© 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. 22766
    Prev
    Next
    INDEX
    Explanations

    phrases expressing generalizations or underlying implications

    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
    98
    +0.12
    0.7%
    206
    +0.11
    0.6%
    225
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    225
    +0.12
    0.03
    98
    +0.11
    0.03
    335
    +0.11
    0.03
    Negative Logits
    ª
    -1.94
    shit
    -1.39
     experience
    -1.37
    places
    -1.36
     socket
    -1.36
    safe
    -1.35
    Īĺ
    -1.33
    ...)
    -1.33
     sockets
    -1.31
     hence
    -1.30
    POSITIVE LOGITS
    oddsidemargin
    1.57
    escence
    1.52
    éri
    1.50
    herin
    1.47
    quez
    1.47
    itarian
    1.42
    astro
    1.41
    ças
    1.39
     /*!
    1.38
    oulos
    1.37
    Activations Density 0.297%

    No Known Activations