Cell · cited by the thesis

simple_mistral-medium-3.5_canon

Mistral-Medium-3.5 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, mean550
Seconds per model call48.5
Parse error rate2.4%
Settings
FloorSimple floor
ModelMistral-Medium-3.5 (mistral-medium-3.5-128b, instruct)
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=300
Episodesopen one to replay it turn by turn
#outcomefinal HPHP / optstepsmodel callsreplay
1not won550–3616replay
2not won550–3617replay
3not won550–3617replay
4not won550–3617replay
5not won550–3617replay
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_mistral-medium-3.5_canon, never over the thesis result):

python scripts/run_catalog.py --codes VMS VMSa VMS5

Result file: results/reliability/simple_mistral-medium-3.5_canon.json · updated 2026-10-03