-
GPT-4o System Card
Authors:
OpenAI,
:,
Aaron Hurst,
Adam Lerer,
Adam P. Goucher,
Adam Perelman,
Aditya Ramesh,
Aidan Clark,
AJ Ostrow,
Akila Welihinda,
Alan Hayes,
Alec Radford,
Aleksander Mądry,
Alex Baker-Whitcomb,
Alex Beutel,
Alex Borzunov,
Alex Carney,
Alex Chow,
Alex Kirillov,
Alex Nichol,
Alex Paino,
Alex Renzin,
Alex Tachard Passos,
Alexander Kirillov,
Alexi Christakis
, et al. (395 additional authors not shown)
Abstract:
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 mil…
▽ More
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.
△ Less
Submitted 25 October, 2024;
originally announced October 2024.
-
GHM Wavelet Transform for Deep Image Super Resolution
Authors:
Ben Lowe,
Hadi Salman,
Justin Zhan
Abstract:
The GHM multi-level discrete wavelet transform is proposed as preprocessing for image super resolution with convolutional neural networks. Previous works perform analysis with the Haar wavelet only. In this work, 37 single-level wavelets are experimentally analyzed from Haar, Daubechies, Biorthogonal, Reverse Biorthogonal, Coiflets, and Symlets wavelet families. All single-level wavelets report si…
▽ More
The GHM multi-level discrete wavelet transform is proposed as preprocessing for image super resolution with convolutional neural networks. Previous works perform analysis with the Haar wavelet only. In this work, 37 single-level wavelets are experimentally analyzed from Haar, Daubechies, Biorthogonal, Reverse Biorthogonal, Coiflets, and Symlets wavelet families. All single-level wavelets report similar results indicating that the convolutional neural network is invariant to choice of wavelet in a single-level filter approach. However, the GHM multi-level wavelet achieves higher quality reconstructions than the single-level wavelets. Three large data sets are used for the experiments: DIV2K, a dataset of textures, and a dataset of satellite images. The approximate high resolution images are compared using seven objective error measurements. A convolutional neural network based approach using wavelet transformed images has good results in the literature.
△ Less
Submitted 16 April, 2022;
originally announced April 2022.
-
Enhancing Channel Shortening Based Physical Layer Security Using Coordinated Multipoint
Authors:
Muhammad Sohaib J. Solaija,
Hanadi Salman,
Huseyin Arslan
Abstract:
Wireless networks have become imperative in all areas of human life. As such, one of the most critical concerns in next-generation networks is ensuring the security and privacy of user data/communication. Cryptography has been conventionally used to tackle this, but it may not be scalable (in terms of key exchange and management) with the increasingly heterogeneous network deployments. Physical la…
▽ More
Wireless networks have become imperative in all areas of human life. As such, one of the most critical concerns in next-generation networks is ensuring the security and privacy of user data/communication. Cryptography has been conventionally used to tackle this, but it may not be scalable (in terms of key exchange and management) with the increasingly heterogeneous network deployments. Physical layer security (PLS) provides a promising alternative, but struggles when an attacker boasts a better wireless channel as compared to the legitimate user. This work leverages the coordinated multipoint concept and its distributed transmission points, in conjunction with channel shortening, to address this problem. Results show significant degradation of the bit-error-rate experienced at the eavesdropper as compared to state-of-the-art channel shortening-based PLS methods.
△ Less
Submitted 29 September, 2021;
originally announced September 2021.
-
Generalized Coordinated Multipoint Framework for 5G and Beyond
Authors:
Muhammad Sohaib J. Solaija,
Hanadi Salman,
Abuu B. Kihero,
Mehmet Izzet Saglam,
Huseyin Arslan
Abstract:
The characteristic feature of 5G is the diversity of its services for different user needs. However, the requirements for these services are competing in nature, which impresses the necessity of a coordinated and flexible network architecture. Although coordinated multipoint (CoMP) systems were primarily proposed to improve the cell edge performance in 4G, their collaborative nature can be leverag…
▽ More
The characteristic feature of 5G is the diversity of its services for different user needs. However, the requirements for these services are competing in nature, which impresses the necessity of a coordinated and flexible network architecture. Although coordinated multipoint (CoMP) systems were primarily proposed to improve the cell edge performance in 4G, their collaborative nature can be leveraged to support the diverse requirements and enabling technologies of 5G and beyond networks. To this end, we propose generalization of CoMP to a proactive and efficient resource utilization framework capable of supporting different user requirements such as reliability, latency, throughput, and security while considering network constraints. This article elaborates on the multiple aspects, inputs, and outputs of the generalized CoMP (GCoMP) framework. Apart from user requirements, the GCoMP decision mechanism also considers the CoMP scenario and network architecture to decide upon outputs such as CoMP technique or appropriate coordinating clusters. To enable easier understanding of the concept, popular use cases, such as vehicle-to-everything (V2X) communication and eHealth, are studied. Additionally, interesting challenges and open areas in GCoMP are discussed.
△ Less
Submitted 14 August, 2020;
originally announced August 2020.
-
Energy Aware Wireless System based Software Defined Radio
Authors:
H. Salman,
R. Balatiah,
A. Masri,
Y. A. S. Dama
Abstract:
Development of green telecommunication systems is already being considered highly attractive by standard bodies and recently is attracting research attention. While most of the research focuses on modeling and simulation, in this work we implement a lab setup to test an energy aware wireless system based on software defined radio and solar energy power system. In addition, we proposed an energy aw…
▽ More
Development of green telecommunication systems is already being considered highly attractive by standard bodies and recently is attracting research attention. While most of the research focuses on modeling and simulation, in this work we implement a lab setup to test an energy aware wireless system based on software defined radio and solar energy power system. In addition, we proposed an energy aware adaptive modulation algorithm that considers the state of charge of the solar energy batteries before setting up the modulation order. Moreover, the algorithm adapts to user preferences between the connectivity mode and the quality mode.
△ Less
Submitted 12 July, 2019;
originally announced July 2019.
