CoolFace
Modelpublic

DoesntKnowAI/Plasma-8B

sourceHugging Faceupdated 2y agoView on Hugging Face
2likes8downloads
README.md68 linesDownload Raw Back to root
1---2base_model:3- DoesntKnowAI/MentalNitrogenOxide-8B4- arcee-ai/Llama-3.1-SuperNova-Lite5tags:6- merge7- mergekit8- lazymergekit9- DoesntKnowAI/MentalNitrogenOxide-8B10- arcee-ai/Llama-3.1-SuperNova-Lite11---12 13# Plasma-8B14 15Plasma-8B is a merge of the following models using [LazyMergekit](https://colab.research.google.com/drive/1obulZ1ROXHjYLn6PPZJwRR6GzgQogxxb?usp=sharing):16* [DoesntKnowAI/MentalNitrogenOxide-8B](https://huggingface.co/DoesntKnowAI/MentalNitrogenOxide-8B)17* [arcee-ai/Llama-3.1-SuperNova-Lite](https://huggingface.co/arcee-ai/Llama-3.1-SuperNova-Lite)18 19I ran out of naming ideas when naming this. Use if you want but note that this model was only created so that I can merge it with another model20 21GGUF: [DoesntKnowAI/Plasma-8B-Q8_0-GGUF](https://huggingface.co/DoesntKnowAI/Plasma-8B-Q8_0-GGUF)22 23## 🧩 Configuration24 25```yaml26slices:27  - sources:28      - model: DoesntKnowAI/MentalNitrogenOxide-8B29        layer_range: [0, 32]30        weight: 0.8631      - model: arcee-ai/Llama-3.1-SuperNova-Lite32        layer_range: [0, 32]33        weight: 0.1434merge_method: slerp35parameters:36  t:37    - model: DoesntKnowAI/MentalNitrogenOxide-8B38      value: 1.039    - model: arcee-ai/Llama-3.1-SuperNova-Lite40      value: 1.041base_model: DoesntKnowAI/MentalNitrogenOxide-8B42dtype: bfloat1643```44 45## 💻 Usage46 47```python48!pip install -qU transformers accelerate49 50from transformers import AutoTokenizer51import transformers52import torch53 54model = "DoesntKnowAI/Plasma-8B"55messages = [{"role": "user", "content": "What is a large language model?"}]56 57tokenizer = AutoTokenizer.from_pretrained(model)58prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)59pipeline = transformers.pipeline(60    "text-generation",61    model=model,62    torch_dtype=torch.float16,63    device_map="auto",64)65 66outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)67print(outputs[0]["generated_text"])68```