© 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. 27807
    Prev
    Next
    INDEX
    Explanations

    structured data, specifically related to coding and programming elements

    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
    240
    +0.16
    0.9%
    6
    +0.16
    0.9%
    302
    +0.15
    0.9%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    6
    +0.16
    0.13
    310
    +0.16
    0.10
    400
    +0.15
    0.03
    Negative Logits
    Ĩ
    -1.73
    ķ
    -1.49
    missive
    -1.44
    ļ
    -1.39
    aler
    -1.36
    ¢
    -1.36
    ¤
    -1.35
     alone
    -1.34
    Ĵ
    -1.34
    Ĥ¬
    -1.33
    POSITIVE LOGITS
    uities
    1.76
    err
    1.64
    ções
    1.61
    ftware
    1.44
    arity
    1.43
    él
    1.43
    ção
    1.38
    uity
    1.38
    ()</
    1.34
    StackTrace
    1.32
    Activations Density 4.721%

    No Known Activations