metadata

license: apache-2.0
tags:
  - moe
  - mixtral

MetaModel_moe

This model is a Mixure of Experts (MoE) made with mergekit (mixtral branch). It uses the following base models:

🧩 Configuration

base_model: gagan3012/MetaModel
gate_mode: hidden
dtype: bfloat16
experts:
- source_model: gagan3012/MetaModel
- source_model: jeonsworld/CarbonVillain-en-10.7B-v2
- source_model: jeonsworld/CarbonVillain-en-10.7B-v4
- source_model: TomGrc/FusionNet_linear

💻 Usage

!pip install -qU transformers bitsandbytes accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "gagan3012/MetaModel_moe"

tokenizer = AutoTokenizer.from_pretrained(model)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    model_kwargs={"torch_dtype": torch.float16, "load_in_4bit": True},
)

messages = [{"role": "user", "content": "Explain what a Mixture of Experts is in less than 100 words."}]
prompt = pipeline.tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
print(outputs[0]["generated_text"])