© 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. 1-RES-SM
    11. 23910
    Prev
    Next
    INDEX
    Explanations

    references to sodium and its various chemical contexts

    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/1-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.1.hook_resid_post
    Hook Layer
    1
    Architecture
    standard
    Context Size
    128
    Dataset
    EleutherAI/the_pile_deduplicated
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Neuron Alignment
    Index
    Value
    % of L₁
    369
    +0.26
    1.5%
    410
    +0.12
    0.7%
    457
    +0.11
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    111
    +0.26
    0.02
    369
    +0.12
    0.01
    334
    +0.11
    0.02
    Negative Logits
    acial
    -1.71
     else
    -1.57
    uppose
    -1.51
    >/
    -1.43
     bother
    -1.43
    cip
    -1.41
    alls
    -1.41
     prosecute
    -1.41
    rapeut
    -1.39
    oving
    -1.38
    POSITIVE LOGITS
    issance
    2.27
    iki
    1.88
    ître
    1.82
     Bros
    1.76
    esta
    1.68
    leen
    1.66
    ÅĽci
    1.55
    ïve
    1.53
     yourself
    1.51
    iko
    1.50
    Activations Density 0.102%

    No Known Activations