XGRAG uses graph perturbations to quantify component contributions in GraphRAG and achieves 14.81% better explanation quality than text-based baselines on QA datasets, with correlations to graph centrality.
Deepseek-r1 thoughtology: Let's think about llm reasoning
9 Pith papers cite this work. Polarity classification is still indexing.
9
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
background 1representative citing papers
DecompSR is a large, symbolically verified benchmark dataset and generation framework that independently varies productivity, substitutivity, overgeneralisation, and systematicity to probe compositional multihop spatial reasoning in LLMs.
Reasoning gaps between base LLMs and LRMs concentrate on ~8% of early planning tokens; intervening with the reasoning model only at high-disagreement positions recovers performance.
ReasoningGuard is an inference-time method that uses attention mechanisms to inject safety aha moments and scaling sampling to defend large reasoning models against jailbreak attacks.
LRMs exhibit complete accuracy collapse beyond certain puzzle complexities, with reasoning effort rising then declining, outperforming standard LLMs only on medium-complexity tasks.
Non-reasoning LLMs fail the equivalence class problem while reasoning LLMs perform better but remain incomplete, with difficulty peaking at phase transition for the former and maximum diameter for the latter.
SFI-Bench shows current multimodal LLMs struggle to integrate spatial memory with functional reasoning and external knowledge in video tasks.
The paper introduces Language Specific Knowledge (LSK) and shows that selecting an optimal non-English language for a query can improve LLM performance on cultural and social norm datasets.
citing papers explorer
-
XGRAG: A Graph-Native Framework for Explaining KG-based Retrieval-Augmented Generation
XGRAG uses graph perturbations to quantify component contributions in GraphRAG and achieves 14.81% better explanation quality than text-based baselines on QA datasets, with correlations to graph centrality.
-
DecompSR: A dataset for decomposed analyses of compositional multihop spatial reasoning
DecompSR is a large, symbolically verified benchmark dataset and generation framework that independently varies productivity, substitutivity, overgeneralisation, and systematicity to probe compositional multihop spatial reasoning in LLMs.
-
Reasoning Can Be Restored by Correcting a Few Decision Tokens
Reasoning gaps between base LLMs and LRMs concentrate on ~8% of early planning tokens; intervening with the reasoning model only at high-disagreement positions recovers performance.
-
ReasoningGuard: Safeguarding Large Reasoning Models with Inference-time Safety Aha Moments
ReasoningGuard is an inference-time method that uses attention mechanisms to inject safety aha moments and scaling sampling to defend large reasoning models against jailbreak attacks.
-
The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexity
LRMs exhibit complete accuracy collapse beyond certain puzzle complexities, with reasoning effort rising then declining, outperforming standard LLMs only on medium-complexity tasks.
-
How Well Do LLMs Perform on the Simplest Long-Chain Reasoning Tasks: An Empirical Study on the Equivalence Class Problem
Non-reasoning LLMs fail the equivalence class problem while reasoning LLMs perform better but remain incomplete, with difficulty peaking at phase transition for the former and maximum diameter for the latter.
-
From Where Things Are to What They Are For: Benchmarking Spatial-Functional Intelligence in Multimodal LLMs
SFI-Bench shows current multimodal LLMs struggle to integrate spatial memory with functional reasoning and external knowledge in video tasks.
-
Language Specific Knowledge: Do Models Know Better in X than in English?
The paper introduces Language Specific Knowledge (LSK) and shows that selecting an optimal non-English language for a query can improve LLM performance on cultural and social norm datasets.
- When LLMs Stop Following Steps: A Diagnostic Study of Procedural Execution in Language Models