Search

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.

In this paper, an approach to training and evaluating an adapter model for the popular language model "zephyr-7b-beta" is described. The adapter was developed to improve the performance of the base model in tasks related to programming and understanding the Russian language. Considering the high quality of the original model in tasks in the English language, the goal of the research was to expand its linguistic and technical spectrum. The proposed adapter was trained using a large and diverse dataset, including question-answer pairs related to programming, as well code-related texts in Russian language. The applied training methodology ensures an improvement in the model's quality of answers in understanding and generating Python code based on Russian instructions. We evaluated the performance of the base model with the installed adapter using various metrics, comparing it to the base model as well as other state-of-the-art models in this field. The obtained results showed significant improvement, both in tasks related to writing Python code and in processing the Russian language, confirming the effectiveness of the proposed adapter., Comment: 10 pages, 4 figures

We present here a new geometry for laser amplifiers based on bulk gain media. The overlapped seed and pump beams are repetitively refocused into the gain medium with a Herriott-type multipass cell. Similar to a waveguide, this configuration allows for a confined propagation inside the gain medium over much longer lengths than in ordinary single pass bulk amplifiers. Inside the gain medium, the foci appear at separate locations. A proof-of-principle demonstration with Ti:sapphire indicates that this could lead to higher amplification due to a distribution of the thermal load., Comment: to be published in Optics Express

In quantum mechanics, randomness is postulated as a separate axiom. De Broglie's theory allows one to reproduce quantum phenomena from completely deterministic formalism. But the question of the quantum randomness emergency in the de Broglie-Bohm theory needs special attention. In the work [G. Tastevin, F. Lalo\"e, Comptes Rendus. Physique, 2021, 22, 1, pp. 99-116], it was shown that it arises as a result of the device microscopic state influence on the measurement result. In our work, we investigate the genesis of the quantum randomness in the de Broglie's theory in more details. Namely, we investigate the target system and the device behaviour in the decoherence process and model the measurement of canonical-conjugate observables. We propose a thought experiment which tests the opportunity of the information transition using beable-parameters violating the uncertainty relation. We show that in the measurement process, the strong stochastic fluctuations of beable parameters arise randomising the system in accordance with the uncertainty relation. Nevertheless, we find anomalous models of measurement in which these fluctuations can be neglected. These special models require further investigation., Comment: 8 pages (2 columns), 5 figures

We report an amplification-free thin-disk laser oscillatory system delivering 0.9 GW peak power. The 120 fs pulses at 14 MHz containing 12.8 $\mu$J delivered by thin-disk oscillator were compressed by factor 15 down to 8.0 fs with 148 W average output power and 82 % overall efficiency. Additionally, we showed that even a sub-two-cycle operation with 6.2 fs can be reached with this technology. The system will be a crucial part of the XUV frequency comb being developed and a unique high-repetition rate driver for attosecond pulse generation., Comment: 4 pages, 6 figures, 28 short-styled references. arXiv admin note: text overlap with arXiv:2211.00362

The properties of kagome metals are governed by the interdependence of band topology and electronic correlations resulting in remarkably rich phase diagrams. Here, we study the temperature evolution of the bulk electronic structure of the antiferromagnetic kagome metal FeGe using infrared spectroscopy. We uncover drastic changes in the low-energy interband absorption at the 100 K structural phase transition that has been linked to a charge-density-wave (CDW) instability. We explain this effect by the minuscule Fe displacement in the kagome plane, which results in parallel bands in the vicinity of the Fermi level. In contrast to conventional CDW materials, however, the spectral weight shifts to low energies, ruling out the opening of a CDW gap in FeGe.

Faraday isolators are usually limited to Faraday materials with strong Verdet constants. We present a method to reach the 45{\deg} polarization rotation angle needed for optical isolators with materials exhibiting a weak Faraday effect. The Faraday effect is enhanced by passing the incident radiation multiple times through the Faraday medium while the rotation angle accumulates after each pass. Materials having excellent thermos-optical properties in the ultraviolet and mid-infrared range become available for optical isolators. Herriott-type multipass cells offer a simple and compact way to realize the desired propagation length in usual optical materials of standard sizes. A proof-of-principle experiment was carried out, demonstrating polarization rotation of a 532 nm laser beam by an angle of 45{\deg} in anti-reflection-coated fused silica surrounded by a standard neodymium ring magnet.

The magneto-optical properties of (001)-oriented NbAs single crystals have been studied in the spectral range from 5 to 150 meV and in magnetic fields of up to 13 T. A rich spectrum of inter-Landau-level transitions is revealed by these measurements. The transitions follow a square-root-like dependence with magnetic field, but the simple linear-band approximation is unable to accurately reproduce the observed behavior of the transitions in applied fields. We argue that the detected magneto-optical spectra should be related to crossing hyperbolic bands, which form the W1 cones. We propose a model Hamiltonian, which describes coupled hyperbolic bands and reproduces the shape of the relevant bands in NbAs. The magneto-optical spectra computed from this Hamiltonian nicely reproduce our observations. We conclude that the hyperbolic-band approach is a minimal model to adequately describe the magneto-optical response of NbAs and that the chiral (conical) bands do not explicitly manifest themselves in the spectra., Comment: 10 pages, SM included

We measured the magneto-reflectivity spectra (4 - 90 meV, 0 - 16 T) of the triple-point semimetal GdPtBi and found them to demonstrate two unusual broad features emerging in field. The electronic bands of GdPtBi are expected to experience large exchange-mediated shifts, which lends itself to a description via effective Zeeman splittings with a large g factor. Based on this approach, along with an ab initio band structure analysis, we propose a model Hamiltonian that describes our observations well and allows us to estimate the effective g factor, g* = 95. We conclude that we directly observe the exchange-induced $\Gamma_{8}$ band inversion in GdPtBi by means of infrared spectroscopy., Comment: 9 pages, SM included

Scholars have not asked why so many governments created ad hoc scientific advisory bodies (ahSABs) to address the Covid-19 pandemic instead of relying on existing public health infrastructure. We address this neglected question with an exploratory study of the US, UK, Sweden, Italy, Poland, and Uganda. Drawing on our case studies and the blame-avoidance literature, we find that ahSABs are created to excuse unpopular policies and take the blame should things go wrong. Thus, membership typically represents a narrow range of perspectives. An ahSAB is a good scapegoat because it does little to reduce government discretion and has limited ability to deflect blame back to government. Our explanation of our deviant case of Sweden, that did not create and ahSAB, reinforces our general principles. We draw the policy inference that ahSAB membership should be vetted by the legislature to ensure broad membership., Comment: 2 tables, 1 figure

This paper describes a method for using Grovers algorithm to create a quantum vector database, the database stores embeddings based on Controlled-S gates, which represent a binary numerical value. This value represents the embeddings value. The process of creating meaningful embeddings is handled by a classical computer and the search process is handled by the quantum computer. This search approach might be beneficial for a large enough database, or it could be seen as a very qubit-efficient (super dense) way for storing data on a quantum computer, since the proposed circuit stores many embeddings inside one quantum register simultaneously., Comment: 12 pages, 11 figures, Accepted as part of publication "Environmental, Social, and Corporate Governance Perspectives in Development of Quantum Vector Databases Based on Grover's Algorithm" in IEEE Conference "2023 Intelligent technologies and electronic devices in vehicle and road transport complex (TIRVED-2023)"

The diverging relaxation time in approaching hypothetical ideal glass transition is a subject of hot debate. In the current paper we demonstrate, how diverging relaxation time and turning excess entropy to zero (which is an essence of Kauzmann's paradox) can be avoided, using as an example the model molecular glassformer, propylene carbonate. For this purpose we compare its thermodynamic and dielectric relaxation properties, both known from the literature. The agreement between two sets of data can be achieved, if we suppose, that enthalpy of supercooled liquid propylene carbonate is governed by activation law, and relaxation time follows double exponential law. We propose the generalized Adam-Gibbs law to reconcile this two dependencies, and qualitatively discuss its implications.

Since its first demonstration in 2016, the multi-pass spectral broadening technique has covered impressive ranges of pulse energy (3 $\mu$J - 100 mJ) and peak power (4 MW - 100 GW). Energy scaling of this technique into the joule-level is currently limited by phenomena such as optical damage, gas ionization and spatio-spectral beam inhomogeneity. These limitations can be overcome by the novel multi-pass convex-concave arrangement, which exhibits crucial properties such as large mode size and compactness. In a proof-of-principle experiment, 260 fs, 15 $\mu$J and 200 $\mu$J pulses are broadened and subsequently compressed to approximately 50 fs with 90 % efficiency and excellent spatio-spectral homogeneity across the beam profile.

We report supercontinuum generation and pulse compression in two stacked multipass cells based on dielectric mirrors. The 230 fs pulses at 1 MHz containing 12 $\mu$J were compressed by factor 33 down to 7 fs, corresponding to 1.0 GW peak power and overall transmission of 84 %. The source is particularly interesting for such applications as time-resolved ARPES, photoemission electron microscopy, and nonlinear spectroscopy., Comment: 4 journal pages, 6 figures

This paper showcases a method of parametric synthesis of quantum circuits for training perceptron neural networks. Synapse weights are found using Grover's algorithm with a modified oracle function. The results of running these parametrically synthesized circuits for training perceptrons of three different topologies are described. The circuits were run on a 100-qubit IBM quantum simulator. The synthesis of quantum circuits is carried out using quantum synthesizer "Naginata", which was developed in the scope of this work, the source code of which is published and further documented on GitHub. The article describes the quantum circuit synthesis algorithm for training single-layer perceptrons. At the moment, quantum circuits are created mainly by manually placing logic elements on lines that symbolize quantum bits. The purpose of creating Quantum Circuit Synthesizer "Naginata" was due to the fact that even with a slight increase in the number of operations in a quantum algorithm, leads to the significant increase in size of the corresponding quantum circuit. This causes serious difficulties both in creating and debugging these quantum circuits. The purpose of our quantum synthesizer is enabling users an opportunity to implement quantum algorithms using higher-level commands. This is achieved by creating generic blocks for frequently used operations such as: the adder, multiplier, digital comparator (comparison operator), etc. Thus, the user could implement a quantum algorithm by using these generic blocks, and the quantum synthesizer would create a suitable circuit for this algorithm, in a format that is supported by the chosen quantum computation environment. This approach greatly simplifies the processes of development and debugging a quantum algorithm., Comment: 12 pages, 8 figures

At the moment, quantum circuits are created mainly by manually placing logic elements on lines that symbolize quantum bits. The purpose of creating Quantum Circuit Synthesizer "Naginata" was due to the fact that even with a slight increase in the number of operations in a quantum algorithm, leads to the significant increase in size of the corresponding quantum circuit. This causes serious difficulties both in creating and debugging these quantum circuits. The purpose of our quantum synthesizer is enabling users an opportunity to implement quantum algorithms using higher-level commands. This is achieved by creating generic blocks for frequently used operations such as: the adder, multiplier, digital comparator (comparison operator), etc. Thus, the user could implement a quantum algorithm by using these generic blocks, and the quantum synthesizer would create a suitable circuit for this algorithm, in a format that is supported by the chosen quantum computation environment. This approach greatly simplifies the processes of development and debugging a quantum algorithm. The proposed approach for implementing quantum algorithms has a potential application in the field of machine learning, in this regard, we provided an example of creating a circuit for training a simple neural network. Neural networks have a significant impact on the technological development of the transport and road complex, and there is a potential for improving the reliability and efficiency of their learning process by utilizing quantum computation, through the introduction of quantum computing., Comment: 11 pages, 8 figures. arXiv admin note: text overlap with arXiv:2209.09496

We report a new approach to phase matching of nonlinear materials based on the free space multipass cells. This concept quasi-phase matches crystalline quartz and increases the second harmonic generation efficiency by a factor 40.