Vision-Language Models (VLMs) such as GPT-4V have recently demonstrated incredible strides on diverse vision language tasks. We dig into vision-based deductive reasoning, a more sophisticated but less explored realm, and find previously unexposed blindspots in the current SOTA VLMs. Specifically, we leverage Raven's Progressive Matrices (RPMs), to assess VLMs' abilities to perform multi-hop relational and deductive reasoning relying solely on visual clues. We perform comprehensive evaluations of several popular VLMs employing standard strategies such as in-context learning, self-consistency, and Chain-of-thoughts (CoT) on three diverse datasets, including the Mensa IQ test, IntelligenceTest, and RAVEN. The results reveal that despite the impressive capabilities of LLMs in text-based reasoning, we are still far from achieving comparable proficiency in visual deductive reasoning. We found that certain standard strategies that are effective when applied to LLMs do not seamlessly translate to the challenges presented by visual reasoning tasks. Moreover, a detailed analysis reveals that VLMs struggle to solve these tasks mainly because they are unable to perceive and comprehend multiple, confounding abstract patterns in RPM examples.
Recent methods have demonstrated that Large Language Models (LLMs) can solve reasoning tasks better when they are encouraged to solve subtasks of the main task first. In this paper we devise a similar strategy that breaks down reasoning tasks into a problem decomposition phase and a problem solving phase and show that the strategy is able to outperform a single stage solution. Further, we hypothesize that the decomposition should be easier to distill into a smaller model compared to the problem solving because the latter requires large amounts of domain knowledge while the former only requires learning general problem solving strategies. We propose methods to distill these two capabilities and evaluate their impact on reasoning outcomes and inference cost. We find that we can distill the problem decomposition phase and at the same time achieve good generalization across tasks, datasets, and models. However, it is harder to distill the problem solving capability without losing performance and the resulting distilled model struggles with generalization. These results indicate that by using smaller, distilled problem decomposition models in combination with problem solving LLMs we can achieve reasoning with cost-efficient inference and local adaptation.
Large language models are trained on massive scrapes of the web, which are often unstructured, noisy, and poorly phrased. Current scaling laws show that learning from such data requires an abundance of both compute and data, which grows with the size of the model being trained. This is infeasible both because of the large compute costs and duration associated with pre-training, and the impending scarcity of high-quality data on the web. In this work, we propose Web Rephrase Augmented Pre-training ($\textbf{WRAP}$) that uses an off-the-shelf instruction-tuned model prompted to paraphrase documents on the web in specific styles such as "like Wikipedia" or in "question-answer format" to jointly pre-train LLMs on real and synthetic rephrases. First, we show that using WRAP on the C4 dataset, which is naturally noisy, speeds up pre-training by $\sim3x$. At the same pre-training compute budget, it improves perplexity by more than 10% on average across different subsets of the Pile, and improves zero-shot question answer accuracy across 13 tasks by more than 2%. Second, we investigate the impact of the re-phrasing style on the performance of the model, offering insights into how the composition of the training data can impact the performance of LLMs in OOD settings. Our gains are attributed to the fact that re-phrased synthetic data has higher utility than just real data because it (i) incorporates style diversity that closely reflects downstream evaluation style, and (ii) has higher 'quality' than web-scraped data.
Current approaches to evaluating large language models (LLMs) with pre-existing Knowledge Graphs (KG) mostly ignore the structure of the KG and make arbitrary choices of which part of the graph to evaluate. In this paper, we introduce KGLens, a method to evaluate LLMs by generating natural language questions from a KG in a structure aware manner so that we can characterize its performance on a more aggregated level. KGLens uses a parameterized KG, where each edge is augmented with a beta distribution that guides how to sample edges from the KG for QA testing. As the evaluation proceeds, different edges of the parameterized KG are sampled and assessed appropriately, converging to a more global picture of the performance of the LLMs on the KG as a whole. In our experiments, we construct three domain-specific KGs for knowledge assessment, comprising over 19,000 edges, 700 relations, and 21,000 entities. The results demonstrate that KGLens can not only assess overall performance but also provide topic, temporal, and relation analyses of LLMs. This showcases the adaptability and customizability of KGLens, emphasizing its ability to focus the evaluation based on specific criteria.
Due to the expanding scope of machine learning (ML) to the fields of sensor networking, cooperative robotics and many other multi-agent systems, distributed deployment of inference algorithms has received a lot of attention. These algorithms involve collaboratively learning unknown parameters from dispersed data collected by multiple agents. There are two competing aspects in such algorithms, namely, intra-agent computation and inter-agent communication. Traditionally, algorithms are designed to perform both synchronously. However, certain circumstances need frugal use of communication channels as they are either unreliable, time-consuming, or resource-expensive. In this paper, we propose gossip-based asynchronous communication to leverage fast computations and reduce communication overhead simultaneously. We analyze the effects of multiple (local) intra-agent computations by the active agents between successive inter-agent communications. For local computations, Bayesian sampling via unadjusted Langevin algorithm (ULA) MCMC is utilized. The communication is assumed to be over a connected graph (e.g., as in decentralized learning), however, the results can be extended to coordinated communication where there is a central server (e.g., federated learning). We theoretically quantify the convergence rates in the process. To demonstrate the efficacy of the proposed algorithm, we present simulations on a toy problem as well as on real world data sets to train ML models to perform classification tasks. We observe faster initial convergence and improved performance accuracy, especially in the low data range. We achieve on average 78% and over 90% classification accuracy respectively on the Gamma Telescope and mHealth data sets from the UCI ML repository.
Datasets that pair Knowledge Graphs (KG) and text together (KG-T) can be used to train forward and reverse neural models that generate text from KG and vice versa. However models trained on datasets where KG and text pairs are not equivalent can suffer from more hallucination and poorer recall. In this paper, we verify this empirically by generating datasets with different levels of noise and find that noisier datasets do indeed lead to more hallucination. We argue that the ability of forward and reverse models trained on a dataset to cyclically regenerate source KG or text is a proxy for the equivalence between the KG and the text in the dataset. Using cyclic evaluation we find that manually created WebNLG is much better than automatically created TeKGen and T-REx. Guided by these observations, we construct a new, improved dataset called LAGRANGE using heuristics meant to improve equivalence between KG and text and show the impact of each of the heuristics on cyclic evaluation. We also construct two synthetic datasets using large language models (LLMs), and observe that these are conducive to models that perform significantly well on cyclic generation of text, but less so on cyclic generation of KGs, probably because of a lack of a consistent underlying ontology.
We present a practical asynchronous data fusion model for networked agents to perform distributed Bayesian learning without sharing raw data. Our algorithm uses a gossip-based approach where pairs of randomly selected agents employ unadjusted Langevin dynamics for parameter sampling. We also introduce an event-triggered mechanism to further reduce communication between gossiping agents. These mechanisms drastically reduce communication overhead and help avoid bottlenecks commonly experienced with distributed algorithms. In addition, the reduced link utilization by the algorithm is expected to increase resiliency to occasional link failure. We establish mathematical guarantees for our algorithm and demonstrate its effectiveness via numerical experiments.
Speech representation learning has improved both speech understanding and speech synthesis tasks for single language. However, its ability in cross-lingual scenarios has not been explored. In this paper, we extend the pretraining method for cross-lingual multi-speaker speech synthesis tasks, including cross-lingual multi-speaker voice cloning and cross-lingual multi-speaker speech editing. We propose a speech-text joint pretraining framework, where we randomly mask the spectrogram and the phonemes given a speech example and its transcription. By learning to reconstruct the masked parts of the input in different languages, our model shows great improvements over speaker-embedding-based multi-speaker TTS methods. Moreover, our framework is end-to-end for both the training and the inference without any finetuning effort. In cross-lingual multi-speaker voice cloning and cross-lingual multi-speaker speech editing tasks, our experiments show that our model outperforms speaker-embedding-based multi-speaker TTS methods. The code and model are publicly available at PaddleSpeech.
In-context learning using large language models has recently shown surprising results for semantic parsing tasks such as Text-to-SQL translation. Prompting GPT-3 or Codex using several examples of question-SQL pairs can produce excellent results, comparable to state-of-the-art finetuning-based models. However, existing work primarily focuses on English datasets, and it is unknown whether large language models can serve as competitive semantic parsers for other languages. To bridge this gap, our work focuses on cross-lingual Text-to-SQL semantic parsing for translating non-English utterances into SQL queries based on an English schema. We consider a zero-shot transfer learning setting with the assumption that we do not have any labeled examples in the target language (but have annotated examples in English). This work introduces the XRICL framework, which learns to retrieve relevant English exemplars for a given query to construct prompts. We also include global translation exemplars for a target language to facilitate the translation process for large language models. To systematically evaluate our model, we construct two new benchmark datasets, XSpider and XKaggle-dbqa, which include questions in Chinese, Vietnamese, Farsi, and Hindi. Our experiments show that XRICL effectively leverages large pre-trained language models to outperform existing baselines. Data and code are publicly available at https://github.com/Impavidity/XRICL.
Class-based language models (LMs) have been long devised to address context sparsity in $n$-gram LMs. In this study, we revisit this approach in the context of neural LMs. We hypothesize that class-based prediction leads to an implicit context aggregation for similar words and thus can improve generalization for rare words. We map words that have a common WordNet hypernym to the same class and train large neural LMs by gradually annealing from predicting the class to token prediction during training. Empirically, this curriculum learning strategy consistently improves perplexity over various large, highly-performant state-of-the-art Transformer-based models on two datasets, WikiText-103 and Arxiv. Our analysis shows that the performance improvement is achieved without sacrificing performance on rare words. Finally, we document other attempts that failed to yield empirical gains, and discuss future directions for the adoption of class-based LMs on a larger scale.