yumraj

Got curious, searched, saw this on the Fathom page: “ Dealer sets selling price”, closed the page.

We’re going to see a repeat of Maverick with inflated prices which will kill the demand.

So we’re more worried about a kid’s social line than education and future employability. Got it.

Thinking back, different era and country, but fear of failing was much much bigger than not getting an A.

Currently there are no short term consequences to failing. Won’t this provide a motivation to a set of students to do better. The ones who’re going to fail are the ones suited to trade or other jobs, and other avenues later in like to get high school diploma. Maybe this is not the right time in their life and they need different help than being mindlessly promoted in school.

I wonder that even though I don’t have personal hardware to run these at the moment, if it makes sense to download and keep a copy of some models like these in case access/licensing changes in future and/or hardware becomes cheaper.

Can always replace the backup copy with a better one on next release.

Or is someone running an initiative like this, such as Archive.org folks?

I hate the whole concept of QR code for a menu. I have no idea who came up with the idea and why restaurants thought that is a good idea. Yes, physical menus cost a little money to print, but last forever. Unless the restaurant has a new menu every day, it just doesn’t make sense.

Maybe next time I’ll just walk out and let the restaurant know why I’m walking out.

According to Reuters which refers to this WSJ article [0], $30T is the TAM. Calling it potential revenue is a huge stretch.

Not justifying as this number as it is stupidly large, but theoretically if AI replaced all White collar workers in the World it will come to around $30T or so (Gemini said it’s $36T-$37T based on some calculations using global GDP, labor share, white collar share etc)

TAM is always stupidly large in decks, disconnected from reality. Every startup has a deck with TAM in billions.

[0] https://www.reuters.com/business/media-telecom/anthropic-exp...

I don’t think that’s peppermint. That’s pudina so spearmint. Mentha spicata.

It seems to aid in digestion, soothe upset stomach, so maybe it does.

There are two types of peppermint supplements being sold: mentha pipereta that this study looked at, and mentha arvensis.

I wonder if mentha arvensis has the same effect. Supplements vs mint tea would be another interesting thing to study in future.

Try telling the LLM that it is OK to disagree with you and you encourage that since the LLM is the expert or something like that depending on your context.

I do that while working with Claude code, all models (sonnet, opus and fable) and it frequently disagrees with me and tells me why it is disagreeing.

Now, I get that disagreeing is not saying no, but I believe guardrails do do that.

Why are mass weddings stupid? They save each person shit ton of cash and end result is the same.

I’ll argue that over the top destination weddings are the ones that are stupid. Save the money for the down payment for a home.

No it doesn’t.

Why? Because I don’t use it as a TV AND a set top box. I use it just as a TV that is connected to an AppleTV

And before when I was using the inbuilt smart TV, I had disabled various android services related to ads on my Sony TV, and run pi-hole that blocks ads.

I'll look into them, thanks..

Similar work flow.

I have to, at times keep reminding it if it did code review before committing, which it fails to do at times. Also, even a glance at times has revealed issues that I’ve had to discuss and undo etc.

But yes, it gets very tiring just having to click yes yes yes..

Makes sense. Actually to expand, I believe this can make a lot of sense for industrial robots and such which have a more or less fixed job and latency matters more, so a well tested model may be more valuable than need to keep updating them

How does that work, as in re-etching of silicon? Any pointers to read?

Given the fast churn of the models, how does it work out?

Won’t the silicon etched model already be 1 or more versions behind by the time the silicon comes out.

Though if it’s cheap enough, there certainly can be a market for cheaper model inferences.

There’s a difference between things getting incrementally better and a step function.

LLMs will obviously keep getting incrementally better, but in order to get the kind on jump that LLMs themselves were, the sentiment is that we need something more.

Recently I was playing around in this space and came across Dafny, rather I should confess that Claude code pointed be to it.

For software projects it seemed very approachable, even for a complete newbie like me. Probably not for Math though.

Any way to identify or block these proxy and ad click services in the router? Say a Ubiquiti or even pfsense?

I’m not using any of these boxes for especially this reason, but about 10-15 years ago had noticed my treadmill pinging a Chinese portal. I removed the WiFi access from the treadmill but am curious if there might be other devices.

Any specific ports, etc these guys use or are they mostly impossible to distinguish from regular internet traffic?

My another worry has been if these can monitor other Internet traffic, though I think HTTPS should mostly prevent that.

Does it not work if you don’t connect to Internet and not create an account?

Yahoo finance, Finnhub, Alpaca, Edgar.

I have been building few tiny financial apps for analyzing the market, domains and stocks to see if I can use them to make investment decisions.

Adding small tools to help understand specific option strategies, and once executed understand optimal action (exercise early, let it expire, wait)

Personal tools, for my use. No commercial angle in mind.

Russians had conducted similar experiments, Znamya, in the 1990s:

https://en.wikipedia.org/wiki/Znamya_%28satellite%29

This may be the reason why OpenAI reportedly delayed its IPO.

They might have had an inkling that this was coming.

I asked it to do code review of some personal projects, that will only run on my system, and it keeps giving me:

This model's safeguards flagged this message. This sometimes happens with safe, normal conversations. Switched to Opus 4.8. Send feedback with /feedback or learn more

Not sure if others are having more success.

This is an opportunity for MS to make a contrarian bet and keep supporting physical media. IMO they will benefit from acquiring gamers who want to keep using physical media.

Though, I think they will follow what Sony is doing.

LinkedIn, Twitter and tons of money

Such as?

Several bigger projects, more like startup ideas.

Few tools:

1) a combination of Python scraper and Claude skill to help family members find job by matching jobs to resume, to rank best fits

2) similar to above, but for stock data and financial news to identify movers and why they moved and see if anything is interesting.

3) a couple of attempts to import EPIC medical data from hospital into an offline app. Needs more work, data export from EPIC is crappy and a mix of images, pdfs, text, HTML and .jsonb files. Not useful at the moment

4) an application that downloads stock market data to run 15-20 strategies and back testing to identify stocks that match multiple and then run sentiment analysis on news feeds. Interesting, but semi useful. Needs lot more work.