# Copyright 2022 Cerebras Systems.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""
Config classes of Sparsity Configs
"""
from dataclasses import dataclass
from typing import List, Literal, Union
from cerebras.modelzoo.config_manager.config_classes.base.base_config import (
BaseConfig,
required,
)
from cerebras.pytorch.sparse import map_sparsity_algorithm
[docs]@dataclass
class SparsityAlgorithmConfig(BaseConfig):
algorithm: Literal["static", "gmp", "set", "rigl"] = "static"
"""
Sparsity Algorithm to apply. ["static", "gmp", "set", "rigl"]
"""
params: dict = required
"""
Parameters such as init_method, sparsity, param_name_patterns, schedule etc
These vary based on the type of sparsity chosen.
Please see sparsity documentaion for valid parameters for each type
"""
def __init__(self, **kwargs):
for field_name, field_type in self.__annotations__.items():
if field_name in kwargs:
setattr(self, field_name, kwargs.pop(field_name))
self.params = kwargs
super().__init__()
def __post_init__(self):
# This will error out if we are not able to map the params to
# the mentioned sparsity class
map_sparsity_algorithm(algorithm=self.algorithm)
super().__post_init__()
[docs]@dataclass
class SparsityBaseConfig(BaseConfig):
groups: List[SparsityAlgorithmConfig] = required
# Custom init for sparsity, where we want to capture all fixed sparsity params as members.
# sparsity params is a dict that is populated by checking all additional params supplied to us.
# These are sparsity specific and we use signature of that sparsity to validate these.
def __init__(self, *args, **kwargs):
"Sparsity related params: see Sparsity documentation"
if len(args) > 0:
if len(kwargs) > 0:
raise ValueError(
"Expected a list of dicts or a single dict, not both"
)
self.groups = [SparsityAlgorithmConfig(**group) for group in args]
else:
self.groups = [SparsityAlgorithmConfig(**kwargs)]
super().__init__()
SparsityConfig = Union[float, SparsityBaseConfig]