Bio-inspired Learning and Decision-Making with Probabilistic In-Memory Computing Hardware: Part 1
该研究通过生物启发的框架,利用概率内存计算硬件模拟动物学习和决策过程中的贝叶斯推断,以解决不确定性问题。
该研究通过生物启发的框架,利用概率内存计算硬件模拟动物学习和决策过程中的贝叶斯推断,以解决不确定性问题。
本文通过使用概率模拟内存计算处理器解决了大规模概率能量模型在GPU上执行时面临的可扩展性挑战,实现了超过1000倍的加速。
This work addresses the inefficiency of existing Arabic pre-trained models in processing long texts due to the quadratic complexity of self-attention mechanisms. To overcome this limitation, we propose the first bidirectional Mamba encoder tailored for Arabic, leveraging a selective state space architecture that enables linear-complexity modeling of long sequences. The model is pretrained from scratch via masked language modeling on Arabic Wikipedia and CulturaX corpora. Experimental results demonstrate that our approach achieves highly efficient training on consumer-grade GPUs and delivers strong performance across multiple benchmarks, including 96.37% accuracy on the HARD sentiment classification task, as well as competitive results on ARCD question answering and ANERcorp named entity recognition. Although it slightly underperforms Transformer baselines on XNLI inference, the model overall matches or surpasses current state-of-the-art methods in Arabic NLP.
This work addresses the high memory and computational costs of multilingual neural machine translation models stemming from their large vocabularies, a challenge inadequately tackled by existing compression methods that fail to effectively eliminate lexical redundancy. The authors propose a corpus-driven framework that synergistically combines vocabulary pruning with targeted fine-tuning, thereby breaking free from reliance on the original vocabulary structure for the first time. Applied to prominent models such as M2M100, NLLB-200, and mBART-50, the approach reduces vocabulary size from 128,000 to approximately 10,000 tokens—cutting memory usage by 60%—while simultaneously enhancing translation quality. Notably, the pruned M2M100 model surpasses specialized bilingual baselines, achieving state-of-the-art performance with BLEU and COMET scores of 42.04 and 0.8730, respectively, thus demonstrating both efficiency and high fidelity in multilingual translation.
This work addresses the challenge of precisely analyzing functional correctness and resource consumption—such as termination and expected cost—in hybrid quantum programs featuring unbounded loops. The paper presents the first semi-automatic static analysis framework that uniformly handles both functional verification and resource estimation by introducing integer hybrid path summaries (IHPS) to characterize execution paths. The approach integrates loop invariants, extended path forms, quantum Hoare logic, and symbolic execution techniques within a unified formalism. A prototype implementation in Haskell demonstrates the effectiveness of the method on multiple benchmarks, establishing a foundation for building a comprehensive static analyzer for resource-aware verification of hybrid quantum programs.
该研究通过生物启发的框架,利用概率内存计算硬件模拟动物学习和决策过程中的贝叶斯推断,以解决不确定性问题。
本文通过使用概率模拟内存计算处理器解决了大规模概率能量模型在GPU上执行时面临的可扩展性挑战,实现了超过1000倍的加速。
This work addresses the inefficiency of existing Arabic pre-trained models in processing long texts due to the quadratic complexity of self-attention mechanisms. To overcome this limitation, we propose the first bidirectional Mamba encoder tailored for Arabic, leveraging a selective state space architecture that enables linear-complexity modeling of long sequences. The model is pretrained from scratch via masked language modeling on Arabic Wikipedia and CulturaX corpora. Experimental results demonstrate that our approach achieves highly efficient training on consumer-grade GPUs and delivers strong performance across multiple benchmarks, including 96.37% accuracy on the HARD sentiment classification task, as well as competitive results on ARCD question answering and ANERcorp named entity recognition. Although it slightly underperforms Transformer baselines on XNLI inference, the model overall matches or surpasses current state-of-the-art methods in Arabic NLP.
This work addresses the high memory and computational costs of multilingual neural machine translation models stemming from their large vocabularies, a challenge inadequately tackled by existing compression methods that fail to effectively eliminate lexical redundancy. The authors propose a corpus-driven framework that synergistically combines vocabulary pruning with targeted fine-tuning, thereby breaking free from reliance on the original vocabulary structure for the first time. Applied to prominent models such as M2M100, NLLB-200, and mBART-50, the approach reduces vocabulary size from 128,000 to approximately 10,000 tokens—cutting memory usage by 60%—while simultaneously enhancing translation quality. Notably, the pruned M2M100 model surpasses specialized bilingual baselines, achieving state-of-the-art performance with BLEU and COMET scores of 42.04 and 0.8730, respectively, thus demonstrating both efficiency and high fidelity in multilingual translation.
This work addresses the challenge of precisely analyzing functional correctness and resource consumption—such as termination and expected cost—in hybrid quantum programs featuring unbounded loops. The paper presents the first semi-automatic static analysis framework that uniformly handles both functional verification and resource estimation by introducing integer hybrid path summaries (IHPS) to characterize execution paths. The approach integrates loop invariants, extended path forms, quantum Hoare logic, and symbolic execution techniques within a unified formalism. A prototype implementation in Haskell demonstrates the effectiveness of the method on multiple benchmarks, establishing a foundation for building a comprehensive static analyzer for resource-aware verification of hybrid quantum programs.