File size: 1,275 Bytes
cb591f7 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 | default_stage:
default_modifiers:
SmoothQuantModifier:
smoothing_strength: 0.9
mappings:
- !!python/tuple
- ['re:.*q_proj', 're:.*k_proj', 're:.*v_proj']
- re:.*input_layernorm
- !!python/tuple
- ['re:.*gate_proj', 're:.*up_proj']
- re:.*post_attention_layernorm
ignore: []
QuantizationModifier:
config_groups:
group_0:
targets: [Linear]
weights:
num_bits: 4
type: float
symmetric: true
group_size: 16
strategy: tensor_group
block_structure: null
dynamic: false
actorder: null
observer: mse
observer_kwargs: {}
input_activations:
num_bits: 4
type: float
symmetric: true
group_size: 16
strategy: tensor_group
block_structure: null
dynamic: local
actorder: null
observer: minmax
observer_kwargs: {}
output_activations: null
format: null
targets: [Linear]
ignore: ['re:.*lm_head.*', 're:.*multi_modal_projector.*', 're:.*vision_tower.*', 're:.*model.norm.*',
're:.*embed_tokens.*']
|