How to read the parameter info

compiler
1

Hi All,

I’m trying to learn how to read the parameters (weights, bias, etc.,). In the below code, the convolution node has the primals_1 and primals_2 as weight and bias

class GraphModule(torch.nn.Module):
def forward(self, primals_1: f32[64, 3, 3, 3], primals_2: f32[64], primals_3: f32[1, 3, 224, 224]):
# File: sample_conv.py:12, code: x = self.conv(x)
convolution: f32[1, 64, 222, 222] = torch.ops.aten.convolution.default(primals_3, primals_1, primals_2, [1, 1], [0, 0], [1, 1], False, [0, 0], 1); primals_2 = None

    # File: sample_conv.py:13, code: x = self.relu(x)
    relu: f32[1, 64, 222, 222] = torch.ops.aten.relu.default(convolution);  convolution = None
    return [relu, primals_3, primals_1, relu

Can someone give me pointers on how to read the parameters data?

2

Not pretty sure, but see the api aten.convolution here