TensorRT
TensorRT copied to clipboard
PyTorch/TorchScript/FX compiler for NVIDIA GPUs using TensorRT
# Description - In TRT 10, the version string registers as less than `"7.0"`, causing an error for certain test cases ## Type of change - Bug fix (non-breaking change...
# Description New feature to support aten.as_strided converter converter. Our implementation focuses on accurately calculating the indices required Fixes # ([issue](https://github.com/pytorch/TensorRT/issues/2734)) ## Example Usage Given a tensor `x`: ```python x...
Bumps [transformers](https://github.com/huggingface/transformers) from 4.36.0 to 4.38.0. Release notes Sourced from transformers's releases. v4.38: Gemma, Depth Anything, Stable LM; Static Cache, HF Quantizer, AQLM New model additions 💎 Gemma 💎 Gemma...
# Description Update the aten::layer_norm converter to use INormalizationLayer. Resolves warning about precision: `WARNING: [Torch-TensorRT TorchScript Conversion Context] - Running layernorm after self-attention in FP16 may cause overflow. Exporting the...
## Bug Description FAILED runtime/test_python_runtime.py::TestLowRankInputs::test_1D_input - AssertionError: 61.15180206298828 != 0 within 7 places (61.15180206298828 difference) : 1D-Tensor TRT outputs don't match with the original model. ## To Reproduce Steps to...
## Bug Description When I used torch-tensorrt to complite, some errors came out suddenly, which made me confuse. ## To Reproduce Code for method 1: ``` import torch import tensorrt...
Implement `aten.empty_strided` converter https://pytorch.org/docs/stable/generated/torch.empty_strided.html#torch-empty-strided
aten op converter for `sym_storage_offset` https://github.com/pytorch/pytorch/blob/d0211e207c78fafac2edaf2e14954f668e898b4a/aten/src/ATen/TensorGeometry.h#L88 https://github.com/pytorch/pytorch/blob/d0211e207c78fafac2edaf2e14954f668e898b4a/aten/src/ATen/TensorGeometry.h#L139
# Context For models requiring fallback to Torch due to converter capabilities, custom operators, or other needs, each of the `TRTEngine` objects is compiled, initialized, inserted into the Torch `nn.Module`,...