public_git/whisper.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/whisper.cpp.git synced 2026-10-04 21:41:24 +02:00
Code Issues Packages Projects Releases Wiki Activity

talk-llama : sync llama.cpp

Browse Source
This commit is contained in:
Georgi Gerganov
2026-01-14 09:11:59 +02:00
parent 13dc9a912b
commit ecfcc65fbf
43 changed files with 3208 additions and 668 deletions
Download Patch File Download Diff File Expand all files Collapse all files
examples/talk-llama/models/modern-bert.cpp
+5 -15
View File
@@ -1,7 +1,6 @@
#include "models.h"
template <bool iswa>
llm_build_modern_bert<iswa>::llm_build_modern_bert(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
llm_build_modern_bert::llm_build_modern_bert(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v;
const int64_t n_embd_gqa = hparams.n_embd_v_gqa();
@@ -24,13 +23,8 @@ llm_build_modern_bert<iswa>::llm_build_modern_bert(const llama_model & model, co
auto * inp_attn = build_attn_inp_no_cache();
for (int il = 0; il < n_layer; ++il) {
float freq_base_l = 0.0f;
if constexpr (iswa) {
freq_base_l = model.get_rope_freq_base(cparams, il);
} else {
freq_base_l = freq_base;
}
const float freq_base_l = model.get_rope_freq_base(cparams, il);
const float freq_scale_l = model.get_rope_freq_scale(cparams, il);
cur = inpL;
@@ -55,13 +49,13 @@ llm_build_modern_bert<iswa>::llm_build_modern_bert(const llama_model & model, co
// RoPE
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
n_rot, rope_type, n_ctx_orig, freq_base_l, freq_scale,
n_rot, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
ext_factor, attn_factor, beta_fast, beta_slow
);
Kcur = ggml_rope_ext(
ctx0, Kcur, inp_pos, nullptr,
n_rot, rope_type, n_ctx_orig, freq_base_l, freq_scale,
n_rot, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
ext_factor, attn_factor, beta_fast, beta_slow
);
@@ -120,7 +114,3 @@ llm_build_modern_bert<iswa>::llm_build_modern_bert(const llama_model & model, co
res->t_embd = cur;
ggml_build_forward_expand(gf, cur);
}
// Explicit template instantiations
template struct llm_build_modern_bert<false>;
template struct llm_build_modern_bert<true>;