Skip to content

Commit dc570e4

Browse files
peri044narendasan
peri044
and
narendasan
authored
Upgrade stack to Pytorch 2.0 + CUDA 11.7 + TRT 8.5 GA (#1477)
Signed-off-by: Dheeraj Peri <[email protected]> Co-authored-by: Naren Dasan <[email protected]>
1 parent 5fa6374 commit dc570e4

File tree

126 files changed

+631
-512
lines changed

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

126 files changed

+631
-512
lines changed

‎.circleci/config.yml

Lines changed: 185 additions & 175 deletions
Large diffs are not rendered by default.

‎README.md

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -114,10 +114,10 @@ torch.jit.save(trt_ts_module, "trt_torchscript_module.ts") # save the TRT embedd
114114
These are the following dependencies used to verify the testcases. Torch-TensorRT can work with other versions, but the tests are not guaranteed to pass.
115115

116116
- Bazel 5.2.0
117-
- Libtorch 1.13.0 (built with CUDA 11.7)
117+
- Libtorch 2.0.0.dev20230103 (built with CUDA 11.7)
118118
- CUDA 11.7
119119
- cuDNN 8.5.0
120-
- TensorRT 8.5.0
120+
- TensorRT 8.5.1.7
121121

122122
## Prebuilt Binaries and Wheel files
123123

‎WORKSPACE

Lines changed: 11 additions & 11 deletions
Original file line numberDiff line numberDiff line change
@@ -41,7 +41,7 @@ local_repository(
4141
new_local_repository(
4242
name = "cuda",
4343
build_file = "@//third_party/cuda:BUILD",
44-
path = "/usr/local/cuda-11.6/",
44+
path = "/usr/local/cuda-11.7/",
4545
)
4646

4747
new_local_repository(
@@ -56,17 +56,17 @@ new_local_repository(
5656
http_archive(
5757
name = "libtorch",
5858
build_file = "@//third_party/libtorch:BUILD",
59-
sha256 = "b565c662435fd58ec295fa0791388ea52ad0f5fd33517b2d7c0fdcc91b6db531",
59+
sha256 = "59b8b5e1954a86d50b79c13f06398d385b200da13e37a08ecf31d3c62e5ca127",
6060
strip_prefix = "libtorch",
61-
urls = ["https://download.pytorch.org/libtorch/nightly/cu116/libtorch-cxx11-abi-shared-with-deps-1.14.0.dev20221114%2Bcu116.zip"],
61+
urls = ["https://download.pytorch.org/libtorch/nightly/cu117/libtorch-cxx11-abi-shared-with-deps-2.0.0.dev20230103%2Bcu117.zip"],
6262
)
6363

6464
http_archive(
6565
name = "libtorch_pre_cxx11_abi",
6666
build_file = "@//third_party/libtorch:BUILD",
67-
sha256 = "fbb37446c33b05c1e26256c09f6ffb46cea1f6ff9ee2ad5b79b146d09023b0c1",
67+
sha256 = "e260fc7476be89d1650953e8643e9f7363845f5a52de4bab87ac0e619c1f6ad4",
6868
strip_prefix = "libtorch",
69-
urls = ["https://download.pytorch.org/libtorch/nightly/cu116/libtorch-shared-with-deps-1.14.0.dev20221114%2Bcu116.zip"],
69+
urls = ["https://download.pytorch.org/libtorch/nightly/cu117/libtorch-shared-with-deps-2.0.0.dev20230103%2Bcu117.zip"],
7070
)
7171

7272
# Download these tarballs manually from the NVIDIA website
@@ -76,20 +76,20 @@ http_archive(
7676
http_archive(
7777
name = "cudnn",
7878
build_file = "@//third_party/cudnn/archive:BUILD",
79-
sha256 = "ec96d2376d81fca42bdd3d4c3d705a99b29a065bab57f920561c763e29c67d01",
80-
strip_prefix = "cudnn-linux-x86_64-8.4.1.50_cuda11.6-archive",
79+
sha256 = "5454a6fd94f008728caae9adad993c4e85ef36302e26bce43bea7d458a5e7b6d",
80+
strip_prefix = "cudnn-linux-x86_64-8.5.0.96_cuda11-archive",
8181
urls = [
82-
"https://developer.nvidia.com/compute/cudnn/secure/8.4.1/local_installers/11.6/cudnn-linux-x86_64-8.4.1.50_cuda11.6-archive.tar.xz",
82+
"https://developer.nvidia.com/compute/cudnn/secure/8.5.0/local_installers/11.7/cudnn-linux-x86_64-8.5.0.96_cuda11-archive.tar.xz",
8383
],
8484
)
8585

8686
http_archive(
8787
name = "tensorrt",
8888
build_file = "@//third_party/tensorrt/archive:BUILD",
89-
sha256 = "8d7c2085c1639dcc73875048c23598a8526ce3089136876e31d90258e49e4f61",
90-
strip_prefix = "TensorRT-8.4.3.1",
89+
sha256 = "39cc7f077057d1363794e8ff51c4cf21a5dbeccf1116b0020ba0dae0f3063076",
90+
strip_prefix = "TensorRT-8.5.1.7",
9191
urls = [
92-
"https://developer.nvidia.com/compute/machine-learning/tensorrt/secure/8.4.3/tars/tensorrt-8.4.3.1.linux.x86_64-gnu.cuda-11.6.cudnn8.4.tar.gz",
92+
"https://developer.nvidia.com/compute/machine-learning/tensorrt/secure/8.5.1/tars/TensorRT-8.5.1.7.Linux.x86_64-gnu.cuda-11.8.cudnn8.6.tar.gz",
9393
],
9494
)
9595

‎core/conversion/converters/impl/conv_deconv.cpp

Lines changed: 14 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -102,6 +102,20 @@ bool add_conv_deconv(ConversionCtx* ctx, const torch::jit::Node* n, args& args)
102102
}
103103

104104
auto w = Weights(ctx, args[1].unwrapToTensor());
105+
// TODO: Remove this when conv3d with kernel size=1 bug is fixed.
106+
// Github issue: https://github.com/pytorch/TensorRT/issues/1445
107+
bool is_kernel_size_one = true;
108+
bool is_3d_kernel = w.kernel_shape.nbDims == 3;
109+
for (int64_t i = 0; i < w.kernel_shape.nbDims; i++) {
110+
if (w.kernel_shape.d[i] != 1.0f) {
111+
is_kernel_size_one = false;
112+
}
113+
}
114+
if (is_kernel_size_one && is_3d_kernel) {
115+
LOG_WARNING(
116+
"Conv3d layer with kernel size = 1 configuration incurs a failure with TensorRT tactic optimizer in some cases. \
117+
Github issue: https://github.com/pytorch/TensorRT/issues/1445. Other conv variants do not have this issue.");
118+
}
105119
auto dims = in->getDimensions();
106120
auto orig_dims = dims;
107121
LOG_DEBUG("Input dims: " << orig_dims);

‎core/runtime/TRTEngine.cpp

Lines changed: 46 additions & 46 deletions
Original file line numberDiff line numberDiff line change
@@ -68,8 +68,8 @@ TRTEngine::TRTEngine(
6868
uint64_t inputs = 0;
6969
uint64_t outputs = 0;
7070

71-
for (int64_t x = 0; x < cuda_engine->getNbBindings(); x++) {
72-
std::string bind_name = cuda_engine->getBindingName(x);
71+
for (int64_t trt_idx = 0; trt_idx < cuda_engine->getNbIOTensors(); trt_idx++) {
72+
std::string bind_name = cuda_engine->getIOTensorName(trt_idx);
7373
LOG_DEBUG("Binding name: " << bind_name);
7474
auto delim = bind_name.find(".");
7575
if (delim == std::string::npos) {
@@ -80,46 +80,45 @@ TRTEngine::TRTEngine(
8080
<< bind_name
8181
<< "\nEnsure module was compiled with Torch-TensorRT.ts or follows Torch-TensorRT Runtime conventions");
8282
}
83-
8483
std::string idx_s = bind_name.substr(delim + 1);
85-
uint64_t idx = static_cast<uint64_t>(std::stoi(idx_s));
84+
uint64_t pyt_idx = static_cast<uint64_t>(std::stoi(idx_s));
8685

87-
if (cuda_engine->bindingIsInput(x)) {
86+
if (cuda_engine->getTensorIOMode(bind_name.c_str()) == nvinfer1::TensorIOMode::kINPUT) {
8887
inputs++;
89-
in_binding_map[x] = idx;
90-
LOG_DEBUG("TRT Binding: " << x << ": PYT Input: " << idx);
88+
in_binding_map[trt_idx] = pyt_idx;
89+
LOG_DEBUG("TRT Binding index: " << trt_idx << "corresponds to PYT Input index: " << pyt_idx);
9190
} else {
9291
outputs++;
93-
out_binding_map[x] = idx;
94-
LOG_DEBUG("TRT Binding: " << x << ": PYT Output: " << idx);
92+
out_binding_map[trt_idx] = pyt_idx;
93+
LOG_DEBUG("TRT Binding index: " << trt_idx << "corresponds to PYT Output: " << pyt_idx);
9594
}
9695
}
9796

9897
num_io = std::make_pair(inputs, outputs);
9998
in_binding_names.resize(inputs);
10099
out_binding_names.resize(outputs);
101-
102-
for (int64_t x = 0; x < cuda_engine->getNbBindings(); x++) {
103-
std::string bind_name = cuda_engine->getBindingName(x);
104-
if (cuda_engine->bindingIsInput(x)) {
100+
for (int64_t x = 0; x < cuda_engine->getNbIOTensors(); x++) {
101+
std::string bind_name = cuda_engine->getIOTensorName(x);
102+
if (cuda_engine->getTensorIOMode(bind_name.c_str()) == nvinfer1::TensorIOMode::kINPUT) {
105103
in_binding_names[in_binding_map.at(x)] = bind_name;
106104
} else {
107105
out_binding_names[out_binding_map.at(x)] = bind_name;
108106
}
109107
}
110108
} else {
111-
uint64_t inputs = _in_binding_names.size();
112-
in_binding_names.resize(inputs);
113-
for (size_t pyt_idx = 0; pyt_idx < inputs; pyt_idx++) {
109+
uint64_t inputs_size = _in_binding_names.size();
110+
in_binding_names.resize(inputs_size);
111+
for (size_t pyt_idx = 0; pyt_idx < inputs_size; pyt_idx++) {
114112
auto binding_name = _in_binding_names[pyt_idx];
115113
auto trt_idx = cuda_engine->getBindingIndex(binding_name.c_str());
116-
TORCHTRT_CHECK((trt_idx >= 0), "Could not find a TensorRT engine binding for input named " << binding_name);
114+
std::string engine_binded_name = cuda_engine->getIOTensorName(pyt_idx);
117115
TORCHTRT_CHECK(
118-
cuda_engine->bindingIsInput(trt_idx),
116+
(binding_name == engine_binded_name),
117+
"Could not find a TensorRT engine binding for input named " << binding_name);
118+
TORCHTRT_CHECK(
119+
(cuda_engine->getTensorIOMode(binding_name.c_str()) == nvinfer1::TensorIOMode::kINPUT),
119120
"Binding " << binding_name << " specified as input but found as output in TensorRT engine");
120-
LOG_DEBUG(
121-
"Input binding name: " << binding_name << " (trt binding idx: " << trt_idx << ", "
122-
<< "pyt arg idx: " << pyt_idx << ")");
121+
LOG_DEBUG("Input binding name: " << binding_name << "pyt arg idx: " << pyt_idx << ")");
123122
in_binding_map[trt_idx] = pyt_idx;
124123
in_binding_names[pyt_idx] = _in_binding_names[pyt_idx];
125124
}
@@ -129,17 +128,18 @@ TRTEngine::TRTEngine(
129128
for (size_t pyt_idx = 0; pyt_idx < outputs; pyt_idx++) {
130129
auto binding_name = _out_binding_names[pyt_idx];
131130
auto trt_idx = cuda_engine->getBindingIndex(binding_name.c_str());
132-
TORCHTRT_CHECK((trt_idx >= 0), "Could not find a TensorRT engine binding for output named " << binding_name);
131+
std::string engine_binded_name = cuda_engine->getIOTensorName(inputs_size + pyt_idx);
132+
TORCHTRT_CHECK(
133+
(binding_name == engine_binded_name),
134+
"Could not find a TensorRT engine binding for output named " << binding_name);
133135
TORCHTRT_CHECK(
134-
!cuda_engine->bindingIsInput(trt_idx),
136+
!(cuda_engine->getTensorIOMode(binding_name.c_str()) == nvinfer1::TensorIOMode::kINPUT),
135137
"Binding " << binding_name << " specified as output but found as input in TensorRT engine");
136-
LOG_DEBUG(
137-
"Output binding name: " << binding_name << " (trt binding idx: " << trt_idx << ", "
138-
<< "pyt return idx: " << pyt_idx << ")");
138+
LOG_DEBUG("Output binding name: " << binding_name << "pyt return idx: " << inputs_size + pyt_idx << ")");
139139
out_binding_map[trt_idx] = pyt_idx;
140140
out_binding_names[pyt_idx] = binding_name;
141141
}
142-
num_io = std::make_pair(inputs, outputs);
142+
num_io = std::make_pair(inputs_size, outputs);
143143
}
144144

145145
#ifndef NDEBUG
@@ -149,10 +149,10 @@ TRTEngine::TRTEngine(
149149
}
150150

151151
TRTEngine::~TRTEngine() {
152+
rt.reset();
152153
trt_engine_profiler.reset();
153154
exec_ctx.reset();
154155
cuda_engine.reset();
155-
rt.reset();
156156
}
157157

158158
void TRTEngine::disable_profiling() {
@@ -164,7 +164,7 @@ void TRTEngine::disable_profiling() {
164164
}
165165

166166
void TRTEngine::dump_engine_layer_info_to_file(const std::string& path) {
167-
auto inspector = cuda_engine->createEngineInspector();
167+
auto inspector = make_trt(cuda_engine->createEngineInspector());
168168
std::ofstream f(path);
169169
f << std::string(inspector->getEngineInformation(nvinfer1::LayerInformationFormat::kJSON));
170170
f.close();
@@ -208,23 +208,23 @@ std::string TRTEngine::to_str() const {
208208
std::stringstream ss;
209209
ss << "Torch-TensorRT TensorRT Engine:" << std::endl;
210210
ss << " Name: " << name << std::endl;
211-
ss << " Bindings: {" << std::endl;
212-
for (int64_t x = 0; x < cuda_engine->getNbBindings(); x++) {
213-
if (cuda_engine->bindingIsInput(x)) {
214-
const uint64_t pyt_idx = in_binding_map.at(x);
215-
ss << " (" << x << ": " << in_binding_names.at(pyt_idx) << ") Input: [" << std::endl;
216-
ss << " pytorch arg idx: " << pyt_idx << std::endl;
217-
ss << " shape: " << exec_ctx->getBindingDimensions(x) << std::endl;
218-
ss << " dtype: " << util::TRTDataTypeToScalarType(exec_ctx->getEngine().getBindingDataType(x)) << std::endl;
219-
ss << " ]" << std::endl;
220-
} else {
221-
const uint64_t pyt_idx = out_binding_map.at(x);
222-
ss << " (" << x << ": " << out_binding_names.at(pyt_idx) << ") Output: [" << std::endl;
223-
ss << " pytorch return idx: " << pyt_idx << std::endl;
224-
ss << " shape: " << exec_ctx->getBindingDimensions(x) << std::endl;
225-
ss << " dtype: " << util::TRTDataTypeToScalarType(exec_ctx->getEngine().getBindingDataType(x)) << std::endl;
226-
ss << " ]" << std::endl;
227-
}
211+
ss << " Inputs: [" << std::endl;
212+
for (uint64_t i = 0; i < num_io.first; i++) {
213+
ss << " id: " << i << std::endl;
214+
ss << " shape: " << exec_ctx->getTensorShape(std::string("input_" + str(i)).c_str()) << std::endl;
215+
ss << " dtype: "
216+
<< util::TRTDataTypeToScalarType(exec_ctx->getEngine().getTensorDataType(std::string("input_" + str(i)).c_str()))
217+
<< std::endl;
218+
}
219+
ss << " ]" << std::endl;
220+
ss << " Outputs: [" << std::endl;
221+
for (uint64_t o = 0; o < num_io.second; o++) {
222+
ss << " id: " << o << std::endl;
223+
ss << " shape: " << exec_ctx->getTensorShape(std::string("output_" + str(o)).c_str()) << std::endl;
224+
ss << " dtype: "
225+
<< util::TRTDataTypeToScalarType(
226+
exec_ctx->getEngine().getTensorDataType(std::string("output_" + str(o)).c_str()))
227+
<< std::endl;
228228
}
229229
ss << " }" << std::endl;
230230
ss << " Device: " << device_info << std::endl;

‎core/runtime/execute_engine.cpp

Lines changed: 19 additions & 24 deletions
Original file line numberDiff line numberDiff line change
@@ -121,36 +121,30 @@ std::vector<at::Tensor> execute_engine(std::vector<at::Tensor> inputs, c10::intr
121121
}
122122
}
123123

124-
std::vector<void*> gpu_handles;
125-
std::vector<at::Tensor> contig_inputs{};
126124
{
127125
std::unique_ptr<torch::autograd::profiler::RecordProfile> input_profiler_guard;
128126
if (compiled_engine->profile_execution) {
129127
input_profiler_guard =
130128
std::make_unique<torch::autograd::profiler::RecordProfile>(compiled_engine->input_profile_path);
131129
}
132-
133-
contig_inputs.reserve(inputs.size());
134-
135130
for (size_t i = 0; i < inputs.size(); i++) {
136-
uint64_t pyt_idx = compiled_engine->in_binding_map[i];
131+
std::string name = compiled_engine->in_binding_names[i];
137132
TORCHTRT_CHECK(
138-
inputs[pyt_idx].is_cuda(),
139-
"Expected input tensors to have device cuda, found device " << inputs[pyt_idx].device());
140-
auto expected_type = util::TRTDataTypeToScalarType(compiled_engine->exec_ctx->getEngine().getBindingDataType(i));
133+
inputs[i].is_cuda(), "Expected input tensors to have device cuda, found device " << inputs[i].device());
134+
auto expected_type =
135+
util::TRTDataTypeToScalarType(compiled_engine->exec_ctx->getEngine().getTensorDataType(name.c_str()));
141136
TORCHTRT_CHECK(
142-
inputs[pyt_idx].dtype() == expected_type,
143-
"Expected input tensors to have type " << expected_type << ", found type " << inputs[pyt_idx].dtype());
144-
auto dims = core::util::toDimsPad(inputs[pyt_idx].sizes(), 1);
137+
inputs[i].dtype() == expected_type,
138+
"Expected input tensors to have type " << expected_type << ", found type " << inputs[i].dtype());
139+
auto dims = core::util::toDimsPad(inputs[i].sizes(), 1);
145140
auto shape = core::util::toVec(dims);
146-
contig_inputs.push_back(inputs[pyt_idx].view(shape).contiguous());
147-
LOG_DEBUG("Input shape: " << dims);
148-
compiled_engine->exec_ctx->setBindingDimensions(i, dims);
149-
gpu_handles.push_back(contig_inputs.back().data_ptr());
141+
LOG_DEBUG("Input Name: " << name << " Shape: " << dims);
142+
compiled_engine->exec_ctx->setInputShape(name.c_str(), dims);
143+
compiled_engine->exec_ctx->setTensorAddress(name.c_str(), inputs[i].view(shape).contiguous().data_ptr());
150144
}
145+
151146
TORCHTRT_CHECK(
152-
compiled_engine->exec_ctx->allInputDimensionsSpecified(),
153-
"Not enough inputs provided (torch.ops.tensorrt.execute_engine)");
147+
compiled_engine->exec_ctx->allInputShapesSpecified(), "Not enough inputs provided (runtime.RunCudaEngine)");
154148
}
155149

156150
std::vector<at::Tensor> outputs(compiled_engine->num_io.second);
@@ -163,26 +157,27 @@ std::vector<at::Tensor> execute_engine(std::vector<at::Tensor> inputs, c10::intr
163157

164158
for (size_t o = inputs.size(); o < (compiled_engine->num_io.first + compiled_engine->num_io.second); o++) {
165159
uint64_t pyt_idx = compiled_engine->out_binding_map[o];
166-
auto out_shape = compiled_engine->exec_ctx->getBindingDimensions(o);
167-
LOG_DEBUG("Output shape: " << out_shape);
160+
std::string name = compiled_engine->out_binding_names[pyt_idx];
161+
auto out_shape = compiled_engine->exec_ctx->getTensorShape(name.c_str());
162+
LOG_DEBUG("Output Name: " << name << " Shape: " << out_shape);
168163
auto dims = core::util::toVec(out_shape);
169-
auto type = util::TRTDataTypeToScalarType(compiled_engine->exec_ctx->getEngine().getBindingDataType(o));
164+
auto type = util::TRTDataTypeToScalarType(compiled_engine->exec_ctx->getEngine().getTensorDataType(name.c_str()));
170165
outputs[pyt_idx] = std::move(at::empty(dims, {at::kCUDA}).to(type).contiguous());
171-
gpu_handles.push_back(outputs[pyt_idx].data_ptr());
166+
compiled_engine->exec_ctx->setTensorAddress(name.c_str(), outputs[pyt_idx].data_ptr());
172167
}
173168
}
169+
174170
{
175171
std::unique_ptr<torch::autograd::profiler::RecordProfile> enqueue_profiler_guard;
176172
if (compiled_engine->profile_execution) {
177173
enqueue_profiler_guard =
178174
std::make_unique<torch::autograd::profiler::RecordProfile>(compiled_engine->enqueue_profile_path);
179175
}
180-
181176
c10::cuda::CUDAStream stream = c10::cuda::getCurrentCUDAStream(inputs[0].device().index());
182177

183178
// nvinfer1::IExecutionContext::enqueue is not thread safe and we need a mutex for it.
184179
std::unique_lock<std::mutex> lock(compiled_engine->mu);
185-
compiled_engine->exec_ctx->enqueueV2(gpu_handles.data(), stream, nullptr);
180+
compiled_engine->exec_ctx->enqueueV3(stream);
186181
if (compiled_engine->profile_execution) {
187182
LOG_INFO(std::endl << *compiled_engine->trt_engine_profiler);
188183
dump_trace(compiled_engine->trt_engine_profile_path, *compiled_engine->trt_engine_profiler);

‎docker/WORKSPACE.docker

Lines changed: 4 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -51,17 +51,17 @@ new_local_repository(
5151
http_archive(
5252
name = "libtorch",
5353
build_file = "@//third_party/libtorch:BUILD",
54-
sha256 = "8d9e829ce9478db4f35bdb7943308cf02e8a2f58cf9bb10f742462c1d57bf287",
54+
sha256 = "59b8b5e1954a86d50b79c13f06398d385b200da13e37a08ecf31d3c62e5ca127",
5555
strip_prefix = "libtorch",
56-
urls = ["https://download.pytorch.org/libtorch/cu113/libtorch-cxx11-abi-shared-with-deps-1.11.0%2Bcu113.zip"],
56+
urls = ["https://download.pytorch.org/libtorch/nightly/cu117/libtorch-cxx11-abi-shared-with-deps-2.0.0.dev20230103%2Bcu117.zip"],
5757
)
5858

5959
http_archive(
6060
name = "libtorch_pre_cxx11_abi",
6161
build_file = "@//third_party/libtorch:BUILD",
62-
sha256 = "90159ecce3ff451f3ef3f657493b6c7c96759c3b74bbd70c1695f2ea2f81e1ad",
62+
sha256 = "e260fc7476be89d1650953e8643e9f7363845f5a52de4bab87ac0e619c1f6ad4",
6363
strip_prefix = "libtorch",
64-
urls = ["https://download.pytorch.org/libtorch/cu113/libtorch-shared-with-deps-1.11.0%2Bcu113.zip"],
64+
urls = ["https://download.pytorch.org/libtorch/nightly/cu117/libtorch-shared-with-deps-2.0.0.dev20230103%2Bcu117.zip"],
6565
)
6666

6767
####################################################################################

‎docs/_cpp_api/classtorch__tensorrt_1_1DataType.html

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -10,7 +10,7 @@
1010

1111
<meta name="viewport" content="width=device-width, initial-scale=1.0">
1212

13-
<title>Class DataType &mdash; Torch-TensorRT v1.4.0dev0+544654f documentation</title>
13+
<title>Class DataType &mdash; Torch-TensorRT v1.4.0dev0+b638e78 documentation</title>
1414

1515

1616

@@ -215,7 +215,7 @@
215215

216216

217217
<div class="version">
218-
v1.4.0dev0+544654f
218+
v1.4.0dev0+b638e78
219219
</div>
220220

221221

‎docs/_cpp_api/classtorch__tensorrt_1_1Device_1_1DeviceType.html

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -10,7 +10,7 @@
1010

1111
<meta name="viewport" content="width=device-width, initial-scale=1.0">
1212

13-
<title>Class Device::DeviceType &mdash; Torch-TensorRT v1.4.0dev0+544654f documentation</title>
13+
<title>Class Device::DeviceType &mdash; Torch-TensorRT v1.4.0dev0+b638e78 documentation</title>
1414

1515

1616

@@ -215,7 +215,7 @@
215215

216216

217217
<div class="version">
218-
v1.4.0dev0+544654f
218+
v1.4.0dev0+b638e78
219219
</div>
220220

221221

0 commit comments

Comments
 (0)