Werea-NanoSOC-8B / evaluation.json
GoktugD's picture
Mirror GoktugD/Llama-NanoSOC1-8B@main as Werea company flagship
ddaf379 verified
Raw
History Blame Contribute Delete
1.51 kB
{
"release": "v2",
"training_stage": "Kademe 2 / F3 61-90",
"new_holdout": {
"count": 49,
"v1_baseline": {
"json_valid_rate": 1.0,
"decision_accuracy": 0.673469387755102,
"mitre_attack_exact_accuracy": 0.0
},
"v2": {
"json_valid_rate": 1.0,
"decision_accuracy": 0.8775510204081632,
"mitre_attack_exact_accuracy": 0.22448979591836735,
"peak_vram_mib": 5705.67,
"confusion": {
"true_suspicious": 32,
"true_insufficient_evidence": 11,
"false_positive": 5,
"false_negative": 1
},
"scenario_breakdown": {
"cobalt_strike_dns_beacon": {
"count": 48,
"decision_correct": 43,
"mitre_exact": 11
},
"powershell_dns_txt": {
"count": 1,
"decision_correct": 0,
"mitre_exact": 0
}
}
}
},
"v1_regression_holdout": {
"count": 40,
"v1": {
"json_valid_rate": 1.0,
"decision_accuracy": 0.925,
"mitre_attack_exact_accuracy": 0.4
},
"v2": {
"json_valid_rate": 1.0,
"decision_accuracy": 0.925,
"mitre_attack_exact_accuracy": 0.8,
"peak_vram_mib": 5719.84,
"scenario_breakdown": {
"dll_sideload": {
"count": 35,
"decision_correct": 32,
"mitre_exact": 32
},
"base64_powershell": {
"count": 5,
"decision_correct": 5,
"mitre_exact": 0
}
}
}
}
}