© 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. 9817
    Prev
    Next
    INDEX
    Explanations

    various programming syntax elements and code structures

    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
    156
    +0.68
    5.0%
    397
    +0.11
    0.8%
    390
    +0.10
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    156
    +0.68
    0.27
    111
    +0.11
    0.17
    271
    +0.10
    -0.03
    Negative Logits
    itian
    -1.31
    olin
    -1.19
    olf
    -1.15
    its
    -1.13
    1451450014514500
    -1.09
    thia
    -1.09
    uality
    -1.08
    ferr
    -1.08
    sein
    -1.08
    ftware
    -1.06
    POSITIVE LOGITS
    ĻĤ
    3.13
    ©
    3.08
    ·¸
    3.06
    Īĺ
    3.04
    Ķ
    2.96
    ¦
    2.92
    £
    2.92
    Ģ
    2.87
    »¿
    2.87
    IJ
    2.86
    Activations Density 5.322%

    No Known Activations