© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    5. Pythia-70M-Deduped
    7. Residuals
    9. 1-RES-SM
    11. 29200
    Prev
    Next
    INDEX
    Explanations

    the word "commonly" indicating frequent or usual occurrences

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    ctigges/pythia-70m-deduped__res-sm_processed/1-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.1.hook_resid_post
    Hook Layer
    1
    Architecture
    standard
    Context Size
    128
    Dataset
    EleutherAI/the_pile_deduplicated
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    87
    +0.15
    0.9%
    23
    +0.13
    0.7%
    39
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    39
    +0.15
    0.02
    87
    +0.13
    0.02
    257
    +0.12
    0.01
    Negative Logits
     integrity
    -2.11
    ts
    -1.61
     capacity
    -1.55
    tle
    -1.54
    tube
    -1.50
    leg
    -1.50
    ľ
    -1.45
    fully
    -1.44
    mit
    -1.44
    ew
    -1.42
    POSITIVE LOGITS
     cited
    1.66
     speaking
    1.59
    places
    1.56
     penicillin
    1.50
     voiced
    1.50
    ympt
    1.49
     encountered
    1.49
     abbreviated
    1.48
     PCT
    1.43
     referred
    1.42
    Activations Density 0.041%

    No Known Activations