adding new model

main.py

@@ -35,8 +35,8 @@ class PredictRequest(BaseModel):
 #     device_map="auto",
 #     # attn_implementation="flash_attention_2",
 # )
-
-checkpoint = "Qwen/Qwen2.5-VL-7B-Instruct"
+# checkpoint = "Qwen/Qwen2.5-VL-7B-Instruct"
+checkpoint = "Qwen/Qwen2-VL-72B-Instruct-GPTQ-Int4"
 min_pixels = 256 * 28 * 28
 max_pixels = 1280 * 28 * 28
 processor = AutoProcessor.from_pretrained(