Discover AI Creator Profile

Meta Reinforcement Fine-Tuning AI vs GRPO (CMU)

Discover AI March 13, 2025

Meta Reinforcement Fine-Tuning AI is a method that redefines how AI models optimize test-time compute by embedding rewards into each …

CoT to SoT: NEW Abstract AI Reasoning

Discover AI March 12, 2025

The video is about the video title 'CoT to SoT: NEW Abstract AI Reasoning' and the captions discuss the capabilities …

NEW: Topological AI for Reasoning (SOLAR by CMU)

Discover AI March 11, 2025

The video discusses the limitations of current AI reasoning models and introduces a new approach called Topological AI for Reasoning, …

New QwQ 32B: Perfect Long CoT Stability (TEST)

Discover AI March 10, 2025

The video showcases the Long CoT Stability feature on the Qwen QwQ 32B model, comparing its performance to a Gemini …

NEW L1 LLM w/ GRPO to LCPO for Scaling RL (CMU)

Discover AI March 9, 2025

The video is about a new Length Controlled Policy Optimization (LCPO) method for scaling reinforcement learning, specifically its application to …

AI 2.0: The New Core of AI is ...

Discover AI March 8, 2025

The video is about designing the next generation of AI, incorporating optimizations and new ideas from current LLM systems.

Scaling RL: 3B AI w Long Chain-of-Thought & 4 Patterns

Discover AI March 7, 2025

The video discusses new AI research studies on scaling RL, cognitive behaviors, and long chain-of-thought reasoning. These studies demonstrate that …

Beyond CoT: ReasonFLUX (Princeton Univ)

Discover AI March 6, 2025

The video is about a new AI research called ReasonFlux, which builds upon the Chain-of-Thought approach and uses a template …

The CoT Black Box: Beyond our Knowledge

Discover AI March 5, 2025

The video is about exploring the understanding of the Chain-of-Thought (CoT) Black Box, which is an AI reasoning process, and …

The AI Reasoning Lie

Discover AI March 4, 2025

The video is about AI reasoning lies regarding Large Language Models (LLMs), discussing insights from recent research on the absence …

SONNET 3.7 "Extended Thinking 32K": Exceptional LOGIC?

Discover AI March 3, 2025

The video compares the performance of the CLAUDE SONNET 3.7 Extended Thinking 32K and the DeepSeek R1 in logic tests, …

UNexpected: GPT-4.5 vs SONNET 3.7 vs R1 Reasoning

Discover AI March 2, 2025

The video compares the GPT-4.5, SONNET 3.7, and R1 Reasoning models, testing their causal reasoning performance, with a focus on …

LLM + Knowledge Graph + GNN = TRUTH by AI?

Discover AI March 1, 2025

The video discusses the power of LLMs, Knowledge Graphs, and GNNs in finding truth. It also mentions a code implementation …

EASY: Multi-LLM Protocol for Local & Cloud AI (Stanford)

Discover AI February 28, 2025

The video is about a new protocol called MinionS that enables cost-efficient collaboration between on-device and cloud language models, allowing …

Graph RAG Evolved: PathRAG (Relational Reasoning Paths)

Discover AI February 27, 2025

The video is about the latest development of Graph RAG (Relational Reasoning Graphs) model, specifically PathRAG, which shows significant improvements …

Better than AutoGen & LangChain: OctoTools (Stanford AI)

Discover AI February 26, 2025

The video discusses OctoTools, a novel agentic framework that addresses complex reasoning in Large Language Models and modern VLMs, providing …

How to Choose the Best LLM (Grok 3) for Your AI Agent

Discover AI February 25, 2025

The video discusses how to choose the best LLM (Grok 3) for an AI agent, highlighting the benefits of specialization …

Multi AI AGENTS w/ LLMSelector

Discover AI February 25, 2025

The video discusses a new AI tech framework called Multi AI AGENTS w/ LLMSelector, which combines multiple AI agents with …

S* for AI CODE Generation: Plus 100%

Discover AI February 24, 2025

The video discusses the S* framework for AI code generation, which substantially improves coverage and selection accuracy of LLM generated …

Google Stanford AI Co-Scientist: The SCIENCE YOU want

Discover AI February 23, 2025

The video discusses the concept of an AI co-scientist, a theoretical role that aims to combine the expertise of human …

Knowledge Graphs w/ AI Agents form CRYSTAL (MIT)

Discover AI February 22, 2025

The video is about knowledge graphs and how AI agents can store, organize, and reason about knowledge in a dynamic …

3 Logic TESTS of GROK 3 - THINK mode (hardest problems in math)

Discover AI February 21, 2025

The video discusses the testing of GROK 3, a new VLM, on three standard logic causal reasoning tests. The tests …

AI Computer Use: Why we need a REWARD VLM (ARMAP)

Discover AI February 20, 2025

The video discusses the need for a self-generating AI model, specifically a REWARD VLM (ARMAP), to optimize computer use by …

INSANE NEW TTS: Recurrent Depth (New Transformer)

Discover AI February 19, 2025

The video discusses the latest research on Test-Time Compute Scaling (TTS) using recurrent depth blocks that build a new transformer …

Discover AI