mtmd: use ggml_rope_set_offset (#27521)
* mtmd: use ggml_rope_set_offset * add comment
This commit is contained in:
@@ -44,51 +44,31 @@ ggml_cgraph * clip_graph_gemma4v::build() {
|
||||
|
||||
// similar to build_rope_2d, but use neox ordering
|
||||
auto add_pos = [&](ggml_tensor * cur, const clip_layer &) {
|
||||
const int64_t n_dim = cur->ne[0];
|
||||
const int64_t n_head = cur->ne[1];
|
||||
const int64_t n_pos = cur->ne[2];
|
||||
const int64_t n_dim = cur->ne[0];
|
||||
|
||||
// first half
|
||||
ggml_tensor * first;
|
||||
{
|
||||
first = ggml_view_4d(ctx0, cur,
|
||||
n_dim/2, n_head, n_pos, n_batch,
|
||||
cur->nb[1],
|
||||
cur->nb[2],
|
||||
cur->nb[3],
|
||||
0);
|
||||
first = ggml_rope_ext(
|
||||
ctx0,
|
||||
first,
|
||||
pos_x, // positions
|
||||
nullptr, // freq factors
|
||||
n_dim/2, // n_dims
|
||||
GGML_ROPE_TYPE_NEOX, 0, hparams.rope_theta,
|
||||
1.0f, 0.0f, 1.0f, 0.0f, 0.0f
|
||||
);
|
||||
}
|
||||
// first half, dims [0, n_dim/2)
|
||||
cur = ggml_rope_ext(
|
||||
ctx0,
|
||||
cur,
|
||||
pos_x, // positions
|
||||
nullptr, // freq factors
|
||||
n_dim/2, // n_dims
|
||||
GGML_ROPE_TYPE_NEOX, 0, hparams.rope_theta,
|
||||
1.0f, 0.0f, 1.0f, 0.0f, 0.0f
|
||||
);
|
||||
|
||||
// second half
|
||||
ggml_tensor * second;
|
||||
{
|
||||
second = ggml_view_4d(ctx0, cur,
|
||||
n_dim/2, n_head, n_pos, n_batch,
|
||||
cur->nb[1],
|
||||
cur->nb[2],
|
||||
cur->nb[3],
|
||||
n_dim/2 * ggml_element_size(cur));
|
||||
second = ggml_rope_ext(
|
||||
ctx0,
|
||||
second,
|
||||
pos_y, // positions
|
||||
nullptr, // freq factors
|
||||
n_dim/2, // n_dims
|
||||
GGML_ROPE_TYPE_NEOX, 0, hparams.rope_theta,
|
||||
1.0f, 0.0f, 1.0f, 0.0f, 0.0f
|
||||
);
|
||||
}
|
||||
// second half, dims [n_dim/2, n_dim)
|
||||
cur = ggml_rope_ext(
|
||||
ctx0,
|
||||
cur,
|
||||
pos_y, // positions
|
||||
nullptr, // freq factors
|
||||
n_dim/2, // n_dims
|
||||
GGML_ROPE_TYPE_NEOX, 0, hparams.rope_theta,
|
||||
1.0f, 0.0f, 1.0f, 0.0f, 0.0f
|
||||
);
|
||||
cur = ggml_rope_set_offset(cur, n_dim/2);
|
||||
|
||||
cur = ggml_concat(ctx0, first, second, 0);
|
||||
return cur;
|
||||
};
|
||||
|
||||
|
||||
Reference in New Issue
Block a user