mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2025-11-16 11:27:03 +00:00
ggml : add ops SOFTPLUS, EXPM1, TRI, SOLVE_TRI, CUMSUM (#17063)
* Add ops needed for new hybrid models: SOFTPLUS, EXPM1, TRI, SOLVE_TRI, CUMSUM * Update ggml/include/ggml.h Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Update tests/test-backend-ops.cpp Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Code review * Whitespace * Update tests/test-backend-ops.cpp Co-authored-by: Diego Devesa <slarengh@gmail.com> * This is actually sigmoid, duh. * Add CONST, remove TRI_KEEP, other changes from review * Update tests/test-backend-ops.cpp Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Update ggml/src/ggml.c Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Update ggml/src/ggml.c Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * Update ggml/src/ggml-cuda/unary.cu Co-authored-by: Aman Gupta <amangupta052@gmail.com> * Remove extra script * Update ggml/src/ggml.c Co-authored-by: Diego Devesa <slarengh@gmail.com> * Update tests/test-backend-ops.cpp Co-authored-by: Diego Devesa <slarengh@gmail.com> * moving changes from laptop [no ci] * pre-rebase * Update tests/test-backend-ops.cpp Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com> * Update tests/test-backend-ops.cpp Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com> * Refactor tests * ggml : cleanup * cont : fix ggml_fill srcs * tests : add note * ggml : add ggml_fill_inplace * ggml : add asserts * ggml : fix ggml_fill constant cast * cont : ggml_tri minor * Use TENSOR_LOCALS * Fix regression from #14596, regenerate * Don't make commits at night... --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> Co-authored-by: Diego Devesa <slarengh@gmail.com> Co-authored-by: Aman Gupta <amangupta052@gmail.com> Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>
This commit is contained in:
Piotr Wilkin (ilintar)
committed by
GitHub
GitHub
parent
a19bd6f7ce
commit
389ac78b26
@@ -475,6 +475,7 @@ extern "C" {
|
||||
GGML_OP_COS,
|
||||
GGML_OP_SUM,
|
||||
GGML_OP_SUM_ROWS,
|
||||
GGML_OP_CUMSUM,
|
||||
GGML_OP_MEAN,
|
||||
GGML_OP_ARGMAX,
|
||||
GGML_OP_COUNT_EQUAL,
|
||||
@@ -530,6 +531,8 @@ extern "C" {
|
||||
GGML_OP_TIMESTEP_EMBEDDING,
|
||||
GGML_OP_ARGSORT,
|
||||
GGML_OP_LEAKY_RELU,
|
||||
GGML_OP_TRI,
|
||||
GGML_OP_FILL,
|
||||
|
||||
GGML_OP_FLASH_ATTN_EXT,
|
||||
GGML_OP_FLASH_ATTN_BACK,
|
||||
@@ -542,6 +545,7 @@ extern "C" {
|
||||
GGML_OP_RWKV_WKV6,
|
||||
GGML_OP_GATED_LINEAR_ATTN,
|
||||
GGML_OP_RWKV_WKV7,
|
||||
GGML_OP_SOLVE_TRI,
|
||||
|
||||
GGML_OP_UNARY,
|
||||
|
||||
@@ -576,6 +580,8 @@ extern "C" {
|
||||
GGML_UNARY_OP_HARDSWISH,
|
||||
GGML_UNARY_OP_HARDSIGMOID,
|
||||
GGML_UNARY_OP_EXP,
|
||||
GGML_UNARY_OP_EXPM1,
|
||||
GGML_UNARY_OP_SOFTPLUS,
|
||||
GGML_UNARY_OP_GELU_ERF,
|
||||
GGML_UNARY_OP_XIELU,
|
||||
GGML_UNARY_OP_FLOOR,
|
||||
@@ -620,6 +626,13 @@ extern "C" {
|
||||
GGML_TENSOR_FLAG_LOSS = 8, // ...defines loss for numerical optimization (multiple loss tensors add up)
|
||||
};
|
||||
|
||||
enum ggml_tri_type {
|
||||
GGML_TRI_TYPE_UPPER_DIAG = 0,
|
||||
GGML_TRI_TYPE_UPPER = 1,
|
||||
GGML_TRI_TYPE_LOWER_DIAG = 2,
|
||||
GGML_TRI_TYPE_LOWER = 3
|
||||
};
|
||||
|
||||
struct ggml_init_params {
|
||||
// memory pool
|
||||
size_t mem_size; // bytes
|
||||
@@ -957,6 +970,22 @@ extern "C" {
|
||||
struct ggml_context * ctx,
|
||||
struct ggml_tensor * a);
|
||||
|
||||
GGML_API struct ggml_tensor * ggml_expm1(
|
||||
struct ggml_context * ctx,
|
||||
struct ggml_tensor * a);
|
||||
|
||||
GGML_API struct ggml_tensor * ggml_expm1_inplace(
|
||||
struct ggml_context * ctx,
|
||||
struct ggml_tensor * a);
|
||||
|
||||
GGML_API struct ggml_tensor * ggml_softplus(
|
||||
struct ggml_context * ctx,
|
||||
struct ggml_tensor * a);
|
||||
|
||||
GGML_API struct ggml_tensor * ggml_softplus_inplace(
|
||||
struct ggml_context * ctx,
|
||||
struct ggml_tensor * a);
|
||||
|
||||
GGML_API struct ggml_tensor * ggml_sin(
|
||||
struct ggml_context * ctx,
|
||||
struct ggml_tensor * a);
|
||||
@@ -983,6 +1012,10 @@ extern "C" {
|
||||
struct ggml_context * ctx,
|
||||
struct ggml_tensor * a);
|
||||
|
||||
GGML_API struct ggml_tensor * ggml_cumsum(
|
||||
struct ggml_context * ctx,
|
||||
struct ggml_tensor * a);
|
||||
|
||||
// mean along rows
|
||||
GGML_API struct ggml_tensor * ggml_mean(
|
||||
struct ggml_context * ctx,
|
||||
@@ -2187,6 +2220,23 @@ extern "C" {
|
||||
int shift2,
|
||||
int shift3);
|
||||
|
||||
// Convert matrix into a triangular one (upper, strict upper, lower or strict lower) by writing
|
||||
// zeroes everywhere outside the masked area
|
||||
GGML_API struct ggml_tensor * ggml_tri(
|
||||
struct ggml_context * ctx,
|
||||
struct ggml_tensor * a,
|
||||
enum ggml_tri_type type);
|
||||
|
||||
// Fill tensor a with constant c
|
||||
GGML_API struct ggml_tensor * ggml_fill(
|
||||
struct ggml_context * ctx,
|
||||
struct ggml_tensor * a,
|
||||
float c);
|
||||
|
||||
GGML_API struct ggml_tensor * ggml_fill_inplace(
|
||||
struct ggml_context * ctx,
|
||||
struct ggml_tensor * a,
|
||||
float c);
|
||||
|
||||
// Ref: https://github.com/CompVis/stable-diffusion/blob/main/ldm/modules/diffusionmodules/util.py#L151
|
||||
// timesteps: [N,]
|
||||
@@ -2356,6 +2406,27 @@ extern "C" {
|
||||
struct ggml_tensor * b,
|
||||
struct ggml_tensor * state);
|
||||
|
||||
/* Solves a specific equation of the form Ax=B, where A is a triangular matrix
|
||||
* without zeroes on the diagonal (i.e. invertible).
|
||||
* B can have any number of columns, but must have the same number of rows as A
|
||||
* If A is [n, n] and B is [n, m], then the result will be [n, m] as well
|
||||
* Has O(n^3) complexity (unlike most matrix ops out there), so use on cases
|
||||
* where n > 100 sparingly, pre-chunk if necessary.
|
||||
*
|
||||
* If left = false, solves xA=B instead
|
||||
* If lower = false, assumes upper triangular instead
|
||||
* If uni = true, assumes diagonal of A to be all ones (will override actual values)
|
||||
*
|
||||
* TODO: currently only lower, right, non-unitriangular variant is implemented
|
||||
*/
|
||||
GGML_API struct ggml_tensor * ggml_solve_tri(
|
||||
struct ggml_context * ctx,
|
||||
struct ggml_tensor * a,
|
||||
struct ggml_tensor * b,
|
||||
bool left,
|
||||
bool lower,
|
||||
bool uni);
|
||||
|
||||
// custom operators
|
||||
|
||||
typedef void (*ggml_custom1_op_t)(struct ggml_tensor * dst , const struct ggml_tensor * a, int ith, int nth, void * userdata);
|
||||
|
||||
Reference in New Issue
Block a user