© 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 EvalsExportsAPI 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. 12887
    Prev
    Next
    INDEX
    Explanations

    occurrences of the substring "ur" within words

    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/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
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    21
    +0.13
    0.7%
    507
    +0.11
    0.6%
    450
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    257
    +0.13
    0.03
    9
    +0.11
    0.02
    410
    +0.11
    0.03
    Negative Logits
     interest
    -1.79
    er
    -1.73
     alike
    -1.65
    edly
    -1.56
    erty
    -1.47
    o
    -1.46
    a
    -1.41
    enburg
    -1.41
     same
    -1.40
    ensively
    -1.39
    POSITIVE LOGITS
    ricane
    2.57
    ½
    2.54
    isdiction
    2.27
    anium
    2.11
    º
    2.10
    ricanes
    2.05
    vival
    2.04
    ¼
    1.98
    ·
    1.95
    Į
    1.95
    Activations Density 0.133%

    No Known Activations