© 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. 11222
    Prev
    Next
    INDEX
    Explanations

    references to the term "corruption."

    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
    300
    +0.12
    0.7%
    441
    +0.11
    0.6%
    485
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    437
    +0.12
    0.02
    300
    +0.11
    0.01
    441
    +0.11
    0.01
    Negative Logits
    ers
    -1.88
     attention
    -1.86
    ires
    -1.65
    hed
    -1.63
     grasp
    -1.61
    ering
    -1.59
    ually
    -1.56
    ureus
    -1.48
    yer
    -1.44
    etter
    -1.44
    POSITIVE LOGITS
    rected
    2.66
    rections
    2.63
    Ĵ
    2.27
    ruption
    2.25
    responding
    2.23
    ĨĴ
    2.17
    Ļª
    2.14
    ĻĤ
    2.10
    rente
    2.09
    rection
    2.07
    Activations Density 0.108%

    No Known Activations