public_git/whisper.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/whisper.cpp.git synced 2026-10-06 06:21:41 +02:00
Code Issues Packages Projects Releases Wiki Activity

talk-llama : sync llama.cpp

Browse Source
This commit is contained in:
Georgi Gerganov
2026-01-30 15:56:40 +02:00
parent b529c0610f
commit 953e503fd9
38 changed files with 2299 additions and 730 deletions
Download Patch File Download Diff File Expand all files Collapse all files
examples/talk-llama/llama-adapter.cpp
+7 -13
View File
@@ -146,11 +146,9 @@ llama_adapter_lora_weight * llama_adapter_lora::get_weight(ggml_tensor * w) {
return nullptr;
}
static void llama_adapter_lora_init_impl(const char * path_lora, llama_adapter_lora & adapter) {
static void llama_adapter_lora_init_impl(llama_model & model, const char * path_lora, llama_adapter_lora & adapter) {
LLAMA_LOG_INFO("%s: loading lora adapter from '%s' ...\n", __func__, path_lora);
llama_model & model = adapter.model;
ggml_context * ctx_init;
gguf_init_params meta_gguf_params = {
/* .no_alloc = */ true,
@@ -413,17 +411,17 @@ static void llama_adapter_lora_init_impl(const char * path_lora, llama_adapter_l
}
}
// update number of nodes used
model.n_lora_nodes += adapter.get_n_nodes();
// register adapter with model
model.loras.insert(&adapter);
LLAMA_LOG_INFO("%s: loaded %zu tensors from lora file\n", __func__, adapter.ab_map.size()*2);
}
llama_adapter_lora * llama_adapter_lora_init(llama_model * model, const char * path_lora) {
llama_adapter_lora * adapter = new llama_adapter_lora(*model);
llama_adapter_lora * adapter = new llama_adapter_lora();
try {
llama_adapter_lora_init_impl(path_lora, *adapter);
llama_adapter_lora_init_impl(*model, path_lora, *adapter);
return adapter;
} catch (const std::exception & err) {
LLAMA_LOG_ERROR("%s: failed to apply lora adapter: %s\n", __func__, err.what());
@@ -473,12 +471,8 @@ int32_t llama_adapter_meta_val_str_by_index(const llama_adapter_lora * adapter,
return snprintf(buf, buf_size, "%s", it->second.c_str());
}
void llama_adapter_lora_free(llama_adapter_lora * adapter) {
// update number of nodes used
GGML_ASSERT(adapter->model.n_lora_nodes >= adapter->get_n_nodes());
adapter->model.n_lora_nodes -= adapter->get_n_nodes();
delete adapter;
void llama_adapter_lora_free(llama_adapter_lora *) {
// deprecated: adapters are freed by llama_model's destructor
}
uint64_t llama_adapter_get_alora_n_invocation_tokens(const struct llama_adapter_lora * adapter) {