© 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. 4-RES-SM
    11. 14840
    Prev
    Next
    INDEX
    Explanations

    the word "which" 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/4-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.4.hook_resid_post
    Hook Layer
    4
    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
    307
    +0.35
    2.1%
    306
    +0.18
    1.1%
    466
    +0.11
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    307
    +0.35
    0.06
    506
    +0.18
    0.02
    174
    +0.11
    0.04
    Negative Logits
    nikov
    -1.68
    illa
    -1.50
    cias
    -1.45
     huh
    -1.41
    lined
    -1.41
     cui
    -1.38
    ietz
    -1.35
    arium
    -1.33
    Äį
    -1.32
    der
    -1.32
    POSITIVE LOGITS
    ĻĤ
    2.87
    ĥ½
    2.46
     they
    2.35
    Ļ
    2.15
     we
    2.13
    Ĭ
    2.10
    upon
    2.07
    ķ
    2.04
    ¤
    2.04
     there
    2.04
    Activations Density 0.336%

    No Known Activations