© 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. 31401
    Prev
    Next
    INDEX
    Explanations

    phrases and structures that indicate legal or formal 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
    419
    +0.17
    1.0%
    286
    +0.12
    0.7%
    197
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    419
    +0.17
    0.03
    23
    +0.12
    0.05
    69
    +0.11
    0.04
    Negative Logits
    /.
    -1.78
    heit
    -1.64
    /"
    -1.59
     player
    -1.55
    osition
    -1.46
    nergy
    -1.45
     stance
    -1.41
    iances
    -1.41
    /-/
    -1.39
    pa
    -1.38
    POSITIVE LOGITS
     deaths
    1.74
     fatal
    1.73
    plain
    1.72
    apine
    1.63
     justice
    1.58
     severe
    1.54
     death
    1.54
    opsies
    1.52
    circum
    1.48
    oubt
    1.48
    Activations Density 0.565%

    No Known Activations