dbrx-instruct-exl2 / README.md
turboderp's picture
Update README.md
cedaba5 verified
EXL2 quants of [dbrx-instruct](https://huggingface.co/alpindale/dbrx-instruct).
Tokenizer from [here](https://huggingface.co/Xenova/dbrx-instruct-tokenizer)
[2.20 bits per weight](https://huggingface.co/turboderp/dbrx-instruct-exl2/tree/2.2bpw)
[2.30 bits per weight](https://huggingface.co/turboderp/dbrx-instruct-exl2/tree/2.3bpw)
[2.75 bits per weight](https://huggingface.co/turboderp/dbrx-instruct-exl2/tree/2.75bpw)
[3.00 bits per weight](https://huggingface.co/turboderp/dbrx-instruct-exl2/tree/3.0bpw)
[3.40 bits per weight](https://huggingface.co/turboderp/dbrx-instruct-exl2/tree/3.4bpw)
[3.75 bits per weight](https://huggingface.co/turboderp/dbrx-instruct-exl2/tree/3.75bpw)
[measurement.json](https://huggingface.co/turboderp/dbrx-instruct-exl2/blob/main/measurement.json)