jina-v5-omni-nano-test-time-image-tagging kyr0 · COMPLETED
Open-vocab multi-label image tagging with a single frozen model (jina-embeddings-v5-omni-nano-mlx). Zero training, no second model. Tokenizer-vocab labels + patch-level + CWR multi-crop test-time compute. mAP 0.710 on COCO-150.
github.com/kyr0/jina-v5-omni-nano-test-time-image-tagging · ★ 0 · Forks 0 · Size 4.6 MB
SUMMARY
Technologies 4
Scored 4
Observed 4
Practices 6
Evidence 5
Skips 0
COVERAGE
Analyzed 40 files · 9 commits · 0 API calls
TECHNOLOGIES & DEPTH
Markdown LANGUAGE Depth 70
6 files · PRODUCTION
Python LANGUAGE Depth 70
17 files · PRODUCTION
pip BUILD_TOOL Depth 80
1 files · CONFIGURATION
NumPy LIBRARY Depth 78
17 files · PRODUCTION
PRACTICES
documentation · observedautomated_tests · absentcontinuous_integration · absentcontainerization · absentlinting · absentformatting · absent
ACTIVITY & OWNERSHIP
First commit 2026-07-11
Last commit 2026-07-22
Active months 1
Commits 9