Automated audio captioning (AAC) aims to describe audio data with captions using natural language. Most existing AAC methods adopt an encoder-decoder structure, where the attention based mechanism is a popular choice in the decoder (e.g., Transformer decoder) for predicting captions from audio features. Such attention based decoders can capture the global information from the audio features, however, their ability in extracting local information can be limited, which may lead to degraded quality in the generated captions. In this paper, we present an AAC method with an attention-free decoder, where an encoder based on PANNs is employed for audio feature extraction, and the attention-free decoder is designed to introduce local information. The proposed method enables the effective use of both global and local information from audio signals. Experiments show that our method outperforms the state-of-the-art methods with the standard attention based decoder in Task 6 of the DCASE 2021 Challenge.
Realizing general-purpose language intelligence has been a longstanding goal for natural language processing, where standard evaluation benchmarks play a fundamental and guiding role. We argue that for general-purpose language intelligence evaluation, the benchmark itself needs to be comprehensive and systematic. To this end, we propose CUGE, a Chinese Language Understanding and Generation Evaluation benchmark with the following features: (1) Hierarchical benchmark framework, where datasets are principally selected and organized with a language capability-task-dataset hierarchy. (2) Multi-level scoring strategy, where different levels of model performance are provided based on the hierarchical framework. To facilitate CUGE, we provide a public leaderboard that can be customized to support flexible model judging criteria. Evaluation results on representative pre-trained language models indicate ample room for improvement towards general-purpose language intelligence. CUGE is publicly available at cuge.baai.ac.cn.
Standard multi-task benchmarks are essential for driving the progress of general pretraining models to generalize to various downstream tasks. However, existing benchmarks such as GLUE and GLGE tend to focus on short text understanding and generation tasks, without considering long text modeling, which requires many distinct capabilities such as modeling long-range commonsense and discourse relations, as well as the coherence and controllability of generation. The lack of standardized benchmarks makes it difficult to fully evaluate these capabilities of a model and fairly compare different models, especially Chinese pretraining models. Therefore, we propose LOT, a benchmark including two understanding and two generation tasks for Chinese long text modeling evaluation. We construct the datasets for the tasks based on various kinds of human-written Chinese stories. Besides, we release an encoder-decoder Chinese long text pretraining model named LongLM with up to 1 billion parameters. We pretrain LongLM on 120G Chinese novels with two generative tasks including text infilling and conditional continuation. Extensive experiments on LOT demonstrate that LongLM matches the performance of similar-sized pretraining models on the understanding tasks and outperforms strong baselines substantially on the generation tasks.
In recent years, the size of pre-trained language models (PLMs) has grown by leaps and bounds. However, efficiency issues of these large-scale PLMs limit their utilization in real-world scenarios. We present a suite of cost-effective techniques for the use of PLMs to deal with the efficiency issues of pre-training, fine-tuning, and inference. (1) We introduce knowledge inheritance to accelerate the pre-training process by exploiting existing PLMs instead of training models from scratch. (2) We explore the best practice of prompt tuning with large-scale PLMs. Compared with conventional fine-tuning, prompt tuning significantly reduces the number of task-specific parameters. (3) We implement a new inference toolkit, namely InfMoE, for using large-scale PLMs with limited computational resources. Based on our cost-effective pipeline, we pre-train two models: an encoder-decoder bilingual model with 11 billion parameters (CPM-2) and its corresponding MoE version with 198 billion parameters. In our experiments, we compare CPM-2 with mT5 on downstream tasks. Experimental results show that CPM-2 has excellent general language intelligence. Moreover, we validate the efficiency of InfMoE when conducting inference of large-scale models having tens of billions of parameters on a single GPU. All source code and model parameters are available at https://github.com/TsinghuaAI/CPM.
Current storytelling systems focus more ongenerating stories with coherent plots regard-less of the narration style, which is impor-tant for controllable text generation. There-fore, we propose a new task, stylized story gen-eration, namely generating stories with speci-fied style given a leading context. To tacklethe problem, we propose a novel generationmodel that first plans the stylized keywordsand then generates the whole story with theguidance of the keywords. Besides, we pro-pose two automatic metrics to evaluate theconsistency between the generated story andthe specified style. Experiments demonstratesthat our model can controllably generateemo-tion-driven orevent-driven stories based onthe ROCStories dataset (Mostafazadeh et al.,2016). Our study presents insights for stylizedstory generation in further research.
Generating long and coherent text is an important but challenging task, particularly for open-ended language generation tasks such as story generation. Despite the success in modeling intra-sentence coherence, existing generation models (e.g., BART) still struggle to maintain a coherent event sequence throughout the generated text. We conjecture that this is because of the difficulty for the decoder to capture the high-level semantics and discourse structures in the context beyond token-level co-occurrence. In this paper, we propose a long text generation model, which can represent the prefix sentences at sentence level and discourse level in the decoding process. To this end, we propose two pretraining objectives to learn the representations by predicting inter-sentence semantic similarity and distinguishing between normal and shuffled sentence orders. Extensive experiments show that our model can generate more coherent texts than state-of-the-art baselines.
Automatic metrics are essential for developing natural language generation (NLG) models, particularly for open-ended language generation tasks such as story generation. However, existing automatic metrics are observed to correlate poorly with human evaluation. The lack of standardized benchmark datasets makes it difficult to fully evaluate the capabilities of a metric and fairly compare different metrics. Therefore, we propose OpenMEVA, a benchmark for evaluating open-ended story generation metrics. OpenMEVA provides a comprehensive test suite to assess the capabilities of metrics, including (a) the correlation with human judgments, (b) the generalization to different model outputs and datasets, (c) the ability to judge story coherence, and (d) the robustness to perturbations. To this end, OpenMEVA includes both manually annotated stories and auto-constructed test examples. We evaluate existing metrics on OpenMEVA and observe that they have poor correlation with human judgments, fail to recognize discourse-level incoherence, and lack inferential knowledge (e.g., causal order between events), the generalization ability and robustness. Our study presents insights for developing NLG models and metrics in further research.
The transformer based model (e.g., FusingTF) has been employed recently for Electrocardiogram (ECG) signal classification. However, the high-dimensional embedding obtained via 1-D convolution and positional encoding can lead to the loss of the signal's own temporal information and a large amount of training parameters. In this paper, we propose a new method for ECG classification, called low-dimensional denoising embedding transformer (LDTF), which contains two components, i.e., low-dimensional denoising embedding (LDE) and transformer learning. In the LDE component, a low-dimensional representation of the signal is obtained in the time-frequency domain while preserving its own temporal information. And with the low dimensional embedding, the transformer learning is then used to obtain a deeper and narrower structure with fewer training parameters than that of the FusingTF. Experiments conducted on the MIT-BIH dataset demonstrates the effectiveness and the superior performance of our proposed method, as compared with state-of-the-art methods.