Gav

Dad. Husband. Biohacker. Threads PM. 4 years sober (choice not necessity). 7x great-grandson of a forgotten American founder. follow = perspective ≠ endorsement

[Gav @GavSays]

ChatGPT Paid Users: give this Best Self GPT a whirl and lmk what you think!

https://t.co/X9e3d1q7di

[Andrej Karpathy @karpathy]

LLM Knowledge Bases

Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So:

Data ingest:
I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki includes summaries of all the data in raw/, backlinks, and then it categorizes data into concepts, writes articles for them, and links them all. To convert web articles into .md files I like to use the Obsidian Web Clipper extension, and then I also use a hotkey to download all the related images to local so that my LLM can easily reference them.

IDE:
I use Obsidian as the IDE "frontend" where I can view the raw data, the the compiled wiki, and the derived visualizations. Important to note that the LLM writes and maintains all of the data of the wiki, I rarely touch it directly. I've played with a few Obsidian plugins to render and view data in other ways (e.g. Marp for slides).

Q&A:
Where things get interesting is that once your wiki is big enough (e.g. mine on some recent research is ~100 articles and ~400K words), you can ask your LLM agent all kinds of complex questions against the wiki, and it will go off, research the answers, etc. I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files and brief summaries of all the documents and it reads all the important related data fairly easily at this ~small scale.

Output:
Instead of getting answers in text/terminal, I like to have it render markdown files for me, or slide shows (Marp format), or matplotlib images, all of which I then view again in Obsidian. You can imagine many other visual output formats depending on the query. Often, I end up "filing" the outputs back into the wiki to enhance it for further queries. So my own explorations and queries always "add up" in the knowledge base.

Linting:
I've run some LLM "health checks" over the wiki to e.g. find inconsistent data, impute missing data (with web searchers), find interesting connections for new article candidates, etc., to incrementally clean up the wiki and enhance its overall data integrity. The LLMs are quite good at suggesting further questions to ask and look into.

Extra tools:
I find myself developing additional tools to process the data, e.g. I vibe coded a small and naive search engine over the wiki, which I both use directly (in a web ui), but more often I want to hand it off to an LLM via CLI as a tool for larger queries.

Further explorations:
As the repo grows, the natural desire is to also think about synthetic data generation + finetuning to have your LLM "know" the data in its weights instead of just context windows.

TLDR: raw data from a given number of sources is collected, then compiled by an LLM into a .md wiki, then operated on by various CLIs by the LLM to do Q&A and to incrementally enhance the wiki, and all of it viewable in Obsidian. You rarely ever write or edit the wiki manually, it's the domain of the LLM. I think there is room here for an incredible new product instead of a hacky collection of scripts.

[Gav @GavSays]

@sotally_tobers Shoulda

[Gav @GavSays]

Just ate a pound of a few days past sell by turkey with a heap of over a week old homemade pesto. Ask me anything

[Gav @GavSays]

Man @JooliesDates are insane!

Nature’s @krispykreme

[Darshak Rana ⚡️ @thedarshakrana]

https://t.co/xbbByL7nE1

[Gav @GavSays]

Next time you think you’re enlightened try to have a real conversation with @AmazonAlexa from across the room

[DAN KOE @thedankoe]

https://t.co/7l7Jef99QZ

[Gav @GavSays]

Macro sense of urgency.

Micro sense to stop and smell the roses.

Put in your daily work.

Don’t let it get in the way of happiness.

You have time ☯️.

(Things I’d tell my younger self)

[Gav @GavSays]

If you peel the onion back on most political views, you’ll find eventually they’re usually self-serving.

It takes work to separate from yourself enough to have a “best for all” view.

Anyone that does this work usually ends up without a clear party they call theirs.

👋🏻 It me

[Gav @GavSays]

The further from your best self you are on your deathbed, the more regret you'll feel.

The ultimate fuck around, and find out, if you will.

[Gav @GavSays]

Being busy isn’t a flex (I use to think it was, too)

The biggest flex is being so on top of your stuff that you always have extra bandwidth to jump in and help others

The biggest flex is doing good for others because you can

[Gav @GavSays]

Idk who needs to hear this but your anxiety goes away when you shine the light of attention on it.

Gnarly work project? Dig in.

Person making you uncomfortable? Talk to them.

Etc..

[Gav @GavSays]

Might need to start tweeting again. Found some old gold

[Gav @GavSays]

When you hear never stop learning.. that applies inward too.

Keep getting to know yourself better.

Love yourself even when you don’t like what you find.

Adjust.

That’s where the growth is.

[Gav @GavSays]

Rescinding my prior @drinksanzo recommendation it seems they’ve added sugar among other changes..

[Gav @GavSays]

Dear @Costco loyalty product manager out there - my soon to be 70 year old Dad just said to my Mom

“Costco did something new that I’m absolutely thrilled about

Executive members get to shop at 9am”

You go baby

[Gav @GavSays]

Let it be known - I was a big em dash guy pre chatGPT.

I will not be changing my writing style!

[Gav @GavSays]

Goat already with Catan in the shot

[Seattle Seahawks @Seahawks]

It all starts up front. With the 18th pick, we've selected G Grey Zabel!

#GoHawks x @lumentechco

[Gav @GavSays]

@bengreenfield Dope! Thanks for all the great content!

[Gav @GavSays]

@bengreenfield did I just see you at SeaTac airport? Didn’t want to disturb I think you were on a call

[Gav @GavSays]

Parlor Palms

[Tim Urban @waitbutwhy]

I made the mistake of putting a CO2 meter in my office where I am all day and now I'm aware that I'm sitting in a cloud of CO2 all day losing IQ points. Even with the window open, the level is often 1300+. Anyone know of a way beyond ventilation to reduce a room's CO2?