Spaces:
Running on Zero
Running on Zero
Commit History
-
Bump actions/checkout to v7 for the Node 24 runtime
Sanidhya Raviv4 targets Node 20, which GitHub deprecated. Runners already force those actions onto Node 24 and warn on every run, so this only removes the warning and pins us ahead of the removal.c4a5585 -
Fix the sync workflow refspec and fail clearly without a token
Sanidhya Ravi`git push <url> main` needs a local ref named main, but actions/checkout only creates the branch that triggered the run. Dispatching manually from any other branch failed with "src refspec main does not match any" -- which is exactly what you would do to test the deploy before merging. HEAD:main works for both the push trigger and manual dispatch. Also fails fast with a pointed message when HF_TOKEN is missing, rather than surfacing it as an opaque git authentication error.d035625 -
Rename to doc-agent and rewrite the README
Sanidhya RaviThe project moved to sanidhya1910/doc-agent on both GitHub and the Hub, so point the sync workflow at the new Space and rebrand the app. The Space is under sanidhya1910, not blackburn1910 -- the old workflow would have pushed to the wrong account. Also removes .github/workflows/main.yml. It still carried the literal placeholder HF_USERNAME/SPACE_NAME, so leaving it alongside sync-to-hub.yml would have failed on every push to main. The README is rewritten around what the agent does for the reader: what it gives back per document type, that it never leaves you empty-handed, the tool table, how the GPU budget shaped the design, and how to call it over MCP or as a library.8895204 -
Turn the TrOCR demo into an agentic document pipeline
Sanidhya RaviThe Space did not run. `from_pretrained(".")` loads weights from the repo root, but no model files were ever committed -- git history shows only .gitattributes, README.md, app.py and requirements.txt, and the live Space holds the same four. It crashed on boot. TrOCR now loads from the Hub. Beyond that fix, the demo becomes an agent: drop in any document, optionally say what you want, and it works out what the document is, reads it with the right model, and writes the structured output that suits it -- a spreadsheet for tabular data, JSON for a form, a summary PDF for prose. The same tools are published over MCP so external agents can call the Space directly. Pipeline: ingest -> read -> classify -> tables -> route or plan -> export. * ingest.py takes images, multi-frame TIFF, HEIC, PDF, DOCX, PPTX, XLSX, text and ZIP. pypdfium2 bundles its own PDFium, so no poppler/packages.txt. * Pages arriving with a text layer are used as-is and never reach the GPU. * router.py settles unambiguo…5eb40dc
-
Create main.yml
Sanidhya Ravi9bed480