I recently watched Andrej Karpathy's excellent "Deep Dive into LLMs like ChatGPT." LLM performance depends greatly on the initial dataset. This led me to wonder:
How can the Rails community enhance the public Ruby on Rails datasets used by LLMs?
@TateHackert Want to put someone’s bias for action on steroids, remove the possibility of the outcome being perceived as a failure by themselves or others.
@TateHackert Thinking out loud but I think bias for action or lack thereof is related to a need for perfection, need for perfection is related to fear of failure, fear of failure is related to caring about others’ opinion of you and more importantly one’s inner critic.
@TateHackert Maybe a stretch but I read an interesting case study that actually recommended to focus on the extreme ends. i.e. in your example it might be focus on the extreme power user AND the extreme novice. The middle will take care of itself. https://t.co/ML338ALhWc
@Shpigford Curious where you'd put the clipper in that priority list? I feel no app has matched Evernote's clipping capability even 10 years later. Second for me was killer OCR (handwriting, driving by a sign take a photo and it would all show up in search)
@TateHackert@Austen 1. What would he say to the guy who asks “I can launch one of these in 3 months what’s the big deal?” i.e. what are the hard fought lessons that he could only have learned having been in it for this long?
2. What do people underestimate about schools? education? learning? jobs?
Today on #CollisionsYYC: Discover how @ZayZoon empowers employees to access earned wages ahead of time, breaking free from the paycheck-to-paycheck cycle. Gain insights into Zayzoon's 5x revenue growth and mission with @dtuer: https://t.co/vYNzYyfZdl
@dhh@Dell would be great to read how you approach these purely technical projects as compared to the large amount @37signals has already shared about product development via ShapeUp