RichardErkhov/allknowingroger_-_MultiverseMath-12B-MoE-gguf

Quantization made by Richard Erkhov.

Github

Discord

Request more models

MultiverseMath-12B-MoE - GGUF

• Model creator: https://huggingface.co/allknowingroger/
• Original model: https://huggingface.co/allknowingroger/MultiverseMath-12B-MoE/

Name	Quant method	Size
MultiverseMath-12B-MoE.Q2_K.gguf	Q2_K	4.43GB
MultiverseMath-12B-MoE.IQ3_XS.gguf	IQ3_XS	4.95GB
MultiverseMath-12B-MoE.IQ3_S.gguf	IQ3_S	5.22GB
MultiverseMath-12B-MoE.Q3_K_S.gguf	Q3_K_S	5.2GB
MultiverseMath-12B-MoE.IQ3_M.gguf	IQ3_M	5.35GB
MultiverseMath-12B-MoE.Q3_K.gguf	Q3_K	5.78GB
MultiverseMath-12B-MoE.Q3_K_M.gguf	Q3_K_M	5.78GB
MultiverseMath-12B-MoE.Q3_K_L.gguf	Q3_K_L	6.27GB
MultiverseMath-12B-MoE.IQ4_XS.gguf	IQ4_XS	6.5GB
MultiverseMath-12B-MoE.Q4_0.gguf	Q4_0	6.78GB
MultiverseMath-12B-MoE.IQ4_NL.gguf	IQ4_NL	6.85GB
MultiverseMath-12B-MoE.Q4_K_S.gguf	Q4_K_S	6.84GB
MultiverseMath-12B-MoE.Q4_K.gguf	Q4_K	7.25GB
MultiverseMath-12B-MoE.Q4_K_M.gguf	Q4_K_M	7.25GB
MultiverseMath-12B-MoE.Q4_1.gguf	Q4_1	7.52GB
MultiverseMath-12B-MoE.Q5_0.gguf	Q5_0	8.26GB
MultiverseMath-12B-MoE.Q5_K_S.gguf	Q5_K_S	8.26GB
MultiverseMath-12B-MoE.Q5_K.gguf	Q5_K	8.51GB
MultiverseMath-12B-MoE.Q5_K_M.gguf	Q5_K_M	8.51GB
MultiverseMath-12B-MoE.Q5_1.gguf	Q5_1	9.01GB
MultiverseMath-12B-MoE.Q6_K.gguf	Q6_K	9.84GB
MultiverseMath-12B-MoE.Q8_0.gguf	Q8_0	12.75GB

Original model description:

license: apache-2.0 tags: - moe - frankenmoe - merge - mergekit - lazymergekit - allknowingroger/MultiverseEx26-7B-slerp - DT12the/Math-Mixtral-7B base_model: - allknowingroger/MultiverseEx26-7B-slerp - DT12the/Math-Mixtral-7B

NeuralPipe-7B-slerp

NeuralPipe-7B-slerp is a Mixture of Experts (MoE) made with the following models using LazyMergekit:

• allknowingroger/MultiverseEx26-7B-slerp
• DT12the/Math-Mixtral-7B

🧩 Configuration

base_model: allknowingroger/MultiverseEx26-7B-slerp
experts:
  - source_model: allknowingroger/MultiverseEx26-7B-slerp
    positive_prompts: ["what"]
  - source_model: DT12the/Math-Mixtral-7B
    positive_prompts: ["math"]

💻 Usage

!pip install -qU transformers bitsandbytes accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "allknowingroger/MultiverseMath-12B-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"])