Create app.py

app.py

@@ -0,0 +1,56 @@
+import gradio as gr
+from llama_cpp import Llama
+from huggingface_hub import hf_hub_download
+import os
+
+# Persistent Storage
+PERSISTENT_DIR = "/data"
+MODEL_DIR = os.path.join(PERSISTENT_DIR, "models")
+os.makedirs(MODEL_DIR, exist_ok=True)
+
+os.environ["HF_HOME"] = os.path.join(PERSISTENT_DIR, "hf_cache")
+os.environ["HF_HUB_CACHE"] = os.path.join(PERSISTENT_DIR, "hf_cache")
+
+# أحدث Quantization متوازنة (Q4_K_M ≈ 806 MB)
+filename = "gemma-3-1b-it-Q4_K_M.gguf"
+
+model_path = hf_hub_download(
+    repo_id="unsloth/gemma-3-1b-it-GGUF",
+    filename=filename,
+    local_dir=MODEL_DIR,
+    resume_download=True,
+    force_download=False
+)
+
+print(f"✅ النموذج محمل بنجاح: {model_path}")
+
+llm = Llama(
+    model_path=model_path,
+    n_ctx=8192,
+    n_threads=4,          # زد إلى 6-8 إذا أردت
+    n_gpu_layers=0,
+    n_batch=512,
+    verbose=False,
+    chat_format="gemma"
+)
+
+def chat(message, history):
+    messages = history + [{"role": "user", "content": message}]
+    response = llm.create_chat_completion(
+        messages=messages,
+        temperature=0.7,
+        max_tokens=1024,
+        stop=["<end_of_turn>"]
+    )
+    return response["choices"][0]["message"]["content"]
+
+demo = gr.ChatInterface(
+    fn=chat,
+    title="🧠 Gemma 3 1B IT (GGUF) - Docker",
+    description="أحدث إصدار • Persistent Storage • يُحمّل مرة واحدة فقط",
+    examples=["مرحبا، كيف حالك؟", "اكتب قصة قصيرة بالعربية عن تونس"],
+    theme=gr.themes.Soft()
+)
+
+if __name__ == "__main__":
+    demo.launch(server_name="0.0.0.0", server_port=7860)