Louis Mai

On Monday, 10 July 2023, I plan to work a bit on the technologies around ChatGPT. The technology I am talking about is Reinforcement Learning with Human Feedback (RLHF). In the simplest words, RLHF includes a policy network and a reward network. The reward network teaches the policy network by reinforcement learning. The concept of two networks is not new. Actor-Critic or Student-Teacher are some examples. The difference in ChatGPT is that, the reward network is taught by human feedback. The simple law: it is very difficult to product something, however, it is very easy to evaluate. For instance, it is so difficult to draw, but it is so easy to say which paints are more beautiful. Applying this simple law to RLHF: we the human train the reward network to know how to evaluate two summaries of an essay, while the policy network needs to learn how to generate the summaries. Then we expand to other question-answers. The new approach, RLHF, has at least two important revolutions: The summar...

This week since July 02, 2023, I plan to learn 2000 new French words in one week. Here is the summary of the project: Motivation: To escape the limbo of A2 level in French, by pouring my heart to this language. Detailed plan: accumulate new words or phrase from any source. Execution:  I follow the "grand plan" on Monday. I reach about 100 new words. I shift the plan on Tuesday, to learn by heart some French songs. Indeed, I learned by heart 3 songs. I shift again the plan on Thursday. Why? Because I found that I need to be exposed to the language (I mean, to listen to) as much as possible. So I put all effort on listening, news then later on movies. On Saturday, I added grammar and learning new vocabularies to the shifted plan. So my curriculum now is to listen to movies on Youtube while walking, and sometimes learning grammar or new words on newspapers.  Result: I would say that I have failed the project, however I have discovered the curriculum that is best suited for me. I...