asketeddy's picture
Publish bounded Gooo feedback judgment models, synthetic data and actual native dogfood
4a52615 verified
Raw History Blame Contribute Delete
5.22 kB
{
"actual_generated_function_evaluations": 16,
"actual_generated_go_processes": 1,
"actual_local_predictions": 108,
"actual_separate_input_evaluations": 9,
"candidate_attempts": 384,
"decision": "PASS",
"distinct_separate_inputs": 9,
"execution_expected": [
-4,
8,
10,
-18,
-10,
22,
24,
-190,
6,
-16,
-14,
-12,
26,
210,
10,
8
],
"execution_inputs": [
-7,
-1,
0,
1,
5,
6,
7,
-100,
-2,
2,
3,
4,
8,
100,
-9223372036854775808,
9223372036854775807
],
"feedback_predictions": 72,
"native_calls": 6,
"native_revision": "6a3e45d7f185c74f5136ab81b560a14a1366f367",
"observations": [
{
"actual_local_predictions": 18,
"candidate_attempts": 64,
"capture_sha256": "0ac596b9c397e3e03bc353566ad8b4259780a7439e3832ee46bedf3840e97955",
"feedback_predictions": 12,
"finite_cases": 7,
"finite_passed": 6,
"generated_go_sha256": "0db2c6f0da2b2fbed66e56a7f5d9ae20f628167df67500b54ad73072d3cd1c72",
"language": "en",
"process_metrics": {
"wall_ns": 54774625,
"user_cpu_ns": 9232000,
"system_cpu_ns": 9472000,
"child_max_rss_bytes": 21626880,
"process_cpu_percent_of_one_core_over_wall": 34.14719863440416
},
"variant": "fp32"
},
{
"actual_local_predictions": 18,
"candidate_attempts": 64,
"capture_sha256": "b14089f53ee6a2fa6645db1086037cbbbef168c48c9608db7281db84e2721f8a",
"feedback_predictions": 12,
"finite_cases": 7,
"finite_passed": 6,
"generated_go_sha256": "0db2c6f0da2b2fbed66e56a7f5d9ae20f628167df67500b54ad73072d3cd1c72",
"language": "en",
"process_metrics": {
"wall_ns": 11386459,
"user_cpu_ns": 8605000,
"system_cpu_ns": 2988000,
"child_max_rss_bytes": 22003712,
"process_cpu_percent_of_one_core_over_wall": 101.81391774211808
},
"variant": "ptq_ternary"
},
{
"actual_local_predictions": 18,
"candidate_attempts": 64,
"capture_sha256": "f7ea12946330e6bd018447b68b7fbbd8e2eec417dd4ffdec61362ae9ef33c926",
"feedback_predictions": 12,
"finite_cases": 7,
"finite_passed": 6,
"generated_go_sha256": "0db2c6f0da2b2fbed66e56a7f5d9ae20f628167df67500b54ad73072d3cd1c72",
"language": "en",
"process_metrics": {
"wall_ns": 10784959,
"user_cpu_ns": 8310000,
"system_cpu_ns": 2622000,
"child_max_rss_bytes": 21807104,
"process_cpu_percent_of_one_core_over_wall": 101.3633895131173
},
"variant": "qat_ternary"
},
{
"actual_local_predictions": 18,
"candidate_attempts": 64,
"capture_sha256": "56a41fb97937ffdf85be0337f78a4e5d4bb448fc66bcc19c66eb173c83d0418b",
"feedback_predictions": 12,
"finite_cases": 7,
"finite_passed": 6,
"generated_go_sha256": "0db2c6f0da2b2fbed66e56a7f5d9ae20f628167df67500b54ad73072d3cd1c72",
"language": "ko",
"process_metrics": {
"wall_ns": 10051250,
"user_cpu_ns": 7960000,
"system_cpu_ns": 2240000,
"child_max_rss_bytes": 21577728,
"process_cpu_percent_of_one_core_over_wall": 101.4799154334038
},
"variant": "fp32"
},
{
"actual_local_predictions": 18,
"candidate_attempts": 64,
"capture_sha256": "664d0ac2b867c471e90ba4fa4f06c1152dd8aab37756553dd6a021cf6fdeb072",
"feedback_predictions": 12,
"finite_cases": 7,
"finite_passed": 6,
"generated_go_sha256": "0db2c6f0da2b2fbed66e56a7f5d9ae20f628167df67500b54ad73072d3cd1c72",
"language": "ko",
"process_metrics": {
"wall_ns": 10358833,
"user_cpu_ns": 8197000,
"system_cpu_ns": 2408000,
"child_max_rss_bytes": 21905408,
"process_cpu_percent_of_one_core_over_wall": 102.37639703236842
},
"variant": "ptq_ternary"
},
{
"actual_local_predictions": 18,
"candidate_attempts": 64,
"capture_sha256": "4367c9fc1dc8f2f25dd13cd93aec8b2a94d344e885c7262e850958493bf062ba",
"feedback_predictions": 12,
"finite_cases": 7,
"finite_passed": 6,
"generated_go_sha256": "0db2c6f0da2b2fbed66e56a7f5d9ae20f628167df67500b54ad73072d3cd1c72",
"language": "ko",
"process_metrics": {
"wall_ns": 10063250,
"user_cpu_ns": 7994000,
"system_cpu_ns": 2227000,
"child_max_rss_bytes": 22216704,
"process_cpu_percent_of_one_core_over_wall": 101.56758502471865
},
"variant": "qat_ternary"
}
],
"repeated_finite_cases": 42,
"repeated_finite_passes": 36,
"repeated_separate_input_policy_observations": 54,
"runner_revision": "1cfeb4753fdcb62aa2b9cfad3002c1c0172b5328",
"schema": "gooo/trained-feedback-dogfood/v1",
"scope": "Actual new-model native dogfood on one existing compound source, with six policy/language views and preserved 6/7 partial contracts. Exact emitted Go is deduplicated and executed on seven selection inputs plus nine old disjoint inputs; reused policy observations are not independent tasks. Host utilization is not sampled."
}