Cell

tower_qwen3.5-397b_final5_reactive

Qwen3.5-397B on the tower (MT0→MT23w), 2 runs. Not cited by the thesis (development or earlier run).

Outcome
Highest floor reached, per runMT8, MT8
Died0
Final HP, mean1237
Seconds per model call36.6
Parse error rate1.8%
Settings
FloorTower MT0→MT23w
ModelQwen3.5-397B (qwen3.5-397b-a17b, thinking)
Planningreactive
Promptengine-default-tower-basic
Temperature0
Lookahead depth2
Mechanical overrideson (logged per turn)
Engine settings (environment, 8)
THINKING_MAX_TOKENS=8192
CONVERSATION_HISTORY_TURNS=0
MOTA_ORDER_TWO_SIDED=1
MOTA_FIGHT_KEY_FLOOD=1
MOTA_SHOP_ROUTE=1
MOTA_REMOTE_ACTIONS=1
MOTA_WAY_UP=1
LLM_TIMEOUT=300
Runsopen one to replay it turn by turn
#outcomefinal HPHP / optturnsmodel callsreplay
1MT8332–20787replay
2MT82141–20484replay
Re-run this cell

In a checkout of mt-ai-lab with an API key set (the result goes to a new file rerun_<time>_tower_qwen3.5-397b_final5_reactive, never over the thesis result):

python scripts/run_catalog.py --codes T5QR T5QRa T5QRb

Result file: results/reliability/tower_qwen3.5-397b_final5_reactive.json · updated 2026-10-01