© 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. 9277
    Prev
    Next
    INDEX
    Explanations

    parenthetical expressions or mathematical notations

    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
    +0.18
    1.1%
    78
    +0.17
    1.0%
    121
    +0.09
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    451
    +0.18
    0.00
    398
    +0.17
    0.01
    297
    +0.09
    0.01
    Negative Logits
     ordinary
    -1.45
     accord
    -1.41
    nament
    -1.24
     squarely
    -1.24
     purposes
    -1.24
     reck
    -1.22
    iginally
    -1.20
     proportions
    -1.20
     call
    -1.19
     section
    -1.19
    POSITIVE LOGITS
    ĥ½
    2.89
    ŀ
    2.81
    ĵ
    2.57
    ¬
    2.57
    ¯
    2.48
    §
    2.46
    ĨĴ
    2.44
    ij
    2.41
    IJ
    2.41
    ¾
    2.40
    Activations Density 0.017%

    No Known Activations