We demonstrate a proof-of-concept of a large language model conducting corporate lobbying related activities. An autoregressive large language model (OpenAI's text-davinci-003) determines if proposed U.S. Congressional bills are relevant to specific public companies and provides explanations and confidence levels. For the bills the model deems as relevant, the model drafts a letter to the sponsor of the bill in an attempt to persuade the congressperson to make changes to the proposed legislation. We use hundreds of novel ground-truth labels of the relevance of a bill to a company to benchmark the performance of the model, which outperforms the baseline of predicting the most common outcome of irrelevance. We also benchmark the performance of the previous OpenAI GPT-3 model (text-davinci-002), which was the state-of-the-art model on many academic natural language tasks until text-davinci-003 was recently released. The performance of text-davinci-002 is worse than simply always predicting that a bill is irrelevant to a company. These results suggest that, as large language models continue to exhibit improved core natural language understanding capabilities, performance on corporate lobbying related tasks will continue to improve. If AI begins to influence law in a manner that is not a direct extension of human intentions, this threatens the critical role that law as information could play in aligning AI with humans. This paper explores how this is increasingly a possibility. Initially, AI is being used to simply augment human lobbyists. However, there may be a slow creep of less and less human oversight over automated assessments of policy ideas and the written communication to regulatory agencies and Congressional staffers. The core question raised is where to draw the line between human-driven and AI-driven policy influence.
Despite advances in deep learning, current state-of-the-art speech emotion recognition (SER) systems still have poor performance due to a lack of speech emotion datasets. This paper proposes augmenting SER systems with synthetic emotional speech generated by an end-to-end text-to-speech (TTS) system based on an extended Tacotron architecture. The proposed TTS system includes encoders for speaker and emotion embeddings, a sequence-to-sequence text generator for creating Mel-spectrograms, and a WaveRNN to generate audio from the Mel-spectrograms. Extensive experiments show that the quality of the generated emotional speech can significantly improve SER performance on multiple datasets, as demonstrated by a higher mean opinion score (MOS) compared to the baseline. The generated samples were also effective at augmenting SER performance.
Recently, there has been growing interest in the ability of Transformer-based models to produce meaningful embeddings of text with several applications, such as text similarity. Despite significant progress in the field, the explanations for similarity predictions remain challenging, especially in unsupervised settings. In this work, we present an unsupervised technique for explaining paragraph similarities inferred by pre-trained BERT models. By looking at a pair of paragraphs, our technique identifies important words that dictate each paragraph's semantics, matches between the words in both paragraphs, and retrieves the most important pairs that explain the similarity between the two. The method, which has been assessed by extensive human evaluations and demonstrated on datasets comprising long and complex paragraphs, has shown great promise, providing accurate interpretations that correlate better with human perceptions.
It was observed that large language models exhibit a power-law decay of cross entropy with respect to the number of parameters and training tokens. When extrapolated literally, this decay implies that the entropy rate of natural language is zero. To understand this phenomenon -- or an artifact -- better, we construct a simple stationary stochastic process and its memory-based predictor that exhibit a power-law decay of cross entropy with the vanishing entropy rate. Our example is based on previously discussed Santa Fe processes, which decompose a random text into a process of narration and time-independent knowledge. Previous discussions assumed that narration is a memoryless source with Zipf's distribution. In this paper, we propose a model of narration that has the vanishing entropy rate and applies a randomly chosen deterministic sequence called a multiperiodic sequence. Under a suitable parameterization, multiperiodic sequences exhibit asymptotic relative frequencies given by Zipf's law. Remaining agnostic about the value of the entropy rate of natural language, we discuss relevance of similar constructions for language modeling.
This paper uses textual data contained in certified (q-graded) coffee reviews to predict corresponding scores on a scale from 0-100. By transforming this highly specialized and standardized textual data in a predictor space, we construct regression models which accurately capture the patterns in corresponding coffee bean scores.
Recent advances in the healthcare industry have led to an abundance of unstructured data, making it challenging to perform tasks such as efficient and accurate information retrieval at scale. Our work offers an all-in-one scalable solution for extracting and exploring complex information from large-scale research documents, which would otherwise be tedious. First, we briefly explain our knowledge synthesis process to extract helpful information from unstructured text data of research documents. Then, on top of the knowledge extracted from the documents, we perform complex information retrieval using three major components- Paragraph Retrieval, Triplet Retrieval from Knowledge Graphs, and Complex Question Answering (QA). These components combine lexical and semantic-based methods to retrieve paragraphs and triplets and perform faceted refinement for filtering these search results. The complexity of biomedical queries and documents necessitates using a QA system capable of handling queries more complex than factoid queries, which we evaluate qualitatively on the COVID-19 Open Research Dataset (CORD-19) to demonstrate the effectiveness and value-add.
Social bias in Pretrained Language Models (PLMs) affects text generation and other downstream NLP tasks. Existing bias testing methods rely predominantly on manual templates or on expensive crowd-sourced data. We propose a novel AutoBiasTest method that automatically generates sentences for testing bias in PLMs, hence providing a flexible and low-cost alternative. Our approach uses another PLM for generation and controls the generation of sentences by conditioning on social group and attribute terms. We show that generated sentences are natural and similar to human-produced content in terms of word length and diversity. We illustrate that larger models used for generation produce estimates of social bias with lower variance. We find that our bias scores are well correlated with manual templates, but AutoBiasTest highlights biases not captured by these templates due to more diverse and realistic test sentences. By automating large-scale test sentence generation, we enable better estimation of underlying bias distributions
Prompting pre-trained language models leads to promising results across natural language processing tasks but is less effective when applied in low-resource domains, due to the domain gap between the pre-training data and the downstream task. In this work, we bridge this gap with a novel and lightweight prompting methodology called SwitchPrompt for the adaptation of language models trained on datasets from the general domain to diverse low-resource domains. Using domain-specific keywords with a trainable gated prompt, SwitchPrompt offers domain-oriented prompting, that is, effective guidance on the target domains for general-domain language models. Our few-shot experiments on three text classification benchmarks demonstrate the efficacy of the general-domain pre-trained language models when used with SwitchPrompt. They often even outperform their domain-specific counterparts trained with baseline state-of-the-art prompting methods by up to 10.7% performance increase in accuracy. This result indicates that SwitchPrompt effectively reduces the need for domain-specific language model pre-training.
In Spoken language understanding (SLU), a natural solution is concatenating pre-trained speech models (e.g. HuBERT) and pretrained language models (PLM, e.g. T5). Most previous works use pretrained language models with subword-based tokenization. However, the granularity of input units affects the alignment of speech model outputs and language model inputs, and PLM with character-based tokenization is underexplored. In this work, we conduct extensive studies on how PLMs with different tokenization strategies affect spoken language understanding task including spoken question answering (SQA) and speech translation (ST). We further extend the idea to create T5lephone(pronounced as telephone), a variant of T5 that is pretrained using phonemicized text. We initialize T5lephone with existing PLMs to pretrain it using relatively lightweight computational resources. We reached state-of-the-art on NMSQA, and the T5lephone model exceeds T5 with other types of units on end-to-end SQA and ST.
Robot learning methods have the potential for widespread generalization across tasks, environments, and objects. However, these methods require large diverse datasets that are expensive to collect in real-world robotics settings. For robot learning to generalize, we must be able to leverage sources of data or priors beyond the robot's own experience. In this work, we posit that image-text generative models, which are pre-trained on large corpora of web-scraped data, can serve as such a data source. We show that despite these generative models being trained on largely non-robotics data, they can serve as effective ways to impart priors into the process of robot learning in a way that enables widespread generalization. In particular, we show how pre-trained generative models can serve as effective tools for semantically meaningful data augmentation. By leveraging these pre-trained models for generating appropriate "semantic" data augmentations, we propose a system GenAug that is able to significantly improve policy generalization. We apply GenAug to tabletop manipulation tasks, showing the ability to re-target behavior to novel scenarios, while only requiring marginal amounts of real-world data. We demonstrate the efficacy of this system on a number of object manipulation problems in the real world, showing a 40% improvement in generalization to novel scenes and objects.