diff options
Diffstat (limited to 'bitsandbytes/optim')
| -rw-r--r-- | bitsandbytes/optim/adagrad.py | 3 | ||||
| -rw-r--r-- | bitsandbytes/optim/adamw.py | 2 | ||||
| -rw-r--r-- | bitsandbytes/optim/lars.py | 6 | ||||
| -rw-r--r-- | bitsandbytes/optim/rmsprop.py | 11 | ||||
| -rw-r--r-- | bitsandbytes/optim/sgd.py | 6 |
5 files changed, 11 insertions, 17 deletions
diff --git a/bitsandbytes/optim/adagrad.py b/bitsandbytes/optim/adagrad.py index 84ade3c..4f51250 100644 --- a/bitsandbytes/optim/adagrad.py +++ b/bitsandbytes/optim/adagrad.py @@ -2,11 +2,8 @@ # # This source code is licensed under the MIT license found in the # LICENSE file in the root directory of this source tree. -import torch from bitsandbytes.optim.optimizer import Optimizer1State -torch.optim.Adagrad - class Adagrad(Optimizer1State): def __init__(self, params, lr=1e-2, lr_decay=0, weight_decay=0, initial_accumulator_value=0, eps=1e-10, optim_bits=32, args=None, min_8bit_size=4096, percentile_clipping=100, block_wise=True): diff --git a/bitsandbytes/optim/adamw.py b/bitsandbytes/optim/adamw.py index 7761f3b..c4f0355 100644 --- a/bitsandbytes/optim/adamw.py +++ b/bitsandbytes/optim/adamw.py @@ -2,9 +2,7 @@ # # This source code is licensed under the MIT license found in the # LICENSE file in the root directory of this source tree. -import torch from bitsandbytes.optim.optimizer import Optimizer2State -import bitsandbytes.functional as F class AdamW(Optimizer2State): def __init__(self, params, lr=1e-3, betas=(0.9, 0.999), eps=1e-8, diff --git a/bitsandbytes/optim/lars.py b/bitsandbytes/optim/lars.py index 40dede7..912520d 100644 --- a/bitsandbytes/optim/lars.py +++ b/bitsandbytes/optim/lars.py @@ -12,7 +12,7 @@ class LARS(Optimizer1State): weight_decay=0, nesterov=False, optim_bits=32, args=None, min_8bit_size=4096, percentile_clipping=100, max_unorm=0.02): if momentum == 0: - raise NotImplementError(f'LARS without momentum is not supported!') + raise NotImplementedError(f'LARS without momentum is not supported!') super(LARS, self).__init__('lars', params, lr, (momentum, dampening), 0.0, weight_decay, optim_bits, args, min_8bit_size, percentile_clipping, max_unorm=max_unorm, block_wise=False) @@ -21,7 +21,7 @@ class LARS8bit(Optimizer1State): weight_decay=0, nesterov=False, args=None, min_8bit_size=4096, percentile_clipping=100, max_unorm=0.02): if momentum == 0: - raise NotImplementError(f'LARS without momentum is not supported!') + raise NotImplementedError(f'LARS without momentum is not supported!') super(LARS8bit, self).__init__('lars', params, lr, (momentum, dampening), 0.0, weight_decay, 8, args, min_8bit_size, percentile_clipping, max_unorm=max_unorm, block_wise=False) @@ -30,7 +30,7 @@ class LARS32bit(Optimizer1State): weight_decay=0, nesterov=False, args=None, min_8bit_size=4096, percentile_clipping=100, max_unorm=0.02): if momentum == 0: - raise NotImplementError(f'LARS without momentum is not supported!') + raise NotImplementedError(f'LARS without momentum is not supported!') super(LARS32bit, self).__init__('lars', params, lr, (momentum, dampening), 0.0, weight_decay, 32, args, min_8bit_size, percentile_clipping, max_unorm=max_unorm, block_wise=False) diff --git a/bitsandbytes/optim/rmsprop.py b/bitsandbytes/optim/rmsprop.py index 99b718e..7909d5d 100644 --- a/bitsandbytes/optim/rmsprop.py +++ b/bitsandbytes/optim/rmsprop.py @@ -2,16 +2,15 @@ # # This source code is licensed under the MIT license found in the # LICENSE file in the root directory of this source tree. -import torch from bitsandbytes.optim.optimizer import Optimizer1State class RMSprop(Optimizer1State): def __init__(self, params, lr=1e-2, alpha=0.99, eps=1e-8, weight_decay=0, momentum=0, centered=False, optim_bits=32, args=None, min_8bit_size=4096, percentile_clipping=100, block_wise=True): if alpha == 0: - raise NotImplementError(f'RMSprop with alpha==0.0 is not supported!') + raise NotImplementedError(f'RMSprop with alpha==0.0 is not supported!') if centered: - raise NotImplementError(f'Centered RMSprop is not supported!') + raise NotImplementedError(f'Centered RMSprop is not supported!') super(RMSprop, self).__init__('rmsprop', params, lr, (alpha, momentum), eps, weight_decay, optim_bits, args, min_8bit_size, percentile_clipping, block_wise) @@ -19,9 +18,9 @@ class RMSprop8bit(Optimizer1State): def __init__(self, params, lr=1e-2, alpha=0.99, eps=1e-8, weight_decay=0, momentum=0, centered=False, args=None, min_8bit_size=4096, percentile_clipping=100, block_wise=True): if alpha == 0: - raise NotImplementError(f'RMSprop with alpha==0.0 is not supported!') + raise NotImplementedError(f'RMSprop with alpha==0.0 is not supported!') if centered: - raise NotImplementError(f'Centered RMSprop is not supported!') + raise NotImplementedError(f'Centered RMSprop is not supported!') super(RMSprop8bit, self).__init__('rmsprop', params, lr, (alpha, momentum), eps, weight_decay, 8, args, min_8bit_size, percentile_clipping, block_wise) @@ -30,7 +29,7 @@ class RMSprop32bit(Optimizer1State): min_8bit_size=4096, percentile_clipping=100, block_wise=True): if alpha == 0: - raise NotImplementError(f'RMSprop with alpha==0.0 is not supported!') + raise NotImplementedError(f'RMSprop with alpha==0.0 is not supported!') if centered: raise NotImplementError(f'Centered RMSprop is not supported!') super(RMSprop32bit, self).__init__('rmsprop', params, lr, (alpha, momentum), eps, diff --git a/bitsandbytes/optim/sgd.py b/bitsandbytes/optim/sgd.py index 926d804..0529879 100644 --- a/bitsandbytes/optim/sgd.py +++ b/bitsandbytes/optim/sgd.py @@ -9,7 +9,7 @@ class SGD(Optimizer1State): weight_decay=0, nesterov=False, optim_bits=32, args=None, min_8bit_size=4096, percentile_clipping=100, block_wise=True): if momentum == 0: - raise NotImplementError(f'SGD without momentum is not supported!') + raise NotImplementedError(f'SGD without momentum is not supported!') super(SGD, self).__init__('momentum', params, lr, (momentum, dampening), 0.0, weight_decay, optim_bits, args, min_8bit_size, percentile_clipping, block_wise) @@ -18,7 +18,7 @@ class SGD8bit(Optimizer1State): weight_decay=0, nesterov=False, args=None, min_8bit_size=4096, percentile_clipping=100, block_wise=True): if momentum == 0: - raise NotImplementError(f'SGD without momentum is not supported!') + raise NotImplementedError(f'SGD without momentum is not supported!') super(SGD8bit, self).__init__('momentum', params, lr, (momentum, dampening), 0.0, weight_decay, 8, args, min_8bit_size, percentile_clipping, block_wise) @@ -27,6 +27,6 @@ class SGD32bit(Optimizer1State): weight_decay=0, nesterov=False, args=None, min_8bit_size=4096, percentile_clipping=100, block_wise=True): if momentum == 0: - raise NotImplementError(f'SGD without momentum is not supported!') + raise NotImplementedError(f'SGD without momentum is not supported!') super(SGD32bit, self).__init__('momentum', params, lr, (momentum, dampening), 0.0, weight_decay, 32, args, min_8bit_size, percentile_clipping, block_wise) |