NeuralStar_FusionWriter_4x7b

NeuralStar_FusionWriter_4x7b is a Mixture of Experts (MoE) made with the following models using LazyMergekit:

⚡ Quantized Models

Special thanks to MRadermacher for the static and imatrix quantized models

.GGUF https://huggingface.co/mradermacher/NeuralStar_FusionWriter_4x7b-GGUF

IMatrix https://huggingface.co/mradermacher/NeuralStar_FusionWriter_4x7b-i1-GGUF

🧩 Configuration

base_model: mlabonne/AlphaMonarch-7B
experts: 
 - source_model: mlabonne/AlphaMonarch-7B
 positive_prompts: 
 - "chat"
 - "assistant"
 - "tell me"
 - "explain"
 - "ideas"
 - source_model: OmnicromsBrain/Eros_Scribe-7b
 positive_prompts:
 - "adult"
 - "sex"
 - "explicit"
 - "nsfw"
 - "gory"
 - source_model: SanjiWatsuki/Kunoichi-DPO-v2-7B
 positive_prompts:
 - "story"
 - "character"
 - "scene"
 - "plot"
 - "editor"
 - source_model: OmnicromsBrain/NeuralStar_Fusion-7B
 positive_prompts:
 - "codex"
 - "write"
 - "outline"
 - "scenebeat"
 - "prose"

💻 Usage

!pip install -qU transformers bitsandbytes accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "OmnicromsBrain/NeuralStar_FusionWriter_4x7b"

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"])