© 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. 9169
    Prev
    Next
    INDEX
    Explanations

    historical references or events related to nobility and royalty

    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
    210
    +0.15
    0.8%
    136
    +0.14
    0.8%
    168
    +0.13
    0.8%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    111
    +0.15
    0.17
    510
    +0.14
    0.16
    413
    +0.13
    0.12
    Negative Logits
     video
    -2.09
     YouTube
    -1.97
     CNN
    -1.94
     microphone
    -1.89
     workplace
    -1.86
     behavioral
    -1.80
     headphones
    -1.79
     audio
    -1.79
     Video
    -1.78
     videos
    -1.77
    POSITIVE LOGITS
     Emperor
    3.29
     royal
    3.19
     king
    3.16
     monarch
    3.08
     kings
    2.99
     emperor
    2.98
     princes
    2.70
     throne
    2.69
     palace
    2.66
     heir
    2.65
    Activations Density 2.036%

    No Known Activations