OPT-30B-Erebus-4bit-128g

Model description

Warning: THIS model is NOT suitable for use by minors. The model will output X-rated content.

This is a 4-bit GPTQ quantization of OPT-30B-Erebus, original model: https://huggingface.co/KoboldAI/OPT-30B-Erebus

Quantization Information

Quantized with: https://github.com/0cc4m/GPTQ-for-LLaMa

python opt.py --wbits 4 models/OPT-30B-Erebus c4 --groupsize 128 --save models/OPT-30B-Erebus-4bit-128g/OPT-30B-Erebus-4bit-128g.pt
python opt.py --wbits 4 models/OPT-30B-Erebus c4 --groupsize 128 --save_safetensors models/OPT-30B-Erebus-4bit-128g/OPT-30B-Erebus-4bit-128g.safetensors

Output generated in 54.23 seconds (0.87 tokens/s, 47 tokens, context 44, seed 593020441)

Command text-generation-webui:

https://github.com/oobabooga/text-generation-webui

call python server.py --model_type gptj --model OPT-30B-Erebus-4bit-128g --chat --wbits 4 --groupsize 128 --xformers --sdp-attention

Credit

https://huggingface.co/notstoic

License

OPT-30B is licensed under the OPT-175B license, Copyright (c) Meta Platforms, Inc. All Rights Reserved.