{"total":16,"items":[{"citing_arxiv_id":"2606.27981","ref_index":294,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"ToxiREX: A Dataset on Toxic REasoning in ConteXt","primary_cat":"cs.CL","submitted_at":"2026-06-26T11:30:42+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"ToxiREX is a new dataset of 128k Reddit comments in six languages with hierarchical annotations for implicit toxicity in conversational context based on an existing reasoning schema.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.19005","ref_index":13,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Sumi: Open Uniform Diffusion Language Model from Scratch","primary_cat":"cs.CL","submitted_at":"2026-06-17T12:32:46+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":8.0,"formal_verification":"none","one_line_summary":"Sumi is an openly released 7B parameter uniform diffusion language model pretrained from scratch on 1.5T tokens that matches autoregressive models on several benchmarks.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.18767","ref_index":29,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Output Vector Editing for Memorization Mitigation in Large Language Models","primary_cat":"cs.CL","submitted_at":"2026-06-17T07:29:18+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"Output vector editing on MLP neurons suppresses memorization in LLMs up to 87.9% on 6831 sequences in OLMo-7B with a 2.7x gap over zero ablation, ensemble covering 96.5%.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.18246","ref_index":12,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Variable-Width Transformers","primary_cat":"cs.CL","submitted_at":"2026-06-16T17:59:03+00:00","verdict":"CONDITIONAL","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"×-shaped variable-width transformers outperform parameter-matched uniform baselines on language modeling loss with 22% fewer FLOPs and 15% smaller KV cache.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.11105","ref_index":26,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"PhantomBench: Benchmarking the Non-existential Threat of Language Models","primary_cat":"cs.CL","submitted_at":"2026-06-09T17:03:19+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"PhantomBench is a new benchmark of 60K+ non-existent terms showing language models hallucinate at rates up to 86.7 percent even when inputs assume the concepts exist.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.11081","ref_index":12,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Unifying Local Communications and Local Updates for LLM Pretraining","primary_cat":"cs.LG","submitted_at":"2026-06-09T16:40:54+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"GASLoC generalizes communication acceleration to the outer optimizer to enable gossip-based decentralized LLM pretraining that supports adaptive optimizers, local steps, and outperforms prior decentralized methods on standard tasks while matching DiLoCo in multi-step regimes.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.06267","ref_index":36,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Many Circuits, One Mechanism: Input Variation and Evaluation Granularity in Circuit Discovery","primary_cat":"cs.CL","submitted_at":"2026-06-04T15:10:14+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"Structurally distinct circuits for literal sequence copying across token frequency bands implement the same computation, shown by broad transfer of band-specific edges, a shared core recovering 99% performance, and interchangeable representations via causal interventions.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.05616","ref_index":52,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"What's in a Name? Morphological Shortcuts by LLMs in Pharmacology","primary_cat":"cs.CL","submitted_at":"2026-06-04T02:40:42+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"LLMs induce pharmacological meaning primarily from affix cues in drug names, as revealed by a framework applied to 653 drugs and localized via activation patching to early-mid layers.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.01196","ref_index":9,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Low-Resource Safety Failures Are Action Failures, Not Representation Failures","primary_cat":"cs.CL","submitted_at":"2026-05-31T12:19:40+00:00","verdict":"CONDITIONAL","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"Low-resource safety failures are action failures because the harmfulness representation transfers but the decision calibration does not; this is fixed by recalibrating a high-resource gate with 1-4 target-language examples.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.14169","ref_index":54,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"BOOKMARKS: Efficient Active Storyline Memory for Role-playing","primary_cat":"cs.CL","submitted_at":"2026-05-13T22:48:24+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"BOOKMARKS introduces searchable bookmarks as reusable answers to storyline questions, enabling active initialization and passive synchronization for more consistent role-playing agent memory than recurrent summarization.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.15220","ref_index":32,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Always Learning, Always Mixing: Efficient and Simple Data Mixing All The Time","primary_cat":"cs.CL","submitted_at":"2026-05-13T02:29:19+00:00","verdict":"CONDITIONAL","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"OP-Mix is an on-policy data mixing method that uses low-rank adapter interpolation to find near-optimal data mixtures throughout language model training with reduced compute.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.26687","ref_index":12,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"COPUS: Co-adaptive Parallelism and Batch Size Selection in Large Language Model Training","primary_cat":"cs.DC","submitted_at":"2026-04-29T13:52:38+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"COPUS co-adapts batch size and parallelism during LLM training via goodput to deliver 3.9-8% average faster convergence than fixing one while tuning the other.","context_count":1,"top_context_role":"background","top_context_polarity":"background","context_text":"isolation. In state-of-the-art distributed training frameworks like Megatron-LM [41] and DeepSpeed [39], the 3D paral- lelism topology must be statically declared at initialization. Consequently, even the most advanced industrial pre- training runs often rely on compromises that can become suboptimal. For instance, during the training of AI2's OLMo 65B [12], the training recipe relies on a static configura- tion where the batch size is scheduled to repeatedly double (e.g., from roughly 2M to roughly 16M tokens) to track the evolving critical batch size. However, because the frame- work's parallel mesh is locked at launch, the system can only increase gradient accumulation steps to absorb the larger global batch size, leaving the underlying 3D parallelism strat-"},{"citing_arxiv_id":"2604.15873","ref_index":14,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"How Hypocritical Is Your LLM judge? Listener-Speaker Asymmetries in the Pragmatic Competence of Large Language Models","primary_cat":"cs.CL","submitted_at":"2026-04-17T09:22:16+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"LLMs perform substantially better as pragmatic listeners judging language than as speakers generating it, revealing weak alignment between the two roles.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.06005","ref_index":1,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Disentangling MLP Neuron Weights in Vocabulary Space","primary_cat":"cs.CL","submitted_at":"2026-04-07T15:39:45+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":8.0,"formal_verification":"none","one_line_summary":"ROTATE disentangles MLP neurons into faithful vocabulary channels by optimizing weight rotations to maximize vocabulary-space kurtosis, outperforming activation-based baselines for neuron descriptions.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2506.02132","ref_index":15,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Model Internal Sleuthing: Finding Lexical Identity and Inflectional Features in Modern Language Models","primary_cat":"cs.CL","submitted_at":"2025-06-02T18:01:56+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"Inflectional features stay linearly decodable across all layers while lexical identity weakens with depth in modern transformers.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2502.12120","ref_index":17,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"LLMs on the Line: Data Determines Loss-to-Loss Scaling Laws","primary_cat":"cs.LG","submitted_at":"2025-02-17T18:45:25+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Pretraining data determines loss-to-loss scaling laws in LLMs, while model size, optimization, tokenizer, and architecture have limited impact.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null}],"limit":50,"offset":0}