lavawolfiee
/

Mixtral-8x7B-Instruct-v0.1-offloading-demo

Text Generation

Transformers

Safetensors

5 languages

mixtral

text-generation-inference

conversational

Inference Endpoints

8-bit precision

Model card Files Files and versions Community

Edit model card

Attention quantization: HQQ 4-bit, groupsize 64, compress zero, compress scale with groupsize 256
Experts quantization: HQQ 2-bit, groupsize 16, compress zero, compress scale with groupsize 128

Downloads last month: 4,732

Safetensors

Model size

17.9B params

Tensor type

FP16