Meow. This an experimental mixture of expert model with just 2 experts based on Llama 3 Instruct plain in combo with finetune. Specifically, it is built on top of the Meta-Llama-3-8B-Instruct model and finetune is trained on Argilla Capybara dataset.

Experimental mixture of 2 experts Llama3-8b-Instruct

Built with Llama 3

Downloads last month
12
Safetensors
Model size
13.7B params
Tensor type
BF16
·
Inference Providers NEW
This model is not currently available via any of the supported Inference Providers.

Model tree for nisten/llama3-2x8b-MoE-41k-experiment1

Finetuned
(535)
this model

Dataset used to train nisten/llama3-2x8b-MoE-41k-experiment1