The Flow has not summarised this story yet — read it at arXiv Computation and Language.
Sparse autoencoders (SAEs) are proposed to extract numerous features from large language model (LLM) representations, yet explaining these features still relies primarily on external observation. This reliance leads to superficial explanations inferred from observed model behavior and computational inefficiency from collecting such behavioral evidence at scale.
arXiv:2604. 15280v2 Announce Type: replace-cross Abstract: Understanding emotions is a fundamental ability for intelligent systems to be able to interact with humans.
The paper introduces DAN, a training‑free inference‑time framework that improves affective reasoning in multimodal large language models. It combines a Hierarchical Emotional Reasoning Chain (HERC) to better capture fine‑grained visual cues and a Contrastive Discriminative Visual Pruning (CDVP) module to isolate discriminative tokens for semantically similar emotions. Experiments show significant gains, notably a +10.47% improvement on the WebEmo25 benchmark with Qwen3‑VL‑8B‑Instruct.
The study examines how emotions are represented across layers of large language models (LLMs) by probing eight 1B–9B open‑weight models on three datasets (Twitter, Reddit, autobiographical narratives). It finds that the optimal probing layer varies systematically with the dataset, moving from near‑input layers to deeper layers, and that targeted forward‑pass interventions on these layers degrade performance more than random interventions. Additionally, the selected layers transfer across datasets and emotion categories, and early‑exit representations from these layers outperform full‑depth exits by an average of 6.9 percentage points.
arXiv:2609.01322v1 Announce Type: cross Abstract: In many settings, studying causal questions based on text data requires adjusting for confounding information within texts. Yet there is a tradeoff i...
arXiv:2610. 29430v1 Announce Type: new Abstract: Multimodal multi-label emotion recognition is important for affect-aware systems, but models trained with complete text, visual, and acoustic streams often degrade sharply when some streams are unavailable at deployment.