head_type: luh | |
feature_extractor: | |
- name: luh.feature_extractors.lookback_lens | |
- name: luh.feature_extractors.basic_hidden_states | |
layer_nums: | |
- -1 | |
- -5 | |
- -15 | |
- -30 | |
- name: luh.feature_extractors.basic_attention | |
layer_nums: | |
- -1 | |
- -2 | |
- -3 | |
- -4 | |
- -5 | |
- -10 | |
- -15 | |
- -20 | |
- -25 | |
- -30 | |
attn_history_sz: 2 | |
pool: false | |
- name: luh.feature_extractors.token_probabilities | |
top_n: 4 | |
uncertainty_head: | |
head_dim: 128 | |
n_layers: 2 | |
n_heads: 4 | |
dropout: 0.2 | |
enable_feature_projection_layer: true | |
interim_dim: 128 | |