Cell · cited by the thesis

hand03_qwen3.5-397b_canon

Qwen3.5-397B 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, mean47
Seconds per model call63.0
Parse error rate0.0%
Settings
Floorhand03
ModelQwen3.5-397B (qwen3.5-397b-a17b, 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=300
Episodesopen one to replay it turn by turn
#outcomefinal HPHP / optstepsmodel callsreplay
1not won32–3715replay
2not won114–3513replay
3not won47–3614replay
4not won4–3513replay
5not won40–3614replay
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_qwen3.5-397b_canon, never over the thesis result):

python scripts/run_catalog.py --codes VQ3 VQ3a VQ35

Result file: results/reliability/hand03_qwen3.5-397b_canon.json · updated 2026-10-04