Multi-modal retrieval is an important problem for many applications, such as recommendation and search. Current benchmarks and even datasets are often manually constructed and consist of mostly clean samples where all modalities are well-correlated with the content. Thus, current video-text retrieval literature largely focuses on video titles or audio transcripts, while ignoring user comments, since users often tend to discuss topics only vaguely related to the video. Despite the ubiquity of user comments online, there is currently no multi-modal representation learning datasets that includes comments. In this paper, we a) introduce a new dataset of videos, titles and comments; b) present an attention-based mechanism that allows the model to learn from sometimes irrelevant data such as comments; c) show that by using comments, our method is able to learn better, more contextualised, representations for image, video and audio representations.

In 2020 The Guardian published an article that had been written by AI. It was about the increasing use of AI in journalism, and how it is changing the landscape of the industry. It discussed how AI is being used to generate news stories, and how it is being used to help reporters with their work. It was so natural that it was hard to believe that it was written by a software called GPT-3 developed by OpenAI, a research company. The Guardian isn't the only news organization using algorithms to write articles.

I think there is little time left before someone builds AGI (median 2030). Once upon a time, I didn't think this. This post attempts to walk through some of the observations and insights that collapsed my estimates. A single invocation of GPT-3, or any large transformer, cannot run any algorithm internally that does not run in constant time complexity, because the model itself runs in constant time. It's a very large constant, but it is still a constant. They don't have any learnable memory about their internal state from previous invocations. They just have the input stream. Despite all their capability, transformers are fundamentally limited.[1] This is part of the reason why asking GPT-3 to do integer division on large numbers in one shot doesn't work. GPT-3 is big enough to memorize a number of results, so adding small numbers isn't too hard even without fine tuning. And GPT-3 is big enough to encode a finite number of unrolled steps for more complex algorithms, so in principle, fine tuning it on a bunch of arithmetic could get you better performance on somewhat more complex tasks. But no matter how much retraining you do, so long as you keep GPT-3's architecture the same, you will be able to find some arithmetic problem it can't do in one step because the numbers involved would require too many internal steps. So, with that kind of limitation, obviously transformers fail to do basic tasks like checking whether a set of parentheses are balanced... Oh wait, GPT-3 was just writing dialogue for a character that didn't know how to balance parentheses, and then wrote the human's side of the dialogue correcting that character's error. And it writes stories with a little assistance with long-run consistency. And it can generate functioning code. Some of this is already productized. This is an architecture that is provably incapable of internally dividing large integers, and it can handle a variety of difficult tasks that come uncomfortably close to human intuition. Could the kind of intelligence we care about be algorithmically simpler than integer division? This can't be literally true, if we want to include integer division as something a generally intelligent agent can do. But it sure looks like tractable constant time token predictors already capture a bunch of what we often call intelligence, even when those same systems can't divide! I'm raising my eyebrows right now to emphasize it!

Structured knowledge grounding (SKG) leverages structured knowledge to complete user requests, such as semantic parsing over databases and question answering over knowledge bases. Since the inputs and outputs of SKG tasks are heterogeneous, they have been studied separately by different communities, which limits systematic and compatible research on SKG. In this paper, we overcome this limitation by proposing the UnifiedSKG framework, which unifies 21 SKG tasks into a text-to-text format, aiming to promote systematic SKG research, instead of being exclusive to a single task, domain, or dataset. We use UnifiedSKG to benchmark T5 with different sizes and show that T5, with simple modifications when necessary, achieves state-of-the-art performance on almost all of the 21 tasks. We further demonstrate that multi-task prefix-tuning improves the performance on most tasks, largely improving the overall performance. UnifiedSKG also facilitates the investigation of zero-shot and few-shot learning, and we show that T0, GPT-3, and Codex struggle in zero-shot and few-shot learning for SKG. We also use UnifiedSKG to conduct a series of controlled experiments on structured knowledge encoding variants across SKG tasks. UnifiedSKG is easily extensible to more tasks, and it is open-sourced at https://github.com/hkunlp/unifiedskg.

We propose a new paradigm for zero-shot learners that is format agnostic, i.e., it is compatible with any format and applicable to a list of language tasks, such as text classification, commonsense reasoning, coreference resolution, and sentiment analysis. Zero-shot learning aims to train a model on a given task such that it can address new learning tasks without any additional training. Our approach converts zero-shot learning into multiple-choice tasks, avoiding problems in commonly used large-scale generative models such as FLAN. It not only adds generalization ability to models but also significantly reduces the number of parameters. Our method shares the merits of efficient training and deployment. Our approach shows state-of-the-art performance on several benchmarks and produces satisfactory results on tasks such as natural language inference and text classification. Our model achieves this success with only 235M parameters, which is substantially smaller than state-of-the-art models with billions of parameters. The code and pre-trained models are available at https://github.com/IDEA-CCNL/Fengshenbang-LM .

Understanding what constitutes safe text is an important issue in natural language processing and can often prevent the deployment of models deemed harmful and unsafe. One such type of safety that has been scarcely studied is commonsense physical safety, i.e. text that is not explicitly violent and requires additional commonsense knowledge to comprehend that it leads to physical harm. We create the first benchmark dataset, SafeText, comprising real-life scenarios with paired safe and physically unsafe pieces of advice. We utilize SafeText to empirically study commonsense physical safety across various models designed for text generation and commonsense reasoning tasks. We find that state-of-the-art large language models are susceptible to the generation of unsafe text and have difficulty rejecting unsafe advice. As a result, we argue for further studies of safety and the assessment of commonsense physical safety in models before release.

Existing vision-text contrastive learning like CLIP aims to match the paired image and caption embeddings while pushing others apart, which improves representation transferability and supports zero-shot prediction. However, medical image-text datasets are orders of magnitude below the general images and captions from the internet. Moreover, previous methods encounter many false negatives, i.e., images and reports from separate patients probably carry the same semantics but are wrongly treated as negatives. In this paper, we decouple images and texts for multimodal contrastive learning thus scaling the usable training data in a combinatorial magnitude with low cost. We also propose to replace the InfoNCE loss with semantic matching loss based on medical knowledge to eliminate false negatives in contrastive learning. We prove that MedCLIP is a simple yet effective framework: it outperforms state-of-the-art methods on zero-shot prediction, supervised classification, and image-text retrieval. Surprisingly, we observe that with only 20K pre-training data, MedCLIP wins over the state-of-the-art method (using around 200K data). Our code is available at https://github.com/RyanWangZf/MedCLIP.

Existing zero-shot cross-lingual transfer methods rely on parallel corpora or bilingual dictionaries, which are expensive and impractical for low-resource languages. To disengage from these dependencies, researchers have explored training multilingual models on English-only resources and transferring them to low-resource languages. However, its effect is limited by the gap between embedding clusters of different languages. To address this issue, we propose Embedding-Push, Attention-Pull, and Robust targets to transfer English embeddings to virtual multilingual embeddings without semantic loss, thereby improving cross-lingual transferability. Experimental results on mBERT and XLM-R demonstrate that our method significantly outperforms previous works on the zero-shot cross-lingual text classification task and can obtain a better multilingual alignment.

During Tesla's AI Day event, CEO Elon Musk unveiled a robot that is "intended to be friendly". Musk has been one of the most prominent figures to warn that AI is a "danger to the public" and potentially the "biggest risk we face as a civilisation". In 2017, he even said there was just a "five to 10 percent chance of success [of making AI safe]". Speaking about London-based DeepMind in a New York Times interview last year, Musk said: "Just the nature of the AI that they're building is one that crushes all humans at all games.

BIG-Bench (Srivastava et al., 2022) is a diverse evaluation suite that focuses on tasks believed to be beyond the capabilities of current language models. Language models have already made good progress on this benchmark, with the best model in the BIG-Bench paper outperforming average reported human-rater results on 65% of the BIG-Bench tasks via few-shot prompting. But on what tasks do language models fall short of average human-rater performance, and are those tasks actually unsolvable by current language models? In this work, we focus on a suite of 23 challenging BIG-Bench tasks which we call BIG-Bench Hard (BBH). These are the task for which prior language model evaluations did not outperform the average human-rater. We find that applying chain-of-thought (CoT) prompting to BBH tasks enables PaLM to surpass the average human-rater performance on 10 of the 23 tasks, and Codex (code-davinci-002) to surpass the average human-rater performance on 17 of the 23 tasks. Since many tasks in BBH require multi-step reasoning, few-shot prompting without CoT, as done in the BIG-Bench evaluations (Srivastava et al., 2022), substantially underestimates the best performance and capabilities of language models, which is better captured via CoT prompting. As further analysis, we explore the interaction between CoT and model scale on BBH, finding that CoT enables emergent task performance on several BBH tasks with otherwise flat scaling curves.