san-torch
san-torch copied to clipboard
Torch implementation for Stacked Attention Networks
I got 32g cpu ram and 2 gpu (gtx1080 8G) on my machine. why it cannot afford 15G memory? rzai@rzai00:~/prj/san-torch/prepro$ CUDA_VISIBLE_DEVICES=1 th prepro_img_vgg.lua -input_json ../data/vqa_data_prepro.json -image_root /home/rzai/mscoco.org-visualqa.org/ -cnn_proto /home/rzai/VGG_ILSVRC_19_layers_deploy.prototxt -cnn_model...
rzai@rzai00:~/prj/san-torch/data$ python vqa_preprocess.py split 2 usage: vqa_preprocess.py [-h] [--download DOWNLOAD] [--split SPLIT] vqa_preprocess.py: error: unrecognized arguments: split 2 rzai@rzai00:~/prj/san-torch/data$ python vqa_preprocess.py --split 2 parsed input parameters: { "download": 1, "split":...
Hi, Thanks for sharing the code. I ran the code with default settings and only got 45% accuracy (in the paper authors have mentioned over 57% accuracy). I noticed that...