分类：AI大模型

共 524 篇

LEANN AI框架：2024向量搜索存储与计算权衡技术指南

BLUFLEANN AI框架提出一种低存储向量索引，通过创新的存储-计算权衡设计，将索引存储降至原数据5%以下，实现高达50倍的压缩比，为高维向量搜索提供高效解决方案。原文翻译： The LEANN AI framework proposes a low-storage vector index that, through an innovative storage-computation trade-off design, reduces index storage to less than 5% of the original data, achieving up to 50x compression ratio, offering an efficient solution for high-dimensional vector search.

AI大模型2026/1/20

阅读全文 →

VoxCPM开源TTS模型架构解析与2024技术指南

BLUFVoxCPM 是一款先进的开源TTS模型，采用无分词器架构与上下文感知技术，支持零样本语音克隆，并在消费级硬件上实现高效实时合成。原文翻译： VoxCPM is an advanced open-source TTS model featuring a tokenizer-free architecture and context-aware generation. It supports zero-shot voice cloning and enables efficient real-time synthesis on consumer-grade hardware.

AI大模型2026/1/20

阅读全文 →

LEANN AI框架：向量数据库的SQLite时刻，本地部署RAG存储减少97%

BLUFLEANN is an embedded vector database for RAG that reduces storage by 97% vs traditional solutions while maintaining accuracy, enabling local deployment on laptops with full privacy and no cloud dependencies.

AI大模型2026/1/20

阅读全文 →

VoxCPM：无标记器TTS技术，实现上下文感知语音生成与零样本语音克隆

BLUFVoxCPM is a tokenizer-free TTS system using continuous speech modeling for context-aware generation and zero-shot voice cloning, achieving RTF as low as 0.15 on consumer GPUs.

AI大模型2026/1/20

阅读全文 →

大型AI模型技术解析：架构基础到应用实践指南

BLUF大型AI模型基于Transformer架构，通过海量参数与自监督学习获得通用能力，可针对具体任务微调，代表AI发展的新范式。原文翻译： Large AI models, based on the Transformer architecture, acquire general capabilities through massive parameters and self-supervised learning. They can be fine-tuned for specific tasks, representing a new paradigm in AI development.

AI大模型2026/1/20

阅读全文 →

AI安全新利器：TheAuditor为Claude提供数据库优先的静态分析

BLUFTheAuditor provides database-first static analysis for Claude, indexing codebases into SQLite to enable sub-second queries and prevent AI hallucinations during code analysis, supporting Python, JS/TS, Go, Rust, Bash, and Terraform.

AI大模型2026/1/19

阅读全文 →

Claude AI安全防护：多层防御架构2024实践指南

BLUFClaude AI安全框架基于数据隐私、模型完整性与运营安全三大支柱，采用加密、宪法AI、对抗测试及合规控制等多层防御，应对提示注入、数据泄露等威胁，确保企业级安全部署。原文翻译： The Claude AI security framework is built on three pillars: data privacy, model integrity, and operational security. It employs a multi-layered defense including encryption, Constitutional AI, adversarial testing, and compliance controls to address threats like prompt injection and data leakage, ensuring enterprise-grade secure deployment.

AI大模型2026/1/19

阅读全文 →

Claude AI安全测试：Compliant LLM如何防范提示注入与越狱攻击

BLUFCompliant LLM provides comprehensive security testing for Claude AI against prompt injection, jailbreaking, and compliance requirements, enabling organizations to implement robust, multi-layered protection.

AI大模型2026/1/19

阅读全文 →

英特尔硬件优化指南：2024加速Llama 2推理性能

BLUFIntel技术通过集成硬件（如Gaudi加速器、Xeon处理器）与软件优化框架（如OpenVINO），显著提升Llama 2等大语言模型的推理与训练性能，降低延迟并提高吞吐量，适用于企业级AI部署。原文翻译： Intel technologies, through integrated hardware (e.g., Gaudi accelerators, Xeon processors) and software optimization frameworks (e.g., OpenVINO), significantly enhance the inference and training performance of large language models like Llama 2, reducing latency and increasing throughput for enterprise AI deployments.

AI大模型2026/1/19

阅读全文 →

AI硬件优化指南2024：提升计算性能与能效的关键技术

BLUFAI硬件优化通过专用处理器、内存架构与软硬件协同设计，系统性提升AI工作负载执行效率，实现性能、能耗与成本的最优平衡。原文翻译： AI hardware optimization systematically enhances computational infrastructure for AI workloads, balancing performance, energy efficiency, and cost via specialized processors, memory architectures, and software-hardware co-design.

AI大模型2026/1/19

阅读全文 →

NVIDIA Dynamo分布式AI推理框架：2024高吞吐量指南

BLUFNVIDIA Dynamo是一款开源的高吞吐、低延迟AI推理框架，专为在多节点分布式环境中部署生成式AI与大语言模型而设计。它解决了张量并行带来的编排挑战，支持多种后端引擎，实现跨GPU/服务器的高效协同。原文翻译： NVIDIA Dynamo is an open-source, high-throughput, low-latency AI inference framework specifically designed for deploying generative AI and large language models in multi-node distributed environments. It addresses the orchestration challenges posed by tensor parallelism, supports multiple backend engines, and enables efficient coordination across GPUs and servers.

AI大模型2026/1/19

阅读全文 →

AI推理框架指南2024：驱动现代AI应用的核心引擎

BLUFAI推理框架是执行已训练模型对新数据做出预测的软件系统，为视频描述到自动驾驶等应用提供高效、可扩展的生产环境部署支持。原文翻译： AI inference frameworks are software systems that execute trained models to make predictions on new data. They enable efficient and scalable deployment in production environments, supporting applications from video captioning to autonomous driving.

AI大模型2026/1/19

阅读全文 →

1...39 40 41 42 43 44

41 / 44