You can deploy this model and start earning money today!

RichardErkhov/
AmeerH_-_Mistral-Math-2x7b-mix-gguf

Name: AmeerH_-_Mistral-Math-2x7b-mix-gguf
Availability: InStock

Views

No views yet

Quantization made by Richard Erkhov.

Github

Discord

Request more models

Mistral-Math-2x7b-mix - GGUF

Model creator: https://huggingface.co/AmeerH/
Original model: https://huggingface.co/AmeerH/Mistral-Math-2x7b-mix/

Name	Quant method	Size
Mistral-Math-2x7b-mix.Q2_K.gguf	Q2_K	4.43GB
Mistral-Math-2x7b-mix.IQ3_XS.gguf	IQ3_XS	4.95GB
Mistral-Math-2x7b-mix.IQ3_S.gguf	IQ3_S	5.22GB
Mistral-Math-2x7b-mix.Q3_K_S.gguf	Q3_K_S	5.2GB
Mistral-Math-2x7b-mix.IQ3_M.gguf	IQ3_M	5.35GB
Mistral-Math-2x7b-mix.Q3_K.gguf	Q3_K	5.78GB
Mistral-Math-2x7b-mix.Q3_K_M.gguf	Q3_K_M	5.78GB
Mistral-Math-2x7b-mix.Q3_K_L.gguf	Q3_K_L	6.27GB
Mistral-Math-2x7b-mix.IQ4_XS.gguf	IQ4_XS	6.5GB
Mistral-Math-2x7b-mix.Q4_0.gguf	Q4_0	6.78GB
Mistral-Math-2x7b-mix.IQ4_NL.gguf	IQ4_NL	6.85GB
Mistral-Math-2x7b-mix.Q4_K_S.gguf	Q4_K_S	6.84GB
Mistral-Math-2x7b-mix.Q4_K.gguf	Q4_K	7.25GB
Mistral-Math-2x7b-mix.Q4_K_M.gguf	Q4_K_M	7.25GB
Mistral-Math-2x7b-mix.Q4_1.gguf	Q4_1	7.52GB
Mistral-Math-2x7b-mix.Q5_0.gguf	Q5_0	8.26GB
Mistral-Math-2x7b-mix.Q5_K_S.gguf	Q5_K_S	8.26GB
Mistral-Math-2x7b-mix.Q5_K.gguf	Q5_K	8.51GB
Mistral-Math-2x7b-mix.Q5_K_M.gguf	Q5_K_M	8.51GB
Mistral-Math-2x7b-mix.Q5_1.gguf	Q5_1	9.01GB
Mistral-Math-2x7b-mix.Q6_K.gguf	Q6_K	9.84GB
Mistral-Math-2x7b-mix.Q8_0.gguf	Q8_0	12.75GB

Original model description:

license: apache-2.0 tags:

moe
frankenmoe
merge
mergekit
lazymergekit
mistralai/Mistral-7B-Instruct-v0.2
meta-math/MetaMath-Mistral-7B base_model:
mistralai/Mistral-7B-Instruct-v0.2
meta-math/MetaMath-Mistral-7B

Mistral-Math-2x7b-mix

Mistral-Math-2x7b-mix is a Mixure of Experts (MoE) made with the following models using LazyMergekit:

🧩 Configuration

yaml

base_model: mistralai/Mistral-7B-Instruct-v0.2
dtype: float16
gate_mode: cheap_embed
experts:
  - source_model: mistralai/Mistral-7B-Instruct-v0.2
    positive_prompts: ["You are helpful assistant."]
  - source_model: meta-math/MetaMath-Mistral-7B
    positive_prompts: ["You are an assistant good at math."]

💻 Usage

python

!pip install -qU transformers bitsandbytes accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "AmeerH/Mistral-Math-2x7b-mix"

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"])

RichardErkhov/AmeerH_-_Mistral-Math-2x7b-mix-gguf

Original model description:

Mistral-Math-2x7b-mix

🧩 Configuration

💻 Usage

RichardErkhov/
AmeerH_-_Mistral-Math-2x7b-mix-gguf