issue/581 - 添加linear的实现,接口和测试

Author: pengcheng888

include/infinicore/ops/linear.hpp

@@ -0,0 +1,12 @@
+#pragma once
+
+#include "common/op.hpp"
+#include <pybind11/pybind11.h>
+
+namespace infinicore::op {
+
+Tensor linear(Tensor input, Tensor weight, pybind11::object bias);
+
+void linear_(Tensor out, Tensor input, Tensor weight, pybind11::object bias);
+
+} // namespace infinicore::op

python/infinicore/nn/functional.py

@@ -65,3 +65,24 @@ def swiglu(input: Tensor, other: Tensor, *, out=None):
     _infinicore.swiglu_(out._underlying, input._underlying, other._underlying)
 
     return out
+
+
+def linear(input: Tensor, weight: Tensor, bias=None, *, out=None) -> Tensor:
+    r"""Applies a linear transformation to the incoming data: y=xA^T+b."""
+
+    if out is None:
+        return Tensor(
+            _infinicore.linear(
+                input._underlying,
+                weight._underlying,
+                None if bias is None else bias._underlying,
+            )
+        )
+
+    _infinicore.linear_(
+        out._underlying,
+        input._underlying,
+        weight._underlying,
+        None if bias is None else bias._underlying,
+    )
+    return out

src/infinicore/ops/linear/linear.cc

@@ -0,0 +1,57 @@
+#include "infinicore/ops/linear.hpp"
+#include "infinicore/ops/add.hpp"
+#include "infinicore/ops/matmul.hpp"
+
+namespace infinicore::op {
+
+Tensor linear(Tensor input,
+              Tensor weight,
+              pybind11::object bias) {
+    Size ndim = input->ndim();
+    Size out_features = weight->shape()[0];
+
+    // Assign memory to out variables
+    auto output_shape = input->shape();
+    output_shape[ndim - 1] = out_features;
+    auto out = Tensor::empty(output_shape, input->dtype(), input->device());
+
+    // Inplace Calculate
+    linear_(out, input, weight, bias);
+    return out;
+}
+
+void linear_(Tensor out,
+             Tensor input,
+             Tensor weight,
+             pybind11::object bias) {
+
+    auto weight_shape = weight->shape();
+    Size out_features = weight_shape[0];
+    Size in_features = weight_shape[1];
+
+    Size ndim = input->ndim();
+    assert(out->ndim() == ndim);
+
+    // Calculate the number of features
+    Size N = 1;
+    auto input_shape = input->shape();
+    for (size_t i = 0; i < ndim - 1; ++i) {
+        N *= input_shape[i];
+    }
+
+    // linear transformation
+    Tensor out_view = out->view({N, out_features});
+    matmul_(out_view,
+            input->view({N, in_features}),
+            weight->permute({1, 0}));
+
+    // Add bias
+    if (!bias.is_none()) {
+        Tensor bias_tensor = bias.cast<Tensor>();
+        add_(out_view,
+             out_view,
+             bias_tensor->as_strided({N, out_features}, {0, 1}));
+    }
+}
+
+} // namespace infinicore::op

src/infinicore/pybind11/ops.hpp

@@ -5,6 +5,7 @@
 #include "ops/add.hpp"
 #include "ops/attention.hpp"
 #include "ops/causal_softmax.hpp"
+#include "ops/linear.hpp"
 #include "ops/matmul.hpp"
 #include "ops/rearrange.hpp"
 #include "ops/rms_norm.hpp"
@@ -19,6 +20,7 @@ inline void bind(py::module &m) {
     bind_add(m);
     bind_attention(m);
     bind_causal_softmax(m);
+    bind_linear(m);
     bind_matmul(m);
     bind_rearrange(m);
     bind_rms_norm(m);

src/infinicore/pybind11/ops/linear.hpp

@@ -0,0 +1,27 @@
+#pragma once
+
+#include "infinicore/ops/linear.hpp"
+#include <pybind11/pybind11.h>
+
+namespace py = pybind11;
+
+namespace infinicore::ops {
+
+inline void bind_linear(py::module &m) {
+    m.def("linear",
+          &op::linear,
+          py::arg("input"),
+          py::arg("weight"),
+          py::arg("bias") = py::none(),
+          R"doc(Applies a linear transformation to the incoming data: y=xA^T+b.)doc");
+
+    m.def("linear_",
+          &op::linear_,
+          py::arg("out"),
+          py::arg("input"),
+          py::arg("weight"),
+          py::arg("bias") = py::none(),
+          R"doc(In-place, applies a linear transformation to the incoming data: y=xA^T+b.)doc");
+}
+
+} // namespace infinicore::ops

test/infinicore/ops/linear.py

@@ -0,0 +1,140 @@
+import sys
+import os
+
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), ".."))
+
+import torch
+import infinicore
+from framework.base import BaseOperatorTest, TensorSpec, TestCase
+from framework.runner import GenericTestRunner
+from framework.utils import is_broadcast
+
+# ==============================================================================
+# Operator-specific configuration
+# ==============================================================================
+_TEST_CASES_DATA = [
+    # bs, n, in_features, out_features, bias
+    (1, 5, 2048, 5632, True, None, None, None),
+    (1, 1, 2048, 32000, False, None, None, None),
+    (2, 5, 2048, 5632, True, None, None, None),
+    (2, 5, 256, 2048, False, None, None, None),
+    (None, 5, 256, 2048, False, None, None, None),
+    (None, 1, 2048, 5632, True, None, None, None),
+]
+
+# Tolerance configuration
+_TOLERANCE_MAP = {
+    infinicore.float16: {"atol": 0, "rtol": 1e-2},
+    infinicore.float32: {"atol": 0, "rtol": 1e-3},
+    infinicore.bfloat16: {"atol": 0, "rtol": 5e-2},
+}
+
+# Data types to test
+_TENSOR_DTYPES = [infinicore.float16, infinicore.bfloat16, infinicore.float32]
+
+
+def parse_test_cases():
+    """
+    Parse test case data and return list of TestCase objects for linear operation.
+    Each test case contains all necessary information for execution and validation.
+    """
+    test_cases = []
+
+    for data in _TEST_CASES_DATA:
+        bs = data[0]
+        n, in_features, out_features = data[1], data[2], data[3]
+        bias = data[4]
+        input_strides = data[5] if len(data) > 5 else None
+        weight_strides = data[6] if len(data) > 6 else None
+        out_strides = data[7] if len(data) > 7 else None
+
+        # Determine shapes based on batch dimension
+        if bs is None:
+            input_shape = (n, in_features)
+            weight_shape = (out_features, in_features)
+            out_shape = (n, out_features)
+        else:
+            input_shape = (bs, n, in_features)
+            weight_shape = (out_features, in_features)
+            out_shape = (bs, n, out_features)
+
+        if bias is True:
+            bias_shape = (out_features,)
+        else:
+            bias_shape = None
+
+        # Check if tensors support in-place operations
+        c_supports_inplace = not is_broadcast(out_shape)
+
+        # Generate test cases for all data types
+        for dtype in _TENSOR_DTYPES:
+            tolerance = _TOLERANCE_MAP.get(dtype, {"atol": 0, "rtol": 1e-3})
+
+            # Create typed tensor specs
+            input_spec = TensorSpec.from_tensor(input_shape, input_strides, dtype)
+            weight_spec = TensorSpec.from_tensor(weight_shape, weight_strides, dtype)
+            out_spec = TensorSpec.from_tensor(out_shape, out_strides, dtype)
+
+            if bias_shape is not None:
+                bias_spec = TensorSpec.from_tensor(bias_shape, None, dtype)
+            else:
+                bias_spec = None
+
+            # Test Case 1: Out-of-place (return value)
+            test_cases.append(
+                TestCase(
+                    inputs=[input_spec, weight_spec, bias_spec],
+                    kwargs={},
+                    output_spec=None,
+                    comparison_target=None,
+                    tolerance=tolerance,
+                    description=f"Linear - OUT_OF_PLACE",
+                )
+            )
+
+            # Test Case 2: In-place with explicit output tensor (Linear(a, b, out=c))
+            if c_supports_inplace:
+                test_cases.append(
+                    TestCase(
+                        inputs=[input_spec, weight_spec, bias_spec],
+                        kwargs=None,
+                        output_spec=out_spec,  # Specify the output tensor spec
+                        comparison_target="out",
+                        tolerance=tolerance,
+                        description=f"Linear - INPLACE(out)",
+                    )
+                )
+
+    return test_cases
+
+
+class OpTest(BaseOperatorTest):
+    """Linear operator test with simplified implementation"""
+
+    def __init__(self):
+        super().__init__("Linear")
+
+    def get_test_cases(self):
+        return parse_test_cases()
+
+    def torch_operator(self, input, weight, bias, out=None, **kwargs):
+        """PyTorch linear implementation"""
+        result = torch.nn.functional.linear(input, weight, bias)
+        if out is not None:
+            out.copy_(result)
+            return out
+        return result
+
+    def infinicore_operator(self, input, weight, bias, out=None, **kwargs):
+        """InfiniCore linear implementation"""
+        return infinicore.nn.functional.linear(input, weight, bias, out=out)
+
+
+def main():
+    """Main entry point"""
+    runner = GenericTestRunner(OpTest)
+    runner.run_and_exit()
+
+
+if __name__ == "__main__":
+    main()