Upload 15 files

app.py

@@ -57,15 +57,21 @@ autoencoder = load_ae(
 autoencoder.eval()
 
 print("init ovis")
-ovis_path = hf_hub_download(
-    repo_id="AIDC-AI/Ovis-Image-7B", 
+# ovis_path = hf_hub_download(
+#     repo_id="AIDC-AI/Ovis-Image-7B", 
+#     subfolder="Ovis2.5-2B", 
+#     token=hf_token,
+# )
+ovis_tokenizer = build_ovis_tokenizer(
+    "AIDC-AI/Ovis-Image-7B",
     subfolder="Ovis2.5-2B", 
     token=hf_token,
 )
-ovis_tokenizer = build_ovis_tokenizer(ovis_path)
 ovis_encoder = OvisEmbedder(
-    model_path=ovis_path,
+    model_path="AIDC-AI/Ovis-Image-7B",
     random_init=False,
+    subfolder="Ovis2.5-2B", 
+    token=hf_token,
     low_cpu_mem_usage=True,
     torch_dtype=torch.bfloat16,
 ).to(device=device, dtype=_dtype)

ovis_image/model/tokenizer.py

@@ -25,7 +25,7 @@ class OvisTokenizer:
         **hf_kwargs
     ):
         super().__init__()
-        self._tokenizer = AutoTokenizer.from_pretrained(model_path)
+        self._tokenizer = AutoTokenizer.from_pretrained(model_path, **hf_kwargs)
         self.system_prompt = "Describe the image by detailing the color, quantity, text, shape, size, texture, spatial relationships of the objects and background: "
         self.user_prompt_begin_id = 28
         self._max_length = max_length + self.user_prompt_begin_id
@@ -64,11 +64,12 @@ class OvisTokenizer:
         return self._tokenizer.decode(t, skip_special_tokens=False)
 
 
-def build_ovis_tokenizer(tokenizer_path):
+def build_ovis_tokenizer(tokenizer_path, **hf_kwargs):
     max_ovis_encoding_len = 256
     ovis_tokenizer = OvisTokenizer(
         tokenizer_path, 
         max_length=max_ovis_encoding_len,
+        **hf_kwargs,
     )
     return ovis_tokenizer