© 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. 10908
    Prev
    Next
    INDEX
    Explanations

    variations of the word "slow" and related concepts implying reduced speed or pace

    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
    128
    +0.17
    0.9%
    361
    +0.14
    0.8%
    477
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    361
    +0.17
    0.03
    148
    +0.14
    0.01
    253
    +0.13
    0.02
    Negative Logits
    crets
    -1.82
    OLOG
    -1.51
    stanbul
    -1.49
    xtures
    -1.44
     dollars
    -1.43
    supervised
    -1.43
    ICE
    -1.41
     Agency
    -1.41
    sources
    -1.41
    documents
    -1.40
    POSITIVE LOGITS
     downward
    1.85
     streak
    1.82
     descent
    1.77
     onset
    1.73
     gait
    1.72
     pace
    1.72
     convergence
    1.64
     period
    1.64
     down
    1.62
    down
    1.62
    Activations Density 0.272%

    No Known Activations