Large Language Models (LLMs) have ignited an innovative surge of AI applications, marking a new era of exciting possibilities equipped with extended context windows. However, hosting these models is cost-prohibitive mainly due to the extensive memory consumption of KV Cache involving long-context modeling. Despite several works proposing to evict unnecessary tokens from the KV Cache, most of them rely on the biased local statistics of accumulated attention scores and report performance using unconvincing metric like perplexity on inadequate short-text evaluation. In this paper, we propose NACL, a general framework for long-context KV cache eviction that achieves more optimal and efficient eviction in a single operation during the encoding phase. Due to NACL’s efficiency, we combine more accurate attention score statistics in Proxy-Tokens Eviction with the diversified random eviction strategy of Random Eviction, aiming to alleviate the issue of attention bias and enhance the robustness in maintaining pivotal tokens for long-context modeling tasks. Notably, our method significantly improves the performance on short- and long-text tasks by 80% and 76% respectively, reducing KV Cache by up to 5× with over 95% performance maintenance. Code available at https://github.com/PaddlePaddle/Research/tree/master/NLP/ACL2024-NACL.

PaddleSpeech is an open-source all-in-one speech toolkit. It aims at facilitating the development and research of speech processing technologies by providing an easy-to-use command-line interface and a simple code structure. This paper describes the design philosophy and core architecture of PaddleSpeech to support several essential speech-to-text and text-to-speech tasks. PaddleSpeech achieves competitive or state-of-the-art performance on various speech datasets and implements the most popular methods. It also provides recipes and pretrained models to quickly reproduce the experimental results in this paper. PaddleSpeech is publicly avaiable at https://github.com/PaddlePaddle/PaddleSpeech.

Human generates responses relying on semantic and functional dependencies, including coreference relation, among dialogue elements and their context. In this paper, we investigate matching a response with its multi-turn context using dependency information based entirely on attention. Our solution is inspired by the recently proposed Transformer in machine translation (Vaswani et al., 2017) and we extend the attention mechanism in two ways. First, we construct representations of text segments at different granularities solely with stacked self-attention. Second, we try to extract the truly matched segment pairs with attention across the context and response. We jointly introduce those two kinds of attention in one uniform neural network. Experiments on two large-scale multi-turn response selection tasks show that our proposed model significantly outperforms the state-of-the-art models.