LlamaIndex
Log in

Find answers from the community

Home
Members
aelita
a
aelita
Offline, last seen 3 months ago
Joined September 25, 2024
aaelita
·

Hi guys, using LlamaParse is suddenly

Hi guys, using LlamaParse is suddenly taking a lot longer than it was before, and I'm now getting this error repeatedly before my document actually begins parsing: "HTTP Request: GET https://api.cloud.llamaindex.ai/api/parsing/job/206bca8a-b28c-4d35-a48e-91d4af8ba45d/result/text "HTTP/1.1 404 Not Found""
4 comments
a
L
aaelita
·

Is it possible to use an ingestion

Is it possible to use an ingestion pipeline with a node parser or are they mutually exclusive?
5 comments
L
a
aaelita
·

When trying to use LlamaParse, I'm

When trying to use LlamaParse, I'm getting "Error while parsing the PDF file: Currently, only PDF files are supported." even though I am uploading a PDF 🤔
6 comments
L
a
aaelita
·

Hi all, when trying to use OpenAI's '

Hi all, when trying to use OpenAI's 'text-embedding-3-large' embeddings with a Qdrant client, I get a "ValueError: shapes (469,3072) and (1536,) not aligned: 3072 (dim 1) != 1536 (dim 0)" error.

I think this has to do with the embedding model producing vectors of a size that Qdrant doesn't expect, but I'm not sure how to fix. I've tried setting the Qdrant client vector parameter of size to 3072 but this doesn't help -- the error becomes "ValueError: operands could not be broadcast together with shapes (469,3072) (1536,)". Any ideas would be appreciated, thanks!

Plain Text
client = qdrant_client.QdrantClient(location=":memory:")
    client.create_collection(collection_name="collection_name",
    vectors_config=VectorParams(size=3072, distance=Distance.EUCLID), #FIXME
)

    vector_store = QdrantVectorStore(client=client, collection_name="collection_name")
8 comments
L
a
aaelita
·

Hi all -- I'm building a RAG application

Hi all -- I'm building a RAG application with LlamaIndex and am trying to use DeepEval for testing. In particular, I'm working on faithfulness testing and am trying to figure out how to extract the context string for the retrieval_context parameter in DeepEval. I'm playing around with the no_text response synthesizer but am running into issues extracting an actual context string. I'm new to LLMs/RAG so any help would be appreciated, thanks!
2 comments
r
aaelita
·

What are some techniques for improving reliability of responses for RAG? I set my LLM temp

@kapa.ai What are some techniques for improving reliability of responses for RAG? I set my LLM temperature to 0 in settings but using all the same parameters, am seemingly randomly either getting the correct response or getting something like "the context doesn't provide information about ___".
3 comments
k
aaelita
·

I see download_loader is deprecated now

I see download_loader is deprecated now and the solution is downloading the actual package itself. I'm trying to load PDFReader but am not sure what the actual package is called (it's not PDFReader as far as I can tell)
8 comments
a
L
aaelita
·

I'm in the process of adapting to

I'm in the process of adapting to version 0.10.3 and am getting "ImportError: cannot import name 'DFFullProgram' from 'llama_index.core.program'" -- any thoughts?
2 comments
a
L
aaelita
·

Hi everyone, I'm building a RAG

Hi everyone, I'm building a RAG application for complex PDFs and am running into a strange issue where my indexing step is suddenly taking a lot longer than it used to. I am hosting my app on Streamlit community cloud, and indexing still takes a reasonable amount of time there (~45 seconds) whereas it suddenly takes a good couple of minutes+ locally (and shows a lot more progress bars than it used to). The code hasn't changed at all (I also tested this with an earlier version of my code that ran fine locally before), so I'm confused as to what could've caused this. I've tried multiprocessing during my ingestion pipeline to speed things up, but it still doesn't index nearly as fast as it used to. The data I'm using has not changed at all, either. I'm assuming the issue is with my machine or perhaps dependencies, but any ideas would be much appreciated, thanks!
3 comments
a
L