Subnostr

John

bcbf9644d3f475d00eb9c6e467385ce16d4546c1a24222ccfa542bf776eaba95

ML Engineer (Former) attorney

https://www.freedomlawschool.org/w4-withholding-package

Looking in to this, feels off. Like if I was an employer someone doing this might be too much trouble, and I'd just fire them

Need to dig into the law.

To summarize, it was mostly a bunch of people who have no clue what crypto or bitcoin is talking about leading innovation in digital assets. lol. The last reporter’s question brought in the only mention of bitcoin. She asked specifically about Trump’s campaign promise of a bitcoin reserve. The answer was that the feasibility of a bitcoin reserve is the first thing they will be looking into, and mentioned that it was something separate from the sovereign wealth fund.

Another notable quote to demonstrate their level of competency: “You’ve got your actual crypto, like Bitcoin, but then what it runs on is blockchains, and there’s a lot you can do with that new computing technology.” LMFAO these people have no clue wtf they’re talking about and they’re going to “lead” in the space 🤣

nostr:note1ury9xcyyav40ygt6rnalasxjv0xeuku0c6jt53vlfhfe3qjetgqqe7lywl

There's a lot you can do with a really slow database

nostr:nprofile1qqsyvrp9u6p0mfur9dfdru3d853tx9mdjuhkphxuxgfwmryja7zsvhqpzamhxue69uhhv6t5daezumn0wd68yvfwvdhk6tcpz9mhxue69uhkummnw3ezuamfdejj7qgwwaehxw309ahx7uewd3hkctcscpyug I wish links to tweets shows the content somehow, visiting twitter makes me feel dirty

The current side project is an LLM based law school tutor.

The LLM generates essays, grades essays, gives mid-essay hints/conversations.

I'm adding rag based on course notes.

The current issue is how best to incorporate the course notes to DBs so that I can pass the relevant notes to the LLM during essay prompt generation.

Courses are divided into topics, sub topics etc. I want to log which essays are generated that target each topic so that you ensure complete coverage. And grade essays and record performance for the same.

This way I can add an Anki-style spaced repetition but via essays.

The big items I need to solve are how to parse course notes into the database, how to structure it generally, and the Anki style stuff.

Is it just for me that goose will just refuse to answer, act, or do anything at all most of the times? Sometimes it answers, sometimes it shows a loading icon but then it vanishes and nothing happens, but most of the times not even a loading icon is shown.

nostr:nevent1qqsw7v47pwy7c4nsf2ms8gnv7a387nde2khdnr04n9nt0kan844lr8qpz4mhxue69uhhyetvv9ujuerpd46hxtnfduhsz9mhwden5te0wfjkccte9ehx7uewwdhkx6tpdshs5quz20

I'm using OpenRouter and I've tried multiple different models. They all work _sometimes_.

I tried with Claude and had a similar experience, worked maybe half the time and my requests were simple like "go to this website and tell me something"

YouTube's algorithm is so great that it never ever shows me channels I'm subscribed to, but fills my homepage with videos from channels I've browsed in the past 3 days, including many things I've already watched, and completely forgets everything I've done before that. That's the amazing winner UX we should be aiming at!

Use freetube/newpipe/tubular

Please trump announce aliens cracked sha256

“You’re really not scared of this dip?”

No darling, are you?

I'm praying it lasts till payday (it never does)

DETERMINISTIC OPTIMISM 🌞

Openai is probably killing web ads.

When you ask anything you bypass 20 websites you'd have open to research something and murdered those impressions.

Instead of sponsored content it will be sponsored training data.

SEO will be replaced with Weights optimization // bribing to be weighted higher

Can somebody give me a tldr of why people here are so hyped about #cashu and not #fedimint? Or am I reading the room wrong? #asknostr

Fediment feels like a nothingburger, there was a lot of hype like 8 months ago when the fedi guy made podcast rounds but there's nothing that's come of it. Meanwhile with cashu theres constant new use cases/demos of shit on cashu. It's basically devs stuffing cashu into everything

Please stay dipped until payday

Extremely excited for p2pk

3090s are appreciating lmao

New voice cloning SOTA model?

reference: https://voca.ro/1iZeFEtrvjUg

output: https://voca.ro/1bhowitux8wp

More info: https://huggingface.co/HKUSTAudio/Llasa-8B

Damn it's really over this time

I should make an AI curated feed, feels wasteful tho. Like you feed every note through an llm asking if it's good enough for the current feed.

Can select for positivity/insight/shitposts or whatever

Let's say it's 2035, 10 years in the future. Nostr has won. Describe what do you do on your favorite Nostr client. 👇

My diy neuralink feeds shitposts straight into my brain generated by my 9090 GPU cluster, bangers are automatically reshared to ai agents and popular ones will net 10-20 SATs, which should be enough for a week's groceries

Walked 5 miles in the woods and listened to iron john

Bjj was fun, kimura still OP, I'm outta shape tho, met some cool folks

I feel like this has broader implications...

The DOJ gave three rationales for the argument that "Money is not necessarily 'property' for constitutional purposes": (1) the government creates money, so you can't own it; (2) the government can tax your money, so you don't own it; and (3) the Constitution allows the government to spend money for the "general welfare."

https://reason.com/2025/01/31/the-government-says-money-isnt-property-so-it-can-take-yours/?utm_source=twitter&utm_medium=social&utm_campaign=reason_brand&utm_content=autoshare&utm_term=post

Lmao this is retarded

#1) real estate is created and enforced by government, so that isn't property either?

#2) you own nothing??

Load More