| {"model_name": "gpt2-xl", "model_class_name": "HookedTransformer", "hook_point": "blocks.20.hook_resid_pre", "hook_point_eval": "blocks.{layer}.attn.pattern", "hook_point_layer": 20, "hook_point_head_index": null, "dataset_path": "Skylion007/openwebtext", "streaming": true, "is_dataset_tokenized": false, "context_size": 128, "use_cached_activations": false, "cached_activations_path": "activations/Skylion007_openwebtext/gpt2-small/blocks.1.hook_resid_pre", "d_in": 1600, "d_sae": 51200, "b_dec_init_method": "geometric_median", "expansion_factor": 32, "activation_fn": "relu", "normalize_sae_decoder": true, "noise_scale": 0.0, "from_pretrained_path": null, "apply_b_dec_to_input": true, "decoder_orthogonal_init": false, "decoder_heuristic_init": false, "init_encoder_as_decoder_transpose": false, "n_batches_in_buffer": 1, "training_tokens": 2000000, "finetuning_tokens": 0, "store_batch_size": 32, "train_batch_size": 4096, "normalize_activations": false, "device": "cpu", "seed": 42, "dtype": "torch.float16", "prepend_bos": true, "autocast": false, "compile_llm": false, "llm_compilation_mode": null, "compile_sae": false, "sae_compilation_mode": null, "adam_beta1": 0, "adam_beta2": 0.999, "mse_loss_normalization": null, "l1_coefficient": 8e-05, "lp_norm": 1, "scale_sparsity_penalty_by_decoder_norm": false, "l1_warm_up_steps": 0, "lr": 0.0004, "lr_scheduler_name": null, "lr_warm_up_steps": 5000, "lr_end": 4e-05, "lr_decay_steps": 0, "n_restart_cycles": 1, "finetuning_method": null, "use_ghost_grads": false, "feature_sampling_window": 1000, "dead_feature_window": 5000, "dead_feature_threshold": 1e-08, "n_eval_batches": 10, "n_eval_seqs": null, "log_to_wandb": true, "log_activations_store_to_wandb": false, "log_optimizer_state_to_wandb": false, "wandb_project": "mats_sae_training_gpt2_small_resid_pre_5", "wandb_id": null, "run_name": "24576-L1-8e-05-LR-0.0004-Tokens-3.000e+08", "wandb_entity": null, "wandb_log_frequency": 100, "eval_every_n_wandb_logs": 100, "resume": false, "n_checkpoints": 10, "checkpoint_path": "checkpoints/mm179kd2/qb58sohc", "verbose": true, "model_kwargs": {}, "sae_lens_version": "1.7.0", "sae_lens_training_version": "1.7.0", "tokens_per_buffer": 524288} |