data: name: PerturbationDataModule kwargs: toml_config_path: '' embed_key: X_state output_space: all pert_rep: onehot basal_rep: sample num_workers: 0 pin_memory: true n_basal_samples: 1 basal_mapping_strategy: random should_yield_control_cells: true batch_col: gem_group pert_col: gene cell_type_key: cell_line control_pert: non-targeting map_controls: true perturbation_features_file: null store_raw_basal: false int_counts: false barcode: true output_dir: null debug: true model: name: state checkpoint: null device: cuda kwargs: cell_set_len: 512 blur: 0.05 hidden_dim: 696 loss: energy confidence_head: false n_encoder_layers: 4 n_decoder_layers: 4 predict_residual: true softplus: true freeze_pert: false transformer_decoder: false finetune_vci_decoder: false residual_decoder: false batch_encoder: false nb_decoder: false mask_attn: false use_effect_gating_token: false distributional_loss: energy init_from: null transformer_backbone_key: llama transformer_backbone_kwargs: max_position_embeddings: 512 hidden_size: 696 intermediate_size: 2784 num_hidden_layers: 8 num_attention_heads: 12 num_key_value_heads: 12 head_dim: 58 use_cache: false attention_dropout: 0.0 hidden_dropout: 0.0 layer_norm_eps: 1.0e-06 pad_token_id: 0 bos_token_id: 1 eos_token_id: 2 tie_word_embeddings: false rotary_dim: 0 use_rotary_embeddings: false training: wandb_track: false weight_decay: 0.0005 batch_size: 16 lr: 0.0001 max_steps: 1000 train_seed: 42 val_freq: 2000 ckpt_every_n_steps: 2000 gradient_clip_val: 10 loss_fn: mse devices: 1 strategy: auto wandb: entity: your_entity_name project: state local_wandb_dir: ./wandb_logs tags: [] name: _nw0 output_dir: '' use_wandb: true overwrite: false return_adatas: false pred_adata_path: null true_adata_path: null