Spaces:
Running
Running
File size: 6,472 Bytes
e0414df 0c38e08 2ecbe1e 9333eeb 2ecbe1e 9333eeb 2ecbe1e 9333eeb 2ecbe1e e0414df 30b8389 b3186de 30b8389 e0414df 30b8389 31e04c0 81834db ab46f92 c8a1b71 ab46f92 c8a1b71 ab46f92 81834db c465e3f 9018414 a9df0e9 9018414 c8a1b71 9018414 24b126e c465e3f 81834db ab46f92 81834db ab46f92 24b126e e0414df 0c38e08 2ecbe1e 9333eeb 81834db c465e3f 9018414 a9df0e9 9018414 c8a1b71 9018414 24b126e c465e3f 81834db c465e3f 81834db 0c38e08 2ecbe1e 9333eeb 81834db 9018414 e0414df | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 | {
"harness": "stock Codex CLI 0.160.0",
"model": "gpt-6-astra",
"effort": "high",
"seed": 0,
"episodes_per_task": 1,
"mode": "stepped",
"safety_timeout_sec": 28800,
"fresh_workspace": true,
"resume_trajectory": false,
"source_jobs": [],
"families": [
{
"id": "task01",
"name": "RoboCasa",
"total": 10,
"completed": 10,
"pending": 0,
"successes": 2,
"valid_results": 10,
"success_rate": 0.2,
"input_tokens": 63285354,
"cached_input_tokens": 62436736,
"output_tokens": 204733,
"usage_complete": 10,
"control_frequency_hz": 20,
"max_control_steps": 6000,
"preflight_results": 0,
"formal_results": 10,
"modified_results": 10,
"original_results": 0
},
{
"id": "task02",
"name": "LIBERO Long",
"total": 10,
"completed": 10,
"pending": 0,
"successes": 9,
"valid_results": 10,
"success_rate": 0.9,
"input_tokens": 12586499,
"cached_input_tokens": 12106752,
"output_tokens": 93969,
"usage_complete": 10,
"control_frequency_hz": 20,
"max_control_steps": 6000,
"preflight_results": 0,
"formal_results": 10,
"modified_results": 5,
"original_results": 5
},
{
"id": "task03",
"name": "RoboTwin",
"total": 10,
"completed": 10,
"pending": 0,
"successes": 7,
"valid_results": 10,
"success_rate": 0.7,
"input_tokens": 15621828,
"cached_input_tokens": 15138688,
"output_tokens": 105566,
"usage_complete": 10,
"control_frequency_hz": 25,
"max_control_steps": 7500,
"preflight_results": 0,
"formal_results": 10,
"modified_results": 7,
"original_results": 3,
"deferred_tasks": 0
},
{
"id": "task04",
"name": "RoboDojo",
"total": 42,
"completed": 42,
"pending": 0,
"successes": 31,
"valid_results": 42,
"success_rate": 0.7380952380952381,
"input_tokens": 259078099,
"cached_input_tokens": 254678912,
"output_tokens": 971405,
"usage_complete": 42,
"control_frequency_hz": 25,
"max_control_steps": 7500,
"preflight_results": 0,
"formal_results": 42,
"modified_results": 34,
"original_results": 8
},
{
"id": "task06",
"name": "SIMPLE G1",
"total": 12,
"completed": 12,
"pending": 0,
"successes": 7,
"valid_results": 12,
"success_rate": 0.5833333333333334,
"usage_complete": 12,
"control_frequency_hz": 50,
"max_control_steps": null,
"max_control_steps_by_task": {
"task06/01": 10000,
"task06/02": 15000,
"task06/03": 15000,
"task06/04": 15000,
"task06/05": 15000,
"task06/06": 15000,
"task06/07": 15000,
"task06/08": 10000,
"task06/09": 15000,
"task06/10": 15000,
"task06/11": 15000,
"task06/12": 15000
},
"preflight_results": 0,
"formal_results": 12,
"modified_results": 0,
"original_results": 12,
"input_tokens": 106763761,
"cached_input_tokens": 105331968,
"output_tokens": 303566
}
],
"usage_accounting": "reported-responses",
"retry_policy_note": "The first six completed results use original Codex request defaults. Remaining tasks and two authorized fresh replacements use a 50-retry request policy. All new RoboCasa and LIBERO runs use explicit 50-retry SSE request policy; no automatic episode retry.",
"family_protocols": {
"task01": {
"episodes": 1,
"seed": 0,
"max_steps": 6000,
"control_frequency_hz": 20
},
"task02": {
"episodes": 1,
"seed": 0,
"init_state_index": 0,
"max_steps": 6000,
"control_frequency_hz": 20
},
"task04": {
"episodes": 1,
"seed": 0,
"max_steps": 7500,
"control_frequency_hz": 25
},
"task03": {
"episodes": 1,
"seed": 0,
"max_steps": 7500,
"control_frequency_hz": 25,
"safety_timeout_sec": 28800,
"execution_mode": "stepped",
"stop_on_success": true,
"deferred_tasks": [],
"native_tcp_tasks": [
"robotwin/lift-pot-aloha-agilex"
],
"runtime_provenance": "See each episode: the other nine tasks retain their previously measured runtime."
},
"task06": {
"episodes": 1,
"seed": 0,
"control_frequency_hz": 50,
"max_control_steps": null,
"max_control_steps_by_task": {
"task06/01": 10000,
"task06/02": 15000,
"task06/03": 15000,
"task06/04": 15000,
"task06/05": 15000,
"task06/06": 15000,
"task06/07": 15000,
"task06/08": 10000,
"task06/09": 15000,
"task06/10": 15000,
"task06/11": 15000,
"task06/12": 15000
},
"timeout_s": 28800,
"mode": "stepped",
"instruction_policy": "original_native",
"max_steps": null,
"execution_mode": "stepped",
"stop_on_success": true,
"sim_mode": "mujoco_isaac",
"dr_level": 0,
"arm_gravity_compensation": true,
"recording": true,
"model": "gpt-6-astra",
"reasoning_effort": "high",
"service_tier": "default",
"max_request_retries": 50,
"automatic_episode_retries": 0,
"fresh_session": true
}
},
"robotwin_harness": "Stock Codex CLI 0.160.0, gpt-6-astra, high, default service tier; current RoboEnv SDK/CLI; no Kinex agent runtime.",
"native_tcp_repair_note": "Lift pot uses native RoboTwin TCP frames after the RoboEnv correction, with the existing reviewed instruction, unchanged native verifier, seed 0, 7500 action frames at 25 Hz, and one fresh stock Codex 0.160.0 / GPT-6 Astra high episode. Other results retain their historical runtime and protocol.",
"simple_comparability": "Kinex 0.10.3 and ordinary Codex CLI 0.160.0 share the frozen RoboEnv source, native task instructions, seed and protocol. Each task records its measured budget and runtime. Only earlier native failures were rerun at 15000 steps; retained successes use 10000 steps. The current selection mixes budgets and is not a uniform-budget agent comparison. Prior interrupted attempts are retained separately; fresh sessions inherit no trajectories or resources. Wall time is not a controlled speed comparison."
}
|