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
48
Columbidae/Voxtral-Mini-3B-Text-Only-hf
5B
•
Updated
•
201
•
2
Columbidae/gemma-3-12b-it-text
12B
•
Updated
•
2
Columbidae/voxtral-mini-3b-text-only
Updated
•
3
Columbidae/GLM-4-32B-Tulu-Instruct
33B
•
Updated
•
8
Columbidae/CavesOfQwen3-exp16-Unhealed
Text Generation
•
31B
•
Updated
•
15
•
1
Columbidae/medgemma-27b-multimodal
Updated
•
237
•
1
Columbidae/Qwen3-16B-A3B-Base
16B
•
Updated
•
554
Columbidae/Qwen3-16B-A3B-Base-Old-Archived
16B
•
Updated
•
2
Columbidae/Qwen3-16B-A3B-Tulu-Mini
Text Generation
•
16B
•
Updated
Columbidae/Qwen3-21B-Base-Zeroed
Text Generation
•
21B
•
Updated
•
14
•
1
datasets
0
None public yet