© 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. 32256
    Prev
    Next
    INDEX
    Explanations

    references to technical data and formal documentation

    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
    320
    +0.22
    1.3%
    302
    +0.11
    0.6%
    346
    +0.10
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    320
    +0.22
    0.10
    55
    +0.11
    0.09
    433
    +0.10
    0.04
    Negative Logits
    ľĵ
    -1.89
     instance
    -1.86
     feature
    -1.45
    while
    -1.44
    done
    -1.43
    ĻĤ
    -1.39
    substack
    -1.38
    $;
    -1.36
    č↵
    -1.35
    $)
    -1.34
    POSITIVE LOGITS
    etine
    1.51
     Ranch
    1.46
     aid
    1.43
     rub
    1.38
    iella
    1.38
    stand
    1.28
    ström
    1.28
     Estate
    1.27
     botan
    1.26
    ply
    1.26
    Activations Density 2.508%

    No Known Activations