Consolidate duplication accrued across the ML tagging + settings backend,
behavior-preserving (over-DRY guard applied — the three auto-apply sweep
BODIES stay separate; only their shared inner helpers are extracted).
- _sigmoid / _conflict_scores / _insert_presentation_review (heads.py): the
score→prob transform (6 inlined sites), the presentation conflict signal
(2 sites), and the ring-loud PresentationReview insert (2 sites, single-
sourced so the mode column can't drift on the shared composite PK).
- _applied_or_rejected (training_data.py): the per-tag "applied ∪ rejected"
skip-set, byte-identical at 3 sweep sites (heads.py x2, tasks/ml.py ccip).
- ccip sweep divergence fixes: import ccip._FIGURE_KINDS + training_data._l2norm
instead of local copies that silently drift when the canonical changes.
- MLSettings.load / .load_sync classmethods (mirror ImportSettings); route all
8 scalar_one singleton reads through them (the session.get None-path stays).
- GET serializers for MLSettings + ImportSettings are now table-driven off the
same _EDITABLE tuples PATCH writes, so a new field can't be silently absent
from GET (the split that historically dropped fields).
- AUTO_APPLY_THRESHOLD_MIN/MAX constant single-sources the [0.5,0.999] operating
range across the service clamp + the 5 API validators.
- test_ml_dry_helpers.py pins _applied_or_rejected + _sigmoid.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NsmJSQxnNxGgtM5Yz4GAqi
The Explore walk got stuck in dense signatures — neighbours all too similar, so
forward-arrow couldn't escape and Random was the only exit. Root cause: MMR only
diversifies WITHIN the nearest ~400 pool; in a dense cluster that whole pool is
near-identical, so there's no escape route in it.
- gallery_service.similar(reach=0.0, exclude_ids=None): reach>0 widens the pool
(cap 400→1000) and _reach_sample strides across an outward-growing distance span
so the set handed to MMR spans near→mid-far (guaranteed escape routes), not just
the tight cluster. exclude_ids drops already-walked images. Gallery 'more like
this' (reach=0) is unchanged.
- api/gallery similar: parse reach + exclude_ids.
- explore store: default reach 0.4 (auto-diversifies without touching the dial),
pass the breadcrumb as exclude_ids, setReach action.
- ExploreView: a Near↔Far reach slider in the trail.
- tests: _reach_sample math (deeper ranks with higher reach, near kept); similar
exclude_ids drops walked + reach path runs clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Extends WIP title-tagging to lower-precision cues (sketch/doodle/scribble) safely.
- wip_title.py: soft matcher (word-anchored; sketchbook/kadoodle don't trip it);
WIP_TITLE_SOFT_SOURCE + soft SQL prefilter; apply_wip_image_tags takes a source arg.
- training_data._AUTO_SOURCES += 'wip_title_soft' → the soft tier is PROVISIONAL and
never trains the wip head (a finished "sketch" can't pollute it). Only the hard
tier (wip_title) + manual train.
- ImportSettings.wip_soft_title_tagging_enabled (OFF by default, opt-in). Migration 0087.
- importer: hard tier wins, soft is the fallback (source wip_title_soft).
- backfill: refactored into a shared _backfill_wip_tier; hard always, soft when enabled.
- heads.soft_wip_conflict_audit + daily beat: score soft-tagged images against content
heads, flag ring-loud ones (PresentationReview mode=process) for the review strip —
the operator's "measure if they got falsely tagged" safety.
- api settings toggle; ImportFiltersForm soft toggle.
- tests: soft matcher pos/neg; soft source not a training positive; audit flags
ring-loud + spares quiet.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Backend for the system-tag behavior refactor (milestone #157). editor screenshot
moves from chrome (hidden) to the PROCESS group (shown, like wip); wip+editor gain
provisional auto-apply so they stop needing endless manual identification —
without a runaway loop.
- tag.py: split PRESENTATION_SYSTEM_TAGS → CHROME_SYSTEM_TAGS (banner) +
PROCESS_SYSTEM_TAGS (wip, editor screenshot).
- heads.py: generalize presentation_auto_apply_sweep → system_tag_auto_apply_sweep
(mode chrome|process). Same Guard 1 (skip human/confirmed) + Guard 2 (ring-loud
conflict → PresentationReview). process mode uses source 'process_auto' and does
NOT hide (hide is a gallery-query effect of group membership).
- training_data._AUTO_SOURCES += 'process_auto' → the head never trains on its own
auto-applied output; only wip_title/manual train it (the runaway break).
- ml_settings: process_auto_apply_enabled (OFF, opt-in) + threshold + conflict
threshold. presentation_review.mode ('chrome'|'process'). Migration 0086.
- gallery_service: default-hide reads CHROME only (editor now shows); Explore
neighbors exclude the whole PROCESS group.
- tasks/ml + celery beat: scheduled_process_auto_apply (daily, opt-in); prune
covers both modes.
- api: ml_admin process_* CRUD+validation; hidden-review returns mode.
- tests: rename chrome sweep calls; new test_process_auto_apply (apply, guards,
mode flag, no-self-train); gallery test asserts editor now visible.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Auto-apply the `wip` system tag to posts whose TITLE explicitly declares
work-in-progress ("WIP" / "work in progress") — a deterministic, high-precision
complement to the image-based ML `wip` head. WIP images are excluded from the
Explore/gallery browse, so honouring the artist's own label keeps unfinished
pieces out of the main browse.
- services/wip_title.py: precision-first token-anchored matcher (swipe/wiped
never trip it) + sync apply helpers (source='wip_title', ON CONFLICT DO
NOTHING, chunked under the psycopg param ceiling).
- importer: live hook on FRESH import only (never on deep-scan/supersede), so a
manually-removed WIP tag is never re-applied by a routine re-scan.
- maintenance.backfill_wip_title_tags: operator-triggered back-catalogue sweep
(coarse SQL prefilter + regex confirm, keyset-paginated). Deliberately NOT a
beat — a periodic re-run would silently undo manual removals.
- ImportSettings.wip_title_tagging_enabled (default ON, migration 0085) gating
the live hook; GET/PATCH + POST /settings/wip-title/scan.
- Settings UI: toggle + "Scan existing posts" button.
- Tests: pure matcher unit tests + integration (apply idempotency, backfill
precision).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The gate at a fixed 0.80 couldn't catch the real pain: Interpreter (fresh ==
cached, verified by probe) confidently mis-detects short ASCII English like
"... WIP Part 1" as German at 0.86 — above the floor — so it was accepted and a
re-translate reproduced it. Confidence alone can't separate the 0.86 collision
(genuine German lands there too), and single-word mis-flags sit at a confident
1.0 no floor catches.
Two operator-approved levers:
- Acceptance floor is now a live Settings value (ImportSettings.
translation_min_confidence, default 0.90; surfaced in the Translation card), so
it's tunable without a redeploy. _accept takes the threshold as a parameter.
- Per-post sticky override (Post.translation_override: auto/force/original).
'force' stores a translation even below the floor (rescue a skipped
legit-foreign title); 'original' keeps the original and clears any stored
translation (kill a confident mis-flag no floor catches). The sweep honors it
on every run and _reset_translations skips 'original', so the choice survives a
Re-translate-all. POST /api/posts/<id>/translation-override applies it
immediately (translate now when the service is up, else queue for the sweep).
UI: PostTranslationControl on the posts-feed card.
Migration 0084 (both columns + a CHECK on the override). The feed + provenance
serializers expose translation_override.
With a stricter floor the rollback finally works: raise it -> Re-translate all ->
the 0.86 mis-flags are rejected and restored to the original; force /
keep-original handle the residual either way.
Tests: gate thresholds against the param (0.86 rejected at 0.90, explicit-floor
cases); sweep force/original + re-translate-skips-original; override endpoint
(validation, original clears, force queues when disabled, feed exposes it);
settings min_confidence default/save/validate.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CgZP9v2otxVJymiYsnVuMy
Every tag suggestion is a canonical DB tag now (tagging-v2 #114: heads + CCIP
score EXISTING concept tags). The pre-heads apparatus for model-predicted tags
that didn't exist in the DB — creates_new_tag / raw_name / via_alias, the
/suggestions/alias endpoint + add_alias_and_accept, AliasPickerDialog, and the
store's aliasAccept/removeAlias — was dead and is removed.
The type-to-add dropdown was TWO row sources (server autocomplete + the image's
ML suggestions) merged with a dedup that dropped the %-bearing suggestion row
when the debounced server hit landed — the operator's "confidence % flickers
then vanishes". Now it's ONE list of DB-tag matches, each annotated with the
model's confidence (join by canonical_tag_id) when the tag was scored for this
image. No dedup, no flicker; picking a suggested tag still records acceptance
via TagPanel.findPending.
Single per-image fetch: score_image now reports above_threshold per row
(computed vs the head's own suggest cut, separate from the inclusion floor), so
the rail makes ONE min=0 request and derives the panel (above_threshold) and the
dropdown (all, text-filtered) client-side — the two /suggestions calls collapse
to one. Manual "Create 'X' as <kind>" (novel typed names) is unchanged; the
alias table + tag-side alias admin + auto-apply alias matching are untouched.
Tests: gate/serializer assertions updated (above_threshold; dropped dead-field
+ alias-endpoint checks); frontend spec seeds via the single load and covers the
byCategory/aboveByCategory split.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CgZP9v2otxVJymiYsnVuMy
Throughput: translate_posts now runs every 8h (was daily) as the
steady-state cadence for newly-imported posts, and the Settings
"Translate now" button runs it in drain mode (run-until-done, no reset)
so one press clears the whole untranslated backlog instead of a single
300-post chunk. The interrupt/backoff re-enqueue now preserves the drain
flag so a bulk drain resumes cleanly after an Interpreter restart.
Misdetection groundwork: surface the detector's confidence from the
Interpreter client (it was in the detectedLanguage payload but discarded)
and add a read-only "Test translation" box — POST /settings/translation/
probe + TranslationCard UI — that shows detected language + confidence +
engine + result for pasted text, without saving. Lets the operator see
why a short/abbreviation-heavy English title gets mis-detected so the
detection guard (min-length + confidence floor) can be tuned from real
numbers. The guard itself follows once the mis-detected cases are probed.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CgZP9v2otxVJymiYsnVuMy
translation/status now reports `active` (a translate/retranslate sweep is
running, from the TaskRun table) and `last_run` (the most recent finished run's
task + status). The Settings card polls live while a sweep runs, showing a
spinner + "Translating… N remaining" that ticks down, and flags a last run that
ended in error/timeout. No migration.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Explore's neighbour grid (/api/gallery/similar → gallery_service.similar) now
takes an Explore-only exclude_wip flag that drops `wip` system-tagged images
from the candidates, alongside the banner/editor presentation tags. The
gallery's own "similar" button is unchanged (keeps wip, #1274) — only the
Explore store passes exclude_wip=1. The anchor itself may still be a WIP; only
neighbours are filtered.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
retranslate_posts resets the 5 translation columns to NULL for a scoped set of posts (all, or WHERE artist_id IN ids) then reuses the untranslated sweep to re-run them, chasing the tail until drained (run-until-done). Interpreter cache keys on engine_version so a changed model re-translates, an unchanged one is cache-fast. Reset only happens when the service is configured+healthy so translations are never wiped when they can't be rebuilt. New POST /settings/translation/retranslate (artist_id | all=true). UI: per-artist 'Re-translate posts' on the Artist Management tab + 'Re-translate all' in the Settings Translation card, both with confirm dialogs. No migration (reuses m143 columns).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
New POST /api/settings/translation/test pings /v1/health for a GIVEN base URL (not
the saved one), so the operator can verify a URL before enabling it. TranslationCard
gains a Test-connection button that reports reachable/unreachable inline and
updates the status dot. Endpoint test.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
tasks/translation.py — translate_posts: picks untranslated posts (title OR
description non-empty), per-post [title, description] batch via the Interpreter
client, stores translations + detected lang + engine_version; passthrough /
already-target posts are marked handled with no stored translation. 503 or a
connection error interrupts (retry next cycle), 400 stops (fix config), per-post
commit keeps progress; wall-clock bounded. Wired into celery (maintenance_long
lane) + a daily beat. No-op unless enabled + base URL set + healthy. GET
/settings/translation/status + POST .../run for the Settings card. Task tests
(stubbed client, monkeypatched session).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
Post gains post_title_translated / description_translated / translated_source_lang
/ translation_engine_version / translated_at — filled by the translate sweep so
viewing is instant. ImportSettings gains translation_enabled (OFF by default),
interpreter_base_url (EMPTY — no default host; the operator points it at their own
Interpreter proxy behind a reverse proxy) and translation_target_lang (en),
exposed + validated via /settings/import. Migration 0083. Settings defaults +
patch + validation test.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
GET /api/gallery/hidden-review lists unresolved presentation auto-hide flags
(image + presentation tag + conflict tag/score), most-concerning first. POST
.../keep resolves the flag (the tag stays). POST .../unhide removes the
presentation tag (image returns to the gallery), records a TagSuggestionRejection
so the head learns it misfired, and resolves the flag. Tests for list/keep/unhide.
Frontend review strip (shown when Show-hidden is on) next.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
MLSettings gains presentation_auto_apply_enabled / _threshold (default 0.90) +
presentation_conflict_threshold (default 0.50): banner/editor auto-hide with a
FLAT threshold (decoupled from content-head graduation), plus the "also looks
like content" conflict cut. New presentation_review table (image, presentation
tag, conflict tag + score, created/resolved_at) records auto-hides flagged for
review. Migration 0082 (columns + table), ml_admin API (editable + get_settings
+ _validate bounds), settings roundtrip/bounds test. The sweep that reads these
knobs + the Settings UI land in step 4.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
The /similar route splats **filters into similar(), which doesn't take the new
include_hidden kwarg → TypeError → 500 (test_gallery_similar). Drop it like
post_id; similar() has its own presentation exclusion (#1274), so the
gallery-browse flag doesn't apply there.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
The default gallery + facets now implicitly exclude images carrying a
presentation system tag (banner / editor screenshot), reusing the tag-scope
EXISTS machinery. Suppressed when the operator explicitly filters FOR a
presentation tag OR passes include_hidden (the Hidden view — step 2). `wip` is
NOT hidden (real, in-progress art). include_hidden threaded through
scroll/timeline/jump_cursor/facets + the gallery API _parse_filters. Test covers
default-hide, include_hidden, explicit-filter-shows, and wip-stays-visible.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
PreviewDialog.vue was orphaned — nothing mounted it — so the dry-run preview
was unreachable. Rather than wire it up, remove the whole chain: its unique
value (a capped 3-page count of what a backfill would grab) is low, and its
adjacent needs are already covered — auth validation by verify_source_credential,
and actually fetching recent posts by "Check now". Operator decision 2026-07-06.
Removed:
- frontend: PreviewDialog.vue + sources store previewSource()
- backend: POST /api/sources/<id>/preview route, download_backends.preview_source,
IngestCore.preview() + its now-unused NativeIngestError import
- tests: the 3 ingester preview tests
Nothing else referenced the chain (verified). Shared campaign-resolution and
ledger helpers stay — they're used by run()/verify_source_credential.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
Applied tags aren't scored live, so compute the grounding on demand: run the
tag's head over the image's max-over-bag (whole-image + concept crops), argmax
→ the region that best explains the tag on this image, mirroring what
score_image records for live suggestions.
- heads.py: extract _image_bag (now shared by score_image) + ground_applied_tag.
Returns (grounding, has_head): has_head False = no head to localize with →
no overlay; grounding None = the whole-image vector won → whole-image frame.
- tags.py: GET /api/images/<id>/tags/<id>/grounding → {grounding, has_head}.
- TagChip/TagPanel: applied chips inject fcSuggestionHover and fetch grounding
on hover (cached per image+tag, race-guarded), reusing Step 3's overlay in
both the modal and Explore. No new frontend overlay code.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
score_image now keeps the ARGMAX beside the max-over-bag: which bag row won each
head. The region query also selects bbox/kind/detector_version, a parallel
bag_meta maps each row → its region (None for the whole-image vector), and every
hit gains grounding {bbox,kind,detector} (null when the global vector won). Threaded
through SuggestionService (new Suggestion.grounding field) → /api/.../suggestions
payload. This is the data the #1206 hover-overlay draws. CCIP-only hits ground null
for now (figure grounding = step 2). Tests: winning crop grounds the tag with its
bbox+kind; whole-image win → grounding None.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
Exposes the detector config (per-proposer enable + weights + confidence, caps,
dedupe IoU) in Settings → Tagging, backed by MLSettings via /api/ml/settings.
ml_admin adds the detector fields to _EDITABLE + GET payload + validation (conf
0..1, caps >=1, IoU 0..1). New CropProposersCard.vue (mirrors HeadsCard) with
working defaults pre-filled, per-field live-save (no restart — the agent picks
changes up on its next lease), weights-format help, switch-revert on error.
Closes milestone #134: all three proposers are on out-of-the-box and tunable in
the UI. Test: detector defaults GET + patch round-trip + range validation.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
The GPU lease now carries the crop-proposer config from MLSettings in a per-job
'detectors' block (same pattern as embed_model_name). The agent's worker builds
its Proposers from the announced config via _effective_cfg (lease block overlaid
on env) + _proposers_for (rebuilds only when a config signature changes) — so an
operator's UI edit takes effect on the next lease with NO restart, and env is now
just the bootstrap fallback until the server announces. enabled-off maps to empty
weights (proposer skipped); dedupe_iou + max_regions also come from the effective
cfg. Test: lease announces the detectors block with the seeded default weights.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
Final piece of the artist decoupling. (1) Identity-by-source: quick_add_source
resolves the artist by an existing (platform, url) Source first, so a re-add
reuses the artist even after it was renamed (its frozen slug no longer matches
the name) — a slug-based lookup would have duplicated it. (2) Pixiv naming: a new
pixiv source resolves the real display name via the app API (PixivClient
.resolve_display_name → /v1/user/detail) using the stored token, so the artist is
'Kurotsuchi Machi' not '12345678' — and its name-derived slug matches what a
native download produces, unifying them. Falls back to the numeric id when no
token/crypto. ExtensionService gains the crypto seam; the endpoint passes it.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
Operator ask: a surface to merge new sources into existing artists (consolidate
the singleton artist a fresh add spins up). Enabled by the #130 slug decoupling —
the storage path is immutable, so re-attribution moves NO files. SourceService
.reassign moves the source, re-points its posts (Post.source_id==S) and the
images it contributed (ImageProvenance via S, scoped to the old artist so shared
images aren't stolen), and deletes the old artist if it's left fully empty (else
clears its subscription flag). POST /api/sources/<id>/reassign. Frontend: a
'Move…' action per source on the artist Management tab → artist-autocomplete
picker → confirm → routes to the target (whose slug is stable).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
First step of decoupling artist identity/storage/display. migration 0077 drops
uq_artist_name so the display name is free text (two genuinely different creators
can share a name); the slug stays the immutable, unique storage/identity key (the
on-disk path component — untouched, so nothing moves). ArtistService.rename +
PATCH /api/artists/<id> change the name ONLY. Frontend: inline pencil-edit on the
artist header (mirrors TagCard), slug/route unaffected so no navigation. Fixes the
operator's 'no surface to rename an artist' + the name-collision fragility.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
pixiv joins NATIVE_INGESTER_PLATFORMS: download/verify/preview and the
recover/recapture UI actions now route through PixivIngester. Campaign id is
parsed straight from the source URL (numeric user id — no network resolver),
with a platform-aware resolution-failure message. auth_token now rides the
uniform adapter construction (token platforms use it, cookie platforms
accept-and-ignore), and the preview endpoint fetches/threads it. The legacy
gallery-dl pixiv path is fully removed (PLATFORM_DEFAULTS entry + the
refresh-token config branches in download/verify) per no-legacy policy;
gallery-dl keeps hentaifoundry/discord/deviantart until they migrate/retire.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
UI: shield marker + tooltip on TagChip and TagCard; system tags hide
rename/merge/delete affordances (chip kebab entirely — set-fandom never
applies to their general kind; remove stays, un-tagging is normal use).
Aliases stay available: mapping model outputs ONTO a system tag is
useful. Directory cards carry is_system.
Every destructive path that could take out a system row is now guarded,
found by sweeping run 1891s off-by-three failures — each one was a
surface that would have eaten the seeded tags:
- prune-unused: predicate exempts is_system (they ship with zero
applications and matched every unused condition)
- reset-content: predicate exempts is_system AND keeps their
applications — hygiene flags describe the file, not content tagging
- admin tag DELETE: refused with system_tag error
- normalize_existing_tags: scan excludes is_system — canonicalization
would recase wip -> Wip behind TagService.rename's guard, breaking
the name-keyed presentation lookup
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
Training hygiene step 1 (milestone #128). Migration 0075 adds
tag.is_system and seeds wip / banner / editor screenshot (kind=general),
ADOPTING an existing same-(name,kind) tag case-insensitively instead of
duplicating. These rows drive the upcoming training exclusions, so they
are protected: rename and merge-away refuse system tags (merge-INTO
stays allowed — folding an operator's old hygiene tag into the system
row is the intended move; merge is the only tag-delete path, so that
guard covers deletion). is_system rides every tag serialization.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
Operator-flagged: the recovered defective files live DEEP in their artists'
back-catalogues — the normal download cadence (by design, via the seen-gates)
will never re-walk them, so recovery's source re-check alone can't bring them
back. The durable per-post handle is the ExternalLink row, which survives the
image delete:
- services/external_links.refetch_links_for_post: reset settled links to
pending (fresh attempt budget, in-flight left alone) + dispatch their
fetches; sha-dedupe at import discards payload files that still exist, so
only the missing file lands.
- recover_defective_image now captures the image's post ids BEFORE the delete
cascades provenance away and resets those posts' links — future recoveries
are surgical automatically (response gains links_reset; source re-check
stays for gallery-dl-native files within walk reach).
- POST /api/admin/posts/refetch-external {external_post_id, source_id?} — the
manual tool for the three files recovered before this fix existed.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
The ml-worker's ONLY processing role is now the CPU whole-image embed fallback
(tag_and_embed renamed embed_image — Camie tagging was retired #1189 and the
name kept implying otherwise; videos were already handled agent-style: frame
sampling + mean-pool). Detection/cropping/CCIP stay GPU-agent-only, and their
completion is judged per-pipeline: ccip by gpu_job rows, siglip by concept
regions at the current model version — never by image_record.siglip_embedding.
A CPU embed therefore can NEVER close crop work for the agent (regression test
pins this; only the whole-image 'embed' job, the same artifact, is satisfied).
Making removal actually safe (operator will drop the container):
- GPU-queue coordination (enqueue_gpu_backfill, recover_orphaned_gpu_jobs,
reprocess_gpu_jobs) moved verbatim to tasks/gpu_queue.py on the maintenance
quick lane — it lived on the 'ml' queue only by module colocation, which made
the ml-worker a hard dependency of the whole agent pipeline.
- New ml_settings.cpu_embed_enabled (migration 0074, default ON so agent-less
installs keep working): OFF stops the four import hooks queueing embed work
nothing will consume and no-ops the manual backfill; switch lives on the
renamed 'CPU embedding backfill' card.
- NB heads training / auto-apply still run on the ml image (sklearn) — a stack
that removes the container gives those up too.
Deploy note: in-flight messages under the old task names are dropped by the
new workers; the 60s orphan sweep + hourly backfill re-fire under the new
names immediately.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
Autoscaler (agent 2026-07-02.5): the buffer-occupancy signal alone would peg
downloaders at DL_MAX while the bandwidth CAP — not concurrency — is the real
constraint (8 streams sharing 8 MB/s move no more data than 4). Growth is now
gated on the pipe having headroom (net < 85% of cap) and a pipe pinned at the
cap (>= 95%) sheds streams down to 3; dead band prevents flapping. The UI hint
says 'holding at the bandwidth cap' and /status reports bw_capped, so the
behavior is legible without tests that need the ML stack.
Reset content tagging: stays a FULL-instance reset (operator's call), but now
lives in a fenced 'Danger zone' section on Cleanup and the apply is gated by a
preview-derived confirm token (mirrors the Tier-C bulk-delete pattern — stale
counts are rejected server-side). Copy no longer claims suggestions repopulate:
it says plainly the heads' training examples are deleted and re-tagging starts
fresh. Moved out of TagMaintenanceCard into DangerZoneCard.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
The head-vs-centroid eval (#1130) existed to prove the 'frozen embedding +
trained head' spine; the operator accepted the tagging system and dropped the
harness. Removed per rule 22: TagEvalCard + store, /api/tag_eval blueprint,
tag_eval_run ml task, recover-stalled-tag-eval-runs sweep + beat entry,
TagEvalRun model + table (migration 0073), and its tests.
The eval's data loaders + metric helpers were NOT eval-specific — the nightly
heads trainer runs on them — so they moved verbatim to
services/ml/training_data.py (heads.py import updated; behavior unchanged).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
An errored GPU job's stored reason is a suspicion; the file probe is the
verdict. A 15-min beat sweep (triage_gpu_errors) runs verify_integrity's own
probe (sha256 + decode) on each errored image ONCE and writes both verdicts:
ImageRecord.integrity_status and the new GpuJob.triage_status ('defect' |
'file_ok', migration 0072). Every classification logs at WARNING so it
surfaces in Logs/System Activity.
- 'defect' rows are excluded from /retry_errors (re-running a known-bad file
burns agent time re-minting the tombstone); response now reports
defects_kept and the GpuAgentCard toast says so.
- GET /api/gpu/errors: triage view — reason buckets (classify_reason),
probe verdicts, per-job detail. POST /errors/triage runs the sweep now.
- POST /api/gpu/errors/<id>/recover: reuses the Layer-2 refetch pattern —
delete the defective copy + record (full cascade takes the tombstones too)
and re-poll its subscription Source so a fresh copy re-imports and re-enters
the pipeline; 'no_source' when nothing pollable resolves.
- New 'Failed processing' card (GpuTriageCard) in Maintenance: verdict counts,
reason summary, probe-now, defect list with thumbnails + per-image Recover.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
- ml-backfill-daily: the CPU tag_and_embed backfill raced the GPU agent's
daily embed backfill for the same NULL-embedding images at ~100x the cost
(B1 audit verdict, milestone #124). The backfill TASK stays — the manual
/api/ml/backfill button remains the deliberate CPU fallback pending B3.
- purge-legacy: one-time IR-migration cleanup, dry-run verified 0 targets on
the live library before removal (A2 audit, milestone #123). Fully retired
per rule 22: tile, store action, route, service fn, tests.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
The hourly ccip backfill's skip-list lacked 'error' (and the daily
siglip/embed variants re-gated failures on their missing results), so every
permanently-bad file got a fresh doomed job each run — ~24 duplicate error
rows/day per file, the perpetual 'unprocessable' flood. An errored job is now
a TOMBSTONE: no backfill re-enqueues it; retry is deliberate-only via
/retry_errors (an errored back-catalogue needs one button press after a
model swap).
One shared set of dedupe DELETEs (services/ml/gpu_jobs.error_dedupe_statements)
runs before every backfill and inside /retry_errors: error rows made moot by a
later pending/leased/done row go first, then older duplicates (newest reason
survives) — so the error count reads as distinct failing files and a retry
can't fan one file out into duplicate pending jobs. /retry_errors now returns
{requeued, pruned} and the toast shows both.
Poison-loop guards (release and lease-expiry burn no attempts, so a job that
stalls its transfer or crashes the agent every time cycled forever —
operator-observed jobs 99044/125288/131594/143131):
- agent: 3 in-session transient bounces (fetch or submit) → fail with the real
reason instead of another release; strikes never count while stopping, and
clear on submit success. Agent build 2026-07-02.3.
- server: the 60s orphan sweep (statements shared between the beat task and
GpuJobService so they can't drift) converts expired leases with >=5 lease
grants and pending jobs with >=10 to 'error', preserving the last stored
failure reason. Backstops old agent builds.
Tests: tombstone rule across all three backfill variants, moot-row pruning,
poison conversions, and the extended /retry_errors dedupe contract.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM
After an agent-side fix (e.g. the short-video sampler), the errored jobs
(~2.8k) have exhausted their 3 attempts and stay parked: backfill skips
images that already have a job, and /reprocess is the nuclear option (it
resets the 179k DONE jobs too). There was no way to re-run just the errors.
POST /api/gpu/retry_errors resets every status='error' job (all task types)
to pending with attempts=0 and the stored error cleared — a small inline
UPDATE that returns {requeued: n} so the UI toast can show the count.
UI: a "Retry errored jobs" button on the GPU-agent card, right under the
queue tiles; disabled when errored==0. With the agent now logging ffmpeg's
stderr on failure, retrying also reveals which errors were real vs victims
of the fps-filter bug.
Test: retry_errors requeues the errored job (fresh attempts, error cleared)
and leaves done work untouched; asserts via column selects (Core-DML
gotcha), not ORM refresh.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The gallery's newest/oldest sort keys off image_record.effective_date =
COALESCE(primary post's post_date, created_at). The primary post is often the
repost/download the file came from, so the grid led with download dates rather
than when content was first posted (operator-flagged).
Add a second materialized sort key, earliest_post_date = MIN(post_date) across
ALL of an image's provenance posts (every post it appears in), else created_at —
the original publish date. Mirrors the effective_date pattern so the sort stays a
forward index scan.
- alembic 0071: add earliest_post_date + index (DESC, id DESC); backfill
created_at baseline then MIN over image_provenance ⋈ post.
- importer: recompute earliest_post_date whenever a dated post is linked (MIN over
the image's provenance, which now includes the just-added row).
- gallery_service: new sorts posted_new / posted_old key off earliest_post_date;
cursor + year/month grouping follow the active column transparently.
- api: accept posted_new|posted_old; DEFAULT is now posted_new so the grid leads
with original publish date. newest/oldest (effective_date) still available.
- frontend: sort dropdown gains "Newest/Oldest post date" (default Newest post
date); existing effective-date sorts relabelled "Newest/Oldest added".
- tests: service test asserts posted_new/posted_old key off earliest_post_date;
frontend default-sort omission test updated.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ttrj5P7upUTueSfoJcxEqa
- Migration 0069: new installs default to SigLIP 2 (so400m, 512px, 1152-d drop-in)
— UPDATE applies ONLY where no image is embedded yet (fresh install), so an
existing library is NOT silently invalidated; it switches deliberately via the
dropdown → Re-embed → Retrain. Column server_defaults moved to SigLIP 2.
- GET /api/ml/embedder-models: server-authoritative supported list (SigLIP 2 512
recommended / 384 faster / SigLIP 1 384 original) so the UI never free-types.
- GpuAgentCard: the two name/version text fields → a single model dropdown;
Save sets name+version from the picked option (the current model is always
selectable even if off-list).
- embedder.py DEFAULT_MODEL_NAME unchanged (stays the baked local-dir SigLIP 1)
to avoid a local-dir/weights mismatch; SigLIP 2 loads by HF name, cached on the
ml-worker's persistent HF_HOME.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ttrj5P7upUTueSfoJcxEqa
The siglip/ccip backfills skip images that already have current-version regions,
so adding crop detectors only affected NEW images — the back-catalogue would
never be re-cropped. Add a reprocess trigger that resets every done/error job of
a task back to pending, so the agent re-runs the FULL pipeline (figure detection
+ CCIP + concept/panel crops) over the whole library under the current detectors.
- reprocess_gpu_jobs(task='ccip') task + POST /api/gpu/reprocess.
- gpu store reprocess() + GpuAgentCard "Re-process library (re-detect + re-crop)"
button with a confirm (it's heavy).
- Test: a done job resets to pending (attempts cleared).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ttrj5P7upUTueSfoJcxEqa
Fast, read-only, indexed aggregates shaped for ANALYSIS (not the paged UI
directory, which is alphabetical + builds previews and timed out at 10 min on a
full count sweep).
- GET /api/tags/top — top tags by image count, desc. ?kind, ?limit (cap 500),
?min_count, ?source=all|human|manual|accepted|auto (human=manual+ml_accepted,
auto=head_auto+ccip_auto+ml_auto). One GROUP BY over image_tag (indexed on
tag_id).
- GET /api/tags/<id>/stats — per-tag dataset health: total + per-source counts
(manual/accepted/head_auto/ccip_auto), human vs auto rollups, rejection count,
and whether a trained head exists. Backs concept-readiness + source-split
analysis.
Plain-HTTP homelab posture, no auth change. Tests cover ranking, source filter,
min_count, the source breakdown, and 404.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ttrj5P7upUTueSfoJcxEqa
Hygiene follow-up to the Camie retirement (#1189) — these were left inert to
bound that change; nothing reads them now. Migration 0068 drops:
- ml_settings: tagger_store_floor, tagger_model_version, suggestion_threshold_
character/general (already dead pre-retirement — scoring uses per-head
thresholds), video_min_tag_frames (only the deleted video-prediction
aggregator used it).
- image_record: tagger_model_version (no writer), centroid_scores (dead JSON
cache, no reader).
Also: ml_admin _EDITABLE/GET/_validate pruned (dropped the store-floor invariant
+ video_min_tag_frames check); MLThresholdSliders trimmed to a video-embedding
card (interval + max frames only); importer no longer resets the dropped cols;
download_models drops the Camie fetch; stale CASCADE comments in cleanup_service
no longer name the removed tables. Tests updated.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ttrj5P7upUTueSfoJcxEqa
Heads + CCIP are the tag source and head auto-apply is the earned propagation.
The Camie tagger ran only to feed the allowlist bulk-apply (its ImagePrediction
rows had no other consumer), and the allowlist was a SECOND, un-earned auto-apply
path firing in parallel with heads on every accept — exactly the un-earned spray
the v2 pivot replaced. Retire both.
Behavior change: accepting a suggestion now applies the tag to THAT image only
(source='ml_accepted', a head-training positive) — it no longer allowlists +
fans the tag across the library via Camie. Propagation is heads' earned
auto-apply. (Loses instant cold-start propagation for booru-vocab tags; that was
un-earned and bypassed the precision gate.)
- tag_and_embed is now EMBED-ONLY (no Camie load/infer, no ImagePrediction
writes); backfill enqueues it for images with no embedding.
- Removed: services/ml/tagger.py, apply_allowlist_tags + helpers + daily beat +
every enqueue caller (accept/alias/merge/per-image), api/allowlist.py +
blueprint, ImagePrediction + TagAllowlist models/tables (migration 0067),
AllowlistTable.vue + allowlist store, the accept coverage-projection payload.
- AllowlistService gutted to accept/dismiss/undismiss/reject (the rejection store
the rail still needs); accept returns nothing, API returns {accepted, tag_id}.
- tag merge no longer repoints/triggers the allowlist; _keep_as_alias now keys on
ML-applied image_tag sources (incl. head_auto) instead of the allowlist.
- UI: MLBackfillCard relabelled to embedding-only; accept toast simplified;
MaintenancePanel drops the allowlist tile.
Left for a follow-up hygiene pass (now-inert, harmless): the dead settings
columns (tagger_store_floor, tagger_model_version, suggestion_threshold_*,
video_min_tag_frames), image_record.tagger_model_version, MLThresholdSliders
trim, and the Camie model download in download_models.py.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ttrj5P7upUTueSfoJcxEqa
The v2 pivot replaced per-tag SigLIP centroids with learned heads + CCIP.
Centroids were still recomputed (on every tag merge + a daily beat) but NOTHING
read them — suggestions come from heads+CCIP and apply_allowlist_tags applies
via Camie predictions, not centroids. Pure dead wiring; remove it.
Removed: CentroidService, recompute_centroid/recompute_centroids tasks, the
daily beat, POST /api/ml/recompute-centroids, the recompute-on-merge trigger,
the tag_reference_embedding table + model, the centroid_similarity_threshold +
min_reference_images settings (migration 0066), the CentroidRecomputeCard +
its store action + MaintenancePanel tile, and the centroid slider in
MLThresholdSliders. _keep_as_alias drops its vestigial has-centroid branch (the
allowlist branch already covers "could re-emit"); tag merge no longer clears a
table that no longer exists.
NOT touched (still live, parallel to heads): the Camie tagger, ImagePrediction,
and the allowlist bulk-apply — accepting a suggestion still allowlists + applies
it across the library. The tag-eval "centroid" baseline metric is unrelated
(in-memory) and stays. (image_record.centroid_scores JSON column also remains —
separate legacy field, its own micro-cleanup.)
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ttrj5P7upUTueSfoJcxEqa
Make the SigLIP embedder an operator choice (drop-in to SigLIP 2:
google/siglip2-so400m-patch16-512 is a verified 1152-d model at 512px → no
schema change, better small-cue fidelity). A swap = set model + re-embed +
retrain, all operator-driven; the GPU agent does the re-embed so it's fast.
- settings: embedder_model_name is now a setting (migration 0065) alongside the
existing embedder_model_version; both editable + validated (non-empty) in the
ml admin API. The server embedder loads by HF name (AutoImageProcessor/Model,
model-agnostic), preferring the pre-downloaded local dir for the default so
existing deploys don't re-download; rebuilds on a name change.
- agent: new 'embed' job = whole-image SigLIP embedding (mean-pool video frames)
under the lease-announced model → POST /jobs/submit_embedding writes
image_record.siglip_embedding + siglip_model_version. The lease now announces
the model FROM THE SETTING (not a constant).
- re-embed routing: enqueue_gpu_backfill('embed') selects unembedded + stale-
version images; 'siglip' now re-embeds concept crops whose version != current
(so a swap re-triggers crops, not just the never-embedded back-catalogue). The
CPU ml-worker backfill no longer re-embeds on a version mismatch (it can't
churn the library at 512px) — the GPU agent owns version re-embeds. Daily
'embed' + 'siglip' beats self-heal.
- scoring: score_image only bags embeddings in the CURRENT model's space (whole-
image gated by siglip_model_version, concept regions by embedding_version) so a
mid-swap stale vector isn't scored by new-space heads; legacy NULL = current.
- UI: GpuAgentCard "Embedding model (advanced)" — edit name/version, Save, and
"Re-embed library (GPU)" (queues embed + siglip); points at SigLIP 2.
Tests: lease announces model + submit_embedding round-trip; enqueue 'embed'
selects stale/unembedded; stale-version excluded from scoring; embedder model
settable + empty rejected; siglip gate updated to current-version concept.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ttrj5P7upUTueSfoJcxEqa
Lift recall on small/local concepts (glasses, cum, stomach-bulge, xray,
lactation) that the whole-image SigLIP vector washes out: the GPU agent now
embeds figure crops with SigLIP too, stored as kind='concept' regions, and the
suggestion rail scores each image as a BAG (whole-image + every concept crop),
taking each head's MAX over the bag. The whole-image vector is always in the
bag, so this can never score lower than before.
Model-agnostic by construction: the server ANNOUNCES the embedding model
(HF name + version) in the lease, so the agent loads whatever the heads were
trained in and stays in lock-step — a model swap is a server setting + a
re-embed migration, never an agent change.
- agent: model-agnostic CropEmbedder (torch/transformers get_image_features,
fp16 on CUDA, inference-locked); worker branches on job.task — 'ccip' emits
figure(CCIP)+concept(SigLIP) in one pass, 'siglip' emits concept-only so the
back-catalogue backfill never churns figure/CCIP regions; torch cu124 +
transformers in the image.
- server: lease announces embed_model_name/embed_version; score_image is
max-over-bag (version-filtered region embeddings); enqueue_gpu_backfill
'siglip' gates on a missing concept region (drains the back-catalogue,
retries failures, no double-enqueue); daily siglip-backfill beat; UI button;
/api/ccip/overview reports images_with_concept_siglip.
- v1 scope: suggestion rail only — auto-apply stays whole-image (conservative;
heads' thresholds were calibrated on whole-image). Bulk-apply bag = follow-up.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ttrj5P7upUTueSfoJcxEqa
Works through the optional CCIP ideas + the "keep moving even if I forget" ask:
AUTOMATION (no button needed):
- Hourly beat auto-enqueues CCIP backfill — new images get embedded (and errored
ones retried) on their own; the queue never goes idle waiting for a click.
- CCIP auto-apply: a daily sweep tags confident matches (source='ccip_auto') so
identity tags keep flowing. ON by default (opt-out, like head auto-apply);
ml_settings.ccip_auto_apply_enabled + _threshold (0.92, above the suggest cut),
migration 0064. Vectorized (one matmul + reduceat per image), reversible, skips
already-applied/rejected. Switch + threshold in the GPU agent card; GET/PATCH
/api/ml/settings; auto_applied count in /api/ccip/overview.
REFERENCE QUALITY (the over-fire root cause):
- character_references now draws ONLY from single-character images — on a
multi-character image the tag is image-level, so every figure would otherwise
pollute each character's prototypes (a 2-char image tagged 'Velma' made
Daphne's figure a Velma reference). This is the contamination behind residual
over-firing.
- Cached on a cheap signature (char-tag count + ccip-region count/max-id) so the
reference load isn't redone on every modal open.
Tests: multi-character image not used as a reference; auto-apply tags a confident
match as ccip_auto.
NEXT (not done, confirmed): comic-panel cropping + SigLIP concept crops ("spot
interesting content").
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ttrj5P7upUTueSfoJcxEqa
Live data showed the v1 flat 0.75 cosine over-fired — ~64% of matched images got
3-10 character guesses dominated by the most-referenced characters (a 27-ref
character clears a low bar on many images). A sweep showed 0.85 collapses the
noise (noisy multi-matches 47→3) while keeping the confident single-character
matches.
- ml_settings.ccip_match_threshold (migration 0063, default 0.85); match_image
reads it (override still accepted). DEFAULT_SIM_THRESHOLD fallback 0.75→0.85.
- Exposed in GET/PATCH /api/ml/settings (validated 0.5–0.999).
- Slider in the GPU agent card ("Character-match strictness") — tune live, no
redeploy, same observe-and-tune loop as auto-apply.
Test: a ~0.9-cosine figure matches at 0.85, dropped at 0.95.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ttrj5P7upUTueSfoJcxEqa
