A 1B-parameter VLA with joint (non-autoregressive) action prediction claims OpenVLA-comparable training behavior at 4-7x lower inference time, but supports this only with early training curves.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.RO 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
EdgeVLA: Efficient Vision-Language-Action Models
A 1B-parameter VLA with joint (non-autoregressive) action prediction claims OpenVLA-comparable training behavior at 4-7x lower inference time, but supports this only with early training curves.