© 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. 12547
    Prev
    Next
    INDEX
    Explanations

    phrases expressing belief or conviction

    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
    25
    +0.15
    0.8%
    390
    +0.12
    0.7%
    106
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    33
    +0.15
    0.06
    485
    +0.12
    0.04
    106
    +0.12
    0.04
    Negative Logits
    ¼
    -3.85
    Īĺ
    -3.67
    ¿
    -3.60
    į
    -3.39
    ¿½
    -3.32
    Ļª
    -3.19
    Ŀ
    -3.18
    ¸
    -3.15
    »
    -3.15
    µ
    -3.12
    POSITIVE LOGITS
     that
    2.42
     it
    2.36
     they
    2.22
     there
    2.15
    that
    1.97
     we
    1.77
     you
    1.67
     this
    1.66
     humans
    1.63
     most
    1.62
    Activations Density 0.364%

    No Known Activations