@qvac/sdk 0.13.5 → 0.14.1

package/dist/examples/translation/translation-llm-context.js

@@ -2,11 +2,15 @@ import { translate, loadModel, unloadModel, SALAMANDRATA_2B_INST_Q8, } from "@qv
 try {
     const modelId = await loadModel({
         modelSrc: SALAMANDRATA_2B_INST_Q8,
-        onProgress: (progress) => {
-            console.log(progress);
+        onProgress: (p) => {
+            const mb = (n) => (n / 1e6).toFixed(1);
+            const line = `▸ Downloading ${p.percentage.toFixed(0)}% (${mb(p.downloaded)}/${mb(p.total)} MB)`;
+            process.stderr.write(process.stderr.isTTY ? `\r${line}` : `${line}\n`);
+            if (p.percentage >= 100)
+                process.stderr.write("\n");
         },
     });
-    console.log(`✅ Model loaded: ${modelId}`);
+    console.log(`▸ Model loaded: ${modelId}`);
     // With explicit source language
     const engText = "bank";
     const resultExplicit = translate({
@@ -23,6 +27,6 @@ try {
     await unloadModel({ modelId, clearStorage: false });
 }
 catch (error) {
-    console.error("❌ Error:", error);
+    console.error("✖", error);
     process.exit(1);
 }

package/dist/examples/translation/translation-llm.js

@@ -2,8 +2,12 @@ import { translate, loadModel, unloadModel, SALAMANDRATA_2B_INST_Q4, } from "@qv
 try {
     const modelId = await loadModel({
         modelSrc: SALAMANDRATA_2B_INST_Q4,
-        onProgress: (progress) => {
-            console.log(progress);
+        onProgress: (p) => {
+            const mb = (n) => (n / 1e6).toFixed(1);
+            const line = `▸ Downloading ${p.percentage.toFixed(0)}% (${mb(p.downloaded)}/${mb(p.total)} MB)`;
+            process.stderr.write(process.stderr.isTTY ? `\r${line}` : `${line}\n`);
+            if (p.percentage >= 100)
+                process.stderr.write("\n");
         },
     });
     // With explicit source language
@@ -32,6 +36,6 @@ try {
     await unloadModel({ modelId, clearStorage: false });
 }
 catch (error) {
-    console.error("❌ Error:", error);
+    console.error("✖", error);
     process.exit(1);
 }

package/dist/examples/translation/translation-stream.js

@@ -8,9 +8,9 @@ try {
             to: "es",
         },
     });
-    console.log(`✅ Model loaded: ${modelId}`);
+    console.log(`▸ Model loaded: ${modelId}`);
     const text = "Hello, how are you today? I hope you are having a wonderful day!";
-    console.log("\n--- Streaming Translation ---");
+    console.log("▸ Streaming Translation");
     const streamResult = translate({
         modelId,
         text,
@@ -24,11 +24,11 @@ try {
     console.log();
     const stats = await streamResult.stats;
     if (stats) {
-        console.log(`Processing stats:`, stats);
+        console.log(`▸ Processing stats:`, stats);
     }
     await unloadModel({ modelId, clearStorage: false });
 }
 catch (error) {
-    console.error("❌ Error:", error);
+    console.error("✖", error);
     process.exit(1);
 }

package/dist/examples/tts/chatterbox.js

@@ -12,14 +12,22 @@ try {
             ttsEngine: "chatterbox",
             language: "en",
             s3genModelSrc: TTS_S3GEN_EN_CHATTERBOX.src,
+            streamChunkTokens: 25,
+            streamFirstChunkTokens: 10,
+            cfmSteps: 1,
+            threads: 8,
             ...(referenceAudioSrc ? { referenceAudioSrc } : {}),
         },
-        onProgress: (progress) => {
-            console.log(progress);
+        onProgress: (p) => {
+            const mb = (n) => (n / 1e6).toFixed(1);
+            const line = `▸ Downloading ${p.percentage.toFixed(0)}% (${mb(p.downloaded)}/${mb(p.total)} MB)`;
+            process.stderr.write(process.stderr.isTTY ? `\r${line}` : `${line}\n`);
+            if (p.percentage >= 100)
+                process.stderr.write("\n");
         },
     });
-    console.log(`Model loaded: ${modelId}`);
-    console.log("🎵 Testing Text-to-Speech...");
+    console.log(`▸ Model loaded: ${modelId}`);
+    console.log("▸ Testing Text-to-Speech...");
     const result = textToSpeech({
         modelId,
         text: `QVAC SDK is the canonical entry point to QVAC. Written in TypeScript, it provides all QVAC capabilities through a unified interface while also abstracting away the complexity of running your application in a JS environment other than Bare. Supported JS environments include Bare, Node.js, Expo and Bun.`,
@@ -27,23 +35,23 @@ try {
         stream: false,
     });
     const audioBuffer = await result.buffer;
-    console.log(`TTS complete. Total bytes: ${audioBuffer.length}`);
-    console.log("💾 Saving audio to file...");
+    console.log(`▸ TTS complete. Total bytes: ${audioBuffer.length}`);
+    console.log("▸ Saving audio to file...");
     createWav(audioBuffer, CHATTERBOX_SAMPLE_RATE, "tts-output.wav");
-    console.log("✅ Audio saved to tts-output.wav");
-    console.log("🔊 Playing audio...");
+    console.log("▸ Audio saved to tts-output.wav");
+    console.log("▸ Playing audio...");
     const audioData = int16ArrayToBuffer(audioBuffer);
     const wavBuffer = Buffer.concat([
         createWavHeader(audioData.length, CHATTERBOX_SAMPLE_RATE),
         audioData,
     ]);
     playAudio(wavBuffer);
-    console.log("✅ Audio playback complete");
+    console.log("▸ Audio playback complete");
     await unloadModel({ modelId });
-    console.log("Model unloaded");
+    console.log("▸ Model unloaded");
     process.exit(0);
 }
 catch (error) {
-    console.error("❌ Error:", error);
+    console.error("✖", error);
     process.exit(1);
 }

package/dist/examples/tts/llm-to-tts-streaming.js

@@ -12,7 +12,7 @@
  * Usage:
  *   bun run examples/tts/llm-to-tts-streaming.ts
  */
-import { completion, loadModel, textToSpeechStream, unloadModel, LLAMA_3_2_1B_INST_Q4_0, TTS_EN_SUPERTONIC_Q8_0, } from "@qvac/sdk";
+import { completion, loadModel, textToSpeechStream, unloadModel, LLAMA_3_2_1B_INST_Q4_0, TTS_MULTILINGUAL_SUPERTONIC3_Q8_0, } from "@qvac/sdk";
 import { createWav, playPcmInt16Chunk } from "./utils.js";
 const SUPERTONIC_SAMPLE_RATE = 44100;
 /**
@@ -57,18 +57,24 @@ function appendPcmSamples(target, chunk) {
     }
 }
 try {
-    console.log(`Loading LLM from registry: ${LLAMA_3_2_1B_INST_Q4_0.name}`);
+    console.log(`▸ Loading LLM from registry: ${LLAMA_3_2_1B_INST_Q4_0.name}`);
     const llmModelId = await loadModel({
         modelSrc: LLAMA_3_2_1B_INST_Q4_0,
         modelConfig: {
             ctx_size: 2048,
         },
-        onProgress: (progress) => console.log(`LLM load: ${progress.percentage.toFixed(1)}%`),
+        onProgress: (p) => {
+            const mb = (n) => (n / 1e6).toFixed(1);
+            const line = `▸ Downloading ${p.percentage.toFixed(0)}% (${mb(p.downloaded)}/${mb(p.total)} MB)`;
+            process.stderr.write(process.stderr.isTTY ? `\r${line}` : `${line}\n`);
+            if (p.percentage >= 100)
+                process.stderr.write("\n");
+        },
     });
-    console.log(`LLM ready: ${llmModelId}`);
-    console.log("Loading Supertonic TTS (registry)…");
+    console.log(`▸ LLM ready: ${llmModelId}`);
+    console.log("▸ Loading Supertonic TTS (registry)…");
     const ttsModelId = await loadModel({
-        modelSrc: TTS_EN_SUPERTONIC_Q8_0,
+        modelSrc: TTS_MULTILINGUAL_SUPERTONIC3_Q8_0,
         modelConfig: {
             ttsEngine: "supertonic",
             language: "en",
@@ -76,11 +82,18 @@ try {
             ttsSpeed: 1.05,
             ttsNumInferenceSteps: 10,
         },
-        onProgress: (progress) => console.log(`TTS load: ${progress.percentage.toFixed(1)}%`),
+        onProgress: (p) => {
+            const mb = (n) => (n / 1e6).toFixed(1);
+            const line = `▸ Downloading ${p.percentage.toFixed(0)}% (${mb(p.downloaded)}/${mb(p.total)} MB)`;
+            process.stderr.write(process.stderr.isTTY ? `\r${line}` : `${line}\n`);
+            if (p.percentage >= 100)
+                process.stderr.write("\n");
+        },
     });
-    console.log(`TTS ready: ${ttsModelId}`);
+    console.log(`▸ TTS ready: ${ttsModelId}`);
     const prompt = "What is a constellation?";
-    console.log(`\nUser: ${prompt}\nAssistant (streaming):`);
+    console.log(`▸ User: ${prompt}`);
+    console.log("▸ Assistant (streaming):");
     const result = completion({
         modelId: llmModelId,
         history: [{ role: "user", content: prompt }],
@@ -102,7 +115,7 @@ try {
                 const preview = typeof m.sentenceChunk === "string"
                     ? m.sentenceChunk.replace(/\s+/g, " ").trim().slice(0, 72)
                     : "";
-                console.log(`\n[TTS phrase ${phraseIndex}] ${m.buffer.length} samples${preview ? ` — "${preview}${preview.length >= 72 ? "..." : ""}"` : ""}`);
+                console.log(`\n▸ [TTS phrase ${phraseIndex}] ${m.buffer.length} samples${preview ? ` — "${preview}${preview.length >= 72 ? "..." : ""}"` : ""}`);
                 await playPcmInt16Chunk(m.buffer, SUPERTONIC_SAMPLE_RATE);
             }
         }
@@ -118,17 +131,17 @@ try {
     await drainPcm;
     const stats = await result.stats;
     if (stats) {
-        console.log("\nLLM stats:", stats);
+        console.log("\n▸ LLM stats:", stats);
     }
     const outWav = "llm-to-tts-streaming-output.wav";
-    console.log(`\nWriting ${combinedPcm.length} samples to ${outWav} (full utterance; phrases were already played above).`);
+    console.log(`\n▸ Writing ${combinedPcm.length} samples to ${outWav} (full utterance; phrases were already played above).`);
     createWav(combinedPcm, SUPERTONIC_SAMPLE_RATE, outWav);
     await unloadModel({ modelId: llmModelId, clearStorage: false });
     await unloadModel({ modelId: ttsModelId, clearStorage: false });
-    console.log("Models unloaded.");
+    console.log("▸ Models unloaded.");
     process.exit(0);
 }
 catch (error) {
-    console.error("❌ Error:", error);
+    console.error("✖", error);
     process.exit(1);
 }

package/dist/examples/tts/supertonic-multilingual.js

@@ -1,10 +1,10 @@
-import { loadModel, textToSpeech, unloadModel, TTS_MULTILINGUAL_SUPERTONIC2_Q8_0, } from "@qvac/sdk";
+import { loadModel, textToSpeech, unloadModel, TTS_MULTILINGUAL_SUPERTONIC3_Q8_0, } from "@qvac/sdk";
 import { createWav, playAudio, int16ArrayToBuffer, createWavHeader, } from "./utils.js";
-// Supertonic 2 TTS (GGML): multilingual synthesis (en/ko/es/pt/fr).
+// Supertonic 3 TTS (GGML): multilingual synthesis across 31 languages.
 const SUPERTONIC_SAMPLE_RATE = 44100;
 try {
     const modelId = await loadModel({
-        modelSrc: TTS_MULTILINGUAL_SUPERTONIC2_Q8_0,
+        modelSrc: TTS_MULTILINGUAL_SUPERTONIC3_Q8_0,
         modelConfig: {
             ttsEngine: "supertonic",
             language: "es",
@@ -12,12 +12,16 @@ try {
             ttsSpeed: 1.05,
             ttsNumInferenceSteps: 5,
         },
-        onProgress: (progress) => {
-            console.log(progress);
+        onProgress: (p) => {
+            const mb = (n) => (n / 1e6).toFixed(1);
+            const line = `▸ Downloading ${p.percentage.toFixed(0)}% (${mb(p.downloaded)}/${mb(p.total)} MB)`;
+            process.stderr.write(process.stderr.isTTY ? `\r${line}` : `${line}\n`);
+            if (p.percentage >= 100)
+                process.stderr.write("\n");
         },
     });
-    console.log(`Model loaded: ${modelId}`);
-    console.log("🎵 Testing Text-to-Speech...");
+    console.log(`▸ Model loaded: ${modelId}`);
+    console.log("▸ Testing Text-to-Speech...");
     const result = textToSpeech({
         modelId,
         text: `Hola mundo. Esta es una demostración de síntesis de voz con Supertonic en español.`,
@@ -25,23 +29,23 @@ try {
         stream: false,
     });
     const audioBuffer = await result.buffer;
-    console.log(`TTS complete. Total samples: ${audioBuffer.length}`);
-    console.log("💾 Saving audio to file...");
+    console.log(`▸ TTS complete. Total samples: ${audioBuffer.length}`);
+    console.log("▸ Saving audio to file...");
     createWav(audioBuffer, SUPERTONIC_SAMPLE_RATE, "supertonic-multilingual-output.wav");
-    console.log("✅ Audio saved to supertonic-multilingual-output.wav");
-    console.log("🔊 Playing audio...");
+    console.log("▸ Audio saved to supertonic-multilingual-output.wav");
+    console.log("▸ Playing audio...");
     const audioData = int16ArrayToBuffer(audioBuffer);
     const wavBuffer = Buffer.concat([
         createWavHeader(audioData.length, SUPERTONIC_SAMPLE_RATE),
         audioData,
     ]);
     playAudio(wavBuffer);
-    console.log("✅ Audio playback complete");
+    console.log("▸ Audio playback complete");
     await unloadModel({ modelId });
-    console.log("Model unloaded");
+    console.log("▸ Model unloaded");
     process.exit(0);
 }
 catch (error) {
-    console.error("❌ Error:", error);
+    console.error("✖", error);
     process.exit(1);
 }

package/dist/examples/tts/supertonic.js

@@ -1,24 +1,28 @@
-import { loadModel, textToSpeech, unloadModel, TTS_EN_SUPERTONIC_Q8_0, } from "@qvac/sdk";
+import { loadModel, textToSpeech, unloadModel, TTS_MULTILINGUAL_SUPERTONIC3_Q8_0, } from "@qvac/sdk";
 import { createWav, playAudio, int16ArrayToBuffer, createWavHeader, } from "./utils.js";
-// Supertonic TTS (GGML): fast English synthesis with baked-in voices.
+// Supertonic 3 TTS (GGML): fast multilingual synthesis with baked-in voices.
 // Uses registry model constants — downloads automatically from QVAC Registry.
 const SUPERTONIC_SAMPLE_RATE = 44100;
 try {
     const modelId = await loadModel({
-        modelSrc: TTS_EN_SUPERTONIC_Q8_0,
+        modelSrc: TTS_MULTILINGUAL_SUPERTONIC3_Q8_0,
         modelConfig: {
             ttsEngine: "supertonic",
             language: "en",
             voice: "F1",
             ttsSpeed: 1.05,
-            ttsNumInferenceSteps: 5,
+            ttsNumInferenceSteps: 5
         },
-        onProgress: (progress) => {
-            console.log(progress);
+        onProgress: (p) => {
+            const mb = (n) => (n / 1e6).toFixed(1);
+            const line = `▸ Downloading ${p.percentage.toFixed(0)}% (${mb(p.downloaded)}/${mb(p.total)} MB)`;
+            process.stderr.write(process.stderr.isTTY ? `\r${line}` : `${line}\n`);
+            if (p.percentage >= 100)
+                process.stderr.write("\n");
         },
     });
-    console.log(`Model loaded: ${modelId}`);
-    console.log("🎵 Testing Text-to-Speech...");
+    console.log(`▸ Model loaded: ${modelId}`);
+    console.log("▸ Testing Text-to-Speech...");
     const result = textToSpeech({
         modelId,
         text: `QVAC SDK is the canonical entry point to QVAC. Written in TypeScript, it provides all QVAC capabilities through a unified interface while also abstracting away the complexity of running your application in a JS environment other than Bare. Supported JS environments include Bare, Node.js, Expo and Bun.`,
@@ -26,23 +30,23 @@ try {
         stream: false,
     });
     const audioBuffer = await result.buffer;
-    console.log(`TTS complete. Total samples: ${audioBuffer.length}`);
-    console.log("💾 Saving audio to file...");
+    console.log(`▸ TTS complete. Total samples: ${audioBuffer.length}`);
+    console.log("▸ Saving audio to file...");
     createWav(audioBuffer, SUPERTONIC_SAMPLE_RATE, "supertonic-output.wav");
-    console.log("✅ Audio saved to supertonic-output.wav");
-    console.log("🔊 Playing audio...");
+    console.log("▸ Audio saved to supertonic-output.wav");
+    console.log("▸ Playing audio...");
     const audioData = int16ArrayToBuffer(audioBuffer);
     const wavBuffer = Buffer.concat([
         createWavHeader(audioData.length, SUPERTONIC_SAMPLE_RATE),
         audioData,
     ]);
     playAudio(wavBuffer);
-    console.log("✅ Audio playback complete");
+    console.log("▸ Audio playback complete");
     await unloadModel({ modelId });
-    console.log("Model unloaded");
+    console.log("▸ Model unloaded");
     process.exit(0);
 }
 catch (error) {
-    console.error("❌ Error:", error);
+    console.error("✖", error);
     process.exit(1);
 }

package/dist/examples/tts/utils.js

@@ -44,7 +44,7 @@ export function createWav(audioBuffer, sampleRate, filename) {
     const wavHeader = createWavHeader(audioData.length, sampleRate);
     const wavFile = Buffer.concat([wavHeader, audioData]);
     writeFileSync(filename, wavFile);
-    console.log(`WAV file saved as: ${filename}`);
+    console.log(`▸ WAV file saved as: ${filename}`);
 }
 /**
  * Play a WAV buffer by streaming it into ffplay over stdin.

package/dist/examples/vla-pi05.js

@@ -22,21 +22,25 @@ import { close, loadModel, PI05_BASE_Q_AGGRESSIVE, unloadModel, vla, vlaHparams,
 const modelSrcOverride = process.argv[2];
 const modelSrc = modelSrcOverride ?? PI05_BASE_Q_AGGRESSIVE;
 try {
-    console.log("Loading π₀.₅ (pi05) model...");
+    console.log("▸ Loading π₀.₅ (pi05) model...");
     const modelId = await loadModel({
         modelSrc,
         modelType: "ggml-vla",
         modelConfig: { backend: "cpu" },
-        onProgress: (p) => typeof modelSrc === "string"
-            ? undefined
-            : process.stdout.write(`\rDownloading: ${p.percentage.toFixed(1)}%`),
+        onProgress: (p) => {
+            const mb = (n) => (n / 1e6).toFixed(1);
+            const line = `▸ Downloading ${p.percentage.toFixed(0)}% (${mb(p.downloaded)}/${mb(p.total)} MB)`;
+            process.stderr.write(process.stderr.isTTY ? `\r${line}` : `${line}\n`);
+            if (p.percentage >= 100)
+                process.stderr.write("\n");
+        },
     });
     if (typeof modelSrc !== "string")
-        process.stdout.write("\n");
-    console.log(`Model loaded: ${modelId}`);
+        process.stderr.write("\n");
+    console.log(`▸ Model loaded: ${modelId}`);
     const { hparams, backendName } = await vlaHparams({ modelId });
-    console.log(`Backend: ${backendName ?? "(unknown)"}`);
-    console.log("Hparams:", hparams);
+    console.log(`▸ Backend: ${backendName ?? "(unknown)"}`);
+    console.log("▸ Hparams:", hparams);
     // Build synthetic inputs sized to the model's expectations. A real
     // consumer would: read camera frames, tokenize the instruction with the
     // model's tokenizer, and (for π₀.₅) inline the robot state into the prompt.
@@ -54,7 +58,7 @@ try {
     // into the prompt), so pass an empty Float32Array. π₀.₅ requires `noise`.
     const state = new Float32Array(0);
     const noise = new Float32Array(hparams.chunkSize * hparams.maxActionDim);
-    console.log("Running VLA inference...");
+    console.log("▸ Running VLA inference...");
     const { actions, actionDim, chunkSize, stats } = await vla({
         modelId,
         images,
@@ -65,20 +69,20 @@ try {
         mask,
         noise,
     });
-    console.log(`Got ${chunkSize} action steps of dim ${actionDim}.`);
-    console.log("First step:", Array.from(actions.subarray(0, actionDim)));
+    console.log(`▸ Got ${chunkSize} action steps of dim ${actionDim}.`);
+    console.log(Array.from(actions.subarray(0, actionDim)));
     if (stats) {
-        console.log(`Timing: vision=${stats.vision_ms?.toFixed(0)}ms ` +
+        console.log(`▸ Timing: vision=${stats.vision_ms?.toFixed(0)}ms ` +
             `prefill=${stats.prefill_total_ms?.toFixed(0)}ms ` +
             `ode=${stats.ode_ms?.toFixed(0)}ms ` +
             `total=${stats.total_ms?.toFixed(0)}ms`);
     }
     await unloadModel({ modelId, clearStorage: false });
-    console.log("Model unloaded.");
+    console.log("▸ Model unloaded.");
     process.exit(0);
 }
 catch (error) {
-    console.error("π₀.₅ example failed:", error);
+    console.error("✖", error);
     await close();
     process.exit(1);
 }

package/dist/examples/vla-smolvla.js

@@ -16,21 +16,25 @@ import { close, loadModel, SMOLVLA_LIBERO_VISION_Q8, unloadModel, vla, vlaHparam
 const modelSrcOverride = process.argv[2];
 const modelSrc = modelSrcOverride ?? SMOLVLA_LIBERO_VISION_Q8;
 try {
-    console.log("Loading SmolVLA model...");
+    console.log("▸ Loading SmolVLA model...");
     const modelId = await loadModel({
         modelSrc,
         modelType: "ggml-vla",
         modelConfig: { backend: "cpu" },
-        onProgress: (p) => typeof modelSrc === "string"
-            ? undefined
-            : process.stdout.write(`\rDownloading: ${p.percentage.toFixed(1)}%`),
+        onProgress: (p) => {
+            const mb = (n) => (n / 1e6).toFixed(1);
+            const line = `▸ Downloading ${p.percentage.toFixed(0)}% (${mb(p.downloaded)}/${mb(p.total)} MB)`;
+            process.stderr.write(process.stderr.isTTY ? `\r${line}` : `${line}\n`);
+            if (p.percentage >= 100)
+                process.stderr.write("\n");
+        },
     });
     if (typeof modelSrc !== "string")
-        process.stdout.write("\n");
-    console.log(`Model loaded: ${modelId}`);
+        process.stderr.write("\n");
+    console.log(`▸ Model loaded: ${modelId}`);
     const { hparams, backendName } = await vlaHparams({ modelId });
-    console.log(`Backend: ${backendName ?? "(unknown)"}`);
-    console.log("Hparams:", hparams);
+    console.log(`▸ Backend: ${backendName ?? "(unknown)"}`);
+    console.log("▸ Hparams:", hparams);
     // Build synthetic inputs sized to the model's expectations. A real
     // consumer would: read camera frames, tokenize the instruction with the
     // SmolVLM2 tokenizer, and read the robot's current end-effector pose.
@@ -45,7 +49,7 @@ try {
     mask[0] = 1;
     const state = vlaPadState([0, 0, 0, 0, 0, 0], hparams.maxStateDim);
     const noise = new Float32Array(hparams.chunkSize * hparams.maxActionDim);
-    console.log("Running VLA inference...");
+    console.log("▸ Running VLA inference...");
     const { actions, actionDim, chunkSize, stats } = await vla({
         modelId,
         images: [front, wrist],
@@ -56,20 +60,20 @@ try {
         mask,
         noise,
     });
-    console.log(`Got ${chunkSize} action steps of dim ${actionDim}.`);
-    console.log("First step:", Array.from(actions.subarray(0, actionDim)));
+    console.log(`▸ Got ${chunkSize} action steps of dim ${actionDim}.`);
+    console.log(Array.from(actions.subarray(0, actionDim)));
     if (stats) {
-        console.log(`Timing: vision=${stats.vision_ms?.toFixed(0)}ms ` +
+        console.log(`▸ Timing: vision=${stats.vision_ms?.toFixed(0)}ms ` +
             `smollm2=${stats.smollm2_total_ms?.toFixed(0)}ms ` +
             `ode=${stats.ode_ms?.toFixed(0)}ms ` +
             `total=${stats.total_ms?.toFixed(0)}ms`);
     }
     await unloadModel({ modelId, clearStorage: false });
-    console.log("Model unloaded.");
+    console.log("▸ Model unloaded.");
     process.exit(0);
 }
 catch (error) {
-    console.error("VLA example failed:", error);
+    console.error("✖", error);
     await close();
     process.exit(1);
 }

package/dist/examples/voice-assistant/voice-assistant.js

@@ -69,11 +69,11 @@ function sleep(ms) {
 for (const tool of ["ffmpeg", "ffplay"]) {
     const r = spawnSync(tool, ["-version"], { stdio: "ignore" });
     if (r.error || r.status !== 0) {
-        console.error(`${tool} not found on PATH. Install ffmpeg (ffplay ships with it) and retry.`);
+        console.error(`✖ ${tool} not found on PATH. Install ffmpeg (ffplay ships with it) and retry.`);
         process.exit(1);
     }
 }
-console.log("Loading whisper-tiny + Silero VAD...");
+console.log("▸ Loading whisper-tiny + Silero VAD...");
 const asrModelId = await loadModel({
     modelSrc: WHISPER_TINY,
     modelConfig: {
@@ -89,14 +89,14 @@ const asrModelId = await loadModel({
         vad_params: VAD_PARAMS,
     },
 });
-console.log("Loading Llama 3.2 1B...");
+console.log("▸ Loading Llama 3.2 1B...");
 const llmModelId = await loadModel({
     modelSrc: LLAMA_3_2_1B_INST_Q4_0,
     modelConfig: {
         ctx_size: 4096,
     },
 });
-console.log("Loading Supertonic TTS...");
+console.log("▸ Loading Supertonic TTS...");
 const ttsModelId = await loadModel({
     modelSrc: TTS_EN_SUPERTONIC_Q8_0,
     modelConfig: {
@@ -107,7 +107,7 @@ const ttsModelId = await loadModel({
         ttsNumInferenceSteps: 5,
     },
 });
-console.log("All models loaded.\n");
+console.log("▸ All models loaded.\n");
 const ffmpeg = startMicrophone({
     sampleRate: MIC_SAMPLE_RATE,
     format: "f32le",
@@ -129,7 +129,7 @@ async function cleanup() {
     if (shuttingDown)
         return;
     shuttingDown = true;
-    console.log("\n\nStopping...");
+    console.log("\n\n▸ Stopping...");
     ffmpeg.kill();
     try {
         session.end();
@@ -140,21 +140,21 @@ async function cleanup() {
     await unloadModel({ modelId: ttsModelId }).catch(() => { });
     await unloadModel({ modelId: llmModelId }).catch(() => { });
     await unloadModel({ modelId: asrModelId }).catch(() => { });
-    console.log("Done.");
+    console.log("▸ Done.");
     process.exit(0);
 }
 process.on("SIGINT", () => void cleanup());
 process.on("SIGTERM", () => void cleanup());
-console.log("🎙️  Listening. Speak a question and pause. Ctrl+C to quit.\n");
+console.log("▸ Listening. Speak a question and pause. Ctrl+C to quit.\n");
 for await (const rawText of session) {
     if (!isMeaningfulTranscript(rawText))
         continue;
     const userText = rawText.trim();
-    console.log(`🗣️  You: ${userText}`);
+    console.log(`▸ You: ${userText}`);
     history.push({ role: "user", content: userText });
     isSpeaking = true;
     try {
-        process.stdout.write("🤖 Assistant: ");
+        console.log("▸ Assistant:");
         const llmResult = completion({
             modelId: llmModelId,
             history,
@@ -188,10 +188,10 @@ for await (const rawText of session) {
         }
     }
     catch (turnError) {
-        console.error("\n⚠️  Turn failed:", turnError instanceof Error ? turnError.message : turnError);
+        console.error("\n✖ Turn failed:", turnError instanceof Error ? turnError.message : turnError);
     }
     finally {
         isSpeaking = false;
-        console.log("\n🎙️  Listening...\n");
+        console.log("\n▸ Listening...\n");
     }
 }

package/dist/index.d.ts

@@ -1,5 +1,4 @@
-import "#polyfill-bare-globals";
-export { completion, deleteCache, loadModel, downloadAsset, heartbeat, startQVACProvider, stopQVACProvider, unloadModel, transcribe, transcribeStream, bciTranscribe, bciTranscribeStream, embed, finetune, translate, cancel, ragChunk, ragIngest, ragSaveEmbeddings, ragSearch, ragDeleteEmbeddings, ragReindex, ragListWorkspaces, ragCloseWorkspace, ragDeleteWorkspace, textToSpeech, textToSpeechStream, getModelInfo, getLoadedModelInfo, loggingStream, ocr, invokePlugin, invokePluginStream, diffusion, type DiffusionProgressTick, classify, video, type VideoProgressTick, upscale, modelRegistryList, modelRegistrySearch, modelRegistryGetModel, type ModelRegistrySearchParams, suspend, resume, state, vla, vlaHparams, vlaPreprocessImage, vlaPadState, VLA_DEFAULT_IMAGE_SIZE, type FinetuneHandle, } from "./client/api/index";
+export { completion, deleteCache, loadModel, downloadAsset, heartbeat, startQVACProvider, stopQVACProvider, unloadModel, transcribe, transcribeStream, bciTranscribe, bciTranscribeStream, embed, finetune, translate, cancel, ragChunk, ragIngest, ragSaveEmbeddings, ragSearch, ragDeleteEmbeddings, ragReindex, ragListWorkspaces, ragCloseWorkspace, ragDeleteWorkspace, textToSpeech, textToSpeechStream, getModelInfo, getLoadedModelInfo, loggingStream, subscribeServerLogs, type ServerLogHandler, ocr, invokePlugin, invokePluginStream, diffusion, type DiffusionProgressTick, classify, video, type VideoProgressTick, upscale, modelRegistryList, modelRegistrySearch, modelRegistryGetModel, type ModelRegistrySearchParams, suspend, resume, state, vla, vlaHparams, vlaPreprocessImage, vlaPadState, VLA_DEFAULT_IMAGE_SIZE, type FinetuneHandle, } from "./client/api/index";
 export { close } from "./client/index";
 export { plugins } from "./client/plugins-factory";
 export { type LifecycleState, type ModelProgressUpdate, type LoadModelOptions, type LoadCustomPluginModelOptions, type DownloadAssetOptions, type Tool, type ToolCall, type ToolCallWithCall, type ToolCallError, type ToolCallEvent, type CompletionEvent, type CompletionFinal, type CompletionRun, type CompletionStats, type EmbedStats, VERBOSITY, type Attachment, type TranscribeStreamSession, type TranscribeStreamMetadataSession, type TranscribeStreamConversationSession, type TranscribeStreamEvent, type VadStateEvent, type EndOfTurnEvent, type TranscribeSegment, type BciConfig, type BciTranscribeClientParams, type BciTranscribeStreamClientParams, type BciTranscribeStreamSession, type BciTranscribeStreamMetadataSession, type BciStreamOpts, type NeuralInput, type TextToSpeechStreamSession, type TextToSpeechStreamResponse, type TextToSpeechStreamClientParams, type CompletionParams, type ToolDialect, type RagSearchResult, type RagSaveEmbeddingsResult, type RagReindexResult, type RagEmbeddedDoc, type RagDoc, type RagWorkspaceInfo, type RagCloseWorkspaceParams, type RagDeleteWorkspaceParams, type RagIngestStage, type RagReindexStage, type RagSaveStage, SDK_CLIENT_ERROR_CODES, SDK_SERVER_ERROR_CODES, RAG_ERROR_CODES, type QvacConfig, type ModelInfo, type GetModelInfoParams, type GetLoadedModelInfoParams, type LoadedModelInfo, type LoadedInstance, type CacheFileInfo, toolSchema, TOOLS_MODE, type ToolsMode, type McpClient, type McpClientInput, type OCRClientParams, type OCRTextBlock, type OCROptions, type ClassifyClientParams, type ClassificationResult, type DiffusionClientParams, type DiffusionStreamResponse, type DiffusionStats, type VideoClientParams, type VideoStreamResponse, type VideoStats, type UpscaleClientParams, type UpscaleStreamResponse, type UpscaleStats, type VlaConfig, type VlaClientRunParams, type VlaClientRunResult, type VlaHparams, type VlaStats, definePlugin, defineHandler, defineDuplexHandler, type QvacPlugin, type CreateModelParams, type PluginModelResult, type ModelRegistryEntry, type ModelRegistryEntryAddon, PLUGIN_LLM, PLUGIN_EMBEDDING, PLUGIN_WHISPER, PLUGIN_BCI, PLUGIN_NMT, PLUGIN_TTS, PLUGIN_OCR, PLUGIN_DIFFUSION, PLUGIN_VLA, PLUGIN_CLASSIFICATION, SDK_DEFAULT_PLUGINS, type BuiltinPlugin, type ProfilerMode, type FinetuneValidation, type FinetuneRunParams, type FinetuneGetStateParams, type FinetuneStopParams, type FinetuneParams, type FinetuneStatus, type FinetuneProgress, type FinetuneStats, type FinetuneResult, } from "./schemas/index";
@@ -10,8 +9,8 @@ export { SUPPORTED_AUDIO_FORMATS } from "./constants/audio";
 export { InferenceCancelledError } from "./utils/errors-server";
 export type { InferenceCancelledPartial } from "./utils/errors-server";
 export { ContextOverflowError, RequestIdConflictError, RequestNotFoundError, RequestRejectedByPolicyError, } from "./utils/errors-server";
-export { BareRuntimeBinaryNotFoundError, WorkerCrashedError, WorkerShutdownError, } from "./utils/errors-client";
-export { getLogger, SDK_LOG_ID } from "./logging/index";
+export { BareRuntimeBinaryNotFoundError, WorkerCrashedError, WorkerShutdownError, RequestValidationFailedError, } from "./utils/errors-client";
+export { getLogger, SDK_LOG_ID, SDK_ALL_LOG_ID } from "./logging/index";
 export type { Logger, LogTransport, LoggerOptions } from "./logging/index";
 export { profiler } from "./profiling/index";
 export type { ProfilerRuntimeOptions, ProfilerExport } from "./profiling/index";

package/dist/index.js

@@ -1,6 +1,5 @@
-import "#polyfill-bare-globals";
 // Public API exports only
-export { completion, deleteCache, loadModel, downloadAsset, heartbeat, startQVACProvider, stopQVACProvider, unloadModel, transcribe, transcribeStream, bciTranscribe, bciTranscribeStream, embed, finetune, translate, cancel, ragChunk, ragIngest, ragSaveEmbeddings, ragSearch, ragDeleteEmbeddings, ragReindex, ragListWorkspaces, ragCloseWorkspace, ragDeleteWorkspace, textToSpeech, textToSpeechStream, getModelInfo, getLoadedModelInfo, loggingStream, ocr, invokePlugin, invokePluginStream, diffusion, classify, video, upscale, modelRegistryList, modelRegistrySearch, modelRegistryGetModel, suspend, resume, state, vla, vlaHparams, vlaPreprocessImage, vlaPadState, VLA_DEFAULT_IMAGE_SIZE, } from "./client/api/index.js";
+export { completion, deleteCache, loadModel, downloadAsset, heartbeat, startQVACProvider, stopQVACProvider, unloadModel, transcribe, transcribeStream, bciTranscribe, bciTranscribeStream, embed, finetune, translate, cancel, ragChunk, ragIngest, ragSaveEmbeddings, ragSearch, ragDeleteEmbeddings, ragReindex, ragListWorkspaces, ragCloseWorkspace, ragDeleteWorkspace, textToSpeech, textToSpeechStream, getModelInfo, getLoadedModelInfo, loggingStream, subscribeServerLogs, ocr, invokePlugin, invokePluginStream, diffusion, classify, video, upscale, modelRegistryList, modelRegistrySearch, modelRegistryGetModel, suspend, resume, state, vla, vlaHparams, vlaPreprocessImage, vlaPadState, VLA_DEFAULT_IMAGE_SIZE, } from "./client/api/index.js";
 export { close } from "./client/index.js";
 export { plugins } from "./client/plugins-factory.js";
 export { VERBOSITY, SDK_CLIENT_ERROR_CODES, SDK_SERVER_ERROR_CODES, RAG_ERROR_CODES, toolSchema, TOOLS_MODE, definePlugin, defineHandler, defineDuplexHandler, PLUGIN_LLM, PLUGIN_EMBEDDING, PLUGIN_WHISPER, PLUGIN_BCI, PLUGIN_NMT, PLUGIN_TTS, PLUGIN_OCR, PLUGIN_DIFFUSION, PLUGIN_VLA, PLUGIN_CLASSIFICATION, SDK_DEFAULT_PLUGINS, } from "./schemas/index.js";
@@ -39,8 +38,8 @@ export { ContextOverflowError, RequestIdConflictError, RequestNotFoundError, Req
 // spawn because the platform's `bare-runtime-<platform>-<arch>` package is
 // missing (common under pnpm). Exported so consumers can pattern-match with
 // `instanceof`.
-export { BareRuntimeBinaryNotFoundError, WorkerCrashedError, WorkerShutdownError, } from "./utils/errors-client.js";
+export { BareRuntimeBinaryNotFoundError, WorkerCrashedError, WorkerShutdownError, RequestValidationFailedError, } from "./utils/errors-client.js";
 // Logging exports
-export { getLogger, SDK_LOG_ID } from "./logging/index.js";
+export { getLogger, SDK_LOG_ID, SDK_ALL_LOG_ID } from "./logging/index.js";
 // Profiler exports
 export { profiler } from "./profiling/index.js";