-
### Describe the issue
when i use gemm_float8 to run with input A(fp8 e5m2), input B(fp8 e4m3), can not run, but input A(fp8 e4m3), input B(fp8 e4m3) will run right,
### To reproduce
run gemm_floa…
-
### Describe the issue
Running the htdemucs model on DML gives incorrect results compared to the CPU execution.
### To reproduce
The model we are using is the htdemucs v4 model from here: https://g…
-
### Describe the issue
I am profiling performance of onnx model converted from PyTorch2.3.0-cu11.8, it shows that the performance is little bit slower than the torch version. Is there something i mis…
-
### Is your feature request related to a problem? Please describe.
Currently, it is not possible to run the headless on ARM CPUs without a bit of tinkering.
The architecture is also officially uns…
-
We need to define a plugin architecture that allows platform providers to easily write integrations for Kalabox.
Some sort of standard spec around what these should look like would be good. The plug…
pirog updated
10 years ago
-
### Describe the issue
use shape_inference.quant_pre_process to preprocess will result in error even if i set skip_optimization=True
![image](https://github.com/microsoft/onnxruntime/assets/12644192…
-
Hi, I've tried to change lambda memory and architecture to arm.
How do write the configs?
```
///
export default $config({
app(input) {
return {
name: `MyAPP`,
remov…
-
Related issues: #664 #359
With the in-tree provider becoming obsolete, we need basic documentation that covers the following:
* Key concepts about the out-of-tree provider (what is it, why/when…
-
### Describe the issue
Below command produces an embedded TRT engine at `./test/model_ctx.onnx`:
```
.\onnxruntime_perf_test.exe -I -e tensorrt -r 10 -i "trt_timing_cache_enable|1 trt_engine_cache…
-
### **Feature request**:
On the course the current code is headed, it can get really messy to manage state and other code.
It'd be best if architecture is implemented when the code hasn't grown t…