Imagen de Portada

Improving Romanian LLM Pretraining Data using Diversity and Quality Filtering

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Negoita, Vlad, Masala, Mihai, Rebedea, Traian
Formato:	Preprint
Publicado:	2025
Materias:	Computation and Language
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Improving Legal Judgement Prediction in Romanian with Long Text Encoders
por: Masala, Mihai, et al.
Publicado: (2024)

"Înţelegi Româneşte?'' A Recipe for Romanian Vision-Language Models
por: Masala, Mihai, et al.
Publicado: (2026)

OpenLLM-Ro -- Technical Report on Open-source Romanian LLMs
por: Masala, Mihai, et al.
Publicado: (2024)

"Vorbeşti Româneşte?" A Recipe to Train Powerful Romanian LLMs with English Instructions
por: Masala, Mihai, et al.
Publicado: (2024)

FuLG: 150B Romanian Corpus for Language Model Pretraining
por: Bădoiu, Vlad-Andrei, et al.
Publicado: (2024)

Unsupervised Extraction of Dialogue Policies from Conversations
por: Sreedhar, Makesh Narsimhan, et al.
Publicado: (2024)

Safety Through Reasoning: An Empirical Study of Reasoning Guardrail Models
por: Sreedhar, Makesh Narsimhan, et al.
Publicado: (2025)

Aegis2.0: A Diverse AI Safety Dataset and Risks Taxonomy for Alignment of LLM Guardrails
por: Ghosh, Shaona, et al.
Publicado: (2025)

Semi-Supervised Learning for Large Language Models Safety and Content Moderation
por: Dinuta, Eduard Stefan, et al.
Publicado: (2025)

Towards Inference-time Category-wise Safety Steering for Large Language Models
por: Bhattacharjee, Amrita, et al.
Publicado: (2024)

Training a General Purpose Automated Red Teaming Model
por: Padmakumar, Aishwarya, et al.
Publicado: (2026)

LLMic: Romanian Foundation Language Model
por: Bădoiu, Vlad-Andrei, et al.
Publicado: (2025)

Towards Zero-Shot & Explainable Video Description by Reasoning over Graphs of Events in Space and Time
por: Masala, Mihai, et al.
Publicado: (2025)

From Vision To Language through Graph of Events in Space and Time: An Explainable Self-supervised Approach
por: Masala, Mihai, et al.
Publicado: (2025)

The Data-Quality Illusion: Rethinking Classifier-Based Quality Filtering for LLM Pretraining
por: Saada, Thiziri Nait, et al.
Publicado: (2025)

SmellBench: Evaluating LLM Agents on Architectural Code Smell Repair
por: Dinu, Ion George, et al.
Publicado: (2026)

GIT-CXR: End-to-End Transformer for Chest X-Ray Report Generation
por: Sîrbu, Iustin, et al.
Publicado: (2025)

CantTalkAboutThis: Aligning Language Models to Stay on Topic in Dialogues
por: Sreedhar, Makesh Narsimhan, et al.
Publicado: (2024)

QuaDMix: Quality-Diversity Balanced Data Selection for Efficient LLM Pretraining
por: Liu, Fengze, et al.
Publicado: (2025)

Neural Grammatical Error Correction for Romanian
por: Cotet, Teodor-Mihai, et al.
Publicado: (2026)

Pluralistic Behavior Suite: Stress-Testing Multi-Turn Adherence to Custom Behavioral Policies
por: Varshney, Prasoon, et al.
Publicado: (2025)

HistNERo: Historical Named Entity Recognition for the Romanian Language
por: Avram, Andrei-Marius, et al.
Publicado: (2024)

Enhancing Romanian Offensive Language Detection through Knowledge Distillation, Multi-Task Learning, and Data Augmentation
por: Matei, Vlad-Cristian, et al.
Publicado: (2024)

Evaluating Large Language Models for Diacritic Restoration in Romanian Texts: A Comparative Study
por: Nadas, Mihai, et al.
Publicado: (2025)

MultiMatch: Multihead Consistency Regularization Matching for Semi-Supervised Text Classification
por: Sirbu, Iustin, et al.
Publicado: (2025)

Control the Temperature: Selective Sampling for Diverse and High-Quality LLM Outputs
por: Troshin, Sergey, et al.
Publicado: (2025)

Defenses & Enablers For Skill Injection Attacks on Terminal Based Agents
por: Fujinuma, Yoshinari, et al.
Publicado: (2026)

RoIt-XMASA: Multi-Domain Multilingual Sentiment Analysis Dataset for Romanian and Italian
por: Avram, Andrei-Marius, et al.
Publicado: (2026)

Beyond Filtering: Adaptive Image-Text Quality Enhancement for MLLM Pretraining
por: Huang, Han, et al.
Publicado: (2024)

Ultra-FineWeb: Efficient Data Filtering and Verification for High-Quality LLM Training Data
por: Wang, Yudong, et al.
Publicado: (2025)

Judging Quality Across Languages: A Multilingual Approach to Pretraining Data Filtering with Language Models
por: Ali, Mehdi, et al.
Publicado: (2025)

RoD-TAL: A Benchmark for Answering Questions in Romanian Driving License Exams
por: Man, Andrei Vlad, et al.
Publicado: (2025)

IFDID: Information Filter upon Diversity-Improved Decoding for Diversity-Faithfulness Tradeoff in NLG
por: Meng, Han, et al.
Publicado: (2022)

TF3-RO-50M: Training Compact Romanian Language Models from Scratch on Synthetic Moral Microfiction
por: Nadas, Mihai Dan, et al.
Publicado: (2026)

Building Large-Scale English-Romanian Literary Translation Resources with Open Models
por: Nadas, Mihai, et al.
Publicado: (2025)

Improving Pretraining Data Using Perplexity Correlations
por: Thrush, Tristan, et al.
Publicado: (2024)

AboutMe: Using Self-Descriptions in Webpages to Document the Effects of English Pretraining Data Filters
por: Lucy, Li, et al.
Publicado: (2024)

Agentic Video Generation: From Text to Executable Event Graphs via Tool-Constrained LLM Planning
por: Cudlenco, Nicolae, et al.
Publicado: (2026)

What Are They Filtering Out? An Experimental Benchmark of Filtering Strategies for Harm Reduction in Pretraining Datasets
por: Stranisci, Marco Antonio, et al.
Publicado: (2025)

Beyond URLs: Metadata Diversity and Position for Efficient LLM Pretraining
por: Fan, Dongyang, et al.
Publicado: (2025)