pytorch > 1?

[sharifza]

is there any pytorch > 1 fork of this work? There is no way to run it in this shape with newer GPUs such as RTX as they do not support cuda versions less than 10 while pytorch versions < 1 don't support cuda10 !

[zaynmi]

I have the same problem with 2080Ti, throwing out error

RuntimeError: cuda runtime error (48) : no kernel image is available for execution on the device at /home/tang/pytorch/torch/lib/THC/generic/THCTensorMathPairwise.cu:21

Looking for solution

[sharifza]

Yes, exactly the same error.

[truetone2022]

Anyone solved this problem? I meet it too.

[TP0609]

is there any pytorch > 1 fork of this work? There is no way to run it in this shape with newer GPUs such as RTX as they do not support cuda versions less than 10 while pytorch versions < 1 don't support cuda10 !

Hello !
Is your problem solved? I am using RTX2070, not supporting version less than cuda10.0, but cuda10.0 needs to use pytorch>1.0, can this program be run with pytorch>1.0? !
Is your problem solved? I am using RTX2070, not supporting version less than cuda10.0, but cuda10.0 needs to use pytorch>1.0, can this program be run with pytorch>1.0?

[GreenTeaHua]

torchvision.ops.nms(boxes, scores, iou_threshold)
Performs non-maximum suppression (NMS) on the boxes according
to their intersection-over-union (IoU).

NMS iteratively removes lower scoring boxes which have an
IoU greater than iou_threshold with another (higher scoring)
box.

Parameters
----------
boxes : Tensor[N, 4])  bounding boxes坐标. 格式：(x1, y1, x2, y2)
    boxes to perform NMS on. They
    are expected to be in (x1, y1, x2, y2) format
scores : Tensor[N]   bounding boxes得分
    scores for each one of the boxes
iou_threshold : float IoU过滤阈值
    discards all overlapping
    boxes with IoU < iou_threshold

Returns
-------
keep : Tensor
    int64 tensor with the indices
    of the elements that have been kept
    by NMS, sorted in decreasing order of scores
    NMS过滤后的bouding boxes索引（降序排列）
"""

[progvasu]

is there any pytorch > 1 fork of this work? There is no way to run it in this shape with newer GPUs such as RTX as they do not support cuda versions less than 10 while pytorch versions < 1 don't support cuda10 !

Hello !
Is your problem solved? I am using RTX2070, not supporting version less than cuda10.0, but cuda10.0 needs to use pytorch>1.0, can this program be run with pytorch>1.0? !
Is your problem solved? I am using RTX2070, not supporting version less than cuda10.0, but cuda10.0 needs to use pytorch>1.0, can this program be run with pytorch>1.0?

Hey,

Where you able to figure this out? Thanks!

[Spichon]

Hi guys,
So i did figure out how to go through the "no kernel image is available for execution" issue if you are running the code with a RTX2080.
You have to edit the file /lib/lstm/highway_lstm/make.sh with the following value : CUDA_MODELS=(75).

Nevertheless, pytorch 0.4.1 can't deal with cuda10. You can refer to the following zoo model : https://download.pytorch.org/whl/cu100/torch_stable.html

The solution would be to get this code working with torch > 1.0.

[TP0609]

charge 2070 to GTX1080TI (11G), The problem is solved! because the 2070 insufficient memory At 2020-02-24 23:42:07, "Simon Pichon" <notifications@github.com> wrote: Hi guys, So i did figure out how to go through the "no kernel image is available for execution" issue if you are running the code with a RTX2080. You have to edit the file /lib/lstm/highway_lstm/make.sh with the following value : CUDA_MODELS=(75). Nevertheless, pytorch 0.4.1 can't deal with cuda10. You can refer to the following zoo model : https://download.pytorch.org/whl/cu100/torch_stable.html The solution would be to get this code working with torch > 1.0. — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.