© 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. 32059
    Prev
    Next
    INDEX
    Explanations

    instances of the word "identify" and its variations in the context of recognition and categorization

    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
    470
    +0.13
    0.7%
    277
    +0.11
    0.6%
    148
    +0.10
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    148
    +0.13
    0.01
    277
    +0.11
    0.02
    470
    +0.10
    0.03
    Negative Logits
    ´
    -1.66
    onz
    -1.54
    ainen
    -1.53
    lett
    -1.44
    olver
    -1.43
    ĭ
    -1.40
    oj
    -1.35
    typeof
    -1.35
    lev
    -1.35
    poons
    -1.34
    POSITIVE LOGITS
    ably
    2.11
     landmarks
    1.89
     gaps
    1.89
     barriers
    1.75
     mistake
    1.68
     errors
    1.66
     obstacles
    1.57
     clues
    1.56
     duplicates
    1.54
    etect
    1.53
    Activations Density 0.155%

    No Known Activations