result

openr/infer.sh

@@ -80,9 +80,9 @@ ulimit -u 4125556
 cd /nfs_global/S/zhangxiaoyun/prm/openr
 export PYTHONPATH=$(pwd)
 
-bash reason/llm_service/create_service_qwen2.5_math_vllm_gold_prm_speed.sh --acc 1.0 --policy_model_name Qwen2.5-Math-1.5B-Instruct
+bash reason/llm_service/create_service_qwen2.5_math_vllm_gold_prm_speed.sh --acc 1.0 --policy_model_name Qwen2.5-Math-7B-Instruct
 sleep 100s
-bash scripts/eval/beam_search.sh --acc 1.0 --policy_model_name Qwen2.5-Math-1.5B-Instruct
+bash scripts/eval/beam_search.sh --acc 1.0 --policy_model_name Qwen2.5-Math-7B-Instruct
 
 #- End
 echo "Job end at $(date "+%Y-%m-%d %H:%M:%S")"

openr/scripts/eval/beam_search.sh

@@ -15,7 +15,7 @@ done
 
 
 VALUE_MODEL_NAME=${POLICY_MODEL_NAME}_RM
-SAVE_DIR="${POLICY_MODEL_NAME}/${ACC}"
+SAVE_DIR="results/${POLICY_MODEL_NAME}/${ACC}"
 
 echo "POLICY_MODEL_NAME is $POLICY_MODEL_NAME"
 echo "VALUE_MODEL_NAME is $VALUE_MODEL_NAME"