Axiomi — Axioms of intelligence. Live feed https://axiomi.si/ A curated directory of open-weight models, ML frameworks, fine-tuning tools, interpretability libraries, datasets and benchmarks — with live arXiv and Hugging Face feeds. en Tue, 06 Oct 2026 01:26:45 +0000 Dust: Pretraining Transformers Without Backpropagation https://qlabs.sh/research/dust [Hacker News] 96 points · 15 comments Hacker News Mon, 05 Oct 2026 21:15:07 +0000 axiomi.si:hn-49970871 Beam: Reflection's 501B open-weight model https://reflection.ai/blog/introducing-beam [Hacker News] 302 points · 78 comments Hacker News Mon, 05 Oct 2026 19:16:35 +0000 axiomi.si:hn-49969183 "Torturing" LLMs in a Robot Prison Has Triggered the Dumbest Debate in AI Yet https://www.404media.co/someone-torturing-llms-in-a-robot-prison-has-triggered-the-dumbest-debate-in-ai-yet/ [Hacker News] 46 points · 102 comments Hacker News Sun, 04 Oct 2026 08:02:33 +0000 axiomi.si:hn-49951684 Aleph Alpha Kolibri: How the sovereign German LLM works https://tej.as/blog/aleph-alpha-kolibri [Hacker News] 420 points · 12 comments Hacker News Sat, 03 Oct 2026 10:43:51 +0000 axiomi.si:hn-49943034 Kolibri: A Sovereign Open-Weight Model https://aleph-alpha.com/en/blog/kolibri-has-landed-a-sovereign-open-weight-model/ [Hacker News] 695 points · 335 comments Hacker News Sat, 03 Oct 2026 09:36:04 +0000 axiomi.si:hn-49942706 From the creator of Redis; run LLM locally with ds4 https://dwarfstar.sh/ [Hacker News] 362 points · 104 comments Hacker News Fri, 02 Oct 2026 18:01:16 +0000 axiomi.si:hn-49936575 Language Models that Play Chess and Explain Their Moves https://arxiv.org/abs/2610.03695v1 [arXiv] Modern chess engines are silent experts: they play at a superhuman level, but do not offer explanations for their play. On the other hand, language models (LMs) can generate plausible-sounding explanations, but their weak playing strength limits the utility of their explanations.… arXiv Fri, 02 Oct 2026 17:54:22 +0000 axiomi.si:arxiv-2610.03695v1 FrugalEvo: Towards Cost-Aware LLM-Guided Program Evolution https://arxiv.org/abs/2610.03675v1 [arXiv] LLM-guided evolutionary methods, such as AlphaEvolve, have emerged as powerful approaches for challenging computational optimization problems, such as circle packing. However, prior work typically optimizes performance gain over a fixed number of iterations. We argue that practic… arXiv Fri, 02 Oct 2026 17:44:05 +0000 axiomi.si:arxiv-2610.03675v1 Pivot-SD: Efficient Self-Distillation for Masked Diffusion Language Models https://arxiv.org/abs/2610.03665v1 [arXiv] Masked diffusion language models (dLMs) offer a promising parallel alternative to autoregressive models for complex reasoning. However, they face a distinct credit-assignment challenge, since a few commitments during denoising sharply reduce the uncertainty over the remaining mas… arXiv Fri, 02 Oct 2026 17:37:51 +0000 axiomi.si:arxiv-2610.03665v1 Writerslogic at the CLEF 2026 SimpleText Track: Multi-Candidate LLM Simplification and Stacked Complexity Spotting https://arxiv.org/abs/2610.03567v1 [arXiv] We describe the Writerslogic team's participation in the CLEF 2026 SimpleText shared task, addressing Task 1 (text simplification) and Task 2 (complexity spotting). For Task 1, we develop a multi-candidate generation pipeline using GPT-4o-mini that produces five simplification ca… arXiv Fri, 02 Oct 2026 16:44:53 +0000 axiomi.si:arxiv-2610.03567v1 Objects Without Morphisms: What LLMs for Mathematics Do Not Represent https://arxiv.org/abs/2610.03551v1 [arXiv] Large language models (LLMs) have reached expert-level performance on competition mathematics largely through the volume of search placed around them: candidate solutions are sampled in quantity and retained only when an external criterion accepts them. Such a procedure improves … arXiv Fri, 02 Oct 2026 16:33:32 +0000 axiomi.si:arxiv-2610.03551v1 Author Representation Strategies for Zero-Shot Authorship Attribution: A Comparative Study of LLM-Based and Embedding-Based Approaches https://arxiv.org/abs/2610.03531v1 [arXiv] Authorship Attribution (AA) requires capturing fine-grained stylistic characteristics, making it particularly challenging in zero-shot (ZS) settings where no task-specific supervision is available. In this work, we investigate the effect of author representations on ZS AA by eval… arXiv Fri, 02 Oct 2026 16:16:47 +0000 axiomi.si:arxiv-2610.03531v1 Single-Pass Uncertainty Heads for Claim-Level Hallucination Detection in Persian Medical Language Models https://arxiv.org/abs/2610.03482v1 [arXiv] Hallucination detection is particularly important for medical language models, but repeated-sampling approaches are expensive and existing uncertainty-head resources do not directly transfer to a new backbone and language. We adapt the LLM Uncertainty Head (LUH) framework to Aya-… arXiv Fri, 02 Oct 2026 15:50:46 +0000 axiomi.si:arxiv-2610.03482v1 Generalization of Transformer-Based Neural Quantum States via In-Context Learning https://arxiv.org/abs/2610.03463v1 [arXiv] Neural quantum states based on modern deep learning architectures have emerged as powerful representations for quantum many-body systems. In particular, Transformer-based neural quantum states provide expressive models capable of capturing long-range correlations, and their empir… arXiv Fri, 02 Oct 2026 15:38:10 +0000 axiomi.si:arxiv-2610.03463v1 Passing the Test You Trained On: Re-evaluating Prompt-Injection Detectors for LLM Agents https://arxiv.org/abs/2610.03448v1 [arXiv] LLM agents increasingly screen tool outputs with small prompt-injection detectors, and teams choose among detectors by their scores on public benchmarks. We ask whether those scores predict how a detector behaves inside an agent. We replay the ground-truth tool calls of two agent… arXiv Fri, 02 Oct 2026 15:30:11 +0000 axiomi.si:arxiv-2610.03448v1 From Patching to Pruning Visual Computation in Vision Language Models https://arxiv.org/abs/2610.03389v1 [arXiv] Vision language models (VLMs) incur substantial inference cost because every visual token is processed by the attention and MLP projections of every decoder layer, even when token-specific visual computation is unnecessary at many depths. We introduce Patch-to-Prune (P2P), inspir… arXiv Fri, 02 Oct 2026 14:37:53 +0000 axiomi.si:arxiv-2610.03389v1 SyntaxBench: A Statistical Diagnostic Framework for Character-Level Reasoning in Large Language Models https://arxiv.org/abs/2610.03329v1 [arXiv] Large language models are increasingly used where small syntactic errors matter, yet character-level reasoning is still evaluated mostly through isolated probes and aggregate accuracy. We introduce SyntaxBench, a diagnostic benchmark and statistical evaluation framework for chara… arXiv Fri, 02 Oct 2026 14:02:56 +0000 axiomi.si:arxiv-2610.03329v1 Decision models like Jev don't beat LLM-as-a-judge or traditional classifiers https://developers.redhat.com/articles/2026/10/02/benchmarking-ai-decision-models-against-traditional-guardrails [Hacker News] 153 points · 70 comments Hacker News Fri, 02 Oct 2026 13:47:41 +0000 axiomi.si:hn-49933476 Don't be fooled–LLMs don't reason https://www.technologyreview.com/2026/10/02/1145639/dont-be-fooled-llms-dont-reason/ [Hacker News] 76 points · 159 comments Hacker News Fri, 02 Oct 2026 13:45:42 +0000 axiomi.si:hn-49933459 JOVE: Joint Execution and Verification for Resource-Aware LLM Task Graphs https://arxiv.org/abs/2610.03296v1 [arXiv] Complex reasoning queries can be decomposed into directed acyclic task graphs and distributed across heterogeneous LLMs, reducing latency through parallelism and enabling smaller models to solve complex tasks. In practice, however, the suitability of an LLM for a given subtask ma… arXiv Fri, 02 Oct 2026 13:37:59 +0000 axiomi.si:arxiv-2610.03296v1 Wrong Organ, Right Physics: Transferring Echocardiography Pretraining to Lung Ultrasound for Tuberculosis Screening https://arxiv.org/abs/2610.03290v1 [arXiv] Lung ultrasound (LUS) is attractive for tuberculosis (TB) screening at primary-care level, but labelled cohorts are small. Echocardiography carries no such constraint, while sharing the same underlying ultrasound imaging physics, signal processing and B-mode appearance as LUS. We… arXiv Fri, 02 Oct 2026 13:33:18 +0000 axiomi.si:arxiv-2610.03290v1 SPEAR: A Spectral-Disentangled MoE Neural Operator with Knowledge-Guided Expert Aggregation for Large-Scale PDE Pretraining https://arxiv.org/abs/2610.03265v1 [arXiv] Large-scale pre-training has improved the generalization of neural operators across diverse PDEs. However, existing PDE foundation models still struggle with heterogeneous dynamics, where shared representations may cause knowledge interference, while mixture-of-experts (MoE) arch… arXiv Fri, 02 Oct 2026 13:10:09 +0000 axiomi.si:arxiv-2610.03265v1 D2K-Bench: Can LLM Agents Turn Expert Designs into Efficient GPU Kernels? https://arxiv.org/abs/2610.03226v1 [arXiv] GPU kernels generated by large language model (LLM) agents can remain less efficient than expert implementations, but runtime alone does not reveal how the gap relates to design discovery and implementation. We introduce D2K-Bench, a diagnostic benchmark of 26 tasks and 85 worklo… arXiv Fri, 02 Oct 2026 12:42:12 +0000 axiomi.si:arxiv-2610.03226v1 Predicting and Repairing Merge Collapse in Large Language Models https://arxiv.org/abs/2610.03199v1 [arXiv] Large language models fine-tuned from a shared base can be merged by averaging their task vectors, but some merges collapse far below the base model, and common merge operators give no warning before evaluation. We show that one statistic of the specialists' task vectors both pre… arXiv Fri, 02 Oct 2026 12:11:43 +0000 axiomi.si:arxiv-2610.03199v1 Greg Kroah-Hartman – Security in the LLM Age [video] https://www.youtube.com/watch?v=NnV_cWeoo5Q [Hacker News] 338 points · 127 comments Hacker News Fri, 02 Oct 2026 02:51:27 +0000 axiomi.si:hn-49929391 Native Action-Prior Learning from Videos for World Action Models https://huggingface.co/papers/2610.03391 [HF Daily Papers] 75 upvotes HF Daily Papers Thu, 01 Oct 2026 20:00:00 +0000 axiomi.si:hfp-2610.03391 HyperBrowseComp: A Multilingual and Multimodal Stress Test for Web-Browsing Agents https://huggingface.co/papers/2610.03574 [HF Daily Papers] 51 upvotes HF Daily Papers Thu, 01 Oct 2026 20:00:00 +0000 axiomi.si:hfp-2610.03574 Pivot-SD: Efficient Self-Distillation for Masked Diffusion Language Models https://huggingface.co/papers/2610.03665 [HF Daily Papers] 50 upvotes HF Daily Papers Thu, 01 Oct 2026 20:00:00 +0000 axiomi.si:hfp-2610.03665 Clef: Open-weight decision models, and new RL fine-tuning platform https://blog.cloudflare.com/clef-decision-models/ [Hacker News] 640 points · 217 comments Hacker News Thu, 01 Oct 2026 16:18:57 +0000 axiomi.si:hn-49923692 RealCompanion: Benchmarking Human Understanding from Reasoning over Longitudinal Real-World Conversations https://huggingface.co/papers/2610.01780 [HF Daily Papers] 250 upvotes HF Daily Papers Wed, 30 Sep 2026 20:00:00 +0000 axiomi.si:hfp-2610.01780