| { | |
| "D_head": 3, | |
| "D_out": 3, | |
| "R_iX_shape": [ | |
| 2000, | |
| 3 | |
| ], | |
| "R_ix_external_shape": [ | |
| 2000, | |
| 1, | |
| 32 | |
| ], | |
| "d": 32, | |
| "eps": 1e-12, | |
| "h": 1, | |
| "metric_rank": null, | |
| "seed": 0, | |
| "sigma2": 1e-06, | |
| "use_W_O": false, | |
| "\u03b2": 3.298452854156494 | |
| } |
| { | |
| "D_head": 3, | |
| "D_out": 3, | |
| "R_iX_shape": [ | |
| 2000, | |
| 3 | |
| ], | |
| "R_ix_external_shape": [ | |
| 2000, | |
| 1, | |
| 32 | |
| ], | |
| "d": 32, | |
| "eps": 1e-12, | |
| "h": 1, | |
| "metric_rank": null, | |
| "seed": 0, | |
| "sigma2": 1e-06, | |
| "use_W_O": false, | |
| "\u03b2": 3.298452854156494 | |
| } |