extern/whisper.cpp
1
0
Fork 1
mirror of https://github.com/ggerganov/whisper.cpp.git synced 2025-08-09 05:35:26 +02:00
Code Issues Packages Projects Releases Wiki Activity

talk-llama : sync llama.cpp

Browse Source 
ggml-ci
This commit is contained in:
Georgi Gerganov
2025-05-13 13:20:19 +03:00
parent a14c89aefa
commit f890560575
25 changed files with 2847 additions and 1125 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
examples/talk-llama/llama-model-loader.cpp
View File

@ -301,12 +301,12 @@ namespace GGUFMeta {
GGUFMeta::GKV<GGUFMeta::ArrayInfo>::get_kv(meta.get(), kid);
switch (arr_info.gt) {
case GGUF_TYPE_FLOAT32: GGML_ASSERT((std::is_same<T, float>::value)); break;
case GGUF_TYPE_INT32: GGML_ASSERT(
(std::is_same<T, int32_t>::value) ||
(std::is_same<T, uint32_t>::value)); break;
case GGUF_TYPE_UINT32:
case GGUF_TYPE_INT32: GGML_ASSERT((std::is_same<T, int32_t>::value) ||
(std::is_same<T, uint32_t>::value)); break;
case GGUF_TYPE_FLOAT32: GGML_ASSERT((std::is_same<T, float>::value)); break;
default:
throw std::runtime_error(format("%s is not a float32, int32 array", key.c_str()));
throw std::runtime_error(format("%s is not a float32/uint32/int32 array", key.c_str()));
}
result.resize(arr_info.length);
@ -330,12 +330,12 @@ namespace GGUFMeta {
GGUFMeta::GKV<GGUFMeta::ArrayInfo>::get_kv(meta.get(), kid);
switch (arr_info.gt) {
case GGUF_TYPE_FLOAT32: GGML_ASSERT((std::is_same<T, float>::value)); break;
case GGUF_TYPE_INT32: GGML_ASSERT(
(std::is_same<T, int32_t>::value) ||
(std::is_same<T, uint32_t>::value)); break;
case GGUF_TYPE_UINT32:
case GGUF_TYPE_INT32: GGML_ASSERT((std::is_same<T, int32_t>::value) ||
(std::is_same<T, uint32_t>::value)); break;
case GGUF_TYPE_FLOAT32: GGML_ASSERT((std::is_same<T, float>::value)); break;
default:
throw std::runtime_error(format("%s is not a float32, int32 array", key.c_str()));
throw std::runtime_error(format("%s is not a float32/uint32/int32 array", key.c_str()));
}
if (arr_info.length > N_MAX) {
@ -823,6 +823,10 @@ void llama_model_loader::init_mappings(bool prefetch, llama_mlocks * mlock_mmaps
mmaps_used.reserve(files.size());
for (const auto & file : files) {
auto * reg = ggml_backend_dev_backend_reg(ggml_backend_dev_by_type(GGML_BACKEND_DEVICE_TYPE_CPU));
if (!reg) {
throw std::runtime_error(format("%s: no CPU backend found", __func__));
}
auto * is_numa_fn = (decltype(ggml_is_numa) *) ggml_backend_reg_get_proc_address(reg, "ggml_backend_cpu_is_numa");
std::unique_ptr<llama_mmap> mapping = std::make_unique<llama_mmap>(file.get(), prefetch ? -1 : 0, is_numa_fn());
mmaps_used.emplace_back(mapping->size(), 0);