When Activation Oracles Learn Not to Read: Concept-Specific Blind Spots in Fine-Tuned Oracles
Paper • 2607.23379 • Published • 18
Andrey Getmanov
AndreyGetmanov
·
AI & ML interests
None yet
Recent Activity
upvoted a paper 8 days ago
CADENA: Stepwise CAD Reverse Engineering upvoted a paper 23 days ago
RAGU: A Multi-Step GraphRAG Engine with a Compact Domain-Adapted LLMOrganizations
None yet
upvoted a paper 2 days ago
upvoted a paper 8 days ago
upvoted a paper 23 days ago
upvoted a collection about 1 month ago
upvoted a paper 2 months ago
upvoted 2
papers 3 months ago
RaguTeam at SemEval-2026 Task 8: Meno and Friends in a Judge-Orchestrated LLM Ensemble for Faithful Multi-Turn Response Generation
Paper • 2605.04523 • Published • 48
RADIO-ViPE: Online Tightly Coupled Multi-Modal Fusion for Open-Vocabulary Semantic SLAM in Dynamic Environments
Paper • 2604.26067 • Published • 75
upvoted 3
papers 6 months ago
AIRS-Bench: a Suite of Tasks for Frontier AI Research Science Agents
Paper • 2602.06855 • Published • 83
AudioSAE: Towards Understanding of Audio-Processing Models with Sparse AutoEncoders
Paper • 2602.05027 • Published • 63
SLIME: Stabilized Likelihood Implicit Margin Enforcement for Preference Optimization
Paper • 2602.02383 • Published • 30
upvoted a paper 7 months ago
upvoted 4
papers 9 months ago
Multimodal Evaluation of Russian-language Architectures
Paper • 2511.15552 • Published • 79
Unveiling Intrinsic Dimension of Texts: from Academic Abstract to Creative Story
Paper • 2511.15210 • Published • 91
Souper-Model: How Simple Arithmetic Unlocks State-of-the-Art LLM Performance
Paper • 2511.13254 • Published • 140
MADD: Multi-Agent Drug Discovery Orchestra
Paper • 2511.08217 • Published • 57