Cell · cited by the thesis

simple_qwen3.8-27b_final2_plan2_t06

Qwen3.8-27B on the simple floor, 3 episodes. Cited by the thesis in:

Outcome
Wins (boss defeated, alive)2/3 [0.21, 0.94]
Final HP / exact optimum, mean over wins15%
Died0
Final HP, mean180
Seconds per model call383.4
Parse error rate3.6%
Settings
FloorSimple floor
ModelQwen3.8-27B (qwen3.8-27b, thinking)
Planningreactive
Promptengine-default-floor-basic
Temperature0.6
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=2
LLM_TIMEOUT=1800
Episodesopen one to replay it turn by turn
#outcomefinal HPHP / optstepsmodel callsreplay
1won215%369replay
2won215%369replay
3not won535–3610replay
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_qwen3.8-27b_final2_plan2_t06, never over the thesis result):

python scripts/run_catalog.py --codes SQSP2

Result file: results/reliability/simple_qwen3.8-27b_final2_plan2_t06.json · updated 2026-09-30