ChipNeMo: Domain-Adapted LLMs for Chip Design
This work addresses the need for specialized AI tools in chip design, showing that domain adaptation can enhance LLM effectiveness in this technical field, though it is incremental in applying existing adaptation methods to a new domain.
The authors tackled the problem of applying large language models to industrial chip design by developing domain adaptation techniques including tokenization, continued pretraining, and alignment, resulting in their ChipNeMo-70B model outperforming GPT-4 on engineering assistant chatbot and EDA script generation tasks while maintaining competitive performance on bug analysis.
ChipNeMo aims to explore the applications of large language models (LLMs) for industrial chip design. Instead of directly deploying off-the-shelf commercial or open-source LLMs, we instead adopt the following domain adaptation techniques: domain-adaptive tokenization, domain-adaptive continued pretraining, model alignment with domain-specific instructions, and domain-adapted retrieval models. We evaluate these methods on three selected LLM applications for chip design: an engineering assistant chatbot, EDA script generation, and bug summarization and analysis. Our evaluations demonstrate that domain-adaptive pretraining of language models, can lead to superior performance in domain related downstream tasks compared to their base LLaMA2 counterparts, without degradations in generic capabilities. In particular, our largest model, ChipNeMo-70B, outperforms the highly capable GPT-4 on two of our use cases, namely engineering assistant chatbot and EDA scripts generation, while exhibiting competitive performance on bug summarization and analysis. These results underscore the potential of domain-specific customization for enhancing the effectiveness of large language models in specialized applications.