© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    5. Pythia-70M-Deduped
    7. Residuals
    9. 2-RES-SM
    11. 29646
    Prev
    Next
    INDEX
    Explanations

    references to child exploitation and abuse

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    ctigges/pythia-70m-deduped__res-sm_processed/2-res-sm
    Prompts (Dashboard)
    32,768 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    32,768
    Data Type
    torch.float32
    Hook Name
    blocks.2.hook_resid_post
    Hook Layer
    2
    Architecture
    standard
    Context Size
    128
    Dataset
    EleutherAI/the_pile_deduplicated
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    87
    +0.13
    0.7%
    359
    +0.12
    0.7%
    132
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    18
    +0.13
    0.02
    13
    +0.12
    0.01
    339
    +0.12
    0.01
    Negative Logits
    hin
    -1.57
    HECK
    -1.53
     entirety
    -1.47
     UK
    -1.43
    Compat
    -1.39
    ogether
    -1.39
     forthcoming
    -1.39
    ../../
    -1.37
    PHP
    -1.36
    heed
    -1.35
    POSITIVE LOGITS
    ytes
    1.75
    liography
    1.68
    s
    1.65
    dle
    1.65
    lic
    1.60
    th
    1.59
    ilic
    1.59
    dling
    1.51
    ortune
    1.51
    obic
    1.50
    Activations Density 0.087%

    No Known Activations