axonml_nn/

loss.rs

//! Loss functions for training neural networks.
//!
//! 1436 lines. Each loss struct has `new(reduction)` and `compute(input,
//! target) -> Variable` (graph-tracked, backpropagable). `MSELoss`,
//! `L1Loss`, `CrossEntropyLoss` (log-softmax + NLL, class weights +
//! label smoothing), `NLLLoss`, `BCELoss`, `BCEWithLogitsLoss` (fused
//! sigmoid + BCE for numerical stability), `SmoothL1Loss` (Huber).
//! Reduction modes: Mean (default), Sum, None. All losses support
//! batched inputs.
//!
//! # File
//! `crates/axonml-nn/src/loss.rs`
//!
//! # Author
//! Andrew Jewell Sr. — AutomataNexus LLC
//! ORCID: 0009-0005-2158-7060
//!
//! # Updated
//! April 14, 2026 11:15 PM EST
//!
//! # Disclaimer
//! Use at own risk. This software is provided "as is", without warranty of any
//! kind, express or implied. The author and AutomataNexus shall not be held
//! liable for any damages arising from the use of this software.

use std::any::Any;

use axonml_autograd::no_grad::is_grad_enabled;
use axonml_autograd::{GradFn, GradientFunction, Variable};
use axonml_tensor::Tensor;

use crate::module::Module;

// =============================================================================
// Reduction Enum
// =============================================================================

/// Specifies how to reduce the loss over elements.
#[derive(Debug, Clone, Copy, PartialEq, Default)]
pub enum Reduction {
    /// No reduction - return loss per element.
    None,
    /// Mean of all losses.
    #[default]
    Mean,
    /// Sum of all losses.
    Sum,
}

// =============================================================================
// MSELoss
// =============================================================================

/// Mean Squared Error loss.
///
/// loss = mean((input - target)^2)
#[derive(Debug, Clone, Copy)]
pub struct MSELoss {
    reduction: Reduction,
}

impl MSELoss {
    /// Creates a new MSELoss with default reduction (Mean).
    pub fn new() -> Self {
        Self {
            reduction: Reduction::Mean,
        }
    }

    /// Creates MSELoss with specified reduction.
    pub fn with_reduction(reduction: Reduction) -> Self {
        Self { reduction }
    }

    /// Computes the loss.
    pub fn compute(&self, input: &Variable, target: &Variable) -> Variable {
        let diff = input.sub_var(target);
        let squared = diff.pow(2.0);

        match self.reduction {
            Reduction::None => squared,
            Reduction::Mean => squared.mean(),
            Reduction::Sum => squared.sum(),
        }
    }
}

impl Default for MSELoss {
    fn default() -> Self {
        Self::new()
    }
}

impl Module for MSELoss {
    fn forward(&self, input: &Variable) -> Variable {
        // For Module interface, we can't easily pass two inputs
        // This is primarily used via compute() method
        input.clone()
    }

    fn name(&self) -> &'static str {
        "MSELoss"
    }
}

// =============================================================================
// L1Loss
// =============================================================================

/// Mean Absolute Error loss.
///
/// loss = mean(|input - target|)
#[derive(Debug, Clone, Copy)]
pub struct L1Loss {
    reduction: Reduction,
}

impl L1Loss {
    /// Creates a new L1Loss with default reduction (Mean).
    pub fn new() -> Self {
        Self {
            reduction: Reduction::Mean,
        }
    }

    /// Creates L1Loss with specified reduction.
    pub fn with_reduction(reduction: Reduction) -> Self {
        Self { reduction }
    }

    /// Computes the loss.
    pub fn compute(&self, input: &Variable, target: &Variable) -> Variable {
        let input_data = input.data();
        let target_data = target.data();
        // diff = input - target (Tensor op, auto-dispatches to GPU)
        let diff_tensor = input_data.sub(&target_data).expect("tensor sub failed");
        // |diff| = relu(diff) + relu(-diff), using Tensor ops that auto-dispatch to GPU
        let relu_diff = axonml_tensor::ops::clamp_min(&diff_tensor, 0.0);
        let relu_neg_diff = axonml_tensor::ops::clamp_min(&diff_tensor.neg(), 0.0);
        let abs_tensor = relu_diff.add(&relu_neg_diff).expect("tensor add failed");

        let requires_grad = (input.requires_grad() || target.requires_grad()) && is_grad_enabled();
        let loss_var = if requires_grad {
            let grad_fn = GradFn::new(L1LossBackward {
                next_fns: vec![input.grad_fn().cloned(), target.grad_fn().cloned()],
                diff_tensor,
            });
            Variable::from_operation(abs_tensor, grad_fn, true)
        } else {
            Variable::new(abs_tensor, false)
        };

        match self.reduction {
            Reduction::None => loss_var,
            Reduction::Mean => loss_var.mean(),
            Reduction::Sum => loss_var.sum(),
        }
    }
}

impl Default for L1Loss {
    fn default() -> Self {
        Self::new()
    }
}

// =============================================================================
// L1LossBackward
// =============================================================================

/// Gradient function for L1Loss.
///
/// d/d(input) = sign(input - target)
/// d/d(target) = -sign(input - target)
///
/// Stores diff as Tensor<f32> so it stays on GPU when applicable.
#[derive(Debug)]
struct L1LossBackward {
    next_fns: Vec<Option<GradFn>>,
    diff_tensor: Tensor<f32>,
}

impl GradientFunction for L1LossBackward {
    fn apply(&self, grad_output: &Tensor<f32>) -> Vec<Option<Tensor<f32>>> {
        // sign(diff): +1 where diff > 0, -1 where diff < 0, 0 where diff == 0
        // Compute as: diff / (|diff| + eps) which gives sign and handles GPU
        let eps_tensor = Tensor::full(self.diff_tensor.shape(), 1e-12);
        let eps_on_device = if self.diff_tensor.device().is_gpu() {
            eps_tensor.to_device(self.diff_tensor.device()).unwrap()
        } else {
            eps_tensor
        };
        // |diff| approximated as sqrt(diff^2 + eps)  — but simpler: diff * diff then sqrt
        let diff_sq = self
            .diff_tensor
            .mul(&self.diff_tensor)
            .expect("tensor mul failed");
        let diff_sq_eps = diff_sq.add(&eps_on_device).expect("tensor add failed");
        // sqrt via exp(0.5 * ln(x))
        let abs_diff = diff_sq_eps.ln().mul_scalar(0.5).exp();
        let sign_diff = self.diff_tensor.div(&abs_diff).unwrap();

        // grad_input = sign(diff) * grad_output
        let gi = sign_diff.mul(grad_output).unwrap();
        // grad_target = -grad_input
        let gt = gi.neg();
        vec![Some(gi), Some(gt)]
    }

    fn name(&self) -> &'static str {
        "L1LossBackward"
    }

    fn next_functions(&self) -> &[Option<GradFn>] {
        &self.next_fns
    }

    fn as_any(&self) -> &dyn Any {
        self
    }
}

// =============================================================================
// CrossEntropyBackward
// =============================================================================

/// Gradient function for CrossEntropyLoss.
///
/// The gradient of CE w.r.t. logits is: softmax(logits) - one_hot(target).
/// For per-sample losses, each sample's gradient is scaled by the upstream
/// gradient (from reduction).
#[derive(Debug)]
struct CrossEntropyBackward {
    next_fns: Vec<Option<GradFn>>,
    /// Softmax probabilities computed during forward pass, shape (N, C).
    /// Stays on GPU if input was on GPU.
    softmax_probs: Tensor<f32>,
    /// Target class indices as f32, shape (N,). Stays on GPU if input was on GPU.
    targets: Tensor<f32>,
    batch_size: usize,
    num_classes: usize,
}

impl GradientFunction for CrossEntropyBackward {
    fn apply(&self, grad_output: &Tensor<f32>) -> Vec<Option<Tensor<f32>>> {
        // Always use the exact CPU path for correctness, then transfer
        // result to GPU if needed. The CUDA cross_entropy_bwd kernel has
        // precision issues with approximate reciprocal/exp that cause
        // training to stall. The CPU backward is fast enough since CE
        // backward is O(N*C) — negligible vs the forward pass matmuls.
        let softmax_vec = self.softmax_probs.to_vec();
        let target_vec = self.targets.to_vec();
        let grad_vec = grad_output.to_vec();
        let mut grad_input = vec![0.0f32; self.batch_size * self.num_classes];

        // Handle both per-sample grad_output [N] and scalar [1] (from mean reduction)
        let is_scalar_grad = grad_vec.len() == 1;

        for b in 0..self.batch_size {
            let grad_scale = if is_scalar_grad {
                grad_vec[0]
            } else if b < grad_vec.len() {
                grad_vec[b]
            } else {
                1.0 / self.batch_size as f32
            };
            let offset = b * self.num_classes;
            let tc = target_vec[b] as usize;
            for c in 0..self.num_classes {
                let mut g = softmax_vec[offset + c];
                if c == tc {
                    g -= 1.0;
                }
                grad_input[offset + c] = g * grad_scale;
            }
        }

        let mut grad_tensor = Tensor::from_vec(grad_input, &[self.batch_size, self.num_classes])
            .expect("tensor creation failed");
        // Transfer to GPU if the forward was on GPU
        if self.softmax_probs.device().is_gpu() {
            grad_tensor = grad_tensor.to_device(self.softmax_probs.device()).unwrap();
        }
        vec![Some(grad_tensor)]
    }

    fn name(&self) -> &'static str {
        "CrossEntropyBackward"
    }

    fn next_functions(&self) -> &[Option<GradFn>] {
        &self.next_fns
    }

    fn as_any(&self) -> &dyn Any {
        self
    }
}

// =============================================================================
// CrossEntropyLoss
// =============================================================================

/// Cross entropy loss with log softmax.
///
/// This combines LogSoftmax and NLLLoss in a single class.
///
/// # Shape
/// - Input: (N, C) where C = number of classes
/// - Target: (N,) with class indices
#[derive(Debug, Clone, Copy)]
pub struct CrossEntropyLoss {
    reduction: Reduction,
}

impl CrossEntropyLoss {
    /// Creates a new CrossEntropyLoss with default reduction (Mean).
    pub fn new() -> Self {
        Self {
            reduction: Reduction::Mean,
        }
    }

    /// Creates CrossEntropyLoss with specified reduction.
    pub fn with_reduction(reduction: Reduction) -> Self {
        Self { reduction }
    }

    /// Computes the loss.
    ///
    /// # Arguments
    /// * `input` - Logits of shape (N, C)
    /// * `target` - Class indices of shape (N,) as f32 (will be cast to usize)
    pub fn compute(&self, input: &Variable, target: &Variable) -> Variable {
        let input_data = input.data();
        let target_data = target.data();
        let shape = input_data.shape().to_vec();
        let batch_size = shape[0];
        let num_classes = shape[1];

        // GPU fast path: fused softmax + NLL loss kernel
        #[cfg(feature = "cuda")]
        if input_data.device().is_gpu() {
            // Ensure targets are on GPU
            let targets_gpu = if target_data.device().is_gpu() {
                target_data.clone()
            } else {
                target_data.to_device(input_data.device()).unwrap()
            };

            let (loss_tensor, softmax_tensor) = input_data.cross_entropy_fwd_cuda(&targets_gpu);

            let loss_var = if input.requires_grad() {
                let grad_fn = GradFn::new(CrossEntropyBackward {
                    next_fns: vec![input.grad_fn().cloned()],
                    softmax_probs: softmax_tensor,
                    targets: targets_gpu,
                    batch_size,
                    num_classes,
                });
                Variable::from_operation(loss_tensor, grad_fn, true)
            } else {
                Variable::new(loss_tensor, false)
            };

            return match self.reduction {
                Reduction::None => loss_var,
                Reduction::Mean => loss_var.mean(),
                Reduction::Sum => loss_var.sum(),
            };
        }

        // CPU path
        let input_vec = input_data.to_vec();
        let target_vec = target_data.to_vec();

        let mut losses = vec![0.0f32; batch_size];
        let mut softmax_probs_vec = vec![0.0f32; batch_size * num_classes];
        let mut target_classes = vec![0usize; batch_size];

        for b in 0..batch_size {
            let offset = b * num_classes;

            // Numerically stable log-softmax
            let max_val = (0..num_classes)
                .map(|c| input_vec[offset + c])
                .fold(f32::NEG_INFINITY, f32::max);

            let mut sum_exp = 0.0f32;
            for c in 0..num_classes {
                let exp_val = (input_vec[offset + c] - max_val).exp();
                softmax_probs_vec[offset + c] = exp_val;
                sum_exp += exp_val;
            }

            for c in 0..num_classes {
                softmax_probs_vec[offset + c] /= sum_exp;
            }

            let log_sum_exp = max_val + sum_exp.ln();

            let tc = target_vec[b] as usize;
            target_classes[b] = tc;
            losses[b] = log_sum_exp - input_vec[offset + tc];
        }

        let loss_tensor = Tensor::from_vec(losses, &[batch_size]).expect("tensor creation failed");
        let softmax_tensor = Tensor::from_vec(softmax_probs_vec, &[batch_size, num_classes])
            .expect("tensor creation failed");
        let targets_f32: Vec<f32> = target_classes.iter().map(|&tc| tc as f32).collect();
        let targets_tensor =
            Tensor::from_vec(targets_f32, &[batch_size]).expect("tensor creation failed");

        let loss_var = if input.requires_grad() {
            let grad_fn = GradFn::new(CrossEntropyBackward {
                next_fns: vec![input.grad_fn().cloned()],
                softmax_probs: softmax_tensor,
                targets: targets_tensor,
                batch_size,
                num_classes,
            });
            Variable::from_operation(loss_tensor, grad_fn, true)
        } else {
            Variable::new(loss_tensor, false)
        };

        match self.reduction {
            Reduction::None => loss_var,
            Reduction::Mean => loss_var.mean(),
            Reduction::Sum => loss_var.sum(),
        }
    }
}

impl Default for CrossEntropyLoss {
    fn default() -> Self {
        Self::new()
    }
}

// =============================================================================
// NLLLoss
// =============================================================================

/// Negative Log Likelihood loss.
///
/// Expects input to be log-probabilities.
#[derive(Debug, Clone, Copy)]
pub struct NLLLoss {
    reduction: Reduction,
}

impl NLLLoss {
    /// Creates a new NLLLoss with default reduction (Mean).
    pub fn new() -> Self {
        Self {
            reduction: Reduction::Mean,
        }
    }

    /// Creates NLLLoss with specified reduction.
    pub fn with_reduction(reduction: Reduction) -> Self {
        Self { reduction }
    }

    /// Computes the loss.
    pub fn compute(&self, input: &Variable, target: &Variable) -> Variable {
        let input_data = input.data();
        let target_data = target.data();
        let shape = input_data.shape().to_vec();
        let batch_size = shape[0];
        let num_classes = shape[1];

        // NLL forward still needs per-sample gather (index into class dimension).
        // We pull target indices to CPU for the gather but keep input on device.
        let target_vec = target_data.to_vec();
        let input_vec = input_data.to_vec();

        let mut losses = vec![0.0f32; batch_size];
        for b in 0..batch_size {
            let tc = target_vec[b] as usize;
            losses[b] = -input_vec[b * num_classes + tc];
        }

        let mut loss_tensor =
            Tensor::from_vec(losses, &[batch_size]).expect("tensor creation failed");
        if input_data.device().is_gpu() {
            loss_tensor = loss_tensor.to_device(input_data.device()).unwrap();
        }

        let requires_grad = input.requires_grad() && is_grad_enabled();
        let loss_var = if requires_grad {
            let grad_fn = GradFn::new(NLLLossBackward {
                next_fns: vec![input.grad_fn().cloned()],
                target_tensor: target_data.clone(),
                batch_size,
                num_classes,
            });
            Variable::from_operation(loss_tensor, grad_fn, true)
        } else {
            Variable::new(loss_tensor, false)
        };

        match self.reduction {
            Reduction::None => loss_var,
            Reduction::Mean => loss_var.mean(),
            Reduction::Sum => loss_var.sum(),
        }
    }
}

impl Default for NLLLoss {
    fn default() -> Self {
        Self::new()
    }
}

// =============================================================================
// NLLLossBackward
// =============================================================================

/// Gradient function for NLLLoss.
///
/// d/d(input)[b, c] = -1 if c == target[b], else 0
///
/// Stores targets as Tensor<f32> (GPU-resident when applicable).
/// The scatter in backward still uses CPU indexing since it's a sparse write,
/// but the result is moved to GPU if needed.
#[derive(Debug)]
struct NLLLossBackward {
    next_fns: Vec<Option<GradFn>>,
    target_tensor: Tensor<f32>,
    batch_size: usize,
    num_classes: usize,
}

impl GradientFunction for NLLLossBackward {
    fn apply(&self, grad_output: &Tensor<f32>) -> Vec<Option<Tensor<f32>>> {
        let grad_out_vec = grad_output.to_vec();
        let target_vec = self.target_tensor.to_vec();
        let mut grad_input = vec![0.0f32; self.batch_size * self.num_classes];

        for b in 0..self.batch_size {
            let g = if grad_out_vec.len() == 1 {
                grad_out_vec[0]
            } else {
                grad_out_vec[b]
            };
            let tc = target_vec[b] as usize;
            grad_input[b * self.num_classes + tc] = -g;
        }

        let mut gi = Tensor::from_vec(grad_input, &[self.batch_size, self.num_classes])
            .expect("tensor creation failed");
        if grad_output.device().is_gpu() {
            gi = gi.to_device(grad_output.device()).unwrap();
        }
        vec![Some(gi)]
    }

    fn name(&self) -> &'static str {
        "NLLLossBackward"
    }

    fn next_functions(&self) -> &[Option<GradFn>] {
        &self.next_fns
    }

    fn as_any(&self) -> &dyn Any {
        self
    }
}

// =============================================================================
// BCELoss
// =============================================================================

/// Binary Cross Entropy loss.
///
/// Expects input to be probabilities in [0, 1].
#[derive(Debug, Clone, Copy)]
pub struct BCELoss {
    reduction: Reduction,
}

impl BCELoss {
    /// Creates a new BCELoss with default reduction (Mean).
    pub fn new() -> Self {
        Self {
            reduction: Reduction::Mean,
        }
    }

    /// Creates BCELoss with specified reduction.
    pub fn with_reduction(reduction: Reduction) -> Self {
        Self { reduction }
    }

    /// Computes the loss.
    pub fn compute(&self, input: &Variable, target: &Variable) -> Variable {
        let input_data = input.data();
        let target_data = target.data();

        // Clamp predictions to [eps, 1-eps] using Tensor ops
        let eps = 1e-7f32;
        let p_clamped = axonml_tensor::ops::clamp(&input_data, eps, 1.0 - eps);

        // loss = -(t * ln(p) + (1 - t) * ln(1 - p))
        let ln_p = p_clamped.ln();
        let one_minus_p = p_clamped.neg().add_scalar(1.0);
        let ln_one_minus_p = one_minus_p.ln();
        let one_minus_t = target_data.neg().add_scalar(1.0);

        // t * ln(p)
        let term1 = target_data.mul(&ln_p).expect("tensor mul failed");
        // (1-t) * ln(1-p)
        let term2 = one_minus_t.mul(&ln_one_minus_p).expect("tensor mul failed");
        // -(term1 + term2)
        let loss_tensor = term1.add(&term2).expect("tensor add failed").neg();

        let requires_grad = input.requires_grad() && is_grad_enabled();
        let loss_var = if requires_grad {
            let grad_fn = GradFn::new(BCELossBackward {
                next_fns: vec![input.grad_fn().cloned()],
                input_tensor: input_data,
                target_tensor: target_data,
            });
            Variable::from_operation(loss_tensor, grad_fn, true)
        } else {
            Variable::new(loss_tensor, false)
        };

        match self.reduction {
            Reduction::None => loss_var,
            Reduction::Mean => loss_var.mean(),
            Reduction::Sum => loss_var.sum(),
        }
    }
}

impl Default for BCELoss {
    fn default() -> Self {
        Self::new()
    }
}

// =============================================================================
// BCELossBackward
// =============================================================================

/// Gradient function for BCELoss.
///
/// d/dp BCE = (p - y) / (p * (1 - p))
///
/// Stores input/target as Tensor<f32> (GPU-resident when applicable).
#[derive(Debug)]
struct BCELossBackward {
    next_fns: Vec<Option<GradFn>>,
    input_tensor: Tensor<f32>,
    target_tensor: Tensor<f32>,
}

impl GradientFunction for BCELossBackward {
    fn apply(&self, grad_output: &Tensor<f32>) -> Vec<Option<Tensor<f32>>> {
        let eps = 1e-7f32;
        // p_clamped = clamp(input, eps, 1-eps)
        let p_clamped = axonml_tensor::ops::clamp(&self.input_tensor, eps, 1.0 - eps);
        // (p - y)
        let p_minus_y = p_clamped
            .sub(&self.target_tensor)
            .expect("tensor sub failed");
        // p * (1 - p)
        let one_minus_p = p_clamped.neg().add_scalar(1.0);
        let denom = p_clamped.mul(&one_minus_p).expect("tensor mul failed");
        // grad = grad_output * (p - y) / (p * (1 - p))
        let ratio = p_minus_y.div(&denom).unwrap();
        let grad_tensor = grad_output.mul(&ratio).expect("tensor mul failed");
        vec![Some(grad_tensor)]
    }

    fn name(&self) -> &'static str {
        "BCELossBackward"
    }

    fn next_functions(&self) -> &[Option<GradFn>] {
        &self.next_fns
    }

    fn as_any(&self) -> &dyn Any {
        self
    }
}

// =============================================================================
// BCEWithLogitsBackward
// =============================================================================

/// Gradient function for BCEWithLogitsLoss.
///
/// The gradient of BCE w.r.t. input logits is: sigmoid(input) - target.
///
/// Stores input/target as Tensor<f32> (GPU-resident when applicable).
#[derive(Debug)]
struct BCEWithLogitsBackward {
    next_fns: Vec<Option<GradFn>>,
    input_tensor: Tensor<f32>,
    target_tensor: Tensor<f32>,
}

impl GradientFunction for BCEWithLogitsBackward {
    fn apply(&self, grad_output: &Tensor<f32>) -> Vec<Option<Tensor<f32>>> {
        // sigmoid(input) - target, all via Tensor ops (auto-dispatch to GPU)
        let sig = self.input_tensor.sigmoid();
        let sig_minus_t = sig.sub(&self.target_tensor).expect("tensor sub failed");
        let grad_tensor = grad_output.mul(&sig_minus_t).expect("tensor mul failed");
        vec![Some(grad_tensor)]
    }

    fn name(&self) -> &'static str {
        "BCEWithLogitsBackward"
    }

    fn next_functions(&self) -> &[Option<GradFn>] {
        &self.next_fns
    }

    fn as_any(&self) -> &dyn Any {
        self
    }
}

// =============================================================================
// BCEWithLogitsLoss
// =============================================================================

/// Binary Cross Entropy with Logits.
///
/// Combines sigmoid and BCE in a numerically stable way.
#[derive(Debug, Clone, Copy)]
pub struct BCEWithLogitsLoss {
    reduction: Reduction,
}

impl BCEWithLogitsLoss {
    /// Creates a new BCEWithLogitsLoss with default reduction (Mean).
    pub fn new() -> Self {
        Self {
            reduction: Reduction::Mean,
        }
    }

    /// Creates BCEWithLogitsLoss with specified reduction.
    pub fn with_reduction(reduction: Reduction) -> Self {
        Self { reduction }
    }

    /// Computes the loss.
    pub fn compute(&self, input: &Variable, target: &Variable) -> Variable {
        let input_data = input.data();
        let target_data = target.data();

        // Numerically stable: max(x, 0) - x*t + log(1 + exp(-|x|))
        // max(x, 0) = relu(x) = clamp_min(x, 0)
        let relu_x = axonml_tensor::ops::clamp_min(&input_data, 0.0);
        // x * t
        let x_times_t = input_data.mul(&target_data).expect("tensor mul failed");
        // |x| via clamp trick: max(x, 0) + max(-x, 0) = relu(x) + relu(-x)
        let neg_x = input_data.neg();
        let relu_neg_x = axonml_tensor::ops::clamp_min(&neg_x, 0.0);
        let abs_x = relu_x.add(&relu_neg_x).expect("tensor add failed");
        // exp(-|x|)
        let exp_neg_abs = abs_x.neg().exp();
        // log(1 + exp(-|x|))
        let log_term = exp_neg_abs.add_scalar(1.0).ln();
        // loss = relu(x) - x*t + log(1 + exp(-|x|))
        let loss_tensor = relu_x
            .sub(&x_times_t)
            .expect("tensor sub failed")
            .add(&log_term)
            .expect("tensor add failed");

        let loss_var = if input.requires_grad() {
            let grad_fn = GradFn::new(BCEWithLogitsBackward {
                next_fns: vec![input.grad_fn().cloned()],
                input_tensor: input_data,
                target_tensor: target_data,
            });
            Variable::from_operation(loss_tensor, grad_fn, true)
        } else {
            Variable::new(loss_tensor, false)
        };

        match self.reduction {
            Reduction::None => loss_var,
            Reduction::Mean => loss_var.mean(),
            Reduction::Sum => loss_var.sum(),
        }
    }
}

impl Default for BCEWithLogitsLoss {
    fn default() -> Self {
        Self::new()
    }
}

// =============================================================================
// SmoothL1Backward
// =============================================================================

/// Gradient function for SmoothL1Loss.
///
/// The gradient is: diff/beta if |diff| < beta, else sign(diff).
/// Returns gradients for both input and target (negated for target).
///
/// Stores diff as Tensor<f32> (GPU-resident when applicable).
#[derive(Debug)]
struct SmoothL1Backward {
    next_fns: Vec<Option<GradFn>>,
    diff_tensor: Tensor<f32>,
    beta: f32,
    shape: Vec<usize>,
}

impl GradientFunction for SmoothL1Backward {
    fn apply(&self, grad_output: &Tensor<f32>) -> Vec<Option<Tensor<f32>>> {
        // Compute |diff| = sqrt(diff^2 + eps) to stay differentiable and device-agnostic
        let eps = 1e-12f32;
        let diff_sq = self
            .diff_tensor
            .mul(&self.diff_tensor)
            .expect("tensor mul failed");
        let diff_sq_eps = diff_sq.add_scalar(eps);
        let abs_diff = diff_sq_eps.ln().mul_scalar(0.5).exp();

        // sign(diff) = diff / |diff|
        let sign_diff = self.diff_tensor.div(&abs_diff).unwrap();

        // For the L2 region (|diff| < beta): grad = diff / beta
        // For the L1 region (|diff| >= beta): grad = sign(diff)
        // Blend: mask = clamp(|diff| / beta, 0, 1), but we need a hard cutoff.
        // Use a smooth approximation via where: if |diff| < beta -> diff/beta, else sign
        // Since we need element-wise branching and don't have a GPU where_cond,
        // we use: grad = (1 - mask) * (diff / beta) + mask * sign(diff)
        // where mask = clamp((|diff| - beta) * large_value, 0, 1) approximates step function.
        // Actually simpler: mask_l2 = clamp(1 - |diff|/beta, 0, 1) gives 1 in L2 region, 0 in L1
        // BUT this gives a soft transition. For exact correctness, use CPU branching on the mask.
        //
        // Practical approach: compute both branches with Tensor ops, build mask on CPU, blend.
        let grad_l2 = self.diff_tensor.mul_scalar(1.0 / self.beta); // diff / beta
        let grad_l1 = sign_diff; // sign(diff)

        // Build mask tensor: 1.0 where |d| < beta, 0.0 otherwise
        let abs_vec = abs_diff.to_vec();
        let beta = self.beta;
        let mask_vec: Vec<f32> = abs_vec
            .iter()
            .map(|&a| if a < beta { 1.0 } else { 0.0 })
            .collect();
        let mut mask = Tensor::from_vec(mask_vec, &self.shape).expect("tensor creation failed");
        if self.diff_tensor.device().is_gpu() {
            mask = mask.to_device(self.diff_tensor.device()).unwrap();
        }
        let inv_mask = mask.neg().add_scalar(1.0);

        // grad_per_elem = mask * grad_l2 + (1 - mask) * grad_l1
        let blended = mask
            .mul(&grad_l2)
            .unwrap()
            .add(&inv_mask.mul(&grad_l1).expect("tensor add failed"))
            .unwrap();

        // gi = blended * grad_output
        let gi = blended.mul(grad_output).unwrap();
        let gt = gi.neg();
        vec![Some(gi), Some(gt)]
    }

    fn name(&self) -> &'static str {
        "SmoothL1Backward"
    }

    fn next_functions(&self) -> &[Option<GradFn>] {
        &self.next_fns
    }

    fn as_any(&self) -> &dyn Any {
        self
    }
}

// =============================================================================
// SmoothL1Loss
// =============================================================================

/// Smooth L1 loss (Huber loss).
///
/// Uses L2 loss when |x| < beta, L1 loss otherwise.
#[derive(Debug, Clone, Copy)]
pub struct SmoothL1Loss {
    reduction: Reduction,
    beta: f32,
}

impl SmoothL1Loss {
    /// Creates a new SmoothL1Loss with default beta (1.0).
    pub fn new() -> Self {
        Self {
            reduction: Reduction::Mean,
            beta: 1.0,
        }
    }

    /// Creates SmoothL1Loss with specified beta.
    pub fn with_beta(beta: f32) -> Self {
        Self {
            reduction: Reduction::Mean,
            beta,
        }
    }

    /// Computes the loss.
    pub fn compute(&self, input: &Variable, target: &Variable) -> Variable {
        let input_data = input.data();
        let target_data = target.data();
        let diff_tensor = input_data.sub(&target_data).expect("tensor sub failed");
        let shape = diff_tensor.shape().to_vec();

        // Compute |diff| via relu(diff) + relu(-diff)
        let relu_diff = axonml_tensor::ops::clamp_min(&diff_tensor, 0.0);
        let relu_neg_diff = axonml_tensor::ops::clamp_min(&diff_tensor.neg(), 0.0);
        let abs_diff = relu_diff.add(&relu_neg_diff).expect("tensor add failed");

        // L2 branch: 0.5 * diff^2 / beta
        let diff_sq = diff_tensor.mul(&diff_tensor).expect("tensor mul failed");
        let l2_loss = diff_sq.mul_scalar(0.5 / self.beta);

        // L1 branch: |diff| - 0.5 * beta
        let l1_loss = abs_diff.add_scalar(-0.5 * self.beta);

        // Build mask: 1.0 where |diff| < beta, 0.0 otherwise
        let abs_vec = abs_diff.to_vec();
        let beta = self.beta;
        let mask_vec: Vec<f32> = abs_vec
            .iter()
            .map(|&a| if a < beta { 1.0 } else { 0.0 })
            .collect();
        let mut mask = Tensor::from_vec(mask_vec, &shape).expect("tensor creation failed");
        if diff_tensor.device().is_gpu() {
            mask = mask.to_device(diff_tensor.device()).unwrap();
        }
        let inv_mask = mask.neg().add_scalar(1.0);

        // loss = mask * l2_loss + (1-mask) * l1_loss
        let loss_tensor = mask
            .mul(&l2_loss)
            .unwrap()
            .add(&inv_mask.mul(&l1_loss).expect("tensor add failed"))
            .unwrap();

        let loss_var = if input.requires_grad() || target.requires_grad() {
            let grad_fn = GradFn::new(SmoothL1Backward {
                next_fns: vec![input.grad_fn().cloned(), target.grad_fn().cloned()],
                diff_tensor,
                beta: self.beta,
                shape,
            });
            Variable::from_operation(loss_tensor, grad_fn, true)
        } else {
            Variable::new(loss_tensor, false)
        };

        match self.reduction {
            Reduction::None => loss_var,
            Reduction::Mean => loss_var.mean(),
            Reduction::Sum => loss_var.sum(),
        }
    }
}

impl Default for SmoothL1Loss {
    fn default() -> Self {
        Self::new()
    }
}

// =============================================================================
// Tests
// =============================================================================

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_mse_loss() {
        let loss_fn = MSELoss::new();
        let input = Variable::new(
            Tensor::from_vec(vec![1.0, 2.0, 3.0], &[3]).expect("tensor creation failed"),
            false,
        );
        let target = Variable::new(
            Tensor::from_vec(vec![1.0, 2.0, 3.0], &[3]).expect("tensor creation failed"),
            false,
        );
        let loss = loss_fn.compute(&input, &target);
        assert!((loss.data().to_vec()[0] - 0.0).abs() < 1e-6);
    }

    #[test]
    fn test_mse_loss_nonzero() {
        let loss_fn = MSELoss::new();
        let input = Variable::new(
            Tensor::from_vec(vec![1.0, 2.0, 3.0], &[3]).expect("tensor creation failed"),
            false,
        );
        let target = Variable::new(
            Tensor::from_vec(vec![2.0, 3.0, 4.0], &[3]).expect("tensor creation failed"),
            false,
        );
        let loss = loss_fn.compute(&input, &target);
        // Each diff is 1.0, squared is 1.0, mean is 1.0
        assert!((loss.data().to_vec()[0] - 1.0).abs() < 1e-6);
    }

    #[test]
    fn test_cross_entropy_loss() {
        let loss_fn = CrossEntropyLoss::new();
        let input = Variable::new(
            Tensor::from_vec(vec![1.0, 2.0, 3.0, 1.0, 2.0, 3.0], &[2, 3])
                .expect("tensor creation failed"),
            false,
        );
        let target = Variable::new(
            Tensor::from_vec(vec![2.0, 0.0], &[2]).expect("tensor creation failed"),
            false,
        );
        let loss = loss_fn.compute(&input, &target);
        assert!(loss.data().to_vec()[0] > 0.0);
    }

    #[test]
    fn test_bce_loss() {
        let loss_fn = BCELoss::new();
        let input = Variable::new(
            Tensor::from_vec(vec![0.5, 0.5], &[2]).expect("tensor creation failed"),
            false,
        );
        let target = Variable::new(
            Tensor::from_vec(vec![1.0, 0.0], &[2]).expect("tensor creation failed"),
            false,
        );
        let loss = loss_fn.compute(&input, &target);
        // -[1*ln(0.5) + 0*ln(0.5)] - [0*ln(0.5) + 1*ln(0.5)] = -2*ln(0.5) / 2 = -ln(0.5) = 0.693
        assert!((loss.data().to_vec()[0] - 0.693).abs() < 0.01);
    }

    #[test]
    fn test_cross_entropy_gradient_flow() {
        use axonml_autograd::backward;

        // Create input logits with requires_grad=true
        let input = Variable::new(
            Tensor::from_vec(vec![2.0, 1.0, 0.1, 0.5, 2.5, 0.3], &[2, 3])
                .expect("tensor creation failed"),
            true,
        );
        let target = Variable::new(
            Tensor::from_vec(vec![0.0, 1.0], &[2]).expect("tensor creation failed"),
            false,
        );

        let loss_fn = CrossEntropyLoss::new();
        let loss = loss_fn.compute(&input, &target);

        // Loss should be positive
        let loss_val = loss.data().to_vec()[0];
        assert!(loss_val > 0.0, "Loss should be positive, got {}", loss_val);

        // Backward pass
        let ones = Tensor::from_vec(vec![1.0], &loss.shape()).unwrap();
        backward(&loss, &ones);

        // Input should have gradient
        let grad = input
            .grad()
            .expect("Input should have gradient after backward");
        let grad_vec = grad.to_vec();

        // Gradient should be non-zero
        let grad_norm: f32 = grad_vec.iter().map(|g| g * g).sum();
        assert!(
            grad_norm > 1e-10,
            "Gradient should be non-zero, got norm {}",
            grad_norm
        );

        // Gradient shape should match input shape
        assert_eq!(grad.shape(), &[2, 3]);

        // For the correct class, gradient should be negative (softmax - 1 < 0)
        // Sample 0, class 0 (target): grad should be (softmax[0,0] - 1) / 2
        assert!(
            grad_vec[0] < 0.0,
            "Gradient for correct class should be negative"
        );
        // Sample 1, class 1 (target): grad should be (softmax[1,1] - 1) / 2
        assert!(
            grad_vec[4] < 0.0,
            "Gradient for correct class should be negative"
        );

        // For wrong classes, gradient should be positive (softmax > 0)
        assert!(
            grad_vec[1] > 0.0,
            "Gradient for wrong class should be positive"
        );
        assert!(
            grad_vec[2] > 0.0,
            "Gradient for wrong class should be positive"
        );
    }

    #[test]
    fn test_cross_entropy_perfect_prediction() {
        // When logits strongly favor the correct class, loss should be near zero
        let loss_fn = CrossEntropyLoss::new();
        let input = Variable::new(
            Tensor::from_vec(vec![10.0, -10.0, -10.0], &[1, 3]).expect("tensor creation failed"),
            false,
        );
        let target = Variable::new(
            Tensor::from_vec(vec![0.0], &[1]).expect("tensor creation failed"),
            false,
        );
        let loss = loss_fn.compute(&input, &target);
        assert!(
            loss.data().to_vec()[0] < 0.001,
            "Perfect prediction should have near-zero loss"
        );
    }

    #[test]
    fn test_cross_entropy_uniform_prediction() {
        // When logits are all equal, loss should be ln(num_classes)
        let loss_fn = CrossEntropyLoss::new();
        let num_classes = 16;
        let input = Variable::new(
            Tensor::from_vec(vec![0.0; num_classes], &[1, num_classes])
                .expect("tensor creation failed"),
            false,
        );
        let target = Variable::new(
            Tensor::from_vec(vec![0.0], &[1]).expect("tensor creation failed"),
            false,
        );
        let loss = loss_fn.compute(&input, &target);
        let expected = (num_classes as f32).ln(); // ln(16) ≈ 2.77
        let actual = loss.data().to_vec()[0];
        assert!(
            (actual - expected).abs() < 0.01,
            "Uniform logits should give ln(C)={}, got {}",
            expected,
            actual,
        );
    }

    #[test]
    fn test_bce_with_logits_gradient_flow() {
        use axonml_autograd::backward;

        let input = Variable::new(
            Tensor::from_vec(vec![0.5, -0.5, 1.0, -1.0], &[4]).expect("tensor creation failed"),
            true,
        );
        let target = Variable::new(
            Tensor::from_vec(vec![1.0, 0.0, 1.0, 0.0], &[4]).expect("tensor creation failed"),
            false,
        );

        let loss_fn = BCEWithLogitsLoss::new();
        let loss = loss_fn.compute(&input, &target);
        assert!(loss.data().to_vec()[0] > 0.0);

        let ones = Tensor::from_vec(vec![1.0], &loss.shape()).unwrap();
        backward(&loss, &ones);

        let grad = input.grad().expect("Input should have gradient");
        let grad_vec = grad.to_vec();
        assert_eq!(grad_vec.len(), 4);

        // For target=1, grad = sigmoid(x) - 1 < 0
        assert!(grad_vec[0] < 0.0);
        // For target=0, grad = sigmoid(x) > 0
        assert!(grad_vec[1] > 0.0);
    }

    #[test]
    fn test_smooth_l1_gradient_flow() {
        use axonml_autograd::backward;

        let input = Variable::new(
            Tensor::from_vec(vec![1.0, 2.0, 5.0], &[3]).expect("tensor creation failed"),
            true,
        );
        let target = Variable::new(
            Tensor::from_vec(vec![1.5, 1.5, 1.5], &[3]).expect("tensor creation failed"),
            false,
        );

        let loss_fn = SmoothL1Loss::new();
        let loss = loss_fn.compute(&input, &target);
        assert!(loss.data().to_vec()[0] > 0.0);

        let ones = Tensor::from_vec(vec![1.0], &loss.shape()).unwrap();
        backward(&loss, &ones);

        let grad = input.grad().expect("Input should have gradient");
        let grad_vec = grad.to_vec();
        assert_eq!(grad_vec.len(), 3);
        // Gradients should be non-zero
        let grad_norm: f32 = grad_vec.iter().map(|g| g * g).sum();
        assert!(grad_norm > 1e-10);
    }

    // =========================================================================
    // MSE Loss Comprehensive
    // =========================================================================

    #[test]
    fn test_mse_loss_gradient_correctness() {
        use axonml_autograd::backward;

        // MSE gradient = 2*(input - target) / N
        let input = Variable::new(Tensor::from_vec(vec![3.0, 1.0], &[2]).unwrap(), true);
        let target = Variable::new(Tensor::from_vec(vec![1.0, 1.0], &[2]).unwrap(), false);

        let loss = MSELoss::new().compute(&input, &target);
        // loss = ((3-1)^2 + (1-1)^2) / 2 = 4/2 = 2.0
        assert!(
            (loss.data().to_vec()[0] - 2.0).abs() < 1e-5,
            "MSE should be 2.0"
        );

        let ones = Tensor::from_vec(vec![1.0], &loss.shape()).unwrap();
        backward(&loss, &ones);

        let grad = input.grad().expect("Should have gradient");
        let gv = grad.to_vec();
        // dL/dx = 2*(x-t)/N = 2*(3-1)/2 = 2.0 for first, 0.0 for second
        assert!(
            (gv[0] - 2.0).abs() < 0.1,
            "Grad[0] should be ~2.0, got {}",
            gv[0]
        );
        assert!(gv[1].abs() < 0.1, "Grad[1] should be ~0.0, got {}", gv[1]);
    }

    #[test]
    fn test_mse_loss_reduction_sum() {
        let input = Variable::new(Tensor::from_vec(vec![2.0, 4.0], &[2]).unwrap(), false);
        let target = Variable::new(Tensor::from_vec(vec![1.0, 1.0], &[2]).unwrap(), false);
        let loss = MSELoss::with_reduction(Reduction::Sum).compute(&input, &target);
        // sum = (2-1)^2 + (4-1)^2 = 1 + 9 = 10
        assert!((loss.data().to_vec()[0] - 10.0).abs() < 1e-5);
    }

    // =========================================================================
    // L1 Loss
    // =========================================================================

    #[test]
    fn test_l1_loss_basic() {
        let input = Variable::new(Tensor::from_vec(vec![1.0, 5.0, 3.0], &[3]).unwrap(), false);
        let target = Variable::new(Tensor::from_vec(vec![1.0, 2.0, 4.0], &[3]).unwrap(), false);
        let loss = L1Loss::new().compute(&input, &target);
        // mean(|0| + |3| + |-1|) = 4/3 ≈ 1.333
        assert!((loss.data().to_vec()[0] - 4.0 / 3.0).abs() < 1e-4);
    }

    #[test]
    fn test_l1_loss_zero() {
        let input = Variable::new(Tensor::from_vec(vec![1.0, 2.0], &[2]).unwrap(), false);
        let target = Variable::new(Tensor::from_vec(vec![1.0, 2.0], &[2]).unwrap(), false);
        let loss = L1Loss::new().compute(&input, &target);
        assert!(
            loss.data().to_vec()[0].abs() < 1e-6,
            "Identical inputs should give 0 loss"
        );
    }

    // =========================================================================
    // BCE Loss Edge Cases
    // =========================================================================

    #[test]
    fn test_bce_loss_perfect_prediction() {
        let loss_fn = BCELoss::new();
        // Near-perfect predictions
        let input = Variable::new(Tensor::from_vec(vec![0.999, 0.001], &[2]).unwrap(), false);
        let target = Variable::new(Tensor::from_vec(vec![1.0, 0.0], &[2]).unwrap(), false);
        let loss = loss_fn.compute(&input, &target);
        assert!(
            loss.data().to_vec()[0] < 0.01,
            "Perfect prediction should have near-zero loss"
        );
    }

    #[test]
    fn test_bce_loss_worst_prediction() {
        let loss_fn = BCELoss::new();
        // Worst predictions (inverted)
        let input = Variable::new(Tensor::from_vec(vec![0.001, 0.999], &[2]).unwrap(), false);
        let target = Variable::new(Tensor::from_vec(vec![1.0, 0.0], &[2]).unwrap(), false);
        let loss = loss_fn.compute(&input, &target);
        // Should be high
        assert!(
            loss.data().to_vec()[0] > 3.0,
            "Worst prediction should have high loss"
        );
    }

    // =========================================================================
    // BCEWithLogitsLoss Comprehensive
    // =========================================================================

    #[test]
    fn test_bce_with_logits_numerical_stability() {
        let loss_fn = BCEWithLogitsLoss::new();
        // Very large logits should not produce NaN/Inf
        let input = Variable::new(
            Tensor::from_vec(vec![100.0, -100.0, 50.0, -50.0], &[4]).unwrap(),
            false,
        );
        let target = Variable::new(
            Tensor::from_vec(vec![1.0, 0.0, 1.0, 0.0], &[4]).unwrap(),
            false,
        );
        let loss = loss_fn.compute(&input, &target);
        let val = loss.data().to_vec()[0];
        assert!(
            val.is_finite(),
            "Loss should be finite for large logits, got {}",
            val
        );
        assert!(val >= 0.0, "BCE loss should be non-negative");
    }

    #[test]
    fn test_bce_with_logits_zero_logits() {
        let loss_fn = BCEWithLogitsLoss::new();
        // Zero logits → sigmoid(0) = 0.5 → random prediction
        let input = Variable::new(Tensor::from_vec(vec![0.0, 0.0], &[2]).unwrap(), false);
        let target = Variable::new(Tensor::from_vec(vec![1.0, 0.0], &[2]).unwrap(), false);
        let loss = loss_fn.compute(&input, &target);
        // Should be ln(2) ≈ 0.693
        assert!((loss.data().to_vec()[0] - 0.693).abs() < 0.01);
    }

    #[test]
    fn test_bce_with_logits_reduction_none() {
        let loss_fn = BCEWithLogitsLoss::with_reduction(Reduction::None);
        let input = Variable::new(Tensor::from_vec(vec![0.0, 0.0, 0.0], &[3]).unwrap(), false);
        let target = Variable::new(Tensor::from_vec(vec![1.0, 0.0, 1.0], &[3]).unwrap(), false);
        let loss = loss_fn.compute(&input, &target);
        // Should return per-element losses, not reduced
        assert_eq!(loss.shape().len(), 1);
        assert_eq!(loss.shape()[0], 3);
    }

    // =========================================================================
    // SmoothL1 Loss
    // =========================================================================

    #[test]
    fn test_smooth_l1_small_error() {
        // For |diff| < beta=1.0: loss = 0.5 * diff^2 / beta
        let loss_fn = SmoothL1Loss::new();
        let input = Variable::new(Tensor::from_vec(vec![1.0], &[1]).unwrap(), false);
        let target = Variable::new(Tensor::from_vec(vec![1.3], &[1]).unwrap(), false);
        let loss = loss_fn.compute(&input, &target);
        // diff=0.3, |0.3| < 1.0, loss = 0.5 * 0.09 / 1.0 = 0.045
        assert!((loss.data().to_vec()[0] - 0.045).abs() < 0.01);
    }

    #[test]
    fn test_smooth_l1_large_error() {
        // For |diff| >= beta=1.0: loss = |diff| - 0.5*beta
        let loss_fn = SmoothL1Loss::new();
        let input = Variable::new(Tensor::from_vec(vec![0.0], &[1]).unwrap(), false);
        let target = Variable::new(Tensor::from_vec(vec![5.0], &[1]).unwrap(), false);
        let loss = loss_fn.compute(&input, &target);
        // diff=5.0, |5| >= 1.0, loss = 5.0 - 0.5 = 4.5
        assert!((loss.data().to_vec()[0] - 4.5).abs() < 0.1);
    }

    // =========================================================================
    // Cross-Entropy Batch Consistency
    // =========================================================================

    #[test]
    fn test_cross_entropy_batch_independence() {
        let loss_fn = CrossEntropyLoss::new();

        // Single sample
        let input1 = Variable::new(
            Tensor::from_vec(vec![2.0, 1.0, 0.1], &[1, 3]).unwrap(),
            false,
        );
        let target1 = Variable::new(Tensor::from_vec(vec![0.0], &[1]).unwrap(), false);
        let loss1 = loss_fn.compute(&input1, &target1).data().to_vec()[0];

        // Same sample duplicated in batch
        let input2 = Variable::new(
            Tensor::from_vec(vec![2.0, 1.0, 0.1, 2.0, 1.0, 0.1], &[2, 3]).unwrap(),
            false,
        );
        let target2 = Variable::new(Tensor::from_vec(vec![0.0, 0.0], &[2]).unwrap(), false);
        let loss2 = loss_fn.compute(&input2, &target2).data().to_vec()[0];

        // Mean reduction should give same result for duplicated batch
        assert!(
            (loss1 - loss2).abs() < 1e-5,
            "Duplicated batch should give same loss: {} vs {}",
            loss1,
            loss2
        );
    }

    #[test]
    fn test_cross_entropy_high_class_count() {
        // Test with many classes (like BirdCLEF 234 species)
        let n_classes = 100;
        let mut logits = vec![0.0f32; n_classes];
        logits[42] = 5.0; // Correct class has high logit

        let loss_fn = CrossEntropyLoss::new();
        let input = Variable::new(Tensor::from_vec(logits, &[1, n_classes]).unwrap(), false);
        let target = Variable::new(Tensor::from_vec(vec![42.0], &[1]).unwrap(), false);
        let loss = loss_fn.compute(&input, &target);
        let val = loss.data().to_vec()[0];
        assert!(val.is_finite(), "Should handle 100 classes");
        assert!(val < 1.0, "Correct class should have low loss, got {}", val);
    }
}