haraldwolff/llama.cpp
1
0
Fork 0
Code Issues Pull requests Packages Projects Releases Wiki Activity

ggml-cuda : support stablelm rope (#4156)

Browse source 
* ggml-cuda : support stablelm rope

* remove unused freq_base kernel parameter

* add n_dims parameter to llm_build_k_shift, default to n_rot via overload

* llama : fix llm_build_k_shift args

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
This commit is contained in:
slaren 2023-11-24 18:04:31 +01:00 committed by GitHub
parent 189d68446e
commit 8a052c131e
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
2 changed files with 24 additions and 17 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

37
ggml-cuda.cu
View file

@ -4610,8 +4610,8 @@ static __global__ void rope(
template<typename T, bool has_pos>
static __global__ void rope_neox(
const T * x, T * dst, int ncols, const int32_t * pos, float freq_scale, int p_delta_rows, float freq_base,
float ext_factor, float attn_factor, rope_corr_dims corr_dims
const T * x, T * dst, int ncols, int n_dims, const int32_t * pos, float freq_scale, int p_delta_rows,
float ext_factor, float attn_factor, rope_corr_dims corr_dims, float theta_scale, float inv_ndims
) {
const int col = 2*(blockDim.y*blockIdx.y + threadIdx.y);
@ -4620,23 +4620,25 @@ static __global__ void rope_neox(
}
const int row = blockDim.x*blockIdx.x + threadIdx.x;
const int i = row*ncols + col/2;
const int ib = col / n_dims;
const int ic = col % n_dims;
const int i = row*ncols + ib*n_dims + ic/2;
const int i2 = row/p_delta_rows;
// simplified from `(ib * ncols + col) * (-1 / ncols)`, where ib is assumed to be zero
const float cur_rot = -float(col)/ncols;
float cur_rot = inv_ndims * ic - ib;
const int p = has_pos ? pos[i2] : 0;
const float theta_base = p*powf(freq_base, cur_rot);
const float theta_base = p*freq_scale*powf(theta_scale, col/2.0f);
float cos_theta, sin_theta;
rope_yarn(theta_base, freq_scale, corr_dims, cur_rot, ext_factor, attn_factor, &cos_theta, &sin_theta);
const float x0 = x[i + 0];
const float x1 = x[i + ncols/2];
const float x1 = x[i + n_dims/2];
dst[i + 0] = x0*cos_theta - x1*sin_theta;
dst[i + ncols/2] = x0*sin_theta + x1*cos_theta;
dst[i + 0] = x0*cos_theta - x1*sin_theta;
dst[i + n_dims/2] = x0*sin_theta + x1*cos_theta;
}
static __global__ void rope_glm_f32(
@ -5739,20 +5741,26 @@ static void rope_cuda(
template<typename T>
static void rope_neox_cuda(
const T * x, T * dst, int ncols, int nrows, const int32_t * pos, float freq_scale, int p_delta_rows,
const T * x, T * dst, int ncols, int n_dims, int nrows, const int32_t * pos, float freq_scale, int p_delta_rows,
float freq_base, float ext_factor, float attn_factor, rope_corr_dims corr_dims, cudaStream_t stream
) {
GGML_ASSERT(ncols % 2 == 0);
const dim3 block_dims(1, CUDA_ROPE_BLOCK_SIZE, 1);
const int num_blocks_x = (ncols + 2*CUDA_ROPE_BLOCK_SIZE - 1) / (2*CUDA_ROPE_BLOCK_SIZE);
const dim3 block_nums(nrows, num_blocks_x, 1);
const float theta_scale = powf(freq_base, -2.0f/n_dims);
const float inv_ndims = -1.0f / n_dims;
if (pos == nullptr) {
rope_neox<T, false><<<block_nums, block_dims, 0, stream>>>(
x, dst, ncols, pos, freq_scale, p_delta_rows, freq_base, ext_factor, attn_factor, corr_dims
x, dst, ncols, n_dims, pos, freq_scale, p_delta_rows, ext_factor, attn_factor, corr_dims,
theta_scale, inv_ndims
);
} else {
rope_neox<T, true><<<block_nums, block_dims, 0, stream>>>(
x, dst, ncols, pos, freq_scale, p_delta_rows, freq_base, ext_factor, attn_factor, corr_dims
x, dst, ncols, n_dims, pos, freq_scale, p_delta_rows, ext_factor, attn_factor, corr_dims,
theta_scale, inv_ndims
);
}
}
@ -6707,15 +6715,14 @@ inline void ggml_cuda_op_rope(
GGML_ASSERT(false);
rope_glm_f32_cuda(src0_dd, dst_dd, ne00, nrows, pos, freq_scale, ne01, freq_base, n_ctx, main_stream);
} else if (is_neox) {
GGML_ASSERT(ne00 == n_dims && "ne00 != n_dims is not implemented for CUDA yet");
if (src0->type == GGML_TYPE_F32) {
rope_neox_cuda(
(const float *)src0_dd, (float *)dst_dd, ne00, nrows, pos, freq_scale, ne01, freq_base, ext_factor,
(const float *)src0_dd, (float *)dst_dd, ne00, n_dims, nrows, pos, freq_scale, ne01, freq_base, ext_factor,
attn_factor, corr_dims, main_stream
);
} else if (src0->type == GGML_TYPE_F16) {
rope_neox_cuda(
(const half *)src0_dd, (half *)dst_dd, ne00, nrows, pos, freq_scale, ne01, freq_base, ext_factor,
(const half *)src0_dd, (half *)dst_dd, ne00, n_dims, nrows, pos, freq_scale, ne01, freq_base, ext_factor,
attn_factor, corr_dims, main_stream
);
} else {

4
llama.cpp
View file

@ -3469,7 +3469,7 @@ static void llm_build_k_shift(
struct ggml_cgraph * graph,
llm_rope_type type,
int64_t n_ctx,
int64_t n_rot,
int n_rot,
float freq_base,
float freq_scale,
const llm_build_cb & cb) {
@ -3501,7 +3501,7 @@ static void llm_build_k_shift(
// we rotate only the first n_rot dimensions
ggml_rope_custom_inplace(ctx,
ggml_view_3d(ctx, kv.k,
n_rot, n_head_kv, n_ctx,
n_embd_head, n_head_kv, n_ctx,
ggml_element_size(kv.k)*n_embd_head,
ggml_element_size(kv.k)*n_embd_gqa,
ggml_element_size(kv.k)*n_embd_gqa*n_ctx*il),