© 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. 22915
    Prev
    Next
    INDEX
    Explanations

    instances of contrasting statements or "but" clauses within the text

    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
    88
    +0.13
    0.8%
    424
    +0.13
    0.7%
    451
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    274
    +0.13
    0.10
    470
    +0.13
    0.12
    247
    +0.12
    0.11
    Negative Logits
     pleasure
    -1.68
     glad
    -1.67
     forward
    -1.61
    ureus
    -1.46
     us
    -1.44
    ://
    -1.44
     pleased
    -1.37
     shone
    -1.36
     follow
    -1.33
     hello
    -1.31
    POSITIVE LOGITS
    °
    1.88
    «
    1.82
    ĥ½
    1.72
    ¤
    1.72
     Illegal
    1.70
     underest
    1.67
    Ľ
    1.65
    ¥
    1.63
     riots
    1.61
    ¿
    1.61
    Activations Density 2.278%

    No Known Activations