extern/whisper.cpp
1
0
Fork 1
mirror of https://github.com/ggerganov/whisper.cpp.git synced 2025-02-05 04:50:18 +01:00
Code Issues Packages Projects Releases Wiki Activity

ggml : refactor rope norm/neox (llama/7634)

Browse Source 
* ggml : unify rope norm/neox (CPU)

* ggml : fix compile warning

* ggml : remove GLM rope mode

ggml-ci

* metal : better rope implementation

ggml-ci

* cuda : better rope implementation

ggml-ci

* naming : n_orig_ctx -> n_ctx_orig

ggml-ci

* dev : add reminders to update backends

ggml-ci

* vulkan : fix ggml_rope_ext() usage

* cuda : fix array size + indents

ggml-ci
This commit is contained in:
Georgi Gerganov 2024-06-05 11:29:20 +03:00
parent e666315fa8
commit abab4500fa
8 changed files with 380 additions and 628 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

275
ggml-cuda/rope.cu
View File

@ -1,7 +1,7 @@
#include "rope.cuh"
struct rope_corr_dims {
float v[4];
float v[2];
};
static __device__ float rope_yarn_ramp(const float low, const float high, const int i0) {
@ -13,8 +13,7 @@ static __device__ float rope_yarn_ramp(const float low, const float high, const
// MIT licensed. Copyright (c) 2023 Jeffrey Quesnelle and Bowen Peng.
static __device__ void rope_yarn(
float theta_extrap, float freq_scale, rope_corr_dims corr_dims, int64_t i0, float ext_factor, float mscale,
float * cos_theta, float * sin_theta
) {
float * cos_theta, float * sin_theta) {
// Get n-d rotational scaling corrected for extrapolation
float theta_interp = freq_scale * theta_extrap;
float theta = theta_interp;
@ -29,27 +28,38 @@ static __device__ void rope_yarn(
*sin_theta = sinf(theta) * mscale;
}
// rope == RoPE == rotary positional embedding
template<typename T, bool has_pos>
static __global__ void rope(
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 int col = 2*(blockDim.y*blockIdx.y + threadIdx.y);
template<typename T, bool has_ff>
static __global__ void rope_norm(
const T * x, T * dst, int ne0, 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, const float * freq_factors) {
const int i0 = 2*(blockDim.y*blockIdx.y + threadIdx.y);
if (col >= ncols) {
if (i0 >= ne0) {
return;
}
const int row = blockDim.x*blockIdx.x + threadIdx.x;
const int i = row*ncols + col;
if (i0 >= n_dims) {
const int i = row*ne0 + i0;
dst[i + 0] = x[i + 0];
dst[i + 1] = x[i + 1];
return;
}
const int i = row*ne0 + i0;
const int i2 = row/p_delta_rows;
const int p = has_pos ? pos[i2] : 0;
const float theta_base = p*powf(freq_base, -float(col)/ncols);
const float theta_base = pos[i2]*powf(theta_scale, i0/2.0f);
float cos_theta, sin_theta;
rope_yarn(theta_base, freq_scale, corr_dims, col, ext_factor, attn_factor, &cos_theta, &sin_theta);
const float freq_factor = has_ff ? freq_factors[i0/2] : 1.0f;
float cos_theta;
float sin_theta;
rope_yarn(theta_base/freq_factor, freq_scale, corr_dims, i0, ext_factor, attn_factor, &cos_theta, &sin_theta);
const float x0 = x[i + 0];
const float x1 = x[i + 1];
@ -58,23 +68,20 @@ static __global__ void rope(
dst[i + 1] = x0*sin_theta + x1*cos_theta;
}
template<typename T, bool has_pos, bool has_freq_facs>
template<typename T, bool has_ff>
static __global__ void rope_neox(
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, const float * freq_factors
) {
const int col = 2*(blockDim.y*blockIdx.y + threadIdx.y);
const T * x, T * dst, int ne0, 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, const float * freq_factors) {
const int i0 = 2*(blockDim.y*blockIdx.y + threadIdx.y);
if (col >= ncols) {
if (i0 >= ne0) {
return;
}
const int row = blockDim.x*blockIdx.x + threadIdx.x;
const int ib = col / n_dims;
const int ic = col % n_dims;
if (ib > 0) {
const int i = row*ncols + ib*n_dims + ic;
if (i0 >= n_dims) {
const int i = row*ne0 + i0;
dst[i + 0] = x[i + 0];
dst[i + 1] = x[i + 1];
@ -82,16 +89,17 @@ static __global__ void rope_neox(
return;
}
const int i = row*ncols + ib*n_dims + ic/2;
const int i = row*ne0 + i0/2;
const int i2 = row/p_delta_rows;
const int p = has_pos ? pos[i2] : 0;
const float freq_factor = has_freq_facs ? freq_factors[ic/2] : 1.0f;
const float theta_base = pos[i2]*powf(theta_scale, i0/2.0f);
const float theta_base = p*powf(theta_scale, col/2.0f)/freq_factor;
const float freq_factor = has_ff ? freq_factors[i0/2] : 1.0f;
float cos_theta, sin_theta;
rope_yarn(theta_base, freq_scale, corr_dims, ic, ext_factor, attn_factor, &cos_theta, &sin_theta);
float cos_theta;
float sin_theta;
rope_yarn(theta_base/freq_factor, freq_scale, corr_dims, i0, ext_factor, attn_factor, &cos_theta, &sin_theta);
const float x0 = x[i + 0];
const float x1 = x[i + n_dims/2];
@ -100,144 +108,81 @@ static __global__ void rope_neox(
dst[i + n_dims/2] = x0*sin_theta + x1*cos_theta;
}
static __global__ void rope_glm_f32(
const float * x, float * dst, int ncols, const int32_t * pos, float freq_scale, int p_delta_rows, float freq_base,
int n_ctx
) {
const int col = blockDim.x*blockIdx.x + threadIdx.x;
const int half_n_dims = ncols/4;
if (col >= half_n_dims) {
return;
}
const int row = blockDim.y*blockIdx.y + threadIdx.y;
const int i = row*ncols + col;
const int i2 = row/p_delta_rows;
const float col_theta_scale = powf(freq_base, -2.0f*col/ncols);
// FIXME: this is likely wrong
const int p = pos != nullptr ? pos[i2] : 0;
const float theta = min(p, n_ctx - 2)*freq_scale*col_theta_scale;
const float sin_theta = sinf(theta);
const float cos_theta = cosf(theta);
const float x0 = x[i + 0];
const float x1 = x[i + half_n_dims];
dst[i + 0] = x0*cos_theta - x1*sin_theta;
dst[i + half_n_dims] = x0*sin_theta + x1*cos_theta;
const float block_theta = ((float)max(p - n_ctx - 2, 0))*col_theta_scale;
const float sin_block_theta = sinf(block_theta);
const float cos_block_theta = cosf(block_theta);
const float x2 = x[i + half_n_dims * 2];
const float x3 = x[i + half_n_dims * 3];
dst[i + half_n_dims * 2] = x2*cos_block_theta - x3*sin_block_theta;
dst[i + half_n_dims * 3] = x2*sin_block_theta + x3*cos_block_theta;
}
template<typename T>
static void rope_cuda(
const T * x, T * dst, int ncols, 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);
static void rope_norm_cuda(
const T * x, T * dst, int ne0, int n_dims, int nr, 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 float * freq_factors, cudaStream_t stream) {
GGML_ASSERT(ne0 % 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);
if (pos == nullptr) {
rope<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
);
const int n_blocks_x = (ne0 + 2*CUDA_ROPE_BLOCK_SIZE - 1) / (2*CUDA_ROPE_BLOCK_SIZE);
const dim3 block_nums(nr, n_blocks_x, 1);
const float theta_scale = powf(freq_base, -2.0f/n_dims);
if (freq_factors == nullptr) {
rope_norm<T, false><<<block_nums, block_dims, 0, stream>>>(
x, dst, ne0, n_dims, pos, freq_scale, p_delta_rows, ext_factor, attn_factor, corr_dims,
theta_scale, freq_factors
);
} else {
rope<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
);
rope_norm<T, true><<<block_nums, block_dims, 0, stream>>>(
x, dst, ne0, n_dims, pos, freq_scale, p_delta_rows, ext_factor, attn_factor, corr_dims,
theta_scale, freq_factors
);
}
}
template<typename T>
static void rope_neox_cuda(
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, const float * freq_factors, cudaStream_t stream
) {
GGML_ASSERT(ncols % 2 == 0);
const T * x, T * dst, int ne0, int n_dims, int nr, 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 float * freq_factors, cudaStream_t stream) {
GGML_ASSERT(ne0 % 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 int n_blocks_x = (ne0 + 2*CUDA_ROPE_BLOCK_SIZE - 1) / (2*CUDA_ROPE_BLOCK_SIZE);
const dim3 block_nums(nr, n_blocks_x, 1);
const float theta_scale = powf(freq_base, -2.0f/n_dims);
if (pos == nullptr) {
if (freq_factors == nullptr) {
rope_neox<T, false, false><<<block_nums, block_dims, 0, stream>>>(
x, dst, ncols, n_dims, pos, freq_scale, p_delta_rows, ext_factor, attn_factor, corr_dims,
if (freq_factors == nullptr) {
rope_neox<T, false><<<block_nums, block_dims, 0, stream>>>(
x, dst, ne0, n_dims, pos, freq_scale, p_delta_rows, ext_factor, attn_factor, corr_dims,
theta_scale, freq_factors
);
} else {
rope_neox<T, false, true><<<block_nums, block_dims, 0, stream>>>(
x, dst, ncols, n_dims, pos, freq_scale, p_delta_rows, ext_factor, attn_factor, corr_dims,
theta_scale, freq_factors
);
}
} else {
if (freq_factors == nullptr) {
rope_neox<T, true, false><<<block_nums, block_dims, 0, stream>>>(
x, dst, ncols, n_dims, pos, freq_scale, p_delta_rows, ext_factor, attn_factor, corr_dims,
rope_neox<T, true><<<block_nums, block_dims, 0, stream>>>(
x, dst, ne0, n_dims, pos, freq_scale, p_delta_rows, ext_factor, attn_factor, corr_dims,
theta_scale, freq_factors
);
} else {
rope_neox<T, true, true><<<block_nums, block_dims, 0, stream>>>(
x, dst, ncols, n_dims, pos, freq_scale, p_delta_rows, ext_factor, attn_factor, corr_dims,
theta_scale, freq_factors
);
}
}
}
static void rope_glm_f32_cuda(
const float * x, float * dst, int ncols, int nrows, const int32_t * pos, float freq_scale, int p_delta_rows,
float freq_base, int n_ctx, cudaStream_t stream
) {
GGML_ASSERT(ncols % 4 == 0);
const dim3 block_dims(CUDA_ROPE_BLOCK_SIZE/4, 1, 1);
const int num_blocks_x = (ncols + CUDA_ROPE_BLOCK_SIZE - 1) / CUDA_ROPE_BLOCK_SIZE;
const dim3 block_nums(num_blocks_x, nrows, 1);
rope_glm_f32<<<block_nums, block_dims, 0, stream>>>(x, dst, ncols, pos, freq_scale, p_delta_rows, freq_base, n_ctx);
static void rope_norm_cuda_f16(
const half * x, half * dst, int ne0, int n_dims, int nr, 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 float * freq_factors, cudaStream_t stream) {
rope_norm_cuda<half>(x, dst, ne0, n_dims, nr, pos, freq_scale, p_delta_rows, freq_base, ext_factor, attn_factor, corr_dims, freq_factors, stream);
}
static void rope_cuda_f16(
const half * x, half * dst, int ncols, 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) {
static void rope_norm_cuda_f32(
const float * x, float * dst, int ne0, int n_dims, int nr, 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 float * freq_factors, cudaStream_t stream) {
rope_cuda<half>(x, dst, ncols, nrows, pos, freq_scale, p_delta_rows, freq_base, ext_factor, attn_factor, corr_dims, stream);
}
static void rope_cuda_f32(
const float * x, float * dst, int ncols, 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) {
rope_cuda<float>(x, dst, ncols, nrows, pos, freq_scale, p_delta_rows, freq_base, ext_factor, attn_factor, corr_dims, stream);
rope_norm_cuda<float>(x, dst, ne0, n_dims, nr, pos, freq_scale, p_delta_rows, freq_base, ext_factor, attn_factor, corr_dims, freq_factors, stream);
}
static void rope_neox_cuda_f16(
const half * x, half * dst, int ncols, int n_dims, int nrows, const int32_t * pos, float freq_scale, int p_delta_rows,
const half * x, half * dst, int ne0, int n_dims, int nr, 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 float * freq_factors, cudaStream_t stream) {
rope_neox_cuda<half>(x, dst, ncols, n_dims, nrows, pos, freq_scale, p_delta_rows, freq_base, ext_factor, attn_factor, corr_dims, freq_factors, stream);
rope_neox_cuda<half>(x, dst, ne0, n_dims, nr, pos, freq_scale, p_delta_rows, freq_base, ext_factor, attn_factor, corr_dims, freq_factors, stream);
}
static void rope_neox_cuda_f32(
const float * x, float * dst, int ncols, int n_dims, int nrows, const int32_t * pos, float freq_scale, int p_delta_rows,
const float * x, float * dst, int ne0, int n_dims, int nr, 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 float * freq_factors, cudaStream_t stream
) {
rope_neox_cuda<float>(x, dst, ncols, n_dims, nrows, pos, freq_scale, p_delta_rows, freq_base, ext_factor, attn_factor, corr_dims, freq_factors, stream);
rope_neox_cuda<float>(x, dst, ne0, n_dims, nr, pos, freq_scale, p_delta_rows, freq_base, ext_factor, attn_factor, corr_dims, freq_factors, stream);
}
void ggml_cuda_op_rope(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
@ -258,16 +203,22 @@ void ggml_cuda_op_rope(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
const int64_t ne00 = src0->ne[0];
const int64_t ne01 = src0->ne[1];
const int64_t nrows = ggml_nrows(src0);
const int64_t nr = ggml_nrows(src0);
//const int n_past = ((int32_t *) dst->op_params)[0];
const int n_dims = ((int32_t *) dst->op_params)[1];
const int mode = ((int32_t *) dst->op_params)[2];
const int n_ctx = ((int32_t *) dst->op_params)[3];
const int n_orig_ctx = ((int32_t *) dst->op_params)[4];
//const int n_past = ((int32_t *) dst->op_params)[0];
const int n_dims = ((int32_t *) dst->op_params)[1];
const int mode = ((int32_t *) dst->op_params)[2];
//const int n_ctx = ((int32_t *) dst->op_params)[3];
const int n_ctx_orig = ((int32_t *) dst->op_params)[4];
// RoPE alteration for extended context
float freq_base, freq_scale, ext_factor, attn_factor, beta_fast, beta_slow;
float freq_base;
float freq_scale;
float ext_factor;
float attn_factor;
float beta_fast;
float beta_slow;
memcpy(&freq_base, (int32_t *) dst->op_params + 5, sizeof(float));
memcpy(&freq_scale, (int32_t *) dst->op_params + 6, sizeof(float));
memcpy(&ext_factor, (int32_t *) dst->op_params + 7, sizeof(float));
@ -275,38 +226,28 @@ void ggml_cuda_op_rope(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
memcpy(&beta_fast, (int32_t *) dst->op_params + 9, sizeof(float));
memcpy(&beta_slow, (int32_t *) dst->op_params + 10, sizeof(float));
const float * freq_factors = nullptr;
const int32_t * pos = nullptr;
const bool is_neox = mode & 2;
const bool is_glm = mode & 4;
pos = (const int32_t *) src1_d;
const int32_t * pos = (const int32_t *) src1_d;
if (is_neox) {
if (src2 != nullptr) {
freq_factors = (const float *) src2->data;
}
} else {
GGML_ASSERT(src2 == nullptr && "TODO: freq_factors not implemented for !is_neox");
const float * freq_factors = nullptr;
if (src2 != nullptr) {
freq_factors = (const float *) src2->data;
}
rope_corr_dims corr_dims;
ggml_rope_yarn_corr_dims(n_dims, n_orig_ctx, freq_base, beta_fast, beta_slow, corr_dims.v);
ggml_rope_yarn_corr_dims(n_dims, n_ctx_orig, freq_base, beta_fast, beta_slow, corr_dims.v);
// compute
if (is_glm) {
GGML_ASSERT(false);
rope_glm_f32_cuda(src0_d, dst_d, ne00, nrows, pos, freq_scale, ne01, freq_base, n_ctx, stream);
} else if (is_neox) {
if (is_neox) {
if (src0->type == GGML_TYPE_F32) {
rope_neox_cuda_f32(
(const float *)src0_d, (float *)dst_d, ne00, n_dims, nrows, pos, freq_scale, ne01, freq_base, ext_factor,
(const float *)src0_d, (float *)dst_d, ne00, n_dims, nr, pos, freq_scale, ne01, freq_base, ext_factor,
attn_factor, corr_dims, freq_factors, stream
);
} else if (src0->type == GGML_TYPE_F16) {
rope_neox_cuda_f16(
(const half *)src0_d, (half *)dst_d, ne00, n_dims, nrows, pos, freq_scale, ne01, freq_base, ext_factor,
(const half *)src0_d, (half *)dst_d, ne00, n_dims, nr, pos, freq_scale, ne01, freq_base, ext_factor,
attn_factor, corr_dims, freq_factors, stream
);
} else {
@ -314,14 +255,14 @@ void ggml_cuda_op_rope(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
}
} else {
if (src0->type == GGML_TYPE_F32) {
rope_cuda_f32(
(const float *)src0_d, (float *)dst_d, ne00, nrows, pos, freq_scale, ne01, freq_base, ext_factor,
attn_factor, corr_dims, stream
rope_norm_cuda_f32(
(const float *)src0_d, (float *)dst_d, ne00, n_dims, nr, pos, freq_scale, ne01, freq_base, ext_factor,
attn_factor, corr_dims, freq_factors, stream
);
} else if (src0->type == GGML_TYPE_F16) {
rope_cuda_f16(
(const half *)src0_d, (half *)dst_d, ne00, nrows, pos, freq_scale, ne01, freq_base, ext_factor,
attn_factor, corr_dims, stream
rope_norm_cuda_f16(
(const half *)src0_d, (half *)dst_d, ne00, n_dims, nr, pos, freq_scale, ne01, freq_base, ext_factor,
attn_factor, corr_dims, freq_factors, stream
);
} else {
GGML_ASSERT(false);

13
ggml-kompute.cpp
View File

@ -1192,7 +1192,7 @@ static void ggml_vk_rope(
const std::shared_ptr<kp::Tensor>& inB,
const std::shared_ptr<kp::Tensor>& out,
uint32_t inAOff, uint32_t inBOff, uint32_t outOff,
ggml_type src0t, int32_t n_dims, int32_t mode, int32_t n_orig_ctx,
ggml_type src0t, int32_t n_dims, int32_t mode, int32_t n_ctx_orig,
float freq_base, float freq_scale, float ext_factor, float attn_factor, float beta_fast, float beta_slow,
int32_t ne01, int32_t ne02, int32_t ne03,
uint32_t nb00, uint32_t nb01, uint32_t nb02, uint32_t nb03,
@ -1221,14 +1221,14 @@ static void ggml_vk_rope(
struct PushConstants {
uint32_t inAOff, inBOff, outOff;
int32_t n_dims, mode, n_orig_ctx;
int32_t n_dims, mode, n_ctx_orig;
float freq_base, freq_scale, ext_factor, attn_factor, beta_fast, beta_slow;
uint32_t nb00, nb01, nb02, nb03;
int32_t ne0;
uint32_t nb0, nb1, nb2, nb3;
} pushConsts {
safe_divide(inAOff, type_size), safe_divide(inBOff, 4), safe_divide(outOff, type_size),
n_dims, mode, n_orig_ctx,
n_dims, mode, n_ctx_orig,
freq_base, freq_scale, ext_factor, attn_factor, beta_fast, beta_slow,
nb00, nb01, nb02, nb03,
ne0,
@ -1692,13 +1692,16 @@ static void ggml_vk_graph_compute(struct ggml_kompute_context * ctx, struct ggml
#pragma message(" https://github.com/ggerganov/llama.cpp/pull/7225")
GGML_ASSERT(dst->src[2] == nullptr && "phi3 frequency factors not implemented yet");
#pragma message("TODO: update rope NORM mode to match NEOX mode")
#pragma message(" https://github.com/ggerganov/llama.cpp/pull/7634")
GGML_ASSERT(ne10 == ne02);
GGML_ASSERT(src0t == dstt);
// const int n_past = ((int32_t *) dst->op_params)[0];
const int n_dims = ((int32_t *) dst->op_params)[1];
const int mode = ((int32_t *) dst->op_params)[2];
// skip 3, n_ctx used in GLM RoPE, unimplemented in Vulkan
const int n_orig_ctx = ((int32_t *) dst->op_params)[4];
const int n_ctx_orig = ((int32_t *) dst->op_params)[4];
float freq_base, freq_scale, ext_factor, attn_factor, beta_fast, beta_slow;
memcpy(&freq_base, (int32_t *) dst->op_params + 5, sizeof(float));
@ -1708,7 +1711,7 @@ static void ggml_vk_graph_compute(struct ggml_kompute_context * ctx, struct ggml
memcpy(&beta_fast, (int32_t *) dst->op_params + 9, sizeof(float));
memcpy(&beta_slow, (int32_t *) dst->op_params + 10, sizeof(float));
ggml_vk_rope(
seq, id_src0, id_src1, id_dst, off_src0, off_src1, off_dst, src0t, n_dims, mode, n_orig_ctx,
seq, id_src0, id_src1, id_dst, off_src0, off_src1, off_dst, src0t, n_dims, mode, n_ctx_orig,
freq_base, freq_scale, ext_factor, attn_factor, beta_fast, beta_slow,
ne01, ne02, ne03, nb00, nb01, nb02, nb03, ne0, nb0, nb1, nb2, nb3
);

54
ggml-metal.m
View File

@ -172,8 +172,10 @@ enum ggml_metal_kernel_type {
GGML_METAL_KERNEL_TYPE_MUL_MM_ID_IQ1_M_F32,
GGML_METAL_KERNEL_TYPE_MUL_MM_ID_IQ4_NL_F32,
GGML_METAL_KERNEL_TYPE_MUL_MM_ID_IQ4_XS_F32,
GGML_METAL_KERNEL_TYPE_ROPE_F32,
GGML_METAL_KERNEL_TYPE_ROPE_F16,
GGML_METAL_KERNEL_TYPE_ROPE_NORM_F32,
GGML_METAL_KERNEL_TYPE_ROPE_NORM_F16,
GGML_METAL_KERNEL_TYPE_ROPE_NEOX_F32,
GGML_METAL_KERNEL_TYPE_ROPE_NEOX_F16,
GGML_METAL_KERNEL_TYPE_IM2COL_F16,
GGML_METAL_KERNEL_TYPE_IM2COL_F32,
GGML_METAL_KERNEL_TYPE_UPSCALE_F32,
@ -626,8 +628,10 @@ static struct ggml_metal_context * ggml_metal_init(int n_cb) {
GGML_METAL_ADD_KERNEL(GGML_METAL_KERNEL_TYPE_MUL_MM_ID_IQ1_M_F32, mul_mm_id_iq1_m_f32, ctx->support_simdgroup_mm);
GGML_METAL_ADD_KERNEL(GGML_METAL_KERNEL_TYPE_MUL_MM_ID_IQ4_NL_F32, mul_mm_id_iq4_nl_f32, ctx->support_simdgroup_mm);
GGML_METAL_ADD_KERNEL(GGML_METAL_KERNEL_TYPE_MUL_MM_ID_IQ4_XS_F32, mul_mm_id_iq4_xs_f32, ctx->support_simdgroup_mm);
GGML_METAL_ADD_KERNEL(GGML_METAL_KERNEL_TYPE_ROPE_F32, rope_f32, true);
GGML_METAL_ADD_KERNEL(GGML_METAL_KERNEL_TYPE_ROPE_F16, rope_f16, true);
GGML_METAL_ADD_KERNEL(GGML_METAL_KERNEL_TYPE_ROPE_NORM_F32, rope_norm_f32, true);
GGML_METAL_ADD_KERNEL(GGML_METAL_KERNEL_TYPE_ROPE_NORM_F16, rope_norm_f16, true);
GGML_METAL_ADD_KERNEL(GGML_METAL_KERNEL_TYPE_ROPE_NEOX_F32, rope_neox_f32, true);
GGML_METAL_ADD_KERNEL(GGML_METAL_KERNEL_TYPE_ROPE_NEOX_F16, rope_neox_f16, true);
GGML_METAL_ADD_KERNEL(GGML_METAL_KERNEL_TYPE_IM2COL_F16, im2col_f16, true);
GGML_METAL_ADD_KERNEL(GGML_METAL_KERNEL_TYPE_IM2COL_F32, im2col_f32, true);
GGML_METAL_ADD_KERNEL(GGML_METAL_KERNEL_TYPE_UPSCALE_F32, upscale_f32, true);
@ -2285,7 +2289,7 @@ static enum ggml_status ggml_metal_graph_compute(
const int n_dims = ((int32_t *) dst->op_params)[1];
const int mode = ((int32_t *) dst->op_params)[2];
// skip 3, n_ctx, used in GLM RoPE, unimplemented in metal
const int n_orig_ctx = ((int32_t *) dst->op_params)[4];
const int n_ctx_orig = ((int32_t *) dst->op_params)[4];
float freq_base;
float freq_scale;
@ -2302,21 +2306,22 @@ static enum ggml_status ggml_metal_graph_compute(
memcpy(&beta_slow, (int32_t *) dst->op_params + 10, sizeof(float));
const bool is_neox = mode & 2;
const bool is_glm = mode & 4;
GGML_ASSERT(!is_glm && "GLM RoPE not implemented in Metal");
if (!is_neox) {
GGML_ASSERT(id_src2 == nil && "TODO: freq_factors not implemented for !is_neox");
}
id<MTLComputePipelineState> pipeline = nil;
switch (src0->type) {
case GGML_TYPE_F32: pipeline = ctx->kernels[GGML_METAL_KERNEL_TYPE_ROPE_F32].pipeline; break;
case GGML_TYPE_F16: pipeline = ctx->kernels[GGML_METAL_KERNEL_TYPE_ROPE_F16].pipeline; break;
default: GGML_ASSERT(false);
};
if (!is_neox) {
switch (src0->type) {
case GGML_TYPE_F32: pipeline = ctx->kernels[GGML_METAL_KERNEL_TYPE_ROPE_NORM_F32].pipeline; break;
case GGML_TYPE_F16: pipeline = ctx->kernels[GGML_METAL_KERNEL_TYPE_ROPE_NORM_F16].pipeline; break;
default: GGML_ASSERT(false);
};
} else {
switch (src0->type) {
case GGML_TYPE_F32: pipeline = ctx->kernels[GGML_METAL_KERNEL_TYPE_ROPE_NEOX_F32].pipeline; break;
case GGML_TYPE_F16: pipeline = ctx->kernels[GGML_METAL_KERNEL_TYPE_ROPE_NEOX_F16].pipeline; break;
default: GGML_ASSERT(false);
};
}
[encoder setComputePipelineState:pipeline];
[encoder setBuffer:id_src0 offset:offs_src0 atIndex:0];
@ -2345,14 +2350,13 @@ static enum ggml_status ggml_metal_graph_compute(
[encoder setBytes:&nb3 length:sizeof(uint64_t) atIndex:19];
[encoder setBytes:&n_past length:sizeof( int) atIndex:20];
[encoder setBytes:&n_dims length:sizeof( int) atIndex:21];
[encoder setBytes:&mode length:sizeof( int) atIndex:22];
[encoder setBytes:&n_orig_ctx length:sizeof( int) atIndex:23];
[encoder setBytes:&freq_base length:sizeof( float) atIndex:24];
[encoder setBytes:&freq_scale length:sizeof( float) atIndex:25];
[encoder setBytes:&ext_factor length:sizeof( float) atIndex:26];
[encoder setBytes:&attn_factor length:sizeof( float) atIndex:27];
[encoder setBytes:&beta_fast length:sizeof( float) atIndex:28];
[encoder setBytes:&beta_slow length:sizeof( float) atIndex:29];
[encoder setBytes:&n_ctx_orig length:sizeof( int) atIndex:22];
[encoder setBytes:&freq_base length:sizeof( float) atIndex:23];
[encoder setBytes:&freq_scale length:sizeof( float) atIndex:24];
[encoder setBytes:&ext_factor length:sizeof( float) atIndex:25];
[encoder setBytes:&attn_factor length:sizeof( float) atIndex:26];
[encoder setBytes:&beta_fast length:sizeof( float) atIndex:27];
[encoder setBytes:&beta_slow length:sizeof( float) atIndex:28];
[encoder dispatchThreadgroups:MTLSizeMake(ne01, ne02, ne03) threadsPerThreadgroup:MTLSizeMake(nth, 1, 1)];
} break;

200
ggml-metal.metal
View File

@ -1654,8 +1654,7 @@ static float rope_yarn_ramp(const float low, const float high, const int i0) {
// MIT licensed. Copyright (c) 2023 Jeffrey Quesnelle and Bowen Peng.
static void rope_yarn(
float theta_extrap, float freq_scale, float corr_dims[2], int64_t i0, float ext_factor, float mscale,
thread float * cos_theta, thread float * sin_theta
) {
thread float * cos_theta, thread float * sin_theta) {
// Get n-d rotational scaling corrected for extrapolation
float theta_interp = freq_scale * theta_extrap;
float theta = theta_interp;
@ -1672,55 +1671,20 @@ static void rope_yarn(
// Apparently solving `n_rot = 2pi * x * base^((2 * max_pos_emb) / n_dims)` for x, we get
// `corr_fac(n_rot) = n_dims * log(max_pos_emb / (n_rot * 2pi)) / (2 * log(base))`
static float rope_yarn_corr_factor(int n_dims, int n_orig_ctx, float n_rot, float base) {
return n_dims * log(n_orig_ctx / (n_rot * 2 * M_PI_F)) / (2 * log(base));
static float rope_yarn_corr_factor(int n_dims, int n_ctx_orig, float n_rot, float base) {
return n_dims * log(n_ctx_orig / (n_rot * 2 * M_PI_F)) / (2 * log(base));
}
static void rope_yarn_corr_dims(
int n_dims, int n_orig_ctx, float freq_base, float beta_fast, float beta_slow, float dims[2]
int n_dims, int n_ctx_orig, float freq_base, float beta_fast, float beta_slow, float dims[2]
) {
// start and end correction dims
dims[0] = max(0.0f, floor(rope_yarn_corr_factor(n_dims, n_orig_ctx, beta_fast, freq_base)));
dims[1] = min(n_dims - 1.0f, ceil(rope_yarn_corr_factor(n_dims, n_orig_ctx, beta_slow, freq_base)));
dims[0] = max(0.0f, floor(rope_yarn_corr_factor(n_dims, n_ctx_orig, beta_fast, freq_base)));
dims[1] = min(n_dims - 1.0f, ceil(rope_yarn_corr_factor(n_dims, n_ctx_orig, beta_slow, freq_base)));
}
typedef void (rope_t)(
device const void * src0,
device const int32_t * src1,
device const float * src2,
device float * dst,
constant int64_t & ne00,
constant int64_t & ne01,
constant int64_t & ne02,
constant int64_t & ne03,
constant uint64_t & nb00,
constant uint64_t & nb01,
constant uint64_t & nb02,
constant uint64_t & nb03,
constant int64_t & ne0,
constant int64_t & ne1,
constant int64_t & ne2,
constant int64_t & ne3,
constant uint64_t & nb0,
constant uint64_t & nb1,
constant uint64_t & nb2,
constant uint64_t & nb3,
constant int & n_past,
constant int & n_dims,
constant int & mode,
constant int & n_orig_ctx,
constant float & freq_base,
constant float & freq_scale,
constant float & ext_factor,
constant float & attn_factor,
constant float & beta_fast,
constant float & beta_slow,
uint tiitg[[thread_index_in_threadgroup]],
uint3 tptg[[threads_per_threadgroup]],
uint3 tgpig[[threadgroup_position_in_grid]]);
template<typename T>
kernel void kernel_rope(
kernel void kernel_rope_norm(
device const void * src0,
device const int32_t * src1,
device const float * src2,
@ -1743,8 +1707,7 @@ kernel void kernel_rope(
constant uint64_t & nb3,
constant int & n_past,
constant int & n_dims,
constant int & mode,
constant int & n_orig_ctx,
constant int & n_ctx_orig,
constant float & freq_base,
constant float & freq_scale,
constant float & ext_factor,
@ -1758,69 +1721,130 @@ kernel void kernel_rope(
const int64_t i2 = tgpig[1];
const int64_t i1 = tgpig[0];
const bool is_neox = mode & 2;
float corr_dims[2];
rope_yarn_corr_dims(n_dims, n_orig_ctx, freq_base, beta_fast, beta_slow, corr_dims);
rope_yarn_corr_dims(n_dims, n_ctx_orig, freq_base, beta_fast, beta_slow, corr_dims);
device const int32_t * pos = src1;
const int64_t p = pos[i2];
const float theta_base = (float)p;
const float theta_base = (float) pos[i2];
const float inv_ndims = -1.f/n_dims;
if (!is_neox) {
for (int64_t i0 = 2*tiitg; i0 < ne0; i0 += 2*tptg.x) {
float cos_theta;
float sin_theta;
for (int64_t i0 = 2*tiitg; i0 < ne0; i0 += 2*tptg.x) {
if (i0 < n_dims) {
const int64_t ic = i0/2;
const float theta = theta_base * pow(freq_base, inv_ndims*i0);
float cos_theta, sin_theta;
rope_yarn(theta, freq_scale, corr_dims, i0, ext_factor, attn_factor, &cos_theta, &sin_theta);
const float freq_factor = src2 != src0 ? src2[ic] : 1.0f;
rope_yarn(theta/freq_factor, freq_scale, corr_dims, i0, ext_factor, attn_factor, &cos_theta, &sin_theta);
device const T * const src = (device T *)((device char *) src0 + i3*nb03 + i2*nb02 + i1*nb01 + i0*nb00);
device T * dst_data = (device T *)((device char *) dst + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
const T x0 = src[0];
const T x1 = src[1];
const float x0 = src[0];
const float x1 = src[1];
dst_data[0] = x0*cos_theta - x1*sin_theta;
dst_data[1] = x0*sin_theta + x1*cos_theta;
}
} else {
for (int64_t ic = 2*tiitg; ic < ne0; ic += 2*tptg.x) {
if (ic < n_dims) {
const int64_t i0 = ic/2;
} else {
device const T * const src = (device T *)((device char *) src0 + i3*nb03 + i2*nb02 + i1*nb01 + i0*nb00);
device T * dst_data = (device T *)((device char *) dst + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
const float freq_factor = src2 != src0 ? src2[i0] : 1.0f;
const float theta = theta_base * pow(freq_base, inv_ndims*ic);
float cos_theta, sin_theta;
rope_yarn(theta/freq_factor, freq_scale, corr_dims, ic, ext_factor, attn_factor, &cos_theta, &sin_theta);
device const T * const src = (device T *)((device char *) src0 + i3*nb03 + i2*nb02 + i1*nb01 + i0*nb00);
device T * dst_data = (device T *)((device char *) dst + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
const float x0 = src[0];
const float x1 = src[n_dims/2];
dst_data[0] = x0*cos_theta - x1*sin_theta;
dst_data[n_dims/2] = x0*sin_theta + x1*cos_theta;
} else {
const int64_t i0 = ic;
device const T * const src = (device T *)((device char *) src0 + i3*nb03 + i2*nb02 + i1*nb01 + i0*nb00);
device T * dst_data = (device T *)((device char *) dst + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
dst_data[0] = src[0];
dst_data[1] = src[1];
}
dst_data[0] = src[0];
dst_data[1] = src[1];
}
}
}
template [[host_name("kernel_rope_f32")]] kernel rope_t kernel_rope<float>;
template [[host_name("kernel_rope_f16")]] kernel rope_t kernel_rope<half>;
template<typename T>
kernel void kernel_rope_neox(
device const void * src0,
device const int32_t * src1,
device const float * src2,
device float * dst,
constant int64_t & ne00,
constant int64_t & ne01,
constant int64_t & ne02,
constant int64_t & ne03,
constant uint64_t & nb00,
constant uint64_t & nb01,
constant uint64_t & nb02,
constant uint64_t & nb03,
constant int64_t & ne0,
constant int64_t & ne1,
constant int64_t & ne2,
constant int64_t & ne3,
constant uint64_t & nb0,
constant uint64_t & nb1,
constant uint64_t & nb2,
constant uint64_t & nb3,
constant int & n_past,
constant int & n_dims,
constant int & n_ctx_orig,
constant float & freq_base,
constant float & freq_scale,
constant float & ext_factor,
constant float & attn_factor,
constant float & beta_fast,
constant float & beta_slow,
uint tiitg[[thread_index_in_threadgroup]],
uint3 tptg[[threads_per_threadgroup]],
uint3 tgpig[[threadgroup_position_in_grid]]) {
const int64_t i3 = tgpig[2];
const int64_t i2 = tgpig[1];
const int64_t i1 = tgpig[0];
float corr_dims[2];
rope_yarn_corr_dims(n_dims, n_ctx_orig, freq_base, beta_fast, beta_slow, corr_dims);
device const int32_t * pos = src1;
const float theta_base = (float) pos[i2];
const float inv_ndims = -1.f/n_dims;
float cos_theta;
float sin_theta;
for (int64_t i0 = 2*tiitg; i0 < ne0; i0 += 2*tptg.x) {
if (i0 < n_dims) {
const int64_t ic = i0/2;
const float theta = theta_base * pow(freq_base, inv_ndims*i0);
const float freq_factor = src2 != src0 ? src2[ic] : 1.0f;
rope_yarn(theta/freq_factor, freq_scale, corr_dims, i0, ext_factor, attn_factor, &cos_theta, &sin_theta);
device const T * const src = (device T *)((device char *) src0 + i3*nb03 + i2*nb02 + i1*nb01 + ic*nb00);
device T * dst_data = (device T *)((device char *) dst + i3*nb3 + i2*nb2 + i1*nb1 + ic*nb0);
const float x0 = src[0];
const float x1 = src[n_dims/2];
dst_data[0] = x0*cos_theta - x1*sin_theta;
dst_data[n_dims/2] = x0*sin_theta + x1*cos_theta;
} else {
device const T * const src = (device T *)((device char *) src0 + i3*nb03 + i2*nb02 + i1*nb01 + i0*nb00);
device T * dst_data = (device T *)((device char *) dst + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
dst_data[0] = src[0];
dst_data[1] = src[1];
}
}
}
typedef decltype(kernel_rope_norm<float>) kernel_rope_norm_t;
typedef decltype(kernel_rope_neox<float>) kernel_rope_neox_t;
template [[host_name("kernel_rope_norm_f32")]] kernel kernel_rope_norm_t kernel_rope_norm<float>;
template [[host_name("kernel_rope_norm_f16")]] kernel kernel_rope_norm_t kernel_rope_norm<half>;
template [[host_name("kernel_rope_neox_f32")]] kernel kernel_rope_neox_t kernel_rope_neox<float>;
template [[host_name("kernel_rope_neox_f16")]] kernel kernel_rope_neox_t kernel_rope_neox<half>;
typedef void (im2col_t)(
device const float * x,

74
ggml-sycl.cpp
View File

@ -8928,49 +8928,6 @@ static void rope_neox(
dst[i + n_dims/2] = x0*sin_theta + x1*cos_theta;
}
static void rope_glm_f32(
const float * x, float * dst, int ncols, const int32_t * pos, float freq_scale, int p_delta_rows, float freq_base,
int n_ctx
, const sycl::nd_item<3> &item_ct1) {
const int col = item_ct1.get_local_range(2) * item_ct1.get_group(2) +
item_ct1.get_local_id(2);
const int half_n_dims = ncols/4;
if (col >= half_n_dims) {
return;
}
const int row = item_ct1.get_local_range(1) * item_ct1.get_group(1) +
item_ct1.get_local_id(1);
const int i = row*ncols + col;
const int i2 = row/p_delta_rows;
const float col_theta_scale = dpct::pow(freq_base, -2.0f * col / ncols);
// FIXME: this is likely wrong
const int p = pos != nullptr ? pos[i2] : 0;
const float theta = sycl::min(p, n_ctx - 2) * freq_scale * col_theta_scale;
const float sin_theta = sycl::sin((float)theta);
const float cos_theta = sycl::cos((float)theta);
const float x0 = x[i + 0];
const float x1 = x[i + half_n_dims];
dst[i + 0] = x0*cos_theta - x1*sin_theta;
dst[i + half_n_dims] = x0*sin_theta + x1*cos_theta;
const float block_theta =
((float)sycl::max(p - n_ctx - 2, 0)) * col_theta_scale;
const float sin_block_theta = sycl::sin((float)block_theta);
const float cos_block_theta = sycl::cos((float)block_theta);
const float x2 = x[i + half_n_dims * 2];
const float x3 = x[i + half_n_dims * 3];
dst[i + half_n_dims * 2] = x2*cos_block_theta - x3*sin_block_theta;
dst[i + half_n_dims * 3] = x2*sin_block_theta + x3*cos_block_theta;
}
static void k_sum_rows_f32(const float * x, float * dst, const int ncols,
const sycl::nd_item<3> &item_ct1) {
const int row = item_ct1.get_group(1);
@ -12520,22 +12477,6 @@ static void rope_neox_sycl(const T *x, T *dst, int ncols, int n_dims, int nrows,
}
}
static void rope_glm_f32_sycl(const float *x, float *dst, int ncols, int nrows,
const int32_t *pos, float freq_scale,
int p_delta_rows, float freq_base, int n_ctx,
dpct::queue_ptr stream) {
GGML_ASSERT(ncols % 4 == 0);
const sycl::range<3> block_dims(1, 1, SYCL_ROPE_BLOCK_SIZE / 4);
const int num_blocks_x = (ncols + SYCL_ROPE_BLOCK_SIZE - 1) / SYCL_ROPE_BLOCK_SIZE;
const sycl::range<3> block_nums(1, nrows, num_blocks_x);
stream->parallel_for(sycl::nd_range<3>(block_nums * block_dims, block_dims),
[=](sycl::nd_item<3> item_ct1) {
rope_glm_f32(x, dst, ncols, pos, freq_scale,
p_delta_rows, freq_base, n_ctx,
item_ct1);
});
}
static void sum_rows_f32_sycl(const float *x, float *dst, const int ncols,
const int nrows, dpct::queue_ptr stream) {
const sycl::range<3> block_dims(1, 1, WARP_SIZE);
@ -14066,8 +14007,8 @@ inline void ggml_sycl_op_rope(const ggml_tensor *src0, const ggml_tensor *src1,
//const int n_past = ((int32_t *) dst->op_params)[0];
const int n_dims = ((int32_t *) dst->op_params)[1];
const int mode = ((int32_t *) dst->op_params)[2];
const int n_ctx = ((int32_t *) dst->op_params)[3];
const int n_orig_ctx = ((int32_t *) dst->op_params)[4];
//const int n_ctx = ((int32_t *) dst->op_params)[3];
const int n_ctx_orig = ((int32_t *) dst->op_params)[4];
// RoPE alteration for extended context
float freq_base, freq_scale, ext_factor, attn_factor, beta_fast, beta_slow;
@ -14087,7 +14028,9 @@ inline void ggml_sycl_op_rope(const ggml_tensor *src0, const ggml_tensor *src1,
}
const bool is_neox = mode & 2;
const bool is_glm = mode & 4;
#pragma message("TODO: update rope NORM mode to match NEOX mode")
#pragma message(" https://github.com/ggerganov/llama.cpp/pull/7634")
if (is_neox) {
pos = (const int32_t *) src1_dd;
@ -14100,13 +14043,10 @@ inline void ggml_sycl_op_rope(const ggml_tensor *src0, const ggml_tensor *src1,
}
rope_corr_dims corr_dims;
ggml_rope_yarn_corr_dims(n_dims, n_orig_ctx, freq_base, beta_fast, beta_slow, corr_dims.v);
ggml_rope_yarn_corr_dims(n_dims, n_ctx_orig, freq_base, beta_fast, beta_slow, corr_dims.v);
// compute
if (is_glm) {
GGML_ASSERT(false);
rope_glm_f32_sycl(src0_dd, dst_dd, ne00, nrows, pos, freq_scale, ne01, freq_base, n_ctx, main_stream);
} else if (is_neox) {
if (is_neox) {
if (src0->type == GGML_TYPE_F32) {
rope_neox_sycl(
(const float *)src0_dd, (float *)dst_dd, ne00, n_dims, nrows, pos, freq_scale, ne01, freq_base, ext_factor,

22
ggml-vulkan.cpp
View File

@ -3898,11 +3898,6 @@ static vk_pipeline ggml_vk_op_get_pipeline(ggml_backend_vk_context * ctx, const
{
const int mode = ((const int32_t *) dst->op_params)[2];
const bool is_neox = mode & 2;
const bool is_glm = mode & 4;
if (is_glm) {
return nullptr;
}
if (is_neox) {
if (src0->type == GGML_TYPE_F32 && dst->type == GGML_TYPE_F32) {
@ -4401,7 +4396,7 @@ static void ggml_vk_rope(ggml_backend_vk_context * ctx, vk_context * subctx, con
const int n_dims = ((int32_t *) dst->op_params)[1];
const int mode = ((int32_t *) dst->op_params)[2];
// const int n_ctx = ((int32_t *) dst->op_params)[3];
const int n_orig_ctx = ((int32_t *) dst->op_params)[4];
const int n_ctx_orig = ((int32_t *) dst->op_params)[4];
const float freq_base = ((float *) dst->op_params)[5];
const float freq_scale = ((float *) dst->op_params)[6];
const float ext_factor = ((float *) dst->op_params)[7];
@ -4410,12 +4405,12 @@ static void ggml_vk_rope(ggml_backend_vk_context * ctx, vk_context * subctx, con
const float beta_slow = ((float *) dst->op_params)[10];
const bool is_neox = mode & 2;
const bool is_glm = mode & 4;
GGML_ASSERT(!is_glm);
#pragma message("TODO: update rope NORM mode to match NEOX mode")
#pragma message(" https://github.com/ggerganov/llama.cpp/pull/7634")
float corr_dims[2];
ggml_rope_yarn_corr_dims(n_dims, n_orig_ctx, freq_base, beta_fast, beta_slow, corr_dims);
ggml_rope_yarn_corr_dims(n_dims, n_ctx_orig, freq_base, beta_fast, beta_slow, corr_dims);
if (is_neox) {
const float theta_scale = powf(freq_base, -2.0f/n_dims);
@ -6485,9 +6480,8 @@ GGML_CALL static bool ggml_backend_vk_supports_op(ggml_backend_t backend, const
case GGML_OP_ROPE:
{
const int mode = ((const int32_t *) op->op_params)[2];
const bool is_glm = mode & 4;
return !is_glm;
return true;
} break;
case GGML_OP_NONE:
case GGML_OP_RESHAPE:
@ -6992,15 +6986,15 @@ static void ggml_vk_check_results_0(ggml_backend_vk_context * ctx, ggml_compute_
} else if (tensor->op == GGML_OP_ROPE) {
const int n_dims = ((int32_t *) tensor->op_params)[1];
const int mode = ((int32_t *) tensor->op_params)[2];
const int n_ggml_ctx = ((int32_t *) tensor->op_params)[3];
const int n_orig_ggml_ctx = ((int32_t *) tensor->op_params)[4];
//const int n_ctx_ggml = ((int32_t *) tensor->op_params)[3];
const int n_ctx_orig_ggml = ((int32_t *) tensor->op_params)[4];
float freq_base = ((float *) tensor->op_params)[5];
float freq_scale = ((float *) tensor->op_params)[6];
float ext_factor = ((float *) tensor->op_params)[7];
float attn_factor = ((float *) tensor->op_params)[8];
float beta_fast = ((float *) tensor->op_params)[9];
float beta_slow = ((float *) tensor->op_params)[10];
tensor_clone = ggml_rope_ext(ggml_ctx, src0_clone, src1_clone, src2_clone, n_dims, mode, n_ggml_ctx, n_orig_ggml_ctx, freq_base, freq_scale, ext_factor, attn_factor, beta_fast, beta_slow);
tensor_clone = ggml_rope_ext(ggml_ctx, src0_clone, src1_clone, src2_clone, n_dims, mode, n_ctx_orig_ggml, freq_base, freq_scale, ext_factor, attn_factor, beta_fast, beta_slow);
} else if (tensor->op == GGML_OP_UNARY) {
switch (ggml_get_unary_op(tensor)) {
case GGML_UNARY_OP_SILU:

334
ggml.c
View File

@ -6250,16 +6250,13 @@ static struct ggml_tensor * ggml_rope_impl(
struct ggml_tensor * c,
int n_dims,
int mode,
int n_ctx,
int n_orig_ctx,
int n_ctx_orig,
float freq_base,
float freq_scale,
float ext_factor,
float attn_factor,
float beta_fast,
float beta_slow,
float xpos_base,
bool xpos_down,
bool inplace) {
GGML_ASSERT((mode & 1) == 0 && "mode & 1 == 1 is no longer supported");
@ -6280,15 +6277,13 @@ static struct ggml_tensor * ggml_rope_impl(
struct ggml_tensor * result = inplace ? ggml_view_tensor(ctx, a) : ggml_dup_tensor(ctx, a);
int32_t params[13] = { /*n_past*/ 0, n_dims, mode, n_ctx, n_orig_ctx };
int32_t params[11] = { /*n_past*/ 0, n_dims, mode, /*n_ctx*/ 0, n_ctx_orig };
memcpy(params + 5, &freq_base, sizeof(float));
memcpy(params + 6, &freq_scale, sizeof(float));
memcpy(params + 7, &ext_factor, sizeof(float));
memcpy(params + 8, &attn_factor, sizeof(float));
memcpy(params + 9, &beta_fast, sizeof(float));
memcpy(params + 10, &beta_slow, sizeof(float));
memcpy(params + 11, &xpos_base, sizeof(float));
memcpy(params + 12, &xpos_down, sizeof(bool));
ggml_set_op_params(result, params, sizeof(params));
result->op = GGML_OP_ROPE;
@ -6305,10 +6300,9 @@ struct ggml_tensor * ggml_rope(
struct ggml_tensor * a,
struct ggml_tensor * b,
int n_dims,
int mode,
int n_ctx) {
int mode) {
return ggml_rope_impl(
ctx, a, b, NULL, n_dims, mode, n_ctx, 0, 10000.0f, 1.0f, 0.0f, 1.0f, 0.0f, 0.0f, 0.0f, false, false
ctx, a, b, NULL, n_dims, mode, 0, 10000.0f, 1.0f, 0.0f, 1.0f, 0.0f, 0.0f, false
);
}
@ -6317,10 +6311,9 @@ struct ggml_tensor * ggml_rope_inplace(
struct ggml_tensor * a,
struct ggml_tensor * b,
int n_dims,
int mode,
int n_ctx) {
int mode) {
return ggml_rope_impl(
ctx, a, b, NULL, n_dims, mode, n_ctx, 0, 10000.0f, 1.0f, 0.0f, 1.0f, 0.0f, 0.0f, 0.0f, false, true
ctx, a, b, NULL, n_dims, mode, 0, 10000.0f, 1.0f, 0.0f, 1.0f, 0.0f, 0.0f, true
);
}
@ -6331,8 +6324,7 @@ struct ggml_tensor * ggml_rope_ext(
struct ggml_tensor * c,
int n_dims,
int mode,
int n_ctx,
int n_orig_ctx,
int n_ctx_orig,
float freq_base,
float freq_scale,
float ext_factor,
@ -6340,8 +6332,8 @@ struct ggml_tensor * ggml_rope_ext(
float beta_fast,
float beta_slow) {
return ggml_rope_impl(
ctx, a, b, c, n_dims, mode, n_ctx, n_orig_ctx, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow, 0.0f, false, false
ctx, a, b, c, n_dims, mode, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow, false
);
}
@ -6352,8 +6344,7 @@ struct ggml_tensor * ggml_rope_ext_inplace(
struct ggml_tensor * c,
int n_dims,
int mode,
int n_ctx,
int n_orig_ctx,
int n_ctx_orig,
float freq_base,
float freq_scale,
float ext_factor,
@ -6361,8 +6352,8 @@ struct ggml_tensor * ggml_rope_ext_inplace(
float beta_fast,
float beta_slow) {
return ggml_rope_impl(
ctx, a, b, c, n_dims, mode, n_ctx, n_orig_ctx, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow, 0.0f, false, true
ctx, a, b, c, n_dims, mode, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow, true
);
}
@ -6372,8 +6363,7 @@ struct ggml_tensor * ggml_rope_custom(
struct ggml_tensor * b,
int n_dims,
int mode,
int n_ctx,
int n_orig_ctx,
int n_ctx_orig,
float freq_base,
float freq_scale,
float ext_factor,
@ -6381,8 +6371,8 @@ struct ggml_tensor * ggml_rope_custom(
float beta_fast,
float beta_slow) {
return ggml_rope_impl(
ctx, a, b, NULL, n_dims, mode, n_ctx, n_orig_ctx, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow, 0.0f, false, false
ctx, a, b, NULL, n_dims, mode, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow, false
);
}
@ -6392,8 +6382,7 @@ struct ggml_tensor * ggml_rope_custom_inplace(
struct ggml_tensor * b,
int n_dims,
int mode,
int n_ctx,
int n_orig_ctx,
int n_ctx_orig,
float freq_base,
float freq_scale,
float ext_factor,
@ -6401,21 +6390,11 @@ struct ggml_tensor * ggml_rope_custom_inplace(
float beta_fast,
float beta_slow) {
return ggml_rope_impl(
ctx, a, b, NULL, n_dims, mode, n_ctx, n_orig_ctx, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow, 0.0f, false, true
ctx, a, b, NULL, n_dims, mode, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow, true
);
}
struct ggml_tensor * ggml_rope_xpos_inplace(
struct ggml_context * ctx,
struct ggml_tensor * a,
struct ggml_tensor * b,
int n_dims,
float base,
bool down) {
return ggml_rope_impl(ctx, a, b, NULL, n_dims, 0, 0, 0, 10000.0f, 1.0f, 0.0f, 1.0f, 0.0f, 0.0f, base, down, true);
}
// ggml_rope_back
struct ggml_tensor * ggml_rope_back(
@ -6425,16 +6404,13 @@ struct ggml_tensor * ggml_rope_back(
struct ggml_tensor * c,
int n_dims,
int mode,
int n_ctx,
int n_orig_ctx,
int n_ctx_orig,
float freq_base,
float freq_scale,
float ext_factor,
float attn_factor,
float beta_fast,
float beta_slow,
float xpos_base,
bool xpos_down) {
float beta_slow) {
GGML_ASSERT(ggml_is_vector(b));
GGML_ASSERT(b->type == GGML_TYPE_I32);
GGML_ASSERT(a->ne[2] == b->ne[0]);
@ -6450,15 +6426,13 @@ struct ggml_tensor * ggml_rope_back(
struct ggml_tensor * result = ggml_dup_tensor(ctx, a);
int32_t params[13] = { /*n_past*/ 0, n_dims, mode, n_ctx, n_orig_ctx };
int32_t params[11] = { /*n_past*/ 0, n_dims, mode, /*n_ctx*/ 0, n_ctx_orig };
memcpy(params + 5, &freq_base, sizeof(float));
memcpy(params + 6, &freq_scale, sizeof(float));
memcpy(params + 7, &ext_factor, sizeof(float));
memcpy(params + 8, &attn_factor, sizeof(float));
memcpy(params + 9, &beta_fast, sizeof(float));
memcpy(params + 10, &beta_slow, sizeof(float));
memcpy(params + 11, &xpos_base, sizeof(float));
memcpy(params + 12, &xpos_down, sizeof(bool));
ggml_set_op_params(result, params, sizeof(params));
result->op = GGML_OP_ROPE_BACK;
@ -14227,8 +14201,7 @@ static float rope_yarn_ramp(const float low, const float high, const int i0) {
// MIT licensed. Copyright (c) 2023 Jeffrey Quesnelle and Bowen Peng.
static void rope_yarn(
float theta_extrap, float freq_scale, float corr_dims[2], int64_t i0, float ext_factor, float mscale,
float * cos_theta, float * sin_theta
) {
float * cos_theta, float * sin_theta) {
// Get n-d rotational scaling corrected for extrapolation
float theta_interp = freq_scale * theta_extrap;
float theta = theta_interp;
@ -14245,18 +14218,19 @@ static void rope_yarn(
// Apparently solving `n_rot = 2pi * x * base^((2 * max_pos_emb) / n_dims)` for x, we get
// `corr_dim(n_rot) = n_dims * log(max_pos_emb / (n_rot * 2pi)) / (2 * log(base))`
static float ggml_rope_yarn_corr_dim(int n_dims, int n_orig_ctx, float n_rot, float base) {
return n_dims * logf(n_orig_ctx / (n_rot * 2 * (float)M_PI)) / (2 * logf(base));
static float ggml_rope_yarn_corr_dim(int n_dims, int n_ctx_orig, float n_rot, float base) {
return n_dims * logf(n_ctx_orig / (n_rot * 2 * (float)M_PI)) / (2 * logf(base));
}
static void ggml_rope_cache_init(
float theta_base, float freq_scale, float corr_dims[2], int64_t ne0, float ext_factor, float mscale,
float * cache, float sin_sign, float theta_scale
) {
float theta_base, float freq_scale, const float * freq_factors, float corr_dims[2], int64_t ne0, float ext_factor, float mscale,
float * cache, float sin_sign, float theta_scale) {
// ref: https://github.com/jquesnelle/yarn/blob/master/scaled_rope/LlamaYaRNScaledRotaryEmbedding.py
float theta = theta_base;
for (int64_t i0 = 0; i0 < ne0; i0 += 2) {
const float ff = freq_factors ? freq_factors[i0/2] : 1.0f;
rope_yarn(
theta, freq_scale, corr_dims, i0, ext_factor, mscale, &cache[i0 + 0], &cache[i0 + 1]
theta/ff, freq_scale, corr_dims, i0, ext_factor, mscale, &cache[i0 + 0], &cache[i0 + 1]
);
cache[i0 + 1] *= sin_sign;
@ -14265,11 +14239,11 @@ static void ggml_rope_cache_init(
}
GGML_CALL void ggml_rope_yarn_corr_dims(
int n_dims, int n_orig_ctx, float freq_base, float beta_fast, float beta_slow, float dims[2]
int n_dims, int n_ctx_orig, float freq_base, float beta_fast, float beta_slow, float dims[2]
) {
// start and end correction dims
float start = floorf(ggml_rope_yarn_corr_dim(n_dims, n_orig_ctx, beta_fast, freq_base));
float end = ceilf(ggml_rope_yarn_corr_dim(n_dims, n_orig_ctx, beta_slow, freq_base));
float start = floorf(ggml_rope_yarn_corr_dim(n_dims, n_ctx_orig, beta_fast, freq_base));
float end = ceilf(ggml_rope_yarn_corr_dim(n_dims, n_ctx_orig, beta_slow, freq_base));
dims[0] = MAX(0, start);
dims[1] = MIN(n_dims - 1, end);
}
@ -14289,15 +14263,11 @@ static void ggml_compute_forward_rope_f32(
float freq_base, freq_scale, ext_factor, attn_factor, beta_fast, beta_slow;
// these two only relevant for xPos RoPE:
float xpos_base;
bool xpos_down;
//const int n_past = ((int32_t *) dst->op_params)[0];
const int n_dims = ((int32_t *) dst->op_params)[1];
const int mode = ((int32_t *) dst->op_params)[2];
const int n_ctx = ((int32_t *) dst->op_params)[3];
const int n_orig_ctx = ((int32_t *) dst->op_params)[4];
//const int n_ctx = ((int32_t *) dst->op_params)[3];
const int n_ctx_orig = ((int32_t *) dst->op_params)[4];
memcpy(&freq_base, (int32_t *) dst->op_params + 5, sizeof(float));
memcpy(&freq_scale, (int32_t *) dst->op_params + 6, sizeof(float));
@ -14305,8 +14275,6 @@ static void ggml_compute_forward_rope_f32(
memcpy(&attn_factor, (int32_t *) dst->op_params + 8, sizeof(float));
memcpy(&beta_fast, (int32_t *) dst->op_params + 9, sizeof(float));
memcpy(&beta_slow, (int32_t *) dst->op_params + 10, sizeof(float));
memcpy(&xpos_base, (int32_t *) dst->op_params + 11, sizeof(float));
memcpy(&xpos_down, (int32_t *) dst->op_params + 12, sizeof(bool));
GGML_TENSOR_UNARY_OP_LOCALS
@ -14336,20 +14304,15 @@ static void ggml_compute_forward_rope_f32(
const float theta_scale = powf(freq_base, -2.0f/n_dims);
float corr_dims[2];
ggml_rope_yarn_corr_dims(n_dims, n_orig_ctx, freq_base, beta_fast, beta_slow, corr_dims);
ggml_rope_yarn_corr_dims(n_dims, n_ctx_orig, freq_base, beta_fast, beta_slow, corr_dims);
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");
if (src2 != NULL) {
GGML_ASSERT(src2->type == GGML_TYPE_F32);
GGML_ASSERT(src2->ne[0] >= n_dims / 2);
freq_factors = (const float *) src2->data;
}
// backward process uses inverse rotation by cos and sin.
@ -14364,95 +14327,51 @@ static void ggml_compute_forward_rope_f32(
const int64_t p = pos[i2];
float * cache = (float *) params->wdata + (ne0 + CACHE_LINE_SIZE_F32)*ith;
if (!is_glm && !is_neox) { // TODO: cache sin/cos for glm, neox
ggml_rope_cache_init(p, freq_scale, corr_dims, ne0, ext_factor, attn_factor, cache, sin_sign, theta_scale);
}
ggml_rope_cache_init(p, freq_scale, freq_factors, corr_dims, ne0, ext_factor, attn_factor, cache, sin_sign, theta_scale);
for (int64_t i1 = 0; i1 < ne1; i1++) {
if (ir++ < ir0) continue;
if (ir > ir1) break;
float theta_base = (float)p;
if (is_glm) {
theta_base = MIN(p, n_ctx - 2);
float block_theta = MAX(p - (n_ctx - 2), 0);
for (int64_t i0 = 0; i0 < ne0 / 4; i0++) {
const float cos_theta = cosf(theta_base);
const float sin_theta = sinf(theta_base) * sin_sign;
const float cos_block_theta = cosf(block_theta);
const float sin_block_theta = sinf(block_theta) * sin_sign;
theta_base *= theta_scale;
block_theta *= theta_scale;
const float * const src = (float *)((char *) src0->data + i3*nb03 + i2*nb02 + i1*nb01 + i0*nb00);
float * dst_data = (float *)((char *) dst->data + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
const float x0 = src[0];
const float x1 = src[n_dims/2];
const float x2 = src[n_dims];
const float x3 = src[n_dims/2*3];
dst_data[0] = x0*cos_theta - x1*sin_theta;
dst_data[n_dims/2] = x0*sin_theta + x1*cos_theta;
dst_data[n_dims] = x2*cos_block_theta - x3*sin_block_theta;
dst_data[n_dims/2*3] = x2*sin_block_theta + x3*cos_block_theta;
}
} else if (!is_neox) {
for (int64_t i0 = 0; i0 < ne0; i0 += 2) {
if (!is_neox) {
for (int64_t i0 = 0; i0 < n_dims; i0 += 2) {
const float cos_theta = cache[i0 + 0];
const float sin_theta = cache[i0 + 1];
// zeta scaling for xPos only:
float zeta = xpos_base != 0.0f ? powf((i0 + 0.4f * ne0) / (1.4f * ne0), p / xpos_base) : 1.0f;
if (xpos_down) zeta = 1.0f / zeta;
const float * const src = (float *)((char *) src0->data + i3*nb03 + i2*nb02 + i1*nb01 + i0*nb00);
float * dst_data = (float *)((char *) dst->data + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
const float x0 = src[0];
const float x1 = src[1];
dst_data[0] = x0*cos_theta*zeta - x1*sin_theta*zeta;
dst_data[1] = x0*sin_theta*zeta + x1*cos_theta*zeta;
dst_data[0] = x0*cos_theta - x1*sin_theta;
dst_data[1] = x0*sin_theta + x1*cos_theta;
}
} else {
// ref: https://github.com/jquesnelle/yarn/blob/master/scaled_rope/LlamaYaRNScaledRotaryEmbedding.py
for (int64_t ic = 0; ic < ne0; ic += 2) {
if (ic < n_dims) {
const int64_t i0 = ic/2;
for (int64_t i0 = 0; i0 < n_dims; i0 += 2) {
const int64_t ic = i0/2;
const float freq_factor = freq_factors ? freq_factors[i0] : 1.0f;
const float cos_theta = cache[i0 + 0];
const float sin_theta = cache[i0 + 1];
float cos_theta, sin_theta;
rope_yarn(
theta_base/freq_factor, freq_scale, corr_dims, ic, ext_factor, attn_factor,
&cos_theta, &sin_theta
);
const float * const src = (float *)((char *) src0->data + i3*nb03 + i2*nb02 + i1*nb01 + ic*nb00);
float * dst_data = (float *)((char *) dst->data + i3*nb3 + i2*nb2 + i1*nb1 + ic*nb0);
sin_theta *= sin_sign;
theta_base *= theta_scale;
const float x0 = src[0];
const float x1 = src[n_dims/2];
const float * const src = (float *)((char *) src0->data + i3*nb03 + i2*nb02 + i1*nb01 + i0*nb00);
float * dst_data = (float *)((char *) dst->data + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
const float x0 = src[0];
const float x1 = src[n_dims/2];
dst_data[0] = x0*cos_theta - x1*sin_theta;
dst_data[n_dims/2] = x0*sin_theta + x1*cos_theta;
} else {
const int64_t i0 = ic;
const float * const src = (float *)((char *) src0->data + i3*nb03 + i2*nb02 + i1*nb01 + i0*nb00);
float * dst_data = (float *)((char *) dst->data + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
dst_data[0] = src[0];
dst_data[1] = src[1];
}
dst_data[0] = x0*cos_theta - x1*sin_theta;
dst_data[n_dims/2] = x0*sin_theta + x1*cos_theta;
}
}
for (int64_t i0 = n_dims; i0 < ne0; i0 += 2) {
const float * const src = (float *)((char *) src0->data + i3*nb03 + i2*nb02 + i1*nb01 + i0*nb00);
float * dst_data = (float *)((char *) dst->data + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
dst_data[0] = src[0];
dst_data[1] = src[1];
}
}
}
}
@ -14477,8 +14396,8 @@ static void ggml_compute_forward_rope_f16(
//const int n_past = ((int32_t *) dst->op_params)[0];
const int n_dims = ((int32_t *) dst->op_params)[1];
const int mode = ((int32_t *) dst->op_params)[2];
const int n_ctx = ((int32_t *) dst->op_params)[3];
const int n_orig_ctx = ((int32_t *) dst->op_params)[4];
//const int n_ctx = ((int32_t *) dst->op_params)[3];
const int n_ctx_orig = ((int32_t *) dst->op_params)[4];
memcpy(&freq_base, (int32_t *) dst->op_params + 5, sizeof(float));
memcpy(&freq_scale, (int32_t *) dst->op_params + 6, sizeof(float));
memcpy(&ext_factor, (int32_t *) dst->op_params + 7, sizeof(float));
@ -14514,20 +14433,15 @@ static void ggml_compute_forward_rope_f16(
const float theta_scale = powf(freq_base, -2.0f/n_dims);
float corr_dims[2];
ggml_rope_yarn_corr_dims(n_dims, n_orig_ctx, freq_base, beta_fast, beta_slow, corr_dims);
ggml_rope_yarn_corr_dims(n_dims, n_ctx_orig, freq_base, beta_fast, beta_slow, corr_dims);
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");
if (src2 != NULL) {
GGML_ASSERT(src2->type == GGML_TYPE_F32);
GGML_ASSERT(src2->ne[0] >= n_dims / 2);
freq_factors = (const float *) src2->data;
}
// backward process uses inverse rotation by cos and sin.
@ -14542,43 +14456,14 @@ static void ggml_compute_forward_rope_f16(
const int64_t p = pos[i2];
float * cache = (float *) params->wdata + (ne0 + CACHE_LINE_SIZE_F32)*ith;
if (!is_glm && !is_neox) { // TODO: cache sin/cos for glm, neox
ggml_rope_cache_init(p, freq_scale, corr_dims, ne0, ext_factor, attn_factor, cache, sin_sign, theta_scale);
}
ggml_rope_cache_init(p, freq_scale, freq_factors, corr_dims, ne0, ext_factor, attn_factor, cache, sin_sign, theta_scale);
for (int64_t i1 = 0; i1 < ne1; i1++) {
if (ir++ < ir0) continue;
if (ir > ir1) break;
float theta_base = (float)p;
if (is_glm) {
theta_base = MIN(p, n_ctx - 2);
float block_theta = MAX(p - (n_ctx - 2), 0);
for (int64_t i0 = 0; i0 < ne0 / 4; i0++) {
const float cos_theta = cosf(theta_base);
const float sin_theta = sinf(theta_base) * sin_sign;
const float cos_block_theta = cosf(block_theta);
const float sin_block_theta = sinf(block_theta) * sin_sign;
theta_base *= theta_scale;
block_theta *= theta_scale;
const ggml_fp16_t * const src = (ggml_fp16_t *)((char *) src0->data + i3*nb03 + i2*nb02 + i1*nb01 + i0*nb00);
ggml_fp16_t * dst_data = (ggml_fp16_t *)((char *) dst->data + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
const float x0 = GGML_FP16_TO_FP32(src[0]);
const float x1 = GGML_FP16_TO_FP32(src[n_dims/2]);
const float x2 = GGML_FP16_TO_FP32(src[n_dims]);
const float x3 = GGML_FP16_TO_FP32(src[n_dims/2*3]);
dst_data[0] = GGML_FP32_TO_FP16(x0*cos_theta - x1*sin_theta);
dst_data[n_dims/2] = GGML_FP32_TO_FP16(x0*sin_theta + x1*cos_theta);
dst_data[n_dims] = GGML_FP32_TO_FP16(x2*cos_block_theta - x3*sin_block_theta);
dst_data[n_dims/2*3] = GGML_FP32_TO_FP16(x2*sin_block_theta + x3*cos_block_theta);
}
} else if (!is_neox) {
for (int64_t i0 = 0; i0 < ne0; i0 += 2) {
if (!is_neox) {
for (int64_t i0 = 0; i0 < n_dims; i0 += 2) {
const float cos_theta = cache[i0 + 0];
const float sin_theta = cache[i0 + 1];
@ -14592,41 +14477,30 @@ static void ggml_compute_forward_rope_f16(
dst_data[1] = GGML_FP32_TO_FP16(x0*sin_theta + x1*cos_theta);
}
} else {
// ref: https://github.com/jquesnelle/yarn/blob/master/scaled_rope/LlamaYaRNScaledRotaryEmbedding.py
for (int64_t ic = 0; ic < ne0; ic += 2) {
if (ic < n_dims) {
const int64_t i0 = ic/2;
for (int64_t i0 = 0; i0 < n_dims; i0 += 2) {
const int64_t ic = i0/2;
const float freq_factor = freq_factors ? freq_factors[i0] : 1.0f;
const float cos_theta = cache[i0 + 0];
const float sin_theta = cache[i0 + 1];
float cos_theta, sin_theta;
rope_yarn(
theta_base/freq_factor, freq_scale, corr_dims, ic, ext_factor, attn_factor,
&cos_theta, &sin_theta
);
const ggml_fp16_t * const src = (ggml_fp16_t *)((char *) src0->data + i3*nb03 + i2*nb02 + i1*nb01 + ic*nb00);
ggml_fp16_t * dst_data = (ggml_fp16_t *)((char *) dst->data + i3*nb3 + i2*nb2 + i1*nb1 + ic*nb0);
sin_theta *= sin_sign;
theta_base *= theta_scale;
const float x0 = GGML_FP16_TO_FP32(src[0]);
const float x1 = GGML_FP16_TO_FP32(src[n_dims/2]);
const ggml_fp16_t * const src = (ggml_fp16_t *)((char *) src0->data + i3*nb03 + i2*nb02 + i1*nb01 + i0*nb00);
ggml_fp16_t * dst_data = (ggml_fp16_t *)((char *) dst->data + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
const float x0 = GGML_FP16_TO_FP32(src[0]);
const float x1 = GGML_FP16_TO_FP32(src[n_dims/2]);
dst_data[0] = GGML_FP32_TO_FP16(x0*cos_theta - x1*sin_theta);
dst_data[n_dims/2] = GGML_FP32_TO_FP16(x0*sin_theta + x1*cos_theta);
} else {
const int64_t i0 = ic;
const ggml_fp16_t * const src = (ggml_fp16_t *)((char *) src0->data + i3*nb03 + i2*nb02 + i1*nb01 + i0*nb00);
ggml_fp16_t * dst_data = (ggml_fp16_t *)((char *) dst->data + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
dst_data[0] = src[0];
dst_data[1] = src[1];
}
dst_data[0] = GGML_FP32_TO_FP16(x0*cos_theta - x1*sin_theta);
dst_data[n_dims/2] = GGML_FP32_TO_FP16(x0*sin_theta + x1*cos_theta);
}
}
for (int64_t i0 = n_dims; i0 < ne0; i0 += 2) {
const ggml_fp16_t * const src = (ggml_fp16_t *)((char *) src0->data + i3*nb03 + i2*nb02 + i1*nb01 + i0*nb00);
ggml_fp16_t * dst_data = (ggml_fp16_t *)((char *) dst->data + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
dst_data[0] = src[0];
dst_data[1] = src[1];
}
}
}
}
@ -18327,9 +18201,9 @@ static void ggml_compute_backward(struct ggml_context * ctx, struct ggml_tensor
//const int n_past = ((int32_t *) tensor->op_params)[0];
const int n_dims = ((int32_t *) tensor->op_params)[1];
const int mode = ((int32_t *) tensor->op_params)[2];
const int n_ctx = ((int32_t *) tensor->op_params)[3];
const int n_orig_ctx = ((int32_t *) tensor->op_params)[4];
float freq_base, freq_scale, ext_factor, attn_factor, beta_fast, beta_slow, xpos_base, xpos_down;
//const int n_ctx = ((int32_t *) tensor->op_params)[3];
const int n_ctx_orig = ((int32_t *) tensor->op_params)[4];
float freq_base, freq_scale, ext_factor, attn_factor, beta_fast, beta_slow;
memcpy(&freq_base, (int32_t *) tensor->op_params + 5, sizeof(float));
memcpy(&freq_scale, (int32_t *) tensor->op_params + 6, sizeof(float));
@ -18337,8 +18211,6 @@ static void ggml_compute_backward(struct ggml_context * ctx, struct ggml_tensor
memcpy(&attn_factor, (int32_t *) tensor->op_params + 8, sizeof(float));
memcpy(&beta_fast, (int32_t *) tensor->op_params + 9, sizeof(float));
memcpy(&beta_slow, (int32_t *) tensor->op_params + 10, sizeof(float));
memcpy(&xpos_base, (int32_t *) tensor->op_params + 11, sizeof(float));
memcpy(&xpos_down, (int32_t *) tensor->op_params + 12, sizeof(bool));
src0->grad = ggml_add_or_set(ctx,
src0->grad,
@ -18348,16 +18220,13 @@ static void ggml_compute_backward(struct ggml_context * ctx, struct ggml_tensor
src2,
n_dims,
mode,
n_ctx,
n_orig_ctx,
n_ctx_orig,
freq_base,
freq_scale,
ext_factor,
attn_factor,
beta_fast,
beta_slow,
xpos_base,
xpos_down),
beta_slow),
zero_table);
}
} break;
@ -18367,9 +18236,9 @@ static void ggml_compute_backward(struct ggml_context * ctx, struct ggml_tensor
//const int n_past = ((int32_t *) tensor->op_params)[0];
const int n_dims = ((int32_t *) tensor->op_params)[1];
const int mode = ((int32_t *) tensor->op_params)[2];
const int n_ctx = ((int32_t *) tensor->op_params)[3];
const int n_orig_ctx = ((int32_t *) tensor->op_params)[4];
float freq_base, freq_scale, ext_factor, attn_factor, beta_fast, beta_slow, xpos_base, xpos_down;
//const int n_ctx = ((int32_t *) tensor->op_params)[3];
const int n_ctx_orig = ((int32_t *) tensor->op_params)[4];
float freq_base, freq_scale, ext_factor, attn_factor, beta_fast, beta_slow;
memcpy(&freq_base, (int32_t *) tensor->op_params + 5, sizeof(float));
memcpy(&freq_scale, (int32_t *) tensor->op_params + 6, sizeof(float));
@ -18377,8 +18246,6 @@ static void ggml_compute_backward(struct ggml_context * ctx, struct ggml_tensor
memcpy(&attn_factor, (int32_t *) tensor->op_params + 8, sizeof(float));
memcpy(&beta_fast, (int32_t *) tensor->op_params + 9, sizeof(float));
memcpy(&beta_slow, (int32_t *) tensor->op_params + 10, sizeof(float));
memcpy(&xpos_base, (int32_t *) tensor->op_params + 11, sizeof(float));
memcpy(&xpos_down, (int32_t *) tensor->op_params + 12, sizeof(bool));
src0->grad = ggml_add_or_set(ctx,
src0->grad,
@ -18388,16 +18255,13 @@ static void ggml_compute_backward(struct ggml_context * ctx, struct ggml_tensor
src2,
n_dims,
mode,
n_ctx,
n_orig_ctx,
n_ctx_orig,
freq_base,
freq_scale,
ext_factor,
attn_factor,
beta_fast,
beta_slow,
xpos_base,
xpos_down,
false),
zero_table);
}

36
ggml.h
View File

@ -1465,7 +1465,6 @@ extern "C" {
// rotary position embedding
// if mode & 1 == 1, skip n_past elements (NOT SUPPORTED)
// if mode & 2 == 1, GPT-NeoX style
// if mode & 4 == 1, ChatGLM style
//
// b is an int32 vector with size a->ne[2], it contains the positions
// c is freq factors (e.g. phi3-128k), (optional)
@ -1474,8 +1473,7 @@ extern "C" {
struct ggml_tensor * a,
struct ggml_tensor * b,
int n_dims,
int mode,
int n_ctx);
int mode);
// in-place, returns view(a)
GGML_API struct ggml_tensor * ggml_rope_inplace(
@ -1483,8 +1481,7 @@ extern "C" {
struct ggml_tensor * a,
struct ggml_tensor * b,
int n_dims,
int mode,
int n_ctx);
int mode);
// custom RoPE
GGML_API struct ggml_tensor * ggml_rope_ext(
@ -1494,8 +1491,7 @@ extern "C" {
struct ggml_tensor * c,
int n_dims,
int mode,
int n_ctx,
int n_orig_ctx,
int n_ctx_orig,
float freq_base,
float freq_scale,
float ext_factor,
@ -1511,8 +1507,7 @@ extern "C" {
struct ggml_tensor * c,
int n_dims,
int mode,
int n_ctx,
int n_orig_ctx,
int n_ctx_orig,
float freq_base,
float freq_scale,
float ext_factor,
@ -1526,8 +1521,7 @@ extern "C" {
struct ggml_tensor * b,
int n_dims,
int mode,
int n_ctx,
int n_orig_ctx,
int n_ctx_orig,
float freq_base,
float freq_scale,
float ext_factor,
@ -1542,8 +1536,7 @@ extern "C" {
struct ggml_tensor * b,
int n_dims,
int mode,
int n_ctx,
int n_orig_ctx,
int n_ctx_orig,
float freq_base,
float freq_scale,
float ext_factor,
@ -1552,17 +1545,9 @@ extern "C" {
float beta_slow),
"use ggml_rope_ext_inplace instead");
struct ggml_tensor * ggml_rope_xpos_inplace(
struct ggml_context * ctx,
struct ggml_tensor * a,
struct ggml_tensor * b,
int n_dims,
float base,
bool down);
// compute correction dims for YaRN RoPE scaling
GGML_CALL void ggml_rope_yarn_corr_dims(
int n_dims, int n_orig_ctx, float freq_base, float beta_fast, float beta_slow, float dims[2]);
int n_dims, int n_ctx_orig, float freq_base, float beta_fast, float beta_slow, float dims[2]);
// rotary position embedding backward, i.e compute dx from dy
// a - dy
@ -1573,16 +1558,13 @@ extern "C" {
struct ggml_tensor * c,
int n_dims,
int mode,
int n_ctx,
int n_orig_ctx,
int n_ctx_orig,
float freq_base,
float freq_scale,
float ext_factor,
float attn_factor,
float beta_fast,
float beta_slow,
float xpos_base,
bool xpos_down);
float beta_slow);
// clamp
// in-place, returns view(a)