rlvrmulti-qwen2.5-1.5b/eval-results/ifbench/metrics.json

{
  "ifbench": {
    "pass@1": {
      "num_prompts": 294,
      "num_instructions": 335,
      "average_score": 19.80150269062849,
      "prompt_strict_accuracy": 16.666666666666664,
      "instruction_strict_accuracy": 18.507462686567163,
      "prompt_loose_accuracy": 20.74829931972789,
      "instruction_loose_accuracy": 23.283582089552237,
      "num_entries": 294,
      "avg_tokens": 495,
      "gen_seconds": 23
    }
  }
}