Cell · cited by the thesis

hand03_deepseek-v4-flash_canon

DeepSeek-V4-Flash on hand03, 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, mean126
Seconds per model call42.4
Parse error rate1.4%
Settings
Floorhand03
ModelDeepSeek-V4-Flash (deepseek-v4-flash-0731, thinking)
Planningreactive
Promptengine-default-floor-basic
Temperature0
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=1800
Episodesopen one to replay it turn by turn
#outcomefinal HPHP / optstepsmodel callsreplay
1not won98–3614replay
2not won135–3715replay
3not won132–3715replay
4not won132–3715replay
5not won132–3715replay
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>_hand03_deepseek-v4-flash_canon, never over the thesis result):

python scripts/run_catalog.py --codes VD3 VD3a VD35

Result file: results/reliability/hand03_deepseek-v4-flash_canon.json · updated 2026-10-03