| revision = "main" | |
| [data] | |
| path = "data/french_rap_lyrics" | |
| [corpus] | |
| encoding = "r50k_base" | |
| split_train_val = 0.85 | |
| [model] | |
| context_length = 256 | |
| num_heads = 8 | |
| num_layers = 18 | |
| hidden_dim = 424 | |
| [training] | |
| num_training_steps = 20000 | |
| seed = 42 | |
| lr = 1e-3 | |
| batch_size = 64 | |
| [scheduler] | |
| gamma = 0.925 | |
| [evaluation] | |
| sample_text = "Les vrais savent que " | |
| new_tokens = 50 | |
| artist_token = 1 | |
| [wandb] | |
| mode = "online" # offline or online or disabled | |
| group = "fullrun" | |
| tags = [] | |