© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW 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. 8834
    Prev
    Next
    INDEX
    Explanations

    patterns or occurrences of numerical data, specifically focusing on sequences or quantities

    oai_token-act-pair · gpt-4o-mini
    Triggered 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
    Neuron Alignment
    Index
    Value
    % of L₁
    306
    +0.27
    1.7%
    307
    +0.26
    1.6%
    21
    +0.26
    1.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    450
    +0.27
    0.02
    255
    +0.26
    -0.06
    306
    +0.26
    0.05
    Negative Logits
     \[
    -2.13
     [@
    -2.11
    âĢħ
    -2.02
    .[^
    -1.98
     \[*
    -1.89
    \'
    -1.88
    \_
    -1.88
     realised
    -1.85
     £
    -1.84
     [(\[
    -1.83
    POSITIVE LOGITS
     ordinal
    1.43
     Percent
    1.31
    degree
    1.26
     histologic
    1.24
     County
    1.23
    Ļª
    1.21
    IJ
    1.16
     flavors
    1.16
    avorite
    1.15
     Representative
    1.14
    Activations Density 3.421%

    No Known Activations