Files
ModelHub XC fe294906a5 初始化项目,由ModelHub XC社区提供模型
Model: RedHatAI/Llama-3.2-1B-Instruct-quantized.w8a8
Source: Original Platform
2026-05-06 12:14:02 +08:00

19 lines
489 B
YAML

quant_stage:
quant_modifiers:
SmoothQuantModifier:
smoothing_strength: 0.7
mappings:
- - ['re:.*q_proj', 're:.*k_proj', 're:.*v_proj']
- re:.*input_layernorm
- - ['re:.*gate_proj', 're:.*up_proj']
- re:.*post_attention_layernorm
- - ['re:.*down_proj']
- re:.*up_proj
GPTQModifier:
sequential_update: true
dampening_frac: 0.01
ignore: [lm_head]
scheme: W8A8
targets: [Linear]
observer: mse