diff options
| author | YurenHao0426 <blackhao0426@gmail.com> | 2026-06-13 12:35:36 -0500 |
|---|---|---|
| committer | YurenHao0426 <blackhao0426@gmail.com> | 2026-06-13 12:35:36 -0500 |
| commit | 66e0d8b9fd4d0f7a2231d689c055e26fdf1cf04a (patch) | |
| tree | c29cba61124018755a19b02c9d33e3ad5f2e05cc /scripts/run_hrm_sudoku.sh | |
Curated export for clone-and-run Maze training (2x A6000) + diagnostics.
trm/hrm pretrain.py carry trajectory-augmentation code (backward-compatible).
Heavy artifacts (checkpoints/wandb/npz) gitignored; see PROVENANCE.md.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Diffstat (limited to 'scripts/run_hrm_sudoku.sh')
| -rwxr-xr-x | scripts/run_hrm_sudoku.sh | 15 |
1 files changed, 15 insertions, 0 deletions
diff --git a/scripts/run_hrm_sudoku.sh b/scripts/run_hrm_sudoku.sh new file mode 100755 index 0000000..04d3a1c --- /dev/null +++ b/scripts/run_hrm_sudoku.sh @@ -0,0 +1,15 @@ +#!/usr/bin/env bash +# 启动 HRM Sudoku 1k 训练 (HRM 官方推荐配置) +# 单 GPU 约 10h on RTX 4070; A6000 应该更快 +set -euo pipefail +REPO_ROOT="$(cd "$(dirname "$0")/.." && pwd)" +source "$(conda info --base)/etc/profile.d/conda.sh" +conda activate rrm + +cd "$REPO_ROOT/hrm" +OMP_NUM_THREADS=${OMP_NUM_THREADS:-8} \ +WANDB_MODE=${WANDB_MODE:-online} \ +python pretrain.py \ + data_path="$REPO_ROOT/data/sudoku-extreme-1k-aug-1000" \ + epochs=20000 eval_interval=2000 global_batch_size=384 \ + lr=7e-5 puzzle_emb_lr=7e-5 weight_decay=1.0 puzzle_emb_weight_decay=1.0 "$@" |