[lower_to_mlir] Use named_attrs from tt-forge ops instead of hardcodi…

…ng attributes for each op, while lowering to mlir.

pybuda/csrc/passes/lower_to_mlir.cpp

@@ -45,6 +45,23 @@ using namespace tt;
 /**
  * @brief Implementation of TT-MLIR emission from the TTForge graph.
  */
+mlir::Attribute convert_to_mlir_attribute(const  tt::BudaOpAttr& value, mlir::OpBuilder& builder) {
+    return std::visit([&builder](auto&& arg) -> mlir::Attribute {
+        using T = std::decay_t<decltype(arg)>;
+        if constexpr (std::is_same_v<T, std::string>) {
+            return builder.getStringAttr(arg);
+        } else if constexpr (std::is_same_v<T, bool>) {
+            return builder.getBoolAttr(arg);
+        } else if constexpr (std::is_same_v<T, int>) {
+            return builder.getI32IntegerAttr(arg);
+        } else if constexpr (std::is_same_v<T, float>) {
+            return builder.getF32FloatAttr(arg);
+        } else {
+            // If type not handled, throw an exception or handle it appropriately
+            throw std::runtime_error("Unhandled attribute type");
+        }
+    }, value);
+}
 class MLIRGenerator
 {
     public:
@@ -204,15 +221,15 @@ class MLIRGenerator
             ::llvm::ArrayRef<::llvm::StringRef> operation_attributes = TTIROp::getAttributeNames();
             for(auto attribute_name: operation_attributes)
             {
-                if(attribute_name.equals("operand_constraints"))
+                if(attribute_name == "operand_constraints")
                 {
                     // Create operation constraint attributes
                     mlir::NamedAttribute operand_constraints_attribute = builder_.getNamedAttr(
                         "operand_constraints",
                     builder_.getArrayAttr(get_mlir_operand_constraint_attributes(graph, op_node)));
                     attributes.push_back(operand_constraints_attribute);
                 }
-                else if(attribute_name.equals(mlir::OpTrait::AttrSizedOperandSegments<void>::getOperandSegmentSizeAttr()))
+                else if(attribute_name == mlir::OpTrait::AttrSizedOperandSegments<void>::getOperandSegmentSizeAttr())
                 {
                     // Create operation segment sizes attributes
                     mlir::NamedAttribute operand_segment_sizes_attribute = builder_.getNamedAttr(
@@ -225,15 +242,13 @@ class MLIRGenerator
                 }
             }
 
-            // Workaround for now, need to figure out how to handle this properly
-            if(op_node->op_name() == "softmax")
+            for(const auto & attribute: op_node->op_type().named_attrs)
             {
-                log_info("Softmax");
-                int32_t dimension = std::get<int>(op_node->op_attrs()[0]);
-                mlir::NamedAttribute dimension_attribute = builder_.getNamedAttr(
-                        "dimension",
-                    builder_.getSI32IntegerAttr(dimension));
-                attributes.push_back(dimension_attribute);
+                // convert atribute to mlir atribute
+                auto mlir_atribute = convert_to_mlir_attribute(attribute.second, builder_);
+                mlir::NamedAttribute named_attribute = builder_.getNamedAttr(
+                                        attribute.first, mlir_atribute);
+                attributes.push_back(named_attribute);
             }
 
             auto op = builder_.create<TTIROp>(

pybuda/pybuda/op/nn.py

@@ -51,7 +51,7 @@ def Softmax(
     Tensor
         Buda tensor
     """
-    return op("softmax", name, operandA, attrs=(dim, stable)).get_tensor()
+    return op("softmax", name, operandA, attrs=(dim, stable), dimension=dim).get_tensor()
 
 
 def LogSoftmax(
@@ -82,7 +82,7 @@ def LogSoftmax(
     Tensor
         Buda tensor
     """
-    return op("log_softmax", name, operandA, attrs=(dim, stable)).get_tensor()
+    return op("log_softmax", name, operandA, attrs=(dim, stable), dimension=dim).get_tensor()
 
 def Layernorm(
         name: str,