Computational Benefits and Limitations of Transformers and State-Space Models
Explore mechanisms enabling retrieval, copying, and length generalization in language models. Compare Transformers' excellence with LSTM and state-space models' limitations in sequence modeling tasks.
IM-RAG: Multi-Round Retrieval-Augmented Generation Through Learning Inner Monologues
Explore multi-round retrieval-augmented generation through inner monologue learning, enhancing AI's ability to process and generate information iteratively.
Evaluating Fairness in ChatGPT - OpenAI Research Overview
Explore OpenAI's latest research on assessing fairness in ChatGPT, gaining insights into AI ethics and evaluation methodologies for large language models.
Interpretability for Deep Learning: Theory, Applications and Scientific Insights
Dive into deep learning interpretability, exploring methods to enhance transparency, safety, and trustworthiness in AI deployment while uncovering scientific insights in various domains.
How Neural Networks Learn Features from Data
Uncover the mechanism behind neural networks' feature learning through the average gradient outer product (AGOP). Explore its applications in various architectures and its role in developing interpretable models for scientific use.
Controlling Distribution Shifts in Language Models: A Data-Centric Approach
Explore data-centric approaches to control distribution shifts in language models, focusing on algorithmic filtering and domain-specific data synthesis for improved generalization.
When Calibration Goes Awry: Hallucination in Language Models
Explore the phenomenon of hallucinations in language models, their connection to calibration, and potential mitigation strategies in this insightful talk by Adam Kalai from OpenAI.
Fine-tuning Gemma for Non-English Languages
Dive into Gemma post-training techniques and explore adaptation strategies for non-English languages, enhancing multilingual AI capabilities.
Improving Customer Experiences with Speech to Text and Text to Speech
Explore speech-to-text and text-to-speech technologies, custom voice creation, and real-world applications. Featuring discussions on TypeScript and JavaScript with special guest Anders Hejlsberg.
Unlocking the Power of OpenAI for Startups - Microsoft for Startups
Explore Azure OpenAI Service and GitHub Copilot for startups. Learn to deploy GPT-3, fine-tune language models, and leverage AI-powered code suggestions for various use cases in this Microsoft presentation.
Everything You Wanted to Know About Machine Learning but Didn't Know Whom to Ask - Sanjeev Arora
Explore machine learning fundamentals, from text classification to image recognition, with insights on optimization, unsupervised learning, and language models in this comprehensive seminar.
Building Intelligent AI Infrastructure with ORI - Dynamic Query Routing and Model Management
Discover how ORI dynamically routes AI queries to optimal language models, featuring chain atomic queries, benchmark matching, and ethical frameworks integrated with Ray Serve and vLLM for high-speed inference.
How Do Transformers Work? - A Deep Dive into Neural Network Architecture
Dive into the fundamental mechanics and architecture of transformer models, exploring their key components and operational principles through expert insights from MIT's research.
Edit Distance Robust Watermarks - Beyond Substitution Channels in Language Models
Delve into advanced watermarking techniques for language models, focusing on robust detection methods that withstand text modifications while maintaining cryptographic undetectability.
Distortion-Free Mechanisms for Language Model Provenance - Watermarking and Training Independence
Explore groundbreaking watermarking techniques and independence testing methods for language models, focusing on text generation provenance and model weight verification.
Defense Against Prompt Injection Attacks in LLM-Integrated Applications
Explore effective strategies for securing LLM applications against prompt injection attacks through structured queries and specialized instruction tuning techniques.
Learning to Reason - Insights from Language Modeling
Explore the fascinating intersection of language modeling and reasoning processes through Stanford researcher Noah Goodman's insightful analysis of cognitive learning mechanisms.
Szumna inteligencja - Czy można już panikować?
Analiza ChatGPT i jego wpływu na społeczeństwo. Omówienie budowy, konsekwencji i porównania z ludzką inteligencją. Czy sztuczna inteligencja to powód do paniki?
Les IA ne sont pas aussi intelligentes que vous le pensez
Découvrez les réalités des IA et des modèles de langage avec Yann Carbonne. Une exploration lucide et critique pour comprendre leurs capacités réelles et leurs limites actuelles.
Prompt Engineering Best Practices for Productive Customer Interactions
Explore effective prompt engineering strategies for natural language customer interfaces, focusing on RAG techniques to navigate diverse company content and address customer queries seamlessly.
Manipulating Chatbots Through Creative Language - Our Words Are Our Weapons
Entdecken Sie kreative Methoden zur Manipulation von Chatbots durch geschickte Wortwahl. Lernen Sie, wie Social Engineering bei KI funktioniert und wie Sie Sprachmodelle für Ihre Zwecke nutzen können.