Numpy API

PaddlePaddle API

safetensors.paddle.load_file

<source>

( filename: typing.Union[str, os.PathLike]device = 'cpu' ) → Dict[str, paddle.Tensor]

Parameters

• filename (str, or os.PathLike)) — The name of the file which contains the tensors

• device (Dict[str, any], optional, defaults to cpu) — The device where the tensors need to be located after load. available options are all regular paddle device locations

Returns

Dict[str, paddle.Tensor]

dictionary that contains name as key, value as paddle.Tensor

Loads a safetensors file into paddle format.

Example:

Copied

from safetensors.paddle import load_file

file_path = "./my_folder/bert.safetensors"
loaded = load_file(file_path)

safetensors.paddle.load

<source>

( data: bytesdevice: str = 'cpu' ) → Dict[str, paddle.Tensor]

Parameters

• data (bytes) — The content of a safetensors file

Returns

Dict[str, paddle.Tensor]

dictionary that contains name as key, value as paddle.Tensor on cpu

Loads a safetensors file into paddle format from pure bytes.

Example:

Copied

from safetensors.paddle import load

file_path = "./my_folder/bert.safetensors"
with open(file_path, "rb") as f:
    data = f.read()

loaded = load(data)

safetensors.paddle.save_file

<source>

( tensors: typing.Dict[str, paddle.Tensor]filename: typing.Union[str, os.PathLike]metadata: typing.Union[typing.Dict[str, str], NoneType] = None ) → None

Parameters

• tensors (Dict[str, paddle.Tensor]) — The incoming tensors. Tensors need to be contiguous and dense.

• filename (str, or os.PathLike)) — The filename we’re saving into.

• metadata (Dict[str, str], optional, defaults to None) — Optional text only metadata you might want to save in your header. For instance it can be useful to specify more about the underlying tensors. This is purely informative and does not affect tensor loading.

Returns

None

Saves a dictionnary of tensors into raw bytes in safetensors format.

Example:

Copied

from safetensors.paddle import save_file
import paddle

tensors = {"embedding": paddle.zeros((512, 1024)), "attention": paddle.zeros((256, 256))}
save_file(tensors, "model.safetensors")

safetensors.paddle.save

<source>

( tensors: typing.Dict[str, paddle.Tensor]metadata: typing.Union[typing.Dict[str, str], NoneType] = None ) → bytes

Parameters

• tensors (Dict[str, paddle.Tensor]) — The incoming tensors. Tensors need to be contiguous and dense.

• metadata (Dict[str, str], optional, defaults to None) — Optional text only metadata you might want to save in your header. For instance it can be useful to specify more about the underlying tensors. This is purely informative and does not affect tensor loading.

Returns

bytes

The raw bytes representing the format

Saves a dictionnary of tensors into raw bytes in safetensors format.

Example:

Copied

from safetensors.paddle import save
import paddle

tensors = {"embedding": paddle.zeros((512, 1024)), "attention": paddle.zeros((256, 256))}
byte_data = save(tensors)