Add dedicated "monkeyocr" library tag
#11
by
nielsr
HF Staff
- opened
It seems this model isn't compatible with Transformers. So instead let's add "monkeyocr" as a dedicated library to the hub.
We can then track downloads as explained here: https://huggingface.co/docs/hub/models-download-stats
nielsr
changed pull request status to
closed
Hello, I'm also quite confused — it seems that our download count is not being tracked properly. Could you please advise me on how to resolve this issue?
Yes, so the way we track downloads is explained here: https://huggingface.co/docs/hub/models-download-stats.
Ideally, each model checkpoint lives in its own model repository.
- So if you would place the files of https://huggingface.co/echo840/MonkeyOCR/tree/main/Recognition at the root of the repo, downloads will be tracked by default. This model repository could then be called "echo840/MonkeyOCR-recognition". We can then add
library_name: transformersto the model card, since the model is compatible with Transformers. This ensures a dedicated "Use this model" button appears at the top right with a sample code snippet. - Similarly you could have model repos called "echo840/MonkeyOCR-relation" and "echo840/MonkeyOCR-structure" (assuming each is a standalone model). This way you can track downloads for each. Those seem custom, so for that we can add
library_name: monkeyocrand then open a PR to the huggingface.js library as this one: https://github.com/huggingface/huggingface.js/pull/1543. This allows you to customize which file extension you want to track downloads for, have a custom code snippet, etc
Thank you very much, I understand now.