Cover Image

From Reasoning to Answer: Empirical, Attention-Based and Mechanistic Insights into Distilled DeepSeek R1 Models

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Zhang, Jue, Lin, Qingwei, Rajmohan, Saravan, Zhang, Dongmei
Format:	Preprint
Published:	2025
Subjects:	Artificial Intelligence Computation and Language
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Contrastive Attribution in the Wild: An Interpretability Analysis of LLM Failures on Realistic Benchmarks
by: Tan, Rongyuan, et al.
Published: (2026)

LLM Reasoning as Trajectories: Step-Specific Representation Geometry and Correctness Signals
by: Sun, Lihao, et al.
Published: (2026)

MEETING DELEGATE: Benchmarking LLMs on Attending Meetings on Our Behalf
by: Hu, Lingxiang, et al.
Published: (2025)

Navigating the Unknown: A Chat-Based Collaborative Interface for Personalized Exploratory Tasks
by: Peng, Yingzhe, et al.
Published: (2024)

Distill Not Only Data but Also Rewards: Can Smaller Language Models Surpass Larger Ones?
by: Zhang, Yudi, et al.
Published: (2025)

AutoRAG-HP: Automatic Online Hyper-Parameter Tuning for Retrieval-Augmented Generation
by: Fu, Jia, et al.
Published: (2024)

G-KV: Decoding-Time KV Cache Eviction with Global Attention
by: Liao, Mengqi, et al.
Published: (2025)

RealSafe-R1: Safety-Aligned DeepSeek-R1 without Compromising Reasoning Capability
by: Zhang, Yichi, et al.
Published: (2025)

Text2Grad: Reinforcement Learning from Natural Language Feedback
by: Wang, Hanyang, et al.
Published: (2025)

Learning to Refine: Self-Refinement of Parallel Reasoning in LLMs
by: Wang, Qibin, et al.
Published: (2025)

DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
by: DeepSeek-AI, et al.
Published: (2025)

DoVer: Intervention-Driven Auto Debugging for LLM Multi-Agent Systems
by: Ma, Ming, et al.
Published: (2025)

The Vision of Autonomic Computing: Can LLMs Make It a Reality?
by: Zhang, Zhiyang, et al.
Published: (2024)

Safety Evaluation of DeepSeek Models in Chinese Contexts
by: Zhang, Wenjing, et al.
Published: (2025)

Call Me When Necessary: LLMs can Efficiently and Faithfully Reason over Structured Environments
by: Cheng, Sitao, et al.
Published: (2024)

DeepSeek-V3 Technical Report
by: DeepSeek-AI, et al.
Published: (2024)

An evaluation of DeepSeek Models in Biomedical Natural Language Processing
by: Zhan, Zaifu, et al.
Published: (2025)

Self-Evolved Reward Learning for LLMs
by: Huang, Chenghua, et al.
Published: (2024)

Enabling Autonomic Microservice Management through Self-Learning Agents
by: Yu, Fenglin, et al.
Published: (2025)

EfficientRAG: Efficient Retriever for Multi-Hop Question Answering
by: Zhuang, Ziyuan, et al.
Published: (2024)

Output Length Effect on DeepSeek-R1's Safety in Forced Thinking
by: Li, Xuying, et al.
Published: (2025)

Mixture of Tunable Experts -- Behavior Modification of DeepSeek-R1 at Inference Time
by: Dahlke, Robert, et al.
Published: (2025)

A Comparison of DeepSeek and Other LLMs
by: Gao, Tianchen, et al.
Published: (2025)

Privacy in Action: Towards Realistic Privacy Mitigation and Evaluation for LLM-Powered Agents
by: Wang, Shouju, et al.
Published: (2025)

AXIS: Efficient Human-Agent-Computer Interaction with API-First LLM-Based Agents
by: Lu, Junting, et al.
Published: (2024)

Towards Understanding the Safety Boundaries of DeepSeek Models: Evaluation and Findings
by: Ying, Zonghao, et al.
Published: (2025)

Explainable Sentiment Analysis with DeepSeek-R1: Performance, Efficiency, and Few-Shot Learning
by: Huang, Donghao, et al.
Published: (2025)

Zipage: Maintain High Request Concurrency for LLM Reasoning through Compressed PagedAttention
by: Liao, Mengqi, et al.
Published: (2026)

Safety Evaluation and Enhancement of DeepSeek Models in Chinese Contexts
by: Zhang, Wenjing, et al.
Published: (2025)

DeepSeek performs better than other Large Language Models in Dental Cases
by: Zhang, Hexian, et al.
Published: (2025)

Towards Economical Inference: Enabling DeepSeek's Multi-Head Latent Attention in Any Transformer-based LLMs
by: Ji, Tao, et al.
Published: (2025)

DeepSeek LLM: Scaling Open-Source Language Models with Longtermism
by: DeepSeek-AI, et al.
Published: (2024)

Reasoning and the Trusting Behavior of DeepSeek and GPT: An Experiment Revealing Hidden Fault Lines in Large Language Models
by: Li, Rubing, et al.
Published: (2025)

DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model
by: DeepSeek-AI, et al.
Published: (2024)

DeepSeek-Prover-V2: Advancing Formal Mathematical Reasoning via Reinforcement Learning for Subgoal Decomposition
by: Ren, Z. Z., et al.
Published: (2025)

Are Large Language Models Capable of Deep Relational Reasoning? Insights from DeepSeek-R1 and Benchmark Comparisons
by: So, Chi Chiu, et al.
Published: (2025)

Challenges in Ensuring AI Safety in DeepSeek-R1 Models: The Shortcomings of Reinforcement Learning Strategies
by: Parmar, Manojkumar, et al.
Published: (2025)

AgentGen: Enhancing Planning Abilities for Large Language Model based Agent via Environment and Task Generation
by: Hu, Mengkang, et al.
Published: (2024)

RuAG: Learned-rule-augmented Generation for Large Language Models
by: Zhang, Yudi, et al.
Published: (2024)

Large Language Model-Brained GUI Agents: A Survey
by: Zhang, Chaoyun, et al.
Published: (2024)