© 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. 31678
    Prev
    Next
    INDEX
    Explanations

    references to HTML or XHTML document 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/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
    345
    +0.14
    0.8%
    365
    +0.12
    0.7%
    334
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    345
    +0.14
    0.03
    334
    +0.12
    0.03
    19
    +0.11
    0.02
    Negative Logits
     quo
    -1.59
     whilst
    -1.57
    izability
    -1.53
     weighted
    -1.52
     while
    -1.49
     Jr
    -1.46
     patented
    -1.43
     reversal
    -1.41
     elimination
    -1.41
    %).
    -1.39
    POSITIVE LOGITS
    docs
    1.79
    antics
    1.64
    antic
    1.60
    ish
    1.52
    comments
    1.50
    tica
    1.44
    ienn
    1.39
    iska
    1.39
     composer
    1.39
    inki
    1.38
    Activations Density 0.067%

    No Known Activations