Vadim Kantorov

This seems to succeed: ```yaml name: tritonservercppsharp on: workflow_dispatch env: PLATFORM: x64 FRAMEWORK: net80 FRAMEWORKDOT: net8.0 jobs: tritonservercppsharp: runs-on: ubuntu-22.04 steps: - name: Clone and build CppSharp run: | git...

Same problem. And even without re-attaching to a smaller screen (in fact re-attaching to the same screen, but maybe tmux thinks it's a smaller screen somehow). But what's true is...

Are there any workarounds for this? Just hit this again, but when all of tqdm progress certainly fits a single terminal line

Is it because somehow a `\n` char gets printed somewhere? And so the progress is repeated because it starts in the next line? Is there any environment variable to disable...

Related: - https://github.com/pytorch/pytorch/issues/35666 - https://github.com/pytorch/pytorch/issues/41527 - https://github.com/pytorch/pytorch/issues/66707 Regarding the eps, there is `torch.finfo(x).tiny`, but I think torch.finfo still is not scriptable: - https://github.com/pytorch/pytorch/issues/41492

Also struggling how to build a binary correctly with both a static zlib and a static libarchive. If I add it as `-Wl,--whole-archive libz.a` to the target binary, I'm getting...

I wrote down a primer of a minimalistic workflow file I meant above, looking to generate C API bindings to https://github.com/triton-inference-server/core/blob/main/include/triton/core/tritonserver.h Does CppSharp absolutely need to have access to CppSharp...

> Also to note, for some applications, the CppSharp CLI might be useful since it does not need a custom C# code setup Is there an example of using this...

For some reason I'm getting problems loading the generated `exe` (depsite that I copy all the `.dll`s and `.so`s in the current dir: ``` Unhandled exception. System.IO.FileNotFoundException: Could not load...

Only on CI. It appears that dotnet does not like the `libSystem.Native.so` which is found successfully on system path and so it proceeds checking the local directory, it cannot find...