You cannot select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
335 lines
15 KiB
Python
335 lines
15 KiB
Python
"""
|
|
A patch to bitsandbytes 0.34.0 that introduces an option to run backward pass in default (fast) matrix layout.
|
|
Authors: modification by @borzunov, original code by @timdettmers. Please disregard commit authors in this file.
|
|
|
|
Core idea: layouts apply the same permutation to every tile in the matrix. We can treat this as (batched) gather ops.
|
|
Reshape input tensor so that ij-th gather operation op will apply to ij-th elements in each tile.
|
|
Prototype: https://colab.research.google.com/drive/1EJ0MKifajXSSVq7O2_QGwtb0l6gRAGrh?usp=sharing
|
|
Based on: https://github.com/TimDettmers/bitsandbytes/blob/main/csrc/kernels.cu#L2130-L2136
|
|
Exact match tests: see $REPO/tests/test_linear8bitlt.py
|
|
"""
|
|
import dataclasses
|
|
import logging
|
|
from typing import Optional, Tuple
|
|
|
|
import bitsandbytes.functional as F
|
|
import torch
|
|
from bitsandbytes.autograd._functions import GlobalOutlierPooler, MatMul8bitLt, MatmulLtState, prod
|
|
from bitsandbytes.nn import Linear8bitLt
|
|
|
|
|
|
def get_inverse_transform_indices(transform_tile: callable, tile_size: Tuple[int, int]):
|
|
"""
|
|
Compute a permutation of indices that invert the specified (tiled) matrix transformation
|
|
|
|
:param transform_tile: a function that applies forward transform to a tensor of shape [dim1, dim2]
|
|
:param tile_size: higher-level tile dimensions, i.e. (8, 32) for Turing and (32, 32) for Ampere
|
|
:note: we assume that tile_transform applies to a cpu-based int8 tensor of shape tile_size
|
|
:example: transform_tile function for the turing layout (bitsandbytes.functional as F)
|
|
:returns: indices
|
|
"""
|
|
d1, d2 = tile_size
|
|
assert 0 < d1 * d2 < 2**64
|
|
tile_indices = torch.arange(d1 * d2, dtype=torch.int64).view(d1, d2)
|
|
# encode each position in tile as a tuple of <= 8 unique bytes
|
|
permuted_tile_indices = torch.zeros_like(tile_indices)
|
|
for i in range(8):
|
|
# select i-th byte, apply transformation and trace where each index ended up
|
|
ith_dim_indices = torch.div(tile_indices, 256**i, rounding_mode="trunc") % 256
|
|
sample_tile_i = (ith_dim_indices - 128).to(torch.int8).contiguous()
|
|
assert torch.all(sample_tile_i.int() + 128 == ith_dim_indices), "int overflow"
|
|
permuted_tile_i = transform_tile(sample_tile_i)
|
|
ith_permuted_indices = permuted_tile_i.to(tile_indices.dtype) + 128
|
|
permuted_tile_indices += ith_permuted_indices * (256**i)
|
|
if d1 * d2 < 256**i:
|
|
break # if all indices fit in i bytes, stop early
|
|
return permuted_tile_indices
|
|
|
|
|
|
def undo_layout(permuted_tensor: torch.Tensor, tile_indices: torch.LongTensor) -> torch.Tensor:
|
|
"""
|
|
Undo a tiled permutation such as turing or ampere layout
|
|
|
|
:param permuted_tensor: torch tensor in a permuted layout
|
|
:param tile_indices: reverse transformation indices, from get_inverse_transform_indices
|
|
:return: contiguous row-major tensor
|
|
"""
|
|
(rows, cols), (tile_rows, tile_cols) = permuted_tensor.shape, tile_indices.shape
|
|
assert rows % tile_rows == cols % tile_cols == 0, "tensor must contain a whole number of tiles"
|
|
tensor = permuted_tensor.reshape(-1, tile_indices.numel()).t()
|
|
outputs = torch.empty_like(tensor) # note: not using .index_copy because it was slower on cuda
|
|
outputs[tile_indices.flatten()] = tensor
|
|
outputs = outputs.reshape(tile_rows, tile_cols, cols // tile_cols, rows // tile_rows)
|
|
outputs = outputs.permute(3, 0, 2, 1) # (rows // tile_rows, tile_rows), (cols // tile_cols, tile_cols)
|
|
return outputs.reshape(rows, cols).contiguous()
|
|
|
|
|
|
# the rest of this file is just a patch to bitsandbytes that modifies Linear8bitLt and dependencies
|
|
|
|
|
|
class CustomLinear8bitLt(Linear8bitLt):
|
|
def __init__(self, *args, memory_efficient_backward: bool = False, **kwargs):
|
|
assert not memory_efficient_backward, "memory_efficient_backward is no longer used"
|
|
super().__init__(*args, **kwargs)
|
|
old_state, self.state = self.state, CustomMatmulLtState()
|
|
self.state.threshold = old_state.threshold
|
|
self.state.has_fp16_weights = old_state.has_fp16_weights
|
|
self.state.memory_efficient_backward = old_state.memory_efficient_backward
|
|
if old_state.threshold > 0.0 and not old_state.has_fp16_weights:
|
|
self.state.use_pool = True
|
|
|
|
def forward(self, x: torch.Tensor):
|
|
self.state.is_training = self.training
|
|
if self.weight.CB is not None:
|
|
self.init_8bit_state()
|
|
|
|
# weights are cast automatically as Int8Params, but the bias has to be cast manually
|
|
if self.bias is not None and self.bias.dtype != x.dtype:
|
|
self.bias.data = self.bias.data.to(x.dtype)
|
|
|
|
out = custom_matmul8bitlt(x, self.weight, bias=self.bias, state=self.state)
|
|
if not self.state.has_fp16_weights:
|
|
if self.state.CB is not None and self.state.CxB is not None:
|
|
# we converted 8-bit row major to turing/ampere format in the first inference pass
|
|
# we no longer need the row-major weight
|
|
del self.state.CB
|
|
self.weight.data = self.state.CxB
|
|
return out
|
|
|
|
|
|
@dataclasses.dataclass(init=True)
|
|
class CustomMatmulLtState(MatmulLtState):
|
|
tile_indices: Optional[torch.Tensor] = None
|
|
force_no_igemmlt: bool = False
|
|
|
|
def get_tile_size(self):
|
|
assert self.formatB in (
|
|
"col_turing",
|
|
"col_ampere",
|
|
), f"please find this assert and manually enter tile size for {self.formatB}"
|
|
return (8, 32) if self.formatB == "col_turing" else (32, 32)
|
|
|
|
|
|
def custom_matmul8bitlt(
|
|
A: torch.Tensor,
|
|
B: torch.Tensor,
|
|
out: torch.Tensor = None,
|
|
state: CustomMatmulLtState = None,
|
|
threshold=0.0,
|
|
bias=None,
|
|
):
|
|
state = state or MatmulLtState()
|
|
if threshold > 0.0:
|
|
state.threshold = threshold
|
|
return CustomMatMul8bitLt.apply(A, B, out, bias, state)
|
|
|
|
|
|
class CustomMatMul8bitLt(MatMul8bitLt):
|
|
# forward is the same, but we added the fallback for pre-turing GPUs
|
|
# backward is mostly the same, but adds one extra clause (see "elif state.CxB is not None")
|
|
|
|
@staticmethod
|
|
def forward(ctx, A, B, out=None, bias=None, state=CustomMatmulLtState):
|
|
using_igemmlt = torch.cuda.get_device_capability(device=A.device) >= (7, 5) and not state.force_no_igemmlt
|
|
# default to pytorch behavior if inputs are empty
|
|
ctx.is_empty = False
|
|
if prod(A.shape) == 0:
|
|
ctx.is_empty = True
|
|
ctx.A = A
|
|
ctx.B = B
|
|
ctx.bias = bias
|
|
if A.shape[-1] == B.shape[0]:
|
|
return torch.empty(A.shape[:-1] + B.shape[1:], dtype=A.dtype, device=A.device)
|
|
else:
|
|
return torch.empty(A.shape[:-1] + B.shape[:1], dtype=A.dtype, device=A.device)
|
|
|
|
# 1. Quantize A
|
|
# 2. Quantize B
|
|
# 3. Matmul
|
|
# 4. Mixed-precision decomposition matmul
|
|
# 5. Save state
|
|
formatB = state.formatB
|
|
input_shape = A.shape
|
|
if state.outlier_pool is None:
|
|
state.outlier_pool = GlobalOutlierPooler.get_instance()
|
|
|
|
# Cast A to fp16
|
|
if A.dtype != torch.float16:
|
|
logging.debug(f"MatMul8bitLt: inputs will be cast from {A.dtype} to float16 during quantization")
|
|
|
|
# 1. Quantize A
|
|
if len(A.shape) == 3:
|
|
A = A.view(-1, A.shape[-1]).contiguous()
|
|
CA, CAt, SCA, SCAt, coo_tensorA = F.double_quant(A.to(torch.float16), threshold=state.threshold)
|
|
|
|
if state.threshold > 0.0 and coo_tensorA is not None:
|
|
if state.has_fp16_weights:
|
|
idx = torch.unique(coo_tensorA.colidx).long()
|
|
CA[:, idx] = 0
|
|
CAt[:, idx] = 0
|
|
subA = A[:, idx]
|
|
state.subB = B[:, idx].t().contiguous()
|
|
state.idx = idx
|
|
else:
|
|
if state.CxB is None and using_igemmlt:
|
|
# B in in 8-bit row-major, we can transform it back to 16-bit to extract outlier dimensions
|
|
# we also need to convert it to the turing/ampere format
|
|
state.CxB, state.SB = F.transform(state.CB, to_order=formatB)
|
|
else:
|
|
if not state.has_fp16_weights and state.CxB is None and using_igemmlt:
|
|
state.CxB, state.SB = F.transform(state.CB, to_order=formatB)
|
|
subA = None
|
|
|
|
# 2. Quantize B
|
|
if state.has_fp16_weights:
|
|
has_grad = True if (getattr(B, "grad", None) is not None) else False
|
|
is_transposed = not B.is_contiguous() and B.shape[0] == B.stride(1)
|
|
if is_transposed:
|
|
B = B.contiguous()
|
|
|
|
if (state.is_training and not has_grad) or state.CxB is None:
|
|
state.reset_grads()
|
|
(
|
|
CB,
|
|
state.CBt,
|
|
state.SCB,
|
|
state.SCBt,
|
|
coo_tensorB,
|
|
) = F.double_quant(B.to(torch.float16))
|
|
if using_igemmlt:
|
|
state.CxB, state.SB = F.transform(CB, to_order=formatB)
|
|
else:
|
|
state.CB = CB
|
|
else:
|
|
has_grad = False
|
|
|
|
if coo_tensorA is not None and not state.has_fp16_weights:
|
|
# extract outliers
|
|
|
|
outlier_idx = torch.unique(coo_tensorA.colidx)
|
|
state.idx = outlier_idx
|
|
# state.outlier_pool.add_outliers(outlier_idx, A.shape[-1])
|
|
# if state.use_pool and state.outlier_pool.model_dim == A.shape[-1]:
|
|
# # do not use pool for 2nd FFN layer
|
|
# state.idx = state.outlier_pool.get_current_outlier_idx().to(A.device)
|
|
# else:
|
|
# state.idx = outlier_idx
|
|
if state.CxB is not None:
|
|
outliers = F.extract_outliers(state.CxB, state.SB, state.idx.int())
|
|
else:
|
|
outliers = state.CB[:, state.idx.long()].clone()
|
|
|
|
state.subB = (outliers * state.SCB.view(-1, 1) / 127.0).t().contiguous().to(A.dtype)
|
|
CA[:, state.idx.long()] = 0
|
|
CAt[:, state.idx.long()] = 0
|
|
subA = A[:, state.idx.long()]
|
|
|
|
shapeB = state.SB[0] if state.SB else B.shape
|
|
|
|
if len(input_shape) == 3:
|
|
output_shape = (input_shape[0], input_shape[1], shapeB[0])
|
|
else:
|
|
output_shape = (input_shape[0], shapeB[0])
|
|
|
|
# 3. Matmul
|
|
if using_igemmlt:
|
|
C32A, SA = F.transform(CA, "col32")
|
|
out32, Sout32 = F.igemmlt(C32A, state.CxB, SA, state.SB)
|
|
if bias is None or bias.dtype == torch.float16:
|
|
# we apply the fused bias here
|
|
output = F.mm_dequant(out32, Sout32, SCA, state.SCB, bias=bias)
|
|
output = output.to(A.dtype)
|
|
else: # apply bias separately
|
|
output = F.mm_dequant(out32, Sout32, SCA, state.SCB, bias=None)
|
|
output = output.to(A.dtype).add_(bias)
|
|
|
|
else:
|
|
A_wo_outliers = A.clone()
|
|
if state.idx is not None:
|
|
A_wo_outliers[:, state.idx.long()] = 0
|
|
output = torch.nn.functional.linear(A_wo_outliers, state.CB.to(A.dtype))
|
|
output = output.mul_(state.SCB.unsqueeze(0).mul(1.0 / 127.0))
|
|
if bias is not None:
|
|
output = output.add_(bias)
|
|
|
|
# 4. Mixed-precision decomposition matmul
|
|
if coo_tensorA is not None and subA is not None:
|
|
output += torch.matmul(subA, state.subB)
|
|
|
|
# 5. Save state
|
|
ctx.state = state
|
|
|
|
ctx.formatB = formatB
|
|
ctx.grad_shape = input_shape
|
|
ctx.dtype_A, ctx.dtype_B, ctx.dtype_bias = A.dtype, B.dtype, None if bias is None else bias.dtype
|
|
|
|
if any(ctx.needs_input_grad[:2]):
|
|
ctx.tensors = (CAt, subA)
|
|
ctx.tensor_states = (SCAt, state.idx)
|
|
else:
|
|
ctx.tensors = [None, None]
|
|
ctx.tensor_states = (None, None)
|
|
ctx.save_for_backward(None, None)
|
|
|
|
clone_func = torch.clone if len(output_shape) == 3 else lambda x: x
|
|
return clone_func(output.view(output_shape))
|
|
|
|
@staticmethod
|
|
def backward(ctx, grad_output):
|
|
if ctx.is_empty:
|
|
bias_grad = None if ctx.bias is None else torch.zeros_like(ctx.bias)
|
|
return torch.zeros_like(ctx.A), torch.zeros_like(ctx.B), None, bias_grad, None
|
|
req_gradA, req_gradB, _, req_gradBias, _ = ctx.needs_input_grad
|
|
CAt, subA = ctx.tensors
|
|
SCAt, idx = ctx.tensor_states
|
|
formatB = ctx.formatB
|
|
state = ctx.state
|
|
grad_A = grad_B = grad_bias = None
|
|
|
|
if req_gradBias:
|
|
# compute grad_bias first before changing grad_output dtype
|
|
grad_bias = grad_output.sum(0, dtype=ctx.dtype_bias)
|
|
|
|
# Cast grad_output to fp16
|
|
if len(grad_output.shape) == 3:
|
|
grad_output = grad_output.reshape(-1, grad_output.shape[-1]).contiguous()
|
|
|
|
Cgrad, Cgradt, SCgrad, SCgradt, coo_tensor = F.double_quant(grad_output.to(torch.float16))
|
|
if req_gradB:
|
|
CxAt, SAt = F.transform(CAt, formatB, transpose=True)
|
|
C32grad, Sgrad = F.transform(Cgradt, "col32", transpose=True)
|
|
gradB32, SgradB32 = F.igemmlt(C32grad, CxAt, Sgrad, SAt)
|
|
grad_B = F.mm_dequant(gradB32, SgradB32, SCgradt, SCAt)
|
|
if state.threshold > 0.0 and subA is not None:
|
|
grad_B[:, idx] += torch.matmul(grad_output.t(), subA)
|
|
|
|
if req_gradA:
|
|
if state.CBt is not None:
|
|
C32grad, Sgrad = F.transform(Cgrad, "col32")
|
|
if state.CxBt is None:
|
|
state.CxBt, state.SBt = F.transform(state.CBt, to_order=formatB, transpose=True)
|
|
gradA32, SgradA32 = F.igemmlt(C32grad, state.CxBt, Sgrad, state.SBt)
|
|
grad_A = F.mm_dequant(gradA32, SgradA32, SCgrad, state.SCBt).view(ctx.grad_shape).to(ctx.dtype_A)
|
|
|
|
elif state.CB is not None:
|
|
CB = state.CB.to(ctx.dtype_A, copy=True).mul_(state.SCB.unsqueeze(1).mul(1.0 / 127.0))
|
|
grad_A = torch.matmul(grad_output, CB).view(ctx.grad_shape).to(ctx.dtype_A)
|
|
elif state.CxB is not None:
|
|
|
|
if state.tile_indices is None:
|
|
order, tile_size = state.formatB, state.get_tile_size()
|
|
transform = lambda x: F.transform(x.cuda(), from_order="row", to_order=order)[0].to(x.device)
|
|
with torch.no_grad():
|
|
state.tile_indices = get_inverse_transform_indices(transform, tile_size).to(state.CxB.device)
|
|
|
|
CB = (
|
|
undo_layout(state.CxB, state.tile_indices)
|
|
.to(ctx.dtype_A)
|
|
.mul_(state.SCB.unsqueeze(1).mul(1.0 / 127.0))
|
|
)
|
|
grad_A = torch.matmul(grad_output, CB).view(ctx.grad_shape).to(ctx.dtype_A)
|
|
else:
|
|
raise Exception("State must contain either CBt or CB or CxB matrix for backward")
|
|
|
|
return grad_A, grad_B, None, grad_bias, None
|