{ "title": "Dynamic Service Generation", "short_description": "Generate IoT services dynamically from task descriptions and runtime context.", "long_description": "Given a task description and optional baseline context (e.g., existing schemas), generate a complete IoT service implementation at runtime. Submissions are evaluated using CodeBERTScore metrics (Precision, Recall, F1, F3) measuring semantic similarity between the generated service and a reference implementation. Efficiency is also assessed via token consumption.", "paper_link": "https://arxiv.org/abs/2502.00689", "metrics": [ { "name": "Precision", "description": "CodeBERTScore precision measuring code correctness" }, { "name": "Recall", "description": "CodeBERTScore recall measuring code completeness" }, { "name": "F1", "description": "CodeBERTScore F1 measuring overall code similarity" }, { "name": "F3", "description": "CodeBERTScore F3 emphasizing code completeness" }, { "name": "Tokens", "description": "Total token usage for service generation" } ], "entries": [ { "name": "CodeQwen1.5-7B", "precision": 0.86, "recall": 0.79, "f1": 0.83, "f3": 0.80, "tokens": 3482.00, "date": "2025-02-02", "link": "https://github.com/sa4s-serc/SAS_llm_query" }, { "name": "DeepSeek-V2.5", "precision": 0.91, "recall": 0.85, "f1": 0.88, "f3": 0.86, "tokens": 4376.25, "date": "2025-02-02", "link": "https://github.com/sa4s-serc/SAS_llm_query" }, { "name": "GPT-4o-mini", "precision": 0.90, "recall": 0.85, "f1": 0.87, "f3": 0.85, "tokens": 2063.17, "date": "2025-02-02", "link": "https://github.com/sa4s-serc/SAS_llm_query" } ], "type": "dynamic", "token_limit": "8K", "response_time": "30 seconds", "example_available": true, "dataset_download": true, "dataset_link": "https://github.com/sa4s-serc/SAS_llm_query/blob/iot-prototype/dataset.csv" }