We introduce Phi-4-Mini and Phi-4-Multimodal, compact yet highly capable language and multimodal models. Phi-4-Mini is a 3.8-billion-parameter language model trained on high-quality web and synthetic data, significantly outperforming recent open-source models of similar size and matching the performance of models twice its size on math and coding tasks requiring complex reasoning. This achievement is driven by a carefully curated synthetic data recipe emphasizing high-quality math and coding datasets. Compared to its predecessor, Phi-3.5-Mini, Phi-4-Mini features an expanded vocabulary size of 200K tokens to better support multilingual applications, as well as group query attention for more efficient long-sequence generation. Phi-4-Multimodal is a multimodal model that integrates text, vision, and speech/audio input modalities into a single model. Its novel modality extension approach leverages LoRA adapters and modality-specific routers to allow multiple inference modes combining various modalities without interference. For example, it now ranks first in the OpenASR leaderboard to date, although the LoRA component of the speech/audio modality has just 460 million parameters. Phi-4-Multimodal supports scenarios involving (vision + language), (vision + speech), and (speech/audio) inputs, outperforming larger vision-language and speech-language models on a wide range of tasks. Additionally, we experiment to further train Phi-4-Mini to enhance its reasoning capabilities. Despite its compact 3.8-billion-parameter size, this experimental version achieves reasoning performance on par with or surpassing significantly larger models, including DeepSeek-R1-Distill-Qwen-7B and DeepSeek-R1-Distill-Llama-8B.

Large language models (LLMs) have shown limitations in tasks requiring complex logical reasoning and multi-step problem-solving. To address these challenges, researchers have employed carefully designed prompts and flowcharts, simulating human cognitive processes to enhance LLM performance, such as the Chain of Thought approach. In this paper, we introduce MTMT (Multi-thinking Modes Tree), a novel method that interacts with LLMs to construct a thought tree, simulating various advanced cognitive processes, including but not limited to association, counterfactual thinking, task decomposition, and comparison. By breaking down the original complex task into simpler sub-questions, MTMT facilitates easier problem-solving for LLMs, enabling more effective utilization of the latent knowledge within LLMs. We evaluate the performance of MTMT under different parameter configurations, using GPT-4o mini as the base model. Our results demonstrate that integrating multiple modes of thinking significantly enhances the ability of LLMs to handle complex tasks.

We introduce phi-3-mini, a 3.8 billion parameter language model trained on 3.3 trillion tokens, whose overall performance, as measured by both academic benchmarks and internal testing, rivals that of models such as Mixtral 8x7B and GPT-3.5 (e.g., phi-3-mini achieves 69% on MMLU and 8.38 on MT-bench), despite being small enough to be deployed on a phone. The innovation lies entirely in our dataset for training, a scaled-up version of the one used for phi-2, composed of heavily filtered publicly available web data and synthetic data. The model is also further aligned for robustness, safety, and chat format. We also provide some initial parameter-scaling results with a 7B and 14B models trained for 4.8T tokens, called phi-3-small and phi-3-medium, both significantly more capable than phi-3-mini (e.g., respectively 75% and 78% on MMLU, and 8.7 and 8.9 on MT-bench). Moreover, we also introduce phi-3-vision, a 4.2 billion parameter model based on phi-3-mini with strong reasoning capabilities for image and text prompts.

Ground Penetrating Radar (GPR) has been widely used to estimate the healthy operation of some urban roads and underground facilities. When identifying subsurface anomalies by GPR in an area, the obtained data could be unbalanced, and the numbers and types of possible underground anomalies could not be acknowledged in advance. In this paper, a novel method is proposed to improve the subsurface anomaly detection from GPR B-scan images. A normal (i.e. without subsurface objects) GPR image section is firstly collected in the detected area. Concerning that the GPR image is essentially the representation of electromagnetic (EM) wave and propagation time, and to preserve both the subsurface background and objects' details, the normal GPR image is segmented and then fused with simulated GPR images that contain different kinds of objects to generate the synthetic data for the detection area based on the wavelet decompositions. Pre-trained CNNs could then be fine-tuned with the synthetic data, and utilized to extract features of segmented GPR images subsequently obtained in the detection area. The extracted features could be classified by the one-class learning algorithm in the feature space without pre-set anomaly types or numbers. The conducted experiments demonstrate that fine-tuning the pre-trained CNN with the proposed synthetic data could effectively improve the feature extraction of the network for the objects in the detection area. Besides, the proposed method requires only a section of normal data that could be easily obtained in the detection area, and could also meet the timeliness requirements in practical applications.

With the rapid expansion of urban areas and the increasingly use of electricity, the need for locating buried cables is becoming urgent. In this paper, a noval method to locate underground cables based on Ground Penetrating Radar (GPR) and Gaussian-process regression is proposed. Firstly, the coordinate system of the detected area is conducted, and the input and output of locating buried cables are determined. The GPR is moved along the established parallel detection lines, and the hyperbolic signatures generated by buried cables are identified and fitted, thus the positions and depths of some points on the cable could be derived. On the basis of the established coordinate system and the derived points on the cable, the clustering method and cable fitting algorithm based on Gaussian-process regression are proposed to find the most likely locations of the underground cables. Furthermore, the confidence intervals of the cable's locations are also obtained. Both the position and depth noises are taken into account in our method, ensuring the robustness and feasibility in different environments and equipments. Experiments on real-world datasets are conducted, and the obtained results demonstrate the effectiveness of the proposed method.

Ground Penetrating Radar (GPR) is widely used as a non-destructive approach to estimate buried utilities. When the GPR's detecting direction is perpendicular to a pipeline, a hyperbolic characteristic would be formed on the GPR B-scan image. However, in real-world applications, the direction of pipelines on the existing pipeline map could be inaccurate, and it is hard to ensure the moving direction of GPR to be actually perpendicular to underground pipelines. In this paper, a novel model is proposed to estimate the direction and radius of pipeline and revise the existing pipeline map from GPR B-scan images. The model consists of two parts: GPR B-scan image processing and Ellipse Iterative Inversion Algorithm (EIIA). Firstly, the GPR B-scan image is processed with downward-opening point set extracted. The obtained point set is then iteratively inverted to the elliptical cross section of the buried pipeline, which is caused by the angle between the GPR's detecting direction and the pipeline's direction. By minimizing the sum of the algebraic distances from the extracted point set to the inverted ellipse, the most likely pipeline's direction and radius are determined. Experiments on real-world datasets are conducted, and the results demonstrate the effectiveness of the method.

Reinforcement learning competitions have formed the basis for standard research benchmarks, galvanized advances in the state-of-the-art, and shaped the direction of the field. Despite this, a majority of challenges suffer from the same fundamental problems: participant solutions to the posed challenge are usually domain-specific, biased to maximally exploit compute resources, and not guaranteed to be reproducible. In this paper, we present a new framework of competition design that promotes the development of algorithms that overcome these barriers. We propose four central mechanisms for achieving this end: submission retraining, domain randomization, desemantization through domain obfuscation, and the limitation of competition compute and environment-sample budget. To demonstrate the efficacy of this design, we proposed, organized, and ran the MineRL 2020 Competition on Sample-Efficient Reinforcement Learning. In this work, we describe the organizational outcomes of the competition and show that the resulting participant submissions are reproducible, non-specific to the competition environment, and sample/resource efficient, despite the difficult competition task.

Recently, deep reinforcement learning (DRL)-based approach has shown promise in solving complex decision and control problems in power engineering domain. In this paper, we present an in-depth analysis of DRL-based voltage control from aspects of algorithm selection, state space representation, and reward engineering. To resolve observed issues, we propose a novel imitation learning-based approach to directly map power grid operating points to effective actions without any interim reinforcement learning process. The performance results demonstrate that the proposed approach has strong generalization ability with much less training time. The agent trained by imitation learning is effective and robust to solve voltage control problem and outperforms the former RL agents.