metadata

license: apache-2.0
tags:
  - moe
  - merge
  - mergekit
  - lazymergekit
  - Felladrin/TinyMistral-248M-SFT-v4
  - Locutusque/LocutusqueXFelladrin-TinyMistral248M-Instruct
  - Locutusque/TinyMistral-248M-v2-Instruct
  - Locutusque/TinyMistral-248M-v2.5

🌟 Buying me coffee is a direct way to show support for this project.

TinyMixtral-4x248M-MoE

TinyMixtral-4x248M-MoE is a Mixure of Experts (MoE) made with the following models using LazyMergekit:

💻 Usage

!pip install -qU transformers bitsandbytes accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "Isotonic/TinyMixtral-4x248M-MoE"

tokenizer = AutoTokenizer.from_pretrained(model)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    tokenizer=tokenizer
)

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"])