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 /research/flossing/launch_trm_official_gbs768_multi4.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 'research/flossing/launch_trm_official_gbs768_multi4.sh')
| -rwxr-xr-x | research/flossing/launch_trm_official_gbs768_multi4.sh | 37 |
1 files changed, 37 insertions, 0 deletions
diff --git a/research/flossing/launch_trm_official_gbs768_multi4.sh b/research/flossing/launch_trm_official_gbs768_multi4.sh new file mode 100755 index 0000000..5bd9575 --- /dev/null +++ b/research/flossing/launch_trm_official_gbs768_multi4.sh @@ -0,0 +1,37 @@ +#!/usr/bin/env bash +set -eo pipefail + +GPU="${1:-0}" + +source /home/yurenh2/miniconda3/etc/profile.d/conda.sh +conda activate rrm + +cd /home/yurenh2/rrm/trm +export WANDB_MODE=offline +export CUDA_VISIBLE_DEVICES="${GPU}" + +python pretrain.py \ + arch=trm \ + data_paths='[/home/yurenh2/rrm/data/sudoku-extreme-1k-aug-1000]' \ + data_paths_test='[]' \ + evaluators='[]' \ + +project_name='Sudoku-extreme-1k-aug-1000-ACT-torch' \ + +run_name='pretrain_mlp_t_sudoku_official_gbs768_multi4_loguniform_repro' \ + +checkpoint_path='checkpoints/Sudoku-extreme-1k-aug-1000-ACT-torch/pretrain_mlp_t_sudoku_official_gbs768_multi4_loguniform_repro' \ + +load_checkpoint=null \ + epochs=50000 eval_interval=2500 min_eval_interval=0 checkpoint_every_eval=true \ + global_batch_size=768 \ + lr=0.0001 lr_min_ratio=1.0 lr_warmup_steps=2000 \ + beta1=0.9 beta2=0.95 weight_decay=1.0 \ + puzzle_emb_lr=0.0001 puzzle_emb_weight_decay=1.0 \ + ema=true ema_rate=0.999 freeze_weights=false \ + arch.mlp_t=true arch.pos_encodings=none arch.L_layers=2 arch.H_cycles=3 arch.L_cycles=6 \ + +trajectory_augment=true \ + +trajectory_n=4 \ + +trajectory_noise_std=0.001 \ + +trajectory_noise_min=0.00003 \ + +trajectory_noise_max=0.003 \ + +trajectory_noise_sampling=loguniform \ + +trajectory_sigma_start=0.0 \ + +trajectory_sigma_ramp_steps=5000 \ + +trajectory_perturb=both |