pranavm-nvidia

@yeandy With Sphinx, you can specify modules to mock in your `conf.py` ([example](https://github.com/NVIDIA/TensorRT/blob/main/tools/Polygraphy/docs/conf.py#L34)). With that, you won't need to have TensorRT installed.

@Fuzhiyuan Are you trying to build from source, or use the pre-built wheels? For the pre-built wheels, all you need to do is: ``` python -m pip install colored polygraphy...

Thanks for pointing this out, will fix

@deephog You can automate it using the `--check` option, similar to [debug reduce](https://github.com/NVIDIA/TensorRT/tree/main/tools/Polygraphy/examples/cli/debug/02_reducing_failing_onnx_models#reducing-failing-onnx-models). For example: 1. Dump out golden outputs with: ``` polygraphy run test.onnx --onnxrt --save-outputs golden.json ``` 2....

Looks like a `CUBLAS_STATUS_ALLOC_FAILED`: ``` cuBLAS initialization failed: 3 ``` Maybe you're running out of memory on your GPU? Do other networks work on this GPU? I'm wondering if it...

Can you try marking fewer outputs?

Can you try increasing the workspace size? e.g. `--pool-limit workspace:1G`

It sounds like a TensorRT bug. Can you share the extracted model where you're seeing this?

Thanks, I've filed an internal issue to track this (internal id: 3742810). In the meantime, you should be able to work around this by either marking a specific set of...

`debug reduce` will remove layers from the model until you're left with a minimal model that reproduces the failure. So it won't show you the precision loss at each layer,...