Oleg Rybakov
YOU?
Author Swipe
View article: The Mystery of Power and the Spirit of the Law - A Philosophical and Legal Analysis: Review of the Book by I.A. Isaev “Historical Metaphysics of Power and Law: Return to the Origins: Essays on Philosophy of History and Philosophy of Law”
The Mystery of Power and the Spirit of the Law - A Philosophical and Legal Analysis: Review of the Book by I.A. Isaev “Historical Metaphysics of Power and Law: Return to the Origins: Essays on Philosophy of History and Philosophy of Law” Open
-
View article: Pretraining Large Language Models with NVFP4
Pretraining Large Language Models with NVFP4 Open
Large Language Models (LLMs) today are powerful problem solvers across many domains, and they continue to get stronger as they scale in model size, training set size, and training set quality, as shown by extensive research and experimenta…
View article: Family and Demographic Policy in the Vector of Ensuring National Security of Russia
Family and Demographic Policy in the Vector of Ensuring National Security of Russia Open
Based on the analysis of the main normative documents defining the directions of ensuring national security and development of Russia, the authors substantiate the need to form a strategic program and target document in this area, integrat…
View article: Methods of improving LLM training stability
Methods of improving LLM training stability Open
Training stability of large language models(LLMs) is an important research topic. Reproducing training instabilities can be costly, so we use a small language model with 830M parameters and experiment with higher learning rates to force mo…
View article: Immanuel Kant’s Philosophy through the Prism of Time: on the 300th Anniversary of the Philosopher’s Birth
Immanuel Kant’s Philosophy through the Prism of Time: on the 300th Anniversary of the Philosopher’s Birth Open
-
View article: Overview of the International Scientific and Practical Conference «The Philosophy of Immanuel Kant and Modernity. To the 190th Birthday of the Philosopher»
Overview of the International Scientific and Practical Conference «The Philosophy of Immanuel Kant and Modernity. To the 190th Birthday of the Philosopher» Open
.
View article: USM RNN-T model weights binarization
USM RNN-T model weights binarization Open
Large-scale universal speech models (USM) are already used in production. However, as the model size grows, the serving cost grows too. Serving cost of large models is dominated by model size that is why model size reduction is an importan…
View article: SimulTron: On-Device Simultaneous Speech to Speech Translation
SimulTron: On-Device Simultaneous Speech to Speech Translation Open
Simultaneous speech-to-speech translation (S2ST) holds the promise of breaking down communication barriers and enabling fluid conversations across languages. However, achieving accurate, real-time translation through mobile devices remains…
View article: Human Rights: From the Origins of Philosophical and Legal Thought to Modern Geopolitical Reality
Human Rights: From the Origins of Philosophical and Legal Thought to Modern Geopolitical Reality Open
The paper presents the author’s understanding of the unity of the logical and historical connection between the processes of formation of the institution of human rights and the formation of the idea of human rights in philosophical and le…
View article: USM-Lite: Quantization and Sparsity Aware Fine-tuning for Speech Recognition with Universal Speech Models
USM-Lite: Quantization and Sparsity Aware Fine-tuning for Speech Recognition with Universal Speech Models Open
End-to-end automatic speech recognition (ASR) models have seen revolutionary quality gains with the recent development of large-scale universal speech models (USM). However, deploying these massive USMs is extremely expensive due to the en…
View article: 2-bit Conformer quantization for automatic speech recognition
2-bit Conformer quantization for automatic speech recognition Open
Large speech models are rapidly gaining traction in research community. As a result, model compression has become an important topic, so that these models can fit in memory and be served with reduced cost. Practical approaches for compress…
View article: The Need for the Development of a Constitutional and Legal Model for Ensuring the Information Security of Minors
The Need for the Development of a Constitutional and Legal Model for Ensuring the Information Security of Minors Open
Based on the analysis of the main risks of the information environment for minors, the authors substantiate the need to form a constitutional and legal model for ensuring the information security of minors. The authors represent their unde…
View article: STEP: Learning N:M Structured Sparsity Masks from Scratch with Precondition
STEP: Learning N:M Structured Sparsity Masks from Scratch with Precondition Open
Recent innovations on hardware (e.g. Nvidia A100) have motivated learning N:M structured sparsity masks from scratch for fast model inference. However, state-of-the-art learning recipes in this regime (e.g. SR-STE) are proposed for non-ada…
View article: Streaming Parrotron for on-device speech-to-speech conversion
Streaming Parrotron for on-device speech-to-speech conversion Open
We present a fully on-device streaming Speech2Speech conversion model that normalizes a given input speech directly to synthesized output speech. Deploying such a model on mobile devices pose significant challenges in terms of memory footp…
View article: 4-bit Conformer with Native Quantization Aware Training for Speech Recognition
4-bit Conformer with Native Quantization Aware Training for Speech Recognition Open
Reducing the latency and model size has always been a significant research problem for live Automatic Speech Recognition (ASR) application scenarios. Along this direction, model quantization has become an increasingly popular approach to c…
View article: A Scalable Model Specialization Framework for Training and Inference using Submodels and its Application to Speech Model Personalization
A Scalable Model Specialization Framework for Training and Inference using Submodels and its Application to Speech Model Personalization Open
Model fine-tuning and adaptation have become a common approach for model specialization for downstream tasks or domains. Fine-tuning the entire model or a subset of the parameters using light-weight adaptation has shown considerable succes…
View article: Real time spectrogram inversion on mobile phone
Real time spectrogram inversion on mobile phone Open
We present two methods of real time magnitude spectrogram inversion: streaming Griffin Lim(GL) and streaming MelGAN. We demonstrate the impact of looking ahead on perceptual quality of MelGAN. As little as one hop size (12.5ms) of lookahea…
View article: The Symbolic Subject in the Optics of Symbolic State Power
The Symbolic Subject in the Optics of Symbolic State Power Open
Political subjectivity today takes on a symbolic or digital form.The symbolic subject is gradually replacing the physical subject from the political sphere.The state power in its activity essentially relies on symbolic politics, an element…