| {"version": 3, "model_name": "BlueSunflower/Pythia-70M-chess", "layers": [4], "setting": "residual", "tensor_name": "gpt_neox.layers.{layer}.mlp.dense_4h_to_h", "sparsity": 51, "num_epochs": 10, "kl": false, "reconstruction": false, "dataset_name": null, "device": "cuda:0", "seed": 0, "model_batch_size": 4, "buffer_mult": 100, "lr": 0.0003, "l1_coeff": 0.013572517793487848, "beta1": 0.9, "beta2": 0.999, "dict_mult": 4, "seq_len": 256, "d_model": 512, "enc_dtype": "fp32", "remove_rare_dir": false, "num_tokens": 100000000, "d_autoencoder": 2048} |