File size: 2,188 Bytes
c50b27f |
1 |
{"model_name": "gemma-2b", "model_class_name": "HookedTransformer", "hook_point": "blocks.12.hook_resid_post", "hook_point_eval": "blocks.{layer}.attn.pattern", "hook_point_layer": 12, "hook_point_head_index": null, "dataset_path": "HuggingFaceFW/fineweb", "streaming": true, "is_dataset_tokenized": false, "context_size": 1024, "use_cached_activations": false, "cached_activations_path": null, "d_in": 2048, "d_sae": 16384, "b_dec_init_method": "zeros", "expansion_factor": 8, "activation_fn": "relu", "normalize_sae_decoder": false, "noise_scale": 0.0, "from_pretrained_path": null, "apply_b_dec_to_input": false, "decoder_orthogonal_init": false, "decoder_heuristic_init": true, "init_encoder_as_decoder_transpose": true, "n_batches_in_buffer": 16, "training_tokens": 1228800000, "finetuning_tokens": 0, "store_batch_size_prompts": 8, "train_batch_size_tokens": 4096, "normalize_activations": true, "device": "cuda", "seed": 42, "dtype": "torch.float32", "prepend_bos": true, "autocast": true, "compile_llm": true, "llm_compilation_mode": "max-autotune", "compile_sae": false, "sae_compilation_mode": null, "adam_beta1": 0.9, "adam_beta2": 0.999, "mse_loss_normalization": null, "l1_coefficient": 2, "lp_norm": 1.0, "scale_sparsity_penalty_by_decoder_norm": true, "l1_warm_up_steps": 15000, "lr": 5e-05, "lr_scheduler_name": "constant", "lr_warm_up_steps": 0, "lr_end": 5e-06, "lr_decay_steps": 0, "n_restart_cycles": 1, "finetuning_method": null, "use_ghost_grads": false, "feature_sampling_window": 5000, "dead_feature_window": 5000, "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": "gemma_2b_exploration", "wandb_id": null, "run_name": "16384-L1-2-LR-5e-05-Tokens-1.229e+09", "wandb_entity": null, "wandb_log_frequency": 50, "eval_every_n_wandb_logs": 10, "resume": false, "n_checkpoints": 5, "checkpoint_path": "checkpoints/0bzfiqi2/konwqovz/g2ydv7jo", "verbose": true, "model_kwargs": {}, "model_from_pretrained_kwargs": {}, "sae_lens_version": "2.1.3", "sae_lens_training_version": "2.1.3", "tokens_per_buffer": 67108864} |