Introducing our work on general-purpose LLM Agents | GoodAI
goodai.com
Introducing our work on general-purpose LLM Agents | GoodAI
An agent can be thought of as a logical wrapper around an LLM, allowing us to add several features to our AI systems, primarily:
- Tool usage, such as calling APIs for info, executing code,
- Internal thoughts over multiple generation steps
- Ability to use various tools and reasoning steps to answer more complex queries.
- Parallel agents can go and complete
James Briggs • LLMs Are Not All You Need | Pinecone
3 days reading through the code of the paper that solved this, here are some notes:
- Long term interactions: the promise of LLM agents is to potentially run over weeks. A lot of the ideas in the paper tackle exactly this problem
- Memory: It’s easy to write down interactions of agents. But how do we track these over time? How do these change with... See more
- Long term interactions: the promise of LLM agents is to potentially run over weeks. A lot of the ideas in the paper tackle exactly this problem
- Memory: It’s easy to write down interactions of agents. But how do we track these over time? How do these change with... See more
Sanyam Bhutani • Tweet
… to build AI systems with social cognition, so that they can interact meaningfully with the user over the long term, we first need to solve two foundational problems in how memory systems work. These problems, called oneshot learning and continual learningare currently (as of early 2024) missing in publicly available LLMs.