| { | |
| "model_name": "1l-gelu", | |
| "model_class_name": "HookedTransformer", | |
| "hook_name": "blocks.0.hook_resid_post", | |
| "hook_eval": "NOT_IN_USE", | |
| "hook_layer": 0, | |
| "hook_head_index": null, | |
| "dataset_path": "ghidav/arithmetics_reversed", | |
| "dataset_trust_remote_code": true, | |
| "streaming": true, | |
| "is_dataset_tokenized": true, | |
| "context_size": 18, | |
| "use_cached_activations": false, | |
| "cached_activations_path": null, | |
| "architecture": "jumprelu", | |
| "d_in": 512, | |
| "d_sae": 8192, | |
| "b_dec_init_method": "zeros", | |
| "expansion_factor": 16, | |
| "activation_fn": "relu", | |
| "activation_fn_kwargs": {}, | |
| "normalize_sae_decoder": true, | |
| "noise_scale": 0.0, | |
| "from_pretrained_path": null, | |
| "apply_b_dec_to_input": false, | |
| "decoder_orthogonal_init": false, | |
| "decoder_heuristic_init": false, | |
| "init_encoder_as_decoder_transpose": false, | |
| "n_batches_in_buffer": 128, | |
| "training_tokens": 50000000, | |
| "finetuning_tokens": 0, | |
| "store_batch_size_prompts": 8, | |
| "train_batch_size_tokens": 1024, | |
| "normalize_activations": "none", | |
| "device": "cuda", | |
| "act_store_device": "cuda", | |
| "seed": 42, | |
| "dtype": "float32", | |
| "prepend_bos": false, | |
| "autocast": false, | |
| "autocast_lm": 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": 2.0, | |
| "lp_norm": 1, | |
| "scale_sparsity_penalty_by_decoder_norm": false, | |
| "l1_warm_up_steps": 2441, | |
| "lr": 0.0005, | |
| "lr_scheduler_name": "constant", | |
| "lr_warm_up_steps": 0, | |
| "lr_end": 5e-05, | |
| "lr_decay_steps": 9765, | |
| "n_restart_cycles": 1, | |
| "finetuning_method": null, | |
| "use_ghost_grads": false, | |
| "feature_sampling_window": 2000, | |
| "dead_feature_window": 1000, | |
| "dead_feature_threshold": 1e-06, | |
| "n_eval_batches": 10, | |
| "eval_batch_size_prompts": null, | |
| "log_to_wandb": true, | |
| "log_activations_store_to_wandb": false, | |
| "log_optimizer_state_to_wandb": false, | |
| "wandb_project": "sae-feature-circuits", | |
| "wandb_id": null, | |
| "run_name": "L0_hook_resid_post_L1_2_0_rev", | |
| "wandb_entity": null, | |
| "wandb_log_frequency": 30, | |
| "eval_every_n_wandb_logs": 100, | |
| "resume": false, | |
| "n_checkpoints": 0, | |
| "checkpoint_path": "checkpoints/220duyfn", | |
| "verbose": false, | |
| "model_kwargs": {}, | |
| "model_from_pretrained_kwargs": { | |
| "center_writing_weights": false | |
| }, | |
| "sae_lens_version": "3.20.5", | |
| "sae_lens_training_version": "3.20.5", | |
| "tokens_per_buffer": 2359296 | |
| } |