Towards Effective Experiential Learning: Dual Guidance for Utilization and Internalization

Recently, reinforcement learning~(RL) has become an important approach for improving the capabilities of large language models~(LLMs). In particular, reinforcement learning from verifiable rewards~(RLVR) has emerged as a promising paradigm for reasoning tasks. However, existing RL-based training still remains only a rough approximation to human learning. Human learners leverage both external and internal experience to guide exploration and gradually internalize useful trajectories into stable knowledge. Motivated by this gap, we ask: how can LLMs better utilize and internalize experience during RLVR training? To answer this question, we propose \textbf{D}ual \textbf{G}uidance \textbf{O}ptimization~(\textbf{DGO}), a unified framework that leverages \emph{external} and \emph{internal experience} to improve training effectiveness. Specifically, DGO first constructs an experience bank from previously explored trajectories. The policy then performs exploration under the joint guidance of the experience bank and the model's internal knowledge. The resulting trajectories are further used to refine the experience bank and optimize model parameters, forming a closed loop of experience utilization and internalization. Experiments show that DGO consistently outperforms baseline methods, suggesting that better utilization and internalization of experience lead to more effective reasoning.

IQuest-Coder-V1 Technical Report

In this report, we introduce the IQuest-Coder-V1 series-(7B/14B/40B/40B-Loop), a new family of code large language models (LLMs). Moving beyond static code representations, we propose the code-flow multi-stage training paradigm, which captures the dynamic evolution of software logic through different phases of the pipeline. Our models are developed through the evolutionary pipeline, starting with the initial pre-training consisting of code facts, repository, and completion data. Following that, we implement a specialized mid-training stage that integrates reasoning and agentic trajectories in 32k-context and repository-scale in 128k-context to forge deep logical foundations. The models are then finalized with post-training of specialized coding capabilities, which is bifurcated into two specialized paths: the thinking path (utilizing reasoning-driven RL) and the instruct path (optimized for general assistance). IQuest-Coder-V1 achieves state-of-the-art performance among competitive models across critical dimensions of code intelligence: agentic software engineering, competitive programming, and complex tool use. To address deployment constraints, the IQuest-Coder-V1-Loop variant introduces a recurrent mechanism designed to optimize the trade-off between model capacity and deployment footprint, offering an architecturally enhanced path for efficacy-efficiency trade-off. We believe the release of the IQuest-Coder-V1 series, including the complete white-box chain of checkpoints from pre-training bases to the final thinking and instruction models, will advance research in autonomous code intelligence and real-world agentic systems.

InCoder-32B: Code Foundation Model for Industrial Scenarios

Recent code large language models have achieved remarkable progress on general programming tasks. Nevertheless, their performance degrades significantly in industrial scenarios that require reasoning about hardware semantics, specialized language constructs, and strict resource constraints. To address these challenges, we introduce InCoder-32B (Industrial-Coder-32B), the first 32B-parameter code foundation model unifying code intelligence across chip design, GPU kernel optimization, embedded systems, compiler optimization, and 3D modeling. By adopting an efficient architecture, we train InCoder-32B from scratch with general code pre-training, curated industrial code annealing, mid-training that progressively extends context from 8K to 128K tokens with synthetic industrial reasoning data, and post-training with execution-grounded verification. We conduct extensive evaluation on 14 mainstream general code benchmarks and 9 industrial benchmarks spanning 4 specialized domains. Results show InCoder-32B achieves highly competitive performance on general tasks while establishing strong open-source baselines across industrial domains.

SFedHIFI: Fire Rate-Based Heterogeneous Information Fusion for Spiking Federated Learning

Spiking Federated Learning (SFL) has been widely studied with the energy efficiency of Spiking Neural Networks (SNNs). However, existing SFL methods require model homogeneity and assume all clients have sufficient computational resources, resulting in the exclusion of some resource-constrained clients. To address the prevalent system heterogeneity in real-world scenarios, enabling heterogeneous SFL systems that allow clients to adaptively deploy models of different scales based on their local resources is crucial. To this end, we introduce SFedHIFI, a novel Spiking Federated Learning framework with Fire Rate-Based Heterogeneous Information Fusion. Specifically, SFedHIFI employs channel-wise matrix decomposition to deploy SNN models of adaptive complexity on clients with heterogeneous resources. Building on this, the proposed heterogeneous information fusion module enables cross-scale aggregation among models of different widths, thereby enhancing the utilization of diverse local knowledge. Extensive experiments on three public benchmarks demonstrate that SFedHIFI can effectively enable heterogeneous SFL, consistently outperforming all three baseline methods. Compared with ANN-based FL, it achieves significant energy savings with only a marginal trade-off in accuracy.

FPNet: Joint Wi-Fi Beamforming Matrix Feedback and Anomaly-Aware Indoor Positioning

Channel State Information (CSI) provides a detailed description of the wireless channel and has been widely adopted for Wi-Fi sensing, particularly for high-precision indoor positioning. However, complete CSI is rarely available in real-world deployments due to hardware constraints and the high communication overhead required for feedback. Moreover, existing positioning models lack mechanisms to detect when users move outside their trained regions, leading to unreliable estimates in dynamic environments. In this paper, we present FPNet, a unified deep learning framework that jointly addresses channel feedback compression, accurate indoor positioning, and robust anomaly detection (AD). FPNet leverages the beamforming feedback matrix (BFM), a compressed CSI representation natively supported by IEEE 802.11ac/ax/be protocols, to minimize feedback overhead while preserving critical positioning features. To enhance reliability, we integrate ADBlock, a lightweight AD module trained on normal BFM samples, which identifies out-of-distribution scenarios when users exit predefined spatial regions. Experimental results using standard 2.4 GHz Wi-Fi hardware show that FPNet achieves positioning accuracy above 97% with only 100 feedback bits, boosts net throughput by up to 22.92%, and attains AD accuracy over 99% with a false alarm rate below 1.5%. These results demonstrate FPNet's ability to deliver efficient, accurate, and reliable indoor positioning on commodity Wi-Fi devices.

Non-volatile Programmable Photonic Integrated Circuits using Mechanically Latched MEMS: A System-Level Scheme Enabling Power-Connection-Free Operation Without Performance Compromise

Programmable photonic integrated circuits (PPICs) offer a versatile platform for implementing diverse optical functions on a generic hardware mesh. However, the scalability of PPICs faces critical power consumption barriers. Therefore, we propose a novel non-volatile PPIC architecture utilizing MEMS with mechanical latching, enabling stable passive operation without any power connection once configured. To ensure practical applicability, we present a system-level solution including both this hardware innovation and an accompanying automatic error-resilient configuration algorithm. The algorithm compensates for the lack of continuous tunability inherent in the non-volatile hardware design, thereby enabling such new operational paradigm without compromising performance, and also ensuring robustness against fabrication errors. Functional simulations were performed to validate the proposed scheme by configuring five distinct functionalities of varying complexity, including a Mach-Zehnder interferometer (MZI), a MZI lattice filter, a ring resonator (ORR), a double ORR ring-loaded MZI, and a triple ORR coupled resonator waveguide filter. The results demonstrate that our non-volatile scheme achieves performance equivalent to conventional PPICs. Robustness analysis was also conducted, and the results demonstrated that our scheme exhibits strong robustness against various fabrication errors. Furthermore, we explored the trade-off between the hardware design complexity of such non-volatile scheme and its performance. This study establishes a viable pathway to a new generation of power-connection-free PPICs, providing a practical and scalable solution for future photonic systems.

Universal Reasoning Model

Universal transformers (UTs) have been widely used for complex reasoning tasks such as ARC-AGI and Sudoku, yet the specific sources of their performance gains remain underexplored. In this work, we systematically analyze UTs variants and show that improvements on ARC-AGI primarily arise from the recurrent inductive bias and strong nonlinear components of Transformer, rather than from elaborate architectural designs. Motivated by this finding, we propose the Universal Reasoning Model (URM), which enhances the UT with short convolution and truncated backpropagation. Our approach substantially improves reasoning performance, achieving state-of-the-art 53.8% pass@1 on ARC-AGI 1 and 16.0% pass@1 on ARC-AGI 2. Our code is avaliable at https://github.com/UbiquantAI/URM.

Agent Bain vs. Agent McKinsey: A New Text-to-SQL Benchmark for the Business Domain

In the business domain, where data-driven decision making is crucial, text-to-SQL is fundamental for easy natural language access to structured data. While recent LLMs have achieved strong performance in code generation, existing text-to-SQL benchmarks remain focused on factual retrieval of past records. We introduce CORGI, a new benchmark specifically designed for real-world business contexts. CORGI is composed of synthetic databases inspired by enterprises such as Doordash, Airbnb, and Lululemon. It provides questions across four increasingly complex categories of business queries: descriptive, explanatory, predictive, and recommendational. This challenge calls for causal reasoning, temporal forecasting, and strategic recommendation, reflecting multi-level and multi-step agentic intelligence. We find that LLM performance drops on high-level questions, struggling to make accurate predictions and offer actionable plans. Based on execution success rate, the CORGI benchmark is about 21\% more difficult than the BIRD benchmark. This highlights the gap between popular LLMs and the need for real-world business intelligence. We release a public dataset and evaluation framework, and a website for public submissions.

VMDNet: Time Series Forecasting with Leakage-Free Samplewise Variational Mode Decomposition and Multibranch Decoding

In time series forecasting, capturing recurrent temporal patterns is essential; decomposition techniques make such structure explicit and thereby improve predictive performance. Variational Mode Decomposition (VMD) is a powerful signal-processing method for periodicity-aware decomposition and has seen growing adoption in recent years. However, existing studies often suffer from information leakage and rely on inappropriate hyperparameter tuning. To address these issues, we propose VMDNet, a causality-preserving framework that (i) applies sample-wise VMD to avoid leakage; (ii) represents each decomposed mode with frequency-aware embeddings and decodes it using parallel temporal convolutional networks (TCNs), ensuring mode independence and efficient learning; and (iii) introduces a bilevel, Stackelberg-inspired optimisation to adaptively select VMD's two core hyperparameters: the number of modes (K) and the bandwidth penalty (alpha). Experiments on two energy-related datasets demonstrate that VMDNet achieves state-of-the-art results when periodicity is strong, showing clear advantages in capturing structured periodic patterns while remaining robust under weak periodicity.

On Minimization/Maximization of the Generalized Multi-Order Complex Quadratic Form With Constant-Modulus Constraints

In this paper, we study the generalized problem that minimizes or maximizes a multi-order complex quadratic form with constant-modulus constraints on all elements of its optimization variable. Such a mathematical problem is commonly encountered in various applications of signal processing. We term it as the constant-modulus multi-order complex quadratic programming (CMCQP) in this paper. In general, the CMCQP is non-convex and difficult to solve. Its objective function typically relates to metrics such as signal-to-noise ratio, Cram\'er-Rao bound, integrated sidelobe level, etc., and constraints normally correspond to requirements on similarity to desired aspects, peak-to-average-power ratio, or constant-modulus property in practical scenarios. In order to find efficient solutions to the CMCQP, we first reformulate it into an unconstrained optimization problem with respect to phase values of the studied variable only. Then, we devise a steepest descent/ascent method with fast determinations on its optimal step sizes. Specifically, we convert the step-size searching problem into a polynomial form that leads to closed-form solutions of high accuracy, wherein the third-order Taylor expansion of the search function is conducted. Our major contributions also lie in investigating the effect of the order and specific form of matrices embedded in the CMCQP, for which two representative cases are identified. Examples of related applications associated with the two cases are also provided for completeness. The proposed methods are summarized into algorithms, whose convergence speeds are verified to be fast by comprehensive simulations and comparisons to existing methods. The accuracy of our proposed fast step-size determination is also evaluated.