-
Mezzo Content Guard Demo
🐠Analyze text for violence, hate, sexual, toxic, and self‑harm content
-
RyanStudio/Mezzo-Content-Guard-Large
Text Classification • 0.4B • Updated • 3 -
RyanStudio/Mezzo-Content-Guard-Base
Text Classification • 0.1B • Updated • 3 -
RyanStudio/Mezzo-Content-Guard-Small
Text Classification • 82.1M • Updated • 16
Ryan Studio
RyanStudio
AI & ML interests
My mission is to make small, but high quality models that can be comfortably run on consumer GPUs for VRAM poor people like me
Recent Activity
published a model about 9 hours ago
RyanStudio/Mezzo-Content-Guard-Small published a model about 9 hours ago
RyanStudio/Mezzo-Content-Guard-Base published a model about 9 hours ago
RyanStudio/Mezzo-Content-Guard-LargeOrganizations
A series of prompt guard models built on DeBERTa-v3
Prompt Guard Models trained to detect jailbreaking and prompt injections. Made with xlm-roberta and distilbert
-
Mezzo Prompt Guard Demo
🛡1A demo for Mezzo Prompt Guard v2 models
-
RyanStudio/Mezzo-Prompt-Guard-v2-Large
Text Classification • 0.6B • Updated • 42 • 3 -
RyanStudio/Mezzo-Prompt-Guard-v2-Base
Text Classification • 0.3B • Updated • 5 -
RyanStudio/Mezzo-Prompt-Guard-v2-Small
Text Classification • 0.1B • Updated • 39
Qwen 3.5 Models finetuned for Rag Query Rewriting based on a conversation history
-
Mezzo Content Guard Demo
🐠Analyze text for violence, hate, sexual, toxic, and self‑harm content
-
RyanStudio/Mezzo-Content-Guard-Large
Text Classification • 0.4B • Updated • 3 -
RyanStudio/Mezzo-Content-Guard-Base
Text Classification • 0.1B • Updated • 3 -
RyanStudio/Mezzo-Content-Guard-Small
Text Classification • 82.1M • Updated • 16
Prompt Guard Models trained to detect jailbreaking and prompt injections. Made with xlm-roberta and distilbert
-
Mezzo Prompt Guard Demo
🛡1A demo for Mezzo Prompt Guard v2 models
-
RyanStudio/Mezzo-Prompt-Guard-v2-Large
Text Classification • 0.6B • Updated • 42 • 3 -
RyanStudio/Mezzo-Prompt-Guard-v2-Base
Text Classification • 0.3B • Updated • 5 -
RyanStudio/Mezzo-Prompt-Guard-v2-Small
Text Classification • 0.1B • Updated • 39
A series of prompt guard models built on DeBERTa-v3
Qwen 3.5 Models finetuned for Rag Query Rewriting based on a conversation history