wip: T10 batched forward (validation)

crates/xtrain-cuda/src/ffi.rs

@@ -125,7 +125,9 @@ unsafe extern "C" {
     pub fn launch_silu_f32(x: *const f32, y: *mut f32, n: i32, s: CudaStream);
     pub fn launch_silu_dx_f32(x: *const f32, dy: *const f32, dx: *mut f32, n: i32, s: CudaStream);
 
-    // RoPE (rotate_half), x:[tokens,heads,head_dim], position = token index.
+    // RoPE (rotate_half), x:[tokens,heads,head_dim], position = (token index %
+    // period). `period` = sequence length, so a flattened batch of sequences gets
+    // per-sequence positions; period == tokens reproduces the single-sequence case.
     pub fn launch_rope_f32(
         x: *const f32,
         y: *mut f32,
@@ -133,6 +135,7 @@ unsafe extern "C" {
         heads: i32,
         head_dim: i32,
         theta: f32,
+        period: i32,
         s: CudaStream,
     );
     pub fn launch_rope_dx_f32(
@@ -142,6 +145,7 @@ unsafe extern "C" {
         heads: i32,
         head_dim: i32,
         theta: f32,
+        period: i32,
         s: CudaStream,
     );