Chore: Add stop token for Gemma 2b (#3125)

* add stop token

* Bump version

extensions/inference-nitro-extension/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@janhq/inference-cortex-extension",
   "productName": "Cortex Inference Engine",
-  "version": "1.0.13",
+  "version": "1.0.14",
   "description": "This extension embeds cortex.cpp, a lightweight inference engine written in C++. See https://nitro.jan.ai.\nAdditional dependencies could be installed to run without Cuda Toolkit installation.",
   "main": "dist/index.js",
   "node": "dist/node/index.cjs.js",

extensions/inference-nitro-extension/resources/models/gemma-2b/model.json

@@ -8,7 +8,7 @@
   "id": "gemma-2b",
   "object": "model",
   "name": "Gemma 2B Q4",
-  "version": "1.2",
+  "version": "1.3",
   "description": "Gemma is built from the same technology with Google's Gemini.",
   "format": "gguf",
   "settings": {
@@ -22,7 +22,7 @@
     "top_p": 0.95,
     "stream": true,
     "max_tokens": 8192,
-    "stop": [],
+    "stop": ["<end_of_turn>"],
     "frequency_penalty": 0,
     "presence_penalty": 0
   },