Collected molecules will appear here. Add from search or explore.
A comparative analysis and implementation of emotion vector extraction and steering methods (generation-based vs. comprehension-based) across multiple Small Language Model (SLM) architectures.
citations
0
co_authors
1
The project is a fresh academic research release (4 days old, 0 stars) providing a systematic comparison of emotion representations in SLMs. While the methodology (RepE) is established, applying it across 9 model families to compare generation vs. comprehension extraction is a novel contribution to SLM interpretability. It lacks defensibility as an open-source project but serves as a valuable reference for developers building emotion-aware edge applications.
TECH STACK
INTEGRATION
reference_implementation
READINESS