© 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. 1-RES-SM
    11. 24780
    Prev
    Next
    INDEX
    Explanations

    phrases indicating the potential or capability of an action or event

    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/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
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    69
    +0.23
    1.3%
    478
    +0.15
    0.9%
    411
    +0.11
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    69
    +0.23
    0.17
    151
    +0.15
    0.09
    411
    +0.11
    0.10
    Negative Logits
     asleep
    -1.55
    through
    -1.43
    alia
    -1.38
     from
    -1.37
    het
    -1.36
    liament
    -1.35
     based
    -1.33
    neys
    -1.32
    ogg
    -1.31
    ãĤī
    -1.31
    POSITIVE LOGITS
    ĥ½
    2.24
    pless
    1.94
     satisfy
    1.73
    ilers
    1.71
    iling
    1.71
     exclude
    1.70
     accommodate
    1.69
    ·¸
    1.67
     reject
    1.65
     grasp
    1.65
    Activations Density 1.017%

    No Known Activations