This is an attempt to apply the pruning method that created kalomaze/Qwen3-16B-A3B to Qwen/Qwen3-30B-A3B-Base.

Pruning was done based on the assumption that the same experts could be pruned from the base model as were pruned from the instruct model. This may or may not be correct.

Downloads last month
1,287
Safetensors
Model size
16B params
Tensor type
BF16
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for Columbidae/Qwen3-16B-A3B-Base

Finetuned
(23)
this model
Adapters
1 model
Finetunes
2 models
Quantizations
2 models