© 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. 4-RES-SM
    11. 9497
    Prev
    Next
    INDEX
    Explanations

    phrases indicating uncertainty or hedging language

    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
    +0.11
    0.6%
    314
    +0.10
    0.6%
    171
    +0.10
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    111
    +0.11
    0.33
    156
    +0.10
    0.26
    510
    +0.10
    0.22
    Negative Logits
    ĥ½
    -2.86
    ĨĴ
    -2.67
    Ļª
    -2.40
    Īĺ
    -2.30
    ľĵ
    -2.30
    İ
    -2.27
    ¢
    -2.22
    ĭ
    -2.08
    ¿
    -2.07
    ij
    -2.07
    POSITIVE LOGITS
     luck
    2.18
     heck
    1.73
     slightest
    1.71
     shame
    1.70
     shit
    1.70
     sheer
    1.70
     obvious
    1.68
     necessity
    1.66
     implication
    1.65
     occasion
    1.64
    Activations Density 1.892%

    No Known Activations