Instructions to use prince-canuma/Llama-3-6B-v0 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use prince-canuma/Llama-3-6B-v0 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="prince-canuma/Llama-3-6B-v0")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("prince-canuma/Llama-3-6B-v0")
model = AutoModelForCausalLM.from_pretrained("prince-canuma/Llama-3-6B-v0", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use prince-canuma/Llama-3-6B-v0 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "prince-canuma/Llama-3-6B-v0"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "prince-canuma/Llama-3-6B-v0",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/prince-canuma/Llama-3-6B-v0

SGLang

How to use prince-canuma/Llama-3-6B-v0 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "prince-canuma/Llama-3-6B-v0" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "prince-canuma/Llama-3-6B-v0",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "prince-canuma/Llama-3-6B-v0" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "prince-canuma/Llama-3-6B-v0",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use prince-canuma/Llama-3-6B-v0 with Docker Model Runner:
```
docker model run hf.co/prince-canuma/Llama-3-6B-v0
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Model Summary

Introducing the world's first Llama-3 base model with 6B parameters. This model is a untrained model which was created from Meta-Llama-3-8B using a technique called downcycling .

You can check trained version of this model here: https://huggingface.co/prince-canuma/Llama-3-6B-v0.1

Model Description

Developed by: Prince Canuma
Sponsored by: General
Model type: Llama
License: Llama-3

Model Sources

Repository: https://github.com/Blaizzy/Coding-LLMs-from-scratch/tree/main/Llama-3
Video: https://youtube.com/playlist?list=PLDn_JsyofyfTH5_5V1MNb8UYKxMl6IMNy&si=5Y4cm-6wrMOD1Abr

Citation

BibTeX:

@misc{prince2024downcycling,
      title={Efficient LLM Downcycling: Generating Diverse Model Sizes from Pretrained Giants},
      author={Prince Canuma},
      year={2024},
}

Thank You!

I want to extend my heartfelt thanks to the community for the invaluable expertise and unwavering support.

Additionally, I would like to thank Viet from General Catalyst (GC) for providing me with the much needed compute.

This is my most ambitious project yet, and it wouldn't have been possible without the incredible open-source ML community!

Developers, I am eager to see and hear about the innovative fine-tunes and applications you create.

Users, I am excited to learn about your experiences and use cases.

Thank you for your interest and support!

References:

@misc{komatsuzaki2023sparse,
      title={Sparse Upcycling: Training Mixture-of-Experts from Dense Checkpoints}, 
      author={Aran Komatsuzaki and Joan Puigcerver and James Lee-Thorp and Carlos Riquelme Ruiz and Basil Mustafa and Joshua Ainslie and Yi Tay and Mostafa Dehghani and Neil Houlsby},
      year={2023},
      eprint={2212.05055},
      archivePrefix={arXiv},
      primaryClass={cs.LG}
}

@misc{sanyal2024pretraining,
      title={Pre-training Small Base LMs with Fewer Tokens}, 
      author={Sunny Sanyal and Sujay Sanghavi and Alexandros G. Dimakis},
      year={2024},
      eprint={2404.08634},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}

Downloads last month: 29

Safetensors

Model size

6B params

Tensor type

F16

Model tree for prince-canuma/Llama-3-6B-v0

Finetunes

1 model

Quantizations

13 models

Papers for prince-canuma/Llama-3-6B-v0

Pre-training Small Base LMs with Fewer Tokens

Paper • 2404.08634 • Published Apr 12, 2024 • 36

Sparse Upcycling: Training Mixture-of-Experts from Dense Checkpoints

Paper • 2212.05055 • Published Dec 9, 2022 • 6