burn-onnx 0.21.0-pre.3

Library for importing ONNX models into the Burn framework
Documentation
use super::prelude::*;
use burn_store::TensorSnapshot;

impl NodeCodegen for onnx_ir::conv1d::Conv1dNode {
    fn inputs(&self) -> &[Argument] {
        // Filter inputs only dynamic and constant
        &self.inputs
    }

    fn outputs(&self) -> &[Argument] {
        &self.outputs
    }

    fn field(&self) -> Option<Field> {
        let name = Ident::new(&self.name, Span::call_site());
        let weight_shape = self.inputs[1]
            .ty
            .static_shape_known()
            .expect("Conv1d: weight tensor shape must be known at codegen time");
        let groups = self.config.groups;
        let channels_in = (weight_shape[1] * groups).to_tokens();
        let channels_out = weight_shape[0].to_tokens();
        let kernel_size = self.config.kernel_size.to_tokens();
        let stride = self.config.stride.to_tokens();
        let dilation = self.config.dilation.to_tokens();
        let groups = groups.to_tokens();
        let bias = self.inputs.len() == 3;

        let shape = self.inputs[0].ty.static_shape_known();
        let input_spatial = shape.as_deref().map(|s| &s[2..]);
        let padding = crate::burn::codegen::resolve_auto_pad_1d(
            &self.config.auto_pad,
            &self.config.padding,
            input_spatial,
            self.config.kernel_size,
            self.config.stride,
            self.config.dilation,
        )
        .to_tokens();

        Some(Field::new(
            self.name.clone(),
            quote! {
                Conv1d<B>
            },
            quote! {
                let #name = Conv1dConfig::new(#channels_in, #channels_out, #kernel_size)
                    .with_stride(#stride)
                    .with_padding(#padding)
                    .with_dilation(#dilation)
                    .with_groups(#groups)
                    .with_bias(#bias)
                    .init(device);
            },
        ))
    }

    fn forward(&self, scope: &mut ScopeAtPosition<'_>) -> TokenStream {
        let input = scope.arg(self.inputs.first().unwrap());
        let output = arg_to_ident(self.outputs.first().unwrap());
        let field = Ident::new(&self.name, Span::call_site());

        quote! {
            let #output = self.#field.forward(#input);
        }
    }

    fn register_imports(&self, imports: &mut BurnImports) {
        imports.register("burn::nn::PaddingConfig1d");
        imports.register("burn::nn::conv::Conv1d");
        imports.register("burn::nn::conv::Conv1dConfig");
    }

    fn collect_snapshots(&self, field_name: &str) -> Vec<TensorSnapshot> {
        use crate::burn::node_traits::create_lazy_snapshot;
        let mut snapshots = vec![];

        // Weight tensor (input index 1)
        if let Some(weight_input) = self.inputs.get(1) {
            let weight_path = format!("{}.weight", field_name);
            if let Some(snapshot) = create_lazy_snapshot(weight_input, &weight_path, "Conv1d") {
                snapshots.push(snapshot);
            }
        }

        // Bias tensor if present (input index 2)
        if let Some(bias_input) = self.inputs.get(2) {
            let bias_path = format!("{}.bias", field_name);
            if let Some(snapshot) = create_lazy_snapshot(bias_input, &bias_path, "Conv1d") {
                snapshots.push(snapshot);
            }
        }

        snapshots
    }
}

#[cfg(test)]
mod tests {
    use super::super::test_helpers::*;
    use burn::tensor::DType;
    use insta::assert_snapshot;
    use onnx_ir::conv1d::{Conv1dConfig, Conv1dNode, Conv1dNodeBuilder};
    use onnx_ir::padding::{AutoPad, PaddingConfig1d};

    fn create_conv1d_node(name: &str) -> Conv1dNode {
        let config =
            Conv1dConfig::new(3, 1, 1, 1, PaddingConfig1d::Explicit(1, 1), AutoPad::NotSet);

        Conv1dNodeBuilder::new(name)
            .input_tensor("input", 3, DType::F32)
            .input_static_tensor_shape("weight", vec![64, 3, 3], DType::F32)
            .input_static_tensor_shape("bias", vec![64], DType::F32)
            .output_tensor("output", 3, DType::F32)
            .config(config)
            .build()
    }

    fn create_conv1d_node_asymmetric(name: &str) -> Conv1dNode {
        // Asymmetric padding: left=1, right=2
        let config =
            Conv1dConfig::new(3, 1, 1, 1, PaddingConfig1d::Explicit(1, 2), AutoPad::NotSet);

        Conv1dNodeBuilder::new(name)
            .input_tensor("input", 3, DType::F32)
            .input_static_tensor_shape("weight", vec![64, 3, 3], DType::F32)
            .input_static_tensor_shape("bias", vec![64], DType::F32)
            .output_tensor("output", 3, DType::F32)
            .config(config)
            .build()
    }

    #[test]
    fn test_conv1d_forward() {
        let node = create_conv1d_node("conv1");
        let code = codegen_forward_default(&node);
        assert_snapshot!(code, @r"
        pub fn forward(&self, input: Tensor<B, 3>) -> Tensor<B, 3> {
            let output = self.conv1.forward(input);
            output
        }
        ");
    }

    #[test]
    fn test_conv1d_forward_with_clone() {
        let node = create_conv1d_node("conv1");
        let code = codegen_forward_with_clone(&node);
        assert_snapshot!(code, @r"
        pub fn forward(&self, input: Tensor<B, 3>) -> Tensor<B, 3> {
            let output = self.conv1.forward(input.clone());
            output
        }
        ");
    }

    #[test]
    fn test_conv1d_field_init_auto_pad_same_upper() {
        let config = Conv1dConfig::new(3, 1, 1, 1, PaddingConfig1d::Valid, AutoPad::SameUpper);
        let node = Conv1dNodeBuilder::new("conv1")
            .input_tensor_shape("input", vec![1, 3, 7], DType::F32)
            .input_static_tensor_shape("weight", vec![64, 3, 3], DType::F32)
            .input_static_tensor_shape("bias", vec![64], DType::F32)
            .output_tensor("output", 3, DType::F32)
            .config(config)
            .build();
        let code = codegen_field_init(&node);
        assert_snapshot!(code, @r"
        let conv1 = Conv1dConfig::new(3, 64, 3)
            .with_stride(1)
            .with_padding(PaddingConfig1d::Explicit(1, 1))
            .with_dilation(1)
            .with_groups(1)
            .with_bias(true)
            .init(device);
        ");
    }

    #[test]
    fn test_conv1d_field_init_asymmetric_padding() {
        let node = create_conv1d_node_asymmetric("conv1");
        let code = codegen_field_init(&node);
        // Asymmetric padding is passed directly to the module
        assert_snapshot!(code, @r"
        let conv1 = Conv1dConfig::new(3, 64, 3)
            .with_stride(1)
            .with_padding(PaddingConfig1d::Explicit(1, 2))
            .with_dilation(1)
            .with_groups(1)
            .with_bias(true)
            .init(device);
        ");
    }
}