arXiv subjects

Huan Lin

Publications and source records attributed to Huan Lin.

At least 19 recordsLinked to original sources

arXiv2026

PACT: Post-route Agentic Checkpoint Tuning for FPGA Timing Closure

Late-stage FPGA timing closure often starts from an implemented design whose remaining violations are visible in timing reports. Engineering change order (ECO) optimization is a standard mechanism for applying localized changes to such designs without restarting the full implementation flow. Automating post-route ECO optimization remains challenging. A post-route change must improve timing without violating routing legality, hold or pulse-width timing constraints, checkpoint replayability or functional equivalence. This paper presents PACT, a Post-route Agentic Checkpoint Tuning framework for Vivado design checkpoints (DCP). PACT represents post-route tuning as validation-gated transitions between accepted and candidate checkpoints. From checkpoint-derived evidence, an agent selects localized backend actions through a profile-driven recipe planner over typed Vivado and RapidWright skills, and probes tool behavior in isolated workspaces. PACT records each trial as an evidence-gated case to guide candidate generation and suppress unsafe, unsupported or ineffective actions. Across 35 UltraScale+ post-route checkpoints, PACT improves validation-clean $F_{\max}$ (maximum operating frequency) by a geometric mean of $+22.30\%$ over the original implementations, compared with $+15.14\%$ for DATuner and $+9.78\%$ for the Codex Agent. On shared designs, PACT achieves these gains $6.4\times$ faster than the uncapped DATuner and at an average token cost of only \$0.16 per DCP ($24.5\times$ lower than the free-form Codex Agent). The source code is available in an anonymous repository

cs.AR↗

arXiv2026

Qwen-Scope: Turning Sparse Features into Development Tools for Large Language Models

Large language models have achieved remarkable capabilities across diverse tasks, yet their internal decision-making processes remain largely opaque, limiting our ability to inspect, control, and systematically improve them. This opacity motivates a growing body of research in mechanistic interpretability, with sparse autoencoders (SAEs) emerging as one of the most promising tools for decomposing model activations into sparse, interpretable feature representations. We introduce Qwen-Scope, an open-source suite of SAEs built on the Qwen model family, comprising 14 groups of SAEs across 7 model variants from the Qwen3 and Qwen3.5 series, covering both dense and mixture-of-expert architectures. Built on top of these SAEs, we show that SAEs can go beyond post-hoc analysis to serve as practical interfaces for model development along four directions: (i) inference-time steering, where SAE feature directions control language, concepts, and preferences without modifying model weights; (ii) evaluation analysis, where activated SAE features provide a representation-level proxy for benchmark redundancy and capability coverage; (iii) data-centric workflows, where SAE features support multilingual toxicity classification and safety-oriented data synthesis; and (iv) post-training optimization, where SAE-derived signals are incorporated into supervised fine-tuning and reinforcement learning objectives to mitigate undesirable behaviors such as code-switching and repetition. Together, these results demonstrate that SAEs can serve not only as post-hoc analysis tools, but also as reusable representation-level interfaces for diagnosing, controlling, evaluating, and improving large language models. By open-sourcing Qwen-Scope, we aim to support mechanistic research and accelerate practical workflows that connect model internals to downstream behavior.

cs.CL↗

arXiv2026

Zero-Shot Scalable Resilience in UAV Swarms: A Decentralized Imitation Learning Framework with Physics-Informed Graph Interactions

Large-scale Unmanned Aerial Vehicle (UAV) failures can split an unmanned aerial vehicle swarm network into disconnected sub-networks, making decentralized recovery both urgent and difficult. Centralized recovery methods depend on global topology information and become communication-heavy after severe fragmentation. Decentralized heuristics and multi-agent reinforcement learning methods are easier to deploy, but their performance often degrades when the swarm scale and damage severity vary. We present Physics-informed Graph Adversarial Imitation Learning algorithm (PhyGAIL) that adopts centralized training with decentralized execution. PhyGAIL builds bounded local interaction graphs from heterogeneous observations, and uses physics-informed graph neural network to encode directional local interactions as gated message passing with explicit attraction and repulsion. This gives the policy a physically grounded coordination bias while keeping local observations scale-invariant. It also uses scenario-adaptive imitation learning to improve training under fragmented topologies and variable-length recovery episodes. Our analysis establishes bounded local graph amplification, bounded interaction dynamics, and controlled variance of the terminal success signal. A policy trained on 20-UAV swarms transfers directly to swarms of up to 500 UAVs without fine-tuning, and achieves better performance across reconnection reliability, recovery speed, motion safety, and runtime efficiency than representative baselines.

cs.LG↗

arXiv2026

EPD-Serve: A Flexible Multimodal EPD Disaggregation Inference Serving System On Ascend

With the widespread adoption of large multimodal models, efficient inference across text, image, audio, and video modalities has become critical. However, existing multimodal inference systems typically employ monolithic architectures that tightly couple the Encode, Prefill, and Decode stages on homogeneous hardware, neglecting the heterogeneous computational characteristics of each stage. This design leads to inefficient resource utilization and limited system throughput. To address these issues, we propose EPD-Serve, a stage-level disaggregated inference serving system for multimodal models. EPD-Serve decouples the inference pipeline into independent Encode, Prefill, and Decode stages, enabling logical isolation and flexible co-located deployment through dynamic orchestration. Leveraging the Ascend interconnect topology, EPD-Serve introduces asynchronous feature prefetching between Encode and Prefill stages and a hierarchical grouped KV cache transmission mechanism between Prefill and Decode stages to improve cross-node communication efficiency. In addition, EPD-Serve incorporates multi-route scheduling, instance-level load balancing, and multi-stage hardware co-location with spatial multiplexing to better support diverse multimodal workloads. Comprehensive experiments on multimodal understanding models demonstrate that, under high-concurrency scenarios, EPD-Serve improves end-to-end throughput by 57.37-69.48% compared to PD-disaggregated deployment, while satisfying strict SLO constraints, including TTFT below 2000 ms and TPOT below 50 ms. These results highlight the effectiveness of stage-level disaggregation for optimizing multimodal large model inference systems.

cs.DC↗

arXiv2025

Bio-inspired Integrated Networking and Control for Large-Scale Swarm: A Hierarchical Co-design

Unmanned aerial vehicle (UAV) swarms encounter the challenge of high overhead due to both network management and formation control requirements. In this paper, we propose a Bio-inspired Integrated Networking and Control (BINC) scheme, enabling efficient formation management for swarms comprising thousands of UAVs. The scheme forms a two-layer hierarchical structure, where network clusters and formations share the same groups so that cross-cluster control is eliminated. For networking, we design a fused routing message together with control information to reduce overhead, and limit clusters' size to local two-hop topologies for fast command transmission. For controlling, we develop a hybrid bio-inspired control approach, including a pigeon-like leader-follower algorithm within formations under the consideration of cluster topology maintenance, and a starling-like algorithm among formations that helps to improve the ability of obstacle avoidance. We establish a simulation platform for UAV swarms with over 1000 nodes, and experimental results show that the proposed BINC scheme can achieve highly maneuverable swarm formation marching with significant reduction on communication overhead.

cs.NI↗

arXiv2025

Multi-Level Damage-Aware Graph Learning for Resilient UAV Swarm Networks

Unmanned aerial vehicle (UAV) swarm networks leverage resilient algorithms to restore connectivity from communication network split issues. However, existing graph learning-based approaches face over-aggregation and non-convergence problems caused by uneven and sparse topology under massive damage. In this paper, we propose a novel Multi-Level Damage-Aware (MLDA) Graph Learning algorithm to generate recovery solutions, explicitly utilizing information about destroyed nodes to guide the recovery process. The algorithm first employs a Multi-Branch Damage Attention (MBDA) module as a pre-processing step, focusing attention on the critical relationships between remaining nodes and destroyed nodes in the global topology. By expanding multi-hop neighbor receptive fields of nodes to those damaged areas, it effectively mitigating the initial sparsity and unevenness before graph learning commences. Second, a Dilated Graph Convolution Network (DGCN) is designed to perform convolution on the MBDA-processed bipartite graphs between remaining and destroyed nodes. The DGCN utilizes a specialized bipartite graph convolution operation to aggregate features and incorporates a residual-connected architecture to extend depth, directly generating the target locations for recovery. We theoretically proved the convergence of the proposed algorithm and the computational complexity is acceptable. Simulation results show that the proposed algorithm can guarantee the connectivity restoration with excellent scalability, while significantly expediting the recovery time and improving the topology uniformity after recovery.

cs.NI↗

arXiv2025

Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Models

In this work, we introduce the Qwen3 Embedding series, a significant advancement over its predecessor, the GTE-Qwen series, in text embedding and reranking capabilities, built upon the Qwen3 foundation models. Leveraging the Qwen3 LLMs' robust capabilities in multilingual text understanding and generation, our innovative multi-stage training pipeline combines large-scale unsupervised pre-training with supervised fine-tuning on high-quality datasets. Effective model merging strategies further ensure the robustness and adaptability of the Qwen3 Embedding series. During the training process, the Qwen3 LLMs serve not only as backbone models but also play a crucial role in synthesizing high-quality, rich, and diverse training data across multiple domains and languages, thus enhancing the training pipeline. The Qwen3 Embedding series offers a spectrum of model sizes (0.6B, 4B, 8B) for both embedding and reranking tasks, addressing diverse deployment scenarios where users can optimize for either efficiency or effectiveness. Empirical evaluations demonstrate that the Qwen3 Embedding series achieves state-of-the-art results across diverse benchmarks. Notably, it excels on the multilingual evaluation benchmark MTEB for text embedding, as well as in various retrieval tasks, including code retrieval, cross-lingual retrieval and multilingual retrieval. To facilitate reproducibility and promote community-driven research and development, the Qwen3 Embedding models are publicly available under the Apache 2.0 license.

cs.CL↗

arXiv2025

TDCOSMO 2025: Cosmological constraints from strong lensing time delays

We present cosmological constraints from 8 strongly lensed quasars (hereafter, the TDCOSMO-2025 sample). Building on previous work, our analysis incorporated new deflector stellar velocity dispersions measured from spectra obtained with the James Webb Space Telescope (JWST), the Keck Telescopes, and the Very Large Telescope (VLT), utilizing improved methods. We used integrated JWST stellar kinematics for 5 lenses, VLT-MUSE for 2, and resolved kinematics from Keck and JWST for RXJ1131-1231. We also considered two samples of non-time-delay lenses: 11 from the Sloan Lens ACS (SLACS) sample with Keck-KCWI resolved kinematics; and 4 from the Strong Lenses in the Legacy Survey (SL2S) sample. We improved our analysis of line-of-sight effects, the surface brightness profile of the lens galaxies, and orbital anisotropy, and corrected for projection effects in the dynamics. Our uncertainties are maximally conservative by accounting for the mass-sheet degeneracy in the deflectors' mass density profiles. The analysis was blinded to prevent experimenter bias. Our primary result is based on the TDCOSMO-2025 sample, in combination with $\Omega_{\rm m}$ constraints from the Pantheon+ Type Ia supernovae (SN) dataset. In the flat $\Lambda$ cold dark matter (CDM), we find $H_0=71.6^{+3.9}_{-3.3}$ km s$^{-1}$ Mpc$^{-1}$. The SLACS and SL2S samples are in excellent agreement with the TDCOSMO-2025 sample, improving the precision on $H_0$ in flat $\Lambda$CDM to 4.6%. Using the Dark Energy Survey SN Year-5 dataset (DES-SN5YR) or DESI-DR2 baryonic acoustic oscillations (BAO) likelihoods instead of Pantheon+ yields very similar results. We also present constraints in the open $\Lambda$CDM, $w$CDM, $w_0w_a$CDM, and $w_{\phi}$CDM cosmologies. The TDCOSMO $H_0$ inference is robust and consistent across all presented cosmological models, and our cosmological constraints in them agree with those from the BAO and SN.

astro-ph.CO↗

arXiv2025

Qwen3 Technical Report

In this work, we present Qwen3, the latest version of the Qwen model family. Qwen3 comprises a series of large language models (LLMs) designed to advance performance, efficiency, and multilingual capabilities. The Qwen3 series includes models of both dense and Mixture-of-Expert (MoE) architectures, with parameter scales ranging from 0.6 to 235 billion. A key innovation in Qwen3 is the integration of thinking mode (for complex, multi-step reasoning) and non-thinking mode (for rapid, context-driven responses) into a unified framework. This eliminates the need to switch between different models--such as chat-optimized models (e.g., GPT-4o) and dedicated reasoning models (e.g., QwQ-32B)--and enables dynamic mode switching based on user queries or chat templates. Meanwhile, Qwen3 introduces a thinking budget mechanism, allowing users to allocate computational resources adaptively during inference, thereby balancing latency and performance based on task complexity. Moreover, by leveraging the knowledge from the flagship models, we significantly reduce the computational resources required to build smaller-scale models, while ensuring their highly competitive performance. Empirical evaluations demonstrate that Qwen3 achieves state-of-the-art results across diverse benchmarks, including tasks in code generation, mathematical reasoning, agent tasks, etc., competitive against larger MoE models and proprietary models. Compared to its predecessor Qwen2.5, Qwen3 expands multilingual support from 29 to 119 languages and dialects, enhancing global accessibility through improved cross-lingual understanding and generation capabilities. To facilitate reproducibility and community-driven research and development, all Qwen3 models are publicly accessible under Apache 2.0.

cs.CL↗

arXiv2025

Enhancing LLM Language Adaption through Cross-lingual In-Context Pre-training

Large language models (LLMs) exhibit remarkable multilingual capabilities despite English-dominated pre-training, attributed to cross-lingual mechanisms during pre-training. Existing methods for enhancing cross-lingual transfer remain constrained by parallel resources, suffering from limited linguistic and domain coverage. We propose Cross-lingual In-context Pre-training (CrossIC-PT), a simple and scalable approach that enhances cross-lingual transfer by leveraging semantically related bilingual texts via simple next-word prediction. We construct CrossIC-PT samples by interleaving semantic-related bilingual Wikipedia documents into a single context window. To access window size constraints, we implement a systematic segmentation policy to split long bilingual document pairs into chunks while adjusting the sliding window mechanism to preserve contextual coherence. We further extend data availability through a semantic retrieval framework to construct CrossIC-PT samples from web-crawled corpus. Experimental results demonstrate that CrossIC-PT improves multilingual performance on three models (Llama-3.1-8B, Qwen2.5-7B, and Qwen2.5-1.5B) across six target languages, yielding performance gains of 3.79%, 3.99%, and 1.95%, respectively, with additional improvements after data augmentation.

cs.CL↗

arXiv2024

Qwen2.5 Technical Report

In this report, we introduce Qwen2.5, a comprehensive series of large language models (LLMs) designed to meet diverse needs. Compared to previous iterations, Qwen 2.5 has been significantly improved during both the pre-training and post-training stages. In terms of pre-training, we have scaled the high-quality pre-training datasets from the previous 7 trillion tokens to 18 trillion tokens. This provides a strong foundation for common sense, expert knowledge, and reasoning capabilities. In terms of post-training, we implement intricate supervised finetuning with over 1 million samples, as well as multistage reinforcement learning. Post-training techniques enhance human preference, and notably improve long text generation, structural data analysis, and instruction following. To handle diverse and varied use cases effectively, we present Qwen2.5 LLM series in rich sizes. Open-weight offerings include base and instruction-tuned models, with quantized versions available. In addition, for hosted solutions, the proprietary models currently include two mixture-of-experts (MoE) variants: Qwen2.5-Turbo and Qwen2.5-Plus, both available from Alibaba Cloud Model Studio. Qwen2.5 has demonstrated top-tier performance on a wide range of benchmarks evaluating language understanding, reasoning, mathematics, coding, human preference alignment, etc. Specifically, the open-weight flagship Qwen2.5-72B-Instruct outperforms a number of open and proprietary models and demonstrates competitive performance to the state-of-the-art open-weight model, Llama-3-405B-Instruct, which is around 5 times larger. Qwen2.5-Turbo and Qwen2.5-Plus offer superior cost-effectiveness while performing competitively against GPT-4o-mini and GPT-4o respectively. Additionally, as the foundation, Qwen2.5 models have been instrumental in training specialized models such as Qwen2.5-Math, Qwen2.5-Coder, QwQ, and multimodal models.

cs.CL↗

arXiv2024

Multi-hop Differential Topology based Algorithms for Resilient Network of UAV Swarm

Unmanned aerial vehicle (UAV) swarm networks face severe challenges of communication network split (CNS) issues caused by massive damage in hostile environments. In this paper, we propose a new paradigm to restore network connectivity by repositioning remaining UAVs based on damage information within local topologies. Particularly, the locations of destroyed UAVs distributed in gaps between disconnected sub-nets are considered for recovery trajectory planning. Specifically, we construct the multi-hop differential sub-graph (MDSG) to represent local damage-varying topologies. Based on this, we develop two distinct algorithms to address CNS issues. The first approach leverages an artificial potential field algorithm to calculate the recovery velocities via MDSG, enabling simple deployment on low-intelligence UAVs. In the second approach, we design an MDSG-based graph convolution framework to find the recovery topology for high-intelligence swarms. As per the unique topology of MDSG, we propose a novel bipartite graph convolution operation, enhanced with a batch-processing mechanism to improve graph convolution efficiency. Simulation results show that the proposed algorithms expedite the recovery with significant margin while improving the spatial coverage and topology degree uniformity after recovery.

cs.NI↗

arXiv2024

mGTE: Generalized Long-Context Text Representation and Reranking Models for Multilingual Text Retrieval

We present systematic efforts in building long-context multilingual text representation model (TRM) and reranker from scratch for text retrieval. We first introduce a text encoder (base size) enhanced with RoPE and unpadding, pre-trained in a native 8192-token context (longer than 512 of previous multilingual encoders). Then we construct a hybrid TRM and a cross-encoder reranker by contrastive learning. Evaluations show that our text encoder outperforms the same-sized previous state-of-the-art XLM-R. Meanwhile, our TRM and reranker match the performance of large-sized state-of-the-art BGE-M3 models and achieve better results on long-context retrieval benchmarks. Further analysis demonstrate that our proposed models exhibit higher efficiency during both training and inference. We believe their efficiency and effectiveness could benefit various researches and industrial applications.

cs.CL↗

arXiv2024

Qwen2 Technical Report

This report introduces the Qwen2 series, the latest addition to our large language models and large multimodal models. We release a comprehensive suite of foundational and instruction-tuned language models, encompassing a parameter range from 0.5 to 72 billion, featuring dense models and a Mixture-of-Experts model. Qwen2 surpasses most prior open-weight models, including its predecessor Qwen1.5, and exhibits competitive performance relative to proprietary models across diverse benchmarks on language understanding, generation, multilingual proficiency, coding, mathematics, and reasoning. The flagship model, Qwen2-72B, showcases remarkable performance: 84.2 on MMLU, 37.9 on GPQA, 64.6 on HumanEval, 89.5 on GSM8K, and 82.4 on BBH as a base language model. The instruction-tuned variant, Qwen2-72B-Instruct, attains 9.1 on MT-Bench, 48.1 on Arena-Hard, and 35.7 on LiveCodeBench. Moreover, Qwen2 demonstrates robust multilingual capabilities, proficient in approximately 30 languages, spanning English, Chinese, Spanish, French, German, Arabic, Russian, Korean, Japanese, Thai, Vietnamese, and more, underscoring its versatility and global reach. To foster community innovation and accessibility, we have made the Qwen2 model weights openly available on Hugging Face and ModelScope, and the supplementary materials including example code on GitHub. These platforms also include resources for quantization, fine-tuning, and deployment, facilitating a wide range of applications and research endeavors.

cs.CL↗

arXiv2024

TDCOSMO. XVI. Measurement of the Hubble Constant from the Lensed Quasar WGD$\,$2038$-$4008

Time-delay cosmography is a powerful technique to constrain cosmological parameters, particularly the Hubble constant ($H_{0}$). The TDCOSMO collaboration is performing an ongoing analysis of lensed quasars to constrain cosmology using this method. In this work, we obtain constraints from the lensed quasar WGD 2038-4008 using new time-delay measurements and previous mass models by TDCOSMO. This is the first TDCOSMO lens to incorporate multiple lens modeling codes and the full time-delay covariance matrix into the cosmological inference. The models are fixed before the time delay is measured, and the analysis is performed blinded with respect to the cosmological parameters to prevent unconscious experimenter bias. We obtain $D_{\Delta t} = 1.68^{+0.40}_{-0.38}$ Gpc using two families of mass models, a power-law describing the total mass distribution, and a composite model of baryons and dark matter, although the composite model is disfavored due to kinematics constraints. In a flat $\Lambda$CDM cosmology, we constrain the Hubble constant to be $H_{0} = 65^{+23}_{-14}\, \rm km\ s^{-1}\,Mpc^{-1}$. The dominant source of uncertainty comes from the time delays, due to the low variability of the quasar. Future long-term monitoring, especially in the era of the Vera C. Rubin Observatory's Legacy Survey of Space and Time, could catch stronger quasar variability and further reduce the uncertainties. This system will be incorporated into an upcoming hierarchical analysis of the entire TDCOSMO sample, and improved time delays and spatially-resolved stellar kinematics could strengthen the constraints from this system in the future.

astro-ph.CO↗

arXiv2024

A Cohesive Deep Drilling Field Strategy for LSST Cosmology

The Vera C. Rubin Observatory Legacy Survey of Space and Time (LSST) will image billions of astronomical objects in the wide-fast-deep primary survey and in a set of minisurveys including intensive observations of a group of deep drilling fields (DDFs). The DDFs are a critical piece of three key aspects of the LSST Dark Energy Science Collaboration (DESC) cosmological measurements: they provide a required calibration for photometric redshifts and weak gravitational lensing measurements and they directly contribute to cosmological constraints from the most distant type Ia supernovae. We present a set of cohesive DDF strategies fulfilling science requirements relevant to DESC and following the guidelines of the Survey Cadence Optimization Committee. We propose a method to estimate the observing strategy parameters and we perform simulations of the corresponding surveys. We define a set of metrics for each of the science case to assess the performance of the proposed observing strategies. We show that the most promising results are achieved with deep rolling surveys characterized by two sets of fields: ultradeep fields (z<1.1) observed at a high cadence with a large number of visits over a limited number of seasons; deep fields (z<0.7), observed with a cadence of ~3 nights for ten years. These encouraging results should be confirmed with realistic simulations using the LSST scheduler. A DDF budget of ~8.5% is required to design observing strategies satisfying all the cosmological requirements. A lower DDF budget lead to surveys that either do not fulfill photo-z/WL requirements or are not optimal for SNe Ia cosmology.

astro-ph.CO↗

arXiv2023

PolyLM: An Open Source Polyglot Large Language Model

Large language models (LLMs) demonstrate remarkable ability to comprehend, reason, and generate following nature language instructions. However, the development of LLMs has been primarily focused on high-resource languages, such as English, thereby limiting their applicability and research in other languages. Consequently, we present PolyLM, a multilingual LLM trained on 640 billion (B) tokens, avaliable in two model sizes: 1.7B and 13B. To enhance its multilingual capabilities, we 1) integrate bilingual data into training data; and 2) adopt a curriculum learning strategy that increases the proportion of non-English data from 30% in the first stage to 60% in the final stage during pre-training. Further, we propose a multilingual self-instruct method which automatically generates 132.7K diverse multilingual instructions for model fine-tuning. To assess the model's performance, we collect several existing multilingual tasks, including multilingual understanding, question answering, generation, and translation. Extensive experiments show that PolyLM surpasses other open-source models such as LLaMA and BLOOM on multilingual tasks while maintaining comparable performance in English. Our models, alone with the instruction data and multilingual benchmark, are available at: \url{https://modelscope.cn/models/damo/nlp_polylm_13b_text_generation}.

cs.CL↗

arXiv2023

Bridging the Domain Gaps in Context Representations for k-Nearest Neighbor Neural Machine Translation

$k$-Nearest neighbor machine translation ($k$NN-MT) has attracted increasing attention due to its ability to non-parametrically adapt to new translation domains. By using an upstream NMT model to traverse the downstream training corpus, it is equipped with a datastore containing vectorized key-value pairs, which are retrieved during inference to benefit translation. However, there often exists a significant gap between upstream and downstream domains, which hurts the retrieval accuracy and the final translation quality. To deal with this issue, we propose a novel approach to boost the datastore retrieval of $k$NN-MT by reconstructing the original datastore. Concretely, we design a reviser to revise the key representations, making them better fit for the downstream domain. The reviser is trained using the collected semantically-related key-queries pairs, and optimized by two proposed losses: one is the key-queries semantic distance ensuring each revised key representation is semantically related to its corresponding queries, and the other is an L2-norm loss encouraging revised key representations to effectively retain the knowledge learned by the upstream NMT model. Extensive experiments on domain adaptation tasks demonstrate that our method can effectively boost the datastore retrieval and translation quality of $k$NN-MT.\footnote{Our code is available at \url{https://github.com/DeepLearnXMU/RevisedKey-knn-mt}.}

cs.CL↗