Video 2 of my career. Learn how to build Evaluation Pipelines with Haystack's latest release that introduced a number of both statistical and model-based evaluators.

In this coding session, we're walking through the Evaluating RAG tutorial (link in comments).
Retrieval…

📊 Evaluating RAG Pipelines

RAG pipelines consist of two fundamental steps: Retrieval and Generation.

To evaluate a full RAG pipeline, we must assess each step individually and as part of the overall process. The retrieval step can be evaluated using statistical metrics that…

⭐️ Haystack 2.1.0 ⭐️

Seems like yesterday we released Haystack 2.0, and today we're back with Haystack 2.1, fresh out of the oven 🥐

This release includes brand-new functionalities for Haystack.

📊 New Evaluators for both model-based and statistical evaluation of RAG pipelines…

Are you new to Haystack? Start by learning the fundamentals of Haystack pipelines and components.

Check out this run-through by Tuana 👇

youtube.com/watch?v=SxAwye…

Going to PyCon US US? Looking to connect with other engineers in AI?

🥂 We are partnering with @datastax to throw a happy hour just a few blocks from the conference. Come hang with us!

📆 May 17th, 2024, 6pm - 9pm
🍻 The Standard Market & Pint House
lu.ma/pycon-ai-happy…

🔎 Sparse embedding retrieval in Haystack!

Keyword-based retrieval methods like BM25 are efficient but lack semantic understanding.
On the other hand, dense vector retrieval requires considerable computational resources and may struggle in new domains.

💡 SPLADE, a sparse…

🍁 Fallback paths 🍁
These are quite handy if you'd like to implement some sort of 'behavior' for your application when certain conditions are met.

In the most recent example by Tuana, we create a fallback path for cases where a question cannot be answered by the SQL…

🐍 Bilge was on the PyConDE & PyData Berlin stage explaining fallback mechanisms to improve the LLM-based applications using Haystack!

🧑🏻‍🏫 If you couldn't catch her insightful talk, learn how to implement fallbacks with conditional routing in our tutorial 👉 haystack.deepset.ai/tutorials/36_b…

RAG systems encounter challenges in real-world scenarios, such as struggling to generate meaningful answers from retrieved data or providing any answer at all.

Join Bilge at PyConDE & PyData Berlin tomorrow as she explores strategies for improving LLM-based applications with fallback…

The new Llama3 models by AI at Meta are impressive. There are 4 variants: 8B, 70B, 8B instruct, 70B instruct. More to come! 🚀

🧑‍🍳🥂 Example Cookbook of RAG on Oscar night data 👉
Stack: Haystack + Snowflake Arctic embeddings + Llama3 8B instruct
🧡 colab.research.google.com/github/deepset…

Are you at PyConDE & PyData Berlin? Catch our Haystack team at the conference 🚀

🎙️ Mon, 12:15 PM: Silvano Cerza on “Haystack 2.0: the story of a rewrite” pretalx.com/pyconde-pydata…

🎙️ Wed, 12:15 PM: Bilge on 'Improve LLM-based Applications with Fallback Mechanisms'
pretalx.com/pyconde-pydata……

🌟 MARK YOU CALENDARS 🌟

Get ready for a great evening exploring the world of LLMs, vector stores, knowledge graphs, and cutting-edge models! Join us for insightful talks featuring experts from deepset, Neo4j, Qdrant, and mixedbread.ai!

​🎙️Talks

​1.…

Join the Haystack Neo4j Qdrant and mixedbreadai teams next week for an AI meetup in Berlin 🚀👇

Congrats to the Qdrant team. Try out the new Qdrant Hybrid Cloud with Haystack 👇

Congratulations to the Qdrant team for the release of Qdrant Hybrid Cloud 🚀
Stay tuned to discover how to use this with you Haystack pipelines 👏

Did you miss Tuana and Craig Dennis's livestream yesterday on using Llama Guard to add guardrails to your LLM apps?
📺 You can watch it now on YouTube
Go and see how they will keep you from talking about pizza with an LLM if they want to 🍕

youtu.be/iCjccmLaJ7U?si…

Congrats to cohere on the release of their new reranking models. Want to give them a spin? Bring them into your AI pipelines with Haystack 🚀

Livestream with Haystack and Cloudflare Developers this Friday! Learn how to use Llama Guard for your LLM interactions 👇

It's great to see this collaboration take shape! Thank you to the MongoDB team for their support in this partnership!

I'm happy to announce that the MongoDB integration for Haystack has been expanding, MongoDB Atlas is now available for Haystack 2.0, and you can now try it…

We released some small fixes and improvements to Haystack yesterday 💪

· streaming_callback added to the HuggingFaceLocalGenerator to allow you to stream responses from this generator as well
· A new class: 'SparseEmbedding' is added, which will allow us to expand retrieval to…