© 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. 2-RES-SM
    11. 27983
    Prev
    Next
    INDEX
    Explanations

    licensing information and permissions related to software distribution

    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/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
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    34
    +0.16
    0.9%
    233
    +0.13
    0.7%
    450
    +0.10
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    364
    +0.16
    0.04
    353
    +0.13
    0.03
    53
    +0.10
    0.04
    Negative Logits
    onica
    -1.58
    apan
    -1.52
    hline
    -1.52
    arer
    -1.49
    hell
    -1.47
    .[]{
    -1.44
    ena
    -1.44
     respectively
    -1.40
    aster
    -1.35
     \[[
    -1.33
    POSITIVE LOGITS
     citizenship
    1.73
     autonomy
    1.59
    ================================================================
    1.48
     grant
    1.40
     CONTRIBUTORS
    1.37
     licenses
    1.37
     fellowship
    1.33
    TRODUCTION
    1.32
    sible
    1.30
    License
    1.30
    Activations Density 0.183%

    No Known Activations