© 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. 32114
    Prev
    Next
    INDEX
    Explanations

    mathematical operations and relationships

    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
    219
    +0.13
    0.7%
    354
    +0.13
    0.7%
    255
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    67
    +0.13
    0.02
    187
    +0.13
    0.02
    125
    +0.12
    0.01
    Negative Logits
    Ļª
    -1.80
    ĻĤ
    -1.74
    ĨĴ
    -1.73
    )",
    -1.69
    ı
    -1.62
    )">
    -1.60
     (@
    -1.47
    Ĭ
    -1.42
     vice
    -1.41
     flags
    -1.40
    POSITIVE LOGITS
    bard
    1.35
    ABA
    1.33
    Decided
    1.26
     comprehension
    1.19
    assisted
    1.18
    Submitted
    1.18
    keeper
    1.17
    aching
    1.16
    Illustration
    1.16
     Redistributions
    1.15
    Activations Density 0.068%

    No Known Activations