Shane
Upload GPTQ quantized model (group_size=32)
0f4738f verified
default_stage:
default_modifiers:
GPTQModifier:
config_groups:
group_0:
targets: [Linear]
weights: {num_bits: 4, type: int, symmetric: true, group_size: 32, strategy: group,
observer: minmax}
input_activations: null
output_activations: null
ignore: [lm_head]
dampening_frac: 0.1