Although scaling language models improves performance on a range of tasks, there are apparently some scenarios where scaling hurts performance. For instance, the Inverse Scaling Prize Round 1 identified four ''inverse scaling'' tasks, for which performance gets worse for larger models. These tasks were evaluated on models of up to 280B parameters, trained up to 500 zettaFLOPs of compute. This paper takes a closer look at these four tasks. We evaluate models of up to 540B parameters, trained on five times more compute than those evaluated in the Inverse Scaling Prize. With this increased range of model sizes and training compute, three out of the four tasks exhibit what we call ''U-shaped scaling'' -- performance decreases up to a certain model size, and then increases again up to the largest model evaluated. One hypothesis is that U-shaped scaling occurs when a task comprises a ''true task'' and a ''distractor task''. Medium-size models can do the distractor task, which hurts performance, while only large-enough models can ignore the distractor task and do the true task. The existence of U-shaped scaling implies that inverse scaling may not hold for larger models. Second, we evaluate the inverse scaling tasks using chain-of-thought (CoT) prompting, in addition to basic prompting without CoT. With CoT prompting, all four tasks show either U-shaped scaling or positive scaling, achieving perfect solve rates on two tasks and several sub-tasks. This suggests that the term "inverse scaling task" is under-specified -- a given task may be inverse scaling for one prompt but positive or U-shaped scaling for a different prompt.

Finetuning language models on a collection of datasets phrased as instructions has been shown to improve model performance and generalization to unseen tasks. In this paper we explore instruction finetuning with a particular focus on (1) scaling the number of tasks, (2) scaling the model size, and (3) finetuning on chain-of-thought data. We find that instruction finetuning with the above aspects dramatically improves performance on a variety of model classes (PaLM, T5, U-PaLM), prompting setups (zero-shot, few-shot, CoT), and evaluation benchmarks (MMLU, BBH, TyDiQA, MGSM, open-ended generation). For instance, Flan-PaLM 540B instruction-finetuned on 1.8K tasks outperforms PALM 540B by a large margin (+9.4% on average). Flan-PaLM 540B achieves state-of-the-art performance on several benchmarks, such as 75.2% on five-shot MMLU. We also publicly release Flan-T5 checkpoints, which achieve strong few-shot performance even compared to much larger models, such as PaLM 62B. Overall, instruction finetuning is a general method for improving the performance and usability of pretrained language models.

人们普遍认为，对于准确的语义细分，必须使用昂贵的操作（例如，非常卷积）结合使用昂贵的操作（例如非常卷积），从而导致缓慢的速度和大量的内存使用。在本文中，我们质疑这种信念，并证明既不需要高度的内部决议也不是必需的卷积。我们的直觉是，尽管分割是一个每像素的密集预测任务，但每个像素的语义通常都取决于附近的邻居和遥远的环境。因此，更强大的多尺度功能融合网络起着至关重要的作用。在此直觉之后，我们重新访问常规的多尺度特征空间（通常限制为P5），并将其扩展到更丰富的空间，最小的P9，其中最小的功能仅为输入大小的1/512，因此具有很大的功能接受场。为了处理如此丰富的功能空间，我们利用最近的BIFPN融合了多尺度功能。基于这些见解，我们开发了一个简化的分割模型，称为ESEG，该模型既没有内部分辨率高，也没有昂贵的严重卷积。也许令人惊讶的是，与多个数据集相比，我们的简单方法可以以比以前的艺术更快地实现更高的准确性。在实时设置中，ESEG-Lite-S在189 fps的CityScapes [12]上达到76.0％MIOU，表现优于更快的[9]（73.1％MIOU时为170 fps）。我们的ESEG-LITE-L以79 fps的速度运行，达到80.1％MIOU，在很大程度上缩小了实时和高性能分割模型之间的差距。

我们在变压器中重新审视设计选择，并提出方法来解决它们在处理长序列中的弱点。首先，我们提出了一个名为“门控注意单元”的简单层，该层允许使用较弱的单头注意，而质量损失最小。然后，我们提出了一种与该新层的线性近似方法互补的，该方法对加速器友好且质量高度竞争。最终的型号（名为Flash）与短（512）和长（8K）上下文长度相匹配，在WIKI-40B上达到高达4.9 $ \ times $的训练速度和PG上的12.1 $ \ times $，在PG上达到了4.9 $ \ times $的困惑。-19用于自动回归语言建模，C4的4.8 $ \ times $用于掩盖语言建模。

具有更多数据，计算和参数的缩放语言模型在自然语言处理方面取得了重大进展。例如，由于缩放，GPT-3能够在内心学习任务上实现强烈结果。但是，培训这些大密度模型需要大量的计算资源。在本文中，我们提出并开发了名为Glam（通用语言模型）的语言模型系列，它使用稀疏激活的专家架构来规模模型容量，同时与致密变体相比，也产生显着更少的训练成本。最大的Glam具有1.2万亿参数，比GPT-3大约为7倍。它仅消耗了用于训练GPT-3的1/3的能量，并且需要一半的计算拖鞋进行推理，同时仍然在29个NLP任务中实现更好的整体零射击和一次性性能。

我们提出了一种称为基本的组合缩放方法，可在ImageNet ILSVRC-2012验证集上实现85.7％的前1个零点精度，超越了最佳发布的零拍模型 - 剪辑并对齐 - 达9.3％。我们的基本模式还显示出鲁棒性基准的显着改进。例如，在5个测试集中，具有自然分布换档，如想象的 - {A，R，V2，素描}和ObjectNet，我们的车型实现了83.7％的前1个平均精度，只有一个小幅度从其原始的想象精度下降。为实现这些结果，我们扩大了剪辑的对比学习框架，并在三个方面对齐：数据大小，型号大小和批量大小。我们的数据集具有6.6B噪声图像文本对，比对齐的4倍，比夹子大16倍。我们最大的型号具有3B重量，参数比为3.75倍，拖鞋比对齐和夹子更大。我们的批量尺寸为65536，比剪辑的2倍，4倍超过对齐。缩放的主要挑战是我们的加速器的内存有限，如GPU和TPU。因此，我们提出了一种在线渐变缓存的简单方法来克服这个限制。

我们总结了使用巨大的自动语音识别（ASR）模型的大量努力的结果，该模型使用包含大约一百万小时音频的大型，多样的未标记数据集进行了预训练。我们发现，即使对于拥有数万个小时的标记数据的非常大的任务，预训练，自我培训和扩大模型大小的组合也大大提高了数据效率。特别是，在具有34K小时标记数据的ASR任务上，通过微调80亿个参数预先训练的构象异构体模型，我们可以匹配最先进的（SOTA）性能（SOTA）的性能，只有3％的培训数据和通过完整的训练集可以显着改善SOTA。我们还报告了从使用大型预训练和自我训练的模型来完成一系列下游任务所获得的普遍利益，这些任务涵盖了广泛的语音域，并涵盖了多个数据集大小的大小，包括在许多人中获得SOTA性能公共基准。此外，我们利用预先训练的网络的学会表示，在非ASR任务上实现SOTA结果。

本文探讨了提高语言模型的零次学习能力的简单方法。我们表明，指令调整 - 通过对说明书中所述的任务集合微调语言模型 - 大幅提升零射门上看不见任务中的表现。我们采取预训练的语言模型和指令调整它通过自然语言指令模板语言表达了60NLP任务137B参数。我们评估这种指令调整模型，我们称之为FLAN，在看不见的任务类型。FLAN显着改善其未修饰的对应的性能和超过25的20个任务，我们评估零射门175BGPT-3。FLAN甚至GPT-3通过在安利，RTE，BoolQ，AI2-ARC，OpenbookQA和StoryCloze大比分胜过几拍。消融研究显示任务和模型的规模，这个数字是指令调整取得成功的关键组成部分。

Transformers have attracted increasing interests in computer vision, but they still fall behind state-of-the-art convolutional networks. In this work, we show that while Transformers tend to have larger model capacity, their generalization can be worse than convolutional networks due to the lack of the right inductive bias. To effectively combine the strengths from both architectures, we present CoAtNets (pronounced "coat" nets), a family of hybrid models built from two key insights:(1) depthwise Convolution and self-Attention can be naturally unified via simple relative attention; (2) vertically stacking convolution layers and attention layers in a principled way is surprisingly effective in improving generalization, capacity and efficiency. Experiments show that our CoAtNets achieve state-of-the-art performance under different resource constraints across various datasets: Without extra data, CoAtNet achieves 86.0% ImageNet top-1 accuracy; When pre-trained with 13M images from ImageNet-21K, our CoAtNet achieves 88.56% top-1 accuracy, matching ViT-huge pre-trained with 300M images from JFT-300M while using 23x less data; Notably, when we further scale up CoAtNet with JFT-3B, it achieves 90.88% top-1 accuracy on ImageNet, establishing a new state-of-the-art result.1 The initial projection stage can be seen as an aggressive down-sampling convolutional stem.

This paper introduces EfficientNetV2, a new family of convolutional networks that have faster training speed and better parameter efficiency than previous models. To develop these models, we use a combination of training-aware neural architecture search and scaling, to jointly optimize training speed and parameter efficiency. The models were searched from the search space enriched with new ops such as Fused-MBConv. Our experiments show that EfficientNetV2 models train much faster than state-of-the-art models while being up to 6.8x smaller.Our training can be further sped up by progressively increasing the image size during training, but it often causes a drop in accuracy. To compensate for this accuracy drop, we propose an improved method of progressive learning, which adaptively adjusts regularization (e.g. data augmentation) along with image size.With progressive learning, our EfficientNetV2 significantly outperforms previous models on Im-ageNet and CIFAR/Cars/Flowers datasets. By pretraining on the same ImageNet21k, our Effi-cientNetV2 achieves 87.3% top-1 accuracy on ImageNet ILSVRC2012, outperforming the recent ViT by 2.0% accuracy while training 5x-11x faster using the same computing resources. Code is available at https://github.com/google/ automl/tree/master/efficientnetv2.