TGTGInsighttelegram intelligenceLIVE / telegram public index
← GitHub Trends

TGINSIGHT SIMILAR POSTS

Find similar content

Source channel @githubtrending · Post #14993 · Jul 24

#jupyter_notebook Retrieval Augmented Generation (RAG) helps large language models (LLMs) answer questions using up-to-date or private information by connecting them to external data sources, unlike fine-tuning which retrains the model on specific data. RAG is useful when you need current, dynamic information without costly retraining, making it ideal for tasks like customer support or knowledge management. Fine-tuning is better for deep expertise in a specialized field but requires more data and effort. Using RAG lets you get accurate, relevant answers quickly by combining the model’s language skills with fresh, specific data, improving usefulness and reliability. https://github.com/langchain-ai/rag-from-scratch

Results

1 similar post found

Search: #openfuture

当前筛选 #openfuture清除筛选
AI & Law

@ai_and_law · Post #341 · 06/28/2024, 07:04 AM

Implementing Transparency in AI: A Step Forward Zuzanna Warso and Paul Keller from Open Future, alongside Maximilian Gahntz from Mozilla, have published a proposal to implement the EU AI Act’s training data transparency requirement for general-purpose AI (GPAI). Article 53 1(d) of the Act mandates GPAI model providers to publish detailed summaries of their training content, covering data sources and sets with narrative explanations. The proposed template emphasizes a comprehensive scope and sufficient technical detail to benefit both experts and laypeople. These summaries should list primary data collections, provide narrative explanations of other data sources, and clearly distinguish between 'data sources' (origins) and 'datasets' (processed data points). This transparency requirement aims to enhance accountability, enable research and scrutiny, and strengthen individuals' and organizations' ability to exercise their rights in the AI development process. #AI#Transparency#AIAct#DataGovernance#OpenFuture#Mozilla