© 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. 27142
    Prev
    Next
    INDEX
    Explanations

    articles and their frequency in the text

    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
    98
    +0.20
    1.1%
    342
    +0.12
    0.6%
    478
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    98
    +0.20
    0.01
    436
    +0.12
    0.03
    68
    +0.11
    0.04
    Negative Logits
    enance
    -2.08
    ions
    -1.81
    ups
    -1.78
    itures
    -1.72
     errors
    -1.68
    fulness
    -1.59
    ocols
    -1.54
    lessness
    -1.52
    ttes
    -1.49
     discrepancies
    -1.49
    POSITIVE LOGITS
     practitioner
    1.89
     representative
    1.84
     member
    1.82
     measure
    1.81
     predictor
    1.69
     matter
    1.68
     source
    1.67
     mean
    1.65
     colleague
    1.64
    uster
    1.64
    Activations Density 0.190%

    No Known Activations