DefTruth
DefTruth
## Bug Description ## To Reproduce Steps to reproduce the behavior: 1. 2. 3. ## Expected behavior ## Environment > Build information about Torch-TensorRT can be found by turning on...
- add fp16 dtype support for paddle parser
Features: * export_onnx.py to export dynamic and static onnx files. * infer_onnx.py, test demo. * C++ inference for MGMatting export onnx: ```shell # cd to code-base, then run PYTHONPATH=. python3...
QA_corpus数据集方便介绍一下吗？
FILL IN THE PR DESCRIPTION HERE FIX https://github.com/vllm-project/vllm/issues/4171 allow non-power-of-two head sizes in prefix prefill with alibi, this is a small fix based on https://github.com/vllm-project/vllm/pull/4128. **BEFORE SUBMITTING, PLEASE READ THE...
## error log | 日志或报错信息 | ログ 想把RobustVideoMatting（稳健视频抠图模型）用ncnn跑起来，经过一番折腾后成功转成静态维度的onnx，并且使用了onnxsim去除了胶水op，用onnxruntime可以顺利跑通转换后的模型；然后将其转换成ncnn，在转换时报错： ```shell onnx2ncnn rvm_mobilenetv3_fp32-480-480-sim.onnx rvm_mobilenetv3_fp32-480-480.param rvm_mobilenetv3_fp32-480-480.bin Unsupported split axis ! Unsupported split axis ! ``` 在使用ncnn的c++和python接口进行推理时，出现错误，c++提示空指针，python的则是直接崩溃。  所有ex.input返回的错误码都是0，应该是正常的，但是在ex.extract("fgr",fgr)开始，返回的错误码是-100，data是一个空指针，导致cv::Mat初始化出错，程序崩溃。折腾了2天也没解决，目前依然没有头绪🤔，于是决定来这里占个坑~ 🙄 ## model | 模型...
### Your current environment ```text The output of `python collect_env.py` Collecting environment information... PyTorch version: 2.3.0+cu121 Is debug build: False CUDA used to build PyTorch: 12.1 ROCM used to build...