David Marx

not sure what the etiology is here, but figured we should document it and maybe centralize repair discussion here since it's apparently a known issue. I think with CARPFilip we...

Stream ModelOutputs

10

# What does this PR do? (NB: PR is still work in progress, but close to ready) The "streaming" feature for text generation is currently constrained to streaming token ids....

add a TOC

3

for real, let's just add a TOC. shit, could probably even decompose this into something workbench could process ...ooo what if i pull each of these articles down, parse them,...

desperately needs a better structure

1

* improve at-a-glance readability * facilitate exploration * facilitate topic navigation * distinguish "core" papers for a topic in such a way that we have room to add additional relevant...

I might actually be able to just straight up embed the PDFs directly using panel: https://panel.holoviz.org/reference/panes/PDF.html?highlight=pdf so that's neat

model architectures and pretrained models to support

36

## installable - [ ] https://github.com/salesforce/LAVIS - https://github.com/salesforce/BLIP - https://github.com/salesforce/ALBEF - [ ] https://github.com/facebookresearch/multimodal - FLAVA - LateFusion - ALBEF - MDETR - OMNIVORE - video-gpt - [ ] https://github.com/ai-forever/ru-clip...

https://github.com/archinetai/surgeon-pytorch

I suspect this is the issue with clip-fa too

`pip install clip-anytorch` https://github.com/rom1504/CLIP