-
I tried this code: [[playground]](https://play.rust-lang.org/?version=nightly&mode=debug&edition=2021&gist=1ffcf365aa942927bfb5f465c4dc4e99)
```rust
#![feature(inline_const)]
pub fn cause_l…
CAD97 updated
8 months ago
-
# 🐛 Bug
I have an optimization problem that consists of 4 input and 5 output parameters. It takes approximately 3 hours to evaluate one sample of the true function.
I have successfully implemente…
-
I have a dev environment for a large project. It's a 3d game, and contains gigabytes of assets. Each time a flake computation occurs, due to https://github.com/NixOS/nix/issues/3121 the flake evaluati…
-
微博内容精选
-
- [x] Research heuristic study
- [x] Design activity
- [x] Write copy
-
Currently, the TorchInductor C++ backend uses OpenMP, though especially for small kernels there is pretty high warmup time to launch threads. And we have seen our 1-thread performance give bigger spe…
-
### What happened?
For both CPU default flags and data-tiling+ukernel paths, we see an accuracy error on GPT2 for batch sizes 48 and 64 (possibly higher but haven't tried). Accuracy is correct for …
-
### Overview
We are about to deploy the application, and it would be nice to have a list of all the things that should be changed to improve the user flow. That way we can chose to make changes or n…
-
Some engines have a special option : the 'contempt' factor.
This is a kind of 'optimism' when playing (=my words).
Recently i found this nice article about this, it also contains some links for mo…
-
So I was thinking about the following idea.
It is probably completely bogus, but I would definitely investigate it when and if I had the time to, so maybe someone else would be interested as well.
…