© 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. 1-RES-SM
    11. 25792
    Prev
    Next
    INDEX
    Explanations

    variations of the word "Straight."

    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/1-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.1.hook_resid_post
    Hook Layer
    1
    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
    369
    +0.16
    0.9%
    59
    +0.15
    0.8%
    233
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    59
    +0.16
    0.03
    369
    +0.15
    0.02
    386
    +0.13
    0.02
    Negative Logits
     dots
    -1.61
     expense
    -1.57
     interest
    -1.54
    acular
    -1.48
     ears
    -1.48
    tons
    -1.46
     behalf
    -1.46
     natural
    -1.45
     Ideas
    -1.44
     interests
    -1.40
    POSITIVE LOGITS
    ī
    1.80
    pless
    1.76
    rens
    1.67
    hene
    1.64
    fty
    1.63
    igh
    1.63
    eless
    1.60
    Ļ
    1.57
    ³
    1.55
    Īĺ
    1.54
    Activations Density 0.134%

    No Known Activations