© 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. 4-RES-SM
    11. 9892
    Prev
    Next
    INDEX
    Explanations

    the word "Cent" in various 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/4-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.4.hook_resid_post
    Hook Layer
    4
    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
    320
    +0.15
    0.8%
    9
    +0.15
    0.8%
    345
    +0.13
    0.8%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    9
    +0.15
    0.02
    218
    +0.15
    0.03
    349
    +0.13
    0.03
    Negative Logits
    matically
    -2.27
    pite
    -1.78
    iginally
    -1.77
    mith
    -1.65
    jee
    -1.56
    writing
    -1.55
    soever
    -1.55
     utmost
    -1.51
    written
    -1.40
    jos
    -1.40
    POSITIVE LOGITS
    ĭ
    3.08
    enary
    3.06
    ennial
    3.05
    Ģ
    3.05
    Ī
    2.99
    ®
    2.94
    ĨĴ
    2.89
    Ħ
    2.87
    ª
    2.83
    ķ
    2.75
    Activations Density 0.225%

    No Known Activations