© 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. 31112
    Prev
    Next
    INDEX
    Explanations

    references to coding or software-related components, particularly involving Git and GitHub

    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
    231
    +0.16
    0.9%
    17
    +0.12
    0.6%
    263
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    231
    +0.16
    0.14
    172
    +0.12
    0.08
    489
    +0.11
    0.02
    Negative Logits
     purpose
    -1.61
    err
    -1.60
    injury
    -1.60
    erton
    -1.54
    atism
    -1.53
    že
    -1.53
    ÅŁ
    -1.51
    cknowled
    -1.45
    ôle
    -1.45
     meaning
    -1.42
    POSITIVE LOGITS
    ĻĤ
    2.75
    ¿½
    2.67
    ĭ
    2.57
    ĨĴ
    2.54
    ¹
    2.54
    Ĵ
    2.35
    Īĺ
    2.32
    Ģ
    2.30
    ±
    2.29
    ª
    2.27
    Activations Density 5.231%

    No Known Activations