© 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. 4-RES-SM
    11. 13698
    Prev
    Next
    INDEX
    Explanations

    numeric values and their mathematical relationships

    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/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
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    255
    +0.16
    1.0%
    21
    +0.15
    0.9%
    506
    +0.13
    0.8%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    111
    +0.16
    0.08
    71
    +0.15
    0.06
    156
    +0.13
    0.06
    Negative Logits
    ting
    -1.60
     various
    -1.49
     life
    -1.46
     everywhere
    -1.43
     creation
    -1.41
     articles
    -1.40
     same
    -1.36
    ities
    -1.32
     lately
    -1.32
     assembled
    -1.31
    POSITIVE LOGITS
    ŀ
    4.40
    ¢
    4.27
    Ī
    4.08
    ¤
    4.08
    ķ
    4.07
    IJ
    4.00
    ¬
    3.99
    Ĵ
    3.96
    Ŀ
    3.92
    Ļ
    3.90
    Activations Density 0.153%

    No Known Activations