| { |
| "evaluation_set": "response_strategy_source_disjoint_eval_532", |
| "model": "qwen3_8b_method1_0shot", |
| "base_model": "Qwen/Qwen3-8B", |
| "evaluation_protocol": "autoregressive generation followed by the released output parser", |
| "source_disjoint": true, |
| "num_source_utterances": 100, |
| "num_examples": 532, |
| "num_parse_errors": 0, |
| "parse_error_rate": 0.0, |
| "macro_f1": 0.9857480574663244, |
| "micro_f1": 0.990301724137931, |
| "primary_strategy_accuracy": 0.9924812030075187, |
| "exact_match": 0.9793233082706767, |
| "hamming_loss": 0.0021146616541353382, |
| "jaccard_micro": 0.9807897545357525, |
| "jaccard_macro": 0.9728299106392682, |
| "jaccard_samples": 0.9906328320802005, |
| "average_true_labels_per_example": 1.7462406015037595, |
| "average_predicted_labels_per_example": 1.7424812030075187, |
| "release_note": "Sanitized source-disjoint validation metrics; local paths, runtime, and hardware identifiers removed." |
| } |
|
|
