-
From the [blog](https://flashinfer.ai/2024/02/02/introduce-flashinfer.html) I noticed that FlashInfer implements low-precision attention kernels so that we can achieve nearly linear speedup to the com…
-
# Bug Report
0.8.0
**Current behavior:**
This may be a problem with how I am interpreting how GPJax handles combination kernels, so sorry if I've missed something.
It seems that kernels wh…
-
The commit 650ff11a37993a9e5670744130335ac7925e5e06 broke building Toybox for J2 the Turtleboard:
```
glaubitz@node54:/data/home/glaubitz/toybox> make root CROSS_COMPILE=sh2eb-linux-muslfdpic- LIN…
-
### Description
I initially submitted the issue here:
https://github.com/deepmind/dm-haiku/issues/724
But then realized it was a jax issue.
In short, I've been trying to use Conv2DTranspo…
-
When using `chakra_trace_link` on AMD Instinct MI210 GPUs, a bunch of warnings crop up when linking the Kineto and ET JSON files:
```
[2024-07-11 19:37:52,768] trace_linker.py:721 [WARNING]: Missi…
-
The kernel split in orchestration is very high, leading to a kernel fragmentation that is very damaging to CPU (and not great on GPU).
We need to understand _deeply_ why and plan for resolving this i…
-
In ALT Linux we have package rebuilds and occasionally it reports build errors for toybox. (It builds with `make -j8`).
I think this is related to `generated/help.h` generated incorrectly. I manually…
-
### System Info
- `transformers` version: 4.44.0
- Platform: macOS-14.6.1-arm64-arm-64bit
- Python version: 3.12.3
- Huggingface_hub version: 0.24.6
- Safetensors version: 0.4.4
- Accelerate v…
-
First of all, thank you for the great work you're doing in maintaining this project. I encoutered what seems to be a bug when attempting to use a support vector classifier with a precomputed Gram matr…
-
I've just tried to use coz for some rust programs but I don't know how to run it correctly. When I try the default example `toy` provided in the `rust/` directory I'm getting the following error:
```…