© 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. 9827
    Prev
    Next
    INDEX
    Explanations

    phrases indicating association or engagement with specific actions or entities

    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
    478
    +0.23
    1.3%
    358
    +0.16
    0.9%
    47
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    358
    +0.23
    0.04
    278
    +0.16
    0.03
    246
    +0.12
    0.04
    Negative Logits
    thal
    -1.71
    burg
    -1.61
     contributor
    -1.61
    berg
    -1.50
    aden
    -1.49
    OUS
    -1.45
    ium
    -1.45
    ard
    -1.43
    conduct
    -1.42
     liable
    -1.41
    POSITIVE LOGITS
     respect
    2.58
     regard
    2.28
     intention
    2.06
     nothing
    1.97
     luck
    1.92
     nowhere
    1.85
     regards
    1.83
     intentions
    1.78
     lots
    1.74
     purpose
    1.73
    Activations Density 0.139%

    No Known Activations