Discover AI Courses

Nuclear Power AI: Reinforcement Learning vs Pre-Training for High-Risk Technical Environments

Explore how reinforcement learning amplifies pre-trained behaviors in AI systems, with implications for high-risk environments like nuclear power plants, based on Harvard research.

• YouTube
• 29 minutes
• Self-Paced
• Free Video

Agent2Agent and MCP Protocol in Multi-Agent AI

Explore Google's Agent2Agent Protocol and its compatibility with MCP protocol for tool use by LLMs, featuring cross-ecosystem compatibility with LangGraph and crew.ai, plus Python code examples and JSON schema details.

• YouTube
• 23 minutes
• Self-Paced
• Free Video

Reasoning LLM Overthinking - Easy Solution

Discover a simple, free solution to improve distilled reasoning LLMs that tend to overfit and underperform, based on new research findings about enhancing reasoning capabilities with less overthinking.

• YouTube
• 29 minutes
• Self-Paced
• Free Video

A Framework for Objective-Driven Dynamical Stochastic Fields

Explore a framework for objective-driven dynamical stochastic fields in quantum AI, presented by Stanford researchers Zhang and Koyejo.

• YouTube
• 47 minutes
• Self-Paced
• Free Video

Multi-Agents Become Smarter: The AI Dream Team

Explore how multi-agent systems evolve through reinforcement learning, fine-tuning, and complex reasoning to create more intelligent AI collaborations.

• YouTube
• 36 minutes
• Self-Paced
• Free Video

Value-Aware Policy Optimization for Advanced Reasoning Tasks

Explore how ByteDance and Tsinghua University researchers evolved DeepSeek's GRPO into DAPO and VAPO algorithms with four key techniques for efficient and reliable reinforcement learning in advanced reasoning tasks.

• YouTube
• 26 minutes
• Self-Paced
• Free Video

Llama 4 Scout: 10M Token Context Length Explained

Discover how Llama 4 Scout achieves 10M token context length through innovative softmax scaling and optimized RoPE/NoPE layer configurations, and evaluate its reasoning capabilities.

• YouTube
• 29 minutes
• Self-Paced
• Free Video

LLMs at Their Breaking Point: Performance Analysis and Model Selection - Part 1

Gain insights into LLM model selection, performance metrics, and scaling limitations through practical examples and mathematical analysis of pass@k probability in AI systems.

• YouTube
• 33 minutes
• Self-Paced
• Free Video

GraphCHECK: Improving Factuality in LLM Outputs with Graph Neural Networks for Knowledge-Graph Enhanced Verification

Discover how GraphCHECK combines LLMs, Knowledge Graphs, and Graph Neural Networks to enhance factuality verification in AI outputs, with practical Python implementations and research insights.

• YouTube
• 30 minutes
• Self-Paced
• Free Video

EASY: Multi-LLM Protocol for Local and Cloud AI - MinionS

Discover EASY protocol for cost-efficient collaboration between on-device and cloud language models, integrating local PCs with open source LLMs to achieve premium performance at reduced costs.

• YouTube
• 25 minutes
• Self-Paced
• Free Video

The CoT Black Box: Beyond our Knowledge - AI Reasoning We Can't Decode

Delve into the mysteries of Chain-of-Thought (CoT) reasoning in AI models like OpenAI o1 and DeepSeek-R1, exploring the current limitations in understanding how these Large Reasoning Models actually work.

• YouTube
• 27 minutes
• Self-Paced
• Free Video

The AI Reasoning Lie: Debunking Logical Intelligence Emergence in LLMs

Uncover the truth about reasoning in Large Language Models (LLMs), exploring research that challenges assumptions about emergent intelligence in AI systems like o1 and o3.

• YouTube
• 38 minutes
• Self-Paced
• Free Video

Comparing SONNET 3.7 Extended Thinking 32K and DeepSeek R1 Logic Performance

Explore performance comparisons between CLAUDE SONNET 3.7 Extended Thinking 32K and open-source DeepSeek R1 through advanced logic tests, plus see how test-time-compute scaling models compare to classical AI models.

• YouTube
• 13 minutes
• Self-Paced
• Free Video

AI Mathematics Explained - The Easy Way: Reinforcement Learning and Multi-Agent Systems

Discover the mathematics behind reinforcement learning, focusing on multi-agent RL and meta-thinking concepts like REMA, explained in a clear, accessible manner for AI enthusiasts.

• YouTube
• 52 minutes
• Self-Paced
• Free Video

Multi DeepSeek R1: Learning to Reason with Multimodal Large Language Models via Step-wise GRPO

Explore how StepGRPO's step-wise rewards enhance reasoning in multimodal language models, improving accuracy and validity through continuous feedback rather than passive imitation, as demonstrated across multiple benchmarks.

• YouTube
• 23 minutes
• Self-Paced
• Free Video