Copertina

Circuit Fingerprints: How Answer Tokens Encode Their Geometrical Path

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Saurez, Andres, Sengar, Neha, Har, Dongsoo
Natura:	Preprint
Pubblicazione:	2026
Soggetti:	Machine Learning Computation and Language
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

Why Linear Interpretability Works: Invariant Subspaces as a Result of Architectural Constraints
di: Saurez, Andres, et al.
Pubblicazione: (2026)

Continuous Adversarial Text Representation Learning for Affective Recognition
di: Son, Seungah, et al.
Pubblicazione: (2025)

Steering Sparse Autoencoder Latents to Control Dynamic Head Pruning in Vision Transformers (Student Abstract)
di: Lee, Yousung, et al.
Pubblicazione: (2026)

PAST: Phonetic-Acoustic Speech Tokenizer
di: Har-Tuv, Nadav, et al.
Pubblicazione: (2025)

On the Semantic and Syntactic Information Encoded in Proto-Tokens for One-Step Text Reconstruction
di: Bondarenko, Ivan, et al.
Pubblicazione: (2026)

Understanding Token Probability Encoding in Output Embeddings
di: Cho, Hakaze, et al.
Pubblicazione: (2024)

Optimized Multi-Token Joint Decoding with Auxiliary Model for LLM Inference
di: Qin, Zongyue, et al.
Pubblicazione: (2024)

Do LLMs Encode Functional Importance of Reasoning Tokens?
di: Singh, Janvijay, et al.
Pubblicazione: (2026)

Discursive Circuits: How Do Language Models Understand Discourse Relations?
di: Miao, Yisong, et al.
Pubblicazione: (2025)

Parity-Aware Byte-Pair Encoding: Improving Cross-lingual Fairness in Tokenization
di: Foroutan, Negar, et al.
Pubblicazione: (2025)

Student Answer Forecasting: Transformer-Driven Answer Choice Prediction for Language Learning
di: Gado, Elena Grazia, et al.
Pubblicazione: (2024)

Language Model Cascades: Token-level uncertainty and beyond
di: Gupta, Neha, et al.
Pubblicazione: (2024)

Detecting and Suppressing Reward Hacking with Gradient Fingerprints
di: Wang, Songtao, et al.
Pubblicazione: (2026)

On the Robustness of Answer Formats in Medical Reasoning Models
di: Taveekitworachai, Pittawat, et al.
Pubblicazione: (2025)

On Next-Token Prediction in LLMs: How End Goals Determine the Consistency of Decoding Algorithms
di: Trauger, Jacob, et al.
Pubblicazione: (2025)

Promote, Suppress, Iterate: How Language Models Answer One-to-Many Factual Queries
di: Yan, Tianyi Lorena, et al.
Pubblicazione: (2025)

Antidistillation Fingerprinting
di: Xu, Yixuan Even, et al.
Pubblicazione: (2026)

Found in the Middle: How Language Models Use Long Contexts Better via Plug-and-Play Positional Encoding
di: Zhang, Zhenyu, et al.
Pubblicazione: (2024)

Representation Consistency for Accurate and Coherent LLM Answer Aggregation
di: Jiang, Junqi, et al.
Pubblicazione: (2025)

How Do Transformers Learn to Associate Tokens: Gradient Leading Terms Bring Mechanistic Interpretability
di: Im, Shawn, et al.
Pubblicazione: (2026)

The Truncation Blind Spot: How Decoding Strategies Systematically Exclude Human-Like Token Choices
di: Arias, Esteban Garces, et al.
Pubblicazione: (2026)

DropBP: Accelerating Fine-Tuning of Large Language Models by Dropping Backward Propagation
di: Woo, Sunghyeon, et al.
Pubblicazione: (2024)

Graph-based Molecular In-context Learning Grounded on Morgan Fingerprints
di: Al-Lawati, Ali, et al.
Pubblicazione: (2025)

On the Geometry of Positional Encodings in Transformers
di: Cirrincione, Giansalvo
Pubblicazione: (2026)

Judge Circuits
di: Feldhus, Nils, et al.
Pubblicazione: (2026)

Fingerprint Vector: Enabling Scalable and Efficient Model Fingerprint Transfer via Vector Addition
di: Xu, Zhenhua, et al.
Pubblicazione: (2024)

InterrogateLLM: Zero-Resource Hallucination Detection in LLM-Generated Answers
di: Yehuda, Yakir, et al.
Pubblicazione: (2024)

Uncertainty-Aware Answer Selection for Improved Reasoning in Multi-LLM Systems
di: Agrawal, Aakriti, et al.
Pubblicazione: (2025)

How Important Is Tokenization in French Medical Masked Language Models?
di: Labrak, Yanis, et al.
Pubblicazione: (2024)

X-Token: Projection-Guided Cross-Tokenizer Knowledge Distillation
di: Sreenivas, Sharath Turuvekere, et al.
Pubblicazione: (2026)

TokenShapley: Token Level Context Attribution with Shapley Value
di: Xiao, Yingtai, et al.
Pubblicazione: (2025)

Token Distillation: Attention-aware Input Embeddings For New Tokens
di: Dobler, Konstantin, et al.
Pubblicazione: (2025)

When Answers Stray from Questions: Hallucination Detection via Question-Answer Orthogonal Decomposition
di: Yao, Siyang, et al.
Pubblicazione: (2026)

DOTResize: Reducing LLM Width via Discrete Optimal Transport-based Neuron Merging
di: Verma, Neha, et al.
Pubblicazione: (2025)

Merging Feed-Forward Sublayers for Compressed Transformers
di: Verma, Neha, et al.
Pubblicazione: (2025)

TARAZ: Persian Short-Answer Question Benchmark for Cultural Evaluation of Language Models
di: Iranmanesh, Reihaneh, et al.
Pubblicazione: (2026)

Domaino1s: Guiding LLM Reasoning for Explainable Answers in High-Stakes Domains
di: Chu, Xu, et al.
Pubblicazione: (2025)

On the Bias of Next-Token Predictors Toward Systematically Inefficient Reasoning: A Shortest-Path Case Study
di: Alberghi, Riccardo, et al.
Pubblicazione: (2025)

LLMs Have Rhythm: Fingerprinting Large Language Models Using Inter-Token Times and Network Traffic Analysis
di: Alhazbi, Saeif, et al.
Pubblicazione: (2025)

Token Sparse Attention: Efficient Long-Context Inference with Interleaved Token Selection
di: Jo, Dongwon, et al.
Pubblicazione: (2026)