Cell · cited by the thesis

complex_qwen3.5-397b_canon_planmem

Qwen3.5-397B on the complex floor, 5 episodes. Cited by the thesis in:

Outcome
Wins (boss defeated, alive)0/5 [0.00, 0.43]
Final HP / exact optimum, mean over wins–
Died0
Final HP, mean80
Seconds per model call42.7
Parse error rate1.4%
Settings
FloorComplex floor
ModelQwen3.5-397B (qwen3.5-397b-a17b, thinking)
Planningreactive
Promptengine-default-floor-basic
Temperature0
Lookahead depth3
Mechanical overrideson (logged per turn)
Engine settings (environment, 6)
THINKING_MAX_TOKENS=32768
CONVERSATION_HISTORY_TURNS=0
MOTA_ORDER_TWO_SIDED=1
MOTA_FIGHT_KEY_FLOOD=1
MOTA_PLAN_MEMORY=1
LLM_TIMEOUT=300
Episodesopen one to replay it turn by turn
#outcomefinal HPHP / optstepsmodel callsreplay
1not won86–3615replay
2not won6–3715replay
3not won100–3313replay
4not won80–3313replay
5not won130–3514replay
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>_complex_qwen3.5-397b_canon_planmem, never over the thesis result):

python scripts/run_catalog.py --codes PMQC

Result file: results/reliability/complex_qwen3.5-397b_canon_planmem.json · updated 2026-10-04