Update containers/llamacpp/llamacpp.container

containers/llamacpp/llamacpp.container

@@ -23,6 +23,14 @@ Environment=LLAMA_ARG_LOAD_MODE=none
 Environment=LLAMA_ARG_CTX_SIZE=131072
 Environment=LLAMA_ARG_HF_REPO=unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF:Q2_K
 
+# Improvements
+Environment=LLAMA_ARG_CACHE_REUSE=256
+Environment=LLAMA_ARG_BATCH=4096
+Environment=LLAMA_ARG_UBATCH=1024
+Environment=LLAMA_ARG_CACHE_TYPE_K=q4_1
+Environment=LLAMA_ARG_CACHE_TYPE_V=q4_1
+#Environment=LLAMA_CHAT_TEMPLATE_KWARGS={"thinking":true,"reasoning_effort":"medium"}
+
 # MTP
 Environment=LLAMA_ARG_SPEC_TYPE=draft-mtp
 Environment=LLAMA_ARG_SPEC_DRAFT_N_MAX=6