© 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. 10768
    Prev
    Next
    INDEX
    Explanations

    mentions of the word "pan" 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/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
    156
    +0.17
    1.0%
    59
    +0.14
    0.8%
    369
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    386
    +0.17
    0.01
    59
    +0.14
    0.01
    391
    +0.12
    0.01
    Negative Logits
    ĥ½
    -2.59
    ¥
    -2.44
    Ļ
    -2.33
    ł
    -2.28
    ¢
    -2.27
    ļ
    -2.27
    Ĥ¬
    -2.16
    ¡
    -2.10
    ij
    -2.04
    Ķ
    -1.98
    POSITIVE LOGITS
    ason
    1.96
    icked
    1.94
    acea
    1.81
    aman
    1.77
    creas
    1.73
    iero
    1.71
    icking
    1.67
    icky
    1.67
    tera
    1.66
    icks
    1.63
    Activations Density 0.102%

    No Known Activations