.

  • Data and analytics
  • Intermediate
  • 1 to 3 hours
  • From the README
  • Claude Code
  • Codex
  • Cursor
  • Gemini CLI

platform. AGPL-3.0: commercial closed-source use needs an Ultralytics enterprise license.

Who it is forA team building detection for its own objects, like products on a shelf.

The problemPretrained classes rarely match the objects you care about.

The outcomeA trained model with validation metrics.

What you need first

  • Python 3.8 or newer with pip (PyTorch is installed with the package requirements)
  • A GPU is optional for prediction and recommended for training
  • Images or video to run on, or a dataset config for training

Steps

  1. Install with pip install ultralytics
  2. In Python load the model with YOLO("yolo26n.pt")
  3. Call model.train with data set to your dataset yaml (the README uses coco8.yaml), epochs and imgsz
  4. Call model.val() to evaluate on the validation set

Prompt for your agent

With ultralytics/ultralytics, write a Python script that loads YOLO("yolo26n.pt"), trains on <my dataset yaml> for 100 epochs at imgsz 640 like the README example, runs model.val(), and prints the key metrics.

Caveat. Training on CPU works per the README example but is slow.

Use this in your AI agent

Ask your agent

Using the Deduped MCP, get the setup steps for ultralytics/ultralytics and "Train and validate a YOLO model on your dataset"

Add Deduped to Claude Code once

claude mcp add --transport http deduped https://dedupedai.com/mcp

Cursor, ChatGPT, others

Share: X LinkedIn Hacker News Email Copy link

Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer