@qvac/sdk 0.15.0 → 0.17.0

package/dist/schemas/text-to-speech.js

@@ -63,6 +63,20 @@ export const TTS_SUPERTONIC_LANGUAGES = [
     'uk', // Ukrainian
     'vi' // Vietnamese
 ];
+export const TTS_PARLER_EMOTIONS = [
+    'command',
+    'anger',
+    'narration',
+    'conversation',
+    'disgust',
+    'fear',
+    'happy',
+    'neutral',
+    'proper noun',
+    'news',
+    'sad',
+    'surprise'
+];
 // Supertonic languages not already present in the Chatterbox set, used to keep
 // TTS_LANGUAGES a true union across engines without duplicates.
 const TTS_SUPERTONIC_ONLY_LANGUAGES = [
@@ -88,14 +102,72 @@ export const TTS_LANGUAGES = [
 ];
 const ttsChatterboxLanguageSchema = z.enum(TTS_CHATTERBOX_LANGUAGES);
 const ttsSupertonicLanguageSchema = z.enum(TTS_SUPERTONIC_LANGUAGES);
+const ttsParlerEmotionSchema = z.enum(TTS_PARLER_EMOTIONS);
 const ttsIntegerSchema = z.number().int();
 const ttsNonNegativeIntegerSchema = ttsIntegerSchema.nonnegative();
 const ttsPositiveIntegerSchema = ttsIntegerSchema.positive();
+const ttsInt32Schema = ttsIntegerSchema.min(-2147483648).max(2147483647);
+const ttsNonNegativeInt32Schema = ttsInt32Schema.nonnegative();
+const ttsPositiveInt32Schema = ttsInt32Schema.positive();
 // Desired output sample rate in Hz. Matches the @qvac/tts-ggml addon's
 // accepted range; omit to keep the engine's native rate (or 48 kHz when the
 // LavaSR enhancer is active). Supertonic-only: the Chatterbox engine does not
 // yet resample its output, so the field is not exposed on that config.
 const ttsOutputSampleRateSchema = ttsIntegerSchema.min(8000).max(192000);
+const ttsParlerDescriptionFieldsShape = {
+    description: z.string().min(1).optional(),
+    voiceDescription: z.string().min(1).optional(),
+    voice: z.string().min(1).optional(),
+    emotion: ttsParlerEmotionSchema.optional(),
+    pitch: z.string().min(1).optional(),
+    pace: z.string().min(1).optional(),
+    expressivity: z.string().min(1).optional(),
+    noise: z.string().min(1).optional(),
+    reverb: z.string().min(1).optional(),
+    quality: z.string().min(1).optional()
+};
+const ttsParlerTemplateFieldNames = [
+    'voice',
+    'emotion',
+    'pitch',
+    'pace',
+    'expressivity',
+    'noise',
+    'reverb',
+    'quality'
+];
+function refineParlerDescriptionFields(config, ctx) {
+    if (config.description !== undefined && config.voiceDescription !== undefined) {
+        ctx.addIssue({
+            code: 'custom',
+            path: ['voiceDescription'],
+            message: 'description and voiceDescription are mutually exclusive.'
+        });
+    }
+    if (config.description === undefined && config.voiceDescription === undefined)
+        return;
+    const conflictingField = ttsParlerTemplateFieldNames.find((field) => config[field] !== undefined);
+    if (conflictingField) {
+        ctx.addIssue({
+            code: 'custom',
+            path: [conflictingField],
+            message: 'description and voiceDescription are mutually exclusive with Parler voice-template fields.'
+        });
+    }
+}
+function refineParlerRuntimeConfig(config, ctx) {
+    refineParlerDescriptionFields(config, ctx);
+    const nativeStreamingEnabled = (config.streamChunkTokens ?? 0) > 0;
+    if (nativeStreamingEnabled &&
+        config.outputSampleRate !== undefined &&
+        config.outputSampleRate !== 44100) {
+        ctx.addIssue({
+            code: 'custom',
+            path: ['outputSampleRate'],
+            message: 'Parler native streaming emits at 44100 Hz; omit outputSampleRate, use 44100, or disable native streaming.'
+        });
+    }
+}
 export const ttsChatterboxRuntimeConfigSchema = z.object({
     ttsEngine: z.literal('chatterbox'),
     language: ttsChatterboxLanguageSchema,
@@ -105,6 +177,7 @@ export const ttsChatterboxRuntimeConfigSchema = z.object({
     streamChunkTokens: ttsNonNegativeIntegerSchema.optional(),
     streamFirstChunkTokens: ttsNonNegativeIntegerSchema.optional(),
     cfmSteps: ttsNonNegativeIntegerSchema.optional(),
+    cfgRate: z.number().nonnegative().optional(),
     threads: ttsPositiveIntegerSchema.optional(),
     nGpuLayers: ttsIntegerSchema.optional(),
     seed: ttsIntegerSchema.optional()
@@ -116,11 +189,32 @@ export const ttsSupertonicRuntimeConfigSchema = z.object({
     ttsSpeed: z.number().optional(),
     ttsNumInferenceSteps: z.number().optional(),
     useGPU: z.boolean().optional(),
-    outputSampleRate: ttsOutputSampleRateSchema.optional()
+    outputSampleRate: ttsOutputSampleRateSchema.optional(),
+    vulkanCacheDir: z.string().min(1).optional()
 });
+export const ttsParlerRuntimeConfigSchema = z
+    .object({
+    ttsEngine: z.literal('parler'),
+    ...ttsParlerDescriptionFieldsShape,
+    useGPU: z.boolean().optional(),
+    outputSampleRate: ttsOutputSampleRateSchema.optional(),
+    streamChunkTokens: ttsNonNegativeInt32Schema.optional(),
+    streamFirstChunkTokens: ttsNonNegativeInt32Schema.optional(),
+    threads: ttsPositiveInt32Schema.optional(),
+    nGpuLayers: ttsInt32Schema.optional(),
+    seed: ttsInt32Schema.optional(),
+    temperature: z.number().nonnegative().optional(),
+    topK: ttsNonNegativeInt32Schema.optional(),
+    topP: z.number().positive().max(1).optional(),
+    maxFrames: z.union([z.literal(0), ttsInt32Schema.min(10)]).optional(),
+    minNewTokens: ttsInt32Schema.min(-1).optional(),
+    normalizeNumbers: z.boolean().optional()
+})
+    .superRefine(refineParlerRuntimeConfig);
 export const ttsRuntimeConfigSchema = z.discriminatedUnion('ttsEngine', [
     ttsChatterboxRuntimeConfigSchema,
-    ttsSupertonicRuntimeConfigSchema
+    ttsSupertonicRuntimeConfigSchema,
+    ttsParlerRuntimeConfigSchema
 ]);
 // Optional LavaSR post-processing model sources, shared across engines. Supply
 // the enhancer GGUF to bandwidth-extend the output to 48 kHz, and/or the
@@ -142,6 +236,7 @@ export const ttsChatterboxLoadConfigSchema = ttsChatterboxRuntimeConfigSchema.ex
 export const ttsSupertonicLoadConfigSchema = ttsSupertonicRuntimeConfigSchema.extend({
     ...ttsLavasrLoadFieldsShape
 });
+export const ttsParlerLoadConfigSchema = ttsParlerRuntimeConfigSchema;
 function refineChatterboxTokenizerAssets(data, ctx) {
     if (data.ttsEngine !== 'chatterbox')
         return;
@@ -161,7 +256,11 @@ function refineChatterboxTokenizerAssets(data, ctx) {
     }
 }
 export const ttsLoadConfigSchema = z
-    .discriminatedUnion('ttsEngine', [ttsChatterboxLoadConfigSchema, ttsSupertonicLoadConfigSchema])
+    .discriminatedUnion('ttsEngine', [
+    ttsChatterboxLoadConfigSchema,
+    ttsSupertonicLoadConfigSchema,
+    ttsParlerLoadConfigSchema
+])
     .superRefine(refineChatterboxTokenizerAssets);
 // === Legacy ONNX modelConfig fields (deprecated) ===
 //
@@ -193,25 +292,41 @@ const legacyTtsOnnxFieldsShape = LEGACY_TTS_ONNX_MODEL_CONFIG_FIELDS.reduce((acc
 // `loadConfigSchema`. Permits deprecated ONNX field names so
 // `resolveConfig` can raise LegacyTtsModelDeprecatedError instead of a
 // generic Zod error; other unknown keys are still rejected by `.strict()`.
-export const ttsConfigSchema = z.discriminatedUnion('ttsEngine', [
+export const ttsConfigSchema = z
+    .discriminatedUnion('ttsEngine', [
     ttsChatterboxLoadConfigSchema.extend(legacyTtsOnnxFieldsShape).strict(),
-    ttsSupertonicLoadConfigSchema.extend(legacyTtsOnnxFieldsShape).strict()
-]).superRefine(refineChatterboxTokenizerAssets);
-export const ttsClientParamsSchema = z.object({
+    ttsSupertonicLoadConfigSchema.extend(legacyTtsOnnxFieldsShape).strict(),
+    ttsParlerLoadConfigSchema.strict()
+])
+    .superRefine(refineChatterboxTokenizerAssets);
+const ttsClientParamsShape = {
     modelId: z.string(),
     inputType: z.string().default('text'),
     text: z.string().trim().min(1, 'text must not be empty or whitespace-only'),
     stream: z.boolean().default(true),
     sentenceStream: z.boolean().default(false),
     sentenceStreamLocale: z.string().optional(),
-    sentenceStreamMaxChunkScalars: z.number().positive().optional()
-});
-export const ttsRequestSchema = ttsClientParamsSchema.extend({
+    sentenceStreamMaxChunkScalars: z.number().positive().optional(),
+    ...ttsParlerDescriptionFieldsShape
+};
+// Requests carry only an opaque modelId, so client-side validation cannot know
+// which TTS engine is loaded. Description conflicts are intentionally rejected
+// as malformed request shapes before the server performs engine compatibility
+// validation in assertParlerJobOptionsSupported.
+export const ttsClientParamsSchema = z
+    .object(ttsClientParamsShape)
+    .superRefine(refineParlerDescriptionFields);
+export const ttsRequestSchema = z
+    .object({
+    ...ttsClientParamsShape,
     type: z.literal('textToSpeech')
-});
+})
+    .superRefine(refineParlerDescriptionFields);
 export const ttsStatsSchema = z.object({
     audioDuration: z.number().optional(),
-    totalSamples: z.number().optional()
+    totalSamples: z.number().optional(),
+    enhancerBackendDevice: z.number().optional(),
+    enhancerBackendId: z.number().optional()
 });
 export const ttsResponseSchema = z.object({
     type: z.literal('textToSpeech'),
@@ -224,17 +339,21 @@ export const ttsResponseSchema = z.object({
 // Internal: kept un-exported to present a single request-schema surface to
 // consumers. The inferred `TextToSpeechStreamClientParams` type below uses
 // this shape via `typeof`, no runtime export needed.
-const textToSpeechStreamRequestBaseSchema = z.object({
+const textToSpeechStreamRequestBaseShape = {
     modelId: z.string(),
     inputType: z.string().default('text'),
     accumulateSentences: z.boolean().optional(),
     sentenceDelimiterPreset: z.enum(['latin', 'cjk', 'multilingual']).optional(),
     maxBufferScalars: z.number().positive().optional(),
-    flushAfterMs: z.number().positive().optional()
-});
-export const textToSpeechStreamRequestSchema = textToSpeechStreamRequestBaseSchema.extend({
+    flushAfterMs: z.number().positive().optional(),
+    ...ttsParlerDescriptionFieldsShape
+};
+export const textToSpeechStreamRequestSchema = z
+    .object({
+    ...textToSpeechStreamRequestBaseShape,
     type: z.literal('textToSpeechStream')
-});
+})
+    .superRefine(refineParlerDescriptionFields);
 export const textToSpeechStreamResponseSchema = z.object({
     type: z.literal('textToSpeechStream'),
     buffer: z.array(z.number()),

package/dist/schemas/tools.d.ts

@@ -31,7 +31,7 @@ export declare const toolSchema: z.ZodObject<{
                 integer: "integer";
             }>;
             description: z.ZodOptional<z.ZodString>;
-            enum: z.ZodOptional<z.ZodArray<z.ZodString>>;
+            enum: z.ZodOptional<z.ZodArray<z.ZodUnion<readonly [z.ZodString, z.ZodNumber, z.ZodBoolean, z.ZodNull]>>>;
         }, z.core.$strip>>;
         required: z.ZodOptional<z.ZodArray<z.ZodString>>;
     }, z.core.$strip>;

package/dist/schemas/tools.js

@@ -14,6 +14,7 @@ export const TOOLS_MODE = {
     static: 'static',
     dynamic: 'dynamic'
 };
+const jsonSchemaEnumValueSchema = z.union([z.string(), z.number(), z.boolean(), z.null()]);
 export const toolSchema = z.object({
     type: z.literal('function'),
     name: z.string(),
@@ -23,7 +24,7 @@ export const toolSchema = z.object({
         properties: z.record(z.string(), z.object({
             type: z.enum(['string', 'number', 'integer', 'boolean', 'object', 'array']),
             description: z.string().optional(),
-            enum: z.array(z.string()).optional()
+            enum: z.array(jsonSchemaEnumValueSchema).optional()
         })),
         required: z.array(z.string()).optional()
     })

package/dist/schemas/transcription-config.d.ts

@@ -78,6 +78,7 @@ export declare const whisperConfigSchema: z.ZodObject<{
         addon: z.ZodOptional<z.ZodUnion<readonly [z.ZodEnum<{
             embeddings: "embeddings";
             diffusion: "diffusion";
+            audiogen: "audiogen";
             tts: "tts";
             ocr: "ocr";
             vla: "vla";
@@ -91,6 +92,7 @@ export declare const whisperConfigSchema: z.ZodObject<{
             "parakeet-transcription": "parakeet-transcription";
             "ggml-ocr": "ggml-ocr";
             "sdcpp-generation": "sdcpp-generation";
+            "audiogen-ggml": "audiogen-ggml";
             "ggml-vla": "ggml-vla";
             "ggml-classification": "ggml-classification";
             llm: "llm";

package/dist/schemas/translate.d.ts

@@ -1,6 +1,6 @@
 import { z } from 'zod';
-import type { NmtLanguage } from './translation-config';
-import { type NmtModelTypeInput, type LlmModelTypeInput } from './model-types';
+import type { NmtLanguage } from './translation-config.js';
+import { type NmtModelTypeInput, type LlmModelTypeInput } from './model-types.js';
 declare const translateParamsSchema: z.ZodDiscriminatedUnion<[z.ZodObject<{
     modelId: z.ZodString;
     text: z.ZodUnion<readonly [z.ZodString, z.ZodArray<z.ZodString>]>;

package/dist/schemas/translate.js

@@ -58,14 +58,18 @@ const translateRequestIdField = z
     .optional()
     .describe('Stable identifier for this in-flight translation, generated by the client at call time. Surfaces on the registry so callers can target it with `cancel({ requestId })`. Optional on the wire — the server falls back to a server-generated id when the field is missing.');
 export const translateRequestSchema = z.discriminatedUnion('modelType', [
-    translateParamsNmtSchema.extend({
+    translateParamsNmtSchema
+        .extend({
         type: z.literal('translate'),
         requestId: translateRequestIdField
-    }),
-    translateParamsLlmSchema.extend({
+    })
+        .meta({ title: 'TranslateNmtRequest' }),
+    translateParamsLlmSchema
+        .extend({
         type: z.literal('translate'),
         requestId: translateRequestIdField
     })
+        .meta({ title: 'TranslateLlmRequest' })
 ]);
 // Valid model types for translation (aliases and canonical)
 const validTranslationModelTypes = ['nmt', 'nmtcpp-translation', 'llm', 'llamacpp-completion'];
@@ -78,11 +82,13 @@ export const translateServerParamsSchema = translateParamsSchema
     .refine((data) => {
     if (!llmModelTypes.includes(data.modelType))
         return true;
-    // For LLM, check from/to exist
+    // For LLM, `to` is required; `from` is resolved by the worker
+    // (explicit, else auto-detected in server/bare/ops/translate.ts), so an
+    // absent `from` is valid here — it no longer has to arrive on the wire.
     const llmData = data;
-    return llmData.from && llmData.to;
+    return !!llmData.to;
 }, {
-    message: "Both 'from' and 'to' languages are required for LLM translation models"
+    message: "A 'to' language is required for LLM translation models"
 })
     .transform((data) => ({
     ...data,

package/dist/schemas/translation-config.d.ts

@@ -146,6 +146,7 @@ export declare const nmtConfigBaseSchema: z.ZodDiscriminatedUnion<[z.ZodObject<{
         addon: z.ZodOptional<z.ZodUnion<readonly [z.ZodEnum<{
             embeddings: "embeddings";
             diffusion: "diffusion";
+            audiogen: "audiogen";
             tts: "tts";
             ocr: "ocr";
             vla: "vla";
@@ -159,6 +160,7 @@ export declare const nmtConfigBaseSchema: z.ZodDiscriminatedUnion<[z.ZodObject<{
             "parakeet-transcription": "parakeet-transcription";
             "ggml-ocr": "ggml-ocr";
             "sdcpp-generation": "sdcpp-generation";
+            "audiogen-ggml": "audiogen-ggml";
             "ggml-vla": "ggml-vla";
             "ggml-classification": "ggml-classification";
             llm: "llm";
@@ -183,6 +185,7 @@ export declare const nmtConfigBaseSchema: z.ZodDiscriminatedUnion<[z.ZodObject<{
         addon: z.ZodOptional<z.ZodUnion<readonly [z.ZodEnum<{
             embeddings: "embeddings";
             diffusion: "diffusion";
+            audiogen: "audiogen";
             tts: "tts";
             ocr: "ocr";
             vla: "vla";
@@ -196,6 +199,7 @@ export declare const nmtConfigBaseSchema: z.ZodDiscriminatedUnion<[z.ZodObject<{
             "parakeet-transcription": "parakeet-transcription";
             "ggml-ocr": "ggml-ocr";
             "sdcpp-generation": "sdcpp-generation";
+            "audiogen-ggml": "audiogen-ggml";
             "ggml-vla": "ggml-vla";
             "ggml-classification": "ggml-classification";
             llm: "llm";
@@ -233,6 +237,7 @@ export declare const nmtConfigBaseSchema: z.ZodDiscriminatedUnion<[z.ZodObject<{
             addon: z.ZodOptional<z.ZodUnion<readonly [z.ZodEnum<{
                 embeddings: "embeddings";
                 diffusion: "diffusion";
+                audiogen: "audiogen";
                 tts: "tts";
                 ocr: "ocr";
                 vla: "vla";
@@ -246,6 +251,7 @@ export declare const nmtConfigBaseSchema: z.ZodDiscriminatedUnion<[z.ZodObject<{
                 "parakeet-transcription": "parakeet-transcription";
                 "ggml-ocr": "ggml-ocr";
                 "sdcpp-generation": "sdcpp-generation";
+                "audiogen-ggml": "audiogen-ggml";
                 "ggml-vla": "ggml-vla";
                 "ggml-classification": "ggml-classification";
                 llm: "llm";
@@ -270,6 +276,7 @@ export declare const nmtConfigBaseSchema: z.ZodDiscriminatedUnion<[z.ZodObject<{
             addon: z.ZodOptional<z.ZodUnion<readonly [z.ZodEnum<{
                 embeddings: "embeddings";
                 diffusion: "diffusion";
+                audiogen: "audiogen";
                 tts: "tts";
                 ocr: "ocr";
                 vla: "vla";
@@ -283,6 +290,7 @@ export declare const nmtConfigBaseSchema: z.ZodDiscriminatedUnion<[z.ZodObject<{
                 "parakeet-transcription": "parakeet-transcription";
                 "ggml-ocr": "ggml-ocr";
                 "sdcpp-generation": "sdcpp-generation";
+                "audiogen-ggml": "audiogen-ggml";
                 "ggml-vla": "ggml-vla";
                 "ggml-classification": "ggml-classification";
                 llm: "llm";
@@ -307,6 +315,7 @@ export declare const nmtConfigBaseSchema: z.ZodDiscriminatedUnion<[z.ZodObject<{
             addon: z.ZodOptional<z.ZodUnion<readonly [z.ZodEnum<{
                 embeddings: "embeddings";
                 diffusion: "diffusion";
+                audiogen: "audiogen";
                 tts: "tts";
                 ocr: "ocr";
                 vla: "vla";
@@ -320,6 +329,7 @@ export declare const nmtConfigBaseSchema: z.ZodDiscriminatedUnion<[z.ZodObject<{
                 "parakeet-transcription": "parakeet-transcription";
                 "ggml-ocr": "ggml-ocr";
                 "sdcpp-generation": "sdcpp-generation";
+                "audiogen-ggml": "audiogen-ggml";
                 "ggml-vla": "ggml-vla";
                 "ggml-classification": "ggml-classification";
                 llm: "llm";
@@ -541,6 +551,7 @@ export declare const nmtConfigSchema: z.ZodPipe<z.ZodDiscriminatedUnion<[z.ZodOb
         addon: z.ZodOptional<z.ZodUnion<readonly [z.ZodEnum<{
             embeddings: "embeddings";
             diffusion: "diffusion";
+            audiogen: "audiogen";
             tts: "tts";
             ocr: "ocr";
             vla: "vla";
@@ -554,6 +565,7 @@ export declare const nmtConfigSchema: z.ZodPipe<z.ZodDiscriminatedUnion<[z.ZodOb
             "parakeet-transcription": "parakeet-transcription";
             "ggml-ocr": "ggml-ocr";
             "sdcpp-generation": "sdcpp-generation";
+            "audiogen-ggml": "audiogen-ggml";
             "ggml-vla": "ggml-vla";
             "ggml-classification": "ggml-classification";
             llm: "llm";
@@ -578,6 +590,7 @@ export declare const nmtConfigSchema: z.ZodPipe<z.ZodDiscriminatedUnion<[z.ZodOb
         addon: z.ZodOptional<z.ZodUnion<readonly [z.ZodEnum<{
             embeddings: "embeddings";
             diffusion: "diffusion";
+            audiogen: "audiogen";
             tts: "tts";
             ocr: "ocr";
             vla: "vla";
@@ -591,6 +604,7 @@ export declare const nmtConfigSchema: z.ZodPipe<z.ZodDiscriminatedUnion<[z.ZodOb
             "parakeet-transcription": "parakeet-transcription";
             "ggml-ocr": "ggml-ocr";
             "sdcpp-generation": "sdcpp-generation";
+            "audiogen-ggml": "audiogen-ggml";
             "ggml-vla": "ggml-vla";
             "ggml-classification": "ggml-classification";
             llm: "llm";
@@ -628,6 +642,7 @@ export declare const nmtConfigSchema: z.ZodPipe<z.ZodDiscriminatedUnion<[z.ZodOb
             addon: z.ZodOptional<z.ZodUnion<readonly [z.ZodEnum<{
                 embeddings: "embeddings";
                 diffusion: "diffusion";
+                audiogen: "audiogen";
                 tts: "tts";
                 ocr: "ocr";
                 vla: "vla";
@@ -641,6 +656,7 @@ export declare const nmtConfigSchema: z.ZodPipe<z.ZodDiscriminatedUnion<[z.ZodOb
                 "parakeet-transcription": "parakeet-transcription";
                 "ggml-ocr": "ggml-ocr";
                 "sdcpp-generation": "sdcpp-generation";
+                "audiogen-ggml": "audiogen-ggml";
                 "ggml-vla": "ggml-vla";
                 "ggml-classification": "ggml-classification";
                 llm: "llm";
@@ -665,6 +681,7 @@ export declare const nmtConfigSchema: z.ZodPipe<z.ZodDiscriminatedUnion<[z.ZodOb
             addon: z.ZodOptional<z.ZodUnion<readonly [z.ZodEnum<{
                 embeddings: "embeddings";
                 diffusion: "diffusion";
+                audiogen: "audiogen";
                 tts: "tts";
                 ocr: "ocr";
                 vla: "vla";
@@ -678,6 +695,7 @@ export declare const nmtConfigSchema: z.ZodPipe<z.ZodDiscriminatedUnion<[z.ZodOb
                 "parakeet-transcription": "parakeet-transcription";
                 "ggml-ocr": "ggml-ocr";
                 "sdcpp-generation": "sdcpp-generation";
+                "audiogen-ggml": "audiogen-ggml";
                 "ggml-vla": "ggml-vla";
                 "ggml-classification": "ggml-classification";
                 llm: "llm";
@@ -702,6 +720,7 @@ export declare const nmtConfigSchema: z.ZodPipe<z.ZodDiscriminatedUnion<[z.ZodOb
             addon: z.ZodOptional<z.ZodUnion<readonly [z.ZodEnum<{
                 embeddings: "embeddings";
                 diffusion: "diffusion";
+                audiogen: "audiogen";
                 tts: "tts";
                 ocr: "ocr";
                 vla: "vla";
@@ -715,6 +734,7 @@ export declare const nmtConfigSchema: z.ZodPipe<z.ZodDiscriminatedUnion<[z.ZodOb
                 "parakeet-transcription": "parakeet-transcription";
                 "ggml-ocr": "ggml-ocr";
                 "sdcpp-generation": "sdcpp-generation";
+                "audiogen-ggml": "audiogen-ggml";
                 "ggml-vla": "ggml-vla";
                 "ggml-classification": "ggml-classification";
                 llm: "llm";
@@ -820,7 +840,7 @@ export declare const nmtConfigSchema: z.ZodPipe<z.ZodDiscriminatedUnion<[z.ZodOb
         engine?: string | undefined;
         expectedSize?: number | undefined;
         sha256Checksum?: string | undefined;
-        addon?: "embeddings" | "diffusion" | "tts" | "ocr" | "vla" | "llamacpp-completion" | "whispercpp-transcription" | "bci-whispercpp-transcription" | "llamacpp-embedding" | "nmtcpp-translation" | "onnx-tts" | "tts-ggml" | "parakeet-transcription" | "ggml-ocr" | "sdcpp-generation" | "ggml-vla" | "ggml-classification" | "llm" | "whisper" | "bci" | "nmt" | "parakeet" | "classification" | "vad" | undefined;
+        addon?: "embeddings" | "diffusion" | "audiogen" | "tts" | "ocr" | "vla" | "llamacpp-completion" | "whispercpp-transcription" | "bci-whispercpp-transcription" | "llamacpp-embedding" | "nmtcpp-translation" | "onnx-tts" | "tts-ggml" | "parakeet-transcription" | "ggml-ocr" | "sdcpp-generation" | "audiogen-ggml" | "ggml-vla" | "ggml-classification" | "llm" | "whisper" | "bci" | "nmt" | "parakeet" | "classification" | "vad" | undefined;
     } | undefined;
     dstVocabSrc?: string | {
         src: string;
@@ -833,7 +853,7 @@ export declare const nmtConfigSchema: z.ZodPipe<z.ZodDiscriminatedUnion<[z.ZodOb
         engine?: string | undefined;
         expectedSize?: number | undefined;
         sha256Checksum?: string | undefined;
-        addon?: "embeddings" | "diffusion" | "tts" | "ocr" | "vla" | "llamacpp-completion" | "whispercpp-transcription" | "bci-whispercpp-transcription" | "llamacpp-embedding" | "nmtcpp-translation" | "onnx-tts" | "tts-ggml" | "parakeet-transcription" | "ggml-ocr" | "sdcpp-generation" | "ggml-vla" | "ggml-classification" | "llm" | "whisper" | "bci" | "nmt" | "parakeet" | "classification" | "vad" | undefined;
+        addon?: "embeddings" | "diffusion" | "audiogen" | "tts" | "ocr" | "vla" | "llamacpp-completion" | "whispercpp-transcription" | "bci-whispercpp-transcription" | "llamacpp-embedding" | "nmtcpp-translation" | "onnx-tts" | "tts-ggml" | "parakeet-transcription" | "ggml-ocr" | "sdcpp-generation" | "audiogen-ggml" | "ggml-vla" | "ggml-classification" | "llm" | "whisper" | "bci" | "nmt" | "parakeet" | "classification" | "vad" | undefined;
     } | undefined;
     normalize?: number | undefined;
     pivotModel?: {
@@ -848,7 +868,7 @@ export declare const nmtConfigSchema: z.ZodPipe<z.ZodDiscriminatedUnion<[z.ZodOb
             engine?: string | undefined;
             expectedSize?: number | undefined;
             sha256Checksum?: string | undefined;
-            addon?: "embeddings" | "diffusion" | "tts" | "ocr" | "vla" | "llamacpp-completion" | "whispercpp-transcription" | "bci-whispercpp-transcription" | "llamacpp-embedding" | "nmtcpp-translation" | "onnx-tts" | "tts-ggml" | "parakeet-transcription" | "ggml-ocr" | "sdcpp-generation" | "ggml-vla" | "ggml-classification" | "llm" | "whisper" | "bci" | "nmt" | "parakeet" | "classification" | "vad" | undefined;
+            addon?: "embeddings" | "diffusion" | "audiogen" | "tts" | "ocr" | "vla" | "llamacpp-completion" | "whispercpp-transcription" | "bci-whispercpp-transcription" | "llamacpp-embedding" | "nmtcpp-translation" | "onnx-tts" | "tts-ggml" | "parakeet-transcription" | "ggml-ocr" | "sdcpp-generation" | "audiogen-ggml" | "ggml-vla" | "ggml-classification" | "llm" | "whisper" | "bci" | "nmt" | "parakeet" | "classification" | "vad" | undefined;
         };
         mode?: "full" | undefined;
         beamsize?: number | undefined;
@@ -870,7 +890,7 @@ export declare const nmtConfigSchema: z.ZodPipe<z.ZodDiscriminatedUnion<[z.ZodOb
             engine?: string | undefined;
             expectedSize?: number | undefined;
             sha256Checksum?: string | undefined;
-            addon?: "embeddings" | "diffusion" | "tts" | "ocr" | "vla" | "llamacpp-completion" | "whispercpp-transcription" | "bci-whispercpp-transcription" | "llamacpp-embedding" | "nmtcpp-translation" | "onnx-tts" | "tts-ggml" | "parakeet-transcription" | "ggml-ocr" | "sdcpp-generation" | "ggml-vla" | "ggml-classification" | "llm" | "whisper" | "bci" | "nmt" | "parakeet" | "classification" | "vad" | undefined;
+            addon?: "embeddings" | "diffusion" | "audiogen" | "tts" | "ocr" | "vla" | "llamacpp-completion" | "whispercpp-transcription" | "bci-whispercpp-transcription" | "llamacpp-embedding" | "nmtcpp-translation" | "onnx-tts" | "tts-ggml" | "parakeet-transcription" | "ggml-ocr" | "sdcpp-generation" | "audiogen-ggml" | "ggml-vla" | "ggml-classification" | "llm" | "whisper" | "bci" | "nmt" | "parakeet" | "classification" | "vad" | undefined;
         } | undefined;
         dstVocabSrc?: string | {
             src: string;
@@ -883,7 +903,7 @@ export declare const nmtConfigSchema: z.ZodPipe<z.ZodDiscriminatedUnion<[z.ZodOb
             engine?: string | undefined;
             expectedSize?: number | undefined;
             sha256Checksum?: string | undefined;
-            addon?: "embeddings" | "diffusion" | "tts" | "ocr" | "vla" | "llamacpp-completion" | "whispercpp-transcription" | "bci-whispercpp-transcription" | "llamacpp-embedding" | "nmtcpp-translation" | "onnx-tts" | "tts-ggml" | "parakeet-transcription" | "ggml-ocr" | "sdcpp-generation" | "ggml-vla" | "ggml-classification" | "llm" | "whisper" | "bci" | "nmt" | "parakeet" | "classification" | "vad" | undefined;
+            addon?: "embeddings" | "diffusion" | "audiogen" | "tts" | "ocr" | "vla" | "llamacpp-completion" | "whispercpp-transcription" | "bci-whispercpp-transcription" | "llamacpp-embedding" | "nmtcpp-translation" | "onnx-tts" | "tts-ggml" | "parakeet-transcription" | "ggml-ocr" | "sdcpp-generation" | "audiogen-ggml" | "ggml-vla" | "ggml-classification" | "llm" | "whisper" | "bci" | "nmt" | "parakeet" | "classification" | "vad" | undefined;
         } | undefined;
         normalize?: number | undefined;
     } | undefined;
@@ -924,7 +944,7 @@ export declare const nmtConfigSchema: z.ZodPipe<z.ZodDiscriminatedUnion<[z.ZodOb
         engine?: string | undefined;
         expectedSize?: number | undefined;
         sha256Checksum?: string | undefined;
-        addon?: "embeddings" | "diffusion" | "tts" | "ocr" | "vla" | "llamacpp-completion" | "whispercpp-transcription" | "bci-whispercpp-transcription" | "llamacpp-embedding" | "nmtcpp-translation" | "onnx-tts" | "tts-ggml" | "parakeet-transcription" | "ggml-ocr" | "sdcpp-generation" | "ggml-vla" | "ggml-classification" | "llm" | "whisper" | "bci" | "nmt" | "parakeet" | "classification" | "vad" | undefined;
+        addon?: "embeddings" | "diffusion" | "audiogen" | "tts" | "ocr" | "vla" | "llamacpp-completion" | "whispercpp-transcription" | "bci-whispercpp-transcription" | "llamacpp-embedding" | "nmtcpp-translation" | "onnx-tts" | "tts-ggml" | "parakeet-transcription" | "ggml-ocr" | "sdcpp-generation" | "audiogen-ggml" | "ggml-vla" | "ggml-classification" | "llm" | "whisper" | "bci" | "nmt" | "parakeet" | "classification" | "vad" | undefined;
     } | undefined;
     dstVocabSrc?: string | {
         src: string;
@@ -937,7 +957,7 @@ export declare const nmtConfigSchema: z.ZodPipe<z.ZodDiscriminatedUnion<[z.ZodOb
         engine?: string | undefined;
         expectedSize?: number | undefined;
         sha256Checksum?: string | undefined;
-        addon?: "embeddings" | "diffusion" | "tts" | "ocr" | "vla" | "llamacpp-completion" | "whispercpp-transcription" | "bci-whispercpp-transcription" | "llamacpp-embedding" | "nmtcpp-translation" | "onnx-tts" | "tts-ggml" | "parakeet-transcription" | "ggml-ocr" | "sdcpp-generation" | "ggml-vla" | "ggml-classification" | "llm" | "whisper" | "bci" | "nmt" | "parakeet" | "classification" | "vad" | undefined;
+        addon?: "embeddings" | "diffusion" | "audiogen" | "tts" | "ocr" | "vla" | "llamacpp-completion" | "whispercpp-transcription" | "bci-whispercpp-transcription" | "llamacpp-embedding" | "nmtcpp-translation" | "onnx-tts" | "tts-ggml" | "parakeet-transcription" | "ggml-ocr" | "sdcpp-generation" | "audiogen-ggml" | "ggml-vla" | "ggml-classification" | "llm" | "whisper" | "bci" | "nmt" | "parakeet" | "classification" | "vad" | undefined;
     } | undefined;
     normalize?: number | undefined;
     pivotModel?: {
@@ -952,7 +972,7 @@ export declare const nmtConfigSchema: z.ZodPipe<z.ZodDiscriminatedUnion<[z.ZodOb
             engine?: string | undefined;
             expectedSize?: number | undefined;
             sha256Checksum?: string | undefined;
-            addon?: "embeddings" | "diffusion" | "tts" | "ocr" | "vla" | "llamacpp-completion" | "whispercpp-transcription" | "bci-whispercpp-transcription" | "llamacpp-embedding" | "nmtcpp-translation" | "onnx-tts" | "tts-ggml" | "parakeet-transcription" | "ggml-ocr" | "sdcpp-generation" | "ggml-vla" | "ggml-classification" | "llm" | "whisper" | "bci" | "nmt" | "parakeet" | "classification" | "vad" | undefined;
+            addon?: "embeddings" | "diffusion" | "audiogen" | "tts" | "ocr" | "vla" | "llamacpp-completion" | "whispercpp-transcription" | "bci-whispercpp-transcription" | "llamacpp-embedding" | "nmtcpp-translation" | "onnx-tts" | "tts-ggml" | "parakeet-transcription" | "ggml-ocr" | "sdcpp-generation" | "audiogen-ggml" | "ggml-vla" | "ggml-classification" | "llm" | "whisper" | "bci" | "nmt" | "parakeet" | "classification" | "vad" | undefined;
         };
         mode?: "full" | undefined;
         beamsize?: number | undefined;
@@ -974,7 +994,7 @@ export declare const nmtConfigSchema: z.ZodPipe<z.ZodDiscriminatedUnion<[z.ZodOb
             engine?: string | undefined;
             expectedSize?: number | undefined;
             sha256Checksum?: string | undefined;
-            addon?: "embeddings" | "diffusion" | "tts" | "ocr" | "vla" | "llamacpp-completion" | "whispercpp-transcription" | "bci-whispercpp-transcription" | "llamacpp-embedding" | "nmtcpp-translation" | "onnx-tts" | "tts-ggml" | "parakeet-transcription" | "ggml-ocr" | "sdcpp-generation" | "ggml-vla" | "ggml-classification" | "llm" | "whisper" | "bci" | "nmt" | "parakeet" | "classification" | "vad" | undefined;
+            addon?: "embeddings" | "diffusion" | "audiogen" | "tts" | "ocr" | "vla" | "llamacpp-completion" | "whispercpp-transcription" | "bci-whispercpp-transcription" | "llamacpp-embedding" | "nmtcpp-translation" | "onnx-tts" | "tts-ggml" | "parakeet-transcription" | "ggml-ocr" | "sdcpp-generation" | "audiogen-ggml" | "ggml-vla" | "ggml-classification" | "llm" | "whisper" | "bci" | "nmt" | "parakeet" | "classification" | "vad" | undefined;
         } | undefined;
         dstVocabSrc?: string | {
             src: string;
@@ -987,7 +1007,7 @@ export declare const nmtConfigSchema: z.ZodPipe<z.ZodDiscriminatedUnion<[z.ZodOb
             engine?: string | undefined;
             expectedSize?: number | undefined;
             sha256Checksum?: string | undefined;
-            addon?: "embeddings" | "diffusion" | "tts" | "ocr" | "vla" | "llamacpp-completion" | "whispercpp-transcription" | "bci-whispercpp-transcription" | "llamacpp-embedding" | "nmtcpp-translation" | "onnx-tts" | "tts-ggml" | "parakeet-transcription" | "ggml-ocr" | "sdcpp-generation" | "ggml-vla" | "ggml-classification" | "llm" | "whisper" | "bci" | "nmt" | "parakeet" | "classification" | "vad" | undefined;
+            addon?: "embeddings" | "diffusion" | "audiogen" | "tts" | "ocr" | "vla" | "llamacpp-completion" | "whispercpp-transcription" | "bci-whispercpp-transcription" | "llamacpp-embedding" | "nmtcpp-translation" | "onnx-tts" | "tts-ggml" | "parakeet-transcription" | "ggml-ocr" | "sdcpp-generation" | "audiogen-ggml" | "ggml-vla" | "ggml-classification" | "llm" | "whisper" | "bci" | "nmt" | "parakeet" | "classification" | "vad" | undefined;
         } | undefined;
         normalize?: number | undefined;
     } | undefined;

package/dist/schemas/vla.d.ts

@@ -19,6 +19,11 @@ export declare const vlaHparamsSchema: z.ZodObject<{
         discrete: "discrete";
         continuous: "continuous";
     }>>;
+    imageInputMode: z.ZodOptional<z.ZodEnum<{
+        pixels: "pixels";
+        patches: "patches";
+    }>>;
+    imagePatchElems: z.ZodOptional<z.ZodNumber>;
 }, z.core.$strip>;
 export type VlaHparams = z.infer<typeof vlaHparamsSchema>;
 export declare const vlaStatsSchema: z.ZodObject<{
@@ -78,6 +83,11 @@ export declare const vlaHparamsResponseSchema: z.ZodObject<{
             discrete: "discrete";
             continuous: "continuous";
         }>>;
+        imageInputMode: z.ZodOptional<z.ZodEnum<{
+            pixels: "pixels";
+            patches: "patches";
+        }>>;
+        imagePatchElems: z.ZodOptional<z.ZodNumber>;
     }, z.core.$strip>;
     backendName: z.ZodNullable<z.ZodString>;
 }, z.core.$strip>;