Lens Network Traffic Classification Benchmark

Twelve classification tasks derived from publicly available academic traffic datasets serve as a benchmark for evaluating the Lens knowledge-guided foundation model. The dataset bundles these tasks as HuggingFace configurations with standardized train, validation, and test splits. It was created by Charles59 and last updated on June 29, 2026.

Use Cases

• Benchmarking network traffic classification models based on the 12 bundled tasks.
• Evaluating knowledge-guided foundation models like Lens on downstream traffic analysis.
• Comparing model performance across standardized train, validation, and test splits.

Strengths

• Provides 12 distinct classification tasks for comprehensive evaluation.
• Features standardized train, validation, and test splits for each task.
• Derived from publicly available academic datasets via the NetBench benchmark.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect temporal or source bias inherent to the original academic datasets.

Provenance

Source

Publicly available academic traffic datasets obtained via the NetBench benchmark.

Collection Method

Derived and bundled from existing benchmark datasets.

Freshness

Last updated 2026-06-29 16:38:27.