Update app.py

app.py

@@ -86,7 +86,7 @@ MODEL_NAME  = "gpt-3.5-turbo-16k"
 
 #HuggingFace--------------------------------
 #repo_id = "meta-llama/Llama-2-13b-chat-hf"   
-repo_id = "HuggingFaceH4/zephyr-7b-alpha"   #das Modell ist echt gut!!!
+repo_id = "HuggingFaceH4/zephyr-7b-alpha"   #das Modell ist echt gut!!! Vom MIT
 #repo_id = "meta-llama/Llama-2-70b-chat-hf"    
 #repo_id = "tiiuae/falcon-40b"  
 #repo_id = "Vicuna-33b"
@@ -277,7 +277,7 @@ def generate_prompt_with_history_langchain(prompt, history):
 
 ###################################################
 #Funktion von Gradio aus, die den dort eingegebenen Prompt annimmt und weiterverarbeitet
-def invoke (prompt, history, rag_option, openai_api_key,  temperature=0.9, max_new_tokens=512, top_p=0.6, repetition_penalty=1.3,):
+def invoke (prompt, history, rag_option, model_option, openai_api_key,  temperature=0.9, max_new_tokens=512, top_p=0.6, repetition_penalty=1.3,):
     global splittet
     print(splittet)
     #Prompt an history anhängen und einen Text daraus machen
@@ -304,12 +304,14 @@ def invoke (prompt, history, rag_option, openai_api_key,  temperature=0.9, max_n
         ###########################
         #LLM auswählen (OpenAI oder HF)
         ###########################
-        #Anfrage an OpenAI ----------------------------
-        llm = ChatOpenAI(model_name = MODEL_NAME,  openai_api_key = openai_api_key, temperature=temperature)#, top_p = top_p)
-        #oder an Hugging Face --------------------------
-        #llm = HuggingFaceHub(repo_id=repo_id, model_kwargs={"temperature": 0.5, "max_length": 64}) 
-        #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,)
+        if (model_option == "OpenAI"):
+            #Anfrage an OpenAI ----------------------------
+            llm = ChatOpenAI(model_name = MODEL_NAME,  openai_api_key = openai_api_key, temperature=temperature)#, top_p = top_p)
+        else:
+            #oder an Hugging Face --------------------------
+            llm = HuggingFaceHub(repo_id=repo_id, model_kwargs={"temperature": 0.5, "max_length": 64}) 
+            #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,)
         
 
         #zusätzliche Dokumenten Splits aus DB zum Prompt hinzufügen (aus VektorDB - Chroma oder Mongo DB)
@@ -355,6 +357,7 @@ def vote(data: gr.LikeData):
 additional_inputs = [
     #gr.Radio(["Off", "Chroma", "MongoDB"], label="Retrieval Augmented Generation", value = "Off"),
     gr.Radio(["Aus", "An"], label="RAG - LI Erweiterungen", value = "Aus"),
+    gr.Radio(["OpenAI", "HuggingFace"], label="Modellauswahl", value = "OpenAI"),
     gr.Textbox(label = "OpenAI API Key", value = "sk-", lines = 1), 
     gr.Slider(label="Temperature", value=0.9, minimum=0.0, maximum=1.0, step=0.05, interactive=True, info="Höhere Werte erzeugen diversere Antworten", visible=False),
     gr.Slider(label="Max new tokens", value=256, minimum=0, maximum=4096, step=64, interactive=True, info="Maximale Anzahl neuer Tokens", visible=False),