{
 "schema": "local-ai-bench/result/v1",
 "run_id": "20261003T221725Z-llama.cpp-qwen3.8-flash-next-iq4xs-128k-3x-fit-gpu012-stability",
 "ticket": "T11",
 "kind": "stability",
 "model": "qwen3.8-flash-next",
 "config_id": "llama.cpp-qwen3.8-flash-next-iq4xs-128k-3x-fit-gpu012",
 "outcome": "fail",
 "failed_stage": null,
 "exit_status": 0,
 "started_at": "2026-10-03T22:17:25.793653+00:00",
 "finished_at": "2026-10-03T22:19:19.727398+00:00",
 "launch": {
  "file": "launches/llama.cpp-qwen3.8-flash-next-iq4xs-128k-3x-fit.json",
  "sha256": "f7129eb274fa7a838d7a8928e9e72cec61e157259a237393b1f99b1945e722ab",
  "image": "ghcr.io/ggml-org/llama.cpp:server-cuda12-b11312@sha256:6cdf9529493b9581c421fec5628dcd4b8dd7c09507fc7d27ad739cc9205d66f5",
  "argv": [
   "/app/llama-server",
   "--model",
   "/models/Qwen3.8-Flash-Next-IQ4_XS/Qwen3.8-Flash-Next-IQ4_XS-00001-of-00003.gguf",
   "--ctx-size",
   "131072",
   "--flash-attn",
   "on",
   "--cache-type-k",
   "q8_0",
   "--cache-type-v",
   "q8_0",
   "--fit",
   "on",
   "--parallel",
   "1",
   "--threads",
   "15",
   "--jinja",
   "--no-mmproj",
   "--metrics",
   "--host",
   "0.0.0.0",
   "--port",
   "8080",
   "--split-mode",
   "layer"
  ],
  "env": {
   "CUDA_DEVICE_ORDER": "PCI_BUS_ID"
  },
  "engine": "llama.cpp",
  "engine_source_commit": null,
  "mtp": null,
  "entrypoint": "/app/llama-server",
  "port": 8080,
  "shm": "16g",
  "flags": [],
  "weights_at": [
   "/models"
  ]
 },
 "weights": [
  {
   "repo": "bartowski/Qwen3.8-Flash-Next-GGUF",
   "revision": "928589fdb66c6ff07f22ac561e3fbce76553548f",
   "manifest": "models/bartowski-Qwen3.8-Flash-Next-IQ4_XS.json"
  }
 ],
 "host": {
  "hostname": "omarchy-gpu",
  "cpu": "AMD Ryzen 9 5950X 16-Core Processor",
  "ram_total_gib": 125.7,
  "ram_layout": "DIMM_A1 32 GiB DDR4 3200 MT/s, DIMM_A2 32 GiB DDR4 3200 MT/s, DIMM_B1 32 GiB DDR4 3200 MT/s, DIMM_B2 32 GiB DDR4 3200 MT/s (AM4: 2 channels)",
  "kernel": "7.2.5-3-omarchy",
  "driver": "610.57.04",
  "cuda": "13.3",
  "cpus_online": "0-1,3-17,19-31",
  "cpus_offline": "2,18",
  "machine_checks_this_boot": 6
 },
 "gpus": [
  {
   "index": 0,
   "uuid": "GPU-816b43e4-8d65-dfd7-2234-8517b3dfbf2d",
   "name": "NVIDIA GeForce RTX 3090",
   "bus_id": "00000000:04:00.0",
   "pcie_gen_max": 4,
   "pcie_width": 8,
   "display_active": false,
   "used_by_run": true
  },
  {
   "index": 1,
   "uuid": "GPU-c67ac872-3371-a88c-aa92-d969daf6405e",
   "name": "NVIDIA GeForce RTX 3090",
   "bus_id": "00000000:0B:00.0",
   "pcie_gen_max": 4,
   "pcie_width": 8,
   "display_active": false,
   "used_by_run": true
  },
  {
   "index": 2,
   "uuid": "GPU-76d3c6af-7f18-69d2-3145-899103de1722",
   "name": "NVIDIA GeForce RTX 3090",
   "bus_id": "00000000:0C:00.0",
   "pcie_gen_max": 4,
   "pcie_width": 8,
   "display_active": false,
   "used_by_run": true
  }
 ],
 "layout": {
  "name": "current",
  "cards": 3,
  "display": "off"
 },
 "context": {
  "configured": 131072,
  "occupied_max": {
   "state": "unavailable",
   "reason": "no successful request"
  },
  "headroom_min": {
   "state": "unavailable",
   "reason": "configured context unknown (set CONFIGURED_CTX)"
  }
 },
 "cache_state": {
  "cold": false,
  "reset": []
 },
 "provenance": {
  "bench_commit": "d9944c1c5f811326cc98ee2701073db1981ab9b7",
  "bench_dirty": false,
  "pins_sha256": "858564b6e07125ac48c2ef6c5178d0a53596f21eb5f0ebda43906472b01b4d78",
  "registry_commit": "d21258dd744e7c78be28177c90060c6af8e10b7e",
  "kit_commit": "ef883d269e50ecbea290f919f095e2f3ca633b42",
  "harness_commit": "04d809ceab9df28f9adaed044884180159172930"
 },
 "metrics": {
  "duration_s": {
   "state": "unavailable",
   "reason": "driver produced no timing"
  },
  "tool_calls": {
   "value": 0,
   "unit": "calls",
   "detail": {
    "per_min": null,
    "requests": 0
   }
  },
  "parse_failures": {
   "value": 0,
   "unit": "responses",
   "detail": {
    "wire": {},
    "harness_format_errors": {},
    "union_rule": "a response counts once even if both the wire check and the harness flag it",
    "counted": "malformed tool calls only (decision 0003); truncations and other format errors are separate metrics"
   }
  },
  "output_truncations": {
   "value": 0,
   "unit": "responses",
   "detail": {
    "rule": "cut off at max_tokens with no tool call"
   }
  },
  "agent_format_errors": {
   "value": 0,
   "unit": "responses",
   "detail": {
    "rule": "agent protocol errors other than malformed tool calls"
   }
  },
  "repetition_hits": {
   "value": 0,
   "unit": "hits",
   "detail": {
    "ngram64x3": 0,
    "same_call_x4": 0,
    "ngram64x3_cross_field_not_counted": 0,
    "tokenizer": null,
    "saved": "artifacts: proxy/repetition/"
   }
  },
  "decode_window_min_tps": {
   "state": "unavailable",
   "reason": "only 0.0 s of generation; need > 120 s for one window"
  },
  "decode_whole_run_tps": {
   "state": "unavailable",
   "reason": "no generation recorded"
  },
  "tasks_attempted": {
   "value": 0,
   "unit": "attempts",
   "detail": {
    "unfinished_at_cap": 0,
    "task_ids": [
     "django__django-11099",
     "sympy__sympy-21612",
     "psf__requests-2317",
     "matplotlib__matplotlib-23299",
     "scikit-learn__scikit-learn-13439",
     "astropy__astropy-12907",
     "pytest-dev__pytest-7373",
     "sympy__sympy-20590"
    ],
    "task_set_sha256": "8e7487ca7d059df6359fc7d9b86c7021256a8a81b8af049616750f8f83400324"
   }
  },
  "tasks_solved": {
   "state": "skipped",
   "reason": "no attempts"
  },
  "crash_or_oom": {
   "value": false,
   "unit": "bool",
   "detail": {
    "upstream_connection_failures": 0,
    "upstream_http_errors": 0,
    "source": "proxy-observed; the run wrapper adds container death/OOM"
   }
  },
  "gates_after": {
   "state": "skipped",
   "reason": "filled by qualify step: the lab run that follows this run (tools/qualify.py)"
  },
  "agent_attempts": {
   "value": [],
   "unit": "list"
  },
  "tasks_over_64k": {
   "value": [],
   "unit": "instance ids"
  },
  "completion_tokens": {
   "value": 0,
   "unit": "tokens",
   "detail": {
    "finish_reasons": {}
   }
  },
  "mtp_accepted_tps": {
   "state": "skipped",
   "reason": "speculative decoding off in this launch"
  },
  "mtp_acceptance_rate": {
   "state": "skipped",
   "reason": "speculative decoding off in this launch"
  }
 },
 "eligibility": "none",
 "artifacts": [
  "artifacts/runs/20261003T221725Z-lab-llama.cpp-qwen3.8-flash-next-iq4xs-128k-3x-fit-gpu012-stability"
 ],
 "notes": "stability run, 10 min cap, floor 50.0 tok/s; mini-swe-agent 04d809ceab9d (benchmarks/swebench.yaml, native tool calls); sampling {\"temperature\": 0.7, \"top_p\": 0.95, \"seed\": 1234, \"max_tokens\": 8192} | repetition tokenizer: server /tokenize | 0 tool calls in this run count toward the config's 100-call minimum (\u00a75) | FAIL: decode floor not verifiable (too little generation); agent driver exited 1"
}