© 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. 2-RES-SM
    11. 26250
    Prev
    Next
    INDEX
    Explanations

    characters or symbols that are commonly used in code or URL encoding

    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/2-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.2.hook_resid_post
    Hook Layer
    2
    Architecture
    standard
    Context Size
    128
    Dataset
    EleutherAI/the_pile_deduplicated
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Neuron Alignment
    Index
    Value
    % of L₁
    478
    +0.17
    0.9%
    205
    +0.15
    0.8%
    369
    +0.15
    0.8%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    396
    +0.17
    0.03
    47
    +0.15
    0.03
    377
    +0.15
    0.03
    Negative Logits
    manship
    -1.85
    casters
    -1.83
    ars
    -1.78
    stroke
    -1.65
    writers
    -1.59
    writer
    -1.58
    ening
    -1.54
    caster
    -1.51
    ing
    -1.48
    ball
    -1.43
    POSITIVE LOGITS
    nbsp
    3.11
    quot
    2.19
    rsquo
    2.13
    lt
    1.98
    gt
    1.95
    agra
    1.85
    llll
    1.58
     expect
    1.57
    amp
    1.53
    Īĺ
    1.48
    Activations Density 0.105%

    No Known Activations