Deliberative Searcher: Improving LLM Reliability via Reinforcement Learning with constraints

Fuente: arXiv

Gespeichert in:

Bibliographische Detailangaben
Hauptverfasser:	Yin, Zhenyun, Wang, Shujie, Wang, Xuhong, Ma, Xingjun, Wang, Yinchun
Format:	Preprint
Veröffentlicht:	2025
Schlagworte:	Artificial Intelligence
Online-Zugang:	Acceder al recurso
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!

Ähnliche Einträge

Constrained Intrinsic Motivation for Reinforcement Learning
von: Zheng, Xiang, et al.
Veröffentlicht: (2024)

DynaSearcher: Dynamic Knowledge Graph Augmented Search Agent via Multi-Reward Reinforcement Learning
von: Hao, Chuzhan, et al.
Veröffentlicht: (2025)

Q*: Improving Multi-step Reasoning for LLMs with Deliberative Planning
von: Wang, Chaojie, et al.
Veröffentlicht: (2024)

Towards Human-AI Deliberation: Design and Evaluation of LLM-Empowered Deliberative AI for AI-Assisted Decision-Making
von: Ma, Shuai, et al.
Veröffentlicht: (2024)

AceSearcher: Bootstrapping Reasoning and Search for LLMs via Reinforced Self-Play
von: Xu, Ran, et al.
Veröffentlicht: (2025)

Explore with Long-term Memory: A Benchmark and Multimodal LLM-based Reinforcement Learning Framework for Embodied Exploration
von: Wang, Sen, et al.
Veröffentlicht: (2026)

VERIRL: Boosting the LLM-based Verilog Code Generation via Reinforcement Learning
von: Teng, Fu, et al.
Veröffentlicht: (2025)

R1-Searcher: Incentivizing the Search Capability in LLMs via Reinforcement Learning
von: Song, Huatong, et al.
Veröffentlicht: (2025)

R1-Searcher++: Incentivizing the Dynamic Knowledge Acquisition of LLMs via Reinforcement Learning
von: Song, Huatong, et al.
Veröffentlicht: (2025)

Deliberative Dynamics and Value Alignment in LLM Debates
von: Sachdeva, Pratik S., et al.
Veröffentlicht: (2025)

MemSearcher: Training LLMs to Reason, Search and Manage Memory via End-to-End Reinforcement Learning
von: Yuan, Qianhao, et al.
Veröffentlicht: (2025)

Improving LLM Code Generation via Requirement-Aware Curriculum Reinforcement Learning
von: Yin, Shouyu, et al.
Veröffentlicht: (2026)

Improving Medical Reasoning with Curriculum-Aware Reinforcement Learning
von: Rui, Shaohao, et al.
Veröffentlicht: (2025)

BoostAPR: Boosting Automated Program Repair via Execution-Grounded Reinforcement Learning with Dual Reward Models
von: Li, Yuanhao, et al.
Veröffentlicht: (2026)

LLM4EFFI: Leveraging Large Language Models to Enhance Code Efficiency and Correctness
von: Ye, Tong, et al.
Veröffentlicht: (2025)

LightSearcher: Efficient DeepSearch via Experiential Memory
von: Lan, Hengzhi, et al.
Veröffentlicht: (2025)

Aligning Large Language Models with Searcher Preferences
von: Wu, Wei, et al.
Veröffentlicht: (2026)

Atom-Searcher: Enhancing Agentic Deep Research via Fine-Grained Atomic Thought Reward
von: Deng, Yong, et al.
Veröffentlicht: (2025)

From Order to Distribution: A Spectral Characterization of Forgetting in Continual Learning
von: Xu, Zonghuan, et al.
Veröffentlicht: (2026)

Uncovering LLM-Generated Code: A Zero-Shot Synthetic Code Detector via Code Rewriting
von: Ye, Tong, et al.
Veröffentlicht: (2024)

Dynamic Adversarial Reinforcement Learning for Robust Multimodal Large Language Models
von: Bao, Yicheng, et al.
Veröffentlicht: (2026)

Beyond Surface Judgments: Human-Grounded Risk Evaluation of LLM-Generated Disinformation
von: Xu, Zonghuan, et al.
Veröffentlicht: (2026)

VAPO: Efficient and Reliable Reinforcement Learning for Advanced Reasoning Tasks
von: Yue, Yu, et al.
Veröffentlicht: (2025)

BandPO: Bridging Trust Regions and Ratio Clipping via Probability-Aware Bounds for LLM Reinforcement Learning
von: Li, Yuan, et al.
Veröffentlicht: (2026)

Flash-Searcher: Fast and Effective Web Agents via DAG-Based Parallel Execution
von: Qin, Tianrui, et al.
Veröffentlicht: (2025)

Reasoning over Precedents Alongside Statutes: Case-Augmented Deliberative Alignment for LLM Safety
von: Jin, Can, et al.
Veröffentlicht: (2026)

DORAEMON: Decentralized Ontology-aware Reliable Agent with Enhanced Memory Oriented Navigation
von: Gu, Tianjun, et al.
Veröffentlicht: (2025)

LM-Searcher: Cross-domain Neural Architecture Search with LLMs via Unified Numerical Encoding
von: Hu, Yuxuan, et al.
Veröffentlicht: (2025)

Privacy Risks of LLM-Empowered Recommender Systems: An Inversion Attack Perspective
von: Wang, Yubo, et al.
Veröffentlicht: (2025)

MindSearch: Mimicking Human Minds Elicits Deep AI Searcher
von: Chen, Zehui, et al.
Veröffentlicht: (2024)

Efficient Differentiable Causal Discovery via Reliable Super-Structure Learning
von: Ma, Pingchuan, et al.
Veröffentlicht: (2026)

Tree Search for LLM Agent Reinforcement Learning
von: Ji, Yuxiang, et al.
Veröffentlicht: (2025)

SimpleDeepSearcher: Deep Information Seeking via Web-Powered Reasoning Trajectory Synthesis
von: Sun, Shuang, et al.
Veröffentlicht: (2025)

Deliberative Reasoning Network: An Uncertainty-Driven Paradigm for Belief-Tracked Inference with Pretrained Language Models
von: Xu, Anran, et al.
Veröffentlicht: (2025)

BackdoorLLM: A Comprehensive Benchmark for Backdoor Attacks and Defenses on Large Language Models
von: Li, Yige, et al.
Veröffentlicht: (2024)

Reinforcement Learning for Self-Improving Agent with Skill Library
von: Wang, Jiongxiao, et al.
Veröffentlicht: (2025)

Hybrid Deep Searcher: Scalable Parallel and Sequential Search Reasoning
von: Ko, Dayoon, et al.
Veröffentlicht: (2025)

Toward Scientific Reasoning in LLMs: Training from Expert Discussions via Reinforcement Learning
von: Yin, Ming, et al.
Veröffentlicht: (2025)

Formalization Driven LLM Prompt Jailbreaking via Reinforcement Learning
von: Wang, Zhaoqi, et al.
Veröffentlicht: (2025)

PaperAsk: A Benchmark for Reliability Evaluation of LLMs in Paper Search and Reading
von: Wu, Yutao, et al.
Veröffentlicht: (2025)

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
ui	"standard"
VUFIND_SESSION	"483cdd3c7beb746b0e55297779baedcc"
language	"de"

Session

Key	Value
__Laminas	array:2 [`"_REQUEST_ACCESS_TIME" => 1790979693.284 "_VALID" => array:1 [ "Laminas\Session\Validator\Id" => "483cdd3c7beb746b0e55297779baedcc" ]`]
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	"ui=standard; VUFIND_SESSION=483cdd3c7beb746b0e55297779baedcc; language=de"
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	"216.73.217.142"
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	"61684"
REDIRECT_URL	""" /Record/\n arxiv_https___arxiv_org_abs_2507_16727/Similar """
GATEWAY_INTERFACE	"CGI/1.1"
SERVER_PROTOCOL	"HTTP/1.1"
REQUEST_METHOD	"GET"
QUERY_STRING	""
REQUEST_URI	"/Record/%0Aarxiv_https___arxiv_org_abs_2507_16727/Similar"
SCRIPT_NAME	"/index.php"
PHP_SELF	"/index.php"
REQUEST_TIME_FLOAT	1790979693.2795
REQUEST_TIME	1790979693

Environment Variables empty

Registered Handlers

0. Whoops\Handler\PrettyPageHandler