jdeaton

I once worked with someone who was doing performance benchmarking of two systems, and made a duel axis chart with the lines right on top of eachother when in fact one system was like 5x faster than the other. it drove me nuts because I didn't even realize the dual axis at first and thought that they literally had identical performance

Facial expressions are confusing and ambiguous too. Maybe the emojis are WAI

Why not just call Eigen from rust

In python fast means calling libraries that are implemented in C/C++ for example numpy.

I would be interested to know exactly which video of Robert Greene's this is. Just finished reading Mastery this morning. Great author and thinker. He is very in tune with human nature and instinct.

Paperpile is fantastic and you can make a shared folder with your lab/team.

truth is helpful

sensing sarcasm, I wonder how you see this as lacking humility. there are no claims about myself here

Mediocre engineers send in their resume. Good engineers network to find a connection in the company. Great engineers constantly have people banging on their door asking to work with them.

you linked a paper about vision transformers...

This sounds like a classic case of someone (/whole team) mistaking their title for their role.

Sounds like free adversarial data augmentation.

Which field are you in now?

Thanks thats helpful. One place where the parallelizability of this method falls short of the transformer is not being able to pack multiple varying length examples into the same array during training with block diagonal attention pattern. If I understand correctly thats not possible with this architecture and its an important practical concern in large scale transformer training.

Very cool ive read this line of paper originating from hippo, s4, hyena, mamba etc but can someone please explain how this isnt just an RNN/LSTM variant??

Critique is nice but the ML suggested edits are a waste of attention and cant be disabled.

I find it absurd that a programming language which compells someone to set up a containerized OS just to manage their build toolchain could be considered "modern".

Why should ad hominem argument be taken seriously here

> Extensions are unavailable while your Bard Activity is off.

bummer, guess I wont be using it

> your customer is not the company's customer, it's your management chain

Another: some coworkers will have spent so long in large corporations that they've internalized the pathological dynamics of large organizations as universal truths.

This enforcement has helped me a lot reduce habitual youtube use in moments of boredom. Now when I hit the playback block, it gives me a moment of pause to remember that that watching youtube isn't something I really want to be spending my time doing anyways, so I just close the tab.

Always Break Time is an AI anime on twitch that is pretty fun and should be included here too :)

> Smart people don’t consider themselves better than people around them.

Do you have any direct evidence of this or are you merely making an observation.

That's like asking why anyone would want to use an if statement. "Why would you want to change which instructions that get executed during the program? Just decide up front!". Because the decision cannot be made ahead of time.

One of the biggest issues I've observed with workflow systems is an inability to get return values from each subprocess and then decide based on those which subprocesses to instantiate next. It always seems like the workflow execution graph is static in my experience. Is that correct?

Hallucinate means false perception which doesn't describe what LMs do when they're making up plausible sounding things

If you could retroactively give me 7 years software support on my pixel 5 that would be great thanks

I think its great that someone is trying to make haskell not suck to use. Better still that its very ambitious as it means people will learn much if or when it doesn't meet its aspirations.

> This sounds basically of just asking a GPT to suggest ideas for experiments?

I think actually the second idea extends quite a bit further than that. The idea is that you enable the agent to interact with an entire collection of laboratory equipment through tool-use. The agent not only generates hypotheses and designs experiments to test them, but then also actually executes the experiments through tool-use and iterates.

That would be less fun though