docs(space): weights DOI

docs(space): author ORCID

docs(space): Zenodo DOI

docs(space): 0.4.0 benchmark of record (10.91 vs 11.58 on corrected truth)

docs(space): corrected benchmark headline

feat(space): install facet-nmr from PyPI; drop the vendored package and weights

fix(space): bump Gradio SDK to 6.26.0 to clear the tab-switch effect loop

chore: add .mailmap for the canonical author identity

Sync mask-safe shift-retrieval fallback for HA-missing residues (v0.3.1)

rename: residual-SS tabs → no 'SS' acronym, state-enumerated

version: bump to 0.3.0 to match facet-nmr

feat(viz): leave 'other'/'coil' blank in stacked-bar residual SS

feat(ui): rename residual-SS tabs, add "when to use" captions, stacked-bar figure

feat: add retrieval-free SS populations to residue inspector

README: document structural mode + ensemble export (mirror)

phase 3.5: structural-mode inspector + third residual-SS tab

rci: faithful Berjanskii-Wishart + References section (mirror)

Phase 2 polish — HF app mirror

app: revert light-mode override, follow visitor OS preference

app: force light mode + trim deuteration options to protonated/perdeut

app: add 'Auto-correct shift referencing' checkbox in the UI

README: IDP basin-interpretation note (mirror)

visualization: align unassigned-residue rendering across both plots (mirror)

app: put sequence-SS and residual-SS figures in switchable tabs

visualization: fix empty-plot bug + align residual plot styling (mirror)

referencing check: per-nucleus sanity + app warning block (mirror)

Phase 1 polish: coil glyph, IDP figure, per-residue uncertainty (mirror)

README: update to honest post-index-cleanup benchmark numbers (mirror)

inference: demote near-origin 'other'-basin retrievals regardless of SS (mirror)

retrieval index: strip 33K corrupt (phi=psi=0) entries (mirror)

retrieval: tier ordering + SS-basin consistency check (mirror)

retrieval: tighten spread threshold 50 deg -> 30 deg (mirror)

retrieval: reject spread clusters from DBSCAN chain-merging (mirror)

Unify tier vocabulary to High/Medium/Low/Flexible

v0.2.0: retrieval-augmented inference

Promote deuteration notice to standalone callout

Fix HF Spaces startup: disable Gradio SSR mode

Sync: propagate polymer sequence from readers to plot

Sync: sequence/SS plot gap-preserving layout

Sync: deuteration heuristic SS-aware with significance test

Sync: inspector enhancements (chi1 probs, neighbor strip, etc.)

Sync: deuteration flag for 13C isotope shift correction

Sync: interactive Ramachandran inspector + chi1 in summary

Sync: confidence tier rename (High / Medium / Low / Flexible)

Sync: visualization polish (legend placement, colors, Dynamic rename)

Robust checkpoint resolution with multiple fallback paths

Fix: checkpoint path + disable example caching

Sync: hardening (multi-chain, terminals, shift sanity, caching)

Sync: input validation (non-canonical AAs, proton-only)

Sync: sequence+SS figure