© 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. 8615
    Prev
    Next
    INDEX
    Explanations

    phrases related to restrictions or constraints

    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
    247
    +0.19
    1.0%
    348
    +0.14
    0.8%
    308
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    247
    +0.19
    0.03
    198
    +0.14
    0.04
    308
    +0.13
    0.02
    Negative Logits
    ±
    -1.93
    Ļª
    -1.93
    ¶
    -1.92
    ¹
    -1.83
    ²
    -1.75
    bery
    -1.73
    Īĺ
    -1.67
    »
    -1.66
    ¿
    -1.64
    ĥ½
    -1.60
    POSITIVE LOGITS
     edition
    2.86
     scope
    2.25
     availability
    2.00
     editions
    1.95
     access
    1.91
    ly
    1.77
     liability
    1.76
     applicability
    1.75
     lifespan
    1.75
     extent
    1.71
    Activations Density 0.270%

    No Known Activations