pub fn read_k_lr_slot(
ctx: &CudaCtx,
pred: &Predictor,
layer: usize,
block: usize,
) -> Result<Vec<bf16>>Expand description
Helper for tests / debug: copy the predictor’s K_lr slot for a given
(layer, block) to host as BF16. Layout [num_kv_heads, block_size, r].