A training-free two-stage token compression method speeds up video language model inference by about 1.5x and cuts memory use, while keeping accuracy close to the uncompressed model.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
DyCoke: Dynamic Compression of Tokens for Fast Video Large Language Models
A training-free two-stage token compression method speeds up video language model inference by about 1.5x and cuts memory use, while keeping accuracy close to the uncompressed model.