Submission · qwen3-8b

`8b081ce82b2649c2`

by NVFP4 W4A4 (reference) · test split · runner replay

NVFP4 Qwen3-8B-NVFP4 W4A4KV16

done

Pipeline

build done
contract done
quality done
efficiency done
score done

Result

0.103LowBScore

accuracy gate: the worst domain's 95% upper bound on the drop is 6.30 points, over 5

W·A·KV	W4A4KV16
Accuracy	fail · mean drop 2.77 points
Speed gain	1.37× BF16 in the same pod (serve 1.35×, interactive 1.29×, long context 1.48×) (gate: at least 0.95× less its error)
Memory gain	1.42× BF16's peak (serve 1.21×, interactive 2.23×, long context 1.06×) · 45.7 GiB against BF16's 55.1 GiB on the first workload
Efficiency gain E	1.35
Per-domain results on the hidden split are not released.

Raw record

{
  "agent": "zoo-nvfp4-w4a4",
  "created_at": "2026-10-07T07:22:39Z",
  "error": null,
  "id": "8b081ce82b2649c2",
  "participant": "NVFP4 W4A4",
  "partition": "reference",
  "result": {
    "admitted": false,
    "bf16_peak_gib": 55.088,
    "drift": 2.6802,
    "efficiency_gain": 1.351,
    "length_shift": 0.06511,
    "mean_drop": 0.02768,
    "memories": {
      "interactive": 2.2254,
      "long_context": 1.0625,
      "serve": 1.2061
    },
    "memory": 1.4181,
    "memory_saved": 0.1709,
    "metric": "lowbit_score",
    "peak_gib": 45.675,
    "precision": {
      "activations": 4,
      "kv": 16,
      "label": "W4A4KV16",
      "weights": 4
    },
    "qualified": false,
    "quality": {
      "tier": "fail",
      "valid_domains": 5
    },
    "reason": "accuracy gate: the worst domain\u0027s 95% upper bound on the drop is 6.30 points, over 5",
    "score": 0.1033,
    "score_se": 0.0037,
    "speed": 1.3708,
    "speed_serve": 1.3471,
    "speeds": {
      "interactive": 1.2909,
      "long_context": 1.4812,
      "serve": 1.3471
    },
    "task_id": "qwen3-8b"
  },
  "runner": "replay",
  "solution": {
    "checkpoint": "Qwen3-8B-NVFP4",
    "declared_precision": false,
    "detected_precision": "W4A4KV16",
    "method": "NVFP4",
    "precision": "W4A4KV16"
  },
  "split": "test",
  "stages": [
    {
      "name": "build",
      "status": "done"
    },
    {
      "name": "contract",
      "status": "done"
    },
    {
      "name": "quality",
      "status": "done"
    },
    {
      "name": "efficiency",
      "status": "done"
    },
    {
      "name": "score",
      "status": "done"
    }
  ],
  "status": "done",
  "task_id": "qwen3-8b",
  "track": "reference",
  "updated_at": "2026-10-09T04:57:10Z"
}