Update app.py

README.md

@@ -1,8 +1,8 @@
 ---
-title: Gradio Chatbot
-emoji: 💬
+title: Gemma 2 Baku 2B Instruct
+emoji: 🐼
 colorFrom: yellow
-colorTo: purple
+colorTo: yellow
 sdk: gradio
 sdk_version: 4.36.1
 app_file: app.py

app.py

@@ -1,63 +1,148 @@
+import os
 import gradio as gr
-from huggingface_hub import InferenceClient
+import spaces
+from transformers import AutoTokenizer, AutoModelForCausalLM, TextIteratorStreamer
+from threading import Thread
 
+
+# Set an environment variable
+HF_TOKEN = os.environ.get("HF_TOKEN", None)
+
+model_icon = "🐼"
+model_name = "Gemma 2 Baku 2B Instruct"
+model_url = "https://huggingface.co/rinna/gemma-2-baku-2b-it"
+model_id = "rinna/gemma-2-baku-2b-it"
+base_model_url = "https://huggingface.co/google/gemma-2-2b"
+base_model_name = "Gemma 2 2B"
+press_url = "https://rinna.co.jp/news/2024/07/20240725.html"
+logo_url = "https://huggingface.co/rinna/gemma-2-baku-2b/resolve/main/rinna.png"
+
+LICENSE = """
+---
+<div>
+<p>License: <a href="https://ai.google.dev/gemma/terms">Gemma Terms of Use </a><p>
+<p>This space is implemented based on <a href="https://huggingface.co/spaces/ysharma/Chat_with_Meta_llama3_8b">ysharma/Chat_with_Meta_llama3_8b</a>.</p>
+</div>
 """
-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
+
+
+DESCRIPTION = f"""
+<div>
+<p>{model_icon} <a href="{model_url}"><b>{model_name}</b> ({model_id})</a>は、<a href="https://rinna.co.jp">rinna株式会社</a>が<a href="{base_model_url}">{base_model_name}</a>に日本語継続事前学習およびインストラクションチューニングを行った大規模言語モデルです．{base_model_name}の優れたパフォーマンスを日本語に引き継いでおり、日本語のチャットにおいて高い性能を示しています。</p>
+<p>🤖 このデモでは、{model_name}とチャットを行うことが可能です。</p>
+<p>📄 モデルの詳細については、<a href="{press_url}">プレスリリース</a>、および、<a href="https://rinnakk.github.io/research/benchmarks/lm/index.html">ベンチマーク</a>をご覧ください。お問い合わせは<a href="https://rinna.co.jp/inquiry/">こちら</a>までどうぞ。</p>
+</div>
 """
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
 
+PLACEHOLDER = f"""
+<div style="padding: 30px; text-align: center; display: flex; flex-direction: column; align-items: center;">
+   <img src="{logo_url}" style="width: 80%; max-width: 550px; height: auto; opacity: 0.55;  "> 
+   <h1 style="font-size: 28px; margin-bottom: 2px; opacity: 0.55;">{model_name}</h1>
+</div>
+"""
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+css = """
+h1 {
+  text-align: center;
+  display: block;
+}
+#duplicate-button {
+  margin: auto;
+  color: white;
+  background: #1565c0;
+  border-radius: 100vh;
+}
+"""
+
+
+# Load the tokenizer and model
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto")
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
 
-    messages.append({"role": "user", "content": message})
+@spaces.GPU(duration=120)
+def chat_llama3_8b(message: str, 
+              history: list, 
+              temperature: float, 
+              max_new_tokens: int
+             ) -> str:
+    """
+    Generate a streaming response using the llama3-8b model.
+    Args:
+        message (str): The input message.
+        history (list): The conversation history used by ChatInterface.
+        temperature (float): The temperature for generating the response.
+        max_new_tokens (int): The maximum number of new tokens to generate.
+    Returns:
+        str: The generated response.
+    """
+    conversation = []
+    for user, assistant in history:
+        conversation.extend([{"role": "user", "content": user}, {"role": "assistant", "content": assistant}])
+    conversation.append({"role": "user", "content": message})
 
-    response = ""
+    # Need to set add_generation_prompt=True to ensure the model generates the response.
+    input_ids = tokenizer.apply_chat_template(conversation, add_generation_prompt=True, return_tensors="pt").to(model.device)
+    
+    streamer = TextIteratorStreamer(tokenizer, timeout=10.0, skip_prompt=True, skip_special_tokens=True)
 
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
+    generate_kwargs = dict(
+        input_ids=input_ids,
+        streamer=streamer,
+        max_new_tokens=max_new_tokens,
+        do_sample=True,
         temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
+        repetition_penalty=1.1,
+    )
+    # This will enforce greedy generation (do_sample=False) when the temperature is passed 0, avoiding the crash.             
+    if temperature == 0:
+        generate_kwargs['do_sample'] = False
+        
+    t = Thread(target=model.generate, kwargs=generate_kwargs)
+    t.start()
 
-        response += token
-        yield response
+    outputs = []
+    for text in streamer:
+        outputs.append(text)
+        yield "".join(outputs)
+        
 
-"""
-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)",
-        ),
-    ],
-)
+# Gradio block
+chatbot=gr.Chatbot(height=450, placeholder=PLACEHOLDER, label='Gradio ChatInterface')
 
+with gr.Blocks(fill_height=True, css=css) as demo:
+    
+    gr.Markdown(DESCRIPTION)
+    gr.ChatInterface(
+        fn=chat_llama3_8b,
+        chatbot=chatbot,
+        fill_height=True,
+        additional_inputs_accordion=gr.Accordion(label="⚙️ パラメータ", open=False, render=False),
+        additional_inputs=[
+            gr.Slider(minimum=0,
+                      maximum=1, 
+                      step=0.05,
+                      value=0.9, 
+                      label="生成時におけるサンプリングの温度（ランダム性）", 
+                      render=False),
+            gr.Slider(minimum=128, 
+                      maximum=4096,
+                      step=1,
+                      value=512, 
+                      label="生成したい最大のトークン数", 
+                      render=False),
+            ],
+        examples=[
+            ["日本で有名なものと言えば"],
+            ["ネコ: 「お腹が減ったニャ」\nから始まる物語を書いて"],
+            ["C言語で素数を判定するコードを書いて"],
+            ["人工知能とは何ですか"],
+        ],
+        cache_examples=False,
+                     )
+    
+    gr.Markdown(LICENSE)
+    
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()

requirements.txt

@@ -1 +1,3 @@
-huggingface_hub==0.22.2
+accelerate
+transformers
+SentencePiece