policy_gradient_loss.hpp

Go to the documentation of this file.

 
#pragma once
 
#include <cstdint>
#include <vector>
 
#include "pulsatrix/device_backend.hpp"
#include "pulsatrix/tensor.hpp"
 
namespace pulsatrix {
 
class PolicyGradientLoss {
public:
    explicit PolicyGradientLoss(DeviceBackend* backend);
 
    [[nodiscard]] float forward(const Tensor& logits, const Tensor& actions, const Tensor& returns);
 
    [[nodiscard]] Tensor backward() const;
 
private:
    DeviceBackend* backend_;
    Tensor last_probs_;
    Tensor last_returns_;
    Tensor last_action_indices_;
    bool has_forwarded_ = false;
};
 
}  // namespace pulsatrix