Submission · qwen3-8b

d55250a215c34184

by trial2-8b (participant) · dev split · runner kubernetes

FP8-W8A8+FP8head+INT4emb+KV8+seqs12+bt4096-trial2-8b lowbarena-trial-8b2-fp8-fp8head-int4emb W8A8KV8

done

Pipeline

  1. build done
  2. contract done
  3. quality done
  4. efficiency done
  5. score skipped

Result

1.792LowBScore on the development split (never ranked)
W·A·KV W8A8KV8 declared · the service reads W8A8KV8
Accuracytier B · mean drop 0.96 points · worst domain commonsense (upper bound 3.8)
Speed gain1.02× BF16 in the same pod (serve 0.76×, interactive 1.27×, long context 1.11×)
Memory gain3.71× BF16's peak (serve 3.51×, interactive 2.01×, long context 7.24×) · 15.8 GiB against 55.6 GiB on the first workload
E · Defficiency gain 1.95 · drift 0.84 points
Contractpassed, 8.00 bits per weight stored
GPU-hours0.70
Raw record
{
  "agent": "trial2-8b",
  "created_at": "2026-10-13T07:13:14Z",
  "error": null,
  "id": "d55250a215c34184",
  "participant": "trial2-8b",
  "partition": "participant",
  "result": {
    "contract": {
      "activation_bits": 8,
      "bits_per_weight": 8.003226902173912,
      "foreign_tensors": {
        "bytes": 0,
        "count": 0,
        "largest": [],
        "share": 0.0
      },
      "format": "fp8-w8a8",
      "kv_bits": null,
      "model_bits_per_weight": 8.003282765413925,
      "ok": true,
      "quantization": {
        "activations": {
          "actorder": null,
          "block_structure": null,
          "dynamic": true,
          "group_size": null,
          "num_bits": 8,
          "observer": null,
          "observer_kwargs": {},
          "strategy": "token",
          "symmetric": true,
          "type": "float"
        },
        "format": "fp8-w8a8",
        "ignore": [],
        "kv_quantized": false,
        "sparse": false,
        "weights": {
          "actorder": null,
          "block_structure": null,
          "dynamic": false,
          "group_size": null,
          "num_bits": 8,
          "observer": "mse",
          "observer_kwargs": {},
          "strategy": "channel",
          "symmetric": true,
          "type": "float"
        }
      },
      "rope_artifact": {
        "rope_theta": 1000000,
        "rope_type": "default"
      },
      "rope_source": {
        "rope_theta": 1000000,
        "rope_type": "default"
      },
      "safetensor_bytes": 7925060048,
      "sparse": false,
      "violations": [],
      "weight_bits": 8
    },
    "dev_quality": {
      "domains": {
        "code": {
          "bf16": 0.3380681818181818,
          "bf16_se": 0.047431027842082016,
          "candidate_lost": 0.0,
          "chance": 0.0,
          "drop": 0.004261363636363636,
          "headroom": 0.3380681818181818,
          "items": 88,
          "quantized": 0.3338068181818182,
          "recovery": 0.9873949579831932,
          "recovery_se": 0.022716247757476812,
          "se": 0.007679640577101536,
          "status": "valid",
          "upper": 0.016894372385695664
        },
        "commonsense": {
          "bf16": 0.7466666666666667,
          "bf16_se": 0.024134209165109787,
          "candidate_lost": 0.0,
          "chance": 0.24983333333333332,
          "drop": 0.015833333333333335,
          "format_contract": {
            "bf16": 1.0,
            "quantized": 1.0
          },
          "headroom": 0.49683333333333335,
          "items": 300,
          "quantized": 0.7308333333333333,
          "recovery": 0.9681314994968131,
          "recovery_se": 0.02655378171889604,
          "se": 0.01319280388400485,
          "status": "valid",
          "upper": 0.03753549572252131
        },
        "expert": {
          "bf16": 0.44166666666666665,
          "bf16_se": 0.08491318532563476,
          "candidate_lost": 0.0,
          "chance": 0.25,
          "drop": -0.1125,
          "format_contract": {
            "bf16": 0.9916666666666667,
            "quantized": 0.9916666666666667
          },
          "headroom": 0.19166666666666665,
          "items": 30,
          "quantized": 0.5541666666666667,
          "recovery": 1.5869565217391308,
          "recovery_se": 0.23052528561356556,
          "se": 0.04418401307593339,
          "status": "R5_diagnostic",
          "upper": -0.03981729849008957
        },
        "knowledge": {
          "bf16": 0.44666666666666666,
          "bf16_se": 0.027939588170592746,
          "candidate_lost": 0.0,
          "chance": 0.10940343915343916,
          "drop": 0.015833333333333335,
          "format_contract": {
            "bf16": 0.92,
            "quantized": 0.9283333333333333
          },
          "headroom": 0.3372632275132275,
          "items": 300,
          "quantized": 0.43083333333333335,
          "recovery": 0.9530534845139251,
          "recovery_se": 0.03225071708506634,
          "se": 0.010876980933725463,
          "status": "valid",
          "upper": 0.033725966969311724
        },
        "long_context": {
          "bf16": 0.8561666666666666,
          "bf16_se": 0.02867970594529983,
          "candidate_lost": 0.0,
          "chance": 0.0,
          "drop": 0.0020833333333333337,
          "headroom": 0.8561666666666666,
          "items": 100,
          "quantized": 0.8540833333333333,
          "recovery": 0.9975666731555383,
          "recovery_se": 0.012682665850035529,
          "se": 0.010858475745272085,
          "status": "valid",
          "upper": 0.019945525934305913
        },
        "math": {
          "bf16": 0.89,
          "bf16_se": 0.019057569487578137,
          "candidate_lost": 0.0,
          "chance": 0.0,
          "drop": 0.01,
          "headroom": 0.89,
          "items": 200,
          "quantized": 0.88,
          "recovery": 0.9887640449438202,
          "recovery_se": 0.009924285730611847,
          "se": 0.008832614300244544,
          "status": "valid",
          "upper": 0.024529650523902276
        }
      },
      "evaluations": 1,
      "first_quality": null,
      "length": {
        "length_generations": 4544,
        "length_shift": -0.012300378816014312,
        "length_shift_se": 0.005468344699969847,
        "length_tokens": 146.31822183098592,
        "length_tokens_bf16": 148.14040492957747
      },
      "recovery": 0.978982132018658,
      "recovery_se": 0.010041078524902235,
      "reference_rows": "/work/results/qwen-qwen3-8b/bf16-lb",
      "tier": "B",
      "valid_domains": 5,
      "worst_domain": "commonsense",
      "worst_drop": 0.015833333333333335,
      "worst_upper": 0.03753549572252131
    },
    "dev_score": {
      "bf16_peak_gib": 55.615,
      "drift": 0.8403,
      "efficiency_gain": 1.9489,
      "mean_drop": 0.0096,
      "memories": {
        "interactive": 2.0074,
        "long_context": 7.2432,
        "serve": 3.5108
      },
      "memory": 3.7096,
      "memory_saved": 0.7152,
      "peak_gib": 15.841,
      "precision": {
        "activations": 8,
        "kv": 8,
        "label": "W8A8KV8",
        "weights": 8,
        "weights_average": 8.0
      },
      "qualified": true,
      "reason": null,
      "score": 1.7918,
      "speed": 1.0239,
      "speeds": {
        "interactive": 1.2744,
        "long_context": 1.1085,
        "serve": 0.7598
      }
    },
    "efficiency": {
      "interactive": 331.5895082163212,
      "long_context": 815.8783272087899,
      "serve": 1464.539631706047
    },
    "engine": {
      "audit": null,
      "budget": {
        "interactive@1.02": {
          "budget_gib": 8.938,
          "cap_gib": 12.938,
          "factor": 1.02,
          "full_running": 1.0,
          "healthy": true,
          "running": 1.0,
          "value": 337.8146476362128
        },
        "long_context@1.02": {
          "budget_gib": 22.498,
          "cap_gib": 26.498,
          "factor": 1.02,
          "full_running": 12.0,
          "healthy": true,
          "running": 12.0,
          "value": 808.8421235454565
        },
        "serve@1.02": {
          "budget_gib": 16.158,
          "cap_gib": 20.158,
          "factor": 1.02,
          "full_running": 12.0,
          "healthy": true,
          "running": 12.0,
          "value": 1479.2302514680964
        }
      },
      "hard_cap": true,
      "startup": {
        "cuda_graph_gib": 0.06,
        "engine_init_seconds": 64.13,
        "kv_bytes_per_token": 73729.4260506672,
        "kv_cache_dtype": "fp8",
        "kv_cache_gib": 152.48,
        "kv_cache_tokens": 2220608.0,
        "load_seconds": 5.160326,
        "max_concurrency": {
          "sequences": 67.77,
          "tokens_per_request": 32768
        },
        "speculative_config": "None",
        "vllm_version": "0.1.dev18498+g7a33ba95f",
        "weights_gib": 7.42
      },
      "stored_gib": 7.381
    },
    "gpu_hours": 0.698,
    "kl": {
      "kl": 0.009092044286058707,
      "kl_median": 0.009791557695287487,
      "kl_per_token": 0.008995099241270968,
      "kl_se": 0.001290113780927231,
      "sequences": 64,
      "tokens": 15418
    },
    "split": "dev",
    "workload": {
      "bf16": {
        "peak_gib": 55.615,
        "rate": 1927.5479023209139
      },
      "submission": {
        "peak_gib": 15.841,
        "rate": 1464.539631706047
      }
    }
  },
  "runner": "kubernetes",
  "solution": {
    "checkpoint": "lowbarena-trial-8b2-fp8-fp8head-int4emb",
    "declared_precision": true,
    "detected_precision": "W8A8KV8",
    "method": "FP8-W8A8+FP8head+INT4emb+KV8+seqs12+bt4096-trial2-8b",
    "precision": "W8A8KV8"
  },
  "split": "dev",
  "stages": [
    {
      "name": "build",
      "status": "done"
    },
    {
      "name": "contract",
      "status": "done"
    },
    {
      "name": "quality",
      "status": "done"
    },
    {
      "name": "efficiency",
      "status": "done"
    },
    {
      "name": "score",
      "status": "skipped"
    }
  ],
  "status": "done",
  "task_id": "qwen3-8b",
  "track": "engine",
  "updated_at": "2026-10-13T07:56:49Z"
}