chore: repack the upstream mirrors without the download cache

docs: state the working layout and the pseudo-label taxonomy

docs: state the upstream licences and make the usage block runnable

docs: scope the snapshot_download caveat to the affected version

docs: drop em dashes from the markdown

docs: finish the MIT relicense and state the input transforms

docs: rewrite the model card in the style of the original

chore: relicense to MIT

docs: fix the verification step and ship the homophily tail metrics

docs: correct the reproduction instructions

docs: correct the SPIDER probe path

docs: finish the GrapHist++ rename

docs: rename the model to GrapHist++ in prose

Fix stale paths and resync the shipped modeling code

Correct the repository tree description

Bring the studies contents in line with what actually ships

Update README.md

Link the code repository; point script paths at its layout

Link the code repository

Link the authors to their sites

Link the authors to their sites

Cite every source cohort, note LTS4 supervision

Cite every source cohort, note LTS4 supervision and the separate code repository

Move analysis and pipeline code to the code repository; keep the model, its loader, the data and the docs

Remove a non-runnable duplicate survival script (PATH/TO placeholders)

Drop LFS rules for the old data/ layout

Make every command runnable from the repo root; fix the folder-limit and gate wording

Restore the full parameter counts, correct the control value, clarify what inference needs

Remove hardcoded paths from src_at_ef0b17d.tar

Remove hardcoded paths from src_at_9d74e93.tar

Replace hardcoded paths with placeholders: pooled_edge_homophily.json

Replace hardcoded paths with placeholders: compute_pooled_edge_homophily.py

Replace hardcoded paths with placeholders: run_cellvit_pipeline.sh

Replace hardcoded paths with placeholders: fat_tail_followup.py

Replace hardcoded paths with placeholders: run_fullregen.sh

Replace hardcoded paths with placeholders: nucls_super_source_args.json

Replace hardcoded paths with placeholders: run_optuna_repro.sh

Replace hardcoded paths with placeholders: commands.txt

Replace hardcoded paths with placeholders: rebase_graph_paths.py

Ship the parameterized survival scripts (no hardcoded paths, CLI args)

Add authorship, remove em dashes

Correct library versions, add authorship, remove em dashes

Remove internal reproduction notes; their content is now in DATA.md

Rewrite DATA.md for a general audience: no branches, per-dataset reference, absorbs the study numbers

Move data docs to DATA.md and rewrite every path for the new structure

New model card: GrapHist v2 only, shorter, matches the new layout

Restructure: move batch 30/30

Restructure: move batch 29/30

Restructure: move batch 28/30

Restructure: move batch 27/30