© 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 EvalsExportsinterp-engineNEWAPI 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. 13999
    Prev
    Next
    INDEX
    Explanations

    instances of the word "contrast" and similar comparative terms

    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
    335
    +0.13
    0.8%
    452
    +0.12
    0.7%
    46
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    431
    +0.13
    0.01
    335
    +0.12
    0.02
    46
    +0.11
    0.01
    Negative Logits
    ĻĤ
    -2.32
    »¿
    -2.06
    IJ
    -2.05
    ½
    -2.05
    Ĥ
    -1.92
    Īĺ
    -1.91
    Ļ
    -1.90
    ĺ
    -1.89
    ľĵ
    -1.88
    Ħ
    -1.87
    POSITIVE LOGITS
    iveness
    1.69
    ingly
    1.68
     material
    1.60
    actic
    1.59
    mann
    1.59
    ius
    1.56
    ]{.
    1.54
    gery
    1.53
     screening
    1.53
     groups
    1.53
    Activations Density 0.144%

    No Known Activations