evaluation/tin-code-humaneval.json

{
 "version": "oeon.tin-code.v1",
 "benchmark": "HumanEval (openai/openai_humaneval), first problems in order",
 "problems": 40,
 "model": {
  "file": "Qwen3.5-9B-Q4_K_M.gguf",
  "sha256": "03b74727a860a56338e042c4420bb3f04b2fec5734175f4cb9fa853daf52b7e8",
  "base": "Qwen/Qwen3.5-9B"
 },
 "training": "none: the base is used as published",
 "repairs": 2,
 "firstDraft": {
  "passed": 37,
  "rate": 0.925,
  "wilson95": [
   0.8014,
   0.9742
  ]
 },
 "afterRepair": {
  "passed": 38,
  "rate": 0.95,
  "wilson95": [
   0.835,
   0.9862
  ]
 },
 "repairUsesOnly": "the docstring's own examples; the hidden tests are run once, for scoring",
 "secondsTotal": 10081.2,
 "wallSeconds": 10120.9
}