AlexWortega

/

miqu-1-70b-AQLM-2Bit-1x16-hf

like 27

Unfortunately, the AlexWortega/miqu-1-70b-AQLM-2Bit-1x16-hf model requires more memory than what is available on a 16GB V100 GPU, as evidenced by the OutOfMemoryError:

OutOfMemoryError: CUDA out of memory. Tried to allocate 56.00 MiB. GPU 0 has a total capacity of 15.77 GiB of which 2.38 MiB is free. 
Including non-PyTorch memory, this process has 15.77 GiB memory in use. Of the allocated memory 15.46 GiB is allocated by PyTorch, 
and 4.05 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting 
PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation.  See documentation for Memory 
Management  (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)

No, better to use 3090

@lxyuan this model requires at least 18.8Gb to be loaded

Does it work fine on 2x16 gpus?