| set -euo pipefail | |
| export REPOBRIDGE_OUTPUT_DIR="/home/student/YouZheng/jobs/taotern/taotern-200m-branch-only-chat-20260514/outputs" | |
| export TAOTERN_CHECKPOINT_DIR="/home/student/YouZheng/jobs/taotern/taotern-200m-branch-only-chat-20260514/checkpoints" | |
| cd "/home/student/YouZheng/repo" | |
| REMOTE_REPO=/home/student/YouZheng/repo PYTHON_BIN=/home/student/.venv/bin/python SSM_REPO_PATH=/home/student/YouZheng/gamma_ssm_repo DATA_PATH=/home/student/Data/TaoData/pretrain.jsonl SFT_DATA_PATH=/home/student/Data/TaoData/sft.jsonl TOKENIZER_PATH=/home/student/YouZheng/tokenizers/taodata_pilot_8k/tokenizer.model SEQ_LEN=512 BATCH_SIZE=8 PRETRAIN_TOKENS=4000000000 SFT_STEPS=50000 PRETRAIN_LR=0.0008 SFT_LR=0.00005 WEIGHT_DECAY=0.01 LOG_EVERY=100 SAVE_EVERY=100000 SFT_SAVE_EVERY=10000 TOKENIZER_THREADS=8 SAMPLES_PER_CHUNK=2000 BLOCK_RESIDUAL_RMS_CAP= bash scripts/remote/run_200m_branch_only_chat.sh | |