Jianyu Zhang

Jiao Zhang, Xu-Yao Zhang, Cheng-Lin Liu

There are ubiquitous distribution shifts in the real world. However, deep neural networks (DNNs) are easily biased towards the training set, which causes severe performance degradation when they receive out-of-distribution data. Many methods are studied to train models that generalize under various distribution shifts in the literature of domain generalization (DG). However, the recent DomainBed and WILDS benchmarks challenged the effectiveness of these methods. Aiming at the problems in the existing research, we propose a new domain generalization task for handwritten Chinese character recognition (HCCR) to enrich the application scenarios of DG method research. We evaluate eighteen DG methods on the proposed PaHCC (Printed and Handwritten Chinese Characters) dataset and show that the performance of existing methods on this dataset is still unsatisfactory. Besides, under a designed dynamic DG setting, we reveal more properties of DG methods and argue that only the leave-one-domain-out protocol is unreliable. We advocate that researchers in the DG community refer to dynamic performance of methods for more comprehensive and reliable evaluation. Our dataset and evaluations bring new perspectives to the community for more substantial progress. We will make our dataset public with the article published to facilitate the study of domain generalization.

Zhuoming Chen, Ranajoy Sadhukhan, Zihao Ye et al. · uw

Large language models (LLMs) with long context windows have gained significant attention. However, the KV cache, stored to avoid re-computation, becomes a bottleneck. Various dynamic sparse or TopK-based attention approximation methods have been proposed to leverage the common insight that attention is sparse. In this paper, we first show that TopK attention itself suffers from quality degradation in certain downstream tasks because attention is not always as sparse as expected. Rather than selecting the keys and values with the highest attention scores, sampling with theoretical guarantees can provide a better estimation for attention output. To make the sampling-based approximation practical in LLM generation, we propose MagicPIG, a heterogeneous system based on Locality Sensitive Hashing (LSH). MagicPIG significantly reduces the workload of attention computation while preserving high accuracy for diverse tasks. MagicPIG stores the LSH hash tables and runs the attention computation on the CPU, which allows it to serve longer contexts and larger batch sizes with high approximation accuracy. MagicPIG can improve decoding throughput by up to $5\times$ across various GPU hardware and achieve 54ms decoding latency on a single RTX 4090 for Llama-3.1-8B-Instruct model with a context of 96k tokens. The code is available at https://github.com/Infini-AI-Lab/MagicPIG.

Jianyu Zhang, Niklas Nolte, Ranajoy Sadhukhan et al.

Memory Mosaics are networks of associative memories working in concert to achieve a prediction task of interest. Like transformers, memory mosaics possess compositional capabilities and in-context learning capabilities. Unlike transformers, memory mosaics achieve these capabilities in comparatively transparent way ("predictive disentanglement"). We illustrate these capabilities on a toy example and also show that memory mosaics perform as well or better than transformers on medium-scale language modeling tasks.

Jingtong Su, Jianyu Zhang, Karen Ullrich et al.

Common Large Language model (LLM) evaluations rely on demonstration examples to steer models' responses to the desired style. While the number of examples used has been studied and standardized, the choice of how to format examples is less investigated. In evaluation protocols and real world usage, users face the choice how to separate in-context examples: use a comma? new line? semi-colon? hashtag? etc.? Surprisingly, we find this seemingly minor choice can dramatically alter model response quality. Across leading model families (Llama, Qwen, Gemma), performance on MMLU for example can vary by $\pm 23\%$ depending on the choice of delimiter. In fact, one can manipulate model rankings to put any model in the lead by only modifying the single character separating examples. We find LLMs' brittleness pervades topics, model families, and doesn't improve with scale. By probing attention head scores, we find that good-performing delimiters steer attention towards key tokens in the input. Finally, we explore methods to improve LLMs' robustness to the choice of delimiter. We find specifying the selected delimiter in the prompt boosts robustness and offer practical recommendations for the best-performing delimiters to select.

Xinyu Li, Yanyi Zhang, Jianyu Zhang et al.

Process modeling and understanding are fundamental for advanced human-computer interfaces and automation systems. Most recent research has focused on activity recognition, but little has been done on sensor-based detection of process progress. We introduce a real-time, sensor-based system for modeling, recognizing and estimating the progress of a work process. We implemented a multimodal deep learning structure to extract the relevant spatio-temporal features from multiple sensory inputs and used a novel deep regression structure for overall completeness estimation. Using process completeness estimation with a Gaussian mixture model, our system can predict the phase for sequential processes. The performance speed, calculated using completeness estimation, allows online estimation of the remaining time. To train our system, we introduced a novel rectified hyperbolic tangent (rtanh) activation function and conditional loss. Our system was tested on data obtained from the medical process (trauma resuscitation) and sports events (Olympic swimming competition). Our system outperformed the existing trauma-resuscitation phase detectors with a phase detection accuracy of over 86%, an F1-score of 0.67, a completeness estimation error of under 12.6%, and a remaining-time estimation error of less than 7.5 minutes. For the Olympic swimming dataset, our system achieved an accuracy of 88%, an F1-score of 0.58, a completeness estimation error of 6.3% and a remaining-time estimation error of 2.9 minutes.