Update app.py

app.py

@@ -103,7 +103,10 @@ repo_id = "HuggingFaceH4/zephyr-7b-alpha"   #das Modell ist echt gut!!! Vom MIT
 MODEL_NAME_HF  = "mistralai/Mixtral-8x7B-Instruct-v0.1"
 MODEL_NAME_OAI_ZEICHNEN = "dall-e-3"
 #Alternativ zeichnen: Stabe Diffusion from HF:
+#Zeichnen
 API_URL = "https://api-inference.huggingface.co/models/stabilityai/stable-diffusion-2-1"
+#Textgenerierung
+API_URL_TEXT = "https://api-inference.huggingface.co/models/argilla/notux-8x7b-v1"
 
 ################################################
 #HF Hub Zugriff ermöglichen
@@ -435,10 +438,11 @@ def generate_text (prompt, chatbot, history, rag_option, model_option, openai_ap
             #oder an Hugging Face --------------------------
             print("HF Anfrage.......................")
             model_kwargs={"temperature": 0.5, "max_length": 512, "num_return_sequences": 1, "top_k": top_k, "top_p": top_p, "repetition_penalty": repetition_penalty}
-            llm = HuggingFaceHub(repo_id=repo_id, model_kwargs=model_kwargs) 
+            #llm = HuggingFaceHub(repo_id=repo_id, model_kwargs=model_kwargs) 
             #llm = HuggingFaceChain(model=MODEL_NAME_HF, model_kwargs={"temperature": 0.5, "max_length": 128}) 
             #llm = HuggingFaceHub(url_??? = "https://wdgsjd6zf201mufn.us-east-1.aws.endpoints.huggingface.cloud", model_kwargs={"temperature": 0.5, "max_length": 64}) 
             #llm = HuggingFaceTextGenInference( inference_server_url="http://localhost:8010/", max_new_tokens=max_new_tokens,top_k=10,top_p=top_p,typical_p=0.95,temperature=temperature,repetition_penalty=repetition_penalty,)
+            
             print("HF")
             #Prompt an history anhängen und einen Text daraus machen
             history_text_und_prompt = generate_prompt_with_history(prompt, history) 
@@ -453,10 +457,18 @@ def generate_text (prompt, chatbot, history, rag_option, model_option, openai_ap
             #db = document_retrieval_mongodb(llm, history_text_und_prompt)
             #result = rag_chain(llm, history_text_und_prompt, db)
         else:
-            splittet = False
+            #splittet = False
             print("LLM aufrufen ohne RAG: ...........")
-            resulti = llm_chain(llm, history_text_und_prompt)
-            result = resulti.strip()
+            if (model_option == "OpenAI"):
+                resulti = llm_chain(llm, history_text_und_prompt)
+                result = resulti.strip()
+            else:
+                data = {"inputs": prompt}
+                response = requests.post(API_URL, headers=HEADERS, json=data)
+                result = response.json()
+                print("result. HF API.....................")
+                print(result)
+                
 
         #Wenn keine Antwort möglich "Ich weiß es nicht" etc., dann versuchen mit Suche im Internet.
         if (result == None or is_response_similar(result)):