license: apache-2.0
tags:
- MoE
- merge
- mergekit
- Mistral
- Microsoft/WizardLM-2-7B
WizardLM-2-4x7B-MoE
WizardLM-2-4x7B-MoE is an experimental MoE model made with Mergekit. It was made by combining four WizardLM-2-7B models using the random gate mode.
Please be sure to set experts per token to 4 for the best results! Context length should be the same as Mistral-7B-Instruct-v0.1 (8k tokens). For instruction templates, Vicuna-v1.1 is recommended.
Quanitized versions
EXL2 (for fast GPU-only inference):
6_0bpw: https://huggingface.co./Skylaude/WizardLM-2-4x7B-MoE-exl2-6_0bpw (for GPU's with 20+ GB of vram)
4_25bpw: [coming soon] (for GPU's with 16+ GB of vram)
3_0bpw: https://huggingface.co./Skylaude/WizardLM-2-4x7B-MoE-exl2-3_0bpw (for GPU's with 12+ GB of vram)
GGUF (for mixed GPU+CPU inference or CPU-only inference):
https://huggingface.co./mradermacher/WizardLM-2-4x7B-MoE-GGUF
Thanks to Michael Radermacher for making these quants!
Evaluation
I don't expect this model to be that great since it's something that I made as an experiment. However, I will submit it to the Open LLM Leaderboard to see how it matches up against some other models (particularly WizardLM-2-7B and WizardLM-2-70B).
Mergekit config
base_model: models/WizardLM-2-7B
gate_mode: random
dtype: float16
experts_per_token: 4
experts:
- source_model: models/WizardLM-2-7B
- source_model: models/WizardLM-2-7B
- source_model: models/WizardLM-2-7B
- source_model: models/WizardLM-2-7B