© 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. 14190
    Prev
    Next
    INDEX
    Explanations

    topics related to various features and functionalities in software and technology

    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
    127
    +0.14
    0.7%
    424
    +0.13
    0.7%
    335
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    386
    +0.14
    0.03
    424
    +0.13
    0.02
    222
    +0.13
    0.03
    Negative Logits
    iction
    -1.68
    hip
    -1.61
    oco
    -1.61
    robe
    -1.57
    ellar
    -1.50
    ome
    -1.48
    ars
    -1.47
    itch
    -1.47
    rums
    -1.44
    aging
    -1.44
    POSITIVE LOGITS
     -----
    1.61
     fault
    1.51
    voir
    1.42
     Contin
    1.41
    loat
    1.41
     breath
    1.40
     dawn
    1.38
     fog
    1.36
    âĢĤ
    1.34
    1.34
    Activations Density 0.087%

    No Known Activations