Andrew Chang

Building @runalph
My knowledge graphs @neode_ai

[Andrew Chang @achang1618]

@dasanil Immediately? Even with rule of 72, you had to divide twice and know ~2^8-2^9 off the top of your head. Impressive to those who get it. Magic to those who don't.

[Valeriano Velasquez @Valeriano1010]

What an inspiring night at AI Tinkerers Boston! From @achang1618 showcasing AI-built data apps to @vyahhi’s deep dive into MCP servers and @natea’s Minecraft AI build, we saw open source innovation at its best. Missed it? Subscribe for updates https://t.co/y6dzogRDN4 #AITinkerers

[Andrew Chang @achang1618]

@jeremyphoward @levelsio Yikes. I’m all for AI chat for question answering, but those results seem way too opinionated and biased for such a nuanced question. Lol at the “research ppls” too.

Also confused about the outrage over llms.txt? Seems obvious from the nature of LLMs..especially instruct models

[Andrew Chang @achang1618]

@judegomila Obligatory post of my favorite pro-thorium video by Kirk Sorensen https://t.co/ocr99DgeoC

[Andrew Chang @achang1618]

We're able to launch our hugging face notebook environment on 200+ GPU options from 10+ providers ranging from lambda labs to crusoe! Our mix ranges from 1xA100s to 8xH100 clusters 🚀

We built out an API to launch notebook servers that to make it easier to start using GPU tools like pytorch and transformers: https://t.co/HHFO6XXMTb

[Andrew Chang @achang1618]

Create a python networkx graph with one line of ...text?

@americandatasci's Agent Alph can create and run code in notebooks with your simple text directions. However, it's not perfect. Anyone see any issues with some of the data that GPT-4o provided in this tiny knowledge graph?

#notebooks #ai

[Andrew Chang @achang1618]

Excited to demo LLM-powered knowledge graph generation using @americandatasci at tonight's meetup at C10 Labs in Cambridge, MA!

We'll release the notebooks tonight along with info on what GPU environments they run in at https://t.co/ahplK11aVe

Check out @AITinkerers and the event details: https://t.co/G44QFvKXaO

[Andrew Chang @achang1618]

@jeremyphoward The number of blog posts I see criticizing notebooks for functions it wasn’t even meant to be used for is a testament to how useful it actually is 😂

[Andrew Chang @achang1618]

@judegomila Funny and maybe ironic because sigmoid activation functions were a commonly used layer for deep learning neural networks that allowed models to scale in capability by introducing nonlinearity.

Observable universe is finite so chained sigmoid progression may not be a bad thing

[Andrew Chang @achang1618]

@immad $5m FDIC insurance, that's 20x more than the typical personal checking account. Haven't seen any other banks making sweep networks with personal accounts this easy

[Andrew Chang @achang1618]

@judegomila One of my favorite fun facts to share when it pops up; I like to say in a room of 50 there’s >95% chance of two sharing the same birthday

[Andrew Chang @achang1618]

Proud to announce the release of our v2 dashboard at American Data Science.

https://t.co/Wd6odXUK6n

[Andrew Chang @achang1618]

🎉 It's launch day! 🎉
Our American Data Science Dashboard is now live on Product Hunt!

Effortlessly launch and manage Python notebook environments
https://t.co/TmE6Ul1Ehn

Try out the actual dashboard! https://t.co/dGKyQmv4Au

#DataScience #ProductHunt #Launch #JupyterLab

[Carnegie Mellon Chemical Engineering @CMU_ChemE]

Congratulations to @johnkitchin, who will receive the Award for Innovation in Chemical Engineering Education at the @ChEnected annual meeting in November!
https://t.co/tqhntKIzRR

[Dalton Caldwell @daltonc]

Smart people being compensated to make correct predictions within their areas of expertise can still be dead wrong. Even when (especially when!) they are acting in good faith.

In addition, it’s insidiously hard to make laws which seek to ensure that experts don’t make mistakes.

[Andrew Chang @achang1618]

@TheZachMueller @americandatasci @fastdotai @OpenAI @pinecone Thanks zach, hope you're doing well man. We've got HF environments too 😉

[Andrew Chang @achang1618]

Big life event, I've gone full-time on my startup at @americandatasci

You can try out our product which provides familiar pre-built development environments for tools like @fastdotai, @OpenAI, @pinecone, and more. Find more info at https://t.co/N8K1f901eJ

#AI #NLP #Jupyter

[Jeremy Howard @jeremyphoward]

There's a new programming language in town - it's Mojo! I'm more than a little excited about it. It's Python, but with none of Python's problems.

You can write code as fast as C, and deploy small standalone applications like C.

My post is below, and a 🧵
https://t.co/0IWGqcpEY7

[Andrew Chang @achang1618]

@grigoriy_kogan @judegomila Latency on pinecone is great; we're still testing it out at scale, but I'm very impressed. I've worked pretty extensively with in-house vector search tools like faiss and scann, so pinecone was a nice surprise

[Andrew Chang @achang1618]

Check out our new AI demo that leverages the @Golden Knowledge Graph to enhance fact retrieval processes with GPT-3! Our web3 knowledge graph is constantly under verification processes done by the open community...data quality and growth is constantly increasing #AI #NLP

[Jude Gomila @judegomila]

1/ I’d like to share a fun little demo we have been hacking on @golden

We attempt to answer factual questions in a more accurate way using the Golden Knowledge Graph by providing a retrieval enhancement to GPT-3.

Golden Retriever: https://t.co/X6Xwa0QMfe

Thread👇