Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -216,11 +216,11 @@ def load_models():
|
|
216 |
|
217 |
# 加载模型
|
218 |
try:
|
219 |
-
text_encoder = LlamaModel.from_pretrained("hunyuanvideo-community/HunyuanVideo", subfolder='text_encoder', torch_dtype=dtype).to(model_device)
|
220 |
-
text_encoder_2 = CLIPTextModel.from_pretrained("hunyuanvideo-community/HunyuanVideo", subfolder='text_encoder_2', torch_dtype=dtype).to(model_device)
|
221 |
-
tokenizer = LlamaTokenizerFast.from_pretrained("hunyuanvideo-community/HunyuanVideo", subfolder='tokenizer')
|
222 |
-
tokenizer_2 = CLIPTokenizer.from_pretrained("hunyuanvideo-community/HunyuanVideo", subfolder='tokenizer_2')
|
223 |
-
vae = AutoencoderKLHunyuanVideo.from_pretrained("hunyuanvideo-community/HunyuanVideo", subfolder='vae', torch_dtype=dtype).to(model_device)
|
224 |
|
225 |
feature_extractor = SiglipImageProcessor.from_pretrained("lllyasviel/flux_redux_bfl", subfolder='feature_extractor')
|
226 |
image_encoder = SiglipVisionModel.from_pretrained("lllyasviel/flux_redux_bfl", subfolder='image_encoder', torch_dtype=dtype).to(model_device)
|
@@ -237,11 +237,11 @@ def load_models():
|
|
237 |
transformer_dtype = torch.float32
|
238 |
cpu_fallback_mode = True
|
239 |
|
240 |
-
text_encoder = LlamaModel.from_pretrained("hunyuanvideo-community/HunyuanVideo", subfolder='text_encoder', torch_dtype=dtype).to('cpu')
|
241 |
-
text_encoder_2 = CLIPTextModel.from_pretrained("hunyuanvideo-community/HunyuanVideo", subfolder='text_encoder_2', torch_dtype=dtype).to('cpu')
|
242 |
-
tokenizer = LlamaTokenizerFast.from_pretrained("hunyuanvideo-community/HunyuanVideo", subfolder='tokenizer')
|
243 |
-
tokenizer_2 = CLIPTokenizer.from_pretrained("hunyuanvideo-community/HunyuanVideo", subfolder='tokenizer_2')
|
244 |
-
vae = AutoencoderKLHunyuanVideo.from_pretrained("hunyuanvideo-community/HunyuanVideo", subfolder='vae', torch_dtype=dtype).to('cpu')
|
245 |
|
246 |
feature_extractor = SiglipImageProcessor.from_pretrained("lllyasviel/flux_redux_bfl", subfolder='feature_extractor')
|
247 |
image_encoder = SiglipVisionModel.from_pretrained("lllyasviel/flux_redux_bfl", subfolder='image_encoder', torch_dtype=dtype).to('cpu')
|
|
|
216 |
|
217 |
# 加载模型
|
218 |
try:
|
219 |
+
text_encoder = LlamaModel.from_pretrained("hunyuanvideo-community/HunyuanVideo-I2V", subfolder='text_encoder', torch_dtype=dtype).to(model_device)
|
220 |
+
text_encoder_2 = CLIPTextModel.from_pretrained("hunyuanvideo-community/HunyuanVideo-I2V", subfolder='text_encoder_2', torch_dtype=dtype).to(model_device)
|
221 |
+
tokenizer = LlamaTokenizerFast.from_pretrained("hunyuanvideo-community/HunyuanVideo-I2V", subfolder='tokenizer')
|
222 |
+
tokenizer_2 = CLIPTokenizer.from_pretrained("hunyuanvideo-community/HunyuanVideo-I2V", subfolder='tokenizer_2')
|
223 |
+
vae = AutoencoderKLHunyuanVideo.from_pretrained("hunyuanvideo-community/HunyuanVideo-I2V", subfolder='vae', torch_dtype=dtype).to(model_device)
|
224 |
|
225 |
feature_extractor = SiglipImageProcessor.from_pretrained("lllyasviel/flux_redux_bfl", subfolder='feature_extractor')
|
226 |
image_encoder = SiglipVisionModel.from_pretrained("lllyasviel/flux_redux_bfl", subfolder='image_encoder', torch_dtype=dtype).to(model_device)
|
|
|
237 |
transformer_dtype = torch.float32
|
238 |
cpu_fallback_mode = True
|
239 |
|
240 |
+
text_encoder = LlamaModel.from_pretrained("hunyuanvideo-community/HunyuanVideo-I2VHunyuanVideo-I2VHunyuanVideo-I2VHunyuanVideo-I2VHunyuanVideo-I2V", subfolder='text_encoder', torch_dtype=dtype).to('cpu')
|
241 |
+
text_encoder_2 = CLIPTextModel.from_pretrained("hunyuanvideo-community/HunyuanVideo-I2VHunyuanVideo-I2VHunyuanVideo-I2VHunyuanVideo-I2V", subfolder='text_encoder_2', torch_dtype=dtype).to('cpu')
|
242 |
+
tokenizer = LlamaTokenizerFast.from_pretrained("hunyuanvideo-community/HunyuanVideo-I2VHunyuanVideo-I2VHunyuanVideo-I2V", subfolder='tokenizer')
|
243 |
+
tokenizer_2 = CLIPTokenizer.from_pretrained("hunyuanvideo-community/HunyuanVideo-I2VHunyuanVideo-I2V", subfolder='tokenizer_2')
|
244 |
+
vae = AutoencoderKLHunyuanVideo.from_pretrained("hunyuanvideo-community/HunyuanVideo-I2V", subfolder='vae', torch_dtype=dtype).to('cpu')
|
245 |
|
246 |
feature_extractor = SiglipImageProcessor.from_pretrained("lllyasviel/flux_redux_bfl", subfolder='feature_extractor')
|
247 |
image_encoder = SiglipVisionModel.from_pretrained("lllyasviel/flux_redux_bfl", subfolder='image_encoder', torch_dtype=dtype).to('cpu')
|