/* Copyright 2025 The xLLM Authors. All Rights Reserved.

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

    https://github.com/jd-opensource/xllm/blob/main/LICENSE

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
==============================================================================*/

#pragma once

#include <torch/torch.h>

#include "dense_mlp.h"
#include "framework/model/model_args.h"
#include "framework/model/model_input_params.h"
#include "framework/parallel_state/parallel_args.h"
#include "framework/quant_args.h"
#include "framework/state_dict/state_dict.h"
#include "framework/state_dict/utils.h"
#include "fused_moe_base.h"
#include "linear.h"

namespace xllm {
namespace layer {

// FusedMoE common implementation - placeholder for unsupported backends
// Actual implementations are in backend-specific fused_moe.h files.
class FusedMoEImpl : public torch::nn::Module {
 public:
  FusedMoEImpl() = default;
  FusedMoEImpl(const ModelArgs& model_args,
               const FusedMoEArgs& moe_args,
               const QuantArgs& quant_args,
               const ParallelArgs& parallel_args,
               const torch::TensorOptions& options);

  torch::Tensor forward_experts(const torch::Tensor& hidden_states,
                                const torch::Tensor& router_logits,
                                bool enable_all2all_communication);
  torch::Tensor forward(const torch::Tensor& hidden_states,
                        const ModelInputParams& input_params);
  void load_state_dict(const StateDict& state_dict);
};
TORCH_MODULE(FusedMoE);

}  // namespace layer
}  // namespace xllm