© 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. 2-RES-SM
    11. 32365
    Prev
    Next
    INDEX
    Explanations

    questions starting with "Does" or "does"

    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/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
    Neuron Alignment
    Index
    Value
    % of L₁
    36
    +0.13
    0.7%
    219
    +0.12
    0.7%
    295
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    295
    +0.13
    0.04
    2
    +0.12
    0.04
    219
    +0.11
    0.04
    Negative Logits
     whom
    -2.00
    ais
    -1.97
    ubert
    -1.92
     suspects
    -1.91
    atted
    -1.82
    aja
    -1.81
    ungs
    -1.76
    atics
    -1.75
    places
    -1.66
     learns
    -1.65
    POSITIVE LOGITS
    typeof
    1.77
    oming
    1.65
     hereby
    1.58
    ocyanate
    1.50
    chitz
    1.48
    't
    1.46
     seem
    1.46
    OOGLE
    1.44
     entail
    1.43
    change
    1.41
    Activations Density 0.095%

    No Known Activations