Merge pull request #3665 from janhq/fix/phi3-medium-prompt-template

fix: correct prompt template for Phi3 Medium model
janhq · Sep 16, 2024 · daf372f · daf372f
2 parents 6e0c582 + 4087b94
commit daf372f
Show file tree

Hide file tree

Showing 2 changed files with 3 additions and 3 deletions.
diff --git a/extensions/inference-nitro-extension/package.json b/extensions/inference-nitro-extension/package.json
@@ -1,7 +1,7 @@
 {
   "name": "@janhq/inference-cortex-extension",
   "productName": "Cortex Inference Engine",
-  "version": "1.0.16",
+  "version": "1.0.17",
   "description": "This extension embeds cortex.cpp, a lightweight inference engine written in C++. See https://jan.ai.\nAdditional dependencies could be installed to run without Cuda Toolkit installation.",
   "main": "dist/index.js",
   "node": "dist/node/index.cjs.js",

diff --git a/extensions/inference-nitro-extension/resources/models/phi3-medium/model.json b/extensions/inference-nitro-extension/resources/models/phi3-medium/model.json
@@ -8,12 +8,12 @@
     "id": "phi3-medium",
     "object": "model",
     "name": "Phi-3 Medium Instruct Q4",
-    "version": "1.3",
+    "version": "1.4",
     "description": "Phi-3 Medium is Microsoft's latest SOTA model.",
     "format": "gguf",
     "settings": {
       "ctx_len": 128000,
-      "prompt_template": "<|user|> {prompt}<|end|><|assistant|><|end|>",
+      "prompt_template": "<|user|> {prompt}<|end|><|assistant|>",
       "llama_model_path": "Phi-3-medium-128k-instruct-Q4_K_M.gguf",
       "ngl": 33
     },