Remove unnecessary tensor clone #154896 Sign in to view logs Summary Jobs assign Run details Usage Workflow file Triggered via issue March 2, 2025 16:16 pytorchmergebot commented on #148159 1d7397a Status Success Total duration 11s Artifacts – assigntome-docathon.yml on: issue_comment assi...
Tensors and Dynamic neural networks in Python with strong GPU acceleration - Remove unnecessary tensor clone · pytorch/pytorch@67cb36c
分析一下:在上面的 pytorch 介绍和示例演示中可以看出,select 可以通过 类似切片的操作 + 取我们想要的数据 来完成,自然在做 TensorRT 的实现的时候也可以往这个思路走。torch.select 主要由 dim 、index 两个因子来控制取数据的粒度,而TensorRT 在用 Slice 去切 Tensor 的时候,一般由 start、size、stride ...
1. update date of nightly package 20250125< Thanks for sending a pull request What type of PR is...
Tensors and Dynamic neural networks in Python with strong GPU acceleration - [BE][Ez]: Remove redundant empty tensor copies in meta-reg (#147978) · pytorch/pytorch@f423531
// if using dropout, we produce 1 random number for each element of the // attention tensor // TODO(eqy): should state be advanced per thread (local) amount or per call/launch (global) amount philox_state = gen->philox_cuda_state(batch_size * num_heads * max_seqlen_batch_q * ...
Tensors and Dynamic neural networks in Python with strong GPU acceleration - [cuDNN][SDPA] Remove `TORCH_CUDNN_SDPA_ENABLED=1`, enable cuDNN SDPA …· pytorch/pytorch@f845a7a
from_tensor(x, multipliers={0: 4, 1: 4}) 2615 - assert hasattr(pad_x, "multipliers"), breakpoint() 2616 - eager_out = f(pad_x) 2617 - 2618 - for _ in range(3): 2619 - compiled_out = compiled_f(pad_x) 2620 - compiled_out = compiled_f(pad_x) 2621 - 2622...
Summary: unwrap_tensor_subclass is incorporated in export stack natively after pytorch/pytorch#127431 so we can remove this workaround now Test Plan: python test/quantization/test_quant_api.py pyth...
remove padded tensor test #85979 Sign in to view logs Summary Jobs get-label-type manywheel-py3_9-cuda11_8-build build manywheel-py3_9-cuda12_6-build build manywheel-py3_9-cuda12_8-build build manywheel-py3_9-cuda11_8-test manywheel-py3_9-cuda12_6-test manywheel...