Skip to content

OP报错信息优化 #27301

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
gavin1332 merged 2 commits into from
Sep 17, 2020
Merged

OP报错信息优化 #27301

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
0b30c88
paddle/fluid/operators/distributed_ops OP报错信息优化
gavin1332 Sep 14, 2020
5548530
fix typo
gavin1332 Sep 14, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions paddle/fluid/operators/distributed_ops/fake_init_op.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -43,9 +43,9 @@ class FakeInitOp : public framework::OperatorBase {
tensor = out_var.GetMutable<framework::SelectedRows>()->mutable_value();
tensor->Resize(framework::make_ddim(Attr<std::vector<int64_t>>("shape")));
} else {
PADDLE_THROW(
PADDLE_THROW(platform::errors::InvalidArgument(
"fake init op's output only"
"supports SelectedRows and LoDTensor");
"supports SelectedRows and LoDTensor"));
}
}
};
Expand Down
45 changes: 35 additions & 10 deletions paddle/fluid/operators/distributed_ops/listen_and_serv_op.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -134,7 +134,10 @@ void ListenAndServOp::RunSyncLoop(
auto optimize_blocks =
Attr<std::vector<framework::BlockDesc *>>(kOptimizeBlocks);
PADDLE_ENFORCE_GE(num_blocks, 2,
"server program should have at least 2 blocks");
platform::errors::PreconditionNotMet(
"Invalid number of blocks in server program. Expected "
"equal or greater than 2. Recieved %zu",
num_blocks));

// Prepare all the server block
std::vector<int> optimize_blocks_list;
Expand Down Expand Up @@ -218,7 +221,8 @@ void ListenAndServOp::ResetReceivedVars(framework::Scope *recv_scope,
VLOG(3) << "reset sparse var: " << varname;
var->GetMutable<framework::SelectedRows>()->mutable_rows()->clear();
} else {
PADDLE_THROW("The type of sparse var should be SelectedRows");
PADDLE_THROW(platform::errors::PreconditionNotMet(
"The type of sparse var should be SelectedRows"));
}
}
if (UNLIKELY(reset_all)) {
Expand All @@ -235,7 +239,8 @@ void ListenAndServOp::ResetReceivedVars(framework::Scope *recv_scope,
math::set_constant(*dev_ctx, var->GetMutable<framework::Tensor>(),
static_cast<float>(0));
} else {
PADDLE_THROW("The type of dense var should be in [LoDTensor, Tensor]");
PADDLE_THROW(platform::errors::PreconditionNotMet(
"The type of dense var should be in [LoDTensor, Tensor]"));
}
}
}
Expand All @@ -254,8 +259,15 @@ void ListenAndServOp::RunAsyncLoop(framework::Executor *executor,
std::vector<std::string> pieces;
split(grad_and_id, ':', &pieces);
VLOG(3) << "after split, key = " << pieces[0] << ", id=" << pieces[1];
PADDLE_ENFORCE_EQ(pieces.size(), 2);
PADDLE_ENFORCE_EQ(out_map->count(pieces[0]), 0);
PADDLE_ENFORCE_EQ(pieces.size(), 2,
platform::errors::PreconditionNotMet(
"Invalid format of grad_and_id argument. "
"Expected \"grad:block_id\". Recieved %s",
grad_and_id.c_str()));
PADDLE_ENFORCE_EQ(out_map->count(pieces[0]), 0,
platform::errors::AlreadyExists(
"The gradient name %s has already existed in out_map",
pieces[0].c_str()));

int block_id = std::stoi(pieces[1]);
(*out_map)[pieces[0]] = block_id;
Expand All @@ -267,7 +279,10 @@ void ListenAndServOp::RunAsyncLoop(framework::Executor *executor,

size_t num_blocks = program->Size();
PADDLE_ENFORCE_GE(num_blocks, 2,
"server program should have at least 2 blocks");
platform::errors::PreconditionNotMet(
"Invalid number of blocks in server program. Expected "
"equal or greater than 2. Recieved %zu",
num_blocks));
std::vector<int> block_list;
for (size_t blkid = 1; blkid < num_blocks; ++blkid) {
block_list.push_back(blkid);
Expand Down Expand Up @@ -342,9 +357,9 @@ void ListenAndServOp::CacheVarsType(const std::vector<std::string> &varnames,
var->IsType<framework::Tensor>()) {
dense_vars_.push_back(varname);
} else {
PADDLE_THROW(
PADDLE_THROW(platform::errors::PreconditionNotMet(
"The type of received var should be in [SelectedRows, LoDTensor, "
"Tensor].");
"Tensor]."));
}
}
}
Expand Down Expand Up @@ -450,7 +465,12 @@ void ListenAndServOp::RunImpl(const framework::Scope &scope,
split(prefetch_var_name_and_id, ':', &pieces);
VLOG(3) << "after split, prefetch_var = " << pieces[0]
<< ", id=" << pieces[1];
PADDLE_ENFORCE_EQ(pieces.size(), 2);
PADDLE_ENFORCE_EQ(
pieces.size(), 2,
platform::errors::PreconditionNotMet(
"Invalid format of prefetch_var_name_and_id argument. "
"Expected \"xxx:xxx\". Recieved %s",
prefetch_var_name_and_id.c_str()));

int block_id = std::stoi(pieces[1]);
prefetch_block_id_list.push_back(block_id);
Expand All @@ -476,7 +496,12 @@ void ListenAndServOp::RunImpl(const framework::Scope &scope,
sparse_grad_name_to_param_name_str) {
std::vector<std::string> pieces;
split(sparse_grad_name_and_param_name, ':', &pieces);
PADDLE_ENFORCE_EQ(pieces.size(), 2);
PADDLE_ENFORCE_EQ(
pieces.size(), 2,
platform::errors::PreconditionNotMet(
"Invalid format of sparse_grad_name_and_param_name argument. "
"Expected \"xxx:xxx\". Recieved %s",
sparse_grad_name_and_param_name.c_str()));
VLOG(3) << "after split, sparse_grad_name = " << pieces[0]
<< ", param_name = " << pieces[1];
sparse_grad_name_to_param_name[pieces[0]] = pieces[1];
Expand Down