© 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. 2-RES-SM
    11. 31508
    Prev
    Next
    INDEX
    Explanations

    instances of the word "everything."

    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
    365
    +0.20
    1.1%
    246
    +0.12
    0.7%
    356
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    165
    +0.20
    0.03
    365
    +0.12
    0.03
    246
    +0.12
    0.03
    Negative Logits
    Ŀ
    -2.95
    ĥ½
    -2.79
    ¿
    -2.67
    µ
    -2.65
    ŀ
    -2.64
    Ĺ
    -2.60
    ľ
    -2.54
    Ĵ
    -2.50
    ®
    -2.48
    ¹
    -2.48
    POSITIVE LOGITS
     else
    2.83
     ELSE
    2.09
     conceivable
    1.77
     Else
    1.76
     except
    1.70
     happens
    1.68
     contained
    1.68
     happening
    1.65
     stored
    1.59
     happen
    1.58
    Activations Density 0.066%

    No Known Activations