@qvac/sdk 0.14.0 → 0.15.0

package/dist/examples/llamacpp-cache.js

@@ -1,28 +1,28 @@
-import { completion, deleteCache, LLAMA_3_2_1B_INST_Q4_0, loadModel, unloadModel, VERBOSITY, } from "@qvac/sdk";
-const sharedCacheKey = "llamacpp-cache-demo-shared";
-const isolatedCacheKey = "llamacpp-cache-demo-isolated";
+import { completion, deleteCache, LLAMA_3_2_1B_INST_Q4_0, loadModel, unloadModel, VERBOSITY } from '@qvac/sdk';
+const sharedCacheKey = 'llamacpp-cache-demo-shared';
+const isolatedCacheKey = 'llamacpp-cache-demo-isolated';
 const systemPrompt = {
-    role: "system",
-    content: "You are a concise travel assistant. Keep track of the user's preferences across turns.",
+    role: 'system',
+    content: "You are a concise travel assistant. Keep track of the user's preferences across turns."
 };
 function printStats(stats) {
-    console.log("\n▸ Performance Stats:", {
+    console.log('\n▸ Performance Stats:', {
         timeToFirstToken: stats?.timeToFirstToken,
         tokensPerSecond: stats?.tokensPerSecond,
-        cacheTokens: stats?.cacheTokens,
+        cacheTokens: stats?.cacheTokens
     });
 }
 async function runCachedCompletion(params) {
     console.log(`\n▸ ${params.label}`);
-    console.log(`▸ Cache key: ${params.kvCache || "disabled"}`);
-    console.log("▸ AI Response:");
+    console.log(`▸ Cache key: ${params.kvCache || 'disabled'}`);
+    console.log('▸ AI Response:');
     const result = completion({
         modelId: params.modelId,
         history: params.history,
         stream: true,
-        kvCache: params.kvCache,
+        kvCache: params.kvCache
     });
-    let text = "";
+    let text = '';
     for await (const token of result.tokenStream) {
         text += token;
         process.stdout.write(token);
@@ -33,61 +33,61 @@ async function runCachedCompletion(params) {
 }
 let modelId;
 try {
-    console.log("▸ Loading llama.cpp model with cache support...");
+    console.log('▸ Loading llama.cpp model with cache support...');
     modelId = await loadModel({
         modelSrc: LLAMA_3_2_1B_INST_Q4_0,
         modelConfig: {
             ctx_size: 4096,
-            verbosity: VERBOSITY.ERROR,
+            verbosity: VERBOSITY.ERROR
         },
         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}`);
     await deleteCache({ kvCacheKey: sharedCacheKey });
     await deleteCache({ kvCacheKey: isolatedCacheKey });
     const firstUserMessage = {
-        role: "user",
-        content: "I am planning a weekend in Lisbon. I prefer museums, seafood, and walking routes. Suggest a day plan in 3 bullets.",
+        role: 'user',
+        content: 'I am planning a weekend in Lisbon. I prefer museums, seafood, and walking routes. Suggest a day plan in 3 bullets.'
     };
     const firstTurn = await runCachedCompletion({
-        label: "Turn 1: create a named cache",
+        label: 'Turn 1: create a named cache',
         modelId,
         kvCache: sharedCacheKey,
-        history: [systemPrompt, firstUserMessage],
+        history: [systemPrompt, firstUserMessage]
     });
     await runCachedCompletion({
-        label: "Turn 2: reuse the same cache key",
+        label: 'Turn 2: reuse the same cache key',
         modelId,
         kvCache: sharedCacheKey,
         history: [
             systemPrompt,
             firstUserMessage,
-            { role: "assistant", content: firstTurn.text.trim() },
+            { role: 'assistant', content: firstTurn.text.trim() },
             {
-                role: "user",
-                content: "Now revise the plan for a rainy day, but keep the same preferences.",
-            },
-        ],
+                role: 'user',
+                content: 'Now revise the plan for a rainy day, but keep the same preferences.'
+            }
+        ]
     });
     await runCachedCompletion({
-        label: "Turn 3: use a different cache key for an isolated session",
+        label: 'Turn 3: use a different cache key for an isolated session',
         modelId,
         kvCache: isolatedCacheKey,
         history: [
             systemPrompt,
             {
-                role: "user",
-                content: "I am planning a weekend in Berlin. I prefer nightlife and late starts. Suggest a day plan in 3 bullets.",
-            },
-        ],
+                role: 'user',
+                content: 'I am planning a weekend in Berlin. I prefer nightlife and late starts. Suggest a day plan in 3 bullets.'
+            }
+        ]
     });
-    console.log("\n▸ Summary:");
+    console.log('\n▸ Summary:');
     console.log(`▸ - Reusing "${sharedCacheKey}" keeps conversation context across turns.`);
     console.log(`▸ - Switching to "${isolatedCacheKey}" starts a separate cache session.`);
     console.log('▸ - Use deleteCache({ kvCacheKey: "your-session" }) to clear a saved session.');
@@ -102,9 +102,9 @@ catch (error) {
             await unloadModel({ modelId, clearStorage: false });
         }
         catch (cleanupError) {
-            console.warn("✖ cleanup failed:", cleanupError);
+            console.warn('✖ cleanup failed:', cleanupError);
         }
     }
-    console.error("✖", error);
+    console.error('✖', error);
     process.exit(1);
 }

package/dist/examples/llamacpp-dynamic-tools.js

@@ -12,51 +12,51 @@
  *   bun run build
  *   bun run bare:example dist/examples/llamacpp-dynamic-tools.js
  */
-import { z } from "zod";
-import { completion, loadModel, unloadModel, QWEN3_1_7B_INST_Q4, } from "@qvac/sdk";
+import { z } from 'zod';
+import { completion, loadModel, unloadModel, QWEN3_1_7B_INST_Q4 } from '@qvac/sdk';
 const weatherSchema = z.object({
-    city: z.string().describe("City name"),
+    city: z.string().describe('City name')
 });
 const horoscopeSchema = z.object({
-    sign: z.string().describe("An astrological sign, e.g. Taurus or Aquarius"),
+    sign: z.string().describe('An astrological sign, e.g. Taurus or Aquarius')
 });
 const dateSchema = z.object({});
 const toolSchemas = {
     get_weather: weatherSchema,
     get_horoscope: horoscopeSchema,
-    get_date: dateSchema,
+    get_date: dateSchema
 };
 const weatherTools = [
     {
-        name: "get_weather",
-        description: "Get current weather for a city",
-        parameters: weatherSchema,
-    },
+        name: 'get_weather',
+        description: 'Get current weather for a city',
+        parameters: weatherSchema
+    }
 ];
 const horoscopeTools = [
     {
-        name: "get_horoscope",
+        name: 'get_horoscope',
         description: "Get today's horoscope for an astrological sign",
-        parameters: horoscopeSchema,
-    },
+        parameters: horoscopeSchema
+    }
 ];
 const dateTools = [
     {
-        name: "get_date",
+        name: 'get_date',
         description: "Get today's date",
-        parameters: dateSchema,
-    },
+        parameters: dateSchema
+    }
 ];
 function executeToolCall(call) {
-    if (call.name === "get_weather") {
+    if (call.name === 'get_weather') {
         const args = call.arguments;
         return `The weather in ${args.city} is rainy, 8°C with heavy clouds.`;
     }
-    if (call.name === "get_horoscope") {
+    if (call.name === 'get_horoscope') {
         const args = call.arguments;
         return `Horoscope for ${args.sign}: a great day for new beginnings.`;
     }
-    if (call.name === "get_date") {
+    if (call.name === 'get_date') {
         return new Date().toISOString().slice(0, 10);
     }
     return `Unknown tool: ${call.name}`;
@@ -67,7 +67,7 @@ async function chatTurn({ modelId, kvCache, history, tools }) {
         history,
         tools,
         kvCache,
-        stream: true,
+        stream: true
     });
     const tokensTask = (async () => {
         for await (const token of result.tokenStream) {
@@ -83,7 +83,7 @@ async function chatTurn({ modelId, kvCache, history, tools }) {
     const text = await result.text;
     const toolCalls = await result.toolCalls;
     if (toolCalls.length === 0) {
-        history.push({ role: "assistant", content: text });
+        history.push({ role: 'assistant', content: text });
         return;
     }
     for (const call of toolCalls) {
@@ -95,9 +95,9 @@ async function chatTurn({ modelId, kvCache, history, tools }) {
             }
         }
     }
-    history.push({ role: "assistant", content: text });
+    history.push({ role: 'assistant', content: text });
     for (const call of toolCalls) {
-        history.push({ role: "tool", content: executeToolCall(call) });
+        history.push({ role: 'tool', content: executeToolCall(call) });
     }
     // Follow-up turn so the model can incorporate the tool results.
     await chatTurn({ modelId, kvCache, history, tools });
@@ -108,45 +108,45 @@ async function main() {
         modelConfig: {
             ctx_size: 4096,
             tools: true,
-            toolsMode: "dynamic",
+            toolsMode: 'dynamic'
         },
         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: ${modelId}`);
     const kvCache = `dynamic-tools-${Date.now()}`;
     const history = [
         {
-            role: "system",
-            content: "You are a helpful assistant that uses tools when they are available. " +
-                "User's cat is named Windy and dog is named Butch.",
-        },
+            role: 'system',
+            content: 'You are a helpful assistant that uses tools when they are available. ' +
+                "User's cat is named Windy and dog is named Butch."
+        }
     ];
     // Turn 1 — only weather tools available.
-    history.push({ role: "user", content: "What's the weather in Tokyo?" });
-    console.log("\n▸ Turn 1 (tools=weather):\n");
+    history.push({ role: 'user', content: "What's the weather in Tokyo?" });
+    console.log('\n▸ Turn 1 (tools=weather):\n');
     await chatTurn({ modelId, kvCache, history, tools: weatherTools });
     // Turn 2 — same session, swap to horoscope tools. Dynamic mode lets the
     // model see a different tool set without invalidating the kv-cache.
     history.push({
-        role: "user",
-        content: "Now check my horoscope for Aquarius.",
+        role: 'user',
+        content: 'Now check my horoscope for Aquarius.'
     });
-    console.log("\n\n▸ Turn 2 (tools=horoscope):\n");
+    console.log('\n\n▸ Turn 2 (tools=horoscope):\n');
     await chatTurn({ modelId, kvCache, history, tools: horoscopeTools });
     // Turn 3 — swap to a parameterless tool to confirm empty-arg flows work.
-    history.push({ role: "user", content: "What's today's date?" });
-    console.log("\n\n▸ Turn 3 (tools=date):\n");
+    history.push({ role: 'user', content: "What's today's date?" });
+    console.log('\n\n▸ Turn 3 (tools=date):\n');
     await chatTurn({ modelId, kvCache, history, tools: dateTools });
-    console.log("\n\n▸ Done.");
+    console.log('\n\n▸ Done.');
     await unloadModel({ modelId, clearStorage: false });
 }
 main().catch((err) => {
-    console.error("✖", err);
+    console.error('✖', err);
     process.exit(1);
 });

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);
 }