gpt3m-flash-rotary.yaml 185 Bytes
Newer Older
Tri Dao's avatar
Tri Dao committed
1
2
3
4
5
6
7
8
# @package _global_
defaults:
  - /experiment/pile/gpt3m-flash.yaml

model:
  config:
    max_position_embeddings: 0  # Disable absolute position embedding
    rotary_emb_fraction: 0.5