llama.cpp

llama-model : support Qwen2 embedding models and pooling_mode_lasttoken

#13245

Merged