File size: 1,355 Bytes
278786e 06cda4b 278786e |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 |
EXL2 quants of Mistral-7B-instruct
Converted from [Mistral-7B-Instruct-v0.1](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.1). This is a
straight conversion, but I have modified the `config.json` to make the default context size 7168 tokens, since in
initial testing the model becomes unstable a while after that. It's possible that sliding window attention will
allow the model to use its advertised 32k-token context, but this hasn't been tested yet.
[2.50 bits per weight](https://huggingface.co/turboderp/Mistral-7B-instruct-exl2/tree/2.5bpw)
[2.70 bits per weight](https://huggingface.co/turboderp/Mistral-7B-instruct-exl2/tree/2.7bpw)
[3.00 bits per weight](https://huggingface.co/turboderp/Mistral-7B-instruct-exl2/tree/3.0bpw)
[3.50 bits per weight](https://huggingface.co/turboderp/Mistral-7B-instruct-exl2/tree/3.5bpw)
[4.00 bits per weight](https://huggingface.co/turboderp/Mistral-7B-instruct-exl2/tree/4.0bpw)
[4.65 bits per weight](https://huggingface.co/turboderp/Mistral-7B-instruct-exl2/tree/4.65bpw)
[5.00 bits per weight](https://huggingface.co/turboderp/Mistral-7B-instruct-exl2/tree/5.0bpw)
[6.00 bits per weight](https://huggingface.co/turboderp/Mistral-7B-instruct-exl2/tree/6.0bpw)
[measurement.json](https://huggingface.co/turboderp/Mistral-7B-instruct-exl2/blob/main/measurement.json) |