README.md

metadata

license: apache-2.0
tags:
  - moe
  - frankenmoe
  - merge
  - mergekit
  - lazymergekit
  - winglian/Llama-3-8b-64k-PoSE
  - PygmalionAI/pygmalion-2-7b
base_model:
  - winglian/Llama-3-8b-64k-PoSE
  - PygmalionAI/pygmalion-2-7b

Llama-3-8b-pygmalion-2-7b-v1

Tried merging LLAMA 2 and LLAMA 3. This model does not provide and legible output. Llama-3-8b-pygmalion-2-7b-v1 is a Mixture of Experts (MoE) made with the following models using LazyMergekit:

• winglian/Llama-3-8b-64k-PoSE
• PygmalionAI/pygmalion-2-7b

🧩 Configuration

dtype: float16
gate_mode: cheap_embed
experts:
  - source_model: winglian/Llama-3-8b-64k-PoSE
    positive_prompts: ["You are an intelligent bot that is smart and sassy"]
  - source_model: PygmalionAI/pygmalion-2-7b
    positive_prompts: ["You are a sexy girl that loves to roleplay"]```

## 💻 Usage

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

from transformers import AutoTokenizer
import transformers
import torch

model = "farhananis005/Llama-3-8b-pygmalion-2-7b-v1"

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"])