test_stage: | |
quant_modifiers: | |
vLLMQuantizationModifier: | |
ignore: [lm_head] | |
config_groups: | |
group_0: | |
weights: {num_bits: 4, type: int, symmetric: true, strategy: channel} | |
targets: [Linear] | |
SparseGPTModifier: {sparsity: 0.0, quantize: true, sequential_update: false} | |