Scaling Makes It Possible: How Large Models Master Impossible Languages
cs.CL, cs.AI
Submitted: 2026-11-01
Updated: 2026-11-01
Terminology
Sources
- Comparative Study of Sand Drawings in Oceania and Africa
- The Llama 3 Herd of Models
- SpinQuant: LLM quantization with learned rotations
- Qwen2.5 Technical Report
- Call for Papers -- The BabyLM Challenge: Sample-efficient pretraining on a developmentally plausible corpus
- Scaling Laws for Neural Language Models
- What do you learn from context? Probing for sentence structure in contextualized word representations
- Transformer Feed-Forward Layers Are Key-Value Memories
- What Does BERT Look At? An Analysis of BERT's Attention
- BERT Rediscovers the Classical NLP Pipeline
- The Bottom-up Evolution of Representations in the Transformer: A Study with Machine Translation and Language Modeling Objectives
- Analyzing Multi-Head Self-Attention: Specialized Heads Do the Heavy Lifting, the Rest Can Be Pruned
- Robust stabilization of polytopic systems via fast and reliable neural network-based approximations
- Achieving Domain-Independent Certified Robustness via Knowledge Continuity
- Reachability Analysis of Deep Neural Networks with Provable Guarantees
- Assessing BERT's Syntactic Abilities
- Emergent effects of scaling on the functional hierarchies within large language models
- Between Circuits and Chomsky: Pre-pretraining on Formal Languages Imparts Linguistic Biases
- Deep contextualized word representations for detecting sarcasm and irony
- Language Models as Knowledge Bases?
Related papers
- ALARM: Audio-Language Alignment for Reasoning Models
- ThinkSum: Probabilistic reasoning over sets using large language models
- Coherence boosting: When your pretrained language model is not paying enough attention
- From Facts to Insights: A Persona-Driven Dual Memory Framework and Dataset for Role-Playing Agents
- Improving General Role-Playing Agents via Psychology-Grounded Reasoning and Role-Aware Policy Optimization
- The Art of Asking: Multilingual Prompt Optimization for Synthetic Data