update

app.py

@@ -1,63 +1,73 @@
 import gradio as gr
-from huggingface_hub import InferenceClient
+from transformers import AutoTokenizer, AutoModelForCausalLM
+import torch
 
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
+model_name = "elyza/Llama-3-ELYZA-JP-8B"
 
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.float16, device_map="auto")
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+SYSTEM_MESSAGE = """
+あなたは関西弁で話す生命保険の営業マンです。お客様の状況を理解し、適切な保険プランを提案することが仕事です。以下の点に注意してください：
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+1. 丁寧で親しみやすい関西弁を使う
+2. 応答は必ず250文字以内に収める
+3. お客様の基本情報（年齢、家族構成、職業など）を聞き出す
+4. 現在の経済状況や将来の不安について理解を深める
+5. お客様のニーズに合わせた保険商品を簡潔に説明する
+6. 保険の重要性と利点を分かりやすく説明する
+7. お客様からの質問に簡潔に回答する
+8. 押し売りにならないよう、お客様の意思を尊重する
 
-    messages.append({"role": "user", "content": message})
-
-    response = ""
+それでは、お客様とのやり取りを始めてください。
+"""
 
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
+def create_prompt(message, history):
+    prompt = f"システム: {SYSTEM_MESSAGE}\n\n"
+    for human, assistant in history:
+        prompt += f"人間: {human}\n助手: {assistant}\n"
+    prompt += f"人間: {message}\n助手: "
+    return prompt
 
-        response += token
-        yield response
+def respond(message, history, max_tokens, temperature, top_p):
+    prompt = create_prompt(message, history)
+    
+    input_ids = tokenizer.encode(prompt, return_tensors="pt").to(model.device)
+    
+    with torch.no_grad():
+        output = model.generate(
+            input_ids,
+            max_new_tokens=min(max_tokens, 125),  # 約250文字
+            temperature=temperature,
+            top_p=top_p,
+            do_sample=True,
+            pad_token_id=tokenizer.eos_token_id,
+        )
+    
+    generated_text = tokenizer.decode(output[0], skip_special_tokens=True)
+    assistant_response = generated_text.split("助手: ")[-1]
+    
+    truncated_response = assistant_response[:250]
+    last_punctuation = max(
+        truncated_response.rfind('。'),
+        truncated_response.rfind('！'),
+        truncated_response.rfind('？')
+    )
+    if last_punctuation != -1:
+        truncated_response = truncated_response[:last_punctuation + 1]
+    
+    return truncated_response
 
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
 demo = gr.ChatInterface(
     respond,
     additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
+        gr.Slider(minimum=1, maximum=125, value=100, 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=0.95, step=0.05, label="Top-p"),
     ],
+    title="生命保険営業顧問AI",
+    description="生命保険の営業について質問してください。",
 )
 
-
 if __name__ == "__main__":
     demo.launch()