I love science and engineering with a passion, it is what I think about and work on purely for the awe and wonder of it. My interest and work focuses on natural language understanding, conversational agents and reinforcement learning. My path to artificial intelligence is non-traditional, prior to my career in machine learning I was a practicing ophthalmologist. Today I am tackling the the continual learning, controlled generation and inverse reinforcement learning problems that could be used to build more human like AI
Will ChatGPT Do X? Build Your Own Mini-ChatGPT and Then Decide for Yourself
Speakers: Carson Lam
The talk consist of implementing your own reinforcement learning from human feedback large language model in your own colab notebook, then, discussing what you might expect to see when such an agent is scaled up to the sizes of GPT3.5 and beyond using some of the literature on what exactly emerges with scaling up the parameters and training data.
- Not Interested