Update config.py

config.py

@@ -3,13 +3,15 @@ import os
 from dotenv import load_dotenv
 import torch
 
-load_dotenv() # Load variables from .env file for local development
+load_dotenv()
 
 # 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")
-IMAGE_LCM_LORA_NAME = "latent-consistency/lcm-lora-sdv1-5" # Example LCM LoRA
-VIDEO_MODEL_NAME = os.getenv("VIDEO_MODEL_NAME", "zeroscope_v2_576w")
+# >>> CORRECTED VIDEO MODEL NAME <<<
+VIDEO_MODEL_NAME = os.getenv("VIDEO_MODEL_NAME", "cerspense/zeroscope_v2_576w")
+# Optional: For LCM speedup (requires 'peft' in requirements.txt)
+IMAGE_LCM_LORA_NAME = os.getenv("IMAGE_LCM_LORA_NAME", "latent-consistency/lcm-lora-sdv1-5")
 
 # API Security
 ALLOWED_API_KEYS_STR = os.getenv("ALLOWED_API_KEYS", "")
@@ -17,15 +19,16 @@ ALLOWED_API_KEYS = set(ALLOWED_API_KEYS_STR.split(',')) if ALLOWED_API_KEYS_STR
 
 # 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 Limiting
 RATE_LIMIT = "10/minute"
 
+# Print config at startup for verification
 print(f"--- Configuration ---")
 print(f"Text Model: {TEXT_MODEL_NAME}")
 print(f"Image Model: {IMAGE_MODEL_NAME}")
+print(f"Image LCM LoRA: {IMAGE_LCM_LORA_NAME if IMAGE_LCM_LORA_NAME else 'Not Configured'}") # Added LCM print
 print(f"Video Model: {VIDEO_MODEL_NAME}")
 print(f"Device: {DEVICE}")
 print(f"Dtype: {DTYPE}")