README.md

metadata

license: apache-2.0
tags:
  - moe
  - frankenmoe
  - merge
  - mergekit
  - lazymergekit
  - argilla/notus-7b-v1
  - GritLM/GritLM-7B
base_model:
  - argilla/notus-7b-v1
  - GritLM/GritLM-7B

GK-MoE-0.1

GK-MoE-0.1 is a Mixture of Experts (MoE) made with the following models using LazyMergekit:

• argilla/notus-7b-v1
• GritLM/GritLM-7B

🧩 Configuration

base_model: GritLM/GritLM-7B
experts:
  - source_model: argilla/notus-7b-v1
    positive_prompts:
    - "chat"
    - "assistant"
    - "tell me"
    - "explain"
    - "I want"
    - "reason"
    - "math"
    - "mathematics"
    - "solve"
    - "count"
  - source_model: GritLM/GritLM-7B
    positive_prompts:
    - "code"
    - "VB.NET"
    - "vb.net"
    - "programming"
    - "algorithm"
    - "develop"
    ```

## 💻 Usage

```python
!pip install -qU transformers bitsandbytes accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "powermove72/GK-MoE-0.1"

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"])