© 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. 26340
    Prev
    Next
    INDEX
    Explanations

    evidence of data structure definitions and manipulations in programming code

    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
    31
    +0.20
    1.1%
    153
    +0.14
    0.8%
    326
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    31
    +0.20
    0.21
    123
    +0.14
    0.14
    93
    +0.13
    0.16
    Negative Logits
     Netflix
    -1.45
     Theater
    -1.42
    herty
    -1.36
     raft
    -1.36
     bott
    -1.36
     Cinema
    -1.33
     Collaboration
    -1.32
     Foundation
    -1.31
     Chair
    -1.31
    mere
    -1.28
    POSITIVE LOGITS
    ĻĤ
    2.54
    ¿½
    2.40
    ·
    2.36
    Į
    2.32
    ·¸
    2.28
    Ŀ
    2.21
    ģ
    2.11
    ľĵ
    2.10
    ¸
    2.06
    Īĺ
    2.06
    Activations Density 7.246%

    No Known Activations