Use torch.compile to speed up GPTQ algo

src/llmcompressor/modifiers/quantization/gptq/gptq_quantize.py

@@ -3,6 +3,8 @@
 from typing import Dict, Optional, Tuple, Union
 
 import torch
+import torch._dynamo.config
+import torch._inductor.config
 import transformers
 from compressed_tensors.quantization import (
     ActivationOrdering,
@@ -68,7 +70,7 @@ def accumulate_hessian(
     return H, num_samples
 
 
-def quantize_weight(
+def quantize_weight_original(
     module: torch.nn.Module,
     quant_args: QuantizationArgs,
     hessians_dict: Dict[torch.nn.Module, torch.Tensor],
@@ -279,6 +281,281 @@ def quantize_weight(
     )
 
 
+@torch.compile(dynamic=True)
+def _process_block(
+    W1: torch.Tensor,
+    Hinv1: torch.Tensor,
+    scale_slice: torch.Tensor,
+    zero_slice: torch.Tensor,
+    mask_slice: Optional[torch.Tensor],
+    quant_min: int,
+    quant_max: int,
+    sym: bool,
+) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]:
+    """Process a single block of weight columns using with torch.compile support."""
+    count = W1.shape[1]
+    Q1 = torch.zeros_like(W1)
+    Err1 = torch.zeros_like(W1)
+    losses1 = torch.zeros_like(W1)
+
+    for i in range(count):
+        w = W1[:, i]
+        d = Hinv1[i, i]
+
+        s = scale_slice[:, i]
+        z = zero_slice[:, i]
+
+        if sym:
+            z = torch.zeros_like(z)
+
+        scaled = w / s
+        if not sym:
+            scaled -= z
+        q = torch.clamp(torch.round(scaled), quant_min, quant_max)
+        dq = q * s
+        if not sym:
+            dq += z * s
+
+        err1 = (w - dq) / d
+        loss_col = (w - dq) ** 2 / d**2
+
+        Q1[:, i] = dq
+        Err1[:, i] = err1
+        losses1[:, i] = loss_col
+
+        w1_err = err1.unsqueeze(1) @ Hinv1[i, i:].unsqueeze(0)
+        if mask_slice is not None:
+            mask_block = mask_slice[:, i:]
+            W1[:, i:] -= w1_err * mask_block
+        else:
+            W1[:, i:] -= w1_err
+
+    return Q1, Err1, losses1
+
+
+def _quantize_core(
+    W: torch.Tensor,
+    Hinv: torch.Tensor,
+    scale_map: torch.Tensor,
+    zero_map: torch.Tensor,
+    W_nz_mask: Optional[torch.Tensor],
+    blocksize: int,
+    quant_min: int,
+    quant_max: int,
+    sym: bool,
+    num_rows: int,
+    num_columns: int,
+) -> Tuple[torch.Tensor, torch.Tensor]:
+    """Core GPTQ quantization loop processing weights in blocks."""
+    losses = torch.zeros(num_rows, device=W.device, dtype=W.dtype)
+
+    for i1 in range(0, num_columns, blocksize):
+        i2 = min(i1 + blocksize, num_columns)
+
+        # Extract current block and corresponding Hessian/quantization params
+        W1 = W[:, i1:i2].clone()
+        Hinv1 = Hinv[i1:i2, i1:i2].contiguous()
+        scale_slice = scale_map[:, i1:i2]
+        zero_slice = zero_map[:, i1:i2]
+        mask_slice = None
+        if W_nz_mask is not None:
+            mask_slice = W_nz_mask[:, i1:i2]
+
+        # Quantize the current block
+        Q1, Err1, losses1 = _process_block(
+            W1, Hinv1, scale_slice, zero_slice, mask_slice, quant_min, quant_max, sym
+        )
+
+        W[:, i1:i2] = Q1
+        losses += losses1.sum(dim=1) / 2
+
+        # Propagate block error to remaining unprocessed columns
+        w_err = Err1 @ Hinv[i1:i2, i2:]
+        if W_nz_mask is not None:
+            mask_rest = W_nz_mask[:, i2:]
+            W[:, i2:] -= w_err * mask_rest
+        else:
+            W[:, i2:] -= w_err
+
+    return W, losses
+
+
+def quantize_weight(
+    module: torch.nn.Module,
+    quant_args: QuantizationArgs,
+    hessians_dict: Dict[torch.nn.Module, torch.Tensor],
+    blocksize: int = 128,
+    percdamp: float = 0.01,
+) -> Tuple[float, torch.Tensor, torch.Tensor, Union[torch.Tensor, None], torch.Tensor]:
+    """
+    Quantize a module weight according to the GPTQ algorithm
+
+    This version is faster than the original one with torch.compile support
+
+    :param module: module with weight being quantized
+    :param quant_args: quantization arguments used to find quantization parameters
+    :param hessian_dict: dictionary containing preaccumulated hessian for quantization
+    :param blocksize: chunk size of quantization updates
+    :param percdamp: dampening factor on hessian diagonal
+    :return: loss, quantized_weight, scale, zero_point, g_idx
+    """
+    strategy = quant_args.strategy
+    actorder = quant_args.actorder
+    n_bits = quant_args.num_bits
+    sym = quant_args.symmetric
+    final_shape = module.weight.shape
+    final_dtype = module.weight.dtype
+    W = module.weight.clone()
+    H = hessians_dict[module]  # unfortunately python does not have a `move` keyword
+    del hessians_dict[module]  # so we have to delete the original reference manually
+
+    # create observer for calculating quantization parameters
+    observer = Observer.load_from_registry(
+        quant_args.observer,
+        quantization_args=quant_args,
+        averaging_constant=1.0,  # ignore moving average
+    )
+
+    # standardize shape and dtype
+    if isinstance(module, torch.nn.Conv2d):
+        W = W.flatten(1)
+    elif isinstance(module, transformers.Conv1D):
+        W.transpose_(0, 1)
+    W = W.to(dtype=GPTQ_PRECISION)
+    num_rows = W.shape[0]
+    num_columns = W.shape[1]
+
+    if strategy == QuantizationStrategy.GROUP:
+        # mapping from column index to group index
+        g_idx = (
+            torch.arange(num_columns, device=W.device, dtype=torch.int)
+            // quant_args.group_size
+        )
+
+        if actorder == ActivationOrdering.GROUP:
+            # permute by activation order first, then update groups
+            W, H, perm = _apply_activation_ordering(W, H)
+            scale, zero_point = observer(W, g_idx=None)
+
+            # use identity g_idx (invert permutation later)
+
+        elif actorder == ActivationOrdering.WEIGHT:
+            # update groups first, then permute by activation order
+            scale, zero_point = observer(W, g_idx=None)
+            W, H, perm = _apply_activation_ordering(W, H)
+
+            # permute g_idx to maintain identity mapping after unpermutation
+            g_idx = g_idx[perm]
+
+        else:
+            scale, zero_point = observer(W, g_idx=None)
+    else:
+        scale, zero_point = observer(W, g_idx=None)
+
+    scale = scale.to(W.device)
+    zero_point = zero_point.to(W.device)
+
+    # sparsity mask
+    sparsity = tensor_sparsity(W)
+    preserve_zeros = sparsity >= SPARSITY_THRESHOLD
+    W_nz_mask = (
+        (~torch.isclose(W, torch.zeros(1, device=W.device, dtype=W.dtype))).float()
+        if preserve_zeros
+        else None
+    )
+
+    # mask dead hessian values
+    dead = torch.diag(H) == 0
+    H[dead, dead] = 1
+    W[:, dead] = 0
+
+    # compute inverse hessian in place to save memory
+    try:
+        damp = percdamp * torch.mean(torch.diag(H))
+        diag = torch.arange(H.shape[0], device=H.device)
+        H[diag, diag] += damp
+        H = torch.linalg.cholesky(H)
+        H = torch.cholesky_inverse(H)
+        H = torch.linalg.cholesky(H, upper=True)
+        Hinv = H
+    except torch._C._LinAlgError:
+        logger.warning(
+            "Failed to invert hessian due to numerical instability. Consider "
+            "increasing GPTQModifier.dampening_frac, increasing the number "
+            "of calibration samples, or shuffling the calibration dataset. "
+            "Falling back to round-to-nearest for this module."
+        )
+        Hinv = H = torch.eye(num_columns, dtype=H.dtype, device=H.device)
+
+    # See section 3.4 of https://arxiv.org/abs/2203.07259
+    # quantize column
+    if strategy == QuantizationStrategy.TENSOR:
+        scale_map = scale.expand(num_rows, num_columns)
+        zero_map = zero_point.expand(num_rows, num_columns)
+    elif strategy == QuantizationStrategy.CHANNEL:
+        scale_map = scale.expand(-1, num_columns)
+        zero_map = zero_point.expand(-1, num_columns)
+    elif strategy == QuantizationStrategy.GROUP:
+        # get the group index for the current column
+        scale_map = scale[:, g_idx]
+        zero_map = zero_point[:, g_idx]
+    else:
+        raise ValueError(f"Quantization strategy is not supported for GPTQ: {strategy}")
+
+    if sym:
+        quant_min = -(2 ** (n_bits - 1))
+        quant_max = 2 ** (n_bits - 1) - 1
+    else:
+        quant_min = 0
+        quant_max = 2**n_bits - 1
+
+    W, losses = _quantize_core(
+        W=W,
+        Hinv=Hinv,
+        scale_map=scale_map,
+        zero_map=zero_map,
+        W_nz_mask=W_nz_mask,
+        blocksize=blocksize,
+        quant_min=quant_min,
+        quant_max=quant_max,
+        sym=sym,
+        num_rows=num_rows,
+        num_columns=num_columns,
+    )
+
+    has_gidx = False
+    if strategy == QuantizationStrategy.GROUP:
+        if actorder == ActivationOrdering.WEIGHT:
+            # restore original permutation
+            invperm = torch.argsort(perm)
+            W = W[:, invperm]
+
+        elif actorder == ActivationOrdering.GROUP:
+            # restore original permutation
+            invperm = torch.argsort(perm)
+            W = W[:, invperm]
+            g_idx = g_idx[invperm]
+
+            # only save g_idx if mapping is not identity
+            has_gidx = True
+
+    if not has_gidx:
+        g_idx = None
+
+    if isinstance(module, transformers.Conv1D):
+        W.transpose_(0, 1)
+    W = W.reshape(final_shape).to(final_dtype)
+
+    loss = torch.sum(losses).item()
+    return (
+        loss,
+        W,
+        scale.to(dtype=final_dtype),
+        zero_point.to(dtype=quant_args.pytorch_dtype()),
+        g_idx,
+    )
+
+
 def _apply_activation_ordering(
     W: torch.Tensor, H: torch.Tensor
 ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]: