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

RoPE: fix back, CUDA support for back + noncont. (#11240)

Browse source 
* RoPE: fix back, CUDA support for back + noncont.

* fix comments reg. non-cont. RoPE support [no-ci]
This commit is contained in:
Johannes Gäßler 2025-01-15 12:51:37 +01:00 • committed by GitHub
parent 0ccd7f3eb2
commit 432df2d5f9
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
9 changed files with 269 additions and 258 deletions
Download patch file Download diff file Expand all files Collapse all files

19
ggml/include/ggml.h
View file

@ -1500,7 +1500,7 @@ extern "C" {
// rotary position embedding backward, i.e compute dx from dy
// a - dy
GGML_API struct ggml_tensor * ggml_rope_back(
GGML_API struct ggml_tensor * ggml_rope_ext_back(
struct ggml_context * ctx,
struct ggml_tensor * a, // gradients of ggml_rope result
struct ggml_tensor * b, // positions
@ -1515,6 +1515,23 @@ extern "C" {
float beta_fast,
float beta_slow);
GGML_API struct ggml_tensor * ggml_rope_multi_back(
struct ggml_context * ctx,
struct ggml_tensor * a,
struct ggml_tensor * b,
struct ggml_tensor * c,
int n_dims,
int sections[4],
int mode,
int n_ctx_orig,
float freq_base,
float freq_scale,
float ext_factor,
float attn_factor,
float beta_fast,
float beta_slow);
// clamp
// in-place, returns view(a)
GGML_API struct ggml_tensor * ggml_clamp(