Home

Today, the most powerful large language model is GPT-4, developed by OpenAI, which has even outperformed the average human in many tests. What’s the secret behind GPT-4? Although OpenAI has not disclosed the model’s specifics, they have scaled the GPT-3 model 100 times to GPT-4, totaling 1.76 trillion parameters. However, training such a large-...

On Tesla’s AI Day (September 30th, 2022), the company unveiled its first humanoid robot, offering a glimpse into the future of AI-powered robotics. Leveraging advanced perception techniques from its self-driving architecture, Tesla has integrated this technology into its robotics. Prior to Tesla’s breakthrough, Boston Dynamics captivated...

Rainbow is a method to combine all six improvements in Deep Q-Network(DQN). Since the emergence of DQN, scholars have proposed various improvements to it. This post will focus on six of the major improvements, which are eventually integrated into the Rainbow model. Deep Q-Network As I introduced in my previous post, the Dynamic Programming, Mo...

In the real world, many problems can be formed through Reinforcement Learning(RL). A typical RL problem generally consists of an agent that makes the decision and the environment where the agent is surrounded. This post will go over some important basics from Sutton and Button’s Reinforcement Learning book and introduce several of the classic RL...

Hanabi(花火) originally means “fireworks” in Japanese. As a common Japanese cultural symbol, it was used by a French game designer Antoine Bauza to create a forward-thinking card game. Unlike most of the popular competitive card games on the market, Hanabi game is designed as a cooperative game. The rules of the game are simple, yet cleverly cons...

Welcome to Tchen’s personal blog. In this blog, I will upload some useful posts sharing food 🍱🍲🥘 and life experiences 🎞. In addition, I will also share some tech miscellanies📱 and interesting tools 🛠(including tutorials). The original purpose of this blog was to share some of the technology I’ve been learning about. I am a researcher focusing ...