Log Anomaly Detection on BGL
97.67F1 ScoreNLLog
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| NLLogEvaluation Protocol=Reproduced under matched protocol2026.06 | 97.67 | 99.46 | 95.94 | — | |
| DeBERTa-v3Category=Fine-Tuned2026.04 | 97.4 | — | — | — | |
| RoBERTa-baseCategory=Fine-Tuned2026.04 | 96.8 | — | — | — | |
| CLADParser=false2026.04 | 96.45 | 97.68 | 95.3 | — | |
| BERT-baseCategory=Fine-Tuned2026.04 | 96.1 | — | — | — | |
| LogAnomalyEvaluation Protocol=Published results (different protocols)2026.06 | 96 | 97 | 94 | — | |
| LogLLaMAEvaluation Protocol=Published results (different protocols)2026.06 | 95.93 | 92.75 | 99.33 | — | |
| LogGPTEvaluation Protocol=Published results (different protocols)2026.06 | 95.8 | 94 | 97.7 | — | |
| LogGPT2026.06 | 95.8 | 94 | 97.7 | — | |
| CNNParser=true2026.04 | 95.47 | 95.3 | 95.64 | — | |
| LogRobustParser=true2026.04 | 93.73 | 92.61 | 94.89 | — | |
| LogMILP2026.05 | 93.42 | 92.64 | 94.21 | 94.64 | |
| LogLLMEvaluation Protocol=Published results (different protocols)2026.06 | 91.6 | 86.1 | 97.9 | — | |
| GPT-4 + SLCP (5-shot)Category=LLM (Prompt), Shot-count=5, Prompting strategy (SLCP)=true2026.04 | 91.5 | — | — | — | |
| LogNEO2026.06 | 91.3 | 91.9 | 90.4 | — | |
| Drain + RFCategory=Traditional2026.04 | 91.2 | — | — | — | |
| Drain + RF (best)Category=Traditional2026.04 | 91.2 | — | — | — | |
| LogDeep(FL)Learning Protocol=Federated Learning2026.04 | 91.18 | 96.88 | 86.11 | — | |
| LogBERTEvaluation Protocol=Reproduced under matched protocol2026.06 | 90.83 | 89.4 | 92.32 | — | |
| LogBERT2026.06 | 90.5 | 91.7 | 89.2 | — | |
| GPT-4 (5-shot)Category=LLM (Prompt), Shot-count=5, Prompting strategy (SLCP)=false2026.04 | 89.3 | — | — | — | |
| Drain + LRCategory=Traditional2026.04 | 88.7 | — | — | — | |
| GPT-4 + SLCP (zero)Category=LLM (Prompt), Shot-count=0, Prompting strategy (SLCP)=true2026.04 | 88.7 | — | — | — | |
| MIDLog2026.05 | 88.3 | 94.94 | 82.54 | 97.52 | |
| Spell + SVMCategory=Traditional2026.04 | 87.4 | — | — | — | |
| LogAnomaly2026.06 | 86.7 | 88.4 | 85 | — | |
| NeuralLogParser=false2026.04 | 86.19 | 90.22 | 82.5 | — | |
| DeepLogEvaluation Protocol=Reproduced under matched protocol2026.06 | 86.12 | 89.74 | 82.78 | — | |
| DeepLog2026.06 | 86.1 | 79.2 | 94.6 | — | |
| FlogTinyLLMModel Architecture=DeepSeek, Learning Protocol=Federated Learning2026.04 | 86.09 | 92.68 | 80.57 | — | |
| GPT-4 (zero-shot)Category=LLM (Prompt), Shot-count=0, Prompting strategy (SLCP)=false2026.04 | 85.6 | — | — | — | |
| Federated LogBERTLearning Protocol=Federated Learning2026.04 | 84.77 | 75.74 | 96.25 | — | |
| FlogTinyLLMModel Architecture=Phi 1.5, Learning Protocol=Federated Learning2026.04 | 82.6 | 96.66 | 75.69 | — | |
| LogBERT2026.05 | 81.93 | 88.94 | 75.94 | 94.08 | |
| LLaMA-3 (zero-shot)Category=LLM (Prompt), Shot-count=0, Prompting strategy (SLCP)=false2026.04 | 81.3 | — | — | — | |
| FlogTinyLLMModel Architecture=OPT 1.3B, Learning Protocol=Federated Learning2026.04 | 80.51 | 87.37 | 77.3 | — | |
| GPT-3.5 (zero-shot)Category=LLM (Prompt), Shot-count=0, Prompting strategy (SLCP)=false2026.04 | 79.1 | — | — | — | |
| FlogTinyLLMModel Architecture=TinyLLama, Learning Protocol=Federated Learning2026.04 | 78.22 | 90.41 | 68.99 | — | |
| DeepLog2026.05 | 76.43 | 70.04 | 84.09 | 92.33 | |
| LogCluster2026.06 | 76.2 | 94.1 | 64.1 | — | |
| Federated LogGPTLearning Protocol=Federated Learning2026.04 | 75.62 | 92.11 | 74.82 | — | |
| LogAnomaly2026.05 | 73.05 | 75.12 | 71.09 | 94.38 | |
| LogFormer2026.05 | 71.19 | 70.85 | 71.54 | 92.16 | |
| OC4Seq2026.06 | 39.1 | 44.1 | 35.2 | — | |
| CAT2026.06 | 19 | 17.7 | 21 | — | |
| OCSVM2026.06 | 12.1 | 7.3 | 34.5 | — | |
| iForest2026.06 | 6.3 | 49.1 | 3.7 | — | |
| PCA2026.06 | 5.4 | 11.7 | 3.5 | — |