Skip to content

ChipNeMo

Adapt large language models to chip design

Visit official site
Category AI models

What ChipNeMo is

ChipNeMo is described as an NVIDIA project that explores applying large language models to industrial chip design through domain adaptation instead of off-the-shelf models.

What you can do with it

  • Apply a custom tokenizer to chip design text
  • Run domain-adaptive continued pre-training
  • Summarise bugs found in design data

Example uses

  • A team looks for bugs with the model
  • EDA scripts are generated to speed up design flows
  • Error messages from design work are analysed and summarised

Model notes

  • Supervised fine-tuning uses domain-specific instructions
  • A domain-adapted retrieval model is part of the approach
  • The source describes the work as a research publication

Pros & cons

✓ What we like

  • Domain-specific instructions are used in fine-tuning
  • A domain-adapted retrieval model is listed

! What to watch out for

  • The material describes it as an NVIDIA research project
  • Adaptation draws on domain-specific continued pre-training

FAQ

Who published it?

The material names NVIDIA as the publisher.

What adaptation techniques are named?

A custom tokenizer, domain-adaptive continued pre-training, supervised fine-tuning and a domain-adapted retrieval model.

Last reviewed: 2026-11-15

More LLM API platform tools

View all →

How we review