ChatGPT and Elasticsearch: OpenAI meets private data setup walkthrough
Вставка
- Опубліковано 31 лип 2023
- Follow along as Global Solutions Architect Peter Titov walks you through connecting Elasticsearch to OpenAI’s LLM for Retrieval Augmented Generation (RAG).
This detailed walkthrough will show you step by step how to implement the process described in this blog post: www.elastic.co/blog/chatgpt-e...
Additional Resources:
Start the 14-day trial for free! No credit card required: cloud.elastic.co/registration
Subscribe to Elastic’s Community YT channel: / officialelasticcommunity
Connect with us on social media:
LinkedIn: / elastic-co
Twitter: / elastic
Facebook: / elastic.co
About Elastic
Elastic is the leading platform for search-powered solutions, and we help everyone - organizations, their employees, and their customers - find what they need faster, while keeping applications running smoothly, and protecting against cyber threats. When you tap into the power of Elastic Enterprise Search, Observability, and Security solutions, you’re in good company with brands like Netflix, Uber, Slack, Microsoft, and thousands of others who rely on us to accelerate results that matter.
#Elasticsearch #generativeai #openai - Наука та технологія
Well done, thanks!
When adding an inference pipeline your demo shows 4 steps (configure, fields, test, review) however when I do try it I see 5 steps (configure, fields, mappings, test, review) for some reason even though I have dense showing in my mappings when I run a crawl I won't get any documents. Also I tried running removing the crawl rules and then I get a ton of documents but none of them are 'title-vector' so my chatbot can't find them.
great!
what if i want to crawl paywalled pages? (for which I have a paid subscription and credentials ofc)
Cool
@Elastic I am trying this out and the docker run command to deploy the model gives me the following error.
_SentenceTransformerWrapper.forward() missing 2 required positional arguments: 'token_type_ids' and 'position_ids'
facing same problem
is there any resolution to this problem
Do we have to possess an X thousand dollars license to use that solution ?
Really hard to follow because of version change and environment difference.
the link might not work, try to add /search-labs before /blog. Thank me later.