Update app.py

app.py

@@ -15,6 +15,8 @@ from datetime import datetime
 from transformers import AutoTokenizer, AutoModelForCausalLM ,pipeline
 from langchain_huggingface.llms import HuggingFacePipeline
 import spaces
+from langchain_huggingface.llms import HuggingFacePipeline
+from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
 
 
 # Configure logging
@@ -35,7 +37,7 @@ class ChatHistory:
         timestamp = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
         self.messages.append(Message(role=role, content=content, timestamp=timestamp))
     
-    def get_formatted_history(self, max_messages: int = 5) -> str:
+    def get_formatted_history(self, max_messages: int = 10) -> str:
         """Returns the most recent conversation history formatted as a string"""
         recent_messages = self.messages[-max_messages:] if len(self.messages) > max_messages else self.messages
         formatted_history = "\n".join([
@@ -99,7 +101,7 @@ db = Qdrant(
 # Create retriever
 retriever = db.as_retriever(
     search_type="similarity",
-    search_kwargs={"k": 5}
+    search_kwargs={"k": 3}
 )
 
 
@@ -117,14 +119,11 @@ retriever = db.as_retriever(
 #     timeout=None
     
 # )
-repo_id = "CohereForAI/c4ai-command-r7b-12-2024"
-
-llm = HuggingFaceEndpoint(
-    repo_id=repo_id,
-    max_length=8192,
-    temperature=0,
-    huggingfacehub_api_token=HF_TOKEN,
-)
+model_id = "CohereForAI/c4ai-command-r7b-12-2024"
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForCausalLM.from_pretrained(model_id)
+pipe = pipeline("text-generation", model=model, tokenizer=tokenizer, max_new_tokens=8192 )
+llm = HuggingFacePipeline(pipeline=pipe)