© 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. 4-RES-SM
    11. 14859
    Prev
    Next
    INDEX
    Explanations

    phrases beginning with "but."

    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
    307
    +0.29
    1.7%
    380
    +0.11
    0.7%
    128
    +0.10
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    128
    +0.29
    0.07
    380
    +0.11
    0.08
    316
    +0.10
    0.07
    Negative Logits
     worden
    -1.80
    .](
    -1.66
    )->
    -1.66
    oster
    -1.62
    ){#
    -1.59
     été
    -1.59
    )].
    -1.58
    ---|---|---
    -1.57
     {#
    -1.56
    .^[@
    -1.52
    POSITIVE LOGITS
    tons
    2.86
    tig
    2.58
     what
    2.08
     wait
    2.02
    what
    1.99
    how
    1.98
     unlike
    1.97
    why
    1.90
     why
    1.87
     how
    1.77
    Activations Density 0.482%

    No Known Activations