{"total":195,"rows":[{"id":"2102.12092","title":"Zero-Shot Text-to-Image Generation","created":"2021-02-24","cited_by_count":4288,"area":null},{"id":"2212.04356","title":"Robust Speech Recognition via Large-Scale Weak Supervision","created":"2022-12-06","cited_by_count":4057,"area":null},{"id":"2309.00614","title":"Baseline Defenses for Adversarial Attacks Against Aligned Language Models","created":"2023-09-01","cited_by_count":651,"area":"alignment"},{"id":"2310.12815","title":"Formalizing and Benchmarking Prompt Injection Attacks and Defenses","created":"2023-10-19","cited_by_count":364,"area":"safety"},{"id":"2207.14255","title":"Efficient Training of Language Models to Fill in the Middle","created":"2022-07-28","cited_by_count":255,"area":null},{"id":"2402.08983","title":"SafeDecoding: Defending against Jailbreak Attacks via Safety-Aware Decoding","created":"2024-02-14","cited_by_count":247,"area":"safety"},{"id":"2103.06874","title":"CANINE: Pre-training an Efficient Tokenization-Free Encoder for Language Representation","created":"2021-03-11","cited_by_count":232,"area":null},{"id":"2309.10668","title":"Language Modeling Is Compression","created":"2023-09-19","cited_by_count":225,"area":null},{"id":"2402.11753","title":"ArtPrompt: ASCII Art-based Jailbreak Attacks against Aligned LLMs","created":"2024-02-19","cited_by_count":216,"area":"safety"},{"id":"2305.13707","title":"Do All Languages Cost the Same? Tokenization in the Era of Commercial Language Models","created":"2023-05-23","cited_by_count":192,"area":null},{"id":"2405.14782","title":"Lessons from the Trenches on Reproducible Evaluation of Language Models","created":"2024-05-23","cited_by_count":179,"area":null},{"id":"2106.12672","title":"Charformer: Fast Character Transformers via Gradient-based Subword Tokenization","created":"2021-06-23","cited_by_count":160,"area":null},{"id":"2112.10508","title":"Between words and characters: A Brief History of Open-Vocabulary Modeling and Tokenization in NLP","created":"2021-12-20","cited_by_count":132,"area":null},{"id":"2109.00486","title":"Survey of Low-Resource Machine Translation","created":"2021-09-01","cited_by_count":92,"area":null},{"id":"2402.18376","title":"Tokenization Is More Than Compression","created":"2024-02-28","cited_by_count":89,"area":null},{"id":"2407.01599","title":"JailbreakZoo: Survey, Landscapes, and Horizons in Jailbreaking Large Language and Vision-Language Models","created":"2024-06-26","cited_by_count":70,"area":"mllm"},{"id":"2402.01035","title":"Getting the most out of your tokenizer for pre-training and domain adaptation","created":"2024-02-01","cited_by_count":66,"area":null},{"id":"2503.13423","title":"SuperBPE: Space Travel for Language Models","created":"2025-03-17","cited_by_count":63,"area":null},{"id":"2204.08832","title":"Impact of Tokenization on Language Models: An Analysis for Turkish","created":"2022-04-19","cited_by_count":62,"area":null},{"id":"2403.06265","title":"Unpacking Tokenization: Evaluating Text Compression and its Correlation with Model Performance","created":"2024-03-10","cited_by_count":61,"area":null},{"id":"2306.16842","title":"Tokenization and the Noiseless Channel","created":"2023-06-29","cited_by_count":59,"area":null},{"id":"2207.06991","title":"Language Modelling with Pixels","created":"2022-07-14","cited_by_count":55,"area":null},{"id":"2508.15763","title":"Intern-S1: A Scientific Multimodal Foundation Model","created":"2025-08-21","cited_by_count":53,"area":"moe"},{"id":"2105.01735","title":"HerBERT: Efficiently Pretrained Transformer-based Language Model for Polish","created":"2021-05-04","cited_by_count":51,"area":null},{"id":"2410.05864","title":"From Tokens to Words: On the Inner Lexicon of LLMs","created":"2024-10-08","cited_by_count":48,"area":null},{"id":"2104.08211","title":"Robust Open-Vocabulary Translation from Visual Text Representations","created":"2021-04-16","cited_by_count":47,"area":null},{"id":"2405.05417","title":"Fishing for Magikarp: Automatically Detecting Under-trained Tokens in Large Language Models","created":"2024-05-08","cited_by_count":46,"area":null},{"id":"2411.03766","title":"Number Cookbook: Number Understanding of Language Models and How to Improve It","created":"2024-11-06","cited_by_count":42,"area":null},{"id":"2106.08226","title":"Consistency Regularization for Cross-Lingual Fine-Tuning","created":"2021-06-15","cited_by_count":41,"area":null},{"id":"2202.05148","title":"Identifying Weaknesses in Machine Translation Metrics Through Minimum Bayes Risk Decoding: A Case Study for COMET","created":"2022-02-10","cited_by_count":41,"area":null},{"id":"2103.08490","title":"Multi-view Subword Regularization","created":"2021-03-15","cited_by_count":40,"area":null},{"id":"2110.00672","title":"Low Frequency Names Exhibit Bias and Overfitting in Contextualizing Language Models","created":"2021-10-01","cited_by_count":39,"area":null},{"id":"2506.09443","title":"LLMs Cannot Reliably Judge (Yet?): A Comprehensive Assessment on the Robustness of LLM-as-a-Judge","created":"2025-06-11","cited_by_count":37,"area":"safety"},{"id":"2206.02608","title":"What do tokens know about their characters and how do they know it?","created":"2022-06-06","cited_by_count":36,"area":null},{"id":"2305.01778","title":"SLTUNET: A Simple Unified Model for Sign Language Translation","created":"2023-05-02","cited_by_count":35,"area":null},{"id":"2403.01289","title":"Greed is All You Need: An Evaluation of Tokenizer Inference Methods","created":"2024-03-02","cited_by_count":34,"area":null},{"id":"2311.11509","title":"Token-Level Adversarial Prompt Detection Based on Perplexity Measures and Contextual Information","created":"2023-11-20","cited_by_count":33,"area":"safety"},{"id":"2104.00290","title":"Many-to-English Machine Translation Tools, Data, and Pretrained Models","created":"2021-04-01","cited_by_count":31,"area":null},{"id":"2409.04599","title":"BPE Gets Picky: Efficient Vocabulary Refinement During Tokenizer Training","created":"2024-09-06","cited_by_count":31,"area":null},{"id":"2412.03719","title":"From Language Models over Tokens to Language Models over Characters","created":"2024-12-04","cited_by_count":31,"area":null},{"id":"2109.02550","title":"You should evaluate your language model on marginal likelihood over tokenisations","created":"2021-09-06","cited_by_count":30,"area":null},{"id":"2110.02782","title":"How BPE Affects Memorization in Transformers","created":"2021-10-06","cited_by_count":29,"area":null},{"id":"2110.08191","title":"Why don't people use character-level machine translation?","created":"2021-10-15","cited_by_count":27,"area":null},{"id":"2203.01976","title":"Overlap-based Vocabulary Generation Improves Cross-lingual Transfer Among Related Languages","created":"2022-03-03","cited_by_count":25,"area":null},{"id":"2206.07615","title":"The SIGMORPHON 2022 Shared Task on Morpheme Segmentation","created":"2022-06-15","cited_by_count":25,"area":null},{"id":"2502.00894","title":"MorphBPE: A Morpho-Aware Tokenizer Bridging Linguistic Complexity for Efficient LLM Training Across Morphologies","created":"2025-02-02","cited_by_count":25,"area":null},{"id":"2203.13838","title":"Analyzing Generalization of Vision and Language Navigation to Unseen Outdoor Areas","created":"2022-03-25","cited_by_count":24,"area":null},{"id":"2408.04811","title":"h4rm3l: A language for Composable Jailbreak Attack Synthesis","created":"2024-08-09","cited_by_count":24,"area":"safety"},{"id":"2409.03788","title":"HSF: Defending against Jailbreak Attacks with Hidden State Filtering","created":"2024-08-31","cited_by_count":23,"area":"safety"},{"id":"2103.06716","title":"Learning Word-Level Confidence For Subword End-to-End ASR","created":"2021-03-11","cited_by_count":22,"area":null},{"id":"2205.14336","title":"Gating Dropout: Communication-efficient Regularization for Sparsely Activated Transformers","created":"2022-05-28","cited_by_count":22,"area":"moe"},{"id":"2307.05355","title":"UniCoRN: Unified Cognitive Signal ReconstructioN bridging cognitive signals and human language","created":"2023-07-06","cited_by_count":22,"area":null},{"id":"2407.11606","title":"The Foundations of Tokenization: Statistical and Computational Concerns","created":"2024-07-16","cited_by_count":22,"area":null},{"id":"2504.05747","title":"SEA-LION: Southeast Asian Languages in One Network","created":"2025-04-08","cited_by_count":22,"area":null},{"id":"2408.08541","title":"Where is the signal in tokenization space?","created":"2024-08-16","cited_by_count":21,"area":null},{"id":"2406.11687","title":"Tokenization Falling Short: On Subword Robustness in Large Language Models","created":"2024-06-17","cited_by_count":20,"area":null},{"id":"2109.06772","title":"Improving Zero-shot Cross-lingual Transfer between Closely Related Languages by injecting Character-level Noise","created":"2021-09-14","cited_by_count":19,"area":null},{"id":"2304.10158","title":"Does Manipulating Tokenization Aid Cross-Lingual Transfer? A Study on POS Tagging for Non-Standardized Languages","created":"2023-04-20","cited_by_count":19,"area":null},{"id":"2308.00189","title":"Generative Models as a Complex Systems Science: How can we make sense of large language model behavior?","created":"2023-07-31","cited_by_count":19,"area":null},{"id":"2404.03608","title":"Sailor: Open Language Models for South-East Asia","created":"2024-04-04","cited_by_count":19,"area":null},{"id":"2407.00869","title":"Large Language Models Are Involuntary Truth-Tellers: Exploiting Fallacy Failure for Jailbreak Attacks","created":"2024-07-01","cited_by_count":19,"area":"safety"},{"id":"2306.17757","title":"Should you marginalize over possible tokenizations?","created":"2023-06-30","cited_by_count":18,"area":null},{"id":"2406.16829","title":"Understanding and Mitigating Tokenization Bias in Language Models","created":"2024-06-24","cited_by_count":18,"area":null},{"id":"2502.13581","title":"ActionPiece: Contextually Tokenizing Action Sequences for Generative Recommendation","created":"2025-02-19","cited_by_count":18,"area":null},{"id":"2206.03352","title":"Searching for Optimal Subword Tokenization in Cross-domain NER","created":"2022-06-07","cited_by_count":17,"area":null},{"id":"2506.19004","title":"Broken Tokens? Your Language Model can Secretly Handle Non-Canonical Tokenizations","created":"2025-06-23","cited_by_count":17,"area":null},{"id":"2206.02457","title":"Improving Contrastive Learning of Sentence Embeddings with Case-Augmented Positives and Retrieved Negatives","created":"2022-06-06","cited_by_count":16,"area":null},{"id":"2404.17808","title":"Scaffold-BPE: Enhancing Byte Pair Encoding for Large Language Models with Simple and Effective Scaffold Token Removal","created":"2024-04-27","cited_by_count":16,"area":null},{"id":"2105.12410","title":"Joint Optimization of Tokenization and Downstream Model","created":"2021-05-26","cited_by_count":15,"area":null},{"id":"2111.11133","title":"L-Verse: Bidirectional Generation Between Image and Text","created":"2021-11-22","cited_by_count":15,"area":null},{"id":"2410.09303","title":"Exact Byte-Level Probabilities from Tokenized Language Models for FIM-Tasks and Model Ensembles","created":"2024-10-11","cited_by_count":15,"area":null},{"id":"2106.06125","title":"Bridging Subword Gaps in Pretrain-Finetune Paradigm for Natural Language Generation","created":"2021-06-11","cited_by_count":14,"area":null},{"id":"2112.14569","title":"Fine-Tuning Transformers: Vocabulary Transfer","created":"2021-12-29","cited_by_count":14,"area":null},{"id":"2402.14614","title":"Two Counterexamples to Tokenization and the Noiseless Channel","created":"2024-02-22","cited_by_count":13,"area":null},{"id":"2109.04556","title":"Subword Mapping and Anchoring across Languages","created":"2021-09-09","cited_by_count":12,"area":null},{"id":"2410.01188","title":"Gold Panning in Vocabulary: An Adaptive Method for Vocabulary Expansion of Domain-Specific LLMs","created":"2024-10-02","cited_by_count":12,"area":null},{"id":"2210.07095","title":"Incorporating Context into Subword Vocabularies","created":"2022-10-13","cited_by_count":11,"area":null},{"id":"2310.03477","title":"Tik-to-Tok: Translating Language Models One Token at a Time: An Embedding Initialization Strategy for Efficient Language Adaptation","created":"2023-10-05","cited_by_count":11,"area":null},{"id":"2506.03101","title":"Beyond Text Compression: Evaluating Tokenizers Across Scales","created":"2025-06-03","cited_by_count":11,"area":null},{"id":"2602.04705","title":"ERNIE 5.0 Technical Report","created":"2026-02-04","cited_by_count":11,"area":"moe"},{"id":"2204.00665","title":"CipherDAug: Ciphertext based Data Augmentation for Neural Machine Translation","created":"2022-04-01","cited_by_count":9,"area":null},{"id":"2308.00683","title":"CodeBPE: Investigating Subtokenization Options for Large Language Model Pretraining on Source Code","created":"2023-08-01","cited_by_count":9,"area":null},{"id":"2406.18118","title":"SafeAligner: Safety Alignment against Jailbreak Attacks via Response Disparity Guidance","created":"2024-06-26","cited_by_count":9,"area":"alignment"},{"id":"2109.05772","title":"Wine is Not v i n. -- On the Compatibility of Tokenizations Across Languages","created":"2021-09-13","cited_by_count":8,"area":null},{"id":"2111.08634","title":"NVIDIA NeMo Neural Machine Translation Systems for English-German and English-Russian News and Biomedical Tasks at WMT21","created":"2021-11-16","cited_by_count":8,"area":null},{"id":"2209.04126","title":"MaxMatch-Dropout: Subword Regularization for WordPiece","created":"2022-09-09","cited_by_count":8,"area":null},{"id":"2210.08779","title":"Towards Summary Candidates Fusion","created":"2022-10-17","cited_by_count":8,"area":null},{"id":"2305.18294","title":"Transformer Language Models Handle Word Frequency in Prediction Head","created":"2023-05-29","cited_by_count":8,"area":null},{"id":"2306.01497","title":"Data-Efficient French Language Modeling with CamemBERTa","created":"2023-06-02","cited_by_count":8,"area":null},{"id":"2310.11628","title":"Learn Your Tokens: Word-Pooled Tokenization for Language Modeling","created":"2023-10-17","cited_by_count":8,"area":null},{"id":"2406.11007","title":"Threat Modelling and Risk Analysis for Large Language Model (LLM)-Powered Applications","created":"2024-06-16","cited_by_count":8,"area":"safety"},{"id":"2109.01100","title":"How Suitable Are Subword Segmentation Strategies for Translating Non-Concatenative Morphology?","created":"2021-09-02","cited_by_count":7,"area":null},{"id":"2212.09912","title":"Tokenization Consistency Matters for Generative Models on Extractive NLP Tasks","created":"2022-12-19","cited_by_count":7,"area":null},{"id":"2305.17906","title":"Byte-Level Grammatical Error Correction Using Synthetic and Curated Corpora","created":"2023-05-29","cited_by_count":7,"area":null},{"id":"2404.00397","title":"An Analysis of BPE Vocabulary Trimming in Neural Machine Translation","created":"2024-03-30","cited_by_count":7,"area":null},{"id":"2406.07096","title":"Fast Context-Biasing for CTC and Transducer ASR models with CTC-based Word Spotter","created":"2024-06-11","cited_by_count":7,"area":null},{"id":"2407.16382","title":"TookaBERT: A Step Forward for Persian NLU","created":"2024-07-23","cited_by_count":7,"area":null},{"id":"2504.02458","title":"Retrieval-Augmented Purifier for Robust LLM-Empowered Recommendation","created":"2025-04-03","cited_by_count":7,"area":"rag"},{"id":"2506.01687","title":"StochasTok: Improving Fine-Grained Subword Understanding in LLMs","created":"2025-06-02","cited_by_count":7,"area":null},{"id":"2506.02442","title":"Should LLM Safety Be More Than Refusing Harmful Instructions?","created":"2025-06-03","cited_by_count":7,"area":"safety"}],"next_cursor":"100"}