-
**Describe the bug**
The current implementation of TDIGEST_MERGE when used in a group by context launches separate GPU operations (kernels/memory copies) on the order of the number of groups in the o…
-
-
I am using trtllm 0.8.0 (added moe support following llama's implementation). we serve models with trtllm_backend (docker images triton-trtllm-24.02)
[qwen2-moe-57B-A14B](https://huggingface.co/Qwe…
-
### Feature description
I find myself limited by nebari with respect to working on larger analyses (multiple notebooks spread across directory tree). Locally, I would either:
- start JupyterLab with…
-
I'm encountering an error when running nixos-rebuild on a freshly infected system.
```
updating GRUB 2 menu...
cannot copy /nix/store/i1i6pn1wz0i25pvwcfhdra4bz6hcqjp7-linux-6.1.74/Image to /boot/…
-
I am trying to download a large number of kernels from a competition. It seems kernels returned from `kaggle kernels list` for a specific competition are capped at 1100 kernels no matter how I set pag…
-
The jupyter team provides [`jupyter_client` ](https://github.com/jupyter/jupyter_client/) package, which is the reference implementation of the Jupyter protocol.
I'm considering to use this in Hermes…
ngr-t updated
2 weeks ago
-
_This is a copy of a presentation for MIOpen team I held a couple of years ago, when we've introduced and implemented the Solver/Solution architecture. It does not cover the recent additions like GetW…
-
## 🚀 Feature
An FFT backend for `kornia.filters.filter2d`.
## Motivation
`kornia.filters.filter2d` is very slow for large kernels as it currently only performs convolutions in the spatial domai…
-
## Description
When launching `voila` with preheated kernels enabled and a `pool_size=2` in a directory with 4 notebooks, the server crashes with an inscrutable error. The `voila` server runs succe…