Update app.py

app.py

@@ -7,6 +7,7 @@ from custom_llm import CustomLLM
 from pydantic import BaseModel
 from langchain.prompts import PromptTemplate
 from langchain_huggingface import HuggingFacePipeline
+from langchain_huggingface import HuggingFaceEndpoint
 
 
 class ConversationPost(BaseModel):
@@ -16,19 +17,30 @@ class ConversationPost(BaseModel):
 
 
 API_TOKEN = os.environ['HF_API_KEY']
+
+os.environ["HUGGINGFACEHUB_API_TOKEN"] = API_TOKEN
+
 app = FastAPI()
 prompt = PromptTemplate.from_template("""<|im_start|>system
-Kamu adalah Asisten AI yang dikembangkan oleh Jonthan Jordan. Jawablah pertanyaan user secara ketat dalam Bahasa Indonesia<|im_end|>
+Kamu adalah Asisten AI yang dikembangkan oleh Jonthan Jordan. Answer strictly in Bahasa Indonesia<|im_end|>
 <|im_start|>user
 {question}<|im_end|>
 <|im_start|>assistant
 """)
-llm = prompt | HuggingFacePipeline.from_model_id(
-    model_id="Qwen/Qwen2-1.5B-Instruct",
+# llm = prompt | HuggingFacePipeline.from_model_id(
+#     model_id="Qwen/Qwen2-1.5B-Instruct",
+#     task="text-generation",
+#     pipeline_kwargs={
+#         "max_new_tokens": 150,
+#         "return_full_text":False
+#     },
+# )
+
+llm = prompt | HuggingFaceEndpoint(
+    repo_id="Qwen/Qwen-VL-Chat",
     task="text-generation",
-    pipeline_kwargs={
-        "max_new_tokens": 150,
-    },
+    max_new_tokens=150,
+    do_sample=False,
 )
 
 # llm = prompt | CustomLLM(repo_id="Qwen/Qwen-VL-Chat", model_type='text-generation', api_token=API_TOKEN, max_new_tokens=150).bind(stop=['<|im_end|>'])