MoE'd up:

Which were the two most interesting llama3 finetunes as of yet. Resulting model seems OK. It's not on Miqu's level, anyway.

Blah, blah, llama 3 license (no tag for it yet). Also not going to name my model Llama-3-Copus. Come at me, Zuck.

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 lodrick-the-lafted/Copus-2x8B

Quantizations