Bitsandbytes 文档
AdEMAMix
并获取增强的文档体验
开始使用
AdEMAMix
AdEMAMix 是 Adam 优化器的一种变体。
bitsandbytes 还支持分页优化器，这些优化器利用 CUDA 的统一内存，在 GPU 内存耗尽时将内存从 GPU 传输到 CPU。
AdEMAMix
class bitsandbytes.optim.AdEMAMix
< 源代码 >( params: typing.Iterable[torch.nn.parameter.Parameter] lr: float = 0.001 betas: typing.Tuple[float, float, float] = (0.9, 0.999, 0.9999) alpha: float = 5.0 t_alpha: typing.Optional[int] = None t_beta3: typing.Optional[int] = None eps: float = 1e-08 weight_decay: float = 0.01 optim_bits: typing.Literal[8, 32] = 32 min_8bit_size: int = 4096 is_paged: bool = False )
__init__
< 源代码 >( params: typing.Iterable[torch.nn.parameter.Parameter] lr: float = 0.001 betas: typing.Tuple[float, float, float] = (0.9, 0.999, 0.9999) alpha: float = 5.0 t_alpha: typing.Optional[int] = None t_beta3: typing.Optional[int] = None eps: float = 1e-08 weight_decay: float = 0.01 optim_bits: typing.Literal[8, 32] = 32 min_8bit_size: int = 4096 is_paged: bool = False )
AdEMAMix8bit
class bitsandbytes.optim.AdEMAMix8bit
< 源代码 >( params: typing.Iterable[torch.nn.parameter.Parameter] lr: float = 0.001 betas: typing.Tuple[float, float, float] = (0.9, 0.999, 0.9999) alpha: float = 5.0 t_alpha: typing.Optional[int] = None t_beta3: typing.Optional[int] = None eps: float = 1e-08 weight_decay: float = 0.01 min_8bit_size: int = 4096 is_paged: bool = False )
__init__
< 源代码 >( params: typing.Iterable[torch.nn.parameter.Parameter] lr: float = 0.001 betas: typing.Tuple[float, float, float] = (0.9, 0.999, 0.9999) alpha: float = 5.0 t_alpha: typing.Optional[int] = None t_beta3: typing.Optional[int] = None eps: float = 1e-08 weight_decay: float = 0.01 min_8bit_size: int = 4096 is_paged: bool = False )
AdEMAMix32bit
class bitsandbytes.optim.AdEMAMix32bit
< 源代码 >( params: typing.Iterable[torch.nn.parameter.Parameter] lr: float = 0.001 betas: typing.Tuple[float, float, float] = (0.9, 0.999, 0.9999) alpha: float = 5.0 t_alpha: typing.Optional[int] = None t_beta3: typing.Optional[int] = None eps: float = 1e-08 weight_decay: float = 0.01 min_8bit_size: int = 4096 is_paged: bool = False )
__init__
< 源代码 >( params: typing.Iterable[torch.nn.parameter.Parameter] lr: float = 0.001 betas: typing.Tuple[float, float, float] = (0.9, 0.999, 0.9999) alpha: float = 5.0 t_alpha: typing.Optional[int] = None t_beta3: typing.Optional[int] = None eps: float = 1e-08 weight_decay: float = 0.01 min_8bit_size: int = 4096 is_paged: bool = False )
PagedAdEMAMix
class bitsandbytes.optim.PagedAdEMAMix
< 源代码 >( params: typing.Iterable[torch.nn.parameter.Parameter] lr: float = 0.001 betas: typing.Tuple[float, float, float] = (0.9, 0.999, 0.9999) alpha: float = 5.0 t_alpha: typing.Optional[int] = None t_beta3: typing.Optional[int] = None eps: float = 1e-08 weight_decay: float = 0.01 optim_bits: typing.Literal[8, 32] = 32 min_8bit_size: int = 4096 )
__init__
< 源代码 >( params: typing.Iterable[torch.nn.parameter.Parameter] lr: float = 0.001 betas: typing.Tuple[float, float, float] = (0.9, 0.999, 0.9999) alpha: float = 5.0 t_alpha: typing.Optional[int] = None t_beta3: typing.Optional[int] = None eps: float = 1e-08 weight_decay: float = 0.01 optim_bits: typing.Literal[8, 32] = 32 min_8bit_size: int = 4096 )
PagedAdEMAMix8bit
class bitsandbytes.optim.PagedAdEMAMix8bit
< 源代码 >( params: typing.Iterable[torch.nn.parameter.Parameter] lr: float = 0.001 betas: typing.Tuple[float, float, float] = (0.9, 0.999, 0.9999) alpha: float = 5.0 t_alpha: typing.Optional[int] = None t_beta3: typing.Optional[int] = None eps: float = 1e-08 weight_decay: float = 0.01 min_8bit_size: int = 4096 )
__init__
< 源代码 >( params: typing.Iterable[torch.nn.parameter.Parameter] lr: float = 0.001 betas: typing.Tuple[float, float, float] = (0.9, 0.999, 0.9999) alpha: float = 5.0 t_alpha: typing.Optional[int] = None t_beta3: typing.Optional[int] = None eps: float = 1e-08 weight_decay: float = 0.01 min_8bit_size: int = 4096 )
PagedAdEMAMix32bit
类 bitsandbytes.optim.PagedAdEMAMix32bit
< 源码 >( params: typing.Iterable[torch.nn.parameter.Parameter] lr: float = 0.001 betas: typing.Tuple[float, float, float] = (0.9, 0.999, 0.9999) alpha: float = 5.0 t_alpha: typing.Optional[int] = None t_beta3: typing.Optional[int] = None eps: float = 1e-08 weight_decay: float = 0.01 min_8bit_size: int = 4096 )
__init__
< 源码 >( params: typing.Iterable[torch.nn.parameter.Parameter] lr: float = 0.001 betas: typing.Tuple[float, float, float] = (0.9, 0.999, 0.9999) alpha: float = 5.0 t_alpha: typing.Optional[int] = None t_beta3: typing.Optional[int] = None eps: float = 1e-08 weight_decay: float = 0.01 min_8bit_size: int = 4096 )