Copertina

Not All Correct Answers Are Equal: Why Your Distillation Source Matters

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Tian, Xiaoyu, Ji, Yunjie, Wang, Haotian, Chen, Shuaiting, Zhao, Sitong, Peng, Yiping, Zhao, Han, Li, Xiangang
Natura:	Preprint
Pubblicazione:	2025
Soggetti:	Computation and Language
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

1.4 Million Open-Source Distilled Reasoning Dataset to Empower Large Language Model Training
di: Zhao, Han, et al.
Pubblicazione: (2025)

Leveraging Reasoning Model Answers to Enhance Non-Reasoning Model Capability
di: Wang, Haotian, et al.
Pubblicazione: (2025)

DeepDistill: Enhancing LLM Reasoning Capabilities via Large-Scale Difficulty-Graded Data Training
di: Tian, Xiaoyu, et al.
Pubblicazione: (2025)

AM-Thinking-v1: Advancing the Frontier of Reasoning at 32B Scale
di: Ji, Yunjie, et al.
Pubblicazione: (2025)

Think Twice: Enhancing LLM Reasoning by Scaling Multi-round Test-time Thinking
di: Tian, Xiaoyu, et al.
Pubblicazione: (2025)

How Difficulty-Aware Staged Reinforcement Learning Enhances LLMs' Reasoning Capabilities: A Preliminary Experimental Study
di: Ji, Yunjie, et al.
Pubblicazione: (2025)

Exploring the Potential of Offline RL for Reasoning in LLMs: A Preliminary Study
di: Tian, Xiaoyu, et al.
Pubblicazione: (2025)

Not All Proofs Are Equal: Evaluating LLM Proof Quality Beyond Correctness
di: Petrov, Ivo, et al.
Pubblicazione: (2026)

Not All Parameters Are Created Equal: Smart Isolation Boosts Fine-Tuning Performance
di: Wang, Yao, et al.
Pubblicazione: (2025)

Not All Contexts Are Equal: Teaching LLMs Credibility-aware Generation
di: Pan, Ruotong, et al.
Pubblicazione: (2024)

Beyond Answers: Transferring Reasoning Capabilities to Smaller LLMs Using Multi-Teacher Knowledge Distillation
di: Tian, Yijun, et al.
Pubblicazione: (2024)

Mitigating Spurious Correlations Between Question and Answer via Chain-of-Thought Correctness Perception Distillation
di: Xie, Hongyan, et al.
Pubblicazione: (2025)

Not All Languages are Equal: Insights into Multilingual Retrieval-Augmented Generation
di: Wu, Suhang, et al.
Pubblicazione: (2024)

Memorizing is Not Enough: Deep Knowledge Injection Through Reasoning
di: Xu, Ruoxi, et al.
Pubblicazione: (2025)

Answer is All You Need: Instruction-following Text Embedding via Answering the Question
di: Peng, Letian, et al.
Pubblicazione: (2024)

Not All Errors Are Created Equal: ASCoT Addresses Late-Stage Fragility in Efficient LLM Reasoning
di: Zhang, Dongxu, et al.
Pubblicazione: (2025)

All Claims Are Equal, but Some Claims Are More Equal Than Others: Importance-Sensitive Factuality Evaluation of LLM Generations
di: Wanner, Miriam, et al.
Pubblicazione: (2025)

Sandwich Reasoning: An Answer-Reasoning-Answer Approach for Low-Latency Query Correction
di: Zhang, Chen, et al.
Pubblicazione: (2026)

Evaluating and Calibrating LLM Confidence on Questions with Multiple Correct Answers
di: Wang, Yuhan, et al.
Pubblicazione: (2026)

Not All Synthetic Data Is Yours to Learn From
di: Alemohammad, Sina, et al.
Pubblicazione: (2026)

LexPro-1.0 Technical Report
di: Chen, Haotian, et al.
Pubblicazione: (2025)

ASTRA: Automated Synthesis of agentic Trajectories and Reinforcement Arenas
di: Tian, Xiaoyu, et al.
Pubblicazione: (2026)

MMBench: Is Your Multi-modal Model an All-around Player?
di: Liu, Yuan, et al.
Pubblicazione: (2023)

SARI: Structured Audio Reasoning via Curriculum-Guided Reinforcement Learning
di: Wen, Cheng, et al.
Pubblicazione: (2025)

Your Co-Workers Matter: Evaluating Collaborative Capabilities of Language Models in Blocks World
di: Wu, Guande, et al.
Pubblicazione: (2024)

Every Answer Matters: Evaluating Commonsense with Probabilistic Measures
di: Cheng, Qi, et al.
Pubblicazione: (2024)

Seeing Isn't Knowing: Do VLMs Know When Not to Answer Spatial Questions (and Why)?
di: Zhang, Yue, et al.
Pubblicazione: (2026)

LLM-based Privacy Data Augmentation Guided by Knowledge Distillation with a Distribution Tutor for Medical Text Classification
di: Song, Yiping, et al.
Pubblicazione: (2024)

Enhancing Multimodal Sentiment Analysis for Missing Modality through Self-Distillation and Unified Modality Cross-Attention
di: Weng, Yuzhe, et al.
Pubblicazione: (2024)

LLMs Can Generate a Better Answer by Aggregating Their Own Responses
di: Li, Zichong, et al.
Pubblicazione: (2025)

Not All Errors Are Equal: Investigation of Speech Recognition Errors in Alzheimer's Disease Detection
di: Kang, Jiawen, et al.
Pubblicazione: (2024)

All Entities are Not Created Equal: Examining the Long Tail for Ultra-Fine Entity Typing
di: Deshmukh, Advait, et al.
Pubblicazione: (2024)

Is That Your Final Answer? Test-Time Scaling Improves Selective Question Answering
di: Jurayj, William, et al.
Pubblicazione: (2025)

Not All Thoughts are Generated Equal: Efficient LLM Reasoning via Multi-Turn Reinforcement Learning
di: Ning, Yansong, et al.
Pubblicazione: (2025)

Not All LLM-Generated Data Are Equal: Rethinking Data Weighting in Text Classification
di: Kuo, Hsun-Yu, et al.
Pubblicazione: (2024)

Same Question, Different Source, Different Answer: Auditing Source-Dependence in Medical Multi-Source RAG
di: Li, Yubo, et al.
Pubblicazione: (2026)

Not All Uncertainty Is Equal: How Uncertainty Granularity Shapes Human Verification in LLM-Assisted Decision Making
di: Villavicencio, Mauricio, et al.
Pubblicazione: (2026)

Correct Answers from Sound Reasoning: Verifiable Process Supervision for Language Models
di: Kim, Kyuyoung, et al.
Pubblicazione: (2026)

Your Teacher Can't Help You Here: Combating Supervision Fidelity Decay in On-Policy Distillation
di: Liu, Yanjiang, et al.
Pubblicazione: (2026)

Not All Preference Pairs Are Created Equal: A Recipe for Annotation-Efficient Iterative Preference Learning
di: Yang, Sen, et al.
Pubblicazione: (2024)