Cover Image

SOD: Step-wise On-policy Distillation for Small Language Model Agents

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Zhong, Qiyong, Zheng, Mao, Song, Mingyang, Lin, Xin, Sun, Jie, Jiang, Houcheng, Wang, Xiang, Fang, Junfeng
Format:	Preprint
Published:	2026
Subjects:	Computation and Language Artificial Intelligence
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

SimCT: Recovering Lost Supervision for Cross-Tokenizer On-Policy Distillation
by: Sun, Jie, et al.
Published: (2026)

A Survey of On-Policy Distillation for Large Language Models
by: Song, Mingyang, et al.
Published: (2026)

Rubric-based On-policy Distillation
by: Fang, Junfeng, et al.
Published: (2026)

AnyEdit: Edit Any Knowledge Encoded in Language Models
by: Jiang, Houcheng, et al.
Published: (2025)

Neuron-Level Sequential Editing for Large Language Models
by: Jiang, Houcheng, et al.
Published: (2024)

Reinforced Lifelong Editing for Language Models
by: Li, Zherui, et al.
Published: (2025)

Model Merging in the Era of Large Language Models: Methods, Applications, and Future Directions
by: Song, Mingyang, et al.
Published: (2026)

A Survey of Query Optimization in Large Language Models
by: Song, Mingyang, et al.
Published: (2024)

AlphaEdit: Null-Space Constrained Knowledge Editing for Language Models
by: Fang, Junfeng, et al.
Published: (2024)

Goal-Aware Identification and Rectification of Misinformation in Multi-Agent Systems
by: Li, Zherui, et al.
Published: (2025)

Counting-Stars: A Multi-evidence, Position-aware, and Scalable Benchmark for Evaluating Long-Context Large Language Models
by: Song, Mingyang, et al.
Published: (2024)

Symbolic Chain-of-Thought Distillation: Small Models Can Also "Think" Step-by-Step
by: Li, Liunian Harold, et al.
Published: (2023)

STEPER: Step-wise Knowledge Distillation for Enhancing Reasoning Ability in Multi-Step Retrieval-Augmented Language Models
by: Lee, Kyumin, et al.
Published: (2025)

Full-Step-DPO: Self-Supervised Preference Optimization with Step-wise Rewards for Mathematical Reasoning
by: Xu, Huimin, et al.
Published: (2025)

FastCuRL: Curriculum Reinforcement Learning with Stage-wise Context Scaling for Efficient Training R1-like Reasoning Models
by: Song, Mingyang, et al.
Published: (2025)

Walk Before You Run! Concise LLM Reasoning via Reinforcement Learning
by: Song, Mingyang, et al.
Published: (2025)

UniVLR: Unifying Text and Vision in Visual Latent Reasoning for Multimodal LLMs
by: Jiang, Houcheng, et al.
Published: (2026)

DualEdit: Mitigating Safety Fallback in LLM Backdoor Editing via Affirmation-Refusal Regulation
by: Jiang, Houcheng, et al.
Published: (2025)

PLAN-TUNING: Post-Training Language Models to Learn Step-by-Step Planning for Complex Problem Solving
by: Parmar, Mihir, et al.
Published: (2025)

Chain-based Distillation for Effective Initialization of Variable-Sized Small Language Models
by: Shi, Boyu, et al.
Published: (2026)

PRISM: Probability Reallocation with In-Span Masking for Knowledge-Sensitive Alignment
by: Xu, Chenning, et al.
Published: (2026)

GRP: Goal-Reversed Prompting for Zero-Shot Evaluation with LLMs
by: Song, Mingyang, et al.
Published: (2025)

Beyond the Illusion of Consensus: From Surface Heuristics to Knowledge-Grounded Evaluation in LLM-as-a-Judge
by: Song, Mingyang, et al.
Published: (2026)

CodeDelegator: Mitigating Context Pollution via Role Separation in Code-as-Action Agents
by: Fei, Tianxiang, et al.
Published: (2026)

Contrastive Weak-to-strong Generalization
by: Jiang, Houcheng, et al.
Published: (2025)

DeepRAG: Thinking to Retrieve Step by Step for Large Language Models
by: Guan, Xinyan, et al.
Published: (2025)

HardMTBench: Stress-Testing Chinese-English Translation on Knowledge-Intensive Domains
by: Li, Zheng, et al.
Published: (2026)

MiMoTable: A Multi-scale Spreadsheet Benchmark with Meta Operations for Table Reasoning
by: Li, Zheng, et al.
Published: (2024)

TAT-R1: Terminology-Aware Translation with Reinforcement Learning and Word Alignment
by: Li, Zheng, et al.
Published: (2025)

PodBench: A Comprehensive Benchmark for Instruction-Aware Audio-Oriented Podcast Script Generation
by: Xu, Chenning, et al.
Published: (2026)

Iterative Layer-wise Distillation for Efficient Compression of Large Language Models
by: Kovalev, Grigory, et al.
Published: (2025)

Skip-Thinking: Chunk-wise Chain-of-Thought Distillation Enable Smaller Language Models to Reason Better and Faster
by: Chen, Xiao, et al.
Published: (2025)

Few-Step Diffusion Language Models via Trajectory Self-Distillation
by: Zhang, Tunyu, et al.
Published: (2026)

EL4NER: Ensemble Learning for Named Entity Recognition via Multiple Small-Parameter Large Language Models
by: Xiao, Yuzhen, et al.
Published: (2025)

IFMTBench: A Comprehensive Benchmark for Multilingual Translation Instruction Following
by: Sun, Mingrui, et al.
Published: (2026)

Stop Rewarding Hallucinated Steps: Faithfulness-Aware Step-Level Reinforcement Learning for Small Reasoning Models
by: Nie, Shuo, et al.
Published: (2026)

SmallPlan: Leverage Small Language Models for Sequential Path Planning with Simulation-Powered, LLM-Guided Distillation
by: Pham, Quang P. M., et al.
Published: (2025)

Uncertainty-Aware Step-wise Verification with Generative Reward Models
by: Ye, Zihuiwen, et al.
Published: (2025)

Capturing Nuanced Preferences: Preference-Aligned Distillation for Small Language Models
by: Gu, Yanggan, et al.
Published: (2025)

Step-DPO: Step-wise Preference Optimization for Long-chain Reasoning of LLMs
by: Lai, Xin, et al.
Published: (2024)