2-bit HQQ quantized version of Meta-Llama-3.1-405B (base version). I expect the quality to be very poor, useful mainly for testing. Quantization parameters:
nbits=2, group_size=128, quant_zero=True, quant_scale=True, axis=0
Shards have been split with "split", to recombine:
cat qmodel_shard* > qmodel.pt
