adv-optm 1.2.0__py3-none-any.whl

adv_optm/optim/AdamW_adv.py

@@ -0,0 +1,374 @@
+import torch
+from typing import Optional, Callable
+
+from ..util.BF16_Stochastic_Rounding import add_stochastic_
+from ..util.Effective_Shape import _get_effective_shape
+from ..util.NNMF import _nnmf,_unnmf
+from ..util.OrthoGrad import _orthogonalize_gradient
+from ..util.One_Bit_Boolean import _pack_bools, _unpack_bools
+from ..util.Kourkoutas import KourkoutasHelper
+
+class AdamW_adv(torch.optim.Optimizer):
+    """
+    Implements an advanced AdamW algorithm.
+    This is an advanced version of AdamW with optional features like
+    low-rank factorization of optimizer states (SMMF), OrthoGrad, etc.
+
+    Args:
+        params (iterable): iterable of parameters to optimize or dicts defining
+            parameter groups
+        lr (float): learning rate (default: 1e-3)
+        betas (tuple[float, float]): coefficients used for computing running
+            averages of gradient and its square (default: (0.9, 0.999))
+        eps (float): term added to the denominator to improve
+            numerical stability (default: 1e-8)
+        weight_decay (float): weight decay (L2 penalty) (default: 0).
+        use_bias_correction (bool): whether to use bias correction for the first
+            and second moment estimates, as in the original Adam paper.
+            (default: True)
+        vector_reshape (bool): whether to reshape 1D vectors into 2D
+            matrices to apply low-rank compression (default: True).
+        stochastic_rounding (bool): whether to use stochastic
+            rounding for BF16 parameter updates (default: True).
+        use_atan2 (bool): whether to use the atan2 update rule. (default: False)
+        grams_moment (bool): whether to use Grams-style updates. (default: False)
+        cautious_mask (bool):  whether to use cautious masking to align the gradient's
+            direction with the first moment's.  (default: False)
+        orthogonal_gradient (bool): whether to use OrthoGrad.  (default: False)
+        use_AdEMAMix (bool): whether to enable the AdEMAMix feature. This adds
+            a second, slow-moving average of the momentum (`mt_slow`) which is
+            combined with the primary momentum (`mt`) to stabilize updates,
+            especially in noisy, small-batch settings. If `False`, the
+            optimizer behaves as standard AdamW. (default: False)
+        beta3_ema (float): The decay rate for the slow exponential moving average of
+            the momentum (only used when `use_AdEMAMix` is `True`). A higher
+            value (e.g., 0.9999) gives the EMA a longer memory, making it more
+            stable but slower to adapt. A lower value (e.g., 0.999) is often
+            better for shorter training runs. (default: 0.9999)
+        alpha (float): The mixing coefficient that scales the slow momentum term
+            before it is added to the fast momentum term (`update = mt + alpha * mt_slow`).
+            A higher value increases the stabilizing influence of the slow
+            momentum. (default: 5.0)
+        t_alpha (Optional[int]): The number of steps for a linear warmup of the
+            `alpha` parameter (only used when `use_AdEMAMix` is `True`). This is
+            highly recommended to prevent instability at the beginning of training,
+            as it gradually introduces the stabilizing slow momentum term. During
+            the warmup, `alpha` ramps from 0 to its target value. If `None`,
+            the scheduler is disabled. (default: None)
+        kourkoutas_beta (bool): whether to enable the layer-wise dynamic β₂ logic.
+            If `False`, the optimizer behaves as standard AdamW. (default: False)
+        beta2_min (float): The minimum value for dynamic β₂, used during periods of
+            high gradient variance ("sunspikes"). Must be less than `betas[1]`.
+            (default: 0.88)
+        ema_alpha (float): The decay rate for the Exponential Moving Average (EMA) of
+            the pooled gradient norms. Corresponds to `α` in the paper.
+            (default: 0.93)
+        tiny_spike (float): A small constant added to the denominator of the
+            "sunspike" ratio calculation to prevent division by zero. Corresponds
+            to `ε_spike` in the paper. (default: 1e-9)
+        k_warmup_steps (int): The number of initial steps during which β₂ is held
+            at a fixed beta2 value before the
+            dynamic logic activates. (default: 0)
+        k_logging (int): if > 0 and kourkoutas_beta=True, enables periodic console
+            logging of Kourkoutas-β statistics (min, max, mean of `β₂` across layers)
+            every logging steps. Useful for debugging and tuning. Set to 0 to disable
+            logging (default: 0). 
+        layer_key_fn (Optional[Callable]): A function that takes a parameter `p`
+            and returns a unique, hashable key representing its "layer" or "bucket".
+            If `None`, parameters are bucketed by their memory ID (tensor-wise).
+            (default: None)
+        nnmf_factor (bool): whether to use the factorization or disable it to use
+            the uncompressed optimizer. (default: False)
+    """
+
+    def __init__(
+        self,
+        params,
+        lr: float = 1e-3,
+        betas: tuple[float, float] = (0.9, 0.999),
+        eps: float = 1e-8,
+        weight_decay: float = 0.0,
+        use_bias_correction: bool = True,
+        vector_reshape: bool = True,
+        stochastic_rounding: bool = True,
+        use_atan2: bool = False,
+        cautious_mask: bool = False,
+        grams_moment: bool = False,
+        orthogonal_gradient: bool = False,
+        use_AdEMAMix: bool = False,
+        beta3_ema: float = 0.9999,
+        alpha: float = 5.0,
+        t_alpha: int | None = None,
+        kourkoutas_beta: bool = False,
+        beta2_min: float = 0.9,
+        ema_alpha: float = 0.95,
+        tiny_spike: float = 1e-9,
+        k_warmup_steps: int = 0,
+        k_logging: int = 0,
+        layer_key_fn: Optional[Callable] = None,
+        nnmf_factor: bool = False,
+    ):
+        if not (lr >= 0.0):
+            raise ValueError(f"Learning-rate should be >= 0.0. Got {lr}")
+        if not (0.0 <= betas[0] < 1.0 and 0.0 <= betas[1] < 1.0):
+            raise ValueError(f"Betas should be in [0.0, 1.0). Got {betas}")
+        if not (eps >= 0.0):
+            raise ValueError(f"Epsilon should be >= 0.0. Got {eps}")
+        if not (weight_decay >= 0.0):
+            raise ValueError(f"Weight-decay should be >= 0.0. Got {weight_decay}")
+        if kourkoutas_beta and not (betas[1] > beta2_min): raise ValueError(f"For Kourkoutas-β, betas[1] (as beta2_max) must be > beta2_min. Got {betas[1]} and {beta2_min}")
+
+        if cautious_mask and grams_moment:
+            print("Warning: cautious is incompatible with grams, Disabling cautious.")
+            cautious_mask = False
+
+        defaults = {
+            "lr": lr, "betas": betas, "eps": eps, "weight_decay": weight_decay,
+            "vector_reshape": vector_reshape, "use_atan2": use_atan2,
+            "orthogonal_gradient": orthogonal_gradient, "use_bias_correction": use_bias_correction,
+            "beta3_ema": beta3_ema, "alpha": alpha, "t_alpha": t_alpha,
+            "kourkoutas_beta": kourkoutas_beta, "beta2_min": beta2_min, "ema_alpha": ema_alpha,
+            "tiny_spike": tiny_spike, "k_warmup_steps": k_warmup_steps, "k_logging": k_logging,
+        }
+        self.stochastic_rounding = stochastic_rounding
+        self.cautious_mask = cautious_mask
+        self.grams_moment = grams_moment
+        self.use_AdEMAMix = use_AdEMAMix
+        self.factored = nnmf_factor
+        self.kourkoutas_beta = kourkoutas_beta
+        self.layer_key_fn = layer_key_fn
+        super().__init__(params, defaults)
+
+        if self.kourkoutas_beta:
+            self.kourkoutas_helper = KourkoutasHelper(self)
+
+    @property
+    def supports_fused_back_pass(self):
+        return True
+
+    @property
+    def supports_memory_efficient_fp16(self):
+        return True
+
+    @property
+    def supports_flat_params(self):
+        return False
+
+    @torch.no_grad()
+    def step_parameter(self, p: torch.Tensor, group: dict, i: int | None = None):
+        if p.grad is None:
+            return
+
+        grad = p.grad
+        if grad.dtype != torch.float32 and self.factored:
+            grad = grad.float()
+        if group["orthogonal_gradient"]:
+            grad = _orthogonalize_gradient(p, grad)
+        state = self.state[p]
+
+        # State Initialization
+        if 'step' not in state:
+            state['step'] = 0
+
+            should_factor = (
+                self.factored and
+                not (len(p.shape) == 1 and not group['vector_reshape'])
+            )
+
+            state['factored'] = should_factor
+
+            dtype = torch.float32 if self.factored else p.dtype
+            device = p.device
+
+            if state['factored']:
+                state['effective_shape'] = _get_effective_shape(p.numel())
+                d1, d2 = state['effective_shape']
+
+                # First moment (m)
+                if group['betas'][0] > 0:
+                    state['mu_m_nmf'] = torch.zeros(d1, device=device, dtype=dtype) 
+                    state['mv_m_nmf'] = torch.zeros(d2, device=device, dtype=dtype)
+                    if not self.grams_moment:
+                        packed_d2 = (d2 + 7) // 8
+                        state['sign'] = torch.zeros((d1, packed_d2), dtype=torch.uint8, device=device)
+                if self.use_AdEMAMix:
+                    state['mu_m_slow_nmf'] = torch.zeros(d1, device=p.device, dtype=dtype) 
+                    state['mv_m_slow_nmf'] = torch.zeros(d2, device=p.device, dtype=dtype)
+                    packed_d2 = (d2 + 7) // 8
+                    state['sign_slow'] = torch.zeros((d1, packed_d2), dtype=torch.uint8, device=p.device)
+                # Second moment (v)
+                state['mu_v_nmf'] = torch.zeros(d1, device=device, dtype=dtype) 
+                state['mv_v_nmf'] = torch.zeros(d2, device=device, dtype=dtype)
+            else:  # Fallback to standard AdamW for non-factored tensors
+                if group['betas'][0] > 0:
+                    state['exp_avg'] = torch.zeros_like(p, device=device, dtype=dtype)
+                if self.use_AdEMAMix:
+                    state['exp_avg_slow'] = torch.zeros_like(p, device=device, dtype=dtype)
+                state['exp_avg_sq'] = torch.zeros_like(p, device=device, dtype=dtype)
+
+        beta1, beta2 = group['betas']
+
+        current_step = state['step']
+        if group.get('kourkoutas_beta', False):
+            # Call prepare_step() once at the beginning of the step for all params
+            self.kourkoutas_helper.maybe_prepare_step(current_step)
+            # Accumulate current grad's norm for the *next* step
+            self.kourkoutas_helper.accumulate_gradient_sq_norm(p, grad)
+            # Get the dynamic beta2 calculated in prepare_step()
+            beta2 = self.kourkoutas_helper.get_beta2(p, group, current_step)
+
+        step = state['step'] + 1
+        if group['use_bias_correction']:
+            bias_correction1 = 1.0 - beta1 ** step
+            if group.get('kourkoutas_beta', False):
+                bias_correction2 = 1.0 - group['betas'][1] ** step
+                # Use beta2_max for bias correction
+            else:
+                bias_correction2 = 1.0 - beta2 ** step
+        else:
+            bias_correction1 = 1
+            bias_correction2 = 1
+        step_size = group['lr'] / bias_correction1
+
+        if self.use_AdEMAMix:
+            beta3_ema = group['beta3_ema']
+            alpha = group['alpha']
+            t_alpha = group['t_alpha']
+            alpha_t = alpha
+            if t_alpha is not None and t_alpha > 0 and step < t_alpha:
+                alpha_t = min(step * alpha / t_alpha, alpha)
+
+        if state['factored']:
+            d1, d2 = state['effective_shape']
+            grad_reshaped = grad.view(d1, d2)
+
+            # Reconstruct momentum from previous step's factors
+            if beta1 > 0:
+                mt = _unnmf((state['mu_m_nmf'], state['mv_m_nmf']))
+                if not self.grams_moment:
+                    unpacked_sign = _unpack_bools(state['sign'], original_m=d2)
+                    torch.where(unpacked_sign, mt, -mt, out=mt)
+                    del unpacked_sign
+                # Update momentum in full-size
+                mt.mul_(beta1).add_(grad_reshaped, alpha=1.0 - beta1)
+                if self.grams_moment:
+                    update_mt = (grad_reshaped.sign().mul_(mt.abs()))
+                elif self.cautious_mask:
+                    mask = (mt * grad_reshaped > 0).to(grad_reshaped.dtype)
+                    mask.div_(mask.mean().clamp_(min=1e-3))
+                    update_mt = mt.mul(mask)
+                    del mask
+                else:
+                    update_mt = mt.clone()
+
+            vt = _unnmf((state['mu_v_nmf'], state['mv_v_nmf']))
+            vt.mul_(beta2).addcmul_(grad_reshaped, grad_reshaped, value=1.0 - beta2)
+
+            if self.use_AdEMAMix:
+                mt_slow = _unnmf((state['mu_m_slow_nmf'], state['mv_m_slow_nmf']))
+                if state['sign_slow'].dtype != torch.uint8:
+                    state['sign_slow'] = state['sign_slow'].to(torch.uint8)
+                unpacked_sign_slow = _unpack_bools(state['sign_slow'], original_m=d2)
+                torch.where(unpacked_sign_slow, mt_slow, -mt_slow, out=mt_slow)
+                del unpacked_sign_slow
+
+                mt_slow.mul_(beta3_ema).add_(grad_reshaped, alpha=1.0 - beta3_ema)
+                if beta1 > 0:
+                    update = torch.add(update_mt, mt_slow, alpha=alpha_t)
+                else:
+                    update = torch.add(grad_reshaped, mt_slow, alpha=alpha_t)
+            else:
+                update = update_mt if beta1 > 0 else grad_reshaped.clone()
+            del grad_reshaped
+
+            if group['use_atan2']:
+                a = 1.2732395
+                denom = (vt.sqrt() / (bias_correction2**0.5))
+                update.atan2_(denom).mul_(a)
+            else:
+                denom = (vt.sqrt() / (bias_correction2**0.5)).add_(group['eps'])
+                update.div_(denom)
+            del denom
+
+            update = update.view(p.shape).mul_(step_size)
+
+            # Compress updated moments and store new factors
+            if beta1 > 0:
+                if not self.grams_moment:
+                    state['sign'] = _pack_bools(mt > 0)
+                _nnmf(mt.abs(), out=(state['mu_m_nmf'], state['mv_m_nmf']))
+                del mt
+            if self.use_AdEMAMix:
+                state['sign_slow'] = _pack_bools(mt_slow > 0)
+                _nnmf(mt_slow.abs(), out=(state['mu_m_slow_nmf'], state['mv_m_slow_nmf']))
+                del mt_slow
+            _nnmf(vt, out=(state['mu_v_nmf'], state['mv_v_nmf']))
+            del vt
+
+        else:  # Standard AdamW logic for non-factored tensors
+            exp_avg_sq = state['exp_avg_sq']
+
+            if beta1 > 0:
+                exp_avg = state['exp_avg']
+                exp_avg.mul_(beta1).add_(grad, alpha=1 - beta1)
+                if self.grams_moment:
+                    update_mt = grad.sign().mul_(exp_avg.abs())
+                elif self.cautious_mask:
+                    mask = (exp_avg * grad > 0).to(grad.dtype)
+                    mask.div_(mask.mean().clamp_(min=1e-3))
+                    update_mt = exp_avg.mul(mask)
+                    del mask
+                else:
+                    update_mt = exp_avg.clone()
+
+            if self.use_AdEMAMix:
+                exp_avg_slow = state['exp_avg_slow']
+                exp_avg_slow.mul_(beta3_ema).add_(grad, alpha=1 - beta3_ema)
+                if beta1 > 0:
+                    update = torch.add(update_mt, exp_avg_slow, alpha=alpha_t)
+                else:
+                    update = torch.add(grad, exp_avg_slow, alpha=alpha_t)
+            else:
+                update = update_mt if beta1 > 0 else grad.clone()
+
+            exp_avg_sq.mul_(beta2).addcmul_(grad, grad.conj(), value=1 - beta2)
+
+            if group['use_atan2']:
+                a = 1.2732395
+                denom = (exp_avg_sq.sqrt() / (bias_correction2**0.5))
+                update.atan2_(denom).mul_(a)
+            else:
+                denom = (exp_avg_sq.sqrt() / (bias_correction2**0.5)).add_(group['eps'])
+                update.div_(denom)
+            del denom
+
+            update.mul_(step_size)
+
+        # Decoupled weight decay
+        if group["weight_decay"] != 0:
+            if p.dtype == torch.bfloat16 and self.stochastic_rounding:
+                add_stochastic_(p.data, p.data, alpha=-group["weight_decay"] * group["lr"])
+            else:
+                p.data.add_(p.data, alpha=-group["weight_decay"] * group["lr"])
+
+        if p.dtype == torch.bfloat16 and self.stochastic_rounding:
+            add_stochastic_(p.data, -update)
+        else:
+            p.data.add_(-update)
+        del update
+
+        state['step'] += 1
+
+    @torch.no_grad()
+    def step(self, closure=None):
+        """Performs a single optimization step."""
+        loss = None
+        if closure is not None:
+            with torch.enable_grad():
+                loss = closure()
+
+        for group in self.param_groups:
+            for i, p in enumerate(group['params']):
+                self.step_parameter(p, group, i)
+
+        return loss