{ "dataset_file": "construction_risk_natural_scenarios.jsonl", "original_samples": 551, "enhanced_samples": 1145, "cause_weights": { "Caught In Between": 8.193548387096774, "Struck By Object": 3.097560975609756, "Electrocution": 3.4324324324324325, "Falls": 1.0, "Transportation Incidents": 3.4324324324324325, "Other/Unspecified": 12.7, "Contact With Objects": 16.933333333333334, "Exposure Harmful Substances": 254.0 }, "cause_distribution": { "Caught In Between": 31, "Struck By Object": 82, "Electrocution": 74, "Falls": 254, "Transportation Incidents": 74, "Other/Unspecified": 20, "Contact With Objects": 15, "Exposure Harmful Substances": 1 }, "degree_distribution": { "High": 368, "Medium": 165, "Low": 18 }, "top_hazards": { "construction": 338, "blunt force trauma": 258, "blunt force": 196, "fall": 171, "fall from elevation": 81, "construction vehicle": 72, "crushed": 53, "fall protection": 51, "chest": 49, "commercial building": 41, "fracture": 39, "back": 38, "concrete floor": 38, "concrete": 34, "caught between": 32, "carpenter": 30, "contact": 29, "arm": 29, "head": 24, "concussion": 24 }, "model_config": { "base_model": "distilgpt2", "max_length": 384, "lora_r": 16, "lora_alpha": 32, "learning_rate": 0.0003, "epochs": 4 } }