All Episodes

Shadow AI

Meta AI's V-JEPA 2: World Models for Understanding and Planning

NovelSeek Autonomous Scientific Research Framework

Qwen2.5-Math RLVR: Learning from Errors

AlphaEvolve: A Gemini-Powered Coding Agent

OpenAI Codex: Parallel Coding in ChatGPT

Agentic AI Design Patterns

Machine Learning for High-Risk Pregnancy Prediction

AI Mobile Edge Offloading for QoE and Energy Efficiency

Blockchain Chatbot CVD Screening

Deep Learning for Mammographic Breast Density Prediction

RLHF for Large Language Model Fine-Tuning

UB-Mesh: Advancing LLM Training Infrastructure

FASTCURL: Reinforcement Learning for Enhanced AI Reasoning

National AI for Cardiovascular Care: Nature Medicine Analysis

Vision-Language Reward Models: Advancements and Benchmarking

Advancing Vision-Language Reward Models

Mix-LN: A Hybrid Normalization Technique

Direct Q-Function Optimization for LLMs

RAG Attacks on LLMs

SmolAgents: AI Agents in Few Lines of Code

ByteDance's 1.58-bit FLUX AI Model

HuatuoGPT-o1: Advanced Medical Reasoning

FDA Authorizes AI Sepsis Detection Tool

Safe and Efficient Agentic AI

Apple's AI Strategy

Mix-LN: Hybrid Normalization for Transformers

LOTUS 1.0.0: Open-Source Query Engine

OpenAI o3: A Measured Advancement in AI Reasoning

LLM Alignment Faking: A New Threat

TOMG-Bench: A New AI Benchmark for Molecule Generation

Multi-Agent AI Frameworks

Alibaba vs. OpenAI: The AI Race Heats Up

Gemini 2.0: AI Research Assistant Capabilities

Maya: An Open-Source Multilingual AI Model

EXAONE 3.5: Enhanced Bilingual AI

Hugging Face TGI v3.0: Faster Text Generation

Density: A New Metric for Evaluating LLMs

Snowflake's Arctic Embed 2.0

ALAMA: Adaptive Language Model with Auxiliary Memory

Alibaba's AI Challenge to OpenAI

Building Effective AI Agents

Evaluating and Improving LLMs: Four Novel Approaches

AI Scientists: Revolutionizing Scientific Research

TamGen: AI for Antibiotic Discovery

SEALONG: Extending LLM Context Windows

AI Unveils Hidden Climate Extremes

Microsoft GraphRAG: Revolutionizing Data Analysis

Meet OpenCoder: A Completely Open-Source Code LLM Built on the Transparent Data Process Pipeline and Reproducible Dataset

New Scaling Laws for Optimizing Model and Dataset Proportions in Behavior Cloning and World Modeling Tasks

NVIDIA Launches LLaMA-Mesh, a Unified 3D Mesh Generation Method Using LLMs

BLIP3-KALE: An Open-Source Dataset of 218 Million Image-Text Pairs Transforming Image Captioning with Knowledge-Augmented Dense Descriptions

A Robust AI Solution for Managing Memory Constraints and Improving Classification Accuracy in Transformer-Based NLP Models

This AI Paper by Inria Introduces the Tree of Problems: A Simple Yet Effective Framework for Complex Reasoning in Language Models

Is Your LLM Agent Enterprise-Ready? Salesforce AI Research Introduces CRMArena

Databricks Mosaic Research Examines Long-Context Retrieval-Augmented Generation

RT-Affordance: A Hierarchical Method that Uses Affordances as an Intermediate Representation for Policies

Researchers at Peking University Introduce A New AI Benchmark for Evaluating Numerical Understanding and Processing in LLM

FrontierMath: The Benchmark that Highlights AI’s Limits in Mathematics

Databricks Mosaic Research Examines Long-Context Retrieval-Augmented Generation: How Leading AI Models Handle Expansive Information for Improved Response Accuracy

UniMTS: A Unified Pre-Training Procedure for Motion Time Series that Generalizes Across Diverse Device Latent Factors and Activities

Meet Hawkish 8B: A New Financial Domain Model that can Pass CFA Level 1 and Outperform Meta Llama-3.1-8B-Instruct in Math & Finance Benchmarks

MiniCTX: Advancing Context-Dependent Theorem Proving in Large Language Models

How TrigFlow’s Innovative Framework Narrowed the Gap with Leading Diffusion Models Using Just Two Sampling Steps

MathGAP: An Evaluation Benchmark for LLMs’ Mathematical Reasoning Using Controlled Proof Depth, Width, and Complexity for Out-of-Distribution Tasks

Can LLMs Follow Instructions Reliably? A Look at Uncertainty Estimation Challenges

Zhipu AI Releases GLM-4-Voice: A New Open-Source End-to-End Speech Large Language Model

Meta AI Researchers Introduce Token-Level Detective Reward Model (TLDR) to Provide Fine-Grained Annotations for Large Vision Language Models

Google Researchers Introduce UNBOUNDED: An Interactive Generative Infinite Game based on Generative AI Models

This AI Paper Explores If Human Visual Perception can Help Computer Vision Models Outperform in Generalized Tasks

Microsoft To Launch 'AI Agents' to Help You Handle Routine Tasks

IBM unveils new open source AI ‘Granite 3.0’ models for business

Refined Local Learning Coefficients (rLLCs): A Novel Machine Learning Approach to Understanding the Development of Attention Heads in Transformers

This AI Research from Cohere for AI Compares Merging vs Data Mixing as a Recipe for Building High-Performant Aligned LLMs

Are Brains and AI Converging?—an excerpt from ‘ChatGPT and the Future of AI: The Deep Language Revolution’

CREAM: A New Self-Rewarding Method that Allows the Model to Learn more Selectively and Emphasize on Reliable Preference Data

Embed-then-Regress: A Versatile Machine Learning Approach for Bayesian Optimization Using String-Based In-Context Regression

Graph-Constrained Reasoning (GCR): A Novel AI Framework that Bridges Structured Knowledge in Knowledge Graphs with Unstructured Reasoning in LLMs

MMed-RAG: A Versatile Multimodal Retrieval-Augmented Generation System Transforming Factual Accuracy in Medical Vision-Language Models Across Multiple Domains