Yury Malkov

Aaron Hurst, Adam Lerer, Adam P. Goucher et al. · openai

GPT-4o is an autoregressive omni model that accepts as input any combination of text, audio, image, and video, and generates any combination of text, audio, and image outputs. It's trained end-to-end across text, vision, and audio, meaning all inputs and outputs are processed by the same neural network. GPT-4o can respond to audio inputs in as little as 232 milliseconds, with an average of 320 milliseconds, which is similar to human response time in conversation. It matches GPT-4 Turbo performance on text in English and code, with significant improvement on text in non-English languages, while also being much faster and 50\% cheaper in the API. GPT-4o is especially better at vision and audio understanding compared to existing models. In line with our commitment to building AI safely and consistent with our voluntary commitments to the White House, we are sharing the GPT-4o System Card, which includes our Preparedness Framework evaluations. In this System Card, we provide a detailed look at GPT-4o's capabilities, limitations, and safety evaluations across multiple categories, focusing on speech-to-speech while also evaluating text and image capabilities, and measures we've implemented to ensure the model is safe and aligned. We also include third-party assessments on dangerous capabilities, as well as discussion of potential societal impacts of GPT-4o's text and vision capabilities.

Xinyang Zhang, Yury Malkov, Omar Florez et al. · amazon-science

Pre-trained language models (PLMs) are fundamental for natural language processing applications. Most existing PLMs are not tailored to the noisy user-generated text on social media, and the pre-training does not factor in the valuable social engagement logs available in a social network. We present TwHIN-BERT, a multilingual language model productionized at Twitter, trained on in-domain data from the popular social network. TwHIN-BERT differs from prior pre-trained language models as it is trained with not only text-based self-supervision, but also with a social objective based on the rich social engagements within a Twitter heterogeneous information network (TwHIN). Our model is trained on 7 billion tweets covering over 100 distinct languages, providing a valuable representation to model short, noisy, user-generated text. We evaluate our model on various multilingual social recommendation and semantic understanding tasks and demonstrate significant metric improvement over established pre-trained language models. We open-source TwHIN-BERT and our curated hashtag prediction and social engagement benchmark datasets to the research community.

Artem Sevastopolsky, Yury Malkov, Nikita Durasov et al.

State-of-the-art face recognition systems require vast amounts of labeled training data. Given the priority of privacy in face recognition applications, the data is limited to celebrity web crawls, which have issues such as limited numbers of identities. On the other hand, self-supervised revolution in the industry motivates research on the adaptation of related techniques to facial recognition. One of the most popular practical tricks is to augment the dataset by the samples drawn from generative models while preserving the identity. We show that a simple approach based on fine-tuning pSp encoder for StyleGAN allows us to improve upon the state-of-the-art facial recognition and performs better compared to training on synthetic face identities. We also collect large-scale unlabeled datasets with controllable ethnic constitution -- AfricanFaceSet-5M (5 million images of different people) and AsianFaceSet-3M (3 million images of different people) -- and we show that pretraining on each of them improves recognition of the respective ethnicities (as well as others), while combining all unlabeled datasets results in the biggest performance increase. Our self-supervised strategy is the most useful with limited amounts of labeled training data, which can be beneficial for more tailored face recognition tasks and when facing privacy concerns. Evaluation is based on a standard RFW dataset and a new large-scale RB-WebFace benchmark. The code and data are made publicly available at https://github.com/seva100/stylegan-for-facerec.

Gheorghe Comanici, Eric Bieber, Mike Schaekermann et al. · amazon-science, baidu

In this report, we introduce the Gemini 2.X model family: Gemini 2.5 Pro and Gemini 2.5 Flash, as well as our earlier Gemini 2.0 Flash and Flash-Lite models. Gemini 2.5 Pro is our most capable model yet, achieving SoTA performance on frontier coding and reasoning benchmarks. In addition to its incredible coding and reasoning skills, Gemini 2.5 Pro is a thinking model that excels at multimodal understanding and it is now able to process up to 3 hours of video content. Its unique combination of long context, multimodal and reasoning capabilities can be combined to unlock new agentic workflows. Gemini 2.5 Flash provides excellent reasoning abilities at a fraction of the compute and latency requirements and Gemini 2.0 Flash and Flash-Lite provide high performance at low latency and cost. Taken together, the Gemini 2.X model generation spans the full Pareto frontier of model capability vs cost, allowing users to explore the boundaries of what is possible with complex agentic problem solving.

Bilegsaikhan Naidan, Leonid Boytsov, Yury Malkov et al.

This document covers a library for fast similarity (k-NN)search. It describes only search methods and distances (spaces). Details about building, installing, Python bindings can be found online:https://github.com/searchivarius/nmslib/tree/v1.8/. Even though the library contains a variety of exact metric-space access methods, our main focus is on more generic and approximate search methods, in particular, on methods for non-metric spaces. NMSLIB is possibly the first library with a principled support for non-metric space searching.

Rasul Karimov, Yury Malkov, Karim Iskakov et al.

This work is centred around the recently proposed product key memory structure \cite{large_memory}, implemented for a number of computer vision applications. The memory structure can be regarded as a simple computation primitive suitable to be augmented to nearly all neural network architectures. The memory block allows implementing sparse access to memory with square root complexity scaling with respect to the memory capacity. The latter scaling is possible due to the incorporation of Cartesian product space decomposition of the key space for the nearest neighbour search. We have tested the memory layer on the classification, image reconstruction and relocalization problems and found that for some of those, the memory layers can provide significant speed/accuracy improvement with the high utilization of the key-value elements, while others require more careful fine-tuning and suffer from dying keys. To tackle the later problem we have introduced a simple technique of memory re-initialization which helps us to eliminate unused key-value pairs from the memory and engage them in training again. We have conducted various experiments and got improvements in speed and accuracy for classification and PoseNet relocalization models. We showed that the re-initialization has a huge impact on a toy example of randomly labeled data and observed some gains in performance on the image classification task. We have also demonstrated the generalization property perseverance of the large memory layers on the relocalization problem, while observing the spatial correlations between the images and the selected memory cells.

Aliaksandra Shysheya, Egor Zakharov, Kara-Ali Aliev et al.

We present a system for learning full-body neural avatars, i.e. deep networks that produce full-body renderings of a person for varying body pose and camera position. Our system takes the middle path between the classical graphics pipeline and the recent deep learning approaches that generate images of humans using image-to-image translation. In particular, our system estimates an explicit two-dimensional texture map of the model surface. At the same time, it abstains from explicit shape modeling in 3D. Instead, at test time, the system uses a fully-convolutional network to directly map the configuration of body feature points w.r.t. the camera to the 2D texture coordinates of individual pixels in the image frame. We show that such a system is capable of learning to generate realistic renderings while being trained on videos annotated with 3D poses and foreground masks. We also demonstrate that maintaining an explicit texture representation helps our system to achieve better generalization compared to systems that use direct image-to-image translation.

Karim Iskakov, Egor Burkov, Victor Lempitsky et al.

We present two novel solutions for multi-view 3D human pose estimation based on new learnable triangulation methods that combine 3D information from multiple 2D views. The first (baseline) solution is a basic differentiable algebraic triangulation with an addition of confidence weights estimated from the input images. The second solution is based on a novel method of volumetric aggregation from intermediate 2D backbone feature maps. The aggregated volume is then refined via 3D convolutions that produce final 3D joint heatmaps and allow modelling a human pose prior. Crucially, both approaches are end-to-end differentiable, which allows us to directly optimize the target metric. We demonstrate transferability of the solutions across datasets and considerably improve the multi-view state of the art on the Human3.6M dataset. Video demonstration, annotations and additional materials will be posted on our project page (https://saic-violet.github.io/learnable-triangulation).

Dmitry Baranchuk, Artem Babenko, Yury Malkov

This work addresses the problem of billion-scale nearest neighbor search. The state-of-the-art retrieval systems for billion-scale databases are currently based on the inverted multi-index, the recently proposed generalization of the inverted index structure. The multi-index provides a very fine-grained partition of the feature space that allows extracting concise and accurate short-lists of candidates for the search queries. In this paper, we argue that the potential of the simple inverted index was not fully exploited in previous works and advocate its usage both for the highly-entangled deep descriptors and relatively disentangled SIFT descriptors. We introduce a new retrieval system that is based on the inverted index and outperforms the multi-index by a large margin for the same memory consumption and construction complexity. For example, our system achieves the state-of-the-art recall rates several times faster on the dataset of one billion deep descriptors compared to the efficient implementation of the inverted multi-index from the FAISS library.

Leonid Boytsov, David Novak, Yury Malkov et al.

Retrieval pipelines commonly rely on a term-based search to obtain candidate records, which are subsequently re-ranked. Some candidates are missed by this approach, e.g., due to a vocabulary mismatch. We address this issue by replacing the term-based search with a generic k-NN retrieval algorithm, where a similarity function can take into account subtle term associations. While an exact brute-force k-NN search using this similarity function is slow, we demonstrate that an approximate algorithm can be nearly two orders of magnitude faster at the expense of only a small loss in accuracy. A retrieval pipeline using an approximate k-NN search can be more effective and efficient than the term-based pipeline. This opens up new possibilities for designing effective retrieval pipelines. Our software (including data-generating code) and derivative data based on the Stack Overflow collection is available online.