© 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. 26613
    Prev
    Next
    INDEX
    Explanations

    interjections and conversational cues that indicate a speaker's reactions or transitions in dialogue

    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
    2
    +0.12
    0.7%
    478
    +0.12
    0.7%
    477
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    477
    +0.12
    0.05
    2
    +0.12
    0.04
    420
    +0.12
    0.03
    Negative Logits
    ļ
    -2.51
    ħ
    -2.38
    ŀ
    -2.25
    ĵ
    -2.23
    Ļ
    -2.21
    ľĵ
    -2.17
    ij
    -2.06
    Ķ
    -2.06
    ĨĴ
    -2.05
    ĥ
    -2.05
    POSITIVE LOGITS
    áĢº
    2.15
    inki
    1.78
    ert
    1.76
     ÑĩÑĤо
    1.64
    come
    1.54
    strom
    1.52
    BITS
    1.52
    áĢ¬
    1.52
    bye
    1.48
     Round
    1.46
    Activations Density 0.344%

    No Known Activations