© 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. 8606
    Prev
    Next
    INDEX
    Explanations

    references to various spatial positions and orientations

    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
    25
    +0.16
    0.9%
    466
    +0.14
    0.8%
    458
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    6
    +0.16
    0.05
    401
    +0.14
    0.03
    18
    +0.12
    0.06
    Negative Logits
    ĻĤ
    -3.74
    ĵ
    -3.04
    ĩ
    -3.02
    Ģ
    -3.02
    ľĵ
    -2.99
    ²
    -2.89
    °
    -2.82
    ij
    -2.81
    Ĥ
    -2.81
    «
    -2.72
    POSITIVE LOGITS
     thereof
    3.04
    most
    2.85
    ermost
    2.68
     portion
    2.28
     corner
    2.27
     edge
    2.20
     of
    2.17
     side
    2.10
     corners
    2.05
     edges
    2.03
    Activations Density 0.683%

    No Known Activations