Видео

this video contains no info.... not clear why we needed a separate video for this

Friendly advice - you need to introduce your topics before speaking about them. This industry is just getting going, and already strange terminology and acronyms are appearing. You are both talking about very basic concepts, but for some reason speaking in another language that makes it impossible to follow. The only reason that anything is confusing is because people assume listeners know what they know.

Guys! This was awesome! Thanks!

my brain just broke-down on minute 25... I love the in-depth view, pity i was not to level... Luckily we got back to my level soon after... Very good interview...

Thank you so much for joining the podcast Krista! It was awesome getting to discuss all these things with you! Go MIPRO and Go DSPy!!

So there is no direct access to the embeddings storage, i.e. not in the same runtime, only fetching them via API. Which is basically clones vectors in the memory of a server(s) where Weaviate is located. Would be more useful to be able to work directly with the embeddings inside persistence layer. Performing clustering algorithms, launching different regressions algorithms etc. The only benefit i see, is the removal of the round-trip of the data stored in the vector database. I.e. app <-> vector database. As all of the module logic can be implemented in the app, no need to develop module for that.

Is there anywhere I can find this great power point?

Excellent! I found the video very useful

Thanks, great video.

Another great video by Zain. It will be great of you can share the slides or links used in the slides. Thanks in advance.

Bros head is about to fall off

I wonder if it can read data from a sql database directly ?

Think it's pronounced Gleaner as in you glean info from other info. Thank for the video.

can you share code please?

Great discussion!!

Really amazing innovation by Weaviate team! Their workshops speaks about their groundbreaking applications and making it open-source.

is there an implementation for this out there ?

great video

I used RAG in my life coach RUclipsr AI character in Telegram. Prepared a lot of AI processed data of him, with fine-grained segments. The most amazing AI experience for me yet - JulienHimselfBot

In the description

super exciting episode

Congrats Everyone. Well done indeed. Kudos!

I love this video! Thank you very much!!!

Thank you so much for joining Lucas! Odapt!! Super interesting example of an AI-native application, really loved the discussion of native multimodality as well!

This is gotta be one of the weirdest videos I’ve ever seen on RUclips. The content is so awesome and great but the camera creates some kind of a surreal image. It’s like hyper realism or something. The image quality is so immersive. It’s almost distracting so weird.

feels like I am watching bunch of robots talking

Is this a generated video?

Like the tech, but the health use case is really clunky and very simple. Medical data is messy and you would obviously be asking patients most of these questions, not an LLM!

Verba ist wirklich ein super Tool, bin sehr begeistert! Integriert ihr auch GraphRAG (von Microsoft) in das Tool?

Very helpful , I am able to set up on local host , however not able to ingest data, trying to upload a PDF

Love this! Thanks for the shoutout. I love that the signal processing example for quantization is now named after me, but I have to say I didn't come up with it, either. The Wikipedia article for quantization actually mentions the Analog-Digital-Converter example. I think I probably got it from there 🙃

I Have an idea that you could implement.

Thank you so much for joining the podcast both! I loved learning about ACORN to improve Vector Search with Filtering. The whole journey of learning about ACORN has also really helped my understanding of HNSW more broadly. Super exciting as always to see more research in Approximate Nearest Neighbor Search!

Richtig nice!

Always look forward to Connor’s hyped introductions. Great discussion. Nils sentence transformer work got me into this space in the first place. Looking forward to further advancements in multi-attribute embeddings. Would love more podcasts on sota embeddings and new directions. Been meaning to try out newer approaches like nv-embed and gritlm. Have you explored these?

Is the console only available for cloud instances? Or is there a way to have a local console or something for local docker instances?

Thank you so much for joining Josh! Learned so much from this conversation! Window Search Tree!

Another great podcast!

can anyone help me with the alternatives to the verba?

Wow, Victoria, is a natural in front of the Camera, perfect presentation and body language movement just right for the level of presentation required. It almost looks like she went to school for acting, NLP and Micro Expression presentation for marketing.

This is beautiful. I'm working with 5 classmates (electromechanical and software engineering college) on a proyect, we developed a tiny robot able to chat with patients as a co therapist, using a raspberry pi and a LLM. But the hallucinations are way too dangerous here, so i suggested to my team we start implementing RAG. Generating and "validating" the psychology database is really, really, really time consuming, it's hard, tricky and it takes a long time to have good quality examples, but we are pretty sure it's gonna be 100% worth it. I just had knowledge of Ollama and i would love to try out Verba in our prototype, so people in need can start getting attention and we can right away start recollecting data from the final model already deployed. I would love to collaborate with you guys, I'm such an enthusiast of opensource communities and corporations, and I loved the concept you evoke so much. <3

With all that nodding he must be listening to Kanye.

Cool video, but please fix the volume, it gets really low. There are many easy ways to keep a constant volume, even by using a simple phone app. RAG is much more complex than volume levels!!

Question ruclips.net/video/DFqd34ikTH0/видео.html - rewriting 5B tokens with 1$/1M tokens inferences is ~5000$. Nils cites this number as "hunderts of millions" - Where is this number coming from? Thats 1000x.

This is very cool! Thank you! Can I distribute just the chat interface without the configuration behind?

Excellent convo! My team and I are just about to start working on a RAG system for a client and I have to admit we didn’t give embedding models much consideration but now will definitely spend some time optimising the best embedding model for our use case. 🙌🏼

Additional resources: - Product quantization blog (weaviate.io/blog/pq-rescoring) - Binary quantization blog (weaviate.io/blog/binary-quantization) - Weaviate Academy: Vector compression (weaviate.io/developers/academy/py/compression) - Weaviate Documentation: Vector quantization (weaviate.io/developers/weaviate/concepts/vector-quantization) - Lecture: Vector Quantization (Etienne at Brown University) (ruclips.net/video/0diVrgyQwXA/видео.html) - Hugging Face Blog: Binary and Scalar Embedding Quantization (huggingface.co/blog/embedding-quantization)

Your "All Your Base Are Belong To Us" ending just earned a subscription from me. WHAT YOU SAY!

Great!!!