Files
ModelHub XC cb756eb60a 初始化项目,由ModelHub XC社区提供模型
Model: RedHatAI/DeepSeek-R1-Distill-Qwen-7B-quantized.w4a16
Source: Original Platform
2026-09-29 02:56:13 +08:00

11 lines
307 B
YAML

quant_stage:
quant_modifiers:
GPTQModifier:
dampening_frac: 0.01
ignore: [lm_head]
config_groups:
group_0:
weights: {num_bits: 4, type: int, symmetric: true, strategy: group, group_size: 128,
actorder: weight, observer: mse}
targets: [Linear]