Zeroth-order optimization estimates parameter updates by perturbing model weights and comparing objective values, avoiding backpropagation and its activation-memory costs.
Caching frozen model states makes forward-only robot adaptation much faster
VLA-ZO accelerates memory-efficient adaptation by updating only action components and reusing vision-language conditioning across parameter perturbations.
Industry
Jaemin Kim · Jiahn Kim · Taesik Gong
UNIST
Research Digest··2 min read
Kim and colleagues present VLA-ZO, a forward-only method for adapting vision-language-action models to deployment shifts using one target-environment demonstration.
Why this paper
From UNIST
In one line
VLA-ZO adapts vision-language-action models to distribution shifts 25x faster than baseline zeroth-order optimization.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ✓Reports numbers on named benchmarks
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§