© 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. 27157
    Prev
    Next
    INDEX
    Explanations

    references to the Roman Empire and its historical context

    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
    156
    +0.24
    1.4%
    410
    +0.14
    0.8%
    327
    +0.11
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    410
    +0.24
    0.04
    30
    +0.14
    0.02
    446
    +0.11
    0.01
    Negative Logits
     ordinary
    -1.89
     typical
    -1.69
    other
    -1.68
     normal
    -1.68
     other
    -1.60
     measurable
    -1.59
     certain
    -1.56
     sudden
    -1.50
     particular
    -1.48
     discriminate
    -1.40
    POSITIVE LOGITS
    iens
    2.02
    berg
    1.84
    esque
    1.82
    quist
    1.81
    holm
    1.74
    sey
    1.70
    oque
    1.66
     Era
    1.61
    borg
    1.61
     Ages
    1.60
    Activations Density 0.230%

    No Known Activations