Celty: SpMspV GPU Kernel and SIMT Co-Design for Efficient Dual-Sparse LLM Inference

Large Language Models (LLMs) increasingly rely on sparsity to reduce inference cost, but most prior work targets a single sparsity source-either weight or activation-and optimizes for batched multi-user inference. Dual-sparsity, which combines unstructured weight pruning with runtime activation sparsity, offers a compelling tradeoff among model size, accuracy, and latency for single-user decoding, but formulates as a Sparse Matrix-Sparse Vector (spMspV) workload that existing GPU kernels handle poorly. We propose Celty, a co-designed sparse format, GPU kernel, and SIMT microarchitecture for efficient spMspV in LLM inference. At the kernel level, Celty introduces a Run-Length Compressed CSC (RLC-CSC) format that enables vectorized loading of compressed weight columns and exploits both sparsity sources to skip unnecessary memory accesses, with shared memory used for scattered partial-product accumulation. At the microarchitecture level, the Celty Sparse SIMT Core integrates a pipelined RLC decoder to eliminate software-level index reconstruction and repurposes local register files for conflict-free accumulation-operating directly on the same RLC-CSC format without data layout changes. The Celty GPU kernel achieves up to 2.8x speedup over cuBLAS and 2.4x over Flash-LLM. With the Sparse SIMT Core, speedups reach up to 5.3x over cuBLAS at 70% dual-sparsity.

BRIM: Workload-Balanced Dual-Sided Bit-Serial Sparse Inference Accelerator

Bit-serial accelerators exploit bit-level sparsity to reduce DNN inference cost, but existing designs exploit sparsity on only one operand, bounding the speedup. Extending sparsity exploitation to both operands simultaneously yields compounding reductions in partial products but introduces a critical new bottleneck: workload imbalance. Because each concurrent weight - activation pair's execution cost depends on the product of two independently varying operand non-zero bit counts, pairs that must complete together finish at vastly different times, leaving faster computations idle. We show this limits PE utilization to 56 - 64% in existing dual-sided designs. We present BRIM, a hardware - software co-designed dual-sided bit-serial sparse accelerator that directly targets this bottleneck. BRIM combines two integrated mechanisms: 1) Cyclic-Balanced Pruning (CBP), a post-training weight optimization that reshapes weight representations based on profiled activation statistics to equalize expected workloads across concurrently processed pairs offline; and 2) Pairwise Slot Donation, a lightweight hardware mechanism that absorbs residual runtime imbalance with negligible area overhead. Evaluated across CNNs, ViTs, and LLMs under iso-area constraints, BRIM achieves over 90% PE utilization, up to 2.37x speedup, and up to 1.63x energy efficiency improvement over prior dual-sided designs.

KronQ: LLM Quantization via Kronecker-Factored Hessian

Post-training quantization (PTQ) is a widely adopted technique for compressing large language models (LLMs) without retraining. Existing second-order PTQ methods, including GPTQ, construct quantization objectives exclusively from input activation statistics, effectively assuming that all output channels contribute equally to the layer-wise reconstruction objective. We propose KronQ, a PTQ framework that challenges this assumption by introducing the gradient covariance into the quantization pipeline. Under the Kronecker-factored Hessian approximation, the quantization loss depends jointly on both the activation and gradient covariances, and KronQ exploits this at two complementary levels. (1) KronQ introduces bidirectional incoherence processing, extending the existing input-side random rotation to the output dimension using the gradient covariance, reducing weight magnitude variance across both input and output dimensions. (2) KronQ derives a new sensitivity metric for inter-layer mixed-precision allocation, driven by the gradient and activation Hessian traces. Notably, in the case of 2-bit weight-only quantization on LLaMA-3-70B, while GPTQ and GPTAQ diverge or produce degenerate quantizations (>2000 perplexity on WikiText-2), KronQ achieves 7.93 perplexity.

OrbitQuant: Data-Agnostic Quantization for Image and Video Diffusion Transformers

Diffusion transformers (DiTs) achieve state-of-the-art image and video generation, but their multi-step sampling and growing parameter count make inference expensive. Post-training quantization (PTQ) is the natural remedy, yet DiT activations shift across timesteps, prompts, and guidance branches, forcing prior methods to re-fit calibration data for every new checkpoint or modality. We present OrbitQuant, a data-agnostic weight-activation quantizer that bypasses range estimation by quantizing in a normalized, rotated basis. In this basis, a randomized permuted block-Hadamard (RPBH) rotation concentrates each coordinate around one fixed, known marginal regardless of the input, so a single Lloyd-Max codebook serves all timesteps, prompts, and layers of a given input dimension. We extend the same quantizer to weight rows offline, absorbing the rotation into the weights so that it cancels inside each linear layer and only a forward rotation on the activations remains at runtime. The same recipe transfers from image to video with no per-modality tuning. Across FLUX.1, Z-Image-Turbo, Wan 2.1, and CogVideoX, it sets the state of the art for PTQ at several low-bit settings. It also pushes PTQ of image diffusion transformers to W2A4 with usable generation quality.

Optimal Brain Decomposition for Accurate LLM Low-Rank Approximation

Low-rank decomposition has emerged as an important problem in Large Language Model (LLM) fine-tuning and inference. Through Singular Value Decomposition (SVD), the weight matrix can be factorized into low-rank spaces optimally. Previously, a common practice was to decompose the weight in the activation-whitened space, and then achieve satisfying results. In this work, we propose Optimal Brain Decomposition LLM (OBD-LLM), which studies the decomposition problem in the model space by utilizing second-order Hessian information. Through a rigorous Kronecker-factorization of the Hessian, we show that the decomposition needs to consider both input and output information of the layer, and achieves much better decomposition results compared to input only method. Our loss-aware decomposition method involves a bi-directional whitening on the weight matrix. As a result, OBD-LLM is a closed-form solution for the optimal decomposition of weights in the language model. Remarkably, we achieve ~20-40\% better results than previous state-of-the-art decomposition methods, the SVD-LLM.

QuRL: Efficient Reinforcement Learning with Quantized Rollout

Reinforcement learning with verifiable rewards (RLVR) has become a trending paradigm for training reasoning large language models (LLMs). However, due to the autoregressive decoding nature of LLMs, the rollout process becomes the efficiency bottleneck of RL training, consisting of up to 70\% of the total training time. In this work, we propose Quantized Reinforcement Learning (QuRL) that uses a quantized actor for accelerating the rollout. We address two challenges in QuRL. First, we propose Adaptive Clipping Range (ACR) that dynamically adjusts the clipping ratio based on the policy ratio between the full-precision actor and the quantized actor, which is essential for mitigating long-term training collapse. Second, we identify the weight update problem, where weight changes between RL steps are extremely small, making it difficult for the quantization operation to capture them effectively. We mitigate this problem through the invariant scaling technique that reduces quantization noise and increases weight update. We evaluate our method with INT8 and FP8 quantization experiments on DeepScaleR and DAPO, and achieve 20% to 80% faster rollout during training.

Fast and Cost-effective Speculative Edge-Cloud Decoding with Early Exits

Large Language Models (LLMs) enable various applications on edge devices such as smartphones, wearables, and embodied robots. However, their deployment often depends on expensive cloud-based APIs, creating high operational costs, which limit access for smaller organizations and raise sustainability concerns. Certain LLMs can be deployed on-device, offering a cost-effective solution with reduced latency and improved privacy. Yet, limited computing resources constrain the size and accuracy of models that can be deployed, necessitating a collaborative design between edge and cloud. We propose a fast and cost-effective speculative edge-cloud decoding framework with a large target model on the server and a small draft model on the device. By introducing early exits in the target model, tokens are generated mid-verification, allowing the client to preemptively draft subsequent tokens before final verification, thus utilizing idle time and enhancing parallelism between edge and cloud. Using an NVIDIA Jetson Nano (client) and an A100 GPU (server) with Vicuna-68M (draft) and Llama2-7B (target) models, our method achieves up to a 35% reduction in latency compared to cloud-based autoregressive decoding, with an additional 11% improvement from preemptive drafting. To demonstrate real-world applicability, we deploy our method on the Unitree Go2 quadruped robot using Vision-Language Model (VLM) based control, achieving a 21% speedup over traditional cloud-based autoregressive decoding. These results demonstrate the potential of our framework for real-time LLM and VLM applications on resource-constrained edge devices.

Event2Vec: Processing neuromorphic events directly by representations in vector space

The neuromorphic event cameras have overwhelming advantages in temporal resolution, power efficiency, and dynamic range compared to traditional cameras. However, the event cameras output asynchronous, sparse, and irregular events, which are not compatible with mainstream computer vision and deep learning methods. Various methods have been proposed to solve this issue but at the cost of long preprocessing procedures, losing temporal resolutions, or being incompatible with massively parallel computation. Inspired by the great success of the word to vector, we summarize the similarities between words and events, then propose the first event to vector (event2vec) representation. We validate event2vec on classifying the ASL-DVS dataset, showing impressive parameter efficiency, accuracy, and speed than previous graph/image/voxel-based representations. Beyond task performance, the most attractive advantage of event2vec is that it aligns events to the domain of natural language processing, showing the promising prospect of integrating events into large language and multimodal models. Our codes, models, and training logs are available at https://github.com/fangwei123456/event2vec.

GPTQv2: Efficient Finetuning-Free Quantization for Asymmetric Calibration

We introduce GPTQv2, a novel finetuning-free quantization method for compressing large-scale transformer architectures. Unlike the previous GPTQ method, which independently calibrates each layer, we always match the quantized layer's output to the exact output in the full-precision model, resulting in a scheme that we call asymmetric calibration. Such a scheme can effectively reduce the quantization error accumulated in previous layers. We analyze this problem using optimal brain compression to derive a close-formed solution. The new solution explicitly minimizes the quantization error as well as the accumulated asymmetry error. Furthermore, we utilize various techniques to parallelize the solution calculation, including channel parallelization, neuron decomposition, and Cholesky reformulation for matrix fusion. As a result, GPTQv2 is easy to implement, simply using 20 more lines of code than GPTQ but improving its performance under low-bit quantization. Remarkably, on a single GPU, we quantize a 405B language transformer as well as EVA-02 the rank first vision transformer that achieves 90% pretraining Imagenet accuracy. Code is available at github.com/Intelligent-Computing-Lab-Yale/GPTQv2.

Low-power Spike-based Wearable Analytics on RRAM Crossbars

This work introduces a spike-based wearable analytics system utilizing Spiking Neural Networks (SNNs) deployed on an In-memory Computing engine based on RRAM crossbars, which are known for their compactness and energy-efficiency. Given the hardware constraints and noise characteristics of the underlying RRAM crossbars, we propose online adaptation of pre-trained SNNs in real-time using Direct Feedback Alignment (DFA) against traditional backpropagation (BP). Direct Feedback Alignment (DFA) learning, that allows layer-parallel gradient computations, acts as a fast, energy & area-efficient method for online adaptation of SNNs on RRAM crossbars, unleashing better algorithmic performance against those adapted using BP. Through extensive simulations using our in-house hardware evaluation engine called DFA_Sim, we find that DFA achieves upto 64.1% lower energy consumption, 10.1% lower area overhead, and a 2.1x reduction in latency compared to BP, while delivering upto 7.55% higher inference accuracy on human activity recognition (HAR) tasks.