<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[bbbeoy的专栏]]></title><description><![CDATA[]]></description><link>https://blog.csdn.net/bbbeoy</link><language>zh-cn</language><generator>https://blog.csdn.net/</generator><copyright><![CDATA[Copyright &copy; bbbeoy]]></copyright><item><title><![CDATA[From Basics to Advanced: Exploring LangGraph]]></title><link>https://blog.csdn.net/bbbeoy/article/details/143594007</link><guid>https://blog.csdn.net/bbbeoy/article/details/143594007</guid><author>bbbeoy</author><pubDate>Thu, 07 Nov 2024 14:50:06 +0800</pubDate><description><![CDATA[agents]]></description><category></category></item><item><title><![CDATA[Continue PreTrain]]></title><link>https://blog.csdn.net/bbbeoy/article/details/142358733</link><guid>https://blog.csdn.net/bbbeoy/article/details/142358733</guid><author>bbbeoy</author><pubDate>Thu, 19 Sep 2024 14:16:15 +0800</pubDate><description><![CDATA[LLM Continue Pretrain（2024版）

如何更好地继续预训练（Continue PreTraining)


]]></description><category></category></item><item><title><![CDATA[Speculative Sampling]]></title><link>https://blog.csdn.net/bbbeoy/article/details/142208911</link><guid>https://blog.csdn.net/bbbeoy/article/details/142208911</guid><author>bbbeoy</author><pubDate>Fri, 13 Sep 2024 14:15:37 +0800</pubDate><description><![CDATA[.]]></description><category></category></item><item><title><![CDATA[KV cache]]></title><link>https://blog.csdn.net/bbbeoy/article/details/141905019</link><guid>https://blog.csdn.net/bbbeoy/article/details/141905019</guid><author>bbbeoy</author><pubDate>Wed, 04 Sep 2024 22:04:11 +0800</pubDate><description><![CDATA[大模型百倍推理加速之KV cache篇

大模型推理优化技术-KV Cache

Efficient Streaming Language Models with Attention Sinks


]]></description><category></category></item><item><title><![CDATA[RLHF PPO DPO]]></title><link>https://blog.csdn.net/bbbeoy/article/details/141858301</link><guid>https://blog.csdn.net/bbbeoy/article/details/141858301</guid><author>bbbeoy</author><pubDate>Tue, 03 Sep 2024 21:55:41 +0800</pubDate><description><![CDATA[生成式大模型的RLHF技术（一）：基础​​​​​​​





DPO: Direct Preference Optimization 论文解读及代码实践

深入对比 DPO 和 RLHF

深入理解DPO（Direct Preference Optimization）算法
]]></description><category></category></item><item><title><![CDATA[Flamingo：通向多模态AI的里程碑]]></title><link>https://blog.csdn.net/bbbeoy/article/details/141434538</link><guid>https://blog.csdn.net/bbbeoy/article/details/141434538</guid><author>bbbeoy</author><pubDate>Thu, 22 Aug 2024 18:16:04 +0800</pubDate><description><![CDATA[Flamingo：通向多模态AI的里程碑
]]></description><category></category></item><item><title><![CDATA[扩散模型介绍]]></title><link>https://blog.csdn.net/bbbeoy/article/details/141433350</link><guid>https://blog.csdn.net/bbbeoy/article/details/141433350</guid><author>bbbeoy</author><pubDate>Thu, 22 Aug 2024 17:33:34 +0800</pubDate><description><![CDATA[综述 - 扩散模型 - Diffusion Models

轻松学习扩散模型（diffusion model），被巨怪踩过的脑袋也能懂——原理详解+pytorch代码详解（附全部代码）

由浅入深了解Diffusion Model


]]></description><category></category></item><item><title><![CDATA[An Introduction to Diffusion Models and Stable Diffusion]]></title><link>https://blog.csdn.net/bbbeoy/article/details/141399638</link><guid>https://blog.csdn.net/bbbeoy/article/details/141399638</guid><author>bbbeoy</author><pubDate>Wed, 21 Aug 2024 18:31:26 +0800</pubDate><description><![CDATA[theTtt+1βₜt-1t. APfromQt,Encoder Ett-1。]]></description><category></category></item><item><title><![CDATA[Segment Anything(SAM) & FastSAM]]></title><link>https://blog.csdn.net/bbbeoy/article/details/141398306</link><guid>https://blog.csdn.net/bbbeoy/article/details/141398306</guid><author>bbbeoy</author><pubDate>Wed, 21 Aug 2024 17:21:38 +0800</pubDate><description><![CDATA[一篇文章搞懂Segment Anything(SAM)

Segment Anything(sam)项目整理汇总

【SAM】SAM &amp; Fast SAM &amp; Mobile SAM
]]></description><category></category></item><item><title><![CDATA[AI agent]]></title><link>https://blog.csdn.net/bbbeoy/article/details/141227140</link><guid>https://blog.csdn.net/bbbeoy/article/details/141227140</guid><author>bbbeoy</author><pubDate>Fri, 16 Aug 2024 21:37:06 +0800</pubDate><description><![CDATA[LLM Powered Autonomous Agents

万字长文解析AI Agent技术原理和应用


]]></description><category></category></item><item><title><![CDATA[DETR]]></title><link>https://blog.csdn.net/bbbeoy/article/details/141196098</link><guid>https://blog.csdn.net/bbbeoy/article/details/141196098</guid><author>bbbeoy</author><pubDate>Wed, 14 Aug 2024 18:49:15 +0800</pubDate><description><![CDATA[DETR目标检测新范式带来的思考

详细解读DETR，基于transformer的目标检测网络
]]></description><category></category></item><item><title><![CDATA[Multilingual Language Model]]></title><link>https://blog.csdn.net/bbbeoy/article/details/140892024</link><guid>https://blog.csdn.net/bbbeoy/article/details/140892024</guid><author>bbbeoy</author><pubDate>Sat, 03 Aug 2024 16:03:38 +0800</pubDate><description><![CDATA[Multilingual多语言预训练的那些套路
]]></description><category></category></item><item><title><![CDATA[Understanding Low-Rank Adaptation (LoRA) for Efficient Fine-Tuning of Large Language Models]]></title><link>https://blog.csdn.net/bbbeoy/article/details/140716311</link><guid>https://blog.csdn.net/bbbeoy/article/details/140716311</guid><author>bbbeoy</author><pubDate>Fri, 26 Jul 2024 15:12:17 +0800</pubDate><description><![CDATA[This blog post will go into detail about how LoRA works to fine-tune LLMs, following the methodology set out in the “LoRA: Low-Rank Adaptation of Large Language Models” paper
]]></description><category></category></item><item><title><![CDATA[大模型微调]]></title><link>https://blog.csdn.net/bbbeoy/article/details/140554891</link><guid>https://blog.csdn.net/bbbeoy/article/details/140554891</guid><author>bbbeoy</author><pubDate>Fri, 19 Jul 2024 18:26:46 +0800</pubDate><description><![CDATA[觉得这位美女写的好，就拿来了。]]></description><category></category></item><item><title><![CDATA[RAG技术概览]]></title><link>https://blog.csdn.net/bbbeoy/article/details/140554303</link><guid>https://blog.csdn.net/bbbeoy/article/details/140554303</guid><author>bbbeoy</author><pubDate>Fri, 19 Jul 2024 17:04:31 +0800</pubDate><description><![CDATA[RAG]]></description><category></category></item><item><title><![CDATA[Prompt Engineering]]></title><link>https://blog.csdn.net/bbbeoy/article/details/140547378</link><guid>https://blog.csdn.net/bbbeoy/article/details/140547378</guid><author>bbbeoy</author><pubDate>Fri, 19 Jul 2024 14:31:53 +0800</pubDate><description><![CDATA[提示工程(Prompt engineering)是一门相对较新的学科，旨在为各种应用和研究主题开发和优化提示，以有效地利用语言模型（LMs:language models ）。提示工程技能有助于更好地了解大型语言模型（LLMs:large language models ）的能力和局限性。研究人员使用提示工程来提高 LLMs 在各种常见和复杂任务（如问答和算术推理）上的能力。开发人员使用提示工程来设计稳健且有效的提示技术，与 LLMs 和其他工具进行交互。自然语言处理的标准任务之一是文本摘要。]]></description><category></category></item><item><title><![CDATA[大模型基础组件 - Tokenizer]]></title><link>https://blog.csdn.net/bbbeoy/article/details/140544178</link><guid>https://blog.csdn.net/bbbeoy/article/details/140544178</guid><author>bbbeoy</author><pubDate>Fri, 19 Jul 2024 11:17:57 +0800</pubDate><description><![CDATA[这里的动机是一个pair的频率很高，但是其中pair的一部分的频率更高，这时候不一定需要进行该pair的合并。通过这种方式可以更好的处理跨语言和不常见字符的特殊问题(例如，颜文字)，相比传统的BPE更节省词表空间（同等词表大小效果更好），每个token也能获得更充分的训练。1. 根据不同的切分粒度可以把tokenizer分为: 基于词的切分，基于字的切分和基于subword的切分。这是当前大模型的主流分词方案。基于subword的切分能很好平衡基于词切分和基于字切分的优缺点，也是目前主流最主流的切分方式。]]></description><category></category></item><item><title><![CDATA[EfficientNet_V2 ShuffleNet_V2 MobileNets_V3 模型算法详解]]></title><link>https://blog.csdn.net/bbbeoy/article/details/140530321</link><guid>https://blog.csdn.net/bbbeoy/article/details/140530321</guid><author>bbbeoy</author><pubDate>Thu, 18 Jul 2024 18:35:58 +0800</pubDate><description><![CDATA[图像分类】【深度学习】【轻量级网络】【Pytorch版本】EfficientNet_V2模型算法详解
]]></description><category></category></item><item><title><![CDATA[一篇文章搞懂LLaVA-Plus]]></title><link>https://blog.csdn.net/bbbeoy/article/details/140529930</link><guid>https://blog.csdn.net/bbbeoy/article/details/140529930</guid><author>bbbeoy</author><pubDate>Thu, 18 Jul 2024 18:12:01 +0800</pubDate><description><![CDATA[一篇文章搞懂LLaVA-Plus
]]></description><category></category></item><item><title><![CDATA[Next-GPT: Any-to-Any Multimodal LLM]]></title><link>https://blog.csdn.net/bbbeoy/article/details/140529589</link><guid>https://blog.csdn.net/bbbeoy/article/details/140529589</guid><author>bbbeoy</author><pubDate>Thu, 18 Jul 2024 17:56:09 +0800</pubDate><description><![CDATA[https://zhuanlan.zhihu.com/p/658317147https://zhuanlan.zhihu.com/p/663002368]]></description><category></category></item></channel></rss>