Files
modeling_zhaocui/deep_learning/Q1/outputs/tsfa/report_bundle
..
2026-09-23 23:37:34 +08:00

TSFA five-fold result bundle

Seven frozen-alignment comparisons on 100 held-out clips from 37 video IDs. All Python runs use the project uv environment. Read run_manifest.json for the complete protocol and limitations. paired_contrasts.csv reports left minus right with video-ID-cluster bootstrap confidence intervals. The full outputs/tsfa/ directory retains per-clip CSVs and probe checkpoints.

The alignment_only_* summaries pool identical train-fold-standardized raw source features through each method's alignment matrix. They isolate source selection from native model value/output projections. tsfa_temporal_diagnostics_summary.csv reports MVR, span, entropy, and source coverage.