Daniel Huynh

Daniel Huynh

CEO of LaVague and Mithril Security

Daniel Huynh
Members Public

A quick walkthrough into LLMs, Large Action Models and Agents

Large Action Models (LAMs) areLLMs specifically trained for action generation or generalist LLMs like GPT-4 prompted for action. Agents leverage those to achieve users’ goals. LaVague is a framework to build AI Web Agents.

Daniel Huynh
Members Public

Open-source embeddings and LLMs outperform Gemini and OpenAI for Web Navigation while being faster and cheaper

We have developed evaluation tooling of AI models for Web Navigation. We found that local embedding models are as performant as proprietary ones behind APIs like OpenAI while being as fast or even faster and cheaper.

Daniel Huynh
Members Public

Announcing BigAction

BigAction, an open-source initiative to collect datasets to train and evaluate large action models

Daniel Huynh
Members Public

Announcing LaVague

Introducing LaVague, an open-source Large Action Model framework to automate automation.