8-bit optimizers reduce the memory footprint of 32-bit optimizers without any performance degradation which means you can train large models with many parameters faster. At the core of 8-bit optimizers is block-wise quantization which enables quantization accuracy, computational efficiency, and stability.
bitsandbytes provides 8-bit optimizers through the base Optimizer8bit
class, and additionally provides Optimizer2State
and Optimizer1State
for 2-state (for example, Adam
) and 1-state (for example, Adagrad
) optimizers respectively. To provide custom optimizer hyperparameters, use the GlobalOptimManager
class to configure the optimizer.
( params defaults optim_bits = 32 is_paged = False )
( params defaults optim_bits = 32 is_paged = False )
Base 8-bit optimizer class.
( optimizer_name params lr = 0.001 betas = (0.9, 0.999) eps = 1e-08 weight_decay = 0.0 optim_bits = 32 args = None min_8bit_size = 4096 percentile_clipping = 100 block_wise = True max_unorm = 0.0 skip_zeros = False is_paged = False )
( optimizer_name params lr = 0.001 betas = (0.9, 0.999) eps = 1e-08 weight_decay = 0.0 optim_bits = 32 args = None min_8bit_size = 4096 percentile_clipping = 100 block_wise = True max_unorm = 0.0 skip_zeros = False is_paged = False )
Parameters
str
) —
The name of the optimizer. torch.tensor
) —
The input parameters to optimize. float
, defaults to 1e-3) —
The learning rate. tuple
, defaults to (0.9, 0.999)) —
The beta values for the optimizer. float
, defaults to 1e-8) —
The epsilon value for the optimizer. float
, defaults to 0.0) —
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. float
, defaults to 0.0) —
The maximum value to normalize each block with. bool
, defaults to False
) —
Whether to skip zero values for sparse gradients and models to ensure correct updates. bool
, defaults to False
) —
Whether the optimizer is a paged optimizer or not. Base 2-state update optimizer class.
( optimizer_name params lr = 0.001 betas = (0.9, 0.0) eps = 1e-08 weight_decay = 0.0 optim_bits = 32 args = None min_8bit_size = 4096 percentile_clipping = 100 block_wise = True max_unorm = 0.0 skip_zeros = False is_paged = False )
( optimizer_name params lr = 0.001 betas = (0.9, 0.0) eps = 1e-08 weight_decay = 0.0 optim_bits = 32 args = None min_8bit_size = 4096 percentile_clipping = 100 block_wise = True max_unorm = 0.0 skip_zeros = False is_paged = False )
Parameters
str
) —
The name of the optimizer. torch.tensor
) —
The input parameters to optimize. float
, defaults to 1e-3) —
The learning rate. tuple
, defaults to (0.9, 0.0)) —
The beta values for the optimizer. float
, defaults to 1e-8) —
The epsilon value for the optimizer. float
, defaults to 0.0) —
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. float
, defaults to 0.0) —
The maximum value to normalize each block with. bool
, defaults to False
) —
Whether to skip zero values for sparse gradients and models to ensure correct updates. bool
, defaults to False
) —
Whether the optimizer is a paged optimizer or not. Base 1-state update optimizer class.
A global optimizer manager for enabling custom optimizer configs.
( parameters key = None value = None key_value_dict = None )
Override initial optimizer config with specific hyperparameters.
The key-values of the optimizer config for the input parameters are overridden
This can be both, optimizer parameters like betas
or lr
, or it can be
8-bit specific parameters like optim_bits
or percentile_clipping
.
Example:
import torch
import bitsandbytes as bnb
mng = bnb.optim.GlobalOptimManager.get_instance()
model = MyModel()
mng.register_parameters(model.parameters()) # 1. register parameters while still on CPU
model = model.cuda()
# use 8-bit optimizer states for all parameters
adam = bnb.optim.Adam(model.parameters(), lr=0.001, optim_bits=8)
# 2. override: the parameter model.fc1.weight now uses 32-bit Adam
mng.override_config(model.fc1.weight, 'optim_bits', 32)