Cell

complex_deepseek-v4-flash_final2_la3_t06

DeepSeek-V4-Flash on the complex floor, 5 episodes. Not cited by the thesis (development or earlier run).

Outcome
Wins (boss defeated, alive)0/5 [0.00, 0.43]
Final HP / exact optimum, mean over wins–
Died0
Final HP, mean81
Seconds per model call67.9
Parse error rate0.0%
Settings
FloorComplex floor
ModelDeepSeek-V4-Flash (deepseek-v4-flash-0731, thinking)
Planningreactive
Promptengine-default-floor-basic
Temperature0.6
Lookahead depth3
Mechanical overrideson (logged per turn)
Engine settings (environment, 5)
THINKING_MAX_TOKENS=32768
CONVERSATION_HISTORY_TURNS=0
MOTA_ORDER_TWO_SIDED=1
MOTA_FIGHT_KEY_FLOOD=1
LLM_TIMEOUT=300
Episodesopen one to replay it turn by turn
#outcomefinal HPHP / optstepsmodel callsreplay
1not won123–3715replay
2not won55–3514replay
3not won191–2611replay
4not won11–3414replay
5not won25–3715replay
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_deepseek-v4-flash_final2_la3_t06, never over the thesis result):

python scripts/run_catalog.py --codes T6DCB

Result file: results/reliability/complex_deepseek-v4-flash_final2_la3_t06.json · updated 2026-09-30