© 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. 12622
    Prev
    Next
    INDEX
    Explanations

    words that indicate various types of emotional responses or reactions

    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
    21
    +0.17
    1.0%
    350
    +0.12
    0.7%
    9
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    9
    +0.17
    0.04
    181
    +0.12
    0.06
    350
    +0.12
    0.04
    Negative Logits
    ftware
    -1.65
    cript
    -1.61
     ""
    -1.56
    ises
    -1.48
    ños
    -1.48
    izes
    -1.40
     population
    -1.39
    wright
    -1.36
    elves
    -1.34
    unnumbered
    -1.34
    POSITIVE LOGITS
    Ĵ
    1.93
    uen
    1.76
    ighters
    1.74
    Ł
    1.70
    ciparum
    1.69
    orescent
    1.59
    herty
    1.58
    odor
    1.58
    orescence
    1.57
    ĭ
    1.56
    Activations Density 0.360%

    No Known Activations