EngineX-Iluvatar/enginex-mr_series-sherpa-onnx
Archived
16
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

Support building GPU-capable sherpa-onnx on Linux aarch64. (#1500)

Browse Source 
Thanks to @Peakyxh for providing pre-built onnxruntime libraries 
with CUDA support for Linux aarch64.

Tested on Jetson nano b01
This commit is contained in:
Fangjun Kuang
2024-11-01 11:16:28 +08:00
committed by GitHub
parent a3c89aa0d8
commit 9ab89c33bc
41 changed files with 537 additions and 291 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
sherpa-onnx/csrc/online-lstm-transducer-model.cc
View File

@@ -158,9 +158,10 @@ std::vector<Ort::Value> OnlineLstmTransducerModel::StackStates(
h_buf[i] = &states[i][0];
c_buf[i] = &states[i][1];
}
auto allocator = const_cast<OnlineLstmTransducerModel *>(this)->allocator_;
Ort::Value h = Cat(allocator_, h_buf, 1);
Ort::Value c = Cat(allocator_, c_buf, 1);
Ort::Value h = Cat(allocator, h_buf, 1);
Ort::Value c = Cat(allocator, c_buf, 1);
std::vector<Ort::Value> ans;
ans.reserve(2);
@@ -177,8 +178,10 @@ std::vector<std::vector<Ort::Value>> OnlineLstmTransducerModel::UnStackStates(
std::vector<std::vector<Ort::Value>> ans(batch_size);
std::vector<Ort::Value> h_vec = Unbind(allocator_, &states[0], 1);
std::vector<Ort::Value> c_vec = Unbind(allocator_, &states[1], 1);
auto allocator = const_cast<OnlineLstmTransducerModel *>(this)->allocator_;
std::vector<Ort::Value> h_vec = Unbind(allocator, &states[0], 1);
std::vector<Ort::Value> c_vec = Unbind(allocator, &states[1], 1);
assert(h_vec.size() == batch_size);
assert(c_vec.size() == batch_size);