Skip to content
View dr-gareth-roberts's full-sized avatar
🎯
Focusing
🎯
Focusing
31 followers · 80 following

Achievements

Achievement: QuickdrawAchievement: Pull Shark

Achievements

Achievement: QuickdrawAchievement: Pull Shark

Highlights

Block or report dr-gareth-roberts

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Please don't include any personal information such as legal names or email addresses. Maximum 100 characters, markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
dr-gareth-roberts/README.md

Hi! My name is Dr Gareth Roberts

I work as an AI Consultant (specialising in end-to-end implementation of AI solutions both cloud-based and offline)

I have worked as Chief Technology Officer, Head of Artificial Intelligence, and other such titles for a large number of companies across multiple verticles (eCommerce, Property Tech, Mining Exploration, Life Insurance). I have a PhD in Cognitive Neuroscience, Developmental Psychology and Artificial Intelligence. I spent 5 years at the University of Sydney as a research fellow where I was involved in projects across multiple disciplines, institutes and schools (Psychology, Education, Learning, Cognitive Neuroscience, Neuroimaging Laboratories, Sleep Institute)

About Me

  • 🌍  I'm based in Sydney, Australia but travel to Perth quite regularly. Wife and I love travelling around the world.
  • ✉️  You can email me at gareth.roberts@ieee.org.
  • 🧠  I love hacking large language models, running complex prompt chains, probing how they work, etc.
  • ⚡  Heavy cognitive neuroscience background, theoretical cognitive science and AI, commercial mindset.

Current Projects

  • Simulating cognitive processes using LLMs
  • Advanced prompt engineering attacks and mitigation strategies
  • Developing a comprehensive framework for probing the inner worlds of LLMs
  • Trying to get multi-modal LLMs to solve basic fluid intelligence test items that were not in their training set (GPT4o and Claude 3.5 Sonnet perform below chance at the Cattell Culture Fair Intelligence Test - Scale 3 Form A).

Pinned Loading

  1. LLM-Dev LLM-Dev Public

    Python Tools for Developing with LLMs (cloud & offline)

    Python

  2. llm-tester llm-tester Public

    A comprehensive Python library for the systematic investigation of large language models' performance profile.

  3. Powerful-Prompts Powerful-Prompts Public

    Deep dives on a single keyword? Impressive Fermi estimations? Struggling to handle word math problems? Solutions are all here.

  4. insideLLMs insideLLMs Public

    Python library for probing the inner workings of large language models. Systematically test LLMs' zero-shot ability at unseen logic problems, propensity for bias, and vulnerabilities to particular …

  5. baml baml Public

    Forked from BoundaryML/baml

    A programming language for LLM Guardrails, Typesafey, and Observability

    Rust 1

  6. sigoden/aichat sigoden/aichat Public

    All-in-one LLM CLI tool featuring Shell Assistant, Chat-REPL, RAG, AI tools & agents, with access to OpenAI, Claude, Gemini, Ollama, Groq, and more.

    Rust 4.5k 302