| {"seed": 49, "batch_size": 256, "buffer_mult": 16, "lr": 0.0001, "num_tokens": 200000, "l1_coeff": 0.0003, "beta1": 0.9, "beta2": 0.99, "dict_mult": 8, "seq_len": 128, "d_mlp": 2048, "enc_dtype": "fp32", "remove_rare_dir": false, "site": "mlp_out", "layer": 0, "model_name": "gelu-1l", "device": "cuda:0", "model_batch_size": 32, "buffer_size": 4096, "buffer_batches": 32, "act_name": "blocks.0.hook_mlp_out", "act_size": 2048, "dict_size": 16384, "name": "gelu-1l_0_16384_mlp_out"} |