zxienin

Was exactly the first thought I had seeing post on HN. Glad to see the first comment from simon directly pointing it out

I use Abbott Freestyle Libre 3 and it has been deeply insightful about my progress with T2D. An increase in fasting glucose over 2-3 days tells me I need course correction (more movement, choice of food). I also noticed consistent correlation between bad sleep days and raised glucose levels over the day. Made me internalize how important sleep is.

I believe, next major uptick will come from CGM in wearables like Apple/Samsung watches [1][2]. I hope, even non diabetic folks use it to improve their health.

[1] https://www.macrumors.com/2023/09/14/apple-watch-blood-gluco... [2] https://www.macrumors.com/2024/01/23/samsung-racing-to-beat-...

> The assassins have carried out an action resembling an attack on a founding member of NATO and close ally of America’s. In North America.

Even assuming it is sanctioned killing, how would that be an attack on Canada?

My read on lot of justifiable criticism of langchain is, that it is rooted in trying not to lose out the lead it established since Nov 2022. Pace of updates in LLM space have been staggering.

Langchain hasn’t invested enough in quality of lib. A good lib needs good deliberation to build right abstractions that add value, then “get out of the way“.

Looking at the content list, all I need now is Brain - AI interface that “uploads“ all of it to my brain neural net.

</cheeky>

Strangely, no traction on this HN post in last 3 days.

None of Github posts explain, if GPT-4 is now being used instead of Codex for code generation cases. AFAIK, GPT-4 is not fine tunable, and GPT-4 is more general trained than for code. So I wonder how did they achieve a GPT-4 switch. Or maybe GPT-4 is only used for other use cases.

Can anyone share under the hood facts?

Surround yourself with positivity and gratitude.

A neighbour might need help with groceries or meds. Look around, see how you can help others. Don't expect anything back. Know that you made super tiny winy dent in the world. It's powerful energy. If early waking up is your thing, go out and look at dawn sun.

Simple things.

See anything negative? Toxic gossipers? switch off, skip. Period.

You are making choice to see and accentuate the positive around you. This is hard time for you - and it is finite.

Sounds like you have good thing going (your PhD). It's supposed to be hard. Get positive energy, buckle up and just run that course with support system in the research group.

Will it get 1M users using it by next week?

“Our 65B model is better performant than ChatGPT‘s 175B” is missing the point openai made in Nov'22.

That aside, this looks nice incremental research progress.

Not all calories are equal. Calories from carb rich food are metabolized different than calories from “healthy“ fat rich food. Metabolism of former also raises insulin + glucose in blood stream, which influences obesity more than later.

This [1] presentation by Robert Lustig is dense packed with quality insights.

Surrounding daily life with sugar, carbs is the root.

[1] https://youtu.be/dBnniua6-oM

I started reading and listening to Jason Fung’s explanation of body internal system [1]. What actually influences weight is hormonal cycles. I am experiencing positive results so far, working on this explanation.

Take steps to reduce not only amount of insulin presence in blood stream, but reduce its sustained presence. Reduce visceral fat.

The problem is solved by mix of correct nutrition, moderate physical lifestyle and intermittent fasting. Never by “dieting” or sustained calorific deficit.

Fasting should not be primarily thought of as reducing calories. Rather as creating longer gap inside body without insulin running though blood stream and toggling body to use fat as metabolism source. Weight loss (from fat reduction, not muscle) is secondary benefit.

[1] https://youtu.be/RL8x7FTSo-Y

Change of leadership is not necessarily = innovation

model: "text-davinci-003"

I was surprised at first, but I notice you're GPT-3 model, not chatGPT (no API so far)

Does this subscription gives unlimited prompt usage (no of tokens)?

Pretty well played by Hindenberg.

This rout directly benefits their short position.

Invest 2 year to create enough mayhem in market, to cash your chips. Not bad for a business model.

Just TL;DR such corporate news release with ChatGPT.

Is there an open source alternative to this, for personal usage at smaller scale?

This is a very helpful use case, specially for leaders who really want to bypass oddness and get unfiltered feedback. Such a feedback is the only way to stay in reality.

I am not sure why you made it (or started with) SMS. Why not simply use web app?

To the comments here on pricing. It can be largely ignored. I strongly expect OpenAI understands that high cost is barrier to true, widespread, productive adoption in real world. ChatGPT successors will see cost optimisation as key.

By the time, world moves from exploration phase to productive one, a “production ready“ ChatGPT will be out with saner cost base.

A side effect to this would be that it’ll offload a lot of chatGPT exploration traffic away from openai‘s chatGPT. That’s a good for OpenAI.

(speculation) OpenAI monetisation objective, business model and cash flow situation etc. would be different to Microsoft‘s.

OpenAI has indicated interest in generating revenue because compute costs are “eye watering“

And to apply for the service, you must divulge your use case in quality. Because there is zero chance on conflict of interest with Microsoft‘s own plans.

I found it interesting to see a survey wrapped under "join waitlist if you want professional ChatGPT access".

As against, "can you please fill this survey for us?" - which will likely have less takers.

Taking care of day-to-day meals for myself and family. It breaks my routine of what I actually want to do throughout the day. Takeout food is not a real alternative to home cooked meal.

> Mind-boggling compression

> ..We compress raw recording data up to 3,750x times..

Richard Hendricks should've run "Pied Piper" with this use case :-)

> For your privacy, we store all of the recordings locally on your Mac.

how does data transfer look like, when changing your mac?

On similar note: Despite faster hardware every year, laptops don't perform any better, running applications.

Usage experience with typical specs of 2022 (500G SSD and 16G RAM) is as good or bad as ones from 20 years back (say, 20G HDD and 128M RAM).

This doesn’t answer your question, but it may help folks who haven’t ended in this situation.

Amazon sells (at least in EU) CC annual prepaid subscription. Buy that instead and redeem it on Adobe to activate subscription. No direct purchase from adobe, no credit card on file. End of one year, your subscription will just expire out.

who made decision to at Tesla to build own ERP?

Would like to hear of cost/focus tradeoffs after so many years of use.

Anyone at Tesla thought of open sourcing it?

This vaguely reminds me of Silicon Valley season 3 episode, where Hooli CEO throws entire Nucleus team under the bus, to „take responsibility“.

Cracks me up, just how close SV (series) is to out there..