Closed lntzm closed 1 year ago
Hello, for ActivityNet Captions dataset, we use the glove.pkl
file provided by SCN (https://github.com/ikuinen/semantic_completion_network/blob/master/data/activitynet/vocab.pkl). However, they do not provide this file for Charades-STA dataset. Thus, we build this file from the standard glove.6B.300d
.
thanks for your reply!
Hello, for ActivityNet Captions dataset, we use the
glove.pkl
file provided by SCN (https://github.com/ikuinen/semantic_completion_network/blob/master/data/activitynet/vocab.pkl). However, they do not provide this file for Charades-STA dataset. Thus, we build this file from the standardglove.6B.300d
.
Hello ,could you please provide the code to build the 'glove.pkl' file for Charades-STA dataset? I really need it. Look forward to your reply!
Hello, for ActivityNet Captions dataset, we use the
glove.pkl
file provided by SCN (https://github.com/ikuinen/semantic_completion_network/blob/master/data/activitynet/vocab.pkl). However, they do not provide this file for Charades-STA dataset. Thus, we build this file from the standardglove.6B.300d
.Hello ,could you please provide the code to build the 'glove.pkl' file for Charades-STA dataset? I really need it. Look forward to your reply!
I have uploaded the vocab.py
. Please check it.
Hi Author, the provided
glove.pkl
for each dataset is different as it only contains the necessary vocabulary. Specially, it contains:I found the value of vectors is different from the standard
glove.840B.300d
. I want to known how to generate the glove vectors and has it finetuned already in each dataset? How can I build the vectors for other datasets? Thanks!