Papers by Sai Qian Zhang
PipeSpec: Breaking Stage Dependencies in Hierarchical LLM Decoding (2025.findings-acl)
Copied to clipboard
| Challenge: | Speculative decoding is limited by sequential stage dependencies that prevent full hardware utilization. |
| Approach: | They propose a framework that generalizes speculative decoding to use multiple models arranged in a hierarchical pipeline and enables asynchronous execution with lightweight coordination for prediction verification and rollback. |
| Outcome: | The proposed framework achieves 2.25 tokens/unit through pipelined parallelism with multiple models arranged in a hierarchical pipeline. |
T3M: Text Guided 3D Human Motion Synthesis from Speech (2024.findings-naacl)
Copied to clipboard
| Challenge: | Existing methods for speech-driven 3D motion synthesisrely on speech audio . existing methods are inaccurate and inflexible, leading to inflexibility and inefficient synthesis results. |
| Approach: | They propose a text-guided 3D human motion synthesis method that uses text input to generate motions from human speech. |
| Outcome: | The proposed method outperforms existing methods in quantitative and qualitative evaluations. |
Mitigating Sequential Dependencies: A Survey of Algorithms and Systems for Generation-Refinement Frameworks in Autoregressive Models (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Sequential dependencies present a fundamental bottleneck in deploying large-scale autoregressive models . |
| Approach: | They analyze methods based on generation strategies and refinement mechanisms . they examine deployment strategies across computing environments and explore applications spanning text, images, and speech generation. |
| Outcome: | The proposed frameworks can be used to improve the quality of autoregressive models. |
LipoAgent: Coordinating Fine-Tuned LLM Agents for Safer Lipid Design (2026.findings-acl)
Copied to clipboard
| Challenge: | Lipid nanoparticles (LNPs) are among the most clinically mature platforms for nucleic acid delivery, yet designing lipids that are effective and biologically safe remains a major bottleneck. |
| Approach: | They propose a safety-aware multi-agent LLM framework for lipid discovery that enforces toxicity as a prerequisite for efficiency prediction. |
| Outcome: | The proposed framework achieves an average improvement in mRNA transfection efficiency prediction across multiple foundation models. |
DLoRA: Distributed Parameter-Efficient Fine-Tuning Solution for Large Language Model (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Large language models (LLMs) can handle tasks such as text generation, translation and summarization with remarkable proficiency. |
| Approach: | They propose a framework that enables scalable PEFT operations to be performed collaboratively between the cloud and user devices. |
| Outcome: | The proposed framework reduces computation and communication workload over user devices while achieving superior accuracy and privacy protection. |
LARA: LLM-based Agile Power Distribution Network Restoration from Disastrous Events (2026.findings-eacl)
Copied to clipboard
| Challenge: | a large language model generates high-level restoration plans over a compact catalogue of feasible actions. |
| Approach: | They propose a method that generates restoration plans over a catalogue of feasible actions. |
| Outcome: | The proposed model outperforms a time-capped solver on an IEEE 13-node power distribution feeder by 13% while using less than 1% of its wall-clock runtime. |
DREAM-S: Speculative Decoding with Searchable Drafting and Target-Aware Refinement for Multimodal Generation (2026.acl-long)
Copied to clipboard
| Challenge: | Speculative decoding (SD) has proven to be effective for autoregressive generation in large language models (LLMs), however its application to vision-language models (VLMs) remains relatively unexplored. |
| Approach: | They propose a Speculative Decoding framework for vision-language models that integrates a neural architecture search framework and target-aware supernet training to identify optimal interaction strategies. |
| Outcome: | DREAM-S achieves 3.85 speedup compared to baselines on well-established vision-language models. |