© 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. 4-RES-SM
    11. 11587
    Prev
    Next
    INDEX
    Explanations

    mentions of legal professionals or authorities, specifically lawyers or judges

    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
    178
    +0.13
    0.7%
    412
    +0.11
    0.6%
    98
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    110
    +0.13
    0.04
    412
    +0.11
    0.04
    224
    +0.11
    0.03
    Negative Logits
    ction
    -1.65
    ²
    -1.49
    zzle
    -1.45
     generality
    -1.43
    ¸
    -1.39
     millenn
    -1.36
    coming
    -1.33
    fileID
    -1.33
     sight
    -1.32
     priority
    -1.29
    POSITIVE LOGITS
    ubern
    1.97
    arrison
    1.87
    illes
    1.83
    lick
    1.75
    ordon
    1.72
    arcia
    1.70
    rosse
    1.70
    olds
    1.68
    UID
    1.67
    oltz
    1.62
    Activations Density 0.055%

    No Known Activations