mirror of
https://github.com/pytorch/pytorch.git
synced 2025-10-20 21:14:14 +08:00
[torch/utils][Code Clean] Clean asserts in hipify/, jit/, model_dump and tensorboard of torch/utils (#165311)
Including: - `torch/utils/hipify/` - `torch/utils/jit/` - `torch/utils/model_dump/` - `torch/utils/tensorboard/` Fixes part of #164878 Pull Request resolved: https://github.com/pytorch/pytorch/pull/165311 Approved by: https://github.com/albanD
This commit is contained in:
KarhouTam
committed by
PyTorch MergeBot
PyTorch MergeBot
parent
45b8c0f75c
commit
bf5aeb3148
@ -510,9 +510,9 @@ class Mod(sympy.Function):
|
||||
|
||||
# Evaluate if they are both literals.
|
||||
if q.is_Number and p.is_Number:
|
||||
if not (p >= 0):
|
||||
if p < 0:
|
||||
raise AssertionError(p)
|
||||
if not (q >= 1):
|
||||
if q < 1:
|
||||
raise AssertionError(q)
|
||||
return p % q
|
||||
|
||||
|
||||
@ -548,7 +548,8 @@ def get_hip_file_path(rel_filepath, is_pytorch_extension=False):
|
||||
"""
|
||||
# At the moment, some PyTorch source files are HIPified in place. The predicate
|
||||
# is_out_of_place tells us if this is the case or not.
|
||||
assert not os.path.isabs(rel_filepath)
|
||||
if os.path.isabs(rel_filepath):
|
||||
raise AssertionError("rel_filepath must be a relative path")
|
||||
if not is_pytorch_extension and not is_out_of_place(rel_filepath):
|
||||
return rel_filepath
|
||||
|
||||
@ -615,7 +616,8 @@ def get_hip_file_path(rel_filepath, is_pytorch_extension=False):
|
||||
|
||||
|
||||
def is_out_of_place(rel_filepath):
|
||||
assert not os.path.isabs(rel_filepath)
|
||||
if os.path.isabs(rel_filepath):
|
||||
raise AssertionError("rel_filepath must be a relative path")
|
||||
if rel_filepath.startswith("torch/"):
|
||||
return False
|
||||
if rel_filepath.startswith("third_party/nvfuser/"):
|
||||
@ -627,7 +629,8 @@ def is_out_of_place(rel_filepath):
|
||||
|
||||
# Keep this synchronized with includes/ignores in build_amd.py
|
||||
def is_pytorch_file(rel_filepath):
|
||||
assert not os.path.isabs(rel_filepath)
|
||||
if os.path.isabs(rel_filepath):
|
||||
raise AssertionError("rel_filepath must be a relative path")
|
||||
if rel_filepath.startswith("aten/"):
|
||||
if rel_filepath.startswith("aten/src/ATen/core/"):
|
||||
return False
|
||||
@ -658,7 +661,8 @@ def is_special_file(rel_filepath):
|
||||
return False
|
||||
|
||||
def is_caffe2_gpu_file(rel_filepath):
|
||||
assert not os.path.isabs(rel_filepath)
|
||||
if os.path.isabs(rel_filepath):
|
||||
raise AssertionError("rel_filepath must be a relative path")
|
||||
if rel_filepath.startswith("c10/cuda"):
|
||||
return True
|
||||
filename = os.path.basename(rel_filepath)
|
||||
@ -784,7 +788,8 @@ PYTORCH_MAP: dict[str, object] = {}
|
||||
PYTORCH_SPECIAL_MAP = {}
|
||||
|
||||
for mapping in CUDA_TO_HIP_MAPPINGS:
|
||||
assert isinstance(mapping, Mapping)
|
||||
if not isinstance(mapping, Mapping):
|
||||
raise TypeError("Expected each mapping in CUDA_TO_HIP_MAPPINGS to be a Mapping")
|
||||
for src, value in mapping.items():
|
||||
dst = value[0]
|
||||
meta_data = value[1:]
|
||||
|
||||
@ -32,10 +32,14 @@ def make_tensor_from_type(inp_type: torch._C.TensorType):
|
||||
stride = inp_type.strides()
|
||||
device = inp_type.device()
|
||||
dtype = inp_type.dtype()
|
||||
assert size is not None
|
||||
assert stride is not None
|
||||
assert device is not None
|
||||
assert dtype is not None
|
||||
if size is None:
|
||||
raise AssertionError("make_tensor_from_type: 'size' is None (inp_type.sizes() returned None)")
|
||||
if stride is None:
|
||||
raise AssertionError("make_tensor_from_type: 'stride' is None (inp_type.strides() returned None)")
|
||||
if device is None:
|
||||
raise AssertionError("make_tensor_from_type: 'device' is None (inp_type.device() returned None)")
|
||||
if dtype is None:
|
||||
raise AssertionError("make_tensor_from_type: 'dtype' is None (inp_type.dtype() returned None)")
|
||||
return torch.empty_strided(size=size, stride=stride, device=device, dtype=dtype)
|
||||
|
||||
def load_graph_and_inputs(ir: str) -> tuple[Any, list[Any]]:
|
||||
@ -81,7 +85,8 @@ def run_test(ir, inputs, *, warmup_runs=10, test_runs=20) -> float:
|
||||
if isinstance(input, torch.Tensor):
|
||||
is_cpu = input.device.type == "cpu"
|
||||
break
|
||||
assert is_cpu is not None
|
||||
if is_cpu is None:
|
||||
raise AssertionError("No tensor found in inputs")
|
||||
|
||||
out = time_cpu(graph, inputs, test_runs) if is_cpu else time_cuda(graph, inputs, test_runs)
|
||||
return out
|
||||
|
||||
@ -87,19 +87,31 @@ __all__ = ['get_storage_info', 'hierarchical_pickle', 'get_model_info', 'get_inl
|
||||
'burn_in_info', 'get_info_and_burn_skeleton']
|
||||
|
||||
def get_storage_info(storage):
|
||||
assert isinstance(storage, torch.utils.show_pickle.FakeObject)
|
||||
assert storage.module == "pers"
|
||||
assert storage.name == "obj"
|
||||
assert storage.state is None
|
||||
assert isinstance(storage.args, tuple)
|
||||
assert len(storage.args) == 1
|
||||
if not isinstance(storage, torch.utils.show_pickle.FakeObject):
|
||||
raise AssertionError(f"storage is not FakeObject: {type(storage)}")
|
||||
if storage.module != "pers":
|
||||
raise AssertionError(f"storage.module is not 'pers': {storage.module!r}")
|
||||
if storage.name != "obj":
|
||||
raise AssertionError(f"storage.name is not 'obj': {storage.name!r}")
|
||||
if storage.state is not None:
|
||||
raise AssertionError(f"storage.state is not None: {storage.state!r}")
|
||||
if not isinstance(storage.args, tuple):
|
||||
raise AssertionError(f"storage.args is not a tuple: {type(storage.args)}")
|
||||
if len(storage.args) != 1:
|
||||
raise AssertionError(f"len(storage.args) is not 1: {len(storage.args)}")
|
||||
sa = storage.args[0]
|
||||
assert isinstance(sa, tuple)
|
||||
assert len(sa) == 5
|
||||
assert sa[0] == "storage"
|
||||
assert isinstance(sa[1], torch.utils.show_pickle.FakeClass)
|
||||
assert sa[1].module == "torch"
|
||||
assert sa[1].name.endswith("Storage")
|
||||
if not isinstance(sa, tuple):
|
||||
raise AssertionError(f"sa is not a tuple: {type(sa)}")
|
||||
if len(sa) != 5:
|
||||
raise AssertionError(f"len(sa) is not 5: {len(sa)}")
|
||||
if sa[0] != "storage":
|
||||
raise AssertionError(f"sa[0] is not 'storage': {sa[0]!r}")
|
||||
if not isinstance(sa[1], torch.utils.show_pickle.FakeClass):
|
||||
raise AssertionError(f"sa[1] is not FakeClass: {type(sa[1])}")
|
||||
if sa[1].module != "torch":
|
||||
raise AssertionError(f"sa[1].module is not 'torch': {sa[1].module!r}")
|
||||
if not sa[1].name.endswith("Storage"):
|
||||
raise AssertionError(f"sa[1].name does not end with 'Storage': {sa[1].name!r}")
|
||||
storage_info = [sa[1].name.replace("Storage", "")] + list(sa[2:])
|
||||
return storage_info
|
||||
|
||||
@ -124,52 +136,69 @@ def hierarchical_pickle(data):
|
||||
if (
|
||||
typename.startswith(('__torch__.', 'torch.jit.LoweredWrapper.', 'torch.jit.LoweredModule.'))
|
||||
):
|
||||
assert data.args == ()
|
||||
if data.args != ():
|
||||
raise AssertionError("data.args is not ()")
|
||||
return {
|
||||
"__module_type__": typename,
|
||||
"state": hierarchical_pickle(data.state),
|
||||
}
|
||||
if typename == "torch._utils._rebuild_tensor_v2":
|
||||
assert data.state is None
|
||||
if data.state is not None:
|
||||
raise AssertionError("data.state is not None")
|
||||
storage, offset, size, stride, requires_grad, *_ = data.args
|
||||
storage_info = get_storage_info(storage)
|
||||
return {"__tensor_v2__": [storage_info, offset, size, stride, requires_grad]}
|
||||
if typename == "torch._utils._rebuild_qtensor":
|
||||
assert data.state is None
|
||||
if data.state is not None:
|
||||
raise AssertionError("data.state is not None")
|
||||
storage, offset, size, stride, quantizer, requires_grad, *_ = data.args
|
||||
storage_info = get_storage_info(storage)
|
||||
assert isinstance(quantizer, tuple)
|
||||
assert isinstance(quantizer[0], torch.utils.show_pickle.FakeClass)
|
||||
assert quantizer[0].module == "torch"
|
||||
if not isinstance(quantizer, tuple):
|
||||
raise AssertionError("quantizer is not a tuple")
|
||||
if not isinstance(quantizer[0], torch.utils.show_pickle.FakeClass):
|
||||
raise AssertionError("quantizer[0] is not a FakeClass")
|
||||
if quantizer[0].module != "torch":
|
||||
raise AssertionError("quantizer[0].module is not torch")
|
||||
if quantizer[0].name == "per_tensor_affine":
|
||||
assert len(quantizer) == 3
|
||||
assert isinstance(quantizer[1], float)
|
||||
assert isinstance(quantizer[2], int)
|
||||
if len(quantizer) != 3:
|
||||
raise AssertionError("len(quantizer) is not 3")
|
||||
if not isinstance(quantizer[1], float):
|
||||
raise AssertionError("quantizer[1] is not a float")
|
||||
if not isinstance(quantizer[2], int):
|
||||
raise AssertionError("quantizer[2] is not an int")
|
||||
quantizer_extra = list(quantizer[1:3])
|
||||
else:
|
||||
quantizer_extra = []
|
||||
quantizer_json = [quantizer[0].name] + quantizer_extra
|
||||
return {"__qtensor__": [storage_info, offset, size, stride, quantizer_json, requires_grad]}
|
||||
if typename == "torch.jit._pickle.restore_type_tag":
|
||||
assert data.state is None
|
||||
if data.state is not None:
|
||||
raise AssertionError("data.state is not None")
|
||||
obj, typ = data.args
|
||||
assert isinstance(typ, str)
|
||||
if not isinstance(typ, str):
|
||||
raise AssertionError("typ is not a string")
|
||||
return hierarchical_pickle(obj)
|
||||
if re.fullmatch(r"torch\.jit\._pickle\.build_[a-z]+list", typename):
|
||||
assert data.state is None
|
||||
if data.state is not None:
|
||||
raise AssertionError("data.state is not None")
|
||||
ls, = data.args
|
||||
assert isinstance(ls, list)
|
||||
if not isinstance(ls, list):
|
||||
raise AssertionError("ls is not a list")
|
||||
return hierarchical_pickle(ls)
|
||||
if typename == "torch.device":
|
||||
assert data.state is None
|
||||
if data.state is not None:
|
||||
raise AssertionError("data.state is not None")
|
||||
name, = data.args
|
||||
assert isinstance(name, str)
|
||||
if not isinstance(name, str):
|
||||
raise AssertionError("name is not a string")
|
||||
# Just forget that it was a device and return the name.
|
||||
return name
|
||||
if typename == "builtin.UnicodeDecodeError":
|
||||
assert data.state is None
|
||||
if data.state is not None:
|
||||
raise AssertionError("data.state is not None")
|
||||
msg, = data.args
|
||||
assert isinstance(msg, str)
|
||||
if not isinstance(msg, str):
|
||||
raise AssertionError("msg is not a string")
|
||||
# Hack: Pretend this is a module so we don't need custom serialization.
|
||||
# Hack: Wrap the message in a tuple so it looks like a nice state object.
|
||||
# TODO: Undo at least that second hack. We should support string states.
|
||||
@ -223,11 +252,13 @@ def get_model_info(
|
||||
"file_size": zi.file_size,
|
||||
}
|
||||
)
|
||||
assert path_prefix is not None
|
||||
if path_prefix is None:
|
||||
raise AssertionError("path_prefix is None")
|
||||
version = zf.read(path_prefix + "/version").decode("utf-8").strip()
|
||||
|
||||
def get_pickle(name):
|
||||
assert path_prefix is not None
|
||||
if path_prefix is None:
|
||||
raise AssertionError("path_prefix is None")
|
||||
with zf.open(path_prefix + f"/{name}.pkl") as handle:
|
||||
raw = torch.utils.show_pickle.DumpUnpickler(handle, catch_invalid_utf8=True).load()
|
||||
return hierarchical_pickle(raw)
|
||||
@ -285,7 +316,8 @@ def get_model_info(
|
||||
for di, di_next in itertools.pairwise(debug_info):
|
||||
start, source_range, *_ = di
|
||||
end = di_next[0]
|
||||
assert end > start
|
||||
if end <= start:
|
||||
raise AssertionError("end is not greater than start")
|
||||
source, s_start, s_end = source_range
|
||||
s_text, s_file, s_line = source
|
||||
# TODO: Handle this case better. TorchScript ranges are in bytes,
|
||||
|
||||
@ -25,9 +25,10 @@ def make_tsv(metadata, save_path, metadata_header=None):
|
||||
if not metadata_header:
|
||||
metadata = [str(x) for x in metadata]
|
||||
else:
|
||||
assert len(metadata_header) == len(
|
||||
if len(metadata_header) != len(
|
||||
metadata[0]
|
||||
), "len of header must be equal to the number of columns in metadata"
|
||||
):
|
||||
raise AssertionError("len of header must be equal to the number of columns in metadata")
|
||||
metadata = ["\t".join(str(e) for e in l) for l in [metadata_header] + metadata]
|
||||
|
||||
metadata_bytes = tf.compat.as_bytes("\n".join(metadata) + "\n")
|
||||
|
||||
@ -76,10 +76,12 @@ def _prepare_video(V):
|
||||
|
||||
def make_grid(I, ncols=8):
|
||||
# I: N1HW or N3HW
|
||||
assert isinstance(I, np.ndarray), "plugin error, should pass numpy array here"
|
||||
if not isinstance(I, np.ndarray):
|
||||
raise AssertionError("plugin error, should pass numpy array here")
|
||||
if I.shape[1] == 1:
|
||||
I = np.concatenate([I, I, I], 1)
|
||||
assert I.ndim == 4 and I.shape[1] == 3
|
||||
if I.ndim != 4 or I.shape[1] != 3:
|
||||
raise AssertionError("Input should be a 4D numpy array with 3 channels")
|
||||
nimg = I.shape[0]
|
||||
H = I.shape[2]
|
||||
W = I.shape[3]
|
||||
@ -101,13 +103,12 @@ def make_grid(I, ncols=8):
|
||||
|
||||
|
||||
def convert_to_HWC(tensor, input_format): # tensor: numpy array
|
||||
assert len(set(input_format)) == len(
|
||||
input_format
|
||||
), f"You can not use the same dimension shordhand twice. input_format: {input_format}"
|
||||
assert len(tensor.shape) == len(
|
||||
input_format
|
||||
), f"size of input tensor and input format are different. \
|
||||
tensor shape: {tensor.shape}, input_format: {input_format}"
|
||||
if len(set(input_format)) != len(input_format):
|
||||
raise AssertionError(f"You can not use the same dimension shordhand twice. \
|
||||
input_format: {input_format}")
|
||||
if len(tensor.shape) != len(input_format):
|
||||
raise AssertionError(f"size of input tensor and input format are different. \
|
||||
tensor shape: {tensor.shape}, input_format: {input_format}")
|
||||
input_format = input_format.upper()
|
||||
|
||||
if len(input_format) == 4:
|
||||
|
||||
@ -370,9 +370,9 @@ def scalar(name, tensor, collections=None, new_style=False, double_precision=Fal
|
||||
ValueError: If tensor has the wrong shape or type.
|
||||
"""
|
||||
tensor = make_np(tensor).squeeze()
|
||||
assert (
|
||||
tensor.ndim == 0
|
||||
), f"Tensor should contain one element (0 dimensions). Was given size: {tensor.size} and {tensor.ndim} dimensions."
|
||||
if tensor.ndim != 0:
|
||||
raise AssertionError(f"Tensor should contain one element (0 dimensions). \
|
||||
Was given size: {tensor.size} and {tensor.ndim} dimensions.")
|
||||
# python float is double precision in numpy
|
||||
scalar = float(tensor)
|
||||
if new_style:
|
||||
@ -700,7 +700,8 @@ def audio(tag, tensor, sample_rate=44100):
|
||||
if abs(array).max() > 1:
|
||||
print("warning: audio amplitude out of range, auto clipped.")
|
||||
array = array.clip(-1, 1)
|
||||
assert array.ndim == 1, "input tensor should be 1 dimensional."
|
||||
if array.ndim != 1:
|
||||
raise AssertionError("input tensor should be 1 dimensional.")
|
||||
array = (array * np.iinfo(np.int16).max).astype("<i2")
|
||||
|
||||
import io
|
||||
@ -731,7 +732,8 @@ def custom_scalars(layout):
|
||||
for chart_name, chart_metadata in v.items():
|
||||
tags = chart_metadata[1]
|
||||
if chart_metadata[0] == "Margin":
|
||||
assert len(tags) == 3
|
||||
if len(tags) != 3:
|
||||
raise AssertionError("len(tags) != 3")
|
||||
mgcc = layout_pb2.MarginChartContent(
|
||||
series=[
|
||||
layout_pb2.MarginChartContent.Series(
|
||||
|
||||
@ -420,7 +420,8 @@ class SummaryWriter:
|
||||
fw_logdir = self._get_file_writer().get_logdir()
|
||||
for tag, scalar_value in tag_scalar_dict.items():
|
||||
fw_tag = fw_logdir + "/" + main_tag.replace("/", "_") + "_" + tag
|
||||
assert self.all_writers is not None
|
||||
if self.all_writers is None:
|
||||
raise AssertionError("self.all_writers is None")
|
||||
if fw_tag in self.all_writers.keys():
|
||||
fw = self.all_writers[fw_tag]
|
||||
else:
|
||||
@ -930,20 +931,19 @@ class SummaryWriter:
|
||||
fs.makedirs(save_path)
|
||||
|
||||
if metadata is not None:
|
||||
assert mat.shape[0] == len(
|
||||
if mat.shape[0] != len(
|
||||
metadata
|
||||
), "#labels should equal with #data points"
|
||||
):
|
||||
raise AssertionError("#labels should equal with #data points")
|
||||
make_tsv(metadata, save_path, metadata_header=metadata_header)
|
||||
|
||||
if label_img is not None:
|
||||
assert (
|
||||
mat.shape[0] == label_img.shape[0]
|
||||
), "#images should equal with #data points"
|
||||
if mat.shape[0] != label_img.shape[0]:
|
||||
raise AssertionError("#images should equal with #data points")
|
||||
make_sprite(label_img, save_path)
|
||||
|
||||
assert (
|
||||
mat.ndim == 2
|
||||
), "mat should be 2D, where mat.size(0) is the number of data points"
|
||||
if mat.ndim != 2:
|
||||
raise AssertionError("mat should be 2D, where mat.size(0) is the number of data points")
|
||||
make_mat(mat, save_path)
|
||||
|
||||
# Filesystem doesn't necessarily have append semantics, so we store an
|
||||
@ -1094,7 +1094,8 @@ class SummaryWriter:
|
||||
torch._C._log_api_usage_once(
|
||||
"tensorboard.logging.add_custom_scalars_marginchart"
|
||||
)
|
||||
assert len(tags) == 3
|
||||
if len(tags) != 3:
|
||||
raise AssertionError(f"Expected 3 tags, got {len(tags)}.")
|
||||
layout = {category: {title: ["Margin", tags]}}
|
||||
self._get_file_writer().add_summary(custom_scalars(layout))
|
||||
|
||||
|
||||
Reference in New Issue
Block a user