Token-level Ensembling of Models with Different Vocabularies

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Wicks, Rachel, Ravisankar, Kartik, Yang, Xinchen, Koehn, Philipp, Post, Matt
Natura:	Preprint
Pubblicazione:	2025
Soggetti:	Computation and Language
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

Recovering document annotations for sentence-level bitext
di: Wicks, Rachel, et al.
Pubblicazione: (2024)

Escaping the sentence-level paradigm in machine translation
di: Post, Matt, et al.
Pubblicazione: (2023)

Can you map it to English? The Role of Cross-Lingual Alignment in Multilingual Performance of LLMs
di: Ravisankar, Kartik, et al.
Pubblicazione: (2025)

Bridging the Gap between Different Vocabularies for LLM Ensemble
di: Xu, Yangyifan, et al.
Pubblicazione: (2024)

Speech Vecalign: an Embedding-based Method for Aligning Parallel Speech Documents
di: Meng, Chutong, et al.
Pubblicazione: (2025)

Text Style Transfer with Parameter-efficient LLM Finetuning and Round-trip Translation
di: Liu, Ruoxi, et al.
Pubblicazione: (2026)

Learn and Unlearn: Addressing Misinformation in Multilingual LLMs
di: Lu, Taiming, et al.
Pubblicazione: (2024)

Pointer-Generator Networks for Low-Resource Machine Translation: Don't Copy That!
di: Bafna, Niyati, et al.
Pubblicazione: (2024)

PEAR: Pairwise Evaluation for Automatic Relative Scoring in Machine Translation
di: Proietti, Lorenzo, et al.
Pubblicazione: (2026)

Error Norm Truncation: Robust Training in the Presence of Data Noise for Text Generation Models
di: Li, Tianjian, et al.
Pubblicazione: (2023)

Steering Large Language Models with Register Analysis for Arbitrary Style Transfer
di: Yang, Xinchen, et al.
Pubblicazione: (2025)

SLIDE: Reference-free Evaluation for Machine Translation using a Sliding Document Window
di: Raunak, Vikas, et al.
Pubblicazione: (2023)

Exploring Tokenization Strategies and Vocabulary Sizes for Enhanced Arabic Language Models
di: Alrefaie, Mohamed Taher, et al.
Pubblicazione: (2024)

TokAlign++: Advancing Vocabulary Adaptation via Better Token Alignment
di: Li, Chong, et al.
Pubblicazione: (2026)

Navigating the Metrics Maze: Reconciling Score Magnitudes and Accuracies
di: Kocmi, Tom, et al.
Pubblicazione: (2024)

Adaptive BPE Tokenization for Enhanced Vocabulary Adaptation in Finetuning Pretrained Language Models
di: Balde, Gunjan, et al.
Pubblicazione: (2024)

DiffNorm: Self-Supervised Normalization for Non-autoregressive Speech-to-speech Translation
di: Tan, Weiting, et al.
Pubblicazione: (2024)

TokAlign: Efficient Vocabulary Adaptation via Token Alignment
di: Li, Chong, et al.
Pubblicazione: (2025)

Parallel Tokenizers: Rethinking Vocabulary Design for Cross-Lingual Transfer
di: Kautsar, Muhammad Dehan Al, et al.
Pubblicazione: (2025)

When to Ensemble: Identifying Token-Level Points for Stable and Fast LLM Ensembling
di: Yun, Heecheol, et al.
Pubblicazione: (2025)

HiMATE: A Hierarchical Multi-Agent Framework for Machine Translation Evaluation
di: Zhang, Shijie, et al.
Pubblicazione: (2025)

BPE Gets Picky: Efficient Vocabulary Refinement During Tokenizer Training
di: Chizhov, Pavel, et al.
Pubblicazione: (2024)

Over-Tokenized Transformer: Vocabulary is Generally Worth Scaling
di: Huang, Hongzhi, et al.
Pubblicazione: (2025)

Neuron-Level Emotion Control in Speech-Generative Large Audio-Language Models
di: Zhao, Xiutian, et al.
Pubblicazione: (2026)

Interchangeable Token Embeddings for Extendable Vocabulary and Alpha-Equivalence
di: Işık, İlker, et al.
Pubblicazione: (2024)

When the Same Coefficients Reach Different Places: Asymmetric Realizability in Transplanting Tokenizers across Large Language Models
di: Liu, Xiaoze, et al.
Pubblicazione: (2025)

Small Vocabularies, Big Gains: Pretraining and Tokenization in Time Series Models
di: Roger, Alexis, et al.
Pubblicazione: (2025)

X-ALMA: Plug & Play Modules and Adaptive Rejection for Quality Translation at Scale
di: Xu, Haoran, et al.
Pubblicazione: (2024)

Vocabulary-level Memory Efficiency for Language Model Fine-tuning
di: Williams, Miles, et al.
Pubblicazione: (2023)

PyMarian: Fast Neural Machine Translation and Evaluation in Python
di: Gowda, Thamme, et al.
Pubblicazione: (2024)

Optimizing LLMs for Italian: Reducing Token Fertility and Enhancing Efficiency Through Vocabulary Adaptation
di: Moroni, Luca, et al.
Pubblicazione: (2025)

Cut Your Losses in Large-Vocabulary Language Models
di: Wijmans, Erik, et al.
Pubblicazione: (2024)

The Tokenization Bottleneck: How Vocabulary Extension Improves Chemistry Representation Learning in Pretrained Language Models
di: Kalamkar, Prathamesh, et al.
Pubblicazione: (2025)

Exploring the Effect of Segmentation and Vocabulary Size on Speech Tokenization for Speech Language Models
di: Kando, Shunsuke, et al.
Pubblicazione: (2025)

The First to Know: How Token Distributions Reveal Hidden Knowledge in Large Vision-Language Models?
di: Zhao, Qinyu, et al.
Pubblicazione: (2024)

Rethinking Tokenization: Crafting Better Tokenizers for Large Language Models
di: Yang, Jinbiao
Pubblicazione: (2024)

Iterative Auto-Annotation for Scientific Named Entity Recognition Using BERT-Based Models
di: Gupta, Kartik
Pubblicazione: (2025)

CTC-GMM: CTC guided modality matching for fast and accurate streaming speech translation
di: Zhao, Rui, et al.
Pubblicazione: (2024)

Overcoming Vocabulary Constraints with Pixel-level Fallback
di: Lotz, Jonas F., et al.
Pubblicazione: (2025)

DualToken: Towards Unifying Visual Understanding and Generation with Dual Visual Vocabularies
di: Song, Wei, et al.
Pubblicazione: (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
language	"it"
ui	"standard"
VUFIND_SESSION	"d0d1b0ae05acb9ce57def29675bf4c5c"

Session

Key	Value
__Laminas	array:2 [`"_REQUEST_ACCESS_TIME" => 1790958160.1247 "_VALID" => array:1 [ "Laminas\Session\Validator\Id" => "d0d1b0ae05acb9ce57def29675bf4c5c" ]`]
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	"language=it; ui=standard; VUFIND_SESSION=d0d1b0ae05acb9ce57def29675bf4c5c"
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	"54703"
REDIRECT_URL	""" /Record/\n arxiv_https___arxiv_org_abs_2502_21265/Similar """
GATEWAY_INTERFACE	"CGI/1.1"
SERVER_PROTOCOL	"HTTP/1.1"
REQUEST_METHOD	"GET"
QUERY_STRING	""
REQUEST_URI	"/Record/%0Aarxiv_https___arxiv_org_abs_2502_21265/Similar"
SCRIPT_NAME	"/index.php"
PHP_SELF	"/index.php"
REQUEST_TIME_FLOAT	1790958160.1204
REQUEST_TIME	1790958160

Environment Variables empty

Registered Handlers

0. Whoops\Handler\PrettyPageHandler