Open EricMichaelSmith opened 4 years ago
This issue has not had activity in 30 days. Marking as stale.
@EricMichaelSmith this seems done right?
@stephenroller Hmm there are still two different .batchify_image_features()
methods in ImagePolyencoderAgent
and ImageSeq2seqAgent
, so as far as I know this is still an open issue. @klshuster may be able to speak more on this because he's been refactoring some of this code recently
This is still not done, however it seems like it'd be a good change to do
Currently,
ImageSeq2seqAgent
andImagePolyencoderAgent
have two separate.batchify_image_features()
methods for preparing image features for encoding. The main advantage ofImagePolyencoderAgent
's implementation of this method is that the images are packed into a Tensor for data_parallel. It's worth testing whetherImageSeq2seqAgent
really requires a separate.batchify_image_features()
method now: if not, it should be removed, andImagePolyencoderAgent.batchify_image_features()
should be moved intoTorchImageAgent
.