This is fully exp model!

base_model: beomi/EXAONE-3.5-2.4B-Instruct-Llamafied
gate_mode: random
architecture: mixtral
experts_per_token: 2
dtype: bfloat16
experts:
  - source_model: beomi/EXAONE-3.5-2.4B-Instruct-Llamafied
  - source_model: unsloth/Phi-3.5-mini-instruct

Downloads last month: 8

Safetensors

Model size

4.93B params

Tensor type

BF16

Inference API

Other

Inference API (serverless) does not yet support transformers models for this pipeline type.