AI & ML interests

Deep experimentation on ablation and quanting. Goal: trim the fat on models. Question: Is all data, good data, or, is some just noise?

Recent Activity

deucebucket  updated a collection 11 days ago
Gemma 4 - Cerebellum
deucebucket  updated a collection 11 days ago
North-Mini-Code - Cerebellum
deucebucket  updated a collection 11 days ago
Fits 16 GB cards
View all activity

DB-Cerebellum 's collections 8

Fits 16 GB cards
Weights under ~14 GB so a 16 GB GPU runs them fully loaded with room for context. Measured footprints on each card.
Gemma 4 - Cerebellum
Fits 8 GB cards
Cerebellum builds whose weights leave context headroom on 8 GB GPUs. Measured footprints are on each card.
Fits 8 GB cards
Cerebellum builds whose weights leave context headroom on 8 GB GPUs. Measured footprints are on each card.
Fits 16 GB cards
Weights under ~14 GB so a 16 GB GPU runs them fully loaded with room for context. Measured footprints on each card.
Gemma 4 - Cerebellum