NOTE: Safetensor 4bit quant will be uploaded within the day. Cheers.

This is a GPTQ 4 bit quant of ChanSung's Elina 33b. This is a LlaMa based model; LoRA merged with latest transformers conversion. Quantized with GPTQ, --wbits 4 --act-order --true-sequential --save_safetensors c4.

128 groupsize was not used so those running this on a consumer GPU with 24GB VRAM can run it at full context (2048) without any risk of OOM.

Original LoRA: https://huggingface.co/LLMs/Alpaca-LoRA-30B-elina

Repo: https://huggingface.co/LLMs

Likely Author: https://huggingface.co/chansung

Downloads last month
9
Inference Providers NEW
This model is not currently available via any of the supported third-party Inference Providers, and the model is not deployed on the HF Inference API.