Fix CLIP BPE vocab download: derive path without importing the broken maskclip package

app.py

@@ -42,15 +42,24 @@ def _ensure_featup():
 def _ensure_clip_bpe_vocab():
     """FeatUp's maskclip tokenizer expects `bpe_simple_vocab_16e6.txt.gz` to
     sit next to `simple_tokenizer.py`, but the file isn't shipped in the
-    FeatUp package. Download it from OpenAI's CLIP repo if missing."""
+    FeatUp package. Download it from OpenAI's CLIP repo if missing.
+
+    We must NOT `import featup.featurizers.maskclip` to find the path — that
+    package's `__init__.py` runs the tokenizer constructor which will raise
+    the very FileNotFoundError we're trying to prevent. Instead import just
+    the top-level `featup` package and derive the maskclip directory.
+    """
     import urllib.request
-    import featup.featurizers.maskclip as _maskclip_pkg
-    pkg_dir = os.path.dirname(_maskclip_pkg.__file__)
-    bpe_path = os.path.join(pkg_dir, "bpe_simple_vocab_16e6.txt.gz")
+    import featup  # top-level only, does not trigger maskclip chain
+    maskclip_dir = os.path.join(
+        os.path.dirname(featup.__file__), "featurizers", "maskclip"
+    )
+    bpe_path = os.path.join(maskclip_dir, "bpe_simple_vocab_16e6.txt.gz")
     if os.path.isfile(bpe_path):
         return
     url = "https://github.com/openai/CLIP/raw/main/clip/bpe_simple_vocab_16e6.txt.gz"
     print(f"[sab3r] Downloading CLIP BPE vocab to {bpe_path}")
+    os.makedirs(maskclip_dir, exist_ok=True)
     urllib.request.urlretrieve(url, bpe_path)