© 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. 15154
    Prev
    Next
    INDEX
    Explanations

    legal terminology and references related to licensing agreements

    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
    271
    +0.20
    1.2%
    78
    +0.11
    0.7%
    338
    +0.10
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    241
    +0.20
    0.03
    126
    +0.11
    0.00
    441
    +0.10
    0.01
    Negative Logits
     favour
    -1.45
    merce
    -1.39
     UK
    -1.36
    anguage
    -1.30
    ctic
    -1.28
    azine
    -1.27
     agg
    -1.25
    ters
    -1.24
    uximab
    -1.24
     rig
    -1.23
    POSITIVE LOGITS
    Īĺ
    2.35
    «
    2.25
    Ĺ
    1.98
    ¥
    1.91
    ²
    1.89
    £
    1.87
    ī
    1.85
    ¤
    1.85
    Ī
    1.82
    Ļ
    1.81
    Activations Density 0.068%

    No Known Activations