-
### 🐛 Describe the bug
even tho on `Tensor.copy_` we see major improvements on BW on MI300X compared to H100. On a similar memory BW bound op like `sum()`, we were able to achieve a read bandwidth …
-
I have a sxm file that we would like to read and translate into pycroscopy but this does not work.
```
>>> nt = NanonisTranslator(sxm_file)
>>> nt.translate()
-----------------------------------…
-
and
-
A reader will look at the comment that "we time only sections that are entirely on the GPU side and are part of the analytic procedure", note we're only doing single-GPU experiments, and wonder why th…
-
Hello
We are trying to use Deepspeed to load LLama 405b across 2 nodes, of 8 x H100 SXM each.
We want to shard the model across all 16 gpus so that the model will be loaded in shards of 50.6GB e…
-
### Your current environment
```text
PyTorch version: 2.3.0+cu121
Is debug build: False
CUDA used to build PyTorch: 12.1
ROCM used to build PyTorch: N/A
OS: Debian GNU/Linux 11 (bullseye) (x86…
-
### Your current environment
```text
Docker container following build_from _source instruction
```
### How would you like to use vllm
I want to experiment how chunked_prefill can increase…
-
* sxm-client version: 0.2.8
* Python version: 3.8
* Operating System: Ubuntu 20.04
Authentication seems to be broken?
ERROR:/usr/local/lib/python3.8/dist-packages/sxm/client.py:An Exception oc…
-
* sxm-discord version:0.2.1
* Python version: 3.10.6
* Operating System: Ubuntu 22.04
### Description
Installed with PIP created a config file but can't get the player class to work.
### Wh…
-
* sxm-client version: 0.2.8
* Python version: 3.10
* Operating System: MacOS Big Sur
I have an automated recorder that i have made that i use to do analog recordings and I'm adding digital record…