model: add Qwen-2 Math models

mlc-ai · Aug 8, 2024 · 25fd9ec · 25fd9ec
1 parent 35bc80c
commit 25fd9ec
Show file tree

Hide file tree

Showing 4 changed files with 72 additions and 20 deletions.
diff --git a/app/constant.ts b/app/constant.ts
@@ -597,7 +597,6 @@ export const DEFAULT_MODELS: ModelRecord[] = [
     provider: "Alibaba",
     size: "7B",
     quantization: "q4f16_1",
-
     family: "Qwen 2",
     vram_required_MB: 5106.67,
     low_resource_required: false,
@@ -614,7 +613,71 @@ export const DEFAULT_MODELS: ModelRecord[] = [
     provider: "Alibaba",
     size: "7B",
     quantization: "q4f32",
-
+    family: "Qwen 2",
+    vram_required_MB: 5900.09,
+    low_resource_required: false,
+    recommended_config: {
+      temperature: 0.7,
+      presence_penalty: 0,
+      frequency_penalty: 0,
+      top_p: 0.8,
+    },
+  },
+  // Qwen2-Math
+  {
+    name: "Qwen2-Math-1.5B-Instruct-q4f16_1-MLC",
+    display_name: "Qwen",
+    provider: "Alibaba",
+    size: "1.5B",
+    quantization: "q4f16_1",
+    family: "Qwen 2",
+    vram_required_MB: 1629.75,
+    low_resource_required: true,
+    recommended_config: {
+      temperature: 1.0,
+      presence_penalty: 0,
+      frequency_penalty: 0,
+      top_p: 0.8,
+    },
+  },
+  {
+    name: "Qwen2-Math-1.5B-Instruct-q4f32_1-MLC",
+    display_name: "Qwen",
+    provider: "Alibaba",
+    size: "1.5B",
+    quantization: "q4f32_1",
+    family: "Qwen 2",
+    vram_required_MB: 1888.97,
+    low_resource_required: true,
+    recommended_config: {
+      temperature: 1.0,
+      presence_penalty: 0,
+      frequency_penalty: 0,
+      top_p: 0.8,
+    },
+  },
+  {
+    name: "Qwen2-Math-7B-Instruct-q4f16_1-MLC",
+    display_name: "Qwen",
+    provider: "Alibaba",
+    size: "7B",
+    quantization: "q4f16_1",
+    family: "Qwen 2",
+    vram_required_MB: 5106.67,
+    low_resource_required: false,
+    recommended_config: {
+      temperature: 0.7,
+      presence_penalty: 0,
+      frequency_penalty: 0,
+      top_p: 0.8,
+    },
+  },
+  {
+    name: "Qwen2-Math-7B-Instruct-q4f32_1-MLC",
+    display_name: "Qwen",
+    provider: "Alibaba",
+    size: "7B",
+    quantization: "q4f32_1",
     family: "Qwen 2",
     vram_required_MB: 5900.09,
     low_resource_required: false,

diff --git a/app/store/config.ts b/app/store/config.ts
@@ -208,25 +208,14 @@ export const useAppConfig = createPersistStore(
   }),
   {
     name: StoreKey.Config,
-    version: 0.48,
+    version: 0.49,
     migrate: (persistedState, version) => {
-      if (version < 0.48) {
+      if (version < 0.49) {
         return {
           ...DEFAULT_CONFIG,
           ...(persistedState as any),
           models: DEFAULT_MODELS as any as ModelRecord[],
 
-          sendMemory: (persistedState as any).modelConfig?.sendMemory || true,
-          historyMessageCount:
-            (persistedState as any).modelConfig?.historyMessageCount || 4,
-          compressMessageLengthThreshold:
-            (persistedState as any).modelConfig
-              ?.compressMessageLengthThreshold || 1000,
-          enableInjectSystemPrompts:
-            (persistedState as any).modelConfig?.enableInjectSystemPrompts ||
-            false,
-          template: DEFAULT_INPUT_TEMPLATE,
-
           modelConfig: {
             model: DEFAULT_MODELS[0].name,
 

diff --git a/package.json b/package.json
@@ -17,7 +17,7 @@
   "dependencies": {
     "@fortaine/fetch-event-source": "^3.0.6",
     "@hello-pangea/dnd": "^16.5.0",
-    "@mlc-ai/web-llm": "^0.2.52",
+    "@mlc-ai/web-llm": "^0.2.55",
     "@serwist/next": "^9.0.2",
     "@svgr/webpack": "^6.5.1",
     "emoji-picker-react": "^4.9.2",

diff --git a/yarn.lock b/yarn.lock
@@ -1180,10 +1180,10 @@
     "@jridgewell/resolve-uri" "^3.1.0"
     "@jridgewell/sourcemap-codec" "^1.4.14"
 
-"@mlc-ai/web-llm@^0.2.52":
-  version "0.2.52"
-  resolved "https://registry.yarnpkg.com/@mlc-ai/web-llm/-/web-llm-0.2.52.tgz#15c0496b3ce8c35bd0ac5d1dd92d30cdc6c055db"
-  integrity sha512-TRjAQsOCUf7vrwVeOZoBZtB9KnW9InWK2w85rJWL7IBSSW/znYsNdD3/opmEGtf7NTKr4RBIu+8RdnhcZoHk4A==
+"@mlc-ai/web-llm@^0.2.55":
+  version "0.2.55"
+  resolved "https://registry.yarnpkg.com/@mlc-ai/web-llm/-/web-llm-0.2.55.tgz#678c1703df1a9c9d504fc81db6d2f7b7c5d39d7b"
+  integrity sha512-o2Nzmx//sT3ttWrnSYy6s7NTCnBEWPI8ubbzg6WH59dAI34pHSP/+V8UxtDS+ZlEXl0PV0iDgl1i+Y1WeKABlw==
   dependencies:
     loglevel "^1.9.1"