© 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. 4-RES-SM
    11. 13822
    Prev
    Next
    INDEX
    Explanations

    questions or requests beginning with "Can" or "Could."

    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/4-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.4.hook_resid_post
    Hook Layer
    4
    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
    +0.15
    0.8%
    63
    +0.12
    0.7%
    471
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    81
    +0.15
    0.01
    63
    +0.12
    0.01
    48
    +0.12
    0.02
    Negative Logits
    ĥ
    -2.42
    Ķ
    -2.28
    «
    -2.06
    ĸ
    -2.04
    <|outofrange|>
    -2.04
    ↵
    -2.04
    ↵
    -2.04
    ↵  ³³³
    -2.04
    -2.04
    <|outofrange|>
    -2.04
    POSITIVE LOGITS
    't
    2.40
    onical
    2.37
     anyone
    2.28
     someone
    2.12
     anybody
    2.06
    terbury
    1.95
     somebody
    1.92
    cer
    1.72
    ucks
    1.57
    akin
    1.54
    Activations Density 0.015%

    No Known Activations