We show that transformer-based large language models are computationally universal when augmented with an external memory. Any deterministic language model that conditions on strings of bounded length is equivalent to a finite automaton, hence computationally limited. However, augmenting such models with a read-write memory creates the possibility of processing arbitrarily large inputs and, potentially, simulating any algorithm. We establish that an existing large language model, Flan-U-PaLM 540B, can be combined with an associative read-write memory to exactly simulate the execution of a universal Turing machine, $U_{15,2}$. A key aspect of the finding is that it does not require any modification of the language model weights. Instead, the construction relies solely on designing a form of stored instruction computer that can subsequently be programmed with a specific set of prompts.
Retrieval-Augmented Code Generation for Universal Information Extraction
CAS Key Laboratory of Network Data Science and Technology, Institute of Computing Technology, Chinese Academy of Sciences（中国科学院计算技术研究所网络数据科学与技术重点实验室）；School of Computer Science and Technology, University of Chinese Academy of Sciences（中国科学院大学计算机科学与技术学院）；Platform and Content Group, Tencent（腾讯平台与内容事业群）
Huawei Noah’s Ark Lab（华为诺亚方舟实验室）；University of Waterloo（滑铁卢大学）
Data Augmentation (DA) is known to improve the generalizability of deep neural networks. Most existing DA techniques naively add a certain number of augmented samples without considering the quality and the added computational cost of these samples. To tackle this problem, a common strategy, adopted by several state-of-the-art DA methods, is to adaptively generate or re-weight augmented samples with respect to the task objective during training. However, these adaptive DA methods: (1) are computationally expensive and not sample-efficient, and (2) are designed merely for a specific setting. In…
Pre-trained language models (PLMs) have achieved remarkable success on various natural language understanding tasks. Simple fine-tuning of PLMs, on the other hand, might be suboptimal for domain-specific tasks because they cannot possibly cover knowledge from all domains. While adaptive pre-training of PLMs can help them obtain domain-specific knowledge, it requires a large training cost. Moreover, adaptive pre-training can harm the PLM's performance on the downstream task by causing catastrophic forgetting of its general knowledge. To overcome such limitations of adaptive pre-training for PLM…
University of Helsinki（赫尔辛基大学）；Universidade Federal do Pará（帕拉联邦大学）
This paper presents and discusses the first Universal Dependencies treebank for the Apurinã language. The treebank contains 76 fully annotated sentences, applies 14 parts-of-speech, as well as seven augmented or new features - some of which are unique to Apurinã. The construction of the treebank has also served as an opportunity to develop finite-state description of the language and facilitate the transfer of open-source infrastructure possibilities to an endangered language of the Amazon. The source materials used in the initial treebank represent fieldwork practices where not all tokens of …
University of California, Los Angeles（加利福尼亚大学洛杉矶分校）；Facebook AI（脸书人工智能研究院）
In recent years, we have seen a colossal effort in pre-training multilingual text encoders using large-scale corpora in many languages to facilitate cross-lingual transfer learning. However, due to typological differences across languages, the cross-lingual transfer is challenging. Nevertheless, language syntax, e.g., syntactic dependencies, can bridge the typological gap. Previous works have shown that pre-trained multilingual encoders, such as mBERT \cite{devlin-etal-2019-bert}, capture language syntax, helping cross-lingual transfer. This work shows that explicitly providing language syntax…
Retrieval-based language models (R-LM) model the probability of natural language text by combining a standard language model (LM) with examples retrieved from an external datastore at test time. While effective, a major bottleneck of using these models in practice is the computationally costly datastore search, which can be performed as frequently as every time step. In this paper, we present RetoMaton - retrieval automaton - which approximates the datastore search, based on (1) saving pointers between consecutive datastore entries, and (2) clustering of entries into "states". This effectively…
