Notebooks

Interactive walkthroughs rendered into this portal

Notebooks portal

Pages below are git symlinks to repository notebooks/*.ipynb. Quarto renders them with execute: eval: false using stored cell outputs and figure sidecar folders under _site/notebooks/ (published with the whole site to Posit Connect).

Live Jupyter

pip install -e ".[notebooks]"
jupyter notebook notebooks/

Browse

Memo

Pipeline walkthrough

Markdown → classify → extract → vision → summarize with figures.

DICIE

DICIE pipeline walkthrough

Paper Fig. 1 stages, page images, review routing, batch run_file.

Baseline

Random Forest text + handwriting

TF-IDF + Random Forest under OCR / handwriting noise.

Corpus

Sample corpus walkthrough

Seed and query the SQLite medical / salvage store.

SQL

Sample corpus SQL integrations

SQL-oriented views over DocumentStore.

Train

Sample corpus train → test

Export corpus rows into train/eval preparation paths.

RVL-CDIP

RVL-CDIP recreation sampling

SQL queries + 60–70 seeded samples per class (0–15).

OpenRouter

RVL-CDIP OpenRouter eval

Vision + OCR multi-model classify / structured extract PoC.

Regeneration helpers

python scripts/build_sample_corpus_notebooks.py
python scripts/build_rf_notebook.py   # Random Forest notebook builder
python scripts/build_rvl_cdip_recreation_notebook.py
python scripts/build_rvl_cdip_openrouter_eval_notebook.py

--- title: "Notebooks" subtitle: "Interactive walkthroughs rendered into this portal" page-layout: full --- ::: {.hero-kicker} Notebooks portal ::: ::: {.hero-lead} Pages below are git symlinks to repository `notebooks/*.ipynb`. Quarto renders them with **`execute: eval: false`** using **stored cell outputs** and figure sidecar folders under `_site/notebooks/` (published with the whole site to Posit Connect). ::: ::: {.callout-tip} ## Live Jupyter ```bash pip install -e ".[notebooks]" jupyter notebook notebooks/ ``` ::: ## Browse ::::: {.feature-grid} ::: {.feature-card} Memo ### [Pipeline walkthrough](pipeline_walkthrough.ipynb) Markdown → classify → extract → vision → summarize with figures. ::: ::: {.feature-card} DICIE ### [DICIE pipeline walkthrough](docie_pipeline_walkthrough.ipynb) Paper Fig. 1 stages, page images, review routing, batch `run_file`. ::: ::: {.feature-card} Baseline ### [Random Forest text + handwriting](random_forest_text_handwriting_classification.ipynb) TF-IDF + Random Forest under OCR / handwriting noise. ::: ::: {.feature-card} Corpus ### [Sample corpus walkthrough](sample_document_corpus_walkthrough.ipynb) Seed and query the SQLite medical / salvage store. ::: ::: {.feature-card} SQL ### [Sample corpus SQL integrations](sample_corpus_sql_integrations.ipynb) SQL-oriented views over `DocumentStore`. ::: ::: {.feature-card} Train ### [Sample corpus train → test](sample_corpus_train_test_pipeline.ipynb) Export corpus rows into train/eval preparation paths. ::: ::: {.feature-card} RVL-CDIP ### [RVL-CDIP recreation sampling](rvl_cdip_recreation_sampling.ipynb) SQL queries + 60–70 seeded samples per class (0–15). ::: ::: {.feature-card} OpenRouter ### [RVL-CDIP OpenRouter eval](rvl_cdip_openrouter_eval.ipynb) Vision + OCR multi-model classify / structured extract PoC. ::: ::::: ## Regeneration helpers ```bash python scripts/build_sample_corpus_notebooks.py python scripts/build_rf_notebook.py # Random Forest notebook builder python scripts/build_rvl_cdip_recreation_notebook.py python scripts/build_rvl_cdip_openrouter_eval_notebook.py ``` ::: {.see-also} ### See also [Pipeline hub](../pipelines/index.qmd) · [Sample Document Corpus](../sample_document_corpus.md) · [RVL-CDIP SQL Index](../rvl_cdip_sql.md) · [DICIE Pipeline](../docie_pipeline.qmd) · [How-to: Quarto site](../how-to/launch-quarto-site.qmd) :::