ProPD: Dynamic Token Tree Pruning and Generation for LLM Parallel Decoding

Recent advancements in generative large language models (LLMs) have significantly boosted the performance in natural language processing tasks. However, their efficiency is hampered by the inherent limitations in autoregressive token generation. While parallel decoding with token tree verification, e.g., Medusa, has been proposed to improve decoding parallelism and efficiency, it often struggles with maintaining contextual relationships due to its independent token prediction approach and incurs significant verification overhead, especially with large tree sizes and batch processing. In this paper, we propose ProPD, an efficient LLM parallel decoding framework based on dynamic token tree pruning and generation. ProPD features an advanced early pruning mechanism to efficiently eliminate unpromising token sequences to improve verification efficiency. Additionally, it introduces a dynamic token tree generation algorithm to balance the computation and parallelism of the verification phase in real-time and maximize the overall efficiency across different batch sizes, sequence lengths, and tasks, etc. We verify ProPD across a diverse set of datasets, LLMs, and batch sizes and demonstrate ProPD consistently outperforms existing decoding algorithms by 1.1-3.2x.

AttentionLego: An Open-Source Building Block For Spatially-Scalable Large Language Model Accelerator With Processing-In-Memory Technology

Large language models (LLMs) with Transformer architectures have become phenomenal in natural language processing, multimodal generative artificial intelligence, and agent-oriented artificial intelligence. The self-attention module is the most dominating sub-structure inside Transformer-based LLMs. Computation using general-purpose graphics processing units (GPUs) inflicts reckless demand for I/O bandwidth for transferring intermediate calculation results between memories and processing units. To tackle this challenge, this work develops a fully customized vanilla self-attention accelerator, AttentionLego, as the basic building block for constructing spatially expandable LLM processors. AttentionLego provides basic implementation with fully-customized digital logic incorporating Processing-In-Memory (PIM) technology. It is based on PIM-based matrix-vector multiplication and look-up table-based Softmax design. The open-source code is available online: https://bonany.cc/attentionleg.

PiML Toolbox for Interpretable Machine Learning Model Development and Validation

PiML (read $\pi$-ML, /`pai.`em.`el/) is an integrated and open-access Python toolbox for interpretable machine learning model development and model diagnostics. It is designed with machine learning workflows in both low-code and high-code modes, including data pipeline, model training, model interpretation and explanation, and model diagnostics and comparison. The toolbox supports a growing list of interpretable models (e.g. GAM, GAMI-Net, XGB2) with inherent local and/or global interpretability. It also supports model-agnostic explainability tools (e.g. PFI, PDP, LIME, SHAP) and a powerful suite of model-agnostic diagnostics (e.g. weakness, uncertainty, robustness, fairness). Integration of PiML models and tests to existing MLOps platforms for quality assurance are enabled by flexible high-code APIs. Furthermore, PiML toolbox comes with a comprehensive user guide and hands-on examples, including the applications for model development and validation in banking. The project is available at https://github.com/SelfExplainML/PiML-Toolbox.

Explainable Recommendation Systems by Generalized Additive Models with Manifest and Latent Interactions

In recent years, the field of recommendation systems has attracted increasing attention to developing predictive models that provide explanations of why an item is recommended to a user. The explanations can be either obtained by post-hoc diagnostics after fitting a relatively complex model or embedded into an intrinsically interpretable model. In this paper, we propose the explainable recommendation systems based on a generalized additive model with manifest and latent interactions (GAMMLI). This model architecture is intrinsically interpretable, as it additively consists of the user and item main effects, the manifest user-item interactions based on observed features, and the latent interaction effects from residuals. Unlike conventional collaborative filtering methods, the group effect of users and items are considered in GAMMLI. It is beneficial for enhancing the model interpretability, and can also facilitate the cold-start recommendation problem. A new Python package GAMMLI is developed for efficient model training and visualized interpretation of the results. By numerical experiments based on simulation data and real-world cases, the proposed method is shown to have advantages in both predictive performance and explainable recommendation.

Unwrapping The Black Box of Deep ReLU Networks: Interpretability, Diagnostics, and Simplification

The deep neural networks (DNNs) have achieved great success in learning complex patterns with strong predictive power, but they are often thought of as "black box" models without a sufficient level of transparency and interpretability. It is important to demystify the DNNs with rigorous mathematics and practical tools, especially when they are used for mission-critical applications. This paper aims to unwrap the black box of deep ReLU networks through local linear representation, which utilizes the activation pattern and disentangles the complex network into an equivalent set of local linear models (LLMs). We develop a convenient LLM-based toolkit for interpretability, diagnostics, and simplification of a pre-trained deep ReLU network. We propose the local linear profile plot and other visualization methods for interpretation and diagnostics, and an effective merging strategy for network simplification. The proposed methods are demonstrated by simulation examples, benchmark datasets, and a real case study in home lending credit risk assessment.

Hyperparameter Optimization via Sequential Uniform Designs

Hyperparameter tuning or optimization plays a central role in the automated machine learning (AutoML) pipeline. It is a challenging task as the response surfaces of hyperparameters are generally unknown, and the evaluation of each experiment is expensive. In this paper, we reformulate hyperparameter optimization as a kind of computer experiment and propose a novel sequential uniform design (SeqUD) for hyperparameter optimization. It is advantageous as a) it adaptively explores the hyperparameter space with evenly spread design points, which is free of the expensive meta-modeling and acquisition optimization procedures in Bayesian optimization; b) sequential design points are generated in batch, which can be easily parallelized; and c) a real-time augmented uniform design (AugUD) algorithm is developed for the efficient generation of new design points. Experiments are conducted on both global optimization tasks and hyperparameter optimization applications. The results show that SeqUD outperforms related hyperparameter optimization methods, which is demonstrated to be a promising and competitive alternative of existing tools.

An Effective and Efficient Initialization Scheme for Multi-layer Feedforward Neural Networks

Network initialization is the first and critical step for training neural networks. In this paper, we propose a novel network initialization scheme based on the celebrated Stein's identity. By viewing multi-layer feedforward neural networks as cascades of multi-index models, the projection weights to the first hidden layer are initialized using eigenvectors of the cross-moment matrix between the input's second-order score function and the response. The input data is then forward propagated to the next layer and such a procedure can be repeated until all the hidden layers are initialized. Finally, the weights for the output layer are initialized by generalized linear modeling. Such a proposed SteinGLM method is shown through extensive numerical results to be much faster and more accurate than other popular methods commonly used for training neural networks.

GAMI-Net: An Explainable Neural Network based on Generalized Additive Models with Structured Interactions

The lack of interpretability is an inevitable problem when using neural network models in real applications. In this paper, a new explainable neural network called GAMI-Net, based on generalized additive models with structured interactions, is proposed to pursue a good balance between prediction accuracy and model interpretability. The GAMI-Net is a disentangled feedforward network with multiple additive subnetworks, where each subnetwork is designed for capturing either one main effect or one pairwise interaction effect. It takes into account three kinds of interpretability constraints, including a) sparsity constraint for selecting the most significant effects for parsimonious representations; b) heredity constraint such that a pairwise interaction could only be included when at least one of its parent effects exists; and c) marginal clarity constraint, in order to make the main and pairwise interaction effects mutually distinguishable. For model estimation, we develop an adaptive training algorithm that firstly fits the main effects to the responses, then fits the structured pairwise interactions to the residuals. Numerical experiments on both synthetic functions and real-world datasets show that the proposed explainable GAMI-Net enjoys superior interpretability while maintaining competitive prediction accuracy in comparison to the explainable boosting machine and other benchmark machine learning models.

Enhancing Explainability of Neural Networks through Architecture Constraints

Prediction accuracy and model explainability are the two most important objectives when developing machine learning algorithms to solve real-world problems. The neural networks are known to possess good prediction performance, but lack of sufficient model explainability. In this paper, we propose to enhance the explainability of neural networks through the following architecture constraints: a) sparse additive subnetworks; b) orthogonal projection pursuit; and c) smooth function approximation. It leads to a sparse, orthogonal and smooth explainable neural network (SOSxNN). The multiple parameters in the SOSxNN model are simultaneously estimated by a modified mini-batch gradient descent algorithm based on the backpropagation technique for calculating the derivatives and the Cayley transform for preserving the projection orthogonality. The hyperparameters controlling the sparse and smooth constraints are optimized by the grid search. Through simulation studies, we compare the SOSxNN method to several benchmark methods including least absolute shrinkage and selection operator, support vector machine, random forest, and multi-layer perceptron. It is shown that proposed model keeps the flexibility of pursuing prediction accuracy while attaining the improved interpretability, which can be therefore used as a promising surrogate model for complex model approximation. Finally, the real data example from the Lending Club is employed as a showcase of the SOSxNN application.