Create config.py

config.py

@@ -0,0 +1,33 @@
+# config.py
+import os
+from dotenv import load_dotenv
+import torch
+
+load_dotenv() # Load variables from .env file for local development
+
+# Model Configuration
+TEXT_MODEL_NAME = os.getenv("TEXT_MODEL_NAME", "google/flan-t5-base")
+IMAGE_MODEL_NAME = os.getenv("IMAGE_MODEL_NAME", "runwayml/stable-diffusion-v1-5")
+VIDEO_MODEL_NAME = os.getenv("VIDEO_MODEL_NAME", "zeroscope_v2_576w")
+
+# API Security
+ALLOWED_API_KEYS_STR = os.getenv("ALLOWED_API_KEYS", "")
+ALLOWED_API_KEYS = set(ALLOWED_API_KEYS_STR.split(',')) if ALLOWED_API_KEYS_STR else set()
+
+# Inference Device
+DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
+# For Spaces free tier CPU or basic T4 GPU, FP16 might be beneficial if GPU is available
+DTYPE = torch.float16 if DEVICE == "cuda" else torch.float32
+
+# Rate Limiting (Example: 10 requests per minute)
+RATE_LIMIT = "10/minute"
+
+print(f"--- Configuration ---")
+print(f"Text Model: {TEXT_MODEL_NAME}")
+print(f"Image Model: {IMAGE_MODEL_NAME}")
+print(f"Video Model: {VIDEO_MODEL_NAME}")
+print(f"Device: {DEVICE}")
+print(f"Dtype: {DTYPE}")
+print(f"API Keys Loaded: {'Yes' if ALLOWED_API_KEYS else 'No (WARNING: Open Access!)'}")
+print(f"Rate Limit: {RATE_LIMIT}")
+print(f"---------------------")