esp-dl
esp-dl copied to clipboard
espressif
ESP32 Model Deployment Project using TVM (AIV-682)
Hello guys, I am trying to deploy my custom Yolov5 model to Esp32s3, but i am facing that issue while i try to "export_onnx_model.py", I am getting this error.
` Model Information:
Input Name: images
Input Shape: (1, 3, 192, 192)
Input DType: float
Output Name: output0
Output Shape: (1, 2268, 6)
Output DType: float
[13:06:06] /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/convert_layout.cc:99: Warning: Desired layout(s) not specified for op: nn.max_pool2d
[13:06:06] /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/convert_layout.cc:99: Warning: Desired layout(s) not specified for op: nn.max_pool2d
[13:06:06] /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/convert_layout.cc:99: Warning: Desired layout(s) not specified for op: nn.max_pool2d
[13:06:06] /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/convert_layout.cc:99: Warning: Desired layout(s) not specified for op: image.resize2d
[13:06:06] /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/convert_layout.cc:99: Warning: Desired layout(s) not specified for op: image.resize2d
Traceback (most recent call last):
File "../../tools/tvm/export_onnx_model.py", line 237, in
Thank you for any response.
you can try non quantitative models，example：python ../../tools/tvm/export_onnx_model.py --model_path ${onnx_model}_opt.onnx --img_path ${input_file}.npy --target_chip esp32s3 --out_path "." --template_path "../../tools/tvm/template_project_for_model/"
Ok, i'll try and share the results asap! thank you.
upgrade onnx==1.13.0, It works!!!
hare the results asap! thank you.
upgrade onnx==1.13.0, It works!!!
Non quantized solution works. Thanks
Ok, i'll try and share the results asap! thank you.
upgrade onnx==1.13.0, It works!!!
Upgradng onnx is also works, thanks a lot.
Hi,
I have the same problem, i was try to updrade onnx==1.13.0 but show me same problem
and when i use a Non quantized model another problem appear
nst, tvm::Range> > > const&, unsigned long, bool, std::unordered_set<tvm::tir::IterVar, std::hash<tvm::tir::IterVar>, std::equal_to<tvm::tir::IterVar>, std::allocator<tvm::tir::IterVar> > const&, std::unordered_map<tvm::tir::IterVar, tvm::PrimExpr, std::hash<tvm::tir::IterVar>, std::equal_to<tvm::tir::IterVar>, std::allocator<std::pair<tvm::tir::IterVar const, tvm::PrimExpr> > >*, bool) at /home/gansichen/Workspace/projects/local/framework/tvm/src/te/operation/op_utils.cc:206 3: tvm::te::PassUpIndex(tvm::te::Stage const&, tvm::runtime::Map<tvm::tir::IterVar, tvm::Range, void, void> const&, std::unordered_map<tvm::tir::IterVar, tvm::PrimExpr, std::hash<tvm::tir::IterVar>, std::equal_to<tvm::tir::IterVar>, std::allocator<std::pair<tvm::tir::IterVar const, tvm::PrimExpr> > >*, bool) at /home/gansichen/Workspace/projects/local/framework/tvm/src/te/schedule/message_passing.cc:244 2: tvm::indexdiv(tvm::PrimExpr, tvm::PrimExpr, tvm::Span) at /home/gansichen/Workspace/projects/local/framework/tvm/src/tir/op/op.cc:399 1: tvm::floordiv(tvm::PrimExpr, tvm::PrimExpr, tvm::Span) at /home/gansichen/Workspace/projects/local/framework/tvm/src/tir/op/op.cc:409 0: tvm::runtime::Optional<tvm::PrimExpr> tvm::arith::TryConstFold<tvm::tir::FloorDiv>(tvm::PrimExpr, tvm::PrimExpr) at /home/gansichen/Workspace/projects/local/framework/tvm/src/tir/transforms/../../arith/const_fold.h:273 File "/home/gansichen/Workspace/projects/local/framework/tvm/src/arith/const_fold.h", line 302 InternalError: Check failed: pb->value != 0 (0 vs. 0) : Divide by zero
I use yolov5s model Any suggestion please thanks
Hello guys, I am trying to deploy my custom Yolov5 model to Esp32s3, but i am facing that issue while i try to "export_onnx_model.py", I am getting this error.
`
Model Information: Input Name: images Input Shape: (1, 3, 192, 192) Input DType: float Output Name: output0 Output Shape: (1, 2268, 6) Output DType: float [13:06:06] /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/convert_layout.cc:99: Warning: Desired layout(s) not specified for op: nn.max_pool2d [13:06:06] /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/convert_layout.cc:99: Warning: Desired layout(s) not specified for op: nn.max_pool2d [13:06:06] /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/convert_layout.cc:99: Warning: Desired layout(s) not specified for op: nn.max_pool2d [13:06:06] /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/convert_layout.cc:99: Warning: Desired layout(s) not specified for op: image.resize2d [13:06:06] /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/convert_layout.cc:99: Warning: Desired layout(s) not specified for op: image.resize2d Traceback (most recent call last): File "../../tools/tvm/export_onnx_model.py", line 237, in dump_tvm_onnx_project(target_chip=args.target_chip, model_path=args.model_path, img_path=args.img_path, template_path=args.template_path, generated_project_path=args.out_path) File "../../tools/tvm/export_onnx_model.py", line 167, in dump_tvm_onnx_project mod = preprocess_onnx_for_esp(mod, params) File "/home/train/esp-dl/tools/tvm/python/tvm/relay/op/contrib/esp.py", line 61, in preprocess_onnx_for_esp return seq(mod) File "/home/train/esp-dl/tools/tvm/python/tvm/ir/transform.py", line 160, in call return _ffi_transform_api.RunPass(self, mod) File "/home/train/esp-dl/tools/tvm/python/tvm/_ffi/_ctypes/packed_func.py", line 238, in call raise get_last_ffi_error() tvm.error.InternalError: Traceback (most recent call last): 84: 0xffffffffffffffff 83: _start 82: Py_BytesMain 81: Py_RunMain 80: PyRun_SimpleFileExFlags 79: 0x0000000000673b60 78: 0x0000000000673aba 77: 0x0000000000673a40 76: PyEval_EvalCode 75: _PyEval_EvalCodeWithName 74: _PyEval_EvalFrameDefault 73: PyFunction_Vectorcall 72: PyEval_EvalCodeWithName 71: PyEval_EvalFrameDefault 70: PyFunction_Vectorcall 69: PyEval_EvalCodeWithName 68: PyEval_EvalFrameDefault 67: PyObject_MakeTpCall 66: 0x000000000057a42d 65: PyFunction_Vectorcall 64: PyEval_EvalFrameDefault 63: PyObject_MakeTpCall 62: 0x000000000057a42d 61: PyFunction_Vectorcall 60: PyEval_EvalCodeWithName 59: PyEval_EvalFrameDefault 58: PyObject_MakeTpCall 57: 0x00007f834ca199db 56: ctypes_callproc 55: 0x00007f83422be409 54: 0x00007f83422beff4 53: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::IRModule, tvm::transform::PassContext>(tvm::IRModule&&, tvm::transform::PassContext&&) const 52: operator() at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/backend/contrib/esp/generate_constants.cc:325 51: tvm::relay::contrib::esp::GenerateConstants(tvm::IRModule const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/backend/contrib/esp/generate_constants.cc:312 50: tvm::relay::MixedModeMutator::VisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:102 49: ExpandDataflow<tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>, tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)> > at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:509 48: ExpandDataflow<tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>, tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>, tvm::relay::ExpandDataflow(tvm::relay::Expr, FCheckVisited, FVisitLeaf) [with FCheckVisited = tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>; FVisitLeaf = tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>]::<lambda(const Expr&)> > at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:468 47: operator() at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:98 46: tvm::relay::MixedModeMutator::VisitLeaf(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:81 45: tvm::relay::MixedModeMutator::DispatchVisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:94 44: tvm::relay::ExprMutator::VisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:158 43: tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>::VisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:95 42: tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>::InitVTable()::{lambda(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>)#6}::FUN(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:128 41: tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>::InitVTable()::{lambda(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>)#6}::operator()(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>) const at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:128 40: tvm::relay::MixedModeMutator::VisitExpr(tvm::relay::CallNode const*) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:291 39: tvm::RelayExpr tvm::relay::MixedModeMutator::Rewritetvm::relay::CallNode(tvm::relay::CallNode const*) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:313 38: tvm::relay::contrib::esp::GenerateConstantsMutator::Rewrite(tvm::relay::CallNode const*, tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/backend/contrib/esp/generate_constants.cc:273 37: tvm::relay::MixedModeMutator::VisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:102 36: ExpandDataflow<tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>, tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)> > at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:509 35: ExpandDataflow<tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>, tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>, tvm::relay::ExpandDataflow(tvm::relay::Expr, FCheckVisited, FVisitLeaf) [with FCheckVisited = tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>; FVisitLeaf = tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>]::<lambda(const Expr&)> > at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:468 34: operator() at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:98 33: tvm::relay::MixedModeMutator::VisitLeaf(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:81 32: tvm::relay::MixedModeMutator::DispatchVisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:94 31: tvm::relay::ExprMutator::VisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:158 30: tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>::VisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:95 29: tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>::InitVTable()::{lambda(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>)#6}::FUN(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:128 28: tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>::InitVTable()::{lambda(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>)#6}::operator()(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>) const at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:128 27: tvm::relay::MixedModeMutator::VisitExpr(tvm::relay::CallNode const*) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:291 26: tvm::RelayExpr tvm::relay::MixedModeMutator::Rewritetvm::relay::CallNode(tvm::relay::CallNode const*) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:313 25: tvm::relay::contrib::esp::GenerateConstantsMutator::Rewrite(tvm::relay::CallNode const*, tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/backend/contrib/esp/generate_constants.cc:290 24: tvm::relay::contrib::esp::GenerateConstantsMutator::GenerateConv2dExpConstants(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/backend/contrib/esp/generate_constants.cc:214 23: tvm::relay::contrib::esp::GenerateConstantsMutator::ConvertConv2dLayout(tvm::RelayExpr, tvm::relay::Conv2DAttrs const*, tvm::Attrs*) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/backend/contrib/esp/generate_constants.cc:167 22: tvm::relay::transform::FoldConstantExpr(tvm::RelayExpr const&, bool) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/fold_constant.cc:430 21: tvm::relay::transform::FoldConstantExpr(tvm::RelayExpr const&, tvm::IRModule const&, bool) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/fold_constant.cc:423 20: tvm::relay::MixedModeMutator::VisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:102 19: ExpandDataflow<tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>, tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)> > at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:509 18: ExpandDataflow<tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>, tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>, tvm::relay::ExpandDataflow(tvm::relay::Expr, FCheckVisited, FVisitLeaf) [with FCheckVisited = tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>; FVisitLeaf = tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>]::<lambda(const Expr&)> > at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:468 17: operator() at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:98 16: tvm::relay::MixedModeMutator::VisitLeaf(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:81 15: tvm::relay::MixedModeMutator::DispatchVisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:94 14: tvm::relay::ExprMutator::VisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:158 13: tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>::VisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:95 12: tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>::InitVTable()::{lambda(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>)#6}::FUN(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:128 11: tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>::InitVTable()::{lambda(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>)#6}::operator()(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>) const at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:128 10: tvm::relay::MixedModeMutator::VisitExpr(tvm::relay::CallNode const*) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:291 9: tvm::RelayExpr tvm::relay::MixedModeMutator::Rewritetvm::relay::CallNode(tvm::relay::CallNode const*) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:313 8: Rewrite at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/fold_constant.cc:202 7: ConstEvaluate at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/fold_constant.cc:270 6: tvm::relay::Eval(tvm::RelayExpr, tvm::runtime::Map<tvm::GlobalTypeVar, tvm::TypeData, void, void>, std::unordered_set<tvm::runtime::String, std::hashtvm::runtime::String, std::equal_totvm::runtime::String, std::allocatortvm::runtime::String >, DLDevice, tvm::Target, tvm::runtime::Map<tvm::runtime::String, tvm::runtime::ObjectRef, void, void>) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/backend/interpreter.cc:1108 5: tvm::relay::Prepare(tvm::IRModule, tvm::CompilationConfig const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/backend/interpreter.cc:961 4: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::IRModule, tvm::transform::PassContext>(tvm::IRModule&&, tvm::transform::PassContext&&) const 3: operator() at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/type_infer.cc:981 2: tvm::relay::TypeInferencer::Infer(tvm::GlobalVar, tvm::relay::Function) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/type_infer.cc:787 1: tvm::relay::TypeInferencer::Solve() at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/type_infer.cc:623 0: tvm::relay::TypeSolver::Solve() at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/analysis/type_solver.cc:643 86: 0xffffffffffffffff 85: _start 84: Py_BytesMain 83: Py_RunMain 82: PyRun_SimpleFileExFlags 81: 0x0000000000673b60 80: 0x0000000000673aba 79: 0x0000000000673a40 78: PyEval_EvalCode 77: _PyEval_EvalCodeWithName 76: _PyEval_EvalFrameDefault 75: PyFunction_Vectorcall 74: PyEval_EvalCodeWithName 73: PyEval_EvalFrameDefault 72: PyFunction_Vectorcall 71: PyEval_EvalCodeWithName 70: PyEval_EvalFrameDefault 69: PyObject_MakeTpCall 68: 0x000000000057a42d 67: PyFunction_Vectorcall 66: PyEval_EvalFrameDefault 65: PyObject_MakeTpCall 64: 0x000000000057a42d 63: PyFunction_Vectorcall 62: PyEval_EvalCodeWithName 61: PyEval_EvalFrameDefault 60: PyObject_MakeTpCall 59: 0x00007f834ca199db 58: ctypes_callproc 57: 0x00007f83422be409 56: 0x00007f83422beff4 55: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::IRModule, tvm::transform::PassContext>(tvm::IRModule&&, tvm::transform::PassContext&&) const 54: operator() at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/backend/contrib/esp/generate_constants.cc:325 53: tvm::relay::contrib::esp::GenerateConstants(tvm::IRModule const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/backend/contrib/esp/generate_constants.cc:312 52: tvm::relay::MixedModeMutator::VisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:102 51: ExpandDataflow<tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>, tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)> > at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:509 50: ExpandDataflow<tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>, tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>, tvm::relay::ExpandDataflow(tvm::relay::Expr, FCheckVisited, FVisitLeaf) [with FCheckVisited = tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>; FVisitLeaf = tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>]::<lambda(const Expr&)> > at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:468 49: operator() at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:98 48: tvm::relay::MixedModeMutator::VisitLeaf(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:81 47: tvm::relay::MixedModeMutator::DispatchVisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:94 46: tvm::relay::ExprMutator::VisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:158 45: tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>::VisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:95 44: tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>::InitVTable()::{lambda(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>)#6}::FUN(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:128 43: tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>::InitVTable()::{lambda(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>)#6}::operator()(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>) const at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:128 42: tvm::relay::MixedModeMutator::VisitExpr(tvm::relay::CallNode const*) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:291 41: tvm::RelayExpr tvm::relay::MixedModeMutator::Rewritetvm::relay::CallNode(tvm::relay::CallNode const*) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:313 40: tvm::relay::contrib::esp::GenerateConstantsMutator::Rewrite(tvm::relay::CallNode const*, tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/backend/contrib/esp/generate_constants.cc:273 39: tvm::relay::MixedModeMutator::VisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:102 38: ExpandDataflow<tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>, tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)> > at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:509 37: ExpandDataflow<tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>, tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>, tvm::relay::ExpandDataflow(tvm::relay::Expr, FCheckVisited, FVisitLeaf) [with FCheckVisited = tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>; FVisitLeaf = tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>]::<lambda(const Expr&)> > at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:468 36: operator() at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:98 35: tvm::relay::MixedModeMutator::VisitLeaf(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:81 34: tvm::relay::MixedModeMutator::DispatchVisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:94 33: tvm::relay::ExprMutator::VisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:158 32: tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>::VisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:95 31: tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>::InitVTable()::{lambda(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>)#6}::FUN(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:128 30: tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>::InitVTable()::{lambda(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>)#6}::operator()(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>) const at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:128 29: tvm::relay::MixedModeMutator::VisitExpr(tvm::relay::CallNode const*) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:291 28: tvm::RelayExpr tvm::relay::MixedModeMutator::Rewritetvm::relay::CallNode(tvm::relay::CallNode const*) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:313 27: tvm::relay::contrib::esp::GenerateConstantsMutator::Rewrite(tvm::relay::CallNode const*, tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/backend/contrib/esp/generate_constants.cc:290 26: tvm::relay::contrib::esp::GenerateConstantsMutator::GenerateConv2dExpConstants(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/backend/contrib/esp/generate_constants.cc:214 25: tvm::relay::contrib::esp::GenerateConstantsMutator::ConvertConv2dLayout(tvm::RelayExpr, tvm::relay::Conv2DAttrs const*, tvm::Attrs*) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/backend/contrib/esp/generate_constants.cc:167 24: tvm::relay::transform::FoldConstantExpr(tvm::RelayExpr const&, bool) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/fold_constant.cc:430 23: tvm::relay::transform::FoldConstantExpr(tvm::RelayExpr const&, tvm::IRModule const&, bool) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/fold_constant.cc:423 22: tvm::relay::MixedModeMutator::VisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:102 21: ExpandDataflow<tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>, tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)> > at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:509 20: ExpandDataflow<tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>, tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>, tvm::relay::ExpandDataflow(tvm::relay::Expr, FCheckVisited, FVisitLeaf) [with FCheckVisited = tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>; FVisitLeaf = tvm::relay::MixedModeMutator::VisitExpr(const Expr&)::<lambda(const Expr&)>]::<lambda(const Expr&)> > at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:468 19: operator() at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:98 18: tvm::relay::MixedModeMutator::VisitLeaf(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:81 17: tvm::relay::MixedModeMutator::DispatchVisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:94 16: tvm::relay::ExprMutator::VisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/ir/expr_functor.cc:158 15: tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>::VisitExpr(tvm::RelayExpr const&) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:95 14: tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>::InitVTable()::{lambda(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>)#6}::FUN(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:128 13: tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>::InitVTable()::{lambda(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>)#6}::operator()(tvm::runtime::ObjectRef const&, tvm::relay::ExprFunctor<tvm::RelayExpr (tvm::RelayExpr const&)>) const at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:128 12: tvm::relay::MixedModeMutator::VisitExpr(tvm::relay::CallNode const*) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:291 11: tvm::RelayExpr tvm::relay::MixedModeMutator::Rewritetvm::relay::CallNode(tvm::relay::CallNode const*) at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/relay/expr_functor.h:313 10: Rewrite at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/fold_constant.cc:202 9: ConstEvaluate at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/fold_constant.cc:270 8: tvm::relay::Eval(tvm::RelayExpr, tvm::runtime::Map<tvm::GlobalTypeVar, tvm::TypeData, void, void>, std::unordered_set<tvm::runtime::String, std::hashtvm::runtime::String, std::equal_totvm::runtime::String, std::allocatortvm::runtime::String >, DLDevice, tvm::Target, tvm::runtime::Map<tvm::runtime::String, tvm::runtime::ObjectRef, void, void>) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/backend/interpreter.cc:1108 7: tvm::relay::Prepare(tvm::IRModule, tvm::CompilationConfig const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/backend/interpreter.cc:961 6: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::IRModule, tvm::transform::PassContext>(tvm::IRModule&&, tvm::transform::PassContext&&) const 5: operator() at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/type_infer.cc:981 4: tvm::relay::TypeInferencer::Infer(tvm::GlobalVar, tvm::relay::Function) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/type_infer.cc:787 3: tvm::relay::TypeInferencer::Solve() at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/transforms/type_infer.cc:623 2: tvm::relay::TypeSolver::Solve() at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/analysis/type_solver.cc:632 1: tvm::TypedEnvFunc<bool (tvm::runtime::Array<tvm::Type, void> const&, int, tvm::Attrs const&, tvm::TypeReporter const&)>::operator()(tvm::runtime::Array<tvm::Type, void> const&, int, tvm::Attrs const&, tvm::TypeReporter const&) const at /home/gansichen/Workspace/projects/local/framework/tvm/include/tvm/ir/env_func.h:142 0: tvm::relay::SplitRel(tvm::runtime::Array<tvm::Type, void> const&, int, tvm::Attrs const&, tvm::TypeReporter const&) at /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/op/tensor/transform.cc:3070 File "/home/gansichen/Workspace/projects/local/framework/tvm/src/relay/analysis/type_solver.cc", line 643 InternalError: Check failed: (false) is false: [13:09:31] /home/gansichen/Workspace/projects/local/framework/tvm/src/relay/op/tensor/transform.cc:3070: InternalError: Check failed: (reporter->Assert(indices[i] > begin)) is false: indices_or_sections need to be a sorted ascending list `
Thank you for any response.
Hi,
Can you please tell me witch version yolo used (yolov5s, yolov5m, yolov5l ?) and if you have change the architecutre(layer/function) from model
Thanks
Do you guys got running models with TVM ? Where you able to export the yolo model? I used TVM as well but it was inconsistent and buggy.
Do you guys got running models with TVM ? Where you able to export the yolo model? I used TVM as well but it was inconsistent and buggy.
Not yet, I'm still waiting for the return of @YigitKaanBingol
Sorry guys, I've just able to deploy it and run the example, unfortunately i had no time to test accuracy of model.
Hi,
I have the same problem, i was try to updrade onnx==1.13.0 but show me same problem and when i use a Non quantized model another problem appear
nst, tvm::Range> > > const&, unsigned long, bool, std::unordered_set<tvm::tir::IterVar, std::hash<tvm::tir::IterVar>, std::equal_to<tvm::tir::IterVar>, std::allocator<tvm::tir::IterVar> > const&, std::unordered_map<tvm::tir::IterVar, tvm::PrimExpr, std::hash<tvm::tir::IterVar>, std::equal_to<tvm::tir::IterVar>, std::allocator<std::pair<tvm::tir::IterVar const, tvm::PrimExpr> > >*, bool) at /home/gansichen/Workspace/projects/local/framework/tvm/src/te/operation/op_utils.cc:206 3: tvm::te::PassUpIndex(tvm::te::Stage const&, tvm::runtime::Map<tvm::tir::IterVar, tvm::Range, void, void> const&, std::unordered_map<tvm::tir::IterVar, tvm::PrimExpr, std::hash<tvm::tir::IterVar>, std::equal_to<tvm::tir::IterVar>, std::allocator<std::pair<tvm::tir::IterVar const, tvm::PrimExpr> > >*, bool) at /home/gansichen/Workspace/projects/local/framework/tvm/src/te/schedule/message_passing.cc:244 2: tvm::indexdiv(tvm::PrimExpr, tvm::PrimExpr, tvm::Span) at /home/gansichen/Workspace/projects/local/framework/tvm/src/tir/op/op.cc:399 1: tvm::floordiv(tvm::PrimExpr, tvm::PrimExpr, tvm::Span) at /home/gansichen/Workspace/projects/local/framework/tvm/src/tir/op/op.cc:409 0: tvm::runtime::Optional<tvm::PrimExpr> tvm::arith::TryConstFold<tvm::tir::FloorDiv>(tvm::PrimExpr, tvm::PrimExpr) at /home/gansichen/Workspace/projects/local/framework/tvm/src/tir/transforms/../../arith/const_fold.h:273 File "/home/gansichen/Workspace/projects/local/framework/tvm/src/arith/const_fold.h", line 302 InternalError: Check failed: pb->value != 0 (0 vs. 0) : Divide by zeroI use yolov5s model Any suggestion please thanks
I used Yolov5n model and didn't change any layer or function.
Output Shape: (1, 2268, 6)
Output Shape: (1, 2268, 6) the 6 indicates the number of classes, in fact you only trained on these 6 classes, right? if I understand correctly
Thanks
I dont think so, i trained yolov5n model for one class, and export to .onnx format. The reason for Output Shape: (1, 2268, 6), 6 is for x,y,w,h,confidence,class, if you have 80 classes, your shape will be -> (1,2268,85) 2268 is based on your exportted model input shape. I hope i made myself clear.
I dont think so, i trained yolov5n model for one class, and export to .onnx format. The reason for Output Shape: (1, 2268, 6), 6 is for x,y,w,h,confidence,class, if you have 80 classes, your shape will be -> (1,2268,85) 2268 is based on your exportted model input shape. I hope i made myself clear.
thank you for these answers