I haven't written any logs for a while. Past few days I started catching up on
LLM technology, both for career and personal interest. Well mostly for career
development. I read the transformer and gpt papers - very interesting
actually. Later on I may start get hands dirty building the gpt2 toy model
from scratch. Or before that I may dig deeper into the transformer
implementation. For career benefit I probably should stay on the optimization
side, rather than modeling. From optimization point of view, the knowledge is
transferrable between llm and recommendation - NLP in different domains.

It's still useful to get familiar with the ML and modeling side, as it's
important to understand how the business domain is modelled and represented -
both LLM and recommendation, and maybe multi-media GenAI (far fetched). I'll
probably focus mostly in NLP.

On the side of optimization, probably invest time into lower level linear
algebra things and GPU programming. Then connect here with the transformer
technology.

***

On the life side, I'm not any better than previously. Life calmed down a bit.
Or more accurately, I'm tired of getting bothered. There were terbulant days,
still hard and exhausting. I'm getting old. I feel sad when thinking about it.