model : support Qwen3-Embedding (#15023)

2025-08-02 03:44:50 -05:00
parent f906275537
commit 339bd0268c
3 changed files with 18 additions and 2 deletions
--- a/src/llama-model.cpp
+++ b/src/llama-model.cpp
@@ -899,6 +899,7 @@ void llama_model::load_hparams(llama_model_loader & ml) {
            } break;
        case LLM_ARCH_QWEN3:
            {
+                ml.get_key(LLM_KV_POOLING_TYPE, hparams.pooling_type, false);
                ml.get_key(LLM_KV_ATTENTION_LAYERNORM_RMS_EPS, hparams.f_norm_rms_eps);
                switch (hparams.n_layer) {
                    case 28: type = hparams.n_embd == 1024 ? LLM_TYPE_0_6B : LLM_TYPE_1_7B; break;