© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    5. Pythia-70M-Deduped
    7. Residuals
    9. 4-RES-SM
    11. 15174
    Prev
    Next
    INDEX
    Explanations

    mentions of illegal actions or situations

    oai_token-act-pair · gpt-4o-mini
    Triggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    ctigges/pythia-70m-deduped__res-sm_processed/4-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.4.hook_resid_post
    Hook Layer
    4
    Architecture
    standard
    Context Size
    128
    Dataset
    EleutherAI/the_pile_deduplicated
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Neuron Alignment
    Index
    Value
    % of L₁
    136
    +0.14
    0.8%
    156
    +0.11
    0.6%
    274
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    386
    +0.14
    0.01
    120
    +0.11
    0.03
    350
    +0.11
    0.02
    Negative Logits
    e
    -1.61
    eer
    -1.48
    gmail
    -1.42
     validity
    -1.38
    ´
    -1.37
    plit
    -1.37
    equation
    -1.32
     description
    -1.32
    etus
    -1.31
    www
    -1.31
    POSITIVE LOGITS
    legal
    2.25
    irubin
    2.24
    aterally
    2.22
    umines
    1.97
    abs
    1.96
    gium
    1.91
    iterate
    1.87
    itary
    1.86
    inx
    1.86
    ibrary
    1.79
    Activations Density 0.113%

    No Known Activations