Upload .config/inference/inference-config.json with huggingface_hub

.config/inference/inference-config.json

@@ -0,0 +1,12 @@
+{
+  "task_type": "text-generation",
+  "use_chat_template": true,
+  "parameters": {
+    "temperature": 0.7,
+    "top_p": 0.95,
+    "top_k": 50,
+    "repetition_penalty": 1.1,
+    "max_new_tokens": 512,
+    "do_sample": true
+  }
+}