Skip to content Skip to sidebar Skip to footer

Artificial Intelligence

Assessing AI Model Safety via Red Teaming Method: An In-depth Analysis of LLM and MLLM’s Resilience to Jailbreak Assaults and Prospective Enhancements

Large Language Models (LLMs) and Multimodal Large Language Models (MLLMs) are key advancements in artificial intelligence (AI) capable of generating text, interpreting images, and understanding complex multimodal inputs, mimicking human intelligence. However, concerns arise due to their potential misuse and vulnerabilities to jailbreak attacks, where malicious inputs trick the models into generating harmful or objectionable…

Read More

Introducing AnythingLLM: A Publicly Accessible, Comprehensive AI Desktop Application for Local LLMs + RAG.

In the modern business landscape, artificial intelligence (AI) has dramatically reshaped how organizations communicate, particularly when it comes to making use of documents. This is where AnythingLLM comes in – an open-source, innovative full-stack application that uses chatbot technology to enhance how companies interact with their documents. AnythingLLM is designed with an emphasis on efficiency,…

Read More

The automated platform instructs users on the appropriate instances to engage with an AI assistant.

Researchers at MIT and the MIT-IBM Watson AI Lab have developed a method of teaching users when to collaborate with an artificial intelligence (AI) assistant. The model creates a customised onboarding process, educating users on when to trust or ignore an AI model’s advice. The training process can detect situations where the AI model is…

Read More

The team from MIT has published reports on the management of AI.

A committee consisting of leaders and scholars from the Massachusetts Institute of Technology (MIT) have released policy briefs that provide a blueprint for governing artificial intelligence (AI) in the U.S. The committee believes that existing regulatory and liability processes can be extended to regulate AI in the U.S. The goal is to strengthen U.S. leadership…

Read More

A software engineer advances the limits of geometric study.

Justin Solomon is an associate professor in the MIT Department of Electrical Engineering and Computer Science and a member of the Computer Science and Artificial Intelligence Laboratory who is using geometric techniques to solve complex problems in data science and artificial intelligence, among other areas. These techniques draw upon the geometric structures within datasets to…

Read More

Bridging the gap between design and production for optical instruments.

Photolithography, the technique of etching features onto a surface using light manipulation, is commonly used in the manufacturing of computer chips and optical devices. However, small deviations during the manufacturing process often impact the performance of the finished product. To address this, researchers from MIT and the Chinese University of Hong Kong have leveraged machine…

Read More

Deep neural networks exhibit potential as representations of human auditory perception.

MIT researchers have found that computational models based on machine learning that simulate the human auditory system are drawing closer to potentially helping in the creation of improved hearing aids, cochlear implants and brain-machine interfaces. The study is the most comprehensive comparison so far made between these computer models and the human auditory system. Notably,…

Read More

“AutoTRIZ: A Creative AI Instrument that Utilizes Extensive Language Models (LLMs) for the Automation and Improvement of the TRIZ (Innovative Problem-solving) Approach”

The Theory of Inventive Problem Solving (TRIZ) is a widely recognized method of ideation that uses the knowledge derived from a large, ongoing patent database to systematically invent and solve engineering problems. TRIZ is increasingly incorporating various aspects of machine learning and natural language processing to enhance its reasoning process. Now, researchers from both the Singapore…

Read More

Stanford University researchers have unveiled Octopus v2, a tool that enhances on-device language models for improved super agent operations.

Artificial intelligence, particularly large language models (LLMs), faces the critical challenge of balancing model performance and practical constraints such as privacy, cost, and device compatibility. Large cloud-based models that offer high-accuracy rely on constant internet connectivity, raising potential issues of privacy breaches and high costs. Deploying these models on edge devices introduces further challenges in…

Read More

Introducing RAGFlow: A Deep Learning Document Comprehension Engine that Uses Retrieval-Augmented Generation (RAG), Now Open-Source.

In the dynamic environment of Artificial Intelligence (AI), the constant challenge for businesses is managing immense quantities of unstructured data. To this end, the pioneering open-source AI project, RAGFlow is set to redefine the way organizations derive insights and respond to complex inquiries with remarkable truthfulness and precise accuracy. RAGFlow is an avant-garde engine that…

Read More

The Function of Transformers in Natural Language Processing – Training Large Language Models (LLMs) with Transformers, How Does it Work?

Transformers have revolutionized Natural Language Processing (NLP) with Large Language Models (LLMs), such as OpenAI's GPT series, BERT, and Claude series, etc. The advancement of Transformer Architecture brought about a new way of building models designed to understand and accurately generate human language. The Transformer Model was introduced in 2017 through a research paper titled "Attention…

Read More

Google DeepMind Introduces Mixture-of-Depths: Fine-Tuning Transformer Models for Adaptable Resource Management and Improved Computation Efficiency

The transformer model has become a crucial technical component in AI, transforming areas such as language processing and machine translation. Despite its success, a common criticism is its standard method of uniformly assigning computational resources across an input sequence, failing to acknowledge the varying computational demands of different parts of a data sequence. This simplified…

Read More