arXiv subjects

Yibo Zhong

Publications and source records attributed to Yibo Zhong.

8 recordsLinked to original sources

arXiv2026

SpikingBrain2.0: Brain-Inspired Foundation Models for Efficient Long-Context and Cross-Platform Inference

Scaling context length is reshaping large-model development, yet full-attention Transformers suffer from prohibitive computation and inference bottlenecks at long sequences. A key challenge is to design foundation models that maintain performance and long-context efficiency with minimal training overhead. We introduce SpikingBrain2.0 (SpB2.0), a 5B model that advances both architecture and training efficiency of its predecessor. Our contributions are two-fold. (1) Architectural Innovation: We propose Dual-Space Sparse Attention (DSSA), an inter-layer hybrid of Sparse Softmax Attention (MoBA) and Sparse Linear Attention (SSE), achieving an improved performance-efficiency trade-off for long-context modeling. SpB2.0 further supports dual quantization paths: INT8-Spiking coding enables sparse event-driven computation, while FP8 coding accelerates inference on modern GPUs. (2) Enhanced Training Strategy: We develop an optimized Transformer-to-Hybrid (T2H) pipeline with dual conversion paths for LLMs and VLMs using curated open-source data. Empirically, SpB2.0-5B and SpB2.0-VL-5B recover most of the base Transformer (Qwen3-4B) capability with under 7k A100 GPU hours. SpB2.0 achieves a 10.13x TTFT speedup at 4M context and supports over 10M tokens on 8 A100 GPUs under vLLM, where full-attention models exceed memory limits. It also demonstrates strong cross-platform compatibility, enabling FP8 GPU inference (2.52x speedup at 250k) and efficient neuromorphic execution (64.31% sparsity, with 70.6% and 46.5% area and power reduction at 500MHz). Overall, SpikingBrain2.0 provides a practical pathway for lightweight, multimodal, spiking foundation models, highlighting the potential of combining brain-inspired mechanisms with efficient architectures for resource-constrained and edge scenarios.

cs.LG↗

arXiv2025

PEANuT: Parameter-Efficient Adaptation with Weight-aware Neural Tweakers

Fine-tuning large pre-trained foundation models often yields excellent downstream performance but is prohibitively expensive when updating all parameters. Parameter-efficient fine-tuning (PEFT) methods such as LoRA alleviate this by introducing lightweight update modules, yet they commonly rely on weight-agnostic linear approximations, limiting their expressiveness. In this work, we propose PEANuT, a novel PEFT framework that introduces weight-aware neural tweakers, compact neural modules that generate task-adaptive updates conditioned on frozen pre-trained weights. PEANuT provides a flexible yet efficient way to capture complex update patterns without full model tuning. We theoretically show that PEANuT achieves equivalent or greater expressivity than existing linear PEFT methods with comparable or fewer parameters. Extensive experiments across four benchmarks with over twenty datasets demonstrate that PEANuT consistently outperforms strong baselines in both NLP and vision tasks, while maintaining low computational overhead.

cs.LG↗

arXiv2025

ENA: Efficient N-dimensional Attention

Efficient modeling of long sequences of high-order data requires a more efficient architecture than Transformer. In this paper, we investigate two key aspects of extending linear recurrent models, especially those originally designed for language modeling, to high-order data (1D to ND): scanning strategies and attention-hybrid architectures. Empirical results suggest that scanning provides limited benefits, while attention-hybrid models yield promising results. Focusing on the latter, we further evaluate types of attention and find that tiled high-order sliding window attention (SWA) is efficient in both theory and practice. We term the resulting hybrid architecture of linear recurrence and high-order SWA as Efficient N-dimensional Attention (ENA). We then conduct several experiments to demonstrate its effectiveness. The intuition behind ENA is that linear recurrence compresses global information into a state, while SWA complements it by enforcing strict local modeling. Together, they form a simple framework that offers a promising and practical solution for ultra-long high-order data modeling.

cs.LG↗

arXiv2025

Low-Rank Interconnected Adaptation across Layers

Low-rank adaptation (LoRA) is a widely used parameter-efficient fine-tuning (PEFT) method that learns weight updates $ΔW = AB$ for pretrained weights $W$ through low-rank adapters $A$ and $B$. While LoRA ensures hardware efficiency, its low-rank weight updates limit adaptation performance. In this paper, we propose low-rank interconnected adaptation across layers (Lily), a novel PEFT method that introduces an interconnected framework with locally shared $A$ and globally shared $B$ experts. This structure eliminates redundant per-layer $AB$ pairs, enabling higher-rank $ΔW$ with equal or fewer parameters. To enhance expressiveness, we use data-dependent routers to determine $A$-$B$ interconnections, preventing $B$ experts from converging to the same behavior and improving representational power across domains. Experiments across modalities, architectures, and model sizes demonstrate Lily's superior performance and efficiency. GitHub: https://github.com/yibozhong/lily

cs.CV↗

arXiv2025

UORA: Uniform Orthogonal Reinitialization Adaptation in Parameter-Efficient Fine-Tuning of Large Models

This paper introduces Uniform Orthogonal Reinitialization Adaptation (UORA), a novel parameter-efficient fine-tuning (PEFT) approach for Large Language Models (LLMs). UORA achieves state-of-the-art performance and parameter efficiency by leveraging a low-rank approximation method to reduce the number of trainable parameters. Unlike existing methods such as LoRA and VeRA, UORA employs an interpolation-based reparametrization mechanism that selectively reinitializes rows and columns in frozen projection matrices, guided by the vector magnitude heuristic. This results in substantially fewer trainable parameters compared to LoRA and outperforms VeRA in computation and storage efficiency. Comprehensive experiments across various benchmarks demonstrate UORA's superiority in achieving competitive fine-tuning performance with negligible computational overhead. We demonstrate its performance on GLUE and E2E benchmarks and its effectiveness in instruction-tuning large language models and image classification models. Our contributions establish a new paradigm for scalable and resource-efficient fine-tuning of LLMs.

cs.CL↗

arXiv2024

Rethinking Low-Rank Adaptation in Vision: Exploring Head-Level Responsiveness across Diverse Tasks

Low-rank adaptation (LoRA) has shifted the paradigm of adapting pre-trained Vision Transformers (ViT), achieving great efficiency by updating only a subset of tailored parameters to approximate weight updates. However, the multi-head design of the self-attention mechanism, with the heads working in parallel in the computation flow, exhibiting similar visual patterns and requiring update over all of them, incurs unnecessary storage and computational overhead. In this paper, we propose Head-level responsiveness tuning for low-rank adaptation (Heart-LoRA). The proposed method explores redundancy among the heads and selectively activates task-responsive heads, thus enabling fine-grained head-level tuning. Additionally, given the different responsiveness of heads to diverse visual tasks, our proposed method dynamically activates a subset of the approximated heads that are tailored to the current task. Experimental results show that Heart-LoRA yields superior performance over state-of-the-art PETL approaches on visual adaptation benchmark datasets.

cs.CV↗

arXiv2024

Pear: Pruning and Sharing Adapters in Visual Parameter-Efficient Fine-Tuning

Adapters have been widely explored to alleviate computational and storage costs when fine-tuning pretrained foundation models. However, the adapter itself can exhibit redundancy, leading to unnecessary storage overhead and inferior performance. In this paper, we propose Prune and Share (Pear), a novel adapter-pruning framework for efficient fine-tuning of pretrained visual foundation models. Specifically, we prune certain adapters and share the more important unpruned ones with positions where adapters are pruned, allowing continual adaptation at these positions after pruning. Additionally, a knowledge checkpoint strategy is introduced, which preserves the information of the pruned adapters and further boosts performance. Experimental results on visual adaptation benchmark validate the effectiveness and efficiency of the proposed Pear comparing to other competitive methods. Code is in https://github.com/yibozhong/pear.

cs.CV↗

arXiv2020

More of the Bulk from Extremal Area Variations

It was shown recently, building on work of Alexakis, Balehowksy, and Nachman that the geometry of (some portion of) a manifold with boundary is uniquely fixed by the areas of a foliation of two-dimensional disk-shaped surfaces anchored to the boundary. In the context of AdS/CFT, this implies that (a portion of) a four-dimensional bulk geometry can be fixed uniquely from the entanglement entropies of disk-shaped boundary regions, subject to several constraints. In this Note, we loosen some of these constraints, in particular allowing for the bulk foliation of extremal surfaces to be local and removing the constraint of disk topology; these generalizations ensure uniqueness of more of the deep bulk geometry by allowing for e.g. surfaces anchored on disconnected asymptotic boundaries, or HRT surfaces past a phase transition. We also explore in more depth the generality of the local foliation requirement, showing that even in a highly dynamical geometry like AdS-Vaidya it is satisfied.

hep-th↗