nmn2 icon indicating copy to clipboard operation
nmn2 copied to clipboard

raise AttributeError('Layer %s has no keyword argument %s=%s' % (param_type, k, v))

Open haibin894609937 opened this issue 7 years ago • 3 comments

Hi,I meet an error,,,, 2017-07-21 12:17:32,576 DEBUG [root] prepared indices 2017-07-21 12:55:22,728 DEBUG [root] computed image feature normalizers 2017-07-21 12:55:22,789 DEBUG [root] using cvpr chooser 2017-07-21 12:57:36,539 INFO [root] TRAIN2014, VAL2014: 2017-07-21 12:57:36,540 INFO [root] 369861 items 2017-07-21 12:57:36,540 INFO [root] 2002 answers 2017-07-21 12:57:36,541 INFO [root] 877 predicates 2017-07-21 12:57:36,541 INFO [root] 3591 words 2017-07-21 12:57:36,541 INFO [root] 2017-07-21 12:57:56,250 INFO [root] TEST-DEV2015: 2017-07-21 12:57:56,250 INFO [root] 60864 items 2017-07-21 12:57:56,250 INFO [root] 2002 answers 2017-07-21 12:57:56,251 INFO [root] 877 predicates 2017-07-21 12:57:56,251 INFO [root] 3591 words 2017-07-21 12:57:56,251 INFO [root] 2017-07-21 12:59:02,813 INFO [root] TEST2015: 2017-07-21 12:59:02,813 INFO [root] 244302 items 2017-07-21 12:59:02,814 INFO [root] 2002 answers 2017-07-21 12:59:02,814 INFO [root] 877 predicates 2017-07-21 12:59:02,814 INFO [root] 3591 words 2017-07-21 12:59:02,814 INFO [root] Traceback (most recent call last): File "main.py", line 255, in main() File "main.py", line 31, in main do_iter(task.train, model, config, train=True) File "main.py", line 97, in do_iter batch_data, model, config, train, vis) File "main.py", line 118, in do_batch predictions = forward(data, model, config, train, vis) File "main.py", line 170, in forward dropout=(train and config.opt.dropout), deterministic=not train) File "/disk3/hbliu/vqa/nmn2/models/nmn.py", line 440, in forward question_hidden = self.forward_question(question_data, dropout) File "/disk3/hbliu/vqa/nmn2/models/nmn.py", line 648, in forward_question bottoms=[word], param_names=[wordvec_param])) File "/home/hbliu/disk3/vqa/nmn2/apollocaffe/python/apollocaffe/layers/caffe_layers.py", line 203, in init super(Wordvec, self).init(self, name, kwargs) File "/home/hbliu/disk3/vqa/nmn2/apollocaffe/python/apollocaffe/layers/layer_headers.py", line 6, in init self.parse(sublayer, name, kwargs) File "/home/hbliu/disk3/vqa/nmn2/apollocaffe/python/apollocaffe/layers/layer_headers.py", line 57, in parse raise AttributeError('Layer %s has no keyword argument %s=%s' % (param_type, k, v)) AttributeError: Layer Wordvec has no keyword argument vocab_size=3591

haibin894609937 avatar Jul 21 '17 06:07 haibin894609937

@jacobandreas

haibin894609937 avatar Jul 21 '17 13:07 haibin894609937

Are you using my ApolloCaffe fork? Haven't seen this error before. In any case this codebase is kind of out-of-date---you're better off using https://github.com/ronghanghu/n2nmn.

jacobandreas avatar Jul 26 '17 23:07 jacobandreas

I use your apollocaffe fork , thanks for your advice

haibin894609937 avatar Jul 27 '17 02:07 haibin894609937