Submission · qwen3-8b

`49c48eef866f4357`

by FP8 W8A8 + FP8 KV (reference) · test split · runner replay

FP8 per channel + FP8 KV Qwen3-8B-FP8-Dynamic W8A8KV8

done

Pipeline

build done
contract done
quality done
efficiency done
score done

Result

1.455LowBScore

Qualified: the accuracy, speed, and length gates passed.

W·A·KV	W8A8KV8
Accuracy	tier A · mean drop 0.63 points
Speed gain	1.33× BF16 in the same pod (serve 1.35×, interactive 1.24×, long context 1.41×) (gate: at least 0.95× less its error)
Memory gain	1.81× BF16's peak (serve 1.87×, interactive 1.63×, long context 1.95×) · 29.4 GiB against BF16's 55.1 GiB on the first workload
Efficiency gain E	1.54
Per-domain results on the hidden split are not released.

Raw record

{
  "agent": "zoo-fp8-w8a8-kvfp8",
  "created_at": "2026-10-07T07:22:39Z",
  "error": null,
  "id": "49c48eef866f4357",
  "participant": "FP8 W8A8 + FP8 KV",
  "partition": "reference",
  "result": {
    "admitted": true,
    "bf16_peak_gib": 55.084,
    "drift": 0.5813,
    "efficiency_gain": 1.5423,
    "length_shift": 0.01556,
    "mean_drop": 0.00626,
    "memories": {
      "interactive": 1.6347,
      "long_context": 1.9538,
      "serve": 1.8706
    },
    "memory": 1.8145,
    "memory_saved": 0.4654,
    "metric": "lowbit_score",
    "peak_gib": 29.448,
    "precision": {
      "activations": 8,
      "kv": 8,
      "label": "W8A8KV8",
      "weights": 8
    },
    "qualified": true,
    "quality": {
      "tier": "A",
      "valid_domains": 5
    },
    "reason": null,
    "score": 1.4552,
    "score_se": 0.0389,
    "speed": 1.3314,
    "speed_serve": 1.3545,
    "speeds": {
      "interactive": 1.2394,
      "long_context": 1.4057,
      "serve": 1.3545
    },
    "task_id": "qwen3-8b"
  },
  "runner": "replay",
  "solution": {
    "checkpoint": "Qwen3-8B-FP8-Dynamic",
    "declared_precision": false,
    "detected_precision": "W8A8KV8",
    "method": "FP8 per channel + FP8 KV",
    "precision": "W8A8KV8"
  },
  "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"
}