ByteMaster01's picture
Upload quantized model artifacts
576c02d verified
Raw
History Blame Contribute Delete
264 Bytes
default_stage:
default_modifiers:
GPTQModifier:
targets: [Linear]
ignore: [lm_head]
scheme: W4A16
bypass_divisibility_checks: false
block_size: 128
dampening_frac: 0.01
actorder: static
offload_hessians: false