Conditional Scale Entropy: A Wavelet-Derived Tool for Mechanistic Interpretability of Metaphor Processing in Transformers
This paper introduces Conditional Scale Entropy (CSE), a wavelet-derived measure of how transformer computation engages across frequency scales at each layer, and applies it to study metaphor processing in decoder-only language models. The authors prove CSE is invariant to update magnitude, isolating structural computation patterns from intensity. Across architectures ranging from GPT-2 (124M) to LLaMA-2 7B and GPT-oss 20B, metaphorical tokens consistently produce higher spectral breadth than literal tokens in early-to-mid layers, with the effect surviving permutation correction and specificity controls. The work establishes multi-scale coordination as a consistent mechanistic signature of metaphorical language processing and positions CSE as a general interpretability tool for cross-depth structure in transformers.
Related guides (3)
Related events (8)
Transformer embeddings shown to intrinsically encode Russell's circumplex model of emotion geometry
A new arXiv paper investigates whether Transformer-based text and speech encoders (RoBERTa, wav2vec 2.0) recover the geometric structure of Russell's circumplex model of affect — a valence-arousal topology from psychology. Experiments on naturalistic datasets (MSP-Podcast) and LLM-generated stimuli show that multimodal fusion achieves perfect topological alignment with Russell's primary emotion ordering, and zero-shot generic text embeddings place fine-grained emotion terms near their human-mapped coordinates. The authors argue this structure is intrinsically encoded in the representations rather than being an artifact of labeling, bridging psychological theory and representation learning.
Comparative study of semantic geometry in transformer embeddings vs. graph-based lexical models
A preprint from arXiv compares the geometric and topological properties of transformer-based vector embeddings (CamemBERT) against lexical co-occurrence graphs for representing semantic structure. Applied to a French civic debate corpus, the study finds similar local topology but divergent global structure between the two approaches. The authors argue graph-based models offer more interpretable semantic organization and suggest graphs could guide neural architectures toward more stable, interpretable convergence.
Hyperfitting Explained: Terminal Geometric Expansion in Final Transformer Layers Drives Diversity Gains
This paper investigates the 'hyperfitting' phenomenon—where fine-tuning LLMs to near-zero loss on small datasets improves open-ended generation and reduces repetition—and demonstrates it is mechanistically distinct from temperature scaling. Entropy-matched control experiments falsify both the temperature-equivalence and static vocabulary reweighting hypotheses, instead localizing the effect to a 'Terminal Expansion' in the final transformer block where feature-space dimensionality expands by ~80.8 dimensions, enabling promotion of deep-tail tokens via context-dependent rank reordering. The authors introduce Late-Stage LoRA, a targeted fine-tuning strategy updating only the final 5 layers, achieving robust generation with minimal parameter updates.
Dynamic short convolutions yield 1.33–1.60× compute advantage over standard Transformers
A new arXiv preprint introduces dynamic short convolutions as an architectural primitive for Transformers, using input-dependent filters to combine locality bias with increased expressivity. Experiments across 150M–2B parameter language models show consistent perplexity improvements over standard Transformers and static convolution variants, with scaling-law fits indicating a 1.33× compute advantage when applied to key/query/value vectors and 1.60× when added after every linear layer. The technique also improves linear RNNs (Mamba-2, Gated DeltaNet) and mixture-of-experts architectures, with custom Triton kernels making training practical.
LEAF-X: Entropy-guided explainability framework for transformer-based ASR models
Researchers introduce LEAF-X (Listening with Entropy-guided Attention for Faithful explainability), a model-intrinsic XAI framework for transformer-based automatic speech recognition systems like Whisper. The method combines entropy-guided attention weighting, multi-layer attention rollout, and optional causal ablations to produce sparse token-to-frame attributions. Evaluations show 32% improved faithfulness and 35-39% stronger locality/sparsity compared to perturbation-based explainers and raw attention maps, enabling more auditable ASR.
Overview of Natively Supported Quantization Schemes in 🤗 Transformers
This Hugging Face blog post surveys the quantization methods natively integrated into the Transformers library as of September 2023, covering schemes such as GPTQ, bitsandbytes (LLM.int8, NF4), and related techniques. It explains how each method works, their trade-offs in terms of memory reduction and inference speed, and how practitioners can apply them via the Transformers API. The post serves as a practical reference for deploying large language models under memory constraints.
FASE: Fast Adaptive Semantic Entropy for uncertainty quantification in multi-agent code generation
Researchers introduce Fast Adaptive Semantic Entropy (FASE), a metric for approximating functional correctness in LLM-generated code using minimum spanning trees of structural and semantic dissimilarity graphs, replacing costly LLM-driven equivalence checks. Evaluated on HumanEval and BigCodeBench with Qwen3-Embedding-8B, FASE achieves a 25% improvement in Spearman correlation and 19% increase in ROCAUC over prior semantic entropy methods. Critically, it requires only ~0.3% of the runtime cost of traditional semantic entropy approaches, making it practical for real-world multi-agent workflows.
Extracting Concepts from GPT-4: 16 Million Patterns via Sparse Autoencoders
OpenAI applied scaled sparse autoencoders (SAEs) to GPT-4 to automatically identify approximately 16 million interpretable features or patterns in the model's internal computations. This represents a significant scaling of mechanistic interpretability techniques previously demonstrated on smaller models. The work advances the ability to understand what concepts and representations large frontier models encode internally.