Large Language Models (LLMs) often incur an alignment tax: safety post-training can reduce general utility (e.g., reasoning and coding). We argue that this tax primarily arises from continual-learning-style forgetting in sequential alignment, where distribution shift and conflicting objectives cause safety updates to overwrite pre-trained competencies. Accordingly, we cast safety alignment as a continual learning (CL) problem that must balance plasticity (acquiring safety constraints) and stability (preserving general abilities). We propose Orthogonal Gradient Projection for Safety Alignment (OGPSA), a […]

Ver mais

I Built ArtemisFlow: A Local-First Job Tracker for Tech Candidates Modern hiring processes often involve multiple companies, stages, and timelines happening in parallel. At some point, keeping track of everything becomes harder than it should be. Like many developers, I initially used spreadsheets. They worked well enough in the beginning. But over time, they became harder to maintain. I kept adding columns, moving things around, and writing notes in places that didn’t really belong anywhere. Some companies replied […]

Ver mais

arXiv:2603.23520v1 Announce Type: new Abstract: Medicine is an empirical discipline refined through long-term observation and the messy, high-variance reality of clinical practice. Physicians build diagnostic and therapeutic competence through repeated cycles of application, reflection, and improvement, forming individualized methodologies. Yet outcomes vary widely, and master physicians’ knowledge systems are slow to develop and hard to transmit at scale, contributing to the scarcity of high-quality clinical expertise. To address this, we propose Med-Shicheng, a general framework that enables large […]

Ver mais

arXiv:2602.06079v1 Announce Type: new Abstract: The scaling of Large Language Models (LLMs) drives interest in matrix-based optimizers (e.g., Shampoo, Muon, SOAP) for their convergence efficiency; yet their requirement for holistic updates conflicts with the tensor fragmentation in distributed frameworks like Megatron. Existing solutions are suboptimal: synchronous approaches suffer from computational redundancy, while layer-wise partitioning fails to reconcile this conflict without violating the geometric constraints of efficient communication primitives. To bridge this gap, we propose Canzona, a Unified, Asynchronous, […]

Ver mais

Table of Links Abstract and Introduction Definitions Literature Review Argument Development The AI Model’s Potential for Feeling During Inference Conclusion and References Abstract This paper explores the hypothesis that the OpenAI-o1 model–a transformer-based AI trained with reinforcement learning from human feedback (RLHF)–displays characteristics of consciousness during its training and inference phases. Adopting functionalism, which argues that mental states are defined by their functional roles, we assess the possibility of AI consciousness. Drawing on theories from neuroscience, philosophy of […]

Ver mais

arXiv:2601.00263v1 Announce Type: new Abstract: Counterfactuals refer to minimally edited inputs that cause a model’s prediction to change, serving as a promising approach to explaining the model’s behavior. Large language models (LLMs) excel at generating English counterfactuals and demonstrate multilingual proficiency. However, their effectiveness in generating multilingual counterfactuals remains unclear. To this end, we conduct a comprehensive study on multilingual counterfactuals. We first conduct automatic evaluations on both directly generated counterfactuals in the target languages and those derived […]

Ver mais

APIs are now the default interface for modern applications. They expose business logic, identity flows, mobile backends, partner integrations and internal services. That also means API testing often requires more than a generic web scan. Burp Suite is excellent for manual analysis, but API security testing benefits from repeatable automation: checking JWT behavior, validating object-level authorization, detecting mass assignment patterns and confirming whether rate limits are enforced server-side. This article describes a safe design approach for Burp Suite […]

Ver mais

arXiv:2512.04299v2 Announce Type: replace-cross Abstract: Spectral gradient methods, such as the recently popularized Muon optimizer, are a promising alternative to standard Euclidean gradient descent for training deep neural networks and transformers, but it is still unclear in which regimes they are expected to perform better. We propose a simple layerwise condition that predicts when a spectral update yields a larger decrease in the loss than a Euclidean gradient step. This condition compares, for each parameter block, the squared […]

Ver mais

arXiv:2511.20605v3 Announce Type: replace-cross Abstract: We introduce and analyse active learning markets as a way to purchase labels, in situations where analysts aim to acquire additional data to improve model fitting, or to better train models for predictive analytics applications. This comes in contrast to the many proposals that already exist to purchase features and examples. By originally formalising the market clearing as an optimisation problem, we integrate budget constraints and improvement thresholds into the label acquisition process. […]

Ver mais