alexmarques's picture
Upload folder using huggingface_hub
7f1db53 verified
raw
history blame
489 Bytes
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