© 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. 14286
    Prev
    Next
    INDEX
    Explanations

    words related to the concept of being able to do something or capability

    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
    280
    +0.12
    0.7%
    452
    +0.10
    0.6%
    74
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    252
    +0.12
    0.03
    148
    +0.10
    0.01
    213
    +0.10
    0.02
    Negative Logits
     their
    -1.41
     executed
    -1.36
    fected
    -1.31
    oration
    -1.30
    oral
    -1.29
    .."
    -1.25
     AUTHOR
    -1.22
     occasion
    -1.20
    orian
    -1.19
     line
    -1.19
    POSITIVE LOGITS
    £
    2.36
    Ł
    2.23
    ³
    2.06
    ĻĤ
    2.04
    mente
    2.02
    ¦
    2.02
    ¾
    1.98
    Ĵ
    1.94
    į
    1.87
    ´
    1.86
    Activations Density 0.220%

    No Known Activations