As the scope of machine learning broadens, we observe a recurring theme of algorithmic monoculture: the same systems, or systems that share components (e.g. training data), are deployed by multiple decision-makers. While sharing offers clear advantages (e.g. amortizing costs), does it bear risks? We introduce and formalize one such risk, outcome homogenization: the extent to which particular individuals or groups experience negative outcomes from all decision-makers. If the same individuals or groups exclusively experience undesirable outcomes, this may institutionalize systemic exclusion and reinscribe social hierarchy. To relate algorithmic monoculture and outcome homogenization, we propose the component-sharing hypothesis: if decision-makers share components like training data or specific models, then they will produce more homogeneous outcomes. We test this hypothesis on algorithmic fairness benchmarks, demonstrating that sharing training data reliably exacerbates homogenization, with individual-level effects generally exceeding group-level effects. Further, given the dominant paradigm in AI of foundation models, i.e. models that can be adapted for myriad downstream tasks, we test whether model sharing homogenizes outcomes across tasks. We observe mixed results: we find that for both vision and language settings, the specific methods for adapting a foundation model significantly influence the degree of outcome homogenization. We conclude with philosophical analyses of and societal challenges for outcome homogenization, with an eye towards implications for deployed machine learning systems.
Recent multimodal models such as DALL-E and CM3 have achieved remarkable progress in text-to-image and image-to-text generation. However, these models store all learned knowledge (e.g., the appearance of the Eiffel Tower) in the model parameters, requiring increasingly larger models and training data to capture more knowledge. To integrate knowledge in a more scalable and modular way, we propose a retrieval-augmented multimodal model, which enables a base multimodal model (generator) to refer to relevant knowledge fetched by a retriever from external memory (e.g., multimodal documents on the web). Specifically, we implement a retriever using the pretrained CLIP model and a generator using the CM3 Transformer architecture, and train this model using the LAION dataset. Our resulting model, named Retrieval-Augmented CM3 (RA-CM3), is the first multimodal model that can retrieve and generate mixtures of text and images. We show that RA-CM3 significantly outperforms baseline multimodal models such as DALL-E and CM3 on both image and caption generation tasks (12 FID and 17 CIDEr improvements on MS-COCO), while requiring much less compute for training (<30% of DALL-E). Moreover, we show that RA-CM3 exhibits novel capabilities such as knowledge-intensive image generation and multimodal in-context learning.
How do author perceptions match up to the outcomes of the peer-review process and perceptions of others? In a top-tier computer science conference (NeurIPS 2021) with more than 23,000 submitting authors and 9,000 submitted papers, we survey the authors on three questions: (i) their predicted probability of acceptance for each of their papers, (ii) their perceived ranking of their own papers based on scientific contribution, and (iii) the change in their perception about their own papers after seeing the reviews. The salient results are: (1) Authors have roughly a three-fold overestimate of the acceptance probability of their papers: The median prediction is 70% for an approximately 25% acceptance rate. (2) Female authors exhibit a marginally higher (statistically significant) miscalibration than male authors; predictions of authors invited to serve as meta-reviewers or reviewers are similarly calibrated, but better than authors who were not invited to review. (3) Authors' relative ranking of scientific contribution of two submissions they made generally agree (93%) with their predicted acceptance probabilities, but there is a notable 7% responses where authors think their better paper will face a worse outcome. (4) The author-provided rankings disagreed with the peer-review decisions about a third of the time; when co-authors ranked their jointly authored papers, co-authors disagreed at a similar rate -- about a third of the time. (5) At least 30% of respondents of both accepted and rejected papers said that their perception of their own paper improved after the review process. The stakeholders in peer review should take these findings into account in setting their expectations from peer review.
Language models (LMs) are becoming the foundation for almost all major language technologies, but their capabilities, limitations, and risks are not well understood. We present Holistic Evaluation of Language Models (HELM) to improve the transparency of language models. First, we taxonomize the vast space of potential scenarios (i.e. use cases) and metrics (i.e. desiderata) that are of interest for LMs. Then we select a broad subset based on coverage and feasibility, noting what's missing or underrepresented (e.g. question answering for neglected English dialects, metrics for trustworthiness). Second, we adopt a multi-metric approach: We measure 7 metrics (accuracy, calibration, robustness, fairness, bias, toxicity, and efficiency) for each of 16 core scenarios when possible (87.5% of the time). This ensures metrics beyond accuracy don't fall to the wayside, and that trade-offs are clearly exposed. We also perform 7 targeted evaluations, based on 26 targeted scenarios, to analyze specific aspects (e.g. reasoning, disinformation). Third, we conduct a large-scale evaluation of 30 prominent language models (spanning open, limited-access, and closed models) on all 42 scenarios, 21 of which were not previously used in mainstream LM evaluation. Prior to HELM, models on average were evaluated on just 17.9% of the core HELM scenarios, with some prominent models not sharing a single scenario in common. We improve this to 96.0%: now all 30 models have been densely benchmarked on the same core scenarios and metrics under standardized conditions. Our evaluation surfaces 25 top-level findings. For full transparency, we release all raw model prompts and completions publicly for further analysis, as well as a general modular toolkit. We intend for HELM to be a living benchmark for the community, continuously updated with new scenarios, metrics, and models.
Long samples of text from neural language models can be of poor quality. Truncation sampling algorithms--like top-$p$ or top-$k$ -- address this by setting some words' probabilities to zero at each step. This work provides framing for the aim of truncation, and an improved algorithm for that aim. We propose thinking of a neural language model as a mixture of a true distribution and a smoothing distribution that avoids infinite perplexity. In this light, truncation algorithms aim to perform desmoothing, estimating a subset of the support of the true distribution. Finding a good subset is crucial: we show that top-$p$ unnecessarily truncates high-probability words, for example causing it to truncate all words but Trump for a document that starts with Donald. We introduce $\eta$-sampling, which truncates words below an entropy-dependent probability threshold. Compared to previous algorithms, $\eta$-sampling generates more plausible long English documents according to humans, is better at breaking out of repetition, and behaves more reasonably on a battery of test distributions.
Likelihood, although useful as a training loss, is a poor search objective for guiding open-ended generation from language models (LMs). Existing generation algorithms must avoid both unlikely strings, which are incoherent, and highly likely ones, which are short and repetitive. We propose contrastive decoding (CD), a more reliable search objective that returns the difference between likelihood under a large LM (called the expert, e.g. OPT-13b) and a small LM (called the amateur, e.g. OPT-125m). CD is inspired by the fact that the failures of larger LMs are even more prevalent in smaller LMs, and that this difference signals exactly which texts should be preferred. CD requires zero training, and produces higher quality text than decoding from the larger LM alone. It also generalizes across model types (OPT and GPT2) and significantly outperforms four strong decoding algorithms in automatic and human evaluations.
A common approach to transfer learning under distribution shift is to fine-tune the last few layers of a pre-trained model, preserving learned features while also adapting to the new task. This paper shows that in such settings, selectively fine-tuning a subset of layers (which we term surgical fine-tuning) matches or outperforms commonly used fine-tuning approaches. Moreover, the type of distribution shift influences which subset is more effective to tune: for example, for image corruptions, fine-tuning only the first few layers works best. We validate our findings systematically across seven real-world data tasks spanning three types of distribution shifts. Theoretically, we prove that for two-layer neural networks in an idealized setting, first-layer tuning can outperform fine-tuning all layers. Intuitively, fine-tuning more parameters on a small target dataset can cause information learned during pre-training to be forgotten, and the relevant information depends on the type of shift.
Pretraining a language model (LM) on text has been shown to help various downstream NLP tasks. Recent works show that a knowledge graph (KG) can complement text data, offering structured background knowledge that provides a useful scaffold for reasoning. However, these works are not pretrained to learn a deep fusion of the two modalities at scale, limiting the potential to acquire fully joint representations of text and KG. Here we propose DRAGON (Deep Bidirectional Language-Knowledge Graph Pretraining), a self-supervised approach to pretraining a deeply joint language-knowledge foundation model from text and KG at scale. Specifically, our model takes pairs of text segments and relevant KG subgraphs as input and bidirectionally fuses information from both modalities. We pretrain this model by unifying two self-supervised reasoning tasks, masked language modeling and KG link prediction. DRAGON outperforms existing LM and LM+KG models on diverse downstream tasks including question answering across general and biomedical domains, with +5% absolute gain on average. In particular, DRAGON achieves notable performance on complex reasoning about language and knowledge (+10% on questions involving long contexts or multi-step reasoning) and low-resource QA (+8% on OBQA and RiddleSense), and new state-of-the-art results on various BioNLP tasks. Our code and trained models are available at https://github.com/michiyasunaga/dragon.
Recent work has observed that pre-trained models have higher out-of-distribution (OOD) robustness when they are exposed to less in-distribution (ID) training data (Radford et al., 2021). In particular, zero-shot models (e.g., GPT-3 and CLIP) have higher robustness than conventionally fine-tuned models, but these robustness gains fade as zero-shot models are fine-tuned on more ID data. We study this relationship between sample efficiency and robustness -- if two models have the same ID performance, does the model trained on fewer examples (higher sample efficiency) perform better OOD (higher robustness)? Surprisingly, experiments across three tasks, 23 total ID-OOD settings, and 14 models do not reveal a consistent relationship between sample efficiency and robustness -- while models with higher sample efficiency are sometimes more robust, most often there is no change in robustness, with some cases even showing decreased robustness. Since results vary on a case-by-case basis, we conduct detailed case studies of two particular ID-OOD pairs (SST-2 -> IMDb sentiment and SNLI -> HANS) to better understand why better sample efficiency may or may not yield higher robustness; attaining such an understanding requires case-by-case analysis of why models are not robust on a particular ID-OOD setting and how modeling techniques affect model capabilities.
Despite the empirical successes of self-supervised learning (SSL) methods, it is unclear what characteristics of their representations lead to high downstream accuracies. In this work, we characterize properties that SSL representations should ideally satisfy. Specifically, we prove necessary and sufficient conditions such that for any task invariant to given data augmentations, desired probes (e.g., linear or MLP) trained on that representation attain perfect accuracy. These requirements lead to a unifying conceptual framework for improving existing SSL methods and deriving new ones. For contrastive learning, our framework prescribes simple but significant improvements to previous methods such as using asymmetric projection heads. For non-contrastive learning, we use our framework to derive a simple and novel objective. Our resulting SSL algorithms outperform baselines on standard benchmarks, including SwAV+multicrops on linear probing of ImageNet.