Prasad Chalasani

Since we already have docling, and it’s unclear if this is better, we’ll wait to see some comparisons. See this discussion https://www.reddit.com/r/LocalLLaMA/s/0aARsH1h5v

[`pymupdf4llm`](https://pymupdf.readthedocs.io/en/latest/pymupdf4llm/api.html#pymupdf4llm-api) has a way to iterate through the converted markdown page-by-page. Is there a similar way here?

I find `pymupdf4llm` far more convenient - it gives you markdown-formatted text _for each pdf page_, unlike docling which as far as I know gives you the `export_to_markdown()` method that...

Looking at some of the docling examples, I tried to get page-by-page markdown this way in langroid but it is super-slow: https://github.com/langroid/langroid/blob/main/langroid/parsing/document_parser.py#L563

@rickzx yes the git clone works. Can we use `mlc_llm serve` and directly point it to the local cloned model, rather than the HF... argument?

> @pchalasani you may be missing the dependencies in [step 1 here](https://llm.mlc.ai/docs/deploy/ios.html#build-ios-app-from-source) Thanks but I'm not using it on iOS. Please let me know if these deps are needed for...

Thanks @abab-dev I tried it , asked a simple question and it proceeded to get a bunch of HN threads and then quit: ``` ✔ Session complete: hn_sessions/session_20250720_153508 (0 successful)...

I see some errors ``` ✔ Session complete: hn_sessions/session_20250720_155629 (3 successful) Ingesting post_44524707.json... 2025-07-20 15:56:36 - ERROR - Error ingesting HN data: BaseModel.validate() takes 2 positional arguments but 3 were...

> @pchalasani, it was working on my end earlier. Let me take another look and check again. Also are you going to merge pydantic v2 pr anytime soon. so that...

> > @pchalasani, it was working on my end earlier. Let me take another look and check again. Also are you going to merge pydantic v2 pr anytime soon. so...