tmoertel

16 ms·

181.
Pharma Is Pushing $200k Cancer Drugs When Cheaper Doses May Work(bloomberg.com)
6 points by tmoertel 1y ago|0 comments
182.
by tmoertel 1y ago
NOTE: The article of interest begins on page 392. I had linked to it directly, but the Hacker News site stripped the page number out of the URL, so now the link goes to the first page of the issue of BYTE magazine that contains the article.
183.
by tmoertel 1y ago
Woz is rightly celbrated for his hardware designs, but, as this article shows, his software designs were also impressive. This article gives a clear derivation of some clever math and implementation techniques to compute e to over 100,000
184.
The Impossible Dream: Computing E to 116K Places with a Personal Computer [Woz](archive.org)
2 points by tmoertel 1y ago|2 comments
185.
Measuring AI Ability to Complete Long Tasks (2x every 7 months)(metr.org)
3 points by tmoertel 1y ago|0 comments
186.
Statistical Methods in Generative AI(huggingface.co)
2 points by tmoertel 1y ago|0 comments
187.
by tmoertel 1y ago
I've always explained it like this: I would prefer that you work reasonable hours and no more. Just make sure that you're giving your best during those hours. Focus on solving problems, and avoid distractions like social media. Mo
188.
by tmoertel 1y ago
I think that reasonable hours are still the norm at much of Google. I worked there as a SWE for over decade before I left last year. I never once felt pressured to work long hours or extra days. I do recall several times when management fol
189.
by tmoertel 1y ago
Please note that the person who posted this announcement to HN is the event's organizer. Thus it seems reasonable to conclude that "the very general forum of HN" is part of the event's target audience.
190.
by tmoertel 1y ago
This one should. The whole point of announcements is to cast a wide net in hopes of catching people who are new to the scene, and those new folks shouldn't be expected to already know the scene's lingo.
191.
by tmoertel 1y ago
The Gemini Pro subscription includes Deep Research and Veo 3; you don't need the pricey Ultra subscription: https://gemini.google/subscriptions/
192.
by tmoertel 1y ago
When reading this article, be aware that there are some percent signs missing, and their absence might cause confusion. For example: > After 9 peeks, the probability that at least one p-value dips below 0.05 is: 1 − (1 − 0.05)^9 = 37. Th
193.
by tmoertel 1y ago
For the Bacillus subtilis ?
194.
by tmoertel 1y ago
> My understanding is that top talent gets top pay, regardless of their living arrangements. Indeed. Top talent can say no to lower offers because they are confident that companies are unlikely to find other top candidates who will say y
195.
by tmoertel 1y ago
> Is there a good reason why a developer in Thailand or India should be paid less than their colleague who works on the same team, but is based in the US? Yes, and that reason is that people in most of the developed world are free to say
196.
by tmoertel 1y ago
That's Hemingway’s trick: “You write until you come to a place where you still have your juice and know what will happen next and you stop and try to live through until the next day when you hit it again.” https://www.thepar
197.
Mind the Trust Gap: Fast, Private Local-to-Cloud LLM Chat(hazyresearch.stanford.edu)
1 points by tmoertel 1y ago|0 comments
198.
by tmoertel 1y ago
> I get why it is a business strategy to not have limits... What is the strategy? Is is purely market segmentation? (As in: "If you need to worry about spending too much, you're not the big-money kind of enterprise customer we
199.
by tmoertel 1y ago
Quick version: Google AI Edge Gallery is an app that lets you run LLM models from Hugging Face locally on your phone (Apache license). It lets you run one-shot, chat, and image-input flows and provides useful performance measurements.
200.
by tmoertel 1y ago
What's the deal with this notice, presumably injected by GitHub into the hosted code samples in the article: > This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears
201.
by tmoertel 1y ago
According to the authors' blog post: > Does Branch Privilege Injection affect non-Intel CPUs? > No. Our analysis has not found any issues on the evaluated AMD and ARM systems. Source: https://comsec.ethz.ch/resear
202.
by tmoertel 1y ago
The & is a postfix operator that converts an expression into an anonymous function. See “Pure Functions”: https://reference.wolfram.com/language/tutorial/FunctionalOp...
203.
by tmoertel 1y ago
A while ago I tried to create a more self-explanatory implementation: https://github.com/tmoertel/practice/blob/master/libraries%2... It is limited to integer weights only to make it easy to verify that
204.
by tmoertel 1y ago
Two notes on the weighted version. First, the straightforward implementation of selecting the top N when ranked by POW(RANDOM(), 1.0 / weight) has stability problems when the weights are very large or very small. Second, the resulting
205.
by tmoertel 1y ago
Mathematica has this: In[1]: Map((1 + #1)&, {a, b, c}) Out[1]: {a + 1, b + 1, c + 1} See https://reference.wolfram.com/language/ref/Function.html.en for the full story.
206.
by tmoertel 1y ago
I read it. It doesn't seem to show that ”google will sell the fact that you visited that porn website to anyone who is interested.” What am I missing?
207.
by tmoertel 1y ago
In early 1990s I worked at a mid-sized software company making software to help big businesses do big-business stuff. One day, another programmer pops into my cube and says: "Hey, how do you draw a 3d pie chart?" "What?"
208.
by tmoertel 1y ago
> google will sell the fact that you visited that porn website to anyone who is interested. Has Google has actually done this? According to Google, they don't sell personal information: https://about.google/company-i
209.
by tmoertel 1y ago
Indeed!
210.
by tmoertel 1y ago
I think that there is a different reason that an emphasis on simple code often results in faster systems. When you write simple code, you spend less time writing code. Therefore, you have more time left to invest in optimizing the very smal

More ›