© 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. 10480
    Prev
    Next
    INDEX
    Explanations

    adjectives describing the thickness or thinness of objects

    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
    19
    +0.16
    0.9%
    198
    +0.14
    0.8%
    156
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    19
    +0.16
    0.03
    198
    +0.14
    0.02
    452
    +0.13
    0.01
    Negative Logits
    YS
    -1.43
    Developer
    -1.41
     kindly
    -1.39
    ubicin
    -1.37
     Administrator
    -1.35
    APH
    -1.35
     ours
    -1.32
    inoc
    -1.30
     delegates
    -1.30
     promise
    -1.30
    POSITIVE LOGITS
    eners
    2.69
    ning
    2.67
    ening
    2.62
    ened
    2.59
    ningen
    2.34
    ness
    2.31
    ering
    2.25
    nings
    2.17
    est
    2.00
    eling
    1.97
    Activations Density 0.143%

    No Known Activations