© 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. 10848
    Prev
    Next
    INDEX
    Explanations

    references to biological or scientific entities and 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
    397
    +0.16
    0.9%
    21
    +0.14
    0.8%
    71
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    398
    +0.16
    0.04
    229
    +0.14
    0.09
    287
    +0.11
    0.09
    Negative Logits
    į
    -4.85
    <|outofrange|>
    -4.84
    -4.84
    -4.84
    ↵
    -4.84
    č↵
    -4.84
    <|outofrange|>
    -4.84
    <|padding|>
    -4.84
         ↵
    -4.84
    -4.84
    POSITIVE LOGITS
     include
    3.39
     are
    3.29
     contain
    3.23
     were
    3.09
     comprise
    2.95
     involve
    2.88
     consist
    2.86
     constitute
    2.83
     have
    2.72
     belong
    2.71
    Activations Density 0.938%

    No Known Activations