© 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. 4-RES-SM
    11. 15104
    Prev
    Next
    INDEX
    Explanations

    references to specific types of programming or coding elements

    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/4-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.4.hook_resid_post
    Hook Layer
    4
    Architecture
    standard
    Context Size
    128
    Dataset
    EleutherAI/the_pile_deduplicated
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Neuron Alignment
    Index
    Value
    % of L₁
    21
    +0.30
    2.1%
    172
    +0.21
    1.5%
    89
    +0.17
    1.2%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    21
    +0.30
    0.15
    255
    +0.21
    -0.01
    172
    +0.17
    0.18
    Negative Logits
    dles
    -1.27
     final
    -1.25
    irl
    -1.22
     basic
    -1.22
    htra
    -1.15
    han
    -1.14
     tails
    -1.13
    etimes
    -1.11
    fections
    -1.09
     care
    -1.09
    POSITIVE LOGITS
    Ĩ
    1.83
    Īĺ
    1.81
    º
    1.76
    ı
    1.73
    IJ
    1.71
    ĥ½
    1.68
    Ī
    1.64
    ģ
    1.59
    Ĵ
    1.58
    ĻĤ
    1.58
    Activations Density 4.285%

    No Known Activations