| # Copyright (c) Facebook, Inc. and its affiliates. |
| # All rights reserved. |
| # |
| # This source code is licensed under the BSD-style license found in the |
| # LICENSE file in the root directory of this source tree. |
| |
| from torch.testing._internal.common_utils import TestCase, run_tests, is_iterable_of_tensors |
| import torch |
| import torch.nn.functional as F |
| from torch import Tensor |
| import functools |
| import itertools |
| import copy |
| import warnings |
| import unittest |
| from torch.testing._internal.common_device_type import instantiate_device_type_tests, \ |
| skipCUDAIfNoMagma |
| from torch.testing._internal.common_device_type import ops, onlyCPU |
| from functorch_lagging_op_db import functorch_lagging_op_db |
| from common_utils import ( |
| parameterized, |
| instantiate_parameterized_methods, |
| get_fallback_and_vmap_exhaustive, |
| get_exhaustive_batched_inputs, |
| ) |
| import types |
| from torch.utils._pytree import tree_flatten, tree_unflatten, tree_map |
| from functorch import grad, vjp, vmap |
| from functorch._src.eager_transforms import _as_tuple |
| |
| # Version of autograd.grad that handles outputs that don't depend on inputs |
| def _autograd_grad(outputs, inputs, grad_outputs=None, retain_graph=False, create_graph=True): |
| inputs, inputs_spec = tree_flatten(inputs) |
| result = [torch.zeros_like(inp) for inp in inputs] |
| diff_argnums = tuple(i for i, inp in enumerate(inputs) if inp.requires_grad) |
| inputs = tuple(inputs[i] for i in diff_argnums) |
| if grad_outputs is None: |
| diff_outputs = tuple(out for out in outputs if out.requires_grad) |
| else: |
| something = [(out, go) for out, go in zip(outputs, grad_outputs) |
| if out.requires_grad] |
| if len(something) == 0: |
| diff_outputs, grad_outputs = (), () |
| else: |
| diff_outputs, grad_outputs = zip(*something) |
| if len(diff_outputs) == 0: |
| return tuple(torch.zeros_like(inp) for inp in inputs) |
| grad_inputs = torch.autograd.grad(diff_outputs, inputs, grad_outputs, |
| retain_graph=retain_graph, |
| create_graph=create_graph, |
| allow_unused=True) |
| grad_inputs = tuple(torch.zeros_like(inp) if gi is None else gi |
| for gi, inp in zip(grad_inputs, inputs)) |
| for idx, grad_inp in zip(diff_argnums, grad_inputs): |
| result[idx] = grad_inp |
| return tree_unflatten(result, inputs_spec) |
| |
| |
| def diff_arg(arg): |
| if is_iterable_of_tensors(arg): |
| if all([a.requires_grad for a in arg]): |
| return True |
| if all([not a.requires_grad for a in arg]): |
| return False |
| raise RuntimeError("NYI: The test runner can't handle this") |
| return isinstance(arg, Tensor) and arg.requires_grad |
| |
| |
| # Given f, returns an f' such that: |
| # - f' takes only positional arguments |
| # - All arguments to f' are floating-point Tensors |
| # - All outputs of f' are floating-point Tensors |
| def normalize_op_for_vjp2(f, args, kwargs, output_process_fn_grad=None): |
| flat_args, args_spec = tree_flatten(args) |
| diff_argnums = tuple(i for i, arg in enumerate(flat_args) if diff_arg(arg)) |
| assert len(diff_argnums) > 0 |
| primals = tuple(flat_args[i] for i in diff_argnums) |
| |
| @functools.wraps(f) |
| def wrapped(*primals): |
| _args = list(flat_args) |
| for num, arg in zip(diff_argnums, primals): |
| _args[num] = arg |
| _args = tree_unflatten(_args, args_spec) |
| result = f(*_args, **kwargs) |
| if output_process_fn_grad is not None: |
| result = output_process_fn_grad(result) |
| if isinstance(result, tuple): |
| # TODO: Remove the following hack for namedtuples |
| result = tuple(result) |
| result = tuple(r for r in result if torch.is_floating_point(r)) |
| assert len(result) > 0 |
| return result |
| return wrapped, primals |
| |
| |
| def normalize_op_for_vjp(f, sample): |
| args = tuple([sample.input] + list(sample.args)) |
| return normalize_op_for_vjp2(f, args, sample.kwargs, sample.output_process_fn_grad) |
| |
| |
| def ref_vjp(f, *primals): |
| result = f(*primals) |
| |
| def wrapped(cotangents): |
| return _autograd_grad(_as_tuple(result), primals, _as_tuple(cotangents)) |
| |
| return result, wrapped |
| |
| |
| def normalize_op_for_vjp_vjp(f, sample): |
| fn, primals = normalize_op_for_vjp(f, sample) |
| result = fn(*primals) |
| cotangents = _as_tuple( |
| tree_map(lambda x: torch.randn_like(x, requires_grad=True), result)) |
| num_primals = len(primals) |
| args = (*primals, *cotangents) |
| |
| @functools.wraps(f) |
| def wrapped(*args): |
| primals = args[:num_primals] |
| cotangents = args[num_primals:] |
| result, vjp_fn = vjp(fn, *primals) |
| if isinstance(result, torch.Tensor): |
| assert len(cotangents) == 1 |
| cotangents = cotangents[0] |
| return vjp_fn(cotangents) |
| |
| return wrapped, args |
| |
| |
| def is_inplace(op, variant): |
| if hasattr(variant, "__wrapped__"): |
| return variant.__wrapped__ is op.get_inplace() |
| return variant is op.get_inplace() |
| |
| |
| vjp_fail = { |
| '__getitem__', |
| '__rpow__', |
| 'linalg.cholesky', |
| 'linalg.inv', |
| 'linalg.matrix_norm', |
| 'linalg.matrix_power', |
| 'linalg.norm', |
| 'nanquantile', |
| 'quantile', |
| 'tensor_split', |
| 'norm', |
| 'to_sparse', |
| 'cross', |
| } |
| |
| |
| class TestOperators(TestCase): |
| @ops(functorch_lagging_op_db, allowed_dtypes=(torch.float,)) |
| def test_grad(self, device, dtype, op): |
| if op.name in vjp_fail: |
| self.skipTest("Skipped; Expected failures") |
| return |
| |
| if not op.supports_autograd: |
| self.skipTest("Skipped! Autograd not supported.") |
| return |
| |
| samples = op.sample_inputs(device, dtype, requires_grad=True) |
| |
| # TODO: test in-place |
| if is_inplace(op, op.get_op()): |
| self.skipTest("Skipped! NYI: inplace-testing not supported.") |
| return |
| |
| for sample in samples: |
| args = [sample.input] + list(sample.args) |
| kwargs = sample.kwargs |
| |
| diff_argnums = tuple(i for i, arg in enumerate(args) if diff_arg(arg)) |
| assert len(diff_argnums) > 0 |
| diff_args = tuple(args[i] for i in diff_argnums) |
| |
| def wrapped_fn(*args, **kwargs): |
| result = op(*args, **kwargs) |
| if sample.output_process_fn_grad is not None: |
| result = sample.output_process_fn_grad(result) |
| |
| # Reduce into single value for grad |
| if isinstance(result, torch.Tensor): |
| return result.sum() |
| result = sum([res.sum() for res in result]) |
| return result |
| |
| result = grad(wrapped_fn, diff_argnums)(*args, **kwargs) |
| expected = _autograd_grad(_as_tuple(wrapped_fn(*args, **kwargs)), diff_args) |
| |
| self.assertEqual(result, expected) |
| |
| @ops(functorch_lagging_op_db, allowed_dtypes=(torch.float,)) |
| def test_vjp(self, device, dtype, op): |
| if op.name in vjp_fail: |
| self.skipTest("Skipped; Expected failures") |
| return |
| |
| if not op.supports_autograd: |
| self.skipTest("Skipped! Autograd not supported.") |
| return |
| |
| samples = op.sample_inputs(device, dtype, requires_grad=True) |
| |
| # TODO: test in-place |
| if is_inplace(op, op.get_op()): |
| self.skipTest("Skipped! NYI: inplace-testing not supported.") |
| return |
| |
| for sample in samples: |
| fn, primals = normalize_op_for_vjp(op, sample) |
| result = fn(*primals) |
| cotangents = tree_map(lambda x: torch.randn_like(x), result) |
| |
| _, vjp_fn = vjp(fn, *primals) |
| result_vjps = vjp_fn(cotangents) |
| |
| _, vjp_fn = ref_vjp(fn, *primals) |
| expected_vjps = vjp_fn(cotangents) |
| |
| self.assertEqual(result_vjps, expected_vjps) |
| |
| @ops(functorch_lagging_op_db, allowed_dtypes=(torch.float,)) |
| def test_vjpvjp(self, device, dtype, op): |
| op_skip = set({ |
| }) |
| op_skip = op_skip.union(vjp_fail) |
| if op.name in op_skip: |
| self.skipTest("Skipped; Expected failures") |
| return |
| |
| if not op.supports_autograd: |
| self.skipTest("Skipped! Autograd not supported.") |
| return |
| if not op.supports_gradgrad: |
| self.skipTest("Skipped! Operation does not support gradgrad") |
| return |
| |
| samples = op.sample_inputs(device, dtype, requires_grad=True) |
| |
| # TODO: test in-place |
| if is_inplace(op, op.get_op()): |
| self.skipTest("Skipped! NYI: inplace-testing not supported.") |
| return |
| |
| for sample in samples: |
| fn, args = normalize_op_for_vjp_vjp(op, sample) |
| result = fn(*args) |
| cotangents = tree_map(lambda x: torch.randn_like(x), result) |
| |
| # Compute vjp of vjp |
| _, vjp_fn = vjp(fn, *args) |
| result_vjps = vjp_fn(cotangents) |
| |
| # Compute ref_vjp of vjp. We could have done ref_vjp of ref_vjp, |
| # but since we're confident that vjp works by itself, this is |
| # an equivalent way to test that. |
| _, vjp_fn = ref_vjp(fn, *args) |
| expected_vjps = vjp_fn(cotangents) |
| |
| self.assertEqual(result_vjps, expected_vjps) |
| |
| @ops(functorch_lagging_op_db, allowed_dtypes=(torch.float,)) |
| def test_vmapvjp(self, device, dtype, op): |
| op_skip = { |
| 'broadcast_to', |
| 'dsplit', |
| 'dstack', |
| 'einsum', |
| 'gradient', |
| 'hsplit', |
| 'hstack', |
| 'linalg.multi_dot', |
| 'lu', |
| 'moveaxis', |
| 'positive', |
| 'ravel', |
| 'squeeze', |
| 'unfold', |
| 'vsplit', |
| 'vstack', |
| 'resolve_conj', |
| } |
| op_skip = op_skip.union(vjp_fail) |
| if op.name in op_skip: |
| self.skipTest("Skipped; Expected failures") |
| return |
| if not op.supports_autograd: |
| self.skipTest("Skipped! Autograd not supported.") |
| return |
| |
| samples = op.sample_inputs(device, dtype, requires_grad=True) |
| |
| # TODO: test in-place |
| if is_inplace(op, op.get_op()): |
| self.skipTest("Skipped! NYI: inplace-testing not supported.") |
| return |
| |
| for sample in samples: |
| fn, args = normalize_op_for_vjp(op, sample) |
| for loop_out, batched_out in get_fallback_and_vmap_exhaustive(fn, args, {}): |
| self.assertEqual(loop_out, batched_out) |
| |
| @ops(functorch_lagging_op_db, allowed_dtypes=(torch.float,)) |
| def test_vjpvmap(self, device, dtype, op): |
| op_skip = { |
| '__getitem__', |
| '__rpow__', |
| 'broadcast_to', |
| 'dsplit', |
| 'gradient', |
| 'hsplit', |
| 'linalg.cholesky', |
| 'linalg.inv', |
| 'linalg.matrix_norm', |
| 'linalg.matrix_power', |
| 'linalg.norm', |
| 'lu', |
| 'moveaxis', |
| 'nanquantile', |
| 'positive', |
| 'quantile', |
| 'ravel', |
| 'squeeze', |
| 'tensor_split', |
| 'unfold', |
| 'vsplit', |
| 'fill_', |
| 'norm', |
| 'resolve_conj', |
| 'to_sparse', |
| 'clamp', |
| } |
| if op.name in op_skip: |
| self.skipTest("Skipped; Expected failures") |
| return |
| |
| if not op.supports_autograd: |
| # If the op doesn't support autograd, vmap(op) won't either |
| self.skipTest("Skipped! Autograd not supported.") |
| return |
| |
| # TODO: test in-place |
| if is_inplace(op, op.get_op()): |
| self.skipTest("Skipped! NYI: inplace-testing not supported.") |
| return |
| |
| samples = op.sample_inputs(device, dtype, requires_grad=True) |
| |
| for sample in samples: |
| args = [sample.input] + list(sample.args) |
| kwargs = sample.kwargs |
| |
| for batched_args, in_dims, kwargs in get_exhaustive_batched_inputs(args, kwargs): |
| vmapped_op = vmap(op, in_dims) |
| fn, primals = normalize_op_for_vjp2(vmapped_op, batched_args, kwargs, |
| sample.output_process_fn_grad) |
| result = fn(*primals) |
| cotangents = tree_map(lambda x: torch.randn_like(x), result) |
| |
| _, vjp_fn = vjp(fn, *primals) |
| result_vjps = vjp_fn(cotangents) |
| |
| _, vjp_fn = ref_vjp(fn, *primals) |
| expected_vjps = vjp_fn(cotangents) |
| |
| self.assertEqual(result_vjps, expected_vjps) |
| |
| |
| only_for = ("cpu", "cuda") |
| instantiate_device_type_tests(TestOperators, globals(), only_for=only_for) |
| |
| if __name__ == '__main__': |
| run_tests() |
