Benchmarks/Seed-OSS-36B-Instruct-unsloth-Q4_K_M/bench_metrics.json

{
  "llamabench": {
    "backend": "CUDA",
    "log_path": "Seed-OSS-36B-Instruct-unsloth-Q4_K_M/llamabench.md",
    "ngl": "35",
    "raw_row": {
      "backend": "CUDA",
      "model": "seed_oss 36B Q4_K - Medium",
      "ngl": "35",
      "params": "36.15 B",
      "size": "20.26 GiB",
      "t/s": "26.65 \u00b1 0.22",
      "test": "pp8",
      "tps_value": 26.65
    },
    "test": "pp8",
    "tps": 26.65
  },
  "perplexity": {
    "code": {
      "log_path": "Seed-OSS-36B-Instruct-unsloth-Q4_K_M/perplexity_code.log",
      "ppl": 1.4235,
      "ppl_error": 0.00979
    },
    "general": {
      "log_path": "Seed-OSS-36B-Instruct-unsloth-Q4_K_M/perplexity_general.log",
      "ppl": 7.097,
      "ppl_error": 0.17602
    },
    "math": {
      "log_path": "Seed-OSS-36B-Instruct-unsloth-Q4_K_M/perplexity_math.log",
      "ppl": 5.7134,
      "ppl_error": 0.13053
    }
  }
}