微信扫码
添加专属顾问
我要投稿
两个关键问题限制了 RAG 的发展:
新型 RAG 算法之间缺乏全面和公平的比较。
像 LlamaIndex 和 LangChain 这样的开源工具使用了高级抽象,这导致了透明度的缺失,并限制了开发新算法和评估指标的能力。
不同 RAG 库和框架的比较。公平比较指的是在评估过程中对所有基本组件进行对齐,包括随机种子、生成器、检索器和指令。数据收集器指的是能够收集或生成训练和测试数据的能力,无论是通过从现有的原始数据集中抽样,还是通过使用LLM构建标记数据。
RAGLAB提供了一个模块化的架构,允许用户轻松地替换和扩展算法的各个组成部分,包括检索器(retriever)、生成器(generator)和指令(instruction)。
RAGLAB 框架的架构和组件
检索器(Retriever):集成了基于BERT的模型,如Contriever和ColBERT,提供了统一的查询接口和客户端-服务器架构,以及检索缓存机制。
语料库(Corpus):提供预处理的Wikipedia语料库,包括2018年和2023年的版本,以及对应的索引和嵌入。
生成器(Generator):集成了Huggingface Transformers和VLLM,支持量化和低秩适应(LoRA)技术,允许使用大型模型。
指令实验室(Instruction Lab):包含系统指令、任务指令和算法指令,允许用户自定义和组合指令。
训练器(Trainer):集成了Accelerate和DeepSpeed库,支持模型的微调,包括LoRA和量化LoRA技术。
Naive RAG
# read process insruction
Instruction:\n {task_instrucion} \n## Input:\n\n{query}\n\n Now, based on the following passages
and your knowledge, please answer the question more succinctly and professionally. ### Background
Knowledge:\n {passages} \n\n### Response:\n"
RRR
# rewrite process instruction
a better search query for Wikipedia to answer the given question, end the query with '**'. \n\n
Question: Ezzard Charles was a world champion in which sport? \n\n Query: Ezzard Charles
\n\n Question: What is the correct name of laughing gas? \n\n Query: laughing gas
\n\n Question: {query} \n\n Query: " # read process insruction
Instruction:\n {task_instrucion} \n## Input:\n\n{query}\n\n Now, based on the following passages
and your knowledge, please answer the question more succinctly and professionally. ### Background
Knowledge:\n {passages} \n\n### Response:\n"
ITER-RETGEN
# read process insruction
Instruction:\n {task_instrucion} \n## Input:\n\n{query}\n\n Now, based on the following passages
and your knowledge, please answer the question more succinctly and professionally. ### Background
Knowledge:\n {passages} \n\n### Response:\n"
Self ASK
# follow up question instruction
When does monsoon season end in the state the area code 575 is located? Are follow up :
questions needed here: Yes. Follow up: Which state is the area code 575 located in? Intermediate
answer: The area code 575 is located in New Mexico. Follow up: When does monsoon season end in
New Mexico? Intermediate answer: Monsoon season in New Mexico typically ends in mid-September. So the final answer is: mid-September. \n{query} Are follow up questions needed here:" # read process insruction
Instruction:\n {task_instrucion} \n## Input:\n\n{query}\n\n Now, based on the following passages
and your knowledge, please answer the question more succinctly and professionally. ### Background
Knowledge:\n {passages} \n\n### Response:\n"
Active RAG
# read process insruction
Instruction:\n {task_instrucion} \n## Input:\n\n{query}\n\n Now, based on the following passages
and your knowledge, please answer the question more succinctly and professionally. ### Background
Knowledge:\n {passages} \n\n### Response:\n"
Self-RAG
# read process instruction
Instruction:\n{task_instruction}\n\n## Input:\n\n{query}\n\n### Response:\n"
RAGLAB 系统用户评估问卷
RAGLAB: A Modular and Research-Oriented Unified Framework for Retrieval-Augmented Generationhttps://arxiv.org/pdf/2408.11381https://github.com/fate-ubw/RAGLab
53AI,企业落地大模型首选服务商
产品:场景落地咨询+大模型应用平台+行业解决方案
承诺:免费场景POC验证,效果验证后签署服务协议。零风险落地应用大模型,已交付160+中大型企业
2025-04-30
速报!DeepSeek-Prover-V2-671B 悄然上线,或为 R2 铺路?
2025-04-30
“Qwen3之后,我才真正敢投AI应用”
2025-04-30
真·开源MCP平台来了!ACI.dev能一站直连600+工具,让你的智能体秒变全能王!
2025-04-30
n8n:免费+开源的自动化神器,比dify更简单,比Make更强大!
2025-04-30
宝藏发现:Sim Studio,一款让AI工作流搭建变简单的开源利器
2025-04-29
我们有必要使用 Qwen3 吗?
2025-04-29
Qwen3开源发布:Think Deeper, Act Faster!社区推理、部署、微调、MCP调用实战教程来啦!
2025-04-29
阿里 Qwen3 技术详解: 4B 参数媲美 72B,MoE 架构的开源突破
2024-07-25
2025-01-01
2025-01-21
2024-05-06
2024-09-20
2024-07-20
2024-07-11
2024-06-12
2024-08-13
2024-12-26
2025-04-30
2025-04-29
2025-04-28
2025-04-28
2025-04-28
2025-04-21
2025-04-19
2025-04-17