Stablebeluga2

🚀 Stable Beluga 2

Stable Beluga 2 is a fine - tuned Llama2 70B model for text generation, offering efficient weight storage and optimized performance.

🚀 Quick Start

Start chatting with Stable Beluga 2 using the following code snippet:

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline

tokenizer = AutoTokenizer.from_pretrained("stabilityai/StableBeluga2", use_fast=False)
model = AutoModelForCausalLM.from_pretrained("stabilityai/StableBeluga2", torch_dtype=torch.float16, low_cpu_mem_usage=True, device_map="auto")
system_prompt = "### System:\nYou are Stable Beluga, an AI that follows instructions extremely well. Help as much as you can. Remember, be safe, and don't do anything illegal.\n\n"

message = "Write me a poem please"
prompt = f"{system_prompt}### User: {message}\n\n### Assistant:\n"
inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
output = model.generate(**inputs, do_sample=True, top_p=0.95, top_k=0, max_new_tokens=256)

print(tokenizer.decode(output[0], skip_special_tokens=True))

✨ Features

Changes in this fork

This repository contains the model from the stabilityai/StableBeluga2 repository with the following changes:

1. Storing weights in bfloat16 instead of float32. This leads to 2x smaller files and a small quality loss, which is not significant compared to the loss caused by NF4 quantization used in Petals by default.
2. Storing weights in small shards. Each transformer block is stored in its own shard (1.71 GB each). The input and output embeddings and adjacent layernorms are in a separate shard (1.05 GB) too. This way, Petals clients and servers don't have to download any excess data besides the layers they actually use.
3. Using Safetensors instead of Pickle. This allows faster loading with smaller RAM requirements.

📚 Documentation

Model Description

Stable Beluga 2 is a Llama2 70B model finetuned on an Orca style Dataset

Other Beluga Models

• StableBeluga 1 - Delta
• StableBeluga 13B
• StableBeluga 7B

Model Details

Training Dataset

Stable Beluga 2 is trained on our internal Orca - style dataset

Training Procedure

Models are learned via supervised fine - tuning on the aforementioned datasets, trained in mixed - precision (BF16), and optimized with AdamW. We outline the following hyperparameters:

🔧 Technical Details

Ethical Considerations and Limitations

Beluga is a new technology that carries risks with use. Testing conducted to date has been in English, and has not covered, nor could it cover all scenarios. For these reasons, as with all LLMs, Beluga's potential outputs cannot be predicted in advance, and the model may in some instances produce inaccurate, biased or other objectionable responses to user prompts. Therefore, before deploying any applications of Beluga, developers should perform safety testing and tuning tailored to their specific applications of the model.

📄 License

Fine - tuned checkpoints (Stable Beluga 2) is licensed under the STABLE BELUGA NON - COMMERCIAL COMMUNITY LICENSE AGREEMENT

How to cite

@misc{StableBelugaModels, 
      url={[https://huggingface.co/stabilityai/StableBeluga2](https://huggingface.co/stabilityai/StableBeluga2)}, 
      title={Stable Beluga models}, 
      author={Mahan, Dakota and Carlow, Ryan and Castricato, Louis and Cooper, Nathan and Laforte, Christian}
}

Citations

@misc{touvron2023llama,
      title={Llama 2: Open Foundation and Fine - Tuned Chat Models}, 
      author={Hugo Touvron and Louis Martin and Kevin Stone and Peter Albert and Amjad Almahairi and Yasmine Babaei and Nikolay Bashlykov and Soumya Batra and Prajjwal Bhargava and Shruti Bhosale and Dan Bikel and Lukas Blecher and Cristian Canton Ferrer and Moya Chen and Guillem Cucurull and David Esiobu and Jude Fernandes and Jeremy Fu and Wenyin Fu and Brian Fuller and Cynthia Gao and Vedanuj Goswami and Naman Goyal and Anthony Hartshorn and Saghar Hosseini and Rui Hou and Hakan Inan and Marcin Kardas and Viktor Kerkez and Madian Khabsa and Isabel Kloumann and Artem Korenev and Punit Singh Koura and Marie - Anne Lachaux and Thibaut Lavril and Jenya Lee and Diana Liskovich and Yinghai Lu and Yuning Mao and Xavier Martinet and Todor Mihaylov and Pushkar Mishra and Igor Molybog and Yixin Nie and Andrew Poulton and Jeremy Reizenstein and Rashi Rungta and Kalyan Saladi and Alan Schelten and Ruan Silva and Eric Michael Smith and Ranjan Subramanian and Xiaoqing Ellen Tan and Binh Tang and Ross Taylor and Adina Williams and Jian Xiang Kuan and Puxin Xu and Zheng Yan and Iliyan Zarov and Yuchen Zhang and Angela Fan and Melanie Kambadur and Sharan Narang and Aurelien Rodriguez and Robert Stojnic and Sergey Edunov and Thomas Scialom},
      year={2023},
      eprint={2307.09288},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}

@misc{mukherjee2023orca,
      title={Orca: Progressive Learning from Complex Explanation Traces of GPT - 4}, 
      author={Subhabrata Mukherjee and Arindam Mitra and Ganesh Jawahar and Sahaj Agarwal and Hamid Palangi and Ahmed Awadallah},
      year={2023},
      eprint={2306.02707},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}