Skip to content
KernelIndex
Search⌘K

PyTorch eager

PyTorch · python · MIT

Use it

Vendorable · source mirrored · MITView source →

No package. Vendor the mirrored source: 30 lines, MIT.

71_Conv2d_Divide_LeakyReLU.py
curl "https://kernelindex.com/api/v1/implementations/kernelbench-l2-71-conv2d-divide-leakyrelu-torch?include=source"
interfacepython · torch_eager
symbolModel.forward
Compatibility
measured onNVIDIA H100
declared hardwaredeclared only
architectures—
dtypes

Benchmark evidence

2 measurements across 1 GPU, fastest first.

Operation / workload
Hardware
Latency
Rank
Observed
Conv2d Divide LeakyReLUfp32 · [128, 8, 128, 128]
NVIDIA H100
1.80ms±0.00
#2 of 2
2026-03-05
Conv2d Divide LeakyReLUfp32 · [128, 8, 128, 128]
NVIDIA H100
2.64ms±0.00
#2 of 2
2026-03-05

Reported · How evidence levels are derived →

Source and license

sourceavailable
revision digestsha256:a9a3e331c03ab5138d805b76df27b84e2e83097c61cd31ef53678d45e1962768
license declaredMIT
license concludedMIT
imported2026-08-26

Kernel source

71_Conv2d_Divide_LeakyReLU.py30 lines
import torch
import torch.nn as nn

class Model(nn.Module):
    """
    Simple model that performs a convolution, divides by a constant, and applies LeakyReLU.
    """
    def __init__(self, in_channels, out_channels, kernel_size, divisor):
        super(Model, self).__init__()
        self.conv = nn.Conv2d(in_channels, out_channels, kernel_size)
        self.divisor = divisor

    def forward(self, x):
        x = self.conv(x)
        x = x / self.divisor
        x = torch.nn.functional.leaky_relu(x, negative_slope=0.01)
        return x

batch_size = 128
in_channels = 8
out_channels = 64
height, width = 128, 128
kernel_size = 3
divisor = 2

def get_inputs():
    return [torch.rand(batch_size, in_channels, height, width)]

def get_init_inputs():
    return [in_channels, out_channels, kernel_size, divisor]
scrolls · 30 lines total

Source code from KernelBench, © 2023 Anne Ouyang, Simon Guo, Azalia Mirhoseini (Scaling Intelligence Lab, Stanford University), MIT License · MIT

Best evidence level for this revision: reported

JSON