Effective and Efficient Conversation Retrieval for Dialogue State Tracking with Implicit Text Summaries

Few-shot dialogue state tracking (DST) with Large Language Models (LLM) relies on an effective and efficient conversation retriever to find similar in-context examples for prompt learning. Previous works use raw dialogue context as search keys and queries, and a retriever is fine-tuned with annotated dialogues to achieve superior performance. However, the approach is less suited for scaling to new domains or new annotation languages, where fine-tuning data is unavailable. To address this problem, we handle the task of conversation retrieval based on text summaries of the conversations. A LLM-based conversation summarizer is adopted for query and key generation, which enables effective maximum inner product search. To avoid the extra inference cost brought by LLM-based conversation summarization, we further distill a light-weight conversation encoder which produces query embeddings without decoding summaries for test conversations. We validate our retrieval approach on MultiWOZ datasets with GPT-Neo-2.7B and LLaMA-7B/30B. The experimental results show a significant improvement over relevant baselines in real few-shot DST settings.

Self-Supervised Dataset Distillation for Transfer Learning

Dataset distillation methods have achieved remarkable success in distilling a large dataset into a small set of representative samples. However, they are not designed to produce a distilled dataset that can be effectively used for facilitating self-supervised pre-training. To this end, we propose a novel problem of distilling an unlabeled dataset into a set of small synthetic samples for efficient self-supervised learning (SSL). We first prove that a gradient of synthetic samples with respect to a SSL objective in naive bilevel optimization is \textit{biased} due to the randomness originating from data augmentations or masking. To address this issue, we propose to minimize the mean squared error (MSE) between a model's representations of the synthetic examples and their corresponding learnable target feature representations for the inner objective, which does not introduce any randomness. Our primary motivation is that the model obtained by the proposed inner optimization can mimic the \textit{self-supervised target model}. To achieve this, we also introduce the MSE between representations of the inner model and the self-supervised target model on the original full dataset for outer optimization. Lastly, assuming that a feature extractor is fixed, we only optimize a linear head on top of the feature extractor, which allows us to reduce the computational cost and obtain a closed-form solution of the head with kernel ridge regression. We empirically validate the effectiveness of our method on various applications involving transfer learning.

Self-Supervised Set Representation Learning for Unsupervised Meta-Learning

Dataset distillation methods have achieved remarkable success in distilling a large dataset into a small set of representative samples. However, they are not designed to produce a distilled dataset that can be effectively used for facilitating self-supervised pre-training. To this end, we propose a novel problem of distilling an unlabeled dataset into a set of small synthetic samples for efficient self-supervised learning (SSL). We first prove that a gradient of synthetic samples with respect to a SSL objective in naive bilevel optimization is \textit{biased} due to the randomness originating from data augmentations or masking. To address this issue, we propose to minimize the mean squared error (MSE) between a model's representations of the synthetic examples and their corresponding learnable target feature representations for the inner objective, which does not introduce any randomness. Our primary motivation is that the model obtained by the proposed inner optimization can mimic the \textit{self-supervised target model}. To achieve this, we also introduce the MSE between representations of the inner model and the self-supervised target model on the original full dataset for outer optimization. Lastly, assuming that a feature extractor is fixed, we only optimize a linear head on top of the feature extractor, which allows us to reduce the computational cost and obtain a closed-form solution of the head with kernel ridge regression. We empirically validate the effectiveness of our method on various applications involving transfer learning.

Drug Discovery with Dynamic Goal-aware Fragments

Fragment-based drug discovery is an effective strategy for discovering drug candidates in the vast chemical space, and has been widely employed in molecular generative models. However, many existing fragment extraction methods in such models do not take the target chemical properties into account or rely on heuristic rules. Additionally, the existing fragment-based generative models cannot update the fragment vocabulary with goal-aware fragments newly discovered during the generation. To this end, we propose a molecular generative framework for drug discovery, named Goal-aware fragment Extraction, Assembly, and Modification (GEAM). GEAM consists of three modules, each responsible for goal-aware fragment extraction, fragment assembly, and fragment modification. The fragment extraction module identifies important fragments that contribute to the desired target properties with the information bottleneck principle, thereby constructing an effective goal-aware fragment vocabulary. Moreover, GEAM can explore beyond the initial vocabulary with the fragment modification module, and the exploration is further enhanced through the dynamic goal-aware vocabulary update. We experimentally demonstrate that GEAM effectively discovers drug candidates through the generative cycle of the three modules in various drug discovery tasks.

Knowledge-Augmented Reasoning Distillation for Small Language Models in Knowledge-Intensive Tasks

Large Language Models (LLMs) have shown promising performance in knowledge-intensive reasoning tasks that require a compound understanding of knowledge. However, deployment of the LLMs in real-world applications can be challenging due to their high computational requirements and concerns on data privacy. Previous studies have focused on building task-specific small language models (LMs) by fine-tuning them with labeled data or distilling LLMs. However, these approaches are ill-suited for knowledge-intensive reasoning tasks due to the limited capacity of small LMs in memorizing the knowledge required. Motivated by our theoretical analysis on memorization, we propose Knowledge-Augmented Reasoning Distillation (KARD), a novel method that fine-tunes small LMs to generate rationales with augmented knowledge retrieved from an external knowledge base. Moreover, we further propose a neural reranker to obtain documents relevant to rationale generation. We empirically show that KARD significantly improves the performance of small T5 and Flan-T5 models on the challenging knowledge-intensive reasoning datasets, namely MedQA-USMLE and StrategyQA. Notably, our method makes the 250M models achieve superior performance against the fine-tuned 3B models, having 12 times larger parameters, on both MedQA-USMLE and StrategyQA benchmarks.

DiffusionNAG: Task-guided Neural Architecture Generation with Diffusion Models

Neural Architecture Search (NAS) has emerged as a powerful technique for automating neural architecture design. However, existing NAS methods either require an excessive amount of time for repetitive training or sampling of many task-irrelevant architectures. Moreover, they lack generalization across different tasks and usually require searching for optimal architectures for each task from scratch without reusing the knowledge from the previous NAS tasks. To tackle such limitations of existing NAS methods, we propose a novel transferable task-guided Neural Architecture Generation (NAG) framework based on diffusion models, dubbed DiffusionNAG. With the guidance of a surrogate model, such as a performance predictor for a given task, our DiffusionNAG can generate task-optimal architectures for diverse tasks, including unseen tasks. DiffusionNAG is highly efficient as it generates task-optimal neural architectures by leveraging the prior knowledge obtained from the previous tasks and neural architecture distribution. Furthermore, we introduce a score network to ensure the generation of valid architectures represented as directed acyclic graphs, unlike existing graph generative models that focus on generating undirected graphs. Extensive experiments demonstrate that DiffusionNAG significantly outperforms the state-of-the-art transferable NAG model in architecture generation quality, as well as previous NAS methods on four computer vision datasets with largely reduced computational cost.

On Divergence Measures for Bayesian Pseudocoresets

A Bayesian pseudocoreset is a small synthetic dataset for which the posterior over parameters approximates that of the original dataset. While promising, the scalability of Bayesian pseudocoresets is not yet validated in realistic problems such as image classification with deep neural networks. On the other hand, dataset distillation methods similarly construct a small dataset such that the optimization using the synthetic dataset converges to a solution with performance competitive with optimization using full data. Although dataset distillation has been empirically verified in large-scale settings, the framework is restricted to point estimates, and their adaptation to Bayesian inference has not been explored. This paper casts two representative dataset distillation algorithms as approximations to methods for constructing pseudocoresets by minimizing specific divergence measures: reverse KL divergence and Wasserstein distance. Furthermore, we provide a unifying view of such divergence measures in Bayesian pseudocoreset construction. Finally, we propose a novel Bayesian pseudocoreset algorithm based on minimizing forward KL divergence. Our empirical results demonstrate that the pseudocoresets constructed from these methods reflect the true posterior even in high-dimensional Bayesian inference problems.

Self-Distillation for Further Pre-training of Transformers

Pre-training a large transformer model on a massive amount of unlabeled data and fine-tuning it on labeled datasets for diverse downstream tasks has proven to be a successful strategy, for a variety of vision and natural language processing tasks. However, direct fine-tuning of the pre-trained model may be suboptimal if there exist large discrepancies across data domains for pre-training and fine-tuning. To tackle this issue, several previous studies have proposed further pre-training strategies, where we continue to pre-train the model on the target unlabeled dataset before fine-tuning. However, all of them solely focus on language models and we empirically find that a Vision Transformer is vulnerable to overfitting as we continue to pretrain the model on target unlabeled data. In order to tackle this limitation, we propose self-distillation as a regularization for a further pre-training stage. Specifically, we first further pre-train the initial pre-trained model on the target unlabeled data and then consider it as a teacher for self-distillation. Then we take the same initial pre-trained model as a student and enforce its hidden representations to be close to those of the teacher while optimizing the student with a masked auto-encoding objective. We empirically validate the efficacy of self-distillation on a variety of benchmark datasets for image and text classification tasks. Experimentally, we show that our proposed method outperforms all the relevant baselines. Theoretically, we analyze the proposed method with a simplified model to understand how self-distillation for further pre-training can potentially help improve the performance of the downstream tasks.

Set-based Meta-Interpolation for Few-Task Meta-Learning

Meta-learning approaches enable machine learning systems to adapt to new tasks given few examples by leveraging knowledge from related tasks. However, a large number of meta-training tasks are still required for generalization to unseen tasks during meta-testing, which introduces a critical bottleneck for real-world problems that come with only few tasks, due to various reasons including the difficulty and cost of constructing tasks. Recently, several task augmentation methods have been proposed to tackle this issue using domain-specific knowledge to design augmentation techniques to densify the meta-training task distribution. However, such reliance on domain-specific knowledge renders these methods inapplicable to other domains. While Manifold Mixup based task augmentation methods are domain-agnostic, we empirically find them ineffective on non-image domains. To tackle these limitations, we propose a novel domain-agnostic task augmentation method, Meta-Interpolation, which utilizes expressive neural set functions to densify the meta-training task distribution using bilevel optimization. We empirically validate the efficacy of Meta-Interpolation on eight datasets spanning across various domains such as image classification, molecule property prediction, text classification and speech recognition. Experimentally, we show that Meta-Interpolation consistently outperforms all the relevant baselines. Theoretically, we prove that task interpolation with the set function regularizes the meta-learner to improve generalization.

Sequential Reptile: Inter-Task Gradient Alignment for Multilingual Learning

Multilingual models jointly pretrained on multiple languages have achieved remarkable performance on various multilingual downstream tasks. Moreover, models finetuned on a single monolingual downstream task have shown to generalize to unseen languages. In this paper, we first show that it is crucial for those tasks to align gradients between them in order to maximize knowledge transfer while minimizing negative transfer. Despite its importance, the existing methods for gradient alignment either have a completely different purpose, ignore inter-task alignment, or aim to solve continual learning problems in rather inefficient ways. As a result of the misaligned gradients between tasks, the model suffers from severe negative transfer in the form of catastrophic forgetting of the knowledge acquired from the pretraining. To overcome the limitations, we propose a simple yet effective method that can efficiently align gradients between tasks. Specifically, we perform each inner-optimization by sequentially sampling batches from all the tasks, followed by a Reptile outer update. Thanks to the gradients aligned between tasks by our method, the model becomes less vulnerable to negative transfer and catastrophic forgetting. We extensively validate our method on various multi-task learning and zero-shot cross-lingual transfer tasks, where our method largely outperforms all the relevant baselines we consider.