Source code for trainer.optimization.adafactor

# Copyright 2021 Facebook and Zilliz. All rights reserved.
#
# 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.


import torch
import math
from torch.optim import Optimizer


[docs]class Adafactor(Optimizer): """ AdaFactor pytorch implementation as introduced in `Adafactor: Adaptive Learning Rates with Sublinear Memory Cost` https://arxiv.org/abs/1804.04235. Args: params (Iterable[nn.parameter.Parameter]): Iterable of parameters to optimize or dictionaries defining parameter groups. lr (float, optional): The external learning rate. eps (Tuple[float, float], optional): Regularization constants for square gradient and parameter scale respectively. clip_threshold (float, optional): Threshold of root mean square of final gradient update. decay_rate (float, optional): Coefficient used to compute running averages of square. beta1 (float, optional): Coefficient used for computing running averages of gradient. weight_decay (float, optional): Weight decay (L2 penalty). scale_parameter (bool, optional): If True, learning rate is scaled by root mean square. relative_step (bool, optional): If True, time-dependent learning rate is computed instead of external learning rate. warmup_init (bool, optional): Time-dependent learning rate computation depends on whether warm-up initialization is being used. """ def __init__( self, params, lr=None, eps=(1e-30, 1e-3), clip_threshold=1.0, decay_rate=-0.8, beta1=None, weight_decay=0.0, scale_parameter=True, relative_step=True, warmup_init=False, ): if lr is not None and relative_step: raise ValueError("Cannot combine manual `lr` and `relative_step=True` options") if warmup_init and not relative_step: raise ValueError("`warmup_init=True` requires `relative_step=True`") defaults = dict( lr=lr, eps=eps, clip_threshold=clip_threshold, decay_rate=decay_rate, beta1=beta1, weight_decay=weight_decay, scale_parameter=scale_parameter, relative_step=relative_step, warmup_init=warmup_init, ) super().__init__(params, defaults) @staticmethod def _get_lr(param_group, param_state): rel_step_sz = param_group["lr"] if param_group["relative_step"]: min_step = 1e-6 * param_state["step"] if param_group["warmup_init"] else 1e-2 rel_step_sz = min(min_step, 1.0 / math.sqrt(param_state["step"])) param_scale = 1.0 if param_group["scale_parameter"]: param_scale = max(param_group["eps"][1], param_state["RMS"]) return param_scale * rel_step_sz @staticmethod def _get_options(param_group, param_shape): factored = len(param_shape) >= 2 use_first_moment = param_group["beta1"] is not None return factored, use_first_moment @staticmethod def _rms(tensor): return tensor.norm(2) / (tensor.numel() ** 0.5) @staticmethod def _approx_sq_grad(exp_avg_sq_row, exp_avg_sq_col): r_factor = (exp_avg_sq_row / exp_avg_sq_row.mean(dim=-1, keepdim=True)).rsqrt_() c_factor = exp_avg_sq_col.rsqrt() return torch.mm(r_factor.unsqueeze(-1), c_factor.unsqueeze(0))
[docs] def step(self, closure=None): """ Performs a single optimization step Args: closure(callable, optional): A closure that reevaluates the model and returns the loss. """ loss = None if closure is not None: loss = closure() for group in self.param_groups: for p in group["params"]: if p.grad is None: continue grad = p.grad.data if grad.dtype in {torch.float16, torch.bfloat16}: grad = grad.float() if grad.is_sparse: raise RuntimeError("Adafactor does not support sparse gradients.") state = self.state[p] grad_shape = grad.shape factored, use_first_moment = self._get_options(group, grad_shape) # State Initialization if len(state) == 0: state["step"] = 0 if use_first_moment: # Exponential moving average of gradient values state["exp_avg"] = torch.zeros_like(grad) if factored: state["exp_avg_sq_row"] = torch.zeros(grad_shape[:-1]).to(grad) state["exp_avg_sq_col"] = torch.zeros(grad_shape[:-2] + grad_shape[-1:]).to(grad) else: state["exp_avg_sq"] = torch.zeros_like(grad) state["RMS"] = 0 else: if use_first_moment: state["exp_avg"] = state["exp_avg"].to(grad) if factored: state["exp_avg_sq_row"] = state["exp_avg_sq_row"].to(grad) state["exp_avg_sq_col"] = state["exp_avg_sq_col"].to(grad) else: state["exp_avg_sq"] = state["exp_avg_sq"].to(grad) p_data_fp32 = p.data if p.data.dtype in {torch.float16, torch.bfloat16}: p_data_fp32 = p_data_fp32.float() state["step"] += 1 state["RMS"] = self._rms(p_data_fp32) lr = self._get_lr(group, state) beta2t = 1.0 - math.pow(state["step"], group["decay_rate"]) update = (grad ** 2) + group["eps"][0] if factored: exp_avg_sq_row = state["exp_avg_sq_row"] exp_avg_sq_col = state["exp_avg_sq_col"] exp_avg_sq_row.mul_(beta2t).add_(update.mean(dim=-1), alpha=(1.0 - beta2t)) exp_avg_sq_col.mul_(beta2t).add_(update.mean(dim=-2), alpha=(1.0 - beta2t)) # Approximation of exponential moving average of square of gradient update = self._approx_sq_grad(exp_avg_sq_row, exp_avg_sq_col) update.mul_(grad) else: exp_avg_sq = state["exp_avg_sq"] exp_avg_sq.mul_(beta2t).add_(update, alpha=(1.0 - beta2t)) update = exp_avg_sq.rsqrt().mul_(grad) update.div_((self._rms(update) / group["clip_threshold"]).clamp_(min=1.0)) update.mul_(lr) if use_first_moment: exp_avg = state["exp_avg"] exp_avg.mul_(group["beta1"]).add_(update, alpha=(1 - group["beta1"])) update = exp_avg if group["weight_decay"] != 0: p_data_fp32.add_(p_data_fp32, alpha=(-group["weight_decay"] * lr)) p_data_fp32.add_(-update) if p.data.dtype in {torch.float16, torch.bfloat16}: p.data.copy_(p_data_fp32) return loss