callummcdougall commited on
Commit
4b48abc
1 Parent(s): 7fb2626

Upload SAE gpt2-small-layer-5-mlp-transcoder-folded-b_dec_out

Browse files
gpt2-small-layer-5-mlp-transcoder-folded-b_dec_out/cfg.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"architecture": "standard", "d_in": 768, "d_sae": 24576, "dtype": "float32", "device": "cuda", "model_name": "gpt2-small", "hook_name": "blocks.5.ln2.hook_normalized", "hook_layer": 5, "hook_head_index": null, "activation_fn_str": "relu", "activation_fn_kwargs": {}, "apply_b_dec_to_input": true, "finetuning_scaling_factor": false, "sae_lens_training_version": null, "prepend_bos": true, "dataset_path": "Skylion007/openwebtext", "dataset_trust_remote_code": true, "context_size": 128, "normalize_activations": "none", "neuronpedia_id": null, "model_from_pretrained_kwargs": {}}
gpt2-small-layer-5-mlp-transcoder-folded-b_dec_out/sae_weights.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d9a03a3342682be658d3b29b77d1d1a267f358d6edb8a338ac28fcf941cd68dc
3
+ size 151096640