© 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. 14488
    Prev
    Next
    INDEX
    Explanations

    instructions for assembling or constructing items

    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
    25
    +0.21
    1.3%
    466
    +0.14
    0.9%
    325
    +0.14
    0.9%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    325
    +0.21
    0.09
    342
    +0.14
    0.05
    26
    +0.14
    0.08
    Negative Logits
     mood
    -1.35
     era
    -1.31
    '?"
    -1.30
     Breakfast
    -1.29
    stadt
    -1.28
    "?
    -1.25
    's
    -1.22
    ^](#
    -1.21
     NEWS
    -1.20
    anguages
    -1.19
    POSITIVE LOGITS
    hang
    2.10
     edges
    1.96
     corners
    1.95
     bottom
    1.94
     top
    1.90
    board
    1.89
     edge
    1.82
     adjacent
    1.82
     poles
    1.73
    rium
    1.70
    Activations Density 0.704%

    No Known Activations