Submission · qwen35-27b
fec500147e0a4a89
by cyankiwi/Qwen3.5-27B-AWQ-4bit (reference) · test split · runner kubernetes
AWQ g32, cyankiwi cyankiwi/Qwen3.5-27B-AWQ-4bit
W4A16KV16
Pipeline
- build done
- contract done
- quality failed
- efficiency done
- score done
Result
pending
LowBScore
quality not measured
| W·A·KV | W4A16KV16 declared · the service reads W4A16KV16 |
| Drift from BF16 | KL 0.012 nats per token |
| Speed gain | — BF16 in the same pod (gate: at least 0.95× less its error) |
| Memory gain | — BF16's peak · — GiB against BF16's — GiB on the first workload |
| Per-domain results on the hidden split are not released. | |
2.34 GPU-hours
Raw record
{
"agent": "hf.cyankiwi.Qwen3.5-27B-AWQ-4bit",
"created_at": "2026-10-08T18:10:37Z",
"error": null,
"id": "fec500147e0a4a89",
"participant": "cyankiwi/Qwen3.5-27B-AWQ-4bit",
"partition": "reference",
"result": {
"admitted": null,
"anchor_efficiency": {
"interactive": 93.76499076331078,
"long_context": 314.96127726031233,
"serve": 848.5743142199557
},
"anchor_peak_memory": {
"interactive": {
"device_peak_gib": 161.798,
"kv_bytes_per_token": 200806,
"kv_gib": 0.239,
"kv_usage": 0.0023,
"other_gib": 4.555,
"peak_gib": 55.015,
"weights_gib": 50.22
},
"long_context": {
"device_peak_gib": 161.8,
"kv_bytes_per_token": 75895,
"kv_gib": 73.537,
"kv_usage": 0.6913,
"other_gib": 4.555,
"peak_gib": 128.313,
"weights_gib": 50.22
},
"serve": {
"device_peak_gib": 161.798,
"kv_bytes_per_token": 88591,
"kv_gib": 22.98,
"kv_usage": 0.216,
"other_gib": 4.555,
"peak_gib": 77.756,
"weights_gib": 50.22
}
},
"efficiency": {
"interactive": 126.70573058370624,
"long_context": 111.14029589437688,
"serve": 381.34158367457155
},
"gpu_hours": 2.344,
"kl": {
"kl": 0.011903762568120532,
"kl_median": 0.010981784176698761,
"kl_per_token": 0.01212828255402766,
"kl_se": 0.0007701735099757635,
"ppl": {
"hidden": {
"dnll": 0.017014044729516998,
"dnll_se": 0.0005806167970240156,
"ppl": 10.853214867974152,
"ppl_bf16": 10.670119795435681,
"tokens": 262016,
"windows": 128
},
"wikitext2": {
"dnll": 0.008080565660423577,
"dnll_se": 0.0016727030622871151,
"ppl": 6.955336706807456,
"ppl_bf16": 6.8993601176944415,
"tokens": 296815,
"windows": 145
}
},
"sequences": 256,
"tokens": 63668
},
"metric": "lowbit_score",
"peak_memory": {
"interactive": {
"device_peak_gib": 163.175,
"kv_bytes_per_token": 200835,
"kv_gib": 0.239,
"kv_usage": 0.0017,
"other_gib": 3.445,
"peak_gib": 21.625,
"weights_gib": 17.94
},
"long_context": {
"device_peak_gib": 163.177,
"kv_bytes_per_token": 75906,
"kv_gib": 73.548,
"kv_usage": 0.5262,
"other_gib": 3.445,
"peak_gib": 94.933,
"weights_gib": 17.94
},
"serve": {
"device_peak_gib": 163.175,
"kv_bytes_per_token": 88604,
"kv_gib": 22.984,
"kv_usage": 0.1644,
"other_gib": 3.445,
"peak_gib": 44.369,
"weights_gib": 17.94
}
},
"precision": {
"activations": 16,
"kv": 16,
"label": "W4A16KV16",
"weights": 4,
"weights_average": 4.75
},
"qualified": false,
"reason": "quality not measured",
"score": null,
"task_id": "qwen35-27b",
"timed_audit": null
},
"runner": "kubernetes",
"solution": {
"checkpoint": "cyankiwi/Qwen3.5-27B-AWQ-4bit",
"declared_precision": true,
"detected_precision": "W4A16KV16",
"method": "AWQ g32, cyankiwi",
"precision": "W4A16KV16"
},
"split": "test",
"stages": [
{
"name": "build",
"status": "done"
},
{
"name": "contract",
"status": "done"
},
{
"name": "quality",
"status": "failed"
},
{
"name": "efficiency",
"status": "done"
},
{
"name": "score",
"status": "done"
}
],
"status": "done",
"task_id": "qwen35-27b",
"track": "artifact",
"updated_at": "2026-10-09T02:46:18Z"
}