Create app.py

app.py

@@ -0,0 +1,60 @@
+import os
+import subprocess
+import sys
+
+try:
+    import llama_cpp
+except ImportError:
+    print("Installing pre-built llama-cpp-python...")
+    subprocess.check_call([
+        sys.executable, "-m", "pip", "install", 
+        "llama-cpp-python", 
+        "--extra-index-url", "https://abetlen.github.io/llama-cpp-python/whl/cpu"
+    ])
+
+import gradio as gr
+from llama_cpp import Llama
+from huggingface_hub import hf_hub_download
+
+# モデルの設定
+model_id = "kawasumi/Tema_Q-R4.2-GGUF"
+model_file = "Tema_Q-R4.2-Q2_K.gguf"
+
+print("Downloading model...")
+model_path = hf_hub_download(repo_id=model_id, filename=model_file)
+
+print(f"Loading model from {model_path}...")
+# メモリ節約のための設定を強化
+llm = Llama(
+    model_path=model_path,
+    n_ctx=1024,      
+    n_threads=2,     
+    use_mmap=False,  
+    n_batch=128,     
+)
+print("Model loaded.")
+
+def chat_response(message, history):
+
+    prompt = f"<start_of_turn>user\n{message}<end_of_turn>\n<start_of_turn>model\n"
+    
+    output = llm(
+        prompt,
+        max_tokens=512,
+        stop=["<end_of_turn>", "user"],
+        stream=True
+    )
+    
+    response = ""
+    for chunk in output:
+        text = chunk["choices"][0]["text"]
+        response += text
+        yield response
+
+demo = gr.ChatInterface(
+    fn=chat_response, 
+    title="Tema_Q-R4.2 Chat",
+)
+
+if __name__ == "__main__":
+    demo.launch(server_name="0.0.0.0", server_port=7860)