Large language models (LLMs) have been shown to be able to perform new tasks based on a few demonstrations or natural language instructions. While these capabilities have led to widespread adoption, most LLMs are developed by resource-rich organizations and are frequently kept from the public. As a step towards democratizing this powerful technology, we present BLOOM, a 176B-parameter open-access language model designed and built thanks to a collaboration of hundreds of researchers. BLOOM is a decoder-only Transformer language model that was trained on the ROOTS corpus, a dataset comprising hundreds of sources in 46 natural and 13 programming languages (59 in total). We find that BLOOM achieves competitive performance on a wide variety of benchmarks, with stronger results after undergoing multitask prompted finetuning. To facilitate future research and applications using LLMs, we publicly release our models and code under the Responsible AI License.

The crystallization of modeling methods around the Transformer architecture has been a boon for practitioners. Simple, well-motivated architectural variations can transfer across tasks and scale, increasing the impact of modeling research. However, with the emergence of state-of-the-art 100B+ parameters models, large language models are increasingly expensive to accurately design and train. Notably, it can be difficult to evaluate how modeling decisions may impact emergent capabilities, given that these capabilities arise mainly from sheer scale alone. In the process of building BLOOM--the Big Science Large Open-science Open-access Multilingual language model--our goal is to identify an architecture and training setup that makes the best use of our 1,000,000 A100-GPU-hours budget. Specifically, we perform an ablation study at the billion-parameter scale comparing different modeling practices and their impact on zero-shot generalization. In addition, we study the impact of various popular pre-training corpora on zero-shot generalization. We also study the performance of a multilingual model and how it compares to the English-only one. Finally, we consider the scaling behaviour of Transformers to choose the target model size, shape, and training setup. All our models and code are open-sourced at https://huggingface.co/bigscience .

现在，可以使用最先进的神经语言模型通过零射门提示来解决临时语言任务，而无需进行监督培训。近年来，这种方法已广受欢迎，研究人员证明了提示在特定的NLP任务上实现强烈准确的提示。但是，找到新任务的提示需要实验。具有不同措辞选择的不同提示模板会导致明显的准确性差异。提示允许用户尝试及时变化，可视化及时性能，并迭代优化提示。我们开发了一个工作流程，该工作流程允许用户首先使用少量数据专注于模型反馈，然后再进入大型数据制度，该数据制度允许使用任务的定量度量来实现有希望的提示的经验基础。然后，该工具可以轻松部署新创建的临时模型。我们使用多种现实世界用例演示了Fackide（http://prompt.vizhub.ai）和我们的工作流程的实用性。

最近已被证明大型语言模型在各种任务集中获得合理的零射普通化（Brown等，2020）。它已经假设这是语言模型的隐式多任务学习的结果，在语言模型中的预押（Radford等，2019）。可以通过明确的多任务学习直接引起零拍常规化？为了以缩放测试这个问题，我们开发一个系统，以便轻松地将任何自然语言任务映射到人类可读的提示表单中。我们转换一组大量的监督数据集，每个数据集都有多个提示，具有不同的措辞。这些提示的数据集允许基准测试模型执行完全看不见的任务的能力。我们介绍了一个普拉克尔编码器 - 解码器模型（Raffel等，2020; Lester等，2021），覆盖各种任务。该模型在多个标准数据集中达到强大的零点性能，通常优于其尺寸的型号超过16倍。此外，我们的方法对来自Big-替补基准测试的任务子集具有强烈性能，优于其尺寸的6倍。所有提示和培训的型号都可以在https://github.com/ bigscience-workshop / protectsource / httpsource / https：//huggingface.co/bigscience/t0pp。

Recent progress in natural language processing has been driven by advances in both model architecture and model pretraining. Transformer architectures have facilitated building higher-capacity models and pretraining has made it possible to effectively utilize this capacity for a wide variety of tasks. Transformers is an open-source library with the goal of opening up these advances to the wider machine learning community. The library consists of carefully engineered stateof-the art Transformer architectures under a unified API. Backing this library is a curated collection of pretrained models made by and available for the community. Transformers is designed to be extensible by researchers, simple for practitioners, and fast and robust in industrial deployments. The library is available at https://github.com/ huggingface/transformers.

As Transfer Learning from large-scale pre-trained models becomes more prevalent in Natural Language Processing (NLP), operating these large models in on-theedge and/or under constrained computational training or inference budgets remains challenging. In this work, we propose a method to pre-train a smaller generalpurpose language representation model, called DistilBERT, which can then be finetuned with good performances on a wide range of tasks like its larger counterparts. While most prior work investigated the use of distillation for building task-specific models, we leverage knowledge distillation during the pre-training phase and show that it is possible to reduce the size of a BERT model by 40%, while retaining 97% of its language understanding capabilities and being 60% faster. To leverage the inductive biases learned by larger models during pre-training, we introduce a triple loss combining language modeling, distillation and cosine-distance losses. Our smaller, faster and lighter model is cheaper to pre-train and we demonstrate its capabilities for on-device computations in a proof-of-concept experiment and a comparative on-device study.

Selecting the number of topics in LDA models is considered to be a difficult task, for which alternative approaches have been proposed. The performance of the recently developed singular Bayesian information criterion (sBIC) is evaluated and compared to the performance of alternative model selection criteria. The sBIC is a generalization of the standard BIC that can be implemented to singular statistical models. The comparison is based on Monte Carlo simulations and carried out for several alternative settings, varying with respect to the number of topics, the number of documents and the size of documents in the corpora. Performance is measured using different criteria which take into account the correct number of topics, but also whether the relevant topics from the DGPs are identified. Practical recommendations for LDA model selection in applications are derived.

Applying deep learning concepts from image detection and graph theory has greatly advanced protein-ligand binding affinity prediction, a challenge with enormous ramifications for both drug discovery and protein engineering. We build upon these advances by designing a novel deep learning architecture consisting of a 3-dimensional convolutional neural network utilizing channel-wise attention and two graph convolutional networks utilizing attention-based aggregation of node features. HAC-Net (Hybrid Attention-Based Convolutional Neural Network) obtains state-of-the-art results on the PDBbind v.2016 core set, the most widely recognized benchmark in the field. We extensively assess the generalizability of our model using multiple train-test splits, each of which maximizes differences between either protein structures, protein sequences, or ligand extended-connectivity fingerprints. Furthermore, we perform 10-fold cross-validation with a similarity cutoff between SMILES strings of ligands in the training and test sets, and also evaluate the performance of HAC-Net on lower-quality data. We envision that this model can be extended to a broad range of supervised learning problems related to structure-based biomolecular property prediction. All of our software is available as open source at https://github.com/gregory-kyro/HAC-Net/.

Counterfactual explanation is a common class of methods to make local explanations of machine learning decisions. For a given instance, these methods aim to find the smallest modification of feature values that changes the predicted decision made by a machine learning model. One of the challenges of counterfactual explanation is the efficient generation of realistic counterfactuals. To address this challenge, we propose VCNet-Variational Counter Net-a model architecture that combines a predictor and a counterfactual generator that are jointly trained, for regression or classification tasks. VCNet is able to both generate predictions, and to generate counterfactual explanations without having to solve another minimisation problem. Our contribution is the generation of counterfactuals that are close to the distribution of the predicted class. This is done by learning a variational autoencoder conditionally to the output of the predictor in a join-training fashion. We present an empirical evaluation on tabular datasets and across several interpretability metrics. The results are competitive with the state-of-the-art method.

Despite their impressive performance on diverse tasks, large language models (LMs) still struggle with tasks requiring rich world knowledge, implying the limitations of relying solely on their parameters to encode a wealth of world knowledge. This paper aims to understand LMs' strengths and limitations in memorizing factual knowledge, by conducting large-scale knowledge probing experiments of 10 models and 4 augmentation methods on PopQA, our new open-domain QA dataset with 14k questions. We find that LMs struggle with less popular factual knowledge, and that scaling fails to appreciably improve memorization of factual knowledge in the tail. We then show that retrieval-augmented LMs largely outperform orders of magnitude larger LMs, while unassisted LMs remain competitive in questions about high-popularity entities. Based on those findings, we devise a simple, yet effective, method for powerful and efficient retrieval-augmented LMs, which retrieves non-parametric memories only when necessary. Experimental results show that this significantly improves models' performance while reducing the inference costs.