Added functionality for data parallel table #843

csarofeen · 2017-02-25T00:00:50Z

Added:
Scatter of input across dims other than 0
Arbitrary ordered input

Not supported: keyword arguments still not supported

For input:
Variables are scattered
Tensors are broadcasted
Primitive variables are broadcasted
All others are shallow copied
Supports nested lists/tuples and maintains structure

Gather
Nested lists, nested tuples, None and Variables are supported.

…no kwargs). Variables will be scattered/gathered. Nested tuple/list structure will be preserved. Anything else will be shallow copied to each thread.

…le doc fix.

torch/cuda/comm.py

        assert min(chunk_sizes) > 0, "got a negative chunk_size"
        chunks = [tensor.narrow(dim, start - size, size)
                  for start, size in zip(_accumulate(chunk_sizes), chunk_sizes)]
+    chunks = tuple(chunk.contiguous() for chunk in chunks)


torch/nn/parallel/data_parallel.py

            self.module.cuda(device_ids[0])

    def forward(self, *inputs):
+        #leave non-Variable/non-tuples in place


torch/nn/parallel/scatter_gather.py

+        if isinstance(obj, tuple) or isinstance(obj, list):
+            return type(obj)(zip(*map(scatter_map, obj)))
+        if torch.is_tensor(obj):
+            return broadcast(obj, target_gpus)


torch/nn/parallel/scatter_gather.py

-            return Gather(target_device)(*outputs)
-        return type(out)(map(gather_map, zip(*outputs)))
+            return Gather(target_device, dim=dim)(*outputs)
+        if isinstance(out, tuple) or isinstance(out, list):


torch/nn/parallel/data_parallel.py

-            return tuple((map(_to_cuda, obj)))
+            if isinstance(obj, tuple):
+                return tuple((map(_to_cuda, obj)))
+            return obj


…thub_personal

csarofeen · 2017-02-28T00:48:22Z

@apaszke @colesbury included the changes we discussed on slack.

torch/nn/parallel/data_parallel.py

+            if kwargs:
+                gpu_dict = {}
+                for key in kwargs.keys():
+                    gpu_dict[key] = _to_cuda(kwargs[key])


Moved block that _to_cuda's the kwargs in the with cuda.device block

torch/nn/parallel/parallel_apply.py



-def parallel_apply(modules, inputs):
+def parallel_apply(modules, inputs, kwargs):


torch/nn/parallel/parallel_apply.py

+    else:
+        threads = [threading.Thread(target=_worker,
+                                    args=(module, input, kwargs, results, lock))
+                   for module, input, kwargs in zip(modules, inputs, kwargs)]


…thub_personal

csarofeen · 2017-03-01T23:03:17Z

@apaszke @szagoruyko I think I hit all your comments, is there anything else you think needs to be done for this PR?

apaszke

Can you please not change the formatting of the code that doesn't need to be touched by this PR? It makes the diffs smaller, and there's no point in doing that, because we're not merging any PRs that don't pass the linter checks.

torch/nn/parallel/scatter_gather.py

-            return Scatter(target_gpus)(obj)
-        return tuple(zip(*map(scatter_map, obj)))
+            return Scatter(target_gpus, dim=dim)(obj)
+        assert not torch.is_tensor(obj), "Tensors not supported in DataParallel"


torch/nn/parallel/scatter_gather.py

-       them accross given GPUs
+def scatter(input, target_gpus, dim=0):
+    """
+    Slices all variables and tensors into approximately


test/test_nn.py

            self.assertTrue(h_module is module)
            self.assertEqual(input[0].data, torch.ones(5, 5))
-            self.assertEqual(output.data, torch.Tensor(5, 5).fill_(1 / (1 + 1 / math.e)))
+            self.assertEqual(


torch/nn/parallel/data_parallel.py

+            for key in kwargs.keys():
+                scatter_kwargs[key] = self.scatter(
+                    _to_cuda(kwargs[key]), self.device_ids)
+            gpu_dicts = tuple([


torch/nn/parallel/data_parallel.py

+                scatter_kwargs[key] = self.scatter(
+                    _to_cuda(kwargs[key]), self.device_ids)
+            gpu_dicts = tuple([
+                dict([(key, values[i])


torch/nn/parallel/data_parallel.py



-def data_parallel(module, inputs, device_ids, output_device=None):
+def data_parallel(module, inputs, device_ids, module_kwargs=None, output_device=None, dim=0):


torch/nn/parallel/data_parallel.py

+        gpu_dicts = tuple([
+            dict([(key, values[i]) for key, values in scatter_kwargs.items()])
+            for i in device_ids
+        ])


torch/nn/parallel/parallel_apply.py

    # Fast track
    if len(modules) == 1:
-        return (modules[0](*inputs[0]),)
+        return (wrap(modules[0], *inputs[0], **kwargs_tup[0]), )


torch/nn/parallel/parallel_apply.py

+    if kwargs_tup:
+        assert len(modules) == len(kwargs_tup)
+    else:
+        kwargs_tup = tuple({} for gpu in modules)


torch/nn/parallel/parallel_apply.py

    results = {}

-    def _worker(module, input, results, lock):
+    def _worker(module, input, results, lock, **kwargs):


apaszke · 2017-03-04T01:26:21Z

I think that once the comments are fixed and the conflict is resolved it's ready to merge.

Sorry for the delay, we wanted to discuss it a bit first.

…ub_personal Reverted test_nn.py formatting from autopep # Conflicts: # test/test_nn.py

apaszke · 2017-03-05T01:35:51Z

Thanks!

#843) onnx/onnx@73c34ae

pytorch#843) onnx/onnx@73c34ae

…27efbc Previous import was 403ccfbd0161c38f0834413d790bad0874afbf9a Included changes: - **[69894f2](onnx/onnx@69894f2)**: Use op schema.all tensor types in random like definitions (pytorch#865) <Scott McKay> - **[b9d6b90](onnx/onnx@b9d6b90)**: Clarify random like operators (pytorch#846) <Scott McKay> - **[fc6b5fb](onnx/onnx@fc6b5fb)**: Refactor shape inference implementation (pytorch#855) <anderspapitto> - **[b7d8dc8](onnx/onnx@b7d8dc8)**: fix cmake warning message (pytorch#863) <Eric S. Yu> - **[f585c5d](onnx/onnx@f585c5d)**: add pytorch-operator test for tile (pytorch#831) <Wenhao Hu> - **[993fe70](onnx/onnx@993fe70)**: add install step (pytorch#832) <Eric S. Yu> - **[68bc26c](onnx/onnx@68bc26c)**: add type inference for traditional ml ops except classifier ops. (pytorch#857) <Ke Zhang> - **[9cc0cda](onnx/onnx@9cc0cda)**: fix string representation of scalar types (pytorch#858) <G. Ramalingam> - **[1078925](onnx/onnx@1078925)**: fix y in pow test case to scalar (pytorch#852) <Wenhao Hu> - **[c66fb6f](onnx/onnx@c66fb6f)**: Add some math function shape inference (pytorch#845) <anderspapitto> - **[ff667d1](onnx/onnx@ff667d1)**: Refactor return type and docs for ONNXIFI_BACKEND_DIRECTX_ID (pytorch#853) <Marat Dukhan> - **[11c6876](onnx/onnx@11c6876)**: clear initializer names when clear initializer (pytorch#849) <Wenhao Hu> - **[73c34ae](onnx/onnx@73c34ae)**: Clarify FeatureVectorizer description. (pytorch#843) <Scott McKay> - **[1befb9b](onnx/onnx@1befb9b)**: Remove useless text in docs (pytorch#850) <Lu Fang> - **[e84788f](onnx/onnx@e84788f)**: Fix SELU attributes' default values (pytorch#839) <Lu Fang> - **[ebac046](onnx/onnx@ebac046)**: Add tile test case (pytorch#823) <Wenhao Hu> - **[8b7a925](onnx/onnx@8b7a925)**: a few more shape inference functions (pytorch#772) <anderspapitto> - **[9718f42](onnx/onnx@9718f42)**: Make the coefficient non optional for LinearClassifier (pytorch#836) <Jaliya Ekanayake> - **[ef083d0](onnx/onnx@ef083d0)**: Add save_tensor and load_tensor functions for Protos (pytorch#770) <Lu Fang> - **[45ceb55](onnx/onnx@45ceb55)**: Check if CMAKE_BUILD_TYPE set before project(). (pytorch#812) <Sergii Dymchenko> - **[4b3d2b0](onnx/onnx@4b3d2b0)**: [WIP] reenable shape inference tests (pytorch#834) <anderspapitto> - **[22d17ee](onnx/onnx@22d17ee)**: RNN tests: LSTM, GRU, SimpleRNN (pytorch#739) <Peyman Manikashani> - **[de65b95](onnx/onnx@de65b95)**: dimension denotation (pytorch#443) <Tian Jin> - **[eccc76e](onnx/onnx@eccc76e)**: fix field number issue in onnx operator proto and enable its build (pytorch#829) <Ke Zhang> - **[d582beb](onnx/onnx@d582beb)**: disable shape inference test to unbreak ci (pytorch#830) <Lu Fang> - **[485b787](onnx/onnx@485b787)**: function proto for composite op. (pytorch#802) <Ke Zhang> - **[cd58928](onnx/onnx@cd58928)**: specify defaults for attributes of Affine op (pytorch#820) <G. Ramalingam> - **[7ee2cf9](onnx/onnx@7ee2cf9)**: merge the dummy backend back into the main one (pytorch#743) <anderspapitto> - **[1c03a5a](onnx/onnx@1c03a5a)**: [Proposal] ONNX Interface for Framework Integration (previously ONNX Backend API) header and docs (pytorch#551) <Marat Dukhan> - **[3769a98](onnx/onnx@3769a98)**: Rename real model test case from VGG-16 to ZFNet (pytorch#821) <Lu Fang>

* [bootcamp] Improve "Shape" operator to support axes specification To improve .shape operator of Caffe2 to support x.shape(tensor, axes), which takes an optional int array "axes" as input. For example, x.shape(tensor, [1, 0]) will return the dimension for axis 1 and 0 following the specified order. For current version, "axes" input allows duplications and can have arbitrary length. * Back out "Add barrier net that runs before training nets" Original commit changeset: b373fdc9c30f. Need additional changes to some callers to support barrier failures. * Change warning to verbose log to reduce log spam The `LOG(WARNING)` was a bit spammy for regular use so lets just make it a `VLOG`. * Extract the shared code from different caffe2_benchmark binaries The OSS benchmark and Internal benchmark will share most functions in the benchmark. * Support MFR in sequence training As titled. * Make knowledge distillation work with using logged prediction feature as teacher label. 1) Add loading raw dense feature as teacher label. 2) Optional calibration function for teacher label 3) Add teacher label into generic unit test 4) Deprecated TTSN workflow version using feature_options to config teacher label * [C2/CUDA]: unjoined cross entropy sigmoid as desc * Add async_scheduling executor into deferrable_net_exec_test Add async_scheduling into tests and fix some exception cases * Fix Event disabled error When disabling event in RNN ops make sure we don't call Finish on disabled event from op's RunAsync * cuda ensure cpu output op can handle both TensorCPU and TensorCUDA as desc. * [C2 Core] Infer input device option in C2 hypothesis_test checkers Improve how we default input blob device options. Previously it defaults as where op lives but it is not necessarily the case. For example: CopyCPUToGPU * [C2 Op]SplitByLengthsOp CPU/GPU implementation [C2 Op]SplitByLengthsOp CPU/GPU implementation * fix undefined symbol error not sure why we're getting undefined symbol even with link_whole = True Need to figure out why but need this workaround for now * Add tools in DAIPlayground platform to help debugging models Add additional tools to allow Plauground override individual method defined in AnyExp. This will allow user to create module that specificly change certain default method behavior. An example included in this diff is deactivating test model and checkpointing. When debugging any model problems, switching off components helps me quickly narrow down the location of the bug. The technique is extensively used in task T27038712 (Steady memory increase in EDPM, eventually resulting in gloo/cuda.cu:34: out of memory) * add shape and type inference for int8 conversion operator * Fix flaky test for group_norm Fix flaky test for group_norm * Fix group_norm_op_test flaky Fix group_norm_op_test flaky * Implementation of composite learning rate policy In many state-of-the-arts deep learning works, people use a simple trick to schedule the learning rate: use a fixed learning rate until error plateaus and then switch to a different fixed learning rate, and so on. In this diff, we implemented a simple version of the composite learning rate. The user gives a set of learning rates policies and corresponding iteration nums, and the optimizer will change the learning rate policy based on the number of iterations so far. For example, the user give two learning rate policies, one is FixedLearningRate and PolyLearningRate, with an iteration number of 1k. Then the first 1k iteration, we use FixedLearningRate. For the following iterations, we use PolyLearningRate. * Split two use cases of CachedReader into two classes, DBFileReader and CachedReader # Use Cases: 1). input: DB file -> output: DatasetReader. Use DBFileReader. 2). input: Reader -> build cache DB file -> output: DatasetReader. Use CachedReader. # Changes to CachedReader: 1). Move db_path to the constructor. Because in mock reader. cache will always be built ahead. # Changes to tests: 1). Make a separate TestCase class for CachedReader and DBFileReader. 2). Make it possible to add more test functions by adding setUp, tearDown and _make_temp_path. 3). Make delete db_path more general. `db_path` could be a file for `log_file_db`, but could also be a directory for `leveldb`. * Back out "On Mobile phones, call GlobalInit with no arguments in predictor in case we need to perform initialization" Original commit changeset: 4489c6133f11 * Fix LARS bug Fixed a bug in the LARS implementation which caused all subsequent blobs not using LARS to have the LARS learning rate multiplier applied to them. * [tum] support sparse init & add uniformFill option as title * Propagate exception for async nets Capture the exception when an exception is thrown in async nets and re-throw it after wait(). This allows exceptions to be propagated up to the caller. This diff was a part of D7752068. We split the diff so that C2 core files changes are in a separate diff. * Automatic update of fbcode/onnx to 69894f207dfcd72d1e70497d387201cec327efbc Previous import was 403ccfbd0161c38f0834413d790bad0874afbf9a Included changes: - **[69894f2](onnx/onnx@69894f2)**: Use op schema.all tensor types in random like definitions (#865) <Scott McKay> - **[b9d6b90](onnx/onnx@b9d6b90)**: Clarify random like operators (#846) <Scott McKay> - **[fc6b5fb](onnx/onnx@fc6b5fb)**: Refactor shape inference implementation (#855) <anderspapitto> - **[b7d8dc8](onnx/onnx@b7d8dc8)**: fix cmake warning message (#863) <Eric S. Yu> - **[f585c5d](onnx/onnx@f585c5d)**: add pytorch-operator test for tile (#831) <Wenhao Hu> - **[993fe70](onnx/onnx@993fe70)**: add install step (#832) <Eric S. Yu> - **[68bc26c](onnx/onnx@68bc26c)**: add type inference for traditional ml ops except classifier ops. (#857) <Ke Zhang> - **[9cc0cda](onnx/onnx@9cc0cda)**: fix string representation of scalar types (#858) <G. Ramalingam> - **[1078925](onnx/onnx@1078925)**: fix y in pow test case to scalar (#852) <Wenhao Hu> - **[c66fb6f](onnx/onnx@c66fb6f)**: Add some math function shape inference (#845) <anderspapitto> - **[ff667d1](onnx/onnx@ff667d1)**: Refactor return type and docs for ONNXIFI_BACKEND_DIRECTX_ID (#853) <Marat Dukhan> - **[11c6876](onnx/onnx@11c6876)**: clear initializer names when clear initializer (#849) <Wenhao Hu> - **[73c34ae](onnx/onnx@73c34ae)**: Clarify FeatureVectorizer description. (#843) <Scott McKay> - **[1befb9b](onnx/onnx@1befb9b)**: Remove useless text in docs (#850) <Lu Fang> - **[e84788f](onnx/onnx@e84788f)**: Fix SELU attributes' default values (#839) <Lu Fang> - **[ebac046](onnx/onnx@ebac046)**: Add tile test case (#823) <Wenhao Hu> - **[8b7a925](onnx/onnx@8b7a925)**: a few more shape inference functions (#772) <anderspapitto> - **[9718f42](onnx/onnx@9718f42)**: Make the coefficient non optional for LinearClassifier (#836) <Jaliya Ekanayake> - **[ef083d0](onnx/onnx@ef083d0)**: Add save_tensor and load_tensor functions for Protos (#770) <Lu Fang> - **[45ceb55](onnx/onnx@45ceb55)**: Check if CMAKE_BUILD_TYPE set before project(). (#812) <Sergii Dymchenko> - **[4b3d2b0](onnx/onnx@4b3d2b0)**: [WIP] reenable shape inference tests (#834) <anderspapitto> - **[22d17ee](onnx/onnx@22d17ee)**: RNN tests: LSTM, GRU, SimpleRNN (#739) <Peyman Manikashani> - **[de65b95](onnx/onnx@de65b95)**: dimension denotation (#443) <Tian Jin> - **[eccc76e](onnx/onnx@eccc76e)**: fix field number issue in onnx operator proto and enable its build (#829) <Ke Zhang> - **[d582beb](onnx/onnx@d582beb)**: disable shape inference test to unbreak ci (#830) <Lu Fang> - **[485b787](onnx/onnx@485b787)**: function proto for composite op. (#802) <Ke Zhang> - **[cd58928](onnx/onnx@cd58928)**: specify defaults for attributes of Affine op (#820) <G. Ramalingam> - **[7ee2cf9](onnx/onnx@7ee2cf9)**: merge the dummy backend back into the main one (#743) <anderspapitto> - **[1c03a5a](onnx/onnx@1c03a5a)**: [Proposal] ONNX Interface for Framework Integration (previously ONNX Backend API) header and docs (#551) <Marat Dukhan> - **[3769a98](onnx/onnx@3769a98)**: Rename real model test case from VGG-16 to ZFNet (#821) <Lu Fang> * [C2]ReluN Op relu n op. tf reference: https://www.tensorflow.org/api_docs/python/tf/nn/relu6 * Call destructor when assigning a blob value * Add executor overrides Add executor overrides flag to enable migration to async_scheduling executor * Add barrier net that runs before training nets - attempt #2 Add a synchonize barrier net that is run before training nets. With this net, shards that are faster will wait for other shards before start training. This reduce chances of the faster shards timing out during GLOO AllReduce. Removed explicit data_parallel_model.py.synchronize call in holmes workflow. This change was landed previously but caused errors for some EDPM workflows - See https://fb.facebook.com/groups/1426530000692545/permalink/1906766366002237/ - because EDPM assumes any call to CreateOrCloneCommonWorld and Gloo ops are wrapped in exception handlers but in this case exception thrown in the barrier init net is not handled. To address this issue, we add _CreateOrCloneCommonWorld to the param_init_net instead of a new barrier init net. Since errors for param_init_net run is handled gracefully and re-rendezvous, it should fixes the problem. * Handle empty nets in async_scheduling Make sure we don't get stuck on empty nets * use CUDA_ARCH for conditional compile * [C2 fix] infer function for ensure_cpu_output_op * Update group_norm test to reduce flaky test * Fix lr_multiplier for GPU

pytorch#843) onnx/onnx@73c34ae

* [bootcamp] Improve "Shape" operator to support axes specification To improve .shape operator of Caffe2 to support x.shape(tensor, axes), which takes an optional int array "axes" as input. For example, x.shape(tensor, [1, 0]) will return the dimension for axis 1 and 0 following the specified order. For current version, "axes" input allows duplications and can have arbitrary length. * Back out "Add barrier net that runs before training nets" Original commit changeset: b373fdc9c30f. Need additional changes to some callers to support barrier failures. * Change warning to verbose log to reduce log spam The `LOG(WARNING)` was a bit spammy for regular use so lets just make it a `VLOG`. * Extract the shared code from different caffe2_benchmark binaries The OSS benchmark and Internal benchmark will share most functions in the benchmark. * Support MFR in sequence training As titled. * Make knowledge distillation work with using logged prediction feature as teacher label. 1) Add loading raw dense feature as teacher label. 2) Optional calibration function for teacher label 3) Add teacher label into generic unit test 4) Deprecated TTSN workflow version using feature_options to config teacher label * [C2/CUDA]: unjoined cross entropy sigmoid as desc * Add async_scheduling executor into deferrable_net_exec_test Add async_scheduling into tests and fix some exception cases * Fix Event disabled error When disabling event in RNN ops make sure we don't call Finish on disabled event from op's RunAsync * cuda ensure cpu output op can handle both TensorCPU and TensorCUDA as desc. * [C2 Core] Infer input device option in C2 hypothesis_test checkers Improve how we default input blob device options. Previously it defaults as where op lives but it is not necessarily the case. For example: CopyCPUToGPU * [C2 Op]SplitByLengthsOp CPU/GPU implementation [C2 Op]SplitByLengthsOp CPU/GPU implementation * fix undefined symbol error not sure why we're getting undefined symbol even with link_whole = True Need to figure out why but need this workaround for now * Add tools in DAIPlayground platform to help debugging models Add additional tools to allow Plauground override individual method defined in AnyExp. This will allow user to create module that specificly change certain default method behavior. An example included in this diff is deactivating test model and checkpointing. When debugging any model problems, switching off components helps me quickly narrow down the location of the bug. The technique is extensively used in task T27038712 (Steady memory increase in EDPM, eventually resulting in gloo/cuda.cu:34: out of memory) * add shape and type inference for int8 conversion operator * Fix flaky test for group_norm Fix flaky test for group_norm * Fix group_norm_op_test flaky Fix group_norm_op_test flaky * Implementation of composite learning rate policy In many state-of-the-arts deep learning works, people use a simple trick to schedule the learning rate: use a fixed learning rate until error plateaus and then switch to a different fixed learning rate, and so on. In this diff, we implemented a simple version of the composite learning rate. The user gives a set of learning rates policies and corresponding iteration nums, and the optimizer will change the learning rate policy based on the number of iterations so far. For example, the user give two learning rate policies, one is FixedLearningRate and PolyLearningRate, with an iteration number of 1k. Then the first 1k iteration, we use FixedLearningRate. For the following iterations, we use PolyLearningRate. * Split two use cases of CachedReader into two classes, DBFileReader and CachedReader # Use Cases: 1). input: DB file -> output: DatasetReader. Use DBFileReader. 2). input: Reader -> build cache DB file -> output: DatasetReader. Use CachedReader. # Changes to CachedReader: 1). Move db_path to the constructor. Because in mock reader. cache will always be built ahead. # Changes to tests: 1). Make a separate TestCase class for CachedReader and DBFileReader. 2). Make it possible to add more test functions by adding setUp, tearDown and _make_temp_path. 3). Make delete db_path more general. `db_path` could be a file for `log_file_db`, but could also be a directory for `leveldb`. * Back out "On Mobile phones, call GlobalInit with no arguments in predictor in case we need to perform initialization" Original commit changeset: 4489c6133f11 * Fix LARS bug Fixed a bug in the LARS implementation which caused all subsequent blobs not using LARS to have the LARS learning rate multiplier applied to them. * [tum] support sparse init & add uniformFill option as title * Propagate exception for async nets Capture the exception when an exception is thrown in async nets and re-throw it after wait(). This allows exceptions to be propagated up to the caller. This diff was a part of D7752068. We split the diff so that C2 core files changes are in a separate diff. * Automatic update of fbcode/onnx to 69894f207dfcd72d1e70497d387201cec327efbc Previous import was 403ccfbd0161c38f0834413d790bad0874afbf9a Included changes: - **[69894f2](onnx/onnx@69894f2)**: Use op schema.all tensor types in random like definitions (pytorch#865) <Scott McKay> - **[b9d6b90](onnx/onnx@b9d6b90)**: Clarify random like operators (pytorch#846) <Scott McKay> - **[fc6b5fb](onnx/onnx@fc6b5fb)**: Refactor shape inference implementation (pytorch#855) <anderspapitto> - **[b7d8dc8](onnx/onnx@b7d8dc8)**: fix cmake warning message (pytorch#863) <Eric S. Yu> - **[f585c5d](onnx/onnx@f585c5d)**: add pytorch-operator test for tile (pytorch#831) <Wenhao Hu> - **[993fe70](onnx/onnx@993fe70)**: add install step (pytorch#832) <Eric S. Yu> - **[68bc26c](onnx/onnx@68bc26c)**: add type inference for traditional ml ops except classifier ops. (pytorch#857) <Ke Zhang> - **[9cc0cda](onnx/onnx@9cc0cda)**: fix string representation of scalar types (pytorch#858) <G. Ramalingam> - **[1078925](onnx/onnx@1078925)**: fix y in pow test case to scalar (pytorch#852) <Wenhao Hu> - **[c66fb6f](onnx/onnx@c66fb6f)**: Add some math function shape inference (pytorch#845) <anderspapitto> - **[ff667d1](onnx/onnx@ff667d1)**: Refactor return type and docs for ONNXIFI_BACKEND_DIRECTX_ID (pytorch#853) <Marat Dukhan> - **[11c6876](onnx/onnx@11c6876)**: clear initializer names when clear initializer (pytorch#849) <Wenhao Hu> - **[73c34ae](onnx/onnx@73c34ae)**: Clarify FeatureVectorizer description. (pytorch#843) <Scott McKay> - **[1befb9b](onnx/onnx@1befb9b)**: Remove useless text in docs (pytorch#850) <Lu Fang> - **[e84788f](onnx/onnx@e84788f)**: Fix SELU attributes' default values (pytorch#839) <Lu Fang> - **[ebac046](onnx/onnx@ebac046)**: Add tile test case (pytorch#823) <Wenhao Hu> - **[8b7a925](onnx/onnx@8b7a925)**: a few more shape inference functions (pytorch#772) <anderspapitto> - **[9718f42](onnx/onnx@9718f42)**: Make the coefficient non optional for LinearClassifier (pytorch#836) <Jaliya Ekanayake> - **[ef083d0](onnx/onnx@ef083d0)**: Add save_tensor and load_tensor functions for Protos (pytorch#770) <Lu Fang> - **[45ceb55](onnx/onnx@45ceb55)**: Check if CMAKE_BUILD_TYPE set before project(). (pytorch#812) <Sergii Dymchenko> - **[4b3d2b0](onnx/onnx@4b3d2b0)**: [WIP] reenable shape inference tests (pytorch#834) <anderspapitto> - **[22d17ee](onnx/onnx@22d17ee)**: RNN tests: LSTM, GRU, SimpleRNN (pytorch#739) <Peyman Manikashani> - **[de65b95](onnx/onnx@de65b95)**: dimension denotation (pytorch#443) <Tian Jin> - **[eccc76e](onnx/onnx@eccc76e)**: fix field number issue in onnx operator proto and enable its build (pytorch#829) <Ke Zhang> - **[d582beb](onnx/onnx@d582beb)**: disable shape inference test to unbreak ci (pytorch#830) <Lu Fang> - **[485b787](onnx/onnx@485b787)**: function proto for composite op. (pytorch#802) <Ke Zhang> - **[cd58928](onnx/onnx@cd58928)**: specify defaults for attributes of Affine op (pytorch#820) <G. Ramalingam> - **[7ee2cf9](onnx/onnx@7ee2cf9)**: merge the dummy backend back into the main one (pytorch#743) <anderspapitto> - **[1c03a5a](onnx/onnx@1c03a5a)**: [Proposal] ONNX Interface for Framework Integration (previously ONNX Backend API) header and docs (pytorch#551) <Marat Dukhan> - **[3769a98](onnx/onnx@3769a98)**: Rename real model test case from VGG-16 to ZFNet (pytorch#821) <Lu Fang> * [C2]ReluN Op relu n op. tf reference: https://www.tensorflow.org/api_docs/python/tf/nn/relu6 * Call destructor when assigning a blob value * Add executor overrides Add executor overrides flag to enable migration to async_scheduling executor * Add barrier net that runs before training nets - attempt pytorch#2 Add a synchonize barrier net that is run before training nets. With this net, shards that are faster will wait for other shards before start training. This reduce chances of the faster shards timing out during GLOO AllReduce. Removed explicit data_parallel_model.py.synchronize call in holmes workflow. This change was landed previously but caused errors for some EDPM workflows - See https://fb.facebook.com/groups/1426530000692545/permalink/1906766366002237/ - because EDPM assumes any call to CreateOrCloneCommonWorld and Gloo ops are wrapped in exception handlers but in this case exception thrown in the barrier init net is not handled. To address this issue, we add _CreateOrCloneCommonWorld to the param_init_net instead of a new barrier init net. Since errors for param_init_net run is handled gracefully and re-rendezvous, it should fixes the problem. * Handle empty nets in async_scheduling Make sure we don't get stuck on empty nets * use CUDA_ARCH for conditional compile * [C2 fix] infer function for ensure_cpu_output_op * Update group_norm test to reduce flaky test * Fix lr_multiplier for GPU

Correct link

…instead of the Default Cuda Stream. (pytorch#843) * Adding C++ Multihead Attention implementation to contrib. * Add reference test that at least works for forward. * Remove CublasLt support from multihead attention. * Add new Python version of self attention. * Update python model of MHA with backward pass. * Fixed Output Linear connection in MHA. * Clean up compiles and add documentation to PySelfAttention. * Add Encdec Python version of multihead attention. Cleanup files. * Tests for self and encdec multihead attention. * Add reference pytorch implementation of attention with norm and add. * Add cutlass branch definition. * Add cutlass download to compile. * Add norm/add tests. * Add biases to pytorch python versions. * Add tests and fix issues with python version of attention masking. * Create README.md * Update README.md * Update README.md * Update perf test parameters. * Update README.md * Update README.md * Update README.md * Add files via upload * Update README.md * Update README.md * Update README.md * Fix matmul1 output tensor size. Fix tests that missed issue. * Allow for Z dimensions of 64K and greater on batched GEMMs. * remove redundant imports * general cleanup, remove deprecated or unused functions * Update Multihead Attention's softmax to use the Current Stream instead of the default stream. * Fix setup.py that got messed up in merge with upstream. * Update Multihead Attention strided batched gemms to use the current stream instead of the default. Co-authored-by: pbialecki <pbialecki@nvidia.com>

csarofeen added 3 commits February 24, 2017 13:09

Update to data_parallel_table. Support for arbitrary ordered inputs (…

35c2d54

…no kwargs). Variables will be scattered/gathered. Nested tuple/list structure will be preserved. Anything else will be shallow copied to each thread.

Simple scatter_gather cleanup. Test backwards for data parallel. Simp…

ed3ec1e

…le doc fix.

Add none to gather_map in scatter_gather

5f5704f

apaszke reviewed Feb 25, 2017

View reviewed changes

csarofeen added 9 commits February 24, 2017 16:12

Fix whitespace test_nn.py

e241b35

Data_parallel now accepts keyword arguments, rejects tensors.

5e4ae73

Merge branch 'master' of https://github.com/csarofeen/pytorch into gi…

83ab787

…thub_personal

Merge branch 'master' into master

0316696

Quick change to scatter_gather>gather

2c7c753

Merge branch 'master' of https://github.com/csarofeen/pytorch into gi…

b055145

…thub_personal

autopep8

3823ac0

Cleanup debug print statement

dd5d2a5

format

44fe916

apaszke reviewed Feb 28, 2017

View reviewed changes

torch/nn/parallel/data_parallel.py Outdated

if kwargs:

gpu_dict = {}

for key in kwargs.keys():

gpu_dict[key] = _to_cuda(kwargs[key])

This comment was marked as off-topic.

Sign in to view

Single gpu case fix in DataParallel

90fa6a8

Moved block that _to_cuda's the kwargs in the with cuda.device block

szagoruyko reviewed Feb 28, 2017

View reviewed changes

torch/nn/parallel/parallel_apply.py Outdated

def parallel_apply(modules, inputs):

def parallel_apply(modules, inputs, kwargs):

This comment was marked as off-topic.

Sign in to view

szagoruyko reviewed Feb 28, 2017

View reviewed changes

csarofeen added 3 commits February 28, 2017 13:56

Cleaned up parallel_apply logic

5923c03

Merge branch 'master' of https://github.com/csarofeen/pytorch into gi…

d45858e

…thub_personal

Merge branch 'master' into master

52f7b01

apaszke reviewed Mar 4, 2017

View reviewed changes

csarofeen added 3 commits March 4, 2017 16:54

Merge branch 'master' of https://github.com/pytorch/pytorch into gith…

5632263

…ub_personal Reverted test_nn.py formatting from autopep # Conflicts: # test/test_nn.py

Changes requested by @apaszke

35b4f38

Whitespace fix.

f0189e4

apaszke merged commit b1ae7f9 into pytorch:master Mar 5, 2017

onnxbot added a commit that referenced this pull request Apr 28, 2018

[auto] Update onnx to 73c34ae - Clarify FeatureVectorizer description. (

7450e91

#843) onnx/onnx@73c34ae

Jorghi12 pushed a commit to wsttiger/pytorch that referenced this pull request May 10, 2018

[auto] Update onnx to 73c34ae - Clarify FeatureVectorizer description. (

e923714

pytorch#843) onnx/onnx@73c34ae

weiyangfb pushed a commit to weiyangfb/pytorch that referenced this pull request Jun 11, 2018

[auto] Update onnx to 73c34ae - Clarify FeatureVectorizer description. (

a5b8e56

pytorch#843) onnx/onnx@73c34ae

ezyang added the open source label Jun 24, 2019

mrshenli pushed a commit to mrshenli/pytorch that referenced this pull request Apr 11, 2020

Merge pull request pytorch#843 from pytorch/jlin27-quant-url-patch

e8055d7

Correct link

jjsjann123 pushed a commit to jjsjann123/pytorch that referenced this pull request Aug 5, 2021

Translation between welford and persistent normalization (pytorch#843)

c3873c0

KyleCZH pushed a commit to KyleCZH/pytorch that referenced this pull request Sep 20, 2021

Build macOS conda binary with llvm (pytorch#843)

e77a2a5



		def parallel_apply(modules, inputs):
		def parallel_apply(modules, inputs, kwargs):



		def data_parallel(module, inputs, device_ids, output_device=None):
		def data_parallel(module, inputs, device_ids, module_kwargs=None, output_device=None, dim=0):

Added functionality for data parallel table #843

Added functionality for data parallel table #843

Uh oh!

Conversation

csarofeen commented Feb 25, 2017

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

csarofeen commented Feb 28, 2017

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

csarofeen commented Mar 1, 2017

Uh oh!

apaszke left a comment

Choose a reason for hiding this comment

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

Uh oh!

This comment was marked as off-topic.

apaszke commented Mar 4, 2017 •

edited

Loading