© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    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. 30079
    Prev
    Next
    INDEX
    Explanations

    repeated references to the concept of "part."

    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
    500
    +0.14
    0.7%
    289
    +0.12
    0.7%
    502
    +0.12
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    500
    +0.14
    0.03
    20
    +0.12
    0.03
    502
    +0.12
    0.03
    Negative Logits
    ETHOD
    -1.71
    ĻĤ
    -1.60
    keywords
    -1.46
    "){
    -1.43
    ibles
    -1.38
    $)
    -1.37
    ASSERT
    -1.37
     slightest
    -1.37
    www
    -1.36
    biamo
    -1.35
    POSITIVE LOGITS
    isans
    2.45
     of
    2.29
    ridges
    2.25
     thereof
    2.03
    ridge
    1.92
     Of
    1.89
    aking
    1.69
    essential
    1.52
    enberg
    1.52
    olini
    1.49
    Activations Density 0.076%

    No Known Activations