ExternalVendorCode/whisper.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/whisper.cpp.git synced 2025-06-16 13:58:09 +00:00
Code Issues Actions 4 Packages Projects Releases Wiki Activity

talk-llama : sync llama.cpp

Browse Source 
ggml-ci
This commit is contained in:
Georgi Gerganov
2025-06-10 10:12:44 +03:00
parent 96eaf46ec6
commit db264d6220
23 changed files with 911 additions and 437 deletions
Download Patch File Download Diff File Expand all files Collapse all files

50
examples/talk-llama/llama-model.cpp
View File

@ -543,6 +543,12 @@ void llama_model::load_hparams(llama_model_loader & ml) {
uint32_t n_vocab = 0;
ml.get_key(LLM_KV_VOCAB_SIZE, n_vocab, false) || ml.get_arr_n(LLM_KV_TOKENIZER_LIST, n_vocab, false);
// for classifier models
ml.get_arr(LLM_KV_CLASSIFIER_OUTPUT_LABELS, classifier_labels, false);
if (!classifier_labels.empty()) {
hparams.n_cls_out = classifier_labels.size();
}
// arch-specific KVs
switch (arch) {
case LLM_ARCH_LLAMA:
@ -686,7 +692,6 @@ void llama_model::load_hparams(llama_model_loader & ml) {
ml.get_key(LLM_KV_ATTENTION_LAYERNORM_EPS, hparams.f_norm_eps);
ml.get_key(LLM_KV_ATTENTION_CAUSAL, hparams.causal_attn);
ml.get_key(LLM_KV_POOLING_TYPE, hparams.pooling_type, false);
ml.get_arr_n(LLM_KV_CLASSIFIER_OUTPUT_LABELS, hparams.n_cls_out, false);
switch (hparams.n_layer) {
case 3:
@ -956,6 +961,11 @@ void llama_model::load_hparams(llama_model_loader & ml) {
case 46: type = LLM_TYPE_27B; break;
default: type = LLM_TYPE_UNKNOWN;
}
// ref: https://github.com/google/gemma_pytorch/blob/014acb7ac4563a5f77c76d7ff98f31b568c16508/gemma/config.py#L173
hparams.f_attention_scale = type == LLM_TYPE_27B
? 1.0f / std::sqrt(float(hparams.n_embd / hparams.n_head(0)))
: 1.0f / std::sqrt(float(hparams.n_embd_head_k));
} break;
case LLM_ARCH_GEMMA3:
{
@ -976,6 +986,7 @@ void llama_model::load_hparams(llama_model_loader & ml) {
default: type = LLM_TYPE_UNKNOWN;
}
// ref: https://github.com/google/gemma_pytorch/blob/014acb7ac4563a5f77c76d7ff98f31b568c16508/gemma/config.py#L289
hparams.f_attention_scale = type == LLM_TYPE_27B
? 1.0f / std::sqrt(float(hparams.n_embd / hparams.n_head(0)))
: 1.0f / std::sqrt(float(hparams.n_embd_head_k));
@ -4356,6 +4367,15 @@ void llama_model::print_info() const {
LLAMA_LOG_INFO("%s: ssm_d_state = %u\n", __func__, hparams.ssm_d_state);
LLAMA_LOG_INFO("%s: ssm_dt_rank = %u\n", __func__, hparams.ssm_dt_rank);
LLAMA_LOG_INFO("%s: ssm_dt_b_c_rms = %d\n", __func__, hparams.ssm_dt_b_c_rms);
if (!classifier_labels.empty()) {
LLAMA_LOG_INFO("%s: n_cls_out = %u\n", __func__, hparams.n_cls_out);
size_t i = 0;
for (auto label : classifier_labels) {
LLAMA_LOG_INFO("%s: cls_label[%2zu] = %s\n", __func__, i++, label.c_str());
}
}
}
LLAMA_LOG_INFO("%s: model type = %s\n", __func__, type_name().c_str());
@ -8484,14 +8504,7 @@ struct llm_build_gemma2_iswa : public llm_graph_context {
cb(Kcur, "Kcur", il);
cb(Vcur, "Vcur", il);
// ref: https://github.com/google/gemma_pytorch/commit/03e657582d17cb5a8617ebf333c1c16f3694670e
switch (model.type) {
case LLM_TYPE_2B:
case LLM_TYPE_9B:
case LLM_TYPE_27B: Qcur = ggml_scale(ctx0, Qcur, 1.0f / sqrtf(float(n_embd_head))); break;
default: GGML_ABORT("fatal error");
};
cb(Qcur, "Qcur_scaled", il);
Qcur = ggml_scale(ctx0, Qcur, hparams.f_attention_scale);
cur = build_attn(inp_attn, gf,
model.layers[il].wo, NULL,
@ -8632,9 +8645,12 @@ struct llm_build_gemma3_iswa : public llm_graph_context {
cb(Kcur, "Kcur", il);
cb(Vcur, "Vcur", il);
// ref: https://github.com/google/gemma_pytorch/blob/014acb7ac4563a5f77c76d7ff98f31b568c16508/gemma/model.py#L315
Qcur = ggml_scale(ctx0, Qcur, hparams.f_attention_scale);
cur = build_attn(inp_attn, gf,
model.layers[il].wo, NULL,
Qcur, Kcur, Vcur, nullptr, nullptr, hparams.f_attention_scale, il);
Qcur, Kcur, Vcur, nullptr, nullptr, 1.0f, il);
}
cur = build_norm(cur,
@ -13600,6 +13616,18 @@ int32_t llama_model_n_swa(const llama_model * model) {
return model->hparams.n_swa;
}
uint32_t llama_model_n_cls_out(const struct llama_model * model) {
return model->hparams.n_cls_out;
}
const char * llama_model_cls_label(const struct llama_model * model, uint32_t i) {
if (i < model->classifier_labels.size()) {
return model->classifier_labels[i].c_str();
}
return nullptr;
}
// deprecated
int32_t llama_n_ctx_train(const llama_model * model) {
return llama_model_n_ctx_train(model);
@ -13760,7 +13788,7 @@ uint64_t llama_model_size(const llama_model * model) {
}
const char * llama_model_chat_template(const llama_model * model, const char * name) {
const auto key = name ? LLM_KV(model->arch, name)(LLM_KV_TOKENIZER_CHAT_TEMPLATE_N)
const auto key = name ? LLM_KV(model->arch, name)(LLM_KV_TOKENIZER_CHAT_TEMPLATE)
: LLM_KV(model->arch)(LLM_KV_TOKENIZER_CHAT_TEMPLATE);
const auto & it = model->gguf_kv.find(key);
if (it == model->gguf_kv.end()) {