Improving Long-Range Navigation with Spatially-Enhanced Recurrent Memory via End-to-End Reinforcement Learning

Recent advancements in robot navigation, especially with end-to-end learning approaches like reinforcement learning (RL), have shown remarkable efficiency and effectiveness. Yet, successful navigation still relies on two key capabilities: mapping and planning, whether explicit or implicit. Classical approaches use explicit mapping pipelines to register ego-centric observations into a coherent map frame for the planner. In contrast, end-to-end learning achieves this implicitly, often through recurrent neural networks (RNNs) that fuse current and past observations into a latent space for planning. While architectures such as LSTM and GRU capture temporal dependencies, our findings reveal a key limitation: their inability to perform effective spatial memorization. This skill is essential for transforming and integrating sequential observations from varying perspectives to build spatial representations that support downstream planning. To address this, we propose Spatially-Enhanced Recurrent Units (SRUs), a simple yet effective modification to existing RNNs, designed to enhance spatial memorization capabilities. We introduce an attention-based architecture with SRUs, enabling long-range navigation using a single forward-facing stereo camera. Regularization techniques are employed to ensure robust end-to-end recurrent training via RL. Experimental results show our approach improves long-range navigation by 23.5% compared to existing RNNs. Furthermore, with SRU memory, our method outperforms the RL baseline with explicit mapping and memory modules, achieving a 29.6% improvement in diverse environments requiring long-horizon mapping and memorization. Finally, we address the sim-to-real gap by leveraging large-scale pretraining on synthetic depth data, enabling zero-shot transfer to diverse and complex real-world environments.

Vid2Sim: Generalizable, Video-based Reconstruction of Appearance, Geometry and Physics for Mesh-free Simulation

Faithfully reconstructing textured shapes and physical properties from videos presents an intriguing yet challenging problem. Significant efforts have been dedicated to advancing such a system identification problem in this area. Previous methods often rely on heavy optimization pipelines with a differentiable simulator and renderer to estimate physical parameters. However, these approaches frequently necessitate extensive hyperparameter tuning for each scene and involve a costly optimization process, which limits both their practicality and generalizability. In this work, we propose a novel framework, Vid2Sim, a generalizable video-based approach for recovering geometry and physical properties through a mesh-free reduced simulation based on Linear Blend Skinning (LBS), offering high computational efficiency and versatile representation capability. Specifically, Vid2Sim first reconstructs the observed configuration of the physical system from video using a feed-forward neural network trained to capture physical world knowledge. A lightweight optimization pipeline then refines the estimated appearance, geometry, and physical properties to closely align with video observations within just a few minutes. Additionally, after the reconstruction, Vid2Sim enables high-quality, mesh-free simulation with high efficiency. Extensive experiments demonstrate that our method achieves superior accuracy and efficiency in reconstructing geometry and physical properties from video data.

Learning-Augmented Hierarchical Clustering

Hierarchical clustering (HC) is an important data analysis technique in which the goal is to recursively partition a dataset into a tree-like structure while grouping together similar data points at each level of granularity. Unfortunately, for many of the proposed HC objectives, there exist strong barriers to approximation algorithms with the hardness of approximation. Thus, we consider the problem of hierarchical clustering given auxiliary information from natural oracles. Specifically, we focus on a *splitting oracle* which, when provided with a triplet of vertices $(u,v,w)$, answers (possibly erroneously) the pairs of vertices whose lowest common ancestor includes all three vertices in an optimal tree, i.e., identifying which vertex ``splits away'' from the others. Using such an oracle, we obtain the following results: - A polynomial-time algorithm that outputs a hierarchical clustering tree with $O(1)$-approximation to the Dasgupta objective (Dasgupta [STOC'16]). - A near-linear time algorithm that outputs a hierarchical clustering tree with $(1-o(1))$-approximation to the Moseley-Wang objective (Moseley and Wang [NeurIPS'17]). Under the plausible Small Set Expansion Hypothesis, no polynomial-time algorithm can achieve any constant approximation for Dasgupta's objective or $(1-C)$-approximation for the Moseley-Wang objective for some constant $C>0$. As such, our results demonstrate that the splitting oracle enables algorithms to outperform standard HC approaches and overcome hardness constraints. Furthermore, our approaches extend to sublinear settings, in which we show new streaming and PRAM algorithms for HC with improved guarantees.

Unsupervised Post-Training for Multi-Modal LLM Reasoning via GRPO

Improving Multi-modal Large Language Models (MLLMs) in the post-training stage typically relies on supervised fine-tuning (SFT) or reinforcement learning (RL). However, these supervised methods require expensive and manually annotated multi-modal data--an ultimately unsustainable resource. While recent efforts have explored unsupervised post-training, their methods are complex and difficult to iterate. In this work, we are the first to investigate the use of GRPO, a stable and scalable online RL algorithm, for enabling continual self-improvement without any external supervision. We propose MM-UPT, a simple yet effective framework for unsupervised post-training of MLLMs. MM-UPT builds upon GRPO, replacing traditional reward signals with a self-rewarding mechanism based on majority voting over multiple sampled responses. Our experiments demonstrate that MM-UPT significantly improves the reasoning ability of Qwen2.5-VL-7B (e.g., 66.3 %$\rightarrow$72.9 % on MathVista, 62.9 %$\rightarrow$68.7 % on We-Math), using standard dataset without ground truth labels. MM-UPT also outperforms prior unsupervised baselines and even approaches the results of supervised GRPO. Furthermore, we show that incorporating synthetic questions, generated solely by MLLM itself, can boost performance as well, highlighting a promising approach for scalable self-improvement. Overall, MM-UPT offers a new paradigm for continual, autonomous enhancement of MLLMs in the absence of external supervision. Our code is available at https://github.com/waltonfuture/MM-UPT.

Towards Efficient Key-Value Cache Management for Prefix Prefilling in LLM Inference

The increasing adoption of large language models (LLMs) with extended context windows necessitates efficient Key-Value Cache (KVC) management to optimize inference performance. Inference workloads like Retrieval-Augmented Generation (RAG) and agents exhibit high cache reusability, making efficient caching critical to reducing redundancy and improving speed. We analyze real-world KVC access patterns using publicly available traces and evaluate commercial key-value stores like Redis and state-of-the-art RDMA-based systems (CHIME [1] and Sherman [2]) for KVC metadata management. Our work demonstrates the lack of tailored storage solution for KVC prefilling, underscores the need for an efficient distributed caching system with optimized metadata management for LLM workloads, and provides insights into designing improved KVC management systems for scalable, low-latency inference.

Advancing Multimodal Reasoning via Reinforcement Learning with Cold Start

Recent advancements in large language models (LLMs) have demonstrated impressive chain-of-thought reasoning capabilities, with reinforcement learning (RL) playing a crucial role in this progress. While "aha moment" patterns--where models exhibit self-correction through reflection--are often attributed to emergent properties from RL, we first demonstrate that these patterns exist in multimodal LLMs (MLLMs) prior to RL training but may not necessarily correlate with improved reasoning performance. Building on these insights, we present a comprehensive study on enhancing multimodal reasoning through a two-stage approach: (1) supervised fine-tuning (SFT) as a cold start with structured chain-of-thought reasoning patterns, followed by (2) reinforcement learning via GRPO to further refine these capabilities. Our extensive experiments show that this combined approach consistently outperforms both SFT-only and RL-only methods across challenging multimodal reasoning benchmarks. The resulting models achieve state-of-the-art performance among open-source MLLMs at both 3B and 7B scales, with our 7B model showing substantial improvements over base models (e.g., 66.3 %$\rightarrow$73.4 % on MathVista, 62.9 %$\rightarrow$70.4 % on We-Math) and our 3B model achieving performance competitive with several 7B models. Overall, this work provides practical guidance for building advanced multimodal reasoning models. Our code is available at https://github.com/waltonfuture/RL-with-Cold-Start.

MetaBox-v2: A Unified Benchmark Platform for Meta-Black-Box Optimization

Meta-Black-Box Optimization (MetaBBO) streamlines the automation of optimization algorithm design through meta-learning. It typically employs a bi-level structure: the meta-level policy undergoes meta-training to reduce the manual effort required in developing algorithms for low-level optimization tasks. The original MetaBox (2023) provided the first open-source framework for reinforcement learning-based single-objective MetaBBO. However, its relatively narrow scope no longer keep pace with the swift advancement in this field. In this paper, we introduce MetaBox-v2 (https://github.com/MetaEvo/MetaBox) as a milestone upgrade with four novel features: 1) a unified architecture supporting RL, evolutionary, and gradient-based approaches, by which we reproduce 23 up-to-date baselines; 2) efficient parallelization schemes, which reduce the training/testing time by 10-40x; 3) a comprehensive benchmark suite of 18 synthetic/realistic tasks (1900+ instances) spanning single-objective, multi-objective, multi-model, and multi-task optimization scenarios; 4) plentiful and extensible interfaces for custom analysis/visualization and integrating to external optimization tools/benchmarks. To show the utility of MetaBox-v2, we carry out a systematic case study that evaluates the built-in baselines in terms of the optimization performance, generalization ability and learning efficiency. Valuable insights are concluded from thorough and detailed analysis for practitioners and those new to the field.

Time Travel is Cheating: Going Live with DeepFund for Real-Time Fund Investment Benchmarking

Large Language Models (LLMs) have demonstrated notable capabilities across financial tasks, including financial report summarization, earnings call transcript analysis, and asset classification. However, their real-world effectiveness in managing complex fund investment remains inadequately assessed. A fundamental limitation of existing benchmarks for evaluating LLM-driven trading strategies is their reliance on historical back-testing, inadvertently enabling LLMs to "time travel"-leveraging future information embedded in their training corpora, thus resulting in possible information leakage and overly optimistic performance estimates. To address this issue, we introduce DeepFund, a live fund benchmark tool designed to rigorously evaluate LLM in real-time market conditions. Utilizing a multi-agent architecture, DeepFund connects directly with real-time stock market data-specifically data published after each model pretraining cutoff-to ensure fair and leakage-free evaluations. Empirical tests on nine flagship LLMs from leading global institutions across multiple investment dimensions-including ticker-level analysis, investment decision-making, portfolio management, and risk control-reveal significant practical challenges. Notably, even cutting-edge models such as DeepSeek-V3 and Claude-3.7-Sonnet incur net trading losses within DeepFund real-time evaluation environment, underscoring the present limitations of LLMs for active fund management. Our code is available at https://github.com/HKUSTDial/DeepFund.

Ambiguity Resolution in Text-to-Structured Data Mapping

Ambiguity in natural language is a significant obstacle for achieving accurate text to structured data mapping through large language models (LLMs), which affects the performance of tasks such as mapping text to agentic tool calling and text-to-SQL queries. Existing methods of ambiguity handling either exploit ReACT framework to produce the correct mapping through trial and error, or supervised fine tuning to guide models to produce a biased mapping to improve certain tasks. In this paper, we adopt a different approach that characterizes the representation difference of ambiguous text in the latent space and leverage the difference to identify ambiguity before mapping them to structured data. To detect ambiguity of a sentence, we focused on the relationship between ambiguous questions and their interpretations and what cause the LLM ignore multiple interpretations. Different to the distance calculated by dense embedding vectors, we utilize the observation that ambiguity is caused by concept missing in latent space of LLM to design a new distance measurement, computed through the path kernel by the integral of gradient values for each concepts from sparse-autoencoder (SAE) under each state. We identify patterns to distinguish ambiguous questions with this measurement. Based on our observation, We propose a new framework to improve the performance of LLMs on ambiguous agentic tool calling through missing concepts prediction.

Comet: Accelerating Private Inference for Large Language Model by Predicting Activation Sparsity

With the growing use of large language models (LLMs) hosted on cloud platforms to offer inference services, privacy concerns about the potential leakage of sensitive information are escalating. Secure multi-party computation (MPC) is a promising solution to protect the privacy in LLM inference. However, MPC requires frequent inter-server communication, causing high performance overhead. Inspired by the prevalent activation sparsity of LLMs, where most neuron are not activated after non-linear activation functions, we propose an efficient private inference system, Comet. This system employs an accurate and fast predictor to predict the sparsity distribution of activation function output. Additionally, we introduce a new private inference protocol. It efficiently and securely avoids computations involving zero values by exploiting the spatial locality of the predicted sparse distribution. While this computation-avoidance approach impacts the spatiotemporal continuity of KV cache entries, we address this challenge with a low-communication overhead cache refilling strategy that merges miss requests and incorporates a prefetching mechanism. Finally, we evaluate Comet on four common LLMs and compare it with six state-of-the-art private inference systems. Comet achieves a 1.87x-2.63x speedup and a 1.94x-2.64x communication reduction.