วีดีโอ

That's great! Thanks for your video

Coming from reading the paper and watching a couple of explanations and found the code really clear and much simpler than I thought it could be! Thanks for the video man!

What a nice content

Thank you for the informative video on implementing diffusion models. I have been following along and made some modifications to improve the clarity of the generated images, such as increasing the number of steps and epochs and adjusting the learning rate. However, the final images are still not as clear as I would like them to be. Could you please provide some guidance on how to further improve the quality of the output images?

nice

Y=32!×43×11

Wow! Never seen such a good content on here

Great

Great

Thanks for the video, it is very informative. It's amazing that all you had to do was give the SQLDatabaseChain() the reference to the SQLite Db and the LLM and it was able to figure out the correct joins, but what if I had an existing schema that had hundreds of tables and I only wanted to expose a subset of those tables to the LLM for the creation of a SQL statement to satisfy a users inquiry, how would I restrict the tables to be used? Also, you bring up a good point about using OpenAI's LLM and table names becoming public. Have you tried and can recommend other LLM's? Thanks for sharing.

Great work, thx

Can one do zoom and pan in IJulia ?

To the point! Loved it

Nice and clean. This is such a generous gift. I am new at this. I will use your code as a template for future training flows. Thank you for sharing your knowledge.

Thanks for the video and the work you put in sharing this. Very helpful so far! I followed the instructions and it executed just as shown in the video. But where is the fine-tuned model saved to? After execution, I only find an empty folder, named with the current time in the root of the python penv.

Thanks for sharing 💕💕

fantastic work; for someone who isn't a coder, it's a gold mine. I'm new to the LLM space and have an inquiry: since the PDF I extracted isn't formatted, how can I create a dataset with embeddings to finetune the LLM model? Additionally, I'd like your advice on structural modelling for large Txt databases-should I use llm or another method (I want to combine hundreds of PDFs into a single structural hierarchy document that summarises information without losing context. Is this possible locally?)?

This was really useful. I'm trying to build(improve) an LLM that could convert Redshift Sql to Athena Sql or Spark Sql. I'm thinking of using this to train over something like Mistral 7B.

Man I’m glad I subbed to your channel. What a great walkthrough

awesome tutorial, thanks!

Nice video.I am a junior data analyst and was figuring if it is possible to use the same concept on a existing db with like hundreds of tables. Was wondering bc I got some (hard) business inquiries but got little explanation about the db itself and was thinking about a way of having a conversation with your db. Let me know your thoughts

Somehow the 3D Plot isn´t working. What am I doing wrong? (I am using VS Code)

I would say the headline is misleading, as this tutorial is not really training the model, but rather indexing PDF's and exposing the most relevant parts to the OpenAI prompt with a question.

Perfect!

Thanks for this video, it's super helpful :) I'm newer to using langchain and Chroma - how would I get the indexes to persist using Chroma? In another project I used "vectordb = Chroma.from_documents(documents=texts, embedding=embedding,persist_directory=persist_directory)" with a text splitter, but am unsure on how to persist the index created by VectorIndexCreator.

Hi I am trying to simulate the project showed in the video but I am having doubt that how to write a general prompt and then generate sql by passing table and column information.

how to ask questions to github repositories so the model understands programming languages?

great and helpful vid :) ! can this setup be used with a non Open-AI platform ?

I understand that we are setting OpenAI API key in this code, but where are we using it? Where have we mentioned that we want to use some LLM model? Is this cod even using some LLM model?

Can we pluggin LLAMA models by facebook?

After getting all dependencies, I am not able to get langchain to work properly after import os. Do you have any suggestions?

Thank you for putting together this demo. I am going to have to try this out. I worry, however, that it will not do so well with a real-world database, one with various foreign keys, one-to-many relationships, abbreviated table and column names. I have no idea how it would be able to make sense of such a schema seeing as how I barely can.

What website or program are you using to run this? I was trying to run it in VS Code and got an error after index = VectorstoreIndexCreator().from_loaders(loaders) - "Output exceeds the size limit. Open the full output data in a text editor"

Can you use Alpaca with LangChain?

Is it possible any other Llm instead of OpenAI model. without using OpenAI Api key. Is there any way? could you guide ?

I'm not clear why Detectron2 is needed, given that we are not doing object detection here?

I stoped using julia because of plots are really slow 🙁

You can't read anything! The quality of the video is very bad...a pity

I found this useful