Function gdn_verify_fused_conv_kn_batched 

pub fn gdn_verify_fused_conv_kn_batched(
    gpu: &dyn GpuBackend,
    kernel: KernelHandle,
    conv_state: DevicePtr,
    new_input: DevicePtr,
    weight: &DenseWeight,
    output: DevicePtr,
    conv_state_inter: DevicePtr,
    num_tokens: u32,
    d_inner: u32,
    d_conv: u32,
    qk_channels: u32,
    head_dim: u32,
    input_stride: u32,
    output_stride: u32,
    inter_stride: u32,
    l2_eps: f32,
    n_seq: u32,
    conv_state_seq_stride: u32,
    input_seq_stride: u32,
    output_seq_stride: u32,
    inter_seq_stride: u32,
    stream: u64,
) -> Result<()>

2026-09-25: gdn_verify_fused_conv_kn for n_seq sequences in one launch, one grid row (blockIdx.y) per sequence.

The kernel offsets conv_state, new_input, output and conv_state_inter by seq times their *_seq_stride, then runs the same per-sequence body as gdn_verify_fused_conv_kn, so each sequence gets the result a separate call would give it.

Grid: (ceil(d_inner/256), n_seq, 1) Block: (256, 1, 1)