Where did you get that 3 per 100k from? In each of the last 2 years there were 55 homicides in SF from what I can tell. Maybe you're thinking of New York, which is half of SF's rate?
The FBI's Uniform Crime Reports (second link) contains statistics from 2019. However, it seems to be the most recent edition. I have no idea if or where the 2020-2022 ones exist.
If someome could point me to them or explain why they don't exist, I'd appreciate it.
Any merit to this service? It claims without 'hidden fees' so I thought they would get rid of the airbnb-like fees, but it looks like there is still cleaning fees, platform fees, and taxes not included in the nightly rate (for +50% final cost over the nightly).
The install (npx dalai serve) fails silently for me. With --verbose it says `npm info run [email protected] install { code: 1, signal: null }`. Ubuntu 22.04.
How do they get ChatGPT not to hallucinate stuff about the articles? Everything seems fairly accurate, which is not my experience with ChatGPT when talking about technical things. Is it heavily curated/edited by humans?
I noticed that the text often comes out verbatim from the articles, perhaps this indicates a clever prompt that keeps things closer to the truth by requiring verbatim output.
chatGPT hallucinates more the further removed it is from the data. I'm asking it about laravel, and it knows nothing about laravel 9 or 10 changes, but if I feed it an entire article or document it'll hallucinate a lot less because it's fresh.
kinda like how we can recall things closer to the event than months later.
it knows a ton from it's training but it still got it from the web so always question it, but if we can add meta data and other things to strengthen the llms understanding it shouldn't hallucinate much at all.
If you use temperature 0 with an API call it does not hallucinate much at all especially with a good prompt including the information you are asking about.
How does token->embedding lookup work with 1.4T BPE tokens? Since there are more tokens than the 65B parameters it must be doing some sort of interesting thing based on the merge operations. Is it different from what other GPT models with ~100k tokens are doing?
At inference, how many of those tokens are used? (they mention most tokens are used only once during training, so the must be very long sequences.)
Ah, that makes more sense, thank you. Since this was mentioned in the tokenizer section and the number of unique tokens wasn't mentioned I misunderstood.
The options grant appraisal for startup employees seems very optimistic - 70% it's worth nothing. 25% payout is equivalent to all the backpay you would have gotten if you had worked at a big tech co. instead. 4% life changing money but not the richest person you know. 1% you are the richest person you know.
Was this the generally the way people thought about it back then? I remember seeing fairly pessimistic valuations around 2013 though, and I see it more like 10-15% chance of big tech salary equivalent, and 1-2% for life changing.
I've seen this use of constants inside big O notation in leetcode and 'informal' discussions. Is it pedantic to say that O(NM) == O(N) if M is a constant (in this case since it's bound by 26)? Or is this the current and expected usage?
It's correct to omit constants from big O, but in the article W is a distinct input variable so the usage is valid. The size of the window W is bound by N, not a constant 26.
>What happens to all of Anja's clients if the business exits?
>All biological goods stored with Anja will stay in our New Jersey lab regardless
>of what happens to Anja as a business. Client relationship management and
>connection will be securely transferred to a responsible, experienced party if
>necessary.
Another comment asked this and other questions but this one was not answered other than 'there is insurance'.
How does this work? Will customers lose their existing 20 years they paid for and have to pay the new party all over again? How is Anja incentivized to find a good party to transfer to at the time of business failure? How is the funds for the lab guaranteed after Anja goes out of business?
Cryonics organizations have solved this problem with a non-profit structure, but I'm really curious how it is handled in this for-profit case.
Clients sign an agreement stating that it is their property. We shift the management of this property to another company in the case that anything happens. All samples remain in our lab, unless specifically requested otherwise. We have a legal obligation to turn over client relationship management elsewhere.
Can you answer this question: In the case of business failure, will customers lose their existing 20 years they paid for and have to pay the new management company all over again?
These points and the Guzey essay were discussed in depth in Science Fictions by Stuart Ritchie, which features examples of negligence and fraud in science. Having read and enjoyed the message in Why We Sleep, my first reaction before reading based on the length of the essay was that the motivation was axe-grindy, but the further I read the more it made sense and the more egregious Walker's claims seemed. The lack of a response from Walker is another serious concern (especially after Andrew Gelman highlights this). It seems that he is trying to avoid a Streisand Effect by ignoring it completely. If that is the case, Walker's strategy seems to be working because on the comments on new threads like this, many mention mention Walker but seem unaware of the criticisms.
Quote from Gelman:
> We’ve left “super-important researcher too busy to respond to picky comments” territory and left “well-intentioned but sloppy researcher can’t keep track of citations” territory and entered “research misconduct” territory.
Yes, once you have felt and identified a chanterelle with success a few times, it's very easy for an experienced forager to tell the difference between gills and folds (especially in the larger species such as C. californicus). The folds are often veiny and cannot be moved or 'plucked' like gills. In general though, identification should be done using as many factors as possible (e.g. using a identification key which is like a decision tree). For example, besides gills and smell, another way to differentiate the (non- or only mildly-toxic) false chanterelle is the firmness.
https://sfist.com/2023/01/03/sf-sees-exact-same-number-of-ho... https://en.wikipedia.org/wiki/List_of_United_States_cities_b...