stable_diffusion_v1_4使用torchtrt推理时报错

[MaltoseFlower]

ERROR: [Torch-TensorRT] - Unsupported operator: aten::to.dtype_layout(Tensor(a) self, *, ScalarType? dtype=None, Layout? layout=None, Device? device=None, bool? pin_memory=None, bool non_blocking=False, bool copy=False, MemoryFormat? memory_format=None) -> Tensor(a) /workspace/luor/FlagPerf/inference/benchmarks/stable_diffusion_v1_4/pytorch/model_utils/unet2d.py(854): forward /root/miniconda3/envs/fperf/lib/python3.10/site-packages/torch/nn/modules/module.py(1501): _slow_forward /root/miniconda3/envs/fperf/lib/python3.10/site-packages/torch/nn/modules/module.py(1520): _call_impl /root/miniconda3/envs/fperf/lib/python3.10/site-packages/torch/nn/modules/module.py(1511): _wrapped_call_impl /root/miniconda3/envs/fperf/lib/python3.10/site-packages/torch/jit/_trace.py(1074): trace_module /root/miniconda3/envs/fperf/lib/python3.10/site-packages/torch/jit/_trace.py(806): trace /workspace/luor/FlagPerf/inference/inference_engine/nvidia/torchtrt.py(23): call /workspace/luor/FlagPerf/inference/benchmarks/stable_diffusion_v1_4/pytorch/forward.py(209): engine_forward /workspace/luor/FlagPerf/inference/run_inference.py(80): main /workspace/luor/FlagPerf/inference/run_inference.py(137):

Traceback (most recent call last): File "/workspace/luor/FlagPerf/inference/run_inference.py", line 137, in config, p_forward, p_infer, p_forward_core, p_infer_core, val_acc, infer_acc = main( File "/workspace/luor/FlagPerf/inference/run_inference.py", line 80, in main p_infer, p_infer_core, infer_acc = benchmark_module.engine_forward( File "/workspace/luor/FlagPerf/inference/benchmarks/stable_diffusion_v1_4/pytorch/forward.py", line 209, in engine_forward outputs = model(inputs) File "/workspace/luor/FlagPerf/inference/inference_engine/nvidia/torchtrt.py", line 25, in call self.trt_model = torchtrt.compile( File "/root/miniconda3/envs/fperf/lib/python3.10/site-packages/torch_tensorrt/_compile.py", line 185, in compile compiled_ts_module: torch.jit.ScriptModule = torchscript_compile( File "/root/miniconda3/envs/fperf/lib/python3.10/site-packages/torch_tensorrt/ts/_compiler.py", line 151, in compile compiled_cpp_mod = _C.compile_graph(module._c, _parse_compile_spec(spec)) RuntimeError: [Error thrown at core/compiler.cpp:413] Expected conversion::VerifyConverterSupportForBlock(g->block()) to be true but got false Not all operations in graph are supported by the compiler

请问测试时，官方使用的软件版本是多少？ 以下是我测试时的版本： libtorch 2.4.0
torch 2.2.2
torch-tensorrt 2.2.0
torchmetrics 1.4.2
tensorrt 8.6.1.post1
tensorrt-bindings 8.6.1
tensorrt-libs 8.6.1

[MaltoseFlower]

btw，官方写的stable_diffusion_v1_4 pytorch benchmark推理脚本不适配 nvidia inference engine。会报错，不过比较容易修复。

[shh2000]

ngctorch23:06-py3.