Upload SAE blocks.6.hook_resid_post/l0-10

blocks.6.hook_resid_post/l0-10/cfg.json

@@ -0,0 +1 @@
+{"architecture": "topk", "d_in": 2048, "d_sae": 16384, "dtype": "float32", "device": "cuda", "model_name": "Llama-3.2-1B", "hook_name": "blocks.6.hook_resid_post", "hook_layer": 6, "hook_head_index": null, "activation_fn_str": "topk", "activation_fn_kwargs": {"k": 10}, "apply_b_dec_to_input": true, "finetuning_scaling_factor": false, "sae_lens_training_version": "5.5.1", "prepend_bos": true, "dataset_path": "HuggingFaceFW/fineweb", "dataset_trust_remote_code": true, "context_size": 1024, "normalize_activations": "none", "neuronpedia_id": null, "model_from_pretrained_kwargs": {"center_writing_weights": false}, "seqpos_slice": [null], "l1_coefficient": 0.001, "aux_coefficient": 0.03125, "lp_norm": 1.0, "use_ghost_grads": false, "normalize_sae_decoder": false, "noise_scale": 0.0, "decoder_orthogonal_init": false, "init_encoder_as_decoder_transpose": true, "mse_loss_normalization": null, "decoder_heuristic_init": true, "scale_sparsity_penalty_by_decoder_norm": false, "jumprelu_init_threshold": 0.001, "jumprelu_bandwidth": 0.001, "disable_enc_bias": false, "reconstruction_loss": "MSE"}

blocks.6.hook_resid_post/l0-10/sae_weights.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6970c5ef20fb1853498c1116a4a2e255fe092f5a1e892f7cbd5c60209d6c6405
+size 268509504

blocks.6.hook_resid_post/l0-10/sparsity.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7987075a737996e3c6c516399e164484782ca3f7d60c540896629807ddaefa45
+size 65616