models/v004-contextual/config.json

{
  "contextual_action": true,
  "encoder": "mean",
  "lexical_features": true,
  "vocab_size": 1683,
  "width": 64
}