© 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. 26215
    Prev
    Next
    INDEX
    Explanations

    occurrences of the word "reach" and its variations in different contexts

    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
    457
    +0.14
    0.8%
    358
    +0.13
    0.7%
    292
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    457
    +0.14
    0.03
    292
    +0.13
    0.03
    317
    +0.13
    0.03
    Negative Logits
    ¨
    -2.00
    onso
    -1.67
    ²
    -1.67
    ĭ
    -1.61
    ĥ½
    -1.59
    ¿½
    -1.59
    ŀ
    -1.55
    »¿
    -1.54
    ksen
    -1.53
    İ
    -1.52
    POSITIVE LOGITS
    eeper
    1.92
     depths
    1.68
    ables
    1.63
     speeds
    1.63
    able
    1.59
    gmail
    1.57
     confluence
    1.57
    point
    1.50
     maturity
    1.45
     adulthood
    1.45
    Activations Density 0.190%

    No Known Activations