cs-cr/security
IB-arxiv-2607.28100-forensic-qa-accuracy
measured
In forensic question-answering over 30 held-out capture files, a frontier LLM reached 99.3% accuracy from PCAP-LM documents versus 51.0% from a token-budget-matched tshark -V prefix.
Observed quantities
accuracy_pcaplm_pct 99.3 · accuracy_tshark_pct 51.0 · n_files 30
01
Primary source
- Title
- PCAP-LM: An LLM-Native Text Representation for TLS Bulk Traffic Analysis
- Authors
- Xavier Marjou, Lucas Tamic, Ilan Jaffeux-Cheniout
- Version
- v1
- Check
- Support quote found word for word in the arXiv abstract v1, 2026-10-08
In a forensic question-answering evaluation over 30 held-out files, a frontier LLM achieves 99.3% accuracy from PCAP-LM documents versus 51.0% from a token-budget-matched tshark -V prefix
02
Boundary of this block
No additional public limitation was recorded.
03
Connections
Outgoing
No outgoing public edges.
Incoming
No incoming public edges.