© 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 EvalsExportsAPI 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. 12729
    Prev
    Next
    INDEX
    Explanations

    references to time periods and frequency of actions

    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
    397
    +0.19
    1.1%
    337
    +0.11
    0.6%
    330
    +0.10
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    397
    +0.19
    0.04
    500
    +0.11
    0.04
    337
    +0.10
    0.04
    Negative Logits
    Ĩ
    -4.44
    Ľ
    -4.21
    Ŀ
    -4.12
    Ļª
    -4.07
    Ļ
    -3.96
    Ĵ
    -3.94
    ĻĤ
    -3.93
    ¢
    -3.91
    °
    -3.87
    į
    -3.87
    POSITIVE LOGITS
     thereafter
    1.91
    day
    1.85
    jour
    1.65
     whatsoever
    1.59
    ation
    1.59
    hops
    1.58
     except
    1.55
    nis
    1.53
    leep
    1.47
    places
    1.41
    Activations Density 0.516%

    No Known Activations