training_results.json

{
  "experiment_summary": {
    "timestamp": "2025-05-25T19:24:48.173944",
    "model_name": "Qwen/Qwen3-0.6B-Base",
    "use_context_training": true,
    "use_context_evaluation": false,
    "max_samples": null
  },
  "results": {
    "baseline_accuracy": 0.6849514563106797,
    "fine_tuned_accuracy": 0.7169902912621359,
    "improvement": 0.03203883495145621,
    "improvement_percentage": "+3.20%"
  },
  "training_config": {
    "timestamp": "2025-05-25T16:08:49.325915",
    "model_name": "Qwen/Qwen3-0.6B-Base",
    "data_path": "data/mcqa_training_format.tsv",
    "use_context": true,
    "eval_with_context": false,
    "epochs": 5,
    "batch_size": 2,
    "learning_rate": 5e-05,
    "test_size": 0.1,
    "max_samples": null,
    "output_dir": "./qwen-mcqa-ft",
    "baseline_only": false,
    "eval_only": false,
    "device": "cuda",
    "torch_version": "2.5.1+cu124"
  },
  "success": true
}