© 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. 2-MLP-SM
    11. 23859
    Prev
    Next
    INDEX
    Explanations

    instances of the word "know" and its variations

    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/2-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.2.hook_mlp_out
    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
    44
    +0.11
    0.6%
    506
    +0.11
    0.6%
    274
    +0.10
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    506
    +0.11
    0.03
    44
    +0.11
    0.01
    363
    +0.10
    0.02
    Negative Logits
     ([@
    -1.71
     commissioned
    -1.67
     famous
    -1.59
     enacted
    -1.59
     pioneering
    -1.58
    **.
    -1.56
     (**
    -1.53
     literature
    -1.52
     [@
    -1.49
     motivated
    -1.46
    POSITIVE LOGITS
    tone
    1.85
    lett
    1.77
    glich
    1.76
    ħ
    1.61
    ½
    1.56
    bye
    1.51
    blockList
    1.39
    stein
    1.39
    iel
    1.36
    sie
    1.35
    Activations Density 0.097%

    No Known Activations