tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-06-27 03:55:20 +00:00
Code Issues Packages Projects Releases Wiki Activity

llama : fix llama_model_chat_template with template name (LLM_KV with suffix) (#14050)

Browse Source
This commit is contained in:
Sigbjørn Skjæret
2025-06-07 14:13:12 +02:00
committed by GitHub
parent 745aa5319b
commit 0974ad7a7c
3 changed files with 9 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
src/llama-arch.cpp
View File

@ -200,7 +200,6 @@ static const std::map<llm_kv, const char *> LLM_KV_NAMES = {
{ LLM_KV_TOKENIZER_HF_JSON, "tokenizer.huggingface.json" }, { LLM_KV_TOKENIZER_HF_JSON, "tokenizer.huggingface.json" },
{ LLM_KV_TOKENIZER_RWKV, "tokenizer.rwkv.world" }, { LLM_KV_TOKENIZER_RWKV, "tokenizer.rwkv.world" },
{ LLM_KV_TOKENIZER_CHAT_TEMPLATE, "tokenizer.chat_template" }, { LLM_KV_TOKENIZER_CHAT_TEMPLATE, "tokenizer.chat_template" },
{ LLM_KV_TOKENIZER_CHAT_TEMPLATE_N, "tokenizer.chat_template.%s" },
{ LLM_KV_TOKENIZER_FIM_PRE_ID, "tokenizer.ggml.fim_pre_token_id" }, { LLM_KV_TOKENIZER_FIM_PRE_ID, "tokenizer.ggml.fim_pre_token_id" },
{ LLM_KV_TOKENIZER_FIM_SUF_ID, "tokenizer.ggml.fim_suf_token_id" }, { LLM_KV_TOKENIZER_FIM_SUF_ID, "tokenizer.ggml.fim_suf_token_id" },
{ LLM_KV_TOKENIZER_FIM_MID_ID, "tokenizer.ggml.fim_mid_token_id" }, { LLM_KV_TOKENIZER_FIM_MID_ID, "tokenizer.ggml.fim_mid_token_id" },
@ -1707,8 +1706,14 @@ static const std::map<llm_tensor, llm_tensor_info> LLM_TENSOR_INFOS = {
LLM_KV::LLM_KV(llm_arch arch, const char * suffix) : arch(arch), suffix(suffix) {} LLM_KV::LLM_KV(llm_arch arch, const char * suffix) : arch(arch), suffix(suffix) {}
std::string LLM_KV::operator()(llm_kv kv) const { std::string LLM_KV::operator()(llm_kv kv) const {
return suffix ? ::format(LLM_KV_NAMES.at(kv), LLM_ARCH_NAMES.at(arch), suffix) std::string name = ::format(LLM_KV_NAMES.at(kv), LLM_ARCH_NAMES.at(arch));
: ::format(LLM_KV_NAMES.at(kv), LLM_ARCH_NAMES.at(arch));
if (suffix != nullptr) {
name += ".";
name += suffix;
}
return name;
} }
std::string LLM_TN_IMPL::str() const { std::string LLM_TN_IMPL::str() const {

1
src/llama-arch.h
View File

@ -196,7 +196,6 @@ enum llm_kv {
LLM_KV_TOKENIZER_HF_JSON, LLM_KV_TOKENIZER_HF_JSON,
LLM_KV_TOKENIZER_RWKV, LLM_KV_TOKENIZER_RWKV,
LLM_KV_TOKENIZER_CHAT_TEMPLATE, LLM_KV_TOKENIZER_CHAT_TEMPLATE,
LLM_KV_TOKENIZER_CHAT_TEMPLATE_N,
LLM_KV_TOKENIZER_FIM_PRE_ID, LLM_KV_TOKENIZER_FIM_PRE_ID,
LLM_KV_TOKENIZER_FIM_SUF_ID, LLM_KV_TOKENIZER_FIM_SUF_ID,
LLM_KV_TOKENIZER_FIM_MID_ID, LLM_KV_TOKENIZER_FIM_MID_ID,

2
src/llama-model.cpp
View File

@ -13788,7 +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 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); : LLM_KV(model->arch)(LLM_KV_TOKENIZER_CHAT_TEMPLATE);
const auto & it = model->gguf_kv.find(key); const auto & it = model->gguf_kv.find(key);
if (it == model->gguf_kv.end()) { if (it == model->gguf_kv.end()) {