© 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. 14180
    Prev
    Next
    INDEX
    Explanations

    numbers and their significant appearances in context

    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
    153
    +0.13
    0.8%
    203
    +0.13
    0.7%
    498
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    203
    +0.13
    0.07
    34
    +0.13
    0.06
    227
    +0.12
    0.04
    Negative Logits
    ĻĤ
    -2.77
    į
    -2.73
    ¶
    -2.64
    ĥ½
    -2.60
    ¿½
    -2.59
    ĭ
    -2.53
    ¦
    -2.52
    µ
    -2.50
    Į
    -2.44
    Ĭ
    -2.42
    POSITIVE LOGITS
    era
    1.63
     terminus
    1.58
    #####
    1.39
    mart
    1.34
    orig
    1.32
    USP
    1.31
     Suite
    1.30
    XX
    1.30
     beginning
    1.30
     Pages
    1.28
    Activations Density 0.407%

    No Known Activations