Tutorials

This is a collection of tutorials written in Python to help you ramp up using TT-NN for various tasks such as tensor operations, model conversion, and profiling.

These tutorials assume you already have a machine set up with either a wormhole or blackhole device available and that you have successfully followed the instructions for TT-NN / TT-Metalium Installation.

With the recommended virtual environment activated, you can run these tutorials directly:

$ python3 --version
Python 3.10.12
$ python3 example.py
...

• Create and Add Two Tensors
  • Import the necessary libraries
  • Open Tenstorrent device
  • Tensor Creation
  • Perform the addition operation and convert back
  • Full example and output
• Basic Operations with TT-NN
  • Import the necessary libraries
  • Open Tenstorrent device
  • Helper Function for Tensor Preparation
  • Host Tensor Creation
  • Convert Host Tensors to TT-NN Tiled Tensors or Create Natively on Device
  • Tile-Based Arithmetic Operations
  • Simulated Broadcasting (Row Vector Expansion)
  • Full example and output
• MLP inference with TT-NN
  • Import the necessary libraries
  • Open Tenstorrent device
  • Load MNIST Test Data
  • Load Pretrained MLP Weights
  • Basic accuracy tracking, inference, loop, and image flattening
  • Convert to TT-NN Tensor
  • Layer 1 (Linear + ReLU)
  • Layer 2 (Linear + ReLU)
  • Layer 3 (Output Layer)
  • Convert Back to PyTorch and sum results
  • Full example and output