Skip to content

Open-Sora: Democratizing Efficient Video Production for All

Python

add local_files_only

Zangwei Zheng8bf34f4418ee
diff --git a/configs/opensora-v1-1/train/stage1-2.py b/configs/opensora-v1-1/train/stage1-2.py
index 6e6f7dc..a119e37 100644
--- a/configs/opensora-v1-1/train/stage1-2.py
+++ b/configs/opensora-v1-1/train/stage1-2.py
@@ -47,12 +47,14 @@ vae = dict(
type="VideoAutoencoderKL",
from_pretrained="stabilityai/sd-vae-ft-ema",
micro_batch_size=4,
+ local_files_only=True,
)
text_encoder = dict(
type="t5",
from_pretrained="DeepFloyd/t5-v1_1-xxl",
model_max_length=200,
shardformer=True,
+ local_files_only=True,
)
scheduler = dict(
type="iddpm-speed",
diff --git a/opensora/models/text_encoder/t5.py b/opensora/models/text_encoder/t5.py
index 7364eb9..aaf2ecf 100644
--- a/opensora/models/text_encoder/t5.py
+++ b/opensora/models/text_encoder/t5.py
@@ -46,6 +46,7 @@ class T5Embedder:
torch_dtype=None,
use_offload_folder=None,
model_max_length=120,
+ local_files_only=False,
):
self.device = torch.device(device)
self.torch_dtype = torch_dtype or torch.bfloat16
@@ -99,8 +100,17 @@ class T5Embedder:
self.hf_token = hf_token
assert from_pretrained in self.available_models
- self.tokenizer = AutoTokenizer.from_pretrained(from_pretrained, cache_dir=cache_dir)
- self.model = T5EncoderModel.from_pretrained(from_pretrained, cache_dir=cache_dir, **t5_model_kwargs).eval()
+ self.tokenizer = AutoTokenizer.from_pretrained(
+ from_pretrained,
+ cache_dir=cache_dir,
+ local_files_only=local_files_only,
+ )
+ self.model = T5EncoderModel.from_pretrained(
+ from_pretrained,
+ cache_dir=cache_dir,
+ local_files_only=local_files_only,
+ **t5_model_kwargs,
+ ).eval()
self.model_max_length = model_max_length
def get_text_embeddings(self, texts):
@@ -134,6 +144,7 @@ class T5Encoder:
dtype=torch.float,
cache_dir=None,
shardformer=False,
+ local_files_only=False,
):
assert from_pretrained is not None, "Please specify the path to the T5 model"
@@ -143,6 +154,7 @@ class T5Encoder:
from_pretrained=from_pretrained,
cache_dir=cache_dir,
model_max_length=model_max_length,
+ local_files_only=local_files_only,
)
self.t5.model.to(dtype=dtype)
self.y_embedder = None
diff --git a/opensora/models/vae/vae.py b/opensora/models/vae/vae.py
index 9c79c5c..273bcfd 100644
--- a/opensora/models/vae/vae.py
+++ b/opensora/models/vae/vae.py
@@ -8,9 +8,11 @@ from opensora.registry import MODELS
@MODELS.register_module()
class VideoAutoencoderKL(nn.Module):
- def __init__(self, from_pretrained=None, micro_batch_size=None, cache_dir=None):
+ def __init__(self, from_pretrained=None, micro_batch_size=None, cache_dir=None, local_files_only=False):
super().__init__()
- self.module = AutoencoderKL.from_pretrained(from_pretrained, cache_dir=cache_dir)
+ self.module = AutoencoderKL.from_pretrained(
+ from_pretrained, cache_dir=cache_dir, local_files_only=local_files_only
+ )
self.out_channels = self.module.config.latent_channels
self.patch_size = (1, 8, 8)
self.micro_batch_size = micro_batch_size
@@ -70,9 +72,11 @@ class VideoAutoencoderKL(nn.Module):
@MODELS.register_module()
class VideoAutoencoderKLTemporalDecoder(nn.Module):
- def __init__(self, from_pretrained=None, cache_dir=None):
+ def __init__(self, from_pretrained=None, cache_dir=None, local_files_only=False):
super().__init__()
- self.module = AutoencoderKLTemporalDecoder.from_pretrained(from_pretrained, cache_dir=cache_dir)
+ self.module = AutoencoderKLTemporalDecoder.from_pretrained(
+ from_pretrained, cache_dir=cache_dir, local_files_only=local_files_only
+ )
self.out_channels = self.module.config.latent_channels
self.patch_size = (1, 8, 8)