Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/62135
The initial implementation of Adam with Smart Decay had an off-by-one error. This was in the summation of the geometric series used to calculate how much built-up momentum would have been discharged in skipped minibatches.
The unit tests should have caught these, but the testing strategy missed this because k, the "number of skipped minibatches" was always either 0 or so high that the impact of the bug was too small. The impact of the bug was proportional to 1/k. The testing strategy has also been adjusted to cover this bug.
Differential Revision: D29889309
fbshipit-source-id: b086c0efed5c27f621061e726533c73658daffc6
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/62058
This is the second diff in this stack. This diff includes the changes to DPER3; the first diff includes the changes to Caffe2.
We want to decay learning parameters properly. Previously this was not done when a parameter is absent from a minibatch. We fix this by keeping track of missed minibatches and making decay catch up accordingly.
The exponential moving averages (EMA) for the first and second moments used in Adam are updated only for parameters seen in a minibatch. Actually, for these parameters, 0 should be added to the EMAs and the EMAs should then be decayed by multiplying by beta1 and beta2 respectively.
To avoid the computational overhead of touching every parameter for every minibatch, we:
* keep track of the last time a parameter is seen
* instead of decaying the EMAs by multiplying by beta1 and beta2, we multiply by beta1^k and beta2^k, where k is the number of minibatches since the parameter was last seen.
We hope this will significantly improve the inconsistent learning parameter issue we have seen with Adam.
Differential Revision: D29638897
fbshipit-source-id: 18d8e227d72c2e23010ca81e0f6eeb78872c8d3c
Summary:
As GoogleTest `TEST` macro is non-compliant with it as well as `DEFINE_DISPATCH`
All changes but the ones to `.clang-tidy` are generated using following script:
```
for i in `find . -type f -iname "*.c*" -or -iname "*.h"|xargs grep cppcoreguidelines-avoid-non-const-global-variables|cut -f1 -d:|sort|uniq`; do sed -i "/\/\/ NOLINTNEXTLINE(cppcoreguidelines-avoid-non-const-global-variables)/d" $i; done
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/62008
Reviewed By: driazati, r-barnes
Differential Revision: D29838584
Pulled By: malfet
fbshipit-source-id: 1b2f8602c945bd4ce50a9bfdd204755556e31d13
Summary: A test case that triggers db_options with the save operator is missing.
Test Plan: buck test
Differential Revision: D29642719
fbshipit-source-id: 72b7374d40430398abac26dfe91538550525384d
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/61548
We want to decay learning parameters properly. Previously this was not done when a parameter is absent from a minibatch. We fix this by keeping track of missed minibatches and making decay catch up accordingly.
The exponential moving averages (EMA) for the first and second moments used in Adam are updated only for parameters seen in a minibatch. Actually, for these parameters, 0 should be added to the EMAs and the EMAs should then be decayed by multiplying by beta1 and beta2 respectively.
To avoid the computational overhead of touching every parameter for every minibatch, we:
* keep track of the last time a parameter is seen
* instead of decaying the EMAs by multiplying by beta1 and beta2, we multiply by beta1^k and beta2^k, where k is the number of minibatches since the parameter was last seen
* we calculate the amount of momentum that would have been discharged over the missed minibatches and update the weight accordingly.
Differential Revision: D29654246
fbshipit-source-id: 7a6cd7966eb1f31116d99dfce79a78b2d3ee9e3e
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/61551
We aim to enable rate limiter in C2 load, with a fix bandwidth limit.
This diff update LoadOp to pass down the manifold db options.
Test Plan:
```
buck test mode/opt caffe2/caffe2/python/operator_test:load_save_test
```
Differential Revision: D29639102
fbshipit-source-id: cf69549adadf4c7f12a8a2b7f3ca39092cab4b99
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/61488
We want to decay learning parameters properly. Previously this was not done when a parameter is absent from a minibatch. We fix this by keeping track of missed minibatches and making decay catch up accordingly.
The exponential moving averages (EMA) for the first and second moments used in Adam are updated only for parameters seen in a minibatch. Actually, for these parameters, 0 should be added to the EMAs and the EMAs should then be decayed by multiplying by beta1 and beta2 respectively.
To avoid the computational overhead of touching every parameter for every minibatch, we:
* keep track of the last time a parameter is seen
* instead of decaying the EMAs by multiplying by beta1 and beta2, we multiply by beta1^k and beta2^k, where k is the number of minibatches since the parameter was last seen.
Differential Revision: D27978269
fbshipit-source-id: e47524101ddfcb281c46c505b9b7a8f0835bc64a
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/60402
Add float64 data type support for ScatterWeightedSum for cases that 10^7 precision is not sufficient.
Test Plan: buck test caffe2/caffe2/python/operator_test:sparse_ops_test -- testScatterWeightedSum
Reviewed By: jianyuh
Differential Revision: D29190324
fbshipit-source-id: 871a60744694e901a2c7685a67350860745d6729
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/59775
This operator is similar to `GetAllBlobNames` but also returns the estimated
size required to serialize each node.
One goal of this operator is to allow checkpoint saving logic to estimate the
amount of space/bandwidth required to save a checkpoint when first starting
training, without actually serializing any blobs yet. Currently the
checkpointing logic uses `GetAllBlobNames` to determine the blobs to
checkpoint. It can instead be updated to use `EstimateAllBlobSizes` to also
get an estimate for how much space will be required for the checkpoint.
ghstack-source-id: 132275153
Test Plan: Included a new unit test.
Reviewed By: mraway
Differential Revision: D29020227
fbshipit-source-id: 811e5d86c4b59183e84e6424c48c97739be09043
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/60382
Instead of setting weight_decay w uniformly for all ids, for each row i in the sparse embedding table, the actual weight_decay `w_i` becomes `w*freq_i` where `freq_i = halflife/counter_i \in [\log(2), halflife]`. Counter is from `rowwise_counter` with definition `counter_i = 1 + \exp(-iter_{\delta}*\rho)*counter_i`.
Test Plan:
buck test //caffe2/caffe2/python/operator_test:adagrad_test -- test_row_wise_sparse_adagrad
buck test caffe2/caffe2/fb/dper/layer_models/tests/split_1:sparse_nn_test_weight_decay
Reviewed By: 0x10cxR1
Differential Revision: D25581030
fbshipit-source-id: 54b3831b20516c76c559b13d8deb809e2ee3b446
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/60106
In Caffe2, some elementwise in-place compatible ops lack coverage for the in-place case. We add tests for a subset of them here and thereby increase coverage.
Test Plan:
```
buck test //caffe2/caffe2/python/operator_test:elementwise_ops_test
```
Let CI run.
Reviewed By: clrfb
Differential Revision: D29143189
fbshipit-source-id: 83138ad8eff8fe95c40aece53714da3577396a23
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/57080
ONNX optimizer is removed in ONNX 1.9
This PR removes ONNX optimizer from a C++ code path and uses `try-except` block in Python to keep it compatible with both ONNX-1.8 and 1.9.
Test Plan: Imported from OSS
Reviewed By: heitorschueroff
Differential Revision: D28467330
Pulled By: malfet
fbshipit-source-id: 5e4669dd0537648898e593f9e253da18d6dc7568
Co-authored-by: neginraoof <neginmr@utexas.edu>
Co-authored-by: Nikita Shulga <nshulga@fb.com>
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/58062
Make templated function to make sure BatchSparseToDense supports int32 lengths/indices
Test Plan:
```buck test //caffe2/caffe2/python/operator_test:batch_sparse_to_dense_op_test
```
Reviewed By: khabinov
Differential Revision: D28271423
fbshipit-source-id: 41b88b7a3663616b533aaf4731ff35cdf6ec4c85
Summary: Relax test deadlines for c2 tests. We run on loaded machines, and timings are unreliable.
Test Plan: Fixes existing tests
Reviewed By: mruberry
Differential Revision: D28690006
fbshipit-source-id: 457707e81a1ec92548c1f23ea7a0022fa0a3bfda
Summary: Tests are frequently failing with "exceeded the deadline of 1000.00ms", we expect this to happen, so remove the deadline
Test Plan: N/A: Fix breakages
Reviewed By: robieta
Differential Revision: D28581051
fbshipit-source-id: 4825ada9af151fa5d57c45c549138c15ba613705
Summary: When run on very heavily loaded machines, some of these tests are timing out. It's not an issue with the test, it's an issue with the environment. I've removed the timeout so we at least keep unit test coverage.
Test Plan: N/A: Fix breakages
Reviewed By: ngimel
Differential Revision: D28492334
fbshipit-source-id: aed3ee371763161aab2d356f5623c7df053fda6f
Summary:
This is the only line (not in `third_party`) matching the regex `^#!.*python2`, and [it is not the first line of its file](https://github.com/koalaman/shellcheck/wiki/SC1128), so it has no effect. As a followup to https://github.com/pytorch/pytorch/issues/58275, this PR removes that shebang to reduce confusion, so now all Python shebangs in this repo are `python3`.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/58409
Reviewed By: walterddr
Differential Revision: D28478469
Pulled By: samestep
fbshipit-source-id: c17684c8651e45d3fc383cbbc04a31192d10f52f
Summary:
Some machines don't have a versionless `python` on their PATH, which breaks these existing shebangs.
I'm assuming that all the existing versionless `python` shebangs are meant to be `python3` and not `python2`; please let me know if my assumption was incorrect for any of these.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/58275
Test Plan: CI.
Reviewed By: zhouzhuojie
Differential Revision: D28428143
Pulled By: samestep
fbshipit-source-id: 6562be3d12924db72a92a0207b060ef740f61ebf
Summary: Removed the deadline restriction since the first run can take more than the deadline, wile subsequent runs are shorter.
Reviewed By: ngimel
Differential Revision: D28260077
fbshipit-source-id: 8ed2f5c16bc184bf4fae0a59b662fa1da2d4dd0a
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/57296
Seems many trainers disable print(), so we cannot see the thread dumps with CompleteInTimeOrDie(). So log.info() also.
Test Plan: sandcastle
Reviewed By: aalmah
Differential Revision: D28098738
fbshipit-source-id: dfdca8801bacf5c7bccecc2387cb7ef41dadfa46
Summary:
This is an automatic change generated by the following script:
```
#!/usr/bin/env python3
from subprocess import check_output, check_call
import os
def get_compiled_files_list():
import json
with open("build/compile_commands.json") as f:
data = json.load(f)
files = [os.path.relpath(node['file']) for node in data]
for idx, fname in enumerate(files):
if fname.startswith('build/') and fname.endswith('.DEFAULT.cpp'):
files[idx] = fname[len('build/'):-len('.DEFAULT.cpp')]
return files
def run_clang_tidy(fname):
check_call(["python3", "tools/clang_tidy.py", "-c", "build", "-x", fname,"-s"])
changes = check_output(["git", "ls-files", "-m"])
if len(changes) == 0:
return
check_call(["git", "commit","--all", "-m", f"NOLINT stubs for {fname}"])
def main():
git_files = check_output(["git", "ls-files"]).decode("ascii").split("\n")
compiled_files = get_compiled_files_list()
for idx, fname in enumerate(git_files):
if fname not in compiled_files:
continue
if fname.startswith("caffe2/contrib/aten/"):
continue
print(f"[{idx}/{len(git_files)}] Processing {fname}")
run_clang_tidy(fname)
if __name__ == "__main__":
main()
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/56892
Reviewed By: H-Huang
Differential Revision: D27991944
Pulled By: malfet
fbshipit-source-id: 5415e1eb2c1b34319a4f03024bfaa087007d7179
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/56717
The signal_handler was under the caffe2 namespacee but was being used
by PyTorch as well.
I've fixed this my moving it to the c10 namespace where now both C2 and PyTorch
can use it.
The signal_handler interface in caffe2/utils/signal_handler.h is kept the same
for backward compatiblity for C2, but most of the commmon code is moved to c10.
ghstack-source-id: 127446929
Test Plan: waitforbuildbot
Reviewed By: ezyang
Differential Revision: D27946738
fbshipit-source-id: d6228d1a0108f4c807d405e7a0bb799c5375388f
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/56813
When the arg `pass_inputs_as_tensor_list` is True, the input tensors are wrapped into a TensorList and passes in as a single param.
Test Plan: buck test //caffe2/caffe2/python:workspace_test -- TestScriptModule
Reviewed By: dzhulgakov
Differential Revision: D27972928
fbshipit-source-id: 5a199649445b0306f3134086c85bd55da45e1a0b
Summary: `networkx 2.4+` replaced `node` attribute to `nodes` in graph object. This caused failures in `caffe2`'s' `topological_sort_traversal_longest_path` function which uses networkx library for topological sort.
Differential Revision: D27718857
fbshipit-source-id: 812fbb613946565d089cc84a20f3cdf7df046e19
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/55003
Using the `caffe2::setPrintStackTracesOnFatalSignal` utility in
distributed tests to set a signal handler that dumps the state of all threads
for all processes when it receives a FATAL signal. This would help in debugging
tests further.
I had to revert all the python faulthandler code since only one signal handler
function is supported, so running python faulthandler with
`setPrintStackTracesOnFatalSignal` doesn't work.
Sample output:
```
SIGSEGV(11), PID: 3492872, Thread 3492872:
[0] ???(0x7fa7b2d1d61b) in libcaffe2_caffe2_caffe2_cpu.so
[1] ???(0x7fa7b2d1d3fb) in libcaffe2_caffe2_caffe2_cpu.so
[2] ???(0x7fa7b2d1d33d) in libcaffe2_caffe2_caffe2_cpu.so
[3] ???(0x7fa7b2d1d167) in libcaffe2_caffe2_caffe2_cpu.so
[4] ???(0x7fa7ce683150) in libpthread.so.0
[5] ???(0x7fa7be2b233c) in libcaffe2__C_impl_cuda.so
[6] ???(0x7fa7be2ce80c) in libcaffe2__C_impl_cuda.so
[7] ???(0x7fa7be2a0512) in libcaffe2__C_impl_cuda.so
[8] torch::distributed::rpc::TensorPipeAgent::send(torch::distributed::rpc::WorkerInfo const&, torch::distributed::rpc::Message&&, float, std::unordered_map<signed char, signed char, std::hash<signed char>, std::equal_to<signed char>, std::allocator<std::pair<signed char const, signed char> > > const&)+0x24f(0x7fa7be29f71f) in libcaffe2__C_impl_cuda.so
[9] torch::distributed::autograd::sendMessageWithAutograd(torch::distributed::rpc::RpcAgent&, torch::distributed::rpc::WorkerInfo const&, torch::distributed::rpc::Message&&, bool, float, bool)+0x393(0x7fa7b602b203) in libcaffe2_libtorch.so
[10] torch::distributed::rpc::pyRpcPythonUdf(torch::distributed::rpc::WorkerInfo const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >&, std::vector<at::Tensor, std::allocator<at::Tensor> >&, float, bool)+0x201(0x7fa7bd844971) in libcaffe2__C_impl_cuda.so
```
ghstack-source-id: 125630551
Test Plan: waitforbuildbot
Reviewed By: SciPioneer
Differential Revision: D27419714
fbshipit-source-id: 8aca9a14ef688004053d8798124d9c3a3fbe3489
Summary:
*Context:* https://github.com/pytorch/pytorch/issues/53406 added a lint for trailing whitespace at the ends of lines. However, in order to pass FB-internal lints, that PR also had to normalize the trailing newlines in four of the files it touched. This PR adds an OSS lint to normalize trailing newlines.
The changes to the following files (made in 54847d0adb9be71be4979cead3d9d4c02160e4cd) are the only manually-written parts of this PR:
- `.github/workflows/lint.yml`
- `mypy-strict.ini`
- `tools/README.md`
- `tools/test/test_trailing_newlines.py`
- `tools/trailing_newlines.py`
I would have liked to make this just a shell one-liner like the other three similar lints, but nothing I could find quite fit the bill. Specifically, all the answers I tried from the following Stack Overflow questions were far too slow (at least a minute and a half to run on this entire repository):
- [How to detect file ends in newline?](https://stackoverflow.com/q/38746)
- [How do I find files that do not end with a newline/linefeed?](https://stackoverflow.com/q/4631068)
- [How to list all files in the Git index without newline at end of file](https://stackoverflow.com/q/27624800)
- [Linux - check if there is an empty line at the end of a file [duplicate]](https://stackoverflow.com/q/34943632)
- [git ensure newline at end of each file](https://stackoverflow.com/q/57770972)
To avoid giving false positives during the few days after this PR is merged, we should probably only merge it after https://github.com/pytorch/pytorch/issues/54967.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/54737
Test Plan:
Running the shell script from the "Ensure correct trailing newlines" step in the `quick-checks` job of `.github/workflows/lint.yml` should print no output and exit in a fraction of a second with a status of 0. That was not the case prior to this PR, as shown by this failing GHA workflow run on an earlier draft of this PR:
- https://github.com/pytorch/pytorch/runs/2197446987?check_suite_focus=true
In contrast, this run (after correcting the trailing newlines in this PR) succeeded:
- https://github.com/pytorch/pytorch/pull/54737/checks?check_run_id=2197553241
To unit-test `tools/trailing_newlines.py` itself (this is run as part of our "Test tools" GitHub Actions workflow):
```
python tools/test/test_trailing_newlines.py
```
Reviewed By: malfet
Differential Revision: D27409736
Pulled By: samestep
fbshipit-source-id: 46f565227046b39f68349bbd5633105b2d2e9b19
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/54042
Pull Request resolved: https://github.com/pytorch/pytorch/pull/53881
1. Fix position_weighted optimizer: Position weighted layer uses default optimizer but is actually gradient_slice, which will cause problem if we do not handle it properly in the new optimizier. The solution is to use sparseadagrad when it is gradient_slices.
2. Optimizer implementation of v1 and v2: using 1st momentum with/without bias_correction.
3. also implemented decoupled weight decay in the new optimizer.
Test Plan:
buck test //caffe2/caffe2/fb/dper/layer_models/tests/split_1:sparse_nn_test_2 -- test_mlp_optimization
buck test //caffe2/caffe2/python:optimizer_test -- TestDecayAdagrad
buck test //caffe2/caffe2/python/operator_test:decay_adagrad_test
ctr_mbl_feed work flow: f255731660
oc work flow: f255739503
Reviewed By: 0x10cxR1
Differential Revision: D26839668
fbshipit-source-id: 2b6881c1a88540ef5766be40f5e80001257e2199
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/53735
Add an option to BlobSerializationOptions to request that float data be
serialized as bfloat16. This reduces the serialized data size at the expense
of some loss in precision.
ghstack-source-id: 124317910
Test Plan: Included a new unit test.
Reviewed By: mraway
Differential Revision: D26658205
fbshipit-source-id: 74521ed161059066355a3f208488ed01a344dbb5
Summary: Add ability to reset optimizer counter..
Test Plan: will wait for integration tests to run on diff.
Differential Revision: D27248286
fbshipit-source-id: a608df1bd61b64eb317c9ffd9cfdd804c5288f6d
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/54274
Some of the Python tests need to be aware of whether or not FBGEMM is
available, so expose this setting in the pybind extension.
ghstack-source-id: 124317732
Test Plan: Will use this variable in the tests on D26658205.
Reviewed By: mraway
Differential Revision: D27171780
fbshipit-source-id: 4c94144a959bf8bf0e1553b6e029e94a91794e29
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/53402
Add an `options` field to the `Save` operator which accepts options for how to
serialize different blobs. At the moment this simply allows controlling the
existing `chunk_size` behavior, but in the future we can add other options,
such as the ability to control compression settings or other serialization
formats.
ghstack-source-id: 123567034
Test Plan:
Added a new test to `load_save_test.py` that passes in options and verifies
that blobs were serialized with the expected number of chunks.
buck test caffe2/caffe2:caffe2_test_cpu \
caffe2/caffe2/core:serialization_test \
caffe2/caffe2/python/operator_test:load_save_test
Reviewed By: mraway
Differential Revision: D26502577
fbshipit-source-id: 6e302e530bb96990517c2e35c505db7f14a56284
