Update verl_0720_main/9b.sh

verl_0720_main/9b.sh

@@ -49,7 +49,7 @@ pip install --upgrade pip
 pip uninstall -y moxing-framework || true
 
 # ================= 路径配置 =================
-SCRIPT_DIR=/opt/huawei/dataset/zyr_yuyin/lyf/verl-nanoclaw-rl/verl_0720_nanoclaw/verl_0720_main
+SCRIPT_DIR=/verl_0720_main
 if [ -f "${SCRIPT_DIR}/verl/requirements-npu.txt" ]; then
     DEFAULT_WORK_DIR=${SCRIPT_DIR}/verl
 elif [ -f "${SCRIPT_DIR}/requirements-npu.txt" ]; then
@@ -59,13 +59,13 @@ else
 fi
 WORK_DIR=${WORK_DIR:-${DEFAULT_WORK_DIR}}
 INSTALL_DIR=${INSTALL_DIR:-/home/ma-user}
-BKGS=${BKGS:-/opt/huawei/dataset/zyr_yuyin/bkgs}
+BKGS=${BKGS:-//bkgs}
 chmod 755 "${INSTALL_DIR}"
 
 # Nanoclaw 自定义包已随 WORK_DIR 提供：nanoclaw_recipe。
 
 GCC_INSTALL_PREFIX=${GCC_INSTALL_PREFIX:-/home/ma-user/gcc-11.3.0}
-COMPILED_GCC_ARCHIVE_PATH=${COMPILED_GCC_ARCHIVE_PATH:-/opt/huawei/dataset/zyr_yuyin/bkgs/gcc-11.3.0-compiled-aarch64.tar.gz}
+COMPILED_GCC_ARCHIVE_PATH=${COMPILED_GCC_ARCHIVE_PATH:-//gcc-11.3.0-compiled-aarch64.tar.gz}
 
 echo "--> 正在从缓存恢复 GCC 11.3.0..."
 tar -xzf "${COMPILED_GCC_ARCHIVE_PATH}" -C /home/ma-user/
@@ -79,12 +79,12 @@ gcc --version
 cd "${BKGS}"
 cp jemalloc-5.3.0.tar.bz2 "${INSTALL_DIR}"
 
-VLLM_LATEST_PKGS=${VLLM_LATEST_PKGS:-/opt/huawei/dataset/zyr_yuyin/lyf/verl-05-12/verl_new_26_05_09/pkgs}
+VLLM_LATEST_PKGS=${VLLM_LATEST_PKGS:-/pkgs}
 rm -rf "${INSTALL_DIR}/vllm" "${INSTALL_DIR}/vllm-ascend"
 cp -r "${VLLM_LATEST_PKGS}/vllm" "${INSTALL_DIR}"
 cp -r "${VLLM_LATEST_PKGS}/vllm-ascend" "${INSTALL_DIR}"
 
-CANN_BKGS=${CANN_BKGS:-/opt/huawei/dataset/zyr_yuyin/bkgs/cann_0527}
+CANN_BKGS=${CANN_BKGS:-/cann_0527}
 cp "${CANN_BKGS}/Ascend-cann-toolkit_9.0.0_linux-aarch64.run" "${INSTALL_DIR}"
 cp "${CANN_BKGS}/Ascend-cann-910b-ops_9.0.0_linux-aarch64.run" "${INSTALL_DIR}"
 cp "${CANN_BKGS}/Ascend-cann-nnal_9.0.0_linux-aarch64.run" "${INSTALL_DIR}"
@@ -134,19 +134,6 @@ make -j"$(nproc)"
 make install
 export LD_PRELOAD=${INSTALL_DIR}/lib/libjemalloc.so.2:${LD_PRELOAD:-}
 
-# # ================= 可选：安装 MindSpeed 栈 =================
-# # 纯 VERL engine 路线不依赖 MindSpeed-MM YAML。默认不安装，避免和新版 VERL engine 混淆。
-# INSTALL_MINDSPEED_STACK=${INSTALL_MINDSPEED_STACK:-0}
-# if [ "${INSTALL_MINDSPEED_STACK}" = "1" ]; then
-#     MindSpeed_PATH=${MindSpeed_PATH:-/opt/huawei/dataset/zyr_yuyin/lyf/verl-slow-stable}
-#     cd "${MindSpeed_PATH}"
-#     cd Megatron-LM && pip install -e . --no-deps && cd ..
-#     cd MindSpeed && pip install -e . --no-deps && cd ..
-#     cd MindSpeed-MM && mkdir -p logs data ckpt && pip install -e . --no-deps && cd ..
-#     pip install beartype bs4 diffusers==0.30.3 ftfy imageio-ffmpeg pandarallel pytest-mock
-# else
-#     echo "--> Skip MindSpeed/MindSpeed-MM installation for pure VERL engine route."
-# fi
 
 
 
@@ -157,7 +144,7 @@ pip uninstall -y triton
 # 2. 卸载 triton-ascend（增加 -y 自动确认）
 pip uninstall -y triton-ascend
 pip install --no-cache-dir --force-reinstall triton==3.5.0
-pip install --no-deps /opt/huawei/dataset/zyr_yuyin/bkgs/triton_ascend-3.2.1-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl
+pip install --no-deps triton_ascend-3.2.1-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl
 
 
 # ================= 安装新版 VERL =================
@@ -360,7 +347,7 @@ export RAY_EXPERIMENTAL_NOSET_ASCEND_RT_VISIBLE_DEVICES=1
 
 # ================= 路径与数据配置 =================
 HDFS_ROOT=${HDFS_ROOT:-$PWD}
-DATA_ROOT=${DATA_ROOT:-/opt/huawei/dataset/zyr_yuyin/lyf/verl-nanoclaw-rl/nanoclawRL_temp_ckpt}
+DATA_ROOT=${DATA_ROOT:-/nanoclawRL_temp_ckpt}
 
 # Nanoclaw 数据输入支持两种目录，优先推荐 0625 扁平格式：
 #   base_tasks/data_*/env_builder.py
@@ -368,7 +355,7 @@ DATA_ROOT=${DATA_ROOT:-/opt/huawei/dataset/zyr_yuyin/lyf/verl-nanoclaw-rl/nanocl
 #   base_tasks/data_*/workplace_verifier.py
 #   base_tasks/data_*/manifest.json
 # 也兼容旧格式：base_tasks/tasks/data_* + base_tasks/scripts|scrips/data_*。
-DEFAULT_NANOCLAW_BASE_TASKS=${DEFAULT_NANOCLAW_BASE_TASKS:-/opt/huawei/dataset/zyr_yuyin/lyf/datasets/nanoclawRLdata/0710_add1000agent_qwen3_7_max_v1/exported_new_data}
+DEFAULT_NANOCLAW_BASE_TASKS=${DEFAULT_NANOCLAW_BASE_TASKS:-/exported_new_data}
 train_base_tasks=${TRAIN_DATA_PATH:-${BASE_TASKS:-${DEFAULT_NANOCLAW_BASE_TASKS}}}
 val_base_tasks=${VAL_DATA_PATH:-${VAL_BASE_TASKS:-${train_base_tasks}}}
 train_files="['$train_base_tasks']"
@@ -383,8 +370,8 @@ if [ ! -d "${val_base_tasks}" ]; then
     exit 2
 fi
 
-model_path=${MODEL_PATH:-/opt/huawei/dataset/zyr_yuyin/models/Qwen/Qwen3___5-9B}
-verifier_model_path=${VERIFIER_MODEL_PATH:-/opt/huawei/dataset/zyr_yuyin/models/Qwen/Qwen3___5-9B}
+model_path=${MODEL_PATH:-/Qwen3___5-9B}
+verifier_model_path=${VERIFIER_MODEL_PATH:-/Qwen3___5-9B}
 # 纯 VERL engine 路线：不要使用 MindSpeed-MM YAML。
 unset MM_CONFIG_FILE || true
 
@@ -393,7 +380,7 @@ tool_config_path=${TOOL_CONFIG_PATH:-nanoclaw_recipe/nanoclaw_tool_config.yaml}
 nanoclaw_task_glob=${NANOCLAW_TASK_GLOB:-data_*}
 nanoclaw_task_ids=${NANOCLAW_TASK_IDS:-}
 # 多机训练必须用所有节点都能访问的共享目录；不要用 /tmp，否则 reward worker 可能跨节点找不到 workspace。
-nanoclaw_temp_root=${NANOCLAW_TEMP_ROOT:-/opt/huawei/dataset/zyr_yuyin/lyf/verl-nanoclaw-rl/nanoclawRL_temp_workplace_v14_qwen35_9b_sp1_fsdp2_31k_longseq_allopt}
+nanoclaw_temp_root=${NANOCLAW_TEMP_ROOT:-/nanoclawRL_temp_workplace_v14_qwen35_9b_sp1_fsdp2_31k_longseq_allopt}
 # 默认保留每个 step/data_sample 的目录，方便复盘每条 GRPO 采样；磁盘紧张时手动设 NANOCLAW_CLEANUP_WORKSPACES=True。
 nanoclaw_cleanup_workspaces=${NANOCLAW_CLEANUP_WORKSPACES:-False}
 nanoclaw_keep_failed_workspaces=${NANOCLAW_KEEP_FAILED_WORKSPACES:-False}