I wanna know which models vectara is using. And Langchain is involved in this Vectara?
Hi @Muaz_Ashraf and welcome to the forums!
Vectara has its own models. We’ve been working on training new models that are even better and we plan on updating them pretty soon (e.g. in the next few months) via a phased rollout.
As to Langchain, we’ve worked with the Langchain maintainers to add Vectara as a backend (you can see this post and this documentation)
1 Like
Hi, shane thank you for your response, can you tell me how you are using Langchain in Vectara? It would be very helpful for me. One another thing I wanna ask, if I upload same file twice it will again process in your database(Pinecone) or it will just say this file is already uploaded just chat with the existing file?
Also vectara is made in Java? vectara is also available in python?
can you tell me how you are using Langchain in Vectara?
We are not using Langchain in Vectara. We’ve made Vectara available via a Langchain integration. Hopefully that makes sense
One another thing I wanna ask, if I upload same file twice it will again process in your database(Pinecone) or it will just say this file is already uploaded just chat with the existing file?
To be clear, we do not use Pinecone in Vectara, but I think that isn’t that isn’t the point of your question. If you try to upload a file a 2nd time, Vectara will throw an error saying the file is already indexed. If you want to replace a file today, you need to do that by deleting the document first and then re-uploading a new one. We’re working on being able to update a document in place, but for now, this is the workflow for updating documents.
Also vectara is made in Java? vectara is also available in python?
Vectara uses a variety of technologies behind the scenes. But for end-users, you can use any language that can use REST and/or gRPC, which is basically all programming languages. We provide examples of many languages on our docs and also in our getting-started examples.
Thank you shane. I have another query. I integrate langchain with vectara to make a streamlit app but when I upload a file it gives me a response according to the previous uploaded file and in Vectara console, when I also go into the search bar it is also giving me previous file responses not the new one. If I share my code with you can you help me out?
I upload my new file to vectara console then it gave response to me according to new file, when i upload from streamlit it is not working however when i upload prev file from streamlit it works (2 file uploded from streamlit) and it has a source in vectara but when i upload my 3rd file from streamlit it does not have the source in vecatra
.One another thing it is giving me too much short response
Can you help me’
Hello, I want to ask, I deployed app streamlit and langchain using vectara. I created account in vectara add api key, customer id, corpus id in my .env file. I deployed successfully and it works fine for me but for another user it cause error bcz it give message No permissions to add document to Vectara.
Check your corpus ID, customer ID and API key. How can other user use easily without any error regarding vectara id’s
It would be great if you response me with an answer. Thank you. Regards Muaz.
I have to admit I’m not terribly familiar with streamlit or the langchain integration myself. But I’m ccing @ofermend here, as he helped with the langchain integration and he may have some familiarity with streamlit or be able to help guide you further
I just wanna know about in deployment case how we handle customer id, corpus id or api. bcz other user is not able to upload file (No permission) if @ofermend help. It whould be great help.
To find your customer ID, you can click your name in the upper right hand corner of the console:
To find your corpus ID, you can navigate to a corpus and find it here (e.g. 117 here):
You’ll need to create an API key. You can do so at Vectara - Semantic Search
Hi,
Happy to help if I can. Can you please provide more information about the other user - are they using the same customer ID (i.e. same Vectara account) and same corpus? Same API key or not?
Ofer
I’m new to Vectara - just logging in for the first time - and not sure I’m asking this question in the right place.
I don’t have a coding background. I’m an editor.
I’m looking for a system where I can upload long PDFs, mostly of transcripts of conversations. I want to upload an outline - a series of headings. And have the algorithm go through and tag each paragraph in the transcript with the most appropriate heading.
Can Vectara do that? Is there an off-the-shelf way for it to do that? Is this an application for Boomerang?
Thank you.
Hi Art,
Sorry for the late response. From what you describe, it seems like you want to match each paragraph with a heading/tag (from a pre-made list of such tags). Is that right? How many tags do you have, roughly? are those just one-word tags or do the headings constitute of multiple words each?