Search for: All records

In pretraining data detection, the goal is to detect whether a given sentence is in the dataset used for training a Large Language Model LLM). Recent methods (such as Min-K % and Min-K%++) reveal that most training corpora are likely contaminated with both sensitive content and evaluation benchmarks, leading to inflated test set performance. These methods sometimes fail to detect samples from the pretraining data, primarily because they depend on statistics composed of causal token likelihoods. We introduce Infilling Score, a new test-statistic based on non-causal token likelihoods. Infilling Score can be computed for autoregressive models without re-training using Bayes rule. A naive application of Bayes rule scales linearly with the vocabulary size. However, we propose a ratio test-statistic whose computation is invariant to vocabulary size. Empirically, our method achieves a significant accuracy gain over state-of-the-art methods including Min-K%, and Min-K%++ on the WikiMIA benchmark across seven models with different parameter sizes. Further, we achieve higher AUC compared to reference-free methods on the challenging MIMIR benchmark. Finally, we create a benchmark dataset consisting of recent data sources published after the release of Llama-3; this benchmark provides a statistical baseline to indicate potential corpora used for Llama-3 training. 

This work presents SPARC (Spatio-Temporal Adaptive Resource Control), a novel approach for multi-site spectrum management in NextG cellular networks. SPARC addresses the challenge of limited licensed spectrum in dynamic environments. We leverage the O-RAN architecture to develop a multi-timescale RAN Intelligent Controller (RIC) framework, featuring an xApp for near-real-time interference detection and localization, and a MApp for real-time intelligent resource allocation. By utilizing base stations as spectrum sensors, SPARC enables efficient and fine-grained dynamic resource allocation across multiple sites, enhancing signal-to-noise ratio (SNR) by up to 7dB, spectral efficiency by up to 15%, and overall system throughput by up to 20%. Comprehensive evaluations, including emulations and over-the-air experiments, demonstrate the significant performance gains achieved through SPARC, showcasing it as a promising solution for optimizing resource efficiency and network performance in NextG cellular networks. 

An increasingly popular machine learning paradigm is to pretrain a neural network (NN) on many tasks offline, then adapt it to downstream tasks, often by re-training only the last linear layer of the network. This approach yields strong downstream performance in a variety of contexts, demonstrating that multitask pretraining leads to effective feature learning. Although several recent theoretical studies have shown that shallow NNs learn meaningful features when either (i) they are trained on a single task or (ii) they are linear, very little is known about the closer-to-practice case of nonlinear NNs trained on multiple tasks. In this work, we present the first results proving that feature learning occurs during training with a nonlinear model on multiple tasks. Our key insight is that multi-task pretraining induces a pseudo-contrastive loss that favors representations that align points that typically have the same label across tasks. Using this observation, we show that when the tasks are binary classification tasks with labels depending on the projection of the data onto an 𝑟-dimensional subspace within the 𝑑 ≫𝑟-dimensional input space, a simple gradient-based multitask learning algorithm on a two-layer ReLU NN recovers this projection, allowing for generalization to downstream tasks with sample and neuron complexity independent of 𝑑. In contrast, we show that with high probability over the draw of a single task, training on this single task cannot guarantee to learn all 𝑟 ground-truth features. 

A striking property of transformers is their ability to perform in-context learning (ICL), a machine learning framework in which the learner is presented with a novel context during inference implicitly through some data, and tasked with making a prediction in that context. As such, that learner must adapt to the context without additional training. We explore the role of softmax attention in an ICL setting where each context encodes a regression task. We show that an attention unit learns a window that it uses to implement a nearest-neighbors predictor adapted to the landscape of the pretraining tasks. Specifically, we show that this window widens with decreasing Lipschitzness and increasing label noise in the pretraining tasks. We also show that on low-rank, linear problems, the attention unit learns to project onto the appropriate subspace before inference. Further, we show that this adaptivity relies crucially on the softmax activation and thus cannot be replicated by the linear activation often studied in prior theoretical analyses. 

Current wireless networks employ sophisticated multi-user transmission techniques to fully utilize the physical layer resources for data transmission. At the MAC layer, these techniques rely on a semi-static map that translates the channel quality of users to the potential transmission rate (more precisely, a map from the Channel Quality Index to the Modulation and Coding Scheme) for user selection and scheduling decisions. However, such a static map does not adapt to the actual deployment scenario and can lead to large performance losses. Furthermore, adaptively learning this map can be inefficient, particularly when there are a large number of users. In this work, we make this learning efficient by clustering users. Specifically, we develop an online learning approach that jointly clusters users and channel-states, and learns the associated rate regions of each cluster. This approach generates a scenario-specific map that replaces the static map that is currently used in practice. Furthermore, we show that our learning algorithm achieves sub- linear regret when compared to an omniscient genie. Next, we develop a user selection algorithm for multi-user scheduling using the learned user-clusters and associated rate regions. Our algorithms are validated on the WiNGS simulator from AT&T Labs, that implements the PHY/MAC stack and simulates the channel. We show that our algorithm can efficiently learn user clusters and the rate regions associated with the user sets for any observed channel state. Moreover, our simulations show that a deployment-scenario-specific map significantly outperforms the current static map approach for resource allocation at the MAC layer.