finish ALphaDivergenceLoss #78

S1/uucoco_#78/AlphaDivergenceLoss_cuda.py

@@ -0,0 +1,91 @@
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from torch.utils.cpp_extension import load_inline
+
+cuda_source = """
+#include <torch/extension.h>
+#include <cuda_runtime.h>
+
+__global__ void alpha_divergence_kernel(
+    const float* __restrict__ p,
+    const float* __restrict__ q,
+    float* __restrict__ output,
+    int n,
+    float alpha
+) {
+    extern __shared__ float sdata[];
+    int tid = threadIdx.x;
+    int idx = blockIdx.x * blockDim.x + threadIdx.x;
+
+    float local_sum = 0.0f;
+    for (int i = idx; i < n; i += blockDim.x * gridDim.x) {
+        float p_val = p[i];
+        float q_val = q[i];
+        local_sum += powf(p_val, alpha) * powf(q_val, 1.0f - alpha);
+    }
+
+    sdata[tid] = local_sum;
+    __syncthreads();
+
+    for (int s = blockDim.x / 2; s > 0; s >>= 1) {
+        if (tid < s) {
+            sdata[tid] += sdata[tid + s];
+        }
+        __syncthreads();
+    }
+
+    if (tid == 0) {
+        atomicAdd(output, sdata[0]);
+    }
+}
+
+torch::Tensor alpha_divergence_cuda(torch::Tensor p, torch::Tensor q, float alpha) {
+    int batch_size = p.size(0);
+    int num_classes = p.size(1);
+    int n = batch_size * num_classes;
+
+    auto output = at::zeros({1}, p.options());
+
+    int threads = 256;
+    int blocks = 128;
+    int shared_mem = threads * sizeof(float);
+
+    alpha_divergence_kernel<<<blocks, threads, shared_mem>>>(
+        p.data_ptr<float>(),
+        q.data_ptr<float>(),
+        output.data_ptr<float>(),
+        n,
+        alpha
+    );
+
+    float sum_val = output.item<float>();
+    float mean_sum = sum_val / batch_size;
+    float loss = (mean_sum - 1.0f) / (alpha * (alpha - 1.0f));
+
+    return torch::tensor(loss, p.options());
+}
+"""
+
+cpp_source = """
+torch::Tensor alpha_divergence_cuda(torch::Tensor p, torch::Tensor q, float alpha);
+"""
+
+alpha_divergence_loss = load_inline(
+    name="alpha_divergence_loss",
+    cpp_sources=cpp_source,
+    cuda_sources=cuda_source,
+    functions=["alpha_divergence_cuda"],
+    verbose=False
+)
+
+
+class ModelNew(nn.Module):
+    def __init__(self, alpha=0.5):
+        super(ModelNew, self).__init__()
+        self.alpha = alpha
+
+    def forward(self, p, q):
+        p_prob = F.softmax(p, dim=1)
+        q_prob = F.softmax(q, dim=1)
+        return alpha_divergence_loss.alpha_divergence_cuda(p_prob, q_prob, self.alpha)

S1/uucoco_#78/AlphaDivergenceLoss_torch.py

@@ -0,0 +1,33 @@
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+
+class Model(nn.Module):
+    def __init__(self, alpha=0.5):
+        super(Model, self).__init__()
+        self.alpha = alpha
+
+    def forward(self, p, q):
+        p_prob = F.softmax(p, dim=1)
+        q_prob = F.softmax(q, dim=1)
+
+        term = (p_prob ** self.alpha) * (q_prob ** (1 - self.alpha))
+        sum_term = torch.sum(term, dim=1)
+        loss = (sum_term - 1) / (self.alpha * (self.alpha - 1))
+
+        return loss.mean()
+
+
+batch_size = 32
+num_classes = 1000
+
+
+def get_inputs():
+    p = torch.randn(batch_size, num_classes, requires_grad=True)
+    q = torch.randn(batch_size, num_classes)
+    return [p, q]
+
+
+def get_init_inputs():
+    return [0.5]

S1/uucoco_#78/prompt.txt

@@ -0,0 +1,101 @@
+You write custom CUDA kernels to replace the pytorch operators in the given GeGLU architecture to get speedups.
+
+You have complete freedom to choose the set of operators you want to replace. You may make the decision to replace some operators with custom CUDA kernels and leave others unchanged. You may replace multiple operators with custom implementations, consider operator fusion opportunities (combining multiple operators into a single kernel, for example, combining chunk+gelu+elementwise_mul), or algorithmic changes (such as optimized memory access patterns). You are only limited by your imagination.
+Technologies Used in This Code
+Core Libraries & Frameworks
+PyTorch: Deep learning framework
+
+CUDA: NVIDIA's parallel computing platform for GPU acceleration
+
+C++: For high-performance kernel implementation
+
+PyTorch Specific Components
+torch.nn.Module: Base class for neural network modules
+
+torch.nn.functional.F.softmax: Softmax activation function
+
+torch.utils.cpp_extension.load_inline: For inline compilation of CUDA/C++ extensions
+
+PyTorch Tensors: Multi-dimensional arrays with automatic differentiation
+
+torch.tensor(): Tensor creation from Python scalar
+
+CUDA/C++ Implementation Details
+CUDA Kernels: Custom GPU kernel (alpha_divergence_kernel)
+
+CUDA Math Functions: powf() for floating-point exponentiation
+
+Parallel Reduction: Tree-based reduction using shared memory
+
+Shared Memory: Using __shared__ for inter-thread communication
+
+Atomic Operations: atomicAdd for thread-safe global updates
+
+Grid-Stride Loops: Efficient memory access pattern
+
+Mathematical Components
+Alpha Divergence: Rényi/Alpha family of divergences between distributions
+
+Exponential Operations: Power computations with powf()
+
+Normalization: Scaling by (alpha * (alpha - 1.0f))
+
+Statistical Distance: Measures difference between probability distributions
+
+Optimization Techniques
+Shared Memory Reduction: Parallel tree reduction within thread blocks
+
+Grid-Stride Loops: Efficient handling of arbitrary tensor sizes
+
+Fused Computation: Complete divergence calculation in single kernel
+
+Batch Processing: Mean computation across batch dimension
+
+Performance Features
+Massive Parallelization: GPU acceleration for divergence computation
+
+Memory Efficiency: Shared memory for intermediate reduction results
+
+Numerical Stability: Proper handling of alpha parameter range
+
+Host-Device Coordination: CPU post-processing of GPU results
+
+
+
+
+
+
+Here's an example to show you the syntax of inline embedding custom CUDA operators in torch: The example given architecture is:
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+
+class Model(nn.Module):
+    def __init__(self, alpha=0.5):
+        super(Model, self).__init__()
+        self.alpha = alpha
+
+    def forward(self, p, q):
+        p_prob = F.softmax(p, dim=1)
+        q_prob = F.softmax(q, dim=1)
+
+        term = (p_prob ** self.alpha) * (q_prob ** (1 - self.alpha))
+        sum_term = torch.sum(term, dim=1)
+        loss = (sum_term - 1) / (self.alpha * (self.alpha - 1))
+
+        return loss.mean()
+
+
+batch_size = 32
+num_classes = 1000
+
+
+def get_inputs():
+    p = torch.randn(batch_size, num_classes, requires_grad=True)
+    q = torch.randn(batch_size, num_classes)
+    return [p, q]
+
+
+def get_init_inputs():
+    return [0.5]

S1/uucoco_#78/run_code.py

@@ -0,0 +1,77 @@
+###########################################################
+# 性能和精度验证程序
+###########################################################
+import torch
+import torch.nn as nn
+import time
+from AlphaDivergenceLoss_torch import Model, get_inputs, get_init_inputs
+from AlphaDivergenceLoss_cuda  import ModelNew
+
+
+def run_benchmark():
+    # 检查 CUDA 是否可用
+    if not torch.cuda.is_available():
+        print("CUDA 不可用，请确保您有可用的 NVIDIA GPU 并已正确安装 PyTorch CUDA 版本。")
+        return
+    else:
+        device = torch.device("cuda")
+
+    # 初始化模型
+    init_inputs = get_init_inputs()
+    init_inputs = [
+        x.cuda(device=device) if isinstance(x, torch.Tensor) else x for x in init_inputs
+    ]
+    inputs = get_inputs()
+    inputs = [
+        x.cuda(device=device) if isinstance(x, torch.Tensor) else x for x in inputs
+    ]
+
+    torch_model = Model(*init_inputs).cuda()
+    cuda_model = ModelNew(*init_inputs).cuda()
+
+    torch_model.eval()
+    cuda_model.eval()
+
+    print("-------------------- 精度对齐验证 --------------------")
+    with torch.no_grad():
+        output_torch = torch_model(*inputs)
+        output_cuda = cuda_model(*inputs)
+
+    precision_flag = torch.allclose(output_torch, output_cuda, rtol=1e-03)
+    if precision_flag:
+        print("✅ 精度对齐：两个模型的输出结果非常接近。")
+    else:
+        print("❌ 精度不一致！")
+
+    print("\n-------------------- 性能加速比测试 --------------------")
+    num_iterations = 100
+
+    # PyTorch 模型计时
+    torch.cuda.synchronize()
+    start_time = time.time()
+    for _ in range(num_iterations):
+        _ = torch_model(*inputs)
+    torch.cuda.synchronize()
+    torch_time = (time.time() - start_time) / num_iterations
+
+    # 自定义 CUDA 内核计时
+    torch.cuda.synchronize()
+    start_time = time.time()
+    for _ in range(num_iterations):
+        _ = cuda_model(*inputs)
+    torch.cuda.synchronize()
+    cuda_time = (time.time() - start_time) / num_iterations
+
+    print(f"PyTorch torch.relu 平均执行时间: {torch_time:.6f} 秒")
+    print(f"自定义 CUDA 内核 平均执行时间: {cuda_time:.6f} 秒")
+    speedup = 0
+    if cuda_time > 0:
+        speedup = torch_time / cuda_time
+        print(f"加速比 (Speedup): {speedup:.2f}x")
+    else:
+        print("CUDA 内核执行时间为0，无法计算加速比。")
+    return precision_flag, speedup
+
+
+if __name__ == "__main__":
+    precision_flag, speedup = run_benchmark()