| name: rectification | |
| backbone: dit # dit / dimamba / hf_dit | |
| parameterization: mean | |
| time_conditioning: True | |
| T: 0 # 0 (continuous time) / 1000 | |
| subs_masking: False | |
| causal_attention: False | |
| gumbel_tau_log10_start: -1.0 | |
| gumbel_tau_log10_end: -2.0 | |
| curriculum_start: 0 | |
| curriculum_end: 1 | |
| integral_cache_path: ${hydra:runtime.cwd}/integral/${data.tokenizer_name_or_path}.pkl | |
| loss_type: elbo | |
| ignore_bos: False | |
| gamma_min: -3.5 | |
| gamma_max: -1.75 | |
| use_linear_schedule: False | |
| use_simple_loss: False | |
| onestep_mode: False |