© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    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. 2-RES-SM
    11. 28069
    Prev
    Next
    INDEX
    Explanations

    instances of the word "extend" and its variations, indicating a focus on the concept of prolongation or continuation

    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
    247
    +0.14
    0.8%
    317
    +0.14
    0.8%
    280
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    317
    +0.14
    0.02
    380
    +0.14
    0.02
    247
    +0.13
    0.02
    Negative Logits
    ubicin
    -2.52
    ksen
    -1.69
    qui
    -1.69
    oi
    -1.69
    ģ
    -1.64
    ±
    -1.56
    IED
    -1.55
     mistake
    -1.54
    bered
    -1.51
    ubot
    -1.50
    POSITIVE LOGITS
    ibility
    1.99
     lifespan
    1.94
    iveness
    1.92
    ivity
    1.78
    ibilities
    1.69
    enden
    1.65
     beyond
    1.61
    izability
    1.56
     invariance
    1.55
    ible
    1.55
    Activations Density 0.113%

    No Known Activations