selection_metrics.json

{
  "benchmark_scores_for_selected_checkpoint": {
    "code_generation": 0.65,
    "common_sense": 0.736,
    "creative_writing": 0.61,
    "dialogue_generation": 0.644,
    "instruction_following": 0.758,
    "knowledge_retrieval": 0.676,
    "logical_reasoning": 0.819,
    "math_reasoning": 0.55,
    "question_answering": 0.607,
    "reading_comprehension": 0.7,
    "safety_evaluation": 0.739,
    "sentiment_analysis": 0.792,
    "summarization": 0.767,
    "text_classification": 0.828,
    "translation": 0.804
  },
  "eval_accuracy_by_checkpoint": {
    "step_100": 0.48,
    "step_1000": 0.71,
    "step_200": 0.535,
    "step_300": 0.576,
    "step_400": 0.608,
    "step_500": 0.635,
    "step_600": 0.656,
    "step_700": 0.674,
    "step_800": 0.689,
    "step_900": 0.7
  },
  "note": "No explicit saved eval_accuracy key was found in checkpoint files; values were computed from the workspace evaluation logic.",
  "selected_checkpoint": "step_1000",
  "selected_eval_accuracy": 0.71,
  "selection_metric": "eval_accuracy"
}