Initialize rebalanced 2B MoE project

scripts/README.md

@@ -0,0 +1,5 @@
+# Scripts
+
+- `sync_pretrain_data_into_repo.sh`: 200B 数据构建完成后，把数据目录同步到 `dataset/pretrain/data/`，默认优先 hardlink。
+- `wait_and_sync_pretrain_data.sh`: 后台等待当前 200B 构建进程结束，然后自动同步数据。
+- `train_nemo_megatron_2b_moe.sh`: NeMo/Megatron 训练入口占位，包含 image、mount、路径检查。

scripts/sync_pretrain_data_into_repo.sh

@@ -0,0 +1,32 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+SRC="${SRC:-/mnt/beegfs/yi/laoyao_2b_moe_pretraining_dataset/train/pretrain_rebalanced_web40_edu20_chinese10_science10_logic10_math5_code5_200b_v1_20260701}"
+DST="${DST:-/mnt/beegfs/yi/laoyao_2b_moe/dataset/pretrain/data/pretrain_rebalanced_web40_edu20_chinese10_science10_logic10_math5_code5_200b_v1_20260701}"
+LOG="${LOG:-/mnt/beegfs/yi/laoyao_2b_moe_pretraining_dataset/logs/wait_build_rebalanced_200b_20260701_130058.log}"
+
+mkdir -p "$(dirname "$DST")"
+
+if pgrep -f "build_rebalanced_pretrain_dataset.py --target-tokens 200000000000" >/dev/null; then
+  echo "pretrain builder is still running; not syncing yet" >&2
+  exit 2
+fi
+
+if [ ! -d "$SRC" ]; then
+  echo "source dataset directory does not exist: $SRC" >&2
+  exit 1
+fi
+
+if ! grep -q "rebalanced_pretrain_done\|completed\|wrote manifest\|finished" "$LOG" 2>/dev/null; then
+  echo "warning: completion marker not found in log; syncing existing directory anyway" >&2
+fi
+
+rm -rf "$DST"
+mkdir -p "$DST"
+if cp -al "$SRC"/. "$DST"/ 2>/tmp/laoyao_sync_cp_al.err; then
+  echo "synced with hardlinks: $DST"
+else
+  echo "hardlink copy failed, falling back to rsync copy" >&2
+  cat /tmp/laoyao_sync_cp_al.err >&2 || true
+  rsync -a --info=progress2 "$SRC"/ "$DST"/
+fi

scripts/train_nemo_megatron_2b_moe.sh

@@ -0,0 +1,20 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+REPO_ROOT="${REPO_ROOT:-/mnt/beegfs/yi/laoyao_2b_moe}"
+IMAGE="${IMAGE:-nvcr.io/nvidia/nemo:26.06}"
+CONFIG="${CONFIG:-$REPO_ROOT/training/nemo_megatron/pretrain_2b_moe_200b.yaml}"
+DATA_DIR="$REPO_ROOT/dataset/pretrain/data/pretrain_rebalanced_web40_edu20_chinese10_science10_logic10_math5_code5_200b_v1_20260701"
+
+if [ ! -d "$DATA_DIR" ]; then
+  echo "pretrain data is missing under repo; run scripts/sync_pretrain_data_into_repo.sh after the 200B build finishes" >&2
+  exit 1
+fi
+
+mkdir -p "$REPO_ROOT/runs" "$REPO_ROOT/logs"
+
+docker run --rm --gpus all --ipc=host --network=host \
+  -v /mnt/beegfs:/mnt/beegfs \
+  -w "$REPO_ROOT" \
+  "$IMAGE" \
+  bash -lc "echo TODO: launch NeMo/Megatron pretraining with $CONFIG; ls -lh $CONFIG"

scripts/wait_and_sync_pretrain_data.sh

@@ -0,0 +1,8 @@
+#!/usr/bin/env bash
+set -euo pipefail
+INTERVAL_SECONDS="${INTERVAL_SECONDS:-300}"
+while pgrep -f "build_rebalanced_pretrain_dataset.py --target-tokens 200000000000" >/dev/null; do
+  date "+[%F %T] pretrain builder still running; waiting..."
+  sleep "$INTERVAL_SECONDS"
+done
+bash /mnt/beegfs/yi/laoyao_2b_moe/scripts/sync_pretrain_data_into_repo.sh