Skip to content
InfratoApp

NVIDIA · Course

The Art of Compressing LLMs: Pruning, Distillation, and Quantization Demystified (self-paced)

The Art of Compressing LLMs: Pruning, Distillation, and Quantization Demystified is a self-paced NVIDIA course on shrinking large language models while preserving their performance, so they cost less to run and deploy.

Open at NVIDIA ↗Common questions

Who it suits

  • Engineers who must deploy large models under memory, latency or cost limits.
  • Researchers choosing between compression methods for a specific problem.

What it covers

  • The mechanics of pruning, knowledge distillation and quantization, with hands-on practice.
  • The trade-offs between accuracy, latency, cost and hardware requirements for each method.
  • Strategies for combining the methods for maximum efficiency.
  • The optimisation pipeline from compression to efficient deployment and benchmarking with industry-standard tools.

Common questions about The Art of Compressing LLMs: Pruning, Distillation, and Quantization Demystified (self-paced)

Why compress a model?
NVIDIA says the memory footprint of large models drives up computational costs, complicates deployment and adds latency.
Does it cost anything?
NVIDIA sells the self-paced course with Buy Now and Redeem Code options, and says online purchase is supported in select locations. Its page states the price, which changes, so check it.
Is it theory or practice?
NVIDIA says the curriculum bridges theory and application, with hands-on practice for each method.

More like this

Keep learning on Infrato

Where this comes from

Written from NVIDIA’s own pages, read on 2026-10-14. Content, prices and availability change, so the provider’s page is the source. We are not affiliated with NVIDIA and this is not an endorsement.