© 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. 13109
    Prev
    Next
    INDEX
    Explanations

    instances of the word "what" and related interrogative expressions

    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
    198
    +0.13
    0.8%
    302
    +0.12
    0.7%
    312
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    243
    +0.13
    0.05
    302
    +0.12
    0.04
    246
    +0.11
    0.05
    Negative Logits
    idades
    -1.45
     listing
    -1.44
    aciones
    -1.38
    emente
    -1.38
     reservations
    -1.36
    index
    -1.35
     "%
    -1.34
     boycott
    -1.30
     speaking
    -1.30
     vote
    -1.29
    POSITIVE LOGITS
     ours
    1.73
    llll
    1.63
     glue
    1.52
    onic
    1.48
    ––––––––
    1.47
    anding
    1.45
     yours
    1.42
    cms
    1.39
     WARRANTY
    1.39
    yyyy
    1.38
    Activations Density 0.462%

    No Known Activations