Large Language Models (LLMs), despite their remarkable capabilities, are prone to generating hallucinated or outdated content due to their static internal knowledge. While Retrieval-Augmented Generation (RAG) integrated with Reinforcement Learning (RL) offers a solution, these methods are fundamentally constrained by a single-query mode, leading to prohibitive latency and inherent brittleness. To overcome these limitations, we introduce RAG-R1, a novel two-stage training framework centered around multi-query parallelism. Our framework enables LLMs to adaptively leverage internal and external knowledge during the reasoning process while transitioning from the single-query mode to multi-query parallelism. This architectural shift bolsters reasoning robustness while significantly reducing inference latency. Extensive experiments on seven question-answering benchmarks confirm the superiority of our method, which outperforms the strongest baseline by up to 13.7% and decreases inference time by 11.1%.

The emergence of long-context text applications utilizing large language models (LLMs) has presented significant scalability challenges, particularly in memory footprint. The linear growth of the Key-Value (KV) cache, which stores attention keys and values to reduce redundant computations, can significantly increase memory usage and may prevent models from functioning properly in memory-constrained environments. To address this issue, we propose a novel approach called Cache Sparse Representation (CSR), which converts the KV cache by transforming the dense Key-Value cache tensor into sparse indexes and weights, offering a more memory-efficient representation during LLM inference. Furthermore, we introduce NeuralDict, a novel neural network-based method to automatically generate the dictionary used in our sparse representation. Our extensive experiments demonstrate that CSR matches the performance of state-of-the-art KV cache quantization algorithms while ensuring robust functionality in memory-constrained environments.

Estimating per-flow cardinality from high-speed data streams has many applications such as anomaly detection and resource allocation. Yet despite tracking single flow cardinality with approximation algorithms offered, there remain algorithmical challenges for monitoring multi-flows especially under unbalanced cardinality distribution: existing methods adopt a uniform sketch layout and incur a large memory footprint to achieve high accuracy. Furthermore, they are hard to implement in the compact hardware used for line-rate processing. In this paper, we propose Couper, a memory-efficient measurement framework that can estimate cardinality for multi-flows under unbalanced cardinality distribution. We propose a two-layer structure based on a classic coupon collector's principle, where numerous mice flows are confined to the first layer and only the potential elephant flows are allowed to enter the second layer. Our two-layer structure can better fit the unbalanced cardinality distribution in practice and achieve much higher memory efficiency. We implement Couper in both software and hardware. Extensive evaluation under real-world and synthetic data traces show more than 20× improvements in terms of memory-efficiency compared to state-of-the-art.

Estimating per-flow cardinality from high-speed data streams has many applications such as anomaly detection and resource allocation. Yet despite tracking single flow cardinality with approximation algorithms offered, there remain algorithmical challenges for monitoring multi-flows especially under unbalanced cardinality distribution: existing methods adopt a uniform sketch layout and incur a large memory footprint to achieve high accuracy. Furthermore, they are hard to implement in the compact hardware used for line-rate processing.In this paper, we propose Couper, a memory-efficient measurement framework that can estimate cardinality for multi-flows under unbalanced cardinality distribution. We propose a two-layer structure based on a classic coupon collector’s principle, where numerous mice flows are confined to the first layer and only the potential elephant flows are allowed to enter the second layer. Our two-layer structure can better fit the unbalanced cardinality distribution in practice and achieve much higher memory efficiency. We implement Couper in both software and hardware. Extensive evaluation under real-world and synthetic data traces show more than 20× improvements in terms of memory-efficiency compared to state-of-the-art.

Given the enormous number of users and items, industrial cascade recommendation systems (RS) are continuously expanded in size and complexity to deliver relevant items, such as news, services, and commodities, to the appropriate users. In a real-world scenario with hundreds of thousands requests per second, significant computation is required to infer personalized results for each request, resulting in a massive energy consumption and carbon emission that raises concern. This paper proposes GreenFlow, a practical computation allocation framework for RS, that considers both accuracy and carbon emission during inference. For each stage (e.g., recall, pre-ranking, ranking, etc.) of a cascade RS, when a user triggers a request, we define two actions that determine the computation: (1) the trained instances of models with different computational complexity; and (2) the number of items to be inferred in the stage. We refer to the combinations of actions in all stages as action chains. A reward score is estimated for each action chain, followed by dynamic primal-dual optimization considering both the reward and computation budget. Extensive experiments verify the effectiveness of the framework, reducing computation consumption by 41% in an industrial mobile application while maintaining commercial revenue. Moreover, the proposed framework saves approximately 5000kWh of electricity and reduces 3 tons of carbon emissions per day.