Update app.py

app.py

@@ -1,73 +1,78 @@
-import gradio as gr
-from huggingface_hub import InferenceClient
-
-# InferenceClient'ı uygun model ile başlatıyoruz
-client = InferenceClient(model="HuggingFaceH4/zephyr-7b-beta")
-
-def respond(message, history, system_message, max_tokens, temperature, top_p):
-    # Mesajları hazırlıyoruz, sistem mesajı ile başlıyoruz
-    messages = [{"role": "system", "content": system_message}]
-
-    # Geçmişteki kullanıcı ve asistan mesajlarını ekliyoruz
-    for user_msg, assistant_msg in history:
-        if user_msg:
-            messages.append({"role": "user", "content": user_msg})
-        if assistant_msg:
-            messages.append({"role": "assistant", "content": assistant_msg})
-
-    # Son kullanıcı mesajını ekliyoruz
-    messages.append({"role": "user", "content": message})
-
-    response = ""
-    try:
-        # Yanıtları alıyoruz ve yayınlıyoruz
-        for message in client.chat_completion(
-            messages=messages,
-            max_tokens=max_tokens,
-            stream=True,
-            temperature=temperature,
-            top_p=top_p,
-        ):
-            token = message.choices[0].delta.get('content', '')
-            response += token
-            yield response
-    except Exception as e:
-        # Hata durumunda hata mesajını döndürüyoruz
-        yield f"Hata: {e}"
-
-# Gradio arayüzünü oluşturuyoruz
-with gr.Blocks(theme="gradio/seafoam") as demo:
-    system_message = gr.Textbox(value="You are a GPT-2 model chatbot named Simulacra. You are produced by Anezatra.", label="Sistem mesajı")
-    max_tokens = gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Maksimum yeni token sayısı")
-    temperature = gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Sıcaklık (Temperature)")
-    top_p = gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nükleus örnekleme)")
-
-    chatbot = gr.Chatbot()
-    msg = gr.Textbox(label="Mesajınızı yazın")
-    clear = gr.Button("Temizle")
-
-    def user_input(user_message, history):
-        return "", history + [[user_message, None]]
-
-    def bot_response(history):
-        last_message = history[-1][0]
-        response_gen = respond(
-            message=last_message,
-            history=history[:-1],
-            system_message=system_message.value,
-            max_tokens=max_tokens.value,
-            temperature=temperature.value,
-            top_p=top_p.value,
-        )
-        for response in response_gen:
-            history[-1][1] = response
-            yield history
-
-    msg.submit(user_input, [msg, chatbot], [msg, chatbot], queue=False).then(
-        bot_response, chatbot, chatbot
-    )
-    clear.click(lambda: None, None, chatbot, queue=False)
-
-# Uygulamayı başlatıyoruz
+from peft import AutoPeftModelForCausalLM
+from transformers import GenerationConfig
+from transformers import AutoTokenizer
+import torch
+import streamlit as st
+from streamlit_chat import message
+
+st.session_state.clicked=True
+
+def process_data_sample(example):
+
+    processed_example = "<|system|>\n You are a support chatbot who helps with user queries chatbot who always responds in the style of a professional.</s>\n<|user|>\n" + example + "</s>\n<|assistant|>\n"
+
+    return processed_example
+
+@st.cache_resource(show_spinner=True)
+def create_bot():
+
+    tokenizer = AutoTokenizer.from_pretrained("Vasanth/zephyr-support-chatbot")
+
+    model = AutoPeftModelForCausalLM.from_pretrained(
+                                                        "Vasanth/zephyr-support-chatbot",
+                                                        low_cpu_mem_usage=True,
+                                                        return_dict=True,
+                                                        torch_dtype=torch.float16,
+                                                        device_map="cuda"
+                                                    )
+
+    generation_config = GenerationConfig(
+                                            do_sample=True,
+                                            temperature=0.5,
+                                            max_new_tokens=256,
+                                            pad_token_id=tokenizer.eos_token_id
+                                        )
+    
+    return model, tokenizer, generation_config
+
+model, tokenizer, generation_config = create_bot()
+
+bot = create_bot()
+
+def infer_bot(prompt):
+    inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
+    outputs = model.generate(**inputs, generation_config=generation_config)
+    out_str = tokenizer.decode(outputs[0], skip_special_tokens=True).replace(prompt, '')
+    return out_str
+
+def display_conversation(history):
+    for i in range(len(history["assistant"])):
+        message(history["user"][i], is_user=True, key=str(i) + "_user")
+        message(history["assistant"][i],key=str(i))
+
+def main():
+
+    st.title("Support Member 📚🤖")
+    st.subheader("A bot created using Zephyr which was finetuned to possess the capabilities to be a support member")
+
+    user_input = st.text_input("Enter your query")
+
+    if "assistant" not in st.session_state:
+        st.session_state["assistant"] = ["I am ready to help you"]
+    if "user" not in st.session_state:
+        st.session_state["user"] = ["Hey there!"]
+                
+    if st.session_state.clicked:
+        if st.button("Answer"):
+
+            answer = infer_bot(user_input)
+            st.session_state["user"].append(user_input)
+            st.session_state["assistant"].append(answer)
+
+            if st.session_state["assistant"]:
+                display_conversation(st.session_state)
+
 if __name__ == "__main__":
-    demo.launch(share=True)
+    main()
+    
+