© 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. 1-RES-SM
    11. 28754
    Prev
    Next
    INDEX
    Explanations

    references to calories in food or dietary contexts

    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/1-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.1.hook_resid_post
    Hook Layer
    1
    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
    113
    +0.15
    0.9%
    503
    +0.13
    0.8%
    402
    +0.11
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    113
    +0.15
    0.02
    503
    +0.13
    0.02
    28
    +0.11
    0.01
    Negative Logits
    ³
    -1.82
    ĨĴ
    -1.69
    Ĥ¬
    -1.60
    ĥ
    -1.55
    ·¸
    -1.45
     deadly
    -1.40
    ¯
    -1.39
    ĸ´
    -1.38
     private
    -1.38
    lbrace
    -1.38
    POSITIVE LOGITS
    ifornia
    2.12
     equivalents
    1.97
    oupe
    1.89
    endas
    1.82
    culated
    1.80
    leys
    1.77
    endar
    1.75
    ientes
    1.70
    antry
    1.69
    itudes
    1.69
    Activations Density 0.078%

    No Known Activations