model: deepseek/deepseek-r1 provider: Nebius output_dir: results/deepseek-r1_logic max_concurrent: 10 default_size: 50 default_seed: 45 categories: - category: logic datasets: - dataset: aiw - dataset: circuit_logic - dataset: knights_knaves - dataset: propositional_logic - dataset: self_reference - dataset: syllogism - dataset: zebra_puzzles