Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
kevmo314
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
21 ms
·
211.
▲
by
kevmo314
2y ago
This list mentions A/B testing a few times and it's worth noting that A/B testing is great but it's not free. I've seen a nontrivial number of smart engineers get bogged down in wanting to A/B test everything t
212.
▲
by
kevmo314
2y ago
> Do code reviews. Don't just skim PRs (sorry, reader!), but really dig into them: run the branch locally, test it, think critically about the design and the implementation, and provide feedback. Record a video of your review to hig
213.
▲
by
kevmo314
2y ago
Interesting they chose to split prefilling into its own independent service, I hadn't heard of that technique before. I found this paper that researches why that could be beneficial: https://arxiv.org/abs/2401.0967
214.
▲
by
kevmo314
2y ago
Those are configured on the client so luckily you get to keep them. Extensions in VS Code are a bit different because some (many?) run on the host and as a result devcontainers wipe them out. I do agree though, I think devcontainers are in
215.
▲
by
kevmo314
2y ago
I use them as a last resort. They're nice in theory but you'll lose your extensions (which you can configure but if you check it in you're imposing on everyone else...), bash history, and most debugging tools you have install
216.
▲
by
kevmo314
2y ago
Producing low-quality work takes less time than high-quality work even under that assumption.
217.
▲
by
kevmo314
2y ago
The article is arguing that effort and outcome are disconnected and that > When outcome is impossible to predict, the best strategy is volume. but it concludes that > Because in a world of blockbuster or bust, the only thing that matt
218.
▲
by
kevmo314
2y ago
The article didn't claim it was a good proxy.
219.
▲
by
kevmo314
2y ago
I like the use of the functional API here. I learned through a similar route and it was very helpful for me compared to trying to understand `torch.nn.Module`. Here's a gist of my learning path if it's helpful to anyone: https:&#
220.
▲
by
kevmo314
2y ago
If this had come five years ago perhaps TensorFlow could've stood a chance against PyTorch. Switching from TensorFlow to PyTorch was such a breath of fresh air, I definitely could have used something like this.
221.
▲
by
kevmo314
2y ago
I don't think it would be possible out of the box: Docker pulls assume a direct HTTP download for the image. It would be pretty cool to build a proxy that acts as a torrent client for images however it would be a lot less ergonomic to
222.
▲
by
kevmo314
2y ago
> Again, iiuc, this would result in the same conflicts as the rebase way Yeah I mean I agree I would have expected too but it's not what happens to me. Like I said I'm sure I'm using it wrong but that's my gripe with
223.
▲
by
kevmo314
2y ago
If I'm only working on one feature at a time then actually it's largely ok, it's when I have to manipulate the commit graph that things start to really go wrong for me. But rebasing I still don't understand, even after u
224.
▲
by
kevmo314
2y ago
As someone who uses this model with git, the main reason is because git is ubiquitous but I cannot get used to git's incredibly user unfriendly API. I've tried the feature branch approach and it doesn't click for me. Merge co
225.
▲
by
kevmo314
2y ago
Is the UI tooling that does the visualization a library? I really like the UI format, would love to use this for breaking down and debugging video byte streams too. EDIT: Oh it's actually reasonably simple, great use of CSS! https:&#x
226.
▲
by
kevmo314
2y ago
I once had a project that moved a microservice into a monolith on a larger instance type. SRE flagged that larger instance type as expensive, at $1k/mo. They really drilled in, asking "are you sure? that's a very expensive in
227.
▲
by
kevmo314
2y ago
The targets documentation makes this seem like Bazel but everything is a container, so you don't have to learn the ins and outs of a build system versus using whatever tooling is most familiar + a thin Docker config layer. https:/
228.
▲
by
kevmo314
2y ago
Is the news here that someone bolted an LLM onto that? Extensions always had this degree of access. npm or pip install can also do some pretty nasty stuff with a malicious package.
229.
▲
by
kevmo314
2y ago
Retiring, for example.
230.
▲
by
kevmo314
2y ago
> The buyout offer entitles federal employees to stop working more or less immediately and continue to be paid through Sept. 30. > The federal workforce's normal attrition rate is about 6% a year, meaning some of those who'v
231.
▲
by
kevmo314
2y ago
Definitely curious, this looks very similar to Coconut, even down to the CoT encoding process in Figure 2. They go into a lot more detail though, seems like parallel innovation.
232.
▲
In search of the perfect dynamic array growth factor [video]
(youtube.com)
1 points
by
kevmo314
2y ago
|
0 comments
233.
▲
by
kevmo314
2y ago
I used to think inventing new things was the hard part. Now I'm finding that realizing solutions are novel and noteworthy is a lot harder.
234.
▲
by
kevmo314
2y ago
The ultimate monorepo.
235.
▲
by
kevmo314
2y ago
This line in the article resonated with me: > First, it sets the stage for what was to come, and second, while it was unquestionably a success, it was not my success. I used to think that success was being successful the way I wanted and
236.
▲
The Qubit Factory
(github.com)
1 points
by
kevmo314
2y ago
|
0 comments
237.
▲
by
kevmo314
2y ago
I think it's somewhat hard to get the API right. This is what React-based static-site generators are but people love defeating them by introducing runtime dependencies and configuration.
238.
▲
by
kevmo314
2y ago
Fun story, while at FAANG I was once tasked with making a parameter configurable via flag. This wasn't super easy for various reasons, but I pointed out that it was configurable already: all you had to do was submit a PR to change the
239.
▲
by
kevmo314
2y ago
I believe that's similar to the idea behind https://github.com/facebookresearch/blt
240.
▲
by
kevmo314
2y ago
Yeah I've tried that approach. The model ends up needing to learn every combination of tokens. For example, the word "apple" now has six bytes positions it can be split on and the model suddenly needs to learn that all six wi
More ›