vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-15 19:39:42 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Merge branch 'upstream' into concedo_experimental

Browse source 
# Conflicts:
#	.ecrc
#	CMakePresets.json
#	ci/run.sh
#	docs/backend/SYCL.md
#	ggml/src/CMakeLists.txt
#	src/llama.cpp
#	tests/test-backend-ops.cpp
#	tests/test-sampling.cpp
This commit is contained in:
Concedo 2024-08-27 17:46:40 +08:00
commit b2c1ff7a13
30 changed files with 7666 additions and 6889 deletions
Download patch file Download diff file Expand all files Collapse all files

6
otherarch/whispercpp/whisper.cpp
View file

@ -2008,7 +2008,7 @@ static struct ggml_cgraph * whisper_build_graph_encoder(
ggml_element_size(kv_pad.v)*n_state_head,
0);
cur = ggml_flash_attn_ext(ctx0, Q, K, V, nullptr, KQscale, 0.0f);
cur = ggml_flash_attn_ext(ctx0, Q, K, V, nullptr, KQscale, 0.0f, 0.0f);
cur = ggml_reshape_2d(ctx0, cur, n_state, n_ctx);
} else {
@ -2471,7 +2471,7 @@ static struct ggml_cgraph * whisper_build_graph_decoder(
ggml_element_size(kv_self.v)*n_state_head,
ggml_element_size(kv_self.v)*n_state*n_ctx*il);
cur = ggml_flash_attn_ext(ctx0, Q, K, V, KQ_mask_f16, 1.0f, 0.0f);
cur = ggml_flash_attn_ext(ctx0, Q, K, V, KQ_mask_f16, 1.0f, 0.0f, 0.0f);
cur = ggml_reshape_2d(ctx0, cur, n_state, n_tokens);
} else {
@ -2553,7 +2553,7 @@ static struct ggml_cgraph * whisper_build_graph_decoder(
ggml_element_size(wstate.kv_cross.v)*n_state_head,
ggml_element_size(wstate.kv_cross.v)*n_state*n_audio_ctx_pad*il);
cur = ggml_flash_attn_ext(ctx0, Q, Kcross, Vcross, nullptr, KQscale, 0.0f);
cur = ggml_flash_attn_ext(ctx0, Q, Kcross, Vcross, nullptr, KQscale, 0.0f, 0.0f);
cur = ggml_reshape_2d(ctx0, cur, n_state, n_tokens);
} else {