yobbo

8 ms·

91.
by yobbo 1y ago
So besides it being dovecot instead of cyrus, I am right? Btw, cyrus and postfix are far from random.
92.
by yobbo 1y ago
Open-Xchange is most likely a more effective name for the combination Cyrus IMAP, postfix, etc.
93.
by yobbo 1y ago
> Nobody was thinking about “solving” tasks by asking oracle-like models I remember this being talked about maybe even earlier than 2018/2019, but the scale of models then was still off by at least one order of magnitude before it h
94.
by yobbo 1y ago
They count on potential customers not reading the contracts, or being able to do math or research themselves. Typical customers for these types of scams are small offices with no technical person in the loop.
95.
by yobbo 1y ago
> Cars used to be simpler to work on because No, it's because they are designed to be assembled from complete sub-assemblies. Maintenance is not assumed to be done on the sub-assemblies while they are in the final product. Under war
96.
by yobbo 1y ago
I understand your goal. I've run this solution for more than a decade, originally with openvpn and multiple subnets. Yes, it does require your local networks to be subnets of the VPN. If you can't control dhcp, then you can add st
97.
by yobbo 1y ago
You should pick a random network (like in 10.x.x.x if ipv4) and a smaller mask for your VPN than any public wifi would ever choose. Then a yet smaller mask for your home network. The probability of collision then is tiny.
98.
by yobbo 1y ago
Short answer: because lock-in disables competition, and cloud-based business models enable almost perfect lock-in. Software quality only matters when users can switch.
99.
by yobbo 1y ago
> it uses Wireguard if I'm mobile and doesn't if I'm at home (which saves me a hop through the cloud hub) Appropriate netmasks in the routing tables do this regardless of physical interfaces. For example, if your local net
100.
by yobbo 1y ago
He laments youtube comments and health-gadget data in silos and walled gardens, but this is entirely congruent with the original http client/server concept. The protocols created no incentives to protect data and identities from being
101.
by yobbo 1y ago
No, it is not an accusation. If you accept that the world is not "just" (just-world-fallacy), then you will also believe that rewards are indeterministic. It follows that rewards are attributed to luck, while effort and results ar
102.
by yobbo 1y ago
I think it's more correct to attribute what you receive to luck, rather than what you give. On the other hand, if you received nothing in return for your work, would you do it?
103.
by yobbo 1y ago
The argument you started off making was that his "lack of success" was because he didn't have the desire to learn and thus wasted the opportunity. But - he achieved some of the highest grades, so by all accounts he was succes
104.
by yobbo 1y ago
> Often blaming "esoteric" and "irrelevant" subjects such as linear algebra for their lack of success, while not understanding that it is a vital course for any engineer. The author was successful in his particular en
105.
by yobbo 1y ago
Your argument is "I once saw someone do stupid FP, and it is all I have seen. Therefore, I believe all FP is stupid."
106.
by yobbo 1y ago
At a top university in Scandinavia, lecturers will rarely respond (or even read) emails from students. "Office hours" did not exist, but TAs attend scheduled labs/exercises. Professors were even under instruction to not engag
107.
by yobbo 1y ago
He graduated with "some of the highest grades" in his class.
108.
by yobbo 1y ago
The poster was achieved success in all metrics he was exposed to.
109.
by yobbo 1y ago
> University gave me the opportunity to talk to world-class researchers during office hours Neither world-class researchers or office hours exist in most Universities. "Office hours" is entirely an American (and maybe British?)
110.
by yobbo 1y ago
What does the moment look like when one trend ends? ZIRP is one example of a decade-long trend that is ending.
111.
by yobbo 1y ago
Yes, the purpose is to verify the gradient computations which are typically incorrect on the first try for things like self-attention and softmax. It is very slow. It is not necessary for auto-differentiation, but this project does not use
112.
by yobbo 1y ago
Very nice! Next thing to add would be numerical gradient testing.
113.
by yobbo 1y ago
> For this to be true they would have to throw away labeled training data. That's how validation works.
114.
by yobbo 1y ago
It's not clear from the formulas how x=[x1,...,xN] relates to y, μ, and σ since these are defined without x. Assuming y = Wx + b, and μ, σ, and π are learnable parameters for each output dimension. The symbol π seems to mean both weigh
115.
by yobbo 1y ago
As I remember intuitively, it's a convolution over a time window. The size of the time window limits the frequencies that can be detected.
116.
by yobbo 1y ago
Sounds similar to DOT-5 brake fluid. Maybe this: https://en.wikipedia.org/wiki/Tributyl_phosphate "The major uses of TBP in industry are as a component of aircraft hydraulic fluid, brake fluid, and as a solvent fo
117.
by yobbo 1y ago
It could be condensation in expansion tanks, or it could be rain into open containers on the ground, or someone could have mistakenly poured cooling liquid (or something else) into the containers, or into the hydraulic system itself, or ...
118.
by yobbo 1y ago
It might also be their relative confidence in peope vs LLMs for this sort of task. People could be worse when the task itself is trivial but the volume is intangible for a single human.
119.
by yobbo 1y ago
Step 2 does not imply step 3. Step 3 only occurs in a small fraction of cases. Step 4 even smaller fraction.
120.
by yobbo 1y ago
Yep, you're correct. It should be (n+1)/(n+m+2). If you have access to a machine or lookup tables, you might as well plug in the values for the distribution Beta(1+n,1+m) which should correspond to the joint density. (The formula

More ›