| { |
| "dropout": 0.1, |
| "hidden_act": "silu", |
| "hidden_size": [ |
| 384, |
| 768, |
| 768, |
| 768, |
| 384 |
| ], |
| "initializer_range": 0.02, |
| "input_dim": 1, |
| "intermediate_size": [ |
| 1536, |
| 3072, |
| 3072, |
| 3072, |
| 1536 |
| ], |
| "model_type": "bert", |
| "n_heads": [ |
| 4, |
| 4, |
| 8, |
| 8, |
| 8, |
| 4, |
| 4 |
| ], |
| "num_heads": [ |
| 6, |
| 12, |
| 12, |
| 12, |
| 6 |
| ], |
| "num_latent_tokens": 2, |
| "num_layers": [ |
| 1, |
| 3, |
| 3, |
| 3, |
| 2 |
| ], |
| "num_reg_tokens": 1, |
| "quantiles": [ |
| 0.1, |
| 0.2, |
| 0.3, |
| 0.4, |
| 0.5, |
| 0.6, |
| 0.7, |
| 0.8, |
| 0.9 |
| ], |
| "scales": [ |
| 1, |
| 8, |
| 32, |
| 8, |
| 1 |
| ], |
| "transformers_version": "4.57.3", |
| "use_latent_tokens": false |
| } |
|
|