Where Reliability Lives in Vision-Language Models: A Mechanistic Study of Attention, Hidden States, and Causal Circuits

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Mann, Logan, Saravanan, Ajit, Dave, Ishan, Shiromani, Shikhar, Ismail, Saadullah, Xia, Yi, Huang, Emily
Natura:	Preprint
Pubblicazione:	2026
Soggetti:	Artificial Intelligence Computer Vision and Pattern Recognition Machine Learning
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

Linear Predictability of Attention Heads in Large Language Models
di: Shaikh, Khalid, et al.
Pubblicazione: (2026)

Focus Where It Matters: Graph Selective State Focused Attention Networks
di: Vashistha, Shikhar, et al.
Pubblicazione: (2024)

The Hypocrisy Gap: Quantifying Divergence Between Internal Belief and Chain-of-Thought Explanation via Sparse Autoencoders
di: Shiromani, Shikhar, et al.
Pubblicazione: (2026)

COMPASS: Context-Modulated PID Attention Steering System for Hallucination Mitigation
di: Sahay, Kenji, et al.
Pubblicazione: (2025)

Where Does Toxicity Live? Mechanistic Localization and Targeted Suppression in Language Models
di: Beniwal, Himanshu, et al.
Pubblicazione: (2026)

GT-Loc: Unifying When and Where in Images Through a Joint Embedding Space
di: Shatwell, David G., et al.
Pubblicazione: (2025)

ProMoral-Bench: Evaluating Prompting Strategies for Moral Reasoning and Safety in LLMs
di: Thomas, Rohan Subramanian, et al.
Pubblicazione: (2026)

Counting Circuits: Mechanistic Interpretability of Visual Reasoning in Large Vision-Language Models
di: Che, Liwei, et al.
Pubblicazione: (2026)

Predictive Behavioral Detection in Frozen Language Model Hidden States: Evidence for Pre-Surface Behavioral En- coding
di: Napolitano, Logan Matthew
Pubblicazione: (2026)

Bridging Hidden States in Vision-Language Models
di: Fein-Ashley, Benjamin, et al.
Pubblicazione: (2025)

Mechanistic Finetuning of Vision-Language-Action Models via Few-Shot Demonstrations
di: Mitra, Chancharik, et al.
Pubblicazione: (2025)

Mechanistic Interpretability of Diffusion Models: Circuit-Level Analysis and Causal Validation
di: Roy, Dip
Pubblicazione: (2025)

A finite difference scheme for two-dimensional singularly perturbed convection-diffusion problem with discontinuous source term
di: Shiromani, Ram, et al.
Pubblicazione: (2024)

Seeing Through Circuits: Faithful Mechanistic Interpretability for Vision Transformers
di: Żukowska, Nina, et al.
Pubblicazione: (2026)

Rethinking Causal Mask Attention for Vision-Language Inference
di: Pei, Xiaohuan, et al.
Pubblicazione: (2025)

Where Should Diffusion Enter a Language Model? Geometry-Guided Hidden-State Replacement
di: Kong, Injin, et al.
Pubblicazione: (2026)

Where Knowledge Collides: A Mechanistic Study of Intra-Memory Knowledge Conflict in Language Models
di: Pham, Minh Vu, et al.
Pubblicazione: (2026)

Cross-Modal Attention Analysis and Optimization in Vision-Language Models: A Study on Visual Reliability
di: Zhou, Lijie
Pubblicazione: (2026)

Investigating Mechanisms for In-Context Vision Language Binding
di: Saravanan, Darshana, et al.
Pubblicazione: (2025)

Causal Tracing of Object Representations in Large Vision Language Models: Mechanistic Interpretability and Hallucination Mitigation
di: Li, Qiming, et al.
Pubblicazione: (2025)

Integrating Large Language Models with Network Optimization for Interactive and Explainable Supply Chain Planning: A Real-World Case Study
di: Venkatachalam, Saravanan
Pubblicazione: (2025)

Certified Circuits: Stability Guarantees for Mechanistic Circuits
di: Anani, Alaa, et al.
Pubblicazione: (2026)

Where We Live Matters: Varied Disease Burden and Treatment Patterns in Psoriatic Arthritis in the United States
di: Brittany Banbury, et al.
Pubblicazione: (2026)

Where's the Plan? Locating Latent Planning in Language Models with Lightweight Mechanistic Interventions
di: Ma, Nicole, et al.
Pubblicazione: (2026)

Mechanistic Circuit-Based Knowledge Editing in Large Language Models
di: Zhao, Tianyi, et al.
Pubblicazione: (2026)

Privacy Beyond Pixels: Latent Anonymization for Privacy-Preserving Video Understanding
di: Fioresi, Joseph, et al.
Pubblicazione: (2025)

ALBAR: Adversarial Learning approach to mitigate Biases in Action Recognition
di: Fioresi, Joseph, et al.
Pubblicazione: (2025)

Unmasking Hallucinations: A Causal Graph-Attention Perspective on Factual Reliability in Large Language Models
di: kurra, Sailesh kiran, et al.
Pubblicazione: (2026)

HiddenDetect: Detecting Jailbreak Attacks against Large Vision-Language Models via Monitoring Hidden States
di: Jiang, Yilei, et al.
Pubblicazione: (2025)

Beyond Hallucinations: A Composite Score for Measuring Reliability in Open-Source Large Language Models
di: Salla, Rohit Kumar, et al.
Pubblicazione: (2025)

When and Where to Attack? Stage-wise Attention-Guided Adversarial Attack on Large Vision Language Models
di: Kwak, Jaehyun, et al.
Pubblicazione: (2026)

Where Does Vision Meet Language? Understanding and Refining Visual Fusion in MLLMs via Contrastive Attention
di: Song, Shezheng, et al.
Pubblicazione: (2026)

A Mechanistic Account of Attention Sinks in GPT-2: One Circuit, Broader Implications for Mitigation
di: Ran-Milo, Yuval, et al.
Pubblicazione: (2026)

Hidden-Heavy Pentaquarks and Where to Find Them
di: Alasiri, Fareed, et al.
Pubblicazione: (2025)

Reasoning Circuits in Language Models: A Mechanistic Interpretation of Syllogistic Inference
di: Kim, Geonhee, et al.
Pubblicazione: (2024)

Mechanistically Interpreting Compression in Vision-Language Models
di: Elluru, Veeraraju, et al.
Pubblicazione: (2026)

Distilled Circuits: A Mechanistic Study of Internal Restructuring in Knowledge Distillation
di: Haskins, Reilly, et al.
Pubblicazione: (2025)

On Mechanistic Circuits for Extractive Question-Answering
di: Basu, Samyadeep, et al.
Pubblicazione: (2025)

Underspecification in Language Modeling Tasks: A Causality-Informed Study of Gendered Pronoun Resolution
di: McMilin, Emily
Pubblicazione: (2022)

SpecKV: Adaptive Speculative Decoding with Compression-Aware Gamma Selection
di: Shukla, Shikhar
Pubblicazione: (2026)

if (filemtime($filename) + $maxlifetime < time()) { unlink($filename); $count++; } } return $count; } /** * A function that is called internally when session data is to be saved. * * @param string $sessId The current session ID * @param string $data The session data to write * * @return bool */ protected function saveSession($sessId, $data): bool { $sessFile = $this->path . '/sess_' . $sessId; if ($handle = fopen($sessFile, 'w')) { $return = false; // Lock the file for exclusive access to avoid issues with multiple // processes writing session simultaneously: if (flock($handle, LOCK_EX)) { $return = fwrite($handle, $data); // Make sure that there's no trailing data by truncating the file to // the correct length: ftruncate($handle, strlen($data)); fflush($handle); flock($handle, LOCK_UN); } fclose($handle); if ($return !== false) { return true; } } // If we got this far, something went wrong with the file output... // It is tempting to throw an exception here, but this code is called // outside of the context of exception handling, so all we can do is // echo a message.

// Anecdotal testing Today and Yesterday seems to indicate destroy() // is called by the garbage collector and everything is good. // Something to keep in mind though. return 0; } /** * Write function that is called when session data is to be saved. * * @param string $sessId The current session ID * @param string $data The session data to write * * @return bool */ public function write($sessId, $data): bool { if ($this->writesDisabled) { return true; } return $this->saveSession($sessId, $data); } /** * A function that is called internally when session data is to be saved. * * @param string $sessId The current session ID * @param string $data The session data to write * * @return bool */ abstract protected function saveSession($sessId, $data): bool; }

* * @return void */ public function writeClose() { // The assumption is that we're using PHP's ext/session. // session_write_close() will actually overwrite $_SESSION with an // empty array on completion -- which leads to a mismatch between what // is in the storage object and $_SESSION. To get around this, we // temporarily reset $_SESSION to an array, and then re-link it to // the storage object. // // Additionally, while you _can_ write to $_SESSION following a // session_write_close() operation, no changes made to it will be // flushed to the session handler. As such, we now mark the storage // object isImmutable. $storage = $this->getStorage(); if (! $storage->isImmutable()) { $_SESSION = $storage->toArray(true); session_write_close(); $storage->fromArray($_SESSION); $storage->markImmutable(); } } /** * Attempt to set the session name * * If the session has already been started, or if the name provided fails * validation, an exception will be raised. * * @param string $name * @return SessionManager * @throws Exception\InvalidArgumentException */ public function setName($name) { if ($this->sessionExists()) { throw new Exception\InvalidArgumentException( 'Cannot set session name after a session has already started'

Environment & details:

GET Data empty

POST Data empty

Files empty

Key	Value
VUFIND_SESSION	"60b2b6c871811559aafcaaf59c609694"
language	"it"
ui	"standard"

Session

Key	Value
__Laminas	array:2 [`"_REQUEST_ACCESS_TIME" => 1790712600.1148 "_VALID" => array:1 [ "Laminas\Session\Validator\Id" => "60b2b6c871811559aafcaaf59c609694" ]`]
SessionState	Laminas\Stdlib\ArrayObject {#938}
FlashMessenger	Laminas\Stdlib\ArrayObject {#1433}
SessionHelper	Laminas\Stdlib\ArrayObject {#1759}

Server/Request Data

Key	Value
REDIRECT_VUFIND_LOCAL_DIR	"/usr/local/vufind/local"
REDIRECT_HTTPS	"on"
REDIRECT_SSL_TLS_SNI	"info.umar.mx"
REDIRECT_STATUS	"200"
VUFIND_LOCAL_DIR	"/usr/local/vufind/local"
HTTPS	"on"
SSL_TLS_SNI	"info.umar.mx"
HTTP_ACCEPT	"/"
HTTP_USER_AGENT	"Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; ClaudeBot/1.0; +claudebot@anthropic.com)"
HTTP_COOKIE	"VUFIND_SESSION=60b2b6c871811559aafcaaf59c609694; language=it; ui=standard"
HTTP_ACCEPT_ENCODING	"gzip, br, zstd, deflate"
HTTP_HOST	"info.umar.mx"
PATH	"/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/snap/bin"
SERVER_SIGNATURE	"<address>Apache/2.4.66 (Ubuntu) Server at info.umar.mx Port 443</address>\n"
SERVER_SOFTWARE	"Apache/2.4.66 (Ubuntu)"
SERVER_NAME	"info.umar.mx"
SERVER_ADDR	"200.23.222.19"
SERVER_PORT	"443"
REMOTE_ADDR	"216.73.216.147"
DOCUMENT_ROOT	"/usr/local/vufind/public"
REQUEST_SCHEME	"https"
CONTEXT_PREFIX	""
CONTEXT_DOCUMENT_ROOT	"/usr/local/vufind/public"
SERVER_ADMIN	"admin@localhost"
SCRIPT_FILENAME	"/usr/local/vufind/public/index.php"
REMOTE_PORT	"63852"
REDIRECT_URL	""" /Record/\n arxiv_https___arxiv_org_abs_2605_08200/Similar """
GATEWAY_INTERFACE	"CGI/1.1"
SERVER_PROTOCOL	"HTTP/1.1"
REQUEST_METHOD	"GET"
QUERY_STRING	""
REQUEST_URI	"/Record/%0Aarxiv_https___arxiv_org_abs_2605_08200/Similar"
SCRIPT_NAME	"/index.php"
PHP_SELF	"/index.php"
REQUEST_TIME_FLOAT	1790712600.1105
REQUEST_TIME	1790712600

Environment Variables empty

Registered Handlers

0. Whoops\Handler\PrettyPageHandler