CPU: Add support for FLOOR,CEIL,ROUND and TRUNC unary operators

docs/ops.md

@@ -110,6 +110,10 @@ Legend:
 |                       SWIGLU_OAI | ❌ | ❌ | ❌ | ❌ | ❌ | ❌ | ❌ | ❌ | ❌ |
 |                             TANH | ❌ | ✅ | ✅ | 🟡 | 🟡 | ✅ | 🟡 | 🟡 | ❌ |
 |               TIMESTEP_EMBEDDING | ❌ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ❌ |
+
 |                            TRUNC | ❌ | ❌ | ✅ | ❌ | ❌ | ❌ | ❌ | ❌ | ❌ |
+
+|                         TOPK_MOE | ❌ | ❌ | ❌ | ❌ | ❌ | ❌ | ✅ | ❌ | ❌ |
+
 |                          UPSCALE | ❌ | 🟡 | ✅ | ✅ | 🟡 | ✅ | 🟡 | ✅ | ❌ |
 |                            XIELU | ❌ | ❌ | ❌ | ❌ | ❌ | ❌ | ❌ | ❌ | ❌ |

ggml/include/ggml.h

@@ -580,6 +580,7 @@ extern "C" {
         GGML_UNARY_OP_CEIL,
         GGML_UNARY_OP_ROUND,
         GGML_UNARY_OP_TRUNC,
+        GGML_UNARY_OP_XIELU,
 
 
         GGML_UNARY_OP_COUNT,
@@ -1155,6 +1156,7 @@ extern "C" {
             struct ggml_context * ctx,
             struct ggml_tensor  * a);
 
+<<<<<<< HEAD
     GGML_API struct ggml_tensor * ggml_floor(
             struct ggml_context * ctx,
             struct ggml_tensor  * a);
@@ -1195,6 +1197,18 @@ extern "C" {
 
 
 
+    // xIELU activation function
+    // x = x * (c_a(alpha_n) + c_b(alpha_p, beta) * sigmoid(beta * x)) + eps * (x > 0)
+    // where c_a = softplus and c_b(a, b) = softplus(a) + b are constraining functions
+    // that constrain the positive and negative source alpha values respectively
+    GGML_API struct ggml_tensor * ggml_xielu(
+            struct ggml_context * ctx,
+            struct ggml_tensor  * a,
+            float alpha_n,
+            float alpha_p,
+            float beta,
+            float eps);
+
     // gated linear unit ops
     // A: n columns, r rows,
     // result is n / 2 columns, r rows,

ggml/src/ggml-cpu/ops.cpp

@@ -8993,6 +8993,7 @@ void ggml_compute_forward_unary(
             {
                 ggml_compute_forward_exp(params, dst);
             } break;
+<<<<<<< HEAD
         case GGML_UNARY_OP_FLOOR:
             {
                 ggml_compute_forward_floor(params, dst);
@@ -9008,6 +9009,11 @@ void ggml_compute_forward_unary(
         case GGML_UNARY_OP_TRUNC:
             {
                 ggml_compute_forward_trunc(params, dst);
+
+        case GGML_UNARY_OP_XIELU:
+            {
+                ggml_compute_forward_xielu(params, dst);
+
             } break;
         default:
             {

ggml/src/ggml-cpu/unary-ops.cpp

@@ -305,3 +305,17 @@ void ggml_compute_forward_round(const ggml_compute_params * params, ggml_tensor
 void ggml_compute_forward_trunc(const ggml_compute_params * params, ggml_tensor * dst) {
     unary_op<op_trunc>(params, dst);
 }
+
+void ggml_compute_forward_xielu(const ggml_compute_params * params, ggml_tensor * dst) {
+    const float alpha_n = ggml_get_op_params_f32(dst, 1);
+    const float alpha_p = ggml_get_op_params_f32(dst, 2);
+    const float beta = ggml_get_op_params_f32(dst, 3);
+    const float eps = ggml_get_op_params_f32(dst, 4);
+
+    const auto xielu_op_params = [alpha_n, alpha_p, beta, eps](float f) {
+        return op_xielu(f, alpha_n, alpha_p, beta, eps);
+    };
+
+    unary_op_functor(params, dst, xielu_op_params);
+}
+

ggml/src/ggml-cpu/unary-ops.h

@@ -26,6 +26,8 @@ void ggml_compute_forward_floor(const struct ggml_compute_params * params, struc
 void ggml_compute_forward_ceil(const struct ggml_compute_params * params, struct ggml_tensor * dst);
 void ggml_compute_forward_round(const struct ggml_compute_params * params, struct ggml_tensor * dst);
 void ggml_compute_forward_trunc(const struct ggml_compute_params * params, struct ggml_tensor * dst);
+void ggml_compute_forward_xielu(const struct ggml_compute_params * params, struct ggml_tensor * dst);
+
 
 
 

ggml/src/ggml.c

@@ -1143,14 +1143,14 @@ static const char * GGML_UNARY_OP_NAME[GGML_UNARY_OP_COUNT] = {
     "HARDSIGMOID",
     "EXP",
     "GELU_ERF",
+    "XIELU",
     "FLOOR",
     "CEIL",
     "ROUND",
     "TRUNC",
 };
 
-static_assert(GGML_UNARY_OP_COUNT == 19, "GGML_UNARY_OP_COUNT != 19");
-
+static_assert(GGML_UNARY_OP_COUNT == 20, "GGML_UNARY_OP_COUNT != 20");
 
 static const char * GGML_GLU_OP_NAME[GGML_GLU_OP_COUNT] = {
     "REGLU",