Jonathan Cachat
jcachat.bsky.social
Jonathan Cachat
@jcachat.bsky.social
Neuroscience PhD turned Data Scientist & Machine Learning Engineer
www.nature.com/articles/s41...

thought this was an interesting read, & a cool approach - perhaps a step in the right direction - but feel like it falls short of the goal they are after.

had never considered fine-tuning a foundational model on results of psychology studies.

🧵
A foundation model to predict and capture human cognition - Nature
A computational model called Centaur, developed by fine-tuning a language model on a huge dataset called Psych-101, can predict and simulate human nature in experiments expressible in natural language...
www.nature.com
July 7, 2025 at 11:15 AM
so, i have been seeing the AGI is here youtube videos the last 2-3 days....

finally watched @openai's Day 12, and it's worth the watch

but i dont think we are at AGI yet

OpenAI o3 and o3-mini—12 Days of OpenAI: Day 12 https://www.youtube.com/live/SKBG1sqdyIU?si=xbleruIEHH5tkdhC
December 28, 2024 at 11:49 PM
"Let's train the model on what humans look at & forget all this TA stuff."

I came to same conclusions 5 years back

https://research.google/blog/unlocking-the-power-of-time-series-data-with-multimodal-models/

Do y
ou have some ideas for financial analysis or other time-series data worth inve...
December 28, 2024 at 11:49 PM
Even with OpenAI & Google Gemini 2 announcements today, I still think this is my best discovery today: https://github.com/jimmc414/1filellm/tree/main

Prep
your knowledgebase (code, pdfs, URLs) for embedding and vector store, all-in-one, one-and-done
December 28, 2024 at 11:49 PM
$100k baby, we living on the moon
December 28, 2024 at 11:49 PM
here is a peek into my data science side-quests: profiling & visualizing my offroad RZR rides.

https://www.youtube.com/watch?v=GecPx3ippWE
December 28, 2024 at 11:49 PM
been waiting for a grounded, sources cited scientific literature model! of course it was Semantic Scholar & @AllenInstitute⁩ 🙌🏼🔬🥼

Ai2 OpenScholar: Scientific literature synthesis with retrieval-augmented language models https://allenai.org/blog/openscholar?utm_source=tldrai
December 28, 2024 at 11:49 PM
take a break from pumpkin spice to watch @ArmchairExpPod - it's intellectual gold 🥇

but, @harari_yuval - said "a failure of art to explain the reality of bureaucracy" - what about stand up?!

next book "intellectual history of political thought as expressed by comedy"?? 🤓
December 28, 2024 at 11:49 PM
just caught up on @latentspacepod, again 🥇🥇🥇🥇

Building AGI in Real Time (OpenAI Dev Day 2024) https://youtu.be/eOlJglULFnk?si=SDzHIc_itAyz--mp via @YouTube
December 28, 2024 at 11:49 PM
often asked about my current LLM stack, so posting my top MLops app list below:

http://big-agi.com / @enricoros - for comparing (beaming) and merging responses to same prompt.

@llama_index  - for granular control of "your" document repositories & how they are embedded

🧵
December 28, 2024 at 11:49 PM
Pretty Lights - Swirl Bridge tour 2024. all shows, all sets w/ video @ http://prettynights.com Brooklyn Night 2 Show -> starts strong, continues stronger NOLA Night 2 -> Keep that music simple (Dr John sample) is © https://soundcloud.com/prettylightslive/albums
December 28, 2024 at 11:49 PM
for the @PrettyLights fan's out there, if ya don't know about http://prettynights.com - now ya know. a gold mine of all sets from the #swirlbridge tour this year. every city had different sets, and custom mixes influenced by the local flavors - just gold, straight gold!
December 28, 2024 at 11:49 PM
One of the smartest convos happening out there now! Also, check out @Trevornoah & @sama convos the episode previous!

https://open.spotify.com/episode/5AHnN99DoFnajp5AupuxJE?si=ns9O6bdGRR-Bj5B69q0crQ
December 28, 2024 at 11:49 PM
December 28, 2024 at 11:49 PM
Subway Map / Circuit Diagram of #Human #Brain
December 28, 2024 at 11:49 PM
. @elonmusk - @lexfridman & @glukianoff have a FANTASTIC idea here.

Integration of Rhetorical Classifers on X, similar to like button.

https://youtube.com/clip/Ugkxs-k9ujVm5Uw4gHJV9gLEbMXNpnZZSm7x?si=NX6rwpw1zg-L9NWv

Hell
, imagine the value of a crowd sourced training data set of logic /...
December 28, 2024 at 11:49 PM
December 28, 2024 at 11:49 PM
The future is here. #movementpatternanalysis @polarisrzr #data-from-field
December 28, 2024 at 11:49 PM
"How well is it understood to do this dance of different variables for various kinds of growth to happen?" love this discussion @lexfridman @NeriOxman https://blog.jcachat.com/2023/07/07/dialectical-model-of-human-nature/
December 28, 2024 at 11:49 PM
after listening to @LangChainAI with @swyx & @hwchase17 - the overlap btw my Neuroscience & Data Science has emerged in the form of a LangChain Agent Schematic!

whos coming with me!
December 28, 2024 at 11:49 PM
check out this LLM profile of @jcachat from @ringerapps - its dead on!

https://app.birdbrain.bio/results/435c6900-3a4c-4ab1-8d3f-08180dbbff97
December 28, 2024 at 11:49 PM
my last few wks?? Data Analysis w LLMs -

@LangChainAI is a clear leader, & easy to get up. next @llama_index & singularity

intro to topic @
https://ashukumar27.medium.com/dolly2-and-langchain-a-game-changer-for-text-data-analytics-7518d48d0ad7

flex
ibility to work with cognitive age...
December 28, 2024 at 11:49 PM
Loving this @badfriends_pod episode, @bobbyleelive + @joshpeck & juicy = ❤️ long time, first time callin from @CheetoSantino 👌🏼
December 28, 2024 at 11:49 PM
Went over to On Purpose w @jayshetty post @ArmchairExpPod, great (ancient) wisdom served. My favs:

“Even black & white TV, is mostly gray”

“If you are looking for new ideas, read an old book”

“Us & We, not you & me”

🤞🏼for the encore, great convo guys.
December 28, 2024 at 11:49 PM
December 28, 2024 at 11:49 PM