Extend serving rotary cache for code context

runs/frontier-s3-real-v0/run_full_real.sh

@@ -152,13 +152,16 @@ PY
 
 export TOKENIZERS_PARALLELISM=false VLLM_USE_V1=1 TORCH_CUDA_ARCH_LIST=9.0 PREFIX_CACHING=true
 export HF_HUB_OFFLINE=1 TRANSFORMERS_OFFLINE=1 FLASHINFER_WORKSPACE_BASE
+HF_OVERRIDE_ARGS=()
 if [[ "${ALLOW_LONG_CONTEXT_SERVER}" == "true" ]]; then
   export VLLM_ALLOW_LONG_MAX_MODEL_LEN=1
+  HF_OVERRIDE_ARGS=(--hf-overrides "{\"max_position_embeddings\":${MAX_MODEL_LEN}}")
 fi
 ulimit -n 65536
 setsid "${VENV_ROOT}/bin/vllm" serve "${MODEL_ROOT}" \
   --host 127.0.0.1 --port "${PORT}" --served-model-name "${SERVED_MODEL}" \
   --tensor-parallel-size "${TP}" --gpu-memory-utilization 0.92 \
+  "${HF_OVERRIDE_ARGS[@]}" \
   --max-model-len "${MAX_MODEL_LEN}" --max-num-batched-tokens 8192 --max-num-seqs "${MNS}" \
   --enable-prefix-caching --block-size 16 --enable-chunked-prefill --no-enable-log-requests \
   --enable-logging-iteration-details \