defect_zero_iff_one

"This uniform aggregation causes uncontrolled hidden-state growth with depth, progressively diluting each layer’s contribution. We propose Attention Residuals (AttnRes), which replaces this fixed accumulation with softmax attention over preceding layer outputs"

"we design DPO-Positive (DPOP), a new loss function and training procedure which avoids this failure mode"

"Moreover, we find that ChatGPT is 63.41% accurate on average in 10 different reasoning categories under logical reasoning, non-textual reasoning, and commonsense reasoning, hence making it an unreliable reasoner. It is, for example, better at deductive than inductive reasoning."

"Gated SAEs solve shrinkage, are similarly interpretable, and require half as many firing features to achieve comparable reconstruction fidelity"

"we propose MEMRL, a non-parametric approach that evolves via reinforcement learning on episodic memory. By decoupling stable reasoning from plastic memory, MEMRL employs a Two-Phase Retrieval mechanism to filter noise and identify high-utility strategies through environmental feedback."

"We also manually annotate the “win/tie/lose” outcomes of responses from ChatGPT and Vicuna-13B"

"SCoRe addresses these challenges by training under the model’s own distribution of self-generated correction traces and using appropriate regularization to steer the learning process into learning a self-correction behavior that is effective at test time as opposed to fitting high-reward responses for a given prompt."

"This task generates a dense, self-supervised signal that compels the model to learn the underlying dynamics of the driving environment."

"high scores primarily reflect rote memorization of training data rather than genuine task understanding or execution ability"

"CLEVRER includes four types of questions: descriptive (e.g., 'what color'), explanatory ('what is responsible for'), predictive ('what will happen next'), and counterfactual ('what if')."

"We find that attention sink acts more like key biases, storing extra attention scores, which could be non-informative and not contribute to the value computation. We also observe that this phenomenon (at least partially) stems from tokens' inner dependence on attention scores as a result of softmax normalization."

"MemoryBank incorporates a memory updating mechanism, inspired by the Ebbinghaus Forgetting Curve theory, which permits the AI to forget and reinforce memory based on time elapsed and the relative significance of the memory"

"Brain-only participants exhibited the strongest, most distributed networks; Search Engine users showed moderate engagement; and LLM users displayed the weakest connectivity. Over four months, LLM users consistently underperformed at neural, linguistic, and behavioral levels."

"We establish a scalable pipeline to construct a large-scale medical visual question-answering dataset, named PMC-VQA, which contains 227k VQA pairs of 149k images that cover various modalities or diseases."

"the reward is a similarity score (between 0 and 1) of the predicted and the oracle patch calculated by Python’s difflib.SequenceMatcher"

"MIRIX consists of six distinct, carefully structured memory types: Core, Episodic, Semantic, Procedural, Resource Memory, and Knowledge Vault, coupled with a multi-agent framework that dynamically controls and coordinates updates and retrieval."

"TTRL boosts the pass@1 performance of Qwen-2.5-Math-7B by approximately 211% on the AIME 2024 with only unlabeled test data. Furthermore, although TTRL is only supervised by the maj@n metric, TTRL has demonstrated performance to consistently surpass the upper limit of the initial model maj@n."

"Theoretically, we prove that the global optimum to the training objective function of our method is achieved only when the LLM policy aligns with the target data distribution."

"This cost exhibits reciprocity J(x)=J(x⁻¹), vanishes only at x=1, and diverges at boundary regimes x→0⁺ and x→∞"

"we introduce OpenVid-1M, a precise high-quality dataset with expressive captions... propose a novel Multi-modal Video Diffusion Transformer (MVDiT) capable of mining both structure information from visual tokens and semantic information from text tokens"

"We introduce the concept of pixel-space reasoning... VLMs are equipped with a suite of visual reasoning operations, such as zoom-in and select-frame... a reinforcement learning (RL) phase leverages a curiosity-driven reward scheme to balance exploration between pixel-space reasoning and textual reasoning."

"The benchmark includes a diverse set of 110 explicitly malicious agent tasks (440 with augmentations), covering 11 harm categories including fraud, cybercrime, and harassment. In addition to measuring whether models refuse harmful agentic requests, scoring well on AgentHarm requires jailbroken agents to maintain their capabilities following an attack to complete a multi-step task."

"categorize existing works into... (1) model-based efficient reasoning... (2) reasoning output-based efficient reasoning, which aims to dynamically reduce reasoning steps and length during inference; (3) input prompts-based efficient reasoning"

"the graphical model for control as inference... optimality variables... p(Ot=1|st,at)=exp(r(st,at))"

"We introduce SHIFT, where we improve the generalization of a classifier by ablating features that a human judges to be task-irrelevant."

"recent empirical analyses reveal that 26.1% of community-contributed skills contain vulnerabilities, motivating our proposed Skill Trust and Lifecycle Governance Framework—a four-tier, gate-based permission model"

"VALL-E emerges in-context learning capabilities and can be used to synthesize high-quality personalized speech with only a 3-second enrolled recording of an unseen speaker as an acoustic prompt"

"We find the model complies with harmful queries from free users 14% of the time, versus almost never for paid users. In almost all cases where the model complies with a harmful query from a free user, we observe explicit alignment-faking reasoning, with the model stating it is strategically answering harmful queries in training to preserve its preferred harmlessness behavior out of training."

"the learned critic in Reasoner-Zero training effectively identifies and devalues repetitive response patterns"

"When completed, the dataset will contain over one and a half million captions describing over 330,000 images. For the training and validation images, five independent human generated captions will be provided."