Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
rthnbgrredf
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
13 ms
·
151.
▲
Ask HN: How to Find Better Style?
2 points
by
rthnbgrredf
3y ago
|
2 comments
152.
▲
Ask HN: Is it possible to filter out Twitter submissions?
7 points
by
rthnbgrredf
3y ago
|
4 comments
153.
▲
A more mature take on stateless Terraform
(bejarano.io)
77 points
by
rthnbgrredf
3y ago
|
89 comments
154.
▲
by
rthnbgrredf
3y ago
If you're interested in category theory, I have compiled a list of resources quite recently: https://github.com/madnight/awesome-category-theory
155.
▲
by
rthnbgrredf
3y ago
Anyone know what kind of performance, in terms of token per second, one could expect with such a system running llama-2-70b?
156.
▲
by
rthnbgrredf
3y ago
Any statement which is not logically valid (read: always true) is unprovable. The statement ∃x∃y(x>y) is not provable from the theory of linear orders, since it is false in the singleton order. On the other hand, it is not disprovable s
157.
▲
by
rthnbgrredf
3y ago
Kernel is enough attack surface to require reboot, can't ignore it.
158.
▲
by
rthnbgrredf
3y ago
It seems to me that they made the same mistake that I did in my GitHub archive queries, they do not filter bot accounts. JavaScript, without filter, is on top 1 because of dependabot. If you filter all bots then Python is number 1, see: ht
159.
▲
by
rthnbgrredf
4y ago
Author here. I started to filter pull requests from bots like dependapot, because dependabot massively inflates the number of pull requests, especially for JavaScript, in recent years. Unfortunately, I don't have enough BigQuery Credit
160.
▲
Quantization: GPT Model Compression and Efficiency
(beuke.org)
2 points
by
rthnbgrredf
4y ago
|
0 comments
161.
▲
by
rthnbgrredf
4y ago
Finding the theory of everything. ChatGPT: "The theory of everything (TOE) is a hypothetical single, all-encompassing, coherent theoretical framework of physics that fully explains and links together all physical aspects of the univers
162.
▲
by
rthnbgrredf
4y ago
I suggest to just you to think of proofs in a different way. There are different ways to write a proof. I think proofs are sufficient if you are able to convince your audience. If you try to convince a theorem prover like Coq you have to be
163.
▲
by
rthnbgrredf
4y ago
Learn vi/vim/nvim. Interface hasn't changed much since 1976 (first version of vi).
164.
▲
World Elephant
(en.wikipedia.org)
1 points
by
rthnbgrredf
4y ago
|
0 comments