PyTorchKR Papers

최신 논문

총 2,090편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2301.08721공식 코드 · Papers with Code

Batch Prompting: Efficient Inference with Large Language Model APIs

Performing inference on large volumes of samples with large language models (LLMs) can be computationally and financially costly in industry and real-world use. We propose batch prompting, a simple yet effective prompting approach that enables the LLM to run inference in batches, instead of one sample at a time. Our method reduces both token and time costs while retaining downstream performance. We theoretically demonstrate that under a few-shot in-context learning setting, the inference costs decrease almost inverse linearly with the number of samples in each batch. We extensively validate the effectiveness of batch prompting on ten datasets across commonsense QA, arithmetic reasoning, and NLI/NLU: batch prompting significantly~(up to 5x with six samples in batch) reduces the LLM (Codex) inference token and time costs while achieving better or comparable performance. For state-of-the-art Chat-based LLMs, e.g., GPT-3.5 and GPT-4, we show the benefits of batch prompting also hold. Further analysis shows that the number of samples in each batch and the complexity of tasks affect its performance. Moreover, batch prompting can be applied across different reasoning methods using LLMs. Our code can be found at the site https://github.com/xlang-ai/batch-prompting.

arXiv:2301.05402공식 코드 · Papers with Code

In BLOOM: Creativity and Affinity in Artificial Lyrics and Art

We apply a large multilingual language model (BLOOM-176B) in open-ended generation of Chinese song lyrics, and evaluate the resulting lyrics for coherence and creativity using human reviewers. We find that current computational metrics for evaluating large language model outputs (MAUVE) have limitations in evaluation of creative writing. We note that the human concept of creativity requires lyrics to be both comprehensible and distinctive -- and that humans assess certain types of machine-generated lyrics to score more highly than real lyrics by popular artists. Inspired by the inherently multimodal nature of album releases, we leverage a Chinese-language stable diffusion model to produce high-quality lyric-guided album art, demonstrating a creative approach for an artist seeking inspiration for an album or single. Finally, we introduce the MojimLyrics dataset, a Chinese-language dataset of popular song lyrics for future research.

arXiv:2301.05226공식 코드 · Papers with Code

See, Think, Confirm: Interactive Prompting Between Vision and Language Models for Knowledge-based Visual Reasoning

Large pre-trained vision and language models have demonstrated remarkable capacities for various tasks. However, solving the knowledge-based visual reasoning tasks remains challenging, which requires a model to comprehensively understand image content, connect the external world knowledge, and perform step-by-step reasoning to answer the questions correctly. To this end, we propose a novel framework named Interactive Prompting Visual Reasoner (IPVR) for few-shot knowledge-based visual reasoning. IPVR contains three stages, see, think and confirm. The see stage scans the image and grounds the visual concept candidates with a visual perception model. The think stage adopts a pre-trained large language model (LLM) to attend to the key concepts from candidates adaptively. It then transforms them into text context for prompting with a visual captioning model and adopts the LLM to generate the answer. The confirm stage further uses the LLM to generate the supporting rationale to the answer, verify the generated rationale with a cross-modality classifier and ensure that the rationale can infer the predicted output consistently. We conduct experiments on a range of knowledge-based visual reasoning datasets. We found our IPVR enjoys several benefits, 1). it achieves better performance than the previous few-shot learning baselines; 2). it enjoys the total transparency and trustworthiness of the whole reasoning process by providing rationales for each reasoning step; 3). it is computation-efficient compared with other fine-tuning baselines.

arXiv:2301.01181공식 코드 · Papers with Code

Large Language Models as Corporate Lobbyists

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. It 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 the simple baseline. Longer-term, 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. Initially, AI is being used to simply augment human lobbyists for a small portion of their daily tasks. However, firms have an incentive to use 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.

arXiv:2301.00303공식 코드 · Papers with Code

Rethinking with Retrieval: Faithful Large Language Model Inference

Despite the success of large language models (LLMs) in various natural language processing (NLP) tasks, the stored knowledge in these models may inevitably be incomplete, out-of-date, or incorrect. This motivates the need to utilize external knowledge to assist LLMs. Unfortunately, current methods for incorporating external knowledge often require additional training or fine-tuning, which can be costly and may not be feasible for LLMs. To address this issue, we propose a novel post-processing approach, rethinking with retrieval (RR), which retrieves relevant external knowledge based on the decomposed reasoning steps obtained from the chain-of-thought (CoT) prompting. This lightweight approach does not require additional training or fine-tuning and is not limited by the input length of LLMs. We evaluate the effectiveness of RR through extensive experiments with GPT-3 on three complex reasoning tasks: commonsense reasoning, temporal reasoning, and tabular reasoning. Our results show that RR can produce more faithful explanations and improve the performance of LLMs.

arXiv:2212.11214공식 코드 · Papers with Code

Crowd Score: A Method for the Evaluation of Jokes using Large Language Model AI Voters as Judges

This paper presents the Crowd Score, a novel method to assess the funniness of jokes using large language models (LLMs) as AI judges. Our method relies on inducing different personalities into the LLM and aggregating the votes of the AI judges into a single score to rate jokes. We validate the votes using an auditing technique that checks if the explanation for a particular vote is reasonable using the LLM. We tested our methodology on 52 jokes in a crowd of four AI voters with different humour types: affiliative, self-enhancing, aggressive and self-defeating. Our results show that few-shot prompting leads to better results than zero-shot for the voting question. Personality induction showed that aggressive and self-defeating voters are significantly more inclined to find more jokes funny of a set of aggressive/self-defeating jokes than the affiliative and self-enhancing voters. The Crowd Score follows the same trend as human judges by assigning higher scores to jokes that are also considered funnier by human judges. We believe that our methodology could be applied to other creative domains such as story, poetry, slogans, etc. It could both help the adoption of a flexible and accurate standard approach to compare different work in the CC community under a common metric and by minimizing human participation in assessing creative artefacts, it could accelerate the prototyping of creative artefacts and reduce the cost of hiring human participants to rate creative artefacts.

arXiv:2212.10465공식 코드 · Papers with Code

SODA: Million-scale Dialogue Distillation with Social Commonsense Contextualization

Data scarcity has been a long standing issue in the field of open-domain social dialogue. To quench this thirst, we present SODA: the first publicly available, million-scale high-quality social dialogue dataset. By contextualizing social commonsense knowledge from a knowledge graph, we are able to distill an exceptionally broad spectrum of social interactions from a large language model. Human evaluation shows that conversations in SODA are more consistent, specific, and (surprisingly) natural than those in prior human-authored datasets. Using SODA, we train COSMO: a generalizable conversation model that is significantly more natural and consistent on unseen datasets than best-performing conversation models (e.g., GODEL, BlenderBot-1, Koala, Vicuna). Experiments reveal COSMO is sometimes even preferred to the original human-written gold responses. Additionally, our results shed light on the distinction between knowledge-enriched conversations and natural social chitchats. We plan to make our data, model, and code public.

arXiv:2212.10561공식 코드 · Papers with Code

Parsel: Algorithmic Reasoning with Language Models by Composing Decompositions

Despite recent success in large language model (LLM) reasoning, LLMs struggle with hierarchical multi-step reasoning tasks like generating complex programs. For these tasks, humans often start with a high-level algorithmic design and implement each part gradually. We introduce Parsel, a framework enabling automatic implementation and validation of complex algorithms with code LLMs. With Parsel, we automatically decompose algorithmic tasks into hierarchical natural language function descriptions and then search over combinations of possible function implementations using tests. We show that Parsel can be used across domains requiring hierarchical reasoning, including program synthesis and robotic planning. We find that, using Parsel, LLMs solve more competition-level problems in the APPS dataset, resulting in pass rates over 75\% higher than prior results from directly sampling AlphaCode and Codex, while often using a smaller sample budget. Moreover, with automatically generated tests, we find that Parsel can improve the state-of-the-art pass@1 performance on HumanEval from 67\% to 85\%. We also find that LLM-generated robotic plans using Parsel are more than twice as likely to be considered accurate than directly generated plans. Lastly, we explore how Parsel addresses LLM limitations and discuss how Parsel may be useful for human programmers. We release our code at https://github.com/ezelikman/parsel

arXiv:2212.08913공식 코드 · Papers with Code

Claim Optimization in Computational Argumentation

An optimal delivery of arguments is key to persuasion in any debate, both for humans and for AI systems. This requires the use of clear and fluent claims relevant to the given debate. Prior work has studied the automatic assessment of argument quality extensively. Yet, no approach actually improves the quality so far. To fill this gap, this paper proposes the task of claim optimization: to rewrite argumentative claims in order to optimize their delivery. As multiple types of optimization are possible, we approach this task by first generating a diverse set of candidate claims using a large language model, such as BART, taking into account contextual information. Then, the best candidate is selected using various quality metrics. In automatic and human evaluation on an English-language corpus, our quality-based candidate selection outperforms several baselines, improving 60% of all claims (worsening 16% only). Follow-up analyses reveal that, beyond copy editing, our approach often specifies claims with details, whereas it adds less evidence than humans do. Moreover, its capabilities generalize well to other domains, such as instructional texts.

arXiv:2212.09656공식 코드 · Papers with Code

Visconde: Multi-document QA with GPT-3 and Neural Reranking

This paper proposes a question-answering system that can answer questions whose supporting evidence is spread over multiple (potentially long) documents. The system, called Visconde, uses a three-step pipeline to perform the task: decompose, retrieve, and aggregate. The first step decomposes the question into simpler questions using a few-shot large language model (LLM). Then, a state-of-the-art search engine is used to retrieve candidate passages from a large collection for each decomposed question. In the final step, we use the LLM in a few-shot setting to aggregate the contents of the passages into the final answer. The system is evaluated on three datasets: IIRC, Qasper, and StrategyQA. Results suggest that current retrievers are the main bottleneck and that readers are already performing at the human level as long as relevant passages are provided. The system is also shown to be more effective when the model is induced to give explanations before answering a question. Code is available at \url{https://github.com/neuralmind-ai/visconde}.

arXiv:2212.09196공식 코드 · Papers with Code

Emergent Analogical Reasoning in Large Language Models

The recent advent of large language models has reinvigorated debate over whether human cognitive capacities might emerge in such generic models given sufficient training data. Of particular interest is the ability of these models to reason about novel problems zero-shot, without any direct training. In human cognition, this capacity is closely tied to an ability to reason by analogy. Here, we performed a direct comparison between human reasoners and a large language model (the text-davinci-003 variant of GPT-3) on a range of analogical tasks, including a non-visual matrix reasoning task based on the rule structure of Raven's Standard Progressive Matrices. We found that GPT-3 displayed a surprisingly strong capacity for abstract pattern induction, matching or even surpassing human capabilities in most settings; preliminary tests of GPT-4 indicated even better performance. Our results indicate that large language models such as GPT-3 have acquired an emergent ability to find zero-shot solutions to a broad range of analogy problems.

arXiv:2212.09095공식 코드 · Papers with Code

Rethinking the Role of Scale for In-Context Learning: An Interpretability-based Case Study at 66 Billion Scale

Language models have been shown to perform better with an increase in scale on a wide variety of tasks via the in-context learning paradigm. In this paper, we investigate the hypothesis that the ability of a large language model to in-context learn-perform a task is not uniformly spread across all of its underlying components. Using a 66 billion parameter language model (OPT-66B) across a diverse set of 14 downstream tasks, we find this is indeed the case: $\sim$70% of attention heads and $\sim$20% of feed forward networks can be removed with minimal decline in task performance. We find substantial overlap in the set of attention heads (un)important for in-context learning across tasks and number of in-context examples. We also address our hypothesis through a task-agnostic lens, finding that a small set of attention heads in OPT-66B score highly on their ability to perform primitive induction operations associated with in-context learning, namely, prefix matching and copying. These induction heads overlap with task-specific important heads, reinforcing arguments by Olsson et al. (arXiv:2209.11895) regarding induction head generality to more sophisticated behaviors associated with in-context learning. Overall, our study provides several insights that indicate large language models may be under-trained for in-context learning and opens up questions on how to pre-train language models to more effectively perform in-context learning.

arXiv:2212.08108공식 코드 · Papers with Code

Dataflow Analysis-Inspired Deep Learning for Efficient Vulnerability Detection

Deep learning-based vulnerability detection has shown great performance and, in some studies, outperformed static analysis tools. However, the highest-performing approaches use token-based transformer models, which are not the most efficient to capture code semantics required for vulnerability detection. Classical program analysis techniques such as dataflow analysis can detect many types of bugs based on their root causes. In this paper, we propose to combine such causal-based vulnerability detection algorithms with deep learning, aiming to achieve more efficient and effective vulnerability detection. Specifically, we designed DeepDFA, a dataflow analysis-inspired graph learning framework and an embedding technique that enables graph learning to simulate dataflow computation. We show that DeepDFA is both performant and efficient. DeepDFA outperformed all non-transformer baselines. It was trained in 9 minutes, 75x faster than the highest-performing baseline model. When using only 50+ vulnerable and several hundreds of total examples as training data, the model retained the same performance as 100% of the dataset. DeepDFA also generalized to real-world vulnerabilities in DbgBench; it detected 8.7 out of 17 vulnerabilities on average across folds and was able to distinguish between patched and buggy versions, while the highest-performing baseline models did not detect any vulnerabilities. By combining DeepDFA with a large language model, we surpassed the state-of-the-art vulnerability detection performance on the Big-Vul dataset with 96.46 F1 score, 97.82 precision, and 95.14 recall. Our replication package is located at https://doi.org/10.6084/m9.figshare.21225413 .

arXiv:2212.08061공식 코드 · Papers with Code

On Second Thought, Let's Not Think Step by Step! Bias and Toxicity in Zero-Shot Reasoning

Generating a Chain of Thought (CoT) has been shown to consistently improve large language model (LLM) performance on a wide range of NLP tasks. However, prior work has mainly focused on logical reasoning tasks (e.g. arithmetic, commonsense QA); it remains unclear whether improvements hold for more diverse types of reasoning, especially in socially situated contexts. Concretely, we perform a controlled evaluation of zero-shot CoT across two socially sensitive domains: harmful questions and stereotype benchmarks. We find that zero-shot CoT reasoning in sensitive domains significantly increases a model's likelihood to produce harmful or undesirable output, with trends holding across different prompt formats and model variants. Furthermore, we show that harmful CoTs increase with model size, but decrease with improved instruction following. Our work suggests that zero-shot CoT should be used with caution on socially important tasks, especially when marginalized groups or sensitive topics are involved.

arXiv:2212.05339공식 코드 · Papers with Code

Elixir: Train a Large Language Model on a Small GPU Cluster

In recent years, large language models have achieved great success due to their unprecedented size. However, training these models poses a challenge for most researchers as it requires a substantial number of GPUs. To reduce GPU memory usage, memory partitioning, and memory offloading have been proposed. These approaches eliminate memory redundancies and offload memory usage to the CPU and NVMe memory, respectively, enabling training on small GPU clusters. However, directly deploying these solutions often leads to suboptimal efficiency. Only experienced experts can unleash the full potential of hardware by carefully tuning the distributed configuration. Thus, we present a novel solution, Elixir, which automates efficient large-model training based on pre-runtime model profiling. Elixir aims to identify the optimal combination of partitioning and offloading techniques to maximize training throughput. In our experiments, Elixir significantly outperforms the current state-of-the-art baseline. Our optimal configuration achieves up to a 3.4$\times$ speedup on GPT-2 models compared with SOTA solutions. We hope that our work will benefit individuals who lack computing resources and expertise, granting them access to large models. The beta version of Elixir is now available at https://github.com/hpcaitech/ColossalAI/tree/feature/elixir.

arXiv:2212.03597공식 코드 · Papers with Code

DeepSpeed Data Efficiency: Improving Deep Learning Model Quality and Training Efficiency via Efficient Data Sampling and Routing

Recent advances on deep learning models come at the price of formidable training cost. The increasing model size is one of the root causes, but another less-emphasized fact is that data scale is actually increasing at a similar speed as model scale, and the training cost is proportional to both of them. Compared to the rapidly evolving model architecture, how to efficiently use the training data (especially for the expensive foundation model pretraining) is both less explored and difficult to realize due to the lack of a convenient framework that focuses on data efficiency capabilities. To this end, we present DeepSpeed Data Efficiency, a framework that makes better use of data, increases training efficiency, and improves model quality. Specifically, we propose and combine two data efficiency techniques: efficient data sampling via a general curriculum learning library, and efficient data routing via a novel random layerwise token dropping technique. For GPT-3 1.3B language model pretraining, our work achieves 12.5x less data/time/cost (\$3.7K if rent on Azure), while still maintaining 95% of model quality compared to baseline with full data and cost (\$46.3K). For GPT-3 1.3B and BERT-large pretraining, our work can also achieve the same model quality with up to 2x less data/time/cost, or achieve better model quality under same data/time/cost. DeepSpeed Data Efficiency is easy to use and tune, enabling us to easily apply it and verify its benefit on additional tasks including GPT-3 MoE model pretraining and small-scale GPT-2/ViT finetuning.

arXiv:2211.15718공식 코드 · Papers with Code

Contrastive Novelty-Augmented Learning: Anticipating Outliers with Large Language Models

In many task settings, text classification models are likely to encounter examples from novel classes on which they cannot predict correctly. Selective prediction, in which models abstain on low-confidence examples, provides a possible solution, but existing models are often overly confident on unseen classes. To remedy this overconfidence, we introduce Contrastive Novelty-Augmented Learning (CoNAL), a two-step method that generates OOD examples representative of novel classes, then trains to decrease confidence on them. First, we generate OOD examples by prompting a large language model twice: we prompt it to enumerate relevant novel classes, then generate examples from each novel class matching the task format. Second, we train a classifier with a novel contrastive objective that encourages lower confidence on generated OOD examples than training examples. When trained with CoNAL, classifiers improve in their ability to detect and abstain on novel class examples over prior methods by an average of 2.3% in terms of accuracy under the accuracy-coverage curve (AUAC) and 5.5% AUROC across 4 NLP datasets, with no cost to in-distribution accuracy.

arXiv:2211.09085공식 코드 · Papers with Code

Galactica: A Large Language Model for Science

Information overload is a major obstacle to scientific progress. The explosive growth in scientific literature and data has made it ever harder to discover useful insights in a large mass of information. Today scientific knowledge is accessed through search engines, but they are unable to organize scientific knowledge alone. In this paper we introduce Galactica: a large language model that can store, combine and reason about scientific knowledge. We train on a large scientific corpus of papers, reference material, knowledge bases and many other sources. We outperform existing models on a range of scientific tasks. On technical knowledge probes such as LaTeX equations, Galactica outperforms the latest GPT-3 by 68.2% versus 49.0%. Galactica also performs well on reasoning, outperforming Chinchilla on mathematical MMLU by 41.3% to 35.7%, and PaLM 540B on MATH with a score of 20.4% versus 8.8%. It also sets a new state-of-the-art on downstream tasks such as PubMedQA and MedMCQA dev of 77.6% and 52.9%. And despite not being trained on a general corpus, Galactica outperforms BLOOM and OPT-175B on BIG-bench. We believe these results demonstrate the potential for language models as a new interface for science. We open source the model for the benefit of the scientific community.

arXiv:2211.01334공식 코드 · Papers with Code

MemoNet: Memorizing All Cross Features' Representations Efficiently via Multi-Hash Codebook Network for CTR Prediction

New findings in natural language processing (NLP) demonstrate that the strong memorization capability contributes a lot to the success of Large Language Models (LLM). This inspires us to explicitly bring an independent memory mechanism into CTR ranking model to learn and memorize cross features' representations. In this paper, we propose multi-Hash Codebook NETwork (HCNet) as the memory mechanism for efficiently learning and memorizing representations of cross features in CTR tasks. HCNet uses a multi-hash codebook as the main memory place and the whole memory procedure consists of three phases: multi-hash addressing, memory restoring, and feature shrinking. We also propose a new CTR model named MemoNet which combines HCNet with a DNN backbone. Extensive experimental results on three public datasets and online test show that MemoNet reaches superior performance over state-of-the-art approaches. Besides, MemoNet shows scaling law of large language model in NLP, which means we can enlarge the size of the codebook in HCNet to sustainably obtain performance gains. Our work demonstrates the importance and feasibility of learning and memorizing representations of cross features, which sheds light on a new promising research direction.

arXiv:2210.15458공식 코드 · Papers with Code

Arithmetic Sampling: Parallel Diverse Decoding for Large Language Models

Decoding methods for large language models often trade-off between diversity of outputs and parallelism of computation. Methods such as beam search and Gumbel top-k sampling can guarantee a different output for each element of the beam, but are not easy to parallelize. Alternatively, methods such as temperature sampling and its modifications (top-k sampling, nucleus sampling, typical decoding, and others), are embarrassingly parallel, but have no guarantees about duplicate samples. We present a framework for sampling according to an arithmetic code book implicitly defined by a large language model, compatible with common sampling variations, with provable beam diversity under certain conditions, as well as being embarrassingly parallel and providing unbiased and consistent expectations from the original model. We demonstrate the effectiveness of our approach on WMT machine translation, more than halving the standard deviation when estimating expected BLEU score reward, and closing the BLEU score gap between independent sampling and beam search by up to 63%.