© 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. 26080
    Prev
    Next
    INDEX
    Explanations

    references to user interface elements and navigation options

    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
    156
    +0.24
    1.4%
    184
    +0.15
    0.9%
    255
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    184
    +0.24
    0.02
    17
    +0.15
    0.00
    355
    +0.11
    0.02
    Negative Logits
    edes
    -1.70
    onin
    -1.64
    googleapis
    -1.55
    bestos
    -1.45
    monary
    -1.44
     killing
    -1.44
     beating
    -1.41
     mortality
    -1.38
    copyright
    -1.38
    yg
    -1.36
    POSITIVE LOGITS
    ister
    1.61
    erville
    1.61
     choices
    1.58
     drinks
    1.54
     options
    1.54
    oon
    1.54
    VIEW
    1.52
     Navigation
    1.52
    view
    1.51
     drawer
    1.50
    Activations Density 0.176%

    No Known Activations