A semi-supervised conditional flow matching method, guided by a new inter-modal bridge cost, aligns latent spaces of distinct pre-trained models using only a small set of paired samples.
For each, we had three variations–small, medium, and large (Table 1)– with the SiLU activation function applied after every layer in all models
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Model alignment using inter-modal bridges
A semi-supervised conditional flow matching method, guided by a new inter-modal bridge cost, aligns latent spaces of distinct pre-trained models using only a small set of paired samples.