Lion (Evolved Sign Momentum) is a unique optimizer that uses the sign of the gradient to determine the update direction of the momentum. This makes Lion more memory-efficient and faster than AdamW
which tracks and store the first and second-order moments.
( params lr = 0.0001 betas = (0.9, 0.99) weight_decay = 0 optim_bits = 32 args = None min_8bit_size = 4096 percentile_clipping = 100 block_wise = True is_paged = False )
( params lr = 0.0001 betas = (0.9, 0.99) weight_decay = 0 optim_bits = 32 args = None min_8bit_size = 4096 percentile_clipping = 100 block_wise = True is_paged = False )
Parameters
torch.tensor
) —
The input parameters to optimize. float
, defaults to 1e-4) —
The learning rate. tuple(float, float)
, defaults to (0.9, 0.999)) —
The beta values are the decay rates of the first and second-order moment of the optimizer. float
, defaults to 0) —
The weight decay value for the optimizer. int
, defaults to 32) —
The number of bits of the optimizer state. dict
, defaults to None
) —
A dictionary with additional arguments. int
, defaults to 4096) —
The minimum number of elements of the parameter tensors for 8-bit optimization. int
, defaults to 100) —
Adapts clipping threshold automatically by tracking the last 100 gradient norms and clipping the gradient at a certain percentile to improve stability. bool
, defaults to True
) —
Whether to independently quantize each block of tensors to reduce outlier effects and improve stability. bool
, defaults to False
) —
Whether the optimizer is a paged optimizer or not. Base Lion optimizer.
( params lr = 0.0001 betas = (0.9, 0.99) weight_decay = 0 args = None min_8bit_size = 4096 percentile_clipping = 100 block_wise = True is_paged = False )
( params lr = 0.0001 betas = (0.9, 0.99) weight_decay = 0 args = None min_8bit_size = 4096 percentile_clipping = 100 block_wise = True is_paged = False )
Parameters
torch.tensor
) —
The input parameters to optimize. float
, defaults to 1e-4) —
The learning rate. tuple(float, float)
, defaults to (0.9, 0.999)) —
The beta values are the decay rates of the first and second-order moment of the optimizer. float
, defaults to 0) —
The weight decay value for the optimizer. dict
, defaults to None
) —
A dictionary with additional arguments. int
, defaults to 4096) —
The minimum number of elements of the parameter tensors for 8-bit optimization. int
, defaults to 100) —
Adapts clipping threshold automatically by tracking the last 100 gradient norms and clipping the gradient at a certain percentile to improve stability. bool
, defaults to True
) —
Whether to independently quantize each block of tensors to reduce outlier effects and improve stability. bool
, defaults to False
) —
Whether the optimizer is a paged optimizer or not. 8-bit Lion optimizer.
( params lr = 0.0001 betas = (0.9, 0.99) weight_decay = 0 args = None min_8bit_size = 4096 percentile_clipping = 100 block_wise = True is_paged = False )
( params lr = 0.0001 betas = (0.9, 0.99) weight_decay = 0 args = None min_8bit_size = 4096 percentile_clipping = 100 block_wise = True is_paged = False )
Parameters
torch.tensor
) —
The input parameters to optimize. float
, defaults to 1e-4) —
The learning rate. tuple(float, float)
, defaults to (0.9, 0.999)) —
The beta values are the decay rates of the first and second-order moment of the optimizer. float
, defaults to 0) —
The weight decay value for the optimizer. dict
, defaults to None
) —
A dictionary with additional arguments. int
, defaults to 4096) —
The minimum number of elements of the parameter tensors for 8-bit optimization. int
, defaults to 100) —
Adapts clipping threshold automatically by tracking the last 100 gradient norms and clipping the gradient at a certain percentile to improve stability. bool
, defaults to True
) —
Whether to independently quantize each block of tensors to reduce outlier effects and improve stability. bool
, defaults to False
) —
Whether the optimizer is a paged optimizer or not. 32-bit Lion optimizer.
( params lr = 0.0001 betas = (0.9, 0.99) weight_decay = 0 optim_bits = 32 args = None min_8bit_size = 4096 percentile_clipping = 100 block_wise = True )
( params lr = 0.0001 betas = (0.9, 0.99) weight_decay = 0 optim_bits = 32 args = None min_8bit_size = 4096 percentile_clipping = 100 block_wise = True )
Parameters
torch.tensor
) —
The input parameters to optimize. float
, defaults to 1e-4) —
The learning rate. tuple(float, float)
, defaults to (0.9, 0.999)) —
The beta values are the decay rates of the first and second-order moment of the optimizer. float
, defaults to 0) —
The weight decay value for the optimizer. int
, defaults to 32) —
The number of bits of the optimizer state. dict
, defaults to None
) —
A dictionary with additional arguments. int
, defaults to 4096) —
The minimum number of elements of the parameter tensors for 8-bit optimization. int
, defaults to 100) —
Adapts clipping threshold automatically by tracking the last 100 gradient norms and clipping the gradient at a certain percentile to improve stability. bool
, defaults to True
) —
Whether to independently quantize each block of tensors to reduce outlier effects and improve stability. Paged Lion optimizer.
( params lr = 0.0001 betas = (0.9, 0.99) weight_decay = 0 args = None min_8bit_size = 4096 percentile_clipping = 100 block_wise = True )
( params lr = 0.0001 betas = (0.9, 0.99) weight_decay = 0 args = None min_8bit_size = 4096 percentile_clipping = 100 block_wise = True )
Parameters
torch.tensor
) —
The input parameters to optimize. float
, defaults to 1e-4) —
The learning rate. tuple(float, float)
, defaults to (0.9, 0.999)) —
The beta values are the decay rates of the first and second-order moment of the optimizer. float
, defaults to 0) —
The weight decay value for the optimizer. int
, defaults to 32) —
The number of bits of the optimizer state. dict
, defaults to None
) —
A dictionary with additional arguments. int
, defaults to 4096) —
The minimum number of elements of the parameter tensors for 8-bit optimization. int
, defaults to 100) —
Adapts clipping threshold automatically by tracking the last 100 gradient norms and clipping the gradient at a certain percentile to improve stability. bool
, defaults to True
) —
Whether to independently quantize each block of tensors to reduce outlier effects and improve stability. Paged 8-bit Lion optimizer.
( params lr = 0.0001 betas = (0.9, 0.99) weight_decay = 0 args = None min_8bit_size = 4096 percentile_clipping = 100 block_wise = True )
( params lr = 0.0001 betas = (0.9, 0.99) weight_decay = 0 args = None min_8bit_size = 4096 percentile_clipping = 100 block_wise = True )
Parameters
torch.tensor
) —
The input parameters to optimize. float
, defaults to 1e-4) —
The learning rate. tuple(float, float)
, defaults to (0.9, 0.999)) —
The beta values are the decay rates of the first and second-order moment of the optimizer. float
, defaults to 0) —
The weight decay value for the optimizer. int
, defaults to 32) —
The number of bits of the optimizer state. dict
, defaults to None
) —
A dictionary with additional arguments. int
, defaults to 4096) —
The minimum number of elements of the parameter tensors for 8-bit optimization. int
, defaults to 100) —
Adapts clipping threshold automatically by tracking the last 100 gradient norms and clipping the gradient at a certain percentile to improve stability. bool
, defaults to True
) —
Whether to independently quantize each block of tensors to reduce outlier effects and improve stability. Paged 32-bit Lion optimizer.