Zhefeng Wang

Large Language Models (LLMs) for Generative AI have achieved remarkable progress, evolving into sophisticated and versatile tools widely adopted across various domains and applications. However, the substantial memory overhead caused by their vast number of parameters, combined with the high computational demands of the attention mechanism, poses significant challenges in achieving low latency and high throughput for LLM inference services. Recent advancements, driven by groundbreaking research, have significantly accelerated progress in this field. This paper provides a comprehensive survey of these methods, covering fundamental instance-level approaches, in-depth cluster-level strategies, emerging scenario directions, and other miscellaneous but important areas. At the instance level, we review model placement, request scheduling, decoding length prediction, storage management, and the disaggregation paradigm. At the cluster level, we explore GPU cluster deployment, multi-instance load balancing, and cloud service solutions. For emerging scenarios, we organize the discussion around specific tasks, modules, and auxiliary methods. To ensure a holistic overview, we also highlight several niche yet critical areas. Finally, we outline potential research directions to further advance the field of LLM inference serving.

The rapid increase in the size of large language models (LLMs) has significantly escalated their computational and memory demands, posing challenges for efficient deployment, especially on resource-constrained devices. Structured pruning has emerged as an effective model compression method that can reduce these demands while preserving performance. In this paper, we introduce FASP (Fast and Accurate Structured Pruning), a novel structured pruning framework for LLMs that emphasizes both speed and accuracy. FASP employs a distinctive pruning structure that interlinks sequential layers, allowing for the removal of columns in one layer while simultaneously eliminating corresponding rows in the preceding layer without incurring additional performance loss. The pruning metric, inspired by Wanda, is computationally efficient and effectively selects components to prune. Additionally, we propose a restoration mechanism that enhances model fidelity by adjusting the remaining weights post-pruning. We evaluate FASP on the OPT and LLaMA model families, demonstrating superior performance in terms of perplexity and accuracy on downstream tasks compared to state-of-the-art methods. Our approach achieves significant speed-ups, pruning models such as OPT-125M in 17 seconds and LLaMA-30B in 15 minutes on a single NVIDIA RTX 4090 GPU, making it a highly practical solution for optimizing LLMs.

Neural networks often tend to rely on bias features that have strong but spurious correlations with the target labels for decision-making, leading to poor performance on data that does not adhere to these correlations. Early debiasing methods typically construct an unbiased optimization objective based on the labels of bias features. Recent work assumes that bias label is unavailable and usually trains two models: a biased model to deliberately learn bias features for exposing data bias, and a target model to eliminate bias captured by the bias model. In this paper, we first reveal that previous biased models fit target labels, which resulted in failing to expose data bias. To tackle this issue, we propose poisoner, which utilizes data poisoning to embed the biases learned by biased models into the poisoned training data, thereby encouraging the models to learn more biases. Specifically, we couple data poisoning and model training to continuously prompt the biased model to learn more bias. By utilizing the biased model, we can identify samples in the data that contradict these biased correlations. Subsequently, we amplify the influence of these samples in the training of the target model to prevent the model from learning such biased correlations. Experiments show the superior debiasing performance of our method.

We introduce an open source high-quality Mandarin TTS dataset MSceneSpeech (Multiple Scene Speech Dataset), which is intended to provide resources for expressive speech synthesis. MSceneSpeech comprises numerous audio recordings and texts performed and recorded according to daily life scenarios. Each scenario includes multiple speakers and a diverse range of prosodic styles, making it suitable for speech synthesis that entails multi-speaker style and prosody modeling. We have established a robust baseline, through the prompting mechanism, that can effectively synthesize speech characterized by both user-specific timbre and scene-specific prosody with arbitrary text input. The open source MSceneSpeech Dataset and audio samples of our baseline are available at https://speechai-demo.github.io/MSceneSpeech/.

The topic of multimodal conversation systems has recently garnered significant attention across various industries, including travel and retail, among others. While pioneering works in this field have shown promising performance, they often focus solely on context information at the utterance level, overlooking the context-aware dependencies of multimodal semantic elements like words and images. Furthermore, the ordinal information of images, which indicates the relevance between visual context and users’ demands, remains underutilized during the integration of visual content. Additionally, the exploration of how to effectively utilize corresponding attributes provided by users when searching for desired products is still largely unexplored. To address these challenges, we propose PMATE, a P osition-aware M ultimodal di A logue system with seman T ic E lements. Specifically, to obtain semantic representations at the element level, we first unfold the multimodal historical utterances and devise a position-aware multimodal element-level encoder. This component considers all images that may be relevant to the current turn and introduces a novel position-aware image selector to choose related images before fusing the information from the two modalities. Finally, we present a knowledge-aware two-stage decoder and an attribute-enhanced image searcher for the tasks of generating textual responses and selecting image responses, respectively. We extensively evaluate our model on two large-scale multimodal dialogue datasets, and the results of our experiments demonstrate that our approach outperforms several baseline methods.

While recent advancements in large language models (LLMs) bring us closer to achieving artificial general intelligence, the question persists: Do LLMs truly understand language, or do they merely mimic comprehension through pattern recognition? This study seeks to explore this question through the lens of syntax, a crucial component of sentence comprehension. Adopting a natural language question-answering (Q&A) scheme, we craft questions targeting nine syntactic knowledge points that are most closely related to sentence comprehension. Experiments conducted on 24 LLMs suggest that most have a limited grasp of syntactic knowledge, exhibiting notable discrepancies across different syntactic knowledge points. In particular, questions involving prepositional phrase attachment pose the greatest challenge, whereas those concerning adjectival modifier and indirect object are relatively easier for LLMs to handle. Furthermore, a case study on the training dynamics of the LLMs reveals that the majority of syntactic knowledge is learned during the initial stages of training, hinting that simply increasing the number of training tokens may not be the `silver bullet' for improving the comprehension ability of LLMs.

Inspired by early research on exploring naturally annotated data for Chinese word segmentation (CWS), and also by recent research on integration of speech and text processing, this work for the first time proposes to mine word boundaries from parallel speech/text data. First we collect parallel speech/text data from two Internet sources that are related with CWS data used in our experiments. Then, we obtain character-level alignments and design simple heuristic rules for determining word boundaries according to pause duration between adjacent characters. Finally, we present an effective complete-then-train strategy that can better utilize extra naturally annotated data for model training. Experiments demonstrate our approach can significantly boost CWS performance in both cross-domain and low-resource scenarios.