out_dir: results/raw data: kind: tokens dir: data/processed/fineweb_edu_300m splits: - val - test seq_len: 512 batch_size: 16 n_batches: 40 budgets: - 0 - 1 - 2 - 4 - 8 systems: backbone_103m: mode: backbone_only model: from_checkpoint checkpoint: experiments/runs/pretrain103m_seed01_20261029232706/checkpoint controller: false enlarged_102m: mode: backbone_only model: from_checkpoint checkpoint: experiments/runs/pretrain_enlarged102m_seed01_20261030015017/checkpoint controller: false prisme_103m_editor: mode: prisme model: from_checkpoint checkpoint: experiments/runs/editor_pilot_103m_lr3e4_seed01_20261030021200/checkpoint controller: false gr_103m: mode: global_recurrence baseline_kwargs: ffn_w: 2560 model: from_checkpoint checkpoint: experiments/runs/gr_pilot_103m_seed01_20261030015117/checkpoint controller: false