| eval: | |
| eval_datasets: null | |
| json_datasets: null | |
| test_datasets: | |
| - name: generator_mean | |
| length: 96 | |
| generator: PATTERN | |
| num_pairs: 4 | |
| batch_size: 96 | |
| num_tasks_to_show: 32 | |
| task_generator_kwargs: | |
| num_cols: 4 | |
| num_rows: 4 | |
| pattern_size: 2 | |
| - name: generator_gradient_ascent_5 | |
| length: 96 | |
| generator: PATTERN | |
| num_pairs: 4 | |
| batch_size: 96 | |
| inference_mode: gradient_ascent | |
| inference_kwargs: | |
| lr: 0.1 | |
| num_steps: 10 | |
| num_tasks_to_show: 32 | |
| task_generator_kwargs: | |
| num_cols: 4 | |
| num_rows: 4 | |
| pattern_size: 2 | |
| training: | |
| seed: 0 | |
| use_hf: true | |
| kl_coeff: 0.001 | |
| batch_size: 128 | |
| learning_rate: 0.0004 | |
| inference_mode: mean | |
| task_generator: | |
| class: PATTERN | |
| num_cols: 4 | |
| num_rows: 4 | |
| num_pairs: 4 | |
| num_workers: 16 | |
| pattern_size: 2 | |
| train_datasets: null | |
| mixed_precision: false | |
| total_num_steps: 200000 | |
| inference_kwargs: null | |
| eval_every_n_logs: 20 | |
| log_every_n_steps: 1000 | |
| resume_from_checkpoint: null | |
| online_data_augmentation: false | |
| gradient_accumulation_steps: 1 | |
| save_checkpoint_every_n_logs: 200 | |
| decoder_transformer: | |
| _target_: src_v2.models.utils.DecoderTransformerConfig | |
| max_cols: 4 | |
| max_rows: 4 | |
| num_layers: 2 | |
| transformer_layer: | |
| _target_: src_v2.models.utils.TransformerLayerConfig | |
| num_heads: 6 | |
| dropout_rate: 0.0 | |
| mlp_dim_factor: 4.0 | |
| emb_dim_per_head: 12 | |
| attention_dropout_rate: 0.0 | |
| encoder_transformer: | |
| _target_: src_v2.models.utils.EncoderTransformerConfig | |
| max_cols: 4 | |
| max_rows: 4 | |
| latent_dim: 2 | |
| num_layers: 2 | |
| variational: true | |
| transformer_layer: | |
| _target_: src_v2.models.utils.TransformerLayerConfig | |
| num_heads: 6 | |
| dropout_rate: 0.0 | |
| mlp_dim_factor: 4.0 | |
| emb_dim_per_head: 12 | |
| attention_dropout_rate: 0.0 | |
| latent_projection_bias: false | |