{"d_in": 2048, "dtype": "bfloat16", "expansion_factor": 64, "l1_coefficient": 8e-05, "train_batch_size": 2048, "dead_feature_window": 1000, "use_ghost_grads": true, "d_sae": 131072} |
{"d_in": 2048, "dtype": "bfloat16", "expansion_factor": 64, "l1_coefficient": 8e-05, "train_batch_size": 2048, "dead_feature_window": 1000, "use_ghost_grads": true, "d_sae": 131072} |