Cell · cited by the thesis

simple_qwen3.5-397b_final2_maxsteps

Qwen3.5-397B on the simple 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, mean545
Seconds per model call134.6
Parse error rate8.7%
Settings
FloorSimple 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_STEPS=60
LLM_TIMEOUT=1800
Episodesopen one to replay it turn by turn
#outcomefinal HPHP / optstepsmodel callsreplay
1not won488–364replay
2not won542–366replay
3not won564–365replay
4not won565–364replay
5not won566–364replay
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>_simple_qwen3.5-397b_final2_maxsteps, never over the thesis result):

python scripts/run_catalog.py --codes PM PMa

Result file: results/reliability/simple_qwen3.5-397b_final2_maxsteps.json · updated 2026-09-30