name: mini_backbone
seed: 1
mode: backbone_only
model:
  name: mini
  vocab_size: 128
  d_model: 96
  n_layers: 2
  n_heads: 4
  n_kv_heads: 2
  ffn_width: 256
  max_seq_len: 320
  w: 32
  ffn_w: 96
  n_slots: 16
  n_editor_heads: 2
  r_max: 8
  readout_zero_init: false
  readout_init_std: 0.2
  ctrl_input: h
  ctrl_use_entropy: true
init:
  freeze_backbone: false
data:
  kind: synthetic
  tokenizer: char
  splits_dir: data/splits/synthetic_v1
  loss_span: all
  eval_splits:
  - dev_iid
train:
  batch_size: 16
  seq_len: 320
  steps: 150
  lr: 0.001
  warmup: 10
  weight_decay: 0.1
  grad_clip: 1.0
  log_every: 50
  eval_every: 150
  eval_batches: 2
  ckpt_every: 150
em_max_items: 32