AttributeError: 'jittor_core.Var' object has no attribute 'sort'

[YongZh958]

Wed Aug 25 15:01:57 2021 Validating....
0%| | 33/41429 [01:43<36:05:37, 3.14s/it]
Traceback (most recent call last):
File "tools/run_net.py", line 47, in
main()
File "tools/run_net.py", line 40, in main
runner.run()
File "/root/ZY/JDet-master/python/jdet/runner/runner.py", line 85, in run
self.val()
File "/root/anaconda3/envs/Jit/lib/python3.7/site-packages/jittor/init.py", line 89, in inner
ret = func(*args, **kw)
File "/root/anaconda3/envs/Jit/lib/python3.7/site-packages/jittor/init.py", line 257, in inner
ret = func(*args, **kw)
File "/root/ZY/JDet-master/python/jdet/runner/runner.py", line 153, in val
result = self.model(images,targets)
File "/root/anaconda3/envs/Jit/lib/python3.7/site-packages/jittor/init.py", line 737, in call
return self.execute(*args, **kw)
File "/root/ZY/JDet-master/python/jdet/models/networks/s2anet.py", line 35, in execute
outputs = self.bbox_head(features, targets)
File "/root/anaconda3/envs/Jit/lib/python3.7/site-packages/jittor/init.py", line 737, in call
return self.execute(*args, **kw)
File "/root/ZY/JDet-master/python/jdet/models/roi_heads/s2anet_head.py", line 627, in execute
return self.get_bboxes(*outs,self.parse_targets(targets,is_train=False))
File "/root/ZY/JDet-master/python/jdet/models/roi_heads/s2anet_head.py", line 538, in get_bboxes
scale_factor, cfg, rescale)
File "/root/ZY/JDet-master/python/jdet/models/roi_heads/s2anet_head.py", line 597, in get_bboxes_single
cfg.max_per_img)
File "/root/ZY/JDet-master/python/jdet/ops/nms_rotated.py", line 589, in multiclass_nms_rotated
_, inds = scores.sort(descending=True)
AttributeError: 'jittor_core.Var' object has no attribute 'sort'

训练了6小时左右出现这个错误

[li-xl]

python/jdet/ops/nms_rotated.py的589行改成inds,_ = scores.argsort(descending=True)就可以了，我已经更新了JDet，修复了这个问题，感谢支持！

[YongZh958]

[w 0825 16:15:52.108481 20 init.py:980] load parameter fc.weight failed ...
[w 0825 16:15:52.108591 20 init.py:980] load parameter fc.bias failed ...
[w 0825 16:15:52.108702 20 init.py:998] load total 267 params, 2 failed

[w 0825 16:16:02.075836 20 cudnn_conv_Tx:float32__Ty:float32__Tw:float32__XFORMAT:abcd__WFORMAT:oihw__YFORMAT:abcd__J...hash:798cb5ed49dadaa2_op.cc:200] forward_ algorithm cache is full
[w 0825 16:16:02.650392 20 cudnn_conv_backward_w_Tx:float32__Ty:float32__Tw:float32__XFORMAT:abcd__WFORMAT:oihw__YFOR...hash:c41e3d43aa5d4cf7_op.cc:199] backward w algorithm cache is full
[w 0825 16:16:04.062816 20 cudnn_conv_backward_x_Tx:float32__Ty:float32__Tw:float32__XFORMAT:abcd__WFORMAT:oihw__YFOR...hash:74f24b7a5fa4fe17_op.cc:201] backward x algorithm cache is full

请问在加载模型时，出现这些警告会影响模型的训练吗

[li-xl]

不会的，这是jittor在搜索最快的运行算法，只是搜索的缓存满了

[YongZh958]

好的，谢谢帮助