not original author but batching is one very important trick to make inference efficient, you can reasonably do tens to low hundreds in parallel (depending on model size and gpu size) with very little performance overhead

now i know the bar is 1000 feet below the earth with teams but matrix is still only maybe a foot or two above the surface

i really want to like it but every few months i try it and it’s clearly just not ready :(

logos on website if you want to see some of our customers lol

True! Devin Review doesn’t make the kind of judgements you mention, it just does its best to find bugs and help you understand the code faster. I managed to review a PR on an airplane (without starlink) with it earlier this week lol

Obviously the working tree should be a commit like any other! It just makes sense!

(I work at Windsurf but not really intended to be an ad I’m just yapping)

E: I found the paper: https://arxiv.org/pdf/2010.11929

> We use standard learnable 1D position embeddings, since we have not observed significant performance gains from using more advanced 2D-aware position embeddings (Appendix D.4).

Although it looks like that was just ImageNet so maybe this isn't that surprising.

Weirdly enough this also means that if you’re running with TCP you can have retransmits at the NFS/SunRPC level and at the TCP level depending on your configuration.

People have tried this before but usually the UI version is not fully complete so you have to drop to YAML. Now that the spec is good enough it's possible to build a complete UI for this.

My current project is basically this, except it edits your git-ops config repository, so you can click-ops while you git-ops.

- Java is cool, actually

- Java would be just as uncool even if people weren’t required to use it in school

The main downside I can see is that users have to write their workspaces for a particular deployment target. This would be a problem for e.x. open source projects trying to check in a workspace definition file of some kind. We standardize on Kubernetes across clouds and bare metal so it's not an issue for us, but it makes sense that it would be an issue for other use cases.

Taking advantage of the functional nature of Chisel enables a set of generators called Chipyard [0] for things like cores, networking peripherals, neural network accelerators, etc. If you're focusing on exploring the design space of one particular accelerator and don't care too much about the rest of the chip, you can easily get a customized version of the RTL for the rest of your chip. Chisel handles connecting up all the components of the chip. All the research projects in the lab benefit from code changes to the generators.

Chisel even enables undergraduate students (like me!) to tape out a chip on a modern-ish process node in just a semester, letting Chisel significantly reduce the RTL we have to write. Most of the remaining time is spent working on the physical design process.

[0]: https://github.com/ucb-bar/chipyard

[1]: https://classes.berkeley.edu/content/2023-Spring-ELENG-194-0...

Apparently this was just a function of how it was prompted, they claim it works fine if you use a different prompt -- https://www.anthropic.com/index/claude-2-1-prompting

[0]: https://boom-core.org

[1]: https://carrv.github.io/2020/papers/CARRV2020_paper_15_Zhao....

In reality I think distributions that are willing to ship binaries (NixOS, Homebrew on MacOS) ship VSCodium, and other distributions (Alpine) have packages called things like `code-oss` that are basically the distribution's internal compiled version of VSCode and have nothing (?) to do with VSCodium.

Occasionally I'm at the airport (or someplace with the "real" wideband 5G) and I want to download a movie quickly. At that point I turn on 5G and get 2 Gbit down, which is convenient!