thek0tyara/llama-cpp-turboquant
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 4

llama : add support for qwen3 reranker (#15824)

Browse source
This commit is contained in:
Douglas Hanley 2025-09-25 03:53:09 -05:00 • committed by GitHub
parent dfcd53f7ec
commit b5bd037832
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
9 changed files with 166 additions and 78 deletions
Download patch file Download diff file Expand all files Collapse all files

8
common/common.cpp
View file

@ -961,15 +961,13 @@ struct common_init_result common_init_from_params(common_params & params) {
bool has_eos = llama_vocab_eos(vocab) != LLAMA_TOKEN_NULL;
bool has_sep = llama_vocab_sep(vocab) != LLAMA_TOKEN_NULL;
bool has_rerank_prompt = llama_model_chat_template(model, "rerank") != NULL;
if (!has_eos && !has_sep) {
LOG_WRN("%s: warning: vocab does not have an EOS token or SEP token, reranking will not work\n", __func__);
if (!has_eos && !has_sep && !has_rerank_prompt) {
LOG_WRN("%s: warning: vocab does not have an EOS token, SEP token, or rerank prompt. Reranking will not work\n", __func__);
ok = false;
} else if (!has_eos) {
LOG_WRN("%s: warning: vocab does not have an EOS token, using SEP token as fallback\n", __func__);
} else if (!has_sep) {
LOG_WRN("%s: warning: vocab does not have a SEP token, reranking will not work\n", __func__);
ok = false;
}
if (!ok) {