© 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. 2-RES-SM
    11. 29619
    Prev
    Next
    INDEX
    Explanations

    emotional expressions related to love and relationships

    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
    258
    +0.22
    1.3%
    23
    +0.22
    1.3%
    203
    +0.18
    1.1%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    258
    +0.22
    0.20
    23
    +0.22
    0.16
    189
    +0.18
    0.12
    Negative Logits
    tees
    -2.11
    ème
    -1.90
     unpublished
    -1.70
    ister
    -1.65
    book
    -1.64
     remarks
    -1.61
    bourg
    -1.59
    nikov
    -1.58
    aine
    -1.58
     reviewer
    -1.57
    POSITIVE LOGITS
     They
    1.75
     Maybe
    1.69
    rap
    1.68
    while
    1.58
    _**
    1.57
     Especially
    1.54
     habit
    1.51
    mozilla
    1.51
     Holy
    1.49
     Sure
    1.49
    Activations Density 1.715%

    No Known Activations