·
DataBubble
  • Home
  • Models
  • News
  • Compare
  • Boards
  • Pricing
  • About
  • Newsletter
  • Methodology
  • Contact
Latest
What Transfers from Training-Only Priors? Separating Knowledge, Ordinal Targets, and Optimization39m◆SpectrumQuant: Spectrum Alignment for LLM Quantization39m◆A Mixture-Gated Theory of Anisotropic Representation Dynamics in Transformers39m◆Privacy Accounting for Empirical Denoisers39m◆Foundations of the Scaling Law39m◆Lean Finder V2: Semantic Search for Mathlib That Disambiguates User Queries39m◆The Value of System 1 Models in Medicine39m◆Harmful Responses Are Jailbreak Templates: Transferring Autoregressive Jailbreaks to Diffusion Language Models39m◆Stochastic Inverse Problem: stability, regularization and Wasserstein gradient flow39m◆Quantum-aware Transformer model for state classification39m◆Deep Learning Warm Starts for Trajectory Optimization on the International Space Station39m◆BrainRotViT: Transformer-ResNet Hybrid for Explainable Modeling of Brain Aging from 3D sMRI39m◆MedInsightBench: Evaluating Medical Analytics Agents Through Multi-Step Insight Discovery in Multimodal Medical Data39m◆Isotropic Activation Functions Enable Deindividuated Neurons and Adaptive Topologies39m◆Half-Truths Break Similarity-Based Retrieval39m◆How do standard time series forecasting benchmarks translate to finance? A broad study of risk adjusted performance39m◆Temporal Fair Division in Multi-Agent Systems: From Precise Alternation Metrics to Scalable Coordination Proxies39m◆Causal Representation Learning for Generalisable Recommendation39m◆Instance-dependent Stochastic Lipschitz bandit39m◆Driving Video Retrieval for Complex Queries with Structured Grounding39m◆What Transfers from Training-Only Priors? Separating Knowledge, Ordinal Targets, and Optimization39m◆SpectrumQuant: Spectrum Alignment for LLM Quantization39m◆A Mixture-Gated Theory of Anisotropic Representation Dynamics in Transformers39m◆Privacy Accounting for Empirical Denoisers39m◆Foundations of the Scaling Law39m◆Lean Finder V2: Semantic Search for Mathlib That Disambiguates User Queries39m◆The Value of System 1 Models in Medicine39m◆Harmful Responses Are Jailbreak Templates: Transferring Autoregressive Jailbreaks to Diffusion Language Models39m◆Stochastic Inverse Problem: stability, regularization and Wasserstein gradient flow39m◆Quantum-aware Transformer model for state classification39m◆Deep Learning Warm Starts for Trajectory Optimization on the International Space Station39m◆BrainRotViT: Transformer-ResNet Hybrid for Explainable Modeling of Brain Aging from 3D sMRI39m◆MedInsightBench: Evaluating Medical Analytics Agents Through Multi-Step Insight Discovery in Multimodal Medical Data39m◆Isotropic Activation Functions Enable Deindividuated Neurons and Adaptive Topologies39m◆Half-Truths Break Similarity-Based Retrieval39m◆How do standard time series forecasting benchmarks translate to finance? A broad study of risk adjusted performance39m◆Temporal Fair Division in Multi-Agent Systems: From Precise Alternation Metrics to Scalable Coordination Proxies39m◆Causal Representation Learning for Generalisable Recommendation39m◆Instance-dependent Stochastic Lipschitz bandit39m◆Driving Video Retrieval for Complex Queries with Structured Grounding39m◆
News/Copy First, Translate Later: Interpreting Translation Dynamics in Multilingual Pretraining
arxiv
PublishedJune 29, 2026 at 4:00 AM
—neutral

Copy First, Translate Later: Interpreting Translation Dynamics in Multilingual Pretraining

Source
arxiv.orgfull article ↗
Read on arxiv→
Publisher summary· verbatim

arXiv:2604.17633v2 Announce Type: replace Abstract: Large language models exhibit impressive cross-lingual capabilities. However, prior work analyzes this phenomenon through isolated factors and at sparse points during training, limiting our understanding of how cross-lingual generalization emerges-

Stay posted· Newsletter

Three short emails a week: Monday, Wednesday and Friday. Top movers, new models and the most-referenced story.

// no spam · unsubscribe one-click · free forever

Your email is used only to send The Bubble Brief. Privacy policy

Discussion
Mentioned models
01
  • 01
    multilingual 1.7B model
Source
↗
arxiv
Read original ↗All from arxiv →
Tags
04
#multilingual#pretraining#language-models#translation

No replies yet. Be first.

Mentioned models
01
  • 01
    multilingual 1.7B model
Source
↗
arxiv
Read original ↗All from arxiv →
Tags
04
#multilingual#pretraining#language-models#translation

Related coverage

More from ARXIV
arxivWhat Transfers from Training-Only Priors? Separating Knowledge, Ordinal Targets, and Optimization39marxivSpectrumQuant: Spectrum Alignment for LLM Quantization39marxivA Mixture-Gated Theory of Anisotropic Representation Dynamics in Transformers39marxivPrivacy Accounting for Empirical Denoisers39m
The Bubble Brief
MON · WED · FRI

Read multilingual insights every Monday, Wednesday and Friday — top movers, new models, story of the week.

// no spam · unsubscribe one-click · free forever

Your email is used only to send The Bubble Brief. Privacy policy

Originally published on arxiv ↗
PrivacyTermsRefundsLegal notice|AboutMethodologyContact

© 2026 Databubble

Built by Marouane Gazouzi · Readyo, AI interview coach

HomeModelsNews