tinysafe-1 and tinysafe-2

These are ecosystem siblings representing iterative versions of the same safety classifier architecture, where v2 serves primarily as a learning artifact despite marginal performance improvements over v1's more practical 71M parameter footprint.

tinysafe-1
21
Experimental
tinysafe-2
20
Experimental
Maintenance 10/25
Adoption 2/25
Maturity 9/25
Community 0/25
Maintenance 10/25
Adoption 1/25
Maturity 9/25
Community 0/25
Stars: 2
Forks:
Downloads:
Commits (30d): 0
Language: Python
License: MIT
Stars: 1
Forks:
Downloads:
Commits (30d): 0
Language: Python
License: MIT
No Package No Dependents
No Package No Dependents

About tinysafe-1

jdleo/tinysafe-1

71M parameter safety classifier (DeBERTa-v3-xsmall). Dual-head: binary safe/unsafe + 7 category labels. Sub-2ms inference.

About tinysafe-2

jdleo/tinysafe-2

141M param safety model (not much better than v1, but a great learning)

Related comparisons

Scores updated daily from GitHub, PyPI, and npm data. How scores work