© 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 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. 10030
    Prev
    Next
    INDEX
    Explanations

    instances of the word "some" in various contexts

    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
    10
    +0.13
    0.7%
    274
    +0.12
    0.7%
    247
    +0.12
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    409
    +0.13
    0.06
    10
    +0.12
    0.05
    247
    +0.12
    0.06
    Negative Logits
    roll
    -1.69
    roc
    -1.63
    eto
    -1.60
    over
    -1.59
    ED
    -1.55
    eler
    -1.54
    ret
    -1.50
    rait
    -1.47
    rolling
    -1.41
     Einstein
    -1.41
    POSITIVE LOGITS
    what
    2.03
     sort
    1.78
    where
    1.76
    place
    1.73
     kind
    1.69
    how
    1.66
     things
    1.64
     nasty
    1.60
     nights
    1.57
     caution
    1.54
    Activations Density 0.165%

    No Known Activations