Hybrid JEMs at intermediate generative-discriminative balance maximize human alignment on perceptual similarity, gloss, uncertainty, robustness, cue conflict, and feature attribution benchmarks.
• All JEMs were trained using mixed precision (via PyTorch AMP) and torch.compile to improve training efficiency
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Not Too Generative, Not Too Discriminative: The Human Alignment Sweet Spot
Hybrid JEMs at intermediate generative-discriminative balance maximize human alignment on perceptual similarity, gloss, uncertainty, robustness, cue conflict, and feature attribution benchmarks.