Find answers from the community

Updated 2 months ago

Hi guys may I check whether the

Hi guys may I check whether the llamaparse extraction model / method has been changed recently? Previously i had good results (about a week ago around 24th September) extracting the text and pages correctly, but now it looks like the text is not being detected right for most of the pages. Just wondering if any changes has been made to Llamaparse recently in terms of the repo that could have caused this?

Am running this code:
Plain Text
        parser = LlamaParse(
            api_key=llama_api_key,
            result_type="markdown",
            verbose=True,
            disable_ocr=True
            )

parser.get_json_result(self.input_filepath)
W
T
S
16 comments
hi, did you add any new change or any new instruction? since the last time or its exactly same and you are getting different result?

cc: @Sacha Bron
I didn't touch that part of my code no, and it suddenly changed to give different results. Just wondering why it happened and where the change came from
Hi @Teng, we've had a few issues on some specific fonts on the 24th and made some improvements latter.
Can you share more details on the files that were working for you previously but are broken now?
If you cannot share the file directly, can you share the jobId so we can fix your issue?
there are two runs:
  1. f52da8ad-9207-457e-a205-1bfb352aa0a1
  2. a3ceb437-d181-4d3e-adfa-6960909017bf
Let me know if you can fix it, thanks Sacha!
any updates? πŸ™‚ @Sacha Bron
I'm currently looking at it πŸ™‚
I see you had a Cache Hit on theses two jobs and as mention, we've had a couple issues past week, so maybe your cached versions where bad.
Can you retry to send these 2 files with the "invalidate_cache" option set to true?
and if you're not happy with the result, we can deep dive into what you're seeing vs what you'd like to see.
ok let me take a look, thanks!
Oh I am getting this error now: [SSL: UNEXPECTED_EOF_WHILE_READING] EOF occurred in violation of protocol (_ssl.c:1007)
Any way to bypass that?
let me know, thanks @Sacha Bron πŸ™‚
That's strange. How are you calling LlamaParse? API? Lib? UI?
through lib on docker
mostly through lib using
Plain Text
parser = LlamaParse(
            api_key=llama_api_key,
            result_type="markdown",
            verbose=True,
            disable_ocr=True
            )
This error is coming from your side. Maybe you're behind an enterprise proxy.
I found a few things you can try on StackOverflow: https://stackoverflow.com/questions/33410577/python-requests-exceptions-sslerror-eof-occurred-in-violation-of-protocol
Add a reply
Sign up and join the conversation on Discord