{"total":16,"items":[{"citing_arxiv_id":"2606.26986","ref_index":69,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"ReaORE: Reasoning-Guided Progressive Open Relation Extraction Empowered by Large Reasoning Models","primary_cat":"cs.CL","submitted_at":"2026-06-25T12:59:53+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":4.0,"formal_verification":"none","one_line_summary":"ReaORE is a progressive open relation extraction method that applies coarse-to-fine reasoning to improve generalization to unseen relations over clustering or direct LLM generation.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.26466","ref_index":87,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Soft Token Alignment for Cross-Lingual Reasoning","primary_cat":"cs.CL","submitted_at":"2026-06-25T00:01:58+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"SOLAR aligns soft-token probability mixtures across languages in embedding space during SFT and raises multilingual reasoning accuracy by up to 17.7 points over the base model.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.11853","ref_index":224,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Task-Aware Structured Memory for Dynamic Multi-modal In-Context Learning","primary_cat":"cs.CV","submitted_at":"2026-06-10T09:30:25+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"TASM proposes a task-aware structured memory framework using task-vector compression, bipartite token merging, and a Core Memory plus Latent Bank hierarchy to enable efficient dynamic multi-modal in-context learning.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.09932","ref_index":126,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"When RL Fails after SFT: Rejuvenating Model Plasticity for Robust SFT-to-RL Handoff","primary_cat":"cs.LG","submitted_at":"2026-06-07T17:58:58+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"Excessive SFT reduces LLM plasticity for RL; Rejuvenation restores it via base-anchored fusion and targeted neuron resets, yielding better RL performance and OOD generalization.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.08147","ref_index":42,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Biological Reasoning-Informed Regression for Interpretable Regulatory DNA Activity Prediction","primary_cat":"q-bio.GN","submitted_at":"2026-06-06T12:56:08+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"R3LM trains LLMs via two-stage reasoning-then-regression on a new dataset CRE-ReasonBench with mechanistic traces, achieving SOTA enhancer activity prediction across three cell types with interpretable outputs.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.22223","ref_index":14,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"How Many Different Outputs Can a Transformer Generate?","primary_cat":"cs.LG","submitted_at":"2026-05-21T09:26:37+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Transformers are limited to a linearly growing number of accessible output sequences with prompt length, with exponential decay in accessible proportion beyond a critical point, even under unbounded context.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.17978","ref_index":97,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"AutoVecCoder: Teaching LLMs to Generate Explicitly Vectorized Code","primary_cat":"cs.CL","submitted_at":"2026-05-18T07:33:15+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"AutoVecCoder combines VecPrompt for automated intrinsic knowledge synthesis and VecRL for efficiency-aligned RL to train an 8B LLM that achieves SOTA on SimdBench SSE/AVX subsets and sometimes exceeds -O3 compiler results.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.11518","ref_index":40,"ref_count":2,"confidence":0.88,"is_internal_anchor":false,"paper_title":"AutoLLMResearch: Training Research Agents for Automating LLM Experiment Configuration - Learning from Cheap, Optimizing Expensive","primary_cat":"cs.AI","submitted_at":"2026-05-12T04:42:35+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"AutoLLMResearch trains agents in a multi-fidelity LLMConfig-Gym environment formulated as a long-horizon MDP to enable cross-fidelity extrapolation for automating high-cost LLM experiment configurations.","context_count":1,"top_context_role":"background","top_context_polarity":"background","context_text":"regions diverge, a property critical for real-world deployment.Result 10:Our method degrades far more gracefully across both stress tests and recovers quickly, showing that text-based reasoning learns a deeper extrapolation strategy that remains robust under adversarial low-fidelity regimes. Broader Impact and FutureThis work contributes to broader AI Scientist [40]. Our framework can be extended to more LLM configuration scenarios and broadly to any domain where cheap trials can guide expensive decisions (catalyst optimization, etc [41, 42]). Future directions: 1) expanding our Gym with more tasks and fidelity; 2) multi-objective optimization that balances competing goals. More broadly,toward Recursive LLM Design,as foreshadowed in our introduction, AutoLLMResearch"},{"citing_arxiv_id":"2605.08031","ref_index":19,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Object Hallucination-Free Reinforcement Unlearning for Vision-Language Models","primary_cat":"cs.CV","submitted_at":"2026-05-08T17:19:53+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"HFRU is a two-stage reinforcement unlearning method operating on the vision encoder with GRPO optimization and an abstraction reward that achieves over 98% forgetting and retention on object and face tasks with negligible hallucination.","context_count":1,"top_context_role":"background","top_context_polarity":"background","context_text":"Since πref assigns non-zero probability to such sequences, it follows that Zcomp(x, p)> Z pen(x, p).(18) Now consider any hallucinated sequence yh, i.e., a sequence containing tokens in DHallu such that DHallu ∩ O(x) =∅ . By construction, such sequences do not contain valid hypernyms, hence Rabs(yh) = 0. Therefore, their numerators remain unchanged: πcomp(yh |x, p) = πref(yh |x, p) exp \u0010 Rpen(yh) β \u0011 Zcomp(x, p) .(19) SinceZ comp(x, p)> Z pen(x, p), we obtain πcomp(yh |x, p)< π pen(yh |x, p),∀(x, p)∈ D f .(20) Summing over all hallucinated sequences yields Phallu(πcomp |x, p)< P hallu(πpen |x, p),∀(x, p)∈ D f .(21) Taking expectation over the forget setD f completes the proof: E(x,p)∼Df [Phallu(πcomp |x, p)]