Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Apollo Research · Taylor · Identifying Functionally Important Features with End-to-End Sparse Dictionary Learning
    3. GPT2-Small
    4. Residual Similar CE Local
    5. 6-RES_SCL-AJT
    6. 40294
    Prev
    Next
    INDEX
    Explanations

    classified documents and sensitive information being shared or leaked

    oai_token-act-pair · gpt-3.5-turbo
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    neuronpedia/gpt2-small__res_scl-ajt/6-res_scl-ajt
    Prompts (Dashboard)
    12,288 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    46,080
    Data Type
    torch.float32
    Hook Point
    blocks.6.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    apollo-research/Skylion007-openwebtext-tokenizer-gpt2
    Hook Point Layer
    6
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    447
    +0.32
    1.5%
    218
    +0.10
    0.5%
    657
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    554
    +0.32
    -0.10
    673
    +0.10
    0.11
    561
    +0.10
    0.10
    Negative Logits
    Interstitial
    -0.94
    culosis
    -0.82
    stad
    -0.74
    ntil
    -0.69
    iasis
    -0.69
    Õ
    -0.68
    cers
    -0.64
    axis
    -0.63
    icidal
    -0.62
    adium
    -0.62
    POSITIVE LOGITS
     pertaining
    0.92
    heet
    0.88
     detailing
    0.87
     documents
    0.85
     relating
    0.83
     leaked
    0.82
     trove
    0.81
     declass
    0.78
    hig
    0.78
     dumps
    0.77
    Activations Density 11.394%

    No Known Activations