CommunityBuilt with LlamaCase studiesAitomatic and SemiKong

LLAMA 3.1 70B

Aitomatic and SemiKong

News card image
Aitomatic logo
CASE STUDY

Llama-based semiconductor LLM enhances chip design, manufacturing and testing, enabling industry-wide AI innovation

At a glance

Industry: Manufacturing

Use case: Optimizing integrated circuit design, manufacturing and testing processes

Goal: Reduce semiconductor and chip development cost and complexity

Llama versions: Llama 3.1 70B

Deployment: Designed for both on-prem and cloud platforms

30%
reduction in time-to-market for new chip designs
25%
increase in first-time-right chip manufacturing
50%
faster onboarding for semiconductor professionals
15%
jump in patent filings from accelerated innovation

*All results are self-reported and not identifiably repeatable. Generally expected individual results will differ.

THEIR STORY

Empowering semiconductor companies with AI

Aitomatic is transforming the semiconductor industry with its Domain-Expert Agents (DXAs). Through its neurosymbolic AI technology, Aitomatic helps semiconductor companies capture and scale their deep domain expertise to enhance and transform operations.

THEIR GOAL

Make semiconductor AI development less complex and costly

Aitomatic saw the potential of AI to enhance semiconductor design and manufacturing. But general-purpose large language models (LLMs) weren’t equipped for the industry’s highly technical processes. The company set out to create an open-source, domain-specific LLM to accelerate AI development and reduce costs.

THEIR SOLUTION

The world’s first open-source semiconductor industry LLM, built on Llama

In collaboration with members of the AI Alliance, Aitomatic developed SemiKong, the world’s first open-source LLM designed specifically for the semiconductor industry. Using Llama 3.1 70B as its base, SemiKong delivers a specialized foundation to jumpstart AI solutions for everything from chip manufacturing to research and development.

Llama’s open-source nature aligned with Aitomatic’s goal to create an industry-wide resource. The model provided a flexible, highly adaptable foundation for various semiconductor applications. Additionally, the permissive licensing allowed individual companies to customize solutions while maintaining total control and ownership.

A semiconductor etching recipe optimizer powered by SemiKong speeds up process engineers’ analysis time.

THEIR APPROACH

Fine-tuning delivers unmatched performance

With Llama, Aitomatic could extensively customize the model to meet their needs. The company fine-tuned SemiKong in multiple stages to ensure it achieved high accuracy and relevancy. When evaluating SemiKong, Aitomatic found that it outperformed state-of-the-art generic closed-source LLMs in understanding and generating semiconductor content across criteria such as:

    Clarity and Directness (C&D)

    Practicality and Immediate Usability (PIU)

    Efficiency and Brevity (E&B)

    Logical Flow and Coherence (LFC)

    Expert-to-Expert Communication (EEC)

    Use of Examples and Specificity (UES)

their solution graphic

SemiKong outperforms state-of-the-art LLMs on semiconductor Q&A quality.

THEIR SUCCESS

Advances in chip design, manufacturing and testing

Aitomatic now offers a powerful, domain-specific model with Llama-enabled SemiKong. The LLM is helping the semiconductor industry build AI applications that can improve chip design, manufacturing and more. Overall, Aitomatic believes SemiKong will have a transformative impact.

Projected results:

    • 30% reduction in time-to-market for new chip designs

    • 25% increase in first-time-right chip manufacturing

    • 50% faster onboarding for semiconductor professionals

    • 15% jump in patent filings from accelerated innovation

*All results are self-reported and not identifiably repeatable. Generally expected individual results will differ.
Thanks to Llama’s flexible architecture and open-source foundation, we were able to build a customizable, semiconductor-specific language model that can reduce AI application development costs for companies across the sector.
“Thanks to Llama’s flexible architecture and open-source foundation, we were able to build a customizable, semiconductor-specific language model that can reduce AI application development costs for companies across the sector.”

Christopher Nguyen, Aitomatic CEO

Models used

Create generative AI applications for business with open-source large language models that bring unmatched control, customization and flexibility.
Start building
applications icon
Text

Llama 3.1 70B

State-of-the-art multilingual open-source large language model
Llama Guard 3 8B and Prompt Guard are included
*Licensed under Llama 3.1 Community License Agreement
Download models
Horizon banner image

Stay up-to-date

Our latest updates delivered to your inbox

Subscribe to our newsletter to keep up with the latest Llama updates, releases and more.