No module named 'tensorrt_llm.bindings'` error message

[maulikmadhavi]

Installing using [pixi](https://pixi.sh/latest/) library management tool. The pixi.toml manifest file looks like this.

[project]
description = "Add a short description here"
name = "TensorRT-LLM"
platforms = ["linux-64"]
version = "0.1.0"
channels = ["nvidia", "conda-forge", {channel = "pytorch", priority = -1}]

[pypi-options]
index-url = "https://pypi.nvidia.com"
extra-index-urls = ["https://pypi.org/simple"]

[dependencies]
python = "==3.12.3"
mpi4py = ">=4.0.1,<5"

[pypi-dependencies]
torch = { version = "==2.5.1+cu118", index = "https://download.pytorch.org/whl/cu118/" }
tqdm = ">=4.66.5, <5"
torchvision = { version = "==0.20.1+cu118", index = "https://download.pytorch.org/whl/cu118/" }
nvidia-cutlass = ">=3.6.0.0, <4"
tensorrt_cu12_libs = { version = ">=10.7.0", index = "https://pypi.nvidia.com" }
tensorrt_cu12_bindings = { version = ">=10.7.0", index = "https://pypi.nvidia.com" }
tensorrt = { version = ">=10.7.0", index = "https://pypi.nvidia.com" }
tensorrt-llm = {version = "==0.16.0",  index = "https://pypi.nvidia.com" }
onnx = ">=1.17.0, <2"
huggingface-hub = ">=0.27.0, <0.28"

Upon running pixi shell can build the environment setup is successful. However, on checking import tensorrt_llm, it failed with ModuleNotFoundError: No module named 'tensorrt_llm.bindings' error message.

Can anyone suggest me with any missing inputs or anything?

Thanks
Maulik

[nv-guomingz]

#1194
Hi @maulikmadhavi Please refer to above issue link. We don't verify the pixi yet so we still recommend u follow the doc to install/run trt-llm