amoss

16 ms·

211.
by amoss 4y ago
Not to disagree, but you seem to have skipped 0) Increasing the parameter size of LLMs is a path to sentience.
212.
by amoss 4y ago
Ah, ok so you are considering the latency rather than the throughput. I was thinking that being able to perform the update within 16.6ms should not be too hard - there is a sibling comment about SSD latency but the working set should fit in
213.
by amoss 4y ago
Why? In 16.6ms a modern processor will execute about 128 million instructions. We need to locate about a screenful of text and/or cat pictures. We need to perform layout and then render it. The amount of work being done is not very muc
214.
by amoss 4y ago
https://archive.is/N7xvO for what it is worth. But, tl;dr Tories are gonna Tory.
215.
by amoss 4y ago
This assumes that the LAN is the bottleneck. Gigabit ethernet tops out at 120MB/s, which is about the speed of spinning rust on a NAS.
216.
by amoss 4y ago
It can fix the planet or it gets the hose again.
217.
by amoss 4y ago
I wonder how this will apply to code that comes out of systems like copilot?
218.
by amoss 4y ago
lol, of course. I hadn't realised it is so long since I wrote any Haskell...
219.
by amoss 4y ago
The interesting part of the exchange is the contrast between the surface level syntax of the exchange and the semantics below. It reminds me of teaching in a lot of ways. The level of polish on the immediately apparent surface level of the
220.
by amoss 4y ago
This is true, but not currently relevant. The cost (difficulty) of training does not indicate how good a fit the model is for copyright protection. I suspect that part of the arguing this out in court will be making those economic arguments
221.
by amoss 4y ago
That is the point that I made - a baseline for comparison is not defining the method that produces the best understanding of the code. It is defining the worst that we would expect somebody to try.
222.
by amoss 4y ago
A baseline to compare against would be googling the name of the function and reading a random result page instead of the actual code. The OP is demonstrating that the system does not rise above that baseline.
223.
by amoss 4y ago
Those protections rely on "systematic arrangement", i.e. that some effort has been put into finding, verifying and collection a set of facts / individual pieces of data. It is not obvious that would apply to the set of weight
224.
by amoss 4y ago
It helps if you take a deep breath.
225.
by amoss 4y ago
It would seem strange to represent doubt by that squiggle though when clearly it is represented by an x in a blue circle.
226.
by amoss 4y ago
How would you be offended if somebody referred to you by "they"?
227.
by amoss 4y ago
I have not read the paper beyond the abstract so this is something of a throwaway comment. The setup of the study raises questions about construct validity. The main purpose of these coding assistants is to write new code - i.e. code with
228.
by amoss 4y ago
It is reminiscent of the male behaviour in Universe 25 - this description reminded me of "the beautiful ones": https://www.the-scientist.com/foundations/universe-25-1968-1...
229.
by amoss 4y ago
If we flip the question around it seems more reasonable: would your IQ test score be accurate if you did not train beforehand on practice IQ tests? I would presume not - most tests are timed, and if you are spending time on first-time only
230.
by amoss 4y ago
I'm currently taking a sabbatical for a year. My son is still young enough that he finishes school in the early afternoon and we get to spend most of the day together. It is bliss. The projects that I'm working on still get time i
231.
by amoss 4y ago
Thanks for writing this up and sharing. The rawness of the experience comes through very clearly, and while it must have been difficult to write it reads very well. Hopefully this can help other people in the same situation. Your level of p
232.
by amoss 4y ago
Given the limited release of access under a license - is there any decision on whether a model like this would be protected by copyright if it leaked? The output of an algorithm is not a human authored creative work, which given the recent
233.
by amoss 4y ago
Do you think that a real, ancient, stoic philospher would have been able to give a reasonable answer to that question?
234.
by amoss 4y ago
Love it, results are legit. > How should we deal with the inevitable threat of death? Friend, death is an inevitable part of life that we must accept. We should strive to live our lives in a way that honors the gift of life and pre
235.
by amoss 4y ago
I hope so - I've had chronic insomnia my whole life (and a fair mix of night terrors and sleep walking when younger). Without melatonin supplements I would not have any kind of regular sleep schedule at all. Although on the bright side
236.
by amoss 4y ago
This makes sense for Apple - they are aiming at the premium end of the market. Having a process advantage over the competition will translate into a performance / battery win. After all, this is what Intel did for decades while they co
237.
by amoss 4y ago
There is a trade-off between the three kinds of improvements. It is possible to pick some points (for example) that give a less than maximum power reduction as well as a less than maximum speed improvment. I think the comparison would be th
238.
by amoss 12y ago
So it is 10 years before GNU and 15 years before Linux... It is also in the era that Microsoft are convincing people that software should have copyright for the first time. Create the open source movement in a time when there is no dominant