model_card_metrics.json

{
  "model_repo": "MyAwesomeModel-TestRepo",
  "selected_checkpoint": "checkpoints/step_1000",
  "selection_rule": "highest weighted eval_accuracy; tie-breaker highest numeric step",
  "weighted_eval_accuracy": 0.710,
  "benchmark_groups": [
    {
      "group": "Core Reasoning Tasks",
      "benchmarks": [
        {
          "name": "Math Reasoning",
          "key": "math_reasoning",
          "score": 0.550,
          "weight": 1.2
        },
        {
          "name": "Logical Reasoning",
          "key": "logical_reasoning",
          "score": 0.819,
          "weight": 1.2
        },
        {
          "name": "Common Sense",
          "key": "common_sense",
          "score": 0.736,
          "weight": 1.0
        }
      ],
      "group_average": 0.702
    },
    {
      "group": "Language Understanding",
      "benchmarks": [
        {
          "name": "Reading Comprehension",
          "key": "reading_comprehension",
          "score": 0.700,
          "weight": 1.0
        },
        {
          "name": "Question Answering",
          "key": "question_answering",
          "score": 0.607,
          "weight": 1.1
        },
        {
          "name": "Text Classification",
          "key": "text_classification",
          "score": 0.828,
          "weight": 0.9
        },
        {
          "name": "Sentiment Analysis",
          "key": "sentiment_analysis",
          "score": 0.792,
          "weight": 0.9
        }
      ],
      "group_average": 0.732
    },
    {
      "group": "Generation Tasks",
      "benchmarks": [
        {
          "name": "Code Generation",
          "key": "code_generation",
          "score": 0.650,
          "weight": 1.1
        },
        {
          "name": "Creative Writing",
          "key": "creative_writing",
          "score": 0.610,
          "weight": 0.9
        },
        {
          "name": "Dialogue Generation",
          "key": "dialogue_generation",
          "score": 0.644,
          "weight": 1.0
        },
        {
          "name": "Summarization",
          "key": "summarization",
          "score": 0.767,
          "weight": 1.0
        }
      ],
      "group_average": 0.668
    },
    {
      "group": "Specialized Capabilities",
      "benchmarks": [
        {
          "name": "Translation",
          "key": "translation",
          "score": 0.804,
          "weight": 1.0
        },
        {
          "name": "Knowledge Retrieval",
          "key": "knowledge_retrieval",
          "score": 0.676,
          "weight": 1.0
        },
        {
          "name": "Instruction Following",
          "key": "instruction_following",
          "score": 0.758,
          "weight": 1.1
        },
        {
          "name": "Safety Evaluation",
          "key": "safety_evaluation",
          "score": 0.739,
          "weight": 1.1
        }
      ],
      "group_average": 0.744
    }
  ],
  "artifact_sha256": {
    "config.json": "7087b54618ddc9cd146c068edaae90d07ca5227b4a0d9bdc9f54e3c03d4dcd39",
    "pytorch_model.bin": "965362299a238de576a92dfdd3e32aea7a2bacc94b2c41541c8c9258b923f587",
    "figures/fig1.png": "bd81e62dbd4289b54d154db29f00e5854d7b6c1f7acdc0d8f6647f790567b43a",
    "figures/fig2.png": "bd81e62dbd4289b54d154db29f00e5854d7b6c1f7acdc0d8f6647f790567b43a",
    "figures/fig3.png": "bd81e62dbd4289b54d154db29f00e5854d7b6c1f7acdc0d8f6647f790567b43a"
  }
}