mrkn1

5 ms·

1. 31.▲
   Show HN: Local CPU OCR for images, PDFs, webpages(github.com)

   3 points by mrkn1 4mo ago|0 comments
2. 32.▲
   by mrkn1 4mo ago

   https://github.com/kouhxp/fftext
3. 33.▲
   by mrkn1 5mo ago

   For 100% local CPU fact checking, I made this: https://news.ycombinator.com/item?id=48301003
4. 34.▲
   by mrkn1 5mo ago

   A lot of my queries are summarize/explain/fact check, and these are covered 100% on my CPU locally [0], reducing frontier model reliance [0] https://news.ycombinator.com/item?id=48301003
5. 35.▲
   Show HN: Local CPU model for fact-checking, summarizing, explaining text(github.com)

   6 points by mrkn1 5mo ago|0 comments
6. 36.▲
   by mrkn1 5mo ago

   just released new version that implements your ideas
7. 37.▲
   by mrkn1 5mo ago

   new version released is up to 3x faster on CPU. Let me know!
8. 38.▲
   Show HN: CPU transcription for YouTube/TikTok/X, now 3x faster and diarization(github.com)

   3 points by mrkn1 5mo ago|0 comments
9. 39.▲
   Advancing Mathematics Research with AI-Driven Formal Proof Search(arxiv.org)

   1 points by mrkn1 5mo ago|0 comments
10. 40.▲
    by mrkn1 5mo ago

    thank you being thorough clipboard: rn input is treated like any other source, so text gets written to ./textsnaps/clipboard_ocr.txt, and stdout just prints that path. Nothing goes back to the clipboard in this version (stay tuned
11. 41.▲
    by mrkn1 5mo ago

    Great question. I'm not familiar with docling-serv but pretty different beasts from what I gathered. Docling is a heavier pipeline (actually uses GPU).textsnap is the opposite: single-file CLI, small VLM running on plain CPU cores, one
12. 42.▲
    by mrkn1 5mo ago

    thanks! yapsnap is audio to text, and textsnap is image to text. Both have been daily use cases for me for a while. And yes, the feedback on yapsnap encouraged me to also release textsnap on github
13. 43.▲
    Show HN: CPU-only fast OCR for screenshots, images, PDFs, webpages(github.com)

    4 points by mrkn1 5mo ago|9 comments
14. 44.▲
    by mrkn1 5mo ago

    Reading this means a lot, thank you! Even faster version in the coming days, stay tuned and PRs welcome!
15. 45.▲
    by mrkn1 5mo ago

    Added Diarization / Speaker Separation that is fast and CPU only. Thank you all for the great feedback and support. PRs welcome! yapsnap "https://www.youtube.com/watch?v=NzKJ-xO-VhE" --diarize SPEAKER_00
16. 46.▲
    by mrkn1 5mo ago

    done! new version separates speakers on CPU fast
17. 47.▲
    by mrkn1 5mo ago

    done! just pushed a new version with CPU diarization
18. 48.▲
    by mrkn1 5mo ago

    Kroko's website says benchmarks aren't formalized yet. FWIW, this url says 5% WER for English [0]. though it doesn't specify the dataset, so not directly comparable to Parakeet's 6.32 on the Open ASR Leaderboard Best way
19. 49.▲
    by mrkn1 5mo ago

    in the roadmap!
20. 50.▲
    by mrkn1 5mo ago

    will work on it, that would be neat. I love pyannote but not happening on CPU at reasonable speeds lol
21. 51.▲
    by mrkn1 5mo ago

    thank you! good use case, what hetzner box specs have you chosen?
22. 52.▲
    by mrkn1 5mo ago

    small, ONNX-optimized models designed specifically for low-latency CPU streaming, so it avoids overhead of large transformer arch and GPU memory transfers
23. 53.▲
    by mrkn1 5mo ago

    thanks! making a note of the feature request
24. 54.▲
    by mrkn1 5mo ago

    Good point! I haven't found a faster way to consume info than reading. But depends on the type of learner you are (visual, auditory, hands-on/interactive, etc)
25. 55.▲
    by mrkn1 5mo ago

    thank you!
26. 56.▲
    by mrkn1 5mo ago

    Just download the model for your preferred language, all hosted on the Kroko-ASR collection here: https://huggingface.co/Banafo/Kroko-ASR/tree/main Right now you have Dutch, French, Portuguese, Spanish, Germa
27. 57.▲
    by mrkn1 5mo ago

    Yes fair point, asr cached and exposed. I meant to draw the line more on fetchable or not.
28. 58.▲
    by mrkn1 5mo ago

    thanks for running it Niraj. I see something similar on my machine, which still surprises me every time lol
29. 59.▲
    by mrkn1 5mo ago

    I appreciate the perspective! higher ceiling than I'd put on it, but if it gets there awesome. PRs welcome!
30. 60.▲
    by mrkn1 5mo ago

    Youtube has transcripts on most videos, not all. The others don't expose them. If you mean the "transcript APIs" for TikTok/IG/X, they are all transcribing audio like yapsnap does. If you have a way to pull native o