ExternalVendorCode/whisper.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/whisper.cpp.git synced 2025-02-21 01:21:26 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

cuda : fix rope + add tests (llama/7452)

Browse Source 
* cuda : fix rope pos data

ggml-ci

* ggml : drop mode & 1 == 1 support for ggml_rope

ggml-ci

* ggml : support freq_factors for f16 rope (CPU)

ggml-ci

* tests : add rope tests using frequency factors

ggml-ci
This commit is contained in:
Georgi Gerganov 2024-05-22 11:01:35 +03:00
parent c9dcb75118
commit c4d6958b3e
3 changed files with 21 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
ggml-cuda/rope.cu
View File

@ -283,9 +283,9 @@ void ggml_cuda_op_rope(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
const bool is_neox = mode & 2;
const bool is_glm = mode & 4;
if (is_neox) {
pos = (const int32_t *) src1_d;
pos = (const int32_t *) src1_d;
if (is_neox) {
if (src2 != nullptr) {
freq_factors = (const float *) src2->data;
}

20
ggml.c
View File

@ -6245,6 +6245,8 @@ static struct ggml_tensor * ggml_rope_impl(
float xpos_base,
bool xpos_down,
bool inplace) {
GGML_ASSERT((mode & 1) == 0 && "mode & 1 == 1 is no longer supported");
GGML_ASSERT(ggml_is_vector(b));
GGML_ASSERT(b->type == GGML_TYPE_I32);
GGML_ASSERT(a->ne[2] == b->ne[0]);
@ -14413,7 +14415,7 @@ static void ggml_compute_forward_rope_f32(
freq_factors = (const float *) src2->data;
}
} else {
GGML_ASSERT(src2 == NULL && "TODO: freq_factors not implemented for mode 1");
GGML_ASSERT(src2 == NULL && "TODO: freq_factors not implemented for !is_neox");
}
// backward process uses inverse rotation by cos and sin.
@ -14529,6 +14531,7 @@ static void ggml_compute_forward_rope_f32(
}
}
// TODO: deduplicate f16/f32 code
static void ggml_compute_forward_rope_f16(
const struct ggml_compute_params * params,
struct ggml_tensor * dst,
@ -14536,6 +14539,7 @@ static void ggml_compute_forward_rope_f16(
const struct ggml_tensor * src0 = dst->src[0];
const struct ggml_tensor * src1 = dst->src[1];
const struct ggml_tensor * src2 = dst->src[2];
if (params->type == GGML_TASK_TYPE_INIT || params->type == GGML_TASK_TYPE_FINALIZE) {
return;
@ -14588,6 +14592,17 @@ static void ggml_compute_forward_rope_f16(
const bool is_neox = mode & 2;
const bool is_glm = mode & 4;
const float * freq_factors = NULL;
if (is_neox) {
if (src2 != NULL) {
GGML_ASSERT(src2->type == GGML_TYPE_F32);
GGML_ASSERT(src2->ne[0] >= n_dims / 2);
freq_factors = (const float *) src2->data;
}
} else {
GGML_ASSERT(src2 == NULL && "TODO: freq_factors not implemented for !is_neox");
}
// backward process uses inverse rotation by cos and sin.
// cos and sin build a rotation matrix, where the inverse is the transpose.
// this essentially just switches the sign of sin.
@ -14660,10 +14675,11 @@ static void ggml_compute_forward_rope_f16(
// simplified from `(ib * n_dims + ic) * inv_ndims`
float cur_rot = inv_ndims * ic - ib;
float freq_factor = freq_factors ? freq_factors[ic/2] : 1.0f;
float cos_theta, sin_theta;
rope_yarn(
theta_base, freq_scale, corr_dims, cur_rot, ext_factor, attn_factor,
theta_base/freq_factor, freq_scale, corr_dims, cur_rot, ext_factor, attn_factor,
&cos_theta, &sin_theta
);
sin_theta *= sin_sign;

2
ggml.h
View File

@ -1460,7 +1460,7 @@ extern "C" {
struct ggml_tensor * b);
// rotary position embedding
// if mode & 1 == 1, skip n_past elements (DEPRECATED)
// if mode & 1 == 1, skip n_past elements (NOT SUPPORTED)
// if mode & 2 == 1, GPT-NeoX style
// if mode & 4 == 1, ChatGLM style
//