Update app.py

app.py

@@ -1,50 +1,44 @@
 import gradio as gr
-from huggingface_hub import InferenceClient
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import torch
 
-client = InferenceClient("cody82/innopolis_bot_model")
+model_id = "cody82/innopolis_bot_model"
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    # Собираем историю в текст
-    full_prompt = system_message.strip() + "\n"
+# Загружаем токенизатор и модель
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForCausalLM.from_pretrained(model_id)
 
-    for user_msg, bot_msg in history:
-        if user_msg:
-            full_prompt += f"User: {user_msg}\n"
-        if bot_msg:
-            full_prompt += f"Bot: {bot_msg}\n"
+# Убираем cuda, т.к. у нас CPU
+device = torch.device("cpu")
+model = model.to(device)
 
-    full_prompt += f"User: {message}\nBot:"
+def respond(message, history):
+    history = history or []
+    # Объединяем историю и текущий ввод
+    full_input = ""
+    for user, bot in history:
+        full_input += f"User: {user}\nAssistant: {bot}\n"
+    full_input += f"User: {message}\nAssistant:"
 
-    # Генерируем ответ от модели
-    response = ""
-    for token in client.text_generation(
-        prompt=full_prompt,
-        max_new_tokens=max_tokens,
-        temperature=temperature,
-        top_p=top_p,
-        stream=True,
-    ):
-        response += token
-        yield response
+    inputs = tokenizer(full_input, return_tensors="pt").to(device)
+    outputs = model.generate(
+        **inputs,
+        max_new_tokens=200,
+        do_sample=True,
+        temperature=0.7,
+        top_p=0.95,
+        pad_token_id=tokenizer.eos_token_id
+    )
+    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
+    
+    # Извлекаем только ответ ассистента (после последнего "Assistant:")
+    if "Assistant:" in response:
+        response = response.split("Assistant:")[-1].strip()
+    
+    history.append((message, response))
+    return response, history
 
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value="You are a helpful assistant.", label="System message"),
-        gr.Slider(minimum=1, maximum=1024, value=256, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=2.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(minimum=0.1, maximum=1.0, value=0.9, step=0.05, label="Top-p"),
-    ],
-    title="Innopolis Bot",
-    description="Чат с кастомной моделью cody82/innopolis_bot_model",
-)
+chat = gr.ChatInterface(fn=respond, title="Innopolis Chatbot")
 
 if __name__ == "__main__":
-    demo.launch()
+    chat.launch()