TradingMoE: Routing the Right Experts in Evolving Markets

Large language models (LLMs) have shown strong potential for financial analysis and trading, but direct trading remains challenging because the predictive capabilities required can vary across assets, decision fields, and market conditions. Existing LLM-based trading systems either coordinate human-defined external experts or adopt conventional internal Mixture-of-Experts (MoE) routers that do not directly evaluate how individual experts contribute to trading decisions. Moreover, these routers receive no direct signal indicating when an inactive expert has become more suitable as market conditions change. We find that native router scores poorly reflect how much individual experts improve trading decisions, frequently leaving better alternatives unselected. We further reveal that token-specific expert usefulness exhibits a compact low-dimensional structure. Based on these findings, we propose TradingMoE, a trading-oriented sparse MoE that augments a frozen dense LLM with lightweight residual experts. We introduce a Query-Key router that represents the expertise required by each token under the current market context as a low-dimensional query and matches it with learnable expert keys. We further propose a sparse expert selection update mechanism that samples a few inactive experts during training and estimates whether they should replace the weakest expert in the current Top-k route. This mechanism enables the router to update expert selection as market conditions change while preserving sparse computation. Experiments against 22 baselines on stock and cryptocurrency markets show that TradingMoE improves cumulative return over the best-performing baselines by 30.89% and 30.7%, respectively. Rolling paper-trading experiments further demonstrate that its advantage persists under forward-only deployment.

RecGPT-V3 Technical Report

Large language models (LLMs) are transforming recommender systems from matching co-occurrence patterns in historical behavior toward reasoning about the intent that drives it. RecGPT-V1 pioneered this paradigm on Taobao by centering user understanding, and RecGPT-V2 scaled it via coordinated multi-agent reasoning; both are deployed in production with consistent gains in user experience and commercial outcomes. However, operating RecGPT at scale reveals three challenges: (1) stateless behavior modeling, where each request reprocesses full user history, wasting computation and discarding prior analysis; (2) a tag-to-item information bottleneck, where natural-language tags form a lossy channel between user understanding and item grounding; and (3) inefficient explicit reasoning, whose lengthy chain-of-thought incurs untenable latency and compute overhead. We present RecGPT-V3, a stateful, hybrid-modal recommender that reasons over natural language for open-world knowledge and Semantic IDs (SIDs) for concrete item grounding. A Memory Hub maintains structured, continually evolving user memory that distills long-horizon behavior into condensed units, cutting user-modeling computation by 55.8%. A Hybrid-modal Foundation Model allows the LLM jointly reason over text tags and SIDs, opening a high-bandwidth channel into the item space. Latent Intent Reasoning internalizes verbose rationales into compact learnable latent tokens that remain decodable into readable explanations, lowering output token cost by 200x. Deployed in Taobao's "Guess What You Like" feed, RecGPT-V3 achieves consistent gains in large-scale online A/B tests: IPV +1.28%, CTR +1.00%, TC +1.97%, GMV +3.97%, while cutting end-to-end serving resource consumption by 52.4%.

Beyond Noisy Signals: Dual-Level Denoising for Multi-modal Sequential Recommendation

Multi-modal Sequential Recommendation (SR) incorporates rich side information (e.g., textual and visual features) to enhance dynamic user preference modeling. However, existing frameworks inevitably suffer from a \textbf{Dual-Noise Dilemma}: (1) \textit{Feature-level redundancy} stemming from the semantic gap between generic pre-trained representations and fine-grained recommendation intent; and (2) \textit{Sequence-level stochasticity} induced by spurious interactions such as accidental clicks. To break this bottleneck, we propose \textbf{DDMSR}, a novel \textbf{D}ual-level \textbf{D}enoising \textbf{M}ulti-modal \textbf{S}equential \textbf{R}ecommendation framework that systematically purifies signals from both feature-topological and sequence-frequency perspectives. Specifically, we first design a graph-based feature denoising module that leverages Laplacian smoothing on item semantic graphs as a structural low-pass filter, effectively suppressing high-frequency semantic noise while preserving salient features. For sequence purification, we introduce a frequency-domain sequence denoising module, utilizing the Fast Fourier Transform and a learnable frequency filter to adaptively modulate the interaction spectrum and attenuate anomalous signals. Furthermore, a multi-modal contrastive alignment objective is incorporated to bridge the heterogeneity gap and enforce cross-modal semantic consistency. Extensive experiments on four public benchmark datasets demonstrate that DDMSR consistently outperforms state-of-the-art baselines, providing a highly robust and efficient solution for multi-modal sequential recommendation. The source code is available at: ~\href{https://github.com/jluo00/DDMSR}{\textcolor{blue}{https://github.com/jluo00/DDMSR}}.

ShopX: A Foundation Model for Intent-to-Item Fulfillment in Agentic Shopping

The wave of AI-native applications is moving shopping beyond page- and feed-based browsing toward intent-driven experiences orchestrated by LLM agents. A common design wraps an LLM around existing search and recommendation pipelines, forcing complex intents through low-bandwidth retrieval or ranking interfaces and leaving a gap between language understanding and item-space fulfillment. Generative recommendation gives LLMs a direct item-space interface through semantic IDs (SIDs), but existing models mainly generate candidates for retrieval rather than translate flexible intents into item-space outcomes. We propose ShopX to address this bottleneck by unifying intent understanding, execution planning, and flexible SID-native item-space operations into a single foundation model. We deploy ShopX in agentic shopping workflows through a model-native item-fulfillment framework with a serving harness that defines a model-facing action protocol and exposes support surfaces for context access, catalog grounding, and state management. Within this framework, ShopX plans and composes SID-based item-space operations such as SID beam-search retrieval, listwise ranking, or product bundling. This model-centric design reduces lossy hand-offs between agent orchestration and item-space execution. To build ShopX, we design semantically recoverable, LLM-operable SIDs and a training recipe that equips a general LLM for flexible multi-turn item-space fulfillment while retaining the knowledge and instruction-following abilities needed by a shopping agent. We evaluate the ShopX framework against tool-mediated agentic systems on single- and multi-turn fulfillment tasks derived from anonymized Taobao production logs, showing that model-native fulfillment improves overall framework behavior, especially on complex or ambiguous requests.

FlashNav: Ultra-Fast Policy Training for Robot Navigation within 20 Seconds

Deep reinforcement learning has shown strong potential for robot navigation, but its practical deployment is still limited by the long wall-clock cost of policy training. This paper presents FlashNav, a GPU-first framework for ultra-fast range-based robot navigation training. To the best of our knowledge, FlashNav is the first DRL-based robot navigation framework that reaches seconds-level policy training, with the fastest deployable policy trained in less than 20 seconds. The key idea is to align simulation with the navigation MDP: FlashNav preserves the essential components for velocity-level navigation, including occupancy geometry, range sensing, goal-conditioned control, robot motion dynamics, collision handling, termination, and reset, while removing unnecessary rendering and high-fidelity physical details from the training loop. Built on a batched bitmap simulator and a fully GPU-resident training pipeline with our FastDSAC learner, FlashNav generates massive parallel navigation transitions entirely on GPU. Experiments on TurtleBot2 and Unitree Go2 show that FlashNav achieves a 100\% success-rate below 20 seconds on an RTX 5090 and remains within tens of seconds across desktop GPUs. The learned policies further transfer to physical wheeled and legged robots in static and dynamic indoor scenes, demonstrating that DRL-based navigation can be trained at seconds-level speed while preserving deployable obstacle-avoidance behavior.

Do We Really Need Immediate Resets? Rethinking Collision Handling for Efficient Robot Navigation

Should a single collision necessarily terminate an entire navigation episode? In most deep reinforcement learning (DRL) frameworks for robot navigation, this remains the standard practice: every collision immediately triggers a global environment reset and is penalized as a complete task failure. While a collision during deployment naturally indicates task failure, applying the same treatment during training prevents the agent from exploring challenging obstacle configurations, which slows learning progress in the early training phase. In this work, we challenge this convention and propose a Multi-Collision reset Budget (MCB) framework that decouples local collision termination from global environment resets, allowing the agent to retry difficult configurations within the same episode. Experiments on multiple simulated and real-world robotic platforms show that the framework accelerates early-stage exploration and improves both success rate and navigation efficiency over conventional single-collision reset baselines, with a small collision budget producing the largest gains.

Learning from Demonstration with Failure Awareness for Safe Robot Navigation

Learning from demonstration is widely used for robot navigation, yet it suffers from a fundamental limitation: demonstrations consist predominantly of successful behaviors and provide limited coverage of unsafe states. This limitation leads to poor safety when the robot encounters scenarios beyond the demonstration distribution. Failure experiences, such as collisions, contain essential information about unsafe regions, but remain underutilized. The key difficulty lies in the fact that failure data do not provide valid guidance for action imitation, and their naive incorporation into policy learning often degrades performance. We address this challenge by proposing a failure-aware learning framework that explicitly decouples the roles of success and failure data. In this framework, failure experiences are used to shape value estimation in hazardous regions, while policy learning is restricted to successful demonstrations. This separation enables the effective use of failure data without corrupting policy behavior. We implement this design within an offline reinforcement learning (RL) setting and evaluate it in both simulation and real-world environments. The results show that our framework consistently reduces collision rates while preserving the task success rate, and demonstrate strong generalization across different environments and robot platforms.

FSUNav: A Cerebrum-Cerebellum Architecture for Fast, Safe, and Universal Zero-Shot Goal-Oriented Navigation

Current vision-language navigation methods face substantial bottlenecks regarding heterogeneous robot compatibility, real-time performance, and navigation safety. Furthermore, they struggle to support open-vocabulary semantic generalization and multimodal task inputs. To address these challenges, this paper proposes FSUNav: a Cerebrum-Cerebellum architecture for fast, safe, and universal zero-shot goal-oriented navigation, which innovatively integrates vision-language models (VLMs) with the proposed architecture. The cerebellum module, a high-frequency end-to-end module, develops a universal local planner based on deep reinforcement learning, enabling unified navigation across heterogeneous platforms (e.g., humanoid, quadruped, wheeled robots) to improve navigation efficiency while significantly reducing collision risk. The cerebrum module constructs a three-layer reasoning model and leverages VLMs to build an end-to-end detection and verification mechanism, enabling zero-shot open-vocabulary goal navigation without predefined IDs and improving task success rates in both simulation and real-world environments. Additionally, the framework supports multimodal inputs (e.g., text, target descriptions, and images), further enhancing generalization, real-time performance, safety, and robustness. Experimental results on MP3D, HM3D, and OVON benchmarks demonstrate that FSUNav achieves state-of-the-art performance on object, instance image, and task navigation, significantly outperforming existing methods. Real-world deployments on diverse robotic platforms further validate its robustness and practical applicability.

CeRLP: A Cross-embodiment Robot Local Planning Framework for Visual Navigation

Visual navigation for cross-embodiment robots is challenging due to variations in robot and camera configurations, which can lead to the failure of navigation tasks. Previous approaches typically rely on collecting massive datasets across different robots, which is highly data-intensive, or fine-tuning models, which is time-consuming. Furthermore, both methods often lack explicit consideration of robot geometry. In this paper, we propose a Cross-embodiment Robot Local Planning (CeRLP) framework for general visual navigation, which abstracts visual information into a unified geometric formulation and applies to heterogeneous robots with varying physical dimensions, camera parameters, and camera types. CeRLP introduces a depth estimation scale correction method that utilizes offline pre-calibration to resolve the scale ambiguity of monocular depth estimation, thereby recovering precise metric depth images. Furthermore, CeRLP designs a visual-to-scan abstraction module that projects varying visual inputs into height-adaptive laser scans, making the policy robust to heterogeneous robots. Experiments in simulation environments demonstrate that CeRLP outperforms comparative methods, validating its robust obstacle avoidance capabilities as a local planner. Additionally, extensive real-world experiments verify the effectiveness of CeRLP in tasks such as point-to-point navigation and vision-language navigation, demonstrating its generalization across varying robot and camera configurations.

FastDSAC: Unlocking the Potential of Maximum Entropy RL in High-Dimensional Humanoid Control

Scaling Maximum Entropy Reinforcement Learning (RL) to high-dimensional humanoid control remains a formidable challenge, as the ``curse of dimensionality'' induces severe exploration inefficiency and training instability in expansive action spaces. Consequently, recent high-throughput paradigms have largely converged on deterministic policy gradients combined with massive parallel simulation. We challenge this compromise with FastDSAC, a framework that effectively unlocks the potential of maximum entropy stochastic policies for complex continuous control. We introduce Dimension-wise Entropy Modulation (DEM) to dynamically redistribute the exploration budget and enforce diversity, alongside a continuous distributional critic tailored to ensure value fidelity and mitigate high-dimensional value overestimation. Extensive evaluations on HumanoidBench and other continuous control tasks demonstrate that rigorously designed stochastic policies can consistently match or outperform deterministic baselines, achieving notable gains of 180\% and 400\% on the challenging \textit{Basketball} and \textit{Balance Hard} tasks.