search

wkcn / AttentionSampler

Attention-based sampler in TASN (Trilinear Attention Sampling Network)
23 stars 0 forks source link

Hi, i wonder how to use trained model to test accuracy or to classify my images. #5

Open qkyo opened 4 years ago

qkyo commented 4 years ago

Hi, i wonder how to use trained model to test accuracy or to classify my images. The original author seems to too busy to take care of his github. if you have done any work related it, plz give me some guidance. thank u so much

qkyo commented 4 years ago

it seems like there are no test file in this project, or maybe i ignore it. Thanks again!

qkyo commented 4 years ago

btw, sth seems to be wrong. I have added your files,

Error in CustomOp.forward: Traceback (most recent call last): File "E:\Anaconda3\envs\python37\lib\site-packages\mxnet-1.3.1-py3.7.egg\mxnet\operator.py", line 987, in forward_entry aux=tensors[4]) File "E:\Anaconda3\envs\python37\lib\site-packages\mobulaop-2.32-py3.7.egg\mobula\glue\mx.py", line 103, in forward out = self._forward(*in_data) File "./AttentionSampler/attention_sampler\attention_sampler.py", line 45, in forward attxi = F.cumsum(attx, 1) AttributeError: module 'mxnet.ndarray' has no attribute 'cumsum'

this problem arises.I wonder if this is a problem with the mxnet version,

vb123er951 commented 4 years ago

Hi @qkyo , the author has update the code to support old version of mxnet, maybe you can try again.

Also I'm wondering how to do testing in TASN by only use master-net... @wkcn can you give some suggestions?

wkcn commented 4 years ago

Hi @vb123er951 , I am not familiar with it.

You can ask the author @Heliang-Zheng.