whitiger's picture
Upload quantized EXAONE model
9ac5fb1 verified
default_stage:
default_modifiers:
SmoothQuantModifier:
smoothing_strength: 0.8
mappings:
- - - .*norm.*
- [.*(q|k|v)_proj]
ignore: []
GPTQModifier:
targets: [Linear]
ignore: [embed_tokens, lm_head, model\.layers\.29\.self_attn\.(k_proj|v_proj)\.weight,
model\.layers\.29\.mlp\.down_proj\.weight]
scheme: W8A8
kv_cache_scheme:
num_bits: 8
type: int
symmetric: true
group_size: null
strategy: tensor
block_structure: null
dynamic: false
actorder: null
scale_dtype: null
zp_dtype: null
observer: minmax
observer_kwargs: {}
weight_observer: minmax
input_observer: minmax
block_size: 128
dampening_frac: 0.01
actorder: static
offload_hessians: false