© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    5. Pythia-70M-Deduped
    7. Residuals
    9. 2-RES-SM
    11. 32404
    Prev
    Next
    INDEX
    Explanations

    references to nuclear energy and weapons

    oai_token-act-pair · gpt-4o-mini
    Triggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    ctigges/pythia-70m-deduped__res-sm_processed/2-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.2.hook_resid_post
    Hook Layer
    2
    Architecture
    standard
    Context Size
    128
    Dataset
    EleutherAI/the_pile_deduplicated
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Neuron Alignment
    Index
    Value
    % of L₁
    156
    +0.18
    1.1%
    189
    +0.12
    0.7%
    30
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    189
    +0.18
    0.02
    486
    +0.12
    0.01
    145
    +0.12
    0.02
    Negative Logits
     cref
    -2.11
    bbe
    -1.68
    eding
    -1.62
     ours
    -1.60
    hav
    -1.59
    riber
    -1.59
     patient
    -1.59
    abb
    -1.54
    ères
    -1.52
     frequent
    -1.51
    POSITIVE LOGITS
    ¤
    2.37
    Ń
    2.31
    Ĥ
    2.30
    ħ
    2.29
    £
    2.21
    ¬
    2.19
    º
    2.13
    Ħ
    2.12
    ĸ
    2.11
    ¸
    2.07
    Activations Density 0.105%

    No Known Activations