Annotating Log Data For Machine Learning
summary
A reproducible pipeline extracts failure logs, selects top error lines with TF‑IDF, pre‑processes noise, and runs a classifier (Transformer via Hugging Face) or an LLM; a custom pre‑trained model called Vbird and a fine‑tuned classifier reach F1 score .95 across five imbalanced classes, while Llama 2 works for summarization but sometimes misclassifies without fine‑tuning. The team ships an open-source log annotation project that labels, aggregates, summarizes, routes, and can open bugs.
AI-written summary · about
chapters (6)
- 0:00Log annotation intro
- 1:53Manual log analysis challenges
- 4:00ML log analysis workflow
- 7:15Supervised vs LLM classification
- 11:41Open-source log annotation demo
- 15:40Conclusion
related talks
10mML-based Log Analysis for Faster Debugging
46mAI-Driven Techniques for Noise Filtration in Software Validation Logs
30mDebugging at Scale Using Elastic and Machine Learning
28mML-Driven Test Intelligence at Scale - What Works What Fails and Why It Matters
48mUsing Machine Learning to find Linux bugs
29mCritical CV-NLP Data Errors and How to Fix Them with Galileo
26mNLP for security log analysis : Learning to crawl before you run
43mBeyond Logs: Real-Time Anomaly Detection with Machine Learning