Last week, I broke my silence after being in hiding for months (and growing a pretty impressive beard).

And in that space of reflection, I confronted a truth: I need to be more regular in the newsletter and will continue that.

🌸 Agents

✨ PersonaRAG

It enhances RAG models by incorporating user-centric agents, It also adapted retrieval and generation based on real-time user data.That includes components:

• K-docs retrieval

• User interaction analysis (user profile/contextual retrieval/live session/document ranking/feedback agents)

• Cognitive dynamic adaption(selective/collaborative use of agents).

Large Language Models (LLMs) struggle with generating reliable outputs due to outdated knowledge and hallucinations. Retrieval-Augmented Generation (RAG) models address this by enhancing LLMs with external knowledge, but often fail to personalize the retrieval process. This paper introduces PersonaRAG, a novel framework incorporating user-centric agents to adapt retrieval and generation based on real-time user data and interactions. Evaluated across various question answering datasets, PersonaRAG demonstrates superiority over baseline models, providing tailored answers to user needs. The results suggest promising directions for user-adapted information retrieval systems.

• Code 

• Paper 

✨ Internet of Agents

Researchers from Tsinghua University, Peking University, Beijing University of Posts and Telecommunications, and Tencent have created a new framework called the Internet of Agents (IoA) (Chen et al., 2024).

This framework aims to solve problems faced by current systems that use multiple large language model based artificial intelligence agents.

IoA allows for better collaboration among different types of AI agents, enabling them to work together on various tasks more effectively.

A random thought: Have you seen the Chinese Contributions to LLM agents? They're seriously impressive!

The rapid advancement of large language models (LLMs) has paved the way for the development of highly capable autonomous agents. However, existing multi-agent frameworks often struggle with integrating diverse capable third-party agents due to reliance on agents defined within their own ecosystems. They also face challenges in simulating distributed environments, as most frameworks are limited to single-device setups. Furthermore, these frameworks often rely on hard-coded communication pipelines, limiting their adaptability to dynamic task requirements. Inspired by the concept of the Internet, we propose the Internet of Agents (IoA), a novel framework that addresses these limitations by providing a flexible and scalable platform for LLM-based multi-agent collaboration. IoA introduces an agent integration protocol, an instant-messaging-like architecture design, and dynamic mechanisms for agent teaming and conversation flow control. Through extensive experiments on general assistant tasks, embodied AI tasks, and retrieval-augmented generation benchmarks, we demonstrate that IoA consistently outperforms state-of-the-art baselines, showcasing its ability to facilitate effective collaboration among heterogeneous agents. IoA represents a step towards linking diverse agents in an Internet-like environment, where agents can seamlessly collaborate to achieve greater intelligence and capabilities.

• Code 

• Paper 

🌸 AI

✨ fastest.ai 

The site to retrieve the latency of the LLM models.

✨ GPT-2 (124M) in llm.c in 90 minutes for $20

It is the near future - when you can train a llama 3 model from scratch for 1000 dollars.

✨ 25% of top websites fully restricting AI training via robots.txt

🌸 BioTech

• Samsung Bioepis Gets FDA Approval for Biosimilar to AstraZeneca’s Soliris

• Cutting through the AI hype in drug discovery

• Bridge RNA: A new gene editing technique that could overcome the limitations of CRISPR

• 10 clinical trials to watch in the second half of 2024

• Startup led by former J&J executives raises $165M for cancer, immune disease drugs

• The obesity drug race is being fought on the manufacturing floor as GLP-1 rivals make plans to catch leaders

P.S. The response to my last email really touched me. It was inspiring to read so many of your stories and hear the things you’re working through. I’m glad that we’re on this journey together.

Thank you for reading Musings on AI. This post is public so feel free to share it.