188 lines
8.3 KiB
Python
188 lines
8.3 KiB
Python
from collections import defaultdict
|
|
import torch
|
|
import intel_extension_for_pytorch as ipex # pylint: disable=import-error, unused-import
|
|
import intel_extension_for_pytorch._C as core # pylint: disable=import-error, unused-import
|
|
|
|
# pylint: disable=protected-access, missing-function-docstring, line-too-long
|
|
|
|
OptState = ipex.cpu.autocast._grad_scaler.OptState
|
|
_MultiDeviceReplicator = ipex.cpu.autocast._grad_scaler._MultiDeviceReplicator
|
|
_refresh_per_optimizer_state = (
|
|
ipex.cpu.autocast._grad_scaler._refresh_per_optimizer_state
|
|
)
|
|
|
|
|
|
def _unscale_grads_(
|
|
self, optimizer, inv_scale, found_inf, allow_fp16
|
|
): # pylint: disable=unused-argument
|
|
per_device_inv_scale = _MultiDeviceReplicator(inv_scale)
|
|
per_device_found_inf = _MultiDeviceReplicator(found_inf)
|
|
|
|
# To set up _amp_foreach_non_finite_check_and_unscale_, split grads by device and dtype.
|
|
# There could be hundreds of grads, so we'd like to iterate through them just once.
|
|
# However, we don't know their devices or dtypes in advance.
|
|
|
|
# https://stackoverflow.com/questions/5029934/defaultdict-of-defaultdict
|
|
# Google says mypy struggles with defaultdicts type annotations.
|
|
per_device_and_dtype_grads = defaultdict(lambda: defaultdict(list)) # type: ignore[var-annotated]
|
|
# sync grad to master weight
|
|
if hasattr(optimizer, "sync_grad"):
|
|
optimizer.sync_grad()
|
|
with torch.no_grad():
|
|
for group in optimizer.param_groups:
|
|
for param in group["params"]:
|
|
if param.grad is None:
|
|
continue
|
|
if (not allow_fp16) and param.grad.dtype == torch.float16:
|
|
raise ValueError("Attempting to unscale FP16 gradients.")
|
|
if param.grad.is_sparse:
|
|
# is_coalesced() == False means the sparse grad has values with duplicate indices.
|
|
# coalesce() deduplicates indices and adds all values that have the same index.
|
|
# For scaled fp16 values, there's a good chance coalescing will cause overflow,
|
|
# so we should check the coalesced _values().
|
|
if param.grad.dtype is torch.float16:
|
|
param.grad = param.grad.coalesce()
|
|
to_unscale = param.grad._values()
|
|
else:
|
|
to_unscale = param.grad
|
|
|
|
# -: is there a way to split by device and dtype without appending in the inner loop?
|
|
to_unscale = to_unscale.to("cpu")
|
|
per_device_and_dtype_grads[to_unscale.device][to_unscale.dtype].append(
|
|
to_unscale
|
|
)
|
|
|
|
for _, per_dtype_grads in per_device_and_dtype_grads.items():
|
|
for grads in per_dtype_grads.values():
|
|
core._amp_foreach_non_finite_check_and_unscale_(
|
|
grads,
|
|
per_device_found_inf.get("cpu"),
|
|
per_device_inv_scale.get("cpu"),
|
|
)
|
|
|
|
return per_device_found_inf._per_device_tensors
|
|
|
|
|
|
def unscale_(self, optimizer):
|
|
"""
|
|
Divides ("unscales") the optimizer's gradient tensors by the scale factor.
|
|
:meth:`unscale_` is optional, serving cases where you need to
|
|
:ref:`modify or inspect gradients<working-with-unscaled-gradients>`
|
|
between the backward pass(es) and :meth:`step`.
|
|
If :meth:`unscale_` is not called explicitly, gradients will be unscaled automatically during :meth:`step`.
|
|
Simple example, using :meth:`unscale_` to enable clipping of unscaled gradients::
|
|
...
|
|
scaler.scale(loss).backward()
|
|
scaler.unscale_(optimizer)
|
|
torch.nn.utils.clip_grad_norm_(model.parameters(), max_norm)
|
|
scaler.step(optimizer)
|
|
scaler.update()
|
|
Args:
|
|
optimizer (torch.optim.Optimizer): Optimizer that owns the gradients to be unscaled.
|
|
.. warning::
|
|
:meth:`unscale_` should only be called once per optimizer per :meth:`step` call,
|
|
and only after all gradients for that optimizer's assigned parameters have been accumulated.
|
|
Calling :meth:`unscale_` twice for a given optimizer between each :meth:`step` triggers a RuntimeError.
|
|
.. warning::
|
|
:meth:`unscale_` may unscale sparse gradients out of place, replacing the ``.grad`` attribute.
|
|
"""
|
|
if not self._enabled:
|
|
return
|
|
|
|
self._check_scale_growth_tracker("unscale_")
|
|
|
|
optimizer_state = self._per_optimizer_states[id(optimizer)]
|
|
|
|
if optimizer_state["stage"] is OptState.UNSCALED: # pylint: disable=no-else-raise
|
|
raise RuntimeError(
|
|
"unscale_() has already been called on this optimizer since the last update()."
|
|
)
|
|
elif optimizer_state["stage"] is OptState.STEPPED:
|
|
raise RuntimeError("unscale_() is being called after step().")
|
|
|
|
# FP32 division can be imprecise for certain compile options, so we carry out the reciprocal in FP64.
|
|
assert self._scale is not None
|
|
inv_scale = (
|
|
self._scale.to("cpu").double().reciprocal().float().to(self._scale.device)
|
|
)
|
|
found_inf = torch.full((1,), 0.0, dtype=torch.float32, device=self._scale.device)
|
|
|
|
optimizer_state["found_inf_per_device"] = self._unscale_grads_(
|
|
optimizer, inv_scale, found_inf, False
|
|
)
|
|
optimizer_state["stage"] = OptState.UNSCALED
|
|
|
|
|
|
def update(self, new_scale=None):
|
|
"""
|
|
Updates the scale factor.
|
|
If any optimizer steps were skipped the scale is multiplied by ``backoff_factor``
|
|
to reduce it. If ``growth_interval`` unskipped iterations occurred consecutively,
|
|
the scale is multiplied by ``growth_factor`` to increase it.
|
|
Passing ``new_scale`` sets the new scale value manually. (``new_scale`` is not
|
|
used directly, it's used to fill GradScaler's internal scale tensor. So if
|
|
``new_scale`` was a tensor, later in-place changes to that tensor will not further
|
|
affect the scale GradScaler uses internally.)
|
|
Args:
|
|
new_scale (float or :class:`torch.FloatTensor`, optional, default=None): New scale factor.
|
|
.. warning::
|
|
:meth:`update` should only be called at the end of the iteration, after ``scaler.step(optimizer)`` has
|
|
been invoked for all optimizers used this iteration.
|
|
"""
|
|
if not self._enabled:
|
|
return
|
|
|
|
_scale, _growth_tracker = self._check_scale_growth_tracker("update")
|
|
|
|
if new_scale is not None:
|
|
# Accept a new user-defined scale.
|
|
if isinstance(new_scale, float):
|
|
self._scale.fill_(new_scale) # type: ignore[union-attr]
|
|
else:
|
|
reason = "new_scale should be a float or a 1-element torch.FloatTensor with requires_grad=False."
|
|
assert isinstance(new_scale, torch.FloatTensor), reason # type: ignore[attr-defined]
|
|
assert new_scale.numel() == 1, reason
|
|
assert new_scale.requires_grad is False, reason
|
|
self._scale.copy_(new_scale) # type: ignore[union-attr]
|
|
else:
|
|
# Consume shared inf/nan data collected from optimizers to update the scale.
|
|
# If all found_inf tensors are on the same device as self._scale, this operation is asynchronous.
|
|
found_infs = [
|
|
found_inf.to(device="cpu", non_blocking=True)
|
|
for state in self._per_optimizer_states.values()
|
|
for found_inf in state["found_inf_per_device"].values()
|
|
]
|
|
|
|
assert len(found_infs) > 0, "No inf checks were recorded prior to update."
|
|
|
|
found_inf_combined = found_infs[0]
|
|
if len(found_infs) > 1:
|
|
for i in range(1, len(found_infs)):
|
|
found_inf_combined += found_infs[i]
|
|
|
|
to_device = _scale.device
|
|
_scale = _scale.to("cpu")
|
|
_growth_tracker = _growth_tracker.to("cpu")
|
|
|
|
core._amp_update_scale_(
|
|
_scale,
|
|
_growth_tracker,
|
|
found_inf_combined,
|
|
self._growth_factor,
|
|
self._backoff_factor,
|
|
self._growth_interval,
|
|
)
|
|
|
|
_scale = _scale.to(to_device)
|
|
_growth_tracker = _growth_tracker.to(to_device)
|
|
# To prepare for next iteration, clear the data collected from optimizers this iteration.
|
|
self._per_optimizer_states = defaultdict(_refresh_per_optimizer_state)
|
|
|
|
|
|
def gradscaler_init():
|
|
torch.xpu.amp.GradScaler = ipex.cpu.autocast._grad_scaler.GradScaler
|
|
torch.xpu.amp.GradScaler._unscale_grads_ = _unscale_grads_
|
|
torch.xpu.amp.GradScaler.unscale_ = unscale_
|
|
torch.xpu.amp.GradScaler.update = update
|
|
return torch.xpu.amp.GradScaler
|