Julian Risch

icon indicating a website link https://www.deepset.ai/ icon indicating an email [email protected]

icon company deepset icon location Berlin, Germany icon location Engineering Team Lead @deepset-ai working on Haystack, the production-ready open source AI framework

Results 172 comments of Julian Risch

Replace openapi3 library with our own implementation

related PR: https://github.com/deepset-ai/haystack/pull/7514

creating an alternative in-memory bm25 store and retriever

Dear @Guest400123064 thank you for reaching out! Great repo, we saw your LinkedIn post. 👏 🙂 You're right that the current bm25 implementation in Haystack is suboptimal. It really recreates...

creating an alternative in-memory bm25 store and retriever

@Guest400123064 It would suggest to leave out the changes of the filtering and tokenization logic for the first PR, yes. Smaller changes make it easier to review and merge. The...

Add the memory feature to the library to enable chat + RAG

@vblagoje Let's define what needs to be done in this issue together today. I have in mind breaking it up into four smaller issues and we can split the work....

Add the memory feature to the library to enable chat + RAG

Memory components were added to haystack-experimental and the cookbok repo contains a notebook showing Chat + RAG. Let's close this issue and focus instead on new tool abstractions as part...

Update the eval tutorial

We can rename the component to AnswerGroundednessEvaluator if that is more intuitive? --- View entire conversation on ReviewNB

Update the eval tutorial

AnswerFaithfulnessEvaluator or AnswerHallucinationEvaluator are alternatives. --- View entire conversation on ReviewNB

Update the eval tutorial

@TuanaCelik The only change needed in this tutorial that is caused by the [PR](https://github.com/deepset-ai/haystack/pull/7621) I just merged is that FaithfulnessEvaluator's input parameter `responses` was renamed to `predicted_answers`. And if you...

RequestError: RequestError(400, 'mapper_parsing_exception', 'The number of dimensions for field [embedding] should be in the range [1, 2048] but was [12288]')

Hello @AI-Ahmed I hope you are doing well too. I would suggest that you use a different document store, for example FaissDocumentStore. Elasticsearch limits the vector dimensions and I don't...

`params` in Agent `run()` function not being passed to pipelines

Passing params to pipelines is supposed to work while passing params to nodes is not supported. If it doesn't work with a pipeline, we need to further investigate here: https://github.com/deepset-ai/haystack/blob/c20286609385b6d0b5cf0a174ec8775fbbd50c33/haystack/agents/base.py#L85