© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 30259
    Prev
    Next
    INDEX
    Explanations

    prepositions indicating motion or direction

    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/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
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    478
    +0.18
    1.0%
    69
    +0.16
    0.9%
    282
    +0.12
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    69
    +0.18
    0.06
    282
    +0.16
    0.05
    194
    +0.12
    0.04
    Negative Logits
    arily
    -1.99
    itage
    -1.68
    liness
    -1.67
    ishly
    -1.60
    documentclass
    -1.55
    tery
    -1.54
     signs
    -1.50
    teen
    -1.49
    irst
    -1.49
    ter
    -1.46
    POSITIVE LOGITS
    verte
    1.76
    eligible
    1.70
    ordinate
    1.66
    activate
    1.61
    elastic
    1.61
     remission
    1.60
    verts
    1.56
    active
    1.55
    complet
    1.54
    jected
    1.53
    Activations Density 0.140%

    No Known Activations