© 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. 4-RES-SM
    11. 10908
    Prev
    Next
    INDEX
    Explanations

    references to medical conditions and patient demographics

    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
    348
    +0.21
    1.3%
    502
    +0.12
    0.7%
    25
    +0.10
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    502
    +0.21
    0.09
    348
    +0.12
    0.06
    361
    +0.10
    0.08
    Negative Logits
     mood
    -1.51
    las
    -1.51
    cludes
    -1.50
     shame
    -1.43
     apologies
    -1.37
     existence
    -1.35
    endment
    -1.35
    woke
    -1.34
    stood
    -1.31
     sleep
    -1.29
    POSITIVE LOGITS
    pecific
    2.13
    artan
    1.92
    quo
    1.77
     belonging
    1.69
     skilled
    1.56
    omal
    1.56
    erna
    1.54
    nai
    1.52
    erving
    1.50
     deemed
    1.50
    Activations Density 1.158%

    No Known Activations