FEIFEIEIAr
FEIFEIEIAr
For your problem, you should change 'your file path' to your own path, like '/home/.../your_vcf.vcf', in line 78 of Generate_Deletion_Image.py. However, I read some codes of DeepSV, I found a...
> Thanks for the detailed thread. Might be a good candidate for [discussions](https://github.com/huggingface/diffusers/discussions). > > Have you tried warming up the GPU before reporting the numbers? If not, I would...
> To my knowledge, PyTorch makes calculations asynchronously in GPU. So, could you put `torch.cuda.synchronise()` before each `time.time()`? Also, `time.perf_counter()` is recommended most of the time instead of `time.time()`. Thanks!...
> 这个和 #387 大概率是一个问题——你尝试了手动forward port么?这里可能是需要处理一下,但是我现在人在国内,得回到家里之后才有时间去试一下,我没怎么用过remote server。 感觉应该是port的问题,应该是9001端口有什么进程占用了。我在后面一段时间的使用中都没有发生这个 issue 里面的问题了。 我今天试了一下: 1、如果先在 terminal 里使用 `vizviewer ./result.json` ,再右键 result.json,使用 View with VizTracer ,就会一直没响应。(就是这个issue的情况) 2、先右键 result.json,使用 View with VizTracer ,再在 terminal 里使用 `vizviewer...
> 你好,感谢您的贡献,这里是需要使用 apex 的 FusedRMSNorm,不然训练的时候会爆显存,推理的时候可不用。 您好! - 现在的model.py是只引用了apex。因为readme里面写的apex是可选的,如果没装apex的话,这样直接推理会报错  - components.py里是这样import的方式才是正确的。所以改成在model里直接import components就行了。  或者您可能需要修改readme,或者加是否引用apex的参数
You can find the **File** "/home/cai/.local/lib/python3.10/site-packages/torchstat/model_hook.py", line 65, in wrap_call and this line of code `np.array([parameter_quantity], dtype=np.long))` change `dtype=np.long` to `dtype=np.int64` or `dtype=np.int32` or `dtype=np.int_`. I don't think changing this...