How does the choice of training data influence an AI model? This question is of central importance to interpretability, privacy, and basic science. At its core is the data deletion problem: after a reasonable amount of precomputation, quickly predict how the model would behave in a given situation if a given subset of training data had been excluded from the learning algorithm. We present a data deletion scheme capable of predicting model outputs with vanishing error $varepsilon$ in […]

Ver mais

arXiv:2511.08401v4 Announce Type: replace Abstract: We prove that training a source model optimally for its own task is generically suboptimal when the objective is downstream transfer. We study the source-side optimization problem in L2-SP ridge regression and show a fundamental mismatch between the source-optimal and transfer-optimal source regularization: outside of a measure-zero set, $tau_0^* neq tau_S^*$. We characterize the transfer-optimal source penalty $tau_0^*$ as a function of task alignment and identify an alignment-dependent reversal: with imperfect alignment ($0<rho<1$), […]

Ver mais

Many people—especially AI company employees [1] —believe current AI systems are well-aligned in the sense of genuinely trying to do what they’re supposed to do (e.g., following their spec or constitution, obeying a reasonable interpretation of instructions). [2] I disagree. Current AI systems seem pretty misaligned to me in a mundane behavioral sense: they oversell their work, downplay or fail to mention problems, stop working early and claim to have finished when they clearly haven’t, and often seem […]

Ver mais

Large language models (LLMs) suffer from proactive interference (PI): outdated information in the context window disrupts retrieval of current values. This interference degrades retrieval accuracy log-linearly as stale associations accumulate, a bottleneck that persists regardless of context length and resists prompt-engineering mitigations. Biological brains resolve an analogous challenge through sleep-dependent memory consolidation: synaptic downscaling, selective replay, and targeted forgetting. We propose SleepGate, a biologically inspired framework that augments transformer-based LLMs with a learned sleep cycle over the key-value […]

Ver mais

As Generative AI shifts from simple retrieval to autonomous action, enterprises face the engineering challenge of scaling scattered proofs of concept into robust, secure systems. This article deconstructs the architecture of a production-grade AI Agent Platform, positioning it as an internal Platform-as-a-Service (PaaS) that supports both code-first engineers and low-code integrators. By breaking the system down into seven logical containers—Interaction, Development, Core, Foundation, Information, Observability, and Trust—the guide outlines a blueprint for a “factory” capable of serving, securing, […]

Ver mais

This story was originally published in February 2024. The minimum requirements for Tier 1 toolchains targeting Windows will increase with the 1.78 release (scheduled for May 02, 2024). Windows 10 will now be the minimum supported version for the *-pc-windows-* targets. These requirements apply both to the Rust toolchain itself and to binaries produced by Rust. Two new targets have been added with Windows 7 as their baseline: x86_64-win7-windows-msvc and i686-win7-windows-msvc. They are starting as Tier 3 targets, […]

Ver mais

arXiv:2603.16926v1 Announce Type: new Abstract: The Inaugural Music Source Restoration (MSR) Challenge targets the recovery of original, unprocessed stems from fully mixed and mastered music. Unlike conventional music source separation, MSR requires reversing complex production processes such as equalization, compression, reverberation, and other real-world degradations. To address MSR, we propose a two-stage system. First, an ensemble of pre-trained separation models produces preliminary source estimates. Then a set of pre-trained BSRNN-based restoration models performs targeted reconstruction to refine these […]

Ver mais

Meta has been poaching talent from Thinking Machines Lab. But it’s a two-way street.

Ver mais