@qvac/sdk 0.14.1 → 0.16.0

package/dist/examples/llamacpp-filesystem.js

@@ -1,9 +1,9 @@
-import { completion, loadModel, unloadModel } from "@qvac/sdk";
+import { completion, loadModel, unloadModel } from '@qvac/sdk';
 // Get GGUF file path from command line arguments
 const ggufPath = process.argv[2];
 if (!ggufPath) {
-    console.error("✖ Please provide the path to a GGUF file as the first argument");
-    console.error("✖ Usage: bun run examples/llamacpp-filesystem.ts <path-to-gguf-file>");
+    console.error('✖ Please provide the path to a GGUF file as the first argument');
+    console.error('✖ Usage: bun run examples/llamacpp-filesystem.ts <path-to-gguf-file>');
     process.exit(1);
 }
 console.log(`▸ Loading GGUF model from: ${ggufPath}`);
@@ -11,36 +11,34 @@ try {
     // Load model from provided file path
     const modelId = await loadModel({
         modelSrc: ggufPath,
-        modelType: "llamacpp-completion",
+        modelType: 'llamacpp-completion',
         modelConfig: {
-            ctx_size: 4096,
+            ctx_size: 4096
         },
         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");
-        },
+                process.stderr.write('\n');
+        }
     });
     console.log(`▸ Model loaded successfully! Model ID: ${modelId}`);
     // Create conversation history
-    const history = [
-        { role: "user", content: "Explain Bitcoin in 3 key points" },
-    ];
-    console.log("\n▸ AI Response:");
-    process.stdout.write(""); // Start response on new line
+    const history = [{ role: 'user', content: 'Explain Bitcoin in 3 key points' }];
+    console.log('\n▸ AI Response:');
+    process.stdout.write(''); // Start response on new line
     // Stream completion
     const result = completion({ modelId, history, stream: true });
     for await (const token of result.tokenStream) {
         process.stdout.write(token);
     }
     const stats = await result.stats;
-    console.log("\n▸ Performance Stats:", stats);
-    console.log("\n\n▸ Completed!");
+    console.log('\n▸ Performance Stats:', stats);
+    console.log('\n\n▸ Completed!');
     await unloadModel({ modelId, clearStorage: false });
 }
 catch (error) {
-    console.error("✖", error);
+    console.error('✖', error);
     process.exit(1);
 }

package/dist/examples/llamacpp-http.js

@@ -1,42 +1,40 @@
-import { completion, loadModel, unloadModel } from "@qvac/sdk";
+import { completion, loadModel, unloadModel } from '@qvac/sdk';
 // Get HTTP URL from command line arguments or use default HuggingFace URL
 const httpUrl = process.argv[2] ||
-    "https://huggingface.co/bartowski/Llama-3.2-1B-Instruct-GGUF/resolve/main/Llama-3.2-1B-Instruct-Q4_0.gguf";
+    'https://huggingface.co/bartowski/Llama-3.2-1B-Instruct-GGUF/resolve/main/Llama-3.2-1B-Instruct-Q4_0.gguf';
 console.log(`▸ Loading GGUF model from HTTP: ${httpUrl}`);
 try {
     // Load model from HTTP URL
     const modelId = await loadModel({
         modelSrc: httpUrl,
-        modelType: "llamacpp-completion",
+        modelType: 'llamacpp-completion',
         modelConfig: {
-            ctx_size: 4096,
+            ctx_size: 4096
         },
         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");
-        },
+                process.stderr.write('\n');
+        }
     });
     console.log(`▸ Model loaded successfully! Model ID: ${modelId}`);
     // Create conversation history
-    const history = [
-        { role: "user", content: "Explain quantum computing in 3 key points" },
-    ];
-    console.log("\n▸ AI Response:");
-    process.stdout.write(""); // Start response on new line
+    const history = [{ role: 'user', content: 'Explain quantum computing in 3 key points' }];
+    console.log('\n▸ AI Response:');
+    process.stdout.write(''); // Start response on new line
     // Stream completion
     const result = completion({ modelId, history, stream: true });
     for await (const token of result.tokenStream) {
         process.stdout.write(token);
     }
     const stats = await result.stats;
-    console.log("\n▸ Performance Stats:", stats);
-    console.log("\n\n▸ Completed!");
+    console.log('\n▸ Performance Stats:', stats);
+    console.log('\n\n▸ Completed!');
     await unloadModel({ modelId, clearStorage: false });
 }
 catch (error) {
-    console.error("✖", error);
+    console.error('✖', error);
     process.exit(1);
 }

package/dist/examples/llamacpp-multi-gpu.js

@@ -1,4 +1,4 @@
-import { completion, loadModel, unloadModel, VERBOSITY, LLAMA_3_2_1B_INST_Q4_0, } from "@qvac/sdk";
+import { completion, loadModel, unloadModel, VERBOSITY, LLAMA_3_2_1B_INST_Q4_0 } from '@qvac/sdk';
 // Multi-GPU inference distributes a model across multiple GPUs using llama.cpp's
 // built-in split modes. Two strategies are available:
 //
@@ -16,14 +16,14 @@ const modelSrc = process.argv[2] ?? LLAMA_3_2_1B_INST_Q4_0;
 try {
     const modelId = await loadModel({
         modelSrc,
-        modelType: "llamacpp-completion",
+        modelType: 'llamacpp-completion',
         modelConfig: {
-            "split-mode": "layer",
-            "tensor-split": "1,1",
-            "main-gpu": 0,
+            'split-mode': 'layer',
+            'tensor-split': '1,1',
+            'main-gpu': 0,
             ctx_size: 4096,
             gpu_layers: 99,
-            verbosity: VERBOSITY.ERROR,
+            verbosity: VERBOSITY.ERROR
         },
         onProgress: (progress) => {
             if (progress.shardInfo) {
@@ -34,24 +34,24 @@ try {
             else {
                 console.log(`▸ Downloading: ${progress.percentage.toFixed(1)}%`);
             }
-        },
+        }
     });
     const history = [
         {
-            role: "user",
-            content: "Explain the difference between pipeline and tensor parallelism in one paragraph.",
-        },
+            role: 'user',
+            content: 'Explain the difference between pipeline and tensor parallelism in one paragraph.'
+        }
     ];
     const result = completion({ modelId, history, stream: true });
     for await (const token of result.tokenStream) {
         process.stdout.write(token);
     }
     const stats = await result.stats;
-    console.log("\n\n▸ Stats:", stats);
+    console.log('\n\n▸ Stats:', stats);
     await unloadModel({ modelId, clearStorage: false });
     process.exit(0);
 }
 catch (error) {
-    console.error("✖", error);
+    console.error('✖', error);
     process.exit(1);
 }

package/dist/examples/llamacpp-multimodal.js

@@ -1,4 +1,4 @@
-import { completion, loadModel, SMOLVLM2_500M_MULTIMODAL_Q8_0, MMPROJ_SMOLVLM2_500M_MULTIMODAL_Q8_0, unloadModel, } from "@qvac/sdk";
+import { completion, loadModel, SMOLVLM2_500M_MULTIMODAL_Q8_0, MMPROJ_SMOLVLM2_500M_MULTIMODAL_Q8_0, unloadModel } from '@qvac/sdk';
 if (process.argv.length < 3) {
     console.error(`▸ Specify an image file path as the first argument and a second image file path as the second (optional) argument`);
     process.exit(1);
@@ -11,31 +11,31 @@ try {
         modelSrc: SMOLVLM2_500M_MULTIMODAL_Q8_0,
         modelConfig: {
             ctx_size: 1024,
-            projectionModelSrc: MMPROJ_SMOLVLM2_500M_MULTIMODAL_Q8_0,
+            projectionModelSrc: MMPROJ_SMOLVLM2_500M_MULTIMODAL_Q8_0
         },
         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");
-        },
+                process.stderr.write('\n');
+        }
     });
     //Using one particular media
     const history = [
         {
-            role: "user",
+            role: 'user',
             content: "What's in this image?",
-            attachments: [{ path: imageFilePath }],
-        },
+            attachments: [{ path: imageFilePath }]
+        }
     ];
     const result = completion({ modelId, history, stream: true });
     for await (const token of result.tokenStream) {
         process.stdout.write(token);
     }
     const stats = await result.stats;
-    console.log("\n▸ Performance Stats:", stats);
-    console.log("▸ --------------------------------");
+    console.log('\n▸ Performance Stats:', stats);
+    console.log('▸ --------------------------------');
     //Using multiple media
     if (process.argv.length < 4) {
         console.log(`▸ Only one image provided, terminating`);
@@ -44,21 +44,21 @@ try {
     const imageFilePath2 = process.argv[3];
     const history2 = [
         {
-            role: "user",
-            content: "Compare the two newspaper articles",
-            attachments: [{ path: imageFilePath }, { path: imageFilePath2 }],
-        },
+            role: 'user',
+            content: 'Compare the two newspaper articles',
+            attachments: [{ path: imageFilePath }, { path: imageFilePath2 }]
+        }
     ];
     const result2 = completion({ modelId, history: history2, stream: true });
     for await (const token of result2.tokenStream) {
         process.stdout.write(token);
     }
     const stats2 = await result2.stats;
-    console.log("\n▸ Performance Stats:", stats2);
-    console.log("▸ --------------------------------");
+    console.log('\n▸ Performance Stats:', stats2);
+    console.log('▸ --------------------------------');
     await unloadModel({ modelId, clearStorage: false });
 }
 catch (error) {
-    console.error("✖", error);
+    console.error('✖', error);
     process.exit(1);
 }

package/dist/examples/llamacpp-p2p.js

@@ -1,4 +1,4 @@
-import { completion, LLAMA_3_2_1B_INST_Q4_0, loadModel, downloadAsset, unloadModel, VERBOSITY, } from "@qvac/sdk";
+import { completion, LLAMA_3_2_1B_INST_Q4_0, loadModel, downloadAsset, unloadModel, VERBOSITY } from '@qvac/sdk';
 try {
     // First just cache the model
     await downloadAsset({
@@ -8,34 +8,34 @@ try {
             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");
-        },
+                process.stderr.write('\n');
+        }
     });
     // Then load it in memory from cache
     const modelId = await loadModel({
         modelSrc: LLAMA_3_2_1B_INST_Q4_0,
         modelConfig: {
-            device: "gpu",
+            device: 'gpu',
             ctx_size: 2048,
-            verbosity: VERBOSITY.ERROR,
-        },
+            verbosity: VERBOSITY.ERROR
+        }
     });
     const history = [
         {
-            role: "user",
-            content: "Explain quantum computing in one sentence, use lots of emojis",
-        },
+            role: 'user',
+            content: 'Explain quantum computing in one sentence, use lots of emojis'
+        }
     ];
     const result = completion({ modelId, history, stream: true });
     for await (const token of result.tokenStream) {
         process.stdout.write(token);
     }
     const stats = await result.stats;
-    console.log("\n▸ Performance Stats:", stats);
+    console.log('\n▸ Performance Stats:', stats);
     // Change `clearStorage: true` to delete cached model files
     await unloadModel({ modelId, clearStorage: false });
 }
 catch (error) {
-    console.error("✖", error);
+    console.error('✖', error);
     process.exit(1);
 }

package/dist/examples/llamacpp-sharded.js

@@ -1,4 +1,4 @@
-import { completion, loadModel, unloadModel, VERBOSITY } from "@qvac/sdk";
+import { completion, loadModel, unloadModel, VERBOSITY } from '@qvac/sdk';
 // Sharded models can be loaded from:
 // 1. HTTP archives: "https://example.com/model.tar.gz"
 // 2. HTTP pattern: "https://example.com/model-00001-of-00005.gguf"
@@ -7,12 +7,12 @@ import { completion, loadModel, unloadModel, VERBOSITY } from "@qvac/sdk";
 // 5. Local archive: pass the path to the archive file (.tar, .tar.gz, .tgz)
 try {
     const modelId = await loadModel({
-        modelSrc: "https://huggingface.co/Qwen/Qwen2.5-Coder-7B-Instruct-GGUF/resolve/main/qwen2.5-coder-7b-instruct-q4_0-00001-of-00002.gguf",
-        modelType: "llamacpp-completion",
+        modelSrc: 'https://huggingface.co/Qwen/Qwen2.5-Coder-7B-Instruct-GGUF/resolve/main/qwen2.5-coder-7b-instruct-q4_0-00001-of-00002.gguf',
+        modelType: 'llamacpp-completion',
         modelConfig: {
-            device: "gpu",
+            device: 'gpu',
             ctx_size: 2048,
-            verbosity: VERBOSITY.ERROR,
+            verbosity: VERBOSITY.ERROR
         },
         onProgress: (progress) => {
             // For sharded models, progress.shardInfo contains detailed progress for both
@@ -29,25 +29,25 @@ try {
                 console.log(`▸ Progress: ${progress.percentage.toFixed(1)}% ` +
                     `(${(progress.downloaded / 1024 / 1024).toFixed(2)}MB / ${(progress.total / 1024 / 1024).toFixed(2)}MB)`);
             }
-        },
+        }
     });
     const history = [
         {
-            role: "user",
-            content: "What are the benefits of sharding large language models? Use emojis in your response.",
-        },
+            role: 'user',
+            content: 'What are the benefits of sharding large language models? Use emojis in your response.'
+        }
     ];
     const result = completion({ modelId, history, stream: true });
-    console.log("\n▸ Model response:");
+    console.log('\n▸ Model response:');
     for await (const token of result.tokenStream) {
         process.stdout.write(token);
     }
     const stats = await result.stats;
-    console.log("\n\n▸ Performance Stats:", stats);
+    console.log('\n\n▸ Performance Stats:', stats);
     await unloadModel({ modelId, clearStorage: false });
     process.exit(0);
 }
 catch (error) {
-    console.error("✖", error);
+    console.error('✖', error);
     process.exit(1);
 }

package/dist/examples/llamacpp-structured-output.js

@@ -1,4 +1,4 @@
-import { loadModel, completion, unloadModel, QWEN3_600M_INST_Q4, } from "@qvac/sdk";
+import { loadModel, completion, unloadModel, QWEN3_600M_INST_Q4 } from '@qvac/sdk';
 // Demonstrates per-request structured output via `responseFormat`.
 // Three modes are exercised against the same model so you can eyeball the
 // difference between free-form text, free-form JSON, and a strict JSON Schema.
@@ -6,24 +6,24 @@ import { loadModel, completion, unloadModel, QWEN3_600M_INST_Q4, } from "@qvac/s
 // Usage:
 //   bun run examples/llamacpp-structured-output.ts
 const PERSON_SCHEMA = {
-    type: "object",
+    type: 'object',
     properties: {
-        name: { type: "string" },
-        age: { type: "integer" },
-        occupation: { type: "string" },
+        name: { type: 'string' },
+        age: { type: 'integer' },
+        occupation: { type: 'string' }
     },
-    required: ["name", "age", "occupation"],
-    additionalProperties: false,
+    required: ['name', 'age', 'occupation'],
+    additionalProperties: false
 };
 const HISTORY = [
     {
-        role: "system",
-        content: "You extract structured information about people from short bios. /no_think",
+        role: 'system',
+        content: 'You extract structured information about people from short bios. /no_think'
     },
     {
-        role: "user",
-        content: "Hi, I'm Alice, I'm 30 years old and I work as a data engineer.",
-    },
+        role: 'user',
+        content: "Hi, I'm Alice, I'm 30 years old and I work as a data engineer."
+    }
 ];
 // `json_object` mode only enforces that the output is *some* valid JSON object
 // — it doesn't pin the keys. Small models (Qwen3-0.6B in this example) will
@@ -34,13 +34,13 @@ const HISTORY = [
 // grammar itself forces the keys.
 const JSON_OBJECT_HISTORY = [
     {
-        role: "system",
-        content: 'You extract structured information about people from short bios and reply ONLY with a JSON object containing "name", "age", and "occupation". /no_think',
+        role: 'system',
+        content: 'You extract structured information about people from short bios and reply ONLY with a JSON object containing "name", "age", and "occupation". /no_think'
     },
     {
-        role: "user",
-        content: "Hi, I'm Alice, I'm 30 years old and I work as a data engineer.",
-    },
+        role: 'user',
+        content: "Hi, I'm Alice, I'm 30 years old and I work as a data engineer."
+    }
 ];
 // Drains the run via the canonical `events` / `final` surface and returns the
 // content text. `tokenStream` / `text` / `stats` still work but are legacy and
@@ -48,11 +48,11 @@ const JSON_OBJECT_HISTORY = [
 // await `final.contentText` for the aggregated result.
 async function runToContent(run) {
     for await (const event of run.events) {
-        if (event.type === "contentDelta")
+        if (event.type === 'contentDelta')
             process.stdout.write(event.text);
     }
     const final = await run.final;
-    process.stdout.write("\n");
+    process.stdout.write('\n');
     return final.contentText;
 }
 try {
@@ -63,49 +63,49 @@ try {
             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");
-        },
+                process.stderr.write('\n');
+        }
     });
     console.log(`\n▸ Model loaded: ${modelId}\n`);
-    console.log("▸ --- 1. responseFormat: text (baseline, free-form) ---");
+    console.log('▸ --- 1. responseFormat: text (baseline, free-form) ---');
     await runToContent(completion({
         modelId,
         history: HISTORY,
         stream: true,
-        responseFormat: { type: "text" },
+        responseFormat: { type: 'text' }
     }));
-    console.log("\n▸ --- 2. responseFormat: json_object (any valid JSON) ---");
+    console.log('\n▸ --- 2. responseFormat: json_object (any valid JSON) ---');
     const jsonObjectOut = await runToContent(completion({
         modelId,
         history: JSON_OBJECT_HISTORY,
         stream: true,
-        responseFormat: { type: "json_object" },
+        responseFormat: { type: 'json_object' }
     }));
-    console.log("▸ parsed:", JSON.parse(jsonObjectOut.trim()));
-    console.log("\n▸ --- 3. responseFormat: json_schema (strict shape) ---");
+    console.log('▸ parsed:', JSON.parse(jsonObjectOut.trim()));
+    console.log('\n▸ --- 3. responseFormat: json_schema (strict shape) ---');
     const jsonSchemaOut = await runToContent(completion({
         modelId,
         history: HISTORY,
         stream: true,
         responseFormat: {
-            type: "json_schema",
+            type: 'json_schema',
             json_schema: {
-                name: "person",
-                schema: PERSON_SCHEMA,
-            },
-        },
+                name: 'person',
+                schema: PERSON_SCHEMA
+            }
+        }
     }));
     // Output is guaranteed schema-valid JSON: object with exactly
     // {name: string, age: integer, occupation: string}, no extras.
     const parsed = JSON.parse(jsonSchemaOut.trim());
-    console.log("▸ parsed:", parsed);
-    console.log("▸ schema-valid:", typeof parsed.name === "string" &&
+    console.log('▸ parsed:', parsed);
+    console.log('▸ schema-valid:', typeof parsed.name === 'string' &&
         Number.isInteger(parsed.age) &&
-        typeof parsed.occupation === "string" &&
-        Object.keys(parsed).sort().join(",") === "age,name,occupation");
+        typeof parsed.occupation === 'string' &&
+        Object.keys(parsed).sort().join(',') === 'age,name,occupation');
     await unloadModel({ modelId });
 }
 catch (error) {
-    console.error("✖", error);
+    console.error('✖', error);
     process.exit(1);
 }

package/dist/examples/logging-file-transport.js

@@ -1,13 +1,13 @@
-import { getLogger, loadModel, completion, unloadModel, LLAMA_3_2_1B_INST_Q4_0, VERBOSITY, } from "@qvac/sdk";
-import { writeFileSync, appendFileSync } from "fs";
-import { join } from "path";
+import { getLogger, loadModel, completion, unloadModel, LLAMA_3_2_1B_INST_Q4_0, VERBOSITY } from '@qvac/sdk';
+import { writeFileSync, appendFileSync } from 'fs';
+import { join } from 'path';
 try {
-    const logFile = join(process.cwd(), "model-logs.txt");
+    const logFile = join(process.cwd(), 'model-logs.txt');
     // Clear previous logs
-    writeFileSync(logFile, "");
+    writeFileSync(logFile, '');
     // Create logger with file transport
-    const logger = getLogger("llm-file-demo", {
-        level: "debug",
+    const logger = getLogger('llm-file-demo', {
+        level: 'debug',
         enableConsole: false, // Disable console output (logs go to file only). Set to true for both console + file
         transports: [
             // File transport - append logs to file
@@ -15,42 +15,40 @@ try {
                 const timestamp = new Date().toISOString();
                 const logLine = `[${timestamp}] ${level.toUpperCase()} ${namespace}: ${message}\n`;
                 appendFileSync(logFile, logLine);
-            },
-        ],
+            }
+        ]
     });
-    logger.info("Starting model demo with file logging");
+    logger.info('Starting model demo with file logging');
     // Load model with integrated logging
     const modelId = await loadModel({
         modelSrc: LLAMA_3_2_1B_INST_Q4_0,
         modelConfig: {
             ctx_size: 2048,
             temp: 0.7,
-            verbosity: VERBOSITY.DEBUG,
+            verbosity: VERBOSITY.DEBUG
         },
-        logger, // All model logs will be captured
+        logger // All model logs will be captured
     });
     logger.info(`Model loaded: ${modelId}`);
     // Run a simple completion
-    logger.info("Starting completion...");
-    const messages = [
-        { role: "user", content: "Tell me a short joke about programming." },
-    ];
-    let responseText = "";
+    logger.info('Starting completion...');
+    const messages = [{ role: 'user', content: 'Tell me a short joke about programming.' }];
+    let responseText = '';
     const result = completion({ modelId, history: messages, stream: true });
     for await (const token of result.tokenStream) {
         responseText += token;
         process.stdout.write(token);
     }
-    console.log("\n");
+    console.log('\n');
     logger.info(`Completion finished, ${responseText.length} characters generated`);
     // Unload model
     await unloadModel({ modelId });
-    logger.info("Model unloaded successfully");
+    logger.info('Model unloaded successfully');
     // Show user where logs were saved
     console.log(`▸ Logs saved to ${logFile}`);
-    console.log("▸ Inspect it to see detailed model operation logs");
+    console.log('▸ Inspect it to see detailed model operation logs');
 }
 catch (error) {
-    console.error("✖", error);
+    console.error('✖', error);
     process.exit(1);
 }

package/dist/examples/logging-global.js

@@ -2,10 +2,10 @@
 // transport on both the client and the worker, so the only logs printed are
 // the ones our handler below receives.
 const configDir = import.meta.dirname ?? process.cwd();
-process.env["QVAC_CONFIG_PATH"] = `${configDir}/config/logging/logging.config.json`;
-const { loadModel, completion, unloadModel, subscribeServerLogs, LLAMA_3_2_1B_INST_Q4_0 } = await import("@qvac/sdk");
+process.env['QVAC_CONFIG_PATH'] = `${configDir}/config/logging/logging.config.json`;
+const { loadModel, completion, unloadModel, subscribeServerLogs, LLAMA_3_2_1B_INST_Q4_0 } = await import('@qvac/sdk');
 try {
-    console.log("▸ Starting global logging demo...\n");
+    console.log('▸ Starting global logging demo...\n');
     // One subscription captures every server-side log (SDK, models, RAG, …)
     // without having to know any stream IDs ahead of time.
     const unsubscribe = subscribeServerLogs((log) => {
@@ -13,14 +13,14 @@ try {
     });
     const modelId = await loadModel({
         modelSrc: LLAMA_3_2_1B_INST_Q4_0,
-        modelConfig: { ctx_size: 2048 },
+        modelConfig: { ctx_size: 2048 }
     });
     const result = completion({
         modelId,
-        history: [{ role: "user", content: "Count from 1 to 5." }],
-        stream: true,
+        history: [{ role: 'user', content: 'Count from 1 to 5.' }],
+        stream: true
     });
-    console.log("▸ Response:\n");
+    console.log('▸ Response:\n');
     for await (const token of result.tokenStream) {
         process.stdout.write(token);
     }
@@ -28,7 +28,7 @@ try {
     unsubscribe();
 }
 catch (error) {
-    console.error("✖", error);
+    console.error('✖', error);
     process.exit(1);
 }
 export {};