© 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. 1-RES-SM
    11. 24608
    Prev
    Next
    INDEX
    Explanations

    the presence of the substring "th" in words or characters

    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/1-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.1.hook_resid_post
    Hook Layer
    1
    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
    404
    +0.19
    1.1%
    503
    +0.11
    0.6%
    314
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    404
    +0.19
    0.04
    101
    +0.11
    0.02
    239
    +0.11
    0.02
    Negative Logits
    ¥
    -1.90
    ¦
    -1.80
     sexes
    -1.61
     conflicting
    -1.61
    °
    -1.59
    ¬
    -1.50
    ½
    -1.50
    ABILITY
    -1.46
    Ĥ
    -1.39
    ħ
    -1.37
    POSITIVE LOGITS
    neys
    2.01
    ousand
    1.99
    ailand
    1.88
    êm
    1.87
    ousands
    1.87
    umbs
    1.84
    orne
    1.83
    ieves
    1.82
    ritis
    1.80
    ongs
    1.77
    Activations Density 0.064%

    No Known Activations