It's still early days for repsubsamp--while I have plans to publish it as a complete software package on CRAN, I have reached the point in its development where I wanted to share it with the world and get some feedback. If you give it a try, I hope you enjoy it!
What do you do when you need the smallest-necessary "representative" subsample from a dataframe? I'm pleased to share repsubsamp: a data analysis tool for R that I've been working on over the last-year. (currently "in beta") https://t.co/Cysu28yiyl
While we anxiously await the moment we can publish and share our paper, today I wanted to share the software behind it. Repsubsamp can take any dataframe with metadata columns, find the "best" subsample from many candidates, and benchmark it against the population data.
@aaronbrand20469@canna_brain@talktodenene Hubermann frequently conflates dopamine signaling, bioavailability, and release. Example, in the WM episode, he cites dop. bioavailability as related to WM span, but then prescribes practices that stimulate dopamine release, which is putting the cart before the horse.
@canna_brain@aaronbrand20469@talktodenene It’s not even theoretical—it’s anti-theoretical. A lot of it boils down to “more dopamine good”, a message the RL/dopamine field has perpetually been struggling to undo.
He sure does spin it to promote his supplements though.
Funny remark from Dr. Tim Holy in seminar: “the nasal cavity is not only important for olfaction, but also appears to be important for this very important function called ‘breathing’”
@NonLocalityGuy Yeah, this has been a problem since the beginning. If you ask for an essay with citations, it will do the same thing. As if it is simply looking at templates of related citations and generating new ones by predicting likely words that would belong.
@SenguptaLab In general I don’t think you have much to fear at the moment. Any effort plagiarizing chatGPT on this assignment is going to come out quite bad.
@SenguptaLab Currently, ChatGPT isn’t great with length or deep insight. When I use it, I can ask it for outlines on a topic. Then ask it to flesh out paragraphs, and at the end proofreading. It’s 👍 for speeding up busy work, but good logic, pathos & insights still need to come from within
(7) From talking to friends in other fields: chatGPT can suggest a detailed (and accurate) patient management plan for complex scenarios. It can draft a competent (if rather standard) supreme court amicus brief. The possibilities seem endless.
Thread: over the last week, I've been exploring how new sophisticated chatbots like @OpenAI#chatGPT could revolutionize #neuroscience work. Read on to see what I've done so far:
(6) That being said, I echo the sentiment that chatGPT signals a tech revolution comparable to Google or the iPhone. Advanced AI assistants can make us to be faster and more efficient than ever before, and it will be important for us to find how to get the most out of them.