PaddlePaddle · co63oc · Aug 10, 2026
diff --git a/swe-paddle/tasks/PaddlePaddle__Paddle-73570/README.md b/swe-paddle/tasks/PaddlePaddle__Paddle-73570/README.md
@@ -0,0 +1,51 @@
+# PaddlePaddle__Paddle-73570
+
+This directory converts Paddle PR #73570 into a SWE-Paddle community task candidate.
+
+## Source
+
+| Field | Value |
+| --- | --- |
+| Repo | `PaddlePaddle/Paddle` |
+| PR | [73570](https://github.com/PaddlePaddle/Paddle/pull/73570) |
+| PR title | `[0-size Tensor Job2 No.87] Add 0-size Tensor support for masked_fill` |
+| Base commit | `3efb8dbb51547f0235a402135c54ed83c2f12d61` |
+| Gold commit | `70574f3ff130128d7cfed5a7bc50f2842137cc98` |
+| Merged at | `2025-07-01` |
+| Task type | `bug_fix` |
+| Resource | CPU |
+| Scope | C++ Operator Kernel |
+
+## Summary
+
+Fix `paddle.masked_fill` and `paddle.diag` to correctly handle 0-size tensors in CPU/GPU/XPU kernels by adding early-return logic and fixing gradient shape handling.
+
+## Why This Is A Good SWE-Paddle Candidate
+
+- It is derived from a merged Paddle bug-fix PR rather than a synthetic issue.
+- The target behavior is isolated to the C++ operator kernel level and requires rebuilding Paddle from source.
+- The failure is deterministic: the base revision fails when processing 0-size tensors due to missing early-return logic and incorrect gradient shape handling.
+- The task has clear regression coverage for existing non-zero-size behavior.
+- The task runs on CPU and does not require distributed execution, external services, or additional datasets.
+
+## Files
+
+- `proposal.md`: candidate proposal for maintainer triage.
+- `instruction.md`: self-contained problem statement for the coding agent.
+- `solution/code.patch`: gold implementation patch (C++ kernel changes).
+- `tests/test.patch`: tests exposing the target behavior.
+- `tests/test.sh`: minimal target test command.
+- `environment/README.md`: environment and reproduction notes.
+
+## Verification
+
+```bash
+bash tests/test.sh
+```
+
+Expected behavior:
+
+| Revision state | Existing behavior (P2P) | masked_fill/diag F2P |
+| --- | ---: | ---: |
+| Base + `tests/test.patch` | PASS | FAIL |
+| Base + `tests/test.patch` + `solution/code.patch` | PASS | PASS |
diff --git a/swe-paddle/tasks/PaddlePaddle__Paddle-73570/environment/README.md b/swe-paddle/tasks/PaddlePaddle__Paddle-73570/environment/README.md
@@ -0,0 +1,52 @@
+# Environment Notes
+
+## Expected Environment
+
+- Repository: `PaddlePaddle/Paddle`
+- Base commit: `3efb8dbb51547f0235a402135c54ed83c2f12d61`
+- Gold commit: `70574f3ff130128d7cfed5a7bc50f2842137cc98`
+- Resource: CPU
+- GPU required: no
+- Patch type: C++ kernel (CPU/GPU/XPU backends) + InferMeta + Symbolic Shape
+- Python dependencies: PaddlePaddle (source build), NumPy
+
+The verifier should execute against the Paddle source revision represented by the selected patch state. A source build is required since the patch modifies C++ kernel code, InferMeta, and symbolic shape inference.
+
+## Build Instructions
+
+1. Check out `PaddlePaddle/Paddle` at the base commit.
+2. Apply `tests/test.patch`.
+3. Build Paddle from source (CPU-only build is sufficient):
+   ```bash
+   mkdir build && cd build
+   cmake .. -DWITH_GPU=OFF -DWITH_TESTING=ON -DCMAKE_BUILD_TYPE=Release
+   make -j$(nproc)
+   ```
+4. Install the built Paddle package.
+
+## Run Order
+
+1. Check out `PaddlePaddle/Paddle` at the base commit.
+2. Build and install Paddle from source.
+3. Apply `tests/test.patch`.
+4. Run the P2P tests; existing non-zero-size behavior should pass.
+5. Run the 0-size tensor tests; the target case should fail before the fix.
+6. Apply `solution/code.patch`.
+7. Rebuild Paddle from source.
+8. Reinstall Paddle package.
+9. Run `bash tests/test.sh`; all target tests should pass.
+
+## Minimal Test Command
+
+```bash
+bash tests/test.sh
+```
+
+## Expected Matrix
+
+| Revision state | P2P | masked_fill/diag F2P |
+| --- | ---: | ---: |
+| Base + test patch | PASS | FAIL |
+| Base + test patch + solution patch | PASS | PASS |
+
+No GPU, distributed runtime, external service, or additional dataset is required.
diff --git a/swe-paddle/tasks/PaddlePaddle__Paddle-73570/instruction.md b/swe-paddle/tasks/PaddlePaddle__Paddle-73570/instruction.md
@@ -0,0 +1,51 @@
+# 修复 `paddle.masked_fill` 和 `paddle.diag` 对 0-size Tensor 的处理
+
+## 详细描述
+
+当 `paddle.masked_fill(x, mask, value)` 或 `paddle.diag(x)` 的输入为 0-size Tensor 时，当前 CPU/GPU/XPU kernel 实现会直接进入后续计算逻辑，导致 kernel 内部对空数据执行计算或产生其他错误。
+
+典型表现包括：
+
+- kernel 在执行填充或对角线提取时崩溃或报错
+- 0-size Tensor 输入无法通过 `masked_fill` 或 `diag` 算子
+- 梯度计算时形状处理不正确
+
+例如：
+
+```python
+import numpy as np
+import paddle
+
+paddle.disable_static()
+
+# masked_fill 0-size tensor 输入
+x = paddle.to_tensor(np.random.rand(0, 3).astype('float32'))
+mask = paddle.to_tensor(np.random.randint(0, 2, (0, 3)).astype('bool'))
+value = paddle.to_tensor(np.array([1.0]).astype('float32'))
+out = paddle.masked_fill(x, mask, value)
+# 期望返回 shape 为 (0, 3) 的空 Tensor
+
+# diag 0-size tensor 输入
+x = paddle.to_tensor(np.random.rand(10, 0).astype('float64'))
+out = paddle.diag(x, offset=1)
+# 期望返回正确 shape 的空 Tensor
+```
+
+上述调用中输入包含 0-size 维度。按照 API 语义，0-size Tensor 的操作应正常返回正确 shape 的空 Tensor。
+
+需要在 CPU/GPU/XPU kernel 层添加 0-size 早期返回处理，并修复梯度 kernel 中的形状处理逻辑。
+
+## 验收说明
+
+- 当输入为 0-size Tensor 时，`paddle.masked_fill` 和 `paddle.diag` 应正常完成，返回正确 shape 的空 Tensor
+- 输出的 shape 应与输入一致
+- 非 0-size Tensor 输入下的 masked_fill/diag 行为不得退化
+- 梯度计算也应正常工作（0-size Tensor 的梯度也为空 Tensor，或保持正确形状）
+
+## 技术要求
+
+- 熟悉 C++ 和 Paddle PHI kernel 开发
+- 了解 Tensor shape、0-size Tensor 和 kernel 执行路径
+- 了解 masked_fill 和 diag 算子的输入输出语义
+- 了解 Paddle CPU/GPU/XPU kernel 的实现模式
+- 需要从源码编译 Paddle 以验证修改
diff --git a/swe-paddle/tasks/PaddlePaddle__Paddle-73570/proposal.md b/swe-paddle/tasks/PaddlePaddle__Paddle-73570/proposal.md
@@ -0,0 +1,59 @@
+# SWE-Paddle Task Proposal: PaddlePaddle__Paddle-73570
+
+## 1. 来源信息
+
+- Instance ID: `PaddlePaddle__Paddle-73570`
+- PR 链接: https://github.com/PaddlePaddle/Paddle/pull/73570
+- PR 标题: `[0-size Tensor Job2 No.87] Add 0-size Tensor support for masked_fill`
+- Base commit: `3efb8dbb51547f0235a402135c54ed83c2f12d61`
+- Gold commit: `70574f3ff130128d7cfed5a7bc50f2842137cc98`
+- Merged at: 2025-07-01
+- 你的身份: contributor
+
+## 2. 问题一句话
+
+`paddle.masked_fill` 和 `paddle.diag` 在输入为 0-size Tensor 时，CPU/GPU/XPU kernel 未处理 0-size 边界情况导致崩溃或报错，需要在 kernel 入口添加 0-size 早期返回逻辑，并修复梯度 kernel 中的形状处理。
+
+## 3. 为什么适合作为 SWE-Paddle 样本
+
+- **真实性**: 来自 Paddle「0-size Tensor 机制建设」系列任务，是真实研发需求。
+- **代表性**: 覆盖 C++ kernel 层面的 0-size Tensor 边界处理，涉及 CPU/GPU/XPU 三端 kernel、梯度 kernel 和 InferMeta。
+- **边界清楚**: 目标仅限输入为 0-size 时的 kernel 早期返回和梯度形状处理；正向非零尺寸输入不应受影响。
+- **非平凡性**: 修复需要在多个 kernel 中添加 `numel() == 0` 的早期返回，并修复梯度 kernel 中使用 `phi::Full` 填充 0 以保持正确形状，涉及对 kernel 执行流程和梯度计算的理解。
+- **回归护栏明确**: 目标 F2P 可覆盖 0-size Tensor 输入的 `masked_fill` 和 `diag` 算子测试；同文件中已有的标准测试用例可作为 P2P 护栏。
+
+## 4. 任务类型和标签
+
+- 任务类型: `bug_fix`
+- 执行后端: `cpu`
+- 设备范围: `cpu_only`
+- 模块标签: `[operator_kernel, masked_fill, diag, 0-size_tensor, cpu_kernel, gpu_kernel, xpu_kernel]`
+
+## 5. 验证思路
+
+- 目标测试命令: `bash tests/test.sh`
+- 目标测试文件:
+  - `test/legacy_test/test_diag_v2.py`（`TestDiagV2Op_ZeroSize`）
+  - `test/legacy_test/test_masked_fill.py`（`TestMaskedFillAPI_ZeroSize2`）
+- P2P 候选: 同文件中已有的 `TestDiagV2Op`、`TestMaskedFillAPI` 等标准算子测试用例。
+- 修复前预期: `base_commit` + `tests/test.patch` 后，0-size Tensor 输入的算子测试失败（kernel 崩溃或报错）。
+- 修复后预期: 继续应用 `solution/code.patch` 并重新编译后，0-size Tensor 输入正常返回空 Tensor，P2P 存量测试仍然通过。
+
+## 6. 环境与资源
+
+- 是否能提供 Docker: 无
+- Dockerfile 或镜像地址: 暂无
+- Paddle 来源: `PaddlePaddle/Paddle` source checkout at `base_commit`，需要源码编译。
+- OS / Python / CUDA / cuDNN / 其他关键依赖: Linux CPU + Python + numpy；编译需要 CMake、GCC；不要求 CUDA/cuDNN（CPU 编译即可验证）。
+- 硬件: CPU 即可（编译和测试均不需要 GPU）。
+- patch 类型: 含 C++ kernel 修改（CPU/GPU/XPU 三端）+ InferMeta + 符号推导，需要重新编译 Paddle。
+- 最小测试命令: `bash tests/test.sh`
+- 是否有 oracle 日志: 无
+
+## 7. 风险自查
+
+- 泄露风险: 正式 `instruction.md` 只描述「masked_fill/diag 对 0-size Tensor 输入的行为异常」，不指出具体 `numel() == 0` 分支逻辑或具体代码位置。
+- 环境风险: 中。任务涉及 C++ kernel 修改，需要源码编译 Paddle，编译时间较长。
+- flaky 风险: 低。测试使用固定的 0-size Tensor 构造，不依赖随机数差异或多设备同步。
+- 拆分风险: 低。该 PR 目标集中在 `masked_fill` 和 `diag` 的 CPU/GPU/XPU kernel 0-size 早期返回和梯度形状处理，测试明确指向新增的 ZeroSize 测试类，适合作为一个独立样本。
+- 其他不确定点: 完整任务包阶段应确认新增 F2P 在 `base_commit` 编译后确实失败。注意 `masked_fill` 的梯度处理需要保持正确形状（x_grad 可能非 0-size）。
diff --git a/swe-paddle/tasks/PaddlePaddle__Paddle-73570/solution/code.patch b/swe-paddle/tasks/PaddlePaddle__Paddle-73570/solution/code.patch
@@ -0,0 +1,176 @@
+diff --git a/paddle/fluid/pir/dialect/operator/interface/infer_symbolic_shape/unary_infer_sym.cc b/paddle/fluid/pir/dialect/operator/interface/infer_symbolic_shape/unary_infer_sym.cc
+index 0a3ca9505ef93..fca5f88da5e26 100644
+--- a/paddle/fluid/pir/dialect/operator/interface/infer_symbolic_shape/unary_infer_sym.cc
++++ b/paddle/fluid/pir/dialect/operator/interface/infer_symbolic_shape/unary_infer_sym.cc
+@@ -899,6 +899,9 @@ bool DiagOpInferSymbolicShape(pir::Operation *op,
+           size_ = x_shape[1].dyn_cast<int64_t>();
+         }
+       }
++      if (size_ < 0) {
++        size_ = 0;
++      }
+       infer_context->SetShapeOrDataForValue(
+           op->result(0), symbol::TensorShapeOrDataDimExprs({size_}));
+     } else {
+diff --git a/paddle/phi/infermeta/unary.cc b/paddle/phi/infermeta/unary.cc
+index 161e9b57b84d4..7fdc1bca049c3 100644
+--- a/paddle/phi/infermeta/unary.cc
++++ b/paddle/phi/infermeta/unary.cc
+@@ -955,6 +955,9 @@ void DiagInferMeta(const MetaTensor& x,
+         size_ = x_dims[1];
+       }
+     }
++    if (size_ < 0) {
++      size_ = 0;
++    }
+     out->set_dims({size_});
+     out->set_dtype(x.dtype());
+   } else {
+diff --git a/paddle/phi/kernels/cpu/diag_grad_kernel.cc b/paddle/phi/kernels/cpu/diag_grad_kernel.cc
+index b53582604b5b6..01205d2dd9117 100644
+--- a/paddle/phi/kernels/cpu/diag_grad_kernel.cc
++++ b/paddle/phi/kernels/cpu/diag_grad_kernel.cc
+@@ -28,6 +28,7 @@ void DiagGradKernel(const Context& dev_ctx,
+                     int offset,
+                     DenseTensor* x_grad) {
+   T* dx_data = dev_ctx.template Alloc<T>(x_grad);
++  if (x_grad && x_grad->numel() == 0) return;
+   const T* dout_data = out_grad.data<T>();
+   auto dx_dims = x_grad->dims();
+   auto dout_dims = out_grad.dims();
+diff --git a/paddle/phi/kernels/cpu/diag_kernel.cc b/paddle/phi/kernels/cpu/diag_kernel.cc
+index 5f21a4b26d307..675763fbe8f72 100644
+--- a/paddle/phi/kernels/cpu/diag_kernel.cc
++++ b/paddle/phi/kernels/cpu/diag_kernel.cc
+@@ -30,6 +30,7 @@ void DiagKernel(const Context& dev_ctx,
+   auto* x_data = x.data<T>();
+   auto x_dims = x.dims();
+   T* out_data = dev_ctx.template Alloc<T>(out);
++  if (out && out->numel() == 0) return;
+   auto out_dims = out->dims();
+
+   int64_t i = 0;
+diff --git a/paddle/phi/kernels/cpu/masked_fill_grad_kernel.cc b/paddle/phi/kernels/cpu/masked_fill_grad_kernel.cc
+index 5824e327ea569..73a0cb3e2a740 100644
+--- a/paddle/phi/kernels/cpu/masked_fill_grad_kernel.cc
++++ b/paddle/phi/kernels/cpu/masked_fill_grad_kernel.cc
+@@ -20,9 +20,9 @@
+ #include "paddle/phi/kernels/empty_kernel.h"
+ #include "paddle/phi/kernels/expand_grad_kernel.h"
+ #include "paddle/phi/kernels/expand_kernel.h"
++#include "paddle/phi/kernels/full_kernel.h"
+ #include "paddle/phi/kernels/funcs/common_infer_shape_functions.h"
+ #include "paddle/phi/kernels/funcs/common_shape.h"
+-
+ namespace phi {
+
+ template <typename T, typename Context>
+@@ -33,6 +33,19 @@ void MaskedFillGradKernel(const Context& dev_ctx,
+                           const DenseTensor& out_grad,
+                           DenseTensor* x_grad,
+                           DenseTensor* v_grad) {
++  if (out_grad.numel() == 0 || mask.numel() == 0) {
++    // x shape [2, 1, 3], mask shape [2, 0, 3], x_grad shape [2, 1, 3]
++    if (x_grad) {
++      phi::Full<T, Context>(
++          dev_ctx, phi::IntArray(common::vectorize(x_grad->dims())), 0, x_grad);
++    }
++    if (v_grad) {
++      phi::Full<T, Context>(
++          dev_ctx, phi::IntArray(common::vectorize(v_grad->dims())), 0, v_grad);
++    }
++    return;
++  }
++
+   auto x_grad_dims = x_grad->dims();
+   auto mask_dims = mask.dims();
+   bool expand_x = false;
+diff --git a/paddle/phi/kernels/cpu/masked_fill_kernel.cc b/paddle/phi/kernels/cpu/masked_fill_kernel.cc
+index ac3949ef4de6c..7edace7f8ebec 100644
+--- a/paddle/phi/kernels/cpu/masked_fill_kernel.cc
++++ b/paddle/phi/kernels/cpu/masked_fill_kernel.cc
+@@ -29,6 +29,11 @@ void MaskedFillKernel(const Context& dev_ctx,
+                       const DenseTensor& mask,
+                       const DenseTensor& value,
+                       DenseTensor* out) {
++  if (x.numel() == 0 || mask.numel() == 0) {
++    dev_ctx.template Alloc<T>(out);
++    return;
++  }
++
+   auto x_dims = x.dims();
+   auto mask_dims = mask.dims();
+
+diff --git a/paddle/phi/kernels/gpu/diag_grad_kernel.cu b/paddle/phi/kernels/gpu/diag_grad_kernel.cu
+index a93990993bc7a..13cd7cc08ae60 100644
+--- a/paddle/phi/kernels/gpu/diag_grad_kernel.cu
++++ b/paddle/phi/kernels/gpu/diag_grad_kernel.cu
+@@ -57,6 +57,7 @@ void DiagGradKernel(const Context& dev_ctx,
+                     int offset,
+                     DenseTensor* x_grad) {
+   T* dx_data = dev_ctx.template Alloc<T>(x_grad);
++  if (x_grad && x_grad->numel() == 0) return;
+   auto* dout_data = out_grad.data<T>();
+   auto dx_dims = x_grad->dims();
+   auto dout_dims = out_grad.dims();
+diff --git a/paddle/phi/kernels/gpu/diag_kernel.cu b/paddle/phi/kernels/gpu/diag_kernel.cu
+index 3d10b2aef24c7..036431c3ae3a3 100644
+--- a/paddle/phi/kernels/gpu/diag_kernel.cu
++++ b/paddle/phi/kernels/gpu/diag_kernel.cu
+@@ -63,6 +63,7 @@ void DiagKernel(const Context& dev_ctx,
+   auto* x_data = x.data<T>();
+   auto x_dims = x.dims();
+   T* out_data = dev_ctx.template Alloc<T>(out);
++  if (out && out->numel() == 0) return;
+   auto out_dims = out->dims();
+
+   auto GetBlockGridSize = [&dev_ctx](int64_t size) {
+diff --git a/paddle/phi/kernels/gpu/masked_fill_grad_kernel.cu b/paddle/phi/kernels/gpu/masked_fill_grad_kernel.cu
+index 60bb8b00d5b44..638032595b328 100644
+--- a/paddle/phi/kernels/gpu/masked_fill_grad_kernel.cu
++++ b/paddle/phi/kernels/gpu/masked_fill_grad_kernel.cu
+@@ -303,13 +303,14 @@ void MaskedFillGradKernel(const Context& dev_ctx,
+                           DenseTensor* x_grad,
+                           DenseTensor* v_grad) {
+   if (out_grad.numel() == 0 || mask.numel() == 0) {
+-    if (x_grad != nullptr) {
+-      x_grad->Resize({0});
+-      dev_ctx.template Alloc<T>(x_grad);
++    // x shape [2, 1, 3], mask shape [2, 0, 3], x_grad shape [2, 1, 3]
++    if (x_grad) {
++      phi::Full<T, Context>(
++          dev_ctx, phi::IntArray(common::vectorize(x_grad->dims())), 0, x_grad);
+     }
+-    if (v_grad != nullptr) {
+-      v_grad->Resize({0});
+-      dev_ctx.template Alloc<T>(v_grad);
++    if (v_grad) {
++      phi::Full<T, Context>(
++          dev_ctx, phi::IntArray(common::vectorize(v_grad->dims())), 0, v_grad);
+     }
+     return;
+   }
+diff --git a/paddle/phi/kernels/gpu/masked_fill_kernel.cu b/paddle/phi/kernels/gpu/masked_fill_kernel.cu
+index 28d5e55557bf5..0261cb316c4ab 100644
+--- a/paddle/phi/kernels/gpu/masked_fill_kernel.cu
++++ b/paddle/phi/kernels/gpu/masked_fill_kernel.cu
+@@ -209,7 +209,6 @@ void MaskedFillKernel(const Context& dev_ctx,
+                       const DenseTensor& value,
+                       DenseTensor* out) {
+   if (x.numel() == 0 || mask.numel() == 0) {
+-    out->Resize({0});
+     dev_ctx.template Alloc<T>(out);
+     return;
+   }
+diff --git a/paddle/phi/kernels/xpu/diag_kernel.cc b/paddle/phi/kernels/xpu/diag_kernel.cc
+index 8fb84f6a04aab..ad22c19bd7a2f 100644
+--- a/paddle/phi/kernels/xpu/diag_kernel.cc
++++ b/paddle/phi/kernels/xpu/diag_kernel.cc
+@@ -29,6 +29,7 @@ void DiagKernel(const Context& dev_ctx,
+   using XPUType = typename XPUTypeTrait<T>::Type;
+   auto* x_data = reinterpret_cast<const XPUType*>(x.data<T>());
+   dev_ctx.template Alloc<T>(out);
++  if (out && out->numel() == 0) return;
+   auto* out_data = reinterpret_cast<XPUType*>(out->data<T>());
+
+   auto x_shape = common::vectorize<int64_t>(x.dims());