© 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. MLP Post
    9. 0-MLP-SM
    11. 26194
    Prev
    Next
    INDEX
    Explanations

    occurrences of the word "through."

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    ctigges/pythia-70m-deduped__mlp-sm_processed/0-mlp-sm
    Prompts (Dashboard)
    32,768 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    32,768
    Data Type
    torch.float32
    Hook Name
    blocks.0.hook_mlp_out
    Hook Layer
    0
    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
    279
    +0.12
    0.6%
    14
    +0.11
    0.6%
    90
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    117
    +0.12
    0.07
    107
    +0.11
    0.06
    95
    +0.11
    0.06
    Negative Logits
    Ĵ
    -1.97
    ŀ
    -1.82
    ĨĴ
    -1.79
    ģ
    -1.68
    ¿
    -1.59
    Ĩ
    -1.58
    ı
    -1.56
    ¯
    -1.51
    ĥ½
    -1.50
    -1.49
    POSITIVE LOGITS
    than
    1.63
    ttes
    1.49
    forces
    1.47
    LAND
    1.42
    nae
    1.41
    nez
    1.40
    mare
    1.39
    worker
    1.38
    end
    1.35
    holds
    1.35
    Activations Density 0.056%

    No Known Activations