| {"seed": 49, "batch_size": 4096, "buffer_mult": 128, "lr": 5e-05, "num_tokens": 320000000, "l1_coeff": 2, "beta1": 0.9, "beta2": 0.999, "d_in": 2048, "dict_size": 16384, "seq_len": 2048, "enc_dtype": "fp32", "model_name": "pythia1.3b30v90l18", "site": "resid_pre", "device": "cuda:0", "model_batch_size": 4, "log_every": 100, "save_every": 30000, "dec_init_norm": 0.08, "hook_point": "blocks.18.hook_resid_pre", "wandb_project": "pythia crosscoder 1.3b training", "wandb_entity": "xiaoxiaoanddali-imperial-college-london"} |