New Links | ML News

ML Newsnew | past | best | rss | submit

331.	Towards Physics of Multimodal Pretraining (arxiv.org)
	1 point by gmays 43 days ago \| hide \| past \| pdf \| discuss
332.	FreeToken: Efficient Edge-Native Moe Serving with Bandwidth-Adaptive Execution (arxiv.org)
	4 points by throwaway2027 43 days ago \| hide \| past \| pdf \| discuss
333.	Every skill-evolution method authored unsafe skills across four agent harnesses (arxiv.org)
	2 points by zwdommy 43 days ago \| hide \| past \| pdf \| discuss
334.	AVO: Agentic Variation Operators for Autonomous Evolutionary Search (arxiv.org)
	3 points by ijidak 43 days ago \| hide \| past \| pdf \| discuss
335.	Mind Viruses: Self-Propagating Ideas in Multi-Agent LLM Systems (arxiv.org)
	2 points by jdcaballerov 43 days ago \| hide \| past \| pdf \| discuss
336.	Phantom Gains: Auditing Self-Improvement Against a Measured Null (arxiv.org)
	4 points by sbulaev 43 days ago \| hide \| past \| pdf \| discuss
337.	Bounded Agents: Delegation Security for Multi-Agent AI Systems (arxiv.org)
	3 points by xmuruaga 44 days ago \| hide \| past \| pdf \| discuss
338.	Inadvertent Context Leakage in Language Models (arxiv.org)
	1 point by sbulaev 44 days ago \| hide \| past \| pdf \| discuss
339.	Compress and Forget: Bitsandbytes Quantization Amplifies Proactive Interference (arxiv.org)
	1 point by sbulaev 44 days ago \| hide \| past \| pdf \| discuss
340.	Can LRMs Do Analogical Reasoning Under Perceptual Uncertainty? (2025) (arxiv.org)
	2 points by wslh 44 days ago \| hide \| past \| pdf \| discuss
341.	Spade: Self-Play in Adaptive Synthetic Executable Environments (arxiv.org)
	2 points by ironyman 44 days ago \| hide \| past \| pdf \| discuss
342.	The Spectral Neuron (arxiv.org)
	4 points by E-Reverance 44 days ago \| hide \| past \| pdf \| discuss
343.	VulnGym: Benchmarking Coding Agents for Repository-Level Vulnerability Detection (arxiv.org)
	2 points by wslh 44 days ago \| hide \| past \| pdf \| discuss
344.	DiffusionGemma Technical Report (arxiv.org)
	165 points by gmays 44 days ago \| hide \| past \| pdf \| 40 comments
345.	Do Chatbot LLMs Talk Too Much? (arxiv.org)
	12 points by bretkoppel 44 days ago \| hide \| past \| pdf \| 4 comments
346.	Test-Time Scaling in the Wild: Why Exploitation, Not Exploration (arxiv.org)
	1 point by sbulaev 45 days ago \| hide \| past \| pdf \| discuss
347.	Fool's Gold: Defensive Deception Against Safety-Removal Attacks on Open-Weight (arxiv.org)
	2 points by sbulaev 45 days ago \| hide \| past \| pdf \| discuss
348.	Cost-Aware Optimization for Agentic Query Execution (arxiv.org)
	1 point by matt_d 45 days ago \| hide \| past \| pdf \| discuss
349.	Stealing Reasoning Traces from Proprietary LLM APIs (arxiv.org)
	3 points by doppp 45 days ago \| hide \| past \| pdf \| discuss
350.	Universality of Gradient Descent Neural Network Training (arxiv.org)
	39 points by E-Reverance 45 days ago \| hide \| past \| pdf \| 2 comments
351.	Stealing Reasoning Traces from Proprietary LLM APIs (arxiv.org)
	2 points by amai 45 days ago \| hide \| past \| pdf \| 1 comment
352.	Chain-of-Thought Reasoning in the Wild Is Not Always Faithful (2025) (arxiv.org)
	66 points by florianherrengt 45 days ago \| hide \| past \| pdf \| 38 comments
353.	Improving the Matrix Multiplication Exponent (arxiv.org)
	6 points by aaraujo002 45 days ago \| hide \| past \| pdf \| discuss
354.	Adding Error Bars to Evals: A Statistical Approach to Language Model Evaluations (arxiv.org)
	3 points by Anon84 45 days ago \| hide \| past \| pdf \| discuss
355.	Sampling More, Getting Less: Calibration Is the Diversity Bottleneck in LLMs (arxiv.org)
	2 points by clukic 45 days ago \| hide \| past \| pdf \| discuss
356.	Stop Anthropomorphizing Intermediate Tokens as Reasoning/Thinking Traces (2025) (arxiv.org)
	316 points by nunodonato 45 days ago \| hide \| past \| pdf \| 281 comments
357.	AutoResearch: Insight In, Hallucination Out (arxiv.org)
	2 points by johnbarron 45 days ago \| hide \| past \| pdf \| discuss
358.	Benchmarking General-Purpose Agents on Market-Validated End-to-End Workflows (arxiv.org)
	3 points by tcp_handshaker 45 days ago \| hide \| past \| pdf \| discuss
359.	Fragility of Self-Improving Agents: Variance, Task Order, and Underspecification (arxiv.org)
	3 points by tcp_handshaker 45 days ago \| hide \| past \| pdf \| discuss
360.	Model Hypnosis: Strong control of AI via additive subliminal effects (arxiv.org)
	3 points by sbulaev 46 days ago \| hide \| past \| pdf \| discuss
	More