[Admin maintenance] Support new ZeroGPU hardware

Thank you so much for having shared this Space with the community on this demo. We have upgraded the ZeroGPU infra-structure to run on modern blackwell architecture.
For that, we need to upgrade your demo to support that. This PR fixes your demo to work with the new architecture. As this is something we broke on our end, we may merge this PR autonomously. If this breaks unexpectedly or brings unintended consequences, feel free to revert, modify or otherwise. Any issues you can email apolinario@huggingface.co

README.md

@@ -4,7 +4,7 @@ emoji: 🔍➡️📝
 colorFrom: gray
 colorTo: green
 sdk: gradio
-sdk_version: 4.44.0
+sdk_version: 5.50.0
 app_file: app.py
 pinned: false
 ---

app.py

@@ -12,13 +12,6 @@ if not hasattr(huggingface_hub, "HfFolder"):
             try: huggingface_hub.logout()
             except Exception: pass
     huggingface_hub.HfFolder = HfFolder
-import subprocess  # 🥲
-
-subprocess.run(
-    "pip install flash-attn --no-build-isolation",
-    env={"FLASH_ATTENTION_SKIP_CUDA_BUILD": "TRUE"},
-    shell=True,
-)
 import spaces
 import gradio as gr
 import re
@@ -36,7 +29,7 @@ os.environ["HF_HUB_ENABLE_HF_TRANSFER"] = "1"
 model = Qwen2_5_VLForConditionalGeneration.from_pretrained(
     "Qwen/Qwen2.5-VL-7B-Instruct",
     torch_dtype=torch.bfloat16,
-    attn_implementation="flash_attention_2",
+    attn_implementation="sdpa",
     device_map="auto",
 )
 processor = AutoProcessor.from_pretrained(