Qwen2.5 models with the tokenizers padded with extra unused special tokens to match the model embedding size. Workaround for various bugs.
Columba Labs
community
AI & ML interests
Birb make bots go brrr
Recent Activity
View all activity
models 67
Columbidae/gemma4-31b-pt-embed-it
Image-Text-to-Text • 33B • Updated • 12
Columbidae/gemma-3-4b-pt-text-only
4B • Updated • 6
Columbidae/Ministral-3-14B-Instruct-TextOnly-BF16
14B • Updated • 4
Columbidae/Ministral-3-14B-Base-TextOnly-BF16
14B • Updated • 8 • 1
Columbidae/apertus-12b-chatml-untrained
12B • Updated • 1
Columbidae/Apertus-8B-2509-ChatML-trained-embeddings
Text Generation • 4B • Updated • 3
Columbidae/granite-4.0-h-small-half-slerp
32B • Updated • 3 • 1
Columbidae/GLM-Air-Q4KM-Yarn-65k
Updated • 2
Columbidae/GLM-Air-Q4KM-Yarn
Updated • 9
Columbidae/Nemo-Base-Pruned-LP-Double
8B • Updated • 3
datasets 0
None public yet