© 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. 15187
    Prev
    Next
    INDEX
    Explanations

    technical terms and code structure elements relevant to programming

    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
    78
    +0.25
    1.6%
    21
    +0.17
    1.0%
    48
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    78
    +0.25
    0.02
    21
    +0.17
    0.06
    126
    +0.12
    0.01
    Negative Logits
    soever
    -1.60
    hydration
    -1.30
    LLY
    -1.27
    Errno
    -1.25
    balls
    -1.25
    .',
    -1.25
    strap
    -1.24
    jets
    -1.21
     prompt
    -1.21
    npmjs
    -1.20
    POSITIVE LOGITS
    ī
    2.01
    Ķ
    1.99
    ¸
    1.95
    ķ
    1.87
    ĵ
    1.83
    Īĺ
    1.83
    §
    1.83
    ¼
    1.81
    ĸ
    1.80
    IJ
    1.75
    Activations Density 0.431%

    No Known Activations