InternVideo
InternVideo copied to clipboard
InternVideo-MM-L-14 pretraining datasets
Hi! Could you kindly clarify about InternVideo v1 released models?
On what was InternVideo-MM-L-14 pretrained? The GitHub page says WebVid10M+Self-collected (14M), while in the paper it’s WebVid2M, WebVid10M, and HowTo100M. Was the released model also fine-tuned on Kinetics 710?