cfg.json

{"model_name": "wkcn/TinyCLIP-ViT-40M-32-Text-19M-LAION400M", "model_class_name": "HookedTransformer", "hook_point": "blocks.2.hook_mlp_out", "hook_point_eval": "blocks.{layer}.attn.pattern", "hook_point_layer": 2, "hook_point_head_index": null, "dataset_path": "NeelNanda/c4-tokenized-2b", "streaming": true, "is_dataset_tokenized": true, "context_size": 50, "use_cached_activations": false, "cached_activations_path": null, "d_in": 512, "d_sae": 8192, "b_dec_init_method": "geometric_median", "expansion_factor": 16, "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": 32, "training_tokens": 195000000, "finetuning_tokens": 0, "store_batch_size_prompts": 32, "train_batch_size_tokens": 4096, "normalize_activations": false, "device": "cuda", "seed": 42, "dtype": "torch.float32", "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": "constant", "lr_warm_up_steps": 5000, "lr_end": 4e-05, "lr_decay_steps": 0, "n_restart_cycles": 1, "finetuning_method": null, "use_ghost_grads": true, "feature_sampling_window": 1000, "dead_feature_window": 5000, "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": "vit_sae_training", "wandb_id": null, "run_name": "QZaT_wkcn/TinyCLIP-ViT-40M-32-Text-19M-LAION400M_blocks.{layer}.hook_mlp_out_2", "wandb_entity": null, "wandb_log_frequency": 100, "eval_every_n_wandb_logs": 10, "resume": false, "n_checkpoints": 10, "checkpoint_path": "/network/scratch/s/sonia.joseph/saes/QZaT_wkcn/TinyCLIP-ViT-40M-32-Text-19M-LAION400M_blocks.{layer}.hook_mlp_out_2/mjt69mp0/j3jj2z2z/4hdvpai4", "verbose": true, "model_kwargs": {}, "model_from_pretrained_kwargs": {}, "sae_lens_version": "2.1.3", "sae_lens_training_version": "2.1.3", "store_batch_size": 32, "tokens_per_buffer": 6553600}