Update app.py

app.py

@@ -6,13 +6,17 @@ from threading import Thread
 from transformers import TextIteratorStreamer, AutoTokenizer, AutoModelForCausalLM
 from PIL import ImageDraw
 from torchvision.transforms.v2 import Resize
-import subprocess
+#import subprocess
 
 #subprocess.run('pip install flash-attn --no-build-isolation', env={'FLASH_ATTENTION_SKIP_CUDA_BUILD': "TRUE"}, shell=True)
 
 #subprocess.run('cp -r moondream/torch clients/python/moondream/torch')
 #subprocess.run('pip install moondream[gpu]')
 
+
+#model_id = "vikhyatk/moondream2"
+#revision = "2025-01-09"
+
 #def load_moondream():
 #    """Load Moondream model and tokenizer."""
 #    model = AutoModelForCausalLM.from_pretrained(
@@ -21,14 +25,6 @@ import subprocess
 #    tokenizer = AutoTokenizer.from_pretrained("vikhyatk/moondream2")
 #    return model, tokenizer
 
-"""Load Moondream model and tokenizer."""
-moondream = AutoModelForCausalLM.from_pretrained(
-    "vikhyatk/moondream2", trust_remote_code=True, device_map={"": "cuda"}
-)
-tokenizer = AutoTokenizer.from_pretrained("vikhyatk/moondream2")
-
-#model_id = "vikhyatk/moondream2"
-#revision = "2025-01-09"
 #tokenizer = AutoTokenizer.from_pretrained(model_id, revision=revision)
 #moondream = AutoModelForCausalLM.from_pretrained(
 #    model_id, trust_remote_code=True, revision=revision,
@@ -37,6 +33,15 @@ tokenizer = AutoTokenizer.from_pretrained("vikhyatk/moondream2")
 
 #moondream.eval()
 
+"""Load Moondream model and tokenizer."""
+tokenizer = AutoTokenizer.from_pretrained("vikhyatk/moondream2")
+moondream = AutoModelForCausalLM.from_pretrained(
+  "vikhyatk/moondream2",
+  revision="2025-01-09",
+  trust_remote_code=True,
+  device_map={"": "cuda"},
+)
+
 @spaces.GPU(durtion="150")
 def answer_questions(image_tuples, prompt_text):
     result = ""