© 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. 2-RES-SM
    11. 12588
    Prev
    Next
    INDEX
    Explanations

    instances of the word "rapid" in various contexts, indicating a focus on speed or quick changes

    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
    477
    +0.16
    0.9%
    405
    +0.13
    0.7%
    317
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    148
    +0.16
    0.01
    477
    +0.13
    0.02
    114
    +0.12
    0.01
    Negative Logits
     behalf
    -1.69
     particular
    -1.62
     coli
    -1.48
    oard
    -1.44
    clusively
    -1.44
     coff
    -1.40
     behold
    -1.40
     Journal
    -1.36
     favor
    -1.34
    TRODUCTION
    -1.34
    POSITIVE LOGITS
    ity
    2.33
    ¢
    2.06
    ¡
    2.02
    ķ
    2.00
    ¨
    1.93
    Ł
    1.87
    ł
    1.78
    »
    1.77
    Ģ
    1.73
     heartbeat
    1.73
    Activations Density 0.069%

    No Known Activations