It is a bit misleading as in the title you talk about SVG, while you save them in png... too bad, would have been interesting to see that

Tomaz, thank you for sharing!

nice video good work

Why u reading from paper silly man? Condescending.

Thank you for the divine guidance

Thank you ,I have already built a text2sql module, which is not a query engine but a complicated workflow in itself with run time retrieval of rows, etc as suggested in other videos. Can I wrap this complicated text2sql workflow in a custom query engine and use the same steps

When is llamaindex cloud available? Waitlist is there quite long now

Predicting a DAG beforehand may be useful for research problems but for practical industry problems, generally, you cannot plan beyond 1st level in tree and therefore need to go back to plan again after each level thus losing advantage of this approach. Real world systems are extremely stochastic.

AttributeError: 'LlamaParse' object has no attribute 'get_charts'

wow! very Impressive!!!

did tomaz commit suicide immediatley after this presentation?

How to break up questions that have multiple parts, where understanding/answering one part affects how you form questions for the succeeding parts and answer these parts.

do i need paid subscription for llamaparse?

Thanks for informative video. I have a question. How to add support for integrating Google drive through OAuth ? I need this support in my SaaS where my customers can add their Google drive through OAuth

So when i understand it correctly, with RAFTDatasetPack i get a dataset to fine-tune for RAFT. Which fine-tuning method do i pick now though? Or in general: How is the fine-tuning done then. Any sources for that?

so verbose. who is going to use it like this ...

Good overview of more advanced workflows functionality. Thanks! Nevertheless, it seems that the code you are walking us through is not the same as the code in the linked repository? Could you provide us with (a link to) the updated code? Thanks in advance.

dude, it's so crazy, I watched this for work and when I got home my housemate asked me how hard it would be to make a data retrieval tool for Pub Med. so stoked!

How are pain points 11 and 12 related to RAG?

is it basically a NL2SQL method?

Hmm now I'm gonna have to optimize my resume for query XD

i cannot find the code...can yu please provide this one exactly?

Why wasn't the OpenAI Agent used here?

Hi, you can make a video about extracting financial data from PDF to calculate financial indicators. I am using llamaindex to research this aspect. I encountered a bottleneck. Sometimes llm cannot find the document content🥲🥲 Thanks

this is powerful stuff - if you make it easily useable/operatable for an average smb owner, this is easily multi-million dollar micro SaaS

What was the tool/lib mentioned for web search @2:45? Thank you.

can you use the same basic flow, but instead of wikipedia, use local pdf files parsed with pyMypdf4llm , which extracts imagse and dumps them out to local directory, just like in your cookbook?

Any examples with typescript?

llama index provides a framework for AI systems with a simple interface that is easy to integrate with other systems. I love its philosophy.

Colpali (Newer - ColQwen2 is Amazing). It retrieves the accurate pages, every single time really fast for each query that I passed. The only problem is there isn't any good OpenSource VLM's to run Locally. Qwen2 is good but not well for writing a proper response. I used Pixtral 12B. It was good enough but I observed Hallucination as well. Gonna share it in Colpali repository's readme section soon. Do check it out. Regards @DCparam

Hope this will truly help some use cases of advanced reporting in data engineering

Cheer~~~add (comments or supplementary information) at the end of a speech or text.😊

Very small fonts, and hard to see clearly. Suggest improve please.

I'm excited about LlamaIndex. When I started coding, everything I did basically revolved around Pandas. I get the sense that the next few years of my career will revolve around LlamaIndex. Thanks for the great videos!

Love this.

One of the cleanest approaches I’ve seen. Very clearly explained. Thanks

I love llama-index ❤, I think it has just the right level of abstraction to maintain control over the process out of the box and several ways to break down the abstractions to get more control where needed. It feels very comfortable to work with and the documentation is very clear. Thanks for this awesome framework 🎉🎉🎉🎉

Awesome

Are there any examples of report generation from structured data using LLMs?

As of November 4th 2024, the QueryPipeline is deprecated

Wow !! Amazing video - what a practical way of depicting rather than just talking of AI Hype !

hi. i've been trying to get acces to this for awhile now. No one is responding to any inquiries from your company. can you please let us know how to use this?

Could the agent incorporate articles from PubMed?

What is everyone's take on this vs LangChain. LangChain for me was quite overwhelming but I am a beginner so there's that.

How can we integrate the free SEMANTIC SCHOLAR api? How about PubMed please? Thank you!

There's a bewildering variety of useable and decently documented orchestration frameworks, and I'd hesitate to pick a favorite (yet), but ...LlamaIndex seems to have an outstanding commitment to avoiding unnecessary complexity.

What if i need to do the same but with PubMed to research adversial effects on a molecules ? what would be the differences ? I would need to look at the full text when available or title + abstract and make a custom report in a certain format ?

What app are you using to create the workflow diagrams?

If you want to explain the LlamaIndex workflows, I believe you would be far better off simplifying the example: less agents, less tools, less events ... for a first-time viewer this is getting quickly confusing and convoluted.

Thank you for the tutorial. Maybe I am missing something but it seems like this has the potential of getting into an infinite loop. Can can you limit the number of iterations of tool calls so you don't get a big bill from someone because the system got into an infinite loop of using the LLM and querying cloud hosted sql database? Also, does the text to sql have the ability to see if the sql statement was created correctly?