© 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. 2-RES-SM
    11. 13476
    Prev
    Next
    INDEX
    Explanations

    expressions of personal identity and self-description

    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
    56
    +0.15
    0.9%
    218
    +0.14
    0.8%
    5
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    56
    +0.15
    -0.01
    91
    +0.14
    0.10
    326
    +0.13
    0.05
    Negative Logits
     sides
    -1.60
     symbols
    -1.55
    mente
    -1.53
    yline
    -1.50
     respectively
    -1.50
     corresponding
    -1.49
     their
    -1.46
     resultant
    -1.45
     represents
    -1.45
     successive
    -1.42
    POSITIVE LOGITS
     habit
    1.69
    python
    1.62
    buntu
    1.56
     minded
    1.54
    arner
    1.52
    ologie
    1.48
     trouble
    1.46
    adow
    1.46
    working
    1.37
    kern
    1.37
    Activations Density 3.330%

    No Known Activations