agent-harness / configs /reliability /E11_repeat_cells.json
cuber12's picture
Publish agent harness research code and paper artifacts
d61821a verified
Raw
History Blame Contribute Delete
955 Bytes
{
"cells": [
{"context_budget": 65536, "harness_id": "H000", "model_id": "M002", "task_id": "TASK_S4_R001_003"},
{"context_budget": 65536, "harness_id": "H007", "model_id": "M003", "task_id": "TASK_S4_R001_003"},
{"context_budget": 65536, "harness_id": "H000", "model_id": "M003", "task_id": "TASK_S4_R002_004"},
{"context_budget": 65536, "harness_id": "H007", "model_id": "M004", "task_id": "TASK_S4_R002_004"},
{"context_budget": 65536, "harness_id": "H000", "model_id": "M004", "task_id": "TASK_S4_R003_010"},
{"context_budget": 65536, "harness_id": "H007", "model_id": "M002", "task_id": "TASK_S4_R003_010"}
],
"description": "Six non-oracle groups balanced across three models, three repositories, and H000/H007; repeated at temperature 0.2 with seeds 0/1/2.",
"experiment_id": "E11",
"schema_version": 1,
"seeds": [0, 1, 2],
"selection_salt": "study4-ancillary-20260719",
"temperature": 0.2,
"top_p": 1.0
}