| { |
| "auxiliary_active_at_checkpoint": false, |
| "auxiliary_coefficient_at_checkpoint": 0.0, |
| "calibration_minibatch": 0, |
| "checkpoint_words": 30000000, |
| "condition": "shared", |
| "factor_scales": { |
| "associative": 0.5272428393363953, |
| "hierarchical": 9.905651092529297, |
| "order": 0.9976174831390381 |
| }, |
| "initial_aux_gradient_ratio": 0.25, |
| "input_output_embeddings_tied": false, |
| "lambda0": 0.14438274502754211, |
| "projectors": { |
| "concatenated_condition_number": 1.7477143918426399, |
| "cross_block_overlap_frobenius": [ |
| 2.4472843287594483, |
| 2.4276792538495555, |
| 2.4482626734373665 |
| ], |
| "mode": "shared", |
| "ranks": [ |
| 128, |
| 128, |
| 128 |
| ], |
| "within_block_orthonormal_error": [ |
| 4.8733968194270005e-08, |
| 4.908529305236442e-08, |
| 4.901802364010166e-08 |
| ] |
| }, |
| "schema_version": 1, |
| "variant": "dial_shared_s5", |
| "withdrawal_words": 10000000 |
| } |
|
|