© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 29089
    Prev
    Next
    INDEX
    Explanations

    references to facades or architectural features

    oai_token-act-pair · gpt-4o-miniTriggered 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
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    194
    +0.14
    0.8%
    203
    +0.13
    0.7%
    429
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    111
    +0.14
    0.10
    271
    +0.13
    0.04
    156
    +0.11
    0.05
    Negative Logits
    ingale
    -1.71
     pleas
    -1.58
     certain
    -1.54
     thoughts
    -1.52
     vague
    -1.51
     plain
    -1.47
     repetitive
    -1.46
    rift
    -1.45
    ctic
    -1.45
     hardest
    -1.44
    POSITIVE LOGITS
    idity
    1.81
    ettes
    1.74
    iders
    1.58
    thouse
    1.57
    utions
    1.57
    encies
    1.54
    leys
    1.50
    emen
    1.50
    istes
    1.48
    etto
    1.46
    Activations Density 0.351%

    No Known Activations