tower_qwen3.5-397b_final3_reactive
Qwen3.5-397B on the tower (MT0→MT23w), 1 run. Cited by the thesis in:
| Highest floor reached, per run | MT8 |
| Died | 0 |
| Final HP, mean | 2220 |
| Seconds per model call | 36.0 |
| Parse error rate | 2.4% |
| Floor | Tower MT0→MT23w |
| Model | Qwen3.5-397B (qwen3.5-397b-a17b, thinking) |
| Planning | reactive |
| Prompt | engine-default-tower-basic |
| Temperature | 0 |
| Lookahead depth | 2 |
| Mechanical overrides | on (logged per turn) |
Engine settings (environment, 7)
THINKING_MAX_TOKENS=8192 CONVERSATION_HISTORY_TURNS=0 MOTA_ORDER_TWO_SIDED=1 MOTA_FIGHT_KEY_FLOOD=1 MOTA_SHOP_ROUTE=1 MOTA_REMOTE_ACTIONS=1 LLM_TIMEOUT=300
| # | outcome | final HP | HP / opt | turns | model calls | replay |
|---|---|---|---|---|---|---|
| 1 | MT8 | 2220 | – | 204 | 85 | replay |
In a checkout of mt-ai-lab with an API key set (the result goes to a new file rerun_<time>_tower_qwen3.5-397b_final3_reactive, never over the thesis result):
python scripts/run_catalog.py --codes T3QR
Result file: results/reliability/tower_qwen3.5-397b_final3_reactive.json · updated 2026-10-01