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

ggml : add unary and binary map operations (#874)

Browse source 
* GGML map ops proof of concept.

* Various cleanups.

Add handling for task setting.

Add handling for ggml_compute_backward.

Rename functions to ggml_map_unary_f32 and ggml_map_binary_f32

Fix compiler warnings related to casting function pointers and `void *`

Reorder functions and definitions based on the GGML op number.

Use typedefs for map op function pointer types.

* Fix position of map ops cases in ggml_compute_forward
This commit is contained in:
Kerfuffle 2023-04-14 08:43:55 -06:00 • committed by GitHub
parent a32f7acc9f
commit c9a59b70a5
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
2 changed files with 237 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

18
ggml.h
View file

@ -253,6 +253,9 @@ enum ggml_op {
GGML_OP_FLASH_ATTN,
GGML_OP_FLASH_FF,
GGML_OP_MAP_UNARY,
GGML_OP_MAP_BINARY,
GGML_OP_COUNT,
};
@ -652,6 +655,21 @@ struct ggml_tensor * ggml_flash_ff(
struct ggml_tensor * c0,
struct ggml_tensor * c1);
// Mapping operations
typedef void (*ggml_unary_op_f32_t)(const int, float *, const float *);
typedef void (*ggml_binary_op_f32_t)(const int, float *, const float *, const float *);
struct ggml_tensor * ggml_map_unary_f32(
struct ggml_context * ctx,
struct ggml_tensor * a,
const ggml_unary_op_f32_t fun);
struct ggml_tensor * ggml_map_binary_f32(
struct ggml_context * ctx,
struct ggml_tensor * a,
struct ggml_tensor * b,
const ggml_binary_op_f32_t fun);
//
// automatic differentiation
//