Soft Token Alignment for Cross-Lingual Reasoning
cs.CL
Submitted: 2026-06-25
Updated: 2026-11-05
Terminology
Sources
- Soft Tokens, Hard Truths
- Reasoning Beyond Language: A Comprehensive Survey on Latent Chain-of-Thought Reasoning
- DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
- Aligning Multilingual Reasoning with Verifiable Semantics from a High-Resource Expert Model
- Revisiting Multilingual Data Mixtures in Language Model Pretraining
- TAPO: Translation Augmented Policy Optimization for Multilingual Mathematical Reasoning
- Learn Globally, Speak Locally: Bridging the Gaps in Multilingual Reasoning
- Language-Specific Latent Process Hinders Cross-Lingual Performance
- Decoupled Weight Decay Regularization
- Cross-Lingual Activation Steering for Multilingual Language Models
- DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
- Gemma 3 Technical Report
- LLMs are Single-threaded Reasoners: Demystifying the Working Mechanism of Soft Thinking
- A Formal Comparison Between Chain of Thought and Latent Thought
- Qwen3 Technical Report
- How Well Does Mathematical Reasoning Transfer Across Languages in English-Centric LLMs?
- Crosslingual Reasoning through Test-Time Scaling
- Think Natively: Unlocking Multilingual Reasoning with Consistency-Enhanced Reinforcement Learning
- Align to the Pivot: Dual Alignment with Self-Feedback for Multilingual Math Reasoning
- A Comprehensive Evaluation of Multilingual Chain-of-Thought Reasoning: Performance, Consistency, and Faithfulness Across Languages
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