Upload gsaefinetune_gemma-2b-it_layer12_college_bio_cleaned_l1=2_expansion=16_tokens=8192000_cfg.json with huggingface_hub

gsaefinetune_gemma-2b-it_layer12_college_bio_cleaned_l1=2_expansion=16_tokens=8192000_cfg.json

@@ -0,0 +1 @@
+{"first_activation_pos": 0, "gsae_repo": null, "gsae_filename": null, "gsae_cfg_filename": null, "gsae_release": null, "gsae_id": null, "control_dataset_path": null, "is_control_dataset_tokenized": false, "control_mixture": 0, "save_final_checkpoint_locally": true, "model_name": "gemma-2b-it", "model_class_name": "HookedTransformer", "hook_name": "blocks.12.hook_resid_pre", "hook_eval": "NOT_IN_USE", "hook_layer": 12, "hook_head_index": null, "dataset_path": "jacobcd52/college_bio_cleaned", "dataset_trust_remote_code": true, "streaming": true, "is_dataset_tokenized": false, "context_size": 128, "use_cached_activations": false, "cached_activations_path": null, "architecture": "gated", "d_in": 2048, "d_sae": 32768, "b_dec_init_method": "mean", "activation_fn": "relu", "activation_fn_kwargs": {}, "expansion_factor": 16, "normalize_sae_decoder": false, "noise_scale": 0.0, "from_pretrained_path": "/root/specialised-SAEs/sae_lens/jacob/temp_sae", "apply_b_dec_to_input": true, "decoder_orthogonal_init": false, "decoder_heuristic_init": true, "init_encoder_as_decoder_transpose": true, "n_batches_in_buffer": 64, "training_tokens": 8192000, "finetuning_tokens": 0, "store_batch_size_prompts": 32, "train_batch_size_tokens": 4096, "normalize_activations": "none", "device": "cuda", "act_store_device": "cuda", "seed": 42, "dtype": "bfloat16", "prepend_bos": true, "autocast": false, "autocast_lm": false, "compile_llm": false, "llm_compilation_mode": null, "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": false, "l1_warm_up_steps": 0, "lr": 0.0001, "lr_scheduler_name": "constant", "lr_warm_up_steps": 0, "lr_end": 1e-05, "lr_decay_steps": 400, "n_restart_cycles": 1, "finetuning_method": null, "use_ghost_grads": false, "feature_sampling_window": 1000, "dead_feature_window": 100000, "dead_feature_threshold": 1e-08, "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-it-layer12-gsae-ft", "wandb_id": null, "run_name": "gsaefinetune_gemma-2b-it_layer12_college_bio_cleaned_l1=2_expansion=16_tokens=8192000", "wandb_entity": null, "wandb_log_frequency": 20, "eval_every_n_wandb_logs": 20, "resume": false, "n_checkpoints": 0, "checkpoint_path": "gsaefinetune_gemma-2b-it_layer12_college_bio_cleaned_l1=2_expansion=16_tokens=8192000/1gws7650", "verbose": true, "model_kwargs": {}, "model_from_pretrained_kwargs": {"dtype": "bfloat16"}, "sae_lens_version": "3.14.0", "sae_lens_training_version": "3.14.0", "tokens_per_buffer": 33554432}