-
Hello, thank you for your work!
I noticed that you are using 8 Nvidia V100 to run the model. I would like to know if the V100 has 16GB or 32GB of video memory? If the translation part is removed, can…
-
I ran mem_spd_test.py and got the following error:
RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cuda:1!
I did not make any changes except …
-
Hi there! I am trying to run evaluation with multiple GPUs. I have ran everything in './scripts/minimal_example.sh'. Running on 1 GPU works perfectly, however when I pass in more than 1 GPU, all Huggi…
-
Hi,
in train.sh the configuration sets ns, nv, max radius and batch_size to 16, 4, 5 and 2, but in the args file in the checkpoint folder for the best model provided these parameters are set to dif…
-
### Prerequisites
- [X] I have read the [documentation](https://hf.co/docs/autotrain).
- [X] I have checked other issues for similar problems.
### Backend
Local
### Interface Used
CLI
…
-
I have 2 4090 and I want to merge 8 7B models. But I get out of memory.
And only one GPU is used. So, how to use 2 4090 simultaneously.
Or there is other method to solve this?
-
do we support the old V100 GPUs with FlexAttention ?
-
Hello, is there any way to run a inference with 2 or more GPUs?
-
```
100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 80/80 [00:00
-
The situation on AMD GPUs for develop is a mess. Here are a list of crashes I am seeing on LUMI-G (MI250x) when running the benchmark driver (only `-t 79 --norms` passed):
- You usually have to build…