segment1_qwen3.5-397b_basic_k1k2
Qwen3.5-397B on the tower (MT5→MT8), 2 runs. Not cited by the thesis (development or earlier run).
| Highest floor reached, per run | MT7, MT7 |
| Died | 2 |
| Final HP, mean | -1 |
| Seconds per model call | 30.8 |
| Parse error rate | 5.1% |
| Floor | Tower MT5→MT8 |
| 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) |
| Walkthrough checkpoint | 1 |
Engine settings (environment, 1)
THINKING_MAX_TOKENS=8192
In a checkout of mt-ai-lab with an API key set (the result goes to a new file rerun_<time>_segment1_qwen3.5-397b_basic_k1k2, never over the thesis result):
python scripts/run_catalog.py --codes W1b
Result file: results/reliability/segment1_qwen3.5-397b_basic_k1k2.json · updated 2026-09-26