© 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 EvalsExportsinterp-engineNEWAPI 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. 14350
    Prev
    Next
    INDEX
    Explanations

    references to coding or programming concepts

    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/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
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    78
    +0.21
    1.2%
    397
    +0.15
    0.9%
    9
    +0.14
    0.8%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    397
    +0.21
    0.11
    9
    +0.15
    0.10
    78
    +0.14
    0.02
    Negative Logits
     Plaintiffs
    -1.43
     journal
    -1.36
     toward
    -1.33
    izing
    -1.24
    ^\[[@
    -1.20
    journal
    -1.20
     regards
    -1.20
    jac
    -1.18
     regard
    -1.18
    aries
    -1.17
    POSITIVE LOGITS
     '
    3.88
    .'
    3.77
    ?'
    3.58
    '
    3.54
    ,'
    3.54
    .'"
    3.40
    '.
    3.38
    ',
    3.37
    '."
    3.35
    ,'"
    3.32
    Activations Density 1.372%

    No Known Activations