sft-qwen2.5-1.5b/eval-results/gsm8k/metrics.json

{
  "gsm8k": {
    "pass@1": {
      "num_entries": 1319,
      "avg_tokens": 322,
      "gen_seconds": 43,
      "symbolic_correct": 74.14708112206216,
      "no_answer": 0.37907505686125853
    }
  }
}