HERO: Hierarchical Encoder for Video+Language Omni-representation Pre-training
30841Knowledge Triples
| Source Entity | Relationship | Target Entity |
|---|---|---|
|
Hero
Task |
includes |
Frame Order Modeling
Task |
|
Hero
Task |
includes |
Masked Frame Modeling
Method |
|
Hero
Task |
includes |
Masked Language Modeling
Method |
|
Hero
Task |
includes | |
|
Hero
Task |
introduces |
How2QA
Material |
|
Hero
Task |
introduces |
How2R
Material |
|
Hero
Task |
trainedOn |
HowTo100M
Material |
|
Hero
Task |
trainedOn |
TV datasets
Material |
|
Hero
Task |
uses |
Cross modal Transformer
Method |
|
Hero
Task |
uses |
Temporal Transformer
Method |