© 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. 4-RES-SM
    11. 12544
    Prev
    Next
    INDEX
    Explanations

    phrases that relate to the concept of goodness and moral intent

    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/4-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.4.hook_resid_post
    Hook Layer
    4
    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
    147
    +0.13
    0.7%
    294
    +0.12
    0.7%
    312
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    67
    +0.13
    0.00
    23
    +0.12
    0.06
    312
    +0.12
    0.06
    Negative Logits
    ·
    -2.62
    «
    -2.58
    ¡
    -2.57
    £
    -2.57
    Ģ
    -2.54
    ¦
    -2.51
    Ń
    -2.47
    ¸
    -2.42
    °
    -2.41
    ³
    -2.39
    POSITIVE LOGITS
     Good
    3.27
     bad
    3.25
     Bad
    3.08
     Poor
    2.82
     good
    2.73
    good
    2.72
     ugly
    2.66
     Excellent
    2.66
     GOOD
    2.54
     poor
    2.52
    Activations Density 1.385%

    No Known Activations