© 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. 1-RES-SM
    11. 29955
    Prev
    Next
    INDEX
    Explanations

    instances of quotation marks in the text

    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/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
    Neuron Alignment
    Index
    Value
    % of L₁
    478
    +0.21
    1.2%
    421
    +0.11
    0.6%
    369
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    421
    +0.21
    0.02
    147
    +0.11
    0.02
    165
    +0.11
    0.01
    Negative Logits
    izing
    -1.71
    '">
    -1.62
    \]](
    -1.59
    heses
    -1.54
    ization
    -1.50
    pection
    -1.47
    ened
    -1.45
    itation
    -1.45
     erupted
    -1.44
     strikes
    -1.41
    POSITIVE LOGITS
     yourselves
    2.07
     harmless
    1.66
     gay
    1.36
    ascii
    1.35
     bac
    1.34
     ourselves
    1.33
     certain
    1.33
     yer
    1.32
     Gent
    1.31
     our
    1.31
    Activations Density 0.083%

    No Known Activations