dfreq-tug / evaluation.json
djduecaster's picture
Publish reviewed source-authored Tug baseline
c234222 verified
Raw History Blame Contribute Delete
1.09 kB
{
"official_format_local": {
"candidate": "brace_lean002",
"score_cm": -12.041612342000008,
"runner_version": 5,
"seeds_finished": 5,
"per_seed": [
{
"seed": 1,
"status": "lost",
"time_s": 45.0,
"score_cm": -18.656958639621735
},
{
"seed": 2,
"status": "lost",
"time_s": 45.0,
"score_cm": -5.963634327054024
},
{
"seed": 3,
"status": "lost",
"time_s": 45.0,
"score_cm": -8.068934828042984
},
{
"seed": 4,
"status": "lost",
"time_s": 45.0,
"score_cm": -7.6185621321201324
},
{
"seed": 5,
"status": "lost",
"time_s": 45.0,
"score_cm": -19.899971783161163
}
],
"trace_seed": 3,
"model_sha256": "4d913bef9eb5b9db45bb9741836ab084e37024a8ba64315eaa5b51bf55bd963e"
},
"held_out_mean_ground_cm": -11.86637431383133,
"source_generation": "Original private experiment evidence retained by author; per-seed rating results included here."
}