From 68932b55c5ef01b8eb1481e0823b494ec78bed8d Mon Sep 17 00:00:00 2001 From: LyAhn Date: Fri, 3 Jul 2026 09:21:39 +0100 Subject: [PATCH] fix(tagger): separate JoyTag confidence threshold Store confidence thresholds per tagger model so JoyTag no longer inherits WD tuning. Refresh the active threshold when switching models, guard stale threshold saves, and keep UI Lab mocks in sync. Also tightens the onboarding model selector so the segmented control no longer stretches across the row. --- src-tauri/src/commands.rs | 7 ++- src-tauri/src/tagger.rs | 57 +++++++++++++------- src/components/SettingsModal.tsx | 11 ++-- src/components/onboarding/StepAiFeatures.tsx | 2 +- src/dev/mockBackend.ts | 15 +++++- src/store.ts | 21 +++++--- src/taggerModels.ts | 4 +- 7 files changed, 82 insertions(+), 35 deletions(-) diff --git a/src-tauri/src/commands.rs b/src-tauri/src/commands.rs index a10f80b..96ad3b0 100644 --- a/src-tauri/src/commands.rs +++ b/src-tauri/src/commands.rs @@ -2159,6 +2159,7 @@ pub struct SetTaggerModelParams { #[derive(Deserialize)] pub struct SetTaggerThresholdParams { pub threshold: f32, + pub model: Option, } #[derive(Deserialize)] @@ -2258,7 +2259,11 @@ pub async fn set_tagger_threshold( params: SetTaggerThresholdParams, ) -> Result { let app_dir = app.path().app_data_dir().map_err(|e| e.to_string())?; - tagger::set_tagger_threshold(&app_dir, params.threshold).map_err(|e| e.to_string()) + match params.model { + Some(model) => tagger::set_tagger_threshold_for_model(&app_dir, model, params.threshold), + None => tagger::set_tagger_threshold(&app_dir, params.threshold), + } + .map_err(|e| e.to_string()) } #[tauri::command] diff --git a/src-tauri/src/tagger.rs b/src-tauri/src/tagger.rs index 9cac6a3..d2a7f03 100644 --- a/src-tauri/src/tagger.rs +++ b/src-tauri/src/tagger.rs @@ -16,6 +16,7 @@ pub const WD_TAGGER_MODEL_NAME: &str = "wd-swinv2-tagger-v3"; const TAGGER_ACCELERATION_FILE: &str = "settings/tagger_acceleration.txt"; const TAGGER_THRESHOLD_FILE: &str = "settings/tagger_threshold.txt"; +const JOYTAG_THRESHOLD_FILE: &str = "settings/joytag_threshold.txt"; const TAGGER_BATCH_SIZE_FILE: &str = "settings/tagger_batch_size.txt"; const TAGGER_MODEL_FILE: &str = "settings/tagger_model.txt"; @@ -27,8 +28,7 @@ pub const JOYTAG_MODEL_NAME: &str = "joytag"; // NCHW layout (not NHWC). These are the OpenAI CLIP normalization constants. const JOYTAG_MEAN: [f32; 3] = [0.481_454_66, 0.457_827_5, 0.408_210_73]; const JOYTAG_STD: [f32; 3] = [0.268_629_54, 0.261_302_6, 0.275_777_1]; -// JoyTag's recommended detection threshold (used as the default; the user's -// tagger_threshold setting still overrides it). +// JoyTag's recommended detection threshold. const JOYTAG_DEFAULT_THRESHOLD: f32 = 0.4; // Shared ONNX Runtime DLLs (live in the caption model's `onnxruntime/` dir and @@ -64,8 +64,8 @@ pub const TAGGER_INFER_CHUNK: usize = 4; /// claim the GPU/CPU between dispatches. Negligible against per-chunk inference. pub const TAGGER_INFER_YIELD_MS: u64 = 40; -/// Set to `true` by `set_tagger_acceleration` so the tagging worker loop -/// knows to drop its cached `WdTagger` and reload with the new EP. +/// Set to `true` by tagger setting changes so the tagging worker loop knows to +/// drop its cached model session and reload with the current settings. pub static TAGGER_SESSION_DIRTY: AtomicBool = AtomicBool::new(false); // --------------------------------------------------------------------------- @@ -112,6 +112,20 @@ impl TaggerModel { } } + fn threshold_file(self) -> &'static str { + match self { + Self::Wd => TAGGER_THRESHOLD_FILE, + Self::JoyTag => JOYTAG_THRESHOLD_FILE, + } + } + + fn default_threshold(self) -> f32 { + match self { + Self::Wd => DEFAULT_THRESHOLD, + Self::JoyTag => JOYTAG_DEFAULT_THRESHOLD, + } + } + /// Hugging Face repo the model files are fetched from. fn repo_id(self) -> &'static str { match self { @@ -280,21 +294,33 @@ pub fn set_tagger_acceleration( Ok(acceleration) } -pub fn tagger_threshold(app_data_dir: &Path) -> f32 { - let path = app_data_dir.join(TAGGER_THRESHOLD_FILE); +fn tagger_threshold_for_model(app_data_dir: &Path, model: TaggerModel) -> f32 { + let path = app_data_dir.join(model.threshold_file()); let Ok(value) = std::fs::read_to_string(path) else { - return DEFAULT_THRESHOLD; + return model.default_threshold(); }; value .trim() .parse::() - .unwrap_or(DEFAULT_THRESHOLD) + .unwrap_or_else(|_| model.default_threshold()) .clamp(0.01, 1.0) } +pub fn tagger_threshold(app_data_dir: &Path) -> f32 { + tagger_threshold_for_model(app_data_dir, tagger_model(app_data_dir)) +} + pub fn set_tagger_threshold(app_data_dir: &Path, threshold: f32) -> Result { + set_tagger_threshold_for_model(app_data_dir, tagger_model(app_data_dir), threshold) +} + +pub fn set_tagger_threshold_for_model( + app_data_dir: &Path, + model: TaggerModel, + threshold: f32, +) -> Result { let clamped = threshold.clamp(0.01, 1.0); - let path = app_data_dir.join(TAGGER_THRESHOLD_FILE); + let path = app_data_dir.join(model.threshold_file()); if let Some(parent) = path.parent() { std::fs::create_dir_all(parent)?; } @@ -1057,17 +1083,10 @@ fn load_joytag_labels(path: &Path) -> Result> { Ok(labels) } -/// JoyTag detection threshold. Honours the shared `tagger_threshold` setting if -/// the user has set it, otherwise uses JoyTag's recommended default (0.4). +/// JoyTag detection threshold. Uses JoyTag's own setting so WD tuning does not +/// leak into the general/photo-friendly model. fn joytag_threshold(app_data_dir: &Path) -> f32 { - match std::fs::read_to_string(app_data_dir.join(TAGGER_THRESHOLD_FILE)) { - Ok(value) => value - .trim() - .parse::() - .unwrap_or(JOYTAG_DEFAULT_THRESHOLD) - .clamp(0.01, 1.0), - Err(_) => JOYTAG_DEFAULT_THRESHOLD, - } + tagger_threshold_for_model(app_data_dir, TaggerModel::JoyTag) } fn sigmoid(x: f32) -> f32 { diff --git a/src/components/SettingsModal.tsx b/src/components/SettingsModal.tsx index cee18b0..6ab464e 100644 --- a/src/components/SettingsModal.tsx +++ b/src/components/SettingsModal.tsx @@ -461,6 +461,9 @@ export function SettingsModal() { current={taggerModel} onSelect={(nextModel) => { if (nextModel === taggerModel) return; + setTaggerThresholdDraft(null); + setTaggerThresholdError(null); + if (thresholdErrorTimerRef.current) clearTimeout(thresholdErrorTimerRef.current); setTaggerModelSwitching(true); setTaggerModelSwitchError(null); void setTaggerModel(nextModel) @@ -562,12 +565,12 @@ export function SettingsModal() { className="w-20 rounded-md border border-white/10 bg-white/5 px-2 py-1.5 text-xs text-white focus:border-white/20 focus:outline-none" value={thresholdDisplay} onChange={(event) => setTaggerThresholdDraft(event.target.value)} - onBlur={() => { - const value = parseFloat(thresholdDisplay); + onBlur={(event) => { + const value = parseFloat(event.currentTarget.value); if (!isNaN(value) && value >= 0.05 && value <= 0.99) { setTaggerThresholdError(null); setTaggerThresholdSaving(true); - void setTaggerThreshold(value) + void setTaggerThreshold(value, taggerModel) .catch((error: unknown) => setTaggerThresholdError(String(error))) .finally(() => { setTaggerThresholdDraft(null); @@ -584,7 +587,7 @@ export function SettingsModal() { {taggerThresholdError ? (

{taggerThresholdError}

) : ( -

{taggerThresholdSaving ? "Saving..." : `Default: ${taggerModel === "joytag" ? "0.4" : "0.35"}`}

+

{taggerThresholdSaving ? "Saving..." : `Default: ${TAGGER_MODELS[taggerModel].defaultThreshold}`}

)} diff --git a/src/components/onboarding/StepAiFeatures.tsx b/src/components/onboarding/StepAiFeatures.tsx index c32b679..901a426 100644 --- a/src/components/onboarding/StepAiFeatures.tsx +++ b/src/components/onboarding/StepAiFeatures.tsx @@ -75,7 +75,7 @@ export function StepAiFeatures() { The AI tagger model labels images so you can search with{" "} /t — tags look like:

-
+
{(["wd", "joytag"] as const).map((model) => ( = { + wd: 0.35, + joytag: 0.4, +}; type AnyPayload = Record | undefined; type Rgb = [number, number, number]; @@ -42,6 +46,10 @@ function isRgb(value: unknown): value is Rgb { return Array.isArray(value) && value.length === 3 && value.every((entry) => typeof entry === "number"); } +function isTaggerModel(value: unknown): value is TaggerModel { + return value === "wd" || value === "joytag"; +} + function colorDistanceSq(left: Rgb, right: Rgb): number { return (left[0] - right[0]) ** 2 + (left[1] - right[1]) ** 2 + (left[2] - right[2]) ** 2; } @@ -528,8 +536,13 @@ export async function handleMockCommand(cmd: string, payload?: unknown): Promise case "set_tagger_acceleration": return p.acceleration ?? "auto"; case "get_tagger_threshold": + return mockTaggerThresholds[mockTaggerModel]; case "set_tagger_threshold": - return p.threshold ?? 0.35; + { + const model = isTaggerModel(p.model) ? p.model : mockTaggerModel; + mockTaggerThresholds[model] = p.threshold ?? mockTaggerThresholds[model]; + return mockTaggerThresholds[model]; + } case "get_tagger_batch_size": case "set_tagger_batch_size": return p.batch_size ?? 8; diff --git a/src/store.ts b/src/store.ts index 9f0933b..1cd1d50 100644 --- a/src/store.ts +++ b/src/store.ts @@ -589,7 +589,7 @@ interface GalleryState { loadTaggerModel: () => Promise; setTaggerModel: (model: TaggerModel) => Promise; loadTaggerThreshold: () => Promise; - setTaggerThreshold: (threshold: number) => Promise; + setTaggerThreshold: (threshold: number, model?: TaggerModel) => Promise; loadTaggerBatchSize: () => Promise; setTaggerBatchSize: (batchSize: number) => Promise; probeTaggerRuntime: () => Promise; @@ -2326,11 +2326,14 @@ export const useGalleryStore = create((set, get) => ({ const taggerModel = await invoke("set_tagger_model", { params: { model }, }); - // Switching models changes which files are required on disk, so refresh the - // status too — the download/ready UI must reflect the newly-selected model. + // Switching models changes both readiness and the active threshold setting, + // so refresh them together for the selected model. try { - const taggerModelStatus = await invoke("get_tagger_model_status"); - set({ taggerModel, taggerModelStatus, taggerModelError: null, taggerRuntimeProbe: null }); + const [taggerModelStatus, taggerThreshold] = await Promise.all([ + invoke("get_tagger_model_status"), + invoke("get_tagger_threshold"), + ]); + set({ taggerModel, taggerModelStatus, taggerThreshold, taggerModelError: null, taggerRuntimeProbe: null }); } catch (error) { set({ taggerModel, taggerRuntimeProbe: null, taggerModelError: String(error) }); } @@ -2345,11 +2348,13 @@ export const useGalleryStore = create((set, get) => ({ } }, - setTaggerThreshold: async (threshold) => { + setTaggerThreshold: async (threshold, model) => { const taggerThreshold = await invoke("set_tagger_threshold", { - params: { threshold }, + params: { threshold, model }, }); - set({ taggerThreshold }); + if (!model || get().taggerModel === model) { + set({ taggerThreshold }); + } }, loadTaggerBatchSize: async () => { diff --git a/src/taggerModels.ts b/src/taggerModels.ts index e3ac289..53a3546 100644 --- a/src/taggerModels.ts +++ b/src/taggerModels.ts @@ -1,15 +1,17 @@ import type { TaggerModel } from "./store"; -export const TAGGER_MODELS: Record = { +export const TAGGER_MODELS: Record = { wd: { name: "WD SwinV2 Tagger v3", tab: "WD (anime)", + defaultThreshold: 0.35, description: "Anime-focused vision model by SmilingWolf. Generates booru-style tags with configurable confidence thresholds.", }, joytag: { name: "JoyTag", tab: "JoyTag (general)", + defaultThreshold: 0.4, description: "Booru-schema tagger that also handles photographic content and is strong on NSFW concepts. The explicitness rating is derived from its tags.", },