Entity · paper

Beyond Independent Labels: Schwartz-Geometry Decoding for Human Value Detection

paperactiveprovisionalbeyond-independent-labels-schwartz-geometry-decoding-for-human-value-detection-e81c6a70·1 events·first seen 44h ago

Aliases: Beyond Independent Labels: Schwartz-Geometry Decoding for Human Value Detection

Co-occurring entities

Qwen2.5-7B-Instruct-1M DeBERTa-v3

More like this (12)

Who Needs Labels? Adapting Vision Foundation Models With the Metadata You Already Have Measuring Human Value Expression in Social Media Texts: Calibrated LLM Annotation and Encoder Transfer lambda-scaled structural decoding Schwartz Value Theory Supervised Semantic Differential Scaling LLM Reasoning from Minimal Labels: A Semi-Supervised Framework with a Lightweight Verifier Schwartz's Theory of Basic Human Values Influcoder: Distilling Decoders' Gradient Influence Rankings into an Encoder for Data Attribution Seeing Is Not Sharing: Some Vision-Language Models Overestimate Common Ground in Asymmetric Dialogue Selective Disclosure Watermarking for Large Language Models The Value Axis: Language Models Encode Whether They're on the Right Track The Geometry of Updates: Fisher Alignment at Vocabulary Scale

Recent events (1)

3arXiv · cs.CL·44h ago·source ↗

Schwartz-Geometry Decoding improves coherence in human value detection without sacrificing F1

A new arXiv preprint proposes injecting the circular Schwartz value continuum as an output-space geometry into multi-label classifiers for human value detection. The authors compare training-time geometry-aware objectives against a post-hoc energy decoder on a DeBERTa-v3-base model, finding that the decoder improves label-set coherence with the theoretical continuum without degrading Macro-F1 or Micro-F1. Training-time geometry injection yields only marginal gains, no better than a random ordering. A Qwen2.5-72B-Instruct diagnostic shows that supplying the continuum at inference shifts behavior but does not match supervised structured prediction.

Evaluation and Benchmarking Qwen2.5-7B-Instruct-1M Beyond Independent Labels: Schwartz-Geometry Decoding for Human Value Detection DeBERTa-v3