bjackman

19 ms·

331.
by bjackman 1y ago
Well, you are just directly contradicting the concrete claims made by the post so one of you is wrong... FWIW my interpretation of this is that the hallucination vector encodes the behaviour that a the model produces bullshit despite having
332.
by bjackman 1y ago
I think in a highly financialised free market like the US the executive is pretty powerless to really hide important economic data like this, even within their own country. There would be so much alpha in knowing stuff like the true employm
333.
by bjackman 1y ago
I have never personally had any impulse at all to get a tattoo, until I saw this guy getting "Ötzi's" tattoos: https://youtu.be/_BqarmmtLwc?si=fI3Lg9RXTabOuesG&utm_source=... As soon as I saw the title of
334.
by bjackman 1y ago
I think it's quite rare that the most important beliefs you should derive from the news can be justified by a collection of absolute statements. I think you'd be better served by a tool for calculating chains of Bayesian reasoning
335.
by bjackman 1y ago
Here in Switzerland bank transfers only take place during business hours I believe this is because Ürs has to load my silver pieces onto the donkey and drive it to the other bank.
336.
by bjackman 1y ago
Yeah exactly. If I move elsewhere for like 30s I'll get bitten. Or my ankles will be exposed or something. Same with incense-style repellents, I'm pretty convinced they reduce the number of mosquitoes but eventually one will be at
337.
by bjackman 1y ago
Some species of mosquitoes are able to survive their egg site drying out for quite some time. They go dormant and then hatch when they are rehydrated. I believe this is quite dependent on the timing in the lifecycle though so maybe there&#x
338.
by bjackman 1y ago
FWIW, I have found that icaridin-based repellents also work. It's dramatically less unpleasant than DEET! (If you find that a fan works, that's even better! But I never really find these environmental interventions are good enough
339.
by bjackman 1y ago
An idle observation about bureaucracy: In my big tech job we have a pretty small (by public sector standards) red-tape burden but it does exist. It does slow work down and it does increase the activation energy such that some small projects
340.
by bjackman 1y ago
Yeah I actively prefer diet drinks. Consuming 40g of sugar felt good when I was 10, but since around puberty it mostly makes my mouth feel sticky and my body feel dehydrated. Often it gives me a headache. However despite my strong objection
341.
by bjackman 1y ago
Is that issue the reason you can only work for 1-2 hours? Or is there something else limiting it?
342.
by bjackman 1y ago
I wonder if we'll reach a point soon where the monitor can be replaced by AR glasses? Beginning to dream of a Corne keyboard + AR glasses + phone setup. There's a chance we're beginning to approach fully pocketable computing
343.
by bjackman 1y ago
I have installed NixOS on mainstream hardware and it doesn't involve any of the faffing discussed here. The HW-specific aspect of setup is normally completely automatic.
344.
by bjackman 1y ago
It's rather bleak that these special "Linux computers" require this much fiddling. None of my laptops are sold as running Linux but all of them were easier to set up than this (only the MacBook Air 2013 required anything at a
345.
by bjackman 1y ago
You can be in case 1 and find meaning in what you do. That's where the blacksmith is. There's absolutely nothing wrong with doing a good job for 40 hours a week in return for a salary. Being a competent professional who does quali
346.
by bjackman 1y ago
I'm gonna copy paste a comment I wrote yesterday that I think fits perfectly here: As an engineer if you are gonna be a rank and file employee you need to do it for your own reasons. I think the main good reasons to do it are: 1. It&#x
347.
by bjackman 1y ago
Yeah but each only exists once. If you wanna clean up on ERR and also use that trap for debugging you need to implement some janky layer on top of the trap. (FWIW my take away from issues like this is always: Bash is not a serious programmi
348.
by bjackman 1y ago
But trap doesn't "stack" (like e.g. defer in Go) so if you do this it's not available for other purposes like cleanup
349.
by bjackman 1y ago
No don't worry, there are plenty of extremely shady free "VPN" services that this law will provide with a nice stream of victims
350.
by bjackman 1y ago
As an engineer if you are gonna be a rank and file employee you need to do it for your own reasons. I think the main good reasons to do it are: 1. It's relatively chill and you value the stability. You deliver competence from 9-5 then
351.
by bjackman 1y ago
Yeah as others mentioned it's really all about open source. If your project is open source but still interacts with the main body of Google code in a significant way, you can still be in the monorepo and then mirror it to Git via Copyb
352.
by bjackman 1y ago
I haven't used JJ with the git backend so there may be something I'm missing here, but I don't see any reason they wouldn't play nicely. (I use Gerrit daily with pain git). JJ does create commits without you explicitly a
353.
by bjackman 1y ago
We are talking about hardware here so ultimately you need to trust some manufacturer, software algorithms don't help. With SEV-SNP and Intel TDX I think it's possible to build a hardware platform that doesn't require the user
354.
by bjackman 1y ago
I don't think you can necessarily draw conclusions about Intel vs TSMC from Intel vs AMD. Yes, building top-of-the-line CPUs is hard and it's impressive that we saw the dominance flip in the course of just a few years. But I think
355.
by bjackman 1y ago
If you think the org that produced the hardware might have backdoored it, architecting your software to avoid the TPM or whatever is dumb. Targeting Google HW at all is an unavoidable act of complete trust so you might as well use the HW pr
356.
by bjackman 1y ago
Yeah I guess it confirms your point in a way too. There's nothing keeping me from switching except the activation energy cost. Almost every aspect of working in my area (Linux kernel) is painful so I'm constantly investing in tool
357.
by bjackman 1y ago
Counter point: I adopted it internally at Google (there's a backend for Piper, Google's monorepo Perforce thingy). I don't do my day-to-day work in the monorepo but I still jump in there once or twice a week. I adopted JJ bec
358.
by bjackman 1y ago
Given the Noam Brown comment ("It was a surprise even to many researchers at OpenAI") it seems extra surprising if multiple labs achieved this result at once. There's a comment on this twitter thread saying the Google model w
359.
by bjackman 1y ago
All the alternatives have a risk of setting off D&R tripwires. Assuming these things can spoof their device IDs so they look like a Logitech keyboard etc, I think the cost of the hardware setup is gonna easily pay for itself in terms of
360.
by bjackman 1y ago
I think the important question here is: once they've got your face or ID, do they store it? Is this effectively forbidding pseudonymity? It's worth thinking twice about this in a country that is actively censoring mainstream polit

More ›