Glossary · Term

PyTorch

← all terms

Definition

Plain language

A widely used free toolkit for building and training AI models, written in the Python language.

As stated in the literature

An open-source deep-learning framework with dynamic computation graphs and automatic differentiation; the de facto research standard, with custom CUDA kernels often benchmarked against its reference implementations.

Why it matters: It is the go-to toolkit that lets researchers build and train AI models without reinventing the underlying machinery.

For example, a researcher might use PyTorch to define a neural network in a few lines of Python and have it automatically work out the math for training.

Heard on the show

“One — every speedup here is measured against PyTorch eager, the unoptimized default.”
Episode 177 — Why Raw Profiler Data Made an AI Worse at Writing GPU Code

Mentioned in 7 episodes

  1. 177
    Why Raw Profiler Data Made an AI Worse at Writing GPU Code
  2. 124
    A Cheap Model With the Blueprints Beats Expensive Models Working Blind
  3. 065
    One Loop to Optimize Them All: A Universal API for LLM-Driven Discovery
  4. 053
    An AI Agent Swapped In Focal Loss And Beat A Human-Tuned Training Script
  5. 036
    Sparse Attention Was the Wrong Frame. Treat It as Geometry Instead.
  6. 033
    Echo: The Paper Arguing You Never Needed a KV Cache for Retrieval
  7. 027
    When AI Agents Build the Serving Stack: A Bet on Bespoke Infrastructure

Related terms