az226

40 ms·

301.
by az226 3y ago
230x8 MoE.
302.
by az226 3y ago
No motivation needed whatsoever. Just check out page 27, table 5. The data speaks for itself.
303.
by az226 3y ago
I tried. It didn’t generate any code. It felt as though I got a reply back from ChatGPT.
304.
by az226 3y ago
Unironically — it happens that Carmack said he got it from someone else.
305.
by az226 3y ago
Agreed. Racial discrimination preferences won’t be missed.
306.
by az226 3y ago
In a typical class of 2500 students at Harvard, removing legacy preference would increase the minority population by 25. Removing athletic preference would and another 25. Removing minority racial preference would remove 450. 450 is 18x tim
307.
by az226 3y ago
Use an LLM to go through the first 50 links, explore the content, and give you a summary of your query along with the link to the best result.
308.
by az226 3y ago
A typical class at Harvard has 2,500 student admits. Of those, 700 are black or Hispanic. If Harvard did away with legacy admissions, 25 additional Hispanic or black applicants would be admitted, so 725. However, if they did away with affir
309.
by az226 3y ago
A typical class at Harvard has 2,500 student admits. Of those, 700 are black or Hispanic. If Harvard did away with legacy admissions, 25 additional Hispanic or black applicants would be admitted, so 725. However, if they did away with affir
310.
by az226 3y ago
A typical class at Harvard has 2,500 student admits. Of those, 700 are black or Hispanic. If Harvard did away with legacy admissions, 25 additional Hispanic or black applicants would be admitted, so 725. However, if they did away with affir
311.
by az226 3y ago
Patently incorrect. Minority preference is ~20x times larger than legacy preference at Harvard. A typical class at Harvard has 2,500 student admits. Of those, 700 are black or Hispanic. If Harvard did away with legacy admissions, 25 additio
312.
by az226 3y ago
If Harvard were to go from a race and legacy aware policy to a race and legacy blind policy, 20 minority students who would have been accepted would not for every legacy student that similarly would not. That is, racially discriminating is
313.
by az226 3y ago
Elon Musk is African American.
314.
by az226 3y ago
In America middle eastern counts as white, not black. But nobody is enforcing these things and now they won’t matter because racially discriminating on admissions is illegal.
315.
by az226 3y ago
Harvard has said that 85% of students who apply would do well academically, but they only accept like 4%.
316.
by az226 3y ago
Patently false
317.
by az226 3y ago
Legacy is always brought up in these discussions. The amount of students who get into Harvard because of the boost from their race outnumbers the number of legacy students who get boosted from their status 20 to 1. Basically a race and lega
318.
by az226 3y ago
Once I’m used to GPT4 output and once in a while try and see if GPT3.5 can also get it right, it fails miserably. Every single time. I’m sure it’s fine for some things, simple things.
319.
by az226 3y ago
We’re still in the early innings.
320.
by az226 3y ago
Norway should have at the same time made a tax for expatriating wealth, to make capital flight very expensive.
321.
by az226 3y ago
Yes and no. Some LLMs can’t get the right answer no matter how many rolls of the dice they get. The bigger BS is that several questions weren’t questions and some didn’t have enough detail to be answered and it somehow got 100% on those. Th
322.
by az226 3y ago
But then this dumb statement should have been taking about the upload to GitHub and not the paper and how collaborators weren’t aware, didn’t get permission to run the analysis, etc. Sure, they’re all connected to the paper but they’re sepa
323.
by az226 3y ago
But that’s something they should deal with internally. It is quite strange actually. The statement should also come from MIT’s public affairs office and not the professors themselves. So it looks like they are big mad and acting on their ow
324.
by az226 3y ago
Paper didn’t include the exams themselves, please correct me if my understanding is incorrect.
325.
by az226 3y ago
So lame they focus on gatekeeping the exams and crying about permission as opposed to challenging the paper’s shit methodology.
326.
by az226 3y ago
Ehm. I answered that the cop and ambulance were in violation of the rule, but it’s sometimes okay to break the rules. That doesn’t mean they didn’t break the rules — it means they were justified in breaking the rules.
327.
by az226 3y ago
Pressure cookers that are rated for 15 PSI typically come apart at 75 PSI. But that’s the thing about ratings. The manufacturer guarantees safe usage at 15 PSI and anything above that all bets are off and you’re on your own. Same with milk
328.
by az226 3y ago
No chance GitHub is anywhere near using private repo source code to train the public GitHub Copilot. On the other hand, getting a personalized Copilot within your org is something entirely different and probably will be out in the not too d
329.
by az226 3y ago
Carbon fiber would get worn down very rapidly compared to steel.
330.
by az226 3y ago
OceanGateGate

More ›