© 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 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. 10650
    Prev
    Next
    INDEX
    Explanations

    instances of the word "Hello" across various contexts

    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
    201
    +0.14
    0.8%
    336
    +0.11
    0.6%
    268
    +0.10
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    268
    +0.14
    0.02
    334
    +0.11
    0.02
    112
    +0.10
    0.01
    Negative Logits
    ató
    -1.73
    aments
    -1.73
    éd
    -1.69
    ych
    -1.67
    º
    -1.67
    artan
    -1.62
    ¯
    -1.56
    ::
    -1.53
    į
    -1.53
    yr
    -1.53
    POSITIVE LOGITS
     skies
    2.05
     worlds
    2.03
     world
    1.97
     thunder
    1.81
     heavens
    1.75
     life
    1.75
     World
    1.74
    world
    1.72
     universe
    1.69
     heaven
    1.68
    Activations Density 0.227%

    No Known Activations