hand01_qwen3.6-35b_final2
Qwen3.6-35B on hand01, 3 episodes. Not cited by the thesis (development or earlier run).
| Wins (boss defeated, alive) | 3/3 [0.44, 1.00] |
| Final HP / exact optimum, mean over wins | 64% |
| Died | 0 |
| Final HP, mean | 9 |
| Seconds per model call | 16.4 |
| Parse error rate | 0.0% |
| Floor | hand01 |
| Model | Qwen3.6-35B (qwen3.6-35b-a3b, thinking) |
| Planning | reactive |
| Prompt | engine-default-floor-basic |
| Temperature | 0 |
| Lookahead depth | 3 |
| Mechanical overrides | on (logged per turn) |
Engine settings (environment, 4)
THINKING_MAX_TOKENS=16384 CONVERSATION_HISTORY_TURNS=0 MOTA_ORDER_TWO_SIDED=1 MOTA_FIGHT_KEY_FLOOD=1
In a checkout of mt-ai-lab with an API key set (the result goes to a new file rerun_<time>_hand01_qwen3.6-35b_final2, never over the thesis result):
python scripts/run_catalog.py --codes HS1
Result file: results/reliability/hand01_qwen3.6-35b_final2.json · updated 2026-09-29