© 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. 2-RES-SM
    11. 10259
    Prev
    Next
    INDEX
    Explanations

    programming constructs related to code safety and structure

    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/2-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.2.hook_resid_post
    Hook Layer
    2
    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
    419
    +0.13
    0.7%
    170
    +0.12
    0.7%
    44
    +0.11
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    136
    +0.13
    -0.04
    431
    +0.12
    0.06
    484
    +0.11
    0.06
    Negative Logits
     Minister
    -1.74
    uds
    -1.55
    metrics
    -1.30
    UV
    -1.28
    ricanes
    -1.26
     journalist
    -1.26
    chaft
    -1.26
     Indies
    -1.25
    bod
    -1.23
    plot
    -1.23
    POSITIVE LOGITS
    completely
    1.67
    unnumbered
    1.56
     cm
    1.54
     apart
    1.48
    alian
    1.43
    aston
    1.41
    square
    1.38
     proof
    1.34
    }}+
    1.31
     '[
    1.31
    Activations Density 3.822%

    No Known Activations