© 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
    2. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    3. Pythia-70M-Deduped
    4. Residuals
    5. 2-RES-SM
    6. 28460
    Prev
    Next
    INDEX
    Explanations

    occurrences of the word "attend" and its variations

    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
    258
    +0.12
    0.7%
    362
    +0.12
    0.7%
    24
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    24
    +0.12
    0.03
    282
    +0.12
    0.02
    387
    +0.12
    0.03
    Negative Logits
    Ģ
    -1.83
    į
    -1.73
    «
    -1.65
    ¿½
    -1.63
    arity
    -1.63
    Eq
    -1.58
    CHANTABILITY
    -1.56
    naire
    -1.55
    İ
    -1.55
     sense
    -1.49
    POSITIVE LOGITS
     school
    1.93
     meetings
    1.83
     college
    1.80
    ugin
    1.70
     conferences
    1.66
     university
    1.66
     least
    1.61
    wan
    1.61
     graduating
    1.60
    antry
    1.58
    Activations Density 0.236%

    No Known Activations