© 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. 32425
    Prev
    Next
    INDEX
    Explanations

    the presence of specific phrases or clauses that indicate conditional scenarios or 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/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
    243
    +0.17
    1.0%
    376
    +0.11
    0.6%
    16
    +0.10
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    243
    +0.17
    0.08
    282
    +0.11
    0.06
    152
    +0.10
    0.04
    Negative Logits
    ?:
    -1.65
     ladder
    -1.60
    ?
    -1.59
    .*
    -1.53
     angles
    -1.53
    edly
    -1.51
    ?!
    -1.48
    !
    -1.48
    ystem
    -1.46
    +.
    -1.41
    POSITIVE LOGITS
    jection
    2.44
    fection
    2.35
    »
    2.31
    Ń
    2.20
    Ī
    2.19
    »¿
    2.12
    ¢
    2.10
    2.09
    ↵
    2.09
    ↵
    2.09
    Activations Density 0.166%

    No Known Activations